Description

Molecular formula: SiO₂

Silicon–oxygen tetrahedra (SiO₄) are the basic units of all crystalline and amorphous forms reported in Section 1.1.1 (with the exception of stishovite, in which, under extreme pressure conditions, silicon is forced to bind to six oxygen atoms in an octahedral coordination). In each silicon–oxygen tetrahedron, each silicon atom is surrounded by four oxygen atoms; each oxygen atom is shared by two tetrahedra.

The three-dimensional framework of crystalline silicas is determined by the regular arrangement of the tetrehedra, which share each of their corners with another tetrahedron. Differences in the orientation and position of the tetrahedra create the differences in symmetry and cell parameters that give rise to the various polymorphs. In the case of quartz, the structural feature is a helix composed of tetrahedra along the c-axis. The helices have a repeat distance of three tetrahedra. The winding of the helices can be left- or right-handed, which results in the enantiomorphism of quartz crystals (Frondel, 1962).

The phases of silica and their crystalline structures have been extensively studied and several surveys have been carried out (e.g. Frondel, 1962; Wycoff, 1963; Sosman, 1965). Table 2 reports symmetry, lattice parameters (the unit cell dimensions a, b, c), density and the strongest lines (d values) obtained by X-ray diffraction of the various natural polymorphs stable or metastable at room temperature.

Table 2

Crystallographic data of silica polymorphs.

The silicon–oxygen bond is regarded as partially ionic (that is, close to 1 : 1 ionic to covalent bond character). The mean Si–O distance in tetrahedral polymorphs is 0.161–0.162 nm and the mean O–O distance 0.264 nm. The variation in the Si–O–Si bond angles and the almost unrestricted rotation of adjacent tetrahedra around the bridging oxygen atom account for the variability of silica frameworks (Flörke & Martin, 1993).

The ²⁹Si nuclear magnetic resonance (NMR) peaks of the framework of silica polymorphs appear at the highest field of the ²⁹Si chemical shift range of silicates. The shifts observed for silica polymorphs range from −107 to −121 ppm. The chemical shift differences observed for the various polymorphs are due only to changes in the structural arrangement of the SiO₄ tetrahedra within the silica backbone. Quantitative correlations between the observed chemical shifts and several geometrical parameters, typically bond angles, have been established (Engelhardt & Michel, 1987). In this context, NMR appears to be a promising technique for a better insight into the crystal structure and for the identification of the various polymorphs.

The amorphous silica forms are also composed of tetrahedra sharing their oxygen atoms. However, in these silicas the orientation of the bonds is random and lacks any long-range periodicity. The lack of crystal structure is shown by the absence of sharp lines in an X-ray diffraction, although some short-range organization may still be present.

A large variety of amorphous silicas have been prepared for different uses (Section 1.1.1), the properties of which are described by Iler (1979) and Bergna (1994). These amorphous silicas differ in particle form and size, porous structure and residual water content.

The micromorphology of silica particulates to which people are exposed (respirable size range) depends not only upon crystallinity but also upon the way in which the silica particulates were formed. Ground samples — whether from crystalline or vitreous forms — have very acute edges and a marked heterogeneity in particle size; smaller particles are held at the surface of bigger ones by surface charges (Fubini et al., 1990). Diatomaceous earths and even cristobalite particles derived from diatomaceous earths have an almost infinite variety of shapes; this variation has its origins in the living matter from which they originated (Iler, 1979). Pyrogenic amorphous silicas are aggregates of non-porous, smooth, round particles and are totally different from the forms found in diatomaceous earths (Ettlinger, 1993; Ferch & Toussaint, 1996). In precipitated silicas, the size of the particle morphology and the extent of the inherently porous structure are dependent upon the procedure used in their preparation.

The surface areas of ground samples of crystalline or vitreous silica depend on the grinding procedure and vary between 0.1 and 10–15 m²/g. Diatomites have a rather broad range of surface areas, which, after calcination, fall mostly into the range 2–20 m²/g. Pyrogenic amorphous silicas have surface areas ranging from 50 to 400 m²/g. Precipitated amorphous silicas have a very variable specific surface area ranging between 50 and nearly 1000 m²/g because of their porous structure and the small size of the particles.

Quartz particles often have a perturbed external amorphous layer (known as the Beilby layer; Fubini, 1997). Removal of this layer by etching improves the crystallinity and increases the fibrogenic potential of the dust (King & Nagelschmidt, 1960).

(a) Solubility in water

Silica is rather poorly soluble in water and solubility is higher for the amorphous than for the crystalline morphologies. The solubility of the various phases of silicas is very complex and depends upon several factors (Iler, 1979). Solubility increases with temperature and pH and is affected by the presence of trace metals. Particle size influences the rate of solubility. The external amorphous layer in quartz (the Beilby layer) is more soluble than the crystalline underlying core.

(b) Reactivity

Silica is attacked by alkaline aqueous solutions, by hydrofluoric acid and by catechol (Iler, 1979). The rates of etching in hydrofluoric acid vary in the following sequence (King & Nagelschmidt, 1960; Flörke & Martin, 1993): stishovite < coesite < quartz < tridymite, cristobalite < vitreous silica (Coyle, 1982).

Etching in hydrofluoric acid eliminates the Beilby layer (Fubini et al., 1995) on quartz (see Section 1.1.2). Stishovite is almost insoluble in hydrofluoric acid and coesite reacts at a much lower rate than quartz or vitreous silica. Hydrofluoric acid solutions can thus be used to separate the various polymorphs (Stalder & Stöber, 1965).

(a) Hydration and hydrophilicity

The surface of silica reacts with water vapour from the ambient air to form an external layer of silanols (SiOH). This process may be extremely slow (smooth surfaces with stable siloxane bridges, Si–O–Si) or very fast (fresh defective surfaces, strained siloxane bridges). The part of the surface covered by a dense layer of silanols is is hydrophobic (Bolis et al., 1991). Under the same conditions of humidity, the various polymorphs show a different degree of hydrophilicity — quartz and stishovite being the most hydrophilic and pyrogenic amorphous silica the most hydrophobic (Cerrato et al., 1995).

(b) Mechanical fracture

Cleavage seldom occurs on defined crystal planes and fractures are conchoidal. Dusts originated by quartz grinding have a peculiar reactivity arising from the homolytic and heterolytic rupture of the silicon–oxygen bonds, which leaves unsatisfied valencies as unpaired electrons (surface radicals) and surface charges (Antonini & Hochstrasser, 1972; Fubini et al., 1989). A similar, even more pronounced effect takes place with tridymite and cristobalite (Fubini et al., 1989, 1990). The effect is less pronounced with coesite and does not occur with stishovite (Fubini et al., 1995). If grinding is performed in dry air, oxygen or hydrogen peroxide aqueous solutions, reactive oxygen species (ROS) — SiO₂ and Si⁺O₂: —are formed (Dalal et al., 1989; Fubini et al., 1989, 1990; Fubini, 1997). Conversely, if grinding takes place in a wet atmosphere, silanols are formed rather than surface radicals (Volante et al., 1994).

In aqueous suspensions, freshly ground surfaces generate ROS (Vallyathan et al., 1988). Whether the ROS arise from the silica itself or from certain impurities exposed at the surface during the grinding procedure is still under debate; acid washing decreases the radical yield (Miles et al., 1994).

(c) Thermal treatments

The presence and extent of silanols at the surface of a silica sample determines its hydrophilicity. Upon heating, silanols condense into siloxanes with elimination of water:

This reaction progressively converts hydrophilic surfaces to hydrophobic ones (Fubini et al., 1995). When cooling down under ambient conditions, some water uptake takes place, with partial reconversion of siloxanes into silanols. However, high temperature and prolonged heating stabilize surface siloxane with consequent inhibition of rehydroxylation. The surface is thus metastably hydrophobic and remains as such for very long periods of time.

The above reaction occurs more readily with amorphous silicas, in which the silicon tetrahedra are able to move more easily than in the crystalline forms where the silanols are stabilized in ordered arrays. As a consequence, crystalline particles are more hydrophilic than amorphous ones when submitted to the same heating procedure.

Heating also removes defects and radicals originated by grinding (Fubini et al., 1989).

(d) Etching

Etching with hydrofluoric acid, alkaline hydroxides or catechol modifies the surface of silica samples (Iler, 1979). External layers are attacked with the progressive elimination of surface radicals (Costa et al., 1991). With hydrofluoric acid, the external surface is smoothed out and the specific surface area decreases. This effect is due to the smoothing out of the fractal part of the surface and to the total dissolution of smaller particles rather than the simple reduction in the dimensions of each particle (Fubini et al., 1995). As a consequence, the size distribution of an etched sample reveals a higher proportion of larger particles.

Etching also eliminates impurities that can modulate silica toxicity (King & Nagelschmidt, 1960; Nash et al., 1966; Nolan et al., 1981).

(e) Metal contaminants

Metal impurities modify the surface reactivity of silica samples. Aluminium decreases silica solubility (Iler, 1979). Transition metal ions (typically iron), adsorbed at the surface, activate the production of free radicals in aqueous suspensions (Vallyathan et al., 1988).

(a) Air sampling and analysis for silica

In the past, assessment strategies for airborne crystalline silica were generally based on particle count procedures. Using sampling instruments such as the konimeter, thermal precipitator or impinger, airborne dust samples were collected and then examined by light microscopy. In some cases, selective counting rules were used to reject particles not considered to be respirable (Hearl & Hewett, 1993) or, in the case of mixed dust, not considered to be silica (see Section 1.3.2 for further information on historical sampling methods for occupational exposure).

Currently, filter collection methods, coupled with X-ray diffraction or infrared spectrophotometry (IR) are favoured for the assessment of the silica concentration of airborne dusts. In the case of crystalline silica, most countries (e.g. the United States, the United Kingdom, Germany, Japan and Australia) require that the sample be restricted to the respirable fraction. In contrast, amorphous silica can also be assessed using a total dust sample. Based on the information available, it appears that, internationally, the X-ray diffraction and IR methods are equally acceptable, except in Sweden and Japan in which only the X-ray diffraction method is permitted (Madsen et al., 1995).

One standard procedure in the United States for crystalline silica (NIOSH Method 7500) employs a sampling train fitted with a 10-mm nylon cyclone and a polyvinyl chloride (PVC) membrane filter, running at a 1.7 L/min flow rate. After sampling is complete, the filter is removed and subjected to low-temperature ashing or dissolution, and the resulting dust is assessed for crystalline silica using X-ray diffraction. NIOSH Method 7602 is similar, but uses IR for analysis.

In NIOSH Method 7501 for amorphous silica, the sample is subjected to X-ray diffraction analysis before and after heating to 1500 °C (fumed silica) or to 1100°C (other amorphous silica). The concentration of amorphous silica is calculated from the difference in the two cristobalite concentrations (Eller & Cassinelli, 1994).

Quartz, tridymite and cristobalite can be distinguished by X-ray diffraction because their strongest reflections (i.e. peaks in the diffractograms) are different (see Section 1.1.2). The detection limit in respirable dust samples is about 5 µ for quartz and 10 µg for cristobalite; these limits approximate to an atmospheric level of 0.01–0.02 mg/m³ for a 0.5 m³ air sample (Bye et al., 1980; Bye, 1983).

(b) Surface analysis

In view of the most recent results, bulk analysis alone does not appear to be sufficient to predict the level of biological activity — it is largely the exposed surface of silica that determines its toxicity. Clay occlusion of respirable quartz particles may be detected by low-voltage scanning electron microscopy X-ray analysis (Wallace et al., 1990). An alternative way is to determine the surface composition by laser microprobe mass analysis, which examines the outermost layers of individual particles (Tourmann & Kaufmann, 1994). The technique uses a laser to vaporize and ionize a small volume of material near the surface of a single particle. The ions generated are identified with a time-of-flight mass spectrometer.

Several techniques can be used to examine the various surface properties of particulate materials such as adsorption capacity, hydrophilicity and potential for free radical release; these techniques are described in Fubini (1997). Detailed surface analysis can be carried out with X-ray photoelectron spectroscopy, scanning electron microscopy with energy dispersion X-ray analysis, and several adsorption techniques. These techniques are too sophisticated for routine analysis.

Recent data reveal variation in the biological responses to crystalline silica samples that are identical in their bulk properties (Hemenway et al., 1994; Daniel et al., 1995; Fubini et al., 1995). These differences must be related to surface properties — hydrophilicity, surface radicals, defects — or to different levels of surface impurities. In both hypotheses, surface analysis is required to define the potential hazard of a given dust.

(a) Sand and gravel

Silica-bearing deposits are found on every continent and from every geological era. The majority of deposits that are mined for silica sands consist of free quartz, quartzites and sedimentary deposits, such as sandstone (Harben & Bates, 1984).

Industrial sand and gravel, often referred to as ‘silica sand’ and ‘quartz sand’, include high-silica-content sand and gravel (United States Department of the Interior, 1994). Table 4 summarizes recent data on the production of silica sand in major producing countries.

Table 4

Silica sand and gravel production.

Processing operations depend both on the nature of the deposit and on the end product required. They generally include crushing and milling for refining particle size and wet/dry screening to separate very fine particles (Davis & Tepordei, 1985).

(b) Quartz crystals

Two different kinds of production may be distinguished: (i) the processing of naturally occurring quartz; and (ii) the hydrothermal culturing of quartz.

The largest reserves of highly pure quartz occur in Brazil. Minor deposits are found in Angola, India, Madagascar and the United States.

Hydrothermally cultured quartz crystals are of major economic importance and their use is growing rapidly. Hydrothermal synthesis consists of crystal growth or reaction at high pressure and temperature in aqueous solution in sealed steel autoclaves (Flörke & Martin, 1993). Synthetic quartz crystal production is concentrated in Japan, Russia and the United States. Smaller production capacity exists in Belgium, Brazil, Bulgaria, China, France, Germany, the Republic of South Africa and the United Kingdom (United States Department of the Interior, 1994).

(c) Refractory silica

Silica bricks are manufactured from mixtures of ground quartz arenite and quartzite and are fired at 1450–1600 °C. They are used in certain high-temperature processes and are generally produced in batches. Silica used in refractories must contain > 93% in weight SiO₂. Quartz is mostly transformed into cristobalite, but tridymite is also formed under the action of mineralizers (mainly CaO). The brick consists of nearly equal amounts of cristobalite, tridymite, residual quartz and a glass phase (Flörke & Martin, 1993).

(d) Diatomite

Diatomite is obtained from sedimentary rocks that are mainly composed of the skeletons of diatoms. These skeletons are composed of opal-like amorphous silica and exhibit a wide range of porous fine structures and shapes, which are altered upon calcination. Even the calcined and crystallized product, however, partially retains the original biogenic micromorphology (Iler, 1979). Particle-size distribution, shape and fine structure vary from one deposit to another (Benda & Paschen, 1993).

The most notable commercial source of diatomite is in California, United States, where there is a marine deposit of unusual purity over 300 m thick. Other major deposits that are mined occur in Algeria, Denmark, France, Iceland and Romania (Dickson, 1979; Reimarsson, 1981; Harben & Bates, 1984; Benda & Paschen, 1993).

Diatomite is mined almost exclusively by opencast methods, using bulldozers and other similar equipment to remove the material. Some diatomite is mined underground in Europe, Africa, South America and Asia. In one operation in Iceland, where the mineral lies under water, slurried material is transferred by a pipeline to a processing plant (Kadey, 1975). The processing methods for crude material are fairly uniform worldwide. The general procedure is described by Benda & Paschen (1993) and can be schematized as follows:

Preliminary size reduction → drying, grinding → dried, fine diatomite

Dried, fine diatomite → furnace, 800–1000 °C → grinding → calcined diatomite

Calcined diatomite → alkaline flux, 1000–1200 °C → grinding → flux-calcined diatomite

Calcination and, even more so, flux calcination yield a considerable amount (up to 65%) of crystalline material (cristobalite) (Benda & Paschen, 1993). During calcination, porosity area and specific surface strongly decrease. Some chemical and physical properties of commercially available diatomites used for filtering or as fillers are reported in Table 5.

Table 5

Chemical and physical properties of some commercial diatomites.

The major producing country is the United States, followed by Denmark and France. Diatomite production by region during the years 1970–94 is presented in Table 6.

Table 6

World diatomite production 1970-94.

(e) Synthetic amorphous silicas

Commercial/synthetic amorphous silicas have been classified (Ferch & Toussaint, 1996) as ‘wet process’ silicas (including precipitated silicas and silica gels), pyrogenic (‘fumed’) silicas and surface-modified silicas. Surfaces of the modified silicas have been rendered hydrophobic, for example, by silylation with dimethyl dichlorosilane.

Worldwide production of synthetic amorphous silicas in 1995 was estimated at 1100 thousand tonnes, including 900 thousand tonnes precipitated silicas, 90 thousand tonnes silica gels and 110 thousand tonnes pyrogenic silicas (Ferch & Toussaint, 1996).

(ii) Pyrogenic silicas

The manufacturing process for pyrogenic silicas is based mainly on the combustion of volatile silanes, especially silicon tetrachloride, in an oxygen–hydrogen burner. Primary particles (7–50 nm particle size) of amorphous silica fuse together in the high-temperature flame to yield stable aggregates of between 100 and 500 nm in diameter. These aggregates form micron-sized agglomerates. The finely divided silica is separated from the hydrochloric acid-containing off-gas stream in filter stations. The hydrochloric acid content of the product is commonly reduced to less than 100 ppm by desorbing the hydrochloric acid with air in a fluid-bed reactor. Pyrogenic silica appears as a fluffy white powder (Ettlinger, 1993; Ferch & Toussaint, 1996). Physico-chemical characteristics of pyrogenic and ‘wet-process’ silicas are given in Table 7.

Table 7

Characteristic properties of synthetic amorphous silicas.

(a) Sand and gravel

Silica sand has been used for many different purposes for many years; its most ancient and principal use throughout history has been in the manufacture of glass (Davis & Tepordei, 1985). Sands are used in ceramics, foundry, abrasive, hydraulic fracturing applications and many other uses (Table 8).

Table 8

Industrial sand and gravel sold or used by United States producers in 1994, by major end use.

As illustrated in Table 8, several uses require the material to be ground. In some uses (e.g. sandblasting, abrasives), grinding also occurs during the use.

Refractory silica bricks, in which silica is converted by heat into cristobalite and tridymite are used in sprung arches of open-hearth furnaces, covers of electric furnaces, roofs of glass-tank furnaces, blast pre-heaters and coke and gas ovens (Flörke & Martin, 1993).

(b) Quartz crystals

Quartz has been used for several thousand years in jewellery as a gem stone (e.g. amethyst, citrine). Large quantities of pure crystals were required when the application of pure quartz in the electronics industry was discovered. At present, the major demand comes from both electronics and optical components industries.

An electronic-grade quartz crystal is a single-crystal silica that is free from defects and has piezoelectric properties that permit its use in electronic circuits for accurate frequency control, timing and filtering. These uses generate most of the demand for electronic-grade quartz crystals. A smaller amount of optical-grade quartz crystal is used in windows and lenses in specialized devices, including some lasers. Cultured (synthetic) quartz has replaced natural crystal in most of these applications (United States Department of the Interior, 1994).

(c) Diatomites

The main uses of diatomites are in filtration (60% of world production), as fillers (25% of world production) and in other uses (insulators, absorption agents, scourer in polishes and cleaners, catalyst supports, packing material) (Benda & Paschen, 1993).

The intricate microstructure and high pore-space volume of diatomite have made it a major substrate for filtration. Diatomite has been used to filter or clarify dry-cleaning solvents, pharmaceuticals, beer, wine, municipal and industrial water, fruit and vegetable juices, oils and other chemical preparations (Kadey, 1975).

The next most important application of diatomite is as a filler in paint, paper and scouring powders. It imparts abrasiveness to polishes, flow and colour qualities to paints and reinforcement to paper. It is also used as a carrier for pesticides, a filler in synthetic rubber goods, in laboratory absorbents and in anti-caking agents (Kadey, 1975; Sinha, 1982).

(d) Synthetic amorphous silicas

Consistent with their physico-chemical and morphological properties, the different classes of synthetic amorphous silicas find uses in very different areas of application. However, most of the applications are related to the reinforcement of various elastomers, the thickening of various liquid systems, the free-flow of powders or as a constituent of matting, absorbents and heat insulation material (Ferch & Toussaint, 1996). Table 9 lists the major applications of synthetic amorphous silicas.

Table 9

Major applications of finely divided synthetic amorphous silicas.

Crystalline silica

Quartz in its α form is abundant in most rock types, sands and soils. Table 10 reports the average quartz composition of major igneous and sedimentary rocks. Important differences can be observed in the composition of the various rocks. In igneous rocks, quartz is a common component of acid (granitic) and intermediate (e.g. syenites, andesites) plutonic rocks. However, quartz occurs at very low levels or is absent from the basic and ultra-basic varieties (e.g. trachytes, gabbros, olivines, peridotite). Quartz may also be present in a variety of volcanic tuffs (United States Bureau of Mines, 1992).

Table 10

Average quartz composition of major igneous and sedimentary rocks.

Quartz, being a hard, inert and insoluble mineral, endures through the various weathering processes and is found in trace to major amounts in a variety of sedimentary rocks. It is a major component of soils, composing 90–95% of all sand and silt fractions in a soil. There are a variety of sandstones, including orthoquartzite in which the grains are 95% quartz and the cement is a precipitate or a film of clay. Greywacke is considered a variety of sandstone. Quartz is also common in siltstone. Sand is composed of quartz predominantly, while gravel is of variable composition. Argillaceous rocks, including shales, clays and mudstones, may contain substantial amounts of quartz, depending on the varieties. Wyoming bentonite, a valuable clay, contains up to 24% crystalline silica (quartz and cristobalite). In coal, quartz constitutes typically up to 20% of the mineral matter (Greskevitch et al., 1992). Illinois coal has been reported to contain 1.2–3.1% quartz. Diatomaceous earth typically contains 0.1–4% quartz. Limestones may contain a small proportion of quartz (Atkinson & Atkinson, 1978; Harben & Bates, 1984; United States Bureau of Mines, 1992; Klein, 1993; Ross et al., 1993; Parkes, 1994; Weill et al., 1994).

In metamorphic rocks, quartz is a common constituent either an an original constituent, as a product of the metamorphic process or by crystallization from silica-bearing fluids. It is an important constituent of metamorphic phyllites, mica schists, migmatites, gneiss and quartzites. It has been reported to comprise 31–45% of the mineral content of Ardennes slate and 20–50% of taconite (Atkinson & Atkinson, 1978; United States Bureau of Mines, 1992; Heaney & Banfield, 1993; Ross et al., 1993).

Quartz is the primary gangue (or matrix) mineral in the metalliferous veins of ore deposits. In nature, quartz can also be found in important colour varieties — amethyst, citrine, smoky quartz, morion, tiger's eye — which are valued as semi-precious stones. Quartz crystals are frequently found in cavities and also occur in hollow globular forms called geodes (Atkinson & Atkinson, 1978).

Tridymite and cristobalite, formed during the devitrification of siliceous volcanic glass, can be found as fine-grained crystals in acid volcanic rocks. Furthermore, cristobalite is present in some bentonite clays and may be present as traces in diatomite (Heaney & Banfield, 1993; Ross et al., 1993; Parkes, 1994).

Coesite and stishovite have been found in rocks that equilibrated in short-lived high pressure environments, such as meteoritic impact craters. Keatite has been found in high-altitude atmospheric dusts, which are believed to originate from volcanic sources (Heaney & Banfield, 1993; Guthrie & Heaney, 1995).

Amorphous silica

Amorphous silica is widespread in nature as biogenic silica and non-biogenic silica glass.

Silica glass forms as volcanic glass (obsidian) from extrusive magmas, as lechate-lierite within tektites associated with meteorite impact craters and as fulgurite resulting from lightning strikes on unconsolidated sand or soil (Heaney & Banfield, 1993).

Silica of biological origin is produced by diatoms, radiolarians and sponges which extract silica dissolved in water to form their structures or shells. Biogenic amorphous silica levels in diatoms vary with species and range from less than 1% to almost 50% by weight. Siliceous oozes on the sea floor, which derive from the skeletons of diatoms, solidify to form opaline deposits. Opaline materials characterize diatomaceous earth deposits and are also found in bentonite clays. Diatomaceous earth is typically 90% amorphous silica (Heaney & Banfield, 1993; Ross et al., 1993; Rabovsky, 1995).

Biogenic silica is also produced by a variety of plants. Internal silicification of plant tissues promotes structural integrity and affords protection against plant pathogens and insects. The silica content is especially high in grasses, and silica can account for approximately 20% of the dry weight of rushes, rice and sugar cane. Amorphous silica in plants may be deposited as nodules or phytoliths — very tiny pure amorphous silica grains of a myriad of shapes and sizes — in many plants and trees (Heaney & Banfield, 1993). Some of the amorphous silica in plants (e.g. sugar cane, canary grass, wheat, rice, conifer needles) exists as fibres or spicules of various forms. Plant biogenic silica is released to the soil through burning or normal decay; soil concentrations are typically in the range of < 1 to 3% (Newman, 1986; Boeniger et al., 1988; Lawson et al., 1995).

Crystalline silica

Because of the extensive natural occurrence of crystalline silica in the earth's crust and the wide uses of the materials in which it is a constituent, workers may be exposed to crystalline silica in a large variety of industries and occupations. Thus, between 1980 and 1992, compliance officers of the United States Occupational Safety and Health Administration found respirable quartz to be present in samples taken in 255 industries of differing Standard Industrial Classification codes, excluding mining. In 48% of those industries, average overall exposure exceeded permissible exposure levels (Freeman & Grossman, 1995).

Crystalline silica is probably one of the most documented workplace contaminants; the severity of its health effects and the widespread nature of exposure have been long recognized. Reviews on occupational exposures to crystalline silica can be found in a number of reports (United States National Institute for Occupational Safety and Health, 1974, 1983; World Health Organization, 1986; Hilt, 1993; Weill et al., 1994). Table 11 presents a number of industries, jobs or operations where occupational exposure to crystalline silica has been reported, together with the origin or source of the silica.

Table 11

Main activities in which workers may be exposed to crystalline silica.

Although not exhaustive, the following section focuses on representative data in the main industries where quantitative exposure levels are available in the published literature and/or where major occupational health studies have been conducted. These include mines and quarries, foundries and other metallurgical operations, ceramics and related industries, construction, granite, crushed stone and related industries, sandblasting of metal surfaces, agriculture and miscellaneous other operations.

The reporting of exposure levels to crystalline silica in the scientific literature has changed considerably over the years with the evolution of the various sampling techniques and strategies, the development of improved analytical methods and the formulation of occupational exposure limits reflecting advances in the understanding of particle penetration and effects in the respiratory system.

In the first half of the twentieth century, sampling techniques varied from country to country, and airborne particles were collected with a variety of devices, such as koni-meters, Owen's jets, electrostatic or thermal precipitators, and impingers (Patty, 1958; Ayer, 1969; Harris & Lumsden, 1986). Exposure levels were usually reported as number of particles per unit volume, with particles counted by microscopy. No relationship could be established between the results of these various older methods.

In the United States, impinger methods (Greenburg-Smith or midget impingers) were commonly in use until the early 1970s. Dust levels, whether based on counts from an impinger or on mass collected on a filter, were associated frequently with data on the crystalline silica content of the dust. Considerable differences in estimates of crystalline silica content obtained by these methods may result, depending on the nature of interfering materials, on the analytical techniques used (whether chemical, petrographic or spectroscopic) and on the origin of the dust sample being analysed (Patty, 1958; Harris & Lumsden, 1986).

Crystalline silica content has been found to be usually smaller in airborne than in settled dust and in respirable than in total airborne dust (Ayer, 1969; Hearl & Hewett, 1993). However, there are exceptions to this general rule (Jorna et al., 1994).

Various limitations of the impinger method led to its decreasing use; sampling times were too short (10–30 min), the complexity of the sampling procedure prevented personal samples being taken, the impinger could not trap particles < c. 0.5–0.7 µm in size, and there was also found to be large inter-observer variability (Patty, 1958; Ayer, 1969; Hearl, 1996). On the other hand, however, total mass concentration, as collected on a filter, had the disadvantage of not being able to take into account particle size, which plays a major role in the hazards associated with crystalline silica inhalation (Ayer, 1969).

The introduction in the 1970s and the current generalized use of respirable mass sampling methods in most countries has made it possible to compare data realistically between various studies. In addition, conversion factors can be applied to filter-respirable mass concentration (in mg/m³) and impinger-particle count levels (in million particles per cubic foot; mppcf) to integrate past and present evaluations. Conversion factors may differ, however, depending on the nature of the dust (Sheehy & McJilton, 1987; Montgomery et al., 1991).

A number of factors remain to be taken into account when evaluating present data. There are uncertainties in the interpretation of analytical data for microcrystalline silica, or data taken in the presence of various interfering substances; in addition, in cases where the particle size distribution is widely different from that of the standards used, interpretation can be uncertain (Hearl & Hewett, 1993). More importantly, the representativeness of the data has to be evaluated in view of the sampling strategy used, recognizing that compliance inspection data usually have been obtained using a worst-case scenario strategy (Hearl & Hewett, 1993; Lippmann, 1995). A further complication results from the fact that respirable crystalline silica exposure levels are most often not reported directly in mg/m³ but indirectly in terms of a total respirable mass concentration. This total respirable mass concentration has to be compared (e.g. in the form of a severity factor) with an occupational exposure limit that varies with the content of crystalline silica in the dust. Furthermore, the current practice of collecting only respirable dust has been questioned by Lippmann (1995) who has argued that thoracic particles (i.e. those available for deposition within the airways of the thorax) may be more important for end-points such as lung and stomach cancer. Finally, it may be noted that industrial hygiene measurement practices do not take into account the surface area of particles, which may well be a relevant indicator of exposure. During the 1960s, crystalline silica dust exposure was measured in a study of South African gold mines as respirable surface area. This measurement was reported to be more strongly related to silicosis than the respirable particle count (Beadle, 1971).

(a) Mines

Occupational exposure to crystalline silica in mines originates from the dust generated from the ore being extracted or its associated rock. Mines are usually classified as surface or underground, coal, metal or non-metal; mines may be associated with various milling operations. In the United States, coal is the main mineral being mined primarily underground, together with antimony, lead, tungsten, molybdenum and silver. Surface mining accounts, however, for most of the metallic and non-metallic ores (Burgess, 1995). Exposure to crystalline silica in quarries, the crushed stone and related industries is detailed in a separate section. Exposure to silica in coal mines is covered in the monograph on coal dust in this volume.

The quartz content was determined in 2075 bulk settled dust samples collected from 1984 to 1989 in a representative sample of United States mines (491) from 66 different mineral commodities, including coal. Approximately 50% of all samples had a percentage of quartz above 5%; the overall average was 14%. Commodities with an average quartz percentage above 40% were sand/gravel and sandstone; those between 20 and 40% were copper, granite, lithium, mica, molybdenum, phosphate rock, shale, slate, stone, titanium and uranium–vanadium. For most commodities, wide variations were observed between the various samples. Labourers (surface) and bin pulley/truck loader workers were potentially exposed to bulk dust containing the highest percentage quartz. These data are only indicative of potential risk since settled dust composition may not be representative of inhalable or respirable dusts. Ninety-one per cent of samples analysed for cristobalite yielded non-detectable levels (< 0.75%) and only 4% contained more than 1% cristobalite, most of which came from diatomite calcining facilities (Greskevitch et al., 1992).

Respirable quartz levels of nearly 22 000 samples taken by inspectors from 1988 to 1992 in United States mines are summarized by commodity in Table 12. Mean exposure levels were usually below 0.1 mg/m³ but a significant percentage of samples were found to exceed the compliance limit. Mean quartz content of samples by commodity was rarely greater than 15%. Occupations at greatest risk of overexposure were found to be scoop tram, crusher, jackleg stoper drill and load–haul–dump operators (underground occupations); jackhammer and pneumatic drill operators (surface occupations); and packing, packaging or loading, labourer and bullgang workers (milling occupations). The authors indicate various limitations to the representativeness of this data set, the main ones being the compliance sampling strategy and the exclusion of samples containing less than 1% quartz or corresponding to less than 0.1 mg/m³ respirable dust. These effects would tend towards an overestimation of the exposure indicators (Watts & Parker, 1995).

Table 12

Respirable quartz exposures by commodity in United States mines (1988–92).

Estimates of exposure to respirable crystalline silica during the period 1950–87 in 20 Chinese mines (10 tungsten, six iron–copper and four tin) have been derived from industrial hygiene data and other historical exposure information. A 10-fold decrease was found between the periods 1950–59 and 1981–87 and the following arithmetic mean levels of respirable silica dust in mg/m³ were estimated to be as follows (older and most recent period, respectively): underground mining (4.89, 0.39), surface mining (1.75, 0.27), ore dressing (3.45, 0.42), tungsten mines (4.99, 0.64), iron and copper mines (0.75, 0.20) and tin mines (3.49, 0.45). In the surface mining operations, transport and service occupations generally had higher levels of exposure than mine production occupations; the opposite pattern was true in underground mining occupations, while the ore preparation workers were generally more exposed than ore separation or service workers in ore-dressing operations (Dosemeci et al., 1995).

Indications of past and present exposure levels of gold miners have been reported in a number of epidemiological studies. In South Africa, a high crystalline silica content of 30% in respirable dust at the Witwatersrand mine was reported. The levels of dust exposure were reduced during the 1930s to a level ranging from 0.05 to 0.84 mg/m³ for respirable quartz in underground dust (Beadle & Bradley, 1970). At the Homestake gold mine (South Dakota, United States), respirable dust contained 13% crystalline silica and, since engineering improvements in the early 1950s, levels of respirable silica have decreased substantially to within legal limits (Brown et al., 1986). In Ontario (Canada) gold mines, crystalline silica content in hard rock has been reported to vary from 4 to 12% and, based on konimeter data, past levels could have been significantly above current exposure limits (Kabir & Bilgi, 1993).

In two Sardinian mines (one for lead ore and the other for zinc ore), similar concentrations of respirable dust were estimated to be 3–5 mg/m³ in 1945–60 and 1.6–1.7 mg/m³ in 1981–88. However, quartz content differed significantly between the two mines (median values of 1.2% and 12.8%, respectively) because of differing wall rock composition (Carta et al., 1994). In a copper mine in Finland, respirable dust contained on average 18.3% quartz; the mean concentration of respirable quartz in the general mine air decreased from about 0.16 mg/m³ before 1965 to 0.08 mg/m³ after 1981 and the mean concentration where loading operations took place decreased from 0.8 before 1965 to 0.15 mg/m³ since 1975. In an old copper mine, respirable quartz was estimated to be above 2 mg/m³ during dry-drilling operations before 1940 (Ahlman et al., 1991). The mining and milling of diatomaceous earth may entail exposures to crystalline silica, notably to cristobalite formed from amorphous silica during the calcination process. Further details on occupational exposures in the diatomaceous earth industry may be found in the section on amorphous silica.

Beside crystalline silica, several other toxic hazards can be found in mines, such as carbon monoxide and nitrogen dioxide from blasting and engine exhausts, nickel and arsenic, depending on rock composition, aldehydes and polycyclic aromatic hydrocarbons from diesel engine exhausts, various metallic and non-metallic compounds such as asbestos, and ionizing radiation from radon daughters (Burgess, 1995). The extent of such exposures is strongly dependent on work practices and varies with commodity and specific vein composition. In the gold mines in the United States and Canada (Ontario) and the Sardinian lead and zinc mines mentioned above, and in a tin mine in south-east China, average working levels of radon daughters have been reported as ranging up to 0.3, which is within the accepted standard; substantial levels of radon daughters have been observed in Chinese copper mines and in some South African gold mines (Brown et al., 1986; Hnizdo & Sluis-Cremer, 1991; Wu et al., 1992; Kabir & Bilgi, 1993; Carta et al., 1994; Fu et al., 1994). Arsenic has been measured at average levels of a few (µg/m³ in the United States and Canadian gold mines; at the United States gold mine, amphibole asbestos fibres have been measured at mean levels of 0.44 and 1.16 fibres/mL for miners and surface crushers, respectively (Brown et al., 1986).

(b) Granite quarrying and processing, crushed stone and related industries

Granite rock, containing from 10 to about 30% quartz, is obtained in quarries and further processed into structural (dimensional) stone or crushed for road materials. Other rocks rich in crystalline silica such as sandstone, flint and slate are also subjected to various quarrying, milling and processing operations to produce building or road materials (Weill et al., 1994; Burgess, 1995). Respirable quartz exposure levels measured in various countries for various jobs in the granite quarrying and processing industries as well as the crushed stone and related industries are summarized in Table 13. Exposure data collected by inspectors in the United States appear in Table 12.

Table 13

Occupational exposure to crystalline silica in the granite quarrying and processing industries and the crushed stone and related industries in various countries.

Respirable crystalline silica levels are related to the crystalline silica content of the rock being quarried or milled; for example, levels have been found to be higher with flint than with granite (Guénel et al., 1989a), and with granite or sandstone than with limestone or traprock (Davies et al., 1994; Kullman et al., 1995). The higher exposure levels have usually been associated with the following jobs or operations: rock and stone drilling and cutting in quarries; dimensional stone cutting and finishing in sheds usually outside quarries; rock crushing, sieving and transport within or outside quarries. In three Russian quarries producing sand and gravel mixtures, the average respirable quartz levels in 1990 ranged from [0.44 to 4.46 mg/m³] for various stone crusher locations in cold periods of the year and from [0.77 to 1.87 mg/m³] in hot periods (Kiselev, 1990). In Hong Kong quarries producing crushed stone, average respirable quartz levels in 1982 were measured at 0.93 mg/m³ for rock drillers, from 0.10 to 0.42 mg/m³ for various crusher locations and from 0.11 to 0.19 mg/m³ for screening locations (Ng et al., 1987a). In United States granite quarries and sheds, control measures put in place during the late 1930s and the 1940s resulted in 10–100-fold reductions in what were very elevated dust levels (Davis et al., 1983). Granite stone-cutting is now usually associated with mean levels of respirable quartz below 0.1 mg/m³. In the industry as a whole, present control measures include water-mist injection during drilling, local exhaust ventilation, wet methods for cutting granite and the use of control cabins (Health and Safety Executive, 1992a; Davies et al., 1994; Burgess, 1995).

The presence of cristobalite has been reported in a limited number of samples in the road materials industry in Denmark (Guénel et al., 1989a) and traprock crushing operations in the United States (Kullman et al., 1995). Asbestos fibres and other fibrous minerals were found in one of 19 stone crushing facilities investigated in the United States (Kullman et al., 1995). Other constituents of the dusts would depend on the mineral being mined or milled (e.g. silicates, carbonates) (Kullman et al., 1995) and could include abrasives such as silicon carbide and aluminium oxides (Eisen et al., 1984).

Slate-pencil workers in India are exposed to respirable crystalline silica originating from the sawing of silica-rich [c. 40–50%] slate slabs. A survey of five plants in 1991 found personal respirable dust levels of 0.06–1.12 mg/m³ (average, 0.61 mg/m³; mean free silica content, 15%). Previous surveys in 1982 and 1971, before control measures were implemented, had found levels 10–100-fold higher (Fulekar & Alam Khan, 1995).

(c) Foundries

Occupational exposure to crystalline silica in foundries originates mainly in the use of sands in the making of moulds and cores. These sands have quartz contents of 5 to nearly 100%. Quartz and cristobalite, the latter being formed from quartz during the pouring of metal, may further contaminate the work environment during the knocking-out or shaking-out operations and during the removal of adherent sand from the castings by grinding or abrasive blasting operations. Other potential sources of crystalline silica are parting powders such as silica flour applied on the moulds as well as the maintenance and repair of silica-rich refractory materials used in furnaces and ladles (Weill et al., 1994; Burgess, 1995). Detailed descriptions of metal founding operations can be found in MacBain & Strange (1983), IARC (1984) and Burgess (1995).

Respirable quartz exposure levels measured for various jobs in foundries of various countries are summarized in Table 14. In general, the various studies concur in identifying high-exposure jobs as being related to sand preparation and reclamation, knocking-out or shaking-out, cleaning of castings (fettling, grinding, sandblasting), furnace and ladle refractory relining and repair. In two United States foundries where mullite sand was used as a refractory in moulds, personal respirable dust samples contained cristobalite up to 41% and the occupational exposure limit was exceeded 10 to 20 times in several operations, depending on the plant, notably during dipping, grinding and shaking-out. Cristobalite was present in the original mullite refractory and was also generated by heating the colloidal silica binder used in mould making (Janko et al., 1989). Lower levels usually found in non-ferrous foundries compared to iron and steel foundries have been explained by the lower pouring temperatures of the metal, which results in lower sand contamination of the castings. Other factors may be related to the size of foundries, the size of castings and production rates (Oudiz et al., 1983).

Table 14

Occupational exposure to crystalline silica in foundries in various countries.

Improvement in plant ventilation and work practices have been credited in a 10–20-fold lowering in respirable crystalline silica exposure levels of fettlers and coremakers between 1977 and 1983 in a United States grey iron foundry (Landrigan et al., 1986). Effective controls include well-designed and maintained local ventilation, baffles and air jets on the ventilation equipment of grinding machines, good housekeeping, use of vacuum systems and of wet sweeping, as well as isolation to prevent cross-contamination (Ayalp & Myroniuk, 1982; United States National Institute for Occupational Safety and Health, 1983; O'Brien et al., 1987, 1992; Health and Safety Executive, 1992b). The substitution of siliceous sands with olivine (olivine is a magnesium iron silicate that contains almost no free silica; Davis, 1979) sands results in decreased exposure levels (Gerhardsson, 1976; O'Brien et al., 1992), but contamination from processes using silica sand must be controlled (Davis, 1979). Silica flour parting powders can be replaced by low-silica powders such as those containing olivine or zircon (Landrigan et al., 1986, Weill et al., 1994).

Although crystalline silica represents a major potential air contaminant, the foundry environment is complex and several other exposures have been documented. For example, polycyclic aromatic hydrocarbons may originate from the thermal decomposition of organic material (such as coal-tar pitch, coal, mineral oils, synthetic resins, vegetable matter) present as additives or binders in sands. Other exposures include various metal fumes and dusts depending on type of metal or alloy produced, carbon monoxide, sulfur dioxide, nitrogen oxides, formaldehyde, amines, phenols, furfuryl alcohol and aliphatic and aromatic hydrocarbons (e.g. benzene) (IARC, 1984; Palmer & Scott, 1986; Burgess, 1995).

(e) Ceramics, cement and glass industries

In the manufacture of structural clay products (bricks, pipes, tiles), exposure to crystalline silica depends mainly on the quartz content of the clay or shale that is the principal raw material. Refractory bricks are made with minerals of very high quartz content. In the case of pottery and sanitary ware, flint (100% quartz) is added to clay as a raw material going into the manufacture of the slip. Sand, which may be used as dusting powder, may also contribute to airborne silica in the ceramics industry, as well as the decorative material (glaze) that may be added to the surface (Weill et al., 1994; Burgess, 1995).

Respirable quartz exposure levels measured for various jobs in the ceramics industry of various countries are summarized in Table 15. Mixing, moulding, glaze spraying and finishing jobs have been associated with the higher exposure levels, often in the range of 0.1–03 mg/m³. Successful reduction of exposure levels has been accomplished by simple control measures such as enclosure, use of moisture or water mist, use of non-siliceous dusting compounds, better housekeeping and ventilation (Buringh et al., 1990; Health and Safety Executive, 1992c; Cooper et al., 1993; Burgess, 1995). It has been estimated that silica dust exposure 20–30 years ago in Italian ceramics factories was three- to five-fold higher than in the early 1990s (Cavariani et al., 1995). Cristobalite may be released during repair of refractory materials used in the fabric of kilns (Health and Safety Executive, 1992c).

Table 15

Occupational exposure to crystalline silica in the ceramics industry in various countries.

Even though crystalline silica constitutes the main health hazard in the ceramics industry, other exposures may be found in certain operations. For example, talc is sometimes used in the body of clay products and as a parting compound in sanitary ware manufacture; various metal compounds, such as chromates and lead compounds, are used as pigments in glazes (Thomas et al., 1986; Burgess, 1995).

In the cement industry, crystalline silica exposure may occur during the handling of raw materials that may contain some quartz, such as clay and volcanic tuff, as well as the sand dust may be added in the process. However, once manufactured, normal Portland cement contains little crystalline silica (Prodan, 1983). In a Swedish plant, the quartz content of dust was generally < 5% and respirable quartz concentrations in areas where raw materials were handled was generally less than 0.1 mg/m³ Substantially lower concentrations are reported for workers handling clinker and finished cement (Jakobsson et al., 1993).

In a survey of 17 Italian cement factories, median respirable dust concentrations varied from 0.9 to 7 mg/m³ depending on sites, but most samples contained < 1% crystalline silica (Pozzoli et al., 1979).

Sand is a major raw material in the manufacture of glass (IARC, 1993), including fibreglass. When washed sand is used, airborne dust from the mixed batch commonly contains only 1–5% crystalline silica. In the manufacture of fibreglass, the silica is added to the batch as a finely divided powdered sand of 98.5% or higher silica content (Powell, 1982). In the glass industry in general, the manual unloading of dry sand and the use of crushed quartz are considered to be hazardous procedures. Hazards associated with hand filling of pots in the pot process, more common in the past, have been eliminated in the more modern tank process. Refractory blocks and bricks used in the construction of furnaces and tanks contain crystalline silica including cristobalite and tridymite and exposure may occur during their cutting, sawing and chipping to size (Cameron & Hill, 1983).

Respirable quartz and cristobalite have been measured in the range of 0.004– 0.71 mg/m³ and 0.1–0.25 mg/m³, respectively, in United States man-made mineral fibre plants (Manville, CertainTeed and Owens-Corning Fiberglass companies, 1962–87).

In seven European ceramic fibre plants, respirable crystalline silica was detected in eight of 17 groups where samples were collected. In general the levels were low — individual measurements ranged from 0.01 to 0.25 mg/m³ Cristobalite was found in a single sample collected from a bricklayer dismantling de-vitrified ceramic fibre insulation (Cherrie et al., 1989). Exposures to man-made mineral fibres in the glass manufacturing industry have been covered previously in the IARC Monographs series (IARC, 1988, 1993).

Amorphous silica

Even though it may be present in a variety of work environments, exposure to amorphous silica has been the object of only a few quantitative published reports. This can be explained in good part by the fact that most varieties of amorphous silica have been considered to be of low toxicity compared to other occupational contaminants such as crystalline silica. Also, amorphous silicas have often not been reported specifically, being part of ‘nuisance dusts’ measured by non-specific gravimetric methods. Dust levels reported in a few studies, including a large compilation of data from the synthetic amorphous silica industry, can be found in Table 16.

Table 16

Occupational exposure to different types of amorphous silica.

(a) Air

Quartz is a major mineral component of desert dust, which consists of fine particles smaller in size than 10 µm that can be transported by winds over thousands of kilometres and brought down by rainfall onto water or land surfaces (Klein et al., 1993). Exposure to quartz from dust storms has been suggested as a cause of non-occupational pneumoconioses reported in certain regions of the world (Weill et al., 1994). In the western Himalayas, 80% of the dust collected during dust storms was respirable and its silica content ranged between 60 and 70% (Saiyed et al., 1991). Levels of exposure to quartz attained during dust storms have not been documented. Dust samples collected in the windy season in two communes in a sandy area of Gansu Province in China ranged from 8.35 to 22 mg/m³. Deposited dust in these places consisted mainly in fine particles (< 5 µm) and had a free silica content of 15–26% (Xu et al., 1996).

Crystalline silica has been reported as a possible important constituent of volcanic ash collected at high altitude or as settled dust at ground level. Cristobalite and keatite are reported to constitute 35% of El Chichón (Mexico) ash collected at 34–36 km altitude (Klein et al., 1993); crystalline silica has been identified as present at levels of 3–7% in Mount St Helens (Washington State, United States) settled ash samples (Dollberg et al., 1986).

There is no extensive data set on levels of silica in ambient air. Ambient levels of quartz, based on inhalable particulate measurements taken in 1980 in 22 United States cities have been reported. Fine quartz levels (particles < 2.5 µm aerodynamic diameter) were from 0 to 1.9 µg/m³, while coarse levels (from 2.5 to 15 µm) went from 1.0 to 8.0 µg/m³. Quartz represented on average 4.9% of the coarse particle mass and 0.4% of the fine particle mass (Davis et al., 1984). It has been estimated that crystalline silica concentrations in the range of 1–10 µg/m³ are common in urban and rural settings (Hardy & Weill, 1995).

No data are available for ambient levels of amorphous silica, except for some measurements of silica fibres taken in the vicinity of farming operations. Thus, amorphous silica fibres were identified as smoke constituents in three of seven area samples located near burning sugar cane fields in Hawaii (Boeniger et al., 1991). Amorphous silica fibres were observed at 0.02 fibres/mL in one of 11 samples collected upwind of rice farming operations in California, in one of two 1.5-km downwind samples and in two of four field-edge downwind samples; a mean level of 0.004 fibres/mL was detected for all downwind samples. For community samples collected in neighbouring towns on days when there was rice burning, fibres were detected in four of 14 samples; the mean level for all samples was < 0.004 fibres/mL (Lawson et al., 1995).

Non-occupational inhalation of crystalline silica may also occur during the use of a variety of consumer or hobby products, such as cleansers, cosmetics, art clays and glazes, pet litter, talcum powder, caulk and putty, paint, mortar and cement (United States Bureau of Mines, 1992). In a study on the possible contamination of homes with crystalline silica on work clothing, no difference was found between the levels of cristobalite in outside ambient air and in the laundry areas of three homes investigated (Versen & Bunn, 1989).

(b) Water

Silica may be present in water as quartz particles and diatom fragments. No quantitative data on levels of quartz or other silica forms in potable or other forms of water were available to the Working Group. Silica dissolves to a small extent in water as monomeric silicic acid. Levels range from 1 ppm to almost 100 ppm (mg/L) depending on the climate, the petrographic nature of the aquifer, the depth and the activity of various biological processes (Siever, 1978).

(c) Food

Amorphous silica (such as fumed silica) is incorporated in a variety of food products as anti-caking agent at levels up to 2% by weight (such foods include beverage mixes, salad dressings, sauces, gravy mixes, seasoning mixes, soups, spices, snack foods, sugar substitutes, desserts). Amorphous silica is also used as an anti-caking agent and as an excipient in pharmaceuticals for various drug and vitamin preparations. Other possible uses include the following: retention of volatiles, microencapsulation, dispersion agent, clarification of beverages, viscosity control, anti-foaming agent and dough modifier (Villota & Hawkes, 1985).

Gold ore miners

McDonald et al. (1978) conducted a study of a cohort of 1321 miners from one gold mine in South Dakota (United States) who had at least 21 years' employment at the mine. SMRs based on South Dakota mortality rates showed excess mortality from all causes (631 observed; SMR, 1.15 [95% CI, 1.06–1.24]); there were 37 cases of pneumoconiosis (none expected) and 39 of tuberculosis (3.6 expected). There was no overall excess of respiratory cancer (17 observed; SMR, 1.03 [95% CI, 0.60–1.65]), although, in the first half of the follow-up period (1937–55), six deaths from lung cancer were observed against 3.4 expected. Using dust exposure data from company midget impinger samples and the estimated average silica content of 39%, the authors examined the mortality risks in five categories of dustiness; they showed clear linear relationships for tuberculosis and pneumoconiosis (McDonald & Oakes, 1984). Using five categories of dustiness, no correlation with respiratory cancer was found (McDonald et al., 1978).

In a cohort previously followed by Brown et al. (1986), Steenland and Brown (1995) followed up 3328 white male United States gold miners from South Dakota who worked underground for at least one year between 1940 and 1965. The follow-up was through to 1990. Primary exposures were to (non-asbestiform) amphibole minerals in the cummingtonite-grunerite series and to silica. The silica content of respirable dust in the mid-1970s was estimated to be 13%. The median respirable silica decreased from 0.15 mg/m³ in 1930 to 0.05 mg/m³ after 1950. Exposure to arsenic and radon were below United States Occupational Safety and Health Administration standards (radon daughters, 0–0.17 WL). Yearly measurements from 1937 to 1975 were used to calculate exposure levels for five job categories and to calculate cumulative dust (dust-days). Smoking data (never/occasional, current and ex-smoker) were volunteered by 602 of the men in a 1960 Public Health Service Silicosis Survey. Compatible age- and race-specific data on smoking from a 1955 survey of a sample of the United States population were used to estimate the effect of smoking differences on SMR for lung cancer. Of the cohort, 2% was lost to follow-up. Mortality from all causes was elevated (1551 observed; SMR, 1.13; 95% CI, 1.07–1.19). The SMR for all cancers was not elevated (303 observed; SMR, 1.01; 95% CI, 0.90–1.13). None of the cancer sites had a greatly elevated SMR. The SMR for lung cancer was 1.13 (115 observed; 95% CI, 0.94–1.36) when the United States population was used for comparison. However, the SMR for lung cancer was elevated for person-years for those workers whose first exposure (first job underground) was more than 30 years before (90 observed; SMR, 1.27; 95% CI, 1.02– 1.55). The SMR for lung cancer was also elevated in the highest exposure category (28 observed; SMR, 1.31 [95% CI, 0.87–1.89]), but the trend with duration of exposure was inconsistent: SMRs, 1.02 [95% CI, 0.79–1.30], 1.55 [95% CI, 1.08–2.16] and 1.01 [95% CI, 0.56–1.67] for < 10 years, 10–20 years and ≥ 20 years of exposure, based on 65, 35, 15 observed deaths, respectively. The SMR for lung cancer was increased mainly in men hired before 1930 (21 observed; SMR, 1.30 [95% CI, 0.80–1.99]). However, the smoking-adjusted SMR using the United States population was 1.07 [95% CI, 0.88– 1.28]). Mortality was increased for non-malignant respiratory disease (170 observed; SMR, 1.86; 95% CI, 1.58–2.16), asthma (7 observed; SMR, 2.61; 95% CI, 1.09–5.61) and pneumoconiosis and other respiratory diseases (92 observed; SMR, 2.61; 95% CI, 2.11–3.20). Mortality was also increased for non-Hodgkin's lymphoma (13 observed; SMR, 1.63; 95% CI, 0.86–2.78).

Hnizdo and Sluis-Cremer (1991) followed up a cohort of 2209 white South African gold miners whose exposure started during 1936–43 and who were studied for respiratory disorders during 1968–71, when 45–54 years of age. The mortality follow-up was through 1986. Vital status was established from the Gold Miners' Provident Fund records, medical files and the Department of Interior; miners not reported dead were assumed to be alive. The cause of death was established independently by two medical doctors from the best available evidence (death certificates, medical files and autopsy reports available on 84% of dead miners). The average level of respirable dust in the gold mine in 1968 was 0.3 mg/m³ of which approximately 30% was crystalline silica. Uranium was mined in some gold mines as a main product or as a by-product. Levels of radon daughters ranged from 0.1 to 3.0 working levels (WL) in most deep mines (average, 0.4 WL); in a few shallow mines, up to 6 WL was measured. Cumulative dust exposure was evaluated in terms of respirable surface area (RSA)-years (Beadle & Bradley, 1970) and the duration of dust exposure were calculated from personal records of dusty shifts. Smoking history was obtained in 1968–71 and pack-years were calculated. There were 77 cases of primary lung cancer. The estimated excess risk of lung cancer for every 1000 RSA-years, standardized for smoking, year of birth and age (estimated from the proportionate hazards model), was 2.3% (95% CI, 0.5%–4.2%). For miners in the highest exposure category (≥ 41 000 RSA-years), the estimated relative risk of lung cancer was 2.92 (95% CI, 1.02–8.4). No association between lung cancer and silicosis of the parenchyma or pleura at autopsy was found, but a significant association with hilar gland fibrosis was observed (adjusted odds ratio, 3.9; 95% CI, 1.2–12.7). [The Working Group noted that arsenic is not known to be present in the dust of South African gold mines but that radon was a potential confounding factor.]

In an extended follow-up of a study reported by Wyndham et al. (1986), Reid and Sluis-Cremer (1996) followed up a cohort of 4925 white South African gold miners. These miners were born in 1916–30, were working in gold mines in the vicinity of Johannesburg on 1 January 1970 and were then aged 39–54 years. The follow-up was through 1989. Daily cigarette consumption was obtained from medical files. Exposure to mining was measured as duration of dusty exposure obtained from a record of dusty shifts, and as cumulative dust exposure (duration weighted by an average dust level for an occupational category measured in the late 1960s, in years-mg/m³). Vital status was established for 4875 miners. The age- and year-specific mortality rates for white South African men were applied to calculate standardized mortality ratios (SMRs). The SMR was increased for all deaths (2032 observed; SMR, 1.30; 95% CI, 1.24–1.35). There was no increased risk for all cancers (341 observed; SMR, 1.10; 95% CI, 0.99–1.23), but the SMR for lung cancer was increased (143 observed; SMR, 1.40; 95% CI, 1.18–1.65). SMRs were also increased for pulmonary tuberculosis (20 observed; SMR, 2.95; 95% CI, 1.81–4.58), pneumonia (68 observed; SMR, 1.46; 95% CI, 1.13–1.85), pneumoconiosis (16 observed; SMR, 21.3; 95% CI, 12.2–34.7) and chronic obstructive pulmonary disease (i.e. emphysema, bronchitis, asthma) (176 observed; SMR, 1.89; 95% CI, 1.62–2.19). The relative risk for lung cancer and cumulative dust exposure for five years before death, adjusted for smoking, estimated from a nested case–control study was 1.12 (95% CI, 0.97–1.3) (mg/m³)-years and that for chronic obstructive pulmonary disease and cumulative dust exposure was 1.20 (95% CI, 1.0–1.4) (mg/m³)-years. [The Working Group noted the possible overlap with the study of Hnizdo and Sluis-Cremer (1991). The Working Group considered that radon is a potential confounding factor for this study.]

Kusiak et al. (1991) followed up a cohort of 13 603 male non-uranium gold miners who worked in Ontario, Canada. This cohort consisted of all who had been examined in chest clinics in Ontario in 1955 or later and who had been employed for at least two weeks in dusty jobs in Ontario mines after 1954 and for at least 60 months in dusty jobs in the mining industry anywhere. Deaths that occurred between 1955 and 1986 were identified from a national mortality database (about 6% of deaths up to 1977 were missing). Miners who reported that they had worked in asbestos mines were excluded. Before 1950, dust concentrations were often above 1000 particles/mL; by 1959 they had dropped to 400 particles/mL by 1959 and by 1967 to 200 particles/mL. The percentage of silica in respirable dust measured in 1978 survey ranged from 4.3 to 11.8% in different mines. Arsenic was present in most gold mines and was also associated with gold specks. Measurements of radon decay products (all post 1961) ranged from 0.001 to 0.335 WL. Smoking data were obtained from a random sample of miners. Expected deaths were calculated from Ontario male death rates. The overall SMR for lung cancer in gold miners was increased (SMR, 1.29; 95% CI, 1.15–1.45). The SMR was increased among miners who started gold mining before 1946 and never mined nickel was 1.40 (95% CI, 1.22–1.59). No increase in lung cancer risk was observed in gold miners who started mining after 1945. The authors attributed the increased risk of lung cancer in Ontario gold miners to the duration of underground mining and the associated exposure to arsenic and radon decay products. In a nested case–control study, Kusiak et al. (1993) reported an increased mortality from stomach cancer in this cohort of gold miners (104 observed; SMR, 1.52; 95% CI, 1.25–1.85), which was attributed to exposure to chromium. [The Working Group noted the lack of cumulative exposure measurements to silica.]

Iron ore miners

Lawler et al. (1983) examined the mortality of 10 403 white male employees of a Minnesota (USA) haematite ore mining company (1937–1978) and contrasted it with that of US white males. Chemical analyses of the ore showed an average silica content of 8% in 1943 but 20–25% in the ore being mined in the late 1970s. For the total cohort (underground and above-ground miners), the SMR for all causes was 0.93 (4699 observed [95% CI, 0.90–0.96]). Mortality from tuberculosis (33 observed, SMR 0.45 [95% CI, 0.31–0.63]) and respiratory disease (234 observed, SMR 0.79 [95% CI, 0.69–0.90) was lower than expected; no elevated risk from these two causes of death was seen for underground miners. For stomach cancer, underground miners had an SMR of 1.67 (77 observed [95% CI, 1.32–2.09]) and above-ground miners had an SMR of 1.81 (49 observed [95% CI, 1.34–2.40]). For lung cancer, the SMR was 1.00 for underground miners (117 observed [95% CI, 0.82–1.19]) and 0.88 for above-ground miners (95 observed [95% CI, 0.71–1.07]). No data on smoking habits or exposure to radon daughters were obtained. [The Working Group noted that exposures in the Minnesota iron ore mines were complex and included fibrous amphiboles as well as silica.]

A group of 1173 iron miners in Lorraine (France) was observed for five years following clinical examinations and lung function tests (Pham et al., 1983). During this period, there were 40 deaths versus 39 expected on the basis of rates for the general male population of Lorraine. There were 13 deaths from lung cancer (SMR, 3.5; 95% CI, 1.9– 6.0). All the lung cancer cases were found among underground workers; they were all smokers and they had had a longer mean length of employment underground (23.6 years) than the whole underground group (16.7 years). Measured levels of radon daughters were approximately 0.03 WL in the mine and 0.07 WL in the return air. The prevalence of smoking was higher (66%) in the study population than in a general population sample (52%). [The Working Group noted that the methods for ascertaining causes of death for cases and controls were not comparable.]

Kinlen and Willows (1988) followed a cohort of 1947 iron ore miners in Cumbria (United Kingdom) from 1939 to 1982. Miners were compared with men of a similar social class from England and Wales. Mortality data were analysed by proportional mortality. Radon levels measured in the mines in 1969 in the closed area ranged from 0.4–3.2 WLs (median 2.0). There were 1604 deaths. The proportionate mortality ratios (PMR) were increased for tuberculosis (88 observed; PMR, 3.55; [95% CI, 2.85–4.37]) and non-malignant respiratory disease (292 observed; PMR, 1.62; [95% CI, 1.44–1.82]). There was an elevated increase for cancer of the stomach (49 observed; PMR, 1.24; [95% CI, 0.82–1.64]). There was no elevated increase for lung cancer (84 observed; PMR, 0.97; [95% CI, 0.77–1.20]). Risk of lung cancer was increased when population rates for the rural population were used (PMR, 1.59; [95% CI, 1.27–1.97]).

Chen et al. (1990) followed a cohort of 6444 men employed on 1 January 1970 in two iron ore mines in Longyan and Taochong in China through to 1982. Vital status was ascertained in 8534 of 8641 (99%) miners; 2090 miners with exposure for less than one year were excluded. Occupational history and smoking habits were assessed retrospectively by a questionnaire. Job titles were used to assign exposure level. Mechanical ventilation was introduced in 1955 in the Longyan mine and in 1963 in the Taochong mine, reducing total dust from several hundred mg/m³ to 3.8 mg/m³ (23–28% of settled dust was iron). Traces of 3,4-benzo[a]pyrene, titanium, arsenic, chromium, nickel, cobalt, cadmium and beryllium were found in the dust. Levels of radon daughters found in 1984 at the working face were higher (0.2 WL) than at other workplaces (0.1 WL), With improvements in ventilation, there were parallel reductions in the radon and dust concentrations over the years. The expected deaths were based on sex- and age-specific death rates for China for the years 1973–75. Diagnosis of silicosis was obtained from routine X-ray examinations carried out on a periodical basis and read by a panel. There were 550 deaths. The SMR for total cancer mortality (98 cancers) was 1.1 (95% CI, 0.9–1.3). The SMR for lung cancer was increased (29 observed; SMR, 3.7; 95% CI, 2.5–5.3) and was higher in those who worked prior to usage of mechanical ventilation (20 observed; SMR, 4.8; 95% CI, 2.9–7.4) than in those who started after it was introduced (9 observed; SMR, 2.4; 95% CI, 1.1–4.6). There was an increasing trend with low, medium and heavy exposure (SMRs, 2.6, 2.6 and 4.2, respectively), mainly in smokers. [The Working Group noted that the exposure-response analysis was based on cumulative exposure estimates generated from single job titles, which may not reflect complete job history.] A high proportion of deaths (41%) was due to non-malignant respiratory disease (227 observed). A total of 1226 silicotics, diagnosed at routine periodic examinations, had an SMR for lung cancer of 5.3 (14 deaths; 95% CI, 2.9–8.8) and for non-silicotics the SMR was 2.9 (15 deaths; 95% CI, 1.6–4.7). In current smokers, subjects with silicosis (n = 962) were at higher risk for lung cancer (13 observed; SMR, 6.7; 95% CI, 3.6–11.5) than subjects without silicosis (n = 3123) (12 observed; SMR, 3.0; 95% CI, 1.6–5.3). Subjects with silicotuberculosis (n = 389) were also at higher risk for lung cancer (7 observed; SMR, 9.3; 95% CI, 3.8–19.2).

Other ore miners

Hodgson and Jones (1990) followed up a cohort of 3010 miners who had at least 12 months' mining experience between 1941 and 1984 in two tin ore mines in Cornwall, United Kingdom. The follow-up was through 1986. SMRs were calculated using the national age- and year-specific death rates. Radon daughter levels had been monitored since 1967 and the average exposure was estimated as 8–12 working level months (WLM)/year in mine A and 9–19 WLM/year in mine B. Arsenic was also mined in mine A. The SMR for all causes of death was increased (851 observed; SMR, 1.27; [95% CI, 1.18–1.35]) and that for lung cancer was significantly increased (105 observed; SMR, 1.58; [95% CI, 1.29–1.91]). There was a strong dose-response trend with duration of underground exposure; the SMRs increased as follows: 0.91 [95% CI, 0.51–1.50] for 1–5 years; 1.72 [95% CI, 0.94–2.88] for 5–10 years; 1.76 [95% CI, 1.09–2.7] for 10–20 years; 3.55 [95% CI, 2.07–5.69] for 20–30 years; and 4.47 ([95% CI, 2.50–7.37]) for more than 30 years. There were 49 deaths from silicosis and 33 deaths from silicotuberculosis. Smoking and radon daughters were considered to be the main risk factors for lung cancer. [The Working Group noted the high exposure to radon.]

Ahlman et al. (1991) followed up a cohort of 597 miners employed between 1954 and 1973 for at least three years either in a copper ore mine (n = 398) or zinc ore mine (n = 199) in eastern Finland. The follow-up was through to 1986 (person-years, 14 782). Vital status was obtained via the Population Data Register. Regional age-specific data rates were used for comparison. Occupational histories and smoking data were obtained through a questionnaire. In the copper mine, mean respirable silica dust concentrations decreased from 0.16 to 0.08 mg/m³ over the years and average radon daughter levels decreased from 1.7 to 0.7 WL. In the zinc mine, the highest concentration measured was 11 WL and the mean concentration was 0.4 WL. Diesel-powered machines were introduced in the 1960s. Overall mortality was increased [SMR, 1.04; 95% CI, 0.85–1.27]; 102 observed; 97.8 expected based on regional rates). Mortality from lung cancer was increased (10 observed, 4.3 expected; 6.9 expected based on regional rates). Five of the lung cancer deaths (SMR, 2.94; [95% CI, 0.96–6.86]) came from the zinc mine (1.7 expected). [The Working Group noted the high exposure to radon.]

A total of 9912 (369 silicotics and 9543 non-silicotics) white male metal ore miners in the United States who volunteered for a standard medical examination during 1959–61 were followed up for lung cancer mortality through 1975 (Amandus & Costello, 1991). The ores that were mined consisted of copper, lead-zinc, iron, mercury, lead silver, gold and gold-silver, tungsten and molybdenum. Miners who were employed in non-uranium mines and had not been exposed to diesel exhausts were studied. Silicosis was diagnosed from radiograms taken at the examination according to the ILO 1959 classification (1, 2, 3 small rounded opacities and large opacities). Lung cancer was increased in silicotics (14 deaths; SMR, 1.73; 95% CI, 0.94–2.90) in comparison with non-silicotics (118 deaths; SMR, 1.18; 95% CI, 0.98–1.42). Age- and smoking-adjusted lung cancer risk in silicotics was 1.96 (95% CI, 1.19–3.23) times that in non-silicotics. In those who had smoked cigarettes for over 25 years, SMRs were 2.69 in silicotics (8 deaths; 95% CI, 1.16–5.30) and 1.76 in non-silicotics (64 deaths; 95% CI, 1.36–2.26). The SMR for lung cancer was increased mainly in silicotics in lead-zinc mines (4 observed; SMR, 2.42; 95% CI, 0.66–6.21) and in mercury mines (3 observed; SMR, 14.03; 95% CI, 2.89–40.99). SMRs were significantly increased in non-silicotics who had worked for over 20 years in an underground metal mine (SMR, 1.52; 95% CI, 1.10–2.03); and who had been employed at a mercury mine (SMR, 2.66; 95% CI, 1.15–5.24). After excluding mercury miners, the SMR for lung cancer was 1.39 in silicotics and 1.14 in non-silicotics. Among those who had worked in mines with low radon exposure, age- and smoking-adjusted lung cancer risk ratio between silicotics and non-silicotics was 2.59 (95% CI, 1.44–4.68). [The Working Group noted that this is a cohort of volunteers based on a medical survey of 50 underground mines. Participation rate was not discussed nor was percentage follow-up defined.]

Chen et al. (1992) identified a cohort of 70 179 workers employed from 1972 through 1974 for at least one year in one of four industrial groups: (i) 10 tungsten ore mines, (ii) six copper-iron ore mines, (iii) four tin ore mines and (iv) eight pottery factories and one clay mine all in south central China. Mortality follow-up was through 1989. Silica dust exposure was estimated by merging individual job titles and time of exposure against job-time-specific measurements of total dust and percentage of free silica collected, mostly on a monthly basis for most dust-exposed jobs. Subjects were classified into four exposure levels according to the job title with the highest dust level in which the subject worked for at least one year. The average annual total dust levels were 6.1 mg/m³ (range, 2.0–26.3 mg/m³) for tungsten ore mines, 5.6 mg/m³ (range, 3.8–16.1 mg/m³) for copper and iron ore mines, and 7.7 mg/m³ (range, 3.4–29.7 mg/m³) for tin ore mines. The lower ranges represent more recent levels. Confounding factors studied were arsenic and polycyclic aromatic hydrocarbons. Vital status and cause of death were obtained from employment registers, accident records, medical records and personal contact. Cause of death was coded according to the Chinese coding system. For subjects who died of primary lung cancer, medical reports and X-rays were sought. Silica-exposed workers had yearly radiograms and cases of silicosis (Chinese categories: suspected, 1, 2 or 3) were reported to factory registries. [The Working Group noted that this system is very close to the ILO system.] Silicotics had more frequent medical examinations. The expected deaths for selected causes were based on age- and sex-specific rates computed as the average rates obtained from national mortality surveys carried out during 1973–75 and in 1987. Vital status was identified for 68 241 (97.2%) of the subjects (28 442 in tungsten mines, 18 231 in copper–iron mines, 7849 in tin mines). Mortality from all causes was slightly increased (6192 observed; SMR, 1.06; 95% CI, 1.04–1.09). Mortality from all cancers was decreased (1572 observed; SMR, 0.86; 95% CI, 0.81–0.90). However, increased mortality was found for cancer of the nasopharynx (78 observed; SMR, 1.54; 95% CI, 1.22–1.93), due to a significant increase in tungsten ore and tin ore mines, and for liver cancer (474 observed; SMR, 1.15; 95% CI, 1.05–1.26), due to a significant increase in copper-iron ore and tin ore mines. Cancer sites with significantly decreased mortality were the oesophagus, stomach, colorectum and lung. The overall SMR for lung cancer was decreased (330 observed; SMR, 0.79; 95% CI, 0.71–0.88), although it was increased in tin ore miners (SMR, 1.98; 95% CI excludes 1.0) and in silicotic workers (SMR, 1.22; 95% CI, 0.9–1.6, compared to non-silicotics). Other causes of death with increased mortality were other respiratory diseases (925 observed; SMR, 1.48; 95% CI, 1.39–1.58), due to an increase in tungsten ore miners and pottery workers, and pulmonary heart disease (695 observed; SMR, 5.81; 95% CI, 5.38–6.26). Mortality from pulmonary tuberculosis was decreased in all groups (overall 312 observed; SMR, 0.77; 95% CI, 0.69–0.86) and that from pneumoconiosis was increased in all groups (overall 199 observed; SMR, 36.25 [95% CI, 31.4–41.7]). Relative risks, adjusted for decade of birth, sex, factory type and age, that showed a statistically significant trend with low, medium and high dust exposure levels (p < 0.01) were for respiratory disease (relative risks: low, 1.0; medium, 2.39 (95% CI, 1.9–3.0); and high, 3.65 (95% CI, 3.0–4.5), pneumoconiosis (relative risks, 1.0; 7.29 (95% CI, 4.5–11.8); and 13.57 (95% CI, 8.9–21.0), respectively) and pulmonary heart disease (relative risks, 1.0; 1.27 (95% CI, 1.0–1.6); and 1.93 (95% CI, 1.6–2.4). For lung cancer the respective relative risks were 1.0, 1.38 (95% CI, 1.0–1.9) and 1.10 (95% CI, 0.9–1.4).

McLaughlin et al. (1992) conducted a nested case–control study of this same cohort. Using 316 male lung cancer cases and 1352 controls, these investigators found an increasing trend in the age- and smoking-adjusted odds ratios for lung cancer with cumulative dust (p = 0.02) and cumulative respirable silica (p = 0.004) in tin ore miners only: the odds ratio in the highest level of cumulative silica dust was 3.10. A trend with increasing arsenic levels (p = 0.0004) was also observed in tin miners. Exposure to arsenic and to polycyclic aromatic hydrocarbons was highly correlated with exposure to silica dust. Subjects with silicosis had an increased risk of lung cancer in iron–copper ore miners (15 observed; odds ratio, 3.1) and in tin ore miners (37 observed; odds ratio, 2.0) but not in tungsten ore miners where a significant decreasing trend for lung cancer and respirable dust and respirable silica dust was observed (p < 0.05). [The Working Group noted that exposure to arsenic confounded the potential dose-response relationship between silica exposure and lung cancer risk in tin miners.]

In an update of previous studies (e.g. Higgins et al., 1983), Cooper et al. (1992) followed a cohort of 3431 men who had worked prior to 1959 for at least three months in ‘taconite’ surface mines and the mill in a Minnesota (United States) iron ore mine through 1988. A total of 1058 subjects were found to be dead through employment records, Social Security Administration records (contributing to or receiving pension), from the National Death Index and from previous searchers. Those not found were assumed to be alive. Death certificates were obtained from state offices of vital statistics in the State of residence or death. Death certificates were obtained for 1039 (98.2%) subjects known to be dead. The United States white male population was used as a reference. Up to 28–40% of free silica in air samples was reported and subjects were exposed to elongated dust particle fragments of non-asbestiform amphibole minerals. There was no underground mining. SMRs were significantly decreased for all causes (1058 observed; SMR, 0.83; 95% CI, 0.78–0.88), all cancers (232 observed; SMR, 0.87; 95% CI 0.76–0.99) and all respiratory system cancers (65 observed; SMR, 0.67; 95% CI, 0.52–0.85) and lung cancers (62 observed; SMR, 0.67; 95% CI, 0.52–0.86). The SMR for respiratory cancers displayed a significant negative trend with duration of employment. The SMR for non-malignant respiratory disease was significantly decreased (55 observed; SMR, 0.71; 95% CI, 0.54–0.93). The use of Minnesota death rates increased the above SMRs, but not above 1.00. [The Working Group noted that the absence of an increase in mortality from non-malignant respiratory disease in this cohort suggests low worker exposure to free silica. In the first study of this cohort (Higgins et al., 1983), the investigators also noted relatively low silica exposures in this study population.]

Cocco et al. (1994a) and Carta et al. (1994) followed up a cohort of 4740 male workers who had at least one year of employment between 1932 and 1971 and were working during 1960–71 in a lead ore (A) and a zinc ore (B) mine in Sardinia. The mortality follow-up was through 1988. Vital status was ascertained for 99.5% of the cohort. SMRs were based on the regional five-year age- and year-specific death rates. The average respirable dust concentrations in underground workplaces in the two mines were similar — 2.5–2.6 mg/m³ in 1962–70 and they decreased to 1.6–1.8 mg/m³ in 1981–88 (with median quartz concentration of 1.2% and 12.8%, respectively) in mines A and B. Surface workers were exposed to less than 1 mg/m³ in both mines from the 1970s. The mean exposure to radon daughters was higher in mine A (0.13 WL) than in mine B (0.011 WL) among underground miners. Smoking habits were comparable between the two mines. Of the cohort, 2096 worked in mine A and 2603 in mine B, and 41 in both. In underground workers from mine A, the SMR for all causes of death was not increased (325 observed; SMR, 1.03; 95% CI, 0.92–1.14) nor was that for all cancers (84 observed; SMR, 0.99; 95% CI, 0.80–1.23) nor that for lung cancer (28 observed; SMR, 1.15; 95% CI, 0.77–1.67); the SMRs for cancer of the peritoneum and retroperitoneum (4 observed; SMR, 9.17; 95% CI, 2.50–23.47) and for respiratory diseases (68 observed; SMR, 2.46; 95% CI, 1.91–3.12) were increased. In underground miners from mine B, the SMR for all causes of death was increased (472 observed; SMR, 1.20; 95% CI, 1.09–1.31). Mortality from all cancers (101 observed; SMR, 0.92; 95% CI, 0.76–1.12) and lung cancer (26 observed; SMR, 0.79; 95% CI, 0.52–1.16) were not increased. Increases were SMRs for infectious and parasitic diseases (29 observed; SMR, 4.16; 95% CI, 2.79–5.97), pulmonary tuberculosis (29 observed; SMR, 7.06; 95% CI, 4.73–10.14) and respiratory disease (156 observed; SMR, 5.18; 95% CI, 4.40–6.06). Death from silicosis was included under non-malignant respiratory disease. Surface workers from both mines had a similar pattern of SMRs. SMRs for lung cancer did not show a systematic increasing trend with increasing duration of underground employment in any of the mines.

A cohort of 310 women employed in surface jobs (belt pickers) in the two mines (reported above) and 173 women not exposed to silica were also studied for lung cancer risk (Cocco et al., 1994b). There were 163 deaths in the total cohort and the risk of lung cancer was elevated in the exposed (5 cases; SMR, 2.83; 95% CI, 0.91–6.60) and in the unexposed women (1 case; SMR, 1.22; 95% CI, 0.02–6.78). [The Working Group noted the small number of cancer cases.]

Ceramics

Thomas (1982) examined the mortality of members of the United States Potters and Allied Workers Union for 1955–77. In men, there were elevated PMRs for tuberculosis (62 observed; PMR, 3.39; [95% CI, 1.36–1.74]), non-malignant respiratory disease (frequently noted as silicosis; 268 observed; PMR, 1.54; [95% CI, 1.36–1.74]) and lung cancer (178 observed; PMR, 1.21; [95% CI, 1.04–1.40]). The lung cancer excess appeared to be localized among workers in the sanitary-ware divisions (62 observed; PMR, 1.80; [95% CI, 1.38–2.31]). Silica exposure was said to be similar in sanitary-ware divisions and in other parts of the plants but to be characterized by the use of talc to dust moulds. [The Working Group noted the possibility that the talc was contaminated with asbestos.]

A cohort mortality study reported by Thomas and Stewart (1987) and Thomas (1990) was based on 2055 white men employed for one year or more in three plants manufacturing ceramic sanitary ware between 1939–1966 and followed up until 1 January 1981. Exposures were predominately to quartz but in some processes also to fibrous (tremolitic) talc until 1976 and non-fibrous (non-asbestiform) talc. Against United States rates for white males, the number of deaths from all causes was significantly fewer than expected (578 deaths; SMR, 0.90; [95% CI, 0.83–0.98]). There was an excess of lung cancer deaths (52 observed; SMR, 1.43; [95% CI, 1.07–1.88]) but a deficit of deaths from digestive cancer (19 observed; SMR, 0.52; [95% CI, 0.31–0.81]). Mortality from non-malignant respiratory disease was also increased (64 observed; SMR, 1.73; [95% CI, 1.33–2.21]). The lung cancer mortality risk increased with number of years of exposure to non-fibrous talc but was unrelated to years of exposure to silica. Information was not available on smoking. [The Working Group noted that the degree of overlap between these studies was not clear.]

A cohort of 1784 male Dutch ceramic workers was constructed based on a nationwide cross-sectional silicosis survey between 1972 and 1982. Follow-up took place between time of medical examination and 31 December 1991 (Meijers et al., 1996). Only those persons with a total working history of more than two years in the ceramics industry were selected for analysis. No usable quantitative exposure measurements were available, but each worker was classified as having low, medium or high silica exposure according to job description. Cause-, age- and calendar time-specific death rates of the total male Dutch population were applied to calculate expected numbers of deaths and SMRs. Overall lung cancer mortality risk was lower than expected (30 observed; SMR, 0.88 [95% CI, 0.59–1.26]). For silica exposure, there was no exposure–,response relationship with respect to cumulative dust exposure (low: 9 observed; SMR, 0.82; [95% CI, 0.37–1.55]; medium: 10 observed; SMR, 0.75; [95% CI, 0.36–1.38]; high: 11 observed; SMR, 1.15; [95% CI, 0.57–2.05]). Stomach cancer was not evaluated in this study.

Pottery

In a large cohort mortality study from southern central China, described in detail in Section 2.1.2 (Chen et al., 1992), 13 719 pottery workers were included with average annual dust exposure of 11.4 mg/m³ (9.4–23.8 mg/m³). The SMRs among these pottery workers were 1.44 (p < 0.05) for respiratory disease and 0.58 (p < 0.05) for lung cancer. In a nested analysis of 316 male lung cancer cases and 1352 controls (62 cases and 238 controls in pottery workers) (McLaughlin et al., 1992), also described in Section 2.1.2, the odds ratios for lung cancer for pottery were 2.0, 1.7 and 1.5 for low, medium and high total dust exposure as compared to no exposure. The trend for cumulative respirable silica exposure was not significant. There was no association with silicosis. Smokers of more than 20 cigarettes a day were at greatly increased risk (OR, 7.4).

In the British pottery industry, a study of mortality in a cohort of 4093 men was made by Winter et al. (1990). The subjects had been included in a survey of respiratory disease in the pottery industry conducted in 1970–71. Difficulties were encountered in ensuring the full tracing of the cohort and the investigators decided to limit their study to men and women under 60 years of age in 1970–71 (n = 3669). Among these subjects, 390 deaths were observed by the end of 1985 against 363.4 expected from national rates (SMR, 1.07) and 394.7 against local rates (SMR, 0.99). The SMRs for the 60 deaths observed from lung cancer were 1.40 (95% CI, 1.07–1.80) for national rates and 1.32 (95% CI, 1.00–1.69) for local rates. Adjustments for recorded smoking habits made very little difference to these SMRs, but possible exposure to other hazardous dusts was not considered. There was some indication of a relation between risk and estimated cumulative exposure to respirable quartz. Mean respirable quartz concentrations obtained in the workplace in each pottery were used to form four cumulative exposure groups, which assumed that current exposure levels applied to the entire occupational history in the pottery. The smoking adjusted lung cancer SMRs for the four cumulative exposure groups were 1.08 [95% CI, 0.35–2.54] for 0–0.14 (mg/m³) × years, 0.99 [95% CI, 0.43–1.95] for 0.15–0.49 (mg/m³) × years, 1.62 (95% CI, 1.05–2.39) for 0.50–1.49 (mg/m³) × years and 1.51 [95% CI, 0.93–2.31] for 1.50 (mg/m³) x years or more. [The Working Group noted the investigators' concern about possible bias in the follow-up and by the fact that mortality results were linked to men under 60 years of age in 1970–71.]

A further investigation in the British pottery industry was based on a cohort of 7020 male pottery workers in Staffordshire, born 1916–45, a few of whom were possibly included in the cohort of Winter et al. (1990). This study had three phases: in the first, proportional mortality was analysed in the 1016 men who had died by 30 June 1992 (McDonald et al., 1995); in the second, SMRs were examined in a cohort reduced to 5115 after exclusion of men who had worked in foundries, asbestos or other dusts (Cherry et al., 1995); and, finally, risks were assessed in detail taking account of radiographic changes, exposure estimates and smoking habit (Burgess et al., 1997; Cherry et al., 1997; McDonald et al., 1997). In the first phase of the study, after exclusion of recorded asbestos exposure, the PMR for lung cancer was found to be 1.22 (112 deaths [95% CI, 1.01–1.47]) against national rates but 1.04 (112 deaths; [95% CI, 0.86–1.25]) against local rates. The PMR for lung cancer in those with pneumoconiosis on their death certificate (30) was 1.75 (7 deaths [95% CI, 0.7–3.6]). A nested case–control study of 75 lung cancer cases and 75 controls matched on date of birth and date of first exposure suggested that the risk of lung cancer was associated with smoking history and past asbestos exposure. A further analysis based on 47 case–control pairs, in which both cases and referents were smokers showed evidence that risk was related to the duration of silica exposure (≥ 10 years) in pottery work (odds ratio, 2.8; 90% CI, 1.1–7.5) (McDonald et al., 1995). In the second phase of the study, SMRs against national mortality rates for the period 1985 through June 1992 were as follows: lung cancer, 1.91 (68 deaths [95% CI, 1.48–2.42]); and non-malignant respiratory disease, 2.87 [95% CI, 2.17–3.72]). Against local rates, the corresponding SMRs were 1.28 [95% CI, 0.99–1.62] and 2.04 [95% CI, 1.55–2.65] (Cherry et al., 1995). In the third phase, the three following related analyses were reported (Burgess et al., 1997; Cherry et al., 1997; McDonald et al., 1997): a radiographic validation of the exposure matrix; findings from a nested case–control study of mortality in relation to exposure, smoking and radiological changes using conditional logistic regression; and detailed findings from a sub-cohort of 1083 men used in the radiographic validation. The case–control analysis was based on 52 cases employed for 10 or more years (and 3–4 times as many controls). These three sets of analyses, taken together, showed (i) that a relationship existed between cumulative exposure and small radiographic opacities, and that this relationship was dominated by the intensity of exposure, and (ii) that in both the full cohort and sub-cohort, lung cancer risk was dominated by smoking but in neither was it related to cumulative exposure. However, lung cancer risk was increased in workers whose average intensity of exposure was 200 µg/m³ or greater (odds ratio, 1.88; 90% CI, 1.06–3.34) and in workers whose maximum exposure was 400 µg/m³ or greater (odds ratio, 2.16; 90% CI, 1.11–4.18). The latter risk was limited to workers in firing and post-firing occupations. Eight per cent levels of cristobalite were recorded in dust samples from this industry but these were not specific to firing and post-firing operations. [The Working Group noted that this study was the only epidemiological examination of peak exposure effects in lung cancer risk. Whereas the findings do not support a relation with cumulative exposure, the possibility remains that high-intensity exposures (≥ 400 µg/m³) may increase risk.]

Refractory brick

A series of reports on refractory brick workers in Genoa (Puntoni et al., 1985, 1988) was updated by Merlo et al. (1991). In this latter study, a cohort of 1022 factory brick male workers for six months or more between 1 January 1954 and 31 December 1977 was followed through 1986. Geometric mean concentration of respirable dust ranged from 200–560 µg/m³; crystalline silica was 30–65%. Observed deaths were compared with mortality for the Italian male population and smoking habits recorded for 285 workers actively employed in 1984 were noted. The overall mortality based on 243 deaths was somewhat above expectation (SMR, 1.10; 95% CI, 0.97–1.25). An excess was more definite for lung cancer (28 observed; SMR, 1.51; 95% CI, 1.00–2.18), urinary bladder cancer (7 observed; SMR, 2.78; 95% CI, 1.12–5.71) and non-malignant respiratory diseases (40 observed, SMR, 2.41; 95% CI, 1.72–3.28). The excess mortality from lung cancer and other respiratory diseases was almost entirely due to the experience of men first employed before 1957. Mortality was stratified by both length of employment and by years since first employment. SMRs for workers with > 19 years since first employment and for the category ≤ 19 years since first employment and > 19 years tenure were: lung cancer, SMR, 1.75 (8 deaths; 95% CI, 0.75–3.46) and SMR, 2.01 (13 deaths; 95% CI, 1.07–3.44); respiratory disease, SMR, 1.58 (7 deaths; 95% CI, 0.64–3.25) and SMR, 3.89 (28 deaths; 95% CI, 2.59–5.63) and bladder cancer, SMR, 5.75 (4 deaths; 95% CI, 1.57–14.74) and SMR, 0.99 (1 death; 95% CI, 0.25–5.49), respectively. A comparison of the smoking habits of the 285 men employed in 1984 and those of the Italian male population showed no significant difference. [The Working Group noted that information was not available on levels of exposure to crystalline silica or on the degree of conversion from quartz to cristobalite.]

A separate analysis was conducted on male silicotics and non-silicotics among 231 workers from the same refractory brick plant employed on 1 January 1960 and followed for mortality through 1979 (Puntoni et al., 1988). Included were 136 silicotics, identified from compensation files. SMRs were calculated using age-specific Genova mortality rates during the follow-up period as the reference. The SMR for all causes was 1.63 (57 deaths; 95% CI, 1.23–2.11) in silicotics and 0.64 (16 deaths; 95% CI, 0.36–1.03) in non-silicotics. Significant non-cancer excesses in the silicotics were reported for cardiovascular (SMR, 1.73) and non-malignant respiratory (SMR, 5.00) diseases. The SMRs for all cancer was 1.42 (16 deaths; 95% CI, 0.81–2.30) in silicotics and 0.88 (7 deaths; 95% CI, 0.35–1.81) in non-silicotics. With six deaths, the SMR for lung cancer was 1.67 (95% CI, 0.61–3.64) in silicotics (two non-smokers) and, with five deaths, 2.08 (95% CI, 0.67–4.84) in non-silicotics (one non-smoker). Laryngeal cancer was in excess in silicotics (3 deaths; SMR, 6.82; 95% CI, 1.40–19.9), while no laryngeal cancer deaths occurred in non-silicotics.

A further cohort mortality study from China was made in 11 refractory brick plants (Dong et al., 1995). Entry to the study was restricted to 6266 men first employed before 1962, almost all between 1950 and 1959. By 1985, 871 (13.9%) had died and 263 (4.2%) were lost to follow-up. Almost all cohort members had been subject to periodic health examination and chest X-ray; the latter classifying in the Chinese system silicosis as follows: category I, 20%; category II, 7%; and category III, 3%. Smoking habits were also recorded. Standardized rate ratios (SRRs) were calculated by age and cause of death in comparison with a population of 11 470 male workers from 10 rolling steel mills. The overall SRR was 1.44 (871 deaths; 95% CI, 1,35 –1.54]); among non-silicotics, the SRR for all causes of death was 1.04 (390 deaths: [95% CI 0.94 –1.15]) and among silicotics (categories I. II III) the SRR was 2.10 (481 deaths; [95% CI, 1.92–2.30]). The corresponding SRRs for cardiorespiratory disease were 1.25 (255 deaths; [95% CI, 1.10–1.41]), 0.96 (111 deaths: [95% CI; 0.79–1.16]) and 1.65 (144 deaths: [95% CI, 1.40–1.94]) and for lung cancer 1.49 (65 deaths; [95% CI 1.15–1.90]), 1.11 (30 deaths; [95% CI, 0.75–1.58]) and 2.10 (35 deaths; [95% CI, 1.46–2,92]). In men with 20 or more years of exposure, the SRR for lung cancer increased significantly with duration of exposure. In men without silicosis, the SRR for lung cancer was 1.20 (21 deaths; (95% CI, 0,74–1.83]) in smokers and 0.85 (7 deaths; [95% CI, 0.34–1.75]) in non-smokers. The corresponding SRRs for men with silicosis were 2.34 (21 deaths: [95% CI, 1.45 –3.58]) and 2.13 (12 deaths: [95% CI, 1.10–3.72]), respectively.

Diatomaceous earth

Checkoway et al. (1993) conducted a cohort mortality study of 2570 diatomaceous earth industry workers from two plants in Southern California, United States, In this industry, the raw material is calcined at temperatures ranging from 800 °C to 1000 °C with conversion of the amorphous silica mainly to cristobalite. The main study cohort was defined as while men workers employed for at least 12 months' cumulative service. Follow-up was performed for the years 1942–87. The analysis focused on exposures to crystalline silica. Semi-quantitative exposure to airborne dust was estimated for each cohort memher and so far as possible workers thought to have been exposed to asbestos were excluded. Vital status was ascertained for 91% of the cohort and certified cause of death obtained for 94% of the 628 deaths. Only 129 workers from the cohort (5%) were classified as only having had amorphous silica exposure, from opencast mining of the ore. Compared with white United States males, the SMR for all causes was 1.12 (95% CI, 1.03–1.21), the excess largely explained by increased risks for lung cancer (59 deaths; SMR, 1.43; 95% CI, 1.09–1.84) and non-malignant respiratory disease (77 deaths; SMR, 2.27; 95% CI, 1.79–2.83). Results obtained by use of local county mortality rates were not shown but the SMR for lung cancer was reported as 1.59. Internal exposure–response analyses were performed for lung cancer and non-malignant respiratory disease mortality with respect to cumulative exposure to crystalline silica. Evidence supportive of dose–response was produced for lung cancer; the rate ratio in the highest exposure category reached 2.74 (19 observed; 95% CI, 1.38–5.46), assuming a 15-year latency. A similar gradient was found for non-malignant respiratory disease (excluding pneumonia and infectious respiratory diseases. Limited data available on cigarette smoking did not suggest that this factor could account, for these trends.

In view of the possibility that exposure to asbestos might have been more extensive than originally thought, further analyses were later undertaken to study this question in detail (Checkoway et al., 1996). This examination was restricted to a subset of 2266 workers from the larger of the two diatomite plants in the original cohort of 2570 white men; for these workers, it was possible to add individual assessments of asbestos exposure to those of crystalline silica. Workers hired before 1930 were excluded because of uncertainties of the asbestos exposure data. There were 52 deaths from lung cancer in this subset giving an overall SMR of 1.41 (95% CI, 1.05–1.85). Of the 52 deaths, 22 were in men in the lowest category of silica exposure (SMR, 1.16; 95% CI, 0.73–1.75); 15 of the 22 deaths occurred in men not exposed to asbestos (SMR, 1.13; 95% CI, 0.63–1.86); a total of 31 deaths were seen in men exposed to silica (all categories) but not asbestos (SMR, 1.34; 95% CI, 0.91–1.91). An exposure–response gradient for lung cancer was detected with respect to the crystalline silica index, lagged by 15 years. The rate ratio reached 1.83 (10 observed; 95% CI, 0.79–4.25) in the highest exposure category. Following adjustment for asbestos exposure, the exposure–response gradient for crystalline silica was virtually identical (rate ratio, 1.79; 95% CI, 0.77–4.18).

3.1.1. Mouse

A group of 60 female BALB/cBYJ mice, six weeks old, was exposed by inhalation in chambers to quartz (Min-U-Sil, a crystalline silica containing more than 96% quartz) for 8 h per day on five days per week. Subgroups of six to 16 mice each were exposed for total periods of 150, 300 or 570 days, and mice were necropsied either immediately after the end of the exposure period or following a holding period of 30 or 150 days. The average exposure concentrations of particles (diameter < 2.1 µm [diameter not further specified]) were approximately 1475, 1800 and 1950 µg/m³ for the subgroups exposed for 150, 300 and 570 days, respectively. A similar group of 59 controls consisting of subgroups of seven to 13 mice each was not exposed to silica but was sacrificed by the same schedule. Pulmonary adenomas (type II, Clara-cell and mixed type II and Clara-cell tumours) were found in both silica-exposed mice (overall incidence, 9/60) and in controls (overall incidence, 7/59); these incidences were not significantly different. The overall incidences of severe pulmonary lymphoid cuffing and heavy alveolar macrophage accumulation were 37/60 and 39/60, respectively, in silica-treated mice and 5/59 and 3/59, respectively, in controls; the difference in incidences between the silica-treated and control animals was statistically significant (p < 0.05) (Wilson et al., 1986). [The Working Group noted the small numbers of animals in the subgroups and the variable exposure and observation periods.]

3.1.2. Rat

Groups of 72 male and 72 female Fischer 344 rats, 3 months old, were exposed by inhalation in chambers to 0 or 51.6 mg/m³. Quartz (Min-U-Sil 5; mass median aerodynamic diameter, 1.7–2.5 µm; geometric standard deviation, 1.9–2.1) for 6 h per day on five days per week for 24 months. After four, eight, 12 and 16 months of the experiment, 10 males and 10 females per group were removed from the chambers; five were sacrificed and five were retained with no further exposure. All survivors were killed at 24 months. Mean survival was 688 ± 13 days for controls and 539 ± 13 days for rats exposed to quartz until death, the difference being statistically significant (p < 0.05). The incidence of epidermoid carcinomas of the lungs in treated rats still alive at 494 days, when the first pulmonary tumour appeared, was 10/53 (19%) females and 1/47 (2%) males. Three of five female rats that received no further exposure to quartz after four months also developed epidermoid carcinomas; metastasis to the mediastinal lymph nodes was reported in one of these female rats. None of the 42 male or 47 female controls developed a lung tumour. Additional lesions in quartz-treated rats included areas of pulmonary adenomatosis and nodular fibrosis, cuboidal metaplasia of the alveolar epithelium, as well as alveolar proteinosis and peribronchiolar lymphoreticular hyperplasia (Dagle et al., 1986). [The Working Group noted that, due to inadequate reporting, it cannot be determined from which exposure subgroups animals surviving at 494 days were derived; no statistical analysis of lung carcinoma incidences was reported.]

One group of 62 female Fischer 344 rats [age unspecified] was exposed by nose-only inhalation to 12 ± 5 mg/m³ quartz (Min-U-Sil; mass median aerodynamic diameter: 2.24 ± 0.2 µm, and a geometric standard deviation of 1.75 ± 0.3; respirable fraction 70 ± 3% according to the criteria of the American Conference of Governmental and Industrial Hygienists; all particles < 5.0 µm) for 6 h per day on four days per week for 83 weeks; the animals were observed for the duration of their life span. Controls were sham-exposed to filtered air (62 females) or were unexposed (15 females). Mean survival times were 683 ±108 days for quartz-exposed rats and 761 ± 138 days for sham-exposed controls. [The survival time of the unexposed controls was not specified.] Of the quartz-exposed rats, 18/60 had lung tumours (three squamous-cell carcinomas, 11 adenocarcinomas and six adenomas), all of which were observed after 17 months or more of exposure. No lung tumour was observed in 54 sham-exposed controls; 1/15 unexposed controls had an adenoma of the lung. Most of the quartz-exposed rats still alive after 400 days developed pronounced pulmonary fibrosis, lung granulomas and silicotic nodules, often accompanied by emphysema and alveolar proteinosis (Holland et al., 1983, 1986). A morphological description of the tumours is given by Johnson et al. (1987). The peripheral adenomatous lung tumours were found to be composed predominantly of alveolar type II pneumocytes.

Groups of 50 male and 50 female viral antibody-free SPF (specific pathogen-free) Fischer 344 rats, eight weeks old, were exposed by inhalation in chambers to 0 or 1 mg/m³ silica (silicon dioxide, type DQ 12; 87% crystallinity as quartz; mass median aerodynamic diameter about 1.3 µm, with a geometric standard deviation of 1.8; respirable fraction 74% according to the criteria of the American Conference of Governmental and Industrial Hygienists) for 6 h per day, five days per week for 24 months; the rats were then kept without further exposure for another six weeks. Mean survival in the treated and control groups was comparable; at the termination of the study at 25.5 months, 40% of the control and 35% of the silica-treated animals survived (not statistically different by the Kaplan-Meier method using a life-test programme). The incidences of primary lung tumours in rats exposed to silica were 7/50 males (one adenoma, three adenocarcinomas, two benign cystic keratinizing squamous-cell tumours, one adenosquamous carcinoma and one squamous-cell carcinoma; one animal had an adenoma and an adenocarcinoma) and 12/50 females (two adenomas, eight adenocarcinomas and two benign cystic keratinizing squamous-cell tumours); only 3/100 controls [sex unspecified] had primary lung tumours (two adenomas and one adenocarcinoma). The combined incidence of benign and malignant lung tumours in silica-treated rats (19%) was significantly elevated compared to the incidence of 3% in the control group (using simple tests for homogeneity of contingency tables using χ²-statistics or Fisher's exact methods) [no p values were given]. The first tumour in silica-exposed rats was observed after 21 months of exposure. In a 21-month parallel serial sacrifice study, one further lung tumour (an adenoma) was found among 13 silica-exposed rats versus no lung tumours in a total of 11 controls. Nodular bronchoalveolar hyperplasia, interpreted as borderline to adenoma, was found in 13/100 silica-exposed rats (distributed about equally between the sexes) and was not reported to occur in controls. Other nonneoplastic pulmonary lesions occurring in high incidences in silica-exposed rats included the following: multifocal lipoproteinosis with and adjacent to fibrotic areas, foamy macrophages containing lipoid substances; intra-alveolar and interstitial inflammatory cell infiltrates mainly consisting of polymorphonuclear leukocytes; moderate degrees of multifocal (predominantly subpleural and peribronchiolar) fibrosis; and alveolar- and bronchiolar-type bronchoalveolar hyperplasia. The severity of these pulmonary lesions, in particular the fibrosis, increased with increasing exposure time (Muhle et al., 1989, 1991, 1995).

Two groups of 70 male and 70 female (Cpb: WU, Wistar random) rats, six weeks old, were exposed by inhalation in chambers to 0 (controls) or 58.5 ± 0.7 mg/m³ quartz (Sikron [Sykron] F300 obtained from Guertz Werke, Frechen, Germany, crystalline, hydrophilic, pH 7, 99% SiO₂; BET-surface area, < 1.5 m²/g; geometric diameter (mean), 8 µm with a global range of 0.1-25 µm; no agglomeration; edges coarse, irregular and sharp) for 6 h per day on five days per week for 13 weeks. At the end of the exposure period and at 26, 39, 52 and 65 weeks after the start of exposure, 20, 10, 10, 10 and 20 rats per sex per group were killed, respectively. Only one respiratory tract tumour was observed, namely a small squamous-cell carcinoma in the lung parenchyma of a quartz-treated female killed at 65 weeks. In addition, a focus of squamous metaplasia in the periphery of the lung was found in one quartz-treated male killed at 65 weeks. Major non-neoplastic pulmonary changes in quartz-treated animals were the accumulation of alveolar macrophages, granulomatous inflammation, interstitial fibrosis, bronchiolo-alveolar hyperplasia and fibrotic granulomas. Associated lymph nodes contained many macrophages with or without cellular necrosis and slight fibrosis (Reuzel et al., 1991). [The Working Group noted the short duration of the study, the lack of information on survival and that only a small proportion of the quartz particles was respirable to rats.]

Three groups of 90 female Wistar rats, six to eight weeks old, were exposed by nose-only inhalation to 0, 6.1 ±0.36 or 30.6 ± 1.59 mg/rn³ quartz (DQ 12; mass median aerodynamic diameter, 1.8 µm with a geometric standard deviation of 2.0) for 6 h per day, on five days per week for 29 days. In each group, interim sacrifices of two to six rats each were made directly after quartz exposure and six, 12 and 24 months later; the terminal sacrifice was made 34 months after exposure. The mean survival times were 741 ± 179 and 739 ± 191 days for the control and low-dose groups, respectively. The mean survival in the high-dose group, reported as a survival curve only, seemed to be slightly lower than that in the two other groups, particularly in the final few months of the study (Kaplan-Meier curves). Twenty-four months after treatment, the numbers of rats with lung tumours were 8/37 and 13/43 in the low- and high-dose groups, respectively. The total incidences of lung tumours were 37/82 (45.1%) and 43/82 (52.4%) for the low- and high-dose groups, respectively. No lung tumour was observed in controls. In many animals, more than one lung tumour of the same type or different types were found; 62 tumours (eight bronchiolo-alveolar adenomas, 17 bronchiolo-alveolar carcinomas, 37 squamous-cell carcinomas, one anaplastic carcinoma) were found in the low-dose group and 69 (13 bronchiolo-alveolar adenomas, 26 bronchiolo-alveolar carcinomas, 30 squamous-cell carcinomas) in the high-dose group. Metastases were observed most frequently in the tracheobronchial lymph nodes and occasionally in the kidneys and the heart. Treatment- and dose-related non-neoplastic pulmonary lesions included increased numbers of alveolar macrophages, thickening of the alveolar walls, perivascular accumulation of inflammatory cells, degeneration of alveolar macrophages, alveolar proteinosis, granulomas, emphysema, interstitial fibrosis and proliferation of alveolar and bronchiolar epithelium. Only in single cases were bronchiolo-alveolar carcinomas accompanied by marked fibrosis, indicating, at most, a weak influence of marked fibrosis on lung-tumour development in female rats (Spiethoff et al., 1992).

3.3.1. Mouse

In a screening study based on the induction of lung adenomas in strain A mice, a group of 30 male strain A/J mice, 11–13 weeks old, received weekly intratracheal instillations of 2.9 mg (9.75 mg/kg bw) silica (Min-U-Sil 216 quartz purchased from Whittaker, Clark and Daniels, Inc., NJ, United States; 1–5 µm) [size not further specified] in 0.02 mL vehicle [vehicle unspecified] for 15 weeks. A group of 20 mice was treated similarly but with the vehicle only. A positive control group of 30 mice received a single intraperitoneal injection of 0.1 mL urethane (64.1 mg/kg bw) in sterile saline. Survivors (all animals but one of the vehicle control group) were killed 20 weeks after study initiation. The incidences of lung adenomas were 9/29 (31%), 4/20 (20%) and 18/30 (60%) in the vehicle control, silica-treated and positive control groups, respectively. The average numbers of lung adenomas per mouse were 0.31 ± 0.09, 0.20 ± 0.09 and 0.97 ±0.19 for vehicle controls, silica-treated mice and positive controls, respectively. The differences in tumour incidence and multiplicity were statistically significant between positive and vehicle controls (Fisher's exact test; p < 0.05 for tumour incidence, p < 0.01 for tumour multiplicity). Differences in tumour incidence and multiplicity between silica-treated and vehicle controls were not statistically different (McNeill et al., 1990).

Two groups of 26 male mice from each of three strains (A/JCr, BALB/cAnNCr and (athymic nude) NCr-NU) received one single intratracheal instillation of either 10 mg/animal quartz (Min-U-Sil < 5; 99% pure with 0.1% iron [presence of iron in Min-U-Sil is not uncommon]; surface area 3.15 m²/g; particle size distribution mostly between 0.5 and 2.0 µm) or 10 mg/animal tridymite (area surface 5.24 m²/g) [particle size and particle size distribution unspecified] in 0.1 mL saline. Survivors for more than six months were studied at unscheduled death up to 24 months. The incidences of lung tumours in animals treated with Min-U-Sil and tridymite were 2/15 (one adenoma and one adenocarcinoma) and 4/16 (four adenomas, one of the adenomas not in a silicotic area) in A/JCr mice, 2/26 (one adenoma and one adenocarcinoma, the adenocarcinoma not in a silicotic area) and 2/22 (two adenomas) in BALB/cAnNCr mice and 1/4 (one adenoma) and 0/5 in NCr-NU mice, respectively. In view of the incidence of spontaneous lung adenomas in strain A mice and the low incidence of the lung tumours in tridymite-treated mice, the authors regarded the observed tumours as unrelated to treatment. Non-neoplastic pulmonary changes were analogous in the three strains of mice for both types of silica, and mainly consisted of silicotic granulomas with large necrotic centres, alveolar proteinosis, transient hyperplasia of bronchial and bronchiolar epithelium and only sporadic and transient hyperplasia of the alveolar epithelium (Saffiotti, 1990; 1992; Saffiotti et al., 1996). [The Working Group noted both the lack of information on survival and the absence of a control group.]

3.3.2. Rat

A group of 40 Sprague-Dawley rats [sex and age unspecified] received weekly intratracheal instillations of 7 mg quartz (Min-U-Sil; mean particle size 1.71 ± 1.86 µm; all particles < 5 µm) in 0.2 mL saline for 10 weeks. A group of 40 rats received saline only and another group of 20 animals was untreated. All animals were observed for the duration of their life span. Lung tumours were reported in 6/36 quartz-treated rats (one adenoma and five carcinomas) [type of carcinomas unspecified] and in 0/40 saline-treated and 0/18 untreated controls. Focal and diffuse pulmonary fibrosis was only observed in quartz-treated animals (Holland et al., 1983). [The Working Group noted the absence of information on survival.]

Groups of 85 male Fischer 344 rats, obtained when weighing 180 ± 15 g and treated two weeks later, received a single intratracheal instillation of 20 mg quartz into the left lung either as Min-U-Sil (particle size, 0.1% ≥5 µm; surface area, 4.3 m²/g) or as novaculite (from Malvern Minerals Co., Hot Springs, AR, United States; particle size, 2.2% ≥ 5 µm; surface area, 1.6 m²/g) in a suspension of filtered, deionized water [volume unspecified]. Controls received the suspension vehicle alone. Interim sacrifices of 10 rats each were made at six, 12 and 18 months; terminal sacrifice was made at 22 months. In the Min-U-Sil-treated group, the incidences of lung tumours were 1/10 at 12 months, 5/10 at 18 months, 5/17 in rats that died between 12 and 22 months, and 19/30 at 22 months; total incidence was 30/67 (45%). All tumours were adenocarcinomas, some of which had squamous and/or undifferentiated areas. The incidences of lung tumours in the novaculite-treated group were as follows: 1/10 at 12 months, 2/10 at 18 months, 2/17 in rats that died between 12 and 22 months, and 16/35 at 22 months; total incidence was 21/72 (29%). One tumour was an epidermoid carcinoma; all others were adenocarcinomas; 87% of the tumours were in the left lung. In the control group, 1/44 had a lung tumour (an adenocarcinoma) at 22 months; total incidence was 1/75. The total lung tumour incidences in Min-U-Sil- or novaculite-treated rats were significantly different from that in controls (Fisher's exact test; p < 0.001). The Min-U-Sil-treated group had larger lung tumours and more extensive granulomatous and fibrotic lung lesions than the novaculite-treated group (Groth et al., 1986).

Groups of male and female F344/NCr rats [initial numbers unspecified], four to five weeks old, received one single intratracheal instillation of 12 or 20 mg/animal quartz Min-U-Sil 5 (99% pure with 0.1% iron; surface area 3.15 m²/g; particle size distribution mostly between 0.5 and 2.0 µm) in 0.3 and 0.5 mL saline, respectively [one source mentions 0.3 mL saline for the 20 mg dose], 12 mg hydrofluoric acid-etched Min-U-Sil 5 (prepared as described by Saffiotti, 1962) (99% pure with no iron; surface area 2.98 m²/g; particle size distribution mostly between 0.5 and 2.0 (µm) in 0.3 mL saline, or 20 mg ferric oxide (haematite, Fe₂O₃; non-fibrogenic dust) in 0.3 mL saline [or 0.5 mL saline; see above]. A group of untreated controls was also observed. The number of animals in each group [not further specified], the number of animals observed at interim kills or after unscheduled death and the incidences, total numbers, multiplicity and types of lung tumours found are summarized in Table 24. Type, degree and incidences of nonneoplastic pulmonary changes were very similar in each of the quartz-treated groups, and included the following: macrophage reaction; interstitial fibrosis; hyperplasia of peribronchial lymphoid tissue; silicotic granulomas increasing in size and becoming more fibrotic with time; and hypertrophy, hyperplasia and adenomatoid proliferation of alveolar epithelium. The mediastinal lymph nodes showed reactive hyperplasia (Saffiotti, 1990; 1992; Saffiotti et al., 1996).

Table 24

Incidence, numbers and types of lung tumours in F344/NCr rats after a single intratracheal instillation of quartz.

Six groups of female Wistar rats, 15 weeks old, received one single intratracheal instillation or 15 weekly intratracheal instillations of one of three quartz preparations (DQ 12, Min-U-Sil, quartz ± 600 [sources unspecified]) in 0.4 mL 0.9% sodium chloride solution (see Table 25). An additional control group of rats received 15 weekly instillations of the sodium chloride solution only. To retard silicosis development, two of the experimental groups of rats each received seven subcutaneous injections of 2 mL 2% polyvinylpyridine-N-oxide (PVNO) in saline; the first injection was given one day before the first intratracheal instillations of quartz, and the remaining six injections were given at four-month intervals. The animals died spontaneously or were killed when moribund or at 131 weeks. Animals treated with quartz DQ 12 developed severe silicosis and had a relatively short survival (median survival time about 15 months as visible from mortality curves). Owing to the protective effect of PVNO against silicosis, the groups treated with DQ 12 or Min-U-sil and PVNO developed more pulmonary squamous-cell carcinomas (Pott et al., 1994).

Table 25

Incidence of lung tumours in female Wistar rats after intratracheal instillation of quartz.

3.3.3. Hamster

Two groups of 48 Syrian hamsters [sex and age unspecified] received intratracheal instillations of 3 or 7 mg quartz (Min-U-Sil; mean particle size 1.71 ± 1.86 µm; all particles < 5 µm) in 0.2 mL saline once a week for 10 weeks. A group of 68 animals received saline only and another group of 72 animals was untreated. All animals were observed for the duration of their life span. No lung tumour was observed among 31 low-dose animals, 41 high-dose animals, 58 saline controls or 36 untreated controls. Both the incidence and severity of pulmonary fibrosis were minimal. Pneumonitis-pneumonia complex occurred in 13/31 and 21/41 of animals receiving the low and high dose, respectively, late in the exposure period (Holland et al., 1983). [The Working Group noted absence of information on survival]

Groups of 25–27 male outbred (LAK:LVG) Syrian golden hamsters, 11 weeks old, received weekly intratracheal instillations of 0.03, 0.33, 3.3 or 6.0 mg quartz (Min-U-Sil; particle diameter: median, 0.84 ±0.07 µm; average, 1.06 ± 0.07 µm; mass median, 3.14 ±0.24 µm; mass aerodynamic, 5.13 ±0.40 µm) in saline [volume unspecified] for 15 weeks. Groups of 27 saline-treated and 25 untreated hamsters served as controls. Animals were killed when moribund or when survival within the group reached 20%; any remaining groups were killed at 24.5 months of age. The average survival times were 498 ± 44, 506 ± 41, 383 ± 31 (p < 0.005 compared with saline-treated controls) and 348 ± 26 days (p < 0.005 compared with saline-treated controls) for the groups treated with 0.03, 0.33, 3.3 and 6.0 mg quartz, respectively, and 534 ± 35 and 595 ± 14 days for the saline-treated hamsters and untreated controls, respectively. No pulmonary tumour was observed in any of the groups. In animals treated with quartz, dose-related alveolar septal fibrosis of slight to moderate degree, granulomatous inflammation and alveolar proteinosis were observed in the lungs, but no animal developed nodular fibrosis or foci of dense fibrous tissue in the lung (Renne et al., 1985).

Three groups of 50 male outbred Syrian golden hamsters, seven to nine weeks old, received weekly intratracheal instillations of 1.1 mg quartz as Sil-Co-Sil (Ottawa Silica Sand; Sil-Co-Sil 395–325 grain fineness number; surface area 0.0021 m²), 0.7 mg Min-U-Sil (5) µm; surface area 0.0021 m²) or 3.0 mg ferric oxide (particulate negative control) in 0.2 mL saline for 15 weeks. A group of 50 vehicle controls received instillations of 0.2 mL saline alone. Survivors were killed 92 weeks after first treatment. Survival was significantly lower in the Sil-Co-Sil-treated group than in the Min-U-Sil-treated group and in the saline control group (p < 0.05) [survival not further specified; method of statistical analysis unspecified]. One adenosquamous carcinoma of the bronchi and lung was observed in the Min-U-Sil-treated group at week 68 (effective number of animals, 35). No respiratory tract tumour was found in the 50 hamsters treated with Sil-Co-Sil or in the 48 saline-treated controls. In the ferric oxide-treated group, one benign tumour of the larynx (papilloma or adenoma) was observed at week 62 (effective number of animals, 34). Bronchiolo-alveolar hyperplasia was occasionally seen in the particulate-treated animals. No pulmonary fibrosis was observed; however, pulmonary granulomatous inflammation was significantly increased in Sil-Co-Sil- and Min-U-Sil-treated hamsters compared to saline controls (p < 0.001) (Niemeier et al., 1986).

3.5.1. Mouse

In a study reported as an abstract, three groups of 37–43 male Marsh mice, three months of age, received a single intrathoracic injection [method of administration not further specified] of 10 mg/animal tridymite (prepared in the laboratory from silicic acid with a 0.002% heavy metal-iron content; particle size, 20% < 3.3 µm and 40% in the range 6.6–15 µm) in saline, 5 mg/animal chrysotile (acid washed, containing 0.4% iron and 0.05% copper) in saline or saline alone. After 19 months, the effective numbers of mice were 32–34 per group. Among the animals given tridymite one developed a lung adenocarcinoma and two intrapleural lymphoid tumours; there was one lung adenocarcinoma and no lymphoid tumour in saline controls; there were four lung adenocarcinomas and four lymphoid tumours in the chrysotile group. Lesions reported as ‘lymph node reactive hyperplasia simulating malignancy’ were found in 19/32 tridymite-, 1/32 chrysotile- and 1/34 saline-treated mice; the differences between the tridymite-treated mice and the chrysotile- and saline-treated were highly statistically significant (p < 0.02; Yates correction) (Bryson et al., 1974).

3.5.2. Rat

Two groups of 48 male and 48 female SPF Wistar rats and two groups of 48 male and female standard Wistar rats, six weeks old, received a single intrapleural injection of 20 mg/animal quartz (alkaline-washed silica supplied by Dr G. Nagelschmitt, Safety in Mines Research Establishment who prepared it from Snowit, a silica sand produced commercially in Belgium; particle size < 5 µm) suspended in 0.4 mL saline or 0.4 mL saline alone, and were observed for their life span. The 50% survival of quartz-treated rats was about 850 days and that of quartz-treated standard rats about 700 days [distribution of survival times of males and females together given as bar diagrams]. Mean survival times of saline-treated controls (males and females) were 883 and 725 days for SPF and standard rats, respectively. Malignant tumours of the reticuloendothelial system involving the thoracic region were observed in 39/95 quartz-treated SPF rats (23 histiocytic lymphomas, five Letterer-Siwe or Hand-Schüller-Christian's disease-like tumours, one lymphocytic lymphoma, four lymphoblastic lymphosarcomas and six spindle-cell sarcomas) and in 31/94 standard rats (30 histiocytic lymphomas, one spindle-cell sarcoma) compared to 8/96 SPF controls (three lymphoblastic lymphosarcomas, five reticulum-cell sarcomas) and 7/85 standard controls (one lymphoblastic lymphosarcoma and six reticulum-cell sarcomas) [p < 0.001]. The earliest quartz-induced tumour occurred 296 days after injection in SPF rats and 58 days after injection in standard rats, but the next tumour in standard rats did not occur until more than 300 days after injection. Most tumours occurred between 300 and 1000 days after injection [time to appearance of reticuloendothelial thoracic tumours in controls unspecified]. These tumours were predominantly observed in the upper mediastinum, the pericardium, the diaphragm and the lungs, and their distribution corresponded to that of silicotic nodules. In addition to the reticuloendothelial tumours in about one third of the quartz-treated animals, another third (21 standard and 30 SPF-rats) showed ‘hyperplastic reaction’ (granulomatous lesions) only, mainly in the thoracic cavity. A variety of other tumours did not appear to be associated with treatment. Standard rats often had accompanying infections that were absent in the SPF rats (Wagner & Berry, 1969; Wagner, 1970; Wagner & Wagner, 1972).

In a larger study, a total of 23 malignant reticuloendothelial tumours (21 malignant lymphomas of the histiocytic type (MLHT) with often widespread dissemination, two lymphosarcomas/thymomas/spindle-cell sarcomas) was observed in a group of 80 male and 80 female Caesarean-derived SPF inbred Wistar rats [distribution of the tumours over the sexes unspecified], on average 39 days old, that received a single intrapleural injection of 20 mg/animal quartz (alkaline-washed quartz (see above); particle size, < 5 µm) suspended in 0.4 mL saline. Two males and two females were sacrificed every five weeks; at 120 weeks, the remaining rats were killed. No MLHT and one thymoma/-lymphosarcoma occurred in a group of 15 saline-treated controls. In addition to tumours, (widespread) silicotic nodules occurred in most of the quartz-treated rats examined (Wagner, 1976).

A group of 16 male and 16 female Caesarean-derived SPF inbred Wistar rats, on average 39 days of age, received a single intrapleural injection of 20 mg/animal quartz (Min-U-Sil, a naturally occurring, commercial, fine quartz said to be 99% pure) in 0.4 mL saline. The animals were killed when moribund (mean survival, 678 days). Eight of 32 rats developed MLHT and 3/32 developed thymomas/lymphosarcomas [sex unspecified]. In 15 controls treated with saline only (mean survival, 720 days), no MLHT but one thymoma/lymphosarcoma was found. In addition to tumours, ‘hyperplastic reaction’ was reported to occur in 16/32 quartz-treated rats and in none of the rats treated with saline (Wagner, 1976).

A group of 16 male and 16 female Caesarean-derived SPF inbred Wistar rats, on average 39 days of age, received a single intrapleural injection of 20 mg/animal cristo-balite (prepared by heating Loch Aline sand for 1 h at 1620 °C; containing 0.6 × 10⁶ particles/µg; particle size distribution: 58.7% 0–1 µm, 28.9% 1–2 µm, 10.4% 2–4.6 µm; Wagner et al., 1980) in 0.4 mL saline. The animals were killed when moribund; mean survival time was 714 days. Eighteen of 32 rats developed malignant lymphoma (13 MLHT and five thymomas/lymphosarcomas) [sex unspecified]. In 15 controls treated with saline only (mean survival, 720 days), no MLHT but one thymoma/lymphosarcoma was found. In addition to tumours, ‘hyperplastic reaction’ was reported to occur in 13/32 cristobalite-treated rats and in none of the rats treated with saline (Wagner, 1976).

Groups of 16 male and 16 female Wistar-derived Alderley-Park rats, five to six weeks of age, received a single intrapleural injection of 20 mg of one of four quartz preparations (Table 26) in 0.4 ml saline. The incidence of MLHT observed in each treated group (except that receiving DQ 12) over the life span was statistically significantly higher than that in saline controls (Wagner et al., 1980).

Table 26

Incidences of malignant lymphoma of the histiocytic type (MLHT) in rats after an intrapleural injection of 20 mg/animal quartz.

Groups of 16 male and 16 female Wistar-derived Alderley-Park rats, 12 male and 12 female PVG rats and 20 male and 20 female Agus rats, five to six weeks of age, received a single intrapleural injection of 20 mg quartz (Min-U-Sil) in 0.4 mL saline. Groups of 16 male and 16 female Wistar rats, 12 male and 12 female Agus rats and eight male and four female PVG rats were injected with saline alone. All rats were observed for the duration of their life span. Mean survival times for quartz-treated animals were 545 days for Wistar rats, 666 days for PVG rats and 647 days for Agus rats [mean survival times of controls unspecified]. MLHT was seen in 11/32 (34%) Wistar-derived, 2/24 (8.3%) PVG and 2/40 (5%) Agus rats [sex unspecified]. Tumour morphology was similar in all strains, except that the Wistar rats showed histological evidence of tumour spread below the diaphragm. No MLHT was found in any saline-injected control rat (Wagner et al., 1980).

A group of 16 male and 16 female Wistar-derived Alderley-Park rats, five to six weeks of age, received a single intrapleural injection of 20 mg/animal cristobalite (see above; containing 0.6 × 10⁶ particles/µg; particle size distribution: 58.7% 0–1 µm, 28.9% 1–2µm, 10.4% 2–4.6 µm) in 0.4 mL saline. Mean survival was 597 days. Of 32 rats observed for life span, four developed MLHT [sex unspecified]; no such tumour was found in 16 male and 16 female saline controls (mean survival, 717 days) (Wagner et al., 1980).

A group of 16 male and 16 female Wistar-derived Alderley-Park rats, five to six weeks of age, received a single intrapleural injection of 20 mg/animal tridymite (prepared by Safety-in-Mines Research Laboratories, Sheffield, United Kingdom, by dissolving impurities from silica cement that had had long service at approximately 1380 °C in a gas-retort house; the sample contained 0.35 × 10⁶ particles/µg; particle size distribution: 34.9% 0-1 µm, 44.9% 1–2 µm, 21.2% 2–4.6 µm) in 0.4 ml saline. Mean survival was 525 days. Of 32 rats observed for life span, 16 developed MLHT [sex unspecified]. No such tumour was found in 16 male and 16 female saline controls (mean survival, 717 days) (Wagner et al., 1980).

Two groups of 36 male SPF non-inbred Sprague-Dawley rats, two months old, received a single intrapleural injection of 20 mg/animal quartz (DQ 12) in 1 mL saline or 1 mL saline only. A group of 27 male rats served as untreated controls. All rats were allowed to live until they died or were moribund. Mean survival times were 769 ± 155, 809 ± 110 and 780 ± 132 days for untreated, saline- or quartz-treated groups, respectively; differences between groups were not statistically significant (Student's Mest). Six malignant histiocytic lymphomas (17%; observed between 899 and 911 days after treatment) and two malignant Schwannomas (6%; observed between 885 and 911 days after treatment) were found in the quartz-treated group. One chronic lymphoid leukaemia and one fibrosarcoma were observed in the saline and untreated groups, respectively. ‘Granulomatous reactions’ were observed in 5/34 quartz-treated rats but in none of the controls (Jaurand et al., 1987).

3.9.1. Inhalation exposure

Rat: Two sets of three groups of 90 female Wistar rats, six to eight weeks of age, were exposed by nose-only inhalation to 0, 6.1 ± 0.36 or 30.6 ± 1.59 mg/m³ quartz (DQ 12; mass median aerodynamic diameter, 1.8 µm with a geometric standard deviation of 2.0) for 6 h per day on five days per week for 29 days. Immediately after the last exposure, five rats of both the low- and high-quartz exposure group and two sham-exposed control animals were sacrificed. One week after the end of the exposure period, all 90 rats of one of the two sham-exposed control groups, and of one of the two low- and high-quartz exposure groups received a single intravenous injection of 600 µL enriched Thorotrast (2960 Bq ²²⁸Th per mL) in saline. In each of the six groups, interim sacrifices of three or six animals each were made six, 12 and 24 months after the end of the exposure period. Survival was reduced and deaths occurred earlier (Kaplan-Meier curves) in the rats exposed to low- and high-quartz levels combined with Thorotrast as compared with their quartz-exposed but Thorotrast-free counterparts; the differences were highly statistically significant (p < 0.001; log-rank test). A similar difference was found between sham-exposed controls and rats treated with Thorotrast only. The reduction in survival was caused by a higher incidence of (fatal) lung cancer at earlier times, by the occurrence of Thorotrast-induced (fatal) liver and spleen tumours and by Thorotrast-treated non-specific life-shortening effects. Incidences, numbers and types of lung tumours, and total incidences of liver and spleen tumours in the six groups are presented in Table 27. Comparison of the cumulative rates of animals with fatal and incidental lung tumours (Kaplan–Meier curves) in the groups exposed to quartz and treated with Thorotrast with those in the corresponding Thorotrast-free groups revealed for the Thorotrast treatment a marked, positive trend of high statistical significance (p < 0.001). This trend suggests a pronounced interactive effect of Thorotrast and quartz on pulmonary carcinogenesis in female rats. Non-neoplastic pulmonary changes in quartz-exposed rats with or without Thorotrast treatment included the following: degeneration of alveolar macrophages; alveolar proteinosis; granulomas; interstitial inflammation and early fibrosis; emphysema; and hyperplasia of alveolar and bronchiolar epithelium. These non-neoplastic changes were more pronounced in the high- than in the low-quartz-exposure group, but were not aggravated in animals also given Thorotrast. Marked pulmonary fibrosis occurred only in a few quartz-exposed or quartz-exposed plus Thorotrast-treated animals, and only occasionally were bronchiolo-alveolar carcinomas accompanied by extensive scar tissue, indicating at most a weak influence of fibrosis on lung tumour development. Results obtained in animals treated with quartz only are reported Section 3.1 (Spiethoff et al., 1992).

Table 27

Numbers of animals with lung, liver and spleen tumours, numbers and types of lung tumours, and total incidences of liver and spleen tumours in female Wistar rats after inhalation exposure to quartz and Thorotrast.

3.9.2. Intratracheal administration

(b) Hamster

Groups of 50 male outbred Syrian golden hamsters, seven to nine weeks of age, received the following weekly intratracheal administrations in 0.2 mL saline for 15 weeks: 3 mg/animal benzo[a]pyrene; 3 mg ferric oxide; 3 mg ferric oxide with 3 mg benzo[a]pyrene; 1.1 mg/animal Sil-Co-Sil from Ottawa Silica Sand; 1.1 mg of the Sil-Co-Sil with 3 mg benzo[a]pyrene; 0.7 mg Min-U-Sil; 0.7 mg Min-U-Sil with 3 mg benzo[a]pyrene; 7 mg/animal Min-U-Sil plus 0.3 mg/animal ferric oxide; 7 mg Min-U-Sil plus 0.3 mg ferric oxide plus 3 mg benzo[a]pyrene. Control animals received administrations of 0.2 mL saline alone. Survivors were killed 92 weeks after the first treatment. In addition to the tumour data presented in Table 28, bronchiolo-alveolar hyperplasia was commonly seen in the particulate plus benzo[a]pyrene groups and only occasionally in the particulate control groups. No pulmonary fibrosis was observed; however, pulmonary granulomatous inflammation was significantly increased compared to saline controls in the groups receiving Sil-Co-Sil, Min-U-Sil or Min-U-Sil plus ferric oxide alone or in combination with benzo[a]pyrene. Results obtained in animals treated with quartz only are discussed in Section 3.3 (Niemeier et al., 1986). [The Working Group noted the inadequate reporting of survival times.]

Table 28

Incidences of respiratory tract tumours in hamsters after intratracheal administration of quartz with or without benzo[a]pyrene.

3.9.3. Intrapleural administration

Rat: Eighty male SPF Sprague-Dawley rats, three months of age, were exposed by inhalation to ²²²Ra at 100% equilibrium with radon daughters for 10 h per day on four days per week for 10 weeks (dose rate of 3000 WL/day; total dose of 6000 working-level months). Sixty rats received no further treatment. Two weeks after exposure to radon, two groups of 10 rats each received a single intrapleural injection of 2 mg/animal of either DQ 12 quartz (particle size, 90% < 0,5 µm) or BRGM quartz (French quartz from Fontainblau prepared by the Bureau de Recherches Géologiques Minières, Orléans la Source, France; particle size, 90% < 4 <m) in 0.5 mL saline. The animals were observed for life span, and all were necropsied. Of the group exposed only to radon by inhalation, 17/60 developed bronchopulmonary carcinoma (28%) and 0/60 pleural or combined pulmonary-pleural tumours. In the group receiving radon plus DQ 12 quartz, 4/10 developed bronchopulmonary carcinomas and 2/10 combined pulmonary-pleural tumours. In the group receiving radon plus BRGM quartz, 1/10 developed a bronchopulmonary carcinoma and 3/10 pulmonary-pleural tumours (Bignon et al., 1983). [The Working Group noted that groups receiving quartz alone or vehicle alone were not included and that the groups receiving combined treatment were comprised of small numbers of animals.]

3.1.1. Mouse

Four groups of 40 male and 40 female B6C3F, mice, five weeks old, were fed diets containing 0 (controls), 1.25, 2.5 or 5% food-grade micronized silica (Syloid 244; SiO₂. xH₂O; a fine white silica powder). The total intake of silica was 38.45, 79.78 and 160.23 g/mouse for males, and 37.02, 72.46 and 157.59 g/mouse for females in the low-, mid- and high-dose group, respectively. After six and 12 months, 10 animals per sex per group were killed; the remaining animals were killed at 21 months. Survival was high in all groups (data presented as cumulative survival rate curves). Mean survival was greatest in the 5%-dose group for both sexes, but there were no statistically significant differences in survival rate between groups (Mantel-Hanszel χ²-test). Tumour response in the silica-fed mice was not statistically significantly different from that in controls (Fisher's exact test; Cochran-Armitage test for trend) (Takizawa et al., 1988) [The Working Group noted the development of tumours in the haematopoietic organs, particularly malignant lymphoma/leukaemia in females of the 2.5%-dose group, but considered the increased incidence to be random since no dose–response relationship was observed.]

3.1.2. Rat

Four groups of 40 male and 40 female Fischer rats, five weeks old, were fed diets containing 0 (controls), 1.25, 2.5 or 5% food-grade micronized silica (Syloid 244; SiO₂. xH₂O; a fine white powder). The total silica intake was 143.46, 179.55 and 581.18 g/rat for males, and 107.25, 205.02 and 435.33 g/rat for females in the low-, mid-and high-dose group, respectively. After six and 12 months, 10 animals per sex per group were killed; the remaining animals were sacrificed at 24 months. Survival was high in all groups (data presented as cumulative survival rate curves) and highest in the 5%-dose group, but there were no statistically significant differences in mean survival rate between groups (Mantel–Hanszel χ²-test). Tumour response in silica-fed rats was not increased significantly in comparison to that in controls (Fisher's exact test; Cochran-Armitage test for trend) (Takizawa et al., 1988).

3.2.1. Mouse

Groups of 75 mice of a mixed strain, divided approximately equally by sex, about three months old, were either untreated or exposed by inhalation in a chamber (capacity of 600 L) to about 0.5 g per day precipitated silica [source unspecified] (particle size: ‘many appeared to be about 5 µm or less in diameter’) or to ferric oxide dust once an hour for 6 h on five days per week for one year. The animals were observed for life span. Survival at 600 days was 12/74 and 19/75 for the silica-treated and ferric oxide-treated mice, respectively, and 17/75 in the control group for silica and 13/73 in the control group for ferric oxide. The incidences of pulmonary tumours (adenomas and adenocarcinomas) in mice surviving 10 months or longer were 13/61 (21.3%) for silica-exposed animals and 5/63 (7.9%) for the controls, and 17/52 (32.7%) for ferric oxide-exposed animals and 5/52 (9.6%) for the controls. Nodular fibrotic overgrowth or hyperplasia of the tracheobronchial lymph nodes was found in 18/61 (29.5%) silica-treated mice, in 26/52 (50%) ferric oxide-treated mice and in 9/63 (14.3%) and 7/52 (13.4%) of the respective controls that survived 10 months or more (Campbell, 1940). [The Working Group noted the inadequate description of the test material and the exposure conditions.]

3.2.2. Rat

Two groups of 35 male SPF-bred Han: Wistar rats, about 10 weeks old, were exposed by inhalation in chambers (capacity 2 m³) to 10.91 mg/m³ quartz-glass (amorphous glass dust VP 203-006 with an infrared spectrum corresponding to that of silicic acid; 50%-value for the particle size distribution in the inhalation chamber was 0.42 µm) or 11.12 mg/m³ crystalline quartz (DQ 12; 99% of the particles < 4 µm; 50%-value for the particle size distribution in the inhalation chamber was 0.40 µm) for 7 h per day on five days per week for a maximal period of 12 months (in total, 251 exposure days during 56 weeks). A similar group of unexposed male rats served as controls. After four and eight months, five rats from each group and, after 12 months, 15 rats of each of the exposed groups and 10 controls were killed. The remaining survivors were kept for a 12-month post-exposure period. Six rats of the quartz-glass group, three rats of the crystalline-quartz group and three controls died or were killed because they were seriously injured during fighting with their cage mates, resulting in 4/35, 7/35 and 7/35 survivors in the respective groups at the end of the study [survival not further specified]. Only one primary respiratory tract tumour was found, namely a squamous-cell carcinoma of the lung in a crystalline-quartz-treated animal. The major non-neoplastic pulmonary change in quartz-glass-exposed rats was slight, focal cellular reaction with minimal fibrosis; lungs of crystalline-quartz-exposed rats showed severe macrophage reaction, fibrosis, emphysema and focal adenoid transformation of type II pneumocytes. Mediastinal lymph nodes in both exposed groups were strongly enlarged and showed severe fibrosis with bundles of hyalinized collagen fibres (Rosenbruch et al., 1990). [The Working Group noted the small number of animals surviving to two years.]

3.5.1. Intratracheal administration

Hamster: Groups of 24 male and 24 female randomly bred Syrian golden hamsters, six to seven weeks of age, received weekly intratracheal instillations of 3 mg/animal silica (fine particles) [the nature of the silica sample was not further described, except that it was obtained from Sigma Chemical Co., St Louis, MO, United States; the company's catalogues first described the item as amorphous silica and subsequently as a mixture of amorphous and crystalline particles, particle size unspecified] or 1.5 mg/animal manganese dioxide (fine particles) [particle size not further specified], 3.0 mg/animal benzo[a]pyrene (ground for 24 h in a mullite mortar; particle size, 100% < 20 µm, 98% < 10 µm, 79% < 5 µm, 5% < 1 µm), a mixture of 3.0 mg/animal silica and 3.0 mg/animal benzo[a]pyrene (prepared by ball-milling the suspensions together for seven days) [particle size of the mixed dust unspecified] in 0.2 mL saline for 20 weeks. A control group of 24 males and 24 females received saline alone, and a group of 50 males and 50 females served as untreated controls. Survival at 50 weeks was 18/48 saline controls, 13/48 silica-treated, 9/48 manganese dioxide-treated, 15/46 benzo[a]pyrene-treated, 19/48 silica plus benzo[a]pyrene-treated and 16/46 manganese dioxide plus benzo[a]pyrene-treated animals and 75/100 untreated controls. The incidences of respiratory-tract tumours were 0/48 saline controls, 0/48 silica-treated, 0/48 manganese dioxide-treated, 5/48 benzo[a]pyrene-treated (one papilloma and one squamous-cell carcinoma of the larynx, four papillomas of the trachea), 21/48 silica plus benzo[a]pyrene-treated (eight papillomas of the trachea, one squamous-cell carcinoma of the larynx, two of the trachea and three of the bronchus/lung, three adenocarcinomas and six adenomas of the bronchus/lung) [p < 0.001 as compared to benzo[a]pyrene alone] and 5/46 manganese dioxide plus benzo[a]pyrene-treated (one papilloma of the larynx and three of the trachea, one squamous-cell carcinoma of the respiratory tract) animals and 0/100 untreated controls. Bronchiolar and alveolar adenomatoid lesions were frequently encountered in animals treated with silica plus benzo[a]pyrene; these lesions occurred much more frequently in animals treated with manganese dioxide plus benzo[a]pyrene and were not seen at all in any of the other groups (Stenbäck & Rowland, 1979). The authors later reported similar effects with silica and with other dusts, such as ferric oxide, titanium dioxide and talc, mixed with benzo[a]pyrene (Stenbäck et al., 1986). [The Working Group noted that the silica tested might have been a mixture of amorphous and crystalline silica.]

(a) Deposition

The deposition of a respirable particle is a function of its aerodynamic diameter, which is defined as the diameter of a sphere of unit density having the same terminal settling velocity as the particle itself (Jones, 1993). The site of deposition in the respiratory tract is dictated by the aerodynamic diameter. In humans, while large particles with an aerodynamic diameter greater than 10 µm will deposit in the upper respiratory tract, only those below 10 µm will deposit with any efficiency in the tracheobronchial region; for the alveolar region, deposition only begins to become substantial with aerodynamic diameters well below 10 µm (Task Force Group on Lung Dynamics, 1966).

Deposition of particles in the respiratory bronchioles and proximal alveoli results in slow clearance, interaction with macrophages and a greater likelihood of lung injury. This contrasts with deposition on the conducting airways where the majority of the particles are cleared by the mucociliary escalator. Therefore, quartz particles with an aerodynamic diameter below 10 µm are likely to be the most harmful to humans.

(b) Distribution and clearance

There are few data on human lung quartz-dust burdens that allow conclusions to be drawn about deposition or clearance. However, quartz is found in the bronchoalveolar macrophages and sputum of silicotic patients (Sébastien, 1982; Porcher et al., 1993). Also, at autopsy, there is wide variation in the masses and proportions of quartz retained in the lung (Verma et al., 1982; Gibbs & Wagner, 1988). For example, Verma et al. (1982) reported 25–264 mg per single lung at autopsy in hard-rock miners with 14–36 years of exposure; these miners had variable amounts of pathological response but there was not a good correlation between lung crystalline quartz content and pathological score. The well-documented effect of smoking on clearance (Morgan, 1984) is a further confounding factor in drawing conclusions about clearance kinetics in humans.

(c) Biopersistence of silica

The physico-chemical changes in quartz that result from residence in the lung could be an important factor in determining the continuing toxicity of quartz to the lung following deposition. As a response to the rejection of the ‘mechanical model’ of silicosis, which had propounded that any particle with ‘sharp or jagged edges’ might injure tissue, a solubility theory of silicosis was proposed. The solubility theory was based on the release from silica of silicic acid, which was considered to be a ‘protoplasmic poison’ (King & McGeorge, 1938; King, 1947). In fact very little dissolution occurs; for example, 9 mg SiO₂(0.45%) was released from 2 g crystalline silica placed in ascitic fluid for two weeks (King & McGeorge, 1938). Current theories no longer consider that the dissolution of quartz contributes substantially to its clearance or to changes in its biological activity (Vigliani & Pernis, 1958; Heppleston, 1984). Indeed there is evidence of enrichment of crystalline silica in lungs of individuals exposed to hard rock compared to the dust in the air they breathed (Verma et al., 1982), suggesting that crystalline silica is less-efficiently cleared, either by dissolution or mechanical clearance, than the non-silica mineral components of the dust. In the study of Pairon et al. (1994), biopersistence was assessed in occupationally exposed subjects by counting silica particles in bronchoalveolar lavage (BAL) fluid after varying periods of time since their last occupational exposure. Crystalline silica was found to be among the most biopersistent of non-fibrous mineral particles.

(a) Deposition

Animals such as the rat demonstrate different alveolar deposition patterns from humans, with negligible deposition of particles of aerodynamic diameter above 6 µm. This variation arises because of differences in the mode (mouth and nose) and pattern (cycle period and tidal volume) of inhalation between the species (Jones, 1993); these factors need to be considered in the interpretation of animal studies.

Quartz particles with an aerodynamic diameter below 6 µm are likely to be most harmful in rats. Brody et al. (1982) confirmed, in rats exposed short term to 109 mg/m³ quartz (high purity Thermal Americal Fused Quartz Co.), that there was a substantial deposition on the alveolar duct/terminal bronchiolar surfaces of silica particles with an average aerodynamic diameter of 1.4 µm (range, 0.3–4.0 µm). In another inhalation study in rats, which used Min-U-Sil silica of aerodynamic diameter 3.7 µm, more than 80% of the particles that deposited peripherally were found on the alveolar ducts, particularly their bifurcations, and on the distal terminal bronchioles (Warheit et al., 1991a).

(b) Distribution and clearance

Immediately following deposition of quartz on the surface of the mammalian lung, there is either rapid mucociliary clearance if deposition is in the upper airways, or phagocytosis by alveolar macrophages and slower clearance if deposition is in the lung periphery (Brody et al., 1982; Warheit et al., 1991a). There are differences between species in terms of clearance rates (Oberdörster, 1988; Jones, 1993), with clearance from the lungs of humans, dogs and guinea-pigs being slower than from the lungs of rats and hamsters.

Clearance by mucociliary mechanisms is generally considered to be efficient; clearance from the lung periphery is slow and incomplete, i.e. there is a sequestered dust fraction that is never cleared (Morgan, 1984; Vacek et al., 1991).

A number of possible fates of particles after deposition in the lung periphery have been suggested: (i) phagocytosis by macrophages followed by migration to the mucociliary escalator for clearance; (ii) persistent macrophage accumulations in the airspaces (Stöber et al., 1989); (iii) penetration to the interstitium for phagocytosis by interstitial macrophages and possible exudation back on to the alveolar surface (Vacek et al., 1991); (iv) penetration to the interstitium; and (v) translocation to the lymph nodes (McMillan et al., 1989; Absher et al., 1992). All of these possibilities, except the first, would result in slower clearance or sequestration.

The kinetics of deposition and clearance of quartz have been successfully studied in the rat model. In rats, three days after a 3-h inhalation exposure to quartz, it was observed (using scanning electron microscopy) that the particles that had deposited on the terminal bronchiolar and alveolar duct surfaces were translocated into epithelial cells and to the interstitium (Brody et al., 1982). McMillan et al. (1989) reported impaired clearance of an inhaled, relatively innocuous particle of similar size, titanium dioxide, during concomitant inhalation of Sykron F600 quartz. Furthermore, analysis of the lymph nodes revealed that the decreased lung burden could be largely explained by translocation of the quartz to the lymph nodes. More recently, Vacek et al. (1991) monitored the disposition of Min-U-Sil 5 quartz and C&E Mineral Corp. cristobalite in alveolar fluid, free cells, lung tissue and lymph nodes over six months following eight days of exposure of rats for 7 h per day to 11–65 mg/m³ particles with a mass median aerodynamic diameter of around 1.0 µm. Twenty-four hours were allowed to elapse for tracheobronchial clearance and thereafter rats were killed at regular time-points for assessment of the lung burden in the various compartments. The data were then applied to a number of mathematical models and the best fit determined. The model that fitted the data best used no clearance of quartz via the mucociliary escalator, which was explained by the relative toxicity of the cristobalite and Min-U-Sil to macrophages, preventing their movement. Donaldson et al. (1990a) demonstrated that alveolar macrophages from rats inhaling Sykron F600 quartz were indeed impaired in their ability to migrate in response to a standard chemotactic signal, C5a. The model of Vacek et al. (1991) also showed considerable transfer of quartz to the lymph nodes and to another, notional, compartment. The continued accumulation of quartz in the lymph nodes up to 150 days after cessation of exposure in this model of cristobalite silicosis (Absher et al., 1992) reveals the dynamic nature of the redistribution of cristobalite that occurs following deposition. The same laboratory (Hemenway et al., 1990) also described clearance of C&E Mineral Corp. cristobalite and two types of quartz (Min-U-Sil 5 and Thermal American Fused Quartz Co.; TAFQ), which had similar aerodynamic diameters, following inhalation exposure in rats. There were very large differences in the clearance of the three samples, with cristobalite being cleared markedly more slowly than the two types of quartz. These differences have been a result of the greater severity of lung injury and inflammation caused by inhalation of cristobalite compared to the two quartz types.

Heating of CRS cristobalite increased its accumulation in the lungs and lymph nodes. TAFQ quartz heated to 800 °C for 24 h was found in high amounts in the thymus and lymph nodes of rats exposed by inhalation; an unhealed sample was biologically inactive (Hemenway et al., 1994).

A physiologically based kinetic model of quartz deposition and lung response suggested the probable importance of interstitialization of quartz, followed by interstitial inflammation, in the development of silicosis (Tran et al., 1996). Transport of Sykron F600 quartz to the lymph nodes has been found to coincide with the onset of inflammation (Vincent & Donaldson, 1990) in a rat model of ongoing quartz exposure. Inflammatory leukocytes from DQ 12 quartz-exposed lung have been shown to cause loss of integrity and detachment of epithelial cell monolayers in culture (Donaldson et al., 1988a), which may be a factor that promotes the interstitialization of quartz in the inflamed lung.

More experimental evidence for the importance of interstitialization in the pathogenic effects of silica comes from Adamson (1992) who demonstrated that depletion of the macrophage defences in male Swiss-Webster mice by 6.5 Gy whole body irradiation allowed increased interstitial access of quartz particles (Dowson & Dobson). This led to enhanced phagocytosis by interstitial macrophages, which in turn led to a florid interstitial response with fibroplasia and collagen accumulation. This study emphasized the importance of the macrophage response in dealing with deposited quartz. The same laboratory (Adamson et al., 1994) showed that generation of a controlled inflammatory response in the alveolar space by instillation of N-formyl-l-methionyl-leucyl-phenyl-alanine (FMLP), a leukocyte chemotactic factor, ameliorated the harmful effects of quartz. In this case, mice received an instillation of quartz and a subgroup received an instillation of FMLP two to three weeks later. The quartz plus FMLP-treated rats showed significantly lower lung tissue burden and lymph node burden of quartz, which resulted in less fibrosis. This outcome was interpreted to be a consequence of the attraction of quartz-loaded macrophages from the interstitium into the alveolar space, with concomitant lowering of the interstitial dose of quartz and the dose available for lymphoid transport.

Amorphous silica is cleared more quickly from the lungs of rats than quartz. For instance, rats inhaling Ludox colloidal amorphous silica at 50 or 150 mg/m³ showed clearance half-times of 40 and 50 days, respectively (Lee & Kelly, 1992). This contrasts with half-times of > 125 days for rats inhaling cristobalite (Hemenway et al., 1990), while Driscoll et al. (1991) described only 20% clearance of Min-U-Sil quartz 20 days after a five-day inhalation of 50 mg/m³.

Crystalline silica

In humans, exposure to crystalline silica causes the following range of non-neoplastic pulmonary effects:

(a) Inflammation

Bégin (1986) and Rom et al. (1987) described increased uptake of ⁶⁷Ga, an index of inflammatory macrophage activation, in the lungs of silicotics. Increases in neutrophils, macrophages and lymphocytes in the BAL fluid of silica-exposed populations was reported by Bégin (1986), while Rom et al. (1987) found increases only in lymphocytes in a population of silicotics. In another BAL study, healthy granite workers showed only a modest, insignificant increase in neutrophils in BAL fluid, although an increase in lymphocytes was significant (Christman et al., 1985). A study of granite workers with silicosis in Québec, Canada, showed a 2.4-fold increase in macrophage numbers and a 4.4-fold increase in lactate dehydrogenase (LDH), suggesting that cell death occurred in the silicotic lung (Bégin et al., 1993) (see Table 30).

Table 30

Bronchoalveolar lavage (BAL) fluid leukocyte populations in crystalline silica-exposed populations.

(b) Silicosis

Silicosis has been detected by X-ray (e.g. Graham, 1992), lung-function testing (e.g. Ng & Chan, 1992) and computed tomography (CT) scan (e.g. Bégin et al., 1988). Parkes (1994) described the following four different types of silicosis:

1. Nodular fibrosis — comprising collagenous nodular lesions with a substantial content of quartz. These nodules arise focally in macrophage/reticulin complexes within the interstitium at the level of the respiratory bronchioles and become progressively more collagenized until the full-blown silicotic nodule arises; as they evolve, the nodules become more-or-less hyalinized, necrotic or calcified (Graham, 1992). Progressive massive fibrosis (PMF) arises from the agglomeration of nodules (Silicosis and Silicate Disease Committee, 1988) possibly as a result of high focal quartz content (Leibowitz & Goldstein, 1987).
2. Mixed dust fibrosis — less-well-defined stellate fibrotic lesions of radially arranged collagen strands and dust-containing macrophages caused by exposure to free silica plus an inert material (Silicosis and Silicate Disease Committee, 1988).
3. Diffuse interstitial pulmonary fibrosis — this type of focal interstitial change is associated with combined exposure to silica plus other silicate minerals, e.g. in foundries and diatomaceous earth processing plants where cristobalite is present (Silicosis and Silicate Disease Committee, 1988).
4. Rapidly occurring diffuse interstitial pulmonary fibrosis with alveolar lipo-proteinosis (acute or accelerated silicosis) — this condition develops after heavy exposure to silica-containing dust (e.g. during sandblasting) and is progressive in the absence of further exposure (Silicosis and Silicate Disease Committee, 1988); the patient often dies of respiratory failure.

(c) Lymph node fibrosis

Silicotic mediastinal adenopathies were found in two workers exposed to cristobalite during the changing of diatomaceous earth-containing filters in breweries (Nemery et al., 1992).

Preferential transport of quartz to the lymph nodes in lungs exposed to mixed dust has been described by Chapman and Ruckley (1985) and hilar and mediastinal lymph nodes frequently show silicotic nodules at autopsy (Silicosis and Silicate Disease Committee, 1988) with calcification in some cases (Sargent & Morgan, 1980). In a necropsy study, fibrosis of the lymph nodes appeared to be a factor that predisposed to parenchymal silicosis (Murray et al., 1991).

(d) Airways disease

Neukirch et al. (1994) described chronic airflow limitation that was independent of radiographic change in pottery workers exposed to silica dust. Cowie and Mabena (1991) reported chronic airflow limitation afflicting all of a population of South African gold miners who were exposed to silica-containing dust, as well as symptoms of bronchitis in men who worked in the dustiest occupations. Ng and Chan (1992) reported obstructive impairment of lung function in active and retired granite workers and that was related to the extent of radiological opacities. Using a sensitive measure of airway function, Chia et al. (1992) demonstrated significant small airways obstruction associated with silica dust exposure in the absence of radiological evidence of silicosis among currently employed granite quarry workers. Hnizdo (1990) noted a synergistic effect of smoking and gold-mine dust exposure in leading to death from chronic obstructive lung disease, with 5% of deaths from chronic obstructive lung disease being attributable to dust alone, 34% from smoking and 59% from the combined effects of dust and smoking.

(e) Emphysema

An association has been demonstrated between emphysema and exposure to silica-containing dusts or silicosis (Becklake et al., 1987; Hnizdo & Sluis-Cremer, 1991; Cowie et al., 1993; Leigh et al., 1994). In non-smoking South African gold miners with a long duration of exposure, only a minimal degree of emphysema was found at autopsy (Hnizdo et al., 1994). Using CT, 48 out of 70 men who had worked underground for an average of 29 years in the gold mining industry were found to have emphysema (Cowie et al., 1993).

(f) Epithelial effects

Increased permeability of the airspace epithelium to inhaled small molecular weight compounds is a feature of smokers (Minty et al., 1981) and is considered to play a role in the development of lung inflammation in smokers. Nery et al. (1993) reported that the airspace epithelium of non-smoking silicotics was more permeable than that of normal individuals and that there was an additive effect in smoking silicotics, who showed a markedly increased permeability. Hyperplastic type II cells were found in increased numbers in the BAL of silicotics (Schuyler et al., 1980) even years after cessation of exposure to silica, suggesting ongoing injury and proliferation.

(g) Tuberculosis

The highly fatal consumptive disease of the lungs in hard-rock miners, described by G. Agricola in the sixteenth century, is thought to have resulted from exposure to quartz, arsenic and uranium in the presence of tuberculosis (TB). However, despite the dramatic reduction in the prevalence of TB in the twentieth century, a South African study recently reported that the annual incidence of TB was 981/100 000 in men without silicosis and 2707/100 000 in men with silicosis (Cowie, 1994). In a study of 5406 underground haematite miners in China, 25% of the workers had silicosis and 42% of these silicotics had TB (Chen et al., 1989).

(h) Extra-pulmonary effects of silica

Silica exposure has been found to have a number of extra-pulmonary effects and indeed the term ‘extrapulmonary silicosis’ has been coined (Slavin et al., 1985). This term encompasses the spread of lesions to the liver, spleen, kidneys, bone marrow and extrathoracic lymph nodes. Silicosis of the liver has been especially well documented (reviewed in Slavin et al., 1985).

Abnormal renal function has been recorded in silica-exposed individuals with and without silicosis (Hotz et al., 1995). Also, a relationship has been described between length of exposure to silica and to severity of renal dysfunction (Ng et al., 1993). However, in another case–control study, silicosis was associated with renal alterations but there was no relationship between the loss of renal function and the length of exposure or severity of silicosis (Boujemaa et al., 1994). Persistence of renal effects after cessation of silica exposure was reported in the study of Ng et al. (1992). A relationship between rapidly progressive glomerulonephritis and silica exposure was shown in a hospital-based case–control study by Gregorini et al. (1993), and Michigan men with exposure to silica were found to have an elevated odds ratio for end-stage renal disease (Goldsmith & Goldsmith, 1993). The presence of silica within the renal tubules was reported in one case study of silica-related glomerulonephritis (Osorio et al., 1987). Systemic sclerosis-like (scleroderma-like) disorders have been reported following exposure to silica (Cowie & Dansey, 1990; Haustein et al., 1990; Rustin et al., 1990).

Abrasion-related deterioration in dental health has been recorded in Danish granite workers (Petersen & Henmar, 1988), and evidence of increased incidence of rheumatoid arthritis was found in Finnish granite workers (Klockars et al., 1987). Occasionally, cutaneous exposure to silica causes granulomas that may mimic cutaneous sarcoidosis (Mowry et al., 1991) or granulomatous cheilitis (Harms et al., 1990).

(a) In-vivo effects of silica

(i) Inflammation

Exposure of rats to crystalline silica results in a marked inflammatory response characterized by a high percentage of neutrophils (see Table 31).

Table 31

Bronchoalveolar lavage (BAL) cell populations in rats exposed to crystalline silica.

Female Fischer 344 rats were exposed by inhalation to air, 0.1, 1.0 or 10 mg/mg³ quartz (Min-U-Sil) for 6 h per day on five days per week for four weeks (Henderson et al., 1995). The mass median aerodynamic diameter of the aerosol was 1.3–2.0 µm. Lung responses were characterized by analysis of BAL fluid one, eight and 24 weeks after exposure and by histopathology 24 weeks after exposure. Mean lung burdens, determined one week after the end of exposure, were 43, 190 and 720 µg/mg quartz for the low-, mid- and high-exposure levels. Exposure to 10 mg/m³ resulted in lung injury and inflammation demonstrated by progressive increases in BAL fluid neutrophils and lactate dehydrogenase. Exposure to 1.0 mg/m³ quartz resulted in a transient increase in BAL fluid neutrophils one week after exposure. Histopathology 24 weeks after exposure to 10 mg/m³ demonstrated an active-chronic inflammatory response associated with the bronchial-associated lymphoid tissues, interstitium and intrapleural regions. In this study, exposure to 0.1 mg/m³ quartz had no apparent effects with no changes in BAL fluid or histopathology.

Exposure of rats to quartz (Sykron F600; Min-U-Sil 5) by inhalation produced a time-dependent and dose-dependent accumulation of macrophages and neutrophils in the BAL fluid (Donaldson et al., 1988b; Warheit et al., 1991a; Velan et al., 1993). The inflammation persisted after the end of exposure and progressed in the rats that had received high exposures (Donaldson et al., 1988b, 1990b) suggesting a mechanism for the well documented progressive nature of silicosis. In contrast, a similar airborne mass concentration of Ludox colloidal amorphous silica caused only very modest neutrophilic inflammation that resolved over a three-month recovery period (Warheit et al., 1991b; Lee & Kelly, 1993). Following long-term, moderate inhalation exposure of rats, guinea-pigs and adult male cynomolgus monkeys to amorphous silica (origin not stated), Groth et al. (1981) reported that, histologically, only the monkeys showed evidence of inflammatory macrophage accumulations and early silicotic lesions, suggesting species differences in deposition, clearance or response to this material. [The Working Group noted that no information was provided on species differences in lung dust burdens.]

Instillation of quartz in rats (Dowson & Dobson; DQ 12; Moores et al., 1981; Brown et al., 1991), guinea-pigs (Min-U-Sil; Lugano et al., 1982) and Syrian hamsters (Min-U-Sil; Beck et al., 1982) caused neutrophilic inflammation that persisted over time. Quartz (Dowson & Dobson)-induced inflammation is reflected in increases in BAL lysosomal enzyme levels in mice (Adamson & Bowden, 1984). Increased phospholipids were also recovered, which may arise from type II cell proliferation in rats in response to epithelial injury caused by Min-U-Sil (Heppleston et al., 1974) or natural sand (Eklund et al., 1991). In general, these responses were more persistent and of greater magnitude than those seen with low-toxicity dusts particles such as latex, titanium dioxide or iron.

In comparative tests, the inflammation and acute lung injury caused by freshly fractured (milled) quartz was much greater in intensity than that caused by aged quartz (Iota standard quartz sand), even though the aerodynamic diameters were very similar in the two samples (Shoemaker et al., 1995; Vallyathan et al., 1995). [The Working Group noted that the milled samples contained 222 (µg/g iron, compared with 7 (µg/g in the unmilled sample.] Inflammatory leukocytes from DQ 12 quartz-instilled lung caused detachment of epithelial cells in culture and degraded extracellular matrix (Donaldson et al., 1988b) by a largely protease-mediated mechanism which appeared to be mediated by the neutrophils and not the macrophages (Donaldson et al., 1992). In a sheep model, following multiple quartz [origin not stated] instillations, BAL cells showed increased release of superoxide (Cantin et al., 1988) but there was no such increase in BAL cells of rats following a single instillation of quartz (DQ 12; Donaldson et al., 1988c).

Groups of male and female Wistar rats (Cpb:WU, Wistar random) were exposed by inhalation in chambers to three types of amorphous silica (Aerosil 200, Aerosil R 974, Sipernat 22S), for 6 h per day on five days per week for 13 weeks. Groups of rats were killed at the end of the exposure period and at weeks 13, 26, 39 and 52. Non-neoplastic pulmonary changes seen in rats killed at the end of the exposure period comprised slight to severe accumulation of alveolar macrophages, intra-alveolar granular material, cellular debris and polymorphonuclear leukocytes in the alveolar spaces, and increased septal cellularity, seen as an increase in the number of type II pneumocytes and macrophages within the alveolar walls. In general, the most severe changes were found in rats exposed to Aerosil 200, and the mildest changes were seen in rats exposed to Sipernat 22S. Alveolar bronchiolization occurred mainly in males exposed to 5.9 or 31 mg Aerosil 200/m³ or to Aerosil R 974. During the post-exposure periods, no recovery from lung lesions was observed in a comparison group of quartz (Sikron [Sykron] F300)-exposed rats, whereas in rats exposed to the amorphous silicas, the changes disappeared partly or completely. In rats exposed to 31 mg Aerosil 200/m³ or to quartz, accumulations of alveolar macrophages were still found 52 weeks after the end of exposure. In rats exposed to Sipernat 22S or Aerosil R 974, lesions were found until week 39 after exposure. Accumulation of intra-alveolar granular material, cellular debris and polymorphonuclear leukocytes were occasionally found in the group exposed to 31 mg Aerosil 200/m³ and in all quartz-exposed rats during the post-exposure period. Rats exposed to Sipernat 22S recovered completely from the slight increases in septal cellularity that were observed at the end of the exposure period. A lesser degree of recovery was observed in rats exposed to Aerosil 200 or Aerosil R 974, and no recovery occurred in rats exposed to quartz. Alveolar bronchiolization persisted mainly in quartz-exposed animals and in some rats exposed to Aerosil 200. Focal interstitial fibrosis was first observed 13 weeks after exposure in all exposed group. During the subsequent postexposure period, this condition disappeared completely in rats exposed to Aerosil R 974 or quartz at the end of the exposure period. This lesion disappeared completely in rats of the Aerosil 200 group within 13 weeks after the end of exposure, but in rats exposed to Aerosil R974, recovery took more than 39 weeks. Slight fibrosis was observed in the granulomas in animals of the quartz group (Reuzel et al., 1991).

The substantially lower inflammatory effects of synthetic, precipitated, amorphous silica relative to crystalline silica has been demonstrated in several other inhalation studies (Hemenway et al., 1986 (precipitated Zeofree 80); Lee & Kelly, 1992 (Ludox colloidal); Warheit et al., 1995 (Zeofree 80 and Ludox)). For example, a four-week exposure of rats to airborne mass concentrations of 150 mg/m³ colloidal silica showed a return to virtually normal lung morphology after a three-month recovery period (Lee & Kelly, 1993).

(b) In-vitro cellular effects of silica

Crystalline silica

Amorphous silica

Unique or multiple (four times) epidermal application of biogenic silica fibres (mean length, 150 µm) [size distribution not described; dose unknown] in female skin promotion-responsive mice (SENCAR) resulted in an induction of ornithine decarboxylase activity in epidermal cells (Bhatt et al., 1992). Induction was maximum at 4–6 h and inhibitor studies revealed some similarities with 12-O-tetradecanoylphorbol-13-acetate. [The Working Group noted that no statistical analysis was available. The relevance of this paper in the field of the effects of amorphous silica may be questioned as only extremely long silica fibres were evaluated.]