Historically, cervical cancer screening, also known as secondary prevention of cervical cancer, was based on examining cells collected from the surface of the cervix by Pap smear (cytology), followed by colposcopy for women with abnormal smears and histological assessment, followed by surgical treatment for histologically proven cancer precursors. This approach resulted in dramatic reductions in cervical cancer incidence and mortality in health systems that were robust enough to support relatively complex screening programs effectively. However, very few LMICs have been able to initiate or sustain cytology-based screening programs because of lack of adequate resources or health care or laboratory infrastructure.
For screening and treatment of precancerous lesions, several new tools have been developed that are better suited to low-resource settings. Depending mainly on the target age group and frequency of screening, these tools may be effective in reducing cervical cancer rates. The new interventions include the following:
Impact of Cervical Cytology-Based Screening Programs
Cytology-based cervical cancer screening, which began in the early 1960s in the Scandinavian countries, was not evaluated in randomized trials to assess the impact of screening on cervical cancer incidence or mortality. The marked reduction in cervical cancer incidence and mortality after cytology-based screening programs were initiated in a variety of LMICs was interpreted as strong nonexperimental support for organized cervical cancer screening programs.
The International Agency for Research on Cancer (IARC) conducted a comprehensive analysis of data from several of the largest screening programs in the world in 1986; the analysis showed that well-organized, cytology-based screening programs were effective in reducing cervical cancer incidence and mortality (Hakama 1986). In the Nordic countries, following the introduction of nationwide screening in the 1960s, mortality rates from cervical cancer fell between 84 and 11 percent, respectively, corresponding to the country with the shortest screening interval and widest age range (Iceland) and to the country with only 5 percent population coverage by an organized screening program (Norway) (Laara, Day, and Hakama 1987).
Further, the age-specific trends indicated that the target age range of a screening program was a more important determinant of risk reduction than the frequency of screening within that age range. This finding was in agreement with the estimates of the IARC Working Group on Cervical Cancer Screening that for interscreen intervals of up to five years, the protective effect of organized screening exceeded 80 percent throughout the targeted age group (IARC Working Group on Cervical Cancer Screening 1986a, 1986b). It is clear that the extent to which screening programs have succeeded or failed to decrease the incidence of and mortality from cervical cancer is largely a function of three factors:
The extent of screening coverage of the population at risk
The target age of women screened
The reliability of cytology services in the program.
Gakidou, Nordhagen, and Obermeyer (2008) evaluated screening programs in 57 countries and found that the levels of effective screening coverage using cytology vary widely across countries, from over 80 percent in Austria and Luxembourg to less than 1 percent in Bangladesh, Ethiopia, and Myanmar. Many women in low-income countries (LICs) had never had a pelvic examination. This proportion of women is largest in Bangladesh, Ethiopia, and Malawi, where more than 90 percent of women report never having had a pelvic examination, compared with 9 percent of women living in the richest global wealth decile. Although crude coverage rates are high for women in the richest wealth deciles, effective coverage rates are overall low, with rates of around 60 percent and less than 10 percent in the poorest countries.
Screening efforts have failed to produce the expected reductions in cervical cancer mortality in many places, even when large numbers of Pap smears were performed, because the wrong women have been screened (for example, younger women attending antenatal clinics), coverage of the most at-risk population was too low (that is, women ages 35–64 years), the quality of cervical smears was poor (Irwin, Oberle, and Rosero-Bixby 1991; Lazcano-Ponce and others 1994; Sankaranarayanan and Pisani 1997), and follow-up of screen-positive women was incomplete. In all cases, funds were spent for little gain.
Alternative Approaches to Cytology for Cervical Cancer Screening
Visual Inspection with Acetic Acid
VIA involves applying a 3–5 percent acetic acid solution to the cervix and then examining it with the naked eye using a bright light source. No expensive equipment or supplies are needed, and screening takes less than five minutes. A well-defined aceto-white area close to the transformation zone indicates a positive test.
VIA is inexpensive and simple and can be carried out by primary care staff. Most important, VIA provides an immediate result that can be used to decide on treatment, usually with cryotherapy, which requires training but no surgery or anesthetic.
It is difficult to recommend VIA unconditionally, however, because its sensitivity and specificity are lower than those of other screening methods (). VIA sensitivity and specificity are variable, because they are highly dependent on the training and skill of the staff carrying out the examinations. The accuracy of the test decreases with the increasing age of the women screened. In cross-sectional studies, the sensitivity and specificity of VIA compared favorably with cytology in detecting high-grade cervical cancer precursor lesions and cervical cancer. Sensitivity has varied from 49 to 96 percent, and specificity has varied from 49 to 98 percent (Denny, Quinn, and Sankaranarayanan 2006). However, many of these studies suffer from verification bias, where the true status of disease in test-negative women is unknown. Sauvaget and others (2011) performed a meta-analysis of 26 studies of VIA with confirmatory testing, using high-grade squamous intraepithelial lesions (HSIL) as the disease threshold. Sauvaget and others (2011) report a sensitivity of 80 percent specificity (range 79–82 percent) and 92 percent specificity (range 91–92 percent) for VIA, with a positive predictive value of 10 percent. They conclude that in very low-resource settings where the infrastructure for laboratory-based testing is not available, VIA is a reasonable alternative to cytology. However, in more recent randomized studies, VIA has performed less well.
Performance and Characteristics of Screening Methods.
Despite its limitations, the possibility of immediate diagnosis and treatment makes VIA the only possible alternative in many low-resource settings. One potential use of VIA that would have a significant impact is following an HPV test, for HPV-positive women only, to make treatment decisions. The utility of VIA in this context is promising but yet to be proven.
Case Study of Upscaling VIA
From 2005 through 2009, the World Health Organization (WHO) sponsored a VIA demonstration project in six Sub-Saharan African countries: Madagascar, Malawi, Nigeria, Tanzania, Uganda, and Zambia (WHO 2012). In all, 19,579 women were screened with VIA. Of these, 1,980 were VIA-positive (11.5 percent); cancer was suspected in 326 (1.7 percent). Of the VIA-positive women, 1,737 were eligible for cryotherapy (87.7 percent); of these, 1,058 (60.9 percent) were treated, 601 (34.6 percent) were lost to follow-up, and 78 women were not treated. Of the women treated, 243 (39.1 percent) were treated during the same visit as the screening.
No information was available for 230 of the 326 women in whom cancer was suspected (70.5 percent); of the 96 women investigated, cancer was confirmed in 79, but no staging information was recorded; 77 of the women were treated, mostly with radiation.
This is an interesting study of “real world” VIA screening, with all of the difficulties of any screening program, even with a test as simple as VIA. These difficulties range from achieving adequate coverage; to losing to follow-up the large number of women needing treatment (only 60 percent of eligible women were treated); to treating women on the same day as screening (“screen and treat”), which occurred for less than 40 percent of the women. The failure to refer over 70 percent of women with suspicious lesions for further evaluation—possibly because cervical biopsy is not a free service in any of these countries and most women could not afford to pay—is disturbing. The greatest utility of VIA in countries that cannot afford any alternative is to establish the necessary infrastructure to provide health care services to older women. Once VIA becomes successfully implemented, it should be relatively easy to introduce more sensitive methods of screening into the system. In many LMICs, establishing a sustainable and appropriate infrastructure is most likely the priority.
HPV Testing
Highly sensitive and reproducible laboratory techniques to detect oncogenic HPV and cervical cancer have been developed and are being used or considered in place of cervical cytology for primary screening, in addition to other potential uses (Cuzick and others 2008). The cervix is sampled with a brush, which is inserted into the endocervix and then removed and placed in a tube containing special transport media. The U.S. Food and Drug Administration has approved five of the many tests available for routine laboratory service:
Hybrid Capture 2 detects 13 oncogenic types of HPV (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68).
Cervista HPV HR detects 14 HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68).
Cervista HPV 16/18 detects only HPV 16 and 18.
Aptima (transcription–mediated amplification test) detects RNA from 14 HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68).
Cobas 4800 (real-time polymerase chain reaction [PCR]–based test) detects 14 HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68).
Other tests that use PCR technology are being used in many clinical studies.
HPV testing is an excellent alternative to cytology for cervical cancer screening (Arbyn and others 2012). In meta-analyses of cross-sectional studies, the sensitivity of the Hybrid Capture 2 (HC2) DNA test, the most commonly used test, was 90 percent to detect CIN2+ and 95 percent to detect CIN3+, with more heterogeneity in studies from LMICs. Compared with cytology, the sensitivity of HC2 is 23–46 percent higher on average, and the specificity is 3–8 percent lower (note we are using the terminology as reported by the authors, hence the switch between cervical intraepithelial neoplasia (CIN) and squamous intraepithelial lesion (SIL) terminology).
Another advantage of HPV testing is the possibility of linking screening to treatment without colposcopy or prior histological sampling, particularly once either simplified or point-of-care HPV tests are developed. A randomized screening trial to evaluate safety investigated the acceptability and efficacy of screening women and treating those with positive tests without colposcopy and histological sampling (Denny and others 2010). A total of 6,555 previously unscreened women, ages 35–65 years, were tested for high-risk types of HPV using HC2 (Qiagen, Gaithersburg, MD, United States) and VIA, performed by nurses in primary care settings. This study found that the HPV screen-and-treat arm was associated with a 3.7-fold reduction in the cumulative detection of CIN2 or greater by 36 months; VIA was associated with a 1.5-fold reduction. For every 100 women screened, the HPV and screen-and-treat strategy averted 4.1 cases of CIN2 and greater compared with VIA-and-treat strategies, which averted 1.8 cases.
A further advantage of HPV testing is that specimens can be obtained by self-collection, with almost complete preservation of the sensitivity and specificity of the screening method. Self-collection, which can be done at home, is accepted by women and could significantly increase participation in screening, particularly by women who are reluctant to undergo a gynecological examination or who live in remote areas.
Another landmark study was a cluster randomized trial of villages and centers where 131,746 women ages 30–59 years were recruited and randomly assigned to one of four groups: HPV testing; cytologic testing; VIA; or the standard of care, which involved no organized or opportunistic screening (Sankaranarayanan and others 2009). The incidence rate of cervical cancer stage 2 or higher and death rates from cervical cancer were significantly higher in the cytologic, VIA, and control groups compared with the HPV testing group. Further, the age-standardized incidence rate (ASIR) of invasive cancer among women who had negative test results on cytological or VIA testing was more than four times the rate among HPV-negative women.
The high negative predictive value of HPV testing (nearly 100 percent) allows the extension of the screening interval, with consequent savings that can offset the possibly higher cost of the test compared with cytology. Screening with HPV testing under age 30 is not recommended, as HPV infection in this group of women is common, and most infections are likely to be transient with a low likelihood of developing into cancer. Screening younger women will add to the costs of the program and may result in significant overtreatment that may be associated with reproductive morbidity, in addition to significant emotional and social problems.
The HPV test is already in use for primary screening in several countries, although in the United States, primary HPV testing has been recommended only in combination with cytology in primary screening or for triage of cytologic abnormalities. A recent study including more than 300,000 women in the United States concluded that HPV testing without cytology might be sufficiently sensitive for primary screening (Katki and others 2011).
Triage of Positive HPV Tests
Even among women over age 30 years, most HPV infections regress; only a minority of women develop persistent infection with high-risk types of HPV that progresses to cervical cancer precursors and cervical cancer. HPV testing identifies women at risk, but not those HPV-positive women who are most likely to have or to develop in the near future significant disease requiring treatment. The challenge is to triage these women by further testing with visual methods, cytology, molecular biomarkers, or a combination of techniques.
Among the visual methods, colposcopy with subsequent biopsy and treatment of visible lesions is the usual procedure in cytology-based programs. However, this method requires highly specialized training and relatively costly equipment. More importantly, the colposcopic impression, colposcopically guided biopsy, and histologic diagnosis are poorly reproducible and have important limitations to the point of reducing the potential of highly sensitive screening tests. The current practice of selecting the most worrisome lesion for biopsy misses up to one-third of prevalent small HSIL lesions. The collection of multiple biopsies from aceto-white lesions can increase the sensitivity of colposcopy (Pretorius and others 2011).
Cytology of HPV-positive women is under strong consideration as a triage method in screening programs, given the high specificity of cytology and ample expertise and infrastructure existing in some areas. This method has the advantage of being highly specific, but it suffers from limited sensitivity. Sensitivity of cytology is influenced by many factors and is complex, but used as a triage test for women already identified as high risk, cytology may suffice. The reduction in the number of cytology tests required and the restriction to HPV-positive women may improve the quality of cytology by reducing the workload and the number of negative slides.
Using DNA biomarkers, limiting further follow-up to women infected with HPV 16 and 18, which are responsible for about 70 percent of cervical cancer and precursors, can reduce the number of women referred to colposcopy while maintaining adequate sensitivity (Castle and others 2011). Overexpression of certain oncoproteins is a marker for increased risk of progression to cervical cancer and may be a better predictor of cancer risk than HPV DNA testing alone, although this is yet to be confirmed (Dockter and others 2009). One biomarker under intensive study is p16ink4a, which is overexpressed in cancerous and precancerous cervical cells. In a meta-analysis of studies using several detection methods, the proportion of smears overexpressing p16ink4a increases with the severity of cytological abnormalities (12 percent of normals and 89 percent of HSIL) and histological abnormalities (2 percent of normals and 82 percent of CIN3) (Sahasrabuddhe, Luhn, and Wentzensen 2011). A rapid test for the E6 oncoproteins of HPV types 16, 18, and 45 is undergoing clinical trials (Schweizer and others 2010).