Familial Pancreatic Cancer

Anecdotal case reports and epidemiologic studies suggest that pancreatic cancer clusters in some families. Although chance aggregations or a common environmental exposure could account for some familial clusters, evidence for a genetic basis continues to accumulate. Ghadirian and colleagues²⁵⁸ performed a population-based case-control study in Quebec, Canada, and found that 7.8% of pancreatic cancer patients reported a positive family history of pancreatic cancer, compared to 0.6% of controls. Falk and colleagues²⁵⁹ studied pancreatic cancer in Louisiana and found similar results, as evidenced by an increased risk for pancreatic cancer among persons reporting any cancer in a close relative (OR = 1.86; 95% CI = 1.42 to 2.44). The highest risk was seen in those with a history of pancreatic cancer in a close relative (OR = 5.25; 95% CI = 2.08 to 13.21).

The National Familial Pancreas Tumor Registry at Johns Hopkins is based upon at least two first-degree relatives who have pancreatic cancer.²⁶⁰ Analysis of these families indicates that even second-degree relatives of affected patients are at increased risk for pancreatic cancer.

When a familial aggregation of pancreatic cancer is recognized, a variety of hereditary cancer disorders needs to be considered. These include HNPCC (MSH2, MLH1), PJS, FAMMM, familial breast cancer (BRCA2), and hereditary pancreatitis.¹⁷⁸ (All of these conditions are discussed elsewhere in this chapter.)

Hereditary pancreatitis is characterized by recurrent episodes of severe chronic pancreatitis. Affected individuals are at risk for pancreatic pseudocysts, pancreatic exocrine failure, diabetes mellitus, and pancreatic cancer. A gene for hereditary pancreatitis has been mapped to chromosome 7q35.²⁶¹ This gene encodes a cationic trypsinogen that, when mutated, fails to inactivate trypsin, resulting in autodigestion of the pancreas. Presumably, chronic epithelial injury and regeneration increase the risk for pancreatic cancer.

Even when known hereditary syndromes are excluded, familial aggregations of pancreatic cancer remain. The inheritance pattern in these families appears to be autosomal dominant.

Surveillance of high-risk patients is evolving. Brentnall and colleagues²⁶² used endoscopic ultrasound, endoscopic retrograde cholangiopancreatography (ERCP), spiral computed tomography, and serum carcinoembryonic antigen and CA19-9 analysis to follow 14 patients from three pancreatic cancer kindreds. Seven patients underwent pancreatectomy on the basis of abnormalities on ultrasonography and ERCP. All seven had histologic dysplasia in the resected specimen; there were no occult carcinomas.

Given the known heterogeneity in pancreatic cancer, it is likely that many differing cancer-susceptibility loci will ultimately be identified. For example, Eberle and colleagues²⁶³ have identified a susceptibility locus which maps to chromosome 4q32–34, in an autosomal dominantly inherited pancreatic cancer family. In the case of BRCA2, which predisposes to the hereditary breast-ovarian cancer syndrome, Murphy and colleagues²⁶⁴ found that BRCA2 mutations significantly increase the risk of pancreatic cancer in individuals with these mutations. Indeed, these investigators indicate that BRCA2 mutations pose the most common inherited genetic alteration to date that has been identified in familial pancreatic cancer.

Familial Gastric Cancer

Gastric cancer has many similarities to esophageal carcinoma in that its incidence shows marked geographic variation. For example, its incidence is exceedingly high in Japan (160 per 100,000 population per year), with high rates also noted in Finland and Chile. In contrast, its incidence is low in the United Kingdom (10 per 100,000 population per year), and in the United States, where rates have been declining during the past several decades.

There are two types of gastric adenocarcinoma that can be distinguished histopathologically: intestinal type and diffuse type. Molecular pathology supports the theory that those differences emerge through specific genetic pathways for the two tumor types.²⁶⁵ The intestinal type shows components of glandular or intestinal architecture and tubular structures. The diffuse type shows noncoherent, single cells, often with signet-ring cell morphology, and it frequently presents as linitis plastica. Separation of gastric cancer into intestinal and diffuse types is exceedingly useful genetically and epidemiologically.

The intestinal type shows a greater frequency than its diffuse counterpart. In general, the intestinal type is more commonly related to environmental exposures; these include diet (salted fish and meat, and smoked foods), smoking, alcohol, and infection with Helicobacter pylori. Indeed, there may be a genetic-environmental interaction in the etiology of gastric carcinoma and H. pylori susceptibility.^266–268 El-Omar and colleagues²⁶⁹ showed that eradication of H. pylori infection leads to resolution of the gastric inflammation and resolution of hypochlorhydria and atrophy in about half of patients. Importantly, will eradication of H. pylori be effective in hereditary forms of gastric cancer?

In contrast, the diffuse type has a low but relatively stable incidence in most parts of the world. In general, it is related to host factors (discussed subsequently).

Familial Risk Factors

We estimate that between 5% and 10% of gastric cancer involves familial clustering. Familial risk for gastric carcinoma was investigated by Hemminki and Jiang,²⁷⁰ who employed the nationwide Swedish Family Cancer Database, which comprises 10.2 million individuals with more than 34,000 gastric carcinomas. This is the largest study on familial gastric cancer published to date. Standardized incidence rates (SIRs) were 1.31 (95% CI, 0.97 to 1.70) and 1.47 (95% CI, 1.08 to 1.92) when a parent presented with gastric cancer or gastric adenocarcinoma respectively. Risk was 1.59 (95% CI, 1.10 to 2.16) in offspring whose diagnosis was at an age greater than 50 years. Sibling risk for gastric cancer was 3.16 and 5.75 when diagnosed earlier than age 50 years. Population-attributable proportion of gastric cancer was 0.45%. These authors concluded that environmental factors, possibly H. pylori infections, provided the main explanation for familial clustering of gastric cancer. The population-attributable proportion of familial gastric cancer was much lower than that cited in the literature.

Palli and colleagues²⁷¹ studied 1,016 gastric cancer patients and 1,623 population-based controls. Their study was adjusted for potential confounders including diet. Their results showed an increased risk for gastric cancer among those with the following relatives affected: sibling (OR = 2.6); parent (OR = 1.7: mother affected, OR = 2.3; father affected, OR = 1.3); two or more siblings (OR = 8.5); both parents (OR = 3.0).

Hereditary Gastric Carcinoma

Perhaps the most celebrated family with gastric cancer is that of Napoleon Bonaparte (Figure 16-1). Lying near death on St. Helena in 1821, 52-year-old Napoleon Bonaparte was heard to murmur “Pylorus…My father's pylorus.”²⁷² His father had died of gastric cancer, as had his father's father, his brother, and three sisters. An autopsy showed that Napoleon, too, succumbed to cancer of the stomach.

Figure 16-1

Reconstructed pedigree of Napoleon Bonaparte. st = stomach, d = death.

The differential diagnosis of hereditary gastric cancer includes its integral involvement in HNPCC (Lynch syndrome).²⁷³ Park and colleagues²⁷⁴ found that after endometrial cancer, gastric cancer is the second most common extracolonic cancer in cases of HNPCC in Korea, an endemic area for gastric cancer, where the relative risk for gastric cancer is known to be increased fourfold in individuals with HNPCC. FAP, Li-Fraumeni syndrome, and Peutz-Jeghers syndrome are other examples of disorders that predispose to multiple cancers of diverse sites.

Hereditary Diffuse Gastric Cancer

Hereditary diffuse gastric cancer (HDGC) is a distinctive syndrome, a subset of which is a result of the E-cadherin gene mutation. Lobular carcinoma of the breast may be an integral lesion in those affected by the E-cadherin mutation. HDGC is an autosomal dominantly inherited syndrome. Germ line mutations in the E-cadherin (CDH1) gene were first identified by Guilford and colleagues in three Maori kindreds.²⁷⁵ Its penetrance is 70% to 80%. The average age of onset of HDGC is about 37 years.

Clinical criteria for defining HDGC families include the following: (1) two or more documented cases in first/second degree relatives, at least one diagnosed at younger than age 50 years; (2) three or more documented diffuse gastric cancers in first/second degree relatives independent of age of onset. Twenty-five percent to 50% of families meeting these criteria have identifiable germ line mutations involving the E-cadherin gene. Other families may have unidentified mutations or yet-to-be-defined genes that contribute to HDGC.

Genetic Counseling and Gastric Cancer

Lynch and colleagues²⁷⁹ performed mutation-based genetic counseling in an HDGC family that harbored the E-cadherin mutation (Figure 16-2). We invited family members as a group to attend what we call a family information service (FIS).²⁸⁰ Twenty-four members attended this session and were thoroughly educated about the clinical and genetic features of HDGC. After giving individual signed consent, they were tested for the E-cadherin mutation, of which 9 were found to be positive and 15 negative. Three of the E-cadherin mutation positives were affected and are deceased as a result of diffuse gastric carcinoma (see Figure 16-2).

Figure 16-2

Pedigree of family showing E-cadherin germ line mutation. d = death, Br = breast carcinoma, St = stomach, Lym = lymphoma, Sk = skin, leu = leukemia.

None of the 19 patients counseled wanted their results sent to their physicians once they recognized the potential for insurance discrimination. None had undergone endoscopic ultrasound. Three who were positive for the E-cadherin mutation expressed strong interest in prophylactic gastrectomy, and one of these has recently undergone prophylactic total gastrectomy and was found to be negative for metastatic disease. This individual (III-3), the proband in this family, during the genetic counseling process, was advised about the importance of undergoing total prophylactic gastrectomy. He realized that he was positive for the E-cadherin germ line mutation. He was told that many patients who had undergone prophylactic total gastrectomy had evidence of microscopic metastatic disease in the surgical specimen. He eventually became resigned to the need for this and underwent the procedure. The examination of the entire stomach revealed three foci of intramucosal signet-cell carcinoma, each < 0.5 mm in dimension. There was no evidence of lymph node involvement or distal metastatic spread (TCS, unpublished data).

In conclusion, surveillance, even with intensive endoscopic examinations and random gastric biopsies, may be inadequate for the early diagnosis and prevention of diffuse gastric cancer in HDGC families, hence justifying consideration of prophylactic surgery, particularly among E-cadherin mutation carriers. The genetic counselor must discuss the option for prophylactic gastrectomy. Clearly, there is a need for a multidisciplinary group of clinicians, geneticists, and genetic counselors to educate and support the family member to enable him or her to make decisions about diagnostic and management options for the future.

Familial Prostate Cancer

Interest in familial prostate cancer risk dates back nearly a half-century to the report by Morganti and colleagues²⁹¹ describing an 11-fold increased risk for carcinoma of the prostate in first-degree relatives of index cases when compared to age-matched controls. Later, Woolf²⁹² showed a threefold increased risk for relatives of affecteds, and in 1982, Cannon and colleagues²⁹³ concluded that prostate cancer had a stronger familial aggregation in a Utah Mormon genealogic database than breast or colorectal carcinoma. More recent epidemiologic studies confirm these findings and indicate that risk is related to the number of family members affected and the age of the proband at diagnosis.²⁹⁴

Segregation analysis has suggested that a highly penetrant autosomal dominant mutant gene could account for 9% of all prostate cancers and 40% of those diagnosed before age 55.²⁹⁵ Some candidate genes have been identified. The HPC1 locus on chromosome 1q24–q25 was identified by linkage analysis of 91 families with three or more affected first-degree relatives.^{296, 297} In contrast, the search for linkage at chromosome 1q42.2–43 was negative in one study.²⁹⁸ A putative prostate cancer susceptibility gene has been mapped to Xq27–28,²⁹⁹ a finding that correlates with suggestions of an X-linked mode of inheritance for prostate cancer susceptibility.³⁰⁰

The risk for prostate cancer is increased three- to fourfold in carriers of BRCA1 or BRCA2 mutations.^{301, 302} Hereditary breast-ovarian cancer syndrome should be ruled out in families with clusters of prostatic carcinoma.

Risks and benefits of surveillance are unproven, but regular screening by serum prostate specific antigen and regular digital rectal examination are reasonable. In high-risk families, screening should begin before age 50 years. The role of transrectal ultrasound and random prostate biopsies is yet to be evaluated.

Lung Cancer

Lung cancer, more than any other solid tumor, is readily identified with an environmental cause, namely, tobacco exposure. Some non-tobacco-related familial clusters are associated with an earlier age of diagnosis, but no genetic loci that might explain these cases have been identified.

In an effort to determine whether a genetic predisposition may exist, Lynch and colleagues³⁰³ evaluated cancer risk in the relatives of 254 consecutively ascertained probands with histologically verified lung cancer and in relatives of 231 probands with other smoking-associated cancers. There was no strong evidence for increased risk of lung cancer in relatives, but there was a significant increased risk for cancers of all anatomic sites among the relatives of lung cancer probands (p < .001). Conversely, there were no significant excesses of cancer at all anatomic sites in relatives of probands with other smoking-associated carcinomas. The observed increased risk for cancer at all anatomic sites in the relatives of lung cancer probands could result from an underlying hereditary susceptibility to cancer in general in these families.

Although no familial lung cancer syndrome has been identified, lung cancer presents a model for how the genetics of toxin metabolism may lead to cancer predisposition. Susceptibility to lung cancer may in part be attributable to interindividual variability in metabolic activation or detoxification of tobacco carcinogens. The glutathione S-transferase M1 (GSTM1) genetic polymorphism has been extensively studied in this context. A recent meta-analysis of the results of 43 case-control studies including > 18,000 individuals identified a slight excess of risk of lung cancer in association with the GSTM1 null genotype (OR = 1.2).³⁰⁴ However, a pooled analysis of 9,500 subjects involved in 21 case-control studies from the International Collaborative Study on Genetic Susceptibility to Environmental Carcinogens (GSEC) data set performed to assess the role of GSTM1 genotype as a modifier of the effect of smoking on lung cancer risk revealed no evidence of increased risk of lung cancer among carriers of the GSTM1 null genotype (age-, gender-, and center-adjusted). In addition, there was no evidence of interaction between GSTM1 genotype and either smoking status or cumulative tobacco consumption.³⁰⁴ A number of additional studies searching for associations between lung cancer and common genetic polymorphisms have been performed, but like the family studies, no convincing evidence for low penetrance susceptibility genes has emerged from this line of investigation either.

Still, millions of people smoke each year, but not all will develop lung cancer. Tobacco-associated lung cancer has been observed to cluster in some families, prompting further work on genetic determinants of toxin metabolism (for review see Hecht³⁰⁵ and Bennett and colleagues³⁰⁶). Interestingly, mouse models of toxin-related lung cancers have led to the identification of a susceptibility gene in that system.³⁰⁷ This gene, known as Pas-1, has been mapped to the mouse chromosome 6.³⁰⁸ Identification of homologs to Pas-1 in humans may eventually lead to breakthroughs in our understanding of the pathogenesis of lung cancer and its genetic determinants.

Renal Cell Carcinoma

Familial renal cell carcinoma (RCC) occurs as two distinct histologic entities: familial papillary renal carcinoma and familial nonpapillary (clear cell) carcinoma. Both have an autosomal dominant pattern of inheritance. The gene for papillary RCC is on chromosome 7q31.1-34. Germ line mutations have been documented in papillary RCC families.³⁰⁹ The gene for familial clear cell RCC has been mapped to 3p14.2 (sporadic clear cell carcinoma frequently has somatic abnormalities of 3p). This gene has not been cloned. Incidences for the two conditions are not known.

As in many hereditary cancer syndromes, early disease onset, multifocal disease, and bilateral disease are clues to the diagnosis. Family history is the mainstay of diagnosis; genetic testing is not available. In familial clusters of clear cell carcinoma, VHL disease and tuberous sclerosis should be ruled out.

Extrarenal neoplasms have been described in families with papillary renal cell carcinoma.³¹⁰ These include carcinomas of the stomach, rectum, breast, lung, pancreas, and bile duct. It is not clear whether these associations are significant.

Individuals from families with RCC should have regular renal imaging beginning at an age 10 years younger than the age at which the youngest member of the family was diagnosed with renal cancer.

Familial Neuroblastoma

Neuroblastoma is a common tumor of childhood, with almost all cases occurring prior to the age of 10 years.³¹¹ Heritable predisposition for the development of the disease has been observed but, to date, no susceptibility genes have been isolated.^312–314 Several chromosomal regions are potential sites for a neuroblastoma susceptibility gene, including 17q, 1p, 11q, and 14q. Chromosome 1p deletions have been reported in as many as 80% of near-diploid neuroblastomas that have been karyotyped, making this a particularly promising locus.³¹⁵ These deletions on chromosome 1 have variable proximal break points, but a region of consistent deletion has been mapped to sub-bands of 1p36, possibly marking the location of a tumor-suppressor gene important in malignant transformation or progression of neuroblastoma. Further support for this locus comes from the study of a sib pair with neuroblastoma. A1p36 deletion was identified in both tumors through use of double-color fluorescence in situ hybridization, and neither tumor showed evidence of MYCN amplification, another common genetic alteration in neuroblastoma. Haplotype analysis showed that the siblings inherited the same paternal 1p36→pter chromosome region in both tumors. However, the maternal 1p region was deleted. These data suggest that the siblings inherited a predisposition to neuroblastoma in the paternal 1p36 region and that tumors developed as a consequence of somatic loss of the maternal 1p36 allele.

Despite the difficulties of identifying a neuroblastoma susceptibility gene, there is convincing evidence that hereditary predisposition to neuroblastoma segregates as an autosomal dominant Mendelian trait. Thus, efforts to identify a susceptibility locus have continued. The most recent progress was reported by Maris and Matthay,³¹⁶ who performed linkage analysis on 10 families' multiple neuroblastomas and identified a locus at 16p12–p13 consistent with linkage (lod = 3.46). Informative recombinants defined a large (26-cM) deletion that is not yet amenable to candidate gene mutation screening. However, subchromosomal deletions were identified in 5 of 12 familial (42%) and 55 of 259 nonfamilial (21%) neuroblastomas at the same locus, narrowing the interval to approximately 13 cM. These data suggest that a gene for hereditary neuroblastoma is located at 16p12–p13 and that inactivation of this gene may contribute to the pathogenesis of sporadic neuroblastomas.³¹⁷

Hematologic Malignancy

Most studies dealing with familial aggregations of hematologic neoplasms have lacked adequately matched controls and have focused on the same hematologic malignancy that was present in the proband. In an attempt to address this issue, Shpilberg and colleagues³¹⁸ investigated the familial aggregation of hematologic cancers in 4,061 family members of 189 patients with a variety of hematologic neoplasms. They compared their findings with control groups of 955 relatives of 36 patients with nonmalignant hematologic disorders and 508 relatives of 33 patients with type II diabetes mellitus. They found that the majority of hematologic cancers among family members differed from those of the probands. Specifically, “The odds ratio for haematological neoplasms among relatives of the index cases adjusted for age, sex, ethnicity, number of relatives in the family, and degree of familial linkage in the study group versus the two control groups was 3.62 (95% confidence interval, 1.44 to 9.07, P < 0.01).” These findings lend support to the idea of a genetic predisposition to cancer in a subset of patients, with diverse familial clustering of the hematologic neoplasms. Therefore, they suggest that the familial aggregation of hematologic cancer is not necessarily disease specific, which is consistent with an abnormality in the pluripotent hematopoietic stem cell.³¹⁸ Alternatively, more differentiated hematologic cells of different types may be susceptible to genetic damage in some families.

Families prone to hematologic malignancies have received a paucity of attention when compared to families with solid tumors. Conversely, hematologic malignancies have had more extensive cytogenetic investigation.³¹⁹

Acute Myelogenous Leukemia

We have studied a family³²⁰ with acute myelogenous leukemia (Figure 16-3) wherein many clues were present that appear to link the presumptive genetic susceptibility of solid tumors to hematologic cancer in this family. Review of the pedigree shows several examples of key individuals with solid tumors having progeny with hematologic cancer and, conversely, patients with hematologic cancer having progeny with solid tumors. The pedigree appears to be consistent with an autosomal dominant predisposition to both hematologic and solid tumors. In the absence of a hereditary cancer disorder such as the Li-Fraumeni syndrome, these findings appear to be unique, as the hematologic cancers are predominantly acute myelogenous leukemia (AML). There are no premonitory stigmata that may associate this susceptibility to AML.

Figure 16-3

Pedigree for hematologic- and solid-tumor-prone family manifesting acute myelogenous leukemia. (Reproduced with permission from Lynch et al.) d=death

Chronic Lymphocytic Leukemia

We present another family³²¹ with hereditary chronic lymphocytic leukemia (Figure 16-4). Note that the pedigree shows chronic lymphocytic leukemia (CLL) in a father (II-1) and all four of his children (III-1, III-2, III-3, III-4). (Finding a male-to-male transmission excluded X-linked inheritance.) This finding is consistent with an autosomal dominant mode of genetic transmission for susceptibility to CLL. The occurrence of CLL in identical twins in this sibship adds to the likelihood of a primary hereditary etiology. Therefore, hereditary factors appear to be a major contributing causal factor in a subset of CLL cases.^322–324 CLL has also been reported in association with breast cancer.³²⁵

Figure 16-4

Pedigree of a family in which the father and all four children are CLL-affected, findings consistent with an autosomal dominantly-inherited predisposition to CLL. (Reproduced with permission from Lynch et al.) d = death.

Yuille and colleagues,³²⁶ employing family history questionnaires on 268 CLL-affected individuals, found that a family history of lymphoproliferative disorder (LPD) in a first-degree relative was present in 33 (12%) of those responding to the survey, while 15 (6%) reported CLL in a first-degree relative. The largest number of CLL affecteds found in any one of their families was three patients. In their literature survey of CLL families, they identified 81 familial reports of which “…64 were two-case families. Thirty-eight contained an affected sibling, and 16 contained an affected parent and affected offspring. Six pairs of monozygotic twins have been reported. In two of these cases, a first-degree relative was also affected. In five of the families with two or more cases of CLL, other relatives were affected.…The largest CLL family reported to date is a six-case family comprising three affected siblings and three affected cousins.”³²⁷ Yuille and colleagues³²⁶ suggest that there is an approximate 30-fold increase in risk for CLL in relatives of CLL affecteds.

To date, there has not been any evidence of predisposition genes identified specifically for CLL. However, hematologic cancers inclusive of CLL have been identified in a litany of mendelian inherited hematologic cancer-prone disorders, particularly such chromosomal breakage syndromes as Bloom syndrome, Fanconi aplastic anemia, and ataxia-telangiectasia.³²⁸

Multiple Myeloma

The etiology of multiple myeloma (MM) remains obscure, although there are reports of familial clustering that suggest a host susceptibility factor which may act in concert with environmental effects. Lynch and colleagues³²⁹ reported a family with MM (Figure 16-5). Their extended MM family comprised a sibship of seven, wherein three patients showed histologically verified MM, while two had a monoclonal gammopathy of unknown significance (MGUS). Other tumors in the family included acute lymphocytic leukemia, malignant melanoma, and prostate cancer. Their review of the literature clearly indicates that a subset of familial MM may also include other hematologic cancers as well as solid tumors.

Figure 16-5

Pedigree of multiple myeloma family with a sibship of seven, showing three siblings with verified MM and two siblings with monoclonal gammopathies of undetermined significance. (Reproduced with permission from Lynch et al.) d = death.

Shoenfeld and colleagues,³³⁰ in their review, identified 36 reports of familial multiple myeloma and added one family of their own. Interestingly, the patients did not show any significant clinical differences from those with nonfamilial myeloma when considering gender, age, distribution of monoclonal proteins, clinical and laboratory data, and prognosis. They did observe an increased incidence of immunoglobulin abnormalities in healthy relatives of the patients manifesting MM.

Horwitz and colleagues³³¹ described a family in which multiple myeloma was present in three siblings, two of whom had a history of a monoclonal gammopathy. Their literature review disclosed 38 pairs of siblings with plasma cell disorders, whereas eight families showed a third affected sibling and four families had a fourth affected relative, thereby suggesting that some cases of multiple myeloma may have an hereditary basis. They suggested that other family members may be at increased risk for developing the disease. These authors also noted in their review of the literature that only two other families had three siblings with MM, but they state that, “…five other sibships had combinations of myeloma and monoclonal gammopathy, and two had three siblings with monoclonal gammopathies only. Furthermore, another close relative was affected [with MM] in four of these families. Since approximately 3% of the elderly have benign [monoclonal] gammopathies, the presence of several individuals with such an abnormality in a single family might not be unusual.³³² On the other hand, frank myeloma in three siblings should be very rare.”

Bizzaro and Pasini³³³ studied a family in which five siblings had a monoclonal gammopathy. MGUS was diagnosed in two of these individuals, following which a family study showed that one sister had died from multiple myeloma, and four of the seven living siblings were discovered to have MGUS.

Hodgkin Disease and Non-Hodgkin Lymphoma

Research dealing with the genetics of Hodgkin disease and non-Hodgkin lymphoma (NHL) is limited.³³⁴ When considering leukemia or lymphoma in relatives with hematologic malignancies, case-control studies show an approximate 3.62-fold risk for these disorders.³¹⁸ An approximate ninefold increased risk for Hodgkin disease among first-degree relatives of Hodgkin disease affecteds has been described, but there was no evidence of an increased risk in NHL.³³⁵

Ferraris and colleagues³³⁶ estimate that approximately 4.5% of all cases of Hodgkin disease are familial. These authors suggest that shared environmental factors might be important, inclusive of Epstein-Barr virus (EBV) and other viral agents, which may be acting in concert with genetic determinants to explain the familial aggregation of Hodgkin disease. Mack and colleagues³³⁷ described high concordance for Hodgkin disease in monozygotic twins as compared with dizygotic twins, suggesting a role for genetic factors in the pathogenesis of Hodgkin disease.

X-Linked Lymphoproliferative Syndrome

Sullivan³³⁸ has reviewed the history of X-linked lymphoproliferative syndrome (XLP), also known as Duncan disease. Some 30 years ago, Purtilo and colleagues performed an autopsy on an 8.5-year-old male who died following infectious mononucleosis (IM). This was the third male sibling in the family to have died following a bout of IM. This led to their seminal publication in 1975, which described X-linked recessive combined variable immunodeficiency.³³⁹ More than 80 families have now been reported to the XLP syndrome registry established by Purtilo and colleagues.³⁴⁰

XLP is characterized by an enormous sensitivity to EBV. The phenotype features severe or fatal IM, acquired hypogammaglobulinemia, and malignant lymphoma. Sullivan³³⁸ lists the following phenotypic features: (1) life-threatening EBV infection (58% with a mortality of 96%); (2) immunodeficiency in 31% with a mortality of 45%; (3) lymphoma or Hodgkin disease in 30% with a mortality of 69%; and (4) aplastic anemia in 3% with a 50% mortality. Most of the lymphomas are extranodal non-Hodgkin lymphomas, with distal small bowel the site at highest risk.

Genetic linkage studies have localized the XLP locus to the long arm of the X chromosome in Xq24–q25. Coffey and colleagues³⁴¹ identified a gene termed SH2D1A “…that is mutated in XLP patients and encodes a novel protein composed of a single SH2 domain. SH2D1A is expressed in many tissues involved in the immune system. The identification of SH2D1A will allow the determination of its mechanism of action as a possible regulator of the EBV-induced immune response.” EBV infection in susceptible infants is initially asymptomatic, but by adolescence or early adulthood, IM will occur with 100% mortality by the age of 40 years.