Family History of Cancer and Associated Risk of Developing Neuroendocrine Tumors: A Case-Control Study

Neuroendocrine tumors (NET) are rare cancers; yet, between 1973 and 1999, a total of 13,715 NETs were identified, indicating a significant increase in the annual incidence from 1.33 to 2.47 and from 2.28 to 4.48 cases per 100,000 persons in White and Black men, respectively (1). Very little is known about the risk factors for NETs, which is due to the rarity of these cancers and the lack of large-scale prospective epidemiologic studies. A genetic etiology of NETs has been suggested in that NETs may occur as part of hereditary syndromes, such as multiple endocrine neoplasia type 1 (MEN-1; refs. 2, 3), von Hippel-Lindau syndrome (3, 4), neurofibromatosis type 1 (5, 6), tuberous sclerosis (7-9), and nonpolyposis colon cancer (10). Of these, MEN-1 syndrome, an autosomal dominant disorder characterized by parathyroid hyperplasia, pituitary adenomas, and pancreatic tumors, is the one in which NETs most frequently occur. The occurrence of familial NETs not associated with hereditary syndromes is rare as most NETs occur as nonfamilial (sporadic) tumors (2).

NETs may develop at many locations and are commonly classified according to their embryonic site of origin: the foregut (lungs, bronchi, stomach, and duodenum), the midgut (jejunum, ileum, appendix, and proximal large bowel), and the hindgut (distal colon and rectum; ref. 11). NETs from different locations may have different genetic abnormalities and may develop via different molecular pathways. Although mutation of the MEN-1 gene is common in foregut NETs, loss of chromosomes 18, 11q, and 16q is observed in midgut NETs. Furthermore, epigenetic and genetic factors may influence disease outcome, including tumor progression and patient survival (12, 13).

Because of the limited epidemiologic studies on risk factors associated with the development of NETs, we embarked on a hospital-based case-control study. This study aimed to determine whether a positive family history of any cancer and NETs in particular, affect a person's risk of developing NETs. The specific risks conferred by a positive family history of several cancers were also assessed. Risk factors were determined for NETs at 5 different sites: the small intestine, stomach, lung, pancreas, and rectum.

We conducted a retrospective hospital-based case-control study at The University of Texas M. D. Anderson Cancer Center. Briefly, we evaluated 740 patients with histologically confirmed NETs treated at the M. D. Anderson Cancer Center and 924 healthy controls. This study was reviewed and approved by the M. D. Anderson Institutional Review Board.

The study patients were identified in a search of the M. D. Anderson Patients Informatics database for all patients who had been diagnosed with NETs between 2000 and 2006. We then confirmed the pathologic diagnosis and site of the NET in each patient by manually reviewing each patients' pathology records. Tumors were classified according to disease site as small bowel (duodenum, jejunum, and ileum) with inclusion of appendiceal tumors (n = 325), stomach (n = 55), lung (n = 146), pancreas (n = 160), and rectum (n = 54) NETs.

The control subjects were healthy individuals with no past history of cancer. All controls were genetically unrelated family members, particularly spouses of patients with cancers other than pancreatic, liver, gastrointestinal, lung, or head and neck cancers (that is, cancers known to be associated with cigarette smoking, alcohol drinking, and dietary habits). Only residents of the United States who could communicate in English were included in the study. Control volunteers were recruited from the institution's Central Diagnostic Radiology Clinic, where all cancer patients and their companions are sent for the initial cancer diagnosis or posttreatment follow-up examination. Controls were personally interviewed and simultaneously enrolled in the study within the same period in which the patients with carcinoids were diagnosed (2000-2006). Information from the controls was electronically saved in a secured database at the Department of Gastrointestinal Medical Oncology.

Questionnaires were used to assess the subjects' history of smoking and alcohol use, their family history of cancer, and their personal medical history. All forms were validated by the Department of Epidemiology and stored in a computer database.

Subjects were questioned about their family history of cancer. A positive family history of cancer (ever) was defined as any first-degree or second-degree relative having a recorded cancer diagnosis. In this study, we restricted our results to first-degree relatives. For cases and controls, we collected information about all first-degree family members including parents, siblings, and offspring. Such information included age, sex, and whether the family member was alive or diseased. Information about the first-degree relatives' history of cancer, types of cancer, relationship to the subject, and age at cancer diagnosis were also collected.

The statistical analyses were conducted using Stata software (Stata Corp.). χ² tests were used to compare the distributions of categorical variables between patients with NETs and control subjects, where P < 0.05 was considered statistically significant.

We did univariate unconditional logistic regression analyses to assess the marginal effects of each factor on risk for each type of NET using maximum-likelihood estimation. We also did multivariate unconditional logistic regression analyses using all variables significant at P < 0.05 in the univariate analyses. For each factor, we calculated the adjusted odds ratio (AOR) and 95% confidence interval (CI). All AORs were adjusted for age, sex, race, and other significant risk factors using the likelihood ratio test. The final model was chosen on the basis of biological plausibility and lowest −2 log-likelihood function.

A total of 740 patients with NETs and 924 controls were analyzed. Most subjects were White (Table 1), and most were from Texas and neighboring states (Louisiana, Arkansas, Oklahoma, New Mexico). Overall, patients were younger than controls, with a mean difference in age of 5 years (P = 0.001); however, we observed no statistical difference in educational levels or gender distribution between patients with NETs at different sites. Moreover, the patients' mean age at diagnosis was similar for all types of NETs (P = 0.9) and for both sexes. The distribution of cigarette smoking and alcohol drinking in patients with NETs at different sites was not statistically significant different from controls (P > 0.05).

Table 2 shows that 74.2% of the patients and 69.2% of the controls had a family history of cancer, yielding elevated risk for NET development (P = 0.02) among those with ever family history of cancer. The significant relationship between first-degree relatives with cancer and NET development was observed in patients with NETs of the small intestine (AOR, 1.6), stomach (AOR, 2.5), lung (AOR, 2.6), and pancreas (AOR, 1.8). However, the association was not statistically significant for rectal NET development after adjusting for the subjects' age, sex, and race, with an AOR of 1.7 (95% CI, 0.7-4.4).

In patients, the most frequently reported family members with a history of cancer were the parents. A parental history of cancer was significantly associated with both pulmonary and small intestinal NETs; however, a history of cancer in siblings or offspring of NET patients was significantly associated with only pulmonary NETs. Table 2 presents the distribution of other malignant tumors in the first-degree relatives of patients with NETs and controls. A first-degree family history of esophageal cancer was significantly associated with pancreatic NETs, with an AOR of 5.6 (95% CI, 1.1-29.6). There was a 70% and 130% increased risk of developing small intestinal NETs among subjects with positive family histories of colorectal cancer and prostate cancer, respectively. Moreover, individuals with a positive family history of lung cancer had a 2-fold increase in risk of developing pulmonary NETs. A family history of gastric, pancreatic, breast, ovarian, kidney, melanoma, skin (nonmelanoma), uterine, cervical, bladder, leukemia, lymphoma, bone, hepatobiliary (liver, bile duct, and gallbladder), and brain cancers showed no significant association with the development of NETs at any site.

Five patients with NETs reported a positive family history of NETs in first-degree relatives (Table 3); none of these patients were related to each other. Four patients (two men and two women) with NETs had first-degree relatives (parents, in all cases) with NETs of the small intestine (n = 2), appendix (n = 1), and lung (n = 1). Parents of two sons and two daughters were the first-degree relatives with past history of NETs. Also, in all cases, the patients' age at diagnosis was less than their affected parents' age at diagnosis. Table 3 shows that three patients had maternal grandfathers with other cancers.

One of these five familial NETs was a 40-year-old woman, related to the MEN-1 syndrome, who reported a family history of pancreatic NETs; her brother was diagnosed at age 42 and her mother at age 51. This patient also reported a family history of other cancers, including renal cancer in her brother and her maternal grandmother and an unknown primary tumor in her maternal grandfather. None of the controls reported a positive family history of NETs among their parents, siblings, or offspring.

We did further analyses according to divisions of embryonic origin: foregut (stomach, lung, and pancreas; n = 361), midgut (small intestine and appendix; n = 325), and hindgut (rectum; n = 54). The estimated AORs (95% CI) of foregut carcinoids were 1.8 (1.3-2.6) for ever positive family history of cancer; 1.8 (1.2-2.7) for first-degree relatives; 1.6 (1.1-2.4) for parents; 1.3 (0.8-2.1) for siblings; and 2.4 (0.9-6.5) for offspring. First-degree family history of esophageal cancer and lung cancer were not associated with the development of NETs when classified by embryonic origin. The estimated AORs (95% CI) were 2.3 (0.6-9.2) and 1.3 (0.7-2.2), respectively. Positive family histories of other specific cancers among first-degree relatives were not significantly related to foregut development.

Comparing all cases (n = 740) and controls (n = 940) showed that the overall AORs (95% CI) for all NETs were 1.5 (1.2-2.1) for ever positive family history of cancer; 1.5 (1.1-2.1) for first-degree relatives; 1.4 (1.1-1.9) for parents; 1.1 (0.7-1.6) for siblings; and 1.4 (0.6-3.4) for offspring. Only first-degree family history of prostate cancer was significantly related to all NETs: AOR 1.7 (95% CI, 1.1-2.8).

This study found that a large number of patients with NETs reported positive family histories of cancer which did not significantly differ between NET sites. Having a first-degree relative with any cancer was a risk factor for the subject's development of small intestinal, gastric, pulmonary, or pancreatic NETs.

Five patients with NETs recalled that a parent had been diagnosed with NETs; however, four of these occurrences were unrelated to the MEN-1 syndrome. The association of parental NETs was found in both men and women. Our observation is supported by several anecdotal reports of familial NETs of the small intestine (14-16), stomach (17), rectum (10), and lung (18), and by two epidemiologic studies, one conducted in the United States (19) and the other in Sweden (20). The Swedish population-based study estimated a 4-fold increase in risk of developing NETs in the offspring of affected parents. Their results were later confirmed by a Mayo Clinic hospital-based study (19) in which the authors reported that 3.7% of 245 patients diagnosed with NETs had one or more first-degree relatives with the same malignancy. The prevalence of a positive family history of any cancer among first-degree relatives in our population (54.1%) was comparable to that of the Mayo study (52.7%). However, familial clusterings of NETs are rarely reported.

In addition to the four patients with familial small intestinal NETs, one Black woman with a pulmonary NET recalled that her mother was also diagnosed with a pulmonary NET, but no other cancers were reported among her other family members. Pulmonary carcinoids generally occur sporadically; however, a genetic etiology has been suggested for some pulmonary NETs in relation to the MEN-1 syndrome (21, 22). In our study, the patient and her mother were diagnosed with pulmonary NETs at age 61 and 66 years, respectively, making a diagnosis of MEN-1 syndrome unlikely because the average age at diagnosis for MEN-1 syndromes related to NETs is ≤50 years. It is possible that the occurrence of this familial case of pulmonary NETs was due to chance. This case is supported by a report from Oliveira et al. (18) on four patients (two sets of first-degree relatives) diagnosed with pulmonary NETs in the absence of MEN-1 syndrome. The authors suggested that there was a rare germ line mutation specific for the development of pulmonary NETs.

The elevated risk of developing NETs extended to individuals with a family history of other cancers (not NETs) among first-degree relatives after adjusting for the potentially confounding effects of age, sex, and race. Such association could be attributed to common genetic abnormalities (e.g., point mutation, chromosomal loss or gain, and DNA methylation) or environmental factors that may cause genetic alterations in members of the same family and may later predispose healthy family members to develop malignancies. This last notion may partially explain the significant association between a positive family history of esophageal cancer and the development of pancreatic NETs in the current study. It has been suggested that P16/^INK4A and P14/^ARF genetic hypermethylations are important events in the carcinogenesis of squamous cell carcinoma, esophageal adenocarcinoma (23-25), and pancreatic NETs (26). Moreover, hypermethylation of the P16/^INK4A gene was observed in 64% of the family members of patients with esophageal squamous cell carcinoma (25). This may raise the question of whether family members with pancreatic NETs shared common environmental risk factors that lead to multiple-organ or single-organ precancerous conditions such as chronic inflammation, which is commonly associated with epigenetic phenomena like P16/^INK4A and P14/^ARF genetic hypermethylations.

The presence of NETs and other cancers in first-degree relatives may be consistent with several inherited cancer syndromes in which genetic mutations are associated with both NETs and cancers of other types, such as hereditary nonpolyposis colorectal cancer and small intestinal NETs (14). Moreover, the risk of patients with small intestinal NETs to later develop colorectal carcinoma is three times higher than what would be expected in the general population (19).

We observed 2-fold increases in the risk of small intestinal NET development in relation to the presence of a family history of prostate cancer among a patient's first-degree relatives. This observation was supported by previous findings of the subsequent development of prostate cancer among patients with small intestinal NETs, with AORs of 2.3 (95% CI, 1.2-4.1; ref. 27) and 2.7 (95% CI, 1.6-3.9; ref. 20). In addition, the coexistence of prostatic adenocarcinoma and a small intestinal NET in two patients was previously described (28, 29). It is not clear why patients with small intestinal NETs have a high tendency to develop prostate cancer. One explanation is the high tendency of patients with NETs to have a positive family history of cancer as compared with healthy individuals. Thus, a possible positive family history of prostate cancer among patients with NETs may increase their likelihood for developing prostate cancer during their lifetime (30-32). This hypothesis may be supported by the results of the Swedish familial study (20) in which a significant risk of developing prostate cancer was observed during long-term follow-up of the subjects.

Our results disagreed with the Mayo study (19), which found no association between a positive family history of colorectal or prostate cancer and the development of small intestinal NETs. However, it is possible that those null results are attributable to underestimation, which may have been due to the investigators' comparing observed rates among patients at a cancer referral center with the expected rate of incidence in the general population, estimated from the Surveillance, Epidemiology, and End Results program. In our study, however, controls were comparable to the study base from which all patients were diagnosed. Moreover, the estimated age-specific prevalence of a family history of lung, colorectal, ovarian, breast, and prostate cancer in our controls was comparable to the race- and age-specific prevalence estimated from the National Health Interview Survey of households in the United States (33).

Our study is the largest case-control study to date to assess the risk of a positive family history of different cancers and the development of NETs at five anatomic locations after adjusting for the potentially confounding effects of age, sex, and race. Family history of cancers among first degree-relatives of cases was confirmed from physician's notes in patients' medical records and from baseline epidemiologic assessment at the initial visit to the M. D. Anderson Cancer Center. Controls were personally interviewed for the same information using the same questionnaire form.

We considered the embryonic classification of carcinoid tumors. However, overall association between first-degree family histories of esophageal and lung cancers were not statistically significant with foregut NETs. This study had some limitations due to the source of subjects (hospital based). We believe that this limitation was also a strength in that we had the ability to review the patients' medical records to confirm pathologic diagnoses and validate the subjects' questionnaire responses.

It is possible that the accuracy of reporting cancers among affected family members may vary by type of cancer. However, we do not believe that such accuracy is different between cases and controls and the effect of any misclassification will most likely be attenuated (bias toward the null).

In summary, most NETs occur sporadically, regardless of disease site. However, we found evidence that a positive family history of cancer is associated with the risk of developing NETs that did not arise in the context of other hereditary syndromes. We support the establishment of a national consortium to investigate environmental and genetic risk factors for NETs by tumor site and to investigate possible gene-environment interactions in patients from different racial groups. Such analyses may help us to better understand the epidemiology of NETs and provide a basic foundation for developing prevention strategies among individuals at high risk.