Abstract
A family history of non–Hodgkin's lymphoma (NHL) confers increased risk of NHL, but it is unknown whether the excess risk in males and females varies by the sex or kinship of the affected relative. We linked nationwide Swedish registries to identify parents and siblings of NHL patients who developed NHL between January 1, 1961 and December 31, 2002. In males, parental risks were approximately the same from fathers and mothers, whereas sibling risks were higher from brothers [standardized incidence ratio (SIR), 1.8; 95% confidence interval (95% CI), 1.0-2.9] than sisters (SIR, 0.9; 95% CI, 0.2-1.9). In females, parental and sibling risks were higher from same-sex relatives (SIR from mothers, 1.9; 95% CI, 1.2-2.7; SIR from sisters, 6.3; 95% CI, 4.0-9.3) than from opposite-sex relatives (SIR from fathers, 1.2; 95% CI, 0.7-1.9; SIR from brothers, 0.7; 95% CI, 0.2-1.6). These findings did not vary substantially by the age of diagnosis of the offspring. Risk of NHL in offspring was also increased among those with a parent diagnosed with multiple myeloma or leukemia. The relative risk of NHL among those with a parent diagnosed with any hematopoietic cancer was 1.5 (95% CI, 1.4-1.7) and that for having a sibling with any hematopoietic cancer was also 1.5 (95% CI, 1.2-1.9). Our results suggest that part of the familial risk of NHL may be attributable to shared environmental exposures, particularly between same-sex siblings. (Cancer Epidemiol Biomarkers Prev 2007;16(11):2496–9)
Introduction
It is well established that non–Hodgkin's lymphoma (NHL) aggregates in families, conferring ∼2-fold excess risk of NHL in first-degree relatives (1-12). Familial risk is typically ascribed to inherited genetic cofactors, although shared environmental causes can also result in familial clustering. Sex differences in familial risk of NHL may be informative about modes of inheritance, particularly if a susceptibility gene is linked to the X or Y chromosome or if, due to genomic imprinting, such a gene is expressed differentially when inherited from the mother versus the father (13). Although a 2:1 male to female ratio in NHL incidence is consistently documented (14), sex- and kindred-specific familial risks of NHL are not well studied. Therefore, we investigated the excess risk of NHL in first-degree relatives of male and female probands in the nationwide Swedish Family-Cancer Database, the largest familial data set of incident cancers.
Materials and Methods
The Swedish Family-Cancer Database is a linked product of four registries: the Multi-Generation Register, Swedish Cancer Registry, national census data, and death notification database. In the early 1990s, Statistics Sweden created the Multi-Generation Register by linking data from several different population-based registries (15). About 7.3 million children born in Sweden in 1932 or later are registered with both biological parents in the Multi-Generation Register. The population-based Swedish Cancer Registry, established in 1958, contains individual data on newly diagnosed malignant tumors within Sweden, with reporting of ∼100% of diagnoses and histologic verification of 97% of tumors (16). We used International Classification of Disease, 7th revision codes 200 and 202 to identify NHL. Information on histopathology (Systemized Nomenclature of Medicine code) has been recorded in Swedish Cancer Registry from 1993 and is subsequently missing for 41% of NHL cases among offspring and 63% of NHL cases among parents. Therefore, we did not assess the risks of histopathology-specific subtypes of NHL. As there were some inaccuracies in the vital status determinations in the first years of cancer registration, cancers in our study were diagnosed between 1961 and 2002. In total, 51,363 offspring had a parental history of NHL, contributing 1,874,039 person-years of follow-up.
Standardized incidence ratios (SIR) were calculated to measure the relative risk of NHL in offspring according to the occurrence of NHL in their family (i.e., using parents or siblings as probands). Follow-up was started for each offspring at birth or January 1, 1961, whichever came later, independently of proband diagnosis. Follow-up ended at diagnosis of first cancer, death, emigration, or December 31, 2002. SIRs were calculated as the ratio of observed to expected number of cases. The expected numbers of cancers were obtained using age-specific (5-year bands), sex-specific, and calendar year–specific (10-year bands) rates as prevailed in the corresponding general population in our database. Parents' ages were not limited but offspring were 0 to 70 years of age. Confidence intervals (95% CI) were calculated assuming a Poisson distribution (17). Risks for siblings were calculated using the cohort method, described elsewhere (18).
Results
In the database, there were 23,390 parents (13,431 fathers and 9,959 mothers) and 6,141 offspring (3,839 sons and 2,302 daughters) with NHL. The median age at diagnosis was 66 years for parents and 47 years for offspring. Altogether, we identified 168 NHL patients with a parent or a sibling with the same cancer. Overall, familial risks of NHL were significantly increased in persons with a parental or a sibling history of NHL. Table 1 shows the sex-specific SIRs of NHL in offspring with a parent or a sibling affected with NHL. In males, parental risks were approximately the same from fathers and mothers, whereas sibling risks were higher from brothers (SIR, 1.8; 95% CI, 1.0-2.9) than sisters (SIR, 0.9; 95% CI, 0.2-1.9). In females, parental as well as sibling risks were higher from the same-sex proband than the opposite-sex proband, with a statistically significant difference between SIRs from sisters (SIR, 6.3; 95% CI, 4.0-9.3) versus brothers (SIR, 0.7; 95% CI, 0.2-1.6).
Familial risks in NHL, estimated by the SIR
Person at risk . | Proband . | Observed . | Expected . | SIR* (95% CI) . |
---|---|---|---|---|
Males | Father | 41 | 22.3 | 1.8 (1.3-2.4) |
Mother | 41 | 20.2 | 2.0 (1.5-2.7) | |
Brother | 14 | 7.6 | 1.8 (1.0-2.9) | |
Sister | 4 | 4.7 | 0.9 (0.2-1.9) | |
Male (father or brother) | 55 | 29.9 | 1.8 (1.4-2.3) | |
Female (mother or sister) | 45 | 24.9 | 1.8 (1.3-2.4) | |
Parent (father or mother) | 81 | 42.2 | 1.9 (1.5-2.4) | |
Sibling (brother or sister) | 18 | 12.6 | 1.4 (0.9-2.2) | |
Females | Father | 18 | 14.4 | 1.2 (0.7-1.9) |
Mother | 25 | 13.0 | 1.9 (1.2-2.7) | |
Brother | 4 | 5.4 | 0.7 (0.2-1.6) | |
Sister | 22 | 3.5 | 6.3 (4.0-9.3) | |
Male (father or brother) | 22 | 19.9 | 1.1 (0.7-1.6) | |
Female (mother or sister) | 47 | 16.5 | 2.8 (2.1-3.7) | |
Parent (father or mother) | 43 | 27.3 | 1.6 (1.1-2.1) | |
Sibling (brother or sister) | 26 | 9.0 | 2.9 (1.9-4.1) | |
Both | Father | 59 | 36.8 | 1.6 (1.2-2.0) |
Mother | 66 | 33.2 | 2.0 (1.5-2.5) | |
Brother | 18 | 13.3 | 1.4 (0.8-2.1) | |
Sister | 26 | 8.2 | 3.2 (2.1-4.5) | |
Male (father or brother) | 77 | 50.0 | 1.5 (1.2-1.9) | |
Female (mother or sister) | 92 | 41.4 | 2.2 (1.8-2.7) | |
Parent (father or mother) | 124 | 69.4 | 1.8 (1.5-2.1) | |
Sibling (brother or sister) | 44 | 21.5 | 2.0 (1.5-2.7) |
Person at risk . | Proband . | Observed . | Expected . | SIR* (95% CI) . |
---|---|---|---|---|
Males | Father | 41 | 22.3 | 1.8 (1.3-2.4) |
Mother | 41 | 20.2 | 2.0 (1.5-2.7) | |
Brother | 14 | 7.6 | 1.8 (1.0-2.9) | |
Sister | 4 | 4.7 | 0.9 (0.2-1.9) | |
Male (father or brother) | 55 | 29.9 | 1.8 (1.4-2.3) | |
Female (mother or sister) | 45 | 24.9 | 1.8 (1.3-2.4) | |
Parent (father or mother) | 81 | 42.2 | 1.9 (1.5-2.4) | |
Sibling (brother or sister) | 18 | 12.6 | 1.4 (0.9-2.2) | |
Females | Father | 18 | 14.4 | 1.2 (0.7-1.9) |
Mother | 25 | 13.0 | 1.9 (1.2-2.7) | |
Brother | 4 | 5.4 | 0.7 (0.2-1.6) | |
Sister | 22 | 3.5 | 6.3 (4.0-9.3) | |
Male (father or brother) | 22 | 19.9 | 1.1 (0.7-1.6) | |
Female (mother or sister) | 47 | 16.5 | 2.8 (2.1-3.7) | |
Parent (father or mother) | 43 | 27.3 | 1.6 (1.1-2.1) | |
Sibling (brother or sister) | 26 | 9.0 | 2.9 (1.9-4.1) | |
Both | Father | 59 | 36.8 | 1.6 (1.2-2.0) |
Mother | 66 | 33.2 | 2.0 (1.5-2.5) | |
Brother | 18 | 13.3 | 1.4 (0.8-2.1) | |
Sister | 26 | 8.2 | 3.2 (2.1-4.5) | |
Male (father or brother) | 77 | 50.0 | 1.5 (1.2-1.9) | |
Female (mother or sister) | 92 | 41.4 | 2.2 (1.8-2.7) | |
Parent (father or mother) | 124 | 69.4 | 1.8 (1.5-2.1) | |
Sibling (brother or sister) | 44 | 21.5 | 2.0 (1.5-2.7) |
Ratio of observed to expected, with 95% CIs.
Because the median age at diagnosis in offspring was relatively young, we investigated whether the pattern of familial risks varied by age group. The overall association with parental or sibling history of NHL did not vary by age at diagnosis in offspring. The SIR by parental history was 1.7 (95% CI, 1.3-2.2; n = 61) for offspring diagnosed before age 50 years and 1.8 (95% CI, 1.4-2.3; n = 63) for offspring diagnosed between ages 50 and 70 years. Sibling relative risks were 2.0 (95% CI, 1.2-3.1; n = 18) for those diagnosed before 50 years and 2.1 (95% CI, 1.4-3.0; n = 26) for those diagnosed at 50 to 70 years. We did not have adequate statistical power to study all sex-specific associations for different diagnostic age groups. However, the strongest relative risk estimate from Table 1—that for risk of NHL among females with sisters with the same malignancy—was not modified by age, with a SIR of 6.1 (95% CI, 2.6-11.1; n = 8) for females diagnosed before age 50 and 6.4 (95% CI, 3.5-10.3; n = 14) for those diagnosed later.
We analyzed the risk of NHL among offspring in association with any hematopoietic cancer in parents and siblings (Table 2). In addition to NHL, associations were noted between NHL in offspring and multiple myeloma or leukemia in parents. The overall familial risks of NHL among offspring by any hematopoietic cancer were identical from parent and sibling, with statistically significant SIRs of 1.5.
Familial risks of NHL in offspring with a parent or sibling with hematopoietic cancer, estimated by the SIR
Proband . | Cancer site in proband . | Observed . | Expected . | SIR* (95% CI) . |
---|---|---|---|---|
Parent | NHL | 124 | 69.4 | 1.8 (1.5-2.1) |
Hodgkin lymphoma | 14 | 11.7 | 1.2 (0.7-1.9) | |
Multiple myeloma | 60 | 41.2 | 1.5 (1.1-1.8) | |
Leukemia | 102 | 74.4 | 1.4 (1.1-1.6) | |
All hematopoietic cancers | 300 | 196.7 | 1.5 (1.4-1.7) | |
Sibling | NHL | 44 | 21.5 | 2.0 (1.5-2.7) |
Hodgkin lymphoma | 11 | 6.8 | 1.6 (0.8-2.7) | |
Multiple myeloma | 6 | 5.5 | 1.1 (0.4-2.1) | |
Leukemia | 17 | 17.2 | 1.0 (0.6-1.5) | |
All hematopoietic cancers | 78 | 51.0 | 1.5 (1.2-1.9) |
Proband . | Cancer site in proband . | Observed . | Expected . | SIR* (95% CI) . |
---|---|---|---|---|
Parent | NHL | 124 | 69.4 | 1.8 (1.5-2.1) |
Hodgkin lymphoma | 14 | 11.7 | 1.2 (0.7-1.9) | |
Multiple myeloma | 60 | 41.2 | 1.5 (1.1-1.8) | |
Leukemia | 102 | 74.4 | 1.4 (1.1-1.6) | |
All hematopoietic cancers | 300 | 196.7 | 1.5 (1.4-1.7) | |
Sibling | NHL | 44 | 21.5 | 2.0 (1.5-2.7) |
Hodgkin lymphoma | 11 | 6.8 | 1.6 (0.8-2.7) | |
Multiple myeloma | 6 | 5.5 | 1.1 (0.4-2.1) | |
Leukemia | 17 | 17.2 | 1.0 (0.6-1.5) | |
All hematopoietic cancers | 78 | 51.0 | 1.5 (1.2-1.9) |
Ratio of observed to expected, with 95% CIs.
Discussion
We observed an excess risk of NHL in males with a father, mother, or brother with NHL and in females with a mother or sister with NHL. The increased risk in siblings and parent-child pairs of the same sex, as well as between mothers and sons, suggests that shared environmental risk factors between family members account for at least part of the 2-fold excess risk of NHL in families. Similar environmental exposures in same-sex family members may result from the tendency of sisters and brothers to engage in sex-typed activities (19, 20), which have been found to persist from early childhood through middle childhood to early adolescence (21, 22). Such activities could lead to shared infections, dietary patterns, sunlight exposure, or other environmental or behavioral factors, which may in turn affect NHL risk, among family members who tend to spend more time together. If this model is correct, then early-life exposures may be important to later NHL development. Our results provide no clear evidence of sex-linked genetic susceptibility to NHL. A sex-linked genetic mechanism might be suggested, for example, by shared risk only between mothers and sons (consistent with a recessive X-linked trait) or only between fathers and sons (consistent with a Y-linked trait).
Our findings differ somewhat from those recently reported by the International Lymphoma Epidemiology Consortium (InterLymph), in which the relative risk of NHL was higher in males and females with a self-reported history of NHL in a father or brother than a mother or sister, respectively (23). However, those authors also concluded that shared environmental risk factors are involved in familial NHL. In a previous study in the Swedish Family-Cancer Database (24), as well as our recent population-based case-control study of lymphomas (11), there was no difference in the relative risk of NHL between male and female offspring of a father, mother, or either parent with NHL or hematopoietic malignancy. However, one study found that familial clustering was stronger in females (10), whereas another observed stronger aggregation in male relatives (8). In addition, we and others previously found a stronger risk association with NHL in siblings than in parents (4, 6, 8, 10, 11), again suggesting that shared environmental exposures in childhood and adolescence may account for part of the familial risk of NHL. Indeed, we estimated that only 10% of susceptibility to NHL is attributable to genetic effects (25).
The analysis of NHL risk among offspring with a parent or sibling with another hematopoietic malignancy revealed previously reported associations of NHL with multiple myeloma (26) and leukemia (23). The overall familial relative risk of NHL associated with having a relative with any hematopoietic cancer in our study was 1.5, which is identical to the estimate recently reported by InterLymph (23). We did not observe any sex-specific differences in these discordant site analyses (data not shown), although our analysis of sibling risks was hampered by small numbers.
Strengths of our study include its large size, virtually complete coverage of the Swedish population (15, 16), and basis on validated familial NHL diagnoses. In contrast, self-reported family history is subject to bias (27), with reliability varying by case-control status as well as several other characteristics, including age, sex, birth order, and recency of a family member's diagnosis. Therefore, differences in these factors between cases and controls in studies relying on self-reported familial data could affect the odds ratio in unpredictable ways. On the other hand, despite the 41 years of follow-up in the database, we observed a limited number of NHL diagnoses among offspring or siblings of NHL probands. We also lacked histopathologic information for a large proportion of cases, preventing us from examining familial risk of NHL subtypes. In addition, given that the median age at diagnosis among offspring in our study was substantially below the average for NHL in general, we cannot exclude the possibility that some sex-specific patterns of familial risks are different for NHL diagnosed at a young age. Still, in our analysis, there was no implication of modification of familial risks by age at diagnosis in offspring.
In conclusion, our study provides persuasive evidence that environmental risk factors are at play in the development of NHL within families. There are few well-established environmental causes of NHL, and there remains no cogent explanation for the steep rise in the international incidence of NHL during the second half of the 20th century (14, 28). Therefore, further investigation into shared exposures between family members may help to illuminate the causes of both familial and nonfamilial NHL.
Grant support: Swedish Cancer Foundation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.