Risk of second malignant neoplasms among lymphoma patients with a family history of cancer

Abstract

Radiotherapy and chemotherapy are known risk factors for second cancers after lymphoma. The role of genetic influences, however, remains largely unknown. We assessed risk of second cancers associated with family history of any cancer in 41,181 patients with Hodgkin lymphoma (HL) (n = 7,476), non-Hodgkin lymphoma (NHL) (n = 25,941), or chronic lymphocytic leukemia (CLL) (n = 7,764), using a large population-based database. Family history of cancer was based on a diagnosis of any cancer in 110,862 first-degree relatives. We found increased relative risk (RR) (1.81, 95% confidence interval (CI): 1.04–3.16) of breast cancer among HL patient with positive (vs. negative) family history of cancer. Among CLL patients with positive (vs. negative) family history of cancer, we observed elevated risks of bladder (RR = 3.53, 95% CI: 1.31–9.55) and prostate cancer (RR = 2.15, 95% CI: 1.17–3.94). For NHL patients with positive (vs. negative) family history of cancer, we observed non-significantly increased risk of non-melanoma skin cancer (RR = 1.94, 95% CI: 0.86–4.38) and lung cancer (RR = 1.99, 95% CI: 0.73–5.39). Our observations suggest that genetic factors, as measured by positive family history of cancer, may be influential risk-factors for selected second tumors following lymphoproliferative disorders. © 2006 Wiley-Liss, Inc.

Therapeutic advances in the treatment of Hodgkin lymphoma (HLL) have translated into current 5-year relative survival rates of almost 85%.1 These improvements in survival, however, have been accompanied by serious late sequelae of successful treatment,2 including the occurrence of new primary cancers. Second malignant neoplasms now comprise the leading cause of death among long-term survivors of HL.3, 4 We recently provided estimates of the cumulative absolute risk of breast cancer among young women treated for HL,5 based on treatment, age at therapy and length of follow-up. Limitations of the underlying dataset,6, 7 however, precluded inclusion of family history of breast cancer in the prediction models.5 In a related report,8 risk of breast cancer following chest radiotherapy for HL in young women did not appear to be modified by family history of breast or ovarian cancer, but data were admittedly incomplete and largely self-reported. In contrast, Nichols et al.9 found an association between family history of cancer and risk of second neoplasms among HL patients. Our current survey uses information on HL patients from the Swedish and Danish population-based cancer registries included in our recent study4 and cancer diagnosis among their first-degree relatives, to examine the association of family history with second malignant neoplasms among HL patients. We also evaluate, for the first time, the effect of family history on the risk of second cancers among non-Hodgkin lymphoma (NHL) and chronic lymphocytic leukemia (CLL) patients.

Material and methods

Data resources used in this study have been described in detail elsewhere.10, 11, 12 Briefly, the Swedish Multi-generation Registry was merged with the Swedish Cancer Registry (all cancers 1958–1998) to identify all Swedish lymphoma cases and their linkable first-degree relatives (alive in 1932 or born ≥1932), including all cancer diagnoses among relatives. A similar database was created using the Danish Cancer Registry (all cancers 1968–1997) and the Danish Central Population Registry (CPR).10, 11, 13, 14

We studied selected second neoplasms, including tumors that are relatively common in the general population and those for which significantly increased risks have been consistently reported after HL, NHL or CLL2, 15, 16, 17 (Table II). Positive family history of cancer denotes any cancer in one or more first-degree family members. To compare risks of second cancers in HL, NHL and CLL patients with positive (vs. negative) family history of cancer, we used Cox proportional hazards model18 fitted on the age scale, adjusted for gender. Follow-up time started at age at diagnosis with lymphoma and ended at age of development of a malignancy, or at censoring, defined as end of study (1998) or age at death, whichever came first.

Approval was obtained from the NIH institutional review board for these studies. Informed consent was waived because we had no contact with study subjects.

Results

A total of 7,476 HL, 25,941 NHL and 7,764 CLL patients, and 110,862 first-degree relatives were evaluated (Table I). Compared to HL patients with a negative family history of cancer, we found increased risk of breast cancer among HL patients with positive family history of cancer (RR = 1.81, 95% CI: 1.04–3.16) (Table II). In a subanalysis stratified by age at HL diagnosis, we observed a 1.59-fold (95% CI: 0.86–2.93; 14 with vs. 38 without family history of cancer) and a 3.11-fold (95% CI: 0.89–10.91; 3 with vs. 15 without family history of cancer) elevated risk for breast cancer among HL patients diagnosed at the age <45 years and ≥45 years, respectively.

Table I. Characteristics of Swedish and Danish Patients and First-Degree Relatives

Variable	HL	NHL	CLL
HL, Hodgkin lymphoma; NHL, non-Hodgkin lymphoma; CLL, chronic lymphocytic leukemia. 1 Values in square brackets indicate interquartile ranges. 2 Values in parentheses indicate percentages. 3 Values in flower braces indicate mean number per probands.
Total number of patients	7,476	25,941	7,764
Age at diagnosis, median	39 [25–57]1	62 [50–72]	67 [59–75]
Gender, n
Male	4,515 (60)2	15,089 (58)	5,141 (66)
Female	2,961 (40)	10,852 (42)	2,623 (34)
Calendar year at diagnosis, median	1982 [1973–1990]	1988 [1980–1993]	1986 [1979–1993]
First-degree relatives, n
Parents	6,444 {0.9}3	10,581 {0.4}	1,222 {0.2}
Siblings	5,190 {0.7}	8,998 {0.3}	1,010 {0.1}
Offspring	11,502 {1.5}	50,156 {1.9}	15,759 {2.0}

Table II. Relative Risk of Second Malignant Neoplasm Among Swedish and Danish Patients with Lymphoid Malignancies in Relation to Family History of Cancer1

Second neoplasm	Hodgkin lymphoma			Non-Hodgkin lymphoma			Chronic lymphocytic leukemia
	FH−	FH+	RR2	FH−	FH+	RR2	FH−	FH+	RR2
FH+, positive family history of cancer; FH−, no family history of cancer; RR, relative risk; CI, confidence interval; Gastrointestinal tumors include esophagus, stomach, small intestine, colon, rectum, liver, gallbladder, and pancreas. 1 Diagnoses of squamous and/or basal cell skin lesions were not counted as cancer; estimates were virtually unchanged when diagnoses of squamous and/or basal cell skin lesions were included as cancer. 2 Models adjusted for gender. 3 Total number of lymphoma cases with positive versus no family history of cancer. 4 Values in parentheses indicate 95% CIs. 5 Undetermined. Italic entries have p-values <0.05. Among affected patients who developed a second malignant neoplasm a positive family history of cancer included the following cancer sites in first-degree relatives:Hodgkin lymphoma patients who developed a secondary breast cancer (n = 17): Pt. no. 1 colon; Pt. no. 2 prostate; Pt. no. 3 liver; Pt. no. 4 melanoma; Pt. no. 5 lung; Pt. no. 6 one relative with prostate and one with breast; Pt. no. 7 ovary; Pt. no. 8 bladder; Pt. no. 9 mycosis fungoides; Pt. no. 10 melanoma; Pt. no. 11 one relative with ovary and one with brain; Pt. no. 12 Non-Hodgkin lymphoma; Pt. no. 13 kidney; Pt. no. 14 lung; Pt. no. 15 melanoma; Pt. no. 16 testis; Pt. no. 17 breast. Non-Hodgkin lymphoma patients who developed a secondary lung cancer (n = 5): Pt. no. 1 breast; Pt. no. 2 testis; Pt. no. 3 cervix; Pt. no. 4 tonsil; Pt. no. 5 breast. Non-Hodgkin lymphoma patients who developed a secondary nonmelanoma skin cancer (n = 7): Pt. no. 1 melanoma; Pt. no. 2 testis; Pt. no. 3 cervix; Pt. no. 4 acute lymphocytic leukemia; Pt. no. 5 cervix; Pt. no. 6 carcinoma of skin; Pt. no. 7 breast. Chronic lymphocytic leukemia patients who developed a secondary bladder cancer (n = 7): Pt. no. 1 bladder; Pt. no. 2 non-Hodgkin lymphoma; Pt. no. 3 one relative with thyroid and one with melanoma; Pt. no. 4 stomach; Pt. no. 5 breast; Pt. no. 6 bone; Pt. no. 7 one relative with non-Hodgkin lymphoma and one with testis. Chronic lymphocytic leukemia patients who developed a secondary prostate cancer (n = 13): Pt. no. 1 breast; Pt. no. 2 one relative with rectum and one with breast; Pt. no. 3 one relative with ovary and one with breast; Pt. no. 4 breast; Pt. no. 5 melanoma; Pt. no. 6 breast; Pt. no. 7 melanoma; Pt. no. 8 one relative with brain and one with stomach; Pt. no. 9 one relative with stomach and one with breast; Pt. no. 10 one relative with colorectal and one with lung; Pt. no. 11 melanoma; Pt. no. 12 ovary; Pt. no. 13 pituitary.
n3	6,391	1,085		25,509	432		7,086	678
Breast cancer	53	17	1.81 (1.04–3.16)4	109	2	1.28 (0.31–5.26)	27	3	1.10 (0.33–3.67)
Lung cancer	59	6	0.90 (0.38–2.11)	175	5	1.99 (0.73–5.39)	97	9	1.22 (0.56–2.65)
Gastrointestinal tumors	79	7	0.67 (0.30–1.46)	349	4	1.06 (0.39–2.84)	115	15	1.52 (0.81–2.84)
Bladder cancer	14	0	–5	98	0	–5	28	7	3.53 (1.31–9.55)
Kidney cancer	18	1	0.46 (0.06–3.58)	74	0	–5	14	2	2.50 (0.54–11.52)
Prostate cancer	30	5	1.29 (0.39–4.24)	191	0	–5	96	13	2.15 (1.17–3.94)
Malignant melanoma	18	4	1.49 (0.49–4.55)	50	1	1.27 (0.17–9.64)	23	2	1.19 (0.28–5.13)
Nonmelanoma skin cancer	68	6	0.53 (0.23–1.24)	278	7	1.94 (0.86–4.38)	173	18	1.41 (0.87–2.29)

Among NHL patients with positive (vs. negative) family history of cancer, nonsignificant excesses of nonmelanoma skin cancer (NMSC) (RR = 1.94, 95% CI: 0.86–4.38) and lung cancer (RR = 1.99, 95% CI; 0.73–5.39) were observed.

In CLL patients, excess cancers of bladder (RR = 3.53, 95% CI: 1.31–9.55) and prostate (RR = 2.15, 95% CI: 1.17–3.94) were observed among patients with positive (vs. negative) family history of cancer.

Discussion

The amplified risks of breast cancer among HL patients with positive family history of cancer suggest the operation of shared genetic factors (e.g., DNA repair19), environmental factors or their interaction. By extrapolating our findings from a prior case-control study,20 it can be estimated that ∼70% of HL patients may have received chest radiotherapy. Although some evidence suggests that family history of breast or ovarian cancer may influence response to radiotherapy, these data were too sparse to permit a detailed examination of the influence of cancer types in first degree relatives on breast cancer risk after HL.8

Increased risks of lung cancer in NHL survivors have been well established.15, 21 We found 2-fold nonsignificant increased risk of lung cancer in NHL patients with positive (vs. negative) family history of cancer. Although radiotherapy20, 22, 23 and alkylating agent chemotherapy20 for HL have been associated with increased risks of lung cancer, with tobacco use multiplying all treatment-related excesses,20, 23 similar analytic studies in NHL patients have not been conducted. NHL is not typically regarded as smoking-related,24 although more recent analyses suggest that certain subtypes may be linked with tobacco use.25, 26 Thus, positive family history of cancer may be a surrogate marker of a reduced inherent ability to repair damage induced by exogenous carcinogens, such as radiotherapy, chemotherapy or tobacco, although the observed association could also be due to shared environmental risk-factors or a combination of influences. Our finding of almost 2-fold increased risks of NMSC associated with positive family history of cancer in NHL survivors is supported by the results of Nugent et al.27

Past treatment approaches to CLL have typically included chlorambucil and less frequently, cyclophosphamide-based regimens.28 Whether positive family history of cancer is a marker of an impaired ability to metabolize these types of cytotoxic drugs is unknown. Cyclophosphamide is a human bladder carcinogen with increasing cumulative dose strongly related to increasing risk.29 More recent approaches to CLL treatment include the use of nucleoside analogs, which have been associated with increased risks of second cancers in some,30 but not all31 studies. The etiology of prostate cancer remains largely unknown. It is of interest that chlorambucil, an established human carcinogen32 can penetrate prostate tissue and was recently found to reinduce hormone sensitivity following failure of conventional prostate cancer chemotherapy.33

The strengths of this study are the large sample size and unbiased assessment of cancer status in relatives. Limitations include lack of information on clinical/treatment and risk-factors, and underreporting of positive family history of cancer because of inherent database truncations.10, 11, 13, 34 Although it would be of interest to match on site of second cancer and family history for the same cancer type, our numbers were not large enough for subset analyses, despite a large population base.

Nonetheless, our results suggest that the role of genetic factors in the occurrence of selected solid tumors in patients following treatment for lymphoproliferative disorders should be further explored. Given that cancer therapy is one of the few settings in which humans are exposed to carefully measured amounts of potential carcinogens, this provides a unique opportunity to explore the role of gene-environment interactions.35 Expansion of our knowledge base could help define high-risk groups of patients for intensified surveillance efforts. Moreover, access to better predictive markers will provide a unique and important opportunity for personalized medicine, with clinical management and treatment strategies tailored to the individual.

Acknowledgements

Dr. O. Landgren, Dr. R.M. Pfeiffer, Ms. L. Stewart and Dr. L.B. Travis designed the study. Ms. G. Gridley, Dr. L. Mellemkjaer, Dr. K. Hemminki and Dr. L.R. Goldin obtained data. Dr. O. Landgren, Dr. R.M. Pfeiffer and Ms. L. Stewart analyzed data. Dr. O. Landgren, Ms. L. Stewart and Dr. L.B. Travis initiated this work. Dr. O. Landgren wrote the report. All authors were involved in the interpretation of the results, read, gave comments and approved the final version of the manuscript. Dr. O. Landgren, Dr. R.M. Pfeiffer and Ms. L. Stewart had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Each author declares that he/she has no conflict of interests relevant to this paper. The authors thank Ms. Emily Steplowski, Information Management Services, Silver Spring, MD, for important efforts in setting up this complex database, and Dr. B.J. Stone for her consultation in creating numerous variables.