Factors related to immune function, certain infections and specific chemical agents, have been associated with non-Hodgkin lymphoma (NHL).1 We previously reported an increased incidence of NHL among long-term users of tricyclic antidepressant medication in a population-based cohort of more than 30,000 users of antidepressant medications, with a standardized incidence rate ratio (IRR) of 2.5 (95% confidence interval [CI] = 1.4–4.2) among those with 5 or more prescriptions.2 Although another population-based study did not confirm our observation, overall,3 it did suggest a possible excess of NHL with tricyclic antidepressant medication among the long-term users. Herein, we report an update of our population-based cohort, examining the incidence of NHL in more than 43,000 users of antidepressant medications; the sample size has increased from 97,237 person-years at risk and 24 cases in the previous study to 264,489 person-years at risk and 92 cases.
Using the nationwide Danish Central Population Register and the Danish Cancer Registry, we identified all 354,551 inhabitants of North Jutland County who were 30 years or older during the period 1 January 1989 to 31 December 2003 and who had no history of cancer (except nonmelanoma skin cancer) prior to study entry (1989 or age 30 years).
The population-based North Jutland Prescription Database established in 19894 retains key information on prescriptions for refundable drugs dispensed from all pharmacies in the county. This includes the personal identification number of the customer, type of drug dispensed according to the Anatomic Therapeutical Chemical (ATC) classification system,5 and the date of dispensing at the pharmacy. The 10-digit personal identification number (which encodes sex and date of birth) ensured that a complete prescription history could be established for each individual. We identified 43,932 individuals who redeemed at least 2 prescriptions for antidepressant medications (ATC code N06A except N06AX12) between 1 January 1989 and 31 December 2003 and who were free of cancer at date of second prescription. Throughout the study period and consistent with our previous study antidepressants were classified as tricyclic antidepressants (ATC codes N06AA0 and N06AA1), selective serotonin reuptake inhibitors (SSRI; N06AB and N06AE02) or “other” antidepressants.
We also obtained prescription data for immunosuppressive drugs (methotrexate, azathioprine, cyclosporine, cyclophosphamide, chlorambucil, penicillamin, aurothiomalat, and auranofin; ATC codes L04AX03, L04AX01, L04AA01, L01AA01, L01AA02, L01BA01, M01CB01, M01CB03, and M01CC01), oral corticosteroid treatment (H02AB) and nonsteroidal anti-inflammatory drugs (NSAIDs; M01A) that might act as confounders of the association between antidepressant medications and risk of NHL.1,6,7
The North Jutland Prescription Database does not contain information on medications given during hospitalizations. We did, however, identify all persons who had an inpatient or outpatient contact for psychiatric disorders. This was done by linking our cohort to the nationwide Central Psychiatric Registry, which contains information on all psychiatric admissions since 1969.8
We identified cancer cases through the nationwide Danish Cancer Registry. This Registry has recorded incident cases of cancer on a nationwide basis since 1943, classified according to a modified Danish version of the International Classification of Diseases version 7 (ICD-7),9 to determine occurrence of NHL (ICD-7 codes 200 and 202). Follow-up was from 1 January 1989 or age 30 years, whichever occurred later, and continued until the date of a NHL diagnosis, another primary cancer (except nonmelanoma skin cancer), death, emigration from North Jutland or 31 December 2003, whichever came first. The person-time of the study subjects was classified according to use of antidepressant medication in exposed time (≥2 prescriptions for antidepressants) and unexposed time (<2 prescriptions for antidepressants).
Log-linear Poisson regression analysis was used to compute incidence rate ratios (IRRs) adjusted for calendar period (1989–1993, 1994–1998, 1999–2003), sex, age (10-year age-groups), and filling of 2 or more prescriptions of NSAIDs, immunosuppressive drugs or corticosteroids prior to censuring events. We performed stratified analyses according to number of prescriptions, years of follow-up, type of antidepressant used (tricyclic antidepressants only, SSRI only, other antidepressants only, or mixed use) and hospitalization (psychiatric inpatient or outpatient contact). Our analyses allowed subjects to change among categories of covariates and exposure variables over time. Individuals who changed from one type of antidepressant to another contributed person-time to a mixed-user category from that date onwards. Within each categorical level, all variables were treated as time-independent. A test for linear trend was used to evaluate dose response with number of prescriptions and years of follow-up. The statistical analyses were performed in SAS 8.02 (SAS Institute, Cary, NC).
Among 43,932 users of antidepressant medications, 11,958 used tricyclic antidepressants 22,695 SSRIs, 4,421 other antidepressants, and 4,858 more than one type. Overall, users of antidepressant medications contributed 264,489 person-years at risk and 92 cases of NHL, of which 17 cases were observed within the first year of follow-up.
The incidence of NHL was increased among users of tricyclic antidepressants with an overall IRR of 1.53 (CI = 1.06–2.21). Trends in risk estimates were found with increasing number of prescriptions and with length of follow-up (Table). The IRR for users with 10 or more prescriptions was 2.50 (CI = 1.43-4.34) if they had 5 or more years of follow-up and 1.75 (CI = 0.87-3.52) if they had less than 5 years of follow-up. For users of SSRIs the overall IRR of NHL was 1.25 (CI = 0.88-1.79); however, when the first year of follow-up was excluded there was no excess incidence, and no trend was observed with either number of prescriptions or length of follow-up (Table).
Among tricyclic antidepressants users who had never been hospitalized, the overall IRR was lower than among those who had been hospitalized; however, the risks according to number of prescriptions were similar to those of the overall analyses (for 20–29 prescriptions, IRR = 3.28 [CI = 1.36–7.93] n = 5; for 30+ prescriptions, 1.79 [CI = 0.74–4.32; n = 5]). In contrast, no significant trends were observed among never-hospitalized SSRI users for either number of prescriptions or length of follow-up (data not shown).
Users of mixed antidepressants had no increased incidence of NHL beyond the first year of follow-up (IRR = 0.95; CI = 0.61–1.47; n = 22), whereas for users of other antidepressants, the IRR excluding first year of follow-up was 1.45 (CI = 0.60–3.51; n = 5); further analysis was not possible due to small numbers (data not shown).
We found a 2.5-fold increase in incidence of NHL among long-term heavy users of tricyclic antidepressants. This supports our previous report using the same data sources2—here with a larger cohort and extension of the study period by 8 years. The previous study also revealed a 2.5-fold increase among users of 5 or more prescriptions of tricyclic antidepressants but with less statistical precision. We were thus able to conduct meaningful analyses by number of prescriptions and to compute separate estimates for long-term heavy use as defined by 10 or more prescriptions and 5 or more years of follow-up. The dose-response relation originally reported was further supported in this extended analysis with increasing risk associated with number of dispensed prescriptions, although the highest risk estimate was observed in individuals with 20–29 prescriptions.
In contrast to some animal studies,10 we did not observe an increased risk of NHL among SSRI users beyond the first year of exposure. This is consistent with the results of the only other population-based study, which included 638 cases of NHL and 1930 controls, identified through population-based registries in Ontario, Canada.3 That study reported an age- and sex-adjusted odds ratio of 1.1 (95% CI = 0.6–1.9; n = 19) for ever-use of SSRIs and 0.6 (0.2–1.8; n = 3) for use of 18 months or more.3 That study also reported a modest increased risk of NHL among persons who used tricyclic antidepressants for 25 months or more, (odds ratio = 1.6; 95% CI = 0.8–1.3; n = 12). However, because of the lack of prescription data, the authors could not evaluate a dose-response relation beyond duration of use.
The main strength of our study is the use of population-based registries4,8,9 with essentially complete coverage, which minimizes recall and selection bias, loss to follow-up, and misclassification of exposures or outcomes. Because we did not have comprehensive data on autoimmune disorders and chronic inflammatory conditions that have a known influence on the risk of NHL,1,7 we used prescriptions for immunosuppressive drugs as the best proxy for these conditions, as well as prescriptions for NSAIDs (which have also been suggested to influence risk of NHL6). Further, clinical depression has been suggested to result in immunosuppression and thereby increase the risk for cancer, although the evidence for this hypothesis is conflicting and there are no reports of increased risk of NHL in studies of depression and cancer risk.11,12 In the present study, some 25% of the exposed cohort had been admitted to a clinic or hospital with a psychiatric disorder and thus probably received antidepressant medication that we could not account for in our analysis. However, heavy use of tricyclic antidepressants was also associated with an increased incidence of NHL among those who were never hospitalized. A limitation of our study was the absence of data on indications for use of the antidepressants. Use of SSRIs in Denmark is recommended as the first choice in mild to moderate depression and tricyclic antidepressants for severe or melancholic depression, and thus a difference in severity of depression would be expected between patients treated with these 2 medications. However, whether such a difference between the exposure groups also includes lifestyle factors, socioeconomic status or other factors possibly affecting risk of NHL could not be evaluated in the present analysis.
In conclusion, the results of this population-based register-based study showed an increased incidence of NHL among long-term heavy users of tricyclic antidepressants. Given the high prevalence of antidepressant use, this finding warrants additional studies with sufficient power to investigate risk by type and dosage of antidepressant medication as well as by lymphoma subgroups.
1. Alexander DD, Mink PJ, Adami HO, et al. The non-Hodgkin lymphomas: a review of the epidemiologic literature. Int J Cancer
. 2007;120(suppl 12):1–39.
2. Dalton SO, Johansen C, Mellemkjær L, et al. Antidepressant medications and risk for cancer. Epidemiology
3. Bahl S, Cotterchio M, Kreiger N, et al. Antidepressant medication use and non-Hodgkin’s lymphoma risk: no association. Am J Epidemiol
4. Gaist D, Sorensen HT, Hallas J. The Danish prescription registries. Dan Med Bull
5. Capella D. Descriptive tools and analyses. In: Dukes MNG, ed. Drug Utilization Studies: Methods and Uses
. Geneva: World Health Organization: WHO Regional Publications; 1993:55–78. European Series No. 45.
6. Chang ET, Smedby KE, Hjalgrim H, et al. Medication use and risk of non-Hodgkin’s lymphoma. Am J Epidemiol
7. Kinlen LJ. Immunosuppressive therapy and acquired immunological disorders. Cancer Res
8. Munk-Joergensen P, Mortensen PB. The Danish Psychiatric Central Register. Dan Med Bull
9. Storm HH, Michelsen E, Clemmensen IH, et al. The Danish Cancer Registry: history, content, quality and use. Dan Med Bull
10. Serafeim A, Grafton G, Chamba A, et al. 5-hydroxytryptamine drives apoptosis in biopsylike Burkett lymphoma cells: reversal by selective serotonin reuptake inhibitors. Blood
11. Penninx BW, Guralnik JM, Pahor M, et al. Chronically depressed mood and cancer risk in older persons. J Natl Cancer Inst
12. Dalton SO, Mellemkjaer L, Olsen JH, et al. Depression and cancer risk: a register-based study of patients hospitalized with affective disorders, Denmark, 1969–1993. Am J Epidemiol