Time Trends in Risk and Risk Determinants of Non-Hodgkin Lymphoma in Solid Organ Transplant Recipients

Authors


Pia Fernberg, pia.fernberg@ki.se

Abstract

Organ transplantation increases risk of non-Hodgkin lymphoma (NHL), but long-term risk and time trends have seldom been evaluated. Immunosuppressive drug load is an important risk determinant, but the details are unclear. We studied NHL risk in a nationwide Swedish cohort of 11 081 graft recipients transplanted 1970–2008. Relative risks (RRs) were estimated within the cohort and versus the general population by age, sex, follow-up time and calendar period. NHL risk was also assessed by cumulative and average doses of immunosuppressive treatments in a nested case-control design throughout 1997 using conditional logistic regression. We observed 153 NHL cases during 97 853 years of follow-up. Compared with the general population, NHL risk was eightfold increased (RR 7.9; 95% confidence interval [CI] 6.6–9.4), and increased risks persisted after ≥15 years of follow-up among kidney (6.1; 95% CI 3.5–10) and nonkidney recipients (44; 14–103). Among nonkidney recipients, NHL risk was lower in the 2000s compared with the 1990s (0.5; 95% CI 0.3–1.0; p = 0.04). A high average dose of antithymocyte immunoglobulin (ATG) conferred an eightfold increased risk of NHL (OR 8.5; 95% CI 1.9–38). To conclude, posttransplant NHL risk decreased during the last decade among nonkidney recipients, possibly because of a more careful use of ATG, the introduction of new drugs, or both.

Abbreviations: 
ATG

antithymocyte immunoglobulin

CI

confidence interval

CMV

cytomegalovirus

EBV

Epstein–Barr virus

ICD

international classification of disease

LRT

likelihood ratio test

NHL

non-Hodgkin lymphoma

OR

odds ratio

PTLD

posttransplant lymphoproliferative disease

RR

relative risk

SIR

standardized incidence ratios

Introduction

Solid organ transplantation is an established and effective treatment for end-stage organ failure. Since the first kidney transplantation, performed in the 1950s, the number of transplant recipients has increased (1) and patient survival has improved (2,3). Despite this positive development, an increased risk of non-Hodgkin lymphoma (NHL) remains a life-threatening complication (4). The risk of NHL has been reported to be highest within the first 6 months to a year following transplantation (5,6). Recent studies also note a persistent risk increase for at least 10 years posttransplantation, albeit at a lower level (5–7), perhaps reflecting differences in the relative importance of transplant-related risk factors in different stages of the posttransplant period (6). Few studies have adequately followed organ transplant patients for lymphoma risk beyond 10 years after transplantation.

The immunosuppressive drug load enabling activation of latent Epstein–Barr virus (EBV) infection, or poor control of a primary EBV infection, is believed to have a pivotal role in posttransplant NHL development. Heart and lung graft recipients are at a markedly higher risk of NHL than kidney recipients (5,8), supporting an association with more aggressive immunosuppressive therapy. Ever use of several drugs such as cyclosporine (9), azathioprine (10), OKT3/muromonab-CD3 and antithymocyte immunoglobulin (ATG) (5,8,11) have been linked with a higher risk of NHL. Most studies have been limited to investigations of ever use of the drugs in question and have not addressed potential dose–response relationships (5,8,12–14). The introduction of mammalian target of rapamycin inhibitors around the year 2000 has been associated with a decrease in risk of posttransplant malignancies overall in several studies, but whether or not that also applies to NHL is unclear (15,16).

To further elucidate risk and time trends of posttransplant NHL, we undertook a population-based cohort study of solid organ transplant recipients in Sweden using data from National Patient and Cancer registers from 1970 to 2008. We also performed a case-control study among patients transplanted from 1970 to 1997, to assess detailed risk determinants of NHL including immunosuppressive treatment and infections.

Materials and Methods

Cohort study

The investigated cohort consisted of all patients hospitalized for organ transplantation in Sweden between 1970 and 2008 (n = 11 081) identified in the National Patient Register (17). Organ transplantations are exclusively carried out at four public university hospital centers in Sweden. The Patient Register covered the regions of two transplantation centers from 1970 and the other two starting in 1972, thus the register is population-based with regard to these procedures from 1972 and onward. Incident cancers were ascertained by linkage of transplant recipients to the Swedish Cancer Register (18). Close to 98% of all newly diagnosed cancers are reported to the Cancer Register, whereof 99% are morphologically verified (18). NHL was defined according to the ICD 7 codes 200, 202 and 204.1. Subjects with a history of NHL before transplantation, and NHL cases diagnosed within 30 days after transplantation were excluded. Subjects were followed until the date of NHL diagnosis, death or December 31, 2008, whichever came first. The Patient Register records do not provide reliable information on graft failure or retransplantation. Hence, subjects continued to contribute follow-up time from the date of first transplantation until one of the above-mentioned end dates of follow-up. The study was approved by the Regional Ethical Review Board (Karolinska Institutet, Stockholm, Sweden).

Casecontrol study

The case-control study was carried out within a subcohort of transplant recipients defined by the above-described registers between the years 1970 and 1997 (described by Ref.19). Cases were all incident NHL during this time from this cohort (n = 45), and three controls per case without a diagnosis of cancer from the remaining cohort, matched to the cases by age (5-year intervals) and calendar period of transplantation (5-year intervals). Controls were followed for an equal amount of time as their corresponding case. For one case, we only found one control fitting the matching criteria, thus 45 cases and 133 controls were identified. All living patients except one control subject gave informed consent to participate. Because of coding errors and problems locating medical records, we had to exclude six cases and 32 controls, and another two cases and three controls were excluded because there was no corresponding case or control left in their matched set. Hence, the final case-control analysis included a total of 134 subjects, 37 cases and 97 controls corresponding to inclusion rates of 88% and 79% of eligible cases and controls, respectively.

Exposure assessment (casecontrol study)

A comprehensive protocol was created to assemble information from the patients’ medical records at the transplantation centers, and at local hospitals if required. Data on patient and donor characteristics, type of transplanted organ, number of transplantations and HLA antigens were recorded by research assistants blinded to case-control status. For the posttransplant period, information on infectious complications, rejection episodes and explantation/retransplantation was extracted. Detailed information was collected on daily doses of immunosuppressants administered or prescribed, dates of dose changes and modes of administration for corticosteroids, azathioprine, cyclosporine (and cyclosporine microemulsion), muromonab-CD3 (OKT3), ATG, antilymphocyte immunoglobulin (ALG), tacrolimus (FK506, Prograf), mycophenolate mofetil and cyclophosphamide. However, information on specific brands or drug formulations was not available in the patient journals. If the exact dose was missing for a shorter time period, the mean dose was imputed between the last known prior and the subsequent dose.

Data on infectious complications were collected for viral infections with cytomegalovirus (CMV); EBV; varicella zoster; herpes simplex; hepatitis A, B or C viruses and fungal and bacterial infections. Patients were defined as having a viral infection based on serologic evidence of IgM antibodies and/or increase in IgG antibodies, or on the basis of a clinical diagnosis or administration of antiviral therapy. Fungal infections (Pneumocystis jiroveci, candida or aspergillus species) were defined as present if fungi had been demonstrated in routine tests and/or if treatment was given. Bacterial infections were recorded based on prescribed or administered antibiotic treatment and/or if a culture stated abundant growth. A rejection episode was defined as a clinically diagnosed acute rejection treated with rejection therapy. The majority of the rejection episodes were also verified by biopsy/cytology. HLA mismatch was defined as a donor phenotype of HLA-A, HLA-B or HLA-DR not present in the recipient HLA. Antigen splits were converted to their broad antigen in the estimation of number of mismatches.

Statistical analyses

In the cohort study, we performed comparisons of risk of NHL relative to the expected risk in the general population, and also internally among transplant recipients only. As measures of relative risk (RR) of NHL versus the general population, we computed standardized incidence ratios (SIR), the ratio of the observed to the expected number of NHL cases. The expected numbers of NHL cases were calculated as the product of age-, sex- and calendar period–specific follow-up time in the cohort and the corresponding incidence of NHL in the general population. SIR by age, sex, follow-up time, calendar period of transplantation and transplant type were calculated. Within the cohort of transplant recipients, we estimated the incidence rate ratio of NHL within the cohort by type of transplanted organ, age, sex, follow-up time and calendar period using Poisson regression. Age, follow-up time and calendar period were treated as time-dependent variables, allowing patients to move between categories as follow-up accrued. All covariates were treated as categorical. Estimates of main effects of each variable were adjusted for all other covariates available in the cohort study. The logarithm of time at risk was used as offset in the Poisson model. Although all cases of NHL recorded in the cancer register could be used in the analysis of risk within the cohort, the comparison with the general population was restricted to posttransplant NHL recorded as the first cancer because of the fact that background population rates were only available for NHL as a first malignancy. The incidence of NHL among all transplant patients, and in kidney and nonkidney recipients separately, in different calendar periods was further estimated and compared with risk score adjusted cumulative hazard functions, computed from the Cox proportional hazard model fit, and plotted graphically by time in years since transplantation. To compute an adjusted cumulative hazard for each period, we extracted a baseline hazard function for each decade and fixed age and sex at their median values (20). In analyses of risk by calendar time, we used the period 1990–1999 as referent because both kidney and nonkidney recipient groups were of considerable size.

In the case-control study, we performed conditional logistic regression to estimate odds ratios (OR) and 95% confidence intervals (CI). Likelihood ratio test p values were computed to compare the likelihood of a null model to the likelihood of the univariate model of interest. In analyses of rare exposures (n < 5), exact logistic regression was utilized. Five controls initially lost their matched case, but could be reallocated to other cases when permitting slightly wider matching criteria (up to 10 years difference in age and date of transplantation and 14 days divergence in follow-up time). Analyses were conducted with and without these additional controls, and because the results did not change meaningfully, they were kept in the final analyses.

In analyses of both average and total accumulated dose of immunosuppressants, the participants were categorized according to the quartile distribution among controls. For ATG, the 50th percentile was used as cutpoint. Orally administered medications were analyzed both separately and combined with the corresponding intravenous preparations. Because the results were similar, the combined results are presented. Average daily dose comprised only orally administered drugs (except for ATG and OKT3 that are not given orally). When combining orally and intravenously administered drugs, differences in biological availability and potency were taken into account. Corticosteroids were converted to the equivalent dose prednisolone (using conversion factors 8.33 × bethametasone, 0.25 × hydrocortisone, 1.25 × methyl-prednisolone, 1 × prednisone). Oral corticosteroids were multiplied by 0.82 when added to intravenously administered corticosteroids. Similarily, oral azathioprine was multiplied by a factor 0.3, cyclosporine by 0.25 and cyclosporine microemulsion by 0.38. In the case-control study, to investigate potential independence of associations observed in the univariate model, we adjusted for each of the covariates one by one in a multivariable model.

Results

Cohort study

In the cohort of 11 081 transplant recipients followed through 2008, the median follow-up time was 7 years, with a maximum of 38 years (Table 1). Kidney recipients constituted 74% of the patients (n = 8177) but generated 82% of the follow-up time. In total, 153 cases of NHL were observed during 97 853 person-years (Table 2). Compared with the general population, male and female kidney recipients were at a five- to sixfold increased risk of NHL whereas nonkidney recipients were at an about 20-fold increased risk. Young graft recipients below 20 years of age were at a 40-fold (kidney) or 400-fold (nonkidney) risk, whereas among individuals 60 years and older, risks fell to 3- and 13-fold increased, respectively. In the posttransplant period, NHL risks were highest during the first year. Among kidney recipients, a sixfold increased risk persisted after 15 years or more of follow-up. Among nonkidney recipients, the risk was 44-fold increased after 15+ years, although based on small numbers. There were no clear time trends in risk when comparing to expected rates in the general population.

Table 1.  Characteristics of solid organ transplant recipients in the cohort study (1970–2008) and of cases with posttransplant non-Hodgkin lymphoma (NHL) and controls in the nested case-control study (1970–1997)
 Cohort, n (person-years) n = 11 081case-control study,1 n (%)
Cases (n = 37)Controls (n = 97)
  1. 1Cases and controls were matched on age (in 5-year intervals) and calendar year of transplantation (in 5-year intervals).

  2. NA = not available.

Age of recipient (years)
 Median (range)48 (0.0–87)49 (14–65)50 (13–68)
Sex   
 Men6841 (59 186)27 (73)54 (56)
 Women4240 (38 667)10 (27)43 (44)
Calendar period of transplantation
 1970–19791061(13 190)3 (8)6 (6)
 1980–19892443 (32 441)12 (32)36 (37)
 1990–19993618 (37 290)22 (59)55 (57)
 2000–20083959 (14 932)
Transplant type   
 Kidney8177 (79 013)20 (54)84 (87)
 Liver1473 (8489)2 (5.4)3 (3.1)
 Heart532 (3930)10 (27)5 (5.2)
 Lung441 (2204)5 (14)5 (5.2)
 Mixed/other458 (3865)
Transplantation center
 Uppsala1749 (14 856)3 (8)14 (14)
 Stockholm2695 (22 364)6 (16)17 (18)
 Göteborg4750 (43 108)22 (60)46 (47)
 Malmö/Lund1887 (17 524)6 (16)20 (21)
Follow-up time (years)
 Median (range)7.1 (0.0–38)1.8 (0.2–16)1.2 (0.2–16)
Time in dialysis (days)
 Median (range) kidney recipients onlyNA229 (0–1864)244 (0–1918)
Table 2.  Number of observed posttransplant non-Hodgkin lymphomas (NHL), follow-up time and relative risk versus the general population in kidney and nonkidney transplant recipients, and relative risk within the cohort of all organ transplant recipients in Sweden 1970–2008
 Kidney recipientsNonkidney recipientsAll organ transplant recipients
Obs/person-yearsRelative risk (SIR; 95% CI) versus general populationObs/person- yearsRelative risk (SIR; 95% CI) versus general populationObs/person-yearsAdjusted relative risk (IRR; 95% CI) within the cohort1
  1. 1The relative risks within the cohort are adjusted for all other covariates listed. The internal analysis included more NHL cases (n = 153) than the comparison with the general population (n = 141), because of a restriction to first cancers in the latter.

  2. 2Including heart and lung.

  3. 3Other included bowel and isolated pancreas transplants.

Sex
 Men49/43 3224.8 (3.6–6.4)40/920119 (14–26)95/59 1861.1 (0.8–1.6)
 Women27/27 7626.0 (4.0–8.8)25/644024 (15–35)58/38 667  1.0 (ref)
Age at transplantation (years)
 0–195/615940 (13–93)9/1708379 (174–720)14/8323  1.0 (ref)
 20–3924/27 86711 (7.1–16)10/440839 (19–72)36/35 0450.7 (0.4–1.4)
 40–5934/30 7414.0 (2.8–5.6)35/804317 (12–24)77/44 6291.0 (0.6–1.9)
 60–13/63173.4 (1.9–5.8)11/148213 (6.5–23)26/98541.7 (0.9–3.4)
Time of follow-up
 <6 months6/38229.7 (3.6–21)12/113363 (32–110)18/4458  1.0 (ref)
 6 months–1 year9/361115 (6.9–28)11/103662 (31–111)20/50471.0 (0.5–1.9)
 1–4 years17/23 8283.9 (2.3–6.2)15/642712 (7.0–21)35/33 1280.3 (0.2–0.5)
 5–9 years17/19 1724.3 (2.5–7.0)18/463218 (10–28)37/26 9370.4 (0.2–0.6)
 10–14 years11/11 0184.4 (2.2–7.9)4/19278.6 (2.3–22)18/15 3050.5 (0.3–1.0)
 15+ years16/96356.1 (3.5–10)5/48644 (14–103)25/12 9770.5 (0.2–1.3)
Calendar period of transplantation
 1970–197910/11 4995.0 (2.4–9.2)0/310.0 (0.0–231)10/13 1900.6 (0.3–1.3)
 1980–198930/26 8375.7 (3.2–8.1)4/164515 (4.1–39)41/32 4411.0 (0.6–1.5)
 1990–199927/23 4825.1 (3.4–7.4)46/958523 (17–31)77/37 289  1.0 (ref)
 2000–20089/92674.4 (2.0–8.4)15/438017 (9.3–27)25/14 9330.6 (0.4–1.0)
Transplant type
 Kidney76/71 0855.2 (4.1–6.5)  85/79 013  1.0 (ref)
 Liver  24/708716 (9.9–23)26/84892.7 (1.7–4.4)
 Heart2  21/362626 (16–40)21/39304.6 (2.7–7.6)
 Lung  15/203036 (20–60)15/22046.4 (3.4–11)
 Mixed  5/258114 (4.7–34)6/38651.6 (0.6–3.3)
 Other3  0/3180.0 (0.0–71)0/353

In the internal analysis, when comparing to other organ transplant recipients as opposed to the general population, NHL incidence still differed by organ type and by duration of follow-up, but not by sex or age at transplantation (Table 2). Over successive calendar periods, there was no clear reduction in risk of NHL among all transplant recipients (Table 2 and Figure 1). However, among nonkidney recipients, there was a statistically significant decrease in NHL risk among individuals transplanted during 2000–2008 compared with 1990–1999 (RR 0.5; 95% CI 0.3–1.0; p = 0.04; Figure 1). In absolute terms, the NHL incidence fell from 48 cases per 10 000 person-years in 1990–1999 to 34 in 2000–2008 among these patients. There was also an apparent difference in risk from 1980–1989 to 1990–1999, but few lymphomas occurred among nonkidney recipients in the 1980s and the risk in 1980–1989 was not statistically significantly different from that in 1990–1999 (RR 0.5; 95% CI 0.2–1.3; p = 0.21). Because of small numbers, trends in risk per specific nonkidney graft type could not be robustly assessed, but point estimates of risk were below one for all types (data not shown). Among kidney recipients, risk of lymphoma did not change significantly from 1990–1999 to 2000–2008 (RR 0.8; 95% CI 0.4–1.7; p = 0.58).

Figure 1.

Figure 1.

Cumulative incidence of NHL among solid organ transplant recipients in Sweden per decade of calendar time (estimated by the cumulative hazard adjusted for age and sex, and for all organs also by organ type). Related relative risks are presented in Table 2 (internal cohort analysis) and text.

Figure 1.

Figure 1.

Cumulative incidence of NHL among solid organ transplant recipients in Sweden per decade of calendar time (estimated by the cumulative hazard adjusted for age and sex, and for all organs also by organ type). Related relative risks are presented in Table 2 (internal cohort analysis) and text.

Casecontrol study

Table 1 displays the characteristics of NHL cases and controls in the nested case-control study. The median time from transplantation to NHL diagnosis was 22 months. Overall, the median follow-up time in the study was shorter among the controls (1.2 years) than among the cases (1.8 years; Table 1), because of a larger loss of controls from the early study period. However, in the statistical analyses, each case was only compared with the matched controls followed for an equal amount of time (through conditional logistic regression).

Most of the participants (n = 194; 78%) had received a kidney transplant. Donor status, rejection episodes, transplantectomy or retransplantation were not significantly associated with NHL risk (Table S1). Two or more HLA-B mismatches were associated with a significant fourfold increase in NHL risk compared to no HLA-B mismatch (Table S1), but the association was not significant after adjustment for type of transplanted organ or ever use of ATG (data not shown).

Ever use of ATG conferred a statistically significant fivefold excess risk of NHL (Table 3). High average daily dose (>50th percentile, corresponding to a cutoff of 200 mg), but not high accumulated dose, was associated with a more pronounced risk (RR 8.5; 95% CI 1.9–38; Table 3). A greater proportion of heart, lung and liver recipients were treated with ATG (93%, 90% and 40%, respectively) in comparison with kidney recipients (18%). A high average daily dose of ATG (>50th percentile) remained significantly associated with NHL risk compared with never use of ATG upon adjustment for organ type (OR 8.5; 95% CI 1.4–50; Table S2), although risk estimation was unstable. Ever use of ATG was associated with early-onset NHL (occurring within the first year after transplantation; OR 5.8; 95% CI 1.4–34) as well as of late-onset NHL presenting after 1 year or more (OR 5.3; 95% CI 1.4–20). All patients had been treated with steroids. Neither a high accumulated dose nor a high average daily dose of corticosteroids was significantly associated with an increased risk of NHL, compared with low doses. Similarly, there was no clear evidence of associations with ever use or increasing doses of azathioprine, cyclosporine (Table 3) or cyclophosphamide (data not shown). There was evidence of an increased risk of NHL associated with ever use of OKT3 or tacrolimus, but power was limited (Table 3). Only one individual had been treated with ALG (a control), which resulted in unstable risk estimation (data not shown).

Table 3.  Immunosuppressive therapy and risk1 of posttransplant non-Hodgkin lymphoma (NHL) in a nested case-control study in Sweden 1970–1997, by ever use, total accumulated and average daily dose of treatments2
 Cases, n (%)Controls, n (%)OR (95% CI)p*
  1. *p-Value of the association computed with the likelihood ratio test (LRT).

  2. 1Odds ratios (OR) and 95% confidence intervals (CI) computed in a conditional regression model adjusting for the matching factors age and calendar period at transplantation.

  3. 2Analyses of average dose were restricted to orally administered drugs (apart from ATG and OKT3 that are exclusively intravenous drugs).

  4. 3All subjects received steroids, hence OR for ever versus never use of corticosteroids could not be estimated. Variation in efficacy of different steroid drugs were accounted for (see statistical analyses).

  5. 4Including cyclosporine microemulsion.

Antithymocyte immunoglobulin (ATG)    
 Never16 (43)74 (76) 1.0 (ref) 
 Ever21 (57)23 (24)5.6 (2.5–14)0.0001
Accumulated ATG dose (i.v.)    
 No ATG16 (43)74 (76)(ref) 
 <50th percentile12 (32)13 (13)5.8 (1.9–18) 
 >50th percentile9 (24)10 (10)5.3 (1.6–17)0.0005
Average dose ATG (i.v.)    
 No ATG16 (43)74 (76)(ref) 
 <50th percentile14 (38)15 (16)5.0 (1.8–13) 
 >50th percentile7 (19)8 (8.3)8.5 (1.9–38)0.0004
Corticosteroids3    
Accumulated corticosteroid dose (i.v., o)    
 1st quartile6 (16)25 (26)(ref) 
 2nd quartile8 (22)24 (25)1.4 (0.4–4.5) 
 3rd quartile7 (19)24 (25)1.6 (0.4–6.8) 
 4th quartile16 (43)24 (25)4.5 (0.8–24)0.32
Average dose corticosteroids (o)    
 1st quartile11 (30)22 (23)(ref) 
 2nd quartile11 (30)23 (24)1.0 (0.3–3.2) 
 3rd quartile7 (19)26 (27)0.6 (0.1–2.5) 
 4th quartile8 (22)26 (27)0.7 (0.2–2.7)0.83
Azathioprine (i.v., o)    
 Never3 (8.1)20 (21)(ref) 
 Ever34 (92)77 (79)2.8 (0.8–16)0.12
Accumulated azathioprine dose (i.v., o)    
 No azathioprine3 (8.1)20 (21)(ref) 
 1st quartile12 (32)19 (20)4.4 (1.0–18) 
 2nd quartile5 (13.5)20 (21)1.7 (0.4–8.6) 
 3rd quartile5 (13.5)19 (20)1.5 (0.3–8.0) 
 4th quartile12 (32)19 (20)5.2 (0.7–40)0.10
Average dose azathioprine (o)    
 No azathioprine3 (8.1)20 (21)(ref) 
 1st quartile10 (27)20 (21)2.9 (0.7–18) 
 2nd quartile10 (27)18 (19)3.7 (0.8–25) 
 3rd quartile5 (14)19 (20)1.4 (0.2–11) 
 4th quartile9 (24)20 (21)3.4 (0.7–23)0.27
Cyclosporine (i.v., o)    
 Never6 (16)13 (13)(ref) 
 Ever31 (84)84 (87)0.9 (0.2–5.8)0.92
Accumulated cyclosporine dose (i.v., o)    
 No cyclosporine6 (16)13 (13)(ref) 
 1st quartile10 (27)21 (22)2.4 (0.2–24) 
 2nd quartile5 (14)21 (22)0.5 (0.0–5.6) 
 3rd quartile6 (16)21 (22)0.5 (0.1–4.4) 
 4th quartile10 (27)21 (22)1.2 (0.1–10)0.34
Average dose cyclosporine (o)4    
 No cyclosporine6 (16)13 (13)(ref) 
 1st quartile7 (19)22 (23)0.4 (0.0–3.4) 
 2nd quartile8 (22)21 (22)0.5 (0.1–4.2) 
 3rd quartile4 (11)25 (26)0.3 (0.1–2.6) 
 4th quartile12 (32)16 (17)2.5 (0.3–20)0.03
Muromonab-CD3 (OKT3)    
 Never32 (87)94 (97)(ref) 
 Ever5 (14)3 (3.1)6.0 (0.9–67)0.06
Tacrolimus    
 No32 (87)96 (99)(ref) 
 Yes5 (14)1 (1.0)11 (1.2–512)0.04
Combined treatment regimens    
 Steroids + Azathioprine6 (16)13 (13)(ref) 
 Steroids + Azathioprine + Cyclosporine17 (46)36 (37)2.0 (0.2–31) 
 Steroids + Cyclosporine2 (5.4)9 (9.3)0.7 (0.04–12) 
 Azathioprine + Cyclosporine12 (32)39 (40)0.7 (0.1–11)0.35

All patients with a posttransplant EBV infection (primary infection n = 2; reactivation n = 5) were cases, implying a strong but unstable estimated association with NHL risk, although a stable risk estimate could not be computed because of the lack of infected controls (Table 4). Also, posttransplant CMV infection was associated with an increased NHL risk. In analyses of all herpes virus group infections (EBV, CMV, varicella zoster and herpes simplex virus infections), a fivefold increased risk of NHL was observed (Table 4), which remained after adjustment for ATG or organ type (Table S2); this association was only evident for early-onset NHL (data not shown). Fungal infections and multiple infection types (bacterial, fungal and viral) were associated with an increased risk of NHL (Table 4) but the association disappeared after adjustments for ATG or organ type (data not shown).

Table 4.  Posttransplant infectious complications and risk1 of posttransplant non-Hodgkin lymphoma (NHL) in a nested case-control study in Sweden 1970–1997
 Cases, n (%)Controls, n (%)OR (95% CI)p*
  1. *p-Value of the association computed with the likelihood ratio test (LRT).

  2. 1Odds ratios and 95% CI in a conditional logistic regression model adjusting for the matching factors age and calendar period at transplantation.

  3. 2Including cytomegalovirus (CMV), Epstein–Barr virus (EBV), herpes simplex virus (HSV), varicella zoster virus (VZV).

  4. 3Any fungal infections included Candida albicans, Pneumocystis jiroveci and Aspergillus.

Infectious complications    
Any herpes virus group infection2    
 No11 (30)63 (65)(ref) 
 Yes26 (70)34 (35)4.9 (1.9–13)0.0004
CMV infection    
 No20 (54)74 (76)(ref) 
 Yes17 (46)23 (24)2.8 (1.2–6.8)0.01
No. of CMV episodes    
 020 (54)74 (76)(ref) 
 113 (35)17 (18)3.0 (1.1–9.0) 
 ≥24 (11)6 (6.2)2.4 (0.4–12)0.04
EBV infection    
 No30 (81)97 (100)(ref) 
 Yes7 (19)0 (0.0)NA
Any fungal infection3    
 No18 (49)73 (75)(ref) 
 Yes19 (51)24 (25)2.9 (1.3–6.6)0.007
No. of Candida infections    
 023 (62)82 (85)(ref) 
 1–212 (32)12 (12)3.2 (1.3–8.5) 
 >22 (5.4)3 (3.1)2.3 (0.2–22)0.02*
Bacterial infection    
 No6 (16)26 (27)(ref) 
 Yes31 (84)71 (73)1.8 (0.7-4.7)0.24
No. of bacterial infections    
 06 (16)26 (27)(ref) 
 1–217 (46)45 (46)1.6 (0.6–4.6) 
 >214 (38)26 (27)2.0 (0.7–6.1)0.43
Multiple types of infections    
 No infections4 (11)20 (21)(ref) 
 Bacterial and viral9 (24)13 (13)2.9 (0.6–17) 
 Bacterial and fungal4 (11)6 (6.2)1.8 (0.2–15) 
 Bacterial, viral and fungal16 (38)14 (14)4.2 (1.1–22)0.012

Discussion

In this large population-based cohort study of solid organ transplant recipients in Sweden, recipients of heart, lung or liver grafts were confirmed to be at a much greater risk of NHL than kidney recipients, and increased risks persisted beyond 15 years after the first transplantation among kidney as well as nonkidney recipients. Interestingly, we observed a decline in NHL risk in the 21st century among nonkidney recipients compared with the 1990s. In a detailed scrutiny of administered immunosuppressive therapy, we observed a strong association between ever use and average dose of ATG and NHL risk. Although tacrolimus was also associated with NHL risk, there was no clear evidence for associations with increasing doses of azathioprine, cyclosporine or corticosteroids. Consistent with previous studies, posttransplant primary EBV infection or reactivation was associated with NHL risk. It is possible that a more careful use of ATG in recent years could, at least in part, explain the observed decrease in risk of NHL among nonkidney graft recipients in the 21st century, although time trends in use of immunsosuppressive drugs could not be specifically investigated.

In the last decade, immunosuppressive regimens have changed, new therapies have been introduced and the immunosuppressive load is generally lower today. Whether the incidence of NHL has been influenced by this development has not been explicitly investigated. With regard to lymphoproliferative malignancies, Libertiny et al. reported in 2001 (21) an increase in incidence among renal graft recipients following the introduction of cyclosporine in the 1980s, and a further increase in the 1990s. An increase in NHL incidence from the 1970s to the 1980s is also indicated in the present data. In a recent large study of renal transplant recipients in Australia (6), no clear time trends were observed up to end of follow-up in 2003, but a several-fold increased risk of NHL was observed after 15 years or more of follow-up, which is in line with our findings. With regard to age at transplantation, we note that young recipients (<20 years of age) were at an especially high risk of lymphoma relative to the low expected incidence in the young general population, in line with previous findings (19,22,23). However, age was not a strong risk determinant among the transplant recipients per se.

Heart-transplanted patients receive what is perceived as more immunosuppression, and the high immunosuppressive load along with rejection therapy in the first posttransplant year is believed to be a major determinant of lymphoma risk in these patients (5,6,8). Opelz et al. showed that induction, as well as rejection, treatment with ATG or OKT3 were associated with an increase in risk of early- and late-onset NHL (5). An increased risk of lymphoma associated with ever use of ATG has been confirmed in cohorts of kidney (14) and nonkidney recipients (7). A few additional studies also support an increased risk of lymphoma in patients treated with OKT3 (12,13,24).

With regard to other immunosuppressive drugs used in this setting, such as cyclosporine, azathioprine and tacrolimus, previous results are inconclusive. Ever use of cyclosporine in combination with azathioprine has been associated with a modest excess risk of NHL (8). However, in another study, lymphomas were more common among patients treated with cyclosporine only, as opposed to a combination of azathioprine and steroids (9). Two studies observed a higher risk of posttransplant lymphoproliferative disorders (PTLD) in tacrolimus-treated patients in comparison with those on cyclosporine regimens (5,14), although no association was detected in another (7). Yet other investigations did not report any significant differences in PTLD or NHL incidence comparing treatment eras (25) or regimens (5).

Taken together, previous and presented findings clearly point to a more important role for specific drugs, primarily ATG, in posttransplant lymphoma development. However, neither we nor others have been able to investigate potential interaction effects between different drugs. It is also important to note that even actual doses of administered immunosuppressive medications may not entirely reflect the resulting biological level of immunosuppression in each individual.

Recipient EBV seronegativity is repeatedly reported as a risk factor for NHL/PTLD (14,24,26). Although we could not assess the role of pretransplant EBV serology, our observation that posttransplant EBV infection/reactivation only occurred in those who subsequently developed NHL, supports an important role of EBV in the development of early posttransplant NHL. In line with this, early-onset NHL/PTLD (occurring within 12 months) is typically EBV positive (27–29). Growth and/or apoptosis-controlling cellular pathways are disturbed by the action of EBV viral proteins (EBNA-1 and LMP-1), resulting in complete immortalization of B cells (30,31). The immunosurveillance by cytotoxic T cells toward the infected cells is impeded by immunosuppressive medications, mainly ATG, OKT3 and tacrolimus, allowing the B-cell blasts to proliferate (30). NHL/PTLD presenting late, on the other hand, are more commonly EBV negative (27,29). In this context, it is interesting to note that use of ATG in our study conferred an increased risk of NHL of both early and late onset.

The strengths of our investigation include the population-based design and the complete assessment of organ transplant recipients and of NHL through nationwide and high-quality registers. With these data, we had the ability to assess long-term risks and recent time trends. To our knowledge, this is also the first study to assess drug doses as they were actually used during the entire posttransplant period up until lymphoma diagnosis, whereas previous studies have been confined to ever use of drugs based on intention-to-treat guidelines or crude measures of administered doses (5,7–9,11,14,24,25). Our study also has limitations. We lacked detailed information on risk determinants in the large cohort study, and could not investigate potential variation in risk during periods on and off immunosuppressive treatment and dialysis (6). Instead, risk of posttransplant NHL among kidney recipients represented an estimate of average risk for the entire follow-up period. The case-control study was small in numbers and follow-up ended in 1997, which limited our ability to investigate effects of drugs introduced in the late 1990s and thereafter, in relation to risk. Our investigation of time trends in NHL incidence covering the period through 2008, only indirectly reflects potential effects of the introduction of new drugs and treatment regimens together with modifications of old treatment schedules.

Despite efforts to adjust for confounding, independent effects of correlated variables such as immunosuppressive load and infections were difficult to disentangle, and residual confounding because of unmeasured alterations of immunosuppressive regimens, misclassification or unknown factors cannot be excluded.

In summary, our study provides evidence that the risk of posttransplant NHL has decreased among nonkidney transplant recipients in the 21st century. If this is because of a more careful use of ATG, or to the introduction of new and less carcinogenic drugs is, however, an open question. In future evaluations of the potential benefit of new drugs such as sirolimus with regard to posttransplant lymphoma risk, effects of a parallel reduction in use of drugs such as ATG should be kept in mind.

Acknowledgments

We are grateful to The Swedish Cancer Society and Filip Lundbergs Foundation for contributing with funding to this project.

Disclosure

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

Ancillary