Main

Organ transplantation is a lifesaving option for individuals with end-stage organ disease, and over 28 000 solid organ transplantations are performed yearly in the United States. However, solid organ transplant patients must receive intensive long-term immunosuppressive therapy to prevent rejection of the transplant, putting them at high risk of developing post-transplant lymphoproliferative disorders (PTLDs). These disorders include a spectrum of potentially deadly lymphoid cell proliferations, including non-Hodgkin lymphoma (NHL) and Hodgkin lymphoma (HL) (Tsao and Hsi, 2007). The NHL represents one of the most common malignancies diagnosed after transplant (Andreone et al, 2003). Risk in transplant recipients is estimated as 3- to 21-fold higher than that in the general population, and perhaps as much as 120-fold higher among children who receive transplants (Kasiske et al, 2004; Busnach et al, 2006; Caillard et al, 2006; Vajdic et al, 2006; Giordano et al, 2007; Grulich et al, 2007; Serraino et al, 2007; Jiang et al, 2008; Baccarani et al, 2009; Vajdic and van Leeuwen, 2009; Quinlan et al, 2010; Engels et al, 2011). The NHL risk exhibits a U-shaped pattern over time following transplantation, with risk being highest in the first year after transplant, falling subsequently, and then increasing again at 5 or more years after transplant (van Leeuwen et al, 2009; Quinlan et al, 2011). HL was once considered aetiologically distinct from NHL, but is now recognised as similar to some B-cell NHL subtypes. Risk of HL among transplant recipients is likely lower than that for NHL, having been reported as 2- to 3.6-fold higher than that in the general population (Quinlan et al, 2010; Engels et al, 2011).

A major gap in our understanding of transplant-related lymphoma involves the spectrum and associated risks of the many histologic subtypes of lymphoma, which are heterogeneous with respect to clinical and epidemiologic characteristics (Morton et al, 2006; Swerdlow et al, 2008). The two prior studies examining the risk of NHL subtypes among transplant recipients reported substantial elevation in the risk of diffuse large B-cell lymphoma (DLBCL, the most common subtype in most populations) but no increase for follicular lymphoma (another common subtype in the general population) (Quinlan et al, 2010; Vajdic et al, 2010). These studies were hampered by small numbers, especially of uncommon NHL subtypes, and did not examine how the occurrence of NHL subtypes varied by demographic characteristics or time since transplantation.

Better characterisation of the risk of lymphoma subtypes in transplant recipients would help clinicians caring for these patients, as different subtypes require different management. This information would also improve understanding of the causal importance of immunosuppression. Therefore, we assessed the risks of individual lymphoma subtypes in the recently completed Transplant Cancer Match Study (Engels et al, 2011). This large cohort of US solid organ transplant recipients, for whom cancer ascertainment was conducted uniformly via linkage with population-based cancer registries, enabled us to quantify the risks of specific subtypes overall and according to recipient age at transplantation, type of transplanted organ, and time since transplant.

Materials and methods

Transplant cancer match study

The Transplant Cancer Match Study is described in detail elsewhere (http://transplantmatch.cancer.gov) (Engels et al, 2011). In brief, the US Scientific Registry of Transplant Recipients (SRTR) was linked with 14 US population-based cancer registries. The SRTR includes structured data regarding all US solid organ transplants occurring since 1987, including recipient demographic characteristics, reason for transplant, and characteristics of the transplanted organs. The cancer registries include standardised information regarding patient demographic characteristics and detailed tumour characteristics.

Serial record linkages were completed between the SRTR and the following central cancer registries, together covering ∼42% of the US transplant patient population: California (years of coverage: 1988–2008), Colorado (1988–2006), Connecticut (1973–2006), Georgia (1995–2008), Hawaii (1973–2007), Illinois (1986–2007), Iowa (1973–2007), Michigan (1985–2006), New Jersey (1979–2006), New York (1976–2007), North Carolina (1990–2007), the Seattle-Puget Sound area of Washington State (1974–2008), Texas (1995–2006), and Utah (1973–2008). Record linkages were accomplished using a computer algorithm followed by manual review and confirmation of potential matches. Analyses were restricted to transplant recipients residing in geographic areas covered by the cancer registries during the specified time periods.

The TCM Study was approved by human subjects research review committees at the National Cancer Institute and the following cancer registries: California, Colorado, Connecticut, Georgia, Hawaii, Illinois, Iowa, Michigan, New Jersey, New York, Seattle-Puget Sound, Texas, and Utah. It was formally exempted from human subjects research approval at the Health Resources and Services Administration and the North Carolina cancer registry.

Lymphoma outcomes and follow-up

The NHLs were identified in transplant recipients through linkage with cancer registries. Lymphoma subtypes were classified using International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) site and histology codes according to current International Lymphoma Epidemiology (InterLymph) Consortium consensus guidelines (Turner et al, 2010). Transplant recipients were considered at risk of lymphoma beginning at the date of transplantation or the start of cancer registry coverage (whichever came later). Hispanics were followed beginning in 1992 to correspond to the years for which general population rates were available for comparison (see below). Follow-up time ended at the first diagnosis of any lymphoma subtype, death, failure of a transplanted organ, subsequent transplant, loss to follow-up, or last date of cancer registry coverage (whichever came first). Recipients were considered at risk separately during successive transplant episodes. Of the 180 210 distinct transplant episodes, we excluded 1265 because racial/ethnic information was missing or did not correspond to a group for which general population cancer rates were available, and a further 160 episodes occurring among recipients with human immunodeficiency virus (HIV) infection reported to the SRTR. We did not exclude episodes if the recipient had a history of cancer before transplantation. Altogether, our final analytic cohort included 178 785 transplant episodes occurring in 165 734 individuals, 1617 NHL, and 48 HL cases.

Statistical analysis

For each lymphoma subtype, we measured the risk in transplant recipients relative to the general population using the standardised incidence ratio (SIR, i.e., observed/expected number of cases). Observed numbers were derived from the cancer registry linkages, as described above. Expected numbers were calculated by applying general population cancer incidence rates obtained from each cancer registry to person-time at risk among transplant recipients, stratified by sex, 5-year age group, race/ethnicity, and calendar year. Rates for whites, blacks, and Asians/Pacific Islanders were derived from participating registries. Rates for Hispanics (available 1992 onward) were derived from the US Surveillance, Epidemiology, and End Results programme, in which 8 of the 14 contributing cancer registries participate. The 95% confidence intervals (CIs) around each SIR were derived assuming that the observed counts follow a Poisson distribution.

We also calculated SIRs and 95% CIs across categories defined by age at transplant, transplanted organ, and successive time intervals (1–360 days, 361–1800 days, and 1801+ days) after transplant. Univariate Poisson regression models were created for each subtype and used to test for heterogeneity across these categories.

In sensitivity analyses we (1) excluded all transplants for which the reported NHL was not the first haematologic malignancy reported, and (2) excluded transplants for which there was a gap between date of transplant and the beginning of cancer registry coverage. Results from these sensitivity analyses were similar to those from our main analysis and thus are not presented here. All CIs and statistical tests were two sided.

Results

We evaluated 178 785 solid organ transplants in 165 734 individuals, with 791 281 person-years of follow-up. Table 1 describes the demographic characteristics of individuals who received these transplants. Of the transplant recipients, 61% were male, and the median age at transplant was 47.0 years, with a quarter of patients under age 35 years at the time of transplant. Recipients were racially and ethnically diverse, with almost 40% of patients being non-white. Kidney transplants were most common, but 41% of transplants were of other organs, and 91% were first transplants.

Table 1 Characteristics of 178,785 solid organ transplants*, United States, 1987–2008
Full size table

Through matches with cancer registry records, 1617 NHL diagnoses and 48 HL diagnoses among transplant recipients were identified. Of the NHLs, specific histologic subtype was reported for 1285 (80%). Compared with cases with specified subtypes, NHLs for which histologic subtype was not reported (i.e., ‘NHL, not otherwise specified’) were similar in terms of age, sex, and organ transplanted, but were significantly (P<0.05) more likely to have disease limited to lymph nodes (i.e., nodal) and to have been diagnosed in the earlier years of the study (e.g., 1987–1994).

Overall, transplant recipients had over six-fold increased risk of developing any kind of NHL compared with the general population (SIR=6.2, 95% CI: 5.9–6.5). Table 2 describes the risk for 19 defined distinct NHL subtypes. Among NHLs for which specific subtypes were reported, 85.5% were B-cell lymphomas and 6.2% were T-cell lymphomas. The NHL subtype with the highest elevation in risk was hepatosplenic T-cell lymphoma, an uncommon NHL subtype for which the risk was elevated 100-fold (SIR=100, 95% CI: 33–234) above the general population. However, this increase was based on only six cases and an incidence rate of less than one per 100 000 person-years. The most common subtype diagnosed was DLBCL, which comprised almost two-thirds of all post-transplant NHLs with specified subtype, and for which the risk was over 13 times higher than in the general population (SIR=13.5, 95% CI: 12.7–14.4). After DLBCL, the next most commonly diagnosed subtype was Burkitt’s lymphoma, for which the risk was ∼25 times elevated (SIR=24.5, 95% CI: 19.6–30.2). Other B-cell lymphomas with significantly elevated risks included lymphoplasmacytic lymphoma (SIR=2.8, 95% CI: 1.6–4.5) and marginal zone lymphoma of mucosa-associated lymphoid tissue (MALT; SIR=2.8, 95% CI: 1.9–4.0). The risk of developing HL was significantly elevated 3.6-fold (SIR=3.6, 95% CI: 2.9–4.4) compared with the general population.

Table 2 Overall SIRs for lymphoma subtypes among solid organ transplant recipients, United States, 1987–2008
Full size table

When T-cell lymphomas were considered as a single entity, the risk was seven-fold higher than in the general population (SIR=7.1, 95% CI: 5.6–8.9; Table 2). Peripheral T-cell lymphomas and anaplastic large-cell lymphomas comprised over 80% of the T-cell lymphomas, with other subtypes occurring rarely. Among the T-cell lymphomas, the greatest risk elevation was for hepatosplenic T-cell lymphoma (SIR=100, as noted above). Other T-cell lymphomas with elevated risks included extranodal natural killer (NK)/T-cell lymphomas, nasal type (SIR=15.0, 95% CI: 6.5–29.6), anaplastic large-cell lymphoma (SIR=12.8, 95% CI: 9.0–17.7), primary cutaneous anaplastic large-cell lymphoma (SIR=13.5, 95% CI: 6.2–25.5), and other peripheral T-cell lymphomas (SIR=3.9, 95% CI: 2.7–5).

Among other specified NHLs, the risk was also elevated for precursor B- or T-cell lymphoblastic leukaemia/lymphoma, compared with the general population (SIR=2.0, 95% CI: 1.23–3.20). In sharp contrast, few cases and no significant elevations in risk were observed for follicular lymphoma, small lymphocytic lymphoma/chronic lymphocytic leukaemia (CLL/SLL), mantle cell lymphoma, or mycosis fungoides/Sézary’s syndrome.

The risks differed substantially according to the age at transplant, with a strong inverse relationship between SIRs and age demonstrated for several NHL subtypes and HL (Table 3). Among recipients aged <20 years at the time of transplantation, the risks were strikingly elevated for Burkitt’s lymphoma (SIR=123, 95% CI: 79.0–183), DLBCL (SIR=379, 95% CI: 318–447), peripheral T-cell lymphoma (SIR=172, 95% CI: 69.1–354), and anaplastic large-cell lymphoma (SIR=96.9, 95% CI: 41.8–191). For many subtypes, it was difficult to assess gradients in risk by age because there were fewer than three cases diagnosed among those aged <20 years at transplant. Notably, however, the risks for most subtypes among individuals aged ≥50 years at the time of transplantation remained elevated compared with the general population.

Table 3 SIRs and 95% CIs for development of lymphoma subtypes in subgroups of solid organ transplant recipients by age at transplant year, United States, 1987–2008
Full size table

Table 4 shows that for some subtypes, the patterns of risk also differed with respect to time since transplant, although for other subtypes, the numbers were small and differences were not significant. The SIRs for DLBCL and anaplastic large-cell lymphoma showed a U-shaped pattern, with risks highest in the first year after transplant, lower at 361–1800 days after transplant, and slightly higher again 1801+ days after transplant. In contrast, SIRs for HL, Burkitt’s lymphoma, peripheral T-cell lymphomas, and hepatosplenic T-cell lymphoma were not statistically significantly elevated in the first year after transplant but increased subsequently, with markedly elevated relative risk observed 1801+ days after transplant.

Table 4 SIRs and 95% CIs for development of lymphoma subtypes in subgroups of solid organ transplant recipients by time since transplant, United States, 1987–2008
Full size table

The risk of lymphoma subtypes differed by the type of organ transplanted (Table 5). For most subtypes, kidney recipients had lower SIRs than liver or thoracic organ (heart and/or lung) recipients (although still elevated compared with the general population in many instances). However, the number of cases for many subtypes was small, and risk did not differ significantly by organ. The risks for DLBCL and anaplastic large-cell lymphoma were highest among pancreas and kidney/pancreas recipients, whereas Burkitt’s lymphoma risk was highest among liver and thoracic organ recipients. None of the recipients who developed hepatosplenic T-cell lymphoma received a liver transplant, rather, four of the six recipients received kidneys.

Table 5 SIRs and 95% CIs for development of lymphoma subtypes in subgroups of solid organ transplant recipients by transplanted organ, United States, 1987–2008
Full size table

The patterns for NHL overall according to age, organ type, and time since transplant generally mirrored those observed for DLBCL, the most common subtype (Tables 3, 4, 5). Similarly, the patterns of risk of unclassified lymphoma (NHL, not otherwise specified) were also similar to those observed for DLBCL.

Discussion

Non-Hodgkin’s lymphoma is among the most common malignancies diagnosed among solid organ transplant recipients. Overall, the risk for NHL is six-fold higher following transplantation than in the general population, but in this large, population-based study we demonstrate that the risks vary dramatically by lymphoma subtype. Elevated risks were particularly striking (>10-fold) for hepatosplenic T-cell lymphoma, Burkitt’s lymphoma, extranodal NK/T-cell lymphoma nasal type, DLBCL, and anaplastic large-cell lymphoma (both systemic and primary cutaneous forms). For DLBCL and Burkitt’s lymphoma, which were the most common lymphoma subtypes after transplant, the relative risks were most strongly pronounced among transplant patients under age 20 years, and differed according to the type of organ transplanted. We also observed moderately elevated risks (i.e., 2- to 3-fold higher than the general population) for HL, lymphoplasmacytic lymphomas, precursor B- or T-cell lymphoblastic leukaemia/lymphoma, other T-cell lymphomas, and marginal zone lymphoma of MALT type, whereas the risks were not increased for follicular or mantle cell lymphomas, CLL/SLL, or mycosis fungoides/Sézary’s syndrome.

The mechanisms explaining the different spectrum of lymphoma subtypes among transplant patients compared with the general population include immunosuppression, including the early, intense induction phase and the later maintenance phase; chronic immune activation because of the presence of donor organ tissue; or the combined effects of these, resulting in long-term, chronic immune dysfunction. The lymphoma subtypes for which the risk is elevated are similar to those that are increased in HIV-infected patients (Mbulaiteye et al, 2003; Biggar et al, 2007; Engels et al, 2008; Vajdic et al, 2010), another population experiencing both chronic immunosuppression and immune activation. Investigators in the United States and Australia have demonstrated especially elevated risks among HIV-infected individuals for ‘high-grade’ NHL subtypes as a group (SIR ∼50–110) and for DLBCL (SIRs 25–100), Burkitt’s lymphoma (SIRs 50–140), and T-cell lymphomas (SIR ∼15) (Mbulaiteye et al, 2003; Engels et al, 2006; van Leeuwen et al, 2009; Vajdic et al, 2010). In contrast, follicular or ‘low-grade’ lymphoma risks were not higher than in the general population in any of the transplant or HIV-infected groups (Mbulaiteye et al, 2003; Quinlan et al, 2010; Vajdic et al, 2010). The overall relative risks observed here are lower than those observed in HIV-infected populations, perhaps because of the different pattern of immunosuppression or the different type of immune stimulus conferred by a donor organ than associated with HIV infection.

These data are informative with regard to several aspects of lymphomagenesis for specific subtypes. For DLBCL, our observations are consistent with the known role of Epstein–Barr virus (EBV)-driven B-cell expansion under conditions of intense immunosuppression, especially in children. Epstein–Barr virus is detected in the tumour cells of most transplant-related DLBCLs (Carbone et al, 2009). We found that the risk of developing DLBCL was concentrated in children (ages 0–19 years), and was highest in the first year after transplant. These findings agree with previous reports of transplant-related NHL overall and PTLD (Opelz et al, 2009; Vajdic and van Leeuwen, 2009) and are expected, given that DLBCL comprises a large fraction of PTLD. Children are often EBV seronegative at transplantation, which puts them at risk of subsequent primary EBV infection and PTLD (Jenson, 2011; Quinlan et al, 2011). For Burkitt’s lymphoma, which is also an aggressive lymphoma, the risk was similarly highest in children, but was not elevated in the first year after transplant and rose with time since transplant. Of interest, only 40–60% of HIV-associated Burkitt’s lymphoma is EBV related (Carbone et al, 2009). We were unable to separate EBV-defined cases, and it is possible that the age and latency patterns that we observed reflect mixed occurrence for EBV-positive and -negative forms of Burkitt’s lymphoma after transplant (Carbone et al, 2009). Transplant-related cases of HL are also commonly EBV positive (Bierman et al, 1996; Knight et al, 2009), and in our study, HL risk also increased with time since transplant.

We also observed statistically significantly elevated risks of marginal zone lymphomas of MALT type and lymphoplasmacytic lymphoma, although the magnitudes of elevation were lower than those for DLBCL and Burkitt’s lymphoma. Outside of the transplant setting, these subtypes are thought to arise under conditions of persistent immune stimulation by chronic microbial infections (e.g., Helicobacter pylori for gastric MALT type lymphoma, hepatitis C virus for lymphoplasmacytic lymphoma) (Engels, 2007; Giordano et al, 2007; de Sanjose et al, 2008). Transplant-related immunosuppression may increase microbial activity or modulate the immune response against these microorganisms. In contrast, for other indolent B-cell lymphomas such as follicular lymphoma and CLL/SLL, our data are not suggestive of major roles for the types of immunosuppression, immune activation, or other immune disturbances experienced by transplant recipients.

Although T-cell lymphomas constituted a small proportion of NHLs in this transplant population, we observed very high relative risks of certain T-cell subtypes, including hepatosplenic T-cell lymphoma and both systemic and primary cutaneous anaplastic large-cell lymphomas. For anaplastic large-cell lymphoma, relative risks were markedly elevated in the first year after transplant. Of interest, EBV is not generally thought to play an important role in the aetiology of T-cell PTLD (Engels, 2007). Although the exceedingly rare hepatosplenic T-cell lymphoma is well recognised as a type of PTLD, EBV is infrequently detected in tumour cells of this malignancy (Engels, 2007). The very high relative risks of hepatosplenic T-cell and anaplastic large-cell lymphomas observed here after transplant, but not in HIV-infected individuals (Opelz and Dohler, 2004; LaCasce, 2006), may suggest that unique aspects of transplant-associated immune disturbance are aetiologically important. Inflammatory and anti-inflammatory activity are of interest because of the apparent association of hepatosplenic T-cell lymphomas with use of tumour necrosis factor-α (TNF-α) inhibitors for Crohn’s disease, arthritis, and other autoimmune diseases (Hellgren et al, 2010; Deepak et al, 2013; Mason and Siegel, 2013). It is also interesting that none of the hepatosplenic T-cell lymphomas occurred in liver recipients, among whom the donor organ provides a strong and sustained local immune stimulus.

Our data also demonstrate substantial variation in the risk of lymphoma subtypes according to age and time since transplant. In general, SIRs for most lymphoma subtypes were elevated among young transplant recipients. This pattern probably reflects the contribution of primary EBV infection in younger recipients, as well as a countervailing age-related rise in NHL incidence in the general population due to other mechanisms. Although we confirmed a U-shaped pattern in overall NHL risk over time, a pattern previously observed by others (van Leeuwen et al, 2009; Quinlan et al, 2011), we did not find a consistent pattern of risk across all the specific subtypes. Intense, T cell-depleting induction immunosuppressive agents used at the time of transplantation may explain why the relative risks of DLBCL and systemic and cutaneous anaplastic large-cell lymphoma were highest nearer the time of transplant. In contrast, the relative risk of Burkitt’s lymphoma increased with longer time since transplant. Among older recipients and among recipients many years post transplant, lymphoma occurrence may be caused by some combination of age-related immune senescence, immunosuppression-related immune dysfunction, and chronic antigenic stimulation from the transplanted graft or infection (Opelz and Dohler, 2004; Vajdic and van Leeuwen, 2009; van Leeuwen et al, 2009).

Our analysis is the first to assess lymphoma subtype occurrence by the type of organ transplanted. The relative risk of DLBCL was substantially higher among pancreas and pancreas/kidney recipients than recipients of other organs, and Burkitt’s lymphoma risk was highest among recipients of liver and thoracic organ (heart and/or lung) transplants. Most prior studies of post-transplant cancer risk have focussed on kidney recipients, but some have reported higher risks of overall NHL for small intestine or lung recipients than heart, liver, or kidney recipients (LaCasce, 2006; Tsao and Hsi, 2007). In the Collaborative Transplant Study, PTLD risk was substantially higher among heart/lung and lung recipients than among other recipients (Opelz and Dohler, 2004). Variation in the risk of DLBCL and other subtypes by organ type may relate to differences in immunosuppressive regimens or the intensity of immunosuppression. Along these lines, some studies have shown that overall NHL risk is higher with certain induction immunosuppression agents, particularly monoclonal anti-CD3 antibody (Bustami et al, 2004; Dharnidharka et al, 2012). Among HIV-infected patients, the risk of DLBCL also has been shown to correlate with the degree of immunosuppression as reflected by CD4 count (Biggar et al, 2007). Alternatively, it is possible that different transplanted organs confer different levels of chronic antigen stimulation relevant to lymphomagenesis.

Although DLBCL is the most commonly diagnosed B-cell malignancy in the general population, it represents only 25–30% of all NHL cases (Morton et al, 2006), whereas it comprises over half of all transplant-related NHLs. Follicular lymphoma and CLL/SLL each comprise ∼15–20% of NHLs diagnosed in the general population, but <3% of NHL cases in our transplant cohort. Notably, transplant recipients thus have higher proportions of aggressive subtypes and lower proportions of indolent B-cell lymphomas. Because many of the subtypes with elevated risk are rare in the general population, our results underscore the importance of expert haematopathologic workup of suspicious lymphoproliferations among transplant patients (Jagadeesh et al, 2012) and, moreover, suggest a specific range of lymphoma subtypes as differential diagnoses among such patients. Correct diagnosis of subtype is important for management of post-transplant lymphoma, most importantly for planning therapeutic regimen, which varies substantially by specific lymphoma subtype.

This study has several important strengths. It is the first cohort study of transplant recipients to have a large enough number of incident lymphoma cases to allow assessing risks separately for a wide spectrum of subtypes. Prior efforts to quantify these risks have been limited by lack of data on lymphoma subtypes, or small size and restriction to special subgroups of transplant patients, for example, kidney recipients (Vajdic et al, 2010) or Medicare beneficiaries over age 65 years (Quinlan et al, 2010). Our cohort included a well-defined, population-based sample of the US transplant population, and linkage with corresponding population-based cancer registries allowed for highly complete, uniform cancer ascertainment.

Among the study’s limitations, we note that lymphoma subtype classifications were derived from cancer registry abstractions of medical records, which may not have been standard and may have been affected by changes in lymphoma diagnostic practice over time. Our prior studies of cancer registry classification of lymphoma subtypes suggest good reliability for some subtypes (e.g., DLBCL, follicular lymphoma) (Clarke et al, 2004). However, misclassification would be more likely for less common NHL subtypes and for cases diagnosed in the earlier years of this study before widespread dissemination of the international consensus guidelines (Clarke et al, 2004). In addition, we could not examine risks across the full spectrum of PTLDs, because US cancer registries collect information only for cases deemed malignant by a pathologist. Despite the large number of NHL cases, we did not have adequate numbers of rare subtypes, including most T-cell lymphomas, to reliably assess risk according to age or time since transplant. We note that our estimated risk for overall NHL included CLL, which is now understood to be the same entity as SLL (Turner et al, 2010). Inclusion of CLL/SLL decreased the SIR for overall NHL, which affects comparisons with previous reports (Kasiske et al, 2004; Caillard et al, 2006; Vajdic et al, 2006; Giordano et al, 2007; Serraino et al, 2007; Jiang et al, 2008; Baccarani et al, 2009; Quinlan et al, 2010; Engels et al, 2011). Finally, we lacked information on tumour EBV status, and hence we could not separately examine the risk for EBV-defined lymphoma subtypes.

In conclusion, we found substantial differences in the risk for individual lymphoma subtypes and varying patterns in association with age, transplanted organ, and time since transplantation. These results highlight that NHL should not be considered a single entity in studies of lymphoid malignancy or PTLD after transplant. There is a characteristic clinical spectrum of NHL subtypes among transplant recipients. Because lymphoma treatment varies by subtype, patients suspected of having lymphoma should receive a detailed haematopathologic workup. Our findings also provide new insight into the importance of immunosuppression for the development of some lymphoma subtypes. Future research will require large and representative case series with detailed pathologic classification to untangle the complex effects of host characteristics, oncogenic viral infections, iatrogenic immunosuppression, chronic antigen stimulation, and age-related immune senescence on development of NHL.