Post-transplant lymphoproliferative disorders (PTLD) constitute a diverse entity with no consensus on optimal management. An accurate but user-friendly prognostic index would facilitate clinical-decision making and comparison across trials. Identification of prognosticators from past studies is difficult owing to the lack of uniform criteria, including transplant type. This confounds analysis as each transplant sub-type has distinct features including organ source, duration and extent of immunosuppression, patient selection, risk to allograft of involvement with PTLD, and scope for reduction of immunosuppression following diagnosis of PTLD. The most frequently performed solid organ transplant (SOT) is kidney allografting. Although risk factors for PTLD development post-renal transplant have been identified (Caillard et al, 2006), to our knowledge, no head to head comparison of prognostic indices for survival has been performed. We devised a novel ‘renal-PTLD’ index utilizing B symptoms and lactate dehydogenase (LDH), and compared this with published prognostic models (The International Non-Hodgkin’s Lymphoma Prognostic Factors Project, 1993; Leblond et al, 2001; Ghobrial et al, 2005; Choquet et al, 2007) in 42 consecutive PTLD patients following renal transplant. The indices were applied to a series of unselected patients with PTLD following renal transplant in a single-centre. Follow-up was longer than in any of the previously reported series (Leblond et al, 2001; Ghobrial et al, 2005; Choquet et al, 2007).
All biopsies underwent central review by a lymphoma pathologist. Patient characteristics are outlined in Table I. At the time of analysis 49% of patients had died. Median follow-up for surviving patients was 76 months (3–268 months). Five patients died prior to receiving therapy. Determination of prognostic factors was conducted using the Kaplan–Meier and Cox regression methods.
|At transplant (median, range)||36 years (17–66)|
|At PTLD (median, range)||42 years (17–70)|
|<25 years||3 (7%)|
|25–60 years||34 (81%)|
|>60 years||5 (12%)|
|Gender||28M : 14F|
|Cyclosporin A||34 (81%)|
|Other monomorphic disease||13 (31%)|
|Cell of origin|
|EBV tissue status (data available in 40 patients)|
|OKT3 and EBV+ve||8/11 (73%)|
|Time of onset|
|Overall (median, range)||6·3 years (0·1–27·1)|
|<1 year||10 (24%)|
|EBV+ve (median, range)||1·9 years (0·1–14·7)|
|EBV−ve (median, range)||10·6 years (0·3–23)|
|Site of disease|
|Extranodal (including graft)||35 (92%)|
|Nodal and extranodal||11 (28%)|
|Graft organ||4 (10%)|
|Bone marrow||13 (40%)|
|Stage 1||10 (25%)|
|Stage 2||4 (10%)|
|Stage 3||3 (8%)|
|Stage 4||23 (58%)|
|B symptoms (data available in 39 patients)||17 (44%)|
|PS >1||21 (50%)|
|PS >2||17 (40%)|
|LDH (data available in 40 patients)|
|Median, range||227 (113–4044)|
|B2-microglobulin (data available in 28 patients)|
|Median, range||5·5 (1·6–68·4)|
|Number of sites (data available in 37 patients)|
|Immunosuppression reduction||34 (81%)|
|Chemotherapy (anthracycline based)||26 (62%)|
|Chemotherapy + rituximab||6 (14%)|
By univariate analysis, only serum LDH > normal, performance status (PS) >1 or >2 and presence of B symptoms (one or more of drenching night sweats, >10% unintentional weight loss in the preceding 6 months and fever >38°C) were significantly associated with survival (Log Rank P = 0·011; P < 0·0001; P < 0·0001; P = 0·003 respectively). Neither age, stage, extranodal disease, number of disease sites, time to PTLD, monomorphic disease, graft organ involvement, Epstein–Barr virus (EBV) tissue status, cell of origin or beta-2-microglobulin (B2M), were discriminatory. Our data set was nearly complete for all parameters with the exception of B2M and this parameter should be tested in future prognostic studies in PTLD patients.
Investigation of multiple factors associated with survival via Cox regression indicated that PS >1 and presence of B symptoms were highly correlated (P = 0·002), and hence could not be used in the one regression model. Presence of B symptoms remained significantly associated with survival after adjustment for age, sex and LDH (P = 0·03). LDH was also significantly associated with survival in this model (P = 0·04), and was combined with presence of B symptoms into a new index (Fig 1A). We identified a significant change in survival between sub-groups (P = 0·005). There was a progressive reduction in survival with the number of positive factors, and median survival was not reached (group 0), 5·4 years (group 1) and only 0·5 years (group 2). When comparing the five indices, all but the Ghobrial index were able to categorize our patients into three approximately equal size groupings and enabled discrimination into low, intermediate and high-risk groupings with highly significant P-values (Fig 1A–E).
Leblond et al (2001) rationalized that the International prognostic index (IPI) was originally applied to lymphomas in immunocompetent patients, and proposed an index consisting of PS >1 and numbers of involved sites. In their series of 61 PTLD patients who had undergone a variety of SOT (including 44% non-renal PTLD), IPI was not of value. The cohort reported by Choquet et al (2007), which had 45% non-renal PTLD, included only patients receiving rituximab monotherapy following failure of primary therapy. Their index included age >60 years, PS >1, and LDH > normal. Ghobrial et al (2005) found Eastern cooperative group PS >2, graft organ involvement and monomorphic disease (although the latter was not significant by their univariate analysis) had superior discriminatory value to the IPI. Their study included 35% renal SOT patients. Our study is in concordance with the findings of other groups, that monomorphic histology is not an adverse prognosticator (Leblond et al, 2001; Tsai et al, 2001; Choquet et al, 2007). None of the three studies mentioned above (Leblond et al, 2001; Ghobrial et al, 2005; Choquet et al, 2007) examined the impact of B symptoms on survival, but it has been previously observed to independently influence survival in a study of 42 PTLD patients (Tsai et al, 2001).
A key feature of our analysis is the restriction to renal SOT recipients. Renal transplant patients generally receive markedly reduced immunosuppression with respect to their other SOT counterparts, and when faced with organ rejection transplant, physicians have the option to sacrifice the kidney and use dialysis. Furthermore, renal SOT patients tend to be older, meaning a reduced risk of primary EBV infection (a major risk-factor for early-onset PTLD). Although our study is at variance with that of Ghobrial et al (2005) regarding the discriminatory value of their proposed standard (Fig 1B), we concur with them and Choquet et al (2007) that the IPI retained prognostic value (Fig 1C). We also found that the indices reported by Leblond et al (2001) and Choquet et al (2007) allows division into groups with significant difference in survival (Fig 1D and E). The indices of Choquet et al (2007) and Leblond et al (2001) are easy to use and can readily be applied by transplant centres participating in multicentre clinical trials. Indeed, the parameters used to make up these indices have striking overlap with the IPI and age-adjusted IPI. Our data suggest that our index may also have applicability to PTLD, at least after renal transplant. In our retrospective cohort, approximately only one-third received rituximab therapy. Nowadays, after failing immunosuppression, patients are increasingly treated with rituximab either as monotherapy or in combination with chemotherapy. It would therefore be important to validate our model in the rituximab era. Indeed, until ours and the other proposed models have been prospectively validated in larger clinical studies, the IPI should continue to be considered the prognostic index of choice for PTLD occurring after renal SOT.