Central Nervous System (CNS) Prophylaxis Does Not Decrease the Rates of CNS Relapse From Diffuse Large B-Cell Lymphoma In the Era of R-CHOP

While some clinical characteristics/sites of diffuse large B-cell lymphoma (DLBCL) are associated with increased rates of central nervous system (CNS) relapse, the role/benefits of CNS prophylaxis are controversial, particularly in the era of better disease control with R-CHOP chemotherapy. We evaluated the benefits of high dose methotrexate (HDMTX 3 g/m2) and intrathecal (IT) chemotherapy (MTX 12 mg) as primary CNS prophylaxis (CNSPr) in patients with DLBCL (de novo or transformed) treated with curative intent R-CHOP chemotherapy between the years of January 2002-December 2008. During this period, we adopted a local ‘informal' clinical practice recommendation for CNS prophylaxis for patients with testicular involvement, increased lactate dehydrogenase enzyme (LDH) and greater than 1 extranodal (EN) site, epidural disease, invasive sinus or skull involvement. Compliance with and efficacy of this treatment recommendation was the catalyst for this retrospective audit.

Using the cancer pharmacy database, we retrospectively identified 214 patients who received 1–8 cycles of R-CHOP chemotherapy (median 6) for DLBCL. Patients with transformed histologies were included if they had not yet received R-CHOP chemotherapy. Patients with HIV or CNS involvement at diagnosis were excluded.

The median age was 64 with 54% male patients. 71% had stage III-IV disease, 49% had an elevated LDH, 57% EN disease (35% >1 EN site), 49% had high or high-intermediate international prognostic index scores (IPI), 14% had transformed disease, 8 patients had testicular lymphoma, 11% had increased LDH + > 1 EN site. 27 patients (12.6%) received some form of CNS prophylaxis (37% IT MTX alone (median 1.5 times (1-3)); 7% with HDMTX 3g/m² alone (median 1.5 times (1-3)) and 56% with both HDMTX and IT chemotherapy (median 2 HDMTX (1-6) and 3 IT chemotherapy (1-9)). Compared with patients that did not receive CNSPr, patients that did had higher stage disease (p=.0061), more EN disease (P <0.0001), more testicular involvement (p<0.0001), higher IPI (p=.029), age adjusted IPI (aaIPI) (p=.048) and revised international prognostic index (R-IPI) (p=.016). Of those deemed to be at higher risk of CNS relapse defined by high IPI (20%; Haioun et al. 2000), high-intermediate and high aaIPI (52%; Feugier et al. 2004), or increased LDH and > 1 EN site (11%; Van Besien et al. 1998), 23%, 18% and 29% received CNSPr respectively, demonstrating imperfect compliance with local practice guidelines. 75% of patients with testicular lymphoma received CNSPr, 83% inclusive of both HDMTX (median 2) and IT chemotherapy (median 3). Eight patients (3.7% of all patients) relapsed in the CNS at a median time of 17 months (6-35 months range). Five patients developed parenchymal CNS relapse, 2 had leptomeningeal disease and 1 had both parenchymal and leptomeningeal involvement. The relapse rates in those that received or did not receive prophylaxis were 7.4% (2/27) and 3.2% (6/187) respectively. Six out of the 8 relapses were isolated relapses in the CNS and 4/8 were in testicular lymphoma patients. If the 8 testicular lymphoma patients were excluded, the overall rate of CNS relapse was 1.9% (0% in the 21 with prophylaxis and 2% in the 185 that did not). 62% (5/8) of those with CNS relapse have died with a median survival post CNS relapse of 2 months (range 0.5–16). Of the 4/8 patients with testicular involvement that relapsed, 3 had received CNS prophylaxis with HDMTX and IT chemotherapy (median 2 (range 1–5)). By multivariate analysis, testicular involvement was the only negative risk factor for CNS relapse (HR 33.5, p<.0001 (95% CI 8.3–135).

R-CHOP chemotherapy may negate the need for CNS prophylaxis in patients with DLBCL, even those formerly identified as higher risk using standard prognostic scoring systems with the exception of testicular lymphoma. Better forms of CNS prophylaxis are needed in these patients. CNS relapses appear to occur later as isolated parenchymal events compared with the pre rituximab era, but survival post CNS relapse remains short.

No relevant conflicts of interest to declare.