Prognostication, Survival and Treatment-Related Outcomes in HIV-Associated Burkitt Lymphoma (HIV-BL): A US and UK Collaborative Analysis

Introduction: There are few data about prognostication and outcomes in patients (pts) with HIV-BL treated in the cART era. Optimal treatment strategies to minimize treatment-related mortality (TRM) remain unclear and current recommendations are based on small studies. We conducted a multicenter international analysis to identify prognostic factors and outcomes in pts with HIV-BL treated in the cART era.

Methods: This retrospective analysis included a subcohort from a recent study across 30 US sites (Evens et al. Blood 2020) augmented by data from 5 UK centers treated 2009-2018. Progression-free (PFS) and overall survival (OS) were estimated by Kaplan-Meier & differences assessed by log-rank test. Univariate (UVA) associations were derived via Cox model and multivariable (MVA) models were constructed by forward selection of significant variables with P<0.05.

Results: 249 (US: 140 & UK: 109) pts with newly diagnosed HIV-BL were included. Clinical features included median age 43 (IQR 35-50 years [yrs]); male sex: 84%; ECOG PS: 2-4: 48%; elevated LDH: 85% (> 3x upper limit of normal (ULN) 49% & >5xULN 39%); >1 extranodal (EN) site: 60%; any CNS involvement (CNSinv) 25%; and +bone marrow (BM) 46%. MYC rearrangement was reported in 93% of pts with t(8;14) in 49%, break-apart probe in 41% and MYC-light chain in 3%; the rest had classical BL with negative MYC testing (4%) or missing result (3%) (otherwise classical BL). Median CD4 count was 217 (IQR 90-392 cells/µL) with 68% pts having CD4>100 cells/µL. At BL diagnosis, HIV viral load was detectable in 55%; 39% of pts were on cART. Baseline features were similar between the US & UK cohorts with significant differences only in ECOG PS 2-4 (32% vs 65%; P<0.001) & baseline CNSinv (30% vs 17%, respectively; P=0.02).

Tx regimens included: CODOX-M/IVAC (Magrath) 60%, DA-EPOCH 25%, HyperCVAD/MA 13%, & other 1%; most pts (87%) received rituximab (R). Similar regimens were used in pts with baseline CNSinv: Magrath 64%, DA-EPOCH 24% & HyperCVAD 12%. In the US, pts most frequently received DA-EPOCH (42%) followed by Magrath (32%) & HyperCVAD/MA (24%), whereas in the UK, 96% received Magrath. R was more frequently given in the US (94% vs 79%, P<0.001). Similar baseline features were seen in US pts selected for DA-EPOCH as those selected for Magrath or HyperCVAD/MA except for lower median CD4 count (144 vs 260 cells/µL; P=0.04). Overall response to Tx was: CR 70%, PR 9%, PD 14%, not evaluable 7%. TRM was 18% following HyperCVAD/MA, 13% after DA-EPOCH & 7% in patients treated with Magrath. Overall, 33% of pts had a relapse of HIV-BL with 23% systemic only & 10% CNS.

With median follow-up of 4.5 yrs, 3-yr PFS & OS were 61% & 66%, respectively, and nearly identical in both countries (Fig A). Pts with CD4>100 cells/µL had better 3-yr PFS (Fig B) & OS (68% vs 57% P=0.03). We observed significantly worse outcomes in pts with baseline CNSinv (3-yr PFS 36% vs 69%, P<0.001; OS 41% vs 73%, P<0.001; Fig C). Magrath was associated with the highest 3-yr PFS (66%) compared with 63% after HyperCVAD/MA & 51% after DA-EPOCH, but the difference was not significant (P=0.13; Fig D). Pts receiving R had numerically higher PFS, but also not statistically significant (63% vs 53% P=0.16). We observed poor outcomes in pts with baseline CNSinv regardless of frontline Tx (3-yr PFS HyperCVAD/MA 40%, Magrath 39%, DA-EPOCH 32%; P=0.93; Fig E).

The incidence of CNS recurrence at 3 yr across all Tx was 11%. Higher incidence was observed with DA-EPOCH (P=0.032 vs other regimens; Fig F) with no difference according to CD4 count. Variables associated with PFS & OS on UVA included: ECOG PS 2-4, >1 EN, +BM, baseline CNSinv, LDH>ULN, CD4 <100 cells/µL. On MVA, the variables independently associated with inferior PFS were ECOG PS 2-4 (HR 1.87 P=0.007); baseline CNSinv (HR 1.70, P=0.023); LDH >5xULN (HR 2.09, P<0.001); and >1 EN sites (HR 1.58 P=0.043). The same variables were significant on MVA for OS. Adjusting for all of the prognostic variables, Tx with Magrath was associated with longer PFS (adjusted HR, 0.45, P=0.005).

Conclusions: These data represent the largest analysis of HIV-BL to date. There were favorable tolerance and outcomes with intensive R-containing regimens with Magrath being associated with lower TRM and the highest PFS. In addition, prognostic factors for pt outcomes were associated with lymphoma characteristics rather than with HIV-related features. Pts with baseline CNSinv represent a high-risk group with unmet therapeutic needs.