MUM1 Expression in Follicular Lymphoma Is a Poor Prognostic Marker in Patients Treated with Immunochemotherapy (SWOG 9800/9911) but Not Chemotherapy Alone (SWOG 8809): A Southwest Oncology Group Correlative Science Study

Background: Follicular lymphoma (FL) is the second most common non-Hodgkin lymphoma in the United States and is typically an indolent lymphoma with a relapsing/remitting course. The FL international prognostic index (FLIPI) predicts outcome in FL. However, some patients with a favorable FLIPI experience unexpectedly rapid progression and short survival. Biologic prognostic markers are needed for better stratification and markers of immune response predict outcome in patients treated with chemotherapy (CT). Therapy has evolved in recent years, with the routine addition of monoclonal antibody therapy to chemotherapeutic agents (immunochemotherapy, ICT). Because biologic prognostic markers may vary by treatment regimen, we evaluated biologic markers of the immune response (FOXP3, CD68) and tumor cells (MUM1) in patients treated with either CT (SWOG 8809) or ICT (SWOG 9800/9911).

Methods: We constructed tissue microarrays with 181 available tissue blocks from SWOG 8809 (CHOP + interferon alpha; n=103) and SWOG 9800/9911 (CHOP+ rituximab/¹³¹I-tositumomab; n=30 and 48 respectively). Immunostains to enumerate regulatory T-cells (Tregs, FOXP3) and lymphoma associated macrophages (LAMs, CD68) were performed. FOXP3 was scored as positive cells/5 high power field (hpf) and by pattern (follicular/perifollicular vs evenly distributed). LAMs were scored as intrafollicular (IF) LAMs/5 hpf and extrafollicular (EF) LAMs/5 hpf. We also evaluated MUM1 expression in tumor cells based on single institution data suggesting MUM1/IRF4 expressing FL demonstrated an aggressive course (Kelley T et al 2007), using 20% positive cells as the cutoff. Associations between potential biomarkers and outcome were assessed using stratified Cox proportional hazards regression.

Results: The median ages were 47.4, 54.8 and 49.8 yrs for S8809, S9800, and S9911, respectively. FLIPI score distributions for S8809 and S9911 were 0 (36%, 35%), 1(38%, 44%), and 2 (26%, 21%); FLIPI score was not available for S9800. S9800 patients were slightly older and S9911 patients were less likely to have elevated LDH than the other studies; otherwise, baseline clinical characteristics were similar across the study populations. The median and interquartile ranges for FOXP3+ cells, CD68 IF LAM, CD68 EF LAM were 169 (106–253), 68 (55–84), and 146 (120–188) respectively. 19.3% of cases showed a follicular/perifollicular pattern of FOXP3. MUM1 expression was seen in 14.4% of cases. Of the clinical variables available for all study populations, age >60 years and LDH>ULN were both associated with worse overall survival (OS, P<.0001). As continuous or dichotomized values (above/below the mean values), LAMs or FOXP3+ cells were not associated with overall survival (OS) in either the entire cohort or in patients receiving IC. Likewise, FOXP3 pattern was also not associated with OS. However, MUM1 expression was associated with poor OS (hazard ratio [HR] 1.91, 95% confidence interval [CI] 1.06–3.46, P=.03). Interestingly, this effect was due entirely to the poor OS of MUM1+ FL in the ICT S9800/S9911group (HR 5.53, 95% CI 2.07–14.75, P=.0006).

Conclusions: MUM1 is expressed in a subset (approximately 15%) of FL. In this series, MUM1 expression appears to identify a group of FL patients with shorter OS in the context of ICT. The biologic basis for this more aggressive behavior is unknown; however, because MUM1 controls expression of cytokines/chemokines involved in both B-cell proliferation and T-cell responses, it may have direct effects on tumor cells as well as tumor immune microenvironment.