Nivolumab for Relapsed/Refractory Diffuse Large B-Cell Lymphoma in Patients Ineligible for or Having Failed Autologous Transplantation

A Single-Arm, Phase II Study

Stephen M. Ansell, MD, PhD; Monique C. Minnema, MD, PhD; Peter Johnson, MD; John M. Timmerman, MD; Philippe Armand, MD, PhD; Margaret A. Shipp, MD; Scott J. Rodig, MD, PhD; Azra H. Ligon, PhD; Margaretha G.M. Roemer, PhD; Nishitha Reddy, MD; Jonathon B. Cohen, MD; Sarit Assouline, MD; Michelle Poon, MD; Manish Sharma, MD; Kazunobu Kato, MD, PhD; Selda Samakoglu, MD, PhD; Anne Sumbul, MS; Andrew Grigg, MD


J Clin Oncol. 2019;37(6):481-489. 

In This Article


Patient Baseline Characteristics and Disposition

Patients were enrolled between March 2014 and August 2015 across 45 sites globally (Appendix Figure A1, online only). Of 121 patients treated, 87 had relapsed/refractory disease after auto-HCT and 34 were auto-HCT ineligible. Baseline characteristics of the two groups are listed in Table 1. In general, patients in the auto-HCT–failed group were younger and had better baseline performance status and longer time since diagnosis than did patients who were ineligible for auto-HCT. Patients had received a median of three prior systemic therapies in each cohort. In the auto-HCT–failed cohort, 40 patients (46%) did not receive any additional lines of systemic or radiation therapy between auto-HCT and the initiation of nivolumab. In the auto-HCT–failed group, 40% of patients had germinal center B-cell–like (GCB) subtype and 32% had non-GCB subtype disease, and in the auto-HCT–ineligible group, 56% had GCB and 18% non-GCB subtype disease. Data were missing in the remainder. All but one patient who had evaluable samples were EBV negative (70 of 121 patients), one patient was EBV positive, and the remainder (n = 50) had either nonevaluable data, mixed results (more than one sample), or missing data.

Figure A1.

Flow diagram.

Median number of doses of nivolumab received were four (range, one to 44 doses) in the auto-HCT–failed cohort and three (range, one to 22 doses) in the auto-HCT–ineligible cohort. At the time of analysis, seven patients (8%) in the auto-HCT–failed group were still on treatment, for whom the median duration of treatment was 14 months from the first study dose. No patients in the auto-HCT–ineligible cohort were still on treatment. Disease progression was the primary reason for discontinuing treatment (83%).


At a median follow-up of 9 months (range, 0.1 to 25 months) in the auto-HCT–failed cohort and 6 months (0.2 to 24 months) in the auto-HCT–ineligible cohort, IRC-assessed ORRs were 10% and 3%, respectively (Table 2). Nearly one third of patients (31%) in the auto-HCT–failed cohort had IRC-assessed stable disease or better response (Figure 1A). Of the nine responders in the auto-HCT–failed cohort, CR occurred in three, with durable response and DOR of 11 or more months, 14 or more months, and 17 months at data cutoff. Of these three patients who achieved CR, two were still on treatment and one developed myelodysplastic syndrome unrelated to study drug and subsequently died. The other six patients achieved PR, with a median DOPR of 7 months (95% CI, 3 to 11 months). The one patient with IRC-assessed PR in the auto-HCT–ineligible group had a DOR of 8.3 months. Patients who achieved CR or PR had early responses, with a median time to response of 1.9 months (Figure 1C). Responses were observed in both the GCB subtype (CR in one and PR in four patients) and non-GCB subtype (CR in one and PR in three patients). Among responders, eight were EBV negative, one was not evaluable, and one had missing data (Appendix Table A1, online only). One patient with PR had transformed lymphoma. Among the seven patients who were still on treatment, two had CR, two PR, and three stable disease. Nineteen patients in the auto-HCT–failed cohort and five patients in the auto-HCT–ineligible cohort were treated with nivolumab beyond disease progression—none achieved response after progression.

Figure 1.

Change in target lesion burden per independent radiology review committee by best overall response in all response-evaluable patients. Best reduction in target lesion in patients (A) who experienced failure with autologous hematopoietic cell transplantation (auto-HCT) and (B) who were ineligible for auto-HCT. Tumor burden change over time in patients (n = 9) of the auto-HCT–failed cohort who had (C) complete remission (CR) or partial remission (PR) and (D) stable disease (SD; n = 15). Horizontal reference line indicates 50% reduction consistent with a response per revised 2007 International Working Group criteria. Response-evaluable patients had assessments at baseline and at one or more postbaseline time point.

Median PFS was 1.9 months in the auto-HCT–failed cohort and 1.4 months in the auto-HCT–ineligible cohort, with a 6-month IRC-assessed PFS rate of 19.1% and 5.2%, respectively (Figure 2A). Median OS was 12.2 months in the auto-HCT–failed cohort and 5.8 months in the auto-HCT–ineligible cohort, and 6-month OS rate was 67% and 47%, respectively (Figure 2B).

Figure 2.

(A) Progression-free survival (PFS) and (B) overall survival (OS) as assessed by independent radiology review committee. Symbols represent censored observations. auto-HCT, autologous hematopoietic cell transplantation.


Grade 3 or 4 AEs were reported in 62% of all treated patients, of which 24% were deemed to be treatment related; none had a frequency of more than 5% (Table 3). The most common treatment-related AEs (TRAEs) were nausea (17%), fatigue (17%), and diarrhea (10%)—these were generally not severe (majority grade 1 to 2). Treatment-related SAEs were reported in 14 patients (12%).

Four patients (3%) had TRAEs as the primary cause of discontinuation of treatment. TRAEs that contributed to treatment discontinuation included neutropenia, thrombocytopenia, diarrhea, pancreatitis, lipase increase, and dermatitis psoriasiform.

Rates of immune-mediated AEs were low in all treated patients with extended follow-up. The most common grade 3 to 4 immune-mediated AEs were nephritis and renal dysfunction (4%), hepatitis and hepatic dysfunction (3%), diarrhea (3%), and rash (2%).

Overall, 78 patients died. The primary reason for death was disease progression in 39 (45%) and 29 (85%) patients in the auto-HCT–failed and auto-HCT–ineligible cohorts, respectively. No death was attributable to study drug toxicity.

FISH and Immunohistochemistry

In total, 74 of 112 patients had evaluable tumor biopsy specimens for 9p24.1 FISH; 34% of DLBCL specimens were disomic and 45% were polysomic at PD-L1/PD-L2. PD-L1/PD-L2 polysomy in these DLBCLs was low level, largely only three copies in these tumors. Only 16% of DLBCLs exhibited copy gain, which was also low level in the majority of cases (three copies in nine of 13 cases). An additional 3% of tumors had amplification, including PDL2–selective amplification in one patient. Patients were classified by the highest-level 9p24.1 alterations (Figs 3A and 3B), as previously described.[11,23] DLBCLs with 9p24.1 amplification had additional tumor cells with copy gain (0% to 86%) and/or disomy (4% to 8%). Tumors with relative copy gain had additional cells with polysomy (2% to 58%) and/or disomy (2% to 82%), and DLBCLs that were categorized as polysomic had additional disomic tumor cells (14% to 98%; Figure 3B). The percentage of residual disomic cells was lowest in DLBCLs with amplification, intermediate in tumors with copy gain, and highest in tumors with polysomy (P < .001; Figure 3C), which is consistent with an ordered spectrum of 9p24.1 genetic alterations.

Figure 3.

Prevalence and spectrum of 9p24.1 genetic alterations and association of residual disomy with 9p24.1 genetic categories. (A) Prevalence of 9p24.1 genetic alterations in evaluated diffuse large B-cell lymphomas (DLBCLs). (B) Spectrum of 9p24.1 alterations in evaluated DLBCLs. Each patient is classified by the highest observed level of 9p24.1 alteration in tumor cells: polysomy, copy gain, amplification (Ampl), or rearrangement (Rearr). Individual patients are visualized as columns on the x-axis. Percentages of tumor cells with monosomy/relative copy loss (gray), disomy (black), polysomy (light red), copy gain (medium red), amplification (dark red), and rearrangement (brown) are depicted on the y-axis. In cases with evaluable 9p24.1 status and PD-L1 immunohistochemistry (n = 46), programmed death ligand 1 (PD-L1) expression (H-score) on PAX5-positive malignant B cells is indicated below the x-axis (membranous PD-L1 in red, cytoplasmic PD-L1 in blue). (C) Percentage of tumor cells with residual 9p24.1 disomy in DLBCLs classified by 9p24.1 genetic categories (P < .001 from ordinary one-way analysis of variance of unpaired t test data). Coampl, coamplification; PD-L2, programmed death ligand 2.

In this series of DLBCLs with infrequent low-level 9p24.1 alterations, membranous PD-L1 expression was only detected in four (9%) of 46 evaluable cases. One of the three patients who achieved CR had high-level 9p24.1 amplification, whereas the other two had normal 9p24.1 copy number or an unavailable biopsy specimen. Of the seven patients who achieved PR, five had available biopsy specimens and either normal 9p24.1 copy numbers (one patient), low-level polysomy (three patients), or copy gain (one patient). The five evaluable patients with PR had no detectable PD-L1 expression on tumor cells (Appendix Table A1).