Cell-of-origin effect of polatuzumab vedotin in diffuse large B-cell lymphoma: no ordinary subgroup analysis

Polatuzumab vedotin (polatuzumab) is an antibody-drug conjugate comprising monomethyl auristatin E (MMAE) linked to an anti-CD79b monoclonal antibody. Polatuzumab demonstrated improved progression-free survival (PFS) in combination with rituximab, cyclophosphamide, doxorubicin, and prednisone (Pola-R-CHP) for upfront treatment of diffuse large B-cell lymphoma (DLBCL) in the phase 3 POLARIX trial, with no difference in overall survival to date.¹ Seminal gene expression profiling studies revealed 3 DLBCL subtypes based on the cell-of-origin (COO): germinal center B-cell (GCB) type, activated B-cell (ABC) type, and not otherwise specified.² COO-informed clinical trials exploring the addition of novel agents, such as ibrutinib (NCT01855750), lenalidomide (NCT02285062), or bortezomib (NCT01324596), to standard rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) were not practice changing; however, ongoing studies seek to further COO-informed approaches as well as prospectively assess the impact of additional molecular subclassifications of DLBCL.³ 

As highlighted at the March 2023 US Food and Drug Administration Oncologic Drugs Advisory Committee meeting regarding polatuzumab, subgroup analysis of POLARIX suggests that Pola-R-CHP offers greater benefit to patients with ABC-type DLBCL than those with GCB-type DLBCL, where it may or may not offer a benefit (hazard ratio [HR] for PFS in ABC: 0.34; 95% confidence interval [CI], 0.13-0.85; vs HR in GCB: 1.18; 95% CI, 0.75-1.84; HR for overall survival in ABC: 0.27; 95% CI, 0.06-1.26; vs HR in GCB: 1.64; 95% CI, 0.87-3.07). This relationship persisted in the POLARIX Asian extension cohort, and a recent meta-analysis highlighted the large magnitude of this differential effect.⁴ However, as POLARIX did not formally test for an interaction between polatuzumab efficacy and COO, the Pola-R-CHP US Food and Drug Administration approval is COO agnostic.

Clinical trial subgroup analyses typically seek to ensure that no patient subgroup experiences a substantially different outcome because of an interaction between the experimental treatment and a pathobiological or clinical factor.⁵ Subgroups are ideally prespecified before the trial, and subgroup analyses play different roles in interpreting trials with positive vs negative results: for positive trials, they can help ensure that the observed benefit applies equally to all participating patients, whereas in negative trials, they can help generate hypotheses for subsequent testing in adequately powered prospective studies.⁶ In the context of a trial with a negative result for the primary end point, subgroup analyses should only be hypothesis generating because they are frequently underpowered to provide statistically valid answers, and the unknown number of unreported subgroup analyses performed leads to multiplicity problems and uncontrolled type 1 error.⁷ US Food and Drug Administration guidance emphasizes that the clinical importance of a subgroup analysis must be put in the context of the representation of the subgroup in the intention-to-treat population and the plausibility of a differential effect in the subgroup.⁶ 

Here, we review preclinical and clinical studies of polatuzumab in DLBCL to provide a nuanced interpretation of the COO subgroup analysis of POLARIX, a trial that met its primary end point.

For variables not expected to interact with polatuzumab, such as patient sex or geography, subgroup analyses are reassuring that there is no differential efficacy. In contrast, strong preclinical evidence predicted the benefit of polatuzumab in ABC-type DLBCL observed in POLARIX. Key genetic and functional studies revealed that chronic active B-cell receptor (BCR) signaling is uniquely required for survival in ABC-type DLBCL.⁸ Furthermore, ABC-type, but not GCB-type, DLBCL is highly susceptible to either (1) genetic knockdown of proximal BCR subunits, including CD79b, a phenotype not linked to specific BCR pathway mutations; or (2) targeted inhibition of BCR signaling by small-molecule drugs such as ibrutinib or dasatinib. Although subsequent work confirmed that both ABC- and GCB-type DLBCL are sensitive to killing by the MMAE payload,⁹ ABC-type DLBCL appears uniquely positioned to be targeted via dual mechanisms by polatuzumab: microtubule disruption by MMAE as well as BCR signaling abrogation, via CD79b internalization triggered by antibody-drug conjugate binding. Thus, mechanistically, replacing vincristine in R-CHOP with polatuzumab may only function as a “like-for-like” substitution in GCB-type DLBCL, compared with the potential additive mechanistic benefits from polatuzumab in ABC-type DLBCL.

Multiple early trials of polatuzumab in relapsed/refractory DLBCL suggest greater benefit from polatuzumab in ABC-type DLBCL (Table 1). In the phase 2 ROMULUS study,¹⁰ the overall response rate to rituximab plus polatuzumab was 83.3% (10/12) in patients with ABC-type DLBCL, compared with 45.5% (5/11) in patients with GCB-type DLBCL. Subsequently, the phase 2 GO29365 study, which led to polatuzumab’s accelerated approval, randomized patients to receive bendamustine plus rituximab (BR) with or without polatuzumab.^11,12 The benefit of polatuzumab was strongly driven by COO: median PFS for patients with ABC-type DLBCL was 10.76 months with BR-polatuzumab vs 1.97 months with BR alone (HR, 0.20; 95% CI, 0.09-0.45), compared with 2.5 months with BR-polatuzumab vs 1.87 months with BR alone (HR, 0.49; 95% CI, 0.23-1.05) in GCB-type DLBCL. Multiple retrospective observational studies of polatuzumab also support this relationship between COO and polatuzumab efficacy (Table 1), irrespective of whether COO is determined by gene expression profiling via NanoString Lymph2Cx assay (as used in POLARIX) or the surrogate immunohistochemistry-based Hans algorithm used in routine clinical practice, which may misclassify 10% to 20% of cases.^13-18 

On the basis of POLARIX, it is reasonable to consider patients with ABC-type DLBCL for Pola-R-CHP where it is available. Whether to use Pola-R-CHP outside ABC-type DLBCL remains unsettled.¹⁹ In POLARIX, all patients received prophylactic granulocyte colony-stimulating factor (G-CSF). Despite this, rates of febrile neutropenia were modestly higher (14.3% vs 8%; P = .004 by unadjusted Fisher exact test) in patients receiving Pola-R-CHP. R-CHOP is routinely given without prophylactic G-CSF for fit patients aged <60 years, and G-CSF can lead to clinically significant bone pain.²⁰ 

In addition to this modest potential increase in toxicity, the high price of polatuzumab, along with its marginal benefit (absolute PFS improvement of ∼6.5% in all comers), brings the cost-effectiveness of Pola-R-CHP into question. Cost-effectiveness analyses of Pola-R-CHP that assume uniform benefit of Pola-R-CHP across patient subgroups have produced mixed results,^21,22 but Pola-R-CHP would be more cost-effective if its use were restricted to ABC-subtype DLBCL, as confirmed in a recent study in the European context.²³ This may be of importance to the many jurisdictions that consider cost-effectiveness in decisions regarding novel therapy reimbursement.

Beyond the frontline setting, polatuzumab retains its indication for relapsed/refractory DLBCL, and it is used frequently as bridging therapy before chimeric antigen receptor T (CAR-T) cell treatment.²⁴ Considering COO in optimizing bridging therapy may be important, because better disease control at the time of CAR-T infusion improves post–CAR-T outcomes,²⁵ and avoiding ineffective and potentially toxic therapy is critical to minimize post–CAR-T cytopenias.

Despite the limitations of these data, we posit that the greater benefit of polatuzumab in ABC-type DLBCL in POLARIX is an expected signal confirming the phenomenon seen across multiple studies. However, key questions remain. How should Pola-R-CHP be integrated into frontline DLBCL treatment? This regimen adds substantial financial and potentially modest symptomatic toxicities, so its optimal use is in patients who clearly derive benefit from it. We encourage studies of polatuzumab to report efficacy stratified by COO, and clinicians to consider whether polatuzumab’s data in GCB-type DLBCL are sufficient to convince them of its benefit in the wider DLBCL population. Improving the accuracy and resolution of DLBCL subclassification methods and determining their clinical utility should be a priority for lymphoma researchers. How is polatuzumab best used in relapsed/refractory DLBCL? Experience targeting ibrutinib and lenalidomide to patients with ABC-type DLBCL provides precedent for applying predictive biomarkers in DLBCL. With expanding treatment options for DLBCL—mostly without head-to-head comparisons—we must incorporate all available information to maximize patient outcomes.

Contribution: All authors conceived of and performed the research and wrote the article.

Conflict-of-interest disclosure: D.A.R.-G. is supported by an American Society of Clinical Oncology (ASCO) Young Investigator Award and an Lymphoma Research Foundation (LRF) Post-doctoral Fellowship, and serves as a consultant for Cartography Biosciences. E.R.S.C. receives research funding from Arnold Ventures outside the submitted work. N.L.B. receives research funding not relared to this work from ADC Therapeutics, Autolus, BMS/Celgene, Forty-Seven, Gilead/Kite, Janssen, Merck, Millennium, Pharmacyclics, Roche/Genentech, and Seattle Genetics, and serves on advisory boards for ADC Therapeutics, Foresight Diagnostics, Kite, Roche/Genentech, Seattle Genetics, BTG, and Acerta.

Correspondence: David A. Russler-Germain, Washington University School of Medicine, 660 S Euclid Ave, Campus Box 8056, St. Louis, MO 63110; email: germaind@wustl.edu.