Emerging biomarkers for CD3×CD20 bispecific antibodies in lymphoma Free

Bispecific antibodies (BsAbs) represent a novel class of targeted immunotherapies that are transforming cancer management. Those targeting CD20 on malignant B cells and CD3 on T cells (CD3×CD20 BsAbs) have rapidly entered existing treatment paradigms of B-cell non-Hodgkin lymphoma (B-NHL), in particular, follicular lymphoma (FL) and diffuse large B-cell lymphoma (DLBCL). Landmark phase 2 trials have led to accelerated approvals across multiple jurisdictions for 4 CD3×CD20 BsAbs including epcoritamab, glofitamab, mosunetuzumab, and odronextamab (Table 1).^1-13 

Despite mechanisms of BsAbs being conceptionally straightforward, that is, redirecting and recruiting T cells to their malignant cell target, and published extensive preclinical data, optimal patient selection for BsAb therapy remains difficult.^2,5,9,12 While numerous variables influence efficacy, predictive methods to delineate populations who benefit or develop resistance, are lacking. Identifying these “biomarkers” is critical to progress efficacious, yet appropriate, personalized use of these high-cost treatments, particularly as their indications expand.

This review explores emerging evidence of markers associated with outcomes from CD3×CD20 BsAbs focusing on results from approved indications of DLBCL and FL (Figure 1; Table 2).

The FL International Prognostic Index (FLIPI) and International Prognostic Indices (IPI and its variants) for DLBCL are validated prognostic tools in immunochemotherapy treated patients, however their utility is less clear in the context of BsAbs.^26,27 Studies of BsAb-treated patients suggest a higher FLIPI or IPI score are associated with inferior responses, however the contribution of particular individual components of these indices in overall prognosis is challenging to isolate in the small cohorts reported.^3,7,8 

Poor performance status is consistently associated with inferior survival, as with most therapies.¹⁴ The impact of sex on efficacy is unclear. While some studies have suggested a trend toward a higher complete response (CR) in females, others have not.^3,4,7,8 

In contrast to the worse outcomes reported with advanced age in conventional chemotherapy, studies have shown superior responses in older patients receiving BsAbs.^1,3,7,8 In a dedicated subgroup analysis of mosunetuzumab for relapsed/refractory (R/R) FL, those ≥65 years demonstrated improved responses compared to younger patients with objective response rates (ORR) 87% vs 77% and CR 70% vs 55%.²⁸ Similar findings (numerically better ORR and CR compared to younger cohorts) are also seen in older cohorts treated with BsAb-chemotherapy combinations.^29,30 This may be due to the tolerability of BsAbs rather than specific biological characteristics, as evident by retained efficacy and safety in cohorts >80 years.³¹ When considering these data it is worth noting that older patients enrolled in clinical trials are often highly selected and may perform favorably due to superior fitness compared to those treated in standard care.

Pretreatment disease burden has consistently influenced BsAb outcomes. Advanced stage (III/IV) disease is associated with numerically lower CR rates compared to limited-stage disease.^1,16,22 This is more pronounced for DLBCL than FL. Notably, extranodal involvement is yet to show a significant alteration to response.^4,15 Rather than site of disease, total burden may be more meaningful for response.

In BsAb monotherapy cohorts, disease bulk (≥6 cm) does not impact response rates for R/R DLBCL and R/R FL.^1,4 However, when utilized in combination with chemotherapy, both mosunetuzumab-polatuzumab vedotin (R/R DLBCL) and mosunetuzumab-CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisolone; de novo DLBCL) achieved lower CR rates with bulky disease (17% vs 48% and 62% vs 96%, respectively) noting larger bulk definitions (≥7.5-10 cm).^15,16 In addition, retrospective analyses have shown disease bulk impacts on progression-free survival (PFS), notwithstanding variable bulk definitions and limited cohort numbers.¹⁴ 

Surrogate biochemical disease burden assessment with pretreatment lactate dehydrogenase (LDH) is an established adverse risk factor in BsAb-treated DLBCL.^4,14,15 In prospective studies of glofitamab and mosunetuzumab, normal LDH was associated with higher CR rates.^4,15 Conversely, elevated LDH has been associated with inferior PFS and overall survival (OS).^14,19 The prognostic utility of LDH in BsAb-treated FL cohorts remains unclear.

Quantitative positron emission tomography/computed tomography (PET/CT) disease burden assessment using 3-dimensional total metabolic tumor volumes (TMTV) has established close associations with survival in immunochemotherapy-treated B-NHL. In glofitamab-treated R/R DLBCL patients a baseline TMTV above or equal to the study cohort median demonstrated inferior 12-month PFS of 16.8% compared to 50.1%.¹⁷ In addition, elevated TMTV serves as a predictive biomarker for the life threatening BsAb toxicity of grade 2 or higher cytokine release syndrome (P < .0001).¹⁷ 

As expected, depth of response to therapy is an important determinant of long-term outcomes in both FL and DLBCL. Achieving CR on PET/CT is associated with better survival compared to a PR, which is mirrored in studies using sensitive quantitative measurement of depth of response with measurable residual disease. In studies of both FL and DLBCL, those who achieve measurable residual disease negativity after treatment consistently demonstrate longer PFS.^{4,7,8,10,13,22} 

As with many therapies in lymphoma relapse, previous treatment-refractory status portends adverse outcomes to BsAbs; with fewer CRs and inferior PFS.^4,7,14 Prior chimeric antigen receptor T-cell (CAR-T) therapy also confers reductions in CR (−7% to −27%) and OS and for those receiving BsAbs after CAR-T therapy, recent real-world studies show relapse within 3 months of CAR-T administration confers worse outcomes.^{3,4,8,11,14,19,32} The duration of response to CAR-T as a marker of BsAb efficacy likely reflects poor disease biology as is commonly the case for other therapies. Even so, durable responses to BsAb are described in primary refractory and CAR-T exposed cohorts, so BsAbs should still be considered. Continued uncertainties remain regarding optimal CAR-T and BsAb sequencing in R/R disease noting that CAR-T responses are maintained post BsAb therapy, even when utilized in later treatment lines.^33-35 

Prior use of bendamustine, a commonly employed therapy in FL, and notorious for its lymphodepleting properties, has not been shown to alter BsAb-associated ORR, PFS, or OS.^14,19,36,37 Large studies from prospective homogeneous data sets, coupled with analysis of recent administration (ie, within 3 months), are required to confirm true effect.

More specific to CD20-targeting therapies, longer intervals between last anti-CD20 treatment and BsAbs correlates with disease responsiveness. Both epcoritamab and mosunetuzumab have shown inferior response rates for patients specifically refractory to recent anti-CD20 therapy, particularly if administered within 3 months.^3,8,16 CD20 loss is a well described resistance mechanism for anti-CD20 antibodies, discussed below in the context of BsAb failure.^38,39 

Absent CD20 expression is strongly associated with lack of BsAb response in the majority of studies, however exceptions exist, with cases of negligible CD20-expressing R/R FL responding to odronextamab.^3,4,13 Although rare, these may be explained by issues with immunohistochemistry (IHC) reliability, interobserver variability or potential intrapatient heterogeneity with testing of an unrepresentative tumor site. Further, there is potential disparity between CD20 IHC vs messenger RNA expression, a more sensitive measure of cellular CD20.²⁴ Regardless, when CD20 expression is present, in vitro de novo tumor studies of BsAbs demonstrate variable cytotoxicity regardless of the degree of CD20 expression.^20,24,40 Although required, the exact degree of CD20 expression for optimal response remains unclear as does the role of CD20 expression loss as a preemptive treatment failure marker.⁴¹ Interesting work has been done to understand the potential biological factors contributing to diminished responses to CD3×CD20 BsAbs and include masking of potential CD20 binding sites as well as inducement of mutations in MS4A1, which codes for CD20 and causes CD20 epitope disruption and/or truncation.^18,23 Due to the importance of CD20 for response, tumor biopsy before BsAb commencement is recommended.

DLBCL cell-of-origin is prognostic in patients treated with numerous regimens. IHC determined non–germinal center (GC)-DLBCL and RNA-sequencing activated B-cell–DLBCL both demonstrate inferior outcomes compared to GC-DLBCL.^42,43 However, the majority of CD3×CD20 BsAb DLBCL studies have identified higher response rates in IHC-defined non–GC-DLBCL.^4,8,16 It is important to note that when cell-of-origin is assessed using RNA-sequencing, no correlation with outcomes was identified, albeit with a small sample size.²⁰ 

Irrespective of upfront or R/R DLBCL settings, double-expressor, double-hit or triple-hit DLBCL are associated with lower response rates to mosunetuzumab and glofitamab.^3,4,15 In addition, DLBCL with upregulated MYC and downregulated TP53 signatures are less likely to achieve CR and TP53, RHOA, GNAI2, and CD274 mutations were associated with inferior OS.^20,21 Similarly for FL, the mutational landscape appears important, as EZH mutations, which are enriched in high-grade FL, have reportedly inferior BsAb responses.^1,44 

Engagement of inhibitory checkpoints with their respective ligands can lead to anergic T cells and an immunosuppressive tumor microenvironment where BsAb efficacy may be attenuated.^45,46 

Chronic antigenic exposure upregulates coexpression of inhibitory molecules, such as programmed death ligand-1 (PD-L1), leading to impaired cytokine secretion, T-cell proliferation, and reduced cytotoxicity.^45-47 Such features are consistent with a state of T-cell “exhaustion” which contributes to BsAb resistance. Retrospective tumor biopsy analyses identified lower PD-1 expression on T cells and higher PD-L1 on tumor cells correlated with superior response rates.^20,24,25 

Furthermore, after treatment with glofitamab, there is spatial reorganization of cytotoxic CD8⁺ T cells to the surrounding tumor infiltrates.²⁰ With epcoritamab, clonal expansion of cytotoxic CD8⁺ T cells has been associated with CR, whereas the expansion of regulatory CD4⁺ T cells and CD4⁺ T follicular helper cells is associated with reduced cytotoxicity and disease progression.⁴¹ Such findings suggest that a naïve T-cell tumor biome primed with inactivated CD4⁺ T cells, unaffected by T-cell exhaustion and anergy, aid cytotoxic T-cell proliferation and target cell death.⁴⁸ One future strategy would be to assess the pretreatment intratumoral T-cell biome as a means of assessing candidacy for CD3×CD20 BsAbs.

Preclinical epcoritamab studies demonstrated increased tumor cytotoxicity when patient lymph node samples were combined with peripheral blood mononuclear cells (PBMCs), as opposed to the absence of PBMCs.⁴⁰ In the same preclinical model a strong correlation was identified between the total CD3⁺ T cells, in a tumor suspension, and epcoritamab-dependent cytotoxicity (r = 0.63; P < .0001).⁴⁰ Mechanistically, PB CD3⁺ T cells aid response, migrating to the sites of disease. Mosunetuzumab and glofitamab studies have reported disease response associations with higher PB CD19⁺ B cells, but not CD3⁺ T cells.^3,25 Patients with higher total B-cell quantification likely represent a cohort who have not been recently or heavily pretreated. Specific T-cell subset enrichment, as opposed to total levels, are essential for response, as evidenced in glofitamab studies, where CD4⁺ effector memory cells were associated with response, while total quantification of CD3⁺ T cells, CD4⁺ T cells, and CD8⁺ T cells were not.^20,25 In a manner similar to initial CD20 expression, it is clear that PB T cells are required to amplify response, however specific quantities and optimal composition remains unclear.

Not only is an individual’s immune repertoire evolutionary, constantly changing over the patient’s lifetime, but it is influenced by a variety of factors such as autoimmunity, medical comorbidities, environment, and medications. The majority of BsAb-treated patients have received prior chemotherapy, which is known to cause transient absolute peripheral T-cell reductions, later restored by preexisting memory CD8⁺ T-cell expansions.⁴⁹ This has been suggested to be beneficial for CAR-T responses and may explain why amplified cytotoxicity and high response rates are seen with BsAb-chemotherapy combinations.^15,16,30,50 Increasing emphasis is now placed on impaired baseline T-cell function and acquired T-cell exhaustion as BsAb resistance mechanisms.⁵¹ For example, continued BsAb exposure causes sustained T-cell activation with depletion of naïve T cells and expansion of exhausted CD8⁺ T-cell clones.⁵² To potentially mitigate this, a preclinical CD3×CD19 BsAb model incorporated temporary treatment breaks, and as a result, circumnavigated T-cell exhaustion.⁴⁷ Strategies optimizing BsAb delivery must consider these important data when developing new dosing, duration, and therapeutic partners. Future investigation of optimal administration of BsAbs is needed to limit these challenges.

Reduced baseline levels of markers critical to adaptive immune regulation, such as C-reactive protein, interleukin-6 (IL-6), and IL-8, appear associated with higher CR rates in patients treated with glofitamab.^20,25 Notably, cytokine release syndrome and the dose-dependent induction of PB inflammatory cytokines (interferon gamma, IL-6, IL-2, IL8, IL-10, IL-15, IL-17) after treatment with BsAbs did not correlate with response, PFS, or OS.²⁰ 

BsAbs are an attractive T-cell redirecting “off the shelf” therapeutic class for B-NHL, however despite their promising activity, a significant proportion do not achieve durable responses, and relapse remains a major challenge. To gain a deeper understanding of why failures occur, robust translational research must focus on delineating multimodality accurate and accessible biomarkers of response and resistance mechanisms.

Several clinical, immune, and tumor factors appear to be emerging for DLBCL and FL as prognostic, and potentially predictive, in the setting of CD3×CD20 BsAbs such as tumor burden (measured most accurately with PET TMTV), recent anti-CD20 therapy exposure and particular markers of immune health, such as higher PB B cells, lower baseline inflammatory cytokines, and absence of T-cell exhaustion.

BsAbs are now being evaluated in earlier treatment lines, when host immunity is more intact and in combination with additional established and emerging therapies. Furthermore, development of BsAbs with costimulatory domains and trispecific molecules continues. Thus, any biomarker developed must also be tested in these new indications.

Prognostic and predictive biomarkers must be prioritized also due to their potential to address key unanswered questions such as optimal treatment sequencing, particularly in relation to CAR T-cell therapy, as well as appropriate BsAb durations and combinations to maintain or enhance efficacy, yet minimize T-cell exhaustion, toxicity, and financial burden. Addressing these questions through comprehensive embedded translational research will enhance our understanding and application of BsAbs, ultimately improving patient outcomes.

E.A.H. has received funding support from the Australian National Health and Medical Research Council for an Emerging Leader Investigator Grant.

Contribution: All authors designed the concept together, reviewed the literature, wrote the manuscript, and approved the final version.

Conflict-of-interest disclosure: A.B. has received advisory fees or honoraria from Roche, Gilead, BeiGene, and Novartis. E.A.H. reports research funding (paid to institution) from Roche, Bristol Myers Squibb, Merck KgA, AstraZeneca, TG Therapeutics, and Merck; consultancy or advisory role fees (paid to institution) from Roche, Merck Sharp & Dohme, AstraZeneca, Gilead, Antengene, Novartis, Regeneron, Janssen, Specialised Therapeutics, and Sobi; and travel expenses from AstraZeneca. C.S.L. declares no competing financial interests.

Correspondence: Eliza A. Hawkes, Oliva Newton-John Cancer Research Institute, 145 Studley Rd Heidelberg, 3084, Victoria, Australia; email: eliza.hawkes@onjcri.org.au.