Predictors of Delayed Responses to Teclistamab in Multiple Myeloma After Initial Non‐Response

In trials of bispecific antibodies (bsAbs) targeting B-cell maturation antigen (BCMA) in multiple myeloma (MM) such as teclistamab, the median time to initial response is almost always 30–40 days 1-3. This Day 30 (D30) assessment corresponds to the first response assessment at the beginning of the second cycle and thus constitutes an important juncture for decision-making. Cumulative risks of continued BCMA-targeted bsAb therapy include T-cell exhaustion, high-grade infections, time toxicity, and increased costs for the healthcare system both for bsAb administration and for supportive care (e.g., immunoglobulin replacement) 4-7. As such, non-responding patients with low likelihood of subsequent clinical benefit should ideally be transitioned to an alternative therapy instead. However, there are no data to predict the likelihood of delayed response conversion following initial non-response. We thus performed a secondary analysis of published real-world (RW) data from the United States MM Immunotherapy Consortium 8. Detailed methods, including Institutional Review Board approval and informed consent, are described in Table S1. Non-response to teclistamab at D30 and/or Day 90 (D90) was defined as failure to achieve at least a partial response (PR) by International Myeloma Working Group (IMWG) criteria: in other words, minimal response (MR), stable disease (SD), or progressive disease (PD). Among living patients with non-response (NR) at D30, patients with a subsequent delayed response (D30 NR followed by D90 response) were compared against continued non-responders (D30 NR followed either by continued D90 NR or death) using logistic regression. Patients who died before D30 or stopped teclistamab before D90 (in the absence of death) were excluded. Covariates of interest included patient characteristics, markers of disease burden including visceral extramedullary disease (EMD), cytokine release syndrome (CRS) characteristics, and longitudinal absolute lymphocyte count (ALC) values based on prior data suggesting their utility to predict overall response rates with bsAb therapy 9, 10. Delayed responders were separately compared against early responders (i.e., patients with a response at/before D30) using Kaplan–Meier analyses landmarked at D90 to account for the time-dependent nature of delayed responses. Of 351 analyzable patients (details in Figure S1 and Table S2), 55% (n = 193) had not achieved ≥ PR by D30: this comprised 23 patients (12%) with D30 MR, 66 (34%) with D30 SD, and 104 (54%) with D30 PD. Of these 193 initial non-responders, 23% (n = 45) later achieved a delayed response with continued teclistamab. The remaining patients either remained alive with continued NR (54% of all D30 non-responders, n = 104) or died by D90 (23%, n = 44, including 5 deaths possibly or definitely related to teclistamab). As shown in Table 1, the delayed response and NR/death groups were comparable in terms of age, prior BCMA exposure, light-chain only disease, EMD, baseline absolute neutrophil count, and C-reactive protein levels. Baseline ferritin levels were higher in the NR/death group but were missing for one third of patients overall. While CRS grades were comparable between groups, CRS occurred earlier in delayed responders (Figure S2). ALC values (Figure S3) were significantly higher among delayed responders at baseline and D30 but not at Day 7. Finally, exact D30 response distributions were different: 44% MR/53% SD/1% PD among D30 non-responders who later achieved a response versus 3% MR/28% SD/70% PD among D30 non-responders who did not. Logistic regression (Figure S4) identified two independent predictors of delayed responses among D30 non-responders: (1) achievement of a D30 MR, as opposed to D30 SD or D30 PD adjusted odds ratio (aOR) 24.3, 95% confidence interval (CI) 3.64–163; and (2) D30 ALC (continuous variable: aOR 2.96, 95% CI 1.06–8.29). Based on these two covariates and re-review of response depth stratified by ALC values (Figure S3), D30 non-responders were classified into four groups (Figure 1). Group 1 comprised patients with D30 MR (n = 231), 87% of whom achieved a delayed response regardless of D30 ALC. Among patients with D30 SD (n = 66), Groups 2 and 3 were differentiated based on D30 lymphopenia (ALC < 0.5 thousand cells per microliter, as defined previously) 9, 10. Of 26 patients with D30 SD, delayed responses occurred in 54% of patients without D30 lymphopenia (Group 2) versus in 25% of patients with D30 lymphopenia (Group 3). Finally, in Group 4 with D30 PD (n = 104), there was only 1 delayed response. Kaplan–Meier curves (Figure 2) for delayed responders (D30 NR followed by D90 response) versus early responders (response at/before D30) demonstrated comparable landmarked PFS: median PFS 16.8 months (95% CI 13.3–17.6 months) for delayed responders versus 15.2 months (95% CI 13.7–16.1 months) for early responders, with hazard ratio 1.06 (95% CI 0.61–1.85). In summary, we conducted the first dedicated analysis of bsAb recipients without a response at D30, which is generally the first time that MM responses are assessed. Continuing bsAb therapy beyond this point can come with risks, most notably the continued risk of high-grade infections or the potential for T-cell exhaustion that can limit the success of subsequent therapies 4, 5. However, nearly a quarter of D30 non-responders in our analysis subsequently achieved a delayed response with continued teclistamab therapy. More remarkably, landmarked PFS was comparable for these delayed responders versus patients who had already achieved a response by D30. Continuation of bsAb therapy is thus clearly warranted in a subset of patients despite initial non-response, and our analysis identifies a novel two-step framework to help identify such patients. For patients who have achieved an MR at D30, the likelihood of a subsequent response is quite high; conversely, for patients with PD by this timepoint, subsequent response conversion is rare. In patients with SD at D30, the presence of concomitant lymphopenia suggests a much lower likelihood of response conversion. Given the dichotomization of percentages inherent to IMWG response criteria—for example, a 49% reduction in serum paraprotein levels would be MR while a 51% reduction would be PR—it is not surprising that many patients who initially achieve an MR deepen their response with continued time on therapy. In patients with high baseline paraprotein levels, it may be helpful to set the expectation that teclistamab should be continued despite the initial achievement of only SD or MR at D30. Our finding that landmarked PFS is equivalent regardless of time to initial response is a helpful component of such discussions. With regard to D30 ALC being a significant predictor of delayed responses, our findings align with similar studies demonstrating the negative impacts of lymphopenia following BCMA-targeted CAR-T therapy or bsAb therapy 9-12. In fact, delayed responders in our analysis had a median D30 ALC more than twice that of patients with continued NR or death. In the absence of detailed T-cell analyses that are not routinely available in practice, our analysis suggests that Day 30 ALC can serve as a surrogate for T-cell fitness to guide decisions around bsAb continuation in this scenario. Given the very high a priori likelihood of delayed response conversion in patients with D30 MR versus the very low corresponding likelihood in patients with D30 PD, D30 ALC values may be most helpful as a predictive tool for the approximately half of non-responders who achieve D30 SD (as visualized in Figure S3). Our retrospective study has several limitations. The observed D30 and D90 ORRs of < 60%, consistent with our primary analysis 8, reflect the RW nature of this cohort including many BCMA-exposed patients. D30 and D90 response assessments were missing for many patients, although baseline characteristics for analyzable and non-analyzable patients were largely comparable (Table S2). Certain variables with potential impacts on bsAb efficacy—for example, plasma cell burden and inflammatory biomarkers—were missing for over a third of patients and thus were excluded from our regression analyses. Additional limitations arise from the complexities of response assessments, which were determined by each center and—consistent with RW practice in MM—did not require validation using every formal element of IMWG criteria (e.g., 24-h urine testing or confirmatory measurements) given their limited relevance to clinical practice 13-15. Sample sizes for visceral EMD (n = 23) and non-secretory MM (n = 3) were too small to draw meaningful conclusions, particularly given that bone marrow biopsies and imaging may not always be done at D30 in these patients. Despite these limitations, our study provides valuable insights to inform decision-making for MM patients who have not achieved at least a PR by the end of the first cycle. In the future, our risk stratification framework may guide treatment modifications at D30 for those with insufficient responses: for example, adding agents to potentiate T-cell fitness (e.g., pomalidomide) or transitioning to a different MM therapy. Our findings also highlight the importance of longitudinal ALC measurements rather than pre-treatment levels alone in terms of predicting responses to T-cell redirection in MM. In the future, we hope to incorporate more detailed T-cell subset analyses and functional assays to better identify patients for whom continued bsAb therapy may be beneficial despite initial non-response. R.B., G.K., and B.M.R. analyzed the data and wrote the first draft of the manuscript. All authors contributed patients to this analysis, provided critical feedback, and approved the final manuscript. The authors wish to acknowledge the patients who participated in both studies as well as the research personnel at all study sites. Figure S1B wsas made using sankeymatic.com. R.B. reports consulting: Abbvie, Adaptive Biotech, BMS, Caribou Biosciences, Genentech, Gilead/Kite, GSK, Janssen, Karyopharm, Legend Biotech, Poseida Therapeutics, Pfizer, Sanofi, SparkCures; Research: Abbvie, BMS, Janssen, Novartis, Pack Health, Prothena, Sanofi. G.K. reports consulting: BMS, Arcellx, Sanofi, Janssen, Cellectar, Pfizer, Kedrion; Research: BMS, Janssen, Abbvie. S.S. reports research: Magenta Therapeutics, BMS, Allogene, Janssen, Novartis, Abbvie; Advisory Board/Consultancy: BMS, Janssen, Sanofi, Oncopeptides, Takeda, Regeneron, Abbvie, Pfizer, BiolineRx, Legend, Kite. T.R. reports consulting: BMS, J steering committees: Gracell Therapeutics, BMS; research support: Janssen, BMS, C4 Therapeutics, Gracell Therapeutics, Heidelberg Pharma; consulting: Genentech, Janssen, BMS, Karyopharm Therapeutics. L.S. reports consulting: BMS. J.K. reports consulting: GPCR, Janssen, Prothena, Legend Biotech; research: Prothena, Ascentage, Janssen, Karyopharm, GPCR. H.C.L. reports consulting: Bristol Myers Squibb, Pfizer, Janssen, Regeneron, GlaxoSmithKline, Sanofi, Abbvie, Takeda Pharmaceuticals, Allogene Therapeutics, Menarini, Alexion Pharmaceuticals; research funding: Amgen, Bristol Myers Squibb, Janssen, GSK, Regeneron, Takeda Pharmaceuticals. K.K.P. reports consulting: BMS, Janssen, AstraZeneca, Legend Biotech, Kite, Genentech, Abbvie, Sanofi, Caribou, Takeda, Regeneron, Poseida. D.K.H. reports consulting: BMS, Janssen, legend Biotech, Pfizer, Karyopharm; Research: BMS, Karyopharm, Adaptive Biotechnologies, and Pentecost Myeloma Research Center. A.A. reports: consulting: BMS, Karyopharm, Sanofi; research: Abbvie, Adaptive, K36 Therapeutics. J.A.D. reports consulting: Janssen; speakers' bureau: Janssen. H.H. reports consulting: Janssen; speaker bureaus: Janssen, Karyopharm. C.J.F. reports consulting: Janssen; Research: Janssen, Regeneron; ownership of publicly traded stock: Affimed. S.M. reports consulting: Janssen, Pfizer; honoraria: Pfizer; stock ownership: Abbvie. P.F. reports consulting: Janssen, Sanofi; research funding: Karyopharm. Y.L. reports consulting: Janssen, Legend, Celgene, Sanofi, BMS, Pfizer, Regeneron, Genentech, NexImmune, Caribou; research funding: Janssen, Celgene, BMS. A.J.C. reports consulting: Abbvie, Adaptive, BMS, HopeAI, Janssen, Sebia, Sanofi; Research: Abbvie, Adaptive Biotechnologies, Caelum, Harpoon, Nektar, BMS, Janssen, Sanofi, OpnaBio, IgM Biosciences, Regeneron. L.D.A.Jr. reports consulting: Janssen, Celgene, BMS, Amgen, GSK, AbbVie, Beigene, Cellectar, Sanofi, Karyopharm, Pfizer, Prothena. Research: BMS, Celgene, GSK, Janssen, Abbvie. A.L.G. reports research funding: Johnson consulting: Johnson DSMB membership for Johnson & Johnson. The remaining authors declare no conflicts of interest. Data sharing statement: Available from the corresponding author upon reasonable request. Data S1: Supporting Information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.