NOTCH1 Mutations Are Associated With Therapy-Resistance in Patients With B-Cell Lymphoma Treated With CD20xCD3 Bispecific Antibodies

Abstract

CD20 × CD3 bispecific antibodies such as glofitamab, epcoritamab, and mosunetuzumab are novel T cell engaging antibodies which have all shown convincing results and obtained FDA and EMA approval for the treatment of relapsed/refractory diffuse large B cell lymphomas (DLBCL) or follicular lymphoma (FL) with ≥ 2 prior lines of treatment [1]. However, approximately 50% of patients do not achieve remission when treated with single agent CD20 × CD3 bispecific antibodies.

Subgroup analysis of pivotal phase I/II trials have identified elevated LDH > 250 U/L, high tumor burden, and refractory disease as risk factors for lack of response to bispecific antibodies for refractory/relapsed DLBCL [1]. Downregulated TP53 target signatures, upregulated expression of MYC target genes, truncating mutations in MS4A1 (the gene encoding CD20), and loss of CD20 antigen have been identified as predictive factors for lack of response [2-5]. Previously, tumor mutations in TP53 have been associated with poor response to both immunochemotherapy (i.e. R-CHOP) and CD19 CAR-T cell therapy in patients with B-cell lymphoma, but have not yet been examined thoroughly with long-term follow-up in patients treated with CD20 × CD3 bispecific antibodies.

In this retrospective study, we included patients from Rigshospitalet, Copenhagen, and Vejle University Hospital, both in Denmark, who received CD20 × CD3 bispecific antibodies between 2017 and 2023 as part of phase 1 or phase 2 clinical trials. The full list of regimens used are shown in Table S1. For exhaustive methods used, refer to the Supporting Information S1.

We collected pre-treatment formalin-fixed and paraffin-embedded (FFPE) archival specimens from 106 patients, of which 56 had sufficient DNA quantity and tumor involvement for clinical grade diagnostic next-generation sequencing (NGS). NGS was performed with a custom lymphoma panel designed in-house covering 59 commonly mutated genes in lymphoid malignancies.

We examined pre-treatment tumors from 56 patients with B-cell NHL, who received CD20 × CD3 bispecific antibodies between 2017 and 2023 and had sufficient tissue for NGS sequencing. The median age at first administration of CD20 × CD3 bispecific antibody was 70 years, 60.7% were male, and the median number of prior lines of therapy was three (Table S1). Median follow-up time was 24.2 months. Mutational profiling was additionally performed on 14 paired post-CD20 × CD3 bispecific antibody relapse samples.

Of all mutations found in pre-CD20 × CD3-treatment biopsies NOTCH1 (detected in 4/56 [7%] of patients, Figure S1), along with BRAF and EZH2, had the strongest association with inferior PFS in a univariate Cox model (HR: 3.46, 95% CI 1.16–10.3, p = 0.026, Tables S2 and S4, Figure S8). Furthermore, pre-CD20 × CD3-treatment NOTCH1 mutated tumors conferred significantly worse outcomes in Kaplan–Meier analysis of both PFS and OS (Figure 1A, log rank p = 0.02 and p = 0.05, respectively). NOTCH1 mutations found in pre-CD20 × CD3-treatment biopsies were primarily located in exon 34, while one was in exon 26 (c.4975G > A, Figure S7).

In patients with DLBCL we found no statistically significant difference in ORR between TP53 mutation and wildtype TP53 (p = 0.24, Figure S5A). Among patients with TP53 mutations, 6/17 (36%) achieved complete remission, while 20/39 (50%) of patients with wildtype TP53 achieved complete remission. This difference was not statistically significant (Fischer's exact test, p = 0.11). Moreover, PFS was similar in patients with TP53 mutations/deletions and wildtype TP53 (HR: 1.07 95% CI 0.47–2.46, p = 0.9 Figure S5B), even after adjustment for age and performance status (HR: 1.10 95% CI 0.45–2.68, p = 0.83). There was no association between survival in molecular subtype clusters according to Lacy et al. (Figure S3) [6]. Number of mutations were not associated with worse ORR (Figure S6).

In FL, there was a statistically significant association between worse survival and TP53 mutations; however, sample sizes were extremely small (Figure S4).

To investigate clonal development, we analyzed 17 post-CD20 × CD3 relapse samples, 14 of which were available for paired comparison with pre-CD20 × CD3 bispecific antibody treatment samples. The most common mutations at relapse after CD20 × CD3 exposure were CREBBP (41%), STAT6 (24%), BCL2 (24%), NOTCH1 (24%), KMT2D (18%), and BCL2 (18%) (Figure S2 and Table S3).

When analyzing the evolution of tumor mutations during therapy with CD20 × CD3 bispecific antibodies (Figure 1C), we observed significant outgrowth of clones with NOTCH1 mutations (Patient 53, 31, 52, and 5). This consisted of four new NOTCH1 clones (not detectable in pre-treatment samples). Of the four new NOTCH1 clones identified at relapse (not detectable in pre-treatment samples), three were in exon 34 (Figure S7). All exon 34 clones were detected in germinal center B cell-like DLBCL samples. Another patient was found to have a mutation in exon 13 and had follicular lymphoma. Two mutations were deletions and two were missense mutations. None were in the common c.7541_7542delCT hotspot (Figures S7, 1C and Table S3). The four patients who develop new tumor NOTCH1 mutations after treatment all had strong IHC CD20 expression before treatment. In the post-therapy relapse biopsy, only one patient had strong CD20 expression, while two had no expression of CD20 and one has a weak CD20 expression (Figure 1D). Four patients had NOTCH1 mutations in the pre-CD20 × CD3-treatment samples; they all died from lymphoma progression within 5 months after initiation of therapy (Figure 1A,B) without available relapse biopsy.

We identified three patients where TP53 clones had declined in size at relapse (patient 31, 61, and 86). However, one patient also had marked growth of TP53 clones (patient 105) during T-cell engaging therapy. Similarly, three patients had declining CREBBP clones at relapse (113, 39, 71).

To our knowledge, this is to date the largest analysis of the prognostic impact of lymphoma mutations and clonal evolution during therapy with CD20 × CD3 bispecific antibodies. We find that especially NOTCH1 mutated clones expand under the pressure of CD20 × CD3 bispecific antibody therapy and may constitute an escape mechanism potentially leading to downstream CD20 antigen loss. This has as previously been described in chronic lymphocytic leukemia (CLL), where NOTCH1 mutations lead to epigenetic dysregulation and following transcriptional repression of CD20 [7]. Furthermore, the four patients who had NOTCH1 mutations prior to CD20 × CD3 bispecific antibodies, all died from progression within 5 months after beginning treatment. Growth of clones with NOTCH1 activating mutations and increased NOTCH1 activity has been associated with relapse in the context of anaplastic large cell lymphoma, T-cell acute lymphoblastic leukemia and CLL. Additionally, in ∼50% of CLL cases without NOTCH1 mutations, the active intracellular portion of NOTCH1 (ICN1) is detectable [8]. This could potentially be mirrored in lymphoma. If this is true, we might underestimate the impact of NOTCH1 activation as a resistance mechanism against CD20 × CD3 bispecific antibodies. Further studies on analyzing impact of ICN1 expression in lymphomas treated with CD20 × CD3 bispecific antibodies is warranted.

We hypothesize that the NOTCH1 signaling pathway may be a future target to avoid relapse following anti-CD20 T-cell engaging therapy. In desmoid tumors, which are characterized by NOTCH1 hyperactivation, γ-secretase inhibitors have shown promising results. γ-secretase inhibitors should thus be considered in clinical trials combined with CD20 × CD3 bispecific antibodies, to circumvent this resistance pathway.

We found low numbers of EZH2 and MYC mutations in our cohort. Our in-house targeted sequencing panel has previously been validated in 298 patients [9]. In this cohort, we found EZH2 mutations in 16% of FL patient and 11% of DLBCL patient samples, which is comparable to frequencies reported in the literature. The low numbers of EZH2 and MYC mutations in this cohort may be a sampling bias due to the limited number of patients.

Surprisingly, we found that clones harboring pathogenic TP53 mutations diminished in size during therapy. A decrease or a stable level of TP53 mutated clones has also been seen in CLL patients treated with ibrutinib [8]. However, interestingly CD19 CAR T cells have been associated with an expansion of TP53 mutated clones. This difference could be explained by the fact that many CAR-T cell recipients receive bridging chemotherapy and that nearly all CAR-T cell protocols include fludarabine/cyclophosphamide or other lymphocyte depleting treatments. Pre-treatment TP53 mutations were not associated with resistance to CD20 × CD3 bispecific antibodies. Outcomes were poor in both groups, which reflects that the study primarily included biopsies from high-risk phase 1 study participants. A prior study on resistance to glofitamab by Bröske et al. in 2022 did not identify a correlation between TP53 mutations and resistance; however, they did find a correlation between downregulation of TP53 target signatures and resistance [3]. In the context of DLBCL treated with anti-CD19 CAR-T cell therapy, TP53 mutations were found to be associated with refractory disease [10]. Multiple studies have found the same association, when investigating patients with B cell lymphomas treated with immunochemotherapy. Our findings may indicate that TP53 mutations are not a specific marker of therapy resistance in patients with B cell lymphomas treated with CD20 × CD3 bispecific antibodies. However, confirmation of this finding in other larger cohorts is warranted to draw firm conclusions. If validated, this could imply that CD20 × CD3 bispecific antibodies should be a preferred treatment option in patients with relapsed/refractory TP53 mutated B cell lymphomas.

To conduct this study, we used a customized targeted panel sequencing panel of 59 genes. We were thus only able to detect mutations in genes known to be abrogated in lymphoma, and deduction of mutational signatures is very challenging.

In summary, we find that pre-treatment NOTCH1 mutations are associated with poor survival and that new tumor clones with genetic aberrations of NOTCH1 markedly expand during CD20 × CD3 bispecific antibody therapy. Additionally, we find pre-treatment TP53 mutations surprisingly do not infer poorer overall response or survival in this cohort. We also find that clones with TP53 mutations in pre-CD20 × CD3-treatment samples either disappear or diminish in size at relapse in most patients. These data indicate that TP53 mutations may not be an adverse marker for response to CD20 × CD3 antibodies. Further studies are warranted to validate the impact of tumor intrinsic factors in patients treated with CD20 × CD3 bispecific antibodies. Additionally, it is warranted to examine, whether combinations of CD20 × CD3 bispecific antibodies with NOTCH1 targeted therapy or other novel agents may yield improved outcomes.

Emil Kyvsgaard wrote the paper. Simon Husby, Martin Hutchings, Kirsten Grønbæk, and Emil Kyvsgaard designed the research. Lene Sjø, Trine Lønbo Grantzau, Linea Cecilie Melchior, Lise Mette Rahbek Gjerdrum, and Marie Breinholt collected samples. Morten Grauslund, Linea Cecilie Melchior, and Emil Kyvsgaard performed sequencing and variant filtration. Emil Kyvsgaard and Simon Husby performed the statistical analysis. Caroline Riley, Michael Roost Clausen, Martin Hutchings, Carsten U. Niemann, Trine Trab, Lærke Sloth Andersen, and Emil Kyvsgaard collected data. All authors contributed to the interpretation of the data and to the writing and final approval of the manuscript.

We obeyed the laws on handling of personal information in accordance with the Danish Scientific Ethical Committees Act (Komitéloven) § 20, stk 1. nr. 4. The study was approved by Datatilsynet under the Capital Region umbrella application (RH-2020-561). Concerning documentation of data security, the demands of Datatilsynet have been met.

Consent was obtained where applicable. Exemption from obtainment of informed consent from these patients was granted by the Danish ethical council through the Research Biobank and Clinical Database of patients with Lymphoproliferative Malignancies (LM).

K.G. received research support from Janssen and is on the advisory board of Nanexa and GSK. M.H. consulting or advisory role at AbbVie, AstraZeneca, Celgene, Genmab, Janssen, Merck, Roche, and Takeda and received research funding from AbbVie, AstraZeneca, Bristol Myers-Squibb, Celgene, Genentech, Genmab, Incyte, Janssen, Merck, Novartis, Roche, Takeda. C.U.N. received research funding and/or consultancy fees from AstraZeneca, Janssen, AbbVie, Beigene, Genmab, CSL Behring, Octapharma, Takeda, and Novo Nordisk. T.S.L.: consultancy fees: Roche, Gilead. Research funding: Genentech. L.M.R.G. consultancy fees from Kyowa Kirin. T.T. received research support from Janssen. M.G. consultancy fees: Amgen, Astra Zeneca, Thermo Fisher Scientific and research support from Merck.