Intratumoral Sotigalimab with Pembrolizumab Induces Rapid Activation of Antigen-Presenting Cells and Drives Antitumor Responses in Non-injected Tumors in Metastatic Melanoma: A Phase I/II Study

Intratumoral sotigalimab plus pembrolizumab engages the CD40 pathway, activates antigen-presenting cells, and remodels the tumor immune microenvironment, supporting in situ immunization and antitumor responses in metastatic melanoma.

Abstract

Immune checkpoint blockers (ICB) improve outcomes in metastatic melanoma (MM), but resistance limits benefit. This phase I/II (NCT02706353) study evaluated intratumoral sotigalimab (anti-CD40 agonist) with pembrolizumab in 32 patients with ICB-naïve MM. Primary endpoints were safety and objective response rate (ORR). Sotigalimab was well tolerated. At the recommended phase II dose, the ORR was 50%, and the disease control rate was 92%, with ORRs of 67% in injected and 50% in non-injected tumors. Multiomic analyses of tumor and blood showed that sotigalimab effectively engaged the CD40 pathway, boosting infiltration and activation of myeloid cells, including CD11c⁺DC-LAMP⁺ dendritic cells and macrophages. The combination therapy activated innate and adaptive immunity in injected tumors and cytotoxic responses in non-injected tumors. T-cell receptor sequencing showed increased T-cell clonality with expanded new clones shared across tumors. Clinical responses correlated with these immunologic changes but not with baseline features associated with response to anti–PD-1 monotherapy.

Significance:

In this study, we provide compelling data that intratumoral sotigalimab combined with pembrolizumab is safe, activates antigen-presenting cells, and elicits broad innate and adaptive immune responses in both injected and non-injected tumors, supporting further randomized phase II trials to evaluate sotigalimab’s potential to enhance anti–PD-1 therapy through “in situ” immunization.

Introduction

Using immune checkpoint blockers (ICB) to stimulate the immune system achieves significant clinical benefit in many patients with metastatic melanoma (MM). However, many patients with MM do not respond to these treatments or progress due to resistance, necessitating new treatment options to further improve clinical outcomes. In addition, the systemic delivery of ICBs can lead to immune-related side effects that can be severe and limit their use. Growing evidence suggests that efficient antigen presentation and T-cell priming are essential for generating an effective antitumor immune response (1–5). Using intratumoral therapy to manipulate antigen-presenting cells (APC) within the tumor may activate local dendritic cells (DC) to process and present tumor antigens to T cells, prime antitumor T cells, and convert immunologically cold tumors into immunologically hot tumors.

CD40, a costimulatory receptor in the tumor necrosis factor (TNF) superfamily, is prominently expressed on a variety of immune cells including APCs such as DCs, macrophages, and B cells (6–8). In preclinical cancer models, agonistic CD40 monoclonal antibodies (mAb) drive the activation of tumor-specific T-cell responses through the activation of APCs (9–11). In addition, CD40 mAbs induce T helper (Th) 1 cytokines such as IL12 and reeducate tumor-associated macrophages toward an M1-like phenotype with the capacity to destroy tumor stroma in a T cell–independent fashion (12–15). The antitumor effects of agonistic CD40 mAbs have been more pronounced when combined with ICBs (10, 16–20). Sotigalimab is a potent human IgG1 CD40 agonistic mAb that binds with high affinity to the CD40 ligand–binding site on CD40. It is designed to stimulate both innate and adaptive immune responses, including the activation of DCs, B cells, and macrophages (21). Our previous preclinical studies demonstrated that intratumoral CD40 activation induces systemic antitumor effects and augments the activity of anti–PD-1 therapy (10). On the basis of these findings, we hypothesized that intratumoral injection of sotigalimab in combination with the systemic delivery of pembrolizumab will activate local APCs to process and present antigens to T cells at the tumor site, thereby modulating the tumor microenvironment and turning immunologically cold tumors into hot tumors. This, in turn, promotes activation and circulation of tumor-specific CD8⁺ T cells, resulting in responses in non-injected tumors. Here, we report the results of a phase I/II study of intratumoral sotigalimab in combination with systemic pembrolizumab in 32 patients with ICB-naïve MM. The clinical study was accompanied by longitudinal multiomic analyses including T-cell receptor (TCR) sequencing, multiplex immunofluorescence (mIF), imaging mass cytometry (IMC), single-cell multiome interrogation by simultaneously profiling gene expression [single-cell RNA sequencing (scRNA-seq)] and open chromatin [single-cell Assay for Transposase-Accessible Chromatin using sequencing (scATAC-seq)], and NanoString nCounter Human PanCancer Immune Profiling of peripheral blood samples and injected and non-injected tumors collected before and during treatment (Fig. 1A).

Figure 1.

Study design, treatment plan, and clinical responses to sotigalimab and pembrolizumab in all treated patients (n = 32). A, All patients provided written informed consent before any protocol-specified procedures. Serial blood samples, as well as pretreatment and on-treatment tumor biopsies from both injected and non-injected tumors, were collected to perform in-depth analysis of immune cells using TCR sequencing, mIF, IMC, single-cell multiome interrogation by simultaneously profiling gene expression (scRNA-seq) and open chromatin (scATAC-seq), and NanoString nCounter Human PanCancer Immune Profiling. Tumor biopsies (injected and non-injected) were collected at screening and week 6 after combination therapy, with additional biopsies taken within 24 hours of sotigalimab injection. Blood samples were collected at baseline and on days 3, 8, and 21 after therapy. All patients received four doses of sotigalimab every 3 weeks, in combination with pembrolizumab at 2 mg/kg. t-SNE, t-distributed stochastic neighbor embedding. B, Waterfall plot and pie chart showing the ORR assessed as the sum of the diameters of tumors compared with baseline according to RECIST 1.1. in all treated patients (n = 32). * Lymph node, CR: pathologic lymph nodes having a reduced short axis of <10 mm. C, Spider plot demonstrating response kinetics (durable and transient) in all treated patients. D, Waterfall plot showing the best percentage change in the size of injected tumors (blue) and non-injected tumors (orange) from baseline. E, Pie charts showing the ORRs for both injected and non-injected tumors in all treated patients (n = 32). F, Pie charts showing the ORRs for both injected and non-injected tumors in patients treated at the RP2D (n = 24).

Results

Patient Characteristics and Treatment

Thirty-two ICB-naïve patients were enrolled in the phase I/II trial from July 2017 to June 2022. In the phase I portion, 14 patients received intratumoral sotigalimab, starting at an initial dose of 0.1 mg, with doses escalating from 0.5 to 10 mg. This was combined with intravenous pembrolizumab (2 mg/kg), administered every 3 weeks. Sotigalimab was given every 3 weeks for up to 4 doses, whereas pembrolizumab continued until disease progression or unacceptable toxicity occurred. The maximum tolerated dose (MTD) was not reached, and the recommended phase II dose (RP2D) for sotigalimab was 10 mg based on safety, clinical activity, and immune effects. In phase II, 18 patients received intratumoral sotigalimab at the RP2D with pembrolizumab. Patient demographics and baseline disease characteristics are shown in Supplementary Table S1. The median age was 64.5 years (range, 32–81 years). Most patients (27/32, 84%) were male. All patients had an Eastern Cooperative Oncology Group (ECOG) performance status of either 0 (56%) or 1 (44%). Eleven (34%) of the 30 patients assessed for the mutation had BRAF mutation–positive disease. Seventeen (53%) patients had an elevated serum lactate dehydrogenase (LDH) level; of these patients, 7 (41%) had baseline LDH levels between the upper limit of normal (ULN) and twice the ULN and 10 (59%) had baseline LDH levels of more than twice the ULN. Seventeen (53%) patients had stage IV M1a or M1b disease, 7 (22%) had stage IV M1c disease, and 8 (25%) had stage III disease. Programmed cell death ligand 1 (PD-L1) status, defined as ≥1% tumor cell positivity for membranous staining (clone 28–8) by immunohistochemistry (IHC), was positive in 8 (25%) patients, negative in 11 (34%) patients, and unknown in 13 (41%) patients. Liver metastases were present in 12 of the 32 patients assessed (38%). Patient demographics and baseline disease characteristics at the RP2D are summarized in Supplementary Table S2.

Safety and Tolerability

The safety results for all 32 patients are summarized in Supplementary Table S3. Eight (25%) patients had grade 3/4 treatment-related adverse events (AE), including colitis (n = 2, 6%), injection site reaction (n = 2, 6%), elevated alanine aminotransferase (ALT, n = 2, 6%), elevated aspartate aminotransferase (AST, n = 1, 3%), pruritus (n = 1, 3%), maculopapular rash (n = 1, 3%), and decreased platelet count (n = 1, 3%). Five (16%) patients had grade 3/4 immune-mediated AEs, including colitis (n = 2, 6%), pruritus (n = 1, 3%), maculopapular rash (n = 1, 3%), and elevated ALT (n = 1, 3%). No patients discontinued the study or died owing to treatment-related AEs. The combination therapy did not induce dose-limiting toxicity (DLT) at any dose level of sotigalimab. Serial lymphocyte counts exhibited a transient posttreatment decline with recovery by approximately week 4, whereas monocyte and platelet counts remained generally stable throughout treatment (Supplementary Fig. S1A–S1C).

Antitumor Activity

For all treated patients, the objective response rate (ORR) by Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 criteria was 47% [15/32; 95% confidence interval (CI), 30.9%–63.6%], which consisted of a complete response (CR) rate of 16% (5/32) and a partial response (PR) rate of 31% (10/32); the stable disease (SD) rate was 44% (14/32), and the progressive disease (PD) rate was 9% (3/32; Fig. 1B). The disease control rate (DCR; CR + PR + SD) was 91% (29/32; Fig. 1B). At the RP2D of sotigalimab (10 mg), the ORR was 50% (12/24; 95% CI, 31.4%–68.6%), and the DCR was 92% (22/24; Supplementary Fig. S2A). Responses occurred early, often after the first treatment, then deepened over time, and remained durable (Fig. 1C). For all patients, the median overall survival (OS) was 38 months, with 6-, 12-, 24-, and 36-month OS rates of 94%, 92%, 75%, and 55%, respectively (Supplementary Fig. S2B). The median progression-free survival (PFS) duration was 16 months, with 6-, 12-, 24-, and 36-month PFS rates of 62.5%, 55%, 31%, and 28%, respectively (Supplementary Fig. S2C). The 6-month DCR was 62.5% (20/32), comprising 5 CRs, 9 PRs, and 6 cases with SD (Supplementary Fig. S2C). Both sotigalimab-injected and non-injected tumors showed clinical responses (Fig. 1D; Supplementary Fig. S2D). The ORRs for non-injected tumors and injected tumors in all patients were 47% (15/32; 95% CI, 30.9%–63.6%) and 60% (19/32; 95% CI, 42.3%–74.5%), respectively (Fig. 1E). At the RP2D, the ORRs were 50% (12/24; 95% CI, 31.4%–68.6%) for non-injected tumors and 67% (16/24; 95% CI, 46.8%–82.1%) for injected tumors (Fig. 1F). The clinical responses of the injected and non-injected tumors were largely concordant (Supplementary Fig. S2E and S2F). Among injected lesions, the ORR and DCR were comparable across lesion types (Supplementary Table S4). Injected lesions included superficial or subcutaneous/soft-tissue lesions (14/32, 44%), lymph nodes (14/32, 44%), and deep-seated lesions such as bone lesions or pulmonary nodules (4/32, 12%; Supplementary Table S4). The list of injected tumors and corresponding patient IDs is provided in Supplementary Table S5.

The subgroup analysis shows that clinical responses occurred regardless of baseline PD-L1 status, LDH level, or BRAF mutation status (Supplementary Fig. S3A–S3C). Specifically, 62.5% (5/8) of patients with positive baseline PD-L1 and 54.6% (6/11) with negative PD-L1 responded to therapy (Supplementary Fig. S4A). Clinical responses were observed in 57% (4/7) of patients with baseline LDH levels elevated between the ULN and 2 × ULN and in 60% (6/10) of patients with LDH levels exceeding 2 × ULN (Supplementary Fig. S4B). In addition, clinical responses were observed in 60% (6/10) of patients with a BRAF mutation (Supplementary Fig. S4C).

Further details on baseline LDH, tumor burden, and clinical responses for all patients are shown in Supplementary Fig. S4D and S4E.

Sotigalimab Effectively Engages the CD40 Pathway and Induces the Rapid Infiltration and Activation of APCs

To determine the impact of the intratumoral sotigalimab on the local immune response, we used the NanoString nCounter Human PanCancer Immune Profiling Panel to analyze gene expression in 23 matched tumor biopsy samples obtained before and 24 hours after sotigalimab administration. Sotigalimab induced nuclear factor (NF)-κB and mitogen-activated protein kinase (MAPK) pathways and significantly increased the expression of a variety of APC-related genes (including EBI3, CLEC5A, LAMP3, PSMB8, PSMB9, and PSMB10) as well as that of inflammatory mediators such as IL6, IL8, and IL1B (Fig. 2A and B). Gene set enrichment analysis revealed that, compared with baseline samples, samples obtained 24 hours after sotigalimab administration had significant enrichment in numerous gene sets related to immune activation and cell signaling pathways (including TNFα signaling via NF-κB, IL6/JAK/STAT3 signaling, inflammatory response, IFNα response, and IFNγ response), consistent with CD40 activation (Fig. 2C). Immune deconvolution suggested an upregulation of CD40 signaling and an enrichment in myeloid cells, including DCs (CCL13, CD209, HSD11B1, CCL17, CCL22, NPR1, and PPFIBP2), activated DCs (CCL1, EBI3, INDO, LAMP3, and OAS3), and macrophages (CD163, CD68, CD84, and MS4A4A), 24 hours after the start of sotigalimab (Fig. 2D).

Figure 2.

Sotigalimab effectively engages the CD40 pathway and upregulates APC-associated genes in injected tumors. A, Volcano plot of gene expression determined from an analysis of RNA extracted from sotigalimab-injected tumors 24 hours after sotigalimab compared with baseline (BL; n = 23 paired samples). Each gene’s −log₁₀ (P value) and log₂ fold change (log₂ FC) with the selected covariate for each gene. Highly statistically significant genes fall at the top of the plot, above the horizontal lines, and highly differentially expressed (DE) genes fall to either side of the plot. The horizontal lines indicate various false discovery rate (FDR) thresholds or, for unadjusted P values, P value thresholds. The colored dots represent the genes with P values below the given FDR or P value threshold. The 25 most statistically significant genes are labeled. B, Heatmap showing differences in the expression of selected genes associated with the CD40 pathway (NF-κB and MAPK) and genes related to a variety of APCs and inflammatory mediators between baseline and 24 hours after sotigalimab (n = 23 paired samples). C, Gene set enrichment analysis of gene expression data from injected tumors at 24 hours after sotigalimab compared with baseline (n = 23 paired samples), following normalized enrichment scores for selected upregulated and downregulated gene sets. D, Immune cell signatures. The values are the mean differences with 95% CI in the expression of marker genes for activated DCs (aDC; CCL1, EBI3, INDO, LAMP3, and OAS3), DCs (CCL13, CD209, HSD11B1, CCL17, CCL22, NPR1, and PPFIBP2), and macrophages (CD163, CD68, CD84, and MS4A4A) between baseline and 24 hours after sotigalimab. iDC, immature DCs; Tfh, follicular Th cells; Tgd cells, γδT cells; TLS, tertiary lymphoid structures. E, Volcano plot of major immune cell populations determined by IMC of the injected tumors at 24 hours after sotigalimab compared with baseline in 23 patients. Forty-seven ROIs were assessed at each timepoint. F, Representative mIF images of tumor tissue stained for DAPI, CD40, CD11c, and DC-LAMP. G, Quantification of immune cell populations (CD40⁺SOX10⁻), DCs (CD11c⁺DC-LAMP⁺), and activated DCs (CD40⁺CD11c⁺DC-LAMP⁺ indicated by white arrows) in injected tumors at baseline and 24 hours after sotigalimab (n = 23 paired samples).

To confirm the inferred enrichment in myeloid cells, we used IMC (Supplementary Fig. S5A) to analyze tumor biopsy samples obtained from twenty-three patients before and 24 hours after sotigalimab administration (Fig. 2E). Sotigalimab induced a significant increase in CD11b⁺ myeloid cells (Fig. 2E; Supplementary Fig. S5B); this increase was more pronounced in patients who exhibited a treatment response (“responders,” n = 12) than in patients who did not have a response (“nonresponders,” n = 11; Supplementary Fig. S5C and S5D). The activation of CD40 was further supported by scRNA-seq data showing the upregulation of several well-known CD40 pathway and antigen presentation genes (e.g., TRAF1, MAPK1, PSMB8, and PSMB10) in macrophages after sotigalimab injection (Supplementary Fig. S5E). mIF demonstrated a significant increase in the density of CD40 in SOX10⁻ nontumor cells, specifically in CD11c⁺DC-LAMP⁺ DCs and CD40⁺CD11c⁺DC-LAMP⁺ DCs after sotigalimab injection (Fig. 2F and G). These changes were more pronounced in responders than in nonresponders (Supplementary Fig. S5F). Tumor cells also had increased CD40 expression, but this difference was not significant (P = 0.14; Supplementary Fig. S5G and S5H). Collectively, these results demonstrate that intratumoral sotigalimab engaged the CD40 pathway and induced the rapid infiltration and activation of APCs.

Circulating Immune Cells Show Early Engagement of the CD40 Pathway and Upregulation of Genes Associated with APCs Followed by T-cell Activation

To evaluate the extent to which intratumoral CD40 activation induces changes in circulating immune cells, we used the NanoString nCounter Human PanCancer Immune Profiling Panel to profile changes in gene expression in six patients’ (four responders and two nonresponders) peripheral blood mononuclear cells (PBMC) from baseline to 3, 8, and 21 days after the initial treatment with sotigalimab plus pembrolizumab. The combination induced the upregulation of APC-related genes, including ITGAX, CD83, PSMB8, PSMB9, PSMB10, CD86, and TAP1, by days 3 and 8 (Fig. 3A; Supplementary Fig. S6A). Furthermore, the combination induced the rapid upregulation of various cellular pathways, including those involved in antigen processing, macrophage function, the TNF superfamily, and Toll-like receptor (TLR) signaling (Fig. 3B; Supplementary Fig. S6B). By day 21, the combination therapy induced the upregulation of T-cell activation–related genes, such as CD3G, CD8A, IFNG, ICOS, TBX21, and GZMB, as well as the upregulation of cellular pathways related to cytotoxicity, the cell cycle, and B-cell and natural killer (NK) cell function (Supplementary Fig. S6C and S6D).

Figure 3.

The combination of sotigalimab and pembrolizumab quickly engages the CD40 pathway and upregulates APC-associated genes in peripheral blood. A and B, NanoString analysis using the nCounter Human PanCancer Immune Profiling Panel performed on PBMC samples obtained from six patients at baseline and 3 days after combination treatment. Heatmap showing the fold change (FC) in the expression of individual genes related to the CD40 pathway, antigen presentation, and T cells and B cells (A) in peripheral blood at day 3 after combination treatment with sotigalimab and pembrolizumab compared with baseline (BL). Heatmaps showing differences (Δ) in pathway signature scores (B) at baseline and day 3 after combination treatment. C–G, Twelve PBMC samples obtained from three patients at baseline and 3, 8, and 21 days after combination treatment subjected to single-cell multiome [simultaneous profiling of gene expression (scRNA-seq) and open chromatin (scATAC-seq) in the same cell] using the 10x Genomics platform. UMAP embeddings (C) of merged scRNA and scATAC profiles of PBMCs collected from patient 1 at 4 timepoints and identification of major immune cell types: Classical (C) monocytes (CD14, CD74, ITGAM, and CD163), non-classical (NC) monocytes (FCGR3A, CSF1R, and CX3CR1), cDCs (CLEC10A, CD1C, CD86, and CD83), plasmacytoid DCs (pDC; SERPINF1, LILRA4, NRP1, and CLEC4C), B cells (IGHD, IGHM, CD19, CD40, IL4R, and MS4A1), plasma cells (JCHAIN, CD38, CD27, and TNFRSF17), NK cells (NCAM1, KIR3DL1, GZMB, NKG7, and PRF1), CD8⁺ cytotoxic T cells (CD8A, CD8B, GZMA, GZMK, GZMH, and IFNG), CD4⁺ T cells (CD4, CD69, SELL, and CCR7), Tregs (CD4, IL2RA, and FOXP3), and Th1 cells (IL18R1, IL12RB2, CCR7, SELL, and CD69). Each dot represents a single cell; each color represents a different cell cluster. Normalized proportions (D) of cell clusters identified in patient 1 at baseline compared with those identified on day 3 after combination treatment. APCs, including myeloid cells and B cells, are labeled in red. Heatmap showing the FC in the expression of individual genes related to the CD40 pathway or antigen presentation (E) in B cells, monocytes, and cDCs from baseline to 3 days after combination treatment (n = 3; each row represents 1 patient). Heatmap depicting dynamic chromatin accessibility and peak-to-gene links (F) from a trajectory analysis of APCs, including myeloid cells and B cells. Heatmaps of TF deviation scores (G) from a trajectory analysis of APCs from scATAC and corresponding expression changes from scRNA.

To explore the heterogeneous transcriptional states and changes in the underlying epigenetic regulation of circulating immune cells in response to sotigalimab plus pembrolizumab, we performed single-cell multiome interrogation by simultaneously profiling PBMCs using scRNA-seq and scATAC-seq collected from three responding patients at days 3, 8, and 21 after initial treatment. Cells from all scRNA-seq samples were clustered and annotated using the expression of canonical marker genes, and these clusters were superimposed on scATAC clusters (Fig. 3C). The immune cell types we identified included CD4⁺ T cells, CD8⁺ T cells, regulatory T cells (Treg), NK cells, B cells, plasma cells, classical monocytes, non-classical monocytes, classical DCs (cDC), and plasmacytoid DCs. In line with the early engagement of the CD40 pathway and upregulation of APC-related genes, the fractions of cDCs (CLEC10A, CD1C, CD86, and CD83), classical monocytes (CD14, CD74, ITGAM, and CD163), and non-classical monocytes (FCGR3A, CSF1R, and CX3CR1) were increased on days 3 and 8 (Fig. 3D; Supplementary Fig. S6E). Moreover, cDCs, monocytes, and B cells were activated after treatment initiation, as evidenced by their upregulated CD40 pathway–related and APC-related genes (e.g., TRAF3, MAPK1, NFKBIA, ITGAX, CD83, CD86, PSMB8, and PSMB9; Fig. 3E).

We next performed a pseudotime trajectory analysis of APC activation and differentiation using three patients’ scATAC-seq data from baseline and 3, 8, and 21 days after initial treatment with sotigalimab plus pembrolizumab. The cis-elements that were accessible in the first major shift in the trajectory included enhancers for CD86, SPI1, CLEC7A, NFKB2, CCL2, CXCL8, JUN, and FOS and other genes critical to APC activation; the corresponding scRNA-seq data showed that these genes were also upregulated on days 3 and 21 (Fig. 3F; Supplementary Fig. S7A). Pseudotime trajectory analysis of transcription factor (TF) activity using chromVAR deviation scores revealed dynamic chromatin changes at the binding sites of CTCF and TCF3 on day 3, suggesting that these TFs play roles in epigenetic reprogramming in response to treatment. These changes were followed by the activity of ATF3, ATF4, CEBP, JUN, and AP1, which is consistent with these TFs’ roles in inflammatory response (Fig. 3G; Supplementary Fig. S7B).

An expanded analysis of the combination therapy–driven changes in lymphocyte subpopulations revealed that the fractions of CD4⁺ T cells, CD8⁺ T cells, NK cells, and B cells were unchanged on day 3 after the start of sotigalimab plus pembrolizumab but were increased on days 8 and 21 (Supplementary Fig. S6E). In addition, gene expression analysis of specific T-cell subset markers showed the upregulation of cytotoxic markers (e.g., GZMA, GZMK, and PRF1) on days 8 and 21 (Supplementary Fig. S7C). Pseudotime trajectory analysis of scATAC-seq data from patients 1 and 2 revealed biphasic TF activation in CD4⁺ cells. The cis-elements that were accessible in the initial cell-state switch involved the activation of TFs such as JUN, AP1, and ATF3, whereas those that were accessible in the later switch involved Krüppel-like factors (KLF), which aligns with these TFs’ roles in T-cell activation and differentiation. A delayed but similar trend in TF activity was observed for patient 2 (Supplementary Fig. S8A). A similar analysis of CD8⁺ T cells showed that the cell-state switch involved the activation of BATF, TCF4, and RUNX2 in patients 1 and 3 and the activation of ATF1, ATF2, KLF3, and KLF9 in patient 2; these TFs have been reported to play a critical role in regulating T-cell cytotoxicity (Supplementary Fig. S8B). The pseudotime trajectory analysis results for the three patients are summarized in Supplementary Fig. S8C.

The Combination Therapy Induces Broad Innate and Adaptive Immune Activation in Injected Tumors and Cytotoxic Cell Signature in Non-injected Tumors

To evaluate the impact of the combination therapy on inducing an inflammatory response and converting immunologically cold tumors to hot tumors, we subjected 21 tumor samples collected at baseline and 6 weeks after treatment initiation to gene expression profiling. At 6 weeks, injected tumors showed the significant upregulation of genes related to T-cell infiltration and APCs (CD8A, IFNG, GZMA, GZMB, ITGAX, LAMP3, PSMB9, PSMB10, CD68, CD19, and CD86; Fig. 4A). In total, 272 genes were significantly upregulated (P ≤ 0.05). These transcriptional changes were accompanied by enhanced pathways involved in antigen processing, macrophage function, and T- and B-cell functions (Supplementary Fig. S9A). Cell-type analysis showed enhanced expression of genes related to CD8⁺ T cells, cytotoxic cells, macrophages, B cells, and Th1 cells (Fig. 4B). These signatures increased after treatment in responders, compared with their baseline samples (Fig. 4B), but remained unchanged in nonresponders (Fig. 4C). There were no significant changes in Th2, Th17, or Treg immune signatures (Fig. 4B and C). IMC analysis of tumor biopsy samples from 24 patients (Fig. 5A) confirmed that the densities of key adaptive and innate immune cells, including CD11b⁺ myeloid cells, macrophages, B cells, CD4⁺ T cells, CD8⁺ granzyme B⁺ T cells, and CD8⁺TIM-3⁺ T cells, were significantly increased at 6 weeks (Fig. 5B and C). A total of 47 regions of interest (ROI) were assessed at each timepoint (∼2 ROIs per patient; Fig. 5B and C; Supplementary Fig. S9B and S9C). Compared with nonresponders, responders had higher densities of myeloid cells and CD8⁺ granzyme B⁺ T cells but lower densities of CD4⁺ T cells (Supplementary Fig. S9D). The combination therapy altered 169 cell interactions, increasing 63.3% and decreasing 36.7% of them (Fig. 5D). Notably, it enhanced spatial interactions, particularly among CD4⁺ T cells, macrophages, CD74⁺HLA-DR⁺ macrophages, B cells, and CD8⁺ granzyme B⁺ T cells; it also enhanced spatial interactions between macrophages and CD8⁺ granzyme B⁺ T cells, as well as among melanoma cells, B cells, and CD4⁺ T cells, indicating a more robust and coordinated immune response, thereby potentially improving tumor targeting and immune activation (Fig. 5E).

Figure 4.

The combination of sotigalimab and pembrolizumab induces broad innate and adaptive immune activation in injected tumors at the transcript levels. NanoString gene analysis with the nCounter Human PanCancer Immune Profiling Panel was performed to quantify the transcript levels of 770 genes. A, Volcano plot of gene expression determined from an analysis of RNA extracted from the injected tumors at week 6 after sotigalimab administration compared with baseline (BL; n = 21 paired samples)Each gene’s −log₁₀ (P value) and log₂ fold change (log₂ FC) with the selected covariate. Highly statistically significant genes fall at the top of the plot above the horizontal lines, and highly differentially expressed (DE) genes fall to either side. The horizontal lines indicate various false discovery rate (FDR) thresholds or, for unadjusted P values, P value thresholds. Colored dots represent genes with P values below the given FDR or P value threshold for unadjusted P values. Genes are colored if the resulting P value is below the given FDR or P value threshold. B and C, Box and whisker plots indicating gene expression–based cellular score in responders (B) and non-responders (C) for macrophages, CD8⁺ T cells, cytotoxic cells, B cells, NK, Th1, Th2, Th17, and Tregs in injected tumors at week 6 after sotigalimab administration compared with baseline (n = 21 paired samples). Responders (R, n = 13). Non-responders (NR, n = 8). Paired t test *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not statistically significant.

Figure 5.

The combination of sotigalimab and pembrolizumab induces broad innate and adaptive immune activation in injected tumors at the protein levels. A, Representative IMC images illustrating tumor cells and major immune cell populations stained for CD20, CD14, CD68, CD163, CD11b, CD74, CD3, Ki67, CD4, CD8a, and FOXP3. B, Volcano plot showing the mean differences in major immune cell populations, as determined by IMC, in injected tumors at week 6 after combination therapy compared with baseline (BL) in 24 patients with 47 ROIs assessed at each timepoint. C, Quantification of macrophages, B cells, CD8⁺ granzyme B⁺ T cells, and CD8⁺TIM3⁺ T cells in terms of their proportion of total cells in injected tumors at baseline and week 6 after combination therapy in 24 patients. Forty-seven ROIs (∼2 per patient) were assessed at each timepoint. D, Pie chart showing the total proportions of interactions that were significantly increased or decreased in injected tumors following combination therapy compared with baseline in 24 patients. E, Network diagram illustrating interactions among various immune cell populations and melanoma cells that were significantly increased or decreased in injected tumors following combination therapy compared with baseline in 24 patients with 47 ROIs assessed at each timepoint. The edge color represents the direction of change (red for increased interactions and blue for decreased interactions); edge thickness indicates significance (only interactions with an FDR < 0.05). Node size and color represent the number of interactions that change (green indicates an increase in interactions, purple represents a decrease, and gray denotes interactions that remain significantly unchanged).

We then investigated whether the broad immune activation we observed in injected tumors also occurred in non-injected tumors. We subjected 16 non-injected tumor samples, collected at baseline and 6 weeks after treatment initiation, to gene expression profiling. In contrast to the broad immune response observed in injected tumors, non-injected tumors exhibited a more selective transcriptional response, marked by limited upregulation of antigen presentation pathways and enrichment of genes associated with T-cell infiltration, activation, and effector function (CD8A, IFNG, ICOS, GZMA, and GZMB; Fig. 6A). Cell-type analysis revealed the enhanced expression of genes for CD8⁺ T cells, cytotoxic cells, and Th1 cells, which were enriched in responders after treatment, compared with their baseline samples, but remained unchanged in nonresponders (Supplementary Fig. S10A). No significant changes were noted in Th2, Th17, and Treg immune signatures (Supplementary Fig. S10A). We validated the increase in T-cell infiltration using mIF, which demonstrated a robust increase in the density of CD8⁺ T cells in injected tumors in 9 of 13 responders and 3 of 13 nonresponders (Supplementary Fig. S10B). CD8⁺ T-cell density also increased in non-injected tumors in 5 of 10 responders and 3 of 9 nonresponders (Supplementary Fig. S10C).

Figure 6.

The combination of sotigalimab and pembrolizumab induces broad innate and adaptive immune activation in both injected and non-injected tumors. A, Volcano plot of gene expression determined from an analysis of RNA extracted from non-injected tumors at week 6 after combination therapy compared with baseline (BL; n = 16 paired samples). DE, differentially expressed; FC, fold change. B and C, TCRβ sequencing was performed to determine the percentage of T-cell infiltration (the mass estimate of the fraction of productive cells obtained through the immunoSEQ analyzer; left) and clonality (right) in biopsy samples of injected tumors (n = 16 paired samples; B) and non-injected tumors (n = 13 paired samples; C),; paired t test. D–G, Differential abundance analysis of injected and non-injected tumor samples collected at baseline and week 6 after combination therapy. D, Scatter plot of T-cell clone frequencies in biopsy samples of injected and non-injected tumors collected before and after combination therapy. The blue dots represent differentially abundant clones in injected tumors at week 6, which were expanded clones. The orange dots denote differentially abundant clones in non-injected tumors at week 6. The dark gray dots denote clones with no significant difference in abundance. The light gray dots represent T-cell clones excluded from the differential abundance analysis. E, Number of shared clones between injected and non-injected tumors from seven responders (R) and 6 nonresponders (NR) at baseline and week 6 after combination therapy. F, Number of shared clones with high frequency between injected and non-injected tumors from seven responders (R) and 6 nonresponders (NR) at baseline and week 6 after combination therapy, including newly expanded and preexisting clones. G, The Morisita overlap index was used to evaluate clonotype similarity between injected and non-injected tumors from seven responders (R) and 6 nonresponders (NR) at baseline and week 6 after combination therapy. An index of 0 indicates no overlap, whereas an index of 1 indicates perfect overlap.

To evaluate the impact of the combination therapy on clonal T-cell populations among tumor T-cell populations within injected and non-injected tumors, we subjected tumor biopsy samples obtained at baseline and 6 weeks after treatment initiation to TCRβ sequencing. The analysis confirmed the increased T-cell infiltration and revealed increased T-cell clonality in both injected and non-injected tumors, indicating a more oligoclonal T-cell repertoire (Fig. 6B and C). These increases were more pronounced in responders compared with nonresponders (Supplementary Fig. S10D and S10E). An analysis of clone sharing and dynamics between injected and non-injected tumors revealed the expansion of the number of the shared clones between these tumors after treatment initiation, including those with high frequency (Fig. 6D–F). In addition, newly expanded and/or newly detected clones emerged in both tumor sites by week 6 (Fig. 6F). Expanding shared clones were seen in 5 of 7 responders and 2 of 6 nonresponders (Fig. 6E and F). Using the Morisita overlap index, we found that low similarity between the T-cell repertoires of the injected and non-injected tumors at baseline but increased similarity at week 6 in 4 of 7 responders and 2 of 6 nonresponders (Fig. 6G). These findings indicate that the therapy induced overlap between injected and non-injected tumors and generated a novel immunologic response that extended beyond the injected tumors, rather than simply an expansion of preexisting clonotypes.

Given the broad immune activation after the initiation of the combination treatment, we hypothesized that benefit from combination treatment could lead to a more effective antitumor immune response through APC activation. To test this hypothesis, we compared the posttreatment gene expression profile of the patients who received the combination treatment in the present study with that of treatment-naïve patients who were enrolled to receive anti–PD-1 monotherapy in the trials by Riaz and colleagues and Gide and colleagues (22, 23). Compared with responders who received anti–PD-1 monotherapy, responders who received the combination treatment had preferential activation of the CD40 pathway and upregulation of APC-related genes (TRAF6, MAPK1, MAPK14, CD40, CD83, CD86, TAP1, TAP2, PSMB8, and PSMB9; Supplementary Fig. S11A) but similar T-cell activity between the two groups. Notably, combination therapy responders exhibited higher increases in cytotoxicity genes (GZMM, GZMB, and PRF1), whereas anti–PD-1 monotherapy responders had higher increases in Th2-, Th17-, and Treg-related genes (GATA3, RORA, RORC, and FOXP3; Supplementary Fig. S11B). These findings suggest that the combination therapy activates distinct pathways, thereby activating APC-related genes and cytotoxicity while downregulating Th2 cells, Th17 cells, and Tregs, potentially boosting the antitumor response.

Baseline Immunologic Features Associated with Response to Anti–PD-1 Monotherapy Do Not Predict Response to the Combination of Sotigalimab and Pembrolizumab

Resistance to anti–PD-1–based immunotherapy is associated with a lack of tumor T-cell infiltration and low levels of MHC-II– or IFNγ-mediated immune activation at baseline (23–28). Given sotigalimab’s impact on APCs, we hypothesized that the combination therapy’s benefit is more strongly associated with the presence of APCs, rather than T cells, at baseline. To test this hypothesis, we first used the nCounter Human PanCancer Immune Profiling Panel to analyze baseline injected tumors from 23 patients (10 nonresponders and 13 responders). The results of the analysis are shown in Fig. 7A. Unlike anti–PD-1 monotherapy, the combination treatment elicited clinical responses in many patients whose baseline biopsy samples had low T-cell infiltrates and low levels of the MHC-II/IFNγ signature. Notably, gene signatures did not differ significantly between responders and nonresponders. An analysis of transcriptional signatures identified few markers that statistically predicted OS or PFS. A high angiogenesis pathway score was associated with longer OS and PFS, whereas neutrophil score was correlated with OS but not PFS. In contrast, Th1 cells and IFNG expression were associated with unfavorable clinical outcomes. We found no significant associations between the gene signatures related to major immune cell subsets (including DCs, B cells, macrophages, NK cells, mast cells, eosinophils, γδT cells, Th cells, CD8⁺ T cells, cytotoxic cells, follicular Th cells, Th2 cells, and Th17 cells) and OS or PFS.

Figure 7.

Clinical responses to the combination of sotigalimab and pembrolizumab do not require MHC-II expression or IFNγ-mediated immune activation at baseline. A, Heatmap showing the log₂ of individual genes related to MHC-II expression, IFNγ-mediated immune activation, and cell-type scores for major immune cell populations using NanoString data from baseline biopsy samples of injected tumors from 10 nonresponders (NR) and 13 responders (R). The heatmap is annotated with clinical response (top) and statistical associations with response, PFS, and OS (right). Black dots indicate statistically significant relationships. aDC, activated DCs; FC, fold change; iDC, immature DCs; Tfh, follicular Th cells; Tgd cells, γδT cells. B, Heatmap showing the log₂ cellular pathway scores related to antigen processing, cytotoxicity, T-cell function, B-cell function, macrophage function, NK cell function, TLRs, and the TNF superfamily using NanoString data for baseline biopsy samples of injected tumors from 10 nonresponders (NR) and 13 responders (R). C, TCRβ sequencing with the immunoSEQ assay was performed to examine T-cell infiltration and T-cell clonality in baseline injected tumors from nine responders (R). D, mIF was used to quantify CD11c⁺ myeloid cells, CD8⁺ T cells, CD40⁺SOX10⁻ cells, DCs (CD11c⁺DC-LAMP⁺), and activated DCs (CD40⁺CD11c⁺DC-LAMP⁺) in injected tumors at baseline from 13 nonresponders (NR) and 13 responders (R). E, Kaplan–Meier curves and log-rank P values for the OS and PFS of patients with high (Hi) densities of both CD11c⁺ myeloid cells and CD8⁺ T cells (top tertile, green) and patients with low densities of other populations (bottom 2 tertiles, purple), as defined by mIF, in injected tumors at baseline. F, Kaplan–Meier curves and log-rank P values for the OS and PFS of patients with high densities of CD11c⁺ myeloid cells (top tertile, green) and patients with low (Lo) densities of CD11c⁺ myeloid cells (bottom 2 tertiles, purple) in injected tumors at baseline. G, Kaplan–Meier curves and log-rank P values for the OS and PFS of patients with high densities of CD8⁺ T cells (top tertile, green) and low densities of CD8⁺ T cells (bottom 2 tertiles, purple) in injected tumors at baseline.

In addition, the cellular pathway signatures related to antigen processing, macrophage function, and B- and T-cell functions did not differ between responders and nonresponders (Fig. 7B). Higher baseline T-cell infiltration and TCR clonality have been linked to clinical response to anti–PD-1 treatment (25, 29). In the present trial, many patients with low T-cell infiltration and T-cell clonality had clinical responses (Fig. 7C). In addition, neither baseline T-cell infiltration nor T-cell clonality differed between responders and nonresponders (Supplementary Fig. S12A and S12B). mIF staining of baseline injected tumors from 26 patients (13 nonresponders and 13 responders) revealed that responders exhibited significant higher baseline of CD11c⁺ myeloid cells (P = 0.01) and CD8⁺ T cells (P = 0.04) compared with nonresponders (Fig. 7D; Supplementary Fig. S12C and S12D), despite many patients whose biopsies showed low levels of CD11c⁺ and CD8⁺ T-cell infiltrates suggesting the heterogeneity of baseline immune cell infiltration among responders (Supplementary Fig. S12E). Furthermore, responders showed increased spatial interactions between APCs and T cells compared with nonresponders (Supplementary Fig. S12F). Importantly, we found that tumors with high baseline levels of both CD11c⁺ myeloid cells and CD8⁺ T-cell infiltrates exhibited improved clinical outcomes (OS and PFS; Fig. 7E), but neither of these cell populations alone correlated with clinical outcomes (Fig. 7F and G). Collectively, baseline immune features did not predict response to combination therapy as they did for anti–PD-1 monotherapy, but the presence of both CD11c⁺ and CD8⁺ T cells in the tumor seem to be crucial for optimal clinical outcomes.

The Combination of PD-1 Inhibition and CD40 Activation Induces Distinct Changes at the Single-Cell Level in Intratumoral Immune Cell Populations in a Preclinical Melanoma Model

We investigated the effect of combining intratumoral CD40 agonist with systemic PD-1 immunotherapy on tumor-infiltrating immune cells at single-cell resolution in a poorly immunogenic murine B16 melanoma model, revealing key differences between the two pathways (Supplementary Fig. S13A). The combination therapy significantly outperformed single-agent treatments, reducing tumor weight by fivefold compared with anti–PD-1 alone (0.058 ± 0.01 g vs. 0.26 ± 0.05 g) at day 14 (Supplementary Fig. S13B). Unsupervised scRNA-seq clustering revealed 26 immune cell clusters defined by classical markers, with Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) showing treatment-dependent shifts in their relative abundance (Supplementary Fig. S13C and S13D). Compared with the control treatment, the CD40 agonist alone or in combination with anti–PD-1 increased the frequency of intratumoral CD8⁺ T cells (Supplementary Fig. S14A). Clustering analysis identified six subsets of tumor-resident CD8⁺ T cells, including terminal effector CD8⁺ T cells (c2), characterized by high expression of Gzmb, Prf1, and Ifng and exhaustion markers (Pdcd1, Havcr2, and Lag3), which expanded in tumors from mice treated with the combination therapy compared with control mice. In addition, the combination therapy induced the expansion of the effector memory CD8⁺ T cells (c6), expressing high levels of Gzmk and low levels of exhaustion markers. The combination therapy also stimulated CD8⁺ T-cell proliferation (c10 and c12) compared with the control treatment (Supplementary Fig. S14B). In contrast, the frequency of Treg (c11) was decreased in response to the combination therapy or CD40 alone (Supplementary Fig. S14C). Clustering analysis of myeloid populations revealed an increase in inflammatory M1 macrophages (c1; Nos2^hi) following combination therapy (Supplementary Fig. S14D). Conversely, the combination therapy reduced the populations of immunosuppressive M2 macrophages (Mrc1^hi) and decreased proliferative M2 cells in tumors (Supplementary Fig. S14D). Combination therapy promoted recruitment and activation of neutrophils in tumors (Supplementary Fig. S14D), specifically increasing activated N1 neutrophils (c8) marked by high Nos2, IL1a, and Cxcl2 expression (Supplementary Fig. S14D). Overall, intratumoral CD40 activation, especially when combined with PD-1 blockade, enhances innate and adaptive immunity while suppressing immunoregulatory cells in the tumor microenvironment.

Discussion

The concept of “in situ” immunization, in which the tumor itself becomes a source of antigens to stimulate systemic immune responses, has gained significant interest in recent years. Several intratumoral therapies have been explored, aiming at eliciting systemic immune effects (30–40). Our study represents the first-in-human trial of the intratumoral sotigalimab combined with pembrolizumab in patients with ICB-naïve MM, providing novel insights into the feasibility and therapeutic potential of this approach. The combination was well tolerated, with no DLTs, treatment-related deaths, or discontinuations and a safety profile similar to pembrolizumab monotherapy (41, 42). Importantly, we did not observe cytokine release syndrome, which is often associated with CD40-based therapies (43–45). Our safety findings align with prior intratumoral CD40 studies, further supporting the safety and feasibility of this therapeutic approach (46, 47). Our patient cohort shared significant baseline characteristics with those in other melanoma trials, such as KEYNOTE-006 (anti–PD-1 monotherapy; ref. 41), ECHO-301/KEYNOTE-252 (anti–PD-1 therapy plus an indoleamine 2,3-dioxygenase 1 inhibitor; ref. 42), and MASTERKEY-265 (anti–PD-1 therapy plus talimogene laherparepvec; ref. 48). However, a notable distinction was the higher proportion of patients with traditionally unfavorable features for anti–PD-1 response in our study, including 57% with negative PD-L1 status and 53% with elevated LDH levels. Despite these baseline features, the combination therapy yielded an ORR of 50% and a DCR of 92% at the RP2D, which may be considered favorable when viewed alongside outcomes previously reported for anti–PD-1 monotherapy with pembrolizumab in patients with MM (41, 42). Furthermore, the combination therapy demonstrated durable clinical benefit with PFS rates of 62.5% and 55% and OS rates of 94% and 92% at 6 and 12 months, respectively. In contrast, the KEYNOTE-006 trial indicated that pembrolizumab monotherapy every 3 weeks reported an estimated 6-month PFS of 46.4% and 12-month OS of 68.4% (41). Similarly, ECHO-301/KEYNOTE-252 indicated that pembrolizumab alone showed 6- and 12-month PFS rates of 45.8% and 36.6% and 6- and 12-month OS rates of 87.2% and 74.1%, respectively (42). These numerically favorable outcomes should be interpreted cautiously due to differences in study design, patient populations, and sample size. Subgroup analysis further revealed that patients with unfavorable baseline characteristics, such as negative PD-L1 status and elevated LDH levels (49, 50), still showed favorable clinical outcomes, with ORRs of 54.6% and 58.5%, respectively. In PD-L1–negative tumors, the observed ORR of 54.6% is clinically encouraging. However, PD-L1–stratified ORRs are not consistently reported across pembrolizumab-based trials (41, 42, 48), and the available data for PD-L1–negative subgroups are not presented in a format directly comparable with our study, limiting cross-trial comparisons. These findings are biologically plausible, given the mechanism of CD40 agonism in enhancing antigen presentation, and should be interpreted as hypothesis generating, warranting further evaluation in larger, biomarker-stratified studies. Subgroup analysis further revealed that patients with elevated LDH, a marker of primary resistance to anti–PD-1 monotherapy, have historically shown modest to limited activity (42, 48, 51), with ORRs of ∼15 to 25%. In contrast, these patients in our study achieved an ORR of 58.5%, suggesting that this combination regimen may overcome resistance in this high-risk population. These findings are consistent with emerging evidence from studies evaluating IDO/PD-L1 vaccines in combination with nivolumab (52, 53). Although these observations are preliminary and derived from a nonrandomized cohort, they may suggest potential activity in patient subgroups typically considered less responsive to anti–PD-1 monotherapy.

Our study investigated the immune dynamics triggered by sotigalimab and the combination therapy, with a particular focus on the activation of the CD40 pathway. A key observation was the rapid activation of the CD40 pathway, leading to rapid infiltration and activation of myeloid cells, including CD11c⁺DC-LAMP⁺ DCs, CD40⁺CD11c⁺DC-LAMP⁺ DCs, and macrophages in the injected tumors within 24 hours after sotigalimab administration. These findings are consistent with recently published data, which demonstrated that intratumoral CD40 injections activate CD40 signaling, promote cDC1 maturation, and enhance antigen presentation (47). In addition, the combination therapy induced a dynamic immune response in peripheral blood cells, characterized by an early increase in the proportion and activation states of cDCs (CLEC10A, CD1C, CD86, and CD83), classical monocytes and non-classical monocytes, and B cells, indicated by the upregulation of CD40 pathway–related and APC-related genes. Together, these changes provide compelling evidence for the targeted activation of the CD40 pathway by sotigalimab, supporting its proposed mechanism of action. Consistent with previous studies, our findings further indicate that CD40 agonists can substantially remodel the tumor immune landscape and enhance APC function (43–45, 54–56). Our preclinical findings in the B16 melanoma model further support this mechanistic framework, demonstrating that combined CD40 activation and PD-1 blockade remodel the tumor microenvironment toward a more inflammatory and cytotoxic state while reducing immunosuppressive populations.

Our findings also suggest that sotigalimab may act as an in situ vaccine, initiating T-cell priming and systemic immune responses. Before treatment, the T-cell repertoires in the injected and non-injected tumors showed limited overlap. However, after combination therapy, several responders exhibited an expansion of shared, newly formed T-cell clones in both injected and non-injected tumors, including clones with significantly increased abundance. Importantly, the immune response induced by sotigalimab and pembrolizumab was distinct from that induced by the TLR9 agonist tilsotolimod, another immunotherapy agent. Whereas tilsotolimod and ipilimumab primarily amplify preexisting responses to shared antigens between injected and non-injected tumors (34), sotigalimab and pembrolizumab initiate new T-cell responses which may occur through T-cell priming and are important for effective antitumor immunity. These results suggest that sotigalimab’s effect on APCs may contribute to its efficacy by initiating a local antitumor immune response and an abscopal effect, as evidenced by the clinical responses in both injected and non-injected tumors. The systemic immune response elicited by this combination therapy is especially noteworthy in patients whose tumors initially lacked significant T-cell infiltrates and exhibited low MHC-II/IFNγ signatures, features commonly associated with resistance to pembrolizumab alone (26). Our findings suggest that the combination therapy can overcome resistance mechanisms and stimulate a systemic immune response by activating APCs, promoting CD8⁺ T-cell infiltration, and upregulating MHC-II/IFNγ signatures in both injected and non-injected tumors. Additionally, the presence of CD11c⁺ myeloid cells and CD8⁺ T cells in the tumor seems to be crucial for optimal clinical outcomes. A larger, randomized phase II trials with a biomarker-driven approach to identify patients with favorable myeloid and T-cell profiles could improve patient selection and maximize the therapeutic benefit of the combination therapy.

This study had some limitations, including a small study population and the absence of a randomized controlled trial with anti–PD-1 monotherapy as a comparator, requiring a cautious interpretation of the efficacy outcomes. Another limitation of this study is the administration of only four doses of sotigalimab to a single tumor lesion. Although this approach demonstrated some positive immune activation and clinical responses, it is possible that a more prolonged dosing regimen may be necessary to achieve optimal immune responses and fully maximize the therapeutic benefit. Moreover, our results, compared with historical monotherapy data (22, 23), demonstrate that the combination therapy activates distinct immune pathways, enhancing APC-related genes and cytotoxic T-cell response, with responders primarily activating the CD40 pathway. These findings support combining CD40 activation with pembrolizumab to overcome PD-1 monotherapy limitations and enhance antitumor immune responses, presenting a promising immuno-oncology strategy.

In conclusion, our study demonstrates that the combination of sotigalimab and pembrolizumab offers a promising therapeutic strategy for patients with melanoma, particularly those with challenging baseline characteristics that typically confer resistance to anti–PD-1 therapy. The combination therapy rapidly activates APCs and enhances immune activation, leading to immune responses in both injected and non-injected tumors. Further research, particularly through larger randomized phase II trials, is needed to better evaluate sotigalimab therapeutic potential in the “in situ” immunization approach.

Methods

Patients, Eligibility, Study Design, and Treatment Administration

This study was written and conducted in accordance with the principles of the Declaration of Helsinki. All patients provided written informed consent before any protocol-specified procedures.

Eligible patients (≥18 years) with histologically or cytologically confirmed cutaneous or mucosal melanoma (ocular excluded), unresectable stage III to IV disease, at least two injectable lesions (≥10 mm), and ICB-naïve tumors were eligible. Patients had an ECOG performance status of 0 to 1, met standard hematologic and organ function requirements, were human immunodeficiency virus seronegative, and, if of childbearing potential, had a negative pregnancy test and agreed to use effective contraception. Key exclusion criteria included prior PD-1/PD-L1 or CD40 therapy, active autoimmune disease, uncontrolled infections, immunodeficiency, recent cardiovascular events, symptomatic brain metastases requiring steroids, recent systemic therapy/radiation/surgery, live vaccines, pregnancy/nursing, or any condition that could increase risk or interfere with study assessments; stable, treated brain metastases and certain low-risk prior cancers were allowed.

This was an open-label, dose-escalation phase I/II study using an accelerated 3 + 3 design. Thirty-two ICB-naïve patients were enrolled. The primary objectives were to assess the safety and tolerability of intratumoral sotigalimab given with systemic pembrolizumab and to identify the MTD, RP2D, and antitumor clinical activity of the combination therapy. The secondary objectives were to evaluate the immunologic impact of the combination therapy in the blood and injected and non-injected tumors. In the phase I portion, 14 patients were enrolled in 5 dosing cohorts of sotigalimab at 0.1 mg (n = 1), 0.5 mg (n = 1), 1 mg (n = 3), 3 mg (n = 3), and 10 mg (n = 6) in combination with standard pembrolizumab at 2 mg/kg. Dose escalation followed standard 3 + 3 rules, and the MTD was defined as the highest dose level at which ≤1 of 6 subjects experienced a DLT. In phase II, 18 patients received sotigalimab at RP2D in combination with pembrolizumab. All patients received sotigalimab every 3 weeks for a total of 4 doses. Intratumoral injections were performed into a single designated lesion per patient. The site of injection was determined by the principal investigator and the interventional radiologist prioritizing safety and accessibility factors.

Safety and Efficacy Assessments

AEs and laboratory abnormalities were categorized according to the Medical Dictionary for Regulatory Activities and were graded according to the National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events, version 4.03 and by changes from baseline vital signs and clinical laboratory results during and following study drug administration. Immune-related AEs included predefined events and any additional events deemed by the investigator to be immune related. Efficacy evaluations, which included clinical examination and computed tomography or magnetic resonance imaging of known sites of disease, occurred at weeks 8, 17, and 29 and then every 3 months. Tumor response was assessed using RECIST 1.1. All assessments of CR, PR, SD, and PD were confirmed by imaging within 4 weeks after the initial documentation of response.

Biomarker Analysis

Serial blood samples and injected and non-injected tumors were collected before and during treatment to perform in-depth multiomic analyses using TCR sequencing, mIF imaging, IMC, single-cell multiome interrogation by simultaneously performing scRNA-seq and scATAC-seq, and NanoString nCounter Human PanCancer Immune Profiling; details are provided below.

Tumor Needle Biopsy

Biopsy samples were collected from both injected and non-injected tumors. For the injected tumors, biopsies were performed during the screening visit (within 21 days before the first treatment), within 24 hours after sotigalimab administration, and at week 6 following combination therapy. Similarly, biopsies of the non-injected tumors were collected during the screening visit and at week 6.

Gene Expression Assay

RNA was extracted from blood and injected and non-injected tumors using the AllPrep DNA/RNA Kit (Qiagen; RRID: SCR_008539) and analyzed with the nCounter gene expression assay (NanoString Technologies). Hybridization was performed with code sets, and scanning was completed using the nCounter digital analyzer following the manufacturer’s protocol. The gene expression analysis was performed with the Human PanCancer Immune Profiling Panel and nSolver 4.0 (NanoString Technologies; RRID: SCR_003420). nSolver with default settings was used for quality control, and all samples passed the quality control check. Background correction was performed with negative controls, followed by 2-step normalization using a geometric mean of 6 positive controls and 40 housekeeping genes.

TCR Sequencing

Fresh tumor tissue was preserved in RNAlater stabilization solution. Genomic DNA was extracted from injected and non-injected tumors using the AllPrep DNA/RNA Kit (Qiagen; RRID: SCR_008539), followed by amplification and survey-level sequencing of the TCRβ repertoire for each sample using the immunoSEQ assay (Adaptive Biotechnologies). The productive clonality and the mass estimate of the fraction of productive cells were obtained through immunoSEQ analyzer. Differential expression analysis to determine the number of high-frequency shared clones between injected and non-injected tumors, as well as a statistical framework to quantify clonal expansion, was performed using the differential abundance tool in the immunoSEQ analyzer.

Single-Cell Multiome Interrogation

Single-cell multiome interrogation by simultaneously profiling gene expression (scRNA) and open chromatin (scATAC) from the same cell on days 3, 8, and 21 after treatment was performed using the platform from 10x Genomics. Blood samples were collected, and nuclei were isolated according to the 10x Genomics protocol (CG000365). Paired scRNA-seq and scATAC-seq libraries were generated on the Chromium iX Single-Cell Multiome ATAC + Gene Expression platform (10x Genomics) following the manufacturer’s protocol (CG000338) and sequenced on a NovaSeq 6000 sequencing system (Illumina). Nuclei were prepared and counted to ensure quality and concentration. Nuclei were then transposed according to the manufacturer’s protocol. For the generation of barcoded gel beads-in-emulsion (GEM) containing partitioned nuclei, the transposed nuclei suspension was loaded onto the Next GEM Chip J targeting 10,000 nuclei and then ran on a Chromium iX instrument. The GEMs were incubated and quenched in a thermocycler based on the manufacturer’s protocol, which yielded 10x barcodes for DNA from the transposed DNA (for ATAC) and 10x barcodes for full-length cDNA from polyadenylated mRNA (for gene expression). The GEMs were then broken down, and the barcoded products were purified. A preamplification polymerase chain reaction was performed to fill the gaps and generate enough input for subsequent ATAC and gene expression library generation. ATAC and gene expression libraries were generated based on the manufacturer’s protocol, and Illumina-specific P5 and P7 indices were added. The final libraries were quantified using the Qubit HS dsDNA Assay Kit (Thermo Fisher Scientific), and library profiles were analyzed using the High Sensitivity D1000 TapeStation (Agilent Technologies). ATAC and gene expression libraries were pooled separately and sequenced on a NovaSeq 6000 instrument using an SP Reagent Kit (100 cycles; Illumina) targeting 50,000 reads/nuclei for ATAC-seq and 50,000 reads/nuclei for gene expression. The analysis was conducted using 12 PBMC samples obtained from 3 patients at baseline and 3, 8, and 21 days after combination treatment. For each patient, we detected a total of 16,342, 13,559, and 17,035 cells, respectively, that passed the scRNA quality control and 21,690, 17,072, and 29,489 cells, respectively, that passed the scATAC quality control.

Raw single-cell multiome ATAC and gene expression data underwent preprocessing via Cell Ranger ARC v2.0.0 (10x Genomics; RRID: SCR_023897). This process involved demultiplexing cellular barcodes, aligning reads to the human reference genome GRCh38 (hg38), and generating the gene count matrix and aligned fragments file. Quality control metrics were evaluated, and cells were carefully filtered to ensure high-quality data for downstream analyses. For quality filtering, cells with low-complexity libraries (transcripts aligned to fewer than 200 genes), cell debris, and apoptotic cells (>15% mitochondrial transcripts) were excluded. Probable doublets or multiplets were identified and removed using a multistep approach. Briefly, cells with high-complexity libraries (>6,000 detected genes) were excluded, and doublets or multiplets were identified using the DoubletFinder algorithm (RRID: SCR_018771; ref. 57), with expected rates estimated from cell counts and 10x Genomics guidelines. The filtered gene–cell matrix was normalized with scTransform Seurat (SCR_016341; ref. 58), and SAVER (59) was used to recover gene expression from noisy and sparse scRNA-seq data.

Highly variable genes were identified using Seurat’s FindVariableFeatures function, followed by principal component analysis on the top 2,000 genes. Significant principal components were determined with an elbow plot using the ElbowPlot function. Seurat’s FindNeighbors and FindClusters functions were used for clustering and identifying cell types and transcriptional states. UMAP was applied, and the two-dimensional visualization of the cell clusters was carried out using UMAP through the RunUMAP function, with the number of principal components for embedding matching those used in clustering. Differentially expressed genes were detected with FindAllMarkers (false discovery rate–adjusted P < 0.05; log₂ fold change > 1.2). The R package ArchR (RRID: SCR_020982; ref. 60) was used for scATAC-seq data processing and analysis. Cells with a transcription start site enrichment score below 5 and fewer than 1,000 unique fragments were excluded. Doublets were removed based on ArchR parameters. Dimensionality reduction was performed with iterative latent semantic indexing (61), and UMAP was used for single-cell embeddings. Peaks were identified using the MACS2 algorithm (62). Further downstream analyses, including cell label predictions, gene scoring, marker peak identification, and motif enrichment with chromVAR, were performed using ArchR’s default functions. Data from different modalities were integrated using the FindTransferAnchors function, with integration refined through shared 16-bp barcode identifiers to ensure accurate alignment and correlation.

IMC

IMC was performed at MD Anderson’s Flow Cytometry and Cellular Imaging Facility using the Helios CyTOF instrument and the Hyperion Imaging System with laser ablation (Fluidigm) as previously described (63). Briefly, a tissue microarray (TMA) was constructed from formalin-fixed, paraffin-embedded blocks; the TMA included two representative regions of each injected tumors obtained at the screening visit, within 24 hours following sotigalimab injection, and at week 6 after combination therapy and two representative regions of each non-injected tumors obtained at the screening visit and at week 6 after combination therapy. The TMA was deparaffinized, rehydrated, and bleached of melanin pigment using 0.5% hydrogen peroxide in Tris-HCl (pH 10) at 80°C for 15 minutes. This was followed by heat-induced antigen retrieval in Tris-EDTA buffer (pH 9.0) at 95°C for 15 minutes. The samples were blocked with 2% bovine serum albumin and 0.5% horse serum in phosphate-buffered saline (PBS) and then incubated overnight at 4°C with heavy metal–labeled antibodies (Supplementary Table S6). After washing with Tris-buffered saline containing 0.1% Tween, slides were incubated with 0.3125 μmol/L Cell-ID Intercalator-Ir (1:400 dilution) to detect nuclear DNA. Metal-conjugated antibodies were visualized using the Hyperion Imaging System and Helios mass cytometer, with tissue ablation at 200 Hz. The following marker combinations were used to define each population: CD8⁺ T cells (CD3⁺CD8⁺), proliferating CD8⁺ T cells (CD8⁺Ki67⁺), cytolytic cells (CD8⁺ Granzyme B⁺), CD4⁺ T cells (CD3⁺CD4⁺), proliferating CD4⁺ T cells (CD4⁺Ki67⁺), Tregs (FoxP3⁺CD4⁺CD3⁺), proliferating Tregs (FoxP3⁺CD4⁺CD3⁺Ki67⁺), B cells (CD20⁺), monocytes (CD14⁺), macrophages (CD68⁺), HLA-DR⁺ macrophages (CD68⁺HLA-DR⁺), CD74⁺ macrophages (CD68⁺CD74⁺), CD163⁺ macrophages (CD68⁺CD163⁺), myeloid cells (CD11b⁺), and tumor cells (SOX10⁺).

Mass spillover compensation, cell segmentation, cell clustering, phenotyping, and spatial analysis were performed using the Enable Medicine software program. Briefly, spillover was corrected with the R package CATALYST, and mean intensities were exported for single-cell analysis. Image quality was assessed, and low-quality samples with extensive out-of-focus regions, tissue artifacts, or poor staining results were omitted from analysis. Cell segmentation was performed using Deepcell, applying a neural network to identify nuclei and dilating the masks to obtain whole-cell segmentation.

Data analysis and visualization were conducted in R 4.1.2 using a custom R package (https://docs.enablemedicine.com/spatialmap) for normalization, phenotyping, and visualization. After segmentation, a single-cell protein expression matrix was generated by computing the mean expression across pixels within each cell. Cell expression values were normalized across all cells in a sample using quantile normalization, arcsinh transformation, and z-score normalization. To phenotype cells, they were clustered using the Leiden graph clustering algorithm and annotated based on their median expression to identify cell types, validated by manual inspection of multiplexed images.

mIF Imaging

The Opal 7-Color Manual IHC Kit (Akoya Biosciences, #NEL811001KT) was used for the mIF imaging assay. Vendor specifications were followed during the staining protocol. Briefly, the TMA was deparaffinized and rehydrated, followed by antigen retrieval with AR6 buffer (Akoya Biosciences) and the EZ-Retriever System v.3 (BioGenex) at 95°C for 15 minutes, followed by cooling at room temperature for 15 minutes. In this assay, the clones and the antibody concentrations and fluorophores used were as follows: CD8a clone BLR044F (RRID: AB_2891843; 1:250; Opal 690), CD40 clone BLR056F (RRID: AB_2891854; 1:250; Opal 520), CD11c clone EP1347Y (1:1,000; Opal 620), CD1c clone 2A7C11 (RRID: AB_3094460; 1:1,000; Opal 480), DC-LAMP/CD208 clone 1010E1.01 (RRID: AB_2827532; 1:250; Opal 570), and Sox10 clone BC34 (RRID: AB_2861289; 1:100; Opal 780).

The antibody target panel is available in Supplementary Table S7. Image acquisition was performed with the Vectra Polaris Automated Quantitative Pathology Imaging System (Akoya Biosciences). Downstream analysis for cell segmentation and phenotyping was performed using the Visiopharm software program.

Animal Studies

Mouse tumor model: All animal procedures were performed under protocols approved by the Institutional Animal Care and Use Committee at the University of Texas MD Anderson Cancer Center. B16-F10 melanoma cells (ATCC, RRID: CVCL_0159) were grown in Dulbecco’s Modified Eagle medium (Gibco) supplemented with penicillin/streptavidin and 10% heat-inactivated fetal bovine serum (FBS; BenchMark, GeminiBio). Cells at passage 5 were used for tumor inoculations in mice. Ten-week-old female C57BL/6 mice were purchased from The Jackson Laboratory. Mice were housed (5 per cage) in standard caging in accredited animal facilities at MD Anderson under specific pathogen–free conditions in 12:12 light:dark cycles at 22°C. At 12 weeks of age, the mice were injected subcutaneously with B16-F10 cells (0.5 × 10⁶ cells in 100 μL of PBS) in the right flank. The B16 cells were at least 90% viable at the time of injection as assessed by trypan blue exclusion. Tumor growth was quantified by digital caliper measurements, and tumor size was expressed as the tumor length multiplied by the tumor width in millimeters. Mice bearing measurable tumors at day 8 after inoculation were selected for the study. On day 8, the first dose of the CD40 agonistic antibody (clone FGK4.5, BioXcel Therapeutics; 50 mg in 20 mL of dilution buffer, RRID: AB_1107647) was administered intratumorally, and control mice were injected intratumorally with the same volume of dilution buffer (BioXcel Therapeutics). On day 9, the first dose of the anti–PD-1 antibody (clone 29F.1A12, BioXcell Therapeutics, RRID: AB_2687796) at a concentration of 200 μg/100 μL or a rat IgG2a isotype control antibody (BioXcell Therapeutics, RRID: AB_1107769) at the same concentration was injected intraperitoneally. On day 11, mice were treated with the second dose of the CD40 agonistic antibody alone or in combination with the anti–PD-1 antibody. All antibodies were diluted in dilution buffer prior to injection.

scRNA-seq analysis: Mice were euthanized 14 days after tumor inoculation, corresponding to 6 days after CD40 treatment initiation. B16 melanoma tumors were excised and mashed through 40-mm filters, and red blood cells were lysed with ammonium–chloride–potassium buffer for 5 minutes at room temperature followed by washing with PBS containing 2% FBS. Single-cell suspensions from six mice per group were pooled, incubated with antibodies against CD16 and CD32, stained with anti-CD45 antibody (clone 30-F11, BD Biosciences) and SYTOX Blue viability dye (Thermo Fisher Scientific), and then sorted for live CD45⁺ cells using BD FACSAria I (BD Biosciences). Cell counts and viabilities were assessed via trypan blue exclusion using a Countess 3 FL automated cell counter (Life Technologies) prior to sequencing.

Cell capture, TCR and B-cell receptor enrichment, and library preparation were conducted according to 10x Genomics’ 5′ scRNA-seq guidelines (CG000331_Chromium_Next_GEM_Single_Cell_5-v2_User Guide_RevD). The final libraries were quantified using the Qubit HS dsDNA Assay Kit (Thermo Fisher Scientific), and library profiles were analyzed using the High Sensitivity DNA Kit (Agilent Technologies). Libraries were normalized to 5 nmol/L for pooling, and the pools were sequenced on a NovaSeq 6000 S4-XP, 200-cycle flow cell (Illumina). Raw single-cell data were demultiplexed and analyzed with Cell Ranger 7.0.0 and preprocessed using the Cell Ranger 5.0 software suite (10x Genomics; ref. 64). Data filtering was performed using thresholds for cell count, gene count, unique molecular identifiers (UMIs), and mitochondrial percentage. Genes present in <0.1% of cells were excluded. Low-quality cells (<200 genes or <200 reads) were removed. Doublets and multiplets were identified using the DoubletFinder algorithm (57) with expected rates estimated from cell counts and 10x Genomics guidelines and removed. Cells with read counts exceeding three standard deviations above the median were also excluded, as they were likely doublets or multiplets. Cells with mitochondrial content exceeding 5% in tumor samples, as determined by distribution and viability rates, were considered apoptotic and removed from further analysis. This filtering yielded 31,504 cells from the original 36,694 in the raw count matrix. scRNA-seq data analysis was performed using Seurat 4.1.0 (65) following the Seurat workflow. This included normalization with SCTransform (66), principal component analysis, UMAP, cell clustering, and differential analysis to identify markers differentially expressed in each cluster compared with all others. All functions used default parameters. Cluster annotation was done manually based on canonical markers.

Statistical Analysis

Summary statistics were used for continuous variables; numbers and percentages were used for categorical variables. For all analyses, each baseline value was defined as the last measurement taken before the first injection of sotigalimab. Statistical analyses for immune profiling were performed using Prism 10 (GraphPad Software; RRID: SCR_002798). A two-tailed paired t test was used to assess difference measurements at two different timepoints (before and after treatment). A two-tailed unpaired t test was used to assess differences between two independent groups. The statistical significance between three or more groups was determined using one-way ANOVA with Bonferroni multiple comparison. A P value of less than 0.05 was considered statistically significant. The Pearson correlation coefficient was used to compute the correlation between two parameters.

Supplementary Material

Supplementary Table 1

Patient demographics and baseline disease characteristics (n = 32)

Supplementary Table 2

Patient demographics and baseline disease characteristics (n = 24)

Supplementary Table 3

Adverse events to sotigalimab + pembrolizumab (n = 32)

Supplementary Table 4

Injected lesion types, proportion of injected lesions, and overall clinical response (ORR and DCR) in all treated patients (n = 32)

Supplementary Table 5

Injected lesion types and corresponding patient IDs

Supplementary Table 6

List of antibodies and metal tags used for IMC

Supplementary Table 7

List of primary antibodies and Opal dyes used for mIF

Supplementary Figure 1

Peripheral blood lymphocyte, monocyte, and platelet dynamics following sotigalimab and pembrolizumab combination treatment

Supplementary Figure 2

Clinical responses to sotigalimab and pembrolizumab in all treated patients (n = 32) and in patients treated at the RP2D (n = 24)

Supplementary Figure 3

Objective responses by baseline PD-L1, LDH, and BRAF status in patients treated with sotigalimab plus pembrolizumab in all treated patients (n = 32)

Supplementary Figure 4

Baseline biomarkers and clinical responses: ORR by PD-L1, LDH, and BRAF Status, and baseline LDH/tumor burden in all treated patients.

Supplementary Figure 5

Sotigalimab effectively engages the CD40 pathway and induced the rapid infiltration and activation of APCs in injected tumors

Supplementary Figure 6

Peripheral blood multi-omics reveal rapid CD40 pathway engagement and up-regulation of genes associated with APCs following sotigalimab and pembrolizumab combination treatment.

Supplementary Figure 7

Circulating immune cells show early engagement of the CD40 pathway and up-regulation of genes associated with APCs followed by T-cell activation after sotigalimab and pembrolizumab combination treatment

Supplementary Figure 8

Pseudo-time trajectory analysis of CD4+ and CD8+ T cells using scATAC and scRNA data in PBMCs obtained from 3 patients at baseline and after combination therapy (day 3, 8, and 21)

Supplementary Figure 9

The combination of sotigalimab and pembrolizumab induces broad innate and adaptive immune activation in injected tumors

Supplementary Figure 10

The combination of sotigalimab and pembrolizumab induces broad innate and adaptive immune activation in injected and non-injected tumors

Supplementary Figure 11

Immunological changes in injected tumors induced by sotigalimab in combination with pembrolizumab, compared to anti-PD-1 monotherapy.

Supplementary Figure 12

Clinical responses to the combination of sotigalimab and pembrolizumab do not require MHC class II expression or IFN-g–mediated immune activation at baseline

Supplementary Figure 13

The impact of intratumoral CD40 agonist in combination with PD1 therapy on tumor immune landscape at single cell resolution in an immune-cold melanoma model

Supplementary Figure 14

Changes in intratumoral immune cell populations in response to different therapies identified through scRNA-seq analysis

Data Availability

Human scRNA-seq and ATAC-seq data were deposited in the Gene Expression Omnibus (GEO) under accession number GSE324815. Mouse tumor scRNA-seq data were deposited in the GEO under accession number GSE325923. Processed data supporting the findings are provided within the article and its Supplementary Materials or available from the corresponding author upon reasonable request.

Authors’ Disclosures

S.-E. Bentebibel reports grants from Apexigen during the conduct of the study. E. Arslan reports grants from MD Anderson Cancer Center outside the submitted work. D.Y. Duose reports grants from MD Anderson Cancer Center during the conduct of the study. C. Haymaker reports grants from Apexigen during the conduct of the study, as well as grants from Sanofi, Bristol Myers Squibb, Summit Therapeutics, EMD Serono, Genentech, Theolytics, KSQ Therapeutics, Obsidian Therapeutics, Iovance Biotherapeutics, BTG Pharmaceuticals, Novartis, 280 Bio, AstraZeneca, Takeda Pharmaceuticals, Artidis, and ImmunoGenesis, personal fees from Pliant Therapeutics and Regeneron Pharmaceuticals, and other support from BriaCell outside the submitted work. A.T. Mayer reports personal fees and other support from Enable Medicine and other support from Vicinity Bio and Ennovate Pharma outside the submitted work. F.J. Hsu reports personal fees and other support from Apexigen (now subsidiary of Pyxis Oncology) during the conduct of the study. D.H. Johnson reports personal fees from Bristol Myers Squibb, AstraZeneca, Merck, Nectin Therapeutics, Daiichi Sankyo, Signalexis, Amgen, and Tempus outside the submitted work. I.C. Glitza Oliva reports other support from Bristol Myers Squibb, Merck, Pfizer, and Replimune outside the submitted work (research support was provided to the institution and does not represent a conflict of interest for this manuscript). S.P. Patel reports personal fees and other support from Bristol Myers Squibb, MSD, IDEAYA Biosciences, Novartis, and TriSalus Life Sciences, personal fees from Daiichi Sankyo, Fortvita Biologics, Natera, Obsidian Therapeutics, Pfizer, Replimune, Scancell, Sun Pharmaceutical, T3 Pharmaceuticals, and Foghorn Therapeutics, and other support from InxMed, Linnaeus Therapeutics, Lygen, Provectus Biopharmaceuticals, Seagen, and Syntrix Bio outside the submitted work. P. Hwu reports personal fees and other support from Dragonfly Therapeutics, Immatics, and Adventris Pharmaceuticals outside the submitted work. H.A. Tawbi reports personal fees from Bristol Myers Squibb, Merck, Novartis, Medicenna Therapeutics, Iovance Biotherapeutics, Eisai, Pfizer, Roche, IO Biotech, Krystal Biotech, Immunocore Holdings, Regeneron Pharmaceuticals, Corcept Therapeutics, Strand Therapeutics, and T-knife Therapeutics and grants from Bristol Myers Squibb, Novartis, Dragonfly Therapeutics, GlaxoSmithKline, Merck, Roche, Regeneron Pharmaceuticals, and Syntrix Pharmaceuticals outside the submitted work. C. Bernatchez reports personal fees from KSQ Therapeutics outside the submitted work. M.A. Davies reports personal fees from Replimune, Nurix Pharmaceuticals, Roche/Genentech, Pfizer, Novartis, Bristol Myers Squibb, Iovance Biotherapeutics, and THERAtRAME, grants and personal fees from ABM Therapeutics, and grants from Lead Pharma outside the submitted work. K. Rai reports other support from Koshika Therapeutics and Jivanu Therapeutics, personal fees from Daiichi Sankyo, and grants from Cyclacel Inc. outside the submitted work. A. Diab reports grants from Apexigen, personal fees from Memgen, Inc., and grants and personal fees from Regeneron Pharmaceuticals during the conduct of the study. No disclosures were reported by the other authors.

Authors’ Contributions

S.-E. Bentebibel: Conceptualization, resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. D.J. McGrail: Formal analysis, validation, investigation, visualization, writing–review and editing. V. Kochat: Resources, formal analysis, validation, investigation, visualization, writing–review and editing. E. Arslan: Resources, software, formal analysis, validation, investigation, visualization. R. Abdel-Wahab: Formal analysis, validation, investigation, project administration, writing–review and editing. B. Pazdrak: Resources, software, formal analysis, investigation, methodology, writing–review and editing. R. Murthy: Resources, writing–review and editing. N.H.M. Tahon: Resources, formal analysis. S.-N. Cho: Resources, formal analysis, methodology. D.Y. Duose: Resources, methodology. K.M. Wani: Resources, methodology. J. Liu: Formal analysis, investigation, visualization. C. Haymaker: Resources, formal analysis, validation, investigation, methodology, writing–review and editing. J.A. Gomez: Software, formal analysis. H. Sonnemann: Software, formal analysis, investigation, methodology. A.S. Katailiha: Investigation, methodology. B. Nassif-Rausseo: Formal analysis, methodology. M. Rahim: Resources, formal analysis, validation, investigation, visualization, methodology. I. Li: Resources, software, validation, investigation, visualization. A.T. Mayer: Resources, software, validation, investigation, visualization. X. Yang: Resources, funding acquisition, investigation, visualization, writing–review and editing. F.J. Hsu: Resources, investigation, writing–review and editing. J. Zhang: Resources, formal analysis, funding acquisition, validation, investigation, visualization, writing–review and editing. D.H. Johnson: Formal analysis, writing–review and editing. R.N. Amaria: Resources, formal analysis, validation, investigation, visualization, writing–review and editing. I.C. Glitza Oliva: Resources, formal analysis, writing–review and editing. S.P. Patel: Resources, writing–review and editing. P. Hwu: Resources, investigation, visualization, writing–review and editing. K.M. Elsayes: Resources, writing–review and editing. H.A. Tawbi: Resources, investigation, visualization, writing–review and editing. J.K. Burks: Resources, software, investigation, visualization, writing–review and editing. C. Bernatchez: Resources, formal analysis, writing–review and editing. M.A. Davies: Resources, writing–review and editing. K. Rai: Resources, software, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. G. Lizée: Resources, formal analysis, writing–review and editing. S. Ekmekcioglu: Resources, software, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. A. Diab: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.