Abstract Background Fumarate hydratase (FH)–deficient renal cell carcinoma (RCC) is a rare and aggressive subtype associated with hereditary leiomyomatosis and RCC (HLRCC) syndrome. These tumors are characterized by early onset, frequent de novo metastatic presentation, and poor prognosis. Treatment standards are poorly defined, and no validated molecular biomarkers exist for real-time disease monitoring. Circulating tumor DNA (ctDNA) is showing promise as a liquid biomarker in clear cell RCC, but its feasibility and clinical utility in FH-deficient RCC have not been evaluated. The distinct metabolic biology of FH-deficient tumors, driven by fumarate accumulation and the Warburg effect, may influence ctDNA shedding differently than other RCC subtypes. Methods We retrospectively identified patients with pathologically confirmed FH-deficient RCC who underwent serial ctDNA assessments with the Signatera assay (Natera, Inc.) at Memorial Sloan Kettering Cancer Center (October 2021–February 2026). Signatera assay is a tumor-informed, personalized multiplex PCR–based target capture assay tracking up to 16 patient-specific somatic variants. Draws were grouped as surveillance (post-resection, no systemic therapy) or treatment response (on active systemic therapy). Standard-of-care imaging at each draw was categorized as no evidence of disease (NED), stable disease, disease response, or progression based on clinical radiology report and treating-clinician assessment. Draws were classified as concordant (ctDNA-positive with progression; ctDNA-negative with NED/stable/response), discordant (ctDNA-positive with non-progression, or ctDNA-negative with progression), or indeterminate (no contemporaneous imaging or unclassifiable). ctDNA-attributed clinical actions were captured by chart review. Results Thirteen patients (median age 37; 77% male) contributed 60 longitudinal ctDNA draws (median 2 per patient; range 1–16); 19 draws in the surveillance (4 patients) and 41 in the treatment-response cohort (9 patients). ctDNA was detected in 25/60 draws (42%). In the surveillance cohort, detection was 1/19 (5%; single positive at 0.04 MTM/mL); 2 draws were indeterminate, and concordance was 100% (17/17 evaluable). No patients converted from ctDNA-negative to positive during surveillance. In the treatment-response cohort, detection was 24/41 (59%; median burden 4.19 MTM/mL, range 0.13–317.46); 4 draws were indeterminate, and concordance was 81% (30/37 evaluable). Among 32 evaluable ctDNA-negative draws, 30 (94%) corresponded to non-progressive imaging (18 NED, 12 stable). Of 22 evaluable ctDNA-positive draws, 12 (55%) coincided with imaging progression and 10 (45%) occurred with stable or responding disease, raising the possibility of molecular lead time. ctDNA prompted clinical action in 12/60 draws (20%): therapy change after imaging confirmation (n = 5), systemic therapy initiation (n = 2), surgery (n = 2), dose modification (n = 1), treatment resumption (n = 1), and diagnostic workup (n = 1); two involved CT-to-FDG-PET/CT escalation that revealed disease not apparent on CT. Conclusions Tumor-informed ctDNA monitoring using the Signatera assay can be implemented in FH-deficient RCC and demonstrates meaningful clinical actionability in the postoperative non-metastatic and advanced disease setting. ctDNA may complement standard imaging by enabling earlier molecular detection of disease progression and prompting escalation of imaging modality, particularly to FDG-PET/CT. Prospective validation with standardized testing protocols is needed to define the role of ctDNA in the management of this aggressive RCC subtype.

Abstract Metastasis is a complex biological process and the principal cause of cancer-related mortality. With dual blood supply and fenestrated sinusoids, the liver’s haematogenous route uniquely favors seeding by circulating tumor cells (CTCs). Nonetheless, while portal drainage promotes frequent liver metastases from proximal gastrointestinal cancers such as colorectal carcinoma, hepatic colonization by systemically disseminated CTCs, including those originating from breast cancer, is highly inefficient. The cellular barriers and evasion mechanisms that govern this selective vulnerability remain poorly understood. Here we show that hepatic type 1 innate lymphoid cells (ILC1s) are selectively required for the immunosurveillance of breast cancer liver metastasis, the evasion of which is driven by both cancer cell-intrinsic and microenvironment-associated immune mechanisms. In a large patient cohort, liver metastasis occurred at a relatively later stage and was associated with poorer survival, with body mass index (BMI) above the WHO optimal range associated with higher risk. This metabolic vulnerability was recapitulated in a murine preclinical model, in which breast cancer CTC colonization of the liver—but not the lung—was promoted by a high-fat diet that induced hepatic steatosis. Mechanistically, NK cells expanded and activated in response to CTC challenge and employed both lytic granule and death receptor 5 (DR5, encoded by Tnfrsf10b) pathways for cancer cell clearance, hepatic ILC1s preferentially utilized DR5-mediated mechanisms and required high DR5 expression in cancer cells. Notably, loss of heterozygosity of TNFRSF10B in breast tumours was associated with a predisposition to liver metastasis, and poorer patient prognosis. Conversely, metabolic dysfunction-associated steatotic liver disease (MASLD), a pathological risk factor for breast cancer liver metastasis, correlated with depletion of hepatic group 1 innate lymphoid cells—a phenomenon recapitulated in high-fat diet-fed mice. Importantly, the high-fat diet-induced reduction in hepatic ILC1s and enhancement of breast cancer liver metastasis were reversed by treatment with glucagon-like peptide-1 (GLP-1) receptor agonists. Collectively, these findings reveal that breast cancer liver metastasis is primarily restrained by liver-resident ILC1s rather than circulating NK cells. Evasion of this immunosurveillance can arise from cancer cell-intrinsic genomic alterations that compromise DR5 sensitivity or from microenvironmental metabolic perturbations that diminish ILC1 function. Restoration of hepatic ILC1 activity by GLP-1 receptor agonists highlights a potential translational opportunity to extend these agents beyond metabolic disease into oncology, with innate lymphocyte-mediated cancer immunosurveillance as a distinct mechanism of action. Citation Format: Peng Li, Jun Li, Andrew Cornish, Jing Zhang, Xian Zhang, Ming Li, . Evasion of hepatic ILC1 immunosurveillance drives breast cancer liver metastasis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 6785.

Abstract Background There is a need for novel systemic therapies in patients with metastatic renal cell carcinoma (RCC) who progress on standard agents. A hallmark of clear cell RCC (ccRCC) is von Hippel Lindau (VHL) inactivation which affects downstream targets including vascular endothelial growth factor (VEGF). It also increases cellular proliferation via activation of the CCND1 gene which encodes cyclin D1, a key partner to cyclin-dependent kinases 4 and 6 (CDK4 and CDK6) in driving cancer cells through the cell cycle. Translocation-associated renal cell cancers (tRCC) are rare variants that constitutes <5% of RCCs; they lack consistently effective treatment, highlighting the need for novel strategies. A recent report demonstrated downstream effects of the tRCC fusion transcripts include CDKs. Indeed, preclinical models of human tRCC confirmed CDK4 dependency and demonstrated antitumor effect with CDK4/6 inhibition in vitro and in vivo (Gupta S, 2025). We hypothesize that targeting these pathways will provide benefit in treatment refractory patients with metastatic ccRCC and tRCC. Methods This study is an investigator-initiated, two-center (Memorial Sloan Kettering Cancer Center and Johns Hopkins University), single-arm, phase 1b/2 trial to evaluate the combination of MET/VEGFR/AXL-directed multireceptor tyrosine kinase inhibitor (TKI) cabozantinib with the CDK 4/6 inhibitor abemaciclib in patients with ccRCC or tRCC. Patients must have received both immune checkpoint blockade and TKI. The phase 1b dose escalation will require between 4 and 24 patients in a 3 + 3 dose escalation design across 4 dose levels. Treatment is once daily cabozantinib plus twice daily abemaciclib (see table) and is continued until disease progression, intolerance, or discontinuation for other reasons. Restaging imaging is performed every 8 weeks for the first 3 assessments and then every 12 weeks thereafter. Pharmacokinetic assessments will be integrated for both agents during phase 1b. The maximal tolerated dose (MTD) will be defined as the dose level at which a dose limiting toxicity occurs in at most 1 out of 6 patients. The recommended phase 2 dose (RP2D) will integrate this information with longer-term tolerance, pharmacokinetics, and efficacy signal. In phase 2, patients will be treated at the RP2D following a Simon minimax two-stage design with the primary endpoint being objective response rate at 24 weeks. In stage 1, if 4 or more of the first 12 patients achieve an objective response, enrollment will be extended to 13 additional patients for a total of 25. We are currently accruing patients to the phase 1b portion. Results Metastatic RCC remains a clinical challenge, particularly for patients who have exhausted standard lines of therapy including immune checkpoint blockade and tyrosine kinase inhibition. This study addresses a critical unmet need by rationally targeting convergent oncogenic dependencies, VHL-driven VEGF signaling and CDK4/6-mediated cell cycle dysregulation, in both clear cell and translocation-associated RCC. The inclusion of tRCC is especially meaningful given its rarity, poor prognosis, and near-total absence of evidence-based systemic options. By combining cabozantinib’s broad receptor tyrosine kinase inhibition with abemaciclib’s CDK4/6 blockade, this trial is among the first to clinically exploit the CDK4 dependency recently identified in tRCC preclinical models, translating compelling laboratory findings into a therapeutic strategy for patients with no established next-line standard of care. Beyond establishing safety and tolerability, this investigator-initiated trial is designed to generate meaningful early efficacy signals through a rigorous phase 1b/2 framework with integrated pharmacokinetic analysis. If successful, this combination could define a novel treatment paradigm for a refractory RCC population and lay the groundwork for broader exploration of CDK4/6 inhibition across molecularly defined RCC subtypes. Conclusions N/A

Clear cell renal cell carcinoma (ccRCC) is driven by biallelic loss of VHL, leading to pseudohypoxia & elevated angiogenesis. Thus, anti-angiogenesis tyrosine kinase inhibitors (TKI) became the effective standard of care for metastatic RCC alongside immune checkpoint blockade (ICB). However, responses are heterogeneous & predictive biomarkers lacking. Tumor-associated macrophages (TAMs) correlate with poor survival & treatment resistance, yet TAM targeting in RCC remains unexplored due to limited preclinical models. Using a spontaneous ccRCC mouse model with Vhl, p53, & Rb1 deletions, we tested TAM depletion with a clinical-grade CSF1R inhibitor (CSF1Ri) combined with TKI or ICB, followed by scRNAseq profiling. Interestingly TAM depletion abolished ICB response by selectively depleting antigen-presenting TAMs, causing T cell exclusion, but had no impact on TKI response. Surprisingly, hypoxyprobe staining showed that both CSF1Ri & TKI induced severe intratumor hypoxia, enabling us to identify a true hypoxia-specific TAM subset distinct from VHL-loss pseudohypoxia. These TAMs predict poor ICB/TKI outcomes from pretreatment tumors & tracked with treatment-induced necrosis in post-treatment tumors. Notably, TKI exacerbated metastases in novel metastatic syngeneic RCC models, cautioning against hypoxia-inducing agents in such tumors. Our findings highlight mechanistic insights into the failure of CSF1Ri trials, hypoxia-driven resistance, & the need for mechanism-guided biomarkers. National Cancer Institute R01 CA258886 Department of Defense Kidney Cancer Research Program KC180165/W81XWH1910792 Tumor Immunology: Checkpoints, Prevention, and Treatment (TIPT)

Vascular endothelial growth factor receptor-targeting tyrosine kinase inhibitors (VEGFR-TKIs) and aPD1 combinations are effective in multiple solid tumors, particularly in clear cell renal cell carcinoma (ccRCC), due to its characteristic pseudo-hypoxic, hyper-angiogenic state driven by biallelic VHL-loss. However, long-term durability is inferior to dual aPD1/aCTLA4 regimens, yet the mechanisms underlying these differences remain unclear. Since tumor-associated macrophages (TAMs) are implicated in therapeutic resistance, we used immunofluorescence staining and scRNAseq (n = 37) to investigate TAM evolution following VEGFR-TKI, aPD1 and combined VEGFR-TKI/aPD1 treatment in a genetically engineered (Ksp1.3cre-ERT2Vhlf/fp53f/f Rb1f/f) ccRCC mouse model. We further corroborate our findings in a novel human scRNAseq cohort (n = 15) of on-treatment tumor samples from patients who received VEGFR-TKI/aPD1 or aPD1/aCTLA4 therapies and experienced a range of clinical responses before undergoing cytoreductive nephrectomy. By quantifying hypoxia using pimonidazole in mice, we reveal that VEGFR-TKIs induced severe intratumor hypoxia. This enabled the identification of hypoxia-responsive SPP1+ TAMs that are absent in baseline pseudo-hypoxic ccRCC tumors. This proxy of true hypoxia tracks with successful response to VEGFR-TKI/aPD1 in both mouse and human on-treatment samples, reflecting treatment-induced hypoxic necrosis. Paradoxically, high levels of pretreatment hypoxic TAM signatures predicted worse outcomes across multiple VEGFR-TKI/aPD1 trials including JAVELIN101 and IMMOTION150/151, as well as an MSKCC real-world data cohort (n = 44). Furthermore, extended exposure to hypoxia-inducing VEGFR-TKIs and aPD1 exacerbated metastasis in two syngeneic models of ccRCC. In conclusion, our study suggests that VEGFR-TKIs induce tumor hypoxia in responsive ccRCC tumors, but preexisting hypoxia renders tumors refractory to VEGFR-TKI/aPD1 therapies. Our data also suggests that chronic therapy-induced hypoxia and inflammation may promote metastatic evolution, thus offering potential mechanistic insight into the poor durability of VEGFR-TKI/aPD1 regimens across multiple cancer types.

Biological insights often depend on comparing conditions such as disease and health, yet we lack effective computational tools for integrating single-cell genomics data across conditions or characterizing transitions from normal to deviant cell states. Here, we present Decipher, a deep generative model that characterizes derailed cell-state trajectories. Decipher jointly models and visualizes gene expression and cell state from normal and perturbed single-cell RNA-seq data, revealing shared and disrupted dynamics. We demonstrate its superior performance across diverse contexts, including in pancreatitis with oncogene mutation, acute myeloid leukemia, and gastric cancer.

Lymphocytes spanning the entire innate-adaptive spectrum can stably reside in tissues and constitute an integral component of the local defense network against immunological challenges. In tight interactions with the epithelium and endothelium, tissue-resident lymphocytes sense antigens and alarmins elicited by infectious microbes and abiotic stresses at barrier sites and mount effector responses to restore tissue homeostasis. Of note, such a host cell-directed immune defense system has been recently demonstrated to surveil epithelial cell transformation and carcinoma development, as well as cancer cell metastasis at selected distant organs, and thus represents a primordial cancer immune defense module. Here we review how distinct lineages of tissue-resident innate lymphoid cells, innate-like T cells, and adaptive T cells participate in a form of multilayered cancer immunity in murine models and patients, and how their convergent effector programs may be targeted through both shared and private regulatory pathways for cancer immunotherapy.

SUMMARY How genetic lesions drive cell transformation and whether they can be circumvented without compromising function of non-transformed cells are enduring questions in oncology. Here we show that in mature T cells—in which physiologic clonal proliferation is a cardinal feature— constitutive MYC transcription and Tsc1 loss in mice modeled aggressive human malignancy by reinforcing each other’s oncogenic programs. This cooperation was supported by MYC-induced large neutral amino acid transporter chaperone SLC3A2 and dietary leucine, which in synergy with Tsc1 deletion overstimulated mTORC1 to promote mitochondrial fitness and MYC protein overexpression in a positive feedback circuit. A low leucine diet was therapeutic even in late-stage disease but did not hinder T cell immunity to infectious challenge, nor impede T cell transformation driven by constitutive nutrient mTORC1 signaling via Depdc5 loss. Thus, mTORC1 signaling hypersensitivity to leucine as an onco-nutrient enables an onco-circuit, decoupling pathologic from physiologic utilization of nutrient acquisition pathways.

Activation status of T and B cells. (A) T cell gating strategy. T cells from Fig. S1, were split into CD4+ and CD4- (CD8+) T subsets, and subsequently analyzed for the activation markers CD25 and HLA-DR. (B) Percentage of activated CD4+ and CD8 (CD4-) T cell subsets. (C) B cell gating strategy. From B cells, the activation marker HLA-DR (black box) was quantified. (D) Percentage of activated B cells on therapy. Biexponential displaying was done for the dot plots. Mean and {plus minus} SD are provided. B = before treatment; P = on anti-PD-1 therapy; n = number of biopsies analyzed. Solid circles represent responders; open circles represent non-responders (n= 26 baseline; n= 22 on therapy).

Supplementary Figure Legends

Combinations of ICB and anti-angiogenesis tyrosine kinase inhibitors (aATKIs) are the mainstay of metastatic ccRCC front line treatment, but heterogenous responses and lack of biomarkers strongly emphasizes the need to understand RCC-specific mechanisms of effective anti-tumor immunity in the treatment context. Although prior work highlights the critical role and heterogeneity of tumor associated macrophages (TAMs) in prognosis and response to ICB and aATKIs, TAM modulation has yet to be tested in RCC. We therefore assessed the impact of TAM depletion using a CSF1R inhibitor (BLZ945) on drug sensitivity in a genetically engineered mouse model (GEMM) of ccRCC, comprising inducible kidney-specific Vhl, p53 and Rb1 deletion (KVpR). Following tamoxifen-induced driver gene deletion, male mice were monitored for multifocal kidney tumor development from 25-52 weeks using micro-ultrasound. Tumors were confirmed and their volumes tracked by weekly MRI imaging. Mice were randomized to receive either no treatment, or single agent Lenvatinib, aPD-1, BLZ945 or combinations. A total of 299 tumors from 86 mice were included in the final analysis. A total of 37 tumors across treatment arms were collected at responsive or resistant time points, for profiling by single cell RNA sequencing (scRNAseq; 10X) to elucidate the underlying mechanisms of response. Surprisingly, we found that TAM depletion negated an effective aPD-1 response. ScRNAseq analysis suggests that this is due to selective depletion of antigen presenting TAMs, and a subsequent striking loss of CD8 T cell presence. Since dendritic cells were not depleted by BLZ945, this suggests that intratumor T cell activation by TAMs is a major mechanism of aPD-1 therapeutic targeting. Beyond enhancement of a conventional CD8 T cell response induced by Lenvatinib and aPD-1, we also highlight the potential roles of type 3 immunity in RCC immune surveillance. Our results emphasize the deleterious impact of pan-macrophage depletion approaches that are currently being deployed in clinical trials with ICB, and highlight the importance of lesser studied immune cell types as promising avenues for future validation as novel biomarkers and targets. Citation Format: Lynda Vuong, Fengshen Kuo, Hui Jiang, Eduardo A. Mascareno, Phillip Rappold, Erich Sabio, Kate Weiss, Andrew Cornish, Stephen Reese, Mojca Adlesic, David Solit, Ian Frew, Oguz Akin, Yingbei Chen, Ming O. Li, Timothy A. Chan, Ari A. Hakimi. Tumor immune microenvironment determinants of response to lenvatinib and a PD-1 blockade in clear cell renal cell carcinoma [abstract]. In: Proceedings of the AACR Special Conference: Advances in Kidney Cancer Research; 2023 Jun 24-27; Austin, Texas. Philadelphia (PA): AACR; Cancer Res 2023;83(16 Suppl):Abstract nr B029.

Gating strategy used to analyze immune cell subpopulations within tumor infiltrating lymphocytes. Flow cytometry dot-plots on the left side and a dendrogram displaying the gating strategy of six populations on the right. We gated alive (7AAD negative) singlet cells that were positive for leukocyte antigen (CD45+). Then, the following gating was applied to designate NK cells (CD45+CD56+CD3-), B cells (CD45+CD19/CD20+CD3-), monocytes (CD45+CD14+), and T cells (CD45+CD3+). From the monocytes, a subset of MDSC with low HLA-DR positive cells was used. Finally, T regulatory cells (Treg) were described among CD4+ cells and from those, the CD25+HighCD127Low.

. Gating strategy used to study T memory stem and T naïve-like T cell subpopulations within tumor-infiltrating lymphocytes. Dendrogram on the top, and the results of applying Boolean gates on the bottom. X-axis subpopulation of T cells as CD4, CD8 and DN T cells. Y axis, frequency of memory stem cells (MSC) and T naïve-like T cells. B = baseline; P = on therapy; n = 11.

Innate lymphocytes are integral components of the cellular immune system that can coordinate host defense against a multitude of challenges and trigger immunopathology when dysregulated. Natural killer (NK) cells and innate lymphoid cells (ILCs) are innate immune effectors postulated to functionally mirror conventional cytotoxic T lymphocytes and helper T cells, respectively. Here, we showed that the cytolytic molecule granzyme C was expressed in cells with the phenotype of type 1 ILCs (ILC1s) in mouse liver and salivary gland. Cell fate-mapping and transfer studies revealed that granzyme C-expressing innate lymphocytes could be derived from ILC progenitors and did not interconvert with NK cells, ILC2s, or ILC3s. Granzyme C defined a maturation state of ILC1s. These granzyme C-expressing ILC1s required the transcription factors T-bet and, to a lesser extent, Eomes and support from transforming growth factor-β (TGF-β) signaling for their maintenance in the salivary gland. In a transgenic mouse breast cancer model, depleting ILC1s caused accelerated tumor growth. ILC1s gained granzyme C expression following interleukin-15 (IL-15) stimulation, which enabled perforin-mediated cytotoxicity. Constitutive activation of STAT5, a transcription factor regulated by IL-15, in granzyme C-expressing ILC1s triggered lethal perforin-dependent autoimmunity in neonatal mice. Thus, granzyme C marks a cytotoxic effector state of ILC1s, broadening their function beyond "helper-like" lymphocytes.

Immune checkpoint therapy with anti-CTLA-4 and anti-PD-1/PD-L1 has revolutionized the treatment of many solid tumors. However, the clinical efficacy of immune checkpoint therapy is limited to a subset of patients with specific tumor types1,2. Multiple clinical trials with combinatorial immune checkpoint strategies are ongoing; however, the mechanistic rationale for tumor-specific targeting of immune checkpoints is elusive. To garner an insight into tumor-specific immunomodulatory targets, we analyzed 94 patients representing five different cancer types, including those that respond relatively well to immune checkpoint therapy and those that do not, such as glioblastoma multiforme, prostate cancer and colorectal cancer. Through mass cytometry and single-cell RNA sequencing, we identified a unique population of CD73hi macrophages in glioblastoma multiforme that persists after anti-PD-1 treatment. To test if targeting CD73 would be important for a successful combination strategy in glioblastoma multiforme, we performed reverse translational studies using CD73-/- mice. We found that the absence of CD73 improved survival in a murine model of glioblastoma multiforme treated with anti-CTLA-4 and anti-PD-1. Our data identified CD73 as a specific immunotherapeutic target to improve antitumor immune responses to immune checkpoint therapy in glioblastoma multiforme and demonstrate that comprehensive human and reverse translational studies can be used for rational design of combinatorial immune checkpoint strategies.

Knowledge of immune cell phenotypes in the tumor microenvironment is essential for understanding mechanisms of cancer progression and immunotherapy response. We profiled 45,000 immune cells from eight breast carcinomas, as well as matched normal breast tissue, blood, and lymph nodes, using single-cell RNA-seq. We developed a preprocessing pipeline, SEQC, and a Bayesian clustering and normalization method, Biscuit, to address computational challenges inherent to single-cell data. Despite significant similarity between normal and tumor tissue-resident immune cells, we observed continuous phenotypic expansions specific to the tumor microenvironment. Analysis of paired single-cell RNA and T cell receptor (TCR) sequencing data from 27,000 additional T cells revealed the combinatorial impact of TCR utilization on phenotypic diversity. Our results support a model of continuous activation in T cells and do not comport with the macrophage polarization model in cancer. Our results have important implications for characterizing tumor-infiltrating immune cells.

Tumor responses to PD-1 blockade therapy are mediated by T cells, which we characterized in 102 tumor biopsies obtained from 53 patients treated with pembrolizumab, an antibody to PD-1. Biopsies were dissociated and single cell infiltrates were analyzed by multicolor flow cytometry using two computational approaches to resolve the leukocyte phenotypes at the single cell level. There was a statistically significant increase in the frequency of T cells in patients who responded to therapy. The frequency of intratumoral B cells and monocytic myeloid-derived suppressor cells (moMDSCs) significantly increased in patients’ biopsies taken on treatment. The percentage of cells with a T regulatory phenotype, monocytes, and NK cells did not change while on PD-1 blockade therapy. CD8 T memory cells were the most prominent phenotype that expanded intratumorally on therapy. However, the frequency of CD4 T effector memory cells significantly decreased on treatment, whereas CD4 T effector cells significantly increased in nonresponding tumors on therapy. In peripheral blood, an unusual population of blood cells expressing CD56 were detected in two patients with regressing melanoma. In conclusion, PD-1 blockade increases the frequency of T cells, B cells, and MDSCs in tumors, with the CD8 T effector memory subset being the major T-cell phenotype expanded in patients with a response to therapy. on November 2, 2017. © 2016 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Author Manuscript Published OnlineFirst on January 19, 2016; DOI: 10.1158/2326-6066.CIR-15-0210

Abstract Tumor responses to programmed cell death protein 1 (PD-1) blockade therapy are mediated by T cells, which we characterized in 102 tumor biopsies obtained from 53 patients treated with pembrolizumab, an antibody to PD-1. Biopsies were dissociated, and single-cell infiltrates were analyzed by multicolor flow cytometry using two computational approaches to resolve the leukocyte phenotypes at the single-cell level. There was a statistically significant increase in the frequency of T cells in patients who responded to therapy. The frequency of intratumoral B cells and monocytic myeloid-derived suppressor cells significantly increased in patients' biopsies taken on treatment. The percentage of cells with a regulatory T-cell phenotype, monocytes, and natural killer cells did not change while on PD-1 blockade therapy. CD8+ memory T cells were the most prominent phenotype that expanded intratumorally on therapy. However, the frequency of CD4+ effector memory T cells significantly decreased on treatment, whereas CD4+ effector T cells significantly increased in nonresponding tumors on therapy. In peripheral blood, an unusual population of blood cells expressing CD56 was detected in two patients with regressing melanoma. In conclusion, PD-1 blockade increases the frequency of T cells, B cells, and myeloid-derived suppressor cells in tumors, with the CD8+ effector memory T-cell subset being the major T-cell phenotype expanded in patients with a response to therapy. Cancer Immunol Res; 4(3); 194–203. ©2016 AACR.