328 Batf3 dendritic cells and 4–1BB/4–1BB ligand axis are required at the effector phase within the tumor microenvironment for anti-PD-L1 efficacy

Background

PD-1/PD-L1 blockade has shown clinical benefit across many cancer types. However, a large fraction of patients are resistant to immune checkpoint blockade therapy and others eventually relapse. Understanding the mechanisms involved in αPD1/PD-L1 immunotherapy efficacy may enable new strategies for improving clinical outcomes. Given that Batf3-lineage dendritic cells (DCs) are needed for spontaneous T cell priming in the tumor-draining lymph node and for recruitment of effector CD8+ T cells to the tumor, in the current work we examined whether Batf3+ DCs are also required during the effector phase of the anti-tumor immune response at the time of anti-PD-L1 administration for therapeutic efficacy.

Methods

We utilized the B16-SIY melanoma model, CD11c-DTR-GFP, and CD11c-DTR-GFP/Batf3 KO bone marrow chimeras to study the role Batf3+ DCs play during anti-PD-L1 immunotherapy. To focus on the effector phase of the immune response, we depleted CD11c+ cells with diphtheria toxin from day seven of tumor injection while simultaneously blocking new T cell entry with FTY720. As flow cytometry revealed high 4-1BBL expression on intratumoral Batf3-DCs, 4-1BB KO mice and anti-4-1BBL blocking antibodies were used. Tumor growth and phenotypic analysis of the tumor infiltrate were evaluated.

Results

Strikingly, we observed that CD11c+ cells, and specifically Batf3+ DCs, were required in the tumor prior to αPD-L1 treatment for immunotherapy efficacy. The normal intratumoral expansion of antigen (Ag)-specific CD8+ tumor-infiltrating T cells (TILs) and increased ratio between Ag-specific CD8+ TILs and regulatory T cells following anti-PD-L1 therapy was eliminated with Batf3+ DC depletion. Batf3+ DCs expressed high levels of 4-1BBL, and increased expression of 4-1BB on antigen-specific CD8+ TILs upon αPD-L1 treatment required Batf3+ DCs. Mechanistic experiments confirmed a requirement for 4-1BB expression on immune cells for αPD-L1 efficacy, and blocking antibodies against 4-1BBL eliminated anti-PD-L1 efficacy as well. Using appropriate bone marrow chimeras, agonistic 4-1BB antibodies were sufficient to bypass the need for CD11c+ DCs at the effector phase for tumor control. In human melanoma samples, co-localization of Batf3+ DCs and CD8+ T cells was observed in T cell-inflamed tumors, which correlated with anti-PD-1 efficacy in metastatic melanoma.

Conclusions

Our results indicate that Batf3+ DCs are necessary during the effector phase of the anti-tumor immune response for anti-PD-L1 efficacy to occur, at least in part through 4-1BB/4-1BBL-mediated reinvigoration of Ag-specific CD8+ TILs.

Ethics Approval

The study obtained ethics approval, IRB protocol 15-0837.

306 Decr2 regulates tumor cell ferroptosis and is required for immunotherapy efficacy

Background

Checkpoint blockade therapies have transformed the landscape of cancer care. Durable clinical responses have been observed in a subset of patients. However, many patients do not respond, and understanding the mechanisms underlying tumor resistance to checkpoint blockade drugs could potentially expand clinical benefit. We therefore performed an unbiased CRISPR library screen to identify novel gene alterations sufficient to mediate resistance to T cell-mediated killing.

Methods

B16.SIY cells were transduced with a genome-scale gRNA lentivirus to generate loss of function mutants. In vitro-primed CD8+ T cells isolated from 2C/Rag2–/– TCR transgenic mice specific for the SIY antigen were co-cultured with transduced B16.SIY tumor cells. Resistant cells were collected and targeted genes were identified by sequencing the gRNAs. The gene encoding Decr2, a peroxisomal 2,4-dienoyl-CoA reductase, was identified. To investigate the role of Decr2 in tumor growth and immune responses in vivo, the Decr2 knock-down or Decr2 overexpressed tumors were transplanted into C57BL/6 mice for studies of anti-PD-L1 efficacy and immune response analysis. Based on a hypothesized role in ferroptosis, RNA-seq was performed and lipid reactive oxygen species were quantified.

Results

Decr2 mutants were relatively resistant to CD8+ T cell killing in vitro. In vivo, Decr2 knock-down tumors showed almost complete loss of anti-PD-L1 efficacy, without affecting early T cell priming or immune cell infiltration. However, tumor-infiltrating T cells failed to expand following anti-PD-L1 therapy in Decr2 knock-down tumors. RNA-seq analysis revealed upregulation of ferroptosis-related genes in Decr2-deficient tumor cells, and these tumor cells were relatively resistant to ferroptosis-inducers in vitro. Decr2 knock-down cells also showed defective induction of lipid ROS in response to CD8+ T cells in vitro, or when analyzed following anti-PD-L1 treatment ex vivo. Analysis of tumor RNA-seq data from human melanoma patients treated with checkpoint blockade revealed decreased Decr2 levels in non-responding patients.

Conclusions

Our results identify tumor cell-expressed Decr2 as novel regulator of ferroptosis and implicate the process of ferroptosis as a major mechanism of immune-mediated tumor cell killing in vivo. Pharmacologic strategies to facilitate tumor cell ferroptosis could have therapeutic utility.

295 Tumor cell- intrinsic expression of FGFR3 drives anti-PDL-1 immunotherapy resistance in a murine bladder cancer model

Background

Immune checkpoint blockade therapy has recently shown efficacy in treating advanced urothelial bladder cancer and anti-PD-1/PD-L1 therapy is currently the standard of care. However, most patients do not respond to these drugs and resistance mechanisms remain elusive. The non-T cell-inflamed tumor microenvironment phenotype correlates with poor prognosis and immunotherapy resistance. We previously found that activating mutations in Fibroblast Growth Factor Receptor 3 (FGFR3) were exclusive to non-T cell-inflamed bladder cancers. We investigated the impact of tumor cell-intrinsic FGFR3 activation on T cell infiltration and tumor responsiveness to anti-PD-1/PD-L1 in a murine model.

Methods

We developed a syngeneic transplantable murine bladder cancer model using the MB49 cell line engineered to express FGFR3 with either the activating G370C mutation (FGFR3-G370C), a kinase-dead mutation K508M (FGFR3-K508M), a truncating mutation resulting in a secreted receptor (FGFR3-sec), or control. Mice were injected subcutaneously into flank and size was measured every 3 days along with PD-L1 therapy (BioXcell clone 10F.9G2). Tumors, draining lymph nodes, and spleens were harvested at endpoint for flow cytometry.

Results

Tumors from mice inoculated with MB49-FGFR3-G370C cells showed diminished CD8+ T cell accumulation within the tumor and were resistant to anti-PD-L1 checkpoint blockade compared to MB49 controls. To determine if FGFR3-mediated immune resistance was dependent on FGFR3 kinase activity, MB49-FGFR3-K508M tumors were evaluated and unexpectedly found to be resistant to anti-PD-L1 treatment. Tumors with a secreted FGFR3 receptor (FGFR3-sec) retained responsiveness to anti-PD-L1 therapy.

Conclusions

In a murine bladder cancer model, FGFR3 activation led to relative T cell exclusion and resistance to immune checkpoint blockade. The mechanism of resistance was not dependent on receptor kinase activity. Further studies are ongoing to determine the biochemical mechanisms of FGFR3-mediated immunotherapy resistance in bladder cancer.

927 Wilms tumor reveals DNA repair gene hyper-expression is linked to lack of immune infiltration

Background

A T cell-rich tumor microenvironment has been associated with improved clinical outcome and response to immune checkpoint blockade therapies in several adult cancers. Understanding the mechanisms for lack of immune cell infiltration is critical for expanding immunotherapy efficacy in the clinic. To gain new insights into the mechanisms of poor tumor immunogenicity, we turned to pediatric cancers, which are generally unresponsive to checkpoint blockade.

Methods

RNAseq and clinical data were obtained for Wilms tumor, rhabdoid tumor, osteosarcoma, and neuroblastoma from the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) database, and adult cancers from TCGA. Using an 18-gene tumor inflammation signature (TIS) representing activated CD8+ T cells, we identified genes significantly anti-correlated with the signature. Immunofluorescence was performed on metastatic melanoma samples for CD8, MSH2, and the tumor cell marker SOX10, and analyzed for relationship to anti-PD-1 efficacy.

Results

Among the four pediatric cancers, we observed the lowest TIS scores in Wilms tumor. Wilms tumors demonstrated significantly lower T cell inflammation signatures than matched normal kidney samples, other pediatric tumor samples, and adult kidney tumor samples. Pathway analysis identified multiple types of DNA repair were upregulated in Wilms tumor and a score generated from the top 50 DNA repair genes strongly anti-correlated with TIS. This striking negative association was also observed in most adult tumor types. The anti-correlation was found to be independent of tumor mutation burden, suggesting that high expression of DNA repair pathway machinery may restrict tumor immunogenicity by mechanisms beyond prevention neoantigen accumulation. MSH2 was one of the top DNA repair genes identified from the Wilms tumor analysis and was confirmed to have a strong anti-correlation with TIS in melanoma samples from TCGA. Immunofluorescence from an independent cohort of metastatic melanoma patients revealed a significant negative correlation between CD8+ T cell numbers and MSH2+ SOX10+ tumor cell numbers. Additionally, non-responders to anti-PD-1 immunotherapy had significantly higher numbers of MSH2+ SOX10+ tumor cells than responders.

Conclusions

Increased tumor expression of DNA repair genes is associated with a less robust immune response in Wilms tumor, and this was also observed in the majority of TCGA tumor types. Surprisingly, the negative relationship between DNA repair score and TIS remained strong across TCGA when correcting for mutation count, indicating a potential role for DNA repair genes outside of preventing the accumulation of mutations. Strategies targeting DNA repair pathways could be considered as new therapeutic interventions to transform non-T cell-inflamed tumors into immune-responsive tumors.

Ethics Approval

The study obtained ethics approval under IRB protocol 15-0837.

918 Doxycycline regulatable ß-catenin demonstrates inducible immune evasion in a melanoma GEM model

Background

Lack of response to checkpoint blockade immunotherapy has been linked to a deficiency of immune cell infiltration within the tumor microenvironment (TME). One demonstrated mechanism sufficient for the non-T cell inflamed TME is tumor cell-intrinsic activation of the β-catenin signaling pathway. Using genetically engineered mouse models (GEMMs), tumors constitutively expressing active β-catenin lack a robust endogenous T cell infiltrate and fail to respond to immunotherapies. In support of these mouse studies, human melanoma metastases with increased active β-catenin signaling exhibit decreased numbers of tumor infiltrating Batf3-driven cDC1 and CD3+ T cells. However, whether temporal activation and inactivation of β-catenin within the same developing tumor would alter immune cell infiltration is not known.

Methods

A model was created in which tamoxifen-regulated Cre-recombinase mediates BRAFV600E oncogene activation and PTEN tumor suppressor gene deletion as well as expression of a doxycycline regulatable reverse transactivator. Upon administration of doxycycline via the drinking water to these animals, a non-degradable form of nuclear β-catenin becomes expressed. Immunofluorescence assays were performed assessing the β-catenin expression status in the tumor cells as well as immune cell infiltration within the TME. Additionally, immunotherapy efficacy experiments were performed.

Results

We observed that administration of doxycycline to these animals drove expression of an active form of nuclear β-catenin. Activation of nuclear β-catenin resulted in a 2-fold decrease in the overall CD3+ T cells infiltration into the TME. Moreover, this decrease in immune infiltration also resulted in loss of anti-PD-L1 + anti-CTLA-4 therapy efficacy. We next performed studies assessing the kinetics with which β-catenin levels diminish upon doxycycline removal. Switching animals to regular drinking water resulted in rapid reduction of nuclear β-catenin levels, including 50 percent reduction after two days of doxycycline removal and almost complete reduction of nuclear β-catenin after four days.

Conclusions

We describe a novel mouse model in which we induce autochthonous melanoma tumors in mice along with inducible expression of a non-degradable, nuclear β-catenin modulated by doxycycline in the drinking water. Activation of β-catenin signaling in melanoma tumors resulted in reduction of immune cells in the TME as well as loss of checkpoint blockade immunotherapy efficacy. This activation can be rapidly reversed by removing doxycycline, allowing for future studies evaluating the consequences of turning off β-catenin once it has already driven a non-T cell-inflamed TME.

Acknowledgements

This work was supported by the Wissler Fellowship from the University of Chicago (SS) K99/R00 (NCI; SS), and R35CA210098 (TG).

642 Decreased host PKC-delta is associated with the T cell-inflamed tumor microenvironment and improves anti-tumor immunity

Background

Favorable clinical responses to immunotherapy have been correlated with a T cell-inflamed tumor microenvironment. The degree of spontaneous immune infiltration in tumors varies widely between individual patients. We hypothesized that germline polymorphisms in immune regulatory genes may affect the host immune response to solid tumors, similar to their influence on autoimmune susceptibility.

Methods

Melanoma TCGA RNAseq and germline SNP data were utilized to identify germline polymorphisms associated with the magnitude of an immune gene signature score. The top GWAS hit associated with increased immune gene expression in melanoma was SNP rs1483185 (p = 8.812e-08, Bonferroni corrected <0.05), within the PKC-delta gene. Using a lymphoblastic cell line GTEX database, this SNP was associated with lower expression PKC-delta, implying a loss of function phenotype. Germline mutations in PKC-delta had previously been associated with familial lupus. To study the role of PKC-delta in anti-tumor immunity, knockout hematopoietic cells and conditional knockout mice were utilized, and implanted tumors were monitored along with detailed immune response analysis.

Results

B16.SIY tumors grew more slowly in chimeras reconstituted with PKC-delta-/- bone marrow compared to WT bone marrow, and this effect was dependent on CD8-beta+ cells. T cell priming in the tumor-draining lymph node was comparable in WT and KO hosts. However, tumors in PKC-delta-/- bone marrow chimeras had increased numbers of CD8+ T cells in the tumor at endpoint, and also responded better to anti-PD-L1 therapy. Single cell RNAseq of the tumor microenvironment revealed that PKC-delta loss primarily altered gene expression in myeloid cell subsets, leading to to increased expression of M1 associated genes and decreased expression of M2 associated genes in PKC-delta-/- chimeras. To follow up further, a conditional PKC-delta KO mouse was developed and crossed to the hematopoietic Vav1-iCre and also to LysM-Cre transgenic mice. In both instances, immune-mediated tumor control was improved, demonstrating that loss of PKC-delta in the myeloid compartment is sufficient to recapitulate the phenotype.

Conclusions

Our results demonstrate that germline variants in immune regulatory genes can profoundly affect anti-tumor immunity and the efficacy of PD-1/PD-L1 blockade. In particular, the myeloid-expressed molecule PKC-delta plays an important regulatory role such that decreased expression/activity mediates improved anti-tumor immunity by altering the M1/M2 ratio. The development of pharmacologic approaches to phenocopy this loss of function phenotype may be attractive to pursue as a novel therapeutic strategy.

Ethics Approval

University of Chicago IRB Protocol 15-0837

322 Melanoma-intrinsic hypoxia-inducible factor-1α results in diminished T cell accumulation within the tumor microenvironment

Background

The hypoxia-inducible factor (HIF) system, consisting of the transcription factors HIF-1α and HIF-2α, mediates cellular adaptation to hypoxia, and can promote cancer progression, invasion, and metastasis. HIF pathway activation in the tumor microenvironment has been implicated in cancer immune evasion; however, a direct causal role for tumor cell-intrinsic HIF-1α and HIF-2α activation in mediating T cell exclusion and cancer cell resistance to immune checkpoint inhibitor therapy has not been demonstrated.

Methods

We performed gene expression analysis of melanoma tumors in the Cancer Genome Atlas (TCGA) data set to determine whether increased HIF-1α pathway activation correlated with reduced T cell-based inflammation. The magnitude of HIF-1α pathway activation across melanoma samples was determined by applying a quantitative scoring system on the expression of a melanocyte-specific hypoxia-induced, HIF-1α-target gene signature consisting of 81 genes. The Pearson correlation test was used to compare the HIF-1α activation score and our 160-gene T-cell-inflamed gene signature. To determine the impact of cancer cell-intrinsic HIF-1α or HIF-2α activation on the endogenous anti-tumor T cell response, we developed an inducible autochthonous mouse melanoma model driven by BRAFV600E expression and PTEN-deletion, with or without inducible expression of either a stabilized variant of HIF-1α or HIF-2α. These murine tumor models are being used to determine the impact of cancer cell-intrinsic HIF-1α or HIF-2α activation on tumor sensitivity to anti-PD-1/PD-L1 and anti-CTLA-4 treatment.

Results

Gene expression analysis of human melanomas in the TCGA demonstrated a statistically significant inverse correlation between the HIF-1α activation score and T cell-inflammation score. Braf/PTEN murine melanomas with and without stabilized HIF-1α expression developed with comparable tumor onset and growth kinetics. Multiparameter immunofluorescence staining of melanoma tissue revealed a significant decrease in tumor-infiltrating T cells within Braf/PTEN melanoma tumors expressing stabilized HIF-1α compared to control Braf/PTEN melanomas.

Conclusions

Our data demonstrate that tumor-cell intrinsic HIF-1α activation leads to diminished T cell accumulation within the tumor microenvironment, which has implications for cancer immunotherapy. The mechanism of this effect is being elucidated. These novel murine models will help elucidate the roles of cancer cell-intrinsic HIF-1α and HIF-2α activation in modulating the anti-tumor T cell response, providing mechanistic insight that will inform the evaluation of novel selective HIF inhibitors, which are showing promising anti-tumor activity in clinical trials in patients with advanced solid tumors.

The Impact of Diabetes and African American Race in COVID-19 Severity

Introduction: COVID-19, an illness caused by the novel coronavirus usually presents as a mild to moderate flu like illness, but can lead to pneumonia, acute respiratory distress syndrome and death in some patients. Since the onset of the COVID-19 pandemic there has been special attention towards patients with diabetes. Not only is diabetes highly prevalent in patients with COVID-19, but diabetes has been reported as a significant predictor of morbidity and mortality. Furthermore, race seems to be a predictor of severity with black people dying at 2.3 times the rate of white people. Methods: Here we performed a retrospective study of 524 cases of COVID-19 at an academic center in Chicago between March 2020 until August 2020. Data were obtained from patients that consented to the study to examine the relationship between diabetes, BMI, age, and race with degree of COVID-19 severity. Not all patients had all clinical and demographic data available. COVID-19 severity was determined using a severity index obtained from the measured SpO2 divided by the FiO2/fraction of inspired oxygen times 100. Numbers ≥ 315 mmHg were defined as low severity with patients generally requiring outpatient care, while 235–314 mm Hg were classified as moderate severity generally requiring inpatient care and≤ 234 mm Hg indicated high severity generally requiring intubation/ICU care. The Pearson correlation coefficient was used for linear correlation analyses. Proportion for categorical values were compared using the Chi squared test, the means for continuous variables were compared using two-tailed t tests or one way ANOVA (with Tukey post-test) for comparisons involving more than two conditions. A multiple linear regression model was used to assess the contribution of different variables. Differences were considered statistically significant at p<0.05 Results: Among 120 patients with an A1c, 55 (46%) patients had diabetes and 65 (54%) did not have diabetes. More patients with a high severity index were seen in the cohort with diabetes compared to those without diabetes (72% compared to 28% p=0.004). Univariate analyses revealed statistically significant positive correlations with higher COVID-19 severity and older age, BMI, and African American race. ANOVA analysis revealed a statistically significant difference between increasing BMI and worse severity category with a BMI mean of 29.3 kg/m² in the low severity category compared to 34.9 kg/m² in the moderate severity category (p=0.006). A multi-variate analysis adjusting for all variables revealed that A1c, older age and race were positively associated with higher COVID-19 severity. Conclusion: Increased A1c, older age and race are positively and independently associated with a higher COVID-19 severity index. Further research regarding the relationship between COVID-19 and these associations is urgently needed.

Immunogenomic determinants of tumor microenvironment correlate with survival in high-risk neuroblastoma

Background:

Tumor-infiltrating CD8+ T cells and neoantigens are predictors of a favorable prognosis and response to immunotherapy in adult cancer, but little is known about their role in pediatric malignancies. Here, we analyzed the prognostic strength of T cell-inflamed (Tinfl) expression and neoantigen load (neo) in high-risk neuroblastoma (NBL). We investigated transcriptional programs associated with the non-T cell-inflamed (non-Tinfl) phenotype for mechanisms of immune exclusion.

Methods:

Using a defined Tinfl gene expression signature, we categorized high-risk NBL from the TARGET (Therapeutically Applicable Research to Generate Effective Treatments, n=123) and the GMKF (Gabriella Miller Kids First, n=48) programs into non-Tinfl, Tinfl, and intermediate groups. Cox proportional hazard models were used to test the association between survival and tumor group, neo, MYCN amplification. Pathways activated in non-Tinfl vs. Tinfl tumors were identified.

Results:

Patients with high-risk Tinfl tumors showed improved overall survival (OS) compared to those with non-Tinfl tumors (P&lt;0.05), independent of MYCN amplification, in both TARGET and GMKF. Higher neo was associated with better event-free survival and OS (P&lt;0.005) and was independent of Tinfl. Activation of MYCN, ASCL1, SOX11, and KMT2A was inversely correlated with the Tinfl signature in both cohorts.

Conclusions:

Our results indicate that tumors from children with high-risk NBL harboring a strong Tinfl signature have a favorable clinical outcome, and neo is a prognosis predictor, independent of Tinfl. Strategies to target SOX11 and other signaling pathways associated with non-Tinfl tumors should be pursued as potential immune-potentiating interventions.

Significant antitumor activity for low-dose ipilimumab (IPI) with pembrolizumab (PEMBRO) immediately following progression on PD1 Ab in melanoma (MEL) in a phase II trial

10004

Background: Combination PD1 + CTLA4 antibodies (Abs) shows greater response rate (RR) versus PD1 Ab alone in MEL, but RR after initial PD1 Ab progression awaits robust investigation. CTLA4 Ab alone after PD1 Ab progression has a historical RR of 13%. We report final results of the first prospective clinical trial evaluating IPI 1mg/kg + PEMBRO immediately following progression on PD1 Ab (NCT02743819). Methods: Patients (pts) with advanced MEL, no prior CTLA4 Ab for metastatic disease, and who had progressed on PD1 Ab as immediately prior therapy (or non-CTLA4 Ab combination) were eligible. Pts received PEMBRO 200 mg + IPI 1 mg/kg Q3W for 4 doses, then PEMBRO alone for up to two years. The primary endpoint was RR by irRECIST. After 35 pts, the study met its primary endpoint with 10/22 evaluable pts achieving a response. The trial was expanded to enroll a total of 70 pts in open-label accrual to further describe the RR for this regimen in an exploratory fashion. The data analysis cutoff was January 30, 2020. Results: 67/70 accrued patients were evaluable for treatment response. Prior treatments included 60 on PD1 Ab alone and 10 on PD1 Ab-based combinations. Of these, 10 pts had progressed in the adjuvant setting. Median length of treatment on prior PD1 Ab was 4.8 months. Response assessments included 4 CR, 17 PR and 16 SD for a RR of 31% (21/67) in evaluable pts, and 30% (21/70) in all enrolled pts. 4 pts with a PR and 6 with SD had unconfirmed responses making the irRECIST response rate 25% (17/67) and 24% (17/70) among evaluable and enrolled pts, respectively. Median progression free survival (PFS) was 4.7 mo (95% CI: 2.8-8.3) and PFS at six months was 45% (95% CI: 33%-57%). 15/70 (21%) pts experienced ≥ grade 3-4 drug-related AEs, the most common being diarrhea, rash and transaminase elevation. PD-L1 positive vs negative status from historical tumor specimens did not associate with RR. Conclusions: This is the largest prospective study of IPI 1mg/kg + PEMBRO, demonstrating significant antitumor activity and tolerability in MEL post-PD1 Ab. Clinical trial information: NCT02743819.

Assessing the effect of immunosuppressive agents for immune-related adverse event management on tumor response

3066

Background: High grade immune-related adverse events (irAEs) to cancer immune checkpoint inhibitors (ICI) require considerable immunosuppression (IS) with high-dose steroids and steroid-sparing IS (SSIS) for steroid-dependent cases. T lymphocyte-specific IS has generally been avoided or used with significant caution due to the fear that these agents may negatively impact ICI efficacy. We sought to determine whether T cell-specific IS agents, such as calcineurin inhibitors (CNIs), have an adverse effect on tumor control when compared to other immunomodulatory drugs (IMDs). Methods: We retrospectively analyzed clinical annotations of adult patients treated with ICIs for malignancy from 1/1/2000-12/31/2019, highlighting patients who were managed with SSIS, specifically those most commonly used for autoimmune disease therapy. Topical IS use was excluded. Patients were categorized as tumor responders or non-responders, and irAEs were graded according to National Cancer Institute’s Common Terminology Criteria for Adverse Events (CTCAE). Progression-free survival (PFS) was assessed via Kaplan-Meier curve. Results: 1331 unique individuals were prescribed ≥1 ICIs, with 526 prescribed systemic steroids (39.5%) and 90 (6.8%) patients prescribed SSIS agents, 25 patients with >1 SSIS: mycophenolate (39), methotrexate (26), leflunomide (5), azathioprine (3), rituximab (24), tocilizumab (3), infliximab (8), etanercept (1), adalimumab (1), golimumab (1) and CNIs (18): cyclosporine, tacrolimus. IMDs hydroxychloroquine (6) and sulfasalazine (5) were also prescribed. The objective response rate was 50.0% in the CNI group compared to 45.5% in the IMD cohort and 45.4% in the irAE group (CTCAE grade matched) with steroids alone without any SSIS. Median PFS were compared between CNI cohort (5.4 months, range 1.3-34 months) to IMD (1.1 months, range 0.4-6.4, p=0.02) and steroid alone (2.4 months, range 0.69-17.7, p=0.48). Multiple regression analysis identified irAE presence as an independent correlates to tumor response (p=0.02). Conclusions: T cell-specific IS should not be excluded from irAE treatment algorithm as we observed that PFS was comparable to immunomodulators and similar efficacy was observed compared to steroids alone. Rapid identification and management of irAEs can help mitigate morbidity but there are virtually no reliable clinical trials to guide irAE management with SSIS. These findings support the need for larger, prospective evaluation of immunosuppression use for high grade irAE therapy.

Enhancing IFN-β production through synergy between the STING pathway and TLR pathways

Previous work from our lab has shown that tumors activate the STING pathway in antigen presenting cells. STING signaling leads to IFN-β production which is required for T cell priming and recruitment. IFN-β transcription is induced by multiple transcription factors including NF-κB and IRF3. Enhancing STING activation and downstream IFN-β production has therapeutic potential, and STING agonists are currently being evaluated in clinical trials. However, innate immune activation in response to a pathogen rarely occurs by activating only one signaling pathway, and activating multiple innate immune pathways similar to a natural infection could improve the efficacy of STING agonists. To test this, we performed experiments with the STING agonist DMXAA alone or in combination with several TLR agonists. We found that LPS + DMXAA induced significantly greater IFN-β transcription than the sum of either agonist alone. To explain this synergy, we assayed each step of STING pathway signaling. LPS did not increase STING aggregation beyond that induced by DMXAA. Additionally, LPS did not increase IRF3 phosphorylation nor IRF3 nuclear translocation in combination with DMXAA. LPS did increase the phosphorylation and nuclear translocation of the NF-κB subunit p65. These results suggest that the synergy in IFN-β production is achieved through IRF3 activation downstream of STING signaling and NF-κB activation downstream of TLR4 signaling. Intratumoral injection of LPS + DMXAA resulted in significantly improved tumor control of B16 melanoma in vivo than either agonist alone, suggesting that combining activation of these pathways improves the anti-tumor innate immune response in a way that is therapeutically beneficial.

Investigating microbiome-based anti-tumor immune modulation using patient-derived mouse models

Growing data supports that the gut microbiota can impact cancer immunotherapy. We previously found certain bacterial taxa associated with response to immune checkpoint blockade. These correlations warrant deeper mechanistic studies using tractable mouse models to establish the role of the microbiota in determining immunotherapy efficacy.

To understand how the microbiota affects anti-tumor immunity we generated patient-based microbiota models. Germ-free mice were colonized using patient feces, bred, and subsequent generations studied using transplantable tumor models. We focused on two representative patient-based models: one from a responder (R) to immunotherapy and one from a non-responder (NR).

Mice established from the R demonstrated greater benefit from anti-PD-L1 therapy compared to NR colonized mice and had superior CD8+ T cell responses with greater accumulation of tumor-specific CD8+ T cells and increased proliferation. In addition, compared to NR-derived mice, R-derived mice had increased intra-tumoral CD103+ dendritic cells (DCs), as well as increased BATF3-lineage DCs in the mesenteric lymph nodes and spleen. In agreement with a stronger anti-tumor immune response, R mice had higher circulating levels of IFN-γ, IL-12p70, and CXCL5 compared to NR mice.

To identify potential systemic signals originating from the gut microbiome and capable of altering global immune populations we profiled serum using an unbiased metabolomics-based approach. We found distinct metabolite profiles in the R- and NR-derived mice, suggesting potential bacteria-derived or -modified metabolite modulating immune function. These data support a role for the gut microbiome determining host sensitivity to cancer immunotherapy.

ACCELERATE and European Medicines Agency Paediatric Strategy Forum for medicinal product development of checkpoint inhibitors for use in combination therapy in paediatric patients

The third multistakeholder Paediatric Strategy Forum organised by ACCELERATE and the European Medicines Agency focused on immune checkpoint inhibitors for use in combination therapy in children and adolescents. As immune checkpoint inhibitors, both as monotherapy and in combinations have shown impressive success in some adult malignancies and early phase trials in children of single agent checkpoint inhibitors have now been completed, it seemed an appropriate time to consider opportunities for paediatric studies of checkpoint inhibitors used in combination. Among paediatric patients, early clinical studies of checkpoint inhibitors used as monotherapy have demonstrated a high rate of activity, including complete responses, in Hodgkin lymphoma and hypermutant paediatric tumours. Activity has been very limited, however, in more common malignancies of childhood and adolescence. Furthermore, apart from tumour mutational burden, no other predictive biomarker for monotherapy activity in paediatric tumours has been identified. Based on these observations, there is collective agreement that there is no scientific rationale for children to be enrolled in new monotherapy trials of additional checkpoint inhibitors with the same mechanism of action of agents already studied (e.g. anti-PD1, anti-PDL1 anti-CTLA-4) unless additional scientific knowledge supporting a different approach becomes available. This shared perspective, based on scientific evidence and supported by paediatric oncology cooperative groups, should inform companies on whether a paediatric development plan is justified. This could then be proposed to regulators through the available regulatory tools. Generally, an academic-industry consensus on the scientific merits of a proposal before submission of a paediatric investigational plan would be of great benefit to determine which studies have the highest probability of generating new insights. There is already a rationale for the evaluation of combinations of checkpoint inhibitors with other agents in paediatric Hodgkin lymphoma and hypermutated tumours in view of the activity shown as single agents. In paediatric tumours where no single agent activity has been observed in multiple clinical trials of anti-PD1, anti-PDL1 and anti-CTLA-4 agents as monotherapy, combinations of checkpoint inhibitors with other treatment modalities should be explored when a scientific rationale indicates that they could be efficacious in paediatric cancers and not because these combinations are being evaluated in adults. Immunotherapy in the form of engineered proteins (e.g. monoclonal antibodies and T cell engaging agents) and cellular products (e.g. CAR T cells) has great therapeutic potential for benefit in paediatric cancer. The major challenge for developing checkpoint inhibitors for paediatric cancers is the lack of neoantigens (based on mutations) and corresponding antigen-specific T cells. Progress critically depends on understanding the immune macroenvironment and microenvironment and the ability of the adaptive immune system to recognise paediatric cancers in the absence of high neoantigen burden. Future clinical studies of checkpoint inhibitors in children need to build upon strong biological hypotheses that take into account the distinctive immunobiology of childhood cancers in comparison to that of checkpoint inhibitor responsive adult cancers.

Abstract 2836: The commensal urinary microbiome as a predictor of response to Bacillus Calmette-Guerin (BCG) immunotherapy in non-muscle invasive bladder cancer

Bacillus Calmette-Guerin (BCG) is an intravesical immunotherapy that is standard of care to prevent recurrence of non-muscle invasive bladder cancer. Even with adequate BCG treatment, recurrences occur in up to 50% of patients within 5 years. The identification of a commensal urinary microbiome has been recently reported, but it has not been studied in the context of bladder cancer treatment. We hypothesized that variation in the urine microbiome is associated with response to BCG immunotherapy. A clinical study was initiated that enrolled patients with a newly diagnosed bladder tumor to evaluate urinary biomarkers. Urine samples were collected via catheterization to minimize contamination at baseline and up to 8 additional time points. Samples were centrifuged and DNA was extracted. The microbial composition for each sample determined by the 16S rRNA gene sequencing-based method and analyzed by Operational Taxonomic Units (OTUs). Interferon gamma and interleukin-2 levels were measured using the Invitrogen ProcartaPlex immunoassay. Thirty-one patients were enrolled with a median age of 69 years, including 9 females and 22 males. All patients were treated with BCG and 10 patients were diagnosed with recurrent disease at a median follow up of 6 months. In baseline samples, a significant difference in microbial composition was observed by distance matrix computation between patients with and without recurrence (Bonferroni-corrected P=0.017). Actinobacteria, Bacteroidetes, Firmicutes, Proteobacteria, and Tenericutes accounted for &gt;99% of the phyla detected across all samples. Proteobacteria had a significantly higher abundance in patients who developed recurrence (P=0.035). In particular, the Enterobacteriacae family was significantly more abundant among patients with recurrence (P=0.001). Lactobacillales was lower in abundance in patients with recurrence versus those without recurrence (P=0.049). A subset of 13 patients, including 6 with recurrence, were analyzed for changes in urine gamma interferon and interleukin-2 levels after BCG induction (week 6), and no difference was observed between patients with and without recurrence (P=0.79 and P=0.96, respectively). Characterizing the commensal urine microbiome in bladder cancer patients is feasible, and variation in composition may predict response to BCG therapy. Further studies are ongoing to validate these findings and determine longitudinal patterns over time.

Citation Format: Randy F. Sweis, Shay Golan, Nimrod Barashi, Elle Hill, Ciro Andolfi, Ryan Werntz, Jeffrey Bloodworth, Thomas Gajewski, Gary D. Steinberg. The commensal urinary microbiome as a predictor of response to Bacillus Calmette-Guerin (BCG) immunotherapy in non-muscle invasive bladder cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2836.

Safety and clinical activity of pembrolizumab immunotherapy and multi-organ site ablative stereotactic body radiotherapy (iMOSART) in patients with advanced solid tumors

20

Background: Clinical responses to PD1 blockade have been linked to IFN-γ associated gene expression in the tumor microenvironment. Stereotactic body radiotherapy (SBRT) can stimulate innate and adaptive immunity to potentially augment immunotherapy. Anti-PD1 treatment outcomes are improved with lower disease burden and multi-site radiation is an emerging paradigm for eradicating metastatic disease. Methods: We conducted a phase I study for patients with metastatic solid tumors who progressed on standard treatment. The primary aim was to evaluate the safety of pembrolizumab Immunotherapy with Multi-Organ Site Ablative stereotactic body Radiation Therapy (iMOSART). Two to four metastases were chosen for each patient and not all sites of disease were targeted with SBRT. Metastases > 65cc were partially irradiated. Pembrolizumab was initiated within 7 days after the final SBRT treatment. Results: 79 patients were enrolled, three patients did not receive any treatment and three patients only received SBRT. The 73 patients included in the analysis were treated with SBRT and at least 1 cycle of pembrolizumab. 94.5% of patients had two lesions treated with SBRT. Median follow-up for toxicity was 5.5 months (IQR 3.3-8.1). Six patients experienced dose-limiting toxicity (DLT) with no radiation dose reductions. RECIST overall objective response rate was 13.5% in the 68 patients with imaging follow-up. Superior control of irradiated lesions was observed compared to non-irradiated target lesions for the 52 patients with paired data (p = 0·0005) with a mean percent tumor burden change -21·7% (SD 24·3%) for irradiated lesions vs. 1·7% (SD 46·3%) for non-irradiated lesions (p = 0·0008). While abscopal response defined by 30% reduction in any single non-irradiated measurable lesion was present in 26·9% of patients, abscopal response was 13·2% when defined by 30% reduction in aggregate diameter of non-irradiated measurable lesions. Conclusions: iMOSART was well tolerated with acceptable toxicity. Further randomized studies are warranted to investigate the potential clinical benefit of this combination approach. Clinical trial information: NCT02608385.

Abstract 2618: Agonist antibodies targeting OX40 and GITR enhance the activity of the IDO1-selective inhibitor epacadostat in preclinical models

The majority of immunotherapeutic agents developed thus far either attempt to stimulate a more productive anti-tumor immune response or to inhibit key proteins in the immunosuppressive tumor milieu. PD-1/PD-L1 axis blockade, CTLA-4 blockade and IDO1 inhibition are examples of the latter approach and have been utilized to reverse the suppressive tumor microenvironment, resulting in clinical benefit for cancer patients. Recent clinical and preclinical data have also demonstrated that combining these approaches results in enhanced therapeutic benefit. Notably, the IDO1-selective inhibitor epacadostat has been shown to increase the efficacy of two checkpoint inhibitors, the anti-CTLA-4 antibody ipilimumab and the anti-PD-1 antibody pembrolizumab, in patients with melanoma. Because both checkpoint receptors and IDO1 serve as negative regulators of the immune response, we also explored the ability of IDO1 inhibition to combine with agents that directly activate T cells through costimulatory receptors of the tumor necrosis factor receptor (TNFR) superfamily. Rodent active surrogate agonist antibodies to 4-1BB, OX40 and GITR were tested with epacadostat in multiple preclinical models. In the B16-SIY melanoma model that does not express IDO1 in tumor cells, both epacadostat and anti-OX40 had little effect, but the combination resulted in enhanced efficacy. This was associated with increased infiltrates of CD8+ T cells and decreased numbers of FoxP3+ TILs. Increased numbers of SIY-reactive T cells were found in both the tumor and the TDLN post-therapy. In contrast, epacadostat did not provide any enhancement to the activity seen with 4-1BB. Clear combinatorial effects were seen with anti-GITR and epacadostat in the more inflamed, IDO1-expressing PAN02 pancreatic cancer model. These data suggest that IDO1 inhibition can be effective in combination with agents that agonize T cell costimulatory receptors as well as with agents that block coinhibitory receptors.

Citation Format: Holly K. Koblish, Brendan Horton, Michael Hansbury, Sybil O'Connor, Kerri Lasky, Christina Stevens, Thomas Condamine, Leslie Hall, Liang-Chuan Wang, Yue Zhang, Horacio Nastri, Gregory Hollis, Reid Huber, Thomas Gajewski, Peggy Scherle. Agonist antibodies targeting OX40 and GITR enhance the activity of the IDO1-selective inhibitor epacadostat in preclinical models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 2618. doi:10.1158/1538-7445.AM2017-2618

Safety of epacadostat 100 mg bid plus pembrolizumab 200 mg Q3W in advanced solid tumors: Phase 2 data from ECHO-202/KEYNOTE-037

3012

Background: The immunosuppressive enzyme indoleamine 2, 3-dioxygenase 1 (IDO1) facilitates immune tolerance in cancer via T-cell suppression, and IDO1 overexpression is associated with poor survival. Epacadostat, an oral inhibitor of IDO1, has been shown to be well tolerated as monotherapy and in combination with checkpoint inhibitors. ECHO-202/KEYNOTE-037 is a phase 1/2 study evaluating the safety and efficacy of oral epacadostat plus IV pembrolizumab in patients (pts) with advanced tumors. Based on phase 1 outcomes, epacadostat 100 mg BID plus pembrolizumab 200 mg Q3W was selected for phase 2 evaluation. This analysis summarizes phase 2 safety experience in the overall population of ECHO-202/KEYNOTE-037 (pooled across tumor types) at an October 29, 2016 data cutoff. Methods: Phase 2 pts were ≥18 years of age with advanced or recurrent melanoma (MEL), non–small cell lung cancer (NSCLC), renal cell carcinoma (RCC), urothelial carcinoma (UC), triple-negative breast cancer, squamous cell carcinoma of head and neck (SCCHN), ovarian cancer, diffuse large B-cell lymphoma, or microsatellite instability–high colorectal cancer. Results: The overall safety population comprised 244 pts receiving ≥1 study treatment dose. Median age was 63 years, 52% were women, and 91% were white. As of data cutoff, 134 study pts (55%) discontinued study treatment, primarily due to disease progression (n = 97). Median exposure to study treatment was 86 days (range, 1–374 days). TRAEs occurring in ≥5% of pts were fatigue (23%); rash (16%); diarrhea and nausea (7% each); increased alanine aminotransferase, increased aspartate aminotransferase, and pruritus (6% each); and pyrexia (5%). A total of 37 pts (15%) had grade ≥3 TRAEs; the most common grade ≥3 TRAEs were increased lipase (asymptomatic) and rash (3% each). TRAEs led to discontinuation in 3% of pts. Conclusions: Epacadostat 100 mg BID plus pembrolizumab 200 mg Q3W was associated with an acceptable safety profile in pts with advanced cancers, supporting continued evaluation of the combination. The phase 3 ECHO-301/KEYNOTE-252 MEL study is ongoing and additional phase 3 studies (NSCLC, UC, RCC, SCCHN) are planned. Clinical trial information: NCT02178722.

Initial efficacy of anti-lymphocyte activation gene-3 (anti–LAG-3; BMS-986016) in combination with nivolumab (nivo) in pts with melanoma (MEL) previously treated with anti–PD-1/PD-L1 therapy

9520

Background: Signaling via LAG-3 and other T-cell inhibitory receptors (eg, PD-1) can lead to T-cell dysfunction and tumor immune escape. Simultaneous blockade of LAG-3 + PD-1 may synergistically restore T-cell activation and enhance antitumor immunity. In a phase 1/2a study, BMS-986016 (IgG4 mAb targeting LAG-3) ± nivo (IgG4 mAb targeting PD-1) demonstrated tolerability, peripheral T-cell activation, and preliminary clinical activity (NCT01968109; Lipson E, et al. J Immunother Cancer. 2016;4[s1]:173 [P232]). Here we describe preliminary efficacy of BMS-986016 + nivo in pts with MEL whose disease progressed on/after prior anti–PD-1/PD-L1 therapy, along with updated safety from all dose expansion pts. Methods: Pts with MEL must have had prior anti–PD-1/PD-L1 (± anti–CTLA-4 or BRAF/MEK inhibitors) and progressive disease (PD). Pts received BMS-986016 80 mg + nivo 240 mg IV Q2W. Primary objectives were safety and objective response rate (ORR; complete [CR] + partial [PR] response), disease control rate (DCR; CR + uCR + PR + uPR + stable disease [SD] > 12 wk), and duration of response (RECIST v1.1). Results: At data cutoff, 43 pts with MEL had been treated with BMS-986016 + nivo following PD on/after prior anti–PD-1/PD-L1 with known prior best responses of 1 CR, 9 PR, 12 SD, and 16 PD. Of the 43 pts, 30 (70%) also had prior anti–CTLA-4, 20 (47%) had ≥ 3 prior therapies, and 15 (35%) had BRAFmutations .In the 31 efficacy-evaluable pts to date, ORR was 16% (confirmed/unconfirmed) and DCR was 45% with benefit observed even in some pts refractory to prior anti–PD-1. Evaluations are ongoing for most pts, with median treatment duration of 10 wk for all 43 pts. Immunopathologic (eg, PD-1/PD-L1 and LAG-3 expression) and clinical characteristics of responders vs nonresponders will be presented. Any grade and grade 3/4 treatment-related AEs occurred in 46% and 9%, respectively, across all dose expansion pts (n = 129). Conclusion: Addition of BMS-986016 to nivolumab demonstrates encouraging initial efficacy in pts with MEL whose disease progressed on/after prior anti–PD-1/PD-L1 therapy, and a safety profile similar to nivolumab monotherapy. Clinical trial information: NCT01968109.

Epacadostat plus pembrolizumab in patients with advanced urothelial carcinoma: Preliminary phase I/II results of ECHO-202/KEYNOTE-037

4503

Background: Pembrolizumab (P), a PD-1 inhibitor, is active and well tolerated in platinum-treated, advanced urothelial carcinoma (UC). Epacadostat (E) potently and selectively inhibits indoleamine 2,3-dioxygenase 1 (IDO1), a tryptophan-catabolizing enzyme that suppresses T-cell–mediated immune surveillance. IDO1 overexpression is associated with tumor progression and shortened patient (pt) survival. ECHO-202/KEYNOTE-037 is an open-label, phase I/II study of E + P in pts with advanced tumors. We report phase I/II efficacy and safety outcomes for the UC cohort at an October 29, 2016 data cutoff. Methods: Adult pts with advanced UC, prior platinum therapy (adjuvant or advanced disease setting) or alternative therapy (if platinum was not appropriate), and no prior checkpoint inhibitor therapy were eligible to participate. In phase I, pts received E (25, 50, 100, or 300 mg PO BID) + P (2 mg/kg or 200 mg IV Q3W); MTD was not exceeded. E (100 mg BID) + P (200 mg Q3W) dosing was selected for phase II. Response was assessed in RECIST 1.1–evaluable pts. Safety was assessed in pts receiving ≥1 E + P dose. Results: A total of 40 pts (phase I, n = 5; phase II, n = 35) were evaluated. Median age was 67 years, 75% were men, 88% were white, 100% had prior platinum therapy, and 75% had 0–1 prior line of therapy for advanced disease. Preliminary ORR (CR+PR) and DCR (CR+PR+SD) for all efficacy-evaluable pts were 35% (13/37; all PR) and 57% (21/37; 13 PR, 8 SD), respectively; for pts with 0–1 prior line of therapy for advanced disease, ORR and DCR were 37% (10/27) and 63% (17/27). At data cutoff, 12/13 responses were ongoing (range, 1+ to 652+ days). PFS and biomarker analyses are ongoing. The most common TRAEs (≥10% of 40 pts) were fatigue (28%), rash (18%), and increased amylase (10%; asymptomatic). Grade ≥3 TRAEs occurred in 20% of pts (rash was the only grade ≥3 TRAE to occur in > 1 pt [n = 3]). Three pts discontinued due to TRAEs (grade 3 rash [n = 1]; grade 3 COPD exacerbation [n = 1], grade 2 diarrhea [n = 1]). Conclusions: E + P was generally well tolerated and associated with increased response compared with previously reported PD-1 inhibitor monotherapy in pts with advanced UC. A phase III UC study is planned. Clinical trial information: NCT02178722.

Correlation of constitutive PD-1 resistance in HNC with GM-CSF expression and presence of myeloid derived suppressor cells (MDSCs)

6049

Background: PD-1 checkpoint blockade is active in head and neck squamous cell carcinomas (HNC) with a response rate of ~18% and significant impact on survival. However, only a subset of patients benefits (Seiwert, Lancet Oncol). Biomarkers such as PD-L1 IHC and the Interferon-Gamma gene expression profile (INF-G GEP) identify inflamed tumors with a higher chance of response (~35-40%). However, it remains unclear why the majority of INF-G inflamed tumors still do not respond, or how to overcome constitutive resistance. Methods: 50 anti-PD-1 treated recurrent/metastatic HNC patients were included. Tumor RNA was analyzed using a 638-gene immune panel on the Nanostring nCounter. HPV status was assessed by HPV E6/E7 mRNA. T-cell inflamed phenotype was calculated using the 6-gene INF-G GEP (geometric mean) using both a low (6) and a high cutpoint. Differential gene expression was determined between inflamed-benefitting (IB) patients (defined as OS ≥ 250days), and inflamed-non-benefitting (INB) patients. Candidate biomarkers were evaluated in the entire cohort. Results: CD8 correlated highly with INF-G GEP (R = 0.80), suggesting T cell-driven inflammation. Comparing inflamed benefitting with inflamed non-benefitting tumors, the most differentially expressed gene was CSF-2, encoding GM-CSF, with 4-fold higher expression in inflamed non-benefitting (INB) tumors (with both cutpoints). In the overall anti-PD-1 treated cohort of 50 patients, CSF-2/GM-CSF correlated strongly with poor overall survival (P = 0.02), outperforming both HPV status, or PD-L1 expression in cox PH multivariate analysis. GM-CSF expression has been linked to myeloid derived suppressor cells (MDSC); MDSC marker CD34, as well as JAK2/IL10 were significantly elevated in inflamed non-benefitting (INB) tumors. There was no difference in M2 macrophage marker CD163 between the two groups. Conclusions: Constitutive resistance to PD-1 checkpoint blockade in inflamed HNC associates with expression of GM-CSF and Myeloid Derived Suppressor Cell (MDSC) markers. Strategies to deplete MDSCs, such as chemotherapy, should be considered in combination or sequentially with anti-PD-1.

A pharmacodynamic study of sirolimus and metformin in patients with advanced solid tumors

TPS11628

Background: Sirolimus is an inhibitor of the mammalian target of rapamycin (mTOR). Metformin has shown anti-cancer activity through its cellular (e.g., AMPK activation) and systemic effects (e.g., inhibition of IGF-1). We conducted a pilot study to test the hypothesis that metformin may potentiate mTOR inhibition by sirolimus. Methods: An open-label, randomized study was conducted in which eligible patients with advanced solid tumors were started on sirolimus (3mg daily) alone for the first 7 days. On day 8, patients were randomized to either receive metformin XL (500 mg daily) plus sirolimus (Arm A) or sirolimus alone (Arm B) for until day 21. From day 22 onwards, all patients recieved metformin XL plus sirolimus. The pharmacodynamic (PD) biomarkers were collected at baseline, day 8 and day 22 of cycle 1. The primary endpoint was to compare the change in PD biomarker phospho-p70S6K, using a two-sample t test (log ratio D22/D8 in arm A vs. arm B). The phospho-p70S6K was measured in peripheral blood T cells using Western blot. The secondary endpoints were to assess objective response rate (RECIST 1.1), toxicity (CTCAE V4.0) and changes in the serum levels of PD biomarkers: fasting glucose, triglycerides, insulin, C-peptide, IGF-1, IGF-1R, IGF-BP, leptin and adiponectin using two-sample t tests. Results: 24 patients were enrolled, at which time an interim futility analysis was conducted. 18 patients were evaluable for the primary endpoint (8 in arm A; 10 in arm B). The mean log ratios D22/D8 in phospho-p70S6K in arms A and B were -0.12 (SD = 0.13) and -0.16 (SD = 0.29), respectively (P = 0.64). Of the 17 pts evaluable for response, the best response was stable disease in 9 patients and progressive disease in 8 patients. There were no dose-limiting or unexpected toxicities. Of the 21 patients evaluable for serum PD biomarkers, there were no significant differences between arms A and B in fasting glucose, triglycerides, insulin, C-peptide, IGF-1, IGF-BP1, IGF-BP3, leptin and adiponectin (P > 0.05 for all). Conclusions: The addition of metformin to sirolimus, although well-tolerated, was not associated with significant changes in phospho-p70S6k and other PD biomarkers. Based on the results of the interim analysis, the trial was terminated. Clinical trial information: NCT02145559.

Epacadostat plus pembrolizumab in patients with advanced RCC: Preliminary phase I/II results from ECHO-202/KEYNOTE-037

4515

Background: Epacadostat (E) is a potent oral inhibitor of indoleamine 2,3-dioxygenase 1 (IDO1), a tryptophan-catabolizing enzyme that induces immune tolerance by T-cell suppression. Preclinical and clinical data suggest that epacadostat has antitumor activity when combined with checkpoint inhibitors, including the PD-1 inhibitor pembrolizumab (P). ECHO-202/KEYNOTE-037 is an ongoing open-label, phase 1/2 (P1/2) study evaluating E + P in multiple tumor types. We report preliminary P1/2 efficacy and safety data for the advanced renal cell carcinoma (RCC) cohort as of a 29OCT2016 data cutoff. Methods: Eligible patients (pts) had advanced clear-cell RCC, prior antiangiogenic therapy (tx), and no prior checkpoint inhibitor tx. In P1 dose escalation (3+3+3), pts received E (25, 50, 100, or 300 mg PO BID) + P (2 mg/kg or 200 mg IV Q3W); MTD was not exceeded. E (100 mg BID) + P (200 mg Q3W) dosing was selected for P2 cohort expansion. Response was assessed in RECIST 1.1 evaluable pts. Safety/tolerability was assessed in pts receiving ≥1 E + P dose. Results: 33 pts (P1, n = 11; P2, n = 22) were enrolled (median age, 63 years; 70% men; 97% white; MSKCC criteria of favorable, intermediate, and poor in 6%, 64%, and 12% of pts, respectively). Of 30 efficacy-evaluable pts, 63% (n = 19) had 0–1 prior tx and 37% (n = 11) had ≥2 prior tx for advanced disease. ORR (CR+PR) and DCR (CR+PR+SD) for pts with 0–1 prior tx was 47% (9/19; 1 CR, 8 PR) and 58% (11/19; 1 CR, 8 PR, 2 SD), respectively; for pts with ≥2 prior tx, ORR and DCR were 0% and 36% (4/11; all SD). At data cutoff, 9/9 responses were ongoing (range, 1+ to 372+ days). PFS and biomarker analyses are ongoing. TRAEs occurring in ≥10% of the 33 pts included fatigue and rash (36% each); and arthralgia, diarrhea, pruritus, and pyrexia (12% each). Grade ≥3 TRAEs occurred in 15% of pts (none in > 1 pt). Two pts discontinued due to TRAEs (grade 3 autoimmune hepatitis, n = 1; grade 3 aseptic meningitis/headache/nausea/vomiting/anxiety, n = 1). Conclusions: E + P was generally well tolerated and associated with encouraging response outcomes in advanced RCC pts with 0–1 prior line of tx. E + P represents a novel immunotherapeutic strategy. A phase 3 RCC study is planned. Clinical trial information: NCT02178722.

Identification of T-cell-inflamed gastric adenocarcinoma in The Cancer Genome Atlas (TCGA)

16

Background: Gastroesophageal adenocarcinoma (GEC) is a significant global health problem. KEYNOTE-012 demonstrated a 22% objective response rate in patients with PD-L1 expressing GEC that received pembrolizumab. A subset of patients (pts) tumors express a T cell “inflamed” (TCI) phenotype, which can be measured using an IFN-γ-based immune signature and may prove more predictive of clinical benefit. Using a 160 gene RNA-Seq immune expression profile, we sought to characterize the molecular environments of TCI versus non-TCI GEC patients in The Cancer Genome Atlas (TCGA). Methods: 395 GEC pts with primary tumors in TCGA were clustered into TCI, non-TCI, and intermediate subtypes using both unsupervised hierarchical and K-means clustering (k = 3). Molecular characteristics were categorized using data acquired via CbioPortal and the UCSC Xena repository. Only non-silent somatic mutations and copy number variations (CNVs) reaching GISTIC2 -2 or +2 thresholds were considered. Statistical comparisons were performed using chi-square, ANOVA, and t-test. Results: The TCI subtype contained patients from all TCGA-defined subtypes - EBV-associated (56%), MSI-high (16%), chromosomal unstable (6%), and genomically stable (27%). No significant differences were seen between TCI and non-TCI for tumor site or stage. Mutations in PTEN, PIK3CA, CDH1, and RHOA were more frequent in TCI patients. ERBB2, CCNE1, and KRAS CNVs were infrequent in TCI patients as were PDE4D deletions ( p< 0.05 ). TCI tumors had higher expression of both co-inhibitory (PD-1, PD-L1/L2, CD28/80, BTLA, LAG3) and co-stimulatory (CD137/40/27, OX40, GITR, ICOS) checkpoint molecules ( p< 10-7). Total mutation burden was no different between TCI and non-TCI pts when excluding MSI-high pts nor when assessing MSI-high alone. Conclusions: The IFN immune phenotype encompassed GEC patients from all TCGA subsets. Correlation of clinical outcome with checkpoint blockade is necessary to confirm these molecular associations and the independent predictive utility of this immune-profile stratification.

A phase II study of cediranib, an oral VEGF inhibitor, in previously untreated patients with metastatic or recurrent malignant melanoma

PURPOSE

A two stage multi-institution Phase II study was undertaken by the Princess Margaret Hospital Consortium to evaluate the efficacy and toxicity of oral cediranib, an inhibitor of vascular endothelial growth factor receptors 1 and 2, in patients with previously untreated advanced malignant melanoma.

PATIENTS AND METHODS

Between May 2006 and April 2008, 24 patients (median age 65 years) with advanced malignant melanoma were treated with oral cediranib. Cediranib was given on a continuous, oral once daily schedule of 45 mg, on a 28 day cycle.

RESULTS

Of the 17 patients evaluable for response, there was stable disease in 8 patients, and progressive disease in 9 patients, with no objective responses seen. Only 2 patients had stable disease >/= 6 months, thus the study was terminated at the end of stage 1 accrual. The overall median survival was 9.9 months, and the median time to progression was 3.5 months. The most frequent non-hematologic adverse events were hypertension (78%), fatigue (69%), diarrhea (69%) and anorexia and nausea (each 57%).

CONCLUSIONS

Although 2 patients had stable disease at 6 months, the short median time to progression and lack of any objective responses indicate that single agent cediranib at this dose and schedule is not sufficiently active to warrant study continuation.

Abstract IA09: Molecular mechanisms controlling the T cell-inflamed tumor microenvironment: Implications for therapy

Two major categories of melanoma metastases have been observed based on gene expression profiling and confirmatory technologies. One subgroup of patients has T cell-inflamed phenotype that includes expression of chemokines, T cell markers, and a type I IFN gene signature. In contrast, the other major subset lacks this phenotype and these tumors show a behavior of immune “exclusion”. The optimal immunotherapeutic intervention may be distinct for these two general subsets. The T cell-inflamed tumor microenvironment subset contains the highest expression of negative regulatory factors, including PD-L1, IDO, and FoxP3+ Tregs, and evidence for T cell-intrinsic anergy has also emerged. In addition, the mechanism of induction of these inhibitory mechanisms has been elucidated—PD-L1 and IDO are induced by IFN-γ, and Tregs are largely recruited by chemokines such as CCL22, produced by activated CD8+ effector T cells. Preclinical experiments have confirmed a critical role for all 4 of these mechanisms in limiting anti-tumor T cell efficacy in vivo, giving a mechanism base for translation back into the clinic. These include anti-PD-1/PD-L1 mAbs, IDO inhibitors, and approaches to deplete CD25+ Tregs and/or reverse anergy. The presence of multiple inhibitory mechanisms in the same tumor microenvironment argues that combination therapies may be advantageous to overcome compensatory effects. Preclinical data indicated synergy between anti-CTLA-4 +/- anti-PD-L1 +/- IDO inhibition. The mechanism of synergy is striking, as it correlates with a marked improvement of IL-2 production and proliferation of tumor-infiltrating CD8+ T cells. Clinical translation of these combination immunotherapies is promising and ongoing. In contrast to the T cell-inflamed melanomas, a new paradigm may be needed to promote de novo inflammation in cases of the non-T cell-infiltrated tumor microenvironment. Recent work has indicated that successful innate immune sensing of tumors involves the STING pathway and type I IFN production. Therefore, STING agonists are being developed to promote effective innate immune activation and generation of an adaptive immune response. Molecular mechanisms that explain the presence or absence of a T cell-inflamed tumor microenvironment in patients are being elucidated, which is opening up new concepts for therapies that may overcome immune exclusion and render patients responsive to immunotherapies.

Citation Format: Thomas Gajewski. Molecular mechanisms controlling the T cell-inflamed tumor microenvironment: Implications for therapy. [abstract]. In: Proceedings of the AACR Special Conference: Tumor Immunology and Immunotherapy: A New Chapter; December 1-4, 2014; Orlando, FL. Philadelphia (PA): AACR; Cancer Immunol Res 2015;3(10 Suppl):Abstract nr IA09.

Inhibition of the STING pathway by the AIM2 inflammasome (INM2P.352)

Tumor-derived DNA triggers the host STING pathway and the subsequent production of IFN-β by antigen-presenting cells, facilitating adaptive immunity against tumors. But cytosolic DNA also activates the AIM2 inflammasome, raising the question of functional impact of this pathway. We found that after DNA stimulation, macrophages or dendritic cells deficient in AIM2, ASC or caspase-1 produce substantially higher IFN-β. This observation suggests possible negative regulation of the STING pathway by the AIM2 inflammasome. We found no evidence for differential delivery of DNA into the cells, increased cell death triggered by activation of the inflammasome, or release of an inhibitory soluble factor. However, biochemical analysis revealed that the AIM2 inflammasome inhibits the early steps of STING pathway activation. Aggregation of STING and phosphorylation of TBK1 and IRF3 were highly enhanced in AIM2 inflammmasome-deficient cells. In addition, pre-treatment with a caspase-1 inhibitor increased the activation of TBK1/IRF3, suggesting that assembly and activation of the inflammasome complex is necessary for the inhibition of the STING pathway. Our data suggest the active AIM2 inflammasome inhibits the STING pathway at or upstream from STING. This crosstalk may have potential consequences for anti-tumor immunity which should be explored in vivo.