McDermott, Joseph

Abstract Metastatic pancreatic cancer patients who previously progressed on nab-paclitaxel/gemcitabine (average of 3 previous lines if treatment) were treated with elraglusib in combination with a rechallenge of this combination. The majority of patients demonstrated a prolonged progression free survival. While this early clinical evidence is highly encouraging, the biological determinants of elraglusib response are unknown. In vitro screens of elraglusib on pancreatic PDX cell lines showed that some tumor cells possess modest intrinsic resistance to elraglusib as measured by LD50 which ranged from 316.9 nM to 1080 nM in a screen of 10 cell lines. To explore this phenomena further, we induced resistance in three cell line models by growing cells in the presence of increasing concentrations of elraglusib. We then sequenced the transcriptome of the parental and resistant lines to identify potential mechanisms of elraglusib resistance. Comparison of RNA sequencing profiles identified upregulation of the Aldo-keto reductase (AKR) family of genes. This upregulation was confirmed by qRT-PCR and by immunoblot demonstrating that AKRs were upregulated at both the transcriptomic and protein level. We further quantified the AKR levels in the remainder of the 10 cell lines and found that many of the more resistant lines had high expression of AKR1C2. However, not all of the resistant lines had high expression which could be because they have a separate mechanism of resistance to elraglusib. A large body of literature has found that AKR genes provide resistance to many other small molecule inhibitors either by direct metabolism or by indirectly relieving oxidative stress. Given these results, we identify aldo-keto reductases as potentially associated with elraglusib resistance, but further investigation is warranted to establish a causal link between AKR and the resistant phenotype. Citation Format: Taylor Weiskittel, Ding Li, Lingling Han, Joseph McDermott, Benedito Caneiro, Hu Li, Daniel Billadeau, Andrew Mazar. Exploring mechanisms of resistance to elraglusib in pancreatic cancer PDX models. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 3857.

Abstract Elraglusib is a selective GSK-3 inhibitor that has activity on a broad array of neoplasms as well as their immune microenvironment. However the determinants of tumor response have not been elucidated. Here, we use genomic panels of cancer driver genes to predict patient response to elraglusib therapy as measured by RECIST criteria. Patient response in the Phase I 1801 trial (NCT03678883) was matched to genomic data from patient tumors, with 80 cases used to train Machine Learning models and 26 cases reserved as a test set for model evaluation. When predicting response from mutation status alone our models reached a modest accuracy of 65% in the test set, which we attributed to the sparsity of genomic data. To remedy this, we engineered pathway-based features to provide combinatorial information while ensuring feature stability. To select the optimal combination of features, we designed an iterative design process that first randomized feature combinations and then created refined models from the highest performing pathway features. This resulted in several final models with greater than 88% accuracy. We retained several models of similar accuracy to more comprehensively identify potential biomarkers and combinatorial relationships. These models were then interpreted using SHapley Additive exPlanations (SHAP) where we were able to identify highly predictive features for both the entire pan-cancer cohort and specific histologies. We found the feature constructed from the Reactome pathway “Chromatin Modifying Enzymes” frequently occurred in high performing models with a high-ranking SHAP value. This feature included genes such as SMARCA4, HDAC2, and KDM5A, and was negatively associated with patient response. Another frequently observed high-ranking pathway feature with a negative response association was based on the “Innate Immune System” Reactome pathway. Many of the features highlighted in our models were negatively associated with elraglusib response and thus could be general markers of poor prognosis. To address this, we altered our pipeline to test only positively associated features which substantially reduced our feature pool to only 285 features but still produced models with accuracies of up to 81%. Analysis of key features used by these models identified several positive markers of elraglusib including mutations in the POLE gene, which previously has been linked to DNA-damage response deficiency and anti-tumor immune response. In sum, we forecast elraglusib response for a variety of tumor histologies and simultaneously reveal potential mechanisms of elraglusib sensitivity and biological action. In future work we will investigate the utility of using POLE and other identified biomarker candidates to predict the likelihood of patient response to elraglusib, as well as whether our machine learning models will be an effective tool to guide patient stratification. Citation Format: Joseph McDermott, Taylor Weiskittel, Daniel Billadeau, Benedito Carneiro, Hu Li, Andrew Mazar. Elraglusib response prediction and mechanistic discovery using iterative machine learning. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5355.

3079 Background: Elraglusib (9-ING-41), a novel inhibitor of GSK-3β, has been evaluated in >230 patients (pts) with advanced malignancies including metastatic melanoma in the 1801 phase 1/2 trial (NCT03678883). Pts with melanoma treated with single agent elraglusib demonstrated clinical benefit including a durable complete response despite multiple prior lines of therapy including checkpoint inhibitors (CPI). Methods: We collected data from numerous -omic modalities in the 1801 trial to characterize markers of elraglusib response with bioinformatics and machine learning (ML). Genomic panels were collected on 105 pts. The most frequent tumor types included colorectal cancer (n=35), pancreas (n=31), lung (n=15) and metastatic melanoma (n=12). Pts were treated with elraglusib as a single agent or in combination with chemotherapy. Pathway based features were generated by quantifying the percent of each Reactome pathway that was mutated. These pathway features were iteratively evaluated for their performance in shallow neural networks predicting patient RECIST with each model using 5-15 randomly selected features at a time. Publicly available genomic data from an anti-PD-1 trial in patients with metastatic melanoma (Hugo et al., 2017) was transformed into pathway-based features and input into the final model to predict elraglusib response in this population. Results: Our final pathway based genomic model for elraglusib response performed with a training accuracy of 93% and a test accuracy of 88% on the 1801 patients. We then tested the model on metastatic melanoma patients who were treated with anti-PD-1 therapy. We found that patients from Hugo et al. were predicted to respond to elraglusib (22/26). In addition, most patients with tumors resistant to anti-PD-1 therapy were also predicted to respond to elraglusib (10/12). Finally, we tested the model on metastatic melanoma patients treated with elraglusib in the 1801 study that had failed CPI and for whom -omic data was available. The model correctly identified 5/5 patients that demonstrated clinical benefit (1 CR, ongoing>1400 days; 4 prolonged OS of 107, 256, 357, and 556 days). Conclusions: We applied robust in-silico modeling developed using 1801 trial data to develop a rationale for the development of elraglusib in the treatment of metastatic melanoma including patients that failed or were refractory to anti-PD-1 treatment. This model correctly identified metastatic melanoma patients in 1801 with good clinical outcomes that had failed prior CPI treatment. Further, we present this in silico approach as a robust way to repurpose prior clinical study genomics and ML modeling for population enrichment during the clinical development of novel drugs. Clinical trial information: NCT03678883 .

Abstract 9-ING-41 (elraglusib) is a potent and selective GSK-3 inhibitor that has shown anti-tumor activity in patient-derived xenograft models and phase1/2 clinical studies in patients (pts) with advanced solid tumors. Preclinical studies have demonstrated that elraglusib downregulates PD-1, TIGIT and LAG-3, upregulates expression of MHC class I proteins in tumor cells from “cold tumors”, and shows synergy when combined with PD-1 blockade in mouse xenografts. Clinical activity has been observed both as a single agent and in combination with standard of care chemotherapies in several advanced cancer histologies. Here we present initial ‘omics data from our phase 1/2 study (NCT03678883; ACT1801) spanning >100 pts with advanced cancer evaluating both single agent and chemotherapy combinations. The patients (pts) included in this study received elraglusib as a ≥3rd line therapy for advanced disease. Most pts with melanoma treated with elraglusib monotherapy (8/9; 89%) stayed on study for >2 cycles. One patient with melanoma refractory to checkpoint inhibitors achieved a confirmed and durable CR. Clinical benefit was also observed among pts with colorectal cancer with 4/12 (33.3%) pts treated with monotherapy and 12/15 (80%) that received elraglusib plus irinotecan rechallenge stayed on study >2 cycles reaching median overall survival of 106 and 211 days, respectively. Based on emerging in vivo and in vitro results demonstrating that elraglusib activates T and NK cells promoting anti-tumor immune responses, we hypothesize immunomodulation by elraglusib may be contributing to anti-tumor immune response in the 1801 trial. We have acquired TCRseq and RNAseq profiles of PBMC samples from seven patients in 1801 during the first two weeks of treatment with elraglusib monotherapy. In these pts, reduced TCR clonality was observed and specific TCR clonotypes expanded after treatment, indicating T cell activation and expansion. These pts also showed changes in PBMC populations during elraglusib therapy as measured by immune deconvolution of PBMC RNAseq. Taken together, these data support a novel, previously unrecognized immunomodulatory mechanism of action for elraglusib and could provide rationale for future clinical development of elraglusib in pts with advanced malignancies. Additional analysis from TCRseq, immune profiling and cytokine analysis from expanded cohort of pts is ongoing. Citation Format: Taylor Weiskittel, Joseph McDermott, Daniel Billadeau, Hu Li, Anwaar Saeed, Devalingam Mahalingam, Solmaz Sahebjam, Pamela Munster, Ludimila Cavalcante, Frank J. Giles, Andrew Mazar, Benedito Carneiro. Multimodal mechanism of action of the GSK-3 inhibitor 9-ING-41 (elraglusib) includes an immunomodulatory component: preliminary results from the 1801 phase 1/2 trial [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr CT222.