Double Impact: Insilico Medicine Delivers Novel Immuno-Oncology Target and Preclinical Candidate Using AI
Expanding the Immuno-oncology Toolbox
with an AI-discovered Target and Novel Inhibitor

We are excited to announce the nomination of the novel preclinical candidate ISM4312A targeting Diacylglycerol Kinase Alpha (DGKA), where we used our artificial intelligence (AI)-based end-to-end platform Pharma.AI to discover a target and design the corresponding small molecule inhibitor. The in vitro and in vivo studies revealed excellent potency, high selectivity, favorable ADME, excellent oral bioavailability, and tolerance -- supporting the ongoing development of ISM4312A as a potential first-in-class DGKA inhibitor in cancer immunotherapy. After designing and evaluating a series of compounds in 15 months, we initiated the IND-enabling studies for the program.

Following a previous "double breakthrough" in discovering a novel target and developing a clinical candidate for fibrosis, Pharma.AI enabled us once again to repeat a rare success story, paving the way to a new strategy, this time in immuno-oncology.
We are committed to leveraging our end-to-end AI platform to discover novel targets to improve therapeutic efficacy in immuno-oncology to address unmet medical needs in cancer.
Feng Ren, PhD
Co-CEO and Chief Scientific Officer of Insilico Medicine
Checkpoint immunotherapy: a revolution with roadblocks
After three decades of failures in about 70 randomized trials, in June 2010, participants of the annual meeting of the American Society of Clinical Oncology (ASCO) witnessed a long-awaited breakthrough in cancer research -- clinical trial results for significant survival advantage improvement in metastatic melanoma, when treated with ipilimumab (Yervoy) developed by Bristol-Myers Squibb. Encouraging clinical trial results presented a promising future for the new monoclonal antibody targeting cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), a switch that limits the ability of T cells to attack cancer cells.

The approval of ipilimumab marked the beginning of the "checkpoint revolution", which would lead to a kaleidoscope of approvals in the coming decade, including such breakthroughs as the anti-PD-1 antibody Keytruda, introduced by Merck to treat multiple advanced solid tumors, nivolumab developed by Bristol-Myers Squibb (brand name Opdivo ), the nivolumab/ipilimumab combination for treating BRAF V600 wild-type melanoma, pembrolizumab for the treatment of advanced non-small cell lung cancer (NSCLC) and so on.

Today, there are nine FDA-approved checkpoint immunotherapies available for patients, including those targeting the PD-1/PD-L1 checkpoint, CTLA-4 checkpoint, and in 2022 -- a newly introduced LAG-3 immune checkpoint pathway.

In contrast to the narrow application of adoptive T-cell therapy, immunotherapy using checkpoint blockade antibodies has shown promising results across a wide range of cancer types, not only in traditionally immunogenic tumors such as melanoma and renal cell carcinoma, but also in lung cancer, bladder cancer, and head and neck cancer.

Yet, the approved checkpoint therapies also brought considerable safety and tolerability concerns. For example, the pioneering ipilimumab (Yervoy) can trigger immune-related adverse events in approximately 60% of patients receiving the drug compared with about 30% of those receiving only gp100. Common autoimmune-related side effects of ipilimumab include fatigue, diarrhea, skin rash, endocrine deficiencies, and colitis. Severe or fatal autoimmune reactions were reported in 12.9% of ipilimumab patients.

The even greater problem is that the clinical benefit of immunotherapy is limited, with only a small percentage of patients experiencing significant improvement (typically around 15-40%, depending on the type of cancer). There is a need to improve immunotherapy to realize its potential fully and potentially provide benefits to patients whose tumors do not respond to current treatments.
Rethinking immuno-oncology with PandaOmics
Immune checkpoint blockade has proven effective in various malignancies. Still, only a small patient population can currently benefit from the immuno-oncology revolution. For instance, while the existing PD-1/PD-L1 axis blockade has proven to be highly efficient, the number of patients that don't respond to PD-1/PD-L1 monotherapy remains high regardless of the cancer type. Therefore, we embarked on a mission to explore new adjuvant immunotherapy strategies.

There are various resistance mechanisms of PD-1/PD-L1 blockade in the cancer-immunity cycle. Each step of the cancer-immunity cycle requires the coordination of numerous factors, which makes finding a novel target a fairly complicated task. Using PandaOmics, the AI-driven target discovery engine of Insilico's Pharma.AI platform, enabled us to crack this code.
We applied PandaOmics to distinguish the critical drivers of the PD-1/PD-L1 therapy resistance and prevent the tumor from evading the immune response, aiming to augment the effect of the checkpoint inhibition. Ideologically, it can be done by comparing the expression profiles of non-responders and responders' patients and deriving the causality of immune escape.

We used six bulk RNAseq datasets, including four for melanoma and two for clear cell renal cell carcinoma, because of data availability and the responder/non-responder clinical status for each patient. AI-driven modeling of the bulk data helped us to calculate PandaOmics TargetID scores and mine targets. PadaOmics pointed us to Diacylglycerol kinase alpha (DGKA), which appeared to be at the top of the list of the most promising target candidates.
DGKA, a protein that has been linked to various aspects of cancer progression, such as survival, migration, and invasion of cancer cells, has been shown to interfere with T-cell function during anti-PD-1 therapy, potentially contributing to resistance to PD-1 blockade. This suggests that inhibiting DGKA may be a promising strategy to enhance the effectiveness of immunotherapy in cancer treatment. While serotonin antagonists ritanserin, R59022, and R59949 have been repurposed as DGKA inhibitors, their potency and selectivity are inadequate. Based on this knowledge and the PandaOmics modeling results, it became clear there is a potentially great opportunity to develop more potent and selective DGKA inhibitors.

We ran additional validation cycles on 2 single-cell melanoma datasets to ensure the right choice. Validation proved that DGKA inhibition could ensure full T-cell activation and avoidance of anergy and subsequently achieve a better response to anti-PD1 therapy.
After calculating 23 predictive AI scores we received a ranked list of potential targets as presented on the heatmap below. DGKA topped the list in both melanoma and ccRCC projects. In the next step, we performed manual analysis and validation on the single-cell level of all the top-scoring targets and found DGKA to be the most promising one.

The results show that DGKA inhibitors may overcome anti-PD-1 resistance and expand the responder patient population in cancer immunotherapy.
On its way to first-in-class
Once we had derived the target, we followed with AI-driven design of the molecules using the generative chemistry engine Chemistry42, another part of the Pharma.AI platform. After designing and evaluating a series of compounds over 15 months, our team nominated ISM4312A as a preclinical candidate and initiated the IND-enabling studies for the program.

ISM4312A is a potential first-in-class DGKA inhibitor with excellent potency and high selectivity in cancer immunotherapy. The compound enhanced T-cell activity in vitro and showed robust anti-tumor activities with or without anti-PD-1 therapy in vivo. It also exhibited favorable ADME, excellent oral bioavailability, and tolerance. The deeper investigation of the biomarker and combination mechanism is ongoing, and we plan to use the findings to advance a new potential therapeutic strategy in immuno-oncology.
This is the 7th preclinical candidate (PCC) delivered using AI in 2022 alone, bringing the total number of AI-enabled PCCs the company has nominated since 2021 to 9, further validating the strength of our end-to-end Pharma.AI platform
Pharma.AI is now used by many pharmaceutical companies to accelerate their pharmaceutical R&D cycles and increase the probabilities of success.
Alex Zhavoronkov, Ph.D.
Founder and CEO of Insilico Medicine
Growing AI-generated pipeline
Over the last several years, we have created and extensively validated the end-to-end AI system, Pharma.AI, which can accelerate drug discovery for various indications at a fraction of a typically incurred R&D cost. Since 2021, our team has managed to deliver 9 preclinical candidates discovered and designed using its AI platform in various disease areas, including fibrosis, inflammation, and oncology, including one for the QPCTL immuno-oncology target in partnership with Fosun Pharma. We have also successfully completed a Phase 0 microdose trial for our drug candidate for treating idiopathic pulmonary fibrosis (IPF) and advanced the molecule to a Phase I clinical trial.

One of the rare events in the drug discovery realm is when a company manages to discover a novel target and follow up with a corresponding potentially first-in-class preclinical candidate. It is a double-impact breakthrough event. Using the AI-first approach, we have done it twice within several years -- the first time for our fibrosis program and now for the immuno-oncology. We firmly believe the AI platform-based drug discovery paradigm is the future of pharmaceutical research.

To reinforce our leading position in the AI-first drug discovery world, we have recently doubled down in our efforts to build a fully automated in-house biology research robotics laboratory which will be a strategic asset in accelerating the generation of big biological data. This will further reinforce our ability to expand our drug discovery pipeline -- both internally and with external partners.
Stay tuned, follow us on social media!