Human-Based Preclinical Models May Improve Toxicity Prediction in Oncology Development

As interest in organoids and other human-based preclinical models continues to grow, questions remain about how these technologies should be incorporated into drug development. During a discussion related to his panel participation at BIO International Convention 2026, Mohit Trikha, PhD, CEO of Kiva Bioscience, shared his perspective on the role these models may play in oncology research and why he believes caution is warranted before replacing traditional animal studies.

Can human-based models improve the prediction of drug toxicity?

Drawing on a career that spans preclinical research, early clinical development, and multiple cancer therapies that advanced from discovery to approval, Trikha described his outlook as one of “cautious optimism.” While encouraged by advances in human-relevant models, he said there is not yet enough evidence to support eliminating animal testing entirely.

“I do not think we should be replacing animal testing with these preclinical model systems, because it's too early,” Trikha said. “You need to do both. You need a more robust package that raises the bar.”

According to Trikha, the oncology field continues to face high rates of clinical trial failure despite significant advances in cancer treatment. In many cases, he noted, traditional animal models have struggled to predict how therapies will perform in patients. Replacing those models without sufficient validation could introduce additional uncertainty as candidates enter first-in-human studies.

Rather than focusing solely on efficacy, Trikha believes the greatest opportunity for next-generation preclinical models lies in improving the prediction of toxicity and tolerability. Many investigational cancer therapies demonstrate promising antitumor activity but ultimately face challenges related to safety, dosing, pharmacology, or a narrow therapeutic index.

At Kiva Bioscience, where the company is developing targeted antibody-drug conjugates (ADCs), Trikha said the goal is to create therapies that are both more effective and better tolerated. He believes future preclinical models should help researchers identify patients at greater risk for adverse events and provide earlier insight into risk-benefit profiles.

“We have a lot of models that predict efficacy,” Trikha said. “There is not enough emphasis on markers of toxicity and markers of tolerability.”

As human-based models continue to evolve, Trikha sees their greatest value not as replacements for existing approaches, but as complementary tools that could help developers build safer, more patient-centered cancer therapies.

BIO 2026 runs June 22–25 in San Diego.

Watch the first segment of Trikha’s interview, and click here for more conference coverage.

About the Speaker

Mohit Trikha, PhD, CEO, Kivu Bioscience

Dr. Mohit (Mo) Trikha is the CEO of Kivu Bioscience. He brings over 25 years of drug development expertise in oncology. He has been instrumental in advancing more than 40 programs from target identification to clinical trials in oncology. Previously, Mo was a venture partner at Apple Tree Partners, and also led oncology early development at AbbVie, where he oversaw the development of discovery-stage programs to clinical proof of concept, focusing on ADCs, bispecifics, and CAR-T therapies. He has contributed to the development of key drugs such as Kadcyla, Polivy, and vismodegib, and chaired collaborations with leading institutions like TeneoBio, Calibr, and The Scripps Research Institute. Earlier in his career, he held roles at Genentech, Centocor/Johnson & Johnson, and Triphase Accelerator. Mo earned his B.S. in Biochemistry from California State University, Los Angeles, and his PhD in Biochemistry and Molecular Biology from the University of Southern California.