Designing Beyond Tolerability Limits: Advancing GB-4362 Into Clinical Development
Some of today’s most effective cancer therapies are limited not by how well they work, but by how well patients can tolerate them. Even highly targeted treatments can expose patients to serious side effects when highly potent drugs escape their intended destination. Addressing this challenge requires new therapeutic strategies.
Antibody-drug conjugates (ADCs) are one of the fastest-growing classes of cancer therapies, delivering highly potent drugs directly to tumor cells. While they have shown strong efficacy in difficult-to-treat cancers, their use is often limited by payload-related toxicity. In the case of monomethyl auristatin E (MMAE)-based ADCs, peripheral neuropathy, or damage to nerves located outside the brain and spinal cord, can affect 60% to 70% of patients in some clinical settings. It can have lasting effects on quality of life and, in many cases, lead to treatment interruption or discontinuation.
GB-4362 is an investigational payload neutralizer designed to selectively bind and neutralize free MMAE in circulation while avoiding interaction with MMAE that remains conjugated to an ADC and active at the tumor site. The program was shaped by integrating computational modeling with clinical insight to define, before entering the clinic, where payload activity must be preserved to drive tumor response and where exposure must be controlled to improve tolerability. By reducing payload-mediated toxicity without compromising efficacy, GB-4362 aims to expand the therapeutic window of an entire class of ADCs.
GB-4362 received FDA Fast Track designation. On Aug. 6, 2026, we announced that all patients in the first cohort had been enrolled and that initial patients had been dosed in the Phase 1 study evaluating GB-4362 in combination with enfortumab vedotin and pembrolizumab for locally advanced or metastatic urothelial cancer.
To discuss how this AI-enabled approach shaped GB-4362 and what it means for improving safety without losing efficacy, we spoke with Gevorg Grigoryan, founder and chief technology officer; Dinesh De Alwis, senior vice president and head of clinical drug development; Kapil Mayawala, vice president of clinical pharmacology and strategic program leadership; and Victoria Szenes, senior director of clinical immunology.
Q: What are the goals of GB-4362, and how could it benefit cancer patients?
A: Kapil Mayawala: We developed GB-4362 with three goals in mind. First, it is designed to reduce toxicities associated with free MMAE, potentially enabling more patients with urothelial cancer to remain on Padcev plus pembrolizumab for longer.
Second, as this combination becomes increasingly integrated into real-world practice, including in community oncology settings, the need to monitor and manage side effects can create barriers to broader use. By reducing treatment-related toxicities, GB-4362 has the potential to ease these challenges and support broader use to Padcev plus pembrolizumab across a wider range of care settings.
Third, data from the ongoing Phase 1 study may help inform the potential development of GB-4362 with other MMAE-based ADCs, potentially extending its benefits to patients across multiple cancer types.
Q: What did AI make possible that traditional approaches could not?
A: Gevorg Grigoryan: The design challenge with GB‑4362 was that we needed an antibody that binds free MMAE in the bloodstream with very high affinity, has negligible interaction with MMAE while it is part of an ADC, preserving the ADC’s tumor‑killing activity, and still behaves like a well‑engineered, developable biologic.
Generative optimization allowed us to treat all of those requirements as a multi-component design objective rather than separate, sequential trade‑offs. By exploring a broad space of sequence variants in silico guided by high‑throughput functional data, we were able to improve neutralization potency for free MMAE while maintaining the system‑level selectivity and developability profile needed to pair safely with MMAE‑based ADCs.
Q: How does this approach differ from traditional oncology drug development?
A: Dinesh De Alwis: One of the things that is very impactful about GB-4362 is that, unlike most cancer drug development, which prioritizes efficacy first and treats toxicity as secondary, this program is fundamentally about benefit-risk.
The bar is not just whether a therapy works, but whether patients can actually tolerate it. Many of today’s therapies deliver strong efficacy, but patients are left to live for years with sometimes irreversible side effects. Peripheral neuropathy, for example, can be permanent in some cases, which is devastating for patients who may survive for several years but must live with a severe, life-altering toxicity.
Recent pivotal-trial data demonstrated strong efficacy with enfortumab vedotin plus pembrolizumab, while longer treatment duration was associated with an increase in peripheral neuropathy. GB-4362 could improve that benefit-risk equation.
Q: Why is addressing peripheral neuropathy critical to patients’ long-term quality of life and ability to remain on cancer treatment?
A: Victoria Szenes: Addressing peripheral neuropathy is critical because it can affect both a patient’s long-term quality of life and their ability to remain on an effective cancer treatment.
We currently have limited options to prevent or reverse treatment-induced peripheral neuropathy. Some medications may help alleviate the symptoms, but they do not repair the nerves or reverse the condition. Mild neuropathy may improve over time, but the likelihood of complete recovery decreases as severity increases. For some patients, symptoms can persist long after completion of their cancer treatment, and the deficit may become permanent.
Symptoms can include numbness, tingling, and burning or shooting pain, typically beginning in the toes and ascending over time. As neuropathy progresses, it can impair balance, making walking unsafe, and interfere with ordinary activities such as buttoning a shirt. Peripheral neuropathy can have a substantial and lasting impact on quality of life.
Compounding this physical burden is the concern that escalating severity of peripheral neuropathy may necessitate treatment delay, dose reduction, or discontinuation. As a result, patients may be unable to receive the full planned dose or duration of an otherwise effective therapy. This adds another layer of stress, leaving patients uncertain whether they will be able to complete their planned course of cancer treatment.
Ultimately, preventing or reducing peripheral neuropathy is not simply about managing a side effect. It is about preserving patients’ function and independence while helping them remain on effective cancer therapy for as long as it is safe and beneficial.
Q: What does the design of GB-4362 suggest about how AI could be used to address toxicity challenges across other ADC payloads or drug classes?
A: Gevorg Grigoryan: If GB‑4362 behaves in patients the way it was designed—lowering free MMAE exposure and related toxicities while preserving the benefit of MMAE‑based ADCs—it would demonstrate that toxicity itself can be treated as an engineering target, not just an unfortunate side effect.
AI is what makes that possible: it allows us to design molecules that are tuned to intercept a toxic species at the system level without rewriting every underlying therapy. That same playbook could, in principle, be applied to other payloads or modalities where the challenge is not “does the drug work?” but “can patients safely stay on it long enough to fully benefit?”
Q: What are you most excited about with regard to AI–drug design in oncology?
A: Gevorg Grigoryan: I’m excited that AI finally lets us design directly around the real constraints in oncology. Over time, as we accumulate more data and more programs, that should turn oncology drug discovery from a largely trial‑and‑error process into a learning system where each drug candidate informs the next.