Northwell Health’s Feinstein Institutes for Medical Research has highlighted peer-reviewed Phase 1 results for DYP688, a Novartis Pharmaceuticals Corporation-developed antibody-drug conjugate designed to deliver the Gq/11 inhibitor SDZ475 to PMEL-expressing, GNAQ or GNA11-mutant melanoma cells. In the 66-patient dose-escalation study, confirmed objective responses occurred in 19.7% of participants, the disease control rate reached 81.8%, and median progression-free survival was 7.2 months. The central commercial complication is that Novartis has made a business decision to halt the programme, leaving the encouraging early clinical signal without a defined development path.
The results were published in Nature Medicine on July 13, 2026, before Northwell drew renewed attention to the study on July 24. That distinction matters because this is not merely a hospital announcement describing unpublished observations. The dataset has undergone peer review and includes detailed safety, pharmacokinetic, biomarker and tumor-response analyses, although peer review does not change the fact that the evidence comes from an open-label, non-randomized, first-in-human dose-escalation trial.
Most participants had advanced and difficult-to-treat disease. Sixty of the 66 patients had metastatic uveal melanoma, more than 92% had received prior anticancer treatment, and nearly 60% had received at least two previous systemic treatment lines. Liver metastases were present in 86.4% of participants, while elevated lactate dehydrogenase levels and substantial circulating tumor DNA indicated that many entered the trial with biologically aggressive disease.
Why could DYP688 matter in metastatic uveal melanoma despite the programme’s uncertain future?
Uveal melanoma is biologically different from the more common cutaneous form of melanoma. Activating mutations in GNAQ or GNA11 occur in approximately 85% to 90% of tumors and drive persistent oncogenic signalling through pathways involved in cell growth and survival. Directly inhibiting the resulting Gq/11 proteins has been scientifically attractive, but systemic exposure to potent Gq/11 inhibitors has previously been associated with risks including hypotension, platelet dysfunction and bleeding.
DYP688 was designed to address that delivery problem. The antibody component binds PMEL, also known as PMEL17 or gp100, a melanocyte-lineage protein expressed by uveal melanoma cells. After the antibody-drug conjugate is internalized, it releases SDZ475, a Gq/11 inhibitor intended to block the disease-driving pathway inside the targeted cell rather than exposing the entire body to comparable concentrations of the free payload.
This is not a conventional antibody-drug conjugate carrying a broadly cytotoxic chemotherapy payload. DYP688 uses what the researchers described as a biology-matched payload, meaning that both the antibody target and the delivered inhibitor are connected to the underlying biology of the tumor. That approach could widen the design space for antibody-drug conjugates beyond familiar microtubule inhibitors, DNA-damaging agents and topoisomerase payloads.
The strategy could be particularly relevant for patients who cannot receive tebentafusp. The United States Food and Drug Administration approved tebentafusp, marketed as Kimmtrak, for HLA-A*02:01-positive adults with unresectable or metastatic uveal melanoma. Because eligibility depends on that specific HLA type, a substantial proportion of patients are excluded from its approved indication.
DYP688 does not depend on HLA-A02:01-mediated presentation of a PMEL peptide to T cells. Responses in the Phase 1 study were reported in HLA-A02:01-negative patients as well as in patients previously treated with tebentafusp. Tumor samples also suggested that PMEL expression remained present after prior tebentafusp exposure, supporting the possibility that the two PMEL-directed mechanisms might be used sequentially rather than being mutually exclusive. That remains an early clinical hypothesis, not an established treatment sequence.
What did the DYP688 first-in-human trial show across the overall and recommended-dose populations?
Patients received intravenous DYP688 at doses ranging from 4 milligrams per kilogram to 24 milligrams per kilogram every two weeks, or 12 milligrams per kilogram to 16 milligrams per kilogram weekly. The primary purpose of the Phase 1 portion was to assess safety, tolerability and dose selection, while response rate, pharmacokinetics and other measures of antitumor activity were secondary or exploratory endpoints.
Thirteen of the 66 treated patients achieved confirmed objective responses, producing an overall response rate of 19.7%. Stable disease was recorded in 41 patients, while 11 had progressive disease as their best response. The resulting disease control rate was 81.8%, and measurable tumor reduction occurred in 47 patients, or 71.2% of the study population.
The estimated median duration of response was 10.5 months, while median progression-free survival across the study was 7.2 months. Fourteen patients remained on treatment for at least 12 months, indicating that the benefit was not limited to brief radiographic changes in every responding or stable patient. Nevertheless, the confidence interval around duration of response remained immature at the upper end, and long-term survival benefit was not established.
The selected recommended dose was 16 milligrams per kilogram every two weeks. Among the 14 patients who started at that dose, five achieved objective responses, equivalent to 35.7%, while median progression-free survival was estimated at 7.4 months. These subgroup figures are notable, but the cohort is too small to provide a dependable standalone estimate of efficacy or to support comparisons with other treatments.
The dose-response pattern was not completely linear. Greater exposure was generally associated with deeper tumor and circulating tumor DNA reductions, but activity appeared weaker than expected in the highest 24-milligram-per-kilogram cohort. Researchers said that group also had less favorable baseline characteristics, including higher lactate dehydrogenase and circulating tumor DNA levels, which may have contributed to the result.
Exploratory biomarker analyses also indicated that patients with greater baseline tumor burden, higher lactate dehydrogenase and higher circulating tumor DNA were less likely to experience tumor regression. Lower baseline circulating tumor DNA was associated with longer progression-free survival, although it remains unclear whether this is primarily a prognostic marker of less aggressive disease or a biomarker capable of predicting DYP688 benefit specifically.
Did DYP688 successfully reduce systemic exposure to its potentially toxic Gq/11 payload?
The pharmacokinetic findings offer some support for the central engineering concept behind DYP688. At the highest every-two-week dose, systemic exposure to free SDZ475 in blood was approximately 80-fold lower than exposure to the active payload while it remained conjugated to the antibody. This suggests that most of the pharmacologically potent material remained attached during circulation rather than being released widely into the body.
Researchers also reported dose-dependent reductions in circulating unbound PMEL, which were considered consistent with target engagement. Paired tumor biopsies showed reduced activity across MAPK and cell-proliferation gene signatures after treatment, adding mechanistic evidence that DYP688 was affecting pathways downstream of GNAQ and GNA11 inside tumors.
These biomarker findings strengthen the biological rationale, but they do not independently prove clinical benefit. Circulating PMEL measurements may be affected by competition between DYP688 and the laboratory assay, while transcriptomic analyses involved a limited number of paired biopsy samples. The most persuasive evidence remains the combination of tumor responses, exposure-response trends and tolerability rather than any single biomarker result.
How manageable was the DYP688 safety profile in the Phase 1 dose-escalation study?
Almost every participant experienced at least one treatment-emergent adverse event, which is common in heavily pretreated early-stage oncology studies. Treatment-related adverse events of any grade were reported in 90.9% of patients, but only five patients experienced Grade 3 treatment-related events. These consisted of hypotension, asymptomatic hypercalcemia, anemia, increased gamma-glutamyl transferase and decreased lymphocyte count.
One patient receiving 24 milligrams per kilogram every two weeks developed Grade 3 hypotension two days after the initial infusion. The event resolved within 24 hours following supportive care, and the patient continued treatment at a reduced dose without another hypotensive episode. This was the only dose-limiting toxicity reported.
The most frequent treatment-related events included hypercalcemia, dry mouth, fatigue, peripheral edema and anemia. Researchers suggested that hypercalcemia could be connected to the payload’s effect on the calcium-sensing receptor pathway, which relies partly on Gα11 signalling. That interpretation remains mechanistic rather than conclusively established.
Treatment-emergent serious adverse events occurred in 39.4% of participants, although only four patients, or 6.1%, experienced serious events judged to be treatment related. No Grade 4 or Grade 5 treatment-related adverse events were reported, and no patient discontinued DYP688 because of a treatment-related adverse event.
Those results support further safety evaluation, particularly because systemic Gq/11 inhibition has historically faced tolerability barriers. They do not establish that DYP688 is safe in a broader population, during prolonged use, in combination regimens or outside the closely monitored conditions of an early clinical trial.
Why can the DYP688 results not be compared directly with Kimmtrak or darovasertib combinations?
The DYP688 trial had no randomized control group, used several doses and schedules, and was not statistically designed to demonstrate efficacy. Tumor assessments were conducted through local radiology review, and the population included patients with different treatment histories, disease burdens and melanoma origins. The reported progression-free survival and response rates should therefore be interpreted as preliminary signals rather than confirmatory estimates.
Kimmtrak occupies a different evidence category because its regulatory approval was supported by a randomized Phase 3 trial demonstrating an overall survival benefit in HLA-A*02:01-positive metastatic uveal melanoma. Its response rate alone does not fully represent its clinical value, making any comparison based only on tumor shrinkage misleading.
The competitive bar has also moved during the period since the DYP688 Phase 1 trial began. IDEAYA Biosciences and Servier have reported randomized Phase 2/3 data for darovasertib plus crizotinib in first-line HLA-A*02:01-negative metastatic uveal melanoma. The OptimUM-02 study produced a median progression-free survival of 6.9 months versus 3.1 months for investigator-selected therapy and an independently reviewed response rate of 37.1% versus 5.8%, with a United States regulatory submission planned for the second half of 2026.
DYP688 therefore cannot advance merely by reproducing an early response signal. A future sponsor would need to identify a commercially and clinically meaningful position, potentially after tebentafusp, after darovasertib-based therapy, in an HLA-unrestricted population, or in a biomarker-selected subgroup. That positioning would then require prospective confirmation.
Why did Novartis halt DYP688 development despite the positive clinical signal?
The Nature Medicine paper records that enrollment was halted on July 8, 2025, after completion of the Phase 1 portion. Northwell subsequently reported that Novartis had made a business decision to stop further development, while Richard D. Carvajal indicated that academic investigators and patient communities continued to see potential in the programme. No public explanation detailing the portfolio, financial or strategic criteria behind the decision was provided in the Northwell announcement.
Clinical results and portfolio decisions often operate on different timelines. A large pharmaceutical company may discontinue an asset that shows activity if the expected development cost, competitive positioning, commercial opportunity, manufacturing complexity or probability of regulatory differentiation does not meet its internal investment threshold. A decision to stop development does not prove that the therapy lacked activity, just as promising Phase 1 data do not prove that continued investment would ultimately produce an approved product.
The publication may still preserve value. It provides human proof of concept for delivering a pathway-specific inhibitor through an antibody-drug conjugate, identifies a recommended dose, documents measurable antitumor activity and offers biomarker information that could guide future trial design. These findings could inform another PMEL-targeting programme or a broader generation of antibody-drug conjugates carrying molecularly targeted rather than conventionally cytotoxic payloads.
Restarting DYP688 itself would be more complicated. It would require rights to the molecule and supporting intellectual property, access to manufacturing and toxicology packages, a sponsor prepared to assume regulatory responsibility, and agreement on a confirmatory development strategy. No licensing, transfer or external development arrangement has been announced.
What evidence would be required before DYP688 could influence routine uveal melanoma treatment?
The next study would need to confirm the recommended dose in a larger and more uniform population, ideally with blinded independent central review and clearly defined groups based on HLA status, prior tebentafusp exposure and previous targeted therapy. Overall survival, durability of response, patient-reported outcomes and longer-term safety would be important, particularly if the programme were positioned against established or emerging systemic treatments.
Further work would also need to clarify whether baseline circulating tumor DNA, lactate dehydrogenase, BAP1 alterations or other molecular characteristics can identify patients more likely to benefit. The Phase 1 analyses generated several plausible signals, but small exploratory subsets can produce associations that do not reproduce in larger studies.
DYP688 has delivered something unusual: a peer-reviewed early clinical dataset with a credible mechanism, measurable activity and comparatively limited high-grade treatment-related toxicity, but no active sponsor-led programme to convert those findings into a registrational trial. The scientific question is no longer whether the molecule can reach its target and produce responses in some patients. The decisive question is whether another developer considers that evidence strong enough to finance the much harder work of proving durable patient benefit.
