A first-in-human clinical trial of roginolisib, a next-generation inhibitor of the delta isoform of phosphatidylinositol 3-kinase (PI3Kδ), has delivered an unusually encouraging combination of tolerability and survival data in patients with advanced cancers, including metastatic uveal melanoma, one of the most treatment-resistant malignancies in oncology. The study, published in Nature Communications, enrolled 44 patients between February 2020 and December 2024 and tested continuous daily oral dosing of the drug across a dose-escalation phase and a subsequent expansion in metastatic uveal melanoma. Unlike earlier PI3Kδ inhibitors, which competed directly with ATP for the kinase’s binding pocket and ultimately ran into prohibitive toxicity, roginolisib locks PI3Kδ into an inactive conformation through allosteric modulation, a mechanism that recent structural work has confirmed as conformation-selective rather than classical non-ATP-competitive inhibition.
The biological rationale for targeting PI3Kδ in solid tumours rests on two complementary mechanisms. Within tumour cells, PI3Kδ signalling supports proliferation in a subset of highly transformed malignancies, and the PIK3CD gene encoding the kinase is expressed in both immune and tumour cells in uveal melanoma, at levels comparable to PIK3CA. Outside the tumour cell, PI3Kδ is essential for the activation of regulatory T cells (Tregs) and contributes to the function of myeloid-derived suppressor cells, both of which build an immunosuppressive shield around tumours. Inhibiting this tumour-extrinsic role has shown potent anti-tumour effects, particularly in combination with immune checkpoint inhibitors. First-generation ATP-competitive inhibitors such as idelalisib demonstrated that blocking PI3Kδ can deplete Tregs while sparing cytotoxic T cells, but their poor selectivity, arising from the conserved nature of the ATP-binding pocket, produced gastrointestinal, hepatic and immune toxicities that curtailed their development in solid tumours.
The trial was designed around a predictive pharmacokinetic-pharmacodynamic model rather than a conventional maximum-tolerated-dose strategy. Part A escalated doses from 10 to 80 milligrams once daily in 24 patients with solid tumours or follicular non-Hodgkin lymphoma, informed at each step by plasma drug levels and by a standardised basophil activation test, which measures PI3Kδ-dependent CD63 expression on basophils as a functional readout of target inhibition. Part B then treated 20 mostly pre-treated patients with metastatic uveal melanoma at 80 milligrams, the dose predicted to maintain plasma concentrations above the inhibitory concentration for 90 percent of PI3Kδ activity (IC90) continuously. Steady-state trough concentrations of roughly 1,980 nanograms per millilitre comfortably exceeded the whole-blood IC90, and pharmacodynamic inhibition of CD63 plateaued at the 80-milligram dose, confirming that the biologically effective dose had been reached.
The safety results stand in sharp contrast to the history of the drug class. No dose-limiting toxicities were observed at any dose level and no maximum-tolerated dose was declared. Only three of 44 patients, or 6.8 percent, experienced drug-related grade 3 adverse events, and these were transient laboratory abnormalities, including decreased platelet and neutrophil counts and elevated lipase, that resolved while patients continued treatment. There were no serious infections, no severe immune-related adverse events, and no need for anti-infective prophylaxis, all of which had been major concerns with first-generation PI3Kδ inhibitors. Total CD4, CD8, NK and B lymphocyte counts and plasma immunoglobulin levels remained stable throughout. Compliance exceeded 97 percent, median exposure reached 144.5 days overall and 221 days for uveal melanoma patients at 80 milligrams, and only one patient discontinued treatment, for a grade 2 uveitis in a patient with similar symptoms on prior pembrolizumab.
Patient-reported outcomes reinforced the clinical picture. Using the PRO-CTCAE questionnaire, 16 of the 20 uveal melanoma patients in the expansion cohort completed serial assessments with a return rate of 88.5 percent. Rather than worsening over time, symptoms improved by week 16 in the domains of fatigue, abdominal pain and pain interference, with symptom deterioration reported only later and coinciding with disease progression. A complementary exploratory analysis using a Burden of Therapy methodology, which weights adverse events by severity and timing, showed no accumulation of toxicity during continuous 80-milligram dosing, with the toxicity profile comparable to that seen at 10 to 40 milligrams. The authors note this is the first time PRO-CTCAE has been applied to capture the patient experience with a PI3K inhibitor.
The pharmacokinetic profile proved equally favourable. Roginolisib was rapidly absorbed, with peak plasma concentrations one to two hours after dosing, exposure increased approximately proportionally across the 20 to 80 milligram range, and the half-life of roughly 24 hours at 80 milligrams supports once-daily continuous administration with a low peak-to-trough ratio. High-sensitivity mass spectrometry detected no reactive metabolites, and each metabolite accounted for less than 6 percent of the total metabolite profile. The investigators argue that three attributes underpin the safety record: the conformation-selective binding mode that stabilises two unique residues in the C-terminal helix of the kinase domain, a pharmacokinetic profile that avoids the peak concentrations at which co-inhibition of PI3Kα and PI3Kβ, and the associated gastrointestinal and hepatic toxicities, would occur, and a clean drug-metabolism profile free of active or reactive metabolites.
Anti-tumour activity was modest by conventional response criteria but notable in its survival implications. Two partial responses were recorded among 32 patients with solid malignancies, one in a patient with metastatic cutaneous melanoma who remains on treatment after more than 71 months of daily dosing, and one in a uveal melanoma patient. Among four follicular lymphoma patients treated at 80 milligrams, one achieved a confirmed partial response by Lugano criteria. More strikingly, in metastatic uveal melanoma, a disease with a historical median overall survival of roughly seven months in the second- and third-line setting, the median overall survival was 20.8 months at the December 2023 data cut-off and 18.4 months at the February 2026 cut-off. This pattern of low response rates paired with unexpectedly long survival mirrors the profile of tebentafusp, the bispecific immunotherapy approved for this disease, and underscores the well-recognised discordance between radiographic response and survival benefit in uveal melanoma.
An exploratory post hoc analysis identified week 16, the time of the second tumour scan, as a potential early predictor of benefit. Among 27 evaluable uveal melanoma patients, 13 had stable disease at week 16 and nine had progressive disease. Patients with stable disease at that landmark had a median overall survival of 28.5 months, compared with 12.3 months for those who had progressed. Supporting correlative studies painted a consistent molecular picture. Radiomic analysis of serial CT scans revealed mixed responses, with some pre-existing lesions shrinking or disappearing while new ones emerged, and identified the quartile coefficient of dispersion, a measure of voxel-intensity variability, as a feature distinguishing responding from non-responding liver metastases. Cell-free DNA trended downward in patients with stable disease and upward in those who progressed, and on-treatment tumour biopsies from patients with stable disease showed downregulation of G-protein-coupled receptor pathways, including those driven by the GNAQ and GNA11 mutations characteristic of uveal melanoma.
Immune monitoring by mass cytometry provided perhaps the most direct evidence of the drug’s mechanism in patients. Regulatory T cells declined over the course of treatment, while activated CD8-positive CD39-positive T cells, a population associated with response to immune checkpoint blockade, increased, and the magnitude of that increase correlated positively with overall survival. Immunosuppressive CD5-positive B cells, CD8-positive CD39-negative T cells and CD56-positive CD16-negative NK cells all decreased, more pronouncedly in patients with stable disease. Plasma proteomics using the Olink platform detected significant changes in 83 of 3,072 measured proteins, including rises in the lymphocyte-activating cytokines IL-15, IL-17D and soluble IFNGR2, a fall in soluble PD-1, and declines in CCL22, a chemokine that recruits Tregs, and in soluble CD5 and CD5-ligand, mirroring the reduction of CD5-positive B cells in blood. Multiplex immunohistochemistry on paired tumour biopsies confirmed a shift in the CD8-to-Treg ratio in favour of cytotoxic T cells within five weeks of starting treatment.
The authors are careful to acknowledge the limitations inherent in a single-arm first-in-human study with a small sample size and exploratory biomarker analyses, and reservations remain about whether roginolisib has sufficient single-agent activity in metastatic uveal melanoma. To address this, a randomised phase 2 study, OCULE-01, has completed recruitment, randomising patients regardless of HLA status after progression on prior immunotherapy to roginolisib or investigator’s choice, with survival as the primary endpoint. A broader clinical programme is also testing roginolisib in combination with tebentafusp in uveal melanoma, with the PD-1 inhibitor dostarlimab in non-small-cell lung cancer, with venetoclax and anti-CD20 antibodies in chronic lymphocytic leukaemia, and with JAK inhibitors in myelofibrosis. If the randomised data confirm the survival signals seen here, roginolisib could become the first PI3Kδ inhibitor to deliver the class’s long-promised immunomodulatory benefit without the toxicity that has repeatedly derailed it.
Subject of Research: First-in-human clinical trial of the conformation-selective PI3Kδ inhibitor roginolisib in advanced and metastatic cancer, including metastatic uveal melanoma
Article Title: The conformation-selective PI3Kδ inhibitor roginolisib in patients with advanced or metastatic cancer, including metastatic uveal melanoma: first-in-human clinical trial
Article References: Di Giacomo, A. M., Simonelli, M., Losurdo, A., Spiliopoulou, P., D’Alonzo, V., Amato, G., Valente, M., Barcenilla, H., Johnsson, A., Gonzalez, L., Tan, Z., Tadepally, L., Brodin, P., Lahn, M., Di Conza, G., Zorrilla, R., van der Veen, L., Bevilacqua, A., Kaur, P., … Maio, M. (2026). The conformation-selective PI3Kδ inhibitor roginolisib in patients with advanced or metastatic cancer, including metastatic uveal melanoma: first-in-human clinical trial. Nature Communications, 17(1), Article 10219. https://doi.org/10.1038/s41467-026-77358-7
Image Credits: AI Generated
DOI: 10.1038/s41467-026-77358-7
Keywords: roginolisib, PI3Kδ inhibitor, metastatic uveal melanoma, first-in-human trial, regulatory T cells, immunotherapy, targeted therapy, pharmacokinetics, overall survival, biomarkers, mass cytometry, Nature Communications
News Source: Nathaniel Bowman. (October 11, 2026). Conformation-Selective PI3Kδ Inhibitor Roginolisib Shows Safety and Survival Signals in First-in-Human Cancer Trial. Scienmag.
