Immune checkpoint inhibitors (ICIs) have changed outcomes across many cancers, but primary and acquired resistanceremain the dominant barrier to durable benefit. A major biologic driver of resistance is TGFβ-mediated immune evasion, which can promote immune exclusion, suppress effector function, and enrich immunosuppressive populations in the tumor microenvironment. Prior attempts to block the TGFβ pathway broadly have been limited by dose-limiting toxicities, in part because pan–TGFβ inhibition disrupts homeostatic roles of TGFβ2/3. This trial evaluates a more selective strategy: targeting latent TGFβ1 only, aiming to relieve immunosuppression while avoiding the toxicity associated with global pathway blockade.
Study Design and Methods
Trial: Multicenter, open-label, phase 1 (DRAGON trial; SRK-181-001)
Agent: Linavonkibart (SRK-181), first-in-class fully human selective anti–latent TGFβ1 monoclonal antibody (IgG4)
Combination partner: Pembrolizumab (anti–PD-1)
Structure (three parts):
- Part A1: Dose escalation, linavonkibart monotherapy
- Part A2: Dose escalation, linavonkibart + anti–PD-(L)1 (combination escalation; pembrolizumab included)
- Part B: Dose expansion, linavonkibart + pembrolizumab in tumor-specific cohorts
Key population (Part B): Patients with advanced solid tumors resistant to prior anti–PD-1 therapy
- Linavonkibart 1,500 mg Q3W (or 1,000 mg Q2W) + pembrolizumab 200 mg Q3W
- Expansion cohorts included ccRCC, melanoma, HNSCC, urothelial cancer, NSCLC, plus a small tumor-basket cohort.
Primary objective: Safety/tolerability (mono and combo)
Secondary objectives: PK, anti-drug antibodies, antitumor activity (Part B), plus biomarker/PD correlates
Results
Safety and tolerability
Selective inhibition of TGFβ1 proved clinically manageable across dose levels. No cytokine release syndrome was observed, and treatment administration was not interrupted in any patient. In the dose-escalation phase, no dose-limiting toxicities or grade 4/5 treatment-related adverse events were reported, even at the highest tested doses. In the expansion cohort, overall toxicity largely mirrored expectations for pembrolizumab-based therapy. The main additional signal was dermatologic toxicity, which emerged as the most consistent incremental risk. Although grade ≥3 events occurred in approximately one-third of patients, grade 4 events were rare and included isolated cases of exfoliative dermatitis, myositis, and pneumonitis. Importantly, no treatment-related deaths were reported. Discontinuations due to adverse events occurred in a meaningful minority, reflecting the clinical fragility typical of heavily pretreated, immunotherapy-refractory populations. Overall, the safety profile contrasts favorably with historical experience of non-selective TGFβ blockade, where systemic toxicity has limited feasibility.
Antitumor activity in anti–PD-1–resistant disease
Clinical activity was observed despite prior resistance to PD-1 inhibition, with the strongest signal emerging in clear cell renal cell carcinoma. In this cohort, objective responses reached 20%, including a complete response, while melanoma demonstrated a similar but slightly lower response rate. More modest activity was seen in head and neck and urothelial cancers, and no responses were observed in NSCLC within this dataset. Importantly, responses—particularly in ccRCC—were often durable, with treatment durations extending beyond what would typically be expected from PD-1 continuation after documented progression. In a setting where immune rechallenge is usually ineffective, these findings suggest a true biological effect rather than random variability.
Immune modulation and translational correlates
Correlative analyses support a mechanism of immune “re-entry” rather than simple immune activation. Treatment was associated with increased CD8⁺ T-cell infiltration and higher proportions of activated cytotoxic T cells in responders. Responding tumors also demonstrated a favorable shift in immune balance, characterized by a lower ratio of regulatory T cells to activated CD8⁺ cells, while patients with progressive disease showed the opposite pattern. In parallel, reductions in circulating granulocytic myeloid-derived suppressor cells were observed in many responders. Together, these findings align with the hypothesis that selective TGFβ1 blockade alleviates dominant immunosuppressive pressures within the tumor microenvironment, allowing PD-1–mediated immunity to regain functional relevance.
Biomarker-enriched activity in ccRCC
Exploratory biomarker analyses suggest that benefit may be restricted to tumors with a pre-existing but constrained immune contexture. In ccRCC, responses clustered in patients with higher baseline CD8⁺ infiltration, elevated regulatory T-cell presence, and increased TGFB1 expression. Conversely, tumors lacking these features did not demonstrate objective responses in this dataset. While hypothesis-generating, these observations raise the possibility that patient selection based on immune and TGFβ-related biomarkers could meaningfully refine future development strategies.
Insights
- Targeting latent TGFβ1 (not pan-TGFβ) is the central innovation: it aims to preserve TGFβ2/3 homeostatic roles and avoid historic toxicity pitfalls while still dismantling a key resistance axis.
- The study supports a biologic model where TGFβ1 blockade converts “suppressed-inflamed” tumors into “productive-inflamed” tumors, improving CD8 function and reducing suppressive pressures (Tregs, gMDSCs).
- ccRCC stands out as the leading signal, and the biomarker story there is unusually coherent for a phase 1 expansion dataset.
Key Takeaway Messages
- Linavonkibart is a first-in-class selective anti–latent TGFβ1 antibody designed to overcome ICI resistance without pan–TGFβ toxicity.
- In ICI-resistant advanced solid tumors, linavonkibart + pembrolizumab showed manageable safety, with dermatologic toxicity as the main added risk signal.
- Efficacy was most compelling in ccRCC (ORR 20% including a CR), with smaller signals in melanoma and limited activity in HNSCC/urothelial cancer; no responses were seen in NSCLC in this report.
- Translational data suggest increased CD8 infiltration/activation and a more favorable effector-to-suppressor balance, supporting on-mechanism activity.
- Biomarkers (baseline CD8/Treg/TGFB1) may help identify patients most likely to benefit—particularly in ccRCC.
Conclusion
This phase 1 DRAGON study provides early clinical proof that selective latent TGFβ1 inhibition can be combined with PD-1 blockade with a feasible safety profile and meaningful activity in PD-1–resistant disease, most notably ccRCC. The biomarker signal suggests this may become a biomarker-guided salvage strategy rather than an indiscriminate add-on. The next step is confirmatory development in larger, more controlled cohorts to validate durability, define the true response ceiling, and formalize a patient selection algorithm.
You Can Read All Article Here