Toronto, ON:  Altis Labs, Inc. ("Altis") is highlighting results presented at the IASLC World Conference on Lung Cancer (WCLC) 2026 from an independent post hoc analysis of the MARIPOSA Phase 3 trial, evaluating whether Imaging-based Prognostication (IPRO)-α Response Rate can serve as an early endpoint aligned with the trial's overall survival (OS) benefit in patients with EGFR-mutated advanced non-small cell lung cancer (NSCLC).

RECIST v1.1 assessments support early endpoints such as objective response rate (ORR) and progression-free survival (PFS), but do not always reflect the magnitude of long-term clinical benefit. IPRO-α is a fully automated AI model that analyzes an entire CT scan to generate a survival prediction, quantifying imaging biomarkers spanning volumetric tumor burden, body composition, and cardiovascular features.

A fully automated pipeline extracted over 5,000 interpretable features from each of approximately 10,000 baseline and on-treatment CT scans from MARIPOSA (n=858). Evaluable baseline and on-treatment IPRO-α scores were generated for 774 participants (90.2%). IPRO-α response was defined as a ≥50% improvement in IPRO-α score relative to baseline, and the treatment effect was estimated as the IPRO-α response rate ratio for the investigational treatment versus the control arm at landmark imaging assessments every 8 weeks. IPRO-α was trained on a real-world dataset of patients with advanced NSCLC drawn from Canadian health systems.

Key findings:

• A significant IPRO-α response rate ratio favoring the investigational treatment appeared at the second on-treatment assessment, Week 16 (1.33; 95% CI 1.03–1.71), and was maintained through Week 48 (Week 24, 1.25; Week 32, 1.23; Week 40, 1.22; Week 48, 1.22, all significant).
• Point estimates attenuated modestly after Week 16 (range 1.22–1.33) while confidence intervals narrowed progressively, reflecting increasing precision as evaluable data accrued.
• At the trial level, the IPRO-α response odds ratio was 1.38 (95% CI 1.03-1.87), while the RECIST-based ORR odds ratio was 1.15 (95% CI 0.78–1.70) and failed to anticipate the significant OS benefit.
• The Week 16 signal would have preceded the trial's primary PFS readout by 11 months and its final OS readout by 26 months.
• Deterioration in IPRO-α was consistently associated with OS detriment and improvement with OS benefit. Grouping patients into tertiles by IPRO-α score at each assessment separated OS in both treatment arms, confirming IPRO-α as a prognostic factor in its own right.

These results show that an AI-derived imaging endpoint can identify treatment benefit that traditional imaging endpoints do not capture, generated automatically from CT imaging already acquired in the trial. Evidence from additional randomized trials may establish IPRO-α response rate as a surrogate endpoint to support clinical development. Future work will assess the performance of IPRO in additional clinical trials.

About Altis Labs

Altis Labs is the AI endpoint company accelerating clinical trials. Altis was founded on the premise that AI's greatest potential lies in generating better evidence, faster. In close partnership with leading cancer centers, Altis has trained a prognostic AI model, IPRO, on the industry's largest multimodal oncology database comprising longitudinal imaging, molecular, clinical, and outcomes data spanning over 500,000 patient years. Top 20 biopharma companies use IPRO to more confidently analyze data from phase 1-4 clinical trials so that they can bring the most effective novel treatments to patients sooner. For more information, visit www.altislabs.com, follow @AltisLabs on social media, or email info@altislabs.com.

‍