Artificial intelligence is increasingly changing how researchers analyse medical images and evaluate the effectiveness of cancer treatments. New findings from Altis Labs suggest that AI-powered imaging analysis could identify signs of treatment benefit earlier than some conventional imaging endpoints used in oncology clinical trials.
At the 2026 World Conference on Lung Cancer (WCLC) in Seoul, Altis Labs presented results from an independent post-hoc analysis of the Phase 3 MARIPOSA clinical trial, which evaluated treatments for patients with EGFR-mutated advanced non-small cell lung cancer (NSCLC).
The analysis focused on Altis Labs’ Imaging-based Prognostication (IPRO) technology. According to the company, the AI system analysed approximately 10,000 radiology scans and generated prognostic measurements at baseline and during treatment.
The findings indicate that IPRO identified a treatment effect beginning at an early stage of the trial, while the conventional RECIST-based objective response rate (ORR) did not show the same predictive signal for the later overall survival benefit.
What Was the MARIPOSA Clinical Trial?
The MARIPOSA Phase 3 trial, registered as NCT04487080, evaluated first-line treatment strategies for patients with locally advanced or metastatic NSCLC carrying specific EGFR mutations.
The study compared amivantamab (Rybrevant) plus lazertinib (Lazcluze) with osimertinib. The trial has generated important long-term survival data, including statistically significant overall survival results reported by Johnson & Johnson in 2025.
The latest Altis Labs analysis does not replace the established clinical endpoints used in the trial. Instead, it examines whether AI-derived imaging measurements could provide an earlier indication of treatment response and potential long-term outcomes.
This distinction is important because clinical trials can take years to generate mature overall survival data.
How Altis Labs’ IPRO Technology Works
Altis Labs’ IPRO system is designed to analyse complete CT scans rather than focusing only on selected tumour measurements.
Traditional imaging assessments such as RECIST primarily evaluate changes in measurable target lesions. This approach has played an important role in oncology trials, but it does not capture every feature visible across a patient’s medical images.
IPRO takes a broader approach by analysing three-dimensional imaging data and extracting prognostic information associated with survival.
The technology can assess imaging characteristics related to areas such as:
- Tumour burden
- Body composition
- Organ health
- Other prognostic imaging biomarkers
- Changes in a patient’s overall imaging profile
The resulting IPRO-α score is intended to provide an AI-generated prediction related to patient survival.
AI Imaging Detected an Early Treatment Signal
One of the key findings from the MARIPOSA analysis was the difference between IPRO Response Rate and traditional ORR.
For the analysis, IPRO Response was defined as a 50% or greater improvement in the IPRO-α score compared with baseline.
Researchers then calculated an IPRO Response Rate ratio comparing the investigational treatment arm with the control arm at different early imaging assessments.
According to the presented findings, the IPRO Response Rate began favouring the investigational treatment at Week 16 and remained favourable at subsequent assessment points.
This is significant because the AI-derived endpoint showed a treatment signal earlier than the conventional response measurement was able to anticipate the later overall survival benefit.
The results suggest that AI-based imaging could potentially provide researchers with additional information while a clinical trial is still underway.
IPRO Response Was Associated With Overall Survival
Another important observation involved individual patient trajectories.
When researchers examined changes in IPRO scores over time, IPRO deterioration was consistently associated with poorer overall survival, while improvement in IPRO measurements was associated with better survival outcomes.
This relationship is particularly relevant to oncology research because overall survival can require lengthy follow-up.
If an imaging-based biomarker can reliably identify patients who are responding to treatment or experiencing deterioration earlier, researchers may gain a more detailed understanding of how treatment effects develop over time.
However, the findings come from a post-hoc analysis, meaning further research would be required to establish how such an AI endpoint should be incorporated prospectively into clinical trial design.
Why Earlier Treatment Signals Matter in Oncology
Overall survival remains one of the most important outcomes in cancer research because it directly measures how long patients live after treatment.
The challenge is that measuring overall survival generally requires patients to be followed for a substantial period.
Other endpoints, including objective response rate and progression-free survival, can provide earlier information about treatment activity.
AI-powered imaging approaches such as IPRO are being investigated as another potential source of earlier evidence.
The potential value is not simply about replacing established endpoints. Instead, AI imaging could complement existing measurements by examining information within medical scans that may not be captured by tumour-size assessments alone.
Moving Beyond Tumour Size
Conventional tumour response assessment often focuses on whether measurable lesions become smaller, remain stable, or grow.
However, cancer progression and patient survival are influenced by multiple biological and physiological factors.
A patient’s overall condition can involve changes in tumour burden, body composition, organ function, and other characteristics visible on medical imaging.
By analysing an entire CT scan, AI systems can potentially identify patterns that are difficult or impossible to measure manually.
This broader imaging analysis is one of the central ideas behind IPRO.
Rather than simply automating RECIST measurements, Altis Labs describes IPRO as an AI-powered prognostic system that uses imaging information to generate survival predictions.
What the Findings Could Mean for Clinical Trials
The potential impact of AI imaging extends beyond a single oncology study.
Clinical trials require significant amounts of time and resources, and researchers need reliable methods to determine whether a treatment is producing meaningful benefits.
An imaging biomarker capable of providing an early efficacy signal could potentially help researchers make more informed decisions during drug development.
For pharmaceutical companies and clinical research organisations, such technology could also support more detailed patient stratification and analysis of treatment response.
However, AI-derived endpoints need to undergo rigorous validation before they can become widely accepted as regulatory or primary trial endpoints.
Factors such as reproducibility, generalisability across patient populations, imaging protocols, scanners, and clinical settings all need to be considered.
Altis Labs’ View on AI in Cancer Research
Altis Labs founder and CEO Felix Baldauf-Lenschen said the findings demonstrate the potential for AI to identify meaningful clinical benefit that traditional imaging endpoints may not detect as early.
The company is positioning IPRO as a technology that can use medical images already collected during clinical trials, potentially adding another layer of information without requiring entirely new imaging procedures.
Altis Labs has also presented other research involving IPRO in oncology, including work examining relationships between IPRO response and overall survival in advanced NSCLC.
The Future of AI-Powered Medical Imaging
The latest MARIPOSA analysis highlights a broader trend in healthcare: using artificial intelligence to extract more information from existing clinical data.
Medical imaging contains significantly more information than simple measurements of tumour diameter. Modern AI systems can process large volumes of imaging data and identify complex patterns that may provide additional prognostic information.
If validated across different clinical trials and cancer types, AI imaging endpoints could become an increasingly important component of oncology research.
The technology could eventually help researchers understand treatment effects earlier, identify patients at higher risk of poor outcomes, and develop more sophisticated ways to measure response.
At the same time, AI should be viewed as a potential complement to established clinical methods rather than an immediate replacement for them.
Conclusion
Altis Labs’ latest findings from the Phase 3 MARIPOSA trial demonstrate how AI-powered medical imaging could provide earlier insights into treatment effects in oncology.
By analysing approximately 10,000 radiology scans, the company’s IPRO technology generated prognostic measurements that showed an early treatment signal beginning around Week 16. The analysis also found an association between changes in IPRO measurements and overall survival.
The results presented at WCLC 2026 add to growing research into AI imaging biomarkers, predictive oncology, and artificial intelligence in clinical trials.
As researchers continue to validate these technologies, AI-based imaging could play an increasingly important role in helping clinical trial teams understand treatment response and patient outcomes earlier in the drug-development process.
Discover more from AiTechtonic - AI & Informative News
Subscribe to get the latest posts sent to your email.