Sign Up And Receive Our Latest Promotions!
.svg){kind=small}
How Isarna Therapeutics ran a 5-site, 8-device retinal trial using a same-day data workflow with RetinAI Discovery®
Isarna Therapeutics develops antisense therapeutics for fibrotic and ophthalmic disease. Its lead retina candidate, ISTH0036, targets the TGF-β pathway implicated in fibrosis-driven vision loss in wet AMD and diabetic macular edema, a novel mechanism of action.
To test ISTH0036 in a Phase 2a trial - the BETTER Study - Isarna needed to enroll and monitor 60 patients across 5 sites, using 8 different imaging devices, without slowing enrollment or compromising the consistency of the trial's primary endpoints.
Isarna ran the study on RetinAI Discovery®, giving sites and the reading center a same-day view of every scan and a standardized read on fluid and central retinal thickness (CRT) across all devices, with dedicated support from our in-house Project Management, Research, and Customer Success teams throughout the study.
The Challenge
Proving a novel mechanism requires evidence that standard imaging endpoints can’t provide
Anti-VEGF therapy is the standard of care for wet AMD and DME. A meaningful share of patients develop fibrosis as their disease progresses and stops responding. ISTH0036 was designed to address that gap directly, which meant Isarna's Phase 2a program had to do something harder than the average trial: generate credible, quantitative evidence of an antifibrotic effect, not just track visual acuity and thickness.
That's a measurement problem before it's anything else. Fibrosis shows up on OCT as hyperreflective material (HRM), a marker that's genuinely difficult to segment and volumetrically quantify by hand, and even harder to quantify consistently when the images come from 8 different acquisition devices (Heidelberg Spectralis, Zeiss Plex Elite, Zeiss Clarus 6000, Optovue, HRA Spectralis, Haag-Streit, Topcon Triton, Eidon FA CenterVue) across 5 sites and 6 required imaging modalities per patient.
If HRM volume is read inconsistently between devices or sites, any treatment effect gets buried in measurement noise or errors, and a novel mechanism like ISTH0036's doesn't get the clean, defensible evidence it needs to move forward.
This is the challenge underneath most early-phase trials built on novel or exploratory anatomical biomarkers: the science and data are only as credible as the measurement behind it, and the measurement has to hold up across every device in a multi-site program, not just for one reference instrument.
The Solution
A same-day data evaluation workflow, and one standardized read instead of eight
Multimodal, multi-device biomarker quantification Isarna could build a scientific case on
Isarna ran the BETTER study on RetinAI Discovery® end to end, using it as the measurement backbone for the trial’s imaging-based evidence.
One standardized reading across 8 devices
RetinAI's CE-marked fluid and layer segmentation model harmonized images from all 8 acquisition devices, so HRM, IRF/SRF, and other anatomical markers were volumetrically quantified the same way regardless of which device or site captured the scan, the precondition for treating those measurements as trial-grade, level-one evidence rather than device-specific approximations.
Same-day visibility across sites and the reading center
Images captured at all 5 sites were uploaded to RetinAI Discovery® and available to the study team for review on the same-day, with each site restricted to seeing only its own patients, keeping enrollment and monitoring moving without waiting on a multi-week review cycle.
Together, this gave Isarna the precise, cross-device biomarker data its team needed as the measurement foundation for its scientific conclusions, not just an operational record of the trial.
The Result
Fibrosis biomarker evidence for a first-in-class mechanism, presented at ARVO 2025
Isarna enrolled all 60 patients across its three target populations within the BETTER study's 9-month window, with same-day data review supporting enrollment decisions at every site throughout.
Isarna presented the study’s clinical results at the 2025 ARVO Annual Meeting. Of the 60 patients enrolled in the study, 43 (mean age 76) were included in this ARVO analysis set. Beyond the trial's primary and secondary outcomes - BCVA and central retinal thickness (CRT) - Isarna's team used RetinAI Discovery® to volumetrically assess intraretinal and subretinal fluid (IRF/SRF), hyperreflective material (HRM) consistent with fibrosis, and additional anatomical biomarkers across the study.
Isarna reported the following findings:
- CRT decreased across all nAMD and DME patient groups from baseline to end of study (-29 to -122 µm).
- In nAMD eyes with HRM present at baseline, HRM volume - the marker for retinal fibrosis measured through RetinAI Discovery® - decreased by 70% in ISTH0036-treated eyes, compared to an approximately 75% increase in fellow eyes receiving standard of care.
- DME patients showed mean IRF volume reductions of -333 nL (treatment-naïve) and -120 nL (anti-VEGF pretreated).
- BCVA improved or stabilized across treatment-naïve and pretreated patient groups; IOP remained stable at a mean of 15 mmHg; 16% of treated eyes showed worsening of cataract.
RetinAI's role was providing the volumetric measurements on which Isarna's analysis is based. For Isarna, the HRM finding is notable: a volumetric, image-based signal consistent with the antifibrotic mechanism ISTH0036 is designed to target, the kind of anatomical marker that's difficult to track consistently by hand across a multi-device, multi-site trial, and the exact measurement RetinAI Discovery® was built to standardize.
Isarna's own ARVO 2025 abstract concludes that ISTH0036 "offers the potential to be a first-in-class antifibrotic agent".
The dataset kept working after the trial closed
RetinAI Discovery®'s value didn't end when the trial closed. The full BETTER imaging dataset stayed in the platform, which meant it could be revisited later without re-collecting a single scan. RetinAI is now collaborating with Isarna’s own clinical and medical leadership on a retrospective analysis of that dataset, reprocessing the original BETTER study images with newer RetinAI algorithms developed since the trial concluded.
The goal is to see how these newer models perform against the original human grader assessments on the same images, as one input into understanding how model and biomarker detection capability have changed over time. No comparison has been completed and no results, quantitative or qualitative, are available yet; this is not a claim that the newer models match, exceed, or are validated against human grading. This work is ongoing, and RetinAI is not yet reporting any outcomes from it.
For sponsors, the retrospective illustrates a platform capability worth noting on its own: images captured once in RetinAI Discovery® remain available for later re-analysis as algorithms evolve, without new patient visits, new scans, or new site coordination, subject to the data retention and reuse terms agreed with each customer. A trial's dataset doesn't stop being useful the day the trial locks.
Check out Discovery for Clinical Studies
to Clinical Studies' processes.
Partners
Check the other clinical studies we are involved in:
*Disclaimer
RetinAI Discovery® platform, RetinAI OCT Atlas® and Macula Biomarkers, as well as LuxIA® (Diabetic Retinopathy Screening model) are CE-marked medical devices in accordance with the Medical Devices Regulation (EU) 2017/745 (CE2797). RetinAI Discovery® platform is also a medical device cleared for clinical use by the FDA. The AI modules for biomarkers, fluid, and layer segmentation and quantification in retinal pathologies are designated as Research Use Only (RUO) in the USA. Please be advised that these tools are not intended to replace professional medical advice, diagnosis, or treatment. We do not warrant or guarantee the clinical accuracy, completeness, or usefulness of any content provided by RUO products. For more information, please read our Privacy Notice (https://www.retinai.com/privacy-policy) and Data Processing Agreement (https://www.retinai.com/data-processing-agreement).
RetinAI Discovery® and RetinAI® are trademarks of Ikerian AG.
Ikerian AG (HQ)
Freiburgstrasse 3
3010 Bern
Switzerland
RetinAI US Inc. (US)
6 Liberty Square
#2498 Boston, MA 02109
United States
Imprint/Legal Notice | Data Processing Agreement | Privacy Notice | Created by Ikerian AG. All rights reserved 2016-2026 ©