Only 7.1% of U.S. adults with cancer enroll in treatment trials (JCO, 2024), a bottleneck that slows drug development and limits access to targeted therapies. Proscia, a Philadelphia-based digital pathology company, aims to address that dynamic with its new tool, Aperture, which uses AI to flag potential candidates right at the point of diagnosis by scanning tissue images, biomarkers, molecular data and clinical records.
Why precision medicine complicates recruitment
As cancer therapies target narrower biomarker-defined subgroups, matching the right patient to the right trial has become more complex, with restrictive eligibility and point-of-care visibility gaps documented as contributing factors. “It’s still tough, and in some ways tougher [to recruit in oncology trials], because pharma has shifted toward precision medicine,” said Nathan Buchbinder, co-founder and chief strategy officer of Proscia. He added that precision therapies require precision diagnostics, yet “pharma companies generally don’t have real-time visibility into the diagnostic testing that determines eligibility,” while providers may lack up-to-date awareness of recruiting studies. Those are gaps Aperture aims to close at the point of diagnosis.
How Aperture Works
Data Pipeline: Pathology images and reports are de-identified within the diagnostic workflow and loaded into Snowflake in a standardized schema for query and analysis.
AI Processing: Multiple AI models work in tandem. LLMs extract structure from unstructured text reports, while computer-vision algorithms analyze tissue images that are often one billion pixels each.
Real-time Matching: As new diagnoses are finalized, the platform compares them with active clinical-trial criteria and flags potential candidates at the point of diagnosis.
Network Scale: Proscia’s Concentriq platform connects labs on track to process about 8 million pathology cases in 2025, with access to over 12 million de-identified tissue images linked to clinical and genomic data.
Proscia says Aperture taps its Concentriq lab network, on track for approximately 8 million pathology diagnoses in 2025, and a real-world data repository exceeding 12 million de-identified tissue images paired with clinical and genomic records, to surface eligible candidates and generate evidence for submissions and CDx work. The system applies AI to pathology images along with AI-derived biomarkers, molecular results, and clinical records to provide real-time insights at diagnosis.
The platform’s engine operates on a substantial data pipeline. “Data generated through the routine diagnostic workflow is de-identified and loaded into Snowflake in a standardized data architecture so it’s queryable,” Buchbinder said. He added that a suite of AI models, including LLMs for text and computer vision for images, extracts features to match patients with trial criteria in near-real time.
For drugmakers, Aperture is positioned to accelerate trial recruitment and support regulatory submissions, companion-diagnostic development, label expansions, and payer negotiations. For labs facing shrinking reimbursements and workforce shortages, it creates new revenue streams and elevates pathology from a cost center to a strategic contributor, according to the company.
Proscia lists a partner ecosystem that spans cloud, workflow, enterprise imaging, and AI: AWS (cloud foundation), Agilent (end-to-end workflow distribution), and Siemens Healthineers (enterprise imaging integration), plus specialized AI from Ibex (cancer detection/grading), Visiopharm (e.g., IVDR-cleared PD-L1 for NSCLC and breast IHC apps), Mindpeak (NSCLC PD-L1 scoring/HER2 work), and Owkin (predictive biomarker screening such as MSI).
Linking diagnosis to treatment decisions
Embedding eligibility checks at sign-out expands pathology’s role. Pathology has long underpinned diagnosis while remaining loosely coupled to downstream treatment choices. Aperture links the two by surfacing trial eligibility at the point of diagnosis and treating the report as the start of patient routing.
“We see Aperture as giving pathologists information tied to outcomes and treatments,” Buchbinder said. “They’re not just puzzle-solvers anymore; they’re more integrated into the patient’s care journey. It extends the pathologist’s impact beyond the slide to a more direct connection with therapies.”
The approach fits the specialty’s data-centric work. While many clinical fields rely on patient-reported symptoms and encounters, pathology centers on interpreting complex, objective data. “Pathologists enjoy making sense of complex data, and doing that at scale as part of routine work is compelling,” Buchbinder added.
For labs facing margin pressure from declining reimbursements and staffing shortages, Proscia says Aperture can create new revenue streams by identifying qualified candidates for clinical trials, potentially helping offset costs and bringing more enrollment-related work into the lab.
Making pathology data actionable at scale requires automation. “Pathology data is often unstructured, free-text reports with inconsistent terminology,” Buchbinder noted. “For labs signing out thousands of cases per day, AI is the only scalable way to normalize.”
A billion pixels of potential
Digital pathology adoption is increasing as more labs move from glass slides to digital sign-out, making routinely generated data accessible for analysis.
“Every whole-slide image contains roughly a billion pixels at the heart of how we understand, diagnose, and treat disease,” Buchbinder said. “Letting that data go untapped in routine workflows is a missed opportunity.”
Beyond individual matches, aggregated, de-identified data may support observational analyses of treatment response, help surface candidate biomarkers, and inform trial design. The company says the dataset spans multiple cancer types and demographic groups; the value of any insights will depend on data quality, representativeness, and study methods.
A governance model is central here: participating laboratories retain control over what leaves the institution. Patient privacy is protected through de-identification and, when needed, third-party tokenization to enable longitudinal linkage without exposing personal identifiers.
“Conceptually, there’s value at the intersection of life sciences and diagnostics. When more diagnostic labs go digital, their data can be used to support discovery, development, and delivery,” Buchbinder said.
Drug-development timelines and costs vary by program and indication. If Aperture performs as described, impact should be assessed with concrete measures: time to first patient enrolled, enrollment rates versus baseline, screen-failure rates and reasons, time from diagnosis to randomization, and cohort diversity relative to historical benchmarks.
Filed Under: Uncategorized
