Beyond Claims and EHR: Imaging-First MASH Patient Identification

Imaging RWD identified 8X more MASH patients than structured data alone.

Metabolic dysfunction-associated steatohepatitis (MASH), formerly known as nonalcoholic steatohepatitis (NASH), has quickly become one of the most important disease areas in hepatology research and biopharma development. The 2023 shift in nomenclature from NAFLD/NASH to MASLD/MASH, along with the 2024 FDA approval of resmetirom (Rezdiffra), have accelerated clinical focus and commercial investment across the space.  

Despite growing awareness and momentum, identifying patients with MASH in traditional real-world data sources remains a challenge. At the same time, prevalence is projected to grow by more than 60%, increasing from an estimated 14.9 million U.S. adults in 2020 to 18.4 million by 2030.1

Claims and EHR datasets often rely on diagnosis coding behavior, structured documentation, or provider-entered disease labels. In clinical practice, however, liver disease is frequently identified through imaging findings long before definitive MASH coding appears in structured records. As a result, many MASH patients remain effectively invisible to traditional real-world evidence methodologies. 

Imaging Captures the Disease, Not Just the Encounter

Traditional claims and EHR approaches are valuable for understanding healthcare utilization, diagnoses, and treatment patterns. But they are inherently dependent on coding behavior and structured documentation. 

Imaging tells a different story. 

In MASH, imaging is often the clinical source of truth, capturing categories of data that cannot be replicated through traditional real-world data sources. Radiology reports, elastography measurements, MRI-PDFF assessments, and fibrosis staging frequently reveal disease presence and progression long before a patient receives a definitive diagnosis code. 

This distinction matters because MASH is historically undercoded. Even published claims-plus-EHR methodologies acknowledge that relying on structured diagnosis data may undercount the true patient population. Imaging-first approaches help address that limitation by identifying evidence of disease directly from radiology and imaging biomarkers. 

This creates an opportunity to identify patients who may otherwise be missed, while also enabling a deeper understanding of fibrosis progression, steatosis burden, and longitudinal disease evolution.

Building an Imaging-First MASH Cohort at Scale

Using an imaging-first methodology aligned to contemporary MASLD/MASH clinical definitions, Avandra has identified a large imaging-evident MASH-spectrum cohort across its federated imaging network — uncovering approximately 8X more potential MASH patients than structured data approaches alone.

Patients in our strict cohort are identified through radiology report findings of quantitative imaging biomarkers including MRI-PDFF 2 , MR elastography 3 and FibroScan VCTE/CAP 4 , liver-context-validated fibrosis staging; and explicit MASLD/MASH/NAFLD/NASH disease terminology in the radiology report. The strict cohort intentionally requires high-specificity imaging evidence - these criteria are individually specific for MASH-spectrum disease in clinical practice, as radiologists do not document PDFF percentages, stage liver fibrosis, or assert MASH terminology incidentally. 

Our broader MASLD/MASH-spectrum cohort is identified through radiology report findings consistent with hepatic steatosis 5 , fibrosis or cirrhosis morphology, or MASLD/MASH/NAFLD/NASH disease terminology in the impression. Inclusion criteria align with the AASLD/EASL/ALEH 2023 MASLD consensus definition, where imaging-evident steatosis is an operationally definitional inclusion criterion 6. Alternate-etiology exclusion criteria - viral hepatitis, alcoholic liver disease, autoimmune hepatitis, and hemochromatosis - follow established conventions in published MASH real-world evidence cohort identification methodology 7. 

Both cohorts apply negation-language detection (e.g., "no evidence of steatosis," "without fatty infiltration") to prevent false-positive matches. Critically, the methodology is independent of K75.81 ICD coding or EHR-recorded MASH diagnoses. 

The current workstream includes two complementary cohort definitions: 

• Broad imaging-evident MASH-spectrum cohort (~450,000 patients):
  Includes patients with imaging findings consistent with steatosis, fibrosis, cirrhosis, or other MASH-spectrum terminology, while excluding alternate liver disease etiologies. Within this population, approximately 170,000 patients have longitudinal imaging documentation (≥ 2 imaging studies), enabling natural history modeling of disease progression and real-world treatment-response analyses.
  

• High-confidence strict cohort (~12,000 patients):
  Includes imaging-evident MASH-spectrum patients supported by quantitative biomarkers (MRI-PDFF, MRE, FibroScan), validated fibrosis staging, or explicit disease terminology in radiology reports. This cohort is well suited for trial enrichment, biomarker validation, and AI/ML training applications.
  

Together, these cohorts demonstrate the scale and methodological flexibility of imaging-first research: broad population reach for epidemiology and burden-of-disease analyses, paired with a rigorously defined high-confidence sub cohort for precision applications including pharma trial enrichment and quantitative biomarker research. 

Importantly, the findings reinforce a growing challenge in MASH research: many patients with imaging-evident disease may never be fully captured through diagnosis codes or structured documentation alone. By incorporating imaging directly into cohort identification, Avandra identified substantially more patients than traditional structured data methodologies. 

These patient records represent imaging-evident disease, quantitative biomarker evidence, and longitudinal follow-up, all critical components for modern MASH research.

Aligning with the Future of MASLD/MASH Research

As clinical definitions continue to evolve, imaging is becoming increasingly central to how MASLD and MASH are identified and studied. Researchers and biopharma organizations increasingly need datasets that move beyond diagnosis codes to include imaging evidence, quantitative biomarkers, and longitudinal disease context. 

Imaging is becoming a foundational layer of next-generation real-world evidence, helping researchers better identify patients, characterize disease progression, and accelerate discovery in MASH and beyond. 

Interested in exploring imaging-first real-world data for MASH, longitudinal biomarker analysis, or AI-driven research? Connect with the Avandra team to learn more. 

1 https://liverfoundation.org/liver-diseases/fatty-liver-disease/nonalcoholic-steatohepatitis-nash/

2 Reeder SB et al., J Magn Reson Imaging 2011; PMID 22025886; Tang A et al., Radiology 2015; PMID 25247408

3 Singh S et al., Clin Gastroenterol Hepatol 2015; PMID 25305349

4 Karlas T et al., J Hepatol 2017; PMID 28039099; Eddowes PJ et al., Gastroenterology 2019; PMID 30689971

5 Hernaez R et al., Hepatology 2011; PMID 21618575

6 Rinella ME et al., Hepatology 2023; 78(6):1966-1986; DOI 10.1097/HEP.0000000000000520

7 Kim Y et al., ISPOR 2024; EPH50