Pan-tumor harmonization of pathologic response assessment for standardized data collection in neoadjuvant trials...

Pathologic response is an emerging surrogate marker for therapeutic efficacy and long-term patient outcomes. Updated guidelines for pan-tumor pathologic response assessment have recently been adopted for ongoing clinical trials and routine clinical care. The goal of this study was to prospectively evaluate the interobserver variability among pathologists in assessments of treatment response across tumor types, anatomic locations, and specimen types.

A multi-institutional, international study led by the Society for Immunotherapy of Cancer was carried out to assess the concordance of pathologic response assessment using the pan-tumor criteria in neoadjuvant-treated resection specimens and on-treatment biopsies. Online lecture-based modules for scoring were developed, and 14 pathologists from multiple institutions were trained. The pathologists then assessed 42 specimens representing 12 different tumor types (total of n = 362 hematoxylin-eosin-stained slides) for the three tissue classes that are assessed: %residual viable tumor (RVT), %regression, and %necrosis. Pathologists were also surveyed regarding interest in and barriers related to pan-tumor pathologic response assessment, as well as quality and effectiveness of the training materials.

Scoring of pathologic response using the pan-tumor system was highly reproducible, irrespective of disease location (primary tumor versus lymph node) or specimen type (resection versus biopsy), with intraclass correlation coefficients (ICCs) > 0.8 for %RVT, %regression, and %necrosis. Subset analyses also showed strong reproducibility of %RVT within almost all individual tumor types evaluated (ICC all ≥ 0.86). The poststudy survey completed by the participating pathologists was used to refine the training materials, and the revised modules are provided as a resource to the wider pathology and oncology communities.

This highly reproducible scoring system enables quantitative assessment of treatment response in pathology specimens across multiple tumor types, anatomic sites, disease stages, and therapies, akin to RECIST for radiographic assessment. We identified potential barriers to implementation and highlight strategies to overcome these challenges.