KEY TAKEAWAYS
- The IMpower110 phase III trial aimed to compare digital and manual PD-L1 TC (SP263) scoring for identifying NSCLC pts who may benefit from atezolizumab.
- OS and PFS were compared between manual and digital PD-L1 scoring, with a subgroup analysis stratified by histology.
- The study revealed digital PD-L1 scoring’s superiority over manual scoring, highlighting a potential benefit for atezolizumab therapy.
Intratumoral heterogeneity and interobserver variability complicate the consistent assessment of tissue-based immune biomarkers like PD-L1, necessitating artificial intelligence and digital pathology. Researchers aimed to compare digital and manual PD-L1 TC (SP263) scoring for identifying NSCLC patients(pts) who may benefit from atezolizumab.
The study compared first-line atezolizumab to chemotherapy in metastatic NSCLC pts with PD-L1 expression ((SP142; PD-L1 ≥1% on TC or IC). PD-L1 quantification on SP263-stained slides utilized a clone-agnostic AI model (AIM-PD-L1; PathAI, Inc) for digital scoring (range 0%-100%) on TC.
OS and PFS based on PD-L1 expression were analyzed, comparing manual and digital scoring with a histology-stratified sub-analysis. Algorithm-derived human interpretable features (HIFs) were investigated for their association with survival and identification of novel biomarkers.
The dataset comprised 509 slides (350 nonsquamous, 123 squamous, and 36 indeterminate histology). Digital scoring, compared to manual scoring, identified 114 and 36 additional PD-L1+ pts at the 1% and 50% cutoff, respectively. Despite differences in PD-L1+ prevalence, OS and PFS analysis demonstrated similar efficacy for manual and digital scoring at both cutoffs (see table).
AIM-PD-L1 facilitated continuous scoring, revealing consistent survival benefits in squamous samples independent of PD-L1, while nonsquamous samples exhibited progressively improved survival with increasing PD-L1 expression. Investigation of HIFs in the tumor microenvironment revealed that in the squamous subtype, patients with a higher density of PD-L1+ lymphocytes in the cancer epithelium exhibited improved PFS in the atezolizumab group.
The study found that digital PD-L1 scoring was superior to manual scoring and identified a subset of squamous histology patients with elevated PD-L1+ lymphocyte density who may benefit from atezolizumab therapy.
Source: https://cattendee.abstractsonline.com/meeting/10925/presentation/831
Clinical Trial: https://clinicaltrials.gov/study/NCT02409342
Herbst RS, Ruderman D, Conway J, Prizant H, Hennek S, Shamshoian J, Abel J, Montalto M, Beck A, Wapinski I, Molinero L, Amin R, Hoang T, Ballinger M, de Marinis F, Giaccone G, Jassem J, Giltnane J, Srivastava MK, Spigel DR. Comparison of Digital vs Manual PD-L1 Tumour Cell Scoring on SP263-Stained Whole Imaging Slides from IMpower110. Presented at: 2023 World Conference on Lung Cancer; September 11, 2023; 2:42 PM-2:52 PM.
