# Deep learning-based assessment of PD-L1 expression in NSCLC predicts outcome for patients treated with anti-PD-1 immunotherapy

## TL;DR

A deep learning model analyzing PD-L1 immunohistochemistry slides can predict outcomes for lung cancer patients receiving anti-PD-1 therapy better than traditional methods.

## Contribution

A novel deep learning approach to PD-L1 IHC slides identifies histological patterns that improve outcome prediction in NSCLC patients.

## Key findings

- Deep learning groups (DLHigh vs. DLLow) showed significantly longer progression-free survival in both training and validation cohorts.
- DL classification provided prognostic information beyond conventional PD-L1 tumor proportion scores.
- Combining DL groups with clinical variables improved prediction of progression-free survival compared to clinical features alone.

## Abstract

PD-L1 expression is widely used as a predictive biomarker for anti-PD-1 therapies in non-small cell lung cancer (NSCLC). However, its prognostic value remains controversial. Here, we investigated whether deep learning (DL) applied to PD-L1 immunohistochemistry (IHC) slides could identify histological patterns predictive of outcome in patients treated with anti-PD-1 therapy.

We analyzed two independent NSCLC cohorts: MSK (n=182, training) and CGFL (n=108, validation). Tumor regions were manually annotated, tiled, stain-normalized, and processed through the UNI foundation model to extract deep features. Clustering of tiles from 10 extreme-outcome MSK cases identified histology-based subgroups. These were then applied to the remaining patients by projection and majority voting. Associations with progression-free survival (PFS) and overall survival (OS) were assessed. DL groups were integrated with clinical covariates in a multivariate model.

Clustering revealed two distinct DL-defined groups (DLHigh vs. DLLow). In the MSK cohort, DLHigh patients had significantly longer PFS than DLLow (median 5.7 vs. 2.5 months; HR = 0.63, 95% CI 0.44–0.89; p=0.01). This prognostic value was independently confirmed in the CGFL cohort (median PFS 15.2 vs. 6.2 months; HR = 0.59, 95% CI 0.36–0.96; p=0.03). OS was numerically higher in DLHigh patients but did not reach significance. DL classification correlated with higher PD-L1 tumor proportion score (TPS). Discordance between DL and TPS was observed, and the DL model further stratified outcomes among patients with TPS ≥50%. A combined model integrating DL groups with clinical variables improved prediction of PFS compared to clinical features alone (HR = 0.50, 95% CI 0.33–0.75; p<0.001 in MSK; HR = 0.54, 95% CI 0.31–0.91; p=0.02 in CGFL).

Deep learning applied to PD-L1 IHC slides identifies reproducible histomorphological patterns associated with outcomes in anti-PD-1–treated NSCLC patients. This approach provides prognostic information beyond conventional PD-L1 scoring and enhances predictive accuracy when combined with clinical factors.