An integrated radiomics-pathomics approach for predicting efficacy and prognosis of immunotherapy in advanced non-small cell lung cancer

To develop a multimodal model for predicting progression-free survival (PFS) and survival probability of immunotherapy in patients with advanced non-small-cell lung cancer (NSCLC). This retrospective study included 182 inoperable NSCLC patients treated with immune checkpoint inhibitors. The primary endpoint was defined as durable clinical benefit (DCB, PFS ≥ 6 months) and non-durable clinical benefit (NDB, PFS < 6 months). Overall survival (OS) was the secondary endpoint for long-term efficacy. Clinical characteristics were selected using univariate and multivariate regression analyses. Radiomics features were extracted from pre-treatment computed tomography (CT) images, and pathomics features were extracted from whole slide images. A multimodal integrated model was constructed by combining clinical characteristics, radiomics-pathomics features, and utilizing multiple machine learning classifiers. The performance of the model was evaluated by plotting the receiver operating characteristic (ROC) curves and calculating the area under the curves (AUC). Kaplan-Meier survival analysis and Cox regression analysis were applied to analyze OS. The multimodal model using Logistic regression classifier demonstrated better predictive performance in distinguishing DCB from NDB than any unimodal models (including the clinical, radiomics and pathomics model), with AUC of 0.90 (95%CI: 0.82–0.98) in the test cohort. Kaplan-Meier analysis and Cox regression analysis revealed significant survival differences between DCB and NDB groups stratified by the multimodal model. Two nomograms were developed to predict DCB and survival probabilities, respectively. The combination of radiomics and pathomics features showed complementary value in predicting the efficacy and prognosis of immunotherapy, potentially improving individualized treatment in NSCLC. • Developing a new biomarker to predict the efficacy and prognosis of immunotherapy in advanced non-small cell lung cancer. • The multimodal combination of radiomics and pathomics features effectively improved the performance of the unimodal prediction models. • Two clinically applicable nomograms were established to predict immunotherapy efficacy and survival probability.