Deep learning on CT scans to predict checkpoint inhibitor treatment outcomes in advanced melanoma
Laurens S Ter Maat1, Rob A J De Mooij2, Isabella A J Van Duin3
1Image Sciences Institute, University Medical Center Utrecht, Utrecht University, Utrecht, The Netherlands.
Abstract:
Immune checkpoint inhibitor (ICI) treatment has proven successful for advanced melanoma, but is associated with potentially severe toxicity and high costs. Accurate biomarkers for response are lacking. The present work is the first to investigate the value of deep learning on CT imaging of metastatic lesions for predicting ICI treatment outcomes in advanced melanoma. Adult patients that were treated with ICI for advanced melanoma were retrospectively identified from ten participating centers. A deep learning model (DLM) was trained on volumes of lesions on baseline CT to predict clinical benefit. The DLM was compared to and combined with a model of known clinical predictors (presence of liver and brain metastasis, level of lactate dehydrogenase, performance status and number of affected organs). A total of 730 eligible patients with 2722 lesions were included. The DLM reached an area under the receiver operating characteristic (AUROC) of 0.607 [95%CI 0.565-0.648]. In comparison, a model of clinical predictors reached an AUROC of 0.635 [95%CI 0.59 -0.678]. The combination model reached an AUROC of 0.635 [95% CI 0.595-0.676]. Differences in AUROC were not statistically significant. The output of the DLM was significantly correlated with four of the five input variables of the clinical model. The DLM reached a statistically significant discriminative value, but was unable to improve over known clinical predictors. The present work shows that the assessment over known clinical predictors is an essential step for imaging-based prediction and brings important nuance to the almost exclusively positive findings in this field.
Insights
Deep learning on CT scans did not improve prediction of advanced melanoma treatment response compared to clinical factors alone. Integrating imaging data with clinical predictors is essential for accurate outcomes in melanoma immunotherapy.
Area of Science:
- Oncology
- Radiology
- Artificial Intelligence
Background:
- Immune checkpoint inhibitors (ICIs) are effective for advanced melanoma but have significant toxicity and cost.
- Biomarkers for predicting ICI response in melanoma are currently lacking.
- Deep learning (DL) on CT imaging is explored for predicting treatment outcomes.
Purpose of the Study:
- To evaluate the predictive value of deep learning models applied to CT imaging of metastatic lesions for ICI treatment outcomes in advanced melanoma.
- To compare DL model performance against established clinical predictors.
- To assess the benefit of combining DL models with clinical predictors.
Main Methods:
- Retrospective analysis of 730 advanced melanoma patients from ten centers treated with ICI.
- A deep learning model (DLM) was trained on metastatic lesion volumes from baseline CT scans.
- The DLM was compared and combined with a clinical predictor model (liver/brain metastasis, LDH, performance status, organ count).
Main Results:
- The DLM achieved an AUROC of 0.607, while the clinical model achieved 0.635.
- The combined model showed no significant improvement over the clinical model alone (AUROC 0.635).
- DLM output correlated with clinical variables, demonstrating discriminative value but not outperforming clinical predictors.
Conclusions:
- Deep learning on CT imaging alone did not enhance prediction of ICI response in advanced melanoma.
- Clinical predictors remain crucial for predicting treatment outcomes.
- Integrating imaging-based assessments with clinical factors is vital for nuanced prediction in melanoma immunotherapy.
Related Concept Videos
01:28Imaging Studies for Cardiovascular System V: CT
01:27Imaging Studies III: Computed Tomography