Early tumor volume change as a novel CT RECIST indicator for predicting pathological response and prognosis in NSCLC patients undergoing immunotherapy

Fan XU

Abstract: The evaluation of tumor response in non-small cell lung cancer (NSCLC) patients undergoing treatment is crucial for assessing prognosis and guiding therapeutic decisions. Traditional RECIST criteria, based on changes in tumor diameter ($\Delta D$), may not accurately capture tumor response, particularly in cases involving immunotherapy. In this study, we propose an alternative CT RECIST response indicator based on tumor volume change ($\Delta V$) to reflect pathologic response to neoadjuvant chemo-immunotherapy and prognosis in non-small-cell lung cancer (NSCLC) patients. We analyzed 916 tumor lesions using deep learning techniques. By analyzing the relationship between tumor diameter and volume, as well as $\Delta D$ and $\Delta V$, we observed inconsistencies, especially in cases with irregular tumor morphology, leading to inconsistencies between CT RECIST response by $\Delta D$. The response measured by $\Delta V$ demonstrated stronger consistency with pathological responses after neoadjuvant therapy. To avoid excessive evaluation in volume-based assessments, an optimal threshold for $\Delta V$ was selected as $-$0.6868 rather than RECIST-derived threshold of $-$30$\%$. Additionally, $\Delta V$, but not $\Delta D$, was statistically significant in relation to overall survival (P-value = 0.0093). Our findings suggest that volume-based response ($\Delta V$) provides a novel and more precise prognostic indicator than diameter-based methods ($\Delta D$) for assessing early immunotherapy response in both resectable and advanced NSCLC. AI-driven, volume-based assessment may offer a more reliable alternative for CT RECIST response evaluation, potentially improving personalized cancer care. Future research should focus on refining $\Delta V$ thresholds and improving segmentation accuracy for complex cases to support broader clinical application.