Purpose: Radiotherapy constitutes a cornerstone in the management of hepatocellular carcinoma (HCC), but its efficacy is limited by radioresistance. Sphingolipids, a class of bioactive lipids, have been implicated in the metabolic reprogramming associated with treatment resistance. However, the potential of circulating sphingolipids as non-invasive biomarkers to predict radiosensitivity in HCC patients remains unexplored. Patients and Methods: This prospective study enrolled 61 HCC patients scheduled for radiotherapy (NCT06864221). Pre-treatment plasma samples were analyzed via LC-MS/MS to quantify 13 sphingolipid species. The primary endpoint was objective response rate (ORR) per mRECIST at 12 weeks. Predictive models were developed using multivariate logistic regression with forward selection and LASSO, evaluated by AUC with bootstrap validation, calibration, and decision curve analysis. Longitudinal analysis was performed in a sub-cohort (n=25) with paired pre- and post-radiotherapy plasma samples. Results: The objective response rate was 54.1%. Univariable analysis identified a distinct sphingolipid signature in responders, characterized by significantly lower S1P and higher levels of CER(d18:1/20:0) and CER(d18:1/24:1). These candidate biomarkers, along with significant clinical variables, were entered into multivariate modeling. The optimal integrated model (Model 1), selected via forward selection, comprised S1P, CER(d18:1/20:0), and the clinical factors ALP and TBIL, and excelled at predicting response (bootstrap-corrected AUC=0.930). A second model based on ceramide/S1P balance (CER(d18:1/26:1)/S1P, Total CER(d18:1)/S1P, AFP) also performed robustly (bootstrap-corrected AUC=0.828). Both models showed clinical utility per decision curve analysis. Longitudinal analysis revealed a coordinated metabolic shift in responders, with reduced S1P and elevated CER(d18:1/26:0), supporting a radiation-induced “sphingolipid rheostat” shift toward apoptosis. Conclusion: This exploratory study provides the first clinical evidence that the baseline plasma sphingolipid profile is a potent, non-invasive predictor of HCC radiosensitivity, validating the “sphingolipid rheostat” theory. Our findings establish a framework for sphingolipid-guided precision radiotherapy and lay the necessary groundwork for future large-scale, multi-center validation trials, which hold significant potential to refine patient stratification and advance the development of novel metabolism-targeted interventions. Plain Language Summary: Not all liver cancer patients benefit equally from radiotherapy, and doctors currently lack a good way to predict who will. Our study asked if a simple blood test could provide the answer by measuring specific fat molecules, called sphingolipids. We analyzed blood from 61 patients before their radiotherapy. We discovered that distinct patterns of these sphingolipids could accurately identify who would respond well to the treatment. We even built two prediction models that showed excellent accuracy. Interestingly, in patients who did respond well, we saw a helpful shift in these molecules after treatment: protective signals decreased while those that encourage cancer cell death increased. This means a straightforward blood test could one day help doctors personalize radiotherapy. By predicting a patient’s response in advance, we can better match them with the most effective therapies. This approach could spare those unlikely to benefit from unnecessary side effects, while ensuring those who will respond get the maximum benefit—ultimately leading to more precise and effective cancer care for everyone. Keywords: biomarker, radiotherapy, predictive model, liquid chromatography-tandem mass spectrometry
Yuan et al. (Sun,) studied this question.