Dual-layer spectral detector CT-based scoring system for noninvasive TP53 expression assessment and adjuvant chemotherapy response prediction in gastric cancer a multicenter cross-sectional study.

Background: No-invasive early prediction of TP53 is a key strategy for improving the prognosis of gastric cancer (GC) patients. To establish a novel, noninvasive and simple scoring system using dual-layer spectral detector computed tomography (DLCT) for preoperative prediction of TP53 expression, prognosis and response to adjuvant chemotherapy (ACT) with retrospective and prospective validation.

Methods: Between April 2021 and March 2025, 568 GC patients were retrospectively and prospectively recruited from two hospitals into a training cohort (TC), a validation cohort (VC), an internal test cohort (ITC), and an external test cohort (ETC). A nomogram prediction model was constructed based on clinical characteristics and DLCT quantitative parameters, which was further simplified to a scoring system. The performance of the system was assessed by discrimination, calibration and clinical applicability. The patients' recurrence-free survival and benefit of ACT were evaluated by survival analysis.

Results: The nomogram outperformed clinical, conventional CT and DLCT models for TP53 prediction. The scoring system exhibited comparable predictive efficacy to the nomogram. The areas under the curve of the scoring system were 0.864 (0.815-0.914) for TC, 0.887 (0.802-0.972) for VC, 0.893 (0.844-0.942) for ITC, and 0.912 (0.838-0.985) for ETC, respectively. There were differences in RFS between patients with different TP53 types predicted by the scoring system (P = 0.006, 0.018, 0.015), in consistent with the results from truth labels. RFS differed between the patients who received the system-recommended ACT regimen and those who did not (P = 0.006, 0.027, 0.032). Furthermore, the scoring system was superior to other currently used testing techniques by using cost-effectiveness analysis (incremental cost-effectiveness ratio and incremental cost-utility ratio were 0).

Conclusion: The simple DLCT-based scoring system enables noninvasively, cost-effectively and rapidly predict TP53 expression, prognosis, and benefit from ACT in GC patient, which is expected to guide individualized treatment.