Diagnostic accuracy of artificial intelligence-assisted 18f-fdg pet/ct for predicting pathological complete response to neoadjuvant chemotherapy in breast cancer: a systematic review and meta-analysis.

Authors

Abstract

We conducted a systematic review and meta-analysis to assess the diagnostic accuracy of artificial intelligence (AI)-assisted 18 F-FDG PET/CT for predicting pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) in breast cancer. A comprehensive search of PubMed, Embase, and Web of Science was conducted for studies, with a cutoff date of August 29, 2025, and updated on October 16, 2025. The QUADAS-2 technique and Grading of Recommendations Assessment, Development and Evaluation framework were employed to evaluate study quality. Diagnosis accuracy was aggregated utilizing a bivariate random-effects model. A total of 49 studies involving 3038 patients were included. The Spearman rank correlation coefficient for AI was determined to be 0.159 (P = 0.662). The pooled sensitivity, specificity, PLR, NLR, DOR of AI-assisted 18 F-FDG PET/CT for predicting pCR to NAC in breast cancer were 0.82 (95% CI 0.76-0.87), 0.83 (95% CI 0.75-0.89), 5.03 (95% CI 3.79-6.69), 0.39 (95% CI 0.31-0.49), and 17.71 (95% CI 10.37-30.25), respectively. Furthermore, the AUC was determined to be 0.83 (95% CI: 0.80-0.86). The Fagan nomogram indicated a positive likelihood ratio of 52% and a negative likelihood ratio of 6%. This meta-analysis demonstrates that AI-assisted 18 F-FDG PET/CT shows good diagnostic accuracy for predicting pCR to NAC in breast cancer, achieving better sensitivity and specificity than MRI and ultrasound, and comparable accuracy to conventional PET/CT with improved specificity. These findings highlight its potential as a reliable tool to aid clinical decision-making, though moderate heterogeneity underscores the need for standardized methods and multicenter prospective validation.

Tags

Topics