AI Cannot Yet Replace Lymph Node Surgery After Negative Prostate PET

A systematic scoping review published in Cancer Imaging concludes that current artificial intelligence evidence does not support omitting extended pelvic lymph-node dissection after a negative prostate-specific membrane antigen PET scan. The study, led by Patricia Rodríguez-Parras and colleagues, examined the potential of AI, machine learning, and radiomics to refine the risk of occult pathological node-positive disease, known as pN1, in prostate cancer patients. A negative prostate-specific membrane antigen (PSMA) PET scan (miN0) does not exclude microscopic pelvic nodal metastasis.

The review followed the PRISMA-ScR guidelines and searched PubMed/MEDLINE and Scopus from January 1, 2018, to August 19, 2026. The search was supplemented by OpenAlex sensitivity searching and backward citation screening. After deduplication, the team screened 496 database records and assessed 99 full reports. Ultimately, 79 AI and radiomics reports were included in the analysis, comprising 29 nodal-prediction studies, 22 PET detection or surrogate-reference studies, and 28 prognostic reports. A separate conventional-model search retained 55 PSMA/MRI-aware comparators.

A central limitation identified in the review is the lack of dedicated validation in miN0 cohorts. No AI study included in the review provided direct decision evidence in a dedicated miN0 population. Four PSMA-PET AI studies used patient-level surgical pN1, but none was miN0-specific. The only explicit external validation of PSMA-radiomics showed a significant drop in performance, with the area under the curve falling from approximately 0.88 to 0.57.

The risk of bias in the included studies was a major concern. Overall risk of bias was high in 25 of the 29 nodal reports, driven mainly by issues in analysis and participant selection. Specifically, 25 of the 29 studies had a high risk of bias in analysis, and 23 of the 29 had a high risk of bias in participant selection. Threshold consequences and absolute pN1 misses were rarely reported.

The review emphasizes the need for dedicated miN0 validation, calibrated probabilities, prespecified thresholds, and prospective decision-impact assessment.

In conclusion, the current AI evidence does not support the omission of extended pelvic lymph-node dissection after a negative PSMA PET scan.

Subject of Research: Artificial intelligence for occult pN1 risk and extended pelvic lymph-node dissection selection in prostate cancer

Article Title: Beyond a negative PSMA PET: artificial intelligence for occult pN1 risk and extended pelvic lymph-node dissection selection in prostate cancer—a systematic scoping review

Article References: Rodríguez-Parras, P., Zambudio-Munuera, A., Millán-Ramos, I., Gutiérrez-Tejero, F., Morales-Martínez, A., Jiménez Pacheco, A., & Arrabal Polo, M. Á. (2026). Beyond a negative PSMA PET: artificial intelligence for occult pN1 risk and extended pelvic lymph-node dissection selection in prostate cancer—a systematic scoping review. Cancer Imaging. https://doi.org/10.1186/s40644-026-01129-2

Image Credits: AI Generated

DOI: 10.1186/s40644-026-01129-2

Keywords: prostate cancer, PSMA PET, artificial intelligence, lymph node dissection, radiomics, systematic review, clinical decision-making, oncology, Beyond, negative, PSMA, artificial