Feasibility and workflow analysis of confocal microscopy in robot‐assisted prostatectomy

Objectives To assess the diagnostic accuracy, workflow feasibility, and downstream compatibility of fluorescence confocal microscopy (Histolog® Scanner [HS]; SamanTree Medical SA, Lausanne, Switzerland) compared with frozen section (FS) for intraoperative margin evaluation (IME) during nerve‐sparing robot‐assisted radical prostatectomy (RARP). Patients and Methods Between May 2024 and April 2025, 68 patients undergoing RARP for non‐metastatic prostate cancer were enrolled; two were excluded due to incomplete data, leaving 66 evaluable cases. The IME was performed using both HS and FS. HS images were interpreted intraoperatively by experienced uropathologists, while surgical decisions were based on FS results. The residual tissue following HS was assessed for compatibility with downstream immunohistochemistry (IHC) and next‐generation sequencing (NGS). Diagnostic performance was evaluated against whole‐mount histology as the reference standard. Results The HS identified carcinoma in four cases (three true positives, one false positive) and negative margins in 55 cases (53 true negatives, two false negatives); seven cases (10.6%) were classified as suspicious and resolved by FS. Considering only definite HS calls, sensitivity was 60.0%, specificity 98.1%, positive predictive value (PPV) 75.0%, and negative predictive value (NPV) 96.4%. When suspicious cases were treated as positive, sensitivity increased to 75.0%, specificity decreased to 91.4%, and PPV to 54.5%, while NPV remained 96.4%. HS processing did not impair IHC or NGS analyses. The overall R1 resection rate was 9.1%. Conclusion Fluorescence confocal microscopy demonstrates high specificity and NPV for IME during RARP. A negative HS result can be interpreted with high confidence, potentially reducing the need for FS. Positive or suspicious HS findings should prompt targeted FS confirmation. Confocal laser scanning microscopy preserves tissue integrity for molecular testing and may be integrated into multimodal oncological workflows.