Intraoperative margin assessment using fluorescence confocal microscopy during robot‐assisted radical prostatectomy: a first experience

Abstract

Objective

To evaluate the feasibility of adopting fluorescence confocal microscopy (FCM) for intraoperative assessment of positive surgical margins (PSMs) during robot‐assisted radical prostatectomy (RARP) in a real‐world clinical setting.

Patients and Methods

We conducted a prospective observational study including consecutive patients undergoing RARP between October 2023 and November 2024 in whom intraoperative margin assessment was performed using FCM (Histolog® Scanner; SamanTree Medical SA, Lausanne, Switzerland). After exclusion of initial learning‐curve cases and non‐interpretable scans, a per‐patient analysis was performed. Prostate specimens were scanned ex vivo intraoperatively, and digital images were reviewed remotely and in real time by an experienced uropathologist blinded to final pathology. Final histopathological evaluation served as the reference standard. Diagnostic performance metrics included sensitivity, specificity, positive predictive value (PPV), negative PV (NPV), and overall accuracy, assessed both for any PSM and for clinically significant PSMs (≥3 mm).

Results

A total of 101 patients were included in the final analysis. The median (interquartile range) overall time from prostate extraction to intraoperative report was 13 (6–16) min. At final pathology, 17 patients (17%) had PSMs. For detection of any PSM, FCM demonstrated a sensitivity of 88.2%, specificity of 91.7%, PPV of 68.2%, and NPV of 97.5%, with an overall diagnostic accuracy of 91.1%. When restricting the analysis to clinically significant PSMs (≥3 mm), diagnostic performance improved, with sensitivity, specificity, PPV, and NPV of 93.8%, 91.8%, 68.2%, and 98.7%, respectively, and an overall accuracy of 92.1%.

Conclusions

Intraoperative surgical margin assessment using FCM is feasible and demonstrates good diagnostic accuracy compared with final histopathology, with rapid turnaround times compatible with routine surgical workflows. Diagnostic performance improves when clinically relevant PSM length thresholds are applied. Larger multicentre studies and randomised trials are warranted to determine whether FCM‐guided intraoperative decision‐making can improve oncological and functional outcomes after RARP.

Abbreviations

FCM

fluorescence confocal microscopy

IFS

intraoperative frozen‐section

IQR

interquartile range

NeuroSAFE

neurovascular structure‐adjacent frozen‐section examination

NSS

nerve‐sparing surgery

NVB

neurovascular bundle

PSM

positive surgical margin

(N)(P)PV

(negative) (positive) predictive value

(RA)RP

(robot‐assisted) radical prostatectomy

Introduction

Radical prostatectomy (RP) was historically associated with high rates of urinary incontinence and erectile dysfunction [1]. Indeed, these adverse outcomes primarily result from surgical trauma to the plexus of parasympathetic nerve fibres (collectively known as the neurovascular bundle [NVB]), which lies in microscopic proximity to the posterolateral surface of the prostatic capsule [2]. More recently, the adoption of robot‐assisted RP (RARP) has enabled refined nerve‐sparing surgery (NSS) achieving robust oncological control while improving postoperative recovery of continence and erectile function [3, 4]. However, NSS may increase the risk of positive surgical margins (PSMs) and, consequently, the likelihood of biochemical recurrence, the need for adjuvant therapy, increased healthcare costs, and reduced quality of life [5, 6]. In this context, intraoperative assessment of surgical margins may serve as a valuable strategy to mitigate this risk by allowing immediate secondary resection when a PSM is detected [7]. Furthermore, recent results from the neurovascular structure‐adjacent frozen‐section examination (NeuroSAFE) trial demonstrated that this approach increases the proportion of patients eligible for NSS and enhances postoperative erectile function, particularly among individuals who would not have undergone bilateral NSS based on standard preoperative criteria [8, 9, 10]. Recently, fluorescence confocal microscopy (FCM) has gained increasing interest for its ability to provide real‐time assessment of surgical margins during RARP. In this study, we report our initial experience with FCM (using the Histolog® Scanner; SamanTree Medical SA, Lausanne, Switzerland), evaluating its diagnostic accuracy for intraoperative margin assessment during RARP in a real‐world setting, using final histopathology as the reference standard.

Patients and Methods

We conducted a prospective observational study including all patients who underwent RARP at our institution between 12 October 2023 and 26 November 2024, in whom intraoperative surgical margin assessment was performed using FCM (Histolog Scanner).

Patient Selection and Preoperative Evaluation

All procedures were performed by highly experienced surgeons, each with a caseload >1000 RARP procedures. Before surgery, each patient's case (including preoperative MRI) was jointly reviewed with the primary urologist. Patients were considered for inclusion if they were deemed suitable candidates for NSS, based on tumour characteristics, imaging findings, and clinical judgement. The use of any robotic system or surgical approach was permitted. In order to minimise the initial learning curve effect, the first 50 cases were excluded. Patients without a Histolog report or a final pathological report were excluded from the analysis. Additionally, cases lacking images of the posterolateral aspect of the prostate, as well as those with inadequately scanned or poorly interpretable images that precluded pathological evaluation, were excluded from the analysis.

Intraoperative Assessment of Prostate Specimen

Prior to the start of the procedure, an Alexis® retractor (Applied Medical, Rancho Santa Margarita, CA, USA) was placed in all patients to facilitate standardised specimen extraction and extracorporeal handling.

At the end of the extirpative phase (before commencing the vesicourethral anastomosis) the prostate specimen was removed from the operative field and processed for intraoperative imaging (Fig. 1). The method used for specimen preparation prior to scanning has been described in detail elsewhere [11]. The prostate was then positioned on a dedicated support and scanned using the Histolog Scanner, with systematic evaluation of all six surfaces of the prostate specimen scanned in the following sequence: posterior, left posterolateral, anterior, right posterolateral, apex and base. Image acquisition was performed by trained operators. Upon completion of the scanning process, the digital images were immediately transmitted to an experienced uropathologist, who performed the intraoperative margin assessment in real‐time and remotely, therefore blinded relative to final pathology. The findings were communicated via telephone to the operating surgeon.

Fig. 1.

Study workflow for intraoperative FCM during RARP.

Integration of FCM Findings Into the Intraoperative Workflow

Although FCM results were provided during the procedure, this was a feasibility study designed to evaluate the diagnostic performance of FCM. For this reason, all clinical decisions (including whether to perform additional resections of the NVB or adjacent tissues) remained entirely at the surgeon's discretion and were not mandated by the FCM report.

Statistical Analysis

A per‐patient analysis was performed to evaluate the diagnostic performance of intraoperative FCM margin assessment. The primary objective of this study was to test the diagnostic performance of FCM for PSM detection. For each case, the intraoperative evaluation provided by the operator was compared with the final histopathological report, which served as the reference standard (Figs 2 and 3). Both intraoperative FCM assessment and final pathology classified margins strictly as positive or negative. In cases where a PSM was identified, the linear extent of the margin (in millimetres) was recorded. Diagnostic performance metrics included sensitivity, specificity, positive predictive value (PPV), negative PV (NPV), as well as overall accuracy. Specifically, diagnostic performance analyses were conducted for accuracy in detecting any PSM and detecting PSM of known clinical significance (≥3 mm). Descriptive statistics were used to summarise baseline demographics and perioperative characteristics. Categorical variables were reported as frequencies and percentages, whereas continuous variables were presented as median values with interquartile ranges (IQRs). All tests were two‐sided, with a significance level of P < 0.05. R software environment for statistical computing and graphics (R, version 4.2.2; R Foundation for Statical Computing, Vienna, Austria) was used for all analyses [12].

Fig. 2.

The PSMs identified by Histolog image assessment and final pathology. Full Histolog Scanner field‐of‐view image of prostatic surfaces are shown with the cancerous infiltrates delineated within the red annotations (5 mm scale bar). Inserts are showing higher magnification of the area delineated within the black frame (0.5 mm scale bar). Top: cancerous infiltrates with glandular morphology. Bottom: higher grade of poorly differentiated cancerous infiltrates that do not present glandular morphology.

Fig. 3.

Histolog image of a PSM not reported by the final pathology assessment. Full Histolog Scanner field of view of a prostatic surface with a small cancerous infiltrates shown by the red annotation (5 mm scale bar). Insert is showing higher magnification of the area delineated within the black frame. Small cancerous infiltrates with glandular morphology are seen with an overall area size of <2 mm (0.5 mm scale bar).

Results

Between October 2023 and November 2024, 127 patients underwent RARP with intraoperative margin assessment with FCM at our institution. Among these, 26 of 127 (20.4%) patients were excluded due to inadequate or non‐interpretable image acquisition. Among the 26 excluded patients, 13 (50.0%) involved the apex, nine (34.6%) the base, and four (15.4%) the posterolateral surface. This resulted in 101 patients in the final cohort. The number of scanned surfaces per case ranged from two to six. Specifically of these 101 patients, 92 (91.1%) had six surfaces scanned, three (3.0%) had five surfaces, two (2.0%) had four surfaces, three (3.0%) had three surfaces, and one (1.0%) had two surfaces scanned (Table S1).

Descriptive Characteristics of the Study Population

The median (IQR) age was 65 (60–70) years and the median (IQR) PSA level was 6.8 (5.1–9.0) ng/mL. At MRI, the median (IQR) prostate volume was 41 (33–52) mL, while 11 of 101 (10.8%) patients were classified as locally‐advanced (stage T3a). The median scanning time was ~1 min per face, resulting in a total scanning time of 2–8 min depending on the number of faces scanned (Table S1). The median (IQR) overall time from prostate extraction to report was 13 (6–16) min. At final pathology, 17 of 101 (17%) of patients showed PSM. These and all other patient characteristics are listed in Table 1.

Table 1.

Baseline characteristics of the 101 patients who underwent RARP and who received intraoperative margin assessment with Histolog at our institution between October 2023 and November 2024.

Characteristic	Value
Number of patients	101
Age, years, median (IQR)	65 (60–70)
Body mass index, kg/m2, median (IQR)	26.5 (24.2–29.0)
ASA score, median (IQR)	2.00 (2.00–2.00)
Charlson Comorbidity Index score, median (IQR)	3.00 (2.00–3.00)
Preoperative PSA level, ng/mL, median (IQR)	6.8 (5.1–9.0)
Prostate volume at MRI, mL, median (IQR)	41 (33–52)
Intraoperative reporting time, min, median (IQR)	13 (6–16)
Previous BPH surgery, n (%)	3 (3)
Previous abdominal surgery, n (%)	26 (26)
Extracapsular extension at MRI, n (%)	9 (9)
Seminal vesicle invasion at MRI, n (%)	4 (4)
Pathological T stage, n (%)
pT2	71 (70)
pT3	30 (30)
Gleason Grade Group at final pathology, n (%)
ISUP 1	12 (12)
ISUP 2	58 (57)
ISUP 3	27 (27)
ISUP 4	2 (2)
ISUP 5	2 (2)
PSM at final pathology	17 (17)
PSM at FCM	22 (22)
Diagnostic performance, n (%)
False negative	2 (2.0)
False positive	7 (7)
True negative	77 (76)
True positive	15 (15)

ISUP, International Society of Urological Pathology.

Diagnostic Performance Considering any PSM

When considering any PSM, ex vivo FCM assessment using the Histolog Scanner identified 15 true positives, 77 true negatives, seven false positives, and two false negatives. Using the final histopathological report as the reference standard, the sensitivity, specificity, PPV, and NPV were 88.2%, 91.7%, 68.2%, and 97.5%, respectively, with an overall diagnostic accuracy of 91.1% (Table 2).

Table 2.

Diagnostic performance of FCM.

Definition of PSM	Sensitivity, %	Specificity, %	PPV, %	NPV, %	Accuracy, %
Overall cohort	88.2	91.7	68.2	97.5	91.1
>3 mm	93.8	91.8	68.2	98.7	92.1

Diagnostic Performance for Clinically Relevant PSM Detection (>3 mm)

The prevalence of patients with a PSM ≥3 mm was 16% (16/101). When considering clinically significant PSMs (≥3 mm), ex vivo FCM assessment identified 15 true positives, 78 true negatives, seven false positives, and one false negative. Using final histopathology as the reference standard, sensitivity, specificity, PPV and NPV were 93.8%, 91.8%, 68.2%, and 98.7%, respectively, with an overall diagnostic accuracy of 92.1% (Table 2).

Discussion

Intraoperative surgical margin assessment during RARP has gained increasing attention in recent years. Notably, results from the NeuroSAFEPROOF trial (ClinicalTrials.gov identifier: NCT03317990) clearly demonstrated that the use of intraoperative margin evaluation to guide NSS leads to improved postoperative erectile function compared with standard practice [8]. Importantly, this approach also enabled bilateral NSS in selected patients who, based on conventional preoperative criteria, would not have been considered suitable candidates [8]. In this context, the recent introduction of FCS offers a valid alternative for intraoperative margin assessment by offering a rapid, high‐resolution imaging of fresh, unprocessed tissue. In this study, we evaluated our initial experience of FCM using the Histolog Scanner, comparing intraoperative findings with the final pathological report as the reference standard, and made several noteworthy observations.

First, compared with more established and validated techniques for intraoperative margin assessment (e.g., NeuroSAFE), FCM represents a simpler and more user‐friendly approach for several reasons. A key advantage of FCM is the speed and ease of sample processing. Indeed, the prostate specimen is scanned en face without the need for sectioning or complex tissue preparation, which are time‐consuming and require specialised expertise, all factors that represent major limitations to the widespread adoption of frozen‐section‐based techniques such as NeuroSAFE [8, 9, 13]. In contrast, FCM preparation is considerably simpler and less time intensive. Turnaround time is a critical aspect when comparing FCM with other intraoperative margin assessment methods. It has been extensively demonstrated that NeuroSAFE requires about 30–60 min to provide a margin assessment, a delay that may limit its applicability in many real‐world surgical settings [8, 9, 11, 13]. Conversely, in the present study we observed a median image acquisition time of 7 min, followed by an additional 2–11 min for pathological interpretation. These findings are highly consistent with previously published data, where highly magnified images are produced within few minutes [14, 15, 16]. Moreover, performing FCM within the operating theatre and in close interaction with the operating surgeon may facilitate a more precise spatial understanding of the location of a PSM within the prostate bed, compared with frozen‐section analysis, where the specimen is sectioned and evaluated remotely in a pathology laboratory. Additionally, FCM offers the possibility of remote pathological assessment through digital image sharing. Unlike conventional frozen‐section analysis, which requires physical transport of the specimen to the pathology laboratory, FCM allows real‐time evaluation via a remote viewer. This approach reduces logistical constraints, shortens turnaround time, and may be particularly beneficial in centres where geographical factors limit immediate access to on‐site pathology services.

Second, we then focused on the diagnostic performance of FCM using the Histolog system. When considering the overall cohort (PSMs vs negative surgical margins), sensitivity, specificity, PPV, and NPV were 88.2%, 91.7%, 68.2%, and 97.5%, respectively, with an overall diagnostic accuracy of 91.1%. These findings are partially consistent with the existing literature. For example, Mayor et al. [13, 17] recently reported a multicentre prospective study showing, on a per‐patient basis, a sensitivity of 48%, specificity of 94%, PPV of 79% and NPV of 80%. The higher sensitivity observed in our series (93.8% vs 88.2%) may reflect multiple factors, including the learning curve associated with FCM interpretation and inherent differences between confocal imaging and conventional histopathology. Moreover, variations in surgical technique (particularly the extent of diathermy use and the consequent thermal artefacts) may influence image quality and interpretation. These findings underscore the importance of standardised surgical handling and evaluation by a dedicated uropathologist with specific expertise in FCM. Furthermore, given that posterolateral margins are sampled in cross‐section at 5‐mm intervals in a single plane, tumour involvement may be missed if it does not intersect the sampled boundary, irrespective of its linear extent [11]. In a similar study, Baas et al. [16] prospectively collected posterolateral surfaces with reported sensitivity and specificity on a per‐surface level of 86% and 99%. Conversely, some other studies compared FCM with intraoperative frozen‐section (IFS) analysis. Almeida‐Magana et al. [18] performed a blinded comparison between FCM and paraffin‐embedded histopathological analysis on posterolateral prostate surfaces. The authors reported high diagnostic accuracy and substantial inter‐rater agreement for the detection of PSMs, with sensitivities ranging from 73% to 91% and specificities from 94% to 100%. Notably, the median length of PSMs in their study was nearly threefold greater than that observed in the present cohort, reflecting selection of patients felt to be likely to have capsular involvement based on MRI, rather than all comers. Similarly, Musi et al. [19] reported that, in the ex vivo FCM setting, agreement between ex vivo FCM and IFS analysis ranged from moderate (κ = 0.62) to strong (κ = 0.86), depending on the diagnostic category. In addition, inter‐observer agreement among pathologists ranged from moderate (κ = 0.74) to almost perfect (κ = 0.90).

In general, the available evidence on FCM is based on small cohorts and often lacks histopathological blinding, whereas the vast majority of evidence on intraoperative margin assessment is derived from the NeuroSAFE technique [7, 8, 9, 11, 20, 21, 22, 23, 24, 25, 26], in which intraoperative margin evaluation has been associated with improved functional outcomes [8]. In theory if FCM can detect significant margins reliably then secondary resection of NVB can be avoided appropriately and the functional benefits observed in the NeuroSAFEPROOF study from increased nerve sparing would be expected to be replicated. Ideally, after an initial phase of methodological standardisation and learning‐curve maturation, a prospective randomised trial should evaluate whether FCM can provide similar functional benefits using a less expensive and less time‐consuming approach.

When considering only PSMs ≥3 mm as positive, diagnostic performance improved, with sensitivity, specificity, PPV, and NPV of 93.8%, 91.8%, 68.2%, and 98.7%, respectively, and an overall diagnostic accuracy of 92.1%. Although the superior accuracy we observed with longer PSMs is limited, it is in keeping with consistent observations within the published literature. Consistent with these observations, the IP8‐FLUORESCE study (International Standard Randomised Controlled Trial Number [ISRCTN] 21536411) demonstrated a marked increase in sensitivity for the detection of PSMs measuring ≥3 mm (79%) [13]. However, performance remained lower than that reported in the largest NeuroSAFE cohorts. This discrepancy likely reflects the limited ability of FCM to identify minimal tumour extensions and emphasises the need to apply PSM length thresholds that are clinically relevant [8]. Importantly, the 3‐mm cut‐off corresponds to the definition of clinically significant PSMs used in the NeuroSAFEPROOF trial, which was associated with improved functional outcomes, although its long‐term oncological implications have yet to be fully elucidated. An additional methodological consideration concerns the definition of PSM when assessed by FCM. Unlike conventional histopathology, where a PSM is reported as a linear measurement (e.g., 3 mm in length) on sectional analysis, FCM provides a two‐dimensional surface evaluation of the sampled margin. As a result, direct equivalence between FCM‐detected margin positivity and a specific linear extent on histopathological sections cannot be assumed. Further research is needed to establish standardised criteria and clinically relevant thresholds that correlate two‐dimensional confocal laser scanning microscopy findings with conventional linear PSM measurements.

Taken together, our findings provide several important observations. Indeed, intraoperative surgical margin assessment using confocal microscopy is technically feasible in a real‐world clinical setting and demonstrates good diagnostic performance when compared with final histopathological evaluation, especially for PSMs ≥3 mm. Although future randomised or large scale multicentre studies are warranted to further validate these findings, the present results support the potential role of FCM as a reliable intraoperative adjunct to guide surgical decision‐making during RARP.

Despite the novelty of our findings, this study has several limitations. First, it is a single‐centre study with a relatively small sample size, which represents a major limitation and is common to many investigations in this field. Second, image interpretation was performed by a single reader, who was an expert uropathologist with extensive experience in FCM reporting. As such, the relatively high diagnostic accuracy observed in this study should be interpreted with caution. Although the initial cases were excluded to mitigate the learning‐curve effect and image interpretation was performed by a dedicated uropathologist with specific expertise in FCM, these factors may have contributed to an optimised diagnostic performance that may not be immediately reproducible in other settings. Furthermore, the single‐centre design and the involvement of a single experienced reader limit the external validity and generalisability of the findings. Variability in surgical technique, specimen handling, and reader expertise across institutions may influence diagnostic accuracy. Therefore, multicentre studies including multiple readers with varying levels of experience are warranted to assess reproducibility and broader applicability. Third, a per‐surface analysis was not feasible due to the retrospective design of the study. Precise anatomical matching between the exact areas evaluated by FCM and those assessed by final histopathology was not systematically recorded, making reliable surface‐by‐surface concordance analysis impossible. Prospective studies with standardised anatomical mapping are needed to allow accurate per‐surface performance assessment, particularly in comparison with IFS analysis. Finally and importantly, as this was a feasibility study, intraoperative surgical management was not modified on the basis of FCM findings. Specifically, no additional surgical steps (e.g., secondary NVB resection vs nerve‐sparing preservation in the presence of a PSM) were undertaken according to the FCM results. Consequently, no conclusions can be drawn regarding its potential impact on oncological or functional outcomes. Prospective studies specifically designed to evaluate FCM‐driven intraoperative management strategies are warranted.

Conclusion

In this prospective real‐world study, intraoperative surgical margin assessment with FCM using the Histolog system proved to be technically feasible and demonstrated good diagnostic accuracy when compared with final histopathological evaluation. The rapid turnaround time and simplified workflow suggest that FCM may represent a practical alternative to frozen‐section‐based techniques for intraoperative margin assessment during RARP. Diagnostic performance was further improved when clinically significant PSM lengths were considered, supporting the relevance of appropriate PSM length thresholds. Although the present findings are encouraging, larger multicentre studies and prospective randomised trials are warranted to determine whether FCM‐guided intraoperative decision‐making can translate into improved oncological and functional outcomes.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not‐for‐profit sectors.

Disclosure of Interests

Nothing is declared.

Supporting information

Table S1. Number of scanned specimen surfaces and corresponding time required for FCM acquisition for each patient (n = 101). Scanning time was conservatively estimated.