Delphi consensus on ex vivo fluorescence confocal microscopy in robot‐assisted radical prostatectomy

Abstract

Objectives

Positive surgical margins (PSMs) after robot‐assisted radical prostatectomy (RARP) increase the risk of prostate cancer recurrence, often requiring salvage treatments that may compromise functional recovery. Intraoperative frozen section can detect PSM but is not feasible as a routine approach in most settings. Ex vivo fluorescence confocal microscopy (FCM) provides a rapid, real‐time alternative, though evidence remains limited and knowledge gaps persist. This international Delphi consensus aimed to define expert recommendations on the clinical application of and workflow and research priorities for the safe integration of FCM during RARP.

Material and Methods

A modified Delphi process was conducted following the RAND/UCLA Appropriateness Method and reported according to the ACCORD checklist. In total, 32 international experts in urology and pathology participated in a hybrid consensus meeting after structured literature review and preparatory online sessions. Forty‐four evidence‐based multiple‐choice items across patient selection, image acquisition, and surgical management domains were developed and anonymously voted on via a secure online platform. Consensus was predefined as ≥80% agreement among respondents.

Results

Consensus was achieved for 24 of 44 items (55%). Key agreements included a risk‐adapted use of FCM (97%), its role as an adjunct to magnetic resonance imaging for nerve‐sparing decisions (87%), adoption of en‐face imaging as the preferred FCM technique (82%), and a maximum reporting time of 30 min (85%). The panel endorsed formal certification for image interpretation (83%), standardized reporting of positive margin length (82%), and feasibility of remote intra‐institutional reporting (84%). Larger (>3 mm) PSM in high‐risk cases warranted complete neurovascular bundle resection (96%).

Conclusions

This consensus provides a structured framework for the intraoperative use of FCM during RARP, defining its indications, workflow standards, and training requirements. Future multicentre studies are needed to assess its oncological and functional impact and to establish standardized implementation pathways.

Abbreviations

ACCORD

ACcurate COnsensus Reporting Document

DRE

digital rectal examination

EPE

extraprostatic extension

FCM

fluorescence confocal microscopy

IFS

intraoperative frozen section

MCQ

multiple‐choice question

MRI

magnetic resonance imaging

NVB

neurovascular bundle

PCa

prostate cancer

PSM

positive surgical margin(s)

RARP

robot‐assisted radical prostatectomy

SC

steering committee

Introduction

Although the introduction of robotic technology to prostate cancer (PCa) surgery has markedly enhanced surgical precision and visualization, with potential benefits in functional recovery, the occurrence of positive surgical margins (PSMs) continues to represent a major concern, as it may compromise local oncological control and necessitate salvage therapies that can negatively affect functional outcomes [1, 2, 3, 4]. Efforts to balance the need for tissue sparing with PSM rates have prompted the exploration of various intraoperative assessment techniques, among which intraoperative frozen section (IFS) analysis remains the current gold standard [5, 6]. The NeuroSAFE randomized trial has reinforced the clinical value of IFS margin assessment during robot‐assisted radical prostatectomy (RARP) in terms of improving nerve‐sparing decisions and postoperative functional outcomes [7]. However, NeuroSAFE implementation is limited by time‐consuming tissue processing (beyond 45 min) and the need for on‐site expert pathology support and specialist equipment, which restricts its routine use to selected, well‐resourced centres [5].

In recent years, fluorescence confocal microscopy (FCM) has emerged as a potential alternative for real‐time intraoperative margin assessment during RARP [8, 9, 10]. Systems such as the VivaScope® 2500 (Multilaser; MAVIG, Munich, Germany) and the Histolog® scanner (SamanTree Medical SA, Lausanne, Switzerland) allow ex vivo imaging of fresh tissue with minimal preparation, enabling margin evaluation within minutes while preserving specimens for subsequent standard histopathology (Fig. 1) [11, 12]. Preliminary studies have shown promising feasibility, accuracy, and workflow integration [12, 13, 14, 15]. Nevertheless, FCM adoption in PCa surgery remains limited by the lack of regulatory guidance, standardized clinical protocols, and large‐scale evidence supporting its intraoperative use.

Fig. 1.

Schematic representation of intraoperative fluorescence confocal microscopy workflows using Histolog® and VivaScope® 2500 systems, highlighting differences in specimen preparation (intact en‐face vs margin sampling), imaging depth, and real‐time reporting.

To address these gaps, an international modified Delphi consensus was convened, assembling leading experts in urology and pathology with recognized expertise in RARP, intraoperative margin assessment, and FCM. The consensus aimed to establish expert‐driven recommendations on the clinical integration, technical workflow, and research priorities required to ensure that the implementation of FCM is safe, effective, and clinically meaningful in the context of RARP.

Methodology

Consensus Framework

A modified Delphi consensus process was conducted according to the RAND/UCLA Appropriateness Method and reported in compliance with the ACcurate COnsensus Reporting Document (ACCORD) checklist. The study was not prospectively registered. The objective was to achieve expert consensus on the clinical implementation of FCM for intraoperative surgical margin assessment during RARP.

Steering Committee

The process was overseen by a seven‐member steering committee (SC) comprising experts in RARP, pathology, intraoperative margin assessment, and FCM (RDG, GS, KVDS, EB, RAM, MdA, TA‐H). RDG and GS, both robotic surgeons with extensive expertise in intraoperative margin assessment and FCM, chaired the SC, and KVDS served as the lead pathology expert. EB, RAM, MdA, and TA‐H, all robotic surgeons experienced in intraoperative FCM and frozen section assessment, contributed to the literature review, identification of the thematic sessions, development of the Delphi questions and response options, and coordinated meeting organization, in continuous discussion with the SC chairs. The Delphi session was chaired by a psychologist (AG) with expertise in consensus methodology, who did not participate in voting. Surgquest® provided logistical support without involvement in study design or data interpretation. ORSI Academy (Melle, Belgium) hosted the meeting and facilitated participant communication and on‐site organization.

Panel Selection and Recruitment

The SC identified an initial list of 44 potential panellists based on predefined criteria, including recognized expertise in PCa, leadership roles in international urological or pathological societies, and experience with FCM or intraoperative margin assessment during RARP. Invitations were sent 4 months before the meeting, with reminders to non‐respondents; only those confirming participation were included. Ultimately, 32 experts from European, American, and Australian centres, including the seven SC members, took part in the voting process, ensuring balanced representation of urology and pathology and a broad international perspective. Table 1 reports the demographic characteristics of the Delphi voting participants across all voting rounds.

Table 1.

Demographic characteristics of the Delphi voting participants across all sessions.

Characteristics	n (%)
Total voting participants at each session	32 (100)
Specialty
Urology	25 (78.1)
Pathology	7 (21.9)
Geographic region a
Europe	26 (81.3)
North America	4 (12.5)
Australia/Oceania	2 (6.3)
Sex b
Male	28 (87.5)
Female	4 (12.5)

^a Geographic region classified according to the participant's primary institutional affiliation.

^b Sex inferred from publicly available institutional profiles.

Evidence Review and Development of Questions

Before the consensus, a review of the literature was performed to identify all relevant evidence regarding intraoperative surgical margin assessment in PCa, with particular focus on FCM. The review was conducted in PubMed, Embase, and the Cochrane Library from inception to May 2025 using combinations of the following terms: prostate cancer, intraoperative, margin assessment, confocal microscopy, fluorescence confocal microscopy, VivaScope, and Histolog. Reference lists of relevant studies and previously published systematic reviews were screened to ensure completeness. Extracted data included methodological design, imaging system used, diagnostic performance (sensitivity, specificity, predictive values), and intraoperative feasibility. The SC synthesized these findings to identify evidence gaps and guide question formulation.

Eight preparatory online meetings were held to define the thematic framework and develop the modified Delphi questionnaire. Topics were grouped into three domains: (1) patient selection; (2) sample processing, image acquisition, and interpretation; and (3) surgical management. Each domain was divided into subtopics from which multiple‐choice questions were generated. Multiple‐choice questions were used to quantify agreement, with consensus defined a priori as ≥80% selecting the same response. This threshold was selected in line with commonly adopted definitions of consensus in Delphi studies (typically ranging between 70% and 80%) and set at a higher level to ensure more stringent agreement, particularly given the use of dichotomous response options rather than a Likert‐type scale [16, 17]. Items meeting this threshold were converted into final consensus statements, whereas others were classified as ‘no consensus reached.’

Pre‐Meeting Preparation

Before the meeting, all participants received a comprehensive information package that included (i) a curated list of references from the literature review; (ii) the preliminary questionnaire with all proposed questions and response options (Appendix S2); (iii) detailed instructions outlining the modified Delphi procedures, voting methodology, and consensus thresholds (Appendix S3); and (iv) two short instructional videos demonstrating the main FCM systems (VivaScope® and Histolog®) used in PCa management (Videos S1 and S2). All participants thus participated in the meeting with shared technical and methodological understanding.

Delphi Procedure

The hybrid Delphi session took place at ORSI Academy (Melle, Belgium) in September 2025, with 25 participants attending in person and seven joining remotely.

Voting was conducted anonymously via the Vevox® online polling platform (Manchester, UK), which allowed secure access through individualized QR codes, real‐time result visualization, and immediate aggregation of data. Each question was displayed simultaneously on a large screen and on participants' personal devices. Voting rounds were time limited, with a countdown timer ensuring structured participation. Consensus was predefined as ≥80% agreement. Items not meeting this threshold were discussed among panellists and, where appropriate, revised before a second round of voting. Depending on the nature of the disagreement, revisions could involve modifications to the question wording, the response options, or both. Questions that failed to reach agreement after the second round were classified as ‘no consensus reached.’ Attendance was monitored digitally through the Vevox® interface. Participants were invited to vote during each round, and minor fluctuations in participation were accepted, considering the hybrid nature of the Delphi process. Panellists had the option to abstain from voting when they felt insufficiently informed or in case of a conflict of expertise.

To ensure accurate documentation, all sessions were audio‐ and video‐recorded with participants' consent. Recordings were used exclusively for verification and qualitative analysis of discussions and were securely stored in compliance with General Data Protection Regulation regulations.

Results

During the modified Delphi session, two questions were excluded for redundancy. The final questionnaire comprised 44 questions that were submitted for formal voting.

Table S1 summarizes all items included in the modified Delphi process, together with corresponding response distributions, consensus outcomes, and an explanatory legend indicating which items were accepted, revised, or classified as ‘no consensus reached.’

Table 2 reports the final consensus statements resulting from the modified Delphi process, derived from the corresponding questionnaire items, along with the percentage of agreement achieved for each statement during the voting process.

Table 2.

Final consensus statements derived from the modified Delphi process, with corresponding agreement rates for each item.

	Statement	Percentage consensus (%)
	Session 1: Patient selection
01	Patients with high‐risk prostate cancer should be considered eligible for nerve‐sparing surgery	87
02	Patients with high‐risk prostate cancer can be considered eligible for nerve‐sparing surgery when FCM is used for real‐time surgical margin analysis	97
03	Nerve‐sparing surgery can be performed in patients without good preoperative erectile function	81
04	Nerve‐sparing surgery can be performed in patients without good erectile function when FCM is available	97
05	When FCM is available, the potential to improve postoperative continence represents a valid indication to pursue nerve‐sparing surgery	88
06	In clinical T3 prostate cancer, MRI should be used as an adjunct to guide nerve‐sparing during real‐time surgical margin analysis	87
07	Real‐time surgical margin analysis should be reserved for patients with preoperative MRI evidence of extra‐prostatic extension risk or those with high‐risk, high‐volume, or high‐grade disease	86
08	Intrafascial nerve‐sparing should not be routinely attempted in all patients	82
	Session 2: Sample processing and image acquisition and interpretation
09	For intraoperative margin assessment, en‐face imaging of the intact specimen, performed directly on the surface without slicing, is the preferred approach	82
10	A scan of adequate quality is defined by the absence of empty spaces within the imaged area, with at least 90% of the anatomical area visible on the posterolateral surfaces	100
11	A scan of adequate quality is defined by the absence of movement artefacts, with no blurring or distortion acceptable in the central field of view and only minimal artefacts tolerated in the periphery	96
12	A scan of adequate quality is defined by sufficient contrast, allowing clear distinction between anatomical structures	100
13	During real‐time surgical margin analysis, not all prostate areas (six faces) should be analysed	88
14	For intraoperative feasibility, the maximum acceptable time delay between specimen excision and pathologist's FCM report should be within 30 min	85
15	A formal certification should be required to interpret FCM scans	83
16	Both certified surgeons and pathologists are qualified to interpret FCM scans	100
17	When reporting positive surgical margins using en‐face FCM imaging, the lesion's maximum dimension length (mm) should be included	82
18	When multiple positive surgical margins are present on the same aspect of the prostate, each should be reported separately	93
19	Remote reporting of FCM scans is acceptable	84
20	Artificial intelligence can be feasibly applied to FCM for the identification of areas of interest to be assessed by the examiner	93
21	Artificial intelligence can be feasibly applied to FCM for image quality assessment	93
	Session 3: Surgical management
22	The Gleason Score is considered a key determinant in the management of positive surgical margins	89
23	In the absence of Gleason Score information, management of a positive surgical margin on FCM should be guided by the margin's dimension and/or the Gleason Score from the preoperative biopsy	89
24	When a positive surgical margin greater than 3 mm is identified intraoperatively in high‐ or high/intermediate‐risk prostate cancer, full resection should be performed	96

FCM, fluorescence confocal microscopy; MRI, magnetic resonance imaging.

Consensus was reached for 20 questions (46%) during the first round of voting and for four additional questions (9%) after a second round, yielding an overall consensus rate of 55% (24/44). When stratified by domain, agreement was highest in session 1 (Patient Selection), where 8/11 items (73%) reached consensus, including one after revision and re‐voting. In session 2 (Sample Processing, Image Acquisition and Interpretation), 13/24 items (54%) achieved consensus, of which two were after revision and re‐voting. In session 3 (Surgical Management), consensus was reached for 3/11 items (27%), including one after revision and one after the addition of a new item proposed by the panellists.

Discussion

This Delphi process aimed to address key aspects required for a successful implementation of FCM, such as patient selection, sample processing, image acquisition and interpretation, and surgical management. Consensus was reached in 55% of statements. This shows the relative novelty of this technology and the substantial gaps in literature. Nonetheless, the consensus results can act as a framework for planning research and guiding the safe clinical implementation of this technology.

Patient Selection

Robust evidence shows that preserving the neurovascular bundle (NVB) is critical for postoperative recovery, particularly for erectile function and early continence [4]. Although nerve‐sparing is essential for functional recovery, its application in high‐risk disease is often restricted by oncological concerns. Therefore, in the first session, the discussion centred on defining which patients are most suitable for intraoperative FCM. Consensus was achieved regarding the eligibility of patients with high‐risk disease for nerve‐sparing procedures (87%). This increased to 97% if FCM was available for real‐time surgical margin assessment. Although some surgeons noted that their nerve‐sparing strategy would not change with FCM, most agreed that its availability could broaden the use of nerve‐sparing by increasing confidence in intraoperative margin assessment. The role of nerve sparing in patients with impaired baseline erectile function was also discussed, given its potential continence benefits both with and without FCM support [4]. Consensus was reached for offering nerve sparing in this subgroup (81%), rising to 97% when FCM was available. This pattern is consistent with NeuroSAFE data, where IFS assessment increased nerve‐sparing rates from 76% to 93% in pT2 disease and from 55% to 83% in pT3 disease [18]. Although agreement was strong among panellists, several experts noted that reported functional benefits may partly reflect technical differences, particularly at the apex and sphincter dissection, rather than pure NVB preservation [19, 20, 21]. The panel agreed that nerve sparing should remain individualized, based on tumour features, magnetic resonance imaging (MRI) and biopsy findings, and the availability of FCM.

Consensus was reached that FCM‐based real‐time margin analysis should complement MRI when planning nerve sparing in clinical T3 disease. MRI alone was deemed insufficient to guide intraoperative decisions because of its limited sensitivity for extra‐prostatic extension (~0.55) and only moderate concordance with final pathology (~65%) [22]. In contrast, FCM offers cellular‐level resolution and real‐time feedback, positioning it as a complementary rather than competing technique. MRI provides preoperative risk mapping, whereas FCM can refine intraoperative decisions. When discrepancies occur, the panel agreed that intraoperative FCM findings should generally guide management.

Initially, no consensus was achieved when considering whether FCM should be used in patients with evidence of extra‐prostatic extension on MRI. Subsequent discussion emphasized that MRI alone may underestimate disease aggressiveness and that broader clinical features, such as high‐grade, high‐volume, or otherwise high‐risk PCa, should also guide selection. After refining the question, consensus (86%) was achieved in favour of a risk‐adapted approach, prioritizing real‐time margin analysis for patients with a higher likelihood of adverse pathology (such as those with high‐grade, high‐volume, or locally advanced disease features). This strategy was viewed as balancing the potential benefits of intraoperative FCM imaging with practical considerations of time and resource efficiency, consistent with principles of personalized, evidence‐informed surgical decision‐making.

The panel evaluated whether digital rectal examination (DRE) remains a useful adjunct to real‐time surgical margin analysis, both for general and side‐specific (unilateral vs bilateral) nerve‐sparing decisions. The discussion revealed considerable variability in clinical practice. DRE has traditionally been integral to preoperative staging in PCa, but its intraoperative value appears limited in the current era of imaging‐guided surgery [23]. Several panellists noted that DRE offers little additional information beyond multiparametric MRI, given its subjective nature and poor reproducibility. However, others emphasized that DRE may still provide complementary insights when imaging findings are equivocal, occasionally revealing disease extension not detected by MRI. The lack of consensus reflects an ongoing transition toward image‐ and technology‐based assessment paradigms, in which DRE may retain a minor, context‐dependent role within multimodal clinical evaluation [23, 24].

Finally, in terms of surgical planning, consensus was reached that intrafascial nerve sparing should not be routinely performed in all cases (82%). The panel emphasized that the indication for intrafascial dissection should be individualized, considering tumour characteristics, stage, grade, and patient factors such as age and baseline erectile function, rather than following a uniform approach. When the impact of intraoperative margin assessment on this decision was discussed, opinions remained divided and no consensus was reached (43% in favour and 57% against intrafascial nerve sparing).

Sample Processing and Image Acquisition

Consensus was reached in favour of the en‐face imaging technique, defined as surface‐based imaging of the surgical margin, as the preferred approach for intraoperative margin assessment over the slicing (cross‐sectional) technique (82%). The panel considered this method the most practical and efficient, enabling rapid evaluation of the entire specimen surface while maintaining tissue integrity for subsequent histopathological analysis. Although two pathologists advocated a slicing approach for irregular or curved areas such as the base or anterior prostatic surfaces, most participants, including other pathologists, emphasized that slicing increases artefacts and technical complexity. Overall, en‐face imaging was endorsed as the most suitable and reproducible strategy for real‐time intraoperative application.

No consensus was reached regarding the optimal scanning depth configuration. Some participants (45%) favoured a fixed scanning depth, citing simplicity, speed, and reduced artefact risk, whereas others (55%) supported an adjustable depth to allow visualization of deeper layers and adaptation to uneven surfaces.

Regarding image quality, consensus was initially achieved in defining adequate visualization as at least 70% of the anatomical area visible within the imaged field. However, several experts argued that this threshold was insufficient for posterolateral regions, which are critical for nerve sparing and oncological control. Following discussion, the criterion was refined to require ≥90% visualization of the anatomical area in these regions, and consensus was reaffirmed (100%). Agreement was also reached on the need for absence of motion artefacts, with blurring acceptable only at the periphery (96%), and for sufficient contrast to delineate anatomical structures clearly (100%). The panel recognized that image quality varies by anatomical site: posterolateral areas generally yield complete, high‐quality scans, whereas the base and apex are more prone to artefacts, air bubbles, and diathermy effects. Pathologists and surgeons concurred that these latter regions contribute less to nerve‐sparing decision‐making. Collectively, the group identified adequate visualization, minimal artefacts, and high contrast as the essential parameters for defining intraoperative scan quality.

Regarding the areas for FCM analysis, the only consensus reached was that scanning the entire prostate (six faces) is not necessary (88%). The discussion emphasized the limited clinical utility and technical challenges of apex scanning. Positive margins at the apex seldom influence intraoperative decision‐making, as additional resection may compromise urethral length and postoperative continence [25]. Moreover, apex imaging is particularly prone to FCM artefacts resulting from inadequate specimen preparation, such as residual fluid or incomplete drying, which may contribute to higher false‐negative rates. Despite the lack of consensus, the panel tended to favour targeted imaging, particularly of the posterolateral region (42%) or the posterolateral plus apex combination (62%).

The panel discussed the maximum acceptable turnaround time between specimen excision and pathologist FCM reporting to maintain intraoperative feasibility. Key factors included the distance between the operating theatre and pathology unit, personnel availability, and the presence of digital pathology infrastructure. Drawing on the NeuroSAFE experience, where processing typically required >45 min, participants agreed that such delays hinder surgical flow [7, 18]. Centres equipped with digital transmission systems or on‐site pathology support reported markedly shorter turnaround times (10–15 min), which were considered optimal for real‐time decision‐making. Overall, consensus was reached that the acceptable timeframe should not exceed 30 min, with a preferred window of 15–30 min for complete interpretation and reporting (85%).

The feasibility of performing Gleason grading with structured FCM during intraoperative assessment was then explored. No consensus was reached regarding the ability to assign formal Gleason scores with this technique (33% in favour vs 67% opposed) or on the capacity to distinguish between low‐risk and high‐risk PCa (45% in favour vs 55% opposed). Participants highlighted the difficulty in reliably distinguishing architectural patterns, particularly between Gleason 3 + 3, 3 + 4, and 4 + 3, on FCM images, noting that differentiation of well‐formed from poorly formed glands remains technically challenging [26]. Although discrimination between benign and malignant tissue appears achievable with FCM, small foci (≤1 mm) may be missed, consistent with reported sensitivities of approximately 48% for all lesions ≥0.1 mm, improving to around 80% for those ≥3 mm [11, 27]. Lesion size currently represents the main determinant of detection sensitivity, as previously published studies have not specifically addressed histological grading [27, 28]. Therefore, FCM enables rapid qualitative identification of cancerous tissue, but accurate cancer grading remains challenging.

The panel then considered whether formal certification should be required for FCM interpretation. There was strong agreement that structured training and certification are necessary for reliable use (83%), as manufacturer instructions alone were viewed as insufficient. Competency should be demonstrated through a standardized process combining theoretical and practical assessment. A tiered model, aligned with existing European micro‐credential frameworks, was deemed appropriate, beginning with basic training on static images and progressing to advanced modules with interactive annotation and supervised review of 20–40 cases [29]. The panel agreed that formal certification is essential for both surgeons and pathologists to ensure accuracy and patient safety (100%). Although concerns were raised about potential bias when surgeons interpret their own images, joint training was encouraged to promote multidisciplinary alignment. From a quality‐assurance standpoint, intraoperative FCM interpretation should complement, rather than replace definitive histopathology and remain subject to structured auditing and certification oversight.

The discussion then focused on standardizing how PSMs should be reported on FCM. Consensus was reached that positive surgical margins should be primarily reported using the maximum linear extent, defined as the maximum transverse diameter (82%). This parameter reflects the transverse sectioning of prostate specimens in routine pathological practice, closely corresponds to measurements obtained on frozen section and definitive histopathology, and provides surgeons with a clear and clinically meaningful estimate of margin extent for intraoperative decision‐making [11, 28]. Alternative metrics, such as two‐dimensional or surface‐area measurements, were considered exploratory and not yet suitable for clinical use. The panel also reached consensus that separate positive margins on the same surface should be reported individually (93%) to maintain diagnostic precision.

There was clear agreement that live remote reporting of FCM scans within the same institutional network is feasible and acceptable, provided that real‐time communication with the surgical team is maintained (84%). Some discussion extended to cross‐institutional or international reporting, which, although potentially valuable for expert consultation, was viewed with caution because of concerns about delayed local skill development, slower certification, and dependence on external specialists.

Regarding the role of artificial intelligence in FCM, the panel agreed that it could eventually support intraoperative prostate margin assessment, but current datasets are too limited for reliable diagnostic algorithms. Insufficient annotated images, large data requirements, and computational demands were identified as major barriers, and several members cautioned that premature implementation might hinder user training [30]. Nonetheless, consensus emerged that early artificial intelligence applications are most feasible in image quality control (93%) and automated detection of areas of interest (93%), whereas no consensus was reached for direct diagnostic interpretation.

Surgical Management

The panel examined the potential role of the Gleason score in guiding the management of intraoperative PSM finding. It was acknowledged that tumour grade is a major prognostic factor in PCa and would, in principle, influence intraoperative decision‐making if reliably available (89%). However, current intraoperative FCM imaging modalities do not yet allow accurate grading of tumour architecture, as previously discussed, limiting the direct use of the Gleason score during surgery. The panel explored in depth the intraoperative management of PSM detected through FCM, with particular attention to how management strategies should vary according to tumour grade, PSM length, and anatomical location. In the absence of intraoperative Gleason grading, a strong consensus was reached that surgical decisions should rely on a combined interpretation of the intraoperative PSM dimension (maximum length in mm) and the preoperative Gleason score, rather than on margin size alone (89%).

For short PSMs (<3 mm) in low‐ and intermediate‐risk patients, no consensus was reached, although most panellists favoured a conservative approach (72% leaving the NVB intact), citing the low likelihood of clinically significant residual disease and the unlikely oncological detriment of preserving the NVB, balanced against the potential functional risks of additional dissection. In high‐risk patients with similarly short margins, opinions were more divided, with preferences split between partial (35%) and full (58%) resection, reflecting the balance between oncological control and preservation of NVB integrity. Conversely, when larger PSMs (>3 mm) were discussed, particularly in high‐risk disease, consensus was achieved in favour of complete NVB resection (96%), considered the most oncologically sound strategy. However, for lower‐risk profiles, heterogeneity persisted, and no unified recommendation could be made (37% for partial resection vs 63% for complete resection).

The discussion also addressed the utility of intraoperative FCM evaluation of tissue resected after partial or full NVB excision. No consensus was reached. Most panellists considered resection alone to be sufficient (69%), citing the limited interpretability of such specimens and logistical constraints, whereas a minority (31%) supported additional FCM assessment for quality assurance.

Regarding apical and basal FCM scanning, panellists noted during the discussion that current technical limitations restrict the clinical relevance of intraoperative decision making at these sites, particularly at the prostate base where image quality is often insufficient for reliable assessment. As this issue was not formally voted on, the panel suggested that its evaluation should be revisited once imaging technology advances. The group underscored the need for prospective, outcome‐driven studies to determine the oncological and functional implications of margin‐directed resection strategies at this level and to establish evidence‐based thresholds for intraoperative management.

Implications and Future Directions

Overall, this Delphi suggests that FCM may represent a useful adjunct to intraoperative decision‐making during RARP, particularly in higher‐risk patients where real‐time margin assessment could support more tailored nerve‐sparing strategies. The findings also indicate that its clinical implementation will depend on not only diagnostic performance but also the development of standardized acquisition protocols, appropriate training, and integration into routine multidisciplinary practice. Although this work focused on PCa surgery, where evidence supporting ex vivo FCM is steadily increasing, this technology is also being explored in other surgical and urological settings, including bladder cancer, suggesting a broader potential role [31, 32, 33]. As further evidence becomes available, its application may extend beyond RARP toward a wider use in intraoperative tissue assessment.

Limitations

The study has several limitations. First, as with all Delphi processes, the conclusions reflect expert opinion rather than empirical evidence and may be influenced by the specific composition of the panel. Although efforts were made to ensure international and multidisciplinary representation, the number of participants was limited, and consensus thresholds do not equate to scientific validation. Second, the modified Delphi design focused on expert agreement rather than quantitative outcome measures; therefore, recommendations should be interpreted as guidance pending prospective validation. Nonetheless, expert opinion from individuals with extensive experience in the field remains valuable, as it helps to consolidate current knowledge and, importantly, to highlight the key areas that require further investigation and development in FCM. Third, despite the inclusion of both urologists and pathologists, most panellists were affiliated with high‐volume academic or referral centres, which may limit the generalizability of some recommendations to lower‐resource settings.

Furthermore, cost considerations were not specifically addressed in this consensus and represent an important limitation. Although FCM may potentially reduce operative time and improve workflow efficiency compared with frozen section analysis, its economic advantage remains to be demonstrated. Dedicated studies are therefore needed to evaluate its cost effectiveness. An additional limitation is the lack of prospective protocol registration. Although the methodology and analysis plan were defined a priori by the SC, the absence of formal registration may increase the risk of bias and limit transparency and reproducibility. In mitigation, we provide detailed documentation of the process, including the questionnaire, preparatory materials, and meeting agenda as Supplementary material. Finally, evidence supporting FCM in prostate surgery remains preliminary, with limited prospective and comparative data. Future multicentre studies are required to validate the proposed workflow, assess its oncological and functional impact, and develop uniform quality assurance criteria to support consistent and safe clinical integration.

Conclusions

This international Delphi consensus provides expert guidance on the intraoperative use of FCM during RARP. FCM is endorsed as a rapid adjunct to MRI for risk‐adapted nerve sparing and real‐time margin assessment, with agreement on en‐face imaging, reporting of maximal margin length, and structured certification for image interpretation. Further multicentre validation is required to confirm the clinical impact of FCM and to establish standardized frameworks for its implementation during RARP.

Disclosure of Interests

Stefano Puliatti reports an ongoing randomized controlled trial on FCM funded by Regione Emilia‐Romagna (Bando FIN‐RER 2020 – Progetto Regione‐Università), investigating the use of ex vivo FCM across multiple oncological settings. He reports no consultancy roles or industry‐related funding. He is also a member of the Robotic Surgery Committee of the Italian Endourological Association and of Uronauti. Alexandre Mottrie reports consulting relationships with Telix, Samantree, and MicroPort; speaker honoraria from Medtronic, Intuitive, and Conmed; and serves as the chief executive officer of ORSI Academy. He is also an honorary member of the European Association of Urology (EAU) Robotic Urology Section and a member of the EAU. Stefania Ferretti reports advisory board participation with Alnylam Pharma (LUX Platform Iperossaluria); consultancy roles with Teleflex, Cook, Rocamed, Coloplast, and Boston Scientific in the past 2 years; and leadership roles in scientific societies, including the Italian Endourological Association and the International Alliance of Urolithiasis. She is also involved in the Endourology Surgquest platform. Declan Murphy reports reimbursement for consulting/speaker duties from Astellas, Johnson & Johnson, Bayer, Novartis, Ipsen, Cipla, Mundipharma and Intuitive Surgical. Mathias Winkler reports that he has been chief investigator on studies evaluating FCM, including IP8‐FLUORESCE (BJU Int 2025, doi: 10.1111/bju.16588; Eur Urol 2026, doi: 10.1016/j.eururo.2025.09.4171). He reports ongoing academic work in this field, including submitted manuscripts on FCM in bladder cancer and on deep learning applications for prediction of positive surgical margins. He has performed unpaid consultancy work for SamanTree. Ricardo Almeida‐Magana reports grants from the National Institute for Health and Care Research (NIHR) (RFPB PB‐PG‐1216‐20 013), the JP Moulton Charitable Foundation (17/0043), and Prostate Cancer UK (MA‐CT20‐11), all supporting salary. He has received travel support from SamanTree Medical (NARUS 2025). He is currently conducting a clinical trial on confocal microscopy for PCa. Greg L. Shaw reports grants from the NIHR (RFPB PB‐PG‐1216‐20013), the JP Moulton Charitable Foundation (17/0043), and Prostate Cancer UK (MA‐CT20‐11), all supporting salary. He has received honoraria and/or travel support from Astellas, Ipsen, and Janssen, and travel support from SamanTree Medical (EAU 2024). He is currently conducting a clinical trial on confocal microscopy for PCa. Anna Silvanto reports involvement in a planned study on FCM for intraoperative prostate margin assessment at University College London Hospitals. Ruben De Groote, Edoardo Beatrici, Mario De Angelis, Vincenzo Cavarra, Francesco Pepillo, Nicola Frego, Natali Rodriguez Peñaranda, and Alexandre Mottrie report authorship of a manuscript on FCM for intraoperative margin assessment during robot‐assisted radical prostatectomy (doi: 10.1111/bju.70262). Cristian Fiori, Jon Oxley, Albino Eccher, Ravi Barod, Luca Reggiani Bonetti, Ruben Olivares, Giampaolo Bianchi, Elio Mazzone, Andrei Plopeanu, and Christopher J. Winkler report no relevant interests to declare.

Funding

This research received no external funding.

Supporting information

Appendix S1. ACCORD (ACcurate COnsensus Reporting Document) checklist used to guide the reporting of the modified Delphi consensus study.

Table S1. Summary of all items included in the modified Delphi process, with corresponding response distributions, consensus outcomes, and an explanatory legend indicating whether items were accepted, revised, or classified as no consensus reached.

Video S1. VivaScope® 2500 workflow for ex vivo fluorescence confocal microscopy, including specimen preparation, image acquisition, and real‐time assessment of surgical margins during robot‐assisted radical prostatectomy.

Video S2. Demonstration of the Histolog® fluorescence confocal microscopy system for ex vivo imaging of prostate specimens. The video illustrates specimen preparation, image acquisition, and real‐time evaluation of surgical margins during robot‐assisted radical prostatectomy.

Appendix S2. Pre‐meeting Delphi questionnaire containing all evidence‐based multiple‐choice questions and response options developed by the Steering Committee and distributed to participants before the consensus meeting.

Appendix S3. Illustrative document outlining the objectives, methodology, voting process, and expected outcomes of the modified Delphi consensus meeting on fluorescence confocal microscopy for intraoperative surgical margin assessment during radical prostatectomy.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.