Diagnostic Accuracy of Indocyanine Green–stained Sentinel Lymph Nodes in Prostate Cancer Patients: A Systematic Review and Meta-analysis

Take Home Message

Per-patient level sensitivity provides a more accurate reflection of the diagnostic performance of indocyanine green–guided sentinel lymph node dissection (ICG-SLND). However, ICG-SLND cannot yet be recommended as the standard approach for lymph node dissection, as extended pelvic lymph node dissection has demonstrated positive effects in preventing metastasis.

Abstract

Background and objective

Indocyanine green-guided sentinel lymph node dissection (ICG-SLND) has demonstrated good diagnostic accuracy for lymph node metastasis in prostate cancer. This study aims to perform a meta-analysis of the diagnostic accuracy of ICG-SLND at both the per-patient and the per-node level.

Methods

We conducted a literature search on PubMed and Embase for relevant studies published up to June 2024. The inclusion criteria were prostate cancer patients undergoing radical prostatectomy, ICG-SLND, and subsequent extended pelvic lymph node dissection (ePLND). Data were extracted to calculate the pooled sensitivity and negative predictive value (NPV) at both the per-patient and the per-node level.

Key findings and limitations

Our search identified 13 relevant studies, comprising a total of 748 patients. All studies were assessed as having a low risk of bias. At the per-patient level, the pooled sensitivity of ICG-SLND for diagnosing lymph node metastasis was 0.87 (95% confidence interval [CI]: 0.77–0.92), with a pooled NPV of 0.95 (95% CI: 0.90–0.98). At the per-node level, the pooled sensitivity was 0.53 (95% CI: 0.45–0.62), and the pooled NPV was 0.98 (95% CI: 0.97–0.98). Significant heterogeneity was observed in the per-node level sensitivity, NPV, and sentinel lymph node detection rate outcomes. The primary limitation is the lack of investigation into the impact of ICG-SLND on survival outcomes.

Conclusions and clinical implications

The per-patient level sensitivity of ICG-SLND for diagnosing lymph node metastases is 87%, which better represents the diagnostic accuracy of ICG-SLND than the per-node level sensitivity. As ePLND has demonstrated a positive impact on oncologic outcomes, ICG-SLND cannot yet be recommended as the standard lymph node dissection approach. The significant heterogeneity observed in the pooled results highlights the need for further research to determine the optimal injection methods.

Patient summary

Indocyanine green–guided sentinel lymph node dissection (ICG-SLND) has demonstrated favorable performance for diagnosing lymph node metastases in prostate cancer. The per-patient level sensitivity of ICG-SLND provides better diagnostic performance than its per-node level sensitivity. However, further research is needed for ICG-SLND to be recommended as the standard approach for lymph node dissection.

1. Introduction

In patients with prostate cancer undergoing radical prostatectomy and pelvic lymph node dissection, ∼13.8% are found to have lymph node metastasis on pathologic examination [1]. For these patients, adjuvant therapy, as opposed to mere observation, provides benefits in both overall survival and cancer-specific survival [2]. Therefore, accurately determining whether a patient has lymph node metastasis is crucial. Currently, there are many nomograms that can help predict the likelihood of lymph node metastasis before surgery [3], [4]. Although these nomograms have been validated externally, these can only serve as indicators of whether lymph node dissection is needed and cannot replace lymph node dissection as a method for diagnosing metastasis [5]. In contrast, prostate-specific membrane antigen positron emission tomography (PSMA PET) has the potential to diagnose lymph node metastasis noninvasively, thereby reducing unnecessary lymph node dissection for patients. However, a meta-analysis shows that PSMA PET has pooled sensitivity of only 58%, meaning that a significant portion of patients with lymph node metastasis might not be diagnosed accurately [6]. Therefore, pelvic lymph node dissection remains necessary for prostate cancer patients to confirm the presence of lymph node metastasis.

Both the European Association of Urology and the American Urological Association recommend using an extended pelvic lymph node dissection (ePLND) template for patients undergoing pelvic lymph node dissection [5], [7]. However, ePLND, which removes a more extensive range of lymph nodes than standard pelvic lymph node dissection or limited pelvic lymph node dissection (lPLND), is more likely to cause complications [8]. To address this, Wawroschek et al [9] in 1999 first attempted sentinel lymph node dissection (SLND) to detect lymph node metastasis with the goal of accurately diagnosing the condition while removing fewer lymph nodes to reduce complications.

In 2011, the first study investigating the accuracy of using indocyanine green (ICG) for SLND was published [10]. After an ICG injection, the sentinel lymph node and lymphatic drainage pathways fluoresce under a near-infrared camera, allowing surgeons to identify and remove sentinel lymph nodes in real time. Since then, numerous studies have been published demonstrating the diagnostic ability of ICG-guided SLND (ICG-SLND) across open (ORP), laparoscopic (LRP), and robot-assisted (RARP) radical prostatectomy [10], [11], [12]. The largest study to date shows that ICG-SLND has sensitivity of 91.4%, which surpasses the diagnostic capability of PSMA PET [6], [13]. While ICG-SLND still involves some degree of invasiveness compared with PSMA PET, it enables removal of fewer lymph nodes, thereby reducing complications. Therefore, ICG-SLND holds promise as an accurate and less invasive method for lymph node staging in prostate cancer patients.

A meta-analysis was published in 2021 that evaluated the diagnostic accuracy of ICG-SLND, but it reported only the per-node level accuracy [14]. Given that ICG does not stain all lymph nodes, SLND alone will inevitably miss some metastatic lymph nodes. However, the purpose of SLND should be to correctly diagnose whether a patient has lymph node metastasis, meaning that per-node level accuracy may underestimate the diagnostic capability of ICG-SLND. In contrast, per-patient level accuracy better reflects the practical utility of ICG-SLND. Therefore, the aim of this study is to perform a meta-analysis of the diagnostic accuracy of ICG-SLND at both the per-patient and the per-node level. The results of this study will provide a more accurate assessment of the diagnostic capabilities of ICG-SLND, guiding future recommendations for the diagnosis of lymph node metastases.

2. Methods

2.1. Protocol and registration

The conduct and reporting of this systematic review and meta-analysis adhered to the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) for Diagnostic Test Accuracy (DTA) studies [15]. The protocol for this study was registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration ID: CRD42024561411).

2.2. Search strategy

We conducted a comprehensive search of relevant literature published up to June 30, 2024, using PubMed and Embase. We also screened references from the identified articles for additional relevant studies. The search keywords were “(prostate cancer) AND ((fluorescence) OR (indocyanine green)) AND (lymph node)”. All prospective and retrospective studies with full-text availability were included in the analysis. Review articles, commentaries, conference abstracts, and images were excluded. If two studies were published by the same research group with overlapping study periods, only the most recent study was included. The language of the included articles was restricted to English. All abstracts and full-text articles were reviewed independently by two authors (Y.-J.C. and C.-L.C.) to determine eligibility. In cases of disagreement, the two authors discussed the study with a third author (Y.-C.T.), and a consensus decision was made regarding whether the study met the inclusion criteria.

2.3. Patient group and intervention

The target population consisted of all male patients diagnosed with prostate cancer who underwent radical prostatectomy. The surgical approach was not restricted and could be open, laparoscopic, or robot-assisted. The index test was a free ICG intraprostatic injection followed by SLND. The reference standard was ePLND performed after the SLND.

2.4. Data extraction and outcome assessment

We extracted the sensitivity and negative predictive value (NPV) of ICG-SLND for diagnosing lymph node metastasis from the included studies. Since false positives (no metastasis, but the sentinel lymph node pathology reports metastasis) are not possible in patients without lymph node metastasis, we did not analyze specificity and positive predictive value [13]. We reported the pooled sensitivity and NPV at both the per-patient and the per-node level.

Per-patient level sensitivity is defined as the number of patients diagnosed as having pN1 by ICG-SLND divided by the total number of patients with lymph node metastasis. Per-patient level NPV is defined as the number of nonmetastatic patients diagnosed as having pN0 by ICG-SLND divided by the total number of patients diagnosed as having pN0 by ICG-SLND.

Per-node level sensitivity is defined as the number of metastatic ICG-stained sentinel lymph nodes divided by the total number of metastatic lymph nodes. Per-node level NPV is defined as the number of nonmetastatic ICG-stained lymph nodes in pN0 patients divided by the total number of nonmetastatic ICG-stained sentinel lymph nodes.

We also reported the detection rate of ICG-stained sentinel lymph nodes, defined as the number of patients with ICG-stained sentinel lymph nodes divided by the total number of patients. We extracted all the necessary numerator and denominator values required to calculate the outcomes from the studies. If these values were not reported directly, we deduced them from other values provided. If neither the direct nor the deduced value was available, the corresponding outcome was not included in the meta-analysis.

2.5. Risk of bias and applicability assessment

All included studies were assessed for the risk of bias and applicability using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool [16]. The risk of bias was evaluated for four domains, each rated as having a low, an unclear, or a high risk: patient selection, index test, reference standard, and flow/timing. The applicability was assessed for three domains, also rated as having a low, an unclear, or a high risk: patient selection, index test, and reference standard. The assessments for both risk of bias and applicability were conducted independently by two authors (Y.-J.C. and C.-L.C.) and consensus was reached. Any disagreements were resolved by a third author (Y.-C.T.).

2.6. Statistical analysis

All the extracted sensitivity, NPV, and sentinel lymph node detection rates were pooled using a random-effect model, due to the variability in surgical techniques, ICG injection approaches, and ICG dosages across the included studies. Results were presented as a forest plot with 95% confidence intervals (CIs). Heterogeneity was assessed using Cochrane’s Q test and quantified using τ² statistic. If the p value of Cochrane’s Q test is <0.05, the included studies will be considered to have significant heterogeneity. For the per-patient level sensitivity and sentinel lymph node detection rate, subgroup analyses were conducted based on the surgical method, route of ICG injection, and total dosage of ICG. Since there are currently no suitable methods for analyzing publication bias in a meta-analysis of proportions, publication bias was not explored [17]. All statistical analyses were performed using R software version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria) and Review Manager 5.4 (Cochrane Collaboration, Oxford, England). A two-tailed p value of <0.05 was considered statistically significant.

3. Results

3.1. Study selection

The process of study selection and the reasons for exclusion are presented in the PRISMA flow diagram (Fig. 1). Of 34 studies that underwent a full-text review, 12 met our inclusion criteria. Additionally, one study was identified through the references of related articles and met the inclusion criteria. Consequently, a total of 13 studies were included in the meta-analysis.

Fig. 1.

PRISMA flow diagram. ICG = indocyanine green; PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-analyses.

3.2. Study characteristics

Among the 13 studies, two were retrospective in design and the remaining were prospective. Out of the 11 prospective studies, one was a randomized controlled trial (RCT). The total number of patients was 748. The characteristics of each study are presented in Table 1. In all studies, ICG-SLND and ePLND were performed before prostatectomy. There was considerable variation in the methods and dosages of ICG injection among the studies. Four studies used a transperineal injection method, three used a transrectal method, one used a percutaneous method, and the remaining five did not specify the injection method. The total dosage of ICG injection ranged from a minimum of 0.2 mg to a maximum of 25 mg.

Table 1.

Characteristics of included studies

Study	Design	Country	Study period	Patient number	Injection route	Injection site	Injection timing	Total dosage (mg)	Total volume (ml)	Surgery	Median age (yr)	Median PSA	Risk group
Inoue (2011) [10]	Prospective	Japan	May 2007–April 20084	14	NR	NR	During prostatectomy	0.25	1	ORP	68 (range: 63–75)	12.16 (range: 4.68–52.67)	NR
Jeschke (2012) [11]	Prospective	Austria	March 2010–October 2011	26	NR	NR	Before prostatectomy	0.5	5	LRP	62 (range: 49–74)	12 (range: 2.9–52.8)	Intermediate: 20; high: 6
Manny (2014) [12]	Prospective	USA	October 2012–April 2013	50	Percutaneous	Anterior base	After Retzius space creation	2	8	RARP	66 (range: 51–73)	6.5 (range: 1.9–32.9)	Low: 15; intermediate: 16; high: 19
Yuen (2015) [31]	Prospective	Japan	January 2012–June 2014	66	Transrectal	Base, apex	Before laparotomy	0.25–0.4	5–8	ORP	NR	<4: 4; 4–10: 41; ≥10: 19	Low: 9; intermediate: 25; high: 32
Hruby (2015) [20]	Retrospective	Austria	January 2012–September 2013	38	Transperineal	Peripheral zone	Before prostatectomy	0.5	5	LRP	67.5 (range: 46–74)	12.2 (range: 0.3–44)	Intermediate: 24; high: 14
Nguyen (2016) [32]	Prospective	Switzerland	November 2012–September 2015	12	Transrectal	Peripheral zone; base, midportion, apex	Before laparotomy	5	10	ORP	67.5 (IQR: 60.8–71.5)	11.2 (IQR: 5.2–46.7)	NR
Chennamsetty (2017) [26]	Prospective	USA	April 2014–April 2015	20	Transperineal	Peripheral zone; base, midportion, apex	After anesthesia induction	1.25, 2.5, 3.75, 5, 7.5	3	RARP	64 (range: 51–79)	7.7 (range: 2.4–179.5)	Intermediate: 12; high: 8
Harke (2018) [19]	Prospective	Germany	August 2014–April 2015	59	Transrectal	Base, apex, and central	Before docking	2.5	NR	RARP	64 (range: 50–75)	8.7 (range: 1.6–54.2)	Intermediate: 43; high: 16
Miki (2018) [33]	Prospective	Japan	January 2014–September 2016	50	NR	Peripheral zone; midprostate	30 min before surgery	0.75	3	LRP	68 (range: 46–76)	8.3 (IQR: 4.1–62.5)	Intermediate: 24; high: 26
Shimbo (2020) [34]	Prospective	Japan	July 2017–December 2018	100	NR	Base, apex	30–60 min before surgery	0.2–0.4	4	RARP	68 (range: 49–83)	7.1 (IQR: 4.70–11.0)	Intermediate: 60; high: 40
Claps (2022) [13]	Prospective	Spain	January 2014–December 2018	219	Transperineal	Middle of transitional zone	Before prostatectomy	25	5	LRP	64 (IQR: 59–69)	6.8 (IQR: 5–9)	Intermediate: 149; high: 70
Özkan (2023) [35]	Retrospective	Turkey	January 2019–December 2021	25	Transperineal	Peripheral zone	During docking	5	2	RARP	68.0 (IQR: 63.3–70.8)	8.0 (IQR: 6.5–11.9)	Intermediate: 13; high: 12
Wit (2023) [36]	Prospective	The Netherlands	August 2014–March 2020	69	NR	Peripheral zone	Before docking	10	2	RARP	69 (IQR: 63.0–71.5)	8.9 (IQR: 6.9–13.5)	NR

IQR = interquartile range; LRP = laparoscopic radical prostatectomy; NR = not reported; ORP = open radical prostatectomy; PSA = prostate-specific antigen; RARP = robot assisted radical prostatectomy.

3.3. Risk of bias and applicability

The results of the risk of bias and applicability assessment are presented in Supplementary Fig. 1. Regarding the risk of bias, three studies were rated as having an unclear risk in the patient selection domain because these did not explicitly state whether the patient population was enrolled consecutively. For the remaining domains of index test, reference standard, and flow and timing, all studies were rated as having a low risk of bias. In terms of applicability, all studies were rated to have a low risk of bias in the domains of patient selection, index test, and reference standard, as these all met the inclusion criteria set by this systematic review.

3.3.1. Diagnostic accuracy at the per-patient level

A total of ten studies provided relevant data for the calculation of the per-patient level sensitivity. Sensitivity across these studies ranged from a minimum of 0% to a maximum of 100%. The pooled sensitivity of ICG-SLND for diagnosing pN1 disease is 0.87 (95% CI: 0.77–0.92; Fig. 2A). In the subgroup analysis, based on the type of surgery, the pooled sensitivity for ORP, LRP, and RARP was 71%, 90%, and 84%, respectively. According to the route of the ICG injection, the pooled sensitivity for the transrectal and transperineal approaches was 82% and 91%, respectively. For different total ICG doses, the pooled sensitivity was 62%, 87%, 81%, and 90% for <0.5, 0.5–1, 1–5, and ≥5 mg, respectively (Supplementary Fig. 2). Regarding NPV, nine studies provided relevant data for determining per-patient level NPV. NPV across these studies ranged from a minimum of 50% to a maximum of 100%. The pooled NPV for ICG-SLND is 0.95 (95% CI: 0.90–0.98; Fig. 2B).

Fig. 2.

Diagnostic accuracy of ICG-guided lymph node dissection at the per-patient level: (A) sensitivity and (B) negative predictive value. CI = confidence interval; ICG = indocyanine green.

3.3.2. Diagnostic accuracy at the per-node level

A total of 11 studies provided relevant data, allowing for the calculation of the sensitivity at the per-node level. Sensitivity across these studies ranged from a minimum of 0% to a maximum of 100%. The pooled sensitivity of ICG-SLND at the per-node level is 0.53 (95% CI: 0.45–0.62), with significant heterogeneity observed across studies (Fig. 3A). Regarding NPV, ten studies provided relevant data for determining the per-node level NPV. NPV across these studies ranged from a minimum of 91% to a maximum of 100%. The pooled NPV for ICG-SLND is 0.96 (95% CI: 0.94–0.97), with significant heterogeneity observed across studies (Fig. 3B).

Fig. 3.

Diagnostic accuracy of ICG-guided lymph node dissection at the per-node level: (A) sensitivity and (B) negative predictive value. CI = confidence interval; ICG = indocyanine green.

3.3.3. Sentinel lymph node detection rate

A total of 11 studies provided relevant data for the calculation of the detection rate of sentinel lymph nodes. The detection rate across these studies ranged from a minimum of 36% to a maximum of 100%. The pooled detection rate following an ICG intraprostatic injection is 0.92 (95% CI: 0.81–0.97), with significant heterogeneity observed across studies (Fig. 4). In the subgroup analysis for detection rate, based on the type of surgery, the pooled detection rates for ORP, LRP, and RARP were 94%, 96%, and 83%, respectively. For the ICG injection route, the pooled detection rates for the transrectal and transperineal approaches were 96% and 95%, respectively. For different total ICG doses, the pooled detection rates were 93%, 92%, 88%, and 94% for <0.5, 0.5–1, 1–5, and ≥5 mg, respectively (Supplementary Fig. 3).

Fig. 4.

Detection rate of sentinel lymph node. CI = confidence interval.

4. Discussion

Our systematic review and meta-analysis revealed that ICG-SLND has the per-patient level sensitivity of 0.87 (95% CI: 0.77–0.92). This indicates that most patients with lymph node metastases can accurately be diagnosed with pN1 disease. A detection rate of 0.92 (95% CI: 0.81–0.97) highlights that ICG is an excellent tracer, as sentinel lymph nodes can be identified and removed in nearly all patients following an intraprostatic injection of ICG. In contrast, the per-node level sensitivity was 0.53 (95% CI: 0.45–0.62), which is significantly lower than the per-patient level sensitivity. This finding confirms that the per-node level sensitivity reported in a previous meta-analysis does not fully reflect the diagnostic accuracy of ICG-SLND [14].

Per-node level sensitivity represents the proportion of metastatic lymph nodes that are removed. According to this study, ICG-SLND leaves 47% of metastatic lymph nodes unremoved. However, even with ePLND, metastatic lymph nodes may not be removed completely. Studies using radioisotopes for SLND showed that approximately 7.7% of metastatic lymph nodes lie outside the ePLND template [18]. Similar findings were observed in studies involving ICG. For instance, Harke et al [19] reported that out of 62 metastatic lymph nodes, seven (11.3%) were outside the ePLND template. Hruby et al [20] found that 16.7% of metastatic lymph nodes were outside the ePLND template. Although the proportion of metastatic lymph nodes missed with ICG-SLND is higher than that missed with ePLND, it may not significantly impact recurrence. An RCT by de Pablos-Rodríguez et al [21] comparing ICG-SLND and ePLND found no significant difference in biochemical recurrence–free survival (BCRFS) with a median follow-up of 16 mo.

Compared with ePLND, ICG-SLND is associated with a significantly lower overall postoperative complication rate (32% vs 70%) [21]. One notable improvement is the reduced incidence of lymphedema. However, the rates of severe complications classified as Clavien-Dindo grade ≥III showed no significant difference between ICG-SLND and ePLND (12% vs 10%). These findings indicate that ICG-SLND plays a role in reducing complications associated with lymph node dissection, although the reduced complications are primarily those that can be managed with noninvasive treatments. To completely avoid complications associated with lymph node dissection, imaging-based methods have been explored to assess lymph node metastasis. Compared with conventional imaging modalities, PSMA PET offers better accuracy in detecting lymph node or bone metastasis, allowing for noninvasive node staging [22]. However, a meta-analysis using ePLND as the reference standard revealed that the per-patient level sensitivity of PSMA PET for detecting lymph node metastasis is only 58% [6]. This accuracy is notably lower than the 87% per-patient level sensitivity observed with ICG-SLND. Furthermore, when focusing on studies involving high-risk cohorts, the per-patient level sensitivity of PSMA PET drops to just 51% [6]. For these high-risk patients, accurate lymph node staging is crucial, as it determines whether they receive appropriate adjuvant therapy. Therefore, among low-invasive diagnostic tools, ICG-SLND is superior to PSMA PET for lymph node staging.

Owing to its ability to remove a greater number of lymph nodes than lPLND, ePLND provides more accurate lymph node staging and is therefore considered the current standard lymph node dissection template [5], [7], [23]. While there is no significant difference between ePLND and lPLND in terms of BCRFS, the largest and longest follow-up RCT to date has demonstrated that ePLND reduces the risk of distant metastases significantly [23], [24], [25]. This indicates that ePLND has a positive impact on oncologic outcomes. Therefore, for prostate cancer patients who meet the lymph node metastasis risk threshold indicated by nomograms, ePLND should be performed. Before ICG-SLND can be recommended as a lymph node dissection method, further evidence is needed to explore its impact on oncologic outcomes.

The method of ICG injection varies significantly across different studies. Among the included studies, three main injection methods were observed: transrectal, transperineal, and percutaneous. Manny et al [12] reported that a percutaneous injection could reduce ICG spillage and does not require additional preparation. However, the detection rate in the study was 76%, lower than that reported in studies using transrectal and transperineal injections. In our subgroup analysis, both the transrectal and the transperineal approach demonstrated favorable per-patient level sensitivity and detection rates. Therefore, transrectal or transperineal injections seem to be the preferred methods. The dose of ICG also varied significantly among studies, ranging from 0.2 to 25 mg. Chennamsetty et al [26] found that higher ICG doses resulted in more sentinel lymph nodes. However, lower doses did not impact sensitivity negatively. In this study, detection rates were similar across different ICG dose subgroups (Supplementary Fig. 3C). For the per-patient level sensitivity, although the pooled sensitivity for studies using a total dose of <0.5 mg was 62%, this result was derived from a limited number of studies and patients. Therefore, it does not necessarily suggest that lower total doses of ICG will result in reduced sensitivity. Additionally, whether to inject ICG systemically or directly into the tumor is another topic worth exploring. An RCT comparing intratumor and systemic injections showed that an intratumor injection had a higher chance of removing all metastatic lymph nodes through SLND, as a systemic injection missed more metastatic lymph nodes that were removed only during ePLND [27]. However, this study included patients with tumors visible on ultrasound, which is not the case for all patients. Performing an intratumor injection may require additional magnetic resonance imaging–ultrasound fusion equipment, potentially increasing preparation time.

In a previous consensus meeting on sentinel lymph node study design, recommendations were made regarding the injection methods for various tracers [28]. It was suggested that the injection site should be in the peripheral zone, with an injection volume of 1–2 ml. Additionally, if using ICG, the injection time should ideally be within 30 min before surgery. Among the 13 included studies, only seven described the injection site, with six describing injection in the peripheral zone and one in the transitional zone. The injection volume was reported in 12 studies, of which three had volumes between 1 and 2 ml, one had a volume below 1 ml, and the remaining eight had volumes >2 ml. Only two studies described the injection timing, and only one administered it within 30 min. In summary, not all studies adhered to the consensus meeting’s recommendations for ICG injection methods. Since these recommendations have not been validated, further research is needed to explore whether injection site, volume, and timing affect detection rates or diagnostic accuracy.

Other tracers besides ICG have also been used to identify sentinel lymph nodes in prostate cancer patients. ICG-^99mTc-nanocolloid can provide a whole-body distribution image of sentinel lymph nodes through single-photon emission computed tomography after an injection [29]. Compared with ICG, this tracer allows surgeons to know the location of sentinel lymph nodes preoperatively. However, this tracer needs to be injected the day before surgery and involves radiation exposure, making it less convenient than ICG. A newer sentinel lymph node tracer, ^99mTc-PSMA-I&S, does not require an intraprostatic injection but needs only intravenous administration, reducing the need for additional preparation and being less invasive [30]. However, it also involves radiation and requires a gamma probe for sentinel lymph node detection. In contrast, ICG-stained sentinel lymph nodes can easily be visualized by the surgeon during the operation, without the need for additional equipment or concerns about radiation exposure to the surgical team, making ICG a simpler and safer option. Further studies are needed to determine whether there are differences in diagnostic accuracy among these tracers.

This meta-analysis has several limitations. First, there is significant variation in the methods of ICG injection among the included studies, potentially affecting the accuracy of the pooled sensitivity in reflecting the true diagnostic accuracy of ICG-SLND. Second, many included studies had small patient populations. Finally, this study only provides the accuracy of ICG in diagnosing lymph node metastasis and cannot determine whether performing only ICG-SLND in prostate cancer patients would impact recurrence and survival negatively. Although a related RCT has been published, it evaluated BCRFS only and not other outcomes such as progression-free, cancer-specific, or overall survival [21]. Additionally, the median follow-up time in that study was only 16 mo. In the future, relevant research should first focus on identifying the optimal injection route and dosage. Once a standardized injection protocol is established, larger randomized controlled studies comparing the impact of ICG-SLND and ePLND on survival outcomes are needed, with stratified comparisons based on varying disease severity. Moreover, Harke et al [19] demonstrated that combining ICG with ePLND can retrieve more lymph nodes than ePLND alone. Investigating whether the combination of ICG and ePLND could further reduce the occurrence of metastases in the future is a valuable area of research. These efforts would help confirm the utility of ICG in prostate cancer patients and potentially refine its application in clinical practice.

5. Conclusions

The per-patient level sensitivity of ICG-SLND for diagnosing lymph node metastases is 87%, which is significantly higher than the per-node level sensitivity reported in previous studies. This result provides a more accurate reflection of the diagnostic performance of ICG-SLND. However, as ePLND has demonstrated a positive effect in preventing distant metastases, ICG-SLND cannot yet be recommended as the standard approach for lymph node dissection. There is considerable variation in the injection methods and doses of ICG across studies, and significant heterogeneity was observed in pooled results. This underscores the need for further research to establish standardized ICG injection protocols. Additionally, the impact of ICG-SLND on oncologic outcomes warrants further investigation.



Author contributions: Yao-Chou Tsai had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.



Study concept and design: Chou.

Acquisition of data: Chou, Chang.

Analysis and interpretation of data: Chou, Chang.

Drafting of the manuscript: Chou.

Critical revision of the manuscript for important intellectual content: Chou, Tsai.

Statistical analysis: Chou, Chang.

Obtaining funding: Tsai.

Administrative, technical, or material support: Tsai.

Supervision: Tsai.

Other: None.



Financial disclosures: Yao-Chou Tsai certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.



Funding/Support and role of the sponsor: None.

Associate Editor: Roderick van den Bergh

Appendix A. Supplementary material

The following are the Supplementary data to this article: