Real-time intraoperative perfusion assessment using indocyanine green fluorescence angiography as a predictor of anastomotic integrity in colorectal surgery: A systematic review and meta-analysis

Abstract

The purpose of this study was to assess the association of intraoperative indocyanine green (ICG) fluorescence angiography with anastomotic leakage and surgical decision-making after colorectal resection. The systematic review and meta-analysis were conducted in accordance with PRISMA guidance. PubMed, Scopus, Web of Science, and Embase were searched for randomized and observational comparative studies reporting anastomotic leakage after colorectal surgery with or without ICG fluorescence assessment. Fourteen studies involving 5264 patients were included. The pooled risk ratio (RR) for anastomotic leakage was 0.59 (95% CI: 0.47–0.73; p < 0.001), corresponding to a 41% relative reduction but an unadjusted absolute difference of approximately 3.2 percentage points. Design-stratified sensitivity analyses remained significant in randomized trials (RR 0.65, 95% CI: 0.50–0.84) and observational studies (RR 0.43, 95% CI: 0.28–0.67), with no statistically significant subgroup interaction (p = 0.12). Surgical strategy was modified in approximately 7% of patients, whereas evidence for reoperation was inconclusive and underpowered. No universally validated qualitative or quantitative threshold was identified, assessment timing varied, and visual interpretation remained operator-dependent. ICG fluorescence angiography should therefore be considered an adjunct to comprehensive surgical assessment rather than an independent predictor of anastomotic integrity.

Graphical abstract

Highlights

• •ICG fluorescence angiography was associated with lower anastomotic leakage risk after colorectal surgery (RR = 0.59).
• •Surgical plans were modified in ~7% of patients; this was not directly comparable with the 41% risk reduction.
• •No universal perfusion threshold or timing was identified; visual ICG interpretation remained operator-dependent.

Introduction

Colorectal resection with primary anastomosis is one of the most basic surgery types used in a variety of gastrointestinal conditions such as colorectal cancer, diverticular disease, or inflammatory bowel disease. Even with significant improvements in surgical methods, perioperative care, and minimally invasive surgeries, anastomotic leakage (AL) is one of the most dreaded complications after colorectal surgery. AL has been reported to occur between 3%and 19%, with greater rates on the left and rectal resections. The complication is linked to high postoperative morbidity, long hospital stays, higher healthcare expenditures and high mortality rates up to 10% in serious form. Moreover, long-term oncological outcomes, such as increased local recurrence and decreased overall survival, can be negatively impacted by anastomotic leakage in colorectal cancer surgery [1], [2].

The pathogenesis of anastomotic leakage is a complicated multifactorial process that includes patient-related, disease-related, and technical factors. Of these determinants, proper perfusion of the bowel ends to be used in anastomosis is core in ensuring that intestinal tissues heal properly. Adequate blood flow is important to keep the tissues oxygenated, to enhance collagen production, and help mucosa regenerate in the initial stages of anastomotic healing. On the other hand, impaired microvascular perfusion can result in tissue ischemia, wound healing, and subsequent anastomotic site breakdown [3]. Therefore, real-time measurement of bowel perfusion during an operation has turned into a pivotal aspect of colorectal surgery.

Historically, surgeons have used subjective measures, including bowel colour, arterial pulsation, mesenteric vessel bleeding and tissue warmth to assess perfusion of the resection margin. These traditional measures are, however, highly qualitative and rely on experience in surgery, so they can be prone to interobserver inconsistency and subject to misinterpretation. Notably, the presence of microvascular insufficiency may go unnoticed with these visual and palpation techniques, which may lead to the emergence of postoperative anastomotic complications [4]. The constraints have led to a growing interest in objective intraoperative technologies that have the potential to offer more reliable and reproducible assessment of intestinal perfusion.

Near-infrared fluorescence imaging with the use of indocyanine green (ICG) has become a promising approach to intraoperative evaluation of bowel perfusion in recent years. Indocyanine green is a water-soluble fluorescent dye which binds quickly to plasma proteins and fluoresces when subjected to near-infrared light. This imaging technology can be used after intravenous delivery to visualize real-time tissue perfusion and microvascular blood flow on the surgical site. ICG fluorescence angiography can inform intraoperative decision-making, such as changing the resection margin or enhancing the anastomotic by observing the vascularisation at the desired transection line and anastomotic site [5].

There is an increasing amount of clinical evidence to support the hypothesis that fluorescence-guided perfusion imaging could enhance surgical outcomes in colorectal surgery. A few prospective studies and randomized clinical trials have indicated that the intraoperative surgical strategy can be changed by the use of ICG fluorescence imaging in a significant percentage of cases and can help to minimize the incidence of anastomotic leakage compared to the traditional visual assessment [4], [6]. Meta-analyses have also shown that fluorescence-guided perfusion assessment can have a considerable effect in reducing the risk of anastomotic leakage, but the effects of these methods are inconsistent across studies [7].

In addition to qualitative visualization, developments in fluorescence imaging permit real-time, characterization of dynamic parameters, including fluorescence-intensity curves, time to fluorescence, time to peak fluorescence, and perfusion-distribution patterns. Although quantitative analysis may help to increase objectivity, a universal validated numerical cut-off or qualitative grading system for revision of the transection line or completed anastomosis has not yet been set. Other parameters have been suggested depending on the imaging platform, ICG dose, camera position and acquisition protocol. These techniques could have implications for left-sided colorectal resections, where the vessels are at risk due to the anatomical arrangement of the inferior mesenteric arterial system and marginal artery of Drummond [8], [9].

Fluorescence-guided surgery is becoming more widely used but there are still some inconsistencies in the imaging procedures, perfusion assessment criteria, timing of assessment and clinical outcome reporting. The interobserver variation can be significant for visual interpretation, and the role of a fluorescence assessment before versus after the completion of the anastomosis is different. Therefore, the use of ICG fluorescence angiography as an independent predictor of anastomotic integrity is not yet established, and its use as an adjunct decision-support tool is uncertain.

The increasing use of fluorescence imaging and the clinical significance of preventing anastomotic leakage is a reason for a synthesis of the available evidence. Therefore, the purpose of this systematic review and meta-analysis was to explore the relationship between the intraoperative ICG fluorescence assessment and the incidence of anastomotic leakage and the contribution to surgical decision making, and to summarise the criteria and time of assessment and their threshold values for modifying the surgical strategy in the various studies.

Methodology

Study design

The study aimed to systematically review and synthesis the evidence regarding the relationship between real-time intraoperative perfusion evaluation with indocyanine green (ICG) fluorescence angiography and anastomotic outcomes after colorectal resection. The methodological framework adhered to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for supporting the transparent and reproducible process of study selection, and data synthesis [10]. The review also included heterogeneity in the timing, the interpretation of ICG, study specific thresholds, and the criteria for changing the operative strategy.

Eligibility criteria

The studies were included in the study based on the criterion that they assessed the application of indocyanine green fluorescence angiography to intraoperative perfusion during colorectal surgery with primary anastomosis. The eligibility criteria were randomized controlled trials, prospective cohort studies, and retrospective observational studies reporting clinically relevant outcomes (e.g., anastomotic leakage, intraoperative alteration of surgical plan, or reoperation due to anastomotic complications). Research studies had to offer adequate numerical data, such as the number of events and the overall sample sizes in both intervention and control groups, to permit the calculation of pooled effect estimates. The studies were filtered out as case reports, narrative reviews, editorials, and conference abstracts and studies that were not experimental and lacked extractable clinical outcome data. Further, these studies that measured perfusion methods, but not fluorescence imaging, or did not use colorectal surgery were eliminated.

Literature search strategy

A thorough literature review was performed using key biomedical databases, such as PubMed, Scopus, Web of Science, and Embase, to find the studies that assessed the use of fluorescence-guided perfusion-assessment in colorectal surgery. The search strategy involved the use of keywords and controlled vocabulary combinations based on intraoperative perfusion imaging and colorectal surgery. Search terms were indocyanine green, ICG fluorescence, fluorescence angiography, bowel surgery, bowel perfusion, anastomotic leak and fluorescence-guided surgery. Only English-language studies in peer-reviewed journals were taken into consideration. The reference lists of the relevant articles were also manually screened to ensure that the search is complete and that more qualified studies are identified.

Study selection

All records identified through the database search were transferred to reference-management software and duplicates were removed. Screening was conducted in two stages: title/abstract review followed by full-text assessment against the prespecified eligibility criteria. Automation-assisted screening was used only to mark records that clearly failed predefined topic criteria; these exclusions were manually checked before final removal. Eligibility decisions were verified by an independent methodological reviewer who did not meet authorship criteria, and disagreements were resolved by consensus.

Data extraction

The author extracted data using a standardized form, and an independent methodological reviewer checked the eligibility decisions and key numerical outcomes. Extracted variables included first author, publication year, study design, country, intervention and control sample sizes, procedure type, ICG dose and imaging system when reported, timing of fluorescence assessment in relation to anastomosis formation, qualitative or quantitative interpretation criteria, the threshold or finding used to modify the surgical plan, anastomotic leakage events, reoperation, and reported adverse consequences. Numerical event data were extracted from both groups for quantitative synthesis. Fourteen studies reported anastomotic leakage, four reported intraoperative surgical-plan modification, and two reported reoperation because of anastomotic leakage. Study-specific ICG assessment criteria and decision rules are summarized in Table 4.

Table 4.

Study-specific ICG fluorescence assessment criteria, timing, and intraoperative decision rules.

Study	Timing and interpretation method	Study-specific threshold/criterion and surgical action
Alekseev et al. [15]	ICG fluorescence angiography was used intraoperatively to assess perfusion of the anastomosed colon and rectal stump; the exact pre-/post-anastomotic sequence was not clearly reported.	Impaired perfusion could prompt a change in the bowel-resection level. No reproducible numerical or ordinal cut-off was reported.
Watanabe et al. [5]	Assessment was performed immediately before anastomosis at the planned proximal anastomotic line. Time from ICG injection and saline flush to visible bowel-wall fluorescence was recorded.	Fluorescence appearing within 60 s was considered adequate; fluorescence after 60 s or absent fluorescence was considered poor. The revised transection line was placed more than 1 cm proximal to the border of the well-perfused zone.
Tsang et al. [21]	ICG was administered after colonic transection, and vascular perfusion was observed before completion of the anastomosis.	The resection plane could be changed when fluorescence was judged inadequate. No explicit numerical threshold was reported.
Tueme-de la Peña et al. [22]	ICG assessment was performed at the selected resection site and repeated at the completed anastomosis using SPY ENV/SPY Contrast imaging.	Visible tissue contrast in less than 60 s indicated adequate perfusion. Poor visualization or a perfusion time greater than 60 s indicated inadequate perfusion and could prompt relocation of the anastomosis.
De Nardi et al. [6]	Perfusion was assessed intraoperatively before construction of the anastomosis by visual interpretation of the fluorescence signal.	Insufficient fluorescence prompted extension of bowel resection. The published report did not provide a universally applicable numerical decision cut-off.
Rinne et al. [23]	Assessment was performed at the clinically planned anastomotic site before bowel transection and repeated after completion of the anastomosis. Time to maximal perceived signal was also recorded.	A subjective 0–2 intensity scale was used: 0, no fluorescence; 1, intermediate fluorescence; 2, fluorescence similar to non-operated bowel. Score 0 required revision, whereas revision for score 1 was left to surgeon discretion.
Kim et al. [24]	Real-time fluorescence was assessed at the planned transection area and/or near the anastomosis in patients considered at risk of bowel ischemia. Perfusion time was explored but was not used as a validated cut-off.	The fluorescence appearance was used qualitatively to confirm competent perfusion. No prespecified numerical threshold for changing the surgical strategy was reported.
Kudszus et al. [14]	Laser fluorescence angiography was applied intraoperatively to assess the perfusion of bowel ends/resection margins.	The surgical plan was modified when perfusion was considered insufficient. No explicit numerical or ordinal decision threshold was reported.
Boni et al. [12]	Perfusion was assessed before and after completion of the anastomosis by real-time qualitative comparison of fluorescence in the proposed bowel segment with adjacent/proximal bowel.	Absent or visibly reduced fluorescence was interpreted as inadequate perfusion and could prompt proximal re-resection. No validated numerical cut-off was reported.
Marquardt et al. [25]	The surgeon first marked the proposed transection line by conventional assessment; ICG fluorescence was then evaluated before resection and again after anastomosis.	The operative strategy was confirmed or changed according to subjective qualitative fluorescence. The system did not provide a quantitative parameter, and no numerical cut-off was used.
Dinallo et al. [26]	ICG fluorescence angiography was performed before creation of the anastomosis. SPY-Q fluorescence intensity was recorded 60 s after the initial fluorescence signal.	Absolute fluorescence values were recorded on a 0–256 grey scale, but the study did not define a prespecified value for changing the anastomotic site.
Impellizzeri et al. [19]	Near-infrared ICG assessment was performed before and after formation of the anastomosis.	Inadequate qualitative perfusion could lead to relocation of the transection site or repetition of the anastomosis. No explicit numerical threshold was reported.
Ishii et al. [13]	Fluorescence assessment was performed at the planned bowel transection/anastomotic site before reconstruction.	The transection line was moved proximally when fluorescence was judged inadequate. No explicit numerical cut-off was reported.
Wojcik et al. [27]	Perfusion was assessed intraoperatively before final colon transection/anastomosis; recorded images were subsequently reviewed by blinded assessors to examine reproducibility.	Inadequate qualitative perfusion prompted a change in the initially planned colon transection. No numerical threshold was reported.

Note: The criteria shown are study-specific and should not be interpreted as universally validated thresholds. ICG, indocyanine green; s, seconds.

Outcomes of interest

The primary outcome was the incidence of anastomotic leakage after colorectal resection. Secondary outcomes were intraoperative modification of the surgical plan following perfusion assessment and reoperation because of anastomotic leakage. The review also descriptively evaluated the timing and criteria used for ICG interpretation and whether the included studies reported potentially unnecessary surgical modification or other adverse consequences.

Statistical analysis

A random-effects model using the inverse-variance method was applied because clinical and methodological variation was expected across studies. Risk ratios with 95% confidence intervals were calculated for dichotomous comparative outcomes, and pooled proportions were used for surgical-plan modification. Forest plots displayed the pooled estimates and individual-study contributions. Anastomotic leakage, surgical-plan modification, and reoperation were analyzed separately. The interpretation of non-significant secondary outcomes considered the width of the confidence interval and the small number of contributing studies; no formal mediation analysis was available to determine whether surgical-plan modification explained the pooled association with leakage.

To examine whether the pooled association was influenced by study design, a design-stratified sensitivity analysis was performed for randomized controlled trials and observational studies. A chi-square test for subgroup differences was used to compare the pooled estimates. No prospective sample-size calculation applies to a meta-analysis; therefore, non-significant findings based on few studies, few events, or wide confidence intervals were interpreted as inconclusive because of the possibility of type II error.

Assessment of heterogeneity

The Cochran Q test and I² statistic were used to measure statistical heterogeneity of studies. I² value was utilized to measure the percentage of variance that can be attributed to between-studies heterogeneity instead of sampling error. The values of around 25%, 50%, and 75% were used to show low, moderate, and high levels of heterogeneity, respectively. Tau² estimates were also derived to further estimate the between-study variance in the random-effects model [11].

Sensitivity considerations

Because randomized and observational studies were combined, the pooled estimate was interpreted as an association rather than definitive evidence of causality. A random-effects model addressed between-study variability but could not eliminate selection bias, residual confounding, or differences in surgical practice and ICG protocols. Secondary analyses based on only two to four studies were considered exploratory and potentially underpowered; consequently, a non-significant result was not interpreted as evidence that no clinically important effect existed.

Results

A total of 2772 records in electronic databases were found in the systematic literature search. Before screening, 1076 duplicate records were eliminated and 850 records were identified as ineligible through automation tools and 286 records were excluded. After these steps, there were 560 records left to screen on titles and abstracts.

Following the screening phase, 284 records were eliminated and 276 reports were requested to retrieve the full-text. Out of these, 156 reports were not retrievable and the remaining 120 reports were to be subjected to full-text eligibility evaluation.

In the full-text review, a number of studies were eliminated on the grounds of non-comparative study design, lack of outcome data and ineligible study design. Ultimately, 14 studies met the inclusion criteria and were included in the final systematic review and quantitative synthesis. The entire selection procedure of the study is outlined in the PRISMA 2020 flow diagram (Fig. 1).

Fig. 1.

PRISMA flow diagram.

Study characteristics

Table 1 summarizes the characteristics of the included studies. The meta-analysis included 14 studies published between 2010 and 2025 and a total of 5, 264 patients undergoing colorectal resection with primary anastomosis. Of these, 2467 patients underwent intraoperative ICG fluorescence, assessment and 2797 underwent conventional perfusion assessment.

Table 1.

Study characteristics of included studies.

Author (year)	Study design	Country	Sample size (ICG / control)	Intervention	Comparator	Primary outcome
[15]	Randomized Clinical Trial	Russia	187 / 190	ICG fluorescence angiography	Visual perfusion assessment	Anastomotic leak rate
[5]	Randomized Clinical Trial	Japan	422 / 417	ICG fluorescence imaging	Standard assessment	Anastomotic leakage
[21]	Prospective Cohort	USA	63 / 68	ICG perfusion assessment	Visual evaluation	Anastomotic leakage
[22]	Retrospective Comparative	Mexico	83 / 85	ICG fluorescence angiography	Conventional assessment	Anastomotic leak
[6]	Multicenter Randomized Trial	Italy	118 / 122	ICG angiography	Visual perfusion evaluation	Anastomotic leak
[23]	Randomized Clinical Trial	Finland	567 / 569	ICG fluorescence imaging	Conventional surgery	Anastomotic leakage
[24]	Prospective Cohort	South Korea	123 / 313	ICG fluorescence imaging	No fluorescence imaging	Anastomotic leak
[14]	Retrospective Matched Study	Germany	201 / 201	Laser fluorescence angiography	Conventional surgery	Anastomotic leakage
[12]	Case-Matched Study	Italy	42 / 38	ICG angiography	Standard surgery	Anastomotic leak
[25]	Retrospective Study	Germany	67 / 66	ICG fluorescence angiography	Visual perfusion assessment	Anastomotic leakage
[26]	Retrospective Cohort	USA	231 / 323	Fluorescence angiography	Conventional surgery	Anastomotic leak
[19]	Retrospective Comparative	Italy	98 / 98	Near-infrared fluorescence angiography	No fluorescence imaging	Anastomotic leak
[13]	Retrospective Cohort	Japan	223 / 265	ICG fluorescence angiography	Conventional surgery	Anastomotic leak
[27]	Prospective Case-Matched	France	42 / 42	Intraoperative fluorescence angiography	No fluorescence angiography	Anastomotic leakage

The studies included were four randomized controlled trials and ten observational studies, which were retrospective cohorts, prospective cohorts, and case-matched comparative studies. These studies were carried out in a wide range of geographic areas such as Japan, Italy, Finland, and Germany, as well as in the United States, Russia, South Korea, and Mexico, which provide a wide international representation of colorectal surgical practice.

Intraoperative ICG fluorescence angiography was used in all included studies to provide real-time visualization of bowel perfusion. However, the timing was not uniform: some studies assessed the planned transection margin before anastomosis formation, some assessed the completed anastomosis, and others performed both assessments. The criteria for adequate perfusion also differed, ranging from subjective fluorescence intensity and distribution to time-based criteria. Several studies reported that fluorescence imaging identified poorly perfused bowel segments not recognized by conventional observation alone [12], [13], [14].

Risk of bias assessment

The methodological quality was assessed based on the study design. Cochrane Risk of Bias 2 was used to evaluate randomized controlled trials, and Newcastle-Ottawa Scale was used to evaluate observational studies, as shown in Table 2, Table 3. All in all, the randomized trials had a low risk of bias on the majority of domains such as randomization process, completeness of outcome data, and outcome measurement. Nevertheless, there were also minor issues with performance bias because of the inability to blind surgeons to the fluorescence imaging opportunities during the intraoperative procedures.

Table 2.

Risk of bias assessment for randomized controlled trials (RoB 2 Tool).

Author (year)	Randomization process	Deviations from intended intervention	Missing outcome data	Outcome measurement	Selective reporting	Overall risk
[15]	Some concerns	Low risk	Low risk	Low risk	Low risk	Some concerns
[5]	Some concerns	Some concerns (open-label surgery)	Low risk	Low risk	Low risk	Some concerns
[6]	Some concerns	Some concerns	Low risk	Low risk	Low risk	Some concerns
[23]	Some concerns	Low risk	Low risk	Low risk	Low risk	Some concerns

Table 3.

Risk of bias assessment for observational / cohort studies (Newcastle–Ottawa Scale).

Author (year)	Selection (4)	Comparability (2)	Outcome (3)	Total score	Quality
[21]	★★★	★★	★★	7/9	Good
[22]	★★★	★	★★	6/9	Moderate
[24]	★★★	★	★★	6/9	Moderate
[14]	★★★★	★	★★	7/9	Good
[12]	★★★	★	★★	6/9	Moderate
[25]	★★★	★	★★	6/9	Moderate
[26]	★★★	★	★★	6/9	Moderate
[19]	★★★	★	★★	6/9	Moderate
[13]	★★★	★★	★★	7/9	Good
[27]	★★★★	★★	★★	8/9	High

The methodological quality of observational studies was moderate to high as the Newcastle-Ottawa scores were between 6 and 8 out of 9, which is acceptable comparability between the intervention and control cohort. The main constraints identified in these studies included retrospective design and possible selection bias which are methodological constraints in surgical observational research.

Primary outcome: anastomotic leak

Fig. 2 represents the pooled meta-analysis comparing the effect of intraoperative indocyanine green (ICG) fluorescence angiography on the incidence of anastomotic leak following colorectal surgery. The quantitative synthesis involved 14 studies with 2467 patients in the ICG group and 2797 patients in the control group.

Fig. 2.

Forest plot showing the pooled effect of ICG fluorescence angiography on the risk of anastomotic leakage following colorectal surgery.

The pooled analysis showed that ICG-guided perfusion assessment was associated with a lower risk of anastomotic leakage than conventional assessment. The pooled risk ratio was 0.59 (95% CI: 0.47–0.73; p < 0.001), corresponding to a 41% relative risk reduction. This value does not indicate that leakage was prevented in 41% of all patients. The crude leakage rates were 4.6% (114/2467) in the ICG group and 7.8% (219/2797) in the control group, an unadjusted absolute difference of approximately 3.2 percentage points.

Most individual studies favored the ICG group, although several confidence intervals included no effect. No statistical heterogeneity was detected (I² = 0.0%, χ² = 10.48, p = 0.6539), and the prediction interval was 0.46–0.75. Nevertheless, the absence of detected statistical heterogeneity does not remove the clinical and methodological differences in study design, patient selection, timing of ICG assessment, and perfusion-interpretation criteria.

These findings support an association between an ICG-assisted operative strategy and lower risk of anastomotic leakage. They do not establish that ICG fluorescence angiography independently predicts anastomotic integrity or that the pooled association is entirely caused by fluorescence-directed changes in surgical management.

In the design-stratified sensitivity analysis, the pooled association remained statistically significant among the four randomized controlled trials (RR 0.65, 95% CI: 0.50–0.84; I² = 0%) and among the ten observational studies (RR 0.43, 95% CI: 0.28–0.67; I² = 0%). The test for subgroup differences was not statistically significant (χ² = 2.44, df = 1, p = 0.12). The direction of association was therefore consistent, although the larger observational estimate remains more susceptible to selection bias and residual confounding.

Secondary outcome: intraoperative surgical plan modification

The pooled proportion of surgical-plan modification during the operation after ICG fluorescence assessment is presented in Fig. 3. This analysis was based on 485 patients in four studies. Approximately 7% (pooled modification rate: 0.07, 95% CI: 0.03–0.13) were evaluated, and the planned operative strategy was altered.

Fig. 3.

Forest plot showing the pooled proportion of intraoperative surgical plan modifications following ICG fluorescence–guided perfusion assessment during colorectal surgery.

Modification rates varied widely depending on type of surgery, time of assessment, and what was used as the criterion of inadequate perfusion. Heterogeneity was moderate to substantial (I² = 71.4%, p = 0.0148), and the prediction interval ranged from 0.00 to 0.28. Because most studies did not apply a validated threshold, the pooled 7% should be interpreted as a heterogeneous rate of surgeon-directed change rather than a standardized intervention rate.

Secondary outcome: reoperation due to anastomotic leak

Fig. 4 shows the meta-analysis of the impact of ICG fluorescence angiography on the reoperation because of anastomotic leakage with 2 studies (464 patients in the ICG group and 459 in the control group).

Fig. 4.

Forest plot showing the pooled effect of ICG fluorescence angiography on the risk of reoperation due to anastomotic leakage after colorectal surgery.

The pooled risk ratio for reoperation was 0.43 (95% CI: 0.12–1.60; p = 0.2069). Although the point estimate favored ICG assessment, the confidence interval and was compatible with substantial benefit, little or no effect, or possible harm. The result was therefore inconclusive rather than evidence of no effect.

Only two studies contributed to the reoperation analysis, and heterogeneity was moderate (I² = 49.8%, p = 0.1583. The small number of studies and wide confidence interval indicate substantial imprecision and a high possibility of a type II error; accordingly, this secondary outcome should be considered underpowered and exploratory.

Discussion

Anastomotic leakage is still among the most severe complications in colorectal surgery that significantly contribute to postoperative morbidity, mortality, hospitalization, and healthcare expenses. The role of impaired bowel perfusion in anastomotic failure has been long established but the traditional method of intraoperative assessment, namely visual inspection and palpation, is also subjective and may not show subtle microvascular compromise. The current meta-analysis examined the efficacy of ICG fluorescence angiography in enhancing perfusion measurements and decreasing anastomotic leakage in colorectal surgery.

The pooled findings indicate an association between an ICG-assisted operative strategy and a lower leakage risk compared with conventional assessment. Several randomized and observational studies reported fewer leaks with ICG fluorescence imaging, including the FLAG trial [15] and the Essential trial [5]. Nevertheless, fluorescence imaging should not be described as completely objective when interpretation is visual. A multicentre assessment demonstrated wide interobserver variation in visually assessed fluorescence parameters and no significant difference in those visual parameters between patients with and without postoperative leakage [16]. A separate video-based study likewise found marked inter-user variability and no concordance when surgeons judged perfusion sufficiency on a binary yes/no basis, reinforcing that visual ICG assessment is not intrinsically objective [17].

Design-stratified analyses improved the methodological interpretation. The randomized-trial estimate was smaller than the observational estimate but remained statistically significant, whereas the interaction test did not demonstrate a statistically detectable difference between designs. This does not establish equivalence between the evidence types because only four randomized trials were available and the subgroup interaction itself may have been underpowered. The more pronounced association in observational studies may partly reflect patient selection, surgeon experience, or unmeasured perioperative differences. The direction and magnitude of the randomized estimate are consistent with a subsequent RCT-only meta-analysis of seven trials involving 4636 participants, which reported an RR of 0.69 (95% CI: 0.58–0.82) [18].

One of the major findings in various studies was that fluorescence imaging could change intraoperative surgical decision-making. In various studies, surgeons adjusted the intended transection line following fluorescence imaging of poor perfusion at the intended anastomotic margin. According to De Nardi et al. (2019), approximately 11% of patients had the planned resection margin changed with the use of fluorescence imaging [6], whereas corresponding results in intraoperative change of surgical plan were reported by [13], [19]. These results indicate that the clinical utility of fluorescence angiography could be related to its capacity to detect poorly perfused bowel loops prior to anastomosis creation, and, therefore, avoid anastomotic failure associated with ischemia.

Table 4 demonstrates that the included studies used heterogeneous study-specific criteria. Two studies applied a time-based criterion of visible fluorescence within 60 s, one randomized trial used a subjective 0–2 intensity scale, and most other studies relied on the surgeon's qualitative judgement without a prespecified numerical threshold. These criteria should not be interpreted as interchangeable or universally validated decision rules.

The timing of ICG angiography is also clinically relevant. Assessment before anastomosis formation primarily informs selection of the transection margin, whereas post-anastomotic assessment evaluates the completed reconstruction and may reveal changes related to tissue tension, stapling, mesenteric orientation, or vascular compression. Several studies used only one time point, and others used both; the available data did not permit a reliable comparison of pre-anastomotic, post-anastomotic, and dual-assessment strategies. A standardized dual assessment may be reasonable for future trials, but current evidence is insufficient to designate post-anastomotic assessment alone as the ideal protocol.

The 7% surgical-plan modification rate does not conflict mathematically with the 41% relative reduction in leakage. The latter is a relative effect derived from a low baseline event rate, with crude leakage rates of 4.6% and 7.8%, rather than an absolute benefit affecting 41% of patients. In addition, the two estimates were derived from different study subsets: four studies contributed to the modification analysis and 14 to the leakage analysis. The included studies did not perform mediation analyses, and ICG may also have confirmed the original plan or prompted minor technical adjustments that were not recorded as formal strategy changes.

Potential harm should also be considered. A false-positive interpretation of poor perfusion may lead to unnecessary proximal resection, further bowel mobilisation, reconstruction of an otherwise adequate anastomosis, a higher or more proximal anastomosis, or diversion. Additional resection may itself increase technical complexity or anastomotic tension. Most included studies did not report whether fluorescence-directed changes were subsequently judged necessary or whether complications were attributable to those changes; therefore, the balance between benefit and unnecessary intervention could not be quantified. ICG administration itself also carries a rare risk of hypersensitivity reactions, including anaphylaxis [20], although adverse-event reporting was inconsistent across the included studies.

The concern regarding type II error applies primarily to individual underpowered studies and the secondary pooled outcomes, rather than to the statistically significant primary meta-analysis. In particular, only two studies contributed to the reoperation analysis and four to the strategy-modification analysis. Their wide or heterogeneous estimates cannot establish either the presence or the absence of a clinically important effect.

Although these results are promising, not every study showed statistically significant changes in leakage rates. Other trials, especially those that had heterogeneous patient groups or lower baseline leak rates, had only weakly or insignificant differences in fluorescence-guided surgery and conventional surgery. Such variation probably indicates variations in surgical methodology, risk profiles of patients, definitions of anastomotic leakage, and the timing of fluorescence measurement. Additionally, the learning curve linked with the technology of fluorescence imaging can also play a role, as more time surgeons are exposed to the technology, the more effective they become in reading the perfusion signals.

The other important factor is that the fluorescence angiography mainly assesses macroscopic perfusion, not microvascular or tissue oxygenation parameters. As a result, anastomotic failure can still be occasioned by non-perfusion factors, such as tension on the anastomosis, patient comorbidities, technical errors or systemic inflammatory responses. Thus, fluorescence imaging must be considered as complementary, but not a solution by itself in preventing anastomotic complications.

Overall, ICG fluorescence angiography may support colorectal surgical decision-making providing real-time visualization of tissue perfusion, particularly during low anterior and left-sided colorectal resections. Its findings should be integrated with anastomotic tension, bowel orientation, technical quality, patient-level risk factors, and the surgeon's conventional assessment rather than used as a stand-alone decision rule.

Conclusion

Intraoperative ICG fluorescence angiography is an adjunctive technique to visualize bowel perfusion and was found to be associated with reduced relative risk of anastomotic leakage in this meta-analysis. However, the evidence does not establish a universal qualitative or quantitative threshold, a standardized optimal timing of assessment, or an independent predictive relationship with anastomotic integrity. Visual interpretation is operator dependent, and the evidence for surgical-plan modification and reoperation is limited to a few studies and less precise. ICG should thus be used in conjunction with comprehensive surgical assessment, and not as a replacement for it. Randomized trials with sufficient power and standardized pre- and post-anastomotic protocols and validated decision criteria are necessary.

Limitations

There are a number of limitations to this review. First, randomized and observational studies were pooled, making the pooled association vulnerable to selection bias, residual confounding and differences in perioperative practice. Design stratified sensitivity analysis demonstrated consistency in direction of effect, but the analysis of subgroup interaction was based on limited number of randomized studies and could have been under powered. Second, the studies included did not employ a common qualitative grading system, or a universally accepted quantitative threshold of inadequate perfusion. Interobserver variation and operator dependence were issues with visual interpretation. Third, ICG dose and imaging platform, camera conditions and assessment time (before or after creation of the anastomosis) were inconsistently reported. The pre-anastomotic and post-anastomotic assessments are focused on different clinical questions and were not separately analyzed. Fourth, the estimate of surgical plan modification of 7% was derived from four studies, and the reoperation analysis was conducted in only two studies, making these secondary analyses prone to type II error and heterogeneous. Fifth, the review was unable to assess whether fluorescence-directed changes were required, or whether false-positive interpretation resulted in unnecessary additional resection, reconstruction, a more proximal anastomosis, increased anastomotic tension or diversion. There was also some inconsistent reporting of potential ICG related hypersensitivity reactions and adverse outcomes from surgical modification. Lastly, screening and data checking were done by a non-author methodological reviewer, and the fact that follow-up was short and most studies were conducted in specialized centres may have reduced generalizability.

Future perspectives

Future multicenter randomized trials should employ standardized doses of ICG, image acquisition parameters and clear time points for assessment. There may be complementary information gained from a protocol that includes pre-anastomosis measurement of the intended transection margin and post anastomosis remeasurement of the completed anastomosis, but its clinical benefit must be assessed in a prospective trial. Prior to using as decision thresholds, quantitative perfusion parameters should be validated externally with clinically relevant leakage outcomes. The exact reason and type of operative modification, verification as necessary, complications of additional resection or reconstruction, events related to ICG, and cost-effectiveness should also be reported. To assess the reproducibility of automated image analysis, it should be compared to clinical reference standards.

CRediT authorship contribution statement

Mohsen Ali Ezzy: Writing – review & editing, Writing – original draft, Validation, Supervision, Methodology, Formal analysis, Data curation, Conceptualization.

Funding statement

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

Declaration of competing interest

The author declares no competing interests or conflicts that could have influenced the content of this work.