The Role of Indocyanine Green With Near-Infrared Imaging for the Intraoperative Detection and Enhancement of Endometriosis Lesions: A Narrative Review

Search Method

Two electronic databases were searched for this review. These were: Ovid MEDLINE and Embase. All searches took place on the 10th of July 2023. Key words in these searches included “endometriosis”, “indocyanine green”, “fluorescent dye”, “fluorescence imaging”, “optical imaging”, “near-infrared imaging” and “near-infrared spectroscopy”. These were combined with Boolean operators “AND” and “OR”. The specific search strategies within these databases are described in Table 1. The output of these searches was exported into EndNote and COVIDENCE for referencing and screening respectively.

OPEN IN VIEWER

Table 1.

Search Strategies.

Database	Search strategy
Ovid MEDLINE	Ovid MEDLINE(R) and Epub Ahead of Print, In-Process, In-Data-Review & Other Non-Indexed Citations and Daily <1946 to July 06, 2023>
	1 endometriosis.mp. or endometriosis/ 33168
	2 indocyanine green.mp. or Indocyanine Green/ 16333
	3 Fluorescent Dyes/ or fluorescent dye*.mp. 95127
	4 fluorescence imaging.mp. or Optical Imaging/ 28393
	5 near-infrared imaging.mp. or Spectroscopy, Near-Infrared/ 16741
	6 2 or 3 or 4 or 5 144946
	7 1 and 6 75
Embase	Embase Classic + Embase <1947 to 2023 July 07>
	1 exp endometriosis/ or endometriosis.mp. 53800
	2 indocyanine green.mp. or indocyanine green/ 27206
	3 fluorescent dye/ or fluorescent dye*.mp. 80285
	4 fluorescence imaging.mp. or fluorescence imaging/ 45980
	5 near infrared spectroscopy/ or near-infrared imaging.mp. 28300
	6 2 or 3 or 4 or 5 166474
	7 1 and 6176

Eligibility Criteria

The key eligibility criteria are listed below:

Screening and Data Extraction

Titles and abstracts of the articles produced by the search were screened independently by 2 reviewers according to the criteria outlined above. The data from the relevant articles was then extracted and summarised into an Excel spreadsheet by the primary reviewer. It was collated according to the sub-categories:

- Title

Dosage, Timing and Safety Data on NIR-ICG

Before considering its possible roles in laparoscopic surgery, it is important to discuss the safety and method of NIR-ICG. While there have been some rare reported cases of allergic reactions to ICG, ²³ there were no reported adverse reactions to ICG within the scope of this review. Importantly, a known ICG allergy was considered an exclusion criterion for the majority of the included studies. Additionally, a comparison of the various methodologies utilised with this technique highlighted the importance of ICG dye dosage and imaging timing. The conclusion demonstrated by Siegenthaler, et al. ²² that longer time intervals between the dye administration and imaging enhanced visualisation of lesions, was reinforced by the fact that all three of the studies which did not support the overall efficacy of ICG used exposure times as low as two and five min within their protocols (Table 5). This difference is thought to be due to the time it takes for the dye to wash-out from the surrounding healthy tissue as the fluorescence remains in the neo-vascularised endometriotic lesion. ¹⁶ With regards to dosages, the studies within this review with the most efficacious results utilised dosages of 0.25-0.30 mg/kg (Table 5). Notably, the use of NIR-ICG within other specialties also utilising the injection of peripheral intravenous ICG is diverse in protocol. The consensus for the evaluation of anastomoses in colorectal surgery is a total ICG dose of 5-10 mg, ²⁴ however biliary tract visualisation is performed with doses ranging from less than 0.05 mg/kg to 0.25-0.30 mg/kg.^25,26 While the total maximum recommended ICG dosage reported in the literature is 2 mg/kg, ²⁵ the highest dose at which it has an established role is 0.50 mg/kg for purposes such as liver tumour visualisation. ²⁶ The reasoning for remaining within this lower range of ICG dosage is unclear, however it is possibly for the sake of cost and patient safety. In future research as well as clinical use, the ideal methodology for the initial assessment using NIR-ICG would thus involve a fixed amount of dye proportional to patient weight prior to surgery (0.25-0.30 mg/kg) followed by a longer waiting time before imaging (10-30 min). This extension of operating time was reported as negligible. ²² The use of additional boluses of ICG provided intra-operatively for repeated assessments throughout the surgery was shown to be a safe technique,^16,18 however its impact on visualisation was not adequately explored within the included studies.

Effectiveness of NIR-ICG in Intraoperative Detection of Endometriosis

The goals of laparoscopic surgery for endometriosis can be summarised into two main aims; to identify the disease, and to safely remove it. This review focuses on the former, and particularly how the intra-operative use of NIR-ICG can improve the gaps of current surgical techniques. As a purely diagnostic tool, NIR-ICG was not found to provide additional value to conventional WL imaging. That is, it was not shown to identify the disease in patients where the disease was unable to be visualised with WL imaging. However, this finding may be limited by the fact that the majority of included participants were of later ASRM disease stage, where there is a larger number of lesions which are easier to identify. Still, its utility is more likely to be in how it can enhance the visualisation provided with WL imaging. It was a common theme across these papers that despite improved detection rates, positive lesions identified through WL imaging were missed by NIR-ICG with a 2D camera to extents ranging from 15% ²⁰ to 65% ¹⁸ (Table 4). Thus, the possible clinical utility of NIR-ICG was suggested to only be in combination with WL imaging as either a confirmatory diagnostic tool, or to aid with more complete excision of the disease.

One previously identified barrier to obtaining complete excision of disease is the concept of occult endometriosis, which is histologically positive endometriosis in the peritoneum that is not normally visible on laparoscopy. Occult endometriosis was first reported in 1986 ²⁷ and has been subsequently confirmed over several studies including by Khan, et al. ²⁸ who reported endometriosis detected in biopsies taken from normal appearing peritoneum, at depths between 10-80 micrometres from the surface and more commonly in women with visually identifiable endometriosis. Furthermore, they reported that these lesions were active and capable of growth. The combination of these conclusions suggests that identifying occult endometriosis for excision is particularly important to prevent symptomatic recurrence for those patients undergoing elective surgical management of their disease. In seven out of the nine included studies in this review, endometriotic lesions were identified through NIR-ICG which were not visualised through WL imaging. The range of additional occult lesions identified across the studies ranged from one ²² to forty-two. ¹⁵ The proportions which these additional lesions represented are summarised in Table 4, describing an inconsistency in the capacity for NIR-ICG to identify occult endometriosis. It is important to note that the significant variability in the methodologies of the included studies could contribute to this inconsistency. As detailed, the dosage and timing utilised had significant impacts on the techniques efficacy and furthermore there was some variety in the histological definition of a positive lesion. Turco, et al. ²⁰ included both the presence of endometriosis foci and sclerosing inflammatory infiltrate in their definition of positive histology, and yielded a PPV of 98.5%, compared to a PPV of 68.8% by Siegenthaler, et al. ²² who only considered endometrial histology. The significance of these results is dependent on the clinical weight of a single unidentified lesion and on removing occult endometriosis, both of which are concepts not explicitly defined in the literature and reliant upon future research looking at long-term outcomes.

NIR-ICG in Detection of Deep Infiltrating Endometriosis

One of the factors that can make endometriosis difficult to detect during surgery is the heterogeneity of lesions in their location and appearances. Four of the included studies sub-categorised the identified endometriotic lesions as DIE or SPE.^17,19,20,22 Some of the locations where DIE was described to be able to be visualised with NIR-ICG include the: periureteral region, ovarian fossa, uterosacral ligament, rectum, rectosigmoid colon, retrocervical region and urinary bladder. The reduced accessibility of some of these locations makes DIE more difficult to visualise on laparoscopy, ²⁹ and provides a surgical challenge of resection without functional damage to the underlying structures.

Notably, NIR-ICG was found to be more effective in the identification of DIE, than for the identification of endometriosis as a whole. Cosentino, et al. ¹⁹ describes an increase in sensitivity from 82% to 89% when considering DIE alone. Furthermore, it was described to be more effective for the detection of lesions in the periureteral region, ovarian fossa and colorectal nodules, and less so for lesions of the bladder. Turco, et al. ²⁰ reported an increase in sensitivity with NIR-ICG from 64.5% to 93.5% for periureteral/ovarian fossa nodules and 86.7% to 90.0% for colorectal nodules, but a decrease from 76.2% to 71.4% for bladder nodules.

The reasons for the differences in detection rates could be due in part to several factors related to the mechanism of ICG. As previously described, NIR-ICG allows real-time visualisation of the dye within the vasculature allowing for visualisation of structures that are not at the surface. Thus, the effectiveness of the fluorescence is dependent on the desired lesion being more vascular than the surrounding tissue. Jayakumaran, et al. ¹⁵ describes that the vascular endometriotic lesions when visualised with NIR-ICG present as a fluorescent island with a surrounding white, fibrotic ring. It has been hypothesised that the pathophysiology of DIE differs from the other subtypes of endometriosis. For instance, DIE has been reported to have a greater degree of vascularity than peritoneal lesions. ³⁰ Furthermore, DIE of the rectum specifically has been reported to have a higher expression of vascular endothelial growth factor (VEGF) and thus vascularity than endometriosis of the bladder or ovaries, ³¹ which could explain the deficiency in the identification of bladder lesions. However, it is important to consider that a diversity of vascular patterns has been reported even within the category of rectosigmoid endometriosis (RSE). In a prospective pilot study utilising NIR-ICG, Raimondo, et al. ³² describes rectosigmoid lesions with both abundant and absent fluorescence indicating their respective hyper- and hypo-vascularity. In fact, they show a majority of RSE nodules (60%) presenting with a hypovascular pattern and an association of these hypovascular lesions with larger nodule size. Notably this study did report observing fluorescence of the RSE nodules up to 3 min. The hypo-vascularity could be explained by the possible role of fibrosis in the pathophysiology of endometriosis. That is, that chronic inflammation leads to the activation of signalling pathways for fibrosis which as it accumulates can both increase nodule size and reduce vascular density. ³² Moreover, several of the included studies excluded participants with ovarian or fallopian tube endometriosis because of the hypo- and hyper-vascularity of these structures respectively.^19,20 Understanding the different degrees of vascularity of lesions is important to establish the limitations of this technique.

Additional Roles of NIR-ICG in Endometriosis Surgery

While this review focuses on the role of NIR-ICG specifically for detection, it is worthwhile mentioning the alternative roles in which this fluorescent technique can be used in laparoscopic surgery for endometriosis, to understand how it can be integrated into practice. Siegenthaler, et al. ²² describe two of these alternative roles. The first is to assist in identifying the margins of resection of DIE, and the second is to allow visualisation of the ureters when excising ureteric endometriosis. Both these serve to maximise the protection of healthy underlying tissue and hence the function of these pelvic organs, preventing infertility and urological dysfunction. In a systematic review of the wider role of ICG in endometriosis surgery, Ianieri, et al. ¹¹ have previously concluded on the same utility for the evaluation of ureter perfusion assessment as well as intestinal anastomoses after segmental resection. Their conclusions about ureteral perfusion assessment are made based on a prospective case series by Raimondo, et al. ¹² which showcased the success of the imaging technique in confirming ureteral ischaemia suspected with white light imaging, prompting ureteral stent placement. There is a larger body of growing evidence supporting the use of NIR-ICG for surgery of RSE. The technique was first reported in 2018 in a video description of a bowel resection for the management of a large rectal endometriotic nodule. In this case, ICG was used to select the margins for rectal resection, and assess rectal vascularisation after the formation of the anastomosis. ³³ The feasibility of the technique was then tested in a larger preliminary study of thirty-two patients, which showed that NIR-ICG could be used in a safe and timely manner to identify sufficient perfusion at the anastomotic line. In this study, NIR-ICG prompted reinforcement of the anastomosis in one case and was followed by an anastomotic leak in another single case. ³⁴ While the majority of the literature surrounding NIR-ICG focuses on the qualitative role of the fluorescence technique, a cohort study utilising the angiography technique after full-thickness bowel resection for RSE by Raimondo, et al. ¹³ went further as to consider quantitative evaluation. Rather than relying only on an individual’s identification of fluorescence patterns, their quantitative analysis used video software to consider fluorescence as well as changes to the intensity of this fluorescence over time. Both qualitative and quantitative methods were established as effective for anastomotic perfusion assessment. However, quantitative compared to qualitative analysis also showed an increased specificity from 71 to 93% in predicting post-operative bowel fistula, suggesting a promising role for the detection of major post-operative complications.

Limitations

There are several limitations to this review. As mentioned earlier, there are very few included studies with limited sample sizes and varied methodologies. Additionally, due to the nature of the intervention, there was no blinding of the surgeons to the NIR-ICG leading to a risk of bias in identification outcomes. The identification of lesions is a highly qualitative, and hence difficult task. The methodologies of the included studies stressed the importance of having surgeons experienced in minimally invasive gynaecological surgery for endometriosis at the centre of these trials for this reason. However, this could also be considered a limitation in the produced evidence as it does not allow a representation of how the technique may have utility for surgeons with less experience. Furthermore, most of the studies only included patients who were in the later stages of the disease (ASRM III-IV), in which endometriosis is often easier to visualise on laparoscopy. Not only does this potentially influence results, it does not accurately represent the clinical need for improved detection tools for people with endometriosis in earlier stages. Lier, et al. ¹⁸ propose that the sensitivity and specificity of imaging techniques on detection could also have different degrees of importance at different stages of the disease. For instance, in earlier stages, specificity may be more essential to confirm the presence or absence of disease whereas in later stages sensitivity is required for more thorough excision. These nuances could not be established with the study populations represented in this review. Finally, many of the statistics reported are purely theoretical, as you cannot truly report on the sensitivity of NIR-ICG as the actual total known number of lesions is impossible to determine without a total peritonectomy. The specificity also depends on whether negative controls were taken, which was not always the case in the included studies.