Biofluorescence (BF) is the physical event observed when electromagnetic radiation (light) is absorbed by a living organism and reemitted at a different wavelength in the visible spectrum. BF is a common phenomenon in the natural world and is observed in a huge variety of biological systems. BF has been unitised in a multitude of medical applications since the late 1940s, when fluorescein was first used to guide the removal of intra-cranial malignancies1. In 1959 indocyanine green (ICG) was approved for human use by the United States Food & Drug Administration and ICG is now utilised in a variety of clinical scenarios, including in colorectal surgery. Due to its pharmacokinetic properties and behaviour under near-infrared excitation, ICG predominates as the primary fluorophore in modern surgical practice, although considerable contention exists within the evidence base for its utilisation.

Despite the need for ongoing research, BF is finding an ever-more established place in routine colorectal practice and is at the forefront of advances in finding equipoise in strategies in surgical oncology. Real-time fluorescence imaging in intraoperative decision making for cancer surgery is advancing how we deliver surgical interventions to patients, in the expectation that both cancer-specific clinical outcomes and surgical morbidity may both be optimised2. Further developments in the dual technologies of ligand-specific fluorophores and artificial intelligence-led quantitative imaging provide an optimistic future for BF in molecular fluorescence guided surgery3.

Biofluorescence (BF) is the physical event observed when electromagnetic radiation (EMR; light) is absorbed by a fluorophore in a living organism and reemitted at a different wavelength, commonly in the visible spectrum. When electromagnetic radiation excites a fluorophore the molecule temporarily enters a higher-energy state and then relaxes to its resting state (known as the Stokes shift), resulting in the emission of a photon which can be detected with the naked eye or through a variety of sensory apparatus4.

Although classically described in marine ecosystems, there is a growing appreciation of the role of biofluorescence in terrestrial animals including fireflies, salamanders, and other amphibians; where it has a role in communication, sexual selection, and visual acuity5,6. Excitation in the natural world is predominated by sunlight, although it may be stimulated by chemical bioluminescence or other sources of EMR. There is limited evidence that BF plays a role in mammalian biology, although its application in medicine has developed over the past 75 years1.

In 1948, Moore utilised the green-yellow fluorophore fluorescein under ultraviolet (UV) excitation to facilitate surgery for intracranial malignancy. This early trial utilised a wavelength of ∼400nm by means of a CH-4 Mercury Vapor Lamp with a Wood's filter but was limited to visualisation ex vivo, and is a methodology that continues to be utilised in dermatology and ophthalmology clinical practice today7. However, despite demonstrating the potential of fluorescence technologies for guiding real-time intra-operative decision making, the ex vivo methods employed in this early work were clearly limiting. Further neurosurgical innovation led to the development of second-generation fluorophore aminolevulinic acid (5-ALA), which breaks down into the compound PpIX to emit violet-red (∼635nm) fluorescence after excitation with blue light at UV wavelengths of 375–440nm, but which can be used in vivo to facilitate in the visualisation of brain tumours8. Use of 5-ALA has demonstrated improved gross total resection over traditional neuronavigation on meta-analysis of almost 1000 cases over 17 years9,10. Although both of these techniques are used in today's clinical practice, the application of UV light is limited to relatively superficial or low-density tissues, as the penetration of light at this wavelength is limited, particularly in dense or pigmented tissues. An ideal fluorophore should be excitable and detectable at practical wavelengths, have predictable and safe pharmacokinetics and toxic profile, and be easily visualisable in target tissues. Methylene blue (MB), a thiazide dye with multiple medical applications, is also a fluorophore and can be excited at a peak of ∼668nm. However, unlike other fluorophores, MB is observable in the visual light spectrum (to the naked eye) and the Stokes shift is small, thus fluorescence is difficult to distinguish from background colour11. MB is also highly hydrophobic and demonstrates poor tissue penetration, further limiting its use as a medical fluorophore.

Indocyanine green (ICG) is an amphiphilic, tricarbocyanine iodine dye suitable for intravenous or intratissue injection. Following administration ICG binds to plasma proteins, primarily albumin, and is transported to the liver where it is excreted into bile via glutathione S-transferase12. The rate of excretion, and therefore intravascular half-life of ICG is determined by the rate of hepatocyte uptake following the rules of first-order kinetics, but is typically between 3 and 4 minutes, although is extended in cirrhosis and other liver diseases13. In fluorescent applications in vivo, the concentration of ICG should be kept below 15mg/L due to the molecule's tendency to aggregate at higher concentrations, leading to mitigation of effective fluorescence and reduced acuity14.

ICG has had US Food and Drug Administration approval sine 1956 for intravenous administration at a concentration of 2.5 mg ml−1, with doses of up to 25 mg in adults, 12.5 mg in children and 6.25 mg in infants15. Although ICG is minimally toxic (it can generate a low concentration of oxygen free-radicles under certain conditions), ICG is not suitable for patients with iodine allergy. ICG is excited with near-infrared (NIR) stimulation at a wavelength of ∼750-800nm and fluoresces at a peak of 823nm. A significant advantage of stimulation at these longer wavelengths is that NIR is able to penetrate more deeply (in excess of 10mm) and thus visualise targets below the surface of tissues; although there is some variability depending on tissue density, pigmentation, and the angle of incidence of the applied EMR16. Furthermore, NIR stimulation of background molecules such as haemoglobin and oxyhaemoglobin results in an effect that leads to greater tissue transparency and less interference with measurement of ICG fluorescence, resulting in greater acuity1. Each of these features makes ICG an optimal fluorophore for in vivo biological applications in medicine and surgery and has thus led to its adoption in a number of fields.

Initial use of ICG in studies of hepatic and cardiac physiology, and later in retinal angiography, drove the growth of ICG through the 1960’s and ‘70’s and validated its utility as a tool in assessing circulatory physiology and vascular anatomy17,18. Refinement in techniques in the application of ICG angiography (ICGA) led to expansion in the 2000’s to assessment of skin-flap viability in general and plastic reconstructive surgery. Semi-quantitative real-time assessment of skin perfusion by ICGA during reconstructive breast surgery has been proven to augment the ability of surgeons to predict, and therefore mitigate, poor flap perfusion above the performance of clinical judgement alone, as well as aid in flap design and vascular anastomotic assessment19. Semi-quantitative methods in ICGA have been further developed to facilitate the assessment of blood flow in a range of scenarios, including assessment of peripheral vascular disease, wound necrosis, and visceral perfusion20, 21, 22. The potential for FB to be utilised in the assessment of perfusion in visceral anastomosis has not been overlooked and has been adopted in small bowel, colorectal, and oesophogastric surgery for over 10 years23, 24, 25.

The excitation-sensor technology employed in NIR fluorescence surgery has developed in conjunction with the evolving application of ICG, and most systems are now tailored specifically to ICG fluorescence26. Most major biotechnology companies now offer NIR-ICG platforms integrated into standard white-light visual-spectrum systems to provide hybrid image overlay in open, laparoscopic, robotic, and endoscopic platforms.

ICG fluorescence has been utilised in colorectal surgery since the turn of the millennium when initial reports of its application as an alternative to India ink for colonic tattooing prior to surgery demonstrated that it was visualisable intraoperatively27. Ultimately, ICG's utility in tattooing is limited due to its relatively early washout from tissue, even after extravascular injection, although its potential for assessing colorectal perfusion has proved more fruitful.

Anastomotic leak (AL) following colorectal anastomosis is a significant and feared complication for both patients and surgeons, and has a significant clinical and economic burden28. Leak-rates vary across the literature but range from approximately 2% to 20% of cases depending on site of anastomosis and a number of other factors. Despite their frequency and a growing body of evidence to support our understanding the pathophysiology of a leak, surgeons have not proved at adept at predicting those cases which will go on to suffer an anastomotic leak29,30. However, poor blood flow to the anastomosis is one factor understood to be important in AL and is potentially amenable to intraoperative assessment and mitigation by the surgeon31. Following its applications in assessing other vascular beds, the potential role of ICG in assessing colonic vasculature in colorectal anastomosis has been investigated in a number of case series and randomised trials.

In an early multicentre non-randomised trial (PILLAR II; 2018), Jafari demonstrated that the leak rate following anterior resection in the group undergoing ICGA was 1.4% compared to 12% in the control group32. The use of ICGA in this study led to a change in the point of proximal colonic transection in 11 patients (8%), of whom, none leaked. The cohort of patients in this study were, however, relatively heterogeneous in terms of the primary pathology (including elective resections for diverticular disease as well as cancers) and the rate of inferior mesenteric artery (IMA) high-ligation. Furthermore, the mean height of anastomosis in this study was 10 ± 4 cm from the anal verge, but no sub-group analysis was performed to indicate whether ICG had a potentially more impactful effect on leak rate in lower, more high-risk, joins. Regarding low joins, the FLAG trial (Russia; 2020) randomised patients to ICGA vs visual assessment of colonic perfusion and found that although the use if ICGA was associated with a reduced risk of leak, this was only observed in low anastomosis, defined as being within 8cm of the anal verge33. Several studies have utilised ICG to examine the blood flow to the rectal stump rather than the colonic conduit to determine if this is a factor in anastomotic leak. Although some association was demonstrated between “delayed time to arterial [stump] perfusion” and anastomotic leak, the intraoperative assessment relied on semi-quantitative analysis of flow combined with vascular anatomy34. Other authors, who also examined leak following non-rectal anastomosis, suggest that leaks occurred most commonly in the subgroup whose anastomotic perfusion was via a marginal vessel rather than by the main native vessel (i.e. following high IMA ligation), but that a pragmatic course of action in such as case would be to prophylactically defunction such as patient and observe them closely postoperatively35. Despite the application of ICG and relatively complex perfusion analysis in these studies, the authors did not arbitrate for a change in anastomotic strategy and seem to argue simply for the accepted wisdom of protecting subjectively high-risk anastomoses and good clinical care.

Subsequent studies have, however, broadly been in keeping with findings of earlier and smaller trial data in supporting ICGA as a tool for reducing the risk of AL in colorectal anastomosis. A systematic review and meta-analysis of 10 studies conducted between 1998 and 2014 and including approximately 1400 patients found that the use of ICGA was associated with a reduced risk of AL (n = 23/693; 3.3 % (95 % CI 1.97-4.63 %) compared with no ICGA assessment (n = 19/223; 8.5 %; 95 % CI 4.8-12.2 %); although the studies were heterogenous in the methods of ICGA assessment and protocols for intraoperative decision making in the case of suspected poor perfusion36. None of the studies had unbiased assessment of the endpoints, nor appropriate power calculation. In 2022, a broadly inclusive systematic review and meta-analysis of over 11,000 patients by Safiejko demonstrated that the colorectal anastomotic leak rate in ICGA and non-ICGA groups was 3.7% vs. 7.6% (p < 0.001) in all trials; 8.1% vs. 12.1% (p = 0.04) in randomized controlled trials (RCTs); and 3.1% vs. 7.3% (p < 0.001) in non-RCTs, respectively37. However, despite the large numbers, there remains a significant degree of heterogeneity in the included studies, some of which included only a small number of patients, some included non-rectal anastomosis, and some were singe-centre or even single-surgeon case series. Furthermore, the multivariate regression analysis reported in the recent multicentre Phase III PILLAR trial (USA; 2021) did not demonstrate any significant difference in leak rates between ICGA and standard assessment of perfusion (OR = 0.845 (95% CI, 0.375-1.905); p = 0.34)38.

Currently, a number of randomised controlled trials are recruiting with the objective of determining the true utility of ICGA in reducing the AL rate in colorectal anastomosis: IntAct (UK & Europe), AVOID (Netherlands), and FLUOCOL-1 (France)39, 40, 41. Together, these trials aim to recruit almost 3000 patients, each has a clear protocol for utilisation of ICG and assessment of perfusion, intraoperative decision making, and have relevant clinical endpoints. Additionally, the IntAct trial will also include a sub-trial investigating the role of the gut microbiome in the pathophysiology of anastomotic leak as there is a growing appreciation that this a critical confounding factor42.

Beyond the assessment of colonic perfusion, ICG currently has a more straightforward role in aiding surgeons define at-risk anatomy during pelvic surgery; specifically, the ureters. Although only adopted sporadically and originally a technique borrowed from colleagues in gynaecology, the technique of injecting ICG into the ureters prior to pelvic colorectal resection has been reported to aid in the localisation of the ureters, particularly in challenging cases involving re-do surgery or sidewall dissection43,44. Although rates of ureteric injury in colorectal surgery are thankfully low, and therefore a reduction in risk by utilisation of ICG difficult to estimate, ICG may facilitate the identification of the approximately thirty percent of ureters that are “difficult to identify” under normal laparoscopic white-light illumination45. However, as ICG is metabolised in the liver and excreted in the bile, it must be introduced to the ureters via direct ureteral catheterisation prior to or during surgery. This is in contrast to the widespread utilisation of ICG in biliary surgery for guiding hepatic resection (including in surgery for colorectal metastasis), identifying bile leaks, and defining biliary anatomy at ductal surgery, where it can be conveniently injected intravenously46, 47, 48. Although there are no studies comparing the relative efficacy of ICG ureteric localisation versus prophylactic ureteric catheterisation/stenting in preventing ureteric injury, instillation of ICG may be performed quickly (mean time ∼10 minutes) and ICG remains visualisable in the ureters for over 8 hours49. There may also be an emerging role for ICG in surgery for endometriosis (both in localisation of nodules and perfusion assessment of treated organs); a surgical domain where colorectal surgeons not infrequently find themselves involved50,51.