Near-Infrared Fluorescence with Indocyanine Green to Assess Bone Perfusion: A Systematic Review

Background: Adequate perfusion of a bone flap is essential for successful reconstruction of osseous defects. Unfortunately, complications related to inadequate bone perfusion are common. Near-infrared fluorescence (NIRF) imaging enables intraoperative visualization of perfusion. NIRF has been investigated in reconstructive surgery to aid the surgeon in clinical perioperative assessment of soft tissue perfusion. However, little is known on the beneficial use of NIRF to assess bone perfusion. Therefore, the aim of this review was to search for studies evaluating NIRF to assess bone perfusion. Methods: A systematic review, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline, was performed. Studies up to October 2021 were included. We extracted data regarding the study population, size and design, reported objective fluorescence parameters and the methodology used for fluorescence imaging and processing. Results: Ten articles were included. Studies reported unevenly on the protocol used for NIRF imaging. Five studies reported objective parameters. Absolute and relative perfusion parameters and parameters derived from maximum fluorescence were reported. The clinical significance of these parameters has not been evaluated in humans. Conclusion: The evidence on bone perfusion as measured with NIRF is limited. More clinical studies are required.


Introduction
Adequate perfusion of tissue is a prerequisite for successful reconstructive surgery. The gold standard for tissue perfusion assessment in reconstructive surgery is based on perioperative clinical judgment of traditional parameters (skin colour, capillary refill and turgor, bleeding at wound edges) [1,2]. Bone perfusion is particularly difficult to clinically assess, as soft tissue related parameters are missing. Subjective observations such as bleeding of the periosteum or from multiple drill holes and bone edges are used. These observations however may be influenced by previous surgery, trauma or infection [3,4]. Unfortunately, complications related to inadequate tissue perfusion are common, despite clinical perioperative perfusion assessment [5][6][7]. Several techniques have been developed over the years to support surgeons in assessing the adequacy of tissue perfusion [8]. The ideal technique would be able to predict tissue viability by providing objective perfusion parameters. Near-infrared fluorescence (NIRF) imaging is a technique that enables intraoperative visualization of anatomic structures and perfusion of tissue after the administration Life 2022, 12, 154 2 of 11 of the fluorescent tracer indocyanine green (ICG) [9] (Figure 1). ICG is a non-specific, FDA approved, fluorescent tracer that binds to plasma proteins and is considered safe [9]. ICG has fluorescent properties when excited with light in the near-infrared range and can be visualized with an NIRF camera. This imaging method is rapidly gaining popularity within different fields of surgery and has been investigated in reconstructive surgery, mostly in breast reconstruction, showing promising results [10][11][12][13]. The advantage of NIRF compared to other real-time optical imaging, such as optical coherence tomography (OCT), sidestream darkfield microscopy (SDF), and laser speckle contrast imaging (LSCI), is its wide-field view, non-contact and less time-consuming technique [14]. Little is known about the use of NIRF to assess bone perfusion. The aim of this study is to present a comprehensive review of the literature concerning the assessment of bone perfusion with NIRF. The focus is to report which NIRF imaging methodology was used and whether objective perfusion parameters are available to allow differentiating between viable and non-viable bone in future research. ideal technique would be able to predict tissue viability by providing objective perfusion parameters. Near-infrared fluorescence (NIRF) imaging is a technique that enables intraoperative visualization of anatomic structures and perfusion of tissue after the administration of the fluorescent tracer indocyanine green (ICG) [9] (Figure 1). ICG is a nonspecific, FDA approved, fluorescent tracer that binds to plasma proteins and is considered safe [9]. ICG has fluorescent properties when excited with light in the near-infrared range and can be visualized with an NIRF camera. This imaging method is rapidly gaining popularity within different fields of surgery and has been investigated in reconstructive surgery, mostly in breast reconstruction, showing promising results [10][11][12][13]. The advantage of NIRF compared to other real-time optical imaging, such as optical coherence tomography (OCT), sidestream darkfield microscopy (SDF), and laser speckle contrast imaging (LSCI), is its wide-field view, non-contact and less time-consuming technique [14]. Little is known about the use of NIRF to assess bone perfusion. The aim of this study is to present a comprehensive review of the literature concerning the assessment of bone perfusion with NIRF. The focus is to report which NIRF imaging methodology was used and whether objective perfusion parameters are available to allow differentiating between viable and non-viable bone in future research. Figure 1. Perioperative photo of a patient who has a viable rectus femoris muscle flap (red inset) in a tibia after debridement of osteomyelitis. Because of ongoing infection and a PET scan suspicious for necrotic bone another surgical debridement was performed. Fluorescence imaging confirmed a lack of signal at a specific tibia region (blue inset) which was removed.

Methods
A systematic review, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline, was performed to assess current concepts regarding the perfusion assessment of bone with fluorescence imaging. A PubMed database search was conducted with the following search strategy: ( References of the selected articles were screened to identify additional papers. The search is up to date to October 2021. Articles written in English and Dutch were included. Two independent reviewers (MMi and MM) screened through the abstracts and selected articles relevant to this review. In case of discordance, a third reviewer (CD) was consulted until consensus was reached. Inclusion criteria were human, cadaveric or animal studies with an evidence level ranging from 1 to 5, investigating bone perfusion by ICG (qualify and quantify). There was no minimum number of participants. In vitro studies were excluded. Methodological quality and risk of bias of the included case series were assessed using the "Joanna Briggs Institute" (JBI) checklist. Level of evidence of the studies was assessed by the American Society of Plastic Surgeons (ASPS) Evidence Rating Scale. Figure 1. Perioperative photo of a patient who has a viable rectus femoris muscle flap (red inset) in a tibia after debridement of osteomyelitis. Because of ongoing infection and a PET scan suspicious for necrotic bone another surgical debridement was performed. Fluorescence imaging confirmed a lack of signal at a specific tibia region (blue inset) which was removed.

Methods
A systematic review, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline, was performed to assess current concepts regarding the perfusion assessment of bone with fluorescence imaging. A PubMed database search was conducted with the following search strategy: ( References of the selected articles were screened to identify additional papers. The search is up to date to October 2021. Articles written in English and Dutch were included. Two independent reviewers (MMi and MM) screened through the abstracts and selected articles relevant to this review. In case of discordance, a third reviewer (CD) was consulted until consensus was reached. Inclusion criteria were human, cadaveric or animal studies with an evidence level ranging from 1 to 5, investigating bone perfusion by ICG (qualify and quantify). There was no minimum number of participants. In vitro studies were excluded. Methodological quality and risk of bias of the included case series were assessed using the "Joanna Briggs Institute" (JBI) checklist. Level of evidence of the studies was assessed by the American Society of Plastic Surgeons (ASPS) Evidence Rating Scale.

Data Collection
From the included articles, we extracted data regarding study population, size and design, study end points, reported objective fluorescence parameters and the methodology used for fluorescence imaging and processing (camera type, positioning, ICG dosage, software used for data analysis). To provide comparable results within studies, we contacted the authors to obtain additional information.
We performed a descriptive and qualitative analysis of the data extracted; no metaanalysis was possible due to the heterogenicity of the studies included.

Results
The initial search in PubMed identified 481 articles. Of these, 464 were excluded because they were not relevant according to title or abstract. 17 articles were fully read, 10 articles met the inclusion criteria and were included into the study ( Figure 2). One article was excluded due to publication bias because the dataset was already included in another paper. Three articles were excluded because these focused on assessment of fluorescence of tissue other than bone (sarcoma, mucosa, flap pedicle), two articles were excluded because these used another method than NIRF with ICG to visualize bone (different probe for fluorescence). One article was excluded because it focused on visualization of necrotic bone.

Data Collection
From the included articles, we extracted data regarding study population, size and design, study end points, reported objective fluorescence parameters and the methodology used for fluorescence imaging and processing (camera type, positioning, ICG dosage, software used for data analysis). To provide comparable results within studies, we contacted the authors to obtain additional information.
We performed a descriptive and qualitative analysis of the data extracted; no metaanalysis was possible due to the heterogenicity of the studies included.

Results
The initial search in PubMed identified 481 articles. Of these, 464 were excluded because they were not relevant according to title or abstract. 17 articles were fully read, 10 articles met the inclusion criteria and were included into the study ( Figure 2). One article was excluded due to publication bias because the dataset was already included in another paper. Three articles were excluded because these focused on assessment of fluorescence of tissue other than bone (sarcoma, mucosa, flap pedicle), two articles were excluded because these used another method than NIRF with ICG to visualize bone (different probe for fluorescence). One article was excluded because it focused on visualization of necrotic bone. The characteristics of the included studies are presented in Table 1. The first study to report the use of ICG to assess bone perfusion was in 2012 by Nguyen et al. [15]. The number of cases per study varied between 2 and 39. Three studies reported on the NIRF assessment of bone perfusion in animals, three on cadavers and four in human beings. Only one prospective study in humans was identified [16]. Perfusion assessment of bone cortex and/or medulla with NIRF was reported in tibia, fibula, radius, iliac crest, femur and scapula.  The characteristics of the included studies are presented in Table 1. The first study to report the use of ICG to assess bone perfusion was in 2012 by Nguyen et al. [15]. The number of cases per study varied between 2 and 39. Three studies reported on the NIRF assessment of bone perfusion in animals, three on cadavers and four in human beings. Only one prospective study in humans was identified [16]. Perfusion assessment of bone cortex and/or medulla with NIRF was reported in tibia, fibula, radius, iliac crest, femur and scapula.

NIRF Imaging Methodology
Not all studies reported clearly on the protocol used for NIRF imaging. Six different imaging systems were used combined with differing NIRF imaging processing software suites ( Table 2). Position of the camera from the surgical field was 30 cm in most cases. The intravenous ICG dose, when reported, varied from 0.036 to 0.3 mg/kg or 7.5 mg to 25 mg. The duration of NIRF imaging video recording to obtain an ICG fluorescence curve over time was reported in two studies only and was 3 to 4 min.

Objective Perfusion Parameters
Quantification of NIRF imaging was reported only in five studies [15][16][17][18][19] (Table 3). Three studies reported on absolute perfusion units, defined as maximum fluorescence intensity after ICG injection [16,17,19]. Additionally, relative perfusion units were reported, which is the fluorescence intensity of the region of interest in relation either to another region of the flap where perfusion is considered to be optimal or to background fluorescence [15]. Three studies calculated several additional perfusion parameters.

Maximum Bone Fluorescence Intensity
Fichter et al. found that the maximum bone marrow fluorescence intensity before and after osteotomy decreased, but not significantly [16]. Valerio et al. described in their study that all their (chimeric) osseous flaps had an absolute fluorescence value higher than 6 on a 0-255 grey scale system assigned by the SPY-Q software [17]. The value on this scale is based on the fluorescence intensity of the image, with a higher number corresponding with a higher fluorescence intensity. The lower threshold of 6 was derived from earlier studies in skin sparing mastectomy patients [20]. The exact absolute fluorescence value per osseous flap was not described. NIRF was used in guiding the resection of inadequately perfused tissue. No complications related to inadequate bone perfusion were reported. A special kinematic model was used by Gitajn et al. in twelve porcine tibias with progressive conditions mimicking progressive bone trauma of increasing severity (Figure 3) [19]. Although maximum fluorescence intensity was not significantly different between normal and injured bone, the proprietary bone-specific kinetic model was able to discriminate between normal and injured bone. This is reported further in paragraph "Other parameters".

Relative Bone Fluorescence Intensity
Nguyen et al. described an intensity ratio (fluorescence at region of interest/background fluorescence) for quantification of fluorescence [15]. Vascularized bone marrow had a signal/background ratio of 2.30 and devascularized bone marrow, simulated by clamping the vascular pedicle, of 1.03.

Other Parameters
Two derived perfusion parameters were reported by Fichter et al., namely the wash in rate (WiR) (defined as the maximum fluorescence slope) and wash in perfusion index (WiPI) (defined as the ratio between the wash in area under the curve (WiAUC), defined within the start of rise of the fluorescence slope and the maximum fluorescence and rise time). Both parameters were significantly decreased before and after osteotomy [16]. No cut-off value of adequate perfusion was suggested as no correlation to clinical outcome was made.
Elliott et al. developed a kinetic analysis model to capture and describe the bipartite blood supply (periosteal and endosteal) to cortical bone [18]. They applied this kinetic model to the ICG fluorescence curves of regions of interest in bone to described recovered values for endosteal (late) and periosteal (early) blood supply to the bone. With the use of this model by Gitajn et al., recovered values of early, late and total blood perfusion (EBP, LBP, TBP) and late perfusion fraction (LPF) significantly differed between injured and normal bone [19]. A combination of LBP and LPF had a 90% sensitivity, 88% specificity and 89% accuracy in diagnosing injured bone.

Discussion
This systematic review reports on studies that have explored the role of NIRF imaging with ICG in assessing bone perfusion. The studies evaluated in this review show that it is a promising way of visualizing bone perfusion, although there is a clear lack of data on this technique.
The goal of real-time bone perfusion imaging in a clinical setting is primarily to differentiate between viable and non-viable tissue, to allow image-guided tailoring of osseous flaps or debridement of necrotic bone. An ideal imaging technique should allow clear differentiation between viable and non-viable tissue, be easy to set up, easy to use and interpret, be safe and minimally invasive for patients, provide consistent results between cases and ideally be cost-effective. Objective perfusion parameters could allow such differentiation, with a clear cut-off value indicating which tissue is deemed non-viable. While objective perfusion parameters derived from NIRF imaging have been reported previously for soft tissue, the evidence for objective levels of bone perfusion is scarce [13]. There is no consensus on which parameters should be used to ideally assess bone perfusion.

Relative Bone Fluorescence Intensity
Nguyen et al. described an intensity ratio (fluorescence at region of interest/background fluorescence) for quantification of fluorescence [15]. Vascularized bone marrow had a signal/background ratio of 2.30 and devascularized bone marrow, simulated by clamping the vascular pedicle, of 1.03.

Other Parameters
Two derived perfusion parameters were reported by Fichter et al., namely the wash in rate (WiR) (defined as the maximum fluorescence slope) and wash in perfusion index (WiPI) (defined as the ratio between the wash in area under the curve (WiAUC), defined within the start of rise of the fluorescence slope and the maximum fluorescence and rise time). Both parameters were significantly decreased before and after osteotomy [16]. No cut-off value of adequate perfusion was suggested as no correlation to clinical outcome was made.
Elliott et al. developed a kinetic analysis model to capture and describe the bipartite blood supply (periosteal and endosteal) to cortical bone [18]. They applied this kinetic model to the ICG fluorescence curves of regions of interest in bone to described recovered values for endosteal (late) and periosteal (early) blood supply to the bone. With the use of this model by Gitajn et al., recovered values of early, late and total blood perfusion (EBP, LBP, TBP) and late perfusion fraction (LPF) significantly differed between injured and normal bone [19]. A combination of LBP and LPF had a 90% sensitivity, 88% specificity and 89% accuracy in diagnosing injured bone.

Discussion
This systematic review reports on studies that have explored the role of NIRF imaging with ICG in assessing bone perfusion. The studies evaluated in this review show that it is a promising way of visualizing bone perfusion, although there is a clear lack of data on this technique.
The goal of real-time bone perfusion imaging in a clinical setting is primarily to differentiate between viable and non-viable tissue, to allow image-guided tailoring of osseous flaps or debridement of necrotic bone. An ideal imaging technique should allow clear differentiation between viable and non-viable tissue, be easy to set up, easy to use and interpret, be safe and minimally invasive for patients, provide consistent results between cases and ideally be cost-effective. Objective perfusion parameters could allow such differentiation, with a clear cut-off value indicating which tissue is deemed non-viable. While objective perfusion parameters derived from NIRF imaging have been reported previously for soft tissue, the evidence for objective levels of bone perfusion is scarce [13]. There is no consensus on which parameters should be used to ideally assess bone perfusion. In cadaveric studies the feasibility of ICG to penetrate cancellous (endosteal) portion of bone through penetrating periosteal capillaries has been shown [21][22][23]. NIRF with ICG should therefore allow adequate assessment of bone perfusion. NIRF with ICG has been used successfully to subjectively assess bone perfusion in two retrospective case studies [24,25]. Yoshimatsu et al. (2017 and even suggested that ICG could be used to define bone specific perfusion regions or bone angiosomes in new osseous flaps when injected into specific arteries. However, cadaveric studies that evaluate bone perfusion are limited and may over-or underestimate the vascular territory of bone flaps as the dynamic aspect of perfusion is not considered.
Five studies reported objective outcomes, including absolute, relative and extrapolated perfusion parameters. Some papers reported human studies and others animal research. Additionally, the NIRF imaging methodology used showed high heterogeneity between the studies included. For example, the studies used different camera systems with different camera chips and optics, resulting in varying camera sensitivities to pick up fluorescence. Moreover, several camera systems employ intensity optimization algorithms to boost the captured image data. Keeping these factors mind, there is a challenge to compare cut-off values between studies using different systems.
Absolute under several injury conditions. The underlying concept of this model is that bone ICG fluorescence curve represents a combination of "late" and "early" bone perfusion, which in turn, reflects the bipartite blood supply network of bone. Bone has a bipartite blood supply, based on a periosteal and an endosteal network [26][27][28]. These networks are connected through small capillaries. This intricate network allows sustained bone perfusion even when the nutrient artery providing the endosteal perfusion is lost, injured or not included in a bone flap [23,28]. With this model Gitajn et al. (2020) and Elliot et al. (2019) extrapolated several parameters and concluded that a combination of LBP and LPF values had a 90% sensitivity, 88% specificity and 89% accuracy in diagnosing injured porcine bone. However, it is difficult to extrapolate their methods and results because others have not used this model yet.
Recent studies suggest that NIRF with ICG in breast reconstruction is cost effective [29]. We did not find any cost-analysis studies on the use of NIRF to assess bone perfusion.
Future studies on NIRF assessment of bone perfusion should report objective outcome measures, including maximum and relative bone fluorescence intensity, after their use is confirmed in fundamental studies. When reporting absolute perfusion parameters, van den Hoven et al. propagated the normalization of the time-intensity curves to minimize the effect of multiple influencing factors, including camera distance and ICG dosage [30]. Bipartite bone perfusion raises the question on ideal NIRF camera position in relation to the evaluated bone. When the fluorescence camera is oriented perpendicular to a transverse bone cut, initially separate fluorescence images of periosteal, cortical and medullary bone could be observed. When the bone is viewed in a sagittal direction, fluorescence from periosteal, cortical and medullary bone will overlap and provide a combined view. The best way to assess bone, and whether there is a clinically relevant difference in evaluation remains unknown. Based on the available data, a camera position of 30 cm, perpendicular to the region of interest, and a dose of 0.1 mg/kg, followed by flushing with saline would be recommended [13]. NIRF depth assessment is dependent on the tissue optical density which is responsible for the absorption and scattering of light. Studies report on the depth of NIRF assessment of soft tissue (liver, skin and fat) being 3 to 15 mm [31,32]. We found no studies evaluating the depth of penetration of NIRF in bone tissue. It is thus unknown to what extent NIRF assessment is reliable in bone tissue. Nevertheless, a transverse cut of bone will be able to both assess periosteal, cortical and medullary bone perfusion. NIRF is able to record fluorescence inflow and outflow [33]. As perfusion is a dynamic process, recording should be conducted for at least 180 s to be able to record a perfusion (fluorescence) curve over time that includes inflow and outflow. To assess venous congestion, which is considered to be a great problem in reconstructive surgery, it could be considered to prolong the recording up to, for example, ten minutes. External parameters should also be included in analyses (temperature, systemic medication during the operation). Finally, it is also recommended to calibrate the system intensity response when several systems are used within a study.
It is a challenge to implement the NIRF findings of bone perfusion in surgery. Regarding reconstructive surgery, NIRF assessment of the bone flap after osteotomy, drilling and plating, e.g., in a case of mandibular reconstruction, will often not lead to an adjustment of the flap itself. However, consequent data of NIRF and clinical follow up, may eventually be useful in assessing the maximal length/surface and number of osteotomies of a bone flap. The more NIRF data available on healthy bone perfusion, the better we will be at assessing inadequate perfusion due to infection, trauma or radiotherapy. Ultimately, it may facilitate in guiding the debridement of an acceptor site before reconstruction.
One of the limitations of this systematic review was the limited amount and heterogeneity of studies included. Therefore, it was not feasible to perform a meta-analysis.

Conclusions
Based on the results of this review, the evidence on NIRF assessment of bone perfusion is limited but promising. A few objective parameters have been reported, but their clinical significance has not been evaluated in humans. All the reported objective values are not yet available in real-time, limiting their applicability during surgery. More studies are required to assess clinical significance and cost effectiveness of NIRF with ICG to differentiate between viable and non-viable tissue, to allow image-guided tailoring of osseous flaps or debridement of necrotic bone.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.