Keywords
fluorescence imaging - ICG - bone perfusion
Adequate bone perfusion is key for successful surgery. It is important for healing
and consolidation, particularly in cases of traumatic crush injuries or when subsequent
radiotherapy is necessary.
Over the years, several techniques have been developed to support surgeons in assessing
adequate tissue perfusion. One of them is near-infrared fluorescence (NIRF) imaging,
using indocyanine green (ICG). ICG is a water-soluble fluorescence dye and when administered
intravenously, it binds to 98% plasma proteins and is therefore an ideal tracer to
measure perfusion. ICG has a half-life of 3 to 4 minutes, and it is cleared exponentially
by the liver with a clearance rate of approximately 20% per minute.[1]
[2] ICG is considered very safe for patients, since anaphylactic reactions are rare
with an incidence of 0.05%.[3] The light required for the excitation of ICG is generated by a light source that
is directly attached to a digital video camera with a specific filter. This light
source emits light between 750 and 800 nm, which is near the infrared range and excites
ICG, which can be viewed around the maximum peak of 832 nm.[1]
[4] With NIRF imaging using ICG, the absorption of ICG can be recorded real time during
surgery and allows perfusion assessment. Owing to its encouraging results, NIRF imaging
has grown in popularity across a variety of surgical specialties, including reconstructive
surgery.[5]
[6]
[7]
[8]
Visualizing bone perfusion is challenging. It would be of great interest to evaluate
the application of NIRF imaging using ICG for osseous tissue. For example, it could
be highly valuable when assessing vascularized bone flaps, especially when they need
to be osteotomized for head and neck reconstructions or extremity reconstructions
where adequate perfusion is key for consolidation. Moreover, it could improve the
early debridement of necrotic bone in extended trauma cases, may improve the identification
of sequesters, and could be useful in the removal of tissue affected by osteoradionecrosis.[9]
[10]
[11]
[12]
To provide objective perfusion parameters and fluorescence intensity curves of bone
tissue, this study investigated the perfusion of the human rib with NIRF imaging using
ICG. The rib is easily accessible when anastomosing a free flap in autologous breast
reconstruction. Bones have a bipartite blood supply consisting of endosteal and periosteal
networks, which are connected through small capillaries.[13]
[14]
[15] The endosteal blood supply is provided by the nutrient artery, which is a branch
of the posterior intercostal artery. The nutrient artery enters the medullary canal
of the rib just beyond the tubercle, which is located posteriorly.[16]
[17] The internal mammary vessels, which arise from the subclavian vessels, branch into
the inferior intercostal vessels ([Fig. 1]). The periosteal circulation of the ribs is based on dual blood supply provided
by the inferior intercostal vessels and the superior supracostal vessels. These vessels
have interconnecting arterioles covering the whole surface of the rib.[18] It is thought that the nutrient artery is necessary for survival of the rib.[19] Conversely, studies have shown that the viability of the rib is sustained on periosteal
blood supply alone.[20]
[21]
[22]
[23]
[24]
Fig. 1 Diagram showing the anatomy of the rib, including the internal mammary artery, also
known as the internal thoracic artery.
Objective fluorescence parameters of bone perfusion are sparse.[25] There are solely five studies in which outcomes are reported, including relative
perfusion and absolute perfusion.[26]
[27]
[28]
[29]
[30] One study reported the relative perfusion parameter defined as the fluorescence
intensity at a region of interest (ROI) divided by the background fluorescence.[26] Others defined absolute perfusion parameters as maximum fluorescence intensity over
time.[27]
[28]
[29] Some of these outcomes are investigated in human studies and others in animal studies.
Quantitative interpretation of fluorescence imaging has as main limitation that it
is subject to inter-user interpretation. Qualitative parameters may overcome this
limitation. There is no consensus on a standardized manner or which parameters should
be used to assess bone perfusion. However, dynamic perfusion parameters such a fluorescence
intensity curves seemed effective for evaluating perfusion.[31] To date, these dynamic parameters have not yet been studied for bone perfusion in
humans.
The aim of this pilot study is to assess the feasibility of fluorescence imaging of
the rib, in order to provide objective perfusion parameters of bone.
Materials and Methods
Participants
This feasibility study was carried out at a tertiary hospital in the Netherlands.
A total of 13 patients undergoing primary or secondary autologous breast reconstruction
from August 2021 to August 2022 were included in this study before surgery. The exclusion
criteria were the following: younger than 18 years, hyperthyroidism, autonomic thyroid
adenoma, epilepsy, renal failure with estimated glomerular filtration rate (eGFR) < 60,
severe liver failure, and patients who are allergic to ICG, iodine, or shellfish.
All the data were collected in Castor (CDMS version 2022.3). The study was approved
by the Ethics Committee of Amsterdam University Medical Center (2021.0142). All patients
provided written informed consent. Preoperative data recorded included patient's age,
height, weight, body mass index (BMI), medical history, history of smoking, and family
history. Intraoperative recorded data included vital parameters such as blood pressure,
heart rate, saturation, and use of vasopressors during administration of ICG.
Surgery and Fluorescence Imaging
In all patients undergoing autologous breast reconstruction, the medial portion of
approximately 2 cm of the second or third rib was removed to perform arterial and
venous anastomoses of the flap to the internal mammary vessels ([Fig. 2]). After revascularization, the rib was exposed, and a camera was positioned at approximately
30 cm above the rib and pointed transverse to the rib with the periosteum visible.
A dose of 0.1 mg/kg of ICG (Verdye 25 mg) was injected intravenously. Subsequently,
fluorescent intensity was captured by the Tivato 700 microscope (Carl Zeiss Meditec
AG, 2019, Jena Germany) for 4 minutes ([Fig. 3]). During the fluorescent assessment, ambient light was dimmed.
Fig. 2 Intraoperative photograph throughout autologous breast reconstruction of the right
breast, showing the visual (a) before and (b) after removal of the rib fragment in which the lateral cross-sectional surface becomes
visible. M., muscle.
Fig. 3 Near-infrared fluorescence (NIRF) imaging using indocyanine green intraoperative
in a patient showing the (a) visual and (b) NIRF fluorescence in the rib (X: endosteal region of interest; Y: periosteal region
of interest).
Quantification of the Fluorescent Signal
Postoperatively, video images were quantified using a tailor-made software written
in the Python v3.8 programming language (Python Software Foundation, https://www.python.org/). For endosteal measurement, the ROI of 5 mm was positioned on the cross-sectional,
lateral surface of the rib by the first author (D.F.B.). The ROI was also positioned
by a second observer (M.M.) to analyze interobserver reliability. Also, an ROI was
positioned on the anastomosed blood vessels where blood perfusion is considered to
be optimal and, if feasible, also on a region of bone with intact periosteum for periosteal
measurement. For all ROIs, the software generated time-intensity curves of the measured
intensity in arbitrary units (a.u.). From these curves, perfusion parameters were
extracted, which are illustrated in detail in [Fig. 4]. Ingress was defined as the increase in fluorescence intensity per second, from
baseline to maximum fluorescence intensity (Imax). Relative perfusion was defined
as the maximum fluorescence intensity of ROI at anastomosed blood vessel divided by
maximum fluorescence intensity at ROI. Mean slope was calculated as Δintensity/Δtime.
Normalized maximum slope is calculated by dividing the mean slope at the steepest
point of the ingress curve by the total slope of the ingress curve ((Δintensity/Δtime)/Imax
– I0). Egress was defined as the decrease in ICG fluorescence intensity per second,
from Imax until last measurement. The starting time of the curves (t0) was defined
as the first moment of increase in intensity compared to baseline.
Fig. 4 Time-intensity curve with extracted perfusion parameters. Imax, maximum intensity.
Tmax is the time at which the intensity is at its maximum.
Patients with videos of blurry image were excluded.
The median values of the measured intensities in the ROI for each video were calculated.
Statistical analysis was performed using IBM SPSS 26.0 (IBM, Armonk, NY, United States).
Normality was assessed using the Shapiro–Wilk test and data were presented as median
with minimum and maximum values. For continuous variables, the Wilcoxon signed-rank
test was used. Interobserver reliability was calculated using the intraclass correlation
coefficient with an absolute agreement definition.
Results
Patient Characteristics and Outcomes
Thirteen patients were included, of which 4 patients were excluded. Two were excluded
due to videos of blurry image and two were excluded because a different surgical technique
was used to remove the rib fragment.
As a result, 9 patients were included and 11 ribs were analyzed. From these patients,
there were seven deep inferior epigastric artery perforator (DIEP) reconstructions
including two bilateral DIEP reconstructions, one superior gluteal artery perforator
(SGAP) reconstruction, and one omental flap reconstruction. The video of patient 5
was only suitable for periosteal measurement and the videos of patient 2 and 5 were
suitable only for endosteal measurement.
Patient characteristics are displayed in [Table 1]. No adverse events related to ICG injection were observed intraoperatively. Intraoperative
vital parameters and patient specifications are shown in [Table 2].
Table 1
Patient characteristics
|
Characteristics
|
Number of patients (n = 9)
|
|
Mean age, y (SD)
|
53.1 (4.6)
|
|
Mean BMI, kg/m2 (SD)
|
28.2 (3.1)
|
|
Diabetes mellitus
|
0
|
|
Hypertension
|
2
|
|
Hypercholesterolemia
|
0
|
|
Active smoking
|
0
|
|
Type of autologous breast reconstruction
|
|
DIEP flap
|
7
|
|
SGAP flap
|
1
|
|
Omental flap
|
1
|
Abbreviations: BMI, body mass index; DIEP, deep inferior epigastric perforator; SD,
standard deviation; SGAP, superior gluteal artery perforator.
Table 2
Patient specifications
|
Characteristics
|
Vital parameters
|
|
Radiation therapy
|
Smoking
|
Cardiovascular comorbidity
|
Blood pressure (mm Hg)
|
Heart rate (beats/min)
|
Saturation (%)
|
|
Patient 1[a]
|
No
|
Former
|
No
|
95/55
|
57
|
100
|
|
Patient 2
|
No
|
No
|
No
|
113/59
|
65
|
100
|
|
Patient 3
|
No
|
No
|
No
|
100/60
|
80
|
98
|
|
Patient 4
|
No
|
No
|
HT
|
118/63
|
60
|
99
|
|
Patient 5
|
No
|
No
|
No
|
93/50
|
81
|
100
|
|
Patient 6
|
Yes
|
Former
|
No
|
115/60
|
65
|
98
|
|
Patient 7
|
No
|
No
|
No
|
96/53
|
46
|
100
|
|
Patient 8
|
No
|
Former
|
No
|
110/63
|
53
|
97
|
|
Patient 9[a]
|
No
|
No
|
HT
|
93/50
|
54
|
98
|
Abbreviations: HT, hypertension.
Note: Vital parameters were collected at the time of injecting of indocyanine green.
The patient is former smoker who quit smoking more than 1 month ago.
a The patient is undergoing a bilateral deep inferior epigastric artery perforator
and will continue as patient “a” and patient “b.”
Time-Intensity Curves
Time-intensity curves for endosteal measurement were generated for 10 ribs and are
shown in [Fig. 5]. There are two distinct patterns in the ICG ingress phase. First, a steep slope
reached within 10 seconds after t0 was observed in patients 2, 6, and 7. In other
patients, ingress was less steep and turned into a flattened slope.
Fig. 5 Time-intensity curves of endosteal measurement of the rib.
Regarding the ICG egress phase, three distinct patterns were observed. Three curves
showed a steep slope, which turned quickly into a flattened slope (patients 2, 6,
and 7). In four curves, egress was clearly prolonged (patients 1b, 3, 4, and 9b).
In the three remaining curves (patients 1a, 8, and 9a), egress had not begun within
the 240-second measurement period.
To summarize, patients 2, 6, and 7 had steep and well-defined ingress and egress.
In all other patients, the curves showed a much more flattened ingress and egress.
Quantitative Analysis
The outcomes of quantitative analysis on the time-intensity curves of the endosteal
and periosteal ROI are shown in [Table 3]. The median maximum intensity was 94.6 a.u. in the endosteal ROI as compared to
89.1 a.u. in the periosteal ROI (p = 0.889). The median mean slope and the median normalized maximum slope were slightly
higher in the periosteal ROI (0.8 vs. 1.2, p > 0.726; 0.1 vs. 1.5, p > 0.161. Among the observers (D.F.B. and M.M.), there was an excellent agreement
about the positioning of the ROI with an intraclass correlation coefficient of 94.3%.
Table 3
Quantitative assessment NIRF imaging
|
Parameter of ingress
|
Endosteal
|
Periosteal
|
|
Median (minimum–maximum)
|
Median (minimum–maximum)
|
|
Imax (a.u.)[a]
|
94.6 (41.5–182.4)
|
89.1 (44.4–110.71)
|
|
Time from t0 to Imax (s)
|
54.4 (1.8–230.8)
|
59.4 (1.0–224.3)
|
|
Relative perfusion (%)
|
2.5 (1.2–4.7)
|
3.1 (1.7–4.6)
|
|
Mean slope
|
0.8 (0.2–17.4)
|
1.2 (0.2–63.9)
|
|
Normalized maximum slope
|
0.1 (0.0–3.9)
|
1.5 (0.1–3.1)
|
|
Parameter of egress
|
|
Number of ribs (n = 10)
|
Number of ribs (n = 9 ribs)
|
|
90% Imax
|
7
|
6
|
|
80% Imax
|
6
|
6
|
|
70% Imax
|
5
|
5
|
|
60% Imax
|
2
|
3
|
|
50% Imax
|
2
|
1
|
Abbreviations: Imax, maximum intensity; NIRF, near-infrared fluorescence.
a Median maximum intensity at the blood vessel was 224.0 a.u. (163.6–274.7)
Discussion
In this pilot study, we were able to quantify perfusion of the human rib with NIRF
imaging using ICG. This was shown with measurements in two different ROIs: endosteal
and periosteal. According to several anatomical studies, the osseous blood supply
is bipartite and depends on the endosteal and periosteal blood supply.[14]
[15]
[32] In the 11 measurements in the study, the endosteal and periosteal parameters show
much agreement with no statistically significant differences. Numerous clinical applications
in soft-tissue surgery have been studied for NIRF imaging using ICG.[33]
[34]
[35]
[36] There are only five previous studies that report on objective NIRF outcomes for
bone perfusion.[26]
[27]
[28]
[29]
[30] It is notoriously difficult to assess the viability of bone based on clinical signs,
but accurate debridement of nonviable bone is crucial. The relative perfusions of
2.5% (endosteal) and 3.1% (periosteal) confirm that perfusion of the rib is quite
low as compared to the perfusion at the level of the anastomosed blood vessels.
Thorough debridement without wasting additional bone is extremely important to accomplish
optimal bone healing, bone reconstruction, and cure. Fluorescence imaging may play
a role in debridement after trauma injury of the extremities, long-lasting infectious
disease of the tibia, femur, and humerus, or osteoradionecrosis after breast (rib),
rectum (sacrum), and oropharyngeal tumors ((neo-)mandible).[12] Moreover, it may have an interesting role in osseous reconstructions such as free
vascularized fibula grafts, especially when multiple osteotomies are necessary.[28] The scientific evidence for using NIRF to visualize bone perfusion is sparse.[25] Yoshimatsu et al evaluated bone perfusion in cadaveric femoral medial condyle.[37] They compared the penetration depth of methylene blue with NIRF imaging using ICG.
Following injection of methylene blue and ICG into the descending genicular artery,
the cancellous area was visible with NIRF imaging due to tiny perforators that penetrated
the periosteum in contrary to the blue dye that was solely visible in the periosteum.
Nguyen et al was able to demonstrate that endosteal perfusion and viability of vascularized
bone flaps can be assessed with NIRF imaging using ICG.[26] This was shown in osteomyocutaneous forelimb flaps and fibula flaps of female Yorkshire
pigs. They compared a devascularized bone flap, in which the pedicle had been ligated,
to a vascularized bone flap using NIRF imaging. The vascularized flap showed NIRF
perfusion at the osteotomy site, whereas the devascularized flap showed a lack of
fluorescence. Gitajn et al demonstrated in porcine models that bone perfusion can
be measured quantitatively from endosteal and periosteal sources using NIRF imaging
with ICG.[29] Fichter et al[28] and Gitajn et al[29] defined absolute perfusion values as maximum fluorescence in number of units at
a specific moment in time and extracted time-intensity curves, whereas Valerio et
al[27] focused on maximum fluorescence at a single unspecified time point without extracting
time-intensity curves. Gitajn et al[29] and Elliot et al[30] studied ICG fluorescence curves of ROIs in bone under various damage situations,
and a new kinetic model was created and used. The underlying idea behind this model
is that the bone ICG fluorescence curve reflects both “early” and “late” bone perfusion
and once again represents the bone's bipartite blood supply network. Gitajn et al
demonstrated that when the periosteal blood supply was disrupted by stripping soft
tissue from the bone, maximum fluorescence intensity decreased by 50% and time to
reach maximum fluorescence intensity increased.[29] This suggests that “late” bone perfusion is connected to endosteal blood supply,
while “early” bone perfusion and periosteal blood supply are connected to one another.
Absolute and relative perfusion values do not offer insight into how intensity changes
over time. Additionally, absolute perfusion is dependent on the measured fluorescence
intensity and hence prone to several influencing factors including camera distance
and camera angle.[31]
[38] Among the endosteal and periosteal ROIs that we obtained, the extracted time-intensity
curves demonstrated several patterns of ICG ingress and egress. In the endosteal view,
a steep ingress was recognized with a steep egress in three ribs and the majority
of the observed curves demonstrated a prolonged egress. In others, ingress is less
steep and egress is flat or does not start at all. For several subjects in this study,
the 4-minute measurement period was insufficient to observe the initiation of venous
outflow. In other studies, these flat curves with slow ingress and slow or absent
egress represent bad perfusion with suboptimal inflow and outflow.[39]
[40]
[41]
[42]
Quantitative perfusion analysis in studies focusing on the esophagus, ileum, and colon
tissue often shows a good inflow and a well-defined egress in case of good tissue
perfusion. We find these curves in three of the ribs that we studied. The majority
of the ribs, however, showed a slow ingress and a flat or absent egress. This may
be explained by a more medial ostectomy, at the level where the rib is most cartilaginous.
Cartilage is notoriously badly perfused. As a result, the intensity curves in cartilaginous
bone tissue may be different from the curves that are often found in well-vascularized
tissue during gastrointestinal surgery.[39]
[40]
[41]
[42] The well-defined curves with a steep ingress and well-defined egress may match the
patients with a more spongy rib or patients in whom the ostectomy of the rib was extended
laterally. However, this cannot be confirmed in hindsight. Other factors that may
influence the shape of the curve are iatrogenic damage of the periosteum during surgical
dissection, environmental influences such as light, heat, and manipulation leading
to vasoconstriction, or blood dripping onto the ROI.
Some limitations were observed in this study. The measurement of maximum fluorescence
intensity is influenced by multiple environmental factors, such as camera distance,
camera angle, ambient light, blood pressure, and use of intraoperative medication.
This may also explain why the fluorescence intensity curves do not start from zero.
Due to the small sample size of this study, it was not possible to establish the effect
of all these parameters on fluorescence intensity and its dynamics. Moreover, it was
hard to position the camera adequately to capture an endosteal and periosteal ROI
in one view. This resulted in two ribs in which only one ROI was visible. Despite
these limitations, this study provides insight into the possibility of quantification
of bone perfusion, showing promising results. This study demonstrated that it is challenging,
but feasible to use NIRF imaging to study the rib. This is a step toward the use of
NIRF imaging with ICG to provide surgeons with quantitative parameters for assessing
bone perfusion.
Based on our first experience of quantifying rib perfusion, we recommend the following:
keep all external parameters stable, prolong the video for more than 5 minutes, and
ensure there is an ROI with sufficient perfusion available in the field of view. A
larger cohort is needed to investigate the value of the inflow parameters in the assessment
of bone perfusion to correlate divergent parameters to patient-specific factors and
for the prediction of clinical outcomes. To minimize movement artifacts due to breathing,
bone perfusion in, for example, a healthy fibula graft should be measured. Taken this
into account, it should be possible to establish reliable cutoff values for normal
bone perfusion. Therefore, more research is needed to investigate the possibilities
of using NIRF imaging with ICG to assess bone perfusion intraoperatively. Cutoff values
are needed to guide a surgeon in the debridement of affected bone or reconstructive
surgery with vascularized bone.
Conclusion
This study demonstrated the feasibility of quantification of perfusion in human ribs
using NIRF imaging with ICG. The result of our study suggests that NIRF imaging using
ICG can provide surgeons with objective parameters for assessing bone perfusion. However,
implementing our NIRF imaging results of bone perfusion into surgery remains a challenge.
