Laser-assisted near-infrared angiography was performed after tumor extirpation and before pharyngeal reconstruction. A, White light image demonstrates the remaining native pharyngeal mucosa in this patient undergoing salvage laryngectomy. B, Grayscale fluorescence image was captured using the SPY imaging system approximately 15 seconds after intravenous indocyanine green dye injection to assess mucosal perfusion. C, Onboard software was then used intraoperatively to quantify relative perfusion values by manually assigning a reference point of maximal perfusion in the center of the mucosa, then selecting multiple points along the cut edges of the mucosa.
White light (A), grayscale fluorescence (B), and heat map fluorescence (C) images of a patient undergoing salvage laryngectomy after tumor extirpation with relatively well-perfused native mucosal edges, as assessed by laser-assisted near-infrared angiography. D, The lowest mean indocyanine green dye percentage of mucosal perfusion was 22.0% in the fistula group vs 34.9% in the nonfistula group (absolute difference, 12.9%; 95% CI, 5.1%-21.7%). Error bars indicate SD.
The ROC curve for the lowest percentage of indocyanine green dye mucosal perfusion with an area under the curve of 0.85 (95% CI, 0.72-0.98), which was significantly greater than the chance diagonal. The 3 lowest percentage of perfusion threshold values are shown with the associated specificity and sensitivity.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Partington EJ, Moore LS, Kahmke R, et al. Laser-Assisted Indocyanine Green Dye Angiography for Postoperative Fistulas After Salvage Laryngectomy. JAMA Otolaryngol Head Neck Surg. 2017;143(8):775–781. doi:10.1001/jamaoto.2017.0187
Can indocyanine green dye angiography be used to determine pharyngofistula formation in patients undergoing salvage laryngectomy?
In this cohort study of 37 patients, patients who developed a fistula after salvage laryngectomy had a significantly mean lower indocyanine green dye mucosal perfusion vs those who did not develop a fistula.
Indocyanine green dye angiography can be used to evaluate pharyngeal mucosa after salvage laryngectomy and anticipate fistula formation.
Pharyngocutaneous fistula formation is an unfortunate complication after salvage laryngectomy for head and neck cancer that is difficult to anticipate and related to a variety of factors, including the viability of native pharyngeal mucosa.
To examine whether noninvasive angiography with indocyanine green (ICG) dye can be used to evaluate native pharyngeal vascularity to anticipate pharyngocutaneous fistula development.
Design, Setting, and Participants
This cohort study included 37 patients enrolled from June 1, 2013, to June 1, 2016, and follow-up was for at least 1 month postoperatively. The study was performed at the University of Alabama at Birmingham, a tertiary care center. Included patients were those undergoing salvage total laryngectomy who were previously treated with chemoradiotherapy or radiotherapy alone.
The ICG dye was injected intraoperatively, and laser-assisted vascular imaging was used to evaluate the native pharyngeal mucosa after the ablative procedure. The center of the native pharyngeal mucosa was used as the reference to compare with the peripheral mucosa, and the lowest mean ICG dye percentage of mucosal perfusion was recorded for each patient.
Main Outcomes and Measures
The primary outcome was the formation of a postoperative fistula, which was assessed by clinical and radiographic assessment to test the hypothesis formulated before data collection.
A total of 37 patients were included (mean [SD] age, 62.3 [8.5] years; 32 [87%] male and 5 [14%] female); 20 had a history of chemoradiotherapy, and 17 had history of radiotherapy alone. Thirty-four patients (92%) had free flap reconstruction, and 3 had primary closure (8%). Ten patients (27%) developed a postoperative fistula. No significant difference was found in fistula rate between patients who underwent neck dissection and those who did not and patients previously treated with chemoradiotherapy and those treated with radiotherapy alone. A receiver operator characteristic curve was generated to determine the diagnostic performance of the lowest mean ICG dye percentage of mucosal perfusion determined by fluorescence imaging, which was found to be a threshold value of 26%. The area under the curve was 0.85 (95% CI, 0.73-0.97), which was significantly greater than the chance diagonal. The overall mean lowest ICG dye percentage of mucosal perfusion was 31.3%. The mean lowest ICG dye percentage of mucosal perfusion was 22.0% in the fistula group vs 34.9% in the nonfistula group (absolute difference, 12.9%; 95% CI, 5.1%-21.7%).
Conclusions and Relevance
Patients who developed postoperative fistulas had lower mucosal perfusion as detected by ICG dye angiography when compared with patients who did not develop fistulas.
Ablation of head and neck cancers frequently results in major defects that can cause significant functional and cosmetic impairments. Advances in reconstructive surgery through the use of local flaps, pedicled flaps, and microsurgical free flaps have helped to improve these impairments but can often lead to serious complications. For example, pharyngocutaneous fistula formation is the most common major complication after total laryngectomy, occurring in approximately 16% to 25% of cases, although rates as high as 65% have been reported in the literature.1,2 In addition to increasing morbidity and delaying initiation of adjuvant therapy, fistula formation carries a substantial financial burden estimated at a mean of $24 915.00 in additional costs per patient.3 Certain patient factors, such as a history of radiotherapy, prior tobacco use, the presence of cardiovascular or peripheral vascular disease, and previous surgery, can alter normal wound healing and potentially lead to poor results.4
In patients undergoing salvage laryngectomy, several factors should be considered to prevent fistula formation. In those patients in whom free tissue is used, flap perfusion is arguably the most critical component but the nature of the native tissue being reconstructed must be considered. Poor tissue perfusion associated with salvage laryngectomy is perhaps the most difficult to assess and presents a high risk of fistula formation.
Laser-assisted near-infrared (NIR) angiography using the NIR fluorescent dye indocyanine green (ICG) with an NIR imaging system (such as the SPY device manufactured by Novadaq Technologies) can provide multiple, real-time, intraoperative assessments of large- and small-caliber blood vessels. Laser-assisted NIR angiography allows for immediate intraoperative assessment of tissue perfusion. The NIR angiography uses ICG dye, a nontoxic, amphiphilic molecule that binds to plasma proteins, which is excellent for assessing perfusion because the molecule remains almost entirely in the intravascular space. It is metabolized in the liver and excreted in the kidneys without any association of hepatic or renal toxic effects. The half-life of ICG dye is very short in humans (3-5 minutes), allowing for multiple uses during the same procedure. The risk of anaphylactic reaction to ICG dye is 1 in 42 000 and is associated with iodine allergy. Because of this low risk, multiple ICG dye injections can be given without ever reaching toxic levels. The type of laser or light diode is safe and has no potential for causing damage to human skin or tissues. Special patient precautions, such as wet towels around the skin, protective eyewear, or laser-safe endotracheal tubes, are not required.
The NIR imaging allows for a real-time evaluation of perfusion intraoperatively. This evaluation can be as subjective as looking at the screen and determining areas that are underperfused, or it can be objective and quantitative. Postimaging software is available and provides analysis tools, including different algorithms and methods for measurement of fluorescence intensity. Tissue perfusion can be assessed only in a qualitative manner because the absolute values are not comparable among patients. For quantitative analysis, values are assigned relative to the amount or intensity of fluorescence in another region. Postimaging software allows for quantitative analysis of perfusion and can decrease subjectivity and increase interobserver reliability.5 The goals of this study are to describe and report our use of laser angiography with ICG dye to evaluate native mucosa perfusion in salvage laryngectomy and how this affects pharyngocutaneous fistula formation. We also hope to use ICG dye angiography to determine a perfusion threshold that can anticipate fistula formation.
Patients undergoing salvage laryngectomy from June 1, 2013, to June 1, 2016, at the University of Alabama at Birmingham were enrolled in the study. Inclusion criteria included total laryngectomy with a history of primary chemoradiotherapy or radiotherapy. Standard demographic data were collected, including medical comorbidities, tumor subsite, prior treatment, time since prior treatment, location of recurrence, whether neck dissections were performed, smoking status, and whether a fistula developed. Comorbidities evaluated were cardiovascular (coronary artery disease, cardiomyopathy), pulmonary (chronic obstructive pulmonary disease, asthma, and emphysema), hypertension, hyperlipidemia, diabetes, gastroesophageal reflux, neurovascular disease (history of cerebrovascular accident, transient ischemic attack), and liver dysfunction. Patients were followed up throughout their hospital course and for at least 4 weeks postoperatively for complications. It is the policy of the University of Alabama at Birmingham to reconstruct the pharyngeal defect in patients undergoing salvage laryngectomy with a patch radial forearm free flap. In certain rare patients, such as those who are very ill or who have severe vascular disease, primary closure is used. For all the cases in this study, the native mucosa was determined to be at least 3 cm in width at the narrowest point of tension-free mucosa, although exact widths for each case were not recorded. The study was approved and informed consent deemed not necessary by the University of Alabama at Birmingham Institutional Review Board. All data were deidentified.
After tumor extirpation and before pharyngeal reconstruction, intravenous ICG dye was administered to the patient. Per a US Food and Drug Administration–approved institutional protocol, 25 mg of ICG dye was reconstituted in 10 mL of 0.9% normal saline for a concentration of 2.5 mg/mL. An open-field imaging device was then positioned over the tissue. Then 10 mg of ICG dye (4 mL) was administered through a peripheral intravenous catheter followed by a 10-mL 0.9% normal saline flush. Image capture began once the ICG dye was visualized in the vasculature on the imaging system, which, in our experience, took approximately 10 to 15 seconds. The edges of native pharyngeal mucosa, where there would be the highest likelihood of fistula formation, were evaluated using the NIR laser angiography capturing system (SPY imaging system). Multiple regions of interest (ROIs) were manually selected and assigned a relative percentage by onboard device software based on a manually selected reference point in the center of the native pharyngeal mucosa to quantitatively assess the perfusion of mucosal edges (Figure 1). Adjacent native pharyngeal mucosa was used as a reference tissue to standardize tissue assessment and eliminate confounding variables in NIR imaging arising from inherent differences in properties among various tissue types. The reference point was assigned in the midline of the posterior pharynx, midway between the base of tongue and esophageal inlet. The ROIs were taken approximately every 1 cm spanning the entire distance of the cut mucosal edge, yielding approximately 20 ROIs per patient. The ROI with the least perfusion at the edges was then reported as the lowest mean ICG dye percentage of mucosal perfusion.
For demographic data, standard statistical methods were used. A 2-tailed, unpaired t test was used to analyze quantitative data, whereas a Fisher exact test was used to analyze proportional data using GraphPad Prism software, version 6.0 (GraphPad Software Inc). The mean (SE) minimum percentage of perfusions was calculated using the same software. A receiver operator characteristic (ROC) curve was generated to determine the diagnostic performance of the lowest mean ICG dye percentage of mucosal perfusion determined by fluorescence imaging to predict postoperative fistula formation. The ROC curve was generated using GraphPad Prism software, version 6.0. The area under the curve (AUC) was estimated with bootstrapped 95% CIs.
Thirty-seven patients (mean [SD] age, 62.3 [8.5] years; 32 [87%] male and 5 [14%] female) who underwent salvage laryngectomy were included in the study (Table 1). Thirty-four patients (92%) underwent free flap reconstruction, and 3 (8%) had primary closure. All patients had prior treatment, with 20 (54%) having chemoradiotherapy and 17 (46%) having radiotherapy alone. The mean time between primary treatment and salvage laryngectomy was 38.2 months (range, 3-264 months). Presence of a fistula was determined by clinical evidence of a fistula within 1 month of the procedure and by esophagography performed at postoperative days 6 to 7 for all patients. Ten patients (27%) developed a postoperative fistula. As indicated in Table 2, there were observed differences in fistula rates when comparing microvascular free flap reconstruction (9 [27%]) vs primary closure (1 [33%]) (observed difference, 6%; 95% CI, –25% to 54%), neck dissection (7 [25%]) vs no neck dissection (3 [33%]) (observed difference, 8%; 95% CI, –20% to 42%), and chemoradiotherapy (6 [30%]) vs radiotherapy only (4 [24%]) (observed difference, 6%; 95% CI, –22% to 32%). None of the differences in observed values are statistically significantly different from the null value. The large width of the CIs implies imprecision in the observed effect and undermines the ability to make any definitive conclusion regarding the association between these factors and fistula formation. All patients in the study had some degree of smoking history, with 14 (38%) reportedly being current smokers at the time of surgery (Table 1). The observed difference in the fistula rates between former (7 [30%]) and current smokers (3 [21%]) was 9% (95% CI, –21% to 34%) (Table 2). In addition, no difference was found in tumor subsite, time since treatment, or medical comorbidities with the exception of diabetes, which had a higher risk of fistula. All surgical mucosal margins were deemed pathologically negative for all patients included in this study. There were no flap failures or return trips to the operating room for flap revision for any of the patients in this study.
The ICG dye angiography data revealed that the overall lowest mean ICG dye percentage of mucosal perfusion was 31.3%. In the fistula group (n = 10), the lowest mean ICG dye percentage of mucosal perfusion was 21.5% (range, 16.0%-29.0%) vs 34.9% (range, 9.0%-59.0%) in the nonfistula group (n = 27) (absolute difference, 12.9%; 95% CI, 5.1%-21.7%) (Figure 2). Various cutoff values for the lowest mean ICG dye percentage of mucosal perfusion were then evaluated to determine the sensitivity and specificity of the test for each value. Using our overall mean of 31.3% as a starting point, we chose to evaluate 3 different cutoff values around this range that had acceptable sensitivities and specificities. With the use of less than 30% as the lowest mean ICG dye percentage of mucosal perfusion, there was a 100% sensitivity (95% CI, 69.2%-100%) and a 70.3% specificity (95% CI, 49.8%-86.2%) for anticipating fistula formation. With less than 26% as the lowest mean ICG dye percentage of mucosal perfusion, sensitivity was 90% (95% CI, 55.5%-99.8%), and specificity was 74.1% (95% CI, 53.7%-88.9%). Finally, with use of less than 24% as a cutoff, the sensitivity was 80.0% (95% CI, 44.4%-97.5%), and the specificity was 85.2% (95% CI, 66.3%-95.8%) (Figure 3). To statistically evaluate the performance of the lowest mean ICG dye percentage of mucosal perfusion determined by fluorescence imaging to predict postoperative fistula formation, we generated a ROC with empirical 95% CIs for the AUC. The AUC was 0.85 (95% CI, 0.73-0.97), and there was a signiﬁcantly greater AUC than the chance diagonal with a 95% CI (Figure 3).
In this study, we demonstrated that ICG dye angiography may be effective for evaluating native mucosa during salvage laryngectomy and found a significant difference between the lowest mean ICG dye percentage of mucosal perfusion in the group who developed pharyngocutaneous fistula and the group who did not. These data suggest that native mucosal perfusion has a distinct effect on outcomes. After these results are confirmed and validated in future studies, the addition of ICG dye angiography could be considered along with microvascular tissue perfusion in the determination of the likelihood of reconstructive success.
In this study, no statistically significant or clinically meaningful difference was found in fistula formation between patients who underwent chemoradiotherapy vs radiotherapy alone, neck dissection vs no neck dissection, and tumor subsite or free flap reconstruction vs primary closure. There was also no association with any medical comorbidities or smoking status with the exception of diabetes, although only 2 patients with diabetes were included in this study, making it difficult to draw conclusions. The large width of the CI, attributable to the small sample size and limited number of fistulas, demonstrates that none of the observed effects are very precise; thus, we are limited in drawing any meaningful conclusions from these results.
We also sought to investigate various thresholds for the lowest mean ICG dye percentage of mucosal perfusion to determine the sensitivity and specificity of this test. The ROC analysis revealed the threshold approach to be a powerful diagnostic tool. The AUC, which is a measure of diagnostic accuracy, for this test (0.85) was greater than the AUC for diagnostic tests routinely used in clinical practice, including positron emission tomography standard uptake values (0.81-0.9), carcinoembryonic antigen antibody (0.6-0.7), and prostate-specific antigen (0.6-0.7). On the basis of our study, it would appear that using 26% as a cutoff value with 90% sensitivity and 74% specificity (7 false-positive results, 1 false-negative result) is a reasonable benchmark for evaluation of the native mucosa. In this study, only 1 patient in the fistula group would have failed this cutoff value (lowest ICG percentage of mucosal perfusion was 29%). With our data, less than 26% perfusion would likely have a high accuracy of predicting viable native mucosa. In addition, our study found that a lowest mean ICG dye percentage of mucosal perfusion threshold of 30% could be safely used with very low likelihood of pharyngocutaneous fistula formation because this cutoff value yielded 100% sensitivity. We hypothesize that, when possible, native mucosa with low perfusion could be resected to optimize native tissue viability and decrease likelihood of fistula formation. This is an area that will require further investigation and validation.
Pharyngocutaneous fistula is a significant morbidity for patients undergoing salvage laryngectomy. Risk factors include prior chemoradiotherapy, salvage laryngectomy within 1 year of treatment, and poor nutritional status.6-9 Other factors, such as concurrent neck dissection, smoking status, and age, have been associated with overall complications after salvage laryngectomy.10 Concurrent neck dissection and its association with fistula formation are still controversial, but studies9,10 have found a trend toward association with fistula. Tumor subsite, particularly the supraglottis, has also been associated with higher rates of pharyngocutaneous fistula.11 Prior radiotherapy or chemoradiotherapy impairs the microvascular circulation of the native tissues, which predisposes patients to adverse outcomes, including pharyngocutaneous fistulas, when compared with primary laryngectomy.11 In the context of salvage laryngectomy, Hasan et al10 found no significant difference in pharyngocutaneous fistula formation between chemoradiotherapy and radiotherapy alone.
In addition, patient factors, such as comorbidities, smoking status, and wound healing, are important to consider. Medical comorbidities, including anemia, diabetes, and vascular disease, can lead to impaired wound healing.11 In particular, cardiopathy, diabetes, and liver disease have been associated with fistula formation.9,11 Indicators, such as prealbumin, glucose, thyrotropin, and preoperative hemoglobin levels, are used to assess tissue viability but are unfortunately not reliable markers.6,11 Although smoking is often associated with vascular compromise and poor wound healing, Liang et al11 found no association between history of smoking and pharyngocutaneous fistula formation.
Complications of head and neck reconstruction are attributable to many factors. Devitalized native tissue present after tumor extirpation is important to note before flap inset because necrosis of this tissue can result in complications similar to those experienced because of partial or complete flap loss. Therefore, it would seem appropriate to assess not only the reconstructive tissue but also the ablative defect for healthy tissue. However, no published data have definitively attributed the true cause of complications, such as salivary leak after free flap reconstruction for a total laryngectomy defect, to be from the ablative site or the reconstructive tissue. Use of laser-assisted NIR angiography with intravenous ICG dye has focused on improving the clinical outcomes of reconstruction. This technology is widely used for breast reconstruction; thus, the equipment needed is already available at many academic medical centers. Beyond the imaging machine, the only additional cost is that of the ICG dye itself, which is approximately $50 per patient. An oft-overlooked opportunity for this technology is the assessment of ablative site tissue quality. This gap in knowledge represents possible new applications and research of ICG dye angiography.
Much of the literature regarding the use of ICG dye for native tissue involves skin-sparing mastectomy flaps for breast reconstruction. An early study12 subjectively analyzed mastectomy flaps in tissue expander–based breast reconstruction and found that there was a 95% correlation between intraoperative imaging and clinical course with 100% sensitivity and 91% specificity. This same group used laser-assisted ICG dye angiography and software algorithms to further quantitate (but only relative to other sites) the perfusion at the skin edges of mastectomy flaps. They found that the mean relative fluorescence of the necrosis and the adequate healing groups was 25.2% and 43.3%, respectively (95% CI, 13.3%-22.9%).13 These results further support this study and our conclusions regarding the lowest mean ICG dye percentage of mucosal angiography perfusion that can lead to wound breakdown and complications. In fact, most major hospitals have access to NIR and ICG dye technology because of its wide acceptance in breast reconstructive surgery.
There are some limitations to this study. The study does not account for other factors, such as microvascular free flap viability (if used in reconstruction), various reconstructive techniques, or surgical complications. Other limitations include the small sample size, few number of patients who developed fistulas, and absence of a validation group. One of the most important implications of the small sample size and few patients with fistulas is the wide CIs for all the observed effects. The wide CIs mean that the estimates we observed are imprecise and limit our ability to make robust conclusions. This small proof-of-principle study cannot be expected to provide thresholds and measures of discrimination that can be generalized with conﬁdence to a broader population. However, this study provides an estimate of the potential to use a commercially available intraoperative fluorescence imaging device to predict fistula formation based on the lowest percentage of mucosal perfusion, thereby introducing a desperately needed tool for standardization during ﬂuorescence perfusion angiography. Ultimately, determining thresholds for general use will require a larger and more general sample.
Pharyngocutaneous fistula formation after salvage laryngectomy is a serious, potentially life-threatening condition and can be attributable to several factors. The viability of native pharyngeal mucosa is an underrecognized factor in fistula formation, and laser-assisted ICG dye angiography is an easy-to-use, safe tool to assess mucosal perfusion intraoperatively. In addition, given its ubiquitous use in breast reconstructive surgery, the technology is widely available at most major medical centers. This quantitative assessment in real time can help guide reconstructive techniques to help avoid wound breakdown and fistula formation. Although this study suggests the value of this technique in assessing the viability of native pharyngeal mucosa, additional studies will need to be performed to validate our findings and further delineate specific predictive values.
Corresponding Author: Erin J. Partington, MD, Department of Otolaryngology, University of Alabama at Birmingham, 563 Boshell Bldg, 1808 Seventh Ave S, Birmingham, AL 35233 (firstname.lastname@example.org).
Accepted for Publication: February 4, 2017.
Published Online: May 18, 2017. doi:10.1001/jamaoto.2017.0187
Author Contributions: Drs Partington and Moore had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Partington, Moore, Warram, Carroll, Rosenthal, Greene.
Acquisition, analysis, or interpretation of data: Partington, Moore, Kahmke, Warram, Carroll, Greene.
Drafting of the manuscript: Partington, Moore, Warram, Rosenthal, Greene.
Critical revision of the manuscript for important intellectual content: Partington, Moore, Kahmke, Carroll, Rosenthal, Greene.
Statistical analysis: Partington, Moore.
Obtained funding: Rosenthal.
Administrative, technical, or material support: Kahmke, Rosenthal, Greene.
Study supervision: Warram, Carroll, Rosenthal, Greene.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Meeting Presentation: The research was presented at the American Head and Neck Society 9th International Conference on Head and Neck Cancer; July 16-20, 2016; Seattle, Washington.
Create a personal account or sign in to: