Effects of Neoadjuvant Chemotherapy and Radiotherapy on Flap Perfusion in a Novel Mouse Model Using Standard Clinical Assessment and Near-Infrared Fluorescence Angiography

1University of Alabama School of Medicine, Birmingham, AL, USA 2Department of Otolaryngology, University of Alabama at Birmingham, Birmingham, AL, USA 3Department of Psychiatry, University of Alabama at Birmingham, Birmingham, AL, USA 4Department of Surgery, University of Alabama at Birmingham, Birmingham, AL, USA 5Department of Radiology, University of Alabama at Birmingham, Birmingham, AL, USA 6Department of Oral and Maxillofacial Surgery, University of Alabama at Birmingham, Birmingham, AL, USA 7Department of Otolaryngology, Stanford University, Stanford, CA, USA #These authors contributed equally

standard-of-care relies on qualitative markers of poor perfusion to guide surgeons in the decision-making process.
This approach is subjective, and thus may be impacted by the surgeon's level of experience and familiarity with the patient population, among other factors. Recent advances in nearinfrared imaging technology have made it possible to assess perfusion in a quantitative manner. The LUNA fl uorescence angiography system employs an indocyanine green (ICG) dye to allow for intraoperative and post-operative visualization of perfusion [8]. Several groups have shown that perfusion assessments utilizing LUNA fl uorescence angiography are reliable and accurate [9,10].
Random-pattern fl aps with length-to-width ratios of 2:1 or less are reliably perfused throughout their length [11].
Flaps with ratios greater than 2:1 have been shown to exhibit ischemia in distal regions due to decreased proximity to the pedicled end [12,13]. Studies conducted in rats bearing overdimensioned 4:1 random-pattern fl aps demonstrated ischemia in distal halves of fl aps that, over time, led to fl ap necrosis in those regions [12,13]. In this study we create over-dimensioned 4:1 random-pattern local fl aps in a murine model, anticipating a loss of perfusion in the distal halves of fl aps. Our intention was to investigate the effects of neoadjuvant chemotherapy and radiotherapy on fl ap perfusion, which were evaluated by nearinfrared imaging technology and standard clinical assessment.

Animal model
Twelve female athymic nude mice were used in this study. Animals were caged in groups of 4 and fed a standard laboratory diet of food and water. Animal treatment and husbandry was in accordance with IACUC standards. Mice were divided into three treatment groups: control (no treatment, n=4), radiation group (36 Gy administered to dorsal skin, n=4), and chemotherapy group (2 mg/kg cisplatin, n=4). In the radiation group, mice received three 12 Gy treatments administered focally to dorsal skin fl aps over a 6-day period. Cisplatin was administered intraperitoneally (IP). Mice were allowed a 15 day recovery period during which they were monitored daily for possible systemic side effects. All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. All procedures performed in studies involving animals were in accordance with the ethical standards (approved IACUC protocols) of the institution or practice at which the studies were conducted.

Radiation therapy
An X-Ray irradiator (XRAD320, Precision X-Ray, Inc. N. Bradford, CT, USA) designed for use with cell monolayers and small animals provided radiation therapy in this study. The XRAD320 was calibrated to administer a tube voltage of 320 kV and a tube current of 12.5 mA to generate high-energy electrons at a mean dose rate of 1.169 Gy/min to a total dose of 36 Gy. The Isovolt Titan (General Electric, Fairfi eld, CT, USA) was utilized to monitor exposure time, x-ray production voltage, and current; the Unidos dosimeter (PTW, Freiburg, Germany) recorded total radiation exposure as well as dose rate. A 20 mm thick aluminum collimator was used to effectively eliminate damaging low-energy x-ray exposure. For the procedure, anesthesia was induced using inhaled isofl urane or 2.5 ug/kg ketamine/xylazine. Animals were placed on either side of a shielded, radiation-safe box in the lateral decubitus position ( Figure 1A). Target skin was exposed through a slit, allowing for focal delivery of radiation. The entire length of the fl ap was exposed to radiation treatment. Nose cones on either end ensured oxygen and isofl urane delivery for the duration of radiation exposure. A dosimeter was placed underneath the exposed skin to measure X-Ray dose. The dosimeter and radiation-safe box were secured to one another and to the base of the XRAD320 to insure homogeneity in radiation exposure among different treatment groups ( Figure 1B). A three-eighths inch lead sheet was placed atop the radiation-safe box to minimize radiation exposure to the body of the mice (Figure 1).

Physician assessment
On POD2, two blinded otolaryngology residents examined the fl aps under white light for qualitative signs of poor perfusion, i.e., skin color, turgor, capillary refi ll, hematoma,

Statistical evaluation
Two-way ANOVA tests were performed to determine signifi cant differences in perfusion across treatment groups, as well as between proximal and distal halves of elevated fl aps. Unpaired, two-sample t-tests were used to compare two specifi c fl ap halves across time or treatment group. Signifi cance is defi ned as P < 0.05.

Flap survival
Mean fl ap perfusion is expressed as a percentage of healthy skin perfusion to account for anatomical differences in mice. As shown in Figure 1

Post-operative day 1 fl uorescence assessment
In the proximal halves of fl aps, perfusion was comparable to that of healthy skin ( Figure 3A perfusion was similar in control and radiation groups (P = 0.5087) but decreased in chemotherapy groups (P = 0.0037). As expected, distal halves of fl aps in each treatment group were found to be more poorly perfused than their proximal half counterparts (P < 0.05). Overall, fl aps on POD4 were more poorly perfused than on POD1. Representative fl uorescence images are also shown from POD4 of the negative control ( Figure 4B), radiation therapy ( Figure 4C), and chemotherapy group ( Figure 4D).

Physician assessment
Clinical assessment performed on POD2 revealed the radiation group to have the healthiest-rated fl aps when compared to control and chemotherapy groups ( Figure 5). Physicians identifi ed the chemotherapy group fl aps as least healthy, both proximally and distally. Distal halves of radiation fl aps received a mean score of 1.375, being rated much healthier than control group fl aps.

Discussion
The tolerance of tissues to radiation and chemotherapy in the perioperative period is not well known. We demonstrate here that the impact of chemotherapy on tissue perfusion after fl ap elevation is signifi cantly greater in the peri-treatment period. During the study, the local skin fl aps exhibited a perfusion gradient with regions proximal to the pedicle having greater perfusion than more distal regions, which was shown to be adversely affected by prior radiotherapy or chemotherapy. This relationship was tested using a preclinical local fl ap murine model. These perfusion models are useful in eliminating complicating or confounding factors, such as variability in fl ap geometry and co-morbid disease, which are inherent in clinical cases. In addition, this study design permitted standardized treatment doses and recovery times following neoadjuvant therapy, which often vary widely in clinical studies [14].
Proximal halves of control group fl aps-the 2cm x 1cm segments nearest the pedicle-remained adequately perfused as determined by LUNA fl uorescence angiography and clinical assessment throughout the experiment, consistent with other groups who employed this model. No statistical difference was found among the treatment groups at POD1, however proximal halves at POD4 of the chemotherapy group fl aps were signifi cantly less perfused than control group fl aps. Overall, the distal (2cm x 1cm) segments of fl aps in this study had poor perfusion. This well-known clinical limitation of randompattern fl aps is why surgeons often select locoregional axialpattern and perforator fl aps, or microvascular free tissue transfers for compromised wound beds [11]. At POD1, a clear perfusion gradient could be visualized in all treatment groups, though the distal-halves of fl aps in the chemotherapy group fared the worst. Clinical assessment performed the following day (POD2) and LUNA imaging performed on POD4 reaffi rmed these fi ndings, indicating a clear link between neoadjuvant chemotherapy treatments and reduced distal half fl ap perfusion leading to necrosis.
Chemotherapy hampers the infl ammatory response, reducing oxygen delivery to the wound and increasing the likelihood of infection [15]. Lawrence et al., found that these delays in wound healing were especially pronounced in cases where chemotherapeutic agents were administered pre-operatively or within 3 weeks post-operatively [7]. Two studies of head and neck cancer patients have established chemotherapy to be among the factors that lead to fl ap failure [16,17]. Despite this, neoadjuvant chemotherapy remains widely used, albeit in head and neck it is used primarily in advanced stage or surgically unresectable tumors.
Neoadjuvant radiotherapy functions to reduce volume and limit spread of locally advanced head and neck cancers to amend the invasive tumor for surgery [18]. It is unclear, however, whether neoadjuvant radiotherapy may inadvertently complicate microsurgical reconstruction in the process. Animal studies suggest it is detrimental to fl ap survival [19][20][21], whereas clinical studies found it to have no appreciable effect [22][23][24]. In our study, irradiated random-pattern fl aps performed similarly to control group fl aps in the proximal half and outperformed control group fl aps in the distal half. We fi nd the superior performance of radiation group fl aps in the distal half to be unremarkable, attributing this fi nding to human error or limited sample size. The statistical signifi cance of the data may have been limited by the small sample size, suggesting that an increased "n" may allow for more accurate fi ndings. A 36-Gy dose of radiation was selected to produce moderate skin damage with low levels of skin contraction, based on a model developed by Rifkin et al. [25]. Higher doses were found to cause increased destruction of epithelial basal cells, leading to desquamation and further reducing viability. It is envisioned that administering greater doses of focal skin irradiation may lead to poorer fl ap performance.
Numerous confl icting studies can be found in the literature regarding the effects of neoadjuvant treatment on fl ap survival. These fi ndings may be due to inherent differences in tumor behavior and patient co-morbidities, as well as fl ap classifi cation by location (local, regional, free), geometric design (advancement, rotation, island), and vasculature  (random, axial, perforator, myocutaneous, fasciocutaneous). Stricter inclusion criteria in clinical trials would reduce participant heterogeneity and potentially streamline results. Even so, it is diffi cult to delineate cause and effect in prospective or retrospective studies due to lack of true controls. In this experimental study of mice, we present evidence that 2-mg/kg Cisplatin preoperatively reduces fl ap perfusion, and that preoperative 36-Gy radiation to the skin does not affect fl ap perfusion.
The current study employs the novel use of an overdimensioned random-pattern skin fl ap in a murine model to study effects of neoadjuvant therapies on fl ap perfusion. The study utilized two independent metrics, LUNA fl uorescence angiography and clinical judgment. Results showed that fl aps receiving neoadjuvant chemotherapy fared worse than fl aps in the control group. Neoadjuvant radiation had no detrimental effects on fl ap perfusion. This study can assist clinicians in making informed decisions pertaining to the use of neoadjuvant therapies, obviating the risk of vascular compromise that may accompany such treatments.