INTRODUCTION
Nasopharyngeal carcinoma (NPC) is one of the most common malignancies in the head and neck area, especially in East and Southeast Asia [
1,
2]. Radiation therapy (RT) either with or without chemotherapy is the mainstay treatment for NPC, and the 5-year overall survival (OS) rates vary from 63.0% to 87.4% due to advancements in RT technique and the application of adjuvant chemotherapy in locoregionally advanced NPC patients [
3-
6]. Although RT has a high response rate and shows favorable survival, it can induce several severe complications, such as postradiation nasopharyngeal necrosis (PRNN) [
7-
9]. Although it is not a common complication, with an incidence rate of 0.8%–1.1% [
8,
10], PRNN can affect the nasopharyngeal mucosa, parapharyngeal tissues, or skull base, thereby inducing severe headache, foul odor, cranial nerve palsy (CNP), and even massive bleeding due to internal carotid artery (ICA) rupture, which can seriously impact patients’ quality of life and survival [
8,
9,
11-
13].
Therefore, timely diagnosis and proper management of PRNN are essential, and endoscopic debridement has shown favorable outcomes with low morbidity [
7,
8]. A vascularized flap is needed to cover the exposed ICA or dura to reduce the risk of catastrophic events such as carotid blow-out or meningitis. Alongside our previous study [
12], several recent studies have reported promising outcomes regarding the use of a nasoseptal flap (NSF) to resurface the defect site, showing improved functional outcomes and better OS [
10,
14-
16]. Although promising results have been obtained using NSFs, there are still cases of failure and incomplete resolution of PRNN. Nonetheless, little has been explained regarding the mechanism of this complication and there is a lack of research on how to treat such intractable cases after failed NSF reconstruction. Thus, focusing on the mucosal resurfacing status of nasopharyngeal wounds, our study aimed to analyze the clinical outcomes of flap reconstruction after endoscopic debridement based on the mucosal resurfacing status and to investigate the effectiveness of salvage operations.
MATERIALS AND METHODS
Patients
We retrospectively reviewed the medical records of NPC patients who were surgically treated for PRNN at our institution between April 2013 and March 2021. We selected patients for surgical treatment based on radiation history for NPC, clinical features (including headache, foul odor, and CNP), endoscopic findings, and radiologic studies (such as computed tomography [CT] and magnetic resonance imaging [MRI]). A total of 31 patients underwent endoscopic debridement for PRNN. Among them, four patients with pathologically confirmed recurrent tumor were excluded. Finally, 27 patients were analyzed in the present study. This study was approved by the Institutional Review Board of Samsung Medical Center (No. 2022-04-039), and the need for informed consent was waived.
According to the method of initial surgery, patients were divided into two groups, debridement with NSF (NSF group, n=21) and debridement only (no NSF group, n=6). The patients with continuous or relapsed symptoms along with persistently necrotic surgical bed were considered candidates for salvage surgery. ICA involvement, defined as necrotic tissue surrounding the ICA, no healthy tissues above the necrotic tissue over the ICA, and evidence of a narrowing of the ICA lumen in comparison to the contralateral ICA, even after surgery for radionecrosis was also considered for salvage surgery.
Ultimately, salvage surgery was performed in 10 patients with reconstruction using an NSF (n=5), anterolateral thigh free flap (ALTFF, n=4), and inferior turbinate (IT) mucosal free graft followed by middle turbinate rotation flap (MTF, n=1). All patients were regularly followed up at 3- to 6-month intervals until at least 5 years after the initial surgery. Nasopharyngeal endoscopic evaluation, contrast-enhanced CT scans, and MRI were performed as follow-up evaluations.
Surgical techniques
Initial surgery
If the ICA was surrounded with necrotic tissue, we performed the balloon occlusion test (BOT) for potential risk of ICA injury and sacrifice. All patients underwent endoscopic debridement with image-guided surgery. After performing frozen biopsy to exclude recurrent tumor, wide debridement was performed using curettes, a microdebrider or Coblator (Coblator II; Smith & Nephew, Memphis, TN, USA). After debriding the devitalized soft tissue, an endoscopic drill was used to remove the necrotic portion of the skull base bone. Careful attention was paid when debriding on the posterolateral side to avoid carotid artery rupture using Doppler ultrasound. If the carotid artery was surrounded by necrotic tissue, the necrotic tissue was intentionally left in place. After debridement, we performed massive povidone-iodine irrigation.
In the NSF group, we covered the defect with NSF based on the septal branch of sphenopalatine artery. NSF was usually harvested from the contralateral side of the main lesion to avoid potential pedicle injury during debridement around the eustachian tube. The NSF was elevated from the mucocutaneous junction (caudal septal margin), including the nasal floor for a wider width. Gelfoam (Pfizer Inc., New York, NY, USA) pledgets and Merocel (Medtronic Xomed, Jacksonville, FL, USA) nasal packing were used to bolster the area and deter flap detachment. The nasopharynx was packed for 7 to 10 days. Following the removal of the packing, the patients started irrigating the lesion with normal saline solution.
Supplementary Video 1 is an example of intraoperative video clip of above endoscopic debridement with NSF reconstruction procedure.
Salvage surgery
When performing salvage surgery, we collected necrotic tissue and remaining deep tissue for frozen biopsy to exclude tumor recurrence. The following debridement procedure was similar to that of the initial surgery. For reconstruction using ALTFF, we used two approaches, the maxillary swing approach for three patients and the transcervical approach for one patient. Regarding the maxillary swing approach, a Weber–Ferguson–Longmire facial incision was performed to allow a view for further wide debridement and adequate space for transferring the ALTFF. Proper ALTFF donor size and shape were based on defect size and the need for adequate coverage of the ICA. Then, a linear skin incision was created from the anterior superior iliac spine to the lateral border of the patella in the non-dominant leg of the patient. To identify the perforator of the descending branch of the lateral circumflex femoral artery, which is a feeding artery to the ALTFF, we performed dissection through the level of the rectus femoris and vastus lateralis. After identifying the muscular perforators entering the vastus lateralis, the designed fasciomuscular flap was harvested with the perforators. The flap was then transferred to the nasopharyngeal area and sutured to the nasopharyngeal wall with Vicryl suture. This procedure was conducted using the transcervical approach for one patient, where a transverse skin incision was performed on the neck level II area along the skin crease. Arterial anastomoses were performed to the facial artery for three and transverse cervical artery for one, and venous anastomoses to the facial vein for two and external jugular vein and transverse cervical vein each in one patient. Two patients required tracheostomy, while nasopharyngeal airway was inserted in addition to nasal packing and endotracheal tube was maintained till extubation on one day postoperation each in one patient. All procedures described above were performed by ENT surgeons. For one patient who received MTF for a second salvage operation, a posterior lateral nasal artery pedicled posterior-based MTF was designed. An incision was made anterior to the middle turbinate, afterwards mucoperiosteum was elevated from the front aspect, and the middle turbinate bone was removed. It is critical to avoid damaging the vascular pedicle as it enters at its lateral attachment. After the flap was harvested, it was gently rotated to the nasopharynx and bolstered for 7–10 days using Gelfoam and Merocel.
Outcome measurements
Pre- and postoperative symptoms of headache and foul odor, CNP, postoperative mucosal status, and salvage operation rate were evaluated to compare outcomes according to initial surgical method. Postoperative mucosal status was classified as complete healing, partial healing, or persistent necrosis (
Fig. 1).
Within the NSF group, patients were subcategorized into the NSF viable subgroup (NSF-V) and the NSF necrotic or uncovered subgroup (NSF-U) according to flap viability after initial surgery for risk factor analysis of flap failure. The risk of flap failure was analyzed according to age, diabetes mellitus (DM), underlying comorbidity, necrosis stage, T stage, ICA involvement, and RT frequency and dose. Based on the clinical features of patient’ symptom, endoscopic examination and radiological findings, necrosis stage was classified as early, middle, or late, as described in previous studies (
Fig. 2) [
8,
12]. Symptoms of headache and foul odor, CNP, final mucosal status, and ICA rupture, along with mortality rates were evaluated for final outcome measurement.
Statistical analysis
Statistical analysis was performed with IBM SPSS 27.0 (IBM Corp., Armonk, NY, USA); P-values for each variable were calculated using the Mann-Whitney U-test. P-values less than 0.05 were considered to represent statistically significant comparisons.
DISCUSSION
Even though PRNN severely impacts patients’ quality of life and survival, no standard treatment has been established. Several therapeutic modalities have been introduced, including conservative management using antibiotics and frequent dressing, hyperbaric oxygen therapy, and endoscopic debridement with or without vascularized flap reconstruction. Among these methods, endoscopic debridement is widely used to remove necrotic tissue and replenish fresh tissue, helping to control infection and improve symptoms in the skull base area [
7,
12]. However, the recovery period is quite long, and necrosis easily relapses; therefore, surgical debridement might need to be repeated [
8,
11,
17]. The exact pathophysiology of PRNN is unclear, but hypoxia, hypovascularity, and hypocellularity from RT are thought to be the key factors in its development and resulting inflammation, sequestration, and erosion of the underlying cortex of the nasopharynx [
18,
19]. Debridement without flap reconstruction seldom achieves complete healing, and failing to resurface defects of the skull base, dura, or ICA can result in life-threatening complications such as meningitis, ICA blowout, and death [
8,
10,
20]. In fact, ICA involvement has been found to be an independent prognostic factor, and it can increase the risk of death from 41.8%–42.9% to 69.2%–72.7% [
7,
8,
10,
11].
Therefore, resurfacing the surgical bed with a vascularized flap and protecting the exposed skull base and ICA are regarded as the most important surgical goals in PRNN treatment. The NSF, pedicled by the posterior septal artery, a branch of the sphenopalatine artery, has been regarded as an effective and reliable reconstruction method for PRNN considering the advancements in image-guided endoscopic skull base surgery [
12,
15,
21]. There have been some challenges in using free flap reconstruction to cover extensive defect sites, after either nasopharyngectomy for recurrent NPC [
22,
23] or endoscopic debridement for osteoradionecrosis in the head and neck area [
24-
28]. Although free flap reconstruction is invasive and has some limitations due to the difficult technique, long operation time, and wound problems, it can cover more extensive defects, unlike the NSF, and plays a role in salvage surgery. With its ample size and high vascularity, the ALTFF is a good option for extensive PRNN, especially for cases involving the ICA, which is a known prognostic factor for symptom aggravation, increased disease extent even after surgery, and death [
21,
25,
26]. A recent study by Zou et al. [
15] analyzed 72 PRNN patients treated with NSFs. It showed successful outcomes of defect area re-epithelialization (70.8%) and symptom improvement without any surgery-related complications or death, and the 2-year OS rate was 77.9%. They also demonstrated that the viability of NSF reconstruction was a protective factor for re-epithelialization, which could serve as a barrier to protect the ICA.
To our knowledge, outcomes have not been compared according to flap and mucosal status. There is also a lack of research on salvage surgery using a vascularized flap, especially for free flap reconstruction, after surgical failure in PRNN patients. As such, we not only analyzed data corresponding to initial curative intent endoscopic debridement with or without NSF, but also included the clinical courses of salvage operations using vascularized flap reconstruction and compared the clinical outcomes according to final mucosal resurfacing status. Moreover, based on the flap viability of the initial NSF group, we attempted to determine the risk factors for NSF failure in PRNN patients. Our data showed that resurfacing the defect area with an NSF in initial surgery showed better outcomes than debridement only. Most patients in the NSF group achieved healthy mucosal status, and only four patients underwent salvage surgery even in the setting of more ICA involvement, while all six patients in the no-NSF group had to undergo salvage surgery due to recurrent necrosis with recurrent symptoms. Among 10 salvage operations, free flaps were also performed in four patients with ALTFFs (cases 4, 5, 21, 23). Interestingly, although not all of the 10 patients maintained completely healed mucosa after salvage surgery, they also showed good clinical outcomes, with most patients reporting symptom relief. Finally, 21 patients (77.8%) had completely healed mucosa on final presentation, whereas six patients (22.2%) did not. No patients suffered from foul odor, and headaches improved in all, but two patients compared to their initial condition. Among 13 patients whose ICA was involved, two ruptures (case 21: 1-month postoperation, case 24: intraoperative) were reported and rescued by coil embolization. Only three patients died (cases 5, 21, 22), and the survival rate was 88.9%. Our results suggest that endoscopic debridement and mucosal resurfacing with NSF reconstruction are essential for the surgical management of PRNN. Even if the initial surgery is ineffective and the NSF does not cover the whole defect area or becomes necrotized, additional reconstructive resurfacing surgery using another vascularized flap is necessary for better clinical outcomes.
Yang et al. [
10] recently reported the clinical outcomes of NPC with PRNN. They demonstrated that osteoradionecrosis, re-irradiation, and ICA involvement affected survival, although only re-irradiation and ICA involvement were independent prognostic factors, with hazard ratios of 1.75 and 1.80, respectively. Additionally, NSF reconstruction was associated with better OS than conservative management. Among 44 patients with NSF in their study, eight flaps (18.2%) did not show favorable re-epithelialization, although an explanation and analysis of risk factors for NSF failure were missing. In our study, when categorizing the NSF group according to flap viability, we found that age, necrosis stage, advanced T stage, ICA involvement, number and dose of RT, and the propensity toward DM and underlying comorbidities tended to be higher in the NSF-U subgroup. Including the variables (e.g., necrosis stage, ICA involvement, and reirradiation) that are already known to be related to the survival of PRNN patients, the factors that we investigated in this study might also play a role in NSF failure due to poor vascularization and perfusion of the underlying nasopharyngeal surface, hindering its healing and mucosalization. Further studies should be conducted with larger cohorts and a sufficient follow-up period, and the pathophysiology should be elucidated to validate these trends.
There were four patients with lower CNP on the initial preoperative examination. Among them, two (cases 23 and 24) experienced partial CNP improvement following salvage surgery, allowing the gastrostomy tube to be removed. During the follow-up period, there were six cases of newly developed CNP, four of which (cases 4, 11, 17, 18; 19.0%) emerged even after final successful flap surgery. One (case 17) of those four cases developed due to a recurrent tumor and another (case 4) after initial surgery without NSF. The other two (cases 11 and 18; 9.5%) in the complete healing group developed new CNP as a consequence of PRNN. In contrast, two patients (cases 1 and 27; 33.3%) with partial or total necrosis newly developed CNP. Therefore, even if the vascularized flap might not fully prevent the development of lower CNP, at least it seems to lessen the likelihood. Nonetheless, we should be cautious that even after successful surgery, necrosis may spread beneath the flap, and CNP may develop and get worse over time. Additionally, although statistically insignificant, a larger proportion of ICA involvement was observed in patients with NSF failure than in the success group (85.7% vs. 42.3%), which implies that if necrotic tissue around the ICA is left in place in order to avoid ICA rupture, there may be a greater chance of flap failure. However, there was a representative example (case 8) of completely healed mucosa and a viable flap with persisting necrotic tissue beneath the mucosal layer on follow-up imaging, but without any symptoms, CNP, or ICA rupture (
Fig. 6). Nonetheless, the necrotic bed poses a high risk for flap failure; therefore, it is a dilemma whether to entirely remove the necrotic tissue around the ICA or not. In such instances, total removal of the necrotic tissue following prophylactic ICA embolization or superficial temporal artery (STA) to middle cerebral artery (MCA) bypass surgery may be an excellent option, as described in a recent publication by Cho et al. [
16]. In their study, they presented a strategy for managing ICA involvement in which coil embolization was conducted if BOT followed by brain perfusion single-photon emission CT revealed no hypoperfusion, while STA-to-MCA bypass surgery was performed if hypoperfusion was seen. This approach allowed extensive sequestrectomy and successful outcomes, providing valuable insights into future directions of ICA handling in PRNN. Nevertheless, more research on this subject is required.
Despite promising advancements from our previous study and some meaningful results herein, the present study has the limitations of a retrospective analysis, including patient selection bias for each surgical modality since we only performed endoscopic debridement without a vascularized flap as the initial operation and free flap for salvage operation in earlier cases, which changed to the current practice pattern of employing the NSF as our experience and evidence for the utility of the NSF gradually accumulated. Other limitations include the small number of patients who were treated at a single institution and operated on by a single surgeon, and a short follow-up period. Hence, further studies with a longer follow-up period and a larger cohort are required to validate our results.
In conclusion, if mucosal resurfacing was initially not done or improper, the prognosis was not good. Reconstruction using an NSF after endoscopic debridement is an effective and reliable modality in the initial surgical management of PRNN. Even if the initial NSF fails, subsequent resurfacing salvage surgery with a vascularized flap, including free flap reconstruction, should be used to improve clinical outcomes.