Nasopharyngectomy for Recurrent Nasopharyngeal Cancer: A 2- to 17-Year Follow-up

Objective  To review the 2- to 17-year outcome of nasopharyngectomy following local recurrence of nasopharyngeal carcinoma.

Design  Retrospective review.

Setting  University medical center.

Patients  Thirty-seven patients with biopsy-proven recurrent nasopharyngeal cancer followed up for a minimum of 2 years after transpalatal, transmaxillary, and/or transcervical resection with and without neck dissection.

Outcome  Clinical examination, magnetic resonance imaging, chest x-ray examination, and liver function tests to determine re-recurrence; unlimited follow-up.

Results  With a mean follow-up of 5.4 years, the crude, 5-year, overall, free-of-disease survival rate was 52%, local control at 5 years was 67%, and the 5-year actuarial survival rate was 60%. Survival by recurrent T stage (rT) was as follows: rT1, 73%; rT2, 40%; rT3, 14%; and rT4, 0%. Complications occurred in 54% and included 1 death from carotid artery injury and 1 patient with permanent pharyngeal plexus paralysis with resultant dysphagia. The remaining patients had transitory complications that spontaneously resolved, required further surgery (closure of palate fistula, debridement, and reapplication of skin graft), or required further medical therapy.

Conclusions  The results of this study are better than most published reports of additional irradiation for rT1 and rT2 lesions. More recent radiation studies that use radiosurgery or implants suggest promising early results. A randomized prospective study comparing surgery with additional irradiation for recurrent disease at the primary site is warranted.

PATIENTS WITH recurrent nasopharyngeal carcinomas continue to challenge physicians' creativity and skill. Although there are no great options for treating metastatic disease, neck dissection is an excellent choice for treating regional recurrence after radiation therapy (XRT) when the primary site is controlled. Recurrence at the primary site has usually been treated with a second course of XRT. Until recently, surgery was considered impossible because of the inaccessible location of the nasopharynx. Lee et al¹ reported that long-term survival after a second course of XRT was less than 30% and was associated with a 26% incidence of complications, including 2% therapy-related death. Such results prompted us to explore the feasibility of surgical resection. This report documents our long-term survival results.

From May 4, 1984, to March 31, 1999, one of the authors (W.E.F.) performed nasopharyngectomy for recurrent nasopharyngeal carcinoma on 44 patients at the Stanford University Medical Center. Initially, the first 9 patients had computed tomographic and magnetic resonance imaging (MRI) scans to localize tumor; subsequently, MRI scans with gadolinium and fat suppression were used to stage disease in addition to the clinical examination, chest x-ray examination, and liver function tests. A retrospective medical record review was performed by 2 surgeons (M.S.M. and E.C.C.) unassociated with the surgery (to eliminate any potential bias), paying particular attention to all complications associated with (but not necessarily due to) the surgical procedure. Follow-up was obtained from patient medical records either at Stanford or from referring physicians.

The surgical approach was variable, depending on the tumor location, but included an intraoral, transpalatal, and sublabial transmaxillary approach; isolation of the internal carotid artery and cranial nerves IX, X, and XII to the skull base; and modified (sparing cranial nerve XI) or radical neck dissection as the clinical condition dictated. No transfacial or lateral infratemporal approaches were used. Since this tumor mimics juvenile angiofibroma in its location, the surgical approaches are similar with the exception of the need for isolating the internal carotid artery and performing a neck dissection. One-centimeter margins were obtained around the tumor and then frozen section margins were obtained from the patient and submitted for analysis; if the margins were positive, additional sections were taken, continuing, if possible, until a negative margin was obtained.

Nasopharyngeal specimens are removed en bloc, although the specimen will occasionally fragment (<10%) when the primary site is extensively ulcerated. After resecting the main specimen, the entire anterior wall of the sphenoid sinus wall is removed, with the sphenoid sinus contents, when necessary. The remaining sphenoid bone is drilled with a diamond burr, with care taken to prevent injury to the internal carotid artery and the optic nerves. Next, the clivus and anterior arch of the atlas and axis are drilled with a cutting burr followed by the diamond drill for 3 to 5 mm until healthy bleeding is encountered and all mucosal, ligamentous, and muscular filamentous attachments to the anterior bone are removed.

The inferior turbinate(s) is helpful to use as a mucosal graft to resurface the nasopharynx and facilitate cleaning of the resultant cavity. If the surface area is too great, then a 0.018-in split-thickness skin graft is applied to the bare bone and secured with antibiotic-impregnated packing, which is held in place for 5 to 7 days. Unilateral or bilateral tympanostomy tubes are placed when the eustachian tube(s) is removed.

After removal of the packing, the nasopharynx is irrigated with warm tap water with added salt and bicarbonate of soda as the patient desires. Initially, patients are encouraged to irrigate 2 to 4 times per day via the nose and oral cavity or, as necessary, to decrease crust formation. Some patients find it helpful to use nasal mucosal emollient (Ponaris; Jamol Laboratories Inc, Emerson, NJ), topically applied to the nose and nasopharynx via nose drops, between irrigations. Others find mineral oil, petroleum jelly, or antibiotic ointment useful. Crusting of the nose and nasopharynx is removed 1 to 2 times per week for 6 to 12 weeks or until the wound is healed; this activity eliminates odoriferous breath.

A temporary soft palate obturator is constructed by a skilled prosthodontist for those patients who require soft palatectomy, and a permanent prosthesis is usually made 4 to 6 months postoperatively.

No patients were excluded from the study except 5 of the 44 patients who had minor salivary gland carcinomas. Four of the 5 patients are free of disease, whereas the fifth patient is alive free of local disease but with pulmonary metastases. Two patients who had tumor extension into the spinal canal were also excluded because they were treated for palliation only because of pain, nasal obstruction, and bleeding; they lived 12 and 80 months, respectively, after resection, with only the latter patient receiving meaningful palliation. These 7 patients were excluded from the final analysis, and the remaining 37 patients had typical World Health Organization class I or IIa or IIb nasopharyngeal carcinoma.

The group consisted of 24 men and 13 women with an average age at the time of operation of 49 years (range, 28-72 years). Disease was restaged based on the preoperative evaluation using the AJCC Cancer Staging Manual (revised 1998). Twenty-two patients were staged as having recurrent T-stage (rT) 1 disease; 5 patients, rT2; 7 patients, rT3; and 3 patients, rT4.

In all patients, primary treatment with XRT had failed to control the neoplasm. The average dose delivered to the nasopharynx was 6800 rad (68 Gy) (range, 6000-7500 rad [60-75 Gy]). The mean number of fractions was 38 (range, 33-60), and the mean length of treatment was 55 days (range, 30-79 days). The mean time to recurrence after irradiation was 28.6 months (range, 2-108 months). In 6 patients, a second course of external beam XRT failed before surgical treatment; 5 (83%) are free of disease more than 5 years after surgery.

Perioperative data are as follows.

Surgical complications occurred in 20 patients (54%), 2 (5%) of which were considered major. Serious complications included 1 death from an intraoperative carotid artery injury and a pharyngeal plexus paralysis, resulting in permanent dysphagia in a second patient.

Transitory complications occurred in 18 patients, including 4 instances of nerve paresis (1 with Horner syndrome, 1 with marginal mandibular branch of the facial nerve, and 2 with 9th, 10th, and 12th nerve paresis with resultant temporary dysphagia). Osteoradionecrosis and skull base osteomyelitis developed in 6 patients (5 of whom failed 2 courses of XRT), followed by complete recovery after treatment with prolonged intravenous antibiotics, debridement, and reapplication of the split-thickness skin graft. Four patients had exposed bone postoperatively, which was treated with local wound care; a single patient had a complete loss of the skin graft that closed secondarily. Five patients developed palatal fistulas, of which 2 required surgical repair, 2 closed spontaneously, and 1 required a palatal obturator. Baroreceptor dysfunction occurred in 1 patient who had bilateral internal carotid artery isolation that required 2½ months of medical therapy (clonidine) to control his hypertension and tachycardia. Thyroid storm developed in 1 patient intraoperatively, with complete recovery. One patient who had severe depression preoperatively remained severely depressed postoperatively, despite intensive treatment. One patient developed pneumonia, requiring readmission and antibiotics, and 1 patient developed nasopharyngeal stenosis. Three patients had more than 1 complication.

Eight patients who had positive surgical margins received postoperative XRT, 3 with intracavitary cesium implants, 3 with stereotactic radiosurgery, 1 with external beam irradiation (photons), and 1 with heavy particle (proton) irradiation. Only 2 of those patients (1 undergoing stereotactic radiosurgery and 1 undergoing proton irradiation) survived 5 years or longer free of disease. The 1 patient who received a second course of external beam XRT died from airway obstruction secondary to supraglottic edema but was free of disease at autopsy. The remainder recurred locally and these patients eventually died from disease.

Patients were followed up for a minimum of 2 to 17 years (mean, 5.4 years; median, 5.2 years) or until death. The survival following nasopharyngectomy is given in Table 1 and actuarial survival in Figure 1. Thirty-three patients have 5 or more years of follow-up; 52% have no evidence of disease. The pattern of recurrence following nasopharyngectomy is given in Table 2, with a local control rate of 67% at 5 or more years. Patient survival free of disease and local control rates, respectively, are as follows: rT1, 73% and 77%; rT2, 40% and 40%; rT3, 14% and 57%; and rT4, 0% and 0%. The apparent disparity in those patients with rT3 tumors (survival vs local control) is attributable to death caused by distant metastases. In those patients who developed postoperative recurrences, the mean time for diagnosis of recurrent disease was 8 months (median, 6 months; range, 1-31 months) and to death was 19 months (median, 17 months; range, 3-38 months).

Sooy² was one of the earliest Americans to recommend electrodesiccation of recurrent nasopharyngeal tumor followed by additional XRT. Fisch³ was the earliest to recommend surgical resection for either recurrent or "persistent" disease. In 1988, Fee et al⁴ and Tu et al⁵ reported their experience with surgical resection on a small series of patients (n = 9) using a transpalatal approach; in both studies, 44% of patients were alive free of disease. Table 3 gives the results of other surgical studies for recurrent disease. Wei et al⁶ described the maxillary swing approach to the nasopharynx in 1991, with a reported 40% actuarial tumor control at 3½ years in 18 patients. By 2000,⁷ they reported a 43% actuarial tumor control at 2½ years in 45 patients who had recurrent nasopharyngeal cancer. With a mean follow-up of 4.5 years, Morton et al⁸ reported a 71% crude survival rate in 7 patients, 6 of whom had rT1 disease, who underwent a transpalatal, transmandibular, or transcervical approach. Hsu et al⁹ described 24 patients treated by a transpalatal, transmaxillary, or transmandibular approach, with 46% surviving free of disease 8 to 50 months after surgery (median, 18 months). King et al¹⁰ reported on 31 patients, 29 of whom had rT1 or rT2 disease. The overall survival rate was 47%, with 43% having tumor control and a 42% actuarial 5-year survival. Finally, To et al¹¹ described 15 patients with a facial degloving approach assisted with a stereotactic navigation guidance system; no follow-up data were provided.

All of these surgical articles (including this one) demonstrate that recurrent nasopharyngeal carcinoma can be resected through a variety and combination of approaches with acceptable outcome and morbidity statistics. Unlike the study by King et al,¹⁰ our study showed little benefit with the addition of postoperative supplemental XRT. With the exception of the transpalatal and the facial degloving approach, all of the other surgical approaches produced significant trismus and/or an undesirable facial scar. The wise surgeon will tailor the operation to the areas involved by tumor and not try to make the tumor resection fit a standard operative approach. For an individual patient, a combination of approaches may be necessary, similar to patients who undergo resection for angiofibroma.

Resurfacing exposed bone with a skin or mucosal graft is a quick and simple solution to a difficult problem. The take rates of these grafts vary from 0% to 100% and average about 70%. For those patients who have had 2 courses of XRT, use of skin or mucosal grafts is not adequate, as demonstrated by 5 of the 6 patients who developed a low-grade osteomyelitis of the skull base. Using flaps such as a radial forearm free flap or temporalis fascia flap will be problematic in anchoring the flap in position, because there is no soft tissue remaining on the lateral or posterior walls of the nasopharynx. The best method of reconstruction of this defect has yet to be determined.

The survival results for re-irradiation are given in Table 4. Wang¹² reported a 5-year actuarial survival rate of 38% for 32 patients with rT1 and rT2 disease and 15% for 6 patients with rT3 and rT4 disease. Pryzant et al¹³ demonstrated a 5-year actuarial local control rate of 35% and an overall survival rate of 21%; those with rT1 or rT2 tumors had a survival rate of 32% vs 9% for those with rT3 or rT4 tumors. This study also showed a 5-year local control rate of 67% in 9 patients who underwent external beam irradiation (2000-3000 rad [20-30 Gy]), followed by 4000 to 5000 rad (40-50 Gy) by intracavitary cesium. The study by Hwang et al¹⁴ of 74 patients showed a 5- and 10-year actuarial survival rate of 37% and 18%, respectively, suggesting that long-term follow-up is needed in this disease.

The report by Lee et al¹ of 706 patients treated for recurrences represents the world's largest series to date and is precisely detailed. They readily admit that the data are old, because they represent patients treated from 1976 to 1985, but much worthwhile information can be obtained. Lee and coauthors reported survival rates of 27% for patients with rT1 disease, 17% for patients with rT2 disease, and 4% for patients with rT3 or rT4 disease. The report by Teo et al¹⁵ of a later period demonstrated a 5-year overall survival rate of 7.6% and a 5-year local control rate of 15.2% with significant complications. Teo et al¹⁵ reported survival rates of 26% for patients with rT1 disease, 8% for patients with rT2 disease, and 3% for patients with rT3 disease. They compared their statistics of additional irradiation of 103 patients with 20 patients (70% of whom received postoperative irradiation) who underwent nasopharyngectomy; those undergoing nasopharyngectomy had a statistically significant better survival than those who received high-dose additional irradiation for rT1 and rT2 lesions; it is noteworthy that the survival curves get very close at 4½ years; therefore, this difference may be only an early advantage. Nonetheless, they concluded that for the "suitably chosen patient, nasopharyngectomy is preferred to re-irradiation mainly with external beam to high dose."

Chua et al¹⁶ analyzed 212 patients treated between 1984 and 1995. The 5-year overall survival rates were 57% for patients with rT1 and rT2 disease, 42% for patients with rT3 disease, and 17% for patients with rT4 disease. Unfortunately, this study combined patients with persistent and recurrent disease even though it is well known that patients with persistent disease do better than those with locally recurrent disease, as documented by a study from the same institution.¹⁷ Kwong et al¹⁷ demonstrated a 5-year overall survival rate of 53.6% and a local control rate of 63% in patients undergoing gold-grain implantation for rT1 lesions. Although this result is not as satisfactory as the 73% reported in this article, it is definitely an improvement from prior XRT articles. Chua et al¹⁸ described 10 patients with locally persistent or recurrent nasopharyngeal carcinomas treated by stereotactic radiosurgery. Although the median follow-up of 10½ months was short, the overall response rate was excellent at 60%, and complications were said to be minimal. Orecchia et al¹⁹ treated 13 patients with hypofractionated stereotactic radiotherapy (2400 rad [24 Gy] in 2 or 4 fractions) with a 3-year overall survival of 31%, and no severe acute or late complications were seen.

Surgical resection offers an advantage over additional irradiation for rT1 and rT2 tumors, but newer modalities of XRT offer great promise. Although not supported by data from this study, patients with rT3 tumors should be treated either with surgery plus postoperative XRT or with XRT alone. Since many of these patients with rT3 disease develop distant metastases, consideration should be given to administering chemotherapy. Surgery would not seem to benefit patients with rT4 tumors, except for palliation. A randomized prospective trial comparing surgery to external beam irradiation plus implant or stereotactic radiosurgery should be performed in countries where the disease is endemic, paying particular attention to 2- to 5-year rates of complications.

Accepted for publication October 2, 2001.

This study was presented at the annual meeting of the American Head and Neck Society, Palm Desert, Calif, May 14, 2001.

Corresponding author and reprints: Willard E. Fee, MD, Stanford University Medical School, Stanford, CA 94305-5328 (e-mail: wfee@stanford.edu).