Factors Affecting the Overall Survival After Salvage Surgery in Patients With Recurrent Nasopharyngeal Carcinoma at the Primary Site: Experience With 60 Cases

Objective  To analyze the factors affecting overall survival after salvage surgery in patients with recurrent nasopharyngeal carcinoma at the primary site after a full course of radiotherapy.

Design  Retrospective analysis of 60 consecutive patients treated by surgical resection of the recurrent tumors, with a mean follow-up of 43.1 months (range, 19-96 months).

Setting  Academic tertiary referral center.

Results  The overall survival and locoregional relapse–free survival were 56% and 60% at 2 years, respectively, and 30% and 40% at 5 years. Twenty-nine (81%) of 36 patients died with uncontrolled local disease. The T stage of the recurrent tumors appeared to be an important prognostic factor. Age, sex, pathologic findings, and disease-free interval (time between previous radiotherapy and local recurrence) were not significant prognosis-affecting factors by the log-rank test. Multivariate analysis showed that patients with recurrent tumors of undifferentiated carcinoma, sarcoma, or small cell carcinoma had unfavorable prognoses. Uncontrolled local disease and the emergence of distant metastasis predicted grave results as well. Postoperative irradiation showed some benefit to patients, but the difference was not statistically significant.

Conclusions  The T stage of the recurrence was the prominent prognosis-affecting factor in patients with recurrent nasopharyngeal carcinoma who received salvage surgery. Patients with local recurrence should be carefully selected for the salvage surgery. We recommend this surgery for patients with rT1, rT2, or limited rT3 lesions. The results of surgical resection in terms of local control and overall survival were slightly better than those of high-dose reirradiation, with fewer late complications.

NASOPHARYNGEAL carcinoma (NPC) is a rare malignant neoplasm in Western countries, but it is common in Taiwan. The treatment of NPC with current techniques of radiotherapy can achieve more than 80% local control. However, local failure (persistence and recurrence) occurs in 15.6% to 48.0% of patients with NPC^1-6 after initial radiotherapy. Reirradiation is an established form of salvage treatment, with further local control rates of 9.4% to 35.0%.^2-6 High-dose reirradiation has significant radiation complications, causing morbidity^3-5 and, occasionally, mortality.^3,5

In view of the drawbacks of reirradiation, several centers have begun to treat local recurrence of NPC by surgical resection. Tu et al⁷ reported that 4 of 9 patients with NPC survived for 5 years after surgical resection of local recurrences. Fee et al⁸ reported that tumor control and survival were only slightly better after surgical resection than after high-dose reirradiation. Wei et al⁹ reported actuarial rates of tumor control and overall survival at 3½ years of 42% and 36%, respectively, with the maxillary swing approach for the surgical resection of local recurrence. Morton et al¹⁰ reported good local control in 5 of 7 patients with NPC who had local recurrence treated by the transcervicomandibulopalatal approach to resect the lesion. Recently, King et al¹¹ reported their 12-year experience in the surgical treatment of recurrent NPC in 31 patients. They stated that surgical resection and postoperative irradiation achieved significant survival and tumor control with an acceptable complication rate in selected cases of recurrent NPC. In addition, they concluded that surgical resection with postoperative radiotherapy was a better salvage treatment than reirradiation alone for selected cases of recurrent NPC.

We began using surgical treatment in selected patients with recurrent NPC in 1992. The results for the first 24 patients were reported previously.¹² The purpose of this article is to analyze the factors affecting the overall survival rates with salvage surgery in patients with recurrent NPC at the primary site after a full course of radiotherapy based on our experience with 60 consecutive patients.

Data from 60 irradiated patients with NPC who underwent salvage surgery for recurrent local disease were reviewed. The duration of postoperative follow-up ranged from 19 to 96 months, with a mean of 43.1 months. All recurrent tumors were histologically verified by nasopharyngeal biopsy and categorized according to the guidelines of the World Health Organization.¹³ The majority (49 cases [82%]) of the specimens were either undifferentiated or differentiated nonkeratinizing carcinoma. Eight patients (13%) had well-differentiated squamous cell carcinoma, a very rare histologic type in previously untreated NPC in Taiwan.¹⁴ Two were diagnosed as having postirradiation sarcoma and 1 as small cell carcinoma.¹⁵

The tumor assessment was based on the results of fiberoptic endoscopy of the nasopharynx, cranial nerve examination, and computed tomographic scans and magnetic resonance imaging of the nasopharynx and its vicinity. Routine physical examination, chest x-ray, liver sonography, and whole-body bone scan were performed as well to exclude the presence of distant metastasis. Clinical staging was determined according to the criteria proposed by the American Joint Committee on Cancer.¹⁶ Generally, the patient's condition was defined as inoperable if the tumor involved cavernous sinus or brain, or encased the internal carotid artery on magnetic resonance imaging. Patients who had local recurrence associated with bilateral neck relapses or with distant metastasis were classified as not suitable for salvage surgery of the nasopharynx.

The operation was performed with the patient under general anesthesia in a supine position. Four approaches—the transpalatal,¹² the transmaxillary,¹² the maxillary swing,¹⁷ and the transmandibular^10,12—were used in the tumor resection, depending on the tumor size, location, and previous radiation doses. A tracheotomy was performed in patients who received the transmandibular approach or who had severe trismus. Otherwise, routine oral endotracheal intubation was performed. Selection criteria for the 4 surgical approaches are shown in Table 1. Frozen-section control was applied routinely during tumor resection to ensure clear surgical margins.

Postoperative irradiation was recommended to patients with positive margins or close margins (<2 mm) of the permanent pathology sections. The dose of 5000 rad (50 Gy) in small portals, ie, with a field size of 5 × 5 cm or 6 × 6 cm confined to the tumor bed, was given to eradicate the residual disease.

The overall survival and locoregional relapse–free survival rates were calculated from the date of surgery with the Kaplan-Meier method and compared by the log-rank test. Univariate and multivariate analyses for prognostic factors affecting overall survival were done with the Cox regression method and compared with the likelihood ratio test. The SAS software (SAS Institute Inc, Cary, NC) was used in these analyses.

The mean age at the time of operation was 50.8 years (range, 29-70 years). The male-female ratio was 3.6:1. Fifty-eight patients had local recurrence and 2 patients had local persistence. Half of the patients (52%) had relapses within 2 years. The intervals between the completion of the last radiotherapy and local recurrence varied from 1 month (persistent tumor) to 170 months, with a median of 22 months.

There was no operative mortality in this series. Some minor complications were encountered in 4 of 11 patients who underwent the transpalatal approach. Two patients with small oronasal fistulas were treated with topical silver nitrate cauterization. Two patients had palatal incompetence without social annoyance. Nasopharyngeal and nasomaxillary crusting persisted in those patients who underwent the transmaxillary or maxillary swing approach. Meticulous daily self-cleansing with saline nasal irrigation alleviated the symptom gradually in about 6 months. Mild trismus developed in the patients with divided pterygoid muscles despite postoperative exercise of the mouth opening. In our early use of the salvage surgery, 2 patients treated by the maxillary swing approach developed an oronasal fistula, which was covered with the dental obturator. Subsequently, 2 figure-of-8 sutures of 2-0 silk were applied to approximate the opposite bony palate and 1 mattressed suture to close the soft palate, successfully preventing this complication. Surgical resection of the eustachian tube resulted in otitis media with effusion; repeated myringotomies and aspirations were applied if the patient complained of a blocked ear.

The overall and locoregional relapse–free survival rates of the 60 patients with NPC who received salvage surgery for their recurrence are demonstrated in Figure 1. At 2 and 5 years, the overall survival rates were 56% and 30%, respectively, while the locoregional relapse–free survival rates were 60% and 40% for the same periods. No patient died after 45 months.

The causes of 36 deaths are shown in Figure 2. Twenty-nine (81%) of 36 patients died with uncontrolled local disease, especially in the patients with rT3 or rT4 disease. One patient with rT1 disease died of a cerebrovascular accident 24 months after surgery, and 1 patient with rT2 disease and poor control of diabetes mellitus died of pneumonia 4 months after surgery. These 2 patients had neither local disease nor distant metastasis in the last follow-up before death.

The log-rank test showed that the most important factors affecting 2-year survival were the rT stage and surgical approach (Table 2). Age, sex, disease-free interval (time between previous radiotherapy and local recurrence), pathologic findings, and postoperative irradiation were not statistically significant prognostic factors. Multivariate analysis showed that patients with pathologic findings of undifferentiated carcinoma, postirradiation sarcoma, or small cell carcinoma had higher risk than patients with differentiated nonkeratinizing carcinoma (Table 3). Although postoperative irradiation did not significantly affect survival in multivariate analysis, the relative risk ratio in patients with good surgical margins without postoperative irradiation was 1.13 (95% confidence interval, 0.42-3.09; P = .81) as compared with the baseline of patients with postoperative irradiation who had positive or close (<2 mm) margins. Three patients who refused postoperative irradiation died within 25 months (Figure 3). This suggests that postoperative irradiation has some benefit to the patients.

We performed salvage surgery in 60 irradiated patients with NPC who had recurrent local lesions. The rT stage and the surgical approach appeared to be the most important prognostic factors. However, surgical approach could not be viewed as an independent factor, because it was chosen on the basis of the tumor location, size, and extent of involvement. Thus, this factor was associated with T stage. For example, the transmandibular approach was used in this study in the patients with rT3 or rT4 lesions, which often invaded the nearby structures. Poor prognosis was thus expected. Therefore, the rT stage was the sole crucial factor affecting patient survival.

Computed tomography and magnetic resonance imaging are complementary tests before surgical intervention. If involvement of the lateral wall of the sphenoid sinus, prevertebral fascia, paravertebral muscle, clivus, or cranial nerve is detected, salvage surgery would not be helpful. These patients usually die of uncontrolled local disease.

Although we operated on patients without distant metastasis, 2 of them developed distant metastasis. One had liver metastasis 2 months after surgery and the other one had bone metastasis 3 months after surgery. They were undergoing chemotherapy¹⁸ at the time of this writing. Close follow-up of patients with NPC for 1 year after surgical resection is highly recommended, since both patients with distant metastasis and most of the treatment failures occurred during this period.

The local control rate and the overall survival rate for patients with recurrent NPC treated by high-dose reirradiation were variable. Pryzant et al⁶ reported 5-year overall survival and local control rates of 21% and 35%, respectively, for the 53 patients treated by external irradiation with or without brachytherapy. Lee et al,¹⁹ in their review of 654 patients, found that the 5-year local control rate was 23% after reirradiation. Teo et al⁵ treated 123 patients by either high-dose reirradiation (n = 103) or nasopharyngectomy (n = 20), with 5-year overall survival and local control rates of 9.4% and 18.7%, respectively. King et al¹¹ described 31 patients who received nasopharyngectomy and postoperative irradiation for treatment of their recurrence and found that the 5-year overall survival and local control rates were 47% and 43%, respectively. Our study showed 5-year overall survival and local control rates of 30% and 40%, respectively. The less satisfactory results in this series might be due to the fact that we operated on many patients with late-stage rT3 or rT4 disease.

King et al¹¹ stated that the patients who received postoperative irradiation fared better in both overall survival and local tumor control. However, postoperative irradiation failed to affect the prognosis significantly in our study (Table 2). The appropriate radiation dose for recurrent NPC has been suggested to be more than 6000 rad (60 Gy),^2,4,19 so our dose of 5000 rad (50 Gy) might be less than optimal. Some investigators recommend postoperative irradiation for all patients as a routine procedure.^7,11 Currently, we treat patients with positive margin or closed surgical margins in the permanent section of resected specimens with postoperative irradiation of 6000 rad (60 Gy) or with concomitant radiotherapy and chemotherapy.²⁰

In conclusion, the rT stage was the prominent prognosis-affecting factor in patients with recurrent NPC who underwent salvage surgery. Age, sex, pathologic findings, and disease-free interval did not appear to be statistically significant in prognosis. Patients with local recurrence of NPC should be carefully selected for salvage surgery. We recommend this surgery for patients with rT1, rT2, or limited rT3 lesions. The local control and overall survival rates were better in patients who underwent surgical resection than in those treated solely with high-dose reirradiaton. Finally, although the salvage surgery was performed in the previously irradiated field, it is a safe procedure and perhaps generates fewer late complications than does high-dose reirradiation.

Accepted for publication March 27, 2001.

Presented at the annual meeting of the American Head and Neck Society, Fifth International Conference on Head and Neck Cancer, San Francisco, Calif, July 30, 2000.

We thank Shian-Fen Huang, MS, for statistical analysis and Meng-Chieh Chiang, MD, PhD, for critical review of the manuscript.

Corresponding author and reprints: Mow-Ming Hsu, MD, Department of Otolaryngology, National Taiwan University Hospital, 7 Chung-Shan S Road, Taipei, Taiwan (e-mail: mmhsu@ha.mc.ntu.edu.tw).