T stage at presentation and at salvage with type of salvage laryngectomy performed. PL indicates partial laryngectomy; TL, total laryngectomy.
Overall survival (OS) for T1 to T2 glottic laryngeal tumors that required salvage partial laryngectomy (SPL) or salvage total laryngectomy (STL) following failed radiation. The 5-year OS was 68% for all patients, 89% for the SPL group, and 50% for the STL group (P = .003).
Disease-specific survival (DSS) for T1 to T2 glottic laryngeal tumors that required salvage partial laryngectomy (SPL) or salvage total laryngectomy (STL) following failed radiation. The 5-year DSS was 72% for all patients, 93% for the SPL group, and 51% for the STL group (P = .002).
Neck recurrence–free survival (NRFS) for T1 to T2 glottic laryngeal tumors that required salvage partial laryngectomy (SPL) or salvage total laryngectomy (STL) following failed radiation. The 5-year NRFS was 90% for all patients, 100% for the SPL group, and 80% for the STL group (P = .04).
Distant recurrence–free survival (DRFS) for T1 to T2 glottic laryngeal tumors that required salvage partial laryngectomy (SPL) or salvage total laryngectomy (STL) following failed radiation. The 5-year DRFS was 83% for all patients, 93% for the SPL group, and 71% for the STL group (P = .06).
Effect of age at recurrence on disease-specific survival (DSS) for T1 to T2 glottic laryngeal tumors that required salvage laryngectomy following failed radiation. The 5-year DSS was 94% for those 60 years or younger and 52% for those older than 60 years (P = .01).
Effect of T stage at recurrence on disease-specific survival (DSS) for patients with T1 to T2 glottic laryngeal tumors that required salvage laryngectomy following failed radiation. The 5-year DSS was 100% for patients with T1 tumors, 84% for patients with T2 tumors, 42% for patients with T3 tumors, and 20% for patients with T4 tumors (P<.001).
Effect of T stage at recurrence on distant recurrence–free survival (DRFS) for patients with T1 to T2 glottic laryngeal tumors that required salvage laryngectomy following failed radiation. The 5-year DRFS was 100% for patients with T1 tumors, 90% for patients with T2 tumors, 83% for patients with T3 tumors, and 25% for patients with T4 tumors (P<.001).
Ganly I, Patel SG, Matsuo J, Singh B, Kraus DH, Boyle JO, Wong RJ, Shaha AR, Lee N, Shah JP. Results of Surgical Salvage After Failure of Definitive Radiation Therapy for Early-Stage Squamous Cell Carcinoma of the Glottic Larynx. Arch Otolaryngol Head Neck Surg. 2006;132(1):59-66. doi:10.1001/archotol.132.1.59
To report the outcome of surgical salvage performed for early-stage squamous cell carcinoma of the glottic larynx that recurred or progressed after definitive radiotherapy.
Retrospective outcome analysis.
Tertiary referral center specializing in head and neck cancer.
Forty-three patients who underwent salvage surgery after definitive radiation therapy for early-stage (T1-T2) glottic cancer were identified from a preexisting database of 662 patients with squamous cell carcinoma of the larynx treated at Memorial Sloan-Kettering Cancer Center between the years 1984 and 1998. The T stage at initial presentation was T1 in 20 (18%) and T2 in 23 (32%). Twenty-one patients (49%) were amenable to salvage partial laryngectomy (SPL), but 22 (51%) required salvage total laryngectomy (STL). Details on patient characteristics, tumor characteristics, postoperative complications, and survival outcome were extracted from the database.
Main Outcome Measures
Overall survival, disease-specific survival, neck recurrence–free survival, and distant recurrence–free survival.
No postoperative death occurred following salvage surgery. The overall incidence of complications was 21%, with no difference between the SPL and STL groups. Patients who required STL had poorer overall survival and disease-specific survival compared with patients who required SPL (overall survival, 50% vs 89%; P = .003; disease-specific survival, 51% vs 93%; P = .002). This difference in survival was associated with a poorer neck recurrence–free survival and distant recurrence–free survival in the STL group compared with the SPL group (neck recurrence–free survival, 80% vs 100%; P = .04; distant recurrence–free survival, 71% vs 93%; P = .06). Univariate analysis showed that age and clinical T stage at recurrence were predictors of overall survival, disease-specific survival, and distant recurrence–free survival.
Careful selection of patients with early-stage glottic tumors that recur or progress after radiation allows patients to be successfully treated by partial laryngectomy with excellent survival outcome. However, despite an aggressive policy of performing partial laryngectomy when feasible, up to 50% of patients will require a total laryngectomy owing to progression of disease. These patients have poorer survival outcomes manifested by local, regional, and distant disease progression.
Early-stage squamous cell carcinoma of the larynx can be treated equally effectively with either partial laryngectomy (transoral endoscopic or open surgical resection) or external beam radiotherapy. The overall survival (OS) rates of primary surgery and postoperative radiotherapy vs radiotherapy with surgical salvage are comparable.1,2 Radiotherapy is often preferred as primary treatment because the quality of voice is generally perceived to be better. However, 5% to 40% of the patients will need salvage surgery for persistent or recurrent disease.3- 8 Although it is well recognized that salvage partial laryngectomy (SPL) is an effective option for selected patients in whom radiation therapy fails, few reports in the literature comprehensively analyze patterns of laryngectomy and compare outcomes in patients who require surgical salvage. Our experience with SPL following radiation failure has been reported in the past,3,4 and for many years our practice has been to perform partial laryngectomy for salvage when feasible. Our selection criteria for patients undergoing partial laryngectomy are the same as those reported by Biller et al9 and Shah et al.3 These criteria include subglottic extension not greater than 5 mm, cartilage not invaded as determined by computed tomographic (CT) scan, extension to contralateral cord not greater than 3 mm, arytenoids (except vocal process) free of tumor, mobility of vocal cord preserved, and supraglottic extension no further than the lateral sinus of Morgagni. Patient- and/or tumor-related factors may render partial laryngectomy unsafe, and some patients require total laryngectomy. This article aims to review our experience and report the outcomes after salvage surgery for patients with early-stage glottic laryngeal cancer in whom radiation therapy has failed.
Forty-three patients who underwent salvage surgery following primary radiation therapy for T1 or T2 N0 tumors of the glottic larynx were identified from a preexisting database of 662 patients with squamous cell carcinoma of the larynx treated at Memorial Sloan-Kettering Cancer Center, New York, NY, between 1984 and 1998. Patients had been treated with primary radiation therapy at a dose of 6600 to 7000 rad (66-70 Gy). Details on patient characteristics, tumor characteristics, prior treatment, surgical outcome, and postoperative complications were extracted from the database.
Table 1 and Table 2 show patient and tumor characteristics, respectively. Overall, 41 patients (95%) were men and 2 (5%) were women. The age range was 24 to 78 years, with a median age of 64 years. Medical comorbidity was present in 26 patients (60%). Thirty-eight patients (88%) were smokers, and 26 (60%) were active drinkers of alcohol. At initial presentation, 16 (37%) had T1a, 4 (9%) had T1b, and 23 (54%) had T2 tumors. All tumors were classified as N0 on presentation. Twenty-three tumors (54%) had anterior commissure involvement, and 15 (34%) had subglottic extension at initial presentation. At recurrence, patients had their tumor restaged by careful endoscopic evaluation and CT scan. Details of T stage at recurrence are shown in Figure 1. Of 20 T1 tumors, 7 remained T1, 10 were upstaged to T2, 2 were upstaged to T3, and 1 was upstaged to T4. Of 23 T2 tumors, 5 were downstaged to T1, 8 remained T2, 6 were upstaged to T3, and 4 were upstaged to T4.
Twenty-one patients (49%) were suitable for SPL, but 22 patients (51%) required salvage total laryngectomy (STL). Univariate analysis of patient and tumor characteristics (Tables 1 and 2) showed that patients who underwent STL were more likely to be older than 60 years (P = .03), have a more advanced T stage at initial presentation (P = .002) and at recurrence (P<.001), and have anterior commissure involvement at initial presentation (77% vs 29%; P = .002). Subglottic extension was also higher in those who required STL, although this was not statistically significant (45% vs 24%; P = .20).
Of 21 patients treated by SPL, 13 (62%) had T1 and 8 (38%) had T2 tumors at initial presentation. Partial laryngectomy consisted of endoscopic laser excision in 1 (5%), cordectomy or laryngofissure in 7 (33%), frontolateral partial laryngectomy in 11 (52%), and supracricoid laryngectomy in 2 (10%). No neck dissection was performed in the patients with SPL.
Of the 22 patients who required STL, 7 (32%) had T1 and 15 (68%) had T2 tumors at initial presentation. One patient required flap reconstruction with pectoralis myocutaneous flap. Three patients had N+ neck disease, and all underwent neck dissection at the time of STL (1 had a bilateral jugular neck dissection, 1 had a radical neck dissection, and 1 had a bilateral jugular neck dissection with bilateral paratracheal lymph node dissection). Nineteen patients had N0 neck disease. Of these, 8 patients underwent elective treatment of the neck at the time of STL (3 had unilateral neck dissection with paratracheal node dissection, 3 had unilateral jugular neck dissection alone, 1 had bilateral jugular neck dissection alone, and 1 had unilateral modified radical neck dissection and paratracheal node dissection) and 11 patients had neck observation.
Complications were categorized into overall, wound (infection, dehiscence, flap necrosis, fistula, carotid rupture, or chyle leak), swallowing (dysphagia or stricture), airway (lung, trachea, or stoma), and systemic (myocardial infarction, urinary tract infection, or pulmonary, renal, or metabolic complications). The 2 laryngectomy groups were compared for complication rates using the Fisher exact test or χ2 test.
Follow-up intervals were calculated in months from the date of salvage surgery. Overall survival, disease-specific survival (DSS), local recurrence–free survival (LRFS), neck recurrence–free survival (NRFS), and distant recurrence–free survival (DRFS) rates were calculated using the Kaplan-Meier method. To identify patient and tumor factors predictive of survival outcome, the following variables were analyzed in univariate analysis using the log-rank test: age, sex, medical comorbidity, smoking and alcohol status, initial T stage and T stage at recurrence, anterior commissure involvement, and subglottic invasion. Multivariate analysis was not attempted because of the small number of patients involved. Statistical analysis was performed with SPSS statistical software for Windows version 11.01 (SPSS Inc, Chicago, Ill) and JMP version 4.0 (SAS Institute Inc, Cary, NC).
No postoperative death occurred following salvage surgery. Table 3 gives the incidence of complications. Overall, 21% of patients had complications. The most common were local wound and fistula complications (16% and 14%, respectively). Univariate analysis showed no significant difference in the incidence of complications for patients suitable for SPL after radiation therapy compared with patients who required STL after radiation therapy. Patients in the SPL group had a higher incidence of pharyngocutaneous fistula (19% vs 9%), but this difference was not statistically significant (P = .41).
The median time to salvage surgery from the time of radiation therapy was 10.7 months (range, 3.3-61.1 months); no difference occurred in the time to salvage between the SPL and STL groups (SPL group: 10.8 months; range, 3.3-61.1 months; STL group: 9.8 months; range, 4.0-47.5 months). The median length of follow-up after salvage surgery was 63 months (range, 1-213 months). The 5-year OS after salvage surgery for all patients was 68% (Figure 2); patients who required STL had a poorer OS compared with those who were amenable to SPL (50% vs 89%, respectively; P = .003). The 5-year DSS for all patients was 72% (Figure 3); patients who required STL had a poorer DSS compared with SPL (51% vs 93%; P = .002). The 5-year LRFS for all patients was 97%; no patient with STL had local recurrence, but 1 patient had local recurrence after SPL and required an STL. One other patient in the SPL group required STL at 18 months after SPL because of a persistent pharyngocutaneous fistula. The 5-year NRFS for all patients was 90% (Figure 4); patients who required STL had a poorer NRFS compared with SPL (80% vs 100%; P = .04). None of the patients in the SPL group had neck recurrence, whereas 2 patients in the STL group had neck recurrence. Of these 2 patients, 1 patient had had neck dissection for N+ disease at the time of STL, and 1 patient had had neck dissection for N0 disease at the time of STL. The 5-year DRFS for all patients was 83% (Figure 5); patients who required STL had a poorer DRFS compared with patients in the SPL group (71% vs 93%; P = .06). The site of distant metastatic disease included pulmonary (n = 2), bone (n = 1), and multiple sites (n = 2).
Univariate analysis showed that age and clinical T stage at recurrence were significant predictors of OS, DSS, and DRFS (Tables 4, 5, and 6; Figures 6, 7, and 8). Subglottic involvement showed a trend toward significance for NRFS (Table 7).
Early-stage squamous cell cancer of the larynx can be treated equally effectively with either surgery or radiotherapy.1,2 However, voice quality is generally perceived to be better with radiation, and for this reason radiotherapy is commonly preferred as primary treatment, with surgery kept in reserve for salvage. Failure rates with primary radiation therapy are reported to be 5% to 10% for T1 and 20% to 40% for T2 glottic lesions.3- 8 Surgical salvage is often performed by total laryngectomy because of lack of experience in the technique of conservation surgery of the larynx, as well as the belief that increased complications are associated with partial laryngectomy of irradiated cartilage and that negative tumor margins are difficult to achieve in a fibrotic edematous larynx. Local control rates with STL have been reported at 65% to 85% for T1 and T2 glottic tumors.10,11 Salvage partial laryngectomy is possible in select patients who do not progress with therapy or who experience recurrence with early-stage disease. The use of SPL to conserve function in patients who experience recurrence with limited disease after radical radiotherapy was first reported by Som.12 Local control rates following SPL have been reported at 66% to 96%.3,4,7,13- 15
Our selection criteria for patients undergoing partial laryngectomy were the same as those reported by Biller et al9 and Shah et al.3 Shah et al3 emphasized the importance of excluding patients in whom the recurrent tumor has extended beyond its original site. They also reported that the main predictor of local recurrence was close margins less than 1 mm, emphasizing the use of intraoperative frozen sections to determine resection margins in patients undergoing partial laryngectomy. McLaughlin et al7 also highlighted the importance of CT scanning to determine if SPL was possible, recommending total laryngectomy if cartilage invasion, vocal cord fixation, extensive subglottic disease, or recurrence beyond the original is seen on CT scan.
On the basis of these selection criteria, 21 (49%) of our 43 patients with recurrent glottic tumors were suitable for SPL, but 22 (51%) required STL. Not surprisingly, those who required STL had a more advanced T stage at both initial presentation and recurrence. Patients who had anterior commissure involvement and subglottic involvement at presentation were more likely to undergo STL. Patients who underwent STL were also more likely to be older than 60 years; these patients are less likely to tolerate the respiratory complications that can be associated with partial laryngectomy procedures. These SPL salvage figures are greater than those reported by Stoeckli et al,16 where only 3 (8%) of 39 patients were suitable for SPL. However, these figures are similar to the results reported by Rodriguez-Cuevas et al,17 who reported SPL in 52% of patients with 48% having total laryngectomy. Factors that may account for these differences are the experience of the surgeon in conservation surgery, differences in patient population with regard to T stage, anterior commissure involvement and subglottic involvement, and possible differences in patient age and cardiopulmonary status.
Overall, in our experience no postoperative mortality occurred for the patients undergoing SPL or STL. Patients treated with SPL had an overall complication rate of 19%, with local wound and fistula complications being the most common at 19% each. This is similar to figures reported by Nibu et al18 (overall postoperative complications of 28% with fistula rate of 14%) and Watters et al15 (overall complication rate of 20% with fistula rate of 8%). Therefore, despite radiation, SPL can be performed relatively safely. For STL, the overall complication rate was 23%, with a local complication rate and fistula rate of 14% and 9%, respectively; these figures are also similar to those reported in the literature.
In our patient series, the patients with SPL had an excellent survival outcome, with 5-year OS and DSS figures of 88% and 93%, respectively. These results are comparable to outcome figures reported by Nibu et al.18 Therefore, our study highlights the fact that by careful selection criteria, it is possible to perform SPL in 50% of patients with excellent survival outcomes. For patients who require STL, Rodriguez-Cuevas et al17 reported survival rates similar to patients who underwent SPL. However, this was not the case in our patient series. In our study, patients who required STL had a much poorer OS and DSS compared with those treated with SPL (OS, 50% vs 88%; DSS, 51% vs 93%). The better survival rates for patients with STL reported by Rodriguez-Cuevas and colleagues are most likely attributable to patient selection; for example, patients suitable for SPL may have been treated with STL, accounting for an improved survival in the STL group. Age and cardiopulmonary status may also have been different compared with our patient population.
The poorer survival in our STL group was due not only to more advanced T stage at recurrence but also to poorer neck and distant disease control compared with the patients who had SPL (NRFS, 80% vs 100%; DRFS, 71% vs 93%; for STL and SPL, respectively). This finding might suggest that glottic tumors that require STL for salvage following radiation failure have an inherently more aggressive tumor phenotype compared with tumors that are amenable to SPL. One may argue that patients who require STL presented late, thus missing an opportunity for SPL. However, no difference occurred in the time to recurrence between our SPL and STL groups; the median time to salvage for patients who required STL was 9.8 months after index radiotherapy compared with 10.8 months for the patients who underwent SPL.
One issue of controversy is whether to perform neck dissection, including paratracheal lymph nodes, in patients with recurrent glottic carcinoma after radiation. For primary T1 and T2 glottic larynx cancers, our policy is to irradiate the larynx with a narrow 6 × 6-cm field with the isocenter on the glottic larynx. The superior border of the field is at the hyoid bone, the inferior border at the lower edge of the cricoid cartilage, the posterior border at the vertebral body, and the anterior border 1 cm anterior to the skin surface. Some of the paratracheal nodes get irradiated in this field, but the jugular chain of lymph nodes are not irradiated. The decision of who gets elective neck dissection (END) at the time of recurrence is determined by the T stage at recurrence and other tumor characteristics, such as cartilage invasion and supraglottic or subglottic extension. Patients with early-stage recurrence who are suitable for SPL generally do not need END, since they are likely to have small tumors whose margins are well delineated and remain confined to the larynx. In our study, the 21 patients who were suitable for SPL all had T1 or T2 tumors at the time of recurrence. None of these patients had a lateral or paratracheal lymph node dissection, and no patient with SPL had neck recurrence. On the other hand, recurrent tumors that fail at a higher T stage and need STL are more likely to have extension into the supraglottis or subglottis or other adverse factors, such as cartilage invasion or extralaryngeal involvement. These patients are obviously at higher risk of nodal metastases and are likely to benefit from END. In our study, of 19 patients with STL who had N0 neck disease, 8 had elective treatment of the neck and 11 had no treatment. Of the 11 patients who had neck observation, 7 had T2 tumors at salvage, with only 4 having T3 or T4 tumors. In contrast, 6 of the 8 patients who underwent END had T3 or T4 tumors. Although this difference was not statistically significant (P = .17), it suggests that patients having neck observation had smaller tumors. Therefore, the decision to perform END in patients who required salvage surgery is based on certain well-recognized tumor characteristics, such as T stage at recurrence. However, there is considerable selection bias involved during decision making for END, influenced by both patient- and surgeon-related factors that are not easy to characterize on a retrospective study such as this.
We analyzed age, sex, presence of medical comorbidity, smoking and alcohol status, T stage at index therapy, anterior commissure involvement and subglottic invasion at index therapy, and T stage at recurrence by univariate analysis to identify factors predictive of outcome. Age and clinical T stage at recurrence were significant predictors of OS, DSS, and DRFS. Patients with T1 or T2 tumors at recurrence had a 2- to 5-fold better survival compared with patients with T3 or T4 tumors, whereas patients 60 years or younger had a 2-fold better survival compared with those older than 60 years. Unfortunately, we could not identify any tumor-related factors that were able to predict which patients would require STL before instituting therapy for the index glottic tumor. Identification of such patients is clearly important because it would allow the physician to customize therapy for patients at initial presentation. Analysis of differences in global gene expression by complementary DNA microarray technology may prove to be extremely useful, and we have already used this approach in patients with advanced laryngeal cancer to identify patients who are resistant to organ preservation chemoradiation therapy.19
We acknowledge that our study has several limitations. It is retrospective and therefore susceptible to the deficiencies in data recording and collection inherent in such study designs. The staging of patients with early-stage laryngeal cancer is also highly dependent on the skill of the surgeon in evaluating the function and anatomy of the larynx by flexible endoscopy, but more important, by endoscopic evaluation under general anesthesia. Adequate visualization of certain anatomic areas within the larynx, such as the anterior commissure, subglottic area, and lateral ventricle, is often difficult in the clinic and requires detailed examination under anesthesia using 0°, 30°, 70°, and 120° rigid endoscopes. In addition, endoscopic technology and radiologic imaging have improved during the period of the study. These factors may therefore have an impact on the accuracy and uniformity of the data collected. Notwithstanding these limitations, we conclude that, using careful selection criteria, SPL is a safe therapeutic option for patients with early glottic larynx cancer in whom primary radiation therapy fails. These patients have an excellent 5-year survival outcome and an acceptable incidence of complications. However, despite an aggressive policy of performing partial laryngectomy when feasible, up to 50% of patients will require STL. These patients who need STL have a poorer survival outcome manifested by local, regional, and distant disease progression.
Correspondence: Jatin P. Shah, MD, Head and Neck Service, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021 (email@example.com).
Submitted for Publication: May 3, 2005; final revision received July 29, 2005; accepted August 17, 2005.
Financial Disclosure: None.