Customize your JAMA Network experience by selecting one or more topics from the list below.
Tassone P, Savard C, Topf MC, et al. Association of Positive Initial Margins With Survival Among Patients With Squamous Cell Carcinoma Treated With Total Laryngectomy. JAMA Otolaryngol Head Neck Surg. 2018;144(11):1030–1036. doi:10.1001/jamaoto.2018.1095
Are positive initial margins associated with survival among patients with squamous cell carcinoma who undergo total laryngectomy and have negative final margins?
In this cohort study of 225 patients with squamous cell carcinoma who underwent total laryngectomy with negative final margins, having positive initial margins was associated with significantly worse disease-free survival and, in particular, worse local disease-free survival among patients who underwent primary total laryngectomy.
Patients who undergo primary total laryngectomy who require re-resection for positive initial margins may be at increased risk for local disease recurrence.
Frozen-section analysis of margin status is routinely performed during total laryngectomy for squamous cell carcinoma; in the case of positive initial frozen margins, re-resection to negative margins can often be accomplished. Some evidence suggests that positive initial margins may be associated with worse survival among patients with head and neck cancer even when final margins are negative, but the significance of positive initial frozen margins in total laryngectomy is poorly understood.
To determine the association between disease-free survival and positive initial margins among patients treated with total laryngectomy for squamous cell carcinoma.
Design, Setting, and Participants
In this retrospective cohort study conducted at a tertiary care center, records of 237 consecutive patients treated with total laryngectomy for squamous cell carcinoma from February 2008 to July 2016 were reviewed for demographic characteristics, disease staging, pathologic markers, and outcomes. Margins were considered to be positive if they contained invasive carcinoma, carcinoma in situ, or severe dysplasia. Data analysis was performed from March to June 2018.
All patients had undergone total laryngectomy for squamous cell carcinoma.
Main Outcomes and Measures
Variables associated with disease-free survival were analyzed using a univariable and multivariable Cox proportional hazards model.
Among all 225 patients (184 [92%] male; mean age, 63.9 years; range, 30-92 years) who underwent total laryngectomy and had negative final margins, 127 patients underwent primary total laryngectomy and 98 underwent salvage total laryngectomy, with mean (SEM) follow-up of 29.3 (1.8) months. Initial frozen margins were positive in 40 of 225 patients (18%); positive initial margins occurred in 21 of 127 patients who underwent primary total laryngectomy (17%) and in 19 of 98 patients who underwent salvage total laryngectomy (19%). On multivariable analysis, only salvage laryngectomy was associated with significantly worse disease-free survival (hazard ratio [HR], 3.35; 95% CI, 1.76-6.36). Among 98 patients who underwent salvage total laryngectomy, positive lymph nodes, lymphovascular invasion, extracapsular nodal extension, and adjuvant therapy were associated with worse disease-free survival on univariable analysis, but no factors remained significant on multivariable analysis. Among 127 patients who underwent primary total laryngectomy, positive nodes, lymphovascular invasion, extracapsular nodal extension, and positive initial margins were associated with worse disease-free survival; however, only positive initial margins were associated with significantly worse disease-free survival on multivariable analysis (HR, 5.01; 95% CI, 1.55-16.2).
Conclusions and Relevance
In this study, positive initial margins were associated with worse disease-free survival among patients who underwent primary total laryngectomy despite negative margins on final pathologic examination. This finding may indicate aggressive tumor behavior in the context of primary laryngeal squamous cell carcinoma.
Total laryngectomy (TL) is the definitive surgical treatment of advanced laryngeal cancer, and approximately 3000 laryngectomies are performed in the United States each year, usually for the treatment of malignant disease.1 Although total laryngectomy has often been used as salvage treatment for patients with laryngeal cancer,2,3 the paradigm of organ preservation for advanced laryngeal and hypopharyngeal cancer may be shifting4; surgery-first strategies may offer better oncologic outcomes for both advanced hypopharyngeal5-8 and advanced laryngeal cancer.9-12
Whether used as initial treatment or as salvage treatment, the goal of TL is to completely remove malignant tissue and achieve negative margins. Positive final margins are a powerful predictive factor associated with worse survival among patients with TL.13-16 In 1978, Byers et al17 described the use of intraoperative frozen margins to predict final margin status in head and neck ablative surgery, and later studies have corroborated the excellent accuracy of intraoperative frozen margins in the head and neck.18-21 Earlier studies have generally considered margins as either positive or negative on the basis of final margins status, without consideration of frozen margin status. Ettl et al22 reported that positive initial margins were associated with worse locoregional control in head and neck resections even when final margins were negative. In this series, however, only 13 of 156 patients (8%) had laryngeal tumors, with 143 patients (92%) having oral cavity tumors. In addition, no patients who received salvage treatment were included. Studies on the methods of obtaining frozen sections have been conducted, but these studies have involved oral cavity tumors and not TL specimens.23-25 In short, the association between positive frozen margins and oncologic outcomes in TL is poorly understood.
In this study, we reviewed outcomes of patients who underwent TL for squamous cell carcinoma to determine whether frozen margin status was associated with oncologic outcomes.
This study was approved by the institutional review board of Thomas Jefferson University, Philadelphia, Pennsylvania, which waived the need for patient consent. After institutional review board approval, the records of 237 consecutive patients who underwent TL for squamous cell carcinoma of either the larynx or hypopharynx from February 2008 to July 2016 were reviewed. All data were deidentified. Data were reviewed for demographic information including age and sex; clinical information including tumor (T) stage, lymph node (N) stage, disease site, and earlier treatment; pathologic information including perineural invasion, lymphovascular invasion, extracapsular nodal extension, initial frozen margin status, and final margin status; treatment data including adjuvant radiation therapy and chemotherapy; and follow-up information including length of follow-up, local or distant recurrence, vital status, and development of fistula. Data were analyzed from March to June 2018.
For an initial analysis, all 237 patients, including those with positive final margins, were included. For additional analyses, patients were excluded if they had positive final margins, defined as invasive carcinoma, carcinoma in situ, or severe dysplasia in separately submitted margins. Although margins are occasionally obtained directly from the TL specimen at Thomas Jefferson University, they are most commonly obtained from the patient after extirpation of the tumor. In most cases, mucosal margins are taken from the left and right superior and inferior pharynx, left and right tongue base, post cricoid mucosa, and trachea, for a total of 8 mucosal margins. Additional deep margins are sent for evaluation when there is deep infiltration of tumor into soft tissues of the tongue or neck. In the case of positive frozen margins, a re-resection is performed, followed by evaluation of a second frozen margin.
The association of prognostic variables with disease-free survival and disease-specific survival was determined using the Cox proportional hazards regression model for both univariable and multivariable analyses. Variables were included in multivariable analysis if they had n > 10 and P < .2 on univariable analysis. P values were 2-sided. Hazard ratios (HRs) were calculated with 95% CIs. Survival analyses were separately performed for all patients, patients who underwent primary laryngectomy, and patients who underwent salvage laryngectomy. Power calculations were performed to determine the power to detect an HR with P < .05. Cumulative incidence survival analysis was performed to determine the association between positive or negative margins and local and distant disease-free survival, with death as a competing risk.26,27 Data were analyzed using R statistics package, version 3.3.3, with extension packages survival and cmprsk (R Foundation).
An initial investigation of the association of margin status with disease-free survival was performed among all 237 patients with TL (male sex, 194; mean [range] age, 62.9 [30-92] years). Twelve of 237 patients (5.1%; 7 patients with salvage TL and 5 patients with primary TL) had positive final margins, 40 patients (16.8%) had positive initial margins but negative final margins, and 185 patients (78.1%) had negative initial margins and negative final margins. First, final margin status alone was analyzed. On Cox proportional hazards analysis, positive final margins were associated with significantly worse disease-free survival compared with negative final margins (HR, 4.89; 95% CI, 2.56-9.32), consistent with earlier studies.13-16 Next, both final and initial margin status were considered. Using negative initial margins and negative final margins as a reference, positive initial margins and positive final margins were associated with significantly worse disease-free survival (positive initial margins: HR, 1.8 [95% CI, 1.03-3.14]; positive final margins: HR, 5.45 [95% CI, 2.81-10.5]) (Figure 1). A Kaplan-Meier survival curve of disease-free survival by margin status is shown in Figure 1.
Because positive final margins are already known to be associated with worse overall survival, most analyses were then performed only for patients with negative final margins. Twelve patients with positive final margins were excluded, and among the remaining 225 patients (184 [92%] male; mean age, 63.9 years [range, 30-92 years]), 40 (18.7%) had positive initial margins that had been reexcised to negative final margins. Twenty-five of 40 (62.5%) had positive initial margins on the basis of the presence of severe mucosal dysplasia or carcinoma in situ, 11 (27.5%) had mucosal carcinoma, and 4 (10%) had submucosal carcinoma. In 8 patients (20%) with positive initial frozen margins, a complete circumferential pharyngectomy was performed to clear final margins. In 4 patients (2 who underwent primary TL and 2 who underwent salvage TL), frozen margins that were initially identified as negative were later determined to be positive on final pathologic examination; these patients underwent re-resection within 2 weeks after the initial laryngectomy.
Demographic characteristics for the 225 patients with negative final margins are shown in Table 1. Among these 225 patients, 127 (56%) underwent primary laryngectomy and 98 (44%) underwent salvage laryngectomy after earlier radiation therapy or chemoradiation therapy. Most patients (163 of 225 [72%]) had laryngeal primary tumors, with the remainder having hypopharyngeal primary tumors. Most of the patients (123 of 225 [55%]) had stage T4a disease, and approximately one-half of the patients (107 of 225 [48%]) had disease positive for nodal metastases. Mean (SEM) follow-up was 29 (1.8) months; during this time, 73 of 225 patients (32%) had disease recurrence, including 43 patients with locoregional recurrence and 40 patients with distant recurrence. Primary TL and salvage TL groups had similar age and sex demographic characteristics. Postoperative fistula occurred in 17 of 127 patients (13%) who underwent primary TL and 29 of 98 patients (30%) who underwent salvage TL, with a significant difference on χ2 comparison (effect size: Cohen d = 0.41; 95% CI, 0.14-0.67).
First, an analysis was performed for all 225 patients who underwent TL and had negative final margins to determine variables associated with disease-free survival. Age, sex, tumor subsite, T and N staging, pathologic markers, initial margin status, adjuvant treatment, and previous treatment were considered for univariable analysis. Deep margins were obtained in addition to mucosal margins in 35 of 225 patients (15%). Several factors were associated with worse disease-free survival on univariable analysis, including disease positive for nodal metastases (HR, 1.88; 95% CI, 1.26-3.22), perineural invasion (HR, 1.77; 95% CI, 1.06-2.96), lymphovascular invasion (HR, 2.31; 95% CI, 1.40-3.79), extracapsular nodal extension (HR, 2.65; 95% CI, 1.62-4.33), and salvage laryngectomy (HR, 1.66; 95% CI, 1.05-2.63). However, only salvage laryngectomy was associated with worse disease-free survival on multivariable analysis (HR, 3.35; 95% CI, 1.76-6.36). Patients who underwent primary laryngectomy had a 5-year disease-free survival of 63%, compared with 47% for patients with salvage laryngectomy.
Next, a survival analysis was performed for 98 patients who underwent salvage TL. The same variables were considered. Several factors were associated with worse disease-free survival on univariable analysis, including disease positive for nodal metastases (HR, 2.69; 95% CI, 1.41-5.11), lymphovascular invasion (HR, 2.01; 95% CI, 1.03-3.94), extracapsular nodal extension (HR, 3.94; 95% CI, 1.83-8.49), adjuvant radiation therapy (HR, 2.44; 95% CI, 1.25-4.78), and adjuvant chemotherapy (HR, 3.29; 95% CI, 1.61-6.75). However, none of these factors remained significant on multivariable analysis.
Next, a survival analysis was performed for 127 patients who underwent primary TL. The same prognostic factors were considered. Several factors were associated with worse disease-free survival on univariable analysis, including disease positive for nodal metastases, lymphovascular invasion, extracapsular nodal extension, and positive initial margins (Table 2). However, only positive initial margins were associated with worse disease-free survival on multivariable analysis (HR, 5.01; 95% CI, 1.55-16.2). Kaplan-Meier curves were generated for patients with primary TL and salvage TL, stratified by initial margin status (Figure 2). Five-year disease-free survival among patients with negative initial margins was 67% compared with 35% among patients with positive initial margins. Positive initial margins were associated with worse disease-free survival among patients with primary TL but not patients with salvage TL despite all patients having negative final margins.
Next, a survival analysis was performed for patients who underwent primary TL with use of the same prognostic factors, with the end point of disease-specific survival. Twenty-six of 127 patients (20%) died of disease during the study period. Variables associated with worse disease-specific survival were similar to those associated with worse disease-free survival; disease positive for nodal metastases, lymphovascular invasion, extracapsular nodal extension, and positive initial margins were associated with worse disease-specific survival on univariable analysis (Table 2). Only positive initial margins were associated with worse disease-specific survival on multivariable analysis (HR, 3.13; 95% CI, 1.14-8.86).
Cumulative incidence survival curves were generated for patients who underwent primary TL for local and distant treatment failure. Local treatment failure was defined as the time from TL to local or regional recurrence (event), death (competing risk), or last follow-up. Distant treatment failure, similarly, was time from TL to metastatic recurrence, death, or last follow-up. Positive initial margins were associated with worse local survival (HR, 4.10; 95% CI, 1.33-12.60) but not worse distant survival (HR, 1.53; 95% CI, 0.44-5.33) among patients who underwent primary TL (Figure 3).
Because TL is often performed to treat advanced laryngeal and hypopharyngeal cancers, it is important to consider the factors associated with successful surgery among these patients. Effective multimodality treatment often begins with effective surgical resection. Although the association of positive final margins with worse survival in TL is well known, our study showed that, despite having negative final margins, patients with positive initial frozen margins had worse disease-free survival. This association was seen among patients who underwent primary TL but not among patients who underwent salvage TL.
In our grouped analysis of 225 patients with negative final margins, only salvage TL status was found to be associated with disease-free survival; this result is not surprising because it is expected that patients who undergo salvage TL will have worse survival compared with patients who undergo primary TL.28 The overall poor survival among patients with salvage TL also likely contributes to the fact that no staging or pathologic features, including positive initial margins, had a significant association with disease-free survival on multivariable survival analysis of patients who underwent salvage TL.
The most interesting result from this investigation is that positive initial margins among patients who had primary TL appeared to be associated with worse disease-free survival, and in particular worse local survival. This association was independent of known prognostic factors, such as disease positive for nodal metastases and extracapsular nodal extension, and it persisted despite most patients having received adjuvant treatment. The oncologic importance of positive initial margins is underscored by their association with worse disease-specific survival. Although it is possible that surgical technique, and in particular cutting across tumors intraoperatively, influenced survival among these patients, all patients eventually achieved negative margins after re-resection. The association with survival may actually indicate an aggressive, infiltrative tumor biology, making it difficult to assess the true edge of a tumor by visual inspection or manual palpation. All of these tumors were approached with the same oncologic principles of complete resection with a safe, wide margin of normal mucosa. Positive frozen margins, therefore, may reflect spread through lymphovascular channels or a wider field of condemned mucosa or field cancerization.29 In these cases, there may be widely diseased mucosa in addition to a discrete tumor, predisposing patients to disease recurrence.
The local influence of initially positive margins is corroborated by our cumulative incidence survival analysis, in which positive initial margins were associated with increased local treatment failure but not increased distant treatment failure. This finding supports the hypothesis of a locally aggressive tumor biology without increased likelihood of distant metastasis.
On our initial analysis of 237 patients with TL, including those with positive final margins, although positive initial margins were not associated with disease-free survival, the association with positive final margins was stronger. Thus, frozen margin status could be considered to be an intermediate risk factor and not as strongly associated with poor outcome as the incomplete resection implied by positive final margins. Although positive final margins are generally accepted as an indication for combination chemotherapy and radiation therapy, further investigation of frozen margin status should be undertaken before it can be used in treatment decisions.
This study is limited by its retrospective nature and by the fact that all patients underwent surgical procedures at a single institution. Patients also were not assessed for medical comorbidities that may be associated with worse overall or disease-specific survival. Our data also do not account for close margins because all margins were submitted separately from the main laryngectomy specimen, and the distance of the tumor from the edge of the main specimen was not routinely measured. In addition, positive margins were defined as both mucosal dysplasia and submucosal spread; these 2 kinds of positive margins may behave differently, and additional studies may be necessary to distinguish between them. Future studies from other institutions are necessary to validate the results reported here. If positive initial margins are associated with a worse prognosis, this information could be valuable for discussions with patients and with multidisciplinary head and neck oncologic teams. Prospectively collected data would be less prone to bias than retrospectively collected data, and future investigations into frozen margin status should be prospective if possible.
Surgical margin status may be an important prognostic factor in head and neck cancer. This study suggests that positive initial frozen margins, even in patients with negative final margins, are associated with increased risk of local treatment failure in the context of primary TL.
Accepted for Publication: May 24, 2018.
Corresponding Author: Patrick Tassone, MD, Department of Otolaryngology–Head & Neck Surgery, Thomas Jefferson University, 925 Chestnut St, Sixth Floor, Philadelphia, PA 19147 (firstname.lastname@example.org).
Published Online: July 26, 2018. doi:10.1001/jamaoto.2018.1095
Author Contributions: Dr Tassone had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Tassone, Luginbuhl, Cognetti.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Tassone, Savard.
Critical revision of the manuscript for important intellectual content: Tassone, Topf, Keane, Luginbuhl, Curry, Cognetti.
Statistical analysis: Tassone, Savard.
Administrative, technical, or material support: Curry.
Supervision: Topf, Luginbuhl, Curry, Cognetti.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Meeting Presentation: This study was presented at the AHNS 2018 Annual Meeting; April 19, 2018; National Harbor, Maryland.
Create a personal account or sign in to: