Customize your JAMA Network experience by selecting one or more topics from the list below.
Sayed Z, Migliacci JC, Cracchiolo JR, et al. Association of Surgical Approach and Margin Status With Oncologic Outcomes Following Gross Total Resection for Sinonasal Melanoma. JAMA Otolaryngol Head Neck Surg. 2017;143(12):1220–1227. doi:10.1001/jamaoto.2017.2011
What is the association of operative approach and margin status with oncologic outcomes following gross total resection of sinonasal mucosal melanoma?
In this cohort study of 72 patients with nonmetastatic sinonasal mucosal melanoma, no differences in overall survival in patients undergoing endoscopic vs open primary resection were found. Local recurrence rates in patients with positive margins following gross total resection were higher than in patients with negative margins; this was not statistically significant but may be clinically significant.
Select patients with sinonasal mucosal melanoma may benefit from endoscopic surgical approaches whereas positive margin status, although a negative prognostic factor, is likely not the main determinant of survival outcomes.
Sinonasal mucosal melanoma (SMM) is a rare malignant neoplasm characterized by a poor prognosis despite aggressive intervention including wide surgical resection. Margin status has previously been cited as an important prognostic factor for local control and overall survival (OS) in patients who undergo either an open or endoscopic surgical approach. No comparisons have been made, however, in patients who have undergone gross total resection with or without positive margins.
To assess the association of margin status and surgical approach with oncologic outcomes in patients with SMM undergoing gross total resection.
Design, Setting, and Participants
In this cohort study, patients with SMM without evidence of regional or distant disease treated with curative intent in part or full at Memorial Sloan Kettering Cancer Center from 1998 through 2016 were retrospectively assessed. Demographic data, prognostic information, and surgical pathology were reviewed. Operative reports and imaging were used to confirm gross total resection of local disease.
Surgical techniques including open maxillectomy, craniofacial resection, and endoscopic resection.
Main Outcomes and Measures
Three-year local recurrence-free survival (LRFS), disease-free survival (DFS), and OS were calculated using the Kaplan-Meier method. Univariate and multivariable analyses of outcomes were carried out using the Cox proportional hazard regression method.
Seventy-two patients (39 [54%] female; mean [SD] age, 67  years) met the eligibility criteria. Thirty-eight patients (53%) underwent open partial or total maxillectomy with or without ethmoidectomy or sphenoidectomy via a transfacial approach. Fourteen patients (19%) had a more extensive craniofacial approach, and 20 patients (28%) underwent endoscopic resection. The 3-year OS for all patients was 52%. The absolute 3-year difference between patients with open/craniofacial resection vs endoscopic resection for LRFS, DFS, and OS was 11% (95% CI, −21% to 43%), 16% (95% CI, −7% to 39%), and 12% (95% CI, −18% to 41%), respectively. The absolute 3-year difference between patients with a negative margin and patients with a positive margin for LRFS, DFS, and OS was 18% (95% CI, −9% to 45%), 5% (95% CI, −17% to 27%), and 15% (95% CI, −9% to 39%), respectively. Multivariable analysis revealed that none of the adjusted variables (margin status, tumor stage, or surgical approach) were significantly associated with OS.
Conclusions and Relevance
Outcomes for patients with SMM remain poor regardless of operative approach or postoperative margin status.
Despite recent clinical advances in treatment, sinonasal mucosal melanoma (SMM), a rare but aggressive disease, has a markedly poor prognosis. Unlike their cutaneous counterparts, neuroectodermal melanocytes residing in the respiratory mucosa, nasal stroma, and secretory glands serve as the major sites of malignant transformation, giving rise to tumors that together account for 1% to 2% of all melanomas.1 Risk factors remain unclear, although an association with preexisting mucosal melanosis, formaldehyde exposure, and African or Japanese ancestry has been observed.2-5 Given its proclivity for distant metastatic disease and local recurrence, 5-year survival rates for SMM have generally remained less than 35% as the overall worldwide incidence increases.6-9 Interestingly, although the intrinsic behavior of melanoma arising from the squamous mucosa of the oral cavity and the respiratory mucosa of the sinonasal tract differs (the latter demonstrating a lower risk of nodal disease but higher rate of perineural invasion), overall prognosis for both forms remains comparable.10,11
Historically, radical primary surgery with wide margins via a transfacial or transcranial procedure was deemed a necessary, if potentially debilitating, approach for complete microscopically margin-negative resection of these tumors. This notion has been bolstered by several studies that have found prognostic significance in margin status, as well as by the relative unreliability of intraoperative frozen tissue analysis for determination of the extent of surgical resection.12,13
More recent work in the treatment of sinonasal malignant neoplasms has demonstrated the efficacy of endoscopic tumor resections that enable a reduction in patient morbidity without compromising oncologic safety. With the refinement and growing availability of this technology, open surgical resection is increasingly reserved for select cases in which transnasal access alone would prove inadequate. Our aim was to evaluate the traditional doctrine of wide-margin surgery in SMM and ascertain whether gross total resection (GTR) with or without positive microscopic margins would be associated with similar survival outcomes through less invasive means.
To that end, we performed a retrospective analysis on patients who were treated surgically with GTR for localized SMM without nodal or distant metastatic disease at Memorial Sloan Kettering Cancer Center (MSK) between the years of 1997 and 2015.
Following approval by the Memorial Sloan Kettering Cancer Center Institutional Review Board with a waiver of consent granted due to the retrospective nature of the study, 72 patients with mucosal melanoma of the nasal cavity and paranasal sinuses without evidence of regional or distant disease treated with curative intent in part or full at MSK from 1997 through 2015 were retrospectively assessed. Operative reports and postoperative imaging (computed tomography, magnetic resonance imaging, or positron emission tomography/computed tomography) were used to limit the study to only those patients who had undergone GTR without visible residual disease. Medical records were then reviewed for demographic information (sex, age, and race), American Joint Committee on Cancer (AJCC) tumor stage, sinonasal subsite, surgery type, pathologic margin status, adjuvant treatment, and documentation of recurrence or death.
Patients who received surgical treatment with GTR were included. Endoscopic resection was characterized by en bloc transnasal excision in the case of small tumors (1-2 cm). Endoscopic excision of larger or more invasive tumors typically involved piecemeal debulking until the site of attachment could be identified and surgically cleared. In both cases, removal of a margin of adjacent normal tissue was attempted when feasible. However, in endoscopic cases in which the tumor was found to have invaded the anterior skull base or medial orbit wall, disease was removed in total without negative margins. Open surgery, defined as partial or total maxillectomy via a lateral rhinotomy or midface degloving approach, was analyzed separately from patients undergoing more extensive combined craniofacial resection (CFR) with the aid of a neurosurgical team. Patients who underwent initial endoscopic biopsy followed by subsequent open surgery were included in the open surgery cohort. Patients who received adjuvant radiotherapy to the primary site (regardless of total dose or number of fractions) were also included. Patients receiving either systemic immunotherapy (ie, ipilimumab) or targeted therapy (ie, imatinib, sorafenib) at any point before or after surgery were stratified for further analysis.
The primary end points were 3-year overall survival (OS), local recurrence-free survival (LRFS), and disease-free survival (DFS). Overall survival was calculated from the date of surgery to the last date the patient was known to be alive or the death date. A death of any cause was considered an OS event. Local recurrence-free survival was calculated from the date of surgery to date of local sinonasal recurrence or the date of last disease assessment. Disease-free survival noted the time interval between any disease recurrence (local, regional, or distant) and the date of surgical intervention. The Kaplan-Meier method was used to determine 3-year OS, LRFS, and DFS. Patients with T4a or T4b tumors were combined for a portion of the analysis given the paucity of T4b tumors in our group. Unadjusted and adjusted hazard ratios for variables predictive of outcome were calculated using Cox proportional hazard regression. Hazard ratios with corresponding confidence intervals were also calculated. IBM SPSS, version 24.0, was used for the analysis.
Seventy-two patients met the specified eligibility criteria. The majority of patients were female (n = 39 [54%]), white (n = 65 [90%]), and 60 years or older (n = 52 [72%]). Based on AJCC staging guidelines, 42 patients (58%) had T3, 24 patients (33%) had T4a, and 6 patients (8%) had T4b tumors at initial presentation. The primary tumor was located in the nasal cavity in 50 patients (69%).
Fifty-eight patients (81%) received adjuvant radiation therapy and 17 patients (24%) received immunotherapy or targeted systemic therapy during treatment. Thirty-eight patients (53%) underwent open partial or total maxillectomy with or without ethmoidectomy or sphenoidectomy via a transfacial approach. Fourteen patients (19%) underwent an open procedure via a combined CFR. Twenty patients (28%) underwent endoscopic resection. Table 1 shows OS and local recurrence as a function of patient, tumor, and treatment variables. Figure 1 demonstrates the OS, LRFS, and DFS of the entire group.
The 3-year OS for all patients was 52%. At 3 years, the difference in the absolute OS between patients with open/CFR (49%) and endoscopic resection (61%) was 12% (95% CI, −18% to 41%). The absolute 3-year difference between patients with open/CFR and endoscopic resection for LRFS, DFS, and OS was 11% (95% CI, −21% to 43%), 16% (95% CI, −7% to 39%), and 12% (95% CI, −18% to 41%), respectively. At 3 years, the difference in absolute OS of all nasal cavity tumors (60%) and paranasal sinus tumors (34%) was 26% (95% CI, −0.1% to 52%). The 3-year absolute difference between patients with T3 (63%) and T4 (39%) disease was 23% (95% CI, −0.7% to 47%). In addition, there was no association that could be determined with the administration of postoperative radiation therapy or targeted treatment/immunotherapy and surgical approach or margin status. The absolute 3-year difference between patients who underwent postoperative radiation therapy for LRFS, DFS, and OS was 19% (95% CI, −13% to 50%), 10% (95% CI, −20% to 39%), and 21% (95% CI, −7% to 50%), respectively. The absolute 3-year difference in OS between patients who did and did not receive immunotherapy or other targeted therapy was 2% (95% CI, −29% to 32%). We did not include the use of immunotherapy as a variable in the analysis of LRFS and DFS because this treatment was indicated for patients who developed distant disease at some point after surgery. When adjusting for stage and margin status on multivariable analysis, surgery type was not associated with of OS (Table 2, model 1). Figure 2 shows the Cox regression survival estimates for model 1 comparing margin status while controlling for stage and surgical approach.
Additional analysis was performed separating the patients who underwent open maxillectomy vs CFR. With patients undergoing open partial or total maxillectomy serving as the reference, the hazard ratio for those undergoing CFR was 1.98 (95% CI, 0.81-4.82) and endoscopic resection was 0.91 (95% CI, 0.40-2.06). Multivariable Cox regression analysis results are reported in Table 2, model 2. As seen in Figure 3, patients who underwent CFR had worse survival than patients who underwent endoscopic or open resection. Grouping open and CFR patients together dilutes the survival effect.
Rates of LRFS at 3 years were 61% and 50% for the open/CFR and endoscopic surgical groups, respectively. The absolute 3-year difference between patients with open/CFR and endoscopic resection was 11% (95% CI, −21% to 43%). Rates of LRFS were 49% and 66% for the positive- and negative-margin groups, respectively. The absolute 3-year difference for patients with negative margins compared with patients with positive margins was 18% (95% CI, −9% to 45%). On multivariable analysis, there was no association of stage, margin, or surgical approach with LRFS (Table 2, model 3).
A total of 54 patients (75%) had evidence of melanoma recurrence at 1 or more sites following surgery. At 3 years, DFS was 31% for open/CFR vs 15% for endoscopic surgical approach. The 3-year absolute difference between the open/CFR and endoscopic surgical approaches was a clinically meaningful 16% (95% CI, −7% to 39%). Disease-free survival for patients with positive vs negative margins was 25% vs 30%, respectively. The 3-year absolute difference in DFS between patients with positive and negative margins was 5% (95% CI, −17% to 27%).
Our contemporary single-institution series analysis attempts to delineate the importance of extent of surgical resection and subsequent margin status for SMM outcomes by examining 3-year OS, LRFS, and DFS. We recognize that direct comparison between surgical approaches poses a challenge because there are substantial selection and treatment biases for patients who undergo endoscopic resection with curative intent. In addition, the relatively low number of patients and a limited staging system make determination of differences in outcome challenging to define. The OS for the whole cohort was 52%, with no associations in outcome between different surgical approaches (open vs endoscopic). The absolute difference of 12% in favor of endoscopic resection is clinically meaningful, yet the wide confidence interval ranging from 18% difference in favor of open/CFR to 41% in favor of endoscopic resection does not support a conclusion of differential outcome by surgical type. Given the wide confidence interval and the location of the null hypothesis, a definitive conclusion regarding the impact on OS cannot be drawn from the data.
Gross tumor resection, that is, the removal of all grossly visible tumor at the time of surgery, has long been the standard treatment for various tumors in the body. Intracranial gliomas and locally advanced liver cancer, for example, are often resected in a tissue-sparing manner. In such cases, the routine use of wide negative margin resection would prove either technically unfeasible or would lead to unacceptable patient morbidity and thus obviate any potential survival benefit.14-16 The outcome of the routine application of this paradigm in SMM, in which the bulk of the tumor is removed through a relatively conservative operation while accepting the increased risk of a positive microscopic margin, has yet to be fully studied. Our findings suggest that positive microscopic margins, although an important prognostic factor, may be less important to long-term survival in SMM than in other less aggressive malignant neoplasms.
The seventh edition of the AJCC mucosal melanoma staging guidelines are based partially on patient prognosis as determined by tumor stage at initial presentation. However, in our study the effect of initial T stage on survival was not statistically significant. We suspect that a study with a larger sample size could capture these differences. Prior studies have shown that SMM of the nasal cavity (particularly of the septum) may have an improved prognosis compared with maxillary and ethmoid sinus subsites. This may be due to the fact that tumors of the nasal cavity may have symptoms at an earlier stage rather than remaining occult.17,18 In our analysis, the 3-year OS of all nasal cavity tumors was 60% compared with 34% in paranasal sinus disease, for an absolute difference in 3-year survival of 26%, which is clinically meaningful and consistent with previous studies. Furthermore, the range of the confidence interval extends from −0.1% to 52%, suggesting that this survival difference may be even larger in the population.
Two prior studies on SMM have reported no association between margin status after resection (regardless of surgical approach) and OS.18,19 In contrast, Penel at al12 in their review of 20 cases of head and neck mucosal melanoma (14 of the oral cavity, 6 of the sinonasal tract) reported that patients with positive margins (30%) were significantly more likely to die (relative risk, 21), thus leading the authors to a conclusion that “obtaining clear margins must be the goal.”12(p993) Few of their patients received postoperative radiation therapy (20%), however, and the clearance of oral cavity margins was potentially less technically challenging due to improved surgical access.12 In our study, the absolute 3-year difference in LRFS for patients with negative margins compared with patients with positive margins was a clinically impressive 18% (95% CI, −9% to 45%), but the wide confidence interval and the location of the null hypothesis do not support a conclusion of a clinically meaningful effect in the population.
Our approach is still to make every reasonable attempt to achieve negative margins during resection, although we admit that this likely plays a lesser role in determining patient outcome due to the highly aggressive nature of the disease overall.
Perhaps not surprisingly, our additional analysis of 14 patients undergoing CFR shows a decreased OS (18% at 3 years), a fact that can likely be attributed to the higher risk of anterior skull base and dural involvement in this subset. In a robust international retrospective analysis of more than 330 patients from 17 institutions, Ganly et al8 noted a 4.5% mortality rate and 33% overall complication rate in patients undergoing CFR for a variety of sinonasal malignant neoplasms. Margin status was deemed an independent predictor of OS although, importantly, SMM accounted for only 6.3% of the patient population in their study.8 Five-year disease-specific survival in this particular subset was 0 in spite of aggressive surgical intervention.8,9
Traditionally, open surgery followed by adjuvant radiation therapy has been used to maximize locoregional control in head and neck mucosal melanoma.20 A large proportion (80%) of patients in our study received adjuvant radiotherapy. A recent review of the MSK experience did not show a survival benefit with adjuvant radiotherapy, although local recurrence rate was improved.21 This is consistent with the effect of adjuvant radiotherapy in cutaneous melanoma.22 The present study was unable to unravel the associations of local control, surgical approach, and adjuvant radiotherapy because of a small sample size and relatively low rate of local recurrence events. In addition, the most effective method of delivering radiotherapy (eg, dose, fractionation, targeting, particle type) remains to be defined. For these reasons, the use of radiotherapy alongside conservative surgical approaches requires further study.
The recent advent of biologic agents, targeted therapies, and immunotherapy for melanoma has also expanded systemic options for treatment.23 It is unclear whether the addition of systemic therapy provides a clinically relevant increase in DFS, although the 2011 US Food and Drug Administration approval of ipilimumab has spurred research in this area.18 Although our analysis contained relatively few patients, the 18 patients receiving either targeted therapy (ie, sorafenib, imatinib) or immunotherapy (ie, ipilimumab) did not show improved OS compared with untreated patients (51% vs 52%) (Table 1). This is almost certainly due to a selection bias whereby only patients who after the surgery went on to develop extensive metastatic or recurrent disease were offered these treatments, often as part of a clinical trial. In many cases, these drugs also served as a “last ditch” treatment option for those whose disease was refractory to first-line therapy. Moreover, one may argue that the relative similarity in survival curves between the 2 groups highlights the therapeutic benefit of targeted therapy/immunotherapy given that a majority of patients receiving this treatment had end-stage disseminated disease. Contemporary case series of patients with widespread metastatic melanoma have shown an objective response rate of between 16% and 30% for imatinib and 12.5% for ipilimumab.23,24 Few of our patients received combined therapy with a programmed cell death 1 checkpoint inhibitor such as nivolumab that may have further improved durable response rates.25 Although the routine use of systemic immunotherapy is not currently indicated as an adjuvant treatment following primary surgery, future longitudinal studies may demonstrate a benefit in this setting.
The goal of surgery in the management of SMM must be clearly defined. To put the patient at considerable risk of complications and loss of function following resection of the orbit or anterior skull base, for example, may not be necessary. When compared with our findings, earlier MSK data from a review of 59 patients with head and neck mucosal melanoma (35 in the sinonasal tract) treated between 1978 and 1998 shows that survival trends have not improved over time.26 Moreover, local recurrence typically continues to occur within 2 years of initial resection (median of 10 months).23 Indeed, radical open resection may be considered futile in certain instances, especially in light of the fact that more than 50% of patients with SMM who achieve initial local control will go on to develop disseminated disease.27 However, relief of troublesome symptoms such as frequent epistaxis and nasal obstruction can still be achieved in cases performed without curative intent. Lund et al1 have also discussed the controversial belief that invasive open surgery for SMM may severely disrupt the host immune response and thereby trigger rapid postoperative recurrence and the hastened progression of metastases.
Attempts at GTR can also play a role in patients with recurrent disease. Ledderose et al28 evaluated 22 patients with recurrent SMM over a 10- year period. In the 12 patients undergoing endoscopic resection, the authors noted a decrease in postoperative pain, as well as decreased hospital length of stay (mean, 4.2 vs 11.2 days) compared with those undergoing extensive open surgical techniques such as maxillectomy.28 Perhaps more tellingly, there was no difference in the subsequent development of distant metastases between the 2 treatment arms, an important determinant in OS.28
Numerous studies have confirmed the feasibility of endoscopic resection for sinonasal malignant neoplasms such as squamous cell carcinoma, esthesioneuroblastoma, and adenoid cystic carcinoma, demonstrating both the oncologic efficacy of endoscopic resection as well as an improvement in quality-of-life metrics over conventional open surgery.29-31 Several authors have argued that accurate tumor margins can be better appreciated and normal intranasal anatomy better preserved with the magnified visualization afforded by the rigid endoscope.32,33 Furthermore, salvage open resection still remains a viable option in the event of endoscopic failure.
Regardless of the surgical approach and extent of resection, overall outcomes in SMM remain consistently poor. Given this fact, a more conservative initial resection may be justified in select patients to minimize morbidity and prioritize patient function over a potentially negligible survival benefit. The importance of the surgeon’s experience with endoscopic techniques cannot be overstated, as those without sufficient experience are unlikely to see this equivalence in outcomes in their own practices. Such an approach should also still strive to achieve negative margins when possible. Appropriate preoperative counseling of patients is also a must because a small number of patients will require conversion to an open approach.
We acknowledge that our study was not without limitations. Although our treatment arms were well matched in terms of demographic data, we did not identify age older than 60 years as a negative prognostic factor. Multiple studies have shown age to be important in both cutaneous and noncutaneous melanomas.34 Moreover, a number of important comparisons on univariate analysis remained underpowered. It should also be noted that there are limitations to the conclusion of lack of statistical difference of hazard ratios on multivariate analysis comparing different surgical approaches because there were a relatively small number of patients and a narrow AJCC staging system range (including only T3 and T4). There may indeed be a clinically significant difference in patient oncologic outcome in the absence of a statistically significant one. There was likely to be selection bias, with those patients with more favorable tumors undergoing transfacial or endoscopic approaches even when they had T4 disease. Additionally, Karnofsky Performance Status and Charlson Comorbidity Index data were not collected for many patients in the beginning of the treatment period so these variables were omitted in our analysis. The omission of a comorbidity index is a limitation in any study comparing OS data because comorbidity is predictive of shorter survival. Therefore, we found it important to assess both LRFS and DFS, recognizing the limitation in OS. It is important to note that the rarity of this disease has made prospective study difficult. To improve the statistical analysis, the accrual of patient data from non–head and neck mucosal subsites, such as the anorectal and genitourinary tracts, is often seen in the literature but may not be clinically relevant to patients with mucosal melanoma in the head and neck.35
Overall outcomes for patients with SMM continue to be poor regardless of operative approach. In our single-institution review, we found no differences in OS in patients undergoing endoscopic vs open/CFR primary resection of localized, nonmetastatic SMM. The rate of local recurrence in patients with positive margins was higher than the rate in patients with negative margins; although this difference may be clinically significant, the imprecision in the estimate prevents us from drawing definitive conclusions. Given our findings, one may consider a primary endoscopic surgical approach in the treatment of select SMM cases, although further study is needed to better quantify outcomes.
Corresponding Author: Marc A. Cohen, MD, MPH, Memorial Sloan Kettering Cancer Center, Head and Neck Service, Department of Surgery, 1275 York Ave, Room C-1075, New York, NY 10065 (email@example.com).
Accepted for Publication: August 26, 2017.
Published Online: October 19, 2017. doi:10.1001/jamaoto.2017.2011
Author Contributions: Dr Cohen had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Sayed, Cracchiolo, Barker, Lee, Tabar, Patel, Cohen.
Acquisition, analysis, or interpretation of data: Sayed, Migliacci, Cracchiolo, Barker, Lee, McBride, Ganly, Morris, Roman, Shoushtari, Cohen.
Drafting of the manuscript: Sayed, Migliacci, Cracchiolo, Lee, Cohen.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Sayed, Migliacci, Lee, McBride, Ganly, Cohen.
Obtained funding: Lee, Cohen.
Administrative, technical, or material support: Lee, Patel, Roman, Cohen.
Study supervision: Barker, Lee, Morris, Cohen.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Barker reports grants from Australia and New Zealand Melanoma Trials Group and Amgen and personal fees from Elsevier, University of Colorado, Charlotte Area Health Education Center, Patient Resouce, University of Florida Health Cancer Center Orlando, and the Driver Group, outside the submitted work. Dr Morris reports other from Merck, outside the submitted work. Dr Shoushtari reports grants and personal fees from Bristol-Myers Squibb and Immunocore and personal fees from Vaccinex and Castle Biosciences, outside the submitted work. No other disclosures are reported.
Funding Support: This research was funded in part through the National Institutes of Health/National Cancer Institute Cancer Center Support Grant P30 CA008748.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Previous Presentations: This study was presented at the American Head & Neck Society 2017 Annual Meeting; April 27, 2017; San Diego, California.