Kaplan-Meier curves for overall (A), disease-specific (B), and disease-free (C) survival using the log-rank test for trend.
Disease-specific survival related to extracapsular extension.
Gregory S. Weinstein, Bert W. O’Malley, Marc A. Cohen, Harry Quon. Transoral Robotic Surgery for Advanced Oropharyngeal Carcinoma. Arch Otolaryngol Head Neck Surg. 2010;136(11):1079–1085. doi:10.1001/archoto.2010.191
To determine the oncologic and functional outcomes in patients undergoing primary transoral robotic surgery followed by adjuvant therapy as indicated with a minimum of 18-month follow-up for advanced oropharyngeal carcinoma.
Prospective single-center cohort study.
Academic university health system and tertiary referral center.
Forty-seven adults with newly diagnosed and previously untreated advanced oropharyngeal carcinoma.
Transoral robotic surgery with staged neck dissection and adjuvant therapy as indicated.
Main Outcome Measures
Margin status, recurrence, disease-specific and disease-free survival, gastrostomy tube dependence, and safety and efficacy end points.
In the 47 patients enrolled with stages III and IV advanced oropharyngeal carcinoma, mean follow-up was 26.6 months. There was no intraoperative or postoperative mortality. Resection margins were positive in 1 patient (2%). At last follow-up, local recurrence was identified in 1 patient (2%), regional recurrence in 2 (4%), and distant recurrence in 4 (9%). Disease-specific survival was 98% (45 of 46 patients) at 1 year and 90% (27 of 30 patients) at 2 years. Based on pathologic risk stratification, 18 of 47 patients (38%) avoided chemotherapy, and 5 patients (11%) did not receive adjuvant radiotherapy and concurrent chemotherapy in their treatment regimen. At minimum follow-up of 1 year, only 1 patient required a gastrostomy tube.
This novel transoral robotic surgery treatment regimen offers disease control, survival, and safety commensurate with standard treatments and an unexpected beneficial outcome of gastrostomy dependency rates that are markedly lower than those reported with standard nonsurgical therapies.
A recent analysis of the US national cancer database by Chen et al1 revealed an increase in the use of combined chemoradiotherapy for the treatment of advanced oropharyngeal carcinoma (AOC) in the United States between 1985 and 2001. The stated rationale behind the increased use of chemoradiotherapy for AOC has been to improve survival and local control rates compared with radiotherapy alone and to improve functional and quality of life (QOL) outcomes compared with “radical open surgical approaches.”2,3 Forastiere and Trotti, in a balanced editorial, acknowledged that “preservation of the organ does not equate to preservation of function. Severe and chronic mucosal injury and tissue fibrosis as a late toxic effect of intensive chemoradiotherapy regimens may leave patients with poor swallowing function and thus be dependent on a gastrostomy tube.”4(p2066) Indeed, 3 recent studies5- 7 evaluating state-of-the-art intensity-modulated radiotherapy (IMRT) and chemotherapy in the treatment of AOC reported permanent gastrostomy tube rates of 9% to 38%. In an effort to improve access for minimally invasive transoral surgical approaches, the team at the University of Pennsylvania established the world's first research and clinical program in transoral robotic surgery (TORS) using the da Vinci Surgical System (Intuitive Surgical Inc, Sunnyvale, California)8,9 The present study analyzes a cohort of patients with newly diagnosed and previously untreated AOC who participated in the first prospective clinical trial of TORS. The primary objectives of this study were (1) to determine the oncologic outcomes of 47 patients with AOC treated with TORS, staged neck dissection, and adjuvant therapy as indicated; (2) to assess dependence on percutaneous gastrostomy (PEG) for nutrition in patients observed for a minimum of 1 year after treatment; and (3) to assess the safety and efficacy end points of this novel strategy.
We used a prospective single-arm cohort study. Data were collected from a pool of 162 patients enrolled in a human subjects protocol to investigate TORS that was approved by the institutional review board of the Hospital of the University of Pennsylvania, Philadelphia. The preoperative inclusion criteria for the initial study included (1) at least 18 years of age at the time of treatment, (2) indications for diagnostic or therapeutic approaches for benign and malignant diseases of the oral cavity or laryngopharynx, and (3) written informed consent or a consent waiver from the institutional review board. The preoperative contraindications for the original study were (1) unexplained fever or untreated active infection, (2) pregnancy, (3) previous head and neck surgery precluding transoral robotic procedures, and (4) the presence of medical conditions contraindicating general anesthesia or transoral surgical approaches. The intraoperative contraindication for the original study was the inability to adequately visualize the anatomy to perform the diagnostic or therapeutic surgical approach transorally. Of the 162 patients who underwent TORS in the original study, 9 were excluded because they chose not to undergo TORS, resulting in 153 patients in the intent-to-treat population. Three patients in the intent-to-treat population were excluded from the trial because of the inability to achieve adequate exposure to perform surgery at the time of TORS. Of the remaining 150 patients treated with TORS, 47 were identified as having previously untreated AOC and underwent TORS between May 24, 2005, and July 16, 2007, at the University of Pennsylvania. The most recent clinic visit assessment was in February 2009.
The indications and contraindications used were chosen to maximize local control and functional outcome and to avoid vascular injury while allowing for healing by secondary intention. The tumor-related indications for oropharyngeal TORS resection included previously untreated biopsy-proved squamous cell carcinoma of the oropharynx (ie, American Joint Committee on Cancer [AJCC] stages III, IVA, and IVB, including AJCC TNM T1, T2, T3, and T4a cancers). The tumor-related contraindications for TORS resection of AOC included (1) stage IVC except for a curable distant metastasis, (2) unresectability of the involved lymph nodes, (3) AJCC TNM T4a except for the unilateral deep/extrinsic muscle of the tongue, (4) tumor-related trismus, (5) AJCC TNM T4b, and (6) any AJCC T category with invasion of the deep tissues lateral to the constrictor muscles or posterior invasion of the prevertebral fascia. This lateral and deep invasion may be noted radiologically but must be confirmed as fixation laterally or posteriorly by palpation. Nodal unresectability was defined as carotid artery encasement with deep neck structure involvement resulting in the prediction that the node could not be grossly resected and skin invasion with dermal metastasis. Non–tumor-related contraindications included (1) benign causes of trismus or other anatomical findings that preclude transoral access, (2) a retropharyngeal internal carotid artery in which the artery is located directly behind the tonsillar fossa (a contraindication for TORS radical tonsillectomy but not for resection of other oropharyngeal sites), and (3) medical contraindications for either general anesthesia or transoral surgery in which an open wound heals by secondary intention (ie, the need for chronic anticoagulation therapy).
The surgical procedures for TORS of the oropharynx have been previously described by O’Malley et al8 and Weinstein et al.9The surgical sites of all the patients were allowed to heal by secondary intention. The da Vinci Surgical System was used in all cases of TORS. A complete description of how pathologic specimens were managed after TORS is described elsewhere.9 In brief, all the specimens were oriented by the surgeon in the pathology department, and when frozen sections were deemed necessary, they were collected directly from the high-risk portions of the specimen. Cervical lymphadenectomy was performed in all the patients. In 45 patients (96%), the neck dissections were staged and performed less than 3 weeks after TORS. One patient had a neck dissection before presentation for TORS. Another patient chose to undergo chemoradiotherapy and then underwent neck dissection for persistent disease in the neck. The rationale for staged neck dissection and the indications for adjuvant therapy with radiation or concurrent chemoradiation after TORS have been previously described.9 The indications for postoperative radiotherapy were standard indications established in randomized trials,10,11 including the presence of pathologic adverse features, such as perineural invasion, lymphovascular invasion, T4 disease, pN2 disease, and extracapsular extension. Margin-positive disease and extracapsular extension were the main indications for concurrent cisplatin chemotherapy. When this therapy was not tolerated, the alternative regimen that was used included carboplatin or cetuximab. When there was an indication for postoperative radiotherapy based on neck indications (ie, N2), we irradiated the primary site to 60 Gy even with margin-negative disease, which at the University of Pennsylvania is defined as greater than 2 mm. Conversely, when there were no adverse features of the neck or primary site that indicated the need for postoperative radiotherapy, we did not irradiate the primary site or the neck.
In accord with standard reporting and publications for gastrostomy tube dependence, we report the gastrostomy rates at the last clinic visit for living patients at minimum follow-up of 1 year.6,12 As per the standard in the literature, patients who were deceased at final analysis of the data are excluded from the functional analysis of swallowing.
The safety and efficacy end points were collected at varying time points during the study. The blood loss value was collected intraoperatively as recorded in the anesthesia record. When the anesthesiologist documented minimal blood loss, this was assigned a value of 25 mL. The presence and duration of tracheostomy and intubation, and complications within 30 days of TORS, were extracted from the patient medical record.
Selection bias could affect the choice of patients undergoing TORS. To avoid this bias, all eligible patients in the otorhinolaryngology–head and neck surgery outpatient clinic at the Hospital of the University of Pennsylvania were offered the option to participate in the study. Measurement bias was also avoided by exclusively reporting on objective measures (ie, PEG dependence), and accuracy was enhanced by having the clinical research coordinator and the principal investigators (G.S.W. and B.W.O.) check all data points against source documents. To increase the level of objectivity and confidence for QOL assessment, we used the objective measure of PEG dependence, which closely correlates with QOL, rather than using a more subjective questionnaire-based QOL instrument.13 Another potential confounder could be that patients who choose to participate in a surgical study vs a nonsurgical one have fewer comorbidities or better overall health status. To ensure comparability with nonsurgical studies in the literature, preoperative functional status and overall health status were assessed using the Karnosfky score and the Charlson Comorbidity Index, respectively.14
Statistical analyses were computed using a software program (SPSS version 16.0; SPSS, Inc, Chicago, Illinois). Nonparametric statistics were determined using Pearson χ2 and Fisher exact tests. Overall survival was defined as the time from TORS to the date of death. Disease-specific survival was defined as the time from TORS to the date of death secondary to AOC. Death without evidence of disease involved censorship at the patient's date of death. Disease-free survival was defined as the time from TORS until disease recurrence or the date of death. Patients who were unavailable for follow-up were censored as per the disease status at the last clinic visit. If the date when lost to follow-up was before 18 months after TORS, these patients were excluded from the current analysis. Three patients with stage III or IV AOC were unavailable for follow-up before 18 months. All were without evidence of disease or PEG dependence and were last seen 2 1/2 to 15 1/2 months after the procedure. The Kaplan-Meier method was used for analysis of survival data, and survival plots were compared using the log-rank test. All the statistics were 2-tailed, and items were considered significant at P < .05.
Between May 24, 2005, and July 16, 2007, 47 consecutive patients with AJCC stage III or IV oropharyngeal squamous cell carcinoma were treated with primary TORS, followed by adjuvant IMRT or concurrent chemotherapy as clinically indicated using the variables listed in the “Methods” section. Thirteen patients received postoperative radiotherapy alone and 2 patients received adjuvant chemotherapy. Patients who received chemotherapy alone had high-risk features on pathologic examination but had previous radiotherapy for other primary cancers, which precluded postoperative radiotherapy. Twenty-seven patients received postoperative concurrent chemotherapy and radiotherapy. Five patients avoided radiotherapy and chemotherapy. The clinical and pathologic findings of the 5 patients who underwent surgery alone were as follows: (1) tongue base: clinical T1N1/pathologic T1N1 (radiotherapy recommended but refused), (2) tonsil: clinical T3N0/pathologic T3N0 (radiotherapy recommended but refused), (3) tongue base: clinical T2N2B/pathologic T2N2B (radiotherapy recommended but refused), (4) palate: clinical T2N1/pathologic T2N0 (no adjuvant therapy recommended because of pathologic downstaging), and (5) tonsil: clinical T3N1/pathologic T3N0 (radiotherapy recommended but refused). Patient and tumor characteristics are given in Table 1 and TNM staging is given in Table 2.
In 47 primary TORS procedures for oropharynx cancers, final pathologic evaluation revealed 1 positive margin (2%), defined as tumor at the inked margin. Using the present treatment regimen of primary TORS and staged neck dissection with adjuvant chemoradiotherapy as indicated, we achieved local disease control in 46 patients (98%), regional control in 45 (96%), and distant control in 43 (91%) at a minimum of 18 months of follow-up. With mean follow-up for all patients being 26 months (range for living patients, 18-44 months), 3 patients have died of AOC, 3 patients have died of unrelated conditions, and 1 patient is currently living with disease. Actuarial overall survival rates for the cohort were 96% (45 of 47) at 1 year and 82% (27 of 33) at 2 years. Actuarial disease-specific survival was 98% (45 of 46) at 1 year and 90% (27 of 30) at 2 years. Disease-free survival was 96% (45 of 47) at 1 year and 79% (26 of 33) at 2 years. Kaplan-Meier curves evaluating overall, disease-specific, and disease-free survival are depicted in Figure 1. There was a statistically significant difference in overall survival rates with reference to extracapsular extension in the metastatic nodal disease (Figure 2). For patients without extracapsular nodal extension, disease-specific survival at 1 year was 100% (27 of 27) and at 2 years was 100% (15 of 15). For those with extracapsular extension, disease-specific survival at 1 year was 95% (18 of 19) and at 2 years was 80% (12 of 15).
Of the 47 patients treated for AJCC stage III and IV lesions with longer than 18 months of follow-up, 41 patients are living. Of these 41 patients, 40 have adequate swallowing function with no PEG tube. The patient with permanent PEG dependence had a clinical T2N1 tumor of the tonsil with multiple pathologic nodes and extracapsular nodal extension and was treated with TORS and concurrent chemoradiotherapy. However, this patient was treated without the use of IMRT and received his concurrent chemoradiotherapy at an outside institution. Therefore, at the last follow-up visit and a minimum of 1 year, the PEG dependency rate was 2.4%.
Given the unknowns concerning airway swelling after TORS at the inception of the institutional review board–approved protocol, airway management after TORS underwent a prudent progression during the study from routine tracheostomy, to routine postoperative intubation, to the present approach, which is postoperative intubation and which is reserved for cases in which the prediction is that significant postoperative edema is likely. The initial 3 patients who received TORS for AOC underwent planned tracheotomy. All 3 of these patients underwent successful decannulation shortly after discharge from the hospital. Twenty-seven patients remained intubated, and the remaining 17 patients were extubated after TORS. The average number of days that the patients remained intubated was 2.89 days (range, 2-3 days). Two patients underwent unplanned tracheostomy postoperatively, 1 for an exacerbation of his sleep apnea after extubation and another for postoperative upper airway edema combined with alcohol-related delirium. The patient with obstructive sleep apnea underwent decannulation 32 days postoperatively, and the other patient was discharged with the tracheostomy.
Complications in the 30-day postoperative period included 2 PEG-related complications, 1 pneumonia, 2 alcohol withdrawals, and 1 seizure, all of which resolved with treatment. There were no intraoperative or perioperative deaths and no life-threatening bleeding incidents. Mean (SD) blood loss during TORS for AOC was 220.2 (246.6) mL. No patient underwent transfusion.
The primary objectives of this study were (1) to determine the oncologic outcomes of 47 patients with AOC treated with TORS, staged neck dissection, and adjuvant therapy as indicated; (2) to assess PEG dependence for nutrition in patients observed for a minimum of 1 year after treatment; and (3) to assess safety and efficacy end points, including blood loss, presence and duration of tracheostomy and intubation, operative times, and overall complications. Two key findings were identified. The first is that this novel TORS approach for AOC resulted in comparable oncologic outcomes compared with published chemoradiotherapy studies, with disease-specific survival of 90% at 2 years. The second finding was an unexpectedly low PEG dependency rate of only 2.4%; this 1 patient was the only patient in this study who did not receive IMRT and who was treated at an outside hospital. The safety end points demonstrated that no patients required blood transfusions in this cohort study. One patient required permanent tracheotomy. There were no perioperative mortalities.
A limitation of this study is that it was a nonrandomized single-center cohort study. However, this is a problem inherent in most surgical research and publications to date. In the 61 years since introduction of the randomized clinical trial, more than 90% of published surgical series are not randomized.15,16 Another potential limitation of this study is a possible selection bias imposed by the inclusion and exclusion criteria. Although we included all AJCC T1 to T3 classifications, we excluded all T4b and most T4a AOC.
There has been recent evidence showing that a subset of patients with human papillomavirus (HPV)–positive tumors have better oncologic outcomes. With respect to the present study, an additional positive or negative selection bias could exist because we did not prospectively assess HPV status in these patients. Nevertheless, the only prospective studies17,18 evaluating the effect of HPV status on oncologic outcomes after chemoradiotherapy used response to induction chemotherapy to triage poor responders to a surgical arm. Furthermore, these studies did not address the oncologic outcomes in the surgical arms as they relate to HPV status. By excluding “bad-acting poor chemoresponders” from the final treatment outcome analysis, these 2 prospective HPV studies17,18 do not accurately reflect outcomes of the entire patient cohort as they relate to HPV status. Therefore, these prospective HPV trials are not comparable with studies that did not use induction chemotherapy, such as this TORS study. In addition, whereas some previously published retrospective chemoradiotherapy studies revealed that HPV-positive patients with head and neck cancer had improved outcomes, other studies contradicted these findings.17
In this series of AJCC stage III and IV lesions treated with primary TORS, only 1 patient required a permanent PEG tube. This is a particularly important issue because Terrell et al13 and others19 have shown that PEG tube dependence has the most negative impact of the clinical predictors of QOL in patients with head and neck cancer. Three recent studies5- 7 evaluating IMRT in the treatment of AOC report permanent PEG tube rates of 9% to 38% after treatment.
Oncologic comparisons with the literature are problematic given the variability in patient selection and treatment approaches that have been reported. In an analysis of a combined series of open and transoral resection of AOC by Preuss et al,20 a positive margin at the time of surgery was the only significant marker for decreased disease-free survival. The positive margin rate in the present TORS study for AOC was 2.0%. The local, regional, and distant recurrence rates in 2 retrospective series5,6 that reported exclusively on patients who had oropharyngeal carcinoma after concurrent chemotherapy and IMRT are summarized along with the present TORS experience in Table 3. Again, differences in patient selection and length of follow-up preclude making definitive comparisons; however, the oncologic outcomes seem similar in the TORS and chemoradiotherapy IMRT groups for AOC.
Although the cure rates for the TORS strategy are similar to those for combined chemotherapy and IMRT, the addition of chemotherapy to a radiotherapy regimen carries the potential for significantly increased morbidity with a higher rate of long-term gastrostomy tube dependency (9%-38%) compared with the present TORS study (2.4%).5- 7,21 The present data parallel the findings of Walvekar et al,22 who demonstrated that a primary surgical approach allowed for pathologic risk stratification for oropharyngeal carcinoma and, thereby, deintensified treatment and resulted in chemotherapy being avoided in 80% of AJCC stage III patients. In the present study of more advanced AJCC stage III and IV AOC, 38% of patients (18 of 47) avoided chemotherapy.22In addition, the high rate of negative margins allowed for deintensification of the postoperative radiation regimen, and 11% of the TORS patients (5 of 47) avoided radiotherapy and chemotherapy altogether. Even more significant is that when postoperative radiotherapy was administered, it was done with a dose painting technique (210-220 cGy to areas of high risk or extracapsular extension), which intensifies the treatment in cases that warrant this approach.
Although most patients in the present series are in the AJCC T1 and T2 categories, a review of the literature reveals that the distribution of T categories in this series is similar to that in other reported surgical and nonsurgical series (Table 4).6,23,24 Laccourreye et al23 reported on the largest series of nonrobotic transoral lateral oropharyngectomies, on which we9 based the original description of TORS radical tonsillectomy. In a previous publication, we9 noted that Laccourreye's nonrobotic approach has more limited indications than TORS radical tonsillectomy and lacks some of the advantages afforded by the robotic technology. Table 4 indicates that Laccourreye et al23 had approximately 10% more T1/T2 cancers than in the present series. Table 4 also reveals that 2 large series6,24 of primary IMRT and concurrent chemotherapy for oropharyngeal cancer had approximately 10% fewer T1/T2 cancers than in the present series. Although Chen25 et al noted that the overall distribution by T category for all oropharyngeal cancers in the national cancer database between 1996 and 2003 was 55% in T1/T2 and 45% in T2/T3, a search of the existing literature did not reveal the distribution of T1/T2 vs T3/T4 for AOC. It seems that in the present series and in other surgical and nonsurgical series that the T-category distribution is weighted toward T1 and T2 cancers.
Although questions remain concerning the ability to generalize the findings of a prospective cohort trial such as this, others have questioned the external validity of prospective randomized clinical trials when applied to a “real-world” situation.26 Although these data for TORS in AOC are exciting and intriguing, particularly the finding of a much lower PEG dependence rate compared with standard nonsurgical treatments, we do not want to overstate our enthusiasm for this new treatment regimen. In our opinion, the next step is a prospective trial comparing concurrent chemotherapy and IMRT with the primary TORS approach that fully assesses QOL, PEG dependence, and costs.
Correspondence: Gregory S. Weinstein, MD, Department of Otorhinolaryngology–Head and Neck Surgery, University of Pennsylvania Medical Center, 3400 Spruce St, 5 Ravdin, Philadelphia, PA 19104 (firstname.lastname@example.org).
Submitted for Publication: January 14, 2010; final revision received March 26, 2010; accepted July 19, 2010.
Author Contributions: All authors contributed equally to this article and should be regarded as joint first authors. Drs Weinstein and O’Malley had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Weinstein, O’Malley, and Cohen. Acquisition of data: Weinstein, O’Malley, and Cohen. Analysis and interpretation of data: Weinstein, O’Malley, Cohen, and Quon. Drafting of the manuscript: Weinstein, O’Malley, Cohen, and Quon. Critical revision of the manuscript for important intellectual content: Weinstein, O’Malley, Cohen, and Quon. Statistical analysis: Cohen and Quon. Administrative, technical, and material support: Cohen. Study supervision: Weinstein, O’Malley, and Cohen.
Previous Presentation: This study was presented at the Annual Meeting of the American Head and Neck Society; May 31, 2009; Phoenix, Arizona; and is the recipient of the 2009 Robert Maxwell Byers Award.
Financial Disclosure: None reported.