Key Points
Question
What are the long-term oncologic outcomes for patients undergoing transoral robotic surgery compared with those undergoing nonrobotic surgery for early-stage oropharyngeal squamous cell carcinoma (SCC)?
Findings
This comparative effectiveness cohort study including 9745 patients with clinical stage T1 or T2 oropharyngeal SCC found that, from 2010 to 2015, the use of transoral robotic surgery increased from 18% to 36% of all surgical procedures for oropharyngeal cancer. Robotic surgery was associated with lower rates of margin positivity, less use of adjuvant chemoradiotherapy, and improved long-term overall survival compared with nonrobotic surgery, even after adjusting for patient- and tumor-related factors.
Meaning
Transoral robotic surgery was associated with improved outcomes, including long-term survival, for patients with early-stage oropharyngeal SCC and should be evaluated in prospective trials.
Abstract
Importance
Transoral robotic surgery has been widely adopted since approval by the US Food and Drug Administration in December 2009, despite limited comparative data.
Objective
To compare the long-term outcomes of transoral robotic surgery with those of nonrobotic surgery for patients with early-stage oropharyngeal cancer.
Design, Setting, and Participants
A retrospective cohort comparative effectiveness analysis was performed of patients in the National Cancer Database with clinical T1 and T2 oropharyngeal squamous cell carcinoma diagnosed between January 1, 2010, and December 31, 2015, who underwent definitive robotic and nonrobotic surgery. Multivariable Cox proportional hazards regression analysis and propensity score matching were performed in patients with known human papillomavirus status to adjust for patient- and disease-related covariates. Survival after robotic and nonrobotic surgery was also compared in 3 unrelated cancers: prostate, endometrial, and cervical cancer. Statistical analysis was performed from April 10, 2019, to May 21, 2020.
Main Outcomes and Measures
Overall survival.
Results
Of 9745 patients (7652 men [78.5%]; mean [SD] age, 58.8 [9.6] years) who met inclusion criteria, 2694 (27.6%) underwent transoral robotic surgery. There was a significant increase in the use of robotic surgery from 18.3% (240 of 1309) to 35.5% (654 of 1841) of all surgical procedures for T1 and T2 oropharyngeal cancers from 2010 to 2015 (P = .003). Robotic surgery was associated with lower rates of positive surgical margins (12.5% [218 of 1746] vs 20.3% [471 of 2325]; P < .001) and lower use of adjuvant chemoradiotherapy (28.6% [500 of 1746] vs 35.7% [831 of 2325]; P < .001). Among 4071 patients with known human papillomavirus status, robotic surgery was associated with improved overall survival compared with nonrobotic surgery in multivariable Cox proportional hazards regression (hazard ratio [HR], 0.74; 95 CI, 0.61-0.90; P = .002). Similar results were seen when analyzing only the subset of facilities offering both robotic and nonrobotic surgery. The 5-year overall survival was 84.8% vs 80.3% among patients undergoing robotic vs nonrobotic surgery in propensity score–matched cohorts (P = .001). By contrast, there was no evidence that robotic surgery was associated with improved survival in other cancers, such as prostate cancer (HR, 0.92; 95% CI, 0.79-1.07; P = .26), endometrial cancer (HR, 0.97; 95% CI, 0.90-1.04; P = .36), and cervical cancer (HR, 1.27; 95% CI, 0.96-1.69; P = .10).
Conclusions and Relevance
This study suggests that transoral robotic surgery was associated with improved surgical outcomes and survival compared with nonrobotic surgery in patients with early-stage oropharyngeal cancer. Evaluation in comparative randomized trials is warranted.
This comparative effectiveness study compares the long-term outcomes of transoral robotic surgery with those of nonrobotic surgery in patients with early-stage oropharyngeal cancer.
Introduction
Patients with early-stage oropharyngeal squamous cell carcinoma (OPSCC) are more likely to receive definitive radiotherapy than surgery because of historical studies showing higher morbidity with surgical approaches.1,2,3 However, the development of minimally invasive approaches via transoral robotic surgery (TORS) has the potential to alter the risk-benefit equation.4,5 Since TORS was approved for treatment of OPSCC by the US Food and Drug Administration in December 2009, numerous studies have shown that TORS is associated with decreased operative time, shorter hospitalizations,6,7,8 and improved postsurgical quality of life vs open surgery.9,10,11
A growing body of evidence supports the effectiveness of TORS for OPSCC.12,13,14,15,16,17,18,19 However, these studies have mostly been single-institution studies with limited sample sizes. Moreover, very few studies compare long-term oncologic outcomes for TORS with other surgical approaches, and survival outcomes having been conflicting. For example, although a retrospective single-institution study reported improved survival with TORS vs open surgery,20 another national retrospective study found similar survival with TORS and nonrobotic surgery.21
Despite the theoretical benefits of minimally invasive surgery, it is not a given that minimally invasive surgical techniques provide superior, or even equivalent, outcomes to other approaches for OPSCC. For example, although randomized trials have shown equivalent survival with minimally invasive surgery vs open surgery in certain cancers,22,23,24 minimally invasive radical hysterectomy causes decreased survival in early-stage cervical cancer.25,26 Thus, the effectiveness of minimally invasive surgery for cancer appears to be context dependent and should be evaluated in each disease site. Given the widespread adoption of TORS for OPSCC despite limited comparative data with other surgical approaches,27,28 it is important to assess the long-term oncologic outcomes of TORS for OPSCC in comparison with nonrobotic surgery.
In this study, we used real-world data from a national cancer registry to compare surgical outcomes and survival among patients with T1 and T2 OPSCC undergoing TORS vs nonrobotic surgery. We also evaluated outcomes for robotic vs nonrobotic surgery in 3 unrelated cancer sites to see if the national-level data on cancer consistently favored one approach vs the other, indicative of widespread patient selection bias across disease sites.
Methods
Data Source
Patient data were obtained from the National Cancer Database (NCDB), a registry sponsored by the American College of Surgeons and the American Cancer Society. The NCDB includes deidentified data from patients treated at more than 1500 Commission on Cancer–accredited facilities in the United States and encompasses more than 70% of newly diagnosed cancer cases nationwide.29 Race/ethnicity and sex were abstracted from the medical record by coders working for the NCDB. Given that this study used deidentified data from a public database, it was deemed exempt from full review and patient consent requirements by the Cedars-Sinai Medical Center Institutional Review Board.
Cohort Selection
We included patients who were 18 years or older with clinical T1 and T2 oropharyngeal cancers (International Classification of Diseases for Oncology, Third Edition [ICD-O-3] codes C01.9, C02.4, C05.1, C09.0, C09.1, C09.8, C09.9, C10.0, C10.1, C10.2, C10.3, C10.4, C10.8, C10.9, and C14.2) of squamous cell carcinoma histologies (ICD-O-3 codes 8050-8084) diagnosed between January 1, 2010, and December 31, 2015, who were undergoing primary surgery (eFigure 1 in the Supplement). We excluded patients with an unknown surgical approach, those not receiving surgery at the diagnosing facility, and those receiving radiotherapy or chemotherapy prior to surgery.
Robotic-assisted or robotic-converted-to-open surgical procedures were classified as TORS. Nonrobotic surgery included open surgery; endoscopic and laparoscopic procedures, including transoral laser surgery; or endoscopic and laparoscopic procedures-converted-to-open surgery. High-volume facilities were defined as the top 5% of centers according to the number of surgical procedures performed from 2010 to 2015.
For patterns of utilization analysis and perioperative mortality assessment, we included patients with any human papillomavirus (HPV) status (positive, negative, or unknown). However, for survival analyses, we excluded patients with unknown HPV status, given the critical prognostic association of this variable with survival in OPSCC.30 We also excluded patients with unknown pathologic T classification, number of examined or positive lymph nodes, margin status, chemotherapy status, or extracapsular extension status to enable adjustments for differences in these covariates between the robotic and nonrobotic surgery groups.
As a separate analysis, we assessed survival for patients with prostate cancer, endometrial cancer, and cervical cancer who underwent surgery to assess for evidence of consistent associations between robotic surgery and outcomes in national registry cancer data. We included women with stage I endometrioid adenocarcinoma diagnosed between 2010 and 2015 who underwent total hysterectomy and women with stage IA2 to IB1 cervical cancer diagnosed between 2010 and 2015 who were undergoing radical hysterectomy with pelvic lymphadenectomy. We also included men with prostate cancer with a Gleason score of 7 to 10 who underwent prostatectomy, but limited analysis to those diagnosed in 2010 given the long natural history of this disease.
Statistical Analysis
Statistical analysis was performed from April 10, 2019, to May 21, 2020. Baseline characteristics were compared using the Welch t test for continuous covariates and the Pearson χ2 test for categorical covariates. We performed multiple imputation by chained equations for patients with unknown race/ethnicity, population density, and median educational level and income according to zip code. Survival estimates were derived via the Kaplan-Meier method and compared using the log-rank test. Ordinary least-squares regression was used to estimate trends in TORS use and 90-day mortality.
Factors associated with 90-day mortality were identified using logistic regression. Univariate and multivariable survival analyses were performed using Cox proportional hazards regression analysis. Two multivariable models were created: 1 model included all covariates and a second parsimonious model included only critical covariates most likely to have an association with survival based on prior studies. Stepwise selection was not performed. The Cox proportional hazards regression assumption was assessed using scaled Schoenfeld residuals, and multicollinearity was assessed using the variance inflation factor.31 Specified interactions of interest between TORS and the following variables were tested one at a time in the main multivariable effects model: HPV status, anatomic site, facility volume, and T classification. For interaction testing, a Bonferroni correction was applied to account for multiple hypothesis testing, with P = .0125 (P = .05/4 comparisons) considered significant.
To account for imbalances in covariates among patients who underwent TORS and nonrobotic surgery, we estimated propensity scores for each patient using a multivariable logistic regression model.32 Propensity score matching was performed using a 1-to-1 nearest-neighbor algorithm with a caliper of 0.05.33
Continuous variables including age, number of positive lymph nodes, and number of examined lymph nodes were modeled using restricted cubic splines to allow for a nonlinear association with survival. The number of knots was determined based on optimizing for the lowest Akaike information criterion and knot placement corresponding to the optimal placements as defined by Harrell.34 Three knots were placed corresponding to the 10th, 50th, and 90th quantiles. All hazard ratio (HR) estimates for these spline functions are presented as the 25th to 75th interquartile change in hazard. Statistical analyses were performed using R, version 3.6.1 statistical software (R Foundation for Statistical Computing)35 with 2-sided tests and P < .05 considered statistically significant unless otherwise specified.
Results
Patterns of Care for TORS Use
From 2010 to 2015, 9745 patients underwent up-front surgery for T1 and T2 OPSCC and met the inclusion criteria for patterns of care analysis (eFigure 1 in the Supplement). A total of 2694 patients (27.6%) underwent TORS and 7051 (72.4%) underwent nonrobotic surgery. Between 2010 and 2015, the proportion of surgical patients undergoing TORS significantly increased, from 18.3% (240 of 1309) in 2010 to 35.5% (654 of 1841) in 2015 (P = .003; Figure 1A). The proportion of facilities performing at least 1 TORS operation also increased, from 6.3% (61 of 966) in 2010 to 13.9% (134 of 966) in 2015 (P = .002; Figure 1B).
Figure 1. Increasing Use of Transoral Robotic Surgery (TORS) Between 2010 and 2015.
A, Primary surgical management of early-stage oropharyngeal squamous cell carcinoma (SCC) over time, stratified by surgical approach (TORS vs nonrobotic surgery). There was a mean absolute 2.9% increase in the use of TORS per year (P = .003). B, Proportion of facilities with at least 1 robotic surgical procedure performed over time (P = .002). In each panel, the P value represents a comparison between the linear regression line and a line with slope of 0. The shaded areas indicate the estimated 95% CIs.
Perioperative Mortality With TORS
During the 2010 to 2015 period, the 90-day perioperative mortality rates for TORS and nonrobotic surgery were similar (1.4% [38 of 2658] vs 1.0% [71 of 6949]; P = .11) (eFigure 2 in the Supplement). Although 90-day perioperative mortality with TORS numerically decreased from 2.5% (6 of 238) in 2010 to 1.2% (8 of 644) in 2015, this decrease did not reach statistical significance (P = .17). TORS had similar 90-day mortality to nonrobotic surgery in univariate logistic regression (odds ratio, 1.53; 95% CI, 0.74-3.18; P = .25) and multivariate logistic regression (odds ratio, 1.56; 95% CI, 0.74-3.31; P = .24).
Surgical Outcomes With TORS
Among patients undergoing definitive surgery, 4071 patients had information regarding HPV status, extranodal extension, and other critical covariables allowing for analysis of outcomes (Table 1; eTable 1 in the Supplement). Patients in this subset who underwent TORS were slightly older (mean [SD] age, 59.2 [9.2] vs 58.5 [9.9] years; P = .01), more often male (83.7% [1462 of 1746] vs 78.2% [1818 of 2325]; P < .001), more likely to have private insurance (64.4% [1124 of 1746] vs 61.5% [1430 of 2325]; P = .03), and more likely to be treated at academic medical centers (85.7% [1497 of 1746] vs 63.3% [1472 of 2325]; P < .001) and higher-volume facilities (73.4% [1282 of 1746] vs 47.3% [1100 of 2325]; P < .001). Patients who underwent TORS were also more likely to have base of tongue cancers (34.2% [598 of 1746] vs 27.5% [640 of 2325]; P < .001) and HPV-positive disease (82.5% [1440 of 1746] vs 69.5% [1617 of 2325]; P < .001).
Table 1. Baseline Characteristics in Patients With Clinical T1 and T2 Oropharyngeal Squamous Cell Carcinoma With Complete Data for Selected Variablesa.
| Characteristic | Patients, No. (%) | P value | ||
|---|---|---|---|---|
| Overall (N = 4071) | Nonrobotic surgery (n = 2325) | TORS (n = 1746) | ||
| Age, mean (SD), y | 58.8 (9.6) | 58.5 (9.9) | 59.2 (9.2) | .01 |
| HPV status | ||||
| Negative | 1014 (24.9) | 708 (30.5) | 306 (17.5) | <.001 |
| Positive | 3057 (75.1) | 1617 (69.5) | 1440 (82.5) | |
| Anatomic site | ||||
| Base of tongue | 1238 (30.4) | 640 (27.5) | 598 (34.2) | <.001 |
| Tonsil | 2583 (63.4) | 1530 (65.8) | 1053 (60.3) | |
| Other | 250 (6.14) | 155 (6.67) | 95 (5.44) | |
| Charlson comorbidity score | ||||
| 0 | 3185 (78.2) | 1802 (77.5) | 1383 (79.2) | .48 |
| 1 | 694 (17.0) | 406 (17.5) | 288 (16.5) | |
| 2 | 148 (3.64) | 92 (3.96) | 56 (3.21) | |
| ≥3 | 44 (1.08) | 25 (1.08) | 19 (1.09) | |
| Pathologic T classification | ||||
| T1 | 1839 (45.2) | 1066 (45.8) | 773 (44.3) | .08 |
| T2 | 1949 (47.9) | 1083 (46.6) | 866 (49.6) | |
| T3 | 214 (5.26) | 129 (5.55) | 85 (4.87) | |
| T4 | 69 (1.69) | 47 (2.02) | 22 (1.26) | |
| Positive lymph nodes, No. | ||||
| Mean (SD) | 2.2 (3.5) | 2.2 (3.3) | 2.3 (3.8) | .53 |
| Median (IQR) | 1.00 (1.00-3.00) | 1.00 (1.00-3.00) | 1.00 (1.00-3.00) | .28 |
| Postoperative radiotherapy | ||||
| No | 1313 (32.3) | 735 (31.6) | 578 (33.1) | .33 |
| Yes | 2758 (67.7) | 1590 (68.4) | 1168 (66.9) | |
| Concomitant chemoradiotherapy | ||||
| No | 2740 (67.3) | 1494 (64.3) | 1246 (71.4) | <.001 |
| Yes | 1331 (32.7) | 831 (35.7) | 500 (28.6) | |
| Margins | ||||
| Negative | 3382 (83.1) | 1854 (79.7) | 1528 (87.5) | <.001 |
| Positive | 689 (16.9) | 471 (20.3) | 218 (12.5) | |
| Extranodal extension | ||||
| No | 2833 (69.6) | 1606 (69.1) | 1227 (70.3) | .43 |
| Yes | 1238 (30.4) | 719 (30.9) | 519 (29.7) | |
| Facility volume | ||||
| Lower | 1689 (41.5) | 1225 (52.7) | 464 (26.6) | <.001 |
| Higher | 2382 (58.5) | 1100 (47.3) | 1282 (73.4) | |
Abbreviations: HPV, human papillomavirus; IQR, interquartile range; TORS, transoral robotic surgery.
Additional baseline data are reported in eTable 1 in the Supplement.
Those receiving TORS were less likely to have positive surgical margins (12.5% [218 of 1746] vs 20.3% [471 of 2325]; P < .001) and were less likely to receive adjuvant concurrent chemoradiotherapy (28.6% [500 of 1746] vs 35.7% [831 of 2325]; P < .001) (Table 1). Hospital length of stay was similar for patients undergoing TORS and those undergoing nonrobotic surgery (mean [SD], 4.3 [4.4] days vs 4.2 [8.0] days; P = .49) (eTable 1 in the Supplement). However, given that patients undergoing nonrobotic surgery were more likely to have tonsillar tumors, which more often involve same-day surgical procedures, we excluded same-day surgical procedures and found that TORS was associated with decreased length of stay (mean [SD], 4.4 [4.4] days vs 5.1 [8.4] days; P = .003) in this subgroup. Last, patients undergoing TORS had more regional lymph nodes examined than those undergoing nonrobotic surgery (mean [SD], 33.7 [17.3] vs 28.5 [18.2]; P < .001) but had similar positive lymph node numbers (mean [SD], 2.3 [3.8] vs 2.2 [3.3]; P = .53) (Table 1).
Survival Analysis
Median follow-up was 37.6 months (interquartile range, 23.6-56.5 months). In comparison with nonrobotic surgery, TORS was associated with improved overall survival (OS) in univariate analysis (HR, 0.64; 95% CI, 0.53-0.76; P < .001) and multivariate analysis in both a parsimonious model restricted to critical variables (HR, 0.74; 95% CI, 0.61-0.90; P = .002) (Table 2) and a model using all covariates (HR, 0.76; 95% CI, 0.62-0.93; P = .007) (eTable 2 in the Supplement). In propensity score–matched cohorts, the 5-year OS rate was 84.8% vs 80.3% with nonrobotic surgery (P = .001) (Figure 2A).
Table 2. Multivariable Cox Proportional Hazards Regression Analysis of Overall Survival in Patients With Early-Stage Oropharyngeal SCC Undergoing Primary Surgery, Using Key Covariates.
| Covariate | Multivariable survival analysis | |
|---|---|---|
| HR (95% CI) | P value | |
| Surgical approach | ||
| Nonrobotic surgery | 1.00 [Reference] | NA |
| TORS | 0.74 (0.61-0.90) | .002 |
| Agea | 1.34 (1.18-1.51) | <.001 |
| Anatomic site | ||
| Base of tongue | 1.00 [Reference] | NA |
| Tonsil | 0.85 (0.70-1.03) | .10 |
| Other | 0.92 (0.66-1.28) | .62 |
| HPV status | ||
| Negative | 1.00 [Reference] | NA |
| Positive | 0.35 (0.29-0.42) | <.001 |
| Facility volume | ||
| Lower | 1.00 [Reference] | NA |
| Higher | 1.02 (0.85-1.24) | .81 |
| Charlson comorbidity score | ||
| 0 | 1.00 [Reference] | NA |
| 1 | 1.61 (1.32-1.97) | <.001 |
| ≥2 | 1.81 (1.33-2.46) | <.001 |
| Pathologic T classification | ||
| T1 | 1.00 [Reference] | NA |
| T2 | 1.47 (1.21-1.78) | <.001 |
| T3 | 1.80 (1.30-2.49) | <.001 |
| T4 | 3.65 (2.37-5.62) | <.001 |
| Positive lymph nodesb | 1.22 (1.06-1.41) | .03 |
| Postoperative radiotherapy | ||
| No | 1.00 [Reference] | NA |
| Yes | 0.59 (0.47-0.74) | <.001 |
| Concomitant chemoradiotherapy | ||
| No | 1.00 [Reference] | NA |
| Yes | 0.87 (0.68-1.10) | .25 |
| Margins | ||
| Negative | 1.00 [Reference] | NA |
| Positive | 1.39 (1.12-1.73) | .003 |
| Extranodal extension | ||
| No | 1.00 [Reference] | NA |
| Yes | 1.64 (1.32-2.05) | <.001 |
Abbreviations: HPV, human papillomavirus; HR, hazard ratio; NA, not applicable; SCC, squamous cell carcinoma; TORS, transoral robotic surgery.
The HR for age is presented as the interquartile change in hazard, from 52 to 65 years of age.
The HR for positive lymph nodes is presented as the interquartile change in hazard, from 1 to 3 positive lymph nodes. For a multivariable model including all covariates, please see eTable 2 in the Supplement.
Figure 2. Overall Survival for Patients With Early-Stage Oropharyngeal SCC Undergoing Either Transoral Robotic Surgery (TORS) or Nonrobotic Surgery in Propensity Score–Matched Cohorts.
Kaplan-Meier estimates of overall survival of patients from all facilities (A) and from facilities that offered both TORS and nonrobotic surgery (B). SCC indicates squamous cell carcinoma.
We performed subgroup analysis to assess for statistical interactions between TORS and variables that could plausibly modulate its association with survival. No significant statistical interactions between the association of TORS with longer survival and key covariates, including HPV status, anatomic site, pathologic tumor classification, and facility volume, were detected in multivariable regression when accounting for multiple hypothesis testing (eFigure 3 in the Supplement).
As a sensitivity analysis, we analyzed the subgroup of patients treated exclusively at facilities performing both TORS and nonrobotic surgery (eTable 3 in the Supplement), given that there could be inherent differences in facilities with TORS capability vs those without TORS capability. Similar to the main analysis, TORS remained associated with improved OS in multivariate analysis in both a model limited to key covariates (HR, 0.72; 95% CI, 0.59-0.88; P = .001) and a full multivariable model (HR, 0.74; 95% CI, 0.61-0.91, P = .005) (eTable 4 in the Supplement). In propensity score–matched cohorts of patients treated at these facilities, 5-year OS was 84.5% in patients undergoing TORS vs 79.3% in those undergoing nonrobotic surgery (P = .007) (Figure 2B). As another sensitivity analysis, we excluded all patients undergoing TORS whose procedure was converted to open surgery, and TORS remained associated with improved OS (HR, 0.76; 95% CI, 0.62-0.92; P = .009) (eTable 4 in the Supplement). Similar results were seen when excluding patients undergoing nonrobotic laparoscopic or endoscopic procedures (HR, 0.71; 95% CI, 0.58-0.87; P = .001) (eTable 4 in the Supplement).
Robotic Surgery in Prostate, Endometrial, and Cervical Cancers
To assess whether national-level cancer data consistently demonstrated longer survival with robotic surgery irrespective of disease site, we evaluated the association of robotic surgery with survival in 3 unrelated disease sites: prostate, endometrium, and cervix. In contrast to what was seen with TORS, there was no evidence that robotic surgery was associated with improved survival in prostate cancer (HR, 0.92; 95% CI, 0.79-1.07; P = .26), endometrial cancer (HR, 0.97; 95% CI, 0.90-1.04; P = .36), or cervical cancer (HR, 1.27; 95% CI, 0.96-1.69; P = .10) in multivariable analysis (eTable 5 in the Supplement). Similarly, propensity score–matched cohorts showed no survival difference between robotic and nonrobotic prostatectomy (5-year OS, 94.9% vs 94.7%; P = .65) (eFigure 4A in the Supplement), robotic and nonrobotic total hysterectomy for endometrial cancer (5-year OS, 90.0% vs 89.9%; P = .28) (eFigure 4B in the Supplement), or robotic and nonrobotic total hysterectomy for cervical cancer (5-year OS, 88.1% vs 90.5%; P = .13) (eFigure 4C in the Supplement).
Discussion
In this study, we observed an increase in the use of TORS for early-stage OPSCC after its approval by the US Food and Drug Administration in December 2009, from 18.3% of all surgical procedures in 2010 to 35.5% in 2015. In addition, the number of facilities offering TORS more than doubled. During this period of increased uptake, perioperative mortality rates remained relatively stable and were similar between those undergoing TORS and those undergoing nonrobotic surgery. TORS was associated with lower rates of margin positivity and use of adjuvant chemoradiotherapy. We found that TORS was also associated with an approximately 25% decrease in relative mortality in comparison with nonrobotic surgery. In propensity score–matched cohorts, this decrease corresponded to an absolute difference of 4.5% in 5-year OS.
Given the unexpected association of TORS with improved survival, we considered the possibility that this finding may be due to unmeasured confounding or patient selection bias. Comparing baseline characteristics of patients undergoing TORS vs nonrobotic surgery, those undergoing TORS were more likely to have certain adverse features associated with worse survival in our models, such as older age and base of tongue primary tumors, but also had somewhat higher prevalence of certain favorable factors such as HPV positivity, private insurance, and residence in higher-income zip codes. Although we adjusted for all these factors in our multivariable models, unmeasured factors, such as smoking status,36 marital status,37 body mass index, and functional status, may have been associated with the survival difference that we observed. Considering this possibility, and that patients treated at institutions offering TORS may be fundamentally different from those treated at institutions without TORS capability, we performed a sensitivity analysis in the subgroup of patients treated at facilities offering both TORS and nonrobotic surgery. This sensitivity analysis minimized most of the demographic differences observed in the main cohort (eTable 3 in the Supplement). We observed that the survival differences between patients undergoing TORS and nonrobotic surgery were at least as pronounced at facilities offering both procedures as in the main cohort, with a 5.2% absolute difference in 5-year OS in propensity score–matched cohorts. In addition, the association of TORS with longer survival was seen across multiple subgroups, including the HPV-positive subgroup, with no specific factor identifying a subgroup with a weaker association between TORS and longer survival based on tests of interaction.
To assess whether patients in real-world data sets undergoing robotic surgery for cancer are generally a more favorable cohort with improved survival owing to nononcologic factors, we examined 3 additional cancers for which robotic surgery is used: prostate, endometrial, and cervical cancer. We found no evidence that robotic surgery was associated with improved survival in any of these sites. The comparison of robotic vs nonrobotic surgery (including both open and nonrobotic laparoscopic approaches) presented here is not identical to previous studies comparing minimally invasive surgery with open surgery in cervical cancer.25,26 When we used our cervical cancer cohort to compare minimally invasive surgery with open surgery, we recapitulated the results of these previous studies showing significantly worse survival with minimally invasive surgery (eFigure 5 in the Supplement). This finding provides supportive evidence that real-world data from national registries can produce concordant results with randomized trials for surgical approaches in cancer. Overall, we did not find evidence that systematic survival differences in robotic vs nonrobotic surgery are present in national registry data. The association of robotic surgery with improved survival appears to be unique to OPSCC.
These results should be considered hypothesis generating. However, there are several reasons that could, in part, explain an association between TORS and improved survival in OPSCC. First, there are technical benefits to TORS for OPSCC, including improved visualization angles, magnified views, increased range of motion, and ability to use en bloc resection techniques despite a minimally invasive approach.38 As evidence to these benefits, patients undergoing TORS had lower rates of margin positivity than those undergoing nonrobotic surgery (12.5% vs 20.3%; P < .001). Alternatively, use of TORS may be a surrogate for surgical acumen, experience, or volume. Patients undergoing TORS had increased numbers of lymph nodes examined (mean, 33.7 vs 28.5; P < .001), a marker of surgical quality that would not be expected to be associated with TORS. This difference was smaller, but still present, in the subgroup of patients treated at facilities offering both TORS and nonrobotic surgery. We adjusted for facility surgical volume, academic center status, and number of lymph nodes examined in our models, but we were unable to directly adjust for individual surgeon volume or experience.
Given the increasing use of TORS for OPSCC, it is reassuring that outcomes in this analysis are not worse than for open surgery, and may even be superior. The effectiveness of TORS vs other surgical approaches is testable in prospective trials, and similar trials have been performed in other cancers.39 Such a trial is more challenging in OPSCC, given that some base of tongue tumors can only be accessed via a robotic approach without splitting the mandible. However, a comparative trial restricted to more accessible subsites, such as the tonsil, soft palate, and accessible base of tongue tumors, could be feasible. An additional reason to pursue prospective trials for TORS is that, despite the promising results we observed, TORS is associated with increased expenses required to purchase, maintain, and use the system, and a better understanding of the clinical benefit would help elucidate the cost-effectiveness of this approach.
An additional caveat is that it is unclear how TORS-based treatment paradigms compare with definitive radiotherapy-based approaches in terms of cancer control, toxic effects, or quality of life. ORATOR (Radiotherapy vs Trans-Oral Robotic Surgery), a phase 2 randomized trial, recently showed similar oncologic outcomes but worse swallowing function in patients with early-stage OPSCC who were treated with TORS vs definitive radiotherapy, although the difference did not meet the trial’s predefined threshold for clinical significance.40 A total of 71% of patients randomized to undergo TORS in ORATOR received postoperative radiotherapy, including 24% receiving postoperative chemoradiotherapy, which is very similar to the patterns of adjuvant therapy use in our study. It is possible that toxicity with TORS-based approaches will be more favorable if TORS enables deintensification of postoperative therapy. Several phase 2 trials have shown promising results with this paradigm,41,42 and phase 3 trials are ongoing (NCT02215265).
Limitations
Several additional limitations should be discussed. Cancer-specific outcomes such as locoregional recurrence, cancer-specific survival, and quality of life were not available for analysis. Also, these results may not be generalizable to patients treated in smaller centers that do not submit data to the NCDB.29 Last, there are other factors that we also were not able to account for, such as type of chemotherapy and radiotherapy quality.
Conclusions
The use of TORS in patients with early-stage OPSCC has increased from 2010 to 2015. Among patients undergoing surgery, TORS was associated with longer survival, lower margin positivity, and less use of adjuvant chemoradiotherapy, compared with nonrobotic surgery. These results are hypothesis generating. The association between robotic surgery and improved survival was unique to OPSCC, and was not observed in prostate, endometrial, or cervical cancers. Further prospective trials would help elucidate the optimal treatment paradigm for early-stage OPSCC.
eFigure 1. Consolidated Standards of Reporting Trials (CONSORT) Diagram Showing Exclusion Criteria for Cohort Analysis
eFigure 2. Perioperative Mortality for Transoral Robotic Surgery (TORS) and Nonrobotic Surgery Between 2010 and 2015
eFigure 3. Forest Plot of the Association of Transoral Robotic Surgery (TORS) With Overall Survival in Multivariable Regression in Selected Subgroups of Patients With Clinical T1-T2 Oropharyngeal Squamous Cell Carcinoma
eFigure 4. Association of Surgical Approach (Robotic vs Nonrobotic) With Overall Survival in Prostate, Endometrial, and Cervical Cancer
eFigure 5. Overall Survival for Women With Stage IA2 or IB1 Cervical Cancer Undergoing Minimally Invasive Surgery (Including Robotic and Nonrobotic Approaches) versus Open Surgery in Propensity Score–Matched Cohorts
eTable 1. Additional Baseline Characteristics in Patients With Clinical T1-T2 Oropharyngeal Squamous Cell Carcinoma With Complete Data for Critical Oncologic Variables Used for Survival Analyses
eTable 2. Univariate and Multivariable Cox Regression Analyses of Overall Survival in Patients With Early-Stage Oropharyngeal Cancer Undergoing Definitive Surgery, Including All Covariates
eTable 3. Baseline Characteristics of Patients With Early-Stage Oropharyngeal Squamous Cell Carcinoma Undergoing Definitive Surgery at Facilities that Perform Both Transoral Robotic Surgery (TORS) and Nonrobotic Surgery
eTable 4. Multivariable Cox Regression Results for the Association Between Transoral Robotic Surgery (TORS) versus Nonrobotic Surgery and Overall Survival in Various Sensitivity Analyses
eTable 5. Multivariable Cox Regression Assessing the Association of Surgical Approach (Robotic vs Nonrobotic) with Overall Survival in Prostate, Endometrial, and Cervical Cancer
References
- 1.Parsons JT, Mendenhall WM, Stringer SP, et al. Squamous cell carcinoma of the oropharynx: surgery, radiation therapy, or both. Cancer. 2002;94(11):2967-2980. doi: 10.1002/cncr.10567 [DOI] [PubMed] [Google Scholar]
- 2.Yeh DH, Tam S, Fung K, et al. Transoral robotic surgery vs. radiotherapy for management of oropharyngeal squamous cell carcinoma—a systematic review of the literature. Eur J Surg Oncol. 2015;41(12):1603-1614. doi: 10.1016/j.ejso.2015.09.007 [DOI] [PubMed] [Google Scholar]
- 3.Cracchiolo JR, Baxi SS, Morris LG, et al. Increase in primary surgical treatment of T1 and T2 oropharyngeal squamous cell carcinoma and rates of adverse pathologic features: National Cancer Data Base. Cancer. 2016;122(10):1523-1532. doi: 10.1002/cncr.29938 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Weinstein GS, O’Malley BW Jr, Snyder W, Sherman E, Quon H. Transoral robotic surgery: radical tonsillectomy. Arch Otolaryngol Head Neck Surg. 2007;133(12):1220-1226. doi: 10.1001/archotol.133.12.1220 [DOI] [PubMed] [Google Scholar]
- 5.O’Malley BW Jr, Weinstein GS, Snyder W, Hockstein NG. Transoral robotic surgery (TORS) for base of tongue neoplasms. Laryngoscope. 2006;116(8):1465-1472. doi: 10.1097/01.mlg.0000227184.90514.1a [DOI] [PubMed] [Google Scholar]
- 6.White H, Ford S, Bush B, et al. Salvage surgery for recurrent cancers of the oropharynx: comparing TORS with standard open surgical approaches. JAMA Otolaryngol Head Neck Surg. 2013;139(8):773-778.Published correction appears in JAMA Otolaryngol Head Neck Surg. 2013;139(12):1290. doi: 10.1001/jamaoto.2013.3866 [DOI] [PubMed] [Google Scholar]
- 7.Lee SY, Park YM, Byeon HK, Choi EC, Kim SH. Comparison of oncologic and functional outcomes after transoral robotic lateral oropharyngectomy versus conventional surgery for T1 to T3 tonsillar cancer. Head Neck. 2014;36(8):1138-1145. doi: 10.1002/hed.23424 [DOI] [PubMed] [Google Scholar]
- 8.Zevallos JP, Mitra N, Swisher-McClure S. Patterns of care and perioperative outcomes in transoral endoscopic surgery for oropharyngeal squamous cell carcinoma. Head Neck. 2016;38(3):402-409. doi: 10.1002/hed.23909 [DOI] [PubMed] [Google Scholar]
- 9.Dean NR, Rosenthal EL, Carroll WR, et al. Robotic-assisted surgery for primary or recurrent oropharyngeal carcinoma. Arch Otolaryngol Head Neck Surg. 2010;136(4):380-384. doi: 10.1001/archoto.2010.40 [DOI] [PubMed] [Google Scholar]
- 10.Dziegielewski PT, Teknos TN, Durmus K, et al. Transoral robotic surgery for oropharyngeal cancer: long-term quality of life and functional outcomes. JAMA Otolaryngol Head Neck Surg. 2013;139(11):1099-1108. doi: 10.1001/jamaoto.2013.2747 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sharma A, Patel S, Baik FM, et al. Survival and gastrostomy prevalence in patients with oropharyngeal cancer treated with transoral robotic surgery vs chemoradiotherapy. JAMA Otolaryngol Head Neck Surg. 2016;142(7):691-697. doi: 10.1001/jamaoto.2016.1106 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Genden EM, Kotz T, Tong CC, et al. Transoral robotic resection and reconstruction for head and neck cancer. Laryngoscope. 2011;121(8):1668-1674. doi: 10.1002/lary.21845 [DOI] [PubMed] [Google Scholar]
- 13.Weinstein GS, Quon H, Newman HJ, et al. Transoral robotic surgery alone for oropharyngeal cancer: an analysis of local control. Arch Otolaryngol Head Neck Surg. 2012;138(7):628-634. doi: 10.1001/archoto.2012.1166 [DOI] [PubMed] [Google Scholar]
- 14.Dabas S, Gupta K, Ranjan R, Sharma AK, Shukla H, Dinesh A. Oncological outcome following de-intensification of treatment for stage I and II HPV negative oropharyngeal cancers with transoral robotic surgery (TORS): a prospective trial. Oral Oncol. 2017;69:80-83. doi: 10.1016/j.oraloncology.2017.04.010 [DOI] [PubMed] [Google Scholar]
- 15.Park YM, Jung CM, Cha D, Kim SH. The long-term oncological and functional outcomes of transoral robotic surgery in patients with hypopharyngeal cancer. Oral Oncol. 2017;71:138-143. doi: 10.1016/j.oraloncology.2017.06.016 [DOI] [PubMed] [Google Scholar]
- 16.de Almeida JR, Li R, Magnuson JS, et al. Oncologic outcomes after transoral robotic surgery: a multi-institutional study. JAMA Otolaryngol Head Neck Surg. 2015;141(12):1043-1051. doi: 10.1001/jamaoto.2015.1508 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Baliga S, Kabarriti R, Jiang J, et al. Utilization of transoral robotic surgery (TORS) in patients with oropharyngeal squamous cell carcinoma and its impact on survival and use of chemotherapy. Oral Oncol. 2018;86:75-80. doi: 10.1016/j.oraloncology.2018.06.009 [DOI] [PubMed] [Google Scholar]
- 18.Doazan M, Hans S, Morinière S, et al. Oncologic outcomes with transoral robotic surgery for supraglottic squamous cell carcinoma: results of the French robotic surgery group of GETTEC. Head Neck. 2018;40(9):2050-2059. doi: 10.1002/hed.25199 [DOI] [PubMed] [Google Scholar]
- 19.Moore EJ, Van Abel KM, Price DL, et al. Transoral robotic surgery for oropharyngeal carcinoma: surgical margins and oncologic outcomes. Head Neck. 2018;40(4):747-755. doi: 10.1002/hed.25055 [DOI] [PubMed] [Google Scholar]
- 20.Ford SE, Brandwein-Gensler M, Carroll WR, Rosenthal EL, Magnuson JS. Transoral robotic versus open surgical approaches to oropharyngeal squamous cell carcinoma by human papillomavirus status. Otolaryngol Head Neck Surg. 2014;151(4):606-611. doi: 10.1177/0194599814542939 [DOI] [PubMed] [Google Scholar]
- 21.Li H, Torabi SJ, Park HS, et al. Clinical value of transoral robotic surgery: nationwide results from the first 5 years of adoption. Laryngoscope. 2019;129(8):1844-1855. doi: 10.1002/lary.27740 [DOI] [PubMed] [Google Scholar]
- 22.Walker JL, Piedmonte MR, Spirtos NM, et al. Recurrence and survival after random assignment to laparoscopy versus laparotomy for comprehensive surgical staging of uterine cancer: Gynecologic Oncology Group LAP2 Study. J Clin Oncol. 2012;30(7):695-700. doi: 10.1200/JCO.2011.38.8645 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Janda M, Gebski V, Davies LC, et al. Effect of total laparoscopic hysterectomy vs total abdominal hysterectomy on disease-free survival among women with stage I endometrial cancer: a randomized clinical trial. JAMA. 2017;317(12):1224-1233. doi: 10.1001/jama.2017.2068 [DOI] [PubMed] [Google Scholar]
- 24.Bregar AJ, Melamed A, Diver E, et al. Minimally invasive staging surgery in women with early-stage endometrial cancer: analysis of the National Cancer Data Base. Ann Surg Oncol. 2017;24(6):1677-1687. doi: 10.1245/s10434-016-5752-8 [DOI] [PubMed] [Google Scholar]
- 25.Ramirez PT, Frumovitz M, Pareja R, et al. Minimally invasive versus abdominal radical hysterectomy for cervical cancer. N Engl J Med. 2018;379(20):1895-1904. doi: 10.1056/NEJMoa1806395 [DOI] [PubMed] [Google Scholar]
- 26.Melamed A, Margul DJ, Chen L, et al. Survival after minimally invasive radical hysterectomy for early-stage cervical cancer. N Engl J Med. 2018;379(20):1905-1914. doi: 10.1056/NEJMoa1804923 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Chen MM, Roman SA, Kraus DH, Sosa JA, Judson BL. Transoral robotic surgery: a population-level analysis. Otolaryngol Head Neck Surg. 2014;150(6):968-975. doi: 10.1177/0194599814525747 [DOI] [PubMed] [Google Scholar]
- 28.Cracchiolo JR, Roman BR, Kutler DI, Kuhel WI, Cohen MA. Adoption of transoral robotic surgery compared with other surgical modalities for treatment of oropharyngeal squamous cell carcinoma. J Surg Oncol. 2016;114(4):405-411. doi: 10.1002/jso.24353 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Boffa DJ, Rosen JE, Mallin K, et al. Using the National Cancer Database for outcomes research: a review. JAMA Oncol. 2017;3(12):1722-1728. doi: 10.1001/jamaoncol.2016.6905 [DOI] [PubMed] [Google Scholar]
- 30.Ang KK, Harris J, Wheeler R, et al. Human papillomavirus and survival of patients with oropharyngeal cancer. N Engl J Med. 2010;363(1):24-35. doi: 10.1056/NEJMoa0912217 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Schoenfeld D. Partial residuals for the proportional hazards regression model. Biometrika. 1982;69(1):239-241. doi: 10.1093/biomet/69.1.239 [DOI] [Google Scholar]
- 32.D’Agostino RB Jr, D’Agostino RB Sr. Estimating treatment effects using observational data. JAMA. 2007;297(3):314-316. doi: 10.1001/jama.297.3.314 [DOI] [PubMed] [Google Scholar]
- 33.Austin PC. A comparison of 12 algorithms for matching on the propensity score. Stat Med. 2014;33(6):1057-1069. doi: 10.1002/sim.6004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Harrell FE. Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. Springer; 2001. doi: 10.1007/978-1-4757-3462-1 [DOI] [Google Scholar]
- 35.R: A language and environment for statistical computing [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2013.
- 36.Grønhøj C, Jensen JS, Wagner S, et al. Impact on survival of tobacco smoking for cases with oropharyngeal squamous cell carcinoma and known human papillomavirus and p16-status: a multicenter retrospective study. Oncotarget. 2019;10(45):4655-4663. doi: 10.18632/oncotarget.27079 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Osazuwa-Peters N, Adjei Boakye E, Chen BY, Tobo BB, Varvares MA. Association between head and neck squamous cell carcinoma survival, smoking at diagnosis, and marital status. JAMA Otolaryngol Head Neck Surg. 2018;144(1):43-50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Hockstein NG, O’Malley BW Jr, Weinstein GS. Assessment of intraoperative safety in transoral robotic surgery. Laryngoscope. 2006;116(2):165-168. doi: 10.1097/01.mlg.0000199899.00479.75 [DOI] [PubMed] [Google Scholar]
- 39.Yaxley JW, Coughlin GD, Chambers SK, et al. Robot-assisted laparoscopic prostatectomy versus open radical retropubic prostatectomy: early outcomes from a randomised controlled phase 3 study. Lancet. 2016;388(10049):1057-1066. doi: 10.1016/S0140-6736(16)30592-X [DOI] [PubMed] [Google Scholar]
- 40.Nichols AC, Theurer J, Prisman E, et al. Radiotherapy versus transoral robotic surgery and neck dissection for oropharyngeal squamous cell carcinoma (ORATOR): an open-label, phase 2, randomised trial. Lancet Oncol. 2019;20(10):1349-1359. doi: 10.1016/S1470-2045(19)30410-3 [DOI] [PubMed] [Google Scholar]
- 41.Ma DJ, Price KA, Moore EJ, et al. Phase II evaluation of aggressive dose de-escalation for adjuvant chemoradiotherapy in human papillomavirus-associated oropharynx squamous cell carcinoma. J Clin Oncol. 2019;37(22):1909-1918. doi: 10.1200/JCO.19.00463 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Swisher-McClure S, Lukens JN, Aggarwal C, et al. A phase 2 trial of Alternative Volumes of Oropharyngeal Irradiation for De-intensification (AVOID): omission of the resected primary tumor bed after transoral robotic surgery for human papilloma virus–related squamous cell carcinoma of the oropharynx. Int J Radiat Oncol Biol Phys. 2020;106(4):725-732. doi: 10.1016/j.ijrobp.2019.11.021 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure 1. Consolidated Standards of Reporting Trials (CONSORT) Diagram Showing Exclusion Criteria for Cohort Analysis
eFigure 2. Perioperative Mortality for Transoral Robotic Surgery (TORS) and Nonrobotic Surgery Between 2010 and 2015
eFigure 3. Forest Plot of the Association of Transoral Robotic Surgery (TORS) With Overall Survival in Multivariable Regression in Selected Subgroups of Patients With Clinical T1-T2 Oropharyngeal Squamous Cell Carcinoma
eFigure 4. Association of Surgical Approach (Robotic vs Nonrobotic) With Overall Survival in Prostate, Endometrial, and Cervical Cancer
eFigure 5. Overall Survival for Women With Stage IA2 or IB1 Cervical Cancer Undergoing Minimally Invasive Surgery (Including Robotic and Nonrobotic Approaches) versus Open Surgery in Propensity Score–Matched Cohorts
eTable 1. Additional Baseline Characteristics in Patients With Clinical T1-T2 Oropharyngeal Squamous Cell Carcinoma With Complete Data for Critical Oncologic Variables Used for Survival Analyses
eTable 2. Univariate and Multivariable Cox Regression Analyses of Overall Survival in Patients With Early-Stage Oropharyngeal Cancer Undergoing Definitive Surgery, Including All Covariates
eTable 3. Baseline Characteristics of Patients With Early-Stage Oropharyngeal Squamous Cell Carcinoma Undergoing Definitive Surgery at Facilities that Perform Both Transoral Robotic Surgery (TORS) and Nonrobotic Surgery
eTable 4. Multivariable Cox Regression Results for the Association Between Transoral Robotic Surgery (TORS) versus Nonrobotic Surgery and Overall Survival in Various Sensitivity Analyses
eTable 5. Multivariable Cox Regression Assessing the Association of Surgical Approach (Robotic vs Nonrobotic) with Overall Survival in Prostate, Endometrial, and Cervical Cancer