Risk of Pathologic Extranodal Extension and Other Adverse Features After Transoral Robotic Surgery in Patients With HPV-Positive Oropharynx Cancer

Key Points

Question

Among patients with human papillomavirus (HPV)-positive oropharynx squamous cell carcinoma (OPSCC), how common are histopathologic features that suggest the use of adjuvant radiotherapy or adjuvant chemoradiotherapy after surgery?

Findings

In this single-institution cohort study of 136 consecutive patients with cT0-cT2 cN0-N2b HPV-positive OPSCC without definite clinical extranodal extension who were carefully selected for surgery, most had histopathologic indications for possible adjuvant radiotherapy. However, approximately one-quarter had pathologic extranodal extension and/or positive surgical margins, which are indications for adjuvant chemoradiotherapy.

Meaning

These findings suggest that with careful selection of patients with HPV-positive OPSCC for primary surgery, approximately one-quarter will meet histopathologic indications for adjuvant chemoradiotherapy.

This cohort study evaluates the risk of adverse histopathologic features that meet indications for possible adjuvant radiotherapy or chemoradiotherapy in patients with human papillomavirus–positive oropharynx squamous cell carcinoma undergoing transoral robotic surgery with neck dissection.

Abstract

Importance

Understanding patient-specific risk of adverse histopathologic findings after primary surgery for human papillomavirus (HPV)-positive oropharynx squamous cell carcinoma (OPSCC) may help guide patient consultations.

Objective

To determine the likelihood of adverse histopathologic features that may indicate adjuvant radiotherapy or chemoradiotherapy after primary surgery for HPV-positive OPSCC according to 2021 National Comprehensive Cancer Network guidelines.

Design, Setting, and Participants

This retrospective cohort study was performed at a single academic tertiary care center. Of 258 patients who underwent transoral robotic surgery (TORS) from March 1, 2012, to March 1, 2021, 136 consecutive, treatment-naive patients with HPV-positive OPSCC without obvious clinical extranodal extension (ENE) who underwent definitive TORS and neck dissection were included in the analysis. Indications for surgical treatment included non–deeply infiltrative oropharynx tumors, minimal soft palate involvement, and low suspicion for pathologic ENE.

Exposures

Primary site TORS with neck dissection.

Main Outcomes and Measures

The primary outcomes were the adverse histopathologic features of pathologic ENE and positive surgical margins (PSM) that are indications for possible adjuvant chemoradiotherapy. Outcomes were compared among varying American Joint Committee on Cancer 7th edition (AJCC-7) T and N categories and patient clinical characteristics.

Results

Of the 136 patients included in the analysis (113 men [83.1%]; median age, 63 [interquartile range, 55-70] years), 109 (80.1%) had at least 1 indication for possible adjuvant radiotherapy. Twenty-seven patients (19.9%) had pathologic ENE and 10 (7.3%) had PSM. Thirty-four patients (25.0%) had pathologic ENE and/or PSM, whereas 3 (2.2%) had both. Age, smoking history, history of alcohol consumption, and clinical T category were not associated with pathologic ENE, PSM, lymphovascular invasion, perineural invasion, or pN2 category or greater. The proportion of pathologic ENE varied by clinical N category: 0 of 16 for cN0, 8 of 48 (16.7%) for cN1, 3 of 23 (13.0%) for cN2a, and 16 of 45 (35.6%) for cN2b. Compared with patients with cN1-cN2a disease, patients with cN2b disease had higher odds of pathologic ENE (odds ratio, 3.01; 95% CI, 1.14-8.10). Clinical and pathologic N category were concordant in 77 patients (56.6%), whereas 42 (30.9%) were upstaged and 17 (12.5%) were downstaged.

Conclusions and Relevance

In this cohort study, approximately one-quarter of carefully selected patients with HPV-positive OPSCC without obvious clinical ENE undergoing primary surgery had pathologic ENE and/or PSM. Patients with AJCC-7 cT0-cT2 cN0-cN2b disease, especially cN0-cN2a, without signs of clinical ENE may represent appropriate candidates for primary surgery when avoidance of adjuvant chemotherapy and/or reduction of adjuvant radiotherapy dose/extent are the goals.

Introduction

The favorable prognosis of human papillomavirus (HPV)-positive oropharynx squamous cell carcinoma (OPSCC) has spurred research into treatment deintensification to improve posttreatment quality-of-life outcomes.^1,2,3 Transoral robotic surgery (TORS) with neck dissection is now a well-established treatment approach for select patients that can provide excellent oncologic results with acceptable quality-of-life impact.^2,4,5 Primary surgery has the benefit of accurate pathologic staging to guide adjuvant therapy.⁶ Many patients undergoing primary surgery subsequently undergo adjuvant radiotherapy; however, a significant proportion may be able to forego chemotherapy according to the National Cancer Database (NCDB).⁷ Definitive chemoradiotherapy and adjuvant chemoradiotherapy for oropharyngeal cancer can have significant detriments to patient-reported quality of life, including salivary and taste dysfunction, long-term pain, financial stress, poor aesthetic outcomes, sexual difficulties, and risk of osteoradionecrosis.^8,9,10,11,12 They also may experience chemotherapy-related adverse effects, including nausea/vomiting, hearing loss, kidney damage, neuropathy, low blood cell counts, and rash, among others. Estimating the frequency of indications for adjuvant radiotherapy and/or adjuvant chemoradiotherapy after primary surgery may help optimize patient selection for surgery to minimize sequelae associated with triple-modality therapy, as well as determine when surgery alone would be sufficient.

According to the 2021 National Comprehensive Cancer Network guidelines, extranodal extension (ENE) and final positive surgical margins (PSM) are current indications for possible adjuvant chemoradiotherapy. Pathologic ENE is difficult to estimate clinically,^13,14 given that the sensitivity of radiographic ENE is imperfect and appears to be reliable only in large, bulky disease presentations.^15,16,17 Determining the individual likelihoods for pathologic ENE and PSM could help patients with HPV-positive OPSCC and their oncology team assess the utility and appropriateness of definitive surgery. In an analysis of patients with HPV-positive OPSCC in the NCDB who underwent surgery, Ochoa et al⁷ found that each higher level of nodal stage had a higher but still relatively low proportion of postsurgical pathologic ENE in patients without clinical ENE. Given the limitations of the NCDB with regard to potential inaccuracies, especially for assessment of PSM, and the absence of important patient-level data such as tobacco and alcohol use, we performed a similar analysis of patients with HPV-positive OPSCC at our institution with more details available. This study was designed to evaluate the risk of adverse histopathologic features that meet indications for possible adjuvant radiotherapy or possible adjuvant chemoradiotherapy in patients with HPV-positive OPSCC selected for primary surgery.

Methods

Study Design

This single-institution, retrospective cohort study included patients with HPV-positive OPSCC treated with primary TORS and neck dissection. Institutional review board approval was obtained from the University of California, San Francisco, Medical Center. Informed consent was not required. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Our primary objective was to determine the proportions of pathologic ENE and PSM per the 7th edition of the American Joint Committee on Cancer (AJCC-7) clinical N category and other possible clinical risk factors. Secondary objectives included determining the proportions of other histopathologic adverse features meeting indications for possible adjuvant radiotherapy and the clinical risk factors for these features.

Patient Selection

We queried a prospectively maintained database as well as our institution-wide cancer registry for all patients with HPV-positive OPSCC from March 1, 2012, to March 1, 2021. Inclusion criteria consisted of a new diagnosis of known or unknown primary HPV-positive OPSCC (positive for p16 and/or high-risk HPV) treated by a primary TORS with a multilevel comprehensive neck dissection. The decision for surgical management was aided by multidisciplinary tumor board review, generally selecting for patients with noninfiltrative disease, minimal soft palate involvement, mostly unilateral disease, and minimal suspicion for pathologic ENE. Exclusion criteria were incomplete definitive surgical resection, including TORS without multilevel neck dissection, palatine tonsillectomy or lingual tonsillectomy only without clearance of margins, neck lymph node open biopsy only, neck dissection without a complete primary tumor resection, initial treatment with nonsurgical modalities, and/or recurrent (or persistent) disease on presentation.

Of 616 patients with HPV-positive OPSCC who were screened, 480 (77.9%) were excluded for the following reasons: 324 (52.6%) underwent primary radiotherapy protocols, 92 (14.9%) underwent TORS resection for unknown primary site or debulking of primary tumor before radiotherapy without comprehensive neck dissection, 14 (2.3%) underwent TORS for recurrent postradiotherapy HPV-positive OPSCC disease, 8 (1.3%) were treated with palliative chemotherapy or immunotherapy, and 26 (4.2%) were lost to follow-up before treatment was initiated. Fifteen (2.4%) of the remaining excluded patients had undergone TORS and neck dissection for either obvious radiographic ENE, showing definite evidence of sternocleidomastoid and/or internal jugular vein invasion by nodal disease or gross clinical ENE with extension into skin for which the patients refused strong recommendations for definitive chemoradiotherapy or were medically unfit to tolerate definitive chemoradiotherapy and were therefore treated surgically. One additional patient (0.2%) had 3 separate noncontiguous primary sites after TORS biopsy findings with prior neck dissection, so definitive surgical management was not pursued. In total, 136 patients with HPV-positive OPSCC (22.1%) and without definite clinical ENE who underwent complete TORS and neck dissection were included in this study.

Outcomes and Independent Variables

We recorded the presence of postsurgical adverse histopathologic features based on current National Comprehensive Cancer Network guidelines.¹⁸ Possible indications for postoperative adjuvant chemoradiotherapy include pathologic ENE and/or PSM. Possible indications for postoperative adjuvant radiotherapy include pT3-pT4 disease, perineural invasion (PNI), lymphovascular invasion (LVI), lymph node greater than 3 cm, multiple positive lymph nodes, or involvement of neck levels IV/V. Degree of pathologic ENE was classified as microscopic (≤1 mm of extension) or macroscopic (>1 mm of extension). Close margins (<1 mm) were not considered as a possible adverse feature given the variability of philosophy on margins and paucity of evidence on the relevance or margins for HPV-positive OPSCC.¹⁹ We recorded patient age, sex, race and ethnicity, tobacco use, alcohol consumption, tumor site (tonsil or base of tongue), tumor size, neck dissection laterality, neck dissection lymph node yield, number and size of positive lymph nodes, primary clinical and pathologic T staging using AJCC-7, and clinical and pathologic N staging using AJCC-7 and AJCC-8. Race and ethnicity categorizations were identified via a review of medical records. We retrospectively staged patients without AJCC-7 or AJCC-8 staging initially. We used the Centers for Disease Control and Prevention definition of heavy alcohol consumption as more than 8 drinks per week for women and more than 15 drinks per week for men.²⁰ We analyzed smoking history by status (never, former, or current) as well as with binary cutoffs of 5, 10, 15, 20, and 25 pack-years. A binary threshold of 20 pack-years demonstrated the largest difference in rates of pathologic ENE and PSM and was therefore used for analyses.

Given the variability of past radiologic assessments without standardized reporting of clinical ENE during the retrospective study period, a neuroradiologist (E.G.) blinded to histopathologic results performed a second review of all available preoperative imaging (computed tomography, magnetic resonance imaging, and/or positron emission tomography with computed tomography) to evaluate for evidence of radiographic ENE. Criteria for features of radiographic ENE included ill-defined or irregular nodal margins, perinodal stranding, loss of fat planes, and a coalescent nodal mass. Patients with evidence of definite invasion of adjacent structures were excluded from the study as described above. Each patient was classified for radiographic ENE as none, equivocal, or suspected. Those without adequate preoperative imaging (7 of 136 [5.1%]) or with preceding excisional biopsy results were excluded from this subanalysis. Patients with no radiographic ENE were analyzed as a subgroup for rates of adverse features.

Statistical Analysis

Statistical analysis was performed from July 1, 2020, to March 15, 2021. We report presurgical characteristics and histopathologic outcomes for the study sample as frequencies (number and percentage) and medians (IQR or range). We first evaluated the frequencies of adverse histopathologic features among all 136 patients who were selected for and underwent primary surgery. We evaluated the presurgical characteristics for associations with the adverse features. We then performed a subgroup analysis that excluded patients with equivocal or suspected radiographic ENE. We also calculated rates of clinicopathologic concordance and the frequencies of primary tumor and nodal upstaging or downstaging using the AJCC-7 staging system. The effect sizes are depicted with percentage risk for each stratum as well as with odds ratios (ORs) with 95% CIs. Missing data are indicated and excluded from relevant analyses. All analyses were performed via STATA, version 16 (StataCorp LLC).

Results

Patient Characteristics and Histopathologic Outcomes

Table 1 shows the patient presurgical characteristics. One hundred thirteen patients were men (83.1%) and 23 were women (16.9%), with a median age of 63 (IQR, 55-70) years. Seven patients (5.1%) were Asian, 2 patients (1.5%) were Hispanic, 1 patient (0.7%) was Native Hawaiian or Other Pacific Islander, and 123 patients (90.4%) were White. Former or current smoking status included 72 patients (52.9%). Fifty-four patients (39.7%) had cT1 disease, and 93 (68.4%) had AJCC-7 cN1 or cN2b disease. Table 2 shows the histopathologic outcomes. At least 1 adverse histopathologic feature was found in most patients (109 of 136 [80.1%]). Thirty-four patients (25.0%) had either pathologic ENE or PSM, including 27 (19.9%) with pathologic ENE, 10 (7.3%) with PSM, and 3 (2.2%) with both. One hundred four patients (76.5%) had 2 or fewer adverse features or less. The most common adverse features of the 47 patients (34.5%) with a single adverse feature were a single positive node larger than 3 cm (21 [15.4%]), multiple positive nodes (15 [11.0%]), and LVI (7 [5.1%]). The most frequent combinations of the 30 patients (22.0%) with 2 adverse features included multiple positive nodes with a positive node larger than 3 cm (10 [7.4%]) and multiple nodes with pathologic ENE (7 [5.1%]).

Table 1. Presurgical Characteristics of Patients Selected for Primary Surgery.

Characteristic	Patient data (N = 136)a
Age, y
Median (IQR)	63 (55-70)
Range	36-87
Sex
Female	23 (16.9)
Male	113 (83.1)
Race and ethnicity
American Indian or Alaska Native	0
Asian	7 (5.1)
Black or African American	0
Hispanic	2 (1.5)
Native Hawaiian or Other Pacific Islander	1 (0.7)
Unknown	3 (2.2)
White	123 (90.4)
Smoking status
Current	8 (5.9)
Former	64 (47.1)
Never	64 (47.1)
Smoking pack-years (n = 68)
Median (IQR)	13.5 (4.8-30.0)
Range	0.25-75.00
Smoking history ≥20 pack-years
No	103 (75.7)
Yes	29 (21.3)
Unknown	4 (2.9)
Heavy alcohol consumptionb	20 (14.7)
ASA class
Class 1	7 (5.1)
Class 2	98 (72.1)
Class 3	31 (22.8)
Preoperative subsite
Base of tongue	38 (27.9)
Tonsil	52 (38.2)
Unknown primary	46 (33.8)
Clinical T category
cT0/Tx	46 (33.8)
cT1	54 (39.7)
cT2	36 (26.5)
Clinical N category (AJCC-7)
cN0	16 (11.8)
cN1	48 (35.3)
cN2a	23 (16.9)
cN2b	45 (33.1)
cN2c	4 (2.9)
Clinical N category (AJCC-8)
cN0	16 (11.8)
cN1	116 (85.3)
cN2	4 (2.9)
Radiographic ENEc
None	103 (79.8)
Equivocal	5 (3.9)
Suspected	21 (16.3)

Abbreviations: AJCC, American Joint Committee on Cancer; ASA, American Society of Anesthesiologists; ENE, extranodal extension.

^a Unless otherwise indicated, data are expressed as number (%) of patients. Percentages have been rounded and may not total 100.

^b Defined as more than 8 drinks per week for women and more than 15 drinks per week for men.

^c Includes 129 patients with adequate preoperative imaging studies available. Results were determined by blinded retrospective review.

Table 2. Histopathologic Outcomes Among Patients Selected for Primary Surgery.

Characteristic	Patient data (N = 136)a
Neck dissection laterality
Ipsilateral	106 (77.9)
Bilateral	30 (22.1)
No. of lymph nodes evaluated
Median (IQR)	33.5 (25.5-43.5)
Range	11-112
No. of positive lymph nodes
Median (IQR)	1.5 (1.0-2.0)
Range	0-12
Largest lymph node size, cm
Median (IQR)	3.4 (2.0-4.3)
Range	0.4-8.0
Tumor size, cm
Median (IQR)	2 (1.3-2.6)
Range	0.2-5.0
Pathologic T category
pT0/Tx	12 (8.8)
pT1	65 (47.8)
pT2	56 (41.2)
pT3	3 (2.2)
Pathologic N category (AJCC-7)
pN0	16 (11.8)
pN1	24 (17.6)
pN2a	26 (19.1)
pN2b	64 (47.1)
2 Nodes	34 (25.0)
3 Nodes	12 (8.8)
4 Nodes	11 (8.1)
≥5 nodes	7 (5.1)
pN2c	2 (1.5)
pN3	4 (2.9)
Pathologic N category (AJCC-8)
pN0	16 (11.8)
pN1	112 (82.4)
pN2	8 (5.9)
NCCN adverse histopathologic features
pT3 primary tumor	4 (2.9)
Perineural invasion	12 (8.8)
Lymphovascular invasion	34 (25.0)
Positive node >3 cm	63 (46.3)
Multiple positive nodes	67 (49.3)
Level IV/V neck involvement	4 (2.9)
Pathologic ENE	27 (19.9)
PSM	10 (7.4)
Degree of pathologic ENE
Macroscopic (≥1 mm)	25 (18.4)
Microscopic (<1 mm)	2 (1.5)
NCCN adjuvant therapy groupb
Any adverse feature	109 (80.1)
Nonpathologic ENE/PSM feature	75 (55.1)
Pathologic ENE and/or PSM	34 (25.0)
No. of adverse features
0	27 (19.9)
1	47 (34.5)
2	30 (22.1)
3	19 (14.0)
≥4	13 (9.5)

Abbreviations: AJCC, American Joint Committee on Cancer; ENE, extranodal extension; NCCN, National Comprehensive Cancer Network; PSM, positive surgical margins.

^a Unless otherwise indicated, data are expressed as number (%) of patients. Percentages have been rounded and may not total 100.

^b Based on 2021 NCCN guidelines. Recommendations for adjuvant radiotherapy include: pT3-pT4, perineural invasion, lymphovascular invasion, positive node larger than 3 cm, multiple positive nodes, and/or level IV/V neck involvement. Recommendations for adjuvant chemoradiotherapy include pathologic ENE and/or PSM.

Risk of Adverse Histopathologic Features

The proportions of adverse histopathologic features varied by AJCC-7 clinical N category (Table 3). Pathologic ENE was present in 0 of 16 patients with cN0, 8 of 48 patients (16.7%) with cN1, 3 of 23 patients (13.0%) with cN2a, 16 of 45 patients (35.5%) with cN2b, and 0 of 4 patients with cN2c. Compared with cN1-cN2a disease, patients with cN2b disease had higher odds of pathologic ENE (OR, 3.01; 95% CI, 1.14-8.10). We were unable to demonstrate an association of cN2b vs cN1-cN2a with PSM (OR, 0.94; 95% CI, 0.14-5.14), LVI (OR, 1.29; 95% CI, 0.50-3.25), or PNI (OR, 2.09; 95% CI, 0.42-11.1). Table 4 demonstrates that the following presurgical factors were not associated with either pathologic ENE or PSM: age 65 years or older (ORs, 1.14 [95% CI, 0.44-2.89] and 0.92 [95% CI, 0.18-4.10], respectively), smoking history of at least 20 pack-years (ORs, 2.28 [95% CI, 0.77-6.35] and 2.59 [95% CI, 0.49-11.79], respectively), heavy alcohol consumption (ORs, 1.42 [95% CI, 0.36-4.70] and 1.50 [95% CI, 0.14-8.39], respectively), clinically known vs unknown primary tumors (ORs, 0.63 [95% CI, 0.21-1.73] and 0.20 [95% CI, 0.004-1.54], respectively), base of tongue vs tonsil primary site (ORs, 1.96 [95% CI, 0.62-6.94] and 2.80 [95% CI, 0.49-28.94], respectively), or cT2 vs cT0-cT1 disease (ORs, 2.31 [95% CI, 0.85-6.08] and 1.21 [95% CI, 0.19-5.66], respectively). Likewise, these same factors were not associated with either LVI or PNI.

Table 3. Likelihood of Adverse Histopathologic Features by AJCC-7 Clinical Nodal Stage Among Patients Selected for Primary Surgery.

Clinical stage	Total No. of patients	No. (%) of patients
		Individual adverse histopathologic events							NCCN adjuvant therapy group
		LVI	PNI	Any pathologic ENE	Macroscopic pathologic ENE	Microscopic pathologic ENE	PSM	Any adverse featurea		Nonpathologic ENE/PSM	Pathologic ENE and/or PSM
cN0	16	3 (18.8)	3 (18.8)	0	0	0	2 (12.5)	7 (43.7)		5 (31.3)	2 (12.5)
cN1	48	13 (27.1)	3 (6.3)	8 (16.7)	8 (16.7)	0	5 (10.4)	36 (75.0)		25 (52.1)	11 (22.9)
cN2a	23	4 (17.4)	1 (4.3)	3 (13.0)	3 (13.0)	0	0	21 (91.3)		18 (78.3)	3 (13.0)
cN2b	45	13 (28.9)	5 (11.1)	16 (35.5)	14 (31.1)	2 (4.4)	3 (6.7)	43 (95.5)		25 (55.5)	18 (40.0)
cN2c	4	1 (25.0)	0	0	0	0	0	2 (50.0)		2 (50.0)	0
All	136	34 (25.0)	12 (8.8)	27 (19.9)	25 (18.4)	2 (1.5)	10 (7.3)	109 (80.1)		75 (55.1)	34 (25.0)

Abbreviations: AJCC, American Joint Committee on Cancer; ENE, extranodal extension; LVI, lymphovascular invasion; NCCN, National Comprehensive Cancer Network; PNI, perineural invasion; PSM, positive surgical margins.

^a Includes pT3-pT4, PNI, LVI, positive node larger than 3 cm, multiple positive nodes, level IV/V neck involvement, pathologic ENE, and PSM.

Table 4. Likelihood of Pathologic ENE or PSM by Presurgical Characteristics Among Patients Selected for Primary Surgery.

Characteristic	No. of patients	Pathologic ENE		PSM
		No. (%)	OR (95% CI)	No. (%)	OR (95% CI)
Age group, y
<65	79	15 (19.0)	1 [Reference]	6 (7.6)	1 [Reference]
≥65	75	12 (16.0)	1.14 (0.44-2.89)	4 (5.3)	0.92 (0.18-4.10)
Smoking history, pack-years
0-20	103	17 (16.5)	1 [Reference]	6 (5.8)	1 [Reference]
≥20	29	9 (31.0)	2.28 (0.77-6.35)	4 (13.8)	2.59 (0.49-11.79)
Heavy alcohol consumptiona
No	116	22 (19.0)	1 [Reference]	8 (6.9)	1 [Reference]
Yes	20	5 (25.0)	1.42 (0.36-4.70)	2 (10.0)	1.50 (0.14-8.39)
Known vs unknown subsite
Known	90	20 (22.2)	1 [Reference]	9 (10.0)	1 [Reference]
Unknown	46	7 (15.2)	0.63 (0.21-1.73)	1 (2.2)	0.20 (0.004-1.54)
Clinical subsite
Base of tongue	38	6 (15.8)	1 [Reference]	2 (5.3)	1 [Reference]
Tonsil	52	14 (26.9)	1.96 (0.62-6.94)	7 (13.5)	2.80 (0.49-28.94)
Clinical T category
cT0-cT1	100	16 (16.0)	1 [Reference]	7 (7.0)	1 [Reference]
cT2	36	11 (30.5)	2.31 (0.85-6.08)	3 (8.3)	1.21 (0.19-5.66)

Abbreviations: ENE, extranodal extension; OR, odds ratio; PSM, positive surgical margins.

^a Defined as more than 8 drinks per week for women and more than 15 drinks per week for men.

The accuracy of radiographic ENE in relation to pathologic ENE, including equivocal or suspected, was as follows: sensitivity, 45.8% (11 of 24); specificity, 85.7% (90 of 105); positive predictive value, 42.3% (11 of 26); and negative predictive value, 87.4% (90 of 103). In a subgroup analysis that excluded patients with suspected or equivocal radiographic ENE, pathologic ENE was found in only 13 of 103 (12.6%) vs 27 of 136 (19.9%) in the entire cohort. The proportions of pathologic ENE in this subgroup were 0 of 16 for cN0, 5 of 40 (12.5%) for cN1, 0 of 13 for cN2a, 8 of 32 (25.0%) for cN2b, and 0 of 2 for cN2c.

Clinicopathologic Concordance

Clinicopathologic concordance according to AJCC-7 occurred in 70 patients (51.5%) based on T category and 77 (56.6%) based on N category (Table 5 and eTable in the Supplement). Sixty-one patients (44.9%) underwent primary tumor upstaging. Most upstaging occurred when cT0/Tx was found to be to pT1 or when cT1 was found to be pT2. Of the 136 patients with cT0-cT2 disease, only 3 (2.2%) were upstaged to pT3 (all owing to size >4 cm), thus meeting an indication for possible adjuvant therapy. Approximately one-third of patients experienced nodal upstaging. One of 16 patients (6.3%) with cN0 and 30 of 48 (62.5%) with cN1 was upstaged to pN2a or greater, meeting an indication for possible adjuvant therapy. Six of 23 patients (26.1%) with cN2a were upstaged to pN2b or greater. When patients with cN0, cN1, and cN2a were upstaged to pN2b, only 2 nodes were identified in 1 of 1 (100%) with cN0, 17 of 23 (73.9%) with cN1, and 2 of 5 (40.0%) with cN2a. Downstaging to pN0-pN1 occurred in 3 of 23 patients (13.0%) with cN2a, 3 of 45 patients (6.7%) with cN2b, and 1 of 4 patients (25.5%) with cN2c. For each AJCC-7 clinical N category, a few patients had 3 or more nodes, including 0 of 16 with cN0, 6 of 48 (12.5%) with cN1, 3 of 23 (13.0%) with cN2a, 22 of 45 (48.9%) with cN2b, and 1 of 4 (25.0%) with cN2c. Even fewer patients had 5 or more nodes, placing them in the AJCC-8 N2 category, including 0 of 16 with cN0, 1 of 48 (2.1%) with cN1, 0 of 23 with cN2a, 7 of 45 (15.5%) with cN2b, and 0 of 4 with cN2c.

Table 5. Clinicopathologic Nodal Staging Concordance Among Patients Selected for Primary Surgery.

N category	Total	Pathologic stage, No. (%)						Staging concordance	Upstaged	Downstaged
		pN0	pN1	pN2a	pN2b	pN2c	pN3
cN0	16	13 (81.3)	2 (12.5)	0	1 (6.3)	0	0	13 (81.3)	3 (18.7)	0
cN1	48	2 (4.2)	16 (33.3)	6 (12.5)	23 (47.9)	0	1 (2.1)	16 (33.3)	30 (62.5)	2 (4.2)
cN2a	23	0	3 (13.0)	14 (60.9)	5 (21.7)	0	1 (4.3)	14 (60.9)	6 (26.1)	3 (13.0)
cN2b	45	1 (2.2)	2 (4.4)	6 (13.3)	33 (73.3)	1 (2.2)	2 (4.4)	33 (73.3)	3 (6.7)	9 (20.0)
cN2c	4	0	1 (25.0)	0	2 (50.0)	1 (25.0)	0	1 (25.0)	0 (0)	3 (75.0)
Total	136	16 (11.8)	24 (17.6)	26 (19.1)	64 (47.1)	2 (1.5)	4 (2.9)	77 (56.6)	42 (30.9)	17 (12.5)

Discussion

We conducted this study to improve identification of individual patient risk for adjuvant radiotherapy and adjuvant chemoradiotherapy given patient preferences for less intensive adjuvant therapy.²¹ This study demonstrates that proper selection of patients with AJCC-7 cT0-cT2 cN0-cN2b disease without clinically obvious ENE may result in pathologic ENE rates of approximately 20% or lower, depending on adherence to radiologic features suggestive of possible radiographic ENE. We believe our results can serve as clinically useful data for selecting and counseling patients with HPV-positive OPSCC about their individual likelihoods of adverse histopathologic features and possible indications for adjuvant therapy should they undergo primary surgery.

Our results are consistent with those of NCDB patients with HPV-positive OPSCC who underwent surgery in that more advanced clinical N categories are associated with a generalized increase likelihood of pathologic ENE.^7,22 The highest risk group in our study were those with cN2b disease, who experienced pathologic ENE approximately one-third of the time in this selected surgical group. Positive surgical margins in this study were uncommon, occurring in 7.4%, and similar to a systematic review and meta-analysis²³ that found a PSM rate of 7.8% after TORS for all patients with OPSCC.

Estimating pathologic ENE based on preoperative imaging has been shown to have limitations by multiple studies.^{12,14,15,16,24,25,26} A recent study from The Johns Hopkins University demonstrated that “absence of perinodal fat plane” was most sensitive, and “irregular nodal margin status” was most specific for pathologic ENE in HPV-positive OPSCC.^17(p1) Through a blinded second review of preoperative imaging, our results suggest it may be possible to use an algorithm including these criteria, among others, to further improve patient selection for primary surgery; however, the positive predictive value of radiographic ENE in our highly selective patient cohort remained low. Thus, given the positive predictive value limitations, a stricter reliance on radiologic features could exclude many patients from surgery who do not actually have pathologic ENE.

Approximately one-half of patients in this study experienced clinicopathologic stage concordance, highlighting the imperfections of clinical staging and the role of primary surgery for accurate staging.⁶ Previous studies have reported similar clinicopathologic discordances.^7,27 Although stage concordance is much higher using AJCC-8,²⁸ this edition is less helpful for guiding treatment. Because the likelihood of pathologic ENE varied by AJCC-7 nodal category and was particularly more likely with cN2b, our results highlight the importance of considering the AJCC-7 staging system for treatment choice and risk tolerance for adjuvant treatment.

Smoking, alcohol consumption, age, and T category did not appear to be associated with risk of pathologic ENE, PNI, LVI, or PSM in our study; however, further study with larger samples may find more subtle associations. As has been seen in past studies⁷ as well as ours, LVI and especially PNI are uncommon in patients with HPV-positive OPSCC who undergo surgery. Our results also show that the presence of LVI and PNI are not associated with the clinical N category.

These findings must be considered in the context of evolving HPV-positive OPSCC guidelines. Extranodal extension and PSM appear to be associated with outcomes in HPV-negative OPSCC; however, some data suggest that this may not be the case with patients with HPV-positive disease.²⁹ Although previous studies using the NCDB have correlated pathologic ENE with overall survival,^30,31,32 a growing body of literature suggests that definitive and adjuvant chemoradiotherapy may not confer a significant clinical benefit over radiotherapy alone for patients with HPV-positive OPSCC and pathologic ENE.^30,33,34,35 Future studies are warranted to further explore the prognosis and role of adjuvant therapy for patients with HPV-positive OPSCC with adverse histopathologic features. Also, in certain situations in which adjuvant radiotherapy would be used without chemotherapy, such as AJCC-7 cT1-cT2 and N0-N2a, PSM as an isolated adverse feature may not justify the addition of adjuvant chemotherapy to adjuvant radiotherapy after surgery. Furthermore, the finding of a single adverse feature after surgery such as a single pathologic lymph node larger than 3 cm (pN2a) or even 2 pathologic lymph nodes in isolation may not be ultimate indications for adjuvant radiotherapy.³⁶

Overall, TORS is likely to play an increasing role in HPV-positive OPSCC management, because studies have revealed excellent oncologic outcomes.^4,37 Several reports have demonstrated many patients are able to avoid any adjuvant therapy.^38,39 Ryan et al³⁶ recently published a 12-institution analysis of 344 patients with HPV-positive OPSCC undergoing surgery alone that showed a 2-year disease-free survival of 91%, disease-specific survival of 98%, and a high salvage rate. Data are currently limited, but randomized clinical trials comparing primary surgery with definitive radiotherapy are under way.^40,41,42,43 With proper selection, primary surgery can lead to satisfactory oncologic outcomes and reduce impairments in quality of life that are associated with radiotherapy and/or chemotherapy.^3,12,21,44 Possible contraindications to definitive surgery to consider include clinical ENE, some cT3-cT4 disease, deep tumors, bilateral base of tongue tumors (especially in the presence of deep parapharyngeal space extension), significant soft palate extension, and possibly bilateral or contralateral neck metastatic disease.

Limitations

We acknowledge that there are several limitations to this study. As a single-institution study, our sample size and event rates prevent us from making definitive conclusions because the 95% CIs around the effect sizes are too wide, particularly regarding potential risk factors such as tobacco use. Our results also reflect only the experience and preferences at our institution. Selection for surgery was determined by a combination of our multidisciplinary tumor board consensus, patient-surgeon discussions, and patient preference, leading to decisions and outcomes that may not be generalizable to other institutions. Several factors could have affected radiographic ENE assessment, including heterogeneous imaging modalities and parameters in this retrospective cohort, variable time from imaging to surgery, interventions such as fine-needle aspiration biopsy, the lack of core needle biopsy specimens before imaging, radiologist experience, and lack of consensus imaging interpretation. In addition, because this is a retrospective study of patients during a 9-year period, the histopathologic analyses were performed by multiple pathologists, generating the possibility of interrater discrepancies on the interpretation and grading of pathologic ENE. Taken together, larger-scale prospective studies are warranted to confirm these results.

Conclusions

With careful selection of patients with HPV-positive OPSCC for TORS and neck dissection who have AJCC-7 cT0-cT2 cN0-N2b disease, no obvious clinical ENE, and noninfiltrative primary site tumors, approximately one-quarter of patients will have pathologic ENE and/or PSM, which are indications for adjuvant chemoradiotherapy according to 2021 National Comprehensive Cancer Network guidelines. However, approximately one-third of patients with cN2b disease will have pathologic ENE after surgery. Presurgical factors such as age, smoking, alcohol consumption, and clinical T category do not appear to be associated with the presence of pathologic ENE and/or PSM. Most patients with HPV-positive OPSCC will meet at least 1 indication for possible adjuvant radiotherapy on histopathologic analysis, usually owing to the presence of a lymph node measuring more than 3 cm and/or multiple pathologic lymph nodes. These data may assist preoperative patient consultations regarding likelihoods of adverse histopathologic features and possible indications for adjuvant therapy after primary surgery.

Supplement.

eTable. Clinicopathologic Primary Staging Concordance Among Patients Selected for Primary Surgery