In this series of OPSCC patients undergoing TORS, the rate of margin positivity was 12 % and the rate of nodal ECE was 44 %. No clinical features predicted positive margins in this single institution study of predominantly T1 and T2 cancers, but nodal size emerged as a significant predictor of ECE. No other clinical features were predictive of ECE. These results appear consistent with those reported from a study of the National Cancer Database (NCDB) of patients receiving TORS for oropharyngeal cancer. In that series, the rate of positive margins was 20.2 % in comparison to 12 % in the current study, suggesting our institutional results compare favorably with national practice patterns. The rate of ECE in the NCDB study was 28.5 %, although in 17 % of cases the value was unknown making it difficult to compare it to the 44 % who had ECE in the current study. Also, only nodal stage, not largest nodal size was reported in that study, again making direct comparison difficult [1].

It is a key finding of our study that the previously described relationship between nodal size and ECE [16–20] holds even in a largely HPV-associated oropharyngeal carcinoma population. HPV-associated SCC often presents with thin-walled cystic lymph nodes [21]. It is hypothesized that these cystic lymph nodes carry fewer viable tumor cells than a solid lymph node of the same size, raising the question of whether lymph node size would correlate with the risk of ECE in HPV-associated oropharynx cancer, as it did in reports from two decades ago [16–20], when HPV was much less prevalent and multiple tumor sites were studied. The current study, however, demonstrates that lymph node size remains a strong predictor of ECE even in a largely HPV-associated population. In the current study, lymph nodes 3.0 cm and larger were associated with a rate of ECE of 92 %, which was statistically significantly higher than the 44 % rate of ECE seen in smaller lymph nodes.

In addition to lymph node size, another recently investigated characteristic has been the correlation between radiographic ECE on computed tomography (CT) imaging and pathologic ECE. In our study, the use of radiographic ECE to predict pathologic ECE was associated with a sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 53, 93, 89, and 65 %, respectively. These findings are fairly consistent with a recent study of 432 patients with oral cavity or laryngeal cancer, which demonstrated a sensitivity, specificity, PPV, and NPV of 43.7, 97.7, 82.6, and 87.3 %, respectively, for the use of radiographic ECE to predict pathologic ECE, though our lower NPV may be at least partially attributable to the higher prevalence of pathologic ECE in our study (44 % vs. 20 %). The authors concluded that these values are reasonable for use in clinical decision making [22]. In another study of 100 patients with head and neck cancer, the sensitivity, specificity, PPV, and NPV by two independent radiologists were (1) 49, 84, 84, and 49 %, respectively, and (2) 65, 54, 71, and 48 %, respectively [23]. The prevalence of pathologic ECE in this study was 63 %. The authors of this study concluded that radiographic ECE is not an adequate predictor of pathologic ECE due in part to a poor NPV. The criteria used to identify radiographic ECE in our study and these two studies were irregular borders of the lymph node capsule as well as perinodal fat infiltration. Although neither of these large studies was limited to oropharyngeal cancer patients and there was a variance in the prevalence of pathologic ECE between studies, the differing conclusions demonstrate the contentious nature surrounding the use of radiographic ECE as a predictor for pathologic ECE.

It should also be noted that debate exists about how the extent of ECE affects prognosis. A study of mostly p16-positive oropharyngeal carcinoma demonstrated that while the risk of recurrence was significantly higher with grade 4 extracapsular extension (defined as no residual nodal tissue or architecture, “soft tissue metastases”), this did not appear to affect disease-free or overall survival [15]. An unrelated study of larynx and hypopharynx carcinoma demonstrated that macroscopic extracapsular extension, but not microscopic extension, was associated with risk of recurrence and death [24]. Another study of 35 patients with head and neck squamous cell carcinoma of unknown primary demonstrated lower overall and cancer-specific survival in patients with ≥ 2 mm of ECE [25]. However, a separate study of 266 patients with oral tongue cancer found no difference in survival when comparing patients with ≤ 2 mm vs. > 2 mm of ECE [26]. Current clinical trials use 1 mm of ECE as a cutoff for analysis [27], and further study is needed in this area.

The findings of the current study, if confirmed in larger series, can be used to help guide patient selection for TORS, among patients with T1 and T2 tumors. TORS is a relatively new technique for operative management of oropharyngeal tumors with less morbidity and improved functional outcomes compared to open surgical approaches for similar tumors [28–31]. Nearly 75 % of newly diagnosed oropharyngeal cancers are now associated with HPV and therefore have a more favorable prognosis, leading to current research studies investigating the benefit of de-escalation of therapy for these patients [10, 11]. Clinical trials currently underway are investigating the effectiveness of reduced-dose adjuvant radiation therapy [27] as well as definitive radiation therapy [32]. In addition, the association between ECE and prognosis in patients with p16-positive oropharyngeal cancer has lately been called into question. Recent retrospective data suggest that ECE is not associated with disease recurrence or survival in these patients [33], and this in turn has cast doubt on the necessity of adjuvant chemoradiation. The question of whether adjuvant chemoradiation is more beneficial than adjuvant radiotherapy alone in patients with ECE in the setting of HPV-related disease is currently being investigated in phase III clinical trials [34, 35]. However, the results from these studies will not be available for some time. Unfortunately, the patients in the present study were treated prior to de-escalation trials for HPV-associated OPSCC being open at our institution, and so were treated using a standard of care approach.

Limitations of this study include its retrospective design and the small sample size. Another consideration is that the patients in this study underwent TORS during the first four years of robotic surgery at our institution, and greater cumulative case volume may lead to improved outcomes. For instance, national data has shown a lower rate of positive margins in high-volume institutions (defined as having more than 10 TORS cases per year) compared to low-volume institutions [1], suggesting the existence of a “learning curve.” That said, the rate of positive margins in our study was favorable to national rates, suggesting that experience was not the critical determinant of margin positivity.