The purpose of this study was to assess the efficacy, tolerance and impact on quality of life of two therapeutic strategies in the treatment of localized and locally advanced HPV-OPSCCs: upfront surgery followed or not by adjuvant RT ± CT and exclusive RT ± CT.
Both study groups were quite homogeneous but differed from each other, as patients from the uS group had a higher performance status and those from the eRT ± CT group presented a higher tumor stage. However, the proportion of stage III OPSCC was the same in both groups. Tumors that developed within the palatin tonsils or the glosso-tonsillar sulcus were more likely to be treated surgically, even though this was not significant. These elements should be taken into account, as they are likely to affect the efficacy outcomes of both treatment groups even if they did not seem to impact PFS in the predictive factor analysis.
We found no significant difference in PFS, OS, DSS, LRPFS and MPFS between uS and eRT ± CT in the whole population, neither in early stages nor in more advanced diseases. Both strategies appeared to be effective in the treatment of stage I-III HPV-OPSCC.
Our results seem consistent with the sparse data available in the literature. Kelly et al. found no difference in OS between a surgical approach and eRCT in a retrospective study of over 3,000 patients with stage I-II HPV-OPSCC (13), as did Kim et al. more recently (19). In more advanced tumors - stage III-IV according to the AJCC 7th version - Kamran et al. reported an OS benefit from surgery over RT-CT (20). This retrospective study did not focus only on HPV-OPSCC, although clinical characteristics were not available for the 5,037 patients with an HPV-related tumor. Additionally, OS was not significantly different between the treatment groups in the multivariate analysis, suggesting that confounding factors might have explained the difference in OS in the univariate analysis. These results came from American databases, and although they allow us to obtain data on large populations, they only provide information on overall survival, and data about other relevant oncological outcomes are missing.
Recently, the GETTEC (Groupe d’Étude des Tumeurs de la Tête Et du Cou) group conducted a study on 382 patients with HPV-OPSCC in 7 French centers (15). There was no difference in OS at 5 years (89.2% and 84.2% in the surgical and nonsurgical groups, respectively), but DSS and RFS were found to be improved in the surgical group in the multivariate analysis. In our study, we found similar survival rates and patterns of relapse to those published by Culié et al. (19% of relapse in our study vs 20%) (15). However, we did not show differences of any kind in terms of efficacy between our 2 groups. Although this might be caused by a low number of events and the presence of confounding factors, it could also be related to good outcomes in our eRT ± CT group. Indeed, Lacau St Guly et al. found no difference in RFS either at 2 years after uS or eRT ± CT in 92 patients with HPV-OPSCC (21).
A meta-analysis attempted to evaluate both strategies in the treatment of OPSCC, irrespective of HPV status (22). Authors showed a significant heterogeneity between studies, and concluded that OS after surgical and nonsurgical treatments was not significantly different. However, this meta-analysis mostly included unicentric single-modality studies with a low number of patients, highlighting the importance of data from well-conducted large retrospective and prospective studies.
Because patients with HPV-OPSCCs have better outcomes than those with non-HPV-related tumors (5, 23), survival is longer, and treatment tolerance and QoL have become an increasing issue (24). Notably, all patients from our study received IMRT, as it was proven to reduce the incidence of xerostomia and improve saliva-related QoL compared to 3D RT (25, 26).
In our study, overall acute and late toxicities did not significantly differ between the treatment groups, and tolerance remained acceptable in both groups, as no grade IV-V side effects were reported.
Interestingly, the phase II ORATOR study, which compared QoL after transoral robotic surgery (TORS) with or without adjuvant treatments to eRT ± CT in the treatment of T1-2 N0-2 OPSCCs (10), found a trend for more grade III dysphagia in the surgery group (26% vs 18%), as in our study, even though the surgical technique was mini-invasive and only 74% of patients received adjuvant RT in the surgical group (10). However, the ORATOR 2 phase II trial comparing TORS ± 50–60 Gy RT and 60 Gy RT plus weekly cisplatin in Ti-2 N0-2 tumors was prematurely discontinued because of excessive toxic effects in the surgical arm (27). Long-term outcomes are awaited with this surgical approach, but caution is required outside expert centers.
Since it is well established that QoL and toxicity assessment by physicians leads to lower incidence and severity of symptoms compared to patients’ reports (28), we decided to prospectively collect patient-reported outcomes in addition to data retrospectively extracted from medical files. We used the multi-item multidomain EORTC QLQ-C30 and QLQ-H&N35 instruments, as they are considered reliable and valid measurement tools of QoL and are used in many international clinical trials (29–32). We observed a trend for a better global QoL (77.6 vs 65.2, p = 0.07) and for a higher role functioning score (91.4 vs 80.3, p = 0.06) according to the QLQ-C30 questionnaire after eRT ± CT. Patients in the nonsurgical group also reported more swallowing issues and sticky saliva, whereas fatigue and appetite loss scores were higher in the uS group. These results are relevant since the global health status from the QLQ-C30 instrument as well as the scores related to the loss of appetite and fatigue are known to be predictive of OS in many cancer types (33–36), including head and neck cancers (37).
OPSCC survivors face clinically important deteriorations in QoL mostly centered on xerostomia, dysphagia and chewing (24, 38). It has been prospectively proven that patients with an HPV-positive tumor have their own course of QoL during and after treatment (39), but studies comparing surgical and nonsurgical strategies are scarce in this particular population. Similar to our findings, Yin et al. showed a worse QoL after surgery alone or upfront surgery followed by RT compared to eRT at 3 and 6 months after treatment (40). Unfortunately, no long-term data were available. In parallel, longitudinal swallowing QoL according to the MD Anderson Dysphagia Inventory (MADI) score was significantly better with RT in the ORATOR trial, contrary to our results (10, 41). However, this difference did not meet the predefined threshold of a clinically relevant change in QoL and was no longer significant at 2 and 3 years.
In our study, no patient underwent surgery without any adjuvant treatment, possibly due to a selection bias. Thus, our findings cannot be extrapolated to pT1-2 pN0-N1 tumors with negative margins and no criteria for adjuvant treatments. Nevertheless, 97% of the patients in the GETTEC study (15) and two-thirds of the patients in the ORATOR trial (10) received adjuvant therapies, suggesting that only a small number of patients are treated by surgery alone. This implies that patients treated by surgery are likely to need a multimodal approach that may increase the risk of toxicity and deteriorate QoL. Patients must be carefully informed of the oncological outcomes of uS and eRT ± CT and of the high rate of adjuvant treatments after surgery. Again, selecting the population that might benefit from surgery alone as well as finding effective treatment de-escalation are consequently paramount.
Our work has several limitations. First, oncological data and physician-reported toxicities were retrospectively obtained, leading to inherent flaws in retrospective studies. Moreover, because of a lack of power due to very few events, only a univariate analysis could be performed. Therefore, our results might have been impacted by some confounding factors. However, our study population was quite homogeneous between the treatment groups and did not differ in terms of alcohol consumption and stage III disease, the only two variables that were found to impact PFS, suggesting that the effects of potential confounding factors are limited. Another weakness is related to the QoL assessment. Although data were prospectively collected, each patient completed the questionnaires at a unique time point after treatment. Consequently, this QoL survey should be examined carefully.
This work also has strengths. It is indeed, to the best of our knowledge, one of the largest European studies in this field of research and one of the rare to focus not only on efficacy but also tolerance and QoL. We present here a multicentric study, thus reducing the center effect on our outcomes. Additionally, data were collected in three academic centers with expertise in the management of OPSCC and that follow international guidelines, leading to meaningful results.
Therefore, our findings contribute to new insights into the management of HPV-OPSCC in the era of therapeutic de-escalation and might help physicians find the most suitable treatment strategy for each individual. Phase III trials comparing surgery and radiotherapy might never exist because of recruitment issues. However, large, international, prospective studies could clarify the debate between surgery and radiotherapy and even determine predictive factors for one strategy or the other.