Postoperative radiotherapy may not be necessary for locally advanced head and neck squamous cell carcinoma: a case-match multicentre study

Head and neck cancer is one of the most common malignancies, and squamous cell carcinoma (SCC) accounts for approximately 90% of all head and neck cancers [1]. Head and neck squamous cell carcinoma (HNSCC) arises in the oral cavity, oropharynx, larynx or hypopharynx and is the sixth leading cancer by incidence worldwide, with more than 600,000 cases diagnosed annually. Only 40%–50% of patients with HNSCC survive for 5 years [2]. Surgery, radiotherapy and chemotherapy are the mainstays of primary treatment in patients with HNSCC [3, 4]. However, despite the progress of multiple treatments and multidisciplinary treatments, the effect of treatment on HNSCC is not good, especially for locally advanced HNSCC (LA-HNSCC) [5, 6]. LA-HNSCC is cancer that has grown outside the origin organ but has not yet spread to distant parts of the body (Stage III/IV disease, pT3-4N0 or pT1-4N+). Most patients present with locoregionally advanced disease, and more than 50% have recurrence within 3 years [7]. At the same time, comprehensive treatment brings more complications and social and economic burden to patients. However, many experienced head and neck cancer surgeons have found that not all patients with LA-HNSCC, especially those with oral squamous cell carcinoma (OSCC) need adjuvant radiotherapy to achieve an ideal prognosis [8, 9]. Our previous study [10] (including some unpublished data) found that patients with locally advanced OSCC did not necessarily need postoperative radiotherapy (PORT) to achieve a good prognosis under the premise of high-quality surgical resection of the tumour.

Based on the above results and the clinical data of our unit, we propose two questions: (1) Is the benefit of PORT for head and neck squamous cell carcinoma as important as we expected and (2) Should the indication of PORT be further discussed? To answer these questions, we performed a case-match cohort analysis on patients with HNSCC at two institutions.

A total of 432 patients were eligible for this study. After performing the case-match-designed matching, only 57 pairs (114 patients) were included. The quality of the matching was excellent, as summarised in Table 1. P values for chi-square tests of homogeneity were greater than 0.4 for all matches and for all variables except groups. 57 matched pairs of patients (110 men and 4 women) with a mean age of 51 ± 9.8 years were identified. The median follow-up period was 40.2 months (range 3–101 months). All patients had LA-HNSCC. Of the 57 patients who received adjuvant treatment, surgery with radiotherapy were 44 patients, other 13 had surgery with chemoradiotherapy. In statistical analysis, some included patients turned from cT3-4N0 to pT1-2N0. In order to be more accurate, we eliminated pT1-2N0 patients. The characteristics were summarised in Table 2. 37 pairs of patients (71 men and 3 women) with a mean age of 48 ± 9.9 years were remained. Of the 37 patients who received adjuvant treatment, surgery with radiotherapy were 28 patients, other 9 had surgery with chemoradiotherapy. The study selection criteria, including the relevant reasons for exclusion, are illustrated in Fig. 2.

The 3-year estimated OS rate was 57.9% for those treated with surgery alone and 54.4% for those receiving surgery plus PORT. There was no significant difference in OS between the two arms (HR 0.88; 95% CI 0.50–1.58; P = 0.79; Fig. 3A). Similarly, there was no significant difference in DFS between the two groups (HR 0.86; 95% CI 0.50–1.50; P = 0.76; Fig. 3B). The 3-year DFS was 52.6% and 49.1% in the surgery-alone group and surgery plus PORT, respectively. Locoregional recurrence occurred in 24 cases in the surgery-alone group and 26 in surgery plus PORT group. There was no statistically significant difference between the two groups. The surgery-alone group had three patients with distant metastasis, and all three patients had pulmonary metastasis, one of whom had multiple metastasis, including adrenal gland and pelvis. The surgery plus PORT group had three distant metastasis: two to the lungs and one to the ribs. After the elimination, the result was the same. No differences in OS (HR 0.90; 95% CI 0.48–1.68; Fig. 4A) or DFS (HR 0.88; 95% CI 0.47–1.65; Fig. 4B) were observed.

Uni- and multi-variable analyses are shown in Table 3. In univariate analysis, only subsite and lymph node metastasis were associated with an increased risk of OS (RR = 1.331; 95% CI 1.063–1.667, 1.638; 95% CI 1.164–2.305, respectively), DFS (RR = 1.345; 95% CI 1.084–1.669, 1.726; 95% CI 1.241–2.400, respectively). Adjuvant chemotherapy resulted in a significant improvement in DFS (RR 0.483; 95% CI 0.248–0.943). In multivariable analysis, after control for age, smoking and alcohol status, no difference in OS (RR 1.184; 95% CI 0.649–2.162) or DFS (RR 1.169; 95%CI 0.659–2.073) was observed with the addition of PORT. Subsite was associated with an increased risk of OS (RR 1.457; 95% CI 1.122–1.890), and DFS (RR 1.423; 95% CI 1.104–1.834). Lymph node metastasis was associated with an increased risk of OS (RR 1.628; 95% CI 1.143–2.317), and DFS (RR 1.725; 95% CI 1.227–2.424). After the elimination, Subsite and lymph node metastasis were associated with an increased risk of OS and DFS in the remaining 37 pairs.

Based on the case-match cohort analysis of 57 pairs, 37 patients relapsed within 1 year after surgery, including 19 patients in the surgery-alone group and 18 patients in the surgery plus PORT group. A total of 46 patients (including the 37 patients who relapsed in 1 year) relapsed 2 years after surgery (22 in the surgery-alone group and 24 in the surgery plus PORT group). A total of 57 patients (27 in the surgery-alone group and 30 in the surgery plus PORT group) relapsed within 3 years after surgery. The mean recurrence time of patients was 33.7 months, the median recurrence time was 30 months, and the recurrence cases within 1 year accounted for 32.5% (37/114) of all the recurrence cases. After eliminating pT1-2N0 patients including 37 pairs, 27 patients relapsed within 1 year after surgery, including 14 patients in the surgery-alone group and 13 patients in the surgery plus PORT group. 33 patients (including the 27 patients who relapsed in 1 year) relapsed 2 years after surgery (16 in the surgery-alone group and 17 in the surgery plus PORT group). A total of 44 patients (21 in the surgery-alone group and 23 in the surgery plus PORT group) relapsed within 3 years after surgery. The mean recurrence time of patients was 33.2 months, the median recurrence time was 29 months, and the recurrence cases within 1 year accounted for 36.5% (27/74) of all the recurrence cases.

This paper is a multi-institutional matched study of surgery-alone compared with surgery plus PORT in patients with LA-HNSCC. Our results showed that PORT did not improve OS or DFS in patients with LA-HNSCC when compared with surgery alone. The results are strengthened by the strict matching criteria. Two large head and neck cancer centres were utilised to recruit patients. Two institutions have adopted very similar treatment paradigms in their approaches to LA-HNSCC with a preference for withholding PORT wherever possible.

PORT has been included in the treatment guidelines for LA-HNSCC for about 50 years [12]. The results of our previous clinical study showed that patients with LA-HNSCC could obtain a good prognosis without PORT [10, 11, 13]. We designed this multicentre case-matching study to validate the conclusions of our previous study. The advantage of the case-matching study is that the factors and conditions affecting the experiment can be controlled in advance to make it as balanced as possible, reduce errors, and at the same time reduce the individual differences of the experimental subjects (patients). In order to eliminate the prescription bias associated with cohort studies comparing outcomes for patients receiving surgery alone with those treated by surgery plus PORT, we applied rigid matching criteria. Although minimised, the prescription bias cannot be completely eliminated. Medical comorbidity seriously affects the survival of patients, and because of the limited amount of data in our database, we did not use this factor as a match. That is unavoidable in all types of clinical studies, including RCTs. Meanwhile, the surgical techniques used in this study are similar to those of some scholars [14, 15]. The basic concept for the anatomic unit (subunit) resection is removal of the entire anatomical subunit in which tumour is contained rather than removing tumour with a 1–2 cm histopathological margin. The results of this clinical study showed that PORT did not significantly improve the OS rate or tumour-free survival rate of patients with LA-HNSCC. Multivariate analysis showed that only lymph node metastasis and tumour site were independent factors affecting survival.

In contrast with our study, many other trials [16‐18] show that PORT shows an improvement in the outcome of patients with LA-HNSCC. The study by Yanamoto et al. [16] showed that after the propensity score analysis, PORT/concurrent chemoradiotherapy significantly improved OS (HR 0.554; 95% CI 0.38–0.80; P = 0.001) and DSS (HR 0.641; 95% CI 0.43–0.96; P = 0.030) compared to surgery only. The finding of Wang et al. [17] suggested that patients undergoing surgery had a 5-year OS of 27% compared with 66% for patients undergoing surgery plus adjuvant radiotherapy (P = 0.003); patients undergoing surgery had a 5-year DFS of 34% compared with 74% for patients undergoing surgery plus adjuvant radiotherapy (P = 0.001). However, compared with our study, these studies had some differences. Firstly, by the prescription bias associated with an unmatched comparative study of this nature, these above data are confounded; case-matching study can minimise the error in this aspect. Secondly, the concept of surgery in our study is different from that in the above studies. Another two case-matching studies [19, 20] came to the similar conclusion as our study: with the addition of adjuvant radiation (or radiation and chemotherapy), no difference in OS or DSS was observed. Another possible reason is that only five of the 114 patients in this study were oropharyngeal SCC and the rest were OSCC.

About the local toxicity associated with PORT on health-related quality of life, the negative impact is well documented [21, 22]. During and after treatment, overall health-related quality of life declines, and it will be at least 1 year before the baseline levels are recovered. However, physical function related to saliva, swallowing and chewing remain persistently low [23]. For conventional radiotherapy compared to intensity-modulated radiotherapy, this effect is worse [24].

Based on the results of our clinical trial, we conclude that PORT does not improve the OS of patients with LA-HNSCC, but may benefit the DFS of patients. Therefore, we believe that PORT is not necessary for patients with LA-HNSCC who are treated for the first time as long as the head and neck cancer surgeon adheres to appropriate surgical concepts (specifically the anatomic unit (subunit) resection). The evidence for this conclusion is not strong enough. Furthermore, the prognoses of the PORT for different surgical approaches need to be confirmed by other high-quality clinical studies.