The optimal management for advanced hypopharyngeal cancer patients remains undetermined. In the pre-CCRT era, according to the results from the two important randomized control trials conducted on advanced hypopharyngeal cancer patients,8,9 the surgery-based treatment seems to have better treatment results in both disease control and patient survival when compared with radiation-based treatment. In one of these two trials, Beauvillain et al.9 compared the treatment results in 90 resectable hypopharyngeal cancer patients (T3–4, N0–3) by dividing them into two groups: induction chemotherapy plus radiotherapy group (RT arm) vs induction chemotherapy plus surgery plus radiotherapy group (S+RT arm). In this trial, more than 80% of patients in both groups first completed 3 cycles of induction chemotherapy as scheduled. Although fewer patients in the S+RT arm (67%) achieved complete or partial response after induction chemotherapy than patients in the RT arm (79%), the 5-year local control (63% vs 39%, p < 0.01) and 5-year OS (37% vs 19%, p = 0.04) were significantly better in the S+RT arm. Notably, in both groups, chemotherapy responders also had better OS than non-responders. Another phase III clinical trial8 from the EORTC Head and Neck Cancer Cooperative Group was conducted to test the feasibility of laryngeal preservation by a non-surgical approach in 202 stage II–IV (T2–4, N0–3) resectable pyriform sinus cancer patients. Patients were first randomized into the immediate surgery group or induction chemotherapy group in this trial. Patients in the immediate surgery group received total laryngectomy + partial pharyngectomy + radiotherapy, while patients in the induction chemotherapy group received 3 scheduled cycles of chemotherapy (cisplatin 100 mg/m2 + 5-FU 1000 mg/m2) first (CT arm). In the CT arm, only patients who achieved complete response after two or three cycles of chemotherapy proceeded to definitive radiotherapy. After induction chemotherapy selection, 35% of patients in the CT arm proceeded to curative surgery instead of definitive radiotherapy. After radiotherapy, a salvage operation had to be performed in 12 additional patients (8 to primary and 4 to the neck) in the CT arm. Although both the 5-year disease-free survival and overall survival were comparable in both arms, considering that: (1) 35% of patients in the CT arm did not achieve complete response after chemotherapy (about half of T3 patients and all of the T4 patients in the CT arm had to receive curative surgery rather than proceeding to subsequent radiotherapy), and (2) salvage operation was performed in 8 additional patients with local failure following radiotherapy in the CT arm, it is reasonable to conclude that primary surgery (followed by radiotherapy) would provide superior local disease control and better survival compared with primary radiotherapy alone if pre-radiotherapy chemotherapy selection were not performed.

In the contemporary era, however, cumulating evidence has demonstrated that IMRT may confer better loco-regional control in head and neck cancers compared with conventional radiotherapy.24 The administration of chemotherapy with IMRT may further improve disease control for patients with locally advanced head and neck squamous cell carcinoma.25 Thus, although there’s still no randomized trial comparing the effectiveness between surgery-based treatment and the radiation-based approach (with chemotherapy) in locally advanced hypopharyngeal cancers, it is natural to postulate that, with the advancement of radiotherapy technology and the addition of chemotherapy, the local disease control advantage and the survival benefit from primary surgery noted in the pre-CCRT era would become insignificant compared with contemporary organ-preservation CCRT. In addition, radiation-based therapy could provide the opportunity for organ preservation in advanced T staged patients who will inevitably require extensive resection of laryngeal and hypopharyngeal structures. Encouraged by the results from an organ-preserving trial conducted on advanced laryngeal cancers,26 in recent decades, there has also been a clear trend of changing from surgery-based approach toward nonsurgical radiation-based therapy for hypopharyngeal cancer patients, even in patients with advanced T stage diseases.2,3,4

Results from our study are somewhat against the assumption that definitive CCRT will provide equal loco-regional disease control compared with OP-CRT. Our results clearly demonstrated that, compared with the CCRT approach, there are still significant local and regional control advantages in advanced T (T3, T4a) or advanced N stage (N2b, N2c) patients receiving OP-CRT as their primary treatment (Table 2). In support of our findings, Lee et al.27 also reported a better local disease control rate in the surgery plus postoperative radiotherapy group than in the CCRT (p < 0.05). With a match-paired analysis conducted on advanced hypopharyngeal cancer patients from multiple institutions (n = 254), Iwae et al.28 also demonstrated a significantly better 5-year loco-regional control rate in the total pharyngolaryngectomy group compared with the chemoradiation group (82.2% vs 63.6%, p < 0.01). Thus, even in the era of CCRT, it seems that surgery-based primary treatment still offers better local and regional disease control in advanced-stage hypopharyngeal cancer patients than those treated with definitive CCRT.

Unexpectedly, in our study, although advance-staged patients in the OP-CRT group had fewer loco-regional recurrences (Table 3), we did notice that a higher proportion of stage IVA patients in the OP-CRT group suffered from distant metastasis after treatment compared with patients in the CCRT group (p = 0.06). The underlying cause for this observation may be explained by two speculations: (1) fewer patients in the OP-CRT group received a full course of induction chemotherapy before curative surgery, and (2) more patients in the CCRT group died from uncontrolled loco-regional disease before distant metastasis became clinically evident. Notably, the EORTC trial8 also cautiously reported a higher distant metastasis rate observed in the primary surgery arm (36%) than in the CT arm (25%).

One strength of our study is that, by providing a detailed failure pattern analysis on local, regional, and distant sites after treatment, we were able to provide additional insights to interpret the survival data in our patients. For instance, in advanced-stage patients (especially in stage IVA), although we noticed a clear, inarguable advantage of the 3-year loco-regional relapse-free survival in the OP-CRT group (stage IVA, p < 0.0001, Table 3), the difference in 3-year disease-free survival between these two groups was less remarkable (stage IVA, p = 0.03, Table 4) due to more distant metastasis in the OP-CRT group (Table 3). And since curative intent salvage treatment for distant failure (more patients in the OP-CRT group) is less likely to succeed than salvage treatment for loco-regional relapse (more patients in the CCRT group), we only noticed a marginally significant benefit of 3-year DSS in the OP-CRT group compared with the CCRT group (stage IVA, 67 vs 50%, p = 0.07, Table 4). Considering that OS may additionally be attributed to factors unrelated to primary treatment, the actual benefit in a specific treatment modality might be obscured in studies that cannot recruit a large number of patients. Our study noted that 30–40% of ‘failure’ was caused by either a second primary tumor or disease-unrelated death in both groups of patients (Table S2). Thus, it is not unexpected to see that the difference in the 3-year OS would become statistically non-significant (52 vs 44%, p = 0.37, Table 4) between these two groups. Taken together, our study advocates that, when comparing different treatment modalities, each survival data should be interpreted cautiously.

Based on the above understandings, it is not surprising that some of the previous studies conducted on advanced-stage hypopharyngeal cancer patients reported better local or loco-regional control rates in the primary surgery group but failed to demonstrate significantly better DSS or OS compared with the CCRT.27,28 It is also reasonable to see that, in retrospective cohort studies comparing DSS or OS between primary surgery and definitive chemoradiation, most studies reported marginally better results, or a trend toward better results, in the surgery group than in the radiation group. For instance, Tsou et al.29 reported better 5-year DSS in the primary surgery followed by CCRT group compared with the primary CCRT followed by early surgical salvage group in both stage III (51 vs 38%, p = 0.03) and stage IV (23.1 vs 11%, p = 0.05) patients. Kim et al.30 reported non-statistically significant improvements in the 5-year DFS (64 vs 51%, p = 0.49) and OS (67% vs 59%, p = 0.83) in resectable stage III/IV patients receiving total laryngectomy compared with patients receiving definitive CCRT. Harries et al.10 reported a trend toward improvement in 5-year relapse-free survival (RFS) (41% vs 35%, p = 0.18) and OS (66 vs 54%, p = 0.09) in advanced-stage patients receiving primary surgery compared with definitive CCRT. After multivariate Cox regression analysis, that study similarly showed a trend toward improved RFS (hazard ratio 2.97, p = 0.12) and OS (hazard ratio 4.78, p = 0.06) in patients receiving surgery.10 In a population-based study, Cheng et al.31 reported OS benefits in both stage III and stage IVA patients receiving primary surgery. Another population-based study similarly reported a better 5-year OS in T4 patients receiving total laryngectomy than patients receiving definitive CCRT (29% vs 25%, p = 0.04).4 In contrast, Jang et al.32 demonstrated comparable 5-year OS for both primary surgery and definitive CCRT groups in 177 resectable, advanced-stage hypopharyngeal cancer patients. In a population-based study, Kim et al.33 failed to demonstrate a 3-year OS advantage in the surgery group compared with the definitive CCRT group, even in T4a patients (30 vs 26%, p = 0.44). The above evidence implies that, while the addition of surgery could confer better loco-regional control in advanced-stage hypopharyngeal cancer patients, the survival advantage of better loco-regional control in the surgery group may not be directly translated into better DFS, DSS, or OS, due to diverse clinical conditions specific to the included study population. Thus, randomized control trials are mandatory to decipher the best approach for locally advanced hypopharyngeal cancer patients.

Another strength of our study is that, by providing a detailed analysis of the failure patterns, we were able to improve our treatment strategies accordingly. For patients with early stage disease (stages I, II), effective disease control can be achieved using either surgery or radiotherapy. Thus, a single modality treatment, such as laryngeal conservation surgery or (chemo-)radiotherapy,34 will be preferred to minimize treatment-related morbidity and maximize functional preservation. For patients with advanced-stage disease who choose to have a primary operation, induction chemotherapy or adjuvant chemotherapy should be considered to decrease the probability of distant metastasis that might occur after curative treatment, especially in T4a, N2b, and N2c patients. For patients who choose definitive CCRT as their primary treatment, we may perform chemotherapy selection8,35 to identify good responders before curative radiation to minimize the chance of difficult salvage surgery after CCRT.34 For patients with very advanced-stage disease (T4b or N3), clinical trials with novel therapeutic approaches may be considered a priority.

Previous investigations have demonstrated the survival significance of tracheostomy and feeding tubes on the survival of patients undergoing (chemo)radiotherapy. In terms of tracheostomy, an earlier retrospective study of 270 irradiated laryngeal cancer patients demonstrated that patients without tracheostomy have better 2-year disease-free survival than those with tracheostomy (74% vs 41%).36 The loco-regional controls differ by anatomic subsite, with the most significant difference occurring in the glottis (tracheostomy vs no-tracheostomy: 69% vs 94%). Of interest, almost all the patients wearing tracheostomy had T3/T4-stage lesions, reflecting the tracheostomy as a surrogate for locally advanced tumors with airway comprises. Later, in a single-institution study investigating the factor predicting organ preservation failure after CCRT, Heukelom et al.37 also identified T4 lesion and pretreatment tracheostomy as independent prognostic factors in laryngeal and hypopharyngeal cancers. Our results are consistent with these findings that the tracheostomy prior to, or during, the radiotherapy may be a surrogate for an advanced T stage. In addition, patients without tracheostomy did show statistically significant oncologic benefits over patients without tracheostomy, although tracheostomy was not identified as an independent prognostic factor for DFS after controlling for other variables. On the other hand, enormous investigations have emphasized the significance of feeding tubes [mainly percutaneous endoscopic gastrostomy (PEG)] in laryngo- and pharyngeal-cancer patients treated with CCRT. A systematic review conducted by Mellors et al.38 demonstrated that the prophylactic PEG was associated with reduced incidence of critical weight loss (> 10% body weight loss) and improved short–term quality of life compared with those receiving reactive nutritional support. The results are consistent with the review by McClelland et al.39, supporting the concept of prophylactic PEG placement in patients at risk for malnutrition during radiation therapy. However, our analysis demonstrated that feeding tube use was negatively associated with disease-free survival, disease-specific survival, and overall survival. We hypothesize that the negative impact was that some of our patients in the CCRT group received reactive feeding tubes instead. The inferior survival in the feeding tube group may be because the patients with reactive nutritional support have already developed significant weight loss40 or poorer tolerance41 to radiation that causes RT interruption or prolongation, which are recognized as major negative prognostic factors for patients receiving (chemo)radiotherapy. Consequently, despite the limited recruited patients and the retrospective design in the current study, our conclusions imply that execution of the upfront surgery rather than CCRT on patients needing FT may improve the overall prognosis of hypopharyngeal cancer. However, a well-designed prospective study is needed to confirm our hypothesis in the future.

There are some limitations to our study. First, the treatment modality used for the OP-CRT and the CCRT groups was not uniform (Table S1). Second, whether the genetic background of each study population impacts the prognosis related to each specific treatment modality (primary surgery vs radiation) is not known. It is noticeable that the other two studies also from Taiwan similarly reported favorable outcomes in advanced-stage hypopharyngeal cancer patients receiving primary surgery.29,31 We also recently demonstrated that pre-diagnosis alcohol consumption and the polymorphisms of ALDH2 impact the prognosis of HNC.42 In a recent study conducted in Japan, heavy drinking combined with the ALDH2*2 allele was an independent poor prognostic factor of OS and DFS for hypopharyngeal cancer patients.43 Since the population in East Asia, including Taiwan and Japan, has the highest prevalence of the poor functional allele (ALDH2*2) in the world,44 the prognostic impact of ALDH2 polymorphism on hypopharyngeal cancer treatment deserves further investigation. Third, the functional results of speech, swallowing, and quality life assessment in our study population require additional study. Knowledge of the nutrition impact symptoms, functional disabilities, and potential morbidity profiles by different treatments are important because some studies have established their associations with quality of life45 and survival46 in head and neck cancer patients. However, the relevant information in the current study is insufficient. Further study is warranted for metrics measurements by different therapeutic strategies to refine our treatment delivery for each HPC patient. Fourth, we have few elders in the current study, which limited the analysis regarding impact of age on survival. Elderly patients are more vulnerable to developing adverse events after interventions.47 Therefore, the best choice of treatment and whether to treat or not should consider many aspects because the associated complications may potentially translate into a higher proportion of complication-related deaths. Fifth, no randomized trial has been conducted to compare the outcome difference between patients receiving CCRT or induction chemotherapy (IC) followed by CRT. However, a recent retrospective study by Chung et al.48 found their 127 staged III/IV HPC patients have comparable 5-year overall survival, disease-free survival after salvages, and laryngeal preservation rates between the two treatments, supporting our methodology by merging CCRT and IC+CRT for further analysis. However, future solid evidence is needed to identify the factors that benefit from either treatment, especially for the patients proceeding with an organ preservation strategy. Sixth, all patients in the current investigation were staged according to the definitions outlined in the AJCC 7th edition because we performed data collection before the general adoption of the AJCC 8th edition in 2018. The major difference between the 8th edition and the 7th is the separation of hypopharyngeal cancer from pharynx cancer and the categorization of the involved lymph node to N3b if an extra-nodal extension (ENE) was present.49 However, Chai et al.50 reported that using contrast-enhanced computed tomography alone to diagnose ENE of regional lymph nodes is not very reliable. Because there is no pathologic evidence for the CCRT group, re-classification from the 7th to 8th edition may lead to mal-categorization, resulting in the misinterpretation of our results. Another limitation is that the time-to-event is a methodologically relevant outcome in a retrospective cohort. However, we hypothesize that the effect should be minimized in the current study because the evaluation modalities are generally uniform before and after the primary treatments. Additionally, we have recommended a more vigilant post-treatment surveillance for patients with a high risk for recurrence and relapse. As for the clinically controversial cases, we will actively arrange a biopsy of the suspicious lesion to confirm the pathologic diagnosis. If the patient declined a biopsy, we reached a consensus through discussions with experts in other fields. Lastly, knowledge regarding hospital admission stay is a specific issue in oncology because of its significant impacts on the health care delivery policy and patients’ health-related quality of life.51 With data from the National Health Insurance program in Taiwan, we might decipher the most optimal treatment strategy for this devastating disease by integrating the analysis of healthcare costs in different treatment modalities.