Long-term results of induction chemotherapy for non-operable esophageal squamous cell carcinoma followed by concurrent chemoradiotherapy: a single-centre experience

Esophageal cancer (EC) is the sixth most common cause of cancer death worldwide, with more than 550,000 new cases of esophageal carcinoma diagnosed each year.^1,2 Unlike most Western countries, esophageal squamous cell carcinoma (ESCC) is still the main pathological type in China.³ Regardless of its histological type, the overall survival (OS) of patients with EC is still poor.⁴ For the management of EC which is deemed a medically unresectable tumor, definitive concurrent chemoradiotherapy (CCRT) is the standard therapy, guaranteeing organ preservation and providing a better quality of life.⁵ However, although definitive CCRT results in encouraging short-term outcomes in the majority of patients, the prognosis remains unfavourable, with 5-year overall survival rates of 20%.⁶ Especially, the rates of locoregional recurrence (LR) and distant metastasis after definitive CCRT can be as high as 50%.^7,8 Therefore, improvement in treatment intensity is greatly needed.

Induction chemotherapy (IC) is an attractive approach but also controversial. Theoretically, the additional IC followed by concurrent chemoradiotherapy (IC-CCRT) has potential benefits for early eradication of micro-metastases, increased tumor radiosensitivity, prevention of tumor progression, and even prolonged OS.⁹ Previous studies have suggested that IC-CCRT has better failure-free survival, overall survival, and distant failure-free survival than CCRT alone in nasopharyngeal cancer, and has been cited by National Comprehensive Cancer Network (NCCN) clinical guidelines.¹⁰ However, the addition of IC to CCRT in the management of ESCC is less reported, and the results of the retrospective studies showed conflicting results.^11,12,13 A randomized controlled trial showed that compared to CCRT alone, the addition of induction chemotherapy with docetaxel plus cisplatin failed to significantly improve the response rate or survival outcomes in unselected ESCC, which were limited by staging and response evaluation issues.⁹ Previous research revealed that IC before CCRT was associated with improvements in pathological complete response (pCR) rate and survival in neoadjuvant therapy settings.^14,15 Therefore, it is important to identify patients who may benefit from IC, especially the patients who are responsive to chemotherapy. Furthermore, the radiotherapy techniques in previous studies were mainly based on 3D-CRT or IMRT.^9,11,12,13 The true value of IC for ESCC patients remains unclear in the era of modern technique IMRT/volumetric modulated arc therapy (VMAT), which could greatly improve the accuracy of radiotherapy with lower toxicity.¹⁶ To the best of our knowledge, there are no published researches to date that compared IC + CCRT vs. CCRT alone for the ESCC patients receiving IMRT/VMAT only.

Therefore, we performed this retrospective study to evaluate the long-term survival outcomes among the ESCC patients who were treated with IC + CCRT to better understand the feasibility, efficacy, and safety of this approach by propensity score matched (PSM) methods. We further performed a stratified analysis to analyze the relationship between tumor response to IC and treatment outcomes.

The efficacy of IC has not been well documented previously for ESCC patients receiving IMRT/VMAT-based CCRT. In the present study, we performed a PSM analysis of patients treated with or without IC before standard CCRT to better understand the efficacy and toxicities of IC. We found that IC + CCRT was not superior to CCRT in terms of 5-year OS, RFS, and DMFS regarding original or well-matched data. The stratified analysis further demonstrated IC + CCRT improved the 5-year OS, RFS, and DMFS for the patients with response (responders or good responders) to IC, whereas it might not have a positive impact for non-responding or poorly responding patients and seemed to have limited benefits in long-term survival. Nearly all of the patients who are alive in our study have completed valid follow-up for 2 years (except for individual deleted data), and the longest followup period was over 7 years. Concerning toxicity, there was no significant difference in toxicity between patients who had IC and those who did not. According to our knowledge, this is the first study to compare the survival benefits of the addition of IC to CCRT and IMRT/VMAT only in ESCC patients. The main strength of our study is that the application of the PSM method balances the baseline characteristics of the included population to reduce potential confounders, thus mimicking the matching observed in randomized controlled trials (RCTs).

Since the Radiation Therapy Oncology Group (RTOG) 85–01 trial indicated that the outcome of CCRT was significantly better than that of RT alone for ESCC patients, definitive CCRT has been a standard treatment.⁵ However, the long-term outcomes remain limited and the failure rate was 50%.^6,7 Updated meta-analyses and systematic reviews of clinical trials have demonstrated that IC + CCRT could prolong short-term survival in unresectable EC patients.¹⁹ Unfortunately, several subsequent similar studies including a prospective randomized clinical trial got negative results.^9,11,12,13 In our study, we found that the IC + CCRT group achieved higher 5-year OS (39.0% vs. 29.3%, p = 0.360), RFS (39.0% vs. 26.9%, p = 0.142), and DMFS (33.6% vs. 27.2%, p = 0.515) compared with the CCRT group, although the difference between the two groups did not reach statistical significance. Rational explanations for the discrepancy may include: 1) The enrolled patients in the current study included clinical stages T1–4N0–3, and the patients with early-stage disease (T1–2N0–1) may not benefit from IC. 2) The regimens of IC in the retrospective studies were not uniform, which may provide a slight bias toward a negative result.

It was reported that the late tumor stage was an important risk factor for poor prognosis in ESCC.²⁰ In this study, multivariate Cox analysis showed AJCC stage was regarded as an independent predictive factor that affects OS, RFS, and DMFS. Later AJCC stage has inferior OS (p = 0.006; HR, 2.533; 95% CI, 1.305–4.916) and RFS (p = 0.003; HR, 2.738; 95% CI, 1.413–5.305), and DMFS (p = 0.001; HR, 2.951; 95% CI, 1.560–5.582) than early AJCC stage ESCC, which was similar to those in Hsieh's report.²¹ We also found radiotherapy treatment time was another independent predictive factor for OS, RFS, and DMFS. In daily clinical practice, unplanned treatment interruptions are inevitable for many reasons. Sher reported that a prolonged total radiotherapy time > 51 days is associated with an inferior overall survival (hazard ratio = 1.63, p = 0.0058). For each additional day required to finish radiotherapy, the hazard rate of death increased by 4.2%.²² Cannon et al. reviewed outcomes of 171 head and neck cancer patients treated with CCRT and found that patients with radiotherapy time ≤ 49 days had a superior 3-year local control rate and OS compared to those with radiotherapy time > 49 days (88% versus 71%, 81% versus 58%, respectively).²³ In our study, the total radiotherapy time ≥ 49 days has the inferior OS (p = 0.025; HR, 1.762; 95% CI, 1.074–2.891), RFS (p = 0.009; HR, 1.920; 95% CI, 1.178–3.131) and DMFS (p = 0.014; HR, 1.827; 95% CI, 1.127–2.961). For patients with total radiotherapy time ≥ 49 days in our study, the vast majority of interruption was due to machine breakdown, machine maintenance, and treatment toxicity (chemoradiotherapy or radiotherapy only). Briefly, 11 patients experienced treatment breaks because of radiation esophagitis. Meanwhile, 6 patients experienced unscheduled interruptions due to chemotherapy toxicities. Therefore, it seems a necessity for radiotherapy without gaps or delays for the sake of improved outcomes and control of disease progression.

So far, there is no evidence to suggest that advanced age is an independent contraindication for CCRT in the retrospective studies.^{20, 24} Wu et al. reported that there was no statistically significant difference between CCRT and RT alone for patients aged 75 years or older.²⁰ In the present study, we found that age is not a predictive factor of OS (p = 0.167; HR, 0.749; 95% CI, 0.498–1.128) and RFS (p = 0.265; HR, 0.794; 95% CI, 0.530–1.191) by multivariate analysis. Interestingly, we found that age < 60 was independently associated with worse DMFS in the current study.²⁵ Similar results have been found in Colzani'study, suggesting that breast cancer patients younger than 50 years at diagnosis had a higher risk of distant metastasis.²⁶ The possible reason may be that the metastases lose aggressive character or that the host defense is better equipped to deal with them in advancing age.²⁷

IC may only benefit a certain subgroup but not unselected patients with ESCC. As is known to all, the T stage was associated with a worse prognosis in esophageal carcinoma.²⁸ Akinori reported that IC for T4 esophageal cancer offered comparable local control and survival to conventional CCRT, and suggested that the strategy of IC followed by CCRT was efficient for T4M0 esophageal cancer.¹³ To identify the subgroups that may benefit from IC, we performed a stratified analysis based on the T stage. Our results indicated that IC + CCRT group achieved better 5-year OS (33.7% vs. 22.5%), RFS (33.7% vs. 19.6%), and DMFS (29.0% vs. 19.6%) compared with the CCRT group, however, there was no significant difference. Moreover, 90% of the symptoms of dysphagia improved significantly after IC, which was consistent with the trial INT 0122.²⁹ In the study by Luo et al.¹², the IC responders (CR or PR) group achieved significantly more favourable OS compared with the IC non-responders (SD or PD) group and the CCRT alone group (p = 0.002). Besides, the post-hoc analysis in prospective research also demonstrated that response to IC was associated with more favourable survival.⁹ Consistent with the results of previous studies, our results suggested that the responders (CR, PR, or SD) to IC had significantly more favourable survival compared with non-responders (PD), or with patients in the CCRT group, with corresponding 5-year OS rates of 41.7%, 14.3%, and 29.3%, respectively. In addition, we further analyzed the potentially important role of IC good responders (CR or PR) from the whole IC group. Our data suggested that the IC good responders might have a significantly prolonged OS, improved locoregional control, and reduced distant metastasis, with corresponding 5-year OS rate, RFS rate, and DMFS of 65.6%, 65.6%, and 62.5%, respectively. Considering the poor prognosis of the IC non-responders or IC poor responders, tumor response after IC could be used to guide subsequent treatment decisions, such as switching to alternative agents included targeted therapies, immunotherapies, or radiosensitizers during radiotherapy for non-responders or IC poor responders. Therefore, further studies are needed to overcome this unfavorable biological characteristic.

In our study, the rate of grade ≥ 3 RE in the IC + CCRT group and the CCRT group had no statistical significance (p = 0.678). There was also no statistical difference in the incidence of myelosuppression (p = 0.944). It has been reported that patients with T4 had an incidence of perforation of 14–23% during CCRT, and the addition of IC before CCRT might reduce the risk of perforation by decreasing the tumor volume before encountering severe esophagitis.^30,31 No esophageal fistula or perforation occurred in our study, which is one of the most troublesome complications caused by CCRT. The possible reason was that ESCC patients by using IMRT/VMAT only are superior to the two-dimensional conformal radiation (2D-CRT) or 3D-CRT. IMRT/VMAT improves the treatment ratio due to the highly conformal dose distributions in the tumor target volume and sharp dose gradients at the transition to the adjacent normal structures. The potential benefits of IMRT/VMAT were investigated in a series of studies.³² Our results are consistent with the outcomes of the studies in the IMRT/VMAT era.

There were several limitations in our study. First, although we used PSM, a method aimed to minimize the impact of observed confounders, the retrospective nature of this study cannot exclude the possibility of bias caused by confounding factors, and adding too many match restrictions would lead to small sample size and might not represent the initial population. Secondly, our study is limited to ESCC patients and could not applied to other types of EC. Finally, due to the retrospective characteristic, IC regimes and concurrence chemotherapy regimens were not uniform. More well-designed prospective, randomized controlled trials are warranted to further confirm the role of IC.

In this study, our results showed the addition of IC to CCRT was not superior to CCRT in unselected ESCC patients, while IC responders might benefit from this regime without an increase in toxicities.