Neoadjuvant versus definitive chemoradiation in patients with squamous cell carcinoma of the esophagus

In patients with locally advanced esophageal squamous cell carcinoma (E-SCC) trimodal therapy including neoadjuvant chemoradiation followed by surgery (nCRT + S) demonstrated its potential to improve overall survival (OS) and disease-free survival (DFS) compared to surgery alone [1‐3]. Therefore, it has been established as the treatment of choice for patients with non-cervical E-SCC who are suitable for surgery [4]. In contrast, patients with locally advanced E-SCC who decline surgery, are medically inoperable or have unresectable tumors should undergo definitive chemoradiation (dCRT). Because of this, patients undergoing dCRT, show in general more advanced tumors and are in a worse general condition than patients who are treated with nCRT + S. Therefore, a meaningful comparison of both treatment options is difficult. Up to the present there are only two randomized trials comparing dCRT and nCRT + S in patients with esophageal cancer (EC) [5, 6]. The study by Stahl and colleagues, which was terminated early, revealed no significant difference for OS between both treatment groups, while nCRT + S was associated with a significantly increased local progression-free survival [5]. However, all patients in this trial received induction chemotherapy, which is not in line with current treatment recommendations and therefore compromises the interpretation of the results. In accordance with that, a French study also found no difference in OS between surgery or continuation of chemoradiation (CRT) in patients with good response to CRT [6]. Furthermore, there was no significant difference regarding local control rate. One of the major problems of this study is that only patients with good response to CRT were randomized to continuation of chemoradiation or surgery. Therefore, the results are only applicable to a specific subgroup of patients. These results are in line with a recently published retrospective trial by Haefner et al. [7] that compared nCRT + S and dCRT in patients with locally advanced EC. After a median follow-up time of 20.4 months, no significant differences were seen regarding progression-free survival (PFS) and OS between both treatment regimes. In contrast, a recent retrospective cohort study by Barbetta et al. [8] demonstrated an improved OS and DFS after nCRT + S compared to dCRT in patients with thoracic or distal E-SCC. In addition, further retrospective trials also demonstrated an increased survival after nCRT + S in EC patients [9‐11]. One major problem with the studies mentioned above is, that most of them included patients with both, adenocarcinoma (AC) and E-SCC. This clearly affects the results because the effect of CRT is higher for patients with E-SCC than for patients with adenocarcinoma [2] and in general these two tumor types – although occurring at the same location - are biology-wise completely different tumor entities. Moreover, many patients were treated with 3-dimensional conformal radiotherapy (3D-CRT). Although no significant differences regarding progression-free survival (PFS) and OS were seen in two retrospective trials comparing 3D-CRT and modern radiation techniques like intensity-modulated radiotherapy (IMRT) for nCRT + S or dCRT in patients with EC [12, 13], the use of modern radiation techniques is at least associated with lower doses to the organs at risk in patients undergoing nCRT + S [12]. Due to the incoherent results of previous trials, our aim is to report the results for patients with only E-SCC, who underwent dCRT or nCRT + S with modern radiation techniques.

This study includes 95 patients with E-SCC who were treated with either dCRT or nCRT + S at our department between 2011 and 2017. Inclusion criteria were histologically proven E-SCC, curative treatment approach with either dCRT or nCRT + S, simultaneous chemotherapy, the use of IMRT and at least one follow-up after the end of therapy. Exclusion criteria were distant metastases (M1) at the time of diagnosis, simultaneous radiotherapy or CRT of a second cancer, discontinuation of therapy due to any reason and a total radiation dose < 50 Gray (Gy) for patients undergoing dCRT or < 41.4 Gy for patients undergoing nCRT + S. All patients were staged with ¹⁸Fludeoxyglucose positron emission tomography combined with computed tomography (¹⁸FDG-PET/CT) (95%) or computed tomography alone (5%). Whenever possible, endoscopic ultrasound was also used to assess T- and N-stage.

In this analysis, we compared outcome of nCRT + S and dCRT in patients with E-SCC. Thereby, we only included patients who were treated with IMRT, to represent current clinical practice. While no significant differences were seen for OS and PFS, the rate of local tumor recurrence was significantly higher in patients treated with dCRT than in those treated with nCRT + S. However, in a multivariate Cox regression analysis, treatment regime was not independently associated with OS, PFS or rate of local tumor recurrence. Instead, the only parameter independently affecting OS, PFS, and rate of local tumor recurrence was tumor location, while patients’ age and tumor grade were independently associated with OS and PFS.

While our absolute data for OS and PFS are comparable with two other recent studies [7, 8], there are conflicting results in terms of the relative difference between patients treated with nCRT + S and patients treated with dCRT. Haefner and colleagues [7] compared dCRT with nCRT + S in patients with esophageal cancer. In contrast to our study, the authors included patients with AC and patients treated with dCRT received two cycles of adjuvant chemotherapy with cisplatin and 5-fluorouracil. After a median follow-up of 20.4 months, no significant differences were visible for median OS (25.9 months vs. 20.6 months) and PFS (14.9 months vs. 15.6 months). In contrast to that, Barbetta and colleagues [8] reported an improved OS (median OS 2.3 years vs. 3.1 years) and DFS (median DFS 1 year vs. 1.8 years) after nCRT + S. The most obvious difference to the present study is the exclusion of patients with cervical or upper thoracic tumors. In our study, a more proximal tumor location was associated with shorter OS and PFS within multivariate Cox regression analysis. While we found no data evaluating the impact of tumor location after nCRT + S or dCRT in E-SCC patients, proximal tumor location was associated with decreased OS in patients with pT2-3N0M0 carcinoma after surgery alone [15]. As patients with cervical tumor location are typically treated with dCRT [16], and therefore the rate of patients with cervical tumor location is frequently higher within the subgroup of patients treated with dCRT, this might explain the improved OS after nCRT + S in some trials. Interestingly, the only study that excluded patients with cervical or upper thoracic tumor location and performed a propensity score-matched analysis, also reported an improved OS for E-SCC patients undergoing nCRT + S [8]. One can speculate that especially after exclusion of patients with cervical or upper thoracic tumor location the number of patients in most trials would be too low to reveal any significant differences in OS. Another factor which might affect the results is the fact that patients who underwent dCRT in our study had a significantly lower rate of lymph node metastases. This might impact the results since it is well established that lymph node involvement is an important and independent prognostic factor in EC patients [17‐19].

In accordance with other studies [5, 8, 10], dCRT was associated with an increased rate of local tumor recurrence in our study. However, Stahl [5] and Liao [10] included patients with different tumor locations including patients with tumors of the cervical esophagus. Based on our results, it is conceivable that the difference in the rate of local tumor recurrence in these studies is biased by differences regarding tumor locations between both cohorts. Although, after excluding patients with cervical tumors, we still recognized a strong trend towards an increased local tumor control after nCRT + S compared to dCRT (p = 0.051). In addition, in a recent trial by Barbetta and colleagues [8], which also demonstrated an increased risk of local tumor recurrence after dCRT compared to nCRT + S, this kind of bias was ruled out by performing a 1:1 propensity score-matching approach and excluding patients with tumors of the upper esophagus.

In accordance with the studies by Haefner et al. [7] and Barbetta et al. [8], the most common reason for treatment failure in patients undergoing dCRT was local or regional tumor recurrence. The absolute rate of local and regional recurrences after dCRT in our study was slightly higher than in the study by Haefner and colleagues (local 38% vs. 24%; regional 13% vs. 4%). This difference might be explained by the fact, that patients treated with dCRT were were more likely to have tumors of the cervical esophagus (47% vs. 16%) and were also more likely to have lymph node metastases (85% vs. 77%). Interestingly, our results for local and regional tumor recurrence are comparable to the results by Barbetta and colleagues, despite patients with upper esophageal carcinoma were excluded in their study. The higher rate of regional tumor recurrence in patients undergoing nCRT + S in the study by Barbetta et al., might be explained by the higher rate of patients with lymph node metastases (100% vs. 85%), which is an independent risk factor for both, locoregional and distant recurrence after nCRT + S [20]. Therefore, the difference in terms of the rate of lymph node metastases might also explain the higher rate of regional recurrence after nCRT + S in our study compared to the results by Haefner et al. (23% vs. 5%) [7].

For both, patients treated with nCRT + S and patients treated with dCRT, the rate of distant disease recurrence in our study is remarkably lower than in other recent trials [7, 8]. While the exclusion of patients with AC, who have a significantly higher risk of distant tumor recurrence after dCRT compared to patients with E-SCC [21], might partially explain the higher distant recurrence rate in the study by Haefner and colleagues [7], the reason for the higher rate of distant tumor recurrences in the study by Barbetta et al. [8] remains unclear. However, we have to point out, that the rate of distant recurrences in our study might be slightly underestimated. Due to the retrospective nature of our study, not all patients underwent periodic computed tomography during follow-up. While survival data of patients were completed by contacting the local registration offices, these data provided no further information about tumor recurrence. The use of different chemotherapy regimens (cisplatin/ 5-fluorouracil or carboplatin and paclitaxel) in our study should not affect results, because two retrospective trials did show significant differences regarding oncologic outcome between those two regimens for E-SCC patients undergoing dCRT or nCRT + S [22, 23].

As it was mentioned before, our study has some limitations. One important limitation of this study is its retrospective nature. Also the moderate imbalances regarding tumor parameters between the patient cohorts (rate of lymph node metastases and tumor location) might affect the results and should be kept in mind. However, we also want to mention some strengths of our study. We only included patients with a curative treatment approach and who received complete treatment. That means that patients in the nCRT + S group had to receive at least 41.4 Gy radiation dose and patients within the dCRT group underwent radiation therapy with at least 50 Gy. In addition, all patients received simultaneous chemotherapy.

In E-SCC patients treated with either dCRT or nCRT + S, a higher rate of local tumor recurrence was seen in patients treated with dCRT than in patients treated with nCRT + S. There was at least a trend towards an improved OS and PFS in patients undergoing nCRT + S. However, this should be interpreted with caution, because proximal tumor location was the only parameter independently affecting the risk of local tumor recurrence.