Comparison of definite chemoradiation therapy with carboplatin/paclitaxel or cisplatin/5-fluoruracil in patients with squamous cell carcinoma of the esophagus

Esophageal cancer is the eighth most common cancer worldwide and in the USA alone almost 17.000 new cases were estimated for 2017 [1, 2]. According to its histology, esophageal cancer is divided into squamous cell carcinoma (SCC) and adenocarcinoma (AC) and epidemiology as well as therapy approaches differ between both subtypes [3, 4].

For patients with locally advanced or node-positive SCCs neoadjuvant chemoradiation therapy (nCRT) with subsequent surgery can improve overall survival (OS) and progression-free survival (PFS) compared with surgery alone [5‐7] and therefore has become the treatment of choice for these patients [4, 8]. However, patients who are not suitable for surgery due to medical or technical reasons or patients who refuse surgery should be treated with definite chemoradiation therapy (dCRT).

nCRT is typically done with either cisplatin and 5-fluorouracil (CDDP/5FU) analogous to the older CALGB-trial [6] or with carboplatin and paclitaxel (Carb/TAX) according to the practice-changing CROSS-trial [5]. Although two retrospective analyses found no significant difference for OS and treatment response between both chemotherapy regimens, nCRT with CDDP/5FU was associated with a significantly increased rate of myelotoxicity [9, 10]. In contrast to the neoadjuvant situation, dCRT with Carb/TAX has only been evaluated in some small, retrospective analyses [11‐13]. However, both chemotherapy combinations are listed as preferred regimens for dCRT by the National Comprehensive Cancer Network (NCCN) guideline [8]. Honing and colleagues performed a retrospective comparison of CDDP/5FU and Carb/TAX for dCRT in patients with esophageal cancer (EC) [14]. In this study, the authors demonstrated comparable results regarding OS and disease-free survival (DFS) for both chemotherapy regimens, but significantly lower rates of hematologic and non-hematologic toxicity in patients receiving concomitant chemotherapy with Carb/TAX. Importantly, half of the patients in this study were diagnosed with AC which – based on the fundamentally different biology of these neoplasms when compared to SCC- might affect the results, although histology was neither in the univariate nor in multivariate analysis an independent factor for OS or DFS. In addition, the median radiation dose was 50.4 Gray (Gy), which is a relatively low dose. In contrast to the North American guidelines, the German S3-Guideline recommends higher irradiation doses of 50-60Gy for dCRT [4].

At our department, dCRT for SCC patients is routinely administered with Carb/TAX since 2014, while previously treated patients received dCRT with CDDP/5FU. In this study, we compared efficiency and toxicity of dCRT with ≥54Gy and either CDDP/5FU or Carb/TAX for patients with SCC.

In this study no significant differences regarding treatment tolerance and oncologic outcome were found between dCRT with either CDDP/5FU or Carb/TAX in patients with squamous cell carcinoma of the esophagus.

While trimodal therapy with nCRT and subsequent surgery has been established as the treatment of choice for SCC patients suitable for surgery, dCRT is recommended for patients unsuitable for surgery due to technical or medical reasons or patients refusal of surgery [4, 8]. Most data in this scenario are derived from studies using concomitant chemotherapy with cisplatin and 5-fluoruracil [15‐17]. However, based on the encouraging results of some smaller trials [11‐13] and the CROSS-trial [5], which evaluated nCRT with carboplatin and paclitaxel, one might speculate that Carb/TAX might also be used effectively for dCRT.

Our SCC patient cohort revealed no significant differences in OS and FFR between both treatment groups. This result is in line with the results of Honing and colleagues [14], who compared dCRT with CDDP/5FU or Carb/TAX in patients with AC and SCC of the esophagus. While OS of patients treated with CDDP/5FU in our analysis was longer than in the analysis by Honing et al. (median OS 24.2 months vs. 16.1 months), results for patients treated with Carb/TAX are difficult to compare. In our analysis median OS was not reached after a median follow-up time of 18.2 months for surviving patients, while Honing et al. reported a median OS of 13.8 months for patients treated with Carb/TAX. A possible explanation for this difference is the higher rate of patients with distant lymph node metastases (M1a) in Honing’s study, who might have a shorter OS than patients without distant metastases (M0). While distant lymph node metastases were seen in 23% (CDDP/5FU) and 9% (Carb/TAX) of patients in the analysis by Honing et al. [14], in our study only one patient (4%), who was scheduled for CDDP/5FU had distant lymph node metastases. In addition, patients included in the present analysis received higher total radiation doses (median total radiation dose: 54Gy vs. 50.4Gy (CDDP/5FU) and 59.4Gy vs. 50.4Gy (Carb/TAX)). Even this relatively small difference in radiation dose might improve treatment effectiveness and therefore might explain differences in OS. In a retrospective trial by Kim et al. [18], high-dose dCRT (median dose 63Gy) was associated with increased OS and loco-regional control than standard-dose dCRT (median dose 50.4Gy). However, it remains an open question if these results are also applicable to lower total dose differences as seen in the present study.

While OS of patients treated with Carb/TAX in our analysis was also higher than in two other studies evaluating dCRT with Carb/TAX in EC patients [12, 13], it was comparable to the results by Ruppert et al. [11]. In their analysis, patients with locally advanced EC, treated with dCRT and Carb/TAX had a 3-year OS of 56.1%. However, in this study patients also received two cycles of additional chemotherapy after dCRT. Regarding patients treated with CDDP/5FU median OS in our study was longer than in the analyses by Herskovic et al. [15] (12.5 months) and Minsky et al. [17] (18.1 months). But in contrast to our study, both analyses used former (2D and 3D) radiation techniques and a lower radiation dose of 50.4Gy.

Regarding freedom from relapse (FFR), we found no other study analyzing FFR in EC patients treated with dCRT with CDDP/5FU. However, when comparing FFR to DFS and PFS our results for patients receiving CDDP/5FU seem to be comparable to the results of Honing et al. [14] (median FFR of 12.1 months vs. median DFS of 11.1 months). Analogous to the results for OS, results for FFR in our trial are different to compare with other trials because median FFR was not reached. However, 2y-FFR in our trial (57%) is comparable to the results by Ruppert et al. [11] (2y-overall FFR 51.3%). In line with the results by Honing and colleagues [14] no significant difference was seen between both treatment groups.

A total of 48 and 20% of patients treated with CDDP/5FU had loco-regional or distant treatment failure. This is in line with data from Herskovic et al. [15]. In their analysis, patients with EC were treated with 50.4Gy and concomitant chemotherapy with CDDP/5FU. After 12 months loco-regional and distant tumor recurrence was seen in approximately 40 and 22% of patients. In our patients treated with Carb/TAX loco-regional or distant treatment failure was seen in 36 and 27% of patients, respectively. In contrast to that two other trials investigating dCRT with Carb/TAX revealed loco-regional recurrence in 42% of patients [12, 13]. However, in both studies median follow-up was longer than in our study, which might explain the higher rate of loco-regional recurrences. This difference in the median follow-up might also explain the difference in the rate of distant recurrences between our study (27%) and the analysis by Haj Mohammad et al. [12] (42%), in which EC patients were treated with dCRT with Carb/TAX.

In contrast to the results of Honing and colleagues [14] our data revealed no significant differences in terms of myelotoxicity between both treatment groups. Compared to the results presented by Honing et al., the rate of ≥ III° myelotoxicity in our study was higher for both, patients treated with CDDP/5FU (52% vs. 19%) and patients treated with Carb/TAX (55% vs. 4%). This difference is remarkable, because especially for patients treated with Carb/TAX the only noticeable difference is seen for radiation dose (median dose 50.4Gy vs. 59.4Gy). It seems likely, that in this case not only the planning target volume, but also the bone marrow will receive higher doses, which might increase the risk of myelotoxicity. In a study by Noronha et al. [13] ≥ III° leukopenia was seen in 49% of patients treated with dCRT with a median radiation dose of 58.7Gy, which is comparable to our results, but we have to point out that almost 42% of patients in their study received induction chemotherapy. Although the rate of ≥ III° myelotoxicity for patients treated with CDDP/5FU in our analysis is much higher than in the study by Honing et al., it is comparable to other studies. After dCRT with 50.4Gy and concomitant chemotherapy with CDDP/5FU severe or worse myelotoxicity was seen in 48% of patients in a study by Herskovic and colleagues [15]. In addition, ≥ III° myelotoxicity was seen in 58% of patients receiving nCRT with CDDP/5FU [10]. While radiation dose was only 45Gy in this analysis, patients received the same amount of chemotherapy. An overview of different studies evaluating overall survival, rate of recurrence and myelotoxicity of different dCRT regimens for SCC is shown in Table 3.

While 16% of patients who received CDDP/5FU were irradiated using 3D-CRT, all patients who received Carb/TAX were irradiated using VMAT, but this difference was not statistically significant (p = 0.112). In two retrospective analyses no significant differences in terms of oncologic outcome were seen between 3D-CRT and VMAT for EC patients undergoing nCRT or dCRT [19, 20]. However, there was a higher rate of leukopenia in patients undergoing nCRT with VMAT compared to 3D-CRT [19]. Considering the lower radiation dose (45Gy) it remains unclear if this result can be transferred to patients undergoing dCRT.

While the question of our study was similar to the study published by Honing et al. [14], with both studies comparing dCRT with either CDDP/5FU or Carb/TAX in patients with esophageal cancer, there are also some relevant differences. In contrast to our study, Honing and colleagues included patients with different histologic subtypes (adenocarcinoma and squamous cell carcinoma). This has to be mentioned, because chemoradiation is more effective in patients with squamous cell carcinoma than in patients with adenocarcinoma. In addition, patients included in the analysis by Honing et al. were treated with a lower radiation dose of 50.4Gy. Because the German S3-Guideline [4] recommends higher radiation doses of up to 60Gy, results presented by Honing and colleagues probably don’t correspond to the treatment regimens used in the daily routine in Germany.

By only including patients with squamous cell carcinoma of the esophagus, our study can exclude that oncologic outcome is biased by different histology subtypes. In addition, results of our study demonstrate, that the therapeutic equivalence of both chemotherapy regimens can be presumed for treatment concepts with higher total radiation doses.

Because of its retrospective nature this study has some limitations. The most obvious problem is the small number of patients, which clearly compromises the power of the study. On the other side, the small number of patients is also caused be the fact, that we specifically compared dCRT with CDDP/5FU or Carb/TAX only in patients with SCC, which clearly is the biologically more rational approach. A further limitation is the difference regarding the follow-up time. Follow-up of patients treated with CDDP/5FU is clearly longer than follow-up of patients treated with Carb/TAX. The reason for this imbalance is that all patients who received CDDP/5FU were treated between 2011 and 2014 while the first patient who received Carb/TAX was treated at the end of 2014. We have to keep in mind that this difference in follow-up time might contribute to differences in survival and rate of recurrences. To consider the short follow-up of patients treated with Carb/TAX, we also compared one-year OS, one-year FFR and rate of recurrences within one year between both treatment groups and found no significant differences. The last limitation is the difference in the total radiation dose. While patients in the CDDP/5FU group received a median radiation dose of 54Gy, patients in the Carb/TAX group received a median radiation dose of 59.4Gy. As explained before this dose difference might also lead to an increased tumor control probability in patients treated with Carb/TAX. In addition, we cannot rule out that even this dose difference might also impair myelotoxicity. However, both doses are within the recommended dose range for dCRT.

Myelotoxicity and oncologic outcome under dCRT were not different for patients with SCC of the esophagus treated with either CDDP/5FU or Carb/TAX. The putative equivalence of dCRT with Carb/TAX in this setting should be further investigated in prospective trials. However, our data reveal that the risk of significant myelotoxicity increases with patient age and therefore other chemotherapy regimens might be evaluated in elderly.