Chemoradiation Followed by Surgery Compared With Chemoradiation Alone in Squamous Cancer of the Esophagus: FFCD 9102

Laurent Bedenne Journal of Clinical Oncology, Vol 25, No 10 (April 1), 2007: pp. 1160-1168

Patients and Methods: Eligible patients had operable T3N0-1M0 thoracic esophageal cancer. Patients received two cycles of fluorouracil (FU) and cisplatin (days 1 to 5 and 22 to 26) and either conventional (46 Gy in 4.5 weeks) or split-course (15 Gy, days 1 to 5 and 22 to 26) concomitant radiotherapy. Patients with response and no contraindication to either treatment were randomly assigned to surgery (arm A) or continuation of chemoradiation (arm B; three cycles of FU/cisplatin and either conventional [20 Gy] or split-course [15 Gy] radiotherapy). Chemoradiation was considered equivalent to surgery if the difference in 2-year survival rate was less than 10%.

Radiotherapy included the macroscopic tumor and lymph nodes, with a 3-cm proximal and distal margin and a 2-cm radial margin. The use of three or four fields and daily treatment of all fields were required. Initially, the following two techniques were allowed, according to the initial choice of the investigator for all of his or her included patients: split-course or conventional radiotherapy. Split-course radiotherapy was delivered in daily fractions of 3 Gy, including two sequences (days 1 to 5 and 22 to 26; 30 Gy) before random assignment and one sequence (days 43 to 47; 15 Gy) after random assignment in arm B (total, 45 Gy). Each sequence was separated by a 2-week rest period. Conventional radiotherapy was delivered in 5 daily fractions per week of 2 Gy during the 4.5 weeks before random assignment (46 Gy) and the 2 weeks after random assignment (20 Gy), for a total of 66 Gy. Beginning January 1999, an amendment based on the results of a randomized study permitted only conventional radiotherapy.

Results: Of 444 eligible patients, 259 were randomly assigned; 230 patients (88.8%) had epidermoid cancer, and 29 (11.2%) had glandular carcinoma. Two-year survival rate was 34% in arm A versus 40% in arm B . Median survival time was 17.7 months in arm A compared with 19.3 months in arm B. Two-year local control rate was 66.4% in arm A compared with 57.0% in arm B, and stents were less required in the surgery arm (5% in arm A v 32% in arm B; P < .001). The 3-month mortality rate was 9.3% in arm A compared with 0.8% in arm B (P = .002). Cumulative hospital stay was 68 days in arm A compared with 52 days in arm B (P = .02).

Conclusion: Our data suggest that, in patients with locally advanced thoracic esophageal cancers, especially epidermoid, who respond to chemoradiation, there is no benefit for the addition of surgery after chemoradiation compared with the continuation of additional chemoradiation.

Our results suggest that chemoradiation alone and chemoradiation followed by surgery are equivalent in terms of survival and quality of life in responders. Results were given from the start of treatment and not from random assignment because, although the differences were nearly the same, the results from the start of treatment better reflected overall survival. Indeed, the treatment administered before random assignment lasted for more than 1 month. Our study results are consistent with the results from the study by Stahl  in which 172 patients with epidermoid esophageal cancer were randomly assigned to either chemoradiation with surgery or chemoradiation without surgery. Median survival time was 16.4 months with surgery compared with 14.9 months without surgery, and 2-year survival rates were 39.9% and 35.4%, respectively. As in our study, freedom from local progression was longer in the surgery group versus the no surgery group (at 2 years, 64.3% v 40.7%). If the patients responding to induction chemotherapy were considered, 3-year survival rates were 58% and 55% in the surgery and no surgery groups, respectively.In the FFCD 9102 study, random assignment was not performed at registration to test the efficacy and tolerance of chemoradiation and, hence, avoid cross over or continuation of an inefficient therapy. A smaller than expected percentage of patients was randomly assigned (57% instead of 75%). The rates of 71% and 87% for complete clinical response, which we based our calculations on, were drawn from phase II studies that included tumors that were not always locally advanced. Moreover,14 patients refusing surgery and 10 patients not fit for surgery were not randomly assigned. Thus, 24 more patients were eligible for chemoradiation (ie, 64%).

A significant difference in therapeutic mortality was observed, and one could consider that chemoradiation increased the postoperative mortality rate, hence undercutting the benefit of surgery. However, a significantly higher operative mortality rate was reported only in two randomized studies comparing preoperative chemoradiation with surgery alone (9% v 4%, respectively, for Walsh; 12% v 4%, respectively for Bosset ). In the latter study, the high dose per fraction (3.7 Gy) was probably responsible. In contrast, Le Prise Urba  and Burmeister  observed similar mortality rates in both the chemoradiation and surgery arms (9% v 7%, 2% v 4%, and 5% v 6%, respectively). Conversely, the benefit of chemoradiation may have been undercut in the first period of the trial by split-course chemoradiation, which was later demonstrated to be inferior to conventional protraction in a randomized study.

The dose of 66 Gy used in our trial seems excessive considering the conclusion of the INT 0123 study that a dose of 64.8 Gy is not superior to 50.4 Gy. However, our study design was different and permitted the delivery of three cycles of concomitant chemoradiation instead of two cycles, as in the INT 0123 trial. Regarding adjuvant chemotherapy, although previous studies were negative, the Medical Research Council OE 02 trial concluded that two preoperative cycles of FU plus cisplatin resulted in a better survival than surgery alone without increasing operative mortality.This raises the question of which of the following is the optimal preoperative treatment: chemotherapy or chemoradiation. Actually, several trials testing preoperative chemoradiation versus surgery showed a trend favoring chemoradiation,and in a recent series, preoperative chemoradiation was predictive of R0 resection, which was a positive prognostic factor. Meta-analyses suggest that preoperative chemoradiation improves 3-year survival and decreases locoregional recurrence rate, although no such beneficial effects are observed after preoperative chemotherapy. However, it is difficult to conclude about the best neoadjuvant treatment because another meta-analysis demonstrated opposite results concerning 2-year survival. In our study, no specific type of surgery was proposed, and this could have produced heterogeneity. However, 94% of the patients had transthoracic esophagectomies, and 4% had transhiatal operation. Moreover, randomized studies or meta-analyses have not demonstrated the superiority of one technique.

In this study, chemoradiation alone prevented 46% of the patients from having high-grade dysphagia until death, compared with previously reported rates of 60% to 67%.Nevertheless, dysphagia was better improved after surgery (63% mild or absent before death).

In conclusion, this study suggests that therapeutic strategies with or without surgery result in similar survival rates for locally advanced thoracic esophageal cancer patients responding to chemoradiation. This study applies especially to patients with epidermoid tumors, who represented almost 90% of the patients, although no difference with adenocarcinomas was observed in multivariate analysis. However, chemoradiation alone entailed fewer early deaths and a shorter hospital stay but more locoregional relapses. Because clinical prognostic factors do not help in choosing between both strategies, further studies comparing surgery and chemoradiation should search for new predictive factors and evaluate new tools to detect early responders. Positron emission tomography scan was reported to discriminate responders from nonresponders as early as 14 days after starting chemoradiation and should be re-evaluated in future studies.