Presented at the Western Thoracic Surgical Association, Victoria, Canada, June 22–25, 2005.
Tumor viability assessed by pathologic analysis of resected specimens in patients with preoperatively treated esophageal adenocarcinoma (EAC) is a prognostic indicator. The feasibility of induction chemotherapy followed by concurrent chemoradiotherapy (CCRT) and surgery for patients with locoregionally advanced EAC has been demonstrated. In this study, the authors evaluated the efficacy of CCRT compared with traditional concurrent chemoradiotherapy (CRT).
The authors retrospectively reviewed 247 consecutive patients with EAC who presented for planned surgery after treatment with either CCRT or CRT from January 1997 through August 2003. Patient demographics, comorbidities, and tumor characteristics were analyzed. Pathologic tumor response, overall survival, and disease-free survival were assessed according to treatment.
One hundred seventeen patients received CCRT, and 130 patients received CRT before planned surgical resection. CCRT resulted in a 64% tumor response rate compared with a 51% tumor response rate in the CRT group (odds ratio, 1.73; P = .035). In the CCRT group, the median overall survival was 55 months, and the 3-year overall survival rate was 59%; in the CRT group, the median overall survival was 25 months, and the 3-year overall survival rate was 41% (hazard ratio [HR], 0.69; P = .041). In the CCRT group, the median disease-free survival was 43 months, and the 3-year disease-free survival rate was 54%; in the CRT group, the median disease-free survival was 18 months, and the 3-year disease-free survival rate was 36% (HR, 0.72; P = .047). Subset analysis of patients with clinical Stage III/IVA disease showed a median overall survival of 51 months with a 3-year overall survival rate of 58% in the CCRT group and a median overall survival of 20 months with a 3-year overall survival rate of 28% in the CRT group (HR, 0.57; P = .019).
Multimodality strategies have been evaluated in patients with locoregionally advanced esophageal cancer in an effort to improve the outcomes achieved with surgery alone.1 Theoretical benefits of neoadjuvant therapy include improved resection rates, pathologic down-staging, and a reduction in disease recurrence. Although the results have been controversial, several retrospective studies2–4 and 2 prospective, randomized trials5, 6 have suggested improved survival associated with neoadjuvant therapy followed by surgery, especially in patients who demonstrate a complete pathologic response to preoperative chemoradiation.7, 8 More recently, we demonstrated that a pathologic partial response (P1) (from 1% to 50% residual carcinoma) also is associated with improved outcomes9 and that the extent of pathologic response to chemoradiation is an independent risk factor for survival.10 Therefore, we reviewed our experience with induction chemotherapy followed by concurrent chemoradiation (CCRT) and surgery compared with traditional preoperative, concurrent chemoradiation (CRT) and surgery to determine whether this additional chemotherapy resulted in improved pathologic response and long-term outcome.
MATERIALS AND METHODS
From a prospectively maintained data base in the Department of Thoracic and Cardiovascular Surgery at The University of Texas M. D. Anderson Cancer Center, a retrospective analysis identified 491 patients who underwent planned esophageal resection from January 1997 through August 2003. Of these 491 patients, we identified 247 patients who had primary esophageal adenocarcinoma (including Type III gastroesophageal cancer) and who received preoperative chemoradiation prior to undergoing surgery. This cohort of patients was analyzed according to 2 treatment groups. The first group included patients who received standard preoperative CRT. The second group included patients who received CCRT. Informed consent was obtained prior to data collection, and the study was performed with approval from The University of Texas M. D. Anderson Cancer Center Institutional Review Board.
All patients had adenocarcinoma that was confirmed with esophagoscopy and biopsy. Endoscopic ultrasound (EUS) studies and computed tomography (CT) scans of the chest and abdomen were obtained to determine pretreatment clinical stage. All patients (100%) in the CCRT group and 106 patients (82%) in the CRT group were staged with pretreatment EUS. More recently, some patients were staged with positron-emission tomography (34 patients [29%] in the CCRT group and 31 patients [24%] in the CRT group). A multidisciplinary team (thoracic surgeon, medical oncologist, radiation oncologist, gastroenterologist, and radiologist) assessed all patients to determine tumor resectability and physiologic ability to tolerate preoperative therapy and surgery.
Of 117 patients who received CCRT, 101 patients were enrolled in institutional protocols. The inclusion criteria for these patients were clinical Stage T2 or T3, N0 or N1, and M0-1a esophageal adenocarcinoma. The remaining 16 patients were treated off protocol but received induction chemotherapy similar to that received by the patients on protocol. Six patients (5%) who received CRT were treated on protocol. In total, 113 patients (97%) in the CCRT group were treated on site at The University of Texas M. D. Anderson Cancer Center, compared with 72 patients (55%) in the CRT group (Table 1). In the CCRT group, 115 patients (98%) received ≥45 grays (Gy) of radiotherapy, and 2 patients received <45 Gy. In the CRT group, 105 patients (95%) received ≥45 Gy, 7 patients received <45 Gy, and 18 patients received an unknown dosage (they were treated at another institution).
Table 1. Pretreatment Characteristics according to Preoperative Chemoradiotherapy Sequence
CRT indicates chemoradiotherapy; C→CRT, chemotherapy followed by CRT; GEJ, gastroesophageal junction; CAD, coronary artery disease; COPD, chronic obstructive pulmonary disease; ASA, American Society of Anesthesiologists.
Preoperative, concurrent CRT.
Induction chemotherapy followed by preoperative, concurrent CRT.
The P value was calculated from mean ages.
Prior cancer excluded basal cell and squamous cell skin cancer.
Patients in the CCRT group received induction chemotherapy followed by concurrent chemoradiotherapy. During the study period, separate Phase II trials were performed to evaluate the additional benefit of induction chemotherapy and preoperative, concurrent chemoradiotherapy. In the first protocol,11 patients received up to 2 courses of induction chemotherapy, which consisted of 5-fluorouracil (750 mg/m2 per day) as a continuous infusion on Days 1 through 5, cisplatin (15 mg/m2 per day) as an intravenous bolus on Days 1 through 5, and paclitaxel (200 mg/m2) as a 24-hour intravenous infusion on Day 1. The second course was repeated on Day 29. This was followed by radiotherapy (45 Gy in 25 fractions) and concurrent administration of 5-fluorouracil (300 mg/m2 per day as a continuous infusion 5 days per week) and cisplatin (20 mg/m2 on Days 1 through 5 of radiotherapy). In the second protocol,12 patients received up to 2 courses of CPT-11 (45 mg/m2), taxotere (33 mg/m2), and 5-fluorouracil (2 g/m2 as a 24-hour infusion). This was followed by concurrent radiotherapy (50.4 Gy in 28 fractions) and CPT-11 (30 mg/m2 per week for 5 weeks), taxotere (20 mg/m2 per week for 5 weeks), and 5-fluorouracil (300 mg/m2 per day). Patients in the CRT group received planned chemoradiotherapy, which consisted of radiotherapy up to 50.4 Gy, and concurrent chemotherapy (a doublet of either cisplatin and 5-fluorouracil or taxol and carboplatin).
Operative resection was performed approximately 3 to 6 weeks after the end of the preoperative regimen. Patients who underwent a salvage esophagectomy13 for disease recurrence after failed definitive primary chemoradiotherapy were excluded from this analysis. Three operative approaches were used. The Ivor–Lewis transthoracic esophagectomy included a 2-field (mediastinal and abdominal) lymphadenectomy. The McKeown esophagectomy involved resection of the entire thoracic esophagus with a 2-field lymphadenectomy and cervical anastomosis. The transhiatal esophagectomy involved resection with abdominal lymphadenectomy and limited mediastinal lymph node resection. Finally, some patients were unresectable at the time of operation and underwent exploration only. All patients who underwent resection received a tube jejunostomy for temporary enteral nutrition.
Dedicated pathologists at The University of Texas M. D. Anderson Cancer Center analyzed all surgical specimens. We defined a pathologic complete response (P0) as complete histologic eradication of tumor cells in the primary tumor site with 0% residual carcinoma, a pathologic partial response (P1) was defined as residual carcinoma of 1% to 50%, and no response (P2) was defined as >50% residual carcinoma. Residual carcinoma was based on the estimated percentage of residual carcinoma in relation to total carcinoma area, including the amount of radiotherapy-induced tissue injury.9 Tumor response did not account for down-staging in the lymph nodes.
Patients were assessed at 1 month, 3 months, 6 months, and 12 months and every 6 months thereafter. The median potential follow-up was 50 months (range, 19–103 months).
Patient characteristics were compared by using an analysis of variance, the Pearson chi-square test, or the Fischer exact test, as appropriate. Perioperative mortality was defined to include operative, 30-day, and same-hospitalization mortality. Univariate logistic regression analysis was performed to identify the factors associated with tumor response. Survival distributions are displayed graphically using the Kaplan–Meier method. Differences in survival distributions were assessed using the log-rank test. Overall survival was calculated from the date of pathologic diagnosis to the date of death or the date of last follow-up and included perioperative mortality. Disease-free survival was calculated similarly from the date of diagnosis to the date of disease recurrence or death. Univariate Cox regression analysis, including only preoperative patient factors, was performed to identify factors associated with survival. A step-wise, variable-selection procedure was used to identify the variables associated most with survival in a multivariate context (P < .10). Two-tailed P values of ≤.05 were considered significant. Data analysis was performed by the departmental biostatistician (A.M.C.) using the Statistical Package for Social Sciences (version 184.108.40.206; SPSS Inc., Chicago, IL).
Demographics and Preoperative Characteristics
Two hundred forty-seven patients with esophageal adenocarcinoma were treated with preoperative chemoradiation followed by surgery. Of these patients, 130 received CRT, and 117 received CCRT. The 2 groups did not differ with respect to gender, race, tumor location, presence of Barrett mucosa, clinical stage, comorbidities, or performance status (Table 1). The median age in the CCRT group (58 years) was slightly younger than in the CRT group (63 years). In addition, there were more patients with prior malignancies in the CRT group (12%) than in the CCRT group (4%).
Although the percentage of patients who had a P0 response was 28% in both groups, more patients in the CCRT group had a P1 response than in the CRT group (36% in the CCRT group vs. 23% in the CRT group). Therefore, the total percentage of patients with clinically meaningful tumor response (P0 and P1) was significantly higher in the CCRT group compared with the CRT group (64% vs. 51%, respectively; P = .049). Fewer patients in the CCRT group, compared with the CRT group, were classified as unresectable at operation (Table 2). Postoperative morbidity, including anastomotic leaks, pneumonia, and atrial arrhythmias, did not differ between the 2 groups. In addition, the perioperative mortality was similar (4%) in both groups. The pattern of recurrence did not differ significantly between groups; 29% of patients in the CCRT group and 37% of patients in the CRT group developed distant recurrences (P = .54).
Table 2. Postoperative Characteristics According to Preoperative Chemoradiotherapy Sequence
The difference in overall survival and disease-free survival between the CCRT group and the CRT group was statistically significant. The median survival and the 3-year overall survival rate, which were computed by using the Kaplan–Meier method (Fig. 1), were 55 months and 59%, respectively, in the CCRT group, compared with 25 months and 41%, respectively, in the CRT group (hazard ratio, 0.69; P = .041). The median survival and the 3-year disease-free survival rate (Fig. 2) were 43 months and 54%, respectively, in the CCRT group and 18 months and 36%, respectively, in the CRT group (hazard ratio, 0.72; P = .047).
Predictive Factors of Tumor Response and Survival
Logistic regression analysis showed that only the use of CCRT was associated with improved tumor response in univariate analyses (odds ratio, 1.73; P = .035 in patients who received CCRT). Factors like age, gender, race, presence of Barrett mucosa, clinical staging, comorbidities, prior cancers, site of induction therapy, and inclusion in institutional protocols were not associated with tumor response (Table 3). In addition, multivariate analysis that specifically included relevant factors, such as age, prior cancer, site of induction therapy, and inclusion in institutional protocols, did not demonstrate that these factors were predictive for tumor response. Stepwise, multivariate regression analysis showed that only the use of CCRT was predictive of tumor response.
Table 3. Univariable Logistic Regression Analysis of Pretreatment Risk Factors for Response to Preoperative Chemoradiotherapy
OR indicates odds ratio; 95% CI, 95% confidence interval; GEJ, gastroesophageal junction; CAD, coronary artery disease; COPD, chronic obstructive pulmonary disease; ASA, American Society of Anesthesiologists; CRT, chemoradiotherapy; C→CRT, chemotherapy followed by CRT.
Performance status was defined according to ASA criteria.
Preoperative, concurrent CRT.
Induction chemotherapy followed by preoperative, concurrent CRT.
Cox regression analyses were performed to identify factors that may predict survival independently. Univariate analysis demonstrated that 4 pretreatment factors were associated with survival: the presence ofcoronary artery disease, prior cancer, American Society of Anesthesiologists (ASA) classification, and the use of CCRT. Stepwise multivariate Cox regression demonstrated that the presence of coronary artery disease and ASA classification were the only independent risk factors for survival in the current analysis.
Survival by Clinical Stage
Subgroup analysis was performed to explore the benefit of induction therapy according to clinical stage. Patients were analyzed according to localized disease (clinical Stage II) versus locoregionally advanced disease (clinical Stage III and IVA). A Kaplan–Meier analysis of patients who had localized disease showed a median survival of 59 months and a median 3-year survival rate of 60% in the CCRT group versus 37 months and 54%, respectively, in the CRT group (P = .61) (Fig. 3). Patients in the locoregionally advanced disease subgroup had a median survival of 51 months and a 3-year survival rate of 58% in the CCRT group versus 20 months and 28%, respectively, in the CRT group (P = .019) (Fig. 4).
The current retrospective analysis demonstrated that patients who received induction chemotherapy prior to neoadjuvant concurrent chemoradiotherapy (CCRT) had a significantly improved local tumor response compared with patients who received concurrent chemoradiotherapy (CRT) alone. To date, 3 prospective, Phase II studies involving induction chemotherapy followed by concurrent chemoradiotherapy and surgery in patients with resectable esophageal adenocarcinoma have been performed at our institution (2 published articles and 1 pending publication).11, 12 The rate of P0 response ranged from 28% to 30% in those trials. The conclusions reached in those trials were that the 3-step strategy was feasible and resulted in high response and resection rates with tolerable morbidity, and patients who responded completely had a favorable survival outcome. Noted by the investigators but not reported in the previous trials was the observation that the addition of induction therapy to the platinum/5-fluorouracil based, concurrent chemoradiotherapy seemed to increase the overall response rate (P0 and P1) compared with previous Phase II trials that featured CRT alone. Furthermore, patients who presented with a more advanced clinical stage responded better to the additional therapy compared with similar historic patients who received CRT alone. To our knowledge, this is the largest study associating the addition of induction chemotherapy to improved tumor response and clinical outcome.
The overall benefit of preoperative chemoradiation in the treatment of esophageal cancer remains controversial; however, several previous publications have substantiated the observation that a P0 response to neoadjuvant therapy within an individual patient is associated with improved survival.7, 8, 10, 14 Recently, we published a study9 that analyzed the outcomes of patients who received neoadjuvant chemoradiotherapy and who had their tumors down-staged significantly at the time of resection. Tumor response in that study was classified according to residual histologic viability of the tumor in the pathologic specimen (P0, 0% residual; P1, 1–50% residual; P2, >50% residual). An important conclusion was that the stratification of patient survival based on posttreatment pathologic stage verified the observation that clinically significant down-staging of esophageal cancer can occur with neoadjuvant therapy. Achieving a partial (P1) or complete (P0) response to therapy was identified as an independent factor for predicting survival.10 In the current study, the rate of P1 response in the CCRT group appears to be associated with a survival benefit, and this is consistent with our previous findings.
To explore which patients benefited most from the use of induction chemotherapy, a subset analysis was performed according to clinical stage. The patients with clinical Stage II disease had similar survival whether they received CCRT or CRT. However, patients with locoregionally advanced, clinical Stage III and IVA disease who received induction chemotherapy had a significant survival advantage over patients of similar stage who received CRT alone. There are several possible explanations for this observation. First, the additional use of induction chemotherapy prior to CRT may benefit patients with larger tumor burden and/or relatively resistant neoplasms and help to achieve pathologic down-staging. Another explanation is that delivery of intensive, systemic therapy with CCRT may help to treat micrometastatic disease and reduce the rate of systemic recurrence. Although the data from our study on distant recurrence did not reach statistical significance (29% in the CCRT group and 37% in the CRT group), the potential difference may warrant further study. Development of novel neoadjuvant regimens should have the objectives both of improving tumor response and of controlling the rate of systemic recurrence. In addition, the use of biologic markers to identify patients who would benefit from CCRT is another potential strategy for individualized therapy.15
It is noteworthy that mortality was not affected by the addition of induction chemotherapy, nor was there an increase in postoperative complications. The rate of anastomotic leaks, pneumonia, and atrial arrhythmias were comparable in both groups. Our results did not demonstrate an increase in anastomotic leaks in the CCRT group (7%) compared with the CRT group (9%). In contrast, the results of a Phase II trial16 performed at another institution reported a leak rate of 19% associated with CCRT. The concern of increased operative morbidity and mortality because of intensive chemotherapy was not demonstrated in this study. The most common toxicities from CCRT were granulocytopenia and esophagitis, as reported from previous Phase II trials. Other toxicities included fatigue, diarrhea, nausea and emesis, neuropathy myalgia, neutropenic fever, anorexia, mucositis, hyperglycemia, and dermatitis. Although patients endured toxicity from the induction chemotherapy in addition to that from the concurrent chemoradiation, no deaths related to the CCRT regimen were reported. Combined results of these 2 Phase II trials indicate that 74 of 81 patients (91%) who received CCRT underwent resection of their cancer. This compares favorably with a similar Phase II trial at other institutions, which reported a 74% resection rate.17 Unfortunately, no comparison can be made in the current study about the resection rate, because the overall number of patients who received CRT (with or without surgery) is unknown. Also noted is the low number of unresectable patients in the CCRT group versus the CRT group. It is possible that additional induction therapy may allow some patients to undergo an R0 resection that otherwise may not have been resectable. Because we can compare overall resection rates only with historic data, this question would be answered best in a randomized, controlled fashion.
It is important to recognize the limitations of the current study. First, this was a retrospective review, and there are inherent difficulties in comparing groups of patients in this manner. There were differences in pretreatment staging seen in the number of patients that had an EUS in the 2 groups; however, postoperative evaluation of local tumor response is an indirect measure of treatment rather than preoperative staging. A true and complete denominator on an intent-to-treat basis is not known accurately, because many of these patients received treatment off protocol. Moreover, many received their neoadjuvant therapy at another institution and were then sent to our institution after they completed induction therapy. Although the dose of radiotherapy was similar whether the patient was treated at The University of Texas M. D. Anderson Cancer Center or at another institution, other differences in fractionation or radiation fields and differences in total doses of administered versus planned chemotherapy are unknown and may have resulted in differing tumor response in patients who were treated at multiple centers. Another bias is the possibility of improved outcomes for patients on study protocols,18 which formed the majority of patients in the CCRT group. The effects of such biases can be minimized only in a prospective, randomized trial. Arguably, the comparison of these groups of patients leads the reader to consider the conclusions very carefully. It is for these reasons that we have chosen to emphasize the observed histologic tumor response, which is a direct measure of cancer cell death within the tumor, rather than overall survival, an end point that clearly is affected by multiple factors.
In conclusion, induction chemotherapy prior to concurrent chemoradiotherapy resulted in an increased tumor response (P0 and P1) compared with traditional concurrent chemoradiotherapy in patients with esophageal adenocarcinoma. The improved tumor response may be associated with longer overall survival and disease-free survival in patients who received CCRT, especially in patients with locoregionally advanced esophageal cancer. Whether improvements in pathologic response and survival justify the potential increased toxicity and cost of additional chemotherapy in CCRT is being evaluated currently at our institution in a randomized Phase II study. The results of that study will allow the selection of a standard treatment regimen (CCRT vs. CRT) for patients with locoregionally advanced esophageal cancer to which novel biologic therapies ultimately may be added in the future.