Preoperative induction of CPT-11 and cisplatin chemotherapy followed by chemoradiotherapy in patients with locoregional carcinoma of the esophagus or gastroesophageal junction




Patients with localized esophageal carcinoma often develop locoregional and distant disease recurrence. The current study investigated the outcome of a new chemotherapy combination as induction therapy before chemoradiotherapy.


Forty-three patients with resectable carcinoma of the esophagus or gastroesophageal junction were enrolled. Most of the tumors were endoscopic ultrasonography (EUS) EUST3 (84%) and EUSN1 (63%). The patients received ≤ 2 6-week cycles of CPT-11 and cisplatin followed by chemoradiotherapy (45 grays with 5-fluorouracil and paclitaxel). Five to six weeks after chemoradiotherapy, the patients underwent staging and surgery. The feasibility, curative resection rates, overall and disease-free survival rates, rate of significant pathologic response, and patterns of disease recurrence were assessed.


Of the 43 patients, 39 (91%) underwent an R0 resection. Two patients (5%) died after surgery. A pathologic complete response (pathCR) was observed in 11 (28%) of the 39 patients (or 26% of the 43 patients). In addition, 16 patients (41% of 39 patients or 37% of 43 patients) had < 10% viable tumor in the surgical specimen (pathPR). A comparison of endoscopic ultrasonograpy T and N classifications with surgical T and N classifications demonstrated significant down-staging (P < 0.01). The median survival period of all 43 patients was 22.1 months. Patients who had achieved a pathCR or pathPR had a longer median survival (25.6 months) than those who achieved less than a pathPR (18.5 months; P = 0.52). None of the clinical parameters examined were found to correlate with survival or pathologic response.


CPT-11–based induction chemotherapy resulted in substantial pathCR and pathPR rates, both of which lead to a favorable survival outcome. The three-step strategy needs to be developed further, with the investigation of targeted therapies with chemotherapy and radiotherapy. Cancer 2004. © 2004 American Cancer Society.

Carcinoma of the esophagus is one of the most virulent malignancies with 5-year survival rates still < 20%.1 The American Cancer Society2 estimates that approximately 13,900 new cases of esophageal carcinoma were diagnosed in the U.S. in 2003.

Patients with locoregional carcinoma of the esophagus or the gastroesophageal junction (GEJ) often present with a T3 primary tumor, positive lymph nodes, or both. A Patterns of Care Study demonstrated an overwhelming preference for treating these patients with preoperative chemoradiotherapy.3 This practice pattern has only strengthened, despite the lack of uncontestable Phase III data to support it.4–6 The focus of several investigations has been the study of new drugs or strategies in Phase II trials.

In late 1998, the combination of CPT-11 and cisplatin was of particular interest,7, 8 with reported high response rates. In patients with upper gastrointestinal (GI) carcinomas, CPT-11 alone has been studied as chemoradiotherapy9 and in combination with cisplatin.10

The so-called “three-step strategy” was shown to be quite feasible in our first such trial.11 In the current trial, we investigated the combination of CPT-11 and cisplatin as induction chemotherapy administered before chemoradiotherapy in uniformly staged patients with locoregional esophageal carcinoma. In the chemoradiotherapy phase, we used a combination of weekly paclitaxel given concurrently with continuous infusion 5-fluorouracil (5-FU) based on previous favorable experience.12

We hypothesized that induction with CPT-11 plus cisplatin would reduce the bulk of the primary carcinoma (and possibly reduce or eliminate distant micrometastases). As a result, chemoradiotherapy would be more effective in increasing the fraction of patients with a substantial pathologic response in the resected specimen, improving the R0 resection rate, and reducing the local disease recurrence rate. All patients with resectable Siewert Type I/II13 tumors in the esophagus or GEJ underwent endoscopic ultrasonography (EUS) staging as well as routine clinical staging.


Patient Selection and Evaluation

All patients with localized and histologically confirmed adenocarcinoma or squamous cell carcinoma of the thoracic esophagus were eligible if they met the protocol eligibility criteria. Patients were evaluated by a chest radiograph, a computed tomography (CT) scan of the chest and abdomen, upper GI double-contrast barium radiographs, an esophagogastroduodenoscopy with EUS, and an electrocardiogram. In addition, levels of SMA-12 and electrolytes, a complete blood count including platelet count, and the serum baseline carcinoembryonic antigen level were monitored. Positron emission tomography (PET) was neither approved nor available at the study institution (M. D. Anderson Cancer Center, Houston, TX) at that time. Patients with T2–3 with any N were eligible as were patients with M1a cancer (celiac lymph nodes associated with a GEJ carcinoma). Patients with T1N1 carcinoma also were eligible. Before registration, all patients were evaluated by a thoracic oncology surgeon, a radiation oncologist, and a medical oncologist. All eligible patients had to have tumor that was considered technically resectable and medically operable based on the clinical staging and evaluation. Nutritional counseling was provided to a select group of patients on an as-needed basis. Written informed consent was obtained from all patients.

Patients with T4 disease and T1N0 lesions were excluded from the study. Patients with uncontrolled medical conditions (such as diabetes, hypertension, heart condition classified as New York Heart Association Class III/IV, or psychiatric illness) were not eligible. Patients who could not comprehend the purpose of the clinical trial or comply with its requirements also were excluded. Finally, patients with any evidence of metastatic disease were not included.

Study Design

The objective of the study was to determine the feasibility of induction chemotherapy with CPT-11 and cisplatin (two cycles maximum) followed by chemoradiotherapy with concurrent paclitaxel and 5-FU before surgery to remove the primary tumor and regional lymph nodes. A second cycle of induction chemotherapy was given if there was no disease progression. When local disease progression occurred, patients received the next step of chemoradiation therapy. If, at any time before the surgery, there was evidence of metastatic carcinoma, the patient was removed from the protocol therapy and offered palliative care. Laparoscopic staging with a placement of a J-tube was considered in selected patients.

If a patient received an R0 resection, no further therapy was planned. Patients who underwent an R1 resection (microscopic carcinoma at the margin) or an R2 resection (macroscopic carcinoma after surgery) or who had M1 disease were offered palliative care.

Step 1: chemotherapy

CPT-11 at a dose of 70 mg/m2 was administered as a 90-minute infusion by central venous access on Days 1, 7, 21, and 28. Cisplatin at a dose of 20 mg/m2 was administered as an intravenous (i.v.) bolus over 1 hour on Days 1, 7, 21, and 28. This chemotherapy cycle was repeated on Day 42, provided the patient had recovered from all toxic side effects (≤ Grade 1). Patients routinely received i.v. hydration and prophylactic antiemetic therapy.

The drug dose was reduced by 25% for Grade 3 nonhematologic toxicity or Grade 4 hematologic toxicity (according to National Cancer Institute Common Toxicity Criteria, Version 2.0). Complete blood counts were evaluated at least once. Serum creatinine, blood urea nitrogen, electrolyte, and magnesium levels were monitored regularly during each cycle.

Step 2: chemoradiotherapy

After completion of induction chemotherapy, patients began to receive chemoradiotherapy. The radiotherapy dose was 45 grays (Gy) (1.8 Gy per fraction). Patients underwent simulation by standard methods using esophagoscopy, esophagography, and CT scan results. Using megavoltage (≥ 6 MV) equipment, the 2-field to 4-field technique was applied. The superior and inferior borders of the field were 5 cm beyond the edges of the tumor whereas the lateral borders of the field were 2 cm beyond the edges.

Concurrently with radiotherapy, 5-FU was administered at a dose of 300 mg/m2 per day × 5 days each week of radiotherapy by continuous infusion. Typically, 5-FU was administered on a Monday and withdrawn on a Friday. Paclitaxel at a dose of 45 mg/m2 was administered i.v. over 3 hours (usually on Mondays; maximum 5 doses).

Step 3: surgery

There was no designated surgical procedure for patients in the current study. A transthoracic approach with regional lymphadenectomy was the most frequently performed surgery. A transhiatal approach also was used occasionally. During surgery, a jejunostomy feeding tube was placed for temporary postoperative nutritional support.

Response, Toxicity Criteria, and Data Management

Upper GI radiographs were performed after each 6-week cycle of induction chemotherapy and just before surgery. Esophagogastroduodenoscopy, CT scans of the abdomen and chest, and all blood tests were repeated before surgery. Some patients also received an EUS evaluation before surgery.

Response criteria included the following: 1) a pathologic complete response (pathCR), defined as the absence of carcinoma in the specimen; 2) a pathologic partial response (pathPR), defined as < 10% of tumor cells in the specimen; 3) a clinical complete response, defined as the absence of tumor after an endoscopy (visual or biopsy); and 4) a minor response, defined as the objective reduction in the evaluable lesion but less than that required for PR.14, 15 Clinical responses were used solely to identify patients who should receive the second cycle of chemotherapy.


Patients were assessed at 3 months, 6 months, and 12 months and then every 6 months until 5 years or death.

Statistical Methods

The study goal was to assess whether the rate of pathCR was ≥ 20% versus an alternative of 5%. A Simon two-stage design was used. In the first stage, 21 treated patients would be assessed if > 1 pathCR was observed, then 22 additional patients would be added. We estimated that 90% of registered patients would undergo surgery.

Patient responses (pathCR and pathCR+pathPR) were crosstabulated by gender, primary tumor location, and baseline stage. Baseline T classification data were given two categories, T1 or T2 (first category) or T3 (second category). Chi-square tests or Fisher exact tests were used to assess the association between each of these factors and response rates. Survival analyses were performed for overall as well as for disease-free survival (DFS) time. Overall survival time was defined as the time from the initiation of treatment to death, if both of these time points were recorded. If the death date was not available, then the date of last follow-up was used in its place. Data from patients who had not died by the time of analysis were counted as censored for the purpose of statistical analysis. DFS time was similarly defined as the time from the initiation of treatment to disease recurrence, or until the last follow-up date if the date of disease recurrence or death was not available. Data from patients who were alive without disease at the time of analysis were counted as censored. Log-rank tests were used to test for differences in survival distributions by gender, tumor location, baseline and postoperative T and N classifications, and down-staging with respect to N or T classification. Univariate Cox proportional hazards models were also fit, yielding estimates of hazard ratios for each of these factors. All statistical tests were two-sided and performed at the 0.05 significance level.


Forty-three patients were included in the current study. Patient characteristics are shown in Table 1. All patients were evaluated for toxicity, response, and survival.

Table 1. Characteristics of 43 Patients
CharacteristicsNo. of patients (%)
  1. GEJ: gastroesophageal junction; PS: performance status; EUS: endoscopic ultrasonography.

Primary sites 
 Lower third20
 squamous cell carcinoma6
Median age (yrs) (range)61 (39–73)
Median PS (range) (0–1) 1 (1)
Baseline EUS (n = 39) 
 N015 (38)
 N124 (62)
Baseline EUS (n = 39) 
 T1/2 6 (15)
 T333 (85)

EUS Results

All 43 patients in the study underwent EUS staging before the initiation of therapy. Thirty-six patients (84%) had a T3 carcinoma, 5 patients (12%) had a T2 carcinoma, and 2 patients (4%) had a T1 carcinoma. Twenty-seven patients (63%) had an N1 carcinoma and 16 patients (37%) had an N0 carcinoma (Table 1).

Response to Induction Chemotherapy

Some degree of definite objective response to chemotherapy was observed in 16 (37%) of the 43 evaluable patients. The carcinoma was unchanged in 24 (56%) patients and progressed in 3 (7%). One patient with stable disease refused a second cycle of chemotherapy and proceeded to chemoradiotherapy. Dysphagia and symptoms of pain associated with swallowing were relieved in 50% of patients.

Response to Chemoradiotherapy

The response to chemoradiotherapy was assessed 4–6 weeks after its completion by repeat barium swallow or endoscopy or both. Improvement was observed in all patients except the one patient who had increased narrowing of the esophagus.

Surgical Findings and Surgical Pathology

Of the 43 patients, 39 (91%) underwent surgery. Four patients did not undergo surgery because one developed metastatic carcinoma after the first cycle of induction chemotherapy, two developed liver metastasis, and one developed pulmonary metastasis at preoperative staging. An R0 resection was achieved in all 39 (100% of the patients who received surgery; 91% of the 43 registered patients).

A pathCR was observed in the surgical specimens of 11 patients (28% or 26% of 43 patients) and a pathPR was observed in the surgical specimens of 16 patients (41% or 37% of 43 patients). Therefore, an overall major pathologic response to therapy occurred in 69% (or 63% of 43 patients) of patients. The primary carcinoma was T3 in 15 (38%), T0 in 13 (33%), T2 in 8 (21%), and T1 in 3 (8%) of the 39 patients who underwent surgery. Twenty-nine (74%) of the 39 patients were N0 (no lymph node metastases) and 10 (26%) were N1. The median number of examined lymph nodes in the surgical specimens was 12 (range, 0–52 lymph nodes) and the median number of lymph nodes with carcinoma was 0 (range, 0–15 lymph nodes).

Correlation between EUS Stage and Pathologic Stage

The EUS and pathologic stages were correlated in the 39 patients who underwent surgery. These data are tabulated in Table 2.

Table 2. Correlation between EUS Staging and Pathologic Staging of 39 patients Undergoing Surgery
CharacteristicsEUS stagingPostsurgical pathologic stagingP value
  1. EUS: endoscopic ultrasonography.

No. of patients with “T3 carcinoma”3314< 0.01
No. of patients with “positive N1”2410< 0.01

Predictors of Pathologic Response (PathCR or PathPR) to Therapy

Of all the factors examined (gender, location of tumor, and pretreatment EUS T and N classifications), only the primary site of the tumor was found to be correlated significantly with pathCR and pathPR to therapy. Patients with cancer involving the GEJ had a statistically significantly higher chance of achieving such a response (89% pathCR or pathPR rate, odds ratio [OR] = 8.5; P = 0.01) compared with patients with cancer not involving the GEJ (50% pathCR or pathPR rate, OR = 1.0; reference). Detailed data are shown in Table 3.

Table 3. Analysis by Chi-Square Test of the Association between PathCR or PathPR and Clinical Factors in 39 Patients Undergoing Surgery
FactorNo. of PatientsNo. of pathCR/PRPathCR/PR (%)OR (95% CI)P value
  • pathCR/PR: pathologic complete and partial response; OR: odds ratio; CI: confidence interval; GEJ: gastroesophageal junction.

  • a

    Fisher exact test.

  • b

    Odds ratio could not be estimated.

 GEJ1917898.5 (1.5–46.9)0.01
Baseline T classification     
 T33323701.15 (0.18–7.33)0.89
Baseline N classification     
 N12416670.73 (0.18–3.02)0.87

Determinants of Down-staging

We also evaluated clinical factors (gender, location of tumor, and pretreatment T or N classification) to determine whether they could predict cancer down-staging. None of these factors was found to be significantly associated with down-staging (P > 0.05; data not shown).


At a median follow-up of > 30 months (minimum follow-up, 28 months; maximum follow-up, ≥ 40 months), 27 of the 43 patients (63%) had died. The median survival duration for the 43 patients was 22.1 months and the 2-year survival rate was 42% (Fig. 1). Of the 27 patients who died, 2 died of myocardial infarction within 15 days of surgery and the remaining 25 died of progressive carcinoma. We examined the effect of tumor location, gender, evidence of T and N down-staging, and pretreatment EUS T and N classification on survival time. None of these factors was found to be significantly correlated with overall survival (Table 4). However, patients with disease involving the GEJ had a longer median survival time than patients who did not have GEJ involvement (26.8 months vs. 20.3 months; P = 0.13). Similarly, patients who achieved a pathCR or pathPR had a longer survival time than patients who did not achieve a pathPR (25.6 months vs. 18.5 months; P = 0.52; Fig. 2).

Figure 1.

Overall survival. The median survival period was 22.1 months. Solid line: survival; dotted line: 95% confidence interval.

Table 4. Associations between Overall Survival Time and Clinical Factors in All 43 Patientsa
CharacteristicsNo. of patientsNo. who died24-Mo survival (95% CI)Median survival time (mos)Unadjusted hazard ratio (95% CI)Log-rank P value
  • GEJ: gastroesophageal junction; 95% CI: 95% confidence interval.

  • a

    Log-rank tests were performed to determine statistical significance.

  • b

    Median survival time was not reached.

Overall43270.42 (0.29–0.60)22.1
 Female430.25 (0.05–1.00)23.01.05 (0.56–1.92) 
 Male39240.44 (0.30–0.63)20.3Reference0.82
 Middle/distal22170.36 (0.21–0.63)20.3Reference 
 GEJ21100.53 (0.35–0.81)26.80.74 (0.50–1.09)0.13
Baseline T classification      
 T1/T2620.83 (0.58–1.00)bReference 
 T337250.35 (0.22–0.55)11.21.80 (0.87–3.70)0.09
Baseline N classification      
 N016100.40 (0.22–0.74)15.7Reference 
 N127170.43 (0.28–0.67)22.20.93 (0.63–1.38)0.81
T down staged postsurgery      
 No940.56 (0.31–0.997)bReference 
 Yes30190.47 (0.31–0.69)23.80.80 (0.73–2.15)0.41
N down staged postsurgery      
 No21140.45 (0.28–0.73)18.5Reference 
 Yes1890.53 (0.34–0.83)26.80.75 (0.49–1.14)0.17
Figure 2.

Survival by clinical response (P = 0.52). Dotted line: no response; dashed/dotted line: pathologic complete response (CR)/partial response (PR).

The median DFS time was 10.2 months. There was no statistically significant effect noted on DFS for gender, location of tumor, pretreatment EUS T and N classification, or extent of pathologic response. However, the median DFS time was longer for patients with tumor involving the GEJ compared with patients with tumor not involving the GEJ (11 months vs. 6.5 months; P = 0.19). The Kaplan–Meier plot is shown in Figure 3. Similarly, a major pathologic response to therapy (pathCR or pathPR) resulted in a longer median DFS compared with patients who did not achieve a pathPR (11 months vs. 2.9 months; P = 0.33).

Figure 3.

Disease-free survival by clinical response (P = 0.33). Dotted line: no response; dashed/dotted line: pathologic complete response (CR)/partial response (PR).

Patterns of Disease Recurrence

Twenty-five patients who died of progressive disease were reviewed for the sites of first and total disease recurrence. The first sites of disease recurrence were locoregional in 5 patients, the brain in 5 patients, and distant with or without local disease recurrence in 15 patients (the most common sites were the bones, liver, and abdomen).

The final metastatic sites were predominantly distant (most common were the liver, bones, and thoracic cavity). Local disease recurrence resulted in the deaths of two patients and all five of the patients who first developed disease recurrence with brain metastases died as a result of progressive brain deterioration.

Chemotherapy-Induced Toxicity

There were no deaths related to chemotherapy. The most frequent hematologic side effect was granulocytopenia and the most frequent nonhematologic side effect was fatigue. The Grade 3 or 4 toxic effects observed are shown in Table 5.

Table 5. Complications from Induction Chemotherapy
Type of toxicityNo. of patients with grade 3 and grade 4 (in the first cycle)No. of patients (%) with grade 3 and grade 4 (in the second cycle)
Nausea and emesis5012
Neutropenic fever2000

Chemoradiotherapy-Induced Toxicity

There were also no reported deaths related to chemoradiotherapy. Chemoradiotherapy was well tolerated, especially in the first 4 weeks. The most common Grade 3 or 4 side effects were fatigue (11 patients), granulocytopenia (4 patients), and esophagitis (3 patients). One patient developed Grade 4 mucositis and required treatment interruption. Treatment field erythema was also a common side effect. The side effects observed during chemoradiotherapy are summarized in Table 6.

Table 6. Toxicity of Chemoradiotherapy
Type of toxicityNo. of patients with
Grade 3Grade 4
Neutropenic “fever”20

Surgical Complications

The deaths of 2 of 39 patients (5%) were attributed to surgery. Both patients died of myocardial infarction. The 39 patients experienced expected postsurgical complications such as weight loss, dumping syndrome, reflux esophagitis, and aspiration, but the most observed postoperative complication was esophageal stricture (in 12 patients), which required periodic dilatations.


Gibson and Forastiere14 articulated their frustrations about not being able to compare outcomes across various studies because of the lack of uniform staging evaluations. They concluded that the future lies with incorporating newer agents while adhering to a uniform cancer staging process. Even if this notion is accepted, the challenge of how to improve results remains a formidable task. The following findings may be important: 1) in a multiinstitutional setting, the R0 resection rate is reported to be 60% after primary surgery15; 2) the local disease recurrence rate is approximately 30%15; 3) the median survival time ranges from 14–18 months; 4) preoperative chemoradiotherapy results in a pathCR rate of approximately 20–25%; 5) the 1-year survival rate is approximately 60%; 6) the 3-year survival rate ranges from approximately 25–30%; and 7) treatment-induced mortality is approximately ≤ 6%.

Patients who achieve a pathCR fare better and patients who achieve a pathPR fare better than those who do not achieve a pathPR. The current study also confirms this finding (Fig. 2), as does an earlier study.16 We analyzed a number of clinical parameters that we believed might be the determinants of a pathCR or pathPR. However, with the exception of carcinoma of the GEJ there were no determinants (this finding needs confirmation by other studies). A PET response might be a predictor of favorable outcome,17 but that possibility needs further study. Our preliminary analysis of postchemoradiotherapy (but preoperative) PET in 83 patients suggests that a low standard unit value (< 3.0) might predict the presence of only microscopic residual cancer18 in the surgical specimen. However, the clinical value of this observation remains to be determined.

In the current trial, we demonstrate that it is possible to use new agents previously studied in combination as induction chemotherapy and to administer paclitaxel combined with 5-FU concurrently with 45 Gy of radiotherapy. The R0 resection rate was high in a uniformly well staged group of patients. These results revealed that the fraction of patients with pathCR plus pathPR was substantial and that the patients achieving pathCR plus pathPR lived longer than those who did not achieve a pathPR (25.6 months vs. 18.5 months). Clearly, further refinements to this approach are necessary. Exploring targeted agents with newer agents such as CPT-11 and radiotherapy is likely to provide an advantage in the future.