Patients with metastatic or unresectable esophageal cancer recurring after previous treatments (chemotherapy, radiation therapy, or surgery) were eligible to participate in this study. The clinical stage of each cancer was evaluated using a complete medical history, physical examination, complete blood count, clinical chemistry, chest radiography, computed tomography of the chest and upper abdomen, and gastrofiberscopy with biopsy. A whole-body bone scan was also performed for cases showing any symptoms suggestive of bony involvement, including abnormalities in serum calcium or alkaline phosphatase levels. Clinical stages for all patients were based on the revised American Joint Committee (AJC) system. Further eligibility requirements were as follows: voluntary informed consent in writing; age <75 yr; Eastern Cooperative Oncology Group (ECOG) performance status ≤2; life expectancy >12 weeks; and adequate hematological, renal, and hepatic function (WBC ≥3,000/µL, granulocytes ≥1,500/µL, platelets ≥100,000/µL, total bilirubin <1.3 mg/dL, and creatinine clearance >60 mL/min). Exclusion criteria were: pre-existing neurotoxicity >WHO criteria grade 1, active infection, other serious underlying medical conditions that would impair the ability of the patient to receive the planned treatment, or inadequate calorie and fluid intake.

Paclitaxel (90 mg/m²) and cisplatin (50 mg/m²) were administrated by intravenous (i.v.) infusion every 2 weeks. Thirty minutes prior to the paclitaxel infusion, each patient received 20 mg dexamethasone, 50 mg ranitidine, and 5 mg chlorpheniramine maleate i.v., to prevent hypersensitivity reactions. After prehydration with at least 1 L of normal saline, the calculated dose of paclitaxel, diluted in 500 mL of normal saline, was infused over 3 hr. The calculated dose of cisplatin was then administered over 3 hr, followed by posthydration with 3 L of normal saline over 24 hr. Ondansetron (8 mg, i.v.) was routinely given. Patients received further cycles of chemotherapy only when the absolute neutrophil count was ≥1,000/µL and platelets ≥150,000/µL. The paclitaxel dose was reduced to 75% of the original dose in subsequent courses if one of the followings occurred: grade 3 neutropenia with infection; grade 4 neutropenia; grade 3 thrombocytopenia with bleeding that required platelet transfusion; and over grade 3 sensory neurotoxicity. In cases of fatigue or asthenia above grade 3, treatment was postponed for 1 week and restarted when the patient had recovered to below grade 2. After three courses of chemotherapy, progressive patients were excluded. Patients showing disease regression received additional courses of chemotherapy, up to a maximum of eight courses.

The clinical restaging procedure included all the studies that had shown abnormal findings prior to treatment. To evaluate toxicity, a physical examination, complete blood count, and blood chemistry were performed at the 8th to 12th day after chemotherapy. To evaluate responses, CT scans of the chest and abdomen, with ultrasonography and endoscopy when appropriate, were repeated after the third and sixth cycles and after the discontinuation of therapy. Responses were evaluated using WHO criteria. A complete response was defined as the complete disappearance of all clinical evidence of disease, persisting for more than 4 weeks. A partial response was defined as a decrease by at least 50% reduction in the sum of the products of the largest perpendicular diameters in all measurable lesions, including esophageal wall thickness or at least a 30% reduction of the largest diameters in uni-dimensional disease (evaluable disease) for at least 4 weeks, with no evidence of progression or of the development of new lesions. Progressive disease was defined as an increase in a previous lesion by more than 25%, or the development of any new lesion. Stable disease was defined as any change in the previous lesion that did not fit into either the partial response or progressive disease categories. DeMeester symptom scores (11) were used by patients to rate their swallowing symptoms on a scale of 0-3 at the beginning of chemotherapy and the best relief during chemotherapy. The scoring system is defined as follows; 0=no dysphagia, 1=mild, i.e. with solids, requiring modification of diet to soft foods, 2=moderate, i.e. difficulty with soft foods, predominantly liquid diet, 3=severe, i.e. obstructed, needing medical intervention for dilatation or bolus obstruction. Toxicities were evaluated according to the National Cancer Institute Common Toxicity Criteria (NCI-CTC).

Overall survival was measured from the date when the treatment began to the date of death, or of the most recent follow-up visit. The duration of response was measured from the date that the response was noted to the date that progression was confirmed, or to the last visit without progression. The time to progression was measured from the date the treatment began to the date of progression, or to the last visit without progression. The statistical analysis was performed using SPSS software (SPSS, Chicago, IL, U.S.A.). All survival data were updated to June 2004. Median survival times and survival curves were estimated using the method of Kaplan and Meier and were compared using a log-rank test. A value of p=0.05 (two-sided) was considered the limit of significance.

Between April 1999 and June 2003, 38 eligible patients with esophageal cancer were enrolled. Of these 38 patients, 6 patients were ineligible: 3 patients had follow-up loss after one cycle of chemotherapy, 2 patients were dropped out because of economic condition and one patient had double primary tumor (esophageal and lung cancer). Therefore, 32 patients were eligible for analysis. Patient characteristics are presented in Table 1. All the patients were male and had confirmed squamous cell carcinoma. Their median age was 60 yr (range 48-74 yr). As indicated in Table 1, 28 patients (88%) had been treated previously. Ten patients (31%) had received neither chemotherapy nor radiotherapy; either metastatic disease was found at diagnosis or they had relapsed after receiving only surgery. Nineteen patients (59%) had received chemotherapy with 5-FU and cisplatin, 18 patients (56%) had received radiation therapy with or without concurrent chemotherapy. Twenty patients (63%) had metastatic lesions. The predominant metastatic sites were the lung, liver, bone, and lymph nodes. All patients evaluated for this study received at least two cycles of chemotherapy.

The median number of cycles received was five, with a range of from two to eight cycles; 151 treatment cycles were delivered in total. Of the 32 patients, 26 pateints (81%) received more than three cycles of chemotherapy, and 13 patients (41%) received more than six cycles. No patients had any further treatment after this combination chemotherapy. The median interval between cycles was 19 days (range, 13-36 days), owing to delayed hematologic recovery and asthenia. The planned and achieved dose intensities for cisplatin were 25 mg/m²/week and 19.2 mg/m²/week, respectively; for paclitaxel, they were 45 mg/m²/week and 33.5 mg/m²/week, respectively.

Among the 32 patients, 28 patients had measurable disease and 4 patients had evaluable disease. One patient (3%) showed a complete response (CR), 12 (38%) had partial responses (PR), 9 (28%) had stable disease (SD), and in 10 patients (31%) the disease progressed (PD) (Table 2). The objective response was 41%. The CR patient still had stable disease after three cycles of chemotherapy, but reached CR after six cycles. Seven partial responses were seen among the 10 chemotherapy-or radiotherapy-naive patients, and 6 of 22 patients previously treated with chemotherapy or radiotherapy responded either completely (one patient) or partially (five patients). The objective response among naive patients was significantly higher than among the others (p=0.049, two-sided Fisher's exact test). Follow-up dysphagia scores were available for 29 patients. 11 patients (34%) had an improvement of at least one grade in dysphagia score after treatment and 12 patients (37%) showed no change of swallowing difficulty. Only 6 patients (19%) had a progression of dysphagia. For responders, the median duration of response was 4.5 months (range, 1.0-49.2 months), and the median time to progression for all patients was 5.0 months (range, 1.3-52.6 months).

The main hematologic toxicities observed were anemia and neutropenia. Over grade 3 of neutropenia and anemia were observed in 6 (19%) and 5 (16%) patients, respectively. Thrombocytopenia was rare. Neutropenic infection developed in 2 patients (6%) with pneumonia, who recovered after treatment with antibiotics. Grade 1 or 2 sensory neurotoxicity was observed in 10 patients (50%); sensory neurotoxicity over grades 3 did not occur. The major non-hematologic toxicity was fatigue, which spontaneously resolved after 1 or 2 weeks of rest. One patient developed grade 3 hepatotoxicity after six cycles of chemotherapy. Therapy was stopped, and the patient recovered spontaneously after 2 months. There was no treatment-related mortality. There were no significant differences in hematologic or non-hematologic toxicities between chemotherapy-or radiotherapy-naive patients and the others (Fig. 1).

The median overall survival time of all patients was 7 months (95% CI, 5.13-8.87). The median overall survival time of responding patients was 10 months (95% CI, 4.95-15.05; range, 5-51 months) compared with 5 months (95% CI 3.85-6.15, range 2-20 months) for non-responding patients. This increased survival time was significant (p=0.0069). The 1-yr and 2-yr survival rates of all patients were 28.1% and 7.1%, respectively (Fig. 2).