Carcinoma of the esophagus is a virulent malignancy that is diagnosed only when patients experience rather severe and advanced symptoms. As a result, it is not uncommon to find M1 cancer at diagnosis or node-positive cancer on initial staging. Approximately 13,900 new cases of esophageal cancer (more than half involving the gastroesophageal junction) are diagnosed each year in the United States. In the West, esophageal cancer has undergone a rapid evolution affecting predominantly Caucasian men and having an adenocarcinoma histology. Association with Barrett's metaplasia, reflux disease, hiatal hernia, and obesity is also evident. This is in contrast to endemic areas where squamous cell histology predominates, and mid- or upper thoracic localization is common. The carcinogenic forces at work are also different. The 5-year survival rate of patients with local-regional esophageal or gastroesophageal carcinoma is approximately 15% to 20%. This cure rate can be obtained by surgery, and by some estimates, also by definitive chemoradiotherapy. The data on definitive chemoradiotherapy are predominantly on squamous cell carcinoma. For patients taken straight to surgery, an R0 (curative) resection rate of 60% and 3-year overall survival rate of approximately 30% is expected. Surgical mortality has been substantially reduced in recent years due to improved supportive care and better patient selection. Nonetheless, it is well understood that mortality and morbidity are directly related to number of surgeries performed at a given institution. For example, the data of Swisher et al demonstrated that hospitals that perform a large number of esophagectomies demonstrated improved outcomes and decreased hospital use. For patients with localized esophageal carcinoma, the challenges include (1) proper selection of patients who are most likely to benefit from the stipulated approach; (2) achievement of an R0 resection when surgery is contemplated; (3) achievement of pathologic complete response or < 10% microscopic cancer in the surgical specimen (two groups that have a better survival than those with > 10% residual cancer); (4) reducing the rate of persistent or locally recurrent cancer when using definitive chemoradiotherapy, (a rate currently above 60%-a desired number would be 30%); (5) reducing morbidity from chemoradiotherapy, surgery, and chemotherapy by providing adequate nutrition, supportive care, and willingness to constantly refine approaches; and (6) improving staging procedures again to select the proper patients. The three-step strategy provides more theoretical advantages than the previously used uni- or dual-modality approaches. We elected to use a combination of irinotecan(Drug information on irinotecan) (CPT-11, Camptosar) and cisplatin that has demonstrated activity in patients with carcinoma of the esophagus. We chose continuous infusion of fluorouracil(Drug information on fluorouracil) (5- FU) and paclitaxel(Drug information on paclitaxel) with radiotherapy, as it has previously resulted in lowered morbidity in patients with carcinoma of the esophagus and gastroesophageal junction in previous studies by our group. Our objectives in this trial were to increase the R0 resection rate, increase the fraction of patients who achieve significant pathologic response, and evaluate morbidity and mortality. The data being discussed are preliminary. Methods Patients with potentially resectable adenocarcinoma or squamous cell carcinoma of the esophagus or gastroesophageal junction with histologic proof of adenocarcinoma underwent full staging including endoscopic ultrasonography. The patients had acceptable hepatic, renal, and marrow function. They were male or female, 18 years or older, with a performance status (PS) ≤ 2. Patients with celiac adenopathy (≥ 1 cm) visualized by CT scan (except when the gastroesophageal junction was involved) were not registered in the study. All patients gave written informed consent. Patients had to be medically fit to withstand surgery. At the time when this protocol was initiated, neither our institution nor others routinely used positron emission tomography scanning. Irinotecan was given at 70 mg/m2, and cisplatin(Drug information on cisplatin) given at 20 mg/m2 IV once a week on days 1, 7, 21, and 28, but not on days 14 and 35 (one cycle). If a patient achieved a clinical response, a second cycle of irinotecan/ cisplatin was given that began on day 42 (day 1 of the second cycle). The patients received radiotherapy at 45 Gy (1.8-Gy fractions) and continuous infusion low-dose 5-FU at 300 mg/m, from Monday through Friday (96-hour infusion), and paclitaxel 45 mg/m2/d on the first day of each radiotherapy week. Four to six weeks after completion of chemoradiotherapy, the patients underwent surgery provided there was no evidence of metastatic disease. The patients received standard premedication prior to paclitaxel, and received standard pre- and post-cisplatin hydration and antiemetics. Statistical Considerations The primary goal of this study was to evaluate the feasibility of preoperative chemotherapy followed by chemoradiation in patients with potentially resectable carcinoma of the esophagus or gastroesophageal junction. Feasibility was defined as 85% (34 out of 43) of patients completing the protocol therapy. Dose modifications were calculated so that if two of the first six patients developed grade 4 nonhematologic toxicity, dose modifications were made to the chemotherapy dose/schedule. The secondary goal was to achieve ≥ 20% rate of pathologic CR; an optimal two-stage design was used. If more than one pathologic CR was observed with the first 21 patients, then 22 more patients were accrued. At the end of the study, if four or less pathologic CRs were reported, this strategy/regimen would not be worth further investigation. Pretreatment Evaluation Histologic confirmation was obtained. Complete history and physical, performance status, urinalysis, SMA-12, complete blood count, differential, platelets, and electrolytes were also obtained. Chest radiograph, upper gastrointestinal radiographs, electrocardiogram, and CT scan of the abdomen and chest were obtained. Endoscopic procedures were performed when indicated. Results Forty-three patients were enrolled (39 males/4 females; median age, 61 [range: 39-73 years]). Thirty-five patients had adenocarcinoma, six had squamous cell carcinoma, and two had both. Thirty-nine of 43 patients enrolled were evaluable for toxicity; all were evaluable for survival. However, the analyses of certain parameters are currently incomplete. Adenocarcinoma and distal esophageal or gastroesophageal junction location of carcinoma were seen frequently. The baseline endoscopic ultrasonography revealed that 36 patients had a T3 tumor, five had a T2 tumor, and two had a T1 tumor. Twentyseven patients had node-positive cancer (N1). Presurgery biopsies were performed in 20 patients, and no cancer cells were identified in the biopsy specimens of 16 patients. This did not correlate with the pathologic complete response. Thirty-nine of the 43 (91%) patients underwent surgery; all had an R0 (curative) resection. A pathologic complete response was noted in 12 of the 39 patients. In addition, 17 patients had only microscopic (< 10%) viable cancer in the specimen. Therefore, major pathologic responses were observed in 29 of 39 (74%) patients or 29 of 43 (67%). With a median follow-up beyond 30 months, 20 patients are alive, and 12 patients remain free of cancer. Toxicity During the induction phase, common toxicities included diarrhea, neutropenia, neutropenia, and some degree of nausea and vomiting. Grade 3 or 4 diarrhea was uncommon. During chemoradiotherapy, common toxicities included fatigue, nausea, and neutropenia. There were no deaths related to therapy. Conclusions Our data suggest that the strategy of induction chemotherapy followed by preoperative chemoradiotherapy is clearly feasible using irinotecan and cisplatin as an induction regimen and giving 5-FU and paclitaxel concurrently with radiotherapy. The study has had a very satisfactory rate (91%) of R0 resection, and the morbidity is acceptable. With the three-step strategy utilized, the fraction of patients with major pathologic response has increased substantially (67% of all 43 patients had < 10% of viable cancer cells in the surgical specimen). Further follow-up is necessary, and, perhaps, more refinement of this approach is warranted to improve further our results.