To the Editor:
The negative result obtained by Kelsen et al. in their trial of preoperative chemotherapy for operable esophageal cancer (Dec. 31 issue) (1) is somewhat surprising, since Herskovic et al. (2) reported positive results for similar chemotherapy with or without radiation therapy. Patient accrual in the study by Kelsen et al. was meager (average, <1 patient per hospital per year). How did the study centers maintain high clinical quality, given the low accrual rate? Were there any differences between the centers that were major contributors of data and the rest of the hospitals? Our own results lead us to believe that induction chemotherapy for esophageal cancer is difficult to do well and requires the efforts of a complex, multidisciplinary team of committed physicians. (3) In the study by Kelsen et al., only 71 percent of patients received all three cycles of chemotherapy, and only 80 percent of patients in the chemotherapy group had surgery. Were the goals of the study more likely to be fulfilled in high-accrual centers?
Of the patients who underwent surgery, the chance of a potentially curative resection (R0) increased from 62 to 78 percent with preoperative chemotherapy. This difference is probably both clinically and statistically significant. The finding that survival was essentially equivalent in the two groups, even though surgery was performed in 16 percent fewer patients in the chemotherapy group than in the immediate-surgery group (80 percent vs. 96 percent, respectively) suggests that a survival advantage is associated with chemotherapy.
What the clinician (and the patient) really want to know is whether, if induction treatment is given before surgery, there will be a benefit. Kelsen et al. could advance this debate by performing a subgroup analysis of patients who received all three cycles of chemotherapy and had surgery. If there is a benefit in this group of patients, attention should be directed to reducing the toxicity of the induction treatment so as to maximize the number of patients who receive all prescribed treatments.
Cameron D. Wright, M.D.
Massachusetts General Hospital
Boston, MA 02114
2. Herskovic A, Martz K, Al-Sarraf M, et al.
Combined chemotherapy and radiotherapy compared with radiotherapy alone
in patients with cancer of the esophagus. N
Engl J Med 1992;326:1593-8.
Return to Text
3. Wright CD, Wain JC, Lynch TJ, et al. Induction
therapy for esophageal cancer with paclitaxel and hyperfractionated
radiotherapy: a phase I and II study. J
Thorac Cardiovasc Surg 1997;114:811-6.
Return to Text
Kelsen et al. are to be congratulated on their study, which reflects the culmination of more than 20 years of effort by the primary investigator to improve the survival of patients with esophageal cancer by the use of preoperative chemotherapy. Although the study convincingly shows that preoperative chemotherapy does not improve overall survival among patients with localized esophageal cancer, several reports indicate that the subgroup of patients who respond to preoperative therapy may still benefit from this approach, (1) whereas neoadjuvant therapy in patients who do not respond may actually be detrimental. (2) An analysis of the subgroups who do and do not have a response to preoperative chemotherapy in comparison with those who undergo primary resection is therefore warranted.
Furthermore, in the study by Kelsen et al., a bias toward more favorable tumor stages in the group undergoing surgery alone, a bias that cannot be excluded, would mask a beneficial effect of neoadjuvant chemotherapy. For example, the number of patients who were not recruited into the program during the trial period and the treatments they received are not given. Perhaps only patients with more advanced disease were included at some of the participating centers. Furthermore, the number of registered but ineligible patients was significantly higher in the chemotherapy-plus-surgery group than in the surgery-only group (P<0.001 by Fisher's exact test). Finally, the expertise of surgeons, and consequently the quality and extent of surgical resection and lymphadenectomy, (3) must have varied substantially, since a wide range of more or less radical surgical procedures were "considered acceptable" and the surgical resections were performed at more than 120 different institutions.
Ulrich Fink, M.D.
Hubert J. Stein, M.D.
Technische Universitat Munchen
D-81675 Munich, Germany
1. Roth JA, Pass HI, Flanagan MM, Graeber GM,
Rosenberg JC, Steinberg S. Randomized clinical trial of preoperative and
postoperative adjuvant chemotherapy with cisplatin, videsine, and
bleomycin for carcinoma of the esophagus. J
Thorac Cardiovasc Surg 1988;96:242-8.
Return to Text
2. Law S, Fok M, Chow S, Chu KM, Wong J.
Preoperative chemotherapy versus surgical therapy alone for squamous
cell carcinoma of the esophagus: a prospective randomized trial. J
Thorac Cardiovasc Surg 1997;114:210-7.
Return to Text
The results reported by Kelsen et al. are difficult to interpret for several reasons. The study was originally designed for the treatment of epidermoid cancer, but 18 months later it was modified to include adenocarcinoma of the esophagus. No reason is given for this decision, nor is the type of involvement of the gastroesophageal junction specified. Thus, some patients may actually have had a cardiac carcinoma.
Furthermore, the recruitment of patients was slow. It took more than five years to register 467 patients. Moreover, surgeons were required to perform at least four esophagectomies per year, but fewer than four patients per participating center were entered into the study. We assume that other patients were treated differently, even if they were eligible. Kelsen et al. state that one of the most common reasons why the full dose of chemotherapy was not administered either before or after surgery was the decision of the physician. This point should be clarified. The study included substantial weight loss and cell type (epidermoid cancer or adenocarcinoma) as stratifying variables. Weight loss is said to be a predictor of poor outcome and to occur more often in cases of epidermoid cancer. However, there was no difference in outcome according to these distinct histologic subtypes. This raises the question whether this study simply lacked the statistical power to detect a difference in overall survival, despite the large number of patients.
Most important, the status of the lymph nodes was not included in the stratification. In our opinion, patients with lymph-node metastasis are not ideally suited for surgery, because this tumor stage often "represents systemic disease beyond the limits of resection." (1) These additional data should be given, although we believe that a post hoc subgroup analysis cannot definitively answer our questions.
Jochen Fleeth, M.D.
Friedrich Begemann, M.D.
General Hospital St. Georg
D-20099 Hamburg, Germany
1. Roth JA, Putnam JB Jr, Rich TA, Forastiere
AA. Cancer of the esophagus. In: DeVita VT Jr, Hellman S, Rosenberg SA,
eds. Cancer -- principles & practice of oncology. 5th ed. Vol. 1.
Philadelphia: Lippincott-Raven, 1997:980-1021.
Return to Text
I would like to suggest two possible reasons why preoperative chemotherapy failed to benefit patients with operable esophageal carcinoma in the study by Kelsen et al. One is that patients assigned to receive chemotherapy before surgery were less likely to receive surgery. The authors report that 217 of 227 patients assigned to the surgery group underwent surgery, but only 171 of 213 assigned to the chemotherapy-plus-surgery group actually underwent surgery. The authors do not report why the latter 42 patients did not have surgery. It would be useful to know the exact reasons.
Another, less obvious possibility is an uneven distribution of stages of disease between the two treatment groups. Because the patients were not stratified according to the stage of the disease and because the trial included patients with tumor stage 1, 2, and 3 disease, it is possible that patients who received chemotherapy had more advanced disease than patients who underwent only surgery. However, Kelsen et al. do not report the stages of disease in the two treatment groups. One potentially confounding factor is that preoperative staging may not reflect the true stage of disease at the time of surgery. Frequently, a patient's disease may be assigned a higher stage because of false positive findings on preoperative computed tomographic scans and the detection of microscopic disease in regional lymph nodes. Therefore, it is highly important to understand the distribution of stages between the two groups of patients in the study by Kelsen et al. and to analyze outcomes according to stage.
Alex Y. Chang, M.D.
Upstate New York Cancer Research and Education Foundation
Rochester, NY 14623
To the Editor:
In response to Drs. Fleeth and Begemann: during the initial steps in the design of our study, the incidence of adenocarcinoma was controversial. When the increased incidence of this tumor became clear, the protocol was amended. Patients with cancers of the gastroesophageal junction could not have more than 2 cm of tumor extension into the stomach. Participating institutions were not required to submit a register of eligible patients who were not entered into the study, and therefore we do not have data regarding their treatment. The preference of a physician not to continue chemotherapy was adjudicated when it was clear that there was no progression of disease, no dose-limiting toxicity, and no refusal of additional treatment by the patient; 10 patients met these criteria. The study was designed to test for a treatment difference with a statistical power of 90 percent, which has been judged to be more than adequate.
Drs. Fleeth and Begemann suggest that patients with lymph-node metastasis are not ideally suited for surgery. We did not stratify patients on the basis of the status of the lymph nodes because of the inaccuracy of currently available noninvasive staging techniques. However, there were no significant differences between the groups according to the clinical nodal stage. Of the group that was treated with surgery only, 59 percent had lymph-node metastases; the corresponding figure for the chemotherapy-plus-surgery group was 49 percent. However, fewer patients in the chemotherapy-plus-surgery group underwent surgical exploration.
In response to Dr. Chang: patients in the chemotherapy group did not undergo surgery for various reasons, including death (10 patients, with death as a result of toxic effects of chemotherapy in 7), inability to tolerate surgery (6), a tumor that was clinically unresectable (5), refusal by the patient (5), development of metastatic disease (3), removal from the protocol (1), and unknown reasons (12). Significantly more patients in the chemotherapy group who underwent surgical exploration had stage 0 or 1 disease (P=0.003), suggesting that their disease was assigned a lower stage, but this difference was offset (for the whole group) by patients in the chemotherapy group who had progression of disease and did not undergo surgery.
Drs. Fink and Stein ask about the survival of patients who had a response to treatment and those who did not respond. Patients in the chemotherapy group who responded to treatment had a significantly better survival rate than did those in the chemotherapy group who did not respond (P=0.002) and those in the surgery-only group (P<0.001).
In response to Dr. Wright: 53 of 78 patients (68 percent) in the three high-volume centers received three courses of induction chemotherapy, as compared with 90 of 126 patients (71 percent) in the institutions with lower accrual. While R0 resections were more likely to be achieved in patients who were assigned to both chemotherapy and surgery and who underwent surgical exploration, there were no significant differences in the overall distribution of R0 resections, since 42 of 213 patients (20 percent) who received chemotherapy did not undergo surgery. We evaluated survival in the subgroup of patients who received all three cycles of chemotherapy followed by surgery as compared with those who underwent surgery only. Whereas patients who received three courses of chemotherapy had significantly better survival than patients who received fewer than three courses of treatment (P=0.04), there was no significant difference in survival between patients who received all three courses and those who underwent surgery only (P=0.80). This result may be due to the presence of progressive disease before surgery in some patients who received fewer than three courses of treatment.
David Kelsen, M.D.
Memorial Sloan-Kettering Cancer Center
New York, NY 10021
Thomas Pajak, Ph.D.
Radiation Therapy Oncology Group Statistical Unit
Philadelphia, PA 19107
Robert Ginsberg, M.D.
Memorial Sloan-Kettering Cancer Center
New York, NY 10021