Long-term Survival Following Induction Chemoradiotherapy and Esophagectomy for Esophageal Carcinoma FREE

ESOPHAGEAL cancer is a devastating disease with unacceptably poor long-term survival after radiotherapy and/or surgery. This poor outcome is primarily owing to the biologic predilection of this malignant tumor for early transmural invasion and lymph node metastasis. A 5-year survival rate of 10% or less has been reported in several studies, with a median survival time of approximately 12 months with locoregional therapy.^1- 2 Current efforts have focused on multimodality therapy to alter the pathogenesis of esophageal carcinoma. Chemoradiotherapy (CRT) before surgical resection theoretically eliminates micrometastases and improves primary tumor resectability. Numerous phase II trials suggest that this approach may improve local-regional control, and select studies have demonstrated prolonged survival in patients undergoing CRT and esophagectomy compared with historical reports of patients treated with esophagectomy alone.^3- 9 However, a limited number of phase III trials have produced conflicting results, doing little to justify the routine use of preoperative CRT.^2,10- 11

Similar to other malignant tumors, the development of esophageal carcinoma is a multistep process in which alterations of specific genetic events occur in a temporal sequence. Tumorigenesis is mediated in part by the mutation or inactivation of tumor suppressor genes that maintain the normal function of esophageal epithelium. Any alterations of tumor suppressor gene function may predict the prognosis of patients with esophageal cancer and allow for individualized treatment. Among the possible genetic disturbances are the mutational damage and hyperexpression of P53, a tumor suppressor gene located on chromosome 17p and found in esophageal tumors.^12- 15 Such mutations may appear early in esophageal tumorigenesis and are present in Barrett neoplasia prior to adenocarcinoma and in dysplasia prior to squamous cell carcinoma.^16- 17 Alterations of the P53 gene may serve as prognostic markers that correlate with clinicopathologic parameters, CRT response, and long-term survival in patients with esophageal cancer.

Patients who are free of disease on pathological examination after CRT are reported to have a more favorable outcome than patients with residual tumor in the resected esophagus.^4- 5,9 A long-term disease-free state is not limited to those who have complete responses. Still, more effective chemotherapy regimens are needed, since complete pathologic response after induction CRT is observed in only 25% of patients, and most patients die of systemic disease. Whether induction CRT is associated with improved survival requires the demonstration of a benefit in large prospective randomized trials that compare this approach with a control arm that receives surgery alone. In this phase II study of 5-fluorouracil, cisplatin, and interferon alfa with concurrent external-beam radiotherapy before esophagectomy, our aim is to identify and characterize long-term survivors using standard clinicopathologic parameters and P53 gene mutational status.

Between August 1991 and January 1995, 44 patients with potentially resectable esophageal cancer with no evidence of distant metastasis were enrolled in this study. These patients had biopsy-proven adenocarcinoma and squamous cell carcinoma of the esophagus or gastroesophageal junction. Patients with celiac lymphadenopathy were eligible for curative resection, whereas patients with documented tracheobronchial tree or mediastinal involvement were ineligible. Criteria for eligibility included (1) a life expectancy of 6 months or longer; (2) being older than18 years; (3) having no history of malignancy, excluding basal cell carcinoma of the skin or carcinoma in situ of the cervix; (4) American Heart Association classification of cardiac performance status from 0 to 1; (5) Eastern Cooperative Oncology Group performance status from 0 to 2; and (6) the following minimum laboratory parameters: serum bilirubin level less than or equal to 34.2 µmol/L (≤2 mg/dL), creatinine level less than or equal to 176.8 µmol/L (≤2 mg/dL), creatinine clearance greater than or equal to 1.0 mL/s (≥60 mL/min), white blood cell count greater than or equal to 4.50 × 10⁹, and platelet count greater than or equal to 150.0 × 10⁹. This study was approved by the institutional review board at the University of Pittsburgh School of Medicine (Pittsburgh, Pa).

Pretreatment evaluation for all patients included a medical history review and physical examination; complete blood cell count; serum electrolyte levels; liver function tests; bilirubin levels; blood urea nitrogen levels; creatinine levels and clearance; chest x-ray film; computed tomography (CT) of the chest and abdomen; electrocardiography; barium esophagography; esophagogastroduodenoscopy with biopsy; and bone scanning. Patients with tumors at or above the level of the carina underwent bronchoscopy. Pretreatment and posttreatment radiographic staging by CT was performed retrospectively by a single radiologist without prior knowledge of the posttreatment pathologic stage. Computed tomography scan criteria for regional lymph node metastases was met when lymph nodes exceeded 1 cm in short-axis diameter.

The preoperative CRT regimens have been previously described in detail and will be briefly summarized.¹⁶ Chemoradiotherapy was administered during a 28-day period (initial 16 patients) or a 21-day period (remaining 28 patients). The length of treatment was shortened by 7 days in an effort to reduce inpatient length of stay. The 28-day regimen consisted of 5-fluorouracil, 300 mg/m² per day, delivered by continuous intravenous infusion on days 1 to 28. Cisplatin was given on days 1 to 5 and 24 to 28 at a dose of 20 mg/m² per day by intravenous infusion for 6 hours. Recombinant interferon alfa-2b (Schering Corp, Kenilworth, NJ) was delivered intravenously at a dose of 3 × 10⁶ U/m² per day on the first and last 5 days of the treatment regimen and by subcutaneous injection on intervening days (Monday, Wednesday, and Friday). The 21-day regimen was similar to the aforementioned regimen with 2 treatment modifications: 5-fluorouracil was given on a dose-escalation schedule of 250 to 300 mg/m², and interferon alfa was given at a dose of 3 × 10⁶ U/m² subcutaneously 3 times per week during the 21-day period.

Concurrent radiotherapy was delivered with a high-energy linear accelerator (Varian, Palto Alto, Calif) (6- to 18-mV photons) with chemotherapy. Patients on the 28-day treatment regimen were treated with radiotherapy once a day up to a total dose of 400 Gy in twenty 20-Gy fractions. Patients on the 21-day chemotherapy regimen received radiotherapy twice a day up to a dose of 450 Gy in thirty 15-Gy fractions with a minimum of 6 hours between treatments. Patients on a once-daily regimen were treated with parallel opposed anterior-posterior fields for the entire course of therapy, whereas patients treated twice daily had spinal cord exclusion from the field at 300 Gy through an oblique arrangement. The maximum spinal cord dose was 350 Gy.

After completion of induction CRT, all patients were restaged by CT scan of the chest and abdomen. Patients with no evidence of disease progression underwent esophagectomy. Although there was no specified surgical approach, most patients underwent transhiatal esophagectomy without thoracotomy. To correct for any weight loss during induction CRT, enteral and total parenteral nutrition were routinely administered. After resection, a feeding jejunostomy tube was used to meet caloric and protein needs until adequate nutritional requirements could be maintained orally. The celiac axis, periesophageal, and subcarinal regional lymph nodes were routinely sampled during esophagectomy.

Response to induction CRT was assessed by pathologic review of resected specimens. The esophagectomy specimens were evaluated through standard protocol involving histologic examination of gross residual tumor and adjacent esophageal and gastric mucosa; proximal, distal, and deep resection margins; and all lymph nodes submitted in entirety. In areas where the tumor site was represented by mucosal ulceration, the entire region was submitted for histologic evaluation. In certain cases, rare clusters of neoplastic cells characterized the remaining tumor. In these instances, immunohistochemical staining for cytokeratins was used to differentiate residual tumor cells from radiation fibroblasts. Cytokeratin staining was also used to detect small clusters of residual tumor cells in lymph nodes with extensive necrosis. All patients were classified as having either a major pathologic response or no response. A major response was defined as either no pathologic evidence for tumor in the resected specimen (complete pathologic response) or microscopic residual tumor in the resected specimen (partial pathologic response). No response was defined as a gross residual tumor identified on pathologic review of the resected specimen. Patients were staged following the guidelines of the American Joint Committee on Cancer criteria. Toxic effects were assessed daily and graded according to the National Cancer Institute common toxicity criteria.

Patients were routinely seen every 3 months for the first 2 years and every 6 months in subsequent years. Follow-up visits included physical examination, complete blood cell count, and serum chemistry profile. Chest and abdominal CT scans were obtained every 6 months for the first year, then once yearly. Esophagogastroduodenoscopy with mucosal brushings and biopsy specimens were performed yearly. Other tests were performed as clinically indicated. The standard clinicopathologic parameters were obtained from medical records and pathology reports: age, sex, histologic findings, CRT regimen, and clinical and pathologic stages. Long-term survivors included all patients who remained alive at follow-up 3 years or more from induction CRT.

Resected tissue obtained from patients for immunohistochemical analysis was also used for detecting the presence and specific genotype of p53 mutations. Topographic genotyping has been previously described.¹⁷ Mutational status of the p53 gene was classified as either normal (tumors with no p53 immunoreactivity and no evidence of point mutations by sequencing) or mutated (tumors with evidence of point mutation by sequencing regardless of immunohistochemical staining results).

Survival time was defined as the time from induction CRT to the date of death or last follow-up. Disease-free survival was characterized as the time from the date of surgery until death or recurrence of disease. Only patients completing induction CRT and undergoing curative resection were included in the disease-free survival analysis. Median survival was calculated for each patient subgroup by the Kaplan-Meier method. Differences in survival compared with prognostic factors were evaluated by a log-rank test. Cox regression models evaluated the joint effect of predictive variables. Results of p53 gene alterations were compared with the clinicopathologic features using the χ² test for qualitative data and the t test for quantitative data with a 2-tailed P value considered significant at .05.

Between August 1991 and January 1995, a total of 44 patients were entered into the study (Table 1). The mean age was 59 years (range, 30-77 years). Most patients were men (82%) and white (93%). The tumors were located either in the middle or distal esophagus with adenocarcinoma as the predominant histologic finding (75%).

Of the 44 patients enrolled in the study, 41 (93%) completed induction CRT (Figure 1). Three patients received either minimal or no calculated CRT doses. Of the 41 patients who completed induction CRT, all but 4 patients were eligible for curative esophagectomy. Disease progression precluded resection for 1 patient who developed hepatic metastasis. The 3 other patients died in the post-CRT period. One of these patients developed intractable cardiac arrhythmias that resulted in eventual death 4 months after the initiation of induction CRT. The causes of the other 2 deaths, which occurred at home, remain unknown.

At surgical exploration, all of the 37 patients eligible for curative esophagectomy qualified for resection. Only 1 patient could not have all gross tumor removed at the time of surgery. The remaining 36 patients underwent curative esophagectomy and were disease-free after resection. In-hospital or 90-day perioperative mortality included 3 (8%) of 37 patients.

Pathologic review of the resected specimens showed major pathologic responses in 33 patients who completed induction CRT. Of this group, 10 patients had no residual tumor (complete pathologic response) and 23 patients had microscopic residual tumor (partial pathologic response) found in either resected or nodal tissue. There was no clear association between histologic features and CRT response. Furthermore, pathologic response demonstrated no relationship to clinical tumor stage or regional lymph node status.

At a median follow-up of 50 months for all surviving patients, 17 of 44 patients are alive, of whom 15 patients show no evidence of recurrent disease. Long-term survival (≥3 years) was observed in 14 of these 17 patients. Of the 14 patients with long-term survival, 1 patient died of disease and another patient is alive with recurrent disease (Table 2). The remaining 12 patients are alive and disease-free (median follow up, 54 months). From the time of study accrual, 27 patients have died, including 23 of disease, 3 of postoperative complications, and 1 of an unrelated cause. Of the 23 patients who died of disease, 3 deaths occurred prior to resection, 1 owing to treatment. The causes of the other 2 deaths remain unknown.

Of the 14 long-term survivors, 6 patients have survived longer than 4 years, and 3 patients longer than 5 years (Figure 2). A second primary tumor developed in 2 patients. The first patient underwent pneumonectomy for lung cancer, and the second patient required a penectomy for squamous cell carcinoma of the penis. Both patients are currently alive with no evidence of disease. One patient experienced a recurrence at 11 months following the start of treatment and remains disease-free 43 months postresection of a single brain metastasis. Of the 7 long-term survivors who had a complete response to CRT, 6 are alive with no evidence of disease. The remaining patient who had a complete CRT response is alive with recurring disease at 36 months and is undergoing radiation therapy. Of the 7 long-term survivors who had a partial response to CRT, 6 are alive with no evidence of disease. The remaining patient with a partial CRT response died of disease. Median survival for the long-term survivors has not been reached.

Six prognostic factors were evaluated for their effect on long-term survival (≥3 years): (1) age (above and below median); (2) sex; (3) treatment regimen (28 days vs 21 days); (4) clinical tumor stage (T2 vs T3 vs T4); (5) clinical nodal stage (N0 vs N1); and (6) P53 mutational status. Of the clinicopathologic parameters tested for association with survival greater than or equal to 3 years, none were significant prognostic factors for survival (Fisher exact test, t test, 2-tailed) (Table 3). While not significant, P53 mutational status suggested long-term survival. In 11 of the 14 patients who are alive and recurrence-free and for whom p53 genotyping was available, 10 patients had no point mutations. Analysis revealed a mutation of p53 in the remaining patient.

The combination of 5-flourouracil, interferon alfa, and cisplatin along with external beam radiotherapy represented the induction regimen that preceded esophagectomy in this phase II study. Similar to other phase II trials examining induction CRT, only 24% of patients had a complete pathologic response.^2,4,7- 9

In this study, a complete pathologic response to CRT was not a reliable predictor of long-term survival. There were no significant differences in the number of long-term survivors based on response (complete vs partial) or in complete pathologic response rate with regard to CRT dose and fractionation schedule. Of the 14 long-term survivors, 7 patients were partial pathologic responders, whereas 7 patients were complete responders to CRT. The pattern of failure found in these patients with esophageal carcinoma clearly implicates distant metastasis as the major problem while underscoring local control of the disease as being more attainable. These findings reinforce the need for more effective systemic therapies for disease treatment. Future controlled trials are needed to identify other novel agents and combination therapies before and/or after esophagectomy that may improve the survival advantage of this patient population.

Although there are several phase II trials examining the use of multimodality treatment for patients with esophageal cancer, only a few phase III trials have been conducted. Several nonrandomized trials using induction CRT report survival rates that are several-fold higher than those reported in past studies for esophageal resection alone.^3- 5,7- 8 However, the true efficacy of multimodality treatment for esophageal cancer remains unclear and controversial owing to the lack of data from well-designed phase III trials comparing induction CRT followed by resection with resection alone. One such single institution trial randomized 113 patients with esophageal adenocarcinoma to receive either induction CRT followed by resection or resection alone.² The complete pathologic response rate was 25%, and the 3-year survival rate was 32% in the induction CRT group compared with 6% in the resection-alone group. Other studies, however, have shown no improvement in overall survival for induction CRT, although patients who responded to CRT did have prolonged survival.^10- 11

Certain criticisms of these studies point to the small sample sizes and possible staging imbalances between CRT-treated and control patients. In 1 study, positive lymph nodes were found in 82% of patients in the surgery group, whereas only 42% were found in patients receiving CRT.² This difference may be owing in part to tumor down-staging after up-front CRT. Alternatively, pretreatment clinical staging is known to be grossly inaccurate, and ensuring equivalent extent of disease between the 2 groups is difficult at best. Therefore, until more data from phase III trials are available for analysis, induction CRT should still be considered experimental. Nevertheless, the results from phase II and phase III trials at this time are intriguing and serve as an impetus for further investigation of this potential multimodality approach.

In trials investigating the therapeutic benefit of CRT alone in treating esophageal cancer, the reported survival rates were essentially similar to those shown in trials for esophagectomy alone.^20- 21 Such studies suggest that surgical resection may not be an essential component of the multimodality approach for esophageal cancer. However, results from our study and others indicate the essential role of surgery. Our data suggest that esophagectomy is critical for treatment since there are patients with persistent tumor in resected esophagus who are alive at 3 years with no observable local or distant recurrences. Indeed, in several studies treating local-regional esophageal carcinoma with chemoradiation alone, high local failure rates were reported.^22- 23 In 1 trial, a survival advantage for concurrent chemoradiation compared with radiation alone in patients with locoregional esophageal cancer was reported with significant differences in both local and distant failure rates.²⁰ However, almost half of all patients in this superior arm had persistent or recurrent local disease after 12 months. The effect of this poor locoregional control on the development of disseminated metastatic disease and long-term survival remains unknown but certainly can contribute to poor quality of life and subsequent death. Although the primary cause of death in patients is related to widespread metastatic disease, esophagectomy may still prove beneficial for those patients with residual disease found only at the primary site of the tumor. This approach is supported by the high local failure rates reported in the aforementioned chemoradiation trials. Furthermore, a significant group of patients do not achieve relief of their dysphagia with nonsurgical therapy. Thus, nonsurgical approaches may not prove to be optimal treatment for this patient population.

In this trial, most of the partial responders (6 of 7 patients) who were alive at study completion with no evidence of disease had residual microscopic tumor either at the primary site only or in regional lymph nodes. It is difficult to imagine that these patients would have an equivalent disease-free survival period if residual tumor had been left behind. Unfortunately, current imaging modalities cannot distinguish these patients from patients who have complete pathologic responses. In an effort to improve long-term survival for patients with esophageal carcinoma, a molecular marker that reliably predicts those who will have complete and partial responses may facilitate the selection of suitable patients for esophagectomy. In previous studies, point mutation of the p53 tumor suppressor gene correlated with residual disease in resected esophageal specimens and was an independent predictor of overall and disease-free survival.^13,19 In this trial, p53 mutational status was suggestive, although not a significant predictor, of long-term survival. The small number of patients may explain this nonsignificance. In 11 of the 14 patients who were alive and recurrence free, p53 genotyping demonstrated no point mutations in 10 patients. Analysis revealed a mutation of p53 in the remaining patient. Other molecular markers such as c-erb-b2 have also been examined as potential predictors of response and outcome in patients with esophageal cancer.¹⁴ Although less invasive and more attractive than current staging methods, these techniques using specific molecular markers require further investigation and confirmation before they can be routinely applied.

Although 5-year survival has not been reached, the survival advantage of this combined modality treatment group at 3 years seems durable. At 3 years, 32% of patients treated with CRT and esophagectomy are alive with no evidence of recurrent disease. In 1 CRT trial, however, 15 of 21 treated patients with squamous cell carcinoma underwent resection, of which 5 had complete pathologic responses.²³ Subsequently, all patients with CRT responses died of disseminated metastasis at 30 to 60 months. Whether the survival curve has reached its plateau at 3 years, and thereby is equivalent to cure, requires continued follow-up.

In conclusion, long-term survival can be achieved in patients with esophageal cancer. Patients surviving for 3 years or longer are less likely to develop tumor recurrences after undergoing induction CRT prior to surgical resection. In the long-term survivors, esophagectomy was a critical component of treatment, since one half of these patients had residual disease after induction CRT. Clinicopathologic parameters and P53 mutational status are not significant predictors of long-term survival. Clearly, the therapeutic benefit of induction CRT and esophagectomy remains to be assessed in larger multi-institutional phase III trials. Until such trials are completed, induction CRT should be offered to patients enrolled in formal research protocols designed to determine the optimal treatment approach. At this time, however, surgery remains the standard treatment for locoregional confined disease in patients with esophageal cancer.

Corresponding author and reprints: Mitchell C. Posner, MD, Department of Surgery, The University of Chicago Hospitals, 5841 S Maryland Ave, MC 5031, Chicago, IL 60637 (e-mail: mposner@surgery.bsd.uchicago.edu).