Concurrent Chemotherapy and Radiation Therapy Followed by Transhiatal Esophagectomy for Local-Regional Cancer of the Esophagus By Arlene A. Forastiere, Mark B. Orringer, Claudia Perez-Tamayo, Susan G. Urba, Sally Husted, Bonnie J. Takasugi, and Marianna Zahurak Forty-three patients with local-regional squamous-cell carcinoma of the esophagus or adenocarcinoma of the esophagus, cardia, or gastroesophageal junction were treated with concurrent cisplatin, vinblastine, fluorouracil (5-FU), and radiation therapy (RT) over 21 days. A transhiatal esophagectomy (THE) was planned on day 42. Seventy-nine percent had T2 primaries by clinical staging and 56% had enlarged regional nodes (N) on computed tomographic (CT) scan. Forty-one patients completed the preoperative treatment and went to surgery (95% operability rate), and 36 (84%) were completely resected. Ten of the 41 operative candidates had no evidence of tumor in the resected esophagus and nodal tissue tumor 0 node,; ToNo), 24% complete response (CR). Myelosuppression was the major toxicity with grade 3 or 4 leukopenia in 93% of patients and two preoperative treatment-related deaths. At a median follow-up of 26 months, the median survival time (MST) of all 43 patients regis-

tered on study has not been reached. The MST of the 36 completely resected patients and the 10 complete responders has not been reached; 70% and 100%, respectively, are alive at 24 months. The MST by histology is 21 months for the 22 squamous patients and has not been reached for the 21 adenocarcinoma patients registered on study. In a prognostic factor analysis, clinical N status, histology, and the percent of cisplatin and vinblastine tolerated were significant predictors for survival. These survival results suggest a significant improvement over the 14-month MST observed in our previous trial using preoperative chemotherapy only in a similar patient population, and a 12-month MST in a historic control group undergoing THE. A randomized trial is now in progress to convincingly determine if survival is prolonged by this therapy. J Clin Oncol 8:119-127. @1990 by American Society of ClinicalOncology.

THE PROGNOSIS for individuals diagnosed

1987. The first trial evaluated preoperative cisplatin, vinblastine, and mitoguazone combination chemotherapy before transhiatal esophagectomy (THE).6 The median survival time for all 34 patients entered into the trial was 14 months, not different from that observed in a series of 100 patients treated with esophagectomy alone at the same institution.7 The low complete response (CR) rate (4%) and the pattern of recurrence, 40% local-regional only, suggested that the administration of RT simultaneous with chemother-

with cancer of the esophagus is exceedingly poor. Fewer than 10% can expect to live beyond 5 years and despite advances in surgical techniques and radiation therapy these survival figures have remained virtually unchanged for over three decades.' Esophagectomy is the standard "curative" therapy for individuals with clinically localized disease. Of those who undergo resection and removal of all macroscopic disease, 5-year survival rates range from 8% to 22%.2 Radiation therapy (RT) neither as a single treatment modality nor added to resection has altered the prognosis.3' 4 However, preoperative radiation therapy may improve local control and resectability. 5 The poor survival rate with local treatment modalities reflects the natural history of this disease to disseminate early. Thus, death results not only from the complications of local-regional disease but also from distant metastases present in over 50% of patients at autopsy. In an effort to improve survival in patients with local-regional disease, we conducted two phase II combined modality trials between July 1983 and October

From the Department of Internal Medicine, Division of Hematology-Oncology, Departmentof Surgery and Department of Radiation Oncology, University of Michigan Medical Center, Ann Arbor, MI; and The Johns Hopkins Oncology Center, BiostatisticsSection, Baltimore,MD. Submitted June 12, 1989; accepted August 23, 1989. Presented in part at the American Society of Clinical Oncology Meeting, May 23, 1988, New Orleans,LA. Address reprintrequests to Arlene A. Forastiere,MD, The Johns Hopkins Oncology Center, 600 N Wolfe St, Baltimore, MD 21205. © 1990 by American Society of Clinical Oncology. 0732-183X/90/0801-0015$3.00/0

Journalof Clinical Oncology, Vol 8, No 1 (January), 1990: pp 119-127

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

119

FORASTIERE ET AL

120 apy could favorably affect outcome. This formed the basis for the design of the second trial, which is the subject of this report. Patients with squamous-cell carcinoma of the esophagus or adenocarcinoma of the esophagus, cardia, or gastroesophageal junction were treated with an intensive 21-day concurrent chemotherapy (cisplatin, vinblastine, and fluorouracil [5FU]) and RT regimen followed by THE. We chose to use cisplatin and 5-FU because of their known antitumor activity in this disease and because they are established radiation sensitizers. 8'9 Both drugs were administered by continuous infusion to increase cytotoxicity and radiation enhancement.' 0-13 The vinca alkaloid, vinblastine, has not been well tested in esophageal cancer; however, based on its comparable activity with the vinca analog vindesine, moderate activity in esophageal cancer is likely. Vindesine has a response rate of 20% to 30% in esophageal cancer.' 4 At our institution, the THE has been the preferred surgical procedure for over a decade.15 Its advantages include cervical esophageal anastomosis and avoidance of a thoractomy. The majority of patients leave the hospital able to swallow within 2 weeks of surgery. In a series of 100 patients undergoing this procedure, the overall mortality was 6%, morbidity was low, and the survival at least as good as that reported using the standard thoraco-abdominal approach.7 The THE was the planned surgical procedure for all patients entered into this trial. The objectives of this pilot trial were to determine the efficacy of concurrent preoperative chemotherapy and RT using the end point of histologic CR rate, to determine disease-free interval and survival, and to assess the toxicity of this combined modality regimen. MATERIALS AND METHODS

Patient Population Between October 1985 and October 1987, 43 patients with biopsy-proven squamous-cell carcinoma of the esophagus or adenocarcinoma of the esophagus, cardia, or gastroesophageal junction were entered onto the study. Patients were newly diagnosed and without prior treatment. To be eligible, disease had to be limited to the esophagus and regional lymph nodes, and all known sites of involvement had to be encompassable within a single (tolerable) radiation field. Celiac adenopathy was acceptable for mid- or distal third tumors because these nodes could be included in the radiation port. There could be no evidence of distant metastases, involvement of the

tracheobronchial tree, or major structures that would preclude removal of tumor. Patients had to be reasonable operative risks without cardiac or other medical contraindications to surgery. Patients were required to have a Karnofsky performance status of at least 60%, adequate bone marrow reserve (WBC > 3,500 cells/pL, platelets > 100,000 cells/ iL), and renal function (creatinine clearance >_50 mg/mL). All patients gave informed consent.

Treatment Plan Chemotherapy and RT were administered over days 1 to 21. This was followed by a 3-week rest, then THE was performed on approximately day 42. Cisplatin was adminis2 tered on days 1 to 5 and 17 to 21 in a dose of 20 mg/m /day by continuous intravenous (IV) infusion. The total dose was mixed in 1 L of D,/0.9 sodium chloride (NaC1) plus 60 mEq potassium chloride (KCI) and infused over 24 hours. An additional 80 mL/hour of Ds/0.9 NaCl was given continuously through the same IV line over the 5 days of treatment. Before starting each 5-day cisplatin infusion, patients were prehydrated with 500 mL of normal saline over 2 hours. Any preexisting dehydration was corrected before the start of treatment. No diuretics were given unless indicated by clinical signs of fluid overload. Vinblastine was administered on days 1 to 4 and 17 to 20, 1 mg/m 2/day by IV bolus. 5-FU was administered over the 21 days of treatment, 300 mg/m 2/ day by continuous IV infusion. The daily dose was mixed in 500 mL normal saline and infused over 24 hours through an infusion port or peripheral IV depending on individual patient venous access. The antiemetic thiethylperazine maleate (Torecan; Roxanne Laboratories, Inc, Columbus, OH) was used on an as-needed basis for nausea; metochlopropamide, lorazepam, and dexamethasone were added for vomiting. Antiemetics were not given prophylactically. RT was administered with a high energy linear accelerator 10 MEV unit. Patients were simulated to encompass the tumor volume with 5 cm longitudinal margins and 2 cm lateral margins. Treatment totals included the mediastinum, celiac, and/or supraclavicular nodes as the pretreatment evaluation dictated. Radiation was delivered through a threeor four-field technique and patients were set up in the prone position in order to maximize the distance between the spinal cord and esophagus. Daily fractions of 250 cGy, 5 days per week to a total of 3,750 cGy over 21 days were planned. This treatment was amended after the first 20 patients (10 squamous-cell, 10 adenocarcinoma) were treated in an attempt to further increase the histologic CR rate. Thereafter, daily fractions of 150 cGy bid, 5 days per week to a total dose of 4,500 cGy over 21 days were given. Spinal cord doses were maintained at or below 3,000 to 3,250 cGy. Treatment modifications for toxicity were as follows: for hematologic toxicity, if granulocytes were less than 1,000 cells/pL or platelets less than 50,000 cells/lL, then 5-FU, cisplatin, and vinblastine were held until counts recovered to greater than these critical values. Chemotherapy was then resumed at full doses. For gastrointestinal toxicity (Southwest Oncology Group [SWOG] criteria), grade 2 or greater stomatitis, 5-FU and vinblastine were held until toxicity resolved. For grade 2 or greater diarrhea, 5-FU was held until toxicity resolved. Stools were routinely checked for Clostridia dificile and toxin. For weight loss, if caloric intake was

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

121

CHEMOTHERAPY, RT, AND THE IN ESOPHAGEAL CA inadequate, enteral feedings were initiated via Dobhoff tube. For nephrotoxicity, creatinine clearance was rechecked before day-17 cisplatin chemotherapy was started. If the clearance decreased to 40 to 50 mL/min, cisplatin dosage 2 was reduced to 15 mg/m /day, days 17 to 21. If the clearance was less than 40 mL/min, no cisplatin was given. Approximately 21 days after completion of chemotherapy and RT, esophagectomy was performed using the technique of blunt THE without thoracotomy."5 Accessible subcarinal, periesophageal, and celiac axis nodes were routinely sampled. Before study entry, the imaging studies performed to evaluate tumor extent consisted of a chest radiograph, barium esophagram, chest and abdominal CT scans, head CT scan, and a bone scan. The barium esophagram, and the chest and abdominal CT scans were repeated during the week before THE to assess response. During treatment, complete blood cell (CBC) counts with differential and platelet count, BUN, serum creatinine, electrolytes, and magnesium were obtained at least once each week. Weight loss and other toxicities were monitored daily.

Criteriafor Response and Toxicity Clinical response to preoperative chemotherapy and radiotherapy was categorized as improvement, stable disease, or progression, and was based on a comparison of the baseline and preoperative imaging studies. Quantitation of the change in tumor size was not made due to the difficulty of obtaining accurate serial measurements from radiographic studies. After resection, patients were categorized as either surgicalpathologic staged complete responders or as having residual tumor in the resected specimen (path positive). Thus, CR was defined as the disappearance of all signs and symptoms of tumor and the absence of histopathologic evidence of residual tumor in the resected esophageal specimen and nodal tissue. Clinical improvement (I) indicated a reduction in tumor mass. Clinically stable disease (S) patients had no appreciable change in tumor mass. Progression (P) was defined as an increase in local-regional tumor mass or the appearance of new lesion(s). Toxicity was graded using SWOG criteria.

StatisticalMethods The major statistical end points of this study were survival, disease-free survival (DFS), and the influence of prognostic factors. Event times for survival were calculated from the date of registration and for DFS from the date of surgery. Event time distributions were estimated using the Kaplan6 Meier method' and compared using log-rank statistics." Variables evaluated for prognostic significance were age, gender, performance status, weight loss, RT total dose, primary (T) clinical staging, clinical nodal (N) status, histology (adenocarcinoma/squamous-cell), Barrett's epithelium, and chemotherapy dose. To study the simultaneous effect of several prognostic factors, covariates having some prognostic value (P < .15) in univariate analyses were en8 tered into a multivariate proportional hazards model.' Statistically nonsignificant factors were removed in a stepwise fashion with reestimation of parameters and significance levels at each step. Time from the start of treatment to surgery was assessed for prognostic significance as a timedependent covariate in a generalization of the proportional

hazards model.19 The risk of recurrence or death conferred by values of prognostic variables is expressed relative to a baseline reference category (hazard ratio). All P values reported are two-sided. To summarize the information gathered from multivariate analyses, a prognostic factor index was calculated for each patient. The index is a weighted average of prognostic factor values with weights taken to be the logarithms of the estimated hazard ratios from the proportional hazards model. Three risk groups of approximately equal sizes were formed for life-table summary by grouping patients according to their individual indices. RESULTS Patient Characteristicsand Response

Forty-three patients were registered on the study. Patient characteristics are detailed in Table 1. Nearly equal numbers of each histologic type were accrued. Thirty-eight percent had a weight loss of at least 10% at study entry. Seventy-nine percent of patients had a clinically staged T2 primary (lesion > 5 cm, circumferential or obstructing) using the 1983 American Joint Committee on Cancer staging criteria. 20 As expected, nearly all adenocarcinomas were distal in location; 12 were associated with Barrett's epithelium. Although nodal stage cannot be determined clinically, 24 (56%) patients had enlarged regional nodes on CT scan suggesting metastatic involvement, 15 of 21 with adenocarcinoma and Table 1. Patient Characteristics Tumor Location

Cervical Thoracic upper middistal Clinical primary stage T1 T2 T3 Clinical node status* "N negative "Npositive

Adenocarcinoma

SquamousCell

Total (%)

0

3

3 (7)

0 1 20

3 10 6

3(7) 11 (26) 26 (60)

4 17 0

4 17 1

8(19) 34 (79) 1(2)

6 15

13 9

19(44) 24 (56)

Number of patients Median age (range) Sex, male/female Race, black/white Median KPSt (range) Histology (%) Adenocarcinoma Squamous-cell

43 62 (35 to 76) 32/11 2/41 90% (60% to 90%)

*Determined by CT scan. fKarnofsky performance score.

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

21(49) 22 (51)

122

FORASTIERE ET AL

nine of 22 patients with squamous-cell histology. This included seven patients (five adenocarcinoma, two squamous-cell) with celiac axis and/ or other abdominal lymph node enlargement. Forty patients had all clinically involved nodes encompassed within the RT field. Three patients had abdominal adenopathy, the entire extent of which could not be encompassed in the radiation field. These patients were inappropriately registered on study. However, because protocol treatment was completed in all cases and metastatic disease was not confirmed by biopsy, they are included in the data analysis. All 43 patients entered on study were evaluable for toxicity and survival. There were two septic deaths during preoperative treatment, leaving 41 evaluable for response. Of these, 18 demonstrated clinical improvement, 20 had stable disease, and three patients had an incomplete preoperative assessment. At surgery, two of the 41 patients had unresectable disease, unsuspected liver metastases in one and extensive local-regional disease in the other. All other patients underwent a THE. Of these 39 patients, gross disease remained in one patient, one patient had a microscopic positive margin, and one patient had a seizure and brain metastases diagnosed within 48 hours of surgery; thus, 36 patients were disease-free postoperatively. The patient with CNS metastases was treated with high-dose corticosteroids, developed an anastomotic leak and infection, and died on postoperative day 42. Before this event, head CT scans had not been included in the staging evaluation. This was the third patient entered into the trial, which was then amended to screen for brain metastases. The results of the histopathologic examination of the resected esophagus and nodal tissue were: CR in six squamous-cell and four adenocarcinomas for a 24% (10 of 41) CR rate, 23% of all patients registered on study. Two additional patients with squamous-cell carcinoma had histologically negative esophageal specimens but positive nodes. Thus, 29% were surgically staged as To after this combined modality treatment. Pathologic staging of the primary for the 39 patients who had a THE performed was: To, 12; T1, 2; T 2, 10; and T 3, 15. Nodal tissue contained tumor in 15; however, since only a sampling of nodes accompanies the THE, negative pathology does not exclude regional node involvement. The oper-

ability rate was 95% (41 of 43), complete resectability rate was 84% (36 of 43), and overall treatment mortality was 7% (three of 43). The three patients with clinically metastatic disease to distant nodes not encompassed in the RT port completed preoperative treatment per protocol. Clinical response was improvement in two patients and stable disease in one. Each underwent THE with complete resection of gross disease; however, histopathologic examination demonstrated residual tumor in the resected esophagus and nodal tissue. Two of the three have died from metastatic disease. The two patients who did not have complete resection of their disease were treated with additional RT for palliation. All other patients were observed without further treatment and followed at prescribed intervals to document disease-free interval and survival time. Toxicity This intensive combined modality regimen was administered in the hospital for the entire 21-day treatment course. There were two treatmentrelated deaths in patients with granulocytopenia and sepsis occurring on days 26 and 28, respectively. RT was terminated early in two patients because of toxicity. All others, 41, received the entire planned dose, either 3,750 cGy or 4,500 cGy, without interruption. The major toxicity was hematologic (Table 2) with 93% of patients experiencing grade 3 or 4 leukopenia, 23% grade 3 or 4 thrombocytopenia, and 33% requiring RBC transfusions. Twenty-seven patients (63%) experienced febrile neutropenia; infection was documented in 14. Eleven patients tolerated 100% of the calculated dosage of all three drugs. Treatment with 100% of the ideal cisplatin and vinblastine doses was correlated with attainment of a histologic CR, P = .007. The mean percent of the ideal dose tolerated was: 5-FU, 84%; cisplatin, 83%; and vinblastine, 86%. This did not Table 2. Hematologic Toxicity Number of Patients (%) Leukopenia Grade 3 1,000 to 1,900 Grade 4 < 1,000 Thrombocytopenia Grade3 25to49,000 Grade 4 < 25,000

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

22 (51) 18(42) 6(14) 4 (9)

123

CHEMOTHERAPY, RT, AND THE IN ESOPHAGEAL CA

differ in relation to RT total dose and fractionation. As well, there was no difference in the incidence of grade 3 or 4 myelosuppression between patients receiving standard fractionation or hyperfractionated RT. Nonhematologic toxicity is listed in Table 3. Eighty-six percent of patients experienced esophagitis and secondary dysphagia; 79% required nutritional support. Weight loss during the preoperative treatment phase averaged 8% with eight or 19% of patients having a loss of more than 10% of their prestudy weight. Twenty-six (60%) patients had mild diarrhea (grade 1 or 2) only. Nausea and vomiting were also mild, grade 3 or 4 toxicity occurring in only two patients. Anorexia and fatigue were universal. Sixty-eight percent of patients had surgery within 1 week of the specified time (day 42). All but one patient was delayed either because of difficulty in the scheduling of operating room time or delayed recovery from bone marrow toxicity. The average length of hospital stay postoperatively was 10 days. Surgical complications were limited to anastomotic leaks in two patients. This was fatal in the one patient with brain metastases diagnosed perioperatively whose wound healing was compromised by the need for corticosteroids. Survival The outcome of all 43 patients registered on study is summarized in Table 4 according to histologic type. At a median follow-up of 26 months, 20 patients have died, 16 from disease, two from treatment-related toxicity, and two from other causes (myocardial infarction, hepatorenal syndrome). The latter two patients died at 5 and 7 months after completing treatment; there was no indication that the therapy contributed to the cause of death in either case. Twenty of the 23 patients alive are disease-free. Fifteen of the 36 completely resected patients have recurred;

Table 4. Outcome by Histology Number of Patients Adenocarcinoma Total registered Preoperative deaths Surgical candidates Unresectable, no THE Palliative THE THE, complete resection Path negative (TONo) Path positive Survival Alive, disease-free Alive with disease Dead of disease (or treatment toxicity) Dead, other causes

SquamousCell

Total

21 0 21 1 0 20 4 16

22 2 20 1 3* 16 6 10

43 2 41 2 3 36 10 26

10 2

10 1

20 3

9 0

7 (2) 2

18 2

*One with positive margin, one with gross residual disease, one with perioperative brain metastases.

Table 5 details the distribution of first recurrences according to histologic type. Sites of distant metastases were lung, liver, brain, abdominal adenopathy, testis, and multiple organs. The median survival time (MST) of all patients registered on study has not been reached with 51% alive at 36 months (95% confidence interval [CI], 0.346 to 0.656) (Fig 1). The MST and median DFS of the 36 patients who were completely resected and the 10 complete responders have not been reached. One hundred percent of CRs are alive at 36 months. One complete responder recurred with a solitary brain metastasis, which was resected and irradiated. He is without other evidence of recurrent disease. Survival and DFS data are listed in Tables 6 and 7. The MST for patients with adenocarcinoma histology has not been reached; 64% are alive at 24 months and 56% at 36 months. For squamouscell histology patients, the MST was 21 months, 49% and 48% alive at 24 and 36 months, respectively (Fig 2). These differences were not significant when compared using the log-rank test (P = .29); however, when adjusted for prognos-

Table 3. Toxicity

Table 5. Patterns of Recurrence

Number of Patients (%)

Number of Patients

Nausea and vomiting (grades 3 and 4) Esophagitis Weight loss (> 10% of prestudy weight) Nutritional support Enteral Total parenteral Renal, ototoxicity peripheral neuropathy

2 (5) 37 (86) 8 (19) 34 (79) 28 (65) 6(14) 0

Adenocarcinoma Local Distant Local and distant Total

Squamous-Cell

Total

0 9 1

1 3 1

1 12 2

10

5

15

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

124

FORASTIERE ET AL

1.00-

Table 7. Disease-Free Survival

0.75-

Complete resection Path negative (ToNo) Path positive

"" ".....

S0.50-

0)

N

Median

12 Mo

24 Mo

36 Mo

36

NR

69%

57%

56%

10 26

NR 19 mo

90% 60%

90% 43%

90% 42%

Abbreviation: NR, median not reached.

0.25-

the

Using

0.00

three

factors

prognostic

demonstrat-

'

20 Mong 0 Months

i40s s tr stica fiacto (Tab e a multivariate prognostic factor index was constructed. Three risk groups were formed from )

4

&

Fiig 1. Survival of all 43 patients registered on study. Med ian survival time has not been reached at a median folio w-up of 26 months.

tic factors (nodal status and chemotherapy dosage ), statistical significance was reached. Attainme nt of a CR had significant impact on survival and DFS times only for squamous-cell histology, P== .012 and .05, respectively.

s

statistical

independent

in

4

patient indices using cut-off values of 1.97 and

3.5 (Fig 3). A significant difference was observed in the estimated 24-month survival of patients in the most favorable risk group compared to the least favorable group, P = .003; the survival differences for the intermediate-risk group approached statistical significance, P = .06. DISCUSSION

Pr )gnosticFactorAnalysis )f the 10 factors evaluated for prognostic imlportance for survival (noted in Statistical Me :thods), N-status, histology, and the percent of the ideal dose of cisplatin and vinblastine recei' ved were statistically significant in univariate anaalyses. The percent of the ideal dosages of cislplatin and vinblastine received were highly cor related with one another (r = .942, P = .00 01) and therefore, a composite variable was for med. This composite variable had the value 1 if t he patient received 100% of both drugs and 0 oth erwise. When adjusted for N status and his tology in a multivariate Cox proportional hazards model (Table 8), the dosage indicator for cisplatin and vinblastine was highly significant, with a hazard ratio of 8.8, P < .001 for patients not tolerating full dose. Making similar adjustments, clinically positive regional nodes and squamous-cell histology were significant poor prognostic indicators with P values of .001 and .012, respectively.

At a median follow-up time of 26 months, the MST for all patients registered on this intensive combined modality pilot trial has not been reached. These survival data include five patients whose disease was not completely resected, two toxicity-related deaths, and two disease-free patients dying from other causes. Twenty-four percent of operable patients were histologic complete responders, 100% of whom are alive at 36 months and 90% of whom are free of recurrence. This suggests a significant improvement over the 14.0 month MST observed in our previous trial using cisplatin-based combination chemotherapy alone before esophagectomy. 6 Both trials were 1.00

S0.75o

a

0.50-

Log Io.·o

0.25-

Table 6. Survival

All patients Complete resection Path negative (TON o) Path positive

N

Median

12Mo

24Mo

36Mo

43 36

NR NR

69% 82%

59% 70%

51% 61%

10 26

NR 27.5 mo

100% 75%

100% 58%

100% 46%

Abbreviation: NR, median not reached.

000-

0

10

20

30

40

50

Months Fig 2. Survival of 21 adenocarcinoma and 22 squamouscell carcinoma patients. The MST for squamous-cell patients is 21 months and has not been reached for adenocarcinoma patients, P = .29 by the log-rank test and P = .01 in a multivariate Cox proportional hazards model.

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

125

CHEMOTHERAPY, RT, AND THE IN ESOPHAGEAL CA

apy dose was the strongest predictor for survival. All patients who ranked in the most favorable prognostic group, depicted in Fig 3, received

Table 8. Cox Proportional Hazards Model for Survival Hazard Ratio

Factor

P Value

95% CI

N-status*

100% of the ideal dose of cisplatin and vinblas-

Negative Positive Histology Adenocarcinoma Squamous-cell

tine. All patients in the group having the least favorable survival outcome received less than

(2.205,

1.000 7.168

.001

23.31)

1.000 3.802

.012

(1.344, 10.75)

100% dosage. Survival differences according to histology were evident in this trial when adjusted

< .001

(2.466, 31.74)

for other prognostic variables. Adenocarcinomas had a better prognosis, regardless of response to treatment, suggesting a more favorable natural

CDDP/Velban

100%

(both) < 100% (either)

1.00 8.847

NOTE. Velban (vinblastine; Eli Lilly and Company, Indianapolis, IN)

suggest that survival of the squamous-cell pa-

Abbreviation: CDDP, cisplatin. *Clinical CT assessment. COnducted

by the same core of investigators and ac crued patients with similar characteristics. Our trial differed from other combined chemoth erapy/radiotherapy preoperative protocols in th e intensity of treatment, which was completed in 3 weeks. 5-FU was administered throughout th e course of treatment by continuous infusion ra ther than over 5 days only. A total of 200 m g/m2 of cisplatin was given during the 3 weeks, aliso by continuous infusion, and vinblastine was inccorporate into the regimen. Our total dose of ra diation therapy was higher, 3,750 cGy or 4,500 cGly using twice-a-day fractionation, and was gi ven in 3 weeks (15 treatment days). Lastly, the TIHE, a low mortality, and low morbidity procedu re was used. Our mortality from this treatment waIs 7%, two preoperative treatment-related de aths and one postoperative death in a patient wi th unsuspected brain metastases. This is in m3arked contrast to mortality rates as high as 30 i%reported by other investigators.2 1 Of the prognostic factors analyzed, chemother.before :.r. ----

.75-

....

-. 2 a·s g, L

i0.50 .....-......

.... "-

0.25

........

................ 0

10

history. The slopes of the survival curves (Fig 2)

30

20

40

5b

Months Fig 3. Survival of all esophageal cancer patients based on their prognostic factor index.

tients plateaued at approximately 12 months, whereas a continuous stepwise fall in the proportion surviving is seen for adenocarcinoma patients to approximately 30 months. Although many investigators have published reports on combined modality approaches to treat local-regional esophageal cancer, few have included adenocarcinoma that has been observed to be increasing in incidence at our institution and elsewhere. Thus, our overall results, and those for the adenocarcinoma patients viewed separately, can only be compared with our own historical controls. The impressive improvement in survival suggested by this data will require a randomized controlled trial for confirmation. For the treatment of squamous-cell carcinoma of the esophagus, both preoperative chemotherapy alone and preoperative chemotherapy combined with radiation therapy have been tested since the 1970s. Kelsen et a122 from Memorial Hospital have focused on preoperative chemotherapy alone before esophagectomy in a series of trials using cisplatin-based combination chemotherapy. Their best results were achieved using cisplatin, bleomycin, and vindesine for two cycles esophagectomy. 22 A 63% partial response rate was observed in the 44 patients entered on study. The median survival time was reported only for 34 patients undergoing resection and was 16.2 months, significantly better than their historic control group. A randomized trial comparing this regimen to esophagectomny was recently reported by Roth et al.23 No survival differences were observed. Two other studies reported by the Memorial group,2 4 25 one a ran-

domized comparison of preoperative cisplatin, bleomycin, and vindesine versus preoperative

radiation therapy, reported shorter MSTs that

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

126

FORASTIERE ET AL

were not improved over historic controls. Our published results using preoperative cisplatin, mitoguazone, and vinblastine6 and the results of other investigators 26 using preoperative chemotherapy alone were similar, and led us to conclude that this single modality before esophagectomy would not have an impact on survival. The Wayne State group has piloted two regimens: mitomycin plus 5-FU, and cisplatin plus 5-FU concurrent with radiotherapy, 3,000 rad given by conventional fractionation.21 In both trials, a 26% histologic CR rate was observed for resected patients. The CR rates for all patients registered on study were slightly lower, 20% and 24%. An 18-month MST was reported for all 21 squamous-cell carcinoma patients treated with the cisplatin plus 5-FU regimen. Although the complete responders had prolonged survival, all developed recurrent disease 30 to 60 months following resection. 27 Both the SWOG and the Radiation Therapy Oncology Group (RTOG) have conducted nonrandomized trials using concurrent cisplatin, 5-FU, and RT to confirm the Wayne State data in larger numbers of patients.28' 29 A high histologic CR rate was confirmed (25% and 29% for resected patients and 17% and 20% for all evaluable patients on the SWOG and RTOG trials, respectively); however, MSTs were not improved. SWOG reported a 12-month MST and a 2-year survival of 28% for all 113 patients registered on study, and a 14-month MST for 71 patients undergoing resection. The RTOG reported a 13-month MST for all evaluable patients with only 15% alive at 2 years. In reviewing both combined modality approaches, preoperative chemotherapy alone and concurrent with radiotherapy, the addition of radiation therapy has increased the histologic CR rates achievable. In our trial thus far, only three of 15 patients have recurred locally (one local only, two local plus distant) suggesting a change in the pattern of recurrence compared with our previous trial. It is of interest that two adenocarcinoma patients developed brain metastases as the site of first recurrence. Two other patients were excluded because of brain metastases detected on the prestudy head CT scan, one squamous-cell and one adenocarcinoma. Both chemotherapy and radiation therapy may

individually be contributing to the improved survival of patients treated in our current trial compared with our previous preoperative chemotherapy protocol. Our intensive, 3-week chemotherapy regimen differed from that used in the preceding trial. Because it was designed to maximize radiation enhancement, it has not been tested as a single modality. In our data analysis, the type of radiation therapy, conventional versus hyperfractionated, was not an important factor; however, the amount of chemotherapy received clearly was. Our surgical procedure, the THE, has been used in both trials in addition to a large historic control series. Our data does suggest that when used as the sole treatment modality, survival outcome may be inferior to combined modality approaches. In conclusion, this brief intensive chemotherapy/radiotherapy protocol for the treatment of adenocarcinoma and squamous-cell carcinoma of the esophagus resulted in a 24% histologic CR rate. The MST for all patients registered on study has not been reached with an estimated 51% alive at 36 months. These results represent a significant improvement compared with our previous approach using two cycles of chemotherapy alone preoperatively, in which a 14-month MST and 4% CR rate were observed,6 and compared with a historic control group undergoing THE. 7 Although our results are very encouraging, patients entered into both of the pilot trials were highly selected. Their overall performance status was excellent and prestudy screening was carefully done to exclude patients with distant metastases or extensive, unresectable local-regional disease. This therapy was associated with considerable toxicity and required hospitalization. Therefore, a randomized trial is essential to demonstrate survival benefit for each histologic type and to justify the added risks and costs. Until such a trial is completed, we believe that combined modality approaches such as ours must be considered investigational. A randomized trial is now in progress comparing this preoperative regimen with THE. ACKNOWLEDGMENT We wish to thank Nancy B. Smith and Donna Sims for excellent secretarial assistance, Arthur Radin, MD, for his critical review of the manuscript, and Steven Piantadosi, MD, PhD, for his assistance with the biostatistical analysis.

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

127

CHEMOTHERAPY, RT, AND THE IN ESOPHAGEAL CA REFERENCES 1. Cancer Facts and Figures. American Cancer Society, 1988 2. Skinner DB: Surgical treatments for esophageal carcinoma. Semin Oncol 11:135-143, 1984 3. Earlam R, Cunha-Melo JR: Oesophageal squamous cell carcinoma: II. A critical review of radiotherapy. Br J Surg 67:457-461, 1980 4. Launois B, Delarue D, Campion JP, et al: Preoperative radiotherapy for carcinoma of the esophagus. Surg Gynecol Obstet 153:690-692, 1981 5. Hancock SL, Glatstein E: Radiation therapy of esophageal cancer. Semin Oncol 9:144-158, 1989 6. Forastiere AA, Gennis M, Orringer MB, et al: Cisplatin, vinblastine, and mitoguazone chemotherapy for epidermoid and adenocarcinoma of the esophagus. J Clin Oncol 5:1143-1149, 1987 7. Orringer M: Transhiatal esophagectomy without thoracotomy for carcinoma of the thoracic esophagus. Am Surg 200:282-288, 1984 8. Dewit L: Combined treatment of radiation and cisdiamminedichloro-platinum (II): A review of experimental and clinical data. Int J Oncol Biol Phys 13:403-426, 1987 9. Pinedo HM, Karim ABMF, Van Vliet WH, et al: Daily cis-dichlorodiam-mineplatinum (II) as a radiation enhancer: A preliminary toxicity report. J Cancer Res Clin Oncol 105:79-82, 1983 10. Rich TA, Lokich JJ, Chaffey JT: A pilot study of protracted venous infusion of 5-fluorouracil and concomitant radiation therapy. J Clin Oncol 3:402-406, 1985 11. Byfield JE, Barone R, Mendelsohn J, et al: Infusional 5-FU and x-ray therapy for non-resectable esophageal cancer. Cancer 45:703-708, 1980 12. Lokich JJ, Shea M, Chaffey J: Sequential infusional 5-fluorouracil followed by concomitant radiation for tumors of the esophagus and gastroesophageal junction. Cancer 60:275-279, 1987 13. Forastiere AA, Belliveau JF, Goren MP, et al: Pharmacokinetic and toxicity evaluation of five-day continuous infusion versus intermittent bolus cis-diamminedichloroplatinum (II) in head and neck cancer patients. Cancer Res 48:3869-3874, 1988 14. Kelsen DP, Bains M, Cvitkovic E, et al: Vindesine in the treatment of esophageal carcinoma: A phase II study. Cancer Treat Rep 63:2019-2021, 1979 15. Orringer MB, Sloan H: Esophagectomy without thoracotomy. J Thoracic Cardiovasc Surg 76:643-654, 1978 16. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958

17. Mantel N: Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 50:163-170, 1966 18. Cox DR: Regression models and life tables (with discussion). J R Stat Soc B 34:187-220, 1972 19. Kalbfleisch JD, Prentice R: The Statistical Analysis of Failure Time Data. New York, NY, Wiley, 1980 20. AJCC Manual for Staging of Cancer (Ed 2). Philadelphia, PA, Lippincott, 1983, pp 61-72 21. Leichman L, Steiger Z, Seydel HG, et al: Combined preoperative chemotherapy and radiation therapy for cancer of the esophagus: The Wayne State University, Southwest Oncology Group and Radiation Therapy Oncology Group experience. Semin Oncol 11:178-185, 1984 22. Kelsen D, Hilaris B, Coonley C, et al: Cisplatin, vindesine, and bleomycin chemotherapy of local-regional and advanced esophageal carcinoma. Am J Med 75:645-652, 1983 23. Roth JA, Pass HI, Flanagan MM, et al: Randomized clinical trial of preoperative and postoperative adjuvant chemotherapy with cisplatin, vindesine, and bleomycin for carcinoma of the esophagus. J Thorac Cardiovasc Surg 96:242-248, 1988 24. Kelsen D, Fein R, Coonley C, et al: Cisplatin, vindesine, and mitoguazone in the treatment of esophageal cancer. Cancer Treat Rep 70:255-259, 1986 25. Kelsen D, Bains M, Burt M, et al: Randomized comparison of preoperative chemotherapy versus radiation in epidermoid esophageal cancer. Proc Am Soc Clin Oncol 7:98, 1988 (abstr 374) 26. Carey RW, Hilgenberg AD, Wilkins EW, et al: Preoperative chemotherapy followed by surgery with possible postoperative radiotherapy in squamous cell carcinoma of the esophagus: Evaluation of the chemotherapy component. J Clin Oncol 4:697-701, 1986 27. Herskovic A, Leichman L, Lattin P, et al: Chemo/ radiation with and without surgery in the thoracic esophagus: The Wayne State experience. Int J Radiat Oncol Biol Phys 15:655-662, 1988 28. Poplin E, Fleming T, Leichman L, et al: Combined therapies for squamous-cell carcinoma of the esophagus: A Southwest Oncology Group Study (SWOG 8037). J Clin Oncol 5:622-628, 1987 29. Seydel HG, Leichman L, Byhardt R, et al: Preoperative radiation and chemotherapy for localized squamous cell carcinoma of the esophagus: A RTOG study. Int J Radiat Oncol Biol Phys 14:33-35, 1988

Downloaded from ascopubs.org by 80.82.78.170 on January 15, 2017 from 080.082.078.170 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.