- Email: [email protected]
Improved results of surgical treatment for esophageal and gastroesophageal junction carcinomas after preoperative combined chemotherapy and radiation
J
THORAC CARDIOVASC SURG
1988;95:415-22
Improved results of surgical treatment for esophageal and gastroesophageal junction carcinomas after preoperative combined chemotherapy and radiation Combined treatment with chemotherapy and radiation (chemoradiation) preceding surgical exploration for esophageal or gastroesophageal squamous cell carcinoma or adenocarcinoma was compared with surgical exploration alone to determine if there was an influence on tumor status at exploration, tumor resectability, disease recurrence, and patient survival. Preoperative chemoradiation resulted in significant tumor response as measured by decreased nodal involvement and 36 % incidence of no residual tumor at resection (total response) and was reflected by an improvement in resectability. Local tumor recurrence was eliminated by preoperative chemoradiation preceding resection. Distant recurrence was not reduced and remained the major cause of death. The 2-year survival rate after tumor resection alone was 33 %, versus 66% after preoperative chemoradiation and resection (p = 0.13). Patient survival after resection alone waspredicted by pathologic extentof local disease as measured by lymph node status. In contrast, survival after chemoradiation and resection was not predicted by pathologic extent of local disease. Surgical resection appears to have been an important component of therapy, primarily because survival was improved in patients after resection of residual local disease.
Steven D. MacFarlane, MD, Lucius D. Hill, MD, Philip C. Jolly, MD, Richard A. Kozarek, MD, and Richard P. Anderson, MD, Seattle. Wash.
Ew
patients with esophageal or gastroesophageal junction squamous cell carcinoma or adenocarcinoma are cured by tumor resection alone. Five-year survival rates vary from 2% to 35% depending on patient population and surgical philosophy." Survival appears to correlate with tumor status at resection as measured by local tumor penetration or lymph node status.!" Preoperative combined chemotherapy and radiation (chemoradiation) is advocated to reduce pathologic extent of local disease and thereby extend survival, as shown in other gastrointestinal carcinomas.' A retrospective analysis of all patients undergoing surgical exploration for esophageal or gastroesophageal junction
From the Section of General, Thoracic, and Vascular Surgery, the Section of Cardiothoracic Surgery, and the Section of Gastroenterology, The Virginia Mason Medical Center, Seattle, Wash. Read at the Thirteenth Annual Meeting of The Western Thoracic Surgical Association, Colorado Springs, Colo., June 24-27, 1987. Address for reprints: Steven D. MacFarlane, MD, 21700 76th Ave. W., Suite 205, Edmonds, WA 98020.
carcinoma was undertaken to determine if, when compared with no preoperative treatment, preoperative chemoradiation influenced the pathologic extent of disease and tumor resectability at exploration and disease recurrence and patient survival after tumor resection.
Patients and methods The clinical record of each patient listed with the diagnosis of esophageal, gastroesophageal junction, or gastric carcinoma by the Medical Records Department of Virginia Mason Medical Center was screened. Two hundred forty-four consecutive patients with esophageal or gastroesophageal junction carcinoma treated between July 1969 and July 1986 were identified. Demographic, operative, and pathologic data for all patients who had surgical exploration were abstracted together with a description of any adjuvant therapy used. Surgical exploration with intent to resect the tumor was undertaken in 149 of 244 patients (61% operability) and resection was accomplished in 96 patients (64% resectability). Surgical resection was considered the primary "curative" treatment modality in all eligible patients during the l7-year period of this study. Exploration in the 149 patients included 114 patients with no preoperative adjuvant therapy (No-Pre) and 13 patients
415
The Journal of
4 16
Thoracic and Cardiovascular Surgery
MacFarlane et al.
Table I. Analysis of 136 patients undergoing exploration for esophageal carcinoma, July 1969 to July 1986 Pre-Chemorad group
No-Pre group (n
Tumorcell type Adenocarcinoma Squamous cell carcinoma Tumor location GE junction Distal third Middle third Proximal third Legend:
(n
%
No.
23 64 (range 35-95)
4
No.
Patients No. of females Mean age (yr)
= 114)
26
= 22)
%
18 57 (range 37-69)
89 25
78 22
13 9
59 41
63 34 13 4
55 30
II
50 23 23 4
II
5 5
4
1
GE. Gastroesophageal.
who arbitrarily received a full course (4000 to 6000 cGy tumor dose) of preoperative radiation (Pre-Rad). In January 1983, preoperative chemoradiation (Pre-Chemorad) was introduced, and 22 patients were subsequently treated in this manner. In the Pre-Chemorad group, patients with squamous cell carcinoma were treated with cisplatin 100 mg/rn" intravenouslyon days I and 29 and 5-ftuorouracil 1000 mg/m 2/day continuous infusion on days I through 4 and 29 through 33. External beam 4 mev linear accelerator radiation treatment was begun on day I and divided into IS treatments over 19 days for a total tumor dose of 3000 cGy. Patients with adenocarcinoma were treated in the same manner except that mitomycin C 7.5 mg/rn? was added on days I and 29. Surgical exploration was performed approximately 60 days into the treatment. Postoperative radiation was receivedby 18 patients for stage III or IV disease found at resection. Fourteen No-Pre patients (22%) received 4000 to 6000 cGy mediastinal radiation. Four Pre-Chemorad patients (21%) completed mediastinal radiation (total 6000 cGy dose) after tumor resection. In the 96 patients who had tumor resection the predominant surgical technique (70 patients) was partial esophageal resection and single-stage gastric reconstruction done through an upper midline laparotomy and right thoracotomy. The resections included at least 5 em in situ proximal and distal margins and all adjacent lymph nodes. Gastric lymph nodes were removed in continuity with the proximal gastrectomy specimen. End-to-side esophagogastrostomies were generally done with an interrupted suture technique. Twenty-six patients had total esophagectomy. Cervical esophagogastrostomy was performed in four patients. Sixteen patients had esophageal reconstruction by colon interposition: two as a single-stage and 14 as a multiple-stage procedure. Six patients died before planned staged reconstruction. Complete follow-up of vital status was obtained in 95 of the
96 patients who underwent resection (99%) and was determined as of March 1987. Status "alive" was confirmed by telephone to patients or their personal physicians. One patient who underwent resection was lost to follow-up at 2 months. Statistical analysis of discrete variables was by Fisher's exact test. Survival and recurrence data were generated by life-table method and compared by the Mantel-Cox test with the BDMPIL computer program (BDMP Statistical Software, Berkeley, Calif.) or log-rank test. The BDMP2L program with Cox's proportional hazard model was used to test the joint effect of patient demographics and tumor characteristics on survival and recurrence. The number of patients (13) in the Pre-Rad group was too small to make any meaningful conclusions. Therefore, this study focuses on the No-Pre and Pre-Chemorad patient groups.
Results Analysis of patient demographics revealed no significant age or sex differences between No-Pre and PreChemorad patients (Table I). Analysis of patients who had tumor resected in each group also showed no age or sex differences. Further analysis of patients who had tumor resected revealed that squamous cell carcinoma occurred more often in Pre-Chemorad than No-Pre patients (47% versus 22%, p < 0.05) and that middlethird esophageal tumor location occurred in more PreChemorad than No-Pre patients (26% versus 6%, p < 0.05). No other significant differences including extent of esophageal resection were noted between patients undergoing resection in the two treatment groups.
Tumor resection and operative pathologic findings. Tumor resection was accomplished in significantly more Pre-Chemorad than No-Pre patients (Table II). No residual disease in the surgical specimen (total response) was found in 36% of Pre-Chemorad patients. Analysis of lymph node status revealed a significantly higher percentage of patients with disease-free lymph nodes in the Pre-Chernorad group (59% versus 25%, p < 0.05). The remaining results that follow concern only the 83 patients in both groups who had resectable tumor.
Operative mortality and morbidity after tumor resection. Operative death after resection occurred in four of 64 No-Pre patients (6.3%); causes of death were acute pancreatitis, subphrenic abscess and sepsis, mediastinitis and sepsis, and pneumonia. Operative death after resection occurred in two of 19 Pre-Chernorad patients (10.5%, p = NS*); causes of these deaths were cardiac arrest and pneumonia. Significant postresection cardiac, pulmonary, or gastrointestinal complications occurred during hospitalization in 22 of 60 (37%)
*NS = Not significant.
Volume 95
Esophageal carcinoma 4 1 7
Number 3
March 1988
Table II. Pathologic findings and tumor resectability in 136 patients, July 1969 to July 1986 No-Pre group (n = 114) Adenocarcinoma (n = 89)
Tumor stage No residual (stage 0) Stage I Stage II Stage III Stage IV Lymph node status No tumor Resected 'p tp
Pre-Chemorad group (n = 22)
Squamous (n = 25)
Squamous (n = 9)
Adenocarcinoma (n = 13)
No.
%
No.
%
No.
%
No.
%
0 3 4 50 32
0 2 2 15 6
8 8 60 24
3 0 2 4 4
23*
3 5 56 36
56* 11
15 31 31
5 1 0 3 0
22 50
25 56
14
32 56
5 10
38 77
8 9
7
33
89*
root
< 0.01 versus No-Pre group. < 0.05 versus No-Pre group.
No-Pre versus four of 17 (24%) Pre-Chemorad operative survivors (p = NS). Anastomotic leak occurred in five of 60 (15%) No-Pre versus none of 17 PreChemorad patients (p = NS). Anastomotic stricture necessitating dilatation occurred in nine of 60 (15%) No-Pre versus five of 17 (29%) Pre-Chemorad patients (p = NS) with one death in the Pre-Chemorad group that resulted from operative revision of a stricture nine months after colon interposition. Patient survival after tumor resection. Mean followup was 27 months (range 0.4 to 138 months) in No-Pre and 17.6 months (range 1.2 to 51 months) in PreChemorad patients. Seven No-Pre patients (one with squamous cell, six adenocarcinoma) were alive at mean follow-up of 62 months (range 24 to 137 months). Eleven Pre-Chemorad patients (six with squamous cell, five adenocarcinoma) were alive at a mean follow-up of 21 months (range 8 to 51 months). Five of the 11 had had no evidence of disease at resection (three with squamous cell, two adenocarcinoma). A trend toward improved survival after preoperative chemoradiation was observed. Two- and 3-year survival rates were 66% versus 32% and 33% versus 21%, respectively, in Pre-Chemorad and No-Pre patients (Fig. 1). However, this improvement could have been due to chance alone (by Mantel-Cox analysis, p = 0.13; by Cox model analysis adjusting survival curves for age, sex, tumor cell type, and postoperative radiation therapy as covariates, p = 0.15). Tumor status at resection and survival in treatment groups. Life-table analysis of survival in relationship to lymph node status revealed a highly significant correlation in the No-Pre group. The 5-year survival rate was
22% in 29 patients with normal lymph nodes versus 6% in 35 patients with diseased lymph nodes (p < 0.01). Survival in relationship to tumor cell type in the No-Pre group revealed no significant difference between patients with squamous cell carcinoma and those with adenocarcinoma (p = 0.52). In contrast, analysis of survival in relation to lymph node status failed to show a significant correlation in the Pre-Chemorad group. The 2-year survival rate ws 66.6% in six patients with diseased lymph nodes versus 67.3% in 13 patients with normal lymph nodes (p = 0.48). In addition, comparison of Pre-Chemorad patients by means of another measure of tumor status, namely presence or absence of local residual tumor at resection, also revealed no significant correlation, as the 3-year survival rate was 33% in each group (p = 0.58). Analysis of postresection survival in relationship to American Joint Committee staging" was not meaningful, as too few patients had stage I or II disease (Table 11). Recurrence after tumor resection. Tumor recurrence was noted in 50 of 77 (65%) resection survivors. Nine of 60 (15%) No-Pre resection survivors had isolated anastomotic or mediastinal recurrences on the basis of endoscopic, computed tomographic, and autopsy findings. No local recurrences were found in 17 Pre-Chemorad resection survivors for the duration of follow-up, which suggested complete local control of disease. Thirty-four of 60 (57%) No-Pre and seven of 17 (41%) Pre-Chemorad resection survivors had distant tumor recurrence in the liver, lungs, lymphatics, bone, or brain. Life-table analysis of distant recurrence revealed 1- and 2-year rates of 41% versus 33% and 57% versus
The Journal of
4 18
MacFarlane et al.
Thoracic and Cardiovascular Surgery
100 - - No-Pre, N=64 0··········0
80
-
O··J..···J···········JL .. .L•....•..••...•••.
C
U
~
Pre-Chemorad, N=19
9
60
'
35
32
Q.
31
:3
29
0"':
28
40
,2
1
Q················J··············o
p
= 0.13
13 1~---.~_--L.-&..,.11 0
20
14
8 7
o
o
6
12
18
24
30 36 Months
42
48
54
60
Fig. 1. Survival rates after resection of squamous cell carcinoma or adenocarcinoma of esophagus in 83 patients; resection alone (No-Pre) versus preoperative chemoradiation and resection (Pre-Chemorad). Numbers on graph are patients at risk. Tick marks designate patients lost to follow-up (first tick on No-Pre graph) or alive at last follow-up (remaining tick marks).
43%, respectively, in No-Pre and Pre-Chernorad patients (p = 0.48). When distant recurrence rates were adjusted for age, sex, tumor cell type, and postoperative radiation treatment as covariates by the Cox model, no significant difference was obtained (p = 0.22). The number of patients who received preoperative radiation treatment alone (Pre-Rad) was small (13 patients), precluding a meaningful analysis of this group. This group appeared to share characteristics of both No-Pre and Pre-Chemorad patients. As in the Pre-Chemorad group, resectability was high (100%), lymph node status was favorable (54% normal nodes), and local recurrence was rare (one patient, 8%). However, life-table analysis of survival showed no significant improvement compared to survival in the No-Pre group (p = 0.33).
Discussion Patient survival after tumor resection with no preoperative treatment correlated well with the pathologic extent of disease as measured by lymph node status. Five-year survival rates were 22% and 6%, respectively, in patients with normal and diseased lymph nodes, (p < 0.01). This correlation is consistent with the observations of Akiyama and co-workers' who reported 5-year survival figures after resection of 53.8% for patients with normal lymph nodes and 15.3% for those
with diseased lymph nodes. Survival in our surgery alone patients was not influenced by tumor cell type. Lymph node status was the most important pathologic prognostic factor after resection alone regardless of whether the tumor was squamous cell carcinoma or adenocarcinoma. Pathologic extent of disease was improved by preoperative chemoradiation on the basis of two findings. First, significant improvement of lymph node status was shown. Second, total response (ie, no residual disease) occurred in 36% of Pre-Chemorad patients. These improvements were similar for both patients with squamus cell carcinoma and those with adenocarcinoma. A total response rate of 56% for squamous cell carcinoma (Table II) compares favorably with the findings of Leichman and associates," who reported a 33% total response rate, and of Poplin and associates'? (Southwest Oncology Group study), who reported a 24% total response rate at exploration with the same preoperative treatment protocol. Tumor regression induced by adjuvant therapy is the most likely reason for improved resectability in Pre-Chemorad patients, although unidentified selection biases may be contributory. On the basis of the observation that survival is related to pathologic extent of disease and the finding that preoperative chemoradiation results in improved tumor status, one should expect improvement in survival.
Volume 95 Number 3 March 1988
Analysis of survival in No-Pre and Pre-Chemorad groups did, in fact, reveal a trend toward significant survival improvement after preoperative chemoradiation. However, the relationship between local tumor status at resection and survival, as noted in the No-Pre patients, was not maintained in Pre-Chemorad patients. Pre-Chemorad patients with normal lymph nodes or total tumor response had a survival rate similar to that of patients with abnormal nodes or resected residual disease. This similarity was unexpected and is at variance with the Southwest Oncology Group findings in which patients with total response had a 40% 3-year survival rate and patients with resected residual disease had only a 12% 2-year survival rate." A likely explanation for failure to find a relationship between local tumor status and survival after preoperative chemoradiation may lie in the nature of tumor recurrence and our inability to stage esophageal carcinoma accurately. Preoperative chemoradiation clearly resulted in adequate local control of disease, as no local recurrences were noted in Pre-Chemorad patients. However, distant metastatic disease, not local recurrence, was ultimately responsible for the majority of deaths. Moreover, preoperative chemoradiation did not significantly reduce the distant recurrence rate. Nondetection of coexisting systemic disease may result in understaging of disease and, therefore, an overestimation of response to therapy, especially in patients with no residual tumor or normal lymph nodes. The importance of surgical resection has been questioned. First, the previously cited findings by Poplin and associates10 have shown that survival remains poor despite resection of residual disease. Second, surgical resection may not improve the status of patients with no residual disease. Given that these two assumptions are true, one could argue that operative risks are too high. However, the evidence suggests that the first assumption is not valid and the second is irrelevant. Our results after preoperative chemoradiation show that the improvement in survival (although possibly due to chance alone) after resection of residual disease was similar to that after total tumor response. If resection were to be eliminated, two thirds of our patients (those with residual disease) would be deprived of local disease control and potential prolongation of survival. The assumption that surgical resection may not improve the status of patients with no residual disease cannot be refuted by this analysis. However, the assumption itself should not be at issue, but rather the inability to exclude the presence of residual local disease without resection. Campbell and associates" have shown that clinical restaging does not correlate with surgical pathologic
Esophageal carcinoma
4I9
findings. Furthermore, this group reported only a 20% 2-year survival rate after chemoradiation in patients with clinically undetectable and unresected local disease. One third of these patients had local recurrences. These results contrasted with a 50% 2-year survival rate with no local recurrences when resection was done. Preoperative chemoradiation appears to result in significant local disease response and improved resectability for both squamous cell carcinoma and adenocarcinoma. The trend toward significant improvement in survival in this series is encouraging and will become more meaningful with longer follow-up. A prospective, randomized trial would, admittedly, provide a clearer answer to these questions but would be difficult given the compromises inherent in the treatment of patients with esophageal cancer. A larger clinical trial, subjected to multivariate analysis, could also provide more certainty to the conclusions. The application of multimodality therapy to adenocarcinoma is relatively recent. Our observation of similar responses by both cell types is supported by the recent, early observations of Wolfe and colleagues" and warrants the inclusion of patients with adenocarcinoma in further studies of multimodality therapy. Surgical resection of esophageal carcinoma should remain an integral part of curative therapy. It is an important aspect of local disease control, and early analysis suggests a trend toward improved survival after resection of residual disease. The solution to the continuing problem of distant recurrence requires the introduction of more effective systemic therapy. The work of Gloria Bailey, PhD, of the Virginia Mason Clinical Research Support Services for help with statistical analysis and of Bette Glass in the preparation of this manuscript is appreciated. 1.
2. 3. 4. 5.
6.
REFERENCES Akiyama H, Tsurumaru M, Kawamura T, Ono y. Principles of surgical treatment for carcinoma of the esophagus. Ann Surg 1981;194:438-46. Dark JF, Mousalli H, Vaughan R. Surgical treatment of carcinoma of the esophagus. Thorax 1981 ;36:891-5. Ellis FH, Gibb SP. Esophagogastrectomy for carcinoma: current hospital mortality and morbidity rates. Ann Surg 1979;190:699-705. Galanduik S, Herman RE, Gassman 11, Cosgrove OM. Cancer of the esophagus: the Cleveland Clinicexperience. Ann Surg 1986;203:101-8. Fein R, Kelson DP, Geller N, Bains M, McCormack P, Brennan MF. Adenocarcinoma of the esophagus and gastroesophageal junction: prognostic factors and results of therapy. Cancer 1985;56:2512-8. Gatzinsky P, Berglin E, Dernevik L, Larsson I, William-
The Journal of
420
7.
8. 9.
10.
11.
12.
MacFarlane et al.
Olsson G. Resectional operations and long-term results in carcinoma of the esophagus. J THORAC CARDIOVASC SURG 1985;89:71-6. Michaelson RA, Magill GB, Quan SHQ, et al. Preoperative chemotherapy and radiation therapy in the management of anal epidermoid carcinoma. Cancer 1983;51: 390-5. Beahrs 0, Myers MH, eds. Manual for staging of cancer. Philadelphia: JB Lippincott Co, 1983. Leichman L, Steiger Z, Seydel HG, Vaitkevicius VK. 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 1984; II :178-85. Poplin E, Leichman L, Seydel HG, Steiger L, Fleming C. SWOG 8073: combined therapy for squamous cell carcinoma of the esophagus (SCCE). Proc Am Soc Clin Oncol 1986;5:80. Campbell WR, Taylor SA, Pierce GE, Hermreck AS, Thomas JH. Therapeutic alternatives in patients with esophageal cancer. Am J Surg 1985;150:665-8. Wolfe WG, Burton GV, Seigler HF, Crocker IR, Vaughn AL. Early results with combined modality therapy for carcinoma of the esophagus. Ann Surg 1987;205:56371.
Discussion Dr. Robert L. Mitchell (Mountain View, Calif).The authors have thoroughly analyzed their data on a combined modality treatment of cancer of the esophagus in 22 patients and compared this with resection alone in 114 patients treated in a previous period. The purpose of the paper is to answer whether combined treatment affects tumor status at exploration, resectability, disease recurrence, and patient survival. Combined treatment unequivocally altered tumor status as, at operation, 36% of the patients had no residual tumor. Resectability appears to be improved, from 56% to 77% for adenocarcinoma and 100% for squamous cell carcinoma. However, the criteria for resectability are not defined, and this is a retrospective study over 17 years. Other recent series, including my own, have reported a high resectability rate without adjuvant therapy. In the combined modality group, local recurrence was zero, but I think it premature to attribute this solely to chemoradiation preoperatively. Survival rates were not statistically altered by combined therapy. However, the authors point out that survival trends favor a combined modality treatment. The obvious danger in this comparative study of two treatment modality groups is drawing major conclusions when the study is not prospective or randomized and contains only 22 patients in the chemoradition group. We have, however, learned from this study that combined chemoradiation affects tumor status at operation and appears to decrease local recurrence. It is interesting that other combined treatment protocols for lung and rectum cancers have also shown a reduction in local recurrence, but survival rates have essentially been unchanged. The question arises whether operation is necessary after chemoradiation, as you have already addressed: Are those
Thoracic and Cardiovascular Surgery
patients who are tumor-free after chemoradiation not improved by operation, and those who have residual tumor doomed anyway? I think for the present, until more data are available, I would concur with the authors that operation is essential to stage the disease, to reduce local recurrence and, we hope, to improve survival rates in the group that has residual tumor. I compliment the authors for studying new modalities of therapy for a disease of high morbidity and mortality with and without treatment. Plaudits are in order for achieving a low morbidity and mortality in your series. Indeed there were no anastomotic leaks in the chemoradiation group. Although not precisely stated in the paper, I surmise that combined treatment is well tolerated by the patients. Possibly the patients were in a more optimum state for resection after this particular chemoradiation protocol. As you know, some previous adjuvant therapy studies have shown that the treatment was worse than the disease. Wolfe and his co-workers at Duke University, as you pointed out, reported in the May issue of the Annals of Surgery on 63 patients treated with combined therapy. All patients showed, in their analysis, a marked decrease in tumor mass and improved swallowing, and became anabolic on oral nutrition. These preliminary studies are encouraging in a disease with a dismal outlook, but must be supported with controlled prospective studies. I have four questions for the authors. First, is the choice of treatment regimen (that is, drugs and radiation) based on scientific data or empiricism? Dr. MacFarlane. It is empirically based on scientific data. Basically, the choice derives from studies that have looked at cloacogenic carcinoma and have shown effectiveness in terms of reduction of tumor mass. At the onset, these studies can really focus only on initial, short-term results to evaluate tumor response. In terms of survival, the long-term data remain to be seen, but for cloacogenic tumors, it appears that there is an improvement in survival, and consequently, this treatment was applied to tumors of the esophagus. Ours is one of the first centers to apply this protocol to adenocarcinoma with the addition of mitomycin C, although the Duke University study now also shows that on early follow-up, adenocarcinoma and squamous carcinoma appear to respond similarly. That is one aspect of this study that is rather exciting. I think we ought to include patients with gastroesophageal junction tumors in further studies of multimodality treatment protocols. Dr Mitchell. Did chemoradiation improve nutrition, as Wolfe and his co-workers have reported? Dr. MacFarlane. The Duke University experience described by Wolfe and associates was surprising, because they were very optimistic about the status of their patients after chemoradiation, although they did not quantify this in any way. They said that after patients received chemotherapy and radiation, they were in better condition for operation than those who did not have preoperative treatment. That is not my empirical impression. I think it does take its toll. We have evaluated white count and that does, in fact, go down, and sometimes requires precautionary hospitalization. The protocol at Virginia Mason is to admit patients for each cycle of chemotherapy. The Duke protocol was quite a bit different from ours, because they used cisplatin and VP-16 rather than cisplatin and 5-ftuorouracil and didn't start radiation until after two cycles of chemotherapy, while we initiated radiation at the beginning. Dr. Mitchell. Dr. Daniel, of Charlottesville, Virginia,
Volume 95 Number 3
March 1988
reported a 38% incidence of nondilatable stricture after chemoradiation before operation. You reported a 29% incidenceof stricture in the combined group after operation. Did yousee a problem with significant stricture with the chemoradiation group before operation? Dr. MacFarlane. Our stricture rate was similar in both groups. It was a little bit higher in the chemoradiation group, to be sure, and whether this relates to the type of operation performed or not is not clear. One was a cervical anastomotic stricture outside the radiation field after total replacement withcolon interposition, so that it did not seem to relate to the preoperative treatment. I do not think we have sufficient numbers to really evaluate cause and effect. Our impression was, first of all, that the operative morbidity was similar in the twogroups, and that anastomotic problems, specifically leaks, did not occur in the chemoradiation group. Benign strictures after combined treatment, aside from one case, were easily dilatedand frequency was similar to that after operation alone. Stricture after chemoradiation, but before operation, was not a problem. Dr. Mitchell. My last question is, is there a place for chemoradiationin the patient whose cancer is nonoperable and who has almost total obstruction of the esophagus? Dr. MacFarlane. Wolfe and co-workers address that question also, and say, in fact, that they get excellent palliative results. Although it was not emphasized in this discussion, we believe that this treatment protocol is very effective as a palliative measure, irrespective of the survival results. In fact, when you can eradicate the local tumor burden and prevent recurrence, these patients, even though they go on to have distant metastases, do not have the debilitating obstructive symptoms. The advent of laser therapy has also provided us with another useful tool for palliation along with preoperative adjuvant treatment. In other words, if the esophagus is almost obstructed, it is very appropriate to use laser excavation to reestablish the lumen and then to place the patient on this combined modality protocol. We have done this once or twice with good results. Dr. James B. D. Mark (Stanford. Calif). We have been using a preoperative combined chemoradiation protocol as well, but just for squamous carcinoma. I was interested to hear about your nice results with adenocarcinoma. My question is this: You have 22 patients on whom you operated after chemoradiation. If your experience is anything like ours, more patientsthan that started a chemoradiation protocol, but never got to operation, usually because of extensive disease or debilitation. Can you give us an idea how many patients began treatment but were never operated on? Dr. MacFarlane. Many patients are initially treated by the oncologists before, and often without, surgical consultation. One of our oncologists has provided this data for adenocarcinomas only. Approximately 15% to 20% do not go on to operation, according to his figures, but some patients were put on the protocol after it had been determined on the basis of initial clinical work-up that the lesions were not resectable. It is hard to get a good grip on what the actual denominator is and who would have been a true surgical candidate to begin with. Dr. Mark. Ideally, it would be a good idea to get in at the beginning with all of these patients, so that the surgeon is on the original staging and decision making. What I think is happening in a number of diseases these days with aggressive pulmonologists, oncologists, and others is that we never see
Esophageal carcinoma
421
some patients with cancer of one kind or another whom we could help, because decisions are made for them by nonsurgeons. Dr. MacFarlane. I agree, and that was one of the reasons for presenting this paper. Dr. Murray I. Sheldon (Concord. Calif). I appreciated that the initial impetus for your study was that your oncologists believed that with the good results of chemoradiation this perhaps would be the only treatment necessary and they wished to drop the surgical arm of the treatment protocol. We have a similar problem at our hospital. Your study evaluated combined chemoradiation plus operation and compared it to operation alone. Do you have any data that I can take back to my hospital that suggest that the combination of chemoradiation and operation is, in fact, beneficial in short- and long-term survival compared with your chemoradiation group alone? Dr. MacFarlane. No, I really do not. The problems are that, first, we do not have the ability to accurately determine the stage of the disease when the patient comes for treatment and, second, as in any nonsurgical arm, we do not have an ability to quantitatively stage the disease after treatment. As you all know, many of these patients have endoscopic and computed tomographic studies and the disease appears to be gone. Then at resection of the specimen a microscopic residual is found. Until we can develop some accurate clinical means to analyze the tumor response, I do not think there is any basis for comparison between surgical and nonsurgical treatment other than just survival. I do not have any good data on survival after nonsurgical chemoradiation treatment. I hope that our data, at the very least, based on local tumor response, suggestion of improved survival rates, and maintenance of acceptable operative mortality and morbidity rates, has shown that there is no basis for the removal of surgical resection as an integral part of treatment. Dr. Sheldon. Are your oncologists publishing the data of chemoradiation alone in any way compared with yours? Dr. MacFarlane. No, only a small number of patients who received chemoradiation did not go to operation, and their survival rate was not very good compared with that of the group having surgical resection. The reason they did not go to operation, however, was that they had some untoward event. Dr. Alden H. Harken (Denver. Colo.). Dr. MacFarlane's sophisticated surgical group is asking a question to which we in the audience, I think, are all sympathetic. It is always easy to poke holes in nonprospective, nonrandomized, multi-institutional trials. Obviously, one of the problems with this study is that it is retrospective and the group sizes differ greatly: The discrepancy between 22 in one group and 114 in the other is bothersome. However, it seems to me that Dr. MacFarlane has raised our awareness of a significant clinical issue: Is there a position for surgery in the therapy of carcinoma of the esophagus? Although our oncologist friends may be building a case that there is not, it seems to me that we must address some of the advantages of surgical therapy, and I believe there are advantages. I would ask whether you have elevated some of them. You persuasively make the case that there is a surgical benefit. In this group of patients we see in the good group a 22% 5-year survival rate and in the bad group a 26% 5-year survival rate, but what happens to the quality of life during that period of time? Is there any way of addressing or assessing the issue of palliation by itself? My bias is that the patients can swallow if they are operated on, and they cannot if not operated on. Did you assess that, and are there any other
The Journal of
422 MacFarlane et ai.
measurable, objective parameters such as anabolism, immunologic response, or ability to swallow that we can look at? Dr. MacFarlane. In this study we compared two surgical groups, so palliation rates are similar in the two groups. The gold standard for palliation in this study is in the 114 patients having operation only. In those patients, we achieved approximately a 92% overall palliation rate, including those who required some repeat dilation. If those were not included, the rate was about 85% palliation. Using that as the gold standard, we have shown that palliation rates were simialr after preoperative treatment and resection. The only other objective information obtained, but not quantified, was my impression in talking to these patients that it took longer to gain weight after the multimodality therapy than it did after operation alone: approximately 2 months compared with 4 weeks after operation alone. We are paying a small price there, but once the patients begin to gain weight, they do well. If they begin to lose weight again, that is a real marker for a recurrence.
Thoracic and Cardiovascular Surgery
Dr. Ivan A. May (Oakland, Calif). Dr. Harken's statement that if you operate on these patients they can swallow and if you do not they cannot swallow is true in my experience as well. Dr. Walter B. Cannon (Palo Alto, Calif). Have you. in the review of all these patients, seen a difference in response in a cervical esophageal cancer versus a middle or lower-third esophageal cancer? Dr. MacFarlane. Only one or two multi modality patients had upper-third cancers. Many of the patients who have upper third cancers we do not evaluate because of our separte approach (medical versus surgical). I agreed with the previous comments of Dr. Mark that there needs to be a consensus on how we screen these patients. I think the surgeon should be included early on, and he often is not. That is one reason we have not been able to analyze your question, Dr. Cannon, because we have seen too few patients with proximal-third tumors.
Bound volumes available to subscribers Bound volumes of THE JOURNAL OF THORACIC AND CARDIOVASCULAR SURGERY are available to subscribers (only) for the 1988 issues from the Publisher, at a cost of $51.00 ($68.00 international) for Vol. 95 (January-June) and Vol. 96 (July-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed. Copies are shipped within 60 days after publication of the last issue of the volume. The binding is durable buckram with the JOURNAL name. volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact The C. V. MosbyCompany, Circulation Department, 11830 Westline Industrial Drive, St. Louis, Missouri 63146, USA; phone (800) 325-4177, ext. 351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular JOURNAL subscription.
THORAC CARDIOVASC SURG
1988;95:415-22
Improved results of surgical treatment for esophageal and gastroesophageal junction carcinomas after preoperative combined chemotherapy and radiation Combined treatment with chemotherapy and radiation (chemoradiation) preceding surgical exploration for esophageal or gastroesophageal squamous cell carcinoma or adenocarcinoma was compared with surgical exploration alone to determine if there was an influence on tumor status at exploration, tumor resectability, disease recurrence, and patient survival. Preoperative chemoradiation resulted in significant tumor response as measured by decreased nodal involvement and 36 % incidence of no residual tumor at resection (total response) and was reflected by an improvement in resectability. Local tumor recurrence was eliminated by preoperative chemoradiation preceding resection. Distant recurrence was not reduced and remained the major cause of death. The 2-year survival rate after tumor resection alone was 33 %, versus 66% after preoperative chemoradiation and resection (p = 0.13). Patient survival after resection alone waspredicted by pathologic extentof local disease as measured by lymph node status. In contrast, survival after chemoradiation and resection was not predicted by pathologic extent of local disease. Surgical resection appears to have been an important component of therapy, primarily because survival was improved in patients after resection of residual local disease.
Steven D. MacFarlane, MD, Lucius D. Hill, MD, Philip C. Jolly, MD, Richard A. Kozarek, MD, and Richard P. Anderson, MD, Seattle. Wash.
Ew
patients with esophageal or gastroesophageal junction squamous cell carcinoma or adenocarcinoma are cured by tumor resection alone. Five-year survival rates vary from 2% to 35% depending on patient population and surgical philosophy." Survival appears to correlate with tumor status at resection as measured by local tumor penetration or lymph node status.!" Preoperative combined chemotherapy and radiation (chemoradiation) is advocated to reduce pathologic extent of local disease and thereby extend survival, as shown in other gastrointestinal carcinomas.' A retrospective analysis of all patients undergoing surgical exploration for esophageal or gastroesophageal junction
From the Section of General, Thoracic, and Vascular Surgery, the Section of Cardiothoracic Surgery, and the Section of Gastroenterology, The Virginia Mason Medical Center, Seattle, Wash. Read at the Thirteenth Annual Meeting of The Western Thoracic Surgical Association, Colorado Springs, Colo., June 24-27, 1987. Address for reprints: Steven D. MacFarlane, MD, 21700 76th Ave. W., Suite 205, Edmonds, WA 98020.
carcinoma was undertaken to determine if, when compared with no preoperative treatment, preoperative chemoradiation influenced the pathologic extent of disease and tumor resectability at exploration and disease recurrence and patient survival after tumor resection.
Patients and methods The clinical record of each patient listed with the diagnosis of esophageal, gastroesophageal junction, or gastric carcinoma by the Medical Records Department of Virginia Mason Medical Center was screened. Two hundred forty-four consecutive patients with esophageal or gastroesophageal junction carcinoma treated between July 1969 and July 1986 were identified. Demographic, operative, and pathologic data for all patients who had surgical exploration were abstracted together with a description of any adjuvant therapy used. Surgical exploration with intent to resect the tumor was undertaken in 149 of 244 patients (61% operability) and resection was accomplished in 96 patients (64% resectability). Surgical resection was considered the primary "curative" treatment modality in all eligible patients during the l7-year period of this study. Exploration in the 149 patients included 114 patients with no preoperative adjuvant therapy (No-Pre) and 13 patients
415
The Journal of
4 16
Thoracic and Cardiovascular Surgery
MacFarlane et al.
Table I. Analysis of 136 patients undergoing exploration for esophageal carcinoma, July 1969 to July 1986 Pre-Chemorad group
No-Pre group (n
Tumorcell type Adenocarcinoma Squamous cell carcinoma Tumor location GE junction Distal third Middle third Proximal third Legend:
(n
%
No.
23 64 (range 35-95)
4
No.
Patients No. of females Mean age (yr)
= 114)
26
= 22)
%
18 57 (range 37-69)
89 25
78 22
13 9
59 41
63 34 13 4
55 30
II
50 23 23 4
II
5 5
4
1
GE. Gastroesophageal.
who arbitrarily received a full course (4000 to 6000 cGy tumor dose) of preoperative radiation (Pre-Rad). In January 1983, preoperative chemoradiation (Pre-Chemorad) was introduced, and 22 patients were subsequently treated in this manner. In the Pre-Chemorad group, patients with squamous cell carcinoma were treated with cisplatin 100 mg/rn" intravenouslyon days I and 29 and 5-ftuorouracil 1000 mg/m 2/day continuous infusion on days I through 4 and 29 through 33. External beam 4 mev linear accelerator radiation treatment was begun on day I and divided into IS treatments over 19 days for a total tumor dose of 3000 cGy. Patients with adenocarcinoma were treated in the same manner except that mitomycin C 7.5 mg/rn? was added on days I and 29. Surgical exploration was performed approximately 60 days into the treatment. Postoperative radiation was receivedby 18 patients for stage III or IV disease found at resection. Fourteen No-Pre patients (22%) received 4000 to 6000 cGy mediastinal radiation. Four Pre-Chemorad patients (21%) completed mediastinal radiation (total 6000 cGy dose) after tumor resection. In the 96 patients who had tumor resection the predominant surgical technique (70 patients) was partial esophageal resection and single-stage gastric reconstruction done through an upper midline laparotomy and right thoracotomy. The resections included at least 5 em in situ proximal and distal margins and all adjacent lymph nodes. Gastric lymph nodes were removed in continuity with the proximal gastrectomy specimen. End-to-side esophagogastrostomies were generally done with an interrupted suture technique. Twenty-six patients had total esophagectomy. Cervical esophagogastrostomy was performed in four patients. Sixteen patients had esophageal reconstruction by colon interposition: two as a single-stage and 14 as a multiple-stage procedure. Six patients died before planned staged reconstruction. Complete follow-up of vital status was obtained in 95 of the
96 patients who underwent resection (99%) and was determined as of March 1987. Status "alive" was confirmed by telephone to patients or their personal physicians. One patient who underwent resection was lost to follow-up at 2 months. Statistical analysis of discrete variables was by Fisher's exact test. Survival and recurrence data were generated by life-table method and compared by the Mantel-Cox test with the BDMPIL computer program (BDMP Statistical Software, Berkeley, Calif.) or log-rank test. The BDMP2L program with Cox's proportional hazard model was used to test the joint effect of patient demographics and tumor characteristics on survival and recurrence. The number of patients (13) in the Pre-Rad group was too small to make any meaningful conclusions. Therefore, this study focuses on the No-Pre and Pre-Chemorad patient groups.
Results Analysis of patient demographics revealed no significant age or sex differences between No-Pre and PreChemorad patients (Table I). Analysis of patients who had tumor resected in each group also showed no age or sex differences. Further analysis of patients who had tumor resected revealed that squamous cell carcinoma occurred more often in Pre-Chemorad than No-Pre patients (47% versus 22%, p < 0.05) and that middlethird esophageal tumor location occurred in more PreChemorad than No-Pre patients (26% versus 6%, p < 0.05). No other significant differences including extent of esophageal resection were noted between patients undergoing resection in the two treatment groups.
Tumor resection and operative pathologic findings. Tumor resection was accomplished in significantly more Pre-Chemorad than No-Pre patients (Table II). No residual disease in the surgical specimen (total response) was found in 36% of Pre-Chemorad patients. Analysis of lymph node status revealed a significantly higher percentage of patients with disease-free lymph nodes in the Pre-Chernorad group (59% versus 25%, p < 0.05). The remaining results that follow concern only the 83 patients in both groups who had resectable tumor.
Operative mortality and morbidity after tumor resection. Operative death after resection occurred in four of 64 No-Pre patients (6.3%); causes of death were acute pancreatitis, subphrenic abscess and sepsis, mediastinitis and sepsis, and pneumonia. Operative death after resection occurred in two of 19 Pre-Chernorad patients (10.5%, p = NS*); causes of these deaths were cardiac arrest and pneumonia. Significant postresection cardiac, pulmonary, or gastrointestinal complications occurred during hospitalization in 22 of 60 (37%)
*NS = Not significant.
Volume 95
Esophageal carcinoma 4 1 7
Number 3
March 1988
Table II. Pathologic findings and tumor resectability in 136 patients, July 1969 to July 1986 No-Pre group (n = 114) Adenocarcinoma (n = 89)
Tumor stage No residual (stage 0) Stage I Stage II Stage III Stage IV Lymph node status No tumor Resected 'p tp
Pre-Chemorad group (n = 22)
Squamous (n = 25)
Squamous (n = 9)
Adenocarcinoma (n = 13)
No.
%
No.
%
No.
%
No.
%
0 3 4 50 32
0 2 2 15 6
8 8 60 24
3 0 2 4 4
23*
3 5 56 36
56* 11
15 31 31
5 1 0 3 0
22 50
25 56
14
32 56
5 10
38 77
8 9
7
33
89*
root
< 0.01 versus No-Pre group. < 0.05 versus No-Pre group.
No-Pre versus four of 17 (24%) Pre-Chemorad operative survivors (p = NS). Anastomotic leak occurred in five of 60 (15%) No-Pre versus none of 17 PreChemorad patients (p = NS). Anastomotic stricture necessitating dilatation occurred in nine of 60 (15%) No-Pre versus five of 17 (29%) Pre-Chemorad patients (p = NS) with one death in the Pre-Chemorad group that resulted from operative revision of a stricture nine months after colon interposition. Patient survival after tumor resection. Mean followup was 27 months (range 0.4 to 138 months) in No-Pre and 17.6 months (range 1.2 to 51 months) in PreChemorad patients. Seven No-Pre patients (one with squamous cell, six adenocarcinoma) were alive at mean follow-up of 62 months (range 24 to 137 months). Eleven Pre-Chemorad patients (six with squamous cell, five adenocarcinoma) were alive at a mean follow-up of 21 months (range 8 to 51 months). Five of the 11 had had no evidence of disease at resection (three with squamous cell, two adenocarcinoma). A trend toward improved survival after preoperative chemoradiation was observed. Two- and 3-year survival rates were 66% versus 32% and 33% versus 21%, respectively, in Pre-Chemorad and No-Pre patients (Fig. 1). However, this improvement could have been due to chance alone (by Mantel-Cox analysis, p = 0.13; by Cox model analysis adjusting survival curves for age, sex, tumor cell type, and postoperative radiation therapy as covariates, p = 0.15). Tumor status at resection and survival in treatment groups. Life-table analysis of survival in relationship to lymph node status revealed a highly significant correlation in the No-Pre group. The 5-year survival rate was
22% in 29 patients with normal lymph nodes versus 6% in 35 patients with diseased lymph nodes (p < 0.01). Survival in relationship to tumor cell type in the No-Pre group revealed no significant difference between patients with squamous cell carcinoma and those with adenocarcinoma (p = 0.52). In contrast, analysis of survival in relation to lymph node status failed to show a significant correlation in the Pre-Chemorad group. The 2-year survival rate ws 66.6% in six patients with diseased lymph nodes versus 67.3% in 13 patients with normal lymph nodes (p = 0.48). In addition, comparison of Pre-Chemorad patients by means of another measure of tumor status, namely presence or absence of local residual tumor at resection, also revealed no significant correlation, as the 3-year survival rate was 33% in each group (p = 0.58). Analysis of postresection survival in relationship to American Joint Committee staging" was not meaningful, as too few patients had stage I or II disease (Table 11). Recurrence after tumor resection. Tumor recurrence was noted in 50 of 77 (65%) resection survivors. Nine of 60 (15%) No-Pre resection survivors had isolated anastomotic or mediastinal recurrences on the basis of endoscopic, computed tomographic, and autopsy findings. No local recurrences were found in 17 Pre-Chemorad resection survivors for the duration of follow-up, which suggested complete local control of disease. Thirty-four of 60 (57%) No-Pre and seven of 17 (41%) Pre-Chemorad resection survivors had distant tumor recurrence in the liver, lungs, lymphatics, bone, or brain. Life-table analysis of distant recurrence revealed 1- and 2-year rates of 41% versus 33% and 57% versus
The Journal of
4 18
MacFarlane et al.
Thoracic and Cardiovascular Surgery
100 - - No-Pre, N=64 0··········0
80
-
O··J..···J···········JL .. .L•....•..••...•••.
C
U
~
Pre-Chemorad, N=19
9
60
'
35
32
Q.
31
:3
29
0"':
28
40
,2
1
Q················J··············o
p
= 0.13
13 1~---.~_--L.-&..,.11 0
20
14
8 7
o
o
6
12
18
24
30 36 Months
42
48
54
60
Fig. 1. Survival rates after resection of squamous cell carcinoma or adenocarcinoma of esophagus in 83 patients; resection alone (No-Pre) versus preoperative chemoradiation and resection (Pre-Chemorad). Numbers on graph are patients at risk. Tick marks designate patients lost to follow-up (first tick on No-Pre graph) or alive at last follow-up (remaining tick marks).
43%, respectively, in No-Pre and Pre-Chernorad patients (p = 0.48). When distant recurrence rates were adjusted for age, sex, tumor cell type, and postoperative radiation treatment as covariates by the Cox model, no significant difference was obtained (p = 0.22). The number of patients who received preoperative radiation treatment alone (Pre-Rad) was small (13 patients), precluding a meaningful analysis of this group. This group appeared to share characteristics of both No-Pre and Pre-Chemorad patients. As in the Pre-Chemorad group, resectability was high (100%), lymph node status was favorable (54% normal nodes), and local recurrence was rare (one patient, 8%). However, life-table analysis of survival showed no significant improvement compared to survival in the No-Pre group (p = 0.33).
Discussion Patient survival after tumor resection with no preoperative treatment correlated well with the pathologic extent of disease as measured by lymph node status. Five-year survival rates were 22% and 6%, respectively, in patients with normal and diseased lymph nodes, (p < 0.01). This correlation is consistent with the observations of Akiyama and co-workers' who reported 5-year survival figures after resection of 53.8% for patients with normal lymph nodes and 15.3% for those
with diseased lymph nodes. Survival in our surgery alone patients was not influenced by tumor cell type. Lymph node status was the most important pathologic prognostic factor after resection alone regardless of whether the tumor was squamous cell carcinoma or adenocarcinoma. Pathologic extent of disease was improved by preoperative chemoradiation on the basis of two findings. First, significant improvement of lymph node status was shown. Second, total response (ie, no residual disease) occurred in 36% of Pre-Chemorad patients. These improvements were similar for both patients with squamus cell carcinoma and those with adenocarcinoma. A total response rate of 56% for squamous cell carcinoma (Table II) compares favorably with the findings of Leichman and associates," who reported a 33% total response rate, and of Poplin and associates'? (Southwest Oncology Group study), who reported a 24% total response rate at exploration with the same preoperative treatment protocol. Tumor regression induced by adjuvant therapy is the most likely reason for improved resectability in Pre-Chemorad patients, although unidentified selection biases may be contributory. On the basis of the observation that survival is related to pathologic extent of disease and the finding that preoperative chemoradiation results in improved tumor status, one should expect improvement in survival.
Volume 95 Number 3 March 1988
Analysis of survival in No-Pre and Pre-Chemorad groups did, in fact, reveal a trend toward significant survival improvement after preoperative chemoradiation. However, the relationship between local tumor status at resection and survival, as noted in the No-Pre patients, was not maintained in Pre-Chemorad patients. Pre-Chemorad patients with normal lymph nodes or total tumor response had a survival rate similar to that of patients with abnormal nodes or resected residual disease. This similarity was unexpected and is at variance with the Southwest Oncology Group findings in which patients with total response had a 40% 3-year survival rate and patients with resected residual disease had only a 12% 2-year survival rate." A likely explanation for failure to find a relationship between local tumor status and survival after preoperative chemoradiation may lie in the nature of tumor recurrence and our inability to stage esophageal carcinoma accurately. Preoperative chemoradiation clearly resulted in adequate local control of disease, as no local recurrences were noted in Pre-Chemorad patients. However, distant metastatic disease, not local recurrence, was ultimately responsible for the majority of deaths. Moreover, preoperative chemoradiation did not significantly reduce the distant recurrence rate. Nondetection of coexisting systemic disease may result in understaging of disease and, therefore, an overestimation of response to therapy, especially in patients with no residual tumor or normal lymph nodes. The importance of surgical resection has been questioned. First, the previously cited findings by Poplin and associates10 have shown that survival remains poor despite resection of residual disease. Second, surgical resection may not improve the status of patients with no residual disease. Given that these two assumptions are true, one could argue that operative risks are too high. However, the evidence suggests that the first assumption is not valid and the second is irrelevant. Our results after preoperative chemoradiation show that the improvement in survival (although possibly due to chance alone) after resection of residual disease was similar to that after total tumor response. If resection were to be eliminated, two thirds of our patients (those with residual disease) would be deprived of local disease control and potential prolongation of survival. The assumption that surgical resection may not improve the status of patients with no residual disease cannot be refuted by this analysis. However, the assumption itself should not be at issue, but rather the inability to exclude the presence of residual local disease without resection. Campbell and associates" have shown that clinical restaging does not correlate with surgical pathologic
Esophageal carcinoma
4I9
findings. Furthermore, this group reported only a 20% 2-year survival rate after chemoradiation in patients with clinically undetectable and unresected local disease. One third of these patients had local recurrences. These results contrasted with a 50% 2-year survival rate with no local recurrences when resection was done. Preoperative chemoradiation appears to result in significant local disease response and improved resectability for both squamous cell carcinoma and adenocarcinoma. The trend toward significant improvement in survival in this series is encouraging and will become more meaningful with longer follow-up. A prospective, randomized trial would, admittedly, provide a clearer answer to these questions but would be difficult given the compromises inherent in the treatment of patients with esophageal cancer. A larger clinical trial, subjected to multivariate analysis, could also provide more certainty to the conclusions. The application of multimodality therapy to adenocarcinoma is relatively recent. Our observation of similar responses by both cell types is supported by the recent, early observations of Wolfe and colleagues" and warrants the inclusion of patients with adenocarcinoma in further studies of multimodality therapy. Surgical resection of esophageal carcinoma should remain an integral part of curative therapy. It is an important aspect of local disease control, and early analysis suggests a trend toward improved survival after resection of residual disease. The solution to the continuing problem of distant recurrence requires the introduction of more effective systemic therapy. The work of Gloria Bailey, PhD, of the Virginia Mason Clinical Research Support Services for help with statistical analysis and of Bette Glass in the preparation of this manuscript is appreciated. 1.
2. 3. 4. 5.
6.
REFERENCES Akiyama H, Tsurumaru M, Kawamura T, Ono y. Principles of surgical treatment for carcinoma of the esophagus. Ann Surg 1981;194:438-46. Dark JF, Mousalli H, Vaughan R. Surgical treatment of carcinoma of the esophagus. Thorax 1981 ;36:891-5. Ellis FH, Gibb SP. Esophagogastrectomy for carcinoma: current hospital mortality and morbidity rates. Ann Surg 1979;190:699-705. Galanduik S, Herman RE, Gassman 11, Cosgrove OM. Cancer of the esophagus: the Cleveland Clinicexperience. Ann Surg 1986;203:101-8. Fein R, Kelson DP, Geller N, Bains M, McCormack P, Brennan MF. Adenocarcinoma of the esophagus and gastroesophageal junction: prognostic factors and results of therapy. Cancer 1985;56:2512-8. Gatzinsky P, Berglin E, Dernevik L, Larsson I, William-
The Journal of
420
7.
8. 9.
10.
11.
12.
MacFarlane et al.
Olsson G. Resectional operations and long-term results in carcinoma of the esophagus. J THORAC CARDIOVASC SURG 1985;89:71-6. Michaelson RA, Magill GB, Quan SHQ, et al. Preoperative chemotherapy and radiation therapy in the management of anal epidermoid carcinoma. Cancer 1983;51: 390-5. Beahrs 0, Myers MH, eds. Manual for staging of cancer. Philadelphia: JB Lippincott Co, 1983. Leichman L, Steiger Z, Seydel HG, Vaitkevicius VK. 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 1984; II :178-85. Poplin E, Leichman L, Seydel HG, Steiger L, Fleming C. SWOG 8073: combined therapy for squamous cell carcinoma of the esophagus (SCCE). Proc Am Soc Clin Oncol 1986;5:80. Campbell WR, Taylor SA, Pierce GE, Hermreck AS, Thomas JH. Therapeutic alternatives in patients with esophageal cancer. Am J Surg 1985;150:665-8. Wolfe WG, Burton GV, Seigler HF, Crocker IR, Vaughn AL. Early results with combined modality therapy for carcinoma of the esophagus. Ann Surg 1987;205:56371.
Discussion Dr. Robert L. Mitchell (Mountain View, Calif).The authors have thoroughly analyzed their data on a combined modality treatment of cancer of the esophagus in 22 patients and compared this with resection alone in 114 patients treated in a previous period. The purpose of the paper is to answer whether combined treatment affects tumor status at exploration, resectability, disease recurrence, and patient survival. Combined treatment unequivocally altered tumor status as, at operation, 36% of the patients had no residual tumor. Resectability appears to be improved, from 56% to 77% for adenocarcinoma and 100% for squamous cell carcinoma. However, the criteria for resectability are not defined, and this is a retrospective study over 17 years. Other recent series, including my own, have reported a high resectability rate without adjuvant therapy. In the combined modality group, local recurrence was zero, but I think it premature to attribute this solely to chemoradiation preoperatively. Survival rates were not statistically altered by combined therapy. However, the authors point out that survival trends favor a combined modality treatment. The obvious danger in this comparative study of two treatment modality groups is drawing major conclusions when the study is not prospective or randomized and contains only 22 patients in the chemoradition group. We have, however, learned from this study that combined chemoradiation affects tumor status at operation and appears to decrease local recurrence. It is interesting that other combined treatment protocols for lung and rectum cancers have also shown a reduction in local recurrence, but survival rates have essentially been unchanged. The question arises whether operation is necessary after chemoradiation, as you have already addressed: Are those
Thoracic and Cardiovascular Surgery
patients who are tumor-free after chemoradiation not improved by operation, and those who have residual tumor doomed anyway? I think for the present, until more data are available, I would concur with the authors that operation is essential to stage the disease, to reduce local recurrence and, we hope, to improve survival rates in the group that has residual tumor. I compliment the authors for studying new modalities of therapy for a disease of high morbidity and mortality with and without treatment. Plaudits are in order for achieving a low morbidity and mortality in your series. Indeed there were no anastomotic leaks in the chemoradiation group. Although not precisely stated in the paper, I surmise that combined treatment is well tolerated by the patients. Possibly the patients were in a more optimum state for resection after this particular chemoradiation protocol. As you know, some previous adjuvant therapy studies have shown that the treatment was worse than the disease. Wolfe and his co-workers at Duke University, as you pointed out, reported in the May issue of the Annals of Surgery on 63 patients treated with combined therapy. All patients showed, in their analysis, a marked decrease in tumor mass and improved swallowing, and became anabolic on oral nutrition. These preliminary studies are encouraging in a disease with a dismal outlook, but must be supported with controlled prospective studies. I have four questions for the authors. First, is the choice of treatment regimen (that is, drugs and radiation) based on scientific data or empiricism? Dr. MacFarlane. It is empirically based on scientific data. Basically, the choice derives from studies that have looked at cloacogenic carcinoma and have shown effectiveness in terms of reduction of tumor mass. At the onset, these studies can really focus only on initial, short-term results to evaluate tumor response. In terms of survival, the long-term data remain to be seen, but for cloacogenic tumors, it appears that there is an improvement in survival, and consequently, this treatment was applied to tumors of the esophagus. Ours is one of the first centers to apply this protocol to adenocarcinoma with the addition of mitomycin C, although the Duke University study now also shows that on early follow-up, adenocarcinoma and squamous carcinoma appear to respond similarly. That is one aspect of this study that is rather exciting. I think we ought to include patients with gastroesophageal junction tumors in further studies of multimodality treatment protocols. Dr Mitchell. Did chemoradiation improve nutrition, as Wolfe and his co-workers have reported? Dr. MacFarlane. The Duke University experience described by Wolfe and associates was surprising, because they were very optimistic about the status of their patients after chemoradiation, although they did not quantify this in any way. They said that after patients received chemotherapy and radiation, they were in better condition for operation than those who did not have preoperative treatment. That is not my empirical impression. I think it does take its toll. We have evaluated white count and that does, in fact, go down, and sometimes requires precautionary hospitalization. The protocol at Virginia Mason is to admit patients for each cycle of chemotherapy. The Duke protocol was quite a bit different from ours, because they used cisplatin and VP-16 rather than cisplatin and 5-ftuorouracil and didn't start radiation until after two cycles of chemotherapy, while we initiated radiation at the beginning. Dr. Mitchell. Dr. Daniel, of Charlottesville, Virginia,
Volume 95 Number 3
March 1988
reported a 38% incidence of nondilatable stricture after chemoradiation before operation. You reported a 29% incidenceof stricture in the combined group after operation. Did yousee a problem with significant stricture with the chemoradiation group before operation? Dr. MacFarlane. Our stricture rate was similar in both groups. It was a little bit higher in the chemoradiation group, to be sure, and whether this relates to the type of operation performed or not is not clear. One was a cervical anastomotic stricture outside the radiation field after total replacement withcolon interposition, so that it did not seem to relate to the preoperative treatment. I do not think we have sufficient numbers to really evaluate cause and effect. Our impression was, first of all, that the operative morbidity was similar in the twogroups, and that anastomotic problems, specifically leaks, did not occur in the chemoradiation group. Benign strictures after combined treatment, aside from one case, were easily dilatedand frequency was similar to that after operation alone. Stricture after chemoradiation, but before operation, was not a problem. Dr. Mitchell. My last question is, is there a place for chemoradiationin the patient whose cancer is nonoperable and who has almost total obstruction of the esophagus? Dr. MacFarlane. Wolfe and co-workers address that question also, and say, in fact, that they get excellent palliative results. Although it was not emphasized in this discussion, we believe that this treatment protocol is very effective as a palliative measure, irrespective of the survival results. In fact, when you can eradicate the local tumor burden and prevent recurrence, these patients, even though they go on to have distant metastases, do not have the debilitating obstructive symptoms. The advent of laser therapy has also provided us with another useful tool for palliation along with preoperative adjuvant treatment. In other words, if the esophagus is almost obstructed, it is very appropriate to use laser excavation to reestablish the lumen and then to place the patient on this combined modality protocol. We have done this once or twice with good results. Dr. James B. D. Mark (Stanford. Calif). We have been using a preoperative combined chemoradiation protocol as well, but just for squamous carcinoma. I was interested to hear about your nice results with adenocarcinoma. My question is this: You have 22 patients on whom you operated after chemoradiation. If your experience is anything like ours, more patientsthan that started a chemoradiation protocol, but never got to operation, usually because of extensive disease or debilitation. Can you give us an idea how many patients began treatment but were never operated on? Dr. MacFarlane. Many patients are initially treated by the oncologists before, and often without, surgical consultation. One of our oncologists has provided this data for adenocarcinomas only. Approximately 15% to 20% do not go on to operation, according to his figures, but some patients were put on the protocol after it had been determined on the basis of initial clinical work-up that the lesions were not resectable. It is hard to get a good grip on what the actual denominator is and who would have been a true surgical candidate to begin with. Dr. Mark. Ideally, it would be a good idea to get in at the beginning with all of these patients, so that the surgeon is on the original staging and decision making. What I think is happening in a number of diseases these days with aggressive pulmonologists, oncologists, and others is that we never see
Esophageal carcinoma
421
some patients with cancer of one kind or another whom we could help, because decisions are made for them by nonsurgeons. Dr. MacFarlane. I agree, and that was one of the reasons for presenting this paper. Dr. Murray I. Sheldon (Concord. Calif). I appreciated that the initial impetus for your study was that your oncologists believed that with the good results of chemoradiation this perhaps would be the only treatment necessary and they wished to drop the surgical arm of the treatment protocol. We have a similar problem at our hospital. Your study evaluated combined chemoradiation plus operation and compared it to operation alone. Do you have any data that I can take back to my hospital that suggest that the combination of chemoradiation and operation is, in fact, beneficial in short- and long-term survival compared with your chemoradiation group alone? Dr. MacFarlane. No, I really do not. The problems are that, first, we do not have the ability to accurately determine the stage of the disease when the patient comes for treatment and, second, as in any nonsurgical arm, we do not have an ability to quantitatively stage the disease after treatment. As you all know, many of these patients have endoscopic and computed tomographic studies and the disease appears to be gone. Then at resection of the specimen a microscopic residual is found. Until we can develop some accurate clinical means to analyze the tumor response, I do not think there is any basis for comparison between surgical and nonsurgical treatment other than just survival. I do not have any good data on survival after nonsurgical chemoradiation treatment. I hope that our data, at the very least, based on local tumor response, suggestion of improved survival rates, and maintenance of acceptable operative mortality and morbidity rates, has shown that there is no basis for the removal of surgical resection as an integral part of treatment. Dr. Sheldon. Are your oncologists publishing the data of chemoradiation alone in any way compared with yours? Dr. MacFarlane. No, only a small number of patients who received chemoradiation did not go to operation, and their survival rate was not very good compared with that of the group having surgical resection. The reason they did not go to operation, however, was that they had some untoward event. Dr. Alden H. Harken (Denver. Colo.). Dr. MacFarlane's sophisticated surgical group is asking a question to which we in the audience, I think, are all sympathetic. It is always easy to poke holes in nonprospective, nonrandomized, multi-institutional trials. Obviously, one of the problems with this study is that it is retrospective and the group sizes differ greatly: The discrepancy between 22 in one group and 114 in the other is bothersome. However, it seems to me that Dr. MacFarlane has raised our awareness of a significant clinical issue: Is there a position for surgery in the therapy of carcinoma of the esophagus? Although our oncologist friends may be building a case that there is not, it seems to me that we must address some of the advantages of surgical therapy, and I believe there are advantages. I would ask whether you have elevated some of them. You persuasively make the case that there is a surgical benefit. In this group of patients we see in the good group a 22% 5-year survival rate and in the bad group a 26% 5-year survival rate, but what happens to the quality of life during that period of time? Is there any way of addressing or assessing the issue of palliation by itself? My bias is that the patients can swallow if they are operated on, and they cannot if not operated on. Did you assess that, and are there any other
The Journal of
422 MacFarlane et ai.
measurable, objective parameters such as anabolism, immunologic response, or ability to swallow that we can look at? Dr. MacFarlane. In this study we compared two surgical groups, so palliation rates are similar in the two groups. The gold standard for palliation in this study is in the 114 patients having operation only. In those patients, we achieved approximately a 92% overall palliation rate, including those who required some repeat dilation. If those were not included, the rate was about 85% palliation. Using that as the gold standard, we have shown that palliation rates were simialr after preoperative treatment and resection. The only other objective information obtained, but not quantified, was my impression in talking to these patients that it took longer to gain weight after the multimodality therapy than it did after operation alone: approximately 2 months compared with 4 weeks after operation alone. We are paying a small price there, but once the patients begin to gain weight, they do well. If they begin to lose weight again, that is a real marker for a recurrence.
Thoracic and Cardiovascular Surgery
Dr. Ivan A. May (Oakland, Calif). Dr. Harken's statement that if you operate on these patients they can swallow and if you do not they cannot swallow is true in my experience as well. Dr. Walter B. Cannon (Palo Alto, Calif). Have you. in the review of all these patients, seen a difference in response in a cervical esophageal cancer versus a middle or lower-third esophageal cancer? Dr. MacFarlane. Only one or two multi modality patients had upper-third cancers. Many of the patients who have upper third cancers we do not evaluate because of our separte approach (medical versus surgical). I agreed with the previous comments of Dr. Mark that there needs to be a consensus on how we screen these patients. I think the surgeon should be included early on, and he often is not. That is one reason we have not been able to analyze your question, Dr. Cannon, because we have seen too few patients with proximal-third tumors.
Bound volumes available to subscribers Bound volumes of THE JOURNAL OF THORACIC AND CARDIOVASCULAR SURGERY are available to subscribers (only) for the 1988 issues from the Publisher, at a cost of $51.00 ($68.00 international) for Vol. 95 (January-June) and Vol. 96 (July-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed. Copies are shipped within 60 days after publication of the last issue of the volume. The binding is durable buckram with the JOURNAL name. volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact The C. V. MosbyCompany, Circulation Department, 11830 Westline Industrial Drive, St. Louis, Missouri 63146, USA; phone (800) 325-4177, ext. 351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular JOURNAL subscription.