Selective therapeutic approach to cancer of the lower esophagus and cardia

J

THoRAc CARDIOVASC SURG

1988;95:42-54

Selective therapeutic approach to cancer of the lower esophagus and cardia The role of curative en bloc resection for carcinoma of tbe lower esophagus and cardia is still controversial. The experience with a selective approacb in 52 patients witb cancer in this location is reviewed. Thirty-two of the cancers were squamous cell, 13 adenocarcinoma, and seven adenocarcinoma associated witb Barrett's esophagus. In 19, tbe tumor was not resectable and all of tbese patients died within a year. In 19 patients, a palliative resection could be done. Actuarial survival was 31 % at 1 year. Only one patient was alive after 5 years. InitiaDy, 16 patients witb noncircumferential lesions on endoscopy and/or no evidence of spread to mediastinallympb nodes on computed tomographic scan were considered to have potentially curable lesions. AU were less than 75 years old and had a forced expiratory volume in 1 second greater than 1.5 L and a resting ejection fraction greater than 40 %. A curative resection consisting of an en bloc tboracic esophagectomy, mediastinal lymphadenectomy, and an 80% gastrectomy with abdominal lymphadenectomy was performed in 14. The left colon was used to reestablisb tbe gastrointestinal continuity. Two patients had more extensive disease discovered at operation, and tbe curative en bloc resection was abandoned. Absence of fuU waD penetration or involvement of four or fewer regional nodes, or both, was correctly predicted by preoperative and intraoperative staging in 86 % of tbe patients. Operative mortality of a curative en bloc resection was 7 % (1/14), and tbe actuarial survival rates were 76%, 66%, and 53% at 1, 2, and 5 years. Inferences are made from these results on tumor characteristics associated with survival, the extent of resection necessary for cure, the difficulty of accomplishing a curative en bloc resection by the transhiatal approach, the contraindication to curative en bloc resection, and the need for a surveillance program for patients with Barrett's esophagus.

Tom R. DeMeester, MD, Giovanni Zaninotto, MD (by invitation), and Karl-Erik Johansson, MD (by invitation), Omaha, Neb.

Current therapy for carcinoma of the esophagus has an aura of pessimism, resulting in an attitude among physicians that cure is impossible. This attitude has given rise to two current treatment philosophies. The first is that palliation of dysphagia is the only reasonable therapeutic goal. Consequently, surgical procedures of touted low operative mortality have been advocated as a means of providing palliation by removal of the primary tumor without regard for the principles of cancer From the Department of Surgery, School of Medicine, Creighton University, Omaha, Neb. Read at the Sixth-seventh Annual Meeting of The American Association for Thoracic Surgery, Chicago, Ill., April 6-8, 1987. Address for reprints: Tom R. DeMeester, MD, Creighton University, School of Medicine, Department of Surgery, 601 North 30th St., Omaha, NE 68131.

42

surgery.' Proponents of this philosophy cite the high mortality of 29% for transthoracic esophageal resection and the miserable and inhumane death from esophageal obstruction as reasons for the approach.l' Cure of the disease with this approach is relegated to a chance phenomenon. The second philosophy is that operation alone is not sufficient therapy and that the addition of adjuvant radiation or chemotherapy is required to achieve cure. This approach predestines patients with limited disease to a delay in resection and to the morbidity and mortality of adjuvant therapy without proven benefit.' This philosophy exists though postoperative staging has shown that in patients with limited disease the results of surgical resection are better than those achieved by any other single or combined therapy.v' The question addressed by this study is this: Can a therapeutic

Volume 95 Number 1

January

Cancer of lower esophagus and cardia

1988

Table I. Classification of esophageal carcinoma Wall penetration WO = Tumor limited to mucosa WI = Tumor penetration through submucosa and into but not through muscle layers W2 = Tumor penetration through muscle layers Lymph node involvement NO = No lymph node involvement Nl = 1 to 4 lymph nodes involved N2 = 5 or more lymph nodes involved Systemic metastasis MO = No metastasis MI = Distant metastasis

approach be developed that identifies patients with early disease in whoman operation can be performed according totheprinciples of a curative en blocresection with a low mortality and increased probability of survival? Patients and methods Patient evaluation. Fifty-two patients with cancer of the lower esophagus or cardia were referred to the senior author (TKO.) from 1980 to 1986. Thirty-two of the cancers were squamous cell carcinoma of the lower esophagus, 13 were adenocarcinoma of the cardia, and seven were adenocarcinoma that developed in a Barrett's esophagus. All patients were assessed as to the extent of the disease by physical exam, esophagogastroduodenoscopic exam, barium roentgenogram of the upper gastrointestinal tracts, computed tomographic scan from thoracic inlet to umbilicus at cuts of I to 2 em, and liver chemistries. Patients were identified preoperatively as to potential curability according to the staging system proposed by Skinner and associates" (Table I). In practice, this meant excluding those patients who had clear evidence of tumor extending through the esophageal wall and lymph node involvement, or the presence of systemic metastasis. Preoperative physiologic assessment. Before therapy was begun, the nutritional status and physiologic function of the patient was assessed. The nutritional assessment was done by obtaining a serum albumin level before any hydration. If the serum albumin level was below 3.4 gm, aggressive hyperalimentation was initiated with a jejunostomy tube inserted operatively by an intramural technique.' Pulmonary reserve was assessed by measuring the forced expiratory volume in I second (FEV 1) . Cardiac reserve was assessed with a gated pool radionuclide study to evaluate wall motion and ejection fraction. Selection of therapy. Surgical therapy was based on the patient's age and physiologic status and the preoperative and intraoperative assessment of the disease. Those with unresectable tumors, systemic metastasis, or poor physiologic status received only palliative therapy. Patients with adequate physiologic status and a resectable tumor but unfavorable pathologicfindings (W2 N2), or who had favorable pathologic findings (no evidence of W2 N2 disease) but were more than 75 years ofage, had a palliative esophagogastrectomy with gastrointestinal continuity reestablished by an esophagogastrostomy. Patients with recurrent local disease after a previous resection

43

Table II. Selected surgical therapy No. of patients

Therapy

Palliation only No operation Exploration Bypass Palliative resection Standard esophagectomy* Transhiatal esophagectomy Repeat esophagogastrectomy Curative resection by radical esophagogastrectomy Total

19 9 4 6 19

10 5 4 14

52

"Two patients switched from a curative resection because of operative findings.

m.

Table Therapy groups versus duration of dysphagia Therapy

Palliation only Palliative resection Curative resection

No. of patients

Duration of symptoms (mo)

19

4.5 ± 3.7 5.2 ± 3.5+ 3.2 ± 2.4+

15*

9t

"Four patients with recurrence after previous operation were excluded. tFive patients without symptoms were excluded.

+P < 0.05. had a palliative resection with gastrointestinal continuity reestablished by a colon interposition. Patients with resectable tumors, favorable pathologic findings (no evidence of W2 N2 disease), adequate physiologic status, and age less than 75 years had a curative en bloc resection of the esophagus and stomach with gastrointestinal continuity reestablished by a colon interposition. If operative findings indicated gross W2 N2 disease in a patient previously thought to have favorable pathologic conditions, the curative resection was abandoned and a palliative resection done. The physiologic criteria for a curative resection were (I) less than 75 years of age, (2) FEV 1 greater than 1.5 L, (3) ejection fraction greater than 40%, (4) normal findings at electrocardiogram, and (5) no cardiac symptoms. Population studied. Table II shows the surgical therapy received by the 52 patients studied. Nineteen patients had advanced or unresectable disease and received only palliative treatment. Six of these 19 had dysphagia relieved with a bypass operation, one had an esophageal stent, and the other 12 received radiation or chemotherapy. Another 19 patients had a palliative resection: four because of recurrent tumor at the anastomosis after a previous esophagogastrectomy for adenocarcinoma ofthe cardia; eight who were treated with preoperative chemotherapy because of clear evidence of W2 N2 disease; three because of marginal physiologic reserve; two because of age more than 75; and two, who had been initially candidates for a curative resection, because more extensive disease was discovered during

44 DeMeester, Zaninotto, Johansson

The Journal of Thoracic and Cardiovascular Surgery

Fig. 1. A, Outline of boundaries for thoracic portion of en bloc resection. 8, Outline of boundaries for abdominal portion of en bloc resection. the curative dissection. Fourteen or 27% (14/52) of the patients had a curative en bloc resection. Table III shows the duration of dysphagia for each of the therapy groups shown in Table II. Five patients who had a curative operation did not have dysphagia at the time the disease was discovered. Four had Barrett's esophagus and the diagnosis of adenocarcinoma was made during their initial endoscopic evaluation for reflux symptoms. The fifth had a squamous cell carcinoma diagnosed during an endoscopic exam done for dyspepsia. The remaining nine patients who had a curative en bloc resection were symptomatic and the diagnosis was made sooner after the onset of dysphagia than in the 15 who had a primary palliative resection. Description of the en bloc resection of the esophagus and stomach. Before this study the technique of en bloc resection of the lower esophagus and cardia was based on the published reports of Logan" and Skinner." After the senior author (T.R.D.) gained experience with their techniques, modifications were made that were thought to comply more faithfully with the principles of an en bloc resection of the lower esophagus and cardia while providing greater safety for the patient. The curative operation done in this study incorporated these modifications. The operation is done through three incisions and in the following order: right posterolateral thoracotomy, en bloc dissection of the esophagus and mediastinum, closure of the thoracotomy, repositioning of the patient in the recumbent position, upper midline abdominal incision, en bloc dissection of the stomach and associated lymph nodes, left neck incision,

division of the esophagus, transhiatal removal of the previously en bloc dissected esophagus and stomach, and reestablishment of gastrointestinal continuity with a left colon interposition. The thoracic portion of the procedure entails an en bloc removal of the thoracic esophagus with its surrounding areolar tissue containing the paratracheal, subcarinal, paraesophageal, and parahiatal lymph nodes; the thoracic duct; the azygos vein down to where it passes into the abdomen on the lateral surface of the vertebra; and a collar of diaphragmatic muscle around the esophageal hiatus (Fig. I, A). The block of tissue is limited anteriorly by the membranous trachea and pericardium; laterally by the right and left mediastinal pleura; and posteriorly by the intercostal arteries, aorta, and anterior vertebral ligaments. The abdominal portion of the procedure entails an en bloc removal of all the posterior peritoneal periaortic areolar tissue down to the celiac axis and the superior border of the common hepatic artery, the splenic artery and spleen, the greater omentum, and the proximal three fourths of the stomach (Fig. I, B). The tissue block includes lymph nodes around the left gastric artery and celiac axis superior to and underneath the common hepatic artery, medial to the portal triad, in the greater omentum, and around the splenic artery down to the celiac axis. This extensive resection is done to incorporate in the surgical specimen all potentially involved regional lymph nodes and submucosal lymphatics of the stomach and esophagus. The pancreas is not removed, but the splenic artery along with its associated nodes is dissected off the superior border of the gland. The procedure is started in the right side of the chest with

Volume 95 Number 1 January 1988

division of the intercostal branches of the azygos vein from its arch down to where it passes into the abdomen on the lateral surface of the vertebra. The segments of posterior pleura between the ligated intercostal veins are divided with an incision parallel to the spine. The posterior dissection is extended across to the left mediastinal pleura along the intercostal arteries to the aorta and over the anterior surface of theaorta into the left side of the chest. The intact azygos vein and its surrounding tissue are thereby elevated anteriorly, which allows the hemiazygos vein or veins to be seen as they cross over the spine underneath the aorta to join the azygos vein. These veins must be identified, ligated, and divided. The aorta becomes visible when the hemiazygos vein is divided and serves as a guide for the extention of the posterior incision into theleftside of the chest. Early division of the azygos vein at its entry into the superior vena cava is avoided because it contributes to excessive bleeding during the posterior mediastinal dissection. At the caudal end of the posterior pleural incision, where the azygos vein was ligated distally, the thoracic duet is identified, divided, and ligated. Anteriorly, the mediastinum is entered through a longitudinal pleural incision made parallel to the posterior margin of the trachea, hilum, and pericardium down to the diaphragm. The anterior dissection is extended across to the left mediastinal pleura along the posterior surface of the trachea, left mainstem bronchus, and pericardium anterior to the subcarinal nodes. To do so requires division of the azygos vein at its entry into the superior vena cava. When both the anterior and posterior dissections of the posterior mediastinum are complete, the esophagus, along with its periareolar tissue containing the paratracheal, subcarinal, paraesophageal, and parahiatal nodes, is pulled into the right thorax and freed by sharp division of a strip of left mediastinal pleura. Care must be taken to avoid damaging the left recurrent laryngeal nerve near the aortic arch. Its common course is to pass directly to the trachea without redundancy and to lie on the left lateral inferior cartilaginous wall as it passes up into the neck. Occasionally it follows the left lateral wall of the esophagus up into the neck. Inferiorly, a collar of diaphragmatic muscle is excised around the esophageal hiatus. Superiorly, the esophagus isbluntlydissected into the neck but not divided. When the dissection is completed, the thoracotomy incision is closed with the specimen remaining in the chest wrapped in a surgical sponge. Next, the patient is moved to the recumbent position. The previously inserted double-lumen endotracheal tube, used for selected deflation of the right lung, is removed and a singlelumen tube is inserted. The anterior neck, chest, and abdomen are prepared and draped and an upper midline abdominal incision made. Exposure for the abdominal dissection is facilitated by a Weinberg retractor that has been welded to a Balfour handle and attached to an over-arm bar. The abdominal dissection is begun with the removal of the greater omentum from the transverse mesocolon and mobilization of the spleen to the midline. This allows access to the left retroperitoneal periaortic areolar tissue that lies just anterior to the aorta. The abdominal dissection is started along the left crus where the thoracic dissection was discontinued. The excision of a collar of the diaphragmatic muscle is continued down the margin of the left crus to the celiac axis. The pancreas is identified, and the splenic artery along with its vein and associated lymph nodes are dissected off the superior

Cancer of lower esophagus and cardia

45

Fig. 2. Reconstruction of gastrointestinal tract with left colon interposition after en bloc resection. Distal anastomosis is made to antral portion of stomach and proximal anastomosis to esophagus in neck. border of the tail of the gland. The splenic vein is ligated approximately midway down the pancreas where it turns inferiorly and leaves the artery. Removal of the splenic artery along with its associated lymph nodes is continued down to the celiac axis, using the palpation of the pulse as a guide to the dissection. At its origin from the celiac axis, the splenic artery is divided, which allows the left gastric artery to be identified. The gastrohepatic ligament is divided along the liver margin up to the esophageal hiatus. The posterior peritoneal tissue underneath and along the superior border of the common hepatic artery and the areolar tissue containing lymph nodes along the medial border of the portal triad are swept toward the celiac axis. Dissection of a collar of diaphragmatic muscle around the esophageal hiatus, which was begun during the thoracic portion of the operation, is continued down the right crus. The dissection of the hiatus is completed with the division of the left gastric artery at its origin from the celiac artery. The esophagus is exposed and divided in the neck through an incision made along the anterior border of the left sternocleidomastoid muscle. Care is taken to identify and protect the left recurrent laryngeal nerve. The surgical specimen is removed from the posterior mediastinum transabdominally trailed by a long suture tied to the esophageal end of the specimen. Gastrointestinal continuity is reestablished between the proximal esophagus in the neck and the remaining gastric

The Journal of

46

Thoracic and Cardiovascular Surgery

Debdeester, Zaninotto, Johansson

Right posterolateral thoracotomy

...

Inspection of tumor

~

~

No evidence of both esophageal wall penetration and lymph node involvement

Gross evidence of esophageal wall penetration and lymph node involvement

'"

T

Curative en bloc diSsection of esophagus

Palliative dissection of esophagus

++ • • • •

I

I

Closure of thoracotomy Reposition Midline laparotomy

______---JII'-

+

Mobilize stomach

~

1

Gross evidence of wall penetration and lymph node involvement

_.

'"

. - - - - - Inspection

'"

No gross evidence of both wall penetration and lymph node involvement

'"

Curative en bloc dissection of stomach

Pullup for neck anastomosis

'"

I

Patient's condition unsatisfactory for colon graft ~ Closure of stomach, cervical esophagotomy, feeding jejunostomy lube. Delay reconstruction.

'"

Patient stable and good arterial supply and venous drainage to colon graft

'"

Colon Interposition

Fig. 3. Algorithm of intraoperative decision-making based on extent of disease encountered.

Table IV. Performance status Karnofsky score Performance

Normal activity Symptomatic but ambulatory; cares for self Ambulatory more than 50% of time; occasionally needs assistance Ambulatory 50% or lessof time; nursing care needed Bedridden; may need hospitalization

AJCCS scale

ECOG scale

HO

o

HI H2 H3

H4

! 2 3 4

(%)

90-100 70-80 50-60 30-40

10-20

Legend: AJCCS, American Joint Committee for Cancer Staging. ECOG, Eastern Cooperative Oncology Group.

antrum in the alxlomen with an isoperistaltic left colon transplant based on the left colic artery and inferior mesenteric vein (Fig. 2). We believe that retention of the antrum and pyloric valve improves postoperative function sufficiently to merit salvage. The colon transplant is pulled up through the posterior mediastinum with the aid of an upside-down Mousseau-Barbin tube using the above-mentioned suture as a guide. Of importance is that if the operation has lasted too long, if the patient is unstable, or if the vascular supply or venous drainage of the colon is not satisfactory, then the colon is not divided, the antral opening of the stomach is closed, an intramural jejunostomy tube inserted, and a cervical esophagostomy constructed. The proximal marginal and midcolic

artery and vein are divided at this time to increase the vascularity of the subsequent colon transplant and to avoid the necessity of dissecting the base of the midcolic artery and vein at a later date. The transverse mesocolon should be opened wide and the whole of the small bowel brought through the opening so that the transverse colon lies under the small bowel low in the alxlomen. This is to prevent its migration into the left upper quadrant and adherence to the denuded posterior peritoneal surface. The patient is discharged on jejunostomy tube feedings to return in 60 to 90 days for reestablishment of gastrointestinal continuity with the previously delayed isoperistaltic left colon graft through a substernal route. The one-stage operation is preferred because postoperative adhe-

Volume 95 Number 1

Cancer of lower esophagus and cardia

January 1988

100

CO > .;; .....

•

80

.\

::l

(J)

60

Q)

40

~

Q)

o,

20

•

Palliative Hesecnon Palliation Only

~

\ \

~ N"19

--.-

-

-.

-

-

--- --N~19

> .;;

80

::::l (J)

60

C

40

-

Q)

o ..... Q) c,

5

3

4

Actuarial

0

5

sions and scarring of the transverse mesocolon can limit the length of the colon transplant, but the reconstruction phase should not be done unless everything is satisfactory. During the thoracic and abdominal dissection. intraoperative staging is done according to the algorithm in Fig. 3. If distant organ metastases are discovered, or obvious tumor extension through the esophageal wall and lymph node involvement is present, the radical dissection is abandoned and a palliative resection performed with the stomach used to reestablish gastrointestinal continuity. Patient follow-up. Patients treated by palliation only or a palliative resection were followed up expectantly. Patients treated with a curative en bloc resection were evaluated every 3 months for the first year and every 4 months thereafter. A barium swallow and CT scans were done every 6 months. The Karnofsky score, as defined in Table IV, was used to classify patient performance status. 10 With one exception, all surviving patients who had a curative en bloc resection were seen in person I month before submission of this manuscript. Statistical methods. Comparison of data was done by the Student's t test. Actuarial and adjusted survival rates were calculated with a previously described formula." The log-rank method was used to compare survival curves for the various therapeutic groups." Results

Fig. 4 shows the overall survival curve for the various therapeutic groups. All 19 patients who received only palliative therapy died within I year. One patient who had a palliative resection survived for 5 years. Fourteen patients underwent a curative en bloc resection. Twelve had a single-stage and two a double-stage procedure. The 5-year actuarial survival rate was 53% and, as suspected, was significantly better than for the palliative group (p < 0.02). The adjusted survival rate is a more accurate assessment of the results of a curative en bloc resection because patients in this age group are susceptible to death from causes other than recurrent tumor. Of the 14 patients who had a curative en bloc resection, two died of causes unrelated to the malignant tumor. This resulted in an adjusted 5-year survival rate of 63% (Fig. 5).

3

2

Years Fig. 4. Actuarial survival of treatment groups.

4

4

20

oL---f}------l------l------'------' 2

Adjusted

.....

.. \

\

C

100

Curative Resection

<>

47

4

5

Years Fig. 5. Actuarial and adjusted survival rates for patients having curative en bloc resection. Numbers between lines show number of patients at risk for each interval of followup.

% Favorable

66

33

20

% Less Favorable

33

40

% Unfavorable

0

66 0

7

Number of

Patients

\

5 3

1.5

40

I

I

:\. 4.5

3

6

7.5

em Fig. 6. Relationship of tumor size to esophageal wall penetration and presence of lymph node metastasis.

Thirty-day hospital mortality for the group undergoing curative en bloc resection was 7% 0/14). The one death occurred in a patient whose vascular supply to the colon was inadequate for interposition; gastrointestinal continuity was reestablished with a gastric tube. On postoperative day 19, he had a massive hemorrhage from a midgastric ulcer that penetrated into the aorta. This was the only patient who had a curative en bloc resection in whom a colon interposition could not be performed. One of five patients who had a blunt esophagectomy and three of the 14 who had a standard esophageal resection for palliation died within 30 days of the procedure. Table V shows the postoperative staging classification of the 14 patients undergoing a curative en bloc resection. Only 14% (2/14) who were considered preoperatively and intraoperatively to have favorable esophageal disease were judged unfavorable postoperatively that is, the tumor had penetrated through the wall of the

48

The Journal of Thoracic and Cardiovascular Surgery

Detdeester. Zaninotto, Johansson

Table V. Postoperative classification after curative resection Favorable

No. of patients

Less favorable

No. of patients

Unfavorable

No. of patients

WONO WI NO

2

WONI WON2 WI NI WI N2 W2 NO W2NI

3 I I

W2 N2

2

I

9

3

Total

I

1

2

Legend: Intraoperative staging 12/14 correct (86%).

Table VI. Lymph node involvement Prevalence Location Para tracheal Subcarinal Paraesophageal Parahiatal Left gastric artery Splenic artery Right gastroepiploic artery

(%)

o 7.1 21.4 35.7 35.7 14.3

o

esophagus and five lymph nodes or more were involved. As mentioned in the Methods section, two other patients were discovered to have advanced disease during the operation, and the resection for cure was abandoned. Thus four of 16 patients thought initially to be candidates for curative en bloc resection had unfavorable pathologic conditions; two discovered at operation and two discovered after histologic examination of the surgical specimen. Overall, preoperative staging had an accuracy of 75% and the combination of preoperative and intraoperative staging had an accuracy of 86%. Fig. 6 shows the relationship of tumor size to the presence of favorable, less favorable, or unfavorable pathologic conditions in the 14 patients having resection for cure. All tumors less than 3 em in length were classified as favorable. The tumor was 4 em or less in 93% of the patients undergoing resection for cure. A total of 599 lymph nodes or 42.7 ± 14.2 nodes per patient were removed by the curative resection. Table VI shows the prevalence of metastasis to the different lymph node groups. Four of the 14 patients who had a curative resection did not have lymph node metastasis. Three are alive and one died of a cause other than cancer. Tables VII and VIII show detailed data of the 14 patients who had a curative resection, eight of whom are still alive. Of the four patients who had an early carcinoma detected in a Barrett's esophagus, three are alive, two without recurrent tumor, and one died as a

result of the surgical resection. Of the four with squamous cell carcinoma, three died, one of a cause other than cancer and two of recurrent tumor. Of the six patients with adenocarcinoma of the cardia, four are alive, all tumor free, and two have died, one of thromboembolism after the second stage of a two-stage procedure. Except for one patient, those alive and free of tumor have an excellent Karnofsky score. Gastrointestinal function is excellent to good in those free of recurrent disease. The current weight of the survivors ranges from 124 to 175 pounds. Discussion Inferences can be drawn from the findings of this study that can improve our understanding of cancer of the distal esophagus and cardia. They are best discussed by posing pertinent questions regarding the surgical management of the disease. What tumor characteristics are associated with a favorable survival, and can these be identified before resection? The current clinical staging for carcinoma of the esophagus uses the TNM system as adopted in 1978 by the American Joint Committee for Cancer Staging and End Results Reporting.'? Experience with this system has shown it to be imprecise, but it has identified several characteristics of esophageal cancer that are associated with improved survival: These include tumors that are less than 5 em in length, that do not extend beyond the esophageal wall, or that do not have lymph node metastasis. We endeavored to sharpen these observations by analyzing 58 patients in whom one or more of these favorable predictors were present." From this analysis it was shown that in patients who had favorable tumors, only metastasis to lymph nodes and tumor penetration of the esophageal wall had a significant and independent influence on prognosis. That is, the beneficial effects of the absence of one factor persisted even when the other was present. Other factors important in the survival of patients with advanced disease, such as tumor size, cell type, degree of cellular differentiation, or location of the tumor in the esophagus, had no effect on

Volume 95 Number 1 January 1988

Cancer of lower esophagus and cardia

49

Table VII. Data on patients who are alive after curative resection (as of March 31, 1987) Follow-up Patients

(rna)

Cell type

Stage

Karnofsky score

I 2 3 4 5

13 23 68 48 55 52 12 5

Adenocacinoma; Barrett's Adenocarcinoma; Barrett's Adenocarcinoma; Barrett's Epidermoid Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma

WONI W2NI WONO W2 N2 WINI W2NO W2Nl WI NO

100 60 100 30 100 100 100 100

6 7 8

Recurrence GI function

of tumor

Excellent Fair Excellent Poor

+*

Good Good Good Good

Legend: GI. Gastrointestinal. 'This patient had an esophageal perforation during dilatation of a benign-appearing stricture. Tumor was diagnosed at operation to close the perforation.

Table vm. Data on patients who died after curative resection Survival time Patient

(Mo!

Cell type

Stage

Cause of death

I 2

0 12 10 24 2 5

Adenocarcinoma; Barrett's Epidermoid Epidermoid Epidermoid Adenocarcinoma Adenocarcinoma

WONI WONI WONO WON2 W2NI W2 N2

Massive hemorrhage Tumor cachexia Stroke Tumor cachexia Thromboembolism Tumor cachexia

3 4 5

6

the survival of patients who had resection of their disease at an early stage. It also indicated that esophageal tumors that met the criteria of no wall penetration or lymph node metastasis could be defined as potentially curable regardless of size, histologic grade, cell type, or location. On the basis of this analysis, Skinner and associates" proposed a new staging system that classifies the depth of wall penetration (W), the presence of lymph node involvement (N), and the presence of systemic metastases (M). Table I shows the definitions for thevarious W, N, and M classifications. Skinner and associates have shown a significant difference in survival curves between those with favorable (WO-I NO-I, W2 NO-I, or WO-I N2) and those with unfavorable (W2 N2) classifications. The current study shows that for cancer of the distal esophagus and cardia, patients with afavorable stage of disease were able to be selected out by preoperative and intraoperative assessment and had a survival after a curative en bloc resection slightly better than reported by Skinner and colleagues for the same classification of disease. These results support a clinical approach to the disease in which an en bloc resection of the esophagus and stomach is advocated for patients most apt to benefit and in which those with more extensive disease who are less likely to benefit are spared the morbidity of the more extensive operation. What is the extent of resection necessary for cure ofcarcinoma of the distal esophagus or cardia? First,

Recurrence of tumor

+ + +

Table IX. Intrathoracic and intra-abdominal lymph node involvement in resected tumors of the distal esophagus and cardia Node location Paratracheal Subcarinal Paraesophageal Parahiatal Left gastric artery Common hepatic artery Splenic artery Right gastroepiploic artery Right gastric artery

Epidermoid" (%)

Adenocarcinoma" (%)

10 14 27 61 21

0 I 20 75 66

10 15 0

54 16

0

0

the resection should be sufficient to incorporate in the operative specimen all the potentially involved regional lymph nodes. Table VI shows the prevalence of intrathoracic and intra-abdominal lymph node involvement in the surgical specimens after radical esophagogastrectomy for our patients with early disease. The pattern is similar to that shown in Table IX, compiled from the work of Akiyama and associates" and Castrini and Pappalardo!' for specimens resected from patients with more advanced disease. Similar data have been collected from autopsy specimens. 15 From this experience, it appears that cancer of the distal esophagus or cardia

The Journal of

50

DeMeester, Zaninotto, Johansson

shows a preference for paraesophageal, parahiatal left gastric, and splenic artery nodes early in the course of the disease. Of importance is that both locations metastasize to abdominal and mediastinal lymph nodes. This is because of the extensive anastomosis between lymphatics draining the proximal stomach and distal esophagus. On the basis of these data, a resection for cure even at an early stage of the disease should include all lymph nodes from the midtracheal level down to the lower border of the celiac axis. These consist of the low paratracheal, subcarinal, paraesophageal, and parahiatal nodes in the thorax and the nodes associated with the left gastric, common hepatic, and splenic arteries in the abdomen. The borders of such a resection are outlined in Fig. 1 and define the en bloc resection done for cure in this study. Second, sufficient stomach and esophagus should be resected to incorporate the extensive submucosal lymphatics of both organs. Injection of the submucosal lymphatic plexus of the esophagus with contrast medium shows that the length of longitudinal lymph flow is about six times the length of the transverse flow.In the upper two thirds of the esophagus the lymph flow tends to move in a cephalad direction and in the lower third in a caudal direction." In the stomach, submucosal lymph channels contain valves that direct the lymph on the right side to the lesser curvature nodes and on the left side to the greater curvature nodes. Obstruction of submucosal lymphatics along the lesser curvature can cause a reversal of lymph flow toward the greater curvature through existing nonvalvular submucosal lymphatic channels." Consequently, cancers located in the lower esophagus or cardia can extend for a considerable length within the submucosal plexus superiorly in the esophagus or inferiorly in several directions in the stomach. On the basis of this knowledge, the only means of assuring complete removal of all submucosal tumor spread is to resect a substantial portion of the stomach and esophagus. This conclusion is supported by clinical experience. In patients with squamous cell carcinoma of the lower esophagus, a high esophageal anastomosis, and therefore a more extensive esophagectomy, tripled survival. I? Similarly, Tam and associates" reported a high incidence of anastomotic recurrences when the proximal esophageal resection margin was less than 10 ern. A similar emphasis on extensive esophageal resection can be made for adenocarcinoma of the cardia. Giuli and Sancho-Gamier," from the International Organization for Statistical Studies of Esophageal Disease data, reported a 19% 5-year survival rate for 340 patients with adenocarcinoma of the cardia. In 75% of the patients who had a late recurrence of tumor, it occurred, as with

Thoracic and Cardiovascular Surgery

squamous cell tumors, in the residual portion of the esophagus. These findings would encourage a more extensive esophagectomy than is currently done for cancer of the lower esophagus and cardia. The need for an extensive gastric resection in patients with adenocarcinoma of the cardia has been well documented. Papachristou and Fortner" have shown that in patients with adenocarcinoma of the cardia, without gross lymph node metastasis or wall penetration, a radical total gastrectomy resulted in an 83% 5-year survival rate, compared to 16% after partial gastrectomy. On the other hand, squamous cell cancers of the lower esophagus infiltrate the stomach only half as frequently as adenocarcinoma of the cardia infiltrates the esophagus. IS As a consequence, the recurrence of squamous cell carcinoma in the unresected portion of the stomach is unusual, and this is the basis for resecting less stomach in patients with this cell type. We have, however, seen recurrent squamous cell carcinoma in the submucosa of the stomach used for esophageal replacement in patients who had a total thoracic esophagectomy. It is evident from this experience that surgical success in the treatment of cancers located in the lower esophagus or cardia is dependent on near total resection of the organ involved with extensive resection of the neighboring organ. Thus, for patients with either adenocarcinoma or squamous cell carcinoma in this area who had resection for cure, we performed a near total removal of both the esophagus and stomach and reestablished gastrointestinal continuity with a colon interposition. Is a more extensive procedure necessary when a less extensive and safer procedure appears just as effective? The answer to this question is no, if indeed the less extensive procedure is safer and as effective in achieving cure. Advocates of transhiatal or blunt esophagectomy claim its advantages over a transthoracic procedure are relative safety and lower incidence of postoperative pulmonary complications." To determine whether this is so will require the ultimate in critical appraisal; namely, a randomized prospective clinical study that is as yet not available and not likely to be for some time. However, pertinent information can be gleaned from current comparative studies between the two operations. Giuli and Sancho-Gamier' reported the results of a prospective multicenter study of 750 patients undergoing a resection for carcinoma of the esophagus, 91 of whom had a transhiatal esophagectomy. Perioperative mortality was 19%, compared with 17% for resection via a left thoracotomy and 13% after a right thoracotomy. The frequency of pulmonary complications was similar to that after thoracotomy, as was the requirement for temporary postoperative ventilatory

Volume 95 Number 1 January 1988

support. Peracchia and Bardini" of the international Groupe European Etude Maladies Oesophage reported similar results in a retrospective multicenter study on 666 patients with malignant esophageal disease treated with a transhiatal esophagectomy. The effect of both of these reports is to question the greater safety and lower prevalence of complications associated with the transhiatal approach when applied universally by a variety of surgeons. Advocates for the transhiatal esophagectomy claim that the procedure is most appropriate for small tumors located in the distal esophagus, because in this location dissection can be done under direct vision." It is unlikely thatan en bloc mediastinal node dissection, as advocated in this study, can be performed through this approach. What effect this will have on long-term survival remains to be seen. Enthusiasts of the transhiatal approach, such as Orringer," claim that their early survival figures compare favorably to those obtained with the more radical procedure advocated by Skinner." This statement, however, is based on a 50% 2-year survival rate in only seven patients whose extent of disease was similar to that in our patients who had a radical operation." Failure to resect most of the stomach by both Orringer and Skinner may account for the similarity of results. It may be more important for survival to extend the resection of the stomach than to do an extensive lymph node dissection. It appears that transhiatal esophagectomy as currently done confers no particular advantage to the patient with carcinoma of the lower esophagus or cardia. It also interferes with accurate surgical staging and may have a distinct disadvantage on long-term survival by compromising an en bloc dissection, two time honored and repetitively proven principles of good surgical oncology. Our efforts have been to select patients by preoperative and intraoperative staging who are most apt to benefit from surgical therapy and to perform a safe and adequate resection based on the knowledge of the local spread of cancers in the cardia and lower esophagus and the principle of en bloc removal of disease. The 53% actuarial 5-year survival rate obtained by this effort supports continued use of the curative en bloc resection in selected patients. What should be assessed in patients considered for curative resection? Part of the preoperative clinical assessment is whether the patient is able to withstand the planned operation. It is futile to select an operation that is aimed at increasing the long-term survival for a patient whose physiologic life expectancy is short. Carcinoma of the esophagus and cardia is predominantly a disease found in patients between the ages of 50 and 70.

Cancer of lower esophagus and cardia 5 I

As a result, the presence of additional chronic illness, such as chronic obstructive lung disease or ischemic heart disease, is the rule rather than the exception. As a consequence, the patient must have sufficient cardiopulmonary reserve to tolerate the operation. The respiratory function is best assessed with an FEY! test, which should have results of 1.5 L or more. Any patient with an FEY! of less than 1 L is not a surgical candidate; first, because he is unlikely to survive the operation and, second, because if he does, his life expectancy is short due to his lung disease." Clinical evaluation and electrocardiogram are not sufficient indicators of cardiac reserve. A gated radionuclide pool scan is noninvasive and provides accurate information on wall motion and ejection fraction. The use of the scan as a preoperative screening tool has resulted in a significant reduction of operative mortality in patients undergoing major vascular reconstruction." In this regard, it is important to note that a normal ejection fraction at rest is 0.40 and remains relatively constant with advancing age." Therefore, a resting ejection fraction of less than 0.40 is an ominous sign regardless of age. We do not advocate extensive en bloc resection for such patients. A substantial number of esophagectomies are performed on elderly patients. It is therefore pertinent to assess the effect of age on the surgical risk. Advanced age is associated with decreased overall performance, regardless of the presence or absence of disease: In a study of marathon runners, who obviously have a high cardiopulmonary reserve, Stones and Kozma" found a rapidly decreasing performance after the age of 70. This decreased physical performance is also manifested in increased operative mortality. Sikes and Detmer," on the basis of a study of 15,930 surgical cases, showed a steady increase in surgical mortality with advancing age and a precipitous incline in mortality over the age of 75. Increased risk may be justified if outweighted by the expected benefits. Thus it is necessary to consider the overall life expectancy of this age group of patients. Since the tum of the century, the average life expectancy from birth has almost doubled, going from 47 years at 1900 to almost 75 in the 1980s. Although these figures are impressive, they are largely the result of advances in the elimination of acute disease and infant mortality. The maximum life span has not changed." There has been no increase in the number of centenarians or the maximum age at which death occurs in the past 150 years. From actuarial data of the federal government, a maximum average life span of somewhat less than 85 years can be calculated. Excluding trauma, suicide, and miscellaneous uncommon conditions, the

52 DeMeester, Zaninotto, Johansson

majority of deaths past the age of 75 are due to cancer, stroke, or heart disease; And since man must die, reducing his chance of death from a cancer in the esophagus will only increase his chance of death from heart disease or stroke. 30 In other words, curing a 75-year-old patient of cancer will only change the cause of death, but not the eventual outcome. As a result, the goal of surgical intervention at an advanced age is to palliate and "add life to years, not years to life." On the basis of this analysis, we have f'oncluded that after the age of 75 a radical operation for cure of carcinoma of the esophagus is unwise because of the additional risk and the reduced benefit. Should patients with Barrett's esophagus enter a surveillance program? Notably, four of our patients who had a resection for cure were discovered to have an early adenocarcinoma in a Barrett's esophagus. It appears that with appropriate surgical therapy these patients can be cured. This raises the question of whether patients with Barrett's esophagus should enter a surveillance program. To answer this question, the data on the risk of adenocarcinoma in Barrett's esophagus must be presented in a way that the clinician can gauge the importance of that risk. To do this, Speckler" has compared the prevalence of esophageal cancer in patients with Barrett's esophagus with that of a common malignant tumor in the general population such as lung cancer. Disease incidence rates are often reported as cases per 100,000 per year. So expressed, the incidence of lung cancer in white males, of the same age group as those with Barrett's esophagus, is 459 per 100,000 per year. In comparison, the incidence of adenocarcinoma in Barrett's esophagus is 500 to 10,000 per 100,000 per year. Speckler also points to areas with a high prevalence of esophageal carcinoma, such as Linxian County of Henan Province in China, where the mortality rate for esophageal cancer in men (similar to the incidence rate) is 161 per 100,000,21 Thus, the prevalence of esophageal cancer for patients with Barrett's esophagus appears to be greater than the risk for lung cancer in the United States and the risk for esophageal cancer in areas of China where mass surveillance programs for malignant disease have been implemented. On the basis of these comparisons and the possibility to cure an early carcinoma that develops in a Barrett's esophagus as shown in this study, patients with Barrett's esophagus should enter an effective cancer surveillance program.

REFERENCES 1. Orringer MB. Palliative procedures for esophageal cell cancer. Surg Clin North Am 1983;63:941-50. 2. Earlam R, Cunha-Melo JR. Oesophageal squamous cell

The Journal of Thoracic and Cardiovascular Surgery

carcinoma. I. A critical review of surgery. Br J Surg 1980; 67:381-90. 3. Giu1i R, Sancho-Garnier H. Diagnostic, therapeutic and prognostic features of cancers of esophagus: results of the international prospective study conducted by the OESO group (790 patients). Surgery 1986;99:614-22. 4. Sweet RH. Late results of surgical treatment of carcinoma of the esophagus. JAMA 1954;155:422-5. 5. Lu YK, Yueh ML, Yue ZG. Cancer of the esophagus and esophagogastric junction: analysis of the results of 1025 resections after 5 to 20 years. Ann Thorac Surg 1987; 43:176-81. 6. Skinner DB, Ferguson MK, Soriano A, Little AG, Staszak YM. Selection of operation for esophageal cancer based on staging. Ann Surg 1986;204:391-401. 7. Fatzinger P, Cimokowski GE, DeMeester TR. The role of nutritional therapy. In: DeMeester TR, Levine B, eds. Cancer of the esophagus. Orlando, Florida. Grune & Stratton, 1985:119-39. 8. Logan A. The surgical treatment of carcinoma of the esophagus and cardia. J THORAC CARDIOVASC SURG 1963; 46:150-61. 9. Skinner DB. En bloc resection for neoplasms of the esophagus and cardia. J THORAC CARDIOVASC SURG 1983; 85:59-71. 10. American Joint Committee on Cancer. Manual for staging of cancer. Philadelphia: J.B. Lippincott Company, 1983. 11. Rimm AA, Hartz AJ, Kalbfleisch JH, Anderson AJ, Hoffmann RG. Basic biostatistics in medicine and epidemiology. Norwalk, Connecticut. Appleton-CenturyCrofts 1980. 12. Skinner DB, Dowlatashy KD, DeMeester TR. Potentially curable cancer of the esophagus. Cancer 1982;50: 2571-5. 13. Akiyama H, Tsurumaru M, Kawamura T, Ono Y. Principles of surgical treatment for carcinoma of the esophagus: analysis of lymph node involvement. Ann Surg 1981;194:438-46. 14. Castrini G, Pappalardo G. Carcinoma of the cardia: tactical problems. J THORAC CARDIOVASC SURG 1981; 82:190-3. 15. Sous HU, Borchard F. Cancer of the distal esophagus and cardia: incidence, tumorous infiltration and metastatic spread. Ann Surg 1986;203:188-95. 16. DeMeester TR. Surgical anatomy of the esophagus. In: Shields TW, ed. General thoracic surgery, 2nd ed. Philadelphia: Lea & Febiger, 1983:82-91. 17. Giuli R. Surgical complications and reasons for failure. In: DeMeester TR, Levin B, eds. Cancer of the esophagus. Orlando, Florida. Grune & Stratton, 1985:198208. 18. Tam PC, Cheng HC, Ma L, Siu KF, Wong J. Local recurrences after subtotal esophagectomy for squamous cell carcinoma. Ann Surg 1987;205:189-94. 19. Papachristou DN, Fortner JG. Adenocarcinoma of the gastric cardia: the choice of gastrectomy. Ann Surg 1980; 192:58-64.

Volume95 Number 1 January 1988

20. Orringer MB. Transhiatal esophagectomy without thoracotomy for carcinoma of the thoracic esophagus. Ann Surg 1984;200:282-8. 21. Peracchia A, Bardini R. Total esophagectomy without thoracotomy: results of a European questionnaire (GEEMO). Int Surg 1986;71:171-5. 22. Wong J. Transhiatal oesophagectomy for carcinoma of the thoracic oesophagus. Br J Surg 1986;73:89-90. 23. Orringer MB. Transhiatal esophagectomy without thoracotomy for carcinoma of the esophagus. Adv Surg 1986; 19:1-49. 24. DienerOF, Furrows B. Further observations of the course and prognosisof chronic obstructive lung disease. Am Rev Respir Dis 1975;III :719-24. 25. Boucher CA, Brewster DC, Darling RC, Okada RD, Strauss HW, Pohost CM. Determination of cardiac risk by dipyridamole-thalium imaging before peripheral vascular surgery. N Engl J Med 1985;312:389-94. 26. Port S, Cobb FR, Coleman E, Jones RH. Effect of age on the response of the left ventricular ejection fraction to exercise. N Engl J Med 1980;303:1133-7. 27. Stones MJ, Kozma A. Adult age trends in record running performances. Exp Aging Res 1980;6:407-16. 28. Sikes ED Jr. Detmer DE. Aging and surgical risk in older citizens of Wisconsin. Wis Med J 1979;78:27-30. 29. Fries JF. Aging, natural death and the compression of senescence. N Engl J Med 1980;303:130-5. 30. Maloney JV. The limits of medicine. Ann Surg 1981; 194:227-55. 31. Speckler SJ. Endoscopic surveillance for patients with Barrett esophagus: does the cancer risk justify the practice? Ann Intern Med 1987;106:902-4.

Discussion Dr. Mark B. Orringer (Ann Arbor. Mich.). Dr. DeMeester's data, I believe, do not justify his conclusions. Of his 52 patients, 19, or 37%, had resectable tumors. Twenty-one, or 40%, required palliative resections for either stage III or IV disease. Thus, as is the case with most such series, the majority ofthese patients, 77%, had incurable disease when first seen and were not candidates for radical esophagectomy, which justifies the pessimism that has been put forth by many of us treating esophageal carcinoma. Ofthe remaining 14 patients described, five, or more than a third, had stage I tumors. These were small, superficial lesions that hadbeen discovered incidentally at the time of endoscopic examination and probably did not merit a radical esophagectomy. We know from the Chinese experience with early detection cytologic screening programs for esophageal cancer thatpatients with such early tumors limited to the mucosa and not even evident on a barium swallow examination can be treated very successfully with better than an 85% 5-year survival rate with esophageal stripping (transhiatal resection) alone. The analogy is the patient with uterine cervical cancer who canbe cured with a conization rather than a hysterectomy because the disease is superficial. Thus we have nine patients of Dr. DeMeester's original series of 52, or 18%, who he thinks can be treated under his "thesmaller the tumor, the bigger the operation" philosophy. We have found no difference in survival between 29 patients

Cancer of lower esophagus and cardia

53

undergoing transhiatal esophagectomy without thoracotomy and those undergoing a standard thoracoabdominal esophagogastrectomy for treatment of carcinoma of the cardia or distal third of the esophagus. Admittedly, the patients having esophagogastrectomy, unlike Dr. DeMeester's, did not undergo formal mediastinal and abdominal lymphadenectomy, splenectomy, and so forth. But does this really matter? Among 75 patients with distal-third tumors we have treated with transhiatal esophagectorny, removing the adjacent lymph nodes but not trying to do a formal en bloc resection, the survival rate for stage I and II tumors has been 100% at I year and between 100% and 69% at 2 years. The mean survival in these patients has been 6 years. Dr. DeMeester's reported survival after radical esophagectomy is virtually the same as ours. I think, in conclusion, that Dr. DeMeester has described an extremely large operation for patients with stage I and II tumors who have a good prognosis from the start. It is not the size of the operation that is determining survival in these patients, but rather the extent and biologic behavior of the individual tumor when it is first discovered. The thoracic surgeon's emerging role in the treatment of esophageal carcinoma is no longer that of providing wishful thinking that cure of this disease is achievable with a radical operation, but rather providing support for an interdisciplinary approach with radiation therapists, oncologists, and tumor immunologists to achieve better long-term survival in these patients. Dr. Watts R. Webb (New Orleans, La.). We are now realizing that the Halstedian principle of radical block excision with all the lymph nodes in the breast makes no difference whatsoever and that the regional spread indicates only the extent of the systemic disease. Do we know enough about the esophagus biologically at this moment to know that what we are doing here is not exactly the same thing: that the lymph node rule is of no value whatsoever and is just an indication of the spread and the extent of the systemic nature of this disease? Dr. DeMeester (Closing). I agree with Dr. Ellis [discussion not published] that an early surveillance program will be required to make an overall impact on the disease. Because of this we should not at the present time submit patients with early lesions to the philosophy that the goal of therapy is only palliation. Rather we should make the effort, as we have with other carcinomas, to identify early lesions and remove them with a curative operation. There currently is an opportunity for early surveillance in this disease. In fact, in four of the patients in our report the disease was diagnosed at an early stage during an evaluation for Barrett's esophagus. If the risk of malignancy in Barrett's esophagus is compared to the risk of carcinoma of the lung, the seriousness of the problem takes on a new perspective. The incidence of lung cancer is about 300 cases per 100,000 population per year. In comparison, if one assumes that patients with Barrett's esophagus have a potential of 0.5% for malignant changes, which is the lowest and most conservative estimate, the incidence would be 500 cases per 100,000 patients with Barrett's per year. This is a considerably higher incidence than the 162 cases of esophageal carcinoma in 100,000 population per year in the endemic areas of China where early-surveillance programs have been organized. Thus there certainly is every reason to institute early surveillance in patients with Barrett's esophagus. Because of this, I believe we should perform a procedure in patients with an early lesion that can give a 63% adjusted, or 53% actuarial, 5-year survival

The Journal of

54

DeMeester, Zaninotto, Johansson

rate. Why cut our gains with a procedure designed to provide only palliation, particularly when the need for palliation in these early lesions is minimal? Dr. Orringer inferred that five of our lesions were limited to the mucosa. Only two were limited to the mucosa; one extended into the wall; and the remainder extended through the wall or had spread to the lymph nodes. When the data from China are stratified according to wall penetration and lymph node involvement, their survival rate is about 60% after a "radical" operation when one or the other, but not both, of these factors are present. This is similar to our results. I would remind Dr. Orringer that II of our patients with early lesions had lymph node metastasis at the time of the operation. Our results and the China data indicate that an adequate resection should be done to remove the regional lymph nodes in patients with early lesions, rather than leaving the possibility of their removal to a chance phenomenon.

Thoracic and Cardiovascular Surgery

Dr. Orringer made a strong statement about his operation being curative for stage I and II disease, but his sample included only II patients with stage I and II disease and he only showed the 2-year survival rate. I believe when we compare his 5-year statistics for these patients with the statistics for those having a more extensive operation, done according to the principles we reported, the latter will provide greater patient survival. In answer to Dr. Webb's comment, this disease is biologically similar to other gastrointestinal carcinoma in which survival depends on both the extent of bowel wall penetration and lymph node metastasis. Consequently, I do not believe that the disease acts similarly to carcinoma of the breast, in which lymph node involvement appears to be the only major determinant of survival.

FDA grants available The Food and Drug Administration's (FDA) Office of Orphan Products Development (OPD) is soliciting applications for grants for clinical trials of products that are useful in treating a rare disease or condition. These products include drugs, biologics, medical devices. and foods for medical purposes. For information contact: Ms. Carol Wetmore, FDAjOPDjHF-35, 5600 Fishers Lane. Rockville, MD 20857, phone 301-443-4903.