Results of aggressive treatment of gastric sarcoma

Annals of Surgical Oncology, 1(3):244-251 Published by Raven Press, Ltd, © 1994 The Society of Surgical Oncology, Inc.

Results of Aggressive Treatment of Gastric Sarcoma

William Carson, MD, Constantine Karakousis, MD, Harold Douglass, MD, Uma Rao, MD, and Michael L. Palmer, MD

Background: Leiomyosarcoma and leiomyoblastoma are subtypes of gastric smooth muscle tumors. These rare tumors are usually treated with surgical resection. However, there is controversy regarding the optimal surgical management for these malignancies and little information is available on the effi- cacy of radiation and chemotherapy in the adjuvant or palliative setting.

Methods: The records of 32 patients with gastric leiomyosarcoma or teiomy- oblastoma were reviewed. Survival data were obtained and patient outcome was analyzed with respect to the type of treatment given. Four different staging systems were compared for their ability to predict survival.

Results: Thirty patients with leiomyosarcoma and two patients with leiomyoblastoma were followed after surgery. All 32 patients were explored, and 21 curative and 11 palliative procedures were performed. Adjacent organs were included in 38% of resections. Only three patients did not undergo gastric resection. Local recurrence developed in eight patients after curative resection for a local control rate of 62%. Eight other patients developed metastatic disease for an overall recurrence rate of 76% after curative resection. Median survival of patients undergoing curative resection was 40 months compared with 8 months for those having a palliative procedure. The estimated 5-year survival was 34% and t0%, respectively (p = 0.05). Twenty-five patients with advanced disease received systemic, hepatic arterial, or intraperitoneal chemotherapy. Eighty percent of patients received a regimen including doxorubi- cin. Four partial and one complete response were noted. Seven patients received postoperative radiation therapy. Fourteen patients underwent debulking surgery of recurrent or persistent disease in conjunction with chemotherapy. Chemotherapy, radiation therapy, and debulking surgery did not result in statistically significant prolongation of survival. Seven patients remain alive, two with liver metastases. Four different staging systems for gastric sarcomas were compared, but none of them were found to be clearly superior in pre- dicting survival.

Conclusions: Curative gastric resection was achieved in 66% of patients and resulted in a significant prolongation of survival as compared with patients who had a palliative procedure. Wedge resection of tumor or partial gastric resection appears to be an acceptable surgical approach to these tumors as long as negative margins can be obtained. Chemotherapy, radiation therapy and debulking surgery did not result in significant prolongation of survival in the face of advanced disease. None of the staging systems for gastric sarcoma currently in use is completely satisfactory. Tumor grade and extent of disease seem to be the most important factors when determining prognosis or considering adjuvant therapy.

Key Words: Gastric leiomyosarcoma--Leiomyoblastoma--Surgery.

Received March 30, 1993, accepted July 21, 1993. From the Departments of Surgical Oncology (W.C., C.K., H.D., M.L.P.) and Pathology (U.R.), Roswell Park Cancer Institute,

Buffalo, New York, USA. Address correspondence and reprint requests to Dr. Michael L. Palmer, Department of Surgical Oncology, Roswell Park Cancer

Institute, Elm and Carlton Streets, Buffalo, NY 14263, USA. The results of this study were presented at the 46th Annual Cancer Symposium of The Society of Surgical Oncology, Los Angeles,

California, March 18-21, 1993.

244

GA S TRIC S A R C O M A 245

Leiomyosarcoma (LMS) and malignant leiomyoblastoma (LMB) comprise the two histologic subtypes of gastric smooth muscle sarcomas. These rare tumors represent - 1 % of primary gastric malignancies. They often attain a large size before diagnosis and are known for their tendency to metas- tasize to the liver. The primary treatment of these tumors is surgical, although the optimal extent of resection has never been clearly defined. Recent reports have cast doubt on the need to obtain ex- tensive margins, and long-term survival has been documented in patients who were treated by simple wedge resection (1,2). Chemotherapy and radiation therapy are generally ineffective in the adjuvant setting or for palliation of advanced disease (3).

This report represents a retrospective review of 32 patients with LMS or LMB seen at the Roswell Park Cancer Institute. The purpose of this study was to characterize the presentation, diagnosis, and surgical management of this malignancy. Results of chemotherapy, radiation, and cytoreductive surgery were examined. In addition, existing staging systems were compared with respect to their ability to predict patient survival.

MATERIALS AND METHODS

Thirty-two patients with a pathologic diagnosis of gastric LMS or malignant LMB were identified by tumor registry search. The medical records of all patients treated after 1970 were included in this analysis. Histologic material was reviewed by one pathologist (U.R.), and tumor grade was assigned according to standard criteria for sarcomas of soft tissue (cellularity, vascularity, mitotic count, necrosis, and character of the stroma).

A curative resection was defined as the complete extirpation of tumor as evidenced by negative gross and microscopic margins. Any gastric resection in the face of residual gross, microscopic, or distant disease was categorized as a palliative resection. Four classification systems were then compared for their ability to predict patient outcome. Tumors were categorized according to grade (high or low), extent of tumor (intact serosa, through serosa/ locally invasive, distant disease), TNM stage (4), and according to the method of Shiu et al. (1) (Ta- bles 1 and 2).

Survival data were calculated by the method of Kaplan and Meier (5). Significance tests were based on the log rank test (6).

TABLE 1. American Joint Committee staging system for soft tissue sarcoma

T Primary tumor TI Tumor <5 cm T2 Tumor >5 cm

G Histologic grade Gt Low G2 Medium G3 High

N Regional lymph nodes NO Negative N1 Positive

M Distant metastasis M0 No distant metastasis M1 Distant metastasis

Stage I IA G1T1NOM0 IB GtT2NOM0

Stage II IIA G2T1NOM0 IIB G2T2NOM0

Stage III IIIA G3T1NOM0 IIIB G3T2NOM0

Stage IV IVA G1-3T1-2N1M0 IVB G1-3T1-2N0-1M1

RESULTS

Thirty patients with gastric LMS and two patients with malignant LMB were identified by tumor registry search. These patients were seen at the Roswell Park Cancer Institute between the years 1971 and 1991.

Twenty-four patients were male and eight were female. The median age at diagnosis was 57 years (range 13-81). Twenty-six patients were white and six were black.

Ten patients had a history of previous tobacco use and six had previously abused alcohol. No patient had a personal or family history of sarcoma. One patient's father had a hemigastrectomy for gastric adenocarcinoma.

Presenting symptomatology is given in Table 3.

TABLE 2. Staging scheme for gastric myosarcoma

Grade

Size

Invasion of adjacent organ

Stage 0 Stage I Stage II Stage HI

High Low <5 cm >5 cm Absent Present No unfavorable factors* One unfavorable factor Two unfavorable factors Three unfavorable factors ~

a Modification of original scheme.

Ann Surg Oncol, Vol. 1, No. 3, 1994

246 W. C A R S O N ET AL .

TABLE 3. Presenting symptoms

Symptoms No. of patients (%)

Pain 13 (41) Bleeding 9 (28) Weight loss 7 (22) Mass 4 (12.5) Nausea 2 (6) Dysphagia 1 (3) Dyspepsia I (3) None 2 (6)

The most frequent presenting symptom was abdominal pain, which occurred in 13 patients (41%). In nine of these patients, abdominal pain was the only symptom. Nine other patients presented with upper gastrointestinal bleeding and no other symptoms. Weight loss was noted in seven patients (22%), and four (12.5%) complained of an abdominal mass.

Physical examination demonstrated a palpable abdominal mass in 12 of 32 patients and another six were tender to palpation but exhibited no discrete mass. One patient presented with ascites. Nine patients had no discernable abnormality on physical examination (Table 4).

The three most commonly used diagnostic modal- ities were EGD (75%), upper gastrointestinal series (66%), and computed tomography scan (CT) of the abdomen (25%). Barium studies and CT scan demonstrated an abnormality in 80% and 87% of cases, respectively. Preoperative endoscopic biopsy of the tumor provided a diagnosis in 12 of 24 patients in which it was used. Thirty-five endoscopic procedures were performed in 24 patients. Sixteen patients underwent just one EDG, whereas, eight had more than one. Fifty percent of the biopsy results were positive for LMS in each group. The diagnosis of LMS was made at the time of surgery in one patient who presented with severe bleeding and re- quired an urgent laparotomy.

All 32 patients in this study were explored. Twenty-one curative and 11 palliative procedures were performed. Major procedures included esoph-

TABLE 4. Findings on physical examination

Finding No. of patients (%)

None 9 (28) Mass 7 (22) Tenderness 6 (t9) OB + 4 (12.5) Mass/cachexia 4 (12.5) MassRenderness 1 (3) Ascites t (3)

agogastrectomy (n = 2), total gastrectomy (n = 3), subtotal gastrectomy (n = 13), and partial gastrectomy (n = 11). All but three of the patients in the palliative group underwent gastric resection. Two of these three patients had exploratory laparotomy with biopsy only. The other patient died intraoper- atively, during the placement of a Denver shunt for ascites. Operative procedures are listed in Table 5. Formal lymph node dissection was not routinely performed. Adjacent organs were included in 38% of resections (12 of 32 patients). En bloc resections were classified as curative in eight patients and palliative in four. The spleen was included in 12 resections and the distal pancreas was included in six (Table 6).

Twenty-four of the 32 tumors were located in the body of the stomach. Other locations were the fun- dus (n = 5) and the cardia (n = 3). The median tumor diameter was 13.5 cm (range 3-60). Gross inspection of the specimen showed tumor ulcer- ation within the lumen of the stomach in 13 patients. Lymph nodes contained metastatic disease in only three of the 29 patients who underwent gastric resection; however, only one patient in this series had a formal lymph node dissection. In nine patients the tumor was contained within an intact serosa; in 12 it had eroded through the serosa or invaded adjacent organs; and in 11 patients the tumor was wide- spread at the time of diagnosis (liver metastases or peritoneal implants).

Eight of 21 patients developed local recurrence after curative resection. This occurred at a median interval of 21.5 months from diagnosis (range 6- 120). The median survival in this group of eight patients was 42 months (range 23-128). Three of these eight patients exhibited distant disease before local recurrence, and four developed metastases at the same time as local recurrence. The most common site of distant metastases was the liver (five patients). Risk of local recurrence did not correlate with the size, grade, or stage of the tumor or with the type of operative procedure.

TABLE 5. Curative and palliative resections

Curative Palliative Total

Esophagogastrectomy 1 1 2 Total gastrectomy 3 - - 3 Subtotal or partial

gastrectomy (including wedge resections) 17 7 24

Biopsy only - - 2 2 Denver shunt - - t 1

Ann Surg Oncol, Vol. I, No. 3, 1994

GASTRIC SARCOMA 247

TABLE 6. En bloc resec t ions

Gastric resection Adjacent organs n

Total gastrectomy Omentum, spleen, distal pancreas, gallbladder, left colon 1

Subtotal gastrectomy Spleen, distal pancreas, left adrenal, left kidney 1

Esophagogastrectomy Spleen, diaphragm, left adrenal 1

Subtotal gastrectomy Spleen, distal pancreas 4 Total gastrectomy Spleen 2 Esophagogastrectomy Spleen 1 Subtotal gastrectomy Spleen 1 Partial gastrectomy Spleen, diaphragm 1

An additional eight patients developed isolated metastatic disease without local recurrence after curative resection. This occurred at a median interval of just 13.5 months (range 1-113). The liver was the site of failure in all eight of these patients. The median survival of this group was 33 months (range 12-151). Thus, 16 of 21 patients who had curative resections developed recurrent disease; however, the local control rate was 62%. Five patients remain alive without evidence of disease after curative resection at 9, 10, 44, 124, and 174 months.

Of the 11 patients undergoing palliative resection, 10 had distant disease at the time of exploration. Nine of these 10 patients had liver metastases and one had peritoneal implantation of tumor. The remaining patient in this group was found to have tumor invading the head of the pancreas, which pre- cluded a curative resection. One patient in this group is alive with metastat ic disease at 176 months. One patient died after attempted resection of a hepatic metastasis. The remaining nine patients had a median survival of just 10.5 months (range 445) .

Twenty-five patients received chemotherapy during the postoperative period for persistent (n = 9) or recurrent disease (n = 16). A variety of cytotoxic compounds were used; however, 20 patients received an adriamycin-based regimen at one point. Other agents used included cis-platinum, dacarba- zine, vincristine, cyclophosphamide, etoposide, I- phosphamide/sodium-2-mercapto-ethane sulfate, actinomycin D, and methyl-cyclohexylnitrosoureas. Intraperitoneal chemotherapy with platinum was used in five patients, and another six received intraarterial chemotherapy for hepatic metastases via hepatic artery catheter. Of these 25 patients, all but five progressed in their disease. Four partial responses to chemotherapy were seen, and one complete response was noted. This occurred in a man

with Neural-based metastases who responded to nine cycles of methyl-CCNU. One partial response occurred with intraperitoneal cis-platinum, and two partial responses were seen after intraarterial therapy of liver metastases with adriamycin-based reg- imens. Unfortunately, the duration of the partial responses was <4 months in all cases. Partial response was defined as a 50% reduction in measurable disease.

Seven patients received radiation therapy after surgery. Two patients received radiation to the gastric bed for residual disease remaining after palliative resection, and two underwent radiation after curative resection. The radiation dosages in these four patients ranged from 3.600 to 4,500 cGy. The remaining three patients were irradiated for metastatic disease.

Fourteen patients underwent one or more additional operations for resection of recurrent or persistent tumor. Six patients had locally recurrent or metastatic disease that was completely resected at a second operation (two local recurrences, one abdominal wall lesion, and three hepatic lesions). The median interval to the second procedure was 13 months. One of these patients died in surgery during a hepatic resection. The survival in the remaining patients was 20, 26, 40, 40, and 45 months. Three of these six patients subsequently underwent additional debulking procedures in conjunction with systemic chemotherapy. All eventually died from liver metastases.

Eight other patients underwent debulking in con- cert with systemic chemotherapy for incompletely resectable local recurrences. The median interval from initial operation to reexploration was 27 months, and the median survival was 30 months. All eight patients eventually died of recurrent disease.

The overall median survival in this study was 34 months, and the estimated 5-year survival was 26%. Seven of 32 patients remain alive. Both patients with LMB are alive without evidence of disease, as are three patients with LMS. These five patients all underwent curative resection of low-grade tumors that had not metastasized. Two other patients with liver metastases from LMS are alive at 45 and 176 months. Twenty-five patients in this series have died. Liver metastases were the cause of death in 18 of 25 patients (72%). Of the 27 patients who developed recurrent disease, all but one had liver metastases. Other major sites of recurrent disease were the peritoneum (n = 17) and lung (n = 4).

Ann Surg Oncol, Vol. t, No. 3, 1994

248 W. C A R S O N E T AL.

Operative procedure (curative vs. palliative), chemotherapy, radiation therapy, and debulking surgery were evaluated for their influence on survival. Curative surgical resection was the only ther- apeutic modality with a statistically significant im- pact on survival. The median survival of patients undergoing curative resection was 40 months compared with 8 months for those undergoing a palliative procedure. Moreover, the estimated 5-year survival after a curative resection was 34% compared with 10% for a palliative resection (p = 0.05) (Ta- ble 7).

Four staging systems were compared. The gastric sarcomas in this series were evaluated by grade, extent of gastric wall penetration, TNM stage, and according to the staging system of Shiu et al. (Table 8). Nineteen tumors were low grade and 13 were high grade. Median survival was 44 months and 23 months respectively. This difference was statistically significant at p = 0.03 by the log rank test.

Nine patients had tumors that were contained within an intact serosa, 12 tumors invaded through the serosa or into adjacent organs, and 11 had given rise to metastases. The median survival for these three stages was 45, 34, and 8 months, respectively. These differences did not reach significance.

Under the TNM staging system, 12 patients were classified as stage I, nine were stage III, and 11 were stage IV. No patient was classified as stage II. The median survival for stages I, III, and IV was 76, 26, and 8 months, respectively. The superior survival of stage I patients approached significance at p = 0.06.

According to the staging system of Shiu et al., three tumors were stage 0, 13 were stage I, nine were stage II, and seven were stage III. The median

T A B L E 7. Therapy o f gastric sarcoma

Therapy

Estimated Median 5-yr survival survival

n (too) (%) p

Resection Curative 21 40 34 0.05 Pallative 11 8 0

Chemotherapy Yes 25 27 19 0.23 No 7 124 67

Radiation Yes 7 40 43 0.19 No 25 24 21

Debulking Yes 14 34 14 0.42 No 18 27 37

TABLE 8. Staging o f gastric sarcoma

Estimated Median 5-yr survival survival

Stage n (too) (%) p

Grade Low 19 44 39 0.03 High 13 23 8

Extent Contained 9 45 50 0.12 Through serosa 12 34 25 Metastatic 11 8 10

TNM I 12 76 55 O.06 III 9 26 11 IV 11 8 9

Shiu 0 3 1 13 45 42 0.05 2 9 24 I1 3 7 7 14

For the three patients staged " O " by Shiu, survivals were 7,44, and 46 months.

survival of stages I, II, and III were 45, 24, and 16 months, respectively. The increased survival of stage I tumors was significant at p = 0.05.

DISCUSSION

Thirty-two patients with gastric LMS or LMB were followed after surgical exploration. The median age was 57 years, which is one decade younger than the average age of the patient with gastric adenocarcinoma (7). As in other studies, male patients predominated (2,8). No patient had a personal history of previous gastric malignancy, and no family history of sarcoma was identified. There was a relatively high rate of tobacco and alcohol abuse in this patient population.

Abdominal pain was the most frequent presenting symptom in this series. It was present at diagnosis in 40% of patients. This pain was not characteristic and often mimicked symptoms of peptic ulcer disease or gastrointestinal reflux. This may be one explanation for the large size of these tumors at diagnosis (2,7-9). Twenty-eight percent of patients presented with melena or hematemesis. Gastroin- testinal bleeding is a common mode of presentation in patients with gastric LMS and may even require urgent laparotomy, as was the case with one patient in this series (9-11).

The presence of a gastric tumor was documented preoperatively in the majority of patients. Sixty-six percent of patients had an upper gastrointestinal series and 25% were studied via CT scan. The sensi-


GASTRIC SARCOMA 249

tivity of these studies in detecting an abnormality was >80%; however, a preoperative diagnosis of LMS was entertained in only two patients. In con- trast, gastroscopy was used in 75% of patients, and a diagnosis of LMS was made in 50% of these cases. Some patients did require more than one endoscopy in order to obtain sufficient material for examination, but in two thirds of cases the diagnosis was made at the first procedure. In previous series, endoscopy has had a low yield, with several investigators reporting no positive biopsy results whatso- ever. Bedikian reported three positive biopsy results in 12 patients, and this is one of the best experiences with endoscopy before our series (2,8, 9). We have found that the diagnosis of LMS can be made on relatively small amounts of tissue as long as areas of necrosis are avoided. Also, the surgeon should not hesitate to repeat the endoscopy if the initial procedure yields insufficient tissue for diagnosis.

All 32 patients were explored, and a curative resection was possible in 21. This translates into a resectability rate of 66%. Eleven patients underwent a palliative procedure, and in 10 of these patients metastatic disease was present at the time of exploration. Therefore, in the absence of distant disease, a curative resection was possible in the majority of patients. The overall local control rate after curative resection was 62%. This is an important finding given the tendency of these tumors to cause abdominal pain and bleeding. Another important consideration is the prolonged survival that is possible in the face of metastatic disease. Eight of 11 patients in the palliative group underwent gastric resection, and only two of these patients had evidence of local recurrence at the time of their death. Therefore, in our experience, gastric resection can provide good local control even for patients with distant disease.

The ability to perform a curative resection carried prognostic significance. The estimated 5-year sur- vi.val of these patients was 34% as compared with 10% for those undergoing palliative procedures. This difference in survival was significant at p = 0.05. Obviously, the poor outcome of patients who could not have a curative resection was probably a function of the high incidence of advanced disease in this group at diagnosis. Nevertheless, the finding of prolonged survival after curative resection sug- gests that every attempt should be made to achieve a complete resection of tumor, even if this should necessitate an extended resection.

The choice of operation for gastric LMS should reflect the size and invasiveness of the tumor. In the present series, only three total gastrectomies and two esophagogastrectomies were performed. Sub- total or partial gastrectomy was used in over 70% of patients, including four of the patients who are currently alive without evidence of disease. The diffi- culty with recommending a particular surgical approach is that the requirements for margin of resection are not known for gastric LMS. Lindsey et al. proposed that a 2-cm margin was adequate. Other investigators have gone so far as to recommend that a 10-cm margin of normal tissue be taken in conjunction with the appropriate lymphatic tissues for all lesions >5 cm in diameter. However, Grant et al. found that the extent of resection was not statistically associated with survival after adjusting for tumor size and grade. And Shiu et al. found no recurrence in six tumors that were removed by wedge resection alone. Apparently, the success of limited resections in gastric LMS is due to the lack of lat- eral mucosal or submucosal extension of tumor. Thus, routine total or subtotal gastrectomy is not warranted for this malignancy. Partial gastrectomy or wedge resection is a reasonable alternative as long as a 2- to 4-cm margin can be obtained. In some cases the entire stomach will be replaced by tumor, and in this situation, the surgeon should not hesitate to undertake a total gastrectomy or en bloc resection in order to obtain adequate margins (1,2,12).

An aggressive surgical approach was used in the treatment of tumors when invasion of adjacent or-. gans was present. En bloc resection of adjacent tissues was necessary in eight patients who underwent curative resection. However, it should be noted that the rate of local recurrence was the same whether or not resection included adjacent organs. In addition, no long-term survivors were seen in the face of local invasion of adjacent organs, even when wide en bloc margins were obtained. Thus, although local invasion of tumor may not preclude a complete resection, it does appear to be a poor prognostic factor.

Lymph node dissection was not routinely performed in these patients, although the low rate of lymph node metastases in this and other series sup- ports this approach (7). Only three patients in this study had lymph node disease, and in all three cases it appeared as if the nodes were involved by direct extension of tumor. In other series, prolonged survival after gastric resection of LMS was achieved without lymph node dissection. In general, aggres-

Ann Surg Oncol, VoL I, No. 3, 1994

250 W. CARSON ET AL.

sive dissection of lymph node-bearing tissue appears to be unnecessary when resecting these tumors (1,2).

Chemotherapy was administered to all patients with persistent or recurrent disease. Chemotherapy was not used in the adjuvant setting. The overall response rate was 20%, but only one complete response was noted. Unfortunately, the duration of the partial responses was very short, ranging from 1 to 4 months. Two of the four partial responses occurred in patients who received adriamycin via the hepatic artery for liver metastases. This route of administration might bear further investigation should more effective cytotoxic agents for sarcoma ever become available.

The median survival of those patients receiving chemotherapy was 27 months. No control group is available for comparison because all patients with recurrent or metastatic disease received chemotherapy. Therefore, it is difficult to assess the effect of chemotherapy on patient survival in this series. In general, there has been little experience with chemotherapy in the setting of advanced disease. In Shiu's report from the Memorial Sloan-Kettering Center, no responses were noted in the five patients receiving chemotherapy for metastatic disease, and median survival was not noted (2). In a report from the M.D. Anderson Hospital, 17 patients received chemotherapy for disseminated disease, and only two partial responses were documented. The median survival of patients with liver metastases who received chemotherapy was 14 months versus 8 months for those who were not treated. This difference did not reach statistical significance (8). Thus, there are no strong data to support the use of chemotherapy in the setting of advanced disease, other than as part of a controlled trial. Another potential use of chemotherapy would be in the adjuvant setting after curative resection. Clearly, no standard regimen exists, and some would question the need for adjuvant therapy in a malignancy that could be controlled in many cases by simple wedge resection (1,3,13).

Radiation therapy was used in seven patients in this study. Only two patients received radiation after curative resection as an adjuvant therapy. This makes it difficult to assess the effectiveness of this modality. In any case, the median survival of patients receiving radiation therapy was 40 months compared with 24 months in the remaining patients. This difference did not reach significance, but it is provocative. The total dose of radiation that can be

administered to the gastric bed by external beam techniques is limited due to the closed proximity of radiosensitive tissues. For this reason, intraopera- rive techniques have been promoted by some investigators as a way of delivering meaningful amounts of radiation to the gastric bed (3).

Fourteen patients were explored for recurrent disease. A complete resection was possible in six of these patients, and their median survival was 40 months. Eight other patients could not be fully resected at the second procedure, and tumor debulking was performed. Their overall median survival was 30 months. The small survival advantage associated with complete resection is not statistically significant and probably reflects a lesser tumor bur- den. There is no proof that the natural history of recurrent gastric LMS can be altered by surgical resection, but in the good-risk patient, there is no reason why a resection of recurrent disease cannot be attempted, especially if symptoms are present.

Four different tumor staging systems were compared with respect to their ability to predict patient survival. Tumor grade, tumor extent, TNM stage, and the staging system of Shiu et al. were examined. Of these, only tumor grade possessed significant prognostic power. Thus, for gastric LMS, survival appears to correlate most highly with the pathological grading of the tumor. Surprisingly, survival was not dependent on the extent of the tumor at diagnosis, although a clear trend in median survival could be seen as tumors progressed to local invasion and distant disease. The TNM staging system approached significance for stage I tumors. Tu- mors were classified as high or low grade only, thus explaining the absence of stage II tumors in our use of the TNM system. It is clear from our data that the lymphatic system is not an important route of spread for gastric LMS. Lymph node status is probably not necessary in staging this tumor. The staging method of Shiu et al. was significant at p = 0.05 for stage I tumors only. As with all the other systems excepting tumor grade, the absolute number of patients in each category was small. This system uses tumor risk factors in staging (high grade, size >5 cm, and presence of local invasion). However, it is not clear that 5 cm is an important cut-off size in gastric LMS. For instance, in this series four of six tumors <5 cm in size were associated with liver metastases at the time of diagnosis. In fact, this system does not directly account for metastatic disease, and in our patient population two of three stage 0 tumors had liver metastases at presentation.

Ann Surg Oncol, VoL I, No. 3, I994

GASTRIC SARCOMA 251

An alternative staging system for gastric LMS might include tumor grade as well as extent. This system would not directly account for tumor size, but we have found that this may not be as important a factor as previously thought. The small numbers of patients available for study greatly hinders the statistical analysis of new staging systems.

REFERENCES

1. Shiu MH, Farr GH, Papchristou DN, Hajdu SI. Myosarco- mas of the stomach. Cancer 1983;49:177-87.

2. Grant CS, Kim CH, Farrugia G, Zinsmeister A, Goellner JR. Gastric leimoysarcoma. Arch Surg 1991 ;126:985-90.

3. Dougherty M J, Compton C, Talbert M, Wood WC. Sarco- mas of the gastrointestinal tract. Ann Surg 1991;214: 569-74.

4. Russell WO. A clinical and pathological staging system for soft tissue sarcomas. Cancer 1977;40:1562-70.

5. Kaplan EL, Meier P. Nonparametic estimation from incom- plete observations. J A m Star Assoc 1958;53:457-86.

6. Cox DR. Regression models and life-tables. J R State Soc 1972;34:187-220.

7. Giberson RG, Dockerty MB, Gray HK. Leimoysarcoma of the stomach. Clinocopathologic study of 40 cases. Surg Gy- necol Obstet 1954;78:186-96.

8. Bedikian AY, Khankhanian N, Valdivieso M, et al. Sarcoma of the stomach: clinicopathologic study of 43 cases. J Surg Oncol 1980;13:121-7.

9. Estes NC, Cherian G, Hailer CC, et al. Advanced gastric leimoysarcoma. Am Surg 1989;55:353-5.

10. WelchJP. Smooth muscle tumors of the stomach. Am J Surg 1975;130:279-85,

11. ReMine WH. Gastric sarcomas. Am J Surg 1979;120:320-3. 12. Lindsay PC, Ordonex N, Raaf JH. Gastric leimoysarcoma:

clinical and pathological review of fifty patients. J Surg On- col 1981 ;18:399-421.

13. McGrath PC, Neifeld JP, Lawrence W, Kay S, Horsley JS, Parker GA. Gastrointestinal sarcomas analysis of prognostic factors. Am Surg 1987;206:706-10.


Documents

Results of aggressive treatment of gastric sarcoma