Gastrointestinal metastases from malignant melanoma

Surgical Oncology 1995; 4: 105-110

Gastrointestinal metastases from malignant melanoma N. RICANIADIS, M. M. KONSTADOULAKIS, D. WALSH* AND C. P. KARAKOUSIS Departments of Surgical Oncology and *Computer Center, Roswell Park Cancer Institute, Buffalo, New York 14263, USA

Between 1980 and 1992, 68 patients with clinical indications of involvement of the gastrointestinal (GI) tract with metastatic melanoma were treated at Roswell Park Cancer Institute. Presenting symptoms were anaemia, abdominal pain, nausea and vomiting. Sites commonly involved were the small bowel (75%), the large intestine (25%), and the stomach (16%). Twenty-one patients were considered unsuitable for surgery; their median survival after diagnosis of GI metastases was 2.9 months. Fortyseven patients underwent abdominal surgery; effective palliation was achieved in most of them. Complete resection of GI metastases was accomplished in 47% of patients. The median survival after operation was 27.6 months for patients with complete resection of GI metastasis and no other disease, 5.1 months for patients with resection of involved GI tract and other metastases present, and 1.9 months for patients who had a by-pass procedure only. The 5-year survival for patients with complete resection of GI metastases and no other evidence of disease was 28.3%. The other groups had only 1-year survivors. Surgical intervention is justified on the basis of these findings, and extended palliation can be achieved in patients with complete resection of metastatic disease. Surgical Oncology 1995; 4: 105-110. Keywords: gastrointestinal metastases, malignant melanoma, management, survival.

The purpose of this study was to investigate the role of surgical intervention in the clinical treatment of the melanoma patient with GI metastasis, and define the group of patients who would benefit the most from an aggressive approach of surgical resection.

INTRODUCTION Clinically detected in only 4-8% of living patients with disseminated melanoma [1,2), gastrointestinal (GI) tract metastases remain a common site of haematogenous spread in patients who eventually develop disseminated disease. Autopsy series report multiple site involvement in 95% of patients who die of melanoma [3). GI tract involvement is reported in almost 60% [2); when liver and spleen are included, the incidence is 86.3% [4). The strictly palliative role of surgery in patients with GI metastasis from melanoma has been previously emphasized [5-10). However, certain patients can be given extensive palliation through resection of the localized metastatic lesions, with long-term survival being reported for carefully selected patients [11-18).

MATERIALS AND METHODS Between 1980 and 1992, 68 patients presented at Roswell Park Cancer Institute (RPCI) with clinical indications of involvement of the GI tract with metastatic melanoma (liver and spleen excluded). Their charts were reviewed and all data concerning the characteristics and treatment of their primary, stage at presentation of the primary, the metastatic pattern prior to onset of the GI symptoms, the presentation, diagnosis and treatment of GI metastases and follow-up were recorded and analysed. The American Joint Committee on Cancer staging system was used [19]. The patients were subdivided for statistical analyses into four

Correspondence: C. Karakousis, MD, PhD, Surgical Oncology Department, Roswell Park Cancer Institute, Elm & Carlton Streets, Buffalo, New York 14263, USA. ©1995 Blackwell Science Ltd

105

106

N. Ricaniadis et al.

groups according to their mode of treatment: • Group 1: Complete resection of GI metastasis; no other disease • Group 2: Resection of involved GI tract; other metastases present in the remaining GI tract or elsewhere • Group 3: By-pass of lesions only • Group 4: No surgery; other treatment or observation There were 43 men and 25 women with a median age at diagnosis of GI . metastasis of 51 years (range : 18-84 years). The site of primary disease was head and neck in 10 patients (15%), trunk in 26 (38%), extremities in 14 (21%), eye in 4· (6%), mucosal in 3 (4%), and unknown in 11 (16%). Median Breslow thickness of the primary lesions was 2 mm (range: 0.20-22.0 mm). Definitive treatment of the primary lesion was carried out in all patients; 16 elective lymphadenectomies were performed, 6 of them with positive nodes. A total of 37 patients (54%) manifested locoregional metastases prior to diagnosis of GI disease. Forty-eight patients (70%) developed other systemic disease prior to diagnosis of GI metastasis; the most common sites included soft tissue in 18 patients (26%), lung in 13 (19%), brain in 5 (7%), bone in 5 (7%), and liver in 4 (6%) . GI metastasis was the first site, or among the first, of systemic metastases in 20 patients (41.6%). The median time from definitive treatment of the primary tumour to the diagnosis of GI metastasis was 33 months (range: 1-390 months). For patients with unknown primary melanoma, the median time to GI metastasis was calculated from the first histologic diagnosis that the patient had melanoma. The median time varied according to the stage of the disease at the time of diagnosis of the primary lesion; it was 50.4 months for patients who were stage I and II at diagnosis, 12.9 months for stage III patients, 1.5 months for those staged IV, and 9.7 months for patients with unknown primaries. The major presenting symptoms at diagnosis of GI metastasis were anaemia in 24 patients (35%) with frank GI haemorrhage in 14 (21%) , abdominal pain in 33 patients (48.5%), nausea andvornltinq in 27 (39.7%) , fatigue in 20 (29.4%), symptoms of complete or recurrent intestinal obstruction in 14 (20.5%), anorexia in 15 (22.0%), and change in bowel habits in 9 (13%). Jaundice was noticed in 4

patients (5.8%) and malabsorption in 1. The duration of symptoms prior to diagnosis of GI metastasis ranged from 1 to 12 weeks. Nineteen patients (27.9%) noted weight loss (median weight loss: 9 kg; range: 5-25 kg). An abdominal mass was detected in 10 patients (15%), and malignant ascites in 5 (7.4%). Upper GI series and/or barium enema were positive for metastases in 25 patients. Barium swallow was positive in 16 (27%) . Endoscopic procedures were positive in 15 patients (22%). Abdominal computed tomography (CT) was abnormal in 31 patients (46%) and negative in 9 (13%). Ultrasound was abnormal in 8 patients (12%). Of a total of 153 diagnostic tests performed in these patients, 97 (63%) were positive. Of the 68 patients, 28 had a single area of involvement in the GI tract and 40 had multiple sites. Twenty-one- patients were not operated on (30%), most of them because of the advanced nature of their disease (group 4). Most of these patients were treated w ith chemotherapy. Patients were considered to have advanced disease when their life expectancy was less than 3 months due to wide dissemination of their disease. These patients were not operated upon when the ir symptoms were vague or not life-threatening, e.g. chronic anaemia secondary to slow GI bleeding. However, in the operative group there were also patients with advanced disease who due to bowel obstruction or rapid GI bleeding required an emergency operation for relief of the acute symptoms. Forty-seven patients had surgery (69%) including 29 men and 18 women who had a median age at operation of. 49 years (range: 18-71 years). GI metastasis was the only site of melanoma at the time of surgery for 18 patients. Other metastatic sites were present synchronously with the GI metastasis in 35 patients; these included liver in 16 patients, soft tissue in 13, lung in 11, brain in 7, distant nodes in 4, bone in 3, and regional nodes in 1. The most commonly involved sites of the GI tract were the jejunum in 22 patients (47%), the ileum in 16 (34CYo), the omentum in 16 (34%), the large bowel in 12 (25%) , the duodenum in 10 (21%), the pancreas in 7 (15%), the gall bladder in 5 (11%), and the stomach in 3 (6%). The lymph nodes in. the mesentery of small and large bowel were involved with tumour in 21 patients (45%). Seven points of © 1995 Blackwell Science Ltd, Surgical Oncology, 4: 105-110

Melanoma GI metastasis

intussusception, 11 points of bleeding, and 1 gastric perforation were found. Complete resection of GI metastases was accomplished in 22 patients (47%); 12 of them were rendered disease-free after the operation, including two patients requiring pancreaticoduodenectomies (group 1). Thirty-two patients had other metastases remaining (group 2) and 4 had by-passes only (group 3). Additional postoperative chemotherapy and/or immunotherapy was given to selected patients, in mos'{ cases for residual disease. Estimates.. of survival were obtained by the method of Kaplan and Meier [20]. Comparisons of survival distributions were performed using the logrank test [21].

107

28.3%, respectively. Groups 2, 3 and 4 had only 1-year survivors (12.9%, 25% and 4.8%, respectively). Breslow thickness, presence or absence of ulceration, and site of primary tumour did not influence survival after operation for GI metastasis. Furthermore, postoperative survival was not influenced by the known versus unknown primary, and age or gender of the patient. Three of the patients who were disease-free after the first operation experienced GI symptoms and were operated on again. The time between surgical treatment of the first GI metastasis and recurrence in the GI tract was 8.9, 20.3 and 25.6 months. Complete resection of GI tumour was achieved in all and 2 were rendered disease-free. Survival after the second operation for these 3 patients was 1.8, 6.9 and 26.8 months.

RESUL TS Thirty-five (73%) of the operated patients were relieved of their preoperative symptoms; the remaining patients experienced only partial relief. Postoperative complications developed in 14 patients (29%). These included postoperative ileus in 4 patients (8%), fever in 4 (8%), sepsis in 2 (4%), wound infection in 2 ~(4%), and pulmonary embolus, intra-abdominal haemorrhage, and myocardial infarction in one patient each (2%). Five of these patients (11%) died within 30 days of surgery. The causes were sepsis, pneumonia, myocardial infarction, intra-abdominal haemorrhage, and prolonged postoperative ileus due to progression of disease. In 33 patients, the postoperative course was uneventful (70%). The median hospital stay was 13 days (range: 1-69 days). The median survival after operation for Gl metastasis was 5.66 months, significantly better than the 2.9 months median survival of patients who were not surgically treated (P = 0.0035). The median survival after the diagnosis of GI metastasis was significantly better for group 1 compared to the other groups (P = 0.0003). Median survival was 27.2 months for group 1 (range: 0.6-66.5 months), 5.06 months for group 2 (range: 0.4-20.2 months), 1.9 months for group 3 (range: 0.7-12.6 months), and 2.9 months for group 4 (range: 0.3-11.6 months). The 1-, 2-, and 5-year survival rate for group 1 was 63.6%, 53.0% and © 1995 Blackwell Science Ltd, Surgical Oncology, 4: 105-110

DISCUSSION A relative delay is often noticed between the onset of symptoms and the diagnosis of GI metastasis, as seen in this and other series [11, 13). At the stage of wide dissemination, the often non-specific GI symptoms may be attributed to benign conditions or to systemic treatment. Complaints such as crampy abdominal pain, fatigue, nausea, and weight loss, among others, are indicative of GI metastasis in melanoma patients. Serious complications can occur with these metastases; two-fifths of our patients had haemoglobin levels of 10 g dl- 1 or less, more than one-fifth had symptoms of intestinal obstruction, and one patient had a gastric perforation. Gl contrast studies and endoscopy proved to be effective in detecting metastases in our series, as others have also found [7, 22, 23]. However, contrast studies and abdominal CT may occasionally be negative, despite the presence of metastases, as seen in this and other studies [13, 15, 24]. Omental, mesentery, and peritoneal implants, although a common operative finding in our patients, were often not detected preoperatively; a similar experience was reported by others [4, 17). Obliteration of mesenteric or omental fat was often not evident on abdominal CT, according to one report; this possibly reflects the fact that metastatic melanoma deposits do not

108

N. Ricaniadis et al.

generally incite a fibrotic reaction, unlike carcinoid tumour or scirrhous adenocarcinoma of the small intestine [22, 24]. GI metastasis was the first or among the first sites of systemic metastases in 20 patients (41.6%). Comparable results (17-33%) have been reported [12, 14, 16]. GI metastasis was the first and only site of distant relapse after regional nodal metastases in 2 patients (2.9%), in accordance with the incidence of ;-3% in previous reports [5, 25]. Surgery was undertaken in about 70% of our patients, similar to 76.6% reported from another institution [14]. In another report, only 38% of the patients with abdominal disease were surgically treated (26). The pattern of Gl tract involvement at operation was similar to that in other reports [6, 11, 12]. Seventy-five per cent of the operated patients had involvement of the small bowel or its mesentery. Large, ulcerated tumour masses penetrating the intestinal wall and infiltrating the mesentery were found in 9 patients in our study; this was also seen in other series [14, 22, 27). Gall bladder involvement was found in 5 of our patients, a high incidence compared to other reports (14). Although melanoma is th~ most common metastatic lesion involving the gall bladder [10], the number of cases descr il:ied in the literature is very low [10, 14, 28). Involvement of this organ is found in 14-15% in autopsy series ' [28). A broad spectrum of operative procedures was performed, depending on the indication for operation and the site of metastasis. It was often necessary to perform two segmental resections in the small bowel; two of our patients had three small bowel resections, although it is generally desirable to limit the number of anastomoses to two and preferably to one (29). More than one anastomosis was used in patients w ith multiple and widely separated lesions, in whom to have one anastomosis with resection of all lesions would have required resection of most of the small bowel. The mortality (11%) and morbidity rates in our series are in agreement with the literature; reported are mortality rates of 0 to 20% [6,7,12-17,30,31). In one report [18], 2 out of 5 patients died after surgery (40%). Previous reports have documented a survival advantage in patients with distant metastases who are treated by surgery, versus those who are not

[26, 32). Moreover, survival after surgery is reported to be better after complete resection of stage IV disease, than after partial resection [6, 9, 26, 33 , 34). The median survival after surgery in this series was significantly better for the patients who were rendered disease-free, having had complete resection of GI metastases. Similar results have been noted by others [6, 16). The median survival of all patients who underwent surgery in our series is in agreement with the literature, where it ranges from 4.5 to 17.3 months [7, 8,11-16]. The patients who were rendered disease-free had a median survival after surgery for GI metastases of 27.2 months, which is comparable to the median of 18 months reported in another study [33]. No difference in survival after operation was noted with respect to age, gender, characteristics of primary, or known versus unknown primary, in agreement with previous reports [11, 26, 34, 35] . The observed palliation of symptoms after operation is similar to that of other series; 79-90% are reported to experience relief of their preoperative ' symptoms [11, 13, 14]. Only one. of the non-surgical patients survived over 1 year; of the surgical patients, 8 patients (17%) survived 1 year after the operation, 3 survived over 2 years (6%), and 2 exceeded 5 years (4%). Extended palliation can be achieved with complete resection of metastatic disease; only patients who were rendered disease-free after the operation survived more than 2 years in this series. Even extensive surgery is justified; one patient was alive and disease-free, 32 months after a pancreaticoduodenectomy. In a study of stage IV disease, patients who were repeatedly rendered tumour-free through several procedures had a significantly longer median survival after resection than patients rendered disease-free only once [32). In the absence of effective systemic treatment, surgical intervention in selected patients with GI metastases from melanoma must be considered as a worthwhile modality. Surgery offers effective palliation to these patients (alleviates symptoms that compromise the patients' quality of life, and prolongs survival). Although long-term survival is not likely, extended palliation can be achieved when complete resection of Gl disease is feasible. Resection 'of asymptomatic GI tract metastases found incidentally may be worthwhile in prolonging © 1995 Blackwell Science Ltd, Surgical Oncology, 4: 105-110

Melanoma GI metastasis

the survival of patients with of other lesions which are disease-free interval from melanoma exceeding 1 year

no or a small number also resectable and a previous presence of (36).

REFERENCES 1. Blessing K, Park K, McLaren K, et al. Gastrointestinal involvement with metastatic melanoma: clinical and pathological features. J R Call Surg Edinb 1986; 5: 293-5. 2. Das Gupta TK, Brasfield RD. Metastatic melanoma of the gastrointestinal tract. Arch Surg 1964; 88: 969-73. 3. Patel JK, Didolkar MS, Pickren JW, et al. Metastatic pattern of malignant melanoma: a study of 216 autopsy cases.Am JSurg 1978; 135: 807-10. 4. Mossiman F, Fontolliet C, Genton A, et al. Resection of metastases to the alirnentarvtract from malignant melanoma.lnt Surg 1982; 67: 257-60. 5. Coit D. Role of surgery for metastatic malignant melanoma: a review. Semin Surg Oncol 1993; 9: 239-45. 6. Wong J, Skinner K, Kim K, et al. The role of surgery in the treatment of nonregionally recurrent melanoma. Surgery 1993; 113: 389-94. 7. Reintgen D, Thompson W, Garbutt J, et al. Radiologic, endoscopic; and surgical considerations of melanoma metastatic tb the gastrointestinal tract. Surgery 1984; 95: 635-9. 8. Wornom IL, Soong SJ, Urist MM, et el. Surgery as palliative treatment for distant metastases of melanoma. Ann Surg 1986; 204: 181-5. 9. Overett TK, Shiu ~MH. Surgical treatment of distant metastatic melanoma: indications and results. Cancer 1985; 56: 1222-30. 10. Lejeune FJ, Lienard D, Sales F. et al. Surgical management of distant melanoma metastases. Semin Surg Onco/1992; 8: 381-91. 11. Caputy GG, Donohue JH, Goellner JR, et al. Metastatic melanoma of the gastrointestinal tract: results of surgical management. Arch Surg 1991; 126: 1353-8. 12. Ihde JK, Coit DG. Melanoma metastatic to stomach, small bowel, or colon. Am J Surg 1991; 162: 208-11. 13. Khadra MH, Thompson JF, Milton GW, et al. The justification for surgical treatment of metastatic melanoma of the gastrointestinal tract. Surg Gynecol Obstet 1990; 171: 413-16. 14. Klaase JM, Kroon BBR. Surgery for metastatic melanoma to the gastrointestinal tract. Br J Surg 1990; 77: 60-1. 15. Goodman PL, Karakousis CPo Symptomatic gastrointestinal metastases from malignant melanoma. Cancer 1981; 48: 1058-9.

©1995 Blackwell Science Ltd, Surgical Oncology, 4: 105-110

109

16. Jorge E, Harvey HA, Simmonds MA, et al. Symptomatic malignant melanoma of the. gastro· intestinal tract. Ann Surg 1986; 199: 328-31. 17. Gutman M, Klausner JM, Inbar M, et al. Surgical approach to malignant melanoma in the gastrointestinal tract. J Surg Onco/1987; 36: 17-20. 18. Giler S, Kott I, Urea I. Malignant melanoma metastatic to the gastrointestinal tract. World J Surg 1979; 3: 375-9. 19. American Joint Committee on Cancer. Manual for Staging of Cancer, 2nd edn. New York, NY: JB Lippincot ce., 1983: 117-21. 20. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457-81. 21. Peto R, Peto J. Asymptomatically efficient rank invariant test procedures. J R Stat Soc 1972; 135: 185-98. 22. Oddson TA, Rice RP, Seigler HF, et al. The spectrum of small bowel melanoma. Gastrointest Radiol 1978; 3: 419-23. 23. Kadakia SC, Parker A, Canales L. Metastatic tumors to the upper gastrointestinal tract: endoscopic experience. Am J Gastroentero/1992; 87: 1418-23. 24. Kawashima A, Fishman EK, Kuhlman JE, et al. CT of malignant melanoma: patterns of small bowel and mesenteric involvement. J Comput Assist Tomog 1991; 15: 570-4. 25. Calabro A, Singletary E, Balch CM. Patterns of relapse in 1001 consecutive patients with melanoma nodal metastases. Arch Surg 1989; 124: 1051-5. 26. Roses DF, Karp NS, Oratz R, et al. Survival with regional and distant metastases from cutaneous malignant melanoma. Surg Gynecol Obstet 1991; 172: 262-8. 27. Silverman JM and Hamlin JA. Large melanoma metastases to the gastrointestinal tract. Gut 1989; 30: 1783-5. 28. Murphy MN, Lorimer SM, Glennon PE. Metastatic melanoma of the gallbladder: a case report and review of the literature. J Surg Oncol 1987; 34: 68-72. 29. Karakousis CPo Surgical treatment of recurrent malignant melanoma. Contemp Surg 1983; 22: 75-86. 30. Klausner JM, Skornick Y, Lelcuk S, et al. Acute complications of metastatic melanoma to the gastrointestinal tract. Br J Surg 1982; 69: 195-6. 31. Wilson BG, Anderson JR. Malignant melanoma involving the small bowel. Postgraduate Med J 1986; 62: 355-7. 32. Karakousis CP, Temple MD, Moore R, et al. Prognostic parameters in recurrent malignant melanoma. Cancer 1983; 52: 575-9. 33. Feun LG, Gutterman J, Burgess A, et al. The natural history of resectable metastatic melanoma (stage IV melanoma). Cancer 1982; 50: 1656-63. 34. Hena MA, Emrich LJ, Nambisan RN, et al. Effect of

N. Rieaniadis et al,

110

surgical treatment on stage IV melanoma. Am J Surg

1987; 153: 270-5. 35. Balch CM. Cutaneous melanoma: prognosis and treatment results worldwide. Semin Surg Oneol 1992;

8: 400-14. 36. Karakousis CP, Velez A, Driscoll DL, Takita H. Metastasectomy in malignant melanoma. Surgery 1991; 115: 295-302.

©1995 Blackwell Science Ltd, Surgical Oncology, 4: 105-110