Prognostic factors influencing survival in gastrointestinal leiomyosarcomas. Implications for surgical management and staging.

The appropriate surgical therapeutic options for either localized or more advanced disease in patients with gastrointestinal leiomyosarcomas remain unclear. A staging classification for this disease has not been adopted nor risk factors identifying patients at risk for recurrence defined. To address these issues, this study evaluated the influence of various clinicopathologic variables on overall and disease-free survival. In an univariate analysis of overall survival involving 191 patients, the Cox proportional hazards model identified four factors that were associated with a significantly better outcome: complete resection without tumor rupture (p less than 0.001), localized lesions (p less than 0.001), low grade of tumor (p = 0.02), and tumors smaller than 5 cm (p = 0.03). When interactive effects of these factors were taken into account, however, type of resection of the tumor was selected as the only significant prognostic factor in a multivariate analysis. Complete resection without tumor rupture improved overall survival of patients with localized disease (median, 46 months) as well as those with contiguous organ invasion (median, 36 months) or peritoneal implants (median, 36 months). In contrast, patients with incomplete resections survived for a median of 21 months. Patients with tumor rupture, despite removal of all gross disease, behaved similarly to those with incomplete resections; median survival was only 17 months. For disease-free survival, important determinants selected from a multivariate analysis were tumor rupture (p = 0.002), contiguous organ invasion (p = 0.02) and high tumor grade (p = 0.02). A staging classification incorporating these prognostic factors of significance was evaluated using a TGM system: T1 (less than 5 cm), T2 (greater than or equal to 5 cm), T3 (contiguous organ invasion or peritoneal implants), T4 (tumor rupture); G: G1 (low grade), G2 (high grade); M: M0 (no metastases), M1 (metastases present). The corresponding 5-year overall survivals for stages I, II, III, IVA, and IVB were 75%, 52%, 28%, 12%, and 7%. Disease-free survival at 2 years after surgery was 89%, 57%, and 47% for stages I, II, and III, respectively. In conclusion, surgery remains the primary modality of treatment for patients with gastrointestinal leiomyosarcomas, and complete resection of all disease without tumor rupture, even of locally advanced disease, improves overall and disease-free survival. A staging classification appears feasible and is recommended to determine outcome in patients with leiomyosarcomas arising from the gastrointestinal tract.