A study was made of 117 patients who presented with gastrointestinal lymphoma. The occurrence was 48 in the stomach, 37 in the small intestine, 13 in the ileocecal region, two in the appendix and 11 in the large intestine. In six cases, multiple sites in the gastrointestinal tract were involved, but in five cases this appeared to be secondary to massive mesenteric or retroperitoneal lymph node disease. Using Rappaport's classification, diffuse histiocytic lymphoma was the most frequent histologic type and constituted 60% of the cases. Nodular lymphomas comprised 10% of the total, nodular poorly differentiated lymphocytic lymphoma forming the majority of this group. Ten of the lymphomas were undifferentiated, 5 of Burkitt's type and 5 non-Burkitt's type. Five were Mediterranean-type lymphomas associated with plasma cell infiltration of the adjacent mucosa, and only two cases of primary Hodgkin's disease were encountered. Two lymphomas could not be classified. Eight percent of the cases showed plasmacytoid changes and were classified as a distinct subgroup of the parent lymphoma rather than as examples of extramedullary plasmacytoma. Gastrointestinally lymphomas occurred most frequently during the fourth to seventh decades. However, nine lymphomas occurred in children younger than 16 years of age. In comparison to adults, the childhood lymphomas showed a number of notable differences with respect to sex distribution, site of involvement and histologic type. Information concerning the extent of the disease at the time of diagnosis was available in 75 cases. Of these, 49% of the lymphomas were confined to the affected viscus and 33% had associated regional lymph node involvement; the remaining 18% had mode widespread disease. In 44 patients information on the spread of disease was available and in 48% there was extra abdominal spread. Prognosis appeared to correlate best with the stage of the disease rather than the histologic type.

One hundred smooth muscle tumors arising in the gastrointestinal tract and retroperitoneum were reviewed in an attempt to define criteria for the diagnosis of leiomyosarcoma in these sites. On the basis of aggressive behavior, 56 of these neoplasms were diagnosed as leiomyosarcoma. Mitoses were found to be the most useful indicator of malignancy; all of the tumors with five or more mitoses/10 HPF behaved aggressively and smooth muscle tumors with this degree of mitotic activity should be diagnosed as leiomyosarcoma. A paucity of mitoses, however, is no assurance of benignity as nearly 40% of the leiomyosarcomas in this series had fewer than five mitoses/10 HPF. Tumor cell necrosis was closely associated with aggressive behavior even when mitoses were infrequent and it is doubtful that benign smooth muscle tumors develop extensive tumor cell necrosis. In the absence of the requisite number of mitoses or tumor necrosis, it is difficult to distinguish some leiomyosarcomas from leiomyomas, but tumor size, cellularity and cellular atypia may be helpful parameters when assessed together. The importance of these criteria in different anatomical sites is discussed. It is emphasized that the criteria for the diagnosis of leiomyosarcoma of the uterus do not apply to non-uterine smooth muscle tumors. The actuarial 2-year survival rate was as follows: gastric leiomyosarcoma, 40%; small intestinal leiomyosarcoma, 60%; and retroperitoneal leiomyosarcoma, 16%.