Background Ovarian cancer is a spectrum of several histologically distinct tumor types which differ in etiology, response to therapy and prognosis. In a resource-limited settings, the diagnosis of ovarian cancer can be challenging. This study describes the distribution of ovarian cancer tumor types in East Africa as well as assessing the diagnostic accuracy by using contemporary methods. Methods Data from 210 women identified from the records with a diagnosis of ovarian cancer in a period of 15 years were included. Two tissue microarrays were constructed and stained with 20 antibodies relevant to ovarian cancer subtyping. An integrated diagnosis was reached by the review of full Haematoxylin and Eosin stained sections, with consideration of immunohistochemical results. The integrated diagnoses were compared with the original diagnoses, and the degree of agreement was evaluated by percentage and Kappa statistics. Results The estimated rates of ovarian cancer were much lower in East Africa compared to a North American population from Alberta, Canada. There was a higher proportion of sex cord stromal tumors and germ cell tumors in the East African population. Diagnostic accuracy for main ovarian tumor type categories was substantial (Kappa 0.70), but only fair for specific ovarian carcinoma histotypes (Kappa 0.34). Poor Haematoxylin and Eosin stain was the main factor hindering correct diagnosis, which was not related to tissue processing. Conclusions In a resource- limited setting, where immunohistochemistry is not routinely carried out, diagnostic accuracy for the main categories of ovarian carcinoma is substantial and could be further improved by standardization of the basic Haematoxylin and Eosin stain.

Abstract Background: Ovarian cancer is a spectrum of several histologically distinct tumor types that differ in etiology, response to therapy, and prognosis. In resource-limited settings, the diagnosis of ovarian cancer can be challenging. This study describes the distribution of ovarian cancer tumor types in East Africa as well as assessing the diagnostic accuracy by using contemporary methods. Methods: Data from 210 women identified from the records with a diagnosis of ovarian cancer in a period of 15 years were included. Two tissue microarrays were constructed and stained with 20 antibodies relevant to ovarian cancer subtyping. An integrated diagnosis was reached by the review of full Haematoxylin and Eosin stained sections, with consideration of immunohistochemical results. The integrated diagnoses were compared with the original diagnoses, and the degree of agreement was evaluated by percentage and Kappa statistics. Results: Though limited by selection bias, the results suggest lower rates of ovarian cancer in East Africa compared to a North American population from Alberta, Canada. There was a higher proportion of sex cord stromal tumors and germ cell tumors in the East African population. Diagnostic accuracy for main ovarian tumor type categories was substantial (Kappa 0.70), but only fair for specific ovarian carcinoma histotypes (Kappa 0.34). Poor Haematoxylin and Eosin stain was the main factor hindering the correct diagnosis, which was not related to tissue processing. Conclusions: In a resource-limited setting, where immunohistochemistry is not routinely carried out, diagnostic accuracy for the main categories of ovarian carcinoma is substantial and could be further improved by standardization of the basic Haematoxylin and Eosin stain.