FlG. 1. Biotin-containing intranuclear inclusions in a borderline clear cell adenofibroma of the ovary. (A, hematoxylin and eosin, and B, avidin-biotin peroxidase.) tranuclear inclusions in tumors of the endometrium and the ovaries. Similarly to Yokoyama's study, the inclusions had a eosinophilic pale ground-glass appear-ance and were surrounded by a thin rim of chromatin (Fig. 1) . In our series, the inclusions were infrequent, but they were found occasionally in tumors with squamous and clear

more »

... cell differentiation. In fact, the presence of these inclusions in ovarian tumors had been reported previously by Tsujimoto and colleagues. 3 The most interesting point about the presence of these inclusions in tumors is that they can pose problems in immunohistochemical interpretation, particularly with regard to nuclear antigens, such as PCNA or p53. We think that it is important to keep in mind these biotincontaining intranuclear inclusions to avoid overestimation of immunostaining for these nuclear antigens in tumor cells. REFERENCES 1. Yokoyama S, Kashima K, Inoue S, et al. Biotin-containing intranuclear inclusions in endometrial glands during gestation and puerperium. Am J Clin Pathol 1993;99:13-17. 2. Mazur MT, Hendrickson MR, Kempson RL. Optically clear nuclei: An alteration of endometrial epithelium in the presence of trophoblast. Am J Surg Pathol 1993;7:415-423. 3. Tsujimoto M, Okano K, Taki I, et al. Biotin-containing intranuclear inclusions demonstrated in three cases of endometrioid adenocarcinoma of the ovary. Transacliones Societatis Pathological' J aponicae 1989;73:157 (in Japanese). Abstr. The Authors' Reply We completely agree with Matias-Guiu and colleagues. Every pathologist has to be careful when estimating various nuclear antigens by the avidin-biotin peroxidase (ABC) method. Fortunately, however, biotin-containing intranuclear inclusions (BCIIs) are quite rare, although I expect that these intranuclear inclusions will be detected in other neoplastic or non-neoplastic conditions besides already known ones. To my knowledge, BCIIs are known to appear in endometrial glands during gestation and puerperium, 1 or with trophoblastic disease, 2 endometrioid adenocarcinoma of the ovary, 3 papillary adenocarcinoma of the thyroid, 4 ' 5 and pulmonary endodermal tumor resembling fetal ling (PET). 6 Nakatani and colleagues called them "optically clear nucleus (OCN) family" and raised a question what is the common denominator between the different conditions. 6 In these conditions, BCIIs showed the same morphological characteristics and immunoreactivities. Although no ultrastructural finding was described in our study, 1 we have made sure that BCIIs are ultrastructurally occupied by filaments reacting with antibody against biotin by immunogold method (Fig. 1) , which was not published. In the endometrium during pregnancy and endometrioid adenocarcinoma of the ovary BCIIs were present in the glandular epithelia, 12 whereas they were in morules of PET 6 or in squamoid structures of thyroid cancer. 4 The squamoid FIG. 1. Gold particles are observed over intranuclear filamentous area corresponding to biotin-containing intranuclear inclusions. structures in the thyroid cancer probably should be called morules, because my coworker has studied BCIIs seen in thyroid cancer, where they were found only in morule but not in squamous metaplasia (personal communication). In his study, morule was different from squamous metaplasia morphologically and immunohistochemically. Squamous and clear differentiation described by Matias-Guiu might be morule. First, to clarify the mechanism for the unique nuclear change, an analysis of the intranuclear filaments should be performed and then the common denominator between the different conditions should be found out.