Detection of residual tumor cells in bladder biopsy specimens: Pitfalls in the interpretation of cytokeratin stains

Some patients who have had prior bladder biopsies or transurethral resections undergo a repeat resection within several months for various reasons. The detection of a few residual tumor cells in bladder specimens with prior biopsy site changes can be challenging based on histology alone. Immunohistochemistry for cytokeratins may be used as an adjunct in this situation. We have noted several cases in which keratin stains were performed and positive cells were noted, raising the issue as to whether the cytokeratin positive cells were residual tumor cells or stromal cells. Immunohistochemistry for a panel of antibodies [AE1/AE3, CAM 5.2, high molecular weight cytokeratin, smooth muscle actin (SMA), desmin, and anaplastic lymphoma kinase (ALK)] was performed on 29 cases of bladder biopsies with prior biopsy site changes. Of 29 patients, 25 had a prior history of bladder tumor: 17 had invasive high-grade urothelial carcinoma (T1, 5 cases; T2, 11 cases; T3,1 case); 7 had noninvasive high-grade papillary urothelial carcinoma; 1 had noninvasive low-grade papillary urothelial carcinoma). One of the patients with noninvasive high-grade papillary urothelial carcinoma and one of the patents with invasive high-grade urothelial carcinoma had associated carcinoma in-situ. Four patients had prior benign bladder diagnoses: cystitis cystica et glandularis; polypoid cystitis; follicular cystitis; and neurogenic bladder with benign prostate hyperplasia. Of the 29 cases, 6 (21%) had cells with staining for at least 2 of the cytokeratin markers. Cytokeratin (CK) AE1/ AE3 was positive for cells in 8/29 cases (28%). In 6 of these cases, cells displayed a spindle cell and 2 cases a more epithelioid morphology. CAM 5.2 was positive in cells in 5/29 cases (17%); 3 of the cases had spindle cell and 2 cases epithelioid morphology. High molecular weight cytokeratin was expressed in cells in 2/29 cases (7%) with 1 case having spindle cell and 1 epithelioid morphology. SMA was positive in cells with a spindle cell morphology and negative in the more epitheloid cytokeratin positive cells. Desmin was positive in 3/6 keratin positive spindle cells and negative in keratin positive epithelioid cells. ALK was negative in all the cases. Three cases with spindle cell morphology and positivity for at least 1 of the keratins and SMA stains were interpreted as aberrant keratin expression in myofibroblastic cells based on the staining and the morphology of the spindle cells. Another 3 cases with concurrent staining for at least 1 of the keratins, SMA and desmin were consistent with smooth muscle cells on the basis of their cellular morphology. Another 2 cases had cells, which expressed at least 2 CK markers but did not express SMA, desmin, or ALK and a more epithelioid morphology. These cells were interpreted as residual tumors cells. When interpreting CK stains for the detection of residual tumor cells, one should pay attention to the nature of the cells and not assume all CK staining cells are residual tumor cells.