Metastatic Pleomorphic Lobular Carcinoma of the Breast to the Urinary Bladder: A Report of 10 Cases and Assessment of TRPS1 in the Differential Diagnosis With Plasmacytoid Urothelial Carcinoma

We performed a natural language search including the words “breast,” “metastatic,” and “bladder” at the laboratory information systems of the University of Miami Miller School of Medicine, Miami, Florida; The Johns Hopkins Medical Institutions, Baltimore, Maryland; and the Virginia Piper Cancer Institute, Minneapolis, Minnesota. Identified cases were then centralized for review. We defined pleomorphic lobular carcinoma according to the definition provided in the World Health Organization Classification of Tumours, 5^th edition, Breast Tumours,²¹  that restricts this diagnosis to carcinomas demonstrating a higher degree of cytologic atypia (ie, larger cells with significant nuclear atypia and nuclear size exceeding 4 times the size of a lymphocyte) and more prominent mitotic activity exceeding that of a conventional invasive lobular carcinoma of the breast. In addition to 2 senior surgical pathologists whose primary academic interest is uropathology (J.I.E. and O.N.K.), all metastatic tumors were reviewed by an experienced breast pathologist (B.S.) to attest that metastatic breast cancer could be classified as pleomorphic lobular carcinoma.

Clinical information about presentation, treatment, and follow-up was collected from patients’ medical records and from the accompanying information submitted by the referring pathologists.

Hematoxylin-eosin–stained slides were reviewed to describe the architectural patterns of growth and cytologic features. The following architectural patterns were included: cellular dyscohesion, single cells and cords of cells, and nests/sheets of neoplastic cells. Cytologic features that were assessed included the percentage of signet ring cells, cytoplasmic color, nuclear size relative to the size of a lymphocyte, and nuclear pleomorphism.

We performed immunostaining on each tumor with the panel of antibodies listed in Table 1. Immunostaining was performed on 4-μm–thick formalin-fixed, paraffin-embedded tissue sections on charged slides using a Leica Bond III automated system according to the manufacturer’s protocol following standard antigen retrieval. A 3,3′-diaminobenzidine chromogen was used as a chromogen. From consecutive in-house specimens of the senior author (O.N.K.), we identified 16 invasive high-grade urothelial carcinomas with plasmacytoid features and stained these tumors with keratin 5/6, p63, and TRPS1 immunostains because these 3 markers were not assessed in a prior immunohistochemical analysis of plasmacytoid urothelial carcinoma.¹⁹  The staining pattern of TRPS1 in breast cancer typically involves nuclear staining. Cytoplasmic and membranous staining was considered nonspecific immunoreactivity. Positive and negative controls were routinely performed for each tested antibody with each test run. When applicable, the pattern of staining in the internal control tissue was assessed.

We identified 10 specimens of metastatic pleomorphic lobular carcinoma to the bladder from 10 female patients ranging in age from 56 to 88 years (median age, 69 years). Of these 10 specimens, 2 were contributed by the Virginia Piper Cancer Institute, 1 specimen was contributed by the University of Miami, and 4 and 3 tumors were retrieved from the consultation practices of the 2 senior authors (J.I.E. and O.N.K., respectively). Past oncologic history was available in 6 patients. The time intervals between the diagnosis of breast cancer and the occurrence of bladder metastases ranged from 6 to 20 years (median, 15 years). Five patients had a history of invasive breast cancer. One patient with a history of high-grade ductal carcinoma in situ 19 years before the identification of bladder metastasis (2003) had liver metastasis reported as metastatic breast carcinoma in 2009, and her latest bladder biopsy (2022) was classified by us as metastatic pleomorphic lobular carcinoma. A total of 6 patients presented with hematuria, and 2 had hydronephrosis. Three patients had a record of metastases to other body sites. Cystoscopy revealed a mucosal mass in 3 patients; in 2 patients, outside urology reports described findings as thickened bladder wall, and 1 patient was reported to have a friable and bleeding mucosa. A total of 6 patients underwent bladder biopsy, and 4 had transurethral resection of bladder tumor.

All metastatic tumors exhibited the characteristic morphology of pleomorphic lobular carcinoma with pleomorphic nuclei (at least 4 times the size of a lymphocyte), many of which were eccentrically located in cells with voluminous eosinophilic cytoplasm (Figure 1). Microscopic patterns included single cells and cords of cells (n = 4), nests/sheets of dyscohesive cells (n = 2), or both (n = 4); see Table 2. In 7 of 10 tumors, we observed cells with a signet ring morphology (Figure 1, C and D). The percent of signet ring cells varied from 10% to 90% of the neoplastic cell population. Targetoid appearance of signet ring cells was noted in 4 tumors (Figure 1, D). All metastases involved the lamina propria. Muscularis propria was not sampled in any of the specimens, including transurethral resections. No surface colonization was seen.

The results of immunohistochemical analysis of pleomorphic lobular carcinoma are presented in Table 3 (Figures 2 and 3). GATA3 was positive in 9 of 10 tumors and was expressed in urothelium in all specimens. Estrogen, progesterone, and androgen receptors were positive in 80%, 43%, and 40% of tumors, respectively. Immunoreactivity for gross cystic disease fluid protein 15 (GCDFP-15) was present in 70% of tumors. High molecular weight keratin (HMWK) was positive in 88% (7 of 8) of the tumors, and keratin 5/6 (n = 9) and p63 (n = 10) were negative in all stained specimens. TRPS1 was positive in 70% of the tumors. During the original workup, mammaglobin analysis was performed in 7 metastatic tumors and all were positive, and in 4 tumors, the original diagnosis was assisted by demonstration of a loss of E-cadherin.

Sixteen cases of invasive high-grade urothelial carcinoma with plasmacytoid features were identified as controls, including 14 transurethral bladder tumor resections, 1 metastatic tumor, and 1 autopsy specimen. Of the 16 cases, 8 were pure plasmacytoid urothelial carcinomas (Figure 4), 1 carcinoma had both micropapillary and plasmacytoid subtypes, 1 predominantly plasmacytoid urothelial carcinoma had focal keratinizing squamous differentiation, and 6 conventional urothelial carcinomas had variable proportions of a plasmacytoid subtype.

A total of 9 of 16 plasmacytoid urothelial carcinomas (56%) were positive for p63 (Figure 5). However, among tumors that were positive, staining was focal (Figure 5, A). Fourteen tumors were assessed with keratin 5/6, and all were negative. TRPS1 showed strong diffuse staining of nuclei of 1 tumor and faint staining of isolated nuclei in 2 tumors (Figure 5, C and D). Of the remaining 12 plasmacytoid urothelial carcinomas, 11 were entirely negative for TRPS1. The plasmacytoid cancer with focal keratinizing squamous differentiation demonstrated a particularly interesting immunophenotype (Figure 6). Like other plasmacytoid cancers in this series, it was negative for keratin 5/6, whereas the area of squamous differentiation was diffusely positive. p63 was faintly immunoreactive in isolated plasmacytoid cancer cells and strongly stained each cell in the focus of keratinizing squamous differentiation. The plasmacytoid subtype was strongly and diffusely immunoreactive with TRPS1, whereas the focus of keratinizing squamous differentiation was negative for this marker.

Pleomorphic lobular carcinoma is rare and accounts for less than 1% of all epithelial malignancies of the breast and approximately 10% of all lobular carcinomas of the breast. It is associated with older age and postmenopausal status.^22,23  Compared with classic lobular carcinoma, pleomorphic lobular carcinoma usually features more worrisome gross and histologic findings, and it often demonstrates more aggressive clinical behavior.²⁴  Histologically, the tumor is composed of dyscohesive single cells with eccentric nuclei and voluminous eosinophilic cytoplasm. Signet ring cells are frequently present. This microscopic picture is remarkably reminiscent of plasmacytoid urothelial carcinoma.²⁰  In its pure form (ie, not being a component of otherwise conventional urothelial carcinoma or urothelial carcinoma demonstrating divergent differentiation), plasmacytoid urothelial carcinoma is an uncommon variant of invasive urothelial carcinoma, characterized by dyscohesive infiltrating cells resembling plasma cells, and it is notorious for its aggressive clinical behavior.²⁰  A histologic distinction is that lobular carcinoma signet ring cells often demonstrate targetoid inclusions that are generally not seen in plasmacytoid urothelial cancer. These inclusions may be helpful but are not specific for pleomorphic lobular carcinoma, because they have also been described in gastric diffuse signet ring cell carcinoma.²⁵  Both pleomorphic lobular carcinoma and plasmacytoid urothelial carcinoma lack extracellular mucin, in contrast to primary signet ring cell adenocarcinoma of the bladder or urachal signet ring cell adenocarcinoma, both of which commonly demonstrate signet ring cells floating in pools of mucin.

In carcinomas metastatic to the urinary bladder, a lack of conventional urothelial carcinoma with an absence of a surface component or tumor invading the luminal surface from the deeper layers of bladder wall should trigger a suspicion of secondary bladder involvement.²⁶  Unfortunately, these findings are not always evident, particularly in small biopsy specimens. In addition, pure plasmacytoid urothelial carcinoma also often lacks a surface component and frequently invades the deep bladder wall layers.^19,20 

Similarly to pleomorphic lobular carcinoma, plasmacytoid urothelial cancer lacks expression of E-cadherin. With the acquisition of plasmacytoid phenotype, both tumors may demonstrate some degree of immunoreactivity with plasma cell markers (eg, CD38 or CD138).²⁷  Given the rarity of metastatic pleomorphic lobular carcinoma to the bladder, its patterns of metastasis and morphology have not been described previously. Here, we conducted such an analysis and assessed the utility of the recently described immunohistochemical marker TRPS1 in distinguishing metastatic pleomorphic lobular carcinoma from plasmacytoid urothelial cancer.¹⁷ 

Overall, the bladder is an uncommon site of metastases. Based on an extensive series of surgical specimens and postmortem studies, breast cancer metastases account for 2.5% of all metastases to the bladder.^26,28  Most breast cancer metastases were diagnosed by cystoscopy and biopsy, some cases were found incidentally at autopsy, and others were identified by imaging studies performed as part of breast cancer follow-up or for urinary symptoms.^28–37  Patients with metastatic breast cancer to the bladder may experience various symptoms, including dysuria, hematuria, changes in voiding, and flank pain. Asymptomatic cases are typically discovered by imaging studies during routine breast cancer follow-up.^28,31,33,38  Imaging signs of bladder involvement include bladder wall thickening, mass formation, and hydroureteronephrosis, all of which were also seen in some of our patients.^36,39–41  Given the late occurrence of pleomorphic lobular carcinoma metastases, there is usually a significant time lag between the primary breast cancer diagnosis and the detection of bladder metastasis. Consequently, the history of breast cancer is often not available when analyzing the bladder specimen. Isolated bladder metastases are less frequent, and concurrent metastases are often seen in bones, lung, and/or liver.⁴²  Metastatic plasmacytoid urothelial carcinoma primarily affects the lung, liver, adrenal glands, retroperitoneum, lymph nodes, and bone.⁴³  Given the overlap in the metastatic sites between pleomorphic lobular breast carcinoma and plasmacytoid urothelial carcinoma, radiologically it is difficult to distinguish between these 2 entities.

The significance of differentiating pleomorphic lobular carcinoma from plasmacytoid urothelial carcinoma is not trivial because the 2 diagnoses lead to divergent treatments. Plasmacytoid urothelial carcinoma is commonly treated with neoadjuvant chemotherapy followed by radical cystectomy.¹²  On the other hand, the treatment of metastatic pleomorphic lobular carcinoma depends on its biomarker profile and may include hormonal treatment, anti-Her2 agents, and targeted therapy (such as CDK4/6 inhibitors), whereas surgery may be the last palliative resource.⁹ 

Parker et al⁴⁴  described HMWK immunoreactivity in other types of breast cancer, rendering it not helpful in distinguishing between breast and urothelial origin. The lack of keratin 5/6 expression in plasmacytoid urothelial carcinoma in our series is not an unexpected finding because most invasive urothelial carcinomas are luminal type, exhibiting diffuse immunoreactivity with HMWK with absent or single-cell patchy staining with keratin 5/6 at the periphery of invasive cancer nests.⁴⁵  A previous study has reported 22% of plasmacytoid urothelial carcinomas were basal subtype with positive keratin 5/6 expression.⁴⁶ 

Another immunohistochemical marker that may be considered as an aid in differential diagnosis is p63. Although conventional urothelial carcinomas are generally positive for p63, Kossai et al⁴⁶  reported only 41% of plasmacytoid urothelial carcinomas were reactive with p63. This finding held true in our series. In addition to problems with lower sensitivity for keratin 5/6 and p63 in plasmacytoid urothelial carcinoma, there are also issues with their specificity. Patchy staining with keratin 5/6 may be observed in invasive lobular carcinoma of the breast,⁴⁷  and extremely rare cases of lobular carcinoma of the breast with hybrid myoepithelial and secretory (aka myosecretory) cell differentiation are immunoreactive with p63 and keratin 5/6 but negative for estrogen and progesterone receptors.⁴⁸ 

GATA3 is among the most sensitive breast markers. However, its utility to assess breast metastasis to bladder is limited because of overlapping expression with urothelium.⁴⁹  Although uroplakin II has been considered a specific but less sensitive urothelial marker, 1 study has reported uroplakin II expression in pleomorphic lobular carcinomas.⁵⁰  Thus, it appears that neither GATA3 nor uroplakin II can reliably separate pleomorphic lobular carcinoma from plasmacytoid urothelial carcinoma.^50,51 

Progesterone and androgen receptors similarly appear to be expressed in low proportions of both tumors.^52,53  Estrogen receptor immunoreactivity may be of slightly more value. Although the expression of SP-1 clone (Roche) was not observed in 45 plasmacytoid urothelial carcinomas tested before,¹⁹  other authors reported immunoreactivity with a different clone (ID5, Dako) in 20% of tumors.⁵⁴  Unlike the classic type of lobular carcinoma of the breast that is characterized by strong diffuse positivity for estrogen receptor, a subset of pleomorphic lobular carcinomas is negative, making estrogen receptor–negative pleomorphic lobular carcinoma a greater pitfall. GCDFP-15 is also reportedly unreliable in this differential diagnosis. As much as 50% of plasmacytoid urothelial carcinomas in women may be GCDFP-15 immunoreactive.¹⁹  Mammaglobin was reported to be negative in all (n = 45) plasmacytoid urothelial carcinomas.¹⁹  Although we did not perform analysis with mammaglobin in the scope of this study in metastatic pleomorphic lobular carcinoma, 7 metastases were stained at the time of original diagnosis, and all were positive for mammaglobin, rendering it useful in this differential diagnosis.

TRPS1 is a novel GATA transcription factor that is a critical activator of the mesenchymal-to-epithelial transition during embryonic development in bone, cartilage, and kidney.⁵⁵  In a study by Ai et al,¹⁷  91% (434 of 479; 90.6%) of breast cancers were immunoreactive with TRPS1. On the contrary, only 2 of 115 urothelial cancers (2%) demonstrated low-intensity staining. GATA3 sensitivity is not superior to that of TRPS1 in breast cancer. Yoon et al⁵⁶  demonstrated strong staining for TRPS1 in all tumors in a cohort of 315 triple-negative breast cancers, but GATA3 was positive in only 94% of tumors. Of 3 TRPS1-negative metastatic pleomorphic lobular carcinomas in our series, 2 demonstrated GATA3 immunoreactivity. One metastatic tumor was negative for both markers.⁵⁴  In the current study, we have demonstrated that a subset of plasmacytoid urothelial carcinomas may be immunoreactive with TRPS1, including strong nuclear immunoreactivity in some tumors. Our cytopathology group has recently described a high prevalence of TRPS1 expression in metastatic carcinomas of Müllerian origin and in a minor subset of metastatic pulmonary adenocarcinomas.⁵⁷ 

As with any retrospective analysis, this study was not devoid of limitations. We conducted a retrospective database search in institutions with focused expertise in breast and urologic surgical pathology, identifying cases that have already been diagnosed rather than assessing a spectrum of consecutive specimens. Moreover, the study cohort was enriched with consultation specimens of the 2 senior authors when obtaining detailed clinical and radiologic information may not have always been possible. Although intuitively it is exceptionally low, the current study design did not allow us to determine the incidence of metastatic pleomorphic lobular carcinoma in consecutive bladder specimens.

In summary, clinical, radiologic, and morphologic presentations of metastatic pleomorphic lobular carcinoma of the breast to urinary bladder share many similarities with the plasmacytoid urothelial carcinoma. Being alert to a patient’s oncologic history, any clinical notes raising the possibility of metastasis, as well as the morphologic indicators (absence of surface component and lack of conventional urothelial carcinoma component) should prompt immunohistochemical investigation to establish the tumor site of origin. Estrogen receptor (particularly clone SP-1 by Roche), mammaglobin, and p63 appear to be helpful in differentiating pleomorphic lobular carcinoma from plasmacytoid urothelial carcinoma, and immunoreactivity for keratin 5/6 may be useful for the distinction of a less frequently encountered basal type of plasmacytoid urothelial carcinoma because it is typically negative in pleomorphic lobular carcinoma. TRPS1 immunoreactivity, both positive and negative, should be interpreted in conjunction with other markers because strong staining or lack of staining can be observed in both tumors. Caution should be exercised when choosing and interpreting the immunohistochemical panel to differentiate these 2 tumors because their immunophenotypes greatly overlap and ramifications of misclassification are major.

The authors thank Dayani Rodriguez for her outstanding administrative support with the conduction of this study.