Article ID: 2024-0017-CR
Invasive urothelial carcinoma (UC) has diverse morphological presentations. Here, we describe the case of a Japanese woman aged in her early 60s with UC with unclear differentiation. The patient presented with distinct glandular differentiation and concurrent cystitis glandularis (CG) and intestinal metaplasia (IM) without a conventional UC component. Up to 2% of patients with bladder cancer develop adenocarcinoma. However, differentiating UC with glandular differentiation (UCg) from adenocarcinoma can be challenging. Although CG and IM are associated with adenocarcinoma, their presence does not necessarily imply that the comorbid cancer is adenocarcinoma. In this case, cytokeratin 7 (CK7) and CK5/6 positivity was assessed to establish the diagnosis of poorly differentiated UCg. A poorly differentiated pure UCg without conventional UC components has not yet been reported, which makes diagnosis extremely difficult. Moreover, because of the highly differentiated glandular structures within poorly differentiated UCs, the mechanism of tumorigenesis remains unclear. Further studies involving a larger case series should be conducted to elucidate the association between CG and IM and investigate the genetic background of these tumors, all of which would improve the accuracy of differentiation between poorly differentiated UC and adenocarcinoma.
Bladder cancer (BC) is characterized by a significant tendency for divergent differentiation and varying histologies, and approximately 2% of cases are adenocarcinomas.1 Urothelial carcinoma (UC) with glandular differentiation (UCg) is a type of glandular lesion in BC. Approximately 75% of BC cases are classified as pure UC, whereas the remaining 25% are histological variants. Pure UC has several variants, which include squamous, glandular, micropapillary, sarcomatoid, small cell, lymphoepithelioma-like, clear cell, plasmacytoid, and poorly differentiated UC.2,3 Glandular differentiation is the second most common type of divergent differentiation after squamous UC, affecting up to 18% of patients with invasive UC.3 The poorly differentiated UC subtype includes carcinomas that do not exhibit morphological features indicative of urothelial origin. Therefore, the presence of conventional UC components or UC in situ can confirm a poorly differentiated UC diagnosis.2
A Japanese woman aged in her early 60s presented with hematuria and dysuria at a clinic. Cystitis was suspected, and antibiotic treatment was initiated. However, her symptoms persisted. Urinary cytology found no malignant cells. However, cystoscopy revealed a nonpapillary broad-based tumor around the bladder trigone that extended from the internal urethral orifice (Fig. 1A). Magnetic resonance imaging and computed tomography revealed a projecting solid tumor (diameter: 30 mm) in the bladder cavity at the bladder neck, without invading the adjacent organs (e.g., uterus, ovary, and rectum) (Fig. 1B). The patient underwent initial transurethral resection (TUR), but the tumor could not be completely resected. Therefore, total cystectomy and hysterectomy with ileal conduit diversion were performed after 3 months. No rectal cancer was observed intraoperatively. No recurrent metastases were observed at 1 year postoperatively.
On cystoscopy before transurethral resection (TUR), a nonpapillary, broad-based tumor was detected (A). T2-weighted magnetic resonance imaging before cystectomy (2 months after TUR) revealed a protruding solid tumor in the bladder cavity at the bladder neck without invasion of the adjacent organs (white arrow) (B).
This investigation was conducted in accordance with the 1975 Declaration of Helsinki. The authors’ institutions do not require ethics committee approval for the publication of case reports. Informed consent was obtained from the patient.
Histological examination revealed an invasive carcinoma consisting of polymorphic or spindle-shaped tumor cells. These cells were growing and infiltrating the nests, sheets, and cords and exhibited massive patterns. Each tumor cell exhibited infiltrative growth, which varied with location and was mixed with abundant infiltrating neutrophils (Fig. 2A,B). No osteoclast-like giant cells were observed. The histology described here dominated most of the tumor. However, distinct glandular structures were interspersed in some regions (Fig. 2C). In some sections of the areas with glandular structures, the cytoplasm of the tumor cells tested positive for mucus on Alcian blue staining (Fig. 2C, inset). The glandular structures gradually lost definition but were frequently observed in poorly differentiated areas showing solid and sheet-like proliferation (Fig. 2D). Neither the papillary UC component nor UC in situ was detected. Cystitis glandularis (CG) and focal intestinal metaplasia (IM) with goblet cells were interspersed around the tumor (Fig. 2E,F). Total cystectomy sample analysis confirmed the TUR findings, and no papillary UC components or UC in situ were observed. Furthermore, the TUR and total cystectomy samples did not contain osteoclast-like giant cells. The tumor also did not exhibit a clear connection with the nonneoplastic urothelium. CG was observed, and the uterus showed no neoplastic changes.
In poorly differentiated areas, polymorphic or spindle-shaped tumor cells (white arrow) proliferated and infiltrated nests, sheets, and cords or exhibited massive patterns depending on the location (A: 200×; B: 400×). Neither the papillary urothelial carcinoma (UC) component nor UC in situ was detected. Distinct glandular structures were mixed in some areas (C: 200×). In this glandular structure, the mucus tested positive on Alcian blue staining of the cytoplasm based on the location (C, inset; 400×). The glandular structures gradually became indistinct and were frequently observed in poorly differentiated areas of solid and sheet-like proliferation (D: 200×). Cystitis glandularis (E: 200×) and focal intestinal metaplasia (F: 400×) with goblet cells (black arrow) were interspersed around the tumor. gl: glandular area; por: poorly differentiated area; Ca: cancer components.
Immunohistochemical studies using an automated Ventana immunostainer (Roche, Basel, Switzerland) revealed that the tumor cells were positive for cytokeratin 7 (CK7) in both the ductal and poorly differentiated areas (Fig. 3), partially positive for CK5/6 in poorly differentiated areas, and positive for CK20 and caudal-related homeobox gene type2 (CDX2) only in areas with distinct ductal structures (Fig. 3). Equivalent poorly differentiated areas were also slightly positive for p63. No tumor cells expressed GATA binding protein 3 (GATA3) or uroplakin III. Based on the clinical and pathological findings, particularly the immunohistochemistry results, we diagnosed the tumor as invasive UC, poorly differentiated with glandular differentiation, distinguishing it from poorly differentiated adenocarcinoma. The antibodies used and the immunohistochemical results obtained are summarized in Table 1.
Immunohistochemical studies revealed that the tumor cells were positive for CK7 in both ductal and poorly differentiated areas, positive for CK20 and CDX2 only in areas with distinct ductal structures, and partially positive for CK5/6 in poorly differentiated areas.
| Protein | Isotype (clone) | Source | Dilution | Tumor cells in this case | |
|---|---|---|---|---|---|
| Poorly differentiated area |
Glandular differentiated area |
||||
| Pan-keratin | Mouse monoclonal (AE1, AE3, PCK26) | Roche | Ready to use | Positive | Positive |
| Cytokeratin 7 | Mouse monoclonal (OV-TL 12/30) | DakoCytomation | ×200 | Positive | Positive |
| GATA 3 | Mouse monoclonal (HG3-31) | Santa Cruz Biotechnology | ×50 | Negative | Negative |
| Uroplakin III | Mouse monoclonal (AU1) | Nichirei Biosciences | Ready to use | Negative | Negative |
| p63 | Mouse monoclonal (DAK-p63) | DakoCytomation | ×50 | Partial positive | Negative |
| Cytokeratin 20 | Mouse monoclonal (AU1) | DakoCytomation | ×50 | Negative | Positive |
| CDX 2 | Mouse monoclonal (DAX-CDX2) | DakoCytomation | ×50 | Negative | Positive |
Unsurprisingly, the pathological evaluation of BC has revealed a wide variety of histopathological patterns.4 UCg substantially overlaps with primary enteric-type adenocarcinoma. These two types of neoplasia can be challenging to differentiate based on morphological and immunohistochemical approaches.2,5,6 In particular, in the present case, only part of the tumor had glandular structures; most of it was comprised of poorly differentiated cancer cells with an unclear structure. In the current case, the key diagnostic question was whether the tumor was a pure adenocarcinoma with dedifferentiation or poorly differentiated UC with glandular differentiation.
Using 20 genes known to be frequently mutated in either muscle-invasive urothelial cancers or colorectal adenocarcinoma, Maurer et al.7 classified gene mutations into urothelial genes (e.g., TERT, RB1, STAG2, KDM6A, CDKN1A, CDKN2A, and ARID1A) or colorectal genes (e.g., KRAS, SMAD4, PTEN, and APC), and they compared the frequencies of each mutation in UCg and bladder adenocarcinoma (BAC). In UCg, in addition to TP53 and PIK3CA alterations, modifications of the urothelial genes associated with muscle-invasive BC were highly prevalent. BAC is related to colorectal genes and some urothelial genes (e.g., ARID1A and RB1). Therefore, although urothelial genes rather than colorectal genes were predominantly involved in the development of UCg (urothelial vs colorectal, 53% vs 19%), both urothelial and colorectal genes were involved in BAC development (29% vs 44%).7
In the present case, CG and IM were observed around the tumor. Studies have also been conducted on the histological background of tumor development and the frequency of the histological types of tumors that develop. BAC reportedly arises in CG glands with IM.8 Smith et al.9 reported that 117 patients with cystic and glandular cystitis exhibited concomitant carcinoma. However, GC is not necessarily specifically involved in the mechanism of BAC development. The incidence rates were as follows9: overall, 33%; adenocarcinoma, less than 1%; squamous cell carcinoma, 3%; and UC, 29%. In 19 patients with IM, concomitant carcinoma was observed with the following incidence: total, 37%; adenocarcinoma, 10%; UC, 24%; and UCg, 5%. IM may be associated with simultaneous carcinoma diagnosis, but the study found no evidence that IM increases the risk of future malignancy.9 Using next-generation sequencing, Pires-Luis et al.10 found that in enteric-type BAC, TP53 (56%) was the most frequently mutated gene, followed by BRCA2 and KMT2B (33%), and then by NOTCH2, KDR, ARID1B, POLE, PTEN, and KRAS (28%). In IM and adenoma, APC and PRKDC (60%) were the most frequently mutated genes, followed by ROS1, ATM, and KMTD2 (50%). Therefore, no clear genetic similarities were indicated between BAC and IM.
The review by McKenney11 revealed that in rare cases, low- or high-grade dysplasia can occur in IM. In cases of high-grade dysplasia (i.e., nuclear swelling, more pronounced hyperchromasia, increased mitotic activity, and cell polarity loss), a subset may be associated with invasive carcinoma, which warrants long-term clinical follow-up. Therefore, a poorly differentiated carcinoma may be considered an adenocarcinoma when it is surrounded by or interspersed with highly atypical intestinal epithelial metaplasia. In the present case, CDX2-positive CG with IM coexisted with poorly differentiated carcinoma. Nevertheless, dysplasia-associated IM was not observed.
Interestingly, previous studies have found that noninvasive UC with glandular differentiation is more likely to progress to invasive lesions than de novo UC in situ, as well as to higher-grade invasive lesions, such as small-cell carcinoma, micropapillary carcinoma, and poorly differentiated UC.12,13 Furthermore, none of the patients in these studies developed pure invasive adenocarcinoma from carcinoma in situ (CIS) with glandular differentiation.12,13
In our study, immunohistochemical analysis revealed that our patient had diffuse positivity for CK7, partial positivity for CK5/6 and p63, and positivity for CD20 and CDX2 only in areas with distinct glandular ductal structures. Paner et al.14 reported the following positive immunohistochemical rates in 130 patients with UCs with or without divergent differentiation in the bladder: GATA3, 50%; uroplakin III, 20%; CK7, 80%; CK20, 55%; p63, 87%; and high molecular weight cytokeratin, 89%. In another study, 86.6% of high-grade (G3) UC tested positive for CK7.15 Moreover, the positivity rates for undifferentiated carcinoma, which may originate from poorly differentiated UC, were as follows14: GATA3, 28%; uroplakin III, 0%; CK7, 50%; CK20, 3%; p63, 50%; and high molecular weight cytokeratin, 49%. Therefore, high-grade UCs reduced staining for CD20, GATA3, and uroplakin III, whereas CK7 expression was maintained. Conversely, the immunohistochemical profile of adenocarcinoma arising from the urinary tract resembles that of enteric adenocarcinoma,2 which usually tests positive for CK20 and CDX2. In poorly differentiated colonic carcinoma, the tumor cells presented with the following immunoreactivity rates16: CDX2, 55%; CK20, 45%; and CK7, 21%. These data indicate that the staining property of CD20 was retained. CD20 and/or CDX2 may also test positive for parts of glandular structures in UCg.2 Therefore, the immunohistochemical results in the present case align with the characteristics of poorly differentiated UC with glandular differentiation.
This case is unique in terms of the mechanism of tumor development. It shows a poorly differentiated UC in which glandular structures are the only remaining highly differentiated components despite the absence of clear differentiation tendencies toward urothelial cells, which are otherwise observed in typical UC, and lacking CIS with glandular differentiation. No other reports on UC similar to this case have been published. In bladder cancer, attempts have been made to compare classification by tumor-cell phenotype with classification by global mRNA expression profiling. This analysis classifies the invasive UC into two major subtypes: urothelial (luminal) and basal type.2,17 The urothelial type has high expression levels of uroplakin and GATA3, whereas the basal type has low GATA3 expression.2,17 Therefore, in molecular profiling, the background of the present case may be a basal type, rather than urothelial type. However, the genetic background of this lesion was not explored, which limits the generalizability of this case report. It is unclear which type is more likely to show the histological findings observed in this case. Further large-scale studies involving a larger number of cases and genetic analysis should explore the association between CG and IM and pure poorly differentiated tumors with glandular differentiation.
In conclusion, this case demonstrates that a portion of poorly differentiated or dedifferentiated carcinoma without the conventional UC component can present with clear glandular differentiation. Furthermore, if a poorly differentiated carcinoma exhibits ductal structures and features CG and IM, clinicians should not automatically assume that the lesions are BAC. Conversely, if a lesion lacks ductal structures and papillary UC or CIS in the adjacent tissues, or if evidence of CG (particularly when accompanied by differentiation into the intestinal epithelium with atypia) exists, the possibility of BAC should be comprehensively considered before making a diagnosis of poorly differentiated UC.
The authors thank Ms. Kanako Kunisawa and Ms. Machi Okamoto (Department of Pathology, Kochi Medical School) for their technical assistance.
The authors have declared that no conflict of interest exists.
