We report a case involving a 45-year-old man with a 12-year history of Wegener granulomatosis, who developed a carcinosarcoma of the urinary bladder after long-term cyclophosphamide therapy. Cyclophosphamide is well recognized as an etiologic agent for urothelial carcinoma of the urinary bladder. However, only 5 cases of carcinosarcoma of the urinary bladder following cyclophosphamide therapy have been reported. We used loss of heterozygosity studies and microsatellite markers to define the molecular basis of this rare neoplasm. These studies revealed evidence supporting a monoclonal origin for the 2 components of this tumor. We also demonstrated allelic loss of chromosome 9p. This loss associated with carcinosarcoma of the urinary bladder is in agreement with previous studies, suggesting a possible role for the tumor suppressor gene p16 in the pathogenesis of this tumor.

Cyclophosphamide therapy has greatly improved the survival of patients with Wegener granulomatosis. However, cyclophosphamide is also a known carcinogen, well documented to cause urothelial carcinoma of the urinary bladder.1 Although cyclophosphamide is clearly linked to the development of carcinoma of the urinary bladder, its role in the development of other malignant bladder tumors is less clear. A recent report documented the interesting association between cyclophosphamide therapy and bladder leiomyosarcoma.2 The authors of the article reported 2 cases and found a total of 7 additional cases of postcyclophosphamide leiomyosarcoma in the literature. Carcinosarcoma of the urinary bladder is an even rarer sequel to cyclophosphamide therapy; only 5 cases have been reported.3,4 

Here, we report a case of carcinosarcoma of the urinary bladder in a patient with Wegener granulomatosis who was given cyclophosphamide therapy. Loss of heterozygosity (LOH) analyses using microsatellite-flanking markers suggested a monoclonal origin for the 2 components of the tumor and provided evidence for allelic loss of chromosome 9p.

A 45-year-old man presented with hematuria and groin pain. He had been diagnosed 12 years earlier with Wegener granulomatosis and was treated with steroids, methotrexate, and cyclophosphamide. During an 8-year period, he received a cumulative dose of approximately 293 g of cyclophosphamide. At the current admission, cyclophosphamide-induced hemorrhagic cystitis was considered the cause of the hematuria, and cyclophosphamide was therefore discontinued. However, the hematuria persisted, and computerized tomography of the abdomen revealed a bladder mass, which was biopsied transurethrally. Results of computerized tomography and bone scans were negative for metastasis. A cystoprostatectomy with ileal conduit formation was performed. The patient was offered adjuvant radiotherapy but declined. In the 12 months since diagnosis, no evidence of recurrence or metastasis of the tumor has been detected.

The LOH analyses were performed using protocols published previously.5 All primer sequences are available from the Web (http://www.ncbi.nlm.nih.gov/genome/guide/human/) except the informative primer pair for the short arm of chromosome 17. This primer pair was designed to flank a (CA)n repeat from BAC clone AC109333 located at 17p13.2 close to the GP1Bα gene. The primers (sequences) for 17p are GT6250F (AGAACTAGGCACAGAGGACACAGTG) and GT6250R (GCCCACGGCATGTTAAAGTTT), which amplify an amplicon of 163 base pairs with 17 (CA)n repeats. Amplification conditions for all primers used involved 35 cycles at 57°C for annealing and 1.5mM magnesium chloride. Microsatellite-flanking markers were selected from all autosomal arms, except the short arms of the acrocentric chromosomes (Table 1). Amplification was performed on DNA extracted from nontumorous prostate (control) and the carcinomatous and sarcomatous components from formalin-fixed paraffin-embedded tissue. Two blocks composed exclusively of sarcomatous and carcinomatous elements were used for DNA extraction, obviating the need for microdissection. If the nontumor tissue was not heterozygous, a second and rarely a third primer pair were used so as that an informative (ie, heterozygous) microsatellite marker for all chromosome arms was obtained.

Table 1.

Microsatellite Marker Analysis to Detect Chromosomal Deletions and Microsatellite Instability

Microsatellite Marker Analysis to Detect Chromosomal Deletions and Microsatellite Instability
Microsatellite Marker Analysis to Detect Chromosomal Deletions and Microsatellite Instability

Gross, Histologic, and Immunohistochemical Findings

Histopathologic examination of the transurethral resection revealed a high-grade, poorly differentiated malignant neoplasm invading the bladder musculature with essentially only vimentin reactivity in the tumor cells. During subsequent radical cystoprostatectomy, a 9.0-cm polypoid tumor was found that occupied 80% of the bladder lumen, grossly invading the muscle wall and extending into the surrounding adipose tissue. Examination of sections revealed a biphasic tumor composed of well-differentiated squamous cell carcinoma, interfacing with poorly to well-differentiated malignant mesenchymal elements (Figure 1, A).

Figure 1.

Junction of carcinoma and sarcoma (hematoxylin-eosin [A], cytokeratin [B], vimentin [C]; original magnification ×60). Figure 2. Liposarcomatous differentiation. A, Well-differentiated liposarcoma (hematoxylin-eosin, original magnification ×24). B, Pleomorphic liposarcoma (hematoxylin-eosin, original magnification ×40). C, Dedifferentiated liposarcoma (hematoxylin-eosin, original magnification ×40). Figure 3. Rhabdomyosarcomatous differentiation (hematoxylin-eosin, original magnification ×40 [A]; hematoxylin-eosin, original magnification ×256 [B]; desmin, original magnification ×256 [C]).

Figure 1.

Junction of carcinoma and sarcoma (hematoxylin-eosin [A], cytokeratin [B], vimentin [C]; original magnification ×60). Figure 2. Liposarcomatous differentiation. A, Well-differentiated liposarcoma (hematoxylin-eosin, original magnification ×24). B, Pleomorphic liposarcoma (hematoxylin-eosin, original magnification ×40). C, Dedifferentiated liposarcoma (hematoxylin-eosin, original magnification ×40). Figure 3. Rhabdomyosarcomatous differentiation (hematoxylin-eosin, original magnification ×40 [A]; hematoxylin-eosin, original magnification ×256 [B]; desmin, original magnification ×256 [C]).

Close modal

The carcinoma was cytokeratin positive and vimentin negative (Figure 1, B and C). Urothelial carcinoma was not detected in multiple sections obtained from the tumor and the adjacent mucosa, although areas of squamous metaplasia were identified. The mesenchymal component was vimentin positive and cytokeratin negative (Figure 1, B and C). Extensive sampling of this component revealed foci resembling a well-differentiated liposarcoma and areas consistent with pleomorphic liposarcoma and dedifferentiated liposarcoma. The well-differentiated liposarcomatous component was composed of relatively mature adipocytes, with lipoblasts and atypical stromal cells present (Figure 2, A). The pleomorphic component contained lipoblasts and pleomorphic spindle and giant cells (Figure 2, B). The dedifferentiated component consisted of a high-grade sarcoma, which appeared nonlipogenic and cellular, with spindle cells and a fascicular pattern of growth (Figure 2, C). Desmin-positive pleomorphic rhabdomyosarcomatous differentiation was also focally detected (Figure 3, A through C). A diagnosis of carcinosarcoma of the urinary bladder was made; the sarcomatous elements had features of a mixed-type liposarcoma with focal rhabdomyosarcomatous differentiation.

Molecular Genetic Analysis

Both the sarcoma and carcinoma samples shared LOH for microsatellite markers on the short arms of chromosomes 9 and 17. They also showed microsatellite instability; they were identical for the same gain of a repeat at the D4S2639 marker on 4p. In addition, the carcinoma but not the sarcoma showed LOH for the long arm of chromosome 22 and the gain of an allele for the D2S362 microsatellite marker on the long arm of chromosome 2. No evidence of microsatellite instability or LOH was evident from the other markers.

A substantial percentage of patients treated with long-term cyclophosphamide develop urothelial carcinoma of the urinary bladder (5% at 10 years and 16% at 15 years).1 Cyclophosphamide's urotoxicity is thought to be mediated by its metabolite acrolein,1,6 which is excreted in the urine, making it especially toxic to the urinary bladder. In contrast to the relatively common occurrence of postcyclophosphamide urothelial carcinomas, there are only 5 previously reported cases of carcinosarcoma of the urinary bladder subsequent to cyclophosphamide therapy (Table 2).3,4 In those cases, however, the patients were being treated for malignancies (lymphoma or recurrent urothelial cancer). To the best of our knowledge, this is the first reported case in which a patient being treated with cyclophosphamide for a nonneoplastic disease (Wegener granulomatosis) has developed a carcinosarcoma of the bladder.

Table 2.

Cyclophosphamide-Induced Bladder Carcinosarcoma: Cases Reported to Date

Cyclophosphamide-Induced Bladder Carcinosarcoma: Cases Reported to Date
Cyclophosphamide-Induced Bladder Carcinosarcoma: Cases Reported to Date

Review of the data in Table 2 reveals that cyclophosphamide-related carcinosarcomas tend to occur in the elderly and to present at an advanced stage and are treated surgically. The prognosis is variably poor and may depend on the extent of resection. Carcinosarcoma of the urinary bladder, even outside the setting of cyclophosphamide therapy, is a rare tumor. A review of the literature revealed only approximately 90 reported cases to date. Liposarcomatous differentiation of the sarcomatous component, as seen in our case, is very rare. Only 2 such cases have been reported.7,8 

The histogenesis of carcinosarcoma is controversial. The divergence theory of a monoclonal origin for both components9 implies that the tumors diverge from a single multipotent common precursor. The convergence theory of multiclonal origin implies 2 stem cells, one epithelial and the other mesenchymal. This theory was originally proposed by Virchow and involved a “collision” between 2 independently developing tumors. Of these conflicting concepts, the monoclonal theory seems to be gaining ground in the published literature.9,10 Our results are consistent with a monoclonal origin for carcinosarcoma of the urinary bladder. The sarcoma and carcinoma in our case show a number of common genetic abnormalities, including LOH on 9p and 17p and microsatellite instability at the D4S2639 marker on 4p. These shared abnormalities suggest a monoclonal origin for these 2 components. The carcinoma shows additional abnormalities (divergence) in the form of LOH at 22q and microsatellite instability at D2S362 on 2q, presumably acquired during tumor progression.

Previous studies on carcinosarcoma of the urinary bladder have used comparative genomic hybridization and LOH to delineate the genetic events occurring in this tumor.10,11 Halachmi et al11 (6 cases) showed loss of 9p, 9q, 8p, and 8q in both components. Gronau et al10 (1 case) showed losses of 9p and 11q in both components. The short arm of chromosome 9 was also lost in our case. Thus, loss of 9p seems to be the most consistent chromosomal aberration in carcinosarcoma of the urinary bladder in the literature to date and may be the key genetic event in the pathogenesis of these tumors. As pointed out by Gronau et al,10 9p is the location of the tumor-suppressor gene p16. The genetic locus for p16 (9p21) is close to the location of the microsatellite we used for chromosome 9 in this study (9p24.1). Losses of 9p, including deletion of p16, have been previously reported in both components in carcinosarcomas of various sites and are an early finding in superficial urothelial carcinomas.12 Another notable feature in the present case is LOH at 17p (in both the carcinoma and the sarcoma), the location of the tumor suppressor gene p53. The p53 gene mutation may have played a role in the development of the tumor; however, loss of 17p was not noted in previous studies on the molecular genetics of carcinosarcoma of the urinary bladder.10,11 Therefore, in contrast to allelic loss of chromosome 9p, loss of chromosome 17p appears to be a finding peculiar to the present tumor rather than a consistent genetic event in urinary bladder carcinosarcomas.

Here, we present a case of carcinosarcoma of the urinary bladder developing in a patient with Wegener granulomatosis that had been given cyclophosphamide therapy. This case adds to the growing evidence that cyclophosphamide may be associated with bladder neoplasms other than urothelial carcinoma. We also present evidence for the monoclonal origin of carcinosarcoma using microsatellite-flanking markers and LOH analyses. The demonstrated allelic loss of chromosome 9p in carcinosarcoma of the urinary bladder suggests direction for further research on the role of the tumor suppressor gene p16 in the development of this tumor.

Talar-Williams
,
C.
,
Y. M.
Hijazi
, and
M. M.
Walther
.
et al
.
Cyclophosphamide-induced cystitis and bladder cancer in patients with Wegener granulomatosis.
Ann Intern Med
1996
.
124
:
477
484
.
Tanguay
,
C.
,
I.
Harvey
,
M.
Houde
,
J. R.
Srigley
, and
B.
Tetu
.
Leiomyosarcoma of the urinary bladder following cyclophosphamide therapy: report of two cases.
Mod Pathol
2003
.
16
:
512
514
.
Sigal
,
S. H.
,
J. E.
Tomaszewski
,
J. J.
Brooks
,
A.
Wein
, and
V. A.
LiVolsi
.
Carcinosarcoma of bladder following long-term cyclophosphamide therapy.
Arch Pathol Lab Med
1991
.
115
:
1049
1051
.
Lopez-Beltran
,
A.
,
A.
Pacelli
, and
H. J.
Rothenberg
.
et al
.
Carcinosarcoma and sarcomatoid carcinoma of the bladder: clinicopathologic study of 41 cases.
J Urol
1998
.
159
:
1497
1503
.
Shrimpton
,
A. E.
,
K. M.
Daly
, and
J. J.
Hoo
.
Mapping of a gene (MRXS9) for X-linked mental retardation, microcephaly and variably short stature to Xq12-q21.31.
Am J Med Genet
1999
.
84
:
293
299
.
Knight
,
A.
,
J.
Askling
, and
A.
Ekbom
.
Cancer incidence in a population-based cohort of patients with Wegener's granulomatosis.
Int J Cancer
2002
.
100
:
82
85
.
Baschinsky
,
D. Y.
,
J. H.
Chen
,
M. S.
Vadmal
,
J. G.
Lucas
,
R. R.
Bahnson
, and
T. H.
Niemann
.
Carcinosarcoma of the urinary bladder—an aggressive tumor with diverse histogenesis: a clinicopathologic study of 4 cases and review of the literature.
Arch Pathol Lab Med
2000
.
124
:
1172
1178
.
Perret
,
L.
,
P.
Chaubert
,
D.
Hessler
, and
L.
Guillou
.
Primary heterologous carcinosarcoma (metaplastic carcinoma) of the urinary bladder: a clinicopathologic, immunohistochemical, and ultrastructural analysis of eight cases and a review of the literature.
Cancer
1998
.
82
:
1535
1549
.
Thompson
,
L.
,
B.
Chang
, and
S. H.
Barsky
.
Monoclonal origin of malignant mixed tumors (carcinosarcomas).
Am J Surg Pathol
1996
.
20
:
277
285
.
Gronau
,
S.
,
C. K.
Menz
,
I.
Melzner
,
R.
Hautmann
,
P.
Moller
, and
T. F.
Barth
.
Immunohistomorphologic and molecular cytogenetic analysis of a carcinosarcoma of the urinary bladder.
Virchows Arch
2002
.
440
:
436
440
.
Halachmi
,
S.
,
A. M.
DeMarzo
, and
N. H.
Chow
.
et al
.
Genetic alterations in urinary bladder carcinosarcoma. Evidence of a common clonal origin.
Eur Urol
2000
.
37
:
350
357
.
Hartmann
,
A.
,
U.
Rosner
, and
G.
Schlake
.
et al
.
Clonality and genetic divergence in multifocal low-grade superficial urothelial carcinoma as determined by chromosome 9 and p53 deletion analysis.
Lab Invest
2000
.
80
:
709
718
.

Author notes

Corresponding author: Sanjay Mukhopadhyay, MD, Department of Pathology, State University of New York–Syracuse, Upstate Medical University, 750 E Adams St, Syracuse, NY 13210 (mukhopas@upstate.edu)