INTRODUCTION

Inflammatory pseudotumor (IP) or myofibroblastic tumor was first recognized in the lung. Presently, extrapulmonary examples of this tumor have been described in virtually all major organs. IP was initially considered nonneoplastic, representing an aberrant inflammatory response despite its gross and microscopic features of a spindle cell neoplasm. Some cases expressed viral genomes (1) and many were accompanied by the constitutional symptoms of an inflammatory process (fever, anemia, weight loss), which resolve after surgical resection. Tumor necrosis factor α may promote fibroblast proliferation and the constitutional symptoms characteristic of most IP (2). However, the hypothesis that IP is reactive in nature with a pathogenesis related to cytokine release applies more readily to cases in the lung than in extrapulmonary sites of involvement. Several pathologic aspects of IP are neoplastic and unlike a purely inflammatory process: it has the potential for local recurrence; multifocal, noncontiguous tumors may develop; local growth is infiltrative; vascular invasion is possible; and five IP at various sites—lung, soft tissue, and bone—were clonal (3, 4) and had aberrant karyotypes (3). From a prognostic perspective, therefore, one can speculate that IP is a soft tissue-mesenchymal tumor of indeterminate or low malignant potential.

The histologic diagnosis of IP of the bladder is difficult owing to its relative rarity and a spectrum of mimics, including leiomyoma and leiomyosarcoma. It is diagnosed in the absence of prior surgery (5) and is thought to have female and young adult preponderances (6). The term postoperative spindle cell nodule (PSCN) of the bladder has been applied to those lesions histologically similar to IP that develop within 3 months of a surgical procedure (7). Distinction of these entities from low-grade inflammatory leiomyosarcoma or fibrosarcoma is particularly vexing, and some pathologists have questioned the existence of IP in the bladder. To our knowledge 81 cases of IP have been described in the bladder, of which 26 included patient follow-up (8, 9, 10, 11, 12, 13, 14). The clinicopathologic criteria for separating IP and PSCN from leiomyosarcoma deserve further exploration. Further, most reported cases of IP have been completely excised, so the outcome after partial resection is not known. Consequently, the untreated natural history of IP of the bladder is uncertain. Our report includes the first follow-up of an incompletely resected IP in which further treatment was precluded.

PATIENTS AND METHODS

We searched the in-house and consultation files of Mayo Clinic, 1928–1998; Christie Hospital, 1982–1999; Vancouver Hospital, 1997–1999; and Bostwick Laboratories, 1999, for all spindle cell bladder tumors; two cases were added from Reading (PA) Hospital and Gainesville (FL) VAMC. All cases but one (12) were not previously reported. The inclusion criterion for in-house cases was the observation by DGB, AGN, JHS, EJ, or LC of focal findings suggestive of IP/PSCN, or mention of those entities in the differential diagnosis (regardless of final diagnosis). The inclusion criterion for outside cases was that the referring pathologist considered the diagnosis of IP or PSCN. Three patients with sarcoma were excluded because the sarcoma did not originate from the bladder, and two because the tumor was indeterminate owing to poor tissue preservation. Remaining were 38 patients, only one of whom was previously reported (12). Patients ranged in age from 2.5 months to 87 years; 28 were male and 9 female. Histologic data were tabulated and distinguished IP and PSCN from sarcoma (see Results). PSCN, rather than IP, was diagnosed if histologically benign tumor arose at any interval after prior bladder instrumentation or surgery. Slides were reviewed by DGB, VG, and KAI to confirm all diagnoses. Spindle cell carcinoma was diagnosed only if associated with a definite invasive or in situ urothelial carcinoma component, regardless of immunostaining. The 1997 TNM system was used for T staging (15); any case without evaluable muscularis propria was stage “indeterminate.” Counts were made of mitotic figures per 10 high-power (0.6 mm diameter) fields (hpf) and inflammatory cells per hpf. Immunohistochemical stains were performed in most cases, using the automated LSAB/Ventana ES (Tucson, AZ), as follows: Keratin (AE1/AE3) Boehringer Mannheim, Indianapolis, IN, primary dilution 1/400; EMA (E29), DAKO, Carpenteria, CA, 1/100; Vimentin (3B4), DAKO, 1/500; α-smooth muscle actin (1A4), DAKO, 1/150; Muscle-specific actin (HHF35), DAKO, 1/50; Desmin (Der 11), DAKO, 1/100, p53 protein (Ki67), DAKO, 1/100. Staining was evaluated as to intensity (0 to 3+) and percent of counted cells positive (0 to 100%). Follow-up was obtained up to March 2001 for 36 of 38 patients, by corresponding with either the patient at home or the patient's surgeon.

RESULTS

Presentation and Demographics

Presenting symptoms were known for 36 patients, and included hematuria in 27, obstruction in 11, and bladder calculi in 2 (Table 1). Two of the patients with PSCN were asymptomatic. After review of all cases and immunohistochemical stains in 30, 22 were diagnosed as benign (16 IP, 5 PSCN, 1 leiomyoma); 13 were diagnosed as sarcoma (7 low-grade; 6 high-grade), and 3 as carcinoma. All 20 patients with IP, PSCN, or leiomyoma were initially diagnosed correctly. Four of 13 patients with sarcoma had been misdiagnosed by referring institutions: three with IP and one with carcinosarcoma.

TABLE 1 Clinical Data in 38 Patients with Spindle Cell Tumor of Bladder; 21 with Bladder Inflammatory Pseudotumor/Postoperative Spindle Cell Nodule, and 17 with Other Spindle Cell Tumors
Full size table

Mean patient age was 38 years (15 to 74) for IP, 65 (49 to 83) for PSCN, 51 (0.2 to 87) for sarcoma, and 76 (69 to 83) for carcinoma/carcinosarcoma. Eleven IP patients were men and six were women. Possible predisposing conditions included neurogenic bladder secondary to cerebrovascular accident in Patient 2 with IP. All patients with PSCN had undergone bladder biopsy/transurethral resection, up to 13 months (Case 21) preoperatively. No sarcoma patients had prior cytoscopy; Patient 30 with sarcoma had posterior uretral valves, and Patient 35 had undergone radiotherapy for rectal carcinoma. Both synchronous and metachronous urothelial carcinoma occurred in one patient with IP (Patient 14); prior or subsequent urothelial carcinoma or papilloma occurred in all four patients with PSCN, and none with sarcoma. No patient with sarcoma, and two of three with carcinoma/carcinosarcoma had prior instrumentation.

Tumor size of IP averaged 4.4 ± 0.7 cm (range 2.5 to 13.0), PSCN was 1.5 ± 0.6 cm (0.4 to 3.0), and sarcoma was 4.0 ± 0.6 cm (0.5 to 7.0). Tumor distribution by site within the bladder was nearly identical for 15 benign and 13 malignant tumors in which it was known. Distribution of benign and malignant cases, respectively, was on side walls, 5 and 8; posterior 4 and 4; anterior 2 and 3; dome 3 and 3; trigone 1 and 1.

Treatment and Follow-Up

Eight of 22 patients with benign tumor underwent partial cystectomy; 14 had transurethral resection only. Four of 13 patients with sarcoma underwent partial cystectomy, 5 radical cystectomy, 3 radiotherapy, and 1 no therapy after transurethral resection. Of three patients with carcinoma/carcinosarcoma, one had partial cystectomy, one radical cystectomy, and one had no further treatment.

Follow-up was available for 36 patients. Among 12 sarcoma patients, 2 had tumor recurrence after partial cystectomy (2 and 26 months) and 2 others died of metastatic tumor (at 3 months: 1 low-grade, stage T3, 1 high-grade, stage indeterminate). A man with sarcoma as an infant (Table 1, Patient 25) underwent cystectomy at age 2, and at age 31 developed a renal abscess, and needed a renal transplant. A 74-year-old man with IP (Patient 3) was not a candidate for definitive tumor ablation after transurethral resection, because of other medical problems (diabetes and hypertension). He experienced an increase in tumor size from 13.0 to 37.5 cm in 5 months, was not further treated, and died of urosepsis. Another patient (Patient 7) with a transurethrally resected bulky 7-cm IP had a histologically identical 3 cm tumor, 5 months postoperatively. The surgeon considered this as residual tumor that was initially incompletely resected, rather than as recurrence (E. Martin, M.D., personal communication). No PSCN recurred. One of three spindle cell carcinoma/carcinosarcoma patients died.

Pathologic Findings

IP was T stage T1 in 3 cases, T2 in 12 cases, and indeterminate in 1. Stage T2 IP, unlike sarcoma, lacked necrosis or myxoid degeneration at the tumor-muscularis propria interface (Fig. 1). Microscopic pattern was biphasic, as in nodular fasciitis. In superficial areas, there was a paucicellular spindle cell proliferation mixed with mostly chronic inflammation, featuring prominent plasma cells (Fig. 2). Only three cases had a sparse component of neutrophils. One case (Patient 8) had 20% eosinophils in the inflammatory infiltrate. Stroma was edematous, acid mucopolysaccharide-rich, with many slit-like blood vessels per low-power field (Fig. 2). The deeper areas of tumor had more densely cellular interlacing storiform fascicles of cells. Cells varied only slightly in size. Larger spindle cells had tapering eosinophilic cytoplasmic processes, normochromatic oval nuclei, and often, visible single nucleoli. Focal nuclear and nucleolar enlargement was attributable to reactive atypia. No cytoplasmic striations or cambium layer were seen, ruling out botryoid rhabdomyosarcoma. There were a mean of 153 chronic inflammatory cells per hpf (range 41 to 412). Mitotic rate was 0 to 1 per 10 hpf in all cases. Ulceration and cystitis were constant findings.

FIGURE 1
figure 1

Inflammatory pseudotumor in 28-year-old woman (Patient 4). Spindle cells without atypia are amid many stromal microvessels. Tumor extends into muscularis propria, without necrosis.

Full size image
FIGURE 2
figure 2

Inflammatory pseudotumor in 26-year-old man (Patient 7). Note spindle cells without atypia, loose matrix, and frequent plasma cells.

Full size image

PSCN was stage T2 in three cases and indeterminate in two. Histologically similar to IP, it differed only by the presence of foreign body giant cells and prominent eosinophils in two cases. Mitotic rate was 0 to 1 per 10 hpf, except in Case 18, which had 15 per 10 hpf. Inflammatory cells per single hpf averaged 140, and in all 4 cases, correlated with postoperative interval: 392 at 1 week but 11 at 13 months. Leiomyoma was like that of the uterus, with no inflammation.

In sarcoma, T2 tumors penetrated deeply into the muscularis (Fig. 3). Low-grade leiomyosarcoma (seven cases) was stage T2 in five cases, T3 in one and indeterminate in one. Spindle tumor cells had moderate nuclear hyperchromasia and atypia (Fig. 4). Mitotic rate ranged from 1 to 16 per 10 hpf; intratumoral variation was seen. Coagulative necrosis was seen in all seven cases, and in six T2 or higher cases at the tumor-muscularis propria junction. Myxoid degeneration was noted in two cases, focally with stromal clefts (Fig. 5). Microvasculature was much less prominent than in IP/PSCN. Admixed inflammatory infiltrate was patchy (mean 24 per hpf, range 8 to 44) with neutrophils predominating over chronic inflammation in six of seven cases.

FIGURE 3
figure 3

Low-grade sarcoma in 87-year-old man (Patient 29). All sarcomas featured necrosis of the tumor (top and left) as it infiltrates the muscularis propria.

Full size image
FIGURE 4
figure 4

Low-grade sarcoma (also Patient 29). Nucleomegaly with prominent or double nucleoli (center) were useful diagnostic findings for low-grade sarcoma. Also, neutrophils or eosinophils usually predominated in the inflammatory infiltrate, with plasma cells rarely seen.

Full size image
FIGURE 5
figure 5

Low-grade sarcoma (Patient 29). Myxoid degeneration was noted in 6 of 13 sarcomas; in this case, amphophilic material fills stromal clefts. Stroma in inflammatory pseudotumor was often edematous but not myxoid, and without dropout of cells.

Full size image

High-grade sarcoma (six cases) was stage T1 in two cases, T2 in one, and indeterminate in three. Growth was infiltrative, with necrosis and in four cases, myxoid degeneration. Cellularity was consistently high, with marked nuclear atypia. Mitotic rate was 0 to 3 per 10 hpf, including atypical mitotic figures. The density of inflammatory infiltrate averaged 101 per hpf (range 7 to 280), with acute inflammation predominating in 4 of 6 cases, concentrated at areas of surface ulceration.

Immunohistochemical results (Table 2) were available in 30 cases. p53 nuclear immunostaining was positive in 9 of 11 sarcomas. There was moderate to strong staining of up to 40% of cells in five low-grade cases and up to 70% of cells in four high-grade cases (Fig. 6). Two cases were negative; however, they came from 1928 and 1929, so these results may reflect lost immunogenicity. In PSCN and IP, p53 immunostaining was either weak (1+), staining 2 to 60% of nuclei (10 of 14 cases) (Fig. 6), or was absent (4 of 14 cases). In IP/PSCN, 7 of 18 cases were weakly reactive for pan-cytokeratin, as were 3 of 8 leiomyosarcomas. Cytokeratin-positive cases were also reactive with at least one mesenchymal marker and, as mentioned before, all had necrosis. Actins did not discriminate between benign and malignant tumor. α-smooth muscle actin was positive in 43% of sarcomas, 63% of IP, and intermediate percentages of PSCN and carcinoma. Three of 11 IP, 2 of 3 PSCN, and none of 8 sarcomas were reactive for desmin. In the IP with juxtamural urothelial carcinoma biopsied separately (Case 14), immunostains for cytokeratin were negative (unlike in the carcinoma) and positive for smooth muscle actin and vimentin.

TABLE 2 Immunohistochemical Reactivity in 30 Cases of Urinary Bladder Spindle Cell Tumor
Full size table
FIGURE 6
figure 6

Immunostaining for p53 showed nuclear reactivity in only rare cells in 9 of 15 inflammatory pseudotumors/postoperative spindle cell nodules (left). A constant feature in sarcoma was more intense nuclear reactivity involving 20–70% of cells (right).

Full size image

Electron microscopy was performed in seven IP cases with available tissue, including the largest IP (Case 3)—13 cm, clinically reported as growing to 37.5 cm—and a PSCN with elevated mitotic rate (Case 18). Findings were similar in IP and PSCN; carcinoma was definitely ruled out in all. Demonstrated features are myofibroblastic (Fig. 7) as opposed to smooth muscle (16): namely, presence of fibronexus junctions (extracellular fibronectin fibrils, intracellular actin), no true lamina external to the plasma membrane, abundant rough endoplasmic reticulum, and rare myofilaments and plasmalemmal caveolae.

FIGURE 7
figure 7

Inflammatory pseudotumor (Patient 16). Tumor cell showing myofibroblastic features of rough endoplasmic reticulum, peripheral smooth muscle myofilaments (long arrows), and fibronectin fibrils (short arrows) (× 1600). (Micrograph courtesy of Dr. Alan Curry, Withington Hospital, Manchester, UK.)

Full size image

Overall clinical and pathologic findings for each entity are summarized (Table 3).

TABLE 3 Findings that Distinguish Benign Spindle Cell Tumors of Bladder from Sarcoma
Full size table

DISCUSSION

Necrosis or myxoid degeneration was key to the diagnosis of sarcoma, seen in all T2 cases; myxoid degeneration was previously noted in 60% of bladder leiomyosarcomas (17). Sarcoma also differed from IP by having moderate-severe nuclear atypia, 1 to 16 mitotic figures per 10 hpf (rather than 0 to 2, as described [6]), predominance of neutrophils over chronic inflammation, and metastatic potential.

Our data show that these histologic distinctions between IP and sarcoma correlated with different natural histories. Both were usually idiopathic, and except for one IP, were not associated with urothelial carcinoma. Both usually caused hematuria and, if large, obstructed the bladder outlet. The two entities overlapped in tumor size, male preponderance (in contrast to assertions about IP [6]), location within the bladder, T stage, and patient age. However, four of 12 followed-up patients with sarcoma had recurrence or death from fatal metastasis. The two metastatic tumors included one low-grade and one high-grade case; both patients were at the extremes of age: 4 and 72 years. Eight other patients were cured by partial or radical cystectomy. Similarly, recurrence and death are reported in one (18) of 15 reported cases (18, 19, 20, 21).

In contrast, no IP or PSCN have any documented instances of metastasis. IP tends to present with larger size than PSCN, and has seemingly limitless potential for local growth. We report the largest IP; untreated, it became 37.5 cm and caused death by obstruction, not metastasis. Seven prior studies traced IP outcome. Horn et al. followed three patients at 12 to 23 months (8); Ro et al. Followed eight patients at 24 to 108 months (9); Young and Scully followed one patient 42 months (10); Lakshmann followed one patient 4 months (11); Coyne followed one patient 9 months (12). Dietrick followed 2 patients 12 to 18 months (13); Jones et al. followed 10 new IP patients (and 1 with PSCN) 6 to 64 months (14). Therapy was size-dependent and was transurethral resection in 15 of the 26 patients, partial cystectomy in 10, and radical cystectomy in one. No tumors recurred or metastasized, suggesting that IP is benign. This finding also agrees with flow cytometry studies of IP that reveal diploid DNA content in 14 of 15 cases (6, 14, 22, 23). A limitation of our study is that the number of IP and PSCN cases precluded statistical analysis.

Separation of PSCN from IP is somewhat of an academic exercise, but PSCN differed by involving older patients (usually those undergoing resections of urothelial carcinoma), smaller average size, presence of eosinophils, and higher mitotic rate in some cases. The criterion that PSCN should be diagnosed only within 3 months of bladder surgery (7) seems too restrictive; lesions with distinctive PSCN histology in our study arose up to 13 months after surgery, at the prior surgical site. Thus we may consider reclassifying one case reported as IP in a 60-year-old man, 15 months after transurethral resection of urothelial carcinoma (14), as PSCN. By analogy, the histology of post-transurethral resection granulomatous prostatitis is so distinctive that it can be diagnosed 7 years after surgery. A novel finding is one IP with synchronous urothelial carcinoma. The patient's age (72 years) and the tumor size were closer to the range seen in PSCN, but there was no prior instrumentation. The association of carcinoma with PSCN reflects a detection bias, because the latter by definition arises after bladder instrumentation. We found no association of urothelial carcinoma with sarcoma, suggesting independent etiologies.

Other studies have emphasized the role of prior inflammation, instrumentation, or surgery in IP or PSCN of bladder or ureter. Implicated associations include hysterectomy for leiomyoma (8), pyelonephritis (8), ruptured diverticulitis of the sigmoid colon with pseudotumor extending to the bladder (24), transurethral bladder resection (7), indwelling catheter (14) and cystitis (23, 25). In our study, antecedent conditions in one of 13 IPs were neurogenic bladder secondary to cerebrovascular accident, and, in all PSCN, bladder biopsy or transurethral resection. No patient with sarcoma had prior bladder instrumentation, and one had an apparent postradiation sarcoma after radiotherapy for rectal carcinoma.

Muscle marker reactivity in myofibroblasts has been characterized as variable (26). Prior reports on IP showed immunoreactivity of IP for actins in most cases and vimentin in all cases (6, 8, 9, 14, 24), reflecting the myofibroblastic nature of the cells. Muscle-specific and α-smooth muscle actin stains were reactive in some IP, PSCN, and sarcomas. Desmin was positive in 3 of 8 of our IP/PSCN, as in prior reports (6, 14, 27, 28), but in no sarcomas; likewise, Mills et al. reported 4 of 12 bladder leiomyosarcomas were desmin-negative (17). Because all our sarcomas were desmin-negative, were variably reactive for actins just like benign entities, and none had electron microscopy performed, we resisted the tendency to label them as leiomyosarcoma. It is conceivable that they are fibrosarcoma or malignant counterparts of IP. Actins were also positive in some carcinomas. The latter observation suggests that these carcinomas might be carcinosarcoma; similarly, α-smooth muscle actin positivity was reported recently in 3 of 14 (29) and 1 of 1 (30) sarcomatoid bladder carcinomas and in 12 of 65 breast carcinomas (31). Desmin reactivity was seen in 3 of 14 bladder carcinomas, cases that may be true carcinosarcoma (29). In uterine smooth muscle neoplasms, p53 protein correlates with diagnosis and separates patients into prognostic groups (32). Although we are uncertain that our 13 sarcomas are leiomyosarcoma, we found that p53 was the only immunostain tested that discriminated IP and PSCN (rare reactive cells) from sarcoma (more diffuse, moderate to strong reactivity).

Epithelial markers had minimal discriminant value. About half of IP and PSCN were cytokeratin-positive in our study, similar to cumulative reports of reactivity in 12 of 31 IP cases (6, 8, 9, 10, 11, 14, 23, 24, 25, 27) and 2 of 5 of PSCN cases (10, 14, 28). Also, three of eight sarcomas were cytokeratin-positive, but all had necrosis and atypia. This is similar to reports of cytokeratin reactivity in 38% of leiomyosarcomas at various sites (33) and one case in bladder (34). The differential diagnostic potential of cytokeratin is further confounded by data from the spindle cell urothelial carcinoma series of Jones and Young (35) who found only 12 of their 19 carcinomas were positive for cytokeratin. Reactivity for epithelial membrane antigen, in our series, was noted in both benign and malignant tumors. Thus, even a substantial panel of the above immunostains may not conclusively distinguish pseudotumor/PSCN, sarcoma, or carcinoma. Fibronectin is reportedly specific for myofibroblasts (16), and perhaps should be studied in an IP/PSCN series like ours.

CONCLUSION

In conclusion, IP/PSCN and leiomyosarcoma of the bladder overlap in clinical and immunohistochemical findings but are histologically distinct, correlating with different natural histories. IP appears benign, not prone to metastasize, and is outside the spectrum of low-grade inflammatory sarcoma of bladder.