Renal cell carcinoma sub-typing by histopathology and fluorescence in situ hybridization on a needle-biopsy specimen

Transcription

1 Original Article RCC SUBTYPING BY FISH AND HISTOLOGY ON NEEDLE BIOPSY BAROCAS et al. Renal cell carcinoma sub-typing by histopathology and fluorescence in situ hybridization on a needle-biopsy specimen Daniel A. Barocas, Susan Mathew, Joseph J. DelPizzo, E. Darracott Vaughan Jr, R. Ernest Sosa, Ronnie G. Fine, Mohamed Akhtar and Douglas S. Scherr Departments of Urology and Pathology of New York Presbyterian Hospital Weill Cornell Medical Center, New York, NY, USA Accepted for publication 12 September 2006 OBJECTIVE To determine the subtype of renal cell carcinoma (RCC) on needle-core biopsies of renal masses using histopathology and fluorescence in situ hybridization (FISH), and to evaluate the use of interphase FISH to augment the accuracy of needle-core biopsies. PATIENTS AND METHODS Histology correlates with prognosis in RCC but, historically, biopsies are inaccurate for histological subtype. As histological subtypes of RCC have distinct cytogenetic abnormalities (loss of 3p in clear cell, trisomy 7 or 17 in papillary and widespread chromosomal losses in chromophobe), we hypothesized that FISH would improve the accuracy of biopsies. Forty patients with renal masses underwent nephrectomy, yielding 42 tumours. Needle-core biopsies were taken of the mass immediately after surgery. Interphase FISH was performed on one core for chromosomes 3, 7, 10, 13, 17, and 21 and the locus 3p Histopathology was performed on a second core. Results were compared in a blinded fashion with final pathology. RESULTS In all, 36 of 42 masses were RCC or oncocytoma. Histopathology of the biopsy correctly identified the tumour subtype in 27 (75%), while four (11%) were incorrectly classified and five (14%) were inadequate for diagnosis. With the addition of FISH, 31 (86%) were correctly subtyped, while two (6%) were incorrect and three (8%) were inadequate. In cases with adequate tissue, histology alone was 87% accurate, while the combined method was 94% accurate. CONCLUSION Needle-core biopsy of renal tumours provides adequate material for evaluation of histological subtype. Adding FISH to histopathology might improve the accuracy of kidney tumour biopsies, providing important prognostic information that can guide management decisions. KEYWORDS carcinoma, renal cell, in situ hybridization, fluorescence, biopsy, needle, cytogenetics, histopathology INTRODUCTION There are cases of RCC annually in the USA [1]. In the past, treatment was limited to radical nephrectomy, but now many are treated with nephron-sparing surgery [2] and ablative therapy is gaining favour in some centres [3]. Some investigators recommend observation for patients with small incidental renal masses [4]. The cohort of patients eligible for nephron-sparing surgery, ablation or observation is expanding as the proportion of small (<4 cm), incidental lesions increases [5]. It would be highly desirable to determine the biological aggressiveness of the tumour before deciding whether to observe or to treat (and if so, how to treat). Histology is known to correlate with prognosis and can serve as a marker of aggressiveness [6]. Clear cell histology is the most aggressive subtype, while papillary carries a slightly better prognosis; chromophobe RCC rarely metastasizes and oncocytomas are benign. Unfortunately, imaging cannot distinguish among the histological subtypes, and noninvasive markers are not available for clinical use. Renal biopsy has been used in selected settings, but the rate of nondiagnostic and inaccurately read biopsies can be >25% [7 9]. Clinicians have been unwilling to rely upon biopsy results to guide therapy due to this high rate of biopsy failure and inaccuracy, plus the theoretical risk of needle-tract seeding. In the present study, we evaluated the use of interphase fluorescence in situ hybridization (FISH) to augment the accuracy of needlecore biopsies of renal tumours. We developed a panel of probes based on the most common genetic aberrations in RCC, i.e. loss of the von Hippel Lindau (VHL) locus of chromosome 3p in clear cell RCC, trisomy 7 and 17 in papillary RCC, and widespread losses including chromosomes 10, 13, 17 and 21 in chromophobe RCC (Table 1) [10 13]. We then tested the FISH panel prospectively on kidney tumour biopsies. PATIENTS AND METHODS Permission was granted by the Institutional Review Board and all the subjects were patients undergoing nephrectomy or partial nephrectomy for renal mass at our institution between December 2004 and April Immediately after surgery, 14 G needle-core biopsies were taken. Tumours were identified by palpation and preoperative imaging, without decapsulating or cutting the kidney. 290 JOURNAL COMPILATION 2006 BJU INTERNATIONAL 99, doi: /j x x

2 RCC SUBTYPING BY FISH AND HISTOLOGY ON NEEDLE BIOPSY TABLE 1 The FISH probes Probe Locus Description CEP 7 7p11.1 7q11.1 α satellite/centromeric CEP 17 17p q11.1 α satellite/centromeric CEP 10 10p q11.1 α satellite/centromeric LSI 13 13q14 Enumeration locus LSI 21 21q q22.2 Enumeration locus VHL 3p25 3p26 Contig of two overlapping BACs that span the VHL locus BAC, bacterial artificial chromosome For histological evaluation, one core was fixed in formalin and paraffin blocks were prepared; slides were made and stained with haematoxylin and eosin. A second core was processed immediately under sterile conditions. The tissue was disaggregated using scalpel blades in a Petri dish with 3 µl of medium (RPMI Invitrogen, Carlsbad, CA, USA) then pipetted vigorously and forced through a 100-µm filter to form a single-cell suspension (SCS) of interphase cells. The SCS was centrifuged, the pellet was re-suspended in freezing buffer (fetal bovine serum with 10% dimethylsulphoxide) and frozen at 80 C. Later, the SCS was thawed rapidly and placed into RPMI medium. The suspension was centrifuged and the pellet was re-suspended in hypotonic KCl (0.075 M, Gibco, Carlsbad, CA, USA) at 37 C and incubated for 15 min to hydro-distend the cells. After centrifuging, the pellet was re-suspended in fixative (three parts 10% acetic acid to one part 100% ethyl alcohol). The suspension was centrifuged and re-suspended in fixative twice more and, after the last centrifugation, re-suspended in a few drops of fixative. Fixed cells were then dropped onto wet slides in a monolayer and allowed to dry on a 55 C hot-plate. Slides were aged overnight at room temperature. We selected commercially available centromere enumeration probes (CEP) for chromosomes 3, 7, 10, and 17, and locusspecific probes (LSI) for chromosomes 13 and 21 (Vysis Inc., Abbott Laboratories, Des Plaines, IL, USA). In addition, we generated a probe for the VHL locus by labelling a set of two overlapping bacterial artificial chromosomes that span the VHL locus (clones RP11 438J1, RP11 572M14, Invitrogen). The clones were amplified by growing them within bacteria in culture, then the complementary DNA sequences were extracted and labelled by the nick-translation method, described previously [14,15]. Each probe was validated on metaphase spreads of human leukocytes. For FISH, slides were pretreated in doublestrength saline sodium citrate (2 SSC, Fisher Scientific, Hampton, NH, USA) at 37 C for 30 min, then dehydrated in an alcohol series (2 min each in 70%, 80% and 100% ethanol). Slides were dried on the bench-top while the probe mixture was prepared. The mixture comprised 1 µl of probe, 7 µl of buffer (Vysis Inc.), and 2 µl of water. Slides with CEP probes only (CEP 7 and 17, CEP 10) were run through the automated Hybrite system (Vysis Inc.) after applying the probe. A standard protocol was used, with denaturation at 74 C for 1 min, and hybridization at 42 C for 24 h. A manual protocol was followed for the remaining slides (LSI 13 and 21, CEP 3 and 3p), with denaturation in 70% formamide in a 75 C water bath for 2 min. The slides were then run through a cold ethanol series. Probe mixtures were denatured in 75 C water bath for 5 min and then applied to the slides. The slides were incubated overnight at 37 C in a humidified chamber. The following day, slides were removed from the Hybrite or humidified chamber, and the slides were washed at 45 C serially in 50% formamide for 10 min, 2 SSC for 5 min twice, and 0.1 SSC for 5 min. Next, the slides were washed at room temperature in phosphate-buffered detergent for 5 min, and then rinsed with distilled water. Slides were allowed to air dry and 10 µl of 4,6,- diamidino-2-phenylindole (Vysis Inc.) counterstain were applied. For FISH analysis, the slides were evaluated under a Nikon Eclipse E800 photomicroscope equipped with epifluorescence, using a 100 objective. Selected images were captured with a charge-coupled-device camera (Kohu Electronics, San Diego, CA, USA) and analysis software (Cytovision Version 3.7, Applied Imaging, San Jose, CA, USA). Two readers (S.M. and D.A.B.) reviewed each slide and each counted 50 cells. A locus or chromosome was considered lost if >10% of cells counted had only one signal for that locus or chromosome [16]. Gains were defined as >5% of cells containing three, five or more signals [16]. Cells containing four signals (tetraploid) were considered normal. Any disagreement between readers was settled by joint review. A final cytogenetic interpretation was rendered for each case based upon the final aggregate count of both readers. One pathologist (M.A.), who was unaware of the final pathology, used WHO criteria [17] to give a biopsy diagnosis, first by reviewing the histology alone, then by reviewing the histology with the cytogenetic information. FISH was considered confirmatory if concordant with biopsy histology; if they were discordant, the histological biopsy diagnosis prevailed. The cytogenetic interpretation prevailed only if the histology was inconclusive or insufficient. Biopsy diagnosis was compared with final pathology diagnosis and sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for histology and histology plus FISH were calculated. Biopsy results with and without FISH were compared using the McNemar test. RESULTS Forty patients underwent surgery for a kidney mass, yielding 42 masses (Table 2). Six masses were not neoplastic (one calcified renal artery aneurysm, one organized haematoma) or not an RCC tumour (one each of urothelial carcinoma, gastrointestinal stromal tumour, GIST, metanephric adenoma, and mixed epithelial and stromal tumour, MEST) on final pathology. These six non-rcc masses were excluded from the FISH analysis. The two nonneoplastic masses could not be identified on histology. The four neoplastic lesions were identified as neoplasms on biopsy histology. The metanephric adenoma and the MEST were correctly subtyped, while the GIST and urothelial carcinoma were read as high-grade tumour and were considered JOURNAL COMPILATION 2006 BJU INTERNATIONAL 291

3 BAROCAS ET AL. incorrectly subtyped on subsequent calculations. The remaining 36 cases (86%) were RCC or oncocytoma (Table 3). Of these, 31 (86%) were readable on biopsy histology, while five (14%) were uninterpretable, because the tissue was necrotic/fibrotic (three), the fragment of tumour was too small (one), or the biopsy contained only normal kidney tissue (one). Histology alone was accurate for subtype in 27 of 36 cases of RCC/oncocytoma and in 27 of the 31 readable cases (87%) (Table 4). The FISH findings are summarized in Table 4; Fig. 1. shows examples. There was loss of 3p in nine of 16 (56%) clear cell RCCs, trisomy 7 or 17 in six of seven papillary RCCs, losses of two or more chromosomes in two of three chromophobic and normal results in three of five oncocytomas. The mucinous tubular and spindle cell carcinoma (MTSCC) had a loss of chromosome 13 in 70% of cells, which is consistent with published reports [17]. Thus, the FISH findings were concordant with final pathology in 21 of 36 (58%) cases and 21 of 31 (68%) cases with evaluable FISH results. Variable Value Number of patients 40 N (%): Male 22 (55) Female 18 (45) Mean (range) age, years 57.4 (34 89) Procedure, n (%) Radical nephrectomy 21 (53) Partial nephrectomy 19 (48) Laparoscopic/hand-assist 18 (45) Open 22 (55) Tumour size RCC/oncocytoma, n 36 Mean (range) size, cm 4.5 ( ) No. <4 cm (%) 19 (53) Pathology stage, n or n (%) 34 T1a 17 (50) T1b 6 (18) T2 3 (9) T3 8 (24) Node positive 1 (3) Metastatic 1 (3) Fuhrman nuclear grade, n or n (%) 27 Grade 1 1 (4) Grade 2 16 (59) Grade 3 8 (30) Grade 4 2 (7) TABLE 2 The patients demographics, surgical procedure and tumour characteristics. The numbers for pathology stage reflect 34 patients with 36 tumours; numbers for nuclear grade reflect grading of clear cell and papillary tumours only, according to WHO standard When the biopsies were re-evaluated in conjunction with the FISH results, three readings were changed from the incorrect subtype to the correct one. Two were originally read as poor quality material; tumour; favour chromophobe and were reread as clear cell when the FISH showed loss of 3p in 13% of cells in one case and 36% in the other. The third case was originally read as chromophobic and changed to papillary when FISH showed trisomy of 7 and 17 in 6% and 9% of cells, respectively. No case was changed from correct to incorrect. Two specimens that were insufficient for histopathological evaluation had enough tissue for the FISH; one had a loss of 3p in 13% of cells and was correctly identified as clear cell, while the other was incorrectly identified as papillary (clear cell on final pathology) due to the presence of trisomy 7 and 17 (in 30% and 29% of cells, respectively) and absence of loss at 3p. Thus, adding FISH facilitated correct identification of four additional tumours (Table 4). FISH improved the accuracy of the biopsy evaluation from 75% to 86% of all RCC/ oncocytomas and from 87% to 94% of RCC/ oncocytomas with readable histology or FISH RCC/oncocytoma Final pathology N All Evaluable* Non-neoplastic 2 Non-RCC/oncocytoma 4 RCC/oncocytoma 36 Clear cell Papillary 7 7 Chromophobe 3 3 Oncocytoma 5 5 MSTCC 1 1 Total number (not significant, McNemar test). Overall diagnostic accuracy of needle-core biopsy combined diagnosis and final pathology diagnosis was 69% for histology alone and 79% for histology and FISH. Sensitivity, specificity, PPVs and NPVs were calculated for the 31 cases with evaluable histology and the 33 cases in which either the FISH or histology was evaluable (Table 5). The FISH panel improved the PPV for chromophobe, as three specimens that were misclassified as chromophobe on histology were re-classified (two clear cell and one papillary) on review with the FISH results. This led to small improvements in the sensitivity and NPV of clear cell and papillary. DISCUSSION TABLE 3 Histological subtype on final pathology diagnosis *Excludes three cases of clear cell that had insufficient tissue on biopsy for histological diagnosis and insufficient tissue for FISH. Renal biopsy has not been used for preoperative diagnosis of renal masses because of concerns about accuracy and tumour-tract seeding, and because of the lack of alternatives to extirpative therapy. Dechet et al. [7] biopsied 100 kidney specimens ex vivo and found 20% were 292 JOURNAL COMPILATION 2006 BJU INTERNATIONAL

4 RCC SUBTYPING BY FISH AND HISTOLOGY ON NEEDLE BIOPSY TABLE 4 Concordance of needle-core biopsy diagnosis with final pathology diagnosis Final pathology Needle core Clear cell Papillary Chromophobe MTSCC Oncocytoma N at final pathology Histological diagnosis Clear Cell Papillary Chromophobe MTSCC Oncocytoma Not readable FISH findings* Loss of 3p Trisomy 7 or Widespread losses Normal Other abnormality Not readable Histology +FISH Clear cell Papillary Chromophobe MTSCC Oncocytoma Not readable *Some specimens had more than one cytogenetic abnormality. not diagnostic. Another 2 5% were incorrectly characterized for benign/ malignant status. They concluded that the high degree of inaccuracy (particularly in predicting benign tumours) precluded recommending needle biopsy to formulate treatment recommendations. Given the abundant data on the prognostic value of histology, other groups have looked more closely at subtyping of biopsy specimens. Wunderlich et al. [9] took renal biopsies by exposing the tumour postoperatively in 50 patients. With five biopsies of each lesion, they could determine the histological subtype and grade in 32 of 50 cases (64%). Neuzillet et al. [8] reported a 92% accuracy for histological subtype among 63 patients undergoing surgery for RCC. However, it is unclear whether their numbers include failed biopsies (three) and inconclusive biopsies (five). Furthermore, the accuracy for 22 patients who did not undergo surgery is unknown. While instances of tumour-tract seeding have been reported, it is rare, and these authors found no microscopic evidence of tract seeding. With improving observation protocols and ablative therapies, as well as the recognition that histological subtype of RCC has prognostic significance, the role for preoperative diagnosis is likely to increase. There is clearly a need to improve the accuracy of needle biopsy, if urologists will use the information to guide treatment/ observation decisions. Although these tumours are difficult to distinguish histologically on a biopsy, there are differences among the subtypes at the genetic, RNA and protein levels that can be exploited for the purpose of augmenting the accuracy of biopsies [18]. For example, our group recently published a series of renal tumour biopsies in which PCR for four markers was used to determine histological subtype [19]. To our knowledge, the present study is the first to prospectively evaluate the utility of FISH for augmenting the accuracy of histological diagnosis of renal tumour biopsies. FISH is a readily available, relatively simple, highly accurate method of detecting losses, gains and translocations of genetic material. It is used routinely in the diagnosis of haematological malignancies and fetal chromosomal derangements, and for surveillance of bladder cancer. Several researchers have shown the feasibility of identifying genetic abnormalities in RCC using FISH [12,13,20]. Akhtar et al. [20] used FISH on six cases of chromophobe RCC to confirm widespread monosomies. Sanjmyatav et al. [12] analysed 25 cases of archival tissue using eight probes and found abnormalities that were consistent with pathology diagnosis in 19 of 25 cases (76%). Siebert et al. [13] found loss of 3p in 15 of 19 clear cell RCCs and not in other subtypes. In the present study, 94% of RCC/ oncocytomas with adequate biopsy material for histology or FISH were correctly subtyped using a combination of FISH and histology. Adding FISH to histology resulted in the correction of four cases that were misclassified or unclassifiable on histology alone. Of the 17 clear cell carcinomas with readable FISH or biopsy histology, all were identified as RCC, and 16 of 17 (94%) were correctly subtyped. The present results show the feasibility of using FISH on renal tumour biopsy specimens, and suggest that it might be a useful adjunct to histology in clarifying equivocal cases and those with insufficient tissue for histological diagnosis. The present study has several limitations. First, there were three failed biopsies (7%), containing only necrosis/fibrosis or normal parenchyma. It is possible that real-time imaging would aid in tumour localization and reduce the likelihood of sampling errors. Second, not all tumours showed the classic cytogenetic abnormality specific to its tumour type; only 67% of cases with readable FISH results were concordant with final pathology. While FISH is excellent for detecting gains, losses and translocations, it can miss mutations, which might be particularly relevant in identifying clear cell carcinoma. There might also have been technical flaws responsible for missing genetic aberrations, or they were simply not present in all cases. Even with this rate of discordant cytogenetic results, FISH has a role in supporting and clarifying the histological diagnosis in difficult and equivocal cases, which might help reduce some of the inaccuracy seen in traditional biopsies. JOURNAL COMPILATION 2006 BJU INTERNATIONAL 293

5 BAROCAS ET AL. FIG. 1. Examples of histology and FISH results. Each tumour type shows classic histological appearance (haematoxylin and eosin stain, 100). The FISH results show loss of 3p in a clear cell RCC, trisomy 7 in a papillary RCC, widespread losses (of chromosomes 17, 3, 10, 13, and 21) in a chromophobe RCC, and diploidy in an oncocytoma. Histology FISH results p Clear cell Papillary Chromophobe Oncocytoma TABLE 5 The percentage sensitivity, specificity, PPV and NPV value of histology alone (H) and histology plus FISH (H + FISH) on needle-core biopsy Final pathology Sensitivity, In conclusion, to our knowledge this is the first study to evaluate FISH as an adjunct to histopathology in the pathology interpretation of needle-core biopsies of renal masses. With the addition of FISH, the diagnostic accuracy for histological subtype improved from 87% to 94% for biopsies with readable histology or FISH. This indicates that FISH might serve to refine the biopsy diagnosis of renal tumours. Specificity, PPV, ACKNOWLEDGEMENTS NPV, Clear cell 87/94 100/ /100 89/94 Papillary 86/ /96 100/88 96/100 Chromophobe 67/67 89/100 40/100 96/97 Oncocytoma 100/100 96/96 83/83 100/100 The authors acknowledge the contributions of Patricia Massino, Galina Aldrete (Cytogenetics Laboratory, Department of Pathology, New York Presbyterian Hospital) and Jim Dalton (Cytogenetics Laboratory, St. Jude Children s Research Hospital) who provided technical assistance. CONFLICT OF INTEREST None declared. Source of funding: departmental. REFERENCES 1 Jemal A, Murray T, Ward E et al. Cancer statistics, CA Cancer J Clin 2005; 55: Derweesh IH, Novick AC. Nephronsparing surgery for renal cell carcinoma. Cancer Treat Res 2003; 116: Weld KJ, Landman J. Comparison of cryoablation, radiofrequency ablation and high-intensity focused ultrasound for treating small renal tumours. BJU Int 2005; 96: Wehle MJ, Thiel DD, Petrou SP, Young PR, Frank I, Karsteadt N. Conservative management of incidental contrast- 294 JOURNAL COMPILATION 2006 BJU INTERNATIONAL

6 RCC SUBTYPING BY FISH AND HISTOLOGY ON NEEDLE BIOPSY enhancing renal masses as safe alternative to invasive therapy. Urology 2004; 64: Russo P. Localized renal cell carcinoma. Curr Treat Options Oncol 2001; 2: Amin MB, Amin MB, Tamboli P et al. Prognostic impact of histologic subtyping of adult renal epithelial neoplasms: an experience of 405 cases. Am J Surg Pathol 2002; 26: Dechet CB, Zincke H, Sebo TJ et al. Prospective analysis of computerized tomography and needle biopsy with permanent sectioning to determine the nature of solid renal masses in adults. J Urol 2003; 169: Neuzillet Y, Lechevallier E, Andre M, Daniel L, Coulange C. Accuracy and clinical role of fine needle percutaneous biopsy with computerized tomography guidance of small (less than 4.0 cm) renal masses. J Urol 2004; 171: Wunderlich H, Hindermann W, Al Mustafa AM, Reichelt O, Junker K, Schubert J. The accuracy of 250 fine needle biopsies of renal tumors. J Urol 2005; 174: Hoglund M, Gisselsson D, Soller M, Hansen GB, Elfving P, Mitelman F. Dissecting karyotypic patterns in renal cell carcinoma: an analysis of the accumulated cytogenetic data. Cancer Genet Cytogenet 2004; 153: Junker K, Weirich G, Amin MB, Moravek P, Hindermann W, Schubert J. Genetic subtyping of renal cell carcinoma by comparative genomic hybridization. Recent Results Cancer Res 2003; 162: Sanjmyatav J, Rubtsov N, Starke H, Schubert J, Hindermann W, Junker K. Identification of tumor entities of renal cell carcinoma using interphase fluorescence in situ hybridization. J Urol 2005; 174: Siebert R, Jacobi C, Matthiesen P et al. Detection of deletions in the short arm of chromosome 3 in uncultured renal cell carcinomas by interphase cytogenetics. J Urol 1998; 160: Mathew S, Raimondi S. FISH, CGH and SKY in the diagnosis of childhood acute lymphoblastic leukemia. In Swansbury J ed., Methods in Molecular Biology. Cancer Cytogenetics Methods and Protocols, Vol Totowa, NJ: Humana Press Inc., 2003: Rigby PW, Dieckmann M, Rhodes C, Berg P. Labeling deoxyribonucleic acid to high specific activity in vitro by nick translation with DNA polymerase I. J Mol Biol 1977; 113: Salama ME, Worsham MJ, DePeralta- Venturina M. Malignant papillary renal tumors with extensive clear cell change: a molecular analysis by microsatellite analysis and fluorescence in situ hybridization. Arch Pathol Lab Med 2003; 127: Lopez-Beltran A, Scarpelli M, Montironi R, Kirkali Z WHO classification of the renal tumors of the adults. Eur Urol 2006; 49: Linehan WM, Walther MM, Zbar B. The genetic basis of cancer of the kidney. J Urol 2003; 170: Barocas DA, Rohan S, Kao J et al. Diagnosis of renal tumors on needle biopsy specimens by histological and molecular diagnosis. J Urol 2006; 176: Iqbal MA, Akhtar M, Ulmer C, Al-Dayel F, Paterson MC. FISH analysis in chromophobe renal-cell carcinoma. Diagn Cytopathol 2000; 22: 3 6 Correspondence: Douglas S. Scherr, 525 East 70th Street, New York, NY 10021, USA. dss2001@med.cornell.edu Abbreviations: FISH, fluorescence in situ hybridization; VHL, von Hippel Lindau; SCS, single-cell suspension; CEP, centromere enumeration probes; LSI, locus-specific probes; SSC, saline sodium citrate; GIST, gastrointestinal stromal tumour, MEST, mixed epithelial and stromal tumour; MTSCC, mucinous tubular and spindle cell carcinoma; (P)(N)PV, (positive) (negative) predictive value. JOURNAL COMPILATION 2006 BJU INTERNATIONAL 295