We thank Roche Diagnostics France for supplying the immunologic reagents.
Taking into consideration the known overexpression of p16INK4a in histologically demonstrated high-grade urothelial malignancies, the objective of the current study was to examine the value of p16INK4a overexpression and of p16/Ki-67 dual labeling versus urinary cytology in the detection of urothelial lesions.
Immunolabeling was performed on demounted and destained Papanicolaou slides after liquid-based ThinPrep processing. Actual diagnoses were ascertained by cystoscopy controls and histopathology. Negative cases, papillary urothelial neoplasia of low malignant potential/low-grade tumor, and high-grade lesions were considered separately.
A total of 216 urine samples were collected from new patients with symptoms who were referred for cystoscopy (92 cases) or patients who were being followed after conservative treatment for lesions involving the bladder (117 cases) or the upper urinary tract (7 cases). p16INK4a positivity was assessed in 171 of the 216 cases (79.2%) and in 93 of 99 high-grade cases with positive cytology (93.9%). Coexpression of p16/Ki-67 in the same cells was observed in 119 of 216 cases (55.1%) and was noted in 18 of 51 cases of negative or papillary urothelial neoplasia of low malignant potential/low-grade tumor (35.3%) and in 80 of 101 high-grade tumors (79.2%) (P < .0001). Thirteen of 14 high-grade intraurothelial lesions (92.8%) were dual labeled. When high-grade tumors, disease progression (increased grade, muscle infiltration, and extension into the upper urinary tract), and cancer-related death were grouped together as an endpoint, dual labeling demonstrated a sensitivity that was slightly higher than that of urinary cytology (82.5% vs 80.8%; P = .8), with 94.9% overall specificity.
Human p16INK4a (p16) protein is the product of the cyclin-dependent kinase inhibitor 2A (CDKN2a) gene, which is located on chromosome 9p21. Physiologically, it acts as a tumor suppressor gene capable of inducing cell cycle arrest in the G1 or G2 phases. Loss by mutation or epigenetic silencing by methylation of the CDKN2a gene are observed in a wide variety of human tumors, including esophageal squamous cell carcinoma, gallbladder carcinoma, endometrial carcinoma, and other human tumors or cell lines.
In low-grade papillary urothelial lesions, p16INK4a is undetectable secondary to a decreased expression, as is reported in many nongynecological carcinomas.[2-5] Conversely, overexpression of p16INK4a protein has been reported in 11% to 100% of urothelial carcinomas (all grades together),[6-9] 80% of high-grade urothelial lesions, and approximately 100% of cases of carcinoma in situ (CIS).[10, 11] In these tumors, as in other premalignant or malignant human neoplasms (eg, prostatic intraepithelial neoplasia, basal-like breast carcinoma, small cell lung carcinoma, human papillomavirus [HPV]-related cervical and head and neck squamous cell carcinoma), p16INK4a overexpression is associated with aggressive disease and may serve as a tool for prognostic and therapeutic response assessment.[12-15] As has been demonstrated in other tumor types, immunoreactivity for p16INK4a has been found to be an independent predictor of disease progression in minimally infiltrating bladder carcinoma.
The p16INK4a protein is intimately connected with p53 and the retinoblastoma gene product (pRb). Abrogation of the p53 and pRb pathways may cause p16INK4a transcription, leading to detectable levels.[5, 16-18] In patients with cervical intraepithelial neoplasia and carcinomatous lesions of the uterine cervix, p16INK4a is overexpressed secondary to interference between the E6 and E7 oncoproteins of high-risk HPV and pRb.[19, 20]
To the best of our knowledge, HPV infection is rarely encountered in patients with bladder carcinoma. Furthermore, there is no correlation between p16INK4a overexpression and HPV in tissue and urine from inverted papilloma or urothelial carcinoma specimens.
In gynecologic cytopathology, p16/Ki-67 dual labeling has been found to provide high sensitivity in the detection of high-grade cervical intraepithelial neoplasia of grade 2 or higher in women with atypical squamous cells of undetermined significance (ASC-US) or low-grade squamous intraepithelial lesions (LSILs). In the urine, we have previously shown that dual labeling could help to identify high-grade cancer cells in new patients with symptoms and in those followed after conservative treatment of bladder cancer. However, accurate assessment was limited by the low number of cases and by a significant percentage of equivocal “atypical” cytology cases. Studying a new series with conventional diagnostic categories was therefore necessary.
The current study was undertaken to investigate whether dual labeling could help to identify new or recurring urothelial carcinoma cases and, more importantly, assess the risk of disease progression after the conservative treatment of urothelial carcinoma of the bladder in a series of patients seen in a urology setting.
MATERIALS AND METHODS
A total of 216 urine samples obtained after cystoscopy were collected from 187 patients who underwent consult mainly in the department of urology at the Centre Hospitalier Lyon Sud in Lyon, France (chief of department: A.R.). There were 34 women and 153 men (81.8%) aged 69 years ± 13 years. Patients were either new patients with symptoms who were referred for cystoscopy (92 patients) or those who were being followed after undergoing complete transurethral resection (TUR), bacillus Calmette-Guerin (BCG) immunotherapy, or surgery for lesions involving the bladder (117 samples) or the upper urinary tract (UUT; 7 samples).
We excluded 2 patients with nonurothelial tumors (1 with a bladder epidermoid carcinoma and 1 with clear cell renal carcinoma) and 14 patients for whom p16/Ki-67 dual labeling was not evaluable.
The results of cystoscopy were recorded in the majority of cases, regardless of the clinical situation. TUR was performed in every case of a papillary bladder lesion, whereas mucosal abnormalities suspicious for CIS were evaluated by biopsies. Some patients underwent surgery for UUT tumors.
Urinary Cytology and Histopathology
The samples were sent to the laboratory within 12 hours of being obtained with a clinical history and cystoscopy data sheet. After centrifugation (600 g for 10 minutes), the cell pellet was suspended in a PreservCyt solution (Cytyc Corporation, Boxborough, Mass) containing methanol and mucolytic and hemolytic agents. After 30 minutes at room temperature, cells were centrifuged and resuspended in 45 mL of CytoLyt solution (Cytyc Corporation) before being treated with the ThinPrep 2000 processor (Hologic Corporation, Marlborough, Mass). The software allows thin-layer cell preparations to be obtained by filtration, as previously described. Slides were stained with a hypochromic Papanicolaou (Pap) procedure and embedded in a permanent mounting medium under coverslips. We first acquired Pap images with a Leica DMR microscope (Leica Microsystems, Buffalo Grove, Ill) using a × 40 objective. Before immunolabeling, areas of interest were carved under the slides and the coverslips were removed by immersion in xylene for 3 days.
Cytopathologic analysis was performed at both low-power and high-power fields. Normal, inflammatory, reactive, and degenerative conditions of the urothelial component were considered as negative, as described by previous studies.[25, 26] Atypical urothelial cells (AUC) were subcategorized as AUC of undetermined significance (AUC-US) and AUC, cannot exclude high-grade (AUC-H). We have previously demonstrated that AUC not otherwise specified match with underlying high-grade lesions in 66% of cases and precede high-grade lesions in 36% of cases at a mean interval of 10 months.[27, 28] In our experience, AUC-H are defined as nonsuperficial urothelial cells with nuclear abnormalities (increased nuclear-to-cytoplasmic ratio, dense chromatin, eccentric nucleus, and/or irregular nuclear shape) in a small number that does not allow their actual categorization as malignant.[27, 28] Specimens in which obvious high-grade cancer cells were evident were considered to be positive.
The 2004 World Health Organization grading system and International Union Against Cancer TNM staging system were used. Histopathological data were separated into 2 groups: 1 group that was positive for high-grade urothelial lesions (pTIS, papillary and invasive) according to the criteria of the World Health Organization classification and 1 group that included negative results and those consistent with low-grade papillary tumors (grade 1 or papillary urothelial neoplasia of low malignant potential [PUN-LMP], grade 2 or low-grade, pTa-pT1 tumors).
Slides destained by 70% ethanol overnight were then submitted to antigen retrieval in a 10× solution of 100 mM of Tris, 10 mM of ethylenediamine tetraacetic acid [pH 9.0], and 15 mmol/L of sodium azide. Simultaneous immunostaining for p16INK4a and Ki-67 was performed using the CINtec PLUS Cytology Kit (Reference 9531; Roche Diagnostics France, Meylan, France) according to manufacturer's instructions in an Autostainer Link 48 (Dako, Glostrup, Denmark). The kit contains a ready-to-use primary antibody cocktail comprising a mouse monoclonal antibody (clone E6H4) directed toward human p16INK4a protein and a rabbit monoclonal antibody (clone 274-11 AC3) directed against the human Ki-67 protein. Ready-to-use reagents comprising 1) a polymer reagent conjugated to horseradish peroxidase and goat antimouse fragment antigen-binding Fab′ antibody fragments and 2) a polymer reagent conjugated to alkaline phosphatase and goat antirabbit Fab′ antibody fragments were used. Horseradish peroxidase -mediated conversion of 3,30-diaminobenzidine chromogen and alkaline phosphatase-mediated conversion of Fast Red chromogen lead to brown and red staining at the p16INK4a and Ki-67 antigen sites, respectively. After counterstaining by alcohol-free hematoxylin, we applied a 2-step mounting procedure, first by using an aqueous mounting medium provided with the kit to prevent alcohol-based fading of the Fast Red signal, followed by a permanent mounting step. Negative and positive controls were included; the C4-1 cervical carcinoma line cells were used as controls for p16/Ki-67 dual labeling and Jurkat T lymphocytic cells were used as controls for Ki-67.
For the interpretation of p16/Ki-67 dual-stained cells, 2 pathologists (E.P. and A.S.A.) and a scientific advisor of Roche Diagnostics France (C.C.) reviewed cases for the presence of p16INK4a or p16/Ki-67 dual labeling. The presence of ≥ 1 urothelial cell(s) simultaneously demonstrating brown cytoplasmic staining and red nuclear staining indicative of p16INK4a and Ki-67 expression, respectively, defined a positive result, irrespective of the interpretation of morphologic abnormalities. Cases without any double immunoreactive cells were determined to be negative for p16/Ki-67 dual labeling.
Patients had histological control after cystoscopy at mean delay of 0 months to 6 months, and those without an actual diagnosis were followed for an additional 12 months. Cystoscopy and urinary cytology were performed at every control according to the European Association of Urology recommendations for follow-up. Disease progression was defined as tumor recurrence at a higher stage or grade (TNM pTa-pT1 with transition from grade 1 or grade 2 to a higher grade, progression from pTa-pT1 to ≥ pT2, and histologically documented metastases or death from urothelial cancer). Single or combined p16INK4a and p16/Ki-67 expression was compared with Pap images that were acquired before labeling. Only nonsuperficial urothelial cells with p16INK4a overexpression were considered to be positive, irrespective of their nuclear features. Dual labeling was evaluated in patients with negative/low-grade tumors and in those with high-grade lesions using the Fisher exact test and calculation on paired series when appropriate (BiostaTGV; INSERM UPMC U707, Paris, France). P values < .05 were considered to be statistically significant.
Among the 124 previously treated cases, 58 (46.8%) were PUN-LMP/low-grade bladder lesions, 57 (45.9%) were high-grade bladder tumors including 1 case of high-grade urothelial dysplasia, 7 (5.6%) were UUT tumors (3 of which were pTXGX, 3 of which were pTaG2, and 1 of which was pT2G3), 1 was a pTXGX bladder tumor, and 1 was a pTX bladder epidermoid carcinoma. Of a total of 92 patients with symptoms, 43 had hematuria (46.7%), 12 had micturition disorders (13.0%), 3 had cystitis (3.3%), and bladder or UUT lesions were viewed by imaging techniques in 2 patients. Clinical data were unavailable for 32 patients (34.8%).
Globally at the time of consultation, 62 of 216 samples (28.7%) had negative cystoscopy findings, 21 (9.7%) had suspicious mucosal aspects leading to bladder or urethral biopsies, 96 (44.5%) had typical papillary growths that lead to bladder TUR, and 13 (6.0%) were found to have UUT tumors. Twenty cases (9.3%) had no cystoscopy report (though it was performed) and 4 cases (1.9%) had no cystoscopy control.
Cytological and Histological Initial Evaluation
In the entire series, with the exception of 64 cases (29.6%) for which there was no histology, TUR, biopsies, and/or surgical evaluations were obtained at a mean delay of 1.6 months ± 2.1 months (range, 0-11 months) in the initial evaluation period. There were 123 bladder TURs performed (80.9%), 11 patients underwent bladder biopsy (7.2%), 8 patients underwent nephroureterectomy (5.3%), 6 patients underwent UUT biopsy (3.9%), 2 patients underwent renal biopsy, 1 patient underwent a cystoprostatectomy, and 1 patient had pelvis surgery specimens available for cytohistologic comparisons.
Overall, 129 tumors were histologically diagnosed. There were 101 high-grade lesions (including 14 high-grade intraurothelial lesions and 8 UUT tumors) and 28 PUN-LMP/low-grade urothelial carcinomas including 2 UUT tumors. There were also 23 negative histological controls (15.1%) including 2 UUT specimens.
Urinary cytology was negative in 66 cases (including 2 cases of decoy cells in renal transplant recipients) and consistent with PUN-LMP/low-grade tumor in 24 cases, 15 cases demonstrated AUC-US (recorded to as negative), 12 cases demonstrated AUC-H (recorded to as positive), and 99 cases were found to have true high-grade cancer cells.
Urinary cytology was positive, high-grade, or demonstrated AUC-H in 83 of 101 histologically proven high-grade urothelial lesions (82.2%), but positivity awaited confirmation by prolonged follow-up in 17 cases (15.3% of AUC-H plus high-grade cytological results taken together). Seven of 24 PUN-LMP/low-grade cytological results (29.2%) and 11 of 81 negative cytological results (13.6%) were undergraded or were diagnosed as falsely negative, with histology demonstrating high-grade urothelial lesions (Table 1).
Table 1. Comparison of Cytological Findings With Cystoscopy and Histology Data When Available at the First Evaluation at 0 to 6 Months
Positivity for p16INK4a was assessed in 171 of the 216 cases (79.2%) (Fig. 1) and in 93 of 99 high-grade cases with positive cytology (93.9%). Comparison with histology demonstrated that p16INK4a was overexpressed in 41 of 51 negative or PUN-LMP/low-grade cases (80.4%) and in 93 of 101 high-grade urothelial tumors (92.1%) (P = .05) (Table 2).
Table 2. p16INK4a Overexpression and p16INK4a/Ki-67 Dual Labeling According to Clinicopathologic Data Including Follow-Up
Mean Follow-Up, Months
Dual Labeling, %
Abbreviation: PUN-LMP/LG, papillary urothelial neoplasia of low malignant potential/low-grade tumor.
Data were calculated for 36 cases (67.9%) that were regularly followed according to the European Association of Urology recommendations. Other patients were not reviewed at 10.4 ± 5.0 months.
Odds ratio of 0.91 (95% confidence interval, 2.9965-43.277).
Data were calculated for 24 cases (one “superficial” papillary lesion was laser fulgurated, without histology).
Odds ratio of 15.41 (95% confidence interval, 3.4568-88.6232).
Odds ratio of 3.01 (95% confidence interval, 1.205-8.2848).
Coexpression of p16INK4a and Ki-67 (dual labeling) in the same cells was observed in 119 of 216 cases in the entire series (55.1%), and in 92 of 111 high-grade plus AUC-H cases (82.9%) (Fig. 2). The omission of AUC-H cases allowed 93 of 99 true high-grade cases with positive cytology (93.9%) to be dual labeled.
Comparison with histology indicated that dual labeling was noted in 18 of 51 negative or PUN-LMP/low-grade cases (35.3%) and in 80 of 101 high-grade lesions (79.2%) (P < .0001). Thirteen of 14 high-grade intraurothelial lesions (92.8%) were dual labeled.
Globally among 66 cases with negative cytology, only 8 cases (12.1%) were dual labeled. Compared with the final results, the 8 cases matched with 2 high-grade tumors, 2 cases of disease progression (delays of 11 months and 21 months, respectively), 1 low-grade tumor, 1 negative control at 8 months, and 1 patient who was lost for control.
Impact of Follow-Up on Diagnostic Values
After the initial evaluation (range, 0 month-6 months), 137 patients (63.4%) had histologically confirmed diagnoses, 1 patient had died of cancer-related complications, 2 patients were aged > 85 years and were lost for control, and 76 patients were followed for an additional mean period of 15 months ± 7.4 months (range, 7 months-38 months) (Table 3).
Table 3. Sensitivity and Specificity Values of Cytology, p16INK4a Overexpression, and Dual Labeling According to the Population Studied, including Follow-Up for All Neoplasms Togethera or Cases of High-Grade Tumors, Disease Progression, and Cancer-Related Death Only
Cases of papillary urothelial neoplasia of low malignant potential/low-grade tumor, low-grade tumor, high-grade tumor, disease progression, and cancer-related death.
All Neoplasms Together
Cases of high-grade tumors and disease progression only
Taking into account initially obtained and follow-up results, sensitivity and specificity results were calculated for 198 cases. Eighteen cases (8.3%) had to be excluded since 13 patients were lost to follow-up and 5 had high-grade positive cytology results that were not confirmed at the time of analysis.
When considering cases with combined negative cystoscopy, negative cytology, and negative follow-up (mean period of 10.4 months ± 9.7 months), 24 of 53 cases (45.3%) demonstrated p16INK4a overexpression. In the same subpopulation, only 4 cases (7.5%) were dual labeled.
Considering the combination of high-grade tumors, disease progression, and cancer-related death as an outcome, dual labeling gave sensitivity results that were slightly higher to those of urinary cytology (82.5% vs 80.8%; P = .8), with a 94.9% identical specificity. In the same population, p16INK4a overexpression allowed for a higher sensitivity (93.3%) to be obtained, with decreased specificity (69.2%).
When appropriately applied to the detection of high-grade lesions (those that might progress to muscle infiltration or metastases), urinary cytology most often achieves sensitivity values > 80%, and specificity values of 90% to 95%.[30-33]
The sensitivity value obtained in the current series (80.8%) is not very different from those we obtained in previously published series: 83.3% and 64.1%, respectively, in new patients and in those followed after conservative treatment or a global sensitivity of 88.9% in high-grade tumors. The majority of series have achieved pooled sensitivity and specificity results of 42% and 96%, respectively. For comparison, fluorescence in situ hybridization (FISH) using the UroVysion assay allows 72% and 83% pooled sensitivity and specificity results to be achieved.
However, sensitivity appears to be lower in the follow-up period of conservatively treated patients compared with new patients. Inflammatory/reactive changes secondary to TUR, BCG immunotherapy, and intravesical chemotherapy may reduce the percentage of diagnostic cells and introduce equivocal variations. For both reasons, many cytology reports are therefore categorized as “suspicious” or “atypical.” Some authors have suggested using new subcategories derived from the Bethesda system for cervical lesions.[28, 37] The terms AUC-US and AUC-H were introduced, assuming that the meaning and use of AUC-US would be analogous to the category of ASC-US in gynecologic cytopathology, and that AUC-H would be better treated as a high-grade positive diagnosis.
The finding that urinary cytology is less efficient after conservative treatment for bladder cancer (most often closer to 80% than 100%) has for years stimulated the search for biomarkers that could improve accuracy in the follow-up period. Among many potential markers, only a few (including Ki-67, uCyt+, p53 protein, and UroVysion FISH) have been recognized as having an impact in urinary cytology specimens.[31, 34, 38] In recent years, one of the most promising biomarkers has been p16INK4a, with particular applicability in gynecopathology in the triage of ASC-US and LSIL Pap cytology results.[19, 20, 39] It is also an interesting marker in other malignancies, because although CDKN2a is clearly a tumor suppressor gene, aberrant expression of its protein is observed in several human cancers.
Functionally, p16INK4a is intimately connected to p53 and pRb. Among other causes, interference or abrogation of the p53 and pRb cell cycle pathways may cause CDKN2a transcription, leading to protein overexpression as observed in high-risk HPV-induced cervical lesions.[19, 39, 40] Overexpression of p16INK4a protein has been reported in 11% to 100% of urothelial carcinomas,[6-9] 80% of high-grade urothelial lesions, and approximately 100% of CIS cases on histologic sections.[10, 11] A recent study even found p16INK4a overexpression in 81.3% and 64.6%, respectively, of low-grade urothelial carcinomas diagnosed by biopsy and cytology. In the current series, we found overexpression of p16INK4a in 97.2% of urinary cytology samples with histologically proven high-grade urothelial carcinomas.
In gynecologic pathology, p16/Ki-67 dual labeling provides high sensitivity and specificity values for the detection of biopsy-confirmed high-grade lesions (cervical intraepithelial neoplasia grade 2 or higher) within a 6-month follow-up period in patients with Pap smears initially categorized as ASC-US or LSIL. In patients with urothelial malignancies, we have previously shown dual labeling in 100.0% of positive high-grade urinary cytology results. Because to the best of our knowledge there is no statistical difference between our previous series and the current study (93 of 99 cases [93.9%] vs 37 of 37 cases [100%]; P = .18), we can conclude that dual labeling allows most, if not all, high-grade positive cytology cases to be confirmed. Moreover, when examining histology, 13 of 14 CIS cases (92.8%) were dual labeled in the current study versus 16 of 17 cases (94.1%) in the previous study. Dual labeling therefore appears to be an excellent tool for the triage of patients with high-grade urothelial malignancies.
We had previously suggested that p16/Ki-67 dual labeling could allow for recognition of most high-grade urothelial cancer cells in new patients and in those individuals who were conservatively treated for urothelial carcinoma. However, p16/Ki-67 dual labeling in the context of positive high-grade urinary cytology is not essential from a diagnostic point of view because the positive predictive value of urinary cytology in patients with high-grade lesions is already very high. More interesting theoretically is the context of equivocal, “atypical” cytology reports, which remain a notorious problem after conservative treatment. In a previous series, we demonstrated that AUC not otherwise specified preceded high-grade lesions in 36.1% of cases at a mean interval of 10.5 months ± 12.0 months and that overall they were associated with or predictive of high-grade lesions in 73.8% of cases. However, as previously shown, p16/Ki-67 dual labeling does not allow for the accurate triage of these atypical urinary cytology cases. Other ancillary methods such as FISH cannot help to clarify equivocal cytology findings, except in patients treated with BCG immunotherapy, in whom a positive FISH result appears to be superior to cytology as a marker of disease recurrence when high-grade positive cytology is excluded from the analysis.
The high sensitivity of dual labeling in patients with high-grade urothelial lesions is not compromised by reduced specificity, but it demonstrates no significant difference compared with conventional urinary cytology. The good results obtained are most likely due to the ThinPrep treatment followed by hypochromic Pap staining, and by the use of restrictive diagnostic criteria in the categorization of atypical cases. In the current series, only cases with TUR, biopsy, or surgical confirmation were included and the follow-up period was restricted to the time necessary for histological control plus 1 additional year.
The main limitation of the current study is that it was performed on destained and demounted material. It is highly probable that antigen retrieval performed on such slides reduces the number of available diagnostic cells for the CINtec PLUS assay. More accurate results should most likely be obtained prospectively on ThinPrep slides without destaining, as it was performed in previous studies on cervical lesions.[22, 39] In those studies, the liquid-based ThinPrep vial contained sufficient residual material for additional slides to be prepared and for HPV testing. In addition, slides had to meet the minimum cellularity criteria as specified in the Bethesda system for reporting cervical cytology. In urine cytopathology, in which low cellularity (even after cystoscopy or bladder lavage) usually does not allow for additional slides to be prepared, the CINtec PLUS assay cannot be performed as a reflex test in conjunction with conventional cytology.
The value of dual labeling, similar to that of any other urinary marker, remains to be demonstrated in cases with negative cytology results. In the current series, only 8 of 66 negative cytology cases (12.1%) were dual labeled. Compared with the follow-up data, 4 of 8 cases (50%) were matched with high-grade urothelial tumors or cases of disease progression. Grouping together negative cystoscopy plus negative cytology plus negative follow-up (mean follow-up period, 10.4 months ± 9.7 months), 45.3% of cases demonstrated p16INK4a overexpression, but only 4 cases (7.5%) were dual labeled. The results of the current study indicate that in addition to its value in detecting high-grade urothelial proliferation, the current marker could do more than play a simple role in identifying tumors that may have prognostic impact (ie, those that may progress from superficial to invasive or metastatic disease).
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
Dr. Carré acts as a member of the Scientific Advisory Board for Roche Diagnostics France and has a financial interest in the company.