DHH is an Independent Prognosticator of Oncologic Outcome of Clear Cell Renal Cell Carcinoma €ger,* Christian Thomas,* Ladan Fazli, Antonio Hurtado-Coll, Estelle Li, Wolfgang Ja Claudia Janssen, Kilian M. Gust, Alan I. So, Michael Hainz, Irene Schmidtmann,† € roff, Walburgis Brenner‡ and Peter C. Black‡,§ Frederik C. Roos, Joachim W. Thu From the Vancouver Prostate Centre and Department of Urologic Sciences, University of British Columbia (WJ, LF, AH-C, EL, CJ, KMG, AIS, PCB), Vancouver, British Columbia, Canada, and Department of Urology (WJ, CT, CJ, FCR, JWT, WB), Institute of Pathology (MH) and Institute of Medical Biostatistics, Epidemiology and Informatics (IS), Johannes Gutenberg University, Mainz, Germany
Abbreviations and Acronyms ccRCC ¼ clear cell RCC CSS ¼ cancer specific survival DHH ¼ desert hedgehog Gli ¼ glioma-associated oncogene HH ¼ hedgehog IHH ¼ Indian hedgehog PTCH1 ¼ patched1 RCC ¼ renal cell carcinoma SHH ¼ sonic hedgehog Smo ¼ smoothened TMA ¼ tissue microarray Accepted for publication July 10, 2014. Study received institutional review board approval. * Equal study contribution. † Financial interest and/or other relationship with Merck-Serono and Charles River Laboratories. ‡ Equal study contribution. § Correspondence: Department of Urologic Sciences, University of British Columbia, Level 6, 2775 Laurel St., Vancouver, British Columbia V5Z 1M9, Canada (telephone: þ1 (604) 875-4818; FAX: þ1 (604) 875-5654; e-mail: pblack@mail. ubc.ca).
Purpose: Aberrant HH signaling has proved important in the pathogenesis of several solid cancers. Limited in vitro analyses suggested an oncogenic role for HH in renal cell carcinoma. In this explorative study we sought to validate aberrant HH expression in patients with renal cell carcinoma. Materials and Methods: A tissue microarray was constructed from 140 radical nephrectomy specimens of patients with clear cell renal cell carcinoma. We performed immunohistochemistry for Ki67 and HH pathway biomarkers, including PTCH1, Smo, SHH, IHH, DHH, Gli1, Gli2 and Gli3. Staining intensity was measured by automated image processing and related to tumor stage and grade. The impact of biomarker expression on cancer specific survival was determined by univariate and multivariate Cox regression analysis. Results: Gli3, PTCH1, DHH and SHH demonstrated markedly higher expression in high than in low grade tumors. Tumor stage was not associated with marker expression. On univariate analysis DHH expression, and tumor grade and stage were associated with cancer specific survival. Multivariate analysis revealed that DHH, grade and stage were independent predictors of cancer specific survival. Conclusions: To our knowledge we report for the first time that a biomarker of the HH pathway is associated with adverse pathological features and poor disease outcomes in patients with clear cell renal cell carcinoma. DHH may serve as an independent predictor of cancer specific survival in clear cell renal cell carcinoma cases. This supports further evaluation of HH signaling to validate the pathway as a target for novel therapy. Key Words: kidney; carcinoma, renal cell; hedgehog proteins; biological markers; mortality
KIDNEY cancer accounts for 3% to 4% of all solid tumors in developed countries.1,2 Until recently its incidence steadily increased about 2% per year.1 It was estimated that in 2013 there would be 65,150 incident cases and 13,680 deaths from kidney cancer in the United States.2 RCC represents 85% of all kidney malignancies,
1842
j
www.jurology.com
of which 80% are ccRCC.3 Surgery is the mainstay of treatment for localized RCC. Surgery is used in conjunction with systemic vascular endothelial growth factor or mammalian target of rapamycin targeted therapy and rarely interleukin-2 therapy in the metastatic setting.3 Despite curative intent metastasis
0022-5347/14/1926-1842/0 THE JOURNAL OF UROLOGY® © 2014 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARCH, INC.
http://dx.doi.org/10.1016/j.juro.2014.07.013 Vol. 192, 1842-1848, December 2014 Printed in U.S.A.
DHH IS PROGNOSTICATOR OF ONCOLOGIC OUTCOME IN CLEAR CELL RENAL CELL CANCER
develops in approximately 30% of patients with localized disease. At this stage only 10% of patients survive 5 years.4 Current gaps in the management of RCC cases include our inability to accurately predict the disease outcome and the pressing need for further novel therapies to improve outcomes.5 In the last decade multiple histopathological biomarkers have been tested for their predictive value for RCC. However, most markers that initially showed promising results could not be validated on multivariate analysis.6 Few biomarkers, such as vascular endothelial growth factor,6 the proliferation marker Ki67 and carbonic anhydrase IX, are independent prognosticators of RCC.7 To date none of these markers has entered routine clinical practice.3 The HH pathway is a key regulatory network during embryonic and postnatal development.8 It is activated by the binding of 1 HH ligand (SHH, IHH or DHH) to the transmembrane receptor PTCH1. PTCH1 binds and suppresses Smo in the absence of ligand but after ligand binding releases this suppression Smo can activate the Gli proteins Gli1, Gli2 and Gli3, which act as specific transcriptional factors.9 HH pathway reactivation was noted in different cancer entities such as ovarian, upper gastrointestinal tract and pancreatic cancer,10e12 including urological malignancies of the prostate and bladder.13,14 Some evidence suggests that the HH pathway may also have an important role in RCC.15 We evaluated expression levels of 9 intermediates of the HH pathway by immunohistochemistry in a cohort of patients who underwent nephrectomy for RCC. Staining intensity of these markers was related to common histopathological features and patient outcomes.
MATERIAL AND METHODS Patients and Specimens Tumor specimens were collected from 140 patients who underwent radical nephrectomy for ccRCC at the Department of Urology at University Medical Center of Johannes Gutenberg-University in Mainz, Germany between 1997 and 2002. Surgical specimens were formalin fixed and paraffin embedded, and examined by a genitourinary pathologist. Staging was determined according to the 7th edition of the AJCC (American Joint Committee on Cancer)/UICC TNM classification16 and grading according to the Fuhrman classification.17 Fuhrman grades 1 and 2 were considered low, and grades 3 and 4 were considered high. Corresponding clinical information on all patients was retrieved from the medical records. Followup was available in 132 patients. The study was approved by the institutional review board and patients provided informed consent.
1843
TMA Construction A TMA was created from formalin fixed, paraffin embedded tissue blocks using a manual tissue microarrayer (Beecher Instruments, Sun Prairie, Wisconsin). Each marked block was sampled twice with a core diameter of 1.0 mm and arrayed in a rectangular pattern with 1.6 mm between the center of each core to create a duplicate TMA layout.
Antibodies Specific antibodies of intermediates of the HH pathway were chosen based on previous publications. Polyclonal antibodies against SHH (1:25, N-19, sc-1194), DHH (1:100, N-19, sc-1193), Smo (1:10, C-17, sc-6367), PTCH1 (1:100, G-19, sc-6149), Gli1 (1:25, H-300, sc-20687) and Gli3 (1:200, H280, sc-20688) were obtained from Santa Cruz Biotechnology, Santa Cruz, California. Monoclonal antibody against IHH (1:100, ab52919) and polyclonal antibody against Gli2 (1:50, ab26056) were obtained from AbcamÒ. Anti-Ki67 (1:150, RM-9106) was obtained from Dako, Burlington, Ontario, Canada.
Histology and Immunohistochemistry TMAs were cut into 4 mm sections. Staining was performed with specific primary antibodies and corresponding secondary antibodies using the Discovery XT autostainer (VentanaÒ) using an enzyme labeled biotin streptavidin system and a solvent resistant 3,30 -diaminobenzidine Map kit.
Immunostaining Evaluation All slides were scored categorically as 0, 1þ, 2þ or 3þ by the institutional pathologist (LF). They were digitized with the SL801 autoloader system and SCN400 scanning system (Leica Microsystems, Concord, Ontario, Canada) at 20 magnification. Images were stored in the SlidePath Digital Image Hub (Leica Microsystems) at Vancouver Prostate Centre. Automated tissue analysis was subsequently performed according to previously described algorithms.18 For each biomarker the staining pattern (nuclear vs cytoplasmic vs membranous) was evaluated. Representative cores (clearly positive, clearly negative and mixed positive/negative) were manually identified to create color segmentation. Tissue image analysis capabilities (Leica Microsystems) with established algorithms were adjusted using these references and morphometric parameters. Cells that stained for each biomarker in each TMA core were subsequently measured by positive pixel identification using the tandem platform tissue image analysis and Digital Image Hub. All results were stored in the database and linked to patient information. Each tumor was represented by duplicate cores so that the mean score of the 2 cores was used for analysis.
Statistical Analysis Results of visual (categorical) and automated (continuous) analysis of staining intensity were analyzed by the Spearman correlation test. All automated analysis results are shown as the median and IQR. Groups were compared using the Mann-Whitney test with differences considered nonrandom at p 0.05. CSS depending on DHH expression was estimated by the Kaplan-Meier method using median DHH expression (307 relative units) as the cutoff.
1844
DHH IS PROGNOSTICATOR OF ONCOLOGIC OUTCOME IN CLEAR CELL RENAL CELL CANCER
Table 1. Pathological specimen tissue expression and staining pattern of HH pathway intermediates No. Tissue Expression (%) Biomarker DHH SHH IHH Smo PTCH1 Gli1 Gli2 Gli3
0
No. Pts 138 139 135 137 140 138 138 137
35 6 49 0 0 29 25 10
(25.4) (4.3) (36.3) (21.0) (18.1) (7.3)
1þ 83 66 67 36 86 86 93 62
(60.1) (47.5) (49.6) (26.3) (61.4) (62.3) (67.4) (45.3)
2þ 20 53 19 74 44 22 20 57
3þ
(14.5) (38.1) (14.1) (54.0) (31.4) (15.9) (14.5) (41.6)
Univariate Cox regression analysis and stepwise backward multivariate Cox regression analysis (enter if p 0.1) were performed for CSS using MedCalc (MedCalc Software, Ostend, Belgium). The Harrell c-index was calculated using the R-function coxph. The c-index is a generalization of the AUC measure to censored data that measures how well the fitted Cox proportional hazard model distinguishes between patients who die and those who survive.
RESULTS Patient Cohort Clinical Data Of the 140 patients with ccRCC 81 were male (58%) and 59 were female (42%). Median age was 65 years (IQR 57e71). Pathological tumor stage was T1 in 71 patients (51%), T2 in 19 (13%) and T3 in 50 (36%). Grade was G1, G2, G3 and G4 in 21 (15%), 66 (47%), 50 (36%) and 3 patients (2%), respectively. No patient had clinical or pathological evidence of lymph node involvement or metastatic disease. Median followup was 67 months (IQR 25e104) and followup data were available on 132 patients. During this observation period 34 patients (24%) died of disease. Immunostaining of HH Pathway Intermediates and Ki67 Visual analysis of immunostaining showed tissue expression of PTCH1 and Smo in 100% of patients, for SHH in 96% and for DHH in 74.6% (table 1). Each biomarker demonstrated a characteristic pattern of staining (nuclear vs cytoplasmic vs
0 14 0 14 10 1 0 8
(10.1) (19.7) (7.1) (0.7) (5.8)
Automated Analysis Spearman R2
Staining Pattern
0.78 0.78 0.40 0.67 0.70 0.70 0.70 0.74
Cytoplasmic Cytoplasmic Stromal Cytoplasmic Membranous þ cytoplasmic Cytoplasmic greater than nuclear Predominantly nuclear Predominantly membranous
membranous), consistent with previous publications (table 2 and fig. 1). The correlation of visually and automatically assessed staining intensities was strong for all biomarkers except IHH (table 1). HH Expression by ccRCC Grade and Stage Immunohistochemical expression of several HH biomarkers was strongly associated with tumor grade. Gli3 (2.2-fold, p ¼ 0.001), PTCH1 (1.5-fold, p ¼ 0.02), SHH (1.7-fold, p ¼ 0.006) and DHH (1.7fold, p ¼ 0.006) were more highly expressed in G3/G4 than in G1/G2 tumors (fig. 2). PTCH1, SHH, Gli1, Gli2, Gli3, IHH, DHH and Smo expression levels were not associated with advanced tumor stage (data not shown). There was a modest trend toward higher expression of the proliferation marker Ki67 in high grade and pT2/3 tumors (p ¼ 0.3 and 0.2, respectively). Univariate and Multivariate CSS Cox Regression Analysis On univariate Cox regression analysis DHH expression (p ¼ 0.0242), high grade tumors (p
Abbreviations and Acronyms ccRCC ¼ clear cell RCC CSS ¼ cancer specific survival DHH ¼ desert hedgehog Gli ¼ glioma-associated oncogene HH ¼ hedgehog IHH ¼ Indian hedgehog PTCH1 ¼ patched1 RCC ¼ renal cell carcinoma SHH ¼ sonic hedgehog Smo ¼ smoothened TMA ¼ tissue microarray Accepted for publication July 10, 2014. Study received institutional review board approval. * Equal study contribution. † Financial interest and/or other relationship with Merck-Serono and Charles River Laboratories. ‡ Equal study contribution. § Correspondence: Department of Urologic Sciences, University of British Columbia, Level 6, 2775 Laurel St., Vancouver, British Columbia V5Z 1M9, Canada (telephone: þ1 (604) 875-4818; FAX: þ1 (604) 875-5654; e-mail: pblack@mail. ubc.ca).
Purpose: Aberrant HH signaling has proved important in the pathogenesis of several solid cancers. Limited in vitro analyses suggested an oncogenic role for HH in renal cell carcinoma. In this explorative study we sought to validate aberrant HH expression in patients with renal cell carcinoma. Materials and Methods: A tissue microarray was constructed from 140 radical nephrectomy specimens of patients with clear cell renal cell carcinoma. We performed immunohistochemistry for Ki67 and HH pathway biomarkers, including PTCH1, Smo, SHH, IHH, DHH, Gli1, Gli2 and Gli3. Staining intensity was measured by automated image processing and related to tumor stage and grade. The impact of biomarker expression on cancer specific survival was determined by univariate and multivariate Cox regression analysis. Results: Gli3, PTCH1, DHH and SHH demonstrated markedly higher expression in high than in low grade tumors. Tumor stage was not associated with marker expression. On univariate analysis DHH expression, and tumor grade and stage were associated with cancer specific survival. Multivariate analysis revealed that DHH, grade and stage were independent predictors of cancer specific survival. Conclusions: To our knowledge we report for the first time that a biomarker of the HH pathway is associated with adverse pathological features and poor disease outcomes in patients with clear cell renal cell carcinoma. DHH may serve as an independent predictor of cancer specific survival in clear cell renal cell carcinoma cases. This supports further evaluation of HH signaling to validate the pathway as a target for novel therapy. Key Words: kidney; carcinoma, renal cell; hedgehog proteins; biological markers; mortality
KIDNEY cancer accounts for 3% to 4% of all solid tumors in developed countries.1,2 Until recently its incidence steadily increased about 2% per year.1 It was estimated that in 2013 there would be 65,150 incident cases and 13,680 deaths from kidney cancer in the United States.2 RCC represents 85% of all kidney malignancies,
1842
j
www.jurology.com
of which 80% are ccRCC.3 Surgery is the mainstay of treatment for localized RCC. Surgery is used in conjunction with systemic vascular endothelial growth factor or mammalian target of rapamycin targeted therapy and rarely interleukin-2 therapy in the metastatic setting.3 Despite curative intent metastasis
0022-5347/14/1926-1842/0 THE JOURNAL OF UROLOGY® © 2014 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARCH, INC.
http://dx.doi.org/10.1016/j.juro.2014.07.013 Vol. 192, 1842-1848, December 2014 Printed in U.S.A.
DHH IS PROGNOSTICATOR OF ONCOLOGIC OUTCOME IN CLEAR CELL RENAL CELL CANCER
develops in approximately 30% of patients with localized disease. At this stage only 10% of patients survive 5 years.4 Current gaps in the management of RCC cases include our inability to accurately predict the disease outcome and the pressing need for further novel therapies to improve outcomes.5 In the last decade multiple histopathological biomarkers have been tested for their predictive value for RCC. However, most markers that initially showed promising results could not be validated on multivariate analysis.6 Few biomarkers, such as vascular endothelial growth factor,6 the proliferation marker Ki67 and carbonic anhydrase IX, are independent prognosticators of RCC.7 To date none of these markers has entered routine clinical practice.3 The HH pathway is a key regulatory network during embryonic and postnatal development.8 It is activated by the binding of 1 HH ligand (SHH, IHH or DHH) to the transmembrane receptor PTCH1. PTCH1 binds and suppresses Smo in the absence of ligand but after ligand binding releases this suppression Smo can activate the Gli proteins Gli1, Gli2 and Gli3, which act as specific transcriptional factors.9 HH pathway reactivation was noted in different cancer entities such as ovarian, upper gastrointestinal tract and pancreatic cancer,10e12 including urological malignancies of the prostate and bladder.13,14 Some evidence suggests that the HH pathway may also have an important role in RCC.15 We evaluated expression levels of 9 intermediates of the HH pathway by immunohistochemistry in a cohort of patients who underwent nephrectomy for RCC. Staining intensity of these markers was related to common histopathological features and patient outcomes.
MATERIAL AND METHODS Patients and Specimens Tumor specimens were collected from 140 patients who underwent radical nephrectomy for ccRCC at the Department of Urology at University Medical Center of Johannes Gutenberg-University in Mainz, Germany between 1997 and 2002. Surgical specimens were formalin fixed and paraffin embedded, and examined by a genitourinary pathologist. Staging was determined according to the 7th edition of the AJCC (American Joint Committee on Cancer)/UICC TNM classification16 and grading according to the Fuhrman classification.17 Fuhrman grades 1 and 2 were considered low, and grades 3 and 4 were considered high. Corresponding clinical information on all patients was retrieved from the medical records. Followup was available in 132 patients. The study was approved by the institutional review board and patients provided informed consent.
1843
TMA Construction A TMA was created from formalin fixed, paraffin embedded tissue blocks using a manual tissue microarrayer (Beecher Instruments, Sun Prairie, Wisconsin). Each marked block was sampled twice with a core diameter of 1.0 mm and arrayed in a rectangular pattern with 1.6 mm between the center of each core to create a duplicate TMA layout.
Antibodies Specific antibodies of intermediates of the HH pathway were chosen based on previous publications. Polyclonal antibodies against SHH (1:25, N-19, sc-1194), DHH (1:100, N-19, sc-1193), Smo (1:10, C-17, sc-6367), PTCH1 (1:100, G-19, sc-6149), Gli1 (1:25, H-300, sc-20687) and Gli3 (1:200, H280, sc-20688) were obtained from Santa Cruz Biotechnology, Santa Cruz, California. Monoclonal antibody against IHH (1:100, ab52919) and polyclonal antibody against Gli2 (1:50, ab26056) were obtained from AbcamÒ. Anti-Ki67 (1:150, RM-9106) was obtained from Dako, Burlington, Ontario, Canada.
Histology and Immunohistochemistry TMAs were cut into 4 mm sections. Staining was performed with specific primary antibodies and corresponding secondary antibodies using the Discovery XT autostainer (VentanaÒ) using an enzyme labeled biotin streptavidin system and a solvent resistant 3,30 -diaminobenzidine Map kit.
Immunostaining Evaluation All slides were scored categorically as 0, 1þ, 2þ or 3þ by the institutional pathologist (LF). They were digitized with the SL801 autoloader system and SCN400 scanning system (Leica Microsystems, Concord, Ontario, Canada) at 20 magnification. Images were stored in the SlidePath Digital Image Hub (Leica Microsystems) at Vancouver Prostate Centre. Automated tissue analysis was subsequently performed according to previously described algorithms.18 For each biomarker the staining pattern (nuclear vs cytoplasmic vs membranous) was evaluated. Representative cores (clearly positive, clearly negative and mixed positive/negative) were manually identified to create color segmentation. Tissue image analysis capabilities (Leica Microsystems) with established algorithms were adjusted using these references and morphometric parameters. Cells that stained for each biomarker in each TMA core were subsequently measured by positive pixel identification using the tandem platform tissue image analysis and Digital Image Hub. All results were stored in the database and linked to patient information. Each tumor was represented by duplicate cores so that the mean score of the 2 cores was used for analysis.
Statistical Analysis Results of visual (categorical) and automated (continuous) analysis of staining intensity were analyzed by the Spearman correlation test. All automated analysis results are shown as the median and IQR. Groups were compared using the Mann-Whitney test with differences considered nonrandom at p 0.05. CSS depending on DHH expression was estimated by the Kaplan-Meier method using median DHH expression (307 relative units) as the cutoff.
1844
DHH IS PROGNOSTICATOR OF ONCOLOGIC OUTCOME IN CLEAR CELL RENAL CELL CANCER
Table 1. Pathological specimen tissue expression and staining pattern of HH pathway intermediates No. Tissue Expression (%) Biomarker DHH SHH IHH Smo PTCH1 Gli1 Gli2 Gli3
0
No. Pts 138 139 135 137 140 138 138 137
35 6 49 0 0 29 25 10
(25.4) (4.3) (36.3) (21.0) (18.1) (7.3)
1þ 83 66 67 36 86 86 93 62
(60.1) (47.5) (49.6) (26.3) (61.4) (62.3) (67.4) (45.3)
2þ 20 53 19 74 44 22 20 57
3þ
(14.5) (38.1) (14.1) (54.0) (31.4) (15.9) (14.5) (41.6)
Univariate Cox regression analysis and stepwise backward multivariate Cox regression analysis (enter if p 0.1) were performed for CSS using MedCalc (MedCalc Software, Ostend, Belgium). The Harrell c-index was calculated using the R-function coxph. The c-index is a generalization of the AUC measure to censored data that measures how well the fitted Cox proportional hazard model distinguishes between patients who die and those who survive.
RESULTS Patient Cohort Clinical Data Of the 140 patients with ccRCC 81 were male (58%) and 59 were female (42%). Median age was 65 years (IQR 57e71). Pathological tumor stage was T1 in 71 patients (51%), T2 in 19 (13%) and T3 in 50 (36%). Grade was G1, G2, G3 and G4 in 21 (15%), 66 (47%), 50 (36%) and 3 patients (2%), respectively. No patient had clinical or pathological evidence of lymph node involvement or metastatic disease. Median followup was 67 months (IQR 25e104) and followup data were available on 132 patients. During this observation period 34 patients (24%) died of disease. Immunostaining of HH Pathway Intermediates and Ki67 Visual analysis of immunostaining showed tissue expression of PTCH1 and Smo in 100% of patients, for SHH in 96% and for DHH in 74.6% (table 1). Each biomarker demonstrated a characteristic pattern of staining (nuclear vs cytoplasmic vs
0 14 0 14 10 1 0 8
(10.1) (19.7) (7.1) (0.7) (5.8)
Automated Analysis Spearman R2
Staining Pattern
0.78 0.78 0.40 0.67 0.70 0.70 0.70 0.74
Cytoplasmic Cytoplasmic Stromal Cytoplasmic Membranous þ cytoplasmic Cytoplasmic greater than nuclear Predominantly nuclear Predominantly membranous
membranous), consistent with previous publications (table 2 and fig. 1). The correlation of visually and automatically assessed staining intensities was strong for all biomarkers except IHH (table 1). HH Expression by ccRCC Grade and Stage Immunohistochemical expression of several HH biomarkers was strongly associated with tumor grade. Gli3 (2.2-fold, p ¼ 0.001), PTCH1 (1.5-fold, p ¼ 0.02), SHH (1.7-fold, p ¼ 0.006) and DHH (1.7fold, p ¼ 0.006) were more highly expressed in G3/G4 than in G1/G2 tumors (fig. 2). PTCH1, SHH, Gli1, Gli2, Gli3, IHH, DHH and Smo expression levels were not associated with advanced tumor stage (data not shown). There was a modest trend toward higher expression of the proliferation marker Ki67 in high grade and pT2/3 tumors (p ¼ 0.3 and 0.2, respectively). Univariate and Multivariate CSS Cox Regression Analysis On univariate Cox regression analysis DHH expression (p ¼ 0.0242), high grade tumors (p
