European Journal of Cancer (2014) 50, 1531–1540

Available at www.sciencedirect.com

ScienceDirect journal homepage: www.ejcancer.com

Poor prognosis and advanced clinicopathological features of clear cell renal cell carcinoma (ccRCC) are associated with cytoplasmic subcellular localisation of Hypoxia inducible factor-2a Nils Kroeger a,b,1, David B. Seligson c,1, Sabina Signoretti d, Hong Yu c, Clara E. Magyar c, Jiaoti Huang c, Arie S. Belldegrun a, Allan J. Pantuck a,⇑ a Institute of Urologic Oncology, Department of Urology, David Geffen School of Medicine, University of California-Los Angeles, Los Angeles, CA, USA b Department of Urology, University Medicine Greifswald, Germany c Department of Pathology and Laboratory Medicine, David Geffen School of Medicine, University of California-Los Angeles, Los Angeles,CA, USA d Department of Pathology, Dana Farber Cancer Institute, Harvard School of Medicine, Boston, MA, USA

Available online 21 February 2014

KEYWORDS Renal cell carcinoma CA-IX VEGF mTOR HIF-1 HIF-2 Tumour suppressor Tumour promotor Cancer specific survival

Abstract Background: Pre-clinical studies have implicated hypoxia inducible factor (HIF)-2a as an important oncogene for clear cell renal cell carcinoma (ccRCC). Generally considered to act as a nuclear transcription factor, a recent study has also implicated HIF-2a as a protein translational initiation complex function within the cytoplasm (Uniacke et al., 2012). We hypothesised that both the absolute expression as well as the sub-cellular localisation of HIF-2a would predict clinicopathological features and cancer specific survival (CSS) in ccRCC. Methods: A tissue microarray (TMA) study was conducted on three hundred and eight ccRCC patients. Survival differences were investigated with the log rank test and associations with CSS with uni- and multivariate Cox regression analyses. Recursive partition tree analysis was used to identify relevant cutoff values. Results: High HIF-2a nuclear (N) (cutoff >32%) expression was associated with smaller tumour sizes (p = 0.002) and lower Fuhrman grades (p = 0.044), whereas tumours with high cytoplasmic (C) HIF-2a (>0%) more often had positive lymph nodes (p = 0.004), distant metastases

⇑ Corresponding author: Address: Institute of Urologic Oncology, Department of Urology, David Geffen School of Medicine at UCLA, 924 Westwood Boulevard, Suite 1050, Los Angeles, CA 90095-7384, USA. Tel.: +1 310 206 2436; fax: +1 310 794 3513. E-mail address: [email protected] (A.J. Pantuck). 1 NK and DBS have contributed equally to this work.

http://dx.doi.org/10.1016/j.ejca.2014.01.031 0959-8049/Ó 2014 Elsevier Ltd. All rights reserved.

1532

N. Kroeger et al. / European Journal of Cancer 50 (2014) 1531–1540

(p = 0.021) and higher Fuhrman grades (p < 0.0001). After adjustment for TNM stage, Eastern Cooperative Oncology Group performance status (ECOG PS), and Fuhrman grade, both continuous (p < 0.0001) and dichotomised (p < 0.0001) HIF-2a C variables remained significant predictors of CSS, while neither HIF-2a N variable was retained. Conclusion: Our investigation supports that HIF-2a may have a unique tumour promoter role in the cytoplasm. This preliminary finding justifies further experimental and mechanistic studies that examine the biological functions of HIF-2a when located in the cytoplasm. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction

2. Patients and methods

The hypoxia inducible factor (HIF) is a heterodimeric transcription factor consisting of an a- and a b-subunit. HIF-a is expressed and then degraded by the von Hippel Lindau (VHL)–ubiquitin ligase complex under physiological conditions. The complex of HIF-a and ARNT (HIF-1b) binds to co-activators (p300/CBP; PKM2) and the promoters of HIF-responsive genes leading to the transcription of proteins that are important for adaption to cellular hypoxia (e.g. VEGF, IGF, EPO, GLUT-1, and CAIX) [1]. The HIF-a family contains at least three different subunits with HIF-1a and HIF-2a being the best investigated members. Clear cell renal cell carcinomas (ccRCC) that have an inactivated VHL gene express either HIF-2a alone or both HIF-1a and HIF-2a [2]. In recent years, numerous pre-clinical investigations have suggested that HIF-2a rather than HIF-1a is the main tumour promoter in ccRCC [3–5]. Much effort has been spent to identify the key regulators that are responsible for the tumour promoter effects of HIF-2a. However, the full mode of action of HIF-2a has not yet been fully characterised. In addition to serving as a well-characterised transcription factor, a recent study has implicated HIF-2a as a member of a protein initiation complex [6]. Authors have shown that protein expression under low oxygen conditions is initiated by a complex that contains HIF-2a, mRNA-binding protein 4 (RBM4) and eukaryotic translation ignition factor 4E type 2 (eIF4E2). This complex assembles on a hypoxia responsive element (rHRE) of messenger ribonucleic acid (mRNA) and targets mRNAs to polysomes for active translation. Yet, relatively few clinical studies have provided evidence for the crucial role of HIF-2a in ccRCC tumour progression, and therefore, the translational relevance of HIF-2a as a tumour promoter in ccRCC currently remains uncertain. Protein translation is a biological process that occurs in the cytoplasm and transcription a process that occurs in the nucleus. Therefore, we hypothesised that both the absolute expression as well as the sub-cellular localisation of HIF-2a would predict clinicopathological features and cancer specific survival (CSS) in ccRCC.

2.1. Patients The study cohort was comprised of 308 ccRCC patients who were treated at the University of California Los Angeles (UCLA) between 1989 and 2000. Clinical and pathologic data were retrospectively gathered from the UCLA Institutional Review Board (IRB) approved kidney cancer database and electronic charts. Clinicopathological data included age, gender, Eastern Cooperative Oncology Group performance status (ECOG PS) [7], TNM stages [8] and Fuhrman grades [9]. 2.2. Immunohistochemistry The sections were deparaffinised, rehydrated and heated in a pressure cooker to 125 °C for 30 s in EDTA for antigen retrieval. After cooling to room temperature, sections were incubated in 3% hydrogen peroxide (Dako, Carpinteria, CA) for 5 min to quench endogenous peroxidase. Sections were then incubated in avidin block for 15 min followed by incubation in biotin block for 15 min (Vector, Burlingame, CA). The sections were then incubated with serum-free protein block for 10 min (Dako, Carpinteria, CA). The primary antibody (clone UP15, kindly provided by Dr. William G. Kaelin, Dana-Farber Cancer Institute) was applied to sections for 1 h at 1:15,000 dilution. Detection was performed by incubation with Dako EnVision+ System HRP labelled polymer (Dako, Cat# K4003 and Cat# K4001) for 30 min followed by incubation with Biotin labelled tyramide (Perkin-Elmer, Cat #SAT700001EA) at a 1:50 dilution for 10 min. The slides were then incubated with LSAB2 Streptavidin-HRP (Dako, Cat # K1016) for 30 min. DAB chromogen (Dako, Cat # K3468) was then applied, and the slides were slightly counterstained with haematoxylin. Formalin-fixed paraffin-embedded cells with high (786-O-vector) or low (786-O-VHL) HIF2-alpha levels were utilised as positive and negative controls, respectively, to validate the specificity of the immunoassay. The immunohistochemical methods and the primary antibodies of other markers investigated in this study were described previously [10].

N. Kroeger et al. / European Journal of Cancer 50 (2014) 1531–1540

2.3. Statistical analyses and identification of cutoff values Continuous variables were reported as means (±standard deviation [SD]) and medians (Inter Quartile Range [IQR]). All proteins investigated as well as age and tumour-sizes were not normally distributed by Kolmogorov–Smirnov test. Therefore, the Mann Whitney U test was used for comparisons of continuous variable pairs and Spearman’s Rho for bivariate correlations. Pearson’s Chi-square and Fisher’s exact tests were used for comparison of categorical variables. The primary study end-point was the association of HIF-2a expression with clinicopathological features and CSS, which was calculated from the date of surgery to the date of death or last follow. The Kaplan Meier method was used for survival estimations and differences between groups were compared with the log rank test. The uni- and multivariable associations of clinicopathological variables with CSS were examined with Cox regression analyses and uni- and multivariable comparisons were performed with the Wald Chi-square test. In multivariable analyses, a backward-stepwise selection with the likelihood ratio criterion (including and exclusion criteria were p < 0.05 and p > 0.10, respectively) was used. Schoenfeld’s global test was used to test the proportional hazards assumption in the Cox models. All tests were performed two-sided to assign statistical significance to p-values 60.05. The statistical software packages IBM SPSS Statistics 19Ò (Chicago, IL) and R4 (www.r-project.org/) were used for all analyses. Two experienced pathologists (D.B.S. and H.Y.) who were blinded to the clinical outcome performed the evaluation of the staining. The sub-cellular localisation and the frequency of positive nuclei (yes versus no) and cytoplasm (yes versus no) of all cells at each spot was graded. The pooled mean of the percentage of positive cells from the three spots of the same tumour case was used for the statistical analyses. In order to identify univariable cutoff values, recursive partition tree analyses of the Kaplan Meier curves were carried out. This statistical method identified an expression of >32% HIF-2 nucleus and >0% HIF-2 cytoplasm as appropriate cutoff values to allow univariate comparisons. We did not perform an external validation of these results.

3. Results 3.1. Evaluation of HIF expression in tumour and metastatic tissue Representative figures of HIF-2a cytoplasm and nucleus are displayed in Fig. 1A–D. HIF-2 a nuclear expression was found in 290/300 (97%) evaluable tumours and HIF-2a cytoplasmic expression was found in 60/299 (20%) tumours. Of the tumours that expressed HIF-2a in the cytoplasm, 58/60 (97%) tumours had a

1533

concomitant HIF-2a expression in the nucleus. Overall, the mean expression was considerably higher in the nuclear than in the cytoplasmic compartment (Fig. 2A). We additionally evaluated the expression of HIF-1a and HIF-2a in 16 metastases whose corresponding primary tumours were part of our study cohort. There was no significant difference in the expression of primary tumours and in their corresponding metastatic sites for either HIF-1a or HIF-2a p = 0.624 (HIF-1a nuclear), p = 0.859 (HIF-1a cytoplasm), p = 0.803 (HIF-2a nuclear), p = 0.722 (HIF-2a cytoplasm) (Fig. 2B). Nuclear expression of HIF-2a was found in almost all corresponding healthy tissue spots, and displayed nuclear protein expression that was predominantly seen in renal epithelial tissues, including normal tubules and glomerular parenchyma, but was only very rarely appreciated in stromal or endothelial tissues. The expression frequency was of HIF-2 nuclear was comparable with the tumour tissue while cytoplasmic HIF-2a was only found in half of the healthy tissue spots and the expression frequency and intensity were overall lower than that seen in tumour tissues. Unfortunately, we are not able to provide statistical evaluations of the healthy tissue because our TMA includes just one corresponding spot of healthy tissue for each tumour case. Furthermore, the staining frequency varied in different anatomical compartments of healthy epithelia (e.g. distal or proximal tubules, collecting ducts), therefore the generation of a single value is complicated by the proportions of each nephron part in each sample of normal tissue. A thorough comparison would make it necessary to examine all anatomical components separately, which with respect to the low number of tissue spots would be impossible as well as beyond the focus of this manuscript.

3.2. Association of HIF expression with survival outcome and clinicopathological features At the time of analysis, 141 (45.8%) patients had died after a median follow up of 2.3 years (IQR 4.2 years) from ccRCC. We did a comparison of clinicopathological features based on the identified cutoff values (Table 1). Tumours with versus without high HIF-2a nuclear expression had smaller tumour sizes (p = 0.002), lower T stages (p = 0.011) and less advanced Fuhrman grades (p = 0.044). Cytoplasmic HIF-2a expression was associated with more frequent positive lymph nodes (p = 0.004), visceral metastases (p = 0.021) and advanced Fuhrman grade (p < 0.0001). Any expression of cytoplasmic HIF-2a was associated with worse survival outcome (p < 0.0001; Fig. 3A). In localised ccRCC, there were only a small number of death events. Therefore, analyses are not meaningful in this sub-group. In metastatic tumours,

1534

N. Kroeger et al. / European Journal of Cancer 50 (2014) 1531–1540

Fig. 1. Expression of nuclear and cytoplasmic HIF-2a. Shown are representative figures of HIF-2a nuclear (panels A and B) and HIF-2a cytoplasmic (panels C and D) expression in ccRCC (Magnification of A and C 10 and 40 of B and D).

Fig. 2. Expression of nuclear and cytoplasmic HIF-2a in primary tumours and corresponding metastatic sites. All hypoxia inducible factors (HIF) were similarly expressed in the cytoplasmic and nuclear compartments (p = 0.624 (HIF-1a nuclear), p = 0.859 (HIF-1a cytoplasm), p = 0.803 (HIF-2a nuclear), p = 0.722 (HIF-2a cytoplasm).

cytoplasmic HIF-2a expression was significantly related to worse survival outcome (median (95% confidence interval (CI)): 0.87 years (0.6–1.2 years) versus 2.12 years (1.2–3.0 years); p < 0.0001). High HIF-2a expression in the nucleus was associated with longer CSS (p = 0.032, Fig. 3B). Metastatic tumours with versus without high HIF-2a nuclear expression had similar CSS (median (95% CI): 1.8 (0.6–3.0) years versus 1.4 (0.7–2.0) years); p = 0.514). Again, analyses of the localised group were not possible due to the low number of death events.

The complete univariate Cox regression analyses are shown in Table 2. Nuclear HIF-2 a expression was only significant when analysed in a dichotomised fashion (p = 0.035). The expression of cytoplasmic HIF-2a, was a significant predictor of CSS when analysed as continuous (p < 0.0001) or dichotomised variable (p < 0.0001) variable. While high HIF-2a nuclear expression lost its significance in multivariable analysis (p = 0.978), HIF-2a cytoplasmic expression was independently associated with CSS as continuous (p < 0.0001; Table 2B) and

N. Kroeger et al. / European Journal of Cancer 50 (2014) 1531–1540

1535

Table 1 Patient and tumour characteristics according to subcellular localisation of HIF-2a. Features

HIF-2a nuclear

HIF-2 a cytoplasm

Low expression 32%

p-value

HIF-2a cyto No

HIF-2a cyto Yes

p-value

Age (y) median (IQR) Size (cm) median (IQR)

61 (14) 7.00 (5)

61 (19) 5.20 (15)

0.638 0.002

61 (16) 7 (17)

59 (16) 7.5 (16)

0.732 0.352

Gender Male Female

155 (66.2) 79 (33.8)

44 (67.7) 21 (32.3)

0.883

164 (68.6) 75 (31.4)

35 (58.3) 25 (41.7)

0.168

ECOG PS 0 1 2 3

76 (32.8) 143 (61.6) 12 (5.2) 1 (0.4)

29 (45.3) 34 (53.1) 1 (1.6) 0 (0.0)

91 (38.6) 135 (57.2) 9 (3.8) 1 (0.4)

14 (23.3) 42 (70.0) 4 (6.7) 0 (0.0)

T stage T1 T2 T3 T4

69 (29.5) 34 (14.5) 114 (48.7) 17 (7.3)

33 (50.8) 5 (7.7) 25 (38.4) 2 (3.1)

85 (35.6) 32 (13.4) 109 (45.6) 13 (5.4)

17 (28.3) 7 (11.7) 30 (50.0) 6 (10.0)

N stage N0 N1 N2

202 (86.3) 12 (5.1) 20 (8.6)

60 (92.3) 2 (3.1) 3 (4.6)

0.430

217 (90.8) 8 (3.3) 14 (5.9)

45 (75.0) 6 (10.0) 9 (15.0)

M stage M0 M1

120 (51.3) 114 (48.7)

40 (61.5) 25 (38.5)

0.161

136 (56.9) 103 (43.1)

24 (40.0) 36 (60.0)

Fuhrman grade low versus high (G1/2 versus G3/4)

135 (57.7) 99 (42.3)

47 (72.3) 18 (27.7)

162 (67.8) 77 (32.2)

20 (33.3) 40 (66.7)

Sarcomatoid Yes No

215 (91.9) 19 (8.1)

63 (96.2) 2 (3.8)

11 (4.6%) 228 (95.4%)

10 (16.7%) 50 (83.3%)

0.203

0.011

0.044 0.269

0.134

0.455

0.004

0.021

Poor prognosis and advanced clinicopathological features of clear cell renal cell carcinoma (ccRCC) are associated with cytoplasmic subcellular localisation of Hypoxia inducible factor-2α.

Pre-clinical studies have implicated hypoxia inducible factor (HIF)-2α as an important oncogene for clear cell renal cell carcinoma (ccRCC). Generally...
2MB Sizes 0 Downloads 3 Views