Activated extracellular signal-regulated kinase is an independent prognostic factor in clinically confined renal cell carcinoma

Authors


Abstract

BACKGROUND:

Extracellular signal-regulated kinase (ERK) promotes proliferation, metastasis, and poor survival in cancers of the breast, lung, and liver. Advanced localized renal cell carcinoma (RCC) is extraordinarily treatment resistant and has high recurrence rates despite surgery. Limited data exist regarding the prognostic significance of activated (phosphorylated) ERK in RCC. The authors hypothesized that activated ERK (pERK) promotes disease progression and metastasis in localized RCC and may be of value as a biomarker to predict disease recurrence.

METHODS:

The expression profile of pERK was examined by immunocytochemistry using a tissue microarray constructed from 174 drug treatment–naive patients who had undergone radical nephrectomy for localized RCC. Levels of tumor-cell specific pERK were scored and correlated with clinicopathologic parameters of RCC and disease-free survival.

RESULTS:

Immunostaining for pERK was present in 36% of all RCCs, with a predominance found in the clear cell histologic subtype. High expression was associated with increased tumor size, increased TNM stage, and vascular invasion. Patients with pERK-positive tumors had a mean disease-free survival of 4.19 years, compared with 6.38 years for patients with pERK-negative tumors (P < .001). Cox regression models revealed pERK to be a significant independent predictor of disease-free survival, with a hazards score of 2.9 (P < .001), a value similar to tumor grade (hazards ratio, 3.01; P < .001).

CONCLUSIONS:

Expression of pERK is an independent prognostic factor in RCC that is associated with advanced and aggressive pathologic features of renal tumors and predicts the onset of metastasis in patients with localized disease. Cancer 2009. © 2009 American Cancer Society.

Renal cell carcinoma (RCC) is a heterogeneous group of highly vascularized invasive kidney tumors, comprising clear cell or conventional (accounting for 60%-80% of all RCC cases), papillary, chromophobe, and collecting duct subtypes.1 Currently, RCC represents 3% of all cancers and 2% of all cancer-related deaths and is also 1 of the few cancers whose incidence and mortality rates are growing steadily worldwide.2, 3 Approximately one-third of all RCC patients will have established metastatic disease at first presentation; the remainder with clinically localized tumors usually undergo curative radical nephrectomy. Approximately 35% to 45% of these will also eventually develop disease recurrence with metastasis at distant sites.2, 3 However, the clinical course of those presenting with localized disease is difficult to predict, even within patients who have similar clinicopathologic parameters. This reflects intrinsic differences in the molecular and biologic profiles of individual tumors.4

Historically, survival rates for metastatic RCC rarely exceed 5%, resulting from the inherently refractive nature of tumors to standard chemotherapy, radiotherapy, and hormonal therapy.5 The last decade, however, has witnessed an explosion in the understanding of the biologic behavior of renal neoplasms and the identification of how key signaling cascades and their molecular determinants integrate to drive disease progression. This preclinical evidence has led to the emergence of rationally designed novel drugs that specifically target these aberrant signaling pathways.6-8 The most developed targeted therapies currently in use for the treatment of advanced localized and metastatic RCC are sorafenib and sunitinib, 2 orally active multikinase inhibitors, and temsirolimus, an inhibitor of the mammalian target of rapamycin (mTOR).9 Indeed, temsirolimus, when used as a single agent, is shown to significantly improve overall survival of patients with metastatic cancer compared with standard therapies for RCC.10 Despite these advances, challenges within the field still exist and include among others the need to 1) optimize the use of such targeted therapies by identifying patients with high-risk disease who are most likely to benefit from their use as single agents or in combination, and 2) evaluate the efficacy of targeted therapies in nonclear cell RCC subgroups for which no specific trials have been reported.11 Therefore the utility of tumor biomarkers that not only identify high-risk patients but also have value in guiding treatment decisions for RCC is increasingly realized.4, 12-14

The epidermal growth factor receptor (EGF-R)/Ras/extracellular signal-regulated kinase (ERK) axis is shown to be mutationally activated or overexpressed in a significant number (>30%) of all cancers.15 Once activated (phosphorylated), the terminal protein kinases, ERK-1/2, are translocated to the nucleus, whereupon an extensive array of transcription factors are activated, including among others Elk-1, Fos, and Stat-3.16 These regulatory mechanisms lead to global changes in gene expression profiles that mainly, but not unequivocally, result in promotion of cell growth. Aberrant ERK signaling is associated with other cancer-related processes, namely tumor angiogenesis, cell migration and invasion, and drug resistance.17, 18 Together this has led to intense pharmaceutical interest in developing tyrosine kinase inhibitors that target signal transduction cascades such as the EGF-R/Ras/ERK and PI-3K/AKT/mTOR cascades.19, 20

Much of the current literature that lends support for ERK-1/2 serving as a cancer driver has been generated by the use of in vitro, ex vivo, and mouse cancer models, with an actual paucity of studies that report its expression, activity, and significance in clinical tumor specimens.21 Of the clinical immunohistologic studies undertaken to date, phosphorylated ERK (pERK) has been shown to correlate with disease progression in melanoma,22 hepatocellular,23 and nonsmall cell lung carcinomas.24 In hepatocellular carcinoma, pERK is predominantly expressed in poorly differentiated tumors, and as such serves as a powerful prognostic factor.23 In this latter study, pERK also was found to be correlated with the presence of intrahepatic metastasis, and could retrospectively predict disease recurrence in patients after the complete surgical removal of initial tumors.23 Trends of increasing levels of pERK have also been observed in late-stage disease in both lung carcinoma23 and melanoma,22 but it is not an independent prognostic factor in either disease when analyzed for patient survival. Conversely, the expression of pERK-1/2 in endometrial cancer25 has been shown to be an independent prognostic indicator associated with a favorable prognosis. Specifically, patients whose tumors exhibited low levels of pERK-1/2 actually had significantly lower disease-free survival and worse overall survival.25 A dichotomy in the impact of activated ERK also exists within the same cancer type. For example, in some breast cancer patient cohorts, high activated ERK-1/2 levels are reportedly associated with early stage tumors and negative lymph node status, and predict long disease-free survival.21 In others, its high expression is reported to correlate with advanced localized or metastatic disease, tamoxifen resistance, and reduced time to disease recurrence.26 These apparent contradictions may reflect the disease stage and context dependency of ERK-1/2. In support of this notion, in vitro studies have shown that variations in the magnitude and spatiotemporal expression of pERK-1/2 can either potentiate proliferation or alternately drive cellular differentiation and senescence.27

The prognostic significance of activated ERK-1/2 in RCC remains undetermined. To test the hypothesis that constitutively activated ERK-1/2 may potentiate disease progression in RCC, we have examined, using tissue microarray technology, the expression of pERK-1/2 in 176 RCCs resected from patients who had undergone radical nephrectomy for the treatment of localized disease. The relative expression of activated ERK-1/2 was correlated with clinical and pathologic variables of RCC, with the overall aim of determining whether pERK-1/2 might have prognostic value in RCC patients.

MATERIALS AND METHODS

Patient Selection

Archival tumor specimens were identified from the pathologic records and consisted of a consecutive series of 174 patients who had undergone curative radical nephrectomy for primary kidney tumors presenting between 1992 and 1996. This series of patients (for which ethical approval was granted by the National Health Service local research ethics committee) has been previously described elsewhere.28 Tissue blocks, histology reports, and slides were available in all cases without a priori knowledge of clinical outcome. For each renal carcinoma, a tissue block was selected that contained a sample of peripheral tumor for making the tissue microarray (TMA). Tumors were classified according to: the Hiedelberg system29 and Furhman nuclear grade30; the presence or absence of any vascular invasion (either microvascular invasion, renal vein invasion, or inferior vena cava invasion)31; and whether or not there was capsular invasion with cellular invasion of perinephric fat.32 No patients had received drug treatment or had evidence of lymph node or distant metastatic disease before or at surgery.

The median age of the 174 patients was 65 years (range, 34 years-88 years); 119 were men and 55 were women. Complete clinical follow-up was performed as previously described.31, 33 Briefly, patients had been reviewed annually as out-patients for 3 to 7 years (mean follow-up, 4.3 years; median follow-up, 3.92 years [range, 0.3 years-8.09 years]) with the following information extracted from their records: date of birth, sex, date of surgery, date patient last seen, date on which recurrent or metastatic disease was first identified, and date and cause of death. For patients in whom the cause of death was recorded as RCC but whose time of recurrence was unavailable, the date of death was considered the endpoint for disease-free survival.

TMA Construction and Immunocytochemistry

The customized TMA was constructed with a Beechers manual tissue arrayer (Beechers Instruments, Inc, Sun Prairie, Wis) from archived paraffin-embedded renal tumor samples. A single core of representative peripheral tumor (0.6 mm in diameter) was punched from each donor block, using a specific orientation transplanted into a premolded recipient paraffin wax block. Peripheral tumor refers to representative viable tumor tissue taken within 5 mm of the growing edge of the tumor. Additional cores were taken from normal renal tissue (adjacent to some of the tumors). Serial sections were cut from the resulting TMA block onto cleaned adhesive glass slides (Superfrost Plus, Microm International, Walldorf, Germany) at 4 μm thickness.

Array sections were dewaxed using a sequential series of graded xylenes and alcohols. The detection of pERK- 1/2 was undertaken using previously published procedures.26 Briefly, after removal of paraffin wax, the endogenous peroxidase activity within the rehydrated tissue was quenched (3% hydrogen peroxide for 5 minutes). Antigen retrieval consisted of microwaving TMA sections in citric acid (0.1 M, pH 6.0) for 30 minutes, followed by slide cooling with running tap water. After draining, the sections were equilibrated in 20% normal human serum (Golden West Biologicals Inc, Temecula, Calif) for 15 minutes at room temperature (diluent, phosphate-buffered saline [PBS]). The primary rabbit antihuman pERK-1/2 antibody (New England Biolabs, Hitchin, UK) was applied to each section at a dilution of 1:25 and incubated overnight at 4°C for a total of 16 hours. Thereafter, the sections were washed (4 × 1 minutes) with PBS (pH 7.3), and tissue was immunostained using the DAKO rabbit Envision staining system (DAKO, Cambridge, UK) according to the manufacturer's instructions. The TMA sections were counterstained with hematoxylin and finally mounted.

Controls and Scoring of Stained Specimens

Human breast cancer specimens, known to be positive for pERK-1/2, were run in parallel and served as an appropriate control for evaluation of staining in the kidney tissue. Negative controls consisted of RCC TMA and breast cancer sections in which the primary antibody had been omitted and replaced with 20% normal human serum (isotypic control).

Scoring of tumor arrays was performed by a pathologist (D.G.) and research associates (L.C. and R.N.) without knowledge of other pathologic and clinical data. Expression of pERK-1/2 was assessed semiquantitatively accordingly to previously described criteria,28, 33 with slight modifications. This scoring system accounts for both the intensity of immunostain within the nucleus and cytoplasm of tumor cells and the percentage of tumor cells involved in each core. Scoring was as follows: 0 indicates no detectable reaction product (deposit) in tumor cells; 1 indicates very light diffuse or focal light deposit in tumor cells; 2 indicates light diffuse or moderate focal deposit (may include very small areas of heavy deposit); and 3 indicates tumor cores containing areas of heavy deposit in most or all tumor cells.

Data and Statistical Analysis

Analysis of disease-free survival of patients with tumors demonstrating different scores of staining for pERK-1/2 was performed using the Kaplan-Meier method using the log-rank test, in which the first appearance of metastasis was considered an event. Patients last seen alive without metastasis or who died because of causes other than RCC were considered censored at the date of last contact or death, respectively. Scores were also converted to a binary simple covariate designated positive or negative by thresholding according to the most informative split on the initial Kaplan-Meier analysis. For pERK-1/2, a score of 0 was negative, and a score of 1 or 2 or 3 was positive. The association of positive pERK- 1/2 expression with recognized histologic tumor prognostic variables (grade, size, vascular invasion, capsular invasion, and tumor type) was examined by cross-tabulation and the chi-square test.

Multivariate analysis by Cox regression was performed to determine whether positive pERK-1/2 expression had influence on prediction of disease outcome in relation to other clinical prognosticators for RCC. Previous analysis had already determined that the most influential covariates predicting disease-free survival of these patients are Fuhrman grade (grades 1 and 2 and grades 3 and 4, respectively, are pooled for analysis), any degree of vascular invasion (histological correlate of stage T2b), and invasion of perinephric tissues (ie, capsular invasion and histological correlate of stage T2a). When these covariates are taken into consideration, tumor size and type were found to have no influence on disease-free survival. Therefore, in this current computation, pERK-1/2 was entered into the multivariate analysis as an independent covariate, using the enter function together with the covariates grade, presence or absence of vascular invasion, and invasion of perinephric tissue.

RESULTS

Immunohistochemistry for Activated ERK- 1/2 in Clinically Confined RCC

Of the 174 cases, 164 were assessable on the array; of these, 60 (36%) stained positive for pERK-1/2. Immunoreactivity for pERK was detected in both the nucleus and cytoplasm of tumor cells. In all cases examined, some degree of cytoplasmic staining coexisted with nuclear staining (Fig. 1). Cases that scored 1, in which staining was predominantly light and diffuse, comprised 50% (30 of 60) of all pERK-positive cases. Moderate staining (score 2) and more extensive heavy immunoreactivity (score 3) accounted for 35% (21 of 60) and 15% (9 of 60) of cases, respectively (Fig. 1).

Figure 1.

This is a representative tissue microarray of clinically confined renal cell carcinoma demonstrating the immunohistochemical expression intensity and pattern of phosphorylated extracellular signal-regulated kinase in tumors that were typically (A) score 0 (negative), (B) score 1, (C) score 2, and (D) score 3.

Cross-tabulation of pERK-1/2 with conventional determinants for RCC is shown in Table 1. Increased expression levels of pERK-1/2 was found to be correlated with tumor size (P = .039), TNM staging (P = .013), and vascular invasion (P = .022). No correlation was observed between tumor grade and capsular invasion. Notably increased expression of pERK-1/2 was found to be significantly associated (P = .006) with nonpapillary tumors (of which 98% were of the clear cell type) when compared with the papillary subtype. It is interesting to note that only 25 of 104 (24%) of patients with pERK-negative tumors developed disease recurrence, compared with 40% (12 of 30), 57% (12 of 21), and 56% (5 of 9) of patients whose tumors had pERK positive scores of 1, 2, and 3, respectively; note that pERK positive-tumors accounted for 53% (29 of 54) of the total number of recurrences within this patient cohort (Table 1).

Table 1. Relation Between pERK-1/2 Expression Score and Conventional Clinicopathologic Parameters in Patients With Clinically Confined RCC*
Prognostic IndicesPhosphorylated ERK-1/2Recurrence, n (%)
0, n=1041, n=302, n=213, n=9Total, n=164
  • pERK indicates phosphorylated extracellular signal-regulated kinase; RCC, renal cell carcinoma.

  • *

    For information purposes, the percentage of disease recurrence in each category is included.

  • For contingency tables of the individual prognostic and pERK-1/2 covariates, the P values derived from chi-square testing are as follows: grade, P = .755; TNM T classification, P = .013; tumor size, P = .039; vascular invasion, P = .022; capsular invasion, P = .083; and nonpapillary versus papillary, P = .006.

  • Denotes statistical significance (P < .05) with respect to pERK-1/2.

Tumor grades 1 and 2752312811825 (21)
Tumor grades 3 and 4297914629 (63)
TNM T15013316710 (15)
TNM T2285824310 (23)
TNM T3 and T426121065434 (63)
Tumor size <7 cm5517437917 (22)
Tumor size >7 cm49131768537 (44)
Vascular invasion absent711511410118 (18)
Vascular invasion present33151056336 (57)
Capsular invasion absent902615613732 (23)
Capsular invasion present144632722 (82)
Nonpapillary842921814251 (36)
Papillary20101223 (14)
Recurrence, n (%)25 (24)12 (40)12 (57)5 (56)54 (36%) 

Activated ERK-1/2 and Disease-free Survival Analysis in Clinically Confined RCC

Univariate survival analysis (Kaplan-Meier) indicated that patients with tumors that did not express pERK-1/2 had a mean disease-free survival of 6.38 years (95% confidence interval [95% CI], 5.80 years-6.96 years) compared with 4.83 years (95% CI, 3.68 years-5.98 years) for patient samples scored as 1, 3.53 years (95% CI, 2.12 years-4.95 years) for patient samples scored as 2, and 3.93 years (95% CI, 2.15 years-5.70 years) for patient samples scored as 3. Therefore, increasing levels of pERK-1/2 in tumor samples resected from RCC patients resulted in a significantly (P = .0017) shorter time to disease recurrence (Fig. 2 Top). This survival analysis also indicated that patients whose tumors displayed any evidence of pERK-1/2 staining had significantly shorter time to disease recurrence than patients whose tumors were negative (Fig. 2 Top). This allowed a separate analysis to be undertaken, in which tumors simply could be categorized as positive or negative. In this analysis, patients with pERK-1/2–positive tumors had a highly significant (P = .0007) reduced disease-free survival of 4.19 years (95% CI, 3.36 years-5.03 years), compared with 6.38 years (95% CI, 5.80 years-6.96 years) for patients with pERK-1/2–negative tumors (Fig. 2 Bottom). By implication, expression of pERK-1/2 per se can be used to predict disease-free survival in patients with RCC.

Figure 2.

Kaplan-Meier metastasis-free survival plots of renal cell carcinoma patients with clinically confined disease are (Top) based on phosphorylated extracellular signal-regulated kinase (pERK) score and (Bottom) stratified by “positive” versus “negative” pERK expression.

Cox Regression Multivariate Analysis

By using multivariate Cox proportional hazards regression models, we next evaluated whether the expression of pERK-1/2 could be of prognostic value in the assessment of primary renal tumors (Table 2). We have previously shown31 the covariates of tumor grade, vascular invasion, and capsular invasion to be influential prognostic factors in RCC, but when tumor size and tumor type were not. Therefore, tumor grade, vascular invasion, and capsular invasion together with pERK were entered into the model. The analysis revealed that pERK-1/2 was a significant influential predictor of shortened disease-free survival with a hazards ratio (HR) of 2.96 (P < .001). Of note, the predictive value of pERK-1/2 was significantly higher than the HR of vascular invasion (HR, 1.58; P = .20) and only slightly lower than that of the powerful and robust prognostic indicator tumor grade (HR, 3.01; P < .001).

Table 2. Multivariate Cox Regression Hazards Model for Time to Disease Recurrence
Prognostic Indice/Model Including Vascular Invasion (No.)HR95% CIP-value
  1. HR indicates hazards ratio; 95% CI, 95% confidence interval; pERK, phosphorylated extracellular signal-regulated kinase.

Grade 1 and 2 (118)1  
Grade 3 and 4 (46)3.011.59-5.72<.001
No capsular invasion present (137)1  
Capsular invasion present (27)4.042.12-7.69<.001
No vascular invasion present (101)1  
Vascular invasion present (63)1.580.78-3.19.20
pERK-1/2 negative (104)1  
pERK-1/2 positive (60)2.961.69-5.67<.001

DISCUSSION

Previous clinicopathologic studies have shown that activated ERK is a reliable predictor of disease progression in several diverse cancers that include malignancies of the liver,23 lung,24 and breast.26 Oka et al34 demonstrated a statistically significant association between pERK expression and grade in a small series of 25 patients with RCC. However, they determined activated ERK levels through Western blot analysis of whole tumor homogenates, an approach that suffers from population averaging effects of contaminating nontumor cells contributed by stromal and vascular compartments. Although experimental studies with renal cancer cell lines and tumor xenografts have provided important insights into the role of pERK in several disease processes, such as invasion and angiogenesis,35-38 the significance of pERK with respect to disease progression and survival of patients with RCC remains unknown. In this current study, we used immunocytochemistry and TMA technologies to examine the prognostic relevance of pERK expression in 164 cases of RCC. To allow the direct assessment of the role of pERK expression in disease progression and to circumvent the issues of drug treatment and coexistent metastatic disease in data interpretation, our unique patient cohort was selected on the basis that subjects: 1) presented with clinically confined disease only, and 2) had not received any therapy before surgery. Herein we provide what to our knowledge is the first demonstration that pERK expression correlates with aggressive and advanced histopathologic indices of RCC and poor disease-free survival. In addition, our analysis also demonstrated that pERK serves as a highly significant and independent prognostic biomarker that predicts disease progression in RCC.

In this study, increased expression of pERK-1/2 was found to be correlated with the conventional clinicopathologic variables of tumor size and vascular invasion. The association of pERK-1/2 with tumor size is perhaps expected, given that ERK activation has been historically connected with cell proliferation. Nevertheless, this correlation is highly relevant, because tumor size is an important prognosticator for patient survival in RCC,39 and is a histopathologic criterion that is incorporated into the universally accepted TNM classification system used for staging renal tumors.40 In addition to its direct effect on cancer cell proliferation, pERK-1/2 may also contribute to tumor size in RCC via the promotion of angiogenesis, thereby maintaining the supply of oxygen and nutrients to the tumor, affording uncompromised growth. Hypoxia-inducible factor 1α (HIF-1α), a key molecule that facilitates the process of angiogenesis in RCC, is a well-documented substrate of ERK in several different cell types.41, 42 The cooperativity of HIF-1α and pERK-1/2 may explain the high vascular nature of RCC tumors per se and hence the correlation between pERK-1/2 and tumor size observed in our current study. Consistent with this view, a recent study has shown that the specific pharmacologic inhibition of ERK is sufficient to suppress the growth and angiogenesis of RCC tumor human xenografts in mice.37

During the final drafting of our current article, a report by Lee et al43 was published that examined the prognostic significance of pERK-1/2 in 328 patients with RCC, comprising patients with localized and metastatic clear-cell subtypes. These authors stratified patients into low and high pERK-1/2 groups, and reported that although pERK-1/2 expression did not reach any independent prognostic significance, high pERK-1/2 expression as compared with low expression was correlated with better disease-free survival, but only in patients whose tumors measured <7 cm in greatest dimension; in tumors measuring >7 cm, they did not demonstrate any correlation between pERK-1/2 and survival. Increased tumor size appears to be positively correlated with RCC progression. In our own study, restricted to localized RCC and stratifying patients into positive and negative pERK-1/2 groups, we were able to demonstrate that pERK-1/2 was an independent poor prognostic factor irrespective of tumor size, although we did note a statistically higher incidence of pERK-1/2 in tumors measuring >7 cm. In our study, papillary tumors represented 13.4% of total RCCs (22 of 164), which is typical of any given RCC patient cohort. Analysis of the nonpapillary cohort still demonstrated that increased pERK-1/2 levels served as a significant predictor of disease recurrence in patients with clinically confined RCC (P = .018). Ethnic differences between the patient cohort used in our study and that of Lee et al cannot be dismissed as a basis for contrasting findings. Ethnicity can provide for different genetic and epigenetic backgrounds in common tumors that ultimately impact on disease progression and response to treatment. For example, EGF-R, a key upstream effector molecule of ERK, demonstrates more frequent mutations in Asian nonsmall cell lung carcinoma populations.44 Although immunohistochemistry provides opportunity to determine both the global and cellular distribution of pERK-1/2 within renal tumors, it does not allow discrimination between the exact involvement of the individual activated ERK isotypes (ie, pERK-1 vs pERK-2) in disease progression. The possibility exists that the mouse monoclonal antibody used in the study by Lee et al43 and the rabbit polyclonal antibody used by ourselves in this current study preferentially detect different isoforms of pERK. Of note, gene ablation studies in mice and in vitro cell lines have recently shown that ERK-2 predominately drives Ras-dependent cell proliferation,45-47 whereas ERK-1 can actually attenuate Ras-dependent tumor formation.47 Using Affymetrix microarray technology, Huang et al37 recently demonstrated elevated ERK-2 levels in 174 cases of clear cell renal tumors.

Multiple reports have documented the importance of microvascular tumor invasion as a robust and accurate determinant of disease progression in RCC. Indeed, in patients with low-grade and localized disease, vascular invasion has been shown to be the most significant independent prognosticator of all pathologic parameters examined. The current study finding that pERK expression is strongly associated with the presence of vascular invasion in clinically confined disease is highly relevant and supports the inclusion of pERK status in staging schemes and prognostic models for RCC that incorporate tumor biomarkers alongside conventional parameters. The clinical correlation between pERK expression and invasion can be reconciled at the molecular level through activated ERK induction of extracellular matrix-digesting enzymes, facilitating cell migration, and invasion into surrounding tissue.48-51 Recently, it has been demonstrated that the invasive capacity of the human renal cell carcinoma cell line Caki-2 is substantially increased on sustained activation of ERK-1/2.38 The enhanced ERK-mediated invasive ability of these cells resulting from up-regulation of metalloproteinases 2 and 9, a mechanism attenuated by the combination of small interfering RNA–mediated down-regulation of ERK-1/2 and chemical inhibition of the RAS/RAF/ERK pathway using PD98059. Furthermore, it has been reported that leptin36 can potentiate the invasiveness of murine renal cancer cells in an ERK-dependent manner. Collectively, such mechanistic studies support the view that activated ERK has a direct involvement in the process of vascular invasion and hence the promotion of metastatic disease in RCC, a view further corroborated by our current clinical findings.

In the current study, multivariate Cox regression analysis determined that pERK was equivalent to tumor grade with respect to its ability to predict disease recurrence, with calculated hazards scores of 2.96 and 3.01, respectively. Surprisingly, we failed to demonstrate a correlation or trend between pERK expression and tumor grade. However, because they are independent of each other and yet represent highly valid independent prognosticators, both pERK and tumor grade may be viewed as complementary determinants that could be effectively used for the improved prediction of metastatic disease in clinically confined RCC. Prognostic models incorporating individual biomarkers with standard clinical parameters have been developed for RCC with good effect.52, 53 For example, malignancy-associated biomarkers that include, among others, PTEN, p53, and Ki-67 have been ranked alongside several conventional RCC staging schemes such as Fuhrman grade, TNM, and Eastern Cooperative Oncology Group performance status. The resultant integrated “clinical/biomarker” allows better prediction of patient survival than either the panel of biomarkers or clinicopathologic parameters alone.

With the emergence of cancer therapies that specifically target growth factor signaling pathways, attention has focused on the immunohistochemical evaluation of suitable biomarkers that can serve as a basis for patient selection and also as markers of treatment response. ERK represents a downstream convergence point for several disparate growth factor signaling cascades, including among others EGF-R, platelet-derived growth factor receptor (PDGF-R), and vascular endothelial growth factor receptor (VEGF-R).16-18 In our study, we demonstrated that pERK-1/2 positivity per se is sufficient to significantly predict disease progression in RCC without the need to discern between various degrees of expression and ascertain its exact subcellular localization. However, pERK-1/2 may also represent a reliable and robust biomarker for the selection of high-risk RCC patients who may benefit from treatment with novel molecular targeted therapies, and serve as an appropriate biomarker by which to measure tumor response to such agents. Indeed, sorafenib and sunitinib, both of which are multitargeted tyrosine kinase inhibitors (TKIs) of the PDGF-R and VEGF-R, are indicated for use in RCC.8, 54 Support for this notion is gained from studies demonstrating that in cancer patients treated with targeted TKIs, tumor tissue exhibited lower levels of pERK post-treatment. Specifically, significant correlations between reduced pERK and disease stabilization or partial responses were observed after therapy with sorafenib55 and erlotinib56 in patients with metastatic carcinomas of the head, neck,55 and colon,56 respectively.

In summary, the results of the current study indicate that pERK-1/2 is an independent prognostic biomarker that significantly predicts the onset of metastasis in clinically confined RCC. In addition, we propose that pERK-1/2 may be used to better select patients with renal cancer who may benefit from treatment with specific molecular targeted therapies and/or as a tumor marker to measure and monitor treatment efficacy when using such agents. Further validation is required in other patient cohorts.

Acknowledgements

We thank Dr. Julia Gee, Mrs. Susan Kyme, and Mrs. Pauline Finley of the Tenovus Research Group (Cardiff University) for advice and assistance with the immunocytochemistry procedures, and Chris von Ruhland for help with image acquisition.

Conflict of Interest Disclosures

The authors made no disclosures.

Ancillary