Prognostic Significance of RNA-Dependent Protein Kinase on Non–Small Cell Lung Cancer Patients

RNA-dependent protein kinase (PKR) has a well-established role in antiviral defense mechanisms and other cellular functions, such as the regulation of growth, apoptosis, proliferation, signal transduction, and differentiation (1–5).

The role of PKR apparently varies with the type of human cancer. For example, Terada et al. (6) reported significant correlations between high PKR scores and vascular invasion and the presence of satellite tumor nodules in thyroid carcinoma cases. Immunohistochemical analysis found minimal PKR immunoreactivity in primary tumors but high levels of PKR protein expression in lymph node metastases in melanoma cases (7). In a small study, Roh et al. (8) found that patients with bronchioloalveolar carcinoma (BAC) and high PKR expression levels had significantly shorter survival durations than did patients with BAC and low PKR expression levels.

Conversely, studies have shown that increased expression of PKR correlates with a better prognosis in patients with head and neck or colon cancer (9, 10). In addition, a positive correlation between PKR expression and the degree of differentiation among cancers of the colon, breast, liver, and head and neck has been shown (11–16). Researchers have suggested that PKR has a role as a tumor suppressor in patients with leukemia and other hematologic malignancies (17–19).

In recent years, we have studied PKR pathways and have found them to be clearly necessary for induction of cell death in some cancer cells including lung cancer after certain treatments (20–25). In the present study, we tested the hypothesis that differences in the levels of expression of PKR proteins could act as a prognostic marker for clinical outcomes in non–small cell lung cancer (NSCLC) patients. Our findings first indicate that PKR expression is an independent prognostic variable in NSCLC patients.

NSCLC and normal human bronchial epithelium tissue microarray (TMA) specimens were from the Lung Cancer Specialized Program of Research Excellence Tissue Bank at The University of Texas M. D. Anderson Cancer Center (Houston, TX; ref. 26). This study was approved by the Institutional Review Board. Tumor tissue specimens for TMA-1 (study set) and TMA-2 (validation set) from 455 lung cancers [278 adenocarcinomas (ACC) and 177 squamous cell carcinomas (SCC)] were histologically examined, classified using the WHO classification system (27), and selected for TMA construction. After histologic examination, the TMAs were constructed using triplicate 1-mm-diameter cores from each tumor. Detailed clinical and pathologic information, including demographic data, smoking history (never and ever smokers), pathologic tumor-node-metastasis (TNM) staging, overall survival, and time of recurrence, was available in most cases (Tables 2 and 3).

The rabbit anti-PKR polyclonal antibody (K-17) used for immunohistochemical staining was obtained from Santa Cruz Biotechnology. Immunohistochemical staining was done as previously described (26).

Immunohistochemical expression was quantified by two pathologists independently for TMA-1 and TMA-2 (M.G.R. and A.P.) using a four-value intensity score (0 for negative, 1 for weak, 2 for moderate, or 3 for strong), and the percentage of tumor cells within each category was estimated. A final score was obtained by multiplying both intensity and extension values (0 × % negative tumor cells + 1 × % weakly stained tumor cells + 2 × % moderately stained tumor cells + 3 × % strongly stained tumor cells). The score ranged from a minimum of 0 to a maximum of 300. Cases with discordant scores between observers were reevaluated.

Human lung (A549, H1299, and H322B) cancer cell lines were obtained from the American Type Culture Collection. Polyinosinic acid:poly-CMP (polyI:C) was obtained from Calbiochem. We measured apoptotic cells by propidium iodide staining and fluorescence-activated cell sorting analysis (FL-3 channel, FACScan; Becton Dickinson; ref. 20).

At 48 hours after transfection, the cell extracts were prepared and immunoblot assays were done as previously described (20). Antibodies to PKR (K-17), phosphorylated PKR (Thr⁴⁵¹), and β-actin (control) were obtained from Santa Cruz Biotechnology.

Survival probability as a function of time was computed by the Kaplan-Meier method. The log-rank test was used to compare patient survival time between groups. Cox regression was used to model the influence of PKR expression on survival time, with adjustment for clinical and histopathologic parameters (age, sex, tumor histology subgroup, and smoking status). The two-sided test was used to test proportions between groups in two-way contingency tables. In univariate analysis, independent sample Student's t and χ² tests were used to analyze continuous and categorical variables, respectively. All of the statistical tests were two-sided; P < 0.05 was considered to be statistically significant. Statistical analyses were done using the S-Plus and the SPSS software.

We investigated PKR expression in NSCLC by immunohistochemical analysis using TMAs from 88 normal bronchial epithelium samples and 231 NSCLC tumor samples (TMA-1) obtained from patients with stage I to IV disease who had not received neoadjuvant or adjuvant therapy and for whom clinical outcomes were known. The samples consisted of 132 ACC and 99 SCC samples. We detected PKR expression in both the cytoplasm and nuclei of bronchial epithelial cells but only in the cytoplasm of NSCLC cells (Fig. 1). Overall, we detected high-level PKR expression (mean score, 282) in normal bronchial epithelium samples and reduced PKR expression (mean score, 170) in NSCLC tumor samples (P < 0.001; Fig. 1; Table 1). Clearly, NSCLC tumors showed reduced mean levels of PKR protein expression.

Using mean PKR expression scores, we did not detect any statistically significant correlations between protein expression and patient's sex, smoking history, or pathologic tumor category (Table 2). Low-level PKR expression was correlated with stage II to IV versus stage I disease (P = 0.025), the presence of lymph node metastasis (P = 0.005), and SCC versus ACC (P = 0.001; Table 2). We next analyzed the relationship between PKR expression and survival duration in pathologic all-stage NSCLCs on TMA-1 (Fig. 2A). The Kaplan-Meier survival curves in Fig. 2A show that low-level PKR expression was related to a significantly reduced probability of survival in the subgroup of patients in all-stage patients (P = 0.003). Multivariate Cox proportional hazards regression analysis showed that expression of PKR was significantly associated with the overall survival after accounting for the effects of age and pathologic T and N classification (P < 0.0001; Table 3).

We next validated our findings using the same PKR antibody in a second TMA set (TMA-2) from 224 NSCLC tumor samples (146 ACCs and 78 SCCs). We confirmed that low-level PKR expression was correlated with the presence of lymph node metastasis (P = 0.044) and SCC versus ACC (P = 0.001; data not shown). The Kaplan-Meier survival curves in Fig. 2B show that low-level PKR expression was related to a significantly reduced probability of survival in the subgroup of patients in all-stage patients (P < 0.0001). Univariate and multivariate Cox proportional hazards regression analysis showed that expression of PKR was significantly associated with overall survival (P < 0.0001; Table 3).

We next determined whether induction of PKR cause cell death in lung cancer cell lines. We first analyzed the PKR and phospho-PKR expression in A549, H1299, and H322B lung cancer cells after treatment with PBS, Ad-Luc (2,500 viral particles/cell), Ad-PKR (2,500 viral particles/cell), or polyI:C (50 ng/μL), which is a synthetic double-stranded RNA molecule that is a potent activator of PKR. Western blot analysis revealed that treatment of A549, H1299, and H322B lung cancer cells with polyI:C and Ad-PKR increased PKR and phospho-PKR levels (Fig. 3A). We next tested whether polyI:C or Ad-PKR induce cell death in A549, H1299, and H322B lung cancer cell lines. As shown by flow cytometric analysis of human lung A549, H1299, and H322B cancer cells, polyI:C or Ad-PKR induced cell death within 48 hours (Fig. 3B). In contrast, Ad-Luc was not toxic to A549, H1299, and H322B lung cancer cells (Fig. 3B).

In this study, we focused on the expression of PKR in NSCLC and its relationship with the clinicopathologic parameters and prognosis of the disease. Expression of PKR was first evaluated in 88 normal tissues and 231 NSCLC tumors by immunohistochemistry using TMA-1. We found PKR to be expressed in both the cytoplasm and nuclei of normal bronchial epithelium, but just in the cytoplasm of NSCLC specimens. The differences between normal and NSCLC cell lines were further confirmed by confocal studies on cell lines (data not shown). We observed PKR to be mainly localized in the cytosol in cancer cell lines. We found a diffuse expression pattern of PKR, which is interesting because the function of PKR in the cytoplasm, but not the nucleus, has been previously defined. It has been shown that the PKR protein is present in the cytoplasm of most cells in an inactive form (1–5). PKR molecules undergo a conformational change when binding to double-stranded RNA or other activator molecules, thus allowing for dimerization and autophosphorylation at Thr⁴⁴⁶ and Thr⁴⁵¹ and leading to activation of the kinase function (1–5). We have not been able to detect the active form of PKR (phospho-PKR) using several antibodies on our TMA-1 specimens and are thus seeking phospho-PKR antibodies for immunohistochemical staining.

Compared with findings in normal bronchial epithelium, reduction of PKR expression was found in 178 (77%) of 231 NSCLCs. We detected a significant association of low-level PKR expression with lymph node metastasis and observed that reduction of PKR was related to a lower probability of survival. Univariate and multivariate Cox proportional hazards regression model analysis also showed that the PKR expression was significantly associated with overall survival after accounting for the effects of age and pathologic T and N classification. In our study, we had the opportunity to validate the use of PKR as a biomarker using additional NSCLC tumor samples (TMA-2). Here, for the first time, we show that low expression of PKR is related to poor prognosis, and high expression of PKR is related to improved prognosis all-stage patients. Our data showed that a high level of PKR was an independent prognostic marker of survival in stage I and all-stage NSCLC patients. In a previous report, Roh et al. (8) found that BAC patients with high PKR expression levels had significantly shorter survival durations than did patients with low PKR expression levels. The results need to be confirming due to small samples (38 cases) and different antibodies used in their study. A BAC tumor represents <1% of our cases, and because of a lack of invasion, we did not include it in the TMAs. We will confirm the significance of PKR on BAC once we collect enough samples.

The mechanism by which PKR inhibits lung tumor progression is not clear yet. It has been shown that overexpression or activation of PKR in many cancer cells leads to apoptosis, which may be due to an increased phosphorylation of eIF2α, but also to the expression of proapoptotic factors such as Fas (28, 29). We speculate that PKR suppresses tumor growth by phosphorylation of eIF2α. In our previous and current study, we have shown that induction of PKR and phospho-PKR in human lung cancer cell causes cell death (20–23). Further study needs to determine the significant of phospho-PKR and phospho-eIF2α expression on these NSCLC tumors.

PKR activity in cancer is positively regulated by the cellular protein MDA-7 (melanoma differentiation–associated gene-7; refs. 20, 23). We have reported that the tumor suppressor MDA-7 interacts physically with PKR, leading to rapid induction of expression of PKR and activation of its downstream targets, resulting in apoptosis induction in human lung cancer cells (20, 23). Ishikawa et al. (30) reported a lack of a significant correlation between the MDA-7/IL-24 status and all patient characteristics, including pathologic stage, in patients with NSCLC. However, subset analyses showed that MDA-7/IL-24 expression was a significant factor predicting a favorable prognosis for ACC (30).

Our results indicate that loss of PKR expression is correlated with a more aggressive behavior all-stage disease and that a high PKR expression predicts a subgroup of patients with a favorable outcome. Our future goal is to determine whether PKR mRNA level correlates with its protein expression in lung tumor samples. We will also attempt to determine the PKR protein expression, gene mutation, and DNA copy number by various molecular techniques, such as the reverse-phase protein array, mutational analysis, fluorescence in situ hybridization, and microarray.

No potential conflicts of interest were disclosed.

We thank Denise M. Woods and Lakshimi Kakarala for their technical assistance and Wanda L. Reese and Deborah L. Waits for their assistance in preparing the manuscript.

Grant Support: NIH Specialized Program of Research Excellence grant 5P50-CA70970-04 (J.A. Roth and I.I. Wistuba), Department of Defense grant W81XWH-07-1-0306 (J.A. Roth and I.I. Wistuba), The University of Texas M. D. Anderson Cancer Center Support Core Grant 2 P30 CA016672, Homer Flower Gene Therapy Fund, Charles Rogers Gene Therapy Fund, Margaret Wiess Elkins Endowed Research Fund, and George P. Sweeney Esophageal Research Fund.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.