Cytoplasmic Clusterin Expression Is Associated with Longer Survival in Patients with Resected Non–Small Cell Lung Cancer

Clusterin is a heterodimeric glycoprotein that was first identified in 1983 (1) and has been subsequently given many different names, including complement lysis inhibitor, apolipoprotein J, testosterone-repressed prostate message-2, and sulfated glycoprotein-2. Clusterin has been regarded as a marker for cell death because its expression was up-regulated in cells undergoing apoptosis in both normal and malignant tissues (2, 3). However, other studies have found that clusterin is expressed in surviving cells and may be a marker for cell survival (4, 5). Much effort has been focused on elucidating the differential effects of clusterin observed in early studies, and many clusterin isoforms have since been identified. It has been shown that clusterin has two main isoforms, a 76- to 80-kDa secreted form that is thought to be cytoprotective and a 55-kDa nuclear form that is believed to be proapoptotic. The precursor to secretory clusterin is translated from the first AUG codon of full-length clusterin mRNA, whereas the precursor to nuclear clusterin is generated by alternative splicing from the second in-frame AUG codon of clusterin mRNA (6). Due to the discovery of several clusterin isoforms and inconsistent data on the role of clusterin in benign and malignant tissues, the role of clusterin in cell survival and cell death is still debated.

To explore the clinical significance of clusterin, several studies have examined its prognostic value in various malignancies. Cytoplasmic clusterin immunostaining was noted to correlate with poor prognosis in patients with renal cell carcinoma (7), hepatocellular carcinoma (8), urothelial bladder carcinoma (9), and prostate adenocarcinoma (10). However, cytoplasmic clusterin staining correlated with good prognosis in pancreatic adenocarcinoma (11) and did not correlate with prognosis in breast carcinoma (12). Thus, retrospective clinical studies show the same uncertainty regarding the role of clusterin in patient survival that has been observed in laboratory studies.

Although previous studies have examined the correlation between clusterin immunostaining and prognosis in several different malignancies, this correlation remains unexplored in lung cancer patients. Lung cancer is the second most common cause of cancer in both men and women in the United States and the leading cause of cancer deaths. Lung cancer has a dismal prognosis with 5-year relative survival rates of only 13.6% for men and 17.5% for women (13), so there is a need for biomarkers that can correlate with prognosis and guide therapy (14). In the present study, we report our analysis of clusterin staining in tumors resected from patients with non–small cell lung cancer (NSCLC).

Archived tissue blocks from patients who presented sequentially and had surgery between 1996 and 2002 were retrieved from the files of Vanderbilt University and the Nashville Veterans Affairs Medical Center pathology departments, according to the approved Institutional Review Board protocol (010178). For all tissue blocks, the H&E-stained sections were reviewed by a pathologist with special interest in lung cancer (A.G.). These diagnoses were reviewed with knowledge of the original pathology report diagnosis but no knowledge of patient outcome. Patients who had incompletely resected tumors (R1 or R2 resection) were excluded from the analysis, so only complete (R0) resections are used in this study. Table 1 summarizes the clinical and molecular characteristics of 113 patients with completely resected NSCLC. Clinical data were obtained from the tumor registry and hospital charts at Vanderbilt Medical Center and Nashville VA Medical Center. The patient follow-up data used in this study is derived from Vanderbilt's Lung Specialized Programs of Research Excellence database. The data managers follow all of the Lung Specialized Programs of Research Excellence patients for survival, progression, and recurrence.

After reviewing the diagnoses, two or three separate representative areas of the tumor were chosen by a surgical pathologist (A.G.). The paraffin-embedded tissues were sampled from archived conventional tissue blocks. The tissue microarrays were constructed with a Beecher instruments tissue arrayer by sampling the three representative areas (0.6 mm) of tumor from the original blocks of tumor and transferring them into a new array block.

With the tissue microarray blocks, paraffin-embedded material was available in a set of 134 individual tumors for evaluation of clusterin staining. Histologic sections from the tissue microarray blocks were deparaffinized in xylene and rehydrated through a series of graded alcohols to TBS buffer. Antigen retrieval was carried out for 25 min in a preheated vegetable steamer with slides immersed in a citrate buffer (Bioigenex). Endogenous peroxide was quenched by applying 0.03% hydrogen peroxide for 5 min and then followed by a buffer wash. Predetermined primary antibody dilution was applied for 1 h at room temperature (Ab, 1:25 dilution of mouse monoclonal anticlusterin α chain; Upstate). After appropriate buffer washes, the specifically bound primary antibody was tagged and visualized using a Dako Monoclonal EnVision+ System (Dako Cytomation). Slides were then counterstained with hematoxylin, dehydrated through alcohols of increasing concentration, placed in xylene, and then cover-slipped using Permount. Three separate tumor samples from each case were examined by immunohistochemistry method on the tissue microarray sections and scored by consensus review (J.M.A. and A.G.). The reviewers were blinded to the original diagnosis and the patient outcomes at the time of review. Staining was assessed in five to ten high-powered fields at 400× magnification. Clusterin immunoreactivity was evaluated semiquantitatively based on the intensity of staining. It was scored as 1+ (weak), 2+ (moderate), and 3+ (intense). The percentage of area which stained positively was also determined. Labeling score was determined by multiplying intensity score by the percentage area which stained positive (intensity × percentage area with positive clusterin staining). The highest labeling score among the three tissue sections was entered for statistical analyses. Samples with no staining or weak staining were considered negative, and samples with moderate to intense staining were considered positive.

The primary analysis focused on detecting the association between clusterin staining and overall/recurrence-free survival. The staining score was categorized into two groups as weak/negative staining (score, <80) and strong staining (or negative/positive; score, ≥80). This threshold was determined by visually determining a clear positive stain supported by histograms of the range of scores. Overall survival was calculated from the date of diagnosis to the date of death or last follow-up. Recurrence-free survival was calculated from the date of diagnosis to the date of recurrence or death date; data were censored for patients who were alive (recurrence-free) at their last follow-up visits. In the analysis, recurrence was defined as recurrence of a tumor after resection or any detected metastases after resection. Kaplan-Meier survival curves were calculated for the subgroups of potential risk factors and were compared using the log-rank test. Cox proportional hazards model was fitted to estimate hazard ratios and confidence intervals adjusting for age, disease stage, and histologic type. Demographic frequencies between weak and strong staining groups were compared using Fisher exact test. All P values are based on two-sided tests, and differences were considered statistically significant when P value is <0.05. Analyses were done with the use of SAS system version 9.1 and R version 2.1.1.

Tissue microarray samples from the tumors of 113 patients with surgically resected NSCLC were immunohistochemically stained with an antibody that detects clusterin. The demographic information for the patients in this study is shown in Table 1. Among the patients, 62% were male and 38% were female; 89% were Caucasian and 11% were African-American. In terms of disease staging at the time of surgical resection, 65% of the patients had stage I disease, 12% had stage II disease, and 23% had stage III disease. Various histologic tumor types were also represented with 37% of patients having adenocarcinoma, 30% squamous cell carcinoma, and 33% other, a category that included bronchioloalveolar carcinoma (n = 5), large cell carcinoma (n = 6), carcinoid tumor (n = 8), large cell neuroendocrine tumor (n = 6), and poorly differentiated NSCLC (n = 12).

To determine the cellular location and distribution of clusterin, specimens were stained with anticlusterin α-chain antibody (Upstate), an antibody that has been previously used to detect both nuclear and cytoplasmic clusterin isoforms (15, 16). We observed cytoplasmic localization of clusterin in many of the tumors, but no nuclear staining was observed. Forty-four of the samples contained appreciable cytoplasmic clusterin staining and were therefore scored as positive for clusterin (Table 1). Figure 1 contains representative samples of the varying staining intensities observed. Because a wide range of staining areas and intensities was present, we conservatively used a staining index of 80 to score a sample as definitively positive, meaning at least 1+ staining in 80% of the tumor cells (or 2+ staining in 40% of tumor cells).

We used the Fisher exact test to determine whether there was a statistically significant difference in clusterin staining between the different demographic variables (Table 1). Significant associations were noted between cytoplasmic clusterin staining and histologic type (P = 0.0075), as well as between clusterin staining and survival (P = 0.0060). An association was also noted between positive staining and disease recurrence, which approached statistical significance (P = 0.0537). No association was seen between clusterin staining and gender, race, or clinical stage.

To determine whether clusterin expression correlates with prognosis in patients with NSCLC, we analyzed the association between clusterin immunostaining and 3-year overall and recurrence-free survival. Results of our analysis using the log-rank test are shown in Table 2. We found that lower clinical stage was significantly associated with longer overall survival (P = 0.0075) and recurrence-free survival (P = 0.0031). Positive clusterin staining was also significantly associated with longer overall survival (P = 0.0108) and recurrence-free survival (P = 0.0231). Gender, race, and histologic type had no association with overall or recurrence-free survival. Figure 2 shows Kaplan-Meier curves for overall survival and recurrence-free survival, comparing positive versus negative clusterin staining (Fig. 2A and B), clinical stage (Fig. 2C and D), and histologic type (Fig. 2E and F).

Due to a low number of stage II patients (n = 13) and a high percentage of noncancer-related deaths in this small group, our analysis showed a lower survival rate for stage II patients than stage III patients. This difference was not statistically significant. However, based on this, the authors decided to conduct a subgroup analysis of stage I patients to confirm the finding that clusterin staining predicts longer survival. We hypothesized that this result would hold true in the more homogenous stage I population. This was indeed the case, as 3-year overall survival was 71% for clusterin-positive patients and only 50% for clusterin-negative patients (P = 0.043). A similar association was seen for recurrence-free survival, but this was not statistically significant (P = 0.109), which is likely due to the small number of patients in the subgroup analysis. The Kaplan-Meier curves for overall survival and recurrence-free survival comparing positive versus negative clusterin staining in the stage I population is shown in Fig. 2G and H.

To determine the independent prognostic value of cytoplasmic clusterin staining, multivariate analysis was done using Cox proportional hazards model (Table 3). Positive clusterin staining was significantly associated with better survival (P = 0.0190) and had an association with reduced recurrence that approached statistical significance (P = 0.0549). The independent relative risk was 0.487 [95% confidence interval (95% CI), 0.27-0.89] for death and 0.345 (95% CI, 0.12-1.02) for recurrence in patients with tumors that stained positive for clusterin. Stage I disease predicted greater survival than stage II disease (P = 0.0105), as well as stage III disease (P = 0.0808). Stage I disease also had less recurrence than stage III (hazard ratio for stage III versus stage I, 4.57; 95% CI, 1.78-11.8), but the difference in recurrence between stage I and stage II did not reach statistical significance (hazard ratio for stage II versus stage I, 3.26; 95% CI, 0.75-14.1), which is likely due to only 13 stage II patients. Additionally, tumor cell histology had no significant association with death or recurrence.

In the present study, we found that cytoplasmic clusterin staining was significantly associated with better 3-year overall survival (P = 0.0108) and recurrence-free survival (P = 0.0231). We also found that lower clinical stage predicted better overall and recurrence-free survival but that disease stage was not correlated with clusterin staining (P = 0.6306). These results suggest that cytoplasmic clusterin staining may be a useful biomarker in NSCLC, independent of clinical staging.

There was also a significant association between clusterin staining and histologic type (P = 0.0075), with positive clusterin staining observed in 57% of adenocarcinomas, 21% of squamous cell carcinomas, and 35% of other NSCLC tumors. However, we did not observe a significant correlation between tumor histology and survival. We found that squamous cell carcinoma had slightly better 3-year overall and recurrence-free survival than adenocarcinoma, but these values were not significant (Fig. 2E and F). Many studies have examined the prognosis of squamous cell carcinoma compared with adenocarcinoma and other histologic types of lung cancer. In general, squamous cell carcinoma has been shown to have a better prognosis with longer survival (17-20). Because we found that adenocarcinoma had the highest staining percentage, but not improved survival, this suggests that the association between positive clusterin staining and survival is independent of tumor histologic type.

Inconsistent results have been reported regarding the association of cytoplasmic clusterin immunostaining and prognosis in cancer patients. Poor prognosis has been associated with clusterin staining in renal cell carcinoma (7), hepatocellular carcinoma (8), urothelial bladder carcinoma (9), and prostate adenocarcinoma (10). However, clusterin staining correlated with good prognosis in pancreatic adenocarcinoma (11) and did not correlate with prognosis in breast carcinoma (12). In this study, we report for the first time the correlation between cytoplasmic clusterin staining and improved overall survival in patients with NSCLC. There are many possible explanations for these mixed results. First, it is possible that clusterin plays different roles in different tissues or that the unique biochemical characteristics of different malignancies result in variable levels of clusterin. Additionally, it has been shown that different antibodies are required to detect the various forms of intracellular clusterin, and many antibodies cross-react with several isoforms (21). One study showed that a 49-kDa precursor of nuclear clusterin resides in the cytoplasm and is posttranslationally modified after exposure to proapoptotic signals and translocated into the nucleus resulting in cell death (22). Several other studies have characterized the nuclear localization of different clusterin isoforms in dying or stressed cells (23-25). Furthermore, confounding this problem is the identification of several other clusterin isoforms that are the result of exon skip (26) or posttranslational modifications (21). Because the antibodies used in staining studies are likely detecting several different isoforms, future studies need to more accurately characterize the isoform(s) detected by their antibodies.

The anticlusterin α-chain antibody used in the present study has been used to detect both cytoplasmic and nuclear clusterin staining patterns, suggesting that it detects several different isoforms (15, 16). Furthermore, previous studies have shown varying results with this antibody. One study found that this antibody detected an intracellular clusterin that localized the nucleus in cancer cells transiently transfected with this intracellular form (25). Another study found that this antibody did not detect the 55-kDa nuclear form of clusterin but did detect the ∼60-kDa precursor to secretory clusterin and the two chains of mature secretory clusterin (23). Therefore, the cytoplasmic clusterin observed in the present study could represent the prosurvival secretory clusterin, a cytoplasmic precursor to the proapoptotic nuclear clusterin, or another isoform that contains the detected α-chain sequence. If the observed cytoplasmic clusterin in this study is the secreted isoform, this would contradict the established role of this protein as a cytoprotective protein, because inhibiting lung cancer cell death would result in poorer prognosis. However, it has been shown (using the same antibody) that an intracellular clusterin isoform results in G₂-M cell cycle arrest and apoptosis in prostate cancer cells, but in that study the transfected clusterin readily localized to the nucleus. Interestingly, these authors found that apoptosis-resistant survivor clones expressed exclusively cytoplasmic clusterin, but that the proliferation rate of these cells was much slower than controls (25). Although this has not been explored in lung cancer models, a slower proliferation rate would result in better clinical outcomes, given that lung cancer is more refractory to standard treatment and has poorer prognosis than prostate cancer.

Chemo- and radioresistance are problematic in NSCLC therapy, and as a result, treatment efficacy has reached a plateau (27-29). Tumor development of standard treatment resistance underscores the importance of development of novel therapeutics. With the use of antisense oligonucleotides, clusterin has been identified as an important therapeutic target for paclitaxel chemosensitization (30) and radiosensitization (31) in NSCLC models both in vitro and in vivo. Additionally, inhibiting clusterin resulted in radiosensitization of tumor vasculature (31), and another study showed that clusterin translocates from the cytoplasm to the nucleus under stress conditions in blood vessels (32). This suggests that tumor vasculature may also be a target of anticlusterin therapy. Because many of the antibodies used in these studies cross-react with various clusterin isoforms, it will be important to accurately characterize the targeted isoforms to maximize the efficacy of anticlusterin cancer therapy. The development of clusterin-targeted therapeutics increases the usefulness of accurately characterizing clusterin in clinical tumor specimens, and clusterin immunostaining would provide a means of monitoring treatment efficacy. In addition, proteomic analysis has shown that different clusterin isoforms can be detected in the serum of patients with colorectal cancer compared with controls (33), and increased serum levels of clusterin were detected in patients with NSCLC compared with control patients (34). These proteomic studies further validate the potential role of clusterin as a biomarker.

In summary, our results suggest that cytoplasmic clusterin staining may be an independent biomarker for prognosis in patients with surgically resected NSCLC. To the authors' knowledge, we are the first to report an association between clusterin staining and prognosis in lung cancer. Larger prospective studies are needed to confirm and fully determine the value of clusterin in predicting NSCLC prognosis. Given the conflicting reports on the role of clusterin in both cell death/survival and clinical outcome, future studies should focus on elucidating the role of the various clusterin isoforms in cell signaling and clinical prognosis.

We thank the participants of the screening trial for making this study possible.