|Home | About | Journals | Submit | Contact Us | Français|
We examined the prognostic value of nuclear and cytoplasmic Survivin expression in men with locally advanced prostate cancer who were enrolled in Radiation Therapy Oncology Group (RTOG) protocol 8610.
RTOG 8610 was a phase III randomized study comparing the effect of radiotherapy (RT) plus short-term androgen deprivation (STAD) with RT alone. Of the 456 eligible cases, 68 patients had suitably-stained tumor material for nuclear Survivin analysis and 65 patients for cytoplasmic Survivin.
Compared to patients with nuclear Survivin intensity scores ≤191.2, those with intensity scores >191.2 had significantly improved prostate cancer survival (hazard ratio (HR) = 0.45, 95% confidence intervals (CI) = 0.20–1.00, p = 0.0452). On multivariate analysis, nuclear Survivin intensity scores >191.2 were significantly associated with improved overall survival (HR = 0.46, 95%CI = 0.25–0.86, p = 0.0156) and prostate cancer survival (HR = 0.36, 95%CI = 0.16–0.84, p = 0.0173). On univariate analysis, compared to patients with cytoplasmic Survivin integrated optical density ≤82.7, those with integrated optical density >82.7 showed a significantly increased risk of local progression (HR = 2.49, 95%CI = 1.03–6.01, p = 0.0421).
Nuclear overexpression of Survivin was associated with improved overall and prostate cancer survivals on multivariate analysis, while cytoplasmic overexpression of Survivin was associated with increased rate of local progression on univariate analysis in patients with locally advanced prostate cancer treated on RTOG 8610. Our results may reflect the different functions of Survivin and its splice variants, which are known to exist in distinct subcellular compartments.
A growing body of evidence indicates that Survivin is a unique member of the inhibitor of apoptosis (IAP) family. First, Survivin is structurally different from other IAPs in that it contains only one copy of the Baculovirus IAP Repeat (BIR) domain, and an extended COON-terminal α-helix, but no RING finger or other identifiable domain 1. Second, unlike the other IAPs, Survivin possesses a sharp differential expression in cancer versus normal tissues 2. Survivin is strongly expressed in embryonic and fetal organs 3, 4, but is undetectable in most terminally differentiated tissues 1. Moreover, it has been described to be selectively expressed in most common human neoplasms 5–12, including prostate cancer 13–17. Its expression in cancer cells is associated with cancer progression and prognosis. Third, Survivin seems to exist in two subcellular pools: cytoplasmic and nuclear 18. Consistent with the multiple subcellular localizations, Survivin is involved in both regulation of cell death and control of cell division 19. Fourth, four splice variants of human Survivin have been identified: Survivin-2α, Survivin-2β, Survivin-3β, and Survivin-δEx3 20–25. The wild type Survivin contains 3 introns and 4 exons. Survivin-2α and Survivin-3β are predicted to be truncated forms with Survivin-2β possessing an additional alternative exon 2 and Survivin-δEx3 omitting exon 3. The corresponding proteins, which display marked changes in their structure at the level of the BIR domain, are characterized by a different anti-apoptotic potential.
Survivin is a potential target in prostate cancer therapy. Survivin is not normally expressed in prostatic secretory epithelial cells, but is strongly expressed in prostate cancer cells 15. Furthermore, Survivin has been associated with established features of biologically aggressive prostate carcinoma, such as higher Gleason scores and metastases to regional lymph nodes 16, 17, 26. Our laboratory has identified Survivin as a potentially important mediator of resistance to chemotherapy and antiandrogen therapy in human prostate cancer. Using human prostate cell lines and xenograft models, we have demonstrated that Survivin mediates paclitaxel-resistance in prostate cancer cells. The inhibition of Survivin sensitizes prostate cancer cells to paclitaxel-induced apoptosis through a caspase-dependant mechanism in vitro and in vivo 13. Furthermore, our preclinical data indicates that the upregulation of Survivin plays a major role in resistance to antiandrogen therapy in prostate cancer cells. During androgen withdrawal, AR-independent upregulation of Survivin via IGFR1/AKT signaling confers resistance to antiandrogen therapy in prostate cancer cells 14.
Given these important findings, we proceeded to evaluate Survivin as a prognostic marker in locally-advanced prostate cancer using specimens from Radiation Therapy Oncology Group (RTOG) protocol 8610. RTOG 8610 is a phase III prospective trial that randomly assigned patients with locally advanced prostate cancers (T2–T4) without evidence of distant metastasis to receive goserelin (3.6 mg) every 4 weeks and flutamide (250 mg) three times a day for 2 months before radiation therapy and during radiation therapy or radiation therapy alone 27. The study opened on April 15, 1987 and closed on June 1, 1991 with a total of 471 patients; 456 of the patients were analyzed (15 were excluded for the following reasons: 4 had no follow-up, 5 had too small tumor, 3 refused all treatment and follow-up, 1 had lung primary, 1 had bone metastases, and 1 had benign disease). Results from the trial demonstrated a significant reduction in local progression and a prolongation of progression-free survival for the patients receiving neoadjuvant hormonal therapy. Given that Survivin isoforms can exist in distinct subcellular compartments and that isoform-specific antibodies have not demonstrated high reliability when used for immunohistochemistry in formalin-fixed tissues, we examined the prognostic value of both nuclear and cytoplasmic Survivin expression using an antibody sensitive to all Survivin isoforms in men with locally advanced prostate cancer enrolled in RTOG 8610.
For this analysis, a subset of patients entered in RTOG 8610 who had sufficient pathologic material available was studied. Diagnostic material (from needle biopsies or transurethral resections) was reviewed centrally for 461 (98%) of the 471 patients, and the tumors were graded according to the Gleason criteria. Tissue blocks were requested from participating institutions (>100) at the time of central pathology review for all cases that were reviewed. During the early years of the study, prostrate-specific antigen (PSA) evaluation on entry was not available. The protocol was later revised to include PSA determinations. Posttreatment PSA values during follow-up were available on all but 52 patients, and a total of 4,456 determinations were analyzed. The median number of PSA values per patient after initial treatment was 10.5. Pretreatment serum PSA determinations were available for less than 15% (10/68) of the eligible patients and hence PSA was not used as an outcome parameter in this study.
Sections were deparaffinized and dehydrated. Microwave unmasking of antigens was performed for 75 minutes in Target Retrieval Solution (Dako, Carpinteria, CA, USA) at 93.0 °C followed by cooling for 20 minutes. Endogenous peroxide was subsequently blocked with peroxidase-blocking solution (Dako) for 15 minutes, followed by washing for 5 minutes with phosphate-buffered saline (PBS). The sections were incubated overnight at 4.0 °C with polyclonal anti-Survivin antibody (Novus Biologicals, Littleton, CO, USA) diluted by antibody diluent (Dako) at 1:300. They were then washed three times in PBS for 5 minutes each and incubated for 30 minutes with labeled-polymer conjugated second antibody (envision+ kit, Dako). They were washed and developed with 3,3′-diaminobenzidine tetrahydrochloride for 2 minutes, lightly counterstained with hematoxylin, dehydrated and mounted.
Formalin-fixed paraffin-embedded tumor samples were analyzed using the ACIS™ (Clarient Chromavision Inc., San Juan Capistrano, CA, USA). The analysis was performed by Z.M. and L-Y.K., who have performed image analysis on RTOG 8610 samples prior to this study after initially reviewing the cases with pathologists28–30. The slides were scanned at 10× objective to give a pixel-to-tissue area ratio of 1:1 (1 pixel: 1 micron2). Using the Threshold Tool, brown and blue pixels were masked in each slide to derive the tumor tissue area and exclude all white areas. The analyses and threshold settings were performed separately for nuclear and cytoplasmic staining. Where possible, at least 6 regions of tumor cells were gated either by freehand around each tumor gland or by using a 40× circle if more than 90% of the region of interest contained tumor cells. Where there was heterogeneity in staining, the tumor area with the highest intensity of staining was chosen.
Nuclear Survivin mean index % and intensity score were calculated by the proprietary software. The Nuclear mean index % was the number of brown pixels divided by the total number of pixels (brown plus blue) in a gated region, taken as a percentage. The nuclear intensity score was the average brown intensity derived from all brown pixels included in a gated region and is scored in arbitrary units using a grayscale of 0–255.
The integrated optical density of gated pixels and intensity score were used to quantify the cytoplasmic Survivin staining. The integrated optical density has been shown to accurately represent the amount of antigen present in an IHC sample31. As prostate tumor glands vary in size, the integrated optical density was standardized by dividing by the tumor tissue area. The cytoplasmic intensity score was the average brown intensity derived from all brown pixels included in a gated region and is scored in arbitrary units using a grayscale of 0–255.
The event for overall survival was defined as death due to any cause. The following is considered an event in assessing prostate cancer survival: death documented as due to prostate cancer, death due to complications of treatment, or death from unknown causes with active malignancy (clinical disease relapse). Local failure was defined as an increase in tumor size of more than 50% by digital rectal exam for cases where complete tumor regression did not occur, or recurrence of a palpable nodule when there was complete regression or a positive biopsy of the prostate after two or more years of follow-up. Distant metastasis was defined as any clinical evidence of distant disease outside the pelvis.
The analysis was performed using the 456 eligible and assessable patients with follow-up. Of these patients, 68 had nuclear Survivin determinations and 65 had cytoplasmic Survivin determinations. Nuclear Survivin mean index % and Survivin intensity score were used for the nuclear Survivin analysis, while the integrated optical density and the intensity score for the cytoplasmic analysis. Each of these variables was modeled as a continuous variable as well as a dichotomized variable in Cox proportional hazards models to identify the impact on overall survival, prostate cancer survival, distant metastasis, and local progression.
Statistical comparisons to assess whether missing Survivin data were dependent on pre-treatment characteristics, treatment arm, or outcome were carried out using the Chi-square test and the Cox proportional hazards model. The five-year estimates for overall and prostate cancer survival were calculated using the Kaplan-Meier method32 and the cumulative incidence method was used to estimate the five-year local progression and distant metastasis failure rates33. Finally, a multivariate Cox proportional hazards model was used to identify the impact of Survivin on overall survival, prostate cancer survival, distant metastasis, and local progression in the presence of other covariates. Correlation between cytoplasmic and nuclear Survivin scores were tested using Pearson correlation test statistics.
Of the 456 eligible and analyzable cases entered in RTOG 8610, 68 patients had adequate and suitably-stained tumor material for nuclear Survivin analysis. There were no statistically significant differences in the pretreatment characteristics (age combined Gleason score, clinical stage, and assigned treatment) of the 456 assessable patients with or without nuclear Survivin determinations (data not shown). The differences with the four end points between the two groups are shown in Table 1a. There were statistically significant differences in prostate cancer survival and distant metastasis between these two groups. This implied that the subset of cases with Survivin is not necessarily a random representation of the entire RTOG 8610 study population and may not be generalizable outside the subset of cases with Survivin data. The distribution of pretreatment characteristics by nuclear Survivin mean index and intensity score are shown in Tables 1b and 1c, respectively. There were no statistically significant differences in the distribution of pretreatment characteristics between these two groups.
Sixty-five patients had sufficient pathologic material available for cytoplasmic Survivin analysis. Table 2a illustrates the differences between patients with cytoplasmic Survivin data versus those without cytoplasmic Survivin data with regards to outcome. There were statistically significant differences in prostate cancer survival and distant metastasis between these two groups. Again, this indicated that the subset of cases with Survivin is not necessarily a random representation of the entire RTOG 8610 study population. Tables 2b and 2c stratify patient characteristics by cytoplasmic Survivin integrated optical density (cut point = 82.7) and intensity score (cut point = 128.2). There were no statistically significant differences in the distribution of pretreatment characteristics between these two groups.
The correlation between nuclear and cytoplasmic Survivin was assessed and found to be positive and statistically significant (r = 0.40, p = 0.0012 for nuclear mean index % vs. cytoplasmic integrated optical density; and r = 0.42, p = 0.0005 for nuclear vs. cytoplasmic intensity score). The positive correlation coefficients (r) indicated that a high nuclear Survivin is correlated with a high cytoplasmic Survivin. Scatter plot matrices were shown in Figure 1A). Examples of weak and strong Survivin immunostaining are illustrated in Figure 1B. Nuclear and cytoplasmic immunostaining were then correlated with clinical outcome.
Univariate and multivariate analyses were performed to determine the relation of nuclear Survivin mean index % to the four end points. None of the analytical results showed a significant association between nuclear Survivin mean index % and any of the end points (data not shown).
Univariate analysis of nuclear Survivin staining stratified by intensity score, with a cut-point of 191.2, revealed a significant increase in prostate cancer survival for the patients with intensity score >191.2 (HR=0.45, 95% CI=0.20–1.00, p = 0.0452). Figure 2 demonstrates the Kaplan-Meier curves corresponding to this association. However, there were no other significant associations with either overall, local progression-free, or distant metastasis-free survivals, using nuclear Survivin intensity score >191.2 as a cut-point (data not shown). Of 68 patients with nuclear Survivin staining, 50 patients received salvage hormone therapy as “clinically indicated” for disease relapse. The protocol did not have guidelines as to type of salvage hormone therapy should be given or for when salvage hormone therapy should be initiated in relapsed patients34. We examined the association between the nuclear Survivin intensity score and the five-year prostate cancer survival from the start of salvage hormone therapy. Compared to patients with Survivin intensity score ≤191.2, those with intensity score >191.2 showed a significant increase in five-year prostate cancer survival (74.3% vs. 47.5%, p = 0.010).
The results of the multivariate analysis can be found in Table 3. For each end point, the model included Gleason grade, clinical stage, and assigned treatment. As many of these patients were entered prior to the PSA era, these values were not available for the multivariate model. Nuclear Survivin intensity score >191.2 remained significantly associated with improved prostate cancer survival (p= 0.0173). Moreover, patients with nuclear Survivin intensity score >191.2 showed a significantly improved overall survival (p = 0.0156).
The results of the univariate analysis of cytoplasmic Survivin can be found in Table 4. The univariate analysis revealed a significantly increased rate of local progression in patients with higher integrated optical density of cytoplasmic Survivin immunostaining. Compared to those with integrated optical density ≤82.7, those who had cytoplasmic integrated optical density >82.7 showed a significantly increased risk of local progression (HR=2.49, 95%CI=1.03–6.01, p= 0.0421,). However, likely due to modest patient numbers, this association did not hold up on multivariate analysis (Table 5). The other word of caution here is that evaluation of local progression without a biopsy is inherently a subjective clinical endpoint in the setting of prostate cancer and must be viewed with caution. There is no significant association between cytoplasmic Survivin intensity score and any of the four end points on either univariate or multivariate analyses (data not shown).
RTOG 8610 is the first major, phase III randomized trial to test the hypothesis that short-term neoadjuvant androgen deprivation therapy (ADT) combined with radiotherapy (RT) would improve treatment outcomes compared with RT alone. Long-Term Results of this study suggests that the addition of 4 months of ADT to RT appears to have a dramatic impact on clinically meaningful end points in men with locally advanced disease with no statistically significant impact on the risk of fatal cardiac events35. Now It is clear that long-term ADT + RT should be used in “high-risk” patients (clinical stage T2c-3, Gleason score 7–10, PSA> 20 ng/mL) and “low-risk” patients (clinical stage T1c-2a, Gleason score 2–6, PSA< 20 ng/mL) may benefit from short-term ADT + RT. Whether “intermediate risk” patients (clinical stage T2b, Gleason score 7, PSA 10–20 ng/mL) benefit from a combination of ADT and RT is not clear. Although the risk stratification system described by D’Amico et al is widely used and is relatively simple, it suffers from heterogeneity within the intermediate risk group36. To address this heterogeneity, it is necessary to create additional subgroups using novel prognostic markers in order to correlate with survival end point. Based on our preclinical studies of Survivin in human prostate cancer cell models and the previous reports of Survivin as a molecular marker in other cancers, we hypothesized that Survivin expression may serve as a prognostic factor for prostate cancer patients and different subcellular localization of Survivin may alter its prognostic role. We tested these hypotheses using samples from men with locally advanced prostate cancer enrolled in RTOG 8610.
The prognostic value of Survivin expression in prostate cancer has remained unclear. Most of the studies which have analyzed the prognostic significance of Survivin expression were either performed without accounting for the subcellular location of the protein, or reported Survivin location to be restricted to the cytoplasm of tumor cells 16, 17, 26. Our study for the first time separately describes the expression of Survivin protein in both the nuclear and cytoplasmic compartments of prostate cancer cells. In our analysis we identified that a different prognostic role is played by the nuclear and cytoplasmic Survivin expression. In particular, the nuclear expression of Survivin identified patients with improved overall and prostate cancer survival, while the cytoplasmic overexpression of Survivin was associated with increased rate of local progression. These findings have been confirmed by univariate and/or multivariate analysis of survival. In the multivariate analyses of nuclear and cytoplasmic Survivin, we used the independent variable “assigned treatment” to examine the risk of RT + hormone therapy as compared to RT alone. None of the analyses shown the significant difference between the two groups. However, among 50 patients who received salvage hormone therapy as “clinically indicated” for disease relapse, those with nuclear Survivin intensity score >191.2 showed a significant increase in five-year prostate cancer survival compared to patients with Survivin intensity score ≤191.2 (74.3% vs. 47.5%, p = 0.010), indicating that this subgroup of relapsed patients with higher nuclear Survivin expression may benefit from salvage hormone therapy.
Although growing evidence suggests that Survivin expression in cancer cells may represent an important prognostic marker to predict disease outcome, current reports are inconsistent and propose opposing conclusions. Cytoplasmic Survivin immunoreactivity has been observed in the vast majority of human tumors and it has been consistently associated with poor prognosis 19. Nuclear staining has been linked with a favorable prognosis in patients with gastric carcinoma 37, breast carcinoma 38, osteosarcoma 39, bladder mucosa and transitional cell carcinoma40, and pancreatic cancer41. In contrast, studies in hepatocellular carcinoma 42, 43, esophageal malignancies 44, and mantle cell lymphoma 45 have shown nuclear Survivin to be associated with poor survival. In addition, Survivin has been detected in the nuclei of non-small-cell lung cancer 46 and ovarian cancer 47 from clinical samples, without any significant relation with clinical outcome. The results of these studies have contributed to the confusion regarding the role of nuclear expression of Survivin. The prognostic difference for nuclear Survivin among these studies is likely tumor-specific.
It has been demonstrated that Survivin exists in two distinct subcellular pools: cytoplasmic and nuclear 18. The two Survivin pools are immunochemically distinct and independently modulated during cell cycle progression. Survivin has been described both as a chromosomal passenger protein in the nucleus and as a microtubule-associated protein in cytoplasm. In the nucleus, Survivin interacts with aurora kinase B and INCENP to complete mitosis 48, suggesting that strong nuclear staining of Survivin may represent increased mitotic events. Conversely, in the cytoplasm, Survivin blocks intrinsic and extrinsic apoptosis by directly inhibiting caspase-3 and -7, as well as interfering with caspase-9 49. This may partly explain the different prognostic implications of cytoplasmic and nuclear Survivin. The distinct roles of Survivin may also be due in part to splice variations. Survivin has four splice variants with different subcellular localizations: Survivin-2α Survivin-2β, Survivin-3β, and Survivin-δEx3 20–25. Survivin-δEx3 is localized preferentially in the nucleus, whereas the wild type Survivin and Survivin 2β isoforms are found in the cytoplasm. Survivin-2β and Survivin-δEx3 may antagonize wild type Survivin by forming heterodimers with reduced antiapoptotic potential, and acting as naturally occurring antagonists of Survivin 24, 50. Different isoforms of Survivin may interact, and the subcellular location ratio of the protein may contribute to the complex regulation of apoptosis and cell division according to these studies. Because the antibody we used in the immunohistochemical staining recognizes all these variants, it is not surprising that the intracellular location of Survivin has different prognostic implications in prostate cancer.
In summary, the present study is the first report to establish the prognostic relevance of the Survivin expression in prostate cancer in relation to its cellular distribution. The nuclear expression of Survivin is associated with improved overall and prostate cancer survival, while the cytoplasmic overexpression of Survivin is associated with increased rate of local progression in patients with locally advanced prostate cancer treated on RTOG 8610. Our data may reflect the divergent functions of Survivin isoforms, which exhibit distinct subcellular localizations. This study represents a retrospective correlative analysis performed on a prospective phase III study. The majority of patients entered this trail prior to the PSA era, and the extent of disease at entry is much more advanced than in the patients we see now who are being diagnosed in the second decade of the PSA era. Our data should be viewed with some caution because of the small dataset, low statistical power and potential cutpoint bias. A larger independent cohort of men with earlier-stage prostate cancer treated with radiotherapy should be investigated to clarify the potential of Survivin in prostate cancer prognosis.
This study was supported by RTOG U10CA21661, CCOP U10CA37422, Stat U10CA32115 from the National Cancer Institute (NCI) to RTOG and the Pennsylvania Department of Health and R01 CA101984-01 (both to A Pollack).
Conflicts of Interest Notification: none.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.