Pro–Prostate-Specific Antigen Measurements in Serum and Tissue Are Associated with Treatment Necessity among Men Enrolled in Expectant Management for Prostate Cancer Free

Prostate cancer is the second leading cause of cancer death among men in the United States, with an anticipated 186,320 newly diagnosed cases and 28,660 deaths in 2008 (1). Overdetection and overtreatment of cases unlikely to cause morbidity represent major dilemmas for prostate cancer management (2). In an effort to reduce the morbidity of overtreatment, expectant management, also known as active surveillance or watchful waiting, with delayed curative intervention has been proposed as a management strategy for low-grade, low-stage prostate cancer (3).

Epstein et al. (4) proposed prostate-specific antigen (PSA) density <0.15 ng/mL/cm³ and favorable diagnostic needle biopsy characteristics (Gleason score <7, <3 cores involved with cancer, ≤50% of any core involved with cancer) as criteria to identify low-grade, low-stage tumors. Men satisfying these criteria are enrolled into a prospective cohort during which they are followed with serial measurements of PSA and repeated biopsies, until tumor characteristics are discovered making them unsuitable for expectant management and treatment is recommended (3). To date, there are very few biomarkers associated with significant outcomes within this cohort (3, 5–9).

Another potential candidate is proPSA (10). Sokoll et al. (11) showed that % [−2]proPSA is the best predictor of prostate cancer, particularly in the 2 to 10 ng/mL total PSA (tPSA) range. proPSA, the precursor of PSA, contains a 7–amino acid pro-leader peptide. Additional truncated forms of proPSA with leader sequences of 5, 4, and 2 amino acids also exist in serum (12). Activational cleavage activity of the leader sequences by human kallikrein-2 and trypsin decreases with decreasing size of the propeptide leader sequence, with [−2]proPSA being resistant to activation. Our group (13) has shown that nuclear structure alterations and [−5/−7]proPSA staining in cancer and benign-adjacent areas (BAA) of prostate tissue can differentiate between patients of native Japanese and American Japanese origin. We sought to assess the association of proPSA staining of biopsy BAA and cancer tissue as well as quantification of serum proPSA from samples taken at diagnosis with unfavorable biopsy conversion on annual surveillance examination in our expectant management cohort.

Patients with signed informed consent were enrolled in our institutional review board–approved expectant management program if they met inclusion criteria [nonpalpable tumor on digital rectal examination (stage T_1c), PSA density ≤0.15 ng/mL/cm³ (PSA before diagnosis divided by prostate volume determined by transrectal ultrasound measurement), and favorable diagnostic needle biopsy characteristics (Gleason score <7, <3 cores involved with cancer, ≤50% of any core involved with cancer)]. Patients were surveyed semiannually with serum tPSA, free PSA (fPSA), and digital rectal examination. An annual surveillance biopsy was also done, and curative intervention was recommended if pathology were unfavorable (Gleason score ≥7, Gleason pattern 4/5, ≥3 cores involved with cancer, >50% of any core involved with cancer).

Serum was obtained before biopsy and stored at −80°C until testing. Serum specimens were analyzed in The Johns Hopkins University Clinical Chemistry Research Laboratory on the Beckman Coulter ACCESS immunoassay system for tPSA, fPSA, and [−2]proPSA. The assays are all dual monoclonal sandwich assays using Hybritech antibodies and a chemiluminescent detection system. The assays for tPSA and fPSA are commercially available, whereas the assay for [−2]proPSA is for research only (14). The [−2]proPSA assay is calibrated using [−2]proPSA purified from the AVA12-PSA mammalian cell line. The assay has linear ranges of <1 to 5,000 pg/mL for [−2]proPSA with intra-assay and interassay precision of 2.3% to 5.3% and 2.7% to 3.5% (range, 9-69 pg/mL), respectively. Cross-reactivity of other PSA isoforms in the assay is minimal.

The [−5/−7]proPSA antibody was provided by Beckman Coulter research and development (Beckman Coulter). Immunohistochemistry was done using a DAKO Autostainer Universal Staining System (DAKOCytomation). After dewaxing and dehydration, biopsy sections were placed in a rice steamer with antigen retrieval solution (DAKO) for 20 minutes. Biopsy sections were next pretreated with 0.3% hydrogen peroxide for 10 minutes to remove endogenous peroxidase activity. Subsequently, biopsy sections were incubated for 1 hour at room temperature with [−5/−7]proPSA antibody (6.31 ng/mL) at 1:320 dilution and followed by detection with the envision plus kit (DAKO). Biopsy sections were incubated with horseradish peroxidase–labeled polymer-secondary antibody for 20 minutes and then with freshly prepared liquid 3,3′-diaminobenzidine + substrate-chromogen solution for 5 minutes. Biopsy sections were counterstained with hematoxylin for 1 minute.

Next, [5/−7]proPSA immunohistochemistry-stained BAA and cancer areas of the same slide were marked as such by J.I.E. for each patient. Images were individually captured using the Zeiss Axioskop microscope at ×200 magnification (Supplementary Fig. S1). Because marked cancer areas on biopsy cores for most patients were very small, only one ×200 field-of-view was captured in each BAA and cancer area. We used Image-Pro Plus version 6.0 (Media Cybernetics) software to quantify [−5/−7]proPSA immunohistochemistry as well as to calculate proPSA (3,3′-diaminobenzidine) % immunohistochemistry area and staining intensity.

Because the original biopsy cores from men in the expectant management cohort had very limited amounts of cancer (only small foci in certain instances), further cutting into those blocks sometimes produced slides without visible cancer. As we wanted to study staining differences in both the cancer and BAA, cases were excluded from immunohistochemical analysis if cancer tissue was not available for a given case.

All data were analyzed using Stata version 10.0 statistical analysis software. The Mann-Whitney test was used to determine distribution differences across favorable and unfavorable biopsy groups. Bivariate Cox proportional hazards regression was used to identify significant prognostic factors for unfavorable biopsy conversion on annual surveillance examination. Ties were handled by the Breslow method. The proportional hazards assumption was verified by examination of residual plots and Schoenfeld residuals. We determined optimal cutoffs to dichotomize continuous variables using the classification and regression tree method. The Kaplan-Meier analysis and the log-rank test were used to test equality of survivor functions across two groups. Statistical significance in this study was set as P ≤ 0.05.

Of the 71 prostate cancer patients from the Johns Hopkins expectant management cohort, with banked serum and tissue available from the time of diagnosis, 39 developed unfavorable biopsies and 32 maintained favorable biopsies on annual surveillance (median follow-up, 3.93 years). Demographic, clinical, and pathologic information for favorable and unfavorable biopsy groups is shown in Table 1.

In bivariate Cox regression analysis, age (P = 0.860), tPSA (P = 0.441), fPSA (P = 0.948), % fPSA (P = 0.433), PSA density (P = 0.816), [−2]proPSA (P = 0.280), number of cores involved with cancer (P = 0.780), maximum % core involved with cancer (P = 0.469), cancer [−5/−7]proPSA % area (P = 0.163), and cancer [−5/−7]proPSA stain intensity (P = 0.301) were not significantly associated with unfavorable biopsy conversion.

Serum [−2]proPSA/% fPSA (P = 0.017), BAA [−5/−7]proPSA % area (P = 0.019), and BAA [−5/−7]proPSA stain intensity (P = 0.003) were significantly associated with unfavorable biopsy conversion (Fig. 1; Table 2). Further, serum [−2]proPSA/% fPSA significantly correlated with BAA [−5/−7]proPSA % area (ρ = 0.403; P = 0.002) and BAA [−5/−7]proPSA stain intensity (ρ = 0.325; P = 0.016) but not with cancer [−5/−7]proPSA % area (ρ = 0.187; P = 0.164) or cancer [−5/−7]proPSA stain intensity (ρ = 0.068; P = 0.614). To avoid multicollinearity, serum [−2]proPSA/% fPSA, BAA [−5/−7]proPSA % area, and BAA [−5/−7]proPSA stain intensity were not evaluated in multivariable analysis due to our small sample size and significant correlations between these variables. Standard clinicopathologic parameters were not evaluated in multivariable logistic regression because none of them were associated with unfavorable biopsy on bivariate analysis.

Overdetection and overtreatment of prostate cancer is an important public health issue among older men in the United States (15). Although the 12-year update of the Scandinavian Prostate Cancer Group Trial showed decrease in prostate cancer–specific mortality across the entire cohort, the absolute decrease in prostate cancer–specific mortality in the subset of men ages ≥65 years randomized to radical prostatectomy versus watchful waiting was only 0.1% [13.1 (8.8-19.5) versus 13.2 (8.9-19.6); ref. 16]. Further, men ages ≥65 years in the radical prostatectomy group had 2.7% higher overall mortality compared with those in the watchful waiting group [42 (35-50.5) versus 39.3 (32.5-47.7); ref. 16]. Thus, a majority of older men diagnosed with screen detected prostate cancer may not gain survival advantage with curative intervention.

Since 1995, our department has prospectively enrolled and followed an expectant management cohort for men with low-grade, low-stage tumors. Curative intervention is suggested by a change in clinical examination or unfavorable pathology on annual surveillance biopsy (3). Although delayed surgical intervention for low-grade, low-stage tumors does not appear to compromise curability compared with immediate surgical intervention (17), it would be helpful to patients, clinicians, and researchers to identify those men at increased risk of developing an unfavorable biopsy earlier in the course of their disease. Because the patients are effectively matched for tPSA, Gleason grade, and clinical stage at the time of entry into the study, we wanted to investigate the role of both tissue and serum proPSA isoforms, collected at the time of entry into expectant management, in determining subsequent unfavorable biopsy conversion.

There has been recent interest in the role of proPSA in prostate cancer early detection and prognosis. Both the overall percentage of proPSA (proPSA/fPSA ratio) and the levels of its truncated forms (particularly [−2]proPSA) have been able to determine the presence of prostate cancer. Sokoll et al. (18) showed that % proPSA could reduce unnecessary biopsies among men with tPSA between 2.5 and 4.0 ng/mL. This result has been validated in a multi-institutional cohort of men with tPSA between 2 and 10 ng/mL (11). Men with tPSA between 4 and 10 ng/mL who were diagnosed with prostate cancer have shown higher fractions of proPSA than men with similar tPSA and no evidence of prostate cancer (19). Catalona et al. (20) showed that higher preoperative proPSA was associated with higher-grade disease and more advanced pathology at the time of surgery. Stephan et al. showed, using a neural network, that ratios of proPSA/% fPSA are associated with features of aggressive prostate cancer among men undergoing prostate cancer screening (21).

There has been a great deal of interest in the research community to determine prognostic biomarkers for prostate cancer managed with watchful waiting. PSA kinetics (8), p53 nuclear staining (22), Ki-67 (6), microvessel density (23), neuroendocrine differentiation (24), TMPRSS2-ERG fusions (5), and apparent diffusion coefficient on diffusion-weighted magnetic resonance imaging (25) have shown prognostic value in patients managed by watchful waiting. Notably, our approach for selecting and monitoring patients (T_1c, Gleason score <7, <3 cores involved with cancer, ≤50% of any core involved with cancer) differs from those described by others, who may enroll men with T₂ lesions and Gleason 7 tumors (5, 6, 8, 16, 22–25), and may be considered more conservative. Because of this more highly stringent entrance criteria, the patients in our cohort are better “matched” and have fewer characteristics differentiating them. Our group has also shown the clinical utility of several biomarkers in this cohort (7, 9).

Recently, Hoshida et al. (26) showed that gene expression profiles of tumor tissue failed to show a significant association with survival in patients with hepatocellular carcinoma, whereas profiles of the surrounding nontumor liver tissue were highly correlated. Alterations in DNA content are present in both BAA and cancer tissue areas of prostate cancer and represent the upregulation of proliferation-related genes including transcription factors, signal transducer, and growth regulators (27, 28). Additionally, mitochondrial DNA alterations are known to be present in prostate cancer and histologically normal-appearing adjacent prostate glands (29).

In the current study, we sought to determine the association of serum and tissue (both cancer and BAA) proPSA levels with outcomes among men enrolled in expectant management for prostate cancer. We showed that serum [−2]proPSA/% fPSA, BAA [−5/−7]proPSA % area, and BAA [−5/−7]proPSA immunohistochemistry stain intensity are associated with unfavorable biopsy conversion on annual surveillance biopsy (Fig. 1; Table 2). Further, serum [−2]proPSA/% fPSA significantly correlated with [−5/−7]proPSA % area in BAA tissue as well as BAA [−5/−7]proPSA stain intensity. However, cancer tissue [−5/−7]proPSA % area and cancer tissue [−5/−7]proPSA immunohistochemistry stain intensity were not associated with unfavorable biopsy conversion and did not correlate with serum [−2]proPSA/% fPSA. Based on these results, we formulated a novel hypothesis: potentially, the increase in serum proPSA/% fPSA is driven by increased proPSA production from prostate cells in BAA of the prostate.

Our study has many strengths. We have one of the largest expectant management cohorts of men with prostate cancer. Our cohort is unique in its stringent entry requirements and close follow-up. It also has the advantage of having banked serum and tissue samples. However, our study has limitations, the most important of which is the limited sample size of men having both serum and tissue samples available for study. Another limitation is the use of prostate biopsy status as a defined endpoint; because prostate biopsies sample only a very small fraction of the gland, our results are prone to verification bias. Our observations and interpretations can only be viewed as exploratory and hypothesis-generating at this juncture because of the problem of examining multiple variables in a small cohort. Future research needs to expand the use of PSA isoforms in combination with other molecular biomarkers to assess prognosis of low-grade, low-stage prostate cancer in expectant management. Such biomarkers must be examined in larger, potentially multi-institutional cohorts in prospective trials with prespecified analysis plans.

In conclusion, measurement of serum and tissue levels of proPSA, at the time of diagnosis, are associated with future unfavorable biopsy conversation and could potentially determine which men enrolled in an expectant management cohort will ultimately require treatment for prostate cancer.

L.J. Sokoll has received research grant support from Beckman Coulter. The other authors have no conflicts of interest to disclose.