Author Correction: Decreased expression of bone morphogenetic protein-2 is correlated with biochemical recurrence in prostate cancer: Immunohistochemical analysis

A correction to this article has been published and is linked from the HTML and PDF versions of this paper. The error has not been fixed in the paper.

for multiple members of the BMP family to be examined independently 6 . Specific isoforms of BMPs have shown oncogenic properties whereas others have demonstrated tumour suppressor activity. The current consensus is that BMPs are involved in both the promotion and inhibition of cancer progression depending on the type of cancer cells and the specific isoforms of BMPs.
BMP-2 plays a role in osteoblast differentiation, osteogenesis, and chondrogenesis. BMP-2 also has potential roles in apoptosis and as a retinoid mediator 5 . In previous studies, BMP-2 was shown to promote tumorigenesis and metastatic features in a PCa cell line; however, other studies showed that BMP-2 did not promote tumorigenesis [7][8][9][10] . Consequently, the biological effects of BMP-2 on PCa development and progression remain unclear, as only limited information is available regarding the role of BMP-2 in human PCa.
The European Association of Urology guidelines state that radical prostatectomy (RP) is the only surgical treatment option for localised prostate cancer 11 . However, it is known that approximately 15-30% of RP-treated patients experience biochemical recurrence (BCR) within five years and 40% within 10 years 12 . Previous reports presented that post-RP BCR is correlated with initial serum prostate-specific antigen (PSA) level, Gleason score (GS), post-operative surgical margins (PSM), and pathological stage 11 . However, the predictive accuracy of BCR was not yet satisfactory, and the time to post-RP BCR is the most important predictor of PCa-specific mortality 13 . Therefore, identification of high-risk patients for post-RP BCR will enable early adjuvant treatment for those patients, reducing the risk for disease progression and PCa-specific mortality 14 . In the present study, we examined the expression level of BMP-2 in cancerous prostate tissues via immunohistochemistry and investigated its prognostic significance in predicting post-RP BCR in patients with localised PCa.

Materials and Methods
Ethic statement. This study was approved by the ethics board of the Korea University Anam Hospital (IRB No. 2011AN0228), and all patients provided written informed consent. Patient information was anonymised and de-identified prior to the study, and we carried out all study procedures in accordance with the Declaration of Helsinki guidelines. All methods in this study were performed in accordance with the relevant guidelines and regulations.
Patients and Definition. This research was designed as a retrospective study. Data were gathered from 137 consecutive patients who underwent RP at Korea University Hospital between 2007 and 2010. Among these patients, 27 who received preoperative neoadjuvant therapy and two lost at follow-up were excluded from the analysis. Therefore, the clinicopathologic parameters of 90 patients were analysed in this study, including age, initial PSA, tumour grade according to the GS, tumour stage (T stage), PSM from specimens, follow-up duration after prostatectomy, and PSA levels during follow-up. Follow-up examinations were performed every three months for the first two years after surgery, every six months over the following three years, and on an annual basis thereafter. BCR after therapy was defined as two consecutive serum PSA concentration measurements >0.2 ng/mL, while clinical progression was defined as local recurrence or distant metastasis observed on imaging studies.
Immunohistochemical staining and assessment. Immunostaining was initially assessed by one investigator (H.C.K) and subsequently reviewed independently by a histopathologist (C.H.K), both of whom were blinded to the patient clinical courses. Using the technique of Kononen and researchers, medium-density tissue microarrays were constructed using tissue core biopsies of 1.5 or 2 mm 15 . For immunohistochemistry, tissue slices were collected on glass slides coated with 3-aminopropyltriethoxysilane (APES, Sigma-Aldrich, St. Louis, MO, USA), air-dried, and stored at 4 °C until processing for indirect immunoperoxidase staining, as described by Bobinac and researchers 16 . Tissue slices were deparaffinised in xylene and rehydrated in ethanol. Endogenous peroxidase and nonspecific binding were blocked by incubation in 0.3% H 2 O 2 diluted with methanol and 5% non-immune serum. The sections were incubated with the primary antibody for 60 minutes.
The anti-BMP-2 antibody purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA) was a goat polyclonal antibody raised against a peptide mapping to the amino terminus of BMP-2. After incubation with a primary antibody, the secondary biotinylated antibody was applied according to the manufacturer's protocol (LSAB ® + Kit Peroxidase, Dako, Carpinteria, CA, USA). Peroxidase-conjugated streptavidin was added, and the site of antigen binding was visualised using 3,3′-diaminobenzidine tetrahydrochloride as a chromogen. Sections were counterstained with haematoxylin. Control slides were processed either with normal serum replacing the specific primary antibodies or with the secondary antibody alone.
The area with the strongest positive staining in the intracellular space was selected, and the images were entered into a personal computer using a charge-coupled device colour camera (ICD-740, Ikegami, Tokyo, Japan). All images were converted to 8-bit grey-scale images consisting of 320 × 200 pixels, and their low-intensity areas were counted as the staining area using image-analysis software (NIH Image, National Institutes of Health; Bethesda, MD, USA). The proportion of low-intensity area to total-image area was calculated, and the staining rate was assessed as the mean of ten images. These investigations were conducted without knowledge of the clinical course of the patients. Cytoplasmic expression was assessed as absent (less than 10%), low (10-30%), moderate (30-50%), or high (50% or more) according to an intensity scale 17,18 . In addition, we classified expression of BMP-2 into two groups: the decreased group (absent or weak expression) and the normal group (moderate or high expression) (Fig. 1). p < 0.05 were included in the multivariate Cox regression model. We calculated the BCR-free survival curves using the Kaplan-Meier method and compared them using the log-rank test.

Results
Patients and tumor characteristics. The clinical demographics of the patients are summarised in Table 1.
Prognostic value of BMP-2 in prostate cancer. Among the 90 RP-treated patients, we found BCR during the follow-up period in 36 patients (40.0%). There were significant differences in baseline characteristics of GS, pathologic stage, and BMP-2 expression between the group with BCR and the group without BCR (Table 2).

Discussion
Although the incidence of PCa has continued to increase, patient mortality has decreased remarkably. This phenomenon is partially due to early PCa detection, resulting from the widespread implementation of PSA screening, as well as significant advances in PCa therapeutics 19 . However, the utility of PSA as a prognostic factor in progressive PCa, unlike early-stage PCa, remains controversial. Semi-quantitative methods to evaluate PSA mRNA   Components of the TGF-β signalling pathway have been implicated in PCa regulation, with either tumour suppressor or tumour promoter activities being attributed to the pathway. More specifically, TGF-β signalling exhibits growth inhibitory effects in the early stages of PCa and promotes malignancy in later stages. Disruption of TGF-β signalling is referred to as a metastasis promoter 21 . BMPs, a subgroup of the TGF-β superfamily, were first described as factors inducing bone and cartilage formation; however, the BMP pathway was subsequently shown to be involved in cellular differentiation, organogenesis, chemotaxis, and cellular proliferation 22 . Among the BMPs, BMP-2 is known for its potent activities in inducing the entire cascade of cartilage and ectopic osteogenesis 23 . In addition to bone formation, BMP-2 plays an important role in cell differentiation, proliferation, morphogenesis, and apoptosis 24,25 . Moreover, BMP-2 is considered a putative tumour-suppressor gene in several cancer types.
A number of studies have examined BMP expression in normal and malignant human prostate tissues. Among the BMP family, only BMP-6 and BMP-7 have been shown to be associated with bone metastasis in PCa. Studies have found that osteoblastic bone lesions caused by PCa expressed BMP-6 and -7 26,27 .
However, the mechanism of BMP-2 expression as an indicator of oncological outcomes has not yet been elucidated. Spanjol and researchers reported high expression of BMP-2/4, -6, and -7 in PCa bone metastases 9 . Lai and colleagues also demonstrated that osteoblast-derived BMP-2 activated β1 integrin and β3 integrin and contributed to PCa cell migration 28 Table 4. Results of univariate and multivariate Cox-proportional hazard analysis of clinico-pathological parameters and BMP-2 expression were differentially expressed with regard to the biochemical recurrencefree interval after radical prostatectomy. BMP = bone morphogenetic proteins; CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen; N/A = not applicable; PSM = positive surgical margin; GS = Gleason's score. PCa cell proliferation, tumour growth, and bone metastasis in an animal model 29 . The findings of these previous studies suggest that BMP-2 expression has potential prognostic value for cancer progression or survival in PCa patients. However, most previous studies examined metastatic PCa or were performed in animal models. Some prior research suggested that BMP-2 expression in neoplastic prostate tissue is associated with PCa. For example, Doak and researchers reported that approximately half of the prostate tumours in their study displayed increased copy numbers of the BMP-2, BMP-5, and BMP-7 gene loci, which may account for their abnormal gene expression patterns in neoplastic prostate tissue 30 . However, the authors did not attempt to correlate the gene abnormalities with GS, positive resection margins, or PSA levels owing to the small study population (n = 12) analysed. In contrast, Horvath and colleagues demonstrated that decreased BMP-2 expression was associated with PCa progression, and its loss was associated with progression to a more aggressive phenotype 31 . The authors demonstrated that negative BMP-2 expression was associated with relapse-free survival time in Kaplan-Meier survival analysis and that loss of BMP-2 expression was correlated with increasing GS.
It was notable in the present analysis that among patients with PCa, decreased BMP-2 expression, rather than normal expression, was associated with BCR. In addition, BMP-2 expression was associated with the proportion of BCR and high GS in our cohort. As mentioned above, Horvath and colleagues also reported that loss of BMP-2 expression was correlated with increasing GS, similar to our results. However, they did not demonstrate a prognostic role for BMP-2 expression in the multivariate analysis.
Two other clinicopathologic factors, T stage and GS, were also confirmed as independent factors for predicting post-RP BCR. Many large centre studies have previously shown a similar result. Mani et al. presented in their large study that the strongest predictors of post-RP BCR were GS (over 8) and pathologic stage (>T3) 32 . In a large multicentre long-term follow-up study, among the 1061 patients with pathologic GS (>8), 80% who underwent RP will have experienced BCR by 15 years 33 . However, although many previous studies indicated that PSM could contribute to BCR, the role of PSMs in the development of BCR in patient subgroups with pT2-3a tumour stages remains controversial 34 . The reason that PSM was not significant in this study may be either because of the above results or because of the relatively small sample size of our study. Although this retrospective study included a relatively small sample, the contributions of our study are as follows: to the best of our knowledge, this study is the first to report HRs of cumulative survival rates for decreased BMP-2 expression with its first revelation as an independent marker in a subsequent multivariate analysis. In addition, this was a relatively long-term follow-up study (over eight years in duration) performed to evaluate the prognostic value of BMP-2 with a focus on BCR in patients with PCa, which may assist clinicians in planning subsequent treatment following radical surgery. Furthermore, the other established prognostic factors, T stage and GS, were also analysed as prognostic factors for BCR-free survival in our study cohort. These results showed that our study cohort was representative of the general population.
There were some limitations to our study. As mentioned above, the study has a retrospective design and a relatively small study cohort from a single centre; moreover, some data were censored for some of the variables. In addition, we could not generate curves for progression-free survival (PFS) and overall survival, because patient progression to CRPC and death due to PCa were rare in this cohort. Nevertheless, to the best of our knowledge, this study constitutes the largest study sample to date for evaluation of the prognostic role of BMP-2 in PCa. If a longer study with a larger sample size could be conducted, PFS could be more clearly determined. Moreover, a more meaningful analysis could have been made based on the time to overall survival as well as PFS. In addition, the mechanism of BMP-2 on PCa is not completely understood. One possible explanation, suggested by Horvath et al., is that downregulation of BMP-2 signalling is related to control of cell proliferation 31 . However, there have been no studies to examine disruption of the BMP pathway in PCa. In addition, it is not clear whether this disruption is a primary or secondary phenomenon in prostate carcinogenesis. Further study is needed to more precisely define the mechanism of BMP pathway disruption in the progression of PCa and how this interacts with TGF-β signalling. Finally, we quantified BMP-2 expression by calculating the proportion of low-intensity area to total-image area. Previous researchers, including Asano and colleagues, have reported the current method 18 . However, the method used in this study has potential bias, which involves subjective factors. The BMP-2 level could have been quantified more accurately with western blotting or real-time PCR analysis. Recently, Yang and investigators used RNA analysis and real-time PCR to quantify BMP-2 expression. However, they did not conduct an analysis to associate BMP-2 levels with PCa prognosis 35 .

Conclusion
Decreased BMP-2 expression in PCa tissue was correlated with a poor GS. In addition, it was also correlated with the important prognostic factors of BCR-free survival in patients with PCa. These data suggest that decreased BMP-2 expression in PCa tissue is related to progression to a more aggressive phenotype.