Serum Macrophage Inhibitory Cytokine-1 Concentrations Correlate with the Presence of Prostate Cancer Bone Metastases Free

Macrophage-inhibitory cytokine-1 (MIC-1), also known as PLAB, prostate-derived factor, GDF-15, and NAG-1, is a divergent member of the transforming growth factor β superfamily (1, 2). MIC-1 protein is synthesized as a 308–amino acid propeptide which, when secreted, binds to local extracellular matrix and subsequently becomes cleaved by a furin-like protease. The mature peptide, which is secreted by an alternate pathway, is a 112–amino acid protein which diffuses rapidly into the circulation (1, 3-5). It is likely that extracellular matrix–bound pro–MIC-1 represents a source of local bioactive MIC-1, whereas the secreted mature MIC-1 has more distant effects (6). The nature of the local and remote effects of MIC-1 secretion are not currently clear. However, MIC-1 has been shown by several groups to induce apoptosis and local MIC-1 expression in the stroma of the malignant prostate gland, and has been linked to improved outcome (6).

Although changes in serum MIC-1 levels are associated with a number of disease conditions (7, 8), they are mostly strongly linked to cancer. Increased MIC-1 expression has been documented in a variety of epithelial cancer cell lines, including breast, pancreas, colorectal, and prostate cancers (9-12). Microarray studies have revealed increased expressions of MIC-1 in patients with breast cancer, and serum MIC-1 levels are the best single predictor of the presence of pancreatic carcinoma (11). In colon cancer, increasing MIC-1 expression is associated with the progression of colonic adenomas to invasive cancer and subsequent metastasis, with serum levels at presentation being an independent predictor of subsequent disease-free and overall survival (13). In the case of prostate cancer, serum MIC-1 levels increase with the progression of disease to metastasis (11, 13, 14).

Recently, the MIC-1 gene locus was linked to familial prostate cancer and the most common polymorphism of the MIC-1 gene was associated with a modified risk for the development of prostate cancer (15-18). MIC-1 expression is up-regulated by androgens in murine prostate tissue, but is down-regulated by both androgens and estrogens in the androgen-dependent human LnCaP prostate cancer cell line (2, 19). MIC-1 expression is also increased during the transition from androgen-sensitivity to androgen-independency in an experimental model (20).

Bone metastases are the most common feature of disease dissemination in prostate cancer and may be linked to MIC-1 expression. MIC-1 mRNA is detected in the cartilaginous tissue of rat embryo and ectopically applied MIC-1 also induced early stages of endochondral bone formation (2). Prostate cancer cells known to express high amounts of MIC-1 form sclerotic bone lesions, whereas prostate cancer cells lacking MIC-1 expression produced lytic bone metastases, further supporting a role for MIC-1 in the process of prostate cancer–induced sclerotic bone metastases (20-24).

The aim of this study was to test the hypothesis that there is an association between serum MIC-1 concentrations and the presence of bone metastasis in patients with prostate cancer. Furthermore, we compared serum MIC-1 concentrations with other serum markers associated with bone metastasis. These included cross-linked carboxy-terminal telopeptide of type I collagen (ICTP) and amino-terminal propeptide of type I procollagen (PINP), and prostate-specific antigen (PSA; refs. 25-27). We show a significant relationship between serum MIC-1 concentrations and the presence of bone metastases and show that only serum MIC-1 levels are predictive of future relapse of disease in the bone.

Male patients that were being treated for prostate cancer (n = 159) at the Department of Oncology, University Hospital of Oulu, Oulu, Finland, were recruited for this study during the years 2001 and 2003. A written, signed consent was obtained from the patients before participating in this study, which was conducted in accordance with the local ethic committees. Blood was drawn from the patients upon their scheduled visits at the clinics, processed, and stored at −80°C until analysis. Serum ICTP, MIC-1, PSA, and PINP concentrations were measured from the same sample. The presence or absence of bone metastasis was detected with native X-ray images and verified with bone scans.

Total PSA (ng/L) was measured with AutoDelfia PROSTATUS PSA Free/Total kit, Wallac (Turku, Finland), according to the manufacturer's recommendations (28).

The serum concentrations of MIC-1 (pg/mL) were analyzed using a sensitive immunoassay as previously described (7, 29). Data defining the sensitivity and specificity of MIC-1 sandwich ELISA have been published (7, 29). All samples were assayed in duplicate, and the coefficient of variation between samples was <12%.

The concentration of cross-linked ICTP (μg/L) and amino-terminal PINP (μg/L) were analyzed by specific RIAs, as previously described (30, 31).

The mean levels of the markers were compared between the patients with and without metastasis by Student' t test. Because of the skewness of the distribution of the markers, the data for all markers was transformed to the log scale for analysis. The best predictors of baseline metastases were determined with stepwise discriminatory analysis. The variable selection was based on Wilks' lambda, the likelihood ratio criterion. Variables that met the significance level for the discriminant analysis were considered for inclusion in a multivariate logistic regression model that predicted the probability of the presence of bone metastases. Receiver operator curves (ROC) were constructed for each marker for the prediction of baseline bone metastases and the areas under the curve were computed. The areas under the curves were compared across the four markers by the method of DeLong et al. using the ROC macro which can be downloaded from http://support.sas.com/ctx/samples/index.jsp?sid=520&tab=output. Multivariate analysis of the prognostic factors (biomarkers) for recurrence was based on the Cox proportional hazards model. The graphical plots for ROC curves were generated in PROC LOGISTIC in SAS version 9.1 (32). For all analyses, P < 0.05 was deemed statistically significant.

A summary of the patient characteristics, including previous cancer treatments, are given in Table 1. Of the studied patients, 124 (78%) did not have bone metastases (BM−) and 35 (22%) did have bone metastasis (BM+) at the time when the blood was drawn. The ages of the patients ranged from 46 to 86 years (mean ± SD, 65 ± 7 years). The mean follow-up time in the whole cohort was 36 ± 7.4 months (mean ± SD, minimum 9 months and maximum 49 months). The mean survival time of the patients during the follow-up was 47.5 months [SE, 0.638; 95% confidence intervals (CI), 46.2-48.7 months]. The mean disease-free survival was 42.6 months (SE, 1.185; 95% CI, 40.3-44.9 months), the median disease-free survival time was 36.0 months. The mean and median times to progression at the various sites were as follows: the mean and median times to bone relapse were 47.5 months (SE, 0.687; 95% CI, 46.1-48.8 months) and 36.9 months, respectively. The mean and median times to local relapse were 48.4 months (SE, 0.494; 95% CI, 47.4-49.3 months) and 37.2 months, respectively. The mean and median times to chemical relapse were 43.9 months (SE, 1.08; 95% CI, 41.7-46.0 months) and 36.0 months, respectively.

The mean values of all the studied biomarkers (ICTP, MIC-1, PSA, and PINP) were statistically significantly higher in the group of patients that had bone metastases (n = 35) at the time point when the blood samples were drawn, as compared with those that did not have bone metastases (n = 124) at the same time point (Table 2). Among the patients who experienced bone relapse (n = 7) during the follow-up (until June 2006), the baseline MIC-1 levels were significantly higher than those who did not (P = 0.03). There was a trend toward higher levels of MIC-1 among patients with local relapse compared with those without, but the differences were not significant. Baseline levels of the other measured markers were not significantly different between no-relapse and relapse groups for any site. It has been suggested that MIC-1 production in some colon cancer cells lines is regulated by nonsteroidal anti-inflammatory drugs. These drugs regulate MIC-1 expression independently of cyclooxygenase inhibition (10, 33). There were 25 patients listed as acetylsalicylic acid (Primaspan) users in this cohort. There were no significant differences in baseline serum levels of the biomarkers (MIC-1, ICTP, and PSA) among patients who used acetylsalicylic acid versus those who did not, stratified by metastasis status at baseline (data not shown). Only PINP had a slight difference in baseline mean levels for acetylsalicylic acid users versus nonusers (42.7 ± 23.4 versus 33.4 ± 19.4; P = 0.04) among those with no baseline metastasis.

There were significant differences in MIC-1 at baseline by tumor grade, not stratifying by baseline metastasis. Grade 3 tumors had MIC-1 values that were significantly higher (mean, 2,326.1) than the lower grades (mean, 2,054.1 and 761.5 for grades 2 and grade 1, P = 0.002 and P < 0.0001, respectively). Similarly, Gleason scored tumors of 8 to 10 had significantly higher MIC-1 serum levels (2,647.2) compared with Gleason score 7 (mean, 1,101.4) and Gleason score 2 to 6 (mean, 756.9; P = 0.0003 and P < 0.0001, respectively).

Results of the stepwise discriminate analysis showed that only PINP, MIC-1, and PSA were significantly related to baseline metastasis in bone. In a multivariate logistic model, both MIC-1 (risk ratios, 7.38; 95% CI, 3.57-15.4; P ≤ 0.0001) and PINP (risk ratios, 3.27; 95% CI, 1.39-7.72; P = 0.007) significantly predicted the presence of baseline prostate cancer bone metastasis. PSA levels and ICTP levels were not significant in the adjusted analyses. There was no significant effect of acetylsalicylic acid usage (P = 0.54) in a model with MIC-1 predicting baseline metastasis.

Results of the Cox regression showed that MIC-1 was borderline significantly related to future bone relapse. In a multivariable model predicting bone relapse with MIC-1, the hazard ratio was 1.88 (95% CI, 0.91-3.88; P = 0.09). Thus, higher serum MIC concentrations were associated with higher risk of bone relapse. In a model predicting local relapses, MIC-1 was not significant (hazard ratio, 1.92; 95% CI, 0.73-5.05; P = 0.19).

ROC analyses were done for each serum marker for the prediction of the presence of bone metastasis. The areas under the curve, SEs, and 95% CIs are given in Table 3 for patients presenting with metastasis for each marker. The ROC curves for the baseline presence of bone metastasis for each biomarker are also shown (Figs. 1– 4). Serum MIC-1 level was the best diagnostic marker of bone metastatic disease (Fig. 3, c = 0.9215) which was significantly better than ICTP (Fig. 1, c = 0.813; P = 0.005), PSA (Fig. 4, c = 0.7126; P = 0.002) and PINP (Fig. 2, c = 0.7982; P = 0.03), as judged by the DeLong test of equality of areas (32).

In this retrospective study of 159 patients, serum MIC-1 level was independently associated with tumor grade and independently predictive of the presence of bone metastasis. Comparison of serum MIC-1 level with other markers shown to predict the presence of bone metastasis in prostate cancer, PINP, ICTP, and PSA (34, 35) by ROC analysis confirmed the superior diagnostic capacity of serum MIC-1 determination. The reason for the superior performance of serum MIC-1 compared with other markers of prostate cancer bone metastases is unclear at the moment, but it may be due to the participation of MIC-1 in the process of sclerosis that is associated with bone metastatic prostate cancer (34-37). However, the particular role for MIC-1 in the pathogenesis of prostate cancer needs to be clarified in animal studies.

Increased serum MIC-1 concentrations were significantly associated with increasing grade and Gleason's score. These findings, which suggest that increased serum MIC-1 concentrations are associated with advanced disease in prostate cancer, are in line with previous findings (11, 35, 38-40) MIC-1 was also the only marker studied that was capable of predicting the future occurrence of bone metastases. Because of the retrospective nature of this study, and the small number of patients that developed bone metastasis throughout the course of the study, further prospective studies in larger cohorts are required to confirm our results (40).

Acetylsalicylic acid, a nonsteroidal anti-inflammatory drug, has been shown to up-regulate MIC-1 in human colon cancer cells (33). It is not known whether acetylsalicylic acid also affects MIC-1 secretion in patients with cancer. We did not, however, detect differences between serum MIC-1 concentrations in patients that were taking acetylsalicylic acid, as compared with those that do not take this drug. Although the number of patients on acetylsalicylic acid was small, our findings do suggest that this frequently used drug does not interfere with the clinical use of serum MIC-1 measurements. There are also other drugs that regulate MIC-1 expression in cancer cells (41). Such drugs are not, however, typically used to treat prostate cancer (42-46).

The biomarkers used in this study represent different biological aspects in the pathophysiology of prostate cancer. PINP and ICTP are products of collagen metabolism, and their increased serum concentrations typically reflect pathologic bone turnover in cancer metastases to bone (38, 47-51). MIC-1 and PSA are derived from malignant cells (52, 53). Apart from possibly reflecting tumor volume within the system, both these proteins may also affect the behavior of the cancer cells. For example PSA, as a kallikrein protease, may regulate prostate cancer cell invasion in a zinc ion–dependent fashion (54). PSA expression, in addition to human kallikrein 4, may also mediate epithelial-mesenchymal transition of prostate cancer cells, further suggesting that PSA may also have a functional role in the progression of prostate cancer through the promotion of tumor cell migration (53). The effects of MIC-1 on cancer cells have been described to be of dual nature. In one hand, MIC-1 promotes invasiveness via up-regulating the urokinase-type plasminogen activator (55). MIC-1 has also been shown to decrease cell adhesion and to induce apoptosis in prostate cancer cells (19). In this light, the increased MIC-1 secretion which has been seen in association with prostate cancer progression is unclear. It may, however, be related to acquired insensitivity to the apoptosis-inducing effects of MIC-1. Similar insensitivity to the growth-inhibitory effects of other members of the transforming growth factor β family has been detected in various cancer cells, and this effect has been attributed to mutations in their receptors. These issues will be solved once the currently unknown cellular receptor for MIC-1 is characterized (56-58). Taken together, in addition to their roles as biomarkers, PSA and MIC-1 may exhibit important functional roles in prostate cancer progression at the cellular level.

The process of bone metastasis involves escape from the primary site, and homing at and growing in bone (59). Although growing in bone, metastatic cancer cells secrete various factors that influence the behavior of the surrounding bone cells in the bone microenvironment. More specifically, breast cancer cells typically secrete factors that activate the bone-resorbing osteoclasts, which results in osteolysis at the site of metastases (59). Prostate cancer bone metastases typically involve osteoblast activation, resulting in the formation of new bone (36). Despite the increased serum MIC-1 concentrations in patients with prostate cancer with bone metastases, the possible role of MIC-1 in their pathophysiology is currently unclear. Whether or not MIC-1 actually stimulates osteoblasts and participates in the formation of sclerotic bone metastases remains to be clarified in animal models of prostate cancer bone metastases, in which the metastasis-associated sclerosis is compared between prostate cancer cells that express various amounts and forms of MIC-1.

In conclusion, we show here that serum MIC-1 concentrations exhibit significant correlation with the presence of bone metastasis in patients with prostate cancer. In this regard, MIC-1 outperformed ICTP and PSA, of which especially the latter is typically clinically used in following the disease course. Furthermore, of the studied biomarkers, only serum MIC-1 exhibited significant prognostic value for future relapse in bone. These findings suggest that MIC-1 may be a helpful, new clinical tool in the clinical management of patients with prostate cancer.

Kari Mononen (R.N.) is acknowledged for gathering the clinical data.