Biochemical Markers for Monitoring Response to Therapy: Evidence for Higher Bone Specificity by a Novel Marker Compared with Routine Markers

Prostate cancer is one of the most common cancer types, representing 19% of all cancers diagnosed in 2002, with 679,000 new cases in the western world (1). Bone metastases are common in prostate cancer patients and arise when the primary tumor metastasizes to the bone, causing a lesion of high bone remodeling, which destroys the bone structure. The consequences are devastating symptoms such as severe bone pain, pathologic bone fractures, increased mortality, hypocalcaemia, spinal cord compression, etc. (2, 3).

Bone metastases occur in >50% to 60% of patients with advanced cancer disease (4) and are incurable. Osteoblasts and osteoclasts are affected by the invasive tumor cells in the bone metastases causing increased number, activity, and survival of these bone remodeling cells (5), a phenomenon known as the vicious cycle. Prostate cancer is characterized mainly by sclerotic bone lesions especially in men failing androgen therapy (6).

To initiate successful therapy, it is crucial that bone metastases are detected as early as possible. Early-stage prostate cancer can usually be cured by surgery or radiation therapy, but this is not effective for metastatic disease (6). Treatment is changed radically when a patient develops bone metastases by adding treatment targeting the skeleton. Bisphosphonates are some of the most used bone-targeting drugs for these patients, and effectively inhibits osteoclasts (7). Soft-tissue metastases are, at present time, normally targeted by chemotherapy or by docetaxel (6).

Bone markers are noninvasive, inexpensive, and potentially serve as helpful tools for detecting bone metastases. In our lab, we previously investigated the relation of eight biomarkers to the extent of skeletal involvement in a group of prostate, lung, and breast cancers (8). From this clinical study, we concluded that collagenous markers were the best discriminators in general. The most sensitive marker for bone metastases was a novel marker of high bone turnover alpha-telopeptide collagen type I (ααCTX), which increased the most with increasing number of bone metastases. This marker provides an index of bone resorption of newly formed bone (8-10) by measuring the release of the nonisomerized form of the CTX epitope. The resorption of mature bone can be assessed by measuring the release of isomerized CTX (ββCTX); however, this is not as sensitive for the detection of bone metastases as ααCTX. The CTX epitope is generated by cathepsin K released by osteoclasts, thus being specific for bone resorption (11).

A characteristic of prostate cancer is the high levels of prostate-specific antigen (PSA) due to the increased numbers of prostate epithelial cells. PSA has become the routine marker for screening for prostate cancer, detection of recurrence, bone metastases, and response to treatment (6). Total alkaline phosphatase (tALP) is a bone formation marker, used as a routine marker for skeletal involvement; however, recent publications have pointed to the fact that bone-specific ALP is a better choice as an index of bone formation due to its higher specificity for bone (12-14).

The use of PSA and tALP as routine markers has become a topic for urologist whether these are the appropriate markers for detection of bone metastasis. Therefore, the aim of this study was a head-to-head comparison between these two routine markers and the novel marker ααCTX in two different prostate cancer groups: (a) a cross-sectional study of newly diagnosed patients with clinically localized prostate cancer before radical prostatectomy (these patients were negative for bone metastases yet some had lymph node metastases and were stratified accordingly) and (b) a prospective study of hormone refractory stage prostate cancer stratified by bone metastases. All received docetaxel as palliative treatment but only patients with bone metastases were additionally administered zoledronate. Furthermore, immunohistochemistry was done on metastatic lymph nodes, nonmetastatic lymph nodes, and bone metastases, using the αCTX antibody, to investigate whether the epitope was detectable in such tissues.

Cross-Sectional Study. In this prospective study, a total of 170 histologic confirmed clinically localized prostate cancer patients were referred to the Department of Urology and Paediatric Urology, Philipps-University Marburg between January 2004 and July 2005 for having a radical retropubic prostatectomy (including regional lymphadenectomy). Sixty-eight age-matched men with benign urologic disorders served as controls. Physical examination, abdominal ultrasonography, and bone scanning using technetium-99 scintigraphy, together with X-ray or magnetic resonance imaging whenever appropriate, were used for clinical staging and to verify the absence of bone metastases.

Second void urine samples and serum samples were collected after fasting prior to radical retropubic prostatectomy and stored at −80°C. After surgery, the disease was pathologically staged by the Institute of Pathology, Philipps-University Marburg according to the 2002 Union Internationale Contra Cancrum classification (15). The study was done in accordance with the Helsinki Declaration II and Standards of Good Clinical Practice.

Longitudinal Study. A total of 40 hormone refractory prostate cancer patients were referred to the Department of Urology and Paediatric Urology, Philipps-University Marburg between December 2003 and July 2005 to receive taxan-based chemotherapy. All patients underwent bone scanning in the same manner as in the cross-sectional study to verify the presence or absence of bone metastases. Patients received four treatment cycles each lasting 4 weeks: 35 mg/m² docetaxel in weeks 1, 2, and 3, and no treatment in week 4. Patients with bone metastases additionally received 4 mg zoledronate in week 1 of every cycle. Second void urine samples and serum samples were collected after overnight fasting at baseline and at each cycle in week 1 before treatment and stored at −80°C. The study was done in accordance with the Helsinki Declaration II and Standards of Good Clinical Practice and approved by an institutional review board at the Medical School, Philipps-University Marburg.

Samples were stored at −80°C until assaying. Urine samples were used for estimation of high bone turnover; bone resorption by assessing the level of ααCTX by urinary ALPHA CrossLaps (ref. 9; Nordic Bioscience). Urinary excretion was corrected for creatinine levels measured by an automated urine analyzer (Hitachi-912, Roche). tALP was measured in heparin plasma on a clinical chemistry analyzer (Hitachi 917, Roche Diagnostics). The concentration of PSA was measured in serum samples on the Elecsys 2010 analyzer (Roche Diagnostics). All samples were tested in a blinded manner.

Data shown are mean ± SD, unless otherwise indicated. Baseline demographic characteristics were compared with Student's t test for unpaired observations. The values of ααCTX and PSA were logarithmically transformed to obtain normality. Comparison between levels of each marker in controls, patients with, and patients without lymph node metastasis was done by ANOVA using the General Linear Models Procedure of the Statistical Analysis System (SAS). The same statistical procedure was used for comparison of levels in patients in the longitudinal study during each treatment cycle. Differences and associations were considered statistically significant if P < 0.05.

The analyses were done on paraffin-embedded pelvic lymph node sections from six prostate cancer patients purchased for immunohistology at Biocat GmbH. Three cases were lymph node positive for prostate cancer (N₁) and three cases were negative (N₀).

Paraffin was removed and antigen retrieval was done by overnight incubation of the sections in a Tris-EDTA buffer (pH 9) at 60°C. Sections were blocked in TBS containing 0.5% casein and incubated overnight at 4°C in a moist atmosphere with primary antibody diluted in TBS containing 0.5% casein or control. The primary antibodies used were monoclonal: F44 raised against the αCTX epitope (Nordic Bioscience; 1:45,000 dilution of a 3.18 mg/mL stock) and another raised against cytokeratin against pan-cytokeratin (KL1 Dako; 1:400 diluted from a 0.13 mg/mL stock). A control for αCTX specificity was done by overnight incubation at room temperature of F44 and 50-fold excess of antigen before use. After incubation, the sections were thoroughly washed in TBS and incubated with secondary peroxidase-labeled antibody (Mouse EnVision, DakoCytomation) for 30 min at room temperature. Finally, the peroxidase activity was visualized with 3,3′-diaminobenzidine+ (Sigma), and the nuclei were counterstained with Harris acidified hematoxylin. The lymph node sections were then dehydrated through a gradient of alcohol (70-99%) and toluene and mounted in dibutyl polystyrene xylene. Pictures were taken using an Olympus BX-60 light microscope equipped with an Olympus DP71 digital camera using the Cell A software from Olympus.

The analyses were done on six iliac crest or transpedicular bone metastasis bone biopsies of the spine of metastatic prostate cancer patients from archive material from the University Hospital Hamburg-Eppendorf, Hamburg, Germany, kindly provided by Günter Delling. Specimens were fixed in buffered formalin (4%), decalcified by ultrasonic sound in combination with constant temperature at 24°C, and embedded in paraffin for preparation of 5-μm-thick sections.

Immunohistochemistry on the bone tumor was done in the same manner as described for the lymph node immunohistochemistry. F44 was diluted 1:45,000 from a 3.18 mg/mL stock. Osteoclasts were visualized by their specific tartrate-resistant acid phosphatase activity.

Patients were stratified by the metastatic involvement of their lymph nodes: 24 patients were lymph node positive (N₁) and 146 patients were lymph node negative (N₀). Demographic data are shown in Table 1. There was no significant difference in age between the groups. As expected, the tumor node metastases (T stage) and Gleason score showed that patients in the N₁ group had more progressed disease than those in the N₀ group (Table 1).

The mean levels of PSA, ααCTX, and tALP were stratified according to metastastic disease in the lymph nodes compared with controls, as seen in Fig. 1. PSA increased significantly in both the N₀ group compared with controls (P < 0.001) and in the N₁ group compared with the N₀ group (P < 0.001; Fig. 1A). ααCTX was not elevated in N₀ patients compared with controls (Fig. 1B), whereas it was highly elevated in N₁ patients compared with N₀ (P < 0.001). No difference could be detected in the tALP across the three groups (Fig. 1C).

Patients were stratified by the metastatic involvement of their bones: 26 patients were positive for bone metastases (+BM) and 14 patients were negative for bone metastases (−BM). Six +BM patients did not participate in follow-up due to discontinuing of treatment. Ten patients were lymph node positive in the −BM group. Hormonal therapy was unchanged during the observation period. Mean age was 72 years (range, 42-87 years).

Relative levels of PSA, ααCTX, and tALP were calculated at each visit as the percentage change of the marker level in relation to the baseline level (Fig. 2). PSA (Fig. 2A) showed a significant decrease in both the docetaxel and docetaxel/zoledronate groups at each visit. In contrast, the relative decrease in ααCTX (Fig. 2B) in patients +BM receiving docetaxel/zoledronate was significantly decreased compared with baseline at all time points (P < 0.01 at months 1 and 4; P < 0.05 at months 2 and 3), indicating a response to zoledronate treatment detected by ααCTX. ααCTX was not affected by the docetaxel treatment in the −BM group. For tALP, the levels were decreased significantly in the docetaxel/zoledronate +BM group at visits 2 to 4 (P > 0.001 at month 2, and P > 0.01 at months 3 and 4) but not after 1 month.

The absolute mean values for all three markers at each visit are seen in Table 2.

Immunohistochemistry was done on pelvic lymph node biopsies from six prostate cancer patients with (N₁) or without (N₀) metastatic disease. Figure 3A and B shows staining with cytokeratin and αCTX, respectively, on N₀ lymph nodes. It was verified that the primary cancer had not spread to the lymph nodes due to the absence of cytokeratin staining in the tissue. In adjacent sections, no staining by αCTX was observed. Cytokeratin staining was detected in the N₁ lymph node, confirming the presence of metastatic cancer (Fig. 3C). Staining by αCTX in the N₁ lymph node was observed as island-like staining in the tissue (Fig. 3D). The negative control for αCTX on lymph node sections showed no staining (data not shown).

Immunohistochemistry was done on bone metastasis biopsies from six prostate cancer patients to investigate whether the αCTX epitope was present. Figure 4A to D shows adjacent sections from bone invaded by prostate cancer cells. Cytokeratin staining and areas of hyperchromatic nuclei confirmed the presence of tumor cells in the bone metastases (Fig. 4A). Numerous osteoclasts were revealed by tartrate-resistant acid phosphatase staining showing abnormally high numbers of osteoclasts in the bone tumor (Fig. 4B). In the proximity of tumor cells and osteoclasts, a diffuse αCTX staining was observed (Fig. 4C) with intense staining at sites of high bone turnover. Control showed no staining (Fig. 4D).

This is the first treatment study revealing longitudinal data for the novel high bone turnover marker ααCTX compared with two routinely used markers in patients with different stages of prostate cancer. Furthermore, these are first data on ααCTX in prostate cancer patients stratified by ±lymph node metastasis supported by immunohistochemistry. We made a head-to-head comparison between the two routine markers PSA and tALP used for determining the progression of prostate cancer and ααCTX in a cross-sectional study of prostate cancer patients stratified by lymph node metastasis as well as in a longitudinal treatment study with patients treated with docetaxel and zoledronate.

Briefly, the findings were as follows: (a) crosssection tALP was not elevated in any of the groups compared with controls, whereas PSA was elevated in both cancer groups compared with controls or lymph node–negative patients. In contrast, ααCTX was significantly elevated in lymph node–positive prostate cancer patients compared with lymph node–negative patients, but not in lymph node–negative patients compared with controls. (b) The percentage decrease in ααCTX and tALP in patients +BM, but not in the −BM group, was significant when compared with baseline level. This was not the case for PSA, which decreased in both the −BM and +BM groups during treatment. (c) Immunohistochemistry on lymph nodes showed staining for αCTX in metastatic lymph node. (d) Immunohistochemistry on bone metastasis from prostate cancer patients revealed staining for αCTX in proximity of tumor cells and high number of osteoclasts.

Emerging evidence suggests that biochemical markers have the potential for early detection of bone metastases compared with traditional imaging techniques (8, 13, 14, 16-19). PSA is the most extensively used biomarker for determining prostate cancer stage and for following the response to systemic treatment of metastatic disease. The general use and confidence in PSA as a marker has become controversial because studies have revealed a high occurrence of false-negative diagnosis of prostate cancer done from PSA assessments (20, 21). This may also compromise the use of this marker for determining skeletal involvement.

Physiologically, it is not anticipated that PSA is specific for bone metastases. PSA is highly expressed by prostate cancer cells as well as by normal prostate epithelial cells, and thus is helpful for following tumor behavior, but its use is compromised as indicator for bone metastases in which osteoclasts and osteoblasts are the main cells involved in the vicious cycle. Total ALP has been the most used marker for detecting increased bone formation in metastatic prostate cancer (12), being highly elevated as they mainly develop sclerotic bone lesions. tALP consists of a pool of isoforms originating from various tissues such as liver, kidney, and bone, and in healthy individuals only ∼50% of the tALP originates from bone remodeling (12). ααCTX is a highly bone-specific resorption marker released during the pathology of bone metastases. Here the normal bone turnover is disassociated, and bone resorption occurs on newly formed bone. Although the bone lesions are mainly sclerotic, a high level of bone resorption is ongoing within these lesions together with the fact that many patients will have a mix of sclerotic and lytic lesions (22).

The data represented here show that PSA could discriminate between patients with and without metastatic disease in the lymphs; however, the elevation in patients without lymph node metastasis compared with controls revealed the nonspecificity for metastases. In contrast, ααCTX was more related to the metastastic disease than were PSA and tALP. We speculate that patients with a high risk of having developing or hidden bone metastases with the first cancer cells spread to the skeleton could be identified by ααCTX.

A limited amount of information is available on markers assessed in prostate cancer patients stratified by lymph node metastases. Part of the author group has previously investigated PSA compared with a number of bone markers in which they found the same pattern for PSA (17). However, they reported that tALP was significantly elevated in patients with localized disease and in lymph node–positive patients, which was not observed in the present cohort. The reason for this discrepancy is unclear. To further elucidate the role of tALP, a study with a higher number of patients should be performed. Another group (13) additionally investigated the ability of a formation marker to diagnose and predict metastatic spread in prostate cancer patients stratified by lymph node and bone metastasis compared with nine other serum markers. They showed that bone formation markers, a bone resorption marker, and two osteoclastogenesis markers were elevated in patients with bone metastases, of which the osteoclastogenesis marker had the best discriminating power.

ααCTX and ALP were unaffected by the docetaxel treatment in the −BM group and simultaneously affected by the zoledronate treatment in the +BM group. However, it is clearly seen that there is a trend that tALP decreases with docetaxel treatment of −BM patients, whereas this is not the case for ααCTX. Additionally, it was observed that ααCTX decreased significantly after the first month of treatment with docetaxel/zoledronate, in contrast to tALP, which was significantly decreased after 2 months of treatment. This observation is in line with the knowledge that resorption measurements are more dynamic than formation measurements, as well as the fact that a formation cycle follows a resorption cycle. This implies higher specificity for monitoring of the bone-targeting treatment of bone metastases by ααCTX. Monitoring by PSA and tALP seemed to be influenced by the docetaxel treatment; especially for PSA, the levels were similar in the two patient groups, which complicates the evaluation of skeletal involvement and response to treatment of bone metastases. The inclusion of control groups would have been interesting to further investigate the sensitivity of ααCTX for bone metastases; however, for ethical reasons such a group did not exist.

ααCTX mean levels in the +BM group reached those of the −BM group at visit 2 subsequent to 2 months of docetaxel/zoledronate therapy, indicating a normalization of the bone turnover of young bone within the bone metastases in these patients at this point. Mean levels of PSA were at the same level in both treatment groups at all visits. This suggests that PSA can be used for monitoring docetaxel treatment of the primary tumor cells without any reflection on bone turnover. Normalization was also seen in tALP, although the percentage decrease relative to baseline was the same in the two different treatment groups.

Other groups have compared bone markers with PSA and tALP. Five bone turnover markers were compared with PSA and tALP in a prostate cancer cohort stratified by ±BM treated with zoledronate (23). In this group, none of the markers were significantly elevated at baseline in the progression group compared with patients without progression. During treatment, the CTX marker for mature bone resorption (ββCTX) decreased the most among all markers. As we have assessed the ααCTX, we cannot directly compare the results, but still it seems that in this cohort a collagenous marker was superior to PSA and tALP. In another study, the markers tartrate-resistant acid phosphatase 5b, matrix metalloproteinase-2, and matrix metalloproteinase-9 were compared with PSA and tALP in a prostate group stratified by ±BM (24). Here tALP and PSA were the only markers significantly increased in patients +BM compared with −BM. The matrix metalloproteinase markers were not recommended for detection of bone metastases. This illustrates that not all markers are superior to PSA and tALP. The matrix metalloproteinase–derived collagenous marker COOH-terminal telopeptide of type I collagen was additionally compared with PSA and tALP in a group of prostate cancer patients and evaluated on the relation to number of bone metastases (14). Here, COOH-terminal telopeptide of type I collagen was not well correlated to bone metastases and was not a superior marker for bone metastases compared with PSA and tALP. Because COOH-terminal telopeptide of type I collagen is degraded by cathepsin K produced by osteoclasts, it was, however, not expected to be bone specific but rather specific for matrix metalloproteinase activity (25).

The lymph node–negative stainings revealed no staining for cytokeratin and ααCTX as expected. Cytokeratin staining was positive in the metastatic lymph nodes showing an invasion by prostate cancer cells; however, the positive staining for ααCTX was not expected due to its cathepsin K specificity. Some groups have revealed that prostate and breast cancer cells are able to produce cathepsin K (26, 27), which could explain why ααCTX is located in the presence of tumor cells. Nevertheless, 10 lymph node–positive patients from a total of 14 were included in the −BM group in the treatment study, and ααCTX did not decrease with docetaxel treatment. The docetaxel treatment should have a diminishing effect on the tumor cells that have metastasized to the lymph nodes, and thus should decrease ααCTX during this treatment. This suggests that the ααCTX assessed in urine measurements originates from osteoclastic bone resorption and thus is specific for bone.

The bone staining showing cytokeratin, tartrate-resistant acid phosphatase, and ααCTX stainings within the bone metastases showed that ααCTX could be found in the proximity of tumor cells invading bone tissue, as previously shown in bone metastasis biopsies from breast cancer patients (10).

ααCTX was the only marker unaffected by the primary cancer and elevated in metastatic disease, indicating a potential for identification of patients at high risk of having hidden bone metastases. In addition, the marker was unaffected by docetaxel treatment in patients without bone metastases, whereas it decreased with docetaxel and zoledronate treatment in patients with bone metastases, suggesting specificity for bone-targeting therapy. The epitope, furthermore, was located in the proximity of invading prostate cancer cells in the lymphs and bone.

P. Qvist, M. Karsdal, D. Leeming, and I. Byrjalsen: Nordic Bioscience employees.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.