Bone Turnover Markers as Predictors of Mortality Risk in Prostate Cancer Patients with Bone Metastases Following Treatment with Zoledronic Acid

Transcription

1 EUROPEAN UROLOGY 59 (2011) available at journal homepage: Prostate Cancer Bone Turnover Markers as Predictors of Mortality Risk in Prostate Cancer Patients with Bone Metastases Following Treatment with Zoledronic Acid Klaus Jung a,b, *, Kurt Miller a, Manfred Wirth c, Michael Albrecht d, Michael Lein a,b,e a Department of Urology, University Hospital Charité, Berlin, Germany b Berlin Institute for Urologic Research, Berlin, Germany c Department of Urology, Hospital of Technical University Carl Gustav Carus, Dresden, Germany d Novartis Pharma GmbH, Nuremberg, Germany e Department of Urology, University Teaching Hospital, Offenbach, Germany Article info Article history: Accepted December 10, 2010 Published online ahead of print on December 22, 2010 Keywords: Prostate cancer Bone metastases Bone markers Overall survival Abstract Background: Clinical data have limited validity for predicting the survival of prostate cancer (PCa) patients with bone metastases. There is a need to improve the predictive evidence both for clinicians and patients. Objective: To evaluate the predictive ability of serum bone markers for mortality risk in PCa patients with bone metastases. Design, setting, and participants: We conducted a survival analysis in relation to bone markers in a subgroup of 52 patients treated with zoledronic acid (4 mg every 4 wk for 15 mo) in a prospective, multicentre trial during , about 4 yr after the end of the trial. Measurements: Serum levels of total and bone-specific alkaline phosphatase, amino-terminal procollagen propeptides of type I collagen (PINP), cross-linked N-terminal (NTx) and cross-linked C-terminal telopeptides of type I collagen (ICTP), C-terminal telopeptides of type I collagen, prostate-specific antigen from the last visit of the treatment study, and clinical data were related to the overall survival (OS) status of patients in the follow-up. Univariate and multivariate Cox regression analyses with internal bootstrapping validation and concordance index calculations were performed. Results and limitations: Out of the 52 patients followed, 34 died within a median followup of 13.8 mo, and 18 patients were alive at a median follow-up of 43.8 mo. The patients who died within the follow-up period had significantly higher concentrations of ICTP, NTx, and PINP than the surviving patients. Cox regression models with clinical data and bone markers showed that ICTP and PINP were most predictive for mortality risk in addition to the occurrence of skeletal-related complications and the continuation of treatment with zoledronic acid. Internal validation confirmed the reliability of the results, although the sample size was small. Conclusions: PINP and ICTP can be considered suitable predictors for the OS of PCa patients with bone metastases. # 2010 European Association of Urology. Published by Elsevier B.V. All rights reserved. * Corresponding author. Department of Urology, University Hospital Charité, Schumannstr. 20/21, Berlin, Germany. Tel ; Fax: address: klaus.jung@charite.de (K. Jung) /$ see back matter # 2010 European Association of Urology. Published by Elsevier B.V. All rights reserved. doi: /j.eururo

2 EUROPEAN UROLOGY 59 (2011) Introduction Several studies of prostate cancer (PCa) patients with bone metastases have suggested that bone markers in serum may be as useful tools for diagnosing bone metastases and predicting skeletal-related events (SRE) as complications [1 6]. Bone markers were also used to monitor the efficacy of bisphosphonate therapy on osseous metastases in these patients [1,5,7,8]. In connection with these trials, bone markers were included in prognosis models for the estimation of overall survival (OS) [1,7,9,10]. However, two limitations impede the evaluation of their prediction capability in terms of their survival probability in relation to bisphosphonate treatment: (1) the use of only single or a few but different bone markers and (2) estimation of prediction based on pretreatment values, while the various serum bone markers differentially decrease following bisphosphonate treatment [8]. We have previously evaluated the clinical validity of the bone formation markers total alkaline phosphatase (talp), bone-specific alkaline phosphatase (balp), and amino-terminal procollagen propeptides of type-i-collagen (PINP) as well as the bone resorption markers cross-linked N-terminal (NTx) and cross-linked C-terminal (CTx) telopeptides of type-i-collagen, and C-terminal telopeptides of type-i-collagen (ICTP) in metastatic prostate cancer patients under treatment with zoledronic acid [5,8]. Using the bone marker values measured at the completion of the treatment after 15 mo, in the present study we aimed to assess the usefulness of these markers for predicting the survival outcome of patients. 2. Materials and methods 2.1. Study group The present study group consisted of a subgroup of 56 patients from a total study population of 308 PCa patients with bone metastases enrolled in a German multicentre clinical trial performed between 2002 and 2005 (Protocol no. CZOL446EDE07). Detailed information about this trial is available elsewhere [5,8]. The primary purpose of the trial (herein referred to as the treatment study) was to evaluate the rate of SRE in PCa patients with bone metastases receiving zoledronic acid (Zometa, Novartis Pharma, Basel, Switzerland; 4 mg every 4 wk over 15 mo). During the administration of study medication, standard antineoplastic therapies, including androgen suppression, and cytotoxic chemotherapy agents were also allowed. In the present study, performed between June and December 2009 as a retrospective evaluation of a patient subgroup, the end point of the analyses was OS. Preconditions for including patients in the present investigation were the availability of all bone marker data at the completion of the study and patient consent that they would be informed of their survival status after completion of the original study by contacting the physician responsible for follow-up Analytical methods PINP, talp, balp, NTx, CTx, ICTP, and prostate-specific antigen (PSA) were measured as previously described [5,8]. We measured talp using the Roche/Hitachi Modular Analytics System (Roche Diagnostics, Mannheim, Germany); balp using the Hybritech Ostase test (Beckman-Coulter, Fullerton, CA, USA); NTx using the enzyme-linked immunosorbent assay (ELISA) Osteomark NTx assay (Ostex, Seattle, WA, USA); ICTP using the ELISA kit from Orion (Espoo, Finland); and CTx, PINP, and PSA using the Elecsys 2010 Chemistry Analyzer (Roche Diagnostics) Statistical analyses Statistical analysis was performed using the PASW Statistics v.18 (SPSS, Chicago, IL, USA) with the bootstrap module and GraphPad Prism v.5.03 (GraphPad Software, La Jolla, CA, USA) with nonparametric tests (Mann-Whitney U test, x 2 test, Fisher exact test, Spearman rank correlation), Kaplan-Meier curves, logistic regression, and Cox proportional hazards regression analyses. Two thousand bootstrap resamplings were used to calculate the bias corrected, accelerated 95% confidence intervals (CI) and to perform internal validations. MedCalc v (MedCalc Software, Mariakerke, Belgium) was used to calculate receiver operating characteristics (ROCs) together with bootstrap software for estimating the overfitting bias for the area under the curve (AUC) [11]. The concordance index was calculated in goodness-of-fit of models [12]; p < 0.05 was considered statistically significant. For sample size and power calculations (GraphPad StatMate v.2.0, GraphPad Software), a proportional difference of 0.40 in the Kaplan- Meier curves was considered a useful measure for differentiating among patients with different risk concerning OS. Under the one-sided conditions of a = 5% and b = 80% that were relevant for this problem, 19 subjects in each group had to be investigated to avoid type I and type II errors. Assuming a clinical test as acceptable with an AUC of about 0.80, a sample size of 22 patients in each group with a different mortality risk was calculated (MedCalc software). Thus, to ensure a power of at least 80% with probably unequal numbers in the groups (ratio of 2:3), about 50 patients had to be investigated. 3. Results 3.1. Patient demographics Questionnaires were sent to the treating centres 49 mo (median: 95% CI, ) after the completion of the treatment study. Information was obtained for 52 of the 56 patients. All patients in the present study were on androgen-deprivation therapy before the initiation of the treatment study; this therapy was accompanied by chemotherapy in 12 patients. Thirty-four patients had died, and 18 patients were alive at the time of the last follow-up. Table 1 summarises the characteristics of the cohort Bone markers and prostate-specific antigen The concentrations of the bone markers and PSA in the two groups at the end of the treatment study revealed a relationship with survival in the follow-up (Table 1). ICTP, NTx, and PINP were higher in the deceased group compared with the survivors ( p < 0.05). PSA, tap, bap, and CTx tended to be associated with survival time ( p < 0.15). However, bone markers in patients on chemotherapy before and during the treatment study who could be considered in the hormone-refractory state were not different compared to the values in patients without chemotherapy (at least p = 0.249). The markers were strongly correlated with each other and with PSA (Table 2).

3 606 EUROPEAN UROLOGY 59 (2011) Table 1 Clinical characteristics of the study groups Variable Total * (n = 52) Patients alive (n = 18) Patients deceased (n = 34) ** Age, yr 68 (67 69) 66 (65 73) 69 (67 73) Cancer duration before the treatment study, mo 17.2 ( ) 5.7 (1.2 35) 28.9 ( ) Cancer duration at completion of treatment study, mo 32.7 ( ) 20.8 ( ) 43.6 ( ) Survival interval after the treatment study, mo 14.5 ( ) 43.8 ( ) 13.8 ( ) SRE before the treatment study, no. (%) 7 (13.5) 1 (5.6) 6 (17.6) SRE during the treatment study, no. (%) 14 (26.9) 2 (11.1) 12 (35.3) SRE until completion of the treatment study, no. (%) 16 (30.8) 2 (11.1) 14 (41.1) SRE before, during, and after the treatment study, no. (%) 26 (50) 7 (38.9) 19 (55.9) ECOG status at the final visit of the treatment study, no. (%) (32.6) 8 (44.4) 9 (26.5) 1 25 (48) 7 (38.9) 18 (52.9) 2 10 (19.2) 3 (16.7) 7 (20.6) Therapy after completion of the treatment study, no. (%) *** Only chemotherapy 1 (1.9) 1 (5.6) 0 (0) Only HT 1 (1.9) 0 (0) 1 (2.9) Only bisphosphonate 2 (3.9) 1 (5.6) 1 (2.9) HT with chemotherapy or RT 3 (5.8) 0 (0) 3 (8.8) Combined therapy (chemotherapy, HT, and/or RT) 42 (80.8) 16 (88.9) 26 (76.5) and bisphosphonate Not declared 3 (5.8) 0 (0) 3 (8.8) Marker values at the end of the treatment study: PSA, mg/l 31.5 ( ) 3.5 ( ) 32.9 ( ) talp, U/l 67 (60 90) 57 (43 67) 87 ( ) balp, ng/l 18.3 ( ) 12.0 ( ) 21.5 ( ) PINP, mg/l 30.3 ( ) 13.9 ( ) 48.6 ( ) NTx, nmol/l 12.6 ( ) 10.2 ( ) 13.4 ( ) CTx, ng/l 53.5 (37 100) 50 (11 54) 56 (37 162) ICTP, mg/l 4.37 ( ) 3.34 ( ) 5.15 ( ) SRE = skeletal-related events; ECOG = Eastern Cooperative Oncology Group; HT = hormone therapy; RT = radiation therapy; PSA = prostate-specific antigen; talp = total alkaline phosphatase; balp = bone-specific alkaline phosphatase; PINP = amino-terminal procollagen propeptides of type I collagen; NTx = crosslinked N-terminal telopeptides of type I collagen; CTx = cross-linked C-terminal telopeptides of type I collagen; ICTP = C-terminal telopeptides of type I collagen; CI = confidence interval. * Values are given as medians, with bias corrected and accelerated 95% CI in parentheses calculated by 2000 bootstrap cycles or numbers of patients with percentages in parentheses. ** Significances were tested by the Mann-Whitney test, with Monte Carlo simulations or the x 2 test and the Fisher exact test for the comparison of proportions. *** Levels of all bone markers were not significantly different between patients with or without a particular therapy (HT, chemotherapy, or RT; s between and 0.861). Therefore, the patients were subdivided as given. Chemotherapy, n = 2; radiation, n =1. HT, n = 8 (living group) and n = 6 (deceased group); chemotherapy, n = 2 and n = 2; HT and chemotherapy, n = 3 and n = 11; HT and RT, n = 1 and n = 2; HT, chemotherapy, and RT, n = 4 and n = Bone markers as predictors of mortality risk The end point of the analyses was OS measured from the date of treatment study completion to the time of the last follow-up or death. To determine the association of the variables with outcome, patients were stratified into two groups according to the dichotomous criteria given in Table 3. Univariate and multivariate risk factor analyses were performed. In univariate analyses, all bone markers, PSA, and SRE were predictive of mortality risk. For the most significant markers (ICTP and PINP), Kaplan-Meier analyses demonstrated distinctly different survival curves (Fig. 1). In multivariable analyses including clinical factors as a full model, only SRE was significant. In a separate analysis including the six bone markers, ICTP and PINP remained a significant combination after backward elimination (Table 3). Therefore, combining this promising bone marker combination with the clinical variables, only SRE as an indicator of advanced disease and the two bone markers Table 2 Correlation coefficients according to Spearman between bone markers and prostate-specific antigen for all patients after completion of the study * balp PINP NTx CTx ICTP PSA talp balp PINP NTx CTx ICTP 0.57 balp = bone-specific alkaline phosphatase; PINP = amino-terminal procollagen propeptides of type I collagen; NTx = cross-linked N-terminal telopeptides of type I collagen; CTx = cross-linked C-terminal telopeptides of type I collagen; ICTP = C-terminal telopeptides of type I collagen; PSA = prostate-specific antigen; talp = total alkaline phosphatase. * Significant correlations in all cases for at least of p < 0.01.

4 Table 3 Univariate and multivariate Cox regression analyses of clinical factors and bone markers for survival prediction after completion of zoledronic treatment Variable Dichotomous criteria * Separate full models with clinical variables or bone markers Multivariate Combined model using clinical variables and bone markers ** Univariate Inclusion Backward Inclusion Backward Clinical data Age, yr ( ) ( ) ( ) Cancer duration until completion of the study, mo SRE until completion of the study ECOG status at the final study visit Treatment after study ( ) ( ) ( ) Absence/ presence 3.40 ( ) ( ) ( ) ( ) ( ) ( ) ( ) Without/with bisphosphonate 0.22 ( ) ( ) ( ) PSA, mg/l ( ) ( ) ( ) Bone markers talp, U/l ( ) < ( ) balp, ng/l ( ) ( ) PINP, mg/l ( ) < ( ) ( ) ( ) ( ) NTx, nmol/l ( ) ( ) CTx, ng/l ( ) ( ) ICTP, mg/l ( ) < ( ) ( ) ( ) ( ) HR = hazard ratio; CI = confidence interval; SRE = skeletal-related events; ECOG = Eastern Cooperative Oncology Group; PSA = prostate-specific antigen; talp = total alkaline phosphatase; balp = bone-specific alkaline phosphatase; PINP = amino-terminal procollagen propeptides of type I collagen; NTx = cross-linked N-terminal telopeptides of type I collagen; CTx = cross-linked C-terminal telopeptides of type I collagen; ICTP = C-terminal telopeptides of type I collagen. * Calculations were performed using dichotomised data based on median values (less than vs greater than or equal to median) for the variables age, cancer duration, PSA, and bone markers at the last visit of the 15-mo zoledronic treatment study or according to the other indicated criteria. ** The combined model with clinical variables and bone markers included the itemised clinical variables, PSA, and the two bone markers ICTP and PINP, which showed s <0.10 in the bone marker full model and remained as the two single markers after backward elimination. Reduced model following backward likelihood elimination (entry p = 0.025, removal p =0.05). EUROPEAN UROLOGY 59 (2011)

5 608 [()TD$FIG] EUROPEAN UROLOGY 59 (2011) showed p < Both bone markers again remained in the model as the only significant markers after backward elimination (Table 3). A similar result was obtained when a Cox regression analysis with backward elimination was performed using all bone markers and clinical variables (Table 4). Internal validation of the models was performed by bootstrapping and estimation of the concordance index [12]. The bootstrap calculations generally confirmed the p values of the conventional Cox regression analyses, but larger ranges of 95% CI of the hazard ratios (HRs) occurred (Table 5). The concordance index, indicating the probability that the Cox regression model would predict the correct order of survival times of two randomly drawn patients, showed acceptable results for the bone markers and particularly for the combination ICTP plus PINP (Table 6). An index value of 1.0 is equivalent to a perfect prediction, while a value of 0.5 corresponds to a coin flip. ROC analyses were applied to evaluate the capability of the markers for predicting death within the following 24 mo (Table 7). Using 2000 bootstrap resamples, CTx, ICTP, and PINP showed AUC values >0.7, with sensitivities/specificities between 70% and 80%. The combination of ICTP plus PINP or the inclusion of all markers did not differentiate better between surviving and deceased patients than ICTP or PINP alone. 4. Discussion Fig. 1 Kaplan-Meier analysis of overall survival in patients after the completion of treatment with zoledronic acid according to the bone markers telopeptides of type I collagen (ICTP) and procollagen propeptides of type I collagen (PINP). The stratification was performed using the median concentration of (A) ICTP and (B) PINP as a dichotomous criterion at the last visit in the treatment study. Differences were tested by the log-rank test. The median survival times for patients with ICTP and PINP values above the corresponding median values were 8.1 mo (95% confidence interval [CI], ) and 11.0 mo (95% CI, ), respectively, whereas patients with marker values below the median had a median survival time of 50.6 mo (95% CI, ) and 50.7 mo (95% CI, ), respectively. PINP = procollagen propeptides of type I collagen; ICTP = telopeptides of type I collagen. This report proves that bone markers especially ICTP and PINP could be used as survival predictors in patients with advanced PCa who are treated with the bisphosphonate zoledronic acid. Reliable information regarding survival probability would be helpful for clinicians when deciding on further treatment management but also for patients who want to be fully informed about their health status in order to control their own advance care planning at the end of life. Numerous prognostic models exist to predict the recurrence of PCa after radical prostatectomy (RP) [13], but only a few models have been developed for survival/mortality prediction in patients with metastatic PCa [14 19]. The tools for metastatic patients generally use clinical data and routine laboratory parameters such as haemoglobin, lactate dehydrogenase, and PSA [14,16,18]. Although the metastatic spread of PCa is characterised by osseous metastases, bone markers have rarely been considered as predictors of survival. In contrast, where they have been used, either only single or a few markers were investigated [2,9,10]. The characteristics of the PCa bone metastases with both increased bone resorption and formation processes [20,21] are typically reflected by increased circulating markers of both processes [6,9,22,23]. Therefore, comparative measurements of formation (talp, balp, PINP) and resorption markers (ICTP, CTx, NTx) as presented in this study are advisable for determining the most suitable markers. Our previous studies in metastatic PCa patients showed that both PINP and ICTP were most indicative in predicting metastatic progression and skeletal complications, respectively [5,8]. Other studies similarly reported on the

6 EUROPEAN UROLOGY 59 (2011) Table 4 Multivariate Cox regression analysis using a full model of all clinical factors and bone markers for survival prediction after completion of zoledronic treatment * Variable Dichotomous criteria Full model Inclusion Backward ** Age, yr ( ) Cancer duration until completion of the study, mo ( ) SRE until completion of the study Absence/presence 1.64 ( ) ECOG status at the final study visit ( ) Treatment after study Without/with bisphosphonate 0.25 ( ) PSA, mg/l ( ) talp, U/l ( ) balp, ng/l ( ) PINP, mg/l ( ) ( ) NTx, nmol/l ( ) CTx, ng/l ( ) ICTP, mg/l ( ) ( ) HR = hazard ratio; CI = confidence interval; SRE = skeletal-related events; ECOG = Eastern Cooperative Oncology Group; PSA = prostate-specific antigen; talp = total alkaline phosphatase; balp = bone-specific alkaline phosphatase; PINP = amino-terminal procollagen propeptides of type I collagen; NTx = crosslinked N-terminal telopeptides of type I collagen; CTx = cross-linked C-terminal telopeptides of type I collagen; ICTP = C-terminal telopeptides of type I collagen. * Calculations were performed using dichotomised data based on median values (less than vs greater than or equal to median) for the variables age, cancer duration, PSA, and bone markers at the last visit of the 15-mo zoledronic treatment study or according to the other indicated criteria. ** Reduced model following backward likelihood elimination (entry p = 0.025, removal p = 0.05). Table 5 Internal validation of univariate and multivariate Cox regression analyses by bootstrapping according to results in Table 3 * Variable Dichotomous criteria Multivariate combined model using clinical variables and bone markers ** Univariate Inclusion Backward Age, yr ( ) ( ) Cancer duration until ( ) ( ) completion of the study, mo SRE until completion of the study ECOG status at the final study visit Absence/presence 3.40 ( ) ( ) ( ) ( ) Treatment after study Without/with bisphosphonate 0.22 ( ) ( ) PSA, mg/l ( ) ( ) talp, U/l ( ) < balp, ng/l ( ) PINP, mg/l ( ) < ( ) ( ) NTx, nmol/l ( ) CTx, ng/l ( ) ICTP, mg/l ( ) < ( ) ( ) HR = hazard ratio; CI = confidence interval; SRE = skeletal-related events; ECOG = Eastern Cooperative Oncology Group; PSA = prostate-specific antigen; talp = total alkaline phosphatase; balp = bone-specific alkaline phosphatase; PINP = amino-terminal procollagen propeptides of type I collagen; NTx = crosslinked N-terminal telopeptides of type I collagen; CTx = cross-linked C-terminal telopeptides of type I collagen; ICTP = C-terminal telopeptides of type I collagen. * Calculations were performed using dichotomised data based on median values (less than vs greater than or equal to median) for the variables age, cancer duration, PSA, and bone markers at the last visit of the 15-mo zoledronic treatment study or according to the other indicated criteria. Bootstrapping was performed with 2000 resamplings as the internal validation approach; the bias corrected, accelerated 95% CIs are given. ** The combined model with clinical variables and bone markers only included the five itemised clinical variables, PSA, and the two bone markers ICTP and PINP, which showed s <0.10 in the bone marker full model and remained as the two single markers after backward elimination. Reduced model following backward likelihood elimination (entry p = 0.025, removal p = 0.05).

7 610 EUROPEAN UROLOGY 59 (2011) Table 6 Concordance indices of the multivariate Cox regression models with different variables Variable of the Cox regression model Concordance index Clinical data * 0.70 Clinical data plus PSA 0.68 Clinical data plus PSA plus ICTP 0.75 Clinical data plus PSA pluspinp 0.74 Clinical data plus PSA plus PINP plus ICTP 0.77 PINP plus ICTP 0.88 PSA = prostate-specific antigen; ICTP = C-terminal telopeptides of type I collagen; PINP = amino-terminal procollagen propeptides of type I collagen; SRE = skeletal-related events; ECOG = Eastern Cooperative Oncology Group. * Clinical data include the five variables age, cancer duration until completion of the study, SRE until completion of the study, ECOG status at the final study visit, and treatment after study as indicated in Table 3. usefulness of PINP and ICTP as diagnostic and prognostic markers in PCa patients [9,22 25]. The present study further substantiated that these two markers are the most suitable survival predictors. Probably as a result of the strong correlations among all bone markers (Table 2), the combination of ICTP and PINP as well as the use of all markers did not improve the prediction of mortality within the following 24 mo in comparison to the single markers (Table 7). Both markers showed better survival predictability than balp, which was recently favoured over the urinary NTx [2]. In contrast, another study reported that increased urinary NTx was more useful than balp in this respect in hormone-refractory PCa patients [10]. These discrepancies might be the result of the different study populations. Regarding the sample type for measurements, assays for Table 7 Receiver operating characteristic analyses of bone markers to predict death within 24 mo Bone marker AUC Sensitivity % Specificity talp, 88 U/l * 59 (36 72) 80 (59 93) balp, 17.5 ng/l * 68 (49 87) 64 (43 82) PINP, 30.9 mg/l ** 67 (45 84) 72 (51 88) NTx, 14.3 nmol/l * 54 (33 74) 80 (59 93) CTx, 540 ng/l ** 67 (45 84) 80 (59 93) ICTP, 4.10 mg/l *** 75 (53 89) 72 (51 88) PINP + ICTP *** 75 (53 89) 72 (51 88) All markers ** 72 (51 88) 72 (51 88) AUC = area under the receiver operation characteristics curve; CI = confidence interval; talp = total alkaline phosphatase; balp = bonespecific alkaline phosphatase; PINP = amino-terminal procollagen propeptides of type I collagen; NTx = cross-linked N-terminal telopeptides of type I collagen; CTx = cross-linked C-terminal telopeptides of type I collagen; ICTP = C-terminal telopeptides of type I collagen. AUCs were calculated using the concentrations of bone markers at the completion of the zoledronic treatment and given as means plus or minus standard error to predict death within 24 mo. Significances in comparison to the null hypothesis value of 0.5 are given as follows: *p < 0.05, **p < 0.01, ***p < Calculations of sensitivity/specificity using the concentrations of bone markers with the highest accuracy (minimal false-negative and false-positive results according to the Youden index) at the completion of the zoledronic treatment to predict death within 24 mo. The AUCs of the combination of ICTP plus PINP and of all markers were calculated using probabilities obtained with binary logistic regression analyses. serum/plasma should be used instead of urine because of the higher preanalytical and biologic variability of bone markers in urine and the inconvenience of urine sampling in elderly patients [26]. PINP in particular fulfils the precondition for routine measurements because an automated assay is available [6]. Although all bone markers were associated with survival, as shown in univariate Cox regression analyses, only ICTP and PINP remained significant variables in the Cox regression model after backward elimination (Table 3). PSA was not associated with survival, which was likewise demonstrated by others [14]; however, PSA monitoring remains essential in the follow-up to control progression. Moreover, the actual survival probabilities were less reliably estimated using the constructed model based only on clinical data rather than on ICTP and PINP. This is demonstrated by comparing the respective concordance indices of 0.66 and 0.88 (Table 6). Prediction models, which were built with clinical data and routine clinico-chemical parameters in other studies of metastatic PCa patients, resulted in comparable concordance indices of [9,14,15,17]. In contrast, the goodness-of-fit for the use of ICTP and PINP was shown by AUCs >0.7 and sensitivities/ specificities of 70 80% (Table 7). The predictive power corresponded to an AUC of 0.69 using the prognostic model of Halabi et al. [16]. Furthermore, it is of particular interest that the continuation of bisphosphonate administration after the completion of the treatment study tended to decrease the (Table 3; combined model). This result concurred with the tendency for higher survival in patients lacking SER (Table 3), which supports the indirect or direct effect of zoledronic acid on improving survival in these patients [27], but it should be clearly considered that the design of this study was not suitable to give a conclusive answer as to the usefulness of zoledronic acid in this context. Some limitations of this study merit discussion. Only a limited number of patients were evaluated. However, the number of patients actually investigated was consistent with the specified preconditions of type I and II error (a = 5%; b = 80%) in the sample size calculations. The results also showed that the risk of type I and II errors as a problem in small studies could be excluded as far as possible. Because an external validation could not be performed, internal validations using bootstrap resampling and concordance index calculations confirmed the accuracy of the established models with a negligible proportion of overfitting. Thus, despite the above-mentioned limitations of the study, valuable information was obtained that should point to ways to future studies. 5. Conclusions ICTP and PINP were shown to be suitable predictors of OS in PCa patients with osseous metastases treated with zoledronic acid in addition to chemotherapy, hormone therapy, and/or radiation therapy. The use of these markers improved stratification compared to using only clinical data. Their use implies an important advantage not only for

8 EUROPEAN UROLOGY 59 (2011) clinicians regarding treatment management but also for patients regarding their own decisions in care planning. Ongoing studies with the objective of improving the clinical outcome of high-risk PCa patients after RP by early administration of zoledronic acid should incorporate measurements of ICTP and PINP into their protocol. Author contributions: Klaus Jung had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Jung, Lein. Acquisition of data: Jung, Lein, Miller. Analysis and interpretation of data: Jung, Lein, Wirth. Drafting of the manuscript: Jung, Lein. Critical revision of the manuscript for important intellectual content: Miller, Wirth, Albrecht. Statistical analysis: Jung, Lein. Obtaining funding: Lein. Administrative, technical, or material support: Miller, Wirth, Albrecht. Supervision: Jung, Lein. Other (specify): None. Financial disclosures: I certify that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/ affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: Dr Albrecht is an employee of Novartis; Drs Miller, Lein, and Wirth are on the advisory board of Novartis; Dr Lein received funding from Novartis. Funding/Support and role of the sponsor: The Urologic Research Foundation, Berlin, and Novartis Pharma GmbH, Nuremberg, Germany provided partial support to Dr Lein. However, the financial support was provided with no further obligation, and the sponsor did not influence the design or conduct of the study; the collection, management, analysis, and interpretation of the data; and the preparation, review, or approval of the manuscript. Acknowledgment statement: The authors acknowledge all the urologists and technicians in the treating centres and Professor Dr Dietmar Schnorr for assistance with the collection of data. References [1] Brown JE, Cook RJ, Major P, et al. Bone turnover markers as predictors of skeletal complications in prostate cancer, lung cancer, and other solid tumors. J Natl Cancer Inst 2005;97: [2] Cook RJ, Coleman R, Brown J, et al. Markers of bone metabolism and survival in men with hormone-refractory metastatic prostate cancer. Clin Cancer Res 2006;12: [3] Smith MR, Cook RJ, Coleman R, et al. Predictors of skeletal complications in men with hormone-refractory metastatic prostate cancer. Urology 2007;70: [4] Seibel MJ. The use of molecular markers of bone turnover in the management of patients with metastatic bone disease. Clin Endocrinol (Oxf) 2008;68: [5] Lein M, Wirth M, Miller K, et al. Serial markers of bone turnover in men with metastatic prostate cancer treated with zoledronic acid for detection of bone metastases progression. Eur Urol 2007; 52: [6] Jung K, Lein M, Stephan C, et al. Comparison of ten serum bone turnover markers in prostate carcinoma patients with bone metastatic spread: diagnostic and prognostic implications. Int J Cancer 2004;111: [7] Coleman RE, Major P, Lipton A, et al. Predictive value of bone resorption and formation markers in cancer patients with bone metastases receiving the bisphosphonate zoledronic acid. J Clin Oncol 2005;23: [8] Lein M, Miller K, Wirth M, et al. Bone turnover markers as predictive tools for skeletal complications in men with metastatic prostate cancer treated with zoledronic acid. Prostate 2009;69: [9] Kamiya N, Suzuki H, Yano M, et al. Implications of serum bone turnover markers in prostate cancer patients with bone metastasis. Urology 2010;75: [10] Rajpar S, Massard C, Laplanche A, et al. Urinary N-telopeptide (untx) is an independent prognostic factor for overall survival in patients with bone metastases from castration-resistant prostate cancer. Ann Oncol 2010;21: [11] Schröder B. Evaluating the predictive performance of species distribution models. Universität Potsdam Web site. geoecology.uni-potsdam.de/users/schroeder/download.html. Accessed September 21, [12] Harrell Jr FE, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15: [13] Lughezzani G, Briganti A, Karakiewicz PI, et al. Predictive and prognostic models in radical prostatectomy candidates: a critical analysis of the literature. Eur Urol 2010;58: [14] Smaletz O, Scher HI, Small EJ, et al. Nomogram for overall survival of patients with progressive metastatic prostate cancer after castration. J Clin Oncol 2002;20: [15] Armstrong AJ, Garrett-Mayer ES, Yang YC, de Wit R, Tannock IF, Eisenberger M. A contemporary prognostic nomogram for men with hormone-refractory metastatic prostate cancer: a TAX327 study analysis. Clin Cancer Res 2007;13: [16] Halabi S, Vogelzang NJ, Kornblith AB, et al. Pain predicts overall survival in men with metastatic castration-refractory prostate cancer. J Clin Oncol 2008;26: [17] Armstrong AJ, Garrett-Mayer E, de Wit R, Tannock I, Eisenberger M. Prediction of survival following first-line chemotherapy in men with castration-resistant metastatic prostate cancer. Clin Cancer Res 2010;16: [18] Nayyar R, Sharma N, Gupta NP. Prognostic factors affecting progression and survival in metastatic prostate cancer. Urol Int 2010; 84: [19] Shamash J, Stebbing J, Sweeney C, et al. A validated prognostic index predicting response to dexamethasone and diethylstilbestrol in castrate-resistant prostate cancer. Cancer 2010;116: [20] Maeda H, Koizumi M, Yoshimura K, Yamauchi T, Kawai T, Ogata E. Correlation between bone metabolic markers and bone scan in prostatic cancer. J Urol 1997;157: [21] Percival RC, Urwin GH, Harris S, et al. Biochemical and histological evidence that carcinoma of the prostate is associated with increased bone resorption. Eur J Surg Oncol 1987;13:41 9. [22] Ramankulov A, Lein M, Kristiansen G, Loening SA, Jung K. Plasma osteopontin in comparison with bone markers as indicator of bone metastasis and survival outcome in patients with prostate cancer. Prostate 2007;67: [23] Koopmans N, de Jong IJ, Breeuwsma AJ, van der Veer E. Serum bone turnover markers (PINP and ICTP) for the early detection of bone metastases in patients with prostate cancer: a longitudinal approach. J Urol 2007;178: [24] Lara Jr PN, Stadler WM, Longmate J, et al. A randomized phase II trial of the matrix metalloproteinase inhibitor BMS in hormonerefractoryprostatecancerpatientswith bone metastases. Clin Cancer Res 2006;12:

9 612 EUROPEAN UROLOGY 59 (2011) [25] Zafeirakis A, Papatheodorou G, Arhontakis A, Gouliamos A, Vlahos L, Limouris GS. Predictive implications of bone turnover markers after palliative treatment with (186)Re-HEDP in hormone-refractory prostate cancer patients with painful osseous metastases. Eur J Nucl Med Mol Imaging 2010;37: [26] Woitge HW, Pecherstorfer M, Li Y, et al. Novel serum markers of bone resorption: clinical assessment and comparison with established urinary indices. J Bone Miner Res 1999;14: [27] Saad F, Schulman CC. Role of bisphosphonates in prostate cancer. Eur Urol 2004;45:26 34.