Edinburgh Research Explorer

Associations Between Serum Bone Biomarkers in Early BreastCancer and Development of Bone Metastasis Results From theAZURE (BIG0104) TrialCitation for published versionBrown J Rathbone E Hinsley S Gregory W Gossiel F Marshall H Burkinshaw R Shulver HThandar H Bertelli G Maccon K Bowman A Hanby A Bell R Cameron D amp Coleman RE 2018Associations Between Serum Bone Biomarkers in Early Breast Cancer and Development of BoneMetastasis Results From the AZURE (BIG0104) Trial JNCI Journal of the National Cancer Institute vol110 no 8 pp 871ndash879 httpsdoiorg101093jncidjx280

Digital Object Identifier (DOI)101093jncidjx280

LinkLink to publication record in Edinburgh Research Explorer

Document VersionPublishers PDF also known as Version of record

Published In JNCI Journal of the National Cancer Institute

Publisher Rights Statementcopy The Author(s) 2018 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution License(httpcreativecommonsorglicensesby40)which permits unrestricted reuse distribution and reproduction in any medium provided the original work isproperly cited

General rightsCopyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s)and or other copyright owners and it is a condition of accessing these publications that users recognise andabide by the legal requirements associated with these rights

Take down policyThe University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorercontent complies with UK legislation If you believe that the public display of this file breaches copyright pleasecontact openaccessedacuk providing details and we will remove access to the work immediately andinvestigate your claim

Download date 10 Jan 2021

ARTICLE

Associations Between Serum Bone Biomarkers in Early Breast

Cancer and Development of Bone Metastasis Results From

the AZURE (BIG0104) Trial

Janet Brown Emma Rathbone Samantha Hinsley Walter Gregory Fatma Gossiel HelenMarshall Roger Burkinshaw Helen Shulver Hasina Thandar Gianfilippo Bertelli KeaneMaccon Angela Bowman Andrew Hanby Richard Bell David Cameron Robert Coleman

Affiliations of authors Academic Unit of Clinical Oncology and Sheffield ECMC University of Sheffield Weston Park Hospital Sheffield UK (JB ER RB HS RC) LeedsInstitute of Cancer and Pathology University of Leeds Leeds UK (JB ER AH) Calderdale and Huddersfield NHS Foundation Trust Huddersfield UK (ER) Clinical TrialsResearch Unit Leeds Institute of Clinical Trials Research University of Leeds Leeds UK (SH WG HM) Academic Unit of Bone Metabolism Metabolic Bone CentreUniversity of Sheffield Northern General Hospital Sheffield UK (FG) Royal Surrey County Hospital Guildford UK (HT) Singleton Hospital Swansea UK (GB) CancerTrials Ireland University College Hospital Galway Ireland (KM) University of Edinburgh Cancer Research Centre Western General Hospital Edinburgh UK (AB DC)Deakin University Geelong Australia (RB)

Authors contributed equally to this work

Correspondence to Janet E Brown MD MSc MBBS Academic Unit of Clinical Oncology University of Sheffield Weston Park Hospital Sheffield S10 2SJ UK(e-mail jebrownsheffieldacuk)

Abstract

Background Adjuvant therapies can preventdelay bone metastasis development in breast cancer We investigated whetherserum bone turnover markers in early disease have clinical utility in identifying patients with a high risk of developing bonemetastasisMethods Markers of bone formation (N-terminal propeptide of type-1 collagen [P1NP]) and bone resorption (C-telopeptide oftype-1 collagen [CTX] pyridinoline cross-linked carboxy-terminal telopeptide of type-1 collagen [1-CTP]) were measured inbaseline (pretreatment blood samples from 872 patients from a large randomized trial of adjuvant zoledronic acid (AZURE-ISRCTN79831382) in early breast cancer Cox proportional hazards regression and cumulative incidence functions (adjustedfor factors having a statistically significant effect on outcome) were used to investigate prognostic and predictive associationsbetween recurrence events bone marker levels and clinical variables All statistical tests were two-sidedResults When considered as continuous variables (log transformed) P1NP CTX and 1-CTP were each prognostic for futurebone recurrence at any time (P frac14 006 P frac14 009 P frac14 008 respectively) Harrellrsquos c-indices were a P1NP of 057 (95 confidenceinterval [CI] frac14 051 to 063) CTX of 057 (95 CIfrac14051 to 062) and 1-CTP of 057 (95 CIfrac14052 to 063) In categorical analysesbased on the normal range high baseline P1NP (gt70 ngmL) and CTX (gt0299 ngmL) but not 1-CTP (gt42 ngmL) were alsoprognostic for future bone recurrence (P frac14 03 P frac14 03 P frac14 10 respectively) None of the markers were prognostic for overalldistant recurrence that is they were bone metastasis specific and none of the markers were predictive of treatment benefitfrom zoledronic acidConclusions Serum P1NP CTX and 1-CTP are clinically useful easily measured markers that show good prognostic ability(though low-to-moderate discrimination) for bone-specific recurrence and are worthy of further study

More than 40 000 women die from breast cancer annually in theUnited States mainly from distant relapse which often occursyears after initial breast cancer diagnosis (1) Bone metastases

ultimately affect more than two-thirds of patients with ad-vanced disease (2) Breast cancer cells can remain dormant formany years in the bone microenvironment escaping the effects

AR

TIC

LE

Received April 11 2017 Revised October 26 2017 Accepted December 6 2017

copy The Author(s) 2018 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby40)which permits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited

1

JNCI J Natl Cancer Inst (2018) 110(8) djx280

doi 101093jncidjx280First published online December 19 2017Article

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

of adjuvant systemic therapies and retaining the potential forfuture activation and proliferation resulting in metastasis inbone andor other distant sites

Because breast cancer cells display this affinity for bonethere is a sound rationale for targeting the bone in the adjuvantsetting Randomized trials of adjuvant bisphosphonates con-firmed by a meta-analysis of all available data (n frac14 18 766) haveindeed shown that development of bone metastases and deathfrom breast cancer can be reduced However the benefits areconfined to postmenopausal patients at the time of bisphosph-onate initiation (3ndash7) strongly suggesting that the postmeno-pausal bone (marrow) microenvironment has a specificinteraction with tumor cell homing to bone andor tumordormancy

In the AZURE trial (ISRCTN79831382) in early breast cancer3360 women with stage IIIII breast cancer were randomized tostandard adjuvant treatment alone or with the addition of zole-dronic acid (zoledronate administered over five years (36)With a median of 842 months (interquartile range [IQR] frac14 66ndash93months) of follow-up zoledronate improved invasive diseasendashfree survival (IDFS) in women who were more than five yearspostmenopausal at diagnosis (n frac14 1041 adjusted hazard ratio[HR] frac14 077 95 confidence interval [CI] frac14 063 to 096) Baseline(pretreatment) serum samples were collected in a subset ofpatients and these provide the opportunity for prespecifiedanalyses of relationships between bone metabolism as deter-mined by serum bone turnover markers and disease outcomeswith or without zoledronate The role of bone turnover markershas been extensively studied in established bone metastasis(89) In the current study our aims were to determine whetherin early breast cancer levels of bone turnover markers predictedeither the risk of disease relapse (both in and outside bone) orthe treatment benefits from zoledronate

Methods

Patients

In the AZURE trial (36) following written informed consentwomen with histologically confirmed breast cancer and eitherlymph node metastasis or T3T4 primary tumor were randomlyassigned to either standard adjuvant therapy (control) or stan-dard adjuvant therapy plus intravenous zoledronate 4 mg (19doses over five years)

At UK centers ethics approval was obtained for this studyand participants gave additional consent for blood donation atstudy entry to be used for biomarker assessment Serum sam-ples were collected and stored under strict standard operatingprocedures temporarily at ndash20 C or ndash80 C at local centers beforeregular transfer to Sheffield for storage at ndash80 C until centralbatch analysis

Laboratory Assays

Bone biomarkers were measured against reference standards ina fully accredited central laboratory (Metabolic Bone UnitUniversity of Sheffield) according to strict standard operatingprocedures Personnel performing and reporting the analyseswere blinded to clinical data

Bone Biomarker AnalysisWe measured two biomarkers of bone resorption C-telopeptideof type-1 collagen (CTX) a measure of cathepsin-K-linked

collagen breakdown and pyridinoline cross-linked carboxy-ter-minal telopeptide of type-1 collagen (1-CTP) which is liberatedby matrix metalloproteinases during degradation of maturetype-1 collagen Because 1-CTP is not produced through cathep-sin-K-mediated bone resorption its concentration is less af-fected by menopause (10) N-terminal propeptide of type-1collagen (P1NP) released during collagen formation is a robustand reliable measure of bone formation and was selected forthis study (11) P1NP and CTX were measured using Cobas e411automated immunoassays (Roche Diagnostic MannheimGermany) and 1-CTP was measured by manual enzyme immu-noassay (Orion Diagnostica UniQ ICTP EIA Espoo Finland)

The P1NP assay has a lower detection limit of 5 ngmL andinterassay coefficient of variation (CV) of 41 Values werecategorized as ldquohighrdquo if greater than 70 ngmL based on advicefrom Roche Diagnostics Results of 70 ngmL or less werecategorized as normal The CTX assay has a measurementrange of 0010 to 600 ngmL and an interassay CV of 40 Theupper limit of normal for premenopausal women (0299 ngmL)was used to categorize results as either ldquohighrdquo (gt0299 ngmL) orldquonormalrdquo (0299 ngmL) We also performed additional analy-ses with a higher threshold (high gt0556 ngmL) to allow closercomparison with an earlier study by Lipton and colleagues (12)The 1-CTP assay was conducted manually with a lower detec-tion limit of 03 ngmL and an upper limit of normal of 42 ngmL The intraassay CV was 81 at 56 ngmL and the interas-say CV was 76 at 48 ngmL

Statistical Analysis

Statistical analysis (consistent with REMARK guidelines) wasperformed on the final AZURE analysis datalock with a medianof 842 months (IQR frac14 66ndash93 months) of follow-up and 966disease-free survival events (6) All analyses were performed onthe intention-to-treat population using SAS version 92 or 94Hypothesis testing was performed at the two-sided 5 level

Cumulative incidence function (CIF) curves were used to in-vestigate time to recurrence as defined below The Cox propor-tional hazards (PH) model was used to assess the relationshipsbetween the bone biomarkers and prognosis and treatment ef-fect with zoledronate The proportional hazards assumptionwas verified by assessing the statistical significance of the inter-action of the relevant bone biomarker and time via an interac-tion term within the Cox model as well as by a manual reviewof the CIF curves Bone marker data were analyzed both as con-tinuous variables (log transformed) and as categorical variablesusing the prespecified high vs normal cut-points for both prog-nostic and predictive relationships

The prespecified end points in the statistical analysis planwere 1) time to bone recurrence whether or not bone was thefirst recurrence (with deaths without prior bone recurrence cen-sored in the Cox PH models and considered competing-riskevents in CIF curves) 2) time to first recurrence in bone includ-ing first recurrence being in bone only or concurrently with recur-rence in another distant site (with deaths without priorrecurrence and nonbone (only) first recurrences censored in theCox PH models and considered competing-risk events in CIFcurves) 3) time to first distant recurrence (with deaths withoutprior distant recurrence censored in the Cox PH models and con-sidered competing-risk events in CIF curves)

Analyses were performed for all participants combined andaccording to menopausal status and were adjusted for mini-mization factors found to be statistically significant for disease

AR

TIC

LE

2 | JNCI J Natl Cancer Inst 2018 Vol 110 No 8

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

outcomes in the main AZURE analyses (ie lymph node involve-ment estrogen receptor [ER] status tumor stage and typetimingof systemic therapy for each end point) as well as treatmentallocation where this was statistically significant in the mainAZURE subgroup analyses Analyses were also adjusted for treat-ment allocation when assessing the interaction of biomarkerswith treatment (predictive analyses) where the interaction termis used to test for heterogeneity between the different biomarkerlevels Exploratory analyses were carried out with a compositeP1NPCTX biomarker in terms of both markers high vs not bothmarkers high

Harrellrsquos c-index was used to assess the discriminatory abil-ity of the markers with a value of 1 representing perfect dis-crimination and 05 being no better than chance Confidenceintervals for Harrellrsquos c-index were calculated as suggested byNewson (13)

Results

Patient Demographics and Baseline Data

Serum samples from 872 UK AZURE participants (441 controlarm 431 treatment arm) were analyzed with a median follow-up of 842 months (IQR frac14 711ndash921 months) Baseline patientdemographics (age lymph node involvement ER progesteronereceptor [PR] and human epidermal growth factor receptor 2(HER2) status menopausal status systemic therapy chemo-therapy and statin use) in this test subpopulation were similarto the overall AZURE patient population (Table 1)

Comparison of IDFS outcomes for the biomarker populationwith the whole AZURE population showed that both the propor-tions of patients with an event and the hazard ratios were simi-lar though the confidence intervals were wider in thebiomarker population due to the smaller number of patientsThis similarity also applied when broken down into post- andnonpostmenopausal subgroups (Supplementary Figure 1 avail-able online)

Baseline data for the three biomarkers (also broken downinto menopausal status) revealed that the proportion ofpatients in each category who fall above the normal ranges forP1NP CTX and 1-CTP for the whole population were 273300 and 505 respectively (Table 2) confirming that thedata were appropriate to test the relationship between acceler-ated baseline bone turnover and subsequent distant recurrenceevents

Bone Biomarker Prognostic Analyses

Figure 1 and Table 3 display key data for prognostic analyses inthe three prespecified recurrence categories The proportionalhazards assumption was also investigated for each Cox propor-tional hazards model applied in our study The majority ofmarkers and end points were not close to violating this assump-tion (ie suggesting no difference in the effects of the markers astime elapses) For the 1-CTP marker the assumption was onlyborderline valid suggesting that the impact of 1-CTP may differas time elapses although the Cox proportional hazards modelwas still appropriate

Bone Recurrence at Any TimeIn adjusted continuous log-transformed analyses (Figure 1)increases in all three markers were strongly associated with sta-tistically significantly increased risk of development of bone

metastasis (P1NP P frac14 006 CTX P frac14 009 1-CTP P frac14 008) In cat-egorical analyses P1NP greater than 70 ngmL (P frac14 03) and CTXgreater than 0299 ngmL (P frac14 03) but not CTX greater than0566 ngmL (P frac14 12) or 1-CTP greater than 42 ngmL (P frac14 10)were statistically significantly prognostic for recurrence in boneat any time (Table 3) Cumulative incidence plots for categoricalanalysis of time to bone metastasis at any time are shown ex-emplified for P1NP in Figure 2 for both control and treatmentarms

Taking P1NP on the basis of the above data as the likelymost sensitive prognostic factor we tested the role of meno-pausal status in P1NP analyses (data not shown) However wedetected no statistically significant prognostic effect of P1NPon bone recurrence in either postmenopausal or non-postmenopausal patients when analyzed with P1NP as either acategorical or continuous variable

Harrellrsquos c-index values (when coded as [log] continuous var-iables) were similar for all three markers P1NP c-index was 057(95 CI frac14 051 to 063) CTX c-index was 057 (95 CI frac14 051 to062) and 1-CTP c-index was 057 (95 CI frac14 052 to 063)

First Recurrence in Bone (12 Concurrent Recurrence Elsewhere)In the adjusted continuous analyses both P1NP (P frac14 03) and1-CTP (P frac14 045) appeared statistically significantly prognosticfor first recurrence in bone (Figure 1 Table 3) However in ad-justed categorical analyses although the hazard ratios for eachmarker were similar to bone recurrence at any time the 95confidence intervals were wide and no statistically significantrelationships between higher marker values and first diseaserecurrence in bone were seen The number of bone-only first re-currence events was too small to justify separate analysis ofthis potential end point of interest

First Distant Recurrence (Whatever the Site)There were no associations in either continuous or categoricalanalyses between baseline P1NP CTX or 1-CTP and develop-ment of distant recurrence at any site (Figure 1 Table 3) clearlydemonstrating that in contrast to recurrence specifically inbone the markers were not prognostic for distant metastasistaken as a whole Categorical data for IDFS were statisticallynonsignificant

Composite P1NP and CTX Biomarker AnalysisAdjusted analyses were performed to assess risks of recurrencefor patients where both P1NP and CTX (using the 0299 ngmLcut-point) were high compared with all other patients detailsare displayed in Table 4 No statistically significant relation-ships were identified between the composite marker and subse-quent recurrence although there was an increased risk for bonerecurrence at any time in the patients with elevation of bothbiomarkers (HR frac14 160 95 CI frac14 099 to 248 P frac14 06)Consideration was given to a joint Cox model containing allthree markers but there were insufficient events to make thismeaningful Further 1-CTP levels represent a different aspect ofthe bone turnover process than P1NP and CTX and are largelyunaffected by inhibitors of osteoclast function such asbisphosphonates (10)

Sensitivity Analyses Assessing Optimum Cut-PointsWe explored the effects of different cut-points for categoricalprognostic analysis of P1NP and bone metastasis at any timeThis analysis (Figure 3) showed that the optimal cut-point for

AR

TIC

LE

J Brown et al | 3 of 9

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

P1NP was approximately 64 nmolmL which we judged wassufficiently close to the prespecified value of 70 nmolmL bear-ing in mind that the number of events was not sufficient to gen-erate a smooth relationship For 1-CTP and CTX similarexploration yielded no clearly optimal cut-point or improve-ment to those preselected (Supplementary Figures 2 and 3available online)

Analyses for Treatment EffectmdashTest for PredictiveBiomarkers

Although P1NP is higher in postmenopausal women and thebenefits of zoledronate are largely restricted to this subset of

patients (67) baseline P1NP did not predict benefit from zoledr-onate when assessed against bone metastasis at any time out-come For example in categorical analyses considering theeffect of P1NP on bone recurrence at any time there were nostatistically significant differences in outcome between thezoledronate and control arms for either high P1NP (HR frac14 09995 CI frac14 052 to 190) or normal P1NP (HR frac14 084 95 CI frac14 052to 137) with a nonsignificant Pinteraction value for the interactionof P1NP and treatment (P frac14 69) (Figure 1B)

We also found no statistically significant interaction withtreatment allocation for either of the other defined lessfrequent outcome categories with any of the bone markers orwith the P1NPCTX composite bone marker However consider-ing the complex inter-relationships between treatment effectand menopausal status in the main AZURE study the numbersof events are likely insufficient for definitive analysis

Corresponding continuous (log-transformed) analyses forbone metastases at any time found no statistically significantinteraction with treatment allocation for any of the markersanalyzed (P1NP P frac14 74 CTX P frac14 47 1-CTP P frac14 31) confirmingthat these baseline markers are not predictive of the treatmentbenefits of zoledronate

Discussion

Our study showed that patients with high serum levels of P1NPCTX or 1-CTP shortly after diagnosis of early breast cancer wereassociated with a higher risk of developing bone metastasis dur-ing the course of their disease P1NP appeared to be the mostsensitive of the markers studied but was not predictive of bene-fit from zoledronate Using CTX and P1NP as a composite bio-marker did not add to the sensitivity of the individual markersThis may be partially because the markers are not independentreporting on essentially linked metabolic processes but mayalso be due to the relatively small number of events in the com-bined group

It has been long-established that the rate of bone loss accel-erates relatively rapidly in perimenopause across the meno-pausal transition with consequent increase in bone turnovermarkers associated with an accelerated decrease in measuredbone mineral density (BMD) (1415) The inverse relationship be-tween loss of BMD and increase in bone turnover markers (in-cluding P1NP and CTX) from premenopause throughperimenopause to postmenopause is well established (16) Asanticipated our data reflect this pattern with baseline bonemarker values all increasing progressively from premenopausethrough the early years following cessation of menses to morethan five years postmenopause

Although all three markers and especially P1NP are good pre-dictors of bone-specific recurrence the calculated values ofHarrellrsquos c-index (each around 057) suggest that they have onlylow-to-moderate discrimination although this is statistically dif-ferent from a chance finding as shown by the lower limit in the95 confidence intervals being greater than 05 From a clinical per-spective however it should be borne in mind that even for the twokey prognostic indicators in everyday use in breast cancer (lymphnodes and stage) c-indices are 062 and 063 respectively (WGregory personal communication) only marginally greater thanthose values reported for the three bone markers in this study

Our findings are consistent with a bone microenvironmentwith increased bone turnover providing a fertile ldquosoilrdquo for thedevelopment of skeletal metastasis By contrast with this clearassociation between baseline bone turnover markers and

Table 1 Baseline demographics of patients in the biomarker subpop-ulation and overall AZURE population

Parameter

Biomarkerpopulation

Overall studypopulation

(n frac14 872) (n frac14 3359)No () No ()

Mean age y 514 515Lymph node status0 16 (18) 62 (18)1ndash3 534 (612) 2075 (618)4 320 (367) 1211 (361)Unknown 2 (02) 11 (03)T stageT1 285 (327) 1065 (317)T2 427 (490) 1717 (511)T3 131 (150) 456 (136)T4 29 (33) 117 (35)Histological grade1 66(76) 285 (85)2 361 (414) 1439 (428)3 428 (491) 1552 (462)ER statusPositive 676 (775) 2634 (784)Negative 192 (220) 705 (210)Unknown 4 (05) 20 (06)PR statusPositive 361 (414) 1423 (424)Negative 205 (235) 806 (240)Unknown 304 (349) 1119 (333)HER2 statusPositive 108 (124) 415 (124)Negative 318 (365) 1251 (372)Unknownnot measured 442 (506) 1672 (498)Neo-adjuvant therapy intended 52 (60) 212 (63)Systemic therapyEndocrine therapy alone 33 (38) 152 (45)Chemotherapy alone 190 (218) 719 (214)Endocrine therapy and

chemotherapy649 (744) 2488 (741)

Use of statins 43 (49) 197 (59)Type of chemotherapyAnthracyclines 819 (939) 3132 (932)Taxanes 178 (204) 775 (231)Menopausal statusPremenopausal 409 (469) 1504 (448)5 y since menopause 123 (141) 490 (146)gt5 y since menopause 266 (305) 1041 (310)Unknown 74 (85) 324 (96)

ER frac14 estrogen receptor HER2 frac14 human epidermal growth factor receptor

PR frac14 progesterone receptorA

RT

ICLE

4 | JNCI J Natl Cancer Inst 2018 Vol 110 No 8

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

recurrence in bone there was no association detectable betweenbone turnover markers and distant recurrence taken as a wholeindicating that bone turnover markers specifically provide prog-nostic information for future recurrence in bone and not for me-tastasis more generally (17ndash19) We acknowledge that in somecases elevation of baseline bone markers may be linked with ac-tive but as yet undetected bone metastases However the rela-tively long follow-up (median frac14 84 months) and few bone eventsin the first two years (lt5) when the cumulative incidencecurves diverge makes it unlikely that the raised markers are sim-ply an early diagnostic indication of bone metastases

There is important literature evidence supporting our studyIn particular Lipton et al (12) investigated b-CTX in 621 postme-nopausal early breast cancer patients in a five-year phase IIItrial of tamoxifen thorn octreotide Over 79 years (median) of fol-low-up 19 (31) patients developed bone-only recurrence asfirst event 47 (75) developed bone and concurrent other re-lapse as first event and 57 (92) developed first recurrence insites excluding bone Using a categorical analysis (cut-point frac14071 ngmL) higher pretreatment b-CTX was associated withshorter bone-only recurrence-free survival (HR frac14 28 95 CI frac14105 to 748 P frac14 03) However there was no statistically

Table 2 Distribution of patients with highnormal bone marker values according to menopausal status

Biomarker Whole population Premenopausal 0ndash5 y postmenopausal gt5 y postmenopausal

P1NPAssay median (IQR) ngmL 551 (412ndash727) 491 (373ndash643) 584 (428ndash761) 648 (481ndash844)No of patients 867 409 121 263 gt 70 ngmL 273 183 301 387CTXAssay median (IQR) ngmL 023 (015ndash032) 018 (013ndash026) 025 (018ndash037) 029 (021ndash041)No of patients 863 408 120 262 gt 0299 ngmL 300 174 374 459 gt 0556 ngmL 42 10 73 791CTPAssay median (IQR) ngmL 425 (326ndash515) 399 (312ndash495) 426 (320ndash502) 460 (377ndash541)No of patients 861 408 118 265 gt 42 ngmL 505 445 496 617

This table does not include data for the patients whose menopausal status was unknown included in the whole study population 1-CTP frac14 pyridinoline cross-linked

carboxy-terminal telopeptide of type-1 collagen CTX frac14 C-telopeptide of type-1 collagen IQR frac14 interquartile range P1NP frac14 N-terminal propeptide of type-1 collagen

Figure 1 Hazard ratios and 95 confidence intervals for adjusted continuous analyses of log-transformed data for baseline N-terminal propeptide of type-1 collagen

C-telopeptide of type-1 collagen and pyridinoline cross-linked carboxy-terminal telopeptide of type-1 collagen and disease outcomes P values were calculated using

the likelihood ratio v2 test statistic and tests were two-sided 1-CTP frac14 pyridinoline cross-linked carboxy-terminal telopeptide of type-1 collagen CI frac14 confidence inter-

val CTX frac14 C-telopeptide of type-1 collagen HR frac14 hazard ratio P1NP frac14 N-terminal propeptide of type-1 collagen

AR

TIC

LE

J Brown et al | 5 of 9

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

significant association with first event in the bone plus concur-rent relapse elsewhere or with first recurrence at other distantsites It should be noted that there were differences in the pa-tient populations and administration of bone-targeted therapybetween the Lipton et al study and our data (the former in-cluded only postmenopausal patients whose tumors weremostly ER positive with consequent lower-risk disease)

A limitation of our study is that only baseline biomarkermeasurements were available although because the propor-tional hazards assumption was not violated this suggests nodifference in the effect of the markers as time elapses Whileour data suggest that the rate of bone turnover at this earlystage of disease when tumor cells may be homing to potentialmetastatic sites is a statistically significant contributing factorto development of bone metastasis changes in subsequentbone turnover may also play a role There is evidence that thismight be the case in a study that assessed paired serum sam-ples at baseline and one year within a large placebo-controlledrandomized study of oral clodronate in early breast cancer (20)Although baseline P1NP was not prognostic for developing bonemetastasis after five years of follow up the incidence of bonemetastasis was statistically significantly higher in womenwhose P1NP value increased by more than 20 in the first year(P lt 02)

A further possible limitation of our analysis is that the bio-marker population comprised slightly more than 25 of the totalAZURE population Although we have shown that both baselinedemographics and outcomes of the biomarker population andthe total trial population are similar this cannot completely ex-clude the possibility of bias in the population analyzed

Because the treatment benefits of adjuvant zoledronate inpostmenopausal women might be related to inhibition of theincreased bone turnover associated with menopause our find-ing that baseline P1NP levels were not predictive for benefitfrom zoledronate was initially surprising However a number offactors (in addition to low event numbers in some analyses)may contribute to this result Administration of multiple dosesof a potent bisphosphonate can confidently be assumed to sup-press bone turnover throughout the five-year treatment periodThis could render baseline marker values less relevant in analy-ses of association Also it should be noted that althoughzoledronate only produced a benefit in overall invasive relapsein patients who were five or more years postmenopause it wasassociated with a reduction in first and subsequent metastasisto bone across all menopausal groups (6) Additionally boneturnover markers reflect activity across the skeleton as a wholewhereas the amount of bone associated with disseminated tu-mor cells likely comprises only a very small fraction of the totalskeletal metabolic activity Finally there is the intriguing possi-bility that the efficacy of zoledronate in the adjuvant settingmay be due to a direct toxic effect on tumor cells in the bone mi-croenvironment and independent of its action on boneturnover

Other recent studies have also addressed the need for prognos-ticpredictive biomarkers relating to adjuvant bone-targetedtreatment in early breast cancer Using primary tumor tissue frompatients in the AZURE study we showed that a novel compositebiomarker comprising the proteins CAPG and GIPC1 was prognosticfor developing bone metastasis (HR frac14 45 95 CI frac14 21 to 98 P lt

001) and predicted response to zoledronate (P frac14 008) (21)

Table 3 Adjusted analyses for high vs normal values of bone markers and analyses for interquartile range change for continuous analyses

Analysis

Bone marker highNo of eventspatients ()

Bone marker normalNo of eventspatients ()

Adjusted categoricalanalysis

Adjusted continuousanalysis

HR (95 CI) P

HR for IQR change(log transformed)

(95 CI)dagger P

PINP (70 ngmL cut-point nfrac14 867)Bone recurrence at any time 37238 (155) 67629 (107) 161 (107 to 242) 03 142 (110 to 182) 006First recurrence in bone 29238 (122) 53629 (84) 158 (100 to 250) 06 138 (104 to 184) 03First distant recurrence (at any site) 54238 (227) 134629 (213) 099 (072 to 137) 96 104 (087 to 124) 64IDFS 69238 (290) 179629 (285) 097 (073 to 129) 83 106 (091 to125) 43CTX (0299 ngmL cut-point nfrac14 863)Bone recurrence at any time 40262 (153) 64601 (106) 155 (105 to 231) 03 141 (109 to 182) 009First recurrence in bone 28262 (107) 54601 (90) 127 (080 to 200) 32 129 (097 to 172) 08First distant recurrence (at any site) 67262 (256) 121601 (201) 126 (093 to 171) 13 111 (092 to 133) 27IDFS 87262 (332) 161601 (268) 124 (096 to 162) 11 115 (099 to 135) 08CTX (0556 ngmL cut-point nfrac14 863)Bone recurrence at any time 737 (189) 97826 (117) 195 (090 to 421) 12 141 (109 to 182) 009First recurrence in bone 637 (162) 76826 (92) 217 (094 to 501) 10 129 (097 to 172) 08First distant recurrence (at any site) 937 (243) 179826 (217) 094 (047 to 186) 85 111 (092 to 133) 27IDFS 1437 (378) 234826 (283) 115 (066 to 200) 62 115 (099 to 135) 081-CTP (42 ngmL cut-point nfrac14 861)Bone recurrence at any time 59440 (134) 45421 (107) 139 (094 to 205) 10 143 (110 to 186) 008First recurrence in bone 45440 (102) 37421 (88) 130 (084 to 202) 24 136 (101 to 184) 045First distant recurrence (at any site) 100440 (227) 87421 (207) 119 (089 to 159) 25 115 (094 to 139) 18IDFS 131440 (298) 116421 (276) 118 (091 to 153) 21 120 (100 to 142) 04

P values were calculated using the likelihood ratio v2 test statistic and tests were performed at the two-sided 5 significance level 1-CTP frac14 pyridinoline cross-linked

carboxy-terminal telopeptide of type-1 collagen CI frac14 confidence interval CTX frac14 C-telopeptide of type-1 collagen HR frac14 hazard ratio IDFS frac14 invasive diseasendashfree sur-

vival IQR frac14 interquartile range P1NP frac14 N-terminal propeptide of type-1 collagen

daggerThese results show the hazard ratio for an interquartile range increase in the log10 (P1NP) or ln (CTX 1-CTP) transformed variables The P value of these analyses is

unchanged from the adjusted continuous analyses shown in Figure 1

AR

TIC

LE

6 | JNCI J Natl Cancer Inst 2018 Vol 110 No 8

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

No at risk

Normal

High

629 605 565 535 507 479 442 318 77 0

238 229 212 199 188 179 160 115 22 0

Cum

ulat

ive

inci

denc

e

000005010015020025030035040045050055060065070075080085090095100

Time from random assignment mo

0 12 24 36 48 60 72 84 96 108

A

No at risk

No ZOL

ZOL

125 119 111 105 100 95 84 61 9 0

113 110 101 94 88 84 76 54 13 0

Cum

ulat

ive

inci

denc

e

000005010015020025030035040045050055060065070075080085090095100

Time from random assignment mo

0 12 24 36 48 60 72 84 96 108

B

Figure 2 A) Cumulative incidence function for time to bone metastasis at any time for categorical analysis of N-terminal propeptide of type-1 collagen (P1NP) level orlt70 ngmL (hazard ratio [HR] for adjusted analyses frac14 161 95 confidence interval [CI] frac14 107 to 242 P frac14 03) B) Cumulative incidence function for time to bone me-

tastasis at any time by treatment arm for participants with high P1NP (70 ngmL HR for adjusted analyses frac14 0989 95 CIfrac140517 to 1895 Pinteraction frac14 69 for the inter-

action between P1NP and treatment ie to assess for differing effects of treatment within the two groups of high or normal P1NP) P values were calculated using the

likelihood ratio v2 test statistic and tests were two-sided 1-CTP frac14 pyridinoline cross-linked carboxy-terminal telopeptide of type-1 collagen CTX frac14 C-telopeptide of

type-1 collagen P1NP frac14 N-terminal propeptide of type-1 collagen ZOL frac14 zoledronate 4 mg (19 doses over 5 years)

AR

TIC

LE

J Brown et al | 7 of 9

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

In another study amplification of the 16q23 chromosomal regionincluding amplification of the MAF gene (22) was predictive ofbreast cancer metastasis to bone (23) However there remains aneed for a simple blood-based test in early breast cancer that canidentify patients with a high risk for development of bone metasta-sis Bone turnover markers are easily measured and are worthy ofadditional investigation in helping to meet this need

Funding

This work was supported by Cancer Research UK (through theawards of a Research Studentship to ER a Clinician ScientistFellowship to JEB and support to the Sheffield Experimental CancerMedicine Centre) and a grant from Novartis Pharmaceuticals

Notes

The funders had no role in the design of the study the collec-tion analysis or interpretation of the data the writing of the

manuscript or the decision to submit the manuscript for publi-cation Novartis provided academic grant support and suppliesof zoledronic acid (Zometa) for the AZURE trial The authors dis-close the following JB received fees from Novartis and Amgenfor Advisory Boards and Speakers Bureaux WG received feesfrom Celgene for statistical consultancy and honoraria fromJanssen DC received nonfinancial support from Novartis RC re-ceived institutional research grants from Amgen and Bayerfees from Novartis for expert testimony and lecture fees fromAmgen All other authors declared no conflicts

We wish to thank the AZURE trial patients who providedblood samples to support this research

References1 US Breast Cancer Statistics 2017 httpwwwbreastcancerorgsymptoms

understand_bcstatistics2 Coleman RE Metastatic bone disease Clinical features pathophysiology and

treatment strategies Cancer Treat Rev 200127165ndash763 Coleman RE Marshall H Cameron D et al Breast-cancer adjuvant therapy

with zoledronic acid N Engl J Med 20113651396ndash1405

Table 4 Adjusted prognostic categorical analyses according to a composite P1NP-CTX marker for both P1NP and CTX high vs not both high

Analysis

P1NP CTX Adjusted analysis

Both higheventscensored ()

Not both higheventscensored () HR (95 CI) P

Bone recurrence at any time 24118 (169) 80641 (111) 160 (099 to 248) 06First recurrence being in bone 18124 (127) 64657 (89) 150 (089 to 254) 14First distant recurrence (at any site) 34108 (239) 154567 (214) 100 (068 to 145) 99First recurrence being in bone only 11131 (77) 47674 (65) 126 (065 to 244) 50

P values were calculated using the likelihood ratio v2 test statistic and tests were performed at the two-sided 5 significance level 1-CTP frac14 pyridinoline cross-linked

carboxy-terminal telopeptide of type-1 collagen CI frac14 confidence interval CTX frac14 C-telopeptide of type-1 collagen HR frac14 hazard ratio IDFS frac14 invasive diseasendashfree sur-

vival P1NP frac14 N-terminal propeptide of type-1 collagen

50 100 150 200

2

4

6

8

10

Cut-point for P1NP

Prognostic analysis - for bone recurrence at any timeadjusted Cox model chi-squares for different P1NP cut-points(optimum cut-point at 64 P = 003)

P = 01

P = 05

P1NP = 70

Figure 3 v2 values from adjusted Cox proportional hazards model analyzing bone metastasis at any time by N-terminal propeptide of type-1 collagen (P1NP) with dif-

fering high vs normal P1NP cut-points Optimum cut-point observed at 64 ngmL with a corresponding P value of 003 P values were calculated using the likelihood ra-

tio v2 test statistic and tests were two-sided P1NP frac14 N-terminal propeptide of type-1 collagen

AR

TIC

LE

8 | JNCI J Natl Cancer Inst 2018 Vol 110 No 8

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

In another study amplification of the 16q23 chromosomal regionincluding amplification of the MAF gene (22) was predictive ofbreast cancer metastasis to bone (23) However there remains aneed for a simple blood-based test in early breast cancer that canidentify patients with a high risk for development of bone metasta-sis Bone turnover markers are easily measured and are worthy ofadditional investigation in helping to meet this need

Funding

This work was supported by Cancer Research UK (through theawards of a Research Studentship to ER a Clinician ScientistFellowship to JEB and support to the Sheffield Experimental CancerMedicine Centre) and a grant from Novartis Pharmaceuticals

Notes

The funders had no role in the design of the study the collec-tion analysis or interpretation of the data the writing of the

manuscript or the decision to submit the manuscript for publi-cation Novartis provided academic grant support and suppliesof zoledronic acid (Zometa) for the AZURE trial The authors dis-close the following JB received fees from Novartis and Amgenfor Advisory Boards and Speakers Bureaux WG received feesfrom Celgene for statistical consultancy and honoraria fromJanssen DC received nonfinancial support from Novartis RC re-ceived institutional research grants from Amgen and Bayerfees from Novartis for expert testimony and lecture fees fromAmgen All other authors declared no conflicts

We wish to thank the AZURE trial patients who providedblood samples to support this research

References1 US Breast Cancer Statistics 2017 httpwwwbreastcancerorgsymptoms

understand_bcstatistics2 Coleman RE Metastatic bone disease Clinical features pathophysiology and

treatment strategies Cancer Treat Rev 200127165ndash763 Coleman RE Marshall H Cameron D et al Breast-cancer adjuvant therapy

with zoledronic acid N Engl J Med 20113651396ndash1405

Table 4 Adjusted prognostic categorical analyses according to a composite P1NP-CTX marker for both P1NP and CTX high vs not both high

Analysis

P1NP CTX Adjusted analysis

Both higheventscensored ()

Not both higheventscensored () HR (95 CI) P

Bone recurrence at any time 24118 (169) 80641 (111) 160 (099 to 248) 06First recurrence being in bone 18124 (127) 64657 (89) 150 (089 to 254) 14First distant recurrence (at any site) 34108 (239) 154567 (214) 100 (068 to 145) 99First recurrence being in bone only 11131 (77) 47674 (65) 126 (065 to 244) 50

P values were calculated using the likelihood ratio v2 test statistic and tests were performed at the two-sided 5 significance level 1-CTP frac14 pyridinoline cross-linked

carboxy-terminal telopeptide of type-1 collagen CI frac14 confidence interval CTX frac14 C-telopeptide of type-1 collagen HR frac14 hazard ratio IDFS frac14 invasive diseasendashfree sur-

vival P1NP frac14 N-terminal propeptide of type-1 collagen

50 100 150 200

2

4

6

8

10

Cut-point for P1NP

Prognostic analysis - for bone recurrence at any timeadjusted Cox model chi-squares for different P1NP cut-points(optimum cut-point at 64 P = 003)

P = 01

P = 05

P1NP = 70

Figure 3 v2 values from adjusted Cox proportional hazards model analyzing bone metastasis at any time by N-terminal propeptide of type-1 collagen (P1NP) with dif-

fering high vs normal P1NP cut-points Optimum cut-point observed at 64 ngmL with a corresponding P value of 003 P values were calculated using the likelihood ra-

tio v2 test statistic and tests were two-sided P1NP frac14 N-terminal propeptide of type-1 collagen

AR

TIC

LE

8 | JNCI J Natl Cancer Inst 2018 Vol 110 No 8

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

4 Gnant M Mlineritsch B Stoeger H et al Adjuvant endocrine therapy pluszoledronic acid in premenopausal women with early-stage breast cancer 62-month follow-up from the ABCSG-12 randomized trial Lancet Oncol 201112631ndash641

5 Paterson AH Anderson SJ Lembersky BC et al Oral clodronate for adjuvanttreatment of operable breast cancer (National Surgical Adjuvant Breast andBowel Project protocol B-34) A multicentre placebo-controlled randomizedtrial Lancet Oncol 201213734ndash742

6 Coleman RE Cameron D Dodwell D et al Adjuvant zoledronic acid inpatients with early breast cancer Final efficacy analysis of the AZURE(BIG 0104) randomized open-label phase 3 trial Lancet Oncol 201415997ndash1006

7 Coleman RE Powles T Paterson A et al Early Breast Cancer Clinical TrialsCollaborative Group (EBCTCG) Adjuvant bisphosphonate treatment in earlybreast cancer Meta-analyses of individual patient data from randomised tri-als Lancet 2015386(10001)1353ndash1361

8 Brown JE Cook RJ Major P et al Bone turnover markers as predictors of skel-etal complications in prostate cancer lung cancer and other solid tumors JNatl Cancer Inst 20059759ndash69

9 Coleman RE Major P Lipton A et al The predictive value of bone resorptionand formation markers in cancer patients with bone metastases receivingthe bisphosphonate zoledronic acid J Clin Oncol 2005234925ndash4935

10 Maemura M Iino Y Yokoe T et al Serum concentration of pyridinolinecross-linked carboxyterminal telopeptide of type I collagen in patients withmetastatic breast cancer Oncol Rep 200071333ndash1338

11 Lee J Vasikaran S Current recommendations for laboratory testing and useof bone turnover markers in management of osteoporosis Ann Lab Med 201232105ndash112

12 Lipton A Chapman J-AW Demers L et al Elevated bone turnover predicts forbone metastasis in postmenopausal breast cancer Results of NCIC CTGMA14 J Clin Oncol 2011293605ndash3610

13 Newson RB Comparing the predictive power of survival models usingHarrellrsquos c or Somersrsquo D Stata J 201010339ndash358

14 Ebeling PR Atley L Guthrie JR et al Bone turnover markers and bone densityacross the menopausal transition J Clin Endocrinol Metab 1996813336ndash3371

15 Riggs BL Melton LJ Involutional osteoporosis N Engl J Med 19863141676ndash1686

16 Botella S Restituto P Monreal I et al Traditional and novel bone remodellingmarkers in premenopausal and postmenopausal women J Clin EndocrinolMetab 201398E1740ndashE1748

17 Ottewell PD Wang N Meek J et al Castration-induced bone loss triggersgrowth of disseminated prostate cancer cells in bone Endocr Relat Cancer201421769ndash781

18 Wang N Reeves KJ Brown HK et al The frequency of osteolytic bone metas-tasis is determined by conditions of the soil not the number of seeds evi-dence from in vivo models of breast and prostate cancer J Exptl Clin CancerRes 201534124

19 Wang N Docherty FE Brown HK et al Prostate cancer cells preferentiallyhome to osteoblast-rich areas in the early stages of bone metastasisEvidence from in vivo models J Bone Miner Res 2014292688ndash2696

20 McCloskey E Paterson A Kanis J et al Effect of oral clodronate on bonemass bone turnover and subsequent metastases in women with primarybreast cancer Eur J Can 201046558ndash565

21 Westbrook JA Cairns DA Peng J et al CAPG and GIPC1 Breast cancer bio-markers for bone metastasis development and treatment J Natl Cancer Inst2016108djv360

22 Pavlovic M Arnal-Estape A Rojo F et al Enhanced MAF oncogene expressionand breast cancer bone metastasis J Natl Cancer Inst 2015107djv256

23 Coleman R Hall A Albanell J et al Effect of MAF amplification on treatmentoutcomes with adjuvant zoledronic acid in early breast cancer a secondaryanalysis of the international open-label randomised controlled phase 3AZURE (BIG 0104) trial Lancet Oncol 2017181543ndash1552

AR

TIC

LE

J Brown et al | 9 of 9

Downloaded from httpsacademicoupcomjnciadvance-article-abstractdoi101093jncidjx2804842040by Edinburgh University useron 14 February 2018

• djx280-TF1
• djx280-TF2
• djx280-TF3
• djx280-TF4
• djx280-TF5