Patient Preferences forTreatments toDelay BoneMetastases

A. Brett Hauber,1* Jorge Arellano,2 Yi Qian,2 Juan Marcos Gonz�alez,1 Joshua D. Posner,1

Ateesha F. Mohamed,1 Francesca Gatta,3 Bertrand Tombal,4 and Jean-Jacques Body5

1RTIHealth Solutions,ResearchTriangle Park,NorthCarolina2GlobalHealth Economics, Amgen Inc.,ThousandOaks,California

3GlobalHealth Economics, Amgen (Europe)GmbH,Zug, Switzerland4CliniquesUniversitaires Saint Luc, Institutde RechercheClinique,Brussels, Belgium

5CHUBrugmann,Universit�e Libre de Bruxelles, Brussels, Belgium

BACKGROUND. Most patients with advanced prostate cancer (PCa) develop bone metasta-ses (BM) and present with bone complications like fracture. Bone-targeted agents that preventmetastasis-induced bone complications can cause adverse events. Understanding how patientsview treatment options may optimize care. This study aimed to quantify how PCa patientsvalue a hypothetical treatment that delays BM but can cause osteonecrosis of the jaw (ONJ).The study also assessed the value patients place on avoiding metastasis-induced bonecomplications versus increased survival.METHODS. PCa patients from the United Kingdom (n¼ 201) and Sweden (n¼ 200) onandrogen-deprivation therapy or hormone therapy for � 3 years completed a 10-questiondiscrete-choice-experiment survey examining whether patients would accept a BM-delayingtreatment. Two time-tradeoff questions assessed patients’ willingness to tradeoff betweensurvival and bone complications. Percentages of patients choosing treatment were summa-rized by levels of treatment efficacy and ONJ risk. Odds ratios from a logit model were used toevaluate how patient and medication characteristics affected treatment choice. Proportions ofpatients choosing each tradeoff scenario were calculated.RESULTS. A majority of patients accepted treatment at the lowest benefit level (5-month BMdelay) and highest risk level (9% ONJ risk). PCa symptoms and prior treatment affectedpatient preferences. Nearly 80% of patients would tradeoff at least 3 months of survival toavoid bone complications.CONCLUSIONS. PCa patients in the UK and Sweden may value a medication that delaysBM, despite the risk of ONJ. Furthermore, patients were willing to tradeoff up to 5 months ofsurvival for prevention of bone complications. Prostate 74:1488–1497, 2014.# 2014 Wiley Periodicals, Inc.

Grant sponsor: Amgen Inc.

Funding: The study was conducted by RTI Health Solutions andfunded by Amgen Inc.

Ateesha F. Mohamed, MA, was employed by RTI Health Solutionsat the time of this analysis. She is currently an employee of BayerHealthcare.

Yi Qian, PhD, and Jorge Arellano, MSc, are employees of Amgen Inc.and hold stock in Amgen Inc.

Francesca Gatta, MSc, is an employee of Amgen (Europe) GmbH.

Bertrand Tombal, MD, PhD, is a member of a standing advisorycommittee or council, and has received consulting fees and researchgrants from Bayer, Amgen, Ferring, Sanofi-Aventis, Medivation, andAstellas.

Jean–Jacques Body, MD, PhD, has received lecture and consultingfees from Amgen Inc. and Novartis.

Disclosure Statement: Brett Hauber, PhD; Juan Marcos Gonz�alez,PhD; and Joshua D. Posner, BA, are employees of RTI HealthSolutions.

Bertrand Tombal and Jean–Jacques Body Co-senior authors�Correspondence to: A. Brett Hauber, RTI Health Solutions 200 ParkOffices Drive, PO Box 12194, Research Triangle Park, NC 27709.E-mail: abhauber@rti.orgReceived 15 May 2014; Accepted 1 July 2014DOI 10.1002/pros.22865Published online 17 August 2014 in Wiley Online Library(wileyonlinelibrary.com).

The Prostate 74:1488^1497 (2014)

� 2014 Wiley Periodicals, Inc.

KEY WORDS: bone pain; discrete-choice experiment; osteonecrosis of the jaw (ONJ);prostate cancer; skeletal-related events (SREs); time tradeoff

INTRODUCTION

Prostate cancer (PCa) is the second most frequentlydiagnosed cancer in men worldwide [1]. In men withnon-metastatic PCa, hormonal therapy is often used.However, when patients become resistant to hormonaltherapies, treatment options are limited before metas-tases develop [2,3]. Advanced PCa is characterized bya very high tropism to bone [4,5]. The skeleton is thefirst metastatic site in 80% of patients, and 90% ofpatients will subsequently have bone metastases [6–8]Bone metastases can alter the normal bone-remodelingprocesses and invade the surrounding structures,resulting in complications, such as pathologic frac-tures, pain, spinal cord compression, or the needfor radiotherapy or surgery to bone, also known asskeletal-related events (SREs) [9].

Several agents are now available that delay first andsubsequent SREs in patients with PCa who developbone metastases [10,11]. In addition, denosumab(Amgen Inc., Thousand Oaks, CA) was shown to haveefficacy in delaying onset of the first bone metastasisin patients with non-metastatic PCa whose diseaseprogresses while receiving hormone therapy [12].However, no medications are currently approved fordelaying bone metastases in non-metastatic castration-resistant PCa patients. Clinical endpoints that couldestablish efficacy for products used to treat patientswith non-metastatic castration-resistant PCa wereconsidered at an FDA Oncologic Drugs AdvisoryCommittee (ODAC) meeting in 2011 [13]. The ODACmembers agreed that prevention or delay of bonemetastases would be a clinically relevant endpoint, butthat it was uncertain what duration of metastasesprevention would be acceptable to prove clinicalbenefit. Understanding how patients with PCa viewtreatment options for delaying bone metastases couldinform healthcare policy [14].

One of the methods most commonly used to assesspreferences for medical interventions, products, orpolicies is the discrete-choice experiment (DCE) [15].With a DCE, a respondent chooses between two ormore treatment options, each defined by a common setof features or attributes. This systematic method elicitstradeoff information to quantify the relative importancerespondents place on various treatment characteristicsor outcomes. The DCE is based on the premise thattreatments are composed of a set of attributesor associated outcomes (e.g., efficacy and risk ofassociated adverse events) and that the relative value of

a particular treatment to an individual is a function ofthese attributes. Numerous studies have demonstratedpatients’ willingness to accept tradeoffs among theattributes associated with a particular medicalintervention [15,16]. Decisions on cancer treatments maybe particularly sensitive to patient preference astreatments often have associated risks, with the optimaltreatment choice being heavily dependent upon howa patient weighs these benefits and risks [17].Since understanding the desire or reluctance for aparticular therapy from the patients’ perspective canhelp to inform healthcare decisions, DCEs are beingincreasingly used in healthcare to ascertain how patientsview treatment options [15,18,19].

The study presented here used a DCE to quantifythe willingness of patients with PCa to receive ahypothetical treatment that provides a benefit ofdelaying bone metastases but may also be associatedwith an adverse event. In addition, this study evaluat-ed how patients value the benefit of preventing SREsafter bone metastases develop using time tradeoffquestions.

MATERIALS AND METHODS

Patients

The Office of Research Protection and Ethics atRTI International reviewed the study research plan.A survey-research company recruited patients in theUnited Kingdom (UK) and Sweden from existingonline panels. Eligible patients were � 18 years old,had a self-reported physician diagnosis of PCa, andhad received androgen-deprivation therapy or hor-mone therapy for at least 3 years. Patients wereinformed that neither benefits nor risks were derivedby participating to this study and provided onlineinformed consent. Preference data were collected usingethical considerations associated with the approvedresearch plan. Demographic information and itemsabout the patients’ experiences with PCa were alsocollected. Patients who completed the survey receivedthe equivalent of s55 in cash.

Survey Instrument

Best practices published by the International Societyof Pharmacoeconomics and Outcomes Research werefollowed in designing and administering the DCE(also known as a choice-format conjoint analysis) [16].

Treatment Preferences to Delay BoneMets 1489

The Prostate

A web-enabled DCE was adapted for online adminis-tration. Research indicates that online DCEs are notstatistically different from DCEs administered via face-to-face interviews [20,21]. The DCE survey consistedof 10 choice questions and was estimated to takeapproximately 25–30min to complete (Fig. 1A). Atleast 150 respondents were needed to estimate apreference model with acceptable confidence intervalsfor all parameters [22,23].

The preference-elicitation DCE that was developedasked respondents to evaluate a series of choicesbetween pairs of constructed treatment-decision alter-natives. The descriptions of the attributes provided tothe patients are shown in Table I. Each question askedpatients to decide whether they would prefer taking ahypothetical prophylactic medication to prevent bonemetastases with a risk of an adverse event (osteonec-rosis of the jaw [ONJ]) or declining treatment with thismedication and, thus, not receiving treatment benefitor risk. Each hypothetical prophylactic medicationwas defined by how long it could delay the occurrenceof bone metastases (a delay of 0–23 months) and by itsassociated risk for ONJ (a risk of 0–9%) (Table I). Thesedefinitions encompass the ranges observed in currentresearch [12,24] as well as the range over whichpatients were willing to accept tradeoffs amongattributes as determined in qualitative interviews withpatients. Bone-targeted agents that are approved toprevent SREs in patients with solid tumors have beenassociated with adverse events, such as ONJ, renaltoxicity, and hypocalcemia [25]. However, not allbone-targeted agents cause renal toxicity, and hypocal-cemia (which is generally preventable and treatable)has been observed infrequently with use of bone-targeted agents [12]. Thus, ONJ was selected as theprimary adverse event in the profile design for thehypothetical medication. A full factorial design wasemployed with all possible combinations of attributelevels. The primary endpoints were the proportions ofrespondents who would choose treatment associatedwith different levels of efficacy and risk. Secondaryendpoints were estimates of the impact of bothmedication attributes and patient characteristics on thelikelihood of choosing prophylaxis.

The survey was validated by conducting open-ended interviews with 15 patients in the United States.These interviews tested the clarity of the surveyinstrument and the appropriateness of the descriptiveinformation, confirmed that the two attributes includ-ed in the survey (delay in bone metastases and riskof ONJ) were salient to patients, tested the level ofdifficulty of the questions, and assessed the willing-ness of patients to accept tradeoffs among attributeswhen evaluating hypothetical medications. In addi-tion, five patients from the UK and five patients from

Sweden were interviewed to assess the cultural appro-priateness of the survey instrument (accuracy of thesurvey’s Swedish translation was also checked in theinterviews with Swedish patients).

TimeTradeoff

To further evaluate patients’ views of the impact ofbone metastases, time-tradeoff (TTO) questions pre-sented scenarios that assessed patients’ willingness totradeoff between life years with and without metasta-sis-induced bone complications (i.e., SREs) (Fig. 1B).

Statistical Analyses

All respondents in the UK and Sweden answeredall 10 choice questions and were included in theanalysis. The percentage of patients in the UK orSweden who chose treatment with the hypotheticalprophylaxis medication were analyzed by the levels oftreatment efficacy (delay in bone metastases) and ofONJ risk.

The effect of treatment attributes and patient char-acteristics on the likelihood of choosing treatment wasexamined using a logit model. Results from the logitmodel were calculated as odds ratios representingthe odds that a patient with given characteristicswould choose a treatment with specific features.The proportions of patients willing to tradeoff 0 to5 months of survival to avoid bone complications weresummarized.

RESULTS

Patient Characteristics

The DCE survey was administered in February andMarch of 2013 to patients with PCa in the UK andSweden. For the UK sample, 243 patients with a self-reported physician diagnosis of PCa were invited toparticipate in the survey; 228 patients (93.8%)responded and 219 (96.1%) respondents were eligibleto participate. Of those patients eligible to participate,201 (91.8%) consented to participate in the survey andanswered all 10 choice questions. For the Swedishsample, 57,000 patients from a household sample wereinvited to participate in the survey; 13,108 patients(23.0%) responded and 1293 (9.9%) respondents wereeligible to participate. Of those patients eligible toparticipate, 200 (15.5%) consented to participate in thesurvey and answered all 10 choice questions.

Patient characteristics between the UK and Swedishgroups were generally similar (Table II). The mean agewas about 72 years, and the majority of patients weremarried, retired, and currently receiving treatment forPCa. The most common treatment was hormone

1490 Hauber et al.

The Prostate

Fig. 1. Treatment-choice example question and time-tradeoff scenarios. A.Treatment-choice example question.ONJ¼ osteonecrosis ofthejaw.B.Survival time tradeoff scenarios.TTO¼ time tradeoff.

Treatment Preferences to Delay BoneMets 1491

The Prostate

therapy (95.5% of UK patients and 89.3% of Swedishpatients) followed by radiotherapy (63.2% of UKpatients and 59.4% of Swedish patients).

Discrete-Choice Experiment

Results from the DCE indicated that in both the UKand Swedish groups, patients were more willing to

receive the hypothetical medication when it wasassociated with increased benefit (i.e., increased delayof bone metastases) and with a lower risk of ONJ(Fig. 2). With a benefit of 23 months of delayed bonemetastases, the majority of patients chose treatment,with the amount of ONJ risk appearing to havelittle effect on this decision. However, even withonly 5 months’ delay of bone metastases, over 50% of

TABLEI. Attributes andLevels for theChoiceQuestions

Attribute Definition Level

Disease progression in 48 monthsStage 1: Prostate cancer that has not spread to

the boneIn this stage a person has prostate cancer,but does not have bone metastases. Inthis stage a person can start takingmedicines that will help delay bonemetastases. Some medicines can delaybone metastases longer than others. Thebetter the medicine is working, thelonger the time until the person getsbone metastases.

48 months (23-month delaya)38 months (13-month delay)30 months (5-month delay)25 months (0-month delay)

Stage 2: Prostate cancer with bone metastasesb In this stage a person has bone metastases.That means the cancer has gotten worseand has spread to the bones. When aperson has bone metastases the doctorwill begin chemotherapy immediately.Chemotherapy is usually given every3 weeks through an intravenous (IV)infusion. During infusions, a nurse willinsert a needle into a vein in the arm orthrough a tube that is attached to a bagfilled with liquid medicine. It takesapproximately 1 hr for all of themedicine to drip from the bag into yourvein. Side effects of chemotherapy caninclude; pain, diarrhea, constipation,mouth sores, hair loss, nausea andvomiting, tiredness that does not goaway even after resting, and bleedingthat is not caused by injury (for e.g.,nosebleeds or bleeding from the gums).Also, while taking chemotherapy aperson’s body may not be able to fightinfections normally.

0 months10 months18 months23 months

Chance of a side effect causing a problem withyour teeth and/or jawbone

Some medicines used to delay bonemetastases have a side effect that cancause problems with your teeth and/orjaw bone. These problems are caused bya condition called ONJ (osteonecrosis ofthe jaw). ONJ occurs when the bone inthe jaw (inside the mouth) is exposed(usually after a tooth is pulled). Theexposed area of the bone can be painful.

None1 out of 100 (1%)4 out of 100 (4%)9 out of 100 (9%)

aRefers to how long the hypothetical medication would delay cancer spreading to the bone.bThe levels for Stage 2 were calculated as the difference between 48 months and the number of months in Stage 1 (e.g., if the level forStage 1 was 30 months, then the level for Stage 2 was 18 months).

1492 Hauber et al.

The Prostate

TABLEII. Characteristics of Respondents Includedin the LogitModel

Characteristic UK n¼ 201 Sweden n¼ 200 P Valuea

Mean (SD) age, years 71.6 (6.7) 71.5 (9.5) 0.94Relationship status, n (%)Married/living as married/civil partnership 178 (88.6%) 163 (81.5%) 0.102Widowed 9 (4.5%) 8 (4.0%)Divorced or separated 8 (4.0%) 19 (9.5%)Single 6 (3.0%) 10 (5.0%)

Employment status, n (%)b

Employed full-time 7 (3.5%) 17 (8.5%) 0.025Employed part-time 8 (4.0%) 1 (0.5%)Self-employed 14 (7.0%) 14 (7.0%)Student 0 1 (0.5%)Retired 168 (83.6%) 164 (82.0%)Unable to work due to disability or health problem 4 (2.0%) 1 (0.5%)Unemployed and looking for work 0 2 (1.0%)

Highest educational qualification, n (%)1–4 O levels/CSEs/GCSEs (any grades), entry level foundation diploma 18 (9.0%) 14 (7.0%) < 0.001First degree (eg, BA) or higher degree (eg, MA, MSc, PhD) 49 (24.4%) 26 (13.0%)Other educational diplomas 60 (29.9%) 146 (73.0%)Professional qualifications (eg, nursing) 46 (22.9%) 0Other work-related or foreign qualifications 17 (8.5%) 14 (7.0%)

No qualifications 11 (5.5%) 0How long ago informed of prostate cancer diagnosis, n (%)< 6 months ago 3 (1.5%) 9 (4.5%) 0.157At least 6 months ago but < 1 year ago 14 (7.0%) 8 (4.0%)At least 1 year ago but < 2 years ago 26 (12.9%) 23 (11.5%)At least 2 years ago but < 5 years ago 74 (36.8%) 63 (31.5%)At least 5 years ago 84 (41.8%) 97 (48.5%)

Currently receiving treatment for prostate cancer, n (%)Yes 133 (66.2%) 124 (62.0%) 0.277No 68 (33.8%) 74 (37.0%)Don’t know 0 2 (1.0%)

Prostate cancer treatment ever received, n (%)Radiotherapy (i.e., external beam radiation; “seed” insertions) 127 (63.2%) 117 (59.4%) 0.437Surgery 36 (17.9%) 36 (18.3%) 0.925Hormone therapy 192 (95.5%) 176 (89.3%) 0.020Other 27 (13.4%) 11 (5.6%) 0.008

Prostate cancer symptoms experienced since first informed of prostate cancer diagnosis, n (%)Trouble urinating 67 (33.3%) 62 (31.0%) 0.617Less force in the stream of urine 109 (54.2%) 104 (52.0%) 0.655Blood in the urine or semen 26 (12.9%) 20 (10.0%) 0.356Swelling in the legs 21 (10.4%) 21 (10.5%) 0.986Discomfort in the pelvic area 39 (19.4%) 26 (13.0%) 0.082Bone pain 33 (16.4%) 13 (6.5%) 0.002Other type of pain 14 (7.0%) 14 (7.0%) 0.989Incontinence 39 (19.4%) 55 (27.5%) 0.056Other 22 (10.9%) 17 (8.5%) 0.409None of the above 38 (18.9%) 49 (24.5%) 0.174

Ever experienced ONJ symptoms, n (%)Yes 3 (1.5%) 7 (3.5%) 0.002No 187 (93.0%) 162 (81.0%)Don’t know 11 (5.5%) 31 (15.5%)

Dental problems treated by a doctor or dentist in the past year, n (%)Yes 91 (45.3%) 34 (17.0%) < 0.001No 110 (54.7%) 163 (81.5%)

(Continued)

Treatment Preferences to Delay BoneMets 1493

The Prostate

patients in the UK and Swedish groups chose treat-ment at each ONJ risk level (Fig. 2). Results from theDCE also indicated that 18% of UK patients and 36%of Swedish patients would take a medication withoutany proven efficacy or toxicity.

Logit Analysis

A logit analysis was conducted to provide insightinto how certain factors or patient characteristics couldimpact a patient’s choice to receive or not receive aprophylactic treatment (Table III). In general, thepresence of symptoms and treatment experiencesaffected patient preference for treatment. In patientsfrom the UK, prior ONJ symptom experience signifi-cantly (P¼ 0.001) increased the likelihood that patientswould choose treatment. Factors that significantlyincreased the likelihood that the Swedish patientswould choose treatment included longer time sincediagnosis of PCa (P¼ 0.029) and currently receivingtreatment for PCa (P¼ 0.032).

Time-Tradeoff Scenarios

Tradeoff scenarios examined months of life patientswere willing to forgo to avoid bone complications dueto metastases (Table IV). Greater than 90% of allpatients were willing to forego survival time to avoidbone complications. Patients from the UK were morelikely to choose shorter life spans without bone compli-cations (i.e., SREs) compared with Swedish patients. Toavoid bone complications, 52.2% of UK patients and

TABLEII. (Continued)

Characteristic UK n¼ 201 Sweden n¼ 200 P Valuea

Don’t know 0 3 (1.5%)Ever received chemotherapy, n (%)

Yes 22 (10.9%) 14 (7.0%) < 0.001No 178 (88.6%) 159 (79.5%)Don’t know 1 (0.5%) 27 (13.5%)

Informed by doctor that cancer has spread to bone, n (%)Yes 42 (20.9%) 23 (11.5%) 0.038No 156 (77.6%) 174 (87.0%)Don’t know 3 (1.5%) 3 (1.5%)

BA¼bachelor of arts; CSE¼ certificate of secondary education; GCSE¼ general certificate of secondary education; MA¼master of arts;MSc¼master of science; ONJ¼ osteonecrosis of the jaw; PhD¼doctor of philosophy; SD¼ standard deviation; UK¼United Kingdom.aP values were computed using either the Pearson chi-square test for categorical variables or the Student’s t-test for continuousvariables.bPercentages displayed excluded missing values.

UK Patients (N = 201)

A UK

B SwedenSweden Patients (N = 200)

Delay in bone metastases (months)

Pro

porti

on o

f res

pond

ents

choo

sing

trea

tmen

t (%

)P

ropo

rtion

of r

espo

nden

tsch

oosi

ng tr

eatm

ent (

%)

100

90

80

70

60

50

40

30

20

10

00 2 4 6 8 10 12 14 16 18 20 22 24 26

82%

91% 91%88%

80%

84%

72%

65%

52%

77%

87% 89%87%82%84%

78%

60%

70%

Delay in bone metastases (months)

0 2 4 6 8 10 12 14 16 18 20 22 24 26

100

90

80

70

60

50

40

30

20

10

0

1% ONJ risk

4% ONJ risk

9% ONJ risk

1% ONJ risk

4% ONJ risk

9% ONJ risk

Fig. 2. Proportion of respondents choosing treatment definedbydelay inbonemetastases andbyriskofONJ. A.Results from theUKpatientgroup.B.Results fromtheSwedishpatientgroup.ONJ¼osteonecrosis of thejaw;UK¼UnitedKingdom.

1494 Hauber et al.

The Prostate

26.0% of Swedish patients were willing to forgo at least5 months of life, while 21.4% of UK patients and 58.5%of Swedish patients were willing to forgo between3 and 5 months of life. Whether differences in timetradeoffs were related to patient characteristics (such as

whether or not prior chemotherapy was received) wasnot tested due to the small size of the subgroups.

DISCUSSION

Understanding how patients view the benefits andrisks associated with treatment may help to optimizepatient care and healthcare policy. In this study, weexamined whether patients with PCa would take amedication that had the possible benefit of delayingbone metastases, but had an associated risk of ONJ,over no treatment. A risk of ONJ was included in theDCE because ONJ is an infrequent but serious compli-cation associated with bone-targeted agents used toprevent SREs in patients with bone metastases fromsolid tumors [26]. The findings presented here indicatethat the UK and Swedish patients surveyed perceivedvalue in a medication that delayed bone metastasesand were willing to receive this medication though it

TABLEIII. Logit Results

VariableUK (n¼ 201)

Odds Ratioa (P value)Sweden (n¼ 200)

Odds Ratioa (P value)

Delay of bone metastases in monthsb 1.075 (< 0.001) 1.063 (< 0.001)Percentage risk of ONJb 0.857 (< 0.001) 0.917 (< 0.001)Medicationc 0.062 (< 0.001) 0.195 (< 0.001)Aged 0.676 (0.136) 0.738 (0.231)Civil statuse 1.675 (0.173) 1.169 (0.582)Educationf 0.872 (0.613) 1.614 (0.129)Employment statusg 1.054 (0.884) 0.703 (0.331)Time since diagnosish 1.163 (0.662) 1.912 (0.029)Currently receiving treatmenti 1.126 (0.697) 1.669 (0.032)Prostatecancer symptomsj 1.402 (0.282) 1.231 (0.508)Prior ONJ symptom experiencek 16.927 (0.001) 2.611 (0.134)Dental problemsl 1.056 (0.840) 0.810 (0.491)Chemotherapy experiencem 0.617 (0.269) 1.247 (0.665)Bone metastasis experiencen 1.055 (0.855) 0.964 (0.943)Constanto 1.992 (0.219) 1.175 (0.744)

ONJ¼ osteonecrosis of the jaw; UK¼United Kingdom.aOdds ratio > 1 means a treatment feature or patient characteristic increases the likelihood of choosing treatment. An odds ratio < 1means a treatment feature or patient characteristic decreases the likelihood of choosing treatment. An odds ratio of 1 means a treatmentfeature or patient characteristic does not affect the likelihood of choosing treatment.bContinuous variable.cShown in last choice question with same levels as opt out.d1 if respondent was older than the median age of 72 years; 0 otherwise.e1 if respondent reported being married, living as married, or civil partnership; 0 otherwise.f1 if respondent reported having a first degree, higher degree, or professional qualifications; 0 otherwise.g1 if respondent was employed full time, part time, or self-employed; 0 otherwise.h1 if respondent was first told of prostate cancer at least 2 years ago; 0 otherwise.i1 if respondent was currently receiving any treatment for prostate cancer; 0 otherwise.j1 if respondent indicated that the symptoms of prostate cancer presented were not being experienced; 0 otherwise.k1 if respondent indicated having experienced symptoms associated with ONJ; 0 otherwise.l1 if respondent indicated having dental problems that were treated by a doctor or dentist; 0 otherwise.m1 if respondent indicated having received chemotherapy; 0 otherwise.n1 if respondent indicated that the cancer had spread to the bones; 0 otherwise.oMean odds ratio of treatment features or personal characteristics not considered in the logit model.

TABLEIV. Willingness toTrade Survival Time toAvoidBoneComplications

Months of life willing to forgoto avoid bone complicationsfrom metastasesa

UKn¼ 201

Swedenn¼ 200

At least 5 months, n (%) 105 (52.2%) 52 (26.0%)Between 3 and 5 months, n (%) 43 (21.4%) 117 (58.5%)Between 1 and 3 months, n (%) 34 (16.9%) 24 (12.0%)0 months, n (%) 19 (9.5%) 7 (3.5%)

UK, United Kingdom.aAssuming longevity of 9 months.

Treatment Preferences to Delay BoneMets 1495

The Prostate

has an associated risk of ONJ. More patients chose toreceive the medication as the number of months’ delayof bone metastases increased and the associated risk ofONJ decreased. A majority of patients in both countrieschose treatment with the medication even at the lowestlevel of bone-metastasis delay (5 months) and highestlevel of ONJ risk (9%). This finding suggests thatpatients with PCa would value a treatment that coulddelay bone metastases by a timeframe as small as5 months. The logit analysis indicated that the factorsmost likely to increase the likelihood of choosingprophylaxis over no treatment included prior ONJsymptoms, time since diagnosis of PCa, and whethertreatment for PCa was currently being received.

It is unclear why 18% of UK patients and 36% ofSwedish patients were willing to take a medicationthat had no proven effect. Possibilities to explain thisfinding include that some patients may have beenwilling to try any available medication and/or somepatients may have assumed there were unexplainedbenefits associated with the medication.

The additional analysis to understand how patientswould value absence of bone complications (i.e., SREscaused by bone metastases) compared with longerduration of survival showed that nearly 80% of patientsfrom both countries would forgo at least 3 months ofsurvival time to avoid an SRE. Results suggested thatcompared with the Swedish patients, those from theUK preferred shorter duration of survival without bonecomplications over longer duration of survival withbone complications. Reasons for this difference mightinclude the different methods used in the two countriesto select patients and the different healthcare systemsin the two countries. The choice of SRE prevention oversurvival time by the majority of patients is worthnoting. Healthcare providers and regulators may as-sume that patients value survival above other consider-ations, and these assumptions may influence treatmentand drug-approval decisions.

A limitation of this study is that simulated clinicaldecisions may not have the same clinical, financial,and emotional consequences of actual decisions.Patients may make different choices when presentedwith actual situations. Another limitation of thisstudy is that some patients surveyed may alreadyhave experienced bone metastases, making it difficultfor them to imagine a situation in which a medicationwas available that would have delayed this event.Finally, only patients from the UK and Sweden weresurveyed (these two countries were selected so as tocover a broad range of expectations and experiencesfor healthcare). However, the similar results in twodifferent countries imply that we would expect tofind similar results in patients with PCa in othercountries, although the numbers may vary. Studies

would need to be conducted to confirm that patientsin other countries are willing to accept a medication,even with risks, to delay the development of bonemetastases.

Currently, medications are available for treatingSREs in patients with PCa; these agents also haveassociated adverse events [10,26]. There is also evi-dence that denosumab may have an effect in prevent-ing bone metastases in PCa [12]. The work presentedhere suggests patients with PCa may perceive a benefitin receiving these medications. It may be useful toincorporate patients’ perspective from studies such asthis one to inform future clinical trial design andregulatory review of new therapies. The lowest delayshown in this study, 5 months, may be reasonable todemonstrate clinical benefit in future trials that assessthe ability of an agent to delay development of bonemetastases in patients with PCa. Since this is a singlestudy conducted in one tumor type, future workshould repeat this experiment in other patient popula-tions with PCa and also in patient populations withother tumor types such as breast cancer.

CONCLUSIONS

This study suggests that the UK and Swedish PCapatients surveyed were willing to receive a medicationthat delays bone metastases despite an associated riskof ONJ. In addition, most patients were willing totradeoff 3 to 5 months of survival to avoid bonecomplications, highlighting the debilitating impact ofthese complications on patients.

ACKNOWLEDGMENTS

The authors thank Linda Rice, PhD, at Amgen Inc.for medical writing assistance and Amy Foreman-Wykert, PhD for editorial assistance. We also thankF. Reed Johnson and Jonas Hjelmgren for assistancewith pretesting the survey instrument in Sweden.

REFERENCES

1. Ferlay J, Soerjomataram I, Ervik M, Dikshit R, Eser S, Mathers C,Rebelo M, Parkin DM, Forman D, Bray F. GLOBOCAN 2012v1.0, Cancer Incidence and Mortality Worldwide: IARC Cancer-Base No. 11 [Internet]. Lyon, France: International Agency forResearch on Cancer 2013. Available from: http://globocan.iarc.fr/ Accessed on 30 January 2014.

2. Amaral TM, Macedo D, Fernandes I, Costa L. Castration-resistant prostate cancer: mechanisms, targets, and treatment.Prostate Cancer 2012;2012:327253.

3. Sharifi N, Gulley JL, Dahut WL. Androgen deprivation therapyfor prostate cancer. JAMA 2005;294:238–244.

4. Clines GA, Guise TA. Molecular mechanisms and treatment ofbone metastasis. Expert Rev Mol Med 2008;10:e7.

1496 Hauber et al.

The Prostate

5. Lamothe F, Kovi J, Heshmat MY, Green EJ. Dissemination ofprostatic carcinoma: an autopsy study. J Natl Med Assoc 1986;78:1083–1086.

6. Bubendorf L, Schopfer A, Wagner U, Sauter G, Moch H, Willi N,Gasser TC, Mihatsch MJ. Metastatic patterns of prostate cancer:an autopsy study of 1,589 patients. Hum Pathol 2000;31:578–583.

7. Nelson JB, Love W, Chin JL, Saad F, Schulman CC, Sleep DJ,Qian J, Steinberg J, Carducci M. Phase 3, randomized, controlledtrial of atrasentan in patients with nonmetastatic, hormone-refractory prostate cancer. Cancer 2008;113:2478–2487.

8. Sartor O, Eisenberger M, Kattan MW, Tombal B, Lecouvet F.Unmet needs in the prediction and detection of metastases inprostate cancer. Oncologist 2013;18:549–557.

9. Fizazi K, Carducci M, Smith M, Damiao R, Brown J, Karsh L,Milecki P, Shore N, Rader M, Wang H, Jiang Q, Tadros S, Dansey R,Goessl C. Denosumab versus zoledronic acid for treatment ofbone metastases in men with castration-resistant prostate cancer:a randomised, double-blind study. Lancet 2011;377:813–822.

10. Hoffman-Censits J, Fu M. Chemotherapy and targeted therapies:are we making progress in castrate-resistant prostate cancer?Semin Oncol 2013;40:361–374.

11. MacVicar GR, Hussain MH. Emerging therapies in metastaticcastration-sensitive and castration-resistant prostate cancer. CurrOpin Oncol 2013;25:252–260.

12. Smith MR, Saad F, Coleman R, Shore N, Fizazi K, Tombal B,Miller K, Sieber P, Karsh L, Damiao R, Tammela TL, Egerdie B,Van Poppel H, Chin J, Morote J, Gomez-Veiga F, Borkowski T,Ye Z, Kupic A, Dansey R, Goessl C. Denosumab and bone-metastasis-free survival in men with castration-resistant prostatecancer: results of a phase 3, randomised, placebo-controlled trial.Lancet 2012;379:39–46.

13. Summary Minutes of the FDA Oncologic Drugs AdvisoryCommittee Meeting, September 14, 2011. Available from: http://www.fda.gov/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/OncologicDrugsAdvisoryCommittee/ucm235829.htm,Accessed on 30 January 2014.

14. de Bekker-Grob EW, Bliemer MC, Donkers B, Essink-Bot ML,Korfage IJ, Roobol MJ, Bangma CH, Steyerberg EW. Patients’and urologists’ preferences for prostate cancer treatment: adiscrete choice experiment. Br J Cancer 2013;109:633–640.

15. Marshall D, Bridges JF, Hauber B, Cameron R, Donnalley L, Fyie K,Johnson FR. Conjoint analysis applications in health - how arestudies being designed and reported?: an update on current practicein the published literature between 2005 and 2008. Patient2010;3:249–256.

16. Bridges JF, Hauber AB, Marshall D, Lloyd A, Prosser LA,Regier DA, Johnson FR, Mauskopf J. Conjoint analysis applica-tions in health–a checklist: a report of the ISPOR Good ResearchPractices for Conjoint Analysis Task Force. Value Health 2011;14:403–413.

17. Blinman P, King M, Norman R, Viney R, Stockler MR. Prefer-ences for cancer treatments: an overview of methods andapplications in oncology. Ann Oncol 2012;23:1104–1110.

18. Hauber AB, Arden NK, Mohamed AF, Johnson FR, Peloso PM,Watson DJ, Mavros P, Gammaitoni A, Sen SS, Taylor SD.A discrete-choice experiment of United Kingdom patients’willingness to risk adverse events for improved function andpain control in osteoarthritis. Osteoarthritis Cartilage 2013;21:289–297.

19. Mohamed AF, Hauber AB, Neary MP. Patient benefit-riskpreferences for targeted agents in the treatment of renal cellcarcinoma. Pharmacoeconomics 2011;29:977–988.

20. Marta-Pedroso C, Freitas H, Domingos T. Testing for the surveymode effect on contingent valuation data quality: A case studyof web based versus in person interviews. Ecol Econ 2007;62:388–398.

21. Nielsen JS. Use of the internet for willingness-to-pay survey:A comparison of face-to-face and web-based interviews. ResEnergy Econ 2011;33:119–129.

22. Orme BK. Getting started with conjoint analysis: strategies forproduct design and pricing research. Madison, WI: ResearchPublishers LLC; 2006.

23. Reed Johnson F, Lancsar E, Marshall D, Kilambi V, Muhlbacher A,Regier DA, Bresnahan BW, Kanninen B, Bridges JF. Constructingexperimental designs for discrete-choice experiments: reportof the ISPOR Conjoint Analysis Experimental Design GoodResearch Practices Task Force. Value Health 2013;16:3–13.

24. Smith MR, Saad F, Oudard S, Shore N, Fizazi K, Sieber P, Tombal B,Damiao R, Marx G, Miller K, Van Veldhuizen P, Morote J, Ye Z,Dansey R, Goessl C. Denosumab and bone metastasis-free survivalin men with nonmetastatic castration-resistant prostate cancer:exploratory analyses by baseline prostate-specific antigen doublingtime. J Clin Oncol 2013;31:3800–3806.

25. El-Amm J, Freeman A, Patel N, Aragon-Ching JB. Bone-targetedtherapies in metastatic castration-resistant prostate cancer:evolving paradigms. Prostate Cancer 2013;2013:210686.

26. Lipton A, Balakumaran A. Denosumab for the treatment ofcancer therapy-induced bone loss and prevention of skeletal-related events in patients with solid tumors. Expert Rev ClinPharmacol 2012;5:359–371.

Treatment Preferences to Delay BoneMets 1497

The Prostate