Cost-effectiveness of population based BRCA testing with varying Ashkenazi Jewish ancestry

12 Pages·2017·1.1 MB·English

Original Research
ajog.org
GYNECOLOGY
Cost-effectiveness of population based BRCA testing with
varying Ashkenazi Jewish ancestry
RanjitManchanda,MRCOG,PhD;ShreeyaPatel,MSc;AntonisC.Antoniou,PhD;EphratLevy-Lahad,PhD;
ClareTurnbull,PhD;D.GarethEvans,PhD;JohnL.Hopper,PhD;RobertJ.Macinnis,PhD;UshaMenon,MD;FRCOG;
IanJacobs,MD,FRCOG;RosaLegood,PhD
BACKGROUND: Population-based BRCA1/BRCA2 testing has been RESULTS: Population testing for BRCA mutations is cost-saving in
foundtobecost-effectivecomparedwithfamilyhistoryebasedtestingin Ashkenazi-Jewishwomenwith2,3,or4grandparents(22-33dayslife-
Ashkenazi-Jewishwomenwere>30yearsoldwith4Ashkenazi-Jewish gained)intheUnitedKingdomand1,2,3,or4grandparents(12-26days
grandparents. However, individuals may have 1, 2, or 3 Ashkenazi- life-gained) in the United States populations, respectively. It is also
Jewish grandparents, and cost-effectiveness data are lacking at these extremely cost-effective in women in the United Kingdom with just 1
lower BRCA prevalence estimates. We present an updated cost- Ashkenazi-Jewish grandparent with an incremental cost-effectiveness
effectiveness analysis of population BRCA1/BRCA2 testing for women ratio of £863 per quality-adjusted life-years and 15 days life gained.
with1,2,and3Ashkenazi-Jewishgrandparents. Results show that population-testing remains cost-effective at the
STUDYDESIGN:Decisionanalysismodel. £20,000e30000 per quality-adjusted life-years and $100,000 per
METHODS: Lifetime costs and effects of population and family quality-adjusted life-years willingness-to-pay thresholds for all 4
historyebasedtestingwerecomparedwiththeuseofadecisionanalysis Ashkenazi-Jewishgrandparentscenarios,with 95%simulationsfound
!
model. 56% BRCA carriers are missed by family history criteria alone. tobecost-effectiveonprobabilisticsensitivityanalysis.Population-testing
Analyses were conducted for United Kingdom and United States pop- remainscost-effectiveintheabsenceofreductioninbreastcancerrisk
ulations.ModelparameterswereobtainedfromtheGeneticCancerPre- fromoophorectomyandatlowerrisk-reducingmastectomy(13%)orrisk-
dictionthroughPopulationScreeningtrialandpublishedliterature.Model reducingsalpingo-oophorectomy(20%)rates.
parametersandBRCApopulationprevalenceforindividualswith3,2,or1 CONCLUSION: Population testing for BRCA mutations with varying
Ashkenazi-Jewishgrandparentwereadjustedfortherelativefrequencyof levels of Ashkenazi-Jewish ancestry is cost-effective in the United
BRCA mutations in the Ashkenazi-Jewish and general populations. In- KingdomandtheUnitedStates.Theseresultssupportpopulationtestingin
cremental cost-effectiveness ratios were calculated for all Ashkenazi- Ashkenazi-Jewish women with 1-4 Ashkenazi-Jewish grandparent
Jewish grandparent scenarios. Costs, along with outcomes, were dis- ancestry.
counted at 3.5%. The time horizon of the analysis is “life-time,” and
perspective is “payer.” Probabilistic sensitivity analysis evaluated model Keywords:ancestry,AshkenaziJewish,BRCA,cost-effectiveness,
uncertainty. populationtesting
Population-based BRCA1/BRCA2 FH-driven clinical criteriaebased Assessments of the full health economic
testing has been investigated testing.1,2 Randomized trial data show implications are critical to inform any
extensivelyintheAshkenazi-Jewish(AJ) thatpopulation-testingdoesnotimpact potential policy change. A health eco-
populationandhasbeenshowntohave psychological well-being or quality-of- nomic assessment allows for the evalua-
severaladvantagescomparedwithfamily life detrimentally.2 There is an overall tionoftheoverallcostsandbenefitsforthe
history (FH)ebased testing. FH-based reduction in anxiety, distress, and un- genetic testing of BRCA1/BRCA2 muta-
testing requires individuals to fulfil certainty,2-4 although higher levels of tionsinwomenofdifferingAJancestry.
stringentclinicalcriteria;however,many cancer-related distress in those women We previously used a decision
BRCA1/BRCA2carriersdonotmeetthis who test positive are reported.3,5 Addi- analytical model to compare the costs
clinical threshold for genetic testing tionally delivery of testing through a and consequences of population-based
based on cancer history in the family. population-based model is acceptable, testing in AJ women who were 30
!
Thisapproachresultsin>50%ofaddi- and testing can be delivered in the years old with 4 AJ grandparents. The
tional at-risk carriers being missed by community outside of hospital-based data used in this model was obtained
genetic clinics with high satisfaction from the Genetic Cancer Prediction
rates.1,5,6 Also, the feasibility of through Population Screening random-
Citethisarticleas:ManchandaR,PatelS,AntoniouAC, population-based genetic testing has ized trial (ISRCTN73338115) that
et al. Cost-effectiveness of population based BRCA been accelerated by the availability of comparedoutcomesofpopulation-and
testing with varying Ashkenazi Jewish ancestry. Am J next-generation sequencing and the FH-basedapproachesforgenetictesting
ObstetGynecol2017;""":"""". decreasingcostsofgenetictesting.7 of women with 4 AJ grandparents. The
0002-9378/$36.00 There are significant resource model showed that overall, when the
ª2017ElsevierInc.Allrightsreserved. implications to consider around down-stream costs of treatment were
http://dx.doi.org/10.1016/j.ajog.2017.06.038
population-basedBRCA1/BRCA2testing. taken into account, population-testing
MONTH2017 AmericanJournalofObstetrics&Gynecology 1.e1
Original Research
GYNECOLOGY ajog.org
FIGURE1
Decisionmodelstructure
Theupperpartofthemodelstructurereflectsapopulation-basedapproachtoBRCAtesting;thelowerpartofthemodeldepictsafamilyhistoryebased
approach.Eachdecisionpointinthemodeliscalleda“node”;eachpaththatextendsfromanodeiscalledadecision“branch.”Eachbranchrepresentsa
1.e2 AmericanJournalofObstetrics&Gynecology MONTH2017
Original Research
ajog.org GYNECOLOGY
was in fact cost-saving compared with when FH criteria are used alone AJ grandparents (0.5*AJprevalence
FH-testing.8 The modelling predicted compared with population screening. 0.5*General-popul¼ationprevalence) and foþr
that this could lead to a significant Genetic counselling and genetic testing 1 AJ grandparent (0.25*AJprevalence
reduction in breast-and-ovarian cancer was offered to all women in the popu- 0.75*General-popu¼lationprevalence). Thþe
incidence and increase life expectancy. lationscreeningarmandonlythosewho probabilityofhavingastrongFHthatful-
However, our original analysis applies fulfilledtheFH-basedcriteriaintheFH filledclinicalgenetictestingcriteriain the
only to women with 4 AJ grandparents arm. The criteria for FH-based testing non-Jewishpopulationwasobtainedfrom
and is not applicable directly to every included personal history of ovarian unselected control population data from
woman with AJ ancestry because 25% cancer (any age), first-degree relative the Australian Breast Cancer Family Reg-
UnitedKingdom(UK)9and44%United with ovarian cancer (any age), first- istry. Similarly, probability parameters P6
States (US)10 Jewish marriages are to degree relative with or personal history (havingapositiveFH)andP8(prevalence
non-Jews. These women thus may have ofbreastcancerat<50yearsold,and/or in FH-negative individuals) were adjusted
just1,2,or3AJgrandparents;therefore, first-degree relative with or personal forrelativeBRCAmutationfrequencyinAJ
theprevalenceofBRCA1/BRCA2muta- historyofmalebreastcancer(anyage).2 andgeneralpopulations.Thiswasdonefor
tions is lower in these groups. Never- Parameter estimates for probabilities, all 3 parameters and their confidence in-
theless,thesewomenremainatelevated costs, and utilities were obtained and tervals for all of the different grandparent
BRCA risk compared with the general adapted from the earlier decision scenarios.Therevisedprobabilitytablefor
(non-Jewish) population. Cost- analyticalmodel. modelparametersisgiveninTable1.
effectiveness data for these varying
lower mutation prevalence levels are Probabilities Qualityadjustedlife-yearsand
unavailable. This important gap in All parameters in the decision analytical costs
knowledge was highlighted at a recent modelwerekeptconstantapartfromthe Utilityweightsexpressthepreferenceofan
meetingofexpertsonpopulation-based following 3 parameters: (1) population individualinaspecifichealthstate.These
AJBRCAtesting inHaifa,Israel,in July prevalenceof BRCA (P1), (2) probability weights are then combined with survival
2016.11 We present an updated cost- of having a positive FH (P6), and (3) inlife-yearstogiveameasurementknown
effectiveness analysis of population BRCA prevalence in the FH-negative in- as a quality adjusted life-years (QALY),
BRCA1/BRCA2 testing for women with dividuals (P8). These 3 parameters are which is the preferred value of health
1,2,and3AJgrandparents. influencedbychangeinthenumberofAJ benefitaccordingtotheNationalInstitute
grandparents. An individual with 3 AJ of Health & Care excellence (NICE).13
Methods grandparents would possess 75% AJ ge- This decision-analytical model that used
We previously developed a decision- netic makeup and 25% from the general revisedestimateswasrunforeachofthe4
analytical model (Figure 1) to calculate population. Someone with 2 AJ grand- scenarios:4,3,2,and1AJgrandparents.
cost-effectiveness of screening women parents would have 50% AJ and 50% Thefollowingutilityscoreswereusedfor
with4AJgrandparents.Modelstructure, general population makeup. Therefore, ovarian cancer (OC): 0.81 for early stage
assumptions, analytic features, advan- BRCApopulationprevalenceforanindi- OC, 0.55 for advanced disease, 0.61 for
tages, and limitations have been vidual with 3, 2, or 1 AJ grandparent is recurrentdisease,0.83forOCremission,
described earlier.8 This model was adjusted for the relative frequency of and 0.16 for end stage OC.14 The
adaptedto model outcomes for women BRCA mutations in the AJ and general following utility scores, which were ob-
with differing AJ ancestry.8 Separate populations. BRCA prevalence estimates tained from NICE guidelines,15-18 were
analyses were performed for both UK for AJ was obtained from the Genetic usedforbreastcancer(BC):0.71forearly/
and US populations. Lifetime costs and Cancer Prediction through Population locally advanced BC, 0.65 for advanced
effects of genetically screening 30-year- Screeningstudy(0.0245[0.0131,0.0416])2 disease, 0.45 for recurrence disease, 0.81
old AJ women for BRCA1/BRCA2 AJ and for the general population (0.0067 for BC remission, and 0.16 for end stage
foundermutationswerecomparedwith [0.00590. 0.0077]) from recent pub- BC. A breakdown of both UK and US
currentpracticeofscreeningwiththeuse lishedestimates.12BRCAprevalencewith costsat2014and2015pricesaregivenin
of FH-based clinical criteria. Fifty-six 3 AJ grandparents (0.75*AJprevalence) Table 2. The analysis covers a health sys-
percent of BRCA carriers are missed (0.25*General-popu¼lationprevalence); for þ2 temorpayerperspective.
=
mutuallyexclusivecourseoroutcome.Eachdecisionisgivenaprobability(probabilities“p1-14”thatwereusedinthemodelareexplainedinTable1),
whichishighlightedinawhiteboxalongthedecisionbranch.Valuesforeachoutcomewerecalculated.Cancerincidencewasestimatedbythesumming
oftheprobabilitiesofpathwaysthatendedinovarianorbreastcancer.Finaloutcomes(blueboxes)ofeachpathincludedevelopmentofbreastcancer,
ovariancancer,andnobreast/ovariancancer.
BC,breastcancer;NoBC,nobreastcancerdeveloped;NoOC,noovariancancerdeveloped;OC,ovariancancer;RRM,risk-reducingmastectomy;RRSO,risk-reducingsalpingo-oophorectomy.
FromManchandaR,LegoodR,BurnellM,etal.Cost-effectivenessofpopulationscreeningforBRCAmutationsinAshkenaziJewishwomencomparedwithfamilyhistoryebasedtesting.JNatlCancerInst
2015;107:380.WithpermissionfromOxfordUniversityPress.
Manchandaetal.Cost-effectivenessofpopulationBRCAtestingwithvaryingJewishancestry.AmJObstetGynecol2017.
MONTH2017 AmericanJournalofObstetrics&Gynecology 1.e3
Original Research
GYNECOLOGY ajog.org
TABLE1
Probabilitiesthatwereusedinthemodel
95%Confidence
Probability Value interval[range] Description Source
P1(4AJGP) 0.0245 0.0131e0.0416 PopulationprevalenceofBRCAFM GeneticCancerPredictionthrough
PopulationScreeningstudy,
Manchanda2,8
P1(3AJGP) 0.0201 0.0113e0.0331 BRCAprevalencewith:3AJ Manchanda,2,8Jervis201512
grandparents (0.75*AJprevalence)
P1(2AJGP) 0.0156 0.0095e0.0247 (0.25*general¼-populationprevalence);þ
P1(1AJGP) 0.011 0.0077e0.0162 2AJgrandparents
(0.5*AJprevalence) ¼(0.5*general-
populationprevalencþe);1AJ
grandparent (0.25*AJprevalence)
(0.75*genera¼l-populationprevalenceþ)
P2 0.52 0.39e0.67 Probabilitythatcarrierwillundergo Evans46
RRM
P3 0.96 [0.8e0.96] Reductioninriskofovariancancer Finch,47Rebbeck33
fromRRSO
P4 0.2987 0.2485e0.3539 ProbabilitythatcarrierwithoutRRSO Chen48
willexperienceovariancancer
P5 0.0185 0.0005e0.0989 Probabilitythatanoncarrierwill CancerResearchUK
experienceovariancancer
0.0128 0.0126e0.0130 Probabilitythatanoncarrierwill Surveillance,Epidemiology,andEnd
experienceovariancancer:US ResultsProgram21
estimate
P6(4AJGP) 0.1238 0.1043e0.1454 Probabilityofhavingapositive GeneticCancerPredictionthrough
FH PopulationScreeningstudy2,8
P6(3AJGP) 0.095 0.079e0.114 Probabilitywith:3AJgrandparents
(0.75*AJprobability) (0.25*general¼-
P6(2AJGP) 0.0668 0.055e0.082 populationprobabilityþ);2AJ
P6(1AJGP) 0.0383 0.0296e0.0498 grandparents (0.5*AJprobability)
(0.5*general-¼populationprobability);þ
1AJgrandparent
(0.25*AJprobability)¼ (0.75*general-
populationprobabilityþ)
P7 0.0938 0.0637e0.1763 BRCAprevalenceinFH-positive GeneticCancerPredictionthrough
individuals PopulationScreeningstudy2,8
P8(4AJGP) 0.0203 0.0114e0.0332 BRCAprevalenceinFH-negative GeneticCancerPredictionthrough
individuals PopulationScreeningstudy2,8
P8(3AJGP) 0.0166 0.0098e0.0266 Probabilitywith:3AJgrandparents
(0.75*AJprobability) (0.25*general¼-
P8(2AJGP) 0.0129 0.0082e0.0199 populationprobabilityþ);2AJ
P8(1AJGP) 0.009 0.006e0.013 grandparents (0.5*AJprobability)
(0.5*general-¼populationprobability);þ
1AJgrandparent
(0.25*AJprobability)¼ (0.75*general-
populationprobabilityþ)
P9 0.91 0.62e0.98 Reductioninbreastcancerriskfrom Rebbeck49
RRMwithoutRRSO
P10 0.53 0.44e0.62 ProbabilitythatcarrierwithoutRRM Chen48
willexperiencebreastcancer
Manchandaetal.Cost-effectivenessofpopulationBRCAtestingwithvaryingJewishancestry.AmJObstetGynecol2017. (continued)
1.e4 AmericanJournalofObstetrics&Gynecology MONTH2017
Original Research
ajog.org GYNECOLOGY
TABLE1
Probabilitiesthatwereusedinthemodel(continued)
95%Confidence
Probability Value interval[range] Description Source
P11 0.13 [0.11e0.14] Probabilitythatanoncarrierwill CancerResearchUK,50Officefor
experiencebreastcancerwith NationalStatistics51
screening:UKestimate
0.124 0.1236e0.1249 Probabilitythatanoncarrierwill Surveillance,Epidemiology,andEnd
experiencebreastcancerwith ResultsProgram22
screening:USestimate
P12 0.55 0.30e0.75 Probabilitythatcarrierwillfollowup Manchanda52
withRRSO
P13 0.49 0.37e0.65 Reductioninriskofbreastcancer Rebbeck33
fromRRSOalone
P14 0.95 0.78e0.99 Reductioninriskofbreastcancer Rebbeck49
fromRRMwithRRSO
AJ,AshkenaziJewish;FH,familyhistory;FM,foundermutations;GP,grandparent;RRM,risk-reducingmastectomy;RRSO,risk-reducingsalpingo-oophorectomy;UK,UnitedKingdom;US,United
StatesofAmerica.
Explanation:TheprobabilitiesP1,P6,andP8havebeenadaptedfordifferentlevelsofAJancestry:4,3,2,and1grandparents.Theothermodelprobabilitiesremainthesame,aspreviously
published.8
P1:TheprobabilityofcarryingaBRCAFMintheAJpopulation(p1 .0245)istakenfromtheGeneticCancerPredictionthroughPopulationScreeningstudy.Thiswastheprobabilitywith4AJ
grandparents.TheprobabilityofhavingaBRCAmutationinthenon-J¼ewishpopulation(0.0067(0.00590.,0.0077)wastakenfromup-to-dateestimatesfromJerivs2015.12BRCAprevalence:3AJ
grandparents (0.75*AJprevalence) (0.25*general-populationprevalence);2AJgrandparents (0.5*AJprevalence) (0.5*general-populationprevalence);1AJgrandparent (0.25*AJprevalence)
(0.75*general¼-populationprevalence).þ ¼ þ ¼ þ
P2:TheprobabilitythatBRCA1/2carrierwillundergoRRMwastakenfromananalysisofUKBRCA1/2carriers.46Acompositeuptakerate(p2 .52)forBRCA1(60%RRMrate)andBRCA2(43%RRM
rate)carriers,weightedfortherelativeprevalenceofBRCA1andBRCA2FMthatwerefoundintheLondonAJpopulation,wascomputed¼.46
P3:ThereductioninovariancancerriskthatwasobtainedfromRRSO(p3 .96)wastakenfrompreviousstudiesthatreporteda4%residualriskofprimaryperitonealcancerafterRRSO.47
¼
P4:TheGeneticCancerPredictionthroughPopulationScreeningstudymodel8usesovariancancerpenetranceestimates(40%forBRCA1,18%forBRCA2)fromametaanalysisthatwascorrected
forascertainment.48Tosimplifytheanalysis,theGeneticCancerPredictionthroughPopulationScreeningstudymodelusedacompositeriskforBRCA1andBRCA2carriersweightedfortherelative
prevalenceofBRCA1andBRCA2FM.TheoverallriskofovariancancerinBRCAcarrierswascalculatedas([0.0132*0.4]/2.45 [0.0113*0.18]/2.45).8
þ
P5:Theriskofovariancancerinalow-riskpopulation(p5 .0185)wasobtainedfromCancerResearchUK53forBritishwomen;SEERdata21wasusedforUSwomen.
¼
P6:TheprobabilityofhavingastrongFHofcancerthatfulfilledthecurrentclinicalcriteria(FH-positive)forwomenwith4AJgrandparentswasobtainedfromtheGeneticCancerPredictionthrough
PopulationScreeningstudy.2,8TheprobabilityofhavingapositiveFHthatfulfilledthenon-AJgenetictestingcriteriawasobtainedfrompreviouslyunpublishedunselectedcontrolpopulationdatafrom
theAustralianBreastCancerFamilyRegistry.Theprobabilityfor3,2,and1grandparentwasadjustedfortherelativeprevalenceinJewishandgeneralpopulations.Theprobability:3AJ
grandparents (0.75*AJprobability) (0.25*General-populationprobability);2 AJ grandparents (0.5*AJprobability) (0.5*General-populationprobability); 1 AJ grandparent (0.25*AJprobability)
(0.75*Genera¼l-populationprobability) þ ¼ þ ¼ þ
P7:TheBRCAprevalenceinFH-positiveindividualswasalsoobtainedfromtheGeneticCancerPredictionthroughPopulationScreeningstudy.2,8
P8:TheBRCAprevalenceinFH-negativeindividualsforwomenwith4AJgrandparentswasobtainedfromtheGeneticCancerPredictionthroughPopulationScreeningstudy.2,8Theprobabilityfor3,
2,and1grandparentwasadjustedfortherelativeprevalenceinJewishandgeneralpopulations.Theprobability:3AJgrandparents (0.75*AJprobability) (0.25*general-populationprobability);2AJ
grandparents (0.5*AJprobability) (0.5*general-populationprobability);1AJgrandparent (0.25*AJprobability) (0.75*general-popula¼tionprobability). þ
¼ þ ¼ þ
P9:ReductioninbreastcancerriskfromRRMinBRCAcarrierswhodidnotundergoRRSOwastakenfromthePreventionandObservationofSurgicalEndpointsstudydata.49
P10:ThebreastcancerpenetranceforBRCAcarriers(57%forBRCA1and49%forBRCA2)wastakenfromametaanalysisthatwascorrectedforascertainment.48Tosimplifytheanalysis,the
GeneticCancerPredictionthroughPopulationScreeningstudymodelusedacompositeriskforBRCA1andBRCA2carriersthatwasweightedfortherelativeprevalenceofBRCA1andBRCA2FM.The
overallriskofbreastcancerinBRCAcarrierswascalculatedas([0.0132*0.57]/2.45 [0.0113*0.49]/2.45).2,8
þ
P11:Theriskofbreastcancerinalow-riskpopulationwastakenfromCancerResearchUKandUKOfficeforNationalStatisticsdata50,51andfromSEERdata22forUSwomen
P12:IntheGeneticCancerPredictionthroughPopulationScreeningstudymodel,8wehaveusedtheRRSOratesthatwerereportedinhigh-riskwomenfromLondon,whichreflectstheviewsof
carriersfromaLondonpopulationandiswithintherangereportedintheliterature.52
P13:ThereductioninbreastcancerriskinpremenopausalwomenwhounderwentRRSOwastakenfromametaanalysisbyRebbecketal.33
P14:ReductioninbreastcancerriskfromRRMinBRCAcarrierswhounderwentRRSOwastakenfromthePreventionandObservationofSurgicalEndpointsstudydata.49
Manchandaetal.Cost-effectivenessofpopulationBRCAtestingwithvaryingJewishancestry.AmJObstetGynecol2017.
Life-years ages for sporadic BC and OC in AJ Analysis
Alifetimehorizon(extendingtotheage women were 57 and 62 years and 63 To calculate the probability of being in
of 83 years) that captured lifetime risks and63yearsintheUK/USpopulations, each branch, the path probabilities of
and consequences was used to model respectively.19-22Five-yearsurvivalrates each branch were multiplied together.
the analysis. Mean ages for BRCA1/ inthegeneralUKpopulationwereused Total costs, life-years, and QALYs were
BRCA2-related BC and OC in AJ in the absence of AJ survival data.23 determined through weighting of the
women of 43.5 years and 54.9 years Costs, QALYs, and life-years were dis- valuesofeachbranchbytheprobability
were used in the analysis. The mean counted at 3.5%.24 ofbeingineachbranch.Toestablishthe
MONTH2017 AmericanJournalofObstetrics&Gynecology 1.e5
Original Research
GYNECOLOGY ajog.org
TABLE2
Costsusedinthemodel(2014prices)
Cost Cost
Item UniteKingdom(£) UnitedStates($) Source
CostofBRCAFoundermutationtesting 50 300 GCaPPS2
Costofgeneticcounselling 43 41 GCaPPS,2PSSRUunitcostsofHealthand
SocialCare,54Schwartzetal201455
Costofrisk-reducingsalpingo- 3,411 8,144 NHSreferencecosts,56BNF57
oophorectomy
(andhormonereplacementtherapyand
osteoporosisprevention)
Costofovariancancerdiagnosisand 14,123 127,995 NHSreferencecosts,56NICEguideline58
treatment
Yearlycostofovariancancertreatmentat 10,050 14,071 NHSreferencecosts,56NICEguideline,58
year1-2 CRUK,59Grannetal201160
Yearlycostofovariancancertreatmentat 14,387 14,071 NHSreferencecosts,56NICEguideline,58
year3-5 CRUK,59Grannetal201160
Terminalcarecostswithovariancancer 15,450 89,424 NationalAuditOffice,61IncisiveHealth
reportforCRUK,62Grannetal201160
Costofrisk-reducingmastectomy 3,901 12,596 NHSreferencecost,56weightedfor21%
complicationrate,4Grannetal201160
Costofbreastscreening 347 1,534 Robertson2011,63NHSreferencecosts,56
CDCguideline64
CostofbreastscreeningBRCAcarriers 4,582 33,530 NHSreferencecosts,56NICEguideline,65
CDCguideline,64Grannetal201160
Costofbreastcancerdiagnosisand 15,527 82,030 NHSreferencecosts,56NICEguideline,16
treatment NICEguideline,15Grannetal201160
Manchandaetal.Cost-effectivenessofpopulationBRCAtestingwithvaryingJewishancestry.AmJObstetGynecol2017. (continued)
cost-effectiveness of population-based available, or they were varied by /- and4AJgrandparentsinUK/USwomen.
þ
screening against FH-based testing, the 10%. Probabilities were assigned a Population testing in women with just
incremental cost-effectiveness ratio beta distribution, costs a gamma distri- 1 AJ grandparent is also cost-effective,
(ICER) was calculated by dividing dif- bution and utilities a log normal distri- with an ICER that equals £863 per
ferencesincostbythedifferencesinef- bution in accordance with published QALYand15dayslifegained.Thistoois
fect. The £20,000e30,000 per QALY literature.29 well below the £20,000 per QALY
cost-effectiveness willingness-to-pay threshold,althoughnotcost-saving.
(WTP)thresholdthathadbeenusedby Results The PSAs for the UK and US
NICE25 was used to compare the cost- BaselineICERresultsforthe4different (Figures 2 and 3) show that, for pop-
effectiveness of population-based grandparent scenarios are given in ulationswith4,3,2,or1AJgrandparent,
screening in comparison with FH- Table 3 alongside discounted and 95% of simulations are cost-effective
!
based testing in the UK. A WTP undiscounted life-years, QALYs, and for population screening at the £20,000
threshold of $100,000 per QALY was lifetime costs for each scenario for perQALYNICEWTPand$100,000per
usedfortheUSanalysis.26,27 womenbothintheUKandUS.Baseline QALYUSWTPthresholds.Thissuggests
To accountforuncertainty,all model resultssuggestthatpopulationtestingin that, compared with current clinical
parameters were varied simultaneously UKwomenwhohad 2AJgrandparents policy of FH-based clinical testing,
!
across their distributions using 10,000 and 1 grandparent for US women re- population testing in 4, 3, 2, and 1 AJ
!
simulations in a probabilistic sensitivity mains cost-saving and highly effective grandparentsishighlycost-effective.
analysis (PSA) in accordance with the compared with traditional testing that
recommendation of NICE methods uses FH-based clinical criteria. This Comment
guidance.28Costswerevariedby 30%, corresponds to a life expectancy gain of Given that a large proportion of mar-
confidence intervals were used t%o vary 15 and 12, 22 and 17, 28 and 22, and riages in the Jewish population are be-
probabilities and utility scores, if 33 and 26 days, respectively, for 1, 2, 3, tweenJewsandnon-Jews,itisimportant
1.e6 AmericanJournalofObstetrics&Gynecology MONTH2017
Original Research
ajog.org GYNECOLOGY
TABLE2
Costsusedinthemodel(2014prices)(continued)
Cost Cost
Item UniteKingdom(£) UnitedStates($) Source
Yearlycostofbreastcancertreatment 2,008 7,738 NHSreferencecosts,56Robertson2011,63
BNF,57NICEguideline,16NICEguideline,15
implementingNICEguidance,16Grannetal
201160
YearlycostofBRCAassociatedbreast 1,917 7,738 NHSreferencecosts,56Robertson2011,63
cancer BNF,57NICEguideline,16NICEguideline,15
implementingNICEguidance,16Grannetal
201160
Terminalcarecostswithbreastcancer 15,450 65,403 NationalAuditoffice,61Grannetal201160
BNF,BritishNationalFormulary;CDC,CentersforDiseaseControlandPrevention;CRUK,CancerResearchUK;GCaPPS,GeneticsCancerPredictionthroughPopulationScreeningstudy;NHS,
NationalHealthService;NICE,NationalInstituteforHealthandClinicalExcellence;PSSRU,PersonalSocialServicesResearchUnit.
Explanation:Costofrisk-reducingsalpingo-oophorectomy:Itisassumedthathormonereplacementtherapyisprovidedtowomenfromtheagethattheyhaverisk-reducingsalpingo-oophorectomy
untiltheageofmenopause(51years).57An80%complianceratewasassumedwithhormonereplacementtherapy;thesecostsareaddedtocostsforsurgery.FortheUnitedKingdom,risk-reducing
salpingo-oophorectomycostsareforanuppergenitaltractlaparoscopic/endoscopicintermediateprocedure.56Tomonitorbonehealth,thecostof3DEXAscansandcalciumandvitamin-D3arealso
included.UnitedStatesprophylacticsalpingo-oophorectomycostsaretakenfromGrannetal60andinflatedwiththeuseofthemedicalcomponentoftheUnitedStatesconsumerpriceindex.
Ovariancancercosts:CostswerebasedonovariancancerguidelinespublishedbyNICE.58Diagnosiscostsincludedultrasoundscan,pelvicexamination,computedtomographyscan,CA125test,
percutaneousbiopsy,andperitonealcytology.Costsoftreatmentincludedreferencecostforloweranduppergenitaltractcomplexmajorprocedureand6cyclesofcarboplatinandpaclitaxel
chemotherapyadministration.Survivorswereassumedtohave3consultantvisits,4CA125testsandacomputedtomographyscaneachyearforthefirst2yearsaftersurgery.Inyears3e5after
surgery,survivorswereassumedtohave2consultantvisitsand2CA125tests.TerminalcancercostsweretakenfromareportsubmittedtotheNationalAuditOffice,UK.61Recurrentovariancancer
costsweretakenfromareportthatwascommissionedbyCRUK.62FortheUnitedStates,thecostofovariancancerdiagnosis,treatment,recurrence,andterminalovariancancercostsweretaken
fromGrannetal2011andinflatedwiththeuseofthemedicalcomponentoftheUnitedStatesconsumerpriceindex.
Breastcancercosts:CostswerebasedonNICEguidelinesonearly/locallyadvancedbreastcancer16andadvancedbreastcancerinUK,15theBNF,57andtheDepartmentofHealthNHSreference
costs.56
Costofbreastscreeningingeneralpopulation:Women50e70yearsoldareofferedmammographyevery3yearsinaccordancewiththeUnitedKingdomNHSbreastcancerscreeningprogram.66
Costofdiagnosisbasedonclinicalexamination,ultrasound,mammographyandbiopsy
Costofbreastscreeningincarriers:Womenareofferedanannualmagneticresonanceimageat30e49yearsoldandanannualmammogramat40e69yearsoldinaccordancewithNICEfamilial
breastcancerguidelines.65
Costofbreastcancertreatment:Innoncarriers,10%ofbreastcancerisnoninvasiveductalcarcinomainsituand90%invasive;95%ofinvasivebreastcancerisearlyandlocallyadvanced,with5%
beingadvancedbreastcancer.16InBRCA1/2carriers,20%arenoninvasiveductalcarcinomainsituand80%invasive.4,67
Yearlycostofbreastcancertreatment:Costsincludesentinellymphnodebiopsyandaxillarylymphnodedissection,asrecommendedinNICEguideline.16Breast-conservingsurgeryandmastectomy
costswithreconstructionareincludedintreatmentcosts.56Radiotherapycoststhatareincludedareofferedforearlyinvasive/locallyadvancedcancer;chemotherapyisofferedalongside
radiotherapyforadvancedcancers.ChemotherapycostsaretakenfromNICEguidelinesbasedonfirst-andsecond-line15,16polychemotherapy.68Coststakeintoaccountthedifferenceinstageat
presentation,with20%ofcancersbeingnoninvasive.CostsarealsotakenintoaccountforthetestingofcancersthatareestrogenreceptorpositiveandHER2positiveinthegeneralandBRCAcarrier
population:70%generalpopulationinvasivebreastcancerswereestrogenreceptorpositive;15%ofearlyinvasivebreastcancersand25%ofadvancedbreastcancersareHER2positive.15,16ER-
positivecancersreceivetamoxifenat20mg/day,ifthewomanispremenopausal,oranastrazole1mg/day,ifthewomanispostmenopausalfor5years;costsofbotharefromtheBNF.57Tooffsetthe
riskofthedevelopmentofbonemetastases,65%ofindividualsareofferedbisphosphonates.PerNICEguidelines,itisassumed50%ofthosewillreceiveintravenouszolendronicacidorpamidronate,
andtheother50%willreceiveoralclodronateandibandronicacid.AsperNICEguidelines,HER2-positivepatientsaregiventrastuzumabat3-weekintervalsforayearoruntildiseaserecurrence.16
RecurrenceratesareincludedforbreastcancerasobtainedthroughtheUnitedStatesNationalSurgicalAdjuvantBreastandBowelProject.69,70Follow-upcostsforbreastcancerinclude,6monthly
consultationsandannualmammogramswithmagneticresonancescansforstage4cancers.Coststakeintoaccounta35%progressionratefromearlyandlocallyadvancedtoadvanceddisease16
and66%relapserateinadvanceddisease.71TerminalcancercarecostswereobtainedfromareportpublishedbytheNationalAuditOffice,UK.61
UnitedStatesbreastcancercosts:ThecostsofbreastcancerdiagnosisandtreatmentwastakenfromGrannetal60andwereinflatedwiththeuseofthemedicalcomponentoftheUnitedStates
consumerpriceindex.
Costofbreastscreeninginnoncarriers:Costsassumethatmammogramsareconductedevery2yearsfrom50yearsold(CDCguidelines).64
Costofbreastscreeningincarriers:CostsassumethatwomenintheUnitedStatesareofferedayearlymammographyandmagneticresonanceimagefrom30yearsold,thenfrom50yearsold
onwardswomenareofferedonlyanannualmammography.64
Terminalbreastcancercosts:CostsaretakenfromGrannetal60andinflatedwiththeuseofthemedicalcomponentoftheUnitedStatesconsumerpriceindex.
Manchandaetal.Cost-effectivenessofpopulationBRCAtestingwithvaryingJewishancestry.AmJObstetGynecol2017.
to explore the cost-effectiveness of pop- grandparentsinadditiontothosewith4 cost-savingforthosewith4,3,and2and
ulationtestinginwomenwithdifferingAJ AJgrandparents.That 95%PSAsimu- 4,3,2,and1AJgrandparentsintheUK/
!
ancestryandBRCAprevalenceratesasa lationsremaincost-effective,despitesig- US populations, respectively. There are
consequence. Our findings confirm the nificant variability in model parameters, notmanyinterventionsthatcansaveboth
cost-effectiveness of population-based issignificantandreassuringforallUK/US livesandmoney.30AJscreeningmightbe
BRCA1/BRCA2 testing (compared with women with differing AJ ancestry. This the most cost-effective and maybe the
testing based on clinical FH criteria) in approachhasthepotentialtoreducethe only cost-saving program among those
unselectedUK&USAJwomenwhoare number of OCs and BCs in the AJ programsthatuseBRCAtesting.31These
30 years old, who have 1, 2, or 3 AJ population. Such a program would be data have important implications for
!
MONTH2017 AmericanJournalofObstetrics&Gynecology 1.e7
Original Research
GYNECOLOGY ajog.org
TABLE3
Modeloutcomesforcosts,life-years,andqualityadjustedlife-yearsfortheUnitedKingdomandtheUnitedStates
UnitedKingdomresults UnitedStatesresults
AshkenaziJewish Qualityadjusted Qualityadjusted
grandparents Screeningarms Cost(£) Life-yearsa life-years Cost($) Life-yearsa life-years
4 Averagepopulation 1861 52.26 23.15 7,254 52.59 23.24
screening
Averagefamilyhistory 1955 52.17 23.12 7,775 52.51 23.22
screening
Incremental(difference) e94 0.090 0.032 e522 0.072 0.027
Incrementalcost- e2960 e19,587
effectivenessratioper
qualityadjustedlife-
years
3 Averagepopulation 1813 52.27 23.16 7,038 52.60 23.25
screening
Averagefamilyhistory 1875 52.19 23.13 7,413 52.53 23.22
screening
Incremental(difference) e62 0.075 0.027 e375 0.061 0.022
Incrementalcost- e2327 e16,788
effectivenessratioper
qualityadjustedlife-
years
2 Averagepopulation 1766 52.28 23.16 6,822 52.61 23.25
screening
Averagefamilyhistory 1792 52.22 23.14 7,033 52.56 23.23
screening
Incremental(difference) e26 0.060 0.021 e212 0.048 0.018
Incrementalcost- e1254 e12,013
effectivenessratioper
qualityadjustedlife-
years
1 Averagepopulation 1718 52.29 23.17 6,606 52.62 23.26
screening
Averagefamilyhistory 1705 52.25 23.15 6,637 52.58 23.24
screening
Incremental(difference) 13 0.042 0.015 e31 0.034 0.012
Incrementalcost- 863 e2,542
effectivenessratioper
qualityadjustedlife-
years
aUndiscountedoutcomesshownforlife-years;costsandqualityadjustedlife-yearsoutcomesarediscounted.
Manchandaetal.Cost-effectivenessofpopulationBRCAtestingwithvaryingJewishancestry.AmJObstetGynecol2017.
populationhealthandcancerprevention numbers are averaged across the popu- appropriately long enough to highlight
and are of value to healthcare providers lation. In health economic terms, these any important variations in costs and
andcarecommissioners. valuesaresignificant;foranindividualin outcomes. The sensitivity and scenario
The number of days of life gained whomcancerisprevented,thisnumber analyses that were undertaken add
rangefrom15and33daysand12and26 ismanyfoldshigher.Ourmodellingin- strength to the study. Although risk-
days, respectively, in the UK and US. corporates the costs of both genetic reducing salpingo-oophorectomy
Although,thesefiguresappearsmall,itis testing and genetic counselling. The (RRSO) reduces the risk of BC in pre-
important to highlight that these time horizon in our modelling is menopausal women, the benefit of
1.e8 AmericanJournalofObstetrics&Gynecology MONTH2017
Original Research
ajog.org GYNECOLOGY
perQALY,£ 1759/$ 14,032perQALY,
FIGURE2 & &
£ 301/$ 7366 per QALY and £2793/
UnitedKingdomprobabilisticsensitivityanalysisforvaryingAshkenazi & &
$7110perQALYforwomenwith4,3,2,
Jewishgrandparentancestry
and 1 AJ grandparents. Thus,
population-based testing in AJ women
of differing ancestry remains cost-
effective in the UK and US, even with
low risk-reducing mastectomy or low
RRSOratestoo.Giventhewidevariation
in genetic testing costs in the US health
system, we also explored thresholds for
cost-effectivenessforpopulationtesting.
Wefindthatpopulationtestingremains
cost-saving for up to 2 AJ grandparents
(cost-effective for 1 AJ grandparent) if
theBRCAfoundermutationtestingcosts
$526 per test. Additionally the program
remains cost-effective (at the $100,000
per QALY WTP threshold) for all 4 AJ
grandparents even if the cost of a test
risesto$1618,$2417,$3185,and$3934
for 1, 2, 3, and 4 AJ grandparents,
respectively.
All surgical interventions have an
associated complication rate. The
Cost-effectiveness acceptability curves are generated when the model undergoes probabilistic
complication rates reported for RRSO
sensitivityanalysis.Intheprobabilisticsensitivityanalysis,allmodelparametersarevariedsimul-
are approximately 4%,38 although that
taneouslyacrosstheirdistributions.Eachofthe4scenarios(4,3,2,1grandparent)weresimulated
10,000times;theresultsshownaretheproportionofthesesimulationsthatwouldbecost-effective for risk-reducing mastectomy is much
atdifferentwillingness-to-pay thresholds.X-axis:Incrementalcost-effectivenessratiointermsof higher, with reports that range from
costperqualityadjustedlife-years.Y-axis:Proportionofsimulations.Thesolidredlinemarksthe 30e64%.39 Another limitation of the
proportionofsimulationsthatwerefoundtobecost-effectiveatthe£20,000thresholdusedbythe analysis is lack of adjustment for any
National Institute for Health and Clinical Excellence. At 95%, simulations are cost-effective for potential negative impact on quality of
!
varyinglevelsofJewishancestryinthisanalysis. life after RRSO. Although worse sexual
£,Britishpoundsterling;AJ,AshkenaziJewish;QALY,qualityadjustedlife-years;UK,UnitedKingdom. functioning and vasomotor symptoms
Manchandaetal.Cost-effectivenessofpopulationBRCAtestingwithvaryingJewishancestry.AmJObstetGynecol2017. have been reported after RRSO, there
was no difference in generic quality of
life.40-42 These issues must be high-
premenopausal oophorectomy on premenopausal oophorectomy. Our lighted clearly when women are coun-
reductioninBCriskrecentlyhasbecome model incorporates risk-reducing mas- selled about these procedures and be
anissueofongoingdebate.SomeDutch tectomyratesthathavebeenseeninthe incorporated into the informed
investigators32 have questioned this UK. However, these rates may be lower decision-makingprocess.Itisreassuring
benefit.However,theperiodoffollowup in Israeli BRCA1/BRCA2 carriers.37 that most women report high satisfac-
in their analysis is short. A number of Hence, we explored a scenario analysis tion rates with surgical prevention,
other investigators have found benefit at the much lower risk-reducing mas- with satisfaction rates that vary from
and disagree with them.33-36 Neverthe- tectomyrateof13%thatwasreportedin 83% for mastectomy39 to 97% for
less,ifweassumenobenefitofreduction Israeli women. The discounted ICERs oophorectomy.40
inBCriskfrompremenopausaloopho- fora13%risk-reducingmastectomyrate Our results support the move for
rectomy,thentheICERperQALYfor1, are £ 1958/$ 11059 per QALY, changingtheparadigmfromFH-basedto
& &
2,3,and4grandparentsis£1971/$2843 £ 1177/$ 7548 per QALY, £196/ population-based BRCA1/BRCA2 testing
per QALY, £ 497/$ 8198 per QALY, $&1255 p&er QALY and £3056/$12,103 across the AJ population. This fulfils the
& & &
£ 1715/$ 13595 per QALY, £ 2420/ perQALYforUK/USwomenwith4,3,2, necessary principles for population
& & &
$ 16697 per QALY in UK/US women, and 1 AJ grandparents, respectively. In screeningforgeneticsusceptibilityofdis-
re&spectively. This suggests that a addition, a scenario with a much lower ease.43Populationtestingofferstheability
populationscreeningapproachwouldbe RRSOuptakeat20%wasexplored.The tomaximizetheopportunityfor preven-
cost-effective, even if there were no discounted ICERs for this scenario in tion in unaffected individuals and to
benefit on reduction in BC risk from UK/US women are £ 2589/$ 17,786 facilitate targeted precision medicine
& &
MONTH2017 AmericanJournalofObstetrics&Gynecology 1.e9
Original Research
GYNECOLOGY ajog.org
3.MetcalfeKA,PollA,RoyerR,etal.Screening
FIGURE3
forfoundermutationsinBRCA1andBRCA2in
UnitedStatesprobabilisticsensitivityanalysisforvaryingAshkenaziJewish unselected Jewish women. J Clin Oncol
grandparentancestry 2010;28:387-91.
4.Nelson HD, Fu R, Goddard K, et al. Risk
assessment, genetic counseling, and genetic
testingfor BRCA-relatedcancer:systematicre-
view to update the us preventive services task
force recommendation. Rockville (MD): Agency
forHealthcareResearchandQuality,USDepart-
mentofHealthandHumanServices;2013.
5.LiebermanS,TomerA,Ben-ChetritA,etal.
Population screening for BRCA1/BRCA2
foundermutationsinAshkenaziJews:proactive
recruitment compared with self-referral. Genet
Med2016.Epubaheadofprint.Doi:10:1038:
gim2016.
6.Manchanda R, Burnell M, Loggenberg K,
et al. Cluster-randomised non-inferiority trial
comparingDVD-assistedandtraditionalgenetic
counsellinginsystematicpopulationtestingfor
BRCA1/2 mutations. J Med Genet 2016;53:
472-80.
7.Shendure J, Ji H. Next-generation DNA
sequencing.NatBiotechnol2008;26:1135-45.
8.Manchanda R, Legood R, Burnell M, et al.
Cost-effectiveness of population screening for
BRCA mutations in Ashkenazi Jewish women
compared with family history-based testing.
JNatlCancerInst2015;107:380.
United States cost-effectiveness acceptability curves are generated when the model undergoes 9.GrahamD,SchmoolM,WatermanS.Jewsin
probabilistic sensitivity analysis. In the probabilistic sensitivity analysis, all model parameters are Britain:asnapshotfromthe2001census,vol.1.
variedsimultaneouslyacrosstheirdistributions.Eachofthe4scenarios(4,3,2,1grandparent)were London: Institute for Jewish Policy Research;
simulated10,000times;theresultsshownaretheproportionofthesesimulationsthatwouldbe 2007.
cost-effectiveatdifferentwillingness-to-paythresholds.X-axis:Incrementalcost-effectivenessratio 10.PewResearchCenter.APortraitofJewish
Americans: Findings from a Pew Research
intermsofcostperqualityadjustedlife-years.($s)/QALY;Y-axis:Proportionofsimulations.Thesolid
Center Survey of US Jews. Washington (DC):
redlinemarkstheproportionofsimulationsthatwerefoundtobecost-effectiveatthe$100,000
PewResearchCenter;2013.
threshold. At 95%, simulations are cost-effective for varying levels of Jewish ancestry in this
! 11.FounderPopulationsandtheircontributionto
analysis.
ourunderstandingofbiologyandhistory:lessons
$,dollars;AJ,AshkenaziJewish;QALY,qualityadjustedlife-years;USA,UnitedStatesofAmerica. fromtheJewishgenome.ThinktankonBRCA
Manchandaetal.Cost-effectivenessofpopulationBRCAtestingwithvaryingJewishancestry.AmJObstetGynecol2017. screening.Haifa,Israel10/7/2016.Availableat:
https://www.foundergenomics.com/. Accessed
October12,2016.
12.JervisS, SongH, Lee A,et al. A riskpre-
approachesinthosewhomayexperience disseminationofinformationandknowl-
diction algorithm for ovarian cancer incorpo-
thedevelopmentofcancer.Thisapproach edge, working in close partnership with
rating BRCA1, BRCA2, common alleles and
has been advocated by us and communitystakeholdersandhealthpro- other familial effects. J Med Genet 2015;52:
others.1,8,44,45 It is also important to fessionals. Moreover, implementation is- 465-75.
highlightthatthosewithfewergrandpar- sues related to health system delivery, 13.National Institute for Health and Clinical
ents, but a significant FH of cancer (ful- referral and management pathways, lo- Excellence.Guidetothemethodsoftechnology
appraisal.London:NationalInstituteforHealth
filling non-Jewish general population gistics,andcontrol,whichcanvaryacross
andClinicalExcellence(NICE);2008.
testing criteria) particularly in non-AJ different models of care in different 14.Havrilesky LJ, Broadwater G, Davis DM,
relatives, should seek genetic advice and countries,remaintobeironedout. n etal.Determinationofqualityoflife-relatedutili-
notbereassuredfalsely.Itisimportantto tiesforhealthstatesrelevanttoovariancancer
diagnosis and treatment. Gynecol Oncol
rule out the presence of a non-founder
References 2009;113:216-20.
mutation in this situation through a full
1.Gabai-KaparaE,LahadA,KaufmanB,etal. 15.National Institute for Health and Clinical
BRCA1/BRCA2screenanalysis.Addition- Population-based screening for breast and Excellence.ClinicalGuideline(CG81):advanced
ally,ourfindingscannotbeextrapolatedor ovariancancerriskduetoBRCA1andBRCA2. breastcancer:diagnosisandtreatment.Cardiff,
generalized to BRCA1/BRCA2 testing in ProcNatlAcadSciUSA2014;111:14205-10. Wales, UK: National Collaborating Centre for
thegeneralnon-Jewishpopulation,which 2.ManchandaR,LoggenbergK,SandersonS, Cancer,NationalInstituteforHealthandClinical
etal.Populationtestingforcancerpredisposing Excellence;2009.
requiresfurtherresearch.Implementation
BRCA1/BRCA2 mutations in the Ashkenazi- 16.National Institute for Health and Clinical
of a population-testing strategy will Jewish community: a randomized controlled Excellence. Early and locally advanced breast
require wide-scale propagation and trial.JNatlCancerInst2015;107:379. cancer:diagnosisandtreatment:NICEClinical
1.e10 AmericanJournalofObstetrics&Gynecology MONTH2017