Res second - KCE REPORTS 172A.

outcom

owed 759

43 .680 686

e high ve

% sholds

r g

ults

dary and other mes

VII Critical appr of study quality Results only vali amongst particip in genetic clinic.

101

raisal y

d pants

102

I Study ID Ruodgari 2007

II Method Design

validation of risk prediction in genetic clinics Funding NIH Setting USA (mayo genetic clinic Sample size 200 families

III Patient characteristics Eligibility criteria 275 Scottish families tested f BRCA1/2 mutations in genetic clinics i

s IV Intervent

for

Index test(s) Four probab estimation m including CO Manchester system (MSS), BOA and Tyrer–C C)

Breast cancer scre

tion(s) V R out )

bility models OS,

scoring ADICEA Cuzick (T–

COS mod the sen and curv maj stru sho sen (91–

(76–

enti How and high for t fam BOA T–C bes prev mut pop

eening

Results primary come

S and MSS dels demonstrated

greatest sitivities

d area under ROC ves for the

ority offamily uctures. They also

wed the highest sitivities

–92%) and AUCs –78%) for the

re dataset overall wever, BOADICEA d T–C had the

hest specificities the majority of the mily structures.

ADICEA and C generated the

t estimates for the valence of tations in the pulation

VI Results secondary an outcomes d

nd other

VII Cr appra qualit Result among in gen

KCE Reports

ritical

aisal of study ty

ts only valid gst participants netic clinic.

s 172

KCE Reports 17

I Study ID Antinou 2008

35, 36

I Study ID Antinou 2008

II Method Design

validation of risk prediction in genetic clinics centers.

Funding NCI Cancer Genetics Network + divers Setting USA (Sample size

2140 families

II Method Design

validation of risk prediction in genetic clinics supported by a grant from the UK Department of Health.

Setting 6 UK genetic clinics

III Patient characteristics Eligibility criteria 1934 families tested for BRCA1/2 mutations.

III Patient characteristics Eligibility criteria 1934 families seen in cancer genetics clinics the UK in whom an index patient had bee screened for BRCA1 and/or BRCA2 mutations.

s IV Intervent a: Index test(s

BRCAPRO, Manchester system and tables,

s IV Intervent a:

s in x

Index test(s carrier prediction a BOADICEA BRCAPRO, Manchester system and tables,

Breast cancer scre

tion(s) V R outc )

IBIS, the r scoring

Myriad

All m simi BRC IBIS 2) .

tion(s) V R outc )

lgorithms A,

IBIS, the r scoring

Myriad

calib

Only calib BOA stat diffe

All m und prob

eening

Results primary come

models showed ilar AUC : CAPRO=0.76, S=0.74)Myriad=0.7

Results primary come

bration

y BOADICEA wel brated (only for ADICEA no

istically significan erence E/O.

models erestimate bability in low risk

VI Results secondary a outcomes

and other VII C of st Resu amon in ge

Critical appraisal udy quality ults only valid

ngst participants enetic clinic.

Critical appraisal udy quality ults only valid

ngst participants enetic clinic.

103

104

I Study ID I

Panchal 2008

D s v p g N fu u s S F T C 1 2

Sample size 2140 families

I Method

Design cross ectional alidation of risk prediction in

enetic centre No source of

unding was sed for this tudy Setting Family cancer Toronto, Canada

00 carriers and 00 non-carriers

III Patient characteristics Eligibility criteria high risk participants in genetic clinics in Canada

IV Interventi

: Index test(s) BRCAPRO, Manchester, Myriad II, FH and BOADIC models

Breast cancer scre

pop

Acc rece cha stat BOA BRC IBIS Man Myr

on(s) V Re outco

Penn II, AT, IBIS CEA

BRCA Myria BOAD have of ap for BR Manc mode AUCs 0.47 At a 1 thres sensi speci BRCA respe

eening

ulation

curacy:

eiver operating racteristic curve istics:

ADICEA=0.77, CAPRO=0.76, S=0.74,

nchester=0.75, riad=0.72)

sults primary ome

APRO, Penn II, ad II, FHAT and DICEA models all

similar AUCs proximately 0.75 RCA status. The chester and IBIS els have lower

s (0. and respectively).

10 % testing hold, the itivities and ificities for a A mutation were, ectively, as

VI Results secondary an outcomes

nd other VII Cri of stu Result among in gen

KCE Reports

itical appraisal dy quality ts only valid gst participants

etic clinic.

s 172

KCE Reports 17

I Study ID I

Lindor 2010

D v p g F m S U g S 2 72

I Method

Design

alidation of risk prediction in

enetic clinics Funding not mentioned Setting USA (mayo

enetic clinic Sample size

00 families

III Patient characteristics Eligibility criteria 200 families see in Mayocancer genetics clinics i whom an index patient had been screened for BRCA1 and/or BRCA2 mutations.

IV Interventi

: n n n

Index test(s) LAMBDA,

; BRCAPRO,

; modified Co tables Myriad II tabl

Breast cancer scre

follow (0.75 Manc (0.58 II (0.9 (0.71 (0.70 (0.20 BOAD 0.65)

on(s) V Re outco

a ouch

All m simila ROC of 0.7 mode LAMB unde numb All m dispe

eening

ws: BRCAPRO 5, 0.62),

chester 8,0.71), Penn

93,0.31), Myriad I ,0.63), FHAT 0,0.63), IBIS 0,0.74),

DICEA (0.70, )

sults primary ome

odels gave ar areas under the

curve 71 to 0.76. All els except BDA substantially

r-predicted the bers of carriers.

odels were too ersed

VI Results secondary an outcomes

nd other VII Cri of stu Result among in gen

itical appraisal dy quality ts only valid gst participants

etic clinic.

105

I S NZ 20 20

106

Study ID ZHTA 007¹¹NZHTA 007¹¹

1 2 3 4

I 1 2 c t s

II Method 1. design system

review 2. Government o

Zealand 3. Search date n

2005 4. Searched

databases Medline and Embase datab the Cochrane Database of Systematic Reviews, the D and HTA data ncluded study des 1. Systematic revie 2. Clinical studies control group for th

ime period beyon systematic reviews

5. Number of inc studies 139

III Patie charact s matic

of New nov

bases,

DARE bases signs ews.

with a he

d key s

cluded

Wom who asse for b canc

nt teristic

IV Interve n(s) men

o were essed breast

cer

1. risk fact for brea can

Breast cancer scre

ntio V Results k

tors ast ncer

1. Effect

• pa an

• du

• At

• lob

• du

• inc

• alc 25

• nu ap

• po

• for an

• ho

• cu 1.

• hig For some

eening

primary outcom

size primary outc

ast history of breas nd 7.4)

uctal hyperplasia R ypical ductal carc bular carcinoma R uctal carcinoma in

creased breast de cohol intake (10%

5g alcohol/day and ulliparity relative ri pproximately 0.09 ost menopausal ob

r the association w nd 1.25 for the obe ormone replaceme urrent or recent us

current users: R 1-4 years after 1.08-1.23) 5-9 years after 1.02-1.13)

>10 years after 0.96-1.05).

gh total energy int risk factors the lev

come(s)

st cancer ( RR be RR 1.5 - 2 cinoma RR 4 RR 6-10 n situ RR 8 - 10

ensity RR 4

% for 10g alcohol/d d 55% for 50g alc sk estimates decr for each additiona besity RR 1.12 with the overweig ese category ent therapy RR 1.

se of oral contrace RR 1.24 (95% CI 1

stopping: RR 1.16 stopping: RR 1.07 stopping: RR 1.0 take 1.4-2.1 vel of increased ri

etween 2.8

day, 25% for cohol/day)

rease by al birth

ht category 2-1.4 eptives

1.15-1.33) 6 (95% CI 7 (95% CI 01 (95% CI

isk was

KCE Reports

VII Critical appr review quality

High quality

The majority studies inclu review used control desig control studie characterized susceptibility selection bia recall bias.

s 172

raisal of

review

of ded in the

the case-gn. Case

es are d by y to

s and

KCE Reports 17

I Study ID

Vrieling 2010

II Method Design systematic review of observational studies funded by the Deutsche Krebshilfe Search date March 2010

III Patient characteristics Adult women

s IV Intervent Index test(s BMI or othe of wheight Comparing highest vers lowest categ of adult weig ER: estroge status PR progeste receptor sta

Breast cancer scre

difficult to d

• ea

• xe

• ph

• sti

tion(s) V R outc )

r measure the sus the

gories ght gain en receptor

erone atus

risk ER+

(11 2.03 2.45 sign hete (p h 0.00 betw mixe pre-post wom com RE 0.86 post wom stud 95%

eening

determine:

arly menarche (like enoestrogens hytoestrogens

lboestrol.

Results primary come

for ER+PR+ and + tumors combine

studies; RE = 3; 95% CI 1.62, 5). Statistically nificant

erogeneity heterogeneity =

02) was shown ween REs for a

ed population of - and

tmenopausal men

mbined (4 studies;

= 1.54; 95% CI 6, 2.22) and for

tmenopausal men only (7 dies; RE = 2.33;

% CI 2.05, 2.60).

ely to be relatively

VI Results secondary a outcomes ed Risk for ER-P

tumors amon postmenopau women (7 stu

= 1.34; 95%

1.63), but statistica significantly d from risk for tumors(p for heterogeneity No associatio observed for ER+PR- tum whereas risk PR+ tumors be assessed

y modest)

and other VII C of re PR-

ng usal udies; RE

CI 1.06, lly different ER+PR+

y\0.0001).

ons were ors for ER-could not .

C im r t u

Critical appraisal eview quality

Clinical mplications of receptor status of

he tumors unclear

107

108

I Study ID

Cumming s 2009⁷ Cumming s 2009⁷ Update of McCormac k 2006

II Method

Search date 2008

Design systematic review of observational studies funded by the National Cancer Institute and the Daniel and Phyllis Da Costa Fund

Meta-analysis 47 prospective studies.

III Patient characteristics Adult women

s IV Intervent Index test(s 3 different m of breast de Wolfe grade BI-RADS

% Of breast is dense

Breast cancer scre

tion(s) V R

)

measures ensity:

t area that Wol N1 ( P1 P2 Dy (

BI-R 1 (fa 2 (s 3 (h den 4 (e

% O

<5 5 – 25 – 50 –

≥ 75 eening

Results primary o

lfe grade

(fatty) 1 1.76 3.05 (most dense)

RADS atty) 1 (ref

cattered densities heterogeneously

se)

extremely dense)

Of breast area that 1 (ref 24 1.74 – 49 2 – 74 2.92 5 4.20

outcome

1 (reference) 6 (1.41 to 2.19) 5 (2.54 to 3.66)

3.98 (2.53 to 6.27

ference)

s) 2.03 (1.61 to 2 2.95 (2.32 to 3

4.03 (3.10 to 5

t is dense ference)

(1.50 to 2.03) 2.15 (1.87 to 2.48

(2.55 to 3.34) (3.61 to 4.89)

VII C of re

2.56) 3.73)

5.26)

)

T a M w t a s a s f b o o s s a

KCE Reports

Critical appraisal eview quality

The meta-analysis by McCormack et al.

was included in his meta-analysis. All studies were adjusted for age;

studies that urther adjust for body mass index or weight

observed somewhat stronger associations

s 172

KCE Reports 17

I Study ID

Key et al, 2006

128

II Method

Search date 2005 Design systemati review & meta-analysis of observational stu Study funded by t Department of He in England 98 studies were included,

involving 75,728 a 60,653 cases in drinker versus nondrinker and dose–respon analyses, respect

III Patien characte stics 5

dies the ealth

and

nse tively.

Adult women

eri IV Intervent

Index test(s Alcohol use

Breast cancer scre

tion(s) V R outc

) exce

with 22%

eac 10 g asso high 5–1 evid bias sign type stat

eening

Results primary come

ess risk associate h alcohol drinking

% (95% CI: 9–37%

h additional g ethanol/day was

ociated with risk her by 10% (95%

5%). There was n dence of publicatio s. Risk did not diffe nificantly by bever e or menopausal

us.

VI Results secondary and other outcomes ed

was

%);

s CI:

no on er rage

Estimated population attributable risks were and 6.0%

USA and U respective

s y r s

VII Critica review qu

n e 1.6 in UK, ely

Consi hetero effects meta-used to exp hetero

al appraisal of uality

derable ogeneity in

s measures, - regression was

but did not help plain the

ogeneity.

109

110

I Study ID

Kahlenbor n et al, 2006

II Method

Search date 2006 Systematic review meta-analysis Supported by Nat Institutes of Healt (US)

34 eligible studies

III Patien characte

6 w and

tional th s

Adult wo nulliparo or multip

nt eristics

IV Inte s) men,

us, parous arous.

Inde Ora con use

Breast cancer scre

ervention( V R prim ex test(s)

ntraceptive e (OC)

Use asso incre prem brea gen 95%

1.29 vario OC stud prov nulli paro sepa use asso brea in bo (OR 1.20 nulli 1.24 0.92

eening

Results

mary outcome e of OCs was

ociated with an eased risk of menopausal ast cancer in

eral (OR, 1.19;

% CI, 1.09- 9) and across

ous patterns of use. Among dies that

vided data on iparous and ous women arately, OC

was ociated with ast cancer risk

oth parous R,1.29; 95% CI,

0-1.40) and iparous (OR, 4; 95% CI, 2-1.67) women.

VI Results secondary and other outcomes Longer duration o use did not substantially alte risk in nulliparous women (OR, 1.2 95% CI, 0.85-1.9 Among parous women, the association was stronger when O were used before first full-te pregnancy (FFTP (OR, 1.44; 95% C 1.28-1.62) than a FFTP (OR, 1.15;

95% CI, 1.06-1.2 The association between OC use breast cancer ris was greatest for parous women w used OCs 4 or m years before FFT (OR,1.52; 95% C 1.26-1.82)

VII Critica review qu

of er

s 9;

96).

OCs erm P) CI, after 26).

e and sk who more

TP CI,

Only case in meta-an

DerSimon random ef used but n measure o heterogen

KCE Reports

al appraisal of uality

e control studies nalysis

nian-Laird ffects model no other

or exploration of neity.

s 172

KCE Reports 17

Appendix 3.2 Appendix 3.2

Systematic rev Table 30 Doub

Reference Dinnes et al, 2001³⁹

2. Technical m 2.1.

Double read

views

ble reading and c

Methodology

• SR

• Funding: UK Department o Health R&D D

• Search date:

between April and July 1999

• Databases: M CINAHL, DHS BIOSIS, Emba BIDS, Cancer NHS EED, CC Dissertation abstracts, PAS Conference P Index, SIGLE, Star, EconLit

• Study design:

prospective an retrospective c studies

• N included stu cohort studies

methods for bre

In document KCE REPORTS 172A. (Page 119-129)