• No results found

eening tool alth technology

In document KCE REPORTS 172A. (Page 58-63)

elines

w on full-field d language restric the authors. Ther he search strategy

ned..

d trials, prospecti ies included in th es were included

country in which group. First, atte cer detection rate esence of subgrou ompare full-field d phy (SFM) for ca men, performed

7

States are mention ck et al, 200758-60. atic women of 40 s) with both FFDM nths. There were

tion by at least o s 172

but call

m

digital ctions refore y and ive he 1

. The h the ention e and ups in

digital ancer

KCE Reports 17

the two readers The next study number of exam 40 years and which 1345 we modalities.

The recall rat comparison to S The difference more significan (28/152 and 3 difference in c FFDM in both s and 42/181in th calculated sen statistically sign (63% (22 of 35) The positive examinations th SFM (3.2-3.3%

statistically sign Results of thes the interval ca interpretation o interpretations) In the study of 200259 were us between SFM, could be found a statistically (p=.008) and significant diffe These results in modalities (83.

standalone mo modalities in m

72

s), of which 507 of Lewin et al, 20 mined women. Of older, 1467 were ere determined by

te of both stud SFM (p<.001): FF in number of bio nt between both s 38/181) and bot cancer detection studies (total num he second study) nsitivity for canc

nificant difference ) (relative sensitiv

predictive value hat led to a diagno

%) than for FFD nificant.

e studies are bias ancers and by th

f the examination , indicating reade Glueck et al, 20 sed for the compa

FFDM and the c using the param significant differ between FFDM erence in ROC w

ndicate the increa .7%) instead of odality. The imple mammographic sc

by FFDM, 746 b 00259, is a similar f the 6736 examin e positive for ad y SFM and 979 by

ies is significan FDM 11.5-11.8%, opsies between S studies: SFM (83

th (31/152 and rate could be fo mber of cancers o

). In the study of er detection con between FFDM ( vity of 95% (21 of 2 e, defined as t

osis of breast can DM (3.7-3.4%) bu

sed because they he high level of ns (17% of all exa

r variability.

0760, the data of arison of the area combination of bo

etric tests. The n rence between M versus combin was found betwe ased cancer detec

SFM (65.3%) o ementation of the creening has som

y SFM and 195 o analysis but with nations of 4489 w dditional examinat

y FFDM and 293 ntly lower for F

SFM: 13.8-14.9%

SFM and FFDM /152 and 87/181) 56/181). No sig ound between S of 35/152 in the fir f Lewin et al, 200 nfirms that there (60%, 31 of 35) a 22) of FFDM to S he fraction of ncer, was slightly l ut the difference y do not take into disagreement ab aminations had dis the study of Lew a under the curve oth. No difference on-parametric tes SFM versus co ned test (p=.000 een SFM versus ction rate when us f FFDM ( 55.1%

e strategy of usi me financial and

Breast cancer scre

on both.

became ), FFDM gnificant FM and

recalled lower for e is not

account bout the scordant win et al, e (ROC) e in Roc sts show ombined 08). No s FFDM.

sing both

%) as a ing both practical

r

eening

estraints, such as not determine whe or the use of bot

ate.

The main conclus s the lack of diffe Only the recall rate 3.3.2.2. The O Skaane et al, 200 SFM and FFDM w breast cancer scre The results of th FFDM) study61, 63 FFDM: no statistic contrast to the stu ound for FFDM.

curve effect of the The increased nu women aged 50-6 045 women FFDM study between S detection rate with

n the group age underlines the com The Oslo II study of which 9 903 u significantly highe detection rate in F

n the Vestfold Co part of the Norwe

ates were found nadequate imagi detection rate wa Dependent on the nvasive tumours

s an increased c ether the number th modalities resu

ion that can be dr rence in cancer d e is significantly lo Oslo studies

03, 2005, 2004, 2 with soft-copy read eening program (O e Oslo I (n= 36

found no differe cally significant di udies of Lewin et The higher recall

readers.

umber of participa 69 years of which

M) enlarges the SFM and FFDM:

h FFDM and a sig d 50-69 years (p mparable performa follow-up results6 underwent SFM a er recall rate w

FDM, but PPVs a ounty study65 (n=

egian national scr d. The authors ing was significa as higher (but n

e type of tumour no significant dif

cost and manpow of readers, the nu ulted in an increa rawn out the resu detection rate betw

ower in FFDM.

200761-64 compare ding in the Norwe Oslo I and Oslo II 683 women unde nce in performan fference in cance al58, 59, a slightly l rate could be ex ants in the Oslo h 10 391 women

minor difference a higher (but n gnificantly higher p<.05). The lack ance of SFM and

64 (n= 13 912 wom and 4 009 unde with FFDM and are comparable.

18 239 women, eening program, state that recal antly lower with ot statistically si r, the cancer det fference could be

wer. The authors umber of compres ased cancer dete ults of the three st

ween SFM and F

ed the performan egian population-b

studies).

erwent both SFM nce between SFM er detection rate a higher recall rate xplained by a lea II study62 (n= 14 underwent SFM a s found in the O not significant) ca recall rate with F

of difference in FFDM.

men aged 50-69 rwent FFDM) sh a significantly h aged 50-69years no difference in ls due to techn FFDM. The ca gnificant) with F tection rate varied e found between

41

could ssions ection tudies FDM. arning 4 436

higher s), as recall nically ancer FDM.

d: for SFM

42

and FFDM, but significantly hig also a differenc The authors co better than SF difference in PP 3.3.2.3. The The Digital Ma 42 760 wome difference in dia a subgroup an Bonferoni meth area under th classification in between SFM a under the curve in AUC 0.15;

dense or extre p=0.003), pre-0.24; p=0.002), These results target groups.

3.3.2.4. Del Italy Analysis of the 706 underwent al, 200767, sho (4.56% versus found in a sub breast density c p=0.03). The re Differences in type of abnorm round: a highe significantly mo

t for ductal carcin gher with FFDM. I ce in PPV (18.5%

onclude that the FM, based on th PV).

e DMIST study mmographic Imag en underwent b

agnostic accuracy nalysis and, adju hod, set the signif he curve.. In the n age or risk gro

and FFDM: no st e (AUC) (p=0.18).

95%CI 0.05-0.25 emely dense brea or perimenopaus , the diagnostic a indicate FFDM m l Turco et al, 200 y

Italian screening FFDM and 21 55 ow a statistically

3.96%, p=0.01) group analysis of categories (only s ecall rate due to po

cancer detection mality, age group er detection rate

ore cancer cases

noma in situ (DCIS In this study, Vige

in FFDM versus performance of he higher cancer

ging Screening T oth SFM and y between SFM a

sting for multiple ficance level at p e overall group oups) the diagnos

tatistically significa . But in women u 5;p=0.002), wome asts (diff in AUC sal women (diff in accuracy is signifi may be of value

07, screening pr g program (n= 36 56 women underw significant highe . This difference f the age group o significant for the v oor technical qua

rate were found p, breast density with FFDM in wo s as well as mic

S) the detection r eland et al, 20076 15.1% in SFM, p

FFDM is equal r detection rate

Trial (DMIST) stud FFDM) investiga and FFDM. They d

e comparisons us p<0.003 for differe of participants stic accuracy was ant difference in t nder age of 50 ye en with heteroge C 0.11; 95%CI 0.

n AUC 0.15; 95%

cantly higher with for screening in rogram in Floren

262 women of w went SFM) by Del

er recall rate with in recall rate w of 50-59 years a very dense breast

lity was lower with in subgroup ana category and sc omen aged 50-59 cro-calcifications f

Breast cancer scre

rate was

(without s similar the area ears (diff eneously 04-0.18;

CI 0.05-h FFDM.

specific nce, creening 9 years,

eening

FFDM (p=0.007) screening. But the SFM and FFDM i FFDM is similar compensated its FFDM, especially with denser breast n the overall con ack of significanc mentioned, but th women with dense 3.3.3. Primary

cohort st 2011 The evidence fou search for primar primary studies w esults are summa Main finding in th anging from no d performance of FF performance than 3.3.3.1. Recall Similar to the abov FFDM is found in 201068 (n= 312,41 268, 501 SFM, wo ate and the false detection of benig

ate, but this rat decrease over tim

ecommend a train stabilization of the

and a higher de e overall analysis s in cancer detectio r, but the addi

higher recall ra in specific groups ts.

nclusion of the rev ce in results) in p

e higher accurac e breasts) is unde Studies: random tudies and cross und in the review ry studies, publis

ere included in th arized in appendix he primary studie ifference between FDM. No studies

FFDM.

l rate

ve-mentioned stud the majority of t 14 screening mam omen aged 50-75y e-positive results

n microcalcificatio e was still highe me could be expla ning in digital scre increase in recal

etection rate with show no significa on rate. The perf tional cancers d te, suggesting a s, such as young view of AETSA t performance for s cy of FFDM in sp erlined.

mized controlled s-sectional studi w of AETSA et a shed between 20

his report. The ch x.

es is the heterog n SFM and FFDM

were found in wh

dies of the review the primary studie mmograms of wh

years), found afte (due to pseudo-ons) a decrease o

er in FFDM com ained by a learnin eening for the ima

l rate.

KCE Reports

h FFDM at incid nt differences bet formance of SFM detected with F a higher sensitiv

er women and wo the heterogeneity screening populat ecific groups (suc d trials, prospecti

ies published 20 al42 is updated w 008 and 2011. F haracteristics and

eneity in perform M to a significant h hich SFM had a b

w, a higher recall r es68-73. Bluekens hich 43,913 FFDM er a peak of the re

lesions and incre over time of the re mpared to SFM.

ng curve. The au age readers to obt

s 172

ifteen main mance higher better

ate in et al, M and eferral eased eferral This uthors tain a

KCE Reports 17

In contradiction 838 mammogra women aged 4 FFDM, p<.001 p<.001)in SFM FFDM. Also H recall rate for S Vinnicombe, 2 201077 (n= 14, rate with FFDM Table 12 Overv

Author, year Bluekens, 20 Sala, 2011 Heddson, 200 Vinnicombe, Juel, 2010 3.3.3.2. Can Several authors an increased ca cancer detectio Domingo et a mammograms women aged 45 performance b tumor characte was significant The conclusion aged 50-64yea recall rate by th increase in can 72

n to the above-me ams of which 17 45-69years), found

) and false-pos . The cancer dete eddson et al, 20 SFM.

00976 (n= 39,65 374 women, age M.

view of recall rat Reca

ncer detection ra s 68, 69 mention th ancer detection ra on rate between S

al, 201178 (Spa of which 171,19 5-69years) confirm etween SFM and ristics and cance higher in FFDM.

ns of the study of ars) contradict th he increased canc

ncer detection ra

entioned studies, S 1,191 SFM and 7 d a higher recall sitive rate (7.6%

ection rate did not 00775 (n= 24,875 51 women, aged ed 49-70 years) fo tes in primary stu

ll rate in SFM

%

ate

he compensation ate but the signific

FM and FFDM is anish Screening 91 SFM and 71,6

ms the results of d FFDM, without r detection rate. O f Feeley et al, 20

e above-mention cer detection rate ate, found in their

Sala et al, 201174 71, 647 FFDM, 1

rate (8.1% SFM SFM vs 5.7%

t differ between S women) found a d 50-70years) an ound no increase

udies

Recall rate in 4.3% (p<.001) 11.7% (p=0.91 1.0% (p<.001) 3.2% (p=.44) 2.4% (p>.05)

of a higher recal cance of the diffe

often lacking.

Program, n=

647 FFDM from the review: a com t significant differ Only the PPV for 01069 (n=107,818 ned compensation e. The authors exp r study, by the im

Breast cancer scre

(n= 242 242,838 103,613 mparable rence in

masses women n of the plain the mproved

d

eening

detection of micr esions’ may lead t n the study of Ka which 56,518 FF combined with C detection perform microcalcifications he recall rate was The increased can of Lipasti, 201079 Heddson, 200775. Pisano et al, 200

mpact of breast between SFM and Pisano et al, 2005

n pre- and perim breasts.

n the study of Van cancer detection Vinnicombe et al, but no differences Table 13 Overview

Author, year Domingo, 2011 Feeley, 2011 Karssemeijer, 2 Lipasti, 2010 Perry, 2010 Vernacchia, 200

rocalcifications. T to an overtreatme arssemeijer et al FDM and 311,08 CAD (computer-ai mance as SFM

s was improved in s increased.

ncer detection rat

9, Perry, 201080, 0881 (n= 49,528 density, age, m d FFDM, conclud 566: a better perfor menopausal wome

n Ongeval et al, 2 rate could be fou 200976 shows a in recall rates or w of cancer dete cance

The detection o ent.

, 200970 (n= 367 82 SFM, aged 5

ided detection), alone. The det n a statistically sig

te with FFDM is a Vernacchia, 200 women) analyse menopausal statu

ding similar findin rmance (area und en younger than 201071 and Juel, 2

und. The meta-an slightly higher de PPVs.

ection rate in prim er detection rate

M

f such ‘minimal ,600 mammogram 50-75years) FFD

resulting in a s tection of DCIS gnificant way, but also found in the 0972, Hambly, 20 es more profound s on the compa ngs as in the stu der the curve) of F 50 years with d 201077, no differen nalysis in the stu etection rate for F

mary studies e Cancer dete

rate in FFDM 0.43% (p=0.59 7.2 per women (p=0.0 0.77% (p=.11) 0.623%

6.4 per women (p<.00 7.9 per women (p=0.0

43

sign ms of DM is similar

and t also study 00973, d the arison udy of

ection 92)

44

Hambly, 2009 Heddson, 200 Van Ongeval, Juel, 2010 Vinnicombe, 3.3.3.3. Diff The fourth edit physical and t minimum stand quality control o of screening:

information an (ALARA). The control and a pr In digital screen

• the direct d The detector is are directly sho photon-counting

• the indirect The imaging de and a remov implementation Most above me systems and D system (such a 3.3.4. Discu The heterogene about the clinic

9 5.2

07 0.3

, 2010 0.6 0.3

2009 0.7

ferent systems o ion of the “Europ technical aspects dards for quality c on the performanc the image qua d the breast do European comm rotocol on dosime ning two systems detection or digita integrated in the own on the scree g systems.

t detection of com etector incorporat vable digital rea

in SFM units.

entioned studies u R systems and th s needle plates) ussion

eity in results ha al performance of

2 per 1000 women 31%

64%

39%

72 per 100 women

of digital screenin pean protocol for

s of mammogra control of mammo

ce standards is bu ality, the minim ose As Low As mission develope etry in Mammogra can be distinguish l radiography (DR

digital mammogr en. The DR syste mputed radiograph

tes a phosphor to ader system is use a mix of both he further investiga

falls out of the sco

ampers to draw o f FFDM. The diffe

n 6.3 per

women (p=

0.49%, (p=0.04) 0.59% (p=0 0.48% (p>0

n 0.68 pe

women (p=

ng

the quality contro phy screening”

ography screenin uilt on three corne um level of di

Reasonably Ac ed a protocol for aphy (EUR16263)

hed83: R)

raphy unit and the ems incorporate hy (CR)

o produce visible s used, facilitat systems. Compa ation into new dev ope of this report.

one consistent co erence in cancer d

Breast cancer scre

1000 erstones agnostic hievable r quality .

e images also the

photons ting the aring CR

veloping .

nclusion detection

r

eening

ate ranges from n with FFDM. Simil majority of the explanations for t variability of the precise detection hypothesis of com detection of cance variables and warn The quality of digi

eading of the dig a specific training

he Belgian expert Next to the clinica nfluence the imp user friendliness, FFDM. Nowadays he information ex he integration of however, like the derived from anoth As regards to the

he clinical perform ecall rate. Othe characteristics are

he performance o cost-effectiveness As regards to the some aspects ha difference in recal an increase or de o specific for each n conclusion coul east similar perfo he population-bas

no difference to a lar range in resu

studies found a this increase in r

readers, the lea of microcalcifica mpensation of the

ers; others contra n for the risk of ov tal screening can ital mammograph in reading of dig ts.

al performance of lementation in cl

data storage, da s the evolution to xchange between FFDM in a screen

high cost and di her technology (V content of this re mance of FFDM,

er performance e not described in of FFDM, other as etc, should also b methodological a mper the interpre

l rate between th crease could be h country.

ld be stated that a rmance of FFDM sed screening pro

significant highe ults can be seen an increase in recall rate are me

rning curve of th ations etc. Some e increased recall adict this positive

vertreatment.

n be guaranteed b hies. The importan gital mammograph a technical moda inical practice. T ata exchange, etc owards the electro n health providers

ning program. Th ifficulty of sharing Van Ongeval et al, port, we decided in particular can indicators, suc this report. In ca spects, such as te be considered.

aspects of the abo etation of the res e countries. Only mentioned, the a all the authors ag

and support the ograms as an equi

KCE Reports

r cancer detection n in the recall ra

recall rate. Diff entioned, such a he readers, the e authors sugges rate by the incre look on the incre by a specific train

nce of the provisi hies is emphasize ality, other factors

hese factors, suc c are in advanta onical medical file s via internet, sup

ere are disadvan g of the digital im

2007)83.

to restrict this stu cer detection rate ch as the tech se of a overall vie chnical character ove-mentioned stu sults. For exampl y the tendency tow

bsolute numbers gree on the better

integration of FFD ivalent to SFM.

s 172

n rate ate. A

ferent as the pports tages mages udy to e and hnical ew on istics, udies,

KCE Reports 17

Key points

Studies on convincing screening

Some sub better in p

FFDM and

Organisat exchange, practice

3.4. Breast

3.4.1. Syste

asses After selection breast ultrasou quality meta-an A summary of c Table 35.

Although the 4 with important sensitivity for u reported range As it was not po most exact esti of the primary evidence was u of the most re discussed in the

72

n the value of FF g proof that it be

program.

bgroup analyses premenopausal w d SFM can be see ional aspects, su , facilitate the im

Ultrasound as

In document KCE REPORTS 172A. (Page 58-63)