static-content.springer.com10.1007... · web viewa systematic review of elastography, electrical...
TRANSCRIPT
A systematic review of elastography, electrical impedance scanning, and digital infrared thermography for breast cancer screening and diagnosis: Supplementary online-only material
Thomas D. Vreugdenburg1, Cameron D. Willis1, Linda Mundy2, Janet E. Hiller3
1Discipline of Public Health, The University of Adelaide, SA, Australia2Health Policy Advisory Committee on Technology, Queensland Government, QLD, Australia3Faculty of Health Sciences, Australian Catholic University, VIC, Australia
eTables
eTable 1: Search strategy
eTable 2: Inclusion and exclusion criteria
eTable 3: NHMRC hierarchy of evidence according to research question
eTable 4: Diagnostic accuracy results of included studies
eTable 5: Investigation of heterogeneity in subgroup likelihood ratios
eTable 6: Examples of advertising claims made by promoters of DITI, EIS and EPI devices
eFigures
eFigure 1: Percentage of all studies fulfilling individual QUADAS criteria.
eFigure 2: Individual study estimates of sensitivity and specificity for USE using ANN
eFigure 3: Individual study estimates of sensitivity and specificity for USE using length ratio
eFigure 4: Individual study estimates of sensitivity and specificity for USE using strain ratio
eTable 1 Search strategy
Query Search Terms#1 breast neoplasms[MH] OR breast neoplasm*[all fields] OR breast[all fields] OR
mammary[all fields]
#2 (thermography[mh]) OR (thermal imag*[tiab]) OR (infrared imag*[tiab]) OR (infra-red
imag*[tiab]) OR (thermology[tiab]) OR (thermometry[tiab]) OR (thermograph[tiab])
#3 (electrical impedance[mh]) OR (electrical impedance[tiab]) OR (electromagnetic
imag*[tiab]) OR (bioelectric* impedance[tiab]) OR (electric impedance[tiab])
#4 (Elasticity Imaging Technique*[MH]) OR (computerised imag*[tiab]) OR
(computerized imag*[tiab]) OR (mechanical imag*[tiab]) OR (tactile imag*[tiab]) OR
(soft tissue elastometer*[tiab]) OR (electronic palpation imag*[tiab]) OR (EPI[tiab])
OR (Model-based imaging[tiab]) OR (Force sensor array[tiab])
#5 #1 AND (#2 OR #3 OR #4)
eTable 2 Inclusion and exclusion criteriaInclusion Exclusion
Population -Human participants -Animal participants-Symptomatic women (diagnostic) -Simulated/prosthetic samples-Asymptomatic women (screening) -Excised tissue samples
Intervention -Screening and/or diagnosis of breast cancer
-Other interventions-Not breast cancer
Index Test -Elasticity imaging -Contact thermography-Electrical impedance imaging -Device not defined/identified-Digital infrared thermal imaging
Outcome - Effectiveness
-Survival rates/reduction in mortality
-Quality of life-Interval cancers (surrogate)-Cancer detection rates (surrogate)-Tumor size/stage (surrogate)
Outcome - Accuracy -Sensitivity/Specificity-False positive/negative rates-Positive/negative predictive values-Positive/negative likelihood ratios
Outcome - Safety -Physical harm-False positive/negative rates-Physical harm
Publication Type -Peer-reviewed journal articles -Letter to the editor-Conference abstract or poster-Narrative review-Editorial
Year of Publication -1981 till March 2011 -1980 or earlier
eTable 3 NHMRC heirarchy of evidence according to research questionLevel Intervention Diagnostic accuracy Screening InterventionI A systematic review of level II
studiesA systematic review of levelII studies
A systematic review of level II studies
II A randomised controlled trial A study of test accuracy with: an independent, blinded comparison with a valid reference standard, among consecutive persons with a defined clinical presentation
A randomised controlled trial
III-1 A pseudorandomised controlled trial(i.e. alternate allocation or some other method)
A study of test accuracy with: an independent, blinded comparison with a valid reference standard, among non-consecutive persons with a defined clinical presentation
A pseudorandomised controlled trial(i.e. alternate allocation or some other method)
III-2 A comparative study with concurrent controls:▪ Non-randomised, experimental trial▪ Cohort study▪ Case-control study▪ Interrupted time series with a control group
A comparison with reference standard that does not meet the criteria required forLevel II and III-1 evidence
A comparative study with concurrent controls:▪ Non-randomised, experimental trial▪ Cohort study▪ Case-control study
III-3 A comparative study without concurrent controls:▪ Historical control study▪ Two or more single arm study ▪ Interrupted time series without a parallel control group
Diagnostic case-control study A comparative study without concurrent controls:▪ Historical control study▪ Two or more single arm study
IV Case series with either post-test or pre-test/post-test outcomes
Study of diagnostic yield (no reference standard)
Case series
eTable 4 Diagnostic accuracy results of included studiesA
Citation Device name
Index testB,C,D
Measurement scale Cut-off score for benign/malignant
Sensitivity Specificity PPV NPV
Arora et al, 2008 BreastScan DITI Risk score 0 (minimal) to 7 (high) in screening mode B = 0 M= 1-7 96.7% 11.8% 65.9%* 66.7%Button et al, 2004 BioScanIR DITI Expert score from 1-3 B= 1 M= 2-3 81% 54% 40.0%* 87.5%Keyserlingk et al, 1998 Bales TIP DITI 6 point imaging criteria (Ville Marie Scale) Any score was disease positive 83.0%* 81.0%* 81.4%* 82.7%*Kontos et al, 2011 Med 2000 DITI 5 point thermal imaging scale (T1-T5) B=T1-T2 M = T4-T5 25% 85% 24% 86%Parisky et al, 2003 BCS2100 DITI Suspicion index scale from 0 to 100 Predetermined threshold 97% 14% 24% 95% Tang et al, 2008 TSI-21 DITI Localised temperature increase B= ↑ < 1C M= ↑ ≥ 1C 93.6% 55.7% 53.% 91.2%Wang et al, 2010 ATIR-M301 DITI Modified Ville Marie Infrared Grading Scale B <0.3 M ≥ 0.3 72.4% 76.6% 81.3% 66.4%Wishart et al, 2010 BreastScan DITI Risk score 0 (minimal) to 5 (high) in screening mode B=0 M = 1-5 53% 41% 59% 36%Diebold et al 2005 TS2000 EIS 5 point level of suspicion index in targeted mode B = 1-3 M = 4-5 75.4% 42.0% 52.7% 66.7%Fuchsjaeger et al 2005 TS2000 EIS 5 point level of suspicion index in targeted mode B = 1-3 M = 4-5 94.6% 74.7% 60.3% 97.1%Glickman et al 2002 TS2000 EIS 5 point level of suspicion index B = 1-3 M = 4-5 86% 71% 39.2%* 95.7%*Malich et al 2003 TS2000 EIS “Bright white spots” on targeted mode B = no spot M = spot 79.8% 64.0% 52.6% 86.4%Malich et al 2007 TS2000 EIS “Bright white spots” on targeted mode B = no spot M = spot 78.8% 58.8% 54.3% 81.7%Martin et al 2002 TS2000 EIS “Bright white spots” on targeted mode B = no spot M = spot 94.0%* 8.3%* 68.1%* 40.0*Melloul et al 1999 TS2000 EIS Presence or absence of “bright white spots” B = no spot M = spot 72.2% 67.0% 27.7%* 93.2%*Stojadinovic et al 2008 TS2000ED EIS Software algorithm scored lesions positive or negative B = red line M = green line 26.4% 80.9%* 28.4% 79.3%Szabo et al 2005 TS2000 EIS 5 point level of suspicion index on targeted mode B = 1-3 M = 4-5 86% 49% 67% 74%Wang et al 2010 EIS 1000 EIS “Bright white spots” on targeted mode B = no spot M = spot 86.7% 72.9% 51.2% 94.1%Wersebe et al 2002 TS2000 EIS “Bright white spots” on targeted mode B = no spot M = spot 62.0% 69.0% 71.0% 59.7%Egorov et al 2009 SureTouch EPI Naïve Bayesian classifier. NR NR 87.5% 84.4% 54.9%* 96.9%*Kaufman et al 2006 SureTouch EPI Subjective interpretation of pressure patterns. N/A N/A 94.0%* N/A N/A N/AAthanasiou et al 2010 ATI HIDI 1000 USE + US Combined BI-RADSTM + elasticity score NR NR 95% 96% 95.0%* 96.4%*Barr 2010 Elegra/Antares USE Ratio of lesions length found on USE:B-mode B < 1 M ≥ 1 100% 95% 84% 100%Burnside et al 2007 Elegra USE Probability of malignancy B <2% M > 2% 99.3% 25.7% 57.6%* 92.3%*Chang et al 2011 EUB-8500 USE High quality images: Modified 3 point Itoh et al scale B = 0-1 M = 2 87.0% 85.1% 50.0%* 97.4%*Chen et al 2006 Voluson 530 USE Support vector machine Automatic classification 85.0% 95.0% 91.9% 90.5%Cho et al 2008 EUB-8500 USE 5 point Itoh et al scale B = 1-2 M = 3-5 82% 84% 52% 96%Cho et al 2010 EUB-8500 USE Strain ratio (fat:mass) B≤ 2.2 M > 2.2 95% 75% 48% 98%
Chung et al 2010 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 58.0% 92.9% 84.8% 75.0%USE Strain ratio B ≤ 212 M > 212 88% 78.6% 74.6%* 90.2%
Evans et al 2010 Aixplorer USE Mean elasticity score (kPa) B< 50 M > 50 97% 83% 88% 95%Fleury et al 2009 Sonix SP USE 4 point scale B = 1-3 M = 4 86.7% 96.0% 76.5% 97.9%Garra et al 1997 Spectra USE Width differences + brightness score differences NR NR 100% 56% 63.2% 100%Giuseppetti et al 2005 EUB-8500 USE 5 point Itoh et al scale B = 1-2 M = 3-5 79% 89% 94.4%* 64.9%*Itohet al 2006 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 86.5% 89.8% 88.2%* 88.3%*Ko et al 2011 EUB-7500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 65.4% 90.4% 72.3% 85.8%Kumm et al 2010 Hi Vision 900 USE Strain ratio B < 4.5 M ≥45 79% 76% 57% 90%
USE 5 point Itoh et al scale B = 1-2 M = 3-5 76% 81% 60% 90%Lee et al 2011 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 35.4% 98.9% 85% 89.5%
USE Strain ratio B ≤ 2 M >2 68.8% 64.8% 26% 92%Leong et al 2010 Antares USE Length ratio B≤ 1.1 M > 1 100% 73.8% 54.2%* 100.0%*Moon et al 2005 Voluson 530 USE Support vector machine Automatic classification 75% 83% 75% 83%Moon et al 2009 EUB-8500 USE Artificial Neural Network B < 0.5 M ≥ 0.5 83.8% 87.6% 80.3% 90.0%Moon et al 2010 EUB-8500 USE Artificial Neural network B < 0.5 M ≥ 0.5 92% 74% 58% 96%
USE 5 point Itoh et al scale B = 1-3 M = 4-5 54% 91% 70% 84%Navarro et al 2011 Antares USE 5 point Itoh et al scale B = 1-3 M = 4-5 69.5% 83.1% 78.9% 75.0%Parajuly et al 2010 Antares USE 5 point Itoh et al scale B = 1-2 M = 3-5 95.7% 96% 94.3% 96.9%Raza et al 2010 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 83.6% 87.4% 76.1% 91.7%Regini et al 2010 EUB-8500 USE Modified 5 point Itoh et al scale B = 1-2 M = 3-5 88.5% 92.7% 86.1%* 94.1%*Regner et al 2006 Elegra USE Width ratio of USE versus B-mode US B ≤ 1.0 M > 1.0 96% 21% 62.0%* 80.0%*
USE Area ratio of USE versus B-mode US B ≤ 1.1 M > 1.1 96% 24% 62.8%* 81.3%*Satake et al 2011 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 81.8% 70.4% 90.0%* 54.3%*Scaperrotta et al 2008 Hi Vision USE Modified 5 point Itoh et al scale B = 1-3 M = 4-5 80.0% 81.0% 71.5% 87.1%Schaefer et al 2009 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 96.9% 76.0% 66.7%* 98.0%*Sohn et al 2009 Antares USE 5 point Itoh et al scale B = 1-3 M = 4-5 65.5% 79.0% 45.8% 89.4%
USE Length ratio B ≤ 0.6 M> 0.6 87.7% 54.7% 35.2% 94.1%Tan et al 2008 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 78.0% 98.5% 93.9%* 94.2%*Thomas et al 2006 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 77.6%* 88.1%* 84.4%* 82.5%*Thomas et al 2007 Aplio80 USE 5 point Itoh et al scale B = 1-3 M = 4-5 96% 80% 82.8%* 95.2%*
Thomas et al 2010 Hi Vision USE 5 point Itoh et al scale B = 1-3 M = 4-5 81% 89% 84% 82.1%*USE Strain ratio B<2.5 M ≥ 2.5 90% 89% 89% 90.2%*
Wojcinski et al 2010 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 81.2% 89.5% 86.8% 84.8%Yerli et al 2011 EUB-7000 USE Modified 5 point Itoh et al scale B = 1-3 M = 4-5 95% 80% 84% 93%
USE Strain ratio B ≤ 3.5 M > 3.5 93% 80% 80% 93%Yoon et al 2011 EUB-7500 USE + US Combined methods Combined methods 98.5% 33.3% 45.1% 97.9%Zhi et al 2007 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 70.1% 95.7% 87.1% 88.5%Zhi et al 2010 EUB-8500 USE 5 point Itoh et al scale B = 1-3 M = 4-5 70.1% 93.0% 77.7% 90.0%
USE Strain ratio B ≤ 3.1 M >3.1 92.5% 91.1% 78.2% 97.2%Zhu et al 2008 EUB-8500 USE 5 point Itoh et al scale B = 1-4 M = 5 85.5% 88.6% 88.1%* 96.1%*ANR = Not ReportedBMG = MammographyCUS = B-mode ultrasoundDDITI = Digital Infrared Thermal Imaging*Calculated by corresponding author with extracted data, not reported in the original article.
eTable 5 Investigation of heterogeneity in subgroup likelihood ratiosDevice class and subgroups Studies (n) I2 (95% CI) Q (p value)EISa
Measure - Bright white spot 6 92 (84-99) 24.78 (p=0.000)Measure - Level of suspicion 4 0 (0-100) 0.79 (p=0.337)Non-blinded index test 4 93 (88-99) 30.51 (p=0.000)Mean age 50-69 4 91 (81-100) 21.34 (p=0.000)DITIb
All studies 7 99 (99-100) 241.95 (p=0.000)Prospective design 6 100 (99-100) 410.33 (p=0.000)USEc
Measure - Elasticity score 26 99 (98-99) 143.31 (p=0.000)Elasticity Score - Exclusions 21 94 (89-99) 34.38 (p=0.000)Measure - Strain ratio 7 0 (0-100) 0.44 (p=0.401)Measure - Size ratio 6 75 (45-100) 8.05 (p=0.009)Measure - Neural network 4 67 (26-100) 6.10 (p=0.024)USE + B-mode 8 97 (94-99) 61.42 (p=0.000)Blinded index test 20 99 (98-99) 139.66 (p=0.000)Level II diagnostic evidence 13 98 (96-99) 90.53 (p=0.000)Level III-2 diagnostic evidence 24 99 (99-100) 261.67 (p=0.000)Prospective design 19 99 (98-99) 177.72 (p=0.000)Retrospective design 7 97 (95-99) 67.29 (p=0.000)Mean age < 50 23 99 (99-100) 243.43 (p=0.000)Mean age 50-69 11 97 (95-99) 71.44 (p=0.000)Device: EUB-8500 20 99 (98-99) 154.23 (p=0.000)Device: Elegra/Antares 6 98 (96-99) 93.06 (p=0.000)aElectrical Impedance ScanningbDigital Infrared Thermal ImagingcUltrasound Elastography
eFigure 1: Percentage of all studies fulfilling individual QUADAS criteria.
eFigure 2: Individual study estimates of sensitivity and specificity for USE using ANN
eFigure 3: Individual study estimates of sensitivity and specificity for USE using length ratio
eFigure 4: Individual study estimates of sensitivity and specificity for USE using strain ratio