Screening with New Modalities: Breast Ultrasound Wendie A. Berg, MD, PhD Professor of Radiology Magee-Womens Hospital of UPMC University of Pittsburgh School of Medicine
Disclosures No personal financial conflicts of interest Philips Healthcare loaning equipment for ultrasound clinical trial
Objectives Describe effect on cancer detection from adding screening US to mammography or tomosynthesis in women with dense breasts Discuss sources of false positives on screening US and ways to reduce them Compare outcomes from different methods of screening breast US
Evidence Supporting Screening Disease-specific mortality reduction Only studied for mammography Reduction in node-positive disease Increase in node-negative invasive cancers Reduction in interval cancers Fewer than 10% of all cancers diagnosed
Failure Analysis Webb ML et al Cancer 2013, epub 9/11/13 7301 invasive breast cancer dx 1990-1999 f/u 2007 609 breast cancer deaths; median age 49 yr at dx 29% ca deaths were among women screened 19% screen detected 10% interval cancers 71% deaths among unscreened women
Interval Cancer Cancer dx by clinical symptoms in interval between recommended screenings Worse prognosis and worse outcome ~1/2 deaths in screened women diagnosed in their 40s are due to interval cancers
Mammography Failure Analysis #1 If not performed at all #2 High-risk women #3 Dense breasts
BI-RADS Density A. Almost entirely fatty B. Scattered fibroglandular density C. Heterogeneously dense which could obscure detection of small masses D. Extremely dense, which lowers the sensitivity of mammography
Breast Density as Function of Age Kerlikowske et al. JNCI 2007; 99:386-395 40% of women of mammographic age have dense breasts
Masking of cancers with increasing breast density Increased risk of developing breast cancer
Interval Cancers and Breast Density Density Odds Ratio 95% CI < 10% 1.0-10-24% 2.1 (0.9, 5.2) 25-49% 3.6 (1.5, 8.7) 50-74% 5.6 (2.1, 15.3) 75% 17.8 (4.8, 65.9) p <.001 Boyd NF, et al. NEJM 2007;356:227-36
Referent Average Pt, Hazard Ratios A B C D Premeno 0.46 1 1.62 2.04 Postmeno no HT 0.57 1 1.35 1.51 Postmeno E+P 0.45 1 1.58 2.09 Kerlikowske K et al J Clin Onc 2010;28:3830-3837
Increased Deaths Chiu SY et al. Cancer Epidemiol Biomarkers Prev 2010;19:1219-28 25 yr f/u Sweden 15,658 women 45-59 12.7% had dense breasts Increased breast cancer mortality with dense breasts RR 1.91 (95%CI 1.26-2.91) Attributed to higher incidence Shorter sojourn time
24 States require some sort of density notification www.densebreast-info.org 3/13/16
Possible tests to add to mammography Modality vs. Mammography alone Absolute Cancer Detection per 1000 screens Clinical breast exam 0.3 Double Read or CAD 1 Tomosynthesis 1-2 Ultrasound 3-4 Molecular Breast Imaging, 7-8 CEDM MRI 10 Copyright Wendie Berg, MD, PhD
Unable to Tolerate MRI: ACRIN 6666 18.5% (1 in 5.4) (95% CI 16.4 to 20.8%) women who had completed 3 years of screening with US and mammography were unable to undergo an MRI Berg WA et al. Radiology 2010;254:79-87
Ultrasound No radiation Not limited by dense tissue No injection of contrast or radioactive material Inexpensive
US to Replace Mammo? Berg WA et al JNCI 2016; 108, epub 12/18/15 111 breast ca dx among 2809 women ACRIN 6666 129 US to detect one cancer, 127 for mammo Of 89 invasive cancers, 53 (60%) seen on US vs. 41 (46%) on mammography, p =.11 More likely node negative when found by US: 34/53 (64%) vs. 18/41 (44%), p =.003
US but not Mammo Inv Ca Detection Density, % US Mammo US, not Mammo 25 0/1 (0) 0/1 (0) 0/1 (0) 26-40 6/10 (60) 6/10 (60) 2/10 (20) 41-60 16/30 (53) 17/30 (57) 8/30 (27) 61-80 22/36 (61) 13/36 (36) 14/36 (39) >80 9/12 (75) 5/12 (42) 6/12 (50) P trend.23.19.06 Berg WA et al JNCI 2016; 108, epub 12/18/15
False Positives Over 3 Years US Mammo P-value Recall Rate 515 (10.7%) 453 (9.4%).03 Biopsy Rate 266 (5.5%) 97 (2.0%) <.001 PPV Biopsies 31/266 (11.7%) 37/97 (38.1%) <.001 Berg WA et al JNCI 2016; 108, epub 12/18/15
US and Mammo Complementary Of 22 DCIS, 18 (82%) seen on mammography vs. 5 (23%) on US, p=.002 Sensitivity of mammography + US 0.76 (0.65-0.85) vs. 0.52 (0.40-0.64) mammo alone (p <.001) Berg WA et al JAMA 2012;307:1394-1404
48F screening Courtesy Dr. Wei Yang, MD Anderson
RT CC MAG RT ML MAG Courtesy Dr. Wei Yang, MD Anderson
Stereotactic biopsy: High nuclear grade DCIS solid type with comedo necrosis, with microinvasion, ER, PR-, HER2 + Skin-sparing mastectomy, 0/4 SLN
Supplemental US Physician Performed Technologist Performed Automated
Handheld US High-frequency transducer, 12-18 MHz linear array Survey scanning transverse and sagittal Document 1 image per quadrant, 1 behind nipple for negative exam Lesions (all studies to date): Orthogonal views ± calipers; optional color or power Doppler image Positive test: BI-RADS 3 or higher assessment, or recommendation for further imaging (BI-RADS 0)
Physician Performed US: Multicenter Results Author N screens ICDR per 1000 Recall Rate (%) Bx Rate (% women) PPV3 Bx Performed Corsetti 9157 4.0 NS 449 (4.9) 50/623 (8.0) Berg yr1 2659 5.3 401 (15.1) 207 (7.8) 14/264 (5.3) Berg yr2-3 4841 3.7 356 (7.4) 242 (5.0) 21/276 (7.6) TOTAL 16,657 4.4 10% 898 (5.4) 85/1163 (7.3) 4.9% of women had biopsies for benign findings
Tech-Performed US (USA): Prevalent Screens Author N ICDR per 1000 Recall Rate (%) Bx Rate (%) PPV3 Bx Performed Kaplan, 2001 1,862 2.7 176 (9.5) 97 (5.2) 6/96 (6.3) Hooley, 2012 648* 4.6 154 (23.8) 46 (7.1) 3/58 (5.2) Weigert, 2012 8,647 2.8 1,196 (13.8) 429 (5.0) 25/418 (6.7) Parris, 2012 5,519 1.8 680 (12.3) 185 (3.3) 10/181 (5.5) Overall 16,676 2.5 2,206 (13.2) 757 (4.5) 47/753 (6.2) *analysis presented for women with negative screening mammograms Berg WA and Mendelson EB. Radiology 2014;272:12-27
Recalls: Tech-Performed HHUS 2,206/16,676 (13.2%) test positive on prevalence screen 1,399 (8.4%) all women BI-RADS 3 757 (4.5%) all women BI-RADS 4 44/753 (5.8%) found to have cancer Only 43/16,676 (0.3%) recalled for additional evaluation (BI-RADS 0) prior to final assessment Berg WA and Mendelson EB. Radiology 2014;272:12-27
Disease Prevalence Affects Yield Moderate Risk* No Known Risks P-value Kolb 2002 14/2914 (4.8 per 1000) 14/7901 (1.8 per 1000).011 Crystal 2003 4/318 (12.5 per 1000) 3/1199 (2.5 per 1000) <.04 Overall 18/3232 (5.6 per 1000) 17/9100 (1.9 per 1000) *Personal hx of breast cancer or first-degree relative with breast cancer vs. no risks
Japan Tohno E et al Breast Cancer 2012;19:138-146 2-day educational program; results of training/testing for 415 technologists and 422 physicians Observers worse with experience < 100 cases Video sensitivity, still image sensitivity, and disease agreement for technologists > for MDs
Node-Negative Invasive Cancers Across 10 series, 475 cancers seen only on US, 415 (87.4%) invasive 273/303 (90.1%) with staging were node negative 22/91 (24%) ILC
By Participant, Yield/1000, ACRIN 6666 Year M+US M Supp. Yield, 95% CI P-value 1 12.8 7.5 5.3 (2.1, 8.4).0001 2 10.0 6.4 3.6 (0.9, 6.4).004 3 13.8 9.9 3.9 (0.9, 6.8).004 Supplemental yield of US is significant each year and similar for incidence and prevalence screens Berg WA et al JAMA 2012;307:1394-404
Weigert: Recalls Incidence Screens Year 1 8647 (12%) women screening US Recall rate 13.8% (n=1196): 767 (8.9%) BR 3; 429 (5.0%) BR 4,5; PPV3 5.6% 24 cancers, CDR 2.8 per 1000 Year 2 10,282 (17.9%) women Recall rate 12.7% (n=1310); CDR 2.3 per 1000 (24) Year 3 4128 (12.8%) women Recall rate 7.7% (n=316); CDR 2.7 per 1000 (11)
Radial Antiradial 60F, 5-yr risk 2.5%, 24-mo US: 12 mm grade 1 IDC-DCIS, N0 Courtesy WP Evans, III, MD
Radial Antiradial 75F personal hx Lt cancer 17 mm grade 3 IDC-DCIS, N0 Seen only on 24-month US Seen in retrospect on mammo Courtesy Gary Whitman, MD, MD Anderson
70F personal hx rt mastectomy, BRCA-1 mutation carrier 24 mo screen US+ 19 mm grade 3 IDC-DCIS, N0 Courtesy Dr. Mary Mahoney, U Cincinnati
ACRIN 6666: Breast Density Density n Yield per 1000 P-value 25% 124 0 26-40% 785 6.4.026 41-60% 2314 3.0.008 61-80% 2807 4.3.0006 >80% 1443 5.5.005 Berg WA, et al., RSNA 2009
Interval Cancer Rate: ACRIN 6666 Yr N Interval N Cancers (%) 1 2 36 5.6 2 4 29 14 3 3 46 6.5 All 9 111 8.1 Interval Ca Rate: 9/7473 screens = 1.2 per 1000 8% of all cancers Berg WA et al JAMA 2012;307:1394-404
Interval Cancer Rate Italy Corsetti V et al Cancer 2011;47:1021-6 Interval cancer rate in fatty breasts 1.0 per 1000 Interval cancer rate in dense breasts after adding screening US 1.1 per 1000
J-START Ohuchi N et al Lancet 2015, epub 11/4/2015 Asymptomatic women aged 40-49 at 42 sites Randomized to M+US or M alone twice in 2 yrs 36,869 to intervention and 36,139 to control group
Results J-START first round Intervention Control P-value Sensitivity 91.1 (87.2-95.0) 77.0 (70.3-83.7).0004 Specificity 87.7 (87.3-88.0) 91.4 (91.1-91.7) <.0001 % Stage 0, I 144/184 (71.3) 79/117 (52.0).019 Interval Cancers 18 (0.05%) 35 (0.10%).034 Ohuchi N et al Lancet 2015, epub 11/4/2015
Time to Perform US: ACRIN 6666 Bilateral scan, not including time discussing results with patient nor creation of report Year Median (min) Mean SD 1 17 19.2 11.9 2 15 16.7 10.4 3 13 14.7 9.2
Reducing False Positives BI-RADS 3 lesions Prevalence of 15-20% of all patients having screening US in prior series (Barr et al; Hooley et al; Chae et al) Across all series, only 1 lesion had suspicious change yielding malignancy at 6-mo follow-up 12-month follow-up reasonable
Orthogonal Views Required for any mass for which future comparison is desirable Not necessary for simple cysts Incomplete characterization without this
RAD ARAD 53F Papillary DCIS with microinvasion Berg WA and Mendelson EB Radiology 2014;262:309-315 Courtesy Dr. Christophe Tourasse
50F invasive ductal carcinoma; echogenic rim in arad view only Berg WA and Mendelson EB Radiology 2014 2014;262:309-315
Cysts ACRIN 6666 1255/2662 (47.1%) women over the three years 998 (37.5%) of 2659 year one 537/1363 (39.4%) post-menopausal participants, had cysts 73 using estrogen replacement 48 (66%) had cysts 1290 no HRT 489 (37.9%) had cysts (p<.0001, less common) 516/793 (65.1%) premenopausal women had cysts (p<.0001) Berg WA, et al Radiol Clin N Amer 2010;48:931-987
Complicated Cysts ACRIN 6666 376 (14.1%) of 2662 participants 301 (80%) had at least one simple cyst 84 (22%) multiple, bilateral Overall 2/475 (0.42%) such lesions malignant Berg WA, et al Radiol Clin N Amer 2010;48:931-987
Complicated Cysts N N Malignant (%) Kolb et al 1998 126 0 Venta et al 1999 308 1 Buchberger et al 1999 133 0 Berg et al 2003 38 0 Chang et al 2007 35 0 Daly et al 2008 228 1 ACRIN 6666 475 2 TOTAL 1343 4 (0.3) Berg WA et al Radiol Clin N Amer 2010,48:931-987
53F incidental finding on US, aspirated, cytology: benign cyst with apocrine cells Cyst or Solid? Radial Antiradial
Radial Antiradial 51F strong FH, incidental finding on US Aspirated to resolution, thick cloudy yellow fluid, cyst
61F with new mass on mammography, prior ipsilateral cancer
Radial Antiradial 12 month follow-up US enlarged: 14-g US-guided bx papillary DCIS
BI-RADS 3 Chae EY et al AJR 2016;206:666-672 With mammographic abnormality, 4/184 (2.2%) malignant Without mammographic abnormality, 4/980 (0.4%) malignant (p=.025)
Clustered Microcysts 3.9 to 5.8% of US examinations 1/235 (0.4%) malignant across 5 series Mean age 48 years (32-71) Short-interval follow-up if uncertainty Caution if new mass on mammogram, post-menopausal woman not on HRT May merit biopsy Berg WA AJR 2005;185:952 Berg WA, et al Radiol Clin N Amer 2010;48:931-987
48F new mass on screening mammogram
60F ipsilateral cancer elsewhere US-guided core biopsy DCIS, intermediate grade
Lesions Synchronous to New Cancer Kim SJ et al AJR 2008;191:653-8 55/482 (11.4%) BI-RADS 3 lesions malignant 36/170 (21.2%) in same quadrant as 1º 12/122 (9.8%) in different quadrant 8/190 (4.2%) in contralateral breast
M-B Circumscribed Masses: US Berg WA et al Radiology 2013:268:673-683 2172 women in ACRIN 6666 135 (6.2%) participants had 153 unique findings described as M-B masses on screening US over 3 annual screens 98 complicated cysts with debris 43 solid, circumscribed, oval masses 7 solid masses with 2-3 lobulations 5 clustered microcysts No malignancies (95%CI up to 2.4%)
Billing CPT codes 76641, unilateral complete right 76641, unilateral complete left Medicare reimbursement averages $165 Subject to deductible and copays
Billing ICD-10 92.2 Inconclusive mammogram Applicable to dense breasts, NOS Inconclusive mammogram due to dense breasts http://www.icd10data.com/icd10cm/codes/r00-r99/r90-r94/r92-/r92.2
Automated Arm US A Tower B Y-axis Gantry & Transducer Carrier C X-axis Gantry D Ultrasound Machine Monitor E Touch Screen / Monitor F Transducer Holster G Patient Bed
Automated Arm Results Kelly KM et al Eur Radiol 2010; 20:734-742 4419 women, 6425 exams, 8 facilities 40% women at intermediate risk 23 cancers mammography 46 cancers M+US Supplemental yield 3.6 per 1000 (95% CI 2.3 to 5.4) 10% recall rate 23/75 (31%) biopsies showed cancer
12 MHz Automated Breast US 15 cm footprint 3 acquisitions per breast in ~15 minutes 3D dataset Transverse Created coronal and sagittal displays
ABUS Results Brem RF et al Radiology 2015;273:663-673 15,318 women BI-RADS 1 or 2 mammo, dense breasts, automated whole breast US 30 (2/1000) cancers only by ABUS 25 detailed: 23 (92%) invasive, mean size 13 mm, 18 (78%) of those N0 20/23 (87%) ER+ 3/22 (14%) stage IIB or higher 13% absolute increase in recall rate immediate additional evaluation, not a final assessment
Time to acquire images HHUS vs. AUS HHUS 13 min (but range up to 90) Training to Yes, months, technologist AUS 15 min Minimal Sensitivity ~85% ~74% Number of images 5-20 1000-1700 Time to interpret < 30 sec 5-10 min Recalls 13% Final assessment typically rendered 13% Incomplete, needs targeted US Interobserver Variability Κ = 0.53 (SE 0.02) Κ = 0.04 to 0.50
Is screening ultrasound still of benefit after tomosynthesis?
ASTOUND trial Tagliafico AS et al JCO 2016;epub 3/9/2016 3231 women with dense breasts, negative mammogram, 5 centers in Italy DBT 13 cancers (ICDR 4.0/1000 95%CI 1.8 to 6.2) US 23 cancers (ICDR 7.1/1000, 95%CI 4.2 to 10.0, p=0.006) False positive recall DBT 53 vs. US 65 (p=0.26)
DBTUST study UPMC Pittsburgh UPMC Hamot, Erie Weinstein Imaging 6200 women DBT and technologist-performed screening US each year for three years NIH and PABCC funding
Three-Step Implementation 1) Does the woman have at least 10-yr life expectancy? No, then CBE only, with mammography only if warranted by symptoms
2) Is the patient at high risk for breast cancer and under age 70? Yes, then MRI annually beginning: When ascertained to be high risk Age 25 if BRCA1/2 or other pathogenic mutation 8yr after chest XRT if XRT before age 30 If unable to tolerate MRI, then US
3) Dense? Yes: Supplement annual mammography with US beginning at age 40-45 No: Tomosynthesis beginning at age 40-45
Mam+US MRI Mammo