- Email: [email protected]
Ultrasound is now better than mammography for the detection of invasive breast cancer
The American Journal of Surgery 188 (2004) 381–385
Scientific paper
Ultrasound is now better than mammography for the detection of invasive breast cancer S.R.C. Benson, M.D., F.R.C.S.*, J. Blue, F.A.S.B.P., K. Judd, M.Med.Rad.D., J.E. Harman, F.R.A.C.S. St. Marks Women’s Health, 10 St. Marks Rd., Remuera, P.O. Box 99455, Auckland, New Zealand Manuscript received May 19, 2004; revised manuscript June 6, 2004 Presented at the Fifth Annual Meeting of the American Society of Breast Surgeons, March 31–April 4, 2004, Las Vegas, Nevada
Abstract Objective: This study investigated the use of ultrasound (US) as a first-line diagnostic tool. Methods: All women attending our breast center underwent bilateral whole-breast US in addition to all other investigations, and results were documented prospectively and preoperatively. Results: Of 796 patients with breast cancer, US was positive in 710 (89%) and mammography in 706 (89%) (P ⫽ not significant). Either US or mammogram was positive in 770 (97%). Of 537 (67%) symptomatic patients, US was positive in 497 (93%) and mammography in 465 (87%). Either US or mammography was positive in 515 (96%). Of 259 (33%) screening patients, 220 (85%) had invasive cancer. US was positive in 195 (89%) and mammography in 203 (92%) (P ⫽ not significant). Either US or mammography was positive in 217 (99%). Of 39 screening patients with ductal carcinoma in situ (5% of all patients), US was positive in 18 (46%) and mammography in 38 (97%). Conclusions: US is significantly better than mammography for detecting invasive breast cancer (92% patients). The combination of US and mammography is significantly better than either modality used alone, together resulting in 9% more breast cancers detected. © 2004 Excerpta Medica, Inc. All rights reserved.
After its introduction in the 1950s and 1960s, physicians struggled for many years to find ultrasonography (US) for an application for US other than a minor adjunct role in the investigation of breast disease. The resolution of the original equipment resulted in diagnostic accuracy far less than the “gold standard” of mammography [1,2]. Its primary use at introduction, therefore, was to distinguish between cystic and solid lesions, thus allowing the aspiration of the former and preventing unnecessary surgery. Increasing confidence with the needle and increasing resolution of US machines expanded the scope of US to more accurately guide diagnostic biopsies [3] and measure tumors [4]. Until recently, these supporting roles remained the principle use of US in breast disease [5]. Further progress in machine resolution and software, the advent of the 6- to 13-MHz probe, and ever-increasing operator skill has led during the last decade to the develop-
* Corresponding author. Tel.: ⫹64-9520-0389; fax: ⫹64-9520-0589. E-mail address: [email protected]
ment of certain US-based diagnostic criteria that enable the distinction of benign from malignant lesions with improving accuracy [6 – 8]. These results, combined with the complete lack of radiation and the maintenance of resolution even in dense breasts, has led to the latest recommendation by the American College of Radiologists that US be used as a first-line investigative tool for palpable masses in pregnant women and women ⬍30 yrs old at high risk of breast cancer [9]. The same guidelines state that US is not currently indicated as a screening tool for microcalcifications. Beyond this small patient population, several studies have now shown a diagnostic accuracy for US approaching or exceeding that of mammography [10 –15]. The majority of these studies, however, appear to have one or both of the following biases: 1. Many studies have used US as the last investigation with the lesion site and diagnosis often already known, i.e., directed US. This introduces a bias that could make the sensitivity of US falsely high. 2. Other studies have only compared US with mammography in younger patients or in those with dense
0002-9610/04/$ – see front matter © 2004 Excerpta Medica, Inc. All rights reserved. doi:10.1016/j.amjsurg.2004.06.032
382
S.R.C. Benson et al. / The American Journal of Surgery 188 (2004) 381–385
breasts. This introduces bias that would make the sensitivity of mammography falsely low overall. Both of these biases would have the effect of improving the results of US when compared with mammography. Many studies also have relatively small numbers of cancers detected, thus making subgroup analysis of diagnostic accuracy virtually impossible. Despite this potential bias, we believed that these studies still detailed a sufficiently improved accuracy for US to warrant its use as a first-line diagnostic tool, and thus New Zealand’s first dedicated breast physician was trained at our breast center in October 1993. The skills of this new “diagnostic breast doctor” were coupled with state-of-the-art US equipment to be used as both a diagnostic and screening tool, and all women were offered bilateral whole-breast US (WBUS) in addition to all other routine investigations. The study presented here thus decreased the previously mentioned biases by using US as a primary diagnostic tool, on an equal basis with mammography, for all patients attending our center.
Methods The majority of new patients attending St. Marks Women’s Health have their initial consultation with a breast physician. Breast physicians in New Zealand now undergo a minimum 3-year full-time training program before qualification, and those at our center perform 1500 to 2000 diagnostic bilateral WBUS examinations/y. The breast physician obtained a full history, performed a clinical breast examination, and then performed WBUS. Both breasts were scanned twice by commencing in the axilla and utilising a clockwise, sequential, overlapping radial approach [16]. When a lesion was identified, hard copy images—including transverse and longitudinal views, were taken. If not already performed, all patients ⱖ30 years old underwent mammography, which was subsequently double read by two radiologists specializing in breast imaging and not involved in the patients’ clinical assessment. Patients were termed “symptomatic” if their referral was driven by any breast symptom. Patients were considered to be “screening” patients if their breasts were completely asymptomatic. This was a heterogeneous group and included both high- and low-risk patients. US was used for both groups of patients, not as a study tool, but as a fully operational diagnostic tool with clinical decisions made based on its findings in conjunction with CBE and mammography. US and mammography were scored on a 5-point system and prefixed with “U” or “M” [11,17]: 1 ⫽ no abnormality detected; 2 ⫽ benign changes; 3 ⫽ abnormal and probably benign; 4 ⫽ suspicious and probably malignant; and 5 ⫽ malignant. Indeterminate US or mammography findings were recorded as “probably benign” or grade 3. Occasionally ad-
ditional imaging downgraded these lesions to grade 2 or “benign”; otherwise all suspicious lesions grades 3 through 5 were further investigated, usually by core biopsy (14-g magnum biopsy system, Bard, Covington, Georgia) using US guidance or mammogram stereotactic guidance as appropriate. Less commonly, prone-table mammotome, fineneedle aspiration cytology, or wire-guided excision biopsy were employed if necessary. All US examinations were performed using 1 of 4 dedicated machines (LOGIQ 400, GE Medical Systems, Milwaukee, Wisconsin) with an LA 39, 6- to 13-MHz linear probe with multifrequency capability and Doppler. All mammographic examinations were performed with dedicated mammography machines (2⫻ senographe DMR or 2⫻ senographe 800T, GE Medical Systems) calibrated to a film–screen combination and using a dedicated processor (X-OMAT multiloader 300plus, Kodak, Rochester, New York) and dedicated mammography cassettes (MIN R 2000, Kodak). Mammotome biopsy examinations were undertaken using a prone table (digital Mammotest plus/S and Mammovision plus/EMC biopsy system, Fischer Imaging Corp, Denver, Colorado) and an 11-g mammotome stereotactic probe (Ethicon Endo-Surgery, Cincinnati, Ohio). The results of the clinical assessments, WBUS (U scores), mammography (M scores), and tissue biopsy, and surgical specimens were collected prospectively. For patients with breast cancer, all data were transferred to the departmental breast cancer database. This study covered the 6-year period from 1997 to 2002 inclusive and included those new patients who had a clinical examination by a breast physician and WBUS at their initial consultation before to any tissue biopsy. All data for the study were collected from the department’s breast cancer database and cross-checked against the patients’ notes, radiology results, or histopathology results as necessary. Either US or mammography was considered to be positive if the U or M score was 3, 4, or 5 and to be negative if the score was 1 or 2. Of 2312 patients undergoing a surgical resection for breast cancer during this period, 834 fulfilled the above criteria, and of these data were available for 802 patients. Six of these had a non– breast cancer primary in the breast, such as lymphoma, leaving 796 patients suitable for analysis. The major reason for patient exclusion was having a mammogram performed elsewhere before patient referral— either organized by the general practitioner for symptomatic patients or performed by the National Screening Program for asymptomatic patients–with tissue biopsy already performed. The hematoma produced was easily visible on US and thus introduced bias. Statistical analyses were performed using a 2 ⫻ 2 contingency table and Fisher’s Exact test (GraphPad InStat version 3.05, GraphPad Software, San Diego, California). P ⬍0.05 was considered significant.
S.R.C. Benson et al. / The American Journal of Surgery 188 (2004) 381–385
383
Table 1 US and mammography positivity for patients with breast cancer Total
No. of patients US ⫹ VE (%) M ⫹ VE (%) US vs. M US or M ⫹ VE (%) US or M vs. US US or M vs. M
796 714 (90) 710 (89) P ⫽ NS 774 (97) P ⬍0.0001 P ⬍0.0001
Symptomatic
Screening
Invasive
In situ
Invasive
In situ
516 488 (95) 453 (88) P ⫽ 0.0002 502 (97) P ⫽ 0.0394 P ⬍0.0001
21 9 (43) 12 (57) P ⫽ NS 13 (62) P ⫽ NS P ⫽ NS
220 195 (89) 203 (92) P ⫽ NS 217 (99) P ⬍0.0001 P ⫽ 0.0021
39 18 (46) 38 (97) P ⬍0.0001 38 (97) P ⬍0.0001 P ⫽ NS
M ⫽ mammography; US ⫽ ultrasonography.
Results Of the 796 patients with proven breast cancer proceeding to surgical resection, 537 (67%) were symptomatic patients, and 259 (33%) were screening patients. Ultrasound and mammography showed equal efficacy overall; however, their combination was significantly better than either modality used alone (P ⬍0.0001, Table 1). Of the 537 symptomatic patients, US was positive in 497 (93%), and mammography was positive in 465 (87%) (P ⫽ 0.0019). Either US or mammography was positive in 515 (96%) patients, 18 more than US alone and 50 more than mammography alone (P ⫽ 0.0301 compared with US alone and P ⬍0.0001 compared with mammography alone), yielding a 10% diagnostic increase over mammography. Of the 259 screening patients, US was positive in 213 (82%) and mammography was positive in 241 (93%) (P ⫽ 0.0003. Either US or mammogram was positive in 255 (99%) patients, 42 more than US alone and 14 more than mammography alone (P ⬍0.0001 compared with US alone and P ⫽ 0.0036 compared with mammography alone), yielding a 6% increase over mammography. Histology of these 796 patients revealed that 604 (76%) had invasive ductal carcinoma, 60 (8%) had ductal carcinoma in situ (DCIS), 98 (12%) had invasive lobular carcinoma, 16 (2%) had a mucinous carcinoma, 7 (1%) had a papillary carcinoma, 6 (1%) had a tubular carcinoma, and 5 (1%) had a medullary carcinoma. In-situ disease comprised 4% of the symptomatic patients and 15% of the screening patients (P ⬍0.0001). Of the 39 screening patients with in-situ disease, US was positive in 18 (46%), and mammography was positive in 38 (97%) (P ⬍0.0001). Either US or mammography was also positive in 38 (97%) (P ⬍0.0001 compared with US and P ⫽ NS compared with mammography) (Table 1). Overall, for all 60 patients with in-situ breast cancer, US was positive in 27 (45%), and mammography was positive in 50 (83%) (P ⬍0.0001). Either US or mammography was positive in 51 (85%) patients (P ⬍0.0001 compared with US and P ⫽ NS compared with mammogram). Thus, there was no difference in the detection of in-situ disease in symptomatic patients between US and mammog-
raphy; however, mammography was significantly more sensitive for the detection of in-situ disease in asymptomatic patients. The imaging negative DCIS patients comprised 6 with nipple discharge and 3 with palpable thickening. Invasive disease Of the 516 symptomatic patients who had an invasive breast cancer, US was positive in 488 (95%), and mammography was positive in 453 (88%) (P ⫽ 0.0002). Either US or mammography were positive in 502 (97%) patients (P ⫽ 0.0394 compared with US alone and P ⬍0.0001 compared with mammography alone) (Table 1). Of the 220 screening patients with invasive breast cancer, US was positive in 195 (89%) and mammography was positive in 203 (92%) (P ⫽ not significant [NS]. Either US or mammography were positive in 217 (99%) patients (P ⬍0.0001 compared with US alone and P ⫽ 0.0021 compared with mammography alone) (Table 1). Overall, for all 736 patients with invasive breast cancer, US was positive in 683 (93%), mammography was positive in 656 (89%; P ⫽ 0.0178), and either test was positive in 719 (98%) patients (P ⬍0.0001 compared with US or mammography alone). Thus, for symptomatic patients and overall for all patients, US was significantly more sensitive than mammography for the detection of invasive breast cancer. In addition, the combination of US and mammography was significantly better for the detection of invasive breast cancer than either modality used alone. There was no difference in invasive breast cancer detection rates for US or mammography when applied to an asymptomatic screening population. The 17 imaging-negative patients comprised 15 patients with a palpable lump, 1 with a palpable thickening, and 1 with nipple changes. In-situ disease Of the 21 symptomatic patients who had in-situ breast cancer, US was positive in 9 (43%), and mammography was positive in 12 (57%) (P ⫽ NS). Either US or mammography was positive in 13 (62%) (P ⫽ NS).
384
S.R.C. Benson et al. / The American Journal of Surgery 188 (2004) 381–385
Comments Mammography has been the “gold standard” in breast cancer detection for ⬎40 years. Limitations in its ability to detect both small and lobular breast cancers, poor resolution in dense breasts, and a lack of significant improvement in cancer detection, despite digital mammography and computer-aided diagnosis, has inevitably lead to a search for other modalities to improve the detection of breast cancer. The advent of the 6- to 13-MHz dedicated breast probe, coupled with the specialty of breast medicine, has lead to the development of breast US as we know it today and as was practiced in our breast center throughout the study period. During this time, more than two thirds of our breast cancer patients were symptomatic at the time of referral. Used as a primary investigative tool in this group of patients, our study demonstrated that US is now significantly more sensitive than mammography. One third of our breast cancer patients were asymptomatic screening or “anxiousreassurance” patients, of whom 85% had invasive breast cancer. In these screening patients with invasive breast cancer, US and mammography yielded equal sensitivity. The only group in whom mammography was superior to US was in asymptomatic patients with DCIS, who comprised 5% of the study population. In this group, mammography is still hugely superior to US because of the lack of soft tissue changes and a reliance on microcalcifications. Mammography remains the investigative study of choice for this small group of patients [18], and thus the role of US in screening requires further investigation. We suggest a trial of bilateral WBUS for all mammographic abnormalities recalled within a screening program. We have demonstrated that US is significantly better than mammography at detecting invasive malignancy (92.5% of patients) and is inferior to mammography in the detection of in-situ disease (7.5% of patients). Overall, US is now sufficiently superior to mammography for the vast proportion of patients in whom it must be considered a first-line diagnostic and screening partner, standing alongside mammography, in dedicated breast centers. It should be considered as an extension of the examining clinician’s fingers and be routine practice in all breast clinics, particularly “one-stop” breast clinics, where it allows for the immediate assessment and biopsy of any lesion of concern. With improving 3-dimensional assessment and computeraided diagnosis, the sensitivity for US is likely to increase further [8,19 –21]. One of the most important findings of this study is that the combination of US and mammography is significantly more sensitive than either modality used alone. Its ease of use, relative low cost, lack of additional radiation, acceptability to the patient and, ability to tissue sample for diagnostic and therapeutic purposes, make this combination— not computed tomography or magnetic resonance imaging—the new “gold standard” in breast cancer imaging. Using both US and mammography results in 9% more
breast cancers detected than using mammography alone. With 8 or 9 breast cancers missed in every 100 patients, “triple assessment” is no longer adequate for the investigation of breast disease [11,13,14]. Best practice in breast cancer detection therefore dictates that US and mammography must be used together, as part of a “quadruple assessment,” in all breast clinics. The increased diagnostic accuracy afforded by such a quadruple assessment benefits the patient by improving breast cancer detection and decreasing patient uncertainty and anxiety. In today’s increasingly litigious society, the improved diagnostic accuracy of the quadruple assessment has an additional benefit. Undertaking the most accurate assessment available not only decreases the possibility of missed diagnoses, it also decreases any criticism associated with such rare missed diagnoses. In fact, in an extremely hostile medicolegal environment, there has not been a single successful claim for a missed cancer diagnosis at our breast center. Limitations of this study included the inability to calculate the exact false-negative and false-positive rates for US without long-term longitudinal data. It has been generally accepted since the 1980s that the false-negative rate for mammography is relatively low, i.e., 5% to 15% [22–24], although it may actually be as high as 15% to 30% [8,10,11,13,14]. Bilateral WBUS detects a proportion of those tumors missed by mammogram. Our findings of a false-negative rate for US of 11% for all patients and 7.7% for symptomatic patients compares favorably with both our figures for mammography (11.5% for all patients and 13.4% for symptomatic patients) and with the literature previously mentioned. There could be a concern that although the addition of breast US to mammography might increase the sensitivity of breast assessment, it could decrease the specificity, thus leading to an increase in false-positive results and an increase in unnecessary biopsies. To assess this, we prospectively studied our biopsy rate. Of our last 1906 core biopsies performed, 445 were malignant, and 1461 were benign, yielding a ration of 1:3.28. This ratio compares favorably with reports in the literature [12]. The addition of bilateral WBUS to mammography increases diagnostic sensitivity without increasing unnecessary biopsies when performed in a specialty breast center using state-of-the-art equipment and staff. In addition to studying the rate of false-positive results, we believe future studies should also address the nature of US false-positive results compared with mammography. With the increased use of US, we have noticed a different pattern emerging. For all screening patients and symptomatic patients not attending a one-stop breast clinic, a mammographic false-positive results entails up to a 2-week, anxiety-provoking delay followed by at least 1 further clinic attendance and additional imaging. The majority of falsepositive results for US, however, comprise indeterminate lesions that are either aspirated or biopsied as a real-time assessment at the patient’s initial consultation. The result
S.R.C. Benson et al. / The American Journal of Surgery 188 (2004) 381–385
can then be telephoned to the patient within days with no recall and no additional imaging, thus considerably decreasing the inconvenience and stress associated with a mammographic false-positive result. A further concern could be that mammography-negative tumors detected by US might be large enough to be palpable, thus negating the benefit of US. The average size of tumors, in this study that were visible on both US and mammography was 19.3 mm. The size of US-negative but mammography-positive tumors was 16 mm, but the size of US-positive and mammography-negative tumors was 15.5mm. In summary, the addition of bilateral WBUS for all patients undergoing triple assessment for breast disease results in significantly improved diagnostic accuracy for breast cancer. The sensitivity of using both US and mammography is higher, the tumor size is smaller, the falsenegative rate is lower, the false-positive rate is probably similar, and the distress of a false-positive result is likely to be decreased when compared with mammography alone. These patient-oriented benefits are also reflected in a medicolegal benefit attained by providing the most accurate assessment currently available, i.e., that of a clinical assessment, both US and mammography, and a histologic or cytologic biopsy. This quadruple assessment is the new “gold standard” in the investigation of breast disease.
References [1] Sickles E, Filly R, Callen P. Breast cancer detection with sonography and mammography: comparison using state-of-the-art equipment. Am J Roentgenol 1983;140:843. [2] Kopans D, Meyer J, Lindfors K. Whole-breast US imaging: four year follow-up. Radiology 1985;157:505. [3] Shah V, Raju U, Chitale D, et al. False-negative core biopsies of the breast: an analysis of clinical, radiological and pathologic findings in 27 consecutive cases of missed breast cancer. Cancer 2003;97:1824 – 31. [4] Madjar H, Ladner H, Sauerbrei W, et al. Preoperative staging of breast cancer by palpation, mammography and high-resolution ultrasound. Ultrasound Obstet Gynecol 2003;3:185–90. [5] Sainsbury R. Breast surgery, in Farndon J, editor. A Companion to Specialist Surgical Practice. 2nd ed. London, England, Saunders; 2001.
385
[6] Rotten D, Levaillant J. The value of ultrasonic examination to detect and diagnose breast carcinomas. Ultrasound Obstet Gynecol 1992;2: 203–14. [7] Berg W, Campassi C, Loffe O. Cystic lesions of the breast: sonographic-pathological correlation. Radiology 2003;227:183–91. [8] Rotten D, Levaillant J, Constancis E, et al. Three-dimensional imaging of solid breast tumours with ultrasound: preliminary data. Ultrasound Obstet Gynecol 2003;1:384 –90. [9] American College of Radiology: ACR practice guidelines for the performance of a breast ultrasound examination. Available at: http:// www.acr.org. [10] Leconte I, Feger C, Galant C, et al. Mammography and subsequent whole-breast sonography of nonpalpable breast cancers: the importance of radiologic breast density. Am J Roentgenol 2003;180: 1675–9. [11] Flobbe K, Bosch A, Kessels A, et al. The additional diagnostic value of ultrasonography in the diagnosis of breast cancer. Arch Intern Med 2003;163:1194 –9. [12] Buchberger W, Niehoff A, Obrist P, et al. Clinically and mammographically occult lesions: detection and classification with high resolution sonography. Semin Ultrasound CT MR 2001;21:325–36. [13] Zonderland H, Coerkamp E, Hermans J, et al. Diagnosis of breast cancer: contribution of US as an adjunct to mammography. Radiology 1999;213:413–22. [14] Kolb T, Lichy J, Newhouse J. Comparison of the performance of screening mammography, physical examination and breast US and evaluation of factors that influence them: an analysis of 27,825 patient evaluations. Radiology 2002;225:165–75. [15] Goergian-Smith D, Taylor K, Madjar H, et al. Sonography of palpable breast cancer. J Clin Ultrasound 2000;28:211– 6. [16] Blue J, McIntyre H, Harman J, et al. Breast ultrasound practices at a specialist breast centre. ASUM Bull 2001. [17] Mendelson E, Berg W, Merritt C. Towards a standardized breast ultrasound lexicon, BI-RADS: ultrasound. Semin Roentgenol 2001; 36:217–25. [18] Marini C, Traino C, Cilotti A, et al: Differentiation of benign and malignant breast microcalcifications: mammography versus mammography-sonography combination. Radiol Med 2003;105:17–26. [19] Kedar R, Cosgrove D, McCready V, et al. Microbubble contrast agent for color doppler US: effect on breast masses. Radiology 1996;198: 679 – 86. [20] Chen D, Chang R, Chen W, et al. Computer-aided diagnosis for 3-dimensional breast ultrasonography. Arch Surg 2003;138:296 –302. [21] Pavic D, Koomen M, Kuzmiak C, et al. Ultrasound in the management of breast disease. Curr Womens Health Rep 2003;3:156 – 64. [22] Baines C, Miller A, Wall C, et al. Sensitivity and specificity of first screen mammography in the Canadian National Breast Screening study: a preliminary report from five centres. Radiology 1986;160: 295– 8. [23] Homer M. Breast imaging: pitfalls, controversies and some practical thoughts. Radiol Clin North Am 1985;23:459 –72. [24] Kopans D, Meyer J, Sadowsky N. Breast imaging. N Engl J Med 1984;310:960 –7.
Scientific paper
Ultrasound is now better than mammography for the detection of invasive breast cancer S.R.C. Benson, M.D., F.R.C.S.*, J. Blue, F.A.S.B.P., K. Judd, M.Med.Rad.D., J.E. Harman, F.R.A.C.S. St. Marks Women’s Health, 10 St. Marks Rd., Remuera, P.O. Box 99455, Auckland, New Zealand Manuscript received May 19, 2004; revised manuscript June 6, 2004 Presented at the Fifth Annual Meeting of the American Society of Breast Surgeons, March 31–April 4, 2004, Las Vegas, Nevada
Abstract Objective: This study investigated the use of ultrasound (US) as a first-line diagnostic tool. Methods: All women attending our breast center underwent bilateral whole-breast US in addition to all other investigations, and results were documented prospectively and preoperatively. Results: Of 796 patients with breast cancer, US was positive in 710 (89%) and mammography in 706 (89%) (P ⫽ not significant). Either US or mammogram was positive in 770 (97%). Of 537 (67%) symptomatic patients, US was positive in 497 (93%) and mammography in 465 (87%). Either US or mammography was positive in 515 (96%). Of 259 (33%) screening patients, 220 (85%) had invasive cancer. US was positive in 195 (89%) and mammography in 203 (92%) (P ⫽ not significant). Either US or mammography was positive in 217 (99%). Of 39 screening patients with ductal carcinoma in situ (5% of all patients), US was positive in 18 (46%) and mammography in 38 (97%). Conclusions: US is significantly better than mammography for detecting invasive breast cancer (92% patients). The combination of US and mammography is significantly better than either modality used alone, together resulting in 9% more breast cancers detected. © 2004 Excerpta Medica, Inc. All rights reserved.
After its introduction in the 1950s and 1960s, physicians struggled for many years to find ultrasonography (US) for an application for US other than a minor adjunct role in the investigation of breast disease. The resolution of the original equipment resulted in diagnostic accuracy far less than the “gold standard” of mammography [1,2]. Its primary use at introduction, therefore, was to distinguish between cystic and solid lesions, thus allowing the aspiration of the former and preventing unnecessary surgery. Increasing confidence with the needle and increasing resolution of US machines expanded the scope of US to more accurately guide diagnostic biopsies [3] and measure tumors [4]. Until recently, these supporting roles remained the principle use of US in breast disease [5]. Further progress in machine resolution and software, the advent of the 6- to 13-MHz probe, and ever-increasing operator skill has led during the last decade to the develop-
* Corresponding author. Tel.: ⫹64-9520-0389; fax: ⫹64-9520-0589. E-mail address: [email protected]
ment of certain US-based diagnostic criteria that enable the distinction of benign from malignant lesions with improving accuracy [6 – 8]. These results, combined with the complete lack of radiation and the maintenance of resolution even in dense breasts, has led to the latest recommendation by the American College of Radiologists that US be used as a first-line investigative tool for palpable masses in pregnant women and women ⬍30 yrs old at high risk of breast cancer [9]. The same guidelines state that US is not currently indicated as a screening tool for microcalcifications. Beyond this small patient population, several studies have now shown a diagnostic accuracy for US approaching or exceeding that of mammography [10 –15]. The majority of these studies, however, appear to have one or both of the following biases: 1. Many studies have used US as the last investigation with the lesion site and diagnosis often already known, i.e., directed US. This introduces a bias that could make the sensitivity of US falsely high. 2. Other studies have only compared US with mammography in younger patients or in those with dense
0002-9610/04/$ – see front matter © 2004 Excerpta Medica, Inc. All rights reserved. doi:10.1016/j.amjsurg.2004.06.032
382
S.R.C. Benson et al. / The American Journal of Surgery 188 (2004) 381–385
breasts. This introduces bias that would make the sensitivity of mammography falsely low overall. Both of these biases would have the effect of improving the results of US when compared with mammography. Many studies also have relatively small numbers of cancers detected, thus making subgroup analysis of diagnostic accuracy virtually impossible. Despite this potential bias, we believed that these studies still detailed a sufficiently improved accuracy for US to warrant its use as a first-line diagnostic tool, and thus New Zealand’s first dedicated breast physician was trained at our breast center in October 1993. The skills of this new “diagnostic breast doctor” were coupled with state-of-the-art US equipment to be used as both a diagnostic and screening tool, and all women were offered bilateral whole-breast US (WBUS) in addition to all other routine investigations. The study presented here thus decreased the previously mentioned biases by using US as a primary diagnostic tool, on an equal basis with mammography, for all patients attending our center.
Methods The majority of new patients attending St. Marks Women’s Health have their initial consultation with a breast physician. Breast physicians in New Zealand now undergo a minimum 3-year full-time training program before qualification, and those at our center perform 1500 to 2000 diagnostic bilateral WBUS examinations/y. The breast physician obtained a full history, performed a clinical breast examination, and then performed WBUS. Both breasts were scanned twice by commencing in the axilla and utilising a clockwise, sequential, overlapping radial approach [16]. When a lesion was identified, hard copy images—including transverse and longitudinal views, were taken. If not already performed, all patients ⱖ30 years old underwent mammography, which was subsequently double read by two radiologists specializing in breast imaging and not involved in the patients’ clinical assessment. Patients were termed “symptomatic” if their referral was driven by any breast symptom. Patients were considered to be “screening” patients if their breasts were completely asymptomatic. This was a heterogeneous group and included both high- and low-risk patients. US was used for both groups of patients, not as a study tool, but as a fully operational diagnostic tool with clinical decisions made based on its findings in conjunction with CBE and mammography. US and mammography were scored on a 5-point system and prefixed with “U” or “M” [11,17]: 1 ⫽ no abnormality detected; 2 ⫽ benign changes; 3 ⫽ abnormal and probably benign; 4 ⫽ suspicious and probably malignant; and 5 ⫽ malignant. Indeterminate US or mammography findings were recorded as “probably benign” or grade 3. Occasionally ad-
ditional imaging downgraded these lesions to grade 2 or “benign”; otherwise all suspicious lesions grades 3 through 5 were further investigated, usually by core biopsy (14-g magnum biopsy system, Bard, Covington, Georgia) using US guidance or mammogram stereotactic guidance as appropriate. Less commonly, prone-table mammotome, fineneedle aspiration cytology, or wire-guided excision biopsy were employed if necessary. All US examinations were performed using 1 of 4 dedicated machines (LOGIQ 400, GE Medical Systems, Milwaukee, Wisconsin) with an LA 39, 6- to 13-MHz linear probe with multifrequency capability and Doppler. All mammographic examinations were performed with dedicated mammography machines (2⫻ senographe DMR or 2⫻ senographe 800T, GE Medical Systems) calibrated to a film–screen combination and using a dedicated processor (X-OMAT multiloader 300plus, Kodak, Rochester, New York) and dedicated mammography cassettes (MIN R 2000, Kodak). Mammotome biopsy examinations were undertaken using a prone table (digital Mammotest plus/S and Mammovision plus/EMC biopsy system, Fischer Imaging Corp, Denver, Colorado) and an 11-g mammotome stereotactic probe (Ethicon Endo-Surgery, Cincinnati, Ohio). The results of the clinical assessments, WBUS (U scores), mammography (M scores), and tissue biopsy, and surgical specimens were collected prospectively. For patients with breast cancer, all data were transferred to the departmental breast cancer database. This study covered the 6-year period from 1997 to 2002 inclusive and included those new patients who had a clinical examination by a breast physician and WBUS at their initial consultation before to any tissue biopsy. All data for the study were collected from the department’s breast cancer database and cross-checked against the patients’ notes, radiology results, or histopathology results as necessary. Either US or mammography was considered to be positive if the U or M score was 3, 4, or 5 and to be negative if the score was 1 or 2. Of 2312 patients undergoing a surgical resection for breast cancer during this period, 834 fulfilled the above criteria, and of these data were available for 802 patients. Six of these had a non– breast cancer primary in the breast, such as lymphoma, leaving 796 patients suitable for analysis. The major reason for patient exclusion was having a mammogram performed elsewhere before patient referral— either organized by the general practitioner for symptomatic patients or performed by the National Screening Program for asymptomatic patients–with tissue biopsy already performed. The hematoma produced was easily visible on US and thus introduced bias. Statistical analyses were performed using a 2 ⫻ 2 contingency table and Fisher’s Exact test (GraphPad InStat version 3.05, GraphPad Software, San Diego, California). P ⬍0.05 was considered significant.
S.R.C. Benson et al. / The American Journal of Surgery 188 (2004) 381–385
383
Table 1 US and mammography positivity for patients with breast cancer Total
No. of patients US ⫹ VE (%) M ⫹ VE (%) US vs. M US or M ⫹ VE (%) US or M vs. US US or M vs. M
796 714 (90) 710 (89) P ⫽ NS 774 (97) P ⬍0.0001 P ⬍0.0001
Symptomatic
Screening
Invasive
In situ
Invasive
In situ
516 488 (95) 453 (88) P ⫽ 0.0002 502 (97) P ⫽ 0.0394 P ⬍0.0001
21 9 (43) 12 (57) P ⫽ NS 13 (62) P ⫽ NS P ⫽ NS
220 195 (89) 203 (92) P ⫽ NS 217 (99) P ⬍0.0001 P ⫽ 0.0021
39 18 (46) 38 (97) P ⬍0.0001 38 (97) P ⬍0.0001 P ⫽ NS
M ⫽ mammography; US ⫽ ultrasonography.
Results Of the 796 patients with proven breast cancer proceeding to surgical resection, 537 (67%) were symptomatic patients, and 259 (33%) were screening patients. Ultrasound and mammography showed equal efficacy overall; however, their combination was significantly better than either modality used alone (P ⬍0.0001, Table 1). Of the 537 symptomatic patients, US was positive in 497 (93%), and mammography was positive in 465 (87%) (P ⫽ 0.0019). Either US or mammography was positive in 515 (96%) patients, 18 more than US alone and 50 more than mammography alone (P ⫽ 0.0301 compared with US alone and P ⬍0.0001 compared with mammography alone), yielding a 10% diagnostic increase over mammography. Of the 259 screening patients, US was positive in 213 (82%) and mammography was positive in 241 (93%) (P ⫽ 0.0003. Either US or mammogram was positive in 255 (99%) patients, 42 more than US alone and 14 more than mammography alone (P ⬍0.0001 compared with US alone and P ⫽ 0.0036 compared with mammography alone), yielding a 6% increase over mammography. Histology of these 796 patients revealed that 604 (76%) had invasive ductal carcinoma, 60 (8%) had ductal carcinoma in situ (DCIS), 98 (12%) had invasive lobular carcinoma, 16 (2%) had a mucinous carcinoma, 7 (1%) had a papillary carcinoma, 6 (1%) had a tubular carcinoma, and 5 (1%) had a medullary carcinoma. In-situ disease comprised 4% of the symptomatic patients and 15% of the screening patients (P ⬍0.0001). Of the 39 screening patients with in-situ disease, US was positive in 18 (46%), and mammography was positive in 38 (97%) (P ⬍0.0001). Either US or mammography was also positive in 38 (97%) (P ⬍0.0001 compared with US and P ⫽ NS compared with mammography) (Table 1). Overall, for all 60 patients with in-situ breast cancer, US was positive in 27 (45%), and mammography was positive in 50 (83%) (P ⬍0.0001). Either US or mammography was positive in 51 (85%) patients (P ⬍0.0001 compared with US and P ⫽ NS compared with mammogram). Thus, there was no difference in the detection of in-situ disease in symptomatic patients between US and mammog-
raphy; however, mammography was significantly more sensitive for the detection of in-situ disease in asymptomatic patients. The imaging negative DCIS patients comprised 6 with nipple discharge and 3 with palpable thickening. Invasive disease Of the 516 symptomatic patients who had an invasive breast cancer, US was positive in 488 (95%), and mammography was positive in 453 (88%) (P ⫽ 0.0002). Either US or mammography were positive in 502 (97%) patients (P ⫽ 0.0394 compared with US alone and P ⬍0.0001 compared with mammography alone) (Table 1). Of the 220 screening patients with invasive breast cancer, US was positive in 195 (89%) and mammography was positive in 203 (92%) (P ⫽ not significant [NS]. Either US or mammography were positive in 217 (99%) patients (P ⬍0.0001 compared with US alone and P ⫽ 0.0021 compared with mammography alone) (Table 1). Overall, for all 736 patients with invasive breast cancer, US was positive in 683 (93%), mammography was positive in 656 (89%; P ⫽ 0.0178), and either test was positive in 719 (98%) patients (P ⬍0.0001 compared with US or mammography alone). Thus, for symptomatic patients and overall for all patients, US was significantly more sensitive than mammography for the detection of invasive breast cancer. In addition, the combination of US and mammography was significantly better for the detection of invasive breast cancer than either modality used alone. There was no difference in invasive breast cancer detection rates for US or mammography when applied to an asymptomatic screening population. The 17 imaging-negative patients comprised 15 patients with a palpable lump, 1 with a palpable thickening, and 1 with nipple changes. In-situ disease Of the 21 symptomatic patients who had in-situ breast cancer, US was positive in 9 (43%), and mammography was positive in 12 (57%) (P ⫽ NS). Either US or mammography was positive in 13 (62%) (P ⫽ NS).
384
S.R.C. Benson et al. / The American Journal of Surgery 188 (2004) 381–385
Comments Mammography has been the “gold standard” in breast cancer detection for ⬎40 years. Limitations in its ability to detect both small and lobular breast cancers, poor resolution in dense breasts, and a lack of significant improvement in cancer detection, despite digital mammography and computer-aided diagnosis, has inevitably lead to a search for other modalities to improve the detection of breast cancer. The advent of the 6- to 13-MHz dedicated breast probe, coupled with the specialty of breast medicine, has lead to the development of breast US as we know it today and as was practiced in our breast center throughout the study period. During this time, more than two thirds of our breast cancer patients were symptomatic at the time of referral. Used as a primary investigative tool in this group of patients, our study demonstrated that US is now significantly more sensitive than mammography. One third of our breast cancer patients were asymptomatic screening or “anxiousreassurance” patients, of whom 85% had invasive breast cancer. In these screening patients with invasive breast cancer, US and mammography yielded equal sensitivity. The only group in whom mammography was superior to US was in asymptomatic patients with DCIS, who comprised 5% of the study population. In this group, mammography is still hugely superior to US because of the lack of soft tissue changes and a reliance on microcalcifications. Mammography remains the investigative study of choice for this small group of patients [18], and thus the role of US in screening requires further investigation. We suggest a trial of bilateral WBUS for all mammographic abnormalities recalled within a screening program. We have demonstrated that US is significantly better than mammography at detecting invasive malignancy (92.5% of patients) and is inferior to mammography in the detection of in-situ disease (7.5% of patients). Overall, US is now sufficiently superior to mammography for the vast proportion of patients in whom it must be considered a first-line diagnostic and screening partner, standing alongside mammography, in dedicated breast centers. It should be considered as an extension of the examining clinician’s fingers and be routine practice in all breast clinics, particularly “one-stop” breast clinics, where it allows for the immediate assessment and biopsy of any lesion of concern. With improving 3-dimensional assessment and computeraided diagnosis, the sensitivity for US is likely to increase further [8,19 –21]. One of the most important findings of this study is that the combination of US and mammography is significantly more sensitive than either modality used alone. Its ease of use, relative low cost, lack of additional radiation, acceptability to the patient and, ability to tissue sample for diagnostic and therapeutic purposes, make this combination— not computed tomography or magnetic resonance imaging—the new “gold standard” in breast cancer imaging. Using both US and mammography results in 9% more
breast cancers detected than using mammography alone. With 8 or 9 breast cancers missed in every 100 patients, “triple assessment” is no longer adequate for the investigation of breast disease [11,13,14]. Best practice in breast cancer detection therefore dictates that US and mammography must be used together, as part of a “quadruple assessment,” in all breast clinics. The increased diagnostic accuracy afforded by such a quadruple assessment benefits the patient by improving breast cancer detection and decreasing patient uncertainty and anxiety. In today’s increasingly litigious society, the improved diagnostic accuracy of the quadruple assessment has an additional benefit. Undertaking the most accurate assessment available not only decreases the possibility of missed diagnoses, it also decreases any criticism associated with such rare missed diagnoses. In fact, in an extremely hostile medicolegal environment, there has not been a single successful claim for a missed cancer diagnosis at our breast center. Limitations of this study included the inability to calculate the exact false-negative and false-positive rates for US without long-term longitudinal data. It has been generally accepted since the 1980s that the false-negative rate for mammography is relatively low, i.e., 5% to 15% [22–24], although it may actually be as high as 15% to 30% [8,10,11,13,14]. Bilateral WBUS detects a proportion of those tumors missed by mammogram. Our findings of a false-negative rate for US of 11% for all patients and 7.7% for symptomatic patients compares favorably with both our figures for mammography (11.5% for all patients and 13.4% for symptomatic patients) and with the literature previously mentioned. There could be a concern that although the addition of breast US to mammography might increase the sensitivity of breast assessment, it could decrease the specificity, thus leading to an increase in false-positive results and an increase in unnecessary biopsies. To assess this, we prospectively studied our biopsy rate. Of our last 1906 core biopsies performed, 445 were malignant, and 1461 were benign, yielding a ration of 1:3.28. This ratio compares favorably with reports in the literature [12]. The addition of bilateral WBUS to mammography increases diagnostic sensitivity without increasing unnecessary biopsies when performed in a specialty breast center using state-of-the-art equipment and staff. In addition to studying the rate of false-positive results, we believe future studies should also address the nature of US false-positive results compared with mammography. With the increased use of US, we have noticed a different pattern emerging. For all screening patients and symptomatic patients not attending a one-stop breast clinic, a mammographic false-positive results entails up to a 2-week, anxiety-provoking delay followed by at least 1 further clinic attendance and additional imaging. The majority of falsepositive results for US, however, comprise indeterminate lesions that are either aspirated or biopsied as a real-time assessment at the patient’s initial consultation. The result
S.R.C. Benson et al. / The American Journal of Surgery 188 (2004) 381–385
can then be telephoned to the patient within days with no recall and no additional imaging, thus considerably decreasing the inconvenience and stress associated with a mammographic false-positive result. A further concern could be that mammography-negative tumors detected by US might be large enough to be palpable, thus negating the benefit of US. The average size of tumors, in this study that were visible on both US and mammography was 19.3 mm. The size of US-negative but mammography-positive tumors was 16 mm, but the size of US-positive and mammography-negative tumors was 15.5mm. In summary, the addition of bilateral WBUS for all patients undergoing triple assessment for breast disease results in significantly improved diagnostic accuracy for breast cancer. The sensitivity of using both US and mammography is higher, the tumor size is smaller, the falsenegative rate is lower, the false-positive rate is probably similar, and the distress of a false-positive result is likely to be decreased when compared with mammography alone. These patient-oriented benefits are also reflected in a medicolegal benefit attained by providing the most accurate assessment currently available, i.e., that of a clinical assessment, both US and mammography, and a histologic or cytologic biopsy. This quadruple assessment is the new “gold standard” in the investigation of breast disease.
References [1] Sickles E, Filly R, Callen P. Breast cancer detection with sonography and mammography: comparison using state-of-the-art equipment. Am J Roentgenol 1983;140:843. [2] Kopans D, Meyer J, Lindfors K. Whole-breast US imaging: four year follow-up. Radiology 1985;157:505. [3] Shah V, Raju U, Chitale D, et al. False-negative core biopsies of the breast: an analysis of clinical, radiological and pathologic findings in 27 consecutive cases of missed breast cancer. Cancer 2003;97:1824 – 31. [4] Madjar H, Ladner H, Sauerbrei W, et al. Preoperative staging of breast cancer by palpation, mammography and high-resolution ultrasound. Ultrasound Obstet Gynecol 2003;3:185–90. [5] Sainsbury R. Breast surgery, in Farndon J, editor. A Companion to Specialist Surgical Practice. 2nd ed. London, England, Saunders; 2001.
385
[6] Rotten D, Levaillant J. The value of ultrasonic examination to detect and diagnose breast carcinomas. Ultrasound Obstet Gynecol 1992;2: 203–14. [7] Berg W, Campassi C, Loffe O. Cystic lesions of the breast: sonographic-pathological correlation. Radiology 2003;227:183–91. [8] Rotten D, Levaillant J, Constancis E, et al. Three-dimensional imaging of solid breast tumours with ultrasound: preliminary data. Ultrasound Obstet Gynecol 2003;1:384 –90. [9] American College of Radiology: ACR practice guidelines for the performance of a breast ultrasound examination. Available at: http:// www.acr.org. [10] Leconte I, Feger C, Galant C, et al. Mammography and subsequent whole-breast sonography of nonpalpable breast cancers: the importance of radiologic breast density. Am J Roentgenol 2003;180: 1675–9. [11] Flobbe K, Bosch A, Kessels A, et al. The additional diagnostic value of ultrasonography in the diagnosis of breast cancer. Arch Intern Med 2003;163:1194 –9. [12] Buchberger W, Niehoff A, Obrist P, et al. Clinically and mammographically occult lesions: detection and classification with high resolution sonography. Semin Ultrasound CT MR 2001;21:325–36. [13] Zonderland H, Coerkamp E, Hermans J, et al. Diagnosis of breast cancer: contribution of US as an adjunct to mammography. Radiology 1999;213:413–22. [14] Kolb T, Lichy J, Newhouse J. Comparison of the performance of screening mammography, physical examination and breast US and evaluation of factors that influence them: an analysis of 27,825 patient evaluations. Radiology 2002;225:165–75. [15] Goergian-Smith D, Taylor K, Madjar H, et al. Sonography of palpable breast cancer. J Clin Ultrasound 2000;28:211– 6. [16] Blue J, McIntyre H, Harman J, et al. Breast ultrasound practices at a specialist breast centre. ASUM Bull 2001. [17] Mendelson E, Berg W, Merritt C. Towards a standardized breast ultrasound lexicon, BI-RADS: ultrasound. Semin Roentgenol 2001; 36:217–25. [18] Marini C, Traino C, Cilotti A, et al: Differentiation of benign and malignant breast microcalcifications: mammography versus mammography-sonography combination. Radiol Med 2003;105:17–26. [19] Kedar R, Cosgrove D, McCready V, et al. Microbubble contrast agent for color doppler US: effect on breast masses. Radiology 1996;198: 679 – 86. [20] Chen D, Chang R, Chen W, et al. Computer-aided diagnosis for 3-dimensional breast ultrasonography. Arch Surg 2003;138:296 –302. [21] Pavic D, Koomen M, Kuzmiak C, et al. Ultrasound in the management of breast disease. Curr Womens Health Rep 2003;3:156 – 64. [22] Baines C, Miller A, Wall C, et al. Sensitivity and specificity of first screen mammography in the Canadian National Breast Screening study: a preliminary report from five centres. Radiology 1986;160: 295– 8. [23] Homer M. Breast imaging: pitfalls, controversies and some practical thoughts. Radiol Clin North Am 1985;23:459 –72. [24] Kopans D, Meyer J, Sadowsky N. Breast imaging. N Engl J Med 1984;310:960 –7.