Breast Biphasic Compression versus Standard Monophasic Compression in X-ray Mammography 1

Francesco Sardanelli, MD Franco Zandrino, MD Andrea Imperiale, MD Emma Bonaldo, MD Maria G. Quartini, RT Nadia Cogorno, RT Breast Biphasic Compression versus Standard Monophasic Compression in X-ray Mammography 1 Index terms: Breast radiography, comparative studies, 00.11 Breast radiography, technology, 00.11 Radiology 2000; 217:576 580 Abbreviations: BC biphasic compression CC craniocaudal MC monophasic compression MLO mediolateral oblique 1 From the Department of Experimental Medicine, Section of Diagnostic Imaging, Genoa University School of Medicine, Genova, Italy (F.S., F.Z., E.B., M.G.Q., N.C.), and the National Institute for Cancer Research, Genova, Italy (A.I.). Received May 6, 1999; revision requested July 16; final revision received January 17, 2000; accepted February 11. Address correspondence to F.S., Department of Radiology, the Biomedical Institute, Via Prà 1b, 16157 Genova, Italy. (e-mail: sardanelli@biomedicalspa.com). RSNA, 2000 Author contributions: Guarantor of integrity of entire study, F.S.; study concepts and design, F.S.; definition of intellectual content, F.S., F.Z.; literature research, F.Z., E.B.; clinical studies, F.S., A.I., M.G.Q., N.C.; data acquisition, F.Z., E.B.; data analysis, F.S., F.Z.; statistical analysis, F.S.; manuscript preparation, editing, and review, F.S., F.Z. Breast biphasic compression (22.5 angled paddle, followed by progressive angle reduction) was compared with standard monophasic compression in x-ray mammography. The presence of the pectoral muscle was recorded for the craniocaudal (CC) view and the presence of the inframammary fold for the mediolateral oblique (MLO) view. The amount of breast in each study and image quality were assessed for both views. For all parameters, biphasic compression performed better than monophasic compression in both CC (P.006) and MLO (P.04) views. The main goal of x-ray mammography is early detection of as much clinically occult breast cancer as possible; this goal requires the highest sensitivity. Good technical performance is essential to obtain high-quality mammograms with optimal tissue visualization and low radiation dose to the patient (1 4). Breast compression is a critical step in mammographic positioning. With standard monophasic compression (MC), the breast is compressed between two parallel paddles while it is held firmly in position by the radiographer (1 3,5,6). The purpose of this study was to compare findings and radiographer performance with MC and biphasic compression (BC), an alternative to MC. Materials and Methods One hundred asymptomatic women referred to our breast unit for periodic mammographic screening were enrolled in our study, which was approved by our institutional review board. The women gave written informed consent and were randomized into one of two groups. The first group of 50 women (age range, 38 76 years; mean age, 54.0 years 10.6 [mean SD]) underwent two craniocaudal (CC) mammographic examinations of the right (n 25) or the left (n 25) breast, one with MC, the other with BC, with a randomized priority of execution. BC was performed as follows: phase one, the compression paddle was angled 22.5 downward; phase 2, the angle was progressively reduced; phase 3, the paddle was parallel to the cassette holder (Fig 1). The second group of 50 women (age range, 38 80 years; mean age, 58.5 years 11.3) underwent two mediolateral oblique (MLO) mammographic examinations of the right (n 25) or the left (n 25) breast, one with MC, the other with BC, with a randomized priority of execution. Angulation for the MLO view ranged from 35 to 55, depending on the patient s axillary conformation. The same mammographic unit (Sophie; Planmed, Helsinki, Finland) was used for both MC and BC mammography. All examinations were performed with automatic exposure control by two trained breast-dedicated radiographers (M.G.Q., N.C.). Each radiographer performed 25 CC and 25 MLO examinations with BC and MC. Compression was always subjectively applied by the radiographer to achieve optimal compression without causing pain. For each mammogram, exposure parameters (radiation exposure and x-ray beam voltage) and thickness of the compressed breast were recorded. To evaluate the CC mammograms, quality parameters included evidence of the pectoral muscle (3 8) and the length of the posterior nipple line (measured by means of manual superimposition of a ruler on mammograms mounted on the light box), extending perpendicularly from 576

Figure 1. Photographs depict mammographic BC with a rubber ball. Left: First phase, the plate is angled 22.5 downward. Middle: Second phase, the angle is progressively reduced. Right: Final phase, the compression paddle is parallel to the cassette holder. TABLE 1 Findings With Breast BC versus Standard MC at X-ray Mammography Findings BC MC Posterior nipple line distance (cm) CC* Mean SD 10.5 2.3 10.2 2.2 Range 6.0 15.3 6.0 14.6 MLO Mean SD 11.0 2.1 10.8 2.1 Range 6.4 15.1 6.2 15.1 Pectoral muscle (n 50) 27 (54) 17 (34) Inframammary fold (n 50) 45 (90) 36 (72) Thickness of compressed breast (cm) CC Mean SD 4.8 1.1 4.7 1.1 Range 2.2 7.2 1.9 7.0 MLO# Mean SD 5.1 1.1 4.8 1.1 Range 2.0 8.2 2.0 8.0 * Difference, 0.35 0.04 (mean standard error); P.001 (Wilcoxon matched pairs signed rank test). Difference, 0.34 0.05; P.002 (Wilcoxon). Data are the number of such findings. Numbers in parentheses are percentages. P.006 (McNemar test). P.022 (McNemar). Difference, 0.20 0.04; difference not significant (Wilcoxon). # Difference, 0.22 0.05; difference not significant (Wilcoxon). the posterior edge of the film to the anterior surface of the nipple (3 5,7). To evaluate the MLO mammograms, quality parameters included the presence of the inframammary fold with distinction between open and closed fold, the former indicating more adequate compression (3 7) and the posterior nipple line, extending perpendicularly from the pectoral muscle (or the border of the film when the pectoral muscle was insufficiently displayed or was not intersected by the posterior nipple line) to the anterior surface of the nipple (3 5,7). For CC and MLO mammograms, a blinded subjective evaluation was performed by consensus by three radiologists (F.S., F.Z., A.I.), who decided if one of the two mammograms had better image quality or if they were equivalent. This qualitative evaluation was based on adequacy of compression (uniform tissue exposure levels, separation of tissues without overlapping, absence of motion artifacts); contrast; adequate exposure; absence of glandular tissue at the posterior border of the film, which indicated the whole gland was included; nipple in profile; and absence of skin folds (4,6). Differences in the performances of the two radiographers were also evaluated. Statistical evaluation of differences between BC and MC mammograms was performed with the Wilcoxon matched pairs signed rank test for the two distance parameters (the posterior nipple line on both CC and MLO views), the exposure parameters (radiation exposure and x-ray beam voltage), and thickness of the compressed breast. The McNemar test was used to test differences in the presence of the pectoral muscle with CC view, the inframammary fold and open versus closed inframammary fold with MLO view, and the subjective assessment of both CC and MLO views. To allow use of the McNemar test (7), the cases were ignored in which results were the same with MC and BC mammograms (presence or absence of the pectoral muscle, inframammary fold, open or closed inframammary fold; equivalent judgement at subjective evaluation). The performances of the two radiographers were compared with the Mann-Whitney U test for the posterior nipple lines with both CC and MLO views and with the 2 and the Fisher exact test for the presence of pectoral muscle with CC view and for the presence of inframammary fold with MLO view (8). Results For the CC view, the pectoral muscle was depicted on 27 (54%) of 50 BC mammograms and on 17 (34%) of 50 MC mammograms (P.006, McNemar test). The length of the posterior nipple line was 6.0 15.3 cm (mean, 10.5 cm 2.3) on BC mammograms and 6.0 14.6 cm (mean, 10.2 cm 2.2) on MC mammograms (P.001, Wilcoxon test); the posterior nipple line was longer on BC than on MC mammograms in 37 women, equal in eight, and longer on MC than on BC mammograms in five. In the subjective evaluation of the 50 CC mammograms, 33 BC mammograms were judged better than MC mammograms, 10 were equivalent, and seven were worse (P.001, McNemar test). In the comparison of exposure parameters, radiation exposure was 40.8 130.0 mas (mean, 69.3 mas 21.8) for MC mammograms and 31.2 133.0 mas (mean, 68.3 mas 23.4) for BC mammograms, and x-ray beam voltage was 24 30 kvp (mean, 27.8 kvp 1.6) for MC mammograms and 24 30 kvp (mean, 27.7 kvp 1.6) for BC mammograms (differences not significant, Wilcoxon test). Thickness of the compressed breast was 1.9 7.0 cm (mean, 4.7 cm 1.1) for MC mammograms and 2.2 7.2 cm (mean, 4.8 cm 1.1) for BC mammograms (differences not significant, Wilcoxon test). An example of the difference between BC and MC mammograms in the CC view is shown in Figure 2. For the MLO view, the length of the posterior nipple line was 6.4 15.1 cm Volume 217 Number 2 Breast Biphasic Compression versus Standard Monophasic Compression 577

(mean, 11.0 cm 2.1) on BC mammograms and 6.2 15.1 cm (mean, 10.8 cm 2.1) on MC mammograms (P.002, Wilcoxon test). The posterior nipple line was longer on BC than on MC mammograms in 29 women, equal in 11, and longer on MC than on BC mammograms in 10. The inframammary fold was depicted on 36 (72%) of 50 MC mammograms and on 45 (90%) of 50 BC mammograms (P.022, McNemar test). At the subjective evaluation, 27 of the 50 BC mammograms were judged better than MC mammograms, 10 equivalent, and 13 worse (P.04, McNemar test). Radiation exposure was 41.5 184.0 mas (mean, 77.8 mas 31.5) for MC mammograms and 43.0 144.0 mas (mean, 77.5 mas 28.4) for BC mammograms (difference not significant, Wilcoxon test), x-ray beam voltage was 25 31 kvp (mean, 28.3 kvp 1.4) for MC mammograms and 25 31 kvp (mean, 28.3 kvp 1.5) for BC mammograms (difference not significant, Wilcoxon test). The thickness of the compressed breast was 2.0 8.0 cm (mean, 5.1 cm 1.1) for MC mammograms and 2.0 8.2 cm (mean, 5.1 cm 1.1) for BC mammograms (difference not significant, Wilcoxon test). An example of the difference between BC and MC mammograms in the MLO view is shown in Figure 3. Data regarding distance parameters for CC and MLO views and visibility of the inframammary fold and pectoral muscle are summarized in Table 1. An open rather than a closed inframammary fold was considered an index of better compression. Of 45 folds depicted on BC mammograms, 23 were open and 22 were closed; of 36 folds depicted on MC mammograms, 16 were open and 20 were closed. Thirty-four open or closed folds were depicted on both BC and MC mammograms; results were concordant between compression modalities in 27 cases (21 open and six closed folds) and discordant in seven (two open on MC mammograms and closed on BC mammograms, five open on BC mammograms and closed on MC mammograms) (difference not significant, McNemar test). No difference was found to be significant between the performances of the two radiographers (Table 2). Discussion To achieve satisfactory visualization of breast tissues, breast positioning is a basic technical factor in x-ray mammography (3 5). Suitable views and optimal breast Figure 2. CC images of the left breast in one patient. (a) With MC, the pectoral muscle (arrows) is depicted, but a spiculated lesion (arrowheads) is only partially visible in the prepectoral area. (b) With BC, a subtle additional amount of the pectoral muscle (arrows) is visible, and a larger portion of the lesion (arrowheads) is depicted. The difference is better appreciated on the magnification images: (c) With MC, an anteroposterior diameter of only 6 mm of the lesion is depicted; (d) with BC, an anteroposterior diameter of 9 mm is depicted. compression are required to study as much of the breast as possible in the best way, by limiting tissue superimposition that may hide lesions (3,4). Two complementary views, CC and MLO, are required for both screening and diagnostic examinations (1,5,6). Additional views may be necessary for lesions not detected or not clearly depicted with the routine views (2,3). When performed best, the CC view should include the anterior profile of the pectoral muscle in at least 30% 40% of mammograms (4). Thus, the presence of the pectoral muscle is a high-quality marker for the CC view, as an indicator of optimal posterior tissue inclusion (3,4, 6,9,10). In this study, the presence of the pectoral muscle in the CC view rose from 34% (17 of 50) on MC mammograms to 54% (27 of 50) on BC mammograms, with a high level of significance, which 578 Radiology November 2000 Sardanelli et al

Figure 3. MLO images of the left breast in one patient. With (a) MC and (b) BC, the inframammary fold (open arrow) is depicted. A larger amount of the whole breast is depicted in b, onthe basis of the positions of the three calcifications (arrowheads) and the round opacity ( ). In addition, a cutaneous fold (solid arrows) is present in the axilla in a. TABLE 2 Performance With BC versus Standard MC at X-ray Mammography Performance BC MC Posterior nipple line distance (cm) CC* Radiographer 1 Mean SD 10.7 2.3 9.9 2.2 Range 6.0 15.0 6.0 14.3 Radiographer 2 Mean SD 10.3 2.3 10.4 2.3 Range 6.1 15.3 6.3 14.6 MLO Radiographer 1 Mean SD 11.0 2.0 10.8 2.2 Range 6.6 14.0 6.2 14.0 Radiographer 2 Mean SD 10.9 2.3 10.8 2.2 Range 6.4 15.0 6.5 15.1 Pectoral muscle (n 25) Radiographer 1 12 (48) 11 (44) Radiographer 2 15 (60) 6 (24) Inframammary fold (n 25) Radiographer 1 22 (88) 15 (60) Radiographer 2 23 (92) 21 (84) * BC, P.449 (not significant [NS]) (Mann-Whitney U test); MC, P.398 (NS). BC, P.899 (NS) (Mann-Whitney); MC, P.712 (NS). BC, P.395 (NS) ( 2 test); MC, P.135 (NS). BC, P.99 (NS) (Fisher exact test); MC, P.059 (NS) ( 2 test). demonstrates a real advantage with BC. In addition, the difference was significantly in favor of BC for the posterior nipple line, another useful quality parameter with which to assess the amount of breast depicted (3 5,9). These results suggest that more breast is included in the CC view with BC, which reduces the possibility of missed lesions. The MLO view is considered the most important and is preferred to the 90 lateral view because more gland is present to be compressed and visualized in the natural oblique long axis (2,3,6). When properly performed, the MLO view shows almost the whole breast including the axillary tail, which is missed in the CC and 90 lateral views (3,6). The intersection between the anterior profile of the pectoral muscle, which must always be depicted, and the posterior border of the film should be at least at the level of the nipple (4 6). Thus, the posterior nipple line is the most useful parameter with which to estimate the amount of breast included in the MLO view (3 5,9). In our study, this distance was significantly longer on BC than on MC mammograms, which demonstrates that BC allows imaging of more breast in the MLO view. Visualization of the inframammary fold in the MLO view ensures that the lower posterior breast tissues, which are not all included in the CC view, have been depicted and that proper compression is applied (3 6,9). In the MLO view, depiction of the inframammary fold rose significantly from 72% (36 of 50) on MC mammograms to 90% (45 of 50) on BC mammograms, which demonstrates the better results with BC. The lack of significance for the open versus closed inframammary folds could be due to the relatively small number of cases. The subjective quality assessment with both CC and MLO views, on the basis of well-known parameters (4,6), demonstrated that mammograms obtained with BC were significantly better than those obtained with MC. In the future, however, receiver operating characteristic studies are necessary for a more accurate analysis. In addition, radiation exposure, x-ray beam voltage, and breast thickness were not significantly influenced by the compression modality; thus, there is no practical drawback to use of BC. What are the reasons for the better performance with BC? The general explanation is that the obliquity of the compression plate allowed earlier and tighter anchoring of the breast, greater access to the patient, and, as a consequence, more effective breast manipulation during the first (angled) compression phase. In fact, radiographers in our study could hold the anterior breast tissues longer with BC than with MC. We do not have a clear explanation for the apparently greater improvement with BC in the CC com- Volume 217 Number 2 Breast Biphasic Compression versus Standard Monophasic Compression 579

pared with the MLO view. We think that the MLO view intrinsically offers fewer opportunities for improvement, owing to the blockage by the axilla for the angle of the cassette holder (ie, two sides) and to the oblique breast position for the radiographer. The absence of significant differences between the performances of the two radiographers with BC indicates that the advantage is real for both, although results by the first radiographer were more improved in the MLO view and by the second radiographer in the CC view. Finally, in our experience, BC did not require more time than MC. Moreover, no woman reported particular breast pain due to BC. To summarize, BC is a good technical improvement for x-ray mammography that allows the study of more of the breast with good image quality. BC should be tested with symptomatic women and in large screening series. References 1. Mammography: a user s guide. Washington, DC: National Council on Radiation Protection and Measurements, 1986; 8 13. 2. Roebuck EJ. Clinical radiology of the breast. 1st ed. Oxford, England: Heinemann Medical Books, 1990; 56 76. 3. Eklund GW, Cardenosa G. The art of mammographic positioning. Radiol Clin North Am 1992; 30:21 53. 4. Eklund GW, Cardenosa G, Parson W. Assessing adequacy of mammographic image quality. Radiology 1994; 190:297 307. 5. Bassett L. Clinical image evaluation. Radiol Clin North Am 1995; 33:1027 1039. 6. Basset LW. Quality determinants of mammography: clinical image evaluation. In: Kopans DB, Mendelson EB, eds. Syllabus: a categorical course in breast imaging. Oak Brook, Ill: Radiological Society of North America, 1996; 57 67. 7. Glantz SA. Primer of biostatistics. New York, NY: McGraw-Hill, 1997; 314 322. 8. Siegel S, Castellan NJ Jr. Statistica non parametrica. Milan, Italy: McGraw-Hill, 1992; 115 191. 9. Bassett LW, Hirbawi IA, DeBruhl N, Hayes MK. Mammographic positioning: evaluation from the view box. Radiology 1993; 188:803 806. 10. Helvie MA, Chan HP, Adler DD, Boyd PG. Breast thickness in routine mammograms: effects on image quality and radiation dose. AJR Am J Roentgenol 1994; 163:1371 1374. 580 Radiology November 2000 Sardanelli et al