点击显示 收起
1 From the Harbor UCLA Medical Center, Torrance, Calif (I.K.); Beth Israel Hosp, Boston, Mass (J.K.B., J.L.C., S.J.S., M.J.H.); Univ of Texas Medical Branch at Galveston (J.V.M.); Women’s Imaging Center of Delaware, Newark (S.L.E.); Aultman Hosp, Canton, Ohio (L.G.H.); St Paul Medical Center, Dallas, Tex (C.E.L.); Hotel Dieu, Montreal, Quebec, Canada (R.T.); Montefiore Medical Center, Bronx, NY (L.M.F.); Memorial Medical Center, Springfield, Ill (C.E.N.), Hosp of the Univ of Pennsylvania, Philadelphia (A.M.S.); JSS Medical Research, Montreal, Quebec, Canada (J.S.S.); and DuPont Pharmaceuticals Company, 331 Treble Cove Rd, 500-2, North Billerica, MA 01862 (S.B.H.). The participating investigators and hospitals are listed at the end of this article. From the 1996 RSNA scientific assembly. Received Jan 2, 2001; revision requested Feb 7; revision received Jun 12; accepted Jul 5. All authors received financial support from DuPont Pharmaceuticals.
ABSTRACT |
---|
Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
---|
MATERIALS AND METHODS: A total of 558 women were prospectively enrolled from 42 centers in North America. Images were interpreted by readers blinded to the subjects’ clinical history, mammographic findings, and other test results. The Breast Imaging Reporting and Data System classification was used to describe breast density. Parenchymal patterns of "heterogeneously dense" and "extremely dense" were used to classify breasts as dense, whereas "almost entirely fat" and "numerous vague densities" defined fatty breasts. Between-group differences were evaluated with the 2 test for categorical variables and Student t test for continuous variables. Accuracy of scintimammography was assessed against the core laboratory histopathologic evaluation, the standard. The 95% CIs around point estimates of sensitivity, specificity, and positive and negative predictive values were calculated with the normal approximation to the binomial distribution.
RESULTS: The analyses were based on 580 breasts with an abnormality; 276 (48%) breasts were dense and 228 had a malignant lesion. Diagnostic properties for scintimammography of fatty versus dense breasts were, respectively, sensitivity, 72% versus 70%; specificity, 80% versus 78%; positive predictive value, 72% versus 67%; negative predictive value, 81% versus 81%; and accuracy, 77% versus 75% (all not significant). Scintimammography led to similar and significant changes in the posttest likelihood of cancer for both dense and fatty breasts.
CONCLUSION: The diagnostic accuracy of scintimammography is not affected by breast density.
Index terms: Breast, radionuclide studies, 00.1216 • Breast neoplasms, diagnosis, 00.31, 00.32 • Breast radiography, comparative studies, 00.11, 00.1216
INTRODUCTION |
---|
Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
---|
Despite the increasing incidence of the disease, age-adjusted mortality rates have been declining in several industrialized regions of the world (2–6). Early detection and effective treatment of women with a diagnosis of breast cancer are major factors contributing to the decline in the mortality rate. It is now well established that early detection prolongs survival of patients with breast cancer and that prognosis becomes markedly worse if the cancer is detected at later stages. Survival duration decreases substantially when metastasis has occurred (2,7).
The reduced sensitivity of mammography in dense breasts, as explicitly stated in the American College of Radiology Breast Imaging Reporting and Data System, or BI-RADS (8), presents a challenge to early detection (9,10). Scintimammography with technetium 99m (99mTc) sestamibi has been suggested as a beneficial adjunct to the current diagnostic procedures for breast cancer (11–16). The biologic principle underlying scintimammography suggests that its accuracy should not be affected by breast density and could therefore be beneficial in the examination of women with dense breasts. The purpose of the present study was to evaluate the accuracy of scintimammography as an adjunct to physical examination and mammography in the detection of breast cancer in women with dense and fatty breasts.
MATERIALS AND METHODS |
---|
Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
---|
Mammography
Mammography was performed by using standard craniocaudal and mediolateral oblique views, with additional views obtained as clinically indicated. All participating sites in the United States had received accreditation by the American College of Radiology Mammography Accreditation Program. The Canadian site had full accreditation in medical imaging from the Diagnostic Accreditation Program in British Columbia. Mammography results were reported by a single investigator at each site. The BI-RADS (8) lexicon was generally used to record information on breast parenchymal pattern, associated findings, mass abnormality description, shape and attenuation density, and type and distribution of calcifications. Parenchymal patterns of "heterogeneously dense" and "extremely dense" were used to classify breasts as dense, whereas "almost entirely fat" and "numerous vague densities" (equivalent to BI-RADS "scattered fibroglandular opacities") defined fatty breasts.
Scintimammography
Subjects received a 20–30-mCi (740–1,110 MBq) bolus of intravenously injected 99mTc sestamibi in the arm, which was contralateral to the suspicious breast abnormality; subjects with bilateral abnormalities received injections in a dorsalis pedis vein. Five minutes after the injection, a 10-minute lateral view of the breast scheduled for biopsy was obtained, with the subject positioned prone on an imaging table overlay so that the breast being imaged was pendent. The subject was then repositioned for acquisition of a lateral view of the contralateral breast followed by a supine anterior view. Lateral imaging was repeated 1 hour after injection. Since greater diagnostic sensitivity has been reported for the early image (17), results from the delayed images are not included in this analysis. Images were acquired with a 128 x 128 matrix by using a high-resolution collimator and a 10% energy window centered at 140 keV.
Scintimammographic images were read at the sites by the principal investigator (including I.K., J.V.M., S.L.E.) at each site (institutional results). Additionally, two groups of three nuclear medicine physicians who were blinded to the subjects’ clinical history, physical examination results (including presence or absence of palpable abnormalities), mammographic findings, and institutional scintimammographic results reviewed the images. Each physician independently evaluated the images. One group of blinded readers read the images obtained in the subjects with nonpalpable abnormalities, and the other group read the images obtained in the subjects with palpable abnormalities. For the blinded read, digital data were converted to a common display format, and the readers were allowed to manipulate image contrast, image intensity, and color scale. The scintimammograms were classified according to the following five-point scale: 0, normal; 1, equivocal; 2, focal uptake of low intensity; 3, focal uptake of medium intensity; and 4, focal uptake of high intensity. Scores of 2, 3, and 4 indicated a positive finding, a score of 0 indicated a negative finding, and a score of 1 indicated an equivocal and hence noninterpretable finding.
Biopsy and Histopathology
Biopsy decisions were based on mammographic results and physical findings and were independent of scintimammographic results. For nonpalpable mammographically detected abnormalities, needle localization followed by specimen radiography was used to confirm that the lesion was correctly sampled. Tumor size was determined from excised tissue specimens by institutional histopathologists. Histopathologic specimens were reinterpreted at a central site (core laboratory) by using excisional biopsy or mastectomy specimens. Evaluation was performed (by either J.L.C. or S.J.S.) with original slides or recuts of the tissue blocks, if originals were unavailable. Ductal carcinoma in situ, infiltrating ductal carcinoma, and infiltrating lobular carcinoma were classified as malignant. Lobular carcinoma in situ was considered nonmalignant.
Statistical Methods
Study participants were classified as having dense or fatty breasts according to the BI-RADS criteria described previously. Differences between patients with dense and those with fatty breasts were assessed by using the Student t test for continuous variables and the 2 statistic for categorical variables. The diagnostic accuracy of scintimammography was evaluated against the core laboratory histopathologic results, the standard. Sensitivity, specificity, positive and negative predictive values, and accuracy were calculated by using standard methods. Ninety-five percent CIs around these estimates were calculated by using the normal approximation to the binomial distribution. The three independent blinded ratings of each scintimammogram were combined into one blinded-read result according to the following rules: (a) If any of the three ratings were positive, the combined rating was positive; (b) if all three ratings were negative or if two were negative and the third was equivocal, the combined rating was negative; and (c) if all three ratings were equivocal or if two were equivocal and the third was negative, the combined rating was equivocal. This approach was adopted to ensure that our analysis was as conservative as possible and that it closely resembled the clinical situation where one positive result would raise sufficient doubt as to be an indication for further investigation.
RESULTS |
---|
Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
---|
|
|
|
|
|
Among the 276 dense breasts, 146 (53%) had a palpable mass, 101 (69%) of which also had positive mammographic findings (Fig 2). Malignant lesions were confirmed in 61 (60%) of 101 breasts. The scintimammograms were positive for 46 (75%) of these breasts, negative for 10 (16%), equivocal for four (7%), and missing for one. For the 40 dense breasts with a palpable mass, positive mammographic findings, and a benign lesion, the scintimammogram was negative for 22 (55%) breasts, positive for 13 (32%), equivocal for three (8%), and missing for two. Of the 45 dense breasts with a palpable mass and negative mammographic findings, seven (16%) had a malignant lesion. For these seven malignant lesions, the scintimammogram was positive for three (43%), negative for two (29%), equivocal for one (14%), and missing for one of the breasts.
|
In patients with fatty breasts, there were 86 true-positive and 32 false-negative scintimammograms. Tumor-size data were available for 80 and 30 of these lesions, respectively. The mean tumor size of the true-positive lesions was 2.26 cm (95% confidence limits: 1.62, 2.90; range, 0.4–25.0 cm); and that of the false-negative lesions, 1.11 cm (95% confidence limits: 0.91, 1.30; range, 0.1–2.2 cm). The difference between these two means was statistically significant (P = .03). Among the women with dense breasts, there were 85 true-positive and 29 false-negative scintimammograms. In this group, tumor-size data were available for 67 and 29 of these lesions, respectively. The mean tumor size of the true-positive lesions was 2.42 cm (95% confidence limits: 2.02, 2.82; range, 0.2–11.0 cm); and that of the false-negative lesions, 1.37 cm (95% confidence limits: 1.01, 1.74; range, 0.3–10.0 cm). This difference was also statistically significant (P = .002).
Table 5 shows the changes in the pre- and posttest probabilities of malignancy that were associated with a sequence of tests beginning with physical examination and followed by mammography and then scintimammography. The prevalence of malignancy in each group was used as the baseline estimate of the pretest probability for cancer. These results show that for patients with fatty breasts, a palpable mass, and a positive mammogram, a positive scintimammogram increased the probability of cancer by 32%, from 63% to 84%. In the same subgroup, a negative scintimammogram decreased the probability of cancer by 53%, from 63% to 30%. Among the patients with positive findings on both physical examination and mammography, a positive scintimammogram increased the probability of cancer by 29%, from 60% to 78%. A negative scintimammogram decreased the probability of cancer by 48%, from 60% to 31%.
|
DISCUSSION |
---|
Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
---|
In the current study, there were 45 dense breasts with a palpable mass but negative mammographic findings. Among the 45 breasts, there were six with a malignancy not detected at mammography. The scintimammogram was positive for three and equivocal for one of the six breasts. With the assumption that a positive or equivocal scintimammogram would be an indication for further testing, scintimammography depicted 67% (four of six) of the cancers that were missed at mammography. In this group, the false-negative rate of scintimammography was 7% (two of 29) compared with 13% (six of 45) for mammography.
In the subgroup of women with dense breasts, no palpable mass, and a positive mammogram, scintimammography was able to correctly depict 73 (80%) of 91 breasts without a malignant lesion and 21 (55%) of 38 breasts with a malignant lesion. The false-positive rate for mammography in this subgroup was 71%, compared with 46% for scintimammography. These data show that due to the identification of cancers missed at mammography, scintimammography can be beneficial to women with dense breasts. These results therefore support the inclusion of scintimammography in the diagnostic examination of women with dense breasts.
The results of this study have also shown that scintimammography could be a potentially beneficial adjunct in the diagnostic evaluation for breast cancer, regardless of patient breast density. Thus, in our sample, scintimammography produced significant improvement over mammography in the posttest likelihood of the presence or absence of malignant lesions, which was similar for women with dense and fatty breasts.
Diagnostic statistics for the blinded readers tended to be lower than that for the institutional readers. This is not an infrequent observation in diagnostic studies and may be attributed to a combination of factors, such as access of the institutional readers to the patient clinical history and mammographic results, lack of familiarity with the image display system of the blinded readers, and fatigue. Scintimammographic images in clinical practice should be interpreted with a full knowledge of patient history, mammographic results, and physical findings. We believe that such procedures will lead to diagnostic results that are closer to our institutional results than to the blinded results.
The estimates of sensitivity, specificity, and positive and negative predictive values for scintimammography in the current study were satisfactory but somewhat lower than generally reported in other studies (11–16). This may have resulted from the method by which subjects were selected for inclusion in the study and the high proportion of women with negative physical examinations (nonpalpable lesions) in the study sample. While the diagnostic statistics for scintimammography for the study subjects with a palpable mass were significantly higher, the over representation of women with negative physical findings may have contributed to these lower overall estimates.
In conclusion, the results of this observational prospective study have shown that scintimammography is highly accurate and could provide significant clinical benefit as an adjunct to physical examination and mammography in the detection of breast cancer. The accuracy and potential benefit of scintimammography are not affected by breast density, and, on the basis of this limited study, sensitivity in dense breasts appears to be better than that with mammography. Therefore, the current study findings further support the implementation of scintimammography in the diagnostic evaluation for breast cancer.
Core Histopathology Laboratory: J. L. Connolly and S. J. Schnitt, Beth Israel Hospital, Boston, Mass.
Core Mammography Laboratory: D. Kopans, Massachusetts General Hospital, Boston, Mass.
Blinded Readers: L. M. Freeman, Montefiore Medical Center, Bronx, NY; L. G. Hanelin, Aultman Hospital, Canton, Ohio; C. Lugo, St Paul Medical Center, Dallas, Tex; C. E. Neal, Memorial Medical Center, Springfield, Ill; A. M. Scheff, Hospital of the University of Pennsylvania, Philadelphia; R. Taillefer, Hotel-Dieu, Montreal, Quebec, Canada.
Principal Investigators: H. Abdel-Dayem, St Vincent’s Hospital and Medical Center, New York, NY; E. J. Andrews, Jr, Medical Center Clinic/West Florida Hospital, Pensacola; B. Barron, University of Texas Health Science Center, Houston; J. Birsner, Antelope Valley Hospital Medical Center, Lancaster, Calif; M. J. Blend, Doctors Hospital of Hyde Park, Chicago, Ill; M. Brown, University of Pittsburgh School of Medicine, Pa; R. R. Butler, Jr, University Medical Center, Jacksonville, Fla; W. Carpentier, Scott and White Clinic and Memorial Hospital, Temple, Tex; R. Carretta, Roseville Community Hospital, Calif; B. D. Collier, Medical College of Wisconsin, Milwaukee; L. P. Davis, Harper Hospital, Detroit, Mich; S. L. Edell, Women’s Imaging Center of Delaware, Newark; A. Fischman, Massachusetts General Hospital, Boston; J. H. Garafola, Lancaster General Hospital, Pa; S. Grossman, Western Pennsylvania Hospital, Pittsburgh; H. Handmaker, Arizona Institute of Nuclear Medicine, Phoenix; R. Jaros, Catholic Medical Center, Manchester, NH; P. Jolles, Medical College of Virginia, Richmond; I. Khalkhali, Harbor-UCLA Medical Center, Torrance, Calif; S. Kipper, Tri-Center Medical Center, Oceanside, Calif; G. Kirk, Loma Linda University Medical Center, Calif; A. Klonecke, Kaiser Foundation Hospital, Sacramento, Calif; M. Lee, Virginia Mason Hospital, Seattle, Wash; J. Levy, Rose Medical Center, Denver, Colo; W. McCartney, University of North Carolina, Chapel Hill; G. W. Moskowitz, University Hospital, Newark, NJ; B. P. Mullan, Mayo Clinic, Rochester, Minn; C. Nagle, William Beaumont Hospital, Troy, Mich; C. Park, Thomas Jefferson University Hospital, Philadelphia, Pa; S. R. Parmett, St Lukes-Roosevelt Hospital Center, New York, NY; P. J. Peller, Lutheran General Hospital, Park Ridge, Ill; N. Peterson, Memorial Mission Hospital, Asheville, NC; H. Phillips, Santa Rosa Radiology Medical Center, Calif; J. Seabold, University of Iowa College, Iowa City; G. Sfakianakis, University of Miami School of Medicine, Fla; H. Shapiro, Mount Sinai Hospital, Hartford, Conn; P. Shtasel, Graduate Hospital of Philadelphia, Pa; C. Stegman, Scottsdale Memorial Hospital, Ariz; F. D. Thomas, SUNY Health Science Center at Syracuse, NY; W. Thompson, Vancouver General Hospital, British Columbia, Canada; M. Troychak, Western Radiology Associates, Seattle, Wash; D. Van Nostrand, Good Samaritan Hospital, Baltimore, Md; J. Villanueva-Meyer, University of Texas Medical Branch at Galveston; A. D. Waxman, Cedars Sinai Medical Center, Los Angeles, Calif; H. A. Ziessman, Georgetown University, Washington, DC.
ACKNOWLEDGMENTS |
---|
REFERENCES |
---|
Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
---|