| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Centro per lo Studio e la Prevenzione Oncologica, Viale A. Volta 171, 50131
Florence, Italy.
2 Screening and Test Evaluation Program, School of Public Health, University of
Sydney, Sydney, Australia.
Received March 23, 2007;
revised May 18, 2007;
Address correspondence to S. Ciatto
(s.ciatto{at}cspo.it).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. In a retrospective study covering 2004–2005, we compared digital with screen-film mammography in two concurrent screening cohorts of women 50–69 years old participating in a screening program operated from mobile units. Each cohort had 14,385 participants matched by age and interpreting radiologist from all participants consecutively registered. We compared recall and cancer detection rates.
RESULTS. The recall rate was higher for digital mammography (4.56% vs 3.96%, p =0.01), particularly when clustered microcalcifications were the only finding (1.05% vs 0.41%, p =10–6) and for younger women (50–59 vs 60–69 years, 5.12% vs 4.17%, p = 0.009). The higher recall rate for digital mammography was mainly evident at incidence screening. The recall rate due to poor technical quality was lower with digital mammography (0.27% vs 0.50%, p = 0.002), possibly because real-time feedback was available. The detection rate was higher for digital mammography (0.72% vs 0.58%, p = 0.14), particularly for cancers depicted as clustered microcalcifications (0.26% vs 0.12%, p = 0.007), in younger (50–59 years) women (0.63% vs 0.42%, p = 0.09), and in denser breasts (1.09% vs 0.53%, p = 0.24). No significant difference was observed in positive predictive value on recall for digital mammography or screen-film mammography. Early cancer (pTis, pT1mic, pT1a) was more frequent in cancer detected with digital mammography than in that detected with screen-film mammography (41.3% vs 27.3%, p =0.06).
CONCLUSION. Digital mammography may be more effective than screen-film mammography in contemporary screening practice in mobile units. The data indicate that digital mammography depicts more tumors than does screen-film mammography, especially lesions seen as microcalcifications. The potential association with improved outcome warrants further study.
Keywords: breast cancer • digital imaging • mammography • mammography screening • women's imaging
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Studies showing the diagnostic accuracy of digital mammography to be comparable with that of screen-film mammography have often focused on symptomatic, diagnostic, and selfreferral settings [3–5]. In only a few studies [6–9] have investigators examined the accuracy of this technique in population-based screening practice. The multicenter Digital Mammographic Imaging Screening Trial [9] showed that digital mammography is more sensitive than screen-film mammography in some groups (women younger than 50 years, women with dense breasts, and premenopausal and perimenopausal women), although overall the two techniques have similar accuracy. There is, however, a paucity of evidence on the diagnostic accuracy of digital mammography as a reliable alternative to screen-film mammography in the context of contemporary screening implemented as part of an organized program.
The screening program in the district of Florence, Italy, provides all population-based organized screening services in that region and operates four mobile mammography units equipped with digital and screen-film mammography. All image interpretation and screening assessments are conducted by the same radiologists at a single institution. The purpose of this retrospective study was to compare the diagnostic accuracy of digital mammography with that of screen-film mammography in concurrent screening cohorts participating in the same population-based organized screening program.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Both the screen-film and digital mammography systems were subjected to a quality control program according to the European guidelines for quality assurance in mammographic screening, including an addendum on digital mammography [10]. To select the appropriate working point according to the aslow-as-reasonably-achievable principle for the digital mammography system, complete physical characterization was performed [10]. For digital mammography, the entrance surface exposure for a standard breast under clinical conditions was 7.1 mGy, resulting in a 1.8-mGy average glandular dose; for screen-film mammography, the exposure was 7.5 mGy, resulting in an average glandular dose of 1.6 mGy. The diagnostic workstation monitors were subjected to a regular quality control program according to the American Association of Physicists in Medicine Task Group 18 report "Assessment of Display Performance for Medical Imaging Systems" [11]. No special image processing was adopted for soft-copy review. Radiologists could modify image brightness and contrast and could use magnification of black and white inversion, but the use of such features was not recorded.
Digital and screen-film mammograms in the study were interpreted by four radiologists, each with more than 5 years of experience in mammographic screening and an average workload of more than 10,000 mammograms interpreted per year. Screen-film mammograms were developed in a separate session at the central unit of our institution. Interpretation was not immediate for either screen-film or digital mammography. At the time of imaging in the mobile unit, no real-time feedback (e.g., to repeat imaging in case of inadequate positioning) was provided to the radiographer for screen-film mammography, but feedback was possible for digital mammography because the digital image obtained was displayed on the monitor. Double review was performed and was not independent, the second reviewer being aware of the first reviewer's report. Thus, recall for further assessment was based on suspicion by either reviewer. At incidence screening, previous mammograms (digital or screen film) were available for comparison at the request of either radiologist. Computer-assisted detection was not used for digital or screen-film mammography because it was not standard practice in most European screening programs.
No special protocol existed regarding the use of digital or screen-film mammography units in different quarters of the city. Women who underwent previous mammography in the digital mammography mobile unit did not automatically undergo follow-up mammography in the digital unit.
Participants
Population-based screening has been ongoing since 1970 in the district of
Florence. The main features of the program have been reported
[12]. The efficacy of the
program was shown in a case–control study
[13]. Screening is offered to
all women 50–69 years old. The cancer detection rate at prevalence
screening is 8.25 cases per 1,000 and at incidence screening is 3.45 cases per
1,000 [12]. Data from the
local cancer registry [14]
indicate that the two areas served by the mobile screening units in this study
had similar breast cancer incidence and detection rates at previous rounds of
screen-film mammography. The detection rates at screening in 2003 were 5.90
cases per 1,000 in the screen-film mammography study area and 5.49 cases per
1,000 in the digital mammography study area. To ensure that concurrent cohorts
were comparable, subjects were included from all participants screened for the
study duration after matching by decade of life and each possible pair of
reviewing radiologists (Fig.
1). All images were interpreted under the same review and
evaluation protocols.
|
The accuracy of digital mammography was compared with that of screen-film mammography for recall, cancer detection rate, and positive predictive value at recall in the overall series and according to variables such as age, breast density, prevalence (first) or incidence (subsequent) screening round, and type of radiologic abnormalities. Because our data showed definite differences in the two mammographic methods for cancers depicted as clustered microcalcifications, we extended the evaluation by reviewing all participants recalled on the basis of the presence of microcalcifications. Two radiologists with long-standing experience in screening mammography reviewed all such cases independently and used consensus in case of disagreement. These radiologists classified the cases according to mammographic morphologic features and level of suspicion. An alternative system for reporting findings on breast imaging was based on a standardized categoric scale (1, negative; 2, benign; 3, equivocal; 4, suspicious; 5, malignant) with published estimates of the likelihood of malignancy [15, 16]. The assessment categories, referred to as R1–R5, were only partially consistent with BI-RADS. The R3 category in our practice prompts assessment rather than early recall; the latter is not a recommended option in screening protocols at our center or in European standards [10]. Statistical significance of observed differences in recall and cancer detection rates was determined by univariate analysis with the chisquare test (Stata 8.0 SE software, StataCorp). The significance level was set at p < 0.05.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Participants
Table 1 shows the
distribution of radiologic breast density, age group, and screening round for
the two mammographic techniques. The results show the two cohorts were
generally comparable.
|
Recall Rate
Table 2 shows that the
overall recall rate, including recall because of poor technical quality,
differed significantly between screen-film mammography (3.96%) and digital
mammography (4.56%) (
2df = 1 =
6.44, p = 0.01). Relative to screen-film mammography, digital
mammography was associated with significantly more recalls because of
radiologic abnormality (4.29% vs 3.46%,
2df = 1 = 13.42, p =
0.0002) and with significantly less recall because of poor technical quality
(0.27% vs 0.50%, 2df = 1 =
9.84, p = 0.002). Digital mammography had a significantly higher
recall rate because of microcalcifications than did screen-film mammography
(1.05% vs 0.41%, 2df = 1 =
39.53, p =10–6), but the recall rates did not differ
for masses (screen-film mammography, 2.85%; digital mammography, 3.07%;
2df = 1 = 1.23; p =
0.25) or distortions (screen-film mammography, 0.19%; digital mammography,
0.16%; 2df = 1 = 0.17;
p = 0.57). Among women 50–59 years old, the recall rate was
significantly higher for digital mammography than for screen-film mammography
(5.12% vs 4.17%, 2df = 1 =
6.84, p = 0.009). Recall rates were higher for digital mammography in
all breast density categories, but the only significant difference was
recorded for very dense (> 75%) breasts (4.85% vs 2.69%,
2df = 1 = 4.48, p =
0.03). Recall rates were similar for screen-film and digital mammography at
prevalence screening, but a significantly higher recall rate was observed for
digital mammography (4.15% vs 3.44%, 2df
= 1 = 8.70, p = 0.003) at incidence screening.
|
Cancer Detection Rate
There were 188 cancers detected (84 in the screen-film mammography group,
104 in the digital mammography group), and 21 benign lesions were subjected to
surgical biopsy (nine in the screen-film mammography group, 12 in the digital
mammography group). There were 144 invasive cancers (69 in the screen-film
mammography, 75 in the digital mammography group), and 44 malignant tumors
were ductal carcinoma in situ (DCIS) (15 in the screen-film mammography group,
29 in the digital mammography group). Thus, overall, 23.4% (44/188) of all
malignant tumors detected in this study were DCIS. The TNM pT category at
diagnosis in the screen-film mammography series was pTis in 15 cases, pT1mic
in one case, pT1a in seven cases, pT1b in 24, pT1c in 26, pT2 in seven, pT4b
in two, and pTx in two cases. The corresponding distribution for the digital
mammography series was pTis in 29 cases, pT1mic in five, pT1a in nine, pT1b in
17, pT1c in 38, pT2 in five cases, and pTx in one case. The frequency of
early-stage lesions (pTis, pT1mic, pT1a) was higher in the digital than in the
screen-film mammography series (41.3% vs 27.3%, 2df =
1 = 3.38, p = 0.06).
Table 3 shows the cancer
detection rate for the screen-film and digital mammography groups according to
type of abnormality on mammography, age group, breast density category, and
screening round. The most notable finding was that significantly more cases of
cancer depicted as microcalcifications were detected on digital mammography
than on screen-film mammography (0.26% vs 0.12%, 2df =
1 =7.15, p = 0.007). In addition, digital mammography had a
higher detection rate than screen-film mammography among women 50–59
years old and a higher overall detection rate than screen-film mammography at
incidence screening. The differences between the two mammographic methods,
however, did not reach statistical significance. The main characteristics of
tumors (invasive or DCIS) depicted as clustered microcalcifications and
detected on screen-film and digital mammography are reported in
Table 4.
|
|
Positive Predictive Value for Recall
The recall rate for digital mammography was higher than for screen-film
mammography, but this rate was compensated by the additional cases of cancers
detected. The positive predictive value for finding cancer among recalled
cases was 15.9% for digital mammography and 14.7% for screen-film mammography
(p = 0.65). Corresponding values for other study variables are
reported in Table 5, but minor
observed differences did not reach statistical significance.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Both cancer detection rates and recall rates in our study were in line with
standards for population screening, and more cases of cancer were detected
with digital than with screen-film mammography. This finding is concordant
with the work of other investigators
[8,
9]. As expected, the overall
need for recall was associated with age and breast density, most likely a
result of greater radiologic uncertainty in the presence of denser breasts,
possibly in combination with a residual proportion of first-round (prevalence
screen) examinations (lack of previous results for comparison) in the 50- to
59year age range. The higher recall rate for digital mammography was evident
at incidence but not at prevalence screening. We have no explanation for this
finding, except for the limited sample size (
10% of each compared cohort)
in the latter group. Overall, digital mammography had a higher recall rate
than screen-film mammography but had a higher cancer detection rate, although
the difference did not reach statistical significance.
Digital mammography was associated with significantly less need for recall because of poor technical quality than was screen-film mammography. It is worth emphasizing that this study was based on mammography performed in mobile units. Developing of screen-film mammograms was deferred, so radiographers could not immediately check proper exposure and positioning. These factors were checked for digital mammography, however, because the digital images were immediately displayed on the monitor in the mobile unit. Although we cannot exclude the possibility that better imaging quality and the opportunity for postprocessing with digital mammography may have contributed to the lower recall rate, lack of immediate image feedback may explain, at least in part, the higher recall rate for screen-film mammography. The findings on this issue may provide incentive for mobile mammography programs to convert to digital systems.
The higher detection rate with digital mammography was more evident among younger women and women with denser breasts, both conditions being traditionally associated with lower sensitivity of screen-film mammography, further confirming the findings reported by other authors [9] for such subgroups. The lack of statistical significance in our data might have been due to sample size.
Digital mammography had a significantly higher detection rate than screen-film mammography for clustered microcalcifications, which caused both the recall and cancer detection rates to be significantly higher for lesions with this feature. Traditional thought has been that calcifications are best detected with higher spatial resolution, which is conventionally lower with digital mammography than with screen-film mammography. It may well be that the compensatory conditions of increased contrast resolution in digital mammography allow enhanced visibility of calcifications. Whether the increased rate of detection of cancers depicted as clustered microcalcifications translates into a greater screening efficacy can be questioned. Cancers associated with this radiologic pattern likely may be more indolent cases, and their detection may not affect mortality [17]. We acknowledge this possibility, more so because microcalcifications detected on digital mammograms in our study were more frequently DCIS, were of lower to intermediate histologic grade, and had a less suspicious morphologic pattern (BI-RADS 3 vs 4–5, or granular vs linear branching). Our findings on this issue differ slightly from those of the Digital Mammographic Imaging Screening Trial [9], in which more cases of high-grade DCIS were detected with digital mammography. Although we have no explanation for such a difference, it should be taken into consideration that DCIS represented almost 31% of the cases of cancer in the Digital Mammographic Imaging Screening Trial and less than 25% of cases of cancer in our series.
The digital mammography series in our study had a higher borderline significant frequency of early-stage cancers (pTis, pT1mic, pT1a) than did the screen-film mammography series. This finding suggests a higher sensitivity of digital mammography than of screen-film mammography for more subtle lesions. It also explains in part why digital mammography did not have an advantage at prevalence screening, at which cancers are less likely to have subtle features relative to incidence screening. These findings warrant further study with larger series of images to determine whether digital mammography does lead to detection of cancers with a more favorable stage distribution.
One possible limitation of our study was that we found a better detection rate with one screening test relative to another, yet this finding cannot on its own be assumed to translate into screening benefit [17]. Neither this study nor other studies of digital mammography for screening were designed to measure mortality as an end point. However, assessing differences in interval cancer rates after sufficient follow-up will provide an indication of expected screening benefit [17]. A study monitoring interval cancers in the two concurrent series used in this study is ongoing.
Another possible limitation was that the matching used in the study design caused subject losses. However, we retained approximately 80% of all consecutive screening participants in the analysis. Most of the nonselected cases were from the screen-film mammography series, but this factor should not account for potential selection bias due to matching characteristics.
This study showed that digital mammography is at least as accurate and effective as screen-film mammography in current screening practice. Our data show that digital mammography leads to additional cancer detection in population screening. In accordance with the few existing studies of digital mammography in screening, our results show a greater detection rate, despite a higher recall rate, for digital mammography than for screen-film mammography and suggest that digital mammography is more sensitive among younger women and women with denser breasts. The higher cancer detection rate for digital mammography is associated with cancers depicted as clustered microcalcifications and possibly cancers with a more favorable stage distribution. Although these findings can be expected to translate into an important screening benefit, future comparison of interval cancer rates [18] may indicate whether these aspects of digital mammography are likely to have an impact on breast cancer mortality.
Acknowledgments
We are grateful to Daniela Ambrogetti, Donata Tanzini, Patrizia Falini,
Elisabetta Gentile, Miriam Gabrielleschi, Lidia Pes, Verusca Giotti, and
Antonello Casini for their invaluable contribution to data collection and for
their support in the implementation of digital mammography in the screening
program.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E J Goldstraw, I Castellano, S Ashley, and S Allen The effect of Premium View post-processing software on digital mammographic reporting Br. J. Radiol., February 1, 2010; 83(986): 122 - 128. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Karssemeijer, A. M. Bluekens, D. Beijerinck, J. J. Deurenberg, M. Beekman, R. Visser, R. van Engen, A. Bartels-Kortland, and M. J. Broeders Breast Cancer Screening Results 5 Years after Introduction of Digital Mammography in a Population-based Screening Program Radiology, November 1, 2009; 253(2): 353 - 358. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. M. Hambly, M. M. McNicholas, N. Phelan, G. C. Hargaden, A. O'Doherty, and F. L. Flanagan Comparison of Digital Mammography and Screen-Film Mammography in Breast Cancer Screening: A Review in the Irish Breast Screening Program Am. J. Roentgenol., October 1, 2009; 193(4): 1010 - 1018. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Sala, M. Comas, F. Macia, J. Martinez, M. Casamitjana, and X. Castells Implementation of Digital Mammography in a Population-based Breast Cancer Screening Program: Effect of Screening Round on Recall Rate and Cancer Detection1 Radiology, July 1, 2009; 252(1): 31 - 39. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Sala, M. Comas, F. Macia, J. Martinez, M. Casamitjana, and X. Castells Implementation of Digital Mammography in a Population-based Breast Cancer Screening Program: Effect of Screening Round on Recall Rate and Cancer Detection Radiology, May 6, 2009; (2009) 2521080696. [Abstract] [Full Text]
|
|
|
|
|
|
S. Vinnicombe, S. M. Pinto Pereira, V. A. McCormack, S. Shiel, N. Perry, and I. M. dos Santos Silva Full-Field Digital versus Screen-Film Mammography: Comparison within the UK Breast Screening Program and Systematic Review of Published Data Radiology, May 1, 2009; 251(2): 347 - 358. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
