|
Breast cancer diagnosis by scintimammography: a meta-analysis and review of the literature |
Liberman M, Sampalis F, Mulder D S, Sampalis J S |
|
|
CRD summary This diagnostic review combined studies of scintimammography in different patient populations, and which used different methods of establishing the true diagnosis. The authors concluded that scintimammography would be useful as an additional test in patients with indeterminate mammogram results. The view has a number of methodological weaknesses and the evidence presented does not support the general conclusion.
Authors' objectives To assess the performance of scintimammography in the diagnosis of breast cancer.
Searching MEDLINE was searched from 1967 to 1999 for articles in English, using the keyword 'scintimammography'. The reference sections of retrieved papers were checked for additional studies.
Study selection Study designs of evaluations included in the reviewThe study designs eligible for inclusion were not stated explicitly, but single case reports were excluded. Both prospective and retrospective study designs were included in the review.
Specific interventions included in the reviewStudies of scintimammography for imaging breast tissue were eligible for inclusion. The most common radiopharmaceutical tracer in the included studies was technetium-99m-sestamibi; other technetium, thallium and indium compounds were also used. Imaging techniques included planar, single-photon emission computed tomography (SPECT), and combined planar-SPECT scintimammography.
Reference standard test against which the new test was comparedNo inclusion criteria relating to the reference standard were specified. The reference standards in the included studies were mastectomy, excisional biopsy, core biopsy, fine-needle aspiration cytology and mammography.
Participants included in the reviewThe authors did not specify any inclusion criteria for the participants, but they stated that the majority of studies in the review selected patients on the basis of an abnormal or indeterminate mammogram. Some studies included patients who had already been diagnosed with breast cancer.
Outcomes assessed in the reviewThe studies had to provide sufficient information to construct a 2x2 contingency table of true positives, false positives, true negatives and false negatives. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and diagnostic accuracy were calculated and reported in the review.
How were decisions on the relevance of primary studies made?The authors did not state how the papers were selected for the review, or how many reviewers performed the selection. Where data from the same study had been published several times, only the most complete data set was included in the review.
Assessment of study quality The authors devised a quality scoring system, with numerical scores for study design (prospective or retrospective), assessment of scintimammograms (how many assessors, whether they were blinded), the reference standard used, patient selection criteria, and the number of patients in the study. Studies could achieve a minimum score of 15 and a maximum of 67. On the basis of the numerical quality score, the studies were also categorised as poor (scores of 39 or below), average (scores of 40 to 59), or excellent quality (scores of 60 or over). Three reviewers (an epidemiologist, a general surgeon and a surgical resident) independently assessed each study and awarded a quality score. If a consensus between the three reviewers could not be reached, the average of their scores was used in the review.
Data extraction The authors did not state how the data were extracted for the review, or how many reviewers performed the data extraction. Data necessary to construct 2x2 contingency tables were extracted, along with patient ages, numbers of malignant and benign lesions, details of the tracer and imaging technique used for scintimammography, and information necessary for the quality assessment.
The reviewers classed intermediate and uninterpretable scintimammography results as positive because, in practice, these would require further investigation in the same way as a positive result. If the primary studies did not state the imaging technique used, this was taken to be planar.
Methods of synthesis How were the studies combined?In each primary study, the numbers of true positives, false positives, true negatives and false negatives were multiplied by the quality score. These weighted values were pooled across studies, and then divided by the sum of the quality assessment scores. This gave an overall 2x2 contingency table, which was used to calculate the sensitivity, specificity, PPV, NPV and test accuracy.
How were differences between studies investigated?The authors acknowledged that the studies were very heterogeneous in terms of design, methodology and patient population, and stated that they used the quality scoring system in an attempt to homogenise the data. Studies, grouped by quality score, were plotted in receiver operating characteristic (ROC) space.
Subgroup analyses were carried out to examine the diagnostic performance in various patient populations (patients with and without palpable breast masses, and those with abnormal mammograms), with different imaging techniques (planar versus SPECT imaging) and for different radiopharmaceutical tracers. It was unclear whether the studies were weighted for the subgroup analyses. It was also unclear whether these were planned in advance.
Results of the review The review included 64 studies. The total number of patients and the numbers in the individual studies were not reported. The individual studies identified a total of 5,354 breast lesions (3,024 malignant and 2,330 benign).
Eight studies were of 'poor' methodological quality, 38 were of 'average' quality, and 18 were 'excellent'.
Values for sensitivity in the primary studies ranged from 50.0% (specificity of 90.0%) to 100.0% (specificity of 95.5 to 100%). Specificity ranged from 0% sensitivity of 87.5%) to 100.0% (sensitivity of 75.0 to 100.0%). Based on the overall weighted 2x2 table, the pooled sensitivity was 84.5% and the pooled specificity 87.5%. The pooled PPV and NPV were 88.2% and 83.6%, respectively, and the overall diagnostic accuracy was 85.9%. Weighted pooled estimates did not differ significantly from unweighted estimates.
Scintimammography was more sensitive in patients with palpable breast masses than in those with non-palpable masses. The pooled sensitivity for a subgroup of 18 studies in patients with palpable masses was 87.8% (pooled specificity of 87.5%), compared with 66.8% (specificity of 86.9%) for 3 studies of patients with non-palpable masses. For 16 studies of patients with abnormal mammograms, the pooled sensitivity was 86.7% and the pooled specificity was 86.6%.
In a comparison of different imaging modalities, pooled sensitivities were similar, but SPECT imaging had a lower pooled specificity than planar and combined planar-SPECT. The pooled sensitivity and specificity were, respectively, 85.2% and 86.8% for 45 studies using planar imaging, 83.7% and 77.3% for 3 studies using SPECT, and 84.9% and 86.2% for 14 studies where planar and SPECT results were combined.
Technetium-99m-sestamibi was used as the radiopharmaceutical tracer in 38 studies, with a pooled diagnostic accuracy of 84.8%.
Authors' conclusions The authors concluded that scintimammography would be an effective adjunct to mammography and physical examination, to assist in obtaining a definitive diagnosis in patients with indeterminate mammograms.
CRD commentary This review addressed a broad research question, covering a range of relevant patient populations and comparing scintimammography with several reference standards. The literature search was restricted to English language articles, only one electronic database was searched, and there was no attempt to locate unpublished research. It is possible that this strategy missed several relevant studies. It is difficult to judge the generalisability of the review, as the inclusion criteria were poorly defined and only limited participant details were reported.
The validity assessment addressed several of the factors that may bias estimates of diagnostic performance. The potential for errors and reviewer bias was reduced since three independent reviewers assessed validity. However, weighting the studies by quality involved a subjective judgement of how certain methodological features would bias the results. In addition, the use of an overall quality score for each study meant that the impact of individual quality items could not be investigated. The authors did not formally test the effect of study quality on the estimate of diagnostic performance. They plotted studies in ROC space, grouped by quality, but their interpretation of this plot was based only on a visual assessment.
Although the review included different reference standards, the effect of the choice of reference standard was not considered in the analysis. Statistical heterogeneity was not formally assessed in the overall set of studies or in subgroups, but a visual examination of the study results suggests that heterogeneity was present. The authors also acknowledged that there was significant clinical heterogeneity. The calculation of summary values from a simple weighted mean 2x2 table was therefore not appropriate, as this method assumes that the studies are statistically homogeneous and represent patients drawn from the same population.
The authors presented insufficient evidence to support their conclusion about the use of scintimammography as an adjunct to mammography. The quantitative results reported should be viewed with caution given the limitations described.
Implications of the review for practice and research Practice: The authors stated that this procedure might be useful for certain patients as an additional test to physical examination and mammography. For women with indeterminate mammograms, scintimammography could assist in diagnosis and help to avoid unnecessary invasive procedures. If mammography was negative, scintimammography might still be performed if the patient had risk factors for breast cancer. Since the technique had lower sensitivity in women with non-palpable breast masses, a more conservative diagnostic approach may be needed in these women. The authors also suggested that scintimammography might be useful in younger women with dense breasts and women with implants, although these groups were not specifically considered in the review.
Research: The authors stated that large, multicentre prospective studies are required for further evaluation of scintimammography.
Funding Fonds de la Recherche en Sante du Quebec, grant number 990750.
Bibliographic details Liberman M, Sampalis F, Mulder D S, Sampalis J S. Breast cancer diagnosis by scintimammography: a meta-analysis and review of the literature. Breast Cancer Research and Treatment 2003; 80(1): 115-126 Indexing Status Subject indexing assigned by NLM MeSH Breast Neoplasms /radionuclide imaging; Female; Humans; Mammography /methods; ROC Curve; Radiopharmaceuticals; Sensitivity and Specificity AccessionNumber 12003001418 Date bibliographic record published 31/05/2005 Date abstract record published 31/05/2005 Record Status This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn. |
|
|
|