A systematic review of positron emission tomography (PET) and positron emission tomography/computed tomography (PET/CT) for the diagnosis of breast cancer recurrence

A systematic review of positron emission tomography (PET) and positron emission tomography/computed tomography (PET/CT) for the diagnosis of breast cancer recurrence

Article Type: Health technology assessment From: Clinical Governance: An International Journal, Volume 16, Issue 2

M. Pennant, Y. Takwoingi, L. Pennant, C. Davenport, A. Fry-Smith, A. Eisinga, L. Andronis, T. Arvanitis, J. Deeks and C. Hyde

Background

Breast cancer (BC) affects 1 in 13 women in their lifetime. Treatment options have developed significantly over the past decade and have had an impact on survival. The diagnosis of BC recurrence is important to allow appropriate treatment. Positron emission tomography (PET) and positron emission tomography/computed tomography (PET/CT) are technologies that have application in the detection and management of cancer. The adoption of PET or PET/CT depends not only on their diagnostic accuracy but also on their comparative advantage over existing diagnostic approaches.

Objectives

This report covers the question of the effectiveness of PET and PET/CT for diagnosing BC recurrence and a second report (to follow) will provide economic modelling to address the question of their cost-effectiveness in this context. The aim of this review was to assess the value of PET and PET/CT, in addition to current practice, for the diagnosis of BC recurrence. The objectives were:

• •

  to assess the diagnostic accuracy of PET compared with conventional diagnostic strategies;

• •

  to assess the diagnostic accuracy of PET/CT compared with conventional diagnostic strategies;

• •

  to assess the diagnostic accuracy of PET and PET/CT compared with magnetic resonance imaging (MRI);

• •

  to compare the accuracy of PET with PET/CT;

• •

  to assess the overall diagnostic accuracy of PET and PET/CT;

• •

  to investigate the impact of PET and PET/CT on patient management; and

• •

  to explore possible mediators of the accuracy of PET and PET/CT.

Methods

A systematic review was conducted. A search for primary studies in MEDLINE (Ovid) and EMBASE (Ovid) was conducted with no language restrictions. Studies of PET or PET/CT in patients with history of BC and suspicion of recurrence were selected for inclusion. Studies were excluded if investigations were conducted for screening or staging of primary BC, if a non-standard PET or PET/CT technology was used, if there was an inadequate or undefined reference standard, or if raw data for calculation of diagnostic accuracy were not available. Both comparative and non-comparative studies were included.

Data extraction and quality assessment were conducted independently by two reviewers with any disagreements resolved by consensus. Direct and indirect comparisons were made between PET and PET/CT and between these technologies and methods of conventional imaging, and a meta-analysis was performed using a bivariate random effects model. Analysis was conducted separately on patient- and lesion-based data. Subgroup analysis was conducted to investigate variation in the accuracy of PET in certain populations or contexts and sensitivity analysis was conducted to examine the reliability of the primary outcome measures.

Results

A total of 28 studies were included in the current review and, of these, 26 investigated the diagnostic accuracy of PET. Twenty-five presented patient-based data and seven presented lesion-based data for PET. Six studies investigated the accuracy of PET/CT, five presenting patient-based data and one presenting lesion-based data. Sixteen studies conducted direct comparisons and, of these, 12 compared the accuracy of PET or PET/CT with conventional diagnostic tests and four compared PET or PET/CT with an MRI technology. Quality varied between studies, and the major quality issue identified was the time delay between conventional tests and PET or PET/CT in comparative studies. The PET or PET/CT technology used was similar across the studies.

• •

  For patient-based data, in studies where direct comparisons were made, PET had significantly higher sensitivity [89 per cent, 95 per cent confidence interval (CI) 83 per cent to 93 per cent vs 79 per cent, 95 per cent CI 72 per cent to 85 per cent, relative sensitivity 1.12, 95 per cent CI 1.04 to 1.21, p=0.005] and significantly higher specificity (93 per cent, 95 per cent CI 83 per cent to 97 per cent vs 83 per cent, 95 per cent CI 67 per cent to 92 per cent, relative specificity 1.12, 95 per cent CI 1.01 to 1.24, p=0.036), compared with conventional imaging tests (CITs) (n=10). Test performance did not appear to vary according to the type of CIT that was compared with PET (p=0.500). Indirect comparisons, where all CIT (n=11) and PET (n=25) studies were included, gave the same findings. For lesion-based data, no significant differences in sensitivity or specificity between PET and CIT were observed for studies making direct comparisons (n=3) or for indirect comparisons for all PET (n=7) and CIT (n=3) studies. In the sensitivity analysis of patient data, for studies in which the time period between PET and comparator tests was clearly less than one month (n=6), differences between PET and CIT tended to be smaller and the difference in sensitivity became non-significant.

• •

  For patient-based data, in all studies where direct comparisons were made (n=4), the CIT used was CT. In these studies, compared with CT, PET/CT had significantly higher sensitivity (95 per cent, 95 per cent CI 88 per cent to 98 per cent vs 80 per cent, 95 per cent CI 65 per cent to 90 per cent, relative sensitivity 1.19, 95 per cent CI 1.03 to 1.37, p=0.015) but the increase in specificity was not significant (89 per cent, 95 per cent CI 69 per cent to 97 per cent vs 77 per cent, 95 per cent CI 50 per cent to 92 per cent, relative specificity 1.15, 95 per cent CI 0.95 to 1.41, p=0.157). Indirect comparisons, where all CIT (n=11) and PET/CT (n=5) studies were included, gave the same findings. No lesion-based data compared PET/CT with CIT. In the sensitivity analysis of patient data, for studies in which the time period between PET/CT and comparator tests was clearly less than one month (n=3) differences between PET/CT and CT became non-significant.

• •

  For patient-based data, three studies compared PET with different types of MRI technology. In each of these studies, there were no significant differences in the sensitivity or specificity of PET compared with MRI. One study compared PET/CT and MRI on a lesion basis and there were no significant differences in sensitivity or specificity for PET/CT compared with MRI.

• •

  For patient-based data, in the analysis of studies directly comparing PET/CT and PET (n=4), PET/CT had significantly higher sensitivity (96 per cent, 95 per cent CI 90 per cent to 98 per cent vs 85 per cent, 95 per cent CI 77 per cent to 91 per cent, relative sensitivity 1.11, 95 per cent CI 1.03 to 1.18, p=0.006), but the increase in specificity was not significant compared with PET (89 per cent, 95 per cent CI 74 per cent to 96 per cent vs 82 per cent, 95 per cent CI 64 per cent to 92 per cent, relative specificity 1.08, 95 per cent CI 0.94 to 1.20, p=0.267). The same pattern of results was observed for the indirect comparison of all PET/CT (n=5) and PET (n=25) studies. In the lesion-based analysis, indirect comparison of PET/CT (n=2) and PET (n=7) showed no significant differences in sensitivity or specificity between PET/CT and PET.

• •

  For overall diagnostic accuracy, on a patient basis, PET/CT (n=5) and PET (n=25) had sensitivities of 96 per cent (95 per cent CI 89 per cent to 99 per cent) and 91 per cent (95 per cent CI 86 per cent to 94 per cent) and specificities of 89 per cent (95 per cent CI 75 per cent to 95 per cent) and 86 per cent (95 per cent CI 79 per cent to 91 per cent) respectively. On a lesion basis, PET/CT (n=2) and PET (n=7) had sensitivities of 96 per cent (95 per cent CI 80 per cent to 99 per cent) and 89 per cent (95 per cent CI 78 per cent to 95 per cent) and specificities of 83 per cent (95 per cent CI 61 per cent to 94 per cent) and 91 per cent (95 per cent CI 83 per cent to 96 per cent), respectively. There was considerable heterogeneity in the spread of results for PET.

• •

  Changes in patient management in study participants ranged from 11 per cent to 74 per cent (median 27 per cent). These changes included initiation and avoidance of medical treatment such as hormone therapy and chemotherapy. In the three studies where only changes in management directly due to PET or PET/CT were considered (patients were not correctly diagnosed by conventional imaging techniques), estimates ranged from 11 per cent to 25 per cent.

• •

  In subgroup analysis, the accuracy of PET did not appear to be related to the location of disease or to whether PET was conducted with or without knowledge of previous clinical history and imaging studies. Characteristics of patient populations varied in many respects and it was not possible to draw definite conclusions about patient characteristics that may have an impact on test accuracy.

Conclusions

• •

  For detection of BC recurrence, in addition to conventional imaging techniques, PET may generally offer improved diagnostic accuracy compared with current standard practice. Uncertainty remains around its use as a replacement, rather than an add-on, to existing imaging technologies.

• •

  PET/CT appears to show a clear advantage over CT for the diagnosis of BC recurrence. Although PET/CT may give an advantage over other CITs, its incremental value over other tests has yet to be directly assessed in studies. Concurrent use with, rather than replacement of, other conventional tests may be appropriate.

• •

  PET/CT appears to show a clear advantage over PET and it is likely to be preferred to PET for use in this context.

• •

  PET and PET/CT appear to have some impact on patient management but there is currently no evidence of the effect of their use on patient outcomes.

Recommendations for future research

• •

  Prospective studies with patient populations clearly defined with regard to their clinical presentation.

• •

  Study of the diagnostic accuracy of PET/CT compared with conventional imaging techniques.

• •

  Study of PET/CT compared with whole-body MRI.

• •

  Studies investigating the possibility of using PET/CT as a replacement for, rather than an addition to, CITs.

• •

  Using modelling of the impact of PET/CT on patient outcomes (to be published in another report) to inform the possibility of conducting large-scale intervention trials to assess impacts on long-term patient outcomes.

Implications for policy

PET/CT has largely superseded PET in current practice, and the apparent advantage of PET/CT over PET found in this review supports that move. On the basis of some of the uncertainties observed, it may be premature to make recommendations about the precise diagnostic role of PET/CT in practice. However, current recommendations for its use for diagnosing metastatic BC following equivocal findings on conventional imaging techniques appear to be justified. It appears that PET/CT may be useful as an addition to current practice for the diagnosis of BC recurrence but this should be reassessed in light of the analysis of its cost-effectiveness.

© Crown Copyright

M. Pennant, C. Davenport and A. Fry-Smith are all based at West Midlands Health Technology Assessment Collaboration, Unit of Public Health, Epidemiology & Biostatistics, University of Birmingham, Birmingham, UK.

Y. Takwoingi and J. Deeks are both based at the Biostatistics group, Unit of Public Health, Epidemiology & Biostatistics, University of Birmingham, Birmingham, UK.

L. Pennant is based at the National Health Service West Midlands Deanery, UK.

A. Eisinga is based at the UK Cochrane Centre, Oxford, UK.

L. Andronis is based at the Unit of Health Economics, University of Birmingham, Birmingham, UK.

T. Arvanitis is based at the School of Electronic, Electrical & Computer Engineering, University of Birmingham, Birmingham, UK.

C. Hyde is based at the Peninsula College of Medicine & Dentistry, University of Exeter, Exeter, UK.