This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.

The study compared three screening strategies aimed at women with a high risk of breast cancer. The strategies compared were mammography (XRM) alone, contrast-enhanced magnetic resonance imaging (CE MRI) alone, and a combination of XRM and CE MRI. CE MRI was conducted using a specific protocol with gadopentetate dimeglumine (Magnevists, Schering Healthcare) as the contrast medium (bolus intravenous injection of 0.2 mmol/kg body weight).

Screening.

Cost-effectiveness analysis.

The study population comprised women aged between 35 and 49 years at high genetic risk of breast cancer (i.e. >0.9% per annum). The inclusion criteria specified that women had to be tested carriers of a deleterious BRCA1, BRCA2 or TP53 mutation, or be a first-degree relative of someone with a BRCA1, BRCA2 or TP53 mutation or a strong family history of breast or ovarian cancer.

The setting was not explicitly reported, but it appears to have been secondary and tertiary care (research setting). The economic study was carried out in the UK.

Patient recruitment for the study took place between 1997 and 2004. The majority of the resource use data were collected alongside the effectiveness trial. However, when required, the data were augmented by authors' assumptions. The cost data were derived from sources published between 2000 and 2005. All costs were reported for the price year 2003.

The effectiveness data were derived from a single study, augmented by authors' assumptions when necessary.

It appears that the costing has been carried out prospectively on the same sample of patients as that used in the effectiveness study.

Details of power calculations, the method of sample selection, and suitability of the study sample were reported in a separate clinical study (Leach et al. 2005, see 'Other Publications of Related Interest' below for bibliographic details). It was reported that, in total, 649 women were included in the study. They were screened with both CE MRI and XRM, which amounted to 1,881 screens (1 to 7 screens over the 7 years of annual follow-up).

The analysis was based on a prospective multi-centre (22 centres in the UK) cohort study. However, details of the study design were not reported in the current study (see Leach et al. 2005).

Details of the methodology employed in the analysis of effectiveness and the patients' characteristics were reported in the separate clinical study (Leach et al. 2005). The primary outcome used in the analysis was the number of additional cancers detected.

The number of cases detected was 13 (0.00691 per screen) with XRM, 27 (0.01435 per screen) with CE MRI alone, and 33 (0.01754 per screen) when the methods were used in combination.

In a sub-group analysis of only patients carrying a BRCA1 mutation, XRM detected 3 out of 13 cancer cases, CE MRI detected 12 out of 13 cases, and the combination strategy detected 12 out of 13 cases.

In a sub-group analysis of only patients carrying a BRCA2 mutation, XRM detected 6 out of 12 cancer cases, CE MRI detected 7 out of 12 cases, and the combination strategy detected 11 out of 12 cases.

The analysis demonstrated that the combination of XRM and CE MRI had a higher sensitivity than the individual methods.

Some measures of effectiveness were based on authors' assumptions and expert opinion.

To determine the number of women recalled for additional testing, it was necessary for the authors to make several assumptions. This was due to the processes followed during the clinical trial.

For recall based on XRM results alone, a woman was assumed to be recalled if the first and second XRM proved abnormal irrespective of the CE MRI results, or if either the first or second XRM was abnormal and the CE MRI was normal.

For recall based on CE MRI results alone, women were recalled if the first and/or second reading of the CE MRI was abnormal, irrespective of the XRM results.

When there was disagreement about XRM reading and one and/or two CE MRI readings were abnormal, the decision to recall depended on expert opinion (radiologist blinded to CE MRI readings).

It was assumed that, when two out of three readers decided on an abnormal mammography result, women were recalled for additional testing whilst, when the decision was based on the CE MRI result alone, women were recalled when the first and/or second CE MRI reading was abnormal.

The measure of benefit used was the number of cancers detected. This was derived directly from the effectiveness study.

The direct health services costs comprised both the cost of screening and any costs of additional testing that were required for a final diagnosis. Specifically, the costs included in the analysis were for screening and repeated XRM, screening and repeated breast CE MRI, ultrasound, core biopsy, MRI-guided biopsy, surgical biopsy and mastectomy. Staff costs, cost of equipment and further consumables were included in the analysis, but these were not reported in detail. The resource quantities were derived from actual data collected during the effectiveness study, complemented by authors' assumptions intended to reflect routine clinical practice, where necessary. Most of the cost estimates were based on actual data, with mean costs from five participating centres included in the study augmented by unit costs from official published sources. The unit costs were only reported for CE MRI, but the quantities of resources used were not reported separately. Equipment costs were adequately discounted. The costs were assumed to have been incurred during the first year, therefore discounting was not carried out. All costs were reported for the price year 2003.

The costs were treated stochastically.

In-line with the perspective adopted, the indirect costs were not included in the analysis.

UK pounds sterling ().

To adjust for sampling variation, a probabilistic sensitivity analysis was carried out using Markov Chain Monte Carlo (MCMC) methods. The analysis was based on a sample of 40,000 runs and was performed using the software WinBUGS. To account for uncertainty due to sampling variation, the net-benefit approach was adopted, employing a statistical model. The statistical model used and all probability distributions assigned were clearly reported. Multi-way sensitivity analyses were performed to investigate variability of the cost-effectiveness results on the sub-groups of women with a BRCA1 (BRCA2) mutation and women with a relative with a BRCA1 (BRCA2) mutation. Various one-way sensitivity analyses were also carried out to assess the robustness of the results to variability in the data. The main parameters tested were costs and resources used related to CE MRI. Specifically, the authors used cost and resource use estimates as indicated by the research protocol. The ranges used were from published sources or were based on actual data collected throughout the effectiveness study.

The reader is referred to the 'Effectiveness Results' section for the cancers detected.

The total costs per screen were reported (any costs of further investigations were included).

The mean total cost per woman screened was 41.7 (standard deviation, SD=9.3) for XRM, 304.0 (SD =204.1) for CE MRI, and 342.4 (SD=13.2) for the combination of XRM and CE MRI.

When the analysis was restricted to BCRA1 patients only, the mean total cost per woman screened was 43.7 (SD=89.8) for XRM, 323.0 (SD=230.0) for CE MRI, and 361.2 (SD=236.5) for the combination of XRM and CE MRI.

When the analysis was restricted to BCRA2 patients only, the mean total cost per woman screened was 55.7 (SD=149.7) for XRM, 317.5 (SD=249.2) for CE MRI, and 369.3 (SD =290.1) for the combination of XRM and CE MRI.

An incremental-cost-effectiveness-analysis was conducted.

The CE MRI screening method was dominated. The combination of XRM and CE MRI resulted in an additional cost of 28,284 per additional cancer detected compared with XRM alone.

For BCRA1 patients only, the combined screening strategy did not result in additional cancers detected. CE MRI resulted in an incremental cost of 11,731 when compared with XRM.

For BCRA2 patients only, the CE MRI screening strategy was dominated. The combined strategy resulted in an additional cost of 15,302 per additional cancer detected when compared with XRM alone.

Cost-effectiveness acceptability curves were produced from the net-benefit statistical model. The following results were obtained. Assuming a willingness-to-pay (WTP) of 20,000 per cancer detected, the combined screening strategy had a 0.07 probability of being cost-effective. When the WTP was raised to 30,000, the probability increased to 0.67. When a gamma-distribution was assigned to the costs, the equivalent probabilities were 0.06 and 0.65, respectively.

When the analysis was restricted to BRCA1 (BRCA2) women, the probability that the combination of CE MRI and XRM was cost-effective was 0.57 (0.82) at a WTP threshold of 20,000 and, 0.71 (0.96) at a WTP threshold of 30,000.

The sensitivity analysis demonstrated that the most sensitive parameters were costs related to the CE MRI test. Using research protocol CE MRI-related costs, the combination strategy resulted in an incremental cost of 44,564 per cancer detected and 22,890 when the analysis was restricted to BCRA2 patients only. When the analysis was restricted to BCRA1 patients only, the CE MRI strategy resulted in an incremental cost of 19,068 when compared with XRM.

Contrast-enhanced magnetic resonance imaging (MRI) might be a cost-effective screening modality for women at high risk, especially for the BRCA1 and BRCA2 sub-groups.

XRM was chosen because it was the recommended screening method in the UK. The choice of CE MRI as the comparator was justified on the strength of evidence in the literature, which demonstrated its higher sensitivity in comparison with XRM. You should decide if these interventions represent current practice or valid comparators in your own setting.

The analysis was based on a prospective multi-centre cohort study. The study enrolled women at high genetic risk of breast cancer and this was reflected in the authors' conclusions. However, it is impossible to comment on the internal validity of the study as the authors referred to a separate clinical paper for details of the clinical study. Some estimates of effectiveness were based on authors' assumptions, which were justified with reference to the literature or to current clinical practice in the authors' setting. Whilst the validity of these estimates is more difficult to determine, the robustness of the assumptions made was investigated in the sensitivity analysis and the results were reported in full.

The measure of benefit (i.e. cancers detected) was obtained directly from the effectiveness analysis (number of cancers detected using each of the screening strategies). This choice of benefit was justified as a suitable proxy for the sensitivity of each screening method. In addition, a net monetary benefit approach was adopted and cost-effectiveness acceptability curves were produced. This enables the likely probability of the interventions being cost-effective over a range of decision-makers' WTP thresholds to be established. All methods used to derive the cost-effectiveness acceptability curves were well reported.

The analysis of the costs was carried out from the perspective of the UK National Health Service. It appears that all the relevant categories of costs related to screening were included in the analysis. However, the consequences of treatment were not appraised, and their inclusion might have further strengthened the authors' conclusions. The costs and the quantities were not reported separately in the publication. Resource use was either taken from the effectiveness study or was assumed by the authors. The costs were treated stochastically and an extensive sensitivity analysis was conducted to assess the robustness of the estimates used.

The authors compared their findings with those from other studies, emphasising that methodological differences as well as differences in study populations should be accounted for when making comparisons. However, the issue of the generalisability of the results was not directly addressed. The authors do not appear to have presented their results selectively and the results of the sensitivity analyses were adequately reported. The authors also reported a number of limitations to their study. First, the recall rates were not investigated separately for each screening strategy as they were based on the combined results of the screening tests combined. Second, the lack of available data meant that the impact of the screening methods on quality of life was not assessed and, therefore, comparisons were restricted to those studies that applied the same measure of benefit in the economic analysis (i.e. cases of cancer detected). Adopting a societal perspective would have enabled the psychosocial costs related to MRI breast screening to be accounted for.

The authors did not make explicit recommendations for changes in policy or practice. They suggested that further research is essential in investigating the impact of early cancer detection on quality of life. In addition, as lower costs related to the CE MRI screening method are expected through technical improvements and wider use of the technology, future economic evaluations should be conducted to provided updated estimates.

Supported by a grant from the UK Medical Research Council.

Because readers are likely to encounter and assess individual publications, NHS EED abstracts reflect the original publication as it is written, as a stand-alone paper. Where NHS EED abstractors are able to identify positively that a publication is significantly linked to or informed by other publications, these will be referenced in the text of the abstract and their bibliographic details recorded here for information.

Leach MO, Boggis CR, Dixon AK, Easton DF, Eeles RA, Evans DG, et al. Screening with magnetic resonance imaging and mammography of a UK population at high familial risk of breast cancer: a prospective multicentre cohort study (MARIBS). Lancet 2005;365:1769-78.