Content of review 1, reviewed on June 12, 2017

This review was originally published as a blog post on the Health Economics and Genomics blog on June 12, 2017.

It is now well documented that health economic evidence to inform commissioning decisions regarding genomic tests is in short supply. This lack of evidence relates to both costs and health outcomes – there is perhaps an understandable tendency to focus on the issues surrounding the measurement of health outcomes in genomics, but data on costs is equally sparse and the generation of such data is also beset by practical and methodological challenges. That said, in the past twelve months we have started to finally see some good quality data emerging on the costs of whole genome and whole exome sequencing, and a recent paper by Kate Tsiplova and colleagues has made a notable contribution to this literature.

The aim of the paper was to estimate the costs of whole exome and whole genome sequencing (WES and WGS) compared to chromosomal microarray analysis (CMA – current practice) in a population of Canadian children with autism spectrum disorder (ASD), then use these costs within what the authors describe as a cost-consequence analysis (which felt to me more like a cost-effectiveness analysis as the incremental cost per additional positive finding was calculated).

Test costs were calculated using a microcosting approach, and were combined with a variety of other costs that would likely be incurred during the diagnostic journey of a child with ASD to estimate the total program costs to service an ASD population over a five year period. My understanding is that the authors essentially calculated this cost for a hypothetical ASD population with specific characteristics. While I think this is an informative exercise, my issue with this approach is that it overlooks a lot of the potential heterogeneity in the diagnostic costs of these patients. For some patients (e.g. those with genetically complex ASD), the costs of diagnosing their ASD could run into millions of dollars/pounds over a long period of time, and it is important to know (a) exactly how much these diagnostic costs are, (b) when these costs are incurred, and (c) whether these patients have specific characteristics. This information can then be used to determine the appropriate point at which to offer WES/WGS. The approach taken in this study potentially obscures these details.

That point aside, overall the costing work in this paper is very detailed and precisely the sort of evidence that we need to generate for genomic tests going forward. Most of the key categories of costs are included, with a couple of exceptions. The authors themselves note that training and implementation costs were excluded from this analysis, which is a limitation of this work. I think it is also important to note that software costs also appear to be excluded – these can run to several thousand dollars/pounds, and become increasingly important if sample throughput is low.

In terms of the costing results, the estimated annual costs per ASD sample were (all Canadian dollars) \$744 for CMA, \$1,655 for WES, \$5,519 for WGS using the HiSeq 2500 and \$2,851 for WGS using the HiSeq X. The costs of CMA and WES feel about right to me, based on current knowledge in the UK, although the costs of WGS are possibly on the low side. Regardless, the WGS costs further support the view that we are not particularly close to the $1,000 genome (assuming that this threshold refers to the cost of using WGS to generate information that can usefully inform clinical practice, rather than the [much less informative] cost of just doing the sequencing).

The economic evaluation compares the incremental costs and diagnostic yield of several hypothetical clinical scenarios to those of current practice. Many of the hypothetical genomic testing scenarios appear to be cost-effective e.g. the incremental cost per additional patient with a positive finding is \$58,959 for the comparison between WGS (HiSeq 2500) and CMA. However, the results are sensitive to the assumed diagnostic yield, which is uncertain. More generally, the authors go on to discuss which test or combination of tests should be recommended based on these results, and do not make a firm recommendation. This is a good demonstration of a key issue encountered in economic evaluation work in this area: it is usually only possible to evaluate a subset of all potential testing strategies in such work, and this adds an additional layer of uncertainty to results that are highly likely to be underpinned by variable data on costs and test outcomes.

The authors go on to note that these cost estimates are likely insufficient alone to fully inform commissioning decisions regarding WGS and WES, as such decisions require information on the costs to a health region or jurisdiction as a whole. In some cases, expensive steps of the testing pathway may need to be centralised in order for testing to represent a cost-effective use of limited health budgets. This requires an extended approach with a wider costing scope and (almost certainly) a lot more data. Such data is likely to be required sooner rather than later as large scale sequencing projects such as the Genomics England 100,000 Genomes Project reach their conclusion, and I expect to see more of these large scale costing studies being published in the next 2-3 years.

The final key point to make on this topic is well stated by the authors in their comprehensive discussion:

“Prospective economic evaluations are also needed to assess the impact of [WGS and WES] on the pathway of care for children with ASD and to weigh the costs of the care pathway against ultimate improvements in health outcomes as a result of testing”

It cannot be overstated how important it is to be able to accurately and reliably link improvements in intermediate measures of genomic test performance (e.g. diagnostic yield) to health outcomes that are comparable across diseases contexts and which can usefully inform commissioning decisions at the national level. Without good evidence on these links, it continues to be difficult to argue a case for the more widespread use of genomic tests, even in light of an improving evidence base on test costs.

Source

    © 2017 the Reviewer (CC BY 4.0).

References

    Kate, T., M., Z. R., R., M. C., J., S. D., L., P. S., Daniele, M., J., Y. E., L., S. W. W., W., S. S., J., U. W. 2017. A microcosting and cost-consequence analysis of clinical genomic testing strategies in autism spectrum disorder. Genetics in Medicine.