Barely a day goes by without a news story or social media post proclaiming that the $1000 genome now exists, and is ushering in a healthcare revolution. Every day, somebody, somewhere in the world, posts these graphs on Twitter. There’s even a Wikipedia page devoted to this topic. It’s a persistent news headline and, frustratingly, it’s currently wrong.
I’ve blogged previously about the real costs of sequencing, noting in particular that the costs of clinical interpretation are likely to be high, and have mostly been ignored to date. I also noted in my previous blog post that:
“There are three key drivers of the cost of sequencing: the assay itself, the bioinformatics analysis required to make sense of the basic test results, and the costs of generating the clinical evidence base, conveying results to patients and making decisions. Of these, only the assay cost is currently falling. The other cost drivers are either static or increasing”
Now, Kathryn Phillips and colleagues have made a further useful contribution to this topic with a recently published article titled “Is the “$1000 Genome” really $1000? Understanding the full benefits and costs of genomic sequencing”.
Phillips and colleagues highlight three reasons why the $1000 genome is unlikely to exist: incidental findings (which may lead to unnecessary tests or treatments which increase costs without improving outcomes), the identification of novel germline mutations (potentially leading to a cascade of further sequencing tests), and the actions that are taken based on test results (i.e. additional testing and treatment after sequencing, the costs of which may overwhelm the direct costs of sequencing). They then go on to describe several scenarios which illustrate these points quite neatly. Finally, some simulations are presented which demonstrate the impact of incidental findings on the cost-effectiveness of genomic testing in the context of Lynch syndrome screening. These results indicate that the cost per QALY gained will vary dramatically depending on whether incremental findings provide net benefits/savings or net harms/costs, with some scenarios yielding ICERs of several hundred thousand dollars per QALY gained, well above the cost per QALY thresholds used by decision-makers in healthcare.
Due to space constraints, these simulations are simple, and more comprehensive estimates are available elsewhere. In addition, other important cost drivers such as bioinformatics and clinical interpretation are not explicitly considered. However, even with these caveats in mind the main conclusion (“the true cost of sequencing extends beyond the test itself and thus the headline of the “$1000 Genome” is not an accurate portrayal of its cost”) holds true. It is pretty clear that at present, we simply do not have enough evidence on the benefits and costs of sequencing to be able to definitively say that the $1000 genome exists and is a good use of limited healthcare resources.
However, we can begin to consider in which clinical contexts the benefits of sequencing are most (and least) likely to exceed the costs in the near future. Phillips and colleagues identify several conditions which will need to be present for this to be the case. One of these is that the additional information provided by sequencing needs to be actionable in some way e.g. resulting in a change in treatment or behaviour, or the use of some intervention. I think this is broadly true, but not wholly true, and it has been demonstrated previously that patients do value tests that only provide non-actionable information. It can, however, sometimes be tricky to effectively demonstrate that genomic tests which provide non-actionable information are cost-effective in the conventional cost per QALY framework, and more work is needed here (by both researchers and decision-makers) to consider the feasibility of alternative approaches.
Overall, the relentless focus on the goal of a $1000 genome worries me, because it has the potential to focus research efforts and scarce research funding on the wrong activities. It’s obviously desirable to reduce the cost of generating genomic information. However, we are soon going to reach the point at which further reductions in these costs are of limited benefit because of bottlenecks in bioinformatics, clinical interpretation and also health economics. Ultimately, this will slow the translation of these technologies into clinical practice and thus have a real effect on population health. To avoid this scenario, funders and other institutions driving the global research agenda need to act now to change the focus from the $1000 genome to the $1000 genomic test package.