Genome sequencing for all babies? We are not there yet!
For decades, leading scientists have predicted that soon, very soon, genome sequencing (GS) of all newborn babies will allow diagnosis and treatment of rare diseases. We are not there yet. I recently explained some of the problems in an article co-authored @AnnieJanvier and Keith Barrington. We focus on the lack of evidence that GS leads to better treatment, the ambiguity of false positives, the unconvincing data about cost-effectiveness, and raise some concerns about purported non-clinical benefits.
1. How often does GS lead to better medical treatment? It is surprisingly difficult to know. There are inspiring case reports, both in lay press and peer-reviewed literature, but many of them describe situations in which the diagnosis or treatment could have been made without GS. Furthermore, the most common outcome reported is not “better treatment,” with documentation, but, instead, the peculiar variable called “change in management” as evidence that GS is useful. The distinction is important. The best case for GS would be if it frequently leads to the diagnosis of a previously unsuspected condition for which there is an effective treatment exists. This rarely occurs even when GS is used in sick babies for whom the diagnosis has remained elusive. It is even less likely to occur in less sick or healthy babies. Change in management is much more ambiguous. We suggest, in our paper, that reported changed in management would often occur anyway by the very trial-and-error nature of clinical management for critically ill babies for whom the diagnosis is uncertain. Doctors try things. Sometimes they help. But, given the nature of intensive care, it is hard to know whether the change in management was either a) directly related to the GS result; b) based on a GS diagnosis that could not have been made in other ways, and c) beneficial. We conclude that from presently available research reports, it is difficult to say how often the genomic diagnosis was either crucial or beneficial.
2. How often are genomic diagnoses false positives? The answer not clear. We know that people with identical genomic findings can have very different clinical symptoms. If we knew the prevalence of variants in asymptomatic people in the general population, we could estimate true and false positivity rates and estimate the positive and negative predictive value of the test. But, with genomics, there are so many variants, they vary among different ethnic groups, and we often do not have the population data to say how often they occur in the absence of symptoms. When that is studied, we usually find that they occur more often than we thought. It is difficult to know when such false positives are the result of complex gene-gene interactions, incomplete penetrance, mild or late-onset disease, or other factors. But to treat genomic findings as precise diagnostic results is dangerous and misleading.
3. How often does GS lead to cost-savings. Another tricky question. Because it is hard to know whether the changes in management are beneficial, it is hard to know whether they save money or generate excess costs. Some studies suggest that genomic diagnosis leads to both earlier discharge or to earlier decisions to withdraw life-sustaining treatment. But, given the heterogeneity of most genetic syndromes, such clinical decisions should not be based on the molecular diagnosis but, instead, should be made based on the clinical indications for discharge or palliative care. We note that overconfidence in the prognostic accuracy of genomic diagnoses may lead to a reticence to withdraw life-support without such a diagnosis and suggest that further study is needed to clarify the appropriate role of genomic diagnoses in decisions about life-support.
4. Does GS have non-clinical benefits. Many studies report that a genomic diagnosis can give parents an “answer” to their questions and doubts about the cause of their child’s condition. Many parents report that a genomic diagnosis allowed them to find support from other families with children who had similar conditions. In some cases, genomic diagnoses allow identification of other family members who may be at risk. There are also potential harms. Many parents hope that GS will reveal a curable condition. It rarely does. Sometimes, a GS result leads to the erroneous conclusion that a baby will quickly die. This can prevent parents from bonding with their child, a problem that is already common in the NICU.
Our conclusions: GS is here to stay. Some babies and families benefit from the GS results. GS is most useful as a diagnostic test in critically ill babies, rather than as a screening test in less sick or healthy babies. At the same time, population testing is essential for refining our ability to interpret results. GS results often confirm an already suspected dismal prognosis. Anticipatory discussions will give parents a realistic understanding of the likely impact of a genetic diagnosis. Both doctors and parents should recognize that the clinical use of diagnostic GS in newborns is very much a work-in-progress. We are only a decade into the age of true genomic medicine. We have much to learn.