La famiglia. As you might already know, our family expanded two weeks ago with the arrival with our newborn son. Mother and baby are well and happy. As with all other newborns in Germany, our son got a heel stick on his third day of life for newborn screening. When my parents visited the following weekend and the kids were in bed one evening, we eventually ended up talking about screening, genome, disease and the possibility to make predictions from your genetic data. Therefore, looking forwards on life from the perspective of a newborn, what could we learn from exome/genome data and do we want to know it?
No need for exomes in the family. No, I am not going to exome-sequence our son, in case you might have been slightly concerned. In this blog post, I wanted to revisit the issues surrounding disease screening and the possibility of using high-throughput genetic data for this. Screening – by definition – is a strategy to look for asymptomatic or unrecognized disease carriers in a population with the idea in mind that early recognition may help prevent or ameliorate the cause of the disease. This of course would be ideal for all disorders, and at least for the genetic diseases, why not full-genome-sequence everybody to anticipate disease and make us healthier? Why not use other tools such as imaging in addition? I vividly remember a discussion with a friend in medical school who was quite concerned by the fact that full body MRI scans are not offered to everybody on a regular basis. He was just on his oncology rotation and convinced that such a full body screening would help identify disease at an early stage and lead to a better outcome. You might already guess that things might not be that simple.
What does it take to screen for a disease? In 1968, Jungner and Wilson suggested criteria for screening programs on behalf of the World Health Organization (WHO). These criteria were recently adapted to the requirements of the genomic era. Amongst other important points, these criteria suggest that screening might be useful for a well-understood disease that can be treated and where early intervention has been shown to be effective. There are other criteria, which pertain to program monitoring, benefit/harm analysis and equity. The choice of words is also interesting. The WHO bulletin does not talk about screening methods but about screening programs. Establishing the diagnostic side with a new method is only one part, but every new screening method needs to have a framework of a comprehensive program for follow-up diagnosis, early treatment, monitoring. The surrounding program is much more expensive than the actual screening test, i.e. for every Euro invested in the actual technology, you might probably want to add 10 Euros for everything else that follows. The sequencing itself would be the least problematic part.
So what would we do if we suddenly started exome sequencing for screening purposes? I think the answer can be divided up into three different groups.
(A) Obvious findings. Most recessive disorders with known alleles would probably be identified including many of the metabolic diseases including in the newborn screening program such as Biotinidase deficiency or Phenylketonuria. However, as you would be moving back from metabolic endophenotype to genetics, some of the interpretation might become difficult. In addition, in congenital hypothyroidism, the disease that is most commonly identified in the newborn screening results, a genetic cause can only be identified in a subset of patients, thereby rendering genetics not suitable for screening.
(b) Questionable findings. Known pathogenic variants might be identified or other variants are found that are predicted to be pathogenic. This would necessitate a stringent follow-up diagnostic machinery for confirmatory testing. In addition, it can be expected that more than 80% of results will be of unknown significance, a difficult situation if this data has been generated for screening rather than for diagnostic purposes.
(c) Problematic findings. This screening might produce information that is controversial. For example, while both Huntingtin CAG expansion (Huntington’s disease) and ApoE4 alleles (Alzheimer’s) can easily be identified, there is an emerging consensus that the decision about generating this data should be left to the legally consentable individual rather than generating this data in minors. Particularly genetic testing for Huntington’s disease has generated an ethical dilemma. In addition, Nobel Prize winner James Watson asked for his genomic data to be blinded with regards to the ApoE4 allele.
Is this applicable to epilepsy? Would a broad population screening for Dravet Syndrome, Rett Syndrome, STXBP1-related encephalopathies make sense? The initial Jungner and Wilson criteria stipulate that such a decision should be based on the fact whether these diseases are relevant (yes), well understood (partially) and causally treatable (not yet). There is widespread hope that with early detection, these diseases can be treated early and that outcome can be improved. However, in most genetic epilepsies, the situation is differently from metabolic disorders, where much of the pathogenic effects can be attributed to a lack of a specific hormone (hypothyroidism) or the accumulation of a toxic metabolite (many recessive metabolic disorders), both of which can be causally treated. Early detection may help improve outcome, but conceptually, early detection is still different from broad population screening. Again, with many epilepsies, we are still in the “diagnostic stage” and so far have little evidence how an early diagnosis may help change patient treatment. I wonder whether this situation might be the ideal starting point to jump directly into neonatal exome screening for epilepsy genes. Genome centers are eager to devour DNA and turn it into sequence. However, within the framework of a screening program, this diagnostic part will only constitute a small puzzle piece. First show me that you’re capable of handling the data, the follow-up and the counseling of patients.
Incidentalomas. An incidentaloma is a tumor found by coincidence without clinical symptoms or suspicion, identified in up to 7% of patients older than 70 years. These tumors are often identified when imaging is indicated for a completely unrelated reason. Accordingly, radiologists are faced with several dilemmas. How can you demonstrate that this lesion is harmless? Sometimes, these accidentially identified tumors may trigger a long follow-up of control investigations. For genomic medicine, the term incidentalome was coined, referring to accidentially identified pathogenic variants.
Fish tonight? On a larger scale, the parallel between imaging and genetics might actually be quite revealing. While there is currently quite some confusion regarding genetic data, much of the discussion on incidentalomas has died down and protocols have been established. In addition, the public interest in full-body MRI has declined and similar phenomena can be expected after the first wave of the genomic hype is over. On Route 1 close to Concordville, PA, there used to be a full-body MRI place. Now it’s turned into a seafood restaurant.