Uncertainty is an inherent aspect of medicine, but nowhere is this more apparent than in genetic medicine. Interpreting genetic data in individual patients is extremely challenging, largely due to the enormous amount of normal genetic variation that plays no (apparent) role in disease. Some of the uncertainty will improve with time – we will learn more, sequence more genomes, find more genes, and develop better tools. But some of it is here to stay, which has important implications for communication.
A recent paper in BMC Medical Genomics, Known unknowns: building an ethics of uncertainty into genomic medicine, explores the questions of how we think about uncertainty in genomics and how we should deal with it. Far from shrinking from the problem, the authors recommend that “uncertainty should be appraised, adapted and communicated about as part of the process of offering and providing genomic information.” This presents an important challenge – uncertainty is generally considered to be bad, something from which patients should be protected. But, in an era of increasing clinical DNA sequencing, it is simply unrealistic and unhelpful to assume that we can protect people from the inherent uncertainties involved.
Types of uncertainty
Uncertainty arising from imperfect or unknown information can arise either in clinical diagnosis or in clinical prognosis. Several different types of uncertainty are identified in the paper:
- Probability – future outcomes are uncertain, e.g. 50% risk of developing cancer.
- Ambiguity – information is imprecise, unknown or conflicting, e.g. novel variant identified.
- Complexity – making sense of result is challenging, e.g. phenotype does not fit.
The probabilistic nature of prognostic testing – where we seek to predict the future development or course of disease –is unlikely to change, no matter how much data we have. The future is inherently uncertain, and our future health is not hard-wired in our genomes. It is sometimes possible to take actions to reduce our future risk of disease, but most conditions are so multifactorial that we can’t know the true cause(s). All we can do is provide clinicians and patients with information to make the best informed choices possible.
Unfortunately, the situation for diagnostic testing – where we are seeking to explain current disease, rather than predict the future – is actually not much better. The ambiguity with which we currently issue genetic results is substantial, and originates from many sources. The relevant gene may not have been assayed; the data quality may be insufficient to find the diagnosis; the relevant variant may have been inappropriately excluded; the result may be so novel that we don’t know how to interpret it. Moreover, there is scant agreement over how to systematically evaluate this uncertainty or how certain we need to be before using the result of a genetic analysis in the clinic – for example, to guide treatment or offer prenatal testing. Part of the aim of the TGMI is to develop tools and systems to reduce this type of ambiguity.
Finally, there is no getting away from the fact that biology is inherently complex. Genetics and genomics give us unique windows into this complexity, but we are really just scratching the surface of disease causality. There are lots of reasons why a known disease-causing genetic variant may be observed in an apparently healthy individual and these can be almost impossible to tease apart. This inherent complexity must give us reason to pause, and to be wary of overdiagnosis.
Awareness of genetic uncertaintyNew DNA sequencing technologies have already revolutionised scientific research, and are now transforming clinical medicine, particularly for the diagnosis of rare diseases. But in addition to celebrating our successes, we must take care to ensure that patients and the public understand that there is still enormous uncertainty surrounding genetic information. Rather than the simplistic black-and-white concept of genetic determinism, the reality in most cases is a rather murkier world of genetic uncertainty.