Much of what we read about precision medicine gives the impression that it provides certainty. That’s a false impression. As David Hunter pointed out in a recent article in The New England Journal of Medicine, it may actually increase uncertainty.
He uses breast cancer gene studies as an example. New tools have extracted information about mutations, epigenetic events, and gene-expression alterations in tumors. Products like Oncotype DX, MammaPrint, PAM50 and others contribute information that goes beyond the traditional clinical prognostic factors like tumor size, grade, and nodal status. The new products are based on evidence about the risk of tumor recurrence, but no direct evidence about which specific treatments were more effective.
A study by Cardoso et al., published in the same issue of the NEJM, looked at the possibility of using the 70-gene MammaPrint test to select patients for adjuvant chemotherapy. They stratified patients into high and low clinical risk (C-high, C-low) and high and low genomic risk (G-high, G-low) and randomly assigned them to either get or not get adjuvant chemotherapy. In the C-low, G-high group, there was no difference between those who did and didn’t get chemotherapy; the genomic information was not helpful for patients at low clinical risk. In patients at high clinical risk, the results were equivocal: forgoing chemotherapy on the basis of low genomic risk resulted in a slightly worse outcome, but the difference was small, only 1.5 percentage points, and the study was not powered to determine whether it was statistically significant. This leaves women at high clinical risk with a difficult decision: if their genomic risk is low, they could avoid the toxic effects of treatment, but it might slightly increase their risk of recurrent cancer.
It’s good that studies like this are being done, but the results don’t provide certainty. In this particular study, dividing a continuous variable into two groups (high and low risk) sacrificed precision for interpretability. Different gene-expression products may result in different risk categorizations, and they improve as technology changes and more data becomes available; the MammaPrint data is 14 years old.
Cancer genomics data is problematic for a number of reasons. Different studies find different information. The raw sequences are usually not published, so results remain ambiguous and questionable. There are many variables, like changes in gene copy numbers and tumor sub-clones, that are not taken into consideration. Precision medicine doesn’t really tailor the treatment to the individual; it lumps large numbers of patients together in order to reach reproducible conclusions.
In the future, we are likely to face a potentially bewildering array of probabilities — estimates of disease risk based on inherited germline sequencing and, once a disease is diagnosed, of prognosis and therapeutic options guided by “-omic” and other analyses. Assessing and acting on these probabilities will require approaches to data presentation, risk quantification, and communication of uncertainty for which we are largely ill equipped and that we already struggle with. In most situations, the best advice will be far from obvious and will often rely on a preliminary estimate as the data mature. In parallel to developing the tools for “-omic” analyses, we urgently need to develop methods to help our patients absorb large amounts of complex information that will help them make choices among increasingly numerous options with increasingly numerous trade-offs. These methods should also help our colleagues answer the age-old question, “What would you do, doctor?”
Conclusion: uncertainty remains
In the early days of genome analysis, the promise seemed infinite. We hoped to pinpoint the genes responsible for every illness. We hoped to find certainty and solutions. We are not there yet, and we are beginning to question whether we ever will be. Uncertainty has always been a part of medicine, and it will continue to be. Precision medicine has great potential, but so far it may only be increasing our uncertainty and making decisions more difficult.
This article was originally published in the Science-Based Medicine Blog.