There are different forms of genetic testing. The genetic testing given to people who seem to have a disease running in their family is most often a DNA sequencing test. While DNA sequencing has been available to many patients for more than three decades, technological advances in the last few years have allowed DNA sequence tests to become routine in many settings. These technological advances have also allowed clinical scientists to begin evaluating the accuracy of previously accepted variant interpretation methods.
In 2014, a large team of scientists published a key article in Nature Genetics exploring whether existing clinical genetic variant interpretations were based on solid evidence. These scientists investigated and corrected variant interpretations in Lynch Syndrome, which is a form of hereditary cancer and is one of the best studied hereditary diseases and has one of the world’s most respected public databases. The scientists independently reviewed the data and collectively downgraded 24% of the positive (pathogenic) classification entries based upon insufficient evidence.
Women with positive Lynch Syndrome variants will often choose to have a hysterectomy and/or remove their ovaries to preempt the high risk of developing deadly tumors, and both men and women will increase their frequency of colonoscopies. A single false positive variant interpretation often impacts numerous patients but the design of this study did not allow the scientists to evaluate the magnitude of the patient impact due to the unsubstantiated positives that were discovered and corrected by their study.
In a 2016 pre-publication white paper, scientists at “Human Longevity, Inc.” asked: If the positive variant interpretations issued by labs and expert committees are to be believed, how many “regular” people from an unselected population would be predicted to be at risk or have a hereditary disease?
They sequenced the DNA and compared the results of these everyday people to the known frequency of genetic diseases in the population using publicly accessible clinical variant classifications (May 2016 ClinVar download). The scientists hypothesized that if the positive classifications within the ClinVar database were all true positives, the frequency of hereditary disease predicted by sequencing the unselected group of people should be no greater than the known frequency of hereditary diseases in the general population (accounting for incomplete penetrance). If the genetically predicted frequency of disease in the unselected group were higher than the known frequency of disease, then the origin of the discrepancy would be false positives from inaccurate variant interpretations found in the clinical ClinVar database.
Human Longevity, Inc. was able to estimate the positive inflation ratio for numerous hereditary diseases due to inaccurate variant interpretation, which they expressed as [observed/expected]. This inflation ratio can be approximated as [(true positive persons + false positive persons) / (true positive persons)] in a tested population. It is important to note that Human Longevity, Inc evaluated only the variant interpretations provided by laboratories who volunteered their classifications into the public ClinVar database.
From their research, positive BRCA1 and BRCA2 hereditary breast and ovarian cancer (HBOC) variants demonstrate false positive rates are of only minor concern once incomplete penetrance is taken into account. However, BRCA1- and BRCA2-related HBOC is one of the most thoroughly studied hereditary diseases. For most other hereditary diseases with defined frequencies in the general population, the false positive rate is astonishingly high, even after accounting for incomplete penetrance.
For example, malignant hyperthermia susceptibility, multiple endocrine neoplasia type 1, Romano-ward long QT and Brugada syndromes, Ehlers-Danlos syndrome (vascular type), Retinoblastoma, and hereditary paraganglioma-pheochromocytoma syndrome all exceed an inflation ratio of 10. Even if we invoke a low 20% penetrance for these so-called positive variants, there is still a greater chance of a person having a false positive than a true positive when receiving a positive genetic test result.
The situation is much worse for rare diseases that primarily impact fetuses, newborns, and children, with inflation ratios exceeding 25x in most rare diseases, and in some cases exceeding an inflation ratio of 1000x. Since these rare diseases may not have very accurate estimates of frequency in the general population, it is possible that the inflation rates may themselves be somewhat overestimated.
CGI has not witnessed significant improvements in the genetic testing industry’s variant interpretations since the 2016 Human Longevity, Inc. study. Therefore, we conclude that for almost all hereditary diseases probed by Human Longevity, Inc., there were much higher odds of a positive genetic test result being false than being true, even when accounting for incomplete penetrance.
We note that Human Longevity, Inc’s study is very recent, and that replication studies will be required to verify or disprove their findings. Thus, CGI tempers its summary of these recent findings and states that more than half of positives are false positives due to inaccurate variant interpretations when averaged across the clinical DNA sequencing industry.
CGI exists to fix the inaccuracies in genetic variant interpretation across the entire industry by requiring a higher burden of evidence for variant interpretations. Until CGI can fix the problem at the source (being the laboratories), CGI offers patients and clinicians the CGI Variant FactChecker™ service, which provides CGI’s research-use-only second opinion service for variants reported by laboratories and companies. Please contact CGI if you believe there is an error in our methodology for this estimate.