geneIQ – Genetics and Diagnostics

The news broke out in March 2018 about mistakes discovered in results returned from DTC genetic tests. While it was not completely unexpected that there will be some mistakes in genotyping performed by various providers, the scale was definitely unexpected. Ambry Genetics, has published a paper in Genetics in Medicine, describing their findings on the topic. Their report found 40 percent of false positives returned back to the customers. While the sample size was relatively small, 49 patients in total, the findings are disturbing due to the scale of false positives. Two main issues were identified. One, incorrect calls in some of the genes. Second, probably more problematic, was incorrect classification of found sequence variants as pathogenic mutations. Classification was performed by DTC companies or by a third party interpreting services.

Some discrepancies between genotyping and sequencing can be expected. The reason is that, genotyping can be compared to “spot testing of DNA” (Genetics in Medicine). Sequencing, on the other hand, gives multiple passes across the given fragment making it more accurate. Many laboratories around the world used genotyping assays for years with great success and reliability, due to standardized assays and methods. So what is different here? Methodologies used by different companies vary substantially as does actual assay design. In many cases exact details are not released by companies.

Based on the publication by Ambry we do not know if any of the DTC companies performed better than others. A systematic comparison of same samples across the different tests available would help in restoring some confidence in quality of raw data produced. Such approach has been already used with microarray  and next generation sequencing platforms. A MicroArray Quality Control (MAQC) Project was conceived to address data reliability initially just for microarrays (Nature Biotechnology 2006). However, the idea has progressed since and in its third stage focused on Next Generation Sequencing (NGS) platforms (Nature Biotechnology 2014). The fourth stage (MAQC-IV or SEQC2), currently ongoing, is focusing on developing standards for analysis of NGS data. Maybe it is time to introduce a similar exercise for genotyping platforms so we can improve standards and reliability for consumer genotyping tests.

There is a clear advantage to genotyping tests such as those offered by 23andme. They are relatively cheap and fast, and can offer a good glimpse into a person’s genome. Now to be really helpful to the consumer, there is a need to share the variant information and use well-curated variant databases. Classification discrepancies can occur of care is not taken with how the data is analyzed.

What does all this mean to the consumer?

Reporting health related findings by DTC companies was initially halted by the FDA and 23andme got itself into trouble with the agency for providing such information. However, the company bounced back in April 2017, as it was allowed to return some of the health findings back to the customer.  Following this, an FDA guidance from November 2017, stated that genetic health risk (GHR) testing will be exempted from premarket review under certain conditions. Such tests will require a one-time review with the agency. This is aimed to allow better access to evaluation of health risks by the general population and improvement of their health management. In March 2018, 23andme got an authorization to return BRCA1/BRCA2 results back to the consumers. In view of the recent findings the obvious question is: “Was the FDA right in granting the right to return health risk results?” And exactly what health results can be returned by the DTC genetic testing companies? And remember that we are not talking just about 23andme.

While we can all agree that the aim of such tests is great and they seem to pose relatively low risk, since no actual diagnosis is returned to a customer. So far so good, however, as Ambry Genetics discovered there is a problem. Although genetics seems simple and basics are not very complicated. Health risk calculations are not as simple as they look. Returning health risks back to a consumer can be problematic.

What could be the problems?

1. Misunderstanding of the information by a consumer
2. Misinterpretation of the information by a health professional not trained in genetics
3. Consumer making health-related decisions on one only test and not talking to a genetic counselor or a medical geneticist
4. Inappropriate interpretation of a presence or absence of a sequence alteration (known as sequence variant) in comparison to a human reference genome, leading to mistaking a benign change for a pathogenic one
5. Risk calculations are based on population frequency of a particular variant. Due to under-representation of some ethnic groups frequencies may not be accurate for a given variant found in a customer
6. DTC tests provide us with only a snapshot of our genome. It is extensive for the most part, designed to give a lot of good information but it is not as exhaustive as genome or even exome sequencing.
7. The snapshots offered by DTC tests are not all the same, depending on the panels used they can be quite different.

It seems that misinterpretation or misunderstanding can be easily tackled with appropriate help by trained professionals. This leads me to a recent article in STAT on a need for interpretation of data. Do we need an interpretation of the data? Not according to 23andme CEO, Anne Wojcicki. She argued in this article, that customers do not need experts to interpret the health risk results from 23andme. Argument is centered around BRCA1 and 2 mutations. In my opinion there are two different parts to the argument she presents. One is the access to the test itself, the second is actual understanding what the test result means for an individual.

I agree that there is a need for easier and better access by public to genetic testing. The ability to perform such tests should not be limited and prevented by insurance companies or physicians. We should be able to ask for a test we want or feel we need. People, whose relatives have been diagnosed with breast or ovarian cancer are frequently told that they are not eligible for free or subsidized screening for one reason or another. This test can break those barriers and help these individuals to assess their risk, especially if a relative had a one of the three mutations screened for in the test.

However, I disagree that we do not need experts to interpret the outcome of a test. As a geneticist I would understand what is in a report, but how this result applies to me based on my family history and my own health history is a different thing. I would want to talk to a genetic counselor to make sure that my understanding is correct. There are a few reasons for this:

1. Three variants tested for are mostly found in individuals of Ashkenazi Jewish ancestry in 23andme test. So for this group it is a perfect test. Simple, quick and easy, no need for a doctor.
2. Negative result in this test cannot eliminate a possibility of you developing breast cancer, especially if you are from a different ethnic group. This disclosure is made in the report, there is also a statement that not all variants are tested for and this test does not diagnose breast cancer. However, this information can be easily missed unless its placement is prominent in the report.
3. What would be the best test to do in case I wish to have one done? What options are out there?
4. BRCA1 and 2 genes are not the only contributors to breast and ovarian cancer, so while somewhat useful this is not a test that can really determine your breast cancer risk.

As with any genetic test, as I said in a few of my previous posts, think about what do you want to get out of the test? This is the most important question as the same price can buy you very different information, for example a much more comprehensive look at the BRCA1 and 2 genes from Color Genomics. This particular test as well as tests from Veritas Genetics require a physician’s referral but that can be arranged by the testing company. And yes, you need a referral from a doctor, which you can avoid by going to 23andme or a similar DTC company. However, you get a completely different information based on more than three gene variants. Also your health risk is based not just on presence of mutations, but also your family history and personal health information.

Now we have come a full circle to the start of this blog piece. The issue was that there are problems with accuracy of the variant calls and interpretation of the variants by DTC companies. So considering the mistakes that are being made should you still do the tests? Will more regulation fix the problem? Or rather what would fix the problem? I am not sure that we would have a one answer to this question, various groups and individuals would have different opinions on how to fix it. While this is being worked on, nobody is going to wait for the outcome, so the question is,

What should you do as a consumer?

• Be aware of the problem
• Find out of if the raw data can be provided back to you
• Have a chat with your GP or a genetic counselor ahead of the test if you are looking for health risk information or any health related results
• Ask your chosen company to give you some more information on how the test is done, how they ensure the accuracy of the data. Companies should be prepared and able to discuss the test with you
• You can ask about the lab accreditation as well. Labs with CLIA and/or CAP accreditation have to follow certain standards. While this is not a guarantee of perfection, it does ensure highest standard of quality
• If your result indicates an increased risk, make sure this is confirmed before you make any serious decisions based on the obtained information. DTC testing often contains a disclaimer that this information is not to be used for medical purposes, so make sure you do not use it in such a way.

Check out the video from Ambry Genetics explaining some of the above in a short and clear way, in just about a minute.