Saturday, November 30, 2013

23andNotMe: A critical evaluation of the limitations of 23andMe and traditional genetic testing

My older sister Heather, who is also a BRCA2 mutation carrier, recently ordered the 23andMe test that’s been in the news recently. As you probably heard, the FDA is trying to shut the company down for making inaccurate claims about the test’s ability to assess genotypic health risks. In a letter to the CEO of 23andMe, the FDA writes, “Some of the uses for which PGS is intended are particularly concerning, such as assessments for BRCA-related genetic risk and drug responses (e.g., warfarin sensitivity, clopidogrel response, and 5-fluorouracil toxicity) because of the potential health consequences that could result from false positive or false negative assessments for high-risk indications such as these. For instance, if the BRCA-related risk assessment for breast or ovarian cancer reports a false positive, it could lead a patient to undergo prophylactic surgery, chemoprevention, intensive screening, or other morbidity-inducing actions, while a false negative could result in a failure to recognize an actual risk that may exist.”

The FDA is clearly trying to set a precedent for how to handle the new availability of over the counter genetic testing. I’m a huge proponent of personalized healthcare, so I was tempted to defend 23andMe. But Heather’s 23andMe results claim she has a normal risk of breast cancer, and even with my experience analyzing this kind of data it took me a full day to figure out why. 

Ultimately I've come to a few conclusions:

* 23andMe doesn't screen for my family's mutation or the vast majority of BRCA mutations.

* The incredible diversity of mutations that can occur in the BRCA genes makes it difficult to study specific types of mutations.

* Traditional genetic testing is much more reliable than 23andMe, BUT geneticists still struggle to fully understand the impact of most BRCA mutations on breast cancer risk. 

And Here's why:

To back up a little, Heather had purchased the test out of curiosity, interested in our family’s genetic heritage and health. We expected her results to indicate a high risk of breast cancer and maybe a few other interesting tid bits. But the results were more shocking than that: They showed nothing. Nada. Niente.

Heather’s DNA results indicated that she has an absolutely average risk of breast cancer, despite the genetic test she had years ago that indicated she has a deleterious mutation in BRCA2. The website suggests there are three main mutations associated with early-onset breast and ovarian cancer and gives no indication of just how many mutations 23andMe leaves out.

Given that my sister and I both had preventative double mastectomies based on our knowledge of our BRCA2 mutations, the 23andMe results raised some major questions. Had we been misled into thinking we were at such a high risk for breast cancer? Who was wrong, our genetic counselors or 23andMe? Being the genetics and data nerd that I am, I naturally launched a full investigation. Yeah yeah, I’m a dork, bear with me.

I had Momma Shaw dig up our family’s original genetic testing results, conducted by Myriad before it lost its gene patents (by the way, yay!). I hadn’t looked at the results since I was first diagnosed nearly six years ago. After a year working with a team at Purdue whose sole job was to catalog epigenetic mutations in breast cancer genes, I expected to be able to understand the language used in the report. Wrong.

I sat with four pages sprawled in front of me, indicating that my mom, grandpa, sister, and I all tested “positive for a deleterious mutation.” The “result” showed a deleterious mutation in “K1026X (3304A>T)” and another mutation in “K1025N (3303G>T)” which was of “uncertain significance”. Mutation K1026X, the results read, 
“may confer as much as an 84% risk of breast cancer and a 27% risk of ovarian cancer by age 70.”

Okay, great, minor flashback to the day I found out my boobs were evil. BUT the more important point was that I'd never seen mutations identified in this way.I needed to understand what our specific mutation meant and how to find it in Heather's 23andMe data. But in hours of research, I couldn't find any studies on our mutation or indicating that it was deleterious.

Finally, I learned that there is an entirely separate nomenclature for BRCA mutations. Complicating matters more, this nomenclature has been changed multiple times so the same exact mutation can be titled two separate ways (Ex. “c3303G>T” or “c3075G>T”). Data sources like 23andMe and NCBI record the position of a mutation or the unique SNP designation as a point relative to the entire genome, whereas BRCA mutations are named as points relative to the gene alone. I needed to translate between these scientific languages to locate the position of our family's mutation in 23andMe data.

Fortunately, the gods of nerdom directed me to the Breast Cancer Information Core (BIC), governed by The only standardized resource I could find, BIC provides a comprehensive database of every single mutation and polymorphism that has been documented in the BRCA1 and BRCA2 genes. Within this database, I finally found our family’s genetic mutation listed clearly in all relevant scientific ‘languages’ if you will. CAN I GETTA HALLELUJAH?

Within the BRCA2 mutation database, our mutations “K1025N” and “K1026X” were listed along with their “Approved Nomenclature” (c.3075G>T and c.3076A>T) and their “Traditional Nomenclature” or nucleotides (3303A>T and 3304G>T). Sure enough, Mutation K1026 is considered clinically significant but has only been observed four times! No wonder I couldn’t find any studies done on it in any scientific journals. There are so many ways that the BRCA genes can mutate that it’s incredibly difficult to study any given mutation in great depth.

The SNP data I got from BIC finally allowed me to search Heather's 23andMe data for our specific mutation. I thought, ‘Maybe the mutation is in the data even if they don’t provide a report on it…”

The search... turned up nothing. 23andMe didn’t even record the area of Heather’s DNA where her unique BRCA2 mutation occurs.

To recap the frustration:
1)     Although my family’s genetic mutation has been deemed deleterious, 23andMe does not screen for it.
2)     There are countless mutations 23andMe doesn’t include (just kidding, I’m totally gonna count them in a minute)
3)     Of the BRCA gene mutations scientists have identified, we really have no idea what impact most of them have on a person’s risk for breast cancer (watch me define ‘most’ riiighht about now)

Given access to BIC’s data, I had to know… 

Just how many mutations have been identified that aren’t included in 23andMe’s test? And of those mutations, how many of them do we really understand?

Welp, here’s the answer.
*Spoiler alert, I’m gonna get really pessimistic before busting out the glass half full mentality.*

  • BIC documents 1776 unique BRCA1 mutations and 1998 unique BRCA2 mutations.
  • 45% of identified BRCA1 and BRCA2 mutations have been found to increase a person’s risk for breast cancer --BUT we really have no idea how risk varies by specific mutation type.
  • 2.9% of the mutations observed in BRCA1 and BRCA2 have been found to be completely and utterly meaningless— Great! If you have one of those mutations, you’ve got nothing to worry about.
  • (BEWARE, DEPRESSING PART à) A whopping 51.5% (“MOST!”) of documented mutations are of unknown clinical importance – AKA we have no idea how dangerous they are, if at all.

Okay so what? Critical Problems...
1)     Even professional genetic testing is still very limited. Unfortunately, we can’t definitively say whether or not more than half of BRCA gene mutations increase a person’s risk for developing breast cancer.
2)     23andMe screens for 3/3664 possible mutations in the BRCA1 and BRCA2 genes that could be associated with higher risk for breast cancer. That’s only 0.08 of one percent of the ways the genes could mutate to increase risk!
3)     I suspect a lot of people may mistakenly interpret a test result like Heather’s as reassurance that they’re not at high risk for breast cancer… and that’s dangerous.

In summary, the Bad News:
23andMe is not a substitute for a thorough, professional genetic test and discussion with a genetic counselor. Although it’s undeniably fascinating (and maybe appeals to the little narcissist in each of us…), 23andMe is best used to satisfy curiosity, not as a means of predicting health outcomes. 

And as promised, the Good News:
If you’re reading this blog, you're probably already smarter than the average bear when it comes to knowing your hereditary risk for breast cancer. You get to kick ass and take names and protect your health however you see fit. Most important, even though we've got a long way to go, we're closer and closer to understanding individual levels of risk. 

The Future (As Envisioned by the All-Knowing Me =P)
BIC is a great example of our progress in organizing our knowledge on hereditary breast cancer. Over time we’ll start to widdle down those ‘unknowns’ and identify which mutations definitively increase risk for breast cancer. Then we can begin to identify how much each unique mutation increases a person’s risk. Eventually, being diagnosed with a BRCA mutation won’t involve an ambiguous risk assessment of 55-85%. One day (according to me and other optimistic scientists) genetic tests will be able to provide each woman with a personalized assessment of risk. Imagine how much easier it would be to navigate your preventative options if you had a more definitive assessment of YOUR personal risk for breast cancer. We’re not there yet… but we’re getting there.

Hope this helped some of you who’ve been as confused as we have been with the whole 23andMe frenzy. As always, questions and comments welcome! <3