Monday, July 13, 2009

The Grand Opening - Looking Inside My Genetic Crystal Ball

When I tell people that I completed a 23andMe test, they all seem to have the same reaction. The question they ask is – so when are you going to die, and of what disease?

I am supposed to know enough about interpreting genetic information to prevent me from responding the same way. However, I have to admit that deep down the same emotionally-based thoughts did exist. Because of our imminent mortality, it is natural for us to ask whether we will learn something about our futures from these types of tests.

After sending in the spit tube and waiting a couple of weeks, I received an email announcing that my results were ready. I was instructed to login at the 23andMe website to an account I had previously created when I ordered the test kit. Once logged onto the site, I found a welcome screen that was filled with links for announcements, new articles, new surveys, and recent community posts. A long main menu block was displayed on the left side of the page, and at the top of the menu was a category titled “health and traits”. The first selection under this category was Clinical Reports. Clicking here would finally open the “secret vault” containing my genetic information.

My Elevated Risk

The initial screen presented after clicking Clinical Reports divided the report into four sections - disease risks, carrier status, traits, and drug response. Under Disease Risks I saw a sub-list of five diseases, but I quickly clicked the link that said see all 10 risk reports … Another screen appeared with 23andMe’s list of these ten diseases divided into three risk levels – Elevated Risk, Decreased Risk, and Typical Risk. (For this post, I am sharing only the Elevated Risk section. Various parts of the complete report will be shown over multiple posts.

Crohn’s disease risk is 3 times!

Wow. So for me Crohn’s disease was the big winner. Interesting. Not something I would have guessed. At first glance, a 300% percent increased risk sounded serious, but then I saw the absolute risk was only 1.5%. Whew. (It dropped even further when I used the odds calculator for my age. See below.) But wait. Should I be more worried that my absolute risk for Prostate cancer was 24%? We’ll come back to that one later. Until then - the question I needed to answer first - what is Crohn’s disease anyway? I’d heard of it, but I needed to learn more. When I clicked on the Crohn’s Disease link I got the following screen.

This screen contained a nice summary explaining Crohn’s disease. I also followed links to even more detailed information on the disease. After feeling that I had acquired a lot of knowledge around Crohn’s, I scrolled further down the same page and found the Odds Calculator.

When I selected my ethnicity and age range, the odds changed. The absolute risk numbers went down significantly from 1.5 out of 100 to only 0.18 out of 100. The relative risk also dropped from 0.49 out of 100 to only 0.06 out of 100. As I played with the calculator, I learned that the odds increased as I got older.

What? It’s not just Genetics?

Next I learned that along with genetics, environmental factors play a big role in the development of disease. In other words, genetics is not the only cause of disease. I know that most people do not know about genetic epidemiology, which is the study of the role of genetic factors in determining health and disease in families and in populations, and the interplay of such genetic factors with environmental factors. So this bit of information is very important.

If only half of the cases of Crohn’s can be attributed to genetics, then half get the disease for reasons other than genetics! I learned that this is true for the other diseases as well – just with different percentages.

What does my Family Health History Tell Me?

I have learned from the time I have been working with the Clinical Genetics Institute that my family health history can also help me understand my genetic risk. The only family history of Crohn’s disease I am aware of is a cousin who developed Crohn’s after getting liver disease. I should look into this further and talk with more realtives.

As I add all of this information together, I decided that worrying about Crohn’s disease was not necessary. I also realized that I didn’t need to run out right away and alert my doctor. But maybe - if sometime in the future I develop symptoms like diarrhea, cramping, or bleeding, and a diagnosis is slow in coming - I can mention to the doctor to look at the possibility of Crohn’s – based on this “new” discovery of my genetic risk. Since 23andMe told me that the clinically-relevant genetic knowledge on this disease and others is changing, I will come back to my report later to see what - if anything - has been updated.

More to Come

As you can see in the screenshot, the page with the summary on Crohn’s disease contains other tabs called ‘How it Works’, ‘Timeline’, ‘MDs Perspective’, etc. I’ll show examples of the content these contain in future posts with other diseases. Additionally, we will look at a graph labeled ‘Marker Effects’. This graph shows the approximate effects of my genotype at the 12 reported markers for Crohn’s.

The 23andMe website content is designed to be mostly educational. This is a key point for customers to understand. Sure I saw my genetic results, but the majority of the menu choices and screens provided me with information to help me understand the disease(s), and how genetics plays a role in the development of the disease. Because my clinical knowledge is extremely limited (I’m an IT guy), I found the educational materials to be very important, sometimes even more important than my actual genetic information. There is a lot more content from my report to share in future posts. We’ll continue this fantastic journey next time.

Perspective from a medical geneticist

Crohn's disease risk

In our previous posts, we’ve written our sections independently in order to try and present our views without any bias that could influence the content. Now that we are diving into Grant’s results, we are going to change. As you may recall in our first post, one of the things we discussed was how this information could be used to inform interactions with medical professionals. With that goal in mind, we thought it important for Janet and me to see Grant’s impression so that we could specifically address any medical questions that arose in his interpretation of the results. In this way we hope not only to reflect both consumer and medical perspectives on the results but also to explore the interaction between the consumer and the health care system, given that 23andMe instructs customers to discuss any changes in health behavior with their provider.

Informed Patient

As I read Grant’s post, my first thought was, wow, I wish all of my patients had this level of insight into risk!! One of the most challenging things in practice is trying to convey concepts of probability and risk. It’s not something most of our brains are very good at. If they were we wouldn’t have Las Vegas and lotteries.

Grant immediately focused on his elevated risk for certain conditions—a very natural reaction. Crohn’s disease jumped out from the list as the relative risk is 3-fold greater than average. What impressed me was how quickly Grant moved from that concept to the more relevant concept of absolute risk—that is, how likely am I to actually develop the disease? In my practice experience a lot of time is spent explaining these concepts with less than satisfactory results.

Just-in-Time Clinical Knowledge

One of the innovations in the 23andMe website is the ability for customers to readily access information that is specific to the question at hand. Grant wasn’t sure exactly what Crohn’s disease was but he was able to click on a link that took him to a brief descriptive paragraph about Crohn’s disease. Embedded in the paragraph are hot links that can provide definitions of concepts that may not be clear. Links are given for those who desire more information. This educational approach is sometimes called “just-in-time”. It is a very sophisticated way to present information to people when they are ready to learn. One of my areas of interest is how the electronic health record can be used to present specific and relevant information to clinicians at the point of care. We’ve been exploring these types of just-in-time strategies to see if that leads to improvement in care. I don’t know if part of 23andMe’s research strategy is to explore the effectiveness of these educational messages, but I hope they spend some time on this as we have a lot to learn.

Odds Calculator

The Odds Calculator is also really cool. In most counseling situations I can’t customize the risks I’m presenting to my patients because I don’t have ready access to all of the information that impacts risk and don’t have the ability to compute the interactions in such a way as to be relevant to the specific patient. The Odds Calculator does this and as Grant noted, his customized absolute risk is much lower than the general risk information presented on the first screen. Now one must assume that the results given by the Odds Calculator is based on good data and knowledge of the interaction between factors. It is after all a black-box—we can’t look under the hood to check the calculator’s math. If a patient came to me and asked about the accuracy of this risk estimate, I would want to know all of the data used, the assumptions made and the literature it was based on to be able to provide a good answer. In general discussion with genetic and non-genetics colleagues, all bemoan that fact that we for the most part aren’t able to provide specific risk information for our patients.

The SNP Train

So how did 23andMe conclude that Grant’s relative risk was 3.01? That, it turns out did not involve a black box. 23andMe reported information on Adjusted Odds Ratios for 12 Single Nucleotide Polymorphisms (SNPs) for which there are reports associating the SNP with Crohn’s disease. Multiplying these Adjusted Odds Ratios together gives a result of 3.04—not different from 3.01. The assumption here is that all of these SNPs independently contribute to Crohn’s risk. This assumption is unproven. For example, three of the SNPs reside in a single gene, NOD2. This raises the question of whether these SNPs travel completely independently, or if they may be linked together so the pattern of SNPs is not random. Let’s use the analogy of a railroad, with an individual car representing a SNP. Looking at the 3 trains in the middle of the track; if the coal car of the top train, the Sinclair tanker of the middle train and the caboose of the bottom train were SNPS, we could most likely treat them as independent since they would all head out to different destinations. However, if the 3 SNPs in NOD2 are in linkage, then they would travel together as the coal car, tanker and caboose would in the train that is on the track. Thus they aren’t independent and the risk conferred by an individual SNP cannot be said to contribute an independent risk.

Combinations of SNPs that travel together are called haplotypes. Haplotypes are increasingly favored for risk prediction because it is easier to address the independence of haplotypes. What I did like was the brief description of what is known about each SNP in relation to Crohn’s disease and the reference list associated with each. It would be great to have a clearinghouse of information like this available as it would be easy to gather necessary information and references quickly when a question arose. That said, I don’t think I’ll cough up (spit up?) the coin to have my genome run just to get access to the information. It is hard to know how useful this information will be to customers.

Understanding Environmental Factors

Finally Grant noted the contribution of environmental factors to the development of Crohn’s disease. If I were counseling Grant (and I guess I am) I would point out that his understanding of the role of genes and environment is not strictly accurate at least as stated. Grant indicates that “If only half of the cases of Crohn’s can be attributed to genetics, then half get the disease for reasons other than genetics!” Actually what is trying to be conveyed is that cases of Crohn’s disease are likely due to a combination of genetic and environmental factors and, on average, half of the factors are genetic and half are environmental. In other words, one can’t divide Crohn’s into two piles, one genetic and one not.

How Your Doctor Will Respond

This leads to the question of how can Grant use this information? Clearly at the present time he’s stuck with the genes he inherited from his parents. So if this information is going to be useful, Grant would have to modify his environment (meaning his relevant behaviors related to the environment) to reduce risk. In the information provided about Genes vs. Environment (and I would not have used vs.) 23andMe states, “…environmental factors are currently unknown.” This means that there is nothing to do to modify Grant’s risk based on present information. So does this information have value? Certainly not if Grant wants to avoid the disease given that he can’t modify his genetic predispositions and we don’t presently know what he should change in his environment. One could argue that if symptoms of diarrhea, weight loss, cramping, etc. were to develop mentioning the increased relative risk for Crohn’s disease could lead to more expeditious diagnosis and treatment. However, it could also lead away from another diagnosis causing delay in diagnosis in treatment. The truth is we don’t know whether this is valuable or not. This position was reinforced in the MD’s Perspective section provided by Dr. Melvin Heyman.

(The following is an excerpt from this section.)

23andMe: Historically, how have doctors used genetics and family history when predicting a patient's risk of developing Crohn's Disease?
Dr. Heyman: The closest doctors have come is to look at family history for initial clues when given a set of symptoms. If a patient presents with symptoms suggestive of Crohn's Disease, such as abdominal pain, weight loss, poor growth in a child, or fever, a positive family history might lead to further testing faster compared to a situation where the family history is negative.
23andMe: Recent genome-wide association studies have enabled scientists to identify genetic markers associated with increased risk of Crohn's Disease. How do you think this information can be put into clinical practice?
Dr. Heyman: Again, until further studies are done to show the clinical utility of a genetic test, a recommendation for clinical practice is premature. However, in the future, such testing might help a doctor decide how aggressively to evaluate a patient for Crohn's Disease if symptoms develop. More research and clinical studies need to be performed before genetic markers for Crohn's Disease can be used to inform clinical practice.
23andMe: If I had the riskier versions of the SNPs known to be associated with Crohn's Disease, what would you suggest that I tell my personal physician?
Dr. Heyman: Because of the lack of clinical studies, we do not currently use genetic information when assessing a patient’s risk of developing Crohn's Disease. As the genetic and clinical research advances, we hopefully will reach a point where this information can be useful for patients and doctors. However, in these early stages, the main message to provide to your personal physician is this: Be aware that these genetic markers may indicate future development of Crohn's Disease.

Dr. Heyman clearly indicates that clinicians do not use genetic information for assessing risk for Crohn’s at the present time although indicates that this may change in the future. I appreciate 23andMe’s willingness to publish an independent perspective that doesn’t oversell the tests’ usefulness. Dr. Heyman does note the importance of family history in assessing risk. Family history captures not only genetic predispositions running in families, but also shared environment that confers additional risk. Grant is aware of a cousin who has Crohn’s disease. A relative such as this (that is beyond a second degree relative) doesn’t significantly increase risk for any disease.

Final Word

So if Grant were to ask me what he should do with regards to his increased risk for developing Crohn’s disease I would say, check back periodically to see if we’ve learned anything about environmental factors. Until then check out your prostate!! (Glimpse of coming attractions)

There are lots of innovative ideas about information sharing that are being deployed in the 23andMe site. I’m going to be learning a lot as we continue to explore Grant’s genome.

Read the next post - Carrier Status – Do you have your tracking number?


  1. Is there anyway that I can see a report from 23andme instead of creating a Demo account as it wont show anything? I love to see how the real report looks like. Is there a way I can see atleast a screen shot of the report?

  2. The report contains information on 116 diseases, traits, carrier status, etc. Each one is clickable with several more pages of support content. This does not allow for a 'screenshot' of the report. In fact, I haven't seen even an option to save the report in a PDF. My guess is this is not offered for both privacy and IP concerns. However, please contact me via the Contributor link on the top right and we can discuss further.