How to tell if your genes could be acted on to help
I typically write my posts with a great focus towards the reader. If I wanted to write strictly for myself, these posts probably would be a lot more boring, with just scientific facts facts facts facts facts (enjoy the mild tongue twister of not so mild twisted outcome). But what I write about has to strike a chord with my inner excitement, and what I want to write about today is pretty exciting to me - very exciting! - so let's see if I can translate this from science to real life.
You see, just like you have favourite artists or sports stars or politicians you like to follow (yeah right to that last one), I have some favourite scientists whose work I like to follow. Some people are just embedded in consistent quality work, especially if it is at the forefront of developing novel strategies. The field of genomics, especially in its current rate of high throughput data production, is an area of science that offers such opportunities, as so many standards, guidelines and measuring metrics have to be created from scratch during this outburst of new technology.
It is akin to when "automobiles" first came out on the scene in the early XX century. They were new, they were cool, they were mysterious, and crazy fast, easily outperforming a horse at a whopping 50 miles/hr speeds! There were no standards at first: no seat belts, no road signs, and not even knowing who yields to who: a horse to a car, or a car to a horse? The only assumed standard that almost everyone agreed to was in regards to not running over people (and even then some probably argued over that exact definition).
So imagine the sudden possibility of being able to decode your most precious biological code, your genome, in its entirety (well nearly, a tiny fraction still eludes modern technological capabilities, but complete genome sounds better than almost complete genome). The amount of knowledge related to the association between your DNA and your traits is growing at a cosmic pace! I use that reference on purpose, as data generated from sequencing genomes is now on par with the data amount produced by YouTube, Twitter, or astronomy!
So that's a lot of data, in a time span shorter than China managed to economically outpace all of the countries in the world to genuinely resemble the 21st century in some of its infrastructure (and that includes genomics landscape too!).
But the standards, which require the brain power of many smart folks, has been lagging behind, despite constant consortia being developed to take part in building the framework and structure in this wild world of data. One of the criticisms frequently ushered towards the application of genome sequencing in context of the general public is the question of its utility. Not enough evidence has been generated to be able to determine how the data genomic testing generates would be of use, especially in the context of determining when individuals should be warned about the consequences of sequencing results.
My role as a consultant means that I aim to pay especially close attention to information that concerns the benefits or protection of the end users of this technology (that’s you!). So what was I so excited about? Earlier this year, one of my favourite scientists published a great paper that offered suggestions on how to start quantifying when a gene with a pathogenic mutation should be considered "actionable"!
Those are the genes that if mutated can lead to serious health complications, but where something can actually be done about it for a patient if discovered. So being able to determine what constitutes a gene with beneficial treatment or intervention to take place is definitely important! This is particularly timely work in light of emerging evidence of genetic testing leading to bad decisions with no benefit to patients.
One such instance that recently made headlines involved a boy with a defibrillator transplanted into his heart after an improperly validated variant was discovered which was suspected of causing the untimely death of his brother. Perhaps a crisis like this could have been averted if the assessment protocol - like the one I want to describe here - was used! And just imagine the trauma for the family involved! Unfortunately, no technology, and no human oversight, is infallible, and such sad mistakes will occur, but hopefully only with decreasing frequency. That is, after all, the hope of all medicine.
So this particular publication discussed the development of a metric for evaluating the clinical action of incidental findings from sequencing one's genome. Before we go on, what are incidental findings? These are secondary discoveries when DNA tests are performed for a specific reason. For example, you and your doctor decide to test genetically based on your DNA family tree and you find out you have a predisposition to breast cancer. That would be a finding that is incidental to the original purpose of the test.
So let’s look at this new metric in a basic description: 5 specific characteristics of the disease or health problem associated with a particular gene mutation (the incidental finding) are looked at. They include the likelihood of disease development, the severity of a disease, the efficacy of a proposed intervention, the burden of such intervention to a patient, and, finally, the knowledge base supporting this information. In hindsight, everything is so logical. :) Each of the categories could be scored from 0 to 3, for a total of 15. The higher the score, the higher the need for considering clinical action. For example, disease severity is scored from minimal health impact (0) to sudden or inevitable death (3). The efficacy of the intervention is scored from ineffective (0) to highly effective intervention (3). The knowledge base is scored from poor evidence (0) to substantial evidence (3), and so on. What I found interesting is that the moment the likelihood of disease development was more than 50%, that was enough to receive the highest score of 3. I would have imagined that higher percentages would be needed for this rating, so it is very suggestive of the wide variation of disease development that can be expected in a population.
What is cool is that the authors tested their scoring system against the highly publicized 56 genes that have previously been suggested by the American College of Medical Genetics and Genomics as those genes associated with disease that should be reported to an individual if found. When first published by ACMG, this report stirred up a great deal of controversy, in terms of how the decisions took place, and what criteria were used to determine this list, not to mention the potential infringement of patients’ rights. Whatever your opinion might be, this list has become the de facto golden standard among the industry on what gene results are reported to an individual when having his or her genome sequenced. Oh, and as of November 2016, this list has been updated to 59!
The authors also tested their new metric against their own expanded list of 161 actionable genes. Maybe you can start to understand why I'm a fan of their work! They push the envelope towards enhanced practicality. And finally, they randomly picked 1000 genes out of the magic bag of known genes (otherwise known as RefSeq database), and decided to assess those too. So their previous work of selecting clinically actionable genes was about to be put to the test! Out of the 1000 random genes, 889 turned out not to be tied to human diseases, so out the window they went. "Gardez genetique l'eau!" - what inhabitants of Edinburgh would shout (for some reason in French) as they poured their waste out the window to warn those below of the incoming refuse. If you were too drunk and looked up in an inopportune moment, that is how one of the most colloquial terms for being drunk was supposedly born. In a similar fashion, if you sequence your genome with poor quality oversight, consider yourself "gene-faced"!
So how did they fare? Of the authors' own actionable genes list, the median score was 12, with a range between 0–15. This means that they previously picked some real duds! 84 of these 161 gene–disease pairs scored ≥12 though, while 29/161 pairs scored <10. So overall that is actually pretty good!
The median score for the ACMG genes was 11, with a range between 7–14. 25/57 gene–disease pairs scored ≥12, and 11/57 pairs scored <10. The one outlier gene that scored 7, the NTRK1 gene, although originally included on ACMG’s recommended list, has since been dropped, so this also checked out. Those previously published genes had been selected pretty well as it turns out, although you can see how such lists need to be carefully monitored and adjusted as appropriate when new evidence surfaces. Also, why it is taking so long to populate this list with new entries!
How about the random genes? The median score of the remaining 111 random genes was 7 (range 1–13). Still 14/111 gene–disease pairs scored ≥12! That's pretty crazy, and pretty significant, because it highlights how much still needs to be assessed in the world of genomics. And if that's the percentage of randomly discovered actionable genes, then this would suggest that as many as 500 actionable genes could be included in the final list, considering that ≥3,000 single-gene disorders are currently suspected. And that would be pretty big list, big enough to clearly indicate clinical utility of genome sequencing.
The authors’ sage conclusion: "gene–disease pairs having a score of 11 or higher represent the top quintile of actionability." So there you have it folks, now at least we have a starting point of measuring how genes involved in disease could make their way onto the pathogenic but actionable list. After all, that is one of the big things that genome sequencing is all about: to help determine what is present (hopefully nothing pathogenic, but that won't always be the case, we all know that), and what can be done about it if found.
Pretty exciting! You know what else would be pretty exciting? These scoring criteria being modified to determine what clearly constitutes a non-actionable pathogenic gene involved in a disease development. Just because a gene does not meet the threshold of 11, does not necessarily make it non-actionable. It just means that perhaps there is fuzzy evidence towards calling it actionable. Just because something doesn't work, it doesn't mean it is broken. Just because your husband developed a beer gut instead of that promised 6-pack (and at this point you would even settle for a 4-pack, maybe even just a 2-pack), doesn't mean he is not a great person! Right? Right?
What would be really exciting is if we had a reverse criteria! Instead of being able to score disease genes, how about scoring disease protective genes, or what I term superhero genes: those that enhance your powers far beyond an ordinary individual in this day and age. That would be pretty sweet to discover in your genome; instead of just wondering if pathogenic information is found, you could wonder if a survival bonus will be found in your genetic cards! ;)
And the final bonus of this story: combining all of the genes assessed in this study that scored 11 or higher revealed a list of 168 genes that could be considered actionable! The data is growing! And so are the incentives to do a DNA check!
Who was my scientific star I mentioned? That was Dr. Jonathan Berg. I actually met him not that long ago at an American Society of Human Genetics conference in Vancouver, and probably made him feel awkward. After all, scientists don't often have groupies. He was presenting data on the studied factors that helped to determine how best to increase the utility of genomic technologies among minorities, which are often disproportionately underrepresented in such studies. This is very important work because ethnic minorities will present their own genetic signatures that need to be understood in the context of medical treatment. I seriously wanted to ask for a photo together but got too embarrassed to ask. Meeting him in person felt cool enough, considering how much I admire the work he oversees (it's actually a group effort of tons of individuals). I was a fan of his work ever since I started studying genomics. So if you don't have your favourite scientist yet, I could suggest a few, and he would be one of them. Probably too soon to produce sports-style cards of our most accomplished scientists, but it's too bad not more attention is paid to individuals who have so much influence in the lives of so many through their dedication to science.
And speaking of dedication, if you enjoy what you read, don’t forget to spread the word. After all, we are a long ways off from getting those world scientist trading cards unless we can talk about science. You would be surprised what a great conversation starter it can be (ok, maybe don’t bring this stuff up right away if you are on a date). Sharing this work will help to accomplish this goal too. And if you have burning questions on where, why and which genome to sequence for yourself, we are always here to help too. Have a great genome!
This article has been produced by Merogenomics Inc. and edited by Kerri Bryant. Reproduction and reuse of any portion of this content requires Merogenomics Inc. permission and source acknowledgment. It is your responsibility to obtain additional permissions from the third party owners that might be cited by Merogenomics Inc. Merogenomics Inc. disclaims any responsibility for any use you make of content owned by third parties without their permission.