The Role of Genetic Testing in Clinical Practice—Promises and Pitfalls: A Clinical Conversation with Helen Messier,BSc,PhD,MD,and Robert Rountree,MD

Helen Messier, BSc, PhD, MD, earned her BSc in genetics from the University of Alberta, in Edmonton, Alberta, Canada. She then continued at the University of Alberta and earned her PhD in molecular immunology, with a focus on the genetic regulation of T-cell receptors. Later on, she completed her postdoctoral studies at the La Jolla Institute of Allergy and Immunology in San Diego, California. Dr. Messier earned her MD at the University of Calgary Medical School, in Calgary, Alberta, completing her residency in family medicine in Calgary. Her postgraduate training included functional, antiaging, nutritional, and integrative medicine. She is an Institute for Functional Medicine–certified practitioner, with experience in rural medicine, primary and walk-in care, geriatric medicine, and hospice care. Dr. Messier recently joined Human Longevity Inc. (HLI) in San Diego, California as the medical director of genomics. At HLI, she will be involved in conducting full genome sequencing, full microbiome sequencing, and full metabolomic assessments, combined with extensive phenotype testing, to create individualized integrative assessments for clients.

Robert Rountree: Not only do you have a BSc in genetics and a PhD in molecular immunology with postdoctoral training, you are also an MD trained in family medicine—a unique combination indeed! What early life experiences led to the intellectual interests that resulted in those career choices?

Helen Messier: I have always been interested in biology; so, right from day 1 in school, biology was always my favorite science. So I received a BSc with a major in biology. As I was going through my program, I realized that I wanted to focus on what I felt was central to all biology, which was human genetics—human DNA. I felt that it was the control center of all of life, and so I decided to do my honors in genetics. It was a natural extension for me to keep researching. Once I started finding things out about it, I was just so excited. I wanted to keep learning more, and ended up receiving my PhD, and then I became a postdoctoral fellow. I then realized that I had access to this amazing information and wanted to start applying it to help people. Obviously, this led to medicine, and I chose family medicine as the field that would give me the broadest ability to apply this knowledge. I got to treat each patient as a whole person.

When I started working in medicine, I realized that we were not really good at integrating the latest science into medicine. That was a big disappointment for me. We were following clinical practice guidelines, wherein the same treatment is applied to everyone with a certain disease. We were not looking at the individual and taking into account individual uniqueness. That is what led me to look more broadly and resulted in my interest in functional medicine.

Functional medicine is consistent with my desire to look at the root cause of disease. I can see what is happening in a patient on a biochemical or genetic level, resulting in the complex of phenotypes that the patient is experiencing, such as disease. By looking at the root cause, and doing something about that, I could have an impact on a patient's final phenotype or final disease state. I realized, when I got into medicine, that the translational gap is typically 15–20 years. That is, it takes that long before information that is published in the scientific literature actually becomes established in clinical practice. I made it my goal to try to close that translational gap, to try to use the knowledge that we have today to help people today, rather than waiting another 20 years.

Dr. Rountree: You are also a faculty member for the Institute for Functional Medicine, teaching in the Immunology Advanced Practice Module. Please contrast the functional medicine approach to immunology to what you learned in graduate school?

Dr. Messier: In the academic setting, I was really looking at what was happening in a cell and in the DNA and how genetic expression was regulated. I looked at the genetic control of the T-cell receptor genes and learned about the molecular control of the immune system. This research was all done in a laboratory setting—in a Petri dish or sometimes an animal model. The applicability to humans seemed a long way off at times. So, in graduate school, I learned the science, but not how to apply it. In functional medicine, I have taken that science and started applying it. It is not perfect; sometimes we have to make our best guess as to whether this is really going to work from a clinical perspective. I always look at risk–benefit. There may be a benefit to applying knowledge, but we also need to know the risks. This means that every patient is looked at from a different approach and given a different treatment. Functional medicine involves assessing each individual rather than following specific protocols.

Dr. Rountree: Please tell us about your current clinical practice? What kinds of cases do you commonly see and how do you make use of your specialty training in addressing the problems you find in your patients?

Dr. Messier: I tend to see the patients who have gone through their diagnostic odysseys in the medical system already. I see the patients who have seen many practitioners. These patients have often had extensive workups and might have diagnoses but do not have plans to help them get better. I spend a lot of time researching every individual person. I will do full genome sequencing for people who desire that. I will look at metabolomics and microbiomics. I will try to figure out the root causes that result in what they are feeling, to see if I can tweak that. I see many people with autoimmune disorders, as well as chronic fatigue and other complex chronic diseases, often ones that don't have names. I also see some people who are really just interested in prevention. They may have family histories of disease and want to prevent their risks of getting the diseases and optimize current functioning.

Dr. Rountree: When the Human Genome Project first began sequencing and mapping the entire catalogue of human genes, there was a lot of hope and promise that this Project would be possible to identify a discrete subset of candidate genes that determine susceptibility for many common diseases, such as obesity or diabetes, Alzheimer's disease, or breast or colon cancer. Now that sequencing can be readily performed and compared to the disease phenotype, how has the understanding evolved regarding the discrete gene hypothesis?

Dr. Messier: You are exactly right. We hoped and assumed that humans would have one gene, one trait, or one gene, or one disease. We thought that it would possible to identify the genes responsible for each of those things. It was the complete opposite of that. Despite having more than 100,000 proteins, humans have only approximately 20,000 protein-coding genes. ¹ Thus, there are not even enough genes to code for all the proteins in the human body. This suggests that there is a very complex interaction, and one particular phenotype can be the result of many different genetic loci. So we realized that this is far more complex than we originally thought. When we realized that only a very small fraction of the human genome was actually coding for proteins, and that there were vast stretches of DNA that appeared without purpose, we named it “junk DNA.”

However, we now know that much of that DNA is regulatory. ² This is the DNA that helps control what genes get turned on and off in cells. Also, we are still figuring out what all of this extra DNA does. We are just at the tip of the iceberg in understanding how to interpret human DNA sequences. Now that we are capable of sequencing full human genomes quickly and relatively inexpensively, we are making—and will be making—great strides in this understanding.

Dr. Rountree: In what other ways has the overall understanding of the human genome—especially the part of the genome related to health and disease—changed from the time you received your undergraduate degree to the current state of the art?

Dr. Messier: In the past, genetics in medicine has always been about single-gene traits or large chromosomal aberrations. For example, we have all heard of trisomy 21 or Down syndrome. So, that was always how we approached it. We would create these pedigrees and look at how traits are passed down from an inheritance, or family history, point of view. This works really well for monogenetic disease conditions that are strongly inherited. However, those conditions account for only about 1% of all diseases. Other people have genetic predisposition for diseases, such as cardiac disease, that can run in families. However, a genetic predisposition does not mean that a person will definitely get that disease. We know that the environment plays a large role in how the genes are expressed, resulting in what a person will actually experience. Therefore, clinical genetics in the past was always very different from molecular genetics.

Given that sequencing was not available, we would use cytogenetic methodologies to look at chromosome structure. We could see additions and deletions of chromosomes and large segments of chromosomes. Now, with molecular genetics, we can sequence the DNA and identify specific variants and collections of variants. We can also look at the regulatory elements. This was common in the laboratory but is now at a point where it can be useful clinically.

Dr. Rountree: Please give us an overview of the major types of DNA-sequence variations that are turning out to have particularly strong clinical relevance? Please also outline the differences among mutations, single-nucleotide polymorphisms [SNPs], and copy number variations [CNVs], and give us some clinically relevant examples of these?

Dr. Messier: People have many different levels of variation in their genomes that make people different from each other. People have the same number of chromosomes and essentially the same number of genes, but obviously have differences that make people unique. There are a few differences as you mentioned—SNPs, INDELs [insertions and deletions], and CNVs. As we know, DNA is a string of four nucleotides in many different combinations. The four nucleotides are: adenine; cytosine; guanine; and thymine. A combination of three of these nucleotides codes for an amino acid, and then the amino acids are strung together in proteins.

A single-nucleotide variant or polymorphism—an SNP—involves a single change in the sequence of these nucleotides. This would be the case as long as if there are at least two versions of the sequence with each present in at least 1% of the general population. ³ Each human being has about 3–4 million SNP's in his or her genome. These SNPs arose from mutations or changes in the DNA that were likely just random. This does not mean that the DNA has been damaged; it has just been changed. If the same mutation occurs often enough in the population it goes from being a mutation to a SNP.

INDELs involve the addition or absence of short sections of DNA from 1 to 10,000 nucleotides. Humans have approximately one INDEL per 7.2 kb of DNA, resulting in more than 400,000 of these in the human genome. ⁴ CNVs are these insertions and deletions at a much bigger level—up to megabase pairs in size. ⁵ All human beings have these examples. I personally have close to 4 million different SNPs, a million different insertions and deletions, and approximately 10 CNVs involving large segments. Most of these are inherited, although some do arise de novo in each person, resulting in individual DNA “fingerprints.” These CNVs can affect many genes, resulting in the possibility of missing certain genes or having multiple copies of the same gene. ⁵ CNVs can play a role in health and disease; for example, a deletion of the glutathione S-transferase (GST) M1 (GSTM1) and T1 (GSTT1) genes results in a higher risk for heart disease. ⁶

There are some clinically important SNPs that can also affect genes. A change in one nucleotide can change the coded amino acid, possibly resulting in a change in the function of that protein. There is even the possibility that the protein will not be made completely, or it will be truncated and nonfunctioning, because, instead of an amino acid, the nucleotide change codes for a stop codon. ⁷ Again—using myself as an example—I have 85 stop mutations, where they actually truncated the genes. Of these, 30 were stop-loss mutations, where the gene would just keep being transcribed with no stop to it. These changes obviously affect the protein structure.

There is a lot that we can interpret from the human genome today, and there is far more that we hope to understand in the future.

How do we use this information? Only some of these changes can actually be interpreted. Examples are eye color, taste ability, or the alcohol flush reaction. However, we don't really know what most of them do. In research, we are trying to correlate these genome sequences with phenotypes, what these sequence variants actually mean. There is a lot that we can interpret from the human genome today, and there is far more that we hope to understand in the future. It is only in the last 10 years or so that we have known that CNVs actually exist and that they have clinical validity.

Dr. Rountree: A number of commercial laboratories are now offering panels to the public and to practitioners that include identification of SNPs and/or CNVs. Some practitioners are claiming to have a high level of expertise in interpreting these tests and are making detailed clinical recommendations based on genetic analysis. What do you see as the pros and cons of this type of testing? What is the level of accuracy?

Dr. Messier: Most of the pros and cons are in the interpretations. The data to determine the meaning of specific variants come from genomewide association studies in the medical literature. ⁸ These association—or case-control—studies suggest that a certain SNP is associated with a certain percent of increased or decreased risk of a disease, based on the number of people in the population with that SNP who have—or don't have—a certain disease. For example, if 60% of the population with a certain SNP has heart disease versus 40% of the population with another SNP or no SNP, then the first SNP is associated with an increased risk of heart disease.

Using myself as an example, one testing company reported that I have a 3.14 times increased risk of rheumatoid arthritis based on testing of variants in nine different genes. However, looking at each of these genes individually showed me that, based on three of these genes, my risk was decreased, and, based on five of the others, it was increased. The clinical validity of this type of analysis is questionable. Part of the problem is that the magnitude of risk for each particular SNP in available research is very, very small. This is likely because there are so many different areas of the genome that are involved in a particular disease. However, as we start doing full genome sequences on increasing numbers of people, I hope that we will have a much better understanding and can use this information in a valid way.

Dr. Rountree: Are there any areas of medicine—such as adjusting drug dosing based on pharmacogenomics or determining increased risk of cancer based on xenobiotic transformation—where SNP testing is proving to be particularly useful? As an example, please tell us how the complex metabolism of tamoxifen illustrates the pros and cons of this testing?

Dr. Messier: I think pharmacogenomics is one area that is “ready for prime time.” We know how certain pharmacologic agents and drugs are metabolized. This means that we can look at the variation in the gene coding for the enzyme that metabolizes that drug, and we can predict how efficiently someone would be able to metabolize it. ⁵ We can predict that a person would need a lower or higher dose of a particular drug, or that the drug is not going to work in that person at all. Understanding this may actually result in an avoidance of some well-characterized reactions to specific drugs, and therefore avoiding morbidity and mortality. For example, patients that metabolize codeine very quickly can overdose on it very easily; in other patients, it won't work at all. ⁹ If we can predict those outcomes, we might be able to prevent serious adverse drug reactions or giving someone a drug that won't work to relieve their pain.

Unfortunately not everything is straightforward. For example, tamoxifen has a very complex metabolism. ^10,11 Cytochrome P450 (CYP) 2D6 is an enzyme that metabolizes tamoxifen to endoxifen. In 2006, the U.S. Food and Drug Administration suggested that everyone should be tested for CYP2D6 because of the possibility that, without appropriate metabolism, tamoxifen would be ineffective. ¹² Like all CYP enzymes, the CYP2D6 gene is highly polymorphic, with multiple alleles that code for different variations of the enzyme with different abilities to metabolize their substrates. ¹⁰

So, already it is difficult to actually do the test to see exactly which CYP2D6 allele is present. The direct-to-consumer testing that currently exists does not allow one to determine which CYP2D6 allele is present and, therefore, whether someone is a fast or slow metabolizer. In addition, there is an issue with tamoxifen metabolism itself; the drug does get converted to endoxifen but also gets converted to many other metabolites, such as 4-hydroxytamoxifen (N-desmethyl tamoxifen), and these metabolites all have inhibitory or active effects. ¹¹ Therefore, people have different reactions, with the drug converting to many metabolites, with different cytochrome enzymes working on them, and each cytochrome enzyme is polymorphic.

Testing for CYP2D6 alone will not let a clinician know if tamoxifen is going to work for that person. Pharmacogenomics testing may work well for many drugs; however, there are a number of drugs for which it is too soon to be able to make valid conclusions. This holds enormous promise, which is why I do it and why I am so excited about it. However, if we do not understand where we are right now—and we do not understand some of the pitfalls of our interpretation and our testing—we do a big disservice by using this information.

Dr. Rountree: I recently listened to an episode of Radiolab (on National Public Radio) that discussed a case of a pregnant woman who discovered that she had twins, only to receive the sad news that one of the twins had anencephaly and would not live long past birth. After the infant died, extensive testing was performed to compare the genes and epigenetics for both twins. As expected, the genes were identical, but there were more than 1000 changes in the epigenetic markers for the anencephalic twin versus the normal one. How is our understanding of epigenetic factors evolving and how do we factor that in when we are trying to understand the influence of genetic sequence variations?

Dr. Messier: Up until now, we have spoken about genetics—the actual sequencing of the DNA. Epigenetics is anything that affects the pathway from gene sequence to RNA coding to protein production and finally protein activity. ^13–14 Epigenetics is anything that controls how much RNA is actually transcribed, how that RNA is processed, how that RNA is read into a protein, how that protein is translated, and how that protein becomes modified to actually work. We have always thought of epigenetics as just being how a gene is read or what controls whether a gene becomes transcribed or not. So for example, if a gene gets methylated by the addition of methyl groups to the CG nucleotides, the gene does not get transcribed.

However, we now know that there are so many other levels of control that are also epigenetic. For example, there is sequence information, including the promoters and enhancers, the regulatory elements, the structure of the DNA, the open-versus-closed state of wrapping around histones, or the modification of histones themselves. We know that RNA has variable splicing, which means that the same RNA sequence can result in a number of different mRNAs or proteins. ² We also have many noncoding RNAs that can silence and control expression of both DNA and of RNA. Finally, when proteins do get made, they are often modified post-translationally. Given that the genetic sequence and the epigenetic control are intertwined, they affect each other and cannot be separated. We know that the environment plays a role in epigenetics in many different ways.

A classic example is the agouti mouse experiments. This strain of mice contains the agouti gene, which, if expressed, will make a mouse fat and dark brown, and it will have a high risk of heart disease and diabetes. If the agouti gene is not expressed, the mouse will be thin, white, and healthy. If the mothers of these genetically identical mice are fed methylation precursors—including green, leafy vegetables containing folate and vitamin B₁₂—the gene is not expressed and the offspring are very healthy. ¹⁵ When their mother's diet was devoid of those methylation factors, the gene was turned on, and the offspring were prone to diabetes, heart disease, and obesity. This was one of the classic experiments that pointed to the role diet has on epigenetics and the expression of DNA.

Interestingly, in the same model, the environmental toxin bisphenol A was given to the mothers and it turned on the expression of the deleterious agouti gene in the offspring. ¹⁶ Folic acid reversed this affect. So, we do know that toxins can change epigenetic regulation and genetic expression, and that diet can play a role in prevention. Yet, testing for toxin exposures in women prior to pregnancy is easier said than done. Therefore, giving good nutrition to pregnant women only makes sense. The fields of nutrigenomics and nutrigenetics are currently emerging. ¹⁷ Nutrigenetics is the field concentrating on how gene-based differences affect response to foods and nutrients, and nutrigenomics explores how nutrients affect genes and gene expression.

Dr. Rountree: Please talk about the role of genetic sequencing of tumors and the use of that information for determining a protocol for cancer treatment?

Dr. Messier: We know that cancer cells have multiple mutations, including SNPs, CNVs, and huge deletions of DNA and even chromosomes. However, we also know that cancer is polyclonal, meaning that different areas of a tumor will have different genetic mutations. ¹⁸ We also need to compare the cancer mutations to each patient's germline DNA as a control. We have discussed the fact that people have CNVs in their normal germline DNA. In the past we would have the sequence for the tumor DNA, but not the germline DNA to act as a control. Then we would attribute a CNV in the cancer cell to the cancer in some way, when, in fact, that same mutation or CNV may very well be a normal thing in that particular individual.

To Contact Dr. Helen Messier

Helen Messier, BSc, PhD, MD

Human Longevity, Inc.

4570 Executive Drive

San Diego, CA 92121

E-mail: drhmessier@gmail.com

Understanding the unique mutations in each person's tumor that might be driving the growth of that tumor has proven to be very useful for designing targeted treatments for that individual. Not all breast or colon cancers are the same. Thus, we should treat them uniquely based on their genetics. However, because of the quick and ready ability of cancer cells to mutate, rendering resistance to these treatments, treatments often only work for a limited period of time. In my opinion, genetics is going to play a really important role in cancer treatment in sequencing the immune system, for example, sequencing T-cells. We can now tell exactly what every single T-cell in the body recognizes, and I hope we can use that information to immunize the body against cancer. ¹⁹

Dr. Rountree: This is a very exciting development. As we wrap up, please tell us what the next frontier will be for you?

Dr. Messier: I am going to be moving down to San Diego to work at Human Longevity, Inc. [HLI], which is truly on the forefront of trying to understand human genetic information. HLI uses full genome sequencing, full microbiome testing, and full metabolomics testing, together with very intricate phenotype data to truly understand what is going on in a person. This helps identify those very first perturbations of disease. This will allow us to possibly not only extend life but also to optimize wellness.

The biggest risk factor for all human disease is aging. So what happens in that process? What is going on in the DNA? Although I will be working with laboratory data and brilliant scientists, I will be applying this information directly to clients. Some of these people may be ill but, often, they are looking for prevention. People can come and get this amazing assessment done.

As we start collecting the data, we are going to get increasingly better at knowing what it means. These assessments will be performed annually, and then, the information will be sent back to primary care providers so that they can use that for care of their patients.   ■