If one were to ask Kjersti Aagaard, MD, PhD, to determine whether a woman is pregnant, she could do so with a test that’s predictive 98.2 percent of the time.
Not that this test will be headed to a clinical laboratory anytime soon. “We call it the $50,000 and six-month pregnancy test,” jokes Dr. Aagaard, assistant professor, obstetrics and gynecology, Baylor College of Medicine, Houston. Her test assesses changes in the vaginal microbiome, sequencing it using a series of analytical pipelines through supervised, or machine, learning to predict a woman’s pregnancy status.
So, yes, a urine dipstick test—or even a glance at the woman in question—still has its merits. Nonetheless, Dr. Aagaard’s work is part of the next frontier in medicine: the human microbiome. Her laboratory at Baylor does both clinical work and basic science; she and her colleagues focus their translational research on trying to understand how perturbations in the intrauterine environment affect future health and disease. But while her focus is on the vaginal microbiome, the implications could be immense.
Dr. Aagaard calls the microbiome “our second genome.” The Human Genome Project, of course, is a household phrase. In its wake now comes the Human Microbiome Project, launched by the NIH in 2008 to identify and characterize microorganisms in healthy individuals as well as those with diseases, using array-based technology to do genome sequencing of bacteria.
“Our microbes outnumber us,” Dr. Aagaard says. Assuming the human body contains approximately 100 trillion cells, she says, 90 trillion of those are microbial. “And that means their genomes outnumber us 100 to one.” The microbiome project resembles the HGP—but is a thousandfold bigger, she says. It’s the scientific equivalent of a stroll down the Las Vegas Strip, a seemingly endless unfolding of new developments. “The HMP is a massive amount of big-team science that will revolutionize how we look at our genomes, our metabolism, and our immune response.” It may even change answers to seemingly basic questions: What is health? What is disease? The line between the two is fluid, a notion that gains more traction as researchers learn more about the human microbiome.
In her own work, Dr. Aagaard has focused on developmental origins of disease, trying, for example, to understand why a maternal high-fat diet can lead to childhood obesity in offspring and bring future risk of adult disease. Subsequently, she says, she and her colleagues became interested in nontraditional genomic inheritance mechanisms, including epigenomic changes, or changes on top of the nucleotide sequence. They’ve looked at mitochondrial DNA mutations. “And then we became interested in metagenomics—the microbiome,” she says.
The premise behind metagenomics research sounds straightforward: trying to compass the totality of organisms in the human body, including microorganisms present in skin, hair, mouth, digestive system, urogenital tract, and respiratory system. Each contains a distinct microbiome, says Dr. Aagaard, with each the result of numerous mechanisms. The microbiome in the gut, for example, is influenced by diet, medications, gender, and even climate exposure. “There are also microbiomes by region in the world,” she says. There is evidence (Hehemann J, et al. Nature. 2010;464:908–912), for example, that the Japanese microbiome contains a gene for a seaweed-digesting enzyme from a marine bacterium called Zobellia galactanivorans; the gene is not expressed in the guts of North Americans.
The microbiome profile is partially acquired vertically; in other words, birth is the first process in establishing the majority of an individual’s microbiome, says Dr. Aagaard. Other investigators have shown that infants’ microbiomes differ depending on whether they arrived vaginally or via cesarean section. “We’re not saying one is better or worse,” Dr. Aagaard cautions. “Just that they’re different.”
This knowledge led her to explore whether the vaginal microbiome itself changes during pregnancy, which led to subsequent questions. Does the vaginal microbiome change in pregnancy compared with nonpregnancy? Does it change at different points in pregnancy? Can it predispose women to certain complications in pregnancy, such as preterm birth?
Tackling that first question, Dr. Aagaard asks, “Do you go along all your life with a kind of predictable vaginal microbiome, at least up to the point of reproduction, where you start to see changes? And then in pregnancy is it remarkably different?”
As it turns out, the microbiome does change in pregnancy—especially in the posterior fornix—with species becoming less diverse and less abundant. Dr. Aagaard finds this fascinating. “Here’s an example where you don’t have a disease—you have a state of health that is absolutely essential to a species propagating, and the vaginal microbiome changes just by being pregnant.”
Her next step has been comparing vaginal microbiome in preterm birth with term birth, as well as possible implications for the offspring. She’s looking at vaginal microbiome in mothers as well as assessing the samples in placenta and the mouth and gut of newborns.
To do their microbiome profiling, researchers take samples of genetic material directly from specimens or whole tissue, and DNA is extracted. From there, researchers can do one of two things.
One is 16S-based enrichment. “You amplify up in the V regions of the bacterial 16S genome—so you enrich for your bacteria,” Dr. Aagaard says. She and her colleagues use the 454 Life Sciences platform for sequencing, though she notes there are other platforms that can also do this work. “That allows you to ascribe different taxa. And that gives you a certain level of profiling.”
The other approach applies shotgun sequencing to metagenomic samples, which “gives you both the host DNA and the nonhost DNA. Then you have to filter out those DNA sequences.”
“The advantage of the whole-genome shotgun is it can get you down to the species level,” she says. “16S-based approaches definitely get you to the genus level, and often get you to the species level, but not consistently.”
“The downside to the whole-genome shotgun is it’s incredibly computationally laden,” she continues. “And it becomes cost-prohibitive to do big population profiling.”
Dr. Aagaard isn’t convinced that will change anytime soon. “You’re talking about terabytes of data.” For now, the sensible approach is to use 16S-based metagenomics to obtain snapshots, then figure out where a deeper look will be useful.
The clinical implications of her work are still unknown. Would it make sense, for example, to monitor the changes in the vaginal microbiome during pregnancy? “We’ve done some of that work initially. It’s a great question, and we hope NIH will want us to answer it,” she says with a laugh.
She and her colleagues have also noted a clear association between microbiome distinctions and obesity. “What’s the chicken and what’s the egg, no one knows,” she says. That has led to work using a nonhuman primate model to explore the link between microbiome, obesity, and high-fat dietary exposure.
For now, the work is largely observational in nature. “We could go on pathogen hunts, but we have to remember that the vast majority of pregnancies are totally normal,” she says. Her pathogen hunt, so to speak, is looking at preterm birth.
“But that varies from one individual to the next. And it probably varies based on where they live in the world, and what type of diets they have, and how many children they’ve had. So for us to make any meaningful interpretations on these pathogen hunts, we first have to know what the whole population looks like, so we can know whether something’s a true pathologic variant or just a normal population variant.” It will take a little bit of time to sort that out, she says—and a lot of specimens.
How will researchers assess the massive amounts of information that ensue? “We can map back these pathways,” she says. Whole-genome shotgun sequencing in particular lets researchers create metabolic reconstructions; they can use these pathways to discover what microbes are likely doing (or not doing).
Bigger picture, she talks about the work emerging from the laboratory of Martin Blaser, MD, at New York University School of Medicine, which suggests that human beings’ microbiomes have shifted in recent years. “Now we eat processed foods, we don’t drink as much wine, the fermentation processes that used to be very important to establishing our microbiomes have changed,” she says, which leads her to another unknown: Are important microbiotic communities disappearing as a result?
Helicobacter pylori levels in the human gut are decreasing, at least in the developed world—and so has the incidence of stomach cancer. But given that esophageal cancer has been increasing simultaneously, it’s possible that H. pylori offers a protective upside that’s not yet fully understood. “There may be some examples of coevolution between H. pylori and humans,” Dr. Aagaard suggests. “So the microbiome opens up the possibility that we can start to understand how we best work together as a community, both as an individual community and as a bigger, human species community.”
It is, she says, a matter of balance: What is commensalism? What is mutualism? And what is pathogenesis? Despite the –isms, this is not a philosophical debate. How the human genome harmonizes with the human metagenome is critical to health, she says, and likely affects disease burden.
None of this will transform diagnostics immediately (see above: $50,000, six-month pregnancy test), but Dr. Aagaard sees a day when pathologists will need to be fluent in the molecular diagnostic techniques for epigenomics and metagenomics. “It’s in the stage of observational science, but it will move quickly to intervention.”
Karen Titus is CAP TODAY contributing editor and co-managing editor.