Special Report on Molecular Diagnostics
Cancer characterization leads to fetal monitoring
The woman lies on the table, her partner at her side, holding her hand as the physician slides the ultrasound probe across her skin, eyes glued to the monitor.
It’s obvious that the growth has changed since the last test. It’s gotten bigger. The tissues have changed.
And where this might be bad news for someone with a diagnosis of cancer, the news is brighter for the expectant mother and her burgeoning family.
Although the comparison of a developing fetus to a cancerous tumor may seem callous, there are a lot of biological and technological parallels between the two.
According to Kim Martin, vice president of global women’s health at Natera, one of the first people to openly describe the parallels between the placenta and a tumor was Dennis Lo of the Chinese University of Hong Kong (CUHK).
“I thought a baby living in a mother is a little bit like cancer growing in a patient,” he said in a 2013 interview with The Guardian. “So, if cell-free tumor DNA can exist in a patient’s blood, surely there can be some cell-free fetal DNA in a pregnant mother’s blood.”
It was this thinking that led Lo to develop what is now described as noninvasive prenatal testing (NIPT) or screening (NIPS).
Within days, he recounted, he had developed a technique for treating a pregnant woman’s plasma and used polymerase chain reaction testing, or PCR, to look for evidence of the Y chromosome. The experiment, later published in The Lancet, was a success.
And yet, it took a while, he emphasized, for the community to see beyond the initial findings.
“They didn’t realise how far this technology could go,” Lo said. “They thought you could use it only to tell the sex of the baby.”
Part of the impetus for NIPT was the problem that traditional fetal diagnostic methods—amniocentesis and chorionic villus sampling (CVS)—came with risks, including a 1 percent or so spontaneous miscarriage rate. A safer way to perform at least a preliminary screen was needed (see sidebar article, “Detecting anomalies”).
Working with Sequenom (now part of Integrated Genetics), Lo launched the first commercial NIPT for Down’s syndrome (trisomy 21) in 2011, and more recently, tests for Edwards (trisomy 18) and Patau (trisomy 13) syndromes. And until recently, broad chromosomal aneuploidies were the sole domain of NIPT.
But as with cell-free DNA (cfDNA) itself, technological innovations in areas such as oncology continue to push those boundaries.
In a recent editorial, Lyn Chitty of Great Ormond Street Hospital for Children suggested that most of the literature on NIPT has focused on chromosomal anomalies, screening for aneuploidy, which has resulted “in a significant reduction in invasive testing and consideration for first-line Down syndrome screening.”
“Far less attention has been given to noninvasive prenatal diagnosis (NIPD) for monogenic disorders,” she continued, “as there is virtually no commercial interest and development is costly and largely done on a bespoke family-by-family basis.”
This imbalance is shifting, however, according to Martin, as cfDNA screening increasingly probes ever-shrinking targets from microdeletions to single-gene disorders where the change might be a single nucleotide. Her company, for one, is pushing that evolution.
“Natera really launched microdeletion pretty early on, shortly after they launched their five-chromosome test: 13, 18, 21, X and Y,” she explains, adding that the most important microdeletion is probably 22q11.2.
It is the most common, with an incidence somewhere between one in 1,000 prenatally to one in 4,000 to 6,000 post-natally, across maternal age.
And knowing the mutation is there opens to door to early intervention into the disorders triggered by the genetic change.
“Early intervention for the cardiac defects, many of which may not be detectable by ultrasound and may have significant morbidity if unrecognized,” she elaborates, “as well as early intervention for developmental disability or intellectual differences.”
As an example, she offers hypocalcemia as a common pathology of 22q.
“There’s clear evidence that early recognition and treatment of low calcium levels made a difference in the intellectual functioning as adults, suggesting effects of hypocalcemia on early brain development,” Martin continues. “We don’t routinely measure calcium in all infants, but a kid who’s at risk for 22q sure needs their calcium evaluated and replaced if they’re hypocalcemic.”
As the targets get smaller, however, the need for a greater number of target molecules becomes more important. This is where issues of fetal fraction—the amount of fetal cfDNA vs. the maternal background—becomes a bigger issue.
“As the fetal fraction is lower, it is kind of a question of how many needles did you put in the haystack,” Martin explains. “The more needles there are, the easier it is to find one. And that is especially true when the needles get really small. A microdeletion is like a very small needle.”
For this reason, the company cannot simply quantify all of the fragments from chromosome 22 and calculate a z-score. Instead, they amplify and sequence more than 2,000 SNPs from the critical region of 22q, giving them a higher sensitivity for the small missing pieces.
“The number of times a specific fragment is amplified, our depth of read, is higher per region than shotgun approaches,” she continues. “We take 14,000 SNPs, but we only divide them between five chromosomes. We take 2,000 SNPs, but we only have them for the critical region of 22q.”
Last year, University of Connecticut Health Center’s Peter Benn and colleagues at Natera, including Martin, examined a revised microdeletion detection protocol that included not only 22q11.2, but also four other microdeletions associated with disorders like cri-du-chat and Angelman syndromes. The new method allowed the researchers to reduce the false-positive rate (FPR) from 0.33 percent to 0.07 percent and increase positive predictive value (PPV) to 44.2 and 31.7 percent for 22q11.2 and the other microdeletions, respectively.
“Prior publications have questioned the use of NIPT as a screening test for microdeletions, citing concerns about high FPRs, low sensitivities, and challenges associated with variants of unknown significance,” the authors wrote. “However, these reports focused on whole-genome sequencing [WGS] approaches that employ counting-based methodologies.”
“By concentrating on specific genomic regions with clinically significant deletions, the targeted nature of the SNP-based method overcomes many of the limitations discussed in these publications,” they continued. “Furthermore, comparison of detection rates for confirmed microdeletions in clinical cohorts and ratios of maternally inherited vs. de-novo deletions to published ratios suggests that SNP-based methods have substantially higher sensitivity than counting-based methods.”
They suggested that their findings supported the idea of offering microdeletion screening as “an adjunct to existing NIPT for the five clinically significant, well-characterized genetic disorders,” something they argued was consistent with American College of Medical Genetics and Genomics (ACMG) guidelines.
Natera is not alone in pursuit of 22q11.2 microdeletions in cfDNA. Last year, Maximillian Schmid and colleagues at Ariosa Diagnostics (part of Roche) and Université Libre de Bruxelles published their efforts to detect microdeletions.
Using a number of microarray platforms focused on a ~3Mb region of 22q11.2, the researchers screened more than 1700 maternal plasma samples. In 122 samples with deletions ranging from 1.96-3.25 Mb, analytical sensitivity was 75.4 percent, while specificity in the remaining presumed unaffected samples was at least 99.5 percent. Clinical validation with samples of known microdeletion status offered similar results.
The next challenge for researchers was then to go from finding short missing fragments to identifying mutations at the single-gene level, which for Natera’s Vistara platform, means 30 specific genes.
As Martin explains, to be selected, those 30 genes needed to meet certain criteria.
“We picked things that were relatively common,” she says, suggesting one in 10,000 as a standard frequency.
“We picked genes that were amenable to sequencing with a high sensitivity,” she adds. “And also disorders that tend not to present with ultrasound findings. These are things where if the ultrasound findings are there at all, they are often not there until after 24 weeks and they are often very nonspecific.”
Another criterion, she explains, was that the diseases caused by the mutations were serious.
“I don’t like the term severe,” Martin offers. “I think severe means different things to different people.”
Rather, she sees serious conditions as those that are going to require medical attention—conditions requiring more pediatric visits than the standard vaccination schedule and medical or surgical interventions.
And perhaps most importantly, the company looked for mutations that were de novo, where the parents remained unaffected.
“They’re in regions that are prone to new mutations,” Martin notes. “That’s why 22q is the most common microdeletion; it’s a region with a high number of repeat sequences that are prone to mismatch during mitosis.”
At the same time, mutations might occur elsewhere in the genome of developing fetuses, and so there has been increasing interest in applying other next-generation sequencing techniques to piece together a more complete image of fetal health.
In a recent review, Chitty and colleagues discussed the promises, pitfalls and practicalities of prenatal whole-exome sequencing (WES).
“Prior to the adoption of next-generation sequencing (NGS) techniques in prenatal diagnostics, the ability to accurately diagnose point mutations or small copy number changes in fetuses has been limited to a small number of genetic diseases with recognizable fetal phenotypes,” the authors wrote. “Many monogenic disorders present prenatally with abnormalities detected by sonography and are likely to account for a significant proportion of cases undiagnosed using cytogenetic techniques, many of which may be detectable using WES.”
Such efforts can also detect genetic anomalies in cases where the prenatal phenotype is not otherwise recognized, they continued, and to identify new candidate genes associated with development.
“A de-novo truncating mutation in the MAP4K4 gene was also found in a fetus with a complex cardiac defect, identifying a possible novel candidate gene for a human developmental disorder given its essential role in embryonic cardiac development in mouse models,” the authors offered as an example.
“Prenatal WES and WGS studies also have the potential to improve understanding of lethal genetic disorders presenting with fetal abnormalities, where the full phenotype is often unknown,” they pressed. “NGS approaches will facilitate gene discovery and may reveal Mendelian inheritance where previously missed.”
At the same time, they acknowledged, there are risks involved in finding genetic variants for which the pathogenic and medical implications are unknown. Such results have the potential to cause great anxiety and uncertainty in the patients to whom they are reported.
“Given the increasing calls to pursue WES/WGS for genetic diagnosis of fetal congenital anomalies and the paucity for most genes of systemically classified variants associated with specific fetal phenotypes, there is strong scientific premise and an urgent unmet need for a formal curation procedure for prenatal genomic variants,” the researchers suggested. “This should ideally be combined with the development of an international registry of the fetal phenotypes associated with (or caused by) these variants.”
Just as cell-free tumor DNA testing led to advances in cell-free fetal DNA, so too have advances in circulating tumor cell (CTC) characterization led to the rejuvenation of the long-known but challenging examination of fetal cell-based testing.
Back to the cell
Before focusing his efforts on fetal cfDNA, CUHK’s Lo had spent a number of years looking for fetal cells within maternal circulation, eventually giving up the effort when the cells simply seemed too rare to be useful. The technology simply wasn’t in place at the time to make this a viable option.
Newer platforms, however, mean researchers are beginning to explore the potential of whole cells.
Unlike Martin, who comes from a clinical background, Purdue University’s Cagri Savran trained as a mechanical engineer, becoming interested in the challenges of microfluidic technology. This work led to the founding of Savran Technologies.
“One of the things that really stood out to him was how slow [current microfluidic systems] were,” recounts Patrick Rivelli, company CEO. “The flow rate was extremely slow, and so it would take literally hours and hours to do one single test.”
As Rivelli explains, this would simply be unworkable for a commercial or hospital diagnostics lab. By adding pores to the bottom of the microfluidic channels, however, Savran was able to accelerate the flow of his microchips and enhance the isolation of the cells of interest.
“Instead of it taking hours, we today process a standard 8-mL patient sample in eight minutes,” Rivelli says.
In developing their platform, however, flow rate was only one important aspect of the design. Another key consideration was that the cells coming out the other end of the system be healthy and reflective of what was happening in the body.
As Rivelli explains, some microfluidic systems fix the cells, thus limiting the possibility of transcriptional analysis. Other systems rely on centrifugation, which subjects the cells to stress. And yet other capture systems, if they’re size-based, squeeze the cells through holes.
Each of these mechanisms, he argues, has the potential to change the metabolic and physical integrity of the cells.
Thus, it was important “to process a whole blood sample,” he continues. “We’ve got to manipulate it in a minimal way so we don’t put forces on these cells, so that at the end, you get a cell that is healthy and representative.”
Adding their voices to the conversation, ARCEDI Biotech’s Ripudaman Singh and colleagues at Aarhus University Hospital weighed in with their checklist of criteria for successful cell-based NIPT, which included:
- It should target specific cell type(s);
- Antibodies both sensitive and specific to those cell types must be available;
- It should be platform- and parameter-independent;
- Fetal cells should be accessible for downstream applications; and
- DNA fidelity from enriched fetal cells should be high enough for genetic analyses via chromosomal microarray or next-gen sequencing.
Last November, ARCEDI launched its own platform in hospitals in Central Denmark, conducting validation studies that compared their cell-based system to cfDNA testing and chorionic villus sampling (CVS).
“The results from the validation study thus far are promising,” said Aarhus University Hospital’s Ida Vogel in the launch announcement. “It is truly groundbreaking that cell-based testing may be able to identify not only chromosomal aneuploidies, but also smaller aberrations, such as microduplication and translocation, that currently can only be viewed through CVS and other invasive procedures.”
In fact, it was in these smaller perturbations where Singh and colleagues noted the advantages of cell-based methods.
“Currently, the NIPT field has largely been dominated by methods targeting cell-free fetal DNA in the maternal blood,” they wrote. “Even though these methods can reliably detect incidence of common aneuploidies (T21, T13 and T18), they have been ineffective in detecting CNVs, which constitute another major chunk of prenatal abnormalities.”
“This is mainly due to the fact that fetal DNA circulating in maternal blood is fragmented and is mixed with maternal DNA,” they explained. “Intact fetal cells circulating in maternal blood can mitigate this shortcoming of the cell-free NIPT, because they are the sources of pure fetal genomes.”
Another advantage of cell-based vs. cfDNA approaches, according to Rivelli, is timing.
“Typically, those cells show up in the mother’s bloodstream earlier in pregnancy,” he elaborates. “Researchers are finding them as early as seven weeks of gestational age. So there may be some circumstances where the earlier you can get information, the better.”
Even if Singh’s platform criteria are achieved, however, cell-based NIPT efforts are challenged by another parallel from the oncology world: cellular heterogeneity.
Where a tumor is comprised of a number of cells occupying different and constantly evolving mutational states, gestation is comprised not just of the fetus but also of the placenta—and these two components do not necessarily experience mutation in the same way, resulting in fetoplacental mosaicism.
“We know pretty clearly the rate of mosaicism for whole chromosome abnormalities because of CVS,” says Natera’s Martin. “For microdeletions, there are reports now for confined placental mosaicism for 22q.”
“So, we don’t know how often a microdeletion is present in the actual fetus, but we know that when you start looking at these rare things or these things like trisomy 16 or trisomy 22, they are much more likely to be confined to the placenta and not in the fetus,” she adds.
One way to compensate for any potential mosaicism is to look specifically for cells from both the fetus and the placenta, as was published last year by Changhua Christian Hospital’s Ming Chen and colleagues.
In their proof-of-concept study, the researchers designed a microfluidic platform they called Cell Reveal, to which they attached anti-CD71 antibodies to capture fetal nucleated red blood cells (fnRBCs) or anti-EpCAM antibodies to capture extravillous cytotrophoblasts (EVT). They then verified the fetal and placental origins of the cells using FISH, whole-genome amplification (WGA) and short tandem repeat (STR) analysis.
Using the microchips, the researchers were able to collect 1-44 nRBC/2 mL and 1-32 EVT/2 mL maternal blood. Likewise, molecular analysis of the harvested cells revealed correct diagnoses of respective chromosomal trisomy in each sample.
“To the best of our knowledge, this report is one of the very few studies on the successful use of circulating fetal cells for noninvasive prenatal diagnosis,” the authors concluded. “The strength of our study is that all the processes of cell capture are automatic which can be performed on a single individual case and completed within 15 h. The captured cells are available for a variety of genetic testing, such as FISH, aCGH and NGS.”
For its part, although the Savran platform currently only looks at trophoblasts, Rivelli is less concerned about the mosaicism issue if only because of how the company designed its system.
“We have the ability to capture and sequence single cells,” he explains.
“Let’s say we have a patient sample and find eight trophoblasts,” he continues. “We’re able to capture each one, one by one by one, and sequence them individually. That helps us tease out mosaicism if that turns out to be a problem.
“Single-cell sequencing is really developing, and single-cell RNA techniques are really coming along, so things you couldn’t do even five years ago are now pretty easy.”
Savran’s platform is still a work in progress, however.
“The workflow is such that it requires a little bit more attention and some manual steps,” Rivelli notes. “We think that to make this a commercially viable product, we need to automate it, we need to simplify the workflow and we need to make it look and act in a way that you’d find in a hospital central lab.”
“We’re in the early stages of automating the system,” he says. “Once we get a frozen, automated design, at that point, we’ll launch into the regulatory approval process.”
That extra time might prove invaluable not just to Savran, but to anyone working in this field, as the expanded capabilities to probe the genetics of a developing fetus come with social and ethical implications that are only getting louder.
Can vs. should
Natera’s Martin uses very blunt language: “The concept of prenatal screening has evolved from search and destroy to search, educate and optimize,” she says, before asking “What does optimize mean?”
“It’s going to mean different things to different families, but it sure means a lot more than interruption of pregnancy,” she offers. “Many families continue affected pregnancies, and the key is that they work with their health providers to develop the absolute best care plan for that infant, whatever that means for that family.”
One part of the challenge is that as with so many other biomedical efforts, including cancer, we’ve quickly moved from academic questions of what can we do to more philosophical questions of what should we do.
“There is no question that the technology has expanded to a point that is beyond the ability of either a provider or patient to make informed decisions about what we should be doing,” Martin says.
What is the goal of this, she asks? How much information should people have, do people want? What do we do when we find variants that we don’t know the significance of? What are the ethics of offering that information to a pregnant family?
“We are at a crossroads,” Martin presses. “Questions are being asked that were never asked in the ‘80s or the ‘90s.
“Nobody asked should we really be screening for Down syndrome. It was taken for granted that because we could, we should.”
And yet, she continues, one of the complaints about cfDNA and sources of resistance to making it available to all women is: What if we find something we don’t know the significance of?
“Why didn’t we ask the questions then, and why are we suddenly saying that now it’s not a good idea?” Martin asks. “It’s okay for 35-year-olds, but it’s not okay for 28-year-olds.”
Rivelli then layers on the possibilities of intervention, given recent developments with technologies such as CRISPR.
“[Recently], there was a successful gene-editing of a mouse fetus using CRISPR technologies,” he recalls. “You’re getting to the point where not only are you able to diagnose something in the fetus, you might also be able to intervene. But then you really get into an interesting conversation.”
For patient-facing Martin, though, it goes back to the individuals receiving the NIPT results.
“Ultimately, the family should decide what information they want,” she asserts. “And it’s our job as medical professionals to be able to give them the tools they need to make those kinds of decisions. Those are the basic principles that I support as an OB geneticist.”
The umbrella term “molecular diagnostics” encompasses a wide variety of techniques and technologies that are employed to analyze biological markers in an individual’s genome and proteome—that is to say, respectively, the genetic code and how cells express their genes in the form of proteins.
Thus, molecular diagnostics brings together the 20th-century development of molecular biology with the more longstanding field of general medical testing to diagnose and monitor disease, detect risk and decide which therapies might work best for individual patients. Of course, given the genomic component, molecular diagnostics is also considered a key part of eventually realizing the dream of true personalized/precision medicine.
Molecular biology itself has its original roots in the 1930s, but found its greatest growth (and clinical applications) in the late 20th century. In fact, in keeping with the prenatal focus of this molecular diagnostics special report, it is worth noting that a key piece in the maturation of molecular biology was work in 1980 by Yuet Wai Kan et al. suggesting a prenatal genetic test for thalassemia that did not rely upon DNA sequencing—which was then still a very young technology—but instead on restriction enzymes that cut DNA where they recognized specific short sequences.
Is my baby healthy?
It’s a primary question for every parent, one that may be asked with some trepidation.
“We know that 3 to 5 percent of babies are born with some birth defect or handicapping condition,” offers Kimberly Martin, vice president of global women’s health at Natera and formerly a practicing obstetrician and gynecologist. “And there are all different types of abnormalities.”
“There are chromosome abnormalities like Down syndrome,” she lists off. “There are single-gene disorders like cystic fibrosis or osteogenesis imperfecta.”
Between those extremes, she adds, there are microdeletions and duplications.
But the biggest group of anomalies, according to Martin, are birth defects such as cleft lip, heart defect or club foot.
“When karyotyping came on the scene in the ‘40s and ‘50s, it suddenly became possible to offer testing,” she says, and so by the ‘60s, the primary question was how old the mother will be when the baby’s born.
“If you said you were 35, you were high-risk, and you were offered testing,” she explains. “If you were less than 35, you were told you were low-risk, and you had a baby.”
By the late ‘80s, maternal hormone measurements were added to the repertoire, including α-fetoprotein, which is part of the quad or quadruple screen. And a decade later, ultrasound features became another screening tool, all designed to give families a greater sense of the risk level for the fetus.
“The thing with many screening tests is that it’s not a yes or no,” Martin cautions. “If I ask how old you are at delivery, and you say 36, that doesn’t mean you’re going to have a baby with Down syndrome. But you’re at higher risk and you get offered invasive testing.”
Likewise, she continues, serum screens identify about 5 percent of pregnancies as being high risk, and yet when given more invasive tests such as amniocentesis or chorionic villus sampling, more than 95 percent of those women were told their babies were fine.
The identification of fetal cfDNA in maternal blood changed the landscape.
“Doctors who had for years been scaring 5 percent of pregnant women with a positive predictive value of 3 to 4 percent, suddenly said well, gee, how does cfDNA perform?” Martin recounts.
“We found out that it has a sensitivity of about 98 percent or greater for Down syndrome,” she says. “It has a false-positive rate of less than 0.5 percent. It has a positive predictive value of about 90 percent.”
For a screening test, she concludes, it seems like a big win.
The win, however, is tempered by a conservative clinical community.
Although several clinical practice guidelines recognize the value of cfDNA screening to identify potential issues of chromosomal trisomies, they are less enthusiastic about its use for microdeletion screening. And in no case do they feel that cfDNA screening is sufficient for clinical decision-making. Rather, they recommend that any positive findings be followed by confirmatory invasive testing.
This caution is to be expected, according to Patrick Rivelli, CEO of Savran Technologies.
“They have things that they’ve been working with and are familiar with,” he says. “But I think that over time, there will be more acceptance. Really, as there is more and more data, I see it shifting.”
For Martin, it must always come back to the families.
“Every day, I saw the impact,” she recalls. “I talked to families who had a high-risk screening test. I talked to families who had a definitive abnormality on ultrasound. I talked to families who had an abnormal amniocentesis.”
“I tried very hard from when I was a resident until I became an attending physician and on into industry to meet the family where they are,” she continues. “To use language that they can understand. To be available and approachable to be asked questions. To try to make a bad situation as supportive as possible.”
Although progress has maybe been slower than she expected, Martin says she knew cfDNA was going to revolutionize how we approached prenatal screening.
“I don’t think it’s going to be too much longer before women are going to have that choice as a first-line choice, and not have to go through all the hoops to get that test,” she enthuses. “And I hope it happens sooner rather than later.”