Preeclampsia, a hypertensive disorder of pregnancy that remains among the leading causes of maternal death worldwide, has long resisted early detection — its clinical signs often arriving only after harm is already done. Researchers have now identified three blood-based molecular markers, MAPK8, CSNK1E, and NOTUM, rooted in a cellular pathway governing placental health, that together distinguish preeclamptic from healthy pregnancies with roughly 92 percent accuracy. The work does not yet offer a clinical test, but it suggests that the biological story of this disease can be read in the bloodst
Scientists identify three blood biomarkers for early preeclampsia detection
Preeclampsia is a serious pregnancy disorder associated with hypertension that poses risks to maternal and fetal health; early diagnostic biomarkers could improve outcomes.
A blood test that catches the disease before symptoms appear
Mark
Why focus on the Wnt pathway specifically? There must be dozens of cellular systems that go wrong in preeclampsia.
Mimi
True, but Wnt signaling is fundamental to how the placenta develops and how the mother's immune system tolerates the pregnancy. When it's dysregulated, you see both placental dysfunction and immune imbalance. It's not random—it's a pathway with known biological weight.
Mark
And the machine learning piece—how much of this is the algorithm finding real biology versus just pattern-matching noise in the data?
Mimi
That's the right skepticism. They used two different algorithms and looked for overlap, which is a safeguard. But you're right that retrospective studies can overfit. That's why they need prospective validation. The lab work helped—when they measured the biomarkers in actual samples, the direction of change matched the prediction.
Mark
So if this works, what changes for a pregnant woman? Does she get a blood test at her first visit?
Mimi
Potentially, yes. Instead of waiting for symptoms like swelling or protein in urine, you'd catch the disease earlier, when intervention might prevent severe complications. But we're not there yet. Right now it's a candidate test. It needs to be tested in thousands of pregnancies, across different populations, before any hospital would offer it.
Mark
What's the timeline for that kind of validation?
Mimi
That depends on funding and research capacity. Could be three to five years if the work gets picked up and resourced. Could be longer. The science is solid enough to justify the investment, but there's no guarantee the biomarkers will perform as well in a prospective cohort as they did in retrospective analysis.
Mark
And if they don't?
Mimi
Then you learn something about the limits of this particular pathway, and you keep looking. Preeclampsia is complex. There may not be a single perfect test. But even a moderately accurate blood test that catches the disease earlier would save lives.
O Pulso
- Preeclampsia still kills, and its danger lies partly in how late it announces itself — by the time blood pressure spikes and protein floods the urine, the window for early intervention may already be closing.
- Researchers focused on Wnt signaling, a cellular pathway that governs how the placenta develops, hypothesizing that its molecular breakdown would leave detectable traces in the blood long before symptoms emerge.
- Machine learning analysis of placental tissue datasets surfaced three candidate biomarkers — MAPK8, CSNK1E, and NOTUM — whose combined diagnostic model achieved an AUC of 0.920, a result that held a consistent, statistically significant pattern in actual blood samples from preeclamptic women.
- The biomarkers also correlated with natural killer cell dysfunction, linking immune disruption at the placenta to the measurable signals in the bloodstream and lending biological coherence to the findings.
- The study's retrospective design means the 92 percent accuracy figure is a proof of concept, not a clinical promise — prospective validation across larger, more diverse populations stands between this discovery and any routine pregnancy test.
Preeclampsia, a hypertensive disorder of pregnancy that remains among the leading causes of maternal death worldwide, has long resisted early detection — its clinical signs often arriving only after harm is already done. Researchers have now identified three blood-based molecular markers, MAPK8, CSNK1E, and NOTUM, rooted in a cellular pathway governing placental health, that together distinguish preeclamptic from healthy pregnancies with roughly 92 percent accuracy. The work does not yet offer a clinical test, but it suggests that the biological story of this disease can be read in the bloodstream before the body sounds its loudest alarms — a possibility that, if validated, could quietly shift the odds for mothers and their children.
Preeclampsia remains one of the leading causes of maternal death worldwide, and its cruelest feature is timing: the clinical signs that doctors have long relied upon — swollen limbs, protein in the urine, a blood pressure cuff reading too high — tend to appear only after the disease has already taken hold. A research team has now identified three blood biomarkers that could shift that timeline, offering the possibility of detection before the damage begins.
The work is grounded in Wnt signaling, a cellular pathway that governs how the placenta grows and functions. When this pathway misfires during pregnancy, the consequences ripple outward. The researchers mined large datasets of placental tissue from preeclamptic and healthy pregnancies, identified eleven Wnt-related genes behaving differently in the disease, then applied machine learning to narrow the field. Three markers emerged: MAPK8, CSNK1E, and NOTUM. A diagnostic model built on these three achieved an area under the curve of 0.920 — roughly 92 percent accuracy in distinguishing preeclamptic from normal pregnancies.
The pattern held in actual blood samples. CSNK1E and NOTUM were elevated in preeclamptic women; MAPK8 moved in the opposite direction. The findings also connected to natural killer cell dysfunction, tying immune disruption at the placenta to the bloodstream signals the researchers had identified — a biological coherence that lends weight to the results.
The caveat is significant. This was a retrospective study, looking backward at existing data rather than following new pregnancies forward. The accuracy achieved is promising but preliminary, and the road from laboratory finding to a test offered at a routine prenatal visit is long. Prospective validation in larger, more diverse cohorts is required before these markers could ever reach clinical practice. What the work establishes, for now, is a proof of concept: that the molecular logic of preeclampsia can be read in the blood, and that machine learning can help decode it.
Preeclampsia kills. It's a pregnancy disorder marked by dangerous spikes in blood pressure, and it remains one of the leading causes of maternal death worldwide. For decades, doctors have relied on clinical observation—swelling, protein in the urine, elevated readings on a cuff—to catch it. But by the time those signs appear, the damage is often already underway. A team of researchers has now identified three blood biomarkers that could change that calculus, offering the possibility of catching the disease earlier, before it takes hold.
The work centers on a cellular pathway called Wnt signaling, which controls how cells grow, move, and die. When this pathway goes wrong in pregnancy, the placenta suffers. The researchers hypothesized that if they could find the molecular fingerprints of that dysfunction in the bloodstream, they might be able to diagnose preeclampsia before a woman's blood pressure spikes dangerously high.
They started by mining two large datasets of placental tissue from preeclamptic and healthy pregnancies. From these, they identified eleven genes involved in Wnt signaling that behaved differently in preeclampsia. Then they applied machine learning algorithms to narrow the field. Three biomarkers emerged: MAPK8, CSNK1E, and NOTUM. When the researchers built a diagnostic model using these three markers, it achieved an area under the curve of 0.920—a score that translates to roughly 92 percent accuracy in distinguishing preeclamptic from normal pregnancies.
The findings held up in preliminary lab work. When the team measured these biomarkers in actual blood samples from preeclamptic women, two of the three—CSNK1E and NOTUM—were elevated compared to controls. MAPK8 went the opposite direction, dropping in preeclamptic samples. The pattern was consistent and statistically significant. The researchers also found that these biomarkers correlated with changes in natural killer cells, immune cells that play a role in placental health and pregnancy tolerance.
But there's a crucial caveat. This was a retrospective study, meaning the researchers looked backward at existing data and samples. The diagnostic accuracy they achieved is promising, but it's preliminary. Before these biomarkers could ever appear on a clinical test that a pregnant woman might take at her routine checkup, they would need to be validated in a prospective study—one that follows new pregnancies forward in time, in larger and more diverse populations. The researchers themselves acknowledge this limitation. They've identified candidates with real potential, but the path from laboratory finding to bedside tool is long and uncertain.
What makes this work significant is the specificity of the approach. Rather than casting a wide net for any blood marker that differs in preeclampsia, the researchers focused on a known biological pathway implicated in placental dysfunction. That focus lends credibility to the findings and suggests these aren't just statistical artifacts. If validated in larger studies, a blood test based on these three markers could offer pregnant women earlier warning, potentially allowing for closer monitoring or earlier intervention before preeclampsia becomes severe. For now, the work stands as a proof of concept—a demonstration that the molecular logic of preeclampsia can be read in the blood, and that machine learning can help decode it.
Citações Notáveis
Before clinical applicability can be assessed, prospective validation in adequately powered and phenotypically diverse cohorts is required— Study authors
Quer a matéria completa?
Leia o original em Nature ↗
Technology