DEVELOPING A BEDSIDE TEST FOR PREECLAMPSIA: OVERCOMING SISYPHUS

OUR CHALLENGE

Preeclampsia affects 5–7% of all pregnancies—over 200,000 per year in the United States alone—and is one of the leading causes of maternal deaths and preterm births worldwide (1,2). This disease presents with vague signs and symptoms and, without much warning, precipitously progresses to widespread organ failure. Preeclampsia catches clinicians by surprise; in the absence of close monitoring or immediate access to care, women present in acute distress with advanced disease, portending adverse outcomes for both mother and baby.

Historically, preeclampsia has been diagnosed using a collection of clinical signs and symptoms and a menu of laboratory tests. Routine prenatal visits consist of blood pressure and urine dipstick assessments. These are considered screening measures for preeclampsia and typically reassure caregivers when results are in the normal range. Over the years, guideline committees have developed disease-defining cutoffs for many parameters, such as 140 mm Hg systolic and 90 mm Hg diastolic for blood pressure, a protein creatinine ratio of ≥ 0.3 mg/dL, and laboratory thresholds defining abnormal liver function tests and platelet counts, but also added less quantitative measures such as abdominal pain and visual disturbances. The lack of specificity of any one measure necessitates a constellation of several continuous measures (e.g., blood pressure, platelet counts) and clinical complaints (e.g., visual disturbances) to serve as the backbone for the diagnosis (3). Detecting and assembling these signs, symptoms, and laboratories also requires routine (and not episodic) prenatal care, which itself poses a challenge. Up until recently, a reliable, robust biomarker specific for preeclampsia had not been developed.

If preeclampsia had a “troponin equivalent” specific test as is available for cardiac ischemia, it could alert the clinician ahead of time, before a woman is truly sick, that the otherwise non-specific signs and symptoms she is experiencing are in fact harbingers of impending severe disease. Although guidelines represent clear-cut thresholds that define the condition, the disease does not turn on or off like a light switch. When women present early, these thresholds are typically not met, they frequently vacillate (e.g., blood pressure and degrees of proteinuria), and their severity is difficult to gauge (e.g., abdominal pain) making it enormously difficult to appropriately triage women. A specific test for preeclampsia that not only indicates impending disease but also distinguishes preeclampsia from conditions that may mimic preeclampsia, such as gestational or chronic hypertension, lupus, or underlying kidney disease, could facilitate care before severe end-organ damage or fetal compromise and define eligibility for a directed therapy in the future.

Exhilarating Early Years

In 2003, Dr. S. Ananth Karumanchi and I agreed to combine our efforts to develop a diagnostic test and a therapeutic intervention for preeclampsia. His team had published a seminal paper (4) hypothesizing angiogenic markers linked to preeclampsia, and our team had established a large pregnancy cohort to identify and test biomarkers and set the stage for therapies (5). In 2004, we joined forces with investigators from the National Institute of Child Health and Human Development to demonstrate that, compared to women who do not progress to preeclampsia, a severalfold increase in levels of soluble fms-like tyrosine kinase 1 (sFlt-1, which is an antiangiogenic factor) and decrease in levels of free placental growth factor (PlGF, which is a proangiogenic factor) antedated the diagnosis of preeclampsia by nearly five weeks (6). Both angiogenic factors are released from the placenta and notably increase in normal pregnancies with advancing gestational age.

We proposed that the ratio of sFlt-1/PlGF would be a clinically simple and useful tool (analogous to urine protein/creatinine ratio) and further demonstrated that this ratio is dramatically altered in women with preterm (before 37 weeks) preeclampsia (7). Our early studies also suggested that low first trimester PlGF (but not sFlt-1) blood levels strongly predicted the development of preeclampsia later in pregnancy (8) and that urinary levels of PlGF displayed favorable predictive characteristics (9).

While these early observations were quite promising, even bordering on quixotic, our biomarkers sat in an ocean of dozens of other promising biomarkers that enjoyed initial enthusiasm but then faded over time in their utility. To avoid a similar fate, we had to demonstrate that our test was robust, consistent across populations, reproducible in the hands of other investigators, and specific for preeclampsia and its related adverse outcomes. After pressure testing the ratio ourselves so as not to be misled, our goal was to transition these markers from simply being research interests to actually becoming clinically useful.

Using the Bradford Hill Criteria (10) as our “GPS,” we embarked on a journey to show that the angiogenic biomarkers were not simply associated with preeclampsia but were intimately tied to the pathogenesis of the disease itself. To strengthen the biomarker-disease relationship, we published a series of early studies supporting a dose-response relationship (7,8,11), demonstrated consistency across different populations including susceptibility among women with diabetes (12,13), found evidence of temporality with the diagnosis (7,14,15), and performed additional experimental studies to substantiate their biological plausibility (16). Furthermore, the advent of automated platform assays we validated against research ELISA assays (17,18) was critical to testing larger, more diverse populations and allowed investigators from around the world to highlight the specificity of the ratio for preeclampsia and not for other conditions that could mimic preeclampsia in pregnancy, such as intrinsic kidney disease (19–23).

Supportive European Experience

In Europe, we were fortunate to have strong champions who could envision with us how the diagnostic test could be incorporated into routine prenatal clinical care. Toward this end, leaders from Europe published a series of supportive studies (21,23–30). Data continued to accumulate, all suggesting that testing with the ratio could aid clinicians in their decision making. Given growing clarity of when to perform the test and how best to use the results, Roche Diagnostics (one of our earliest licensees) obtained the CE mark, which indicates that a product meets the European safety, health, and environmental protection requirements, but not necessarily efficacy standards, required to sell medical devices in the European Union, in 2009. In 2016, Roche Diagnostics funded a multicenter study in Europe to determine whether a ratio below a cutoff predicted the absence of preeclampsia within one week and whether a ratio above that same cutoff predicted the presence of preeclampsia within four weeks (31). Women in the study were primarily Caucasian, and their median blood pressures at enrollment were not above 140 mm Hg systolic or above 90 mm Hg diastolic. At a ratio of 38 or below, the negative predictive value within one week was 99.3%, while the positive predictive value for preeclampsia within four weeks was 36.7% at a ratio above 38. These results were also confirmed in an Asian population (32). Over time, the ratio was included in several guidelines (23,33–35), and its use expanded across Europe (36,37). Progress in the United States stalled, however, despite several attempts led by the Preeclampsia Foundation to galvanize the community (https://www.preeclampsia.org/biomarkers).

U.S. Efforts: Sisyphus Overcome

Up until 2018, studies in the United States were largely single-center experiences (11,38–40). A definitive study in the United States that would support regulatory approval of the ratio as a clinical test would not only have to be racially and ethnically representative of the population of women who give birth in the United States, but it would also have to be adequately powered, demonstrate unequivocal efficacy, include both tertiary care and community hospitals, and have built-in measures to protect women in the event of false positive or negative results. Our pleas to the incumbent in vitro diagnostic (IVD) companies to fund such a study went largely unanswered. They had little appetite for making a sizable investment in any single test, especially given the uncertainty of the return on that investment. We were forced to seek another funding strategy to bring the test to the bedside.

We conducted our early research as young assistant professors in Boston, but the time came for the two of us (Drs. Karumanchi and Thadhani) to search for academic opportunities outside of Boston to catapult our careers. We joined the faculty at Cedars-Sinai in Los Angeles, where Dr. Shlomo Melmed agreed not only to support our research laboratories but also to fund a definitive study in the United States for regulatory approval of our test. Cedars-Sinai was instrumental in allowing us to coordinate all aspects of the study from a single academic home. Along the way, we were fortunate to recruit Dr. Sarah Kilpatrick, Chief of Obstetrics and Gynecology at Cedars-Sinai, to help with the details of design and execution.

If we, as academicians, were entirely responsible for the design, performance, analyses, and sample testing of a study to support the regulatory approval of the ratio, we would have to meet all the rigorous efficacy and safety standards expected by the U.S. Food and Drug Administration (FDA). Therefore, our next step was to seek guidance directly from the FDA. We asked three interested IVD company representatives to accompany us on the visit so they could hear firsthand the FDA’s feedback. In 2018, our team met in person with the FDA to highlight the importance of the study and the details of its design and to seek guidance as to whether it would accept a large, rigorous study performed by academicians (yet eventually submitted to the FDA by established IVD companies based on their own platform specifications). The FDA’s feedback was instructive and strongly supportive.

Clinicians who routinely cared for women with preeclampsia said they did not need a test to diagnose preeclampsia, as the published criteria were clear and key measures could be obtained by clinical assessments and routine laboratories. Instead, they suggested we focus on identifying women at high risk for progressing to preeclampsia with severe features (i.e., those women who would become acutely sick) and determining where early detection could lead to a change in management. Preliminary findings from two small single-center studies we performed in Boston supported such an approach (15,17). Key to our strategy was to design a study that would mitigate the risks and consequences of false positives and false negatives. For this reason, we developed a protocol to test only hospitalized women with a hypertensive disorder of pregnancy to ensure that all elements of care (clinical assessment, standard laboratories), in conjunction with our ratio test, would be used to make clinical decisions and that we would provide the clinician with a window (of two weeks) to guide intense monitoring and interventions (e.g., corticosteroids, step-up care).

We chose the automated KRYPTOR assay (manufactured by Thermo Fisher) that we had previously validated (18). Thermo Fisher provided support for the assays and partial support for the clinical research organization (CRO). Planning of the study began in earnest in 2018. We held an investigator meeting in Boston and initiated the study by the end of that year. Leading maternal fetal medicine experts were enthusiastic, and 18 sites across the United States were accruing subjects in full force by the third quarter of 2019. Just when the study was ahead of schedule, however, the COVID pandemic of 2020 put a damper on enrollment. This was not an interventional study; understandably, the local study staff were called upon to address the priorities of acute patient care. Although enrollment languished, it did not stop. By late 2020, we were back on track given that a blood draw (following consent) required minimal patient interaction.

ICON, the independent CRO, performed initiation and monitoring visits, confirmed proper source documentation of the clinical variables, verified informed consent procedures and adverse effects reporting, and validated biospecimen accountability and storage. Critical electronic entries and data capture variables necessary for the primary and secondary outcomes were carefully audited. During weekly calls with my co-principal investigators (Drs. Karumanchi and Kilpatrick), our study staff, and the CRO team, we focused on troubleshooting enrollment challenges, ensuring data quality, and coordinating sample transfers and testing. The study was completed by the end of 2021. Without knowledge of the sFlt-1/PlGF ratio results or the local site diagnosis, experienced maternal fetal medicine physicians adjudicated the primary outcomes using the American College of Obstetricians and Gynecologists guideline definitions of preeclampsia with severe features.

Results from the study confirmed our hypothesis: a ratio cutoff of 40 yielded a negative predictive value of 96% and a positive predictive value of 65% for the development of preeclampsia, with severe features within two weeks. Importantly, the ratio was also able to predict surrogates of severity of the disease including timing of delivery and adverse maternal and fetal/neonatal outcomes. The ratio performed better than any single or any combination of other standard clinical and laboratory measures used in this setting (41). The elements of a dose-response relationship, consistency across different populations, a temporal relationship with the outcome, and a strong link, not only with the disease, but also with adverse outcomes tied to the disease were also evident. Efficacy and safety data from our clinical study along with assay-specific verification and validation studies were reviewed by the FDA as it deliberated a de novo prognostic test (class II regulatory device) approval, which was granted in May 2023.

End of the Beginning

It took Dr. Karumanchi and me two decades of experimental, observational, and translational studies, with steadfast supporters along the way, to transform our work into an approved test for preeclampsia (42). In parallel with these efforts, we also performed early proof of principle studies to treat human preeclampsia (43,44), which provided the path to developing targeted interventions that are forthcoming (National Clinical Trial number 02923206). We have also supported studies using modern interfering techniques (National Clinical Trial Number 05881993). Interventional studies would not be possible without a specific, robust test for preeclampsia that correlates with adverse outcomes related to the disease. As with any new test, however, additional studies are needed to fully understand the boundaries of when and how best to use the ratio. These boundaries will reveal themselves as we and others perform real-world studies and further explore whether outpatient testing is safe. The latter is necessary because most prenatal care occurs in the outpatient setting since women often present first to an outpatient clinic and not to a hospital. To facilitate these efforts, we are also developing point-of-care screening tests for more widespread use, such as in rural settings, where tertiary and specialty care is frequently limited.