EP3999855A1 - Method of prognosing and diagnosing preeclampsia - Google Patents

Abstract

The present invention provides for a method of prognosing or diagnosing preeclampsia in a subject wherein the method comprises measuring the level of at least four biomarkers in a blood sample from the subject; optionally determining at least one clinical cofactor from the subject; generating a prediction value, wherein the prediction value indicates whether the subject will develop or will not develop preeclampsia or has preeclampsia; the prediction value being based on the levels of the at least four biomarkers in the blood sample from the subject and optionally based on the at least one clinical cofactor.

Description

METHOD OF PROGNOSING AND DIAGNOSING PREECLAMPSIA

Field of the Invention

The present invention relates to a method of prognosing or diagnosing preeclampsia in a subject. The method comprises the measuring of biomarker levels in a sample, e.g. blood sample, taken from the subject, optionally determining at least one clinical cofactor from the subject, and generating a prediction value which indicates whether the subject will develop or will not develop preeclampsia within a time period, or has preeclampsia.

Background

Preeclampsia (PE) is a sudden and severe complication that develops during the second half of gestation and affects both the mother and the unborn child. Despite the general improvement of the socio-economic environment and availability of clinical service in developed countries over the past 20-30 years, the prevalence of PE has remained unchanged. Across 129 international independent studies, the overall estimates for preeclampsia and eclampsia were reported 4.6% (95%CI 2.7-8.2), and 1.4% (95%CI 1.0- 2.0) of all deliveries, respectively (Abalos et al., 2013). In Europe, this extends up to 400,000 cases annually. Although the main clinical symptoms of PE are maternal hypertension and proteinuria, it may be accompanied with thrombocytopenia, renal and liver insufficiency, pulmonary oedema, cerebral and visual impairment and uteroplacental dysfunction. In extreme cases, the disease may progress to eclampsia characterized by severe epileptic seizures due to cerebral oedema and cause maternal and foetal death. The challenge in the clinical management is to distinguish pregnancies with isolated gestational hypertension or increased protein levels in urine from the cases when these symptoms refer to the development of PE. Currently, the only truly curative intervention for PE is delivery of the baby. However, in a large proportion of PE cases it brings along premature birth (<37 gestational weeks) with a broad palette of complications to the new born.

PE is considered a‘disease of placenta’, caused by defective placental formation and/or its impaired functional capacity. Placental secreted molecules that are released into the maternal blood stream have been the most popular source to identify non-invasive biomarkers for PE. It has been acknowledged that increased soluble fms-like tyrosine kinase-1 (sFItl) and decreased placental growth factor (PIGF) represent hallmarks of abnormal placental function and their ratio represents a widely accepted parameter for estimating PE risk. Although many biomarkers have been proposed, only a few show consistent performance. Despite the lack of targeted cure, it is acknowledged that early identification of pregnancies that are susceptible to PE would enable timely personalized clinical management and reduction of the most severe consequences of PE. Currently, initial assessment of a potential risk to develop PE is based on maternal increased blood pressure and/or presence of protein in the urine. Medical history of existing chronic hypertension, nulliparity and other conditions linked with increased risk to PE are also considered. However, this approach lacks sensitivity and specificity as only 35-40% of all cases of PE show any of these pre existing risk factors. The remaining challenge in the clinical management is to distinguish pregnancies with isolated gestational hypertension or increased protein levels in urine from the cases when these symptoms refer to the developing PE.

Since 2014, maternal serum analysis using immunoassay-based tests of sFItl and PIGF has been offered by commercial companies, Roche, Thermo Fischer Scientific and PerkinElmer (only PIGF) (NICE 2016). The tests aim to identify the women whose pregnancy will be complicated by severe PE requiring delivery within 7 days (for the sFlt1/PIGF ratio) or 14 days (for PIGF only) from symptomatic women between 20- and 36-weeks’ gestation. These established tests based on sFlt1/PIGF have several shortcomings. Firstly, although the reported negative predictive value (NPV) of a sFlt1/PIGF ratio <38 to‘rule-out’ or exclude the development of PE within a week is high (e.g. 99.3% in PROGNOSIS study using Elecsys test (Roche)), the calculated positive predictive value (PPV) of sFlt1/PIGF ratio >38 to‘rule- in’ PE within the next 4 weeks was only 36.7%. Applying the test among pregnant women at the 30^th-37^th week of gestation, the PPV of sFlt1/PIGF ratio >38 for the prediction of delivery with PE <1 week was 1.9% and <4 weeks 10.4%, while NPV of sFlt1/PIGF ratio <38 was 99.97% and 99.85%, respectively (Dragan 2017). Although, the current test has good accuracy to predict short term absence of PE, the international guidelines do not recommend routine use in clinical practise without further studies (ISSHP, 2018).

Secondly, the test has an especially high fraction of inconclusive results in predicting PE development after 34 gestational weeks. This clinical group is critical as it represents >90% of all PE cases and typically the patients have no clearly identifiable placental developmental pathology. Thirdly, false-positive predictions based on solely sFlt1/PIGF ratio may arise due to placental malfunction for other reasons or fetal growth restriction, both reflected by low PIGF values. Lastly, the current offered tests are expensive for clinical routine and thus, cannot be developed further as a screening tool to be offered to a wide community of pregnant women.

Dependable and early prediction and/or diagnosis is therefore crucial for successful treatment interventions in hypertensive disorders of pregnancy including inter alia PE. Consequently, provision of further, alternative and preferably improved methods and means for diagnosis, prediction and/or prognosis of hypertensive disorders of pregnancy continues to be of prime importance.

There is therefore a need to provide a robust, cost-effective test to prognose and diagnose PE in expectant mothers.

Summary of the Invention

The Inventors have surprisingly found that the method of the present invention allows for a precise and usable prognosis or diagnosis of preeclampsia (PE), i.e. whether the subject will develop or will not develop PE or whether the subject has PE. In particular, such precision and usability enables accurate early prediction or diagnosis of PE allowing the application of interventions to minimise the worst consequences of PE. The exclusion or ruling out of the development of PE will minimise any further follow up and reduce maternal stress during pregnancy.

Thus, in a first aspect of the present invention, there is provided a method of prognosing or diagnosing preeclampsia in a subject wherein the method comprises, measuring the level of at least four biomarkers in a blood sample from the subject; optionally determining at least one clinical cofactor from the subject; generating a prediction value, wherein the prediction value indicates whether the subject will develop or will not develop preeclampsia, or has preeclampsia; the prediction value being based on the levels of the at least four biomarkers in the sample, e.g. blood sample, from the subject and optionally based on the at least one clinical cofactor; wherein the at least four biomarkers are sFItl , PIGF, ADAM 12, sENG and optionally leptin; the at least one optional clinical cofactor is selected from gestational age and the weight of the subject, wherein a high sFItl , low PIGF, low ADAM12 and high sENG level as compared with control indicates an increased probability of the subject developing preeclampsia or increased probability that the subject has preeclampsia; and if included in the calculation of the prediction value, a high leptin level as compared with control indicates an increased probability of the subject developing preeclampsia or increased probability that the subject has preeclampsia; and if included in the calculation of prediction value, a low gestational age and high weight of the subject at blood sampling indicates an increased probability of the subject to developing preeclampsia or increased probability that the subject has preeclampsia.

Advantageously, it is expected that the method may be used to prognose, i.e. whether the subject will develop PE or will not develop PE, or diagnose PE development in pregnant women on or after the 179^th gestational day. The method comprises measuring the level of at least four biomarkers in a sample, e.g. blood sample, from the subject, preferably the at least four biomarkers are selected from sFItl , PIGF, ADAM 12 and sENG. In at least some embodiments, at least one clinical cofactor from the subject is also determined, preferably the at least one clinical cofactor is selected from gestational age at blood sampling.

In some embodiments of the invention, the biomarkers are sFItl , PIGF, ADAM 12 and sENG or sFItl , PIGF, ADAM12, sENG and leptin and the clinical cofactors at blood sampling are gestational age and the weight of the subject.

The prediction value may, for example, be calculated using the following formulas:

A prediction value (p(i)) equal to, above, higher or greater than the threshold value, indicates that the subject will develop preeclampsia within a two-month period from blood draw/blood sampling or has preeclampsia, i.e. is predictive or diagnostic of PE.

A prediction value (p(i)) lower or less than a threshold value, indicates that the subject will not develop preeclampsia within a two-month period from blood draw/blood sampling, i.e. excludes the risk of developing PE.

Brief Description of the Figures

Figure 1 shows the study design, workflow and key stages in the Luminex® 6PLEX assay development. The workflow included the following steps: selection of preeclampsia biomarkers (I); development and quality assessment of singleplex and multiplex immunoassays (II); experimental evaluation; the applicability and reliability of the developed Luminex® 6PLEX assay implemented on serum samples drawn from pregnant women (Table 5) (III); and development of preeclampsia prediction model based on Luminex® 6PLEX assay (IV).

Figure 2 shows the estimated analytical accuracy of the singleplex measurements of standard reference proteins in General Assay Diluent (GAD). The solid line shows the actual measured median fluorescence (MFI) values of the serial dilutions of each biomarker (from Standard-8 to Standard-1). The analytical accuracy shows the closeness of expected and calculated concentration values of standard proteins, reported in percentage (accuracy %; dashed line). Standard-1 represents only GAD. Additional details are provided in Tables 1 ; 2; 3; 4.

Figure 3 shows the estimated analytical accuracy of the multiplex measurements of standard reference proteins in the General Assay Diluent (GAD). Biomarker measurements were determined with the developed Luminex® 6PLEX assay. The solid line shows the actual measured MFI values of the serial dilutions of each biomarker (from Standard-8 to Standard l The analytical accuracy expresses the closeness of expected and calculated

concentration values of standard proteins, reported in percentage (accuracy %; dashed line). Standard-1 represents only GAD. Additional details are provided in Tables 1 ; 2; 3; 4.

Figure 4 shows the Singleplex vs 6PLEX Luminex® assay performance. (A) The median fluorescence intensity (MFI) values of the reference proteins measured in singleplex and multiplex assay formats were highly correlated across the tested concentration ranges (Spearman correlation coefficient, r = 1.00, P < 0.0005). (B) A statistically significant strong correlation (r ³ 0.83; P < 0.05) was also detected between the singleplex and multiplex assay measurements of ADAM12, SENG, leptin, PIGF and sFItl in the sera of healthy pregnant women (n=6, 12-36 gestational week) and non-pregnant females of reproductive- age (n=2). For PTX3 the range of measured concentrations was low and the sensitivity of the 6PLEX assay was systematically reduced compared to the singleplex test (r = 0.22; P = 0.063). Additional details are provided in Table 11 ; Table 7.

Figure 5 shows the comparison of the performance of the Luminex® 6PLEX assay developed in the current study and the B R A H M S assays for sFItl and PIGF (Thermo Fisher Scientific). In total, 61 serum samples were measured drawn from pregnant women during III trimester of gestation. (A, A’, A”) Correlation (Spearman R) between the two assays for the measured sFItl and PIGF and the estimated sFlt1/PIGF ratio. The diamond symbol refers to sample III-9 with an outlier PIGF value. (B, B’, B”, C, C’, C”) Distribution of the measured sFItl and PIGF and the estimated sFlt1/PIGF ratio in clinical subgroups stratified by the clinical symptoms and the time to the diagnosis of PE relative to the time of the blood draw. Statistical significance for the distributions between subgroups was assessed using Mann-Whitney test.

Figure 6 shows the performance of the Luminex® 6PLEX assay measurements of serum sENG, ADAM12 and leptin among pregnant women. Highly correlated measurements of serum sENG and sFItl levels in pregnant women. (A) R, Pearson correlation coefficient; (B) Distribution of the sENG; (C) ADAM12; (D) leptin. Circulating levels in clinical subgroups stratified by the symptoms and time to the onset of preeclampsia relative from the blood draw. Statistical significance for the distributions between subgroups was assessed using Mann-Whitney test.

Figure 7 shows Receiver Operating Characteristic (ROC) curves for the preeclampsia (PE) prediction derived from the alternative modelling approaches of Luminex® 6PLEX assay data compared to the sFlt1/PIGF ratio based estimates using the B R A H M S

immunoassays (Thermo Fisher Scientific). Model A combined additively the measurements of five biomarkers (ADAM12, sEng, sFIt, leptin, PIGF), whereas Model C excluded leptin in the PE prediction formula. Model J was based on sFlt1/PIGF ratio estimated from either the Luminex® 6PLEX assay or the B R A H M S immunoassay measurements. All models were tested using three alternative analytical settings. Setting 1 utilized solely biomarker data. In Setting 2, biomarker measurements were adjusted to gestational age and in addition, prior to modelling the leptin serum concentrations were corrected to maternal weight as

recommended (Blum et al 1997). In Setting 3 biomarker data were adjusted for gestational age and maternal weight at blood draw. Predictive performance of the models was evaluated using the measurements of prospectively collected serum samples from pregnant women (development of PE within 4-62 days, n=25 vs. no PE during the index pregnancy, n= 32). Building of ROC curves and estimation of their characteristics was performed using general linear models (glm) implemented in R. Detailed information on the models is provided in Table 12 and 13. (A) Comparative ROC curves from alternative settings and combinations of biomarkers. (B) Area under curve (AUC) values and 95% Cl for each presented prediction model.

Detailed Description of the Invention

It is to be understood that different applications of the disclosed methods may be tailored to the specific needs in the art. It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments of the invention only and is not intended to be limiting. In addition, as used in this specification and the appended claims, the singular forms“a”,

“an”, and“the” include plural referents unless the content clearly dictates otherwise. Thus, for example, reference to“an inhibitor” includes two or more such inhibitors, or reference to “an oligonucleotide” includes two or more such oligonucleotide and the like.

The term "biomarker" is widespread in the art and may broadly denote a biological molecule and/or a detectable portion thereof whose qualitative and/or quantitative evaluation in a subject is, alone or combined with other data, predictive and/or informative (e.g., predictive, diagnostic and/or prognostic) with respect to one or more aspects of the subject's phenotype and/or genotype, such as, for example, with respect to the status of the subject as to a given disease or condition. Particularly, biomarkers may be metabolite-, RNA- (esp. mRNA-), peptide-, polypeptide- or protein-based, preferably peptide-, polypeptide- or protein-based.

The terms "gestational age", "age of gestation" and similar are widespread in the art and commonly denote the time as measured in days or weeks, preferably days, from the 1st day of a female's last menstrual period. Thus, gestational age may be defined as the age of the pregnancy from the last normal menstrual period. Normal term pregnancy may be defined as a length of 259 gestational days and longer (up to 294 days), preferably 280 days. A human pregnancy of normal gestation may alternatively be defined as between about 38 and 42 weeks, preferably about 40 weeks.

All publications, patents and patent applications cited herein, whether supra or infra, are hereby incorporated by reference in their entirety.

Prognosing or diagnosing preeclampsia

The method of the invention has utility in a number of applications including prognosing preeclampsia, predicting whether a subject, specifically a pregnant subject, will develop preeclampsia or whether preeclampsia risk can be excluded, and diagnosing preeclampsia. Preferably, the method of the invention prognoses preeclampsia. The method of the invention may also have utility in choosing optimal management of preeclampsia in a subject and/or monitoring the effect of a possible treatment or treatments. Preferably, the method of the invention has utility in prognosing preeclampsia, predicting whether a subject, in particular a pregnant subject, will or will not develop preeclampsia and diagnosing preeclampsia at the beginning, start of, or during the third trimester of pregnancy or after the 179^th gestational day. The method of the invention facilitates determining an appropriate treatment regimen for the subject.

The term“ preeclampsia” or“ pre-eclampsia” (PE) means a multisystem complication of pregnancy comprising high blood pressure or hypertension and one or more additional symptoms selected from: swelling of the hands, feet and/or face (oedema), persisting headache, visual disturbances, liver, renal or hematologic dysfunction (including proteinuria, elevated liver enzymes, hemolysis, thrombocytopenia), epigastric pain and eclamptic seizures and uteroplacental dysfunction. PE generally denotes a pregnancy-associated disease or condition that resolves by 12 weeks postpartum.

Based on the most recent recommendations by the International Society for the Study of Hypertension in Pregnancy (ISSHP) (Brown et al. 2018) preeclampsia can be diagnosed when gestational hypertension is accompanied by ³1 of the following new-onset conditions at or after 20 weeks’ gestation:

• Proteinuria: protein 1 + by dipstick on urine analysis, >300 mg of protein in a 24-hour urine collection, or a single random urine sample having a protein/creatinine ratio >0.3 mg/mg

• Other maternal organ dysfunction, including: acute kidney injury, liver involvement with or without right upper quadrant or epigastric abdominal pain, neurological or haematological complications

• Uteroplacental dysfunction: such as fetal growth restriction, abnormal umbilical artery Doppler wave form analysis, or stillbirth

Gestational hypertension is defined as a systolic blood pressure (BP)>140 mmHg and/or a diastolic BP>90 mmHg after 20 weeks gestation (generally measured on two occasions over 4 hours apart, e.g. about 4 to about 100 hours apart).

Preeclampsia typically occurs in the third trimester of pregnancy, i.e. on or after 28^th week of pregnancy. Preeclampsia may already occur on or after the 20^th week of pregnancy. If preeclampsia is not treated, it can lead to brain oedema that causes eclamptic seizures that are not related to a pre-existing neural condition, and even death of the mother and/or baby. In an embodiment of the present invention, prediction of development or diagnosis of the preeclampsia can be made on or after the 179^th gestational day. Preeclampsia may occur on or after 28^th week of pregnancy.

Prognosing preeclampsia or preeclampsia prognosis means providing a prediction of whether a subject will or is likely to develop preeclampsia. Prognosing preeclampsia or preeclampsia prognosis is a preeclampsia prediction or prediction of preeclampsia onset. Prognosing preeclampsia is a prediction of the subject’s susceptibility or risk of developing preeclampsia; a prediction of the course of disease progression and/or disease outcome, for example expected onset of the preeclampsia, expected severity and course of the preeclampsia, expectations as to whether the preeclampsia will develop into eclampsia; a prediction of the subject’s responsiveness to treatment for the preeclampsia, for example, a prediction of a subject’s responsiveness to treatment for the preeclampsia, for example positive response, a negative response, no response at all. Prognosis includes predicting whether or not an individual will develop preeclampsia, whether or not they will need treatment and/or whether the progress of the disease will be fast or slow. In an embodiment of the present invention, a prediction value is generated wherein the prediction value indicates whether the subject will develop or will not develop preeclampsia, or has preeclampsia.

Diagnosing preeclampsia or preeclampsia diagnosis means providing a determination of whether a subject has preeclampsia. Diagnosing preeclampsia is a determination as to whether a subject, for example a subject that has some clinical symptoms of preeclampsia, a subject that is asymptomatic for preeclampsia but has risk factors associated with preeclampsia, a subject that is asymptomatic for preeclampsia and has no risk factors associated with preeclampsia, is presently affected by preeclampsia; a classification of the subject’s preeclampsia into a subtype of the disease or disorder; a determination of the severity of preeclampsia.

Monitoring preeclampsia means monitoring a subject’s condition, for example to inform a preeclampsia diagnosis, to inform a preeclampsia prognosis and/or to provide information as to the effect or efficacy of a preeclampsia treatment. Treating preeclampsia means prescribing or providing treatment of preeclampsia in a woman and may include preventing the preeclampsia from occurring in a subject which may be predisposed to preeclampsia but has not yet been diagnosed as having preeclampsia; inhibiting preeclampsia, i.e. arresting preeclampsia development; relieving, curing or regressing preeclampsia. If no preeclampsia is predicted in the subject, no special monitoring of the subject is required for at least two months. If preeclampsia is predicted, the subject has to be monitored regularly, for example appointments with the subject each week to measure blood pressure; to test urine protein level; documenting excessive weight gain; monitoring of fetal wellbeing; measuring blood liver enzymes; hemogram; coagulogram; headache frequency; visual disturbances;

epigastric pain, swelling of the hands or face.

Subject and sample

The terms“individual,”“subject,”“host,” and“patient,” are used interchangeably herein and refer to any subject for whom diagnosis, treatment, monitoring or therapy is desired. An individual, subject, host or patient may be a human. The individual, subject, host or patient is preferably a human. The subject may be asymptomatic for preeclampsia. Alternatively, the subject may already have some preeclampsia symptom(s), preferably the subject has some preeclampsia symptoms but has not been diagnosed as having preeclampsia. The subject may have one or more preeclampsia symptoms, preferably one preeclampsia symptom. The subject may not have been diagnosed as having preeclampsia but presents with one or more

preeclampsia symptoms. The preeclampsia symptoms may be selected from the list consisting of increased blood pressure or hypertension, increased urine protein or proteinuria, water retention (oedema), elevated liver enzyme levels, headache, visual disturbances, reduced urine output or oliguria, epigastric pain, eclamptic seizures, reduced platelet count and uteroplacental dysfunction. The subject may have an increased blood pressure or hypertension, and/or an increased urine protein or proteinuria.

The term“biological sample” encompasses a variety of sample types obtained from an organism and can be used in a diagnostic, prognostic or monitoring assay. The biological sample may be any sample derived from the subject. Samples may include, without limitation, whole blood, plasma, serum, red blood cells, white blood cells (e.g., peripheral blood mononuclear cells), saliva, urine, stool (i.e. , faeces), tears, sweat, sebum, nipple aspirate, ductal lavage, tumour exudates, synovial fluid, cerebrospinal fluid, lymph, fine needle aspirate, amniotic fluid, any other bodily fluid, nail clippings, cell lysates, cellular secretion products, inflammation fluid, vaginal secretions, or biopsies such as preferably placental biopsies. Preferred samples may include ones comprising any one or more biomarkers as taught herein in detectable quantities. The biological sample may be a blood sample or other liquid samples of biological origin. Biological sample may refer to samples that have been manipulated in any way after their procurement, such as by treatment with reagents, solubilisation, or enrichment for certain components. The term encompasses a clinical sample and also serum, plasma, biological fluids and tissue samples. Preferably, the sample is a blood sample, for example whole blood, serum or plasma. The sample may be a blood serum sample.

The sample, e.g. blood sample is typically obtained from the individual any time after confirmation of pregnancy, preferably during the first, second or third trimester of gestation. The sample may be collected between the 11^th and 14^th gestational week during the first trimester. The sample may be collected at the 180th gestational day i.e. 25 gestational weeks and 5 gestational days. According to the present invention, the sample, e.g. blood sample may be collected in the third trimester of pregnancy. The sample, for example the blood sample, may be collected after the 140^th, 150^th, 160^th, 170^th, 175^th, 176^th, 177^th or 178^th gestational day. In one embodiment of the present invention, the sample, e.g. blood sample is collected after the 179^th gestational day. Gestation means the duration of pregnancy in a human, i.e. the time interval of development from fertilisation until birth, plus two weeks, i.e. to the first day of the last menstrual period. Thus, gestational age is defined as the age of the pregnancy from the last menstrual period. The second or third trimester means the second or third portions of gestation, each segment being three months long. Thus, for example, the first trimester means from the first day of the last menstrual period through the 13th week of gestation; the second trimester means from the 14th through to the 27th week of gestation; the third trimester means from the 28th week through to birth, i.e. 38-42 weeks of gestation.

A sample may be obtained at about weeks 11 through 42 of gestation, at about weeks 14 through 42 of gestation, at about weeks 18 through 42 of gestation, at about weeks 20 through 42 of gestation, at about weeks 22 through 42 of gestation, at about weeks 24 through 42 of gestation, at about 28 weeks through 42 of gestation, at about weeks 30 through 42 of gestation, at about weeks 34 through 42 of gestation, at about weeks 38 through 42 of gestation. Thus, in some embodiments, the subject sample may be obtained early in gestation, e.g. at week 11 or more of gestation, e.g. at week 11 , 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 , 22, or 23 or more of gestation, more often at week 24, 25, 26, 27, 28, 29, 30, 31 , 32, 33, or week 34 or more of gestation. In other embodiments, the subject sample may be obtained late in gestation, for example, after 34 weeks of gestation, e.g. at week 35, 36, 37, 38, 39, 40, or week 41 of gestation. In one embodiment of the invention, the preeclampsia prognosis or diagnosis is made after the 179^th gestational day. In a further embodiment of the invention, the sample, for example the blood sample, is collected after the 179^th, 180^th, 181^st, 182^nd, 183^rd, 184^th, 185^th or 200^th gestational day. Preferably, the sample should be taken before the 259^th gestational day.

Once a sample is obtained, it can be used directly or frozen. Samples may include samples derived from humans.

Biomarkers

Biomarkers or preeclampsia biomarkers are molecular entities whose representation or determination in a sample, for example a blood sample, is associated with a preeclampsia phenotype. Biomarker concentrations follow tight gestational dynamics, i.e. the normal expected range of concentrations of a specific biomarker is dependent on the gestational age. For example, the concentration of a biomarker in a sample may be high in early pregnancy, but decrease or drop in the late pregnancy or vice versa, the concentration of biomarker in a sample may be low in early pregnancy, but increase in the late pregnancy.

Biomarkers may be differentially represented, i.e. represented at a different level, in a sample from an individual that will develop or has developed preeclampsia as compared to a healthy individual. In some instances, an elevated level of biomarker at the gestational age of sampling is associated with the preeclampsia phenotype. For example, the concentration of biomarker in a sample may be 1.5-fold, 2-fold, 2.5-fold, 3-fold, 4-fold, 5-fold, 7.5-fold, 10- fold, or greater more concentrated in a sample associated with the preeclampsia phenotype than in a sample not associated with the preeclampsia phenotype or the concentration of marker in a sample may be 10%, 20%, 30%, 40%, 50% or greater more concentrated in a sample associated with the preeclampsia phenotype than in a sample not associated with the preeclampsia phenotype. In other instances, a reduced level of the biomarker at the gestational age of sampling is associated with the preeclampsia phenotype. For example, the concentration of marker in a sample may be 1.5-fold, 2-fold, 2.5-fold, 3-fold, 4-fold, 5- fold, 7.5-fold, 10-fold, or greater less concentrated in a sample associated with the preeclampsia phenotype than in a sample not associated with the preeclampsia phenotype or the concentration of the biomarker in a sample may be 10%, 20%, 30%, 40%, 50% or more less concentrated in a sample associated with the preeclampsia phenotype than in a sample not associated with the preeclampsia phenotype.

Preeclampsia biomarkers used in the method of the present invention include sFItl , PIGF, ADAM 12, sENG and leptin. The level of at least three, three, at least four, four, at least five or five biomarkers are measured in the sample, preferably a blood sample, from the patient. The level of three or more, four or more, or five or more biomarkers are measured in the sample, preferably a blood sample, from the patient. The levels of the biomarkers in the sample find use in providing a preeclampsia assessment, for example making a

preeclampsia exclusion, diagnosis, prognosis, monitoring and/or treatment.

In one embodiment of the invention, the method comprises measuring the level of at least three, at least four biomarkers or at least five biomarkers. The method may comprise measuring the level of at least three biomarkers, wherein the at least three biomarkers are selected from sFItl , PIGF and sENG and optionally one or more additional biomarkers selected from ADAM 12 and leptin. The method may comprise measuring the level of at least four biomarkers, wherein the at least four biomarkers are selected from sFItl , PIGF,

ADAM 12 and sENG and optionally one or more additional biomarkers selected from leptin.

In a preferred embodiment of the invention, the method comprises measuring the level of four biomarkers wherein the four biomarkers are sFItl , PIGF, ADAM 12 and sENG. The method may comprise measuring the level of at least five biomarkers, wherein the at least five biomarkers are selected from sFItl , PIGF, ADAM12, sENG and leptin. In a preferred embodiment of the invention, the method comprises measuring the level of five biomarkers wherein the five biomarkers are sFItl , PIGF, ADAM 12, sENG and leptin. Method

The level(s) of preeclampsia arker(s) in the biological sample from an individual are measured or evaluated. The terms“evaluating”,“assaying”,“measuring”,“assessing,” and “determining” are used interchangeably to refer to any form of measurement, including determining if an element is present or not, and including both quantitative and qualitative determinations. Evaluating may be relative or absolute. The level of one or more

preeclampsia markers in the subject sample may be measured or evaluated by any convenient method. For example, such methods may include biochemical assay methods, immunoassay methods, mass spectrometry analysis methods, or chromatography methods, or combinations thereof. The term "immunoassay" generally refers to methods for detecting one or more molecules or analytes of interest in a sample, wherein specificity of an immunoassay for the molecule(s) or analyte(s) of interest is conferred by specific binding between a specific-binding agent, commonly an antibody, and the molecule(s) or analyte(s) of interest. For example, preeclampsia biomarker levels may be detected using any immunoassay-based technology, multiplex platform or other microsphere-based platform, ELISA, RIA, EIA, EMIT, FIA, FPFIA, TRFIA, CLIA, LIA or LIPs. Preeclampsia biomarker levels may be detected using mass spectrometry, proteomic arrays, flow cytometry, western blotting or immunohistochemistry. Preferably the immunoassay-based technology used in the present invention is a multiplex platform e.g. a microsphere-based platform. More preferably, the immunoassay-based technology used in the present invention is a multiplex platform that is a microsphere-based platform e.g. xMAP (sometimes referred to as xMAP technology e.g. as supplied by Luminex). According to the present invention the

microsphere-based platform used in the present invention is xMAP technology.

Clinical cofactors

In some embodiments of the invention, the methods of preeclampsia assessment, e.g.

diagnosing preeclampsia, prognosing preeclampsia, monitoring preeclampsia, may comprise additional assessment(s) that are employed in conjunction with the biomarker measurement. In particular, the subject methods may further comprise measuring one or more clinical cofactors associated with preeclampsia. One such clinical cofactor is gestational age at blood sampling or blood draw. Another is the weight of the subject at blood sampling or blood draw. In an embodiment of the invention, the method comprises measuring at least one or two clinical cofactors. In an embodiment of the invention, the method comprises measuring at least one clinical cofactor that is gestational age. In a preferred embodiment of the invention, the method comprises measuring at least one clinical cofactor selected from gestational age and optionally the weight of the subject. In another preferred embodiment of the invention, the method comprises measuring two clinical cofactors selected from gestational age and weight of the subject.

A subject may be assessed for one or more clinical cofactors at about week 11 or more of gestation, e.g. week 12,13, 14, 15, 16, 17, 18, 19, 20, 21 , 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 , 32, 33, 34, 35, 36, 37 or more of gestation, wherein the clinical cofactors are used in combination with the marker level representation to provide a preeclampsia diagnosis, a preeclampsia prognosis, to monitor the preeclampsia. In some instances, the clinical cofactors may be measured prior to obtaining the preeclampsia marker level representation. In some instances, the clinical cofactors may be measured after obtaining the preeclampsia marker level representation. In some embodiments of the invention, the clinical cofactors are measured at the same time as the sample draw from the subject. In some other embodiments of the invention, the clinical cofactors are measured +/- 1 , 2, 3 or 4 days from the sampling of the subject. In an embodiment of the invention, the method comprises measuring at least one clinical cofactor at the blood draw or blood sampling selected from gestational age and the weight of the subject.

Prediction value

The measurements of the preeclampsia biomarkers and optionally the clinical cofactors may be analysed collectively to arrive at a single preeclampsia prediction value. Prediction value means a single metric value that represents the weighted levels of each of the preeclampsia biomarkers and optionally clinical cofactors measured. As such, the method comprises detecting the level of biomarkers in the sample and optionally determining at least one clinical cofactor and calculating a prediction value based on the levels of the preeclampsia biomarkers and optionally the at least one clinical cofactor. The prediction value is compared to a cut off or threshold value and a prediction of whether the subject will develop or has preeclampsia will be made based on that comparison. The prediction value represents a prediction as to whether the subject will develop or not preeclampsia within a given time period, e.g. 1 week, 2 weeks, 3 weeks, 4 weeks, 5 weeks, 6 weeks, 7 weeks, 8 weeks, 1 month, two months, three month, four months, five months or six months. In an embodiment of the invention the prediction value represents a prediction as to whether the subject will develop or not develop preeclampsia within a two-month period. The prediction value calculated at the 11^th, 12^th, 13^th or 14^th gestational week may predict whether the subject will develop or will not develop preeclampsia with an onset on or after 20^th, 21^st, 22^nd, 23^rd, 24^th, 25^th, 26^th, 27^th 28^th, 29^th, 30^th, 31^st, 32^nd, 33^rd, 34^th, 35^th or 36^th gestational week.

In an embodiment of the invention, the method comprises generating a prediction value based on the levels of the at least four biomarkers in the blood sample from the subject and optionally based on the at least one clinical cofactor and generating a prediction value that is compared to a threshold value. In the case where the prediction value, calculated using the serum biomarker and clinical data of the patient at the time of blood sampling, is equal to, above or larger than the threshold value, for example, the prediction value is above the threshold value, the subject is predicted to have a risk of developing PE, i.e. PE is ruled-in.

In the case where the prediction value, calculated using the serum biomarker and clinical data of the patient at the time of blood sampling, is lower than the threshold value, the risk to PE can is predicted to be zero, i.e. ruled out. The threshold value may be between 0.1 and 0.5. In some embodiments of the invention, the threshold value is 0.175 ± 0.005. In other embodiments of the invention the threshold value is 0.204 ± 0.005. In yet further

embodiments of the invention the threshold value is 0.443 ± 0.005. The method of the present invention predicts whether the subject has, will or will not develop preeclampsia within a two-month period. In a further embodiment of the invention, a high sFItl , low PIGF, low ADAM12 and high sENG level as compared with control indicates an increased susceptibility of the subject developing preeclampsia or increased probability that the subject has preeclampsia and if included in the calculation of prediction value, a high leptin level as compared with control indicates an increased probability of the subject developing preeclampsia or increased probability that the subject has preeclampsia, and if included in the calculation of prediction value, a low gestational age and a high weight of the subject indicates an increased probability of the subject to develop preeclampsia or increased probability that the subject has preeclampsia. In an embodiment of the present invention, the prediction value is based on gestational age and the weight of the subject as clinical cofactors.

The prediction value is typically compared to a threshold value. Thus, the prediction of whether the subject will develop PE is by reference to a threshold value. When the prediction value is compared to a threshold value, it indicates whether the subject will develop or not PE within a time period. The indication that the subject will develop PE typically means that the subject will develop PE within 2, 4, 6 or 8 weeks, 1 , 2, 3, 4, 5, 6, 7 or 8 months. The person will typically develop PE within 1 month e.g. within 1 , 2, 3 or 4 weeks, in 1 month, within 2 months or in 2 months or longer. If the comparison of the prediction value to the threshold indicates that the subject will not develop PE, this means that the subject will not develop PE within 2 months. This is informative to the patient management as no specific follow up needed within 2 months.

A prediction value ( (i )) for a patient sample may be calculated by an algorithm for using biomarker concentrations. The algorithm may be according to the following formula:

wherein the coefficients a, b, c, d, e, f, g, h, and i are integers or numbers a is different from 0. At least 3 of the coefficients selected from b, c, d, e, f, and i may be different from 0 and optionally at least one coefficient selected from g, and h may be different from 0. Coefficient i may be 0 or different from 0.

The algorithm may be according to the following formula:

wherein the coefficients a, b, c, d, e, f, g, and h, are integers or numbers a is different from 0. At least 3 of the coefficients selected from b, c, d, e, and f may be different from 0 and optionally at least one coefficient selected from g, and h may be different from 0. When the coefficients are different from 0, the coefficient values may be as follows: a may be between -175 and -7; and/or b may be between -6 and -0.5; and/or c may be c between 2 and 36; and/or d may be between -20 and -0.5; and/or e may be between -12 and -0.3; and/or f may be between -3 and -0.1 ; and/or g may be between -0.3 and -0.05; and/or h may be between -0.2 and -0.01 ;

The threshold value may be between 0.1 and 0.5, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula: wherein the coefficients a, b, d, e, f, g, h, and i are integers or number a is different from 0. At least 2 of the coefficients selected from b, d, e, f, and i are different from 0 and optionally at least one coefficient selected from g and h is different from 0.

The algorithm may be according to the following formula:

wherein the coefficients a, b, c, d, e, f, are integers or numbers a is different from 0. At least 3 of the coefficients selected from b, c, d, e, and f are different from 0. When the coefficient values are different from 0, the coefficients values are as following: a is comprised between -57 and -15; and/or b is comprised between -1 ,6 and -0.6; and/or c is comprised between 2 and 9.5; and/or d is comprised between -5 and -0,5; and/or e is comprised between -4.5 and -0.9; and/or f is comprised between -0.25 and -0.15;

The threshold value may be between 0.12 and 0.42, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, e, and f are negative integers or numbers and c is a positive integer or number. The coefficients values may be as follows: a = -52 ± 10% b = -1 ± 10% c = 8 ± 10% d = -4 ± 10% e = -4 ± 10% f = -0.2 ± 10% The threshold value may be 0.339 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, e, and f are negative integers or number and c is a positive integer or number. The coefficients values may be as follows: a = -51 ± 10% b = -1 ± 10% c = 8 ± 10% d = -4 ± 10% e = -4 ± 10%

The threshold value may be 0.319 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, e, and f are integers or numbers a is different from 0. At least 2 of the coefficients selected from b, d, e, and f may be different from 0.

The algorithm may be according to the following formula:

wherein the coefficients a, b, c, d, e, f, and g are integers or numbers a is different from 0, g may be different from 0. At least 3 of the coefficients selected from b, c, d, e, and f may be different from 0. When the coefficient values are different from 0, the coefficients values may be as follow: a is comprised between -125 and -17; and/or b is comprised between -3.9 and -0.8; and/or c is comprised between 3.2 and 22; and/or d is comprised between -11.4 and -2,2; and/or e is comprised between -5.5 and -0.3; and/or f is comprised between -2.8 and -2.2; and/or g is comprised between -0.8 and -0.06;

The threshold value may be comprised between 0.26 and 0.492, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, e, f and g are negative numbers and c is a positive number. The coefficients values may be as follows: a = -114±10% b = -4±10% c = 20±10% d = -10±10% e = -5±10% f = -3±10% g = -0.2±10%

The threshold value may be 0.443 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d and e are negative integers or numbers and c is a positive integer or number. The coefficients values may be as follows: a = -64 ± 10% b = -2 ± 10% c = 11 ± 10% d = -5 ± 10% e = -3.9 ± 10% g = -0.08 ± 10%

The threshold value may be 0.447 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, e, f and g are integers or numbers a is different from 0, g may be different from 0. At least 2 of the coefficients selected from b, d, e, and f may be different from 0.

The algorithm may be according to the following formula:

wherein the coefficients a, b, c, d, e, f, g, and h are integers or numbers a is different from 0, g and h may be different from 0. At least 3 of the coefficients selected from b, c, d, e, and f may be different from 0.

When the coefficient values are different from 0, the coefficients values may be as follows: a may be between -175 and -13.9; and/or b may be between -5.3 and -0.9; and/or c may be between 3.2 and 36; and/or d may be between -19.8 and -3.5; and/or e may be between -11.4 and -0.36; and/or f may be between -2.4 and -1.7; and/or g may be between -0.22 and -0.06; and/or h may be between -0.22 and -0.018; The threshold value may be between 0.157 and 0.372, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, e, f, g and h are negative integers or numbers and c is a positive integer or number. The coefficients values may be: a = -158 ± 10% b = -5 ± 10% c = 32 ± 10% d = -18 ± 10% e = -10 ± 10% f = -2 ± 10% g = -0.2 ± 10% h =-0.1 ± 10%

The threshold value may be 0.175 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, e, g and h are negative integers or numbers and c is a positive integer or number. The coefficients values may be: a = -137 ± 10% b = -3 ± 10% c = 26 ± 10% d = -15 ± 10% e = -10 ± 10% g= -0.1 ± 10% h= -0.2 ± 10%

The threshold value may be 0.204 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, g, and h are negative integers or numbers and c is a positive integer or number. The coefficients values may be: a = -53 ± 10% b = -1 ± 10% c = 8 ± 10% d = -4 ± 10% g = -0.1 ± 10% h = -0.07 ± 10%

The threshold value may be 0.341 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

The algorithm may be according to the following formula:

wherein the coefficients a, b, d, e, f, g and h are integers or numbers a is different from 0, g and h may be different from 0. At least 2 of the coefficients selected from b, d, e, and f are different from 0.

The algorithm may be according to the following formula: wherein the coefficients a, b, c, d, e, f, g and h are numbers. Suitably, a, b, d, e, f, g and h are negative numbers and c is a positive number a is different from 0. At least 4 of the coefficients selected from b, c, d and e are different from 0 and optionally at least one coefficient selected from f, g, and h is zero or different from 0.

In an embodiment of the invention, the algorithm is according to formula (I):

wherein a, b, d, e, f, g and h are negative numbers and c is a positive number; or one or more of f, g and h are 0.

In an embodiment of the invention, the algorithm is according to formula (I), wherein the coefficients are as follows: a is -158.3 ±10%; b is -4.8 ±10%; c is 31.9 ±10%; d is -17.8 ±10%; e is -10.3 ±10%; f is -2.1 ±10%; g is -0.2 ±10%; and h is -0.1 ±10%.

Preferably, the threshold value is 0.175 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

In another preferred embodiment, the algorithm is according to formula (II):

wherein a, b, d, e, g and h are negative numbers and c is a positive number; or one or more of g and h are 0.

In an embodiment of the invention, the algorithm is according to formula (II), wherein the coefficients are as follows: a = -136.9 ± 10% b = -3.4 ± 10% c = 26.5 ± 10% d = -14.7 ± 10% e = -9.7 ± 10% g= -0.1 ± 10% h= -0.2 ± 10%

Preferably, the threshold value is 0.204 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia.

In another embodiment of the invention, the algorithm is according to formula (III):

wherein the coefficients a, b, c, d, e, f and g are numbers. Suitably, a, b, d, e, f and g are negative numbers and c is a positive number a is different from 0; or one or more of f and g are 0. g may be different from 0. At least 4 of the coefficients selected from b, c, d, e and f are different from 0.

In a specific embodiment of the invention, the coefficients are as follows: a is -114 ±10%; b is -3.7 ±10%; c is 19.9±10%; d is 10.3 ±10%; e is -5 ±10%; f is -2.5 ±10%; and g is -0.2 ±10%.

Preferably, the threshold value is 0.443 ± 0.005, and a p(i) value equal to or above the threshold value indicates the subject will develop preeclampsia or a p(i) value below the threshold value indicates the subject will not develop preeclampsia. In the calculation of p(i):

[PIGF] is the concentration of PIGF in the sample in [pg/mL]

[sFItl] is the concentration of sFItl in the sample in [pg/mL]

[sENG] is the concentration of sENG in the sample in [ng/mL]

[ADAM12] is the concentration of ADAM12 in the sample in [ng/mL]

[leptin] is the concentration of leptin in the sample in [ng/mL] gest age is the gestational age in days at sampling weight is the maternal weight at sampling in [kg]

[leptin]/weight is the the concentration of leptin in the sample in [ng/mL] adjusted to maternal weight

The caret refers to the exponentiation operator.

In further embodiments, alternative algorithms combining in an additive manner two or more serum biomarkers comprising [PIGF], [sFItl], [sENG], [ADAM12] or [leptin] with or without clinical cofactors can be applied.

Examples

Materials and methods Luminex xMAP® technology xMAP® suspension array technology is based on polystyrene beads with a diameter of 5.6 mm that are internally dyed with various ratios of two spectrally distinct fluorophores (Luminex® Corporation, Austin, TX, USA). As a result, an array of up to 500 different bead sets with specific absorption spectra is created. Various biological molecules, such as individual oligonucleotide probes, proteins or antibodies, can be coupled to alternative sets of beads. These sets are combined to a suspension array and due to their unique absorption spectra, it is possible to measure simultaneously up to 500 different probes in a single multiplex reaction. The technology is capable of performing both protein- and nucleic acid-based analyses, enabling both quantitative protein assays and qualitative DNA-based detection assays.

Luminex® technology: https://www.luminexcorp.com/research/our-technology/xmap- technology/ Reagents for the Luminex® sandwich immunoassays

Development of the Luminex® xMAP based multiplexed assay was targeted to nine previously established maternal serum predictive biomarkers for preeclampsia (Figure 1). Luminex® magnetic microspheres (#MC100) and antibody coupling kit for covalent linking of antibodies and microspheres (Antibody Coupling Kit, #40-50016) were purchased directly from Luminex® Corporation (Austin TX, USA). Capture and detection antibodies, and reference proteins were purchased from R&D Systems (Minneapolis, MN, USA) with the exception of STC1 reference protein (Abnova, Taipei City, Taiwan) (Table 1). Antibodies were covalently linked to the surface of the fluorescently labeled Luminex® microspheres according to the manufacturer s protocol and kept in dark at +4 °C until use. Stock solutions of each reference protein were prepared according to respective manufacturer s instruction. For each biomarker, expected concentration ranges to be measured in the sera of pregnant women were derived from the literature (Table 2). Stock solutions were further diluted in 1 % BSA (#810685; Merck KGaA, Darmstadt, Germany) to prepare the 10X Standard-8 solutions of either individual or multiplexed reference proteins, corresponding to 10-fold higher concentration of the maximum expected value of each analyte (Table 3). Prior to each immunoassay experiment, fresh working solutions the Standard-8 and its two-fold serial dilutions (Standard-7 and subsequent dilutions representing decreasing concentrations) were prepared in General Assay Diluent (GAD; #620; ImmunoChemistry Technologies LLC, Minnesota, USA). GAD represents a mammalian protein-based immunoassay additive and it served as a blank measurement (Standard-1) for each analyzed biomarker in all experiments. All antibodies, stocks and 10X Standard-8 solutions of reference proteins were kept at -80 °C.

For all biomarkers, the optimization series tested 20 alternative combinations of 0.24-30 mg capture antibody (bound to 1.25x106 magnetic beads) and 1-4 mg/mL monoclonal or polyclonal detection antibody diluted in Washing Buffer (WB; 1% BSA in PBS). For the capture and detection antibodies, the range of titrated and tested concentrations followed the recommendations by Luminex® with the aim to identify the most balanced combination with regards to the analytical accuracy and the cost-effectiveness of the assay. Optimal combinations of concentrations were titrated using dilution series of commercially purchased reference proteins for nine biomarkers (Table 3). The best performing combination are presented in Table 4 and in Figure 2.

Serum samples for the immunoassay development

Luminex® singleplex and multiplex immunoassay development and quality assessment (Stage ll-lll, Figure 1) was performed using commercially available standardized pool of human serum samples, referred as Serum Matrix (S1-100ML, Human Serum; Merck KGaA, Darmstadt, Germany). Reference proteins were spiked into the Serum Matrix either in singleplex or multiplex format in three analyte concentrations. The High concentration of each biomarker corresponded to the respective Standard-8 (Table 3), whereas Medium and Low concentrations represented 1 :10 and 1 :100 dilutions of High values, respectively.

Additionally, eight individual human serum samples, available in the in-house ‘Happy Pregnancy’ biobank (see below) and referred throughout the text as the Testing Serum Samples, were included into the experiments of multiplex assay development and quality assessment. The testing set of blood samples had been drawn from healthy pregnant women at 12, 16, 25, 26, 33, and 36 gestational weeks and from two non-pregnant reproductive age women. All blood samples were collected into Becton Dickinson Vacutainer® SST™ Serum Separation Tubes containing spray-coated silica and a polymer gel (Becton Dickinson Company, Franklin Lake, NJ, USA). Serum was separated in the service laboratory (United Laboratories, Tartu University Hospital) using routine procedures according to manufacturer’s instructions (centrifugation at 1 ,800 g for 10 min at room temperature, RT). Serum samples were kept at -80°C before further aliquoting and subsequent analysis.

In immunoassay experiments, the Serum Matrix, the Serum Matrix with the spiked proteins and the Testing Serum Samples were diluted in GAD.

Luminex® sandwich immunoassay protocol

All incubations during the immunoassay protocol were carried out at RT using a microplate shaker with rotational movements (speed set to 500 rpm; #4625, Barnstead Lab-Line, MA, USA). At the end of each incubation step, plates were placed on the Luminex® Magnetic Plate Separator (#CN-0269-01) for 1 min to bring all magnetic microspheres to the bottom of each well, followed by supernatant aspiration. All washing steps between incubations used 100 mL of WB per cell.

Prior immunoassay, 96-well round bottom microplates (#734-1642 Corning, NY, USA) were treated for 10 min with Blocking Buffer (BB; 100 mL per well; 1 % BSA, 0.02% Tween-20 in PBS at pH 7.4). After removal of BB, capture antibody coupled microbead solution prepared in WB (50 mL per well containing 2,500 beads of each analyte) and tested samples ( Standards8- 1' or sera diluted in GAD; 50 mL/well) were pipetted to microplates. The mixture was incubated together for 2 h, followed by a washing step. Next, a biotinylated monoclonal or polyclonal detection antibody (or mixture of antibodies for multiplex assays; 100 mL/ well) diluted in WB was incubated for 1 h with a subsequent washing step. For the labeling of the immunoassay with a fluorescent reporter, incubation with streptavidin-phycoerythrin conjugate (100 mL of 1 mg/mL solution in WB; #PZPJ39S, Europa Bioproducts Ltd, Cambridge, UK) was applied for 30 min, followed by two rounds of washings. Finally, microbeads were resuspended in 75 mI_ of WB, a minimum of 50 biomarker-specific beads were collected from each well and analysed on Luminex® MAGPIX analyzer (Luminex® Corporation, Austin TX, USA) using weighted 5-parameter logistic model implemented in Luminex® xPONENT 4.1 software (Luminex® Corporation, Austin TX, USA).

Happy Pregnancy cohort of pregnant women and cases included into the current study

The‘Happy Pregnancy’ study (full name:‘Development of novel non-invasive biomarkers for fertility and healthy pregnancy’, 2013-2015) recruited 2,334 unselected pregnant women at their first antenatal visit at the Women Clinic of Tartu University Hospital, Estonia. The study was approved by the Ethics Review Committee of Human Research of the University of Tartu, Estonia (permissions no 221/T-6, 17.12.2012; 286/M-18, 31.01.2018). Informed consent was obtained from every subject at first antenatal visit. For every participant, longitudinal anthropometric, epidemiological (three questionnaires across gestation), clinical data and biological material throughout the pregnancy and at the delivery were collected. At every clinical visit, all study participants had been monitored for their weight gain, arterial blood pressure dynamics, symptoms of proteinuria and signs of PE. The data about the further course and pregnancy outcome including fetal parameters were obtained from medical documentation.

In the current study, 61 serum samples (Table 5) from 53 pregnant women (age 18-40 years) with spontaneously conceived single pregnancy were analysed (Table 6). Individuals with multiple pregnancy, pre-existing renal disease or anti-phospholipid syndrome were excluded. During their first antenatal visit, all women were normotensive, including two subjects with medical history of anti hypertensive treatment.

The women selected for the current study included 16 pregnancies with neither symptoms nor development PE, 15 cases with isolated symptoms alerting potentially to PE development but continuing until delivery without additional complications and 22 patients who eventually developed PE. Clinical signs at routine antenatal visits alerting to PE were considered single measurements of increased blood pressure (systolic blood pressure, SBP ³140 mmHg, diastolic blood pressure, DBP ³90 mmHg), signs of proteinuria (amount of protein in urine ³0.3 g/l) or development of gestational hypertension after 20th week of pregnancy. Hypertension was confirmed when the abnormal measurements were obtained in sitting position in two occasions at least 4 hours apart while the patient has been resting for at least 15 minutes.

Diagnosis of PE was based on the criteria recommended by American College of Obstetricians and Gynecologists (ACOG, 2013). In the included cases, PE was assigned at the onset of both hypertension (SBP ³140 mmHg; DBP ³90 mmHg) and proteinuria (2+ protein or greater on dipstick urinalysis, ³300 mg of protein per 24-h urine collection). Alternative to increased urinary protein, other relevant clinical symptoms were considered, such as headache resistant to analgesics or visual disturbances, epigastric pain, severe edema, and oliguria after 20 gestational weeks.

Study design and workflow

The study design is summarized in Figure 1. Briefly, singleplex Luminex^® xMAP tests were developed for preeclampsia biomarkers reported in the literature. The quality assessment included measurements of the reference proteins in a commercially available serum equivalent, as well as the biomarker levels in native human blood samples. A multiplex Luminex^® immunoassay was developed for the analytes that passed all quality criteria in the singleplex test and showed highly correlated performance in the singleplex versus multiplex formats. The established Luminex^® 6PLEX assay was further tested for the technical performance (e.g. inter- and intra-assay variability, the effect of freeze-thaw cycles) and subsequently utilized to measure biomarker levels in 61 serum samples drawn from pregnant women with variable gestational scenarios (Table 5). The data for the measured sFItl and PIGF levels were compared to the currently available commercial test

(B R A H M S). Finally, the improved PE prediction models were developed using alternative combinations of the five informative biomarkers measured with the developed Luminex^® 6PLEX assay. The best performing prediction model was retrospectively applied to the analyzed clinical cases to estimate the rate of false-positive and false-negative PE predictions, and to assess case-by-case the clinical scenarios for biased test performance.

Optimization of the Luminex^® assay conditions

Development of the Luminex^® xMAP based multiplexed assay was targeted to nine previously established maternal serum biomarkers for preeclampsia (Figure 1).

Optimal combinations of antibody concentrations were titrated using dilution series of commercially purchased reference proteins for nine biomarkers (Table 1). For all

biomarkers, the optimization series tested 20 alternative combinations of 0.24-30 mg capture antibody (bound to 1.25x10⁶ magnetic beads) and 1-4 mg/mL monoclonal or polyclonal detection antibody. The range of titrated and tested concentrations followed the

recommendations by Luminex^® and available scientific literature (Table 2; Table 3). The aim of the test series was to identify the most balanced combination with regards to the analytical accuracy and the cost-effectiveness of the assay. The best performing antibody combination are presented in Table 4.

Luminex^® singleplex and multiplex immunoassay development and quality assessment (Stage ll-lll, Figure 1) was performed using commercially available standardized pool of human serum samples, referred as the Serum Matrix throughout the paper (S1-100ML, Human Serum; Merck KGaA, Darmstadt, Germany). Additionally, eight native human serum samples, available in the in-house Happy Pregnancy study biobank (see below) and referred throughout the text the Testing Serum Samples, were included into the experiments of multiplex assay development and quality assessment. The testing set of blood samples had been drawn from healthy pregnant women at 12, 16, 25, 26, 33, and 36 gestational weeks and from two non-pregnant reproductive age women. In immunoassay experiments, the spiked proteins in the Serum Matrix and the Testing Serum Samples were further diluted in General Assay Diluent (GAD; #620; ImmunoChemistry Technologies LLC, Minnesota, USA). The applied Luminex^® sandwich immunoassay protocol followed the guidelines of the manufacturer.

Analytical accuracy, Serum Matrix Coefficient (SMCf) and limit of detection (LoD)

Analytical accuracy expresses the closeness of expected and calculated concentration values of standard proteins in GAD, reported in percentage (accuracy %). It was estimated for each analyzed biomarker in the developed singleplex and multiplex assays based on the recovery of expected concentrations of reference proteins implementing in the analysis of decreasing concentration gradient from the Standards-8 to Standards-1.

Serum Matrix Coefficient (SMCf) represents the ratio of MFI measurements of the analytes spiked into the sample matrix relative to respective proteins diluted in GAD. In the current study, SMCf was estimated using reference proteins spiked into the Serum Matrix ( High , Medium and Low concentrations relative to respective measured values in GAD (Table

7). For SMCf calculation in singleplex experiments, reference proteins spiked the Serum Matrix were further diluted 1 :10 in GAD. For SMCf estimation in multiplex experiment, the Serum Matrix samples spiked with all analyzed proteins were measured for both 1 :10 and 1 :20 dilutions in GAD. In each conducted experiment, MFI measurements were corrected for the background MFI signals detected from either the unspiked Serum Matrix and/or pure GAD solutions.

The limit of detection (LoD) of each biomarker in the final multiplex assay was estimated based on measured MFI background values detected in the Standardl (pure GAD) (Table

8). LoD was calculated as mean + 2SD from four independent MFI measurements. In all subsequent experiments the measured concentration values below LoD were replaced with respective biomarker LoD.

Clinical study material

To assess the applicability of the developed multiplex immunoassay in clinical setting we analyse serum samples drawn at the 2^nd half of gestation (later than 179 gestational days) from pregnant women during a prospective observational cohort study Happy Pregnancy . Upon informed consent, the study had recruited unselected pregnant women at their first antenatal visit at the Women Clinic of Tartu University Hospital, Estonia. All pregnancies were monitored until delivery based on the recommendations of the national guidelines for antenatal care. At every clinical visit, weight gain, arterial blood pressure dynamics, symptoms of proteinuria and signs of PE were documented. PE was diagnosed based on International Society for the Study of Hypertension in Pregnancy (ISSHP) recommendations. Serum samples for research purposes were collected from the study participants in parallel with blood sampling for routine clinical tests and transferred to and stored at -80°C.

The current study included 53 Happy Pregnancy study cases with a spontaneously conceived single pregnancy and representing alternative pregnancy curricula with the respect to the development of PE (Table 6).

At their first antenatal visit all women were normotensive, including two subjects with past medical history of anti hypertensive treatment. During the index pregnancy 22 women had eventually developed PE (age 26.8 ± 5.2 years; pre-pregnancy BMI 26.8 ± 6.2; nulliparity 77%; male newborn 54.5%) and 31 gestations proceeded until delivery without PE (age 28.0 ± 5.2 years; pre-pregnancy BMI 26.8 ± 6.2; nulliparity 51.6%; male newborn 48.4%). The current analysis included in total 61 serum samples drawn between 180-275 gestational days (Table 5). For 45 pregnant women, one serum sample and for 8 pregnant women two consecutive 3^rd trimester serum samples were available for the measurement of circulating PE biomarkers. In six cases, the first serum sample had been drawn before any signs of PE and the second sample either at the onset symptoms (n=3) or diagnosis of PE (n=3). In two cases samples had been drawn at isolated PE symptoms, either further progressing with the disease (n=1) or not (n=1). Further details of the Happy Pregnancy study cohort, clinical characteristics of the analyzed study group and diagnostic criteria for PE are provided in Table 6.

The analyzed clinical serum samples were divided into 6 subgroups. Group I comprised of samples drawn from healthy, uncomplicated pregnancies until delivery (n=23). The blood samples in Groups II - III represent cases without any clinical symptoms of PE at the blood draw but developing the disease 28-62 days (n=6; Group II) or 10-27 days (n=10; Group III) after serum sampling. Group IV included samples from women with isolated hypertension or proteinuria at the blood draw but proceeding with uncomplicated pregnancy until delivery (n=9). Group V represents pregnancies with symptoms alerting to PE at the serum sampling and further proceeding to the disease within 4-43 days (n=9). Serum samples taken from the patients with the diagnosis of PE (group VI, n=4) were used as a comparative reference for biomarker levels at the PE onset.

Biomarker measurements in clinical serum samples with Luminex^® 6PLEX immunoassay

All collected serum samples had been stored in -80° C with no thawing for maximum 1.5 years. Prior to the immunoassay applications, all 61 clinical serum samples selected for the current study were thawed on ice and distributed to smaller (200 ml) aliquots, to be immediately returned to -80° C. All analyzed samples were aliquoted within the same week. One aliquot of each clinical serum sample was utilized in the analysis of biomarker concentrations with the developed Luminex^® 6PLEX immunoassay combining the measurements of sFItl , PIGF, sENG, leptin, ADAM12 and Pentraxin 3 (PTX3). All 61 clinical serum samples (1 :20 dilutions in GAD) were measured in duplicate during the same experiment and the mean of the two parallel MFI values was utilized in subsequent calculations. The concentrations of the analytes were calculated based on the dilution series of the reference proteins kept at -80°C. A mixture of 10X concentrated reference proteins contained sFItl 100 ng/mL, PIGF 5 ng/mL, sENG 100 ng/ml, ADAM12 2000 ng/mL, leptin 100 ng/mL and PTX3 100 ng/mL (Table 3).

Based on the duplicate measurements, the intra-assay variation was calculated for sFItl 5.28%, sENG 5.39%, leptin 5.83%, ADAM12 9.99% and PIGF 11.11 %. Due to the measurements of PTX3 at the bottom distribution of its detection range the respective estimated intra-assay variability (22.26%) was considered unreliable (Table 9). Inter-assay variation for each biomarker was calculated for the artificial control sample (biomarker reference proteins spiked into the Serum Matrix) analyzed in the same experiment on two parallel plates.

We assessed the freeze-thaw stability of the 6PLEX assay measurements in the analysis of two parallel aliquots of three clinical serum samples selected randomly from our biobank (storage in -80°C). Parallel aliquots were subjected to either one (default) or two (additional) rounds of freeze-thaw cycles, followed by simultaneous duplicate measurements with the 6PLEX assay. The thawing procedure was conducted on ice. The biomarker measurement data for the parallel aliquots were compared (Table 10). sFItl and PIGF measurements in clinical serum samples with a commercially available immunoassay

Parallel aliquots of the same 61 clinical serum samples were immediately shipped on dry ice to the commercial service provider Synlab Germany (Leinfelden, Germany) that offers the B R A H M S sFItl Kryptor/ B R A H M S PIGF plus Kryptor PE ratio test developed by Thermo Fisher Scientific. In this commercial test, serum sFItl and PIGF were measured in two separate immunofluorescent automated sandwich assays implemented on KRYPTOR compact PLUS platform (#BM0106172) and using B R A H M S reagents for sFItl

(#845.075) and PIGF plus (#859.075) assays. As instructed by the manufacturer, the results from the assays were combined to estimate the sFIt/PIGF ratio utilized to assess the risk to develop PE. For the sFIt/PIGF ratio, the current rule-out and the rule-in thresholds for the short-term (4 weeks) prediction of PE are considered <38 and >85 for gestational age 20 weeks to 33 weeks and 6 days, or >110 (or gestational age 34 weeks to delivery, respectively (Verlohren et al 2014).

Both, the Luminex^® 6PLEX performed by the first author of this study (K. Ratnik), and the sFItl , PIGF measurements by the commercial service provider were conducted for all samples simultaneously within a month after aliquoting. The experiments were performed blindly without prior knowledge of the pregnancy curriculum of the individual cases or the biomarker measurement data derived from the alternative test.

Statistical analysis

STATA/SE 13.0 (StataCorp LLC; Texas, USA) or the R3.3.3 language and environment (Free Software Foundation, Boston, MA, USA, http://www.r-project.org) was used for statistical analysis. Spearman correlation was used to evaluate comparison of measured biomarkers (both in singleplex vs multiplex evaluation and in Luminex^® 6PLEX vs

B R A H M S). Comparisons between Groups l-VI were performed using Mann-Whitney rank sum test. Chi-square test was used for comparing PE prediction proportions. P-value <0.05 was considered statistically significant. Logistic regression models (glm) implemented on R were applied to investigate independent associations between biomarkers and clinical onset of PE. Due to differences in scale and standard deviations, all biomarker values were centered and scaled for the models. The predictive power of the best models was assessed using the area under the curve (AUC). AUC values > 0.7 were considered to indicate adequate predictive capability. We used function‘predict’ in package‘stats’ to obtain individual predictions from a fitted glm model objects (for individual cases).

Example 1

Biomarker selection and singleplex assay development

In total nine maternal serum biomarkers were initially selected for the assay development (Fig. 1): PIGF, sENG, sFItl , leptin, ADAM12, adiponectin, STC1 , TRAIL, PTX3. A singleplex Luminex® immunoassay was developed using commercial standard proteins and respective antibodies for each biomarker (Table 1). For the standard proteins, the range of clinically relevant biomarker concentrations to be measurable with the assay was derived from the literature data (Table 2). A series of concentrations of the capture and detection antibodies were titrated according to Luminex® recommendations (Table 3) and the most optimal concentrations were selected (Table 4). Applying these selected conditions, satisfactory fluorescence signal response was detected for ADAM 12, adiponectin, sENG, leptin,

Pentraxin3, PIGF and sFItl (values rising >1000 MFI; Figure 2). However, STC1 and TRAIL measurements exhibited low MFI values (<200 MFI) across the entire range of biomarker concentrations and were excluded from the assay development. For all biomarkers, the analytical accuracy in measuring the concentrations of the standard proteins was 90-100%.

ADAM12, adiponectin, sENG, leptin, PTX3, PIGF and sFItl were taken forward to test the performance of the standard proteins spiked into the commercial Serum Matrix (1 :10 dilution) using three alternative biomarker concentrations. The estimated Serum Matrix Coefficient (SMCf) varied the least for the spiked ADAM 12 (78-92%), PIGF (76-98%), PTX3 (78-97%), sENG (80-131 %) and sFItl (128-184%) reference proteins (Table 7). For leptin, the confounding effect of the Serum Matrix was more pronounced for low concentrations of the biomarker (84-157%). In case of adiponectin, a high variability in the confounding effect of the Serum Matrix on the measured MFI values was detected across alternative biomarker concentrations, ranging from 15% to 1611%. Thus, adiponectin was excluded from multiplex immunoassay development due to expected unreliable estimates of the biomarker concentrations.

Example 2

Multiplex assay shows high quality performance

ADAM 12, sENG, leptin, Pentraxin3, PIGF and sFItl were taken forward to the next step aiming to develop a multiplex assay. In brief, the multiplex assay targeted simultaneously all the six retained biomarkers in a single well experiment (as detailed in the materials and methods). From here onward we refer to the developed multiplex testing protocol as a 6PLEX assay.

For the reference proteins diluted in GAD, the MFI values measured in the singleplex and multiplex assay formats were strongly correlated (Spearman correlation, r= 1.00, P < 0.0005), indicating that multiplexing of the reference proteins and antibodies does not interfere the signal intensity of the targeted biomarkers (Figure 4). Multiplexing also did not affect fluorescence signal response (MFI values rising >1000 units) and analytical accuracy (95-109%) detected for ADAM 12, adiponectin, sENG, leptin, Pentraxin3 and sFItl (Figure 3). Although fluorescence signal intensity of the PIGF measurements in the multiplex format (up to 300 MFI units) did not reach as high as in the singleplex assay, the estimated accuracy at each biomarker concentration was stable and high (99-114%).

The effect of serum matrix to 6PLEX assay performance was assessed spiking reference proteins into the commercial Serum Matrix dilutions in GAD (1 :10 and 1 :20) at three alternative biomarker concentrations (Table 11). For all biomarkers, the estimated Serum Matrix Coefficient (SMCf) was less variable and closer to the ideal recovery of 100% for the Serum Matrix dilution 1 :20. The most stable performance of the assay was detected for sENG (SMCf for alternative concentrations 97-99%) followed by ADAM 12 (95-102%), sFItl (100-115%) and PTX3 (72-89%) measurements (Table 11). For PIGF (97-133%) and leptin (106-146%) the presence of serum affected more the measurements at low biomarker concentrations. When the reference proteins were spiked into the Serum Matrix dilution 1 :10, the confounding effect of the latter was more pronounced, and the estimated recovery substantially lower compared to more diluted sera (1 :20 dilution). For PIGF, sFItl , sENG and leptin the SMCf did not exceed 90% for any of the measurements in 1 :10 diluted the Serum Matrix.

The estimated inter-assay variability was consistently low for all targeted biomarkers. The coefficient of variation (CV%) of ADAM12, leptin, PTX3, PIGF, sENG and sFItl between independent experiments was estimated 5.29, 1.88, 2.0, 7.84, 5.60 and 4.18%, respectively (Table 9).

Example 3

6PLEX assay performance in a test set of pregnant and non-pregnant reproductive age women

A pilot assessment of the performance of the 6PLEX test for the analysis of human sera was conducted using samples available in the in-house reproductive biomedicine biobank. The eight Testing Serum Samples had been drawn from healthy pregnant women at 12, 16, 25, 26, 33, and 36 gestational weeks and from two non-pregnant reproductive-age women. ADAM 12, SENG, leptin, PIGF and sFItl showed a statistically significant and strong correlation (Spearman r³ 0.83; P < 0.05) between the biomarker measurements in the singleplex and multiplex assay formats (Figure 4B). Across all samples, the calculated median fold-difference between the measurements in multiplex compared to singleplex format was 0.88, 1.02, 1.07, 1.18 and 1.20 for leptin, ADAM12, sENG, sFItl and PIGF respectively. However, in case of PTX3 the overall range of measured concentrations was low and the sensitivity of the assay appeared to be systematically reduced compared to the singleplex test (Spearman r= 0.22; P = 0.63). For individual samples, median 0.59-fold lower PTX3 concentrations were estimated in the 6PLEX compared to single biomarker measurements.

To evaluate the robustness of the 6PLEX assay in possible real-life clinical laboratory conditions, the freeze-thaw stability was additionally assessed. The median fold-changes between the measurements in sequential freeze-thaw cycles were 1.40 for ADAM 12, 1.23 for PTX3, 1.36 for PIGF, 0.99 for sENG, 1.03 for leptin and 1.03 for sFItl (Table 10).

Example 4

Luminex^® 6PLEX and B R A H M S assay comparison

To validate and compare the Luminex^® 6PLEX assay results with the commercially currently available B R A H M S assay, we analyzed 61 serum samples drawn during the 3rd trimester of human pregnancy (180-275 g.d; Table 5, Table 6). The measured sample set included 25 samples from the women with a later onset of PE (4-62 days after blood draw; median 21 d) and 32 serums representing controls without PE until delivery. Additionally, four samples drawn at the diagnosis of PE were utilized in the analysis as the reference levels of the biomarkers at the disease onset.

The range of measured serum concentrations of both, sFItl (Luminex^® 2,207-39,417 pg/mL vs. B R A H M S 817-11 ,010 pg/mL) and PIGF (7.19 - 28,031 pg/mL vs. 12 - 998 pg/mL) were manifold wider in the Luminex^®6PLEX assay compared to the B R A H M S singleplex fluorescence-based immunoassays. Thus, the currently applied rule-in and rule-out criteria for PE prediction that have been developed for the measurements on the B R A H M S Kryptor platform, could not be automatically transferred to Luminex^® xMAP based test results.

Despite the differences in the measurement ranges, the sFlt1/PIGF ratios exhibited a high correlation between two platforms (Pearson’s r=0.93, P < 0.0001 ; Figure 5A). Among pregnant women without any previous signs of the disease, the sFlt1/PIGF ratio estimated from both assays was able to discriminate with high confidence the cases who developed PE within 27 days (Group I vs. Ill, Mann-Whitney test, P<0.0005; Figure 5 B, C). sFlt1/PIGF ratios of both tests also discriminated equally well patients with isolated clinical symptoms (gestational hypertension, proteinuria) that progressed to PE within 4-43 days (Groups IV vs. V, Mann-Whitney test, P<0.0005). However, only the Luminex^® 6PLEX assay was able to measure statistically significant increased sFItl levels already earlier than 28 days before the onset of PE (Groups I vs. II, Mann-Whitney test, P<0.05). This might reflect increased sensitivity of the sFItl measurements using Luminex^® 6PLEX compared to B R A H M S assay due to a broader measurement range. Notably, neither of the assays showed discriminative values in the PIGF measurements between Group I vs. II.

Example 5

Performance of sENG, ADAM12, leptin and PTX3 in the Luminex^® 6PLEX assay sENG values measured with Luminex^® 6PLEX assay were highly correlated with serum sFItl levels (Pearson’s r=0.86, P<0.0001 ; Figure 6A). Our data robustly confirms that sENG represents an alternative biomarker for PE discrimination irrespective of clinical signs alerting to PE (Group I vs. Group III and Group IV vs. Group V; Mann-Whitney test, P<0.002; Figure 6B). Three additional analyzed biomarkers (ADAM 12, leptin and PTX3) exhibited no individual discriminative measurement values between any of the clinical subgroups (Figure 6C, D). However, the level of PTX3 in term pregnancy serum samples was at the detection limit (Table 8). As the developed Luminex^® 6PLEX assay appeared to be technically unsuitable for accurate measurements of PTX3 during late gestation, PTX3 was not included into the subsequently developed PE prediction models for 2^nd half of pregnancy.

Example 6

High predictive power of the combinatory effect ofsFItl, PIGF, sENG, ADAM12 and leptin in models of Luminex^® 6PLEX assay to predict PE

Statistical models built with alternative combinations of the measured biomarkers were analyzed in order to evaluate their optimal combination for the clinical utility in predicting PE within 2 months (in total 30 tested alternative formulae; Table 12, Table 13). Predictive performance of the analyzed biomarkers was evaluated based on the measurements of sera drawn from pregnant women after 179^th gestational day. These patients had either developed PE within 4-60 days after blood draw (n=25) or progressed uneventfully until delivery (n= 32).

When only the biomarker measurement data from the 6PLEX assay were alternatively additively incorporated into the prediction formula (Setting 1 : 10 Models 1A-J, combining minimum 2 and maximum 5 biomarkers), nearly equal performance was achieved for the combination of sFItl , PIGF, ADAM 12, sENG with leptin (Model 1A: AUC =0.956; 95 Cl% 0.89-0.99) or without leptin (Model 1C: AUC = 0.958; 95 Cl% 0.90-0.99) (Table 12, Figure 4). The usage of the sFlt1/PIGF ratio determined by commercially offered B R A H M S assay exhibited the lowest predictive value (Model J: AUC=0.867; 95 Cl% 0.76-0.95). Example 7

Prediction of PE with the Luminex^® 6PLEX measurements adjusted for gestational age and for leptin, maternal weight at blood draw

As biomarker concentrations follow tight and independent dynamics across pregnancy, the further formulae were developed for all tested biomarker combinations that were adjusted for gestational age at sampling (Setting 2: 10 Models 2A-J combining minimum 2 and maximum 5 biomarkers). Before incorporation into the model, leptin measurements were additionally adjusted for the maternal weight at the blood draw as recommended (Blum et al 1997).

The best performance for the PE prediction was achieved when sFItl , PIGF, sENG,

ADAM 12 and leptin biomarkers measured at the 6PLEX assay were additively combined (Model 2A: AUC=0.984; 95 Cl% 0.96-1.00; Figure 7, Table 12). Model 2A exhibited high sensitivity and specificity (91.7% and 96.9%, respectively) for PE prediction. Alternative Models 2B to 2J built on Luminex^® 6PLEX data performed almost equally well (AUC=0.932- 0.975) with specificity and sensitivity ranging from 91.7-100% and 81.2-93.8%, respectively. Notably, when building the model based on the samples measured less than 28 days before the PE onset, the Luminex^® 6PLEX assay exhibited maximum accuracy (AUC=1.00; 95%CI 1.00-1.00; Table 14).

The Model 2A combining the measurements of sFItl , PIGF, sENG, ADAM 12 and leptin resulted in the PE prediction formula (Table 12; Table 13), where p(i) denotes the prediction for PE development:

In the developed model, p(i) > 0.443 and p(i) £ 0.443 were estimated to represent the rule-in and rule-out threshold for the risk to develop PE within 2 months.

The usage of the sFlt1/PIGF ratio determined by commercially offered B R A H M S assay exhibited the lowest predictive value even when corrected for the gestational age (Model 2-J: AUC=0.872; 95CI% 0.76-0.96) with reduced and high variable sensitivity (79.2%; 95CI%: 17-92) and specificity (90.6%; 95CI%: 41-97).

Example 8

Highly accurate prediction of PE with the Luminex^® 6PLEX measurements adjusted for gestational age and for maternal weight at blood draw In Setting 3, all biomarker measurements were adjusted for both gestational age and maternal weight at blood draw (10 Models 3A-J combining minimum 2 and maximum 5 biomarkers).

Again the superior predictive performance for the PE was achieved when gestational age- and weight- adjusted values of five biomarkers measured by the 6PLEX assay (sFItl , PIGF, sENG, ADAM 12 and leptin) were additively combined (Model 3A: AUC=0.984; 95 Cl% 0.96- 1.00; Figure 7, Table 12, Table 13). Model 3A also exhibited the best combination of sensitivity and specificity (100% and 96.9%, respectively) for PE prediction. When leptin was excluded from the prediction formula, there was only a minor drop in the AUC estimate (model 3C; AUC=0.99) driven by reduced specificity (93%), whereas sensitivity to predict PE was maintained 100%.

The Model 3A combining the measurements of sFItl , PIGF, sENG, ADAM 12 and leptin resulted in the PE prediction formula (Table 13), where p(i) denotes the prediction for PE development:

In the developed model, p(i) > 0.175 and p(i) £ 0.175 were estimated to represent the rule-in and rule-out threshold for the risk to develop PE within 2 months.

The usage of the sFlt1/PIGF ratio estimate derived from the commercially offered

B R A H M S assay outcome exhibited the lowest predictive value also for the analysis Setting 3 (Model 3J: AUC=0.872; 95CI% 0.76-0.96). Sensitivity (79.2%; 95CI%: 16-92) and specificity (90.6%; 95CI%: 69-100) of this test exhibited broad confidence intervals.

Example 9

Comparative utilization of the Luminex® 6PLEX Model 2A, Model 3A and currently applied B R A H M S sFIH/PIGF (Model 1J) to predict PE risk in individual cases

To evaluate the accuracy of the Luminex^® 6PLEX Model 2A, Models 3A compared to currently commercially applied B R A H M S sFlt1/PIGF (Model 1J) in predicting PE development in individual cases, the formula was retrospectively applied to the measured serum samples from either the women who eventually developed PE or respective controls. The 6PLEX assay was able to predict or reject PE development for 53/57 serum samples (93.0%) using Model 2A and 55/57 (96.5%) for Model 3A (Table 15; Table 16). Model 3A resulted in no false-negative predictions (sensitivity 100%). The only false negative result with Model 2A was measured for a sample drawn 62 days before the onset of PE, at gestational day 216 (sample II-6). However, the follow-up serum sample drawn for the same women 48 days later (14 days before the PE confirmation) was already convincingly alerting to the PE risk (sample V-8).

The three false-positive cases predicted by the Model 2A represented other severe adverse pregnancy outcomes than PE, and none of these cases proceeded until normal vaginal delivery. Thus, the test cannot be reliably interpreted as the false-positive outcome, because the probability of eventually developing PE during these gestations cannot be ruled out. One of these samples (I-5) was drawn from a pregnancy case complicated with gestational diabetes and intrahepatic cholestasis with extremely elevated liver enzymes. The patient also developed hypertension 16 days after the blood draw and eventually underwent labour induction and vacuum extraction due to fetal distress (sex: girl; gestational age at birth: 38 weeks and 5 days; birth weight: 3558 grams). In two cases (I-2, I-6) the offspring had been diagnosed at the birth as a small-for-gestational-age newborn. For the case I-6 the emergency Cesarean section had been performed due to a severe fetal growth restriction (sex: girl; gestational age at birth 34 weeks and 4 days; birth weight: 1510 grams) and placental defects such as its small size (only 181 g), symptoms of infarctions and intervillous thrombi. The sample I-2 had been drawn from the women, who developed gestational hypertension 49 days after blood draw that was shortly followed by an elective Cesarean section due to fetal breech position (sex: girl; gestational age at birth: 39 weeks and 2 days; birth weight: 2782 grams). Sample I-2 also provided false-positive PE prediction, when using Model 3A. There was only one unclear PE prediction using Luminex^® 6PLEX assay data (Model 3A, sample 1-10) as no pregnancy complications were observed until delivery. Coincidentally combination of incorporated biomarkers may result in borderline estimates relative to the threshold value of the model. For 1-10, p(i) of the Model 3A was estimated to be 0.1754, very close to the threshold of the formula (0.1750).

When applying the currently recommended thresholds for PE prediction from the

B R A H M S sFlt1/PIGF immunoassay data (rule in >85, rule out <33), the prognostic yield compared to the Luminex^® 6PLEX assay Model 2A and 3A was significantly lower (53/56 (93.0%%) vs 55/57 (96.5%) vs 42/57 (85.7%)). The load of either false-positive (n=2) or false-negative predictions (n=5) of the B R A H M S sFlt1/PIGF test was 12.3% (Table 15; Table 16). The two false-positive outcomes (I-2, I-6; both SGA cases) represented also false-positive predictions for PE by the Model 2A of the Luminex^® 6PLEX assay. It has been discussed before that the altered biomarker profile in preeclampsia and severe fetal growth restriction cases cannot be equivocally distinguished. Consistent with the previous literature reports, in most cases the B R A H M S sFlt1/PIGF test failed to predict PE in long-term (PE ³27 days after blood draw; false-negatives; n=5; inconclusive; n=3). Overall, inconclusive outcome (sFlt1/PIGF 38-85) was detected for 14% of analyzed serums representing both, cases that later developed PE (n=4) or proceeded without complications until delivery (n=4).

Taken together, the B R A H M S sFlt1/PIGF test and the Luminex^® 6PLEX Model 2A and 3A performed with a comparable quality in avoiding false-positive predictions, but the latter appear to be far more accurate in identifying true positives and true negatives (Table 16).

The Luminex^® 6PLEX Models 3A and 3C that incorporating also the presence/absence of PE symptoms at the blood draw provided the most accurate prediction, reaching the outcome of no false-positives and no-false negatives in our dataset. Clinically applicable formulae along with accurate coefficient and thresholds are under development.

Conclusions

Six biomarkers, sFItl , PIGF, sENG, leptin, ADAM12 and PTX3, have been simultaneously measured from human sera in a cost effective way with high precision and accuracy. A precise and usable preeclampsia prognosis through modelling the PE prediction formula has been generated based on the biomarker measurements. An analytically valid protocol for future opportunities to further characterize the presented approach for even more precise PE prediction has been generated.

The high quality performance of the method of the present invention allowed precise measurement of all six biomarkers. Both in singleplex and in multiplex the analytical accuracy was very high. In addition, the multiplexing did not interfere the signal intensities and no cross-disturbance was detected yielding high correlation against singleplex approach.

Simultaneous quantification of six known biomarkers targeting PE prognosis resulted in prediction model that was compared against commercially available B R A H M S assay measuring sFItl and PIGF. The new multiplexed solution using five quantified biomarker values (sFItl , PIGF, sENG, leptin and ADAM12) provided far superior performance in PE prediction in comparison to B R A H M S Kryptor assays. The strength of this study is the use of well characterized third trimester serum samples drawn from pregnant women during a prospective observational cohort study Happy Pregnancy .

Abbreviations: ADAM12, ADAM Metallopeptidase Domain 12; AUC, area under the curve;

Cl, confidence interval; PIGF, placental growth factor; ROC, receiver operating

characteristics; sENG, Endoglin; sFItl , soluble fms-like tyrosine kinase 1. References

Edgardo Abalos, Cristina Cuesta, Ana L. Grosso, Doris Chou, Lale Say“Global and regional estimates of preeclampsia and eclampsia: a systematic review” European Journal of Obstetrics & Gynecology and Reproductive Biology 170 (2013) 1-7.

Blum F Werner, Englaro P, Hanitsch S, Juul A, Hertel T N, Muller J, Skakkebask N E, Heiman L M, Birkett M, Attanasio M A, Kiess W, Rascher W. Plasma Leptin Levels in Healthy Children and Adolescents: Dependence on Body Mass Index, Body Fat Mass, Gender, Pubertal Stage, and Testosterone. J Clin Endocrinol Metab. 1997; 82 (9). 2904- 2910

Brown MA, Magee LA, Kenny LC, Karumanchi SA, McCarthy FP, Saito S, Hall DR, Warren CE8, Adoyi G9, Ishaku S9; International Society for the Study of Hypertension in Pregnancy (ISSHP). Hypertensive Disorders of Pregnancy: ISSHP Classification, Diagnosis, and Management Recommendations for International Practice. Hypertension. 2018;72(1):24-43

Dragan I, Georgiou T, Prodan N, Akolekar R, Nicolaides KH. Screening for pre-eclampsia using sFlt-1/PIGF ratio cut-off of 38 at 30-37 weeks' gestation. Ultrasound Obstet Gynecol. 2017;49:73-77.

Sildver K, Veerus P, Lang K. Sunnikaalukoverad Eestis ja sunnikaalu mojutavad tegurid: registripohine uuring. Eesti Arst, 2015; 8, 465-470.

Verlohren S, Herraiz I, Lapaire O, Schlembach D, Zeisler H, Calda P, Sabria J,Markfeld-Erol F, Galindo A, Schoofs K, Denk B, Stepan H. New gestational phase-specific cut-off values for the use of soluble fms-like tyrosine kinase-1/placental growth factor ratio as diagnostic test for preeclampsia. Hypertension 2014; 63: 346-352.

Zeisler H1 , Llurba E, Chantraine F, Vatish M, Staff AC, Sennstrom M, Olovsson M,

Brennecke SP, Stepan H, Allegranza D, Dilba P, Schoedl M, Hund M, Verlohren S.

Predictive Value of the sFlt-1 : PIGF Ratio in Women with Suspected Preeclampsia. N Engl J Med. 2016 Jan 7;374(1):13-22. doi: 10.1056/NEJMoa1414838.

NICE Diagnostics Guidance No 23, 2016. Triage PIGF test, Elecsys immunoassay sFIt- 1/PIGF ratio, DELFIA Xpress PIGF 1-2-3 test, and BRAHMS sFlt-1 Kryptor/BRAHMS PIGF plus Kryptor PE ratio (https://www.nice.org.uk/guidance/dg23) Table 1

Table 2

Table 4

Table 6

Table 6 (continued)

Table 7

Table 8

Table 9

Table 10

Table 13. Formulas coefficients and thresholds for preeclampsia prediction. Logistic regression models for prediction of preeclampsia (PE) onset 4-62 days after the blood draw using combinations of measured biomarkers.

Claims

Claims

1. A method of prognosing or diagnosing preeclampsia in a subject wherein the method comprises, measuring the level of at least four biomarkers in a blood sample from the subject; optionally determining at least one clinical cofactor from the subject; generating a prediction value, wherein the prediction value indicates whether the subject will develop or will not develop preeclampsia, or has preeclampsia; the prediction value being based on the levels of the at least four biomarkers in the blood sample from the subject and optionally based on the at least one clinical cofactor; wherein the at least four biomarkers are sFItl , PIGF, ADAM12, sENG and optionally leptin; the at least one optional clinical cofactor is selected from gestational age and the weight of the subject; wherein a high sFItl , low PIGF, low ADAM12 and high sENG level as compared with control indicates an increased probability of the subject developing preeclampsia or increased probability that the subject has preeclampsia; and if included in the calculation of the prediction value, a high leptin level as compared with control indicates an increased probability of the subject developing preeclampsia or increased probability that the subject has preeclampsia; and if included in the calculation of prediction value, a low gestational age and high weight of the subject at blood sampling indicates an increased probability of the subject to developing preeclampsia or increased probability that the subject has preeclampsia.

2. The method according to claim 1 wherein the biomarkers are sFItl , PIGF, ADAM12 and sENG.

3. The method according to claim 1 wherein the biomarkers are sFItl , PIGF, ADAM 12, sENG and leptin.

4. The method according to any preceding claim wherein the prediction value is based on gestational age and the weight of the subject as clinical cofactors.

5. The method according to any preceding claim wherein the prediction value is compared to a threshold value.

6. The method according to any preceding claim wherein the prediction value indicates whether the subject will develop or will not develop preeclampsia within a two-month period calculated from blood sampling.

7. The method according to any preceding claim wherein the blood sample is collected on or after the 179^th gestational day.

8. The method according to any preceding claim wherein the prediction value is

calculated according to a formula:

wherein a, b, d, e, f, g and h are negative numbers and c is a positive number; or one or more of f, g and h are 0.

9. The method according to claim 8 wherein, a = -158.3 ± 10% b = -4.8 ± 10% c = 31.9 ± 10% d = -17.8 ± 10% e = -10.3 ± 10% f = -2.1 ± 10% g = -0.2 ± 10% h =-0.1 ± 10%

10. The method according to claim 9, wherein the prediction value is compared to a threshold value of 0.175 ± 0.005.

11. The method according to claim 10, wherein a p(i) value equal to or above the

threshold value indicates the subject will develop preeclampsia within a two-month time period calculated from blood sampling or has preeclampsia, or a p(i) value below the threshold value indicates the subject will not develop preeclampsia within a two-month time period calculated from blood sampling or does not have preeclampsia.

12. The method according to claim 8 wherein the prediction value is calculated according to a formula:

wherein a, b, d, e, g and h are negative numbers and c is a positive number; or one or more of g and h are 0.

13. The method according to claim 12 wherein: a = -136.9 ± 10% b = -3.4 ± 10% c = 26.5 ± 10% d = -14.7 ± 10% e = -9.7 ± 10% g= -0.1 ± 10% h= -0.2 ± 10%

14. The method according to claim 13, wherein the prediction value is compared to a threshold value of 0.204 ± 0.005.

15. The method according to claim 14, wherein a p(i) value equal to or above the

threshold value indicates the subject will develop preeclampsia within a two-month time period calculated from blood sampling or has preeclampsia, or a p(i) value below the threshold value indicates the subject will not develop preeclampsia within a two-month time period calculated from blood sampling or does not have

preeclampsia.

16. The method according to any one of claims 1 to 7 wherein the prediction value is calculated according to a formula:

wherein a, b, d, e, f and g are negative numbers and c is a positive number; or one or more of f and g are 0.

17. The method according to claim 16 wherein: a = -114 ± 10% b = -3.7 ± 10% c = 19.9 ± 10% d = -10.3 ± 10% e = -5 ± 10% f = -2.5 ± 10% g= -0.2 ± 10%

18. The method according to claim 17, wherein the prediction value is compared to a threshold value of 0.443 ± 0.005.

19. The method according to claim 18, wherein a p(i) value equal to or above the

threshold value indicates the subject will develop preeclampsia within a two-month time period calculated from blood sampling or has preeclampsia, or a p(i) value below the threshold value indicates the subject will not develop preeclampsia within a two-month time period calculated from blood sampling or does not have

preeclampsia.

20. The method according to any one of the preceding claims, wherein the biomarkers are measured using a multiplex platform.

21. The method according to claim 20, wherein the multiplex platform is a microsphere- based platform, for example xMAP technology.

22. The method according to any preceding claim wherein the subject is asymptomatic for preeclampsia.

23. The method according to anyone of claims 1 to 21 wherein the subject has not been diagnosed as having preeclampsia but presents with one or more preeclampsia symptoms, preferably preeclampsia symptoms selected from one or more of the following: increased blood pressure or hypertension, increased urine protein or proteinuria, water retention (oedema), elevated liver enzyme levels, headache, visual disturbances, reduced urine output or oliguria, epigastric pain, eclamptic seizures, reduced platelet count and uteroplacental dysfunction, more preferably preeclampsia symptoms selected from increased blood pressure or hypertension and/or an increased urine protein or proteinuria.