Received: 9 February 2022 | Revised: 31 May 2022 | Accepted: 9 June 2022 | First published online: 11 July 2022 DOI: 10.1002/ijgo.14306 CLINICAL ARTICLE Obstetrics First-­trimester screening for pre-­eclampsia and small for gestational age: A comparison of the Gaussian and Fetal Medicine Foundation algorithms Berta Serrano | Erika Bonacina | Carlota Rodo | Pablo Garcia-­Manau | María Ángeles Sanchez-­Duran | María Pancorbo | Cristina Forcada | María Teresa Murcia | Ana Perestelo | Mireia Armengol-­Alsina | Manel Mendoza | Elena Carreras Maternal Fetal Medicine Unit, Department of Obstetrics, Hospital Universitari Vall Abstract d'Hebron, Universitat Autònoma de Objective: Pre-­eclampsia (PE) and small for gestational age (SGA) can be predicted from the Barcelona, Barcelona, Spain first trimester. The most widely used algorithm worldwide is the Fetal Medicine Foundation Correspondence (FMF) algorithm. The recently described Gaussian algorithm has reported excellent results Manel Mendoza and Pablo Garcia-­Manau, Department of Obstetrics, Maternal although it is unlikely to be externally validated. Therefore, as an alternative approach, we Fetal Medicine Unit, Hospital Universitari compared the predictive accuracy for PE and SGA of the Gaussian and FMF algorithms. Vall d'Hebron, Universitat Autònoma de Barcelona, Barcelona, Spain. Methods: Secondary analysis of a prospective cohort study was conducted at Vall Email: mmendoza@vhebron.net (M.M.); d'Hebron University Hospital (Barcelona) with 2641 singleton pregnancies. The areas pgarciamanau@gmail.com (P.G.-­M.) under the curve for the predictive performance for early-­onset and preterm PE and early-­onset and preterm SGA were calculated with the Gaussian and FMF algorithms and subsequently compared. Results: The FMF and Gaussian algorithms showed a similar predictive performance for most outcomes and marker combinations. Nevertheless, significant differences for early-­onset PE prediction favored the Gaussian algorithm in the following combi- nations: mean arterial blood pressure (MAP) with pregnancy-­associated plasma pro- tein A, MAP with placental growth factor, and MAP alone. Conclusions: The first-­trimester Gaussian and FMF algorithms have similar performances for PE and SGA prediction when applied with all markers within a routine care setting in a Spanish population, adding evidence to the external validity of the FMF algorithm. KEYWORDS early-­onset pre-­eclampsia, first trimester, PlGF, pre-­eclampsia, screening, uterine artery doppler 1 | I NTRO D U C TI O N using algorithms that include a combination of maternal charac- teristics, biophysical markers (mean arterial blood pressure [MAP] Pre-­eclampsia (PE) and small for gestational age (SGA) are the main and mean uterine artery pulsatility index [UtAPI]), and biochemical complications of placental disease. First-­trimester PE screening markers (placental growth factor [PlGF] and pregnancy-­associated Berta Serrano and Erica Bonacina contributed equally to this article (co-first authors). This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2022 The Authors. International Journal of Gynecology & Obstetrics published by John Wiley & Sons Ltd on behalf of International Federation of Gynecology and Obstetrics. 150 | wileyonlinelibrary.com/journal/ijgo  Int J Gynecol Obstet. 2023;160:150–160. SERRANO et al. | 151 plasma protein A [PAPP-­A]) can predict PE and SGA.1–­4 The Fetal d'Hebron University Hospital (Barcelona) from October 2015 to Medicine Foundation (FMF) and Gaussian algorithms can identify September 2017. 80%–­90% of pregnant women who will develop PE with delivery A total of 3777 unselected singleton pregnant women attend- <32/<34 weeks of gestation1,5 and 60%–­70% of women who will ing their routine first-­t rimester scan (from 11 ± 0 to 13 ± 6 weeks) develop PE with delivery <37 weeks,1,6 at a 10% false-­positive rate were invited to participate, and 2946 women agreed and provided (FPR). These algorithms can also predict 50%–­60% of SGA with de- their written informed consent. Of those, 305 participants (10.4%) livery <32 weeks and 30%–­4 0% of SGA with delivery <37 weeks. 2,4 had to be excluded for the following reasons: missing outcome Both algorithms use a similar methodology to assess the risk for PE: data (n = 86), major fetal defects or chromosomopathies (n = 13), they combine the a priori risk (based on maternal characteristics and miscarriage or fetal death <24 weeks (n = 15), and insufficient re- obstetric and medical history) with the results of various biochem- maining blood sample to measure PlGF (n = 191). Before the imple- ical and biophysical markers, to estimate the individual a posteriori mentation of the first-­t rimester combined screening for PE in 2018, risk for PE, which is used to classify a pregnant person as at high or no PE screening was performed at the Vall d'Hebron University low risk for PE. In both algorithms, risk for PE can be obtained based Hospital; thus, none of the remaining 2641 participants received on maternal factors alone and in combination with any of the bio- aspirin at any time during their pregnancy. Neonatal birthweight chemical and/or biophysical markers. was not available for 158 participants; therefore, predictive accu- Despite the FMF algorithm being the most used and validated racies for SGA were calculated with 2483 participants and their worldwide, the Gaussian algorithm has some features that confer newborns. advantages in the clinical setting, which is why it has been used for Gestational age was confirmed by fetal crown-­rump length routine first-­trimester PE screening in most maternities in Spain measurement during the first-­trimester scan.13 Maternal charac- since 2018. First, blood samples for measurements of biochemical teristics and medical and obstetric history were recorded at the markers (PAPP-­A and PlGF) can be drawn between 8 ± 0 weeks and first-­trimester ultrasound scan via a patient questionnaire. The fol- 13 ± 6 weeks, as with routine aneuploidy screening, while in the FMF lowing maternal characteristics were recorded: age (years); height algorithm biomarkers should be assessed only between 11 ± 0 and (centimeters); weight (kilograms); ethnicity (white European, South 13 ± 6 weeks.6 Second, UtAPI assessment can be done both transab- American, black, Asian, South-­East Asian, and others); smoking dominally and transvaginally, rendering the algorithm more versatile during pregnancy (yes/no); and conception method (spontaneous/ to different clinical settings, as the UtAPI for the FMF algorithm can assisted reproductive technology/ovulation drugs). Medical history be assessed only transabdominally. Third, likelihood ratios for the a variables included the presence of chronic hypertension (yes/no); priori risk calculation were not derived from the study population diabetes (type 1/type 2/no); renal disease (yes/no); systemic lupus in which the algorithm was investigated but from a larger meta-­ erythematosus (yes/no); and antiphospholipid syndrome (yes/no). 7 analysis that included >25 million pregnancies. This may render the Obstetric history variables included parity (nulliparous/multiparous); Gaussian algorithm less overfitted to a given population and, there- gestational age at birth (weeks) in the last pregnancy; interval be- fore, more adaptable for populations with different characteristics. tween the last delivery and the beginning of the current one (years); The FMF algorithm has been developed and prospectively vali- and personal or family history of PE (yes/no). Biochemical mark- dated in large populations, showing comparable predictive perfor- ers, including serum PAPP-­A and PlGF, were measured at the first-­ mances to the original study.8–­12 By contrast, the Gaussian algorithm trimester routine blood test for aneuploidy screening (from 8 ± 0 has been investigated only in a single cohort of participants. In the to 13 ± 6 weeks) by the fully automated Elecsys assays for PAPP-­A past few years, routine PE screening has been implemented in most and PlGF on an immunoassay platform (cobas e analyzers, Roche hospitals, leaving virtually no women at high risk for PE without as- Diagnostics). Biophysical markers, including MAP and UtAPI, were pirin treatment to prospectively assess the external validity of the assessed at the first-­trimester scan. Blood pressure was measured Gaussian algorithm. Therefore, an indirect approach to test the per- automatically using a calibrated device according to a standard pro- formance of the Gaussian algorithm is to compare it with the most cedure: single measurement in one arm (right or left) while women externally validated combined screening tool for PE worldwide: the were seated and after a 5-­min rest. MAP was calculated as: diastolic FMF algorithm. blood pressure + (systolic-­diastolic blood pressure)/3. UtAPI was The aim of this study was to compare the predictive accuracy for measured following the recommendations of the FMF.14 All exam- PE and SGA of the Gaussian and FMF algorithms. iners were certified by the FMF for PE risk assessment and Doppler ultrasound assessment. SGA newborns were defined as having a birthweight below the 2 | M ATE R I A L S A N D M E TH O DS 10th centile according to customized local charts.15 Indication for elective delivery was based on Doppler ultrasound findings and This is a secondary analysis of previously published data, which was conventional cardiotocogram interpretation, according to the cur- used to test the Gaussian algorithm for early-­onset PE prediction.3 rent protocol.16 Newborns were classified as early SGA if delivery That study was approved by the local ethics committee (CEIC-­VHIR occurred before 32 weeks and as preterm SGA if delivery occurred PR[AMI]265/2018) and conducted in a prospective fashion at Vall before 37 weeks. 152 | SERRANO et al. TA B L E 1 Baseline characteristics of the study population based on PE outcome PE < 34 weeks PE < 37 weeks No PE < 37 weeks (n = 11) (n = 30) (n = 2611) a Age (years) 34 (32–­37) 35.5 (31–­38) 32 (28–­36)c BMI 23.1 (22.5–­32.1) 24.0 (22.5–­27.6) 23.8 (21.3–­27.5) Ethnicity White 10 (90.9%) 25 (83.3%) 2209 (84.6%) Black 0 (0.0%) 1 (3.3%) 71 (2.72%) Mixed 1 (9.1%) 2 (6.7%) 209 (8.0%) Asian 0 (0.0%) 2 (6.7%) 63 (2.41%) Southeast Asian 0 (0.0%) 0 (0.0%) 59 (2.26%) Smoking during pregnancy 1 (9.1%) 3 (10.0%) 309 (11.8%) ART 1 (9.1%) 2 (6.7%) 93 (3.6%) Insemination 1 (9.1%) 1 (3.3%) 16 (0.6%) IVF 0 (0.0%) 1 (3.3%) 77(2.95%) IVF with egg donation 0 (0.0%) 1 (3.3%) 25 (0.96%) Medical history Chronic hypertension 3 (27.3%)a 5 (16.7%)a 24 (0.9%)b,c Diabetes 0 (0.0%) 1 (3.3%) 35 (1.3%) Autoimmune disease 0 (0.0%) 3 (10.0%) 105 (4.0%) APS 0 (0.0%) 1 (3.3%) 8 (0.3%) Obstetric history Nulliparous 2 (18.2%) 13 (43.3%) 1219 (46.7%) Previous PE 2 (18.2%)a 5 (16.7%)a 30 (1.1%)b,c Biophysical variables GA at the time of first-­trimester ultrasound scan 12.7 (12.3–­13.3) 12.7 (12.3–­13.3) 12.6 (12.1–­13) (weeks) MAP (mm Hg) 96 (88.3–­104.3)a 91.2 (86.7–­97.3)a 84.3 (78.7–­90.7)b,c MoM MAP 1.14 (1.10–­1.37)a 1.14 (1.10–­1.29)a 1.06 (0.97–­1.14)b,c a,c a,b Mean UtAPI 2.25 (1.89–­3.05) 1.91 (1.71–­2.31) 1.68 (1.34–­2.05)b,c MoM UtAPI 1.32 (1.12–­2.13)a 1.19 (1.01–­1.44)a 1.03 (0.84–­1-­26)b,c Biochemical variables GA for PAPP-­A + PlGF measurement 11.4 (9.9–­12.3) 10.9 (9.9–­11.7) 10.6 (10–­11.3) PAPP-­A (mU/L) 1373 (607.3–­2291) 1158 (602.3–­2291) 1358 (823.2–­2370) MoM PAPP-­A 0.73 (0.6–­0.93)a 0.72 (0.57–­1.05)a 1.05 (0.73–­1.5)b,c a a PlGF (pg/ml) 22.3 (19.0–­29.8) 25.0 (19.3–­31.7) 32.2 (24.3–­43.0)b,c MoM PlGF 0.69 (0.52–­1.05)a 0.78 (0.63–­0.98)a 0.96 (0.76–­1.19)b,c Note: Categorical data are reported as frequency (percentage) and continuous data as median (interquartile range). Abbreviations: APS, antiphospholipid syndrome; ART, assisted reproductive technique; BMI, body mass index; GA, gestational age; IVF, in vitro fertilization; MAP, mean arterial pressure; MoM, multiple of median; PAAP-­A , pregnancy-­associated plasma protein A; PlGF, placental growth factor; UtAPI, uterine artery pulsatility index. a Significant difference as compared with unaffected women. b Significant difference as compared with early-­onset pre-­eclampsia (PE). c Significant difference as compared with women with preterm PE. PE was defined according to the guidelines of the International accompanied by proteinuria ≥300 mg in 24 h, spot urine protein/ Society for the Study of Hypertension in Pregnancy: systolic blood creatinine ratio ≥0.3 mg/mg, or dipstick urinalysis ≥1+ when a pressure ≥ 140 mm Hg and/or diastolic blood pressure ≥ 90 mm Hg, quantitative method was not available. 17 Early-­onset and preterm confirmed by repeated measurements over a few hours, de- PE were defined as PE requiring delivery before 34 and 37 weeks, veloping after 20 weeks in previously normotensive women, respectively. SERRANO et al. | 153 TA B L E 2 Baseline characteristics of the study population based on SGA outcome No SGA < 32 weeks SGA < 37 weeks SGA < 37 weeks (n = 8) (n = 44) (n = 2439) Age (years) 31.5 (29–­33) 32 (28.5–­37) 32 (28–­36) BMI 23.1 (21.9–­24.5) 23.1 (20.2–­26.4) 23.9 (21.4–­27.6) Ethnicity White 159 (94.6%) 189 (93.6%) 2196 (84.6%) Black 4 (2.4%) 6 (3.0%) 70 (2.7%) Mixed 5 (3.0%) 5 (2.5%) 209 (8.1%) Asian 0 (0.0%) 1 (0.5%) 64 (2.5%) Southeast Asian 0 (0.0%) 1 (0.5%) 58 (2.2%) a Smoking during pregnancy 0 (0.0%) 15 (34.1%) 283 (11.6%)c ART Insemination 2 (1.2%) 3 (1.5%) 16 (0.6%) IVF 6 (3.6%) 7 (3.5%) 77 (3.0%) IVF with egg donation 0 (0.0%) 1 (2.3%) (% del total) 22 (0.9%) Medical history Chronic hypertension 1 (12.5%)a 2 (4.5%) 27 (1.1%)b Diabetes 0 (0.0%) 2 (4.5%) 32 (1.3%) Autoimmune disease 1 (12.5%) 2 (4.5%) 101 (4.1%) APS 0 (0.0%)a 3 (6.8%) 6 (0.2%)b Obstetric history Nulliparous 5 (62.5%) 20 (45.5%) 1126 (46.2%) a Previous PE 1 (12.5%) 2 (4.5%) 32 (1.3%)b Biophysical variables GA at the time of first-­trimester ultrasound scan 12.4 (12.1–­12.6) 12.4 (11.9–­12.9) 12.6 (12.1–­13) (weeks) MAP (mm Hg) 90.8 (85.2–­96) 86.7 (80–­91.1) 84.3 (78.3–­90.7) MoM MAP 1.14 (1.04–­1.17) 1.07 (0.96–­1.17) 1.05 (0.97–­1.15) a Mean UtAPI 1.88 (1.74–­2.67) 1.94 (1.72–­2.45) 1.68 (1.34–­2.04)c MoM UtAPI 1.12 (1.01–­1.60) 1.20 (1.02–­1.47)a 1.02 (0.84–­1.25)c Biochemical variables GA for PAPP-­A + PlGF measurement 11.4 (10.4–­12.3) 10.7 (10–­11.8) 10.6 (10–­11.3) PAPP-­A (mU/L) 1801 (932.2–­2456) 964.25 (631.0–­1794.5) 1355 (816–­2387) MoM PAPP-­A 0.74 (0.6–­0.89) 0.73 (0.55–­1.1)a 1.06 (0.73–­1.51)c a PlGF (pg/ml) 20.0 (18.1–­26.1) 28.2 (19.5–­38.4) 32.1 (24.1–­43.0)b MoM PlGF 0.60 (0.42–­0.79)a 0.72 (0.61–­0.97)a 0.96 (0.75–­1.18)b,c Note: Categorical data are reported as frequency (percentage) and continuous data as median (interquartile range). Abbreviations: APS, antiphospholipid syndrome; ART, assisted reproductive technique; BMI, body mass index; GA, gestational age; IVF, in vitro fertilization; MAP, mean arterial pressure; MoM, multiple of median; PAAP-­A , pregnancy-­associated plasma protein A; PE, pre-­eclampsia; PlGF, placental growth factor; UtAPI, uterine artery pulsatility index. a Significant difference as compared with unaffected women. b Significant difference as compared with early-­onset small for gestational age (SGA). c Significant difference compared with preterm SGA. For the Gaussian algorithm, multiples of the median (MoMs) website.18 We then coded the variables required for the prediction for each marker were calculated according to the methodology de- formulas according to the description provided in the correspond- scribed in a previous study. 3 For the FMF algorithm, MoMs were ing published articles.1,3 For the Gaussian algorithm, the prenatal obtained using the batch calculation tool provided in the FMF screening software SsdwLab 6 (SBP Soft 2007 S.L) was used to 154 | SERRANO et al. TA B L E 3 DR and AUC for prediction of early-­onset PE by the Gaussian and the FMF algorithms PE < 34 ± 0 weeks (n = 11) Gaussian algorithm DR at 5% FPR DR at 10% FPR DR at 15% FPR DR at 20% FPR DR at 25% FPR DR at 30% FPR A priori risk + AUC (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) MAP 0.795 (0.679–­0.912) 36.4 (9.09–­63.6) 45.5 (18.2–­72.7) 54.6 (27.3–­81.8) 54.6 (27.3–­81.8) 72.7 (45.5–­100.0) 72.7 (45.5–­100.0) MAP + PlGF 0.905 (0.844–­0.965) 36.4 (9.09–­63.6) 63.6 (36.4–­90.9) 81.8 (54.6–­100.0) 81.8 (54.6–­100.0) 81.8 (54.6–­100.0) 90.9 (72.7–­100.0) MAP + UtAPI 0.908 (0.840–­0.975) 63.6 (36.4–­90.9) 63.6 (36.4–­90.9) 63.6 (36.4–­90.9) 72.7 (45.5–­100.0) 90.9 (72.7–­100.0) 100.0 (100.0–­100.0) MAP + PAPP-­A 0.833 (0.727–­0.939) 36.4 (9.09–­63.6) 54.6 (27.3–­81.8) 54.6 (27.3–­81.8) 72.7 (45.5–­95.6) 72.7 (45.5–­95.6) 72.7 (45.5–­95.6) MAP + UtAPI + 0.910 (0.844–­0.977) 63.6 (36.4–­90.9) 63.6 (36.4–­90.9) 72.7 (45.5–­100.0) 72.7 (45.5–­100.0) 81.8 (54.6–­100.0) 100.0 PAPP-­A (100.0–­100.0) MAP + UtAPI + 0.951 (0.919–­0.983) 54.6 (27.3–­81.8) 81.8 90.9 (72.7–­100.0) 100.0 100.0 100.0 PlGF (54.5–­100.0) (100.0–­100.0) (100.0–­100.0) (100.0–­100.0) MAP + UtAPI 0.945 (0.912–­0.979) 54.6 (27.3–­81.8) 81.8 90.9 (72.7–­100.0) 100.0 100.0 100.0 + PlGF + (54.6–­100.0) (100.0–­100.0) (100.0–­100.0) (100.0–­100.0) PAPP-­A Note: Comparisons between areas under the curve (AUCs) were performed by two-­t ailed P values. Abbreviations: CI, confidence interval; DR, detection rate; FMF, Fetal Medicine Foundation; FPR, false-­positive rate; MAP, mean arterial pressure; PAPP-­A , pregnancy-­associated plasma protein A; PE, pre-­eclampsia; PlGF, placental growth factor; UtAPI, mean uterine artery pulsatility index. TA B L E 4 DR and AUC for prediction of preterm PE by the Gaussian and the FMF algorithms PE < 37 ± 0 weeks (n = 30) Gaussian algorithm DR at 5% FPR DR at 10% FPR DR at 15% DR at 20% DR at 25% DR at 30% FPR A priori risk + AUC (95% CI) (95% CI) (95% CI) FPR (95% CI) FPR (95% CI) FPR (95% CI) (95% CI) MAP 0.737 0.2667 (13.3–­43.3) 36.7 50.0 53.3 56.7 63.3 (46.7–­8 0.0) (0.648–­0.827) (20.0–­53.3) (33.3–­66.7) (33.3–­70.0) (40.0–­73.3) MAP + PlGF 0.802 26.7 (13.3–­43.3) 46.7 60.0 66.7 73.3 76.7 (60.0–­90.0) (0.722–­0.881) (30.0–­63.3) (43.3–­76.6) (50.0–­82.2) (53.3–­86.7) MAP + UtAPI 0.782 36.7 (20.0–­53.3) 40.0 46.7 56.7 76.7 80.0 (63.3–­93.3) (0.692–­0.872) (23.3–­56.7) (30.0–­63.3) (36.7–­76.7) (60.0–­90.0) MAP + PAPP-­A 0.773 33.3 (20.0–­50.0) 43.3 53.3 63.3 63.3 63.3 (46.7–­8 0.0) (0.687–­0.859) (26.7–­63.3) (33.3–­70.0) (46.7–­8 0.0) (46.7–­8 0.0) MAP + UtAPI 0.797 36.7 (20.0–­53.3) 43.3 53.3 70.0 76.7 83.3 (70.0–­96.7) + PAPP-­A (0.708–­0.886) (26.7–­60.0) (36.7–­73.3) (53.3–­83.4) (56.7–­90.0) MAP + UtAPI 0.798 36.7 (20.0–­56.7) 46.7 56.7 80.0 80.0 80.0 (63.3–­93.3) + PlGF (0.704–­0.893) (30.0–­66.7) (40.0–­73.3) (63.3–­93.3) (63.3–­93.3) MAP + UtAPI 0.782 33.3 (16.7–­50.0) 46.7 63.3 76.7 76.7 76.7 (60.0–­90.0) + PlGF + (0.683–­0.882) (30.0–­66.7) (46.7–­8 0.0) (60.0–­90.0) (60.0–­90.0) PAPP-­A Note: Comparisons between areas under the curve (AUCs) were performed by two-­t ailed P values. Abbreviations: CI, confidence interval; DR, detection rate; FMF, Fetal Medicine Foundation; FPR, false-­positive rate; MAP, mean arterial pressure; PAPP-­A , pregnancy-­associated plasma protein A; PE, pre-­eclampsia; PlGF, placental growth factor; UtAPI, mean uterine artery pulsatility index. calculate early-­onset PE probability scores. For the FMF algorithm, clinical practice. Therefore, there are 15 possible marker combina- the risk calculation tool provided in the FMF website was used.19 tions. Nevertheless, only the seven most clinically relevant have been Besides the a priori risks, the four markers (PAPP-­A , PlGF, MAP, investigated in this study (MAP alone, MAP + PlGF, MAP + UtAPI, and UtAPI) can be incorporated alone or in combination of two, three, MAP + PAPP-­A , MAP + UtAPI + PAPP-­A , MAP + UtAPI + PlGF, and or four for risk calculation, depending on the markers available in the MAP + UtAPI + PlGF + PAPP-­A). SERRANO et al. | 155 FMF algorithm DR at 5% FPR DR at 10% FPR DR at 15% FPR DR at 20% FPR DR at 25% FPR DR at 30% FPR AUC (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) P value 0.758 (0.621–­0.895) 27.3 (0.0–­5 4.6) 27.3 (0.0–­5 4.6) 27.3 (9.09–­63.6) 54.6 (27.3–­90.9) 72.7 72.7 (45.5–­100.0) 0.0214 (45.5–­100.0) 0.858 (0.768–­0.947) 45.5 (18.2–­72.7) 45.5 (18.2–­72.7) 63.6 (36.4–­90.9) 72.7 (45.5–­100.0) 81.8 81.8 (54.6–­100.0) 0.0112 (54.6–­100.0) 0.868 (0.775–­0.961) 45.5 (18.2–­72.7) 54.6 (27.3–­81.8) 63.6 (36.4–­90.9) 72.7 (45.5–­100.0) 81.8 81.8 (54.6–­100.0) 0.1059 (54.6–­100.0) 0.771 (0.631–­0.911) 27.3 (0.0–­5 4.6) 27.3 (0.0–­5 4.6) 54.6 (27.3–­81.8) 63.6 (36.4–­90.9) 72.7 (45.5–­95.6) 72.7 (45.5–­95.6) 0.0022 0.870 (0.768–­0.972) 45.5 (18.2–­72.3) 54.6 (27.3–­81.8) 72.7 (45.5–­90.9) 81.8 (54.6–­100.0) 81.8 81.8 (54.6–­100.0) 0.1374 (54.6–­100.0) 0.923 (0.864–­0.982) 63.6 (36.4–­90.9) 72.7 72.7 90.9 (72.7–­100.0) 90.9 90.9 (72.7–­100.0) 0.1325 (45.5–­100.0) (45.5–­100.0) (72.7–­100.0) 0.945 (0.908–­0.982) 54.6 (27.3–­81.8) 90.9 90.9 90.9 (72.7–­100.0) 100 (100–­100) 100 (100–­100) 0.9651 (54.6–­100.0) (72.7–­100.0) FMF algorithm DR at 5% FPR DR at 10% FPR DR at 15% FPR DR at 20% FPR DR at 25% FPR DR at 30% FPR AUC (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) P value 0.727 26.7 (10.0–­43.3) 26.7 36.7 50.0 60.0 60.0 0.3884 (0.637–­0.817) (13.3–­46.7) (20.0–­53.3) (33.3–­70.0) (43.3–­76.7) (43.3–­8 0.0) 0.790 36.7 (20.0–­53.3) 40.0 53.3 60.0 66.7 66.7 0.4292 (0.712–­0.868) (23.3–­60.0) (33.3–­70.0) (43.3–­76.7) (50.0–­83.3) (50.0–­83.3) 0.786 30.0 (13.3–­50.0) 43.3 46.7 63.3 70.0 73.3 0.8590 (0.701–­0.871) (26.7–­63.3) (30.0–­63.3) (46.7–­8 0.0) (53.3–­86.7) (56.7–­86.7) 0.747 23.3 (10.0–­4 0.0) 36.7 50.0 53.3 60.0 63.3 0.0955 (0.658–­0.836) (20.0–­53.3) (33.3–­66.7) (36.7–­70.0) (40.0–­76.7) (43.3–­8 0.0) 0.800 36.7 (16.7–­53.3) 50.0 56.7 70.0 76.7 76.7 0.8846 (0.714–­0.887) (33.3–­66.7) (40.0–­73.3) (50.0–­86.7) (60.0–­90.0) (60.0–­90.0) 0.818 36.7 (20.0–­53.3) 50.0 56.7 66.7 80.0 80.0 0.4780 (0.739–­0.897) (30.0–­66.7) (40.0–­73.3) (50.0–­83.3) (63.3–­93.3) (66.7–­93.3) 0.818 36.7 (20.0–­56.7) 63.4 70.0 73.3 76.7 76.7 0.3467 (0.728–­0.907) (43.3–­8 0.0) (53.3–­86.7) (53.3–­86.7) (60.0–­90.0) (60.0–­93.3) and percentage, and comparisons between groups were performed 2.1 | Statistical Analysis by chi-­square or Fisher tests, as appropriate. Continuous variables were reported as the median and interquartile range, and the Mann–­ The statistical software RStudio Team (version 1.2.5033 [2019], Whitney U test was used to assess differences between groups. RStudio: Integrated Development for R. RStudio, Inc.) was used for Receiver operating characteristic (ROC) curves were generated and statistical analysis. Categorical data were reported as frequency detection rates (DRs) at fixed 5%, 10%, 15%, 20%, 25%, and 30% 156 | SERRANO et al. TA B L E 5 DR and AUC for prediction of early-­onset SGA by the Gaussian and the FMF algorithms SGA < 32 ± 0 weeks (n = 8) Gaussian algorithm DR at 5% FPR DR at 10% FPR DR at 15% FPR DR at 20% FPR DR at 25% DR at 30% FPR A priori risk + AUC (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) FPR (95% CI) (95% CI) MAP 0.700 12.5 (0.0–­37.5) 12.5 (0.0–­37.5) 37.5 (0.0–­75.0) 37.5 (12.5–­75.0) 62.5 62.5 (25.0–­87.8) (0.546–­0.854) (25.0–­87.8) MAP + PlGF 0.840 25.0 (0.0–­62.5) 37.5 (12.5–­75.0) 75.0 75.0 87.5 (62.5–­100) 87.5 (62.5–­100) (0.710–­0.970) (37.5–­100.0) (37.5–­100.0) MAP + UtAPI 0.740 25.0 (0.0–­62.5) 37.5 (12.5–­75.0) 37.5 (12.5–­75.0) 50.0 (12.5–­87.5) 62.5 62.5 (25.0–­100) (0.564–­0.916) (25.0–­87.8) MAP + PAPP-­A 0.743 12.5 (0.0–­37.5) 37.5 (1.6–­75.0) 37.5 (12.5–­75.0) 62.5 (25.0–­87.8) 62.5 (25.0–­100) 62.5 (25.0–­100) (0.581–­0.905) MAP + UtAPI + 0.757 25.0 (0.0–­62.5) 37.5 (0.0–­75.0) 50.0 (12.5–­87.5) 50.0 (12.5–­87.5) 50.0 75.0 (49.7–­100.0) PAPP-­A (0.589–­0.925) (12.5–­87.5) MAP + UtAPI + 0.811 37.5 (12.5–­75.0) 62.5 (25.5–­87.5) 75.0 75.0 75.0 75.0 (50.0–­100.0) PlGF (0.641–­0.982) (50.0–­100.0) (50.0–­100.0) (50.0–­100.0) MAP + UtAPI 0.806 37.5 (12.5–­75.0) 62.5 (25.5–­87.5) 75.0 75.0 75.0 75.0 (37.5–­100.0) + PlGF + (0.635–­0.978) (37.5–­100.0) (37.5–­100.0) (37.5–­100.0) PAPP-­A Note: Comparisons between areas under the curve (AUCs) were performed by two-­t ailed P values. Abbreviations: CI, confidence interval; DR, detection rate; FMF, Fetal Medicine Foundation; FPR, false-­positive rate; MAP, mean arterial pressure; PAPP-­A , pregnancy-­associated plasma protein A; PlGF, placental growth factor; SGA, small for gestational age; UtAPI, mean uterine artery pulsatility index. TA B L E 6 DR and AUC for prediction of preterm SGA by the Gaussian and the FMF algorithms SGA < 37 ± 0 weeks (n = 44) Gaussian algorithm DR at 5% FPR DR at 10% FPR DR at 15% DR at 20% DR at 25% DR at 30% FPR A priori risk + AUC (95% CI) (95% CI) (95% CI) FPR (95% CI) FPR (95% CI) FPR (95% CI) (95% CI) MAP 0.546 (0.459–­0.632) 9.1 (0.7–­18.2) 18.2 (6.8–­29.6) 22.7 22.7 29.6 36.4 (22.7–­50.0) (11.4–­36.4) (11.4–­36.4) (15.9–­43.2) MAP + PlGF 0.630 (0.540–­0.719) 9.1 (2.3–­20.5) 20.5 (9.1–­31.9) 38.2 43.2 45.5 52.3 (38.6–­65.9) (22.7–­52.3) (29.6–­56.8) (31.8–­61.4) MAP + UtAPI 0.653 (0.57–­0.737) 15.9 25.0 29.6 36.4 52.3 54.6 (40.9–­70.5) (6.8–­27.3) (13.6–­36.7) (18.2–­45.5) (22.7–­52.3) (36.4–­65.9) MAP + PAPP-­A 0.592 (0.505–­0.678) 7.9 (2.3–­18.2) 22.7 25.0 34.1 38.6 40.9 (27.3–­56.8) (11.4–­3 4.1) (13.6–­38.6) (22.5–­50.0) (25.0–­52.3) MAP + UtAPI + 0.670 (0.587–­0.752) 15.9 20.5 (9.1–­3 4.1) 29.5 43.2 52.3 61.4 (47.7–­75.0) PAPP-­A (6.8–­27.3) (15.9–­43.2) (29.6–­61.4) (36.4–­65.9) MAP + UtAPI + 0.697 (0.612–­0.782) 20.5 29.5 45.5 54.6 59.1 63.6 (47.7–­77.3) PlGF (9.1–­3 4.0) (15.9–­43.2) (31.8–­59.1) (38.6–­68.2) (43.2–­72.7) MAP + UtAPI 0.684 (0.598–­0.769) 18.2 29.5 43.2 52.3 54.6 61.4 (47.7–­75.0) + PlGF + (6.8–­29.6) (15.9–­43.2) (27.3–­59.1) (36.4–­65.9) (38.6–­70.5) PAPP-­A Note: Comparisons between areas under the curve (AUCs) were performed by two-­t ailed P values. Abbreviations: CI, confidence interval; DR, detection rate; FMF, Fetal Medicine Foundation; FPR, false-­positive rate; MAP, mean arterial pressure; PAPP-­A , pregnancy-­associated plasma protein A; PlGF, placental growth factor; SGA, small for gestational age; UtAPI, mean uterine artery pulsatility index. FPRs were calculated for both algorithms. The predictive accuracies by the Delong test. 20 Bonferroni correction was used in all tests of both algorithms were compared for a fixed FPR of 10% as well as when multiple comparisons were assessed. Statistical significance for the resulting areas under the curve (AUC), which were compared was set at P < 0.05. SERRANO et al. | 157 FMF algorithm DR at 5% FPR DR at 10% FPR DR at 15% FPR DR at 20% FPR DR at 25% FPR DR at 30% FPR AUC (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) P value 0.722 (0.604–­0.841) 12.5 (0.0–­37.5) 12.5 (0.0–­37.5) 12.5 (0.0–­37.5) 37.5 (12.5–­75.0) 62.5 (25.0–­87.8) 62.5 (25.0–­87.8) 0.4854 0.865 (0.784–­0.945) 37.5 (0.0–­75.0) 37.5 (12.5–­75.0) 50.0 (12.5–­87.5) 87.5 87.5 (62.5–­100) 87.5 (62.5–­100) 0.4625 (50.0–­100.0) 0.777 (0.655–­0.898) 25.0 (0.0–­62.5) 25.0 (0.0–­62.5) 37.5 (12.5–­75.0) 50.0 (12.5–­87.5) 62.5 (25.0–­87.8) 62.5 (25.0–­87.8) 0.4147 0.746 (0.619–­0.873) 12.5 (0.0–­37.5) 12.5 (0.0–­37.5) 37.5 (12.2–­75.0) 50.0 (12.5–­87.5) 62.5 (25.0–­87.8) 62.5 (25.0–­87.8) 0.9418 0.795 (0.663–­0.926) 25.0 (0.0–­62.5) 37.5 (0.0–­75.5) 50.0 (12.5–­87.5) 62.5 (25.0–­87.5) 62.5 (25.0–­87.5) 75.0 (37.5–­100) 0.4514 0.875 (0.774–­0.976) 62.5 (25.0–­87.5) 62.5 (25.0–­87.5) 62.5 (25.0–­87.5) 75.0 75.0 87.5 0.1289 (37.5–­100.0) (37.5–­100.0) (50.0–­100.0) 0.906 (0.834–­0.978) 50.0 (12.5–­87.5) 75.0 75.0 75.0 87.5 100 (100–­100) 0.0582 (37.5–­100.0) (50.0–­100.0) (50.0–­100.0) (62.5–­100.0) FMF algorithm DR at 5% FPR DR at 10% FPR DR at 15% FPR DR at 20% FPR DR at 25% FPR DR at 30% AUC (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) FPR (95% CI) P value 0.563 9.1 (2.3–­18.2) 13.6 (4.5–­25.0) 18.2 (9.1–­29.6) 27.3 34.1 36.4 0.4230 (0.477–­0.649) (15.9–­4 0.9) (20.5–­47.7) (22.7–­52.3) 0.651 13.6 (4.5–­25.0) 22.7 36.4 43.2 50.0 54.6 0.3766 (0.562–­0.739) (11.4–­36.4) (22.7–­52.3) (29.6–­59.1) (36.4–­65.9) (38.6–­68.2) 0.634 13.6 (4.5–­25.0) 20.5 (9.1–­3 4.1) 27.3 38.7 45.5 50.0 0.4437 (0.547–­0.722) (15.9–­43.2) (25.0–­54.6) (31.8–­59.2) (36.4–­65.9) 0.591 6.8 (0.0–­15.9) 11.4 (4.5–­22.7) 27.3 34.1 36.4 45.5 0.9680 (0.504–­0.677) (13.6–­4 0.9) (20.5–­47.7) (22.7–­50.0) (29.6–­61.4) 0.661 13.6 (4.3–­25.0) 25.0 34.1 40.9 47.7 56.8 0.7167 (0.575–­0.746) (11.4–­38.6) (20.5–­47.7) (25.0–­56.8) (34.1–­63.4) (40.9–­70.5) 0.689 18.2 (6.8–­31.8) 34.1 45.5 47.8 56.8 59.1 0.7524 (0.601–­0.776) (20.5–­50.0) (31.8–­59.2) (34.1–­63.6) (40.9–­70.5) (43.2–­72.7) 0.727 22.7 (11.4–­38.6) 40.9 47.7 52.3 63.6 68.2 0.1749 (0.645–­0.809) (25.0–­56.8) (34.1–­63.6) (36.4–­65.9) (49.9–­79.5) (54.6–­81.8) 3 | R E S U LT S newborns, 44 (1.77%) were preterm SGA, including 8 (0.32%) with early-­onset SGA. Among the 2641 participants, 30 (1.14%) women developed pre- Characteristics of the study population are summarized in term PE, including 11 (0.42%) with early-­onset PE. Among the 2483 Table 1 and Table 2. 158 | SERRANO et al. For prediction of early-­onset and preterm PE, and early-­onset The FMF algorithm has been externally validated by several and preterm SGA, the Gaussian and FMF algorithms showed a sim- studies in various populations, showing comparable performance ilar predictive performance with all marker combinations, except to that of the original study. Nevertheless, one study showed that for early-­onset PE prediction with MAP and PAPP-­A (Gaussian some algorithms could underperform when applied to popula- AUC = 0.833 [95% CI, 0.727–­0.939] vs FMF AUC = 0.771 [95% CI, tions that were different to the population where they were de- 0.631–­0.911]; P = 0.002), MAP and PlGF (Gaussian AUC = 0.905 veloped. 22 In this study, we show that performance of the FMF [95% CI, 0.844–­0.965] vs FMF AUC = 0.858 [95% CI, 0.768–­0.947]; algorithm in a Spanish population was similar to the performance P = 0.01), and MAP alone (Gaussian AUC = 0.795 [95% CI, 0.679–­ obtained in the original study, further supporting the external va- 0.912] vs FMF AUC = 0.758 [95% CI, 0.621–­0.895]; P = 0.02), where lidity of the FMF algorithm. By contrast, the predictive ability of the FMF algorithm showed a significantly lower AUC Tables 3–­6. the Gaussian algorithm has not been evaluated in other studies, For early-­onset PE prediction, the Gaussian algorithm showed aside from the original study where it was first validated. It must the greatest AUC when combining maternal history, MAP, UtAPI be noted that the Gaussian algorithm was not developed in our and PlGF (0.951; 95% CI, 0.919–­0.983), followed by the combina- population, but just validated, since this algorithm was constructed tion of all markers (0.945; 95% CI, 0.912–­0.979). The FMF algorithm using previously published data from a large meta-­analysis. This showed the greatest AUC when combining all markers (0.945; 95% might make this algorithm less likely to be overfitted to our pop- CI, 0.908–­0.982). ulation and, therefore, less likely to underperform when applied For preterm PE prediction, the Gaussian algorithm showed the to a different population. Since first-­t rimester PE screening and greatest AUC when combining maternal history, MAP and PlGF aspirin prescription has been implemented in most countries (0.802; 95% CI, 0.722–­0.881), followed by the combination of all across Europe, prospective external validation of the Gaussian markers without PAPP-­A (0.798; 95% CI, 0.704–­0.893). The FMF algorithm in untreated populations seems unlikely. Therefore, algorithm showed the greatest AUC when combining all markers a reasonable indirect approach to assess the predictive perfor- (0.818; 95% CI, 0.728–­0.907). mance of the Gaussian algorithm is to compare it with the FMF For early-­onset SGA prediction, the Gaussian algorithm showed algorithm, which has been extensively validated in various large the greatest AUC when combining maternal history, MAP and PlGF populations. Although our results cannot be considered an exter- (0.840; 95% CI, 0.710–­0.970), followed by the combination of all nal validation of the Gaussian algorithm, the similar accuracies of markers without PAPP-­A (0.811; 95% CI, 0.641–­0.982). The FMF both algorithms suggest that the FMF algorithm is unlikely to out- algorithm showed the greatest AUC when combining all markers perform the Gaussian algorithm in our population where it is being (0.906; 95% CI, 0.834–­0.978). routinely used in most maternities since 2018. For this reason, we For preterm SGA prediction, the Gaussian algorithm showed believe that the Gaussian algorithm might be a reasonable alter- the greatest AUC when combining maternal history, MAP, UtAPI, native to the FMF algorithm for those settings where the latter and PlGF (0.697; 95% CI, 0.612–­0.782), followed by the combina- cannot be applied because of ultrasonographers performing UtAPI tion of all markers (0.684; 95% CI, 0.598–­0.769). The FMF algorithm both transabdominally and transvaginally or for settings measuring showed the greatest AUC when combining all markers (0.727; 95% biomarkers for the aneuploidy and PE screenings before 11 weeks. CI, 0.645–­0.809). The results of this study are relevant since the Gaussian algorithm is already being implemented in other countries aside from Spain. Additionally, as seen in previous studies,23 we confirm that PAPP-­A 4 | DISCUSSION does not increase the predictive accuracy of any of the algorithms when PlGF was being used; however, when PlGF is not available, This study shows that the Gaussian and FMF algorithms have simi- PAPP-­A could increase DR by 5% with some marker combinations. lar predictive accuracies for PE and SGA, except for early-­onset PE, Finally, we observed that a single measurement of MAP could decrease where the FMF algorithm showed a significantly lower AUC with the the predictive accuracy of the FMF algorithm; therefore, the appro- combinations of MAP and PAPP-­A , MAP and PlGF, and MAP alone. priate methodology (the average of two measurements in both arms These significant differences could be partly attributed to the differ- simultaneously) should be performed when using this algorithm. ent methodology required for MAP assessment in both algorithms. In One of the main strengths of the study includes the prospec- this study, MAP was measured once in only one arm and after a 5-­min tive enrollment of patients. Furthermore, this study was performed rest, while the FMF algorithm was designed with an average of two within the context of routine clinical practice and patients were MAP measurements performed at 1-­min intervals in both arms simul- seen by their usual physicians, making the results more reliable taneously after a 5-­min rest.21 This different methodology for MAP and applicable in routine care settings. Moreover, this is the first measurements may have affected the accuracy of all combinations in- study assessing the performance of the FMF algorithm exclusively cluding MAP in the FMF algorithm, but especially MAP alone or those in a Spanish cohort and in a clinical setting where MAP was mea- combinations that included MAP with one other factor. sured once and only in one arm, showing similar results to those SERRANO et al. | 159 reported in the original study, for most combinations of markers. writing the article draft. All authors made substantial revisions to the Despite a previous study showing that prediction of PE is similar article. All authors read and approved the final article. when biomarkers are measured before or after 11 weeks,6 the FMF algorithm was designed with biomarkers assessed between 11 ± 0 AC K N OW L E D G M E N T S and 13 ± 6 weeks. In this study, biomarkers were measured before We thank María del Mar Jiménez Quesada for English-­language cor- 11 ± 0 weeks in 1675 (63.4%) women. Therefore, another remarkable rection of the article, all of the physicians who facilitated the recruit- strength of our work is that it provides evidence of the applicability ment of individuals at the Hospital Universitari Vall d'Hebron, and all of the FMF and Gaussian algorithms before and after 11 weeks for of the participants who agreed to take part in this study. predicting PE and SGA. The main limitation of our study is the low number of cases C O N FL I C T S O F I N T E R E S T with early-­o nset SGA and early-­o nset PE and the relatively low Manel Mendoza, MD, PhD, received lecture fees by Roche diagnos- number of cases with preterm SGA and preterm PE. Additionally, tics. The other authors report no conflicts of interest. indication for elective delivery of SGA fetuses based on Doppler and cardiotocogram findings may be different when using other DATA AVA I L A B I L I T Y S TAT E M E N T fetal growth restriction protocols. However, Doppler and cardio- The data that support the findings of this study are available from tocogram classification is uniform in Spain, where the Gaussian the corresponding author upon reasonable request. algorithm is widely used. Another limitation to be noted is that the technique for MAP measurements may potentially reduce the ORCID FMF algorithm's performance and could explain its lower AUC Berta Serrano https://orcid.org/0000-0002-1264-9195 versus the Gaussian algorithm for some marker combinations. Erika Bonacina https://orcid.org/0000-0002-8042-8247 Carlota Rodo https://orcid.org/0000-0001-8801-5677 Pablo Garcia-­Manau https://orcid.org/0000-0002-2415-1626 5 | CO N C LU S I O N S Mireia Armengol-­Alsina https://orcid.org/0000-0002-2087-5176 Manel Mendoza https://orcid.org/0000-0002-3030-3833 This study shows that the first-­trimester Gaussian and FMF algo- Elena Carreras https://orcid.org/0000-0003-3471-7248 rithms have similar predictive performances for PE and SGA in a Spanish population within a routine care setting. The accuracy of REFERENCES the FMF algorithm in our study was similar to that reported in previ- 1. Tan MY, Syngelaki A, Poon LC, et al. Screening for pre-­eclampsia ous studies, adding evidence to its external validity. by maternal factors and biomarkers at 11–­13 weeks' gestation. Ultrasound Obstet Gynecol. 2018;52:186-­195. AU T H O R C O N T R I B U T I O N S 2. Tan MY, Poon LC, Rolnik DL, et al. Prediction and prevention of Berta Serrano, MD; Erika Bonacina, MD; Pablo Garcia-­Manau, MD; small-­for-­gestational-­age neonates: evidence from SPREE and ASPRE. Ultrasound Obstet Gynecol. 2018;52:52-­59. Manel Mendoza, MD, PhD; and Elena Carreras, MD, PhD, had full 3. Serra B, Mendoza M, Scazzocchio E, et al. A new model for screening access to all of the data in the study and take full responsibility for for early-­onset preeclampsia. Am J Obstet Gynecol. 2020;222:608. the integrity of the data and accuracy of the data analysis. Berta e1-­608.e18. Serrano, MD; Erika Bonacina, MD; Pablo Garcia-­Manau, MD; Manel 4. Mendoza M, Serrano B, Bonacina E, et al. Diagnostic accuracy of the gaussian first-­trimester combined screening for pre-­eclampsia Mendoza, MD, PhD; and Elena Carreras, MD, PhD, conceived and to predict small-­for-­gestational-­age neonates. Int J Gynaecol Obstet. designed the study. Berta Serrano, MD; Erika Bonacina, MD; Carlota 2022;156:322-­330. Rodo, MD, PhD; Pablo Garcia-­Manau, MD; María Ángeles Sanchez-­ 5. Mendoza M, Tur H, Garcia-­Manau P, et al. Cut-­off values for gauss- Duran, MD, PhD; María Pancorbo, MD; Cristina Forcada, MD; María ian first-­trimester screening for early-­onset preeclampsia with ma- ternal history, biochemical markers and uterine artery doppler. J Teresa Murcia, MD; Ana Perestelo, MD; and Mireia Armengol-­Alsina, Gynecol Obstet Hum Reprod. 2021;50:101827. MD, contributed to literature research. Berta Serrano, MD; Erika 6. Mendoza M, Garcia-­Manau P, Arévalo S, et al. Diagnostic accuracy Bonacina, MD; Carlota Rodo, MD, PhD; Pablo Garcia-­Manau, MD; of first-­trimester combined screening for early-­onset and preterm María Ángeles Sanchez-­Duran, MD, PhD; María Pancorbo, MD; preeclampsia at 8–­10 weeks compared to 11–­13 weeks gestation. Ultrasound Obstet Gynecol. 2021;57:84-­90. Cristina Forcada, MD; María Teresa Murcia, MD; Ana Perestelo, 7. Bartsch E, Medcalf KE, Park AL, Ray JG. Clinical risk factors for pre-­ MD; and Mireia Armengol-­Alsina, MD, contributed to data collection eclampsia determined in early pregnancy: systematic review and and confirmation. Berta Serrano, MD; Erika Bonacina, MD; Pablo meta-­analysis of large cohort studies. BMJ. 2016;353:i1753. Garcia-­Manau, MD; and Manel Mendoza, MD, PhD, contributed to 8. Lobo GAR, Nowak PM, Panigassi AP, et al. Validation of Fetal data analysis. Berta Serrano, MD; Erika Bonacina, MD; Pablo Garcia-­ Medicine Foundation algorithm for prediction of pre-­eclampsia in the first trimester in an unselected Brazilian population. J Matern Manau, MD; Manel Mendoza, MD, PhD; and Elena Carreras, MD, Fetal Neonatal Med. 2019;32:286-­292. PhD, contributed to data interpretation. Berta Serrano, MD, Erika 9. O'Gorman N, Wright D, Poon LC, et al. Multicenter screening for Bonacina, MD; and Manel Mendoza, MD, PhD, were in charge of pre-­eclampsia by maternal factors and biomarkers at 11–­13 weeks' 160 | SERRANO et al. gestation: comparison with NICE guidelines and ACOG recommen- 18. The Fetal Medicine Foundation reasearch tools: calculation of dations. Ultrasound Obstet Gynecol. 2017;49:756-­760. MoMs. https://fetal​medic​ine.org/resea​rch/mom. [May 2019]. 10. Chaemsaithong P, Pooh RK, Zheng M, et al. Prospective evaluation 19. The Fetal Medicine Foundation reasearch tools: preeclampsia of screening performance of first-­trimester prediction models for batch risk assessment. https://fetal​medic​ine.org/resea​rch/peRisk. preterm preeclampsia in an Asian population. Am J Obstet Gynecol. [May 2019]. 2019;221:650.e1-­650.e16. 20. DeLong ER, DeLong DM, Clarke-­Pearson DL. Comparing the areas 11. Sonek J, Krantz D, Carmichael J, et al. First-­trimester screening under two or more correlated receiver operating characteristic for early and late preeclampsia using maternal characteristics, curves: a nonparametric approach. Biometrics. 1988;44:837-­8 45. biomarkers, and estimated placental volume. Am J Obstet Gynecol. 21. Poon LC, Shennan A, Hyett JA, et al. The International Federation 2018;218:126.e1-­126.e13. of Gynecology and Obstetrics (FIGO) initiative on pre-­eclampsia: a 12. Park FJ, Leung CHY, Poon LCY, Williams PF, Rothwell SJ, Hyett JA. pragmatic guide for first-­trimester screening and prevention. Int J Clinical evaluation of a first trimester algorithm predicting the risk Gynaecol Obstet. 2019;145(Suppl 1):1-­33. of hypertensive disease of pregnancy. Aust N Z J Obstet Gynaecol. 22. Oliveira N, Magder LS, Blitzer MG, Baschat AA. First-­trimester 2013;53:532-­539. prediction of pre-­eclampsia: external validity of algorithms in 13. Robinson HP, Fleming JE. A critical evaluation of sonar “crown-­rump a prospectively enrolled cohort. Ultrasound Obstet Gynecol. length” measurements. Br J Obstet Gynaecol. 1975;82:702-­710. 2014;44:279-­285. 14. Martin AM, Bindra R, Curcio P, Cicero S, Nicolaides KH. Screening 23. Mazer Zumaeta A, Wright A, Syngelaki A, Maritsa VA, Da Silva AB, for pre-­eclampsia and fetal growth restriction by uterine artery Nicolaides KH. Screening for pre-­eclampsia at 11–­13 weeks' ges- doppler at 11–­14 weeks of gestation. Ultrasound Obstet Gynecol. tation: Use of pregnancy-­associated plasma protein-­a, placental 2001;18:583-­586. growth factor or both. Ultrasound Obstet Gynecol. 2020;56:400-­4 07. 15. Carrascosa A, Yeste D, Copil A, Almar J, Salcedo S, Gussinyé M. Anthropometric growth patterns of preterm and full-­term newborns (24–­42 weeks' gestational age) at the hospital Materno-­Infantil Vall How to cite this article: Serrano B, Bonacina E, Rodo C, et al. d'Hebron (Barcelona)(1997–­2002). An Pediatr. 2004;60:406-­416. 16. Figueras F, Gratacos E. An integrated approach to fetal growth re- First-­trimester screening for pre-­eclampsia and small for striction. Best Pract Res Clin Obstet Gynaecol. 2017;38:48-­58. gestational age: A comparison of the Gaussian and Fetal 17. Brown MA, Magee LA, Kenny LC, et al. Hypertensive disor- Medicine Foundation algorithms. Int J Gynecol Obstet. ders of pregnancy: ISSHP classification, diagnosis, and manage- 2023;160:150-­160. doi: 10.1002/ijgo.14306 ment recommendations for international practice. Hypertension. 2018;72:24-­43.