Advertisement

Predicting first trimester pregnancy outcome: derivation of a multiple marker test

      Objective

      To predict first trimester pregnancy outcome using biomarkers in a multicenter cohort.

      Design

      Case-control study.

      Setting

      Three academic centers.

      Patient(s)

      Women with pain and bleeding in early pregnancy.

      Intervention(s)

      Sera from women who were 5–12 weeks' gestational age with ectopic pregnancy (EP), viable intrauterine pregnancy (IUP), and miscarriage/spontaneous abortion (SAB) was analyzed by ELISA and immunoassay for activin A, inhibin A, P, A Disintegrin And Metalloprotease-12, pregnancy-associated plasma protein A (PAPP-A), pregnancy specific B1-glycoprotein (SP1), placental-like growth factor, vascular endothelial growth factor, glycodelin (Glyc), and hCG. Classification trees were developed to optimize sensitivity/specificity for pregnancy location and viability.

      Main Outcome Measure(s)

      Area under receiver operating characteristic curve, sensitivity, specificity, and accuracy of first trimester pregnancy outcome.

      Result(s)

      In 230 pregnancies, the combination of trees to maximize sensitivity and specificity resulted in 73% specificity (95% confidence interval (CI) 0.65–0.80) and 31% sensitivity (95% CI 0.21–0.43) for viability. Similar methods had 21% sensitivity (95% CI 0.12–0.32) and 33% specificity (95% CI 0.26–0.41) for location. Activin A, Glyc, and A Disintegrin And Metalloprotease-12 definitively classified pregnancy location in 29% of the sample with 100% accuracy for EP. Progesterone and PAPP-A classified the viability in 61% of the sample with 94% accuracy.

      Conclusion(s)

      Multiple marker panels can distinguish pregnancy location and viability in a subset of women at risk for early pregnancy complications. This strategy of combining markers to maximize sensitivity and specificity results in high accuracy in a subset of subjects. Activin A, ADAM12, and Glyc are the most promising markers for pregnancy location; P and PAPP-A for viability.

      Key Words

      Discuss: You can discuss this article with its authors and with other ASRM members at https://www.fertstertdialog.com/users/16110-fertility-and-sterility/posts/13186-21452
      Vaginal bleeding and/or abdominal pain can occur in 25%–30% of viable intrauterine pregnancies (IUP) (
      • Hasan R.
      • Baird D.D.
      • Herring A.H.
      • Olshan A.F.
      • Jonsson Funk M.L.
      • Hartmann K.E.
      Patterns and predictors of vaginal bleeding in the first trimester of pregnancy.
      ). However, these symptoms are often present in failing first trimester IUPs/spontaneous abortions (SABs) as well as ectopic pregnancies (EPs). Ectopic pregnancy, defined as a pregnancy that implants outside the uterine endometrium, is of particular concern as it carries the risk of rupture that can lead to impaired future fertility as well as potentially life-threatening intra-abdominal hemorrhage (
      • Barnhart K.T.
      Clinical practice. Ectopic pregnancy.
      ). Ectopic pregnancies account for 1%–2% of all pregnancies, and although methods for detection and treatment have improved, resulting in an overall reduction in maternal mortality from 6% of all pregnancy-related deaths in the United States from 1990 to 1999 (
      • Chang J.
      • Elam-Evans L.D.
      • Berg C.J.
      • Herndon J.
      • Flowers L.
      • Seed K.A.
      • et al.
      Pregnancy-related mortality surveillance—United States, 1991–1999.
      ) to 0.50 deaths per 100,000 live births (
      • Creanga A.A.
      • Shapiro-Mendoza C.K.
      • Bish C.L.
      • Zane S.
      • Berg C.J.
      • Callaghan W.M.
      Trends in ectopic pregnancy mortality in the United States: 1980–2007.
      ), EP remains an important cause of first trimester maternal morbidity and mortality (
      • Hoover K.W.
      • Tao G.
      • Kent C.K.
      Trends in the diagnosis and treatment of ectopic pregnancy in the United States.
      ,
      • Crochet J.R.
      • Bastian L.A.
      • Chireau M.V.
      Does this woman have an ectopic pregnancy?: the rational clinical examination systematic review.
      ). As such, accurate and expeditious diagnosis of EP is critical to the diagnostic triage of symptomatic early pregnancy.
      Making the diagnosis of EP at the initial point of clinical care can be challenging due to the limitations of available diagnostic modalities. The current standard of care for the diagnosis of EP includes serial serum hCG levels and pelvic ultrasound (
      American College of Obstetrics, Gynecologists
      ACOG Practice Bulletin No. 94. Medical management of ectopic pregnancy.
      ). Unfortunately, ultrasound at first presentation is inconclusive in ≤40% of women because it has limited sensitivity, particularly when performed below the hCG discriminatory zone (
      • Barnhart K.T.
      • Simhan H.
      • Kamelle S.A.
      Diagnostic accuracy of ultrasound above and below the beta-hCG discriminatory zone.
      ). Pregnant patients with nondiagnostic ultrasounds are characterized as having a pregnancy of unknown location and require multiple clinical visits for blood tests, ultrasounds, and possibly surgical procedures before a definitive diagnosis can be made. Due diligence is imperative as ≤50% of patients initially presenting with a pregnancy of unknown location will go on to have an EP (
      • Barnhart K.T.
      • Simhan H.
      • Kamelle S.A.
      Diagnostic accuracy of ultrasound above and below the beta-hCG discriminatory zone.
      ,
      • Banerjee S.
      • Aslam N.
      • Zosmer N.
      • Woelfer B.
      • Jurkovic D.
      The expectant management of women with early pregnancy of unknown location.
      ).
      No single biomarker has been able to predict EP with sufficient and reproducible accuracy; thus, focus has now shifted toward development of a multiple marker test to capture the pathophysiological effects of the various biologic pathways involved in abnormal gestation. Prior studies identified a four-marker test of P, vascular endothelial growth factor (VEGF), inhibin A, and activin A to have 98% sensitivity and 100% specificity for identifying EP in a subset of patients with EP and IUP (
      • Rausch M.E.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • Barnhart K.T.
      Development of a multiple marker test for ectopic pregnancy.
      ). Other groups of investigators have suggested multiple marker tests including combinations of markers of inflammation and corpus luteum (CL) function (
      • Witt B.R.
      • Wolf G.C.
      • Wainwright C.J.
      • Johnston P.D.
      • Thorneycroft I.H.
      Relaxin, CA-125, progesterone, estradiol, Schwangerschaft protein, and human chorionic gonadotropin as predictors of outcome in threatened and nonthreatened pregnancies.
      ), or of trophoblast and CL function (
      • Mueller M.D.
      • Raio L.
      • Spoerri S.
      • Ghezzi F.
      • Dreher E.
      • Bersinger N.A.
      Novel placental and nonplacental serum markers in ectopic versus normal intrauterine pregnancy.
      ). In addition, a disintegrin and metalloprotease-12 (ADAM12) (
      • Beer L.A.
      • Tang H.Y.
      • Sriswasdi S.
      • Barnhart K.T.
      • Speicher D.W.
      Systematic discovery of ectopic pregnancy serum biomarkers using 3-D protein profiling coupled with label-free quantitation.
      ,
      • Rausch M.E.
      • Beer L.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • et al.
      A disintegrin and metalloprotease protein-12 as a novel marker for the diagnosis of ectopic pregnancy.
      ) and placental-like growth factor (
      • Horne A.W.
      • Shaw J.L.
      • Murdoch A.
      • McDonald S.E.
      • Williams A.R.
      • Jabbour H.N.
      • et al.
      Placental growth factor: a promising diagnostic biomarker for tubal ectopic pregnancy.
      ,
      • Daponte A.
      • Pournaras S.
      • Polyzos N.P.
      • Tsezou A.
      • Skentou H.
      • Anastasiadou F.
      • et al.
      Soluble FMS-like tyrosine kinase-1 (sFlt-1) and serum placental growth factor (PlGF) as biomarkers for ectopic pregnancy and missed abortion.
      ) have been suggested as potential biomarker candidates. However, these markers have yet to be assessed when miscarriage is included as a potential outcome of early pregnancy.
      To address the limitations of prior studies and assess the performance of these novel markers in combination, we sought to validate and optimize a multiple marker test for prediction of early pregnancy outcomes in a population that includes the three major outcomes of a symptomatic early pregnancy: IUP, SAB, and EP. Our first aim was to validate a four-marker test for prediction of EP that we previously derived (
      • Rausch M.E.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • Barnhart K.T.
      Development of a multiple marker test for ectopic pregnancy.
      ). Our second aim was to assess a multiple marker test incorporating two novel markers (placental-like growth factor and ADAM12). We hypothesized that marker performance would be synergized by optimizing sensitivity and specificity for pregnancy location (extrauterine (EUP) vs. IUP) and viability (IUP vs. early pregnancy failure).

      Materials and methods

       Study Design

      This is a nested case-control study of 230 women who presented to one of three academic centers (University of Pennsylvania, University of Miami, or University of Southern California) with symptomatic (pain and/or bleeding) early pregnancy, performed within an ongoing prospective cohort. Institutional Review Board approval was obtained at all study sites. Women were included in the initial prospective cohort if they met the following criteria (
      • Hasan R.
      • Baird D.D.
      • Herring A.H.
      • Olshan A.F.
      • Jonsson Funk M.L.
      • Hartmann K.E.
      Patterns and predictors of vaginal bleeding in the first trimester of pregnancy.
      ): complaints of abdominal pain, vaginal bleeding, or both (
      • Barnhart K.T.
      Clinical practice. Ectopic pregnancy.
      ), serum hCG of 100–20,000 mIU/mL (
      • Chang J.
      • Elam-Evans L.D.
      • Berg C.J.
      • Herndon J.
      • Flowers L.
      • Seed K.A.
      • et al.
      Pregnancy-related mortality surveillance—United States, 1991–1999.
      ), 5–12 weeks' gestational age by last menstrual period, AND (
      • Creanga A.A.
      • Shapiro-Mendoza C.K.
      • Bish C.L.
      • Zane S.
      • Berg C.J.
      • Callaghan W.M.
      Trends in ectopic pregnancy mortality in the United States: 1980–2007.
      ) agreement to participate in data and serum collection for the Ectopic Pregnancy Biomarkers Bank after informed consent. Women were excluded if (
      • Hasan R.
      • Baird D.D.
      • Herring A.H.
      • Olshan A.F.
      • Jonsson Funk M.L.
      • Hartmann K.E.
      Patterns and predictors of vaginal bleeding in the first trimester of pregnancy.
      ) they had received any treatment during the current pregnancy before enrollment (
      • Barnhart K.T.
      Clinical practice. Ectopic pregnancy.
      ), they had evidence of gestational trophoblastic disease (
      • Chang J.
      • Elam-Evans L.D.
      • Berg C.J.
      • Herndon J.
      • Flowers L.
      • Seed K.A.
      • et al.
      Pregnancy-related mortality surveillance—United States, 1991–1999.
      ), they were diagnosed with a nontubal EP, or (
      • Creanga A.A.
      • Shapiro-Mendoza C.K.
      • Bish C.L.
      • Zane S.
      • Berg C.J.
      • Callaghan W.M.
      Trends in ectopic pregnancy mortality in the United States: 1980–2007.
      ) there was evidence of multiple gestation.
      Subjects were followed prospectively until a definitive pregnancy outcome was established, as defined by internationally accepted nomenclature (
      • Barnhart K.
      • van Mello N.M.
      • Bourne T.
      • Kirk E.
      • van Calster B.
      • Bottomley C.
      • et al.
      Pregnancy of unknown location: a consensus statement of nomenclature, definitions, and outcome.
      ). Viable IUP was defined as ultrasound detection of an IUP (gestational sac, yolk sac, and fetal pole) with evidence of fetal cardiac activity. Spontaneous abortion or miscarriage was defined as evidence of products of conception on histopathology or if serum hCG declined to <5 mIU/mL. Ectopic pregnancy was defined as laparoscopic or ultrasound visualization of an adnexal mass without evidence of an IUP or increase in hCG level after uterine evacuation.
      The assignment of cases and controls was based on outcome grouping for pregnancy location and viability. For the pregnancy location analysis, or to discriminate an EP from intrauterine processes, EPs were considered the cases, whereas IUP and SABs were the controls. For the pregnancy viability analysis, or to discriminate a nonviable pregnancy from a viable intrauterine gestation, EP and SABs were considered the cases, whereas IUPs were the controls.
      Serum samples were collection from 2009 to 2012. An a priori sample size calculation was performed that focused on the precision for our estimate of sensitivity. Assumptions included a type 1 error rate of 2.5% (given our two comparisons (viability and location)), a desired confidence interval width of ±10%, and a minimum sensitivity of 85% for prediction of extrauterine location (EP). As such, 75 patients would be needed in each pregnancy outcome group.
      A case control design was selected given that EP is a rare disease, and to assure sufficient numbers for estimating sensitivity. As EP was the rarest outcome, SAB and IUP subject selection was frequency balanced for initial hCG and gestational age to that of EP. Definitive outcomes were known at the time of subject selection given the case-control design. Assays were run after study sample selection, thus selection was blind to biomarker profile.

       Data Collection and Biomarker Assays

      Maternal age, gestational age, race, ethnicity, study site, and initial hCG level were collected at the point of presentation from Quantbook v2.0 (Digital Ingenuity, Inc.), a prospective clinical data management system developed for the parent study and used at all three sites. Chart abstraction was used to validate pregnancy outcomes when necessary. Serum was collected at the point of initial presentation, centrifuged at 1,500 × g for 5 minutes, split into 0.5-mL aliquots, and stored at −80°C. Candidate biomarkers were selected for their role in biologic pathways involving trophoblast function, CL function, endometrial function, and angiogenesis on the basis of prior investigation from derivation studies of single markers and marker combinations. These markers included ADAM12, inhibin A, P, activin A, VEGF, pregnancy specific beta1-glycoprotein (SP1), pregnancy-associated plasma protein A (PAPP-A), glycodelin, and placental-like growth factor. Serum was shipped to the primary site of analysis and was stored until assays could be performed in batches. All assays were performed at the Clinical Translational Research Center Core Laboratory at the University of Pennsylvania. Assay characteristics are described in Supplemental Table 1, available online.

       Statistical Analysis

      Differences in baseline characteristics among the three outcome groups were assessed using Kruskal-Wallis for continuous measures, and Pearson's χ2 or Fisher's exact tests for categorical variables as appropriate. Individual biomarker discrimination of pregnancy outcome was assessed using area under the receiver-operating characteristic curve (AUC).
      The previous predictive model of EP was based on the combination of two decision trees (
      • Rausch M.E.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • Barnhart K.T.
      Development of a multiple marker test for ectopic pregnancy.
      ). One decision tree was developed to maximize the sensitivity of prediction, and the second to maximize specificity of prediction for the outcome of interest. To classify a subject as having EP, a given subject would have to be classified similarly by both trees—if one suggested EP and the other did not, the classification for this subject was deemed indeterminate and no prediction would be made. When results from both trees were combined, a prediction would be made in a subgroup of the population and the remaining subjects would be unclassified. Overall accuracy in those classified, and accuracy of each predicted outcome (i.e., EP or IUP) is then reported (
      • Rausch M.E.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • Barnhart K.T.
      Development of a multiple marker test for ectopic pregnancy.
      ,
      • Seeber B.
      • Sammel M.D.
      • Fan X.
      • Gerton G.L.
      • Shaunik A.
      • Chittams J.
      • et al.
      Panel of markers can accurately predict endometriosis in a subset of patients.
      ).
      To validate the previously reported four-marker test (VEGF, P, activin A, inhibin A) the decision tree that was derived to maximize sensitivity for EP (VEGF >28.24 pg/mL or VEGF ≤28.24 pg/mL + P≤12.25 ng/mL) and the decision tree to maximize specificity for EP (P≤6.1 ng/mL + inhibinA ≤8.83 pg/mL + activin A ≤408.65 pg/mL) were applied to data from the current study (
      • Rausch M.E.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • Barnhart K.T.
      Development of a multiple marker test for ectopic pregnancy.
      ).
      Information from each decision rule was translated into categorical variables for each marker and associations with outcome in the new dataset were estimated using logistic regression. Predicted probabilities from the model were used to evaluate these decision rules with respect to prediction of pregnancy location and viability (STATA v 12.0, StataCorp LP). Tree-structured nonparametric classification and regression analysis was used to evaluate discrimination of pregnancy location and viability for the nine putative and novel biomarkers in combination (CARTPro v 6.0, Salford Systems).
      We next developed two decision algorithms using all of our data (including our new markers): one for pregnancy location (EP vs. IUP + SAB) and one for pregnancy viability (IUP vs. EP + SAB) using similar logic and methods. Markers were assessed individually, and in combination, in separate decision trees to maximize sensitivity and specificity of prediction. The decision algorithms were then applied to all participants to assess the proportion of the total sample that was definitively classified with both algorithms. Sensitivity, specificity, and accuracy of prediction of each outcome (i.e., extrauterine location vs. intrauterine location or viability vs. nonviability) was calculated.
      Internal validation was performed by estimating test characteristics and precision estimates (95% confidence intervals (CI)) from 1,000 bootstrap samples drawn with replacement, assuming the same size as the original sample. The performance of our multiple marker panel was also assessed for important subgroups including those with gestational age <6 weeks, and those with hCG <2,000 mIU/mL.

      Results

       Baseline Characteristics

      Two hundred thirty subjects were included in the analysis. After outcome verification, there were 72 EPs, 77 IUPs, and 81 SABs. Baseline characteristics are reported in Supplemental Table 2, available online. Subjects were similar with respect to maternal age, race, ethnicity, and parity. There were no differences in the pregnancy outcome distribution by recruitment site (P=.10). Subjects with an IUP were more likely to be primigravid compared with those with an EP or SAB, among those for whom parity was reported. Those with IUPs presented with higher initial hCG levels compared with EPs and SABs (P=.0001). In addition, SABs presented at a later gestational age (median, 54 days in SAB vs. 42 days in EP and IUP; P=.0001). Although gestational age was unknown in 9% of the sample, the three outcome groups were no different in the distribution of subjects with unknown gestational age (P=.20).

       Validation of Four-Marker Test for EP

      When applying the four-marker test for prediction of EP (VEGF, P, activin A, and inhibin A) (
      • Rausch M.E.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • Barnhart K.T.
      Development of a multiple marker test for ectopic pregnancy.
      ), previously derived on a sample of only EPs and IUPs, to our current study sample including EPs, IUPs, and SABs, the test demonstrated 69% accuracy (95% CI 0.46–0.93) among those conclusively classified (15/230, 6.52% of the total cohort) with 83% sensitivity for prediction of EP (95% CI 0.36–0.99), 57% specificity for prediction of IUP (95% CI 0.21–0.86). However, due to low proportion of subjects conclusively classified, sensitivity and specificity of the test of the whole cohort was poor (sensitivity 6%, 95% CI 0.02–0.14; specificity 3%, 95% CI 0.01–0.07; predictive value of a positive test (PVP) 50%, 95% CI 0.16–0.84; predictive value of a negative test (PVN) 83%, 95% CI 0.36–0.99). When restricting the analyses to just subjects with EP or IUP (excluding those with a miscarriage), the four-marker test had a similar proportion of correct classification (11/149, 7.38%).
      To better characterize the differences observed in the validation samples from the current study, we assessed the accuracy of individual markers for prediction of EP. Although the accuracy of P and activin A were similar in the derivation and validation samples (P AUC of 0.81 vs. 0.86, P=.26; activin A AUC of 0.88 vs. 0.85, P=.62), results for VEGF and inhibin A were not replicated. Discrimination of the outcomes for VEGF and inhibin B were significantly higher in the derivation compared with the validation samples (AUC of 0.77 vs. 0.47, P=.001; AUC of 0.86 vs. 0.64, P=.01).

       Development of a Novel Strategy: Prediction of Early Pregnancy Outcome

      Each serum marker was evaluated individually with respect to prediction of pregnancy location (EP vs. SAB + IUP) and viability (IUP vs. EP + SAB). Distributions and receiver-operating characteristic curves for each marker are reported in Supplemental Table 3, available online, and Table 1. Glycodelin was the best individual marker of extrauterine location (AUC 0.61), whereas activin A was the best marker for intrauterine location (AUC 0.79). Progesterone was the best single marker of viability (AUC 0.88), whereas the best marker of a nonviable state was glycodelin (AUC 0.60).
      Table 1Single marker prediction of pregnancy location and viability, area under receiver-operating characteristic curve.
      Single markerPregnancy location AUC (95% CI)Pregnancy viability AUC (95% CI)
      Activin A0.79
      Marker predicts intrauterine location.
      (0.73–0.85)
      0.56 (0.49–0.64)
      PAPP-A0.61
      Marker predicts intrauterine location.
      (0.54–0.69)
      0.55
      Marker predicts nonviability.
      (0.47–0.62)
      Glycodelin0.61 (0.54–0.68)0.60
      Marker predicts nonviability.
      (0.52–0.68)
      hCG0.61
      Marker predicts intrauterine location.
      (0.53–0.68)
      0.76 (0.70–0.83)
      ADAM120.60
      Marker predicts intrauterine location.
      (0.52–0.69)
      0.51 (0.43–0.58)
      Inhibin A0.60
      Marker predicts intrauterine location.
      (0.53–0.68)
      0.74 (0.67–0.81)
      Progesterone0.58
      Marker predicts intrauterine location.
      (0.50–0.65)
      0.88 (0.83–0.93)
      SP10.51 (0.43–0.59)0.82 (0.76–0.87)
      PLGF0.52 (0.43–0.60)0.51
      Marker predicts nonviability.
      (0.43–0.60)
      VEGF0.52 (0.44–0.60)0.57 (0.49–0.65)
      Note: ADAM12 = a disintegrin and metalloprotease-12; AUC = area under the receiver-operating characteristic curve; CI = confidence interval; PAPP-A = pregnancy-associated plasma protein A; PLGF = placental growth factor; SP1 = pregnancy specific B1-glycoprotein; VEGF = vascular endothelial growth factor.
      a Marker predicts intrauterine location.
      b Marker predicts nonviability.
      Markers were next assessed in combination using a decision algorithm to maximize sensitivity and specificity for pregnancy location and viability. Ten markers were considered in the analysis. The resulting decision trees are presented in Figure 1, Figure 2.
      Figure thumbnail gr1
      Figure 1Sensitivity and specificity trees for pregnancy location. EP = ectopic pregnancy; Intrauterine-miscarriage (SAB) + viable intrauterine pregnancy (IUP).
      Figure thumbnail gr2
      Figure 2Sensitivity and specificity trees for pregnancy viability. IUP = viable intrauterine pregnancy; nonviable-miscarriage (SAB) + ectopic pregnancy (EP).

       Prediction of Pregnancy Location (EP vs. IUP + SAB)

      The tree that maximized sensitivity using activin A and glycodelin for pregnancy location had 100% sensitivity (95% CI 0.95–1.00) and 33% specificity (95% CI 0.26–0.41) when applied to the whole study sample (n = 230). The tree that maximized specificity for pregnancy location using activin A and ADAM12 had 21% sensitivity (95% CI 0.12–0.32) and 100% specificity (95% CI 0.98–1.00). When applying the decision rules for maximizing sensitivity and specificity for extrauterine location (EP) in combination as a single test for the whole study sample, 29% (n = 67) were conclusively classified and in the remaining 71% the test was inconclusive. Among those conclusively classified by the combined decision rules, 100% of the EPs were correctly predicted by activin A, glycodelin, and ADAM12 levels, and the non-EPs (SAB + IUP) were similarly predicted with 100% accuracy. When the test applied to the entire sample, it resulted in modest sensitivity (21%, 95% CI 0.12–0.32) and specificity (33%, 95% CI 0.26–0.41).

       Prediction of Pregnancy Viability (IUP vs. SAB + EP)

      The tree that maximized sensitivity for pregnancy viability using P had 96% sensitivity (95% CI 0.89–0.99) and 73% specificity (95% CI 0.66–0.80), whereas the tree that maximized specificity for pregnancy viability using P and PAPP-A had 31% sensitivity (95% CI 0.21–0.43) and 99% specificity (95% CI 0.96–0.99). When applying the decision rules for maximizing sensitivity and specificity for viability (IUP) in combination as a single test for the whole study sample, 61% of patients (n = 140) were conclusively classified, whereas the test was inconclusive in the remaining 90 patients. Among those conclusively classified, P and PAPP-A accurately predicted 89% of the viable IUPs and 99% of the nonviable gestations (SAB + EP). The test had a specificity of 73% (95% CI 0.65–0.80) and a sensitivity of 31% (95% CI 0.21–0.43) overall.

       Subgroup and Model Comparison Analyses

      Multiple marker tests were superior to hCG and gestational age for prediction of location and viability (P<.001, all tests). We also evaluated test performance in clinically important subgroups including women with a gestational age <6 weeks and those with hCG levels of <2,000 mIU/mL (Table 2) and noted similar trends in test characteristics. Multivariable logistic regression modeling demonstrated that addition of hCG to the other nine markers improved prediction of EP but not IUP (P=.03, P=.83), whereas adding gestational age did not significantly change prediction (P=.48 EP, P=.09 IUP).
      Table 2Test characteristics for prediction of pregnancy location and pregnancy viability with subgroups for hCG <2,000 mIU/mL and gestational age <6 weeks.
      SampleOutcomeSensitivity, % (95% CI)Specificity, % (95% CI)Conclusive classification, n (%)Accuracy among classified, % (95% CI)
      Total (n = 230)Location21 (0.12–0.32)33 (0.26–0.41)67/230 (29)100 (0.94–1.00)
      Viability31 (0.21–0.43)73 (0.65–0.80)140/230 (61)94 (0.88–1.00)
      hCG <2,000 mIU/mL (n = 112)Location18 (0.08–0.32)31 (0.21–0.44)29/112 (26)100 (0.88–1.00)
      Viability31 (0.11–0.59)84 (0.76–0.91)88/112 (79)0.91 (0.75–1.00)
      GA <6 wk (n = 80)Location29 (0.14–0.48)33 (0.20–0.48)25/80 (31)100 (0.86–1.00)
      Viability33 (0.19–0.51)50 (0.35–0.65)37/80 (46)91 (0.80–1.00)
      Note: CI = confidence interval; GA = gestational age.

      Discussion

      The search for a biomarker of EP has proven challenging. Although many markers have been proposed, at present none have been clinically validated. Our data also suggest that a promising panel of markers for the prediction of EP had substantially lower test characteristics than originally proposed (
      • Rausch M.E.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • Barnhart K.T.
      Development of a multiple marker test for ectopic pregnancy.
      ). This finding may be attributable to the inherent biases toward the null seen in validation studies (
      • Altman D.G.
      • Royston P.
      What do we mean by validating a prognostic model?.
      ,
      • Altman D.G.
      • Vergouwe Y.
      • Royston P.
      • Moons K.G.
      Prognosis and prognostic research: validating a prognostic model.
      ), it is likely that this original marker panel was predicting nonspecific early pregnancy failure (as opposed to pregnancy location). This would only become apparent when a third outcome of miscarriage is considered.
      The goal of this endeavor was to investigate a novel use of biomarkers to potentially supplement the current management of women with a possible miscarriage or EP. Although the optimal screening test has high sensitivity and specificity, our study demonstrates that maximizing one may come at a cost of reducing the other. As such, the test performances of these combined algorithms were limited when applied to the total sample. However, the cost of misdiagnosis must be considered. As the cost of misdiagnosis of interrupting and IUP or missing an EP are both grave, we chose to focus on accuracy of diagnosis rather than maximizing the number who could be classified, even if that meant improving accuracy in a subset of the total sample. To achieve this goal, the method of prediction of outcome in this (and previous) analyses uses two different decision trees (one maximizing sensitivity and the other specificity) simultaneously to aid in prediction. To predict an outcome, the prediction for each tree has to agree (i.e., predicted to be an EP by both the trees). This will result in splitting the sample into two groups—a group that will have a predicted outcome and one group will be classified as indeterminate or inconclusive.
      The advantage of combining these two trees is that one does not have to choose which diagnostic error is “worse” (missing an EP or interrupting a desired IUP), but instead can maximize the accuracy of those classified at the expense of classifying fewer subjects. Those who are not conclusively classified (i.e., considered as indeterminate) can be followed clinically with routine care, and thus misdiagnosis with the supplemental use of biomarkers is minimized. In addition, the subgroup classified by our new methodology as EP or IUP could proceed with earlier intervention or referral for obstetric management with confidence. A disadvantage of this approach is when the combined test is applied to the entire population, sensitivity and specificity are artificially low because those not conclusively classified (i.e., classified as indeterminate) are treated as if they are misclassified, even if they were classified correctly by either the “sensitivity” or the “specificity” tree but not both.
      Ideally, clinically useful markers for early pregnancy outcome should be able to distinguish pregnancy location and viability. To overcome this limitation we assessed the ability of multiple markers used concomitantly to discriminate between different two-way comparisons: assessing the markers' ability to assess location, as well as the markers' ability to assess viability of an early gestation. We found that multiple markers may optimize prediction of either location or viability.
      Activin A, glycodelin, and ADAM12 emerged as the most promising markers of pregnancy location (EUP vs. IUP), whereas P and PAPP-A had the best discrimination for viability. These markers reflect cell adhesion, endometrial function, CL and trophoblast function, and as such, it seems plausible that synergies and interactions among these pathways may enhance prediction.
      Activin A, a dimeric glycoprotein in the transforming growth factor-β superfamily, is a placental product that promotes cytotrophoblast invasion (
      • Bearfield C.
      • Jauniaux E.
      • Groome N.
      • Sargent I.L.
      • Muttukrishna S.
      The secretion and effect of inhibin A, activin A and follistatin on first-trimester trophoblasts in vitro.
      ). There have been conflicting results in prior studies applying activin A as a marker of EP. A study (
      • Florio P.
      • Severi F.M.
      • Bocchi C.
      • Luisi S.
      • Mazzini M.
      • Danero S.
      • et al.
      Single serum activin a testing to predict ectopic pregnancy.
      ) of patients with a pregnancy of unknown location found that an activin A level ≤0.37 ng/mL had 100% sensitivity and 99.6% specificity for the diagnosis of EP. However, other published reports (
      • Warrick J.
      • Gronowski A.
      • Moffett C.
      • Zhao Q.
      • Bishop E.
      • Woodworth A.
      Serum activin A does not predict ectopic pregnancy as a single measurement test, alone or as part of a multi-marker panel including progesterone and hCG.
      ,
      • Kirk E.
      • Papageorghiou A.T.
      • van Calster B.
      • Condous G.
      • Cowans N.
      • van Huffel S.
      • et al.
      The use of serum inhibin A and activin A levels in predicting the outcome of “pregnancies of unknown location”.
      ) have not demonstrated a difference in activin A levels when patients with resolving pregnancies of unknown locations or abnormal intrauterine gestations were included. Synthesizing these data supports a conclusion that activin A is a promising biomarker for EP (location) but may offer limited single-marker discrimination for pregnancy viability, as levels could be altered in abnormal intrauterine gestations.
      Glycodelin is an immunomodulatory protein involved in implantation (
      • Vigne J.L.
      • Hornung D.
      • Mueller M.D.
      • Taylor R.N.
      Purification and characterization of an immunomodulatory endometrial protein, glycodelin.
      ) and lower serum levels have been observed in patients with EP compared with incomplete abortions and viable IUPs (
      • Foth D.
      • Romer T.
      Glycodelin serum levels in women with ectopic pregnancy.
      ). Three groups of investigators (
      • Rausch M.E.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • Barnhart K.T.
      Development of a multiple marker test for ectopic pregnancy.
      ,
      • Mueller M.D.
      • Raio L.
      • Spoerri S.
      • Ghezzi F.
      • Dreher E.
      • Bersinger N.A.
      Novel placental and nonplacental serum markers in ectopic versus normal intrauterine pregnancy.
      ,
      • Daponte A.
      • Pournaras S.
      • Zintzaras E.
      • Kallitsaris A.
      • Lialios G.
      • Maniatis A.N.
      • et al.
      The value of a single combined measurement of VEGF, glycodelin, progesterone, PAPP-A, HPL and LIF for differentiating between ectopicand abnormal intrauterine pregnancy.
      ) have evaluated the performance of serum glycodelin in a multiple marker setting: two found that serum levels were significantly lower in patients with EP but underperformed as a diagnostic tool compared with other markers, whereas the third found no difference in levels between EP and abnormal intrauterine gestations.
      ADAM12 is a multidomain glycoprotein that is primarily produced by syncytiotrophoblasts and is involved in syncytial fusion and regulation of insulin-like growth factor (IGF) activity (
      • Shi Z.
      • Xu W.
      • Loechel F.
      • Wewer U.M.
      • Murphy L.J.
      ADAM 12, a disintegrin metalloprotease, interacts with insulin-like growth factor-binding protein-3.
      ). Prior studies have shown conflicting results with respect to the ability of ADAM12 to discriminate EP from intrauterine gestations—one (
      • Rausch M.E.
      • Beer L.
      • Sammel M.D.
      • Takacs P.
      • Chung K.
      • Shaunik A.
      • et al.
      A disintegrin and metalloprotease protein-12 as a novel marker for the diagnosis of ectopic pregnancy.
      ) reported a level of 48.49 ng/mL had 97% sensitivity and 37% specificity for EP, whereas another (
      • Horne A.W.
      • Brown J.K.
      • Tong S.
      • Kaitu'u-Lino T.
      Evaluation of ADAM-12 as a diagnostic biomarker of ectopic pregnancy in women with a pregnancy of unknown location.
      ) reported an accuracy of 66% in a population of pregnancies of unknown location. Low levels of ADAM12 have also been explored as a marker of aneuploidy and preeclampsia. As such, abnormal or even viable intrauterine gestations may have a similar biomarker phenotype, thus explaining the low specificity for extrauterine location. In the current study, the use of activin A, ADAM12, and glycodelin in combination was able to correctly classify 29% of pregnancies on the basis of location. Thus, although the contribution of these three markers for EP is promising, it appears that more than extrauterine location discrimination is necessary for diagnosis.
      The markers that emerged from this study as predictors of viability have known and established roles in the assessment of healthy pregnancy. As a product of the CL in early pregnancy, and later the placenta, P has been extensively evaluated as a predictor of EP and early pregnancy failure. A meta-analysis of 26 studies suggested that serum P levels of <5 ng/mL had good prediction for nonviable pregnancies, but was unable to differentiate EPs from abnormal intrauterine gestations (
      • Mol B.W.
      • Lijmer J.G.
      • Ankum W.M.
      • van der Veen F.
      • Bossuyt P.M.
      The accuracy of single serum progesterone measurement in the diagnosis of ectopic pregnancy: a meta-analysis.
      ). Pregnancy-associated plasma protein-A, produced by the synctiotrophoblast, is used clinically as a marker of aneuploidy, and has been studied as an indicator of early pregnancy failure. A study of patients including EP, SAB, and IUP demonstrated that PAPP-A levels <14.3 ng/mL had 64.5% sensitivity and 99% specificity for pregnancy failure, but noted no difference in levels between EP and abnormal intrauterine gestations (
      • Dumps P.
      • Meisser A.
      • Pons D.
      • Morales M.A.
      • Anguenot J.L.
      • Campana A.
      • et al.
      Accuracy of single measurements of pregnancy-associated plasma protein-A, human chorionic gonadotropin and progesterone in the diagnosis of early pregnancy failure.
      ). Two additional studies (
      • Daponte A.
      • Pournaras S.
      • Zintzaras E.
      • Kallitsaris A.
      • Lialios G.
      • Maniatis A.N.
      • et al.
      The value of a single combined measurement of VEGF, glycodelin, progesterone, PAPP-A, HPL and LIF for differentiating between ectopicand abnormal intrauterine pregnancy.
      ,
      • Ugurlu E.N.
      • Ozaksit G.
      • Karaer A.
      • Zulfikaroglu E.
      • Atalay A.
      • Ugur M.
      The value of vascular endothelial growth factor, pregnancy-associated plasma protein-A, and progesterone for early differentiation of ectopic pregnancies, normal intrauterine pregnancies, and spontaneous miscarriages.
      ) in different cohorts have confirmed that PAPP-A may help identify abnormal pregnancies. Thus, both PAPP-A and P have biologic plausibility as markers of viability independently, and in combination have superior performance to either marker alone. In the current study, these two markers were able to definitively classify viability in 61% of the sample with acceptable test characteristics. Placental-like growth factor was not a strong predictor of pregnancy outcome.
      The strengths of this study include concomitant assessment of multiple biomarkers at the same time in a relatively large, well-characterized group of subjects with documentation of all relevant potential outcomes of early pregnancy. Furthermore this study uses an analytic method that allows for exploration of synergies from multiple biologic pathways to create a tool that may be used in algorithms for clinical decision-making. As a multicenter study performed at widely different geographic locations in the United States, this is an appropriate cross-sectional representation of symptomatic, naturally conceived pregnancies, and thus highly generalizable.
      We acknowledge that this study is limited by its retrospective design and as such may be subject to information bias and the limitations of missing data. However, chart abstraction demonstrated no errors with respect to maternal age or ethnicity, and minimal errors with respect to gestational age (<5%). Missing data were limited to unknown gestational age in 9% of participants. The distribution of missing gestational age suggests that any potential misclassification bias would likely be not differential. Although there is theoretical risk of protein degradation in serum stored over time, should this be a real risk, it would likely bias results toward less discriminatory capacity, therefore the likelihood of overstating accuracy is minimal. All samples were run within 5 years of collection in batches; coefficients of variations (CVs) were acceptable with the exception of SP-1, suggesting minimal batch variation.
      Although these investigated markers are promising biomarkers, their use is not yet recommended for standard clinical care. Results need to be validated in a separate population with consideration for confounding by comorbidities and other clinical processes. Some pathophysiological conditions of pregnancy, such as hypertensive disorders and aneuploidy, may impact biomarker interpretation (
      • Forest J.C.
      • Charland M.
      • Masse J.
      • Bujold E.
      • Rousseau F.
      • Lafond J.
      • et al.
      Candidate biochemical markers for screening of pre-eclampsia in early pregnancy.
      ,
      • Spencer K.
      • Cowans N.J.
      ADAM12 as a marker of trisomy 18 in the first and second trimester of pregnancy.
      ,
      • Torring N.
      • Ball S.
      • Wright D.
      • Sarkissian G.
      • Guitton M.
      • Darbouret B.
      First trimester screening for trisomy 21 in gestational week 8-10 by ADAM12-S as a maternal serum marker.
      ). In addition, some markers, such as P, may have altered expression in a pregnancy resulting from assisted reproduction. Our study demonstrated that prediction was independent of gestational age. However, it is possible that having data from one additional point of care, such as additional serum data from another time point (i.e., 48 hours), may improve the proportion of subjects that could be conclusively classified, thus improving prediction. Any derived multiple marker test would also need to be validated in women with pregnancies of unknown location and further studies are needed to validate this novel strategy. The prediction of first trimester pregnancy outcome is clearly challenging and at this point, the complexity of algorithms that use multiple serum markers alone have limited utility in clinical practice. However, it is possible that refinement of these algorithms and possibly including ultrasound findings and other risk factors may improve prediction. As our understanding of early pregnancy pathophysiology continues to develop, so will our ability to harness this knowledge for clinical management in the first trimester.

      Appendix

      Supplemental Table 1Biomarker assay characteristics.
      BiomarkerMethod/vendorMinimum detectable limitInterassay coefficient of variation, %
      Placental- like growth factor (PIGF)ELISA/R&D System3.9 pg/mL9.8
      A disintegrin and metalloprotease -12 (ADAM12)ELISA/R&D System15 ng/mL8.5
      ProgesteroneImmunoassay/Siemens0.1 ng/mL4.9
      Activin AELISA/R&D System15 pg/mL6.4
      Inhibin AELISA/Beckman15 pg/mL5.7
      Pregnancy-associated plasma protein A (PAPPA)ELISA/R&D System0.15 ng/mL9.6
      Pregnancy specific B1-glycoprotein (SP1)ELISA/Cusabio1 μg/mL18.0
      GlycodelinELISA/Cusabio25 pg/mL10.8
      Vascular endothelial growth factor (VEGF)ELISA/R&D System10 pg/mL1.8
      Note: All intra-assay coefficients of variation <10%.
      Supplemental Table 2Baseline characteristics of study participants.
      CharacteristicsEP (n = 72)IUP (n = 77)SAB (n = 81)P value
      Maternal age (y)30 (18–44)26 (15–43)29 (15–46).06
      Kruskal-Wallis test.
      Hispanic34/72 (49)46/77 (61)51/81 (65).12
      χ2 test.
      White34/72 (47)37/77 (48)37/81 (46).11
      χ2 test.
      Black31/72 (43)23/77 (30)24/81 (30)
      Other7/72 (10)17/77 (22)20/81 (25)
      Primigravid7/41 (17)14/32(44)8/36 (22).03
      χ2 test.
      Nulliparous13/41 (32)17/32 (53)19/36 (53).10
      χ2 test.
      hCG level (mIU/mL)823 (120–18,733)4,271 (124–19,895)632 (102–17,714).0001
      Gestational age (d)42 (18–82)42 (30–75)54 (26–92).0001
      Kruskal-Wallis test.
      Missing gestational age5/72 (7)8/77 (10)8/81 (10).20
      Fisher's exact test.
      Note: Data presented as median (interquartile range) or n (%). EP = ectopic pregnancy; IUP = viable intrauterine pregnancy; SAB = spontaneous abortion.
      a Kruskal-Wallis test.
      b χ2 test.
      c Fisher's exact test.
      Supplemental Table 3Biomarker distribution by outcome.
      BiomarkerEP (n = 72)IUP (n = 77)SAB (n = 81)P value
      PLGF11.0 (9.3–13.7)11.4 (9.1–12.9)11.2 (10.1–12.7).91
      ADAM120.43 (0.26–0.76)0.53 (0.36–0.82)0.59 (0.39–0.87).02
      Progesterone5.0 (2.9–13.5)21.0 (15.0–24.7)2.6 (1.5–6.4).0001
      Activin A206.8 (161.4–254.1)289.3 (225.4–365.8)325.9 (265.6–460.3).0001
      Inhibin A24.2 (15.3–38.7)39.7 (28.7–66.3)24.5 (14.8–33.0).0001
      PAPP-A0.18 (0.13–0.24)0.20 (0.15–0.27)0.26 (0.17–0.34).0004
      SP173.6 (43.1–116.9)119.6 (88.7–177.4)41.3 (22.2–64.9).0001
      Glycodelin0.32 (0.22–0.60)0.2 (0.1–0.41)0.26 (0.1–0.86).008
      VEGF194.7 (90.9–400.9)220 (114.1–411.2)147.3 (55.9–310.4).08
      Note: Data presented as median (interquartile range). ADAM12 = a disintegrin and metalloprotease-12; EP = ectopic pregnancy; IUP = viable intrauterine pregnancy; PAPP-A = pregnancy-associated plasma protein A; PLGF = placental growth factor; SAB = spontaneous abortion; SP1 = pregnancy specific B1-glycoprotein; VEGF = vascular endothelial growth factor.

      References

        • Hasan R.
        • Baird D.D.
        • Herring A.H.
        • Olshan A.F.
        • Jonsson Funk M.L.
        • Hartmann K.E.
        Patterns and predictors of vaginal bleeding in the first trimester of pregnancy.
        Ann Epidemiol. 2010; 20: 524-531
        • Barnhart K.T.
        Clinical practice. Ectopic pregnancy.
        N Engl J Med. 2009; 361: 379-387
        • Chang J.
        • Elam-Evans L.D.
        • Berg C.J.
        • Herndon J.
        • Flowers L.
        • Seed K.A.
        • et al.
        Pregnancy-related mortality surveillance—United States, 1991–1999.
        MMWR Surveill Summ. 2003; 52: 1-8
        • Creanga A.A.
        • Shapiro-Mendoza C.K.
        • Bish C.L.
        • Zane S.
        • Berg C.J.
        • Callaghan W.M.
        Trends in ectopic pregnancy mortality in the United States: 1980–2007.
        Obstet Gynecol. 2011; 117: 837-843
        • Hoover K.W.
        • Tao G.
        • Kent C.K.
        Trends in the diagnosis and treatment of ectopic pregnancy in the United States.
        Obstet Gynecol. 2010; 115: 495-502
        • Crochet J.R.
        • Bastian L.A.
        • Chireau M.V.
        Does this woman have an ectopic pregnancy?: the rational clinical examination systematic review.
        JAMA. 2013; 309: 1722-1729
        • American College of Obstetrics, Gynecologists
        ACOG Practice Bulletin No. 94. Medical management of ectopic pregnancy.
        Obstet Gynecol. 2008; 111: 1479-1485
        • Barnhart K.T.
        • Simhan H.
        • Kamelle S.A.
        Diagnostic accuracy of ultrasound above and below the beta-hCG discriminatory zone.
        Obstet Gynecol. 1999; 94: 583-587
        • Banerjee S.
        • Aslam N.
        • Zosmer N.
        • Woelfer B.
        • Jurkovic D.
        The expectant management of women with early pregnancy of unknown location.
        Ultrasound Obstet Gynecol. 1999; 14: 231-236
        • Rausch M.E.
        • Sammel M.D.
        • Takacs P.
        • Chung K.
        • Shaunik A.
        • Barnhart K.T.
        Development of a multiple marker test for ectopic pregnancy.
        Obstet Gynecol. 2011; 117: 573-582
        • Witt B.R.
        • Wolf G.C.
        • Wainwright C.J.
        • Johnston P.D.
        • Thorneycroft I.H.
        Relaxin, CA-125, progesterone, estradiol, Schwangerschaft protein, and human chorionic gonadotropin as predictors of outcome in threatened and nonthreatened pregnancies.
        Fertil Steril. 1990; 53: 1029-1036
        • Mueller M.D.
        • Raio L.
        • Spoerri S.
        • Ghezzi F.
        • Dreher E.
        • Bersinger N.A.
        Novel placental and nonplacental serum markers in ectopic versus normal intrauterine pregnancy.
        Fertil Steril. 2004; 81: 1106-1111
        • Beer L.A.
        • Tang H.Y.
        • Sriswasdi S.
        • Barnhart K.T.
        • Speicher D.W.
        Systematic discovery of ectopic pregnancy serum biomarkers using 3-D protein profiling coupled with label-free quantitation.
        J Proteome Res. 2011; 10: 1126-1138
        • Rausch M.E.
        • Beer L.
        • Sammel M.D.
        • Takacs P.
        • Chung K.
        • Shaunik A.
        • et al.
        A disintegrin and metalloprotease protein-12 as a novel marker for the diagnosis of ectopic pregnancy.
        Fertil Steril. 2011; 95: 1373-1378
        • Horne A.W.
        • Shaw J.L.
        • Murdoch A.
        • McDonald S.E.
        • Williams A.R.
        • Jabbour H.N.
        • et al.
        Placental growth factor: a promising diagnostic biomarker for tubal ectopic pregnancy.
        J Clin Endocrinol Metab. 2011; 96: E104-E108
        • Daponte A.
        • Pournaras S.
        • Polyzos N.P.
        • Tsezou A.
        • Skentou H.
        • Anastasiadou F.
        • et al.
        Soluble FMS-like tyrosine kinase-1 (sFlt-1) and serum placental growth factor (PlGF) as biomarkers for ectopic pregnancy and missed abortion.
        J Clin Endocrinol Metab. 2011; 96: E1444-E1451
        • Barnhart K.
        • van Mello N.M.
        • Bourne T.
        • Kirk E.
        • van Calster B.
        • Bottomley C.
        • et al.
        Pregnancy of unknown location: a consensus statement of nomenclature, definitions, and outcome.
        Fertil Steril. 2011; 95: 857-866
        • Seeber B.
        • Sammel M.D.
        • Fan X.
        • Gerton G.L.
        • Shaunik A.
        • Chittams J.
        • et al.
        Panel of markers can accurately predict endometriosis in a subset of patients.
        Fertil Steril. 2008; 89: 1073-1081
        • Altman D.G.
        • Royston P.
        What do we mean by validating a prognostic model?.
        Stat Med. 2000; 19: 453-473
        • Altman D.G.
        • Vergouwe Y.
        • Royston P.
        • Moons K.G.
        Prognosis and prognostic research: validating a prognostic model.
        BMJ. 2009; 338: b605
        • Bearfield C.
        • Jauniaux E.
        • Groome N.
        • Sargent I.L.
        • Muttukrishna S.
        The secretion and effect of inhibin A, activin A and follistatin on first-trimester trophoblasts in vitro.
        Eur J Endocrinol. 2005; 152: 909-916
        • Florio P.
        • Severi F.M.
        • Bocchi C.
        • Luisi S.
        • Mazzini M.
        • Danero S.
        • et al.
        Single serum activin a testing to predict ectopic pregnancy.
        J Clin Endocrinol Metab. 2007; 92: 1748-1753
        • Warrick J.
        • Gronowski A.
        • Moffett C.
        • Zhao Q.
        • Bishop E.
        • Woodworth A.
        Serum activin A does not predict ectopic pregnancy as a single measurement test, alone or as part of a multi-marker panel including progesterone and hCG.
        Clin Chimica Acta. 2012; 413: 707-711
        • Kirk E.
        • Papageorghiou A.T.
        • van Calster B.
        • Condous G.
        • Cowans N.
        • van Huffel S.
        • et al.
        The use of serum inhibin A and activin A levels in predicting the outcome of “pregnancies of unknown location”.
        Hum Reprod. 2009; 24: 2451-2456
        • Vigne J.L.
        • Hornung D.
        • Mueller M.D.
        • Taylor R.N.
        Purification and characterization of an immunomodulatory endometrial protein, glycodelin.
        J Biol Chem. 2001; 276: 17101-17105
        • Foth D.
        • Romer T.
        Glycodelin serum levels in women with ectopic pregnancy.
        Eur J Obstet Gynecol Reprod Biol. 2003; 108: 199-202
        • Daponte A.
        • Pournaras S.
        • Zintzaras E.
        • Kallitsaris A.
        • Lialios G.
        • Maniatis A.N.
        • et al.
        The value of a single combined measurement of VEGF, glycodelin, progesterone, PAPP-A, HPL and LIF for differentiating between ectopicand abnormal intrauterine pregnancy.
        Hum Reprod. 2005; 20: 3163-3166
        • Shi Z.
        • Xu W.
        • Loechel F.
        • Wewer U.M.
        • Murphy L.J.
        ADAM 12, a disintegrin metalloprotease, interacts with insulin-like growth factor-binding protein-3.
        J Biol Chem. 2000; 275: 18574-18580
        • Horne A.W.
        • Brown J.K.
        • Tong S.
        • Kaitu'u-Lino T.
        Evaluation of ADAM-12 as a diagnostic biomarker of ectopic pregnancy in women with a pregnancy of unknown location.
        PloS One. 2012; 7: e41442
        • Mol B.W.
        • Lijmer J.G.
        • Ankum W.M.
        • van der Veen F.
        • Bossuyt P.M.
        The accuracy of single serum progesterone measurement in the diagnosis of ectopic pregnancy: a meta-analysis.
        Hum Reprod. 1998; 13: 3220-3227
        • Dumps P.
        • Meisser A.
        • Pons D.
        • Morales M.A.
        • Anguenot J.L.
        • Campana A.
        • et al.
        Accuracy of single measurements of pregnancy-associated plasma protein-A, human chorionic gonadotropin and progesterone in the diagnosis of early pregnancy failure.
        Eur J Obstet Gynecol Reprod Biol. 2002; 100: 174-180
        • Ugurlu E.N.
        • Ozaksit G.
        • Karaer A.
        • Zulfikaroglu E.
        • Atalay A.
        • Ugur M.
        The value of vascular endothelial growth factor, pregnancy-associated plasma protein-A, and progesterone for early differentiation of ectopic pregnancies, normal intrauterine pregnancies, and spontaneous miscarriages.
        Fertil Steril. 2009; 91: 1657-1661
        • Forest J.C.
        • Charland M.
        • Masse J.
        • Bujold E.
        • Rousseau F.
        • Lafond J.
        • et al.
        Candidate biochemical markers for screening of pre-eclampsia in early pregnancy.
        Clin Chem Lab Med. 2012; 50: 973-984
        • Spencer K.
        • Cowans N.J.
        ADAM12 as a marker of trisomy 18 in the first and second trimester of pregnancy.
        J Mater-Fetal Neonat Med. 2007; 20: 645-650
        • Torring N.
        • Ball S.
        • Wright D.
        • Sarkissian G.
        • Guitton M.
        • Darbouret B.
        First trimester screening for trisomy 21 in gestational week 8-10 by ADAM12-S as a maternal serum marker.
        Reprod Biol Endocrinol. 2010; 8: 129