OC107: Growth delay in the first trimester: Sensitivity and specificity of using serial measurements of embryonic growth for the prediction of early pregnancy loss

Cecilia Bottomley, Anneleen Daemen, F. Mukri, Aris T. Papageorghiou, E. Kirk, A. Pexsters, Bart De Moor, D. Timmerman, T. BourneUltrasound 2008 · DOI

Abstract

Early pregnancy loss (EPL) can be preceded by observation of crown rump length (CRL) smaller than expected for gestational age (GA). However a one off small CRL may also be explained by inaccurate menstrual dates. Longitudinal growth studies may better demonstrate true abnormal growth. Our objective was to determine sensitivity and specificity of FLDA (functional linear discriminant analysis) to assess longitudinal growth of CRL, mean gestational sac diameter (MSD) and mean yolk sac diameter (MYD) in prediction of EPL. Prospective cohort study of women with at least 2 ultrasound scans prior to 14w confirming singleton viability. To obtain FLDA performance in distinguishing pregnancies viable at 12w from EPL, a leave-one-out (LOO) cross-validation strategy was applied, resulting in sensitivity and specificity reflecting performance of FLDA on an independent validation set. Of 1078 pregnancies, 773 (71.7%) remained viable at 12 w (class 1) and 305 (28.3%) miscarried (class 2). Increase in CRL with GA was significantly lower for class 2 than class 1 (one-sample t-test, P=2.98e–22). Using rate of change of CRL, FLDA could predict EPL with sensitivity of 60.7% and specificity of 93.1% (PPV 33.3%, NPV 97.7%). Using rate of change of MSD gave sensitivity of 67.6% and specificity of 84.5% (PPV 50%, NPV 91.9%). MSD/CRL ratio gave sensitivity of 44.4% and specificity of 96.6% (PPV 70.6%, NPV of 90.3%). MSD-CRL difference and rate of change of MYD predicted EPL very poorly. There is a difference in rate of change of CRL, MSD and the ratio of the two values for viable pregnancies and those destined to miscarry. This approach can predict EPL with high specificity (up to 96.6%) before 12w gestation. The sensitivity is lower (44.4–67.6%) reflecting the proportion of class 2 patients (miscarriages) without antecedent growth retardation, suggesting two possible mechanisms for miscarriage—one associated with growth delay and another that is not.
BibTeX
@inproceedings{bottomley2008oc,
  author = {Cecilia Bottomley and Anneleen Daemen and F. Mukri and Aris T. Papageorghiou and E. Kirk and A. Pexsters and Bart De Moor and D. Timmerman and T. Bourne},
  title = {{OC107: Growth delay in the first trimester: Sensitivity and specificity of using serial measurements of embryonic growth for the prediction of early pregnancy loss}},
  booktitle = {Ultrasound},
  year = {2008},
  doi = {10.1002/uog.5515},
}

Loading related content...

Figure