Image - Barts and The London logo and link to home page Image - divider Image - divider
 
  Image - Wolfson Institute of Preventive Medicine title and link to Institute home page
Images - preventive medicine, people
  Image - Centre for Environmental and Preventive Medicine title and link to Centre homepage
 
Preliminary results of selected work in progress on Antenatal Screening
  1. Screening in early pregnancy for pre-eclampsia using markers from the Down’s syndrome Quadruple test
  2. Cross trimester marker ratios in antenatal screening for Down’s syndrome
  3. The Application of Multivariate Truncation Limits in Screening for Down’s Syndrome

1. Screening in early pregnancy for pre-eclampsia using markers from the Down’s syndrome Quadruple test
NJ Wald, JK Morris, J Ibison, T Wu, LM George

Computerized obstetric records of women attending for their antenatal care at three East London Hospitals from January 1989 to December 1997 have been linked to the Down’s syndrome screening database at the Wolfson Institute of Preventive Medicine. This record linkage generated 245 cases in which a caesarean section or induction of labour was performed and there was mention of pre-eclampsia in the computerized obstetric record. Manual review of the paper records identified 96 women with a diagnosis of pre-eclampsia that could be confirmed with confidence and for each case we have identified five unaffected pregnancies.

The antenatal serum samples collected between 15 and 20 weeks of gestation (median 16 weeks) have been retrieved from storage and tested for AFP, uE3, hCG, free β-hCG, inhibin-A. Multivariate Gaussian modelling is being used to estimate screening performance in terms of detection rate for a given false-positive rate (a method used previously in connection with antenatal screening for Down’s syndrome).

In the pregnancies with pre-eclampsia inhibin-A and hCG values were significantly raised and uE3 values were significantly lowered, AFP values were not significantly altered. Using the Quadruple test markers (AFP, uE3, hCG [total or free β] and inhibin-A), and a past history of pre-eclampsia, an estimated 41% of pregnancies that developed pre-eclampsia were detected at a 66% false-positive rate. This would offer a very low cost simple method of screening among women already having an antenatal screening test for Down’s syndrome.

 

Top

2. Cross trimester marker ratios in antenatal screening for Down’s syndrome
NJ Wald, J Bestwick, JK Morris

Wright and Bradbury (Wright DE, Bradbury I. Repeated measures screening for Down’s Syndrome. BJOG 2005;112:80-83) proposed the use of the same serum markers in both the first and second trimesters of pregnancy to calculate a woman’s risk of having a Down’s syndrome pregnancy. Their modelling produced better screening performance than would be obtained using the Integrated test. Their analysis was based on using the means, standard deviations and correlation coefficients originally given in the SURUSS report.

We sought to replicate their work and add repeat measures to the Integrated test markers. However the method fails when adding repeat PAPP-A and uE3 measures to the Integrated test markers and the risks of having a Down’s syndrome pregnancy can not be obtained. This is due to the Down’s syndrome correlation matrix having a negative determinant with this and other combination of markers. The modelling requires taking the square root of the determinant, which is not possible if it is negative. A negative determinant for a correlation matrix is more likely to arise when markers are highly correlated. As one would expect, a given analyte measured in the first trimester is highly correlated with the same analyte measured in the second trimester.

We also tested the repeated measures method on our database of MoM values for Down’s syndrome confirmed pregnancies. The high correlation co-efficients caused the calculated risks to be implausibly low, (say, less than 1 in a million) for some of these women.

These problems prompted us to seek an alternative method that would not fail. The discriminating difference between pregnancies with and without Down’s syndrome was probably the difference in the pattern of change in the concentration of a given marker across the first and second trimesters. We therefore used the ratios of the same marker in the first and second trimesters (the cross trimester (CT) ratio) together with the usual marker in the trimester when it is most discriminatory. The SURUSS data were used to calculate an individual woman’s ratio of the same marker (second trimester MoM divided by first trimester MoM) and means, standard deviations and correlation co-efficients were estimated for affected and unaffected pregnancies. As expected, any of the serum markers when they are most discriminatory are less correlated with the ratio of the first and second trimester markers than the marker in the first trimester and the marker in the second trimester. For combinations of markers for which the modelling failed using the method of Wright and Bradbury, the equivalent combinations of markers with CT ratios work successfully. The use of CT ratios for uE3, hCG, PAPP-A and inhibin-A significantly improve screening performance. For example, at a 90% detection rate the false-positive rate is 0.72% with the Integrated test using CT ratios, compared with 2.15% without CT ratios, a reduction of about two-thirds. With the serum Integrated test the corresponding false-positive rates are 2.9% and 8.1%, a similar proportional reduction. The risk estimates applied to a panel of pregnancies with known outcomes are all reasonable.


Initially posted 3rd November 2005
This work was completed and published in
Prenatal Diagnosis (Cross-trimester marker ratios in prenatal screening for Down syndrome, NJ Wald, J Bestwick, JK Morris, 2006; 26: 514-523)

Top

 

3. The Application of Multivariate Truncation Limits in Screening for Down’s Syndrome
J Bestwick and NJ Wald

Conventional truncation limits utilized in multivariate Guassian models from which the risk of a Down’s syndrome pregnancy is estimated are chosen univariately. They are estimated from a separate probability plot of each marker. The limits are chosen for a variety of reasons, particularly where beyond these points, the data are no longer Guassian. This method does not work with repeated measurements of the same marker in the first and second trimesters of pregnancy. The cross trimester (CT) marker ratio overcomes the problem. An alternative approach is to use multivariate truncation limits. The following steps show how such truncation would be performed in the simple case where the screening test consists of the measurement of one marker in the first and second trimesters of pregnancy. This example (please see the PDF version for the figures) uses the parameters for first and second trimester PAPP-A given in the SURUSS report.

 

Initiallly posted 19 January, 2007
 
Top
 
by Wolfson Institute Website Manager © Queen Mary, University of London 2011
Centre for Environmental and Preventive Medicine, Wolfson Institute of Preventive Medicine, Barts and The London, Charterhouse Square, London EC1M 6BQ Tel: 020 7882 3850 Fax: 020 7882 6270