Sequential testing strategies in prenatal screening for down's syndrome.
PhD thesis, University of Glasgow.
Full text available as:
Down’s syndrome is a classic chromosomal disorder with an incidence rate of one in every 750 live births. Early detection of Down’s syndrome pregnancies through screening will provide the option of early termination of pregnancy and better obstetric care to women with affected pregnancies. Some of the screening policies which have been implemented in the UK are second trimester double, triple or quadruple marker tests, first trimester combined ultrasound and biochemical (CUB) screening, and integrated screening. Screening performance can be optimized by applying appropriate correction factors for variables such as maternal smoking, ethnicity and assisted conception. Typical screening performance is around 70% detection of Down’s syndrome pregnancies at a 5% false positive rate for second trimester quadruple marker screening, 90% detection at a 5% false positive rate for CUB screening and 90% detection at a 1-2% false positive rate for integrated screening. The NHS Fetal Anomaly Screening Programme Committee has set a current performance target for Down’s syndrome screening of at least 75% detection at a 3% or lower false positive rate and this can be achieved by CUB or integrated testing by setting a threshold (cut-off) risk of 1 in 150 at term. However, further improvements in performance proposed by the Committee to meet a detection rate of 90% at a false positive rate of 2% or less are unlikely to be reached by single stage testing, and protocols which include some element of sequential testing are required. The Health Technology Assessment Programme is currently reviewing two new approaches to screening, namely, repeated measure and cross trimester testing to evaluate their potential to meet the more challenging standard.
In the present study, using various combinations of maternal serum marker and ultrasound measurements, several screening strategies and refinements are explored to establish their potential for improving detection rates and reducing false positive rates in Down’s syndrome screening. Extensive use has been made of routinely collected screening data from the west of Scotland Regional Screening programme for retrospective analysis using standard Gaussian methods, statistical modeling and SPSS and S-PLUS statistical software. The performance of within- and across-trimester contingent screening programmes have been evaluated and the effects of ethnicity, maternal smoking habit and assisted reproductive technology (ART) on screening markers has been assessed using first and second trimester samples.
Screening within the first trimester
The standard approach to CUB screening is to carry out maternal serum marker measurements (PAPP-A and fβhCG) and ultrasound Nuchal Translucency measurements at 11-13+6 weeks of gestation. This study had also shown that in the CUB screened population in the west of Scotland, adopting a within-trimester contingent screening protocol where all women have serum marker testing but only those women with intermediate risks from the serum markers are offered NT, would have achieved a detection rate of 88.7% at a false positive rate of 5.8% with 29% of women requiring an NT measurement. Using LMP based gestational age this screening protocol would have achieved a detection rate of 83.3% at a false positive rate of 7.4% with 25.9% of women requiring an NT measurement. When analysis was performed only on pregnancies with certain LMP dates, the contingent screening protocol would have achieved a detection rate of 88.9% at a false positive rate of 7.0% with 25.3% of women requiring an NT measurement. Where ultrasound resources are scarce within-trimester contingent screening has the potential to maintain screening performance whilst reducing the number of NT scans required.
Across –trimester screening
Evidence suggests that sequential testing strategies can improve screening performance. This has been explored in this study by statistical modelling using S-PLUS. Various combinations of markers were tested. It was estimated that optimal performance could be achieved by a cross-trimester contingent screening protocol with repeat measures of PAPP-A (NT, PAPP-A, fβhCG in the first trimester followed by AFP, hCG, InhA, uE3, PAPP-A in the second trimester in a sub-set of women with intermediate risks). This could achieve a detection rate of 92.2% at a false positive rate of 1.4% but with only 9.7% of women requiring a second trimester screening test. This meets the aspirational performance standard proposed by the UK NSC. Without NT measurements (i.e. serum only screening), the model indicates that this screening protocol would achieve a detection rate of 86.2% at a false positive rate of 3.0% with 22.3% of women requiring a second trimester screening test. Therefore, the inclusion of NT measurement at the first stage of testing is necessary to achieve the desired performance.
The Effects of Smoking and Ethnicity
Many maternal and pregnancy factors are known to affect serum marker concentrations and small but useful improvements in screening performance can be made by correcting for these. Changes however, vary between trimesters and in this study paired first and second trimester samples have been used to measure the changes in serum marker levels in smokers and between different ethnic groups at each stage of pregnancy.
In this study, the AFP level in smokers was increased in the first trimester by 16.3% when compared with the non-smokers. The hCG level in smokers was decreased by 27.6% and 30.5% in the first and second trimesters respectively. The fβhCG level was decreased in smokers in the second trimester by 17.1% when compared with non-smokers. The PAPP-A level was decreased by 14% and 22.8% in first and second trimesters respectively when compared with non-smokers. These results demonstrate that the effect of smoking is gestation dependant and without appropriate correction factors being applied, these serum marker changes would result in inappropriate risks being estimated for individual women.
The study on the effect of ethnicity on screening markers has shown that South Asian women had higher hCG levels in the first trimester compared with Caucasian women. They also had lower fβhCG and PAPP-A in the second trimester. Oriental women had higher first and second trimester hCG levels when compared with Caucasian women. They also had higher fβhCG and PAPP-A levels in the first trimester. Middle East women had lower first trimester AFP when compared with Caucasian women. Black women had higher hCG in the first trimester when compared with Caucasian women. In Black women, the PAPP-A level was also elevated in both trimesters. While this study confirms that correction for ethnicity is clearly indicated, appropriate correction factors are difficult to derive as there is likely to be some variation in the classification of ethnicity between studies.
Assisted Reproductive Technology
The growing use of ART in developed countries and the variety of different methods employed make accurate correction factors desirable but difficult to derive. In this study, women pregnant after ART had larger NT measurements compared with women who had conceived spontaneously. The PAPP-A level was lower in the IVF or ICSI with fresh eggs group when compared with the controls. Among the ART treatment groups, the NT was higher in the IVF or ICSI with fresh eggs group when compared with the controls. The AFP level was higher in the IVF with donor’s egg group when compared with the controls. The hCG level was higher in the ART group overall when compared with the controls. Women pregnant after IVF or ICSI with fresh eggs and frozen eggs had higher hCG level.
Smoking frequency, birthweight and prematurity
In addition to its effects on serum marker concentrations, smoking in pregnancy is known to be associated with low birth weight and prematurity. It is important therefore that maternal smoking is accurately recorded on screening request forms and in this study, the accuracy of self reported smoking status was assessed by analysis of cotinine in serum. Results showed that the percentage of self-reported smokers (24.1%) at booking was significantly lower than the cotinine-validated estimate of 30.1%. Also, smoking was associated with low birth weight, delivery prior to 39 weeks, increased AFP level (3.1%) and reduced hCG level (28.7%) in the second trimester. An increasing AFP level (but not hCG level) was associated with lower birth weight and delivery prior to 39 weeks in both smokers and non smokers but the effect was most marked in smokers. The difference in birth weight between the highest and the lowest AFP category for non-smokers was 448.3g and for smokers was 619.2g, suggesting that smoking exacerbates the effect of an elevated AFP on birth weight. Overall the difference in birth weight between the lowest AFP category in non smokers and the highest AFP category in smokers was 931.6g.
In summary, this study has shown that a cross-trimester contingent screening protocol with repeat measures has the potential to meet the UK NSC aspirational standard of 90% detection of Down’s syndrome pregnancies with a screen positive rate of less than 2%. Around 90% of women would complete screening in the first trimester without the need for a second stage sequential test. Correcting for factors such as maternal smoking habits, ethnicity and ART would further improve screening performance. Also it has been shown that where ultrasound resources are scarce, within-trimester and across-trimester protocols can reduce the need for NT measurement in all women and still deliver excellent screening performance although this falls short of the higher performance standard. The potential of these new screening protocols now need to be tested in prospective multicentre trials to confirm their performance in prospective practice.
Actions (login required)