Fetuses with Down's Syndrome detected by prenatal screening are more likely to abort spontaneously than fetuses with Down's Syndrome not detected by prenatal screening

Authors


* Dr N. Leporrier, Departement Génétique et Reproduction, CHU Caen, Avenue Côte de Nacre, 14033 Caen cedex, France.

Abstract

Objective Pregnancy with Down's Syndrome is often terminated by miscarriage. We have investigated whether prenatal screening would lead preferentially to the identification of fetuses with Down's Syndrome prone to abort spontaneously.

Design A comparison between the observed and the expected decrease in the prevalence of Down's Syndrome at term following extensive prenatal screening.

Setting A study from 1990 to 1998 in a limited and well controlled area of western France (Basse Normandie), with a birth rate close to 20,000 a year.

Population Women under 38 years of age among whom prenatal screening for Down's Syndrome using biochemical tests and ultrasound findings became progressively extensive in this era.

Methods Our study was based on the registration of Down's Syndrome cases detected prenatally from screening, and live births with Down's Syndrome. Fetal loss rate in the population of prenatally detectable Down's Syndrome was evaluated by comparing the increase in detection of cases of Down's Syndrome following prenatal screening with the expected decrease in the number of cases of Down's Syndrome at birth. The results obtained were compared with published data on the rates of fetal losses postulated to occur among fetuses with Down's Syndrome.

Main outcome measures Comparison of the potential fetal loss rate in a population in whom Down's Syndrome was detected prenatally with the expected fetal loss rate of unselected ones.

Results Prenatal screening resulted in a significant (42%) decrease (P < 0.001) in the prevalence at term of the disorder. Among the 53 fetuses with Down's Syndrome detected prenatally during the last three years of the study, about 50% would have aborted spontaneously if the pregnancy had been allowed to continue. This figure was significantly higher (P < 0.002) than expected on the basis of results from the literature, indicating that current estimates of miscarriage rates among fetuses with Down's Syndrome do not apply to the selected group of cases detectable from prenatal screening.

Conclusions Our results suggest that fetuses with Down's Syndrome detectable prenatally from screening are more likely to abort spontaneously than fetuses with Down's Syndrome which are not detected.

Introduction

Screening by maternal serum biochemical testing and ultrasound examination of the fetus is increasingly being used to identify Down's Syndrome prenatally. In evaluating the impact of an extensive prenatal screening programme for Down's Syndrome which had been developed during a 12-year period in a well defined population of western France, we found a difference larger than expected between the increase in the number of prenatally detected cases and the observed decrease in the number of actual live births. This observation led us to suspect that the natural rate of loss of fetuses with Down's Syndrome detectable prenatally following screening could be higher than for undetectable ones. The present report summarises the data recorded during the course of our study and brings arguments in favour of such a hypothesis.

Methods

Our study concerns pregnant women under 38 years of age living in a well defined area of western France where a pilot study on prenatal screening for Down's Syndrome had been set up in 1990.

Screening for Down's Syndrome was based on a biochemical serum test and an ultrasound test. The biochemical test has been described previously1,2. It consisted of a risk calculation based on the measurement between 15 and 18 weeks of gestation of serum human chorionic gonadotrophin (hCG) and unconjugated estriol (UE3). Such a protocol had been recommended for prenatal screening centres that do not measure α-fetoprotein routinely to detect neural tube defects and are willing to use only two markers instead of three3. A cutoff value of 1/250 was used in order to select women for amniocentesis. More than 120,000 pregnancies were investigated during the study, and the rate of acceptance of the test among women progressively increased to more than 80% of all pregnancies in the last three years of the study. In order to comply with recommendations from the French National Ethical Committee every one of the pregnant women involved in the study had to sign an informed consent letter. Ultrasound screening consisted of selecting for amniocentesis fetuses with anatomical defects suggestive of Down's Syndrome (short femur, congenital heart malformations in the second trimester and, since 1995, fetal nuchal translucency thickness at the first trimester). During the last three years of the study, approximately 90% of all pregnancies had an ultrasound examination and less than 10% of all pregnancies were neither investigated by ultrasound nor by a biochemical test. Comprehensive registration of live births with Down's Syndrome was obtained from the laboratory in charge of all chromosomal analyses. From our results on the increase in prenatal detection of Down's Syndrome, and the decrease in the number of live births with Down's Syndrome, we measured the potential fetal loss rate which would have been observed in the selected group of fetuses which were detected prenatally in our study if the pregnancy had been allowed to continue. We compared these results with reference values from Morris et al.4 concerning the expected fetal loss rates between the first or the second trimester and delivery.

Potential fetal loss rate for Down's Syndrome detected in our study was evaluated by comparing three figures: (i) the increase in the number of cases of Down's Syndrome detected following prenatal screening during three successive periods of three years each, from 1990 to 1998, compared with a reference period of three years from 1987 to 1989 during which limited prenatal screening was performed; (ii) the number of cases of Down's Syndrome observed at birth during the same time; (iii) the number of cases of Down's Syndrome which would have been observed during the same three periods in the absence of any modification in the prenatal screening practice in use during the reference period. This latter figure was corrected by taking into account changes in the number of pregnancies and their age distribution as registered in the geographical area under study by the French National Institute for Statistics and Economic Studies (INSEE), and by using the age-specific risk table for Down's Syndrome established by Palomaki et al.5 from the curves in Cuckle et al.6.

The expected spontaneous fetal loss rate also was calculated for the pregnancies with Down's Syndrome detected prenatally by using reference values obtained by Morris et al.4 from a meta-analysis of the literature about unselected fetuses with Down's Syndrome diagnosed in the first or second trimester.

Statistical significance was defined as a P value < 0.05. Data were analysed with SPSS software version 10.0. Difference in rates were evaluated by the χ2 test. For correlation analysis, Spearman's rank coeffient was used.

Results

Between the beginning and the end of our study, there was a distinct shift in the maternal age distribution of the pregnant women (registered by INSEE), with modal value moving from 25 to 27 years of age. There was also a simultaneous reduction (from 18,690 to 17,142) in the annual number of pregnancies. As indicated in Table 1, during the reference period (1987–1989), a small number of pregnancies with Down's Syndrome were detected prenatally following ultrasound examination of the fetus. The women were 20 to 29 years of age. Four fetuses had congenital heart disease, three had a short femur and one had the Bonnevie Ullrich syndrome. Gestational age at delivery was 16–36 weeks. During the following nine years, the number of chromosomal analyses increased more than tenfold, leading to an increase in the number of pregnancies with Down's Syndrome diagnosed prenatally from 9 to 53 pregnancies (Table 2, line 1). During the third period, there were four times as many chromosomal analyses performed on account of biochemical screening than from ultrasound screening, but each one of the two methods of screening contributed to a similar extent to the prenatal diagnosis of Down's Syndrome (Table 1), indicating that the positive predictive value of ultrasound was much higher than that of biochemical screening. The increase in the number of pregnancies with Down's Syndrome detected following prenatal screening between the reference period and the last three-year period of the study was 44 (Table 2). The impact of this screening programme on the number of live births with Down's Syndrome is seen in Table 2, line 2. The number of live births with Down's Syndrome decreased from 55 to 30 between the reference period and the last three-year period of the study. This decrease was highly significant (ρ= 1.00; P < 0.001). During each three-year period of study, the number of live births with Down's Syndrome which would have occurred in the absence of extension of our screening practice, corrected for the changes in the number and the maternal age distribution of pregnancies, remained practically unchanged (a decrease from 55 to 52; Table 2, line 3).

Table 1.  Number of amniocenteses and diagnosis of Down's Syndrome following ultrasound (first and second trimesters) or serum markers (second trimester) in women under 38 years of age.
 Reference (1987–1989)First period (1990–1992)Second period (1993–1995)Third period (1996–1998)Total (1990–1998)
  1. Period with limited prenatal screening.

Number of amniocenteses2541724247231877383
Following ultrasound (first and second trimesters)225 (0 and 225)291 (6 and 285)489 (49 and 440)640 (100 and 540)1420 (155 and 1265)
Following serum markers291433198325475963
Number of pregnancies with Down's Syndrome identified9253853116
Following ultrasound (first and second trimesters)8 (0 and 8)4 (2 and 2)12 (5 and 7)28 (16 and 12)44 (23 and 21)
Following serum markers121262572
Table 2.  Comparison of potential and expected fetal loss rate in pregnancies with Down's Syndrome.
  Periods
  Reference (1987–1988–1989)First (1990–1991–1992)Second (1993–1994–1995)Third (1996–1997–1998)Total (1990–1998)
  1. DS = Down's Syndrome.

  2. * Significantly different from 0.28 (line 8), P < 0.005.

  3. ** Significantly different from 0.29 (line 8), P < 0.002.

  4. Period with limited prenatal screening.

  5. Using mean values for fetal loss rate at the first and the second trimesters for DS published by Morris et al.4 (e.g. third period = [(0.43 × 16) + (0.23 × 37)] / 53 = 0.29).

1DS prenatal diagnosis (first + second trimesters)9 = (0 + 9)25 = (2 + 23)38 = (5 + 33)53 = (16 + 37)116 = (23 + 93)
2Observed live birth DS55413230103
3Expected DS birth without change in screening525152155
4Difference (3 − 2)11192252
5Gain in detection25 − 9 = 1638 − 9 = 2953 − 9 = 44116 − (3 × 9) = 89
6Potential number of fetal loss (5 − 4)5102237
7Potential fetal loss rate (6/5)0.310.340.50**0.41*
8Expected fetal loss rate0.230.280.260.290.28

Table 2, line 4 shows the difference between the expected number of live births with Down's Syndrome and the actual number of live births with Down's Syndrome. The difference is due to the combination of screening and natural loss.

Table 2, line 5 shows the potential effect of screening for Down's Syndrome (in the absence of natural loss) in preventing live birth with Down's Syndrome. Thus, in the last period of the study, 53 pregnancies with Down's Syndrome were discovered by screening, and so potentially 53 live births with Down's Syndrome were prevented. As the expected number of prevented live births with Down's Syndrome in the third period with the limited screening available in the reference period would be 9, the actual impact of screening improvement on fetal Down's Syndrome detection is 44 (53–9)

However, from Table 2, line 4, only 22 live births with Down's Syndrome were prevented, by termination of pregnancy. Therefore the number of fetal losses occurring naturally is 44 − 22 = 22(Table 2, line 6). The natural fetal loss rate is 22/44 = 0.50 (Table 2, line 7). The expected natural fetal loss rate, from the meta-analysis of Morris et al.4, is 0.29 (Table 2, line 8). By the χ2 test, this difference is significant (P < 0.002).

Discussion

These results show that an extensive prenatal screening programme for Down's Syndrome set up in a well defined area of western France in women under the age of 38 resulted in a significant decrease of 43% (52 to 30) in the occurrence at birth of the disorder. A distinct although less significant decrease had already been observed by the British Registry Data7–9. However, the decrease observed in our study was lower than expected when taking into account the number of fetuses with Down's Syndrome detected prenatally, suggesting that prenatal diagnosis could identify preferentially those chromosomally abnormal pregnancies that are destined to miscarry. Our results suggest that there is a higher fetal loss rate among fetuses with Down's Syndrome detectable prenatally from screening than for undetectable Down's Syndrome, a phenomenon which had not been reported until now.

Our study shows that between the second and the third periods, the number of live births with Down's Syndrome remained practically unchanged, whereas the number of pregnancies with Down's Syndrome detected prenatally increased markedly (Table 2, lines 1 and 2). The fact that this increase concerned screening in the first trimester only and not screening in the second trimester (Table 2, line 1) suggests that fetuses with Down's Syndrome detected by nuchal translucency could be more prone to abort spontaneously than fetuses with Down's Syndrome detected in the second trimester using biochemical markers. An alternative explanation could be that the first trimester loss rate described by Morris et al.4 is too low. Cuckle10 suggested a loss rate in the first trimester of 45%, instead of the 43% quoted by Morris et al.4. Such a change, however, does not seem to be large enough to account for our findings in the first trimester.

A higher fetal loss rate for fetuses with Down's Syndrome detectable prenatally than for undetectable Down's Syndrome could have been predicted. The reason why most fetuses with Down's Syndrome abort spontaneously during the course of pregnancy is not known, but it is likely to be related to the existence of placental or fetoplacental abnormalities. Unusual concentrations in the maternal serum of biochemical substances of placental (hCG) or fetoplacental (UE3) origin may be associated with those placental defects leading to miscarriage. Histological abnormalities of the placenta have been described in cases of fetal chromosomal aneuploidy11. A recent morphometric study of chorionic villi obtained by chorionic villus sampling from pregnancies with Down's Syndrome has shown an inverse relationship between nuchal translucency thickness and chorionic villus diameter and the number of capillaries per villus in cross section12. The results of our study suggest that more research should be performed into the association of miscarriage in Down's Syndrome and pathological and biochemical abnormalities in pregnancy affected by Down's Syndrome.

Acknowledgements

The authors would like to thank J. Alliet, G. Plessis and M. Herlicoviez for providing clinical data, and E. Poiblaud, R. Golba, C. Lebris and D. Gourdier for expert technical assistance. This work was supported by a grant (PHRC) from the Ministère de la Santé, France. The authors would also like to thank Prof Howard Cuckle for useful comments in the manuscript.

Ancillary