The in¯uence of fetal sex in screening for Down syndrome inthe second trimester using AFP and free b-hCG

Kevin Spencer*

Endocrine Unit, Clinical Biochemistry Department, Harold Wood Hospital, Gubbins Lane, Romford, Essex RM3 0BE, UK

In a study of 30 802 unaffected pregnancies and 90 cases of Down syndrome I have analysed secondtrimester maternal serum alpha-fetoprotein (AFP) and free b-human chorionic gonadotrophin (b-hCG) andgrouped the data by fetal sex. In pregnancies with an unaffected female fetus maternal serum AFP wassigni®cantly lower (3%) than in the presence of a male fetus, whilst free b-hCG was signi®cantly raised (7%)in the presence of a female fetus. These differences led to a signi®cantly higher false positive rate in womencarrying a female fetus (5.78% versus 4.64%). In pregnancies affected by Down syndrome the median AFPin the presence of a female fetus was not signi®cantly different to that with a male fetus (0.72 versus 0.73);similarly for free b-hCG the difference (2.37 versus 2.48) was not statistically signi®cant. Although thedetection rate amongst male fetuses was slightly higher than in females (74% versus 66%), this was notstatistically signi®cant. Although fetal gender does appear to have a signi®cant effect on maternal serummarker levels, leading to a higher false positive rate in females, in cases of Down syndrome there is noevidence that such marker differences are signi®cant and consequently little evidence for any fetal genderbias in detection rates. Copyright # 2000 John Wiley & Sons, Ltd.

KEY WORDS: free b-hCG; AFP; trisomy 21; detection rate; false positive rate

INTRODUCTION

As early as 1965, Brody and Carlstrom (1965)demonstrated that maternal serum total hCG levelsin the third trimester of pregnancy were signi®cantlyhigher in women carrying a female fetus compared tothose with a male fetus. This observation wascon®rmed by others (Obiekwe and Chard, 1982;1983) and was in agreement with studies of termplacentae, serum and amniotic ¯uid total hCG(Crosignani et al., 1972; Hobson and Wide, 1974).However, serum studies of early (5±20 weeks) and latepregnancy (21±40 weeks) failed to show statisticallysigni®cant differences between total hCG levels ofmale and female conceptuses (Deville et al., 1980).Similarly, Wide and Hobson (1974) found no differ-ence in the amount of total hCG in placentae frommale and female fetuses during the 10th to 20th weekof gestation. Others such as Boroditsky et al. (1975)have suggested that such maternal serum differencesdo not occur until after the 37th week of gestation.

Since 1990, maternal serum total hCG and its free bsubunit have become increasingly used in secondtrimester screening for Down syndrome in conjunctionwith maternal serum alpha-fetoprotein (AFP) (Waldet al., 1997). Like hCG there is evidence that secondtrimester maternal serum AFP levels also vary withfetal sex. Sowers et al. (1983) and Szabo et al. (1995)have shown higher levels of AFP in the presence of amale fetus. The implications of these changes forDown syndrome screening has only recently beenconsidered. Leporrier et al. (1992) showed that fetal

sex differences were not signi®cantly different for totalhCG prior to the 18th week of pregnancy andtherefore recommended that screening interpretationshould occur only before the 18th week. Bazzett et al.(1998), however, in a much larger study of over 15 000cases showed that maternal serum AFP levels in thepresence of a female fetus were consistently lower thanthose with a male fetus and, conversely, total hCGlevels were higher in pregnancies with a female fetus.These difference existed at all gestations between 14and 20 weeks and led them to the conclusion thatDown syndrome risks for women with female fetuseswould be higher than for those with a male fetus andhence the presence of a female fetus would lead to anincreased false positive rate.

In view of the potential effects of fetal sex on secondtrimester screening performance, in this study I set outto analyse retrospective data from our screeningprogramme using AFP and free b-hCG to see ifmarker changes with fetal sex could be con®rmed andto investigate if this brought about differences in thefalse positive rate or detection rates.

MATERIALS AND METHODS

The Harold Wood centre has provided secondtrimester prenatal screening services to a number ofobstetric units, using maternal serum AFP and freeb-hCG, since 1991 and the performance of thisprogramme has been described previously (Spencer,1999). Data obtained as part of this screeningprogramme were retrieved from the database alongwith pregnancy outcome information. Pregnanciesresulting in multiple pregnancy, neural tube defect,ventral wall defect, fetal death or chromosomal

*Correspondence to: K. Spencer, Endocrine Unit, Clinical Biochem-istry Department, Harold Wood Hospital, Gubbins Lane, Romford,Essex RM3 0BE, UK. E-mail: Kevin_Spencer@Compuserve.com

PRENATAL DIAGNOSIS

Prenat Diagn 2000; 20: 648±651.

Copyright # 2000 John Wiley & Sons, Ltd. Received: 24 January 2000Revised: 5 April 2000

Accepted: 9 April 2000

abnormality were excluded from the analysis, as werecases in which fetal sex was not recorded. A total of30 802 data sets in unaffected pregnancies wereavailable for analysis. Information on fetal sex wasavailable in 90 of the 107 cases of Down syndromescreened during the period 1991±98 (Spencer, 1999).

Data were collated by gestational age and fetal sex.All statistical analyses were performed with Analyse-It(Smart Software, Leeds, UK), a statistical softwareadd-in for Microsoft Excel 7 using analytical resultsconverted to weight-corrected MoMs as previouslydescribed (Spencer, 1999).

RESULTS

Table 1 shows the median AFP MoM and the meanlog10 AFP MoM at the major gestational weeksaccording to fetal sex. The data for free b-hCG issimilarly shown in Table 2. When the log10 analyteMoMs were compared between fetal sexes using t-testsassuming equal variance, the lower levels of AFP andthe high levels of free b-hCG observed in the presenceof a female fetus were highly signi®cant ( p<0.00001).

When the data set was examined by Downsyndrome risk, at the term risk cut-off of 1 in 250,876 of the 15 148 pregnancies (5.78%) with a femalefetus were identi®ed `at increased risk' compared with726 of the 15 654 pregnancies (4.64%) with a malefetus. Thus women carrying a female fetus were 1.25times more likely to be in the `at increased risk'category than those carrying a male fetus. Thisdifference between the number of cases identi®ed atincreased risk in the male and female fetal sex groupswas highly signi®cant when tested by chi-square tests( p<0.0001).

Of the Down syndrome cases which were in thescreened population described previously (Spencer,1999) fetal sex was available for 90 of the 107 cases. Inthis group of 90, 47 cases were of a female fetus and 43were of a male fetus. The median maternal age of thetwo groups was identical at 33.3 years. The medianAFP MoM in the female group was 0.72 and the meanlog10 MoM was x0.149, whilst in the male group thiswas 0.73 and x0.159. The t-test with equal varianceshowed these differences to be not signi®cant( p=0.84). The median free b-hCG MoM in the

female group was 2.37 with a mean log10 MoM of0.356, whilst in the male group this was 2.48 and 0.421.These differences did not reach statistical signi®cance( p=0.26). In the female group 31 of 47 cases (66%)had risks greater than the cut-off compared with 32 of43 (74%) in the male group. The ®sher exactprobability test showed these differences did notreach statistical signi®cance ( p=0.26).

The proportion of male fetuses amongst theDown group was 48% (43/90) and this was notstatistically different to the 51% (15 654/30 802) in theunaffected population (chi-square 0.222, p=0.638).The male : female ratio of 1.033 in the unaffectedpopulation was very similar to the all England andWales ®gure of 1.06 (Of®ce for National Statistics,1998). The male : female ratio in cases of Downsyndrome was 0.91. Unlike the large scale nationalrecords when a male : female ratio of 1.23 wasobserved (Mutton et al., 1996), it was not possibleto demonstrate a marked excess of males in ourlargely second trimester identi®ed cases.

DISCUSSION

In the present study it has been shown that maternalserum AFP values in normal euploid pregnancies areon average 3% lower in the presence of a female fetuscompared to male fetuses, but in pregnancies affectedby Down syndrome the difference is reduced to 1.4%.Sowers et al. (1983) in a smaller series of 2445 womenfound a median maternal serum AFP level some 14%lower in the presence of a female fetus, and Wald andCuckle (1984) observed that maternal serum AFPvalues were signi®cantly associated with fetal sexÐhigher values being more common with a male fetus.Petrikovsky (1989) found an elevated male : femaleratio in cases selected for amniocentesis as a result of araised maternal serum AFP. Calvas et al. (1990) in asmall series of 1480 euploid normal pregnancies founda signi®cant increase in maternal serum AFP in thepresence of a male fetus but concluded that thedifference was too small to affect clinical interpreta-tion. Lockwood et al. (1993) studying a cohort of 5114patients undergoing karyotyping in the 13±22 weekperiod, observed 42 fetuses with Down syndrome. Inthe 3192 euploid cases with fetal sex outcome,

Table 1ÐMedian AFP MoM and mean log10 AFP MoM at various gestational weeks and classi®ed by the presence of amale or female fetus

Females Males

Female : maleMoM ratio

Gestation(week)

MedianAFP MoM

Mean log10

AFP MoM No.MedianAFP MoM

Mean log10

AFP MoM No.

14 0.981 x0.01876 2772 0.9919 x0.00478 2802 0.98915 0.992 x0.00666 8156 1.014 0.007765 8478 0.97816 0.992 x0.00256 2662 1.046 0.014673 2770 0.94817 0.992 x0.00462 1072 1.0246 0.009155 1072 0.96818 0.905 x0.05523 202 1.0137 0.003792 222 0.893

All weeks 0.981 x0.00911 15148 1.0137 0.00648 15654 0.968

INFLUENCE OF FETAL SEX IN DOWN SYNDROME SCREENING 649

Copyright # 2000 John Wiley & Sons, Ltd. Prenat Diagn 2000; 20: 648±651.

maternal serum AFP was 7% lower in women with afemale fetus. Analysis of the 42 cases of Downsyndrome, plus a further 30 cases not from thiscohort, indicated a highly signi®cant lowering ofmaternal serum AFP in the presence of a femaleDown fetus (0.65 versus 0.89 MoM). However,Ghidini et al. (1998), in a series of 49 Down cases,could not show any statistical difference betweenfemales and males (0.75 versus 0.67 MoM). In perhapsthe largest euploid series to date, Bazzett et al. (1998)studied maternal serum AFP data from 15 428 patientsand con®rmed that maternal serum AFP levels in thepresence of a female fetus were consistently 6% lowerthan in males. The results of our studies in cases ofDown syndrome agree with Ghidini et al. (1998) inthat no statistical difference could be demonstrated,and our larger euploid series shows a somewhatsmaller effect than that observed by Bazzett et al.(1998) and is in agreement with the 3% loweringobserved by Spong et al. (1999) in an analysis of 977cases. The ®nding of a small reduction in maternalserum AFP in euploid female fetuses is consistent withour previous ®nding of a 6% reduction of amniotic¯uid AFP from pregnancies with a female fetus andour ®nding of no signi®cant difference between sexesin 91 cases of Down syndrome (Spencer et al., 1997).

In this present study I have also shown thatmaternal serum free b-hCG levels in normal euploidpregnancies are on average 7% higher in the presenceof a female fetus compared to a male fetus; but, inpregnancies affected by Down syndrome, the differ-ence is reversed with levels in the presence of a femalefetus being reduced by 4%, although this is not ofstatistical signi®cance. Although the pattern of highlevels of maternal serum total hCG associated withfemale fetuses has been well documented, its implica-tion for Down syndrome screening has only beenconsidered by a few authors. Leporrier et al. (1992)found signi®cant gender differences only after 17weeks. Wolf et al. (1992) demonstrated a trend tohigher values for free b-hCG and total hCG inamniotic ¯uid of euploid pregnancies with a femalefetus, although this was not signi®cant. They did,however, ®nd striking differences in a series of 21Down cases with a median MoM in the presence of afemale fetus some 3.6 times higher. In a much larger

series of 91 cases (Spencer et al., 1997) this could notbe con®rmed, and a 20±30% elevation of total hCGand free b-hCG was observed in the presence of afemale fetus but did not reach statistical signi®cance.Bazzett et al. (1998), in an analysis of 11 428 euploidpatients, showed a 6% elevation of maternal serumtotal hCG, and Ghidini et al. (1998) showed that in 49cases of Down syndrome maternal serum total hCGwas 5% higher in the presence of a female fetus, butagain this was not signi®cant. In a small study of 977euploid pregnancies, the same group (Spong et al.,1999) could demonstrate no signi®cant differencebetween the sexes for maternal serum total hCG.

The observation of a highly signi®cant elevation ofmaternal serum free b-hCG in the presence of a femalefetus is consistent with the ®ndings of Bazzett et al.(1998) for total hCG and our ®ndings in amniotic ¯uidfor free b-hCG (Spencer et al., 1997). The inability todemonstrate a signi®cant difference in the Downgroup is consistent with our amniotic ¯uid studiesand those of Spong et al. (1999) for total hCG.

What impact do these sex differences have on Downsyndrome detection and false positive rates? Thereduced levels of AFP and the higher levels of totalhCG or free b-hCG in maternal serum of euploidpregnancies with a female fetus should lead to anincreased false positive rate in that group compared tothose with a male fetus. Furthermore, it has beensuggested that if this same pattern occurs in cases withDown syndrome the maternal serum screening wouldbe more ef®cient in detecting female affected pregnan-cies than those with a male affected pregnancy.However, Ghidini et al. (1998) could ®nd no evidencefor this in reviewing sex ratios amongst cases detectedbecause of abnormal serum screening or those detectedbecause of advanced maternal age. Similarly in thisseries of Down syndrome cases collected over the pastseven years I could show no statistical differencebetween detection rates in the presence of a female ormale fetus. It would seem that there is no evidence tosupport any impact of fetal gender on Down syndromedetection rates.

When false positive rates have been looked at, in astudy of 977 women, Spong et al. (1999) found a 14%versus 9% false positive rate (relative risk 1.5) in thepresence of a female fetus. In our study of a much

Table 2ÐMedian free b MoM and mean log10 free b MoM at various gestational weeks and classi®ed by the presence of amale or female fetus

Females Males

Female : maleMoM ratio

Gestation(week)

Median freeb MoM

Mean log10

free b MoM No.Median freeb MoM

Mean log10

free b MoM No.

14 1.110 0.060296 2772 1.060 0.034538 2802 1.04715 1.040 0.027054 8156 0.970 x0.00013 8478 1.07216 1.000 0.002322 2662 0.930 x0.02781 2770 1.07517 0.985 0.00053 1072 0.885 x0.04746 1072 1.11318 0.940 x0.06028 202 0.860 x0.04083 222 1.093

All weeks 1.040 0.025955 15148 0.970 x0.00308 15654 1.072

K. SPENCER650

Copyright # 2000 John Wiley & Sons, Ltd. Prenat Diagn 2000; 20: 648±651.

larger series we con®rm that such differences in thefalse positive rate are highly signi®cant with a falsepositive rate in females of 5.78% compared with 4.64%in males (relative risk 1.25).

Redressing the inequality in false positive rate wouldbe possible by correcting the measured analyte MoMin the presence of a female fetus. However, this wouldrequire prior knowledge of fetal gender. Althoughsome ultrasound studies have shown an overallaccuracy of fetal gender assignment of 97±99% at14±20 weeks of gestation (Meagher and Davison,1996; Harrington et al., 1996), gender assignment atearlier gestations seems to show a decrease in accuracytowards earlier gestations (Efrat et al., 1999; Whitlowet al., 1999). Other complicating issues, such asdisclosure of fetal sex (Harrington et al., 1996), andethical issues surrounding this make correcting forfetal sex too problematical.

REFERENCES

Bazzett LB, Yaron Y, O'Brien JE, Critch®eld G, Kramer RL,Ayoub M, Johnson MP, Evans MI. 1998. Fetal gender impact onmultiple-marker screening results. Am J Med Genet 76: 369±371.

Boroditsky RS, Reyes FI, Winter JSD, Faiman C. 1975. Serumhuman chorionic gonadotropin and progesterone patterns in thelast trimester of pregnancy: relationship to fetal sex. Am J ObstetGynecol 121: 238±241.

Brody S, Carlstrom G. 1965. Human chorionic gonadotropinpattern in serum and its relation to the sex of the fetus. J ClinEndocrinol 25: 792±797.

Calvas P, Bourrouillou G, Smilovici W, Colombies P. 1990.Maternal serum alpha fetoprotein and fetal sex. Prenat Diagn10: 134±136.

Crosignani PG, Nencioni T, Brambati B. 1972. Concentration ofchorionic gonadotrophin and chorionic somatomammotrophin inmaternal serum, amniotic ¯uid and cord blood serum at term.J Obstet Gynaecol Br Commonw 79: 122±126.

Deville JL, Gaspard U, Reuter AM, Franchimont P, Lambotte R.1980. Taux seriques maternels de la choriogonadotrophine et deses sous-unites libres a and b en fonction du sexe foetal. Etudepreliminaire. C R Soc Biol 174: 365±369.

Efrat Z, Akinfenwa OO, Nicolaides KH. 1999. First trimesterdetermination of fetal gender by ultrasound. Ultrasound ObstetGynecol 13: 305±307.

Ghidini A, Spong CY, Grier RE, Walker CN, Pezzullo JC. 1998. Ismaternal serum triple screening a better predictor of Downsyndrome in female than in male fetuses? Prenat Diagn 18:123±126.

Harrington K, Armstrong V, Freeman J, Aquilina J, Campbell S.1996. Fetal sexing by ultrasound in the second trimester: maternalpreference and professional ability. Ultrasound Obstet Gynecol 8:318±321.

Hobson B, Wide L. 1974. Chorionic gonadotrophin in the humanplacenta in relation to the sex of the fetus at term. J Endocrinol 60:75±80.

Leporrier N, Herrou M, Leymarie P. 1992. Shift of the fetal sex ratioin hCG selected pregnancies at risk for Down syndrome. PrenatDiagn 12: 703±704.

Lockwood CJ, Lynch L, Ghidini A, Lapinski R, Berkovitz G,Thayer B, Miller WA. 1993. The effect of fetal gender on theprediction of Down syndrome by means of maternal seruma-fetoprotein and ultrasonographic parameters. Am J ObstetGynecol 169: 1190±1197.

Meagher S, Davison G. 1996. Early second trimester determinationof fetal gender by ultrasound. Ultrasound Obstet Gynecol 8:322±324.

Mutton D, Alberman E, Hook EB. 1996. Cytogenetic andepidemiological ®ndings in Down syndrome, England andWales 1989 to 1993. J Med Genet 33: 387±394.

Obiekwe BC, Chard T. 1982. Human chorionic gonadotrophin levelsin maternal blood in late pregnancy: relation to birthweight, sexand condition of the infant at birth. Br J Obstet Gynaecol 89:543±546.

Obiekwe BC, Chard T. 1983. Placental proteins in late pregnancy:relation to fetal sex. J Obstet Gynaecol 3: 163±164.

Of®ce for National Statistics. 1998. 1996 Birth statistics. Series FM1no 25. HMSO: London.

Petinkovsky B. 1989. Maternal serum alpha fetoprotein concentra-tion and fetal sex. Prenat Diagn 9: 449±450.

Sowers SG, Reish RL, Burton BK. 1983. Fetal sex-relateddifferences in maternal serum a-fetoprotein during the secondtrimester of pregnancy. Am J Obstet Gynecol 146: 786±789.

Spencer K. 1999. Second trimester prenatal screening for Down'ssyndrome using alpha-fetoprotein and free beta hCG: a seven yearreview. Br J Obstet Gynaecol 106: 1287±1293.

Spencer K, Muller F, Aitken DA. 1997. Biochemical markers oftrisomy 21 in amniotic ¯uid. Prenat Diagn 17: 31±37.

Spong CY, Ghidini A, Stanley-Christian H, Meck JM, Seydel FD,Pezzullo JC. 1999. Risk of abnormal triple screen for Downsyndrome is signi®cantly higher in patients with female fetuses.Prenat Diagn 19: 337±339.

Szabo M, Veress L, Munnich A, Papp Z. 1995. Maternal age-dependent and sex-related changes of gestational serum alpha-fetoprotein. Fetal Diagn Ther 10: 368±372.

Wald NJ, Cuckle HS. 1984. Open neural tube defects. In Antenataland Neonatal Screening, Wald NJ (ed). Oxford University Press:Oxford; 25±73.

Wald NJ, Kennard A, Hackshaw A, McGuire A. 1997. Antenatalscreening for Down's syndrome. J Med Screen 4: 181±246.

Whitlow BJ, Lazanakis MS, Economides DL. 1999. The sono-graphic identi®cation of fetal gender from 11 to 14 weeks ofgestation. Ultrasound Obstet Gynecol 13: 301±304.

Wide L, Hobson B. 1974. Relationship between the sex of the fetusand the amount of human chorionic gonadotrophin in placentaefrom the 10th to the 20th week of pregnancy. J Endocrinol 61:75±81.

Wolf GC, Francis WB, McConnell TS, Khazaeli MB. 1992.Amniotic ¯uid levels of human chorionic gonadotropin and itsalpha and beta subunits in second trimester chromosomallyabnormal pregnancies. Prenat Diagn 12: 93±101.

INFLUENCE OF FETAL SEX IN DOWN SYNDROME SCREENING 651

Copyright # 2000 John Wiley & Sons, Ltd. Prenat Diagn 2000; 20: 648±651.