Female Predominance (Low Sex Ratio) in47,+21 Mosaics

Ernest B. Hook,1,2* Philip K. Cross,3 and David E. Mutton4

1School of Public Health, University of California, Berkeley, California2Department of Pediatrics, University of California, San Francisco, California3Bureau of Environmental Health, New York State Department of Health, Albany, New York4Department of Environmental and Preventive Medicine, Wolfson Institute of Preventive Medicine, Medical College ofSt. Bartholomews Hospital, London, United Kingdom

Data from the New York State ChromosomeRegistry on over 10,000 cases of Down syn-drome reported from 1977 to 1996 confirmfindings in the England and Wales Cytoge-netic Register that in mosaic 46/47,+21 casesof Down syndrome the male/female ratio (asinferred from XY and XY karyotypes respec-tively) is less than 1.0 as opposed to the ratioin nonmosaic 47,+21 cases in which the ratiois close to 1.2, or in the general backgroundpopulation with ratio of about 1.05. Theseresults may reflect in part differential pair-ing in 47,+21 somatic cells of X and Y chro-mosomes with a 21 chromosome and/or inin-utero selection. Am. J. Med. Genet. 84:316–319, 1999. © 1999 Wiley-Liss, Inc.

KEY WORDS: Down syndrome; sex ratio;sex proportion; transloca-tion; prenatal diagnosis; mo-saic

INTRODUCTION

Many investigators of Down syndrome have reporteda relative excess of males (for review of the literaturesee Huether et al. [1996]). In the pooled experiencefrom all reports available to them, Huether et al. re-ported a male/female ratio of 1.15 (5,132/4,446) in non-mosaic cases of 47,+21 (different from an expectedbackground ratio of about 1.06 in births, with P <0.001, x2 4 17.3, d.f. 4 1). Mutton et al. [1996] in oneof the largest systematic collections of data from asingle source, in a publication too late to be included inthe review of Huether et al., reported a male/femaleratio of 1.24 in the 5,422 cases between 1989 and 1993in England and Wales, also “significantly” higher than

the ratio in all births in the jurisdiction, of 1.05 (P <0.001, x2 4 36.3, d.f. 4 1). But unexpectedly, amongthe few cases of mosaic 46/47,+21, there was a notableexcess of females. The observed male/female sex ratiowas 23 to 43 or 0.53, a nominally “significant” differ-ence from the ratio in nonmosaics (P < 0.01, x2 4 36.3,d.f. 4 1).

This trend to female excess in mosaics is highly sug-gestive but could result from chance. Here we reportthe results of attempts to confirm the trend in a groupon which extensive systematic data were available onall karyotypes in a population reported to an ongoingChromosome Registry.

MATERIALS AND METHODS

We used data from the New York State ChromosomeRegistry on cases that had been reported for the years1977 (the beginning of systematic collection of all pre-natal diagnoses) to 1996, as reported by December1997, when these years had been “closed out.” Methodsof the Registry are described in Hook et al. [1981]. TheRegistry has made no substantive changes in data col-lection procedures—nor any whatsoever pertinent tosex ratios—since that publication. Here we infer sexfrom karyotype; no attempt was made to determinethat cases diagnosed as XY or XX were phenotypicallymale or female, respectively.

A ruling by the Commissioner of the New York De-partment of Health requires all laboratories within thestate with diagnostic cytogenetic “permits” to reportthe results of diagnosed chromosome abnormalities—except on specimens obtained specifically for cytoge-netic investigation of known or suspected tumors and/or malignancies—regularly to the New York StateChromosome Registry. The department also requiredreports on specific results of all cytogenetic diagnosesof specimens obtained prenatally, whether or not ab-normal. Out-of-state laboratories which may have such“permits” must at least report results on specimensoriginating from within New York State or from NewYork residents who may be out of state at time of sam-pling. The Department issues (and renews) such per-mits if the laboratory demonstrates (and continues to

*Correspondence to: Ernest B. Hook, 140 Warren Hall, MCHSchool of Public Health, University of California, Berkeley, CA94720-7360.

Received 7 April 1998; Accepted 21 January 1999

American Journal of Medical Genetics 84:316–319 (1999)

© 1999 Wiley-Liss, Inc.

demonstrate) satisfactory proficiency in cytogenetic di-agnosis, and meets (and continues to meet) other cri-teria concerning credentials and training of staff, na-ture of the laboratory facilities, record keeping proce-dures, etc. By New York State law and New York StateDepartment of Health regulations and rulings, all labo-ratories offering cytogenetic diagnostic services to NewYork State residents must obtain such a “permit” fromthe New York State Department of Health. No labora-tory anywhere, without such a permit, may receive anyNew York State legally sanctioned reimbursement fora diagnostic cytogenetic procedure. The State may as-sess fines and other penalties against any laboratorywithin the state, and its Director, that charges for acytogenetic procedure if the laboratory lacks the requi-site permit. New York State law specifies penalties forany physician within the State (except those outside itsjurisdiction such as federal institutions; e.g., Veteran’sAdministration hospitals) who refer a specimen for di-agnosis to a laboratory in State or out of state withoutan appropriate permit. (Possible allowable exceptionsinclude studies for purposes of research, but only ifthere is no associated charge to the patient, the State ofNew York, or any other “third party” provider.)

The Registry staff has made no formal evaluation ofthe completeness (or accuracy) of reporting to the Reg-istry derived from independent knowledge of cases ofcytogenetic abnormalities. (Reports to the Registry donot include the name of the individual studied. Thismakes investigation unfeasible that depends on match-ing records.) Confidence in the accuracy and complete-ness of reported results arises from: i) the proficiencyevaluation undergone by laboratories with permits,which establishes their basic competence; ii) regularinspection by staff of the Department of Health of alllaboratories with permit and of their records—although this inspection did not routinely in the pastand does not necessarily at present determine for eachlaboratory whether all results have been reported tothe Registry; iii) continual communication to the labo-ratory of the need for a permit to provide diagnosticservices to New York State residents (and implicitly ofthe associated penalties if such requirement is notmet); iv) continual communication by the Registry staffto the laboratory concerning the necessity of reporting

all pertinent diagnoses to the Registry (and implicitlyof the possible penalties if reporting is not complete nordone on a timely basis, etc.); and v) regular review byRegistry staff of the quantity and nature of diagnosticreports received from each laboratory, with follow-upqueries to the laboratory if the number of reports di-minishes and/or a diagnosis is questionable, ambigu-ous, or inconsistent with other information reported.

The state has not yet had to invoke penalties againstany laboratory to ensure compliance with reporting re-quirements to the Chromosome Registry. Legal actionhas been taken against institutions which refused tocomply with other New York State Department ofHealth–mandated reporting requirements; e.g., thoseof the Cancer Registry, of birth and death certificates,etc.

RESULTS

Observations by 5-year maternal age interval appearin Table I in those diagnosed prenatally and postna-tally. Table II presents the analyses of sex ratios. Asmay be noted for all three major age categories (<35,>34, and unknown), in cases diagnosed prenatally andpostnatally—a total of six different strata—females ex-ceeded males among the mosaics but the reverse pat-tern held for the nonmosaics. By a two-tailed sign testthis consistent trend is “significant” with P 4 0.03. Forall reported cases pooled, the difference between themosaics and nonmosaics (198/231 vs. 5,719/4,999) wassignificant at P < 0.005 (x2 4 8.6, 1 d.f.). The diver-gence from chance was greater in those diagnosed pre-natally (39/54 vs. 1,648/1,350; P < 0.02; x2 4 6.2; 1 d.f.)than postnatally (159/177 vs. 4,071/3,649; P < 0.06; x2

4 3.8; 1 d.f.). The trend to more females in mosaics, asevaluated by the ratios of proportion to males, was alsostronger (i.e., farther from the “null” value of 1.00) inthose diagnosed prenatally than postnatally. These ra-tios (proportion male among mosaics to proportionmale among the nonmosaics) were (0.419/0.550) 4 0.76in those diagnosed prenatally and (0.473/.527) 4 0.90in those diagnosed postnatally.

We also examined the sex ratio in the translocationtrisomies (nonmosaic). Among cases diagnosed postna-tally the sex ratio was 1.14 (234/205), among those di-

TABLE I. Number of Observed XY and XX Karyotypes Among Those With Mosaic and Nonmosaic 47,+21

Maternal age

Prenatal diagnoses Postnatal diagnoses

Nonmosaic +21 Mosaic 46/47,+21 Nonmosaic +21 Mosaic 46/47,+21

XY XX XY XX XY XX XY XX

<15 0 0 0 0 3 1 0 015–19 9 7 0 0 89 85 2 120–24 16 16 6 8 239 214 5 1125–29 124 108 8 5 1,150 958 45 4430–34 182 148 8 15 750 724 28 3135–39 488 405 3 6 324 316 13 1240–44 272 226 0 0 213 194 8 1245–49 18 22 0 0 30 36 0 1>49 0 1 0 1 14 7 0 1Unknown 539 417 14 19 1,259 1,114 14 19Total 1,648 1,350 39 54 4,071 3,649 115 132

Excess Females in 47,+21 Mosaics 317

agnosed prenatally 1.04 (29/28), and among all pooled1.13 (263/233).

DISCUSSION

The resources of the New York State Registry do notenable complete verification of diagnostic informationof all reporting laboratories. As noted above we areconfident that the data analyzed here are fairly accu-rate and complete but cannot exclude formally the pos-sibility of errors or predict the precise extent of theirmagnitude. Nevertheless, for several reasons we thinkthat any such errors do not affect materially the natureof the observed trends.

Errors may result from random factors or from bi-ases of some type. Random errors (or errors resulting innondifferential misclassification) result in a type of ef-fect analogous to that seen with “regression to themean” [Rothman, 1986]. Random errors would tend toobscure, not to enhance, any difference between mosaicand nonmosaics. That is, the observed difference in sexratio between the nonmosaic and mosaic 47,+21 casesobserved in this study tends to occur despite, not be-cause of these errors.

We note two potential biases of pertinence (kindlycalled to our attention by Patricia Jacobs and DorothyWarburton). These biases could result from maternalcell contamination among the prenatally diagnosedcases, and from selective ascertainment of fertile fe-male mosaics among the postnatally diagnosed cases.

Maternal cell contamination would of course contrib-ute to false positive prenatal diagnosis of mosaicism

among 47,XX,+21 but not 47,XY,+21 cases and thus tosome of the observed trend. The proportion of stage IIImosaics (i.e. mosaicism in many colonies not a singleone) attributable to maternal cell contamination hasbeen reported to be about 0.2% [Wilson et al. 1989; seealso Hook, 1992] in all those prenatal diagnoses (nor-mal and abnormal) in which such could be detectedreadily. If this proportion occurred in this series, thenabout three of the cases included as 46,XX/47,XX+21are false mosaic cases. This would mildly weaken thetrend but not abolish it. But if there are 9 or more suchmosaic cases due to maternal contamination among the1404 females diagnosed prenatally in this series, 0.6%or more, then while the trend to a difference wouldremain, adjustment would abolish statistical signifi-cance (P > .05 one tailed) in comparison of the differ-ences between nonmosaic 47,+21 and “true” mosaics.To completely abolish the prenatal mosaic nonmosaicdifference, i.e. for the sex ratio among the mosaics di-agnosed prenatally to be truly equal to that among thenonmosaics, one must postulate that 22 of the 54 pre-natally diagnosed cases, 40% resulted from maternalcell contamination.

It appears likely that maternal contamination doescontribute to a greater observed effect in those diag-nosed prenatally than postnatally. But it appears im-plausible to us that maternal contamination explainsthe entire observed trend among those diagnosed pre-natally. And it has no relevance to the trends in post-natally diagnosed cases.

Among the “postnatally” diagnosed 46/47,+21 mosa-

TABLE II. Male/Female (M/F) Ratios, 90% Confidence Limits on Ratios, Male Proportions, and Standard Error

Male (XY) Female (XX) Total M/F ratio90%

Lower limit90%

Upper limitMale

proportionStandard

error

47,+21Prenatally diagnosed

<35 331 279 610 1.186 1.038 1.357 0.543 0.020>34 778 654 1,432 1.190 1.090 1.299 0.543 0.013Alla 1,648 1,350 2,998 1.221 1.149 1.297 0.550 0.009

Postnatally diagnosed<35 2,231 1,982 4,213 1.126 1.070 1.184 0.530 0.008>34 581 553 1,134 1.051 .953 1.159 0.512 0.015Alla 4,071 3,649 7,720 1.116 1.075 1.158 0.527 0.006

All cases<35 2,562 2,261 4,823 1.133 1.081 1.188 0.531 0.007>34 1,359 1,207 2,566 1.126 1.055 1.202 0.530 0.010Alla 5,719 4,999 10,718 1.144 1.108 1.181 0.534 0.005

Mosaic 46/47,+21Prenatally diagnosed

<35 22 28 50 0.786 0.480 1.250 0.440 0.070>34 3 7 10 0.429 0.066 1.166 0.300 0.145Alla 39 54 93 0.722 0.504 1.014 0.419 0.051

Postnatally diagnosed<35 80 87 167 0.920 0.711 1.186 0.479 0.039>34 21 26 47 0.808 0.487 1.305 0.447 0.073Alla 159 177 336 0.898 0.749 1.075 0.473 0.027

All cases<35 102 115 217 0.887 0.707 1.109 0.470 0.034>34 24 33 57 0.727 0.457 1.121 0.421 0.065Alla 198 231 429 0.857 0.730 1.005 0.462 0.024

aIncludes those where maternal age unknown.

318 Hook et al.

ics, a certain proportion will come to attention becausethey have had affected children and these are muchmore likely to be female than male. But this will onlyapply to those cases diagnosed as adults who representa trivial fraction of cases in the series. The vast major-ity of cases diagnosed have been under age 10, in factwere in the early years of life at the time of diagnosis.

Thus, we interpret the results as demonstrating thatthe difference between the mosaics and the 47,+21 non-mosaics are highly unlikely to be attributable tochance, or some type of idiosyncratic recording or re-porting errors.

We can offer only two conjectures for the differencebetween mosaics and nonmosaics. One relates to pro-duction of mosaicism, the other to embryonic-fetal se-lection. First, in early embryonic life 47,XX+21 cellsmay be more likely than 47,XY,+21 cells to lose anextra 21 chromosome through anaphase loss in mitoticdivision, thus producing a 46 cell line. In a cell in a47,XY+21 conceptus, there are six small chromosomes;perhaps the Y tends to pair with a +21 in mitotic divi-sion (in addition to or instead of with an X chromo-some), making anaphase loss of an extra 21—and thusthis source of mosaicism—less likely in the male. Sec-ond, karyotypic sex may interact strongly with the like-lihood of intrauterine death in a mosaic. That is, thepresence of mosaicism may enhance selectively the

relative likelihood of embryonic and early fetal survivalof a female with 47,+21. These remain hypotheses only.

ACKNOWLEDGMENT

We thank the Directors and staff of the affiliatedlaboratories with the New York State ChromosomeRegistry for their reports.

REFERENCES

Hook EB. 1992. Chromosome abnormalities: prevalence risks and recur-rence. In: Brock DJH, Rodeck CH, Ferguson-Smith MA, editors. Pre-natal diagnosis and screening. Edinburgh: Churchill-Livingstone. p351–392.

Hook EB, Cross PK, Schreinemachers D. 1981. The evolution of the NewYork State chromosome registry. In: Hook EB, Porter IH, editors.Population and biological aspects of human mutation. New York: Aca-demic Press. p 389–428.

Huether CA, Martin RLM, Stoppelman SM, D’Souza S, Bishop JK, TorfsCP, Lorey F, May KM, Hanna JS, Baird PA, Kelley JC. 1996. Sex ratioin fetuses and liveborn infants with autosomal aneuploidy. Am J MedGenet 63:492–500.

Mutton D, Alberman E, Hook EB. 1996. Cytogenetic and epidemiologicalfindings in Down syndrome, England and Wales 1989 to 1993. J MedGenet 33:387–394.

Rothman KJ. 1986. Modern epidemiology. Boston: Little Brown. p 84–89.

Wilson MG, Lin MS, Fujimoto A, Herbert W, Kaplan FM. 1989. Chromo-some mosaicism in 6,000 amniocenteses. Am J Med Genet 32:506–513.

Excess Females in 47,+21 Mosaics 319