Blackwell Science, LtdOxford, UKTRFTransfusion0041-11322006 American Association of Blood BanksFebruary 2006462199203Original ArticleONE- VERSUS TWO-PARENT PATERNITY TESTSWENK ET AL.
PARENTAGE TESTING
Maternal typing and test sufficiency in parentage analyses
Robert E. Wenk, Terry Houtz, and Francis A. Chiafari
M
ost domestic cases of disputed paternity
BACKGROUND: The contribution of maternal typing to involve study of trios consisting of a child, its
paternity analysis was evaluated to determine how many assumed biologic mother, and an alleged
additional loci to study in one-parent cases. father. In many immigration and forensic
STUDY DESIGN AND METHODS: Four groups parentage cases, however, only a child and one alleged
underwent paternity analyses with an eight-locus test parent undergo genetic tests. When these duos are evalu-
battery. Files of 25 case trios were retrieved, in which ated, there is loss of information provided by the certain
alleged fathers had achieved paternity indices of greater knowledge of some of a child’s maternal obligate alleles.
than 100 (“included trios”). Maternal types were omitted One effect of information loss is that a conclusion of
and the cases were reanalyzed (“included duos”). Mother- included paternity may be erroneous. False inclusions of
child pairs of the cases were then coupled with unrelated paternity have been reported anecdotally in immigration
men (“excluded trios”), and the cases were analyzed. cases1-3 and in nonhuman populations.4
Maternal types were omitted from the excluded trios and Study of the biologic mother can reduce the number
cases were reanalyzed (“excluded duos”). of paternal obligate alleles observed in a child. If the
RESULTS: Paternity indices of men in included duos alleged father is not the child’s biologic father, then
were markedly reduced when compared to included trios; restricting the paternal obligate alleles can increase both
odds were sufficiently low in 9 of 25 men that paternity the number of detected genetic inconsistencies and the
remained in doubt. After omission of maternal probability of achieving the criteria for ascertaining
phenotypes, excluded duos exposed 33 percent fewer the man’s nonpaternity. Alternately, if the alleged father is
genetic inconsistencies than excluded trios; 5 of 25 men the child’s biologic father, then the mother’s observed
in excluded duos demonstrated less than two genetic genetic contributions can increase the odds of paternity
inconsistencies and 1 man had none. The specific because the presence of fewer paternal obligate alleles in
probabilities of paternity exclusion in motherless cases a child decreases the probability that an unrelated (“ran-
averaged 61 percent per locus of those in case trios. One dom”) man contributed them.
random man in 52 duos was not excluded by the eight The following study attempts to quantify the contri-
tests versus 1 in 417 trios. bution of maternal typing to the analysis of paternity tests
CONCLUSIONS: Omission of maternal typing from eight when using common microsatellite DNA markers (short-
common microsatellite paternity tests reduced tandem repeats [STRs]). The strength of the evidence for
conclusive evidence for or against paternity by 30 to or against a man’s parentage is considered in light of the
40 percent. False inclusion of random men is an maternal information and current standards for drawing
important failing of tests in motherless cases. Cases conclusions about paternity. Conclusions about test suffi-
involving one parent and child (e.g., in immigration) ciency are applicable to any one-parent case.
would require examination of an additional five similar
loci to compensate for absent maternal data. A change
in standards is suggested. ABBREVIATIONS: AOH = apparent opposite homozygosity; PI(s)
= paternity index(-es); RMNE = random men not excluded; STR(s)
= short-tandem repeat(s).
From BRT Laboratories, Baltimore, Maryland.
Address reprint requests to: R.E. Wenk, BRT Laboratories,
400 West Franklin Street, Baltimore, MD 21201; e-mail:
rwenk@lifebridgehealth.org.
Received for publication March 22, 2005; revision received
June 10, 2005, and accepted June 13, 2005.
doi: 10.1111/j.1537-2995.2006.00701.x
TRANSFUSION 2006;46:199-203.
Volume 46, February 2006 TRANSFUSION 199
, WENK ET AL.
MATERIALS AND METHODS sampling. Five excluded trios in the repeat study group
required study of more loci than the constant eight to
Cases were chosen retrospectively so that the same eight satisfy internal laboratory requirements for parentage
loci, and no other loci, were studied in every case. Individ- exclusion (three genetic inconsistencies with at least one
uals in each case were typed at the following STRs: TPOX, direct inconsistency).
TH01, vWA, D16S539, D7S820, D13S317, D5S818, and Genetic inconsistencies were classified as “direct,”
CSF1PO (PowerPlex system, Promega Corp., Madison, “indirect,” and “apparent opposite homozygosity” (AOH).6
WI). Four groups were examined. Direct inconsistencies involve both a heterozygous
Twenty-five actual case trios that had been investi- alleged parent and a heterozygous child, but the two indi-
gated with only the eight constant loci were chosen unse- viduals either have no alleles in common or share an allele
lectively from files. In each case, paternity of the alleged that is identical to the maternal obligate allele. (Direct
father had been determined according to currently inconsistencies of the second kind are lost when the
accepted criteria—there was no locus with a genetic mother is not typed.) Phenotypes denote genotypes in
inconsistency and the odds (paternity index [PI]) direct inconsistencies. Except for the possibility of muta-
exceeded 100 in every case:5 This group of cases is termed tions, historically, direct inconsistencies have been con-
“included trios.” sidered qualitatively strong exclusionary evidence and
The 25 actual cases were restudied after omitting have paternity indices of zero.
maternal phenotypes (i.e., only types of the father-child Indirect inconsistencies involve either an alleged
pairs were considered). This group is termed “included parental or a child’s phenotype that consists of only one
duos.” The PI in each included duo was compared with the observed allele, raising the possibility of erroneous exclu-
PI of the original included trio and the proportion of duos sionary information6 if an STR allele is silent (because it
was determined that achieved the PI of greater than 100 “dropped out,” was missed, was not recorded, etc.). A
criterion to ascertain parentage. silent allele is defined here as one that is not evident in the
The mother-child pairs from the same 25 actual cases parentage test because of a sporadic technical, observer,
were reconsidered after substituting an unrelated man for or clerical failure. Thus, indirect inconsistencies involve
the original man (biologic father). The substituted man phenotypes that do not denote genotypes and have been
was arbitrarily drawn from another of the 25 cases, except long considered to be weaker exclusionary evidence than
that he was of the same ethnic group as the original man direct inconsistencies. (The actual frequency of misinter-
in the case. This group of simulated cases is termed pretation of monoallelic STR phenotypes in human par-
“excluded trios.” The number of loci was determined that entage analyses is unknown.) Despite more questionable
showed genetic findings inconsistent with paternity and interpretations, indirect inconsistencies are given pater-
the number of cases was recorded that demonstrated two nity indices of zero.
or more genetic inconsistencies, the minimum necessary AOH involves a monoallelic phenotype in both the
to meet standards for concluding nonparentage.5 alleged parent and the child. The two individuals are most
After omitting phenotypic data of biologic mothers, likely homozygotes who are unrelated and possess no alle-
the paternity of the children by the substituted men was les in common.7 There is a possibility, however, that the
retested. This last group of 25 cases is termed “excluded two alleles (genotypically identical or not) are silent, one
duos.” The number of genetic inconsistencies found in in the alleged parent and another in the child. There is a
the excluded duos was compared with the number third possibility too—that the two people are parent and
in excluded trios. The proportion of excluded duos was child and possess a null allele in common. A null STR
determined that met criteria to conclude that substitute allele is defined as one that is not observed because its
alleged fathers were not biologic fathers. heritable DNA sequence cannot be amplified under con-
An additional 25 cases with substituted alleged trolled test conditions. (Usually, null alleles arise from
fathers were simulated to determine the repeatability of mutations in the primer target sequence that flanks the
findings in the first 25 excluded trios and excluded duos. STR repeat.) The possibility of a null allele is the reason
Reasons for the confirmatory study were: a larger pool of why AOH is not considered as a genetic inconsistency,
substitute alleged fathers was desirable for sampling ran- but as a low probability event that is consistent with
dom men and an ascertainment bias may have occurred parentage.8
by selecting cases limited to study of only eight loci. Cases
may have been inadvertently selected for infrequent alle-
RESULTS
les so that they easily met criteria for paternity inclusion.
Substitute alleged fathers were chosen at random from There were 17 US European-American and 8 US African-
ethnic groups of the original men in the cases, but not American trios (and duos) considered in each of the four
from the cases themselves (unlike the original 25 excluded initial study groups. The phenotypes of the eight constant
trios and duos), avoiding a possible bias caused by limited loci were obtained without exception in all individuals.
200 TRANSFUSION Volume 46, February 2006
