 |
 |
 |
 |
 |
|
DNA Testing in Kinship Analysis.
Teresa Aulinskas1, Ph.D., George Riley1 Ph.D., Anne Pace1, B.S., John Houck1, B.S., Mellissa Krug1, B.S., Sarah Bockelman1, B.S., Alexis Howard1, B.S., Bruce Buckingham1, B.S., Howard C. Coleman1, B.S., B.A, and Charles Brenner2, Ph.D.
INTRODUCTION Genelex Corporation has successfully used RFLP (restriction fragment length polymorphism) DNA testing to perform kinship analyses of both extended and limited pedigrees in civil and criminal casework. The analysis develops DNA profiles of test subjects or questioned evidence at up to twelve different VNTR (variable number of tandem repeat) loci. Generally accepted population genetic and Bayesian statistical techniques are used to generate a series of reports comparing the various possible ways in which the subjects may be related. THE LARRY HILLBLOM CASE In the "Hillblom" sibship case, Genelex Corporation was asked to determine if five children, each with a different mother, could be the offspring of Larry Hillblom, the deceased founder of DHL courier service. The mothers had all been teenagers at the time of birth and lived in different Pacific Rim countries such as the Philippines, Viet Nam and Saipan. In this high profile civil case, the Attorney General of California represented the estate valued at approximately $800 million. A histological section from a mole purportedly excised from Hillblom and samples from Hillblom's mother and full brother were not made available for testing, despite repeated requests by the plaintiff's attorneys. The attorneys for four of the children then decided to proceed with testing to determine if the child plaintiffs had the same father. Genelex initially tested four mother/child pairs. Subsequently another mother/child pair came forward. We were asked to determine how many, if any, of the children had a common father. INDIRECT STRATEGY Following testing of twelve VNTR loci by RFLP, an indirect strategy was employed to analyze the data. We calculated the probability of half-sibship or unrelatedness in a series of comparisons between each of the tested mother/child pairs. In general, the more genes that children have in common, the more likely it becomes that they have a common parent. If three children have a common father, at least two of them must share the same paternal allele at each genetic locus. Any two children that have the same father are expected to also have the same paternal allele at approximately one-half of the loci tested. Whenever a third version of the gene appears, it means that there is more than one father or that a mutation has occurred. A common father is impossible if a third gene is observed at more than two loci.
Table 1 (Sibship RFLP results) shows shared paternal alleles at the VNTR loci in the Hillblom case. Matching alleles at each locus (row) are given the same letter designation. Actual sizing data is not presented because of confidentiality requirements. Three paternal alleles were observed at some of the loci in the four children initially tested. In general, the more alleles that two children have in common, the more likely it becomes that they have a common parent. The results indicated that child 2 did not appear to have the same father as child 1, 3 and 4.
Table 1. Shared paternal alleles at the VNTR loci tested in the Hillblom case. Matching alleles at each locus (row) are given the same letter designation. STATISTICAL APPROACH Statistical techniques were used to generate a series of reports comparing the various possible ways in which the subjects may be related. Pair-wise and then three-way comparisons between the mother/child pairs were made. The Baysian approach was used and likelihood ratios (LR) summarized the genetic evidence. A half-sibship likelihood ratio of less than one suggests that the individuals do not have the same father. A half-sibship likelihood ratio greater than one suggests that they do. A series of pair-wise comparison were made. We asked to determine whether two children were half-siblings, sharing a common father or whether they were they unrelated. The data from the twelve loci showed that when child 1 and 3, 1 and 4, and 3 and 4 were compared, the likelihood ratio and probability for half-sibship were very high. Conversely, when child 2 and 1, 2 and 3, or 2 and 4 were compared, the likelihood ratio for sibship fell below one. The data was therefore reported as the likelihood ratio for unrelatedness. The probability of unrelatedness was greater that 99%. Child 2 appeared to be unrelated to the other children.
A series of three-way comparisons were then performed. We asked to determine whether three of the children were half-siblings, sharing a common father or whether two of the children were half-siblings and one unrelated.
The data from the three-way comparisons indicated that when child 1, 3 and 4 were compared with children 1 and 3, the likelihood ratio and the probability for half-sibship were extremely high. Similar results were obtained with all combinations of children 1, 3 and 4.
The data was at least 40 million times more likely if children 1, 3 and 4 had the same father than if any one of them was unrelated. Conversely, when similar comparisons were made including child 2, the likelihood ratio for half-sibship was zero. The results demonstrated that child 2 was unrelated. Seven other DNA testing laboratories were involved in this case either at sample collection, retesting or case review levels. RFLP versus STR Could PCR (polymerase chain reaction) testing of STRs (short tandem repeats) have resolved the Hillblom case? Genelex has been performing STR testing for forensic casework, the typing of sex offenders and for paternity casework since 1997 and has STR typed more than 20,000 individuals. We wanted to compare the results of testing using both RFLP and STR in order to be sure we are using the best method in the complicated cases we are frequently called upon to perform. STR analysis is performed using the GenePrint PowerPlex 1 and 2 Fluorescent STR systems from Promega Corporation, customized for the Hitachi FMBIO II Fluorescent Scanner. Each PowerPlex system allows the simultaneous amplification and single lane detection of eight or nine STR loci. The fluorescent scanner performs automated three color detection of the STR loci and the internal lane marker.
Figure 1. PowerPlex 1.2. DNA was extracted from whole blood using an inorganic extraction method (Dyke salting-out method). Amplification using the GenePrint PowerPlex 1.2 (Promega Corporation) was as recommended by the manufacturer for the Perkin Elmer 480 thermal cycler. Amplified samples were separated by PAGE on 43 cm gels (SA-32, Life Technologies Inc.) using 4.5% acrylamide:bisacrylamide (19:1), 1xTBE, 7M urea. Electrophoresis was at 65 W for 1.5 hr (pre-run ~0.3 hr at 65W). Bands were detected by scanning with an FMBIO II (Hitachi) and were analyzed with Hitachi Analysis 6.0 and STARcall software. PowerPlex 1 allows for the simultaneous analysis of eight STR loci and the amelogenin gender-typing locus. PowerPlex 2 system allows the analysis of six additional STR loci. Three loci are common to the two systems. A total of fourteen STR loci can be typed with the two systems. See Figures 1 and 2. SIBSHIP: RFLP versus STRs Case One is a sibship case similar to the Hillblom case. We asked to determine whether two children were half-siblings, sharing a common father or whether they were they unrelated. Likelihood ratios for half-sibship were calculated using RFLP, PowerPlex 1, and a combination of PowerPlex 1 and 2. Data were calculated with and without the Penta E locus. RFLP results indicated that all children were half-siblings.
Figure 2. PowerPlex 2.1. DNA profiles were developed as described in Fig 1. Amplification using the GenePrint PowerPlex 2.1 (Promega Corporation) was as recommended by the manufacturer for the Perkin Elmer 480 thermal cycler.
The results in this sibship case indicated that PowerPlex 1 and 2 were less powerful and likelihood ratios were generally lower, as expected, due to the lower power of discrimination of STRs. The numbers in parentheses, represent the number of loci at which the children did not share the paternal allele. Interestingly, when child 2 and 4 were compared using the PowerPlex 1 system, they shared the paternal allele at all loci, a pattern typically seen in a parent/child relationship. When all loci were tested, child 2 and 4 shared all loci except one, giving an unusually high likelihood ratio with PowerPlex 1 and 2. In general, RFLP data was clearly more powerful. The second case is an extended pedigree case in which we were asked to determine whether the tested man was the paternal uncle of the child or whether they were unrelated. Only two persons were tested, a challenge even for RFLP.
The results obtained supported the hypothesis that the tested man was the paternal uncle. Likelihood ratios were calculated. Seven RFLP loci gave a reasonable indication that the tested man was the paternal uncle. STR data were much less informative in this case, with the cumulative likelihood ratio for 14 STR loci falling below 60. The Penta E locus was again quite informative. As a result of these comparisons Genelex has chosen to continue routine use of RFLP testing whenever possible in complicated kinship cases. STR testing is reserved for cases involving unusual and degraded samples such as decomposed bone and tissue, amniotic fluid, where turnaround is crucial, and in cases requiring extended typing as a result of mutations. When necessary the systems can be combined to test a total of twenty-eight loci, resulting in an extraordinarily powerful kinship test. REFERENCES 1. Validation of PowerPlex STR multiplex and Amelogenin sex identification typing kits using the FMBIO II fluorescent scanner: Forensic casework and high throughput convicted offender databanking. George R. Riley, Ph.D., et al. In: Proceedings from the Eighth International Symposium on Human Identification. 1997, Promega Corporation pp. 53 - 55 2. Charles Brenner, PhD., DNA View Software Program. 3. Automated Fluorescent Detection of 8-locus and 4-locus STR Multiplexes. Schumm, James W,. et al. In: Proceedings from the Eighth International Symposium on Human Identification. 1997, Promega Corporation pp. 78-81. 4. Automated Fluorescent Detection of STR Multiplexes - Development of the GenePrint PowerPlex and FFFL Multiplexes for Forensic and Paternity Applications. Schumm, James W,. et al. In: Proceedings from the Seventh International Symposium on Human Identification. 1996, Promega Corporation pp. 70-80. 5. Promega Corporation, GenePrint STR System Technical Manuals, # TMD008, TMD011.
|
||||
