Database limitations on the evidentiary value of forensic mitochondrial DNA evidence.

• Author: Kaestle, Frederika A.

ABSTRACT: Mitochondrial DNA (mtDNA) typing is increasingly being offered in criminal jury trials as proof that the defendant is a possible contributor of DNA found at a crime scene. As a prerequisite to introducing such evidence, the prosecution typically must estimate the frequency in the general population of the mtDNA sequence found in the defendant and the crime scene so that jurors can evaluate the probative value of the defendant's inclusion as a potential contributor. The government estimates sequence frequencies by comparing the observed sequence to sequences listed in a racially categorized mtDNA database developed and maintained by the Federal Bureau of Investigation and the Scientific Working Group on DNA Evidence. While mtDNA evidence has significant potential as a law enforcement tool, the SWGDAM database is currently too small and insufficiently representative to provide meaningful estimates of sequence frequencies. Most importantly, the database fails to account for historic and recent human migration patterns that, because mtDNA is maternally inherited and not recombinant, have resulted in significant regional differences in sequence frequencies. With further sampling and study, large regional databases may prove to be an effective and feasible improvement upon the current forensic database for the calculation of meaningful frequency estimates. However, until such databases and meaningful frequency estimates exist, mtDNA evidence is not yet ready for admission in criminal cases to permit inferences that suspects left mtDNA at crime scenes.

INTRODUCTION

Although the public and the legal community are now familiar with nuclear DNA, first admitted as evidence in United States courts almost twenty years ago, (1) nuclear DNA has a less famous counterpart found in the mitochondria of human cells, known as mitochondrial DNA ("mtDNA"). In recent years, law enforcement has increasingly used mtDNA evidence as a tool of both exclusion and inclusion of individuals as suspects in criminal cases. Specifically, because mtDNA exists in greater copy numbers per cell than nuclear DNA, mtDNA is becoming the primary type of forensic evidence extracted and reported from hair shaft samples and degraded DNA. A person may be excluded as a suspect if his mtDNA "profile, " i.e., his mtDNA sequence in particular regions of his full mtDNA strand or "genome, " differs from the profile of the crime scene sample. Such exclusion evidence may be relevant both at the pretrial investigation stage--to prevent wrongful accusation and conviction of innocent persons--and in post-conviction proceedings, such as Innocence Project DNA exonerations, to correct wrongful convictions. Use of mtDNA typing for exclusion purposes has been widely embraced in the scientific and legal communities because it preserves law enforcement resources by removing red herrings from investigations and focusing attention on the true perpetrator(s). (2)

As a tool of inclusion, however, mtDNA typing is more controversial. In criminal cases involving mtDNA, the prosecution typically reports that a defendant is included as a suspect if his mtDNA profile is consistent with, or "matches, " (3) the profile in a crime scene sample. (4) Most courts also require the prosecution to present an estimate of this shared mtDNA profile's frequency in the relevant population, (5) on grounds that, without such an estimate, jurors cannot meaningfully assess the probative value, if any, of the defendant's inclusion as a potential contributor. To estimate such frequencies, an analyst typically compares the suspect's mtDNA profile to a forensic reference database compiled and maintained by the Scientific Working Group on DNA Evidence ("SWGDAM"), a group sponsored by the Federal Bureau of Investigation ("FBI"). (6) In comparing a suspect's sequence to the SWGDAM database, the analyst counts the number of times the sequence appears in various sub-databases organized by the self-reported "race" (7) of the sample contributors, then uses this number to estimate the true frequency of the sequence in each race-based population. (8) Because of the small size of these databases and the diversity of mtDNA profiles, (9) this approach usually yields zero "hits" in the database. Thus, the estimated frequency of the sequence reported out is often less than 0.1%. (10) Such a low estimate is potent evidence in a criminal jury trial because it suggests that a randomly selected individual has only a 1 in 1000 chance of sharing the profile observed in the crime scene sample.

Of course, a frequency estimate is only as good as the statistical method used to calculate it. If the method is invalid, such as if the database is not representative of the relevant population, the resulting frequency estimates may be inaccurate and, if so, will not give the jury a meaningful way to evaluate the probative value of the reported inclusion. In turn, if the probative value is unknown or inaccurately reported, most courts would rule that the evidence of inclusion is inadmissible based on rules of relevance and novel scientific evidence. Thus, whether the SWGDAM database is a valid tool to estimate mtDNA sequence frequencies is a critical question courts should ask in determining whether to admit evidence of mtDNA inclusions. Neither forensic scientists nor attorneys frequently present these questions to courts in a considered way. This Article is an attempt to encourage the legal and scientific communities to view these questions with a more critical eye.

Part I of this Article briefly discusses the fundamentals of mtDNA biology and forensic typing methods and how they differ significantly from nuclear DNA biology and typing methods. Part II explains mtDNA's forensic applications and the particular methods used by the FBI to type mtDNA sequences and estimate sequence frequencies using population databases.

Part III contends that reported mtDNA frequency estimates are currently misleading because the SWGDAM database from which the estimates are calculated is neither representative of the general population nor of the various sub-populations it professes to characterize. First, the SWGDAM database is an incomplete, non-random, non-representative collection of mtDNA profiles compiled without regard to geographic patterns of genetic clustering that have resulted from cultural, political, historical, and economic forces. Second, current reliance in criminal cases on estimates derived from comparisons to the SWGDAM database is misplaced given the poor quality control measures of the database and the manner in which the assessment method is skewed toward reporting an inclusion.

Part IV examines both the general principles governing the admissibility of scientific evidence in most jurisdictions and courts' treatment of mtDNA inclusion evidence. Part IV argues that, because the estimation of profile frequencies using the SWGDAM database is currently controversial and of questionable validity, evidence of mtDNA inclusions does not yet meet most jurisdictions' legal standards for admissibility.

The Article concludes by prescribing various measures to improve the quality and integrity of forensic mtDNA typing. It is the expectation of the authors that, once mtDNA evidence is properly understood in its full scientific context, and once statistically valid databases can place a true probative value on mtDNA evidence, it will be reliable, highly relevant, and properly used in criminal investigations and prosecutions. Today, however, the state of mtDNA evidence presents an unacceptable risk of accusing or convicting the innocent based on inaccurate and misleading scientific evidence.

1. THE BASICS OF MTDNA TYPING AND How IT DIFFERS FROM NUCLEAR DNA TYPING

   1. Differences in the Biology of MtDNA and Nuclear DNA

      Deoxyribonucleic acid ("DNA") exists in every human cell and contains genetic codes inherited from previous generations. Humans have two types of DNA: nuclear DNA ("nDNA") and mtDNA. The mtDNA genome is distinct from the nDNA genome, and the two types of DNA differ in terms of their location within the body, genome size, and genetic makeup. While nDNA is bundled within chromosomes in the nucleus of most human cells, mtDNA exists outside the nucleus in energy-producing organelles called mitochondria. The mtDNA genome is also much smaller than that of nuclear DNA; while the nuclear genome consists of approximately three billion base pairs, (11) the mtDNA genome contains approximately 16, 569 base pairs. (12) The mtDNA genome consists of two primary regions: a coding region, which regulates the production of various biological molecules, and a control region, which regulates replication of the mtDNA molecule itself. (13) The control region, approximately 1125 base pairs long, (14) is the only significant portion of the mtDNA strand that does not code for genes. (15) In contrast, the nDNA genome contains coding regions spread throughout the twenty-three chromosomes that are known to have a genetic purpose, surrounded by regions of so-called "junk" nDNA, for which scientists have yet to find a genetic purpose.

   2. Differences in Forensic Typing of MtDNA and Nuclear DNA

      To distinguish one individual's DNA from that of another, forensic scientists look to particular locations within non-coding regions of the nuclear and mtDNA genomes that are highly variable among humans and therefore have discriminating power. In nuclear DNA typing, scientists typically look to thirteen locations along an individual's nDNA strand identified by the FBI as particularly suitable for forensic testing and used by the FBI to generate the profiles contained in its Combined DNA Index System (CODIS). A person's forensic nDNA "profile" consists of the twenty-six alleles he exhibits at these thirteen "CODIS loci." (16)

      To compare individuals' mtDNA strands, most forensic scientists focus on two regions within the mtDNA control...