Control of contamination must begin with the physical separation of DNA extraction and amplification set-up activities (pre-PCR) from PCR product analysis (post-PCR). In addition, all steps in the extraction and amplification set-up procedures should be performed in a hood. Areas for extracting low levels of mtDNA from skeletal remains should be separated from high level extraction areas. Anterooms or vestibules may be placed at the entrance of each laboratory where mtDNA testing is being performed to help prevent the transfer of PCR product from one laboratory to another. When possible, mixing of the air circulation between pre-PCR and post-PCR areas should be minimized.

Dedicated equipment should be used for amplification setup and in PCR product areas. Positive displacement or plugged tip pipettors should be used for aliquoting samples and PCR reagents. Use of laboratory coats and gloves are essential. Laboratory coats should be disposable or dedicated to the areas where mtDNA PCR product is being handled. Care should be taken when wearing disposable gloves not to touch any surface which may contain a contaminant such as the surface of the skin, eye glasses, clothing or even a cleaned bench-top. As a common practice, before handling evidence or items which come in contact with evidence, always change gloves or wipe gloves with bleach, allowing the gloves to air dry.

General cleaning practices are important for controlling contamination. The universal cleaning agent for PCR contamination is 10% commercial bleach (7 mM sodium hypochlorite). The bench-tops, hoods, and any surface which comes in contact with the evidence or DNA extract should be washed with bleach frequently. In addition, the floors of each laboratory should be periodically washed with bleach.

To illustrate the impact of proper laboratory design and practices, a representative number of cases were evaluated from the AFDIL laboratory. One set of data represented the conditions under which cases were processed in 1991. The second set represented cases performed in 1994. The number of times PCR product was observed in an extraction reagent blank control or a

PCR reagent blank control was counted for each case. For seven cases processed in 1991, 150-. of the extraction reagent blanks and 7% of the PCR reagent blanks showed PCR product following amplification. For five cases processed in 1994, only 8% of the extraction reagent blanks and 3% of the PCR reagent blanks showed PCR product following amplification. These numbers indicate how sensitive mtDNA testing is to contamination, but also illustrate that significant progress can be made to limit the occurrence of contamination with improvements to laboratory design and practices.

Appropriate controls and measures to detect contamination are imperative. Both extraction reagent blank and PCR reagent negative controls should be run in every case to detect the occurrence of contamination. Documentation of contamination will allow for review of the major sources and will assist in developing a comprehensive quality control program for controlling PCR product contamination.

Where a contaminant is detected, testing should be repeated, if possible, unless the sequence from the skeletal remains is otherwise determined to be reliable. There may be instances when contamination prevents the reporting of results.

In general, other than mtDNA testing of skeletal remains, it is uncommon to find amplification product in a reagent blank during any PCR-based testing. Accordingly, the AFDIL rarely finds a positive amplification product in the extraction reagent blank for a whole blood maternal reference.

When contamination does occur, it will typically result in an apparent nonmatch (false exclusion), not a spurious match (false inclusion). Therefore, all exclusions should be carefully scrutinized. The AFDIL procedures preclude a false match due to contamination by a mtDNA blood reference. All blood specimens are tested in a different area of the laboratory. Moreover, the whole blood maternal reference is usually processed after the skeletal remains, and in most cases after the results generated from the skeletal remains have been reported. Only during databasing operations will there be the occasional situation when reference results are generated prior to the skeletal remains results.

An important quality check of the skeletal remains sequence is the independent extraction and testing of multiple bones. Multiple bones are independently extracted in most cases by the AFDIL, unless only a single specimen is available for testing. In addition, the primer sets used for amplification are overlapping, providing further confirmation of the authenticity of the skeletal sequence.

Given the inevitable random contamination inherent in mtDNA testing, redundancy, when possible, at the level of source material is key. Cautionary statements are important where replicate testing cannot be performed.

H. PCR Amplification Ambiguities

Taq polymerase, an enzyme used in current PCR amplification reactions, is known to occasionally and randomly misincorporate erroneous nucleotide bases at a rate of approximately one to ten in 10,000. Generally this is not problematic, because correct sequences will overwhelmingly predominate. However, if the starting target sequence consists of only a few copies, then the chance of a false result from misincorporation during the first few rounds of thermal cycling becomes a possibility.

Nonspecific priming may occur, particularly where the starting conditions are not optimal. This may result in ambiguous or errant sequence when the starting DNA concentration is very low.

During PCR amplification, the polymerase enzyme may stop due to template fragmentation or damage. The partially extended sequence may then anneal to another template fragment in the next cycle. In fact, several fragments may be assembled in this process to recreate the original full length sequence. This process, known as "jumping PCR", may complicate the interpretation of diploid sequences. In mtDNA, where only one sequence is present, "jumping PCR" may have advantages. However, it may also produce errors from incorrect assembly of fragments so small that they have lost their specificity. Duplication of testing results or overlapping sequence data will allow for interpretation of these occurrences.

I. Sequencing Ambiguities

The evaluation of DNA sequencing data is extremely tedious and time consuming. Evaluation of data using manual sequencing methods may result in transcription and reading errors. Automated sequencing methods will minimize reading errors. Nonetheless, sequencing errors do occur in automated analysis; rates have been published for current instrumentation. Most instrumental error can be avoided by limiting the information read to shorter lengths, because the vast majority of errors occur near the end of a sequencing run as the resolution and strength of the base signals diminish. The automated sequencing instrument used by the AFDIL has an error rate of approximately 1 to 2% for the size templates analyzed (200-300 base pairs). This error rate increases to greater than 10% when longer templates (greater than 450 base pairs) are analyzed using current

conventional sequencers. The error rate does not represent errors in the sequence reactions, but instead represent errors in the ability of the instrument software to make an accurate base call.

In large scale sequencing operations, where a maximal quantity of sequencing information is emphasized, some errors can be tolerated. This is not the case in the forensic context. Visual evaluation of the data is mandatory to prevent read errors. Moreover, it is standard practice to confirm sequences, usually by checking complementarity of the reverse strand sequence. With the advent of new sequencing technologies, it may be possible to confirm sequences by a second technology which may not have a tendency toward the same systematic errors. At least two qualified individuals must independently evaluate the sequence in order to ensure that the results accurately represent the data, and to catch transcription errors.

It should be recognized that redundancy in casework also provides an opportunity to catch errors; a case involves multiple bones and at least two family blood reference specimens where available. These procedures can virtually eliminate sequencing errors.

J. Casework Experience

Since October of 1994 when the AFDIL began full production of ten cases per month, it successfully obtained sequence information from Southeast Asia, Korean, and World War II cases representing 37 individuals. The AFDIL has achieved a success rate for obtaining DNA sequence information of greater than 95% from

CILHI casework since late 1992, even for retested cases from 1991 Southeast Asia casework when the original success rate was only 40% This does not mean that the AFDIL obtains sequence information from every bone, nor does it mean that the AFDIL obtains full sequence information in every case. The AFDIL obtains sequence results from little more than half of the bones it tests. CILHI casework is not routine in the sense that, in virtually every case problems are encountered that require retesting, often using modified test conditions. In the first six months of production mtDNA sequencing operations, an average of 3.7 bones were tested per case, 42% of extractions had to be repeated, 37 amplification reactions were required (2.5 times the minimum), and 55.5 sequencing reactions were required (twice the minimum)

[these numbers do not include the number of controls run per case]. Although these numbers are high, they are expected efficiencies for "ancient DNA" testing. Thus, despite difficulties, the AFDIL is now able to obtain mtDNA information from the vast majority of cases. The results from recent Korean cases suggest that Korean war remains may not be any more problematic than those from Southeast Asia, presumably due to better preservation of the remains.

MtDNA identifications from ancient remains is a scientific tour de force, at the cutting edge of today's capabilities. As single DNA molecules can be detected, some degree of cross-contamination is inevitable, especially in a scaled up production facility. The Task Force nevertheless concludes that with appropriate control measures (redundant testing and meticulous lab hygiene) these problems are surmountable, and a good record has been presented in the currently on-going work.

The Task Force finds that the present probability of coincidental matches between mtDNA control region sequences is no more than a few percent. Once sequences from 500 members of a population have been determined, precise statements about the chance of a false association of a set of remains with a family will be able to be made. Published data may be of value, but samples will be needed from Southeast Asian populations. The precision is expected to suffice in the vast majority of cases, given other non-DNA evidence, to effect the mtDNA identification of unassociated Korean remains. It will not be possible to identify every bone. A great deal can be done with anatomical and historical evidence alone.

The Task Force finds that control of contamination is essential to PCR-based laboratory testing. Some contamination is unavoidable, particularly in mtDNA testing of ancient remains, but it does not preclude reliable casework testing where redundancy, good laboratory practices, and appropriate cautionary language are used and constant oversight is maintained.

The Task Force finds that casework experience demonstrates capability to type Korean skeletal remains.

III. RELIABILITY

DSB TOR: To evaluate current and emerging scientific evidence concerning the reliability of the techniques when compared with other current and evolving methodologies. [Is the current MTDNA sequencing identification method reliable? Are quality assurance efforts satisfactory? What further measures would enhance reliability? What studies are necessary to validate a new DNA typing methodology on ancient skeletal remains? What continuing scientific oversight or advisory body should monitor these DNA identification efforts? What quality assurance mechanisms or measures should be implemented?]

The scientific community believes that current mtDNA sequence identification technology is reliable. Forensic laboratories in the United States and Great Britain have begun to use it in casework, recognizing they must be able to defend the technology in court, if needed.

The component technologies of DNA extraction, PCR amplification, and DNA sequencing are all validated, having been used in research and service orientated molecular biology laboratories throughout the world for many years. Furthermore, their application to so-called ancient DNA is well established in the scientific literature.

The studies generally accepted by the forensic community to validate a new DNA testing technology have been articulated as guidelines by the Technical Working Group on DNA Analysis Methods (TWGDAM), sponsored by the Federal Bureau of Investigation (FBI). Such studies include the following: optimized standard source studies (such as studies on fresh body tissues and fluids, stored tissues and fluids, and samples from different tissues from the same individual); variance analysis (studies on measurement precision from known DNA controls); population studies (studies of population frequency distribution in different racial/ethnic groups); preservation studies (studies on tissues and fluids as would be typically found at a scene investigation, eg. dried stains); time/temperature studies (studies on samples incubated at various time and temperature); environmental exposure studies (studies on effects of various commonly encountered substances); evidentiary source studies (studies on nonprobative evidence from typical crime scenes); non-human studies (studies on tissues and fluids from common non-human sources); on-site evaluation

(studies to evaluate methodology transfer to a working forensic laboratory setting). The FBI is completing their validation studies on mtDNA typing in preparation for casework.

The TWGDAM Quality Assurance Guidelines address basic considerations in DNA analysis--such as analyst training, reagent quality control, evidence handling, analytical procedures, proficiency testing requirements, and method validation. Requirements for standard cell line positive controls and extraction controls are included. Requirements specific to RFLP analysis (such as precision of fragment size measurements) and PCR analysis (such as negative reagent blanks and separation of pre and post amplification areas) are also covered. TWGDAM has a subcommittee devoted to the forensic use of mtDNA sequencing and has recently adopted modifications to their guidelines for mtDNA sequence analysis in forensic laboratories. However, they do not concern themselves with application to ancient DNA.

It is not generally appropriate or possible to exhume remains of known individuals which have been buried for 30 years to validate the utility of a DNA typing system. However, indirect evidence of the validity can be found in the contexts presented elsewhere in this text.

Corroboration will ensue from the internal consistency of the findings. In some cases, corroborative evidence for or against an identification may appear from other documentary or physical evidence acquired after the DNA tests have been concluded.

Evidence of the efficacy of skeletal remains identification by mtDNA sequencing can be found in the identification efforts performed to date by the AFDIL. Repeatedly, bones from the same case yield the exact same sequence. The sequences obtained from the bone samples have matched those of the putative family members. often these sequences are unique, never having been seen before. The AFDIL processes the skeletal specimens before testing the family references), eliminating the possibilities of bias or cross contamination. Controls are run with all cases and the results are always checked against the mtDNA sequence of the staff processing the specimens. In one case, an exclusion was found by mtDNA testing; when CILHI received the test results, they suggested a second name association which was then confirmed by a match of the case sequence to that of the second putative family. The AFDIL duplicates the mtDNA sequencing casework of

new personnel as they begin actual casework--this data has not revealed any discrepancies to date.

The quality assurance requirements for adequate mtDNA testing of ancient DNA are demanding. The molecular anthropology community has reiterated many of the concerns in this area, particularly that of contamination and adequacy of controls.

Proficiency surveys do not yet exist specifically for mtDNA sequence identifications, although the AFDIL personnel have been submitting mtDNA sequence data to the College of American Pathologists (CAP) and Cellmark proficiency surveys. However, interlaboratory exchanges have been initiated by TWGDAM members. Data, to date, demonstrate the reliability of this method. The National Institute of Standards and Technology (NIST) is currently developing mtDNA reference material.

The greatest single potential for a mistaken DNA sequence result is from contamination. Some contamination events during laboratory testing of ancient remains are inevitable. However, contamination can be minimized through proper laboratory design and sample handling procedures and can be detected through the use of appropriate controls. The molecular anthropology community advocates the use of multiple negative controls, including a DNA carrier control. TWGDAM recommends negative extraction and amplification controls as well as a known positive amplification control.

The discrepancy case, "X-6", between the AFDIL and the outside laboratory, reinforces the critical importance of quality assurance measures. The AFDIL's ability to blindly replicate their testing results on the case two subsequent times and independent confirmation of AFDIL's results by the British FSS, demonstrate the reliability of results when proper quality assurance measures are employed.

Ambiguities and errors can occur during amplification and sequencing. Nonspecific priming can be reduced through well designed primer sets used under optimized conditions. Sequences should be visually checked and interpreted cautiously, in accordance with appropriate protocols. Sequences should be confirmed. Sequencing results should be independently verified by a second qualified analyst (double reading).

Redundant specimens or replicate extractions should be tested, where possible, to ensure that a given finding is not a chance result of a contaminant. Where redundancy is not possible, cautionary language should be used in the reporting of the results.

A principle of forensic testing that is applied in the military's urine drug testing program is that, after a test result is obtained, it should be confirmed by a second test employing another method. There is not a second method by which current mtDNA testing can be confirmed. However, new technologies are being developed which, when they become available, could have application in testing skeletal remains.

Mutations in the mitochondrial sequence are not common, but do occur. Mutations raise the possibility of a false exclusion. A single base change should therefore be considered "indeterminate", rather than an exclusion or an inclusion. The AFDIL methodology of requiring two (2) family references has already proven to be useful to resolve mutational events.

Continuing scientific oversight is a significant part of the military's quality assurance program. The AFIP is formally reviewed twice a year through a Scientific Advisory Board (SAB). The SAB reports are distributed to the AFIP Director and its Board of Governors, which includes the Army Surgeon General, the executive agent for the AFIP, the Surgeon Generals of the Navy and the Air Force, the Public Health Service, the Medical Director of the Veteran's Administration, and the ASD (HA), who has policy oversight of the AFIP. One SAB member is dedicated to review of the Department of Defense DNA Registry. Currently, this Board member is Maimon Cohen, Ph.D., Chairman of the Division of Genetics, University of Maryland Medical School.

The ASD (HA) has ultimate responsibility for oversight of quality assurance of all identifications of human remains performed on behalf of the military services. A plan has been published of the minimum standards for military or military-contract laboratories to follow in performing mtDNA sequence analysis of ancient skeletal remains (Annex C). These standards were submitted to the American College of Medical Genetics, the College of American Pathologists, TWGDAM, the American Society of Crime Laboratory Directors and the Department of Defense Clinical Laboratory Improvement office for review and comment before