COMMONWEALTH vs. ROBERT W. CURNIN.

WILKINS, J. We consider for the first time the admissibility of the results of tests comparing the DNA of a criminal defendant with DNA found at a crime scene. [Note 1] The defendant

Page 219

was convicted of rape of a child, burglary, and aggravated rape, and indecent assault and battery. We allowed his application for direct appellate review.

In this case, the DNA comparison, sometimes called DNA fingerprinting, [Note 2] tended to prove that semen found on a nightgown of a fourteen year old, handicapped rape victim was that of the defendant. Indeed, the evidence was that only one Caucasian in 59,000,000 has the same distinctive DNA components that were found in the DNA comparison test. Evidence of this nature, based on the scientific principle that every human has unique genetic characteristics and having an aura of infallibility, must have a strong impact on a jury. The erroneous admission of such evidence would undoubtedly be prejudicial in any case where, as here, the identification of the person who committed the crime is in serious dispute. We conclude that the results of DNA testing were improperly admitted in this case. The convictions must be reversed, and the case retried.

The judge considered the admissibility of the results of the DNA testing during a pretrial hearing. An expert witness employed by Cellmark Diagnostics laboratory (Cellmark), the company that conducted the test, carefully described the process by which the DNA in the stain found on the nightgown and DNA taken from a sample of the defendant's blood were put through various steps. Those steps concluded with an autoradiograph that permitted a comparison of genetic material in each sample with respect to four specific

Page 220

sections of human DNA (alleles) that are highly variable among humans. [Note 3]

Everyone agrees that the underlying theory and at least the general processes used by Cellmark are accepted in the scientific community. The defendant did not argue below, and does not argue here, that there were defects in the way the testing procedures were conducted in this case. His challenge is in effect a facial attack on the acceptability of the process Cellmark followed. He presented an expert witness who pointed to various problems that exist or may exist in the use of DNA testing for forensic purposes. The impurity of a sample of blood or semen taken from the crime scene may present problems in the testing process. The amount of DNA found at the scene may be insufficient to permit adequate testing, particularly retesting to verify results. The defendant contends also that there is no general acceptance of how such forensic tests should be conducted, of how controls should be included in the testing process, or of the standards of performance to which a testing laboratory should be held.

The judge rejected the defendant's challenges to the admissibility of the evidence. In findings and rulings issued from the bench, she stated that the process was accepted by the scientific community and that the tests in this case were properly conducted. She did not expressly discuss the defendant's challenge to statistical probabilities determined by Cellmark based on its test results. [Note 4]

Page 221

The use of DNA testing for forensic purposes is of very recent origin. There are several opinions approving the use of DNA test results to prove the identity of the person who committed a crime. In each of these cases, the defendant made no concerted challenge to the admissibility of the evidence by presenting expert testimony. [Note 5] More recently, some courts, instructed by informed challenges backed by expert testimony, have ruled that, at least in the particular case, DNA test results were not admissible. [Note 6] Various studies have been conducted or are under way to determine how, if at all, DNA testing should have a role in the proof of identity in a criminal trial. The problem is not in the scientific concepts involved but rather in how those concepts should be implemented.

We need not resolve the propriety of the forensic DNA testing conducted in this case because we conclude that there

Page 222

is no demonstrated general acceptance or inherent rationality of the process by which Cellmark arrived at its conclusion that one Caucasian in 59,000,000 would have the DNA components disclosed by the test that showed an identity between the defendant's DNA and that found on the nightgown. The question of the acceptance of DNA test results in the trial of cases will, no doubt, have to be determined in voir dire hearings in future cases. [Note 7] A determination that DNA test results are generally admissible in a criminal trial could aid defendants as well as prosecutors. Evidence that a properly conducted test produced no match would be exculpatory of the person tested.

We have usually applied the Frye test in deciding whether evidence produced by a scientific theory or process is admissible, that is, whether the community of scientists involved generally accepts the theory or process. Frye v. United States, 293 F. 1013 (D.C. Cir. 1923). See Commonwealth v.

Page 223

Mendes, 406 Mass. 201, 205 (1989); Commonwealth v. Fatalo, 346 Mass. 266, 269 (1963). The party offering the evidence "has the burden of showing the general acceptance by experts in the field of the reliability" of that evidence. Commonwealth v. Kater, 388 Mass. 519, 527 (1983). In making the determination whether the test is generally accepted, courts may properly consider not only the evidence in the record but also the reasoning and conclusions of other courts and the writings of experts. Id. In these circumstances, an appellate court makes its own determination without regard to the conclusions of the trial or motion judge. [Note 8]

The evidence and other material that may appropriately be considered do not warrant the conclusion that Cellmark followed a generally accepted or obviously logical procedure in deciding the likelihood that someone else would have the same DNA characteristics as those that were identified in the comparison test. The prosecutor presented no expert to support Cellmark's conclusion. The prosecution's expert on Cellmark's test procedures acknowledged that she was not qualified to give an opinion on the subject. She, in fact, made certain concessions about the possible unacceptability of the process that Cellmark used. [Note 9] The judge made no finding or ruling on the probability issue, even though the defense

Page 224

presented as an expert a population geneticist whose testimony indicated that the data on which Cellmark relied were not adequate to support its conclusion and that certain assumptions underlying Cellmark's approach were not shown to be true.

To determine the frequency with which alleles shown on a test appear in the population, one must have gathered and maintained parallel DNA information in a data base. Obviously, there will be a question whether the data base is appropriate, both generally and specifically as applied to a defendant. The approach taken by Cellmark was to determine, allele by allele, its frequency in the population in its data base and then to multiply each fraction to arrive at the probability of someone else having the same DNA characteristics that were revealed in the DNA comparison test. This process is known as the product rule. [Note 10] Cellmark compiled its Caucasian data base by testing 200 blood samples collected at a New York City blood bank.

The prosecution's expert granted that there had been "a lot of discussion recently about the population genetics of these highly variable probes," whether the alleles Cellmark used do "behave the same as other much more well characterized loci that have already been studied in human population genetics." For example, genetic marker analysis of blood stains for blood characteristics has been accepted because the nature of the distribution of those characteristics among the population is well-accepted by the scientific community. See Commonwealth v. Gomes, 403 Mass. 258, 273-275 (1988); Commonwealth v. Beausoleil, 397 Mass. 206, 217-218 n.15 (1986). There is no showing, however, that scientists agree

Page 225

generally that the distribution of the alleles disclosed in Cellmark's testing is random in the Caucasian population so as to warrant the calculations made by Cellmark.

The defendant's expert, Dr. Laurence Mueller, a population geneticist, testified that Cellmark's data base was not adequate for the purpose of producing an estimate of the frequency of finding a particular genotype in the human population. Cellmark did not have data on missing alleles in certain tests reflected in its data base. There was, in addition, the question whether there is significant substructuring (subgroups) within racial groups that would affect probability determinations using Cellmark's data base, and, if so, whether an acceptable statistical adjustment could be made to account for it. Data from a geographically representative population, rather than from a limited area, should be used. He testified that the forensic scientific community does not agree on what population studies are required (the breadth of the sample of humans and the appropriate level of ethnic information). Moreover, there is no agreement on the consequences of using extremely rare alleles. There are disagreements about using the product rule in the circumstances. [Note 11]

Page 226

There are various criticisms of the soundness of accepting certain assumptions made about the distribution of variable alleles, such as Cellmark used. [Note 12] Some criticisms may be invalid. Others may be inconsequential to the estimate produced. In this rapidly developing field, answers and solutions may have already been found about which we know nothing. It may even be that, by the time of the retrial of this case, the prosecution can support the admissibility of evidence of the probability of the alleles disclosed by the DNA test being found elsewhere in the relevant human population (and bolster its position that the DNA testing processes used by Cellmark are generally accepted by the relevant scientific community). It may also be that the relevant scientific community

Page 227

can generally agree on a means of arriving at a conservative estimate of the probability of another person having the same alleles and thus resolve all uncertainties and variables in favor of the defense. See Caldwell v. State, 260 Ga. 278, 290 (1990); People v. Wesley, 140 Misc. 2d 306, 332 (N.Y. 1988).

Based on the absence of the general acceptance or inherent rationality of the process by which Cellmark concluded that only one person in 59,000,000 would have the same alleles as were shown in the test it conducted, we conclude that the admission in evidence of the test results was prejudicial error. The defendant's convictions are vacated, and the case is remanded for retrial.

So ordered.

APPENDIX.

DNA is a six foot long molecule that contains coded information that determines the physical structure and operation of all living things. Because DNA molecules do not vary within a particular individual, a DNA molecule found in one cell of a particular person is identical to the DNA found in any other cell of that person.

In the human body, each DNA molecule is distributed across forty-six sections of the nucleus of the cell. These sections are known as chromosomes. The forty-six chromosomes form twenty-three pairs. One half of each pair is inherited from the mother, and the other half is inherited from the father.

Each DNA molecule is in the form of a double helix, and looks like a ladder which has been twisted to resemble a spiral staircase. The sides of the ladder are composed of repeated sequences of phosphate and deoxyribose sugar molecules. The rungs of the ladder, which are the most important aspect of the DNA molecule for the purposes of DNA testing, are constructed from the nucleotide bases of Adenine (A), Cytosine (C), Guanine (G), and Thymine (T). These bases pair up to form a single rung, known as a "base pair," but can join together only in accordance with the "base pair rule," which states that, because of chemical attraction, A will bond only to T, while C will bond only to G. Therefore, the only possible

Page 228

combinations which can form a rung on the ladder are A-T, T-A, C-G, and G-C.

A single DNA molecule contains approximately three billion rungs, or base pairs. Certain types of human genes that are called "polymorphic" can occur in alternate forms (that is, with differing sequences of base pairs), each of which is capable of occupying a gene's position on the DNA ladder. These alternate forms of genes are called "alleles," and are highly variable from one person to another. Alleles of a particular gene contain a different number of base pairs, and therefore are of different lengths.

Most of the sequences of base pairs in all human DNA molecules are identical. However, roughly three million base pairs are alleles that vary in sequence among humans. The areas on the DNA ladder in which the DNA sequence varies are called "polymorphic sites." Some such sites are more polymorphic than others. Forensic DNA testing makes use of sites which are "highly polymorphic."

If all three million base pairs which vary among humans could be examined and compared, DNA analysis could be completely individualized. However, due to the scope of such an undertaking, forensic DNA testing is limited to the examination and comparison of the length of several highly polymorphic alleles. [Note Appendix-1]

This is accomplished through the following seven step process:

1. Extraction of DNA. It is often necessary to separate the biological material that contains the DNA to be analyzed (e.g. blood or semen) from items such as clothing or carpeting. The biological material is extracted chemically from the surface of the cloth or carpet, and the cells of the biological material are burst open by chemical treatment to release the DNA. An enzyme is then added to digest cellular material that is not DNA, thereby providing a purer DNA sample. A certain minimum amount of DNA must be obtained in order for the analysis to proceed.

2. Restriction Digestion. The "clean" DNA is mixed with proteins called "restriction enzymes," which cut the DNA molecules into fragments at specific base sequences. The restriction enzymes recognize particular sequences of base pairs and cut the DNA ladder at particular locations within that sequence. The resulting fragments of DNA are known as "restriction fragments." Some fragments will contain polymorphic DNA segments, although most will not. Because the alleles differ markedly in length from one person to the next, the restriction fragments which contain these alleles will also differ in length.

3. Gel Electrophoresis. The restriction fragments are separated according to length by a process known as "agarose gel electrophoresis." An agarose gel is prepared with a negative and a positive electrode at opposite

Page 229

ends of the gel. The restriction fragments are placed in the gel near the negative electrode. An electrical current is run through the gel, and the restriction fragments, which are negatively charged in their normal state, are drawn toward the positive electrode. The speed with which a restriction fragment travels through the gel is determined by its length, with the shortest fragments traveling the most quickly. Therefore, gel electrophoresis distributes the restriction fragments along parallel lines across the gel in positions which correspond to their lengths. Because gels may vary and thus cause a difference in the speed with which the same restriction fragment will move through gel, the defendant's sample and the crime scene sample are placed in separate columns in the same gel. Various circumstances could cause a shift in one or more bands that, as will be seen, the testing procedure produces.

4. Southern Transfer. Because the agarose gel is difficult to work with, the restriction fragments are transferred to a nylon membrane by a process known as "Southern transfer," named after a Scottish scientist who developed it. The nylon membrane is placed over the gel, and paper towels are applied to the membrane. Through capillary action, the DNA fragments are drawn out of the gel and attach themselves to the nylon membrane in the same positions they occupied in the gel. The restriction fragments are then treated with a chemical which dissolves the bond between the two nucleotide bases of each rung on the DNA ladder, thereby splitting the ladder in two by sawing through the middle of each rung. This step results in a collection of single strand restriction fragments, and prepares the sample for the next step. The splitting of a fragment does not change its position relative to the other fragments.

5. Hybridization. At this point, the DNA analyst has a large number of restriction fragments, most of which occur in the DNA of all humans. Hybridization allows a DNA analyst to identify and separate from the entire sample certain restriction fragments which contain highly polymorphic alleles. This is achieved by dipping the nylon membrane in solutions of various "genetic probes" (developed by recombinant DNA technology), which are designed to link with identified polymorphic alleles. Each probe is constructed of a single strand of nucleotide bases, and, when introduced to the mass of single strand restriction fragments, will link only to a fragment that contains a nucleotide base sequence found in a particular allele. The probe is "tagged" with a radioactive marker so that, after it links with a particular polymorphic allele, its position relative to the other restriction fragments, which are distributed according to length, can be observed. Most forensic DNA testing uses three or more different probes to examine each DNA sample, with each probe designed to link to a different allele. All testing laboratories do not, however, probe for the same polymorphic alleles.

6. Autoradiography. Autoradiography allows the position of the probes, and their complementary polymorphic restriction fragments, to be recorded.

Page 230

The nylon membrane is placed on a piece of x-ray film. The energy from the radioactive tags on the probes exposes the film, producing a pattern of bands, called an autoradiograph, known as a "DNA print" or an "autorad". Each band represents the presence of a different polymorphic allele, and its position indicates the length of the restriction fragment in which that allele occurs. Because individuals differ in the length of their polymorphic alleles, the position of the bands on DNA prints will tend to differ from person to person as well.

7. Interpretation of the DNA Print. The DNA print of the criminal defendant is compared with the DNA print of the biological material found at the crime scene to determine if both samples of DNA came from the same person. The different DNA prints are inspected visually to determine if there is a match. The comparison can also be done with machines. If the DNA prints of the two samples occupy the same relative positions, the samples contain the same polymorphic allele and are said to match. However, the different DNA prints need not align exactly with one another to constitute a match. A match will be declared if the DNA prints meet certain "matching rules," that is, if they fall within a certain distance of one another. Cellmark declares a match if bands from two DNA prints fall within one millimeter of each other. Of course, if the variation that permits a match to be declared is greater than one millimeter, the number of matches found will increase. Another laboratory (Lifecodes) uses a different matching rule (a standard deviation base match).

8. Calculating the Frequency of Particular DNA Prints. The fact that two DNA samples produce the same DNA prints, and therefore contain the same alleles, is of little probative value in a criminal prosecution until it is determined how often that combination of alleles occurs in a given population. Forensic DNA laboratories have developed procedures to determine the frequency with which a particular DNA print occurs in the relevant general population. The relevant population is determined according to the race of the criminal defendant. For example, in this case, DNA testing attempted to determine the frequency of particular alleles in the Caucasian population. Forensic DNA laboratories have compiled data bases for the frequency of particular alleles in each race by taking DNA prints from a number of people in that race and calculating the percentage of people in which each allele appears. Cellmark consulted its data base and determined that each allele found in the defendant's DNA and in the crime scene DNA occurred in a particular percentage of Caucasians.

The next step is to use the frequency of the alleles to calculate the frequency of particular "genotypes" on an individual's DNA print. A "genotype" is a combination of alleles, one half of which are inherited from each parent. All the alleles in a particular genotype relate to the same bodily structure or function. For example, a person who inherits an A blood type allele from his father and a B blood type allele from his mother is said to have an AB genotype for blood type.

Page 231

If an individual has inherited a different allele from each parent for a particular genotype, that person is said to be "heterozygous" for that genotype. If an individual has inherited the same allele from each parent for a particular genotype, that person is said to be "homozygous" for that genotype. To calculate the frequency of a genotype for which a person is homozygous, the formula "pq" is used, where p and q are the frequencies of the alleles in the genotype. To calculate the frequency of a genotype for which a person is heterozygous, the formula "2pq" is used. The product of p and q is doubled for heterozygous genotypes because it is not known which allele came from which parent; therefore, there are two possibilities for the source of the genotype.

The final step is to calculate the frequency with which a particular DNA print would occur in the relevant population. Cellmark did this by multiplying the frequencies of the particular genotypes found in the DNA print. For example, if four probes were used to create one DNA print, which revealed four genotypes, and if it was determined that genotype A occurs in one of every 1,000 Caucasians, genotype B in five of every 1,000, genotype C in 100 of every 1,000, and genotype D in 200 of every 1,000, the frequency of the entire DNA print in the Caucasian population would equal (.001) x (.005) x (.1) x (.2), or .0000001. According to this calculation, the DNA of only one person out of every ten million Caucasians would produce this particular DNA print. The propriety of using this product rule is discussed in the opinion.

As the acceptable difference in the length or position of the alleles for permitting a match to be declared is increased, the proportion of the population that could be declared to have the same alleles as a defendant increases. After the tests were done in this case, Cellmark changed its view of what constitutes a match, dropping the estimate as to the defendant from one in billions to one in 59,000,000.

The process is described in Hoeffel, The Dark Side of DNA Profiling: Unreliable Scientific Evidence Meets the Criminal Defendant, 42 Stan. L. Rev. 465, 469-475 (1990), and in Thompson & Ford, DNA Typing: Acceptance and Weight of New Genetic Identification Tests, 75 Va. L. Rev. 45, 64-76 (1989). A parallel process used by Cellmark's competitor, Lifecodes, is described in People v. Castro, 144 Misc. 2d 956, 965-969 (N.Y. 1989).