the forensic institute

Low Copy Number DNA and The Forensic Institute

Low Copy Number DNA Technique

 

DNA and LCN

The Omagh Bomb trial and the validation of LCN

Validation of LCN and The Forensic Institute

Recent developments (and disclosure, or not)

2009 Appeal Court decision on LCN

Update from R v C (2011)

Introduction

The Forensic Institute has been one of the main campaigners in bringing the limitations of the use of the Low Copy Number (LCN) DNA technique, and its failure to be internationally accepted, to the attention of UK courts. Given the important successes and the high profile nature of some of the techniques, notwithstanding the ability to claim that England led the World in DNA analysis, a clash with the providers (the Forensic Science Service) was inevitable. Despite considerable abuse we have continued to state our position clearly and succeeded in assisting many defendants who would otherwise have had no scientific basis for their defence.

DNA and LCN

The use of DNA profiling has revolutionised the use of science in legal cases. DNA profiles produced from manufactured kits such as SGMPlus and Identifilier are routinely used as evidence in criminal cases. The kits are designed and validated by the manufacturer to operate within a specific range of amounts of DNA, typically 0.5 – 2.5ng (a nanogram is a 1,000th of a millionth of a gram). The kits work by copying (amplifying) the DNA molecules contained in a sample a number of times to produce enough to be detected in the analyser. The chemistry used by the kits is capable of amplifying just one molecule of DNA. By varying the conditions under which the kit is used some claim to be able to produce profiles from much lower amounts of starting (template) DNA. The general term for these techniques is variously referred to as Low Copy Number (LCN) or Low Template DNA (LTDNA) analysis, although some restrict the term LCN to the process used by the FSS Ltd which uses 34 cycles of amplification instead of the routine 28-cycle process recommended by the manufacturer. However, with very low numbers of template DNA molecules the process may fail to amplify the template. This leads to a number of problems in the interpretation of the resulting profiles. These are caused mostly by sampling, or stochastic, errors caused by the failure of the chemistry to work effectively with such low numbers leading to poor reproducibility of the results.

back to top

The Omagh Bomb trial and the validation of LCN

We were approached by the defence solicitors (Kevin R Winters & Co) in the case of Sean Hoey in Northern Ireland to review the DNA evidence presented against him. Although the trial has been labelled the ‘Omagh Bomb trial’ it was in fact a trial involving twelve other incidents, only three of which were claimed to have any DNA evidence linked to Mr Hoey. Omagh was not one of those three.

We were asked to review the DNA evidence which, we discovered, was based on the LCN technique. Investigation of the published scientific literature revealed that this was not a universally accepted technique and, in fact, was only being used in British courts. We decided that the only way to assess the reliability of the DNA evidence was to assess the experiments that are necessary before any scientific analytical technique can be regarded as fit for purpose. One form of words for that purpose had been stated in a set of guidelines published in the USA by the Specialist Working Group on DNA Analysis and Methods (SWGDAM). Professor Jamieson of The Forensic Institute attached these to one of his statements in the case and one portion was subsequently quoted in the Court’s judgement (and again by the Caddy Review);

“the process whereby the scientific community acquires the necessary information to:

  • Assess the ability of a procedure to obtain reliable results.

  • Determine the conditions under which such results can be obtained.

  • Define the limitations of the procedure.

The validation process identifies aspects of a procedure that are critical and must be carefully controlled and monitored. "

back to top

Validation of LCN and The Forensic Institute

In order to assess the validation we eventually obtained all of the data, including the electronic results, and papers from the FSS Ltd that had been used to validate the technique. To our knowledge we are the only group to have done that until then or since.

Professor Dan Krane from the USA reviewed the profiles produced by the technique by the FSS Ltd.
It became clear that, in our opinion, the data were insufficient to support the claims being made. A statement from Professor Jamieson summarised his position;

In scientific analysis it is essential that we can assess how much we can trust any result. This trust is based upon knowledge of how often the technique gets the correct (i.e. true) value when it is used to measure something. Errors can be classed as false negatives, when the technique says that something is not there when it is, and false positives, when the technique detects something when it is not actually there.

My contention is that the error rate for LCN is unknown, but that the lack of consistency of the results is demonstrably high enough to cause serious concern as to the reliability of this evidence. The prosecution data for the case samples show that they cannot reproduce the electrophoretic profile from any of the key evidential samples, regardless of the allelic designation and consensus. Additionally, the data that was eventually supplied to support the validation of the technique is insufficient to do so.

The issues that the court must be satisfied on are;

  1. That the LCN technique measures the true profile. The technique used to determine a profile must therefore be absolutely dependable. The LCN technique is acknowledged not only to ‘lose’ alleles from the true profile, but also to ‘gain’ alleles. We cannot know which is the case when we have only profiles from the crimestains which, by definition, we do not know the actual profile of.

  2. Given that there is uncertainty, the degree to which this is present in the evidential samples and what account has been taken of this in deriving opinion.

  3. When the profile shows alleles apparently from more than one person, the decision as to who was the donor of the different alleles.

We know that all of these effects are dependent on the amount of starting DNA, yet the LCN technique by definition does not know how much DNA it is testing, and drop-in rates are variable. Therefore it is impossible to know how to reliably incorporate data on allelic drop-out or drop-in.

Although the SGM+ technique is validated satisfactorily, its extension into the LCN technique has not undergone the essential scientific testing and scrutiny to enable it to be regarded as a reliable procedure. This view applies to the technique generally, and its specific deployment within the FSS, and specifically in this case.

In that same statement, Professor Jamieson writes,
To discover the error rate in LCN should have been a fairly simple exercise;

  • Take, say, a blood sample with a known profile

  • Measure the amount of DNA in the sample

  • Divide it into, say 100 aliquots (tubes)

  • Dilute each of these until there is, say 1ng, 500pg (=0.5ng), 200pg, 100pg, 50pg, 20pg, 10pg, 5pg, 1pg, and 0 of DNA. This is a total of 1,000 samples.

  • Run all of these in the LCN system

  • Create a graph like those above recording the number of ‘correct’ results vs ‘wrong’ results for the whole profile and perhaps for each allele.

  • This would have to be repeated with different samples to ensure that all (or at least most of the common) of the alleles had been tested.

Others have generated this type of data although the material eventually supplied by the prosecution did not contain these.

There is an interesting parallel here between this and the experiment suggested by the Caddy Review;
3.15 To provide validation it is normal practice to begin with samples of known provenance and to submit them to the process and then to see how they comply with the expected outcome. This latter may require a statistical evaluation. For the LTDNA analyses the parallel would be the taking of a large number of DNA samples of known profiles reflecting the different alleles at the 10 different loci. These samples would then be serially diluted to provide masses of say 10, 20, 40, 80, 100 and 200pg.

back to top

Recent developments (and disclosure, or not)

In a more recent case we discovered that the FSS Ltd have now performed dilution studies similar to those we suggested. We cannot publish or illustrate these as they were disclosed under confidentiality agreements insisted upon by the FSS Ltd. Again, we are the only organisation that we are aware of to have sought and obtained disclosure of this data which is essential to understanding the reliability of the technique being used in criminal cases.

Similarly, the FSS scientists have claimed in high profile cases, despite the expectation of obtaining mixtures using the LCN technique, that mixture studies were unnecessary. We obtained disclosure of all of the mixture studies now recently performed by the FSS Ltd. They are also subject to a confidentiality agreement.

There has been and continues to be a debate within the international scientific community regarding the reliability of the LCN technique because of the presence of these stochastic effects. We are the first and only UK organisation to have seriously challenged the use of the technique in British Courts. Indeed, in an article by two barristers involved in another of our high profile cases (David Bentley and Peter Lownds of Doughty Street Chambers on R v Broughton), "Though it takes some uncovering, the debate within the forensic community over the use and reliability of LTDNA still rages". They also point out, "Professor Allan Jamieson of the Forensic Institute, Glasgow has been a lone critical voice within this jurisdiction."

The LCN technique was specifically designed to analyse amounts of DNA below 100pg and to produce reliable profiles even in the presence of stochastic effects (allelic dropout in particular). Professor Jamieson (in line with other eminent international experts) argued that the presence of these effects, which increase as the amount of DNA decreases, made the interpretation of these profiles unreliable.

back to top

2009 Appeal Court decision on LCN

The recent Appeal Court decision in Reed vindicates this view of the technique in endorsing the hallmark of reliability as the absence of stochastic effects and explicitly enabling challenges to what the Court would expect to be the ‘rare’ circumstances of evidence being led from cases involving between 100 and 200pg. By extension, although the Court expressed no explicit view, it may be expected that cases in the sub-100pg range will now be extremely rare or absent; in any event, they will invite challenge.

The Court considers, as stated in evidence by Professor Jamieson, that profiles produced with more than 200pg of template DNA, “can produce electrophoretograms which are capable of reliable interpretation”.

However, according to the User’s Manual for the SGMPlus kit it is also a fact that the standard method produces reliable profiles with amounts as low as 250pg. At best, that would seem to leave a ‘useable range’ for LCN of about 50pg.

The Appeal Court also criticised Professor Jamieson’s suitability to give evidence at the Omagh trial. How that is to be squared with ultimately agreeing his main position at that trial and that Judge Weir having listened to, “the very extensive evidence given as to the present reliability or unreliability of the LCN procedure for the purpose of obtaining data of evidential quality”, also agreed that the concerns expressed were valid, is a matter for the Court.

Other complex, and as yet unanswered, questions remain. Many LCN profiles are mixtures, containing small amounts of DNA from two or more people. Does the 200pg limit apply to the total template DNA or to one or other of the contributors which, if the total is 200pg must by definition be less than that? Presumably the answers will only emerge as we take on other cases to present the appropriate challenges again. In that regard, our expertise is recognised;

“… he has given evidence in so many Low Template DNA cases since then on the strength of the observations in R v Hoey that he has acquired a degree of experience from these cases, his discussion with others and his reading of papers.”

back to top

Update from R v C (February 2011)

It did not take long for the question that we raised above to emerge at the Appeal Court.

Close on the heels of Reed & Reed the Appeal Court produced a judgement in R v C . R v C is a case wherein it would appear that mixed DNA profiles with less that 200pgs were involved. We were not, so far as we are aware, involved in the case of R v C.

The judgement in R v C attempts to clarify the ruling in Reed & Reed and in so doing appears to underline our contention that it is not the quantity of DNA, but the quality, specifically the reproducibility (i.e. absence of stochastic variability) that is the essential question;

“Of course if, in the case of a mixed profile, the DNA relating to a particular profile comprises less than 200 picograms, problems may arise. But as was made clear in Reed & Reed and in Broughton, profiles obtained from less than 200 picograms can be reliable. It is reliability that is the issue, not the quantity, though plainly the quantity is relevant (as has been made clear) to the consideration of stochastic effects. ”

In our view, the reliability of the LCN method below the stochastic threshold has not been demonstrated with sufficient numbers of samples and with samples which represent those likely to be discovered in crimestains. The limited extant data must be made available to enable the scientific community to conduct a meaningful assessment of the inferences that can be made from it.

By way of analogy, an aeroplane, designed on the principles of flight, will fly and perform satisfactorily within the parameters of its design. As the speed lowers, there is no change to the aeroplane’s performance according to the laws of aerodynamics, but as the speed lowers further the aircraft will first suffer a loss of control and then simply stop flying (this is termed the stall speed). A graph of the flying performance of the aircraft will be continuous, but will show a precipitous fall-off at the stall speed. The stochastic threshold is the DNA profiling equivalent of this sudden change in performance. Profiling above the stochastic threshold produces consistent and reliable results. Below the stochastic threshold, reliability fails. It is disingenuous to equate the performance of ANY profiling technique above the stochastic threshold, to the performance below; that now appears to have been recognised scientifically and legally.

 

 

back to top

Related links

The Omagh Judgment pdf »


SWGDAM standards for DNA method validation>>


The Caddy Review pdf >>


Commentary on Reed & Reed >>


Judgement in R v C >>


Judgement in R v Broughton>>


Archbold Review article on Broughton>>

Interesting links

LCN - What now? from Barrister Magazine after Reed & Reed 2009>>


LCN DNA - Devil in the Detail » An article by Professor Jamieson


The Forensic Regulator's Response to the Caddy Review (pdf) »


Letter from America on Caddy Review »


FSS factsheet on LCN (pdf) »


Some links to useful LCN research and opinion »