IHMT

Instituto de Higiene e Medicina Tropical

Home / Publicações / Science in court: the myth of HIV fingerprinting.

Science in court: the myth of HIV fingerprinting.

An Editorial in Nature has emphasised the importance of proper scientific validation of forensic methods before their use in court. Classic fingerprinting, DNA fingerprinting, and brain imaging were discussed, and several cases of doubtful interpretation of forensic evidence with potentially incorrect conviction or exoneration of suspects were indicated. Cases of HIV transmission examined in court are experiencing similar problems, whereby the risk of incorrect conviction is substantial because of a flawed view of the science behind forensic phylogenetics. A scientific discussion of HIV fingerprinting in cases of HIV transmission is therefore urgent.

In 1994, phylogenetic analysis, alongside other compelling evidence, first led to a criminal conviction in a case of HIV transmission through rape in Sweden. Thus, the precedent was set for the use of phylogenetic data as circumstantial evidence in court. Since then, forensic phylogenetic analysis has been used repeatedly, albeit in the context of other incriminating evidence. This analysis is increasingly determining convictions by criminal courts. By calling such investigations HIV fingerprinting, scientists raise unrealistic expectations among court rulers. Unlike for DNA fingerprinting, where a likelihood can be calculated for a full match between the evidential DNA and the suspect DNA, there is never a full match between the RNA or the DNA of HIV in two samples, even within an individual.

HIV is constantly evolving, and in forensics, phylogenetic analysis investigates the evolutionary (and thus epidemiological) link between the HIV strains of victim and suspect, trying to establish whether they belong to a transmission chain. When forensic phylogenetics finds a monophyletic cluster joining victim and suspect, this evidence cannot prove transmission beyond reasonable doubt. Proper identification of the transmission source would require two major assumptions: that a phylogenetic tree can flawlessly reconstruct a true epidemic history and that strains from all patients ever infected with HIV are available as controls. Both assumptions are unrealistic. An error margin for clustering in a phylogenetic tree can be estimated by use of bootstrap replicates or Bayesian posterior probabilities for example. However, we can never sample all infected individuals and neither the number of unsampled patients nor their phylogenetic clustering is known. Nevertheless, sufficient controls can be included in the analysis to explore the diversity of HIV strains circulating in the vicinity of suspect and victim. Under such conditions, significant clustering of the suspect and victim virus strains can indicate that they belong to a transmission chain. However an indirect link can never be ruled out.

Full-size image (52 K)Figure. 
Hypothetical phylogenetic tree for an investigation of HIV transmission. Although the suspect and the victim form a monophyletic cluster that is significantly supported, because complete sampling of all patients with HIV (or at least from the epidemiologically relevant population who are local to the parties under investigation) is not feasible, to know if (and how many) other individuals belong to the same transmission chain is not possible. *Statistically significant support of that cluster.
Because the full transmission tree is unknown, no likelihood can be attached to the a priori hypothesis of direct transmission, even if that hypothesis cannot be rejected. In contrast, separate clustering with unlinked individuals that is statistically supported can disprove direct transmission. Therefore, forensic phylogenetics has to be done in the context of hypothesis testing—the court’s hypothesis being the suspect infected the victim, while the expert tries to find evidence to reject this hypothesis. Furthermore, phylogenetic evidence cannot indicate the direction of transmission, but the person who was infected later can be excluded as the source of earlier infections by taking into account the dates of seronegative and seropositive results and the appropriate window period. Such exclusion should not be mistaken for proof of transmission in the other direction. To prevent erroneous use of phylogenetic evidence in court, scientific experts are recommended to do the following: (1) to provide evidence of their expertise in forensic analysis; (2) to use two samples from different times from each party under investigation as close as possible to the alleged time of transmission; (3) to analyse two different viral genomic regions; (4) to use sufficient local controls, with full protection of their identity; (5) to blind the identities of the parties, with unblinding allowed only for testimony; and (6) to word correctly the findings, with no claims made about the likelihood of direct transmission.

Phylogenetic analysis is a powerful technique that can, if properly used, provide valuable circumstantial evidence in forensic investigation for cases of HIV transmission. However, scientists should be aware of the limitations of this analysis, and should emphasise that courts must use other evidence to achieve a conviction.

Science in court: the myth of HIV fingerprinting.

An Editorial in Nature has emphasised the importance of proper scientific validation of forensic methods before their use in court. Classic fingerprinting, DNA fingerprinting, and brain imaging were discussed, and several cases of doubtful interpretation of forensic evidence with potentially incorrect conviction or exoneration of suspects were indicated. Cases of HIV transmission examined in court are experiencing similar problems, whereby the risk of incorrect conviction is substantial because of a flawed view of the science behind forensic phylogenetics. A scientific discussion of HIV fingerprinting in cases of HIV transmission is therefore urgent.

In 1994, phylogenetic analysis, alongside other compelling evidence, first led to a criminal conviction in a case of HIV transmission through rape in Sweden. Thus, the precedent was set for the use of phylogenetic data as circumstantial evidence in court. Since then, forensic phylogenetic analysis has been used repeatedly, albeit in the context of other incriminating evidence. This analysis is increasingly determining convictions by criminal courts. By calling such investigations HIV fingerprinting, scientists raise unrealistic expectations among court rulers. Unlike for DNA fingerprinting, where a likelihood can be calculated for a full match between the evidential DNA and the suspect DNA, there is never a full match between the RNA or the DNA of HIV in two samples, even within an individual.

HIV is constantly evolving, and in forensics, phylogenetic analysis investigates the evolutionary (and thus epidemiological) link between the HIV strains of victim and suspect, trying to establish whether they belong to a transmission chain. When forensic phylogenetics finds a monophyletic cluster joining victim and suspect, this evidence cannot prove transmission beyond reasonable doubt. Proper identification of the transmission source would require two major assumptions: that a phylogenetic tree can flawlessly reconstruct a true epidemic history and that strains from all patients ever infected with HIV are available as controls. Both assumptions are unrealistic. An error margin for clustering in a phylogenetic tree can be estimated by use of bootstrap replicates or Bayesian posterior probabilities for example. However, we can never sample all infected individuals and neither the number of unsampled patients nor their phylogenetic clustering is known. Nevertheless, sufficient controls can be included in the analysis to explore the diversity of HIV strains circulating in the vicinity of suspect and victim. Under such conditions, significant clustering of the suspect and victim virus strains can indicate that they belong to a transmission chain. However an indirect link can never be ruled out.

Full-size image (52 K)Figure. 
Hypothetical phylogenetic tree for an investigation of HIV transmission. Although the suspect and the victim form a monophyletic cluster that is significantly supported, because complete sampling of all patients with HIV (or at least from the epidemiologically relevant population who are local to the parties under investigation) is not feasible, to know if (and how many) other individuals belong to the same transmission chain is not possible. *Statistically significant support of that cluster.
Because the full transmission tree is unknown, no likelihood can be attached to the a priori hypothesis of direct transmission, even if that hypothesis cannot be rejected. In contrast, separate clustering with unlinked individuals that is statistically supported can disprove direct transmission. Therefore, forensic phylogenetics has to be done in the context of hypothesis testing—the court’s hypothesis being the suspect infected the victim, while the expert tries to find evidence to reject this hypothesis. Furthermore, phylogenetic evidence cannot indicate the direction of transmission, but the person who was infected later can be excluded as the source of earlier infections by taking into account the dates of seronegative and seropositive results and the appropriate window period. Such exclusion should not be mistaken for proof of transmission in the other direction. To prevent erroneous use of phylogenetic evidence in court, scientific experts are recommended to do the following: (1) to provide evidence of their expertise in forensic analysis; (2) to use two samples from different times from each party under investigation as close as possible to the alleged time of transmission; (3) to analyse two different viral genomic regions; (4) to use sufficient local controls, with full protection of their identity; (5) to blind the identities of the parties, with unblinding allowed only for testimony; and (6) to word correctly the findings, with no claims made about the likelihood of direct transmission.

Phylogenetic analysis is a powerful technique that can, if properly used, provide valuable circumstantial evidence in forensic investigation for cases of HIV transmission. However, scientists should be aware of the limitations of this analysis, and should emphasise that courts must use other evidence to achieve a conviction.