NEW YORK (GenomeWeb) – Clinical labs tend to assume that PCR is a very reliable technique and artifacts known as allele dropout and drop-in are rare. However, determining the precise frequency of these rare events has been a challenge and little data apart from case reports has been published on the issue.
Now, researchers at Centre Hospitalier Universitaire de Québec and Université Laval in Canada have quantified PCR artifacts — allelic dropout events in which one allele is preferentially amplified, and drop-in events due to sporadic contamination — by analyzing 30,769 patient reports processed over eight years.
"Most laboratories do not use genotyping methods that would allow them to detect allele dropout or drop-in events, so they likely are not aware that such events and errors do occur regularly in their lab," François Rousseau, a co-author on the study and told GenomeWeb in an email.
Rousseau, a professor and researcher at CHU de Québec and Université Laval, said his group was concerned enough about this phenomenon to design its clinical genotyping assays with two independent assays for each allele, wild type and mutant.
This meant that the lab ran four independent PCR reactions for each patient sample.
In results published in Journal of Molecular Diagnostics, Rousseau's group described testing done for four autosomal recessive genetic conditions routinely assessed in the lab: tyrosinemia type I, hereditary hemochromatosis, cystic fibrosis, and Andermann syndrome. The study demonstrated that 94 percent of the dropout and drop-in artifacts were due to non-reproducible PCR failures rather than sequence variation. Thus, they could not have been avoided by careful selection of amplification or detection oligonucleotide sequences, Rousseau explained.
In all, 93 dropout events were detected, but were resolved before the results left the lab. Another 42 artifacts appeared to be drop-in events, due to amplification of alleles not ultimately part of the patient genotype.
The group further calculated that aberrant events from a single assay, with no independent confirmation by a second assay, would yield one erroneous genotype for every 450 genotypes, or an error rate of 0.22 percent, Rousseau said.
"For a clinical molecular lab performing, for instance, 4,000 genotypes per year … this will translate into nine erroneous results per year that will be signed out by the lab," he said.
However, performing two completely independent assays for each locus and requiring concordance between them to ascertain the true result reduces this probability, as the likelihood of essentially a double event becomes one in 200,000.
"For a lab that does 4,000 patient genotypes per year, this would mean one erroneous diagnosis every 50 years," Rousseau said.
The total cost of this workflow is CAN$30 per patient sample including technical time and reagents, Rousseau said, noting that this is only slightly more costly than running a single assay with no independent confirmation of the genotype.
Rousseau said the biggest risk is that these rare events — which can be caused by sequence-independent factors or allele-specific sequence variations — will lead to an incorrect diagnosis of a patient. Since physicians do not usually retest for hereditary mutations, this misdiagnosis could be attached to the case for a long time.
Furthermore, multiplex assays with a single primer pair per mutation have a higher likelihood of error; a laboratorian could expect one erroneous genotype per 100 patient samples in a multiplex test measuring five mutations, for example.
"Our study suggests that clinical grade assays should not be based on a single DNA amplification, nor on a single mutation detection hybridization," Rousseau said.
Another take-home point from the study is that assay design can indeed reduce the likelihood of reporting an error, Rousseau said.
Furthermore, given the low frequency of such events, the likelihood of detecting them in the validation phase of an assay remains slim. A safer strategy appears to be to ensure that the genotyping assays are designed to detect them during routine analysis over the years, triggering the follow-up retesting. Labs can "thus make sure the true genotype is determined in such cases before the test report is signed out," said Rousseau.
Rousseau said his lab will definitely continue designing and using two independent PCRs for each genotype for clinical grade assays, and he believes other clinical labs should do the same. They should also ultimately "avoid using methods that are not amenable to assay designs that will have two totally independent assays with non-overlapping primers for each locus."