Researchers from Weill Cornell Medical College and the Rogosin Institute have developed a next-generation sequencing approach for diagnosing autosomal dominant polycystic kidney disease, which they showed could equal or exceed the efficiency and accuracy of current Sanger sequencing-based diagnosis.
The team published the results of the small retrospective study — which evaluated the accuracy, cost, and turnaround time of an Illumina MiSeq sequencing strategy utilizing individually barcoded long-range PCR libraries relative to the currently used Sanger sequencing — in the Journal of Molecular Diagnostics this month.
According to the authors, the NGS method had a sensitivity of over 99 percent and specificity of 99.9 percent, with costs reduced by up to 70 percent of the cost of Sanger-based testing and turnaround time reduced by about half.
Hanna Rennert, the study's senior author and a professor of pathology at Weill Cornell, told Clinical Sequencing News this week that her team has been working for several years with the Rogosin Institute on a long-term prospective clinical research study of ADPKD genotypes and phenotypes, during which they have updated their genotyping strategy several times.
The group now plans to use its newly-described NGS method for patients in this study. "In practice, we have to wait to batch patients together, but the pipeline we describe [in the paper] is what we use now," Rennert said. "Routinely we run maybe 12 to 15 patients at a time. Technically, we could probably combine even more patients together, but we don’t have the volume."
"We were able to show that the mutation detection rate is as good as Sanger, and maybe also has the capacity to do better, especially because we can capture genomic regions," she added. "So, whereas previous methods focused on exons and areas close to the intron splice junction, now we can learn more about intronic changes that have not really been documented for ADPKD, and might be associated with the disease."
ADPKD is mainly caused by mutations in two genes — PKD1 and PKD2 — spanning 46 and 15 exons, respectively. Analysis of these genes is complicated by six homologous regions, or pseudogenes located on chromosome 16, which share almost 98 percent of their sequence with PKD's exons 1-33.
Previously, molecular diagnosis of ADPKD has relied on long-range PCR using primers located to rare mismatch sites that distinguish PKD1 from its pseudogenes, followed by additional amplification of individual exons and analysis by Sanger sequencing.
More recently, Rennert said, groups have experimented with adopting NGS for this testing, including a recent effort by the Mayo Clinic coupling long-range PCR and Illumina sequencing.
However, this approach lacked real clinical potential, Rennert explained, because samples were pooled and barcoded together, resulting in low sensitivity for individual patient samples as well as slow turnaround time.
In its recent study, the Cornell team developed a method it believes is better tailored to the clinical diagnostic setting by individually barcoding the LR-PCR products of each patient sample, then pooling them for analysis on a single run of the Illumina MiSeq.
To test the method, the researchers evaluated sequencing results using a panel of 25 patient samples from the Rogosin Institute study repository that had been previously analyzed by Sanger.
According to the group, 85 percent of all the sequenced reads using the NGS method mapped to the PKD1 or PKD2 reference genome. Overall, more than 93 percent of the targeted sequences were covered more than 30x, the group reported, with average coverage greater than 103x and minimum coverage of 11x or greater.
For one PKD2 exon — exon1, which contains a very highly GC-rich region — the method achieved only a very low coverage in several patient samples, down to 0x at one to two nucleotides.
Overall, the method was comparable in its performance to the recorded Sanger results, the researchers reported. In one patient, two changes were incorrectly identified by the NGS method as homozygous, rather than heterozygous, but otherwise the results matched exactly.
To further evaluate the method, the team then tested a separate set of 24 samples that had all tested negative for a pathogenic mutation using Sanger sequencing. In this cohort, the NGS method identified pathogenic variants in three patients.
In a third cohort of 25 patients that had not been genotyped previously with any method, the NGS approach identified probable pathogenic mutations in 16 subjects. When the same samples were retested using Sanger, all of both the mutation-positive and mutation-negative calls were confirmed, suggesting a sensitivity and specificity of 100 percent in this group.
On the practical side, the researchers also evaluated the relative costs of NGS versus Sanger sequencing. By pooling 25 patient samples in one MiSeq run, the cost of testing was reduced by about 70 percent compared to Sanger, the authors wrote — from $271 per patient to $82 per patent.
Data analysis was also considerably faster with the NGS method, the authors wrote, taking only one to two weeks versus at least four weeks with Sanger.
According to Rennert, the Cornell team is now using the NGS approach for clinical testing within the long-term study it is conducting with the Rogosin Institute.
In the commercial setting, Rennert said, Athena Diagnostics holds a patent on PKD1 and PKD2 analysis, and currently markets Sanger-based testing for the disease in the US.
But, she said, some European clinicians have expressed interest in the Cornell group's NGS method, and US labs may also potentially be able to adopt the method in the future after recent changes to the legal standing of gene patents.