Long-range PCRs and next-generation sequencing to detect cytomegalovirus drug resistance-associated mutations

Abstract

Cytomegalovirus (CMV) infections occur in 10%-40% of organ or stem cell transplant patients. Despite the low prevalence, CMV antiviral resistance has an important impact on patient outcomes. Guidelines for transplant recipients recommend that resistance should be suspected in cases of unchanged or increasing CMV viral loads after a minimum of 2 weeks of antiviral therapy at an appropriate dose or >6 weeks of ganciclovir exposure. Next-generation amplicon sequencing (NGS) makes it possible to directly target the genes involved in this resistance. Currently, six drugs are available, and six CMV genes (UL54-UL97-UL89-UL56-UL51-UL27 genes) can harbor mutations affecting drug efficacy. Here, we developed different primers targeting these six genes with long-range polymerase chain reaction (PCR). Based on clinical requirements, all genes or a subset could be sequenced in a single run using Oxford Nanopore technology and combined with an automatic bioinformatics pipeline to detect and report mutations. We utilized 46 blood samples, five external quality controls, and 10 mixes of two bacmids provided by the national reference center (CNR) Herpesvirus Limoges, each carrying distinct mutations. Assay performance (sensitivity, specificity, and accuracy) was evaluated through an interlaboratory exchange with CNR Herpesvirus. Long-range PCR combined with next-generation sequencing analysis enables earlier and more comprehensive discrimination of the double population and determines whether the detected single-nucleotide polymorphisms are present on single or multiple CMV strains. We developed a next-generation sequencing assay combined with eight long-range PCRs to sequence all genes involved in CMV antiviral resistance and to detect early low-frequency mutations.

Keywords: antiviral resistance; cytomegalovirus; long-range polymerase chain reaction; next-generation sequencing.

Fig 1

Flow chart of the Bioinformatics pipeline. Fast basecalling is first performed, followed by high-accuracy basecalling to provide higher raw read accuracy. Sequence mapping and alignment used minimap2, which realigns the remaining reads to the CMV AD169 genome reference (GenBank ID X17403.1). ClairS-TO generates the VCF. The list of SNPs obtained is compared with the CNR Herpesvirus, Limoges database (https://www.unilim.fr/cnr-herpesvirus/outils/codexmv/database/) to identify any mutations generating resistance. Finally, the generated report is validated by medical biologists and sent to clinicians. CNR: national reference center; SNP: single-nucleotide polymorphism; VCF: variant calling table.

Fig 2

SNP detection frequencies on UL54 in mixed DNA using NGS technology compared with the reference Sanger technology. Red spots correspond to SNPs from sample 1 and blue spots to those from sample 2. CHUGA: University Hospital Grenoble Alpes; CV: coefficient of variability; PCR: polymerase chain reaction; SD: standard deviation; SNP: single-nucleotide polymorphism.

Fig 3

Visualization of the distribution of SNPs on UL54 raw sequencing data. Haplotype was determined by whatsHap (32) and visualized on IGV (33). Red reads indicate reads belonging to one subpopulation, while blue indicates a second subpopulation. SNPs presented in both populations is described in the table under the figure.