Background
Human cytomegalovirus (HCMV) is the most common pathogen of congenital infection. Although the incidence varies by race or ethnicity, congenital HCMV (cCMV) infection occurs in approximately one in every 100 to 1000 births [
1]. Most infants with cCMV are asymptomatic; however, approximately 10–15% of cCMV cases show physical symptoms [
2,
3]. Common findings of symptomatic cCMV include petechiae, jaundice, hepatomegaly, splenomegaly, microcephaly, and other neurological signs. Thrombocytopenia, transaminitis, direct hyperbilirubinemia, chorioretinitis, and neuroimaging abnormalities are indicative of central nervous system involvement, and sensorineural hearing loss can be found on examination [
4]. Ganciclovir (GCV) and its oral prodrug, valganciclovir (VGCV), are antiviral agents used in the treatment of symptomatic cCMV in infants. Antiviral treatment initiated within 1 month of life improved neurodevelopmental and hearing outcomes [
5]. Furthermore, the 6-month protocol improved hearing and neurodevelopment at the long-term assessment (at 24 months) compared to the 6-week protocol [
6].
CMV antiviral drug resistance has been primarily reported in patients with immunosuppression, such as transplant recipients or those with AIDS [
7]. Recently, antiviral drug resistance has also been reported in several cases of cCMV receiving GCV/VGCV treatment [
8]. Mutations in the viral thymidine kinase gene (
UL97) and the DNA polymerase gene (
UL54) confer resistance to GCV/VGCV [
9]. The current gold standard for genotypic detection of antiviral drug resistance is Sanger sequencing of PCR-amplified
UL97 and
UL54 gene segments [
10]. Although major mutations have been detected in the
UL97 gene [
11], several mutations have been reported in
UL97 and
UL54 [
9]. Comprehensive antiviral mutation detection with Sanger sequencing is labor-intensive because multiple PCR amplicons are needed to sequence the full-length of
UL54 and
UL97. Nanopores and long-read sequencing can provide rapid, near real-time sequencing. In this study, full-length antiviral gene mutation analysis was conducted to detect GNC resistance in patients with cCMV infection using nanopore sequencing.
Discussion
Sanger sequencing is the most frequently used method in antiviral gene mutation assays. As next-generation sequencing (NGS) has become common in clinical settings, an attempt to detect antiviral drug mutations using NGS has been reported [
10,
20‐
23]. Nanopore sequencers can read as long as 100 kbps and are suitable for sequencing the full-length of
UL54 (approximately 4 kbps) and
UL97 (about 2 kbps) genes. As each read covers the full length of the gene, mutations can be detected by a minimum of 30 reads. Minimized input data can reduce the computational load and shorten the time required for data processing.
Data processing can become a bottleneck for NGS assays for clinical researchers who are not familiar with bioinformatics. However, the NGS analysis platform can be used without a bioinformatics background, although there is a limitation in that the user cannot handle the data freely. BugSeq is an online bioinformatics platform for automated microbiology sequencing analysis using nanopore reads [
24]. Bugseq can also be utilized in CMV antiviral drug resistance genotyping [
20]. In this study, a single set of fastq files was run for Bugseq, and the result matched our results (data not shown). The mutation resistance analyzer is also helpful for collating SNVs with drug resistance mutation [
17]. Clinical researchers can utilize such platforms in antiviral research.
The proportion of antiviral mutations can be calculated as the percentage of SNV reads per total reads using NGS [
21]. In this study, the SNV proportion was observed over time in patient 11. The percentage of each mutation subsided after cessation of GCV/VGCV treatment in this patient. Repopulation of the wild type after cessation of GCV/VGCV therapy has been reported previously [
25]. Sahoo et al. reported the ability to detect an antiviral drug resistance lower than 20% using the NGS method [
10]. Although the detection limit of proportion was not set in this study, A594V was detected by Lofreq with a 15% mutation rate in the urine sample of patient 11. Urine samples were also assayed in this study. Interestingly, mutation rates were lower than those in blood samples, although the dynamics of the mutation rates were synchronized with that of blood. This suggests that multiple CMV strains were localized in organs at various rates in the infected individual.
We detected
UL54 mutations in the validation sample (K513N, V787L) and patient 11 (V823A). For the validation sample, the patient received foscarnet after M460V in
UL97 was detected. K513N confers GCV and cidofovir resistance [
26], and V787L confers cidofovir and foscarnet resistance [
27]. For patient 11, the dynamics of the
UL54 V823A mutation synchronized with
UL97 A594V. V823A has been reported to confer cidofovir and GCV resistance [
28]. Most GCV-resistant CMV strains have mutations in
UL97 [
11]; however, specific gene mutations in
UL54 may occur, or in combination with
UL97 [
7]. The
UL54 mutation in combination with the
UL97 mutation increases the level of resistance [
29]. Therefore,
UL97 and
UL54 should also be investigated when the patient is suspected of having antiviral drug resistance using GCV/VGCV.
Several reasons for why drug resistance emerged in patient 11 but no in other infants in the virus persistent group were considered. First, the high viral load before GCV/VGCV therapy in patient 11. This patient might have had a more severe damage in the central nervous system. Retinal abnormalities also suggested an extensive range of viral infection. In contrast, patient 8 had a high viral load before therapy, but antiviral resistance did not emerge. The relatively short period of therapy (3 months) may be one reason. A treatment period longer than 3 months appears to be a risk factor for antiviral drug resistance [
30] in immunocompromised patients. Recently, antiviral drug resistance has been reported in patients with cCMV infection [
8,
31‐
35]. It was observed that GCV resistance emerged after 3 months in patients with cCMV infection [
8]. A 6-month GCV/VGCV treatment is the recommendation for cCMV therapy [
6]; therefore, careful CMV blood load monitoring is needed. Second, an impaired host immune function might be related to persistent viremia. Although the immune function was not fully investigated in patient 11, it was unlikely that the infants had congenital immune dysfunction because the viral load in blood subsided at 14 months of age, 4 months after the cessation of antiviral therapy. The CMV blood load of all patients in this study subsided at approximately 12 months of age. This suggests maturity of the host’s immune system. Third, insufficient blood GCV/VGCV concentrations could not be excluded because of lacking blood GCV/VGCV level measurements. Patient 11 had full-dose VGCV around the period in which the antiviral resistant mutation emerged. In this patient, time-course analysis showed that the mutation emerged after 6 months of therapy. Despite the GCV resistant mutation, it seems VGCV was partially effective because an increase in viral load and transaminases was observed. This may be because the proportion of mutant SNVs did not overwhelm the wild type.
Prolonged detection of CMV in blood could be a risk factor for sensorineural hearing loss [
36] in patients with cCMV infection; however, it is sometimes difficult to continue GCV/VGCV therapy because of side effects such as neutropenia. Therefore, when the CMV blood load increases during GCV/VGCV therapy, it is important to consider whether this increase is affected by the ineffective blood GCV/VGCV level or antiviral drug resistance. It is beneficial to monitor CMV viral blood load levels before, during, and after treatment because an increasing viral load during therapy may provide an indication for antiviral resistance, and therefore, an ineffective antiviral therapy. An increasing viral load after reduction or cessation of antiviral therapy because of adverse effects such as antiviral associated neutropenia may provide indication for resuming therapy. Our study is limited by the small sample size of infants with congenital CMV infection. In addition, although nanopore sequencing provides fast long-read sequencing in a compact and portable format and initial low costs, it involves higher base calling error rates when compared to standard next generation sequencing, such as the Illumina platform. Further study could be necessary to better characterize the application of this technology for CMV resistance testing.
In conclusion, an antiviral gene mutation assay was performed using nanopore sequencing. Antiviral drug resistance can emerge in patients with cCMV during long-term GCV/VGCV therapy.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.