http://orcid.org/0000-0002-0161-1749
Figures
Abstract
Objetives
The aim of this study was to identify CMV drug resistance mutations (DRM) in solid organ transplant (SOT) recipients with suspected resistance comparing next-generation sequencing (NGS) with Sanger sequencing and assessing risk factors and the clinical impact of resistance.
Methods
Using Sanger sequencing as the reference method, we prospectively assessed the ability of NGS to detect CMV DRM in the UL97 and UL54 genes in a nationwide observational study from September 2013 to August 2016.
Results
Among 44 patients recruited, 14 DRM were detected by Sanger in 12 patients (27%) and 20 DRM were detected by NGS, in 16 (36%). NGS confirmed all the DRM detected by Sanger. The additional six mutations detected by NGS were present in <20% of the sequenced population, being located in the UL97 gene and conferring high-level resistance to ganciclovir. The presence of DRM by NGS was associated with lung transplantation (p = 0.050), the administration of prophylaxis (p = 0.039), a higher mean time between transplantation and suspicion of resistance (p = 0.038) and longer antiviral treatment duration before suspicion (p = 0.024). However, the latter was the only factor independently associated with the presence of DRM by NGS in the multivariate analysis (OR 2.24, 95% CI 1.03 to 4.87).
Citation: López-Aladid R, Guiu A, Mosquera MM, López-Medrano F, Cofán F, Linares L, et al. (2019) Improvement in detecting cytomegalovirus drug resistance mutations in solid organ transplant recipients with suspected resistance using next generation sequencing. PLoS ONE 14(7): e0219701. https://doi.org/10.1371/journal.pone.0219701
Editor: Stanislaw Stepkowski, University of Toledo, UNITED STATES
Received: February 20, 2019; Accepted: May 7, 2019; Published: July 18, 2019
Copyright: © 2019 López-Aladid et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: M.A.M. was supported in part by: Agency for Health Technology Assessment and Research Supported by Ministerio de Economía y Competitividad, Instituto de Salud Carlos III (PS12/02131 and PI17/02150); Agència de Gestió d´Ajuts Universitaris I de Recerca, Generalitat de Catalunya, 2017 SGR 794; and Fundació Marató TV3 project code 201824.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Cytomegalovirus (CMV) infection is one of the most common complications in solid organ transplant (SOT) patients. Currently, prophylaxis and preemptive treatments have decreased the incidence of disease and death by CMV. However, mutations associated with resistance to antivirals have become an important problem with a high associated morbidity and mortality [1, 2].
In the transplant setting, the incidence of drug resistance mutations (DRM) is 5–12% depending on the group of patients studied but may be higher than 20% in patients with suspected DRM [3, 4].
Diagnostic laboratory testing should be performed whenever there is a criterion of suspicion of resistance since suboptimal response to treatment may have other causes, and inadequate empiric changes in therapy may have adverse effects [5, 6]. Genotypic testing by Sanger sequencing of the protein kinase gene (UL97) and DNA polymerase gene (UL54) is the usual method for detecting drug resistance and is the basis of the selection of alternative therapy [7]. However, Sanger DNA sequencing is unable to detect and quantify subpopulations of viral mutants of less than 20–30%, which may be crucial for selecting the most adequate therapy [8].
In the last years next generation sequencing (NGS) technologies have provided better understanding of the complexity of variant populations, increasing the detection of subpopulations of minority variants which have a prevalence of 1–3%. Nevertheless, to date, only a few studies have evaluated CMV DRM using NGS and most of these studies were carried out in a reduced number of patients [9– 12].
The present study was conducted within the context of the Group for the Study of Infection in Transplantation (GESITRA), constituting one of its major lines of clinical research. The nationwide network of Spanish hospitals participating in GESITRA has allowed the inclusion of a wide variety of SOT recipients.
The aim of this study was to identify CMV drug resistance mutations in solid organ transplant recipients with suspected resistance comparing NGS with Sanger sequencing and to assessing risk factors and the clinical impact of resistance detected by NGS.
Materials and methods
Settings and study population
We conducted a prospective observational study in nine hospitals included in the Spanish Network for Research in Infectious Diseases (REIPI). Adult SOT patients with suspected DRM were included in the study from September 2013 to August 2016. Resistance was suspected when there was persistent or increasing CMV viral load despite ≥ 2 weeks of appropriate antiviral treatment, or failure to have a significant improvement in clinical symptoms despite 2 weeks of full-dose antiviral therapy.[2,5,6]
The coordinating centre was the Hospital Clinic of Barcelona (Spain), which performed the genotypic resistance testing as well as the data analysis. Patient treatment and follow-up were conducted as per the protocol of each centre. All patients were initially treated with standard-dose ganciclovir/valganciclovir (GCV/VGCV) (adjusted to renal function).
Monitoring of CMV viral load was based on local quantitative PCR-based assays. On the suspicion of resistance, plasma samples were collected and frozen at -80°C in the respective hospital and sent in batches to the Microbiology Laboratory of the Hospital Clinic every 4 months. Thus, genotypic testing was not carried out in real time and all patients received empiric treatment on the presence of suspected drug resistance.
Data at patient inclusion and during follow-up were registered in a clinical database at each participating hospital. We reviewed medical histories in terms of viral response, rejection and mortality at 3 months after suspicion of resistance.
The DRM detected by Sanger sequencing and NGS were also included in the database.
CMV infection and disease and acute cellular rejection were defined as described previously [5, 6].
The study was endorsed by Group for the study of Infection in Transplantation (GESITRA) and approved by the Ethical Committee of Hospital Clínic (Comité Ético de Investigación del Hospital Clínic de Barcelona, CEIC), approval number HCB/2012/7598) as the reference committee for the all participating centres according to legal regulations in Spain.
Microbiological studies
In the coordinating laboratory, all samples were kept at -80°C until processing. DNA extraction was performed in 500 μL of each plasma sample using the QIAsymphony system (Qiagen, Hilden, Germany). CMV viral load was confirmed by CMV Real Time PCR (Nanogen Advanced Diagnostics, Italy) according to the manufacturer’s instructions.
Subsequently, all samples were analysed by Sanger sequencing based on PCR amplification of the CMV UL97 protein kinase gene (codons 400–670) in a single fragment and the UL54 DNA polymerase gene (codons 300–1000) in four fragments [3].
Afterwards, antiviral resistance NGS, using Ion Torrent PGM (Ion Personal Genome Machine, Ion Torrent Life Technologies, South San Francisco, CA) was performed by flanking both the UL97 and UL54 gene.
Primer and library design for NGS
To create DNA library, UL97 and UL54 primers and PCR conditions previously validated and described by Sahoo MK et al were used [10].
After establishing the whole technique, pilot tests were performed in Ion Torrent PGM in order to adjust the reproducibility of the technique. The test of reproducibility was carried out in a total of 6 control samples which included: 2 samples without suspicion of resistance, 2 samples from QCMD with known DRMs, and 2 clinical samples with known DRMs tested by Sanger. The results were the same in two sequencing experiments. Each run had a pool of the 6 samples in 3 different libraries with the same DNA extract. [13]
The limit of detection to avoid false positive results was defined as 5% according to Sahoo et al [10]
Sequencing and bioinformatic analyses
Forty-eight samples were sequenced in the Ion PGMTM 35 System for NGS. Quality assessment of the FASTQ files was performed using FastQC–version 0.11.3 [14]. The resulting reads were aligned to the human herpesvirus 5 strain Merlin reference genome (GenBank: AY446894.2) using the BWA-MEM algorithm from BWA version 0.7.12 with default parameters [15]. Alignments were sorted and indexed using Sambamba version 0.5.1 [16]. Duplicate reads in sorted BAM files were identified and removed using Picard tools [17]. GATK base quality score recalibration, insertion and delection (INDEL) realignment, and duplicate removal were applied [18]. The discovery of single nucleotide polymorphisms (SNP) and INDEL and genotyping were simultaneously performed across 12 samples to optimise this method using standard filtering and variant quality score recalibration according to the GATK Best Practices recommendations [19]. Genetic variant annotation and prediction of the effects of variants on genes (such as amino acid changes) were done using snpEff [20]. To do this a snpEff database was created using the information of the reference genome provided by GenBank. The resulting variant call format (VCF) files were post-processed with in-house R [21] scripts using Bioconductor packages [22] to identify mutations by gene (i.e. UL54 and UL97) in order to find the relative position of each mutation to the first position of the corresponding gene and to filter mutations by missing calls by depth at a sample level (DP> = 5), by allele depth (AD> = 5), and by poor quality calls (in which the percentile of minimum quality accepted was 5%).
An average 247382 raw sequences per sample were obtained in the samples sequenced in this study. After pipeline passing quality filters 142.078 reads were obtained per sample.The mean coverage among the different amplicons was up to 2000X
Statistical analysis
A database was developed for the registry of the patients enrolled in Open Clinica 3.1 [GNULesser General Public License (GNU LGPL)].
Categorical variables are expressed as number and percentage of patients, and the median (interquartile range [IQR]) is used for continuous variables with a non-normal distribution and the mean (standard deviation [SD]) for those with a normal distribution. Categorical variables were compared using the X2 test or the Fisher exact test. Continuous variables were compared using the t test or the non-parametric Mann-Whitney test. Logistic regression analyses were performed to identify variables associated with the presence of mutations. Variables showing a significant result in the univariate analyses (p<0.05) were included in the multivariate model (forward stepwise procedure). The Hosmer-Lemeshow goodness-of-fit test was performed to assess the overall fit of the models [23]. The area under the receiver operating characteristic (ROC) curve of the multivariate model to predict the presence of mutations was calculated. Internal validation of the prediction model was conducted using ordinary nonparametric bootstrapping with 1,000 bootstrap samples and bias-corrected, accelerated 95% CIs [24] (S1 Text; S1 Table). The level of significance was set at 0.05 (2-tailed). All analyses were performed using IBM SPSS Statistics 22.0 (Armonk, New York, USA).
Results
During the study period, 48 adults who had undergone SOT and in whom CMV antiviral resistance was suspected were included in the study. Four samples with a low CMV viral load (< 1.000 copies/ml) were excluded from the analysis because neither Sanger sequencing nor NGS could be carried out. Therefore, 44 patients were finally included. The baseline characteristics of the study population are shown in Table 1.
Based on sequence analysis, 12 out of 44 (27%) patients showed CMV DRM by Sanger sequencing, while 16 patients (36%) showed mutations by NGS (Table 2). All the 16 patients had mutations in the UL97 gene and in addition, two patients had one mutation each in the UL54 gene (D413N and P522A). The NGS identified six low abundance resistance mutations that had not been reported by Sanger sequencing, since they were presented in <20% of the viral population. Patients 8 and 16 showed two mutations in UL97 (C592G + M460V and L595W + A594V respectively); in both cases. In these two patients, M460V and A594V, respectively, were only detected by NGS. These six mutations were reproducibly detected at similar levels in two independent sequencing experiments. All the DRM identified by Sanger were also identified by NGS. The lowest depth coverage obtained related to a resistance mutation was 7% of the viral subpopulations.
The detection of mutations by NGS allowed the differentiation and comparison of the patients with and without mutations (Tables 1 and 3). The presence of mutations was more frequent in lung transplant recipients (p = 0.050) and in patients receiving prophylaxis (42.9% vs. 75%, p = 0.039), regardless of its duration. The median time between transplantation and suspicion of resistance was significantly longer in the group with DRM (211.5 days [87.5; 285.0] vs 108.5 days [43.5; 201.5] p = 0.038) as was the median treatment duration before suspicion of resistance (41.5 days [18.0; 75.5] vs 16.0 days [14.5; 22.5] (p = 0.024). At the time of suspicion of resistance, 75% of patients without mutations were receiving therapy for asymptomatic CMV replication and 50% of patients with DRM were receiving treatment for CMV disease. CMV disease was diagnosed in 54.5% of patients. However, biopsy confirmation was achieved in only 3 patients. All the patients were receiving standard GCV/VGCV doses at the time of suspicion of resistance and in the absence of DRM study results the treatment was empirically modified accordingly. In patients with a DRM, mTORi was added or changed empirically (50%, p = 0.009). No statistically significant differences were observed between the patients with and without mutations in relation to outcome at 3 months (Table 3).
Table 4 shows the clinical and virological data of the SOT recipients with known CMV DRM. Of the variables associated with the presence of resistance mutations in the univariate logistic regression analyses (S2 Text. S2 Table), treatment duration before suspicion of resistance was the only factor independently associated with the presence of mutations in the multivariate analysis (log-transformed scale, OR 2.24, 95% CI 1.03 to 4.87). The area under the ROC curve was 0.71 (95% CI 0.53 to 0.88) for the final model of mutations (S1 Fig).
Discussion
Drug-resistant CMV infection is an important emerging problem in SOT recipients. However, it may be underdiagnosed since mutation analysis is not routinely performed and the current gold-standard Sanger sequencing method lacks sufficient sensitivity [1]. In the present study, a NGS was compared with Sanger sequencing to identify mutations associated with resistance in SOT patients with suspicion of resistance. Among the 44 patients recruited, 14 DRM were detected by Sanger in 12 patients (27%) and 20 DRM were detected with NGS in 16 patients (36%), respectively. NGS showed a higher sensitivity for the detection of mutations present in <20% of the population sequenced which were not found by Sanger sequencing. Univariate analysis of the clinical data revealed that the presence of DRM detected by NGS was associated with lung transplantation (p = 0.050), prophylactic treatment (p = 0.039), a higher mean time between transplantation and suspicion of resistance (p = 0.038) and longer treatment duration with GCV or VGCV before suspicion (p = 0.024). However, the latter was the only factor independently associated with the presence of resistance mutations in the multivariate analysis.
Several aspects differentiate the present stydy from other similar studies in the field reported to date, and these aspects can be summarized as follows:NGS was addressed to the UL97 and UL54 genes, a relatively high number of patients, undergoing different solid organ transplanted were studied, and associated risk factors were also evaluated.
The current gold standard for the genotypic detection of CMV drug resistance is Sanger sequencing of the UL97 and UL54 genes. However, this technique cannot detect viral variants that represent < 20% of the viral population [9, 10]. This lack of sensitivity can lead to continued administration of an antiviral to which resistance has developed and the subsequent expansion of initially minority populations of resistant variants leading to the development of treatment failure, increased morbidity and shortened graft survival. On the other hand, modifications in therapy following the identification of GCV resistance is associated with more rapid viral clearance, further emphasising the importance of timely detection of resistance [7, 11, 12]. Recent studies have shown that NGS methods can provide new insights into viral diversity and in the detection of low-abundance variants by the analysis of thousands of amplicons in a single experiment [10, 25]. However, so far, only a few studies have focused on CMV-resistant mutants. Most studies have included few patients and only some cover the UL97 gene [11, 12]. An exception to this is the study by Sahoo et al which evaluated DRM in both of the genes involved using NGS in a considerable number of specimens [10]. However, this study did not include an analysis of clinical factors. In our study, NGS identified six low abundance resistant mutations that had not been reported by Sanger sequencing, thus suggesting that NGS is more sensitive than Sanger since the use of high throughput sequencing allowed the detection of viral subpopulations with a prevalence of 7%. All six mutations were located in the UL97 gene and showed high-level resistance to GCV [26, 27]. In addition, two of these six DRM identified were present as mixed mutations in two patients which may have important implications in patient management. Therefore, studies with prospective monitoring of the abundance of such mutations over time and an assessment of their association with virologic failure are necessary. UL54 DNA polymerase mutations are typically selected after prolonged GCV exposure and were revealed in 2 lung transplant patients who had received a longer prophylaxis and preemptive treatment [28, 29]. However, compared to Sanger sequencing NGS did not improve the detection sensitivity of DRM in the UL54 gene in concordance with the results of Sahoo et al [10].
In the present study longer treatment duration with antivirals before suspicion of resistance was identified as a risk factor associated with DRM (p = 0.24)[1, 2], and is considered a major factor in the selection and emergence of resistance. Lung transplantation showed the highest incidence of DRM. It is important to note that 4 out of 5 lung transplant patients in our study presented DRM and all had received lengthy preemptive treatment and antiviral prophylaxis [29, 30]. Moreover, the presence of DRM was more frequent among transplant recipients who had received prophylaxis (p = 0.039). Nonetheless, other factors may have also contributed to this finding. The interdependence of host, virus and antiviral therapy is likely associated with the development of DRM [31]. In fact, all the patients included in this study had one or more known risk factors predisposing to DRM [1, 2]. Therefore, according to current consensus guidelines genotypic testing confirmation should be done irrespective of any exclusion criteria whenever antiviral resistance is suspected [5, 6], since resistance might otherwise be underestimated similar to what has been described in previous reports [32]. In the present study, DRM were found in unusual situations, such as during prophylaxis in one patient and in 3 patients with CMV D+/R+ serostatus.
Previous studies have reported antiviral resistance to be associated with increased attributable morbidity and mortality [2]. This was not found in the present study, likely because patients were only clinically followed during the first three months. Moreover, clinicians were unaware of the presence of DRM in real time, and therefore, treatment was empirically modified. Thus, we were unable to evaluate the outcome of the patients according to Sanger or NGS sequencing results. This should be addressed in future studies.
In our cohort, 64% of the patients suspected of having resistance to antivirals did not present DRM. Therefore, as suggested previously other factors might contribute to treatment failure [33]. Emphasis should be given to prioritizing the optimization of host factors rather than switching antiviral medications, thereby adjusting immunosuppressive therapy whenever possible. Additionally, blood GCV levels were not available and were not performed during the clinical care of the patients studied, and thus, sub-therapeutic GCV levels may be a possible cause of non-response to treatment in some patients. On the other hand, a recent study of serial specimens received for genotypic testing showed that the first specimen tested negative for DRM in more than half of the patients, although about two-thirds eventually became positive on repeated testing [4]. It could therefore be suggested that patients without DRM but with suspected criteria of resistance should be closely monitored and new samples should be studied for viral mutations.
Our study has limitations regarding the modest sample size and the lack of balance among groups although it does reflect real life in a nation-wide cohort study. Our results should be confirmed in larger and better balanced cohorts. Another limitation is that the analysis of GCV resistance of the mutations was performed only when there was suspicion; Nonetheless, these patients represented the group with the highest risk for antiviral resistant CMV.
In summary, the development of DRM may be a direct effect of a longer duration of antiviral treatment, and thus, the implementation of effective CMV prevention strategies that minimise drug exposure could lead to lower resistance rates. Although the clinical risk factors for drug resistance have been relatively well defined, genotypic testing should be performed in all patients with suspected criteria of resistance. Despite the lack of standardized protocols and the complexity of the bioinformatics analysis for the use of NGS in routine clinical practice, the results of this study suggests that NGS technology improves the genotypic diagnosis of DRM, fundamentally in low-abundance variants and mixed populations allowing early detection of the emergence of CMV-resistant strains and subsequent targeted adjustment of therapy.
Supporting information
S1 Table. Internal validation of prediction model for mutations using nonparametric bootstrap technique.
https://doi.org/10.1371/journal.pone.0219701.s004
(DOC)
S2 Table. Significant univariate and multivariable logistic regression analyses for mutations.
https://doi.org/10.1371/journal.pone.0219701.s005
(DOC)
S1 Fig. Receiver operating characteristic curve analysis of significant variables derived from the logistic regression model for the ability to predict mutations.
https://doi.org/10.1371/journal.pone.0219701.s006
(DOC)
Acknowledgments
We would like to acknowledge GESITRA-SEIMC, Group for Study of Infection in Transplantation of the Spanish Society of Infectious Diseases and Clinical Microbiology and REIPI, the Spanish Network for Research in Infectious Diseases.
We thank Albert Gabarrus (Statistician in Applied Research in Respiratory Diseases, IDIBAPS, UB, CIBERES) for their statistical analysis. Thanks to Institute for Research in Biomedicine of Barcelona, specially at Jose Luis Mosquera and Camille Stephan to contribute in bioinformatic analysis. We thank Donna Pringle, an English-Spanish translator/reviewer based in Barcelona (Spain) for revision of the manuscript.
References
- 1. Le Page AK, Jager MM, Iwasenko JM, Scott GM, Alain S, Rawlinson WD. Clinical aspects of cytomegalovirus antiviral resistance in solid organ transplant recipients. Clin Infect Dis. 2013; 56: 1018–1029. pmid:23243176
- 2. Fisher CE, Knudsen JL, Lease ED, Jerome KR, Rakita RM, Boeckh M et al. Risk Factors and Outcomes of Ganciclovir Resistant Cytomegalovirus Infection in Solid Organ Transplant Recipients. Clin Infect Dis, 2017; 65:57–63. pmid:28369203
- 3. López-Aladid R, Guiu A, Sanclemente G, López-Medrano F, Cofán F, Mosquera MM et al. Detection of cytomegalovirus drug resistance mutations in solid organ transplant recipients with suspected resistance. JClin Virol. 2017; 90:57–63.
- 4. Kleiboeker S, Nutt J, Schindel B, Dannehl J, Hester J. Cytomegalovirus antiviral resistance: characterization of results from clinical specimens. Transpl Infect Dis. 2014; 16:561–567. pmid:24903568
- 5. Kotton CN, Kumar D, Caliendo AM Huprikar S, Chou S, Danziger-Isakov L, et al. Updated International consensus guidelines on the management of cytomegalovirus in solid organ transplantation. Transplantation. 2018; 102 (6): 900–931. pmid:29596116
- 6. Torre-Cisneros J, Aguado JM, Caston JJ, Almenar L, Alonso A, Cantisán S et al. Management of cytomegalovirus infection in solid organ transplant recipients: SET/GESITRA-SEIMC/REIPI recommendations. Transplant Rev; 2016; 30:119–143.
- 7. Lurain NS, Chou S. Antiviral drug resistance of human cytomegalovirus. Clin Microbiol Rev 2010; 23: 689–712. pmid:20930070
- 8. Chou S. Approach to drug-resistant cytomegalovirus in transplant recipients. Curr Opin Infect Dis 2015; 28: 293–299. pmid:26098499
- 9. Houldcroft CJ, Bryant JM, Depledge DP, Margetts BK, Simmonds J, Nicolaou S et al. Detection of 286 low frequency multi-drug resistance and novel putative maribavir resistance in immunocompromised pediatric patients with cytomegalovirus. Front. Microbiol. 2016; 7: 1317 pmid:27667983
- 10. Sahoo MK, Lefterova M, Yamamoto F, Waggoner JJ, Chou S, Holmes SP et al. Detection of Cytomegalovirus Drug Resistance Mutations by Next Generation Sequencing. J Clin Microbiol. 2013; 51: 3700–3710. pmid:23985916
- 11. Schindele B, Apelt L, Hofmann J, Nitsche A, Michel D, Voigt S et al. Improved detection of mutated human cytomegalovirus UL97 by pyrosequencing. Antimicrob Agents Chemother. 2010; 54: 5234–5241. pmid:20876379
- 12. Garrigue I, Moulinas R, Recordon-Pinson P, Delacour ML, Essig M, Kaminski H et al. Contribution of next generation sequencing to early detection of cytomegalovirus UL97 emerging mutants and viral subpopulations analysis in kidney transplant recipients. J Clin Virol. 2016; 80: 74–81. pmid:27214758
- 13. Lopez Aladid R, Guiu A, Mosquera M, Capón A, Cervilla A, Marcos MA. Detección de mutaciones asociadas a resistencia en citomegalovirus en pacientes transplantados de órgano sólido mediante técnicas de ‘Next Generation Sequencing’. Comparativa con Sanger. Enferm Infecc Microbiol Clin 2017; 35: 42
- 14. Andrews S. FastQC: a quality control tool for high throughput sequence data. Babraham Bioinformatics, 2010.
- 15. Li Heng, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics 2009; 25: 1754–1760. pmid:19451168
- 16. Tarasov A, Vilella A.J, Cuppen E, Nijman IJ, Prins P: Sambamba: fast processing of NGS alignment formats. Bioinformatics 2015; 31: 2032–2034. pmid:25697820
- 17.
Picard: A set of command line tools (in Java) for manipulating high-throughput sequencing (HTS) data and formats such as SAM/BAM/CRAM and VCF. Available online at: https://broadinstitute.github.io/picard/index.html
- 18. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis , Kernytsky A et al. The Genome Analysis Toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010; 20: 1297–1303. pmid:20644199
- 19. Van der Auwera GA, Carneiro MO, Hartl C, Poplin R, Del Angel G, Levy-Moonshine A et al. From FastQ data to high confidence variant calls: the Genome Analysis Toolkit best practices pipeline. Curr Protoc Bioinformatics 2013; 11: 1–33
- 20. Cingolani P, Platts A, Lily Wang L, Coon M, Nquyen T, Wang L et al. A program for annotating 112 and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strain w1118; iso-2; iso-3. Fly 2012; 6: 80–92. pmid:22728672
- 21.
R Core Team. R: A Language and Environment for Statistical Computing. Foundation for Statistical Computing 2016. Available online at: https://www.R117 project.org.
- 22. Huber W, Carey V, Gentleman R, Anders S, Carlson M, Carvalho BS et al. Orchestrating high-throughput genomic analysis with Bioconductor. Nat Methods 2015; 12: 115–121 pmid:25633503
- 23.
Hosmer D, Lemeshow S. Applied logistic regression. Wiley; 1989.
- 24.
Efron B, Tibshirani R. An introduction to the bootstrap (Monographs on statistics and applied probability). New York: Chapman and Hall; 1993.
- 25. Hage E, Wilkie GS, Linnenweber-Held S, Dhingra A, Suárez NM, Schmidt JJ et al. Characterization of human cytomegalovirus genome diversity in immunocompromised hosts by whole-genome sequencing directly from clinical specimens. J Infect Dis. 2017; 215:1673–1683 pmid:28368496
- 26. Smith IL, Cherrington JM, Jiles RE, Fuller MD, Freeman WR, Spector SA. High-level resistance of cytomegalovirus to ganciclovir is associated with alterations in both the UL97 and DNA polymerase genes. J Infect Dis. 1997; 176:69–77. pmid:9207351
- 27. Chou S, Waldemer RH, Senters AE, Michels KS, Kemble GW, Miner RC et al. Cytomegalovirus UL97 phosphotransferase mutations that affect susceptibility to ganciclovir. J Infect Dis. 2002; 185: 162–169. pmid:11807689
- 28. Hakki M, Chou S. The biology of cytomegalovirus drug resistance. Infect Dis 2011; 24: 605–611.
- 29. Bhorade S M, Lurain NS, Jordan A, Leischner J, Villanueva J, Durazo R et al. Emergence of ganciclovir-resistant cytomegalovirus in lung transplant recipients. J. Heart Lung Transplant. 2002; 21:1274–1282. pmid:12490272
- 30. Ramanan P, Razonable RR. Cytomegalovirus Infections in Solid Organ Transplantation: A Review. Infect Chemother. 2013; 45: 260–271. pmid:24396627
- 31. Isada CM, Yen-Lieberman B, Lurain NS, Schilz R, Kohn D, Longworth DL et al. Clinical characteristics of 13 solid organ transplant recipients with ganciclovir-resistant cytomegalovirus infection. Transpl Infect Dis. 2002; 4: 189–194. pmid:12535261
- 32. Couzi L, Helou S, Bachelet T, Moreau K, Martin S, Morel D et al. High incidence of anticytomegalovirus drug resistance among D+R- kidney transplant recipients receiving pre-emptive therapy. Am J Transplant. 2012; 12: 202–209. pmid:21967659
- 33. Limaye AP. Ganciclovir-resistant cytomegalovirus in organ transplant recipients. Clin Infect Dis. 2002; 35: 866–872. pmid:12228824