High-resolution profiling of human cytomegalovirus cell-free DNA in human plasma highlights its exceptionally fragmented nature

Vikas Peddu, Benjamin T. Bradley, Amanda M. Casto, Raj Shree, Brice G. Colbert, Hong Xie, Tracy K. Santo, Meei‐Li Huang, Kwang‐Ting Cheng, Eric Q. Konnick, Stephen J. Salipante, Keith R. Jerome, Christina M. Lockwood, Alexander L. GreningerScientific 2020 · DOI

TL;DR

This paper reveals that human cytomegalovirus (CMV) cell-free DNA (cfDNA) in plasma is significantly more fragmented than human cfDNA. This fragmentation explains why current qPCR-based CMV quantification methods vary so widely between labs and suggests that cfDNA sequencing offers a more accurate and sensitive approach for viral load assessment.

What Problem Does This Paper Address?

Accurately quantifying human cytomegalovirus (CMV) viral loads is crucial for managing CMV infections, particularly in immunocompromised individuals and for preventing congenital CMV disease, a leading cause of newborn impairments. However, current methods, primarily quantitative polymerase chain reaction (qPCR), suffer from significant inter-laboratory variability, with results differing by up to 100-fold for the same sample. A major suspected cause of this variability is the size of the DNA fragment (amplicon) used in the qPCR assay, as shorter amplicons tend to yield higher viral load measurements. Despite this, there has been a lack of precise measurements of CMV cfDNA fragment lengths in plasma, hindering a full understanding of this variability and the development of standardized, accurate quantification methods.

Methods

The researchers leveraged existing plasma cfDNA sequencing data, originally generated for non-invasive prenatal aneuploidy screening. This dataset comprised 2,208 maternal plasma samples. They developed a bioinformatics pipeline to identify CMV reads within this sequencing data. Crucially, their method allowed for the precise measurement of cfDNA fragment lengths, using paired-end sequencing reads to determine insert sizes. To validate their findings, they also sequenced cfDNA from seven plasma samples of solid organ transplant patients with known CMV infections. They further employed a simulation approach to model how different qPCR amplicon lengths would impact viral load measurements based on the observed CMV cfDNA fragment length distributions.

Datasets

  • Maternal Plasma cfDNA Sequencing Data: 2,208 samples collected for non-invasive prenatal aneuploidy screening between May 2017 and November 2018 from pregnant women at the University of Washington.
  • Solid Organ Transplant Patient Plasma Samples: Seven CMV-positive plasma specimens from three solid organ transplant patients.

Key Insights and Ideas

  • CMV cfDNA is Exceptionally Fragmented: The most striking finding is that CMV cfDNA fragments are significantly shorter than human cfDNA fragments circulating in the same plasma. This suggests a distinct biological origin or degradation pathway for viral cfDNA.
  • Fragmentation Explains qPCR Variability: The precise measurement of CMV cfDNA fragment lengths provides a direct explanation for the observed discrepancies in qPCR-based viral load measurements. Shorter fragments are more likely to be captured by shorter qPCR amplicons, leading to higher apparent viral loads.
  • cfDNA Sequencing as a Superior Tool: The study highlights the potential of cfDNA sequencing not just for detection but for highly accurate and sensitive quantification of viral loads, as it can detect fragments that would be missed by standard qPCR assays.
  • HHV-6 as a Comparative Control: The analysis of Human Herpesvirus 6 (HHV-6) in cfDNA provided a valuable comparison, showing that HHV-6 fragments were generally longer and more similar to human cfDNA, suggesting that the extreme fragmentation of CMV might be specific to this virus.

Key Results

  • CMV reads were identified in 120 out of 2,208 maternal plasma samples (5.4%).
  • The median cfDNA fragment size for CMV was significantly shorter than for human cfDNA: 103 bp for CMV vs. 172 bp for human (p < 0.0001). This placed CMV fragments at the 3rd percentile of human cfDNA fragment lengths.
  • In solid organ transplant patients, CMV cfDNA fragments also showed a remarkably short median length of 149 bp.
  • The simulation modeling demonstrated that measured plasma viral loads are strongly affected by qPCR amplicon length, with shorter amplicons yielding higher estimates, consistent with the observed CMV cfDNA fragmentation.
  • HHV-6 cfDNA fragments were found to be longer, with a median length of 167 bp, more closely resembling human cfDNA.

Limitations and Future Work

  • Limited Sample Size for Transplant Patients: While transplant patient data validated the fragmentation finding, the sample size was small (seven specimens).
  • Short-Read Sequencing Limitations: The study acknowledges that short-read sequencing limits fragment length estimation to under 500 bp, meaning potentially longer CMV fragments could exist but were not captured.
  • Clinical Outcome Data: Clinical follow-up data for the maternal NIPT cohort was limited, making it difficult to directly correlate CMV detection with specific congenital outcomes.
  • Unclear CMV Fragmentation Mechanism: The paper hypothesizes potential reasons for CMV's extreme fragmentation (e.g., viral replication, host immune pressure, epigenetic state) but acknowledges that the exact mechanism remains unclear and warrants further investigation.
  • Future Directions: The authors suggest that cfDNA sequencing, with its ability to precisely measure fragment lengths, could be a valuable tool for developing standardized and more accurate CMV quantification methods. They also propose that sequencing fragments <100 bp might be a strategy to enrich for pathogen sequences over human DNA. Further studies are needed to explore the clinical utility of CMV cfDNA fragment size and quantity for predicting congenital infection.

Summary updated

Abstract

Human cytomegalovirus (CMV) infections comprise a leading cause of newborn impairments worldwide and are pervasive concerns among the immunocompromised. Quantification of CMV viral loads is increasingly used to guide definitions of CMV disease but standardization of CMV quantitation remains problematic, mostly due to differences in qPCR amplicon sizes between clinical laboratories. Here, we used plasma cfDNA sequencing data from 2,208 samples sent for non-invasive prenatal aneuploidy screening to detect CMV and precisely measure the length of CMV fragments in human plasma. CMV reads were identified in 120 (5.4%) samples. Median cfDNA fragment size derived from CMV was significantly shorter than cfDNA derived from human chromosomes (103 vs 172 bp, p < 0.0001), corresponding to the 3<sup>rd</sup> percentile of human cfDNA. Sequencing of cfDNA from seven plasma samples from transplant patients positive for CMV confirmed the extraordinarily short nature of CMV cfDNA fragment size with a median length of 149 bp. We further show that these high-resolution measurements of CMV DNA fragment size accurately predict measured discrepancies in serum viral load measurements by different qPCR assays. These results highlight the exceptionally fragmented nature of CMV cfDNA and illustrate the promise of plasma cfDNA sequencing for quantitating viral loads through detection of fragments that would be unrecoverable by qPCR.
BibTeX
@inproceedings{peddu2020high,
  author = {Vikas Peddu and Benjamin T. Bradley and Amanda M. Casto and Raj Shree and Brice G. Colbert and Hong Xie and Tracy K. Santo and Meei‐Li Huang and Kwang‐Ting Cheng and Eric Q. Konnick and Stephen J. Salipante and Keith R. Jerome and Christina M. Lockwood and Alexander L. Greninger},
  title = {{High-resolution profiling of human cytomegalovirus cell-free DNA in human plasma highlights its exceptionally fragmented nature}},
  booktitle = {Scientific},
  year = {2020},
  doi = {10.1038/s41598-020-60655-6},
}
Figure