SAN FRANCISCO (GenomeWeb) – Cell-free DNA shed into urine could be helpful for identifying bacterial and viral infections that often occur in patients who receive kidney transplants, according to a study published this week in Nature Communications.
The work, a collaboration between Iwijn De Vlaminck's lab at Cornell University and John Lee's group at Weill Cornell Medicine, suggests that infections may be better detected from urinary cfDNA than using conventional diagnostic methods. In addition, analyzing cfDNA could have other benefits, such as identifying antimicrobial resistance and tissue damage, and the techniques could potentially be applied not only to kidney transplant patients but also other transplant patients, as well as the general population, to identify infection and pinpoint tissues being damaged by the infection.
In the study, the researchers applied their cfDNA assay to 141 urine samples collected from 82 kidney transplant recipients, including those diagnosed with bacterial or viral infections, and found that the cfDNA results not only agreed with conventional testing but that they also identified bacteria and viruses that conventional methods did not detect.
The work was born out of a five-year $2.3 million National Institutes of Health grant that De Vlaminck received last year to develop noninvasive and more sensitive genomic-based assays to identify kidney transplantation problems. Under the grant, his team is also working on single-cell techniques and methods to harness epigenetic information to identify tissue injury. This is the first published study resulting from the grant.
More than 15,000 kidney transplantations are performed each year in the US, and although they have a more than 90 percent success rate one year after surgery, as time goes on, that success rate drops, with rejection and infection becoming more common. Infection occurs in 20 percent of recipients within the first year and at least half of recipients three years post transplantation. Genomic methods of diagnosing infection offer the promise of greater sensitivity, while analyzing urinary cell-free DNA represents a noninvasive alternative to tissue biopsy for diagnosing rejection.
De Vlaminck said the next steps are to test the method in a larger cohort of patients and in prospective trials.
Michael Oellerich, a professor of clinical chemistry at the University of Göttingen in Germany who was not involved in the study, but whose research focuses on using cell-free DNA in both transplantation and oncology, said that the study demonstrated a "novel, exciting approach that will stimulate further research and that has the potential to improve outcomes in kidney transplantation." Because the technique is a broad approach, it "could detect things that would be missed with current practices."
Oellerich noted, however, that the approach needs to be tested in independent studies and also requires more analytical and clinical validity information, including sensitivity, specificity, and positive and negative predictive values.
In the study, the researchers first isolated cell-free DNA from urinary samples. They then used a previously developed library prep approach that is based on single-stranded ligation. The method has been optimized to work on the highly fragmented cell-free DNA found in urine that is often 50 bases in length or shorter, De Vlaminck said. The more common double-stranded library prep methods often require a size selection step to weed out adapter dimer products, which has the side effect of also filtering out many of those shorter cfDNA fragments. The library prep approach De Vlaminck's team developed, on the other hand, does not require a size selection step, so it is able to keep those shorter fragments and, as a result, is able to begin with a smaller sample size — 1 milliliter of urine as opposed to 10 milliliters.
Another interesting aspect about using urine as a sample is that the ratio of host background DNA to bacterial or viral DNA is not as large as in a blood, De Vlaminck said. That's helpful, he noted, because it can enable a lower sequencing depth, and thus lower costs, to reach the same detection sensitivity. For instance, according to the study, in individuals diagnosed with a urinary tract infection, bacterial cfDNA accounted for up to nearly 35 percent of raw sequencing reads.
In the study, the researchers used the single-stranded library prep method, followed by shotgun sequencing and bioinformatics to analyze the bacterial and viral DNA fragments. They compared their approach to conventional culture-based diagnostics.
There were 43 samples from 31 individuals who had been diagnosed with urinary tract infection with a positive culture sample. In 41 samples, conventional culture was able to report a bacterial organism down to a species level. The NGS-based approach agreed with the culture technique in 40 out of the 41 samples. However, in the one discordant sample, De Vlaminck said that subsequent testing and follow-up with the patient convinced the team that the culture, not the NGS test, was incorrect.
Looking deeper into the data, the team found that in 40 percent of the 43 urinary tract infection samples, the culture technique did not identify the most prevalent organism, which the team attributed to the biases of culture that favor fast-growing species that can be easily isolated and cultured. In addition, they noted, cfDNA analysis enabled identification of a much broader range of species.
The team also used the cfDNA technique to screen for viral infections. Infection with a virus known as BK polyomavirus can occur in 5 percent to 8 percent of kidney transplant recipients.
CfDNA sequencing found detectable levels of virus in nearly half, or 66, of the 141 samples, including multiple polyomavirus species and viruses not routinely screened for in kidney transplant patients. In two patients who were diagnosed with parovirus B19, cfDNA analysis identified infection eight and 80 days, respectively, before conventional diagnosis.
Diagnosing BK polyomavirus noninvasively could be especially useful, De Vlaminck said, because infection with that virus can "cause kidney disease and lead to tissue damage." In addition, the virus is typically diagnosed via an invasive biopsy.
Indeed, in the study, the researchers noted that in samples diagnosed with BK polyomavirus, levels of donor-derived cfDNA were much higher than in patients without viral infection and without any urinary tract infection. Donor-derived cfDNA in the urine can indicate injury to the graft tissue, De Vlaminck noted.
Aside from simply diagnosing infection, cfDNA sequencing also enables antimicrobial resistance profiling, which Oellerich said would be another key advantage of the method. "Drug resistance is extremely important and unfortunately a growing problem," he said.
In the study, the researchers identified genes conferring resistance to various classes of antimicrobials, including vancomycin.
With one test, "you can detect pathogens, acquire phenotypic information, and also get the host response to the infection," De Vlaminck said.
He said the team's next steps are to conduct a larger prospective study of utility, which would help determine performance characteristics and how early infection can be diagnosed.
In addition, he said, the team is already doing studies using single-cell analysis and epigenetic profiling to look at host responses in more detail. Epigenetics, in particular, he said, could enable the work to be applied in a general, non-transplant population to identify tissue injury from infection.