NEW YORK (GenomeWeb) – A team led by researchers from the University of Washington has developed a MALDI-TOF mass spec-based method for detecting vancomycin non-susceptible methicillin-resistant Staphylococcus aureus (MRSA).
Detailed in a January paper in the Journal of Clinical Microbiology, the approach marks a step forward for efforts applying mass spectrometry to the detection of antibiotic resistance and provides a more rapid method of detecting vancomycin intermediate S. aureus (VISA) and heterogeneous VISA (hVISA) organisms, said Susan Butler-Wu, a former UW researcher (now at the University of Southern California) and senior author on the paper.
Of particular note, said Butler-Wu, is the method's ability to detect hVISA, strains in which only a small proportion of the individual organisms may be vancomycin non-susceptible. The gold-standard method for detecting such strains is too expensive and complicated to be used in actual clinical settings, she said.
Butler-Wu and her colleagues performed the work on Bruker's MALDI Biotyper mass spec system, which, along with BioMérieux's competing Vitek MS system, has for several years now been used for microbial identification in clinical settings.
Both systems have 510(k) clearance from the US Food and Drug Administration and have seen significant adoption from clinical microbiology labs due to their ability to provide accurate organism identifications more rapidly and less expensively than many existing methods. The systems have made less headway, though, in identifying antibiotic-resistant organisms and, in recent years, this has emerged as a major area of research interest.
For instance, in recent studies, researchers at the National Institutes of Health have developed a method using the MALDI Biotyper to detect resistance to carbapenems, a class of antibiotics commonly used in hospitals and in patients with infections already resistant to other antibiotics. In that work, they determined that the protein pKpQIL_p019, which is encoded by a gene found on some resistance-conferring plasmids can be identified in spectra collected by the MALDI Biotyper during routine identification assays and can be used as a proxy for carbapenem resistance.
Such an effort, which focuses on identifying a specific MALDI peak present in resistant strains and absent in susceptible strains is characteristic of most current approaches to applying MALDI to the identification of resistant strains, Butler-Wu told GenomeWeb.
She noted that she and her colleagues' approach to detecting VISA was somewhat different in that rather than look for a single distinguishing peak, they used broad proteomic profiles to distinguish between resistant and susceptible organisms.
Such an approach is not particularly novel in protein biomarker research generally. Biodesix's Veristrat lung cancer test, for instance, uses comparisons of broad MALDI-based proteomic profiles to distinguish between likely good and poor responders to therapies. However, Butler-Wu said, it is less common in MALDI-TOF-based clinical microbiology work.
"To our knowledge this is one of the first studies that has taken this approach of not just looking for the absence or presence of a particular [MALDI] peak but rather looking at the spectra overall," she said.
She suggested that such approaches have been rare due to microbiologists' relative inexperience with mass spec. "I was really lucky to have an incredibly talented resident [first author and UW researcher Cheryl Mather]," she said. "This was her project and she had the capability to program in R, which is something most microbiologists aren't doing."
Butler-Wu said that VISA stains exhibit systemic changes compared to susceptible organisms that made her and her colleagues think a broad proteomic profiling approach could prove effective.
"One of the reasons [we] felt this was more likely to work for this particular type of resistance was strains of staph that are intermediately resistant to vancomycin are known to have a lot of cellular changes such as thickened cell walls that you can see by electron microscopy," she said. However, she noted, even in cases where resistance is conferred by a single gene on a plasmid, a broad profiling approach could prove beneficial.
"Even if you just have a plasmid conferring resistance, there is potentially a cost to the organism to having that plasmid, and these may be things that you could pick up by mass spectrometry," she said. "So, I think this is something that in the coming years we are going to see more and more of."
The researchers used a machine learning-based approach to distinguish between the MALDI spectra of VISA, hVISA, and susceptible strains, looking at 21 VISA, 21 hVISA, and 38 susceptible isolates. Using the method they were able to correctly identify 100 percent of VISA and 97 percent of susceptible strains. When they added hVISA, they identified VISA with 100 percent accuracy, susceptible strains with 89 percent accuracy, and hVISA with 76 percent accuracy.
In the case of VISA, the MALDI-based approach provides more rapid turnaround than traditional susceptibility testing, potentially allowing clinicians to more quickly move patients with resistant infections onto a drug other than vancomycin. In the case of hVISA, the MALDI-based test could provide a practical clinical approach where none currently exists.
Detecting hVISA is difficult because such populations consist overwhelmingly of susceptible organisms with only a few resistant organisms mixed in.
"The problem is, when you do susceptibility testing, you only test around 10,000 microorganisms," said Butler-Wu. But, the hVISA population may only be present at levels on the order of one in a million cells.
Testing for such organisms requires a modified population analysis, she said, noting that such analyses are very time consuming and expensive and not practical for a clinical lab. "So, we don't have a good way to test for that phenotype."
While additional validation is needed, the 76 percent accuracy demonstrated in the JCM paper provides enough negative predictive value that the method could be an effective rule-out test for hVISA, Butler-Wu said. "In theory, the negative predictive value is pretty high, and so if you had a negative result, providers could feel fairly confident about knowing it is pretty unlikely there is going to be an hVISA subpopulation."
Butler-Wu said she and her colleagues would like to do additional validation, including longitudinal testing of the method. "We would like to move forward doing some multicenter studies trying to evaluate this," she said.
She declined to speculate on how FDA might view the test, but, she said, independent of regulatory questions, the method would be fairly easy for clinical labs already using MALDI-based microbial ID to implement.
"It is pretty simple to do," she said. "Essentially you run your mass spec like you usually would, and the software is not fancy. It would be pretty easy to teach a [medical laboratory scientist] to run the program and get your answer."