Resistance targets: blood culture ID panel pitfalls

There is no clear data on S. aureus CoNS co-isolation rates for blood culture, Dr. Davis said. “We know this can happen with other specimen types; in MRSA nose swabs, we can get both types.”

CoNS, including S. epidermidis, is the most common blood bottle contaminant, so the risk exists, though reports are scarce, he said.

Newer design panels such as the FilmArray BCID2, or standalone MRSA-detection tests like the Cepheid Xpert MRSA/SA or the BD Max StaphSR, have an additional target that mitigates this potential risk, he said. This additional target is the MREJ, short for mec(SSCmec-orfX) right-extremity junction. “When this target is positive, it not only shows that mecA is present but that it has been inserted specifically into the S. aureus genome,” he said.

When are bloodstream infection resistance determinants, however accurate the results, not as meaningful? “That is usually when they can’t describe or rule out resistance,” Dr. Davis said, noting this relates a lot to Gram-negative antibiotic resistance mechanisms.

There are several diverse mechanisms that contribute to Gram-negative resistance, and few definitive genetic determinants, he said. “So you could not have one but could have other types. Practically speaking, if a Gram-negative resistance gene is detected, it is worthwhile to assume the isolate is resistant. However, if those genes are not detected, can you be sure the isolate is not resistant? Not so much.”

On existing rapid bloodstream infection tests, most of the targets are for genes encoding beta-lactamase enzymes (ESBL or carbapenemase enzymes). (The FilmArray BCID2 also has a target for colistin resistance, mcr-1, which is not beta-lactamase mediated, he said.)

In reviewing Gram-negative resistance mechanisms, Dr. Davis pointed out that it is possible to get a beta-lactamase, bla_X gene—with a different family of a beta-lactamase—either on a plasmid or inserted into a chromosome. “The plasmid-based are usually ESBLs or CREs,” he said, while chromosomes are usually AmpC types. “And when those beta-lactamases are expressed, they can destroy the beta-lactam antibiotics.”

However, Gram-negatives also have porins and efflux pumps that can physically remove or block entry of different antibiotics, he said. “So you have multiple things going on, and no single bloodstream infection test is going to catch everything. The idea is to catch most of it.”

Dr. Davis shared a 2019 meta-analysis of the sensitivity and specificity of rapid diagnostic tests for antibiotic resistance (De Angelis G, et al. Clin Microbiol Infect. 2020;26[3]:271–280). The authors examined 20 studies (3,310 isolates, 2006 to 2019) that compared the Verigene and FilmArray systems with phenotypic and/or genotypic comparator methods. Overall, Dr. Davis said, the rapid tests had high specificity but less sensitivity compared with phenotypic testing.

“Unsurprisingly, when you have a separate comparative test looking at those genes, the rapid detection test targets—like the CTX-M and the carbapenemase genes—perform very well, so there’s very good correlation,” Dr. Davis said.

“The issue then is when you compare the performance of phenotypic tests, meaning you incubate your organism, your Gram-negative rod, and you see what classes of antibiotics it’s resistant to, that’s when you start to see there’s not as great a correlation. It detects a lot of them, but the sensitivity is somewhat low.”

“And when you look at what the overall sensitivity and specificity is, pooled sensitivity only comes to be about 85 percent in terms of capturing all of the potential extended-spectrum beta-lactamase resistance, whereas the specificity is very good.” If you find those genes then, it’s almost certainly going to be resistant, he said. “If you don’t find those genes, you’re only 85 percent sure you’ve ruled it out.”

Acting on bloodstream infection antimicrobial resistance determinants is the key. A Vanderbilt Children’s Hospital retrospective review of children with bloodstream infections and Verigene testing yielded two interesting findings, Dr. Davis said. “They looked at 301 positive blood cultures and found that of those, looking at the chart of what they could have done, in 57 percent [171] of the cases, the Verigene results revealed a chance to change antibiotics.” In 18 percent (30) of the 171 cases, antibiotics could be avoided altogether. In 36 percent (61) of cases, antibiotics could be deescalated. In 16 percent (28) of cases, antibiotics would be escalated (Juttukonda LJ, et al. J Clin Microbiol. 2020;58[4]:e01400–19).

Of the 171 cases in which there was the potential to change antibiotics, change occurred in 119 cases (70 percent). Why no change in 30 percent of cases? “There seems to be a reluctance or an inability to change antibiotics,” Dr. Davis said.

Clinicians were significantly slower to deescalate antibiotics rather than escalate antibiotics, he said, and day shift results led to significantly faster change in antibiotics than night shift results. The worst time for antibiotic changes was between 12 and 6 AM.

“Acting on and performing these tests is a human-based endeavor,” Dr. Davis said, “and you need humans to act on it and decide to make the change when you can.”

Without good genotypic methods to rapidly rule out antibiotic determinants, “we ideally can get faster phenotypic susceptibility testing,” Dr. Davis said. “The gold standard—the best and top-shelf option—is the Accelerate PhenoTest BC system that does identification and susceptibility testing at once.”

But there are also opportunities to use existing technologies, like rapid ESBL screens for Gram-negative rod blood cultures, or direct inoculation of blood culture broth into automated microbroth instruments, he said. His laboratory at Providence Sacred Heart Medical Center uses the BD Phoenix for direct susceptibility and reduced its time to AST report from 44.3 hours to 17.6 hours.

New assays in the pipeline include the T2 system, which has a direct-from-blood (not blood culture) resistance panel. “Whether that makes sense depends on how results are going to be used,” Dr. Davis noted. On the market are broader panels of multidrug resistant genes, he said, citing the Check-MDR Microarray panel (Powell EA, et al. Microb Drug Resist. 2020;26[7]:825–830) and Acuitas MDRO Gene Test (for urine isolates). But those mean extensive testing on all bloodstream infections, which Dr. Davis described as “questionable” in terms of cost/benefit and “challenging to bring on board.” 

Amy Carpenter Aquino is CAP TODAY senior editor.