Colistin Breakpoints Redux – or, the Fun Don’t Stop, Yo!

March 26, 2020

New decade, new pandemic, a new M100 Standard and … (wait for it!) a new set of colistin breakpoints! It’s like 1976, 1980, 2003 and 2014 all over again. (You probably didn’t realize the Clinical and Laboratory Standards Institute (CLSI) updated colistin breakpoints in those years, but they did.)

So, what has changed? Prior to 2020, CLSI recognized colistin breakpoints for Pseudomonas aeruginosa and Acinetobacter baumannii only, not for the Enterobacterales. Technically, CLSI did have a colistin breakpoint for the enterics in the late 1970s, but it was a disk-only breakpoint that was hard to correlate with minimum inhibitory concentration (MIC). The breakpoint of ‘76 was deleted from the M100 in 1980 when colistin was put on the shelf, given the availability of better and less toxic drugs (e.g., cephalosporins). This all happened before breakpoints were added to drug labels, so there are no colistin breakpoints recognized by the U.S. Food and Drug Administration (FDA), something that remains a problem.

Fast forward 40 years and the absence of a colistin Enterobacterales breakpoint has become a bigger and bigger problem. First, some strains of Enterobacterales have become resistant to all other antibiotics, making colistin look like a great option (at least in vitro). Second, the phosphoethanolamine transferase colistin resistance mechanism, typically conferred by plasmid-borne mcr-1 through mcr-8 genes, has emerged in Escherichia coli and Salmonella. Tracking the spread of this resistance mechanism through surveillance of clinical isolates is challenging, given there are no FDA-cleared antibiotic susceptibility tests (ASTs) for colistin. Why? In order to have a test in the U.S., you need an FDA-recognized breakpoint. No breakpoint = no test.

This is where the 21st Century Cures Act (“Cures Act”) comes into play. This exciting piece of legislation (really!) changed the relationship between 3 main actors with regards to breakpoints and AST. The U.S. FDA Center for Drug Evaluation and Research (CDER) now recognizes most of the CLSI M100 breakpoints, and with each update to M100, CLSI submits a rationale document outlining breakpoint changes for CDER consideration. CDER then either accepts, denies or delays each breakpoint change. If CDER accepts the breakpoint, it’s posted online and the U.S. FDA Center for Devices and Radiological Health (CDRH, which regulates AST tests) can accept submissions from diagnostic manufacturers for clearance of their tests based on that breakpoint. Intrigued? CLSI offers a free on-demand webinar for more on this topic, and this handy flow chart summarizes the process:

Flowchart for the process of recognizing antibiotic susceptibility testing breakpoints and getting commercial tests approved as laid out by the 21st Century Cures Act.
Flowchart for the process of recognizing antibiotic susceptibility testing breakpoints and getting commercial tests approved as laid out by the 21st Century Cures Act.
Source: Romney Humphries

Colistin, also known as polymyxin E, was isolated in  Britain in the 1940s as the first antimicrobial to have unique gram-negative activity, sparing gram-positive bacteria. Despite having been around for nearly the entirety of the antimicrobial age, it has remained relatively steeped in mystery. A “Dr.-Jekyll-and-Mr.-Hyde” story for colistin began to emerge - that while the drug was very effective against bacteria, it also damaged renal tubule cells. These claims were never really well-validated. As newer antimicrobial agents came to market, it faded to obscurity and this question was never resolved. The emergence of multi-drug resistant (MDR) gram-negative bacteria in the mid-2000s brought colistin back into action, as it offered tantalizing in vitro activity against these infections. In 2019, CLSI decided to investigate setting a colistin breakpoint for the Enterobacterales. This would allow CLSI to submit a rationale document to CDER, CDER to hopefully recognize the breakpoints and CDRH to clear tests. We’ve taken colistin down this pathway once before for A. baumannii and P. aeruginosa. CDER opted to deny the breakpoint rationale because pharmacokinetic/pharmacodynamic analyses showed that it would be impossible to achieve target concentrations in the lung. Fair enough.

A lot of good work has been done to evaluate the activity and toxicity of colistin since its comeback in the 2010s, which has taught us several things summarized in a recent CLSI / EUCAST position statement:

  1. Most gram-negative bacteria have MICs ≤ 2 μg/mL for colistin.
  2. Pharmacokinetic/pharmacodynamic data suggest an average steady-state plasma concentration of 2 μg/mL is the target for efficacy against isolates with MICs of ≤ 2 μg/mL.
  3. Colistin is indeed responsible for a lot of kidney damage, and it is dose dependent. As many as 65% of patients with colistin trough concentrations >2.2 μg/mL will experience acute kidney injury.
  4. It is nearly impossible to attain an average steady-state concentration in the blood of 2 μg/mL if the patient has normally functioning kidneys.
  5. When given intravenously, colistin concentrations are nowhere near 2  μg/mL in the lungs.
  6. The patient outcomes data associated with using colistin really don’t look very good.

If you look at 1-3 in a vacuum, it is pretty clear: the breakpoint that splits “susceptible” and “resistant” bacteria should be 2 μg/mL. You can’t achieve higher concentrations without frying the kidneys. Lower concentrations bisect the wild type MIC distribution (a no-no if you want a test that works).

The wheels fall off a little bit when you include 4-6. Colistin is administered intravenously as the prodrug, colistin methanesulfonate. This means the human body has to convert it to the active form, colistin, before it begins killing bacteria. It is also rapidly cleared by the kidneys. Unfortunately, the rate of conversion to colistin is slower than the rate of renal clearance. In fact, less than 50% of patients with normal kidney function achieve the target concentration of 2 μg/mL.

Low colistin concentrations in the lungs is well documented and is, in fact, why the FDA did not recognize the CLSI A. baumannii / P. aeruginosa breakpoints. You can learn more about this in a recent international treatment guidance document for the polymyxins.

Finally, many new drugs coming to market with specific activity against MDR A. baumannii and P. aeruginosa have been evaluated in studies against a “best available therapy” comparator, which often includes colistin. In these studies, 24-56% of patients in colistin-containing regimens die within 30 days vs. 10-17% for patients treated with an alternative agent. A quarter to half of patients dying following colistin therapy doesn’t really equate to a “high probability of treatment success,” the current CLSI definition of “susceptible.” As one CLSI member pointed out, giving a patient saline alone might be better, because at least it wouldn’t damage their kidneys.

All that being said, CLSI opted to move ahead with colistin breakpoints for several reasons:

  • Enable the FDA to clear AST tests (which, if anything, could detect “resistant” bacteria and show no role for colistin in treating a patient vs. a small role).
  • Provide guidance on best practices for colistin use.
  • Warn clinicians and laboratories that colistin use is fraught with challenges, including high risk of mortality and renal toxicity, by including a comment to this effect in the M100.

In the end, CLSI assigned an “intermediate” interpretation to any MIC <= 2ug/mL, and a “resistant” interpretation to any MIC > 2 ug/mL. That’s right – there is no susceptible breakpoint. It just wasn’t justified.

What about colistin’s sister molecule, polymyxin B, you ask? In vitro, the activity of both drugs is nearly identical. In vivo, there are differences. First, polymyxin B is administered in its active form, so it is not as hard to achieve target serum drug levels. Second, it seems like the risk of renal toxicity is lower, although that is being further investigated in upcoming clinical trial data. There is no need to renally adjust dosing as the drug is not excreted extensively in the urine. Just like colistin though, polymyxin B does not penetrate the lungs effectively.

How can labs adopt these new breakpoints? First, give some consideration to when colistin testing might be indicated. For an isolate from the respiratory tract or one that is sensitive to other, newer drugs (like ceftazidime-avibactam, meropenem-vaborbactam, etc.), testing probably isn’t warranted. Colistin AST should only be performed and reported as a last resort. Second, how should you report results? It’s probably best to discuss reporting with your ID physicians, ID pharmacists and stewardship team, and to consider adding recommendations for use of maximum doses and addition of a second agent to these reports, as is indicated by CLSI.


Author: Romney Humphries, Ph.D., D(ABMM), M(ASCP)

Romney Humphries, Ph.D., D(ABMM), M(ASCP)
Romney Humphries, Ph.D., D(ABMM), is a Professor of Pathology, Microbiology and Immunology at Vanderbilt University Medical Center as serves as the Director of Infectious Diseases Laboratories & the Division of Laboratory Medicine