Safety in numbers: Advanced biostatistical model resolves questions on vancomycin side effects

3 scientists working on glass board

Biostatistician Yilun Sun [left] is the lead author of a published report on antibiotic (vancomycin) safety that was driven by advanced statistical models. Co-authors include Dr. Li Tang [center], Department of Biostatistics, and Dr. Josh Wolf [right], Department of Infectious Diseases.

We know the antibiotic vancomycin (Vancocin) is very effective at preventing infections, but could it actually be raising the risk of other complications? Through other studies at St. Jude, we have known for some years that vancomycin, given as antibiotic prophylaxis, prevents many infections that would otherwise make kids sick during treatment. But vancomycin can also have serious side effects, so we wanted to be sure that it was the safest choice.

That’s why we began an intensive review of patients who were treated for acute myeloid leukemia (AML) at St. Jude Children’s Research Hospital. Our central question— “is vancomycin safe for AML patients?”— sounds fairly simple, but finding the answer required complex formulas, applied by our statistical experts in the Department of Biostatistics. Our review appears in Antimicrobial Agents and Chemotherapy, published by the American Society for Microbiology.

The study owes its complexity to the multi-layered threats faced by patients with AML

Chemotherapy is a primary treatment for AML, but both the disease and the treatment weaken the body’s ability to fight dangerous infections. These bacterial infections can lead to a loss of kidney or liver function and other serious complications. That’s why we administer antibiotics like vancomycin with the treatment. However, some studies have raised concerns that vancomycin may increase the risk of the same liver and kidney complications we are trying to prevent.

Did those past findings mean we shouldn’t be using vancomycin? Not really. Those studies were not conducted under the same conditions faced by our AML patients. While the drug has been tested for liver and kidney side effects when treating existing infections, it had not undergone such testing when used to prevent potential infections in AML patients. There was nothing wrong with how those studies were conducted, but the cases they examined included many factors that could have confounded the results.

We wanted one of two simple conclusions: Either vancomycin did cause damage to our patients’ organs, or it didn’t. (Spoiler: It didn’t.)

In testing our own use of the antibiotic, St. Jude has the advantage of a large number of previous chemotherapy courses to analyze: AML patients who were given vancomycin under the same clinical protocols by the same teams of doctors. In other words, we were in a good position to compare apples to apples.

To get closer to the answer, the Department of Infectious Diseases began working with members of our Biostatistics and Oncology departments to analyze the cases. We ran statistical models on 476 courses of chemotherapy that had been given to 116 AML patients at St. Jude between 2002 and 2008.

Josh Wolf, MD, explains how the Infectious Diseases and Biostatistics departments resolved safety questions on vancomycin therapy for acute myeloid leukemia patients.

These chemo courses fell into three categories:

  1. Patients given intravenous vancomycin prophylaxis.
  2. Patients given another choice of intravenous antibiotic prophylaxis.
  3. Patients given no intravenous prophylaxis—either oral cephalosporin or no antibiotic at all.

Did vancomycin side effects appear in the results?

  • Kidney function: no damage
  • Liver function: no damage
  • We also checked for Clostridium difficile infections (CDI), which is often caused by antibiotics, and we actually saw a decrease.

These results obviously suggest that vancomycin is reasonably safe for patients with AML. But we still asked ourselves if there is some other explanation for these findings. Take the kidney function tests, for example. We know vancomycin prevented infections. And infections, by causing sepsis, can result in damage to the kidneys.

We wondered if vancomycin could be causing damage to the kidneys in one patient, while preventing it in another, making it look like both patients had about the same result. So we looked at patients who had developed sepsis and used that medical problem as a mediator or connector in our analysis. More confusingly, the chemotherapy itself can also cause kidney and liver problems, so we needed to take that into account as well.

That is where biostatisticians Yilun Sun, MS, and Li Tang, PhD, helped us untangle the different strains of evidence. Yilun, the lead author of our report, ran a series of advanced mathematical operations called generalized linear mixed models that essentially reduced our complex problem to a series of yes-or-no questions. He enabled us to demonstrate that vancomycin not only prevented sepsis, but it did not cause kidney or liver damage, even when sepsis was considered.

This study not only reinforces that what we’re currently doing is safe, it also gives us the impetus to study the proactive use of other antibiotics.

You can read the full article at Adverse effects of intravenous vancomycin-based prophylaxis during therapy for pediatric acute myeloid leukemia. Antimicrob Agents Chemother. March 2018 vol. 62 no. 3 e01838-17.

About the author

Josh Wolf, PhD, MBBS, is a faculty member and clinical investigator of the Infectious Diseases Department and Graduate School of Biomedical Sciences at St. Jude Children’s Research Hospital. View full bio.

More Articles From Josh Wolf

Related Posts

dali-like illustration of clocks in a wasteland
dali-like illustration of clocks in a wasteland

The ticking ‘epigenetic’ clock measures true biological age

Two people in white coats examining a slide.
Two people in white coats examining a slide.

Influenza and gene therapy researchers team up to create more effective flu vaccine

Like kisMET: targeted therapy and radiation synergize in pediatric high-grade glioma

Stay ahead of the curve