By Michael Neely, MD, MSc, FCP
The term “precision medicine” is most commonly associated with targeted therapy selection, or the practice of using biomarkers or genetic data to identify the most effective drug for a specific patient. While targeted therapies have been instrumental for oncology patients in particular, precision medicine encompasses much more than drug selection. The field also includes precision dosing, the process of using multiple patient characteristics to tailor the dosing regimen to the patient, thereby maximizing the concentration-related efficacy of a drug while minimizing the risk of toxicity.
To realize the full promise of precision medicine—the idea that all patients can and should be treated as unique individuals—clinicians must leverage patient data not only to identify the right therapy, but also to determine the precise dose that maximizes the patient’s benefit-to-risk ratio. In the era of value-based care, every facet of precision medicine, from diagnostic technologies to targeted therapies and precision dosing, will play a pivotal role in improving care quality.
The insufficiency of one-size-fits-all drug dosing
Clinicians might like to think that traditional dosing methods—based on monitoring pre-dose trough concentrations—are sufficient for most patients. However, concentrations of a drug vary widely by patient, often by 100% or more, even when sampling is performed within the correct time window. The solution to this variability is model-informed precision dosing (MIPD), which captures prior experience with the drug’s pharmacokinetic behavior in various patient populations to create a model that can be used to optimize dosages for any drug in a single patient.
For many drugs, such as voriconazole or vancomycin, therapeutic drug monitoring has typically revolved around trough concentrations, which function as a surrogate for the area under the curve (AUC) defined by plotting drug concentration changes over time. AUC is a marker of total drug exposure. It is impossible to measure or estimate without multiple blood samples unless one uses MIPD software.
Applying MIPD to a new patient is incredibly powerful, as it recognizes that patients are not “therapeutic islands” divorced from every other patient who has received the drug. When we adopt MIPD, we are not ignoring the learned information about drug behavior in previous patients. Instead, MIPD captures our cumulative experience to create equations that link dosages to concentration levels. The Bayesian approach to pharmacokinetic modeling used in many MIPD algorithms relies on Bayes’ theorem, a mathematical equation that updates the probability of a future event based on the probability of that event in the past combined with new, current data.
Pharmacokinetic models of a particular drug are based on a series of equations that link dosage to concentration in order to estimate the values of parameters such as volume and clearance that describe drug behavior within a particular patient population. When current values are available for a specific patient, clinicians can use a Bayesian pharmacokinetic model to predict how the patient will respond to the drug, developing an a posteriori individualized dosing model.
Model-informed precision dosing in practice
Once a pharmacokinetic model for the individual patient is in place, clinicians can use the patient’s response to an individual dose or dosages—along with relevant covariates such as weight, age, renal function, and other biomarkers—to calculate the projected concentration time profile of the drug. Precision dosing software can test a variety of potential doses by using algorithms to calculate the mean weighted profile in the individual model and comparing that information to the target concentrations. Clinicians can use precision dosing software at the bedside to calculate an optimal dosing regimen, even for patients who have historically been difficult to treat.
As an example, let’s look at a case study. One of my past patients was a 45-year-old woman with HIV who had a long history of medication intolerance. After the patient abandoned several medications due to problematic side effects, she began a fosamprenavir regimen without ritonavir, taking two 700 mg tablets twice daily. The patient complained about daytime fatigue and wanted to take the entire dosage at night. Instead of agreeing to her request—which would risk the development of drug resistance—or changing to yet another medication, our team measured her amprenavir concentration and decided to use precision dosing to identify a better regimen.
The patient’s predicted trough concentration was adequate on her current regimen, but maintaining that concentration with once-a-day dosing would have required a dose almost 30 times the recommended amount. The team ran an alternative regimen through MIPD software and found that, if the patient took one tablet with breakfast and three tablets with dinner (a 10/14-hour schedule), her predicted concentration would be comparable to the current one. Once she was placed on the new regimen, the patient’s fatigue disappeared, her CD4 count rose, and her viral load became undetectable for the first time.
With MIPD, clinicians can accomplish real-world results that they couldn’t otherwise achieve. In this case, we now had a very satisfied patient who was likely to stay on her medication at an effective, safe dosage. As a bonus, the clinic had a much higher probability of retaining that patient, and the pharmaceutical company retained a customer.
The clinical and financial benefit of precision dosing at scale
Insights gleaned from MIPD analysis can be instrumental in the evolution of therapeutic best practices across the industry. For example, by using pharmacokinetic modeling simulations to investigate Bayesian adaptive control for vancomycin, I began to suspect that many patients would likely achieve therapeutic medication levels when treated via AUC monitoring, despite having sub-therapeutic trough concentrations. Therefore, if clinicians were targeting trough concentrations with vancomycin, they were likely overdosing patients and increasing their risk of nephrotoxicity.
Back in 2014, I designed a three-year study to confirm my hypothesis. In year one, we observed standard practice: therapeutic drug monitoring using serum trough level targets of 10–20 mg/L, depending on the severity of illness. In the second year, we used patients’ measured trough data and MIPD to target an AUC to a minimum inhibitory concentration (MIC) ratio (the most accurate measurement of vancomycin activity) of 400 or more. In the final year, we added optimal sampling times to target AUCs.
With MIPD, we were able to collect fewer samples per patient and decrease the number of days patients received vancomycin while in the hospital, as we no longer needed to keep adjusting dosages and checking trough concentrations. Overall, we proved our hypothesis, finding that dosing was therapeutic for 70% of patients based on AUC levels, compared to 19% of patients based exclusively on serum trough levels.
By performing MIPD, we were able to significantly reduce the incidence of nephrotoxicity without compromising outcomes. From a financial perspective, patients who did develop toxicity had an average length of stay of 20 days, more than three times longer than the average (six days) for patients without toxicity, resulting in approximately $146,000 of additional cost per patient.
As a result of this research and other subsequent studies, in 2020 the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, the Pediatric Infectious Diseases Society, and the Society of Infectious Disease Pharmacists revised their cooperative guidelines for vancomycin and now recommend an AUC-based dosing approach.
Across the world, similar population studies using MIPD to optimize treatment are complete or underway, with enormous consequences for how we perform dosing for a wide range of drugs. In Norway, researchers found that using MIPD to determine the best dose for tacrolimus, an anti-rejection medication used to treat patients undergoing kidney, heart, and liver transplants, was preferable to the accepted methodology of adjusting dosages based on physicians’ clinical expertise. In addition to using 28% fewer drug samples, the best-dose methodology was therapeutic for 90% of patients, compared to 78% for traditional methods. Best-dose patients also benefited from better glycemic control and glomerular filtration rates.
As precision dosing gathers momentum, we will undoubtedly begin to practice MIPD for medications in a variety of therapeutic areas. In the area of infectious diseases, beta-lactam antibiotics are an ideal candidate for MIPD research, particularly for critically ill patient populations. Another area ripe for exploration is immunosuppressive and anti-rejection drugs, which are used to treat cancer in transplant patients as well as rheumatology and gastrointestinal autoimmune diseases, such as Crohn’s disease.
Barriers to widespread implementation
Despite growing interest in precision medicine as a whole, the implementation of bedside MIPD continues to challenge many organizations. The rise of precision dosing software that integrates with the electronic medical record has made MIPD much easier and less time-consuming for clinicians, but there are still significant barriers to widespread adoption.
One challenge is the immediate availability of assays. Without on-site laboratories, organizations cannot achieve the rapid turnarounds necessary to measure drug concentrations in time-critical patients, such as those in the ICU. A broader challenge is the need to revamp our drug development and regulatory approval process to focus less on finding the correct dosages and more on defining exposure levels that are linked to safety and efficacy. Revising package inserts to be more user-friendly for clinicians practicing MIPD will go a long way toward legitimizing precision dosing as the new standard of therapeutic monitoring.
The accuracy of precision dosing can transform not only patient outcomes and care quality, but also an organization’s financial performance. In the next decade, precision-dosing advocates will help usher in a new era of personalized medicine for every individual.
Dr. Michael Neely is a professor of pediatrics at the University of Southern California, chief of the Division of Infectious Diseases at Children’s Hospital Los Angeles (CHLA), and director of CHLA’s Laboratory of Applied Pharmacokinetics and Bioinformatics.