How precision drug-dosing supports individualized treatment

The concept of precision drug-dosing has gained ground in recent years, given its ability to boost efficacy and curb side effects. Yet imprecise dosing regimens continue to be common for many drugs, leading to significant rates of adverse drug reactions (ADRs). 

“ADRs are one of the top ten causes of death in the developed world,” said Sirj Goswami, CEO and co-founder of InsightRX. “More than two million serious ADRs occur each year, representing a cost burden of $136 billion annually. More than half of these events are preventable and are dose-related,” Goswami said. 

In the following interview, Goswami shares his perspective on how precision dosing can optimize dosing in clinical trials and improve real-world drug performance. He also touches on the promise of machine learning to drive further dosing-related advances. 

Drug Discovery & Development: How do you think the pandemic has impacted interest in precision medicine?

Goswami: We have heard from health systems that because of the pandemic, they have deprioritized their precision medicine initiatives in favor of patient safety measures. The pandemic has also raised the old question of whether individualizing treatment based on genetic profiles has as much impact as broader population health initiatives (as reflected in this Wall Street Journal article).

We define precision medicine as using patient data to select appropriate targeted therapies and identify the optimal dosage of medications to maximize efficacy and minimize toxicity. Precision medicine thus incorporates complementary approaches with both broad and targeted benefits to patients. 

With that definition in mind, health systems have been very open to the precision dosing aspect of precision medicine. One reason for this may be that precision dosing can already be applied to improve outcomes across a broad range of medications that are in widespread use today. And it can do that without the need to implement additional testing or clinical trial protocols and without running into prior authorization and other cost-containment measures introduced by payers for some of the newer individualized therapies.

Similarly, from the pharmaceutical industry’s point of view, precision dosing is an aspect of precision medicine that can be applied across a drugmaker’s portfolio. Including a precision dosing component early in trials can derisk a drug candidate by improving efficacy and reducing toxicity, regardless of whether the candidate is a genetically targeted or individualized therapy. Not only is the ultimate chance of approval potentially increased, but approved medications are more likely to work as expected when dosing is individualized to a patient. 

Drug Discovery & Development: The InsightRX website describes the platform as a clinical decision support tool. How can biopharma companies use the InsightRX platform? 

Goswami: Biopharma uses the InsightRX platform to individualize drug dosing in clinical trials and identify optimal dosing strategies for new medications. In this scenario, providers are using our clinical decision support tool as part of a clinical protocol.  

The benefits of dose optimization during a clinical trial include:

• Less patient dropout due to toxicity or lack of efficacy.
• The opportunity for understanding individual dose-response relationships during the trial.
• The ability to identify true non-responders to therapy.

One example is dose individualization of the conditioning therapy for gene-therapy trials. Biotechnology companies that are developing gene therapies require personalization of the conditioning therapy administered to a patient before a cellular transplant to improve engraftment success, avoid preventable toxicities and reduce patient drop-off. With the InsightRX platform, biotech companies apply model-informed precision dosing to individualize busulfan dosing and other commonly administered conditioning agents.

Finally, our real-time analytics platform, InsightRX Apollo, can also be used to monitor adverse events and analyze pharmacological data during a clinical trial. The analytics platform can be integrated with electronic data capture systems. It can be used by pharma companies for Phase 1 dose-escalation trials to accurately and efficiently guide the dose-finding process throughout the study.  

Drug Discovery & Development: Could you provide some context for how precision dosing drugs can improve safety and efficacy? 

Goswami: Drugs like vancomycin have what is called a “narrow therapeutic window,” meaning that there is a very tight range of dosages where drug exposure is high enough to be efficacious but low enough to avoid toxicity. In addition, there may be a lot of patient variability, making it riskier to use “one-size-fits-all” dosing. In the case of vancomycin, acute kidney injury is a serious form of toxicity that can substantially increase hospital length of stay and costs.

Precision dosing uses pharmacological models to predict a patient’s response to a given dose based on their individual characteristics and update those predictions as more data become available. In the vancomycin example, our platform will incorporate vancomycin blood levels and serum creatinine levels into each patient’s individual predictions such that the predictions become ever more individualized and more accurate. These predictions allow clinicians to modify dosage regimens to keep the drug levels within a desired target range that maximizes efficacy while minimizing toxicity. 

Drug Discovery & Development: Can your platform also help fine-tune the timing of drug administration as well as dosing? 

Goswami: Yes, in addition to optimizing the dose, the InsightRX Nova platform can also help clinicians determine the optimal dose administration times and dosing intervals (i.e., every 12 hours versus every 24 hours) for an individual patient. 

Drug Discovery & Development: How can your platform help clinicians/pharma companies monitor adverse events? 

Goswami: Our real-time analytics platform, InsightRX Apollo, helps both health systems and pharma companies monitor adverse events and other pharmacological and clinical outcomes relevant to the therapy administered. 

For health systems, the Apollo platform is used by clinicians to monitor and analyze adverse events and drill down and identify additional risk factors linked to the adverse event in question. For vancomycin, Apollo is used to carefully monitor acute kidney injury events and to identify predictors of nephrotoxicity in various patient populations.  

For pharma companies, InsightRX Apollo can also monitor adverse events and analyze pharmacological data for Phase 1 dose-escalation trials to guide the dose-finding process.  

Drug Discovery & Development: Topics such as AI in healthcare have gotten a lot of hype over the years. What sets your continuously learning dosing system apart from the competition?  

Goswami: We apply quantitative pharmacology and machine learning to provide an individualized understanding of a patient’s pharmacology to guide the dose selection process. Through our analytics platform, we also use Big Data from across our customer network to update our pharmacological models continuously, making the platform smarter and more predictive over time.

We believe that no model should be static. Predictive models implemented in a clinical decision support tool should be periodically validated and improved with continued use. Within the context of dosing, this process is critical when dosing vulnerable patient populations such as geriatrics, pediatrics and patients with end-organ dysfunction since pharmacological knowledge in these subpopulations is limited.  

Our learning platform is also free from the “black box” perception often held by the healthcare community regarding AI. The predictive pharmacology models we use to guide dose selection are made transparent to all customers, and we publish our models with open access whenever possible. Furthermore, all model-based dose recommendations have a pharmacological basis, unlike other AI applications where the connection between the input data and the output prediction is inscrutable. Finally, we have a highly trained team of pharmacometricians and data scientists guiding the model-optimization process. The combination of deep domain knowledge with data science expertise sets us apart.