Assessing Drug-Smoking Interactions Using PBPK Modeling

Assessing Drug-Smoking Interactions Using PBPK Modeling

The prevalence of cigarette smoking remains high globally despite abundant evidence showing that it isn’t good for your health. But, did you know that smoking can affect the metabolism of other drugs and even cause serious drug interactions? In this blog post, I’ll discuss the mechanisms by which smoking can impact pharmacokinetics, how physiologically-based pharmacokinetic (PBPK) models can be used to predict these changes, and a case study that illustrates how a client leveraged this technology.

How can cigarette smoke alter the metabolism of other drugs?

Polycyclic aromatic hydrocarbons are some of the major carcinogens found in cigarette smoke. These compounds also induce the liver enzyme cytochrome P450 1A2 (CYP1A2) in a dose-dependent manner. Heavy smoking has been shown to dramatically increase the clearance of drugs metabolized by that pathway, such as caffeine.1 Even exposure to secondhand cigarette smoke (“passive smoking”) has been shown to significantly increase the clearance of the CYP1A2 substrate, theophylline.2 Because of differences in the clearance of CYP1A2 substrates, smokers and passive smokers may require higher drug doses to attain a pharmacological effect that is similar to that of non-smokers.

If smoking is so bad, why do smokers often feel worse after quitting?

When people stop smoking, the induction of CYP1A2 ceases. The decreased levels of CYP1A2 may cause significantly increased plasma concentrations and toxicity in substrate drugs. For drugs with narrow therapeutic indices, smoking cessation could be a serious therapeutic concern. For example, caffeine is almost entirely metabolized by CYP1A2; its clearance is more than 50% higher in smokers compared to non-smokers.3 Thus, median caffeine concentrations are also often two- to threefold higher in non-smokers than smokers. When a patient quits smoking, they should reduce their caffeine intake by half to avoid caffeine toxicity. In fact, the symptoms of caffeine toxicity are similar to that of nicotine withdrawal, often making smokers feel even worse when trying to quit.

Drug safety concerns in a patient population with a high prevalence of smokers

We worked with a global pharmaceutical company that was developing a new anti-schizophrenia drug whose elimination was mediated almost exclusively by CYP1A2.  Schizophrenics are often heavy smokers. With smoking prohibited in hospitals in most countries, our client was concerned about potential increases in drug exposure and ensuing toxicity should a patient need hospitalization.

Given the potential for abrupt changes in smoking habits brought about by hospitalization, the sponsor sought to understand whether it was possible to keep patients’ exposures within the therapeutic window through appropriate clinical management. In particular, they wanted to evaluate whether sudden smoking cessation, and the resulting loss of CYP1A2 induction, might cause the drug’s plasma concentration to spike and increase the risk of an adverse drug reaction (ADR). Excess risk of an ADR would stop development of the drug candidate. Evaluating this potential go/no go decision through additional clinical testing would cost several million dollars and extend development by one to two years. This client came to us seeking a model-based solution.

Using PBPK to inform the risk of drug-smoking interactions

The client worked with my consulting colleagues to better understand likely exposure levels of their drug in smokers, especially following abrupt smoking cessation. They developed a physiologically-based pharmacokinetic (PBPK) model showing the effect of smoking levels on CYP1A2-mediated drug clearance based on published data for two drugs metabolized through that pathway. Simulations using the model reliably predicted clearance values and steady-state and trough plasma concentrations consistent with published observations. The results successfully predicted exposure in smokers and non-smokers, as well as smokers before and after they stop smoking. They then worked with this client to perform dose simulations to predict steady-state and trough concentrations of the candidate drug in smokers before and after smoking cessation. The simulated plasma concentrations were compared to a range of possible ADR thresholds.

The PBPK modeling and simulations showed that while there was a small likelihood of ADRs in a small percentage of patients, this would be clinically manageable. Evaluating this go/no go decision using PBPK modeling and simulation took a few weeks, saving as much as two years and 1–2 million dollars had a conventional clinical evaluation been performed.


References

[1] Tantcheva-Poór I, Zaigler M, Rietbrock S, Fuhr U. Estimation of cytochrome P-450 CYP1A2 activity in 863 healthy Caucasians using a saliva-based caffeine test. Pharmacogenetics. 1999 Apr;9(2):131-44. PMID: 10376760.

[2] Chetty M, Cain T, Rose RH, Jamei M, Rostami-Hodjegan A. A systems approach to predicting differences in pharmacological response to a CYP1A2 substrate, resulting from pharmacokinetic differences in non-smokers, passive smokers and heavy smokers. Presented at the PAGE Meeting. 5-8 June, 2012, Venice, Italy.

[3] Zevin S, Benowitz NL. Drug interactions with tobacco smoking. an update. Clin Pharmacokinet. 1999;36(6):425-438.


Struggling to optimize dosing for a drug to treat a CNS disorder?

Read our case study to learn how a medium-sized bio-pharmaceutical company leveraged Certara’s trial simulation software and PBPK modeling and simulation consulting services to facilitate regulatory approval for a long-acting injectable anti-psychotic drug. Let me know what you think in the comments section!

Theresa Cain

About the Author

Theresa Cain has a degree in Mathematics and an MSc in Medical Statistics. After working for a year as a Statistician at AstraZeneca (Alderley Park, UK), she then joined the Department of Mathematics and Statistics at the University of Sheffield to work on a PhD. The project included some collaboration with the School of Health and Related Research (ScHARR) and involved the Bayesian analysis of discrete choice data to investigate uncertainty in health-state utilities used in economic evaluations. At Certara, she has worked on parameter estimation, pharmacodynamics, SIVA, and the design of clinical studies.