Managing Drug-Drug Interactions in Older Adults

Americans aged ≥65 years totaled 55.7 million (17% of the US population) in 2020 and are projected to reach 94.7 million in 2060.1 Physiological changes associated with age can alter the pharmacokinetics and pharmacodynamics of medications.2 PK changes associated with chronological age occur within the context of other age-associated factors, including chronic health conditions and multimorbidity,3 frailty,4 and polypharmacy.5 Not only does the prevalence of polypharmacy, defined as taking 5 or more medications, increase with age, but the proportion of older adults with polypharmacy has tripled from 1994 to 2014, rising from 13.8% to 42.4%.6 There are sparse data regarding drug–drug interaction (DDI) risks in older adults, as they are generally excluded in clinical trials. Predicting drug exposure and the clinical risk of reaching a threshold exposure associated with a higher risk of an adverse drug event when adding a new medication to a drug regimen becomes challenging to predict. Moreover, less is known about PD changes related to these age-associated factors that may also increase the risk of experiencing clinically relevant DDIs. Potentially serious DDIs are prevalent among older adults, particularly involving drugs acting on the cardiovascular and central nervous systems,7 through altered PDs. Some additional considerations in evaluating the risk for clinically relevant DDIs in older adults include multiple prescribers, the potential for infrequent or inadequate monitoring, and impaired pathways of drug elimination and pharmacogenetic patterns.8 Collectively, a critical gap remains regarding how to manage DDIs in older adults in clinical practice. As noted, DDIs may affect drug PKs and/or PDs, and both are clinically important for older adults. The risk of a DDI increases in older adults experiencing polypharmacy. In people aged 75 years and older, as the number of dispensed medications increases, the odds of having a category D (ie, avoid concurrent use) interaction increases exponentially.9 DDIs are particularly prevalent among older adults with multimorbidity, and most of these interactions are PD in nature.10 Interestingly, regulatory guidance for the assessment of PK DDIs are relatively well developed, whereas a framework for assessing PD interactions is lacking and clinical assessments are not performed during drug development.11 Multidrug interactions, or interactions involving more than 2 drugs, may increase the risk of adverse events in older adults as well. Multidrug interactions often involve psychotropic medications, and the risks of sedative and anticholinergic burden to cognition, falls, and functional status are well documented.12, 13 Outcomes associated with many multidrug interactions are more difficult to predict and are not included in the drug labeling because of a lack of data/evidence. Clinical prescribing decisions for older adults with multimorbidity and polypharmacy are complex. The prescriber must weigh the potential benefits against the potential risks of adding a medication, as well as determining the appropriate dose, for an individual considering age, comorbidities, concurrent medications, frailty, vulnerable functional and cognitive status, and other factors. For older adults with multimorbidity, polypharmacy is often already present. A case of a prescribing decision for an older adult with multimorbidity and polypharmacy is presented in Table 1. Screening the medication regimen, including the proposed new medication, using a DDI screening program can identify potential one-to-one interactions, analyze evidence for risk, and provide a severity category. These DDI screening programs generally do not consider dose, only consider the interactions between each 2-drug combination individually, and do not consider patient-specific factors. In the case presented here, the patient is primarily at risk for multidrug PD DDIs. The drug regimen includes multiple medications that act on the central nervous system (CNS). The American Geriatrics Society (AGS) Beers Criteria for Potentially Inappropriate Medication Use in Older Adults includes potentially clinically important DDIs that should be avoided in older adults,14 and indicates that any combination of 3 or more CNS-active medications (antidepressants, antipsychotics, antiepileptics, benzodiazepine, non-benzodiazepine benzodiazepine receptor agonist hypnotics, and opioids) should be avoided because of the increased risk of falls and fractures.15 Adding an opioid analgesic to the drug regimen in this case would be expected to increase the risk of falls and the patient is at particularly high risk of this outcome if they have a history of previous falls. The drug labeling may not reflect current consensus, such as the AGS Beers Criteria, to guide prescribing and DDI evaluation in older adults. Pharmacokinetic drug interactions to considera (https://www.drugs.com/drug_interactions.html). Accessed July 31, 2022 Pharmacodynamic drug Interactions to consider (https://www.drugs.com/drug_interactions.html). Accessed July 31, 2022 The AGS Beers Criteria recommends minimizing the number of CNS-active medications in the drug regimen for older adults. Evidence-based guidelines for safely deprescribing CNS-active medications are emerging,16 but age-specific evidence is sparse. In addition, data to guide deprescribing are rarely available during the initial regulatory evaluation of a new drug, and specific guidance on discontinuing therapy may not be incorporated into the label post-approval for some time as evidence accumulates. There are also health system factors that contribute to the prevalence of adverse events related to clinically significant DDIs in older adults. Uncoordinated healthcare delivery where prescribers may be unaware of medications prescribed by others may result in inadvertent DDIs. Inadequate healthcare provider training, access to information at the point of prescribing, and DDI alert fatigue also play a role. Insufficient systems for identifying adverse events from DDIs can lead to misdiagnosis, unmanaged symptoms, and prescribing cascades. Recognizing and managing clinically relevant DDIs in older adults remains a challenge, particularly for patients with multimorbidity and polypharmacy. Clinical trials often study a more narrowly defined population via selection criteria or limitations in enrolling specific patient groups. As such, 1 important question that clinical pharmacologists are tasked with is to characterize the drug exposure in a target patient population by assessing intrinsic and extrinsic factors that affect drug PKs. The traditional way of characterizing these effects is to run a dedicated clinical trial. In more recent years, modeling and simulation (M&S) tools have been employed to quantify and predict these effects. For example, the available data and established predictive models can be used to predict exposure changes and to inform dosing recommendations in patients with severe organ (renal or hepatic) impairment in cases where exposure changes in mild and moderate disease severity have been evaluated.17 Advances in mechanistic M&S approaches enable us to better understand the mechanisms driving the effect of aging on drug PKs, and these approaches have the potential to predict the effect of older age or frailty on exposure changes, where clinical data are sparse and the need for proposing dosing recommendation is high. The PBPK models encompass the principles of systems biology and constitute a modeling approach that describes the absorption and disposition of drug molecules, as well as their biological and physiological drivers. An increased understanding of these physiological and biological drivers and their interactions with different drug molecules and formulations provide valuable insights into how drugs will behave in healthy volunteers (HVs) and other population groups, such as patients with diseases or extremes of age.18 Considering the progressive biological and physiological changes that accompany the process of aging, together with coexisting diseases and polypharmacy, PBPK models are ideally suited to predict the impact of these variables on the absorption and disposition of drugs in older patients. There are 2 commercially available platforms for the geriatric population: Simcyp and PK-Sim.18, 19 Chetty et al provided a comprehensive summary of the magnitude of the changes in system parameters in the geriatric population compared with HVs within a PBPK model,18 as briefly summarized below. Absorption There is conflicting evidence on changes in gastric emptying and intestinal transit times in older adults. Consequently, these parameters are left unchanged in the geriatric PBPK population. Distribution Changes to organ blood flow are related to the decline in cardiac output with age. There appears to be very little change in plasma albumin or α1-acid glycoprotein (AAG) concentrations with older age based on the observed data. Elimination There is good evidence that liver size decreases with increasing age in the adult population, with the decrease becoming more pronounced after 60 years of age. This is matched by biopsy data showing a reduced number of hepatocytes in livers from geriatric subjects. The liver blood flow is defined in the geriatric PBPK model as a fixed percentage of cardiac output and hence is reduced with age, in agreement with observed data. Various studies using a combination of in vitro and in vivo approaches have investigated age-related changes in cytochrome P450 (CYP) enzymes, with some conflicting results. Older age is thought to have only a modest, if any, effect on the hepatic glucuronidation of certain drugs in vivo. Consequently, the CYP and UDP-glucuronosyltransferase (UGT) abundance values are left unchanged in the geriatric PBPK population. Kidney aging is associated with an increasing proportion of globally scarred glomeruli, decreasing renal function, and exponentially increasing prevalence of end-stage renal disease (ESRD). The age-related changes in glomerular filtration rate in the geriatric PBPK model are estimated using age-related plasma creatinine data derived from the National Health and Nutrition Examination Survey (NHANES) database. For qualification of the geriatric PBPK population, clinical data on the PKs of CYP and P-glycoprotein substrate drugs, including caffeine, desipramine, digoxin, midazolam, omeprazole, triazolam, and warfarin, in older patients were extracted from the literature. The geriatrics model described above was used to simulate the published studies and the predicted clearance and area under the curve (AUC) were compared with the clinically observed values (pred./obs. ≤ 1.5).18 These results confirm that the observed PK changes in older adults can be explained by decreasing liver size and blood flow rather than changes in specific enzyme expression and activity. Here, we highlighted 2 case examples in the literature applying the PBPK approach to assess the exposure of renally cleared drugs in the geriatric population, especially when combined with renal impairment (Table 2). These case examples hopefully demonstrate that the PBPK platform provides a mechanistic framework to facilitate the extrapolation of drug exposure or DDIs from young HVs to older populations, with or without organ impairment. The prediction of differences in systemic drug exposure or DDIs between younger adults and older adults can be used to inform clinical trial designs and highlight situations where dosage adjustments for older subjects require consideration. PBPK simulations can contribute to a totality of evidence and have the potential to inform drug labeling for the geriatric population. The clinical relevance of these predicted differences in drug exposure with age depends in part on the therapeutic index of the drugs, the severity of the resulting adverse event, and patient factors, such as multimorbidity and frailty status, and is therefore drug and patient specific. The potential for risk is weighed against the potential for benefit in the clinical setting and opportunities to mitigate risk (such as reduced dosage, alternative therapy, and monitoring plans for specific adverse events) are implemented. Although there is relatively more understanding gained by using PBPK models to understand PK-mediated DDIs, PD-mediated DDI changes appear to be more clinically relevant but remain challenging to study. Further studies would be needed to build the PBPK-PD models to better understand the impact of aging on altered DDIs in older adults.20 These PD-mediated DDIs may be particularly relevant for medications acting on the cardiovascular and central nervous systems, based on the evidence documented to date.20, 21 Using real-world evidence to better understand the clinical relevance of DDIs to patient outcomes is an emerging approach to further elucidate the risks and benefits of concurrent therapies and how the balance may shift with age.20 Finally, chronological age may not be the best measure for use in assessing the risk of adverse events resulting from DDIs, considering the increasing heterogeneity of the population with age. Frailty status, multimorbidity, polypharmacy, functional and cognitive status, and function of drug-eliminating organs may be more important than chronological age in determining the risk of adverse effects stemming from DDIs in older adults.20 The assessment of DDIs is routinely conducted during drug development to understand the risk and to inform labeling.24-26 However, DDI assessment for older adults is not routinely performed, leading to knowledge gaps in managing DDIs in older adults. It is not known whether the DDI results obtained in younger adults that may be studied during drug development can be extrapolated to understanding DDIs in older adults. In order to understand whether age can impact DDI outcomes, it is important to understand the effect of age on PKs (such as absorption, metabolism, distribution, and excretion) for PK-based DDIs, or PDs (such as receptor expression or disease state) for PD-based DDIs. If age impacts any of the factors differently between drug pairs, the DDI magnitudes may differ between younger adults and older adults. One recent publication by Stader et al showed that the magnitudes of DDIs did not appear to be impacted by age for DDI scenarios covering a wide range of PK DDI mechanisms, including CYPs, UGTs, and transporters (eg, OATP1B1), based on clinical DDI data and PBPK M&S.27 This finding may allow DDI data obtained in younger adults to be extrapolated to older adults. If we assume that age does not impact the magnitude of DDIs and the exposure–response relationship, DDI-based dosing adjustment recommended for younger adults can be assumed to be the same for older adults. However, age and factors associated with age can impact the PKs and PDs of a drug, which may lead to a different dosing adjustment based on age. For example, the exposure in older adults could be higher than that in younger adults, leading to a higher exposure in the presence of DDIs than in younger adults that may warrant a different dosing adjustment in older adults (Table 1). The impact of age on PDs can lead to a different exposure–response relationship, which may also warrant a different dosing adjustment in older adults from that in younger adults. It is important to understand the impact of age, sex, organ impairment, and other factors on drug PKs via clinical pharmacology studies or population PK approaches during drug development. Mechanistic modeling, including PBPK and clinical data obtained in younger adults, can help predict DDIs in older adults that could take into consideration multiple factors in addition to age. However, limited clinical data in older adults, especially in those aged >85 years, could affect the development and verification of the PBPK models. The US Food and Drug Administration (FDA) has encouraged the enrollment of older adults in clinical trials and has developed many guidance documents related to drug evaluation in older adults.28-31 In 2020, the FDA also held a public workshop on “Roadmap to 2030 for Drug Evaluation in Older Adults.”20 The roadmap makes recommendations on 3 key time frames to collect information for older adults through the drug life cycle: (1) early in development – obtaining clinical pharmacology (PKs, PDs, and DDIs) and disease prevalence data to guide enrollment, dosing, and risk mitigation for older adults in later trials; (2) pivotal clinical trials – achieving the inclusion of representative older adults and the collection of relevant data in efficacy and safety trials to mirror the real-world population; and (3) late phase/post-market – refining, confirming, and continuing the evaluation of safety and the effectiveness in older adults through post-marketing studies/trials and real-world evidence. The FDA's 2006 Physician Labeling Rule has designated a specific Drug Interactions section (Section 7) and other sections where physicians may find relevant information about DDIs, such as Highlights, Dosage and Administration, Contraindications, Warning, and Clinical Pharmacology. Information on drug use in older adults is included in a subsection titled “Geriatric Use” under Section 8, Use in Specific Populations. Specific information on DDIs in older adults and relevant dosing recommendations in the labeling are sparse, mostly because of a lack of evidence/data to support the labeling recommendation. Information on older adults can be collected throughout the drug life cycle, and drug product labeling needs to be updated when new information becomes available. Specifically, the 21 Code of Federal Regulations (CFR) 201.56(a)(2) states that “labeling must be updated when new information becomes available that causes the labeling to become inaccurate, false, or misleading.” The FDA issued an updated draft guidance document on geriatric labeling to promote the consistent placement of relevant information about drug use in geriatric patients that could help the healthcare providers to find the relevant information for decision making.29 Given the heterogeneity of the older adult patient population and the clinical contexts, not all clinically relevant scenarios can be empirically explored during drug development. Modeling approaches may provide an opportunity to elucidate the effect of age on drug PKs, especially when there are multiple influencing factors. The use of these approaches to understand PD DDIs is emerging and holds promise for predicting clinically relevant PD DDIs.11 With the aging population of Americans, more mechanistic studies characterizing the effect of aging on drug transporters and on DDI magnitudes in older adults, including those with comorbidities, are needed. Additionally, more research, particularly to build the evidence base for understanding PD DDIs in older adults is needed. PBPK-PD linked models are also highly desirable to facilitate the assessment of potential PD changes in the older adult population. Increased attention to provider education and communication in health care may also reduce the risk of preventable harm from DDIs in older adults. Collaborations among stakeholders, including researchers, drug developers, regulators, healthcare providers, and patient/caregiver groups, are essential to understand DDIs in older adults and generating data/evidence to support product labeling, which in turn would help healthcare providers to better understand and manage DDIs in older adults (Supporting information). The authors would like to thank Dr Xinning Yang for his valuable comments on the article. The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. A.K. is an employee and shareholder of Certara UK Limited (Simcyp Division), a company that develops and supplies modeling and simulation software and services to the pharmaceutical industry. Z.Z. was supported by the National Institute of General Medical Sciences of the National Institutes of Health under award number R16GM146679. The other authors received no financial support for the research, authorship, and/or publication of this article. The opinions expressed in this article are those of the authors and should not be interpreted as the position of the US Food and Drug Administration or the funding agency. Data sharing not applicable to this article as no datasets were generated or analysed during the current study. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.