FormalPara Key Points

1. New methods in pre-clinical models have allowed for measurement of exposures (such as chronic exposure, polypharmacy and deprescribing) and outcomes (such as health span, functional measures and frailty) that are highly relevant to geriatric pharmacotherapy.

2. These pre-clinical models may better predict drug effects in geriatric patients, in whom medication use is highly prevalent and who are vulnerable to adverse effects.

3. While they are currently costly and time-consuming, there is potential for these pre-clinical models to improve the clinical translation of new drugs and the outcomes of pharmacotherapy in older adults.

1 Why are Pre-clinical Models Needed for Geriatric Pharmacotherapy?

With ageing of the population worldwide comes increasing use of medications by older adults.

Older adults are the most frequent users of multiple medications (polypharmacy) and are at high risk of adverse drug events [1,2,3] . Application of pre-clinical models (including cell culture, animal models and in silico models) that are more relevant to older adults with polypharmacy and multimorbidity will increase the translation of drug development to clinical benefit for older adults in clinical trials and clinical practice.

The increased use of medications imposes great challenges for our healthcare systems [4] . At the individual level, the challenge is to individualise treatment to balance risks related to drug treatment without denying older people valuable drug therapy. At the societal level, the challenge is to reduce drug-related ill health, hospitalisations and associated costs [5] .

Drug treatment in old age is complicated by concomitant use of multiple medications (polypharmacy) [1] . Older individuals often have multiple diseases and impairments (e.g. cardiovascular disease, musculoskeletal disease, psychiatric disorders and cognitive impairment) [6] and therefore use multiple drugs. There is also continuous development of new drugs for age-related diseases. Polypharmacy is very common in old age; about half of the population aged ≥ 65 years is exposed (usually defined as concurrent use of ≥ 5 medications) [7, 8]. The net effects of polypharmacy are at present impossible to foresee in an individual patient. There is limited evidence of the effects of polypharmacy from randomised clinical trials (RCTs), beyond occasional post hoc subgroup analyses [9]. Polypharmacy can lead to unforeseen effects and drug–drug interactions [10], particularly in the context of age-related worsening of renal and hepatic function. [11] This can result in severe adverse outcomes, such as falls [12] and cognitive [10] and functional decline [13]. These are sometimes misattributed to primary disease and can lead to a prescribing cascade [14]. Polypharmacy has increased over time [15, 16] and imposes considerable challenges for the individual patient, the prescriber and for the society at large.

Adverse drug events contribute to up to 30% of hospital admissions for older people [2] , and they entail immense costs for healthcare systems [5] . Older patients are more susceptible to adverse effects of medications than younger adults [3] . Ageing leads to changes in pharmacokinetics and pharmacodynamics, which result in prolonged and increased effects of many drugs. Individual variation in drug response and side effects is large and difficult to predict [2, 17].

Despite this complexity, the majority of knowledge about medications is from monotherapy RCTs in younger and healthier adults [18], which is of limited applicability to older patients [19, 20]. RCTs are performed in a controlled environment, and typically exclude patients at advanced ages with polypharmacy, frailty and multimorbidity [19, 20]. Exclusion criteria for most RCTs include multiple diseases and polypharmacy, which in practice excludes many older patients. RCTs also typically investigate one drug for one disease at a time, which rarely reflects the complexity of drug treatment in old age. This together complicates the translation of RCT results to the treatment of older patients [19].

There are, however, international efforts to increase the representativeness of participants in clinical trials to match the intended real-world users of the medications across a range of characteristics including age, multimorbidity, polypharmacy and frailty [21]. A central concept in ageing research is frailty because it strongly relates to advanced age and adverse health among older adults [22]. Frailty should thus be assessed in RCTs despite the inherent difficulties [21]. Medication agencies and other regulatory bodies should make efforts to increase numbers of frail older adults in RCTs. Post-marketing and epidemiological studies also need to include data on frailty to be able to provide meaningful information from the real-world setting [21]. Furthermore, RCTs need to focus on outcomes that have the highest value to older adults, such as function and independence [21].

Currently, much of the data on medication-related harm in older adults are from epidemiological and pharmacovigilance studies. These real-world studies based on older people with multimorbidity, polypharmacy and frailty will continue to be important for geriatric pharmacotherapy [23]. However, these studies are often limited by small or selected samples as well as bias and other methodological shortcomings. The main challenge is to account for confounding by multiple indications (confounding by multimorbidity and frailty) [24], which means separating the potentially negative effect of polypharmacy from the effect of multiple diseases [15]. There is a need for translational research where large epidemiological data and pre-clinical models in concert help to inform about the underlying mechanisms and effects of pharmacotherapy in old age [25].

A vulnerable geriatric group is people living with dementia, who are particularly susceptible to adverse effects of drugs [26]. The pathological brain changes causal to dementia (e.g. a degenerative process such as Alzheimer’s disease or micro- or macro-vascular compromise as an end-organ effect of cardiovascular disease) lead to increased susceptibility to many drugs [27]. Cognitive deficits and communication problems further complicate the drug treatment and detection of side effects. However, few large studies have assessed the quality of prescribing in people with dementia [26]. Patients with dementia are typically excluded from RCTs and informed consent can be difficult to obtain. These patients are, however, likely medication users in the real-world setting [28]. Even RCTs of anti-dementia drugs do not adequately reflect the real-world setting of patients with dementia. Participants in these RCTs are often younger, with less comorbidities, than real-world patients with dementia [29, 30]. To overcome these challenges, translational research including both human and pre-clinical models of dementia can be informative where evidence would otherwise be difficult to obtain. Furthermore, measures of drug-induced cognitive impairment are important outcomes in pre-clinical and clinical research.

The great complexity of geriatric pharmacotherapy renders great need for precision medicine [25, 31]. Clinicians need tools for improved prediction of expected clinical effects and risk of side effects in individual older patients [25, 32]. For a given patient, the challenge is to individualise treatment to minimise risks of drug treatment without denying older people valuable drug therapy. Artificial intelligence and machine learning have the potential to handle the complexity and variability in data from older adults. This new technology could help create these highly needed tools for optimising and tailoring drug treatment for the individual patient [21]. These techniques can also use real-world data to improve design of RCT eligibility criteria for representative recruitment. They can combine pre-clinical, RCT and real-world data for predictive analysis of adverse drug events and drug–drug interactions [21, 33].

2 Could Pre-clinical Models Provide More Translatable Evidence for Geriatric Pharmacotherapy?

Preclinical testing in the presence of geriatric changes could help detect the effects of ageing on drug toxicity, as well as on pharmacokinetics and efficacy. This information could inform successful translation to the treatment of older adults.

Current guidance for pre-clinical testing of new drugs deliberately excludes evaluation in old age. The Organisation for Economic Co-operation and Development (OECD) guideline for chronic toxicity studies recommends a duration of exposure that is long enough to allow any effects of cumulative toxicity to become manifest, without the confounding effects of geriatric changes [34]. Generally, these studies are performed with chronic oral dosing in rodents for 12 months, which is approximately equivalent to 30 human years for a rat and 40 human years for a mouse. While this drug exposure models the duration of many chronic therapies well, clinically, most chronic therapies are given in the presence of geriatric changes in the second half of the lifespan, not the first half.

New methods and models provide opportunities to test drugs preclinically in more clinically relevant models. Medications that are likely to be used by significant numbers of older adults should be tested in relevant pre-clinical models to investigate age-related pharmacokinetic and pharmacodynamic effects, analogous to the calls for representative recruitment to clinical trials [35]. This is particularly important for drugs that are hypothesised to have a positive or negative impact on health span on in old age, affecting physical and cognitive function and frailty. Here we highlight exposure characteristics, pre-clinical models and outcomes that reflect clinical practice and clinical outcomes, which should produce pre-clinical data more translatable to older patients (Fig. 1).

Fig. 1
figure 1

Pre-clinical models for geriatric pharmacotherapy

Full size image

2.1 Exposures

Exposure to medications in pre-clinical models should reflect the context of which they are used in humans for the conditions of interest to increase clinical translation. This includes the exposure dose, duration and exposure age (or onset of morbidity). For example, people are usually diagnosed with heart failure with preserved ejection fraction in old age and then commence chronic treatment. Therefore, pre-clinical models should begin exposure in old age and test exposure over periods equivalent to the years of clinical treatment. Furthermore, drugs should be tested in the combinations in which they are used, including polypharmacy for multimorbidity. This information can be derived from cluster analyses of population data [36]. Preclinical models should consider these variables to develop clinically relevant models.

It is also important to evaluate the effects of stopping chronic drug treatment, which is common practice in geriatric medicine patients when the ongoing benefits of treatment no longer outweigh the risks, known as deprescribing [37]. In the past decade, mouse models of polypharmacy and deprescribing have been developed [38,39,40,41,42,43,44,45]. These models demonstrated that old animals are more susceptible to polypharmacy-induced harm, chronic polypharmacy and chronic monotherapy in old age can cause physical and cognitive decline, deprescribing can reverse some outcomes and sex influences outcomes. Future studies are required to explore mechanisms of these effects and investigate different drug regimens and the impact of the multiple morbidities for which the drugs are used. Pre-clinical studies can inform deprescribing practice by investigating adverse drug withdrawal effects from fast or gradual withdrawal, including the risk of return of the underlying condition. These models of polypharmacy and deprescribing could be adapted to pre-clinical drug evaluation to ensure that the exposure matches real-world exposures of older people.

2.2 Models

In pre-clinical drug development, the age ranges of the models should reflect the human conditions of interest and the ages of people likely to use the drug clinically. Pre-clinical models range from cell culture through to animals. Bioprinting provides new opportunities to test drugs in models of tissues that are structurally and functionally accurate [46], with potential to replicate the physiology of ageing and the pathologies of multimorbidity.

2.2.1 Cell culture

Cell culture is applied to conduct high throughput screening of drug effects [47]. In ageing research, cell culture commonly involves primary culture of cells from aged animals. Whilst this provides opportunities to understand the effects of drugs on individual cells and for conducting high throughput screening for drug effects on age, it cannot test the dynamic interplay occurring in the body in response to drugs. Translatable clinical factors, including physical function, disease, pain, withdrawal and tolerance, cannot be evaluated in cell lines.

2.2.2 Animal models

Therefore, animal models are commonly employed to measure complex multi-system effects on translatable clinical outcomes. Ageing animals are the most common pre-clinical model used to understand ageing and the effects of interventions in and on ageing, mainly through longitudinal studies or by studying animals of different age groups. Well-characterised animal models of ageing range from yeast to rodents to larger animals and non-human primates, which offer different insights and translatability to humans [48]. Animal models of common age-related diseases are variably translatable to disease in older adults, which often has complex multifactorial pathogenesis. For example, a review of models for vascular cognitive impairment induced through brain lesions, reproducing risk factors (including ageing) or genetic mutations in rodents and larger species, found that no model replicated all pathologic and cognitive aspects of human disease, and a deep understanding of each model could guide selection for different experiments and interpretation of results for translation [49].

Ageing is associated with multimorbidity. Typically, drugs are evaluated in young animals with a single disease for simplicity [50]. Ageing animals develop age-related multimorbidity, with increasing prevalence of conditions from osteoarthritis to renal impairment and/or cancer, depending on the species studied. Some pre-clinical models induce morbidities and comorbidities relevant to the drug being developed, informed by co-morbidity cluster analyses in populations of older adults. This approach will ensure that we understand drug-disease interactions, which are highly relevant to multimorbid older patients. However, to date there are limited studies on multimorbidity due to the complexity of the models, labour intensity, significantly increased cost and need for increased sample size to deal with variability. Examples of comorbid pre-clinical studies are described by Shabir et al. [51]. Work is needed to overcome these challenges to ensure that preclinical findings are applicable to older people.

Rodents are one of the most common species used to investigate ageing and pharmacology due to their relatively short lifespan and similarities to humans. Mice are commonly investigated between the ages of 3–6 months for young mature adults (20–30 years human age), 10–14 months for middle age (38–47 years human age) and 18–24 months old age (56–69 years human years) [52]. Interventions have variable effects in different mouse strains. There are roles for studies in well-characterised inbred strains to understand mechanisms of drug effects and potentially reduce sample size in the absence of genetic variability, as well as in outbred strains to improve generalisability [53].

Investigating ageing in rodents is a lengthy task and premature ageing models have been employed to reduce experiment time [54]. These animals commonly display phenotypes in multiple organ systems that suggest premature ageing and resemble features of natural ageing. These models provide insight into the molecular mechanisms involved but generally represent rare premature ageing conditions in humans, such as Hutchinson–Gilford progeria, Werner syndrome and Cockayne syndrome, limiting the generalisability of the studies.

Other species and organisms have been investigated in studies of ageing and pharmacology. Some, such as yeast, fruit fly, fish and roundworm are simpler to conduct high throughput testing, but they are limited by being simpler organisms [55]. The obvious advantages of using these models include that they are short-lived (compared with humans), offer access to comprehensive resources such as known genetic and transcriptomic data and have available experimental manipulation capabilities and extensive husbandry experience [56]. However, as these species are different to humans and may have different evolutionary toolboxes, this can lead to interpretation biases and inappropriate or false interpretations. In contrast, non-human primates, which share > 92% homology with humans, have been used in ageing research and occasionally in drug development [56]. However, their substantial size, cost, long lifespan and stringent ethics requirements limit the use of this model for drug evaluation unless scientifically essential.

2.3 Outcomes

To facilitate translation of drug evaluation from the bench to bedside of geriatric patients, assessment of outcomes relevant to older people are necessary. Over the past decade, pre-clinical models of frailty have been developed on the basis of the two main models used in humans: the frailty index and the frailty phenotype, which can be applied to drug evaluation [57]. As in clinical trials [21], frailty in pre-clinical models can be applied at baseline to determine whether drug effects vary with frailty [58], and as a clinically important outcome measure. The toolbox for longitudinal assessment of health span in ageing mice was proposed in 2020 [59], which consists of clinically relevant measurements of function of several vital systems such as the cardiovascular (echocardiography), cognitive (novel object recognition), neuromuscular (grip strength, rotarod) and metabolic (glucose tolerance test and insulin tolerance test, body composition and energy expenditure) health. Novel technologies, such as automated behavioural cages, now allow us to explore natural laboratory animal activity, analogous to continuous virtual monitoring in clinical trials [44]. Additionally, advances in machine learning have allowed for the possibility of machine-vision-based analysis of recordings of animal behaviour, opening up opportunities for big data analysis, including the application of this to mouse gait, posture, grooming activity and predicting frailty based on morphometric and gait features [60,61,62]. These techniques can all be applied to pre-clinical evaluation of the effects of new drugs on clinically important outcomes in older people.

3 How Could these Pre-clinical Models for Geriatric Pharmacotherapy be Validated and Implemented?

Implementation of pre-clinical models for assessment of geriatric therapeutics would need validation of the models, proof of concept studies, comprehensive analysis of potential benefits and harms, provision of incentives and regulatory change.

As discussed above, pre-clinical models for ageing, multi-morbidity, polypharmacy and deprescribing have been validated. Models of important geriatric outcomes, such as frailty and health span, have been established. However, there are many ways to consider appropriate models and exposures and to measure each of these outcomes. There is a need to evaluate and select the best models, exposures and outcomes for testing drug effects in old age. For example, different drugs and polypharmacy regimens have different impacts on the mouse frailty phenotype and the mouse clinical frailty index [63]. In mice, as in humans [21], it is not yet clear which frailty measure is most relevant to drug evaluation. Models may also need to be tailored to specific situations to capture the inter-individual variability of older adults, making it difficult to find a consistent approach for testing. For example, different regimens of background polypharmacy are relevant to evaluation of drugs treating different conditions. Proof of concept studies are needed to comprehensively evaluate a range of therapeutic drugs that have already been translated to human therapeutics. Such studies would determine the best pre-clinical models and measures to predict outcomes that were subsequently observed in clinical trial and real-world data from older adults.

Pre-clinical evaluation in models that are more relevant to geriatric therapeutics has potential benefits and harms. The potential benefits are identification of any efficacy or safety issues in old age prior to clinical trials. The benefits of using more translatable models may be less scientific waste because pre-clinical findings more reliably predict the subsequent clinical trial outcomes. Pre-clinical evaluation in models that are more applicable to geriatric patients will increase confidence that drugs are likely to be efficacious or safe in this population, which may facilitate recruitment of older adults to clinical trials. If efficacy or safety issues are detected in pre-clinical models of old age, then they can be investigated further to determine whether the drug should be tested in older adults. This will enable personalised medicine and reduce post-marketing warnings and withdrawals for drugs that are found to be unsafe when used by older adults in practice. The harms of pre-clinical evaluation in these models are primarily cost and time. It is expensive and time consuming to test chronic exposures of drugs in aged animals [50]. Scientific challenges include cost pressure, less availability of historical data (compared with young adult mice) to guide scientists, logistical issues for animal husbandry and stricter labour-intensive animal welfare checks that are necessary for aged animals. The models themselves are labour intensive, such as measurement of health span or frailty and long-term administration of drugs as chronic monotherapy or with background polypharmacy. The increased inter-individual variability in old age may require larger sample sizes to evaluate drug effects. All these factors need to be considered in estimating the cost-effectiveness of using more clinically relevant models to evaluate drugs for geriatric pharmacotherapy.

In addition, in silico models developed from combinations of data from molecular data, preclinical longitudinal studies and available clinical data will enable accurate high-throughput models to predict the effects of drugs in geriatric patients. There is a need to compute this knowledge in a form that can be efficiently translated to clinical practice [64]. Currently, in silico methods have been proposed as a strategy to accelerate the performance of clinical trials targeting human ageing [64] and to identify potential drugs that modulate the ageing process [65]. Future research should consider polypharmacy within these models, as this is the context geroprotective agents will be applied in in most older adults. Research into in silico models for ageing, polypharmacy, multimorbidity and frailty could provide insights for geriatric pharmacotherapy practice.

Implementation of more clinically relevant pre-clinical evaluation of geriatric therapeutics will require incentives. These could include additional funding to test aged or frail animals, analogous to the National Institutes of Health (NIH) funding provided to encourage testing of male and female animals. Ultimately, if these models are found to be useful in improving the pipeline for efficacious and safe drug use in older adults, the efficiency generated may be enough to drive changes in pre-clinical drug evaluation. To ensure consistency of approach, regulatory change to pre-clinical evaluation requirements will be needed. This could include models that are representative of the populations that will use the drug, in terms of age, sex, exposure types and durations and morbidities, as well as measurement of important outcomes for older adults such as frailty and health span.

4 Conclusions

Now, early in the United Nations Decade of Healthy Ageing, there are great opportunities to improve pre-clinical drug evaluation to ensure that new drugs help older people to participate in and contribute to their communities and society. Pre-clinical models for drug evaluation that are more translatable to the older adults, who are major users of medications, can improve effectiveness and safety of medications and inform precision medicine in the geriatric population. The extent to which these pre-clinical studies will need to be performed in vivo versus in silico is an exciting future frontier.