Assessing onset, prevalence and survival in mice using a frailty phenotype

Introduction

Frailty is an age-associated biological syndrome characterized by an increased vulnerability to adverse global health outcomes, a reduced capacity to react to stressors and an overall loss in physiological function [1,2]. Consequently, frail individuals are at a greater risk of falls, dependency, disability, institutionalization, hospitalization and mortality [3,4]. Two of the most commonly used clinical frailty assessments are the phenotype model and the frailty index. The frailty phenotype, proposed by Fried et al. [4], focuses on five specific criteria for unsuccessful physiological aging: unintentional weight loss, weakness, poor endurance/exhaustion, slowness, and low activity. On the other hand, the frailty index developed by Rockwood and colleagues [5,6] conceptualizes frailty using a more multidimensional approach, and is based on a comprehensive geriatric assessment that measures deficits in physiological, psychological, cognitive and social function. Using these approaches approximately 5-10% of community-dwelling individuals 65 years or older are considered frail with the prevalence of frailty increasing with age [1,7,8], highlighting the importance of frailty research.

To date, unanswered questions in the field of frailty are largely due to the inherent limitations of clinical studies, including the ethical, logistical, and biological complications of working with humans, in particular older individuals [9]. With the recent development of mouse models of frailty, reverse translated from the human frailty models, some of these unanswered questions are currently being investigated. Liu et al. [10] developed a frailty phenotype in aged C57BL/6 mice that matched the clinical criteria used by Fried et al. [4], and included grip strength, walking speed, physical activity, and endurance that were evaluated by the inverted-cling grip test, rotarod, voluntary wheel running, and an endurance score obtained from the grip test and rotarod. Similarly, Parks et al. [11] established a frailty index based on accumulated deficits derived from Rockwood and colleagues [5,6], in which aging C57BL/6 mice were assessed using 31 items that included both invasive and noninvasive variables (i.e., activity levels, hemodynamic measures, body composition and basic metabolic status). This group later developed a more simplified version of their 31-item frailty index that only used readily apparent noninvasive signs of clinical deterioration [12]. The development of animal models for frailty represents an important step forward in how we understand frailty, and interventions that may delay or prevent its progression [9].

Although these mouse models are integral to frailty research, they do have some limitations and can therefore be “fine-tuned” to further advance the field. For instance, the original [11] and modified [12] frailty indexes do not include important functional assessments such as walking speed and endurance; physiological variables known to decrease in aging rodents and humans [13,14]. The frailty phenotype established by Liu et al. [10] attempted to match the clinical criteria used in Fried’s human model [4], but did not include a weight factor, lacked an established endurance test and only assessed a small cohort of mice at a single age. However, probably the most important factor missing from these published frailty studies is the use of a phenotypic approach with a longitudinal component that explores important characteristics of frailty, including the onset and prevalence of frailty, and overall healthspan and mortality.

Therefore, the overall goals of this study were first to identify the onset and the prevalence of frailty across the lifespan of C57BL/6 male mice. Second, to determine if a frailty phenotype predicts mortality. To accomplish these goals, we assessed body weight, body fat percentage, walking speed, strength, endurance and physical activity in male C57BL/6 mice longitudinally, in which the same cohort of mice were tested across their entire lifespan.

Results

Discussion

Frailty is a clinical syndrome defined as a late-life vulnerability to adverse health outcomes that ultimately leads to death. The development of animal models for frailty represents an important step forward in how we understand and treat frailty. Here, we expanded on a published mouse frailty phenotype [10] that was adopted from Fried et al. [4] that utilizes a multifaceted approach, incorporating body weight, weakness, poor endurance, slowness and low physical activity. However, our phenotype differed in that we selected a high body weight, rather than unintentional weight loss as a positive marker of frailty. Using our frailty phenotype in a longitudinal study design, we were able to define the onset and prevalence of frailty, and accurately predict mortality. We propose frailty phenotypes, such as the one presented here, are minimal effort, cost effective tools that can be used to determine biomarkers, mechanisms and potential treatments for frailty.

Using a mouse frailty phenotype focused on physiological function it is possible to identify two characteristics of frailty, the onset and the prevalence of frailty across the lifespan. In the present study, the onset of frailty occurred at 17 months when there was a 100% survival rate of our cohort. The prevalence of frailty increased across the lifespan, with nearly every mouse being identified as frail by 32 months (52% survival rate) (Figure 3A). These ages in mice represent human years of ~60 and over [13,16]. Our data are not consistent with a recent report, demonstrating no change in the prevalence of frailty across the lifespan in female ICR/CD1 mice when using the Valencia Score frailty phenotype [17]. Yet, the overall prevalence of frailty increases as humans age, with 5-10% in 60-69 years old to 26-65% in those aged ≥85 years of age [1,7,8]. The increasing prevalence of frailty in both humans and rodents with age suggests common etiologies underlying the frailty syndrome. Interestingly, the degree of prevalence in the very old is different between research performed in the laboratory with rodents and in clinically orientated human research. There could be many underlying reasons for this difference including factors specific to human frailty, such as socioeconomic status and the race/ethnicity of the population. Because these factors can be controlled in rodent studies the assessment of frailty in preclinical models becomes very important. In contrast to the reported prevalence of frailty, to our knowledge the age at which frailty is first observed has not yet been reported in male C57BL/6 mice. The value of identifying the onset of frailty is important in teasing out the underlying etiologies of frailty; however, the identification of this critical time point requires a longitudinal research design, which at times, is prohibitive.

As expected, we report a strong association between mouse frailty status at 23 months of age and overall mortality. Using the five criteria (Table 1), we were able to accurately predict that frail and pre-fail mice die before non-frail mice (Figure 3B). In fact, every frail mouse died within five months (28 months of age), while the majority of the pre-frail mice died in eight months (31 months of age). These finding are evident when comparing the lifespan of frail, pre-frail and non-frail mice; non-frail mice (33.8 months) lived longer than mice identified as pre-frail (29.5 months) and frail (27.1 months). Similar associations have been reported using an accumulation deficit frailty assessment in mice, rats and humans [18–20], and using the Valencia Score in female ICR/CD1 mice [17]. Overall, these findings emphasize the negative outcomes of the frailty status across species.

Early identification of individuals at risk of developing frailty, even at a pre-frail status, has clinical, social and economic benefits due to the burden this syndrome poses on society. For instance, from a quality of life perspective there is evidence attesting that frailty and pre-frailty can be prevented or delayed, by the incorporation of timely and appropriate interventions [21]. The frailty phenotype utilized in the present study focused largely on physiological function, thereby incorporating several tissue/organ systems (e.g., neuromuscular, cardiovascular, pulmonary). It is possible the cellular and molecular processes that maintain these systems were compromised and progressively deteriorated in the frail/pre-frail mice. The increased prevalence of frailty with aging likely represents decreased resilience to ongoing stress, which ultimately could lead to extensive cellular alterations. Some of these alterations include genomic instability, mutations, altered gene expression, loss of cell division potential, cell death and/or impaired intercellular communication [22]. Hence, identifying the onset and prevalence of frailty across the lifespan, in addition to predicting mortality, has the potential to yield information about susceptibility and resilience to the aging processes. Further investigations will be needed to determine which cellular and molecular processes are compromised in frail animals/humans, and if attenuating these compromised processes influence healthspan and/or mortality.

An important and novel component of our frailty model is the criterion of a high body weight (Table 1), which does not follow unintentional weight loss outlined by Fried et al. [4]. Others have applied similar weight loss criterion to mouse models of frailty, for instance Gomez-Cabrera et al. [23] considered mice that lost more than 5% of their weight at 17 months of age to have a positive marker of frailty. However, we would like to emphasize three key concepts. First, our data do indeed support that in a group of healthy mice, body weight steadily decreases from 23 to 32 months of age (p<0.01, Figure 5A). Second, by the mean survival age of our mouse cohort (i.e., 32 months), many of the mice that died before this point were actually the heaviest mice earlier in life (Figure 4A). Lastly, even though the heavy mice died early, it does not necessarily imply these mice are impervious to the weight loss criterion established in human [4] and mouse [23] frailty phenotypes. Rather, it is possible the heavy mice that were identified as frail in this study lost weight soon before dying (i.e., days to weeks), but it was not captured due to the time between assessments. We would also add that it would be advantageous to implement interventions before a significant amount of weight is lost, as death would likely be inevitable at that point.

From this, we determined that in order to accurately predict frailty, one must consider body weight of the mice at all ages. Mice that weighed in the top 20% of our cohort at 23 months were therefore considered to have a positive marker of frailty. Although we are the first to suggest that high body weight be considered a criterion to predict mouse frailty, others have suggested body weight negatively impacts longevity across several mammalian species [24]. In mice, the most basic examples include the Ames dwarf that has superior longevity [25,26] and calorie restriction, in which reducing body weight increases the mean and maximum lifespan [27,28]. Moreover, in a series of 15 mouse stocks that were selected over 22 generations for their different rate of body weight, body weights at 3, 6 and 12 months were significant predictors of longevity (among stocks) [29]. Most relevant to the present study was work done by Miller et al. [30], who reported a significant association between body weight and mortality in mice 2 to 24 months of age. In line with this, our data suggest that frail/pre-frail mice are heavier than non-frail mice throughout adulthood (Figure 4A). Collectively, these data strongly suggest that high body weight early in life negatively impacts healthspan and longevity.

Mice with high body weights also possessed the most body fat (Figure 6A-6F), indicating the “extra weight” was not coming from increased muscle mass. Indeed, the relationship between body weight and body fat has been observed across various species, including humans [31–33]. Furthermore, epidemiological observations provide solid evidence that mortality is associated with excessive body fat (i.e., obesity) [34–36]. Miller et al. [30] suggested that some of the genetic variations that control the speed of body growth in young mammals also lead to alterations in the stress resistance of many cell types. Moreover, these changes in cellular stress resistance endure through most or all of the lifespan, thus influencing the timing and severity of multiple forms of late‐life, potentially fatal, illnesses [30]. This hypothesis aligns with the concept frail mice lack resilience to ongoing stress, in that frail mice may exhibit deficits across various tissue/organ systems, which can be indirectly measured using a phenotypic approach. For instance, our data demonstrate frail/pre-frail mice were heavier, possessed more body fat and were less physically fit across most of their adult lives when compared to non-frail mice (Figure 4A-4C).

Using tests of physiological function it is possible to indirectly assess the health of many tissue/organ systems. Moreover, using our longitudinal study design, we were able to evaluate these systems in the same mouse over a lifetime. Age-related deficits were recorded in walking speed, strength, endurance, and physical activity (Figure 7A-7D), similar to that observed by others using cross-sectional research designs [14,23]. Importantly, our functional variables followed that of the physical phenotype established by Liu et al. [10], with the addition of an established endurance test, time to fatigue on a motorized treadmill. The frailty phenotype presented here is unique when compared to deficits models that mainly use clinical signs of frailty (e.g., fur color, hearing loss, vision loss, loss of whiskers, etc.) [11,12]. In that, our phenotypic model is objective, it functionally challenges the mouse (i.e., maximal treadmill running) and depicts that frail/pre-frail are dysfunctional early in life. In fact, mice identified as frail/pre-frail at 23 months were typically slower and less aerobically fit earlier in life when compared to non-frail mice (Figure 4C & 4E). Taken together, a clear relationship exists with frail/pre-frail mice being unhealthy early in life, as assessed by body weight, obesity and physiological function.

Although our frailty phenotype accurately predicts mortality, future studies will be needed to extend upon these findings. For instance, C57BL/6 male mice were utilized, and although this strain is widely used, other mouse strains may need to be tested for frailty at different time points, depending on their individual lifespans. Moreover, studies will need to assess if frailty in female mice is similar to males, due to declining levels of estrogen that occur with ovarian senescence. Despite continued research, we believe reduced physical function and obesity early in life could be used as “universal” frailty criteria for all strains and across genders.

The frailty phenotype utilized in the present study focused largely on body weight and physiological function, thereby incorporating several tissue/organ systems. Our data demonstrate frail mice were obese and physically unfit throughout the majority of their adult lives, and had a reduced lifespan. Using a frailty phenotypic approach, we were able to identify that the onset of frailty begins the second half of life and steadily increases with age. We suggest that frail mice lack resilience to the aging process, in that multiple tissue/organ systems deteriorate at an accelerated rate compared to non-frail, healthy mice, which ultimately leads to early mortality. In closing, identifying the onset and prevalence of frailty across the lifespan, in addition to predicting mortality, has potential to yield information about susceptibility and resilience to the aging processes.

Materials and Methods