Integrating Frailty and Cognitive Phenotypes: Why, How, Now What?

Abstract

Purpose of Review

This review elucidates the concept of frailty in relationship to reserve and resilience, the relationships and shared pathophysiology between physical frailty and cognitive impairment, the theoretical underpinnings of three integrated phenotypes of physical and cognitive impairments, and the potential of incorporating biomarkers into phenotype refinement and validation.

Recent Findings

The fact that frailty and cognitive impairment are associated and often coexist in older adults has led to the popular view of expanding the definition of frailty to include cognitive impairment. However, there is great variability in approaches to and assumptions regarding the integrated phenotypes of physical frailty and cognitive impairment.

Summary

The development of integrated frailty and cognitive phenotypes should explicate the types of frailty and cognitive impairment they intend to capture and prioritize the incorporation of biological theories that help determine shared and distinct pathways in the progression to physical and cognitive impairments.

Introduction

One of the biggest healthcare challenges worldwide is the medical and economic burden of caring for dependent older adults ravaged by physical and cognitive impairments. Two of the most common geriatric conditions, frailty and cognitive impairment, are known to predict poor health outcomes [1, 2]. An estimated 15% of the older US non-nursing home population aged 65+ is frail [3], 22% of older US adults aged 70+ have mild cognitive impairment [4] and 14% have dementia [5]. Frailty and cognitive impairments often coexist. One study found that 53% of the frail had cognitive impairment [6]. Many of the aging processes underlying frailty may also impact brain aging and cognitive decline [7], but the mechanisms behind the association are unknown.

With increased interest in the frailty syndrome and its relationship with brain aging, recent consensus papers have suggested expanding the definition of frailty to include cognition [8, 9]. Approximately 50% of the frailty instruments in the literature include a measure of cognition [10], although the type of measure varies across studies from global measures (e.g., MMSE [11]) to clinical diagnosis (e.g., dementia [12]). Such inclusion primarily aims to improve the predictive accuracy of frailty for future adverse outcomes [11]. Meanwhile, increasing epidemiological evidence suggests that cognition is separable from physical functioning [13], supported by the finding that 22% of people with Alzheimer’s disease (AD) had no physical indicators of frailty [14]. We therefore advocate caution when creating integrated phenotypes by combining physical and cognitive impairments that may or may not share common etiologies and pathways.

This chapter seeks to clarify the concept of frailty in relationship to reserve and resilience, review bidirectional relationships and shared pathophysiology between physical frailty and cognitive impairment, and elucidate the importance of developing and using integrated phenotypes of physical and cognitive impairments for their intended purposes. To this end, the chapter is organized into five sections. Section 1 reviews the theoretical conceptualization of frailty and its closely related concepts of reserve and resilience. Section 2 provides a brief summary on the bidirectional relationships between frailty and cognitive dysfunction. Section 3 discusses pathophysiological pathways linking frailty and cognitive impairment, focusing on the stress-response systems. Section 4 offers examples of integrated physical and cognitive phenotypes that vary in suppositions about common pathophysiologic pathways. These phenotypes include: frailty index [15, 16], motoric cognitive risk syndrome [17], and cognitive frailty [18]. Section 5 outlines future directions, followed by concluding remarks.

Reserve, Resilience, and Frailty

Frailty is theoretically defined as a clinically recognizable state of increased vulnerability resulting from aging-associated declines in reserve and function across multiple physiologic systems, such that one’s ability to withstand, compensate and overcome stressors is compromised [1]. We further theorize that the frailty state manifests as a medical syndrome, operationally defined by the physical frailty phenotype consisting of muscle weakness, weight loss, slow gait, exhaustion and low activity [19]. Frailty is a clinical entity that both overlaps and is distinct from disability and multimorbidity [20]. Frailty could be viewed as an emergent phenomenon defined as an emergent aggregation of multiple frailty manifestations caused by depletion of reserve and compensatory mechanisms, such that any new deficit leads to failure of the whole organism [21]. As depicted by the schematic diagram in the Figure, we hypothesize that the development of frailty is the result of complex interactions among reserve, resiliency, and type/magnitude of the stressor. Specifically, the progression of frailty may consist of a series of transitions between states of dynamic equilibrium of decreasing integrity before reaching a critical threshold beyond which frailty emerges.

Figure 1.

Hypothesized evolution of degradation of physiologic integrity underlying the progression of frailty. Physiologic integrity of the system is defined by the maintenance of equilibrium in the face of stressors. The level of integrity is theorized to be a function of physiological reserves represented by both the depth of each convex curve and the degree of its curvature, with greater depth and curvature representing greater reserve. Both episodic (e.g. stroke, fall) and chronic insults (e.g. chronic inflammation) are hypothesized to degrade return to equilibrium and decrease integrity of the system to respond to a subsequent stressor. Progression of frailty consists of a series of downward transitions between states of equilibrium of decreasing integrity; and at some critical tipping point, the system becomes overwhelmed and can no longer harness the resources needed to maintain integrity, leading to frailty.

Before addressing connections between frailty and cognition in the following sections, we first aim to crystalize the concepts of vulnerability, reserve and resiliency as they relate to frailty. In the context of frailty, we define vulnerability as lack of ability to resist functional impairment in the presence of intrinsic (e.g., disease pathology) or extrinsic (e.g., air pollution) insults. We posit that vulnerability results from insufficient reserves that impede individual’s ability to withstand stressors, where reserves denote “the means that are available to recover from idiosyncratic adverse events, stressors or nonnormative transitory periods during the life course [22].” In other words, chronic depletion of reserves creates vulnerability.

Studies of physical health have largely confined the discussion of reserve to organ systems or physiological systems. For example, cardiac reserve is an important indicator of “the potential capacity of the heart to function well beyond its basal level, in response to alterations in physiological demands [23].” Its application has since extended beyond diagnosis of coronary artery disease to include determination of post-surgical risks [24]. Recent work has linked energy reserve, another example that is essential to the study of aging processes, to the varying degree of gait slowing with age – a strong and robust predictor of disability and mortality [25, 26]. It is hypothesized that this slowing is the body’s natural adaptive response to shrinking energy reserve as a greater proportion of energy is recruited to combat disease pathologies or declining efficiency in energy utilization [27]. The all-encompassing concept, of “physiologic reserve” is used to characterize “the potential capacity of a cell, tissue, or organ system to function beyond its basal level in response to alterations in physiologic demands [28].”

Closely related to reserve is the concept of resilience, a construct most commonly studied in the field of neuropsychology to characterize individual’s capacity to cope, adapt, or even thrive in the face of adversity in life such as a tragedy, family estrangement, and sudden health problems [29]. The term resilience has extended beyond psychological domain to include emotional, social, physiological, and physical resilience. For example, physical resilience has been defined as “the ability to recover or optimize function in the face of age-related losses or disease,” and may influence or be influenced by other types of resilience [30]. If vulnerability is about susceptibility to stressors and reserve refers to the resource-dependent capacity to cope with stressors, the focus of resilience is on the state of the recovery process itself following a stressor.

Upon reviewing these closely related concepts, we report evidence that vulnerability, reserve, and resilience have permeated the thinking in frailty-related research without making direct reference to them. For example, in the field of immunology, the concept of immunological reserves pertains to “the ability of the immune system to respond to a challenge with greatly increased activity compared to its state of relative rest [31].” Adequate innate and adaptive immunity may be viewed as indicators of immunological reserves. Interestingly, one difference between innate vs. adaptive immunity bears strong resemblance to the concept of brain reserve vs. cognitive reserve [32]: both innate immunity and brain reserve are essentially hardwired, that is - passively acquired through genetic factors and modified by early-life experiences; whereas adaptive immunity, like cognitive reserve, is dynamic and actively acquired throughout the life course. Frailty has been associated with dysregulated innate [33] and adaptive immunity [34], which could explain the blunted antibody response to influenza vaccination among frail older adults, therefore rendering them more susceptible to influenza and its complications [35]. The relationship between frailty and resilience is exemplified by the data showing that glucose and insulin responses to glucose challenge were more exaggerated and prolonged in frail versus nonfrail or prefrail women [36]. At the level of whole-body function, epidemiological data have linked frailty to poor recovery from incident disability [37] or hospitalization-related functional decline [38]. Taken together, these findings support the applicability of the reserve-resilience paradigm in frailty research. Progress made in studies of reserve and resilience in neurocognitive science may therefore facilitate cross-fertilization between disciplines that is necessary to guide future studies of frailty.

Bidirectional relationships between cognitive function and physical frailty

In order to better understand the development of frailty and possible hierarchies and intersecting pathways with cognitive impairments, researchers have studied relationships between frailty (by the physical frailty phenotype [39]) and cognition – both global cognition and domain-specific cognitive status [40]. Cross-sectional studies have reported significant associations between greater prevalence of frailty and increased prevalence of poor global cognition in older adults [40–45]. However, several studies found non-significant associations between global cognition and pre-frailty [43, 46] or frailty [47]. A recent longitudinal study reported that non-frail, cognitively-impaired older adults at baseline had a significantly higher risk for pre-frailty/frailty onset over 4 years when compared to non-frail, cognitively intact participants [48]. In addition, older adults with baseline frailty showed significantly greater risk for onset or worsening of global cognitive impairment over 2–3 years when compared to those without baseline frailty [44, 49]. These results are consistent with previous research, including studies where baseline frailty was associated with incident dementia [7].

Studies of community-dwelling older adults have also observed associations between the physical frailty phenotype and specific cognitive domains, including perceptual-motor function, language, learning and memory, executive function, and complex attention [41, 42, 45, 50]. However, only certain domains – perceptual motor function, and specific measures of language and memory – were consistently associated with frailty across studies. Longitudinal studies of physical frailty and cognitive domains have shown varied associations. In a longitudinal study of processing speed, executive function, and immediate-and-delayed word recall among 331 older women, executive function (measured by Trail Making Test B) was the only cognitive domain whose annual rate of change over time was significantly associated with risk of frailty onset [51]. Another longitudinal study reported that baseline frailty status was not associated with linear change over time in cognitive domains of processing speed, verbal fluency, reaction time/variability, even after adjusting for possible dementia [52]. Overall, the paucity of longitudinal studies on frailty and domains of cognitive function is a notable research gap. We recommend additional longitudinal studies of the natural progression of incident cognitive and physical impairments to improve our understanding of the bi-directional relationships between frailty and cognition.

Physiological Pathways Linking Physical Frailty and Cognitive Impairment

Given that impaired stress response is a hallmark of frailty [53], and cellular energetics is central to stress adaption, we hypothesize that decline in mitochondrial function with age may be the initiator of a pathophysiological cascade, contributing to the organism’s maladaptive multisystemic response to stress through dysregulated energetics, excessive oxidative stress, and overstimulated inflammatory response. Consistent with this hypothesis, a recent study showed that mice with mitochondrial dysfunctions, compared to wild type, had concurrently altered neuroendocrine, inflammatory, metabolic, and transcriptional responses when encountering acute psychological stress [54]. Studies of humans have also generated evidence linking mitochondrial function to three stress-response systems: the innate immune system [55], the hypothalamic-pituitary-adrenal (HPA) axis [56], and the sympathetic nervous system [57, 58]. Mounting evidence implicates these stress-response systems in the pathogenesis of frailty [59]. Chronic activation of these systems and the mutual exacerbation through interactive feedback loops are posited to be the primary driver of the downward spiral of degradation of systems integrity, which at a later stage engenders clinical manifestations and the overt frailty syndrome [53]. Dysregulation of the same stress-response systems has also been associated with decline in cognitive function [60–62]. Of these systems, the innate immune system, and chronic, low-grade inflammation (also termed inflamm-aging [63]) in particular has attracted a great deal of attention in studies of biology of aging. Chronic inflammation marked by both overproduction of pro-inflammatory cytokines (e.g., interleukin-6) and dysregulated balance between pro-inflammatory and anti-inflammatory cytokines (e.g. interleukin-10) may be a key mechanism linking physical frailty and cognitive impairment [7, 64].

Examples of Integrated Physical and Cognitive Phenotypes

There has been a recent surge in the creation of combined physical and cognitive phenotypes likely motivated by the desire to improve: predictive utility [11], early risk detection by exploiting their shared biological and physiological underpinnings [17], or diagnostic accuracy via pathway-specific subtyping of a clinical entity with multifaceted etiology such as cognitive impairment [18]. Having reviewed the shared physiology and bidirectional relationships between frailty and cognitive impairment, we introduce three examples of integrated phenotypes and discuss the scientific rationale, the intended clinical contexts and purposes of use.

Frailty Index

A risk index combines multiple related or unrelated risk factors into a unidimensional measure that is agnostic with regard to the temporal order and underlying etiology of physical and cognitive impairments. A corresponding overall risk score is created by summing the individual scores assigned to the risk factors in the composite according to a set of consistent rules, with a higher risk score indicating a higher level of risk. For example, the Frailty Index (also termed Deficit Accumulation Index) operationalizes frailty by counting the number of deficits accumulated over time, including disability, diseases, physical and cognitive impairments, psychosocial risk factors, and geriatric syndromes (e.g. falls, delirium, and urinary incontinence) [15, 16]. The resulting integrated phenotype constitutes a comprehensive risk index and is viewed as an estimate of biological age [65]. The Frailty Index has proven useful for risk stratification in the general population [66], as well as predicting postoperative outcomes in surgical populations [67]. Its strength in aggregating risk across outcomes poses challenges to identifying and targeting underlying physiologic processes in preclinical and early clinical stages of frailty.

Motoric Cognitive Risk Syndrome

In comparison to the frailty index, Verghese and colleagues coined the term “Motoric Cognitive Risk (MCR) Syndrome” [17] to represent “a transitional state between normal aging and dementia” [68]. The MCR is operationally defined as the presence of both objective slow gait and subjective cognitive complaints in the absence of dementia and difficulty with activities of daily living [17]. The specific criteria used to define slow gait and cognitive complaints however varied across studies [68–70]. This combined phenotype was motivated by the hypothesis of shared brain substrates and pathologies between gait and cognition [71], and the finding that gait slowing preceded decline in objective cognitive performance and subsequent diagnosis of Mild Cognitive Impairment (MCI) a decade later [72]. MCR was shown to be predictive of incident cognitive impairment [69] and dementia onset [17, 69]. In addition, MCR was found to independently predict disability [73], falls [70], and mortality [74]. These findings coupled with the ease, affordability, and noninvasiveness of its assessment make MCR an appealing early marker of dementia applicable to both research and clinical settings.

There are a couple of points worth noting. First, MCR as an early sign of cognitive decline preceding MCI is built on the premise that shared pathophysiology underlying decline in motor function and cognitive function manifests hierarchically over time with slowing gait occurring first. However, given that slow gait may result from non-neurological as well as neurological diseases [75], gait-related immobility may also be causally related to dementia independent of neuropathology, for example, via sedentariness, which has been linked to increased risk of dementia [76], possibly through vascular pathways. The latter could help explain findings that the association between MCR and dementia remained significant after accounting for MCI [69]. The distinction between a marker and a cause has important implications in intervention selection. Interventions targeting gait impairment without addressing the shared pathology would have no impact on dementia if slow gait were merely a risk marker. Second, given the finding of a stronger tie between MCR and vascular dementia [17], it begs the question of whether the relationship between MCR and dementia is specific to certain types of dementia and whether parameters of gait performance other than velocity may also play a role [68].

Cognitive frailty

The third integrated phenotype differs from the preceding two examples by making physical frailty the core phenotype to which cognitive impairment is added to increase predictive utility. Cognitive frailty is a relatively new expansion upon frailty, representing a premorbid cognitive state caused by physical frailty rather than neurodegenerative disorders [18]. The operational definition includes the concurrent presence of physical frailty and cognitive impairment in the absence of clinical diagnosis of dementia [18]. Given the multifaceted etiology of cognitive impairment and the lack of effective treatment strategies beyond short-term symptomatic relief for many neurodegenerative diseases such as AD, it is important to identify a subset of a heterogeneous population for whom preventive or rehabilitative interventions may hold the greatest potential. By focusing on a specific cause of cognitive impairment, i.e., physical frailty, this construct enables a more targeted approach to delay or ameliorate cognitive decline through, for example, behavioral interventions addressing negative consequences of frailty including sedentariness and social isolation. Recent studies have shown that cognitive frailty, or the addition of global cognitive impairment to physical frailty was associated with risks of incident dementia [77–79], functional disability [80, 81], poor quality of life [81], and mortality [78, 81].

Overall, these three phenotypes combine physical and cognitive risk but systematically vary in the focus of interest in and articulation of common neurobiological pathways that may explain progression and inform treatment options.

Future Directions

Twenty-five years of research on frailty across multiple epidemiologic and clinical patient studies demonstrates the value and utility of the concept. However, a critical question facing the field of frailty research remains: whether frailty is ready for widespread clinical use? According to a recent editorial, the answer is no, arguing that while “frailty remains a powerful predictor of patient-centered outcomes but is not yet ready for a role as a full-fledged outcome measure in geriatrics research [82].” Here we propose that addressing this question depends on answering to the fundamental question of what is frailty, and how knowing this will help us advance to answer the more important question of – What we can do about frailty? In our view, knowing that frailty, variously defined, is a strong predictor of adverse outcomes brings little comfort to those whose primary goal is to develop intervention strategies targeting frailty progression. Similarly, combining different types of frailty across physical, cognitive, psychological, social, and environmental domains may be counterproductive or even misleading if the intended goal for such use is not clearly specified [83]. The lasting controversy surrounding the measurement of frailty is rooted in the fact that the public as well as the academics often interpret the word “frailty” in the most generalized sense as vulnerability of all kinds. Overemphasis on predictive validity has also fed into the pursuit of power of prediction without appropriate consideration of construct validity, i.e., the degree to which an index or scale measures what it purports to be measuring [84]. Therefore, we conclude by highlighting three areas of research that are urgently needed in order to bridge the gap between frailty theory and frailty measurement and guide the development and validation of integrated frailty and cognitive phenotypes.

Theory

Although reserve and resilience have been integral parts of the theoretical definition of frailty [85], the relationships between reserve, resilience, and frailty are not well defined. Beginning with reserve and resilience, we find it useful to view the distinction between the two analogousto the difference between hardware and software. For example, in neurocognitive science, reserve is indicative of innate capacity present early in life and is thought to be a function of both brain reserve measured by anatomic parameters (e.g., brain size, neuronal count) and neuronal reserve (i.e., the efficiency of task-related cognitive processing) [32]. As such, reserve can be viewed as the “hardware” in the absence of brain pathology. In contrast, resilience has to do with how the brain processes cognitive tasks and maintain function in the face of brain pathology; therefore, resilience is more in line with the concept of “software”. If frailty is a clinical state of reduced resilience, stress-response experiments arguably give us the best chance to identify such state [86]; and to operationally define frailty, including dynamic measures of the body’s response to a stressor such as glucose challenge is critical to diagnostic accuracy. The challenge then becomes how to select dynamic measures that can capture multisystem physiological dysregulation, which is at the core of frailty biology [53]. On the other hand, if frailty is meant to denote a threshold of reserve, or tipping point (see Figure) below which the ability to mount an effective response to a stressor is greatly diminished, high priority should be placed on identifying that threshold over time, regardless of stressors, in the context of frailty and developing and validating measures accordingly.

Measurement

In order to meaningfully advance frailty measurement, it is critical to distinguish primary from secondary frailty [87, 88]. We consider primary frailty a unique clinical entity in itself and its underlying pathophysiology is separable from other disease-specific processes. By comparison, secondary frailty is clinically in conjunction with signs and symptoms of a preexisting disease (e.g., congestive heart failure) or a direct consequence of the preexisting disease or an acute health event (e.g., hip fracture). The distinction is important for several reasons. First, despite the emerging consensus that frailty is distinct from disability and comorbidity [89], many frailty instruments, such as the Frailty Index, include comorbidity as a key assessment component thereby combining primary and secondary frailty [10]. Second, with declining reserve and resilience being hallmarks of frailty, the validation of frailty requires that observed differences in functional trajectories between the frail and the non-frail are more than a reflection of severity of disease-specific pathology. This is in line with the work on reserves in neuroscience in attempt to explain the mismatch between brain pathology and its clinical manifestations, as evidenced by empirical data including the work by Esiri et al. [90] showing that about 25% of older adults with normal neuropsychological testing scores prior to death meet full pathologic criteria for AD. In the absence of direct measures of reserve, quantifying and accounting for disease pathology become necessary for ascertaining the true meaning and impact of frailty in the presence of comorbidity.

Natural Progression of Frailty and Cognitive Impairment

The cognitive frailty phenotype represents an ongoing effort to understand the heterogeneity in the pathogenesis of cognitive impairment by focusing specifically on MCI caused by physical frailty in the absence of neurodegenerative pathology. However, the stated causal relationship underlying this new clinical entity has not been validated and operational definitions of cognitive frailty so far are exclusively based on cross-sectional data [91]. In order to improve specificity in diagnosis, studies of natural progression of physical frailty and cognitive impairment are urgently needed. Only then can we begin to identify biological and clinical markers that distinguish patterns of coevolution, which may generate insights into the mechanistic pathway(s) implicated in cognitive frailty. A recent study from a US nationally representative sample of 7,439 community-dwelling older adults examined the temporal ordering in frailty and cognitive impairment onset, and found that participants with incident dementia during the 5-year follow-up were at increased risk of developing cognitive impairment first, or frailty and cognitive impairment concurrently. In contrast, dementia onset was associated with reduced risk of physical frailty onset before cognitive impairment [92]. These findings suggest that dementia-related pathology is less likely to be the cause of cognitive impairment if preceded by physical frailty, therefore providing support for the current definition of “cognitive frailty.”

Biomarkers

The term “biomarker” here refers to “objective indications of medical state observed from outside the patient – which can be measured accurately and reproducibly [93].” Biomarkers can help doctors evaluate a patient’s susceptibility to a disease, diagnose a disorder and its severity, predict its likely progression and future outcomes, determine optimal intervention strategies and monitor response to treatment. In the field of preclinical AD detection, cognitive testing remains the primary means to measure preclinical-to-clinical progression to AD; but the last 10 years has seen a considerable increase in the use of biomarkers to predict and track pre-clinical cognitive declines [94]. For example, structural imaging of hippocampal regions related to memory has proved useful in tracking transitions from healthy to MCI to AD. Blood biomarkers (e.g., lipids, inflammatory markers, and hormone levels) are also frequently collected as possible explanatory variables that inform cognitive declines [95–102] and merit similar concurrent study for their utility in predicting transitions to frailty.

Frailty research on biomarkers has focused more on disease-nonspecific mechanisms of the biology of aging. As in the field of AD detection, interest in frailty-related biomarker discovery is expected to intensify rapidly. In addition to disease staging as in the case of detection of AD-related cognitive decline, biomarkers can be helpful for distinguishing etiologies and pathways underlying physical frailty and cognitive impairment. For example, in order to improve diagnostic specificity of cognitive frailty, biomarkers of brain Aβ plaque deposition (e.g., cerebrospinal fluid Aβ42), apolipoprotein Eε4 genotype, as well as advanced neuroimaging techniques can serve to exclude neurodegenerative disease-related pathologies in case-finding of cognitive frailty [18]. While biomarker research holds great promise for a wide range of research and clinical applications, biomarker assays can be cost-prohibitive for large-scale epidemiological surveys and of limited value for population screening. In the case of cognitive frailty, a cost-effective alternative is through proactive health monitoring of older adults as they become frail and flag them at the first sign of cognitive impairment following frailty onset; biomarkers of neurodegenerative pathologies can then be used for case confirmation. From a data collection perspective, as the new generation of electronic medical record systems is increasingly being implemented to allow integrated record keeping and individual-level tracking, monitoring of health change over time is now an attainable goal in the near future.

Conclusions

In this chapter, we summarized the current variability in approaches to and assumptions regarding the integrated relationships between physical frailty and cognitive impairment, the definitions and potential roles of reserve and resilience in buffering or precipitating physical and cognitive impairments (e.g., Figure), and the importance of developing measures that are specific in the types of frailty (e.g., primary vs. secondary) and cognitive impairment (e.g., global vs. domain-specific) they intend to capture. We have prioritized the incorporation of biological theories that help determine shared and distinct pathways in the progression to physical and cognitive impairments (e.g., inflammation) and associated biomarkers that may then be incorporated into phenotype refinement and validation. Importance of bridging the gap between frailty theory and frailty measurement, such that science concerning the biology of frailty can be meaningfully advanced and preventive and treatment strategies can be developed and tested, persists.