We aimed to identify objectively assessed biomarkers that are commonly used in population-based studies and applicable in a range of settings (that is, not limited to use in a laboratory/clinic setting), capable of distinguishing between healthy and unhealthy ageing between individuals at older ages, and which change within individuals over time. Where possible, we sought evidence of replication of the proposed marker in different cohorts and using different study designs. The research base in some domains, for example measures of age-related immune function proved to be less well developed than in others, for example measures of physical capability, so that our recommendations in the former domains are more tentative. To help fill the remaining gaps, we also aimed to identify priorities for further research on biomarkers of healthy ageing and these are summarised below in the sections headed ‘Areas lacking adequate evidence’.
The process used to develop recommendations included: 1) undertaking comprehensive reviews of the literature relevant to each domain using, where available, existing systematic reviews, meta-analyses and other authoritative reports such as the recently launched NIH Toolbox for assessment of neurological and behavioural function, which includes test batteries for cognitive and motor function (the latter described here as physical capability) [14]; 2) we invited international experts to comment on our draft recommendations; and 3) we hosted an experts workshop in Newcastle, UK, on 22–23 October 2012, aiming to help capture the state-of-the-art in this complex area and to provide an opportunity for the wider ageing research community to critique the proposed panel of biomarkers (Fig. 1). In this report we also highlight areas needing further research.
This work has been developed at the request of the Medical Research Council (MRC) to address this gap and a version of the report, including details of the evidence used in their derivation, can be found on the MRC website as a resource for the community [17].
Biomarkers of physical capability
Measures of physical capability, that is, a person’s ability to perform the physical tasks of everyday living, are useful markers of current and future health [18]. Guided by previous work by the Healthy Ageing across the Life Course (HALCyon) research collaboration [3] and the NIH Toolbox, we selected four subdomains: locomotor function; strength; balance; and dexterity (Fig. 2 and Additional file 1: Table S1). Physical capability declines progressively in later life with men performing better than women at all ages [19]. Poor performance in tests of grip strength, walking speed, chair rise time and standing balance are associated with higher mortality rates [18, 20]. In addition, lower levels of physical capability are associated with higher risk of cardiovascular disease (CVD), dementia, institutionalisation and difficulties performing activities of daily living (ADLs) [21].
Areas lacking adequate evidence
Recent work suggests that there is added value, for the prediction of mortality, in assessing different measures of physical capability in midlife [20]. However, there is currently insufficient evidence to recommend an order of priority for these measures or to define, with confidence, the minimum number of measures that should be made across the range of older ages and for different research questions. More research is also needed on the utility of some measures such as performance in the pegboard test (dexterity) which has been understudied.
There is a need for more studies with longitudinal data on change in physical capability, and need to assess physical capability in relation to other positive aspects of health, such as quality of life, that may be important criteria for healthy ageing [22].
Biomarkers of physiological function
Complex molecular changes affecting the structure and function of most cells, tissues and organ systems are a hallmark of ageing [8], and changes in their function can be detected by the third or fourth decades of life [23]. Here we focused on biomarkers of lung function, body composition (including bone mass and skeletal muscle), cardiovascular (CV) function and glucose metabolism (Fig. 2 and Additional file 1: Table S2). From age 25 years, forced expiratory volume (FEV1) declines at approximately 32 ml/year in men and 25 ml/year in women, and there are inverse associations between FEV1 and mortality, cognitive function and fractures [24]. Bone mass declines with age, and bone mass or density predicts risk for future fracture and mortality. Large waist circumference, greater body mass index (BMI) and weight-gain in middle age are all associated with higher mortality or lower healthy survival [25, 26]. In addition, low skeletal muscle mass is associated with increased likelihood of functional impairment and disability. Blood pressure (BP) [27, 28] and blood lipids [29, 30] are currently the strongest predictors of CV morbidity and mortality. Increases in diastolic BP and systolic BP are associated with increased risk of CV mortality [27] and high BP in midlife with cognitive decline in later life. Ageing is associated with reduced metabolic capacity exemplified by diminished glucose homeostasis. Raised fasting blood glucose and glycated haemoglobin (HbA1C) are associated with age, CV events and mortality, cognitive impairment, and dementia, in non-diabetics.
Areas lacking adequate evidence
Emerging biomarkers, for example fibrinogen, plasma cystatin C and brain natriuretic peptide, have been associated with increased risk of CV events and mortality, but it is uncertain if these offer advantages over well-established biomarkers. More research is needed on whether monitoring biomarkers over longer time periods, for example glucose concentration and ambulatory BP over 24 hours, or in response to a challenge, improves their predictive value.
Biomarkers of cognitive function
Cognitive decline may limit independence and signal dementia [31], and, although debated [32], evidence indicates that the onset of cognitive decline is detectable relatively early in adulthood, for example from around 45 years of age or earlier in some functions [33]. We focused on cognitive domains assessed widely in human ageing studies and employed in the NIH Toolbox. We identified nine domains together with tests commonly used for their assessment (Additional file 1: Table S3). Based on current evidence, three domains – executive function, processing speed and episodic memory – are a possible minimum set of domains to be assessed in ageing studies (Fig. 2). If assessment time allows, tests of crystallised cognitive ability and non-verbal reasoning would be useful additions. Executive function is markedly affected by ageing [34], exhibiting an inverted U-shape pattern across the lifespan. Processing speed declines progressively with age [35] and is associated with greater mortality risk [36], CVD and respiratory disease [37]. In addition, episodic memory is sensitive to brain ageing and declines in individuals with mild cognitive impairment and neurodegenerative diseases [38]. A standard deviation advantage in memory is associated with 21 % reduction in mortality risk among older individuals [39].
Areas lacking adequate evidence
To date, computer-based tests are not widely used in major cohorts; availability of tools such as the NIH Toolbox and the imperative to increase cost-effectiveness are likely to drive the migration to digital methodologies. This will require that tests are supported by on-going technical development to ‘future-proof’ operating systems and hardware. Where tests are administered repeatedly in the same individuals problems associated with practice and familiarity need to be addressed. The issue of co-variance among cognitive tests needs more attention because those who score well on one test tend to score well on others [40]. Salthouse and others have highlighted that the causes of cognitive ageing might affect the variance shared by tests or domains or the variance in a specific test or domain [40].
Biomarkers of endocrine function
Age-related changes in the endocrine system, particularly the sex hormones, are well recognised and have established causal links with health outcomes. We focused on sex hormones, the HPA axis, growth hormone IGF-1, melatonin, adipokines and thyroid hormones (Fig. 2 and Additional file 1: Table S4). Strong consensual evidence from longitudinal studies indicates that testosterone, estrogen, DHEAS and growth hormone IGF-1 are linked with risk of premature mortality and physical frailty [41]. For some biomarkers, the relationship with ageing appears to be non-linear, for example both high and low IGF-1 are related to greater mortality rates. DHEAS declines with age from the third decade onwards and low DHEAS is associated with increased mortality in older subjects with concurrent frailty. Hormone replacement studies suggest causal links for both testosterone and estrogen and risk of physical frailty and bone health [42, 43]. Cortisol is associated with age-related disease and disability [44], and abnormal cortisol secretion patterns are associated with increased BP, impaired glucose metabolism and increased incidence of CVD and type 2 diabetes in men [45].
Areas lacking adequate evidence
Longitudinal evidence is needed to enhance understanding of the relationships between cortisol, DHEAS, cortisol:DHEAS ratio, adipokines (adiponectin, leptin, ghrelin), somatostatin, and ageing, frailty and mortality.
Biomarkers of immune function
Whilst the field of immunology is well developed, the study of age-related decline in immunity, termed immunosenescence, is more recent [46]. Here we focused on age-related immune function and inflammatory factors (Fig. 2 and Additional file 1: Table S4). Longitudinal studies comparing immune cells or function with mortality, or with age-related functions such as infection rates or vaccination responses, are scarce [47]. Two octogenarian and nonagenarian studies assessing immune markers (T-cell phenotype, cytomegalovirus serostatus and pro-inflammatory cytokine status) with subsequent mortality have been the basis for the development of the immune risk profile (IRP) [48], which is associated with mortality in those over 60 years [49]. A limitation of the IRP is its narrow scope since it does not consider innate immune factors such as natural killer cell (NK cell) function, which is linked with infection rates and mortality. The best studied aspect of immunosenescence is the age-related increase in systemic inflammatory cytokines, inflammageing [50]. Higher plasma concentrations of IL-6 and TNF-α are associated with lower grip strength and gait speed in older adults [51]. Centenarians show fewer signs of ageing of the immune system although some inflammageing is seen.
Areas lacking adequate evidence
Longitudinal studies should examine relationships between number and function of T cells, neutrophils, NK cells, B cells, and mortality, risk of age-related disease and wellbeing in later life. Given the switch from lymphoid to myeloid cell production with age, the lymphocyte/granulocyte ratio is a potentially useful biomarker of healthy ageing. The IRP needs validation in younger people and should be expanded to include measures of immune function such as infection incidence or vaccination response. Telomere length in leukocytes, including lymphocytes and monocytes, has received much attention. Despite its association with ageing in several cohorts, it is likely that shortened telomeres are also a marker of infection frequency so that leukocyte telomere length may not be a reliable index of biological ageing. Further studies of telomere length and ageing should include investigation of exposure to infections and CMV seropositivity as possible confounders. In the Newcastle 85+ Study, telomere length was uninformative about health status [13].
Sensory functions as potential biomarkers of ageing
Sensory functions are critical for normal levels of independence, for interactions with others and to facilitate enjoyment of life’s experiences. Loss of these functions is more prevalent in older adults, with loss of audition and vision being the most prominent. The prevalence of visual impairment increases with age and may reduce the ability to undertake daily activities such as reading, and may limit mobility and social interactions. Olfactory acuity declines with age, is more common among men, and has been proposed as an indicator of brain integrity in older people. Smell dysfunction is among the earliest ‘preclinical’ sign of neurodegenerative diseases such as Alzheimer’s disease and sporadic Parkinson’s disease [52], and is associated with mortality in the National Social Life, Health and Aging Project [53]. The NIH Toolbox [14] measures audition, vision, olfaction, gustation, vestibular function and pain. Most of these functions, with the exception of pain, decrease across the lifespan, and sensory changes may overlap with changes in cognitive and motor functions. However, the predictive value of measures of sensory function for age-related health outcomes remains uncertain as does the opportunity to modulate ageing-related changes in sensory function through lifestyle or other interventions. Further evidence will be needed before sensory measures can be recommended with confidence as reliable markers of healthy ageing.