Low serum klotho concentration is associated with worse cognition, psychological components of frailty, dependence, and falls in nursing home residents

Serum alpha-klotho (s-klotho) protein has been linked with lifespan, and low concentrations of s-klotho have been associated with worse physical and cognitive outcomes. Although its significance in aging remains unclear, s-klotho has been proposed as a molecular biomarker of frailty and dependence. This study is a secondary analysis of data from a clinical trial performed in a population of 103 older individuals living in 10 nursing homes in Gipuzkoa (Spain). We aimed to elucidate associations between s-klotho (as measured by enzyme-linked immunosorbent assay) and body composition, physical fitness, and cognition, as well as frailty and dependence (determined using validated tests and scales). In addition, we investigated the association of s-klotho concentration with falls in the six months following the initial assessment. Low s-klotho levels were associated with a lower score in the psychological component of the Tilburg Frailty Indicator, a worse score in the Coding Wechsler Adult Intelligence Scale, and a greater dependence in activities of daily living. Moreover, participants with lower s-klotho concentrations suffered more falls during the 6 months after the assessment. Future translational research should aim to validate klotho’s putative role as a biomarker that could identify the risk of aging-related adverse events in clinical practice.

www.nature.com/scientificreports/ association of s-klotho with parameters related to several hierarchical domains of aging 42 (Fig. 1b) in a population of older people living in nursing homes. In particular, we aimed to ascertain the association of s-klotho (1) with the phenotypic (physical fitness and cognition) and functional (frailty and dependence in ADL) domains of aging and (2) with falls in a six-month follow-up (Fig. 1c).

Materials and methods
Study design and participants. This is a secondary analysis of the baseline data from a randomized controlled physical exercise intervention that included a 6-month follow-up for falls. The study was conducted at ten nursing homes in Gipuzkoa, Spain (ACTRN12616001044415), between October 2016 and July 2017 43 . The study adheres to the STROBE guidelines. This study was approved by the Ethics Committee for Research in Humans (CEISH: M10/2016/105) and by the Ethics Committee for Research with Biological Agents of the University of the Basque Country (CEIAB: M30/2016/106). Written informed consent was provided by each participant. The results of this exercise intervention were published previously [44][45][46] . The primary study included 112 men and women living in nursing homes who met the following criteria: ≥ 70 years old; scored ≥ 50 on the Barthel Index for Activities of Daily Living (0-100) 47 ; scored ≥ 20 on the MEC-35 test (0-35), an adapted and validated version of the Mini Mental State Examination (MMSE) in Spanish 48 ; and were capable of standing up and walking independently for at least ten meters. In this secondary analysis, 103 available serum samples from the 112 total participants were analyzed. Serum klotho concentration. Baseline blood samples were collected in the morning following an overnight fast. Following collection, the tubes were centrifuged at 5000×g for 10 min. The serum obtained from each participant was stored in aliquots at − 80 °C for further analysis. A commercial enzyme-linked immunosorbent assay (ELISA) was performed to measure klotho serum concentration according to the manufacturer's protocol (Human soluble α-Klotho Assay Kit JP27998, Immuno-Biological Laboratories Co., Ltd., Gunma, Japan). The quantification was performed spectrophotometrically using a FLUOstar OPTIMA Microplate Reader (ThermoFisher Scientific, Waltham, MA, USA) and Optima Control software version 2.20 (http:// www. tmd. ac. jp/ cmn/ gene/ center/ 0413F 0011A% 20% 20Soft ware% 20Man ual% 20OPT IMA% 20Part% 20II. pdf) (BMG LABTECH GmbH, Offenburg, Germany). Serum samples were measured in duplicate and averaged. Coefficient of variability (%CV) was calculated using the following formula: %CV = standard deviation/mean. Limits for %CV were set at 5% and 10% for intra-assay and inter-assay measurements. respectively. All calculated %CV values were within these limits.
Other analyzed variables. At the baseline of the intervention, sociodemographic data (sex, age and education level) and Barthel Index scores 47 were recorded from the database of the nursing homes. The Barthel Index reflects dependence in activities of daily living (ADL), and its score ranges from 0 to 100, with lower scores indicating a higher degree of dependence.
Regarding anthropometry, waist and hip circumference were measured with a non-elastic anthropometric tape and measured to the nearest 0.1 cm. Height was measured with a Holtain stadiometer to the nearest 0.1 cm, and body mass was measured with an Omron digital scale to the nearest 0.1 kg. Body mass index (BMI) was calculated based on height and mass, and the waist-to-hip ratio was based on waist and hip circumferences. Percentages of muscle and fat were measured using a portable bioelectrical impedance analyzer (Bodystat BIA Quadscan 4000, BOSYSTAT LTD; Isle of Man, UK). This method was supported by the European Working Group on Sarcopenia in Older People (EWGSOP) as a valid and cost-effective alternative to dual X-ray absorptiometry (DXA) 49 .
Physical fitness was assessed through the handgrip strength test using a Jamar dynamometer 50 , and aerobic capacity was evaluated with the 6-Min Walk test of the Senior Fitness Test 51 . Dynamic balance was assessed with the Timed Up and Go Test 52 , and static balance with the Berg Balance Scale 53 . Lower limb function (static balance, walking velocity, and lower limb strength) was evaluated with the Short Physical Performance Battery (SPPB) 54 .
Global cognition was assessed using the Montreal Cognitive Assessment (MoCA) 55 , which included visuospatial/executive function, naming, attention, language, abstraction, delayed recall, and orientation. Trail Making Test A 56 was administered to evaluate executive function, and the Coding test was used to measure processing speed on the Wechsler Adult Intelligence Scale, Fourth Edition (WAIS IV) 57 . Higher scores on these scales indicate better cognitive functioning, with the exception of Trail Making Test A, in which higher scores indicate worse performance. The MoCA, Trail Making Test A and Coding cognitive tests require participants to see and draw. Therefore, participants with visual impairments or difficulties holding a pen were unable to perform (some of) these tests. Regarding Trail Making Test A, some of the participants were unable to finish the test; thus, no result was recorded. For these reasons, the number of participants in cognitive tests varies.
Frailty status was assessed with the Fried Frailty Index (FFI) 25 , the Tilburg Frailty Indicator (TFI) 58 , and the Clinical Frailty Scale (CFS) 59 . Frailty, according to the Fried's frailty phenotype, was identified by the presence of three or more of the following signs/symptoms: unintended weight loss, exhaustion, weakness, slow gait speed, and low physical activit 25 . The TFI contains 15 questions on physical, psychological, and social domains of frailty. Participants with a score of 5 or more were considered frail according to the TFI 58 . For the CFS, frailty status is based on clinical judgment. Possible scores range from 1 to 9, and participants with a score of 6 points or higher were considered frail.
All of the tests were performed by accredited and experienced professionals in each area. The number of falls suffered by each participant within six months after baseline was recorded from the database of the nursing homes. www.nature.com/scientificreports/ Statistical analyses. Statistical analysis was performed using SPSS version 24 (https:// www. ibm. com/ eses/ analy tics/ spss-stati stics-softw are) (IBM, Chicago, IL, USA). The Kolmogorov-Smirnov test was used to test the normality of the data. When data were not normally distributed (s-klotho concentration, hand grip, Timed Up and Go test, all cognition tests, Barthel scale, frailty indicators), they were square root transformed before being analyzed using parametric tests. Qualitative variables are presented as n (%) and quantitative variables as mean ± SD. Due to the exploratory nature of this study, and in order to avoid type II errors, Bonferroni correction was not applied to the statistical analysis 60 .
Pearson correlations were carried out to examine associations between s-klotho concentration and quantitative variables (age, anthropometry, physical fitness, cognition, Barthel Index and frailty scores). Due to the uneven distribution of the number of falls, its association with klotho was calculated using a univariate Poisson regression. Differences in s-klotho levels between groups based on qualitative variables (sex, education level, frailty status, assignation to the control or the intervention group for the exercise program and fallers vs. non fallers) were assessed using Student's t-test.
Results were considered statistically significant at p < 0.05.

Results
Descriptive characteristics of the sample. Baseline descriptive data are shown in Association of s-klotho with quantitative variables. Table 2 shows baseline Pearson's (Table 2A) and Poisson's (Table 2B) correlations of s-klotho with the quantitative variables analyzed. No association was observed between s-klotho concentration and age, anthropometry, or physical fitness. Regarding cognition, participants with lower s-klotho levels showed a trend toward worse scores in cognition-related tests, with statistical significance in the coding test (WAIS IV), which reflects processing speed (r = 0.256, p = 0.012). Regarding the Barthel Index, lower s-klotho was associated with higher dependence in ADL (r = 0.253, p = 0.010). When analyzing frailty scores, the only association found was with the psychological components of the TFI, with lower klotho concentration being associated with higher scores in psychological components of frailty (r =− 0.231, p = 0.026). S-klotho was also negatively associated with the number of falls (B = − 3.311, p < 0.001).
Differences in s-klotho within groups of qualitative variables. Table 3 shows the mean (± standard deviation) s-klotho concentration according to sex, education level, frailty status, assignation to the control or the intervention group for the exercise program and falling status during the 6-month follow-up. S-klotho levels were lower in participants who fell compared to those who did not fall (p = 0.037). No other statistically significant differences were found for qualitative variables.

Discussion
In this study, we have analyzed serum klotho concentrations in a population of 103 older adults living in nursing homes. Lower s-klotho concentration was associated with lower processing speed (higher score in the coding test [WAIS IV]) and more dependence in ADL (lower score in the Barthel Index). In addition, participants with lower s-klotho levels showed a worse score in the psychological components of the TFI. However, no association was found with any other scores on frailty assessment tools, nor were any differences found between the frail and non-frail participants in any of the tools used. Participants with lower s-klotho levels suffered more falls in the six-month follow-up. Due to the described role of klotho in preventing muscle loss 1 , it could be predicted to be positively associated with parameters related to muscle and physical fitness. However, in our work, neither body composition nor scores in physical performance tests were associated with s-klotho. Other authors have described an association between lower s-klotho and less muscle mass 16 . The disagreement between their results and those of our study could be due to the population studied (middle-aged sedentary adults vs. older people living in nursing homes) or in the method employed to determine body composition (DXA vs. bioelectrical impedance analyzer).
Regarding physical performance, lower s-klotho has been associated with less strength 12,13 and worse results in tests of physical fitness 14,15,61 . This discordance between these results and ours could be due to the different populations under study, as these previous studies were conducted in general samples consisting of younger and fitter community-dwelling middle-aged 14,16 or older adults [12][13][14][15] . To our knowledge, no study on the analysis of s-klotho levels in populations living in nursing homes has yet been published.
Although the association of klotho with cognition is not yet well established in humans 15 , lower s-klotho concentration has been associated with worse global cognitive function 15,21 and higher risk of cognitive decline 21 . In our study, participants with lower serum s-klotho concentrations obtained worse results in cognition-related tests, although this association was only significant for the coding test. The coding test measures processing speed 57 , whereas the Trail Making Test A reflects executive function 56 . Similar results that linked global cognitive function but not executive function with s-klotho have been described by Shardell  www.nature.com/scientificreports/ older adults. Better cognition associated with enhanced expression of klotho has been demonstrated in animal models 3,7 as well as in humans 8,15,17,21,22 ; this is likely linked to the action of brain-secreted klotho on the structure of the synapses 18 . Although the mechanisms underlying klotho's regulation in the brain are still unknown, klotho has been proposed as a potential target for therapies aimed at cognitive enhancement 62 . Based on klotho's functions and some preclinical and clinical evidence, Angulo et al. 9 have proposed that its upregulation would prevent phenotypic characteristics of frailty in the elderly. Loss of function of serum s-klotho has also been described as a quantifiable biomarker on the continuum between frailty and resilience 28 and for frailty diagnosis and stratification 28 . However, we did not observe differences in serum s-klotho concentrations between frail and non-frail older people living in nursing homes.
Despite the lack of association between klotho and frailty in our study, participants with lower s-klotho levels exhibited lower scores in the psychological components of the TFI. Psychological frailty encompasses cognitive, www.nature.com/scientificreports/ mood, and motivational components 63 , and some authors have found its association with mental health-related quality of life 64 . In addition, De Roeck et al. 65 have recently proposed that cognitive frailty can occur independent of the other frailty domains, including physical frailty, and can therefore be seen as a distinct concept. These concepts and findings are in line with our result, in which s-klotho was associated with cognition and psychological components of the TFI, but not with the physical or clinical tools for identifying frailty. The lack of significant associations of s-klotho with muscle-related variables and frailty could be explained by the suspicion that the biology and functions of klotho may be more complex than previously hypothesized 6,28,66 . Thus, compensatory/resilient mechanisms could mask the consequences of changes in molecular levels on the phenotypic and functional responses 42 . Moreover, most of the published studies on klotho are related to its serum concentration or its genetic expression. However, klotho is a multifunctional protein with enzymatic activity as a glucuronidase (EC: 3.2.1.31) hydrolyzing extracellular sugars and has great importance in signaling pathways 67 . As it is able to perform multiple functions (hormone, enzyme, and receptor) 66 that differ mechanistically 68 , klotho could be considered a moonlighting enzyme. This could have consequences in the comprehension of its regulatory mechanisms and supposed functions and implications for the functionality of the organism.
Our results also show that individuals with lower s-klotho concentrations were more dependent in their performance of ADL. Similar results have been published by Crasto et al. 15 in a population of communitydwelling older men and women. The loss of independence is one of the ultimate consequences of age-associated decline, but the changes that occur with aging, from the molecular to the functional level, are interconnected in a hierarchical arrangement. Eventually, these changes are reflected at the physical, cognitive, and emotional levels, as well as in social functions and ADL, as a result of the interaction between entropic and compensatory mechanisms 42 . However, literature about age-associated dependence and molecular biomarkers is scarce, and further insight is needed into possible biological mechanisms that may underlie risk of disability 15 . www.nature.com/scientificreports/ We found an independent association of participants with a lower concentration of s-klotho at baseline and falls throughout the six-month follow-up. This association was apparent both when analyzing the number of falls and when analyzing whether participants fell or not. Falls have a significant impact on health outcomes in the elderly and are associated with various factors, including impairments in cognition and dependence in ADL 38 . Due to the consequences of falling, identifying fall prediction tools might lead to novel translational approaches from basic research to clinical practice.
Few studies have focused on the relationship between molecular biomarkers and falls 34 , and to our knowledge, no study has yet analyzed the relationship between klotho and the risk of falling. However, there is evidence for associations between falling risk and two molecules closely related to klotho, vitamin D and FGF-23 39 , that could support our results. Vitamin D promotes secretion of circulating klotho hormone 39 , and the metabolism of these molecules is tightly associated 39,67,69 . Ensuring adequate vitamin D levels has been proposed as a strategy to reduce falls, although the results of studies investigating whether vitamin D supplementation can reduce the number of falls are inconsistent 36,40 . Indeed, it has been shown that community-dwelling older women with lower serum vitamin D concentrations suffered more falls despite the absence of an association between physical performance and vitamin D concentration 70 . These results are in line with the lack of an association between s-klotho levels and physical fitness -related parameters in our study, despite the former's association with falls. Thus, as Uusi-Rasi et al. proposed for vitamin D 70 , s-klotho could modulate individual fall risk, albeit partially. In contrast, FGF-23 is a bone-derived hormone that inhibits α-hydroxylase, reducing serum 1, 25-dihidroxyvitamin D 71 , and α-klotho is a co-receptor for FGF-23 in the kidney 39 . Although FGF-23 has anti-aging effects, it also has some "off-target" actions that could actually be pro-aging 72 . It has been proposed that these off-target FGF-23 pathologies are opposed by secreted klotho 39 . Higher serum FGF-23 levels were associated with an increased risk of falls in patients with chronic kidney disease 41 . The opposite associations of klotho and FGF-23 with falls are in line with the opposing actions of klotho and FGF-23. Overall, these associations between biomolecules and falls support a role for the vitamin D/FGF-23/klotho system, which has been described as crucial for healthy aging, in the assessment of the risk of falls 39 .
This study has some limitations, as the small sample size, the lack of information about the medications the participants were taking, and any cognitive decline during the 6-month follow-up could act as confounders in the analysis. Due to the exploratory nature of this study and in order to increase statistical power, Bonferroni correction was not applied to the statistical analysis. However, this omission may have increased the chance of type I error. Moreover, although the bioelectrical impedance analyzer is accepted by the EWGSOP for body composition measurement and is an economic and portable technology, we recognize that more accurate results could have been obtained if DXA had been used for body composition calculation. In addition, the analysis of Table 3. Student's t-test comparing s-klotho concentrations in groups based on sex, years of education, frailty status, participation in the exercise program and falls. SD standard deviation. www.nature.com/scientificreports/ other molecular biomarkers related to klotho metabolism, such as FGF-23 or vitamin D, could have shown a possible interaction as molecules implicated in the regulation of the physiopathology of falling risk. Nonetheless, this study presents a novel analysis of serum klotho and its association with future falls, presenting a line of research that could be further pursued. The introduction of a low-cost, easily standardized diagnostic tool to estimate the probability of developing falls in the near future would be very useful for routine clinical practice. Promotion of research models from the molecular mechanisms to the functional domains of aging will provide valuable information on both the molecular pathway regulation level and the functional level, which is linked to successful aging. Last, but not least, the studied population should be considered a strength of this work, since people living in nursing homes have specific characteristics and high vulnerability and comprise a poorly studied population. In addition, this study provides knowledge from an integrative perspective on aging-related outcomes, from the molecular to the phenotypic and functional domains of aging.
In conclusion, in a population of nursing home residents, lower s-klotho concentration was associated with worse cognition, scores indicating frailty in the psychological domain of the TFI, higher dependence in ADL, and future falls. These findings provide new and valuable information about the supposed role of klotho as a molecular biomarker linked to aging-related adverse events. Further investigation should be conducted to validate the putative ability of klotho to predict falling. This future research should include panels of associated molecular biomarkers that could be easily translated into clinical practice.