The between-day reliability of fasted circulating irisin concentrations: a cohort study

Irisin, a myokine transiently released after exercise bouts (Bostrom et al. 2012) and shivering (Glazachev et al. 2021), is implicated in the amelioration of obesity (Shoukry et al. 2016), type 2 diabetes (Riddell et al. 2013), and cognitive decline (Lourenco et al. 2020). Irisin was initially identified as a peroxisome proliferator‐activated receptor γ co‐activator‐1α dependent myokine secreted by proteolytic processing of fibronectin type III domain containing 5 with the potential to induce brown‐fat‐like development of white adipose tissue (Bostrom et al. 2012), improve glucose regulation (Ye et al. 2019), mitochondrial quality (Ye et al. 2019), and stimulate brain-derived neurotrophic factor (Wrann et al. 2013). Given the potential health benefits of increased circulating irisin concentrations, there is considerable interest in the effect of various interventions on acute and chronic change of irisin concentrations (Flori et al. 2021). For example, irisin concentrations have been shown to increase acutely in healthy adults in response to a single-bout of exercise (15%) (Fox et al. 2018) and whole-body repeated hypothermia (17%) (Glazachev et al. 2021), with evidence of an increase in the chronic adaptation of irisin to exercise training in some studies (13%–23%) (Kim et al. 2015; Rodziewicz-Flis et al. 2023) but no chronic adaptation overall in a systematic review and meta-analysis of exercise training studies (Qiu et al. 2015).

Given these small but potentially meaningful increases in irisin, it is crucial to ensure that measurements are precise and reproducible. Measurement of the between-day (biological) reliability of irisin can inform whether observed interventional changes are due to the intervention itself or natural biological fluctuations, and evidence the need for larger or smaller sample sizes to detect meaningful effects. Additionally, given that irisin detection methods are prone to within-day (technical) variability, consideration of both technical and biological reliability enables the sources of variance to be identified and potentially controlled.

Several studies measuring daily biological fluctuations in resting irisin concentrations in young adults have reported mixed findings (Anastasilakis et al. 2014; Löffler et al. 2015; Tsuchiya et al. 2015). Large biological fluctuations in resting irisin (29%) were reported in a study of 20 young healthy participants over a 21 h period, indicating that baseline measures taken at different time points may follow a diurnal rhythm (Anastasilakis et al. 2014). However, a study measuring irisin concentrations every 3 h over a 12 h period in seven healthy young men (25 ± 1 years) detected no significant differences between irisin at different time points (Tsuchiya et al. 2015) and a study measuring irisin concentrations in young (18–35 years), healthy (n = 14),and obese adults (n = 14) every hour over a 24 h period reported larger variance in obese adults similar to those reported by Anastasilakis and colleagues but no diurnal variations overall (Löffler et al. 2015). Therefore, it is unknown whether such fluctuations are observed between samples taken at the same time of day over a 24 h period obscuring the impact potential biological variance may have on intervention studies measuring changes in irisin concentrations. Additionally, studies have shown that individual participants may present different resting irisin concentrations based on age (Huh et al. 2012; Löffler et al. 2015; Korkmaz et al. 2019) sex (Anastasilakis et al. 2014; Löffler et al. 2015) and body composition (Huh et al. 2012; Pardo et al. 2014); and as stated evidence exists of increased variance in obese adults (Löffler et al. 2015), however, the effect of these characteristics on between-day reliability has not been explored.

Given the potential variability in irisin levels over consecutive days and the variations seen in resting irisin levels based on factors such as age, sex, and body composition, the aims of this study were to (1) measure the between-day (i.e., biological) reliability of plasma irisin, separated by 24 h in healthy men and women across a large age range, relative to the within-day (i.e., technical) variability, (2) explore the influence of participant age, sex, and body composition on between-day reliability.

Participant recruitment, study outline and participant inclusion and exclusion criteria have previously been described within publication of the primary study (Rose et al. 2022), and was approved by The University of Queensland's Human Research Ethic Committee (UQ NHMRC HREC, #2018000547) and the Sunshine Coast University Human Research Ethic Committee (UniSC HREC, #A201362) and conducted in accordance with the ethical codes of conduct for the treatment of human subjects. All participants provided written, informed consent. Ninety men and women volunteered to participate in the study (Rose et al. 2022), of the n = 90 sample, n = 6 had <2 two available plasma samples and n = 52 had plasma irisin concentrations that were below the assay limit of detection (<standard 1 of the assay). Hence, n = 32 participants were included within analysis (men n = 15, women n = 17) (Fig. 1). Study size was calculated to detect intraclass correlation coefficient (ICC) values of 0.8 ± 0.15 (Bonett 2002). To obtain comparable baseline concentrations fasted venous blood samples were taken twice over two days. Each measurement was taken by the same technician and conducted at the same time of day (06:00–09:30). Fasted venous samples were taken given some evidence of a positive correlation between irisin and blood glucose (Huh et al. 2012), diminishing irisin concentrations after fasting (Varela-Rodríguez et al. 2016) and decreased irisin in response to excess caloric intake (Schlögl et al. 2015). Participants were divided into young (18–35 years, n = 9), middle-aged (40–60 years, n = 17), and older (65–85 years, n = 6) age groups with body composition measured on day one and blood samples taken on day one and two with 24 h between. Briefly, participants presented fasted (12 h) and rested (24 h). A 24 h rest period was deemed sufficient as previous studies measuring the response of irisin to exercise report a return to baseline concentrations within 24 h (Nygaard et al. 2015; Tsuchiya et al. 2015; He et al. 2018). Participants confirmed their compliance to avoidance of moderate to vigorous intensity physical activity on the day prior to testing by completing a pretest preparation form. Participants were asked to be hydrated, refrain from alcohol, caffeine, tobacco, and non-prescribed medications 4 h before testing. Body composition was assessed on day one using 4-compartiment model, including a composite score of dual-energy X-ray absorptiometry, air-displacement plethysmography, and bioelectrical impedance spectroscopy, as well as height and body mass to calculate body mass index (BMI). Resting venous blood (∼20–30 mL) was sampled twice (24 h apart) from the median, cephalic, or basilica vein by a qualified phlebotomist using a 21–23 gauge needle into plasma (EDTA) vacutainers. Samples were taken between 6:00 am and 9:30 am and stored on ice (30 min), with a maximum difference between times on each day of 2 h. Plasma was removed, aliquoted and stored at ≤−80 °C for later analysis.

Undiluted plasma irisin concentrations were measured using MILLIPLEX® xMAP immunoassay analysis (Merck KGaA, Darmstadt, Germany) using a MAGPIX® instrument within the Sunshine Coast Health Institute. The values reported in this present study fell within the normal range of values measurable using MILLIPLEX® xMAP. Data were analysed using SPSS software (IBM Analytics, A (https://www.google.com.au/search?rlz=1C5CHFA_enAU786AU786&q=Armonk&stick=H4sIAAAAAAAAAOPgE-LQz9U3MC5OtlACsyqTqlK0tLKTrfTzi9IT8zKrEksy8_NQOFYZqYkphaWJRSWpRcUAmtLfx0IAAAA&sa=X&ved=2ahUKEwivybuK35HeAhXTXysKHfOgDWkQmxMoATAmegQIBhAg) rmonk, New York, United States). Means, standard deviations, and 95% confidence intervals (95% CI) for participant characteristics were calculated. A histogram plot was used to determine whether data were normally distributed. As resting irisin data were non-parametric, medians, and interquartile ranges were calculated. A repeated measure one-way ANOVA (alpha level 0.05) was conducted to determine age-group differences in body composition and a Kruskal–Wallis for age-group differences in irisin concentrations.

Within- and between-day reliability of irisin (technical variability between duplicates and biological variability between days) was assessed by calculating typical error of the estimate reported as coefficient of variation (CV) (CV = SD/mean × 100) and interclass correlation (ICC), mean error magnitude between measurements expressed as percent (relative to median irisin concentrations) and absolute (raw) values, and 95% CI of absolute error. Irisin data were normalised by Log10 transformation for calculation of the ICC. The ICCs were interpreted as per Koo and colleagues (Koo and Li 2016). Within-day CV was compared to between-day to assess the difference between technical and biological reliability. Systematic and proportional bias of error were evaluated as per primary study (Rose et al. 2022). A standard Bland–Altman plot visually represented within-day and between-day reliability of irisin values and bias, upper and lower limits of agreement (Pereira et al. 2007). Linear regression analysis was used to calculate the influence of sex, age, and body composition (BMI, fat free mass, fat mass, and percentage body fat) on between between-day error.

Participants were within normal to overweight BMI ranges (Jabre and Bland 2021; Zierle-Ghosh and Jan 2024) and %BF (Heo et al. 2012) when compared to normative reference values based on age and sex. There were no significant differences in body composition or resting irisin concentrations between groups (Table 1). Participants had a mean ± standard deviation BMI of 25.8 ± 5.0 kg/m², and young, middle, and older age groups were 23 ± 3 years, 52 ± 7 years, and 72 ± 5 years of age, respectively.

Both between-day (n = 32, ICC = 0.963) and within-day (n = 32, ICC = 0.927) reliability were excellent (Table 2). There was no systematic bias present (n = 32, 95%CI, −0.40, 1.35) (Fig. 2), or evidence of proportional bias for either within-day or between-day error results. Regression analysis revealed no significant influence of age, sex, or body composition on between-day irisin reliability (n = 32, r < 0.158, p > 0.875) (Table 2).

The primary aim of this sub-study was to assess between-day reliability of resting plasma irisin in healthy adults and secondarily to explore the association and potential influence of participant age, sex, and body composition on irisin reliability. Overall, between-day measurement of irisin was reliable and unaffected by age, sex, and body composition.

For investigations exploring the effects of interventions on irisin concentrations, reliable measurements are crucial. Researchers seeking to establish the benefits of interventions that increase irisin need to be confident that observed changes are not merely due to biological variability in resting irisin. Biological variance was observed in irisin as an average 1.1% increase in between-day CV compared to within-day, therefore, studies may use this margin as a reference point, particularly if an intervention demonstrates substantial changes in irisin levels exceeding 1.1% plus within-day variability.

Despite evidence that resting irisin concentrations may be affected by age (Korkmaz et al. 2019), sex (Anastasilakis et al. 2014), and body composition (Huh et al. 2012; Anastasilakis et al. 2014; Pardo et al. 2014), reliability was not influenced by these factors in the present study. Therefore, our reported between-day reliability for irisin is likely to apply to all healthy adults. Between-day reliability was also shown to have no proportional bias indicating that changes in irisin concentration after interventions should not affect reliability.

When considering the impact of irisin reliability on intervention study outcomes, it is of note that only 50% of participants demonstrated a percentage error of less than 15% which is the average improvement in irisin expected after acute exercise (21 studies) (Fox et al. 2018). A similar increase is also reported in response to whole-body repeated hyperthermia (16.7%) (Glazachev et al. 2021) and exercise training (18%) (Kim et al. 2016). Therefore, the use of a control group is recommended in these interventions.

Although Anastasilakis and colleagues (Anastasilakis et al. 2014) reported that resting irisin appears to fluctuate significantly within a day, no predictable change in resting irisin taken at the same time of day over a 24 h period was detected in this study. Given this, it is possible that large differences between sampling time of day pre- and post-intervention could mask intervention effects, therefore, this is an important consideration when planning studies so validity of study outcomes can be enhanced.

This is the first study to investigate the between-day reliability of resting irisin concentrations. Strengths of this study include a comparison between groups to address reliability across multiple ages, body composition and between sexes for a wider application of results. However, there are limitations that need to be considered. Our approach to controlling BMI among age groups (similar range and mean value) to ensure gross similarities to isolate body composition differences may have attenuated the influence of age and limited our ability to detect the effect of body composition on between-day reliability. This may have limited our capacity to discern the influence of body composition on day-to-day reliability.

In conclusion, plasma irisin concentrations had excellent reliability between-days, indicating that significant intervention-induced changes likely represent a meaningful intervention effect. Reliability varied considerably between some individuals between days; however, this was not explained by sex, age, body composition, or resting irisin concentrations. Future studies should implement similar food, fluid, and activity controls prior to blood sampling to ensure this level of reliability. Overall, with pre-testing controls included, results of further research can be interpreted with confidence, when change is beyond between-day error limits reported here.