Predictors of the Acute Postprandial Response to Breaking Up Prolonged Sitting

Purpose: To identify predictors of favourable and glucose levels in response to interrupting prolonged sitting time with standing or light intensity physical activity. Methods: Data were combined from four similarly designed randomised acute crossover trials (n=129; BMI range 19.6 to 44.6kg/m2; South Asian=31.0%; dysglycaemia=27.1%). Treatments included: prolonged sitting (6.5hours) or prolonged sitting broken-up with either standing or light-intensity physical activity (5 minutes every 30 minutes). Time-averaged postprandial responses for insulin and glucose were calculated for each treatment (mean±95% CI). Mutually adjusted interaction terms were used to examine whether anthropometric (BMI), demographic (age, sex, ethnicity (white European vs. South Asian)) and a cardiometabolic variable (HOMA-IR) modified responses. Results : Postprandial insulin and glucose were reduced when individuals interrupted prolonged sitting with bouts of light physical activity, but not with standing. Reductions in time-averaged postprandial insulin were more pronounced if individuals were South Asian compared with white European (-18.9mU/L (-23.5%) vs. -8.2mU/L (-9.3%)), female compared to male (-15.0mU/L (-21.2%) vs. -12.1mU/L (-17.6%)) or had a BMI ≥27.2kg/m2 (-20.9mU/L (-22.9%) vs. -8.7mU/L (-18.2%)). Similarly, being female (-0.4mmol/L (-0.6mmol/L, -0.2mmol/L) (-6.8%) vs. –0.1mmol/L (-0.3mmol/L, 1mmol/L) (-1.7%)) or having a BMI ≥27.2kg/m2 (-0.4mmol/L (-0.6mmol/L, -0.2mmol/L) (-6.7%) vs. – 0.2mmol/L interventions in high-risk participants for whom breaking prolonged sitting time with light activity may yield the greatest therapeutic potential. The present findings suggest that standard demographic and anthropometric outcomes may predict the postprandial response to breaking up prolonged sitting with regular bouts of light intensity physical activity. Being female, South Asian or having a higher BMI, all predicted greater reductions in postprandial insulin, while being female and having a higher BMI predicted greater reduction in postprandial glucose. These results may be used to guide individualised tailored interventions in high risk participants for whom breaking prolonged sitting time could be a viable and effective prevention strategy.


Abstract
Purpose: To identify predictors of favourable changes to postprandial insulin and glucose levels in response to interrupting prolonged sitting time with standing or light intensity physical activity. Methods: Data were combined from four similarly designed randomised acute crossover trials (n=129; BMI range 19.6 to 44.6kg/m2; South Asian=31.0%; dysglycaemia=27.1%).
Treatments included: prolonged sitting (6.5hours) or prolonged sitting broken-up with either standing or light-intensity physical activity (5 minutes every 30 minutes). Time-averaged postprandial responses for insulin and glucose were calculated for each treatment (mean±95% CI). Mutually adjusted interaction terms were used to examine whether anthropometric (BMI), Asian or having a higher BMI, all predicted greater reductions in postprandial insulin, while being female and having a higher BMI predicted greater reductions in postprandial glucose when sitting was interrupted with light physical activity. These results could help to guide personalised

Introduction
Postprandial hyperglycaemia plays a significant role in the development of cardiovascular disease (CVD) in people with type 2 diabetes mellitus (T2DM) (1). The postprandial phase is characterised by a rapid and large increase in blood glucose and insulin levels. Observational evidence suggests that postprandial hyperglycaemia, even in the absence of fasting hyperglycaemia, is associated with higher risks of future cardiometabolic disease (2,3).
Similarly, a hyperinsulinaemic response is closely associated with a number of CVD and T2DM related outcomes (4). Therefore, if these links are in part causal, establishing effective and pragmatic interventions that reduce post-meal hyperglycaemic and hyperinsulinaemic excursions could be important therapeutic targets for the prevention of T2DM and CVD, particularly as individuals spend a large proportion of the day in a postprandial state (5).
Physical activity is known to enhance health and improve postprandial hyperglycaemia (6).
Current physical activity guidelines recommend that adults engage in at least ≥150 minutes of moderate intensity physical activity or ≥75 minutes of vigorous activity and 2-3 resistance exercise sessions per week (7). In addition, current physical activity guidelines now include specific recommendations to reduce and interrupt prolonged sitting (6,8). These guidelines have been informed by emerging research suggesting that sitting time per se is an independent risk factor for cardiometabolic morbidity and mortality (9, 10). Over recent years, epidemiological research has been complemented by acute experimental studies showing that breaking up bouts of prolonged sitting with standing or light intensity activity elicits significant benefits on markers of metabolic health (11)(12)(13)(14)(15).

A C C E P T E D
These results are important as light intensity activities are behaviourally more ubiquitous than moderate to vigorous physical activity (MVPA) and may therefore be appealing interventional targets in the promotion of metabolic health, whilst also being more culturally acceptable to high risk groups (e.g. South Asian women). However, the inter-individual variability in the effectiveness of such interventions is likely to be large. For example, previous experimental research has shown that the magnitude of postprandial dysglycaemia in response to prolonged sitting and the subsequent reduction following breaks may differ considerably according to ethnicity or the degree of underlying insulin resistance (13, 16). Therefore, in order to ensure future T2DM prevention strategies are stratified and targeted at those who could derive the greatest benefit, it is necessary to determine the factors that may predict a favourable response to breaking up prolonged sitting with a low intensity intervention.As such, the aim was to determine whether commonly measured demographic, anthropometric or clinical factors are associated with the postprandial insulin and glucose response when breaking up prolonged sitting, with short bouts of either standing or physical activity, at a light intensity.

Study design
We performed a pooled analysis of data collected from 129 individuals across four separate acute, randomised, crossover experimental studies conducted within the Leicester Diabetes Centre (University of Leicester) (n=99) and the University of Glasgow (n=30), UK (2015-2018); all of which followed the same protocols and standard operating procedures for data collection A C C E P T E D and the same treatment methodology of breaking sitting time with 5 minutes of standing or light physical activity every 30 minutes (see Figure, Supplemental Digital Content 1, protocols and standard operating procedures for data collection, http://links.lww.com/MSS/B865). The research design and methods have been published in detail elsewhere (11)(12)(13)(14). Briefly, participants were recruited from studies previously conducted within the Leicester Diabetes Centre (ACUTE, ARMING HEALTH, STAND UP) or from the public via strategic placement and distribution of promotional materials (STAND UP, FIT2SIT). Detailed inclusion and exclusion criteria can be found in Supplementary Digital Content Table 1 (see Table, Supplemental Participants attended up to four separate visits to their corresponding centre. One to two weeks after an initial familiarisation visit, participants were randomised to the following treatment conditions: 1) prolonged sitting (6.5 hours; plus 60 minute steady state); 2) prolonged sitting broken up with standing for 5 minutes every 30 minutes or 3) prolonged sitting broken up with physical activity (either walking or arm ergometry) for 5 minutes every 30 minutes. As an acute bout of physical activity may enhance insulin sensitivity for up to 48 hours, we used a minimum wash-out period of 7 days between each condition.
All studies were registered with clinicaltrials.gov (ACUTE: NCT02135172; STAND UP: NCT02453204; ARMING HEALTH: NCT02909894; FIT2SIT: NCT02493309). Written informed consent was obtained from all eligible participants and the individual studies had full ethical and governance approval.

Participants
In total, 147 participants were randomised. Causes of drop out between familiarisation and randomisation are detailed in Figure 1. A further 18 individuals were excluded after randomisation: due to cessation of the venous cannula line which resulted in less than 50% of data collection (n=11); illness (n=2); inability to tolerate the standardised meal (n=2), unable to commit time (n=2); or a change in personal circumstance (n=1). This left 129 participants that were included in the analysis.

Familiarisation visit
Before participating in the experimental protocol, participants visited the Leicester Diabetes Centre or University of Glasgow for a familiarisation visit in which they were accustomed to the required power output for the arm ergometry or walking speed (self-perceived light intensity) .
Participants were instructed to walk at a pace they felt was comfortable and registered between 10 and 12 on the Borg RPE scale (17

Experimental treatment overview
Participants were asked to record all food and drink consumed the day before the first experimental condition. They were then asked to replicate this diet before subsequent treatments.
Participants were also requested to avoid alcohol, caffeine and any MVPA for two days prior to each experimental condition (11)(12)(13)(14).
Participants arrived at the laboratory after a 10-hour fast and had a cannula fitted into an accessible arm vein and then asked to sit quietly for 60 minutes. Prolonged sitting (6.5

hours) (ACUTE, STAND UP, ARMING HEALTH, FIT2SIT)
All four studies included a prolonged sitting condition (11)(12)(13)(14), where walking and standing was restricted (lavatory visits were conducted via a wheelchair). Participants sat in a designated room equipped with a chair, desk, laptop and access to books and magazines.

Standing: Sitting (total 5.5 hours) + Standing (total 60 minutes) (ACUTE, STAND UP)
Two studies employed a standing protocol (13, 14) which followed the same procedure as the sitting condition, except that participants were instructed to break their sitting time by standing close to their chair for 5 minutes, every 30 minutes. Individuals were asked to stand in the same, fixed position. In total, individuals accumulated 12 bouts (60 minutes) of standing.

Walking: Sitting (total 5.5 hours) + walking (total 60 minutes) (ACUTE, STAND UP, FIT2SIT)
Three studies employed a walking protocol (12-14) which was similar to the standing condition, but participants conducted 5-minute bouts of walking at a light intensity. Walking speed ranged from 1.5 to 4.4 km/h. In total, individuals accumulated 12 bouts (60 minutes) of walking. For the ACUTE and FIT2SIT trials, the walking breaks were carried out on a treadmill (Spazio Forma

A C C E P T E D
Folding Treadmill/ Excite 700, TechnoGym U.K. Ltd., Bracknell, U.K). For the STAND UP trial participants were instructed to walk up and down a marked track in the laboratory.

HEALTH)
One study employed upper body physical activity through arm ergometry (11). The power output (watts) necessary to elicit the desired energy expenditure during the main experimental condition (equivalent to walking at 3km/h) was established during the familiarisation visit (11). The subsequent power output was implemented for 5 minutes, every 30 minutes. In total, individuals accumulated 12 bouts (60 minutes) of arm ergometry.

Cardiometabolic variables
For the studies conducted solely at the Leicester Diabetes Centre (11,12,14), all samples were analysed within the same location. Plasma glucose was determined using standard enzymatic techniques with commercially available kits (Beckman, High Wycombe, UK) and using stable methodology standardized to external quality assurance reference values. Insulin and glucose samples underwent centrifugation to separate plasma within 15 minutes of collection. Plasma derived from insulin was stored at -80 o C and analysed at the end of data collection using an enzyme immuno-assay (Mercodia, Uppsala, Sweden). Each sample was analysed in duplicate to ensure reliability of readings. Sample values with ≥20% variability were reanalysed.
All samples for STAND UP (13) were analysed at the University of Glasgow. Glucose was analysed using clinically validated automated biochemistry platforms (c311, Roche Diagnostics,

A C C E P T E D
Burgess Hill, UK). Insulin and glucose samples underwent identical preparation (centrifugation and storage) to the Leicester samples and were measured with an equivalent immunoassay platform (e411, Roche Diagnostics, Burgess Hill, UK). The analysers were calibrated and quality controlled using the manufacturer's materials. Coefficient of variation over two levels of controls was less than 3% for biochemistry assays and less than 6% for insulin.
All measurements and analysis were undertaken by individuals blinded to experimental condition.

Statistical analyses
Missing outcome data for participants included in this analysis were imputed using a regression model with key predictor variables (baseline BMI, age, fasting values, ethnicity and treatment) for each time point and outcome. Imputation was used to correct for verification bias (19).  To highlight the direction of significant interactions, modelling responses for insulin values were estimated in white European and South Asian males and females, aged 60, at BMI levels of 25kg/m 2 (normal), 30kg/m 2 (overweight) and 35kg/m 2 (obese).
All data were analysed using SPSS (version 24.0). A p-value of <0.05 was considered statistically significant for main effects and p<0.1 for interactions. Descriptive data are reported as mean (95% CI) in text and tables, unless otherwise stated.

Sensitivity Analyses
In order to aid interpretation and assess the robustness of the outcome, we investigated whether results were affected by removing the ARMING HEALTH participants (n=13), as this protocol did not involve a change in posture. Furthermore, to ascertain whether factors that were found to modify the treatment effect for postprandial responses were driven by higher control values (postprandial response during the sitting condition), we repeated the main analysis after further adjusting for the postprandial response to prolonged sitting (categorised as low, medium or high derived through tertiles).

Results
129 participants were included in this analysis.  The results for interactions are presented in Table 1. Figure 2a, 2b and Table, Supplemental Digital Content 4 display the stratified analysis for both insulin and glucose (see Table, Supplemental Digital Content 4, stratified analysis for insulin and glucose responses during each treatment condition, http://links.lww.com/MSS/B869).

Ethnicity
There was an ethnicity x treatment interaction for insulin (p=<0.001) but not glucose (p=0.354).

BMI
Interactions were seen for both insulin and glucose (both p=<0.001). For those with a BMI above the median split (≥27.2kg/m 2 ), the insulin response was reduced by 20.9mU/L (11.7mU/L,

Age
There was no age x treatment interaction for insulin (p=0.149) or glucose (p=0.811).

HOMA-IR
There was no HOMA-IR x treatment interaction for insulin (p=0.240) or glucose (p=0.549). Figure 3 and

Sensitivity Analyses
The significance levels were largely unaffected when the ARMING HEALTH study was removed from the analysis. These results are presented in In this analysis, females were also shown to derive the greatest metabolic benefit when breaking The current analysis has strengths and limitations. We were able to provide rigorous estimates of the postprandial responses to breaking prolonged sitting, by using data combined from four laboratory-based, randomised cross-over treatments that used the same experimental protocols.      .0)** 5.8 (5.5, 6.1) 5.9 (5.6, 6.2) 5.6 (5.3, 5.9)* BMI ≥27.2kg/m 2 91.6 (76.9, 106.