2019 年 66 巻 1 号 p. 89-105
Angiopoietin-like protein 8 (ANGPTL8) is a newly discovered adipokine plays an important role in energy homoeostasis, obesity and type 2 diabetes (T2D). Although lifestyle modification in obesity and T2D is known to offer metabolic benefits, there is paucity of comprehensive data on change in ANGPTL8. We investigated the effect of lifestyle intervention on ANGPTL8 concentrations. 384 obese/overweight adults with newly diagnosed T2D were randomly assigned (1:1:1) to diet (n = 128), diet + activity (n = 128) or usual care (control, n = 128) groups. All patients received usual care. Besides, the diet group received a calorie-restricted diet aiming for a weight loss of 5–10%. The diet + activity group additionally received a pedometer-based walking program. Primary outcome was change in ANGPTL8 concentration at 6 months. Data were analyzed according to intention-to-treat. From baseline to 6 months, the median ANGPTL8 level changed from 804.38 pg/mL to 792.86 pg/mL in control group. Compared with control, ANGPTL8 decreased with diet (baseline-adjusted between-group difference was –121.00 pg/mL, 95% CI –177.47 to –64.53; p < 0.0001) and diet + activity (–126.16 pg/mL, –181.21 to –71.11; p < 0.0001). There was no greater effect of diet + activity compared with diet (–5.16 pg/mL, –53.63 to 43.31; p = 0.8348). Both effects disappeared after adjusting for change in body fat, but did not differ significantly when adjusting for physical activity. A 6-month intervention inducing weight loss by a calorie-restricted diet or diet + activity, resulted in significant decrease on ANGPTL8 concentration. These effects were established by change in total body fat, and not by change in physical activity.
THE SPREAD OF TYPE 2 DIABETES (T2D) in world is strongly associated with the increasing prevalence of obesity due to the changing lifestyle and dietary habits [1]. Dietary and activity modifications can be effective for the treatment of obesity and T2D [1]. Although the precise mechanisms underlying the relationship of obesity, lifestyle intervention, weight loss and improving diabetic control remain unknown, alterations in hormone signalling, especially adipocyte-derived factors, may play an important role [2, 3].
Angiopoietin-like protein 8 (ANGPTL8), also known as betatrophin, lipasin, RIFL, C19orf80, or TD26, is a novel adipokine correlated with energy homoeostasis, as well as lipid and glucose metabolism [4]. Meta-analysis showed that ANGPTL8 concentrations were increased in obesity and T2D [5]. The serum levels of ANGPTL8 have also been described as a positive predictor of cardiovascular disease (CVD) in patients with T2D [6-8] and could potentially be implemented as a biomarker for T2D [9, 10]. However, there are also studies showing unaltered or even decreased levels of ANGPTL8 in T2D and obesity [11-16]. So the role of ANGPTL8 in obesity and T2D remains incompletely characterized and shows diversity.
Nonetheless, given the connection between ANGPTL8 and obesity, it is reasonable that ANGPTL8 would be regulated by weight loss [17]. Recent weight loss studies have evaluated the effects of bariatric surgery [16, 18-20], diet [21, 22], exercise [23], and diet plus exercise [13] on ANGPTL8 concentrations in diabetic and non-diabetic obese subjects. Unfortunately, contradictory results (increase [16, 19], decrease [18, 20, 22, 23] and not change [13, 21]) have also been reported. These results challenged the potential role of ANGPTL8 as a regulator of obesity and T2D.
The effect of weight loss in overweight/obese patients with newly diagnosed T2D by lifestyle intervention on ANGPTL8 levels has yet to be published. Our hypothesis was that a reduced calorie, weight-loss diet with physical activity (PA) would produce a greater decrease in ANGPTL8 level than diet alone and compared with usual care (control).
This was a single-center, parallel-group, assessor-blinded, 6-month randomized controlled trial (RCT) that took place in Region Xuzhou of China from May 2017 to April 2018, with participants allocated to usual care (control), diet, or diet + activity groups. The study was approved by the Ethics Committee at the First People’s Hospital of Xuzhou of eastern China and registered with the Chinese Clinical Trial Registry (ChiCTR-IOR-17011437, www.chictr.org.cn). All participants provided written informed consent before enrolment. Full trial protocol can be obtained from the corresponding author.
ParticipantsWe recruited patients by searching the records databases of the outpatient facilities at the Department of Endocrinology in the First People’s Hospital of Xuzhou, and of community-based education programmes, and by mass email. Patients were initially screened over the telephone by a research nurse, and potentially eligible subjects were given a face-to-face screening to verify eligibility at least 2 weeks prior to randomization (visit 1), where study nurses obtained informed consent, collected clinical information, carried out a physical examination, and prescribed a standardized diet (50–60% carbohydrate, 10–15% protein, and 20–30% fat (<7% saturated fat)) on the basis of the Dietary Guidelines for Chinese Resident [24]. Patients then returned to perform a one-mile walk test to assess physical fitness (visit 2) and again to remeasure values of baseline variables (visit 3). Between baseline and randomization all patients were seen by a research doctor and dietitian (visit 4).
Inclusion criteria were T2D diagnosed ≤8 months, Hemoglobin A1C (HbA1C) 6.5–9.0%, aged 18–70 years, body mass index (BMI) 24–40 kg/m2, and body weight ≤180 kg. Exclusion criteria included: blood pressure (BP) >160/100 mmHg, low density lipoprotein cholesterol (LDL-C) >4.0 mmol/L, cardiac diseases (a history of unstable angina or myocardial infarction within the previous 6 months or New York Heart Association class III or IV congestive heart failure), diabetic retinopathy, macroalbuminuria (urine albumin-creatinine ratio ≥300 mg/g) or nephropathy (plasma creatinine ≥130 μmol/L), inability to increase PA, mental disorders, pregnancy or planning for conception, insulin-dependence, use of weight-reducing drugs, and taking 3 or more glucose-lowering medications. Key withdrawal criteria included inability to adhere to study protocol, unwillingness to follow the study, or suffered a severe adverse event that prevented them for adhering to study protocols e.g. heart attack.
Randomization and blindingEligible patients were randomized by block randomization method with randomly mixed size (3, 6, or 9) sequenced blocks, to control, diet, or diet + activity groups in a 1:1:1 ratio. Randomization sequence was created by an independent biostatistician, with allocation placed in an opaque, consecutively numbered envelope, which were used in order. The patients, nurses and dietitians were not masked to the group allocation, but the doctors were. All assessments were completed solely by nurses.
ProceduresAll patients received usual care including medical counseling, T2D education, and lifestyle advice by the study nurses at baseline and at 6 months. Appointment times were offered between 0800 and 1900, enabling flexibility for patients who were working. The study doctor, who managed all glucose-lowering, lipid-lowering, and BP–lowering drugs, obtained all clinical variables from the study nurse. To minimize the risk of bias, any changes in drugs adjustments were made by the doctor unaware of treatment allocation according to a prespecified protocol to reach standardization across groups. In the 6-month study period, diabetes therapy was changed only when fasting plasma glucose (FPG) level >12.0 mmol/L, BP >160/90 mmHg, or when patients experienced hypoglycemic events at any visit.
Control group: The control group was instructed to maintain their habitual PA level and continue the standardized diet defined during visit 1, with reviews by the study nurse at baseline and at 3 and 6 months.
Diet group: Patients were asked to maintain their habitual levels of PA. In addition, patients in the diet group were prescribed a caloric deficit of 500 kcal per day compared with the standardized diet, aiming to achieve and maintain 5–10% weight loss throughout the study. This was embedded within a motivational interviewing style consultation to include assessing willingness to change, utilising decisional balance, reflective listening and open-ended questions, to identify needs, motivators and barriers to changing their diet, helping the patients develop and verbalise arguments for change and diminish resistance to it. Patients met with their study dietitian for 60 min at randomization and for 30 min at 3-month intervals for the subsequent 6 months. Reminders and reinforcement were scheduled by monthly group sessions (5–10 participants), in which the nurses spent 15 min discussing diet and another 15 min discussing other issues of concern to the patients. In addition, the nurses contacted patients by phone or email between monthly group sessions for monitoring and motivation.
Diet + activity group: Patients in the diet + activity group received the same dietary intervention as the diet group, but were also asked to undertake brisk walking for at least 30 min on at least 5 days each week over and above their habitual PA levels (according to the joint position statement of ACMS and ADA [25]). PA intensity was increased gradually within 5 weeks and remained at that level thereafter. Activities were recorded by patients on the PA log which has been used in other studies [26, 27] at the end of each day. The diet + activity group sessions were held separately from those of the diet group and nurses discussed with patients included 15 min of diet issues and 15 min of PA logs. All PA logs were checked by the nurses and reviewed with each participant to ensure completeness.
OutcomesThe primary outcome was the between-group difference in the change at 6 months in serum ANGPTL8 level. The secondary outcomes were between-group differences in the changes in weight, BMI, waist, body composition, BP, HbA1C, FPG, fasting insulin, lipid profile, PA, and drug use at 6 months. Additional secondary outcomes including liver fat content (LFC), liver and kidney functions, and high-sensitivity C-reactive protein (hs-CRP). Adverse events reports were recorded.
MeasurementsAll of the following parameters were measured at baseline and repeated 6 months later. PA of each participant was assessed by a pedometer (Yamax Digiwalker, SW-701; Yamax, Tokyo, Japan) for 7 consecutive days, after which the pedometer was returned to study nurse. For control and diet groups, the pedometer viewing windows were concealed with snap-on covers and tamper-proof seals to reduce the likelihood that patients would alter their behaviour in response to the step count value recorded [28]. Pedometer assessed step counts for PA intensity were: <5,000 steps per day for sedentary; 5,000–7,499 steps per day for low-active; 7,500–9,999 steps per day for somewhat active; 10,000–12,499 steps per day for active [29]. In addition to this direct assessment of daily steps, each subject also filled in a validated questionnaire for computation of self-reported total PA intensity score (work and leisure time), which was graded in ten levels (ranges from 1.4 to 2.3 and higher scores indicate higher PA intensity) [30].
Blood specimens were sampled from the antecubital vein after at least 10 h of overnight fasting. Participants refrained from vigorous exercise (24 h) prior to fasting venous blood collection at baseline and 6 months. Samples were directly centrifuged and stored at –80°C. After study completion, the samples were sent to the central laboratory of our hospital for analysis all at once. Multiple samples of each patient were analyzed in the same batch by an investigator blinded to group allocation to minimize batch-to-batch variation. Standard analytical techniques were used to measure HbA1C, FPG, fasting serum insulin, triglyceride (TG), total cholesterol (TC), LDL-C, high density lipoprotein cholesterol (HDL-C), hs-CRP, alanine aminotransferase (ALT), aspartate aminotransferase, and serum creatinine. The homeostatic model of assessment for insulin resistance (HOMA-IR) was calculated from FPG and fasting insulin levels. Circulating full-length ANGPTL8 concentration was determined using an ELISA (catalog number E11644h; Wuhan Eiaab Science, Wuhan, China) with an intra-assay coefficient of variation (CV) of ≤4.8% and an inter-assay CV of ≤7.2%. Anthropometrics including weight, height, waist, BMI and BP were measured according to standard procedures. Each anthropometric measurement was performed by a single well trained observer in the same room. Lean and fat body mass were measured using dual-energy x-ray absorptiometry (Lunar iDXA, Prodigy; GE Healthcare, Little Chalfont, UK). The LFC was assessed by trained ultrasonographers (who were unaware of the clinical details of the patients) using an improved quantitative ultrasound method, as described previously [31].
Statistical analysisThe sample size was determined based on demonstrating the statistical superiority of diet + activity intervention on ANGPTL8 level compared with diet alone. We estimated 36 subjects of each group were required to achieve 90% power, with an assumed intervention difference of 56 pg/mL after 6 months and an assumed standard deviation (SD) of 52 pg/mL for ANGPTL8. A dropout of 20% was taken into account, so at least 44 subjects of each group need to be recruited. The expected effect size and SD of ANGPTL8 were based on the published data in Asian patients with obesity and T2D [23].
Overall outcomes were analyzed according to the intention-to-treat principle. Subjects for whom follow-up values were not available were assumed to have maintained their baseline values (baseline-observation-carried-forward, BOCF). Imputations were not used to replace missing data in the primary analysis, but were included in a sensitivity analysis to assess missing data.
Because of multiple testing, p < 0.025 and < 0.05 were considered significant for the comparisons of both interventions vs. control, and diet + activity vs. diet, respectively. Multivariable linear regression model was applied to compare the diet and diet + activity interventions for primary outcome at 6 months, with adjustment for baseline covariates (age, sex, marriage, duration of diabetes, smoke, alcohol, glucose-lowering, lipid-lowering, and BP–lowering drugs, weight, BMI, systolic BP, diastolic BP, HbA1C, TC, TG, HDL-C, LDL -C, and ANGPTL8), use of glucose-lowering medication at 6 months, and change of HbA1C from baseline to 6 months. Three strategies were combined to select covariates for multivariable adjustment: 1) we adjusted for variables that, when added to this model, changed the regression coefficient of treatment effects by at least 10%; 2) we chose variables that were associated with ANGPTL8 at a level of p < 0.1 in univariate analysis; 3) the decisions as to which covariates to include were based on the evidences from previous trials on similar patient populations. Similar regression models were conducted to assess secondary comparisons of each intervention vs. control for the secondary outcomes at 6 months, with linear or logistic regression models dependent on outcome type. Use of drugs for the treatment of diabetes, hypertension, and lipid regulation were investigated with logistic regression (medication vs. no medication). Stratified analyses were conducted by including interaction in the following regression models: ANGPTL8 concentrations at 6 months by baseline values (sex, age, duration of diabetes, marriage, smoking, alcohol, total fat mass, daily step counts, HbA1C, TC, TG, HDL-C, LDL-C, HOMA-IR, LFC, and medicine use). Finally, if an intervention effect was observed, we explored whether change in weight loss (weight or total fat mass), or change in PA (daily step count or PA score) mediated these effects, by adding each, separately, as a covariate to the model.
Data were analyzed with the use of the statistical packages R (The R Foundation; http://www.r-project.org; version 3.4.3 2018–02-18) and EmpowerStats (www.empowerstats.com; X&Y Solutions Inc.).
A total of 702 subjects initially screened for inclusion, and, of these, 143 were excluded primarily due to having a duration of T2D >8 months and insulin treatment. Additionally, 136 subjects withdrew primarily due to geographical distance. Of the 384 patients enrolled in the study, 128 patients were allocated to each group randomly (Fig. 1). Baseline characteristics of the three groups were similar (Table 1). Additional baseline values of the outcome measures are listed in Tables 2–4. On average, patients were 52 years, had a BMI of 29.1 kg/m2, a diabetic duration of 172 days, an HbA1C of 7.8%, and a daily step counts of 5,769. At 6-month follow-up, complete data on anthropometrics were available for 365 patients; body composition for 364 patients; and pedometer for 368 patients. Blood samples of 365 patients were available (Fig. 1). No significant difference (p = 0.711) in retention rates amongst the groups (96.9% for control, 98.4% for diet group, and 97.7% for diet + activity group) was observed.
Trial profile
| Control (n = 128) | Diet (n = 128) | Diet + activity (n = 128) | |
|---|---|---|---|
| Mean (SD) | |||
| Age (years) | 50.5 (11.4) | 53.1 (10.8) | 51.4 (10.7) |
| Median (interquartile range) | |||
| Duration of diabetes (days) | 184 (162–201) | 186 (158–205) | 174 (145–201) |
| N (%) | |||
| Male | 66 (51.6%) | 62 (48.4%) | 65 (50.8%) |
| Married | 97 (75.8%) | 113 (88.3%) | 109 (85.2%) |
| Smoker | 46 (35.9%) | 40 (31.2%) | 35 (27.3%) |
| Alcohol | 41 (32.0%) | 36 (28.1%) | 39 (30.5%) |
| History of hypertension | 43 (33.6%) | 47 (36.7%) | 53 (41.4%) |
| Glucose-lowering medication | |||
| Metformin | 37 (28.9%) | 35 (27.3%) | 30 (23.4%) |
| Sulphonylurea | 7 (5.5%) | 9 (7.0%) | 13 (10.2%) |
| Thiazolidinedione | 4 (3.1%) | 6 (4.7%) | 3 (2.3%) |
| Alpha-glucosidase inhibitor | 28 (21.9%) | 27 (21.1%) | 21 (16.4%) |
| Other | 7 (5.5%) | 5 (3.9%) | 3 (2.3%) |
| Total | 56 (43.8%) | 57 (44.5%) | 54 (42.2%) |
| Blood pressure-lowering medication | |||
| Renin-angiotensin agent | 21 (16.4%) | 17 (16.4%) | 27 (21.%) |
| Angiotensin-receptor blockade | 10 (7.8%) | 4 (2.4%) | 4 (3.1%) |
| Calcium-channel blocker | 8 (6.2%) | 5 (3.9%) | 6 (4.7%) |
| Thiazide diuretic | 10 (7.8%) | 5 (3.1%) | 8 (6.2%) |
| β blocker | 6 (4.7%) | 8 (6.2%) | 11 (7.0%) |
| Other | 4 (3.1%) | 3 (2.3%) | 4 (2.3%) |
| Total | 29 (22.7%) | 23 (18.0%) | 32 (25.0%) |
| Lipid-lowering medication | |||
| Statin | 31 (24.2%) | 38 (29.7%) | 25 (19.5%) |
| Fibrate | 3 (2.3%) | 1 (0.8%) | 1 (0.8%) |
| Other | 4 (3.1%) | 2 (1.6%) | 2 (1.6%) |
| Total | 37 (28.9%) | 39 (30.5%) | 28 (21.9%) |
| Antiobesity agents | 0 | 0 | 0 |
| Baseline Median | 6 months Median | Change 6 months | % Change 6 months | Treatment effecta (95% CI): intervention vs. control | pb | Treatment effecta (95% CI): diet + activity vs. diet | pc | |
|---|---|---|---|---|---|---|---|---|
| ANGPTL8 (pg/mL) | ||||||||
| Control | 804.38 | 792.86 | –55.07 | –5.95 | ||||
| Diet | 827.91 | 629.19 | –159.73 | –16.77 | –121.00 (–177.47, –64.53) | <0.0001 | ||
| Diet + activity | 897.56 | 711.20 | –161.54 | –18.06 | –126.16 (–181.21, –71.11) | <0.0001 | –5.16 (–53.63, 43.31) | 0.8348 |
aTreatment effect: between-group difference, adjusted for the baseline covariates (age, sex, marriage, duration of diabetes, smoke, alcohol, glucose-lowering medication, blood pressure-lowering medication, lipid-lowering medication, weight, body mass index, liver fat content, daily step counts, hemoglobin A1C, blood lipid profiles, homeostatic model of assessment for insulin resistance, and ANGPTL8 concentrations), use of glucose-lowering medication at 6 months, and change of hemoglobin A1C from baseline to 6 months.
bp < 0.025 is considered significant for the comparison of both intervention groups vs. control.
cp < 0.05 is considered significant for the comparison of diet + activity vs. diet.
The average attendance and average total contact times between the two intervention groups were similar: 14.2 (SD 2.3) visits and 455.3 (SD 68.0) min for diet group and 14.6 (SD 2.4) visits and 467.8 (SD 71.8) min for diet + activity group. Patients in diet + activity group had significantly more PA and at higher intensities at 6 months than did those in the other two groups (Table 5). The diet + activity group increased their mean daily step counts from 5,836 (SD 2,364) to 7,252 (SD 2,401) at 6 months (Table 3). Like step counts, the mean self-reported PA intensity score in the diet + activity group had elevated from 1.55 (SD 0.12) to 1.63 (SD 0.13) at 6 months (Table 3). In addition, from baseline to 6-month, the proportion of patients with somewhat activity and activity in the diet + activity group was approximately 2.12–3.44 times greater than those in the other two groups (Table 5).
| Baseline | 6 months | Change 6 months | % Change 6 months | Treatment effecta (95% CI): intervention vs. control | pb | Treatment effecta (95% CI): diet + activity vs. diet | pc | |
|---|---|---|---|---|---|---|---|---|
| Mean | ||||||||
| Weigh (kg) | ||||||||
| Control | 82.8 | 82.7 | –0.24 | –0.18 | ||||
| Diet | 82.4 | 78.7 | –4.08 | –4.85 | –3.83 (–4.32, –3.33) | <0.0001 | ||
| Diet + activity | 83.0 | 79.1 | –4.31 | –4.95 | –3.99 (–4.48, –3.49) | <0.0001 | –0.16 (–0.65, 0.33) | 0.5237 |
| Total fat mass (kg) | ||||||||
| Control | 30.84 | 30.78 | –0.07 | –0.17 | ||||
| Diet | 30.83 | 27.67 | –3.26 | –10.57 | –3.05 (–3.27, –2.83) | <0.0001 | ||
| Diet + activity | 31.29 | 28.10 | –3.35 | –10.61 | –3.11 (–3.32, –2.89) | <0.0001 | –0.01 (–0.23, 0.21) | 0.9512 |
| Total lean mass (kg) | ||||||||
| Control | 44.78 | 44.77 | –0.13 | –0.17 | ||||
| Diet | 44.43 | 43.95 | –0.83 | –1.69 | –0.74 (–1.04, –0.45) | <0.0001 | ||
| Diet + activity | 44.54 | 43.82 | –0.94 | –1.78 | –0.92 (–1.21, –0.62) | <0.0001 | –0.17 (–0.47, 0.12) | 0.2498 |
| Waist (cm) | ||||||||
| Control | 99.1 | 98.4 | –0.68 | –0.61 | ||||
| Diet | 99.2 | 95.8 | –3.64 | –3.55 | –3.42 (–3.97, –2.87) | <0.0001 | ||
| Diet + activity | 99.9 | 95.3 | –4.79 | –4.63 | –3.52 (–4.08, –2.97) | <0.0001 | –0.10 (–0.65, 0.45) | 0.7171 |
| Systolic BP (mmHg) | ||||||||
| Control | 134 | 134 | 0.13 | 0.64 | ||||
| Diet | 135 | 134 | –0.83 | –0.11 | 1.01 (–0.94, 2.96) | 0.3109 | ||
| Diet + activity | 135 | 134 | –0.36 | 0.35 | 0.24 (–1.70, 2.17) | 0.8102 | –0.77 (–2.71, 1.17) | 0.4356 |
| Diastolic BP (mmHg) | ||||||||
| Control | 81 | 81 | –0.66 | –0.28 | ||||
| Diet | 82 | 81 | –1.17 | –0.56 | 0.56 (–1.15, 2.27) | 0.5220 | ||
| Diet + activity | 82 | 82 | –0.62 | 0.03 | 0.74 (–0.96, 2.44) | 0.3924 | 0.18 (–1.52, 1.89) | 0.8339 |
| Hemoglobin A1C (%) | ||||||||
| Control | 7.83 | 7.82 | 0.01 | 0.12 | ||||
| Diet | 7.85 | 7.61 | –0.24 | –3.10 | –0.25 (–0.30, –0.20) | <0.0001 | ||
| Diet + activity | 7.86 | 7.59 | –0.26 | –3.33 | –0.27 (–0.32, –0.22) | <0.0001 | –0.02 (–0.07, 0.03) | 0.4144 |
| Fasting glucose (mmol/L) | ||||||||
| Control | 7.89 | 7.92 | 0.02 | 0.41 | ||||
| Diet | 8.01 | 7.55 | –0.42 | –5.33 | –0.40 (–0.59, –0.22) | 0.0001 | ||
| Diet + activity | 7.95 | 7.52 | –0.46 | –5.76 | –0.41 (–0.59, –0.22) | <0.0001 | 0.00 (–0.19, 0.18) | 0.9869 |
| Total cholesterol (mmol/L) | ||||||||
| Control | 4.53 | 4.50 | –0.05 | –0.90 | ||||
| Diet | 4.73 | 4.68 | –0.04 | –0.95 | 0.01 (–0.04, 0.06) | 0.6100 | ||
| Diet + activity | 4.71 | 4.66 | –0.06 | –1.42 | –0.02 (–0.07, 0.03) | 0.4841 | –0.03 (–0.08, 0.02) | 0.2244 |
| Triglycerides (mmol/L) | ||||||||
| Control | 1.86 | 1.89 | 0.02 | 3.09 | ||||
| Diet | 1.86 | 1.78 | –0.08 | –1.81 | –0.11 (–0.22, 0.00) | 0.0463 | ||
| Diet + activity | 1.88 | 1.75 | –0.15 | –2.33 | –0.18 (–0.29, –0.08) | 0.0009 | –0.07 (–0.18, 0.03) | 0.1796 |
| HDL-C (mmol/L) | ||||||||
| Control | 1.20 | 1.19 | –0.02 | 2.02 | ||||
| Diet | 1.18 | 1.20 | 0.03 | 6.13 | 0.02 (–0.05, 0.10) | 0.5407 | ||
| Diet + activity | 1.19 | 1.27 | 0.08 | 12.26 | 0.09 (0.02, 0.17) | 0.0172 | 0.07 (–0.01, 0.14) | 0.0751 |
| LDL-C (mmol/L) | ||||||||
| Control | 2.54 | 2.52 | –0.04 | –0.66 | ||||
| Diet | 2.68 | 2.63 | –0.04 | –0.80 | –0.01 (–0.13, 0.11) | 0.9075 | ||
| Diet + activity | 2.69 | 2.63 | –0.09 | –1.77 | –0.03 (–0.15, 0.09) | 0.5844 | –0.03 (–0.14, 0.09) | 0.6638 |
| Steps per day | ||||||||
| Control | 5,820 | 5,969 | 103.12 | 3.30 | ||||
| Diet | 5,653 | 5,917 | 214.88 | 8.56 | 81.35 (–116.32, 279.01) | 0.4205 | ||
| Diet + activity | 5,836 | 7,252 | 1,371.07 | 27.50 | 1,259.03 (1,063.13, 1,454.94) | <0.0001 | 1,177.69 (982.55, 1,372.83) | <0.0001 |
| Physical activity score | ||||||||
| Control | 1.56 | 1.58 | 0.01 | 0.89 | ||||
| Diet | 1.56 | 1.58 | 0.02 | 1.53 | 0.01 (–0.01, 0.03) | 0.2923 | ||
| Diet + activity | 1.55 | 1.63 | 0.07 | 4.57 | 0.04 (0.02, 0.07) | 0.0001 | 0.03 (0.01, 0.05) | 0.0026 |
| Median | ||||||||
| Fasting insulin (mU/L) | ||||||||
| Control | 10.76 | 11.39 | –0.06 | 6.26 | ||||
| Diet | 12.05 | 8.36 | –3.00 | –18.77 | –3.30 (–3.85, –2.75) | <0.0001 | ||
| Diet + activity | 12.16 | 8.27 | –3.19 | –22.09 | –3.47 (–4.01, –2.92) | <0.0001 | –0.17 (–0.71, 0.38) | 0.5460 |
| HOMA-IR | ||||||||
| Control | 3.63 | 3.76 | –0.05 | –1.47 | ||||
| Diet | 4.27 | 2.92 | –1.33 | –25.82 | –1.32 (–1.49, –1.15) | <0.0001 | ||
| Diet + activity | 4.50 | 2.79 | –1.38 | –26.55 | –1.39 (–1.56, –1.22) | <0.0001 | –0.08 (–0.25, 0.10) | 0.3887 |
aTreatment effect: between-group difference, adjusted for the following baseline covariates: age, sex, marriage, duration of diabetes, smoke, alcohol, glucose-lowering medication, blood pressure-lowering medication, lipid-lowering medication, weight, body mass index, liver fat content, daily step counts, hemoglobin A1C, blood lipid profiles, HOMA-IR, and ANGPTL8 concentrations.
bp < 0.025 is considered significant for the comparison of both intervention groups vs. control.
cp < 0.05 is considered significant for the comparison of diet + activity vs. diet.
Abbreviations: HDL-C, high density lipoprotein cholesterol; LDL-C, low density lipoprotein cholesterol; HOMA-IR, homeostatic model of assessment for insulin resistance.
| Baseline | 6 months | Change 6 months | % Change 6 months | Treatment effecta (95% CI): intervention vs. control | pb | Treatment effecta (95% CI): diet + activity vs. diet | pc | |
|---|---|---|---|---|---|---|---|---|
| Mean | ||||||||
| Body mass index (kg/m2) | ||||||||
| Control | 29.0 | 29.0 | –0.07 | –0.18 | ||||
| Diet | 29.2 | 27.8 | –1.44 | –4.85 | –1.35 (–1.52, –1.18) | <0.0001 | ||
| Diet + activity | 29.1 | 27.7 | –1.47 | –4.95 | –1.37 (–1.54, –1.20) | <0.0001 | –0.02 (–0.19, 0.14) | 0.7857 |
| Median | ||||||||
| Liver fat content (%) | ||||||||
| Control | 36.8 | 36.9 | –0.08 | –0.20 | ||||
| Diet | 38.0 | 33.9 | –4.20 | –10.80 | –4.26 (–5.06, –3.46) | <0.0001 | ||
| Diet + activity | 37.6 | 32.2 | –5.49 | –12.15 | –4.38 (–5.17, –3.58) | <0.0001 | –0.12 (–0.91, 0.67) | 0.7739 |
| AST (U/L) | ||||||||
| Control | 22 | 22 | 0.00 | 0.00 | ||||
| Diet | 23 | 23 | –2.00 | –8.00 | –1.15 (–3.20, 0.91) | 0.2749 | ||
| Diet + activity | 24 | 24 | –1.50 | –5.61 | –0.22 (–2.27, 1.82) | 0.8311 | 0.92 (–1.11, 2.96) | 0.3751 |
| ALT (U/L) | ||||||||
| Control | 32 | 31 | 1.00 | 3.19 | ||||
| Diet | 31 | 26 | –6.00 | –14.43 | –6.77 (–11.69, –1.85) | 0.0074 | ||
| Diet + activity | 34 | 29 | –6.00 | –16.00 | –6.39 (–11.29, –1.49) | 0.0110 | 0.38 (–4.51, 5.26) | 0.8798 |
| Creatinine (μmol/L) | ||||||||
| Control | 54 | 53 | 1.00 | 1.85 | ||||
| Diet | 54 | 53 | 0.00 | –0.37 | –1.45 (–6.27, 3.38) | 0.5578 | ||
| Diet + activity | 56 | 56 | –2.00 | –3.05 | 0.28 (–4.52, 5.09) | 0.9082 | 1.73 (–3.06, 6.52) | 0.4799 |
| hs-CRP (mg/L) | ||||||||
| Control | 2.10 | 1.88 | –0.14 | –4.72 | ||||
| Diet | 2.12 | 1.59 | –0.30 | –15.45 | –0.29 (–0.96, 0.38) | 0.4035 | ||
| Diet + activity | 2.38 | 1.71 | –0.43 | –16.07 | –0.37 (–1.04, 0.31) | 0.2870 | –0.08 (–0.75, 0.59) | 0.8137 |
aTreatment effect: between-group difference, adjusted for the following baseline covariates: age, sex, marriage, duration of diabetes, smoke, alcohol, glucose-lowering medication, blood pressure-lowering medication, lipid-lowering medication, weight, body mass index, liver fat content, daily step counts, hemoglobin A1C, blood lipid profiles, homeostatic model of assessment for insulin resistance, and ANGPTL8 concentrations.
bp < 0.025 is considered significant for the comparison of both intervention groups vs. control.
cp < 0.05 is considered significant for the comparison of diet + activity vs. diet.
Abbreviations: AST, aspartate aminotransferase; ALT, alanine aminotransferase; hs-CRP, high-sensitivity C-reactive protein.
| No. (%) of patients | Odds ratio (95% CI) intervention vs. control | pa | Odds ratio (95% CI) diet + activity vs. diet | pb | ||
|---|---|---|---|---|---|---|
| Baseline | 6 months | |||||
| Sedentary (<5,000 steps per day) | ||||||
| Control | 44 (34.4%) | 38 (32.2%) | ||||
| Diet | 45 (35.2%) | 41 (33.1%) | 1.04 (0.61, 1.78) | 0.8865 | ||
| Diet + activity | 48 (37.5%) | 23 (19.0%) | 0.49 (0.27, 0.90) | 0.0204 | 0.48 (0.26, 0.86) | 0.0132 |
| Low active (5,000–7,499 steps per day) | ||||||
| Control | 52 (40.6%) | 51 (43.2%) | ||||
| Diet | 53 (41.4%) | 54 (43.5%) | 1.01 (0.61, 1.69) | 0.9590 | ||
| Diet + activity | 50 (39.1%) | 39 (32.2%) | 0.62 (0.37, 1.06) | 0.0804 | 0.62 (0.37, 1.04) | 0.0688 |
| Somewhat active (7,500–9,999 steps per day) | ||||||
| Control | 23 (18.0%) | 20 (16.9%) | ||||
| Diet | 22 (17.2%) | 23 (18.5%) | 1.12 (0.58, 2.16) | 0.7450 | ||
| Diet + activity | 19 (14.8%) | 41 (33.9%) | 2.51 (1.36, 4.63) | 0.0031 | 2.25 (1.25, 4.06) | 0.0069 |
| Active (≥10,000 steps per day) | ||||||
| Control | 9 (7.0%) | 9 (7.6%) | ||||
| Diet | 8 (6.2%) | 6 (4.8%) | 0.62 (0.21, 1.79) | 0.3724 | ||
| Diet + activity | 11 (8.6%) | 18 (14.9%) | 2.12 (0.91, 4.92) | 0.0818 | 3.44 (1.31, 8.98) | 0.0118 |
ap < 0.025 is considered significant for the comparison of both intervention groups vs. control.
bp < 0.05 is considered significant for the comparison of diet + activity vs. diet.
The primary outcome did not significantly differ between the two intervention groups (Table 2). However, ANGPTL8 concentrations were significantly lower at 6 months in patients who received either intervention than in those who received usual care (Table 2 and Fig. 2). Differences between groups of the secondary outcomes were similar to that of primary outcome, in which the diet and diet + activity groups were not substantially different, and outcomes with either diet or diet + activity were better than those with usual care for weight, waist, BMI, body composition, LFC, HbA1C, FPG, fasting insulin, TG, HDL-C, ALT, and HOMA-IR (Table 3, 4). There were no differences in the proportions using BP-lowering and lipid-lowing medications between interventions and control groups during the period of study, but a larger proportion of the usual care patients seem to increase the use of glucose-lowing medication than the intervention groups at 6 months (Table 6).
Box outlier plot presentation of serum ANGPTL8 levels. Note the significant decrease of ANGPTL8 levels due to diet and diet + activity interventions. Boxes span the 25th and 75th percentile; the whiskers span from 10th to 90th percentile; and the dots indicate the outliers. *p < 0.0001.
| Proportion of patients taking medication (%) | Odds ratioa (95% CI) intervention vs. control | pb | Odds ratioa (95% CI) diet + activity vs. diet | pc | ||
|---|---|---|---|---|---|---|
| Baseline | 6 months | |||||
| Glucose-lowering medication | ||||||
| Control | 43.8 | 50.4 | ||||
| Diet | 44.5 | 44.4 | 0.36 (0.15, 0.91) | 0.0308 | ||
| Diet + activity | 42.2 | 44.7 | 0.44 (0.19, 1.06) | 0.0668 | 1.21 (0.51, 2.88) | 0.6583 |
| Blood pressure-lowering medication | ||||||
| Control | 22.7 | 28.9 | ||||
| Diet | 18.0 | 25.0 | 0.97 (0.46, 2.04) | 0.9262 | ||
| Diet + activity | 25.0 | 35.0 | 1.49 (0.72, 3.07) | 0.2861 | 1.54 (0.75, 3.17) | 0.2423 |
| Lipid-lowering medication | ||||||
| Control | 28.9 | 35.5 | ||||
| Diet | 30.5 | 33.1 | 0.49 (0.18, 1.36) | 0.1724 | ||
| Diet + activity | 21.9 | 29.3 | 0.78 (0.30, 2.04) | 0.6100 | 1.59 (0.58, 4.38) | 0.3722 |
aAdjusted for the following baseline covariates: age, sex, marriage, duration of diabetes, smoke, alcohol, glucose-lowering medication, blood pressure-lowering medication, lipid-lowering medication, weight, body mass index, liver fat content, daily step counts, hemoglobin A1C, blood lipid profiles, homeostatic model of assessment for insulin resistance, and ANGPTL8 concentrations.
bp < 0.025 is considered significant for the comparison of both intervention groups vs. control.
cp < 0.05 is considered significant for the comparison of diet + activity vs. diet.
Stratified analyses according to baseline characteristics of patients are available in Table 7. The effects of interventions on ANGPTL8 concentrations were similar across most subgroup analyses, however, the effect of activity appeared to be more pronounced among patients who had high total fat mass than those with lower (p for interaction <0.0001).
| Subgroup | ANGPTL8 at 6 months, Median (IQR) | β (95% CI) pa | p for interaction | |
|---|---|---|---|---|
| Diet | Diet + activity | |||
| Sex | ||||
| Male | 663.58 (476.03–1,193.92) | 656.81 (416.12–1,060.68) | –5.35 (–77.61, 66.90) 0.8846 | 0.8783 |
| Female | 606.61 (359.07–1,081.39) | 742.69 (473.53–1,272.54) | 10.03 (–62.55, 82.60) 0.7867 | |
| Age (years) | ||||
| <45 | 622.25 (463.92–905.71) | 662.63 (525.41–1,297.64) | –18.14 (–118.67, 82.39) 0.7238 | 0.2922 |
| 45–57 | 526.32 (334.83–1,033.78) | 708.31 (388.02– 972.27) | 51.29 (–31.72, 134.3) 0.2267 | |
| ≥58 | 720.09 (457.34–1,357.12) | 742.69 (501.42–1,228.73) | –34.21 (–120.65, 52.24) 0.4385 | |
| Duration of diabetes (days) | ||||
| <164 | 663.23 (511.23–913.35) | 625.25 (442.23–900.98) | –53.02 (–139.45, 33.41) 0.2300 | 0.4398 |
| 164–195 | 716.25 (351.09–1,230.42) | 692.34 (417.09–1,077.87) | 27.4 (–67.67, 122.47) 0.5725 | |
| ≥196 | 559.53 (367.24–1,373.77) | 863.11 (542.84–1,251.76) | 37.59 (–48.19, 123.38) 0.3910 | |
| Married | ||||
| No | 613.80 (442.44–973.75) | 596.99 (409.93–1,097.47) | 82.00 (–56.53, 220.53) 0.2467 | 0.2562 |
| Yes | 635.77 (376.36–1,186.15) | 728.06 (474.69–1,154.02) | –9.57 (–64.55, 45.41) 0.7333 | |
| Smoking | ||||
| No | 626.04 (363.27–1,056.06) | 697.19 (443.21–1,203.07) | –5.53 (–66.08, 55.02) 0.8580 | 0.5260 |
| Yes | 682.76 (477.06–1,186.15) | 743.98 (443.32–1,063.23) | 19.16 (–76.36, 114.69) 0.6944 | |
| Alcohol | ||||
| No | 626.44 (367.24–1,197.43) | 734.36 (464.22–1,170.19) | –13.09 (–73.68, 47.51) 0.6723 | 0.2294 |
| Yes | 657.15 (440.93–951.84) | 637.59 (415.16–1,049.26) | 34.21 (–59.51, 127.93) 0.4748 | |
| Total fat mass (kg) | ||||
| <28.42 | 603.24 (367.24–744.49) | 569.45 (377.30–983.68) | 53.33 (–25.12, 131.78) 0.1836 | <0.0001 |
| 28.42–32.98 | 626.84 (450.91–936.21) | 716.42 (549.99–1,001.32) | –31.83 (–112.05, 48.39) 0.4372 | |
| ≥32.99 | 739.18 (369.13–1,599.54) | 864.52 (426.41–1,475.03) | –56.15 (–130.26, 17.96) 0.1384 | |
| Daily step counts | ||||
| <4,563 | 782.65 (532.64–1,440.59) | 709.41 (525.92–1,051.42) | –48.90 (–136.48, 38.69) 0.2746 | 0.6104 |
| 4,563–6,347 | 621.81 (434.88–896.73) | 771.39 (518.25–1,426.39) | 36.90 (–51.84, 125.64) 0.4156 | |
| ≥6,348 | 498.11 (333.88–790.89) | 589.88 (376.14–983.68) | 12.77 (–76.21, 101.76) 0.7786 | |
| Hemoglobin A1C (%) | ||||
| <7.4 | 492.83 (310.43–716.25) | 626.66 (434.38–785.30) | 75.06 (–27.69, 177.82) 0.1531 | 0.1446 |
| 7.4–8.2 | 606.61 (373.99–826.12) | 580.81 (380.91–1,096.64) | –26.02 (–105.02, 52.97) 0.5189 | |
| ≥8.3 | 1,137.99 (533.35–1,966.19) | 987.30 (574.27–1,325.09) | –0.01 (–84.17, 84.16) 0.9999 | |
| Total cholesterol (mmol/L) | ||||
| <4.31 | 601.54 (371.27–836.20) | 725.13 (381.25–1,192.30) | 58.30 (–35.2, 151.80) 0.2224 | 0.3246 |
| 4.31–4.97 | 580.51 (367.05–873.07) | 721.64 (574.02–1,075.55) | –6.27 (–89.45, 76.91) 0.8827 | |
| ≥4.98 | 737.35 (472.94–1,658.81) | 696.09 (368.87–1,284.62) | –50.85 (–136.68, 34.99) 0.2464 | |
| Triglycerides (mmol/L) | ||||
| <1.22 | 568.73 (364.81–779.08) | 742.69 (569.45–1,033.25) | 52.31 (–39.71, 144.33) 0.2659 | 0.4746 |
| 1.22–1.87 | 582.46 (370.13–884.33) | 566.93 (376.42–844.03) | –31.14 (–115.25, 52.97) 0.4686 | |
| ≥1.88 | 815.89 (573.56–1,449.89) | 863.11 (443.21–1,553.34) | –7.37 (–97.11, 82.36) 0.8721 | |
| HDL-C (mmol/L) | ||||
| <0.97 | 716.51 (447.89–1,379.32) | 794.04 (372.85–1,296.75) | –76.14 (–165.26, 12.99) 0.0949 | 0.1988 |
| 0.97–1.26 | 635.77 (376.36–1,062.52) | 730.10 (566.93–1,152.94) | 41.15 (–46.25, 128.55) 0.3567 | |
| ≥1.27 | 528.24 (353.07–851.60) | 587.84 (378.85–860.00) | 35.15 (–55.49, 125.79) 0.4477 | |
| LDL-C (mmol/L) | ||||
| <2.35 | 584.41 (351.09–818.35) | 578.40 (380.09–972.27) | 11.38 (–84.85, 107.61) 0.8168 | 0.9352 |
| 2.35–2.94 | 559.37 (339.83–906.89) | 676.03 (439.17–1,042.67) | 17.27 (–69.80, 104.34) 0.6977 | |
| ≥2.95 | 737.35 (516.69–1,579.78) | 763.82 (548.56–1,274.56) | –31.80 (–113.48, 49.89) 0.4460 | |
| HOMA-IR | ||||
| <2.73 | 559.53 (362.39–696.84) | 587.84 (396.88–859.06) | 67.12 (–22.94, 157.18) 0.1450 | 0.5145 |
| 2.73–5.82 | 610.45 (413.38–1,022.45) | 768.10 (550.41–1,128.83) | –3.65 (–94.67, 87.36) 0.9373 | |
| ≥5.83 | 939.75 (482.88–2,011.31) | 758.47 (404.17–1,369.73) | –38.10 (–123.04, 46.85) 0.3800 | |
| Blood pressure-lowering medication | ||||
| No | 633.66 (385.33–1,174.11) | 692.34 (415.16–1,037.63) | 20.08 (–36.89, 77.05) 0.4901 | 0.4488 |
| Yes | 593.15 (347.06–863.63) | 939.52 (474.69–1,396.92) | –25.07 (–134.95, 84.81) 0.6550 | |
| Lipid-lowering medication | ||||
| No | 633.66 (401.03–1,174.11) | 721.64 (486.83–1,154.89) | 15.72 (–44.03, 75.47) 0.6065 | 0.4838 |
| Yes | 615.04 (350.57–971.44) | 613.42 (381.25–1,063.17) | –36.81 (–137.13, 63.52) 0.4726 | |
| Glucose-lowering medication | ||||
| No | 613.80 (365.51–926.43) | 696.09 (491.17–1,152.99) | 2.19 (–64.4, 68.78) 0.9487 | 0.7012 |
| Yes | 703.95 (436.87–1,360.04) | 742.69 (397.52–1,129.78) | 0.71 (–77.4, 78.81) 0.9859 | |
aAdjusted for the baseline ANGPTL8 concentrations.
Each continuous variable was categorized into 3 subgroups by using tertiles.
Abbreviations: HDL-C, high density lipoprotein cholesterol; LDL-C, low density lipoprotein cholesterol; HOMA-IR, homeostatic model of assessment for insulin resistance.
Because the reduced use of glucose-lowering medication from baseline to 6 months was higher in the intervention groups compared with the usual care group (Table 6), we also did subgroup analyses by including interaction terms in the following regression models: ANGPTL8 levels at 6 months by use of metformin at 6 month, by use of sulphonylurea at 6 month, by use of thiazolidinedione at 6 month, and by use of alpha-glucosidase inhibitor at 6 month, respectively. The results showed that the glucose-lowering medication did not play an interactive role in the association between interventions and ANGPTL8 levels (Supplementary Table 1).
Compared to diet alone, diet + activity intervention led to a more significant reduction in the BMI (between-group difference was –0.36 kg/m2, 95% CI –0.64 to 0.09; p = 0.0112), HbA1C (–0.09%, 95% CI –0.15 to –0.02; p = 0.0108), and HOMA-IR (–0.47, 95% CI –0.75 to –0.19; p = 0.0011) in patients with high baseline values than those with low baseline values.
After adjusting for change in weight or total fat mass, intervention effects on change in ANGPTL8 both attenuated and lost statistical significance (Table 8). We observed no impact of change in PA on ANGPTL8.
| Model | Treatment effecta (95% CI): intervention vs. control | pb | Treatment effecta (95% CI): diet + activity vs. diet | pc |
|---|---|---|---|---|
| Change in ANGPTL8 | ||||
| Model 1d: ∆ in weight | ||||
| Control | ||||
| Diet | –12.13 (–71.90, 47.64) | 0.6909 | ||
| Diet + activity | –18.03 (–76.56, 40.50) | 0.5464 | –5.90 (–51.01, 39.22) | 0.7979 |
| Model 2d: ∆ in total fat mass | ||||
| Control | ||||
| Diet | –6.85 (–66.65, 52.95) | 0.8225 | ||
| Diet + activity | –9.37 (–68.50, 49.77) | 0.7564 | –2.52 (–47.43, 42.39) | 0.9126 |
| Model 3d: ∆ in daily step counts | ||||
| Control | ||||
| Diet | –147.10 (–206.77, –87.43) | <0.0001 | ||
| Diet + activity | –157.92 (–207.64, –108.21) | <0.0001 | –10.82 (–69.13, 47.48) | 0.7162 |
| Model 4d: ∆ in physical activity scores | ||||
| Control | ||||
| Diet | –150.14 (–203.87, –96.41) | <0.0001 | ||
| Diet + activity | –157.94 (–207.35, –108.52) | <0.0001 | –7.80 (–59.48, 43.89) | 0.7677 |
aTreatment effect: between-group differences.
bp < 0.025 is considered significant for the comparison of both intervention groups versus control.
cp < 0.05 is considered significant for the comparison of diet + activity versus diet.
dEach model was adjusted for the following baseline covariates: age, sex, marriage, duration of diabetes, smoke, alcohol, glucose-lowering medication, blood pressure-lowering medication, lipid-lowering medication, weight, body mass index, liver fat content, daily step counts, hemoglobin A1C, blood lipid profiles, homeostatic model of assessment for insulin resistance, and ANGPTL8 concentrations.
There were no statistically significant differences in the incidence of adverse events between three groups (Supplementary Table 2).
Sensitivity analyses for the comparison of three groups of the primary and secondary outcomes using LOCF method confirmed robustness of the primary analysis (Supplementary Table 3–7).
ANGPTL8, a novel secreted protein with 198 amino acids, is secreted predominantly by the liver and adipose tissue [32-34], and participates in the regulation of multiple physiology processes, such as energy expenditure, thermogenesis, insulin resistance, inflammation, pregnancy, lipid metabolism, and glucose homeostasis [5, 17, 32-36]. In animal studies, nutritional intake, lipogenesis, obesity, and insulin resistance increase ANGPTL8 mRNA expression in liver and fat tissue, whereas energy restriction and lipolysis suppress the ANGPTL8 transcript level [34, 35]. In human, serum ANGPTL8 levels are increased in obesity and T2D and exert a dual role in glucose and lipids metabolism, thereby establishing it as a potential therapeutic target in the treatment of obesity and T2D [4, 5, 17]. Therefore, the role and regulation of ANGPTL8, particularly in relation to weight loss, have been highlighted recently. However, the relative contributions of caloric restriction and increased PA to weight-loss-induced changes in ANGPTL8 production remain inconsistent. Our main finding was that in overweight/obese patients with newly diagnosed T2D, weight loss, achieved by a hypocaloric diet alone or by diet + activity, resulted in significant reduction in ANGPTL8 concentrations in conjunction with improved metabolic profile compared to usual care arm. The effects on ANGPTL8 appear to be mediated by weight loss, and not by PA.
This study is, to our knowledge, the first RCT of lifestyle intervention in overweight/obese patients with newly diagnosed T2D to report results concerning ANGPTL8 concentrations as an outcome. Our finding on ANGPTL8 is in line with a previous trial [22], in which 57 overweight/obese subjects were randomized to a reduced carbohydrate or reduced fat hypocaloric diets. In that study, both diet groups lost weight together with reduction in ANGPTL8 concentration over 12 weeks.
There are also a few studies in obese subjects evaluated the effects of lifestyle modifications on ANGPTL8 concentrations with interventions which included dietary, exercise, or combined that ranged from 3 months to 12 months in duration [13, 21, 23]. A short-term (3 months) exercise program demonstrated that after a combination of both moderate intensity of aerobic and resistance training, ANGPTL8 levels decreased statistically significant in 62 obese subjects compared with 82 non-obese subjects [23]. However, we find no additional effect when PA was given together with diet. There are several reasons might explain this phenomenon. First, that study did not appear to take into account the influence of diet on serum ANGPTL8 levels. Second, the contribution of exercise has been more difficult to understand, and findings on the effects of increased PA on serum adipokine levels continue to be debated [37]. A few meta-analyses found that there were no statistically significant changes in certain circulating levels of adipokine, such as adiponectin or resistin, in the intervention group with and without exercise [37, 38]. Divergent forms of exercise may have varying effects on adipokine levels [39]. In our study, patients in the diet + activity group increased their step count by about 1,371 steps per day. An increase in PA of 1,371 steps per day would equate to walking about 0.62 mile for the equivalent of an energy expenditure of 50 kcal per day [40]. This modest change in PA may be insufficient or the activity performed even incorrect, because a review revealed that vigorous aerobic activity provided greater effect [39] and a review showed greater effect of aerobic + resistance training over aerobic alone when considering adipokine change [41]. However, the present pragmatic trial was a real-world evaluation of a community-based PA advice in addition to diet. Therefore, our results show the effect of this additional advice, not the effect of PA itself, and this advice led to a statistically significant but modest increase in PA. Third, the modification of two behaviors at the same time may weaken the effects of both. People tend to eat more when they exercise and have a difficult time exercising enough to offset the additional caloric intake [42]. Nevertheless, subjects in the diet + activity group were still active enough, which is indicated by that the diet + activity worked better than diet alone in subjects who had high baseline BMI, HbA1C, and HOMA-IR than in those with lower values. This suggests that more severely affected patients may experience a more obvious PA response. Therefore, the timing of activity intervention may be too early in the course of diabetes to show additional effect on ANGPTL8. In obese patients with T2D, metabolic surgery can reduce ANGPTL8 concentrations after one year follow-up [18]. So if the activity intervention in our study may last for a year or longer, there would be a substantial reduction in the serum levels of ANGPTL8. However, the reduction would depend on whether the intensity of activity could be maintained at no less than the current level. The question should be addressed by continued follow-up of the study participants. Fourth, limitations of exercise self-report, in addition to concomitant extent of weight loss in that study, may also have contributed conflicting results.
Another small-scale 6-month RCT randomized 95 obese patients with metabolic syndrome to isocaloric energy-restricted diets with different levels of proteins (AHA and RESMENA groups, derived 15% and 30% of their energy from protein, respectively) [21]. Evaluations of ANGPTL8 concentrations in all patients together following the nutritional treatments revealed no statistically significant difference after the energy restriction [21], however, the AHA group exhibited significantly higher ANGPTL8 levels than the RESMENA group after the intervention [21]. The authors concluded that the changes of ANGPTL8 seem to be related to dietary regime rather than calorie restriction [21]. In our current study, at 6 months, ANGPTL8 levels decreased remarkably in the dietary caloric restriction group compared to before intervention and to the control group. These conflicting results raise the question as to whether the effects of diet are mediated directly by calorie restriction or via dietary structure. In fact, energy homeostasis plays a critical role in the expression of ANGPTL8. In high-fat diet mice, the expression of ANGPTL8 mRNA in both the liver and adipose tissue was dramatically induced by refeeding, while was strongly suppressed by fasting. The regulations of ANGPTL8 mRNA expression and protein secretion were elevated in response to refeeding following a food deprivation period in mice [32, 33, 43]. In human, ANGPTL8 genes were down-regulated by caloric restriction and regulated in the opposite direction by refeeding in obesity [44]. Serum ANGPTL8 levels were increased at 2 h after a defined meal in non-diabetic subjects [45]. Therefore, current evidences seem to favor caloric restriction for changes of serum ANGPTL8 levels, but because of our study design, no consistent conclusions regarding the specific diet are available at this time. In future additional studies are needed to demonstrate the relationship between macronutrient components and serum ANGPTL8 levels.
Result from a 12-month longitudinal analysis of 69 obese children and 20 normal weight children specifically designed to test the effect of lifestyle intervention through combined exercise, nutrition education, and behavior therapy on ANGPTL8 concentration also did not find changes of ANGPTL8 concentrations [13]. The reason for the different intervention response in adults compared with children is not clear, but it may be owing to the different pathophysiological features (adulthood vs. puberty), and relatively higher degree of chronicity of obesity compared with children [13]. In addition, these varying results between previous and our present study may be partly due to the potential technical issues with the detection of ANGPTL8 levels. ANGPTL8 exist in two different molecular forms in the circulation: Full length or total species (including both the full length protein and C-terminal fragments). There are two most used commercially available ELISA kits (EIAAB and Phoenix kits) that can detect human full length ANGPTL8 and total species ANGPTL8, respectively. Specific comparison of two kits for measuring ANGPTL8 in blood showed that serum ANGPTL8 levels in overweight and obese can be higher (full length) or lower (total species) when compared with lean subjects, respectively [46]. Both kits were accurate, reliable and repeatable in detecting ANGPTL8, different results were likely due to the complicated proteolytic regulation of ANGPTL8 in vivo [46]. Interestingly, in patients with metabolic syndrome, although serum full length ANGPTL8 levels were significantly higher compared to control, but when testing total species ANGPTL8, there was no significant difference between the two groups [47]. Additionally, previous study also showed that both full length and total species ANGPTL8 in obesity were increased [23]. These contradictory findings suggest that future we should use both of these kits simultaneously to fully explain our findings.
Adipose tissue is one of the principal organs of involved in the regulation of ANGPTL8 [4, 7, 22, 32-34]. ANGPTL8 mRNA was abundantly expressed in human adipose tissue [4, 7, 22], and the serum levels of ANGPTL8 were found to be increased in individuals with obesity and T2D and positively correlated with waist and BMI [21, 45, 48]. The increase in ANGPTL8 levels can be reduced in obese or obese T2D subjects who lost significant adipose tissue after bariatric surgery [18, 22]. We found in the intervention groups the observed intervention effects on ANGPTL8 were attenuated markedly and disappeared after adjustment for change in total body fat. Our subgroup analysis further suggested that ANGPTL8 concentration reduction was more evident among patients whose baseline total fat mass were higher than those with lower. Therefore, it seems that the reduction in fat mass following weight loss decreases the ANGPTL8 production in adipose tissue. Overall, our results and the results from other trials suggest that weight loss and/or reduction in body fat might be responsible for the decreased ANGLTL8 concentrations.
ANGPTL8 is a new atypical ANGPTL family member lacking a common feature shared by ANGPTL1–7 structures [32]. Studies on rodents have demonstrated that ANGPTL8 is a crucial regulator of lipid metabolism. ANGPTL8 overexpression increases serum TG levels by inhibiting lipoprotein lipase activity, while ANGPTL8 deficiency results in decreased TG levels due to reduced adipogenesis [33, 34]. The regulation of ANGPTL8 on TG metabolism is ANGPTL3-dependent and co-expression of ANGPTL8 with ANGPTL3 associated with further elevated serum TG level [32]. Study revealed that ANGPTL8 also regulates lipid autophagy by inducing catabolism of lipid droplets in hepatocytes [35]. Furthermore, ANGPTL8 could activate ERK signaling pathway to stimulate the expression of Egr1, which in turn lead to down-regulation of adipose triglyceride lipase and subsequently lipid accumulation in adipose tissue [49]. This provides another insight of regulation of lipid metabolism by ANGPTL8 protein. But even so, the mechanisms of ANGPTL8 in lipid regulation remains to be further elucidated, because previous studies of T2D [12, 36, 45, 50], including ours [51], have shown that serum ANGPTL8 levels were associated [12, 36] or not associated [45, 50, 51] with TG concentrations. In the current study, we found that the mean serum TG levels decreased by 1.81% at 6 months (p = 0.0923) in the diet group and by 2.33% in the diet + activity group (p < 0.0001). Differences between groups for the TG concentrations were similar to ANGPTL8, in that the diet group and diet + activity group did not differ substantially, and TG levels in both were better than those in the control group (Table 3). To explore the link between ANGPTL8 and TG, we further performed a stratified analysis according to serum ANGPTL8 levels (Supplementary Table 8). Although both the ANGPTL8 levels and the TG levels tend to change in the same direction, however, there were no significant interactions between ANGPTL8 levels and interventions with respect to TG levels, suggesting that the decline of TG level might not be associated with a decreasing ANGPTL8 concentration. However, the interpretation of this conclusion should be kept cautious. In fact, a nonlinear relationship between baseline ANGPTL8 and baseline TG was found in the current study (Supplementary Fig. 1). The correlation between baseline ANGPTL8 and baseline TG disappeared with the baseline ANGPTL8 level down to 718.67 pg/mL or up to 2,195.33 pg/mL. That is to say, there is a threshold effect on the relationship between ANGPTL8 and TG (Supplementary Table 9). In the present study, the median ANGPTL8 levels after intervention were 629.19 pg/mL and 711.20 pg/mL, respectively. Both of them were less than 718.67 pg/mL. These values are similar to those of studies [50, 51] which also found no correlation between ANGPTL8 and TG. Here we show that correlation between ANGPTL8 and TG occurred within a relatively narrow ANGPTL8 range (718.67–2,195.33 pg/mL). Taken together, more researches are needed to determine the physiopathological mechanism of ANGPTL8 in lipid metabolism and its relevance to other metabolic pathways.
ANGPTL8 seem to be able to increase the risk of CVD [6-8]. A cross-sectional study highlighted the atherogenic nature of ANGPTL8 and its potential contributions to increased CVD risks [7]. Another cross-sectional study demonstrated that patients with diabetes and CVD had significantly higher ANGPTL8 levels compared with those without CVD [8]. Also another cross-sectional study established the association between elevated ANGPTL8 and increased risk of CVD [6]. Although our subgroup analysis have shown that decreased ANGPTL8 concentration was not interaction with an increased level of HDL-C (Table 7), the trend was borderline insignificant (p for interaction trend test = 0.0804). Our decreases in ANGPTL8, therefore, imply that CVD risk can be positively influenced by a modest amount of weight loss. However, it is difficult to make conclusion to link between reduced ANGPTL8 concentration and favorable prognosis of CVD based on the findings from this study. Therefore, it is necessary to carry out a RCT with the CVD morbidity or mortality as primary outcome and other biomarkers to verify the role of ANGPTL8 on progress of CVD.
Our study has several limitations. Firstly, only patients whose duration of T2D was less than 8 months were included. Thus, the inclusion criteria may limit generalizability, but on the other hand evaluating newly diagnosed T2D patients has the advantage of avoiding more potential confusion with other diseases and medications. Second, the diet + activity arm included several lifestyle elements, which challenges the interpretation of individual effects of each intervention component. Third, the non-supervised PA intervention in this study might be inclined to biases and limitations. However, our trial was originally designed to test the effect from giving advice on PA in addition to diet. Thus, these results reflected the real world. Fourth, to avoid the influence of interactions on lifestyle changes and medications, only a limited number of medications were used. Future studies in individuals with T2D should focus on the other combinations of glucose-lowering medications.
In summary, among obese adults with T2D diagnosed for less than 8 months, a dietary modification resulted in significant reduction of ANGPTL8 concentration vs. usual care following 6 months. The additional PA did not promote further significant decline. Weight and total fat mass appeared to be mediators in this association. Further research is needed to assess generalizability and durability of findings.
This work was supported by the project of Xuzhou social development project in 2016 (KC16SL133), the project of clinical science and technology development fund of Jiangsu University (JLY20160128), grants from the National Natural Science Foundation of China (81570721, 81370965), the Natural Science Foundation of Jiangsu Province, China (BK20151331), the High Caliber Medical Personnel Foundation of Jiangsu Province (LGY2016053), Six talent peaks project in Jiangsu Province (2015-WSN-006), and the Science and Technology Commission of Zhenjiang City, Jiangsu Province (SH2015028).
None of the authors have any potential conflicts of interest associated with this research.
