Replacing with whole grains and legumes reduces Lp-PLA 2 activities in plasma and PBMCs in patients with prediabetes or T2D 1

To determine dietary effects on circulating lipoprotein-associated phospholipase A 2 (Lp-PLA 2 ) activity and enzyme activity in peripheral blood mononuclear cells (PBMCs), 99 patients with impaired fasting glucose, impaired glucose tolerance, or newly-diagnosed T2D

On the sheet, subjects were instructed to record the amount of foods before ingestion and any remaining after ingestion by weighing the foods. During the 2 week run-in period, all participants were advised to continue their usual diet of refi ned rice. Baseline measurements were also performed at the start of the run-in phase. After the run-in period, the subjects in the control group maintained their usual diet of refi ned rice, and the subjects in the whole grain group replaced refi ned rice intake with 1/3 of legumes (black soybeans), 1/3 of barley, and 1/3 of brown rice three times per day as a carbohydrate source and increased vegetable intake to at least 6 units (30-70 g/unit) per day for suffi cient dietary fi ber intake . To check participants' compliance during the whole study period, the dietitian interviewed them during biweekly visits or by telephone. They were interviewed to discern whether they were following the program well, including dietary intake and weight changes. During the study period, all participants were also encouraged to maintain their usual lifestyle and were reminded not to make any changes to their usual dietary habits, except for the whole grain substitution in whole grain group.
Dietary energy values and nutrient content from the 3 day food records were calculated using a computer aided nutritional analysis program (CAN-pro 3.0; Korean Nutrition Society, Seoul, Korea). Total energy expenditure (kcal/day) was calculated from activity patterns including basal metabolic rate, physical activity for 24 h ( 10 ), and specifi c dynamic action of food. Basal metabolic rate for each subject was calculated with the Harris-Benedict equation ( 11 ).

Anthropometric parameters, blood pressure, and blood collection
Body weight and height were measured in the morning in unclothed subjects without shoes to calculate BMI (kg/m 2 ). Waist circumference was measured at the umbilical level with the subjects standing after normal expiration and the hip girth was measured at the widest part of the hip, and the waist and hip ratio (WHR) was calculated. Blood pressure (BP) was measured in the left arm of seated patients with an automatic BP monitor (TM-2654; A and D, Tokyo, Japan) after a 20 min rest. Study subjects were interviewed for smoking and drinking behavior during their visit. After a 12 h fasting period, venous blood specimens were collected in EDTA-treated and plain tubes, centrifuged to produce plasma or serum, and stored at Ϫ 70°C until analysis.

Oral glucose tolerance test, hemoglobin A 1c , and homeostasis model assessment-insulin resistance
All subjects drank a 75 g glucose solution after a 12 h overnight fast . At 0 and 12 weeks, venous specimens were subsequently collected at, before, and 30, 60, and 120 min after loading to determine the serum levels and response areas of glucose, insulin, and C-peptide according to dietary intervention. Glucose was measured by a glucose oxidase method using the Beckman glucose analyzer (Beckman Instruments, Irvine, CA). Insulin and C-peptide were measured by radioimmunoassay with commercial kits from Immuno Nucleo Corp. (Stillwater, MN). Each response of glucose, insulin, C-peptide, and FFA was calculated with the area under each response curve using a trapezoidal method. Hemoglobin A 1c (HbA 1c ) was measured by immunoturbidimetric analyzer using a turbidimeter. Insulin resistance was calculated from the homeostasis model assessment-insulin resistance (HOMA-IR) ( 12 ) using the following equation: HOMA-IR = [fasting insulin ( IU/ml) × fasting glucose (mmol/l)]/22.5.

Serum lipid profi le, FFAs, and serum high-sensitivity C reactive protein
Fasting serum concentrations of total-cholesterol and triglycerides were measured using commercially available kits on a Hitachi 7150 autoanalyzer (Hitachi Ltd., Tokyo, Japan). After the replacement of refi ned rice with whole grains and legumes has been suggested as a way to reduce the risk factors for T2D ( 7,8 ). Although increased whole grain consumption in healthy subjects for a 16 week period did not affect any of the biomarkers of CVD risk, including insulin sensitivity ( 9 ), there was, however, a decreasing effect on blood glucose and insulin by whole grains and legumes reported in diabetic patients ( 7,8 ). Therefore, we sought to determine the effects of dietary intervention (replacement of refi ned rice with whole grains and legumes and a high intake of vegetables) on circulating Lp-PLA 2 activity and enzyme activity in peripheral blood mononuclear cells (PBMCs) in patients with impaired fasting glucose (IFG), impaired glucose tolerance (IGT), or newly diagnosed T2D .

Subjects and study design
Study subjects were selected from participants in a clinical nutrition study conducted by the National Leading Research Laboratory of Clinical Nutrigenetics/Nutrigenomics at Yonsei University. Study subjects were recruited from the Health Service Center during routine check-up at National Health Insurance Corporation Ilsan Hospital, Goyang, Korea (January 2012-December 2012). Based on the data screened from the Health Service Center, subjects who had IFG (100 р fasting glucose < 126 mg/dl) or newly diagnosed T2D (fasting glucose у 126 mg/dl) were referred to the Department of Family Medicine or Internal Medicine. Subjects were excluded if they had any diagnosis of vascular disease, cancer (clinically or by anamnesis), renal disease, liver disease, thyroid disease, chronic alcoholism, and acute or chronic infl ammatory diseases; took antioxidants; were pregnant, breast feeding, or intending to become pregnant during the time of study; or were taking glucose-lowering, lipid-lowering, or antihypertensive medications. A total of 100 subjects were enrolled. The macronutrient composition of each subject's usual diet was that of a typical diet with cooked refi ned rice, which is consumed by a substantial number of Koreans. All subjects provided written informed consent before participation in this study, which was approved by the Institutional Review Board of Yonsei University and performed in accordance with the Declaration of Helsinki.
Independently performed computer randomization was used to allocate numbers and divide the patients into two groups. The present study was carried out in two phases; a 2 week run-in phase consisting of the usual diet with refi ned rice and a 12 week intervention phase consisting of the usual diet of refi ned rice or consumption of whole grains and legumes as a carbohydrate source. At the beginning, a total 120 subjects were screened for the study. Twenty subjects were excluded; 13 subjects were taking medications or supplements which related to hypertension, and 7 subjects were taking blood glucose and lipid lowering drugs. During the run-in period, one subject, not having maintained energy intake, dropped out. The remaining 99 were randomly subdivided into two groups and were assigned to consume either their usual diet (refi ned rice) or the whole grains and legumes as a carbohydrate source for three meals a day during the 12 week intervention.

Dietary education and assessment of dietary intake/ physical activity level
All subjects were given written and verbal instructions by a registered dietitian on completion of a 3 day (2 week days and 1 weekend day) dietary record every 2 weeks throughout the whole study.

Statistical analyses
Statistical analyses were performed using the SPSS version 12.0 for Windows (Statistical Package for the Social Sciences; SPSS Inc., Chicago, IL). A paired t -test was used to evaluate the effects of the dietary intervention. The primary endpoint was to assess changes of Lp-PLA 2 activities in plasma and PBMC between groups. One-way ANOVA with the Bonferroni correction or independent t -test was used to test the inter-group comparison. Pearson and partial correlation coeffi cients were used to determine the relationship between changes in Lp-PLA 2 activities in plasma and PBMC supernatants, and metabolic parameters. General linear model analysis was also used for the comparison with adjustment for confounding factors. Multiple linear regression analyses were performed to identify major factors associated with initial and changed variables, according to changes in Lp-PLA 2 activities in plasma and PBMC supernatants. For descriptive purposes, mean values are presented using untransformed values. Results are expressed as the mean ± standard error. A two-tailed P value of less than 0.05 was considered statistically signifi cant.

RESULTS
After the run-in period, 99 subjects who had maintained their energy intake participated in the intervention program and were randomly subdivided into the control group or the whole grain group. There were no signifi cant differences between control and whole grain groups in baseline characteristics, such as age (55.4 ± 1.5 years vs. 56.3 ± 1.2 years; P = 0.618), male/female distribution (35/14 vs. 32/18; P = 0.429), education level, smoking, and drinking (data not shown). Table 1 shows clinical characteristics, lipid profi les, infl ammatory markers, and nutrient intakes before and after dietary intervention. Frequent interviews with the subjects did not suggest any important changes in physical activity patterns of the individual subjects during the study. As expected, BMI, WHR, total energy expenditure, and total energy intake were similar before and after treatment in both groups. However, IL-6 levels in serum and cultured nonstimulated PBMC supernatants signifi cantly decreased from baseline in the whole grain and legume group. The changes (difference from baseline) in IL-6 levels in PBMC supernatants in the whole grain and legume group were signifi cantly different from those in the refi ned rice group after adjusting for baseline levels. Posttreatment IL-6 levels in PBMC supernatants in the whole grain and legume group were signifi cantly lower than those in the control group. With the substitution of whole grains and legumes for refi ned rice, energy percentages of protein (16.5 ± 0.1% vs. 20.4 ± 0.1%; P < 0.001) and fat (22.1 ± 0.1% vs. 24.9 ± 0.1%; P < 0.001) intake increased signifi cantly, however, those of carbohydrate (61.8 ± 0.2% vs. 54.7 ± 0.1%; P < 0.001) signifi cantly decreased in the whole grain and legume group. Compared with net differences of these macro nutrient parameters, percent energy intakes of protein, fat, and carbohydrate were signifi cantly different between the two groups (all P < 0.001), however, these signifi cant differences disappeared after adjustment for baseline precipitation of apoB-containing lipoproteins with dextran sulfate-magnesium, HDL-cholesterol concentrations in the supernatants were measured enzymatically. LDL-cholesterol was indirectly estimated in participants with serum triglyceride concentrations less than 400 mg/dl using the Friedewald formula: LDL-cholesterol = total-cholesterol Ϫ [HDL-cholesterol + (triglycerides/5)]. FFAs were analyzed with a Hitachi 7150 autoanalyzer (Hitachi Ltd.). Serum high-sensitivity C reactive protein (hs-CRP) concentrations were measured with an Express Plus TM autoanalyzer (Chiron Diagnostics Co., Walpole, MA) using a commercially available high-sensitivity CRP-Latex(II) X2 kit (Seiken Laboratories Ltd., Tokyo, Japan).

Cytokine secretion in nonstimulated PBMCs
Whole blood was mixed with the same volume of RPMI 1640 (Gibco, Invitrogen, Carlsbad, CA). The sample was then centrifuged at 2,000 rpm for 20 min at 10°C. After the separation, a thin layer of PBMCs was isolated and washed twice with RPMI 1640. The pellet was resuspended in RPMI 1640 supplemented with 10% fetal bovine serum (Gibco, Invitrogen) and seeded in 12-well plates (1 × 106 cells/ml; Nunc, Roskilde, Denmark) and incubated at 37°C with 5% CO 2 for 24 h. After 24 h incubation, supernatants were collected and stored at 80°C until interleukin (IL)-1 ␤ , IL-6, TNF-␣ , and Lp-PLA 2 activity levels were assayed ( 13,14 ).

Cytokine assay for IL-1 ␤ , IL-6, and TNF-␣ levels in serum and PBMC supernatants
Levels of IL-1 ␤ , IL-6, and TNF-␣ in serum and PBMC supernatants were measured using the Bio-PlexTM reagent kit on the Bio-PlexTM (Bio-Rad Laboratories, Hercules, CA) according to the manufacturer's instuctions.

Plasma oxidized LDL, LDL particle size, and Lp-PLA 2 activity
Plasma oxidized LDL (ox-LDL) was measured using an enzyme immunoassay (Mercodia, Uppsala, Sweden). The resulting color reaction was read at 450 nm on a Wallac Victor2 multilabel counter (Perkin Elmer Life Sciences, Turku, Finland). Particle size distribution of LDL (1.019-1.063 g/ml) isolated by sequential fl otation ultracentrifugation was examined by a pore gradient lipoprotein system (CBS Scientifi c, San Diego, CA) on commercially available nondenaturing polyacrylamide slab gels containing a linear gradient of 2-16% acrylamide (Alamo Gels Inc., San Antonio, TX). Standards of latex beads (34 nm), thyroglobulin (17 nm), apoferritin (12.2 nm), and catalase (10.4 nm) were used to estimate the relative migration rates of each band. Gels were scanned using a GS-800 calibrated imaging densitometer (Bio-Rad, Graz, Austria). The activity of Lp-PLA 2 in plasma and PBMC supernatants was measured by using a modifi cation of a previously described high-throughput radiometric activity assay ( 15 ).
increased, whereas those of the whole grain and legume group signifi cantly decreased. The ⌬ plasma MDA ( P < 0.001) and ⌬ urinary 8-epi-PGF 2 ␣ ( P = 0.031) in the whole grain and legume group were signifi cantly different from those in the control group after adjustment for the baseline levels. The whole grain and legume group showed lower levels of posttreatment plasma MDA ( P = 0.015) and a trend toward a decrease in posttreatment urinary 8-epi-PGF 2 ␣ ( P = 0.053) than the control group ( Table 2 ).

Glucose, insulin, C-peptide, and oxidative stress
After dietary intervention, fasting glucose and HbA 1c concentrations signifi cantly increased in the control group, whereas fasting glucose, insulin, HOMA-IR index, response areas of glucose, insulin, and C-peptide during oral glucose tolerance test and HbA 1c signifi cantly decreased in the whole grain and legume group ( Table 2 ). The change ( ⌬ ) in fasting glucose ( P < 0.001), ⌬ insulin, ⌬ HOMA-IR index, ⌬ glucose response area, and ⌬ C-peptide response area during oral glucose tolerance test and ⌬ HbA 1c in the whole grain and legume group were signifi cantly different from those in the control group before and after adjustment for the baseline levels. Posttreatment fasting glucose, HOMA-IR index, and glucose response area in the whole grain and legume group were signifi cantly lower than those in the control group. After 12 weeks of dietary intervention, plasma MDA concentrations of the control group signifi cantly Data are expressed as mean ± standard error. SFA, saturated fatty acid . a Tested by logarithmic transformation. b P < 0.05. c P < 0.01. d P < 0.001 compared with initial value in each group tested by paired t -test. e † P < 0.01 compared between two groups at 12 week follow-up. P -value compared with net differences between 2 groups tested by independent t -test. P a value compared with net differences between two groups after adjusting for baseline value. P = 0.002), ⌬ Lp-PLA 2 activity in PBMC supernatants ( r = 0.310, P = 0.002), ⌬ plasma ox-LDL ( r = 0.314, P = 0.002), ⌬ serum IL-6 ( r = 0.218, P = 0.034), and ⌬ urinary 8-epi-PGF 2 ␣ ( r = 0.346, P = 0.001) after adjustment for age, sex, BMI, smoking, and drinking. After adjusting for these confounding variables, ⌬ Lp-PLA 2 activity in PBMC supernatants correlated positively with ⌬ fasting glucose ( r = 0.386, P < 0.001), ⌬ plasma ox-LDL ( r = 0.502, P < 0.001), ⌬ serum IL-6 ( r = 0.233, P = 0.024), and ⌬ IL-6 in PBMC supernatants ( r = 0.269, P = 0.009), and negatively with ⌬ LDL particle size ( r = Ϫ 0.356, P < 0.001) and baseline Lp-PLA 2 activity in PBMC supernatants ( r = Ϫ 0.323, P = 0.002).

Relationship between changes in Lp-PLA 2 activities in plasma and PBMC supernatants and metabolic parameters
When all subjects were considered, ⌬ plasma Lp-PLA 2 activity positively correlated with ⌬ fasting glucose ( r = 0.318,  Fig. 1. Effects of whole grain and legume consumption on circulating Lp-PLA 2 activity (A), Lp-PLA 2 activity in peripheral blood molecular cells (B), plasma ox-LDL (C), and LDL particle size before (white bars) and after (black bars) 12 weeks of dietary treatment (D). Data are expressed as mean ± standard error. § Tested by logarithmic transformation. * P < 0.05, ** P < 0.01, *** P < 0.001 compared with baseline values in each group tested by paired t -test with the Wilcoxon signed rank test. P values derived from independent t -test with the Mann-Whitney U-test. P ′ , P value after adjusting for baseline value. † P < 0.01 compared between two groups at 12 week follow-up.
cell source for molecular studies in humans ( 20 ). The positive and independent relationship between change in PBMC Lp-PLA 2 and change in ox-LDL supports the previous fi nding of the direct effect of ox-LDL on the expression of Lp-PLA 2 in monocytes ( 21 ). Enhanced Lp-PLA 2 results in increasing levels of infl ammatory lysophosphatidylcholine which induces the expression of several infl ammatory cytokines. Inhibition of Lp-PLA 2 has been, however, reported to abolish ox-LDL-induced expression of infl ammatory genes, including IL-6 ( 22,23 ). Thus the signifi cant reductions in circulating and PBMC IL-6 concentrations might be related to the decrease in Lp-PLA 2 activity in the whole grain and legume group. Lp-PLA 2 hydrolyzes oxidized phospholipids in LDL particles and generates bioactive oxidized FFAs and lysophosphatidylcholines ( 24 ). Stafforini et al. ( 25 ) showed that the secreted form of Lp-PLA 2 released F 2 -isoprostanes, the end products of lipid oxidation, from the sn-2 position of phosphatidylcholines with high affi nity. Kono et al. ( 26 ) have reported that intracellular type II Lp-PLA 2 , which shares homology with the plasma enzyme Lp-PLA 2 , is involved in the metabolism of esterifi ed 8-iso-PGF 2 ␣ . Similar to previous fi ndings ( 3,27 ), this study also showed a positive correlation between changes in circulating Lp-PLA 2 activity and urinary 8-epi-PGF 2 ␣ concentrations. Changes in urinary 8-epi-PGF 2 ␣ concentrations were positively correlated with changes in fasting glucose, which were also strongly correlated with changes in plasma MDA, a marker of oxygen-derived free radicals. Similarly, recent studies of dietary treatment in patients with prediabetes or newly diagnosed T2D have shown that changes in MDA positively correlate with changes in fasting glucose and glucose response area during meal tolerance test ( 28 ).
The mechanisms that may link Lp-PLA 2 activity with diet in this study may be through substantial improvement of glucose and insulin metabolism by replacing refi ned rice with whole grains and legumes. Recently, a positive association between Lp-PLA 2 activity and insulin resistance ( 4 ), and a benefi cial effect of improving glycemic control on Lp-PLA 2 activity ( 18 ) have been reported. Alternatively, reduced Lp-PLA 2 activity could be a marker of increased consumption of protein relative to carbohydrate intake in the whole grain and legume group. Substitution of whole grains and legumes for refi ned rice in this study was associated with the replacement of 7% of energy from carbohydrates with energy from protein (about 4%) and fat (about 3%). Hatoum et al. ( 5 ) found that the replacement of 5% of energy from carbohydrates with energy from protein was associated with 2.2 nmol·min Ϫ 1 ·ml Ϫ 1 lower levels of Lp-PLA 2 activity, and this effect of protein intake on reduced Lp-PLA 2 activity unrelated to the protein's effect on lipid profi les. Additionally, whole grains, legumes, and vegetables contain many antioxidants, including vitamins, minerals, and phytochemicals ( 29,30 ). Antioxidants can slow the rate of oxidation of oxidizable substrates ( 29,31 ). In a randomized cross-design study, phytochemicals in soybeans were found to reduce lipid peroxidation in vivo and to increase the resistance of LDL to oxidation ( 32 ). Ox-LDL has been reported to induce Lp-PLA 2 expression in monocytes and macrophages in vitro plasma Lp-PLA 2 activity ( Table 3 ). Age, baseline BMI, baseline fasting glucose, ⌬ fasting glucose, ⌬ HOMA-IR index, ⌬ plasma ox-LDL, ⌬ LDL particle size, baseline plasma Lp-PLA 2 activity, baseline PBMC Lp-PLA 2 activity, and ⌬ PBMC Lp-PLA 2 activity were evaluated. The ⌬ fasting glucose was found to be an independent predictor of ⌬ plasma Lp-PLA 2 activity ( ␤ = 0.113 ± 0.040; P = 0.006) together with baseline plasma Lp-PLA 2 activity ( ␤ = Ϫ 1.004 ± 0.494; P = 0.045).

DISCUSSION
The 12 week intervention with the substitution of whole grains and legumes for refi ned rice was associated with substantial improvement in glucose and insulin metabolism as well as reduction in circulating and PBMC Lp-PLA 2 activities. The positive and independent association between change in plasma Lp-PLA 2 activity and change in fasting glucose suggests that improving glycemic control can reduce circulating Lp-PLA 2 activity without signifi cant changes in lipid profi les in patients with IFG, IGT, or newly-diagnosed T2D. Additionally, the fi nding of the positive correlation between changes in plasma and PBMC Lp-PLA 2 activities suggests that a reduction in PBMC Lp-PLA 2 activity could partly contribute to the reduction in circulating Lp-PLA 2 activity. Indeed, Lp-PLA 2 is known to be secreted from bone marrow-derived cells; mainly monocytes, macrophages, T lymphocytes, and mast cells ( 16,17 ).
Improving glycemic control in the whole grain and legume group resulted in a decrease in plasma ox-LDL and an increase in LDL particle size as well as reducing plasma and PBMC Lp-PLA 2 activities. Sanchez-Quesada et al. ( 18 ) reported that a 4 ± 2 month clinical trial with lifestyle counseling and pharmacologic hypoglycemic therapy in 42 patients with long-term T2D reduced Lp-PLA 2 activity and signifi cantly increased LDL particle size by 0.9%. The mechanism underlying reduced Lp-PLA 2 activity has been suggested to be related to the increase in LDL particle size because the amount of Lp-PLA 2 in small dense LDL is 5-10 times higher than in normal-sized LDL particles ( 19 ). Therefore, the extent of decrease in plasma Lp-PLA 2 activity in the whole grain and legume group of this study may depend on not only PBMC synthesis of Lp-PLA 2 , but also LDL particle size carrying this enzyme in circulation.
PBMCs are constituted mostly by circulating T and B lymphocytes and monocytes, which are considered a useful age, gender, and ethnicity, as well as the type of food, its structure, amount, and preparation. These are all important factors when interpreting metabolic responses, or lack of responses, to whole grain foods ( 41 ).
In summary, the present study showed signifi cant benefi cial effects from the substitution of whole grains and legumes for refi ned rice on circulating Lp-PLA 2 activity, enzyme activity in PBMCs, plasma ox-LDL, and LDL particle size in patients with IFG, IGT, or newly-diagnosed T2D. These effects are likely to substantially reduce the risk factors of CVD in patients with IFG, IGT, or newly-diagnosed T2D. The biological mechanism whereby whole grains and legumes may exert their reducing effects on circulating and PBMC Lp-PLA 2 is not clear in the present study, but it is likely to be due to multiple mechanisms, such as substantial improvement of glucose and insulin metabolism, increased consumption of protein relative to carbohydrates, and antioxidant-reduced lipid peroxides including ox-LDL. Therefore, grains should be consumed in a minimally refi ned form, and frequent consumption of vegetables and legumes should be recommended to reduce cardiovascular risk factors in patients with IFG, IGT, or newly-diagnosed T2D. ( 33 ), supporting the positive and independent correlation between changes in PBMC Lp-PLA 2 and plasma ox-LDL in this study.
There were limitations in this study. Several epidemiologic studies suggest that plasma Lp-PLA 2 is an independent predictor of CVD events in primary and secondary prevention (34)(35)(36)(37). A recent meta-analysis ( 36 ), which included 79,036 participants with or without coronary artery disease from 32 prospective studies, showed that Lp-PLA 2 activity and mass each had a continuous association with the risk for coronary artery disease; as similarly, the present study showed positive correlation between Lp-PLA 2 and CVD risk factors . On the other hand, Holmes et al. ( 38 ) suggested that reducing secretory phospholipase A 2 (sPLA 2 )-IIA mass is unlikely to be a useful therapeutic goal for preventing cardiovascular events , which is associated with increased risk of incident and recurrence of major vascular events. PLA2G2A had large and specifi c effects on circulating sPLA 2 -IIA mass and a small-to-modest effect on sPLA 2 enzyme activity ( 38 ). In addition, sPLA 2 inhibitor, called varespladib, was not a useful strategy to reduce adverse cardiovascular outcomes after acute coronary syndrome ( 39 ). These inconsistencies of PLA 2 families on CVD were emerged by causation and need to be further studied . An earlier study has shown that there were no signifi cant differences in any markers of CVD risk between groups with BMI >25 kg/m 2 ( 9 ). Also, Andersson et al. ( 40 ) did not fi nd a signifi cant effect of whole grain intake on insulin sensitivity, lipid peroxidation, infl ammatory markers, or on any of the metabolic variables studied. However, a 6 week intervention might be too short to achieve significant effects on insulin sensitivity ( 40 ). We believe that the disparity in these fi ndings from the previous observational studies and the present intervention study is caused by variations in subjects' health characteristics,