Association of Lipopolysaccharide-Binding Protein With Aging-Related Adiposity Change and Prediabetes Among African Ancestry Men

Diabetes and obesity are associated with low-level, chronic inflammation (1). In recent years, the gut microbiota have come to be recognized as a contributor to this inflammation (2,3). Gram-negative bacteria contain lipopolysaccharide (LPS) in their outer membranes (4), and through their life cycles the bacteria can shed LPS into the circulation (5). LPS in the circulation can initiate an immune response and promote the release of inflammatory cytokines (4,5). The gut microbiota can therefore be a generator of LPS and a potential contributor to low-level inflammation.

LPS-binding protein (LBP) is produced primarily by the liver and helps mediate the LPS-induced inflammatory response (5,6). When LBP is found at low levels in the serum, it binds LPS and forms complexes with CD14, which then associates with Toll-like receptor 4 (TLR4) on the macrophage and initiates the inflammatory cytokine response (7). However, when found at higher levels, LBP helps attenuate the immune response by transferring LPS to lipoproteins for clearance (8,9).

LBP levels are greater in the presence of cytokines such as interleukin (IL)-6 and IL-1 (5), and although LBP is produced at constitutively low levels during normal physiological states, levels rise rapidly during infection (5). There are several difficulties in accurately measuring LPS in serum using the Limulus amebocyte lysate test, including potential interference by LPS inhibitors (10). LBP is strongly correlated with LPS levels (r ≥ 0.6) in human serum (11) and increases in response to greater LPS in mouse models (12). Therefore, owing to its crucial role in modulating the immune response and its known correlation with LPS, LBP is generally considered to be a reasonable surrogate biomarker for assessing LPS-induced inflammation in humans (13,14).

LBP levels are higher among individuals who are obese, have diabetes, or have metabolic syndrome or glucose intolerance (13–18). Although cross-sectional studies have shown that LBP levels are associated with anthropometric and metabolic measurements (11,15,18), few longitudinal studies have investigated LBP in relation to obesity- and diabetes-related measures (13). To our best knowledge, no longitudinal studies have been conducted among African ancestry men, a population group disproportionately affected by type 2 diabetes (19,20), and thus, in particular, there is need for such study. In the current study, we tested whether baseline LBP is associated with changes in overall, central, and skeletal muscle adiposity, glucose homeostasis, and new cases of prediabetes and type 2 diabetes in a cohort of middle-aged and elderly African ancestry men.

Between 1997 and 2003, 3,170 previously unscreened men were recruited for a population-based prostate cancer screening study on the Caribbean island of Tobago, Trinidad and Tobago (21). To be eligible, men had to be ambulatory, noninstitutionalized, and not terminally ill. Recruitment for the survey was accomplished by flyers, public service announcements, and posters, informing health care workers at local hospitals and health centers, and word of mouth. Approximately 60% of all age-eligible men on the island participated, and participation was similar across the island parishes. All men were invited to participate in a follow-up clinic examination between 2004 and 2007, and 2,031 men (70% of survivors) and 451 new participants completed the visit. Men were invited to complete a dual-energy X-ray absorptiometry (DXA) whole-body scan and a peripheral quantitative computed tomography (pQCT) scan of the lower leg. This visit represented the baseline for the current study. Between 2010 and 2013, we invited these men to return for repeat clinical examinations and DXA and pQCT scans. The baseline and follow-up visits followed the same procedures for questionnaire interviews, biospecimen collection, and DXA and pQCT scans. A total of 1,611 men completed the follow-up assessment (82% of survivors). On the basis of power calculations (see statistical analyses below), we randomly selected 580 of these men for the current study of LBP. The Institutional Review Boards of the University of Pittsburgh and the Tobago Ministry of Health and Social Services approved this study. All participants provided written informed consent before data collection.

A pQCT scan of the calf was performed using the Stratec XCT-2000 to evaluate skeletal muscle fat and muscle cross-sectional areas. Scans were obtained at 66% of the calf length, proximal to the terminal end of the calf. This site was chosen because it is the region of the lower leg with the largest circumference with very little variability across individuals (22). Different tissues in the analyses were separated according to different density thresholds, using the “soft tissue” algorithm. On the basis of this calibration, fat, muscle, and cortical bone are measured with mineral equivalent densities of 0, 80, and 1,200 mg/cm³, respectively. Therefore, changes in muscle tissue to fat-like tissue will be detected as a shift in mineral equivalent density of the muscle from 80 to 0 mg/cm³. Images of the cross-sectional area of skeletal muscle and fat were analyzed using Stratec 5.5D analysis software (Orthometrix, Inc., White Plains, NY). To maintain consistency, all images were analyzed by a single investigator. We assessed intramuscular fat using measures of calf muscle density (mg/cm³). Muscle density is a valid measure of fat accumulation within the skeletal muscle and reflects the fat content such that greater intramuscular fat is associated with lower muscle density (23).

DXA measurement of total body and trunk fat was made using a Hologic QDR 4500 W densitometer (Hologic Inc., Bedford, MA). Scans were analyzed with QDR 8.26a software.

Standing height was measured to the nearest 0.1 cm using a wall-mounted stadiometer. Body weight was recorded to the nearest 0.1 kg without shoes on a balance beam scale. BMI was calculated from body weight and standing height (kg/m²). Waist circumference was measured at the narrowest point of the waist using an inelastic fiberglass tape. If there was no narrowest point, waist circumference was measured at the umbilicus.

All biochemical assays in fasting serum samples were performed in the Heinz Nutrition Laboratory at the University of Pittsburgh. Fasting serum glucose was measured using an enzymatic procedure; the coefficient of variation percentage (CV%) between runs was 1.8%. Insulin was measured using a radioimmunoassay procedure developed by Linco Research, Inc.; the CV% between runs was 2.1%. The degree of insulin resistance (IR) was estimated by HOMA according to the method described by Matthews et al. (24). In previous studies, HOMA-IR has correlated reasonably well with insulin clamp techniques (25). Baseline fasting serum LBP was measured using a Human LBP ELISA kit (Cell Sciences, Canton, MA) according to the manufacturer’s protocol. Manufacturer- reported inter- and intra-assay CV% were 9.8–17.8% and 6.1%, respectively.

Information on lifestyle habits (current smoking [yes/no], walking more than twice in the past week [yes/no], watching 14 or more hours of television per week [yes/no], and current intake of alcohol of more than 3 drinks per week [yes/no]), history of medical conditions, and medication use were assessed using standardized interviewer-administered questionnaires. Men were asked to bring all prescription medications taken in the past 30 days to their clinic visit. Participants also rated their overall health status compared with men their own age. Type 2 diabetes was defined as currently taking an antidiabetic medication, regardless of fasting serum glucose level, or having a fasting serum glucose level of ≥126 mg/dL. Impaired fasting glucose (IFG), also known as prediabetes, was defined as a fasting serum glucose level of 100–125 mg/dL without being on any antidiabetic medication. Hypertension was defined as a systolic blood pressure of ≥140 mmHg and/or diastolic blood pressure of ≥90 mmHg and/or currently taking antihypertensive medication. High triglycerides was defined as a fasting serum triglycerides level of ≥150 mg/dL. Low HDL cholesterol (HDL-c) was defined as a fasting serum HDL-c level of <40 mg/dL.

We estimated that to have the power to detect a univariate correlation of 0.12 (β = 0.8, α = 0.05) between LBP and changes in adiposity or glucose homeostasis, we needed to assay at least 500 samples. Owing to assay requirements and formatting, we measured LBP in a final number of 580 men. We then categorized baseline serum LBP into quartiles (n = 145 each) and tested its association with baseline cross-sectional and longitudinal changes in metabolic measures using a test of linear trend. Cross-sectional models were first adjusted for baseline age and then were additionally adjusted for smoking, walking, history of cancer, perceived health status, and alcohol intake. Longitudinal models were first adjusted for baseline age and baseline metabolic measures and then additionally adjusted for smoking, walking, history of cancer, perceived health status, alcohol intake, and low HDL-c status. We used multivariable logistic regression to test for an association of LBP with new cases of IFG or type 2 diabetes identified at follow-up. Models for IFG were run in only those with normal glucose at baseline, whereas models for type 2 diabetes were run in individuals without diabetes at baseline only. Odds ratios were expressed per 1-SD greater serum LBP. Statistical significance was based on an α = 0.05, and analyses were performed using SAS 9.3 software (SAS Institute, Inc., Cary, NC).

Baseline characteristics for the 580 African ancestry men overall and according to quartiles of LBP are reported in Table 1. The men were an average ± SD age of 59.1 ± 10.5 years. Only 8% of men were current smokers, 9% had moderate alcohol intake, 38% watched more than 14 h of television in a week, and 88% typically walked for exercise. Approximately 5% of men had a history of cancer, and 6% had a history of any cardiovascular disease. Prevalence of IFG, type 2 diabetes, high triglycerides, low HDL-c, and hypertension were high, at 23%, 21%, 19%, 22%, and 53%, respectively. Greater LBP quartile was associated with greater baseline age, having diabetes, low HDL-c status, and taking antidiabetic medication (P < 0.05 for all). However, having a history of cancer decreased with increasing quartiles of LBP (P = 0.04).

Average follow-up time was 6.0 years (range 4.6–8.5). Greater baseline LBP was associated with greater baseline BMI, waist circumference, whole-body and trunk fat percentage, and fasting insulin and HOMA-IR, and inversely associated with skeletal muscle density independent of age and other covariates including smoking, walking, history of cancer, health status, alcohol intake, and low HDL-c status (all P < 0.05) (Table 2).

Baseline LBP was associated with an increase in trunk fat (P = 0.025) and HOMA-IR (P = 0.034) at follow-up after adjustment for all significant covariates (Table 3). LBP was also associated with a decrease in calf skeletal muscle density in minimally adjusted models (P = 0.048) (Table 3). However, the association was attenuated after additional adjustment for smoking, walking, history of cancer, health status, alcohol intake, and low HDL-c status (P = 0.057).

At the follow-up visit, 13.5% of normoglycemic men from baseline had developed IFG and 8.2% of men free of diabetes at baseline had developed type 2 diabetes (Table 4). Baseline LBP levels were positively associated with new cases of IFG, independent of age, trunk fat percentage, physical activity, health status, and low HDL-c status. Each 1-SD greater baseline LBP was associated with an ∼51% increased risk of IFG (95% CI 1.02–2.21). The association was slightly attenuated and of borderline significance after accounting for change in trunk fat (P = 0.067), BMI (P = 0.046), or skeletal muscle density (P = 0.052). Baseline LBP was not significantly associated with new cases of diabetes.

To our knowledge, our study is the first to examine LBP and longitudinal changes in adiposity, glucose homeostasis, and diabetes risk in a population of African ancestry men, who have a high risk of developing type 2 diabetes. We found that greater LBP levels are associated with increasing central and skeletal muscle adiposity and IR as well as with an increased risk for developing IFG. Adjusting for changes in adiposity attenuated the association between LBP and IFG, suggesting that increased adiposity may play a causal role in the LBP association. These data provide further evidence for a potential link between LBP and age-associated increases in adiposity and impaired glucose metabolism.

Chronic inflammation is believed to be a risk factor for obesity and IR (26,27). LPS is derived from the outer membrane of gram-negative bacteria (4,5), and although low levels of LPS can be found in the circulation of healthy individuals, higher levels can produce inflammatory responses (5). LBP is expressed at constitutively low levels and increases in the presence of inflammatory cytokines such as IL-6 and IL-1 (5). LBP functions by binding to LPS, which accelerates LPS binding to CD14 and subsequent presentation to macrophage TLR4 receptors (7). Importantly, Cani et al. (28) showed that chronic infusion of LPS caused weight gain in mice similar to that of a high-fat diet, thereby linking metabolic LPS from the gut microbiome to systemic inflammation and weight gain.

Owing to limitations in accurately measuring LPS levels in the serum (10), not many studies have directly correlated LBP to LPS levels in the serum of humans. However, a single study by Moreno-Navarrete et al. (11) showed that LBP and LPS levels were strongly correlated with coefficients ≥0.6. In addition, LBP has been widely suggested to be a potential marker of gut-derived LPS and consequent LPS-induced inflammation (13–15,17,18). Therefore, most current human research infers that LBP variation is both correlated and caused by variation in LPS. However, it is also possible that variations in LBP in our study may be due to variations in cytokines or an acute-phase reaction unrelated to LPS (29).

Our findings that LBP was significantly and positively associated with baseline obesity measures are in line with other findings in the literature (11–13,16–18). A longitudinal study among 2,529 Chinese individuals found that higher baseline levels of LBP were correlated with an increased number of metabolic syndrome components (13). In contrast, baseline LBP measurements were not related to changes in BMI, waist circumference, whole-body fat, or fasting insulin levels in our sample. However, we found that LBP was associated with an increase in trunk fat over an average of 6 years of follow-up. Our findings raise the possibility that LPS may be more strongly associated with central adiposity, which is a stronger risk factor for type 2 diabetes, than overall adiposity (30). An association of LBP with adiposity is biologically plausible because LBP, which is primarily produced by hepatocytes, is also produced by adipocytes in response to local proinflammatory cytokines (12).

We also found that LBP was positively associated with HOMA-IR at baseline and during follow-up in individuals without diabetes and positively associated with insulin at baseline, although this association was not confirmed by changes in insulin during follow-up. One previous study found that higher levels of serum LBP correlate with HOMA-IR (17), while another found that an association with HOMA-IR may depend on obesity status (16).

We also found, for the first time, that greater LBP levels are associated with a lower skeletal muscle density and its decrease with aging, indicative of an increase in ectopic skeletal muscle adiposity. This finding is in line with other previous data, in particular, a study using mouse models, which showed that LPS injections can lead to changes in muscle quality (31). Ectopic skeletal muscle adiposity is greater among African ancestry individuals than Caucasians (32–36) and has been shown to be an important risk factor for type 2 diabetes (37,38). Whether LBP may partly explain ethnic/racial differences in skeletal muscle adiposity is unclear. Our findings will need to be confirmed in other populations, including those of African ancestry.

In our sample, greater LBP was associated with an increase in IR and increased odds of newly generating an IFG over 6 years of follow-up. However, LBP was not significantly associated with new cases of type 2 diabetes. TLR4 activation on insulin target cells by LPS can lead to activation of the Jun N-terminal kinase and inhibitor of κB kinase pathways, both of which can inhibit insulin’s action by blocking phosphorylation of insulin receptor substrates proteins and increasing insulin receptor substrates degradation. Furthermore, the Jun N-terminal kinase/inhibitor of κB kinase pathways lead to activation of nuclear factor-κB, which increases production of proinflammatory cytokines and further inhibits insulin signaling (1). Skeletal muscle TLR4 expression is known to be higher in individuals who are obese or who have type 2 diabetes and to correlate with IR (39). LBP is a marker of circulating LPS and facilitates the binding of LPS to TLR4; thus, LBP may be an important biomarker for predicting the development of IR due to LPS-induced inflammation, and therefore, prediabetes. Lack of a significant association with new cases of type 2 diabetes in our population may be due to lack of statistical power, although a recent publication by Zhou et al. (40) suggests that LBP measurements alone are not sufficient to predict type 2 diabetes. Alternatively, LBP may be an important marker of only the early metabolic disturbances seen in prediabetes. Future studies are needed to more definitively test this hypothesis.

A strength of our study lies in its longitudinal design. Only one other longitudinal study has been conducted to date, and it was in a Chinese population (13). In addition, the availability of DXA and CT measures allowed us to more accurately describe general and regional body fat distribution associated with LBP, compared with previous studies.

Our study also has some limitations. Our sample included middle-aged and elderly African ancestry men, and thus, our findings may not apply to younger men, women, or other ethnic groups. Also, dietary data were limited in our study and because dietary influences can affect gut bacteria or may affect diabetes risk through other means, having information on food intake could allow for us to have a more holistic picture of how lifestyle might affect LBP. We currently do not have liver function or disease information for our sample. LBP is primarily a hepatically produced protein, and the liver is a site for ectopic fat development; thus, this information may be an important factor to investigate in future studies. Finally, we measured LBP only at baseline and, therefore, cannot examine whether change in LBP is a stronger correlate of metabolic changes than a single LBP measurement.

In conclusion, the current study shows that greater serum LBP concentrations are associated with increases in trunk and skeletal muscle adiposity and IR with aging among African ancestry men. The association with IFG and early changes in IR suggest that LBP may be more informative among individuals with normal serum glucose. Further research is needed to better understand the mechanisms underlying the relationship between LBP, adiposity, and IR and prediabetes.

Acknowledgments. The authors would like to thank all of the participants and the supporting staff from the Tobago Health Study Office and the Calder Hall Medical Clinic.

Funding. This research was supported, in part, by funding or in-kind services from the Division of Health and Social Services and Tobago House of Assembly, by National Institute of Arthritis and Musculoskeletal and Skin Diseases grant R01-AR0-49747, and by National Institute of Diabetes and Digestive and Kidney Diseases grants R01-DK0-97084 and K01-DK0-83029.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.M.Z., C.H.B., and A.L.P. contributed to parent study conception and initiation. C.M.T. contributed to the study design and the acquisition of data and wrote the manuscript. J.M.Z., A.L.K., C.H.B., A.L.P., and I.M. contributed to the study design, data analyses, interpretation of data, and editing of the manuscript. C.S.N. and R.W.E. contributed to the acquisition of data and editing of the manuscript. All authors approved the manuscript before submission. C.M.T. and I.M. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.