Serum Lipopolysaccharide Activity Is Associated With the Progression of Kidney Disease in Finnish Patients With Type 1 Diabetes

Patients with type 1 diabetes were recruited and characterized by the Finnish Diabetic Nephropathy (FinnDiane) Study. The FinnDiane Study is a nationwide multicenter survey, initiated in 1997 to elucidate risk factors for diabetic kidney disease in Finnish patients with type 1 diabetes. The study protocol is in accordance with the Declaration of Helsinki, and it has been approved by the local ethics committee at each study center.

A total of 477 patients participated in this study. Type 1 diabetes was defined as an age at onset <35 years and permanent insulin treatment started within 1 year after the diagnosis. Data on medication and smoking were collected with a standardized questionnaire by the attending physician during the patient's visit. Kidney status was assessed by the albumin excretion rate (AER) in at least two of three overnight or 24-h urine collections: normal albumin excretion (AER <20 μg/min or <30 mg/24 h), microalbuminuria (AER ≥20 < 200 μg/min or ≥30 < 300 mg/24-h), macroalbuminuria (≥200 μg/min or ≥300 mg/24 h), and ESRD (dialysis treatment or kidney transplantation). In addition to the urine collections for classification, AER (24 h) was also measured centrally; this result was used for the statistical analysis. Patients with normal AER (normoalbuminuria) and macroalbuminuria were selected for the study based on their kidney status at the time of reexamination. After the prospective follow-up (5.9 ± 2.1 years on average), all available progressors, with a change from one albuminuria category to a higher level, were matched for sex and smoking with a group of nonprogressors that was twice as large. Altogether 239 normoalbuminuric patients and 238 macroalbuminuric patients were included in this study.

Analysis of fasting blood samples (A1C, serum lipids and lipoproteins, and serum creatinine) were performed as described earlier (12). Serum high-sensitivity C-reactive protein concentrations were measured centrally by photometric immunochemistry at the Helsinki University Central Hospital. Estimated glucose disposal rate (eGDR) was calculated as described earlier (12). The AER was measured centrally by photometric immunochemistry at the Helsinki University Central Hospital. Serum LPS activity was determined with a Limulus amoebocyte lysate chromogenic end point assay from 1:5 diluted samples (Hycult Biotechnology, Uden, the Netherlands). HDL is one of the key factors involved in the detoxification of endotoxins and is known to decline with the severity of diabetic nephropathy. The LPS-to-HDL ratio may reflect the severity of the condition. Serum IgA and IgG class antibodies to the periodontal pathogens Aggregatibacter actinomycetemcomitans and Porphyromonas gingivalis were determined from serum samples by a multiserotype enzyme-linked immunosorbent assay (13).

Data are expressed as means ± SD for normally distributed variables. For nonnormally distributed values, data are expressed as median (interquartile range [IQR]). The levels of significance were tested with ANOVA for normally distributed values, and nonnormally distributed values were analyzed with the Mann-Whitney U test. Correlations between clinical variables were analyzed by Pearson's correlation coefficient. Nonnormally distributed variables were log transformed for the analysis. Differences between frequencies were tested with Pearson's χ² test. The significance of disease-associated variables was tested with multivariate Cox regression analysis. Statistical analyses were performed with SPSS (version 15.0; SPSS, Chicago, IL). P < 0.05 was considered statistically significant.

The principal aim of the study was to investigate whether serum LPS activity is associated with the progression of kidney disease in 477 patients with type 1 diabetes. At the basal visit, 239 patients had a normal AER and 238 had macroalbuminuria. Clinical characteristics of the patients are shown in Table 1.

At baseline, macroalbuminuric patients had significantly higher LPS activity (53 [IQR 38–74] vs. 42 [31–60] EU/ml, P < 0.001) than normoalbuminuric patients. LPS activity did not differ between sexes among all patients. However, in the normoalbuminuric group male patients had slightly higher LPS activity than females (44 [33–63] vs. 38 [29–52] EU/ml, P = 0.037).

In the normoalbuminuric group, 80 of 239 patients developed microalbuminuria during the follow-up. Clinical characteristics of the normoalbuminuric nonprogressor and the normoalbuminuric progressor patients are presented in supplementary appendix 1 (available at http://care.diabetesjournals.org/cgi/content/full/dc09-0467/DC1). Groups were matched for sex and smoking. At baseline, there were no significant differences in antihypertensive (normoalbuminuric nonprogressor 8.2% vs. normoalbuminuric progressor 12.7%) or lipid-lowering (3.1 vs. 5.0%) medications between the groups. The normoalbuminuric nonprogressor patients had significantly higher A1C (9.4 ± 1.7 vs. 8.0 ± 1.4%, P < 0.001), total cholesterol (5.1 ± 1.1 vs. 4.7 ± 0.9 mmol/l, P = 0.001), LDL cholesterol (3.15 ± 0.91 vs. 2.78 ± 0.80 mmol/l, P = 0.002), apolipoprotein (apo) B (94 ± 21 vs. 85 ± 19 g/l, P = 0.001), and triglyceride concentrations (1.11 [IQR 0.88–1.90] vs. 1.06 [0.79–1.33], P = 0.04), as well as AER (16.7 [9.7–37.2] vs. 8.3 [5.9–12.3], P < 0.001) than normoalbuminuric nonprogressor patients. Normoalbuminuric progressor patients were also more insulin resistant (eGDR 6.5 [4.9–8.1] vs. 8.4 [6.1–9.3], P < 0.001). Of note, the normoalbuminuric progressor patients had higher serum LPS activity than normoalbuminuric nonprogressor patients (49 [34–87] vs. 39 [29–54] EU/ml, P < 0.001). However, serum antibody levels to the periodontal pathogen A. actinomycetemcomitans were not different between the groups. The normoalbuminuric progressor patients had slightly lower IgG levels to P. gingivalis than normoalbuminuric nonprogressor patients (Table 2).

In the normoalbuminuric group, variables significantly associated with progression in univariate analysis (A1C, LDL cholesterol, triglycerides, apoB, eGDR, and LPS) were further tested in a Cox regression model. The A1C (hazard ratio 1.28 [95% CI 1.11–1.49], P = 0.001) and eGDR (0.89 [0.80–1.00], P = 0.044) were independent risk factors for incident microalbuminuria (Table 3, model 1). When A1C was removed from the model, both eGDR (0.83 [0.75–0.91], P < 0.001) and lnLPS (1.85 [1.08–3.18], P = 0.026) were associated with disease development (Table 3, model 2). Because progression was defined based on a change in AER, it is natural that AER is a strong predictor of disease progression. Hence, we also included this variable into the regression analyses. As expected, the AER itself was the strongest predictor of disease progression in all models (data not shown).

In the macroalbuminuric group, 79 of 238 patients progressed to ESRD during follow-up. Clinical characteristics of the progressors and nonprogressors are presented in supplementary appendix 2. The groups were matched for sex and smoking. At baseline, macroalbuminuric progressor patients were treated more often with lipid-lowering agents than the macroalbuminuric nonprogressor patients (37 vs. 20%, P = 0.003). No differences were seen in the frequency of antihypertensive medication between the groups. The macroalbuminuric progressor patients had higher A1C (9.3 ± 1.8 vs. 8.7 ± 1.3%, P = 0.016), systolic blood pressure (149 ± 21 vs. 141 ± 19 mmHg, P = 0.01), serum creatinine (275 [IQR 157–364] vs. 117 [94–139] μmol/l; P < 0.001), serum triglyceride (1.71 [1.13–2.63] vs. 1.27 [0.98–1.70 mmol/l], P < 0.001), and serum adiponectin levels (20.4 [13.4–32.1] vs. 13.4 [9.0–19.6], P < 0.001) than macroalbuminuric nonprogressor patients. No difference in insulin sensitivity (eGDR) was found between the groups (P = 0.118).

A high LPS-to-HDL ratio was associated with both the development of microalbuminuria and with the progression of kidney disease (normoalbuminuric nonprogressor 30 [IQR 21–43] vs. normoalbuminuric progressor 39 [26–65], P < 0.001 and macroalbuminuric nonprogressor 47 [28–67] vs. macroalbuminuric progressor 57 [38–82], P = 0.017) (Fig. 1). At baseline, macroalbuminuric patients had higher serum levels of P. gingivalis IgA antibodies (0.81 [0.37–1.40] vs. 0.58 [0.32–0.91]; P < 0.01) than normoalbuminuric patients. The serum levels of A. actinomycetemcomitans IgG antibodies were also slightly lower in the macroalbuminuric patients. In the macroalbuminuric group, neither the serum LPS activity nor IgA/IgG antibody levels to periodontal pathogens were associated with the progression of kidney disease (Table 2).

Serum LPS activity was related to several clinical variables. Among all patients, the best correlations were observed between serum LPS activity and serum lipids: ln(LPS) versus ln(triglycerides) (r = 0.61, P < 0.001), ln(LPS) versus cholesterol (r = 0.34, P < 0.001), ln(LPS) versus apoB (r = 0.34, P < 0.001), and ln(LPS) versus HDL (r = −0.24, P < 0.001) (supplementary appendix 3). Adjustment of serum LPS activity with HDL concentrations strengthened the correlation with serum triglyceride concentration [ln(LPS-to-HDL ratio) vs. ln(triglycerides), r = 0.68, P < 0.001].

Diabetic nephropathy is a devastating health condition that causes morbidity and mortality. It has been estimated that approximately one-third of the patients with type 1 diabetes develop diabetic nephropathy during their lifetime. Although long duration of diabetes and poor glycemic control are associated with the progression of kidney disease, it is evident that other yet-unknown risk factors must exist.

Our living environment is inhabited with viruses and microorganisms, of which some are beneficial and others are detrimental to health. Gram-negative bacteria are considered harmful pathogens, which may colonize in various sites in the human body. In chronic diseases in particular, bacterial infections may cause significant damage to health. Bacterial LPS contributes to acute and chronic inflammations and triggers the innate immune response characterized by cytokine release and immune system activation.

We show here for the first time that high serum LPS activity is associated with the development of microalbuminuria in patients with type 1 diabetes. Although macroalbuminuric patients had a significantly higher serum LPS activity than normoalbuminuric patients, those in the macroalbuminuric group whose disease progressed were no different from those whose disease did not progress. Therefore, other factors seem to play a more important role in the disease progression of macroalbuminuric patients.

LPS has been shown to be associated with various classes of serum lipoproteins. In a normal situation, HDL is thought to be the main factor involved in the detoxification/neutralization of LPS. LPS is released from the cell membranes by HDL and eventually cleared by the liver (6). In certain disease conditions, such as atherosclerosis, periodontitis, and bacterial sepsis, HDL levels can be significantly reduced. This may lead to LPS redistribution toward apoB-containing LDL- and VLDL-rich particles (14,–16). In this respect it is understandable that progression of kidney disease has been linked to both dyslipidemia and metabolic syndrome in patients with type 1 diabetes (2,12,17). Normoalbuminuric patients who developed microalbuminuria displayed several features of the metabolic syndrome, including decreased insulin sensitivity, dyslipidemia, and poor glycemic control. Notably, serum LPS activity showed a strong positive correlation with serum triglyceride and apoB concentrations. A concomitant decrease in serum HDL concentrations could indicate that LPS molecules are distributed toward apoB-containing VLDL/LDL particles. We have recently demonstrated that high LPS activity, especially in combination with a low HDL cholesterol concentration increases the risk of incident cardiovascular disease events in the FINRISK study, a survey on risk factors of chronic diseases in the Finnish population (13). Likewise in the present study, a high LPS-to-HDL ratio was associated with the development of kidney disease in normoalbuminuric patients and with progression in macroalbuminuric patients with type 1 diabetes.

Serum LPS activity in itself does not reveal the type or source of infection. Based on previous findings, Gram-negative bacterial infections are relatively common in patients with periodontitis and chronic kidney disease (8,–10,18). In the present study, we did not see an association between two common periodontal pathogens (A. actinomycetemcomitans and P. gingivalis) and the progression of the kidney disease in patients with type 1 diabetes. This does not, however, rule out detrimental effects of periodontal pathogens in patients with diabetic nephropathy. It is also possible that harmful bacterial pathogens may colonize in other parts of the body. Diabetic patients often have respiratory tract infections. A recent study demonstrates that poor glycemic control and a long duration of type 1 diabetes increase the risk of pneumonia-related hospitalization (19). The gastrointestinal tract may also be inhabited by pathogenic bacteria, e.g., Helicobacter pylori, which is one of the most common causes of chronic infections in humans (20). Recent studies have also demonstrated that in mice fed a high-fat diet, colonization of Gram-negative bacteria in their gut is increased. Changes in gut microbiota may induce metabolic endotoxemia, which is associated with increased inflammation, obesity, and insulin resistance (21,22). A high-fat diet has also been shown to increase plasma levels of bacterial endotoxins in humans (23). A recent study in mice demonstrated that long-chain fatty acids in particular promote LPS absorption from the gut (24). Based on the above observations, it seems that both quantity and quality of lipids influence the translocation of LPS from the gastrointestinal tract to the circulation. These results indicate that infections in the gastrointestinal tract may also contribute to the development of bacterial endotoxemia in humans. It is evident that hyperglycemia increases the risk for bacterial infections. In Danish diabetic patients, a 1 mmol/l increase in plasma glucose was associated with a 6–10% increased relative risk of pneumonia, urinary tract infection, and skin infection (25).

In the present study we showed that poor glycemic control and high serum LPS activity are associated with the development of kidney disease in patients with incident microalbuminuria. Further studies are required to elucidate the role of Gram-negative bacterial infections in patients with diabetes. In the future, identification of bacterial pathogens may help us to find more effective treatments for these patients, which may also retard progression of diabetes complications.

The study was supported by the Folkhälsan Research foundation (to P.-H.G.), Wilhelm and Else Stockmann foundation (to P.-H.G. and M.L.), and the Academy of Finland (grant 118391 to P.J.P.).

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

Parts of this study were presented in abstract form at the annual meeting of the European Diabetic Nephropathy Study Group in Frascati, Rome, Italy on 29–30 May 2009.

The skilled technical assistance of our laboratory technicians M. Parkkonen, A.-R. Salonen, A. Sandelin, T. Soppela, and J. Tuomikangas is acknowledged. We also acknowledge the physicians and nurses at each study center.