Alterations in vitamin D status and anti-microbial peptide levels in patients in the intensive care unit with sepsis

Background Vitamin D insufficiency is common in hospitalized patients. Recent evidence suggests that vitamin D may enhance the innate immune response by induction of cathelicidin (LL-37), an endogenous antimicrobial peptide produced by macrophages and neutrophils. Thus, the relationship between vitamin D status and LL-37 production may be of importance for host immunity, but little data is available on this subject, especially in the setting of human sepsis syndrome and other critical illness. Methods Plasma concentrations of 25-hydroxyvitamin D (25(OH)D), vitamin D binding protein (DBP) and LL-37 in critically ill adult subjects admitted to intensive care units (ICUs) with sepsis and without sepsis were compared to healthy controls. Results Critically ill subjects had significantly lower plasma 25(OH)D concentrations compared to healthy controls. Mean plasma LL-37 levels were significantly lower in critically ill subjects compared to healthy controls. Vitamin D binding protein levels in plasma were significantly lower in critically ill subjects with sepsis compared to critically ill subjects without sepsis. There was a significant positive association between circulating 25(OH)D and LL-37 levels. Conclusion This study demonstrates an association between critical illness and lower 25(OH)D and DBP levels in critically ill patients as compared to healthy controls. It also establishes a positive association between vitamin D status and plasma LL-37, which suggests that systemic LL-37 levels may be regulated by vitamin D status. Optimal vitamin D status may be important for innate immunity especially in the setting of sepsis. Further invention studies to examine this association are warranted.

Plasma concentrations of 25-hydroxyvitamin D (25(OH)D), vitamin D binding protein (DBP) and LL-37 in critically ill adult subjects admitted to intensive care units (ICUs) with sepsis and without sepsis were compared to healthy controls.

This study demonstrates an association between critical illness and lower 25(OH)D and DBP levels in critically ill patients as compared to healthy controls. It also establishes a positive association between vitamin D status and plasma LL-37, which suggests that systemic LL-37 levels may be regulated by vitamin D status. Optimal vitamin D status may be important for innate immunity especially in the setting of sepsis. Further invention studies to examine this association are warranted.

Study Sample and Subjects This study was approved by the Emory University Institutional Review Board. Samples were taken from three patient populations: Group 1 consisted of 24 critically ill subjects in the intensive care unit (ICU) patients diagnosed with sepsis (as defined by the American College of Chest Physicians (ACCP) and Society of Critical Care Medicine (SCCM) consensus panel in 2001 [ 15 ]; Group 2 consisted of 25 ICU subjects without the diagnoses of sepsis, and Group 3 consisted of 21 healthy non-hospitalized controls. Samples were collected between January of 1999 and May of 2006. Group 1 samples were drawn within 2 days of severe sepsis onset and were drawn from the medical intensive care unit between June 2004 and February 2006. Group 2 samples were drawn during the subject's hospital day, which was a mean of 12.8 days. Critically ill subjects (groups 1 and 2) were characterized by sex, race, Acute Physiology and Chronic Health Evaluation II (APACHEII) and sequential organ failure assessment (SOFA) scores (for ICU patients) and whether they were diagnosed with cardiovascular disease (ischemic heart disease or congestive heart failure), liver disease, chronic renal failure, diabetes, HIV, or cancer. Subjects also had baseline laboratory tests performed by standard hospital laboratory methods including albumin, prothrombin time (PT), partial thromboplastin time (PTT), INR (International Normalized Ratio), Alanine aminotranferease (ALT), aspartate aminotransferase (AST), blood urea nitrogen (BUN), creatinine (Cr), hemoglobin, and white blood cell count. Diagnoses and laboratory data were obtained from discharge summaries and computer databases. Healthy control subjects (group 3) were adults without known acute or chronic diseases, no hospitalizations for any illness previous 12 months, not taking any medications or vitamin supplements. They were screened for inclusion by a physician (TRZ) in General Clinical Research Clinic (GCRC) setting to confirm normal history and physical exam and had normal complete blood count, chemistry profile and urinalysis, which were tested within 2 weeks of screening.

We used GraphPad Prism version 4.0 software (La Jolla, CA) to compare means of each of the lab values across the three groups of patients (critical ill subjects with and without sepsis and healthy controls). Multiple regression was used to calculate adjusted least squares means using PROC GLM in SAS 9.1 (SAS Institute Cary, NC) between the three subject populations. Fisher's exact test was used to examine differences in race across the three patient groups. Similarly, patients with darker skin pigmentation are at increased risk for vitamin D deficiency due to the absorption of solar radiation between 290 to 700 nm by melanin [ 15 ], therefore, we controlled for racial differences in all multivariate models. We also examined age and BMI as potential covariates but they were not significantly associated with LL-37. One patient sample was dropped from the analysis because the LL-37 level was more than three standard deviations from the mean. Data are presented as means ± SD.

Vitamin D status differed in the critically ill subjects with sepsis, critically ill subjects without sepsis and healthy controls (p < 0.0001, ANOVA). Race adjusted 25(OH)D concentrations demonstrated no significant differences in 25(OH)D between the two critically ill groups. However, the mean race adjusted 25(OH)D level in the two critically ill groups (16.0 ± 8.5 and 16.2 ± 7.2 ng/mL, sepsis and non-sepsis respectively) was significantly lower than healthy controls (26.0 ± 7.6 ng/mL) (Figure 1 ). The prevalence of vitamin D insufficiency (defined as 25(OH)D < 30 ng/ml) in critically ill subjects with sepsis was 100% (24/24) and 92% (23/25) in critically ill subjects without sepsis. In contrast, the prevalence of vitamin D insufficiency in the healthy controls was 66.5% (14/20) (p = 0.003); there was no significant difference in the prevalence of vitamin D insufficiency between the two critically ill groups (p = 0.15). Figure 1 Vitamin D status in critically ill subjects with sepsis, critically ill subjects without sepsis and healthy subjects . Plasma 25-hydroxyvitamin D levels in critically ill subjects with sepsis (hatched bar) and in critically ill control subjects without sepsis (dark bar) were significantly lower than healthy controls (white bar) (ANOVA, p < 0.0001). 25-hydroxyvitamin D concentrations were adjusted for race. ‡ p < 0.001, critically ill sepsis subjects compared to healthy controls. # p < 0.01, critically ill control subjects compared to healthy controls. There was a statistically significant difference in plasma vitamin D binding protein between the three groups of subjects (ANOVA, p = 0.014) (Figure 2 ). Subjects with sepsis had significantly lower DBP concentrations compared to subjects without sepsis (Figure 2 ). Race was not associated with DBP levels. Cathelicidin (LL-37) concentrations differed in the three groups (p = 0.002, ANOVA). Both groups of critically ill subjects had similar plasma LL-37 concentrations (Figure 3 , 13.7 ± 2.1 ng/mL vs. 10.6 ± 1.4 ng/mL; P = 0.59). However, mean plasma LL-37 levels in healthy controls (27.2 ± 4.9 ng/mL) were significantly higher than compared to the critically ill groups (p < 0.001, Tukey-Kramer for both comparisons) (Figure 3 ). While adiposity has been associated with 25(OH)D levels, we tested the association of BMI with LL-37 and showed no statistical significant relationship (p = 0.20). Therefore, BMI was not used in any of our models. Plasma LL-37 levels also were not significantly associated with age or race. Figure 2 Plasma vitamin D binding protein in critically ill subjects with sepsis, critically ill subjects without sepsis and healthy subjects . Plasma vitamin D binding protein concentrations were significantly lower in critically ill subjects with sepsis (hatched bar) compared to critically ill control subjects (dark bar) (white bar) (ANOVA, p = 0.014). † p = < 0.05, critically ill sepsis subjects compared to critically ill control subjects. Figure 3 Anti-microbial peptide cathelicidin (LL-37) in critically ill subjects with sepsis, critically ill subjects without sepsis and healthy subjects . Plasma LL-37 levels were significantly lower in the two critical ill groups (with sepsis, hatched bar and without sepsis, dark bar) compared to the healthy controls subjects (white bar) (ANOVA, p= 0.002). There was no statistically significant difference between LL-37 levels in the two critically ill groups. ‡ p < 0.001, critically ill sepsis subjects compared to healthy controls. # p < 0.001, critically ill control subjects compared to healthy controls.

We have demonstrated that vitamin D insufficiency is highly prevalent in all three populations. Even in healthy controls, over sixty percent were found to be vitamin D insufficient. However, the prevalence of vitamin D insufficiency is even higher in subjects admitted to the intensive care unit with critical illness. We also demonstrate that vitamin D binding protein levels are significantly lower in critically ill subjects with sepsis compared to critically ill subjects without sepsis and healthy controls. When we examined plasma levels of the endogenous anti-microbial peptide LL-37 in relationship to 25(OH)D, we found that lower levels of 25(OH)D were also associated with lower systemic levels of LL-37. This association supports recent in vivo data that vitamin D plays some roles in regulating the production of antimicrobial peptides such as LL-37 in cultured macrophages [ 3 ]. Since many cells of the immune system possess the vitamin D receptor, vitamin D status may prove to be an important factor in management of sepsis syndrome and other critical illness. Vitamin D insufficiency is a common condition in patients admitted to the intensive care unit [ 5 , 16 - 18 ]. We found that > 95% of our critically ill patients had vitamin D insufficiency. Patients with critical illness likely had vitamin D insufficiency which preceded their hospitalization since several studies have documented a high prevalence of vitamin D insufficiency in hospitalized patients [ 19 - 21 ]. Vitamin D insufficiency continues to remain a health concern in hospitalized patients since few treatment guidelines exist to address vitamin D status. The American Society for Parenteral and Enteral Nutrition recommend only 200 IU of vitamin D daily for hospitalized patients [ 22 ]. Heaney estimates that a dose of 400 IU daily would only raise 25(OH)D concentrations by 2.8 ng/mL, leaving most hospitalized patients vitamin D insufficient [ 23 ]. Van den Berghe et al evaluated increased vitamin D repletion of critically ill subjects with 500 IU of vitamin D; however, 25(OH)D concentrations still remained in the insufficient range [ 6 ]. Thus, these studies suggest that higher recommended doses of vitamin D are likely needed to correct vitamin D insufficiency in hospitalized patients. We found that vitamin D binding protein (DBP, and also known as Gc-globulin) concentrations were also significantly lower in critically ill subjects with sepsis compared to critically ill subjects without sepsis and healthy control subjects. Our findings are consistent with Dahl et al who reported that lowered DBP was associated with sepsis and organ dysfunction [ 14 ]. Vitamin D binding protein is the major carrier protein for circulating 25(OH)D. Adequate levels of DBP are required to recover filtered 25-hydroxyvitamin D lost in the urine [ 24 ]. This process is facilitated by megalin, a protein located on the renal epithelial cell which binds to the DBP-25-hydroxyvitamin D complex to facilitate the recovery of filtered vitamin D metabolites [ 25 ]. Lower DBP results in further loss of urinary 25(OH)D further exacerbating already low levels of circulating 25(OH)D concentrations. Vitamin D binding protein not only is a carrier for the two major circulating forms of vitamin D, 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D, but it is also a scavenger of monomeric actin thus preventing its polymerization into F-actin [ 25 , 26 ]. The actin binding characteristics of DBP may play a protective role in sepsis to prevent polymerization of actin released from injured tissue which can in turn result in microembolization of end-organs [ 25 , 26 ]. Actin binding with DBP results in lowered DBP concentrations which in turn further lowers 25(OH)D due to renal wasting of vitamin D and its metabolites, providing another mechanism to explain why vitamin D insufficiency is common in patients with sepsis. The classic function of vitamin D is to maintain optimal calcium and skeletal homeostasis. Nierman and Mechanick reported the majority of their cohort of chronically ill elderly patients transferred from the intensive care unit had evidence of rapid bone turnover due to vitamin D deficiency [ 5 ]. Recent evidence suggests that vitamin D may also play an important role in enhancing innate immunity against infection. Liu et al demonstrated that 1,25(OH) 2 D 3 treatment of macrophages infected with Mycobacterium tuberculosis in vitro resulted in enhanced production of an endogenous anti-microbial peptide, cathelicidin or LL-37, and in improved killing of the microorganisms [ 3 ]. LL-37 has a broad antimicrobial spectrum and has been demonstrated to possess multiple other immunoregulatory functions, from chemoattraction of inflammatory cells, to promotion of wound healing, and regulation of angiogenesis [ 27 ]. Administration of LL-37 has been demonstrated to be protective in rodent models of sepsis [ 28 , 29 ]. A recent randomized, placebo controlled trial of vitamin D supplement

The authors declare that they have no competing interests.

This research was supported in part by grants from the University Research Committee of Emory University, National Institutes of Health Grant # K23AR054334 and #5T32DK007298. The authors responsibilities were as followed. LJ, LVY and SEJ: contributed to the study design and concept, statistical analysis, and writing of the manuscript. HMB, GM and TRZ: contributed to study design and concept and writing and editing of the manuscript. VT: contributed to the study design and concept, writing and editing of the manuscript and provided financial support for the manuscript.