Persistence of an atherogenic lipid profile after treatment of acute infection with Brucella.

Serum lipid changes during infection may be associated with atherogenesis. No data are available on the effect of Brucellosis on lipids. Lipid parameters were determined in 28 patients with Brucellosis on admission and 4 months following treatment and were compared with 24 matched controls. Fasting levels of total cholesterol (TC), HDL-cholesterol (HDL-C), triglycerides, apolipoproteins (Apo) A, B, E CII, and CIII, and oxidized LDL (oxLDL) were measured. Activities of serum cholesterol ester transfer protein (CETP), paraoxonase 1 (PON1), and lipoprotein-associated phospholipase A2 (Lp-PLA2) and levels of cytokines [interleukins (IL)-1β, IL-6, and tumor necrosis factor (TNFa)] were also determined. On admission, patients compared with controls had 1) lower levels of TC, HDL-C, LDL-cholesterol (LDL-C), ApoB, ApoAI, and ApoCIII and higher LDL-C/HDL-C and ApoB/ApoAI ratios; 2) higher levels of IL-1b, IL-6, and TNFa; 3) similar ApoCII and oxLDL levels and Lp-PLA2 activity, lower PON1, and higher CETP activity; and 4) higher small dense LDL-C concentration. Four months later, increases in TC, HDL-C, LDL-C, ApoB, ApoAI, and ApoCIII levels, ApoB/ApoAI ratio, and PON1 activity were noticed compared with baseline, whereas CETP activity decreased. LDL-C/HDL-C ratio, ApoCII, and oxLDL levels, Lp-PLA2 activity, and small dense LDL-C concentration were not altered. Brucella infection is associated with an atherogenic lipid profile that is not fully restored 4 months following treatment.

Patients that required extended treatment were excluded from the study. Thus, patients with Brucellosis complicated with endocarditis, meningitis, spondylitis, or localized abscesses were not included in this study. Moreover, patients with chronic disease [symptoms for more than 12 months or the appearance of class A immunoglobulins (IgA) in conjunction with class G (IgG) for longer than 6 months before the admission] were excluded. Patients with evidence of malnutrition (serum albumin levels <3.6 g/dl) were also excluded. Patient nutritional status did not change during observation (based on questions about quantity and quality of food intake). All patients received doxycycline 100 mg orally twice daily and rifampin 600 mg orally daily for 6 weeks.
The control group consisted of 24 age-and sex-matched healthy individuals who visited our Outpatient Clinic for a regular checkup. The control group had no detectable antibodies to Brucella. All subjects gave informed consent for inclusion in this study, and the study protocol was approved by our Institutional Ethics Committee.

Biochemical measurements
Fasting serum levels of TC, HDL-C, and TG were determined enzymatically on an Olympus AU600 Clinical Chemistry analyzer (Olympus Diagnostica, Hamburg, Germany). LDL-C was calculated using the Friedewald formula. None of the patients had TG levels exceeding 350 mg/dl. Apo's AI, B, and E as well as Lp(a) levels were measured with a Behring Nephelometer BN100 using reagents (antibodies and calibrators) from Date Behring Holding (Liederbach, Germany). ApoCII and ApoCIII were determined by an immunoturbidimetric assay provided by Kamiya Biomedical (Seattle, WA).
Plasma levels of oxLDL were measured by a competitive ELISA using a specifi c murine monoclonar antibody (4E6) according to the instructions provided by the manufacturer (Mercodia, Uppsala, Sweden). The specifi city of this method was studied by performing the assay in fi ve different plasma samples in which 5 or 15 ng of protein of native LDL or OxLDL was added exogenously. Intra-assay and inter-assay coeffi cients of variation of the assay were 6.0% and 7.0%, respectively ( 34 ). CETP activity was measured by a fl uorometric assay, using a commercially available kit (Roar Biomedical, New York, NY). Briefl y, 2 l of plasma diluted 1:1 with sample buffer (10 mmol/l Tris, 150 mmol/l NaCl, and 2 mmol/l EDTA, pH 7.4) were used as the source of CETP. The assay was performed for 1 h at 37°C, and the CETP activity was determined by the increase in fl uorescence intensity measured in a fl uorescence spectrometer at an excitation wavelength of 465 nm and emission wavelength of 535 nm ( 35,36 ).
Lp-PLA 2 activity was measured by the trichloroacetic acid precipitation procedure using 1-O -hexadecyl-2-[ 3 H-acetyl] sn -glycero-3-phosphocholine (10 Ci/mmol; DuPont-New England Nuclear, Boston, MA) as a substrate at a fi nal concentration of 100 M. Fifty microliters of either serum diluted 1:50, v/v, with HEPES buffer, pH 7.4, or the HDL-containing supernatant after treatment of serum with magnesium chloride-dextran sulfate (HDL-rich serum) (diluted 1:3, v/v, with HEPES) were mixed with HEPES in a fi nal volume of 90 l and used as the source of the enzyme. The reaction was performed for 10 min at 37°C, and Lp-PLA 2 activity was expressed as nmol PAF degraded per ml of serum per min ( 37 ). Reproducibility of Lp-PLA 2 assays was determined by intra-assay determination of coeffi cients of variation (CVs). The CVs for Lp-PLA 2 assays ranged between 4% and 5%.
We measured serum PON1 hydrolyzing activity against paraoxon (PON1 (paraoxonase)] and phenyl acetate [PON1 (arylesterase)]. Both PON1 (paraoxonase) and PON1 (arylesterase) activities in serum were determined in the presence of 2 mM Ca 2+ in 100 mM Tris-HCl buffer (pH 8.0) for paraoxon as leukocytes, and endothelial cells. PAF is degraded by PAF acetylhydrolase, also known as lipoprotein-associated phospholipase A 2 (Lp-PLA 2 ), an enzyme that catalyzes the hydrolysis of the acetyl group at the sn -2 position of PAF ( 17 ). Lp-PLA 2 also hydrolyzes phosphatidylcholine, resulting in the production of lysophosphatidylcholine, a molecule that may mediate various biologic effects of oxidized LDL (oxLDL) ( 17 ). Lp-PLA 2 is mainly distributed on the LDL subclasses (80%), while the remaining is found on HDL (HDL-Lp-PLA 2 ) ( 17 ). Studies have shown that an increase in Lp-PLA 2 mass or activity represents an independent risk factor for the development of atherosclerosis ( 18,19 ). On the other hand, HDL-Lp-PLA 2 may exhibit antiatherogenic properties ( 17 ). Various changes in Lp-PLA 2 plasma activity in response to infection and infl ammation have been reported among different animal species ( 20 ) and in population-based studies ( 21,22 ). Therefore, infection-induced atherogenesis could, at least in part, be mediated by alterations of Lp-PLA 2 activity . Paraoxonase 1 (PON1) plays an important role in HDLmediated anti-atherogenic action. PON1 is an esterase exclusively associated with HDL in plasma and catalyzes the hydrolysis of phospholipid hydroperoxides and cholesteryl ester hydroperoxides formed during LDL oxidation ( 23 ). During infection and infl ammation, serum PON1 activity decreases and acute-phase HDL is unable to protect LDL against oxidation (24)(25)(26).
LDLs consist of a heterogeneous family of particles with different size, composition, and density ( 27,28 ). Several studies have shown that small dense LDL (sdLDL) particles are more atherogenic ( 18,29,30 ) compared with large buoyant ones. The effects of infl ammation on the distribution of LDL subclasses have been investigated ( 9,31 ). An increase in sdLDL particles in response to infection has been reported in some studies ( 2,32 ).
Brucellosis is a zoonosis most commonly encountered in rural Mediterranean areas ( 33 ). The vast majority of cases are attributed to the subtype Brucella melitensis . The effect of Brucellosis on lipid parameters is currently unknown. We undertook this study to evaluate the possible quantitative and qualitative effects of acute Brucellosis on lipoprotein metabolism.

Study population
Twenty-eight consecutive patients who were admitted, diagnosed, and treated for acute Brucellosis in the Department of Internal Medicine, University Hospital of Ioannina, Greece, between January 2003 and September 2005 were included. No patient was receiving any hypolipidaemic agents or had any clinical or laboratory evidence of any disease known to affect lipid metabolism, such as neoplasia, renal or liver dysfunction, and hypoor hyperthyroidism.
All patients were examined on admission and 4 months following successful treatment. No change in the patient dietary habits and body weight was recorded during follow-up.
The diagnosis of acute Brucella infection was established by the presence of specifi c IgM antibodies against Brucella as determined by ELISA. algias, arthralgias, or constitutional symptoms, such as malaise and weakness. Twenty-four age-and sex-matched healthy individuals were included. The clinical and biochemical characteristics of study participants are shown in Table 1 .

Serum lipid profi le
On admission, patients had signifi cantly lower levels of TC, HDL-C, LDL-C, ApoB, and ApoAI compared with controls, while there were no signifi cant differences in TGs, ApoE, and Lp(a) levels ( Table 1 ). Moreover, ApoB/ApoAI and LDL-C/HDL-C ratios were signifi cantly higher in patients on admission compared with controls ( Table 1 ). ApoCII levels did not differ between Brucellosis patients at study entry and controls, whereas levels of ApoCIII were lower in patients on admission compared with controls ( Table 1 ). No difference was observed in oxLDL levels and oxLDL/ApoB ratio between patients on admission and controls ( Table 1 ). CETP activity was higher in patients on admission compared with controls ( Table 1 ).
Four months following successful treatment, signifi cant increases in levels of TC, HDL-C, LDL-C, ApoB, and ApoAI were noticed ( Table 1 ). On the other hand, levels of TGs, ApoB, ApoE, and Lp(a) as well as the LDL-C/HDL-C ratio were not signifi cantly altered. Also, the ApoB/ApoAI ratio signifi cantly decreased ( Table 1 ). ApoCII levels remained unchanged following treatment, whereas a signifi cant increase in ApoCIII levels was noticed ( Table 1 ). In addition, no difference in oxLDL levels and oxLDL/ApoB ratio before and after treatment was observed ( Table 1 ). CETP activity decreased after treatment compared with values on admission ( Table 1 ). Changes in CETP activity ( ⌬ CETP) were negatively correlated with changes in HDL-C levels ( ⌬ HDL-C) during treatment ( r = Ϫ 0.42, P = 0.02).

Hematologic response
No signifi cant difference in numbers of total white blood cells as well as individual components (neutrophils, lymphocytes, and monocytes) was noticed either between patients on admission compared with controls or between patients before and after treatment ( Table 1 ).

Lp-PLA 2 activity
Total plasma and HDL-associated Lp-PLA 2 activity as well as the ratio total plasma Lp-PLA 2 /ApoB did not differ between patients on admission and controls ( Table 2 ).
Cytokine [interleukins (IL)-1 ␤ and IL-6 and tumor necrosis factor (TNFa)] were determined by ELISA using a commercially available immunoassay kit (Quantikine; R and D Systems, Minneapolis, MN). Each sample was measured in duplicate with appropriate sensitivities for IL-1b (<0.1 pg/ml), IL-6 (<0.7 pg/ml), and TNFa (<4.4 pg/ml). The median and mean intra-assay and interassay CVs were <10% for all the above measurements. C-reactive protein (CRP) levels were measured by an immunoturbidimetric assay (Roche Diagnostics). The ELISA employed (Serion ELISA Classic Brucella IgG/IgM/IgA; Institut Virion/Serion, Germany) detected specifi c IgG, IgM, and IgA anti-Brucella antibodies and was used in accordance with the manufacturer's instructions.
LDL subclass analysis was performed electrophoretically using high-resolution 3% polyacrylamide gel tubes and the Lipoprint LDL System (Quantimetrix) according to the manufacturer's instructions (39)(40)(41). Briefl y, 25 l of sample was mixed with 200 l of Lipoprint Loading Gel and placed upon the upper part of the 3% polyacrylamide gel. After 30 min of photopolymerization at room temperature, electrophoresis was performed for 60 min with 3 mA for each gel tube. Each electrophoresis chamber involved two quality controls (sample provided by the manufacturer). For quantifi cation, scanning was performed with a ScanMaker 8700 digital scanner (Mikrotek) and iMac personal computer (Apple Computer). After scanning, electrophoretic mobility (Rf) and the area under the curve were calculated qualitatively and quantitatively with the Lipoprint LDL system Template and the Lipoware software (Quantimetrix, Redondo Beach, CA), respectively. The LDL subfraction was calculated using the Rf between the VLDL fraction (Rf 0.0) and the HDL fraction (Rf 1.0). LDL is distributed from Rf 0.32 to 0.64 as seven bands, whose Rfs are 0.32, 0.38, 0.45, 0.51, 0.56, 0.6, and 0.64 (LDL1 to LDL7, respectively). LDL1 and LDL2 are defi ned as large, buoyant LDL and LDL3 to LDL7 are defi ned as sdLDL. The cholesterol concentration of each LDL subfraction is determined by multiplying the relative area (area under the curve) of each subfraction by the TC concentration of the sample (the TC concentration of the sample is measured independently). Moreover, the Lipoprint LDL System provides a mean LDL particle size (nm).

Statistical analysis
Data are presented as mean (SD) except for non-normal distributed variables, which are presented as median (range). The Kolmogorov-Smirnov test was used to evaluate whether each variable followed a Gaussian distribution. The relationships among study variables were investigated using the Pearson product moment correlation coeffi cient ( r ), whereas correlations including at least one non-normal variable were performed using Spearman correlation coeffi cient (rho). Multivariate analysis was used to determine which factors were independently related to alterations of study parameters. Signifi cance was set at P < 0.05.
Paired Student's t -test (or Wilcoxon's Rank test) was applied for comparisons between study parameters before and after treatment. Unpaired t -test (or Mann-Whitney test as appropriate) was used to compare data between patients and controls. All analyses were carried out with SPSS 13.0 (SPSS, 1989-2004, Chicago, IL).

Characteristics of study population
All patients (18 male and 10 female, mean age 52.9 ± 14.6 years) had a history of occupational exposure to Brucella. They were admitted to hospital because of fever, my-ity, and total Lp-PLA 2 activity/apoB ratio was noticed ( Table 2 ).

PON1 (paraoxonase) and PON1 (arylesterase) activities
On admission, patients had signifi cantly lower activities of PON1 (paraoxonase) and PON1 (arylesterase) compared with controls ( Table 2 ). Four months after successful treatment, a signifi cant increase in PON1 (arylesterase) IL-6, IL-1b, TNFa, and CRP were found. Multiple regression analysis, which included those parameters that were signifi cantly correlated with Lp-PLA 2 in univariate analysis, showed that the two signifi cant predictors of total plasma Lp-PLA 2 activity were serum levels of TNFa and LDL-C ( Table 3 ).
Four months after successful treatment, no change in total Lp-PLA 2 activity, HDL-associated Lp-PLA 2 activ- areas) is exposed to Brucella. Therefore, if Brucellosis is associated with atherogenic changes in lipid profi le that persist following successful treatment, this may be of particular importance from a public health perspective.
Previous studies have shown that lipids are altered in patients with various infections. The majority of studies in humans included patients with sepsis ( 42,43 ). The decrease in cholesterol levels in infections is well known, but the underlying mechanisms remain unclear. Several studies showed that the reduction in cholesterol levels and the alterations in apolipoprotein concentrations were correlated with certain cytokine and acute phase protein levels. Indeed, such correlations were demonstrated in patients with sepsis ( 44 ), in neutropenic patients with fever ( 45 ), and in normal volunteers after a single intravenous endotoxin injection ( 46 ). Moreover, similar associations were noticed in patients with AIDS ( 47 ) and in critically ill surgical patients ( 48 ).
One explanation for the infection-associated decrease in cholesterol levels is that increased levels of IL-6 stimulate LDL receptor expression in hepatic cells and subsequently lead to increased uptake of LDL particles and decreased LDL-C plasma levels ( 49 ). This is consistent with the negative correlation between IL-6 and TC as well as LDL-C levels in our study. Furthermore, cytokines stimulate the displacement of ApoAI by serum amyloid A (an acute-phase protein), thus resulting in modifi cation of HDL particles ( 50,51 ). This mechanism may underlay the negative correlation between HDL-C and ApoAI levels with IL-6 concentration observed in our study. An additional mechanism of cytokine-induced hypocholesterolaemia includes a decrease in the hepatic synthesis and secretion of apolipoproteins ( 48,52 ). This is in agreement with the negative correlation between CRP and ApoAI levels in this study. What is more, the negative correlation between changes in CETP activity and HDL-C levels during treatment suggests that the increased CETP activity during acute infection contributes to the reduction of HDL-C levels.
Different changes in total plasma Lp-PLA 2 activity, a marker of increased CVD risk ( 19,53 ), have been observed in infl ammatory diseases. Indeed, an increase in plasma Lp-PLA 2 activity in patients with sepsis ( 54,55 ) or AIDS ( 21 ) has been reported. On the other hand, some studies showed that total plasma Lp-PLA 2 activity is decreased in sepsis ( 22,(56)(57)(58). In our study, total plasma Lp-PLA 2 and HDL-Lp-PLA 2 activities as well as the ratio of total plasma Lp-PLA 2 /ApoB did not signifi cantly differ between and PON1 (paraoxonase) activities was noticed ( Table 2 ). No signifi cant correlations between PON1 (paraoxonase) or PON1 (arylesterase) and any other lipid or cytokine parameter was found (data not shown).

Cytokines and CRP
On admission, patients had signifi cantly higher levels of CRP, IL-1b, IL-6, and TNFa compared with controls ( Table 2 ). Four months after successful treatment, signifi cant decreases in the levels of CRP, IL-1b, and IL-6 were noticed, whereas TNFa levels were not signifi cantly altered ( Table 2 ). We performed again all the analyses after excluding the two patients with partial CRP response (posttreatment CRP values of 66 and 18 mg/L). No essential change in the results was noticed.

LDL subclass analysis
The large LDL-C concentration and mean LDL particle size was lower, whereas sdLDL-C concentration was higher in patients on admission compared with controls ( Table 1 ). VLDL-C levels did not differ between the two groups ( Table 1 ).
Four months after successful treatment, VLDL-C levels, mean LDL particle size, and sdLDL-C concentration were not signifi cantly altered compared with baseline values ( Table 1 ). On the other hand, large LDL-C levels significantly increased ( Table 1 ).

DISCUSSION
This study shows for the fi rst time that acute Brucellosis is associated with a shift of serum lipids, lipoproteins, and associated enzymes toward a more atherogenic lipid profi le, which is not fully restored 4 months after treatment. Brucellosis is frequent in countries like Greece and others where a large part of the population (mainly in rural  ( 24 ). An increase in PON1 activity (both paraoxonase and arylesterase) was seen 4 months following successful treatment ( Table 2 ). This increase parallels that of HDL-C in the study cohort ( Table 1 ). We observed a shift in LDL subclass distribution toward smaller particles in patients with acute infection compared with controls. Indeed, in patients with acute Brucellosis, sdLDL-C levels increased, whereas large LDL-C levels and LDL size decreased compared with controls. Multiple regression analysis showed that signifi cant predictors of sdLDL-C levels and LDL size were male sex, TGs, and IL-6 levels ( Table 4 ). Of interest, 4 months later, sdLDL-C levels and LDL size were not signifi cantly different compared with those on admission, whereas large LDL-C levels significantly increased ( Table 2 ). There are limited reports in the literature concerning the effect of infection on LDL subclasses. Advanced stages of human immunodefi ciency virus infection are associated with an atherogenic lipid profi le, including a high prevalence of sdLDL particles ( 32 ). Moreover, severe periodontitis is associated with an increase in sdLDL-C levels ( 2 ). But which could be the reasons for the increased sdLDL-C and consequently decreased LDL size during acute infection in the absence of increased TGs? A possible explanation could be the independent correlation of sdLDL-C with IL-6. IL-6 may infl uence key enzymes in the metabolism of sdLDL, such as CETP and hepatic lipase. The fi nding that IL-6 decreased but did not fully reach control levels may partially explain the persistence of increased sdLDL-C 4 months later. It should be noted, however, that no signifi cant correlation between changes in CETP activity ( ⌬ CETP) and sdLDL-C levels ( ⌬ sdLDL-C) as well as between ⌬ CETP and changes in IL-6 levels ( ⌬ IL-6) during treatment were found (data not shown).
Finally, we measured ApoCII and ApoCIII, which are critical determinants of the TG-containing lipoprotein metabolism ( 68 ). Specifi cally, ApoCII is an important activator of LPL, thus enhancing the hydrolysis of TG-rich lipoproteins in the circulation. On the contrary, ApoCIII is a powerful inhibitor of LPL activity. We found a signifi cant decrease of ApoCIII levels at baseline compared with controls and a subsequent increase following treatment. Despite that serum triglycerides did not differ between patients and controls, we noticed a negative correlation between change in ApoCIII ( ⌬ ApoCIII) and change in TGs ( ⌬ TG; rho = Ϫ 0.47, P = 0.04).

Study limitations
The number of study participants is relatively small. However, to our knowledge, this is the fi rst study on the effects of acute infection with Brucella on serum lipids, lipoproteins, and associated enzymes. We used a cross-over design to compare subjects with matched healthy controls. Despite the limitations of this design, a prospective study of lipid alterations in this setting is not possible. In addition, the time that subjects were reevaluated (4 months after successful treatment) may have been too short for lipid alterations to be restored. However, it is very interesting to fi nd that despite subjects being in excellent health, they still had lipid modifi cations. patients on admission and controls ( Table 2 ). Multiple regression analysis showed that, apart from LDL-C levels, the only signifi cant predictor of plasma Lp-PLA 2 was TNFa concentration ( Table 3 ). However, the correlation between Lp-PLA 2 and TNFa was negative ( Table 3 ). Indeed, TNFa is considered as one potent inhibitor of Lp-PLA 2 secretion ( 59 ). On the other hand, total plasma Lp-PLA 2 activity may increase in response to Brucella-derived products, such as lipopolysaccharide (LPS) ( 60 ). LPS increases Lp-PLA 2 mRNA expression in several macrophagerich tissues ( 60 ). Moreover, a previous study indicated the presence of anti-PAF antibodies in patients with Brucella ( 61 ). These antibodies may be the result of a specifi c anti-PAF response. PAF stimulates the synthesis and secretion of Lp-PLA 2 , indicating a feedback regulatory mechanism that may serve to protect the host from the pathological effect of PAF ( 20 ). Thus, LPS and PAF or PAF-like substances may stimulate Lp-PLA 2 production and counteract the possible TNFa-mediated suppression.
One relevant question is where Lp-PLA 2 could be distributed in the presence of decreased LDL-C during infection. It is known that Lp-PLA 2 shows a preferential distribution for sdLDL particles compared with large buoyant ones ( 41 ). In agreement, sdLDL-C concentration increased during acute phase, while large LDL-C levels decreased compared with controls in our study ( Table 2 ). Of note, total plasma Lp-PLA 2 and HDL-Lp-PLA 2 activities as well as the ratio Lp-PLA 2 /ApoB did not signifi cantly change 4 months after successful treatment compared with baseline.
Previous data suggested that PON1 may protect lipoproteins against oxidative modifi cation, perhaps by hydrolyzing phospholipid and cholesteryl-ester hydroperoxides ( 24 ). Serum PON1 activity has been shown to be significantly lower in hepatitis C ( 62 ), human immunodefi ciency virus ( 63 ), and Helicobacter pylori ( 64 ) infected patients. In accordance, PON1 (as expressed by either paraoxonase or arylesterase activity) was reduced in patients with acute Brucellosis compared with controls ( Table 2 ). During infection and infl ammation, serum PON1 activity decreases as a result of changes in synthesis and/or secretion of HDL particles and ApoAI ( 48,(50)(51)(52). Moreover, PON1 activity may decrease in response to Brucella-derived products, such as LPS ( 60 ). Indeed, serum PON1 activity decreases following endotoxin (LPS) administration in Syrian hamsters ( 65 ). Additionally, TNFa and IL-1 treatment of HepG2 cells results in a decrease in PON1 mRNA levels ( 65 ). A previous study demonstrated that oxidized lipids, which are formed during LDL oxidation, result in the inactivation of PON1 ( 66 ). Similarly, it is likely that a possible increased oxidative stress induced by Brucella infection could have led to an increased binding of free radicals to PON1, thus resulting in a less active PON1. However, in our study, we did not observe any difference in oxLDL levels or the oxLDL/ApoB ratio between patients and controls or during treatment. Low PON1 seems to be a risk factor for CVD ( 67 ). Therefore, this decrease in PON1 activity during infection could represent another factor linking the acute phase response with increased atherogenesis CONCLUSIONS Brucella infection is associated with atherogenic changes of serum lipids, lipoproteins, and associated enzymes, which are not fully restored 4 months following successful treatment. Whether these changes contribute to a possible infection-induced acceleration of atherogenesis requires further studies.