Introduction

Obesity is characterized by impaired microvascular and macrovascular function that may contribute to increased risk of cardiovascular disease.1 Clinical and experimental evidence suggest that microvascular dysfunction is a potential factor explaining the association among obesity, hypertension and insulin resistance, which are major cardiovascular risk factors.2,3 Several methods are used to assess microvascular function in clinical research.4 Postocclusive reactive hyperemia (PORH) at forearm skin evaluated by laser Doppler flowmetry (LDF) has been widely used to assess microvascular function due to its non-invasive nature. PORH is the sudden rise in skin blood flow after release of a brief arterial occlusion and provides an overall measurement of microvascular function.5 However, concerns about the reproducibility of this method have been recently raised.5 In particular, reproducibility studies of LDF in obese patients are scarce.

Bariatric surgery has emerged as an effective treatment for morbid obesity on the basis of both its efficacy6 and beneficial effects on obesity-related comorbidities and total mortality.7, 8, 9 It has been previously shown that bariatric surgery could reverse microvascular dysfunction in obese patients although the determinants that mediate this improvement in microvascular function have not been identified.10, 11, 12, 13, 14 Bariatric surgery-induced weight loss also ameliorates the metabolic syndrome (MetS), a group of clinical manifestations that includes obesity, hypertension, insulin resistance and dyslipidemia.15 Interestingly, the association between surgically induced improvement in microvascular function and metabolic syndrome has not been previously investigated.

The purpose of this study was to investigate the effect of surgically induced weight loss on microvascular function in morbidly obese patients with and without MetS. More specifically, we wanted to determine whether changes in microvascular function after bariatric surgery are associated to resolution of MetS. To this end, we prospectively evaluated microvascular function by LDF in morbidly obese patients with and without MetS before and 12 months after bariatric surgery. We also assessed the contribution of different clinical, metabolic and biochemical parameters to surgically induced improvement in microvascular function.

Materials and methods

Study design and subjects

Obese subjects were recruited from the waiting list for bariatric surgery of the Surgery Unit at Hospital Universitario Virgen del Rocío from November 2009 to March 2011. All obese patients were required to meet NIH guidelines for eligibility for bariatric surgery: BMI ⩾40 kg m−2 or ⩾35 kg m−2 with comorbidities (that is, diabetes, hypertension, dilated cardiomyopathy or sleep apnea).16 The inclusion criteria were male and female patients aged 16–65 years, and agreement to participate in the study by providing a signed consent form. The patients with arterial hypertension, diabetes, cardiomyopathy and sleep apnea were under medical treatment for these obesity complications at the time of the evaluation. Exclusion criteria included acute or chronic inflammatory disease, malignant disease, asthma or any history of alcohol or drug abuse. The experimental protocol was approved by the Ethical Committee of the Hospital Universitario Virgen del Rocío. All participants provided written informed consent to participate in the study. Additional written informed consent was obtained after the surgical procedure.

Obese patients were grouped as patients with MetS or without MetS on the basis of the definition of MetS proposed by the Third Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults Panel III.17 Patients fulfilling three or more of the following criteria were considered as having MetS: (1) central obesity (waist circumference >102 cm in men or >88 cm in women); (2) high blood pressure of 130/85 mm Hg or greater or use of antihypertensive therapy; (3) high fasting glucose (⩾110 mg dl−1); (4) hypertriglyceridemia (⩾150 mg dl−1) and (5) low high-density lipoprotein cholesterol (HDLc) (<40 mg dl−1 for males or <50 mg dl−1 for females). Each subject made a visit at baseline and 12 months after the bariatric surgery. For the 12-month follow-up study, patients in the group with MetS before surgery were further subdivided into two subgroups: patients in whom MetS resolved after surgery and patients in whom MetS persisted after surgery. Healthy lean control subjects were also recruited as a comparison group for the evaluations before surgery.

Clinical and biochemical measurements

Clinical and biochemical measurements were performed before surgery (two weeks earlier) and 12 months after surgery. Every measurement was performed after an overnight fast of 10 h. Systolic and diastolic blood pressure was measured right before LDF evaluation. The values were the mean of two measurements with subjects in sitting position. Measures of weight, height, and waist and hip circumferences were also obtained. Blood samples were drawn from an antecubital vein. Plasma glucose, total serum cholesterol, high-density lipoprotein and triglycerides were measured using a Cobas C chemistry analyzer (Roche Diagnostics, Mannheim, Germany). The Friedewald equation was used to calculate LDL cholesterol (LDLc) from total serum cholesterol, HDLc and triglycerides.18 Plasma insulin was measured by electrical chemiluminescence immunoassay using an ElecsysE170 (Roche Diagnostics, Mannheim, Germany). Hemoglobin A1c (HbA1c) was measured by high-pressure liquid chromatography using a Variant II analyzer (BioRad Laboratories, Hercules, CA, USA) The index of insulin resistance (HOMA) was calculated using the formula: glucose (mmol l−1) × insulin (μU ml−1)/22.5. Plasminogen activator inhibitor-1 (PAI-1) concentrations were determined by enzyme-linked immunosorbent assay (American Diagnostica Inc., Stamford, CT, USA). Ultrasensitive C-reactive protein (CRP) was measured with the CardioPhase hsCRPkit (Dade Behring, Marburg, Germany), the intra and interassay CVs being 2.8 and 4.6%, respectively. Oxidized low-density lipoprotein (oxLDL) was measured with the ELISA kit (Immunodiagnostic Systems, Boldon, UK). The intra and interassay CV was 3.9 and 9%, respectively.

Post-occlusive forearm skin reactive hyperemia (PORH) measurements

Studies were performed in the morning in a quiet, temperature-controlled room (22–24 °C). Subjects were asked to avoid smoking and caffeine- and alcohol-containing drinks for 24 h, and from performing vigorous exercises for at least 12 h before the test. Measurements were taken with subjects in a supine position. On the preoperative study, LDF test was performed two weeks before the surgery.

Changes in cutaneous blood flow (flux) was measured by a commercial single-point laser Doppler flowmetry device (Periflux 5000; bandwidth of 15 kHz, Perimed AB, Järfälla, Sweden) with a thermostatic laser Doppler probe (Probe 481-1, Perimed AB, Järfälla, Sweden). The probe has a fiber separation of 0.25 mm and collects perfusion data at a depth of about 0.5–1 mm. Blood flow data were recorded continuously at a sample rate of 40 recordings per second. Data from the laser Doppler perfusion monitor were analyzed using PeriSoft for Windows, version 2.5.5 (Perimed AB, Järfälla, Sweden). Data files were processed for conversion from mV to PU (perfusion units) by division with the gain factor of the instrument (10 mV PU−1). The laser Doppler probe was placed on the volar surface of the right forearm, 10 cm proximal to the wrist. This position was marked so that exactly the same site was used in all measurements. After a baseline measurement of 3 min, the brachial artery was occluded using a pressure cuff placed around the right upper arm that was inflated to 220 mm Hg. Inflating the cuff took less than 5 s. This local ischemia was held for 4 min, and then deflated. Deflating the cuff was practically instantaneous (<40 mm Hg within 0.2 s). The flux recording was continued for at least 5 min (until the signal reaches the baseline flux). Five different PORH parameters were analyzed (Supplementary Figure 1). The value of skin flux at baseline is defined as the average value of the 3-minute baseline period before occlusion. Maximum response (PORHmax) was defined as the maximum absolute change (PORHpeak) from baseline.19 The area under the hyperemic curve (AUCH) was calculated from the time the cuff was released until the end of the measurement. The area under the occlusion curve (AUCO) was calculated from the time the occlusion started until the end of the occlusion. These five parameters were subsequently transformed into cutaneous vascular conductance (CVC), calculated as the flux divided by the mean arterial pressure in mm Hg. Mean arterial pressure was calculated as (2/3(diastolic blood pressure) + 1/3(systolic blood pressure)). To determine the reproducibility of laser Doppler-derived parameters in the measurement of PORH in obese patients, a study of reproducibility was performed before and after bariatric surgery. Specifically, we investigated within-subject reproducibility of the PORH parameters, that is, reproducibility and variability between measurements. These measurements were performed at different times: (0, 15 min and 24 h after the first measurement). For comparison among groups, we used data from measurements at time 0. All measurements were performed by the same investigator.

Statistical methods

Results are shown as the mean±s.d. or median and interquartile range unless otherwise noted. Between-group differences in normally distributed data were assessed by one-way analysis of variance (ANOVA) followed by the Fisher’s multiple comparison tests to identify differences between groups. For non-normally distributed data, comparisons between groups were analyzed by the Kruskal–Wallis test with Dunn’s multiple comparison tests. To analyze changes after bariatric surgery, the two-way ANOVA for repeated measures was chosen with post hoc Tukey’s comparisons. The contribution of clinical and biochemical parameters to variation in microvascular function was assessed by multivariate regression analysis. Statistically significant predictors were included in the models with a stepwise procedure, after adjusting for age and gender, and antihypertensive and diabetic treatment. Two-sided P-values were assessed for all models. Only variables that had a P<0.05 were included in the final model. To determine the repeatability of LDF, the intraclass correlation coefficients were calculated. Values of intraclass correlation coefficient more than 0.80 were considered excellent agreements.20 To examine the intraday and interday reproducibility of each LDF parameter, within subject coefficients of variation (CVs) were calculated as previously described.21 CVs below 35% were considered acceptable.22 Studies were also compared pairwise to check the precision of the method, the existence of magnitude-dependent bias and systematic error by using the Bland & Altman plots. Statistical analyses were performed with the SPSS statistical package (version 17.0; SPSS, Chicago, IL, USA). Power calculation indicated that our sample size provided an 80% power to detect differences in the vascular reactivity with an effect size as low as 0.2 (Cohen’s f), and 90% power to detect an effect size as low as 0.23, on the basis of two-sided tests at the 0.05 significance level. Power calculations were conducted using GPower3.23

Results

Before analyzing the impact of the MetS on microvascular function, we wanted to determine the reproducibility of the different LDF parameters in obese patients. Intraclass correlation coefficients analysis showed that all LDF measurements were above 0.80 (Supplementary Table 1). As expected, intraday CVs were higher than interday CVs in both evaluations (before and after surgery) (Supplementary Table 2). PORH parameters showed less intraday and interday variation than AUC parameters. Bland-Altman plots displayed no apparent trend or evidence of systematic bias (Supplementary Figures 2 and 3). Thus, all LDF parameters showed a high repeatability with good within-observer reproducibility in our study.

Between-group comparisons in LDF measurements are shown in Table 1. AUCH CVC was significantly decreased in both groups of obese patients compared with lean control subjects. There were no significant differences in this variable between the two groups of obese patients (P=0.958).

Table 1 Laser Doppler results related to post-occlusive reactive hyperemia in obese patients before surgery and control lean subjects
Full size table

A total of 62 patients (53 women; age range 19–65 years) completed the 12-month follow-up assessment. Clinical and metabolic characteristics of the 62 obese patients and 30 lean control subjects are described in Table 2. All groups were matched for age and gender. As expected, there was a high prevalence of diabetes and hypertension among obese patients with MetS. Systolic blood pressure, fasting glucose, HbA1c, total cholesterol, LDLc and triglycerides concentrations were higher in obese patients with MetS compared with both obese subjects without MetS and lean subjects. Diastolic blood pressure, heart rate and fasting insulin were higher in the two obese groups. Furthermore, PAI-1, CRP and oxLDL levels were higher in obese patients when compared with lean control subjects.

Table 2 Clinical and metabolic characteristics of obese patients before surgery and control subjects
Full size table

One-year follow-up data for all patients are shown in Table 3. A total of 45 and 17 patients underwent laparoscopic sleeve gastrectomy and Roux-en-Y gastric bypass, respectively. No differences in weight loss between the different types of bariatric surgery were observed (P=0.223). The disparity between male and female patients in the current study (85.5% women) is in agreement with studies reporting that women seek bariatric surgery more often than men.24 However, no significant differences were found among the three study groups in terms of gender and type of surgery (P=0.798 and P=0.963, respectively). Resolution of MetS was observed in 27 patients. Bariatric surgery reduced significantly anthropometric values in all patients. However, patients with MetS after surgery displayed higher BMI than patients without MetS. Higher waist and hip circumference was observed in patients with MetS after surgery compared with patients in whom MetS was resolved after surgery. Bariatric surgery resulted in a decrease in DBP and heart rate in patients in whom MetS was resolved after surgery. All patients reduced fasting insulin, HOMA-IR, HbA1c, PAI-1 and CRP levels after surgery. A significant improvement in fasting glucose, total cholesterol, LDLc and HDLc, triglycerides and oxLDL was observed in patients free of MetS but not in patients with MetS after surgery. Comparison between groups showed elevated fasting glucose, fasting insulin, HOMA-IR, HbA1c, triglycerides, CRP and oxLDL levels in the group of patients in whom MetS persisted after surgery compared with the other groups of obese patients. Low HDL cholesterol was also observed on these patients compared with the other groups.

Table 3 Clinical and metabolic characteristics of obese patients 12 months after surgery
Full size table

Bariatric surgery resulted in a significant increase in AUCH CVC in patients free of MetS, including patients that had MetS before surgery (P<0.05) (Figure 1). In contrast, AUCH CVC did not significantly change in patients in whom MetS was not resolved after surgery (P=0.72). To identify determinants of microvascular function changes after surgery, correlation analyses were performed between all clinical and biochemical variables and AUCH CVC. Increased AUCH CVC after bariatric surgery was significantly associated with an increase in HDLc (Pearson’s R=0.53, P<0.001) and a decrease in oxLDL (R=−0.54, P<0.001), fasting glucose (R=−0.27, P=0.04) and HbA1c (R=−0.29, P=0.03). When these biochemical variables were entered into a stepwise multivariate regression analysis, only changes in HDLc and oxLDL concentrations independently predicted improvement of AUCH CVC after surgery (Table 4). These two variables together accounted for 40.9% of the variability of change in AUCH CVC after surgery. Interestingly, HDLc levels correlated significantly with oxLDL (R=−0.30, P=0.03).

Figure 1
figure 1

AUCH CVC measurements in obese patients before and 1 year after bariatric surgery. Patients were classified in three groups: patients without MetS before surgery (MetS−/MetS−), patients in whom MetS had been resolved after bariatric surgery (MetS+/MetS−), and patients in whom MetS persisted after bariatric surgery (MetS+/MetS−). Data are presented as mean and s.d. **P<0.01; ***P<0.001 (Tukey’s test; within-subject comparison).

Full size image
Table 4 Association between biochemical parameters and improvement of AUCH CVC after bariatric surgery as assessed by stepwise multiple linear regression analysis
Full size table

Discussion

Our prospective study reveals an association between MetS and surgically induced improvement of microvascular dysfunction in obese patients. Thus, our results are in agreement with studies showing a functional association between microvascular dysfunction and metabolic syndrome.25 Indeed, clinical and experimental evidence suggests that microvascular dysfunction may have a central role in the development of obesity-associated hypertension and insulin resistance.3,26

To assess microvascular function, we utilized PORH at forearm skin evaluated by LDF, a method widely used in vascular research.5 PORH has been shown to be fairly reproducible in lean subjects.27, 28, 29 Our reproducibility analysis extends these results demonstrating that all LDF parameters of our study are reproducible both intraday (within-day) and interday (between-day) in obese patients. The reproducibility of PORH parameters in our study was higher than those reported in other studies in non-obese subjects.30 To this regard, it should be noted that in our study the position of laser probe was marked so that exactly the same site was used in all measurements. This was possible due to the short period of time (24 h) between measurements, another difference between our study and previous studies30 that could explain our relatively higher reproducibility values. Several different parameters can be obtained when performing PORH.5 AUCH is a commonly used parameter that simultaneously measures velocity, intensity and duration of the hyperemia response.5 Several studies have demonstrated that AUCH is a reliable indicator of microvascular dysfunction in patients at risk of cardiovascular disease.31, 32, 33, 34 Our LDF studies performed before surgery revealed that AUCH was significantly decreased in obese patients compared with lean control subjects. These results add to the growing body of evidence that microvascular function is impaired in obese patients.1,2

Surgically induced weight loss resulted in a significant improvement in microvascular function in obese patients without MetS, including patients that had MetS before surgery. These results are in agreement with previous studies showing that microvascular dysfunction could be reversed in obese patients after successful bariatric surgery.10, 11, 12, 13, 14 However, our results reveal that obese patients with MetS after surgery, despite significant weight loss, still exhibit microvascular dysfunction. These findings suggest that MetS is a strong determinant of improvement of microvascular function associated with surgically induced weight loss. Stepwise multiple linear regression analysis revealed that low HDLc levels is the component of MetS that best predicts lack of improvement in microvascular dysfunction in obese patients.

In addition to HDLc, our regression analysis identified oxLDL as an independent predictor of improvement of microvascular function in obese patients. An increase in HDLc levels and a decrease in oxLDL levels were independently associated with improvement of microvascular function after surgery. The relationship between HDLc, oxLDL and AUCH CVC can, at least in part, explain the lack of improvement in microvascular function in patients in whom MetS persisted after surgery given that HDLc and oxLDL levels did not significantly change in this group after surgery. Although HDLc and oxLDL are known to have an important role in endothelial function,35,36 the contribution of these factors to obesity-related microvascular dysfunction has been less explored. Our results indicate that HDLc and oxLDL are good predictors of improvement in microvascular function in obese patients after surgery.

Our findings that surgically induced weight loss leads to increased HDLc levels and decreased oxLDL levels are in agreement with previous studies.36,37 Our study further extends these findings showing that HDLc levels correlate negatively with oxLDL levels in patients after bariatric surgery. These results provide a potential mechanism by which surgically induced weight loss might improve microvascular dysfunction in obesity. Experimental data obtained in human subjects38 and in animal models39 have demonstrated that HDL can counteract the inhibitory effect of oxLDL on vascular reactivity. Thus, it is tempting to speculate that bariatric surgery might induce an increase in HDLc levels leading to a reduction in oxidation of LDL and, in turn, an improvement of microvascular function. However, it is important to note that not only the levels but also the functional capacity of HDLc is important for endothelial function.40 To this regard, a recent study has shown that bariatric surgery in obese adolescents increased HDL levels but not functional capacity (determined as its ability to stimulate endothelial nitric oxide synthase activation in vitro).41 Thus, it would be important in future studies to analyze the relationship between HDL functionality and surgically induced improvement in microvascular function.

Why the patients in whom MetS persisted after surgery did not show an improvement in microvascular function remains to be determined. It should be noted that this group of patients, despite significant weight loss after surgery, still remains morbidly obese. Although this might suggest that the lack of microvascular improvement in this group of patients is due to insufficient weight loss, no independent association was found between the degree of surgically induced weight loss and microvascular function in the whole set of obese patients. Nevertheless, an indirect effect of weight loss in microvascular function through changes in HDLc and oxLDL levels cannot be ruled out.

The main limitation of our study is that we cannot conclude whether the surgically induced improvement in microvascular function in obese patients is endothelium-dependent. The major mediators contributing to PORH in the skin include sensory nerves and endothelium-derived hyperpolarizing factors.4 In this regard, a recent study has shown that cytochrome epoxygenase metabolites have a major role in skin PORH.42 It would be interesting to determine the contribution of these pathways in surgically induced microvascular improvement in obese patients.

In summary, the present study shows that bariatric surgery could significantly improve microvascular dysfunction in obese patients, but only in those patients free of MetS after surgery. This improvement is strictly associated to postoperative increase in HDLc levels and decrease in oxLDL levels. Our results suggest that HDLc and oxLDL are good markers of improvement of microvascular function associated with surgically induced weight loss.