Complement activation in polycystic ovary syndrome occurs in the postprandial and fasted state and is influenced by obesity and insulin sensitivity

Abstract Objective Polycystic ovary syndrome (PCOS) is associated with metabolic risk. Complement proteins regulate inflammation and lipid clearance but their role in PCOS‐associated metabolic risk is unclear. We sought to establish whether the complement system is activated in PCOS in the fasting and postprandial state. Design Case‐control study. Patients Fasting complement levels were measured in 84 women with PCOS and 95 healthy controls. Complement activation post‐oral fat tolerance test (OFTT) was compared in 40 additional subjects (20 PCOS, 20 controls). Measurements Activation pathway (C3, C4, C3a(desArg), factor B, factor H, properdin, Factor D) and terminal pathway (C5, C5a, terminal complement complex [TCC]) proteins were measured by commercial or in‐house assays. Results Fasting C3, C3a(desArg) and TCC concentrations were increased in insulin‐resistant (adjusted differences: C3 0.13 g/L [95%CI 0‐0.25]; C3a(desArg) 319.2 ng/mL [19.5‐619]; TCC 0.66 μg/mL [0.04‐1.28]) but not in insulin‐sensitive women with PCOS. C3 and factor H levels increased with obesity. Post‐OFTT, C3 and C4 levels increased to a similar extent in PCOS subjects and controls, whist factor H levels increased more in women with PCOS compared to controls (adjusted differences (area under the curve): 12 167 μg min/mL [4942‐19 392]), particularly in the presence of concomitant obesity. Conclusions Activation and terminal complement pathway components are elevated in patients with PCOS, especially in the presence of insulin resistance and obesity.

The complement system is a key regulator of inflammation and consists of three activation pathways: classical, alternative and lectin, which converge at the level of C3 to form C3 convertases ( Figure S1). Whilst the classical and lectin pathway convertases depend on C2 and C4 cleavage, the alternative pathway convertase requires factor B and factor D. Further activation of the complement system through to the terminal pathway involves C5 cleavage and leads to formation of the membrane attack complex and its fluid-phase by-product, the terminal complement complex (TCC). Both positive and negative regulators exist, including the alternative pathway regulators properdin and factor H.
Components of the complement system, notably C3, have been shown to be increased in patients with metabolic syndrome, 3 type 2 diabetes 4 and cardiovascular disease. 5 Postprandially, C3 activation has been shown to increase lipid clearance and storage in human adipocytes 6,7 ; C3a, a product of C3 activation, is rapidly cleaved in plasma to form C3a(desArg). C3a(desArg) binds to its receptor, C5L2, on adipocytes to increase triglyceride synthesis. 6,8,9 Chylomicrons, transporters of lipids in the postprandial period, have been shown to increase C3 activation, 10 an event that is regulated in vivo by factor H. 11 Furthermore, activation pathway components (C3, C4, factor D and factor B) alter after a meal, and postprandial C3 responses differ in patients with and without cardiometabolic disease. [12][13][14][15] These findings suggest that dysregulated postprandial complement activation may influence metabolic health and contribute to the development of cardiometabolic pathology.
Only a few studies have examined the complement system in women with PCOS, finding increased levels of factor D 16 and C3a(desArg), 17,18 and increased 18,19 or no difference 20,21 in C3 levels compared to matched controls. However, many of these studies are limited by fasting measurements only and small sample sizes. We therefore sought to establish whether the complement system is activated in women with PCOS and whether any abnormalities are evident in the postprandial as well as the fasting state.

| Overall study design
We undertook the study in two parts. We firstly compared fasting plasma complement protein levels between insulin-resistant patients with PCOS (n = 84) and healthy controls (n = 95; cohort 1).
Plasma samples, maintained at −80°C, were obtained from our previous study in which detailed anthropometric, metabolic and cardiovascular phenotyping was undertaken. 22 We then compared fasting and post-oral fat tolerance test (OFTT) complement levels in PCOS women (n = 20) and healthy controls (n = 20; cohort 2), in order to determine any contribution of postprandial lipaemia to complement activation.

| Postprandial study protocol and assessment of insulin sensitivity
Cohort 2 subjects attended our Clinical Research Facility at 09:00 hours after an overnight fast. The standard OFTT meal comprised fresh cream with 40% (weight by volume; w/v) fat emulsion (polyunsaturated:saturated fat ratio of 0.10) that contained 0.001% (w/v) cholesterol and 3% (w/v) carbohydrates and had a total energy content of 3700 kcal/L. 13 The fresh cream was given at a dose of 50 g fat and 3.75 g glucose/m 2 body surface (approximately 200 mL). Subjects were allowed up to 10 minutes to consume the meal. During the test, participants remained supine and were only allowed to drink mineral water. Blood samples were obtained at 0 (fasting baseline), 30, 60, 120, 180 and 240 minutes after consuming the meal, collected into sodium EDTA (2 mg/mL) then centrifuged immediately for 15 minutes at 800 g at 4°C. Plasma was separated and stored in aliquots at −80°C until analysis. The areas under the curves (AUC) for triglycerides and complement components were calculated using the trapezoid method.
On a separate day, after an overnight fast, subjects in cohort 2 underwent basal sampling for measurement of lipids and testosterone. Subjects subsequently underwent a standard 75 g oral glucose tolerance test. Glucose and insulin were measured at 0, 30, 60, 90 and 120 minutes. The AUCs for insulin and glucose were calculated using the trapezoid method. The homeostatic model assessment method was also used to estimate insulin resistance (HOMA-IR).

| Anthropometric measurements
Height, weight, waist and hip circumference were measured according to our published protocol. 23 Abdominal subcutaneous (SF) and visceral (VF) fat areas were measured by X-ray computed tomography (CT; Hawkeye, GE Medical Systems) as previously described. 23 CT images were segmented into fat and nonfat areas according to our previously published protocol. 23

| Biochemical analysis
Plasma total cholesterol, HDL and triglyceride levels were measured using an Aeroset automated analyser (Abbott Diagnostics).
Insulin levels were assessed using an immunometric assay specific for human insulin (Invitron), and glucose was measured using the Aeroset chemistry system (Abbott Diagnostics). Total testosterone was measured by liquid chromatography-tandem mass spectrometry (Quattro™ Premier XE triple quadrupole tandem mass spectrometer; Waters Ltd). Plasma C3 and C4 levels were quantified by nephelometry on a Beckman BN11 nephelometer in the University Hospital of Wales Clinical Immunology laboratory using commercial standards. The assay working range for C3 was 0.02-4.1 g/L, and for C4 was 0.01-1.9 g/L. C5a(desArg), C3a(desArg), factor D and properdin were quantified using their respective commercial assays from Hycult Biotech, as instructed by the manufacturer.
Plasma C5, TCC and factor H were all measured using in-house ELISA. [24][25][26] All assays used purified protein (either C5, TCC or factor H) as a standard. For C5 ELISA, plates were coated with an in-house polyclonal rabbit antihuman C5 antibody (8 μg/mL,  HRP-labelled affinity-purified rabbit antihuman factor H (100 µL; 1 mg/L) was used to detect total factor H. Of note, many papers quote higher levels of serum factor H; the extinction coefficient for factor H standards on which our normal range is based has been validated previously. 26

| Statistical analysis
Linear regression models were used to assess differences in mean bi-

| Fasting complement concentrations in insulinresistant PCOS subjects and healthy controls
The clinical, metabolic and anthropometric characteristics of cohort 1 are shown in Table SI1. As anticipated, PCOS women had higher androgen levels and worse insulin sensitivity, even after adjustment for age and BMI. Table 1 shows the concentration of plasma complement components and activation products in PCOS subjects and controls, before and after adjustment. Plasma C3, C3a(desArg), C3a(desArg)/C3 ratio and TCC levels were significantly increased in the PCOS group compared to controls, even after adjustment.
Factor B, factor H and factor D were all significantly increased in PCOS before, but not after, adjustment for BMI, age and smoking.
Conversely, properdin was significantly increased in the PCOS group but only after adjusting for age, BMI and smoking. Fasting levels of C3, C3a(desArg), C4, FH and TCC were unaffected by oral contraceptive status (data not shown). Figure 1 shows the fasting plasma concentrations of C3, C3a(desArg) and TCC in PCOS subjects and healthy controls, stratified according to BMI. C3 levels increased across BMI categories in both groups but between-group differences were only apparent in obesity (mean difference ± SEM; 0.22 ± 0.1 g/L; P < .05). In contrast, C3a(desArg) and TCC levels were not affected by BMI.

| Associations of fasting complement concentrations and metabolic parameters
The associations between circulating complement levels and a range of anthropometric and metabolic risk measures are shown in Table S2.
Across the whole cohort (PCOS and controls), triglycerides, and to a lesser extent LDL cholesterol, correlated strongly with early complement pathway components and activation products but weakly with terminal components and activation products (C5, C5a(desArg) and TCC). Both early and late complement pathway components were significantly associated with both visceral and subcutaneous fat area. HOMA-IR was most strongly associated with C3, factor H and properdin.

| Effect of postprandial lipaemia on plasma complement concentrations in PCOS subjects and controls
To compare the effects of postprandial lipaemia on complement secretion, PCOS subjects and healthy volunteers in cohort There was a significant difference in the postprandial factor H response between PCOS and control groups ( Figure 2B; Table S5).
In controls, factor H levels fell sharply from baseline during the first hour post-OFTT (mean ± SEM; 250.9 ± 21.49 µg/mL vs 173.6 ± 15.26 µg/mL). In contrast, in the PCOS group, factor H levels increased 30 minutes post-OFTT. Consequently, the AUC for factor H was significantly different between PCOS and controls (Table S5).
Postprandial TCC, C3 and C4 ( Figure 2C-E), but not properdin or C3a(desArg; Figure 2F-G), levels increased from baseline in both controls and PCOS subjects, but no differences were observed between groups (Table S5).

| Effect of obesity on the complement response to OFTT
We subsequently divided the PCOS and control groups into 'obese' (BMI ≥ 30 kg/m 2 ) and 'non-obese' (BMI < 30 kg/m 2 ) groups and compared the effects of lipaemia on plasma levels of factor H, TCC and C3 in these groups ( Figure 3A-C, respectively  Figure 3C).

| D ISCUSS I ON
In this analysis of complement proteins and activation products in young women, we find evidence of complement dysregulation in concentrations were not different in our unadjusted analyses, properdin levels were marginally lower after adjustment for BMI and age.
In contrast, fasting factor H concentrations were higher in PCOS subjects only before adjustment for the group differences in BMI.
This observation is consistent with previous cross-sectional and longitudinal studies which have shown positive associations between factor H concentrations and adiposity. 30,35 We also found elevated TCC concentrations in women with PCOS, indicating increased terminal pathway activation. Increased TCC levels are likely caused by alternative pathway dysregulation since inhibition of the alternative pathway reduces TCC concentrations by >80%. 36,37 In view of the differences we observed in fasting complement protein levels of both activation and terminal pathways, we subsequently sought to compare the kinetics of complement protein generation in response to a fatty meal. We were keen to focus on the impact of lipaemia since postprandial triglycerides predict cardiovascular risk better than fasting levels, 38  In light of our findings of complement dysregulation in women with PCOS, it would be interesting to explore whether therapeutic strategies to reduce alternative pathway activation might alleviate or prevent complications associated with PCOS such as metabolic syndrome, type 2 diabetes and cardiovascular disease. In this context, it is interesting to note that C3a(desArg) levels are reduced by weight loss 46 and physical activity 47 in obesity, whilst C3 and C3a(desArg) levels fall in response to metformin therapy in women with PCOS. 18 Statins have also been shown to reduce C3 and C3a(desArg) levels in patients with cardiovascular risk, both in the fasting and postprandial state. 15,48,49 Additional studies examining the effects of new compounds that selectively reduce alternative complement pathway activation 50 would be of therapeutic interest, not only in PCOS but also in other cardiometabolic disorders associated with enhanced complement activation. However, any such treatments would need a careful evaluation of safety, especially with regard to fertility and pregnancy, in this young, reproductive age population.
Our study has both strengths and limitations. To our knowledge, this is the most comprehensive analysis to date of the complement system in patients with PCOS, benefiting from careful anthropometric and metabolic characterization in addition to measurements in the postprandial as well as the fasting state. Nevertheless, whilst we intentionally sought to compare postprandial complement activation kinetics in a carefully matched population of PCOS subjects and controls, our study was limited by the absence of an additional group of insulin-resistant women with PCOS. We were also limited to a 4-hour study window after the fat challenge.
In summary, we demonstrate evidence of complement activation and dysregulation in women with PCOS which is exacerbated in the presence of obesity and insulin resistance. We show that this extends to the terminal pathway and is evident in the postprandial as well as the fasting state. These disturbances have implications for lipid clearance, inflammation and insulin sensitivity and suggest that studies are needed to explore whether interventions aimed at regulating complement activation in PCOS may be helpful in reducing cardiometabolic risk.

ACK N OWLED G EM ENTS
We thank Professor Stephen Luzio for insulin measurements at the

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.