Acute impact of apheresis on oxidized phospholipids in patients with familial hypercholesterolemia.

We measured oxidized phospholipids (OxPL), lipoprotein (a) [Lp(a)], and lipoprotein-associated phospholipase A2 (Lp-PLA2) pre- and postapheresis in 18 patients with familial hypercholesterolemia (FH) and with low(∼10 mg/dl; range 10–11 mg/dl), intermediate (∼50 mg/dl; range 30–61 mg/dl), or high (>100 mg/dl; range 78–128 mg/dl) Lp(a) levels. By using enzymatic and immunoassays, the content of OxPL and Lp-PLA2 mass and activity were quantitated in lipoprotein density fractions plated in microtiter wells, as well as directly on apoB-100, Lp(a), and apoA-I immunocaptured within each fraction (i.e., OxPL/apoB and Lp-PLA2/apoB). In whole fractions, OxPL was primarily detected in the Lp(a)-containing fractions, whereas Lp-PLA2 was primarily detected in the small, dense LDL and light Lp(a) range. In lipoprotein capture assays, OxPL/apoB and OxPL/apo(a) increased proportionally with increasing Lp(a) levels. Lp-PLA2/apoB and Lp-PLA2/apoA-I levels were highest in the low Lp(a) group but decreased proportionally with increasing Lp(a) levels. Lp-PLA2/apo(a) was lowest in patients with low Lp(a) levels and increased proportionally with increasing Lp(a) levels. Apheresis significantly reduced levels of OxPL and Lp-PLA2 on apoB and Lp(a) (50–75%), particularly in patients with intermediate and high Lp(a) levels. In contrast, apheresis increased Lp-PLA2-specific activity (activity/mass ratio) in buoyant LDL fractions. The impact of apheresis on Lp(a), OxPL, and Lp-PLA2 provides insights into its therapeutic benefits beyond lowering apoB-containing lipoproteins.

were selected for this study ( 17 ). For the purpose of this study, patients were a priori recruited into three groups according to their Lp(a) levels: (low 10 mg/dl; range, 10-11 mg/dl), intermediate ( ‫ف‬ 50 mg/dl; range 30-61 mg/dl), and high (>100 mg/dl; range, 78-128 mg/dl). The diagnosis of FH was validated by DNA analysis in all patients, except for two in which no mutation was found. DNA sequencing was extended to LDLR , APOB , and PCSK9 but not to ARH genes. However, all patients had phenotypic FH by family history. Four patients were homozygous for the LDL-R gene mutation, 6 were heterozygous, and 5 were compound heterozygotes with two mutations of the LDLR gene. One patient was a double heterozygote (mutations in LDLR gene and APOB 3500). All patients were treated once daily by lipidlowering drugs in combination therapy: atorvastatin/ezetimibe (80 mg/10 mg; n = 16), simvastatin/ezetimibe (80 mg/10 mg; n = 1), or rosuvastatin/ezetimibe (20 mg/10 mg; n = 1).
Blood samples were obtained pre-LDL apheresis before coupling the instrument to the cephalic vein and immediately post-LDL apheresis, and were collected into sterile EDTA-containing tubes (fi nal concentration 1 mg/ml). Plasma was immediately separated from blood cells by low-speed centrifugation at 2,500 rpm for 20 min at 4°C and frozen at Ϫ 80°C until analysis. The study was performed in accordance with the ethical principles set forth in the Declaration of Helsinki. Written informed consent was obtained from all subjects.

LDL-apheresis
Three types of columns were used for LDL-apheresis (see Table 1 ). To avoid blood coagulation in the column, 3,000-13,000 units of unfractionated heparin were added during preliminary rinsing of the column and prior to coupling the machine to the patient. Some patients (n = 4) received a heparin bolus (1,000-2,000 U) at the beginning of the procedure and a supplemental heparin infusion during the procedure (500 U/h for three of the four patients and 2000 U/h for one). The duration of each LDL-apheresis session was >2 h.

Isolation of lipoprotein density fractions
Lipoproteins were isolated from plasma by a single-step, nondenaturing density gradient ultracentrifugation in a Beckman SW 41 Ti rotor at 40,000 rpm for 44 h in a Beckman L80 at 15°C by method of Chapman et al., with a slight modifi cation as previously described ( 8,17 ( 3,5,6 ). Lp(a) is potentially more atherogenic than native LDL because it binds with increased affi nity to arterial intimal proteoglycans, resulting in delivery of both atherogenic LDL and proinfl ammatory OxPL. OxPL, oxidized LDL, and Lp(a) trigger apoptosis in endoplasmic reticulum (ER)-stressed macrophages through CD36 and Toll-like receptor 2, suggesting a mechanism through which OxPL/Lp(a) may destabilize atherosclerotic plaques ( 9 ).
OxPL on Lp(a) may be modulated through Lp-PLA 2 , which cleaves OxPL into lysophosphatidylcholine (lysoPC) and a free oxidized fatty acid ( 10,11 ). In normolipidemic individuals, Lp-PLA 2 is mainly associated with LDL and is preferentially associated with atherogenic small dense LDL ( 12 ). In primary hypercholesterolemia, the LDL-associated Lp-PLA 2 is increased in parallel with the severity of hypercholesterolemia, the highest levels being present in homozygous familial hypercholesterolemia (FH) ( 11 ). Interestingly, in individuals with elevated Lp(a) levels, Lp-PLA 2 activity is higher on Lp(a) than on equimolar amounts of LDL ( 13,14 ). Lp-PLA 2 is present in vulnerable plaques, and increased Lp-PLA 2 levels are associated with increased CVD risk in both primary and secondary populations ( 15 ).
Patients with FH generally have signifi cantly higher Lp(a) levels than patients without FH ( 16 ), despite the fact that the LDL receptor is not thought to be involved in its clearance. Lp(a) and its associated proinfl ammatory OxPL and Lp-PLA 2 may be particularly atherogenic and lead to increased CV events. In view of the pathophysiological relationship of OxPL, Lp(a), and Lp-PLA 2 , we assessed the acute changes and distribution of apheresis on these atherogenic moieties in patients with FH as a potential mechanism of clinical benefi t.

Study subjects with varying Lp(a) levels and blood sampling
Eighteen patients with Type IIa FH (10 men and 8 women) from the Haemobiotherapy Unit at Pitie-Salpetriere University Hospital in Paris undergoing LDL-apheresis every 2 or 3 weeks saturating the plates ( Fig. 2 , top panels). The assays were set up in this manner to be able to consistently observe changes in all measurements pre-and postapheresis.

Baseline demographic and biochemical characteristics of the study groups
Baseline characteristics of the three study groups on fi rst presentation and segregated a priori into low ( ‫ف‬ 10 mg/dl, range 7-15 mg/dl), intermediate ( ‫ف‬ 50 mg/dl, range 29-81 mg/dl) or high (>100 mg/dl, range 52-138 mg/dl) Lp(a) levels are shown in Table 1 . These subjects are relatively young-to-middle-aged adults, have very high baseline LDL-C levels, and have been on chronic apheresis for at least two years with a large number of sessions. The effi cacy on the lipid profi le achieved by these apheresis techniques is consistent with fi ndings in the literature ( 17,18 ). Table 2 reports the biochemical variables on blood samples of patients on chronic apheresis that were recruited at the same time for this study. As expected, apheresis led to signifi cant reduction in total cholesterol and LDL-C, apoB, triglycerides, and hsCRP in all Lp(a) groups ( Table 2 ). HDL-C and ApoA-I were decreased by ‫ف‬ 15% in each Lp(a) group. Lp(a) levels were signifi cantly reduced in the intermediate and high Lp(a) groups but not in the low Lp(a) group.

Determination of Lp-PLA 2 mass and activity in plasma and density fractions and sPLA 2 mass and activity in plasma
Plasma Lp-PLA 2 mass in plasma and density gradients were measured in a blinded fashion as previously described (PLAC® Test, diaDexus, Inc.) ( 20,21 ). The assay has a lower detection limit of 2 ng/ml and an interassay coeffi cient of variation (CV) of 6-7%. A threshold of >200 ng/ml is considered elevated ( 22 ). Lp-PLA 2 activity was measured in a 96-well microplate with a colorimetric substrate that is converted on hydrolysis by the phospholipase enzyme (CAM™ assay, diaDexus, Inc.). The specifi c activity of Lp-PLA 2 as the Lp-PLA 2 activity/mass ratio was determined by dividing the Lp-PLA 2 activity by the Lp-PLA 2 mass. Serum sPLA 2 type IIA mass and activity levels were measured only in plasma at the Paris Cardiovascular Research Center, as previously described ( 20,21 ).

Determination of the distribution of apoB-100, apo(a), OxPL, and Lp-PLA 2 mass in density gradient fractions by direct plating of fractions on microtiter well plates
To determine the distribution of apoB, apo(a), OxPL, and Lp-PLA 2 mass in each density fraction directly on microtiter well plates, we incubated an aliquot containing a saturating amount of each fraction (1:200 dilution for apoB, 1:100 for apo(a) and OxPL, and 1:20 for Lp-PLA 2 ) on microtiter wells overnight at 4°C. The content of apoB-100, apo(a), OxPL, and Lp-PLA 2 were determined with the biotinylated secondary antibodies goat antihuman apoB-100 (Pierce), LPA4, E06, and 4B4, respectively, using ELISA ( 3 ), as illustrated in Fig. 1A . Determination of Lp-PLA 2 mass by direct plating was used as a second complementary technique to the diaDexus method. Each sample was assayed in triplicate, and data are expressed as relative light units (RLU) per 100 milliseconds.

Determination of the presence of OxPL and Lp-PLA 2 mass directly on isolated apoB, apo(a), and apoA-I
The content of OxPL ( Fig. 1B ) or Lp-PLA 2 mass ( Fig. 1C ) on isolated apoB-100, apo(a) and apoA-I captured on microtiter well plates with specifi c antibodies were determined by chemiluminescent sandwich ELISA in a similar format as above. For this set of assays, the antibodies MB47, LPA4, and guinea pig anti-human apoA-I were plated on microtiter well plates (5 mg/ml) to capture apoB-100, Lp(a), and apoA-I, respectively. A fraction (1:25 dilution) of each aliquot was added, the unbound plasma was washed off, and the amount of OxPL or Lp-PLA 2 was determined with biotinylated E06 or 4B4, respectively. The assays for OxPL or Lp-PLA 2 present on apoB-100 are reported as OxPL/apoB and Lp-PLA 2 /apoB, on Lp(a) as OxPL/apo(a) and Lp-PLA 2 /apo(a), and on apoA-I as OxPL/apoA-I and Lp-PLA 2 /apoA-I to denote the detection of OxPL or Lp-PLA 2 on the captured lipoproteins. In this format, the amount of apoB-100, Lp(a), and apoA-I captured generally refl ects their content in each aliquot (i.e., nonsaturating). The only exception to this is that because apoB is at a large excess compared with Lp(a) and apoA-I, it was generally  ( Table 2 ). When all three Lp(a) groups were combined, there was a reduction in OxPL/apoB (8,259 ± 7,299 versus 3,913 ± 2,424 RLU, P = 0.003). OxPL/ apo(a) was signifi cantly reduced in all groups, and OxPL/ Lp-PLA 2 mass and activity were signifi cantly reduced in all three groups following apheresis ( Table 2 ). As a comparison, sPLA 2 mass, which is not known to circulate signifi cantly with apoB-containing lipoproteins, was only sig nifi cantly reduced in the high Lp(a) group. There were no signifi cant differences in sPLA 2 activity pre-and postapheresis. When all three Lp(a) groups were combined, there was a reduction in sPLA 2 mass (4.1 ± 7.7 versus 3.1 ± 8.4, P < 0.001) but not sPLA 2 activity (57.6 ± 30.1 versus 62.7 ± 33.7, P = 0.12).  range of both small dense LDL, as previously shown for Lp-PLA 2 activity and Lp(a) ( 12,13,23 ). There appeared to be a similar amount of Lp-PLA 2 mass preapheresis in the low and intermediate Lp(a) groups, whereas the levels were lower in the high Lp(a) groups. Interestingly, Lp-PLA 2 mass was also present in the very dense fractions 23-25, representing densities 1.163 → 1.189 g/ml . This is suggestive of dissociation of a fraction of Lp-PLA 2 from apoB lipoproteins as previously described for separation of lipoproteins using ultracentrifugation compared with FPLC ( 24 ). Lp-PLA 2 activity appeared qualitatively similarly distributed among the three groups and in a similar pattern to Lp-PLA 2 mass, although only two patients in each group were available for measurements. Using the diaDexus sandwich Lp-PLA 2 mass and activity ELISA ( Fig. 3 ), qualitatively similar information was noted compared with the direct-plating assays shown in Fig. 2 . In this analysis, we were able to calculate the specifi c activity of Lp-PLA 2 as the Lp-PLA 2 activity/mass ratio. The specifi c activity was highest in fractions 8-10, which correspond to fractions between LDL-3 and LDL-4 [LDL-3 (d = 1.029-1.039 g/ml), LDL-4 (d = 1.039-1.050 g/ml)], which is in accordance with prior fi ndings showing that despite the high Lp-PLA 2 mass and activity associated with LDL-5 [LDL-5 (d = 1.050-1.063 g/ml)], the enzyme-specifi c activity (activity/mass) in LDL-5 is the lowest among all LDL subfractions ( 12,23 ). This is also apoA-I levels were very low [approximately 40-fold lower than OxPL/apo(a)] and were not affected by apheresis.

ApoB, apo(a), OxPL, and Lp-PLA 2 mass and activity on individual density gradient fractions measured by direct plating on microtiter well plates: effect of apheresis
In patients with low, intermediate, and high Lp(a) levels, apoB is mainly present in fractions 5-12 corresponding to density range 1.019-1.059 g/ml ( Fig. 2 ). At this dilution, apoB was plated to be saturating to provide a comparison for other variables. Thus, the apoB values refl ect the content on the plate rather than the plasma content. Apo(a) and OxPL, which are not saturating at the dilutions plated, are present in fractions 10-20 (density range 1.038-1.139 g/ml) and 11-17 (density range 1.044-1.105 g/ml), respectively. In the low Lp(a) group, OxPL are minimally present. In the intermediate and high Lp(a) groups, the level of apo(a) peaks at tubes 12-14, which represents density range of 1.051-1.075 g/ml. Note that all of the OxPL immunoreactivity coincides with the apo(a) peak and not with the main apoB peak. As the Lp(a) levels increase among groups, the OxPL peaks increase in size accordingly.
Distribution and changes in Lp-PLA 2 mass pre-and postapheresis are shown in the bottom panels of Fig. 2 . Lp-PLA 2 mass was present primarily in fractions 9-16 corresponding to densities 1.033-1.094 g/ml, which represents the

DISCUSSION
Patients with heterozygous FH have higher Lp(a) levels than non-FH patients ( 16 ) and represent a worldwide prevalence of about ten million individuals. Homozygous FH is thought to be present in ten to fi fteen thousand subjects ( 25 ). In this study, apheresis signifi cantly reduced levels of proatherogenic and proinfl ammatory OxPL and Lp-PLA 2 on apoB and Lp(a), particularly in patients with high Lp(a) levels, while increasing Lp-PLA 2 -specifi c activity in buoyant LDL particles. These data provide a scientifi c rationale for understanding how apheresis may provide clinical benefi ts beyond lowering apoB-containing lipoproteins ( 26 ) and in understanding the pathophysiological relationships among Lp(a), OxPL, and Lp-PLA 2 in collectively mediating CVD ( 5,6,9 ).
The clinical relevance of these fi ndings is refl ected by the fact that Lp(a) is now accepted as an independent genetic risk factor of CVD ( 1,2 ), that Lp(a) is a carrier of OxPL and Lp-PLA 2 ( 3,4 ), and that apheresis is an accepted treatment for patients with FH and high Lp(a) levels ( 18 ). This study also demonstrates several novel insights into the potential pathophysiogical role of OxPL, Lp(a), and Lp-PLA 2 in FH: i ) OxPL were almost exclusively detected in the Lp(a)-containing fractions and not on HDL ApoA-I particles; ii ) OxPL/apoB and OxPL/apo(a) increased proportionally with increasing Lp(a) levels; iii ) Lp-PLA 2 /apoB and Lp-PLA 2 /apoA-I levels were highest in patients with low Lp(a) but decreased proportionally with increasing Lp(a) levels; and iv ) Lp-PLA 2 /apo(a) was lowest in patients with low Lp(a) levels and increased proportionally with increasing Lp(a) levels.
Three retrospective, hypothesis-generating studies ( 26-28 ) demonstrated that maximally tolerated medications plus apheresis signifi cantly reduced rate of major cardiac adverse events compared with maximally tolerated medications alone. Apheresis has been shown to signifi cantly reduce apparent in Fig. 2 using the direct-plating method where Lp-PLA 2 mass in fractions 10 and higher almost disappear postapheresis.
Following apheresis, reductions ( ‫ف‬ 50-75%) were noted in apoB, apo(a), OxPL, and Lp-PLA 2 mass, and these changes were most pronounced in the intermediate and high Lp(a) groups ( Figs. 2 and 3 ). Following apheresis, Lp-PLA 2 -specifi c activity actually increased in the intermediate and high Lp(a) groups. This may denote that, after apheresis, the remaining LDL particles are defi cient in small dense LDL particles and enriched in large buoyant particles.

Distribution of OxPL levels on isolated apoB, apo(a), and apoA-I lipoproteins in FH patients: effect of apheresis
OxPL/apoB and OxPL/apo(a) levels progressively increased in parallel with increasing Lp(a) levels, being present primarily in fractions 11-15, representing densities of 1.044-1.084 g/ml ( Fig. 4 ). In contrast, minor amounts of OxPL were detected on apoA-I particles. Following apheresis, reductions were noted in both OxPL/apoB and OxPL/apo(a).

Distribution of levels of Lp-PLA 2 on isolated lipoproteins apoB, apo(a), and apoA-I in FH patients: effect of apheresis
Preapheresis, Lp-PLA 2 /apoB and Lp-PLA 2 /apoA-I were highest in the low Lp(a) group, but they diminished with increasing Lp(a) levels ( Fig. 5 ). In contrast, Lp-PLA 2 / apo(a) was lowest in the low Lp(a) group and increased with increasing Lp(a) levels. Interestingly, a signifi cant amount of Lp-PLA 2 /apo(a) and Lp-PLA 2 /apoA-I were noted in the most dense fractions, suggesting the presence of Lp-PLA 2 mass on poorly lipidated lipoproteins. Following apheresis, reductions were noted in Lp-PLA 2 /apoB and Lp-PLA 2 /apo(a), with more modest changes in Lp-PLA 2 /apoA-I. to ultimately test the hypothesis that lowering elevated Lp(a) a priori results in clinical benefi ts.
We also demonstrate that the majority of immunologically detected OxPL are physically present on Lp(a) particles and not signifi cantly on non-Lp(a)-apoB particles or apoA-I particles. The fact that E06-detectable OxPL are associated with clinical events ( 3 ) suggests that they refl ect a pathophysiologically important and clinically relevant biological activity of Lp(a). In corroborating evidence, we have also shown that OxPL/apoB levels primarily refl ect OxPL associated with high Lp(a) levels and small apo(a) isoforms, i.e., the most atherogenic particles ( 37 ). Because apo(a) isoforms are not normally measured clinically, the OxPL/apoB measurement provides a unique biological assay that summates the cardiovascular risk of both LPA alleles ( 38 ). In preliminary data, we have documented that OxPL are present on Lp(a) in two compartments, one covalently bound to apo(a) and another noncovalently bound in the lipid phase of Lp(a). Further work needs to be performed to understand the relative proportion of OxPL in each component of Lp(a), particularly in different disease states, to defi ne the sites on apo(a)/Lp(a) where OxPL are present, and to determine the specifi c species of OxPL responsible for clinical risk.
The distribution of Lp-PLA 2 mass and activity was shown to be in the small dense LDL fraction, as previously documented by several studies ( 12,23 ), along with a signifi cant Lp(a) concentrations ( 17,18,29,30 ), and a small study suggested it reduces Lp-PLA 2 mass ( 31 ). In this study, we demonstrate the powerful effect of apheresis on reduction of OxPL and Lp-PLA 2 mass and activity physically present on associated apoB-containing lipoproteins. The basis of this decrease is likely due to the fact that the lipoprotein carriers of OxPL and Lp-PLA 2 , Lp(a) and apoB, respectively, also decreased in relative proportion following apheresis. Interestingly, the Lp-PLA 2 -specifi c activity increased in the intermediate and high Lp(a) groups, suggesting that the remaining LDL particles are defi cient in small dense LDL particles with the lowest specifi c Lp-PLA 2 activity and enriched in large buoyant particles exhibiting higher specifi c activity. These data are also consistent with previous data showing that apheresis enhances endotheliumdependent vasodilation in the brachial ( 32 ) and coronary arteries ( 33 ).
Treatments for elevated Lp(a) levels include niacin and the emerging cholesterol ester transport and PCSK9 inhibitors, although their underlying mechanisms of action are not well defi ned. Antisense oligonucleotides to apoB have shown promise in signifi cantly reducing both LDL-C and Lp(a) levels ( 34,35 ). In addition, specifi c antisense oligonucleotides to apo(a) have been shown to reduce apo(a) and their associated OxPL by ‫ف‬ 85% in apo(a)transgenic mice ( 36 ). As they do not affect other lipoprotein levels to any great extent, it may be clinically feasible portion present on lipoproteins present in the poorly lipidated densities. Interestingly, as the Lp(a) levels change, the proportion of Lp-PLA 2 on these particles changes signifi cantly with the underlying lipid content. For example, in patients with low Lp(a) levels, the Lp-PLA 2 is primarily present on apoB. However, as the Lp(a) levels increase, there is proportionately less Lp-PLA 2 on apoB and more on Lp(a). This is consistent with the fact that on a molar basis, Lp(a) contains more Lp-PLA 2 mass (1.5-2 times) and activity (7 times) than LDL ( 13,14 ). However, as LDL is in excess of Lp(a) in most patients, there is globally more Lp-PLA 2 on LDL than Lp(a).
In conclusion, these fi ndings describe a pathophysiological relationship among Lp(a), OxPL, and Lp-PLA 2 in patients with FH and suggest that additional benefi ts of apheresis, beyond lowering apoB-containing lipoproteins, may be to reduce their associated proinfl ammatory OxPL and Lp-PLA 2 .