Methods

The impact of pharmacotherapies for dyslipidemia on circulating PCSK9 levels was assessed in 3 distinct populations, each enrolled at the University of Pennsylvania using institutional review board-approved protocols. Study 1 included 74 subjects with available samples from a previously reported randomized controlled trial involving 120 patients with hyperlipidemia randomly assigned to treatment with 10 mg of atorvastatin daily (QD), 80 mg of atorvastatin QD, or placebo. PCSK9 concentrations were assessed at both baseline and after 16 weeks of therapy.

A second study was derived from a previously described randomized controlled trial ( ClinicalTrials.gov NCT00307307 ) involving participants with carotid atherosclerosis randomized to simvastatin 20 mg QD plus placebo, simvastatin 80 mg QD plus placebo, or simvastatin 20 mg plus extended-release (ER) niacin, titrated up to 2 g QD. PCSK9 levels were measured in 70 of 75 total participants at baseline and after 12 months of therapy.

Study 3 was an open-label lipid kinetics study involving 19 dyslipidemic patients treated with 4 weeks of atorvastatin 10 mg QD followed by sequential addition of fenofibric acid (ABT335) of 135 mg QD for 8 weeks and finally ER niacin titrated to a dose of 2 g QD for 10 weeks ( ClinicalTrials.gov NCT00728910 ). Inclusion criteria included low high-density lipoprotein cholesterol (HDL-C; defined as ≤40 mg/dl in men or ≤50 mg/dl in women and ≤42 mg/dl in men or ≤52 mg/dl in women on statin therapy) and elevated triglycerides (TG/HDL ratio ≥3.5). Participants with LDL-C >190 mg/dl were excluded. PCSK9 concentrations were assessed at each of the 3 study visits.

Fasting plasma samples were obtained as part of the study protocol at the previously referenced time points and stored at −80°C. Lipid subfraction, apolipoproteins levels, and high-sensitivity C-reactive protein (hsCRP) were assayed using enzymatic techniques as previously described. The plasma PCSK9 concentrations in all 3 studies were measured using a validated sandwich enzyme-linked immunosorbent assay designed to recognize free and LDL-bound PCSK9 (ELISA; R&D Systems Inc, Minneapolis, Minnesota). All plasma specimens were assayed in singlicate and controls in duplicates. Intra-assay and inter-assay coefficients of variation for the control samples were 5.7% and 9.7%, respectively.

Categorical variables are presented as frequencies and percentages, and continuous variables are presented as means and SDs or medians and interquartile ranges (IQR) for variables with skewed distributions. The association of PCSK9 with circulating biomarkers was assessed with the use of Pearson’s correlation coefficients. Paired t tests were used to assess the effect of pharmacologic interventions on PCSK9 concentrations. These changes were compared across treatment arms with the use of analysis of covariance, which included the patient’s baseline value and the treatment group as covariates. All reported p values are 2-tailed, with a p value of 0.05 indicating statistical significance.

Results

In study 1, PCSK9 measurements were available in 74 hypercholesterolemic patients treated for 16 weeks with atorvastatin 10 mg QD (n = 26), atorvastatin 80 mg QD (n = 25), or placebo (n = 23). Baseline laboratory values, expressed as mean ± SD were total cholesterol (TC) 242 ± 32 mg/dl, HDL-C 47 ± 11 mg/dl, LDL-C 164 ± 33 mg/dl, apolipoprotein B 110 ± 13 mg/dl, and PCSK9 290 ± 80 ng/ml. Median (IQR) values included TGs 128 (88 to 161) mg/dl and hsCRP 1.4 (0.7 to 3.5) mg/L. No significant differences in baseline characteristics were noted across treatment groups ( Supplementary Table 1 ).

Baseline PCSK9 values were somewhat greater in female patients compared with male patients (303 vs 280 ng/ml), a finding that did not reach statistical significance. No significant relation was noted between baseline PCSK9 levels and TC, LDL-C, or apolipoprotein B ( Table 1 ). A nominally significant positive association was present between levels of PCSK9 and log-transformed levels of both TGs (r = +0.31, p = 0.007) and hsCRP (r = +0.25, p = 0.03).

As expected, statin therapy led to significant reductions in LDL-C levels with mean ± SD percent change of −32 ± 16 with atorvastatin 10 mg and −43 ± 24 with atorvastatin 80 mg compared with −8 ± 31 in the placebo arm. In contrast, PCSK9 levels were increased in a dose-dependent fashion with statin therapy compared with placebo ( Table 2 ). A modest inverse association was noted between percent change in PCSK9 level and corresponding change in LDL-C level (r = −0.24, p = 0.048; Figure 1 ). In contrast, change in PCSK9 did not predict change in either apolipoprotein B (r = −0.19, p = 0.11) or hsCRP (r = −0.09, p = 0.43).

Relation between change in treatment associated changes in LDL-C and PCSK9 levels. (A) Study 1, participant treatment with atorvastatin. (B) Study 3, ER niacin added to participants treated with atorvastatin and fenofibric acid.

Similar analyses were conducted in study 2 participants randomized to 12 months of simvastatin 20 mg/placebo (n = 24), simvastatin 80 mg/placebo (n = 22), or simvastatin 20 mg/niacin titrated up to 2 g QD (n = 24). The study population had a mean age of 70 years and included 49 (70%) men. A total of 64% of subjects were receiving statin therapy and study drug therapy was initiated without a washout period. Baseline lipid parameters were TC mean 191 ± SD 39, HDL-C 46 ± 15 mg/dl, LDL-C 117 ± 32 mg/dl, TG median 139 (IQR 92 to 170), and PCSK9 383 ± 117 ng/ml. Average PCSK9 values were 341 ng/ml for statin-naïve patients and 407 ng/ml for those maintained on chronic statin therapy (p = 0.02). Baseline characteristics were similar across treatment arms ( Supplementary Table 2 ). Baseline lipid subfraction and hsCRP levels did not predict PCSK9 levels at baseline ( Table 1 ).

As previously reported, mean changes in lipids from baseline included the following: simvastatin 20 mg (LDL-C −6%, HDL-C +3%, and TG +23%), simvastatin 80 mg (LDL-C −25%, HDL-C +7%, and TG −7%), simvastatin 20 mg/niacin (LDL-C −41%, HDL-C +24%, and TG −24%). No significant increase in PCSK9 level was noted in the simvastatin 20 mg group, consistent with the minimal reduction in LDL-C that likely reflected prevalence of baseline statin usage. Simvastatin 80 mg led to a substantial increase in mean PCSK9 level from 379 to 530 ng/ml (p <0.0001). In contrast, average PCSK9 levels decreased from 396 to 331 ng/ml with simvastatin 20/niacin therapy (p = 0.002, Table 2 ). Percent PCSK9 change was not a significant predictor of LDL-C level change in study 2 as a whole or in any treatment subgroup.

Study 3 included participants with atherogenic dyslipidemia (n = 19) on a stable dose of atorvastatin 10 mg with sequential addition of 135 mg of fenofibric acid for 8 weeks followed by 2 g ER niacin QD for 10 weeks. Mean age was 51 ± SD 10 years, with 10 men (53%). Baseline laboratory values on atorvastatin 10 mg QD were TC mean 151 ± 26 mg/dl, HDL-C 35 ± 10 mg/dl, LDL-C 85 ± 20 mg/dl, TG median 154 (IQR 121 to 204) mg/dl, and PCSK9 415 ± 86 ng/ml ( Supplementary Table 3 ). As in previous studies, minimal relation was noted between baseline biomarkers and PCSK9 levels ( Table 1 ).

Addition of fenofibric acid to atorvastatin led to average lipid changes of LDL-C +3%, HDL-C +2%, and TG −18%. The subsequent addition of ER niacin resulted in additional changes of LDL-C −14%, HDL-C +12%, and TG −27%. PCSK9 levels increased by +23% with the addition of fenofibric acid and subsequently decreased by −17% with niacin ( Table 1 ). Patients with greater percent reduction in TGs with fenofibric acid tended to have a more notable increase in PCSK9 (r = 0.42, p = 0.07). Change in LDL-C did not predict PCSK9 increase with fenofibrate. However, a significant positive relation was noted between niacin-induced changes in PCSK9 and LDL-C (r = 0.62, p = 0.006; Figure 1 ). Furthermore, PCSK9 levels on combination therapy with atorvastatin and fenofibric acid were linked to subsequent reduction with niacin in both LDL-C (r = −0.59, p = 0.009) and PCSK9 (r = 0.51, p = 0.03).