Antisense inhibition of apolipoprotein (a) to lower plasma lipoprotein (a) levels in humans

Epidemiological, genetic association, and Mendelian randomization studies have provided strong evidence that lipoprotein (a) [Lp(a)] is an independent causal risk factor for CVD, including myocardial infarction, stroke, peripheral arterial disease, and calcific aortic valve stenosis. Lp(a) levels >50 mg/dl are highly prevalent (20% of the general population) and are overrepresented in patients with CVD and aortic stenosis. These data support the notion that Lp(a) should be a target of therapy for CVD event reduction and to reduce progression of aortic stenosis. However, effective therapies to specifically reduce plasma Lp(a) levels are lacking. Recent animal and human studies have shown that Lp(a) can be specifically targeted with second generation antisense oligonucleotides (ASOs) that inhibit apo(a) mRNA translation. In apo(a) transgenic mice, an apo(a) ASO reduced plasma apo(a)/Lp(a) levels and their associated oxidized phospholipid (OxPL) levels by 86 and 93%, respectively. In cynomolgus monkeys, a second generation apo(a) ASO, ISIS-APO(a)Rx, significantly reduced hepatic apo(a) mRNA expression and plasma Lp(a) levels by >80%. Finally, in a phase I study in normal volunteers, ISIS-APO(a)Rx ASO reduced Lp(a) levels and their associated OxPL levels up to 89 and 93%, respectively, with minimal effects on other lipoproteins. ISIS-APO(a)Rx represents the first specific and potent drug in clinical development to lower Lp(a) levels and may be beneficial in reducing CVD events and progression of calcific aortic valve stenosis.

levels ( 28 ). Because human cell lines do not adequately express the apo(a) gene, an apo(a) expression vector containing the 5 ′ -untranslated region, the signal sequence, the fi rst fi ve KIV-like repeats, and 291 bp of the kringle repeat of apo(a) driven by the cytomegalovirus promoter was cotransfected into HepG2 cells with a hemagglutinating virus of Japan (HVJ)-liposome-apo(a) DNA complex and either the apo(a) ribozyme or control oligonucleotide . After 72 h, the apo(a) ribozyme was shown to reduce HepG2 protein secretion by approximately 60%, while no signifi cant change in plasminogen protein was observed. While these initial results were encouraging, the requirement of liposomal formulation of the apo(a) targeting ribozyme prior to delivery would have made in vivo pharmacology studies more challenging.
Mipomersen, a second generation ASO directed to apoB-100 and approved for clinical use in the United States for lowering LDL cholesterol (LDL-C) in patients with homozygous familial hypercholesterolemia ( 29 ), has been shown to lower Lp(a) levels in Lp(a)-transgenic mice ( 30 ) and in humans (31)(32)(33)(34). In humans, four phase 3 trials were performed in 382 patients on maximally tolerated lipid-lowering therapy and randomized 2:1 to weekly subcutaneous mipomersen (200 mg) (n = 256) or placebo (n = 126) for 26 weeks. Mipomersen reduced plasma Lp(a) levels by 21-39%; whereas, no signifi cant change was noted in the placebo groups ( 35 ). Interestingly, in the mipomersen group, only modest correlations were present between percent changes in Lp(a) and apoB ( r = 0.43, P < 0.001) and Lp(a) and LDL-C ( r = 0.36, P < 0.001), suggesting mechanisms of Lp(a) lowering related to liver synthesis of apoB that are not apparent by evaluating plasma apoB levels.
A study in Lp(a)-transgenic mice by Merki et al. ( 30 ) suggested that one potential mechanism of Lp(a) reduction by mipomersen may be through limiting hepatic production of newly formed apoB concomitantly when apo(a) is available to create an Lp(a) particle. Transgenic mice overexpressing both human APOB (h-apoB)-100 plus human LPA to generate genuine Lp(a) particles [human apo(a) does not form a covalent bond with mouse apoB-100] were treated with mipomersen. Mipomersen reduced hepatic apoB production and plasma levels of h-apoB-100 to very low levels (<20 mg/dl) and reduced LDL-C and Lp(a) levels by ‫ف‬ 75%. However, the mice continued to produce similar amounts of apo(a) unbound to h-apoB as before treatment with mipomersen, suggesting that apoB-100 synthesis is a limiting factor in Lp(a) particle generation in this LPA transgenic model ( Fig. 2A ).
In a follow-up investigation, Merki et al. ( 12 ) evaluated ISIS 144367, a second generation apo(a)-specifi c ASO, in the following mouse models: 1 ) LPA transgenic mice expressing a truncated human LPA cDNA with eight KIV repeats [8K-L(a) mice] that has very high Lp(a) levels ( 11, 15,36 ). This construct contains wild-type human apo(a) cDNA encoding kringles, KIV 1 , KIV 2 , a fusion of KIV 3 and KIV 5 , KIV 6 -KIV 10 , KV, and the protease-like domain. The promoter for this construct consists of the apoE hepatic control region LE6 and apoE intron 1. 2 ) Lp(a)-transgenic promising approach in lowering Lp(a) levels in the clinical setting. ASOs represent the third major therapeutic discovery platform, distinct from small molecule and monoclonal antibody approaches, and have shown great promise in specifi c targeting of disease-associated genes in dyslipidemia, oncology, neurological dysfunction, and metabolic disorders. Due to their mode of action, by binding complementary mRNA targets via Watson-Crick base pairing, isoform-specifi c targeting is possible. In the case of Lp(a), this very important feature allows direct targeting of the apo(a) transcript without altering plasminogen transcript levels ASOs TARGETING apo(a) Second generation ASOs are single-stranded chimeric molecules generally 20 nucleotides in length, containing 2 ′ -O -(2-methoxyethyl) (MOE) modifi cations at the 5 ′ and 3 ′ termini (positions 1-5 and 16-20) and DNA-like nucleotides in the central region (position 6-15) with a phosphorothioate (P=S) backbone throughout to enhance nuclease resistance (17)(18)(19). These molecules are up to 15-fold more potent than fi rst-generation P=S only-modifi ed ASO drugs, due to their enhanced mRNA affi nity via the MOE moiety (19)(20)(21), their gapmer design supporting an RNase H1 enzymatic termination mechanism, and their improved pharmacokinetic properties that permit weekly, monthly, or potentially quarterly dosing ( Fig. 1 ). These drugs also have an improved therapeutic index due to reduced pro-infl ammatory properties (22)(23)(24)(25)(26).
Because ASO drugs are metabolized by cellular nucleases and not the cytochrome P450 system, they can be safely co-administered with traditional therapeutic agents with differing modes of action ( 20 ). Additionally, as they are hydrophilic, they may be administered in saline without special formulation via subcutaneous, intravenous, topical, aerosol, enema, intravitreal, intraventricular, intrathecal, and oral routes ( 27 ). The pharmacokinetic properties of ASOs have been extensively quantifi ed in multiple species and in man ( 22 ). Following systemic administration, the liver, kidney, bone marrow, adipose tissue, spleen, and lymph nodes accumulate the highest drug concentrations, while distribution is poor to the intestine, skeletal muscle, heart, lung, reproductive organs, pancreas, and brain. MOE-modifi ed ASOs are resistant to exonuclease degradation, resulting in prolonged tissue half-lives, ranging from 10 to 30 days. In general, ASOs are cleared from tissue by endonucleolytic degradation, producing lower molecular weight metabolites (8-12 nucleotides) that are eliminated by urinary excretion.

REVIEW OF ANTISENSE STUDIES REPORTING Lp(a) LEVELS
In the fi rst described antisense study performed in vitro, a human apo(a) ribozyme oligonucleotide containing P=S DNA and RNA was designed to specifi cally target KIV of the apo(a) mRNA without altering plasminogen transcript mice, and 86% in 12K-apo(a) mice ( Fig. 2B ). The most potent effect was documented in 12K-apo(a) mice expressing apo(a) with multiple KIV 2 repeats containing the natural LPA promoter and regulatory sequences. Importantly, in the 12K-apo(a) mice, ISIS 144367 also significantly reduced plasma-OxPLs on apoB-containing lipoproteins (OxPL-apoB) by 86%.

PHARMACOLOGY OF ISIS-APO(a) Rx IN HUMAN LPA TRANSGENIC MICE
As the apo(a) transcript is not expressed in rodents, in vivo preclinical effi cacy assessments were limited to studies in transgenic mice expressing a 12 kringle KIV apo(a) isoform which expressed the entire human LPA genomic sequence, without h-apoB ( 43 ). Administration of ISIS-APO(a) Rx to 12K-apo(a) mice produced dose-dependent reductions in apo(a) liver mRNA and apo(a) in plasma after 2 weeks of ASO administration at 1.5, 5, 15, and 50 mg/kg/week ( Fig. 3 ). The 50% effective dose values for ISIS-APO(a) Rx apo(a) mRNA and plasma apo(a) reductions were 9.7 and 12.4 mg/kg/week, respectively, in this transgenic model.

PHARMACOLOGY OF ISIS-APO(a) Rx IN LEAN CYNOMOLGUS MONKEYS
While the ISIS-APO(a) Rx binding site in the rhesus monkey contains a single mismatch relative to the nonhuman primate sequence, the potential pharmacodynamic effects of this compound were evaluated in chow-fed cynomolgus monkeys when administered up to 40 mg/kg/week for 12 weeks. Cynomolgus monkeys, in a similar fashion to humans, have a wide range of plasma Lp(a) levels due to variability in the KIV 2 repeats similar to humans, but lack KV of human apo(a ) ( 44 ). As described previously in both humans and nonhuman primates, our data show that cynomolgus monkeys had highly variable hepatic mRNA expression levels ( 44,45 ). Results from this events, lending evidence that the content of pro-infl ammatory OxPLs are key determinants of atherogenic risk mediated by Lp(a) (37)(38)(39)(40)(41)(42).

IDENTIFICATION OF A SECOND GENERATION ANTISENSE DRUG TO HUMAN apo(a )
In light of the promising activity observed in the transgenic mouse studies described above and the desire to identify an optimized human clinical candidate, a high density screen of over 2,200 second generation ASOs designed to complementary sites spanning the entire human apo(a) transcript were tested for their ability to dose-dependently reduce apo(a) mRNA expression in transgenic LPA mouse primary hepatocytes. An optimal ASO, now designated as ISIS -APO(a) Rx , to differentiate it from the prior version ISIS 144367, was identifi ed that binds to the exon 24-25 splice site of the mature human apo(a) transcript (GenBank accession NM_005577.2) at position 3901-3920 bp ( Table 1 ). KIV 2 repeats are identical at the protein level, but are not conserved uniformly at the nucleotide level, which is why the drug only binds to a single splice junction with perfect complementarity. ISIS -APO(a) Rx also has the potential to bind to 11 alternative sites within the transcript containing one to four mismatched nucleotides, relative to the active site. The concentration of ISIS -APO(a) Rx that produced an IC 50 of the apo(a) mRNA in LPA transgenic mouse primary hepatocytes was 0.12 M. In cynomolgus monkey primary hepatocytes, the observed IC 50 was 0.49 M. In contrast, the IC 50 values for control ASOs not targeted to the apo(a) mRNA were >10 M in both primary cell isolates (data not shown).  ( 30 ) and ( 12 ). cohort, after 12 weeks of ISIS-APO(a) Rx administration ( Table 2 ).
In another 13 week study, the effects of ISIS-APO(a) Rx inhibition as a function of dose were evaluated in chow-fed cynomolgus monkeys. At the 4, 8, 12, and 40 mg/kg/week doses, hepatic apo(a) mRNA was reduced to 84 ± 12%, 70 ± 18%, 79 ± 17%, and 97 ± 3%, respectively, of mean saline control expression levels by day 93 of ISIS-APO(a) Rx study demonstrated that ISIS-APO(a) Rx signifi cantly reduced hepatic apo(a) mRNA by 90%, relative to the salineadministered cohort ( Table 2 ). Furthermore, due to some conservation of apo(a) and plasminogen nucleotide sequences (there are three base mismatches within the near homologous binding site), plasminogen mRNA levels were also measured . There was no signifi cant change in hepatic plasminogen mRNA detected, relative to the PBS   Interestingly, when plasma lipids were measured after 13 weeks, there were no signifi cant changes observed in total cholesterol, HDL cholesterol (HDL-C), LDL-C, or apoB levels, even at the highest administered dose ( Table 2 ).
In order to evaluate the heterogeneity of apo(a) allelic expression patterns within the PBS and ISIS-APO(a) Rx cynomolgus monkey treatment groups, Western blotting was performed to directly compare day 1 (predose) and day 93 samples in both saline and 12 mg/kg/week ISIS-APO(a) Rx cohorts ( Fig. 5A ). The apo(a) band intensities observed were consistent with derived plasma apo(a) levels ( Fig. 5B ). Furthermore, as described previously, monkeys in this study were both heterozygous and homozygous for different apo(a) isoform sizes. Importantly, in all four monkeys treated with ISIS-APO(a) Rx , plasma apo(a) levels were reduced to nearly undetectable levels at the 12 mg/kg/week dose by Western blot, irrespective of apo(a) protein isoform size; while in the saline cohort, no changes in expression levels were observed in any of the plasma samples. These results demonstrate that ISIS-APO(a) Rx treatment is highly effective in lowering plasma apo(a) levels in nonhuman primates regardless of individual variation in isoforms or circulating concentrations.

ISIS-APO(a) Rx PHASE 1 TRIAL
A double-blinded placebo-controlled dose-escalation phase I trial in healthy volunteers with Lp(a) concentration of >25 nmol/l (>10 mg/dl) was initiated to assess the effi cacy, safety, and pharmacokinetics of ISIS-APO(a) Rx ( 19 ). A total of 16 subjects were enrolled into the APO(a) Rx single-dose cohorts and 31 subjects into the multiple-dose cohorts. Participants were randomly assigned to receive ISIS-APO(a) Rx by subcutaneous injection (50, 100, 200, or 400 mg) or placebo (3:1) in the single-dose part of the study or to receive six subcutaneous injections of ISIS-APO(a) Rx (100, 200, or 300 mg, for a total dose exposure of 600, 1,200, or 1,800 mg) or placebo (4:1) during a 4 week period in the multi-dose part of the study.
In the multi-dose cohort, ISIS-APO(a) Rx (100-300 mg) resulted in dose-dependent mean percentage decreases in plasma Lp(a) concentration of 39.6% from baseline in the 100 mg group ( P = 0.005), 59.0% in the 200 mg group ( P = 0.001), and 77.8% in the 300 mg group ( P = 0.001). The largest decrease in an individual patient was 88.8% at day 36 after multiple doses of 300 mg ISIS-APO(a) Rx . Maximum plasma concentrations of ISIS-APO(a) Rx were dosedependent over the studied dose range, and maximum administration ( Fig. 4A ). As anticipated, plasma Lp(a) levels were concomitantly reduced by 23 ± 13%, 40 ± 22%, 70 ± 16%, and 90 ± 5% at the 4, 8, 12, and 40 mg/kg/week doses, respectively, relative to day 1 baseline levels ( Fig. 4B ).   ISIS-APO(a) Rx was administered to lean cynomolgus monkeys at 40 mg/kg/week over 13 weeks (n = 4 per group). A loading regimen of three doses in fi rst week followed by once per week subcutaneous administration was utilized.
a Data are expressed as the mean percentage of values observed in saline (±SEM) treated monkeys for apo(a) and plasminogen mRNA levels. b Plasma total cholesterol (TC), HDL-C, LDL-C, and apoB protein are expressed as a percentage of baseline levels. c Indicates statistically signifi cant from saline using Mann Whitney two-tailed test ( P < 0.05). lipoproteins ( Fig. 7 ). No signifi cant changes were noted in OxPLs on plasminogen or plasminogen levels.
In the combined multi-dose cohorts, an inverse correlation was noted between the size of the predominantly expressed apo(a) isoform and baseline plasma Lp(a) and OxPL-apoB concentrations. However, there was no significant correlation between the major apo(a) isoform and the mean percent change from baseline to day 36 in Lp(a) plasma concentration was followed by an initial relatively rapid distribution phase. Post-distribution plasma concentrations in the 300 mg multi-dose cohort reached steady state just before day 36, which coincided with the nadir of Lp(a) and OxPL-apoB and OxPL-apo(a) response ( Fig. 6 ). Similar reductions were observed in the amount of OxPLs associated with apoB-100 (up to 90.2%) and apo(a) (up to 93.1%), but no signifi cant changes were noted in other

ALTERNATIVE THERAPIES TO LOWER Lp(a)
Recent data has demonstrated that Lp(a) can be signifi cantly lowered by 20-40% with ASOs to apoB ( 35 ), monoclonal antibodies to proprotein convertase subtilisin/kexin type 9 (46)(47)(48), and cholesterol ester transfer protein inhibitors ( 49,50 ). However, in patients at or above the 80th percentile, corresponding to ‫ف‬ 50 mg/dl plasma Lp(a) concentrations, much greater reduction than is currently achieved with these indirect therapeutic agents would be required to signifi cantly reduce CVD risk, which is thought to occur at levels which exceed [25][26][27][28][29][30] or OxPL-apoB concentrations, consistent with the independence of lowering of Lp(a) and OxPL-apoB on isoform size ( Fig. 8 ). In the combined single-dose and multi-dose cohorts, at all time points, a strong correlation was noted between Lp(a) concentrations and OxPL-apoB ( r = 0.86, P < 0.0001) and Lp(a) and OxPL-apo(a) ( r = 0.91, P < 0.0001).
In summary, ISIS-APO(a) Rx resulted in potent dose-dependent selective reductions of plasma Lp(a) and represents a potential therapeutic drug to reduce the risk of CVD and calcifi c aortic valve stenosis in patients with elevated Lp(a) concentration.  the need for aortic valve replacement. Elevated Lp(a) levels and the LPA SNP, s10455872, which is associated with elevated Lp(a) levels, have recently been identifi ed in epidemiologic and genome-wide association studies as predictors of aortic valve stenosis, aortic valve replacement, and aortic valve calcifi cation ( 6,8 ). Our group has recently evaluated the role of Lp(a) and OxPL-apoB, which refl ects the biological activity of Lp(a), in predicting the rate of progression of preexisting aortic stenosis in the ASTRONOMER (Aortic Stenosis Progression Observation: Measuring Effects of Rosuvastatin) trial. Elevated levels of both Lp(a) and OxPL-apoB predicted aortic stenosis progression [measured by the annualized increase in peak aortic jet velocity in meters per second per year (m/s/year) by Doppler echocardiography], as well as the need for aortic valve replacement and cardiac death during 3.5 ± 1.2 years of follow-up ( 9 ). The rate of progression was faster in patients in the top tertiles of Lp(a) (peak aortic jet velocity 0.26 ± 0.26 m/s/year vs. 0.17 ± 0.21 m/s/year; P < 0.005) and OxPL-apoB (0.26 ± 0.26 m/s/year vs. +0.17 ± 0.21 m/s/year; P < 0.01) ( Fig. 9 ). These fi ndings support the hypothesis that Lp(a) mediates aortic stenosis progression through its associated OxPLs and provide a rationale for randomized trials of Lp(a)-and OxPL-apoB-lowering therapies in aortic stenosis ( Fig. 10 ). A clinical trial can be performed to assess whether lowering Lp(a) may reduce progression of aortic stenosis and the need for aortic valve replacement.
Finally, the ongoing development of tri-antennary N -acetyl galactosamine conjugates is expected to further enhance mg/dl ( 8 ). Therefore, potent and specifi c inhibitors of Lp(a) represent an unmet medical need for high risk patients.

FUTURE DIRECTIONS
Future studies of ISIS-APO(a) Rx will include gaining more experience on the safety and effi cacy in various populations where it may be used clinically. For example, potential indications may include patients with elevated Lp(a) levels and otherwise controlled risk factors, such as patients with refractory angina ( 51,52 ), recurrent cardiovascular events [including patients undergoing apheresis for elevated Lp(a)] ( 53-56 ), young patients (i.e., <50-60 years old) with CVD ( 57 ), calcifi c aortic valve stenosis ( 9 ), patients with familial hypercholesterolemia of whom 40-50% have Lp(a) levels >50 mg/dl ( 35 ), stroke (particularly in the pediatric age group) ( 58 ), chronic renal disease, and secondary and primary prevention. With the current potent ASO-lowering Lp(a) levels of 80-90%, it may be possible to lower Lp(a) levels in most patients to what is considered least atherogenic, i.e., <25-30 mg/dl, and to test the hypothesis that lowering Lp(a) levels will lead to reduction in CVD events.
An attractive population in which to reduce Lp(a) levels is patients with preexisting aortic valve stenosis. The prevalence of aortic valve stenosis is increasing rapidly due to the aging of the population and there is a clinical need to reduce the progression of aortic stenosis and, ultimately, Fig. 9. Calcifi c aortic valve stenosis progression rate according to plasma levels of Lp(a) and OxPL-apoB. Annualized progression rate of peak velocity across the aortic valve (V peak ) is compared by tertiles in the whole cohort for Lp(a) (A) and for OxPL-apoB (B) and after dichotomization by median age (C, D). * P < 0.05 tertile 3 (>58.5 mg/dl) compared with tertiles 1 and 2 ( р 58.5 mg/dl) of the Lp(a) age р 57 group; † P < 0.05 tertile 3 (>5.5 nM) compared with tertiles 1 and 2 ( р 5.5 nM) of the OxPL-apoB age р 57 group. Error bars = SEM. This fi gure was adapted from ( 9 ) with permission.