Are the PCSK9 inhibitors the panacea of atherosclerosis treatment?

KEYWORDS:

• Cardiovascular disease
• Low density lipoprotein cholesterol
• PCSK9 inhibitors

1. Introduction

Since the initial publication in 2012 of the first human studies showing dramatic reductions in low density lipoprotein cholesterol (LDL-C) by inhibiting PCSK9 with a monoclonal antibody, speculation has focused on the ability of these agents to reduce atherosclerotic cardiovascular disease (CVD) [1,2]. Skeptics, and some guideline committees, suggested further reductions in LDL-C below 70 mg/dL or >50%, achievable by high intensity statin with or without ezetimibe, would have little if any additional CVD benefit as a plateau had been reached [3–5]. They also expressed concern that reductions to very low LDL-C levels, such as ≤25 mg/dL, could be harmful, increasing the risk of hemorrhagic stroke, cognitive impairment, cataracts and diabetes [6–9]. Similar concerns regarding greater LDL-C reductions and low LDL-C plagued the use of later generation, more efficacious, statins, or statins plus ezetimibe, and took decades, until results from PROVE-IT and IMPROVE-IT trials, to resolve [10,11]. However PCSK9 inhibitors, in a mere 5 years, have provided conclusive evidence of CVD benefit from the results of the Further cardiovascular OUtcomes Research with PCSK9 Inhibition in subjects with Elevated Risk (FOURIER) trial with evolocumab[12]. We now know there is continued CVD benefit down to very low LDL-C with none of the predicted untoward adverse effects[12].

The issues surrounding PCSK9 inhibitors, LDL-C reduction, CVD benefit and safety can be best discussed as outlined in the Table 1 below. The current PCSK9 inhibitors are expensive and their cost effectiveness and patient access subject to ongoing debate, however a detailed discussion of these issues is beyond the scope of this review.

Table 1. Summary of Issues Related to PCSK9 inhibitors.

PCSK9 monoclonal antibodies (mAbs) and LDL-C reduction; how much and in whom?

Do PCSK9 monoclonal antibodies reduce CVD risk?

Is there CVD benefit to achieving very low LDL-C of <25 mg/dL?

Are PCSK9 fully human monoclonal antibodies safe?

Are there safety concerns to achieving very low LDL-C < 25 mg/dL?

1.1. PCSK9 monoclonal antibodies and LDL-C reduction

Unlike statins where LDL-C reduction is limited by toxicity, PCSK9 mAbs are limited only by efficacy; once all PCSK9 in plasma is bound to a mAb a maximal ~60% reduction in LDL-C is achieved [1,2]. Increasing doses results in no further LDL-C reduction but provides longer stable LDL-C reductions, without signs of toxicity or adverse events [1,2]. This is important in determining dosing intervals, especially for a drug that is administered by subcutaneous (SC) injection. A rough rule of thumb for alirocumab and evolocumab is that 70–75 mg will maximally bind available PCSK9 and reduce LDL-C 60% for 1 week, double the dose (140–150 mg) for 2 weeks and 3× the 2 week dose (420–450 mg) for 4 weeks [1,2,13]. A stable 60% reduction in LDL-C is seen with appropriate dosing and dosing intervals when alirocumab and evolocumab are added to diet alone, low and maximal dose statin or statin plus ezetimibe [13]. Patients with heterozygous familial hypercholesterolemia (HeFH) and nonFH respond the same and response in HeFH is independent of the underlying LDL receptor mutation [14]. Homozygous FH patients respond half as well, ~30%, as all other patients to evolocumab 420 mg every 4 weeks although the response is related to the underlying LDL receptor defects [15]. A number of studies in patients unable to tolerate statins, or effective doses, show they respond well to, and tolerate, PCSK9 mAbs [16]. In contrast to statins and ezetimibe, PCSK9 inhibition decreases Lp(a) by 25–30% [13].

1.2. Do PCSK9 monoclonal antibodies reduce CVD risk?

For statins, despite robust reductions in LDL-C, there was little if any evidence of CVD risk reduction until the results of the Scandinavian Simvastatin Survival Study (4S), 7 years after regulatory approval and marketing of lovastatin [17]. In contrast, exploratory and post hoc, analyses from phase 3 trials with both evolocumab and alirocumab showed reductions in CVD end points of ~50%, raising expectations for the entire class [18,19]. The CVD benefit was validated by FOURIER, the largest and shortest positive CVD outcome trial for a LDL-C lowering agent [12]. While some in the media and financial world expressed ‘disappointment’ that the predefined CVD end points were ‘only’ reduced 15–20%, this likely reflects both a lack of understanding of trial design and stopping rules along with the raised expectations from the two small exploratory or post hoc studies [20]. As outlined in the FOURIER design publication the trial was planned to continue until >1630 patients experienced the key secondary end point of CVD death, MI and stroke, providing 90% power to detect a relative reduction of at least 15% [21]. Factoring in a lag period of ~1 year, as observed in prior LDL-C lowering trials, a noncompliance rate of 10% per year, and assuming a 2% per year event rate in the placebo arm, it was determined that 27,500 patients followed for a median of ~43 months would be required [21]. However, the ‘hard’ secondary event accumulation rate was nearly double projections and thus the trial was terminated with 1829 such end points after a median of 26 months [12]. While there was no difference in CVD death, the two serious, debilitating and costly CVD events of MI and stroke were reduced by 27% and 21% (p < .001​​ and <.01) respectively [12]. As highlighted by the FOURIER investigators the reduction in CVD events, seen as early as 1 year, increased progressively with duration of treatment in the trial and was similar to that with statin therapy when assessed at 3 years [12].

1.3. Is there CVD benefit to achieving very low LDL-C of <25 mg/dL?

Prior to PCSK9 inhibitors, data from IMPROVE-IT showed patients who achieved an LDL-C <30 mg/dL had a significantly (P < 0.001) lower rate of CVD events when compared to the group with LDL-C >70 mg/dL and that secondary composite end points were significantly reduced for each group achieving a LDL-C <70mg/dL, being numerically lowest for the group <30 mg/dL[11]. In a post hoc analysis from the GLAGOV (Global Assessment of Plaque Regression With a PCSK9 Antibody as Measured by Intravascular Ultrasound) trial, evolocumab-treated patients with a mean LDL-C of 24 mg/dL had significantly more reduction in percent atheroma volume (PAV) compared with the placebo group, mean LDL-C 70.6 mg/dL (−1.97% vs. −0.35%; P < .001). A LOESS (Locally Weighted Polynomial Regression) plot showed a linear relationship between on-trial achieved LDL-C level and PAV progression down to an LDL-C of 20 mg/dL[22]. Supportive evidence from a post hoc analysis of CVD events in 10 phase 3 trials with alirocumab by Ray et al. showed there was no plateau to CVD which continued down to a mean LDL-C of <25 mg/dL (i.e. about half the patients had LDL-C <25 mg/dL) [23]. They showed that LDL-C reduction from 50 mg/dL to 25 mg/dL resulted in the same CVD reduction as reducing LDL-C from 75 mg/dL to 50 mg/dL and the well-established relationship from statin trials of every 40 mg/dL (1 mmol/L) decrease in LDL-C reducing CVD events 24% held firm down to the lowest LDL-C achievable in these trials [23]. The final and most unequivocal evidence was provided in FOURIER, where the median on treatment LDL-C in the evolocumab group was 30 mg/dL, 42% of subjects had LDL-C <25 mg/dL and 25% <20 mg/dL. Based on the lowest achieved LDL-C quartiles, between ~20 and ~25 mg/dL, CVD events were lower than those at the highest two quartiles of ~35 to ~45 mg/dL, and led the FOURIER investigators to conclude ‘lower LDL-C is better’ [12].

1.4. Are PCSK9 fully human monoclonal antibodies safe?

During large phase 2 and 3 clinical trials in >10,000 patients treated for 3+ years no specific or serious clinical or laboratory adverse events were identified with the two fully human PCSK9 mAbs, alirocumab or evolocumab[13]. Although administered subcutaneously, few, and mild, injection site reactions or antidrug antibodies have been reported and they have been well tolerated [13]. No psychological, endocrine or reproductive abnormalities or ‘off target’ effects have been observed in these large trials. The good safety has been recognized by regulatory authorities, such that there is no safety monitoring required for this new class of lipid lowering agents [24,25]. Whether the same safety, minimal injection site reactions and antidrug antibodies will hold true for future PCSK9 inhibitors remains to be seen given the problems encountered with bococizumab, a humanized mAb, which lead to its discontinuation during phase 3 development [26]. However, as with statins, it may take decades of therapy in many hundreds of thousands of patients to detect more subtle side effects not readily apparent in relatively short term trials. Recent Mendelian randomization studies of subjects with loss-of-function variants in PCSK9 suggest there is roughly the same effect as loss-of-function variants in HMG-CoA reductase on the risk of diabetes per unit decrease in LDL-C [27,28]. In the analysis by Ference et al. these variants were independent and additive but the increased risk of diabetes was found only in people with impaired fasting glucose levels for both variants [27]. Furthermore, the risk of diabetes, as confirmed in statin trials, was substantially lower than the protective effect against cardiovascular events, which account for the majority of mortality and morbidity in diabetics. In both the FOURIER and SPIRE outcome trials there was no significant difference between the rates of adjudicated cases of new-onset diabetes between evolocumab or bococizumab and placebo groups [12,26].

1.5. Are there safety concerns to achieving very low LDL-C <25 mg/dL?

Concerns with both the amount of LDL-C reduction and the levels achieved go back more than 40 years and have been recently reviewed in detail [11,13,29]. Concerns ranging from cancer to suicide, often based on faulty interpretation of epidemiological data or post hoc analysis of clinical trials, have been dispelled by additional evidence and decades of use with more efficacious LDL-C lowering agents [29]. However, at every step of more effective LDL-C reduction additional concerns have been raised, most recently centered on hemorrhagic stroke, cognitive impairment, cataracts and diabetes [8,9]. As discussed above, Mendelian genetic variant studies demonstrate a small risk of diabetes, related to per unit LDL-C reduction, in a subgroup of people with impaired glucose tolerance with both HMG-CoA reductase and PCSK9 variants [27,28]. Post hoc analysis from trials such as JUPITER and the alirocumab pooled phase 2 and 3 program have also attempted to relate increased risk of diabetes and cataracts respectively to subgroups achieving very low LDL-C [8,9]. However, both analyses contained substantial methodological flaws. There were significant biases in the baseline characteristics of those achieving very low LDL-C compared to the control or comparator group [8,9]. In JUPITER the low LDL-C group had significantly different (​​p .009 to <.0001) BMI, impaired glucose tolerance, low HDL-C, triglycerides and metabolic syndrome, clearly predisposing them prior to LDL-C reduction toward developing diabetes [8]. Similarly the analysis by Robinson et al. which reported an increase in cataracts with alirocumab in the groups achieving LDL-C <25 mg/dL or <15 mg/dL, was biased at baseline as these groups were older, had 30–50% more diabetics and higher hemoglobin A1c levels than the LDL-C >25 mg/dL or placebo groups[9]. Both analyses were further flawed in that they based the LDL-C cut points on LDL-C calculated by Friedewald formula which when compared to ultracentrifugation, the ‘gold standard’ method, is on average nearly 30% lower and misclassifies at least 36% of patients as having LDL-C <25 mg/dL when they do not [30]. The implication that very low LDL-C is related to increased risk of cataracts as suggested by Robinson et al. has been rapidly repudiated by far more robust data from the IMPROVE-IT, FOURIER and SPIRE (Studies of PCSK9 Inhibition and the Reduction of Vascular Events) trials [11,12,26].  Any relationship between hemorrhagic stroke and very low LDL-C levels has also not been supported by the safety analysis from IMPROVE-IT, FOURIER or SPIRE trials [12,26]. In regard to neurocognitive adverse events the recent detailed safety analysis of the 7 year IMPROVE-IT trial found that the group with LDL-C <30 mg/dL had no signal of increased memory impairment or other neurocognitive disorders. Cognitive impairment was assessed in a prospective sub-study EBBINGHAUS in FOURIER, which found no difference compared to placebo [12].

Both the drug sponsors of alirocumab and evolocumab and the US FDA focused specifically on patients achieving very low LDL-C levels (<25 or <15 mg/dL) and found no differences in adverse events compared to those with higher LDL-C, or those treated with placebo or standard of care [31]. Based on this careful evaluation, the regulators did not require safety monitoring for low LDL-C or include any adverse event warning in their labeling [24,25]. The regulators decision is now further supported by very large longer term trials like FOURIER and SPIRE [12,26].

In conclusion, it is likely that PCSK9 inhibitors added to maximally tolerated statin will eventually achieve a role in the prevention of atherosclerosis and CVD at least equal to that of statin alone, and help minimize, if not eliminate, the component related to LDL. However, it is unlikely that any single drug or intervention will be a panacea for this common, serious and multifactorial disease.

Declaration of interest

TA Turner’s institution has received research support from Amgen and Sanofi for clinical trials and central laboratory analysis. EA Stein has received consulting fees from Amgen, Regeneron, Sanofi, Genentech, Roche, The Medicines Co, ISIS, Catabasis, AstraZeneca, CymaBay, CVS/Caremark and BMS related to PCSK9 inhibitors and other lipid lowering drugs. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Additional information

Funding

This paper was not funded.