Saying that PCSK9 inhibitors are a hot topic in medicine right now is a lot like saying that Donald Trump is a controversial president, or that you shouldn’t reference politics on a blog meant to have broad appeal across the political spectrum. I mean, it’s obvious, right? But as both a primary care doctor and a person who pays taxes, I’m paying careful attention to the rise of PCSK9 inhibitors. After all, adding an expensive monoclonal antibody to the mix for a ubiquitous disease like hyperlipidemia (73.5 million people in the US have elevated LDL levels) has the potential to significantly add to the burden of rising healthcare costs, even as it may allow us to make people live healthier and longer.
So today, we take up SPIRE 1 and SPIRE 2, conveniently presented in one paper for your enjoyment by Paul Ridker and colleagues. You may think that a paper with the phrase “On the basis of the SPIRE lipid-lowering data, the sponsor elected to discontinue further development of bococizumab” would be safe to scroll past. But it’s actually a fantastic example of why the current risk-based lipid-lowering therapy guidelines exist — and it lets me talk about some of my favorite concepts in preventative medicine. So I’m diving in, hold your breath and let’s go.
If You’re Only Going to Read One Paragraph
Bococizumab, a PCSK9 inhibitor with potency roughly equivalent to those currently on the market, reduced a composite outcome of major adverse cardiac events in high risk patients, but not in lower risk patients. This is a great illustration of why it’s essential to select the patients most likely to benefit from lipid-lowering drugs to ensure the benefits of therapy outweigh the potential for adverse events. It reinforces existing risk-based criteria for starting such therapy.
Patients, Inclusion Criteria, Comparators, and Outcomes
In this multi-center, multinational RCT, multiple inclusion criteria were used to recruit patients at high risk for ASCVD. A secondary prevention cohort consisted of patients of any age with MI, ischemic stroke, or coronary or peripheral artery revascularization in the past 5 years. A separate primary prevention cohort included patients with diabetes, peripheral vascular disease, or CKD, along with at least one of known CAD (on imaging), smoking, or low HDL. Patients with familial hyperlipidemia were also included in both cohorts.
With regard to lipid lowering therapy, all patients had to be on a high-dose statin for at least four weeks prior to trial initiation. In the SPIRE 2 trial (higher risk patients), there was exception made for patients with documented intolerance to two different statins, or with history of rhabdo or other reaction that precluded rechallenge.
All patients in the intervention group received 150mg bococizumab subcutaneous every 2 weeks, while the placebo group got, well, placebo. The primary outcome was a composite of nonfatal myocardial infarction, nonfatal stroke, hospitalization for unstable angina requiring urgent revascularization, or cardiovascular death.
I told you there were two trials, but so far I have described only one set of inclusion and exclusion criteria. That’s because SPIRE 1 and SPIRE 2 differed only based on the minimum LDL required for study entry. SPIRE 1 recruited patients whose LDL remained ≥ 70 mg/dL despite statin therapy, whereas in SPIRE 2 they were required to be ≥ 100 mg/dL (including in the statin-intolerant group). This is billed as a low risk versus a high risk cohort. The mean difference ends up being 94 mg/dL in SPIRE-1 and 134 mg/dL in SPIRE-2) —as we talked about yesterday, lowering LDL 40mg/dL for 5 years results in a ~23% risk reduction, which is going to translate to something on the order of 2-3% difference in absolute 10 year risk between groups.
A Bit of Context
PCSK9 inhibitors are truly the new hotness. Can I pull off saying new hotness? Whatever, I’m going with it. In phase 2 trials, PCSK9 inhibitors have been shown to decrease LDL levels by as much as 60%(!). On this basis alone, without clinical endpoint data, the FDA has approved two drugs, evolocumab and alirocumab, for treatment of familial hypercholesterolemia and patients with prior ASCVD whose LDL is not at goal on existing therapy.
Outcomes trials are coming out fast and furious: recently evolocumab, an drug analogous to bococizumab, was shown to reduce cardiovascular events in patients with known ASCVD. The ODYSSEY Outcomes trial of alirocumab is ongoing, looking exclusively at patients with prior ACS. This all goes back to our conversation yesterday — these investigators are focusing on patients at high absolute risk, looking first at the patients most likely to benefit from therapy. The fact that this population is also the most likely to have a major cardiac event (and thus requires a lower sample size needed to see a difference between groups) is just a bonus for the pharmaceutical companies funding the trial.
A quick aside to discuss the so-called statin hypothesis. It is well-established that statins can reduce events related to ASCVD when used in either primary or secondary prevention (if you don’t believe me, look at some of the trials from that same metanalysis above). However, a slew of negative trials of other meds (niacin, fenofibrate, and a cholesterol ester transfer protein inhibitor I’ve never even heard of) had folks stopping to wonder if maybe the benefits of statins wasn’t just lower cholesterol. Theories abounded related to anti-inflammatory effects, or antioxidant effects, or whatever else, effectively generating about as much explanatory usefulness as me waving my hands while saying “cytokines.”
But then finally, the IMPROVE-IT study, published just under two years ago, demonstrated benefit to ezetimibe when added to statin therapy post-MI. The statin therapy isn’t completely disproven — there may be some benefit to statins unrelated to their ability to lower LDL — but the lipid concentration itself matters too. Consider:
Jarcho et al., N Engl J Med 2015; 372:2448-2450
These new positive trials of PCSK9 inhibitors are underlining this fact. Statins aren’t the only show in town, and other ways of improving the lipid profile can still be relevant.
These mammoth trials enrolled 16,817 patients in SPIRE-1 and 10,621 in SPIRE-2, with mean LDL levels of 94 and 134 mg/dL in the two trials respectively. SPIRE-2 patients were more likely to already be on ezetimibe and to not be on a statin (this reflects the inclusion of the statin intolerance group — though 83% of SPIRE-2 patients were on statin therapy). The mean age was 63; 30% of patients were female; and 48% were diabetic. Overall, a reasonable representative population for CAD.
With regard to outcomes, things are interesting. This trial was stopped early, so mean follow up was only 7 months for SPIRE-1 and 12 months for SPIRE-2. In the SPIRE-1 (lower risk) patients, the hazard ratio between the two groups was 0.99, which you will not be shocked to hear did not reach significance (P = 0.94). However, in the SPIRE-2 patients the hazard ratio was 0.79 (P = 0.02). The higher risk patients had more risk reduction and showed a benefit; the lower risk patients did not show benefit over the same number of person-years. And to complete the picture, adverse events? Similar between all groups. So if you’re low risk, you get all the 10.4% injection site reactions from bococizumab without any measurable benefit over this time period.
Certainly the trial’s premature stoppage may have masked an effect in the lower risk cohort. Perhaps treating for 2 years, 5 years, or 10 years would indeed have revealed a significant difference in the primary outcome. Still, it’s a nifty illustration of the importance of attention to baseline risk when selecting which patients should receive a given intervention. Now you know why I made you go through all those graphs yesterday.
So the effect is limited to higher-risk patients. Not ideal, but still not a reason to pull the drug — so why didn’t bococizumab make it to market? The lipid-lowering effect was there; mean change at 14 weeks was −56.0% in the bococizumab group vs +2.9% for the placebo group. However, a weird thing happened, and it was exactly what the pharmaceutical company was afraid of. The mean change actually decreased to -41.8% at 52 weeks.
Fun fact: bococizumab is not a fully humanized monoclonal. It still has about 3% mouse sequence (think infliximab vs adalimumab). And as happens with infliximab, patients developed antibodies to the drug, and became non-responders, with 29% of patients in the lipid-lowering (non-outcomes) trial developing neutralizing antibodies. These weren’t assessed in our current trial, but one does wonder:
What happens when you break out the non-responders? Well, buried in the supplementary appendix…
Ridker et al., N Engl J Med 2017; 376:1527-1539
Those who respond, benefit. And non-responders look a lot like the people who aren’t taking it.
PCSK9 inhibitors are here, they are effective, and we are going to have to figure out how to pay for them in our high risk patients. But selecting patients who benefit most from this costly intervention will be important as we move forward. As for bococizumab, it is no more. But not for nothing, the trial can teach a thing or two. Mostly, I learned to be aware of monoclonal antibodies that aren’t fully humanized.
I’m kidding! I’m kidding. It was the risk thing, seriously.
Read the trial here.