May 5th, 2014

Two Experts Look at a Failed HDL Trial


Despite robust epidemiological evidence suggesting that HDL has a strong protective effect against cardiovascular disease, there has been no good evidence showing that HDL-based therapies are beneficial. Now, the CHI-SQUARE (Can HDL Infusions Significantly QUicken Atherosclerosis Regression) study, published online in the European Heart Journal and the largest to ever study an HDL mimetic, has failed to find even a glimmer of benefit.

Within two weeks of having an acute coronary syndrome, 507 patients were randomized to receive 6 weekly infusions of either placebo or 1 of 3 doses of CER-001, an HDL mimetic from Cerenis Therapeutics. CER-001 had no significant effect on atherosclerosis, as assessed by both intravascular ultrasonography (IVUS) and quantitative coronary angiography (QCA). There were also no significant differences in the number of patients who had at least one major cardiovascular event. 

CardioExchange asked William Boden and Prediman (PK) Shah to comment on CHI-SQUARE. — Larry Husten

William Boden, who is PI of the NIH’s AIM-HIGH trial, was asked whether it was time to write the obituary for HDL:

Boden: You have to be kidding me! The end of what? HDL RIP? How about: RIP suboptimal study design and trial hypotheses? I continue to be amazed that we see nothing but pejorative commentary and noise about the death knell for the HDL hypothesis and HDL-raising therapy when, time after time and trial after trial, we see the same unenlightened study design perpetuated.

What do ILLUMINATE, dal-OUTCOMES, HPS-2 THRIVE, and this CHI-SQUARE trial all have in common? The answer is: an unreasonable study population in which to test HDL-raising therapy. The two CETP inhibitor trials and this one included ACS patients who were not pre-selected for a profile of low HDL-C cholesterol. In fact, I did not see any baseline lipid value for CHI-SQUARE. The baseline Apo-B values were <80 mg/dL in two groups and were 81 and 86 in the other two group0s — values that would be considered “optimal” or ideal. The Apo-A1 values of >130 mg/dL are likewise normal. Hence, we can presume that the baseline LDL-C and HDL-C were normal, or perhaps optimal. Why on earth would one expect a patient with an HDL-C of, say, 50 mg/dL to demonstrate a reduction in coronary atherosclerosis or clinical events when you are making a normal baseline value super-normal with an HDL-raising intervention? Since the epidemiology of HDL-C tells us that the risk of incident CV events is both inverse and curvilinear, if the starting HDL-C is on the flat (normal) part of the event relationship, then why would one expect that raising the HDL-C to 70 or 80 would reduce CV events?

Our recent data (from four separate sources of observational and post hoc RCTs) suggest that baseline HDL-C <30 mg/dL may be the threshold below which one needs to target HDL-raising therapy. This is where the event curve steepens inversely and where one might expect to see an HDL-raising therapeutic benefit.

So this trial tells me nothing new that I haven’t seen in the other trials I’ve mentioned. In our post hoc analysis of AIM-HIGH (admittedly only “hypothesis-generating”) an HDL-C <31 mg/dL was associated with a niacin treatment effect for the primary endpoint. This would actually be the fifth data set to show that it’s the very low HDL-C subset that we need to target, not these “all-comers” designs where patients have normal or high HDL-C to start.

We have yet to see the right trial design. And, of course, since the great majority of AIM-HIGH and HPS-2 patients were receiving statins for 1-5 years, how can you expect, as in HPS-2, to see an incremental HDL-C raising effect when the baseline LDL-C was 63 mg/dL and the baseline HDL-C was ~47 mg/dL? Maybe we need trials of patients who are statin naïve, not such well-treated patients where risk mitigation perhaps cannot be achieved.

PK Shah is the Director of the Atherosclerosis Prevention and Treatment Center and of the Oppenheimer Atherosclerosis Research Center at Cedars-Sinai, he holds the Shapell and Webb Family Chair in Clinical Cardiology, and is a Professor of Medicine and Cardiology.

Shah: This is a very disappointing study showing no short-term effects of rHDL (containing wild type Apo A-I linked to two phospholipid carriers) infusion at doses of 3, 6, 12 mg /kg on non-culprit coronary lesion size as assessed by IVUS and QCA.

If you are a pessimist and disregard all the biological plausibility data on the vascular protective effects of Apo A-I or its mutants, such as Apo A-I Milano shown in preclinical models and small clinical studies, you could conclude that Apo A-I infusions therapy may not realize its promise; on the other hand, if you are an optimist, you could make the argument that a negative study could have been due to:

1. Not measuring plaque composition, which is more likely to change before plaque size changes; i.e., not measuring lipid core size and the inflammation that goes with it. IVUS may not be the best methodology.

2. Not choosing the type of patient most likely to show a change in plaque; i.e., a patient with a lipid-rich plaque rather than any plaque without regard to its composition.

3. Not using a high enough dose (the Apo Milano study in 2003 used 15 and 45 mg/kg/per dose while as doses used in this study were 3, 6, 12 mg/kg/dose).

4. Not using enough infusions to remodel the plaque.

5. The compositional features of the HDL mimetic used in this study may not be optimal; HDL containing Apo A-I Milano, possibly a gain of function mutant, may produce different results as suggested by preclinical studies done in our laboratory.

As a believer in HDL’s vascular protective effects, I remain optimistic that, although HDL has been a tough nut to crack, one of these days we will get it right using the right formulation, right patient, and right dose; research in this field must continue until we get it right, the basic biology is very compelling.


Comments are closed.