July 22nd, 2010

HDL and Residual Risk: A Surprising Finding in JUPITER

CardioExchange welcomes Dr. Paul Ridker to answer questions about his recent paper in the Lancet, which analyzed data from JUPITER and found that HDL may not predict residual risk in patients on high-dose statins who reach very low LDL levels such as those achieved in the treatment group in JUPITER. The findings may surprise some readers, though Ridker et al note in their report that similar findings have been observed in previous trials such as PROVE-IT. HDL was found to be inversely correlated with vascular risk in patients who were randomized to placebo in JUPITER, but there no significant relationships between HDL levels and vascular risk in the rosuvastatin-treated patients.

Is it known what effect HDL has on risk in an untreated population with LDL levels that are already at the post-treatment level in JUPITER? In other words, does the HDL effect disappear because the LDL is below a certain point or because the statin treatment brings it to that point?

There are very little prior data on what effect, if any, HDLC has on risk in populations with an average LDLC of 50 mg/dL. However, we do know from prior population and genetic studies that those lucky enough to have native LDLC levels in this range tend to have very low vascular event rates. In our analysis, we got the LDLC levels down to that range using a potent statin (rosuvastatin 20 mg). When we examine prior studies of high-dose statin therapy (atorvastatin 80 mg in both PROVE-IT and TNT) the same finding was observed, namely that residual risk on the basis of HDLC was marginal once a high-dose statin was used.

Was this same attenuation of residual risk seen across all risk strata, i.e., in the highest-risk as well as the lowest-risk participants?

JUPITER is a primary-prevention trial where the gradient of risk is relatively narrow. However, the similar findings of only marginal relation between on-treatment HDL and subsequent vascular risk in the atorvastatin 80 mg arms of the PROVE-IT and TNT trials noted above suggest that this is also true among those with recent ACS and chronic coronary disease, two very high-risk groups.

How do you think these results should inform and/or impact ongoing efforts to develop HDL-lowering therapies? What are the implications for the design of trials designed to evaluate the efficacy of HDL-raising drugs? Would you recommend that those trials only enroll patients with low HDL and no statin therapy or only those with low HDL and low LDL?

The only way to truly evaluate the role of HDL-raising agents is through carefully designed randomized trials, and I believe it very important that those studies move forward and be completed. That being said, those trials should ensure that all participants are on potent statin therapy (because these agents are already proven and will be generic very soon). More important to me than the baseline HDLC is the question of how large an HDLC increase one anticipates and whether this can be achieved safely. Agents that substantially increase HDLC would seem to hold greater promise. As with any new class of agents, the net benefit-to-risk ratio must be ascertained, so we will also need long-term follow-up in all anticipated trials.

2 Responses to “HDL and Residual Risk: A Surprising Finding in JUPITER”

  1. Lack of proof of relationship does not equal proof of lack of relationship.

    The fact that the incidence of events was too low or the study population too small to reach conclusions does not prove or even suggest the opposite conclusion. To report it as such is intellectual dishonesty or at least statistical ignorance.

    This mistake is often made and reported as fact to the determent of good medicine.

    Competing interests pertaining specifically to this post, comment, or both:
    I like to prevent heart attacks. This requires raising the HDL in a majority of cases.

  2. I agree with Blanchet and disagree with Ridker.
    Ridker is quoted above to say “When we examine prior studies of high-dose statin therapy (atorvastatin 80 mg in both PROVE-IT and TNT) the same finding was observed, namely that residual risk on the basis of HDLC was marginal once a high-dose statin was used”.
    Somehow, he seems to have the TNT analysis backwards. Subjects with LDL-C less than 70 mg/dL, mainly on top-dose atorva 80 mg, had nearly double the CHD risk with HDL-C below 37 vs above 42 in TNT (P Barter, et al. New Engl J Med. 2007;357:1301-1310). (Paradoxically, subjects responding less well to atorva 80 may have been less sensitive to low HDL-C)
    Further, looking at other statin studies, marginal relationships between on-treatment HDL-C and CVD events might be expected when the treatment (statin) has marginal effects on HDL-C.
    Yes, carefully designed clinical trials of HDL-raising are needed, and four are underway (two with niacin and two with CETP-inhibitors).
    Meanwhile, statin-based treatment (probably primarily via LDL-C lowering) remains the best initial therapy due to many placebo controlled and stronger vs weaker statin trials.
    Too bad we have almost NO data whatsoever comparing statins with other drugs (or for that matter, treatment of patients with high vs normal baseline CRP levels).
    In retrospect, it is a shame that JUPITER could not have been a two-by-two trial comparing niacin with rosuvastatin monotherapies, both together, or neither. A 2-3 year follow-up with the same number of subjects (perhaps switching rosuva placebo to active drug halfway through) would probably have answered Dr. Ridker’s question: what are the benefits of HDL-raising in the setting of aggressive LDL-C lowering? So for now, we can only speculate.

    Competing interests pertaining specifically to this post, comment, or both:
    Career-long focus on HDL. Past or present speaker and consultant for all statin manufacturers, and for the maker of the one prescription extended-release niacin product available in the US.