Abstract
Response to Drs Day, Breckenridge, and Temple
I greatly appreciate the thoughtful comments from Drs Day, Breckenridge, and Temple as they provide a broader context for discussions of how investigators and regulators can work together to advance medical care.
The fundamental biologic issue addressed in the JUPITER trial was whether statin therapy as compared to placebo would reduce cardiovascular events among individuals who were at risk for heart disease due to elevated levels of the inflammatory biomarker hsCRP, but who otherwise would not qualify for lipid lowering therapy due to already low levels of LDL cholesterol. Thus, while the US FDA label indication based on JUPITER includes hsCRP (and thus reflects the trials’ core biologic basis), Dr Day is correct in that it came as a surprise to the JUPITER investigative group that the European and Canadian label indications based on the same data are silent on the hsCRP issue. This was particularly puzzling in Canada since hsCRP screening was already part of the Canadian national guidelines for cardiovascular disease prevention when the JUPITER trial was reviewed [1]. In Europe, hsCRP screening is not part of national clinical guidelines and the use of any non-LDL biomarker to define high risk patients is less well accepted by the European regulatory and clinical communities. The European hesitancy to embrace emerging biology, however, does not provide an explanation for an indication that excludes most of those randomized in the trial.
Because the JUPITER trial design fundamentally differed from all prior statin trials, the academic Steering Committee which I chaired was of the impression that the most logical label from JUPITER would be one that closely reflected the patient population studied, those identified at risk on the basis of elevated hsCRP. We further assumed that the trial primary endpoint, a composite of nonfatal MI, nonfatal stroke, coronary revascularizations, and cardiovascular death would be included in any label indication since there was no heterogeneity in the direction of effect for any of the individual components. Last, we were optimistic that total mortality, although formally a secondary endpoint, might also be considered for labeling as this crucial outcome was significantly reduced by 20% in the JUPITER trial, consistent with prior data for statins in secondary prevention.
Despite general concordance of opinion on the academic Steering Committee, I am certain Dr Day is correct in that not all JUPITER investigators in the field feel identically about the role of statins in primary prevention at a policy level. Individual a priori beliefs about the likelihood that statin therapy is an appropriate and cost-effective agent in primary prevention play a substantive role in this latter diversity of opinion. I thus concur with Dr Day that in the end, individual physicians and individual patients will inevitably come to different views of a given trial result.
What the JUPITER experience has taught the Steering Committee is exactly what Dr Breckenridge states succinctly, namely that the regulation of medicine is driven primarily by scientific considerations, but also operates within legal and cultural frameworks that differ widely. There is admittedly tension here as cultural differences are largely geographic; as trialists, we strive to do studies in diverse settings in part to address the generalizability of our science in different populations. Trials that lead to different results in different regions pose a particularly complex set of issues, but this was not the case in JUPITER.
My personal view is most concordant with that of Dr Temple, perhaps not surprising as the US FDA came closest to a label indication that reflects what the JUPITER investigation was all about. The JUPITER Steering Committee also owes considerable thanks to several thoughtful officials at the US FDA who played a critical role in objectively addressing the safety of rosuvastatin and by so doing made it possible for the trial to enroll successfully. Further, based in part on achieving a label indication that explicitly includes inflammation, we have subsequently been successful in obtaining industry as well as NIH support for two new trials that are designed to directly test the inflammatory hypothesis of atherosclerosis, one using an interleukin-1β inhibitor and one using low dose methotrexate. The fact that the US FDA was forward-looking in providing a label for JUPITER that reflected inflammation and hsCRP was crucial to obtaining the funding for these new ventures which, if successful, may provide substantive benefits for our patients.
Dr Temple carefully notes that many in the cardiovascular community make the assumption that drugs proven to be effective in lower risk populations will have similar efficacy in higher risk populations. This argument reflects common thinking in US guidelines circles where it has long been assumed that both relative and absolute risk reductions associated with statin therapy track in a simple manner with increasing absolute risk.
I am not convinced this is true and I worry that this assumption may ignore pathophysiologic signals coming from trials. As an example, despite clear evidence that statins do not reduce coronary artery calcification (CAC), imaging advocates have suggested that because CAC scanning identifies a high risk group, such individuals must by definition benefit from statin therapy. However, the four major placebo-controlled statin trials that failed to show clinical benefit – CORONA, AURORA, GISSI-HF, and 4-D – were trials conducted in the setting of heart failure or renal dysfunction where absolute risk is high and calcified plaques very prevalent. By contrast, the three statin trials with the greatest relative risk reductions – JUPITER, AFCAPS/TexCAPS, and MEGA – were conducted in primary prevention settings where absolute risk is low, calcified plaques infrequent, and the disease process can be interrupted in much earlier stages. We should thus be hesitant before assuming in the absence of evidence that certain patients do or do not benefit from a specific therapy. The method by which one selects an ‘enriched’ population matters greatly, should reflect the biology of the drug of interest, and needs to be reflected in a given labeling indication.