Niacin Alternatives for Dyslipidemia: Fool's Gold or Gold Mine? Part I: Alternative Niacin Regimens

Abstract

Niacin was the first drug demonstrating lowered cholesterol prevents coronary heart disease (CHD) events, with two clinical CHD outcome studies establishing a cardioprotective niacin regimen: 1 g thrice daily with meals. Though cardioprotective, skin toxicity limits niacin's use, fostering several variations to improve tolerability. One of these, an extended-release (ER) alternative, proved immensely successful commercially, dominating clinical practice despite departing from the established regimen in several critical ways. Hence, improved tolerability may have come at the cost of diminished efficacy, posing a conundrum: Does it still help the population at risk for CHD to broaden a drug's acceptance by "watering it down"? This question is crucial at this stage now that the ER alternative failed to recapitulate the benefits of the established cardioprotective niacin regimen in two trials of the alternative approach: AIM-HIGH and HPS2-THRIVE. Part I of this review discusses how vastly the ER alternative departs from the established cardioprotective regimen, why that is important physiologically, and how it may explain the findings of AIM-HIGH and HPS2-THRIVE. Given important gaps left by statin therapy, the established cardioprotective niacin regimen remains an important evidence-based therapy for the statin intolerant or statin averse.

Keywords: Hyperlipidemia; Lipids; Niacin; Niacin conjugates; Niacin prodrugs; Nicotinic acid.

Fig. 1

a Meta-analysis of odds ratio for hard CHD events (CHD death and nonfatal myocardial infarction) comparing the exploratory ER alternative to the established cardioprotective regimen. The event rates for active and comparator groups are as reported in the four cardiovascular outcomes trials (CDP, SIHDS, AIM-HIGH, and HPS2-THRIVE) [, , ••, ••]. We used random-effects meta-analysis techniques to pool the corresponding log-transformed odds ratios (OR) by the metan procedure in Stata v14. Initially, we analyzed the exploratory regimen as if it were equivalent to the established cardioprotective regimen, pooling all four trials. Unsurprisingly, treating such different approaches as equivalents revealed a high degree of heterogeneity, I ², as high as 73 % (p = 0.012), indicating the overwhelming amount of variation between trials is attributable to heterogeneity. Thus, pooling the trials proved unsupportable. In marked contrast, treating the exploratory regimen as distinct from the established cardioprotective regimen rendered heterogeneity insignificant and inconsequential, as shown. Among trials of the exploratory regimen, I ² was very low (13 %, p = 0.3), indicating agreement. Specifically, these trials agreed on inefficacy of the alternative regimen (OR = 1.0, p = 1). Likewise, among trials of the established cardioprotective regimen, I ² was low (36 %, p = 0.2), again indicating agreement. Specifically, these trials agreed on efficacy of the established cardioprotective regimen (OR = 0.75, p = 0.01). The high heterogeneity when pooling all four trials but low heterogeneity when distinguishing the exploratory from the established regimen further supports the concept that the alternative is not an equivalent with respect to outcomes. This is even more apparent when considering the clinical effects on hard CHD events (i.e., no benefit from the alternative, but an odds ratio of 0.75 for the established regimen). Clinically, this analysis affirms a role for the established regimen and denies a role for the ER alternative. CI confidence interval, IRNA immediate-release niacin, PENA pentaerythrityl tetranicotinate, ERNA extended-release niacin. b Meta-regression between log-transformed odds ratio for hard CHD events and percent change in cholesterol. As with Fig. 1, we assessed the odds ratios for hard CHD events for the four cardiovascular outcome trials, now evaluating whether cardioprotection is consistent with the cholesterol hypothesis. Specifically, we regressed the odds ratio for CHD against the drop in total cholesterol for each study by the metareg procedure in Stata v 14. Meta-regression revealed an extraordinarily strong relationship, with virtually no heterogeneity (I ² = 0 %) and nearly perfect correlation (R ² = 98 %). At the low end, it is intuitive that the alternative regimen of AIM-HIGH and HPS2-THRIVE conferred no meaningful CHD benefit, as the reduction in cholesterol was modest at best (∼5 %). This accords with the cholesterol hypothesis, which predicts that minimal reductions in cholesterol translate to minimal benefits on CHD events. On the other hand, the regimens of CDP and SIHDS present progressive reductions in cholesterol that correspond to dose-responsive CHD benefits (dose referring to cholesterol reduction). This also accords with the cholesterol hypothesis, which predicts that greater reductions in cholesterol should translate to greater CHD benefits in a dose-responsive fashion. In addition to distinguishing the alternative regimen from the established regimen, the strong linear relationship between cholesterol lowering and CHD prevention by niacin supports the concept that the lipid-targeting strategy would improve on the already-substantial benefits of the established regimen

Fig. 2

a Meta-analysis of odds ratio for a composite of soft vascular events from the HPS2-THRIVE study, stratified by baseline LDL-C. Importantly, pooling groups across the baseline LDL-C led to a high degree of heterogeneity (I ² = 70.3 %, p < 0.05), thus arguing against pooling all three groups. One group (LDL-C < 58 mg/dL) had an OR > 1.0, and the other two (LDL-C ≥ 58 mg/dL) had OR < 1.0, and the latter two had almost identical OR’s (0.89 and 0.87). Affirming this, when we pooled the latter two groups, heterogeneity was minimized (I ² = 0 %, p = 0.8). Again, those with optimized LDL-C (i.e., LDL-C < 58 mg/dL) differed from those with higher LDL-C, having no discernable benefit from ERN + laropiprant compared to placebo (OR = 1.08, p = 0.2). In contrast, those with LDL-C ≥ 58 mg/dL appear to benefit from ERN + laropiprant (OR = 0.89, CI = 0.81 to 0.97, p = 0.01). Oddly, the study was severely skewed toward people with lower LDL-C, with a minority having LDL-C > 77 mg/dL. These results suggest that a study enrolling people with higher LDL-C (e.g., LDL-C > 70 or >100 mg/dL) might be the ideal way to test the incremental benefit of the ER alternative. b Meta-regression between log-transformed odds ratio for soft vascular events from the HPS2-THRIVE and percent change in LDL-C based on baseline LDL-C. The findings from (a) suggest the study’s primary aim suffered from targeting a population who does not necessarily benefit from further LDL lowering. Conversely, the apparent benefit among those with suboptimal LDL-C suggests higher degrees of LDL suppression do confer benefits, in accordance with the LDL hypothesis. To illustrate this, we conducted a meta-regression showing fewer events with more aggressive LDL-C suppression (R ² = 61 %). Though we used linear regression, there was heterogeneity (I ² = 51 %), suggesting a nonlinear model may fit better. In any case, the relationship is consistent with the LDL hypothesis. As such, this promising result might be exploited to greater effect using the established cardioprotective regimen or better yet, the lipid-targeting strategy of niacin. Both strategies achieve more aggressive LDL-C lowering. According to the LDL hypothesis, this should build upon the promising event reductions from the ER alternative in HPS2-THRIVE. The dark box on the graph represents the hypothesized effect of more robust niacin regimens. Based on HPS2-THRIVE, we predict that a future trial using aggressive LDL-C suppression would have a result somewhere within the dark box (i.e., OR < <0.87)

Fig. 3

Left panel indicates two outcome trials affirmed the HDL hypothesis with the fibrate gemfibrozil, HIT, and HHS [58, 59]. The HIT study is more comparable to AIM-HIGH, having enrolled high-risk patients with low HDL-C at the baseline. Thus, the dashed red line provides a benchmark for an HDL-C increment expected to prevent CHD events. Both the HIT and HHS demonstrated the fibrate prevented hard CHD (OR = 0.66 to 0.76) with an HDL-C increment of 6 to 10 %. Right panel indicates two studies where low-dose niacin was added to a statin-treated background demonstrating low-dose niacin achieves similar to HDL-C increments as the fibrate studies [••, 61]. The study by Wink et al. reported the HDL-C increment from low-dose niacin in two ways. They censored their dataset to exclude one subject on low-dose niacin who had an unusually robust rise in HDL-C, but also reported the full dataset. This suggests a variable response to low-dose niacin, where some subjects are hyper-responders and cause analytical problems due to their very high increments. The AIM-HIGH “control” group received even more immediate-release niacin than in the Wink study, in most cases 150 % of the Wink dose. Not surprisingly, the stronger doses used in AIM-HIGH stimulated a prodigious rise in HDL-C that is not only higher than the Wink study but also considerably higher than the benchmark HIT study. Remarkably, by pushing the dose, the AIM-HIGH investigators managed to double the increment in HDL-C: +12 % (+4.2 mg/dL) in AIM-HIGH vs. +6 % (+2 mg/dL) in HIT. If the HDL hypothesis supported by the fibrate studies also translates to niacin, one would expect the AIM-HIGH “control” group to have similar benefits as HIT or HHS, and since the HDL-C increment is so much more robust, perhaps even greater benefits. Unfortunately, this invalidates the intended control group in AIM-HIGH, because the control by necessity should represent the untreated state, specifically a group lacking an HDL-raising dose of niacin, and preferably lacking niacin altogether. IR immediate release, OR odds ratio, Hard CHD non-fatal myocardial infarction and/or cardiac death, HIT HDL intervention trial, HHS Helsinki heart study