LDL Limbo: How Low is Too Low?

Podcast Patient Case – Very Low LDL

Imagine you receive a message from a primary care provider concerned about a patient with an LDL-C = 22mg/dL. The PCP is worried the LDL-C might be too low.  The patient is taking atorvastatin 80mg daily. How do you reply? What do you recommend?

A well-established, proportional relationship exists between low-density lipoprotein cholesterol (LDL-C) concentrations and the risk of atherosclerotic cardiovascular disease (ASCVD).¹ The ACC/AHA guidelines recommend statin therapy capable of lowering LDL-C by 30-50% to reduce ASCVD risk in patients at intermediate and high-risk for ASCVD events. There has been significant debate regarding the safety of achieving very low LDL-C levels, including a potential negative impact on cognitive function. The current ACC/AHA guidelines (circa 2013) suggest decreasing the statin dose in patients with two consecutive LDL-C levels below 40 mg/dL based on expert opinion.¹ The lack of evidence has been a major challenge for clinicians and it is unclear whether medication doses should be reduced in high-risk patients who may benefit from very low LDL-C levels.

A recently published meta-analysis, which followed the PRISMA guidelines, sought to address this clinical dilemma.² Only studies with mean/median LDL-C levels less than 70 mg/dL at baseline and which reported ASCVD outcomes were included in the meta-analysis. Data from three recent clinical trials as well as patient-level data from a subgroup included in the Cholesterol Treatment Trialists Collaboration (CTTC) meta-analysis were included in this analysis.^2,3  See Table 1.  The risk ratio (RR) of major vascular events per 1 mmol/L (38.7 mg/dL) reduction in LDL-C was calculated for each trial.  See Table 2. The RRR varied in each of these trials. For example, evolocumab reduced LDL-C by 42 mg/dL for a corresponding 22% relative risk reduction (RRR), whereas anacetrapib reduced LDL-C by 11 mg/dL for only a 7% RRR.

Overall, there was 21% RRR for every 1 mmol/L decrease in LDL-C in patients with an LDL-C less than 70 mg/dL at baseline. These results are similar to the original CTTC meta-analysis which demonstrated a 22% RRR for every 1 mmol/L decrease in LDL-C.³ The authors concluded that the reduction in major vascular events per 1 mmol/L reduction in LDL-C is consistent for all patients, regardless of baseline LDL-C, even those patients with a baseline LDL-C less than 70mg/dL.

Table 1. Trial Characteristics²

^{* Achieved LDL-C in the experimental arm as reported in original trials. Reported values from meta-analysis, mmol/L (mg/dL): 1.4 (54), 0.5 (19.31), 1.4 (54) respectively}

Table 2. Effect of 1 mmol/L LDL-C Reduction on ASCVD Event Rates²

The authors assessed several safety outcomes including myalgias/myositis, elevations in aminotransferases, new-onset diabetes, hemorrhagic stroke, and cancer. The risk of developing these adverse outcomes in patients with baseline LDL-C below 70 mg/dL was not significantly different between treatment arms.² However, it should be noted that the trials included in this meta-analysis were relatively short-term (< 7 years of follow-up) so the potential harms with longer duration of use is unknown. Furthermore, this study do not examine other safety concerns, such as cognitive decline.

In 2012, the FDA updated statin safety information to inform clinicians about the potential for cognition-related side effects.⁴ This labeling change was based solely on post-marketing reports of non-serious and reversible effects, and the FDA noted that the data did not suggest that statins were associated with significant cognitive decline. However, with the availability of potent therapies which are often combined with statins, we can now achieve very low levels of LDL-C.  This has prompted renewed interest in understanding how very low LDL-C levels may impact cognition. A pre-specified analysis of IMPROVE-IT analyzed nine safety outcomes, including cognitive adverse events for patients who achieved an LDL-C < 30 mg/dL.  There was no difference after 1 month of therapy.⁵ The EBBINGHAUS trial prospectively analyzed a subgroup of patients from the FOURIER study and assessed the effects of evolocumab on cognitive function using CANTAB, a computerized cognitive assessment tool. There was no significant difference in CANTAB scores between patients who received evolocumab or placebo for a median of 19 months, even in those patients who achieved LDL-C levels < 24 mg/dL (n=661).⁶ Lastly, a meta-analysis of phase 2 and 3 alirocumab trials found no difference in the incidence of neurocognitive events in patients who received alirocumab or placebo, including patients on alirocumab who had 2 consecutive LDL-C values < 25 mg/dL (n=839) over a period of up to 104 weeks.⁷

Another concern for achieving very low LDL-C levels is the impact on steroidogenesis. Cholesterol is the building block for steroid hormones, therefore, intensive LDL-C lowering could adversely impact the production of cortisol and sex hormones. The DESCARTES study compared evolocumab to placebo in regards to steroid hormone levels in 901 patients, including 240 patients with at least one postbaseline LDL-C of < 15 mg/dL on evolocumab.⁸ For patients treated with evolocumab, cortisol levels increased significantly from baseline to week 52 regardless of achieved LDL-C level.  Adrenocorticotropic hormone (ACTH) levels and cortisol:ACTH ratios did not significantly change. For female and male patients, there were no significant correlations between change in LDL-C and estradiol or testosterone levels. The average age of subjects was 56 years and, predictably, there were marked increases in FSH and LH in women at the onset of menopause. While the DESCARTES trial results suggest that there is little to no impact of very low LDL-C concentrations on steroidogenesis, additional data is needed to draw firm conclusions.

Finally, another key issue to consider is the accuracy of the calculated LDL-C when using the Friedwald formula. When the LDL-C is <70 mg/dL, the Friedwald formula underestimates the LDL-C, especially if the TG levels are >150 mg/dL.  This is problematic because it may lead to undertreatment or inappropriate deintensification of therapy if clinicians rely on a calculated LDL-C to make patient management decisions.⁹ Therefore, direct measurements of LDL-C or non-HDL-C (using TC and HDL-C) should be used to monitor lipid-lowering therapy when the LDL-C drops below 70 mg/dL.

So, how low is too low? LDL-C levels in newborns range from 20-40 mg/dL, suggesting a physiologic lower limit.¹⁰ Furthermore, patients with genetic variants that result in very low LDL-C levels do not appear to experience any significant adverse effects.¹¹ Of course, this is not the same as treating patients with medications to achieve very low LDL-C levels. Nevertheless, the findings from this meta-analysis support lowering LDL-C in patients with ASCVD, regardless of the patient’s baseline LDL-C levels. While low levels of LDL-C appear to be safe in the short-term, we’ll need to confirm the long-term safety. Whether intensifying therapy by adding non-statins is cost-effective is another matter!

Should we advocate for more intensive or less intensive LDL-C lowering? Are the reductions in ASCVD events enough to outweigh the potential risks and costs for patients? Back to our case — if the patient has established ASCVD, a dose reduction is unnecessary and continuing atorvastatin 80mg daily will provide maximal protection against future ASCVD events.  A very low LDL-C does not appear to increase the risk of adverse events, at least in the short term. On the other hand, in the absence of ASCVD, it would require a careful discussion with the patient about the (as yet) unknown benefits and risks. What are your thoughts?