Association of lipoprotein(a) levels with risk of serious bleeding : findings from the ODYSSEY OUTCOMES trial

Background

Observational data have suggested an inverse relationship between lipoprotein [Lp](a) and risk of bleeding, including cerebral hemorrhage. This is a potential safety concern for investigational Lp(a)-lowering agents. In a post hoc analysis of the ODYSSEY OUTCOMES trial, we evaluated the association of Lp(a) and bleeding events. The trial population with recent acute coronary syndromes (ACS) was opportune due to high prevalent use of potent antiplatelet therapy.

Methods

ODYSSEY OUTCOMES enrolled 18924 patients 1-12 months post ACS and elevated atherogenic lipoproteins despite optimized statin therapy, randomized to PCSK9 inhibitor alirocumab (ALI) or placebo, and followed for median 2.8 years. Lp(a) was measured by mass assay at baseline, months 4 and 12. Investigator-reported serious bleeding events were identified from serious adverse events lists, blinded to treatment assignment.

An adjusted analysis accounted for study treatment, baseline use of antiplatelet or anticoagulant agents, systolic blood pressure, and history of hypertension or diabetes. Because arterial revascularization occurred more often after randomization in patients with higher Lp(a) and often leads to intensified antiplatelet therapy (and thus risk of bleeding), it was accounted for as a time-varying covariate.

Results

At baseline, median (Q1, Q3) Lp(a) was 21.3 (6.7, 59.6) mg/dL; 89% of patients were treated with at least one antiplatelet agent. Alirocumab reduced time-weighted Lp(a) by a mean 25%. There were 334 first serious bleeding events (including adjudicated hemorrhagic strokes).

The adjusted analysis showed a non-significant trend to higher serious bleeding risk with higher baseline Lp(a) quartile (Figure, Ptrend across Lp(a) quartiles 0.10). There was also no relation between 55 intracranial bleedings and Lp(a) quartiles (Ptrend 0.67)

Overall, there were numerically fewer serious bleeding events with ALI than with placebo: HR (95% CI) = 0.810 (0.653, 1.005), p=0.056. This treatment effect was consistent across quartile of baseline Lp(a) (Ptrend =0.64).

Finally, in the ALI group with 150 serious bleeding events, there was a lower risk of bleeding with time-weighted decreases of log2(Lp(a)) from baseline (adjusted p=0.01): for the mean decrease in Lp(a), the HR for bleeding (95% CI) was 0.86 (0.77, 0.97).

Conclusion

In this post hoc analysis, there was no evidence of an inverse association of Lp(a) levels with serious bleeding or with cerebral hemorrhage. There was no indication that treatment with ALI or decreases in Lp(a) on ALI increased bleeding; to the contrary, decreases in Lp(a) on ALI were associated with decreased bleeding risk, a finding that might be related to ALI prevention of ischemic events and therefore intensity or duration of antiplatelet therapy. These reassuring findings require confirmation in ongoing randomized trials evaluating targeted Lp(a)-lowering agents.Figure