In 47,123 Mexican men and 97,589 women followed prospectively, HDL cholesterol showed non-linear associations with mortality that diverge by sex. Men displayed a J-shaped curve — very high HDL (>95th percentile) raised the hazard ratio for non-CVD deaths to 1.27, rising to 1.82 at the 99th–100th percentile — while women showed a U-shaped pattern, with low HDL elevating CVD mortality risk (HR 1.23) and very high HDL elevating other-cause mortality (HR 1.33). Strikingly, higher small HDL particle count was protective across both sexes: top-quartile small particles were associated with 21% lower all-cause mortality in men (HR 0.79) and 23% lower in women (HR 0.77).
The "HDL hypothesis" — that higher is always better — has been eroding since randomised trials of HDL-raising drugs repeatedly failed to cut cardiovascular events. This large prospective study adds meaningful epidemiological weight to the view that extreme HDL elevation may signal underlying pathology or dysfunctional HDL remodelling rather than protection. The particle-count finding is particularly novel: small HDL particles are typically dismissed as less cardioprotective than large ones, yet here they associate strongly with survival benefit, hinting at reverse-cholesterol transport efficiency over particle size. Limitations include observational design precluding causality, a predominantly Mexican urban cohort limiting generalisability, and residual confounding. As a preprint not yet peer-reviewed, these results — especially the counterintuitive small-particle finding — require independent replication before influencing clinical guidance on HDL targets.