Author + information
- Alawi A. Alsheikh-Ali, MD and
- Richard H. Karas, MD, PhD⁎ ()
- ↵⁎Molecular Cardiology Research Center, Tufts-New England Medical Center, New England Medical Center, 750 Washington Street, Box 80, Boston, Massachusetts 02111
Drs. Rembold and Strandberg appropriately raise the possibility of confounding by age as an explanation for the observed association between achieved low-density lipoprotein cholesterol (LDL-C) and incident cancer (1). To properly address that possibility, we performed a more comprehensive analysis including all the trials in the original study with the addition of 5 large randomized statin trials where incident cancer was reported (Post-CABG [Post Coronary Artery Bypass Graft], GISSI—Prevenzione [Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto Miocardico—Prevention], Kyushu Lipid Intervention, ALERT [Assessment of Lescol in Renal Transplantation], and MEGA [Management of Elevated Cholesterol in the Primary Prevention Group of Adult Japanese]) (2–6). Furthermore, the analysis we performed was weighted and used a random-effects model given the heterogeneity of the trials involved. This more appropriate and comprehensive analysis yields a significant univariate association between age and incident cancer (p = 0.031) as well as a significant inverse univariate association between achieved LDL-C and cancer (p = 0.006). When age and achieved LDL-C are combined in a multivariate model, the inverse association between achieved LDL-C and cancer persists (p = 0.021), whereas that between age and cancer loses statistical significance (p = 0.110). This suggests that the association between achieved LDL-C and cancer is not confounded by age, a finding consistent with previous epidemiologic observations (7). Dr. Kjekshus suggests excluding deaths within the first year after onset of statin treatment to test the hypothesis that the observed relationship is dependent on an association between incipient cancer and low levels of LDL-C. Although such an analysis would be informative, we are unable to perform it because we have no access to individual patient data or dates of cancer diagnoses. Furthermore, it is worth noting that the association between low cholesterol and cancer deaths in the epidemiologic literature persisted after excluding deaths occurring in the first 5 years after measuring cholesterol levels (7). We also agree that a comparative analysis including the placebo groups would be informative, and this is part of ongoing analyses to better understand the association between LDL-C and cancer.
- American College of Cardiology Foundation
- Alsheikh-Ali A.A.,
- Maddukuri P.V.,
- Han H.,
- Karas R.H.
- GISSI Prevenzione Investigators (Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto Miocardico)
- Jacobs D.,
- Blackburn H.,
- Higgins M.,
- et al.