Aernoud Fiolet

369 Clinical implications and future perspectives as large. The effect estimate in this subgroup was directionally consistent, albeit smaller and not statically significant (hazard ratio 0.82, 95% confidence interval 0.61 to 1.11). Whether the differences in observed effect size between strata were statistically significant was not reported. To quantify the relation of treatment effect size and time-to-treatment-initiation, the latter was also analysed as a continuous factor. This analysis confirmed a very early benefit but also had imprecision to detect any signal in patients that started treatment more than 8 days after myocardial infarction. Interestingly, relative risk reduction was greater in patients who initiated treatment more than 21 days after their index infarction. An important caveat in the interpretation of this finding is that not all unequally distributed characteristics were used in the adjustment of these estimators. Secondly, whether this parabolic relationship of treatment initiation and effect size (with a greater effect size seen in both very early [<3 days] and very late [>21 days] treatment initiation) actually reflects separate biological mechanisms of the drug or is a play of chance cannot be distinguished using these data. Early benefits could be based on the potential of colchicine to lower ischemia-reperfusion injury. A 5 day regimen of colchicine (0.5mg twice daily) in patients with ST-elevated myocardial infarction lead to smaller infarct sizes measured using magnetic resonance imaging and troponin T and creatine kinase levels as compared to those taking a placebo. 13 These data suggest that early initiation of low-dose colchicine is justified in patients with myocardial infarction. However, less than 25% of the COLCOT cohort had more than 28 months of follow-up. In contrast to those in the COLCOT cohort, patients from the LoDoCo2 trial were recruited fromoutpatient clinics, irrespective of a prior coronary event, and a fifth of patients in the LoDoCo2 trial completed >4 years of follow-up. In patients with a history of prior acute coronary syndrome, the most recent event occurred more than 4 years prior to randomisation (median time since last acute coronary syndrome 4 years; interquartile range 2–10 years). In all subgroups of patients with a history of prior acute coronary syndrome in the LoDoCo2 trial, benefit of treatment was consistently seen without evidence of a treatment difference for these subgroups (for interaction of the treatment effect and prior acute coronary syndrome status, P = 0.590) (Chapter 8) .