Smoking, blood cells and myeloproliferative neoplasms: meta-analysis and Mendelian randomization of 2·3 million people

Meta-analyses and Mendelian randomization (MR) may clarify the associations of smoking, blood cells and myeloproliferative neoplasms (MPN). We investigated the association of smoking with blood cells in the Danish General Suburban Population Study (GESUS, n = 11 083), by meta-analyses (including GESUS) of 92 studies (n = 531 741) and MR of smoking variant CHRNA3 (rs1051730[A]) in UK Biobank, and with MPN in a meta-analysis of six studies (n (total/cases):1 425 529/2187), totalling 2 307 745 participants. In the meta-analysis the random-effects standardized mean difference (SMD) in current smokers versus non-smokers was 0·82 (0·75-0·89, P = 2·0 * 10-108 ) for leukocytes, 0·09 (-0·02 to 0·21, P = 0·12) for erythrocytes, 0·53 (0·42-0·64, P = 8·0 * 10-22 ) for haematocrit, 0·42 (0·34-0·51, P = 7·1 * 10-21 ) for haemoglobin, 0·19 (0·08-0·31, P = 1·2 * 10-3 ) for mean corpuscular haemoglobin (MCH), 0·29 (0·19-0·39, P = 1·6 * 10-8 ) for mean corpuscular volume (MCV), and 0·04 (-0·04 to 0·13, P = 0·34) for platelets with trends for ever/ex-/current smokers, light/heavy smokers and female/male smokers. Analyses presented high heterogeneity but low publication bias. Per allele in CHRNA3, cigarettes per day in current smokers was associated with increased blood cell counts (leukocytes, neutrophils), MCH, red cell distribution width (RDW) and MCV. The pooled fixed-effects odds ratio for MPN was 1·44 [95% confidence interval (CI): 1·33-1·56; P = 1·8 * 10-19 ; I2 = 0%] in current smokers, 1·29 (1·15-1·44; P = 8·0 * 10-6 ; I2 = 0%) in ex-smokers, 1·49 (1·26-1·77; P = 4·4 * 10-6 ; I2 = 0%) in light smokers and 2·04 (1·74-2·39, P = 2·3 * 10-18 ; I2 = 51%) in heavy smokers compared with non-smokers. Smoking is observationally and genetically associated with increased leukocyte counts and red blood cell indices (MCH, MCV, RDW) and observationally with risk of MPN in current and ex-smokers versus non/never-smokers.