Interactions of platelets with obesity in relation to lung cancer risk in the UK Biobank cohort

Abstract

Background: Platelet count (PLT) is associated positively with lung cancer risk but has a more complex association with body mass index (BMI), positive only in women (mainly never smokers) and inverse in men (mainly ever smokers), raising the question whether platelets interact with obesity in relation to lung cancer risk. Prospective associations of platelet size (an index of platelet maturity and activity) with lung cancer risk are unclear.

Methods: We examined the associations of PLT, mean platelet volume (MPV), and platelet distribution width (PDW) (each individually, per one standard deviation increase) with lung cancer risk in UK Biobank men and women using multivariable Cox proportional hazards models adjusted for BMI and covariates. We calculated Relative Excess Risk from Interaction (RERI) with obese (BMI ≥ 30 kg/m²), dichotomising platelet parameters at ≥ median (sex-specific), and multiplicative interactions with BMI (continuous scale). We examined heterogeneity according to smoking status (never, former, current smoker) and antiaggregant/anticoagulant use (no/yes).

Results: During a mean follow-up of 10.4 years, 1620 lung cancers were ascertained in 192,355 men and 1495 lung cancers in 218,761 women. PLT was associated positively with lung cancer risk in men (hazard ratio HR = 1.14; 95% confidence interval (CI): 1.09-1.20) and women (HR = 1.09; 95%CI: 1.03-1.15) but interacted inversely with BMI only in men (RERI = - 0.53; 95%CI: - 0.80 to - 0.26 for high-PLT-obese; HR = 0.92; 95%CI = 0.88-0.96 for PLT*BMI). Only in men, MPV was associated inversely with lung cancer risk (HR = 0.95; 95%CI: 0.90-0.99) and interacted positively with BMI (RERI = 0.27; 95%CI = 0.09-0.45 for high-MPV-obese; HR = 1.08; 95%CI = 1.04-1.13 for MPV*BMI), while PDW was associated positively (HR = 1.05; 95%CI: 1.00-1.10), with no evidence for interactions. The associations with PLT were consistent by smoking status, but MPV was associated inversely only in current smokers and PDW positively only in never/former smokers. The interactions with BMI were retained for at least eight years of follow-up and were consistent by smoking status but were attenuated in antiaggregant/anticoagulant users.

Conclusions: In men, PLT was associated positively and MPV inversely with lung cancer risk and these associations appeared hindered by obesity. In women, only PLT was associated positively, with little evidence for interaction with obesity.

Keywords: Interaction; Lung cancer; MPV; Obesity; PDW; Platelet count.

Fig. 1

Associations of platelet parameters (continuous scale) with lung cancer risk. BMI body mass index, Cases number of lung cancer cases per group, CI confidence interval, HR hazard ratio, MPV mean platelet volume, NW normal weight BMI = 18.5 to < 25 kg/m²; OW overweight BMI = 25 to < 30 kg/m²; OB obese BMI = 30 to < 45 kg/m²; p-value Wald test for the individual term, PDW platelet distribution width, PLT platelet count, SD standard deviation. Cox proportional hazards models with exposure either PLT, MPV, or PDW (sex-specific z-scores, value minus mean divided by standard deviation after log-transformation), stratified by age at recruitment, region, and in women, menopausal status and hormone replacement therapy use, and adjusted for BMI and height (sex-specific z-scores), recent weight change, smoking status and intensity, alcohol consumption, physical activity, Townsend deprivation index, family history of cancer, time of blood collection, fasting time, diabetes, and use of lipid-lowering drugs, antihypertensive drugs, antiaggregant/anticoagulants, non-steroidal anti-inflammatory drugs, and paracetamol. p_obese & p_sex – p-value comparing the association with PLT, MPV, or PDW between OB and NW/OW or between men and women with the augmentation method of Lunn and McNeil [20]

Fig. 2

Associations of platelet parameters (continuous scale) with lung cancer risk: sensitivity analyses. Cases number of lung cancer cases per model, CI confidence interval, HR hazard ratio, MPV mean platelet volume, p-value Wald test for the individual term, PDW platelet distribution width, PLT platelet count, SD standard deviation. Cox proportional hazards models with exposure either PLT, MPV, or PDW (sex-specific z-scores, value minus mean divided by standard deviation after log-transformation), with the following stratifications, adjustments, and follow-up times: Age—stratified by age at recruitment, with follow-up from recruitment and no adjustment. Smoking—like “Age”, additionally adjusted for smoking status and intensity. Main model—stratified by age at recruitment, region, and in women, menopausal status and hormone replacement therapy use, and adjusted for body mass index (BMI), height, recent weight change, smoking status and intensity, alcohol consumption, physical activity, Townsend deprivation index, family history of cancer, time of blood collection, fasting time, diabetes and use of lipid-lowering drugs, antihypertensive drugs, antiaggregant/anticoagulants, non-steroidal anti-inflammatory drugs, and paracetamol. Follow-up: ≥ 2 years / ≥ 8 years—like “Main model”, excluding participants with less than 2 or 8 years of follow-up and lagging the entry date with 2 or 8 years, correspondingly. Smoking: Never / Former / Current—like “Main model”, in groups according to smoking status (retaining the adjustment for smoking intensity and time since quit). Antiaggregant: No / Yes—like “Main model”, in groups according to antiaggregant/anticoagulant use. p _current & p _{antiaggregant}—p-value comparing the association with PLT, MPV, or PDW between current and never/former smokers or between groups according to antiaggregant/anticoagulant use with the augmentation method of Lunn and McNeil [20]

Fig. 3

Additive and multiplicative interactions of platelet parameters with obesity: sensitivity analyses in men. BMI body mass index, Cases number of lung cancer cases, CI confidence interval, HR hazard ratio, MPV mean platelet volume, p-value p-value for RERI derived with the delta method or p-value from Wald test for the multiplicative interaction term, PLT platelet count, RERI relative excess risk from interaction (additive interaction). Cox proportional hazards models including a cross-classification with obese (additive interaction) or an interaction term with body mass index (BMI) on a continuous scale (multiplicative interaction) for either PLT or MPV in men with the following stratifications, adjustments, and follow-up times: Age—stratified by age at recruitment, with follow-up from recruitment and no adjustment. Smoking—like “Age”, additionally adjusted for smoking status and intensity. Main model—stratified by age at recruitment and region and adjusted for height, recent weight change, smoking status and intensity, alcohol consumption, physical activity, Townsend deprivation index, family history of cancer, time of blood collection, fasting time, diabetes, and use of lipid-lowering drugs, antihypertensive drugs, antiaggregant/anticoagulants, non-steroidal anti-inflammatory drugs, and paracetamol. Follow-up: ≥ 2/8 years—like “Main model”, excluding participants with less than 2 or 8 years of follow-up and lagging the entry date with 2 or 8 years, correspondingly. Smoking: Never / Former / Current—like “Main model”, in groups according to smoking status (retaining the adjustment for smoking intensity and time since quit). Antiaggregant: No / Yes—like “Main model”, in groups according to antiaggregant/anticoagulant use. Groups for cross-classifications (dichotomised to high/low) were defined as follows: BMI ≥ 30 kg/m² (obese) or ≥ median (sex-specific) for PLT (234.0 men; 261.4 women) and MPV (9.17 men; 9.25 women)