Background
Platelets can promote carcinogenesis by releasing growth and angiogenic factors and extracellular vesicles, which induce changes in stromal and tumour cells [
1]. The lung, as well as a major cancer cite [
2], is a major site of terminal platelet production from circulating megakaryocytes [
3]. Correspondingly, platelet count (PLT) is not only higher at the time or shortly prior to lung cancer diagnosis [
4,
5], but lung cancer risk remains consistently higher in individuals with higher PLT for at least ten years prior to diagnosis [
6].
Although general obesity, as reflected in body mass index (BMI), is associated with higher risk of venous thromboembolism [
7], BMI shows an inverse, smoking-related association with lung cancer risk [
8,
9]. Moreover, we have previously shown in UK Biobank that BMI is associated positively with PLT only in women, most strongly in never smokers, but inversely in men, most strongly in ever smokers [
10]. This raises the question whether PLT interacts with obesity in relation to lung cancer risk.
PLT is only one aspect of platelet pathology and does not provide information about platelet functionality, while platelet size could be indicative of platelet activity, as thrombotic conditions are associated with large mean platelet volume (MPV) [
11] and large platelet variability (platelet distribution width, PDW) [
12]. Platelet size could also provide information about platelet maturity, as platelet precursors are larger than mature platelets [
13]. Therefore, examining platelet size in conjunction with PLT could provide more information about potential mechanistic pathways than examining PLT in isolation. Little is known, however, about associations of platelet size with lung cancer risk. The available studies are few, with small number of patients, focused on lung cancer diagnosis and prognosis, and reporting mainly higher MPV and PDW in lung cancer patients compared to healthy controls [
14,
15]. To our knowledge, there are no studies evaluating prospectively associations of MPV or PDW with lung cancer risk.
In this study, we used data from the UK Biobank cohort to investigate the prospective associations of PLT, MPV, and PDW with lung cancer risk and their interactions with obesity in men and women.
Discussion
In this study, PLT was associated positively with lung cancer risk in women and men but showed consistent inverse additive and multiplicative interactions with BMI only in men. Also only in men, MPV was associated inversely with lung cancer risk and showed positive additive and multiplicative interactions with BMI, while PDW was associated weakly positively, with no evidence for interactions with obesity.
Our findings corroborate previous prospective studies reporting positive associations of PLT with lung cancer risk within the year prior to diagnosis [
4,
27] and for at least ten years prior to diagnosis [
6]. Given that lung cancer has unfavourable prognosis, with a global mortality-to-incidence ratio as high as 0.82 [
28], longer-term prospective associations with PLT are compatible with a mechanistic involvement of platelets in lung cancer development. A causal association is further supported by positive associations of genetically predicted PLT, which has high genetic heritability [
29], with lung cancer risk [
30]. To our knowledge, however, there are no previous prospective studies to be able to compare the prospective associations of MPV and PDW with lung cancer risk described in our study. Several case–control studies have previously reported higher MPV at lung cancer diagnosis [
14], in contrast to our findings, but a case–control study examining patients with advanced lung cancer has reported lower MPV at diagnosis, in addition to higher platelet count [
31]. Although prospective associations may be retained to cancer diagnosis, studies recruiting cases and controls at cancer diagnosis would also reflect cancer-related changes and could thus be influenced by reverse causality, hence potentially explaining the differences between previous studies and our findings. Reports of poor prognosis for lower MPV measured at diagnosis [
31,
32] may be more relevant to our findings because these are based on prospective studies and reflect lung cancer progression, which may involve pathways relevant to lung cancer development. Cancer survival, however, is dependent on comorbidities related to platelet activity, as well as on cancer progression and metastasis, and a large meta-analysis has found little evidence for association of MPV measured at diagnosis with overall survival [
33]. Our findings are compatible with a small scale study reporting higher PDW at lung cancer diagnosis [
15], but only for men. The retention of the positive association with PDW to at least 8 years of follow-up in our study suggests that this more likely reflects the influence of platelets on lung cancer development, rather than reverse causality.
A plausible mechanism linking PLT to lung cancer development would be an inflammation-related platelet increase, as platelets are involved in immuno-inflammatory responses [
34] and PLT is high in chronic inflammatory conditions [
35,
36]. Inflammatory markers have, indeed, been associated with higher lung cancer risk, more commonly when measured within the years close to diagnosis and in smokers [
37], but also further away from diagnosis [
38] and in never smokers [
39]. Platelets contribute to cancer-associated inflammation by regulating the migration of haematopoietic and immune cells towards the tumour cite and facilitate cancer progression and metastasis by enabling thrombosis and the formation of neutrophil extracellular traps, which protect cancer cells [
40]. Platelet-derived factors are also involved in immunomodulation, as is the case with TREM-like transcript 1 (TLT-1) protein, which is higher in platelets from patients with lung cancer and promotes cancer progression via suppression of CD8 T-cells [
41]. The lung may be a particularly vulnerable organ to platelet action, as platelets are released in the lung from circulating megakaryocytes [
3]. Although the stronger association with PLT closer to lung cancer diagnosis, described in our and in previous studies [
5,
6], indicates an additional cancer-related PLT increase, the cancer is likely to promote an already operational inflammatory pathway.
Notably, the inverse association of MPV with lung cancer risk in men was observed in the same BMI categories (normal weight and overweight) as the positive association with PLT. This is consistent with the inverse correlation between PLT and MPV, which we have previously shown in UK Biobank for a restricted dataset excluding participants with cardiometabolic conditions [
10] and have confirmed in this study for the unrestricted UK Biobank dataset. A potential explanation for an inverse association with MPV coupled to a positive association with PLT would be a trade-off between platelet size and count related to platelet formation, as immature proplatelet intermediates are larger particles and split into two smaller-size platelets in the process of maturation [
13]. Mutations in megakaryocyte cytoskeleton proteins are, indeed, accompanied with large platelet size coupled to low platelet number [
42]. Therefore, large platelet size may reflect more immature and potentially dysfunctional platelets and, hence, a lower risk of lung cancer development. Large platelet size, however, could also indicate platelet activation, because large platelets are more responsive to stimulation and less susceptible to suppression by aspirin [
43] and MPV is associated positively with markers of platelet activation [
44]. As platelet activation would result in a positive rather than an inverse association with lung cancer risk, a suggestion has previously been offered that larger activated platelets are engaged in thrombotic events, leaving only smaller platelets in the circulation of patients with lung cancer [
31].
Although MPV and PDW are associated positively with each other in UK Biobank ([
10] and this study) and both are higher in conditions involving platelet activation [
45], they were associated with lung cancer risk in opposite directions (inverse for MPV, positive for PDW) and in different groups according to smoking status (current smokers for MPV, never/former smokers for PDW) and antiaggregant/anticoagulant use (non-users for MPV, users for PDW). This suggests that MPV and PDW reflect different underlying mechanisms linking platelets to lung cancer development, with larger MPV more likely reflecting lower lung cancer risk due to platelet immaturity and wider PDW more likely reflecting higher lung cancer risk due to platelet activation. One example of a mechanism of platelet activation differentially affecting MPV and PDW is DNA methylation of platelet-endothelial aggregation receptor 1 (PEAR-1), which is associated positively with PDW but not with MPV [
46]. It is unknown, however, whether PEAR-1 is related to lung cancer risk.
PLT and MPV interacted with BMI in opposite directions in men, with obesity apparently hindering their associations with lung cancer risk, but potentially via different mechanisms. Thus, the inverse interaction of PLT with BMI is likely related to obesity contributing to non-alcoholic fatty liver disease (NAFLD) [
47], which can lead to liver fibrosis, and this in turn can contribute to platelet destruction and removal of platelets from the circulation [
48], as we have previously discussed in relation to the inverse association of BMI with PLT in UK Biobank men [
10]. The positive interaction of MPV with BMI, on the other hand, is likely related to oestrogens, which are generated peripherally by adipose tissue aromatase [
49], and are higher in obese UK Biobank men [
24]. In accordance, oestrogens contribute to lung cancer development and progression [
50], including in never smokers [
51], and polymorphisms in the aromatase gene are associated with higher lung cancer risk [
52]. Supporting a link of high-MPV with oestrogens, MPV is higher in women compared to men [
53], oestrogen containing HRT increases MPV [
54], tamoxifen (an oestrogen receptor modulator with oestrogenic effects outside the breast) also increases MPV [
55], and oestradiol (either synthesised within megakaryocytes or extracellular) stimulates the formation of proplatelets, which are larger than mature platelets [
56]. In addition, oestradiol can induce platelet aggregation via oestrogen receptor beta in men and may thus facilitate platelet action [
57].
The associations of platelet parameters with lung cancer risk and their interactions with obesity showed sex differences, as previously did the associations of platelet parameters with obesity in UK Biobank [
10]. This may be explained by the already higher PLT and higher platelet reactivity in women [
58], which may limit additional influences from variations in PLT and MPV. Female sex and oestrogens also appear protective against NAFLD related fibrosis [
59] and thrombopoietin levels are higher in obese women [
60], potentially resulting in stimulated thrombopoiesis, which would explain the positive association of BMI with PLT [
10] and may be preventing an inverse interaction of PLT with obesity in UK Biobank women.
Despite the detailed adjustment for smoking status and intensity, some residual confounding from smoking may have remained in the positive association of PLT with lung cancer risk, as PLT is higher in smokers [
26]. We did not find, however, evidence for heterogeneity of the positive association with PLT between smoking status categories, although lung cancer cases were fewer in never smokers and power was limited, especially for men. The inverse association with MPV, on the other hand, could not reflect residual confounding from smoking because smoking is associated with higher MPV [
26]. While oestrogens contribute to higher MPV and higher lung cancer risk, as outlined above, tobacco smoke components contribute to oestrogen inactivation [
61], which may explain why the inverse association of MPV with lung cancer risk was relevant specifically to current smokers, with large MPV potentially reflecting platelet immaturity rather than platelet activation at lower oestrogen levels. Although smoking is associated with higher PDW [
26], the positive association with PDW is less likely to be influenced by residual confounding from smoking, because it was not observed in current smokers, possibly because the PDW-related pathways are already activated in current smokers.
Although there is interest in using aspirin for lung cancer prevention [
62], it would be hard to separate in observational settings aspirin use from the conditions requiring aspirin use. Thus, lung cancer risk was higher in antiaggregant/anticoagulant users in our study, which is compatible with higher lung cancer risk described for cardiovascular conditions [
63,
64]. Therefore, the attenuation of the associations and interactions with PLT and MPV in antiaggregant/anticoagulant users most likely corresponds to already higher platelet activity in this group, with little possibility left for further influence of variations in PLT and MPV. On the other hand, PDW was associated positively with lung cancer risk only in antiaggregant/anticoagulant users and may thus reflect the extent of platelet activation in this group. Although our study cannot answer the question whether aspirin use modifies lung cancer risk, we have shown modification of the associations of platelet parameters with lung cancer risk by obesity related factors, at least in men, which supports the possibility for modifying lung cancer risk by modifying PLT and platelet action.
A major strength of our study is the prospective cohort design with available platelet measurements and a sizable number of incident lung cancer cases, which permitted examining cross-classifications. Anthropometric measurements, performed by trained personnel and according to standardised protocols, avoided bias from self-reporting. Information for major lifestyle factors (including smoking intensity and time since quit) and drug use permitted adjustment and minimisation of confounding.
A clear limitation of our study is the lack of information about platelet activation or about blood clotting factors, so we were unable to assess platelet function and thrombosis. We were also unable to examine thrombopoiesis and platelet maturity. Our project did not have access to the information for air pollution available in UK Biobank, as examining this was beyond the scope of our project but merits investigation in future studies because air pollution is associated with platelet activation [
65], as well as with higher risk of lung cancer [
66]. Some residual confounding from smoking is possible for the positive association with PLT, as smoking was the most influential covariate. The number of lung cancer cases was insufficient to assess differences in the interaction patterns between lung cancer subtypes, although no major differences have been reported for the positive associations with PLT between lung cancer subtypes [
5,
30]. Exposures and confounders were measured only once, at cohort recruitment, so changes during follow-up could not be accounted for. UK Biobank participants have healthier lifestyle compared to the general population [
67] and mainly have white ethnic background, preventing investigation of ethnic differences. Last, the reported associations may not be causal, due to the observational nature of the study.