Background
Colorectal cancer (CRC) is one of the most common malignancies and causes of cancer death [
1]. The most important prognostic factor is tumor stage [
2,
3]. However, each tumor and patient is unique [
4], and more exact categorization based on the features of the tumor and the host could improve our understanding on CRC and enable more individualized disease classifications and treatments.
Platelets are anucleate cell fragments generated by megakaryocytes in the bone marrow and also in other organs including the lung [
5,
6]. Their primary function is to contribute to hemostasis and prevent bleeding [
5,
7]. However, they are also involved in multiple other physiological processes, including inflammation, immunity, angiogenesis, and vessel remodeling [
7]. Thrombocytosis, i.e., increased blood platelet count, is either reactive (secondary thrombocytosis) or caused by a clonal bone marrow (myeloproliferative) disorder [
8]. The drivers of secondary thrombocytosis include acute or chronic infection or inflammation, iron deficiency, hemolytic anemia, asplenia, and cancer [
8].
A multitude of evidence links platelets with cancer pathogenesis [
5,
7]. For example, platelets, reacting to the modified tumor vasculature, can release growth factors or enzymes that stimulate cancer cell proliferation or angiogenesis or degrade extracellular matrix [
7]. Additionally, platelets and their released mediators such as cytokines can contribute to the regulation of tumor associated inflammatory reactions [
7]. Thrombocytosis is common in CRC, and three recent meta-analyses, based on 30 studies [
9], 16 studies [
10], and 9 studies [
11] indicate that elevated preoperative platelet count is associated with shorter overall and disease-free survival in CRC. However, comparative analyses including additional prognostic parameters such as lymphatic invasion would be required to establish elevated platelet count as a relevant prognostic indicator in CRC.
Aspirin is one of the most commonly used drugs, with role as analgesic, antipyretic, and agent for cardiovascular prophylaxis [
12]. The cardioprotective effects of aspirin are thought to be primarily based on inhibition of platelet production of TXA2 (thromboxane A2) [
13], resulting in reduced platelet activation and aggregation. Substantial evidence from observational studies and randomized controlled trials support the efficacy of aspirin in the prevention of cancer, especially CRC [
12,
14,
15]. Moreover, aspirin use has been associated with improved outcome in CRC [
12,
16]. Earlier studies have indicated that the benefit of aspirin may be related to the tumor characteristics, such as PTGS2 expression [
16] or PIK3CA mutation [
17], but also platelet-related mechanisms of action have been suggested [
12]. Thus, we hypothesized that the beneficial survival effect of aspirin in CRC could be stronger for patients with high platelet counts.
Colorectal cancer can elicit an anti-tumor immune response that restricts tumor growth and is associated with improved survival [
18‐
20]. However, a systemic inflammatory response to cancer is associated with adverse outcome and may facilitate cancer progression by recruiting or mobilizing tumor-promoting inflammatory cells, attenuating the anti-tumor immune response, enhancing tumor cell migration and circulating tumor cell survival, and modifying the parenchyma of the pre-metastatic sites [
21]. We have previously shown that CRC patients have increased serum levels of IL-6, IL-7, IL-8, and platelet-derived growth factor BB (PDGF-BB), and the patients with distant metastasis have higher serum levels of IL-1ra, IL-4, IL-6, IL-7, IL-8, MCP-1, and PDGF-BB [
22]. Although platelets are considered important regulators of tumor associated inflammatory reactions [
7,
21], the relationships between platelet count, serum cytokine milieu, and tumor infiltrating immune cells in CRC are currently poorly understood.
In this study, we analyzed blood platelet counts in a prospectively recruited cohort of 356 CRC patients and studied its relationships with patient characteristics including aspirin use; markers of systemic inflammation (modified Glasgow Prognostic Score, mGPS; serum levels of CRP, albumin, and 13 cytokines) and blood hemoglobin levels; tumor characteristics including five types of tumor infiltrating immune cells (CD3, CD8, FoxP3, Neutrophil elastase, mast cell tryptase); and survival.
Discussion
Platelets have been implicated with an important role in cancer pathogenesis, especially in regulating inflammatory reactions. The main findings of this study indicate that high platelet count is associated with elevated systemic inflammatory markers in CRC patients. Platelet count, however, was not statistically significantly associated with the density of tumor-infiltrating immune cells or CRC patient survival.
Systemic inflammation can facilitate tumor progression and metastasis [
21]. One of the best validated systemic inflammatory prognostic markers, mGPS is comprised of serum CRP and serum albumin [
29,
39]. The results of this study indicate that high platelet count is associated with elevated mGPS and serum CRP levels, supporting the role of thrombocytosis in cancer associated systemic inflammation.
For a more detailed analysis of the relationships between platelet count and systemic inflammatory regulators, we measured serum levels of 13 cytokines. We utilized Cytoscape, an open source software platform for visualizing complex networks [
37], in creating a 2D visualization of the relationships between blood platelet count and serum cytokine levels, since visual information can facilitate the comprehension of large quantities of correlation data [
40]. The analysis showed that there were strong positive correlations between platelet count and multiple serum cytokine levels, most notably with IL-7, IL-1RA, and PDGF-BB. The mechanisms underlying the observed correlations between platelet count and serum cytokines were not addressed in our study. Thus, these correlations may indicate that (a) platelets contribute to the production of these cytokines or store and release these cytokines, (b) these cytokines contribute to the increased platelet production, or (c) some other unrecognized factors such as shared background factors are involved.
Platelet granules are packed with enzymes, growth factors and cytokines, which are released on platelet activation [
41]. Accordingly, the concentrations of many cytokines in plasma and serum samples are not equal. For example, among the cytokines showing high correlation with platelet count in this study, IL-7 and PDGF-BB are found in platelet granules [
41,
42]. A recent study utilized the same Bio-Rad cytokine panel that was used in this study to compare the cytokine levels in serum and plasma of healthy subjects [
43]. The results indicated that the levels of PDGF-BB, IL-4, IL-7, and CXCL10 were significantly higher in serum samples compared to heparin plasma, suggesting that these cytokines are released from platelets during clotting. However, the same study did not show statistically significant difference in serum levels of IL-1RA, IL-6, IL-8, IL-9, IL-12, IFN-γ, CCL2, CCL4, and CCL11, relative to heparin plasma levels. Of the cytokines potentially released from platelet granules, IL-7 is a major regulator of T cell homeostasis, whereas its role in malignancy is controversial [
44]. PDGF-BB exerts growth factor functions during embryonal development and adult tissue homeostasis and repair such as wound healing, whereas in cancer, this molecule can directly stimulate the growth and survival of tumor cells and tumor stromal cells [
45]. IL-4 is an anti-inflammatory cytokine, which can also contribute to the survival of CRC stem cells [
46]. Thus, the particles released by platelets can contribute to the regulation of tumor growth and tumor associated inflammatory reactions.
Thrombopoietin is regarded as a key hormone in megakaryocyte differentiation and proliferation, resulting in platelet production [
8]. The process is closely regulated, since thrombopoietin in plasma binds to the circulating platelet surface receptors and only the remaining free thrombopoietin is available to promote megakaryocyte proliferation [
8]. Hepatocytes are a major source of thrombopoietin, and inflammation, particularly IL-6, increases thrombopoietin production [
8]. Accordingly, we observed positive correlation between platelet count and serum IL-6 levels, potentially reflecting the up-regulation of thrombopoietin production caused by increased availability of IL-6 in CRC. Notably, our earlier study has indicated that IL-6 is among the cytokines showing the highest increase in CRC, as well as the strongest association with high tumor stage [
22]. Thus, thrombocytosis is one of the potential systemic consequences of cancer-induced liver-reprogramming. The potential effects of cytokines on platelets are not limited to platelet production, since a recent study indicated that the presence of cytokines such as IL-6 and IL-8, both of which show positive correlation with platelet count in our data, can also alter platelet structure causing platelet hyperactivation [
47].
An anti-tumor immune response can control cancer growth [
48], and increased density of tumor infiltrating lymphocytes has been associated with improved survival in CRC [
18,
19]. We hypothesized that, by the release of soluble inflammatory mediators, platelets could contribute to the regulation of tumor immune cell infiltration. However, no statistically significant multivariable adjusted correlations were observed between platelet count and the densities of tumor infiltrating CD3
+, CD8
+, and FoxP3
+ T cells, neutrophils, or mast cells. This suggests that platelet count only account for a minority of the changes in the densities of tumor infiltrating immune cells, while other factors such as microsatellite instability associated with increased tumor immunogenicity are more important [
49,
50].
Improved prognostic parameters are needed to classify CRC into more homogenous and therapeutically relevant groups. Three recent meta-analyses [
9‐
11] suggest that elevated preoperative platelet count is associated with decreased survival in CRC. In our present study, higher platelet counts were observed in higher tumor stage, but platelet count was not significantly associated with survival. Thus, the results do not support the prognostic value of platelet count in unselected stage I–IV CRC patients. Our sample size did not enable comprehensible evaluation of the effect of platelet count in more specific patient subgroups, such as stage II patients. Relative to earlier studies, the major advantage of this study was the inclusion of vast number of additional prognostic parameters such as lymphatic and blood vessel invasion, anemia, and mGPS, for comparison. In addition to elevated systemic inflammatory markers, low blood hemoglobin level was one of the best determinants of high platelet count in this study. The association between reactive thrombocytosis and anemia is also well-known in general population but its pathophysiology remains incompletely understood [
51]. We observed high platelet count in patients with either normocytic or microcytic anemia. Microcytic anemia is most commonly due to iron deficiency, while chronic inflammatory conditions are among the most common causes of normocytic anemia in general population [
52]. In CRC patients, not only normocytic but also microcytic anemia, is associated with systemic inflammation [
23,
53]. Both systemic inflammation and anemia have been associated with adverse clinical outcome in CRC [
29,
53‐
55], and this study highlights the importance to include them as comparison for platelet count also in the subsequent studies evaluating the prognostic significance of platelet count.
Inhibiting cyclooxygenase enzymes and thus platelet activation, aspirin is frequently used in cardiovascular event prevention [
56]. In the setting of cancer, aspirin use has been associated with decreased CRC incidence and improved CRC outcome [
12]. We hypothesized that the beneficial effect of aspirin in CRC could be limited to patients with high platelet counts. However, we did not observe significantly improved survival in the patients who received aspirin regardless of the platelet count. Small patient numbers in some subgroups, in particular, in the aspirin treated group with high platelet count, might have contributed to this negative result, increasing the risk of type 2 error. Moreover, our sample size did not enable us to analyze, whether different aspirin dosages (100 mg/day or 50 mg/day) were differentially associated with tumor or patient characteristics or survival. Thus, additional studies with larger cohorts are warranted. In addition, the aspirin use data was based on medical records, and potential misclassification due to missing data on over the counter aspirin use has to be taken into account.
Also, some other limitations need to be considered in the interpretation of our results. First, tissue microarrays were used in the immunohistochemical analyses. As only a minor fraction of tumor tissue is analyzed, this may not be representative of the whole tissue section. However, numerous studies have generated reproducible results using tissue microarrays [
57]. Moreover, our tissue microarrays were optimized for immune cell counting, including large (diameter 3.0 mm) cores from different parts of the tumor (Fig.
2). An accurate, earlier validated computer assisted immune cell counting method was used [
35]. Second, multiple hypotheses were tested in this observational study. However, we adjusted the level of statistical significance of our exploratory analyses to p = 0.005 [
38] and interpreted the results with p = 0.05–0.005 (considered borderline statistical significance) cautiously. This approach results in some increase in type 2 statistical error but reduce the risk of type 1 error. The advantages were a prospectively recruited study population, with consistent and extensive histological analysis, including additional prognostic parameters such as lymphatic invasion, and systemic inflammatory markers. An assemblage of tumor infiltrating immune cell types and serum cytokines were analyzed, creating a detailed view on the relationships between platelet count and inflammatory markers.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.