Causal Analysis of Platelet Indices and Breast Cancer, Including Estrogen Receptor-Specific Subtypes: A Mendelian Randomization Study

Introduction

Breast cancer is one of the most common malignant tumors among women in the world, with high morbidity and mortality, which brings a heavy burden to women’s health.^1,2 The occurrence of breast cancer is related to many factors, including heredity, endocrine, lifestyle, and environmental factors.^3,4 Especially, estrogen receptor (ER) status is of great significance for the treatment and prognosis of breast cancer.⁵ According to ER status, breast cancer can be divided into ER positive (ER+) and ER negative (ER-) subtypes, with significant differences in biological behavior, treatment response, and prognosis.^6–8 Although significant progress has been made in the diagnosis and treatment of breast cancer and its subtypes, estimating risk remains critical for prevention and treatment strategies.

Platelets, as an important component of blood circulation, have been proven to be associated with the development and progression of tumors such as breast cancer, surpassing their traditional coagulation function.^9,10 It was reported that platelet-secreted factors, such as vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF), could promote angiogenesis by stimulating endothelial cell proliferation and migration, a key process for tumor growth.¹¹ Jiang et al found that platelet-released VEGF stimulated breast cancer cell proliferation through VEGF receptor 2-integrin synergistic signaling.¹² Platelet-derived PDGFB participated in tumor progression by promoting the recruitment of cancer-associated fibroblasts in the tumor microenvironment (TME).¹³ Additionally, platelets could inhibit T-cell activity and enhance pro-tumorigenic functions of myeloid cells (neutrophils, classical monocytes) via surface-overexpressed immune checkpoint molecules, thereby shaping the immunosuppressive TME in breast cancer.¹⁴

Platelet indices, including platelet count (PLT), mean platelet volume (MPV), platelet distribution width (PDW), and plateletcrit (PCT), are important parameters for evaluating platelet function and quantity.¹⁵ In recent years, an increasing number of studies have shown that the change of platelet indices is related to the occurrence, metastasis, and prognosis of tumors such as colorectal cancer, liver cancer, endometrial cancer, and lung cancer.^16–19 For breast cancer, a retrospective cohort study found that PCT predicted clinical outcome and prognosis.²⁰ High levels of MPV, PDW, and PLT were associated with poor prognosis of breast cancer.^21–23 For instance, elevated MPV, indicating platelet activation, was associated with distant metastasis, primary tumor size, and tumor-lymph node metastasis stage of breast cancer.²⁴ The mechanism may be that activated platelets aggregate and wrap around tumor cells to form microthrombi, helping tumor cells evade immune attacks (such as NK cells, macrophages, etc.), thereby promoting metastasis and progression.²⁵ Although these studies may find an association between platelet indices and breast cancer, the causal relationship between platelet indices and the risk of breast cancer and its ER+ and ER- subtypes has not been thoroughly investigated.

Mendelian randomization (MR) analysis is a widely used method in epidemiological research to infer causal relationships between exposures and outcomes.²⁶ Using genetic variations as instrumental variables (IVs) can effectively overcome the confounders and reverse causality problems in traditional observational studies.²⁷ Previous MR analyses have demonstrated a causal relationship between platelet indices, such as PLT, and liver and lung cancer,^28,29 but have not been reported in relation to breast cancer. This study aims to infer the causal relationship between the genetic variation closely related to four platelet indices and breast cancer and its ER+ and ER- subtypes through MR analysis, which is expected to bring new breakthroughs and opportunities for the prevention and treatment of breast cancer.

Material and Methods

Data Source

The genetic tools related to the four platelet indices (PLT, MPV, PDW, and PCT) were selected from a genome-wide association study (GWAS), which comprised 408,112 European participants in the UK Biobank.³⁰ The Breast Cancer Association Consortium (BCAC) provided summary data of overall breast cancer (122,977 cases), ER+ subtype (69,501 cases), ER− subtype (21,468 cases), and control (no breast cancer, 105,974 cases).³¹ All cases and controls were female. The dataset information was shown in Table 1.

Table 1 Genome-Wide Association Study (GWAS) Datasets for MR Analysis

Selection of IVs

Single nucleotide polymorphisms (SNPs) that significantly correlated with four platelet indices (P < 5.0×10⁻⁸) were selected as IVs. To eliminate the influence of linkage disequilibrium (LD) and obtain independent and reliable SNPs, we set specific parameters: r² < 0.001 and clumping distance = 10,000 kb. SNPs associated with exposure were extracted from the GWAS dataset of the outcome variables, which were recorded as IVs, containing information such as effect allele, allele effect size (β), standard error, and P value. SNP harmonization was conducted to ensure consistency in the direction of effect alleles between exposure and outcome datasets, with the following steps: First, the input data containing SNP ID, effect allele, β, and effect allele frequency (EAF) were validated, and then exposure and outcome data were combined through SNP ID. Next, for non-palindromic SNPs (non-A /T or C/G allele combinations), if effect alleles of the exposure and outcome were inconsistent, directional correction was applied to outcome’s effect allele (β were multiplied by −1, EAF were converted to 1-EAF, and effect and non-effect allele labels were exchanged), and SNPs that remain inconsistent after correction were removed. For palindromic SNPs (A/T or C/G) with inconsistent effect alleles, further judgment was based on EAF; SNPs with an EAF within 0.42–0.58 were retained, while those outside this range underwent directional alignment by reversing the outcome’s β (synchronously exchanging allele labels). Ultimately, SNPs missing in the outcome data or remaining unmatchable after the above steps were excluded. The strength of each IV was assessed by F-statistic, calculated as follows: R²= 2×MAF×(1-MAF)×β², F = R²(N−2)/(1−R²), where R² represented the variation proportion of exposure factors explained by each variation coefficient, MAF represented minor allele frequency, β represented the effect size of SNP on exposure, N represented the sample size of the exposure dataset.³² When F > 10, there was no weak IVs bias.

Study Design

To ensure accurate assessment of causal effects in MR analysis, SNPs served as IVs should follow three basic assumptions: firstly, IVs should be significantly correlated with platelet indices (as exposure variable); secondly, IVs must be independent of all potential confounders; finally, IVs can only affect breast cancer (as outcome variable) through platelet indices.

Statistical Analysis

Inverse variance-weighted (IVW), weighted median, MR-Egger regression, simple mode, and weighted mode methods were used to infer causal relationships in MR analysis. Among these methods, IVW was the primary method. It integrated the Wald ratio estimates of the causal effects from different SNPs and provided a consistent estimate of the causal effect of the exposure on the outcome when each genetic variation was applicable to the IV assumptions.³³ The results of IVW method were the most dependable when IVs did not have horizontal pleiotropy.³⁴ Weighted median can provide a consistent estimate of causal effects when at least half of the SNPs serve as valid IVs.³⁵ MR Egger regression was applied to evaluate the horizontal pleiotropy of IVs, and its intercept represented the estimated effect of horizontal pleiotropy.³⁶ When IVs exhibited horizontal pleiotropy, MR Egger regression can still provide unbiased estimates of causal associations. Compared with the MR Egger method, the weighted median method had higher accuracy in the results.³⁷ Simple mode and weighted mode were used as supplementary analysis.³⁸

The heterogeneity test and horizontal pleiotropy assessment of IVs were conducted using the TwoSampleMR package.³⁹ Specifically, the ‘mr_heterogeneity’ function was utilized to calculate the Cochran’s Q value and its corresponding P-value, which was used to evaluate the consistency of effect estimates among different SNPs. The “mr_egger_regression” function was employed for MR-Egger regression analysis, obtaining the intercept estimate value and its P-value for assessing horizontal pleiotropy. The P-value >0.05 indicated the absence of heterogeneity or horizontal pleiotropy. Additionally, the MR-Pleiotropy RESidual Sum and Outlier (MR-PRESSO) test was used to detect and correct for horizontal pleiotropy by removing outliers.⁴⁰ It was performed via the “mr_presso()” function in the MR-PRESSO package⁴¹ with the following parameter settings: OUTLIERtest = TRUE, DISTORTIONtest = TRUE, NbDistribution = 1000, and SignifThreshold = 0.05. Furthermore, the MR-robust adjusted profile score (RAPS) method was used to correct for biases caused by both heterogeneous and systemic pleiotropy (P < 0.05), which provided reliable causal association estimates even in the presence of multiple weak IVs.⁴² We also employed MR Steiger filtering to eliminate SNPs with higher variability in the outcome than exposure and assessed the causal directionality of the retained SNPs with exposure/outcome, which was crucial for avoiding reverse causality bias or pleiotropy risk.⁴³ A direction of “TRUE” and P < 0.05 was statistically significant.

Additionally, we repeatedly analyzed using an independent dataset from the UK Biobank to verify the results’ robustness and assessed their generalizability based on the East Asian population dataset. The relevant dataset information was shown in Supplementary Table 1. Bonferroni correction was used to identify false-positive results generated by multiple tests. P < 0.004 [0.05/(4 exposures × 3 outcomes)] was considered statistically significant, and 0.004 <P < 0.05 was considered suggestive evidence of a potential association. All statistical analyses were performed using R software (version 4.4.0).

Results

Selection of IVs

IVs (P < 5×10⁻⁸) significantly correlated with platelet indices were extracted from GWAS and SNPs with strong LD (r² < 0.001, 10,000 kb) were excluded. Subsequently, palindromic SNPs (ie, A/T or G/C) and SNPs not available in the outcome were removed. Finally, 411 SNPs of PLT, 319 SNPs of PDW, 405 SNPs of PCT, and 377 SNPs of MPV were selected for MR analysis of breast cancer. In total, 411 SNPs of PLT, 322 SNPs of PDW, 407 SNPs of PCT, and 379 SNPs of MPV were selected for MR analysis of ER+ breast cancer. In total, 412 SNPs of PLT, 322 SNPs of PDW, 409 SNPs of PCT, and 376 SNPs of MPV were selected for MR analysis of ER− breast cancer. F-statistics >10 for all IVs indicated no weak IV bias.

Causal Effect of Platelet Indices on Breast Cancer Risk

After Bonferroni correction, MR analysis supported the causal relationship between PCT and the increased risk of overall breast cancer (OR, 1.0693 [95% CI, 1.0281–1.1121]; P = 0.0008) and ER+ breast cancer (OR, 1.0691 [95% CI, 1.0233–1.1169]; P = 0.0028). There was a suggestive causal relationship between PLT and the overall increased risk of breast cancer (OR, 1.0351 [95% CI, 1.0003–1.0711]; P = 0.0483). PDW may have a suggestive association with a lower risk of overall breast cancer and its two subtypes (0.004 <P < 0.05). Although the MR-Egger results of PDW on ER−breast cancer were contrary to other MR methods, they were not statistically significant, and the accuracy of causality revealed by IVW method was better than that of MR-Egger.⁴⁴ In addition, MPV was not associated with breast cancer. The detailed results were shown in Figures 1 and 2.

Figure 1 Forest plot of MR analysis for the effect of platelet indices on breast cancer risk. Abbreviations: MR, Mendelian randomization; SNP, single nucleotide polymorphism; OR, odds ratios; CI, confidence interval; PLT, platelet count; PDW, platelet distribution width; PCT, plateletcrit; MPV, mean platelet volume; ER+, estrogen receptor-positive; ER-, estrogen receptor-negative.

Figure 2 Scatter plots of MR analysis for the effect of platelet indices on breast cancer risk. Abbreviations: MR, Mendelian randomization; SNP, single nucleotide polymorphism; OR, odds ratios; CI, confidence interval; PLT, platelet count; PDW, platelet distribution width; PCT, plateletcrit; MPV, mean platelet volume; ER+, estrogen receptor-positive; ER-, estrogen receptor-negative.

Sensitivity Analysis

Even though Cochran’s Q test results indicated heterogeneity, some heterogeneity was allowed in the primary outcome of the random effects IVW analysis. The MR-Egger intercept (except PDW on ER− breast cancer) indicated no pleiotropy, with P > 0.05. The results were shown in Table 2. In addition, no abnormal IV was found by leave-one-out analysis, which further confirmed the robustness of the results. After global MR-PRESSO testing, it was necessary to exclude some outliers to ensure the effectiveness of the remaining SNPs. MR-PRESSO test results revealed that PCT had a suggestive causal relationship with the increased risk of overall breast cancer and its ER+ subtype in the corrected data. PDW showed a suggestive association with a lower risk of overall breast cancer and its subtypes (ER+ and ER-) in both corrected and uncorrected data. Notably, after outliers were eliminated, the suggestive association between PLT and overall breast cancer disappeared (P = 0.2597), whereas MPV showed a suggestive association with overall breast cancer (P = 0.0291) (Table 3). MR analysis with outliers removed further confirmed the above results, as shown in Table 4. In detail, PCT exhibited a suggestive association with an increased risk of overall and ER+ breast cancer. MPV was also suggestively associated with an increased risk of overall breast cancer. Conversely, PDW demonstrated a suggestive association with a lower risk of overall, ER+, and ER- subtypes. Furthermore, the MR-RAPS test further confirmed the validity of MR causal association (all P < 0.05). The Steiger test revealed that the direction of all MR results was “TRUE” (P < 0.05), indicating no influence from reverse causal effect (Supplementary Table 2).

Table 2 Heterogeneity and Horizontal Pleiotropy Tests of the Impact of Four Platelet Indices on Breast Cancer and Its Subtypes in MR Analysis

Table 3 The MR-PRESSO Test’s Results

Table 4 MR Analysis for the Effect of Platelet Indices on Breast Cancer Risk After Excluding Outliers

External Validation Based on Independent Datasets and East Asian Populations

The repeated validation in the independent dataset showed results consistent with the primary analysis, except for PDW on overall and ER+ breast cancer (Supplementary Tables 3 and 4). These results further supported the suggestive associations of PCT on increased overall and ER+ breast cancer risk, MPV on increased overall breast cancer risk, and PDW on reduced ER- breast cancer risk. However, cross-racial validation of the East Asian population did not find a significant causal association between platelet indices and breast cancer risk, suggesting that this effect may be population-specific (Supplementary Tables 5 and 6).

Discussion

This study explored the causal relationship between four platelet indices and the risk of breast cancer and its subtypes (ER+, ER-). We found that suggestive causal relationships exist between PCT and the increased risk of overall breast cancer and ER+, MPV and the increased risk of overall breast cancer, as well as PDW and the lower risk of overall breast cancer and its two subtypes. However, there was no causal relationship between PLT and the risk of breast cancer and its two subtypes, MPV and the risk of ER+, ER- breast cancer, as well as PCT and the risk of ER- breast cancer. The above findings were repeatedly verified in an independent dataset, but they failed to generalize to the East Asian population. These results provide important clues for understanding the role of platelet indices in the development of breast cancer.

The role of platelet in the pathological mechanism of breast cancer is related to its regulation of tumor angiogenesis and vascular integrity, and its influence on TME.⁴⁵ A study found that platelet-released VEGF and PDGF may influence the development of breast cancer by promoting angiogenesis and tumor growth.⁴⁶ Platelet depletion disrupted vascular integrity in the breast cancer mouse model, causing intratumoral hemorrhage and subsequent cell death.⁴⁷ Platelets interacted with myeloid-derived suppressor cells (MDSCs) to suppress the anti-tumor immunity of T and natural killer (NK) cells and form an immunosuppressive microenvironment that promotes lung metastasis of breast cancer.¹⁰ TME in ER+ breast cancer is usually immunosuppressive.⁴⁸ Antiangiogenic therapy can reverse the immunosuppressive state of breast cancer microenvironment.⁴⁹ As we all know, PCT represents the volume percentage of platelets in the blood. A study showed that it was related to disease-free survival and can be a potential biomarker to predict the clinical results and prognosis of early breast cancer patients.²⁰ These may indicate that the level of PCT affects the risk of breast cancer and its ER+ subtype through the mechanism of influencing angiogenesis, vascular integrity, and TME. The suggestive causal relationship between PCT and the increased risk of overall and ER+ breast cancer found in our MR results echoed these mechanism studies, further suggesting the potential driving role of platelets in the development and progression of breast cancer.

MPV is the average platelet volume in peripheral blood, which is an indicator of platelet activation status.⁵⁰ The interaction between cancer cells and platelets can trigger platelet activation, which enhances the pro-carcinogenic and pro-metastatic functions of platelets.⁵¹ It was reported that the antiplatelet drug Ticagrelor inhibited the metastasis of breast cancer by inhibiting P2Y12-mediated platelet activation.⁵² In addition, significantly higher MPV levels were detected in breast cancer patients, which were positively associated with lymph node metastasis and Ki67 proliferation index in preoperative patients.²¹ These were consistent with the direction of MR suggestive causality that MPV increased the overall risk of breast cancer in this study.

PDW, a parameter reflecting the variability of platelet volume size, is another marker of platelet activation.⁵³ Alterations in the balance between pro- and anti-angiogenic factors released by platelets could affect tumor angiogenesis, thereby inhibiting breast cancer growth and metastasis.⁵⁴ This implied that a higher PDW may indicate a more active state of platelets, and an imbalance in the proportion of various types of factors released by platelets in this active state might affect breast cancer progression. Notably, the association between PDW and breast cancer risk remains controversial in the literatures.^22,55 Liu et al demonstrated that higher PDW levels were associated with better chemotherapy outcomes in breast cancer patients.⁵⁵ Furthermore, Takeuchi et al confirmed that PDW was also associated with ER status.⁵⁶ This is consistent with the direction of suggestive causality between PDW and the lower risk of overall breast cancer and its two subtypes in our MR analysis. However, Huang et al reported the opposite results, indicating that elevated PDW may serve as a marker of poor prognosis in breast cancer.²² These differences may stem from confounders that were difficult to avoid in observational studies. MR analysis overcame these limitations to a certain extent and provided more causal inferential evidence for the relationship between them.

Surprisingly, our study found no causal relationship between PLT and the risk of breast cancer and its two subtypes. However, studies have shown that high PLT are associated with tumor progression, metastasis, and poorer survival in breast cancer patients.^23,57,58 It should be noted that our study may have been limited by sample size, confounding factors, etc., which may have affected the interpretation of the results. Future studies will avoid the above defects to further explore the relationship between them. Collectively, our study revealed suggestive evidence of a causal relationship between platelet indices and breast cancer, providing potential guidance for clinical practice: (i) These platelet indices, as routine items in blood testing, are convenient and low-cost, and are expected to be incorporated into breast cancer risk prediction models as novel biomarkers, especially having potential value for risk stratification in patients with ER subtypes. (ii) Interventions targeting platelet production or functional regulation (such as antiplatelet drugs like aspirin) require subtype-specific evaluation in breast cancer prevention strategies. (iii) The protective association of PDW suggests that platelet heterogeneity may affect the tumor microenvironment through imbalances in the proportions of angiogenic factors.

However, some limitations cannot be ignored. Firstly, this study relied on publicly available GWAS summary data, and although the selected datasets had been applied in other MR studies,^59,60 there may still be potential quality biases. Future studies could further validate the results by incorporating individual-level data. Secondly, the validation in the East Asian population did not replicate the MR results, which might be race-specific or bias caused by insufficient statistical power. Subsequent studies should increase the sample size and validate in diverse populations to enhance the findings’ generalizability. Thirdly, the combined effects of long-term chronic inflammatory status, diet, and other environmental factors were not explored in this study, but they were closely related to platelet function and breast cancer in real life. Given the suggestive results and the limitations of this study, subsequent studies will focus on confirming the reliability and generalizability of these findings through prospective clinical cohort studies and large-scale, cross-racial comparisons. Multi-source data, such as clinical data and biomarker data, will be integrated to more comprehensively assess the impact of genetic variations on breast cancer. The clinical intervention studies will be performed to observe the effects of altered platelet indices by drugs or lifestyle modifications on the risk and prognosis of breast cancer, thereby developing new prevention and treatment strategies. Animal experiments will be conducted to deeply explore the molecular mechanisms of PCT, MPV, and PDW on breast cancer development.

Conclusion

Our MR analysis reveals suggestive evidence of a causal relationship between four platelet indices and breast cancer and its subtypes. Genetic prediction suggests that high PCT increases the risk of overall and ER+ breast cancer, high MPV increases the risk of overall breast cancer, and high PDW is related to a low risk of overall breast cancer and its two subtypes. Although the current results provide suggestive evidence of causal relationships, they offer new insights and directions for breast cancer risk prediction, challenge some previous viewpoints based on traditional observational research, and stimulate further exploration of the relationship between platelet indices and breast cancer risk.