Over the past two decades, hypertension has been reported as a major risk factor for diseases such cardiovascular disease (CVD), diabetes, and kidney disease (1). Hypertension has high incidence and prevalence in developing countries all over the world (2). The Global Burden of Disease study, a collaboration between the World Health Organization and the World Bank, reported detailed risk factors for blood pressure using disability-adjusted life-years, a metric calculated based on premature death due to heart disease (3, 4). Furthermore, recent surveys have reported an increasing prevalence of hypertension in Taiwan, with ~25% of the population being affected (5), and stroke, diabetes mellitus, and CVD are among the leading causes of overall mortality in the Taiwanese population (6). Taiwan has been active in trying to control hypertension, because of its ramifications on public health, and the Taiwan Hypertension Society and the Taiwan Society of Cardiology have been jointly devising policies and methods to monitor hypertension (5). However, not much information is available on the causes of hypertension, and studies designed to understand the factors that have a causal effect on hypertension are needed.

Hypertension is a multi-factorial disease (7–9) that has a strong correlation with a person's platelet count (10). Moreover, American and European clinical guidelines propose antiplatelet drug treatments for reduction of the risk of CVD in patients (11). Similarly, antihypertensive therapies also decrease and prevent the risk of CVD (12, 13). Thus, it can be hypothesized that there exists an association between platelet count and risk of hypertension (14, 15). However, if an exposure has a non-causal association with an outcome, then treatments targeted at the exposure will likely have no real benefit. Therefore, identifying potential risk factors for hypertension and establishing a causal relationship is an emergent public health issue.

Mendelian randomization (MR) studies assess causal inference of a modifiable exposure on an outcome by using genetic alleles as unbiased instruments (16). Observational studies fail to provide conclusive inferences regarding causality, as confounding effects and reverse causality may bias the exposures, thereby leading to overestimation of associations. MR studies are based on the assumption of random assortment of genetic alleles during meiosis and can confer advantages similar to that of a randomized controlled trial by investigating the relationship between genetic alleles that are exclusively associated with an exposure and that affect the disease risk only through exposure (17). The resulting determination of causality is thus free from the effects of confounding and reverse causality. That being said, determining causality alone is not sufficient. The direction of the causality, i.e., whether a rise in platelet counts precedes hypertension or whether hypertension causes elevation of platelet counts, is important for understanding clinical repercussions, which cannot be inferred from simple regression analysis. Therefore, this study aimed to elucidate the bi-directional causality of platelet counts and hypertension using single nucleotide polymorphisms (SNPs) as the instrumental variables (IVs) in a one-sample MR setting, using ~16,000 healthy Taiwanese participants from the Taiwan Biobank (TWB) database.

With hypertension being a serious risk factor for conditions such as CVD, diabetes mellitus, kidney disease, and stroke, detecting elevated levels of blood pressure early on can lead to prevention and control of hypertension. Moreover, several factors such as urbanization, an aging population (high life expectancy), smoking, drinking alcohol, a sedentary lifestyle, and so on, are related to a higher incidence of hypertension (4). Therefore, hypertension has a become a public health issue. Furthermore, >50% of hypertensive individuals are not aware of the condition. Population-based studies, with the goal of identifying causal factors for hypertension, can assist in its early detection, and therefore can help prevent mortality due to CVD. In this study, we conducted a bi-directional causality study for hypertension and platelet count using a general population from Taiwan. A plethora of studies exist in the literature that discuss the elevated platelet counts in CVD patients (38–40). Our study showed that even though both hypertension and platelet count displayed significant association, higher platelet count was observed to have a causal effect on the elevated risk of hypertension.

The prevalence of hypertension rises with age in both males and females; however, rates of developing hypertension in premenopausal women are lower than that of men of the same age (41) while after the age of 50, women apparently have higher rates of hypertension than men (42). Similarly, platelet count has been shown to be a genetic trait regulated by both sex and age (43). Lower platelet counts were observed in men than in women, and platelet count was observed to decrease with age (Supplementary Figure S2). Both age and sex appear to be factors that have significant effects on hypertension and platelet count, and this was reiterated in our analysis (Tables 2, 3). Thus, they were used as covariates for adjustment in our bidirectional MR analysis. Regardless of sex, all people with hypertension are at an increased risk of developing CVD (43). Furthermore, platelet disorders such as thrombocytosis, where the body releases too many platelets, can be triggered by diseases/conditions such as acute bleeding and blood loss, cancer, infections, iron deficiency, splenectomy, and hemolytic anemia (44), while conditions such as alcoholism, autoimmune diseases, bone marrow diseases, and cancer treatments like chemotherapy and radiation therapy can cause thrombocytopenia (low platelet count disorder) (45). However, hypertension has scarcely been reported as a cause for platelet disorders in prior studies. On the other hand, many studies have reported numerous platelet abnormalities including morphologic, biochemical, and functional, to have a causal association with hypertension (10, 14, 46, 47). Our MR analysis confirms all of the above.

Hypertensive drugs may potentially reverse platelet changes, which may occur due to either the direct effect of the antihypertensive drug on the platelets or due to the indirect effect of lowered blood pressure as a result of reduced stress and improved endothelial cell function (47). The effects of antihypertensive drugs on platelet function have been thoroughly characterized through both in vitro experiments and clinical studies. Findings have not been consistent for a given drug class, showing different effects on different parameters of platelet function. In addition, given the wide diversity of study designs, study subjects, and methodologies, it is not yet known if one specific class of drug works better than the others. Although no information about antihypertensive medication use is available in the TWB data, the major conclusion of this study stands, as prior studies have shown consistent findings even after excluding patients who are on antihypertensive drugs (10). Also, there was essentially no reporting bias in this study. This is because among the 3,480 hypertensive patients analyzed in this study, 2,052 were identified based on self-reported questionnaires and 1,428 were identified based on blood pressure measurements. Therefore, patients who took antihypertensive drugs were not classified as healthy, but were identified as hypertensive based on the questionnaire. Hence, these subjects were not neglected due to the treatment effects of drugs.

In hypertension, activation of platelets and endothelial cells leads to thrombotic tendency, which may further result in CVD events such as myocardial infarction and stroke. However, the full pathophysiological mechanism of hypertension has yet to be defined. Hypertension arises due to compromised balance between vasodilators and vasoconstrictors, leading to changes in the vascular structure that result in high blood pressure (48). Elevated platelet counts result in either excessive clotting or abnormal bleeding (49). Prior studies have shown that SNPs associated with platelets may alter platelets in subtle ways to raise the risk of CVD in combination with other risk factors (49). This is precisely what MR analysis illustrates. It infers the causality of an IV-induced modifiable exposure on the outcome. Hence, we believe the SNPs chosen as instruments in the current study had a modifiable effect on the platelet counts, thus demonstrating a positive causality of high platelet count on the risk of developing hypertension. Platelets are blood cells in plasma that stop bleeding by sticking together to form a clot. High platelet counts may cause spontaneous and sudden development of blood clots, and abnormal blood clotting can lead to fatal conditions such as stroke and heart attack. Therefore, based on our findings, higher platelet counts could be used as an early risk indicator of future cardiac events and can be used for public health prevention. Other hematological parameters such as white blood cell count and hematocrits are established prognostic markers of hypertension and CVD and have been shown to be significantly associated through our analysis, indicating that the association may have been on a bone marrow level (50). However, as this was out of the scope of this study, no further analysis was conducted using hematological parameters.

As described above, the validity of IVs in MR depends on 3 assumptions: relevance, independence, and exclusion-restriction. However, only the first assumption can be fully empirically tested, because the second and third assumptions depend on all possible confounders of the exposure-outcome association, both measured and unmeasured. For the IVW method, all of the genetic variants included in the study as IVs need to satisfy the MR assumptions to elucidate a consistent estimate of the causal effect (28). To confirm this, the MR-Egger method and the simple and weighted median methods were conducted as sensitivity analyses. The MR-Egger method estimates the true causal effect consistently under a weaker assumption known as InSIDE, even when all genetic effects are invalid due to violation of the third assumption above (35). However, MR-Egger regression estimates are less precise if the magnitude of association of all genetic variants with the exposure is similar. In contrast, the weighted median method will provide a consistent estimate only if at least 50% of the weight comes from valid genetic variants, under the condition that no single genetic variant contributes to more than 50% of the weight. Compared with the MR-Egger method, the weighted median method allows the MR assumptions to be violated in a more generalized fashion, for the invalid genetic variants (28). Therefore, although a non-significant estimate was observed using the MR-Egger method, we believe that the rest of the data support a causal effect of platelet count on hypertension.

There are some limitations in this study. As only baseline platelet count is available in TWB, this could cause less robustness, leading to an error of estimation due to individual variability, and should be compensated for by the averaging of measurements such as mean platelet volume (MPV), which is an indicator of platelet activation and has been reported to be elevated in hypertensive patients. Unfortunately, MPV isn't available in TWB and therefore, a similar analysis using MPV as the exposure needs to be conducted in an independent Taiwanese cohort to verify the conclusions from this study. Secondly, information such as whether the hypertensive patients were on antihypertensive drugs was also unavailable in TWB. Factors such as this are critical, as antihypertensive drug use may have a confounding effect on the causal effect of platelet count on hypertension. Again, future cohort studies involving hypertensive patients need to be conducted by addressing drug use as a confounding variable in the MR analysis. A total of 2,052 (59.0%) study subjects were categorized as hypertensive based on the self-reported questionnaire. Among the 2,052 self-reported hypertensive patients, 854 also showed abnormal blood pressure from the measurement data. Under such circumstances, some misclassification cannot be ruled out. The TWB dataset is from a cross-sectional study. Due to restrictions imposed by the data, we could only observe one direction of the causal effect. Furthermore, for the MR method, the assumption of independence from confounding factors did not hold in some situations, as there may still be some unknown confounding factors that could affect either platelet count or hypertension, or both, which were not recorded in the TWB data. Therefore, the causality cannot be guaranteed with complete confidence, if unknown factors affect the relationship. The only way to account for and eliminate unknown confounding factors is to reveal them under varying conditions between platelet count and hypertension by conducting plenty of trials with different study designs. According to the exclusion-restriction assumption for MR, the effect of genetic variants on hypertension is exerted through the platelet count only. However, this assumption is difficult to confirm (and to believe). Even though some methods have been developed to solve this problem, such as the weighted median method, there is residual bias that is hard to eliminate, and more domain knowledge and more extensive exploration on the genetic variants are needed. Even though these issues were beyond the scope of our research, we believe the findings of this study have been impacted by them, and they need to be addressed in the future.

This study involved a bi-directional MR analysis of causality between platelet count and hypertension. A significant positive causal effect of platelet count on hypertension was revealed, whereas the causal effect of hypertension on platelet count was reported to be non-significant. Further research on the relationships of other platelet indices with hypertension using larger and multiple data sources can provide more evidence on the relationship between platelet count and hypertension, with the goal of developing better treatments for hypertension.

Higher platelet count was observed to have a significant causal effect on the elevated risk of hypertension through this bi-directional MR study. However, future functional assays are warranted to elucidate the related biological pathways and pathogenic mechanisms.

Researchers can obtain the data by request to Taiwan Biobank through the official application. Requests to access these datasets should be directed to Taiwan Biobank team, https://www.twbiobank.org.tw/new_web.

The studies involving human participants were reviewed and approved by Taiwan Biobank. The patients/participants provided their written informed consent to participate in this study.

The study was conceptualized and designed by T-PL. Data curation was done by T-HH and C-HL. Data analysis and interpretation were done by P-CC, AC, and M-CW. Resources and supervision were done by T-PL and AC. Writing, reviewing, and editing of the manuscript were done by AC and P-CC. All authors have read and approved the final version of the submitted manuscript.

The work was supported in part by the Center of Genomic and Precision Medicine, National Taiwan University, Taiwan (106R8400); the Center for Biotechnology, National Taiwan University, Taiwan (GTZ300); the Taiwan Ministry of Science and Technology (Grant Nos. MOST-109-2314-B-002-151-MY3 and MOST-106-2314-B-002-134-MY2); and the National Taiwan University Higher Education Sprout Project (NTU-110L8810) within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan. The funders had no role in the design and conduct of the study, collection, management, analysis and interpretation of the data, preparation, review, or approval of the manuscript, and decision to submit the manuscript for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.