Abstract
The platelet to highdensity lipoprotein cholesterol ratio (PHR) is an emerging indicator reflecting the balance between thrombogenesis and lipid metabolism inflammation. This study aimed to investigate the relationship between PHR and stroke and its subtypes (ischemic stroke, intracerebral hemorrhage, subarachnoid hemorrhage). This study included 400,318 participants without prior stroke history in the UK Biobank. Participants were grouped according to PHR tertiles. Cox proportional hazards regression models were employed to assess the association between PHR and stroke occurrence and subtypes, progressively adjusting for demographic, lifestyle, and metabolic covariates. Restricted cubic spline (RCS) analysis examined nonlinear relationships, while the Fine and Gray competing risks model evaluated cumulative incidence of stroke across PHR groups. The receiver operating characteristic (ROC) curves were performed to explored the additional predictive value of PHR. A total of 16,169 stroke events occurred during follow-up. After multivariable adjustment, PHR showed no significant association with stroke [1.02 (0.97, 1.07)]. However, subtype analysis indicated a positive association of PHR with ischemic stroke [1.10 (1.03, 1.18)] and inversely with intracerebral hemorrhage [0.77 (0.66, 0.90)]. RCS analysis revealed significant nonlinear associations between PHR and stroke risk: U-shaped for overall stroke, J-shaped for ischemic stroke, and L-shaped for intracerebral hemorrhage. The ROC curves indicated that the inclusion of PHR significantly improved the model’s predictive performance. A non-linear association exists between PHR and stroke risk, with opposing directions for ischemic stroke and intracerebral hemorrhage. PHR may serve as a potential marker reflecting the balance between platelet activation and lipid metabolism inflammation, offering novel insights for stroke risk assessment and prevention strategies.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-62796-6.
Keywords: Platelet to high-density lipoprotein cholesterol ratio, Cohort study, Stroke, UKB
Subject terms: Diseases, Medical research, Neurology, Neuroscience, Risk factors
Introduction
Stroke remains one of the leading causes of mortality and long-term disability worldwide, posing a substantial challenge to global public health systems and placing a heavy socioeconomic burden on societies1. According to the latest Global Burden of Disease (GBD) study, approximately 12 million new cases of stroke occurred globally in 2021, resulting in economic losses of more than $890 billion2,3. Despite advances in therapeutic strategies, the overall incidence and recurrence rates of stroke have not markedly declined in many regions. Therefore, identifying modifiable risk factors and readily available biomarkers for early detection and prevention of stroke is of critical importance.
Several composite indices reflecting inflammatory status and lipid metabolism abnormalities—including the monocyte-to-high-density lipoprotein cholesterol ratio (MHR), the neutrophil-to-high-density lipoprotein cholesterol ratio (NHR), and the high-sensitivity C-reactive protein/high-density lipoprotein cholesterol (Hs-CRP/HDL-C) ratio—have previously been shown to be associated with cardiovascular and cerebrovascular risk4–6. It is well known that platelet activation and dyslipidaemia are central to the pathophysiology of cerebrovascular events. Platelets perform pivotal functions in inflammation, atherosclerotic plaque rupture, and thrombus formation7,8, whereas high-density lipoprotein cholesterol (HDL-C) has long been recognized as a “negative risk factor” for cardiovascular and cerebrovascular diseases owing to its anti-inflammatory, antioxidant, and antithrombotic properties9,10. Recently, the platelet to highdensity lipoprotein cholesterol ratio (PHR) gained prominence in the field of lipid metabolism-related inflammation as a novel biomarker. This ratio captures the balance between procoagulant and anticoagulant forces in circulation and is closely associated with metabolic and inflammatory abnormalities11. Compared with MHR, NHR, and Hs-CRP/HDL-C ratio, PHR more directly integrates platelet-mediated procoagulant effects with HDL-C-mediated anti-inflammatory and antithrombotic effects; therefore, it has greater clinical value in elucidating the mechanisms underlying cerebrovascular diseases, particularly ischemic and hemorrhagic events.
Accumulating evidence has linked elevated PHR levels with an increased risk of atherosclerosis, cardiovascular and cerebrovascular diseases6,12–14, peripheral arterial disease, and acute myocardial infarction. However, its association with stroke—particularly with specific subtypes such as ischemic stroke, intracerebral hemorrhage, and subarachnoid hemorrhage—remains insufficiently explored. Considering the heterogeneous mechanisms underlying different stroke subtypes and the distinct contributions of platelet activation and lipid metabolism to cerebrovascular pathology15, elucidating the relationship between PHR and stroke could provide novel biological insights for risk stratification and precision prevention. Accordingly, this study intended to examine the correlation of PHR with stroke, including its major subtypes, in a large population-based cohort. Our findings might help establish a simple, scalable biomarker for stroke prevention and enhance understanding of the interplay between platelet function, lipid metabolism, and cerebrovascular disease.
Methods
Study population
Data from the UK Biobank were utilized in the present study. Between 2006 and 2010, over 500,000 community-dwelling individuals aged 40–69 years were recruited across England, Scotland, and Wales. Through long-term follow-up, UKB intends to explore the aetiology and prevention strategies for various common diseases. Participants completed detailed questionnaires at baseline alongside physical examinations and laboratory tests. UKB accesses participants’ electronic health records, hospital admissions, and death registrations through linkage with the UK National Health Service (NHS). For this study, we included 502,129 UK Biobank participants. First, 83,572 participants with missing PHR data were excluded. Subsequently, 2,955 individuals with a prior diagnosis of stroke and 15,284 participants with missing covariate data were excluded. Ultimately, 400,318 participants were included in the analysis. (Fig. 1)
Fig. 1.
Flow chart of the study participant selection process.
Assessment of the Platelet-to-HDL Cholesterol Ratio
Platelet counts and HDL-C concentrations were obtained from fasting venous blood samples collected at baseline. Platelet counts were determined using automated hematology analysers, while HDL-C levels were measured through standard enzymatic colorimetric assays. The PHR was calculated as the platelet count (×10⁹/L) divided by the HDL-C concentration (mmol/L):
 |
This composite indicator reflects the dynamic balance between thrombogenic and anti-atherogenic processes.
Diagnosis of stroke and its subtypes
The primary outcome of interest was incident stroke, while secondary outcomes included ischemic stroke, intracerebral hemorrhage, and subarachnoid hemorrhage. Stroke events were identified through NHS inpatient records and death registries using diagnostic codes from the International Classification of Diseases, Ninth and Tenth Revisions (ICD-9 and ICD-10). The detailed code definitions are provided in Supplementary Table 1. The follow-up period was defined as the time from baseline assessment until the first occurrence of stroke, death, loss to follow-up, or 30 October 2022, whichever came first.
Covariates
Potential confounders known to influence the association between PHR and stroke were included as covariates in the analyses. Demographic variables included age, sex, ethnicity, educational attainment, household income, and Townsend deprivation index (TDI). Lifestyle-related factors included smoking status, drinking status, physical activity, diet score, and sleep score.
Smoking status was categorized as never, former, or current smoking according to self-reported smoking history. Drinking status was classified as never, former, or current drinking based on self-reported alcohol consumption history16. Physical activity was assessed using self-reported physical activity questionnaires collected in the UK Biobank and defined according to previously published criteria17. Diet score was calculated based on adherence to healthy dietary components, including fruit, vegetable, fish, and processed meat intake, with higher scores indicating healthier dietary patterns18. Sleep score was derived from multiple sleep-related behaviors, including sleep duration, chronotype, insomnia symptoms, snoring, and daytime sleepiness, with higher scores representing healthier sleep patterns19.
Clinical and metabolic variables included body mass index (BMI), total cholesterol, triglycerides, urate, hypertension, hyperlipidemia, and diabetes mellitus. Hypertension, hyperlipidemia, and diabetes mellitus were identified based on self-reported physician diagnosis, medication use, hospital records, or corresponding ICD codes. All covariates were obtained from standardized baseline questionnaires, physical examinations, laboratory measurements, and linked health records from the UK Biobank.
Statistical analysis
The study population was divided into three groups based on PHR quantiles. Continuous variables are presented as mean ± standard deviation (SD) or median (interquartile range [IQR]) as appropriate, while categorical variables are reported as counts and percentages.
Since there is currently no universally accepted clinical cutoff point for PHR, this study grouped PHR values into tertiles. This method divides the study population into relatively balanced low-, medium-, and high-PHR groups, thereby ensuring adequate statistical power for each group and avoiding potential bias associated with the use of arbitrary cutoffs. Additionally, tertile grouping facilitates the examination of whether a dose-response relationship exists between increasing PHR levels and stroke risk, and enhances the interpretability of the results. To evaluate the association between PHR and stroke (and its subtypes), we verified the validity of the Cox proportional hazards assumption (Supplementary Figs. 1–4). Furthermore, Cox proportional hazards models were constructed sequentially: Model 1: adjusted for age, gender, and ethnicity. Model 2: further adjusted for lifestyle factors (smoking status, physical activity, and alcohol consumption). Model 3: fully adjusted for all covariates.
Potential nonlinear relationships between PHR and stroke risk were examined using restricted cubic spline (RCS) analyses. When nonlinearity was detected, inflection points were identified via a recursive algorithm, followed by two-piecewise linear regression models to estimate associations before and after the threshold. The cumulative incidence of stroke in different PHR groups was visualized using the Fine and Gray competing risks model. To explored the additional predictive value of PHR, we performed receiver operating characteristic (ROC) curves. Subgroup analyses and interactions were further conducted to investigate potential subgroup differences. All analyses were conducted using R software (version 4.2.2). A two-sided P value < 0.05 was considered statistically significant.
Results
Characteristics of the study population
Baseline characteristics of the study population are summarized in Table 1. The mean age of participants was 56.49 (8.08) years; 53.79% were female, and 95.10% were of White ethnicity. During follow-up, a total of 16,169 incident stroke cases were identified, including 7,710 ischemic strokes, 1,511 intracerebral hemorrhages, and 845 subarachnoid hemorrhages. Compared with participants in the lower PHR tertiles, those in the highest tertile of PHR were more likely to be male, non-White, current smokers, and obese, and they tended to have higher TDI, lower household income (<£18,000), and elevated triglyceride and urate concentrations. Moreover, individuals in this group showed a higher prevalence of hypertension, dyslipidemia, and diabetes.
Table 1.
The baseline characteristics of participants.
| Characteristic | Total N = 400,3181 |
T1 N = 133,4401 |
T2 N = 133,4391 |
T3 N = 133,4391 |
P |
|---|---|---|---|---|---|
| Age | < 0.001 | ||||
| Median (Q1, Q3) | 58.00(50.00, 63.00) | 59.00(51.00, 64.00) | 58.00(50.00, 63.00) | 57.00(49.00, 63.00) | |
| Gender | < 0.001 | ||||
| Female | 215,314(53.79%) | 84,869(63.60%) | 70,827(53.08%) | 59,618(44.68%) | |
| Male | 185,004(46.21%) | 48,571(36.40%) | 62,612(46.92%) | 73,821(55.32%) | |
| Ethnicity | < 0.001 | ||||
| Non-White | 19,602(4.90%) | 6,001(4.50%) | 6,001(4.50%) | 7,600(5.70%) | |
| White | 380,716(95.10%) | 127,439(95.50%) | 127,438(95.50%) | 125,839(94.30%) | |
| Drinking status | < 0.001 | ||||
| Never | 16,515(4.13%) | 4,124(3.09%) | 5,202(3.90%) | 7,189(5.39%) | |
| Previous | 13,860(3.46%) | 3,295(2.47%) | 4,173(3.13%) | 6,392(4.79%) | |
| Current | 369,943(92.41%) | 126,021(94.44%) | 124,064(92.97%) | 119,858(89.82%) | |
| Smoking status | < 0.001 | ||||
| Never | 219,336(54.79%) | 75,653(56.69%) | 73,912(55.39%) | 69,771(52.29%) | |
| Previous | 139,509(34.85%) | 47,098(35.30%) | 46,548(34.88%) | 45,863(34.37%) | |
| Current | 41,473(10.36%) | 10,689(8.01%) | 12,979(9.73%) | 17,805(13.34%) | |
| Physical activity | < 0.001 | ||||
| No | 152,177(38.01%) | 45,143(33.83%) | 50,360(37.74%) | 56,674(42.47%) | |
| Yes | 248,141(61.99%) | 88,297(66.17%) | 83,079(62.26%) | 76,765(57.53%) | |
| Education | < 0.001 | ||||
| UnKnown | 68,463(17.10%) | 20,201(15.14%) | 22,717(17.02%) | 25,545(19.14%) | |
| College | 131,963(32.96%) | 48,521(36.36%) | 44,032(33.00%) | 39,410(29.53%) | |
| Other levels | 199,892(49.93%) | 64,718(48.50%) | 66,690(49.98%) | 68,484(51.32%) | |
| BMI | < 0.001 | ||||
| < 30 | 304,270(76.01%) | 115,093(86.25%) | 102,460(76.78%) | 86,717(64.99%) | |
| ≥ 30 | 96,048(23.99%) | 18,347(13.75%) | 30,979(23.22%) | 46,722(35.01%) | |
| TDI | < 0.001 | ||||
| Median (Q1, Q3) | −2.20 (−3.67, 0.41) | −2.31 (−3.74, 0.13) | −2.27 (−3.70, 0.26) | −2(−3.56, 0.84) | |
| Household income | < 0.001 | ||||
| Less than 18,000 | 90,302(22.56%) | 27,703(20.76%) | 29,399(22.03%) | 33,200(24.88%) | |
| 18,000 to 30,999 | 100,317(25.06%) | 33,582(25.17%) | 33,376(25.01%) | 33,359(25.00%) | |
| 31,000 to 51,999 | 104,235(26.04%) | 34,843(26.11%) | 35,146(26.34%) | 34,246(25.66%) | |
| 52,000 to 100,000 | 83,027(20.74%) | 28,663(21.48%) | 27,868(20.88%) | 26,496(19.86%) | |
| Greater than 100,000 | 22,437(5.60%) | 8,649(6.48%) | 7,650(5.73%) | 6,138(4.60%) | |
| Diet score | < 0.001 | ||||
| Median (Q1, Q3) | 3.00(2.00, 4.00) | 3.00(2.00, 4.00) | 3.00(2.00, 4.00) | 3.00(2.00, 4.00) | |
| Sleep score | < 0.001 | ||||
| Median (Q1, Q3) | 1.00(1.00, 2.00) | 1.00(1.00, 2.00) | 1.00(1.00, 2.00) | 2.00(1.00, 2.00) | |
| TG | < 0.001 | ||||
| Median (Q1, Q3) | 1.48 (1.04, 2.14) | 1.15 (0.86, 1.59) | 1.50 (1.09, 2.09) | 1.91 (1.36, 2.70) | |
| Urate | < 0.001 | ||||
| Median (Q1, Q3) | 303(250.50, 360.80) | 282.10 (235.30, 338.60) | 303.70 (251.90, 360.20) | 322.90 (269.90, 378.50) | |
| Platelet Count | < 0.001 | ||||
| Median (Q1, Q3) | 248(213.20, 287.00) | 214.30 (186.00, 243.90) | 247.90 (219.10, 279.00) | 286(252.10, 325.00) | |
| HDL | < 0.001 | ||||
| Median (Q1, Q3) | 1.40 (1.17, 1.68) | 1.73 (1.51, 1.99) | 1.41 (1.24, 1.59) | 1.15 (1.00, 1.31) | |
| Hypertension | < 0.001 | ||||
| No | 273,822(68.40%) | 97,971(73.42%) | 92,140(69.05%) | 83,711(62.73%) | |
| Yes | 126,496(31.60%) | 35,469(26.58%) | 41,299(30.95%) | 49,728(37.27%) | |
| Hyperlipidemia | < 0.001 | ||||
| No | 335,267(83.75%) | 115,735(86.73%) | 112,268(84.13%) | 107,264(80.38%) | |
| Yes | 65,051(16.25%) | 17,705(13.27%) | 21,171(15.87%) | 26,175(19.62%) | |
| DM | < 0.001 | ||||
| No | 365,884(91.40%) | 126,975(95.16%) | 123,539(92.58%) | 115,370(86.46%) | |
| Yes | 34,434(8.60%) | 6,465(4.84%) | 9,900(7.42%) | 18,069(13.54%) | |
| Median follow-up time | < 0.001 | ||||
| Median (Q1, Q3) | 13.63 (12.78, 14.36) | 13.47 (12.68, 14.26) | 13.64 (12.83, 14.37) | 13.73 (12.95, 14.43) | |
| Stroke | < 0.001 | ||||
| No | 384,149(95.96%) | 128,526(96.32%) | 128,274(96.13%) | 127,349(95.44%) | |
| Yes | 16,169(4.04%) | 4,914(3.68%) | 5,165(3.87%) | 6,090(4.56%) | |
| Ischemic stroke | < 0.001 | ||||
| No | 392,608(98.07%) | 131,252(98.36%) | 130,939(98.13%) | 130,417(97.74%) | |
| Yes | 7,710(1.93%) | 2,188(1.64%) | 2,500(1.87%) | 3,022(2.26%) | |
| Intracerebral hemorrhage | 0.045 | ||||
| No | 398,807(99.62%) | 132,897(99.59%) | 132,975(99.65%) | 132,935(99.62%) | |
| Yes | 1,511(0.38%) | 543(0.41%) | 464(0.35%) | 504(0.38%) | |
| Subarachnoid hemorrhage | 0.3 | ||||
| No | 399,473(99.79%) | 133,144(99.78%) | 133,152(99.78%) | 133,177(99.80%) | |
| Yes | 845(0.21%) | 296(0.22%) | 287(0.22%) | 262(0.20%) |
Relationship between PHR and stroke and its subtypes
After adjusting for covariates using three models, the association between PHR and stroke gradually weakened; furthermore, when adjusting for all confounding factors, we found no significant association between PHR and stroke overall [1.02 (0.97, 1.07)] (Table 2). Regarding stroke subtypes, although the associations between PHR and ischemic stroke and intracerebral hemorrhage gradually weakened after adjusting for covariates, they remained statistically significant. PHR showed a positive association with ischemic stroke [1.1(1.03, 1.18)]and a negative association with intracerebral hemorrhage [0.77(0.65, 0.90)]. However, no significant association was found with subarachnoid hemorrhage [0.95 (0.76, 1.18])]. Further analysis, using the second tertile of PHR as reference, revealed a U-shaped trend in the relationship between PHR and stroke, with HR (95% CI) values of 1.05 (1.01, 1.10) and 1.06 (1.02, 1.10). PHR exhibited a J-shaped trend with ischemic stroke, with HR (95% CI) of 0.99 (0.94, 1.05) and 1.06 (1.01, 1.12). PHR exhibited an L-shaped trend with HR (95% CI) of 1.24 (1.09, 1.41) and 1.04 (0.92, 1.19) for intracerebral hemorrhage.
Table 2.
Association of platelet-to-HDL ratio with stroke and its subtypes.
| Stroke | Model 1 h(95%CI) | Model 2 h(95%CI) | Model 3 h(95%CI) |
|---|---|---|---|
| PHR | 1.43(1.36,1.50) | 1.31(1.25,1.37) | 1.02(0.97,1.07) |
| PHR category | |||
| Tertile 1 | 0.96(0.92,0.99) | 0.98(0.94,1.02) | 1.05(1.01,1.10) |
| Tertile 2 | 1.0 | 1.0 | 1.0 |
| Tertile 3 | 1.24(1.20,1.29) | 1.19(1.15,1.24) | 1.06(1.02,1.10) |
| Ischemic stroke | |||
| PHR | 1.58(1.47,1.69) | 1.46(1.36,1.56) | 1.10(1.03,1.18) |
| PHR category | |||
| Tertile 1 | 0.89(0.84,0.95) | 0.91(0.86,0.97) | 0.99(0.94,1.05) |
| Tertile 2 | 1.0 | 1.0 | 1.0 |
| Tertile 3 | 1.25(1.19,1.32) | 1.21(1.14,1.27) | 1.06(1.01,1.12) |
| Intracerebral hemorrhage | |||
| PHR | 0.93(0.80,1.08) | 0.88(0.75,1.02) | 0.77(0.66,0.90) |
| PHR category | |||
| Tertile 1 | 1.17(1.04,1.33) | 1.19(1.05,1.35) | 1.24(1.09,1.41) |
| Tertile 2 | 1.0 | 1.0 | 1.0 |
| Tertile 3 | 1.14(1.00,1.29) | 1.11(0.98,1.26) | 1.04(0.92,1.19) |
| Subarachnoid hemorrhage | |||
| PHR | 1.06(0.86,1.29) | 0.98(0.80,1.20) | 0.95(0.76,1.18) |
| PHR category | |||
| Tertile 1 | 0.99(0.84,1.17) | 1.01(0.86,1.19) | 1.03(0.87,1.21) |
| Tertile 2 | 1.0 | 1.0 | 1.0 |
| Tertile 3 | 0.97(0.82,1.14) | 0.93(0.79,1.10) | 0.91(0.77,1.09) |
Model 1: Age, gender, and ethnicity were adjusted.
Model 2: Age, gender, ethnicity, drinking status, smoking status, and physical activity were adjusted.
Model 3: Age, gender, ethnicity, drinking status, smoking status, physical activity, education, BMI, tdi, household income, diet score, sleep score, TG, urate, hypertension, DM, and hyperlipidemia were adjusted.
We further validated the existence of a nonlinear relationship using RCS (Fig. 2). Around the inflection point, the relationship between PHR and stroke exhibited opposite trends: with HR (95% CI) being 0.78 (0.71, 0.86) before the inflection point and 1.31 (1.19, 1.43) thereafter (Table 3). Conversely, the association between PHR and ischemic stroke was statistically significant only after the inflection point, at [1.26 (1.11, 1.44)]. Notably, the relationship between PHR and intracerebral hemorrhage exhibited a negative correlation only before the inflection point [0.56 (0.42, 0.75)](Table 3). Results from the Fine and Gray competing risks model indicated that individuals in the third tertile of PHR had the highest risk of stroke and ischemic stroke (Fig. 3). Conversely, those in the first tertile of PHR had the highest risk of intracerebral hemorrhage (Fig. 3). In addition, we used ROC curves to assess whether PHR could improve the model’s predictive performance. In the prediction of overall stroke, Model 1—which incorporated PHR—showed a nearly 10% improvement in predictive performance compared to Model 2, which relied solely on baseline covariates (Fig. 4). Similarly, the inclusion of PHR also improved the model’s predictive performance for ischemic stroke and intracerebral hemorrhage (Supplementary Fig. 5).
Fig. 2.
Association between platelet-to-HDL ratio and stroke (A: stroke; B: ischemic stroke; C: intracerebral hemorrhage; D: subarachnoid hemorrhage). Age, gender, ethnicity, drinking status, smoking status, physical activity, education, BMI, TDI, household income, diet score, sleep score, TG, urate, hypertension, DM and hyperlipidemia were adjusted.
Table 3.
Threshold effect analysis of platelet-to-HDL ratio with stroke and its subtypes using a two-piecewise linear regression model.
| Stroke | Adjust HR (95% CI) | P value |
|---|---|---|
| Fitting by standard linear model | 1.02 (0.97, 1.07) | 0.42 |
| Fitting by two-piecewise linear model | ||
| Inflection point | 5.17 | |
| < 5.17 | 0.78 (0.71, 0.86) | < 0.001 |
| > 5.17 | 1.31(1.19, 1.43) | < 0.001 |
| Log-likelihood ratio | < 0.001 | |
| Ischemic stroke | ||
| Fitting by standard linear model | 1.10(1.03, 1.18) | 0.01 |
| Fitting by two-piecewise linear model | ||
| Inflection point | 5.17 | |
| < 5.17 | 0.95(0.83, 1.09) | 0.5 |
| > 5.17 | 1.26(1.11, 1.44) | < 0.001 |
| Log-likelihood ratio | 0.013 | |
| Intracerebral hemorrhage | ||
| Fitting by standard linear model | 0.77(0.66, 0.90) | 0.01 |
| Fitting by two-piecewise linear model | ||
| Inflection point | 5.17 | |
| < 5.17 | 0.56(0.42, 0.75) | < 0.001 |
| > 5.17 | 1.06(0.77, 1.45) | 0.7 |
| Log-likelihood ratio | 0.014 |
Fig. 3.
Cumulative incidence of stroke according to baseline PHR tertiles. (A: stroke; B: ischemic stroke; C: intracerebral hemorrhage; D: subarachnoid hemorrhage).
Fig. 4.
ROC curves of Cox regression models for predicting incident stroke.
Subgroup analyses
To investigate whether subgroup differences exist, we conducted subgroup analyses and interaction tests. No significant differences were observed in the relationship between PHR and stroke across subgroups defined by lifestyle habits and chronic diseases (Fig. 5). In subgroup analyses examining PHR in relation to ischemic stroke and intracerebral hemorrhage, no significant interactions were identified across subgroups(Supplementary Fig. 6).
Fig. 5.
Subgroup analysis of association between platelet-to-HDL ratio and stroke.
Discussion
To our knowledge, this is the first large-scale cohort study to examine the relationship of PHR with stroke and its major subtypes. After comprehensive adjustment for confounders, no significant association was observed between PHR and overall stroke incidence. However, RCS analyses revealed a U-shaped relationship of PHR with stroke, with a threshold-dependent reversal of association around the inflection point. This non-linear pattern was further reflected in subtype analyses, where a J-shaped association emerged for ischemic stroke and an L-shaped association for intracerebral hemorrhage. The ROC curve indicated the significant value of PHR in model predictions. Subgroup analyses indicated that the associations remained consistent across all subgroups.
The observed U-shaped association between PHR and stroke risk suggests that both excessively high and low PHR levels may contribute to an increased likelihood of stroke, particularly in cases of ischemic stroke and intracerebral hemorrhage. This bidirectional association suggests that PHR may provide a certain reference for stroke risk stratification. From a clinical perspective, both individuals with low PHR and those with high PHR should adhere to current evidence-based strategies for primary stroke prevention; however, the risk profiles of these two groups may differ. A higher PHR may reflect an increased platelet-associated thromboinflammatory burden and/or reduced HDL-C levels, suggesting that such individuals may require greater attention to metabolic abnormalities and atherosclerotic risk. In contrast, the clinical significance of a lower PHR should not be simplistically interpreted as a protective phenotype. For example, a low platelet count may indicate reduced platelet production, increased platelet consumption, malnutrition, or an autoimmune disease; conversely, extremely high HDL-C levels may also be associated with HDL dysfunction or complex metabolic states. Therefore, regardless of whether PHR is at a lower or higher level, it must be interpreted comprehensively in conjunction with the individual’s clinical context. Furthermore, as this study is observational, PHR should not be regarded as a causal factor for stroke or a direct target for intervention. The inflection point identified through threshold saturation analysis represents only the lowest-risk interval observed in this study, not a recommended therapeutic target. Further mechanistic studies and external validation are needed to clarify the practical value of PHR in personalized stroke prevention.
Previous research examining the relationship between PHR and stroke remains limited, with most studies focusing on isolated biomarkers and yielding inconsistent findings. The prevailing view posits that higher platelet counts are associated with increased risks of thrombosis and ischemic stroke20. Following the rupture of atherosclerotic plaques, exposed collagen interacts with von Willebrand factor through the platelet glycoprotein Ib/IX/V complex, triggering platelet adhesion and aggregation21. The resulting thrombus formation narrows or occludes the vascular lumen, ultimately leading to ischemic stroke22,23. Consequently, antiplatelet therapy remains a cornerstone in the acute management and secondary prevention of transient ischemic attack and ischemic stroke24,25. However, one study in older adults reported no significant relationship between platelet count and stroke incidence26. High-density lipoprotein (HDL) has traditionally been regarded as protective against cardiovascular disease, demonstrating an inverse association with coronary heart disease (CHD). Nevertheless, findings regarding HDL cholesterol and stroke risk have been heterogeneous27. For instance, an Asia-Pacific cohort study found no significant association between elevated HDL-C levels and reduced stroke risk28, whereas Li et al. identified a U-shaped relationship between cumulative HDL-C concentrations and stroke risk, with both low and high HDL-C levels conferring increased risk29. Among the limited research on the relationship between PHR and stroke, two studies of middle-aged and elderly individuals in China demonstrated a linear positive correlation6,30, while analyses from the United States revealed a J-shaped pattern. Notably, both studies relied on self-reported stroke diagnoses, which may limit their validity13.
PHR integrates two pivotal components of vascular homeostasis—platelets and HDL—each playing distinct but interconnected roles in thrombosis and atherosclerosis. Platelets are the key cellular mediators of thrombus formation, releasing inflammatory cytokines and procoagulant factors that promote endothelial dysfunction and accelerate atherogenesis31. Through the release of platelet factor 4 (PF4), P-selectin, serotonin, and platelet-derived growth factor (PDGF), platelets recruit inflammatory cells and amplify local vascular inflammation32–36. Conversely, HDL confers atheroprotective benefits through reverse cholesterol transport and exerts anti-inflammatory, antioxidative, and endothelial-stabilising effects37,38. Reduced HDL levels not only diminish cholesterol efflux capacity but may also impair HDL functionality, weakening its inhibitory influence on platelet activation.
Ischemic stroke often arises from the progression and rupture of atherosclerotic plaques, wherein exposed collagen triggers rapid platelet activation and aggregation, leading to vessel occlusion39. Elevated PHR may therefore reflect a dual pathological imbalance characterized by enhanced platelet-mediated thrombogenicity and reduced HDL-C-mediated vascular protection, both of which may contribute to an increased thrombotic risk. This mechanistic interplay aligns with our observation of increased ischemic stroke risk in participants with high PHR values. In contrast, the pathophysiology of intracerebral hemorrhage is dominated by vascular structural fragility and impaired coagulation40. Lower PHR levels, frequently accompanied by thrombocytopenia, predispose to bleeding, while the protective vascular effects of HDL become overshadowed by platelet deficiency. This inverse association suggests that PHR exhibits stroke subtype–dependent predictive characteristics, and assessing total stroke risk alone may obscure its divergent implications across subtypes.
Several limitations of this study should be noted. First, PHR was assessed at a single time point, precluding evaluation of longitudinal changes. Future studies should investigate temporal variations in PHR and their relationship to stroke incidence. Second, the predominantly White participant population limits generalisability to other ethnic groups. Third, although extensive covariate adjustment was performed, residual confounding cannot be entirely excluded. Finally, as this study was observational, causality cannot be inferred; further mechanistic and interventional research is required to validate these associations.
Conclusion
In summary, this study utilized large-scale population data from the UK Biobank to reveal a U-shaped relationship between PHR and overall stroke risk, yet demonstrated directionally opposite and non-linear associations with distinct stroke subtypes. Elevated PHR levels correlated with increased risk of ischemic stroke, whereas low PHR levels indicated heightened susceptibility to intracerebral hemorrhage. As a convenient indicator integrating platelet and HDL levels, PHR may hold promise as a valuable tool for identifying risks associated with specific stroke subtypes. Future research should explore its dynamic variations and intervention feasibility to advance its application in precision prevention and clinical risk management.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
We thank the participants of the UKB and the UKB staff.
Author contributions
Fengxiao Wang: Methodology, Data curation, Formal Analysis, Writing-original draft and Funding acquisition. Xinyu Chen: Methodology and Writing-original draft. Zhen Liu: Visualization. Haixiao Gao: Methodology. Xixue Lu: Project administration, Data curation and Supervision; Mingyue Xia: Writing-editing. Chunling Bao: Writing-editing and Validation. Yunfei Chen: Writing-review & editing. Xintong Yu: Funding acquisition and Project administration. All authors reviewed the manuscript.
Funding
This work was supported by the Shanghai Oriental Talent Program(QNWS2024098), Three Year Action Plan for Shanghai to Further Accelerate the Inheritance, Innovation and Development of Traditional Chinese Medicine (2025–2027) (1-1-2),
Shanghai Clinical Research Center for Acupuncture and Moxibustion (20MC1920500), and Key Construction Discipline of National Administration of Traditional Chinese Medicine (ZYYZDXK-2023068).
Data availability
The UK Biobank data are available at https://www.ukbiobank.ac.uk. This research has been conducted using the UK Biobank Resource under Application Number 552996.
Declarations
Competing interests
The authors declare no competing interests.
Ethics approval and consent to participate
An approval for the UKB project was obtained from the North West Research Ethics Committee (11/NW/0382) and informed consents were signed by all participants. All methods were performed in accordance with the relevant guidelines and regulations of the Declaration of Helsinki.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Fengxiao Wang and Xinyu Chen contributed equally to this work.
Contributor Information
Yunfei Chen, Email: icyf1968@163.com.
Xintong Yu, Email: yuxt1117@163.com.
References
- 1.Perera, K. S. et al. Evaluating Rates of Recurrent Ischemic Stroke Among Young Adults With Embolic Stroke of Undetermined Source: The Young ESUS Longitudinal Cohort Study. JAMA Neurol.79, 450–458. 10.1001/jamaneurol.2022.0048 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Collaborators, G. B. D. S. R. F. Global, regional, and national burden of stroke and its risk factors, 1990–2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Neurol.23, 973–1003. 10.1016/S1474-4422(24)00369-7 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Feigin, V. L. et al. World Stroke Organization: Global Stroke Fact Sheet 2025. Int. J. Stroke. 20, 132–144. 10.1177/17474930241308142 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Feng, Q., Miao, C. & Gao, X. Associations of the Hs-CRP/HDL-C ratio with stroke among US adults: Evidence from NHANES 2015–2018. J. Stroke Cerebrovasc. Dis.34, 108353. 10.1016/j.jstrokecerebrovasdis.2025.108353 (2025). [DOI] [PubMed] [Google Scholar]
- 5.Huang, Q. & Li, Y. Association between monocyte-to-high-density lipoprotein cholesterol ratio and stroke: a propensity score matched cross-sectional study. Front. Neurol.17, 1721994. 10.3389/fneur.2026.1721994 (2026). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Feng, Q., Ren, L., Chen, S. & Li, X. Platelet-to-high-density lipoprotein cholesterol ratio as a predictor of stroke risk: a longitudinal analysis of the China health and retirement longitudinal study. Front. Neurol.16, 1503743. 10.3389/fneur.2025.1503743 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wang, X. et al. miRNA-29 aggravates myocardial infarction via inhibiting the PI3K/mTOR/HIF1alpha/VEGF pathway. Aging (Albany NY). 14, 3129–3142. 10.18632/aging.203997 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Holly, S. P. et al. Ether lipid metabolism by AADACL1 regulates platelet function and thrombosis. Blood Adv.3, 3818–3828. 10.1182/bloodadvances.2018030767 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Gluba-Sagr, A. et al. High-density lipoproteins. Part 2. Impact of disease states on functionality. Am. J. Prev. Cardiol.23, 101073. 10.1016/j.ajpc.2025.101073 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Pilard, M., Babran, S. & Martel, C. Regulation of Platelet Function by HDL. Arterioscler. Thromb. Vasc Biol.45, e184–e200. 10.1161/ATVBAHA.124.318260 (2025). [DOI] [PubMed] [Google Scholar]
- 11.Jialal, I., Jialal, G. & Adams-Huet, B. The platelet to high density lipoprotein -cholesterol ratio is a valid biomarker of nascent metabolic syndrome. Diabetes Metab. Res. Rev.37, e3403. 10.1002/dmrr.3403 (2021). [DOI] [PubMed] [Google Scholar]
- 12.Wu, W. et al. Impact of Platelet-to-HDL-Cholesterol Ratio on Long-Term Mortality in Coronary Artery Disease Patients with or Without Type 2 Diabetes: Insights from a Chinese Multicenter Cohort. J. Inflamm. Res.17, 2731–2744. 10.2147/JIR.S458950 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Zhang, H., Xu, Y. & Xu, Y. The association of the platelet/high-density lipoprotein cholesterol ratio with self-reported stroke and cardiovascular mortality: a population-based observational study. Lipids Health Dis.23, 121. 10.1186/s12944-024-02115-y (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wang, B., Wang, J., Liu, C. & Hu, X. The potential of platelet to high-density lipoprotein cholesterol ratio (PHR) as a novel biomarker for heart failure. Sci. Rep.14, 23283. 10.1038/s41598-024-75453-7 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Potter, T. B. H., Tannous, J. & Vahidy, F. S. A Contemporary Review of Epidemiology, Risk Factors, Etiology, and Outcomes of Premature Stroke. Curr. Atheroscler Rep.24, 939–948. 10.1007/s11883-022-01067-x (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zhou, Q. et al. Air pollution, genetic factors, and chronic rhinosinusitis: A prospective study in the UK Biobank. Sci. Total Environ.940, 173526. 10.1016/j.scitotenv.2024.173526 (2024). [DOI] [PubMed] [Google Scholar]
- 17.Zhang, Y. et al. Healthy lifestyle counteracts the risk effect of genetic factors on incident gout: a large population-based longitudinal study. BMC Med.20, 138. 10.1186/s12916-022-02341-0 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Luo, P. et al. Air Pollution and Allergic Rhinitis: Findings from a Prospective Cohort Study. Environ. Sci. Technol.57, 15835–15845. 10.1021/acs.est.3c04527 (2023). [DOI] [PubMed] [Google Scholar]
- 19.Fan, M. et al. Sleep patterns, genetic susceptibility, and incident cardiovascular disease: a prospective study of 385 292 UK biobank participants. Eur. Heart J.41, 1182–1189. 10.1093/eurheartj/ehz849 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Arboix, A. et al. Ischemic stroke as first manifestation of essential thrombocythemia. Report of six cases. Stroke26, 1463–1466. 10.1161/01.str.26.8.1463 (1995). [DOI] [PubMed] [Google Scholar]
- 21.Stoll, G. & Nieswandt, B. Thrombo-inflammation in acute ischaemic stroke - implications for treatment. Nat. Rev. Neurol.15, 473–481. 10.1038/s41582-019-0221-1 (2019). [DOI] [PubMed] [Google Scholar]
- 22.Nieswandt, B., Pleines, I. & Bender, M. Platelet adhesion and activation mechanisms in arterial thrombosis and ischaemic stroke. J. Thromb. Haemost. 9 (Suppl 1), 92–104. 10.1111/j.1538-7836.2011.04361.x (2011). [DOI] [PubMed] [Google Scholar]
- 23.Denorme, F., Vanhoorelbeke, K. & De Meyer, S. F. von Willebrand Factor and Platelet Glycoprotein Ib: A Thromboinflammatory Axis in Stroke. Front. Immunol.10, 2884. 10.3389/fimmu.2019.02884 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Gao, Y. et al. Dual Antiplatelet Treatment up to 72 Hours after Ischemic Stroke. N Engl. J. Med.389, 2413–2424. 10.1056/NEJMoa2309137 (2023). [DOI] [PubMed] [Google Scholar]
- 25.Bhatia, K. et al. Dual Antiplatelet Therapy Versus Aspirin in Patients With Stroke or Transient Ischemic Attack: Meta-Analysis of Randomized Controlled Trials. Stroke52, e217–e223. 10.1161/STROKEAHA.120.033033 (2021). [DOI] [PubMed] [Google Scholar]
- 26.van der Bom, J. G. et al. Platelet count and the risk for thrombosis and death in the elderly. J. Thromb. Haemost. 7, 399–405. 10.1111/j.1538-7836.2008.03267.x (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Emerging, R. et al. Major lipids, apolipoproteins, and risk of vascular disease. JAMA302, 1993–2000. 10.1001/jama.2009.1619 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Woodward, M. et al. Associations between high-density lipoprotein cholesterol and both stroke and coronary heart disease in the Asia Pacific region. Eur. Heart J.28, 2653–2660. 10.1093/eurheartj/ehm427 (2007). [DOI] [PubMed] [Google Scholar]
- 29.Li, H. et al. U-Shaped Relationship of High-Density Lipoprotein Cholesterol and Incidence of Total, Ischemic and Hemorrhagic Stroke: A Prospective Cohort Study. Stroke53, 1624–1632. 10.1161/STROKEAHA.121.034393 (2022). [DOI] [PubMed] [Google Scholar]
- 30.Hou, X. et al. Association between platelet-to-high-density lipoprotein cholesterol ratio and future stroke risk: a national cohort study based on CHARLS. Front. Neurol.15, 1479245. 10.3389/fneur.2024.1479245 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Brill, A. et al. von Willebrand factor-mediated platelet adhesion is critical for deep vein thrombosis in mouse models. Blood117, 1400–1407. 10.1182/blood-2010-05-287623 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Sachais, B. S. et al. Elimination of platelet factor 4 (PF4) from platelets reduces atherosclerosis in C57Bl/6 and apoE-/- mice. Thromb. Haemost. 98, 1108–1113 (2007). [PubMed] [Google Scholar]
- 33.Burger, P. C. & Wagner, D. D. Platelet P-selectin facilitates atherosclerotic lesion development. Blood101, 2661–2666. 10.1182/blood-2002-07-2209 (2003). [DOI] [PubMed] [Google Scholar]
- 34.Blann, A. D., Nadar, S. K. & Lip, G. Y. The adhesion molecule P-selectin and cardiovascular disease. Eur. Heart J.24, 2166–2179. 10.1016/j.ehj.2003.08.021 (2003). [DOI] [PubMed] [Google Scholar]
- 35.Sugiura, T. et al. Serotonin in peripheral blood reflects oxidative stress and plays a crucial role in atherosclerosis: Novel insights toward holistic anti-atherothrombotic strategy. Atherosclerosis246, 157–160. 10.1016/j.atherosclerosis.2016.01.015 (2016). [DOI] [PubMed] [Google Scholar]
- 36.Tillie, R. et al. A Switch from Cell-Associated to Soluble PDGF-B Protects against Atherosclerosis, despite Driving Extramedullary Hematopoiesis. Cells10 10.3390/cells10071746 (2021). [DOI] [PMC free article] [PubMed]
- 37.Jia, C. et al. High-Density Lipoprotein Anti-Inflammatory Capacity and Incident Cardiovascular Events. Circulation143, 1935–1945. 10.1161/CIRCULATIONAHA.120.050808 (2021). [DOI] [PubMed] [Google Scholar]
- 38.Crudele, L. et al. Low HDL-cholesterol levels predict hepatocellular carcinoma development in individuals with liver fibrosis. JHEP Rep.5, 100627. 10.1016/j.jhepr.2022.100627 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Shi, A. et al. The Phenolic Components of Gastrodia elata improve Prognosis in Rats after Cerebral Ischemia/Reperfusion by Enhancing the Endogenous Antioxidant Mechanisms. Oxid Med Cell Longev 7642158, (2018). 10.1155/2018/7642158 (2018). [DOI] [PMC free article] [PubMed]
- 40.Chen, S., Zeng, L. & Hu, Z. Progressing haemorrhagic stroke: categories, causes, mechanisms and managements. J. Neurol.261, 2061–2078. 10.1007/s00415-014-7291-1 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The UK Biobank data are available at https://www.ukbiobank.ac.uk. This research has been conducted using the UK Biobank Resource under Application Number 552996.