2025 年 32 巻 11 号 p. 1425-1433
Aim: The aim of this study was to clarify the association between serum total cholesterol and fatal subarachnoid hemorrhage in Japanese men and women.
Methods: The study involved a pooled analysis of individual data from 12 well-qualified cohort studies conducted in Japan. The participants were classified according to their serum total cholesterol levels: ＜4.14 mmol/L (＜160 mg/dL), 4.14–4.64 mmol/L (160–180 mg/dL), 4.65–5.16 mmol/L (180–199 mg/dL), 5.17–5.68 mmol/L (200–219 mg/dL), 5.69–6.20 mmol/L (220–239 mg/dL), and ≥ 6.21 mmol/L (≥ 240 mg/dL). The outcome of the analysis was death from subarachnoid hemorrhage.
Results: A total of 120,973 participants (70,947 women and 50,026 men) aged 18–96 years at baseline underwent follow-up for a median of 12.7 years, and 261 participants died from subarachnoid hemorrhage during this period. In women, both low (＜5.69 mmol/L [＜220 mg/dL]) and high (≥ 6.21 mmol/L [≥ 240 mg/dL]) serum total cholesterol levels were significantly associated with a higher risk of fatal subarachnoid hemorrhage compared with the reference group (5.69–6.20 mmol/L [220–239 mg/dL]). These associations remained significant after adjustment for confounding factors. In contrast, no associations were observed in men.
Conclusion: Both low and high serum total cholesterol levels were associated with a higher risk of fatal subarachnoid hemorrhage in 70,947 female participants from 12 cohort studies throughout Japan.
The global incidence rate of subarachnoid hemorrhage is estimated at 9 per 100,000 person-years1). However, the rate of subarachnoid hemorrhage in Japan is much higher, at 20–100 per 100,000 person-years1, 2). Subarachnoid hemorrhage is the most lethal type of stroke in Japan3) as well as in other countries4, 5). Effective prevention measures for subarachnoid hemorrhage require strategies based on reliable knowledge of risk factors for its development. Low serum cholesterol levels have been shown to be associated with hemorrhagic stroke, including intracerebral hemorrhage and subarachnoid hemorrhage6-8). However, there is no definitive evidence for a significant association between serum cholesterol and subarachnoid hemorrhage9-12).
The aim of the present study was to clarify the association between serum total cholesterol and fatal subarachnoid hemorrhage using data from an integrated dataset based on 12 cohort studies conducted throughout Japan.
The Evidence for Cardiovascular Prevention from Observational Cohorts in Japan (EPOCH-JAPAN) study was a pooled analysis of data from 15 well-qualified cohort studies conducted in Japan. The details of the project were described previously13). Each cohort obtained written informed consent from the study participants and received ethical approval from the ethics committee of the relevant institute. For the present study, three cohorts in the EPOCH-JAPAN study were excluded due to a lack of data on deaths from subarachnoid hemorrhage. Consequently, 132,826 participants aged 18–96 years at baseline from 12 cohorts were included in the study. The baseline years were 1985 to 2002, and the median follow-up was 12.7 years. After exclusion of 11,853 participants with missing data on death from subarachnoid hemorrhage, serum total cholesterol, age, or sex, a total of 120,973 participants (70,947 women and 50,026 men) were included in the analysis.
MeasurementsBlood samples were collected and serum total cholesterol levels were measured using enzyme methods. Measurement of serum total cholesterol was standardized in five cohorts such that the values would be trackable to the Centers for Disease Control and Prevention reference method. The participants were divided into six groups according to their serum total cholesterol levels: ＜4.14 mmol/L (＜160 mg/dL), 4.14–4.64 mmol/L (160–180 mg/dL), 4.65–5.16 mmol/L (180–199 mg/dL), 5.17–5.68 mmol/L (200–219 mg/dL), 5.69–6.20 mmol/L (220–239 mg/dL), and ≥ 6.21 mmol/L (≥ 240 mg/dL). The reference group was defined as 5.69–6.20 mmol/L (220–239 mg/dL). Blood pressure was measured with a standard sphygmomanometer in all cohorts. In several cohorts where blood pressure was measured more than once, the average values were used for the analysis. Body mass index was calculated as weight in kilograms divided by height in meters squared (kg/m2). Smoking status and alcohol intake were measured by a questionnaire and defined as current or not.
OutcomeFor each deceased subject, the primary underlying cause of death was coded according to the International Classification of Diseases (ICD) for the National Vital Statistics based on the criteria proposed by the World Health Organization. Detailed causes of death were sought using all available sources in each cohort study. In many studies, death certificates were reviewed and/or the National Vital Statistics were utilized after obtaining permission from the Ministry of Internal Affairs and Communications of Japan. Other sources utilized in some studies included autopsy findings, medical records, health examinations, and questionnaires. The outcome of the present analysis was death from subarachnoid hemorrhage (430 in ICD-9 or I60 in ICD-10).
Statistical AnalysisFor the participant characteristics, continuous variables were presented as mean and standard deviation (SD), and categorical variables were shown as number and percentage. Multivariable-adjusted hazard ratios (HRs) for fatal subarachnoid hemorrhage defined by serum total cholesterol levels were calculated using Cox proportional hazards models adjusted for age (model 1) and BMI, blood pressure, smoking, and alcohol drinking (model 2) as covariates. A Fine and Gray model analysis was conducted to account for competing risk of death from any other cause than subarachnoid hemorrhage. The significance level was set at P＜0.05. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).
The characteristics of the participants in the 12 cohorts are shown in Table 1. The mean serum total cholesterol levels ranged from 4.90 to 5.44 mmol/L (189.3 to 210.2 mg/dL). The characteristics of the female participants according to their serum total cholesterol levels are shown in Table 2. Women with higher serum total cholesterol levels were older, more likely to be obese, and less likely to be current smokers. Similar patterns were observed in men (Table 3).
| Cohort | n | Baseline year |
Women n (%) |
Age, years mean (SD) |
Serum TC, mmol/L mean (SD) |
Follow-up, years median (IQR) |
SAH death N |
|---|---|---|---|---|---|---|---|
| Aichi, Workers | 5,589 | 1997 | 1,206 (21.6) | 48.1 (7.1) | 5.44 (0.91) | 10.9 (9.5–10.9) | 2 |
| CIRCS, Osaka | 6,180 | 1985 | 3,952 (63.9) | 55.1 (12.8) | 5.43 (0.95) | 13.5 (10.8–13.8) | 5 |
| Hisayama | 2,735 | 1988 | 1,573 (57.5) | 59.5 (12.1) | 5.38 (1.09) | 14.0 (14.0–14.0) | 16 |
| Iwate | 26,455 | 2002 | 17,300 (65.4) | 62.1 (11.6) | 5.18 (0.86) | 2.2 (2.0–4.1) | 13 |
| JACC | 29,008 | 1988 | 18,460 (63.6) | 57.4 (9.6) | 5.13 (0.95) | 18.5 (17.1–19.5) | 107 |
| JMS | 12,236 | 1992 | 7,437 (60.8) | 55.3 (11.6) | 4.97 (0.91) | 11.1 (10.3–12.0) | 23 |
| Ohasama | 2,712 | 1987 | 1,636 (60.3) | 56.2 (12.2) | 5.06 (0.95) | 14.7 (11.8–14.7) | 12 |
| Osaki | 16,044 | 1994 | 9,232 (57.5) | 62.4 (9.4) | 5.29 (0.91) | 12.7 (12.6–12.8) | 44 |
| RERF | 4,610 | 1986 | 3,105 (67.4) | 62.1 (12.3) | 5.44 (1.03) | 19.9 (12.2–20.6) | 11 |
| Suita | 6,445 | 1989 | 3,355 (52.1) | 55.4 (13.3) | 5.36 (0.96) | 14.9 (13.5–15.9) | 11 |
| Tanno-Sobetsu | 1,972 | 1977 | 1,072 (54.4) | 49.4 (8.1) | 4.90 (0.97) | 27.0 (17.1–29.4) | 13 |
| Toyama, Workers | 6,987 | 1990 | 2,619 (37.5) | 37.9 (10.0) | 4.93 (0.92) | 20.1 (19.2–20.7) | 4 |
TC: total cholesterol; SAH: subarachnoid hemorrhage; SD: standard deviation; IQR: interquartile range; CIRCS: Circulatory Risk in Communities Study; JACC: Japan Collaborative Cohort Study; JMS: Jichi Medical School; RERF: Radiation Effects Research Foundation.
| Serum total cholesterol, mmol/L |
P value |
||||||
|---|---|---|---|---|---|---|---|
|
＜4.14 (＜160 mg/dL) |
4.14–4.64 (160–179 mg/dL) |
4.65–5.16 (180–199 mg/dL) |
5.17–5.68 (200–219 mg/dL) |
5.69–6.20 (220–239 mg/dL) |
≥ 6.21 (≥ 240 mg/dL) |
||
| Number of participants | 6,496 | 10,234 | 15,118 | 15,526 | 11,676 | 11,717 | |
| Age, years | 50.7 (14.1) | 54.0 (13.2) | 56.8 (12.3) | 58.9 (11.0) | 60.1 (10.2) | 60.5 (9.4) | ＜0.001 |
| BMI, kg/m2 | 22.4 (3.2) | 22.9 (3.3) | 23.3 (3.4) | 23.6 (3.3) | 23.8 (3.3) | 24.1 (3.4) | ＜0.001 |
| History of CVD, n (%) | 234 (3.7) | 379 (3.7) | 673 (4.5) | 666 (4.3) | 573 (4.9) | 632 (5.4) | ＜0.001 |
| Smoking, n (%) | |||||||
| Current smokers | 395 (6.4) | 541 (5.6) | 672 (4.7) | 600 (4.1) | 503 (4.6) | 531 (4.8) | ＜0.001 |
| Previous smokers | 154 (2.5) | 192 (2.0) | 252 (1.8) | 238 (1.6) | 189 (1.7) | 233 (2.1) | |
| Alcohol drinking, n (%) | |||||||
| Current drinkers | 1,515 (24.9) | 2,322 (24.0) | 3,187 (22.2) | 3,093 (21.1) | 2,228 (20.3) | 2,135 (19.3) | ＜0.001 |
| Previous drinkers | 92 (1.5) | 147 (1.5) | 195 (1.4) | 244 (1.7) | 183 (1.7) | 181 (1.6) | |
| Total cholesterol, mmol/L | 3.78 (0.30) | 4.41 (0.15) | 4.91 (0.15) | 5.41 (0.15) | 5.92 (0.15) | 6.78 (0.57) | ＜0.001 |
| Systolic BP, mmHg | 121.7 (19.4) | 124.9 (19.6) | 127.3 (19.8) | 129.5 (20.1) | 131.2 (20.4) | 133.6 (20.3) | ＜0.001 |
| Diastolic BP, mmHg | 72.1 (11.5) | 74.2 (11.5) | 75.6 (11.4) | 76.8 (11.4) | 77.6 (11.6) | 79.1 (11.4) | ＜0.001 |
BMI: body mass index; CVD: cardiovascular disease; BP: blood pressure.
| Serum total cholesterol (mmol/L) |
P value |
||||||
|---|---|---|---|---|---|---|---|
|
＜4.14 (＜160 mg/dL) |
4.14–4.64 (160–179 mg/dL) |
4.65–5.16 (180–199 mg/dL) |
5.17–5.68 (200–219 mg/dL) |
5.69–6.20 (220–239 mg/dL) |
≥ 6.21 (≥ 240 mg/dL) |
||
| Number of participants | 8,273 | 9,925 | 11,476 | 9,666 | 5,932 | 4,673 | |
| Age, years | 57.1 (14.5) | 56.8 (13.5) | 57.1 (12.8) | 56.7 (12.3) | 56.7 (11.9) | 55.2 (11.4) | ＜0.001 |
| BMI, kg/m2 | 22.3 (2.9) | 22.8 (2.9) | 23.2 (2.9) | 23.5 (2.9) | 23.8 (2.9) | 24.0 (2.9) | ＜0.001 |
| History of CVD, n (%) | 518 (6.3) | 575 (5.8) | 668 (5.9) | 513 (5.4) | 314 (5.3) | 253 (5.4) | 0.054 |
| Smoking, n (%) | |||||||
| Current smokers | 4,444 (55.6) | 4,800 (49.8) | 5,124 (45.8) | 4,105 (43.6) | 2,410 (41.7) | 1,969 (43.2) | ＜0.001 |
| Previous smokers | 1,697 (21.2) | 2,256 (23.4) | 3,020 (27.0) | 2,735 (29.1) | 1,736 (30.0) | 1,329 (29.2) | |
| Alcohol drinking, n (%) | |||||||
| Current drinkers | 5,947 (74.7) | 7,191 (74.9) | 8,316 (74.5) | 6,968 (74.3) | 4,237 (73.4) | 3,333 (73.2) | ＜0.001 |
| Previous drinkers | 500 (6.3) | 486 (5.1) | 546 (4.9) | 486 (5.2) | 305 (5.3) | 222 (4.9) | |
| Total cholesterol, mmol/L | 3.72 (0.34) | 4.40 (0.15) | 4.90 (0.15) | 5.40 (0.15) | 5.91 (0.15) | 6.74 (0.60) | ＜0.001 |
| Systolic BP, mmHg | 130.6 (19.6) | 130.8 (19.3) | 131.3 (19.1) | 131.7 (18.8) | 132.4 (18.8) | 133.1 (18.8) | ＜0.001 |
| Diastolic BP, mmHg | 77.5 (11.8) | 78.5 (11.7) | 79.3 (11.6) | 80.1 (11.4) | 80.7 (11.6) | 81.7 (11.5) | ＜0.001 |
BMI: body mass index; CVD: cardiovascular disease; BP: blood pressure.
During the median follow-up of 12.7 years, 261 participants died from subarachnoid hemorrhage. The associations between serum total cholesterol and fatal subarachnoid hemorrhage in women and men are shown in Table 4. In women, both high and low serum total cholesterol levels were significantly associated with a higher risk of fatal subarachnoid hemorrhage compared with the reference group. These associations remained significant in the multivariable analysis with adjustment for body mass index, systolic blood pressure, smoking, and alcohol drinking compared with the reference group. Comparable findings were observed in the Fine and Gray model accounting for competing risk of death from any other cause than subarachnoid hemorrhage; the HR (95% confidence interval) values were 2.73 (1.34–5.55) for serum total cholesterol ＜4.14 mmol/L (＜160 mg/dL), 1.97 (1.01–3.83) for serum total cholesterol 4.14–4.64 mmol/L (160–180 mg/dL), 2.01 (1.09–3.70) for serum total cholesterol 4.65–5.16 mmol/L (180–199 mg/dL), 1.95 (1.06–3.56) for serum total cholesterol 5.17 to 5.68 mmol/L (200–219 mg/dL), and 2.17 (1.18–3.97) for serum total cholesterol ≥ 6.21 mmol/L (≥ 240 mg/dL) compared with the reference group. In contrast, there were no significant associations between serum total cholesterol and fatal subarachnoid hemorrhage in men. We also conducted analyses in women stratified by age (65 and 55 years), as shown in Supplemental Tables 1 and 2. Significant positive associations were found for the lowest serum total cholesterol group in women aged ＜65 years and the highest serum total cholesterol group in women aged ≥ 65 years. In women aged ≥ 55 years, all serum total cholesterol groups had a significantly higher risk of fatal subarachnoid hemorrhage compared with the reference group, while no association was found in women aged ＜55 years. Similar associations were found in the analysis excluding participants with history of cardiovascular disease (Supplemental Table 3).
| Serum total cholesterol, mmol/L | ||||||
|---|---|---|---|---|---|---|
|
＜4.14 (＜160 mg/dL) |
4.14–4.64 (160–179 mg/dL) |
4.65–5.16 (180–199 mg/dL) |
5.17–5.68 (200–219 mg/dL) |
5.69–6.20 (220–239 mg/dL) |
≥ 6.21 (≥ 240 mg/dL) |
|
| Women | ||||||
| Number of participants | 6,496 | 10,234 | 15,118 | 15,526 | 11,676 | 11,717 |
| Number of SAH deaths | 17 | 22 | 41 | 40 | 19 | 41 |
| Hazard ratio (95% CI) | ||||||
| Model 1 | 2.31 (1.19–4.47) | 1.64 (0.88–3.03) | 1.91 (1.11–3.29) | 1.68 (0.98–2.92) | Reference | 2.00 (1.16–3.45) |
| Model 2 | 3.05 (1.49–6.25) | 2.07 (1.06–4.03) | 2.08 (1.13–3.82) | 1.96 (1.07–3.59) | Reference | 2.14 (1.17–3.92) |
| Men | ||||||
| Number of participants | 8,273 | 9,925 | 11,476 | 9,666 | 5,932 | 4,673 |
| Number of SAH deaths | 24 | 19 | 14 | 14 | 7 | 3 |
| Hazard ratio (95% CI) | ||||||
| Model 1 | 2.17 (0.93–5.08) | 1.50 (0.63–3.59) | 0.98 (0.39–2.43) | 1.19 (0.48–2.95) | Reference | 0.53 (0.14–2.06) |
| Model 2 | 1.65 (0.69–3.93) | 1.20 (0.49–2.91) | 0.83 (0.33–2.09) | 1.16 (0.47–2.89) | Reference | 0.54 (0.14–2.10) |
SAH: subarachnoid hemorrhage; CI: confidence interval.
Model 1: adjusted for age.
Model 2: adjusted for age, body mass index, systolic blood pressure, smoking, and alcohol drinking.
| Serum total cholesterol, mmol/L | ||||||
|---|---|---|---|---|---|---|
|
＜4.14 (＜160 mg/dL) |
4.14–4.64 (160–179 mg/dL) |
4.65–5.16 (180–199 mg/dL) |
5.17–5.68 (200–219 mg/dL) |
5.69–6.20 (220–239 mg/dL) |
≥ 6.21 (≥ 240 mg/dL) |
|
| Women (＜65 years) | ||||||
| Number of participants | 5,244 | 7,783 | 10,722 | 10,429 | 7,597 | 7,767 |
| Number of SAH deaths | 13 | 10 | 22 | 16 | 10 | 19 |
| Hazard ratio (95% CI) | ||||||
| Model 1 | 2.80 (1.21–6.48) | 1.22 (0.50–2.95) | 1.81 (0.86–3.84) | 1.23 (0.56–2.71) | Reference | 1.74 (0.81–3.75) |
| Model 2 | 3.87 (1.55–9.67) | 1.49 (0.57–3.88) | 2.15 (0.94–4.94) | 1.42 (0.59–3.39) | Reference | 1.76 (0.75–4.11) |
| Women (≥ 65 years) | ||||||
| Number of participants | 1,252 | 2,451 | 4,396 | 5,097 | 4,079 | 3,950 |
| Number of SAH deaths | 4 | 12 | 19 | 24 | 9 | 22 |
| Hazard ratio (95% CI) | ||||||
| Model 1 | 1.54 (0.47–5.01) | 2.29 (0.96–5.46) | 2.05 (0.93–4.54) | 2.25 (1.05–4.85) | Reference | 2.26 (1.04–4.92) |
| Model 2 | 2.01 (0.58–6.91) | 2.96 (1.16–7.58) | 2.05 (0.83–5.03) | 2.61 (1.11–6.12) | Reference | 2.57 (1.08–6.09) |
SAH: subarachnoid hemorrhage; 95% CI: 95% confidence interval.
Model 1: adjusted for age.
Model 2: adjusted for age, body mass index, systolic blood pressure, smoking, and alcohol drinking.
| Serum total cholesterol, mmol/L | ||||||
|---|---|---|---|---|---|---|
|
＜4.14 (＜160 mg/dL) |
4.14–4.64 (160–179 mg/dL) |
4.65–5.16 (180–199 mg/dL) |
5.17–5.68 (200–219 mg/dL) |
5.69–6.20 (220–239 mg/dL) |
≥ 6.21 (≥ 240 mg/dL) |
|
| Women (＜55 years) | ||||||
| Number of participants | 4,052 | 5,212 | 6,172 | 5,077 | 3,228 | 2,993 |
| Number of SAH deaths | 7 | 3 | 5 | 3 | 3 | 8 |
| Hazard ratio (95% CI) | ||||||
| Model 1 | 2.724 (0.682–10.882) | 0.799 (0.159–4.010) | 1.126 (0.267–4.748) | 0.690 (0.139–3.425) | Reference | 2.800 (0.740–10.598) |
| Model 2 | 4.544 (0.865–23.879) | 0.958 (0.132–6.967) | 1.643 (0.294–9.178) | 1.141 (0.189–6.888) | Reference | 3.782 (0.774–18.494) |
| Women (≥ 55 years) | ||||||
| Number of participants | 2,444 | 5,022 | 8,946 | 10,449 | 8,448 | 8,724 |
| Number of SAH deaths | 10 | 19 | 36 | 37 | 16 | 33 |
| Hazard ratio (95% CI) | ||||||
| Model 1 | 2.095 (0.949–4.629) | 1.954 (1.003–3.808) | 2.140 (1.186–3.861) | 1.917 (1.066–3.448) | Reference | 1.866 (1.026–3.393) |
| Model 2 | 2.724 (1.188–6.246) | 2.434 (1.197–4.948) | 2.229 (1.160–4.282) | 2.134 (1.122–4.057) | Reference | 1.949 (1.012–3.754) |
SAH: subarachnoid hemorrhage; CI: confidence interval.
Model 1: adjusted for age.
Model 2: adjusted for age, body mass index, systolic blood pressure, smoking, and alcohol drinking.
| Serum total cholesterol, mmol/L | ||||||
|---|---|---|---|---|---|---|
|
＜4.14 (＜160 mg/dL) |
4.14–4.64 (160–179 mg/dL) |
4.65–5.16 (180–199 mg/dL) |
5.17–5.68 (200–219 mg/dL) |
5.69–6.20 (220–239 mg/dL) |
≥ 6.21 (≥ 240 mg/dL) |
|
| Women | ||||||
| Number of participants | 6,266 | 9,858 | 14,456 | 14,867 | 11,106 | 11,090 |
| Number of SAH deaths | 13 | 19 | 30 | 33 | 16 | 36 |
| Hazard ratio (95% CI) | ||||||
| Model 1 | 2.181 (1.044–4.557) | 1.737 (0.891–3.387) | 1.691 (0.921–3.106) | 1.667 (0.917–3.029) | Reference | 2.102 (1.165–3.792) |
| Model 2 | 2.861 (1.314–6.227) | 2.120 (1.033–4.351) | 1.837 (0.942–3.583) | 1.829 (0.950–3.522) | Reference | 2.133 (1.115–4.082) |
| Men | ||||||
| Number of participants | 7,756 | 9,353 | 10,809 | 9,154 | 5,618 | 4,420 |
| Number of SAH deaths | 23 | 16 | 13 | 13 | 7 | 3 |
| Hazard ratio (95% CI) | ||||||
| Model 1 | 2.118 (0.901–4.979) | 1.280 (0.524–3.126) | 0.922 (0.367–2.317) | 1.110 (0.443–2.784) | Reference | 0.535 (0.138–2.071) |
| Model 2 | 1.603 (0.668–3.848) | 1.066 (0.431–2.640) | 0.774 (0.303–1.975) | 1.085 (0.433– 2.724) | Reference | 0.542 (0.140–2.102) |
SAH: subarachnoid hemorrhage; CI: confidence interval.
Model 1: adjusted for age.
Model 2: adjusted for age, body mass index, systolic blood pressure, smoking, and alcohol drinking.
In the present pooled analysis using data from the integrated EPOCH-JAPAN dataset covering 12 cohort studies throughout Japan, both high (≥ 6.21 mmol/L [≥ 240 mg/dL]) and low (＜5.69 mmol/L [＜220 mg/dL]) serum total cholesterol levels in women were associated with a higher risk of fatal subarachnoid hemorrhage compared with the reference group (5.69–6.20 mmol/L [220–239 mg/dL]). These associations remained significant after adjustment for confounding factors such as body mass index, systolic blood pressure, smoking, and alcohol drinking. Low cholesterol was associated with the risk of fatal subarachnoid hemorrhage in women aged ≤ 65 years, whereas high cholesterol was associated with the risk of fatal subarachnoid hemorrhage in women aged ＞65 years. In contrast, no associations between serum total cholesterol and fatal subarachnoid hemorrhage were noted in men, although a non-significant trend toward an increased risk of fatal subarachnoid hemorrhage was observed for very low serum total cholesterol levels (＜4.14 mmol/L [160 mg/dL]).
Several prospective studies have investigated the association between serum total cholesterol and incidence or mortality of subarachnoid hemorrhage, but no studies have examined the association between serum total cholesterol and risk of subarachnoid hemorrhage separately for men and women. The Honolulu Heart Program found no clear association between serum total cholesterol and incidence of subarachnoid hemorrhage in Japanese American populations9), while the JACC study found no association between serum total cholesterol and mortality from subarachnoid hemorrhage in Japanese subjects11). Furthermore, the Asia-Pacific Cohort Studies Collaboration, a meta-analysis of data for participants in prospective cohort studies carried out in the Asia-Pacific region, revealed no clear association between serum total cholesterol and fatal/nonfatal subarachnoid hemorrhage10). However, a stroke registry study conducted in Akita Prefecture, Japan, suggested an increased incidence of subarachnoid hemorrhage in Japanese people with low serum total cholesterol levels (＜4.14 mmol/L [160 mg/dL])2). A recent systematic review of prospective studies, including the Honolulu Heart Program and JACC study, found that low total serum cholesterol levels were associated with an increased risk of subarachnoid hemorrhage in East Asian populations, while no such associations were observed in non-East Asian populations14). In the present study, low serum total cholesterol levels (＜5.69 mmol/L [220 mg/dL]) were associated with an increased risk of fatal subarachnoid hemorrhage in women. In addition, very low serum total cholesterol levels (＜4.14 mmol/L [160 mg/dL]) showed a tendency to be associated with an increased risk of fatal subarachnoid hemorrhage in men. Based on the above findings, low serum total cholesterol levels have the potential to be associated with an increased risk of subarachnoid hemorrhage in East Asian people.
In this pooled analysis of data from prospective studies, high serum total cholesterol levels (≥ 6.21 mmol/L [≥ 240 mg/dL]) were associated with increased mortality from subarachnoid hemorrhage in women aged ≥ 65 years, but no such association was observed in men. Most previous observational studies2, 9-11) also found no increased risk of subarachnoid hemorrhage associated with high serum total cholesterol levels. A meta-analysis revealed that postmenopausal state was associated with increased serum total cholesterol levels and higher risk of subarachnoid hemorrhage15). However, data on postmenopausal status were not available for the women evaluated in the present study. Since most Japanese women aged ≥ 55 years would be postmenopausal16), we conducted an analysis in women stratified by age of 55 years, and found a stronger association in women aged ≥ 55 years compared with women aged ＜55 years. These findings were consistent with those in a previous study15). Further investigations are required to confirm whether there is a causal relationship between high serum total cholesterol and increased risk of fatal subarachnoid hemorrhage.
The FINRISK study reported that higher TC and LDL-C levels were associated with increased SAH risk, particularly in men17). These findings contrast with our study, which observed no significant relationship between TC and SAH risk in men. Differences in study design, and population characteristics may account for these discrepancies. As shown in Table 3 of our study, the group with the highest TC levels had a smoking rate approximately 12% lower and was about 2 years younger than the group with the lowest TC levels. Although we adjusted for these variables in the analysis, individuals in the highest TC category may have been relatively healthier, contributing to the observed lower SAH risk. Further research is needed to clarify how lipid profiles influence SAH risk across different populations.
The mechanisms underlying the association between low serum total cholesterol and fatal subarachnoid hemorrhage, especially in women aged ＜65 years, remain unclear. A possible mechanism involves the promotion of arterial medial layer smooth muscle cell necrosis associated low serum total cholesterol18). Low serum total cholesterol was also shown to be associated with impaired endothelium in arteries19). Endothelium and smooth muscle cell impairment in the walls of cerebral artery aneurysms may promote rupture of the aneurysms and subsequent hemorrhage in the subarachnoid space.
The reason why this association was found only in women remains unclear. However, several possible mechanisms may explain this association. Studies have shown that arterial stiffness progresses differently in men and women, with postmenopausal women experiencing a rapid increase in arterial stiffness due to estrogen loss20). This hormonal shift may alter vascular integrity, making arteries more susceptible to the effects of both low and high cholesterol. Additionally, differences in vascular structure and lipid metabolism between sexes could contribute to this U-shaped relationship. Further research is needed to fully elucidate these mechanisms.
The strengths of the present study are the very large sample size of 120,973 participants with a single ethnicity (Japanese) based on a pooled analysis of data from prospective studies and the long-term follow-up period. Several limitations of the study should also be discussed. First, because the pooled analysis included participants of community health examinations, who were more likely to be health-conscious, the findings may have been affected by selection bias. Second, the non-standardized measurement methods for serum total cholesterol levels in some cohorts may also have resulted in misclassification of exposure. Some cohorts started earlier than the tradition for collecting data on treatment for dyslipidemia. Thus, medication for dyslipidemia was not considered in the study. Third, the blood samples were obtained during health check-ups conducted as part of the cohort studies. We did not consider fasting status, because it was recorded in only a few cohorts. Finally, medication data, including antiplatelet or anticoagulant use, which might affect the incidence of SAH, were not available for all cohorts.
In conclusion, both low and high serum total cholesterol levels were associated with a higher risk of fatal subarachnoid hemorrhage in 70,947 women from 12 cohort studies conducted throughout Japan.
The Evidence for Cardiovascular Prevention from Observational Cohorts in Japan (EPOCH–JAPAN) Research Group is composed of the following investigators. Co-Chairpersons: Hirotsugu Ueshima (Shiga University of Medical Science), Tomonori Okamura (Keio University School of Medicine), and Yoshitaka Murakami (Toho University). Executive Committee: Hirotsugu Ueshima (Shiga University of Medical Science), Yutaka Kiyohara (Hisayama Research Institute for Lifestyle Diseases), Toshiharu Ninomiya (Kyushu University Graduate School of Medicine), Yutaka Imai (Tohoku Institute for Management of Blood Pressure), Takayoshi Ohkubo (Teikyo University School of Medicine), Hiroyasu Iso (The National Center for Global Health and Medicine), Isao Muraki (Osaka University Graduate School of Medicine), Kazumasa Yamagishi (University of Tsukuba Institute of Medicine), Akiko Tamakoshi (Hokkaido University Faculty of Medicine), Yoshihiro Miyamoto (National Cerebral and Cardiovascular Center), Yoshihiro Kokubo (National Cerebral and Cardiovascular Center), Katsuyuki Miura (Shiga University of Medical Science), Sachiko Tanaka-Mizuno (Kyoto University Graduate School of Medicine and Public Health), Akiko Harada (Shiga University of Medical Science), Shigeyuki Saitoh (Sapporo Medical University), Hirofumi Ohnishi (Sapporo Medical University), Ichiro Tsuji (Tohoku University Graduate School of Medicine), Atsushi Hozawa (Tohoku University), Hideaki Nakagawa (Kanazawa Medical University), Masaru Sakurai (Kanazawa Medical University), Michiko Yamada (Radiation Effects Research Foundation), Yoshimi Tatsukawa (Radiation Effects Research Foundation), Kiyomi Sakata (Iwate Medical University), Kozo Tanno (Iwate Medical University), Akihiko Kitamura (Yao City Public Health Center), Masahiko Kiyama (Institute of Preventive Medicine Inc., Osaka Center for Cancer and Cardiovascular Disease Prevention), Yuji Shimizu (Osaka Institute of Public Health), Akira Okayama (Research Institute of Strategy for Prevention), Shizukiyo Ishikawa (Jichi Medical University), Hiroshi Yatsuya (Nagoya University Graduate School of Medicine), Takeo Nakayama (Kyoto University School of Public Health), Fujiko Irie (Ibaraki Prefecture), and Toshimi Sairenchi (Dokkyo Medical University).
This research was supported by a grant-in-aid from the Ministry of Health, Labour and Welfare, and Health and Labour Sciences research grants, Japan (Research on Health Services: H17–Kenkou–007; Comprehensive Research on Cardiovascular Disease and Life-Related Disease: H18-Junkankitou [Seishuu]-Ippan-012; H19–Junkankitou [Seishuu]-Ippan-012; H20-Junkankitou [Seishuu]-Ippan-013; H23-Junkankitou [Seishuu]-Ippan-005; H26-Junkankitou [Seisaku]-Ippan-001; H29-Junkankitou-Ippan-003; 20FA1002, and 23FA0501).
HA has received research grants from Daiichi Sankyo and Takeda, lecture fees from Bayer, Daiichi Sankyo, Fukuda Denshi, MSD, Takeda, and Teijin, and consultancy fees from Kyowa Kirin. The other authors have no other conflicts of interest to disclose.
This study was registered as a preprint on medRxiv prior to journal submission. The preprint version is accessible at https://www.medrxiv.org/content/10.1101/2024.10.15.24315569v1.full.pdf. No substantive changes have been made to the data or conclusions between the preprint and the current manuscript.
This work has not been presented previously.
