The authors have declared that no competing interests exist.
Conceived and designed the experiments: YS TW T. Konta YU T. Kato T. Kayama IK. Performed the experiments: T. Kimura TN. Analyzed the data: HN YS SI AI KY SA MS YA KN. Contributed reagents/materials/analysis tools: IK. Wrote the manuscript: HN YS.
Accumulating evidence suggests the involvement of an autoimmune mechanism in the pathogenesis of respiratory dysfunction. The aim of this study was to investigate the relationship between pulmonary function and serum antibodies to several connective tissue disease autoantigens (ACTDA) levels, which has not been investigated in a general population.
Blood sampling and spirometry were performed for subjects (n = 3,257) aged ≥40 years who participated in a community-based annual health check in Takahata, Japan, from 2004 to 2006. ACTDA was measured by enzyme immunoassay, and subjects with ACTDA values ≥20 were defined as positive.
In males, there were significant inverse relationships between logarithmically transformed ACTDA values and spirometric parameters, including % predicted values for forced expiratory volume in 1 s (FEV1) and maximal midexpiratory flow (MMF) as well as FEV1/forced vital capacity (FVC). Multiple linear regression analysis revealed that except for the relationship between ACTDA and FEV1/FVC, these relationships were still significant after adjustment for Brinkman index (a measure of inhaled cigarette consumption). The prevalence of positive ACTDA was greater in male never-smokers with mixed ventilation disorders and relatively severe airflow obstruction (% predicted FEV1 below the median value).
Autoimmunity may be involved in the mechanism of impaired pulmonary function in the general population.
Inhalation of cigarette smoke causes respiratory inflammation even in healthy smokers, and long-term smoking causes various respiratory diseases. Of these diseases, chronic obstructive pulmonary disease (COPD) has had the most impact on public health worldwide [
In addition, autoimmunity is associated with the pathogenesis of pulmonary diseases such as COPD [
The prevalence of antinuclear antibody (ANA) in the general population is 13.8% in the United States and 26% in Japan [
It has been reported that there is no relationship between cigarette smoking and ANA positivity [
To date, the association of antinuclear antibodies with spirometric values in the general population has not been elucidated. In this study, we cross-sectionally and longitudinally investigated the relationship between ACTDA values and spirometric parameters in a general Japanese population.
This study was part of the Molecular Epidemiological Study utilizing the Regional Characteristics of 21st Century Centers of Excellence (COE) Program and the Global COE Program in Japan [
This study was based on an annual community health check, in which all residents of Takahata (in northern Japan) ≥40 years of age were invited to participate. From 2004 through 2006 (visit 1), 1,579 men and 1,941 women (a total of 3,520 subjects) were enrolled in the study and underwent initial spirometry. Two hundred sixty-three subjects were excluded from the analysis because spirometry data did not meet the criteria described below; data for 3,257 subjects (1,502 males and 1,755 females) were entered into the final statistical analysis. Subjects used a self-report questionnaire to document their medical histories, smoking habits, current use of medications, and clinical symptoms. One hundred forty-seven of the 542 male current smokers at visit 1 underwent subsequent follow-up spirometry in 2009 (visit 2) [
Fasting blood samples were obtained from an antecubital vein and immediately transferred to chilled tubes. The presence of ACTDA in serum was determined using an enzyme immunoassay (EIA) method (Medical & Biological Laboratories Co. Ltd., Nagoya, Japan). We used the MESACUP ANA EIA kit, that utilized a mixture of purified recombinant or natural antigens, including single- and double-stranded DNA, RNP, Sm, SS-A/Ro, SS-B/La, centromere, topoisomerase I, and Jo-1 antigens. Because EIA detects antibodies to a limited number of autoantigens, the value obtained using this kit is not identical to an ANA titer measured by indirect immunofluorescence assay (IIF); thus, we applied the term
Spirometric parameters of forced vital capacity (FVC) and FEV1 were measured using standard techniques, with subjects performing FVC maneuvers on a CHESTAC-25 part II EX instrument (Chest Corp., Tokyo, Japan) in accordance with the guidelines of the Japanese Respiratory Society (JRS) [
For continuous variables, data are presented as the mean (standard deviation [SD]). Student’s t-test for parametric data and the Mann-Whitney U test for non-parametric data were used to analyze the differences between 2 groups. Analysis of variance followed by Tukey’s test was performed for multivariate comparisons. Univariate regression analysis was used to examine the association between log ACTDA levels and each spirometric measure considered in this study. Multiple linear regression analysis was then performed to determine whether log ACTDA levels contributed to each of these spirometric measures after adjustment for all other variables included in the model. Differences in proportions were evaluated using the chi-square test. Statistical significance was inferred for two-sided
Of the subjects who underwent spirometry on visit 1, 191 males and 372 females were positive for ACTDA. The distribution of ACTDA measurements was skewed towards higher values (25% quartile, 7.2; median, 11.1; 75% quartile, 17.2). Therefore, ACTDA values were logarithmically transformed before analysis. Logarithmically-transformed ACTDA values were significantly associated with age in both males (coefficient, 5.75;
62.4 (10.4) | 65.1 (10.2)* | 67.8 (10.7) | 0.0007 | |
432.2 (489.6) | 481.0 (462.8) | 655.9 (932.3) | 0.11 | |
66 | 70.5 | 72.2 | 0.42 | |
97.5 (14.7) | 96.9 (15.9) |
87.7 (16.6)* | 0.02 | |
95.9 (17.4) | 94.7 (16.8) |
81.2 (22.2)* | 0.002 | |
77.1 (8.9) | 76.6 (8.6) |
71.3 (11.5) * | 0.02 | |
92.1 (37.6) | 89.0 (34.6) | 68.6 (32.0)* | 0.02 |
Of 3,257 subjects who underwent spirometry, ACTDA values were not available for 10 subjects and Brinkman index was not available for 335 subjects because of the lack of precise information about cigarette smoking.
This table shows characteristics of all male subjects according to serum levels of ACTDA.
Values are means (SD) or percentage.
*,
Abbreviations: BI, Brinkman index; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
|
||||
---|---|---|---|---|
54.5 (6.6) | 54.9 (6.3) | 54.4 (9.4) | 0.89 | |
443.7 (440.4) | 466.7 (452.4) | 327.0 (338.2) | 0.8 | |
68.1 | 70.4 | 60 | 0.85 | |
97.9 (12.5) | 97.0 (12.7) | 96.8 (11.6) | 0.83 | |
97.2 (14.7) | 96.5 (12.9) | 93.4 (15.8) | 0.78 | |
78.7 (7.4) | 79.1 (7.3) | 76.2 (4.8) | 0.68 | |
96.2 (33.2) | 95.5 (31.9) | 80.5 (25.9) | 0.56 |
Of 3,257 subjects who underwent spirometry, ACTDA values were not available for 10 subjects and Brinkman index was not available for 335 subjects because of the lack of precise information about cigarette smoking.
This table shows characteristics of male subjects aged younger than 65 years according to serum levels of ACTDA.
Values are means (SD) or percentage.
Abbreviations: BI, Brinkman index; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
71.6 (5.0) | 72.3 (4.9) | 73.0 (5.2) | 0.34 | |
420.9 (512.2) | 491.1 (472.6) | 792.9 (1073.9)* | 0.03 | |
63.4 | 70.6 | 76.9 | 0.24 | |
97.1 (16.8) | 96.8 (17.9) |
84.1 (17.3)* | 0.03 | |
94.3 (20.1) | 93.5 (19.1) |
76.5 (22.9)* | 0.006 | |
75.2 (10.0) | 74.8 (9.1) | 69.4 (12.9) | 0.11 | |
87.3 (41.5) | 84.5 (35.8) | 64.1 (33.8) | 0.11 |
Of 3,257 subjects who underwent spirometry, ACTDA values were not available for 10 subjects and Brinkman index was not available for 335 subjects because of the lack of precise information about cigarette smoking.
This table shows characteristics of male subjects aged 65 years and older according to serum levels of ACTDA.
Values are means (SD) or percentage.
*,
Abbreviations: BI, Brinkman index; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
61.4 (10.3) | 62.7 (10.1) | 65.4 (8.5) |
0.003 | |
18.1 (91.9) | 14.9 (84.9) | 11.6 (62.5) | 0.75 | |
9.5 | 9 | 5.1 | 0.51 | |
100.1 (14.2) | 99.1 (14.3) | 97.4 (17.4) | 0.21 | |
100.0 (15.2) | 99.1 (15.8) | 97.0 (18.2) | 0.24 | |
80.0 (6.4) | 79.8 (6.7) | 79.4 (8.0) | 0.73 | |
100.0 (32.8) | 99.9 (37.1) | 95.3 (35.5) | 0.58 |
Of 3,257 subjects who underwent spirometry, ACTDA values were not available for 10 subjects and Brinkman index was not available for 335 subjects because of the lack of precise information about cigarette smoking.
This table shows characteristics of all female subjects according to serum levels of ACTDA.
Values are means (SD) or percentage.
,
Abbreviations: BI, Brinkman index; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
54.2 (6.6) | 54.6 (6.8) | 57.3 (4.1) | 0.05 | |
25.5 (103.1) | 21.2 (93.8) | 26.0 (92.3) | 0.89 | |
13.4 | 14.5 | 7.7 | 0.64 | |
100.7 (13.1) | 98.4 (13.0) | 99.4 (13.9) | 0.13 | |
100.5 (13.9) | 97.9 (15.2) | 97.9 (17.5) | 0.09 | |
81.0 (5.6) | 80.5 (6.4) | 79.5 (8.1) | 0.31 | |
100.4 (29.4) | 97.2 (32.5) | 95.2 (31.0) | 0.34 |
Of 3,257 subjects who underwent spirometry, ACTDA values were not available for 10 subjects and Brinkman index was not available for 335 subjects because of the lack of precise information about cigarette smoking.
This table shows characteristics of female subjects aged younger than 65 years according to serum levels of ACTDA.
Values are means (SD) or percentage.
Abbreviations: BI, Brinkman index; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
71.4 (4.6) | 71.8 (4.6) | 71.7 (4.9) | 0.59 | |
8.2 (73.3) | 8.3 (74.1) | 0.0 (0.0) | 0.81 | |
4 | 2.7 | 3 | 0.74 | |
99.4 (15.6) | 99.9 (15.6) | 95.8 (19.8) | 0.38 | |
99.3 (16.9) | 100.3 (16.4) | 96.2 (18.9) | 0.45 | |
78.7 (7.2) | 79.0 (6.9) | 79.4 (8.1) | 0.76 | |
99.4 (37.0) | 103.0 (41.6) | 95.5 (39.2) | 0.48 |
Of 3,257 subjects who underwent spirometry, ACTDA values were not available for 10 subjects and Brinkman index was not available for 335 subjects because of the lack of precise information about cigarette smoking.
This table shows characteristics of female subjects aged 65 years and older according to serum levels of ACTDA.
Values are means (SD) or percentage.
Abbreviations: BI, Brinkman index; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
We investigated the relationship between log ACTDA value and spirometric measures (
Graphs show the relationships between annual decline in spirometric parameters (A and B, % predicted FVC; C and D, % predicted FEV1; E and F, FEV1/FVC; G and H, % predicted MMF) and serum ACTDA value relative to sex (male: A, C, E, and G; female: B, D, F and H). A significant relationship of log ACTDA value was observed with FEV1, FEV1/FVC, and MMF in male subjects (A, b = -2.51,
ACTDA values were not available for 10 subjects. n = 1498 and 1746 male and female subjects, respectively.
b, partial regression coefficient; β, standard partial regression coefficient; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow.
Men (n = 191) |
Women (n = 372) |
|||||||
---|---|---|---|---|---|---|---|---|
b | SE | β | b | SE | β | |||
FVC% predicted | -15.51 | 9.3 | 0.097 | -0.12 | -0.53 | 4.43 | 0.905 | 0.01 |
FEV1% predicted | -23.73 | 10.15 | 0.021 | -0.17 | -0.49 | 4.84 | 0.919 | -0.01 |
FEV1/FVC | -10.08 | 5.19 | 0.053 | -0.14 | -1.2 | 2.07 | 0.563 | -0.03 |
MMF% predicted | -41.3 | 19.91 | 0.04 | -0.15 | -10.08 | 11.02 | 0.361 | -0.05 |
Univariate linear regression analyses in ACTDA-positive individuals; explanatory variable, log ACTDA value; dependent variable, each spirometric parameter
Abbreviations: b, partial regression coefficient; β, standard partial regression coefficient; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
Men (n = 18) |
Women (n = 59) |
|||||||
---|---|---|---|---|---|---|---|---|
b | SE | β | b | SE | β | |||
FVC% predicted | 9.03 | 35.3 | 0.801 | 0.06 | 11.2 | 16.13 | 0.491 | 0.09 |
FEV1% predicted | 20.4 | 46.83 | 0.669 | 0.11 | 2.76 | 16.94 | 0.871 | 0.02 |
FEV1/FVC | 11.9 | 24.36 | 0.632 | 0.12 | -6.74 | 7.43 | 0.368 | -0.1 |
MMF% predicted | -6.4 | 67.99 | 0.926 | -0.02 | -29 | 32.89 | 0.382 | -0.1 |
Univariate linear regression analyses in ACTDA-positive individuals; explanatory variable, log ACTDA value; dependent variable, each spirometric parameter
Abbreviations: b, partial regression coefficient; β, standard partial regression coefficient; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
BI (FVC% predicted) | -0.005 | 0.001 | <0.0001 | -0.14 |
Log ACTDA (FVC% predicted) | -1.8 | 1.028 | 0.08 | -0.03 |
BI (FEV1% predicted) | -0.009 | 0.001 | <0.0001 | -0.22 |
Log ACTDA (FEV1% predicted) | -2.56 | 1.143 | 0.025 | -0.04 |
BI (FEV1/FVC) | -0.004 | 0 | <0.0001 | -0.2 |
Log ACTDA (FEV1/FVC) | -0.413 | 0.523 | 0.43 | -0.01 |
BI (MMF% predicted) | -0.019 | 0.002 | <0.0001 | -0.2 |
Log ACTDA (MMF% predicted) | -5.796 | 2.473 | 0.019 | -0.04 |
Multivariate linear regression analyses in all study subjects; explanatory variable (v.), BI and log ACTDA value; dependent v., each spirometric parameter
FVC% predicted, FEV1% predicted and MMF% predicted were standardized values by age, sex, and height (Reference #10). FEV1/FVC was adjusted for age and sex.
Abbreviations: b, partial regression coefficient; β, standard partial regression coefficient; BI, Brinkman index; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flow
To examine if positive ACTDA was the cause of restrictive or obstructive pulmonary disease in never-smokers, we compared ACTDA positivity among the following spirometric classifications: normal, % predicted FVC ≥ 80 and FEV1/FVC ≥ 0.7; restrictive, % predicted FVC < 80 and FEV1/FVC ≥ 0.7; obstructive, % predicted FVC ≥ 80 and FEV1/FVC <0.7; and mixed disorder, % predicted FVC < 80 and FEV1/FVC < 0.7. As shown in
439 | 45 | 10.3 | 7.7–13.4 | 63.3 (10.1) | |
21 | 21 | 14.3 | 5.0–34.6 | 65.4 (11.9) | |
31 | 3 | 9.7 | 3.3–24.9 | 67.7 (8.9) | |
10 | 5 | 50* | 23.7–76.3 | 74.9 (5.3)* | |
725 | 93 | 12.8 | 10.6–15.5 | 60.4 (10.5) | |
75 | 10 | 13.3 | 7.5–23.1 | 64.9 (10.3)* | |
153 | 24 | 15.7 | 7.5–23.1 | 67.6 (8.4)* | |
45 | 8 | 17.8 | 9.3–31.3 | 69.3 (6.5)* | |
1415 | 296 | 20.9 | 18.9–23.1 | 62.0 (10.1) | |
86 | 25 | 29.1 | 20.5–39.4 | 65.8 (10.2)* | |
70 | 15 | 21.4 | 13.4–32.3 | 66.7 (9.8) | |
15 | 5 | 33.3 | 15.2–58.3 | 66.7 (7.5) | |
135 | 28 | 20.7 | 14.8–28.3 | 53.7 (8.9) | |
13 | 1 | 7.7 | 1.3–33.3 | 57.3 (12.2) | |
10 | 1 | 10 | 1.8–40.4 | 63.6 (10.2)* | |
3 | 1 | 33.3 | 6.1–79.2 | 60.7 (7.8) |
Age: mean years-old (standard deviation)
Smokers included both past and current smokers.
ACTDA values were not available in 2 male never-smokers, 1 male past/current smoker, and 7 female never-smokers.
Differences in the percentage of positive ACTDA and the age by type of spirometric disorder were evaluated using the chi-square test and analysis of variance followed by Tukey’s test, respectively. *:
Abbreviations: CI, confidence interval
439 | 45 | 10.3 | 7.7–13.4 | 63.3 (10.1) | |
22 | 2 | 9.1 | 2.5–27.8 | 67.3 (9.1) | |
19 | 6 | 31.6* | 15.4–54.0 | 71.9 (7.8)* | |
1415 | 296 | 20.9 | 18.9–23.1 | 62.0 (10.1) | |
45 | 11 | 24.4 | 14.2–38.7 | 66.3 (10.0)* | |
40 | 9 | 22.5 | 12.3–37.5 | 67.1 (8.7)* |
Age: mean years-old (standard deviation)
Median values of % predicted FEV1 among subjects with obstructive disorder were 75.35% in males and 77.15% in females.
ACTDA values were not available in 2 male never-smokers, 1 male past/current smoker, and 7 female never-smokers.
Differences in the percentage of positive ACTDA and the age by type of spirometric disorder were evaluated using the chi-square test and analysis of variance followed by Tukey’s test, respectively. *:
Abbreviations: CI, confidence interval; FEV1, forced expiratory volume in 1 s
The relationship between decline in spirometric values and the logarithmically transformed value of ACTDA was investigated. As shown in
b | SE | β | ||
---|---|---|---|---|
Change in FVC (%/year) | -2.061 | 0.962 | 0.034 | -0.18 |
Change in FEV1 (%/year) | -2.283 | 1.146 | 0.048 | -0.16 |
Change in FEV1/FVC (%/year) | -0.167 | 0.774 | 0.829 | -0.02 |
Change in MMF (%/year) | -3.816 | 3.383 | 0.261 | -0.09 |
Relationships between spirometric measures and ACTDA values were analyzed using univariate linear regression analyses. Explanatory variable, log ACTDA value; dependent variable, each decline in spirometric parameter
n = 147
Abbreviations: b, partial regression coefficient; β, standard partial regression coefficient; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flowβ, standard partial regression coefficient
b | SE | β | ||
---|---|---|---|---|
Change in FVC (%/year) | -1.715 | 1.037 | 0.101 | -0.15 |
Change in FEV1 (%/year) | -1.688 | 1.015 | 0.099 | -0.15 |
Change in FEV1/FVC (%/year) | -0.036 | 0.763 | 0.962 | -0.01 |
Change in MMF (%/year) | -0.993 | 2.402 | 0.679 | -0.04 |
Relationships between spirometric measures and ACTDA values were analyzed using multivariate linear regression analyses. Explanatory variable, log ACTDA value and Brinkman index; dependent variable, each decline in spirometric parameter
n = 147
Abbreviations: b, partial regression coefficient; β, standard partial regression coefficient; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 s; MMF, maximal midexpiratory flowβ, standard partial regression coefficient
In the present study, we demonstrated an inverse relationship between spirometric measures and ACTDA value in male subjects participating in an annual health check-up. The relationships between ACTDA value and FEV1, and MMF remained significant after adjustment for Brinkman index. In male never-smokers, the prevalence of ACTDA positivity was significantly greater in subjects with more severe airway obstruction (% predicted FEV1 < the median value) than in normal subjects and subjects with less severe airway obstruction (% predicted FEV1 ≥ the median value), although it is possible that the difference in percentage of male never-smokers positive for ACTDA may be attributable to the difference in the mean age of each group.
Accumulating evidence has suggested the involvement of autoimmunity in the pathogenesis of COPD. In addition, some autoimmune diseases cause restrictive lung disorders, namely, interstitial pneumonias [
The precise causes and diagnoses of the pulmonary dysfunction/restrictive lung disorders were not clear in the present study because this study was not hospital-based research and such information was not available. There may be various causes of restrictive pulmonary disorders in Japan, and the prevalence of interstitial pneumonias is low [
It is reasonable to assume that inhalation of cigarette smoke induces impairment of pulmonary function. In this study, we demonstrated a significant positive relationship between ACTDA value and Brinkman index in healthy male subjects, although the habit of smoking did not related to the level of ACTDA (
A primary cause of obstructive disorders may be COPD or bronchial asthma. In Japan, the prevalence of COPD and bronchial asthma is estimated from 8% to 10% and from 1% to 3% in adults, respectively [
In contrast to the positive relationship between ACTDA value and Brinkman index in the present study, we demonstrated that the prevalence of ACTDA positivity was greater in male never-smokers with relatively severe airflow obstruction. Of 501 male never-smokers, 41 subjects had airflow obstruction (31, obstructive; 10 mixed), and 8 of those subjects were positive for ACTDA. The prevalence of ACTDA positivity was greater in male never-smokers with the mixed pulmonary disorder than in subjects with other spirometric patterns. Autoimmunity has been suggested to be involved in the progression of COPD after cessation of cigarette smoking [
Severity of airflow obstruction is usually evaluated by GOLD classification [
As mentioned earlier, the fact that the present study was not hospital based presents several limitations, including the lack of information regarding patients’ (i) history of autoimmune disease; (ii) findings on chest radiography; (iii) current medications, including immunosuppressive agents such as corticosteroids; and (iv) socio-economic status, which has the potential to create sampling bias. Furthermore, we did not measure ANA levels by IIF using HEp-2 cells, which is used in clinical environments to confirm the presence of ANA. The MESACUP ANA ELISA test performed in this study uses a mixture of only 9 antigens recognized by autoantibodies in systemic rheumatic diseases. In contrast, antibodies to hundreds to thousands of antigens are detected in a standard immunofluorescence ANA screening test. Only a small subset of individuals positive for ACTDA by immunofluorescence have autoantibodies to some of the 9 antigens used in MESACUP ANA EIA. In the future, IIF measurement of ANA and disease-specific antibodies should be evaluated in subjects with airflow obstruction without a history of cigarette smoking.
Because this study was performed in a sample of the Japanese population, it is still unknown if this relationship between ACTDA and respiratory function would exist for other ethnicities. ANA levels are reported to be twice as high in Japan than in the USA. Genetic backgrounds including human leukocyte antigens; and lifestyles, such as diet and smoking habit, are different in each ethnicity. Therefore, further studies are needed to confirm this correlation between autoimmunity and respiratory functions for other ethnicities.
In the longitudinal analysis, although significant associations between decline in FVC and FEV1 and ACTDA were observed, these associations seem to be weak because statistical significance did not remain after adjustment for Brinkman Index. The number of subjects participating in the longitudinal study (n= 147) might not have been large enough to achieve statistical significance in this assay. In the future, studies using larger numbers of subjects may be needed to evaluate the relationships between longitudinal changes in spirometric values and autoimmunity.
In conclusion, we have demonstrated inverse relationships between ACTDA value and spirometric parameters in healthy male subjects participating in an annual health checkup, suggesting the involvement of an autoimmune mechanism in the development of pulmonary dysfunction.
We thank Taiko Aita, Emiko Nakamura, and Eiji Tsuchida for their excellent technical assistance. We also thank the following contributors: Michiko Nishiwaki (Saiseikan Yamagata City Hospital, Yamagata, Japan), Toshihiro Wada (Saiseikan Yamagata City Hospital, Yamagata, Japan), Jun-Ichi Machiya (Nihonkai General Hospital, Sakata, Japan), Noriyuki Hirama (NHO Yamagata National Hospital, Yamagata, Japan), Noriaki Takabatake (Tohoku Central Hospital, Yamagata, Japan), and Makoto Sata (National Cerebral and Cardiovascular Center, Osaka, Japan). We also thank Toshiro Tango (Center for Medical Statistics, Tokyo, Japan) for evaluating the statistical analyses.