Prognosis and Prognostic Factors in Patients With Idiopathic Dilated Cardiomyopathy in Japan

Transcription

1 CLINICAL INVESTIGATIONS Circ J 2008; 72: Prognosis and Prognostic Factors in Patients With Idiopathic Dilated Cardiomyopathy in Japan Results From a Nationwide Study Katsuyuki Miura, MD; Akira Matsumori, MD*; Ali Nasermoaddeli, MD; Yoshiyuki Soyama, PhD; Yuko Morikawa, MD; Masaru Sakurai, MD; Akira Kitabatake, MD**; Masaki Nagai, MD ; Yutaka Inaba, MD ; Hideaki Nakagawa, MD Background There have been few large-scale nationwide studies investigating both the prognosis and the prognostic factors of idiopathic dilated cardiomyopathy (IDC). A predictive score that can be used in clinical practice has not been established. Methods and Results A nationwide epidemiological study of the prognosis of IDC was conducted in 1999 among randomly selected hospitals in Japan, and 147 departments participated in the present 5-year follow-up survey. The vital status of 1,554 IDC patients was collected in 2004 using medical records and residence-based registers. The crude 5-year survival rate for those diagnosed in 1998 was 78.6%. Cox s regression model selected 5 independent predictors of mortality: male sex, higher age, higher New York Heart Association functional class, higher left ventricular diameter index, and lower left ventricular ejection fraction. A predictive score using these 5 variables effectively predicted prognosis; 5-year survival rates were 90.6% in patients with a score of 4 or less and 49.0% in patients with a score of 9 or 10. Conclusions This nationwide survey revealed the present prognostic status of IDC in Japan and 5 independent predictors of prognosis that can be used in clinical practice as a predictive score. (Circ J 2008; 72: ) Key Words: Cardiomyopathy; Epidemiology; Prognosis To ascertain both the prognosis and the prognostic factors of a disease is important for assessing the present level of medical care for that disease, for doctors in choosing the treatment method, and for patients in making an informed decision. The prognosis of idiopathic dilated cardiomyopathy (IDC) has been reported in several previous studies conducted in the 1980s and 1990s, 1 9 which showed it has been improving. 10 However, those results need to be carefully assessed for several reasons. First, the widespread use of echocardiography has made possible the early detection of asymptomatic patients, so the proportion of patients with mild IDC may be increasing. Second, most of the previous studies were relatively small in scale, within certain hospitals or in a certain area possibly providing better medical care, so the real prognosis of the disease on a nationwide level may be different. Moreover, large hospitals with specialists may deal with more severe cases. There have (Received June 22, 2007; revised manuscript received October 21, 2007; accepted October 23, 2007) Department of Epidemiology and Public Health, Kanazawa Medical University, Ishikawa, *Department of Cardiovascular Medicine, Kyoto University Graduate School of Medicine, Kyoto, **Department of Cardiovascular Medicine, Hokkaido University Graduate School of Medicine, Sapporo, Department of Public Health, Saitama Medical School, Saitama and Department of Epidemiology and Environmental Health, Juntendo University School of Medicine, Tokyo, Japan Mailing address: Katsuyuki Miura, MD, Department of Epidemiology and Public Health, Kanazawa Medical University, 1-1 Daigaku, Uchinada-machi, Kahoku, Ishikawa , Japan. All rights are reserved to the Japanese Circulation Society. For permissions, please cj@j-circ.or.jp been very few nationwide epidemiological surveys that included small hospitals and clinics. In addition, recent progress in cardiac transplantation in Western countries has made the determination of prognosis more difficult and few studies have been performed to investigate prognosis in Asian patients with IDC. In 1999, the Japanese Research Committees on Epidemiology of Intractable Diseases and on Idiopathic Cardiomyopathies undertook a nationwide epidemiological survey of idiopathic cardiomyopathy in Japan in order to provide a detailed description of the clinico-epidemiological features so that appropriate health service planning could be facilitated. A detailed description is given elsewhere, 11,12 And the purpose of the present study was to evaluate the 5-year survival rate of IDC patients from that nationwide study in Japan, where cardiac transplantation for IDC is not yet popular, and to clarify factors that can predict the prognosis of this disease independently and reliably. We also attempted to create a score for the prediction of prognosis. Methods Nationwide Survey The nationwide survey of cardiomyopathies, including IDC, was designed to reveal their prevalence and clinical features in Japan. 11,12 Briefly, the Japanese Research Committee on Idiopathic Cardiomyopathies prepared classification criteria that were based on the report of the World Health Organization/International Society and Federation of Cardiology (WHO/ISFC) task force on the definition and classification of cardiomyopathies. 13,14 Specific heart muscle

2 344 MIURA K et al. Table 1 Five-Year Survival Rates for the Predictive Variables in Patients With IDC in Japan n (%) Deaths 5-year survival p for rate (%) log-rank test Sex Men 1,128 (72.6) Women 426 (27.4) Age (years) <30 85 (5.5) (43.3) (51.2) <0.001 BMI (kg/m 2 ) < (22.0) (53.6) (24.4) <0.001 Family history of IDC Yes 87 (6.9) ,165 (93.1) History of hypertension Yes 297 (20.1) ,178 (79.9) History of diabetes mellitus Yes 199 (13.3) ,298 (86.7) Habitual drinking before diagnosis Yes 495 (36.3) No 866 (63.7) Habitual smoking before diagnosis Yes 547 (39.6) No 833 (60.6) NYHA functional class I 300 (20.5) II 601 (41.2) III 413 (28.3) IV 146 (10.0) <0.001 Rhythm Sinus rhythm 1,053 (72.6) Atrial fibrillation 378 (26.1) Atrial flutter 20 (1.3) Left high voltage on ECG Yes 562 (39.9) No 846 (60.1) Left bundle branch block Yes 191 (14.4) ,131 (85.6) <0.001 CTR on chest X-ray (%) < (14.8) (23.0) (24.2) (38.0) <0.001 LV diameter at diastole (mm) (age 20 years) <50 78 (5.5) (31.0) (42.3) (21.2) <0.001 LV diameter index (mm/m 2 ) < (14.2) (28.4) (27.6) (16.3) (13.5) <0.001 LVEF (%) < (12.2) (24.6) (29.2) (20.4) (13.4) <0.001 No. of beds in the hospital < (7.0) (9.7) (7.0) (29.5) University hospital 666 (46.8) IDC, idiopathic dilated cardiomyopathy; BMI, body mass index; NYHA, New York Heart Association; CTR, cardiothoracic ratio; LV, left ventricular; LVEF, LV ejection fraction.

3 Prognosis of IDC in Japan disease, which was defined as heart muscle disease of known etiology or associated with disorders of other systems, was excluded from the survey. The hospitals included in each survey were randomly selected by stratified sampling of all departments of internal medicine, cardiovascular medicine and pediatrics throughout Japan, identified in a directory of names, department addresses, and number of hospital beds obtained from the Ministry of Health, Labour and Welfare of Japan. Baseline Survey The survey investigated patients with IDC as either inpatients or outpatients in specific departments in The questionnaire for the first survey of the number of IDC patients was mailed directly to 2,414 departments in January 1999 and of those, 1,409 (58.4%) responded, reporting data on 6,341 patients. A second survey was performed to collect detailed clinical data. From a total of 577 departments that reported 1 or more IDC patients in the first survey, 191 departments agreed to participate in the second survey and detailed clinical data were collected from a total of 1,932 patients. Patients who died prior to 1998 or those visiting a hospital for the first time during 1999 or later were excluded as inappropriate cases, as were patients whose data were reported from more than 1 department (duplicate cases). The questionnaire requested detailed clinico-epidemiological information for each patient, including age, sex, symptoms and New York Heart Association (NYHA) functional class. Data from physical examination and baseline laboratory measurements, including standard 12-lead electrocardiogram, chest X-ray and echocardiography (for measuring left ventricular diameter (LVD) at end-diastole and left ventricular ejection fraction (LVEF)), were available. Left high voltage was detected by ECG (Minnesota code: 3-1 or 3-3). M-mode echocardiographic assessment for LVEF and 2-dimensional echocardiographic assessment for LVD were conducted in 95% of patients. LVD index was calculated as LVD (mm) divided by body surface area (m 2 ). Analyses of LVD were performed only for patients aged 20 years or over. Data on blood tests and cardiac catheterization were obtained for a small proportion of the subjects and therefore we did not include it in our analysis. Patients were generally given medical therapy as reported previously. 12 Follow-up Of the 191 departments that reported 1,932 subjects in the second survey, 147 departments (reporting 1,625 patients) agreed to participate in the 5-year follow-up survey. Patients vital status was reported by doctors, with vital status for 462 withdrawn cases obtained from the residence-based register of the local government for those patients. However, follow-up was not possible for 71 patients, either because their addresses were not obtainable or the refusal of the local government to cooperate, so they were excluded. Therefore, 5-year follow-up was completed for 1,554 (95.6%) patients. To address the possibility of follow-up bias, we found no significant difference in the sex, age, body mass index, and NYHA functional class distribution between those who participated and those who did not participate in the followup. The ethical committees of the Kanazawa Medical University and the Kyoto University Graduate School of Medicine approved the study protocol. A nationwide survey of the prognosis of idiopathic hypertrophic cardiomyopathy was performed at the same time. 15 Age- and sex-adjusted HR (95%CI) Sex Men 1.27 ( )* Women 1 Age (years) < ( ) ( ) BMI (kg/m 2 ) < ( ) ( ) Family history of IDC Yes 1.51 ( ) History of hypertension Yes 1.00 ( ) History of diabetes mellitus Yes 1.11 ( ) Habitual drinking before diagnosis Yes 0.82 ( ) Habitual smoking before diagnosis Yes 0.88 ( ) NYHA functional class I 1 II 1.92 ( ) III 3.33 ( ) IV 2.84 ( ) Rhythm Sinus rhythm 1 Atrial fibrillation 0.69 ( ) Atrial flutter 1.26 ( ) Left high voltage on ECG Yes 0.90 ( ) Left bundle branch block Yes 1.58 ( ) CTR on chest X-ray (%) < ( ) ( ) ( ) LV diameter at diastole (mm) (age 20 years) < ( ) ( ) ( ) LV diameter index (mm/m 2 ) < ( ) ( ) ( ) ( ) LVEF (%) < ( ) ( ) ( ) ( ) 50 1 No. of beds in the hospital < ( ) ( ) ( ) University hospital 0.88 ( ) 345 Table 2 Age- and Sex-Adjusted HR of All-Cause Mortality in Patients With IDC According to Each Predictive Variable *Adjusted only for age; adjusted only for sex. HR, hazard ratios; CI, confidence interval. Other abbreviations see in Table 1.

4 346 MIURA K et al. Table 3 Multivariate Adjusted HR of All-Cause Mortality for Selected Predictors in Patients With IDC Adjusted HR* (95%CI) Wald statistic p value Sex (men vs women) 1.68 ( ) Age (for 10-year increase) 1.40 ( ) 46.8 <0.001 NYHA functional class (for 1 class increase) 1.28 ( ) 11.0 <0.001 LV diameter index (for 10 mm/m 2 increase) 1.42 ( ) 23.2 <0.001 LVEF (for 10% decrease) 1.23 ( ) 14.3 <0.001 *All 5 variables are included in the same Cox s proportional hazard model. The 5 variables were selected from all variables using a stepwise selection method. See Tables 1,2 for abbreviations. Table 4 Scoring by the 5 Predictive Variables Score Sex Men 2 Women 0 Age (years) < NYHA functional class I 0 II 1 III or IV 2 LV diameter index (mm/m 2 ) < LVEF (%) < Total score ranges from 0 to 10. See Table 1 for abbreviations. Table 5 Score Five-Year Survival Rates by the Category of Total Predictive Total predictive score n (%) Deaths 5-year survival rate (%) (26.8) or (36.8) or (27.4) or (9.0) Survival rates were significantly different by log-rank test (p<0.001). Statistical Analyses Survival estimates were calculated using the Kaplan-Meier method, and the 5-year (1,825 days) survival probability was calculated for the overall cohort. Patients were classified on the basis of baseline prognostic factors. Significant differences in survival rates among the classifications were tested by the log-rank test for trends. Hazard ratios (HR) according to baseline characteristics were calculated by Cox s proportional hazard model with 95% confidence intervals up to the longest follow-up time of 2,190 days. Ageand sex-adjusted HRs were calculated for each prognostic factor and then multivariate-adjusted HRs were calculated in a model that included all variables selected by the stepwise selection method (p<0.15). In this model, variables were included as continuous variables and their statistical significance was compared by the Wald statistic. The log minus log plotted against survival time for each covariate did not show any deviation from the proportionality assumption. The data were analyzed with SPSS (version 12.0J; Chicago, IL, USA). All reported significance levels are p<0.05 (2- tailed tests). Results Five-Year Survival Rate Of the 1,554 patients identified at baseline, 420 (27.0%) died during the follow-up period. The probability of actuarial survival was calculated from the time of the baseline survey for 390 patients who were initially diagnosed in The crude 5-year survival rate for those diagnosed in 1998 was 78.6%. The crude 5-year survival rate for the whole cohort was 75.7%. Among the 264 dead patients whose cause of death was determined, 141 (53%) died from heart failure, 30 (11%) from sudden death, 30 (11%) from cancer, and 21 (8%) from arrhythmia. Baseline characteristics of the patients are shown in Table 1. Of the 1,554 patients in the study, 27.4% were women and 51.2% were aged over 60 years. The majority of patients (61.7%) had experienced no or only mild symptoms (NYHA function classes I or II) at baseline. Prognostic Factors Clinical, echocardiographic, and standard 12-lead ECG variables were examined for an association with survival during the follow-up period (Table 1). Crude 5-year survival rates significantly decreased with higher age, lower body mass index, higher NYHA functional class, presence of left bundle branch block, increasing cardiothoracic ratio on chest X-ray, higher LVD, higher LVD index, and lower LVEF. The 5-year survival rate for NYHA class III was 65.1%, whereas that for NYHA class I was 89.1%. The rate for an LVD index 47 mm/m 2 or higher was 60.7%, whereas that for an LVD index less than 32mm/m 2 was 88.2%. There was no significant difference in the crude survival rate of men and women. Family history of IDC, history of hypertension and diabetes, smoking and drinking habits, and hospital size were not associated with 5-year survival rates. Table 2 presents the age- and sex-adjusted HRs according to the prognostic factors. The HR for all-cause mortality among male patients was slightly higher than that for female patients. A significantly higher HR was observed for patients with a body mass index of less than 20kg/m 2 compared with patients with a body mass index of kg/m 2. HRs were significantly higher for family history, higher NYHA functional class, left bundle branch block, and higher cardiothoracic ratio. Patients with an LVD index of 47 mm/m 2 or higher or with an LVEF of less than 20% had an over 3- fold higher HR for death compared with patients with an LVD index of less than 32 mm/m 2 or with an LVEF of 50% or higher. HRs did not differ according to hospital size. Table 3 presents the results of multivariate-adjusted analysis. Five variables (sex, age, NYHA functional class, LVD

5 Prognosis of IDC in Japan index, and LVEF) were selected when the stepwise selection method was performed using all variables in the Cox s regression model. The HRs shown in Table3 were adjusted for the 4 other variables. All 5 variables were independently and significantly related to mortality risk. The Wald statistic showed that age had the strongest association with prognosis. A 10-year increase in age was related to a 40% higher risk of death. The LVD index was the second strongest predictor of death; a 10 mm/m 2 increase was related to a 42% higher risk of death. Male sex independently increased the mortality risk. Predictive Score Using these 5 independently related variables, we created a score for the prediction of prognosis. An approximately 20 30% increase in mortality risk, as shown in Table3, corresponded to a 1-point increase in the score (Table4). In this scoring, the results shown in Tables 1 and 2 were also taken into consideration. The total score from the 5 predictive variables (sex, age, NYHA class, LVD index, and LVEF) ranged from 0 to 10. Table5 shows the 5-year survival rates according to the total predictive score: they were 90.6% in patients with a score of 4 or less and 49.0% in patients with a score of 9 or 10. Survival curves according to the score category are shown in Fig 1. Discussion We report the 5-year survival rates and adjusted HRs for all-cause mortality by baseline prognostic factors from a nationwide study of IDC in Japan. Although there has been a previous survey in Japan, 3 it mainly involved university hospitals, so the present study is the first nationwide followup survey in Japan to include various sized hospitals, and such a nationwide survey has not been done in any other country. The current 5-year survival rate of this disease after diagnosis is 78.6% in Japan. One methodological issue in the present survey is the diagnostic criteria used. In 1995, the WHO/ISEF task force reported a new definition and classification of cardiomyopathy in which the cardiomyopathies were defined simply as diseases of the myocardium associated with cardiac dysfunction. 16 However, we used the definition and classification provided by the earlier task force of 1980, 13,14 in which idiopathic cardiomyopathy was distinguished from other specific heart muscle diseases. Our reasons for doing this were, first, that nearly all cardiologists and specialists in general medicine in Japan have been using this definition in their diagnosis of cardiomyopathies for a long time; and second, that numerous previous reports have also used the same definition, allowing us to compare our data with those reports. Most prior studies of the natural history and prognosis of IDC have been based on populations of selected patients from referral centers, 1 9 and therefore, on the basis of different levels of care and management, various prognoses would be expected. The clinical outcome and perception of prognostic factors in IDC is profoundly affected by a bias in patient selection. However, the present study is free from referral bias because patients were recruited from all over the nation and from different diagnostic centers. Previous studies in the 1980s demonstrated that the 5-year survival rate was approximately 50%, 1 3 but more recent studies into the 1990s reported better survival rates of 70 80%. 4 6,8,9 Survival rates in the present study (ie, 78.6% Cumulative survival rate or less Survival days in newly diagnosed patients and 75.7% in total patients) were similar to those in the studies conducted in the 1990s in which the proportion of asymptomatic patients (NYHA functional class I: 15 20%) was similar to this study. An interesting finding in the present nationwide Japanese study is that survival rates were not significantly different among general hospitals and university hospitals, regardless of their size. Left ventricular dilatation, 4 6,17 lower LVEF, 2,5,6,17,18 and higher NYHA functional class 2,5,6,18 are relatively established predictors of poor prognosis. Older age is also reported to predict poor prognosis. 1,4,6 Few previous reports showed a relationship between male sex and poor prognosis. In the present study, these 5 variables were independently related to prognosis. Most previous studies did not examine the predictive ability of these variables in a model including all variables simultaneously. Therefore, the independent relationships of these predictors to prognosis have been seldom investigated. For example, although prognosis did not differ by sex in the univariate analysis, as observed in previous studies, male sex was a significant predictor in the multivariate analysis in our study. This would be the first study to demonstrate the independent predictive ability of these 5 variables in a large cohort of IDC patients. In the present study, we attempted to create a score to predict prognosis using the 5 related variables, all of which are now routinely measured in the clinical setting, and such a predictive score should prove useful to clinicians. Our large-scale cohort has made the creation of this predictive score and the simulation of prediction possible, which were difficult in previous studies given the small number of patients. Study Limitations First, we failed to differentiate cardiovascular deaths, including sudden deaths and deaths caused by end-stage cardiac failure, from other causes of mortality, and we Fig 1. Kaplan-Meier survival curves by the total predictive score from the 5 predictive variables shown in Table 4.

6 348 MIURA K et al. reported all-cause mortality as our main outcome. Second, because the study was based on a large-scale nationwide survey, the methods used for various measurements were not fully standardized, and the validity of diagnosis may not be sufficiently high in some departments. Third, findings from cardiac catheterization, cardiac biopsy, and blood biomarkers, which were measured in part of the cohort, were not considered in this analysis. Fourth, medical treatment was not taken into consideration because the main purpose of this study was to describe the status of IDC patients with average treatment at present in Japan. Fifth, because a relatively small number of departments in the baseline survey participated in the follow-up survey, there might be a selection bias. Finally, the prediction of prognosis using our predictive score may be difficult to generalize to IDC patients in other countries. In conclusion, a nationwide survey has revealed the present prognostic status of IDC in Japan, and a poorer prognosis for IDC is predicted by higher LVD index, lower LVEF, higher NYHA functional class, older age and male sex. We have created a score for the prediction of prognosis using these 5 variables to assist clinicians in determining the outcome for their IDC patients. Acknowledgments The authors are grateful to the doctors who participated in this nationwide survey. This study was supported by a Grant-in-Aid for the Epidemiology of Intractable Diseases Research Committee and a Grant-in-Aid for Idiopathic Cardiomyopathy Research Committee from the Ministry of Health, Labour and Welfare of Japan. References 1. Fuster V, Gersh BJ, Guiliani ER, Tajik AJ, Brandenburg RO, Frye RL. The natural history of idiopathic dilated cardiomyopathy. Am J Cardiol 1981; 47: Diaz RA, Obasohan A, Oakley CM. Prediction of outcome in dilated cardiomyopathy. Br Heart J 1987; 58: Kawai C, Sakurai T, Kishimoto C, Tomioka N. A follow up survey for the prognosisi of idiopathic cardiomyopathies in Japan. In: Annual Report of the Research Committee on Idiopathic Cardiomyopathy, the Ministry of Health and Welfare, Japan. Tokyo: The Ministry; 1983; (in Japanese). 4. Ikram H, Williamson HG, Won M, Crozier IG, Wells EJ. The course of idiopathic dilated cardiomyopathy in New Zealand. Br Heart J 1987; 57: Komajda M, Jais JP, Reeves F, Goldfarb B, Bouhour JB, Juillieres Y, et al. Factors predicting mortality in idiopathic dilated cardiomyopathy. Eur Heart J 1990; 11: Sugrue DD, Rodeheffer RJ, Codd MB, Ballard DJ, Fuster V, Gersh BJ. The clinical course of idiopathic dilated cardiomyopathy: A population-based study. Ann Intern Med 1992; 117: Redfield MM, Gersh BJ, Bailey KR, Rodeheffer RJ. Natural history of incidentally discovered, asymptomatic idiopathic dilated cardiomyopathy. Am J Cardiol 1994; 74: Felker GM, Thompson RE, Hare JM, Hruban RH, Clemetson DE, Howard DL, et al. Underlying causes and long-term survival in patients with initially unexplained cardiomyopathy. N Engl J Med 2000; 342: Matsumura Y, Takata J, Kitaoka H, Kubo T, Baba Y, Hoshikawa E, et al. Long-term prognosis of dilated cardiomyopathy revisited: An improvement in survival over the past 20 years. Circ J 2006; 70: Dec GW, Fuster V. Idiopathic dilated cardiomyopathy. N Engl J Med 1994; 331: Miura K, Nakagawa H, Morikawa Y, Sasayama S, Matsumori A, Hasegawa K, et al. Epidemiology of idiopathic cardiomyopathy in Japan: Results from a nationwide survey. Heart 2002; 87: Matsumori A, Furukawa Y, Hasegawa K, Sato Y, Nakagawa H, Morikawa Y, et al. Epidemiological and clinical characteristics of cardiomyopathies in Japan: Results from nationwide surveys. Circ J 2002; 66: The WHO/ISFC task force on the definition and classification of cardiomyopathies: Report of the WHO/ISFC task force on the definition and classification of cardiomyopathies. Br Heart J 1980; 44: Research Committee on Idiopathic Cardiomyopathy. Guidelines for the diagnosis of idiopathic cardiomyopathy. In: Annual report of the Research Committee on Idiopathic Cardiomyopathy. Tokyo: Ministry of Health and Welfare, Japan; 1986; 13 5 (in Japanese). 15. Nasermoaddeli A, Miura K, Matsumori A, Soyama Y, Morikawa Y, Kitabatake A, et al. Prognosis and prognostic factors in patients with hypertrophic cardiomyopathy in Japan: Results from a nationwide study. Heart 2007; 93: Richardson P, McKenna W, Bristow M, Maisch B, Mautner B, O Connell J, et al. Report of the 1995 World Health Organization/ International Society and Federation of Cardiology Task Force on the definition and classification of cardiomyopathies. Circulation 1996; 93: Doi YL, Chikamori T, Tanaka J, Yonezawa Y, Poloniecki JD, Ozawa T, et al. Prognostic value of thallium-201 perfusion defects in idiopathic dilated cardiomyopathy. Am J Cardiol 1991; 67: Kelly TL, Cremo R, Nielsen C, Shabetai R. Prediction of outcome in late-stage cardiomyopathy. Am Heart J 1990; 119: