International Journal of Cardiology 173 (2014) 300–304
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Typical and atypical coronary heart disease deaths and their different relationships with risk factors. The Gubbio residential cohort Study Paolo Emilio Puddu a,⁎, Oscar Terradura Vagnarelli b, Mario Mancini c, Alberto Zanchetti d, Alessandro Menotti e a
Department of Cardiovascular, Respiratory, Nephrological, Anesthesiological and Geriatric Sciences, Sapienza University of Rome, Viale del Policlinico 155, I-00161 Rome, Italy Centre of Preventive Medicine, Gubbio, Italy Department of Clinical Medicine and Surgery, Federico II University, Naples, Italy d Istituto Auxologico Italiano, University of Milan, Milan, Italy e Associazione per la Ricerca Cardiologica, Via Latina 49, I-00179 Rome, Italy b c
a r t i c l e
i n f o
Article history: Received 6 December 2013 Received in revised form 30 January 2014 Accepted 9 March 2014 Available online 15 March 2014 Keywords: Coronary heart disease CHD fatalities Gubbio Cohort Study Cox model Prediction
a b s t r a c t Objectives: The Seven Countries Study showed that fatal coronary heart disease (CHD) with only chronic heart failure, arrhythmia or blocks (atypical CHD, A-CHD) may represent a distinct disease as compared to fatal CHD cases with angina pectoris, acute myocardial infarction (AMI) or sudden death (typical CHD, T-CHD). We aimed at validating this, using identical diagnostic criteria, in a separate residential cohort first examined in 1983–85 in Gubbio, central Italy. Material and methods: Forced Cox's models were run to assess 9 classic risk factors and their 20-year predictivity of A-CHD versus T-CHD, in the entire cohort or separately for men and women. Results: There were 3229 subjects aged 30–79 years. Entry mean age was slightly higher in women than men although age at death was lower in men than in women for both T-CHD (71.99 ± 11.38 versus 81.20 ± 9.35 years, p b 0.0001) and A-CHD (80.22 ± 9.44 versus 84.98 ± 8.13 years, p b 0.0001). T-CHDs were predicted by male gender, age, continued smoke, systolic blood pressure (SBP), blood glucose, total and HDL-cholesterol (protective). A-CHDs were predicted by age, continued smoke, SBP, body mass index and blood glucose but neither total nor HDL-cholesterol or gender was significant. In the entire cohort and in men there were predictive differences of T-CHD versus A-CHD fatalities only in relation to age (p b 0.01), SBP (p b 0.05) and total cholesterol (p b 0.01). Conclusion: As age, SBP and total cholesterol had a different predictive role of T-CHD versus A-CHD fatalities also in the Gubbio cohort, the possibility is reinforced that a different etiology exists between these entities. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Coronary heart disease (CHD) may present clinically either with symptoms (angina pectoris, acute myocardial infarction, AMI, with either ST segment elevation or not, the latter being the modern counterpart of the intermediate syndrome of the past) or with painless manifestations (chronic arrhythmia and blocks and congestive heart failure) whereas sudden death is closer to the first than the second category [1]. The availability of specific biological tests has tremendously increased the diagnostic accuracy of AMI [2]. However,
⁎ Corresponding author at: Laboratory of Biotechnologies Applied to Cardiovascular Medicine, Department of Cardiovascular, Respiratory, Nephrological, Anesthesiological and Geriatric Sciences, Sapienza University of Rome, Viale del Policlinico, 155, Rome 00161, Italy. Tel.: +39 06 49972659; fax: +39 06 4453891. E-mail addresses: [email protected] (P.E. Puddu), [email protected] (O. Terradura Vagnarelli), [email protected] (M. Mancini), [email protected] (A. Zanchetti), [email protected] (A. Menotti).
http://dx.doi.org/10.1016/j.ijcard.2014.03.021 0167-5273/© 2014 Elsevier Ireland Ltd. All rights reserved.
the abovementioned symptomatic and asymptomatic categories may have much less distinct boundaries when diagnostic criteria are applied in longitudinal epidemiological studies [3]. On the other hand, precise boundary definitions among different CHD clinical presentations are not provided by the most updated in vivo imaging modalities [4] specifically directed to coronary plaques [5] nor by post mortem anatomo-pathological studies [6–9]. Reports from the Seven Countries Study, a classic of international cardiovascular epidemiology [10,11], showed that fatal cases where the presentation among 10,628 men initially aged 40–59 years during 40 years of follow-up was only congestive heart failure, chronic arrhythmia, blocks or with the generic mention of chronic CHD (arbitrarily called atypical CHD, A-CHD) were not predicted by serum cholesterol [12–14]. On the opposite, fatal CHD cases manifested with angina pectoris, AMI, acute ischemic attack or sudden death (arbitrarily called typical CHD, T-CHD) showed a strong positive association between serum cholesterol and events. Moreover, there was a stronger association with age for A-CHD than for T-CHD [12–14]. Therefore, the
P.E. Puddu et al. / International Journal of Cardiology 173 (2014) 300–304
hypothesis was raised that A-CHD may represent a distinct disease as compared to T-CHD [14]. The aim of the present study was the external validation of the Seven Countries Study results whereby the differential relationship of A-CHD and T-CHD with age and total cholesterol was obtained [12–14]. It is in fact essential in epidemiological studies to try to repeat in independent cohorts the evidence obtained in one investigation [15–17]. We took advantage of a common thread of shared investigators between the Seven Countries Study [10–16] and the Gubbio Cohort Study [18–21]. We applied identical methods for case definitions and we extended the applicability of the results since we also investigated women (not included in the Seven Countries Study) yet assessing very long follow-up periods. The null hypothesis was that A-CHD and T-CHD deaths share identical risk factors and are predicted indistinctly by those factors entering significantly a forced Cox's model during 20 years of follow-up. 2. Material and methods 2.1. Population, risk factors and mortality data The Gubbio Cohort Study is a population study devoted to cardiovascular diseases conducted in the small town of Gubbio in central Italy. The original sample, first examined in 1983–1985, covered the entire population living within the ancient medieval walls, including both genders and ages 5 to 95 (n = 5376), with a participation rate of 92.2%. For the purpose of this analysis only men and women aged 30–79 were considered, for a total of 3553 subjects. A number of potential risk factors were measured at entry examination, as follows: - gender (coded as 0 and 1 for women and men, respectively); - age in years at the time of the visit; - body mass index (BMI in kg/m2) following the procedures suggested by the WHO Cardiovascular Survey Methods Manual [22] for the measurement of height and weight; - systolic blood pressure in mm Hg, measured by a trained personnel, following the procedure described by the WHO Manual [22]; in particular, participants were in sitting position, a mercury sphygmomanometer was used and the first measurement was taken after a 5 min rest; three measurements were recorded, and the average of the second and third measurement was used for analysis; the physicians taking the measurements were trained and tested using the tape-recorded procedures of the London School of Hygiene; - serum total cholesterol (mg/dl), high-density lipoprotein (HDL) serum cholesterol (mg/dl), and fasting plasma glucose (mg/dl) measured using automated enzymatic methods [23,24]; the laboratory involved in these analyses was under quality control of the WHO Lipid Reference Center of Prague; controls included also duplicate blinded analyses on at least 10% of all laboratory samples; - Cigarette smoking habits were derived from a standard questionnaire and people were classified in 3 dummy variables, i.e. current smokers, ex-smokers and never smokers, the last one used as reference. Mortality data were collected through the end of 2007 with a mean follow-up of about 20 years. Information was obtained from the local register office and review of medical records and interviews with spouses or baystanders and coded by a single reviewer (AM) using the ninth revision of the WHO International Classification of Diseases (WHO-ICD-9) [25]. This was the same person responsible for coding fatalities in the Seven Countries Study [12–14]. In case of uncertainty among apparent multiple causes of death, a hierarchical system was adopted giving precedence to violent causes, cancer in advanced stages, CHD, stroke and other causes in that order. The first cause of death was used for this analysis. The end-point for the analysis was cases of CHD classified into 2 groups: – Typical coronary heart diseases (T-CHDs) were cases coded as AMI, acute coronary ischemic attacks, angina pectoris and sudden death, after reasonable exclusions of other causes; – Atypical coronary heart diseases (A-CHDs) were cases classified as congestive heart failure, arrhythmia, blocks and ill-defined chronic CHD (without other specifications), in the absence of AMI, other ischemic attacks, angina pectoris, sudden death or other explicit etiologies. Subjects who carried CHD (including cases with previous coronary surgery), other forms of heart disease, cerebrovascular disease and peripheral artery disease at entry examination were excluded from the analysis (n = 324: prevalent cases). Among the remaining 3229 subjects there were missing data for some of the risk factors selected for this analysis. Therefore, they were imputed by a multivariate normal procedure using as reference the personal characteristics with full data. This process was based on a regression analysis using the variable containing the missing value as the dependent variable and all variables with non missing data in that row as independent variables. The values of these non missing variables from the row containing the missing values are used in the regression equation to compute the imputed values. The proportion
301
of missing values was relatively small (2.3%) and no difference was found between the means of the original and the imputed values. No significant difference was seen in the variance of the overall sample, compared with the original one. Oral consent was obtained from the participants in compliance with the Helsinki declaration. 2.2. Statistical analysis Data are presented as mean ± standard deviation or standard error (as appropriate) and minimum, maximum and range values. Death rates from T-CHD and A-CHD were computed for overall subjects and separately for men and women, including standardized rates for the two genders based on the overall age distribution. Age at death was computed for the two end-points, for the entire cohort and separately for men and women. Cox proportional hazards models were solved (after assessing the proportionally of hazards) using T-CHD and A-CHD fatalities separately and 9 risk factors (gender, age, smoking and ex smoking habits, systolic blood pressure, body mass index, blood glucose, serum total and HDL cholesterol) as possible predictors adopting the forced methods (all covariates were in the Cox's model at step zero), in order to compute their individual relative contribution. Hazard ratios (HR) and 95% confidence intervals (CI) were calculated using standard formulae. For continuous covariates, deltas corresponded approximately to 1 SD whereas the absolute range was considered for dichotomous covariates. T-tests were adopted to test the difference between coefficients of T-CHD versus A-CHD fatalities. Figures were drawn to show the relationship of age, serum cholesterol and systolic blood pressure with T-CHD and A-CHD deaths. Significance was concluded for p values b0.05.
3. Results Baseline descriptive statistics are shown in Table 1 for 3229 individuals aged 30–79 (mean age 52.92 ± 13.16) years. There were 1433 men (44%) and 1796 women (56%). During approximately 20 years of follow-up T-CHD and A-CHD death rates were respectively 5.30 and 3.98% (p = 0.3009) in men and 3.40 and 4.73% (p = 0.0531) in women. T-CHD rates were higher in men than in women (p = 0.0101) whereas A-CHD rates were similar (p = 0.3440). Heart failure was present in 49.3% of A-CHD deaths either as an explicit cause of death or it was annotated among the symptoms present before death. There was also a significant entry mean age difference in men and women, respectively 51.50 ± 13.06 and 53.85 ± 13.16 years (p b 0.0001) with a significantly lower age at death in men (p b 0001) than in women (p = 0.0014) for both T-CHD (71.99 ± 11.38 versus 81.20 ± 9.35 years, p b 0.0001) and A-CHD (80.22 ± 9.44 versus 84.98 ± 8.13 years, p b 0.0001). Table 2 shows HR and 95% CI of forced Cox's model solutions, to predict T-CHD and A-CHD fatalities in the entire cohort and in males and females. In the entire cohort, T-CHDs were predicted by male gender, age, continued smoke, systolic blood pressure, blood glucose, total cholesterol and HDL-cholesterol (the last being protective); A-CHDs were instead predicted by age, continued smoke, systolic blood pressure, body mass index and blood glucose. Notably, in A-CHD, neither total cholesterol nor HDL-cholesterol or gender had a significant role. There were significant differences for age (p b 0.01), systolic blood pressure (p b 0.05) and total cholesterol (p b 0.01) in their predictive capacities of T-CHD versus A-CHD fatalities whereas the remaining covariates considered had a comparable predictive role for either T-CHD or A-CHD. This picture was confirmed in males and females, although the smaller numbers introduced some discrepancies. For example in women, in whom convergence for A-CHD was obtained only after removing body mass index and blood glucose, the predictive capacities of T-CHD versus A-CHD fatalities were different only for age (p b 0.01), there was no difference for systolic blood pressure and total cholesterol had only a borderline significantly different role (p b 0.0674). Fig. 1 illustrates the 20-year risk of death whereby it is clearly shown that starting at age 55 years there is a dissociation between A-CHD and T-CHD risk curves, the former steeply increasing with age. On the other hand, Fig. 2 shows that the 20-year risk of T-CHD death is a function of serum cholesterol levels, but the reverse is true in the case of A-CHD death. These differences are in contrast with those of systolic blood pressure where trends, differentiating T-CHD versus A-CHD deaths,
302
P.E. Puddu et al. / International Journal of Cardiology 173 (2014) 300–304
Table 1 Descriptive statistics in the baseline Gubbio Population Study to predict typical and atypical cardiovascular fatalities during 20-year follow-up. Variables
N
Mean
Error (¶)
Minimum
Maximum
Range
Gender (female = 0; male = 1) Age (years) Ex smokers (no = 0; yes = 1) Smokers (no = 0; yes = 1) (*) SBP (mm Hg) Body mass index (Kg/m2) Blood glucose (mg/dl) Total cholesterol (mg/dl) HDL cholesterol (mg/dl)
3229 3229 3229 3229 3229 3229 3229 3229 3229
44% men 52.92 0.18 0.36 135.95 27.10 91.50 217.81 47.32
– 13.16 0.007 0.008 23.19 4.20 19.80 42.73 12.20
0 30 0 0 85 15.82 52 88 13
1 79 1 1 255 49.02 351 526 98
1 49 1 1 170 33.20 299 438 85
SBP: systolic blood pressure; HDL: high density lipoprotein. (¶): Standard deviation for continuous variables and standard error for dichotomous variables; (*): No smokers were 46%.
serum cholesterol that sometimes had an inverse relationship with events [12–14], the hypothesis was advanced that A-CHD may represent a distinct disease as compared to T-CHD [14]. The present evidence was obtained in a large residential and independent cohort, the Gubbio Population Study [18–21], whose mortality data were analyzed by the same investigator and by applying the same methods for case definition as those used in the previous investigations with the Seven Countries Study material [12–14]. The important extension was that in the Gubbio Study also women were considered [18–21] whereas the 20-year follow-up of the entire cohort was similarly long as compared to the initial results from the Seven Countries Study [12]. A further important extension was related to systolic blood pressure whose relationship was positive for both A-CHD and T-CHD (Table 2) although a steeper relationship with T-CHD was seen, starting approximately at 140 mm Hg (Fig. 3). External validation is an essential methodological requisite to confer reliability to epidemiological evidence which was systematically explored both in Europe and the United States [15–17]. Based on the present validation and extensions we believe the significant differences seen in Table 2 and Figs. 1–3 for age, total cholesterol and systolic blood pressure in their predictive capacities for T-CHD versus A-CHD fatalities, whereas the remaining covariates considered which had a similar predictive role, reinforce the possibility that T-CHD is a different disease compared to A-CHD. The message from the Gubbio epidemiological cohort, in men and women, is that male gender, age, continued smoke, systolic blood pressure, blood glucose, total cholesterol and HDL-cholesterol (the last being protective) predict fatal CHD with painful symptoms (angina pectoris, acute ischemic attacks, AMI and sudden death). On the other hand, although continued smoke, body mass index and blood glucose also have a role to predict fatal painless CHD (congestive heart failure,
were in any case positive. Moreover, Fig. 3 illustrates the dissociation between 20-year risk of death curves for T-CHD and A-CHD, starting at approximately 140 mm Hg and the steeper increase in the former type, opposite to what observed in case of age. Figs. 1–3 correspond and illustrate Cox's coefficients differences between T-CHD and A-CHD in the entire cohort for age (respectively 0.0814 ± 0.0098 versus 0.1827 ± 0.0132, t = − 6.17, p b 0.0001), serum cholesterol (respectively 0.0057 ± 0.0018 versus − 0.0032 ± 0.0022, t = 3.13, p = 0.0018) and systolic blood pressure (respectively 0.0228 ± 0.0036 versus 0.0110 ± 0.0035, t = 2.31, p = 0.0210) whereas the remaining covariate coefficients were not significantly different. However, by gender, these differences were principally due to men in whom Cox's coefficients in T-CHD versus A-CHD were respectively 0.0611 ± 0.0122 versus 0.1732 ± 0.0199 for age (t = − 4.80, p b 0.0001), 0.0057 ± 0.0023 versus − 0.0050 ± 0.0034 for serum cholesterol (t = 2.57, p = 0.0102), and 0.0299 ± 0.0049 versus 0.0129 ± 0.0062 for systolic blood pressure (t = 2.15, p = 0.0360). Indeed, in women, only Cox's coefficients for age were different between T-CHD and A-CHD (respectively 0.1161 ± 0.0166 versus 0.1932 ± 0.0182, t = − 3.13, p = 0.0018) whereas there was only a borderline significant difference for serum cholesterol (respectively 0.0047 ± 0.0030 versus − 0.0028 ± 0.0028, t = 1.83, p = 0.0674) and no difference for systolic blood pressure (respectively 0.0157 ± 0.0052 versus 0.0129 ± 0.0043, t = 0.42, p = 0.6746). 4. Discussion Our results validate and greatly extend previous studies whereby, based on a stronger association with age for A-CHD versus T-CHD fatalities and the observation that A-CHD deaths were not predicted by
Table 2 Hazard ratios (HR) and 95% confidence intervals (CI) of forced Cox's model solutions to predict typical and atypical cardiovascular fatalities during 20-year follow-up. Males plus females Typical
Covariates Gender Age Ex smokers Smokers SBP Body mass index Blood glucose Total cholesterol HDL cholesterol Parameters Processed Failed Censored Log likelihood
Males Atypical
Females
Typical
Atypical
Atypical (¥)
Typical
Deltas
HR
95% CI
HR
95% CI
HR
95% CI
HR
95% CI
HR
95% CI
HR
95% CI
1 15 1 1 25 4 20 45 12
2.02 3.39 1.06 2.21 1.77 1.04 1.14 1.29 0.75
1.32–3.08 2.54–4.52†† 0.62–1.83 1.43–3.41 1.48–2.11† 0.87–1.24 1.01–1.28 1.10–1.51†† 0.61–0.92
1.39 15.50 1.26 1.75 1.32 1.20 1.19 0.87 0.94
0.90–2.15 10.51–22.84 0.76–2.09 1.09–2.81 1.11–1.57 1.03–1.41 1.08–1.32 0.72–1.05 0.77–1.14
– 2.50 0.93 1.87 2.11 1.15 1.10 1.29 0.85
– 1.75–3.58†† 0.48–1.80 1.01–3.48 1.66–2.69† 0.88–1.48 0.90–1.34 1.05–1.58†† 0.65–1.11
– 13.43 1.33 1.78 1.38 1.39 1.16 0.80 1.02
– 7.49–24.09 0.69–2.58 0.83–3.84 1.02–1.87 1.04–1.86 0.96–1.42 0.59–1.08 0.76–1.37
– 5.46 1.26 2.59 1.42 0.96 1.17 1.27 0.63
– 3.35–8.90†† 0.39–4.11 1.43–4.71 1.09–1.84 0.75–1.23 1.00–1.36 0.97–1.66⁎ 0.46–0.86
– 5.70 1.32 2.72 1.48 – – 0.88 0.80
– 10.64–30.93 0.42–3.18 0.98–3.31 1.12–1.70 – – 0.69–1.13 0.62–1.02
3229 137 3092 −959.13
3229 142 3087 −886.18
1433 76 1357 −481.64
1433 57 1376 −311.58
1796 61 1735 −378.26
1796 85 1711 −483.20
See Table 1 for units. SBP: systolic blood pressure; HDL: high density lipoprotein. When 95% CI cross 1, p N 0.05. (¥): convergence obtained after removing BMI and blood glucose. When testing Cox's model coefficients' differences between typical and atypical solutions ††: p b 0.01; †: p b 0.05; *: p = 0.0674.
P.E. Puddu et al. / International Journal of Cardiology 173 (2014) 300–304
1,00
303
0,16
0,90
0,14 T-CHD A-CHD
0,12
20-year risk of death
20-year risk of death
0,80 0,70 T-CHD A-CHD
0,60 0,50 0,40 0,30
0,10 0,08 0,06 0,04
0,20 0,02
0,10 0,00 30
35
40
45
50
55
60
65
70
75
Age (years)
arrhythmia, blocks and ill-defined chronic CHD, without other specifications, in the absence of AMI, other ischemic attacks, angina pectoris, sudden death or other explicit etiologies), age, systolic blood pressure and total cholesterol differentiate them from T-CHD fatalities. It is in fact of note that age increases the 20-year risk of A-CHD deaths much more than that of T-CHD deaths (Fig. 1), total serum cholesterol has a diverging role (Fig. 2) whereas systolic blood pressure has a positive, not diverging role while having a steeper relationship with T-CHD than with A-CHD (Fig. 3) opposite to the relationship with age. In the Seven Countries Study most recent report [14] there were 13 cohorts of 40–59 year-old men followed up for 40 years (N = 9704 CHD-free subjects) and T-CHD was more common in North American and Northern European countries, while A-CHDs were more common 0,08 0,07
20-year risk of death
0,06 0,05
T-CHD A-CHD
0,04 0,03 0,02 0,01 0,00 180
200
220
240
260
120
130
140
150
160
170
180
190
Systolic blood pressure (mmHg)
Fig. 1. The 20-year risk of death was calculated in overall 3229 individuals enrolled in the Gubbio Population Study according to T-CHD versus A-CHD types (see Methods and methods for definitions' details and Results for Cox's coefficient statistical differences) as a function of age: starting from 55 years of age the 2 curves diverge and the A-CHD one increases exponentially. Note the risk up to approximately 90%.
160
0,00
280
300
320
Total serum cholesterol (mg/dl) Fig. 2. The 20-year risk of death was calculated in overall 3229 individuals enrolled in the Gubbio Population Study according to T-CHD versus A-CHD types (see Methods and methods for definitions' details and Results for Cox's coefficient statistical differences) as a function of total serum cholesterol levels: whereas T-CHD curve increases, the opposite is seen for A-CHD curve. The mathematical counterpart is the opposite sign in the Cox's coefficient for serum cholesterol between T-CHD (positive sign) versus A-CHD (negative sign). Note the risk up to approximately 7–8%.
Fig. 3. The 20-year risk of death was calculated in overall 3229 individuals enrolled in the Gubbio Population Study according to T-CHD versus A-CHD types (see Methods and methods for definitions' details and Results for Cox's coefficient statistical differences) as a function of systolic blood pressure: starting from 140 mm Hg the 2 curves diverge and the T-CHD one increases exponentially. An opposite trend, although always positive, is evident between T-CHD and A-CHD, as compared to what seen with age in Fig. 1. The mathematical counterpart is the opposite sign in the t test value between T-CHD versus A-CHD age (negative sign) versus systolic blood pressure (positive sign). Note the risk up to approximately 16%.
in Southern and Eastern Europe. Age at death was 5 years higher for A-CHD than for T-CHD and Cox model's coefficient for age was significantly larger for A-CHD (HR 2.36, CI 2.18–2.26) versus T-CHD (HR 1.50, CI 1.43–1.58) while coefficients for total serum cholesterol were larger and significant for T-CHD (HR 1.29, CI 1.22–1.35) but not for A-CHD (HR 0.93, CI 0.85–1.03) [14]. On the other hand, smoking habits, forced expiratory volume in 3/4 s and diabetes all predicted both conditions almost equally and similar to what seen with systolic blood pressure whose HR for 20 to 40 years were 1.49 (CI 1.40–1.59) to 1.36 (CI 1.30–1.42) and 1.66 (CI 1.47–1.87) to 1.35 (CI 1.25–1.45) respectively for T-CHD and A-CHD. The evidence from the Gubbio population study on systolic blood pressure is therefore new, although it is unclear how much the different age ranges and gender composition at study run-in between the present investigation (30–79 years, both genders) and the Seven Countries Study material (40–59 years, only men) [12–14] might have contributed. It is of interest to note (from Table 2) that the significant difference seen in the entire Gubbio cohort between T-CHD and A-CHD was confined to men, although the lower death rates among women might have lowered the power for showing significances. There has been little attention in past studies about the possible etiology of what we call A-CHD death, although some important suggestions come from very old contributions. In particular, at least three papers published in the early 1960's dealt with pathological investigations showing that the size of myocardial scars had a bimodal distribution, the large scars being strongly associated with gross atheroma and thrombosis of the large coronary arteries, that was not the case for the small scars [6,7,26]. The T-CHD deaths may have a more clearcut definition, at least from the morphological point of view [4,5,8,9]; however, the differentiation may still be difficult pre-mortem, to the extent that the clinical definitions for these events, either fatal or not, have still some areas of uncertainty notwithstanding the advances of biological tests [1,2]. Apoptosis might be a reasonable hypothesis to explain A-CHD cases [27] as opposed to acute thrombosis after coronary atheromatous dislocation [5,9] for T-CHD deaths or non-fatal events, although these processes may rather indicate pathogenetic steps than etiologic factors [14]. Differences may also depend on large versus
304
P.E. Puddu et al. / International Journal of Cardiology 173 (2014) 300–304
small coronary vessel disease: cholesterol may have a clear-cut role on large coronary vessels whereas systolic blood pressure may have a more definite role in small coronary vessel disease [4–9]. Under these perspectives, the significantly steeper relation of systolic blood pressure with T-CHD versus A-CHD in conjunction with the positive versus no relation due to total cholesterol might indicate a predominant proatherosclerotic versus a pro-arteriosclerotic impact of these joint risk factors. However, epidemiological evidence may not provide a clearcut explanation and may just set the stage for hypotheses whose further test may be worthwhile. A coordinated and larger effort is therefore needed from clinicians, pathologists, geneticists and epidemiologists to identify the etiology and pathogenesis of deaths from chronic congestive heart failure and arrhythmia of uncertain etiology, here A-CHD deaths, in order to plan effective preventive steps to deal with this increasing cause of death, especially seen in Southern and Eastern Europe [14]. The validation that we report here strongly supports that need. Although covariates measured at baseline have been repeatedly demonstrated to keep their predictive capacity from 10 [28] to 25 and more years [29,30], it is quite evident that in relation to CHD death risk, there might be some dilution after 25 years of follow-up [14]. At 40 to 50 years from the first survey, when most if not all individuals initially enrolled are dead [31–33], although the cardiovascular component has still a high burden, all-cause fatalities are rather predicted. Nevertheless, cardiovascular risk factors and respiratory and physical fitness measurements contribute to influence very long-term survival in any case [31–33] so that their role in health status is still quite clear [33]. Future studies may address how risk factor variations over time may or not contribute to further differentiate A-CHD versus T-CHD fatalities. There was a differential predictive capacity between T-CHD and A-CHD deaths in our 20-year follow-up duration analysis; however, high blood pressure, cholesterol, cigarette smoking and overweight were selected also among the 10 major risk factors identified from a world-wide perspective, trying to summarize information collected from the most disparate sources in the group of the developed countries [34,35]. It will be thus important to confirm these observations in further epidemiological material, on both sides of the Atlantic and also in Eastern Countries, which may stress the possibility that these clinical forms of CHD represent indeed different diseases. Funding and conflicts Merck Sharp and Dohme (MSD), Italy helped in partial collection of data reported in the present investigation whereas analysis and preparation of this manuscript were not funded. OTV was a professional consultant of MSD Italy for the Gubbio Population Study until December 2011. AM has received a research grant from MSD, Italy, for the Gubbio Population study, through Cardioricerca srl, Rome and Medrisk srl, Rome. Such grant was not related to the preparation of this manuscript. The authors certify that they comply with the Principles of Ethical Publishing in the International Journal of Cardiology. References [1] Nomenclature and criteria for diagnosis of ischemic heart disease, report of the Joint International Society of Cardiology and Federation of Cardiology/World Health Organization task force on standardization of clinical nomenclature. Circulation 1979;59:607–9. [2] Thygesen K, Alpert JS, Jaffe AS, et al. Third universal definition of myocardial infarction. Eur Heart J 2012;33:2551–67. [3] Menotti A, Verdecchia A, Dima F. The estimate of coronary incidence following different case finding procedures. Eur Heart J 1989;10:568–72. [4] Gallino A, Stuber M, Crea F, et al. “In vivo” imaging of atherosclerosis. Atherosclerosis 2012;224:25–36. [5] Ylä-Herttuala S, Bentzon JF, Daemen M, et al. Stabilization of atherosclerotic plaque: an update. Eur Heart J 2013;34:3251–8.
[6] Schwartz CJ, Mitchell JR. The relation between myocardial lesions and coronary artery disease. I. An unselected necropsy study. Br Heart J 1962;24:761–86. [7] Mitchell JR, Schwartz CJ. The relation between myocardial lesions and coronary artery disease ii. A selected group of patients with massive cardiac necrosis or scarring. Br Heart J 1963;25:1–24. [8] Webber BJ, Seguin PJ, Burnett DG, Clark LL, Otto JL. Prevalence of and risk factors for autopsy-determined atherosclerosis among US service members, 2001–2011. JAMA 2012;308:2577–83. [9] Falk E, Nakano M, Bentzon JF, Finn AV, Virmani R. Update on acute coronary syndromes: the pathologists' view. Eur Heart J 2013;34:719–28. [10] Keys A, Aravanis C, Blackburn H, et al. Seven Countries Study. A multivariate analysis of death and coronary heart disease. Cambridge Mass: Harvard Univ Press; 1980 1–381. [11] Kromhout D, Menotti A, Blackburn H. Prevention of coronary heart disease. Diet, lifestyle and risk factors in the Seven Countries Study. Norwell Ma, USA and Dordrecht, NL: Kluwer Publisher; 2002 1–267. [12] Menotti A, Blackburn H, Seccareccia F, et al. Relationship of some risk factors with typical and atypical manifestations of coronary heart disease. Cardiology 1998;89:59–67. [13] Menotti A, Lanti M, Nedeljkovic S, Nissinen A, Kafatos A, Kromhout D. The relationship of age, blood pressure, serum cholesterol and smoking habits with the risk of typical and atypical coronary heart disease death in the European cohorts of the Seven Countries Study. Int J Cardiol 2006;106:157–63. [14] Menotti A, Puddu PE, Lanti M, et al. Epidemiology of typical coronary heart disease versus heart disease of uncertain etiology (atypical) fatalities and their relationships with classic coronary risk factors. Int J Cardiol 2013;168:3963–7. [15] Menotti A, Puddu PE, Lanti M. Comparison of the Framingham risk function-based coronary chart with risk function from an Italian population study. Eur Heart J 2000;21:365–70. [16] Menotti A, Lanti M, Puddu PE, Kromhout D. Coronary heart disease incidence in northern and southern European populations: a reanalysis of the seven countries study for a European coronary risk chart. Heart 2000;84:238–44. [17] D'Agostino RB, Grundy S, Sullivan LM, Wilson PW. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA 2001;286:180–7. [18] Menotti A, Lanti M, Puddu PE, et al. First risk functions for prediction of coronary and cardiovascular disease incidence in the Gubbio Population Study. Ital Heart J 2000;1:394–9. [19] Puddu PE, Lanti M, Menotti A, et al. Serum uric acid for short-term prediction of cardiovascular disease incidence in the Gubbio Population Study. Acta Cardiol 2001;56:243–51. [20] Puddu PE, Lanti M, Menotti A, et al. Red blood cells count in short-term prediction of cardiovascular disease incidence in the Gubbio Population Study. Acta Cardiol 2002;57:177–85. [21] Menotti A, Lanti M, Angeletti M, et al. Twenty-year cardiovascular and all-cause mortality trends and changes in cardiovascular risk factors in Gubbio, Italy: the role of blood pressure changes. J Hypertens 2009;27:266–74. [22] Rose G, Blackburn H. Cardiovascular survey methods. Geneva: World Health Organization; 1968. [23] Morisi G, Macchia T, Angelico P, Pacioni F, Zucca A. Automatic measurement of triglycerides, cholesterol, glucose and uric acid: perspective use in screening for preventive medicine [in Italian]. Ann Ist Super Sanita 1979;15:239–61. [24] Buongiorno AM, Macchia T, Morisi G, Zucca A. HDL cholesterol: comparison of methods and use: perspectives in prevention of arteriosclerosis [in Italian]. G Ital Chim Clin 1982;7:127–38. [25] WHO International Classification of Diseases (WHO-ICD-9) World Health Organization. International classification of diseases and causes of death. 9th revision. Geneva: WHO; 1975. [26] Anonymous. Myocardial fibrosis. Br Med J 1963;18:1304. [27] Bernecker OY, Huq F, Heiat EK, Podesser BK, Hajjar RJ. Apoptosis in heart failure and the senescent heart. Cardiovasc Toxicol 2003;3:183–90. [28] Menotti A, Lanti M, Puddu PE, et al. The risk functions incorporated in Riscard 2002: a software for the prediction of cardiovascular risk in the general population based on Italian data. Ital Heart J 2002;3:114–21. [29] Menotti A, Lanti M, Puddu PE. Twenty-five-year cardiovascular disease incidence among middle-aged men. Disease burden, time shape, predictors, risk probabilities. Ital Heart J 2000;1:749–57. [30] Puddu PE, Menotti A. Artificial neural network versus multiple logistic function to predict 25-year coronary heart disease mortality in the Seven Countries Study. Eur J Cardiovasc Prev Rehabil 2009;16:583–91. [31] Puddu PE, Menotti A, Tolonen H, Nedeljkovic S, Kafatos A. Determinants of 40-year all-cause mortality in the European cohorts of the Seven Countries Study. Eur J Epidemiol 2011;26:595–608. [32] Puddu PE, Menotti A. Artificial neural networks versus proportional hazards Cox models to predict 45-year all-cause mortality in the Italian Rural Areas of the Seven Countries Study. BMC Res Methodol 2011;12:100. [33] Menotti A, Puddu PE, Lanti M, Maiani G, Fidanza F. Cardiovascular risk factors predict survival in middle-aged men during 50 years. Eur J Intern Med 2013;24:67–74. [34] World Health Organization. The world health report 2002. Reducing risks, promoting healthy life. Geneva: World Health Organization; 2002. [35] Ezzati M, Van der Hoorn S, Lopez AD, et al. Comparative quantification of mortality and burden of disease attributable to selected risk factors. In: Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJL, editors. Global burden of disease and risk factors. Washington DC: Oxford University Press and The World Bank; 2006.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Typical and atypical coronary heart disease deaths and their different relationships with risk factors. The Gubbio residential cohort Study Paolo Emilio Puddu a,⁎, Oscar Terradura Vagnarelli b, Mario Mancini c, Alberto Zanchetti d, Alessandro Menotti e a
Department of Cardiovascular, Respiratory, Nephrological, Anesthesiological and Geriatric Sciences, Sapienza University of Rome, Viale del Policlinico 155, I-00161 Rome, Italy Centre of Preventive Medicine, Gubbio, Italy Department of Clinical Medicine and Surgery, Federico II University, Naples, Italy d Istituto Auxologico Italiano, University of Milan, Milan, Italy e Associazione per la Ricerca Cardiologica, Via Latina 49, I-00179 Rome, Italy b c
a r t i c l e
i n f o
Article history: Received 6 December 2013 Received in revised form 30 January 2014 Accepted 9 March 2014 Available online 15 March 2014 Keywords: Coronary heart disease CHD fatalities Gubbio Cohort Study Cox model Prediction
a b s t r a c t Objectives: The Seven Countries Study showed that fatal coronary heart disease (CHD) with only chronic heart failure, arrhythmia or blocks (atypical CHD, A-CHD) may represent a distinct disease as compared to fatal CHD cases with angina pectoris, acute myocardial infarction (AMI) or sudden death (typical CHD, T-CHD). We aimed at validating this, using identical diagnostic criteria, in a separate residential cohort first examined in 1983–85 in Gubbio, central Italy. Material and methods: Forced Cox's models were run to assess 9 classic risk factors and their 20-year predictivity of A-CHD versus T-CHD, in the entire cohort or separately for men and women. Results: There were 3229 subjects aged 30–79 years. Entry mean age was slightly higher in women than men although age at death was lower in men than in women for both T-CHD (71.99 ± 11.38 versus 81.20 ± 9.35 years, p b 0.0001) and A-CHD (80.22 ± 9.44 versus 84.98 ± 8.13 years, p b 0.0001). T-CHDs were predicted by male gender, age, continued smoke, systolic blood pressure (SBP), blood glucose, total and HDL-cholesterol (protective). A-CHDs were predicted by age, continued smoke, SBP, body mass index and blood glucose but neither total nor HDL-cholesterol or gender was significant. In the entire cohort and in men there were predictive differences of T-CHD versus A-CHD fatalities only in relation to age (p b 0.01), SBP (p b 0.05) and total cholesterol (p b 0.01). Conclusion: As age, SBP and total cholesterol had a different predictive role of T-CHD versus A-CHD fatalities also in the Gubbio cohort, the possibility is reinforced that a different etiology exists between these entities. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Coronary heart disease (CHD) may present clinically either with symptoms (angina pectoris, acute myocardial infarction, AMI, with either ST segment elevation or not, the latter being the modern counterpart of the intermediate syndrome of the past) or with painless manifestations (chronic arrhythmia and blocks and congestive heart failure) whereas sudden death is closer to the first than the second category [1]. The availability of specific biological tests has tremendously increased the diagnostic accuracy of AMI [2]. However,
⁎ Corresponding author at: Laboratory of Biotechnologies Applied to Cardiovascular Medicine, Department of Cardiovascular, Respiratory, Nephrological, Anesthesiological and Geriatric Sciences, Sapienza University of Rome, Viale del Policlinico, 155, Rome 00161, Italy. Tel.: +39 06 49972659; fax: +39 06 4453891. E-mail addresses: [email protected] (P.E. Puddu), [email protected] (O. Terradura Vagnarelli), [email protected] (M. Mancini), [email protected] (A. Zanchetti), [email protected] (A. Menotti).
http://dx.doi.org/10.1016/j.ijcard.2014.03.021 0167-5273/© 2014 Elsevier Ireland Ltd. All rights reserved.
the abovementioned symptomatic and asymptomatic categories may have much less distinct boundaries when diagnostic criteria are applied in longitudinal epidemiological studies [3]. On the other hand, precise boundary definitions among different CHD clinical presentations are not provided by the most updated in vivo imaging modalities [4] specifically directed to coronary plaques [5] nor by post mortem anatomo-pathological studies [6–9]. Reports from the Seven Countries Study, a classic of international cardiovascular epidemiology [10,11], showed that fatal cases where the presentation among 10,628 men initially aged 40–59 years during 40 years of follow-up was only congestive heart failure, chronic arrhythmia, blocks or with the generic mention of chronic CHD (arbitrarily called atypical CHD, A-CHD) were not predicted by serum cholesterol [12–14]. On the opposite, fatal CHD cases manifested with angina pectoris, AMI, acute ischemic attack or sudden death (arbitrarily called typical CHD, T-CHD) showed a strong positive association between serum cholesterol and events. Moreover, there was a stronger association with age for A-CHD than for T-CHD [12–14]. Therefore, the
P.E. Puddu et al. / International Journal of Cardiology 173 (2014) 300–304
hypothesis was raised that A-CHD may represent a distinct disease as compared to T-CHD [14]. The aim of the present study was the external validation of the Seven Countries Study results whereby the differential relationship of A-CHD and T-CHD with age and total cholesterol was obtained [12–14]. It is in fact essential in epidemiological studies to try to repeat in independent cohorts the evidence obtained in one investigation [15–17]. We took advantage of a common thread of shared investigators between the Seven Countries Study [10–16] and the Gubbio Cohort Study [18–21]. We applied identical methods for case definitions and we extended the applicability of the results since we also investigated women (not included in the Seven Countries Study) yet assessing very long follow-up periods. The null hypothesis was that A-CHD and T-CHD deaths share identical risk factors and are predicted indistinctly by those factors entering significantly a forced Cox's model during 20 years of follow-up. 2. Material and methods 2.1. Population, risk factors and mortality data The Gubbio Cohort Study is a population study devoted to cardiovascular diseases conducted in the small town of Gubbio in central Italy. The original sample, first examined in 1983–1985, covered the entire population living within the ancient medieval walls, including both genders and ages 5 to 95 (n = 5376), with a participation rate of 92.2%. For the purpose of this analysis only men and women aged 30–79 were considered, for a total of 3553 subjects. A number of potential risk factors were measured at entry examination, as follows: - gender (coded as 0 and 1 for women and men, respectively); - age in years at the time of the visit; - body mass index (BMI in kg/m2) following the procedures suggested by the WHO Cardiovascular Survey Methods Manual [22] for the measurement of height and weight; - systolic blood pressure in mm Hg, measured by a trained personnel, following the procedure described by the WHO Manual [22]; in particular, participants were in sitting position, a mercury sphygmomanometer was used and the first measurement was taken after a 5 min rest; three measurements were recorded, and the average of the second and third measurement was used for analysis; the physicians taking the measurements were trained and tested using the tape-recorded procedures of the London School of Hygiene; - serum total cholesterol (mg/dl), high-density lipoprotein (HDL) serum cholesterol (mg/dl), and fasting plasma glucose (mg/dl) measured using automated enzymatic methods [23,24]; the laboratory involved in these analyses was under quality control of the WHO Lipid Reference Center of Prague; controls included also duplicate blinded analyses on at least 10% of all laboratory samples; - Cigarette smoking habits were derived from a standard questionnaire and people were classified in 3 dummy variables, i.e. current smokers, ex-smokers and never smokers, the last one used as reference. Mortality data were collected through the end of 2007 with a mean follow-up of about 20 years. Information was obtained from the local register office and review of medical records and interviews with spouses or baystanders and coded by a single reviewer (AM) using the ninth revision of the WHO International Classification of Diseases (WHO-ICD-9) [25]. This was the same person responsible for coding fatalities in the Seven Countries Study [12–14]. In case of uncertainty among apparent multiple causes of death, a hierarchical system was adopted giving precedence to violent causes, cancer in advanced stages, CHD, stroke and other causes in that order. The first cause of death was used for this analysis. The end-point for the analysis was cases of CHD classified into 2 groups: – Typical coronary heart diseases (T-CHDs) were cases coded as AMI, acute coronary ischemic attacks, angina pectoris and sudden death, after reasonable exclusions of other causes; – Atypical coronary heart diseases (A-CHDs) were cases classified as congestive heart failure, arrhythmia, blocks and ill-defined chronic CHD (without other specifications), in the absence of AMI, other ischemic attacks, angina pectoris, sudden death or other explicit etiologies. Subjects who carried CHD (including cases with previous coronary surgery), other forms of heart disease, cerebrovascular disease and peripheral artery disease at entry examination were excluded from the analysis (n = 324: prevalent cases). Among the remaining 3229 subjects there were missing data for some of the risk factors selected for this analysis. Therefore, they were imputed by a multivariate normal procedure using as reference the personal characteristics with full data. This process was based on a regression analysis using the variable containing the missing value as the dependent variable and all variables with non missing data in that row as independent variables. The values of these non missing variables from the row containing the missing values are used in the regression equation to compute the imputed values. The proportion
301
of missing values was relatively small (2.3%) and no difference was found between the means of the original and the imputed values. No significant difference was seen in the variance of the overall sample, compared with the original one. Oral consent was obtained from the participants in compliance with the Helsinki declaration. 2.2. Statistical analysis Data are presented as mean ± standard deviation or standard error (as appropriate) and minimum, maximum and range values. Death rates from T-CHD and A-CHD were computed for overall subjects and separately for men and women, including standardized rates for the two genders based on the overall age distribution. Age at death was computed for the two end-points, for the entire cohort and separately for men and women. Cox proportional hazards models were solved (after assessing the proportionally of hazards) using T-CHD and A-CHD fatalities separately and 9 risk factors (gender, age, smoking and ex smoking habits, systolic blood pressure, body mass index, blood glucose, serum total and HDL cholesterol) as possible predictors adopting the forced methods (all covariates were in the Cox's model at step zero), in order to compute their individual relative contribution. Hazard ratios (HR) and 95% confidence intervals (CI) were calculated using standard formulae. For continuous covariates, deltas corresponded approximately to 1 SD whereas the absolute range was considered for dichotomous covariates. T-tests were adopted to test the difference between coefficients of T-CHD versus A-CHD fatalities. Figures were drawn to show the relationship of age, serum cholesterol and systolic blood pressure with T-CHD and A-CHD deaths. Significance was concluded for p values b0.05.
3. Results Baseline descriptive statistics are shown in Table 1 for 3229 individuals aged 30–79 (mean age 52.92 ± 13.16) years. There were 1433 men (44%) and 1796 women (56%). During approximately 20 years of follow-up T-CHD and A-CHD death rates were respectively 5.30 and 3.98% (p = 0.3009) in men and 3.40 and 4.73% (p = 0.0531) in women. T-CHD rates were higher in men than in women (p = 0.0101) whereas A-CHD rates were similar (p = 0.3440). Heart failure was present in 49.3% of A-CHD deaths either as an explicit cause of death or it was annotated among the symptoms present before death. There was also a significant entry mean age difference in men and women, respectively 51.50 ± 13.06 and 53.85 ± 13.16 years (p b 0.0001) with a significantly lower age at death in men (p b 0001) than in women (p = 0.0014) for both T-CHD (71.99 ± 11.38 versus 81.20 ± 9.35 years, p b 0.0001) and A-CHD (80.22 ± 9.44 versus 84.98 ± 8.13 years, p b 0.0001). Table 2 shows HR and 95% CI of forced Cox's model solutions, to predict T-CHD and A-CHD fatalities in the entire cohort and in males and females. In the entire cohort, T-CHDs were predicted by male gender, age, continued smoke, systolic blood pressure, blood glucose, total cholesterol and HDL-cholesterol (the last being protective); A-CHDs were instead predicted by age, continued smoke, systolic blood pressure, body mass index and blood glucose. Notably, in A-CHD, neither total cholesterol nor HDL-cholesterol or gender had a significant role. There were significant differences for age (p b 0.01), systolic blood pressure (p b 0.05) and total cholesterol (p b 0.01) in their predictive capacities of T-CHD versus A-CHD fatalities whereas the remaining covariates considered had a comparable predictive role for either T-CHD or A-CHD. This picture was confirmed in males and females, although the smaller numbers introduced some discrepancies. For example in women, in whom convergence for A-CHD was obtained only after removing body mass index and blood glucose, the predictive capacities of T-CHD versus A-CHD fatalities were different only for age (p b 0.01), there was no difference for systolic blood pressure and total cholesterol had only a borderline significantly different role (p b 0.0674). Fig. 1 illustrates the 20-year risk of death whereby it is clearly shown that starting at age 55 years there is a dissociation between A-CHD and T-CHD risk curves, the former steeply increasing with age. On the other hand, Fig. 2 shows that the 20-year risk of T-CHD death is a function of serum cholesterol levels, but the reverse is true in the case of A-CHD death. These differences are in contrast with those of systolic blood pressure where trends, differentiating T-CHD versus A-CHD deaths,
302
P.E. Puddu et al. / International Journal of Cardiology 173 (2014) 300–304
Table 1 Descriptive statistics in the baseline Gubbio Population Study to predict typical and atypical cardiovascular fatalities during 20-year follow-up. Variables
N
Mean
Error (¶)
Minimum
Maximum
Range
Gender (female = 0; male = 1) Age (years) Ex smokers (no = 0; yes = 1) Smokers (no = 0; yes = 1) (*) SBP (mm Hg) Body mass index (Kg/m2) Blood glucose (mg/dl) Total cholesterol (mg/dl) HDL cholesterol (mg/dl)
3229 3229 3229 3229 3229 3229 3229 3229 3229
44% men 52.92 0.18 0.36 135.95 27.10 91.50 217.81 47.32
– 13.16 0.007 0.008 23.19 4.20 19.80 42.73 12.20
0 30 0 0 85 15.82 52 88 13
1 79 1 1 255 49.02 351 526 98
1 49 1 1 170 33.20 299 438 85
SBP: systolic blood pressure; HDL: high density lipoprotein. (¶): Standard deviation for continuous variables and standard error for dichotomous variables; (*): No smokers were 46%.
serum cholesterol that sometimes had an inverse relationship with events [12–14], the hypothesis was advanced that A-CHD may represent a distinct disease as compared to T-CHD [14]. The present evidence was obtained in a large residential and independent cohort, the Gubbio Population Study [18–21], whose mortality data were analyzed by the same investigator and by applying the same methods for case definition as those used in the previous investigations with the Seven Countries Study material [12–14]. The important extension was that in the Gubbio Study also women were considered [18–21] whereas the 20-year follow-up of the entire cohort was similarly long as compared to the initial results from the Seven Countries Study [12]. A further important extension was related to systolic blood pressure whose relationship was positive for both A-CHD and T-CHD (Table 2) although a steeper relationship with T-CHD was seen, starting approximately at 140 mm Hg (Fig. 3). External validation is an essential methodological requisite to confer reliability to epidemiological evidence which was systematically explored both in Europe and the United States [15–17]. Based on the present validation and extensions we believe the significant differences seen in Table 2 and Figs. 1–3 for age, total cholesterol and systolic blood pressure in their predictive capacities for T-CHD versus A-CHD fatalities, whereas the remaining covariates considered which had a similar predictive role, reinforce the possibility that T-CHD is a different disease compared to A-CHD. The message from the Gubbio epidemiological cohort, in men and women, is that male gender, age, continued smoke, systolic blood pressure, blood glucose, total cholesterol and HDL-cholesterol (the last being protective) predict fatal CHD with painful symptoms (angina pectoris, acute ischemic attacks, AMI and sudden death). On the other hand, although continued smoke, body mass index and blood glucose also have a role to predict fatal painless CHD (congestive heart failure,
were in any case positive. Moreover, Fig. 3 illustrates the dissociation between 20-year risk of death curves for T-CHD and A-CHD, starting at approximately 140 mm Hg and the steeper increase in the former type, opposite to what observed in case of age. Figs. 1–3 correspond and illustrate Cox's coefficients differences between T-CHD and A-CHD in the entire cohort for age (respectively 0.0814 ± 0.0098 versus 0.1827 ± 0.0132, t = − 6.17, p b 0.0001), serum cholesterol (respectively 0.0057 ± 0.0018 versus − 0.0032 ± 0.0022, t = 3.13, p = 0.0018) and systolic blood pressure (respectively 0.0228 ± 0.0036 versus 0.0110 ± 0.0035, t = 2.31, p = 0.0210) whereas the remaining covariate coefficients were not significantly different. However, by gender, these differences were principally due to men in whom Cox's coefficients in T-CHD versus A-CHD were respectively 0.0611 ± 0.0122 versus 0.1732 ± 0.0199 for age (t = − 4.80, p b 0.0001), 0.0057 ± 0.0023 versus − 0.0050 ± 0.0034 for serum cholesterol (t = 2.57, p = 0.0102), and 0.0299 ± 0.0049 versus 0.0129 ± 0.0062 for systolic blood pressure (t = 2.15, p = 0.0360). Indeed, in women, only Cox's coefficients for age were different between T-CHD and A-CHD (respectively 0.1161 ± 0.0166 versus 0.1932 ± 0.0182, t = − 3.13, p = 0.0018) whereas there was only a borderline significant difference for serum cholesterol (respectively 0.0047 ± 0.0030 versus − 0.0028 ± 0.0028, t = 1.83, p = 0.0674) and no difference for systolic blood pressure (respectively 0.0157 ± 0.0052 versus 0.0129 ± 0.0043, t = 0.42, p = 0.6746). 4. Discussion Our results validate and greatly extend previous studies whereby, based on a stronger association with age for A-CHD versus T-CHD fatalities and the observation that A-CHD deaths were not predicted by
Table 2 Hazard ratios (HR) and 95% confidence intervals (CI) of forced Cox's model solutions to predict typical and atypical cardiovascular fatalities during 20-year follow-up. Males plus females Typical
Covariates Gender Age Ex smokers Smokers SBP Body mass index Blood glucose Total cholesterol HDL cholesterol Parameters Processed Failed Censored Log likelihood
Males Atypical
Females
Typical
Atypical
Atypical (¥)
Typical
Deltas
HR
95% CI
HR
95% CI
HR
95% CI
HR
95% CI
HR
95% CI
HR
95% CI
1 15 1 1 25 4 20 45 12
2.02 3.39 1.06 2.21 1.77 1.04 1.14 1.29 0.75
1.32–3.08 2.54–4.52†† 0.62–1.83 1.43–3.41 1.48–2.11† 0.87–1.24 1.01–1.28 1.10–1.51†† 0.61–0.92
1.39 15.50 1.26 1.75 1.32 1.20 1.19 0.87 0.94
0.90–2.15 10.51–22.84 0.76–2.09 1.09–2.81 1.11–1.57 1.03–1.41 1.08–1.32 0.72–1.05 0.77–1.14
– 2.50 0.93 1.87 2.11 1.15 1.10 1.29 0.85
– 1.75–3.58†† 0.48–1.80 1.01–3.48 1.66–2.69† 0.88–1.48 0.90–1.34 1.05–1.58†† 0.65–1.11
– 13.43 1.33 1.78 1.38 1.39 1.16 0.80 1.02
– 7.49–24.09 0.69–2.58 0.83–3.84 1.02–1.87 1.04–1.86 0.96–1.42 0.59–1.08 0.76–1.37
– 5.46 1.26 2.59 1.42 0.96 1.17 1.27 0.63
– 3.35–8.90†† 0.39–4.11 1.43–4.71 1.09–1.84 0.75–1.23 1.00–1.36 0.97–1.66⁎ 0.46–0.86
– 5.70 1.32 2.72 1.48 – – 0.88 0.80
– 10.64–30.93 0.42–3.18 0.98–3.31 1.12–1.70 – – 0.69–1.13 0.62–1.02
3229 137 3092 −959.13
3229 142 3087 −886.18
1433 76 1357 −481.64
1433 57 1376 −311.58
1796 61 1735 −378.26
1796 85 1711 −483.20
See Table 1 for units. SBP: systolic blood pressure; HDL: high density lipoprotein. When 95% CI cross 1, p N 0.05. (¥): convergence obtained after removing BMI and blood glucose. When testing Cox's model coefficients' differences between typical and atypical solutions ††: p b 0.01; †: p b 0.05; *: p = 0.0674.
P.E. Puddu et al. / International Journal of Cardiology 173 (2014) 300–304
1,00
303
0,16
0,90
0,14 T-CHD A-CHD
0,12
20-year risk of death
20-year risk of death
0,80 0,70 T-CHD A-CHD
0,60 0,50 0,40 0,30
0,10 0,08 0,06 0,04
0,20 0,02
0,10 0,00 30
35
40
45
50
55
60
65
70
75
Age (years)
arrhythmia, blocks and ill-defined chronic CHD, without other specifications, in the absence of AMI, other ischemic attacks, angina pectoris, sudden death or other explicit etiologies), age, systolic blood pressure and total cholesterol differentiate them from T-CHD fatalities. It is in fact of note that age increases the 20-year risk of A-CHD deaths much more than that of T-CHD deaths (Fig. 1), total serum cholesterol has a diverging role (Fig. 2) whereas systolic blood pressure has a positive, not diverging role while having a steeper relationship with T-CHD than with A-CHD (Fig. 3) opposite to the relationship with age. In the Seven Countries Study most recent report [14] there were 13 cohorts of 40–59 year-old men followed up for 40 years (N = 9704 CHD-free subjects) and T-CHD was more common in North American and Northern European countries, while A-CHDs were more common 0,08 0,07
20-year risk of death
0,06 0,05
T-CHD A-CHD
0,04 0,03 0,02 0,01 0,00 180
200
220
240
260
120
130
140
150
160
170
180
190
Systolic blood pressure (mmHg)
Fig. 1. The 20-year risk of death was calculated in overall 3229 individuals enrolled in the Gubbio Population Study according to T-CHD versus A-CHD types (see Methods and methods for definitions' details and Results for Cox's coefficient statistical differences) as a function of age: starting from 55 years of age the 2 curves diverge and the A-CHD one increases exponentially. Note the risk up to approximately 90%.
160
0,00
280
300
320
Total serum cholesterol (mg/dl) Fig. 2. The 20-year risk of death was calculated in overall 3229 individuals enrolled in the Gubbio Population Study according to T-CHD versus A-CHD types (see Methods and methods for definitions' details and Results for Cox's coefficient statistical differences) as a function of total serum cholesterol levels: whereas T-CHD curve increases, the opposite is seen for A-CHD curve. The mathematical counterpart is the opposite sign in the Cox's coefficient for serum cholesterol between T-CHD (positive sign) versus A-CHD (negative sign). Note the risk up to approximately 7–8%.
Fig. 3. The 20-year risk of death was calculated in overall 3229 individuals enrolled in the Gubbio Population Study according to T-CHD versus A-CHD types (see Methods and methods for definitions' details and Results for Cox's coefficient statistical differences) as a function of systolic blood pressure: starting from 140 mm Hg the 2 curves diverge and the T-CHD one increases exponentially. An opposite trend, although always positive, is evident between T-CHD and A-CHD, as compared to what seen with age in Fig. 1. The mathematical counterpart is the opposite sign in the t test value between T-CHD versus A-CHD age (negative sign) versus systolic blood pressure (positive sign). Note the risk up to approximately 16%.
in Southern and Eastern Europe. Age at death was 5 years higher for A-CHD than for T-CHD and Cox model's coefficient for age was significantly larger for A-CHD (HR 2.36, CI 2.18–2.26) versus T-CHD (HR 1.50, CI 1.43–1.58) while coefficients for total serum cholesterol were larger and significant for T-CHD (HR 1.29, CI 1.22–1.35) but not for A-CHD (HR 0.93, CI 0.85–1.03) [14]. On the other hand, smoking habits, forced expiratory volume in 3/4 s and diabetes all predicted both conditions almost equally and similar to what seen with systolic blood pressure whose HR for 20 to 40 years were 1.49 (CI 1.40–1.59) to 1.36 (CI 1.30–1.42) and 1.66 (CI 1.47–1.87) to 1.35 (CI 1.25–1.45) respectively for T-CHD and A-CHD. The evidence from the Gubbio population study on systolic blood pressure is therefore new, although it is unclear how much the different age ranges and gender composition at study run-in between the present investigation (30–79 years, both genders) and the Seven Countries Study material (40–59 years, only men) [12–14] might have contributed. It is of interest to note (from Table 2) that the significant difference seen in the entire Gubbio cohort between T-CHD and A-CHD was confined to men, although the lower death rates among women might have lowered the power for showing significances. There has been little attention in past studies about the possible etiology of what we call A-CHD death, although some important suggestions come from very old contributions. In particular, at least three papers published in the early 1960's dealt with pathological investigations showing that the size of myocardial scars had a bimodal distribution, the large scars being strongly associated with gross atheroma and thrombosis of the large coronary arteries, that was not the case for the small scars [6,7,26]. The T-CHD deaths may have a more clearcut definition, at least from the morphological point of view [4,5,8,9]; however, the differentiation may still be difficult pre-mortem, to the extent that the clinical definitions for these events, either fatal or not, have still some areas of uncertainty notwithstanding the advances of biological tests [1,2]. Apoptosis might be a reasonable hypothesis to explain A-CHD cases [27] as opposed to acute thrombosis after coronary atheromatous dislocation [5,9] for T-CHD deaths or non-fatal events, although these processes may rather indicate pathogenetic steps than etiologic factors [14]. Differences may also depend on large versus
304
P.E. Puddu et al. / International Journal of Cardiology 173 (2014) 300–304
small coronary vessel disease: cholesterol may have a clear-cut role on large coronary vessels whereas systolic blood pressure may have a more definite role in small coronary vessel disease [4–9]. Under these perspectives, the significantly steeper relation of systolic blood pressure with T-CHD versus A-CHD in conjunction with the positive versus no relation due to total cholesterol might indicate a predominant proatherosclerotic versus a pro-arteriosclerotic impact of these joint risk factors. However, epidemiological evidence may not provide a clearcut explanation and may just set the stage for hypotheses whose further test may be worthwhile. A coordinated and larger effort is therefore needed from clinicians, pathologists, geneticists and epidemiologists to identify the etiology and pathogenesis of deaths from chronic congestive heart failure and arrhythmia of uncertain etiology, here A-CHD deaths, in order to plan effective preventive steps to deal with this increasing cause of death, especially seen in Southern and Eastern Europe [14]. The validation that we report here strongly supports that need. Although covariates measured at baseline have been repeatedly demonstrated to keep their predictive capacity from 10 [28] to 25 and more years [29,30], it is quite evident that in relation to CHD death risk, there might be some dilution after 25 years of follow-up [14]. At 40 to 50 years from the first survey, when most if not all individuals initially enrolled are dead [31–33], although the cardiovascular component has still a high burden, all-cause fatalities are rather predicted. Nevertheless, cardiovascular risk factors and respiratory and physical fitness measurements contribute to influence very long-term survival in any case [31–33] so that their role in health status is still quite clear [33]. Future studies may address how risk factor variations over time may or not contribute to further differentiate A-CHD versus T-CHD fatalities. There was a differential predictive capacity between T-CHD and A-CHD deaths in our 20-year follow-up duration analysis; however, high blood pressure, cholesterol, cigarette smoking and overweight were selected also among the 10 major risk factors identified from a world-wide perspective, trying to summarize information collected from the most disparate sources in the group of the developed countries [34,35]. It will be thus important to confirm these observations in further epidemiological material, on both sides of the Atlantic and also in Eastern Countries, which may stress the possibility that these clinical forms of CHD represent indeed different diseases. Funding and conflicts Merck Sharp and Dohme (MSD), Italy helped in partial collection of data reported in the present investigation whereas analysis and preparation of this manuscript were not funded. OTV was a professional consultant of MSD Italy for the Gubbio Population Study until December 2011. AM has received a research grant from MSD, Italy, for the Gubbio Population study, through Cardioricerca srl, Rome and Medrisk srl, Rome. Such grant was not related to the preparation of this manuscript. The authors certify that they comply with the Principles of Ethical Publishing in the International Journal of Cardiology. References [1] Nomenclature and criteria for diagnosis of ischemic heart disease, report of the Joint International Society of Cardiology and Federation of Cardiology/World Health Organization task force on standardization of clinical nomenclature. Circulation 1979;59:607–9. [2] Thygesen K, Alpert JS, Jaffe AS, et al. Third universal definition of myocardial infarction. Eur Heart J 2012;33:2551–67. [3] Menotti A, Verdecchia A, Dima F. The estimate of coronary incidence following different case finding procedures. Eur Heart J 1989;10:568–72. [4] Gallino A, Stuber M, Crea F, et al. “In vivo” imaging of atherosclerosis. Atherosclerosis 2012;224:25–36. [5] Ylä-Herttuala S, Bentzon JF, Daemen M, et al. Stabilization of atherosclerotic plaque: an update. Eur Heart J 2013;34:3251–8.
[6] Schwartz CJ, Mitchell JR. The relation between myocardial lesions and coronary artery disease. I. An unselected necropsy study. Br Heart J 1962;24:761–86. [7] Mitchell JR, Schwartz CJ. The relation between myocardial lesions and coronary artery disease ii. A selected group of patients with massive cardiac necrosis or scarring. Br Heart J 1963;25:1–24. [8] Webber BJ, Seguin PJ, Burnett DG, Clark LL, Otto JL. Prevalence of and risk factors for autopsy-determined atherosclerosis among US service members, 2001–2011. JAMA 2012;308:2577–83. [9] Falk E, Nakano M, Bentzon JF, Finn AV, Virmani R. Update on acute coronary syndromes: the pathologists' view. Eur Heart J 2013;34:719–28. [10] Keys A, Aravanis C, Blackburn H, et al. Seven Countries Study. A multivariate analysis of death and coronary heart disease. Cambridge Mass: Harvard Univ Press; 1980 1–381. [11] Kromhout D, Menotti A, Blackburn H. Prevention of coronary heart disease. Diet, lifestyle and risk factors in the Seven Countries Study. Norwell Ma, USA and Dordrecht, NL: Kluwer Publisher; 2002 1–267. [12] Menotti A, Blackburn H, Seccareccia F, et al. Relationship of some risk factors with typical and atypical manifestations of coronary heart disease. Cardiology 1998;89:59–67. [13] Menotti A, Lanti M, Nedeljkovic S, Nissinen A, Kafatos A, Kromhout D. The relationship of age, blood pressure, serum cholesterol and smoking habits with the risk of typical and atypical coronary heart disease death in the European cohorts of the Seven Countries Study. Int J Cardiol 2006;106:157–63. [14] Menotti A, Puddu PE, Lanti M, et al. Epidemiology of typical coronary heart disease versus heart disease of uncertain etiology (atypical) fatalities and their relationships with classic coronary risk factors. Int J Cardiol 2013;168:3963–7. [15] Menotti A, Puddu PE, Lanti M. Comparison of the Framingham risk function-based coronary chart with risk function from an Italian population study. Eur Heart J 2000;21:365–70. [16] Menotti A, Lanti M, Puddu PE, Kromhout D. Coronary heart disease incidence in northern and southern European populations: a reanalysis of the seven countries study for a European coronary risk chart. Heart 2000;84:238–44. [17] D'Agostino RB, Grundy S, Sullivan LM, Wilson PW. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA 2001;286:180–7. [18] Menotti A, Lanti M, Puddu PE, et al. First risk functions for prediction of coronary and cardiovascular disease incidence in the Gubbio Population Study. Ital Heart J 2000;1:394–9. [19] Puddu PE, Lanti M, Menotti A, et al. Serum uric acid for short-term prediction of cardiovascular disease incidence in the Gubbio Population Study. Acta Cardiol 2001;56:243–51. [20] Puddu PE, Lanti M, Menotti A, et al. Red blood cells count in short-term prediction of cardiovascular disease incidence in the Gubbio Population Study. Acta Cardiol 2002;57:177–85. [21] Menotti A, Lanti M, Angeletti M, et al. Twenty-year cardiovascular and all-cause mortality trends and changes in cardiovascular risk factors in Gubbio, Italy: the role of blood pressure changes. J Hypertens 2009;27:266–74. [22] Rose G, Blackburn H. Cardiovascular survey methods. Geneva: World Health Organization; 1968. [23] Morisi G, Macchia T, Angelico P, Pacioni F, Zucca A. Automatic measurement of triglycerides, cholesterol, glucose and uric acid: perspective use in screening for preventive medicine [in Italian]. Ann Ist Super Sanita 1979;15:239–61. [24] Buongiorno AM, Macchia T, Morisi G, Zucca A. HDL cholesterol: comparison of methods and use: perspectives in prevention of arteriosclerosis [in Italian]. G Ital Chim Clin 1982;7:127–38. [25] WHO International Classification of Diseases (WHO-ICD-9) World Health Organization. International classification of diseases and causes of death. 9th revision. Geneva: WHO; 1975. [26] Anonymous. Myocardial fibrosis. Br Med J 1963;18:1304. [27] Bernecker OY, Huq F, Heiat EK, Podesser BK, Hajjar RJ. Apoptosis in heart failure and the senescent heart. Cardiovasc Toxicol 2003;3:183–90. [28] Menotti A, Lanti M, Puddu PE, et al. The risk functions incorporated in Riscard 2002: a software for the prediction of cardiovascular risk in the general population based on Italian data. Ital Heart J 2002;3:114–21. [29] Menotti A, Lanti M, Puddu PE. Twenty-five-year cardiovascular disease incidence among middle-aged men. Disease burden, time shape, predictors, risk probabilities. Ital Heart J 2000;1:749–57. [30] Puddu PE, Menotti A. Artificial neural network versus multiple logistic function to predict 25-year coronary heart disease mortality in the Seven Countries Study. Eur J Cardiovasc Prev Rehabil 2009;16:583–91. [31] Puddu PE, Menotti A, Tolonen H, Nedeljkovic S, Kafatos A. Determinants of 40-year all-cause mortality in the European cohorts of the Seven Countries Study. Eur J Epidemiol 2011;26:595–608. [32] Puddu PE, Menotti A. Artificial neural networks versus proportional hazards Cox models to predict 45-year all-cause mortality in the Italian Rural Areas of the Seven Countries Study. BMC Res Methodol 2011;12:100. [33] Menotti A, Puddu PE, Lanti M, Maiani G, Fidanza F. Cardiovascular risk factors predict survival in middle-aged men during 50 years. Eur J Intern Med 2013;24:67–74. [34] World Health Organization. The world health report 2002. Reducing risks, promoting healthy life. Geneva: World Health Organization; 2002. [35] Ezzati M, Van der Hoorn S, Lopez AD, et al. Comparative quantification of mortality and burden of disease attributable to selected risk factors. In: Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJL, editors. Global burden of disease and risk factors. Washington DC: Oxford University Press and The World Bank; 2006.
