Download Fatty liver index and mortality: The cremona study in the 15th year of

Fatty Liver Index and Mortality: The Cremona Study
in the 15th Year of Follow-Up
Giliola Calori,1 Guido Lattuada,1,4 Francesca Ragogna,1 Maria Paola Garancini,2 Paolo Crosignani,5
Marco Villa,6 Emanuele Bosi,1 Giacomo Ruotolo,1,7 Lorenzo Piemonti,3 and Gianluca Perseghin1,4
A fatty liver, which is a common feature in insulin-resistant states, can lead to chronic liver
disease. It has been hypothesized that a fatty liver can also increase the rates of non–hepaticrelated morbidity and mortality. Therefore, we wanted to determine whether the fatty liver
index (FLI), a surrogate marker and a validated algorithm derived from the serum triglyceride level, body mass index, waist circumference, and c-glutamyltransferase level, was associated with the prognosis in a population study. The 15-year all-cause, hepatic-related,
cardiovascular disease (CVD), and cancer mortality rates were obtained through the Regional Health Registry in 2011 for 2074 Caucasian middle-aged individuals in the Cremona
study, a population study examining the prevalence of diabetes mellitus in Italy. During the
15-year observation period, 495 deaths were registered: 34 were hepatic-related, 221 were
CVD-related, 180 were cancer-related, and 60 were attributed to other causes. FLI was independently associated with the hepatic-related deaths (hazard ratio 5 1.04, 95% confidence
interval 5 1.02-1.05, P < 0.0001). Age, sex, FLI, cigarette smoking, and diabetes were independently associated with all-cause mortality. Age, sex, FLI, systolic blood pressure, and
fibrinogen were independently associated with CVD mortality; meanwhile, age, sex, FLI,
and smoking were independently associated with cancer mortality. FLI correlated with the
homeostasis model assessment of insulin resistance (HOMA-IR), a surrogate marker of insulin resistance (Spearman’s q 5 0.57, P < 0.0001), and when HOMA-IR was included in the
multivariate analyses, FLI retained its association with hepatic-related mortality but not
with all-cause, CVD, and cancer-related mortality. Conclusion: FLI is independently associated with hepatic-related mortality. It is also associated with all-cause, CVD, and cancer
mortality rates, but these associations appear to be tightly interconnected with the risk conferred by the correlated insulin-resistant state. (HEPATOLOGY 2011;54:145-152)
See Editorial on Page 6
N
onalcoholic fatty liver disease (NAFLD) is
common in insulin-resistant subjects1 and
affects 20% to 30% of the adult population
and more than 50% of overweight and obese individuals.2
NAFLD is associated with an increased risk of developing
advanced fibrosis and cirrhosis3 and incident type 2 diabetes.4 Because of its association with metabolic syndrome
and type 2 diabetes, it has been hypothesized that
NAFLD may also be associated with increased rates of cardiovascular disease (CVD)5; in particular, patients with
NAFLD have elevated levels of plasma biomarkers of
Abbreviations: BMI, body mass index; CI, confidence interval; CVD, cardiovascular disease; FLI, fatty liver index; GGT, c-glutamyltransferase; HOMA-IR, homeostasis model
assessment of insulin resistance; HR, hazard ratio; IGT, impaired glucose tolerance; NAFLD, nonalcoholic fatty liver disease; OGTT, oral glucose tolerance test.
From the 1Division of Metabolic and Cardiovascular Sciences, 2Medical Direction, and 3Diabetes Research Institute, H. San Raffaele Scientific Institute, Milan,
Italy; 4Department of Sport, Nutrition, and Health Sciences, University of Milan, Milan, Italy; 5National Cancer Institute, Milan, Italy; 6Epidemiology Service, Local
Health Authority of the Province of Cremona, Cremona, Italy; and 7AstraZeneca R&D, Molndal, Sweden.
Received January 17, 2011; accepted April 1, 2011.
This work was supported by a liberal donation from the family of Angela Musazzi and Mario Stellato.
Giliola Calori contributed to the study concept and design, the acquisition of the data, the analysis and interpretation of the data, the critical revision of the
article for important intellectual content, and the statistical analysis. Guido Lattuada and Francesca Ragogna contributed to the analysis and interpretation of the
data and the critical revision of the article for important intellectual content. Maria Paola Garancini contributed to the study concept and design, the acquisition
of the data, and the analysis and interpretation of the data. Paolo Crosignani and Marco Villa contributed to the acquisition of the data and the critical revision
of the article for important intellectual content. Emanuele Bosi contributed to the critical revision of the manuscript for important intellectual content. Giacomo
Ruotolo and Lorenzo Piemonti contributed to the study concept and design, the analysis and interpretation of the data, and the critical revision of the article for
important intellectual content. Gianluca Perseghin contributed to the study concept and design, the acquisition of the data, the analysis and interpretation of the
data, the drafting of the article, and the acquisition of funding.
145
146
CALORI ET AL.
inflammation, endothelial dysfunction, markers of subclinical cardiovascular risk, and a higher prevalence of clinically manifesting CVD.6 Some studies have also reported
a higher incidence of major outcomes,7-10 such as nonfatal
CVD events,7 deaths due to CVD,8,9 revascularization
procedures,9 and all-cause mortality.10 The data were
obtained from community-based cohorts,7,10 nested casecontrol studies,8,9 the general population,7,10 or patients
with type 2 diabetes8,9; NAFLD was established with abdominal ultrasonography,7,10 liver biopsy,8,9 or c-glutamyltransferase (GGT) levels.10 These studies had a maximum follow-up of 7.3 years.
The fatty liver index (FLI) is a surrogate marker of
a fatty liver. It was validated in a large group of subjects with or without suspected liver disease11 and was
associated with coronary heart disease and early atherosclerosis in a cross-sectional study.12
The aim of this study was to assess the relationship
between FLI and hepatic-related, all-cause, CVD, and
cancer mortality rates in the Cremona study, a 19901991 population survey designed to establish the prevalence of diabetes in Lombardy, Italy.13,14 In this cohort,
the vital status and the time of death were ascertained
through Regional Health Registry files, and the causes of
death were classified with the International Classification
of Diseases. The follow-up was 15 years, which is the
shortest time needed to explore the effects of metabolic
risk factors (diabetes status, insulin resistance, body
fatness, and metabolic syndrome) on mortality.15,16
Patients and Methods
Study Cohort and Follow-Up. The Cremona study
was a large population survey in the health district of
Cremona (38,643 inhabitants from three representative
municipalities: Cremona, Casalbuttano, and Vescovato)
that was performed (1) to estimate the prevalence of
diagnosed and undiagnosed diabetes and impaired glucose tolerance (IGT) according to the oral glucose
tolerance test (OGTT) and the World Health Organization criteria in Northern Italy and (2) to set up a
population cohort in anticipation of a follow-up
study.13,14 After an information campaign, 2074 randomly selected subjects who were more than 40 years
old visited one of the three clinics set up for this program (one in each town), and data for 2011 of these
HEPATOLOGY, July 2011
subjects were available for this analysis. The anthropometric and laboratory features are summarized in Table 1.
The subjects’ medical histories, anthropometric parameters, and clinical data were collected according to a
standard protocol by trained interviewers. Venous
blood samples were collected after 12 hours of overnight fasting and 2 hours after the oral administration
of 75 g of glucose monohydrate. Further details about
the study protocol have been reported previously.13,14
Fifteen years later, the vital status and the time of
death were ascertained through Regional Health Registry files, and the causes of death were classified with
the International Classification of Diseases, Ninth Revision (death codes 401-448 for CVD, death codes
140.0-208.9 for cancer, death codes 155-156 for hepatic diseases related to hepatocarcinoma, and death
codes 571-573 for hepatic diseases related to cirrhosis).
The median follow-up was 180 months, and the median follow-up for those still alive was 182 months
(98% of those who were still alive had a minimum
follow-up period of 174 months). Data for 2011 of
the 2074 individuals were available.
Definition of Diabetes, IGT, and Metabolic
Syndrome. Diabetes was defined by the use of oral
hypoglycemic agents or insulin and according to the
World Health Organization diagnostic criteria for the
OGTT (basal plasma glucose level >7.8 mmol/L or
>11.1 mmol/L after a 2-hour oral glucose load). Patients
with manifest diabetes did not undergo the OGTT. IGT
was defined as a basal plasma glucose level <7.8 mmol/L
and a plasma glucose level >7.8 mmol/L but <11
mmol/L after a 2-hour oral glucose load. Metabolic syndrome was defined according to the National Cholesterol
Education Program Adult Treatment Program III criteria.
Analytical Determinations. Blood, serum, and
plasma substrates were assessed as previously described.13,14
Calculations. The body mass index (BMI) was calculated as the weight (kg) divided by the square of the
height (m2). Alcohol consumption was calculated as
grams of alcohol (20 g for a glass of wine, 30 g for an
aperitif, and 80 g for liquor). The homeostasis model
assessment of insulin resistance (HOMA-IR) score was
calculated as previously described,17 and low-density
lipoprotein cholesterol levels were calculated with the
Friedwald formula. FLI was calculated according to a
previously published report by Bedogni et al.11:
Address reprint requests to: Gianluca Perseghin, M.D., Division of Metabolic and Cardiovascular Sciences, H. San Raffaele Scientific Institute, Via Olgettina
60, Milan, Italy 20132. E-mail: [email protected]; fax: 0039-02-26432771.
C 2011 by the American Association for the Study of Liver Diseases.
Copyright V
View this article online at wileyonlinelibrary.com.
DOI 10.1002/hep.24356
Potential conflict of interest: Nothing to report.
HEPATOLOGY, Vol. 54, No. 1, 2011
CALORI ET AL.
147
Table 1. Anthropometric and Laboratory Parameters of the Cohort
Anthropometric parameters
Age (years)
BMI (kg/m2)
Waist circumference (cm)*
Smoking [n (%)]*
Alcohol consumption (g/day)*
Systolic blood pressure (mm Hg)
Diastolic blood pressure (mm Hg)
Heart rate (beats/minute)
Biochemical laboratory parameters
Glucose (mmol/L)
Cholesterol (mmol/L)
High-density lipoprotein cholesterol (mmol/L)
Triglycerides (mmol/L)*
Alanine aminotransferase (U/L)*
Aspartate aminotransferase (U/L)*
GGT (U/L)*
Alkaline phosphatase (U/L)
Fibrinogen (mg/dL)
Hormones
Insulin (pmol/L)
Other parameters
HOMA-IR
Metabolic syndrome [n (%)]
Diabetes [n (%)]
FLI*
Women (n 5 1126)
Men (n 5 885)
All (n 5 2011)
58 6 11
26.3 6 4.8
85 6 12
170 (14.6)
17 6 26
144 6 22
79 6 12
76 6 12
57 6 11
26.7 6 3.8
95 6 10
290 (31.8)
68 6 71
146 6 20
81 6 12
72 6 13
57 6 11
26.5 6 4.4
89 6 12
460 (22.2)
42 6 58
145 6 21
80 6 12
75 6 12
5.25 6 1.13
6.20 6 1.16
1.46 6 0.36
1.29 6 0.65
21 6 14
25 6 10
29 6 52
180 6 75
297 6 74
5.39 6 1.06
5.91 6 1.06
1.25 6 0.39
1.63 6 1.28
30 6 26
30 6 17
53 6 67
174 6 60
268 6 70
5.31 6 1.10
6.07 6 1.12
1.37 6 0.39
1.44 6 0.99
25 6 20
27 6 14
39 6 61
177 6 69
284 6 73
94 6 61
100 6 83
97 6 72
3.86 6 3.58
393 (34)
109 (9.4)
32 6 26
4.19 6 4.27
313 (34)
87 (9.5)
52 6 26
4.00 6 3.90
706 (34)
196 (9.5)
41 6 28
Unless otherwise indicated, the data are presented as means and standard deviations.
*There was a significant difference between men and women (P < 0.05 according to an independent t test).
FLI ¼ e0:953loge
Triglycerides þ0:139 BMI þ0:718loge GGT þ0:053 Waist circumference 15:745
ð1 þ e
=
0:953loge Triglycerides þ0:139 BMI þ0:718loge GGT þ0:053 Waist circumference 15:745
Statistical Analysis. Analyses were performed with
SAS software (version 9.1). Concentrations are presented
as means and standard deviations unless otherwise stated.
Because of the skewed distributions of serum insulin, triglycerides, fibrinogen, and glucose, log-transformed values were used in the analysis. The association of each
investigated risk factors with all-cause, CVD, cancer, and
hepatic-related mortality rates after the 15-year observation period were estimated with a Cox proportional hazards model with adjustments for age and sex. Hazard
ratios (HRs) and 95% confidence intervals (CIs) are presented. A multivariate Cox proportional model (stepwise), which included parameters with P values <0.1 in
the univariate analysis, was used to investigate the independent association of the risk factors with all-cause,
CVD, cancer, and hepatic-related mortality rates.
Results
Anthropometric and Laboratory Characteristics of
the Study Subjects (Table 1). The population consisted of overweight individuals; 22.2% of the study
Þ 100
subjects were active smokers, and they had higher than
normal systolic blood pressures and total cholesterol
levels. Metabolic syndrome was detected in 34% of
the population, and diabetes was detected in 9.5%.
FLI was significantly higher in men versus women
(P < 0.0001; Table 1). It was also significantly higher
in individuals with type 2 diabetes and IGT versus
individuals with normal glucose tolerance (55 6 28
versus 38 6 27, P < 0.0001).
Hepatic-Related Mortality During the 15-Year
Observation Period. Tables 2 and 3 summarize the
results for hepatic-related mortality. During the 15-year
observation period, 34 hepatic-related deaths were
recorded. Table 2 summarizes the results of the univariate
analysis, and Table 3 summarizes the results of the multivariate analysis. Age, sex, low-density lipoprotein cholesterol, and FLI were all independently associated with allcause mortality. The inclusion of HOMA-IR in the
multivariate analysis did not change the outcome. When
the FLI factors were tested individually in the multivariate model in place of FLI, BMI, waist circumference,
and GGT were associated with hepatic-related mortality.
148
CALORI ET AL.
HEPATOLOGY, July 2011
Table 2. Age- and Sex-Adjusted HRs Associated With 15-Year
Hepatic-Related Mortality By Univariate Analysis
Table 4. Age- and Sex-Adjusted HRs Associated With 15-Year
All-Cause Mortality By Univariate Analysis
Variable
HR
95% CI
P Value
Variable
HR
95% CI
BMI (kg/m2)
Waist circumference (cm)
Fasting glucose (mmol/L)
Fasting insulin (pmol/L)
Diabetes
HOMA-IR
Smoking
Alcohol consumption (g/day)
Fibrinogen (mg/dL)
Low-density lipoprotein
cholesterol (mmol/L)
High-density lipoprotein
cholesterol (mmol/L)
Triglycerides (mmol/L)
GGT (U/L)
Systolic blood pressure (mm Hg)
Diastolic blood pressure (mm Hg)
Heart rate (beats/minute)
FLI
1.128
1.054
1.013
1.036
2.643
1.116
1.071
1.001
0.994
0.987
1.051-1.210
1.024-1.085
1.002-1.024
1.025-1.048
1.172-5.957
1.078-1.156
0.692-1.657
0.996-1.005
0.988-1.000
0.978-0.997
0.0008
0.0004
<0.0203
<0.0001
<0.0191
<0.0001
0.76
0.77
0.0436
0.0084
1.02
1.01
1.01
1.02
1.06
1.74
1.11
1.77
1.00
1.00
0.99
1.00-1.04
0.99-1.02
1.01-1.01
1.01-1.02
1.04-1.07
1.39-2.18
1.02-1.22
1.42-2.20
0.99-1.00
1.00-1.00
0.99-0.99
0.0528
0.0766
<0.0001
<0.0001
<0.0001
<0.0001
0.0187
<0.0001
0.30
0.021
0.880
0.989
0.965-1.013
0.35
1.00
0.99-1.00
0.323
1.002
1.004
1.011
1.035
1.028
1.036
1.001-1.004
1.003-1.006
0.995-1.028
1.009-1.060
1.005-1.052
1.021-1.052
0.001
<0.0001
0.167
0.007
0.0193
<0.0001
BMI (kg/m2)
Waist circumference (cm)
Fasting glucose (mmol/L)
Fasting insulin (pmol/L)
HOMA-IR
Diabetes
Metabolic syndrome
Smoking
Alcohol consumption (g/day)
Fibrinogen (mg/dL)
Low-density lipoprotein
cholesterol (mmol/L)
High-density lipoprotein
cholesterol (mmol/L)
Triglycerides (mmol/L)
GGT (U/L)
Systolic blood pressure (mm Hg)
Diastolic blood pressure (mm Hg)
Heart rate (beats/minute)
FLI
1.00
1.001
1.00
1.00
1.01
1.006
1.00-1.00
1.000-1.002
1.00-1.01
1.00-1.01
1.01-1.02
1.002-1.009
0.0295
0.004
0.131
0.417
0.0019
0.0013
All-Cause Mortality During the 15-Year Observation Period. Tables 4 and 5 summarize the results for
all-cause mortality. During the 15-year observation period, 495 deaths were registered. Table 4 summarizes
the results of the univariate analysis, and Table 5 summarizes the results of the multivariate analysis. Age,
sex, cigarette smoking, diabetes, and FLI were all independently associated with all-cause mortality. When
HOMA-IR was included in the multivariate analysis,
FLI did not retain its independent association. When
the FLI factors were tested individually in the multivariate model in place of FLI, only GGT was associated with all-cause mortality.
CVD Mortality During the 15-Year Observation
Period. Tables 6 and 7 summarize the results for
CVD mortality. During the 15-year observation peTable 3. Cox Proportional Hazards Model (Stepwise
Selection) of the Predictors of 15-Year Hepatic-Related
Mortality By Multivariate Analysis
Variable
HR
95% CI
P Value
Age (years)
Sex
Low-density lipoprotein
cholesterol (mmol/L)
FLI
1.079
0.406
0.981
1.041-1.118
0.178-0.926
0.971-0.991
<0.0001
0.032
0.0003
1.037
1.022-1.053
<0.0001
The summary report related to the multivariate analysis was performed with
all the variables significant at P < 0.1 in the univariate analysis (with the
exclusion of BMI, waist circumference, and serum triglycerides). Only variables
remaining significantly associated are shown. When the FLI factors were individually tested in the multivariate models in place of FLI, the following factors were
associated with hepatic-related mortality: BMI (HR ¼ 1.14, 95% CI ¼ 1.061.21, P < 0.001), waist circumference (HR ¼ 1.06, 95% CI ¼ 1.03-1.09, P ¼
0.0002), and GGT (HR ¼ 1.005, 95% CI ¼ 1.003-1.006, P < 0.0001).
P Value
riod, 221 CVD-related events were registered. Table 6
summarizes the results of the univariate analysis, and
Table 7 summarizes the results of the multivariate
analysis. Age, sex, systolic blood pressure, fibrinogen,
and FLI were independently associated with CVD
mortality. When HOMA-IR was included in the multivariate analysis, FLI did not retain its independent
association. When the FLI factors were tested individually in the multivariate model in place of FLI, only
BMI was associated with CVD mortality.
Cancer Mortality During the 15-Year Observation
Period. Tables 8 and 9 summarize the results for cancer mortality. During the 15-year observation period,
180 cancer-related events were registered. Table 8 summarizes the results of the univariate analysis, and Table
9 summarizes the results of the multivariate analysis.
Table 5. Cox Proportional Hazards Model (Stepwise
Selection) of the Predictors of 15-Year All-Cause Mortality
By Multivariate Analysis
Variable
HR
95% CI
P Value
Age (years)
Sex
Diabetes
FLI
Cigarette smoking
1.106
0.508
1.688
1.004
1.139
1.095-1.116
0.406-0.635
1.338-2.129
1.001-1.007
1.009-1.286
<0.0001
<0.0001
<0.0001
0.0238
0.0354
The summary report related to the multivariate analysis was performed with
all the variables significant at P < 0.1 in the univariate analysis (with the
exclusion of BMI, waist circumference, serum triglycerides, and HOMA-IR). Only
variables remaining significantly associated are shown. When the FLI factors
were individually tested in the multivariate models in place of FLI, the following
factor was associated with all-cause mortality: GGT (HR ¼ 1.001, 95% CI ¼
1.000-1.002, P ¼ 0.046).
HEPATOLOGY, Vol. 54, No. 1, 2011
CALORI ET AL.
Table 6. Age- and Sex-Adjusted HRs Associated With 15-Year
CVD Mortality by Univariate Analysis
149
Table 8. Age- and Sex-Adjusted HRs Associated With 15-Year
Cancer Mortality by Univariate Analysis
P Value
Variable
HR
95% CI
P Value
1.01-1.07
1.00-1.02
1.01-1.02
1.01-1.02
1.04-1.08
1.37-2.59
1.09-1.42
1.02-2.09
0.99-1.00
1.00-1.00
0.99-1.00
0.0193
0.0842
<0.0001
<0.0001
<0.0001
0.0001
0.0011
0.0377
0.652
0.0115
0.191
1.014
1.006
1.007
1.019
1.003
1.606
1.013
2.214
1.002
1.001
0.997
0.980-1.048
0.993-1.019
1.001-1.013
1.011-1.027
1.002-1.005
1.092-2.361
0.880-1.166
1.549-2.912
1.000-1.004
0.999-1.003
0.994-1.001
0.422
0.385
0.015
<0.0001
0.0001
0.016
0.856
<0.0001
0.020
0.322
0.179
0.99
0.98-0.99
0.0506
1.001
0.991-1.010
0.896
1.00
1.001
1.01
1.01
1.02
1.007
1.00-1.00
1.000-1.002
1.01-1.02
1.00-1.02
1.01-1.03
1.002-1.012
0.391
0.21
0.0002
0.0862
0.0033
0.0078
BMI (kg/m2)
Waist circumference (cm)
Fasting glucose (mmol/L)
Fasting insulin (pmol/L)
HOMA-IR
Diabetes
Metabolic syndrome
Smoking
Alcohol consumption (g/day)
Fibrinogen (mg/dL)
Low-density lipoprotein
cholesterol (mmol/L)
High-density lipoprotein
cholesterol (mmol/L)
Triglycerides (mmol/L)
GGT (U/L)
Systolic blood pressure (mm Hg)
Diastolic blood pressure (mm Hg)
Heart rate (beats/minute)
FLI
1.000
1.002
0.998
1.003
1.005
1.006
0.999-1.002
1.001-1.003
0.991-1.005
0.992-1.015
0.994-1.016
1.000-1.011
0.658
0.0002
0.572
0.336
0.403
0.0321
Variable
HR
95% CI
BMI (kg/m2)
Waist circumference (cm)
Fasting glucose (mmol/L)
Fasting insulin (pmol/L)
HOMA-IR
Diabetes
Metabolic syndrome
Smoking
Alcohol consumption (g/day)
Fibrinogen (mg/dL)
Low-density lipoprotein
cholesterol (mmol/L)
High-density lipoprotein
cholesterol (mmol/L)
Triglycerides (mmol/L)
GGT (U/L)
Systolic blood pressure (mm Hg)
Diastolic blood pressure (mm Hg)
Heart rate (beats/minute)
FLI
1.04
1.01
1.01
1.02
1.06
1.88
1.25
1.46
1.00
1.00
0.99
Age, sex, cigarette smoking, and FLI were independently associated with cancer mortality. When HOMAIR was included in the multivariate analysis, FLI did
not retain its independent association. When the FLI
factors were tested individually in the multivariate
model in place of FLI, only GGT was associated with
cancer mortality.
Correlation Between FLI, HOMA-IR, and
Markers of Low-Grade Inflammation. FLI was associated with the surrogate marker of insulin resistance
(HOMA-IR; Spearman’s q ¼ 0.57, P < 0.0001) for
the entire population. The relationship was detectable
regardless of the diabetes status. In fact, FLI was associated with HOMA-IR in individuals with normal glucose tolerance (Spearman’s q ¼ 0.54, P < 0.0001) and
in patients with IGT and type 2 diabetes (Spearman’s
q ¼ 0.57, P < 0.0001).
Table 7. Cox Proportional Hazards Model (Stepwise
Selection) of the Predictors of 15-Year CVD Mortality by
Multivariate Analysis
Variable
HR
95% CI
P Value
Age (years)
Sex
Systolic blood pressure (mm Hg)
Fibrinogen (mg/dL)
FLI
1.138
0.401
1.010
1.002
1.006
1.120-1.157
0.302-0.534
1.004-1.017
1.000-1.004
1.000-1.011
<0.0001
<0.0001
<0.0013
0.0173
0.0387
The summary report related to the multivariate analysis was performed with
all the variables significant at P < 0.1 in the univariate analysis (with the
exclusion of BMI, waist circumference, and HOMA-IR). Only variables remaining
significantly associated are shown. When the FLI factors were individually tested
in the multivariate models in place of FLI, the following factor was associated
with CVD mortality: BMI (HR ¼ 1.033, 95% CI ¼ 1.001-1.066, P ¼ 0.046).
FLI was associated with fibrinogen (Spearman’s q ¼
0.06, P ¼ 0.007) as a surrogate marker of low-grade
inflammation. To corroborate this association, we
looked for other parameters; in previous studies, measurements of surrogate markers of low-grade inflammation were obtained for subgroups of individuals within
the Cremona study. In particular, monocyte chemoattractant protein 1/chemokine (C-C motif ) ligand 2,
tumor necrosis factor a soluble receptor II, and leptin
were measured in a subgroup of 363 individuals to establish their relationship with CVD mortality during a
follow-up period of 7 years.18 FLI was associated with
tumor necrosis factor a soluble receptor II (Spearman’s
q ¼ 0.14, P < 0.011) and with leptin (Spearman’s
q ¼ 0.38, P < 0.0001) but not with monocyte chemoattractant protein 1/chemokine (C-C motif ) ligand
2 (Spearman’s q ¼ 0.011, P ¼ 0.86). The levels of
high-sensitivity C-reactive protein were measured in a
Table 9. Cox Proportional Hazards Model (Stepwise
Selection) of the Predictors of 15-Year Cancer Mortality by
Multivariate Analysis
Variable
HR
95% CI
P Value
Age (years)
Sex
Cigarette smoking
FLI
1.069
0.557
1.259
1.006
1.054-1.084
0.389-0.797
1.041-1.522
1.001-1.011
<0.0001
<0.0014
<0.0175
<0.0304
The summary report was performed with all the variables significant at P <
0.1 in the univariate analysis. Only variables remaining significantly associated
are shown (with the exclusion of HOMA-IR). When the FLI factors were individually tested in the multivariate models in place of FLI, the following factor was
associated with cancer mortality: GGT (HR ¼ 1.002, 95% CI ¼ 1.001-1.003,
P < 0.001).
150
CALORI ET AL.
subgroup of 447 elderly subjects (65 years old) to
establish its interaction with the lipid profile in the
Cremona study19 and was associated with FLI (Spearman’s q ¼ 0.29, P < 0.0001).
Discussion
This work demonstrates an association between FLI
and all-cause mortality in middle-aged individuals. FLI
was associated not only with hepatic-related mortality
but also with CVD and nonhepatic cancer mortality
independently of the diabetes/IGT status and metabolic syndrome. For the first time, this surrogate
marker was validated as a predictor of all-cause mortality in a population study; moreover, for the first time,
an association between NAFLD and mortality rates
was established during a 15-year follow-up period.
Hepatic-related mortality was independent of the concomitant insulin-resistant state; in contrast, CVD, cancer, and all-cause mortality rates were tightly related to
the concomitant insulin-resistant state estimated with
HOMA-IR.
Comparison With Previous Epidemiological
Studies. CVD and cancer were the two most common
causes of death in the Cremona population, and
chronic liver disease (cirrhosis and hepatocellular carcinoma in particular) accounted for 7% of all deaths.
FLI was associated with an absolute reduction of the
survival rate. This finding agrees with the data generated by the population study of Olmsted County,
MN, in which NAFLD was associated with reduced
survival in the general population with a follow-up
period of 7 to 8 years20 and in people with type 2 diabetes with a follow-up period of 11 years.21 Also, the
finding that FLI was associated with hepatic-related
mortality (a combination of hepatocellular carcinoma–
related mortality and cirrhosis-related mortality) is in
agreement with the Olmsted County study, which also
reported higher hepatic-related mortality among people
with NAFLD20 with 7 to 8 years of follow-up. The
association between FLI and CVD mortality is also in
agreement with the reports discussed in the introduction and recently reviewed by Targher et al.,6 who
summarized the robust evidence supporting the link
between NAFLD and CVD in the literature. In all
these studies, fatty livers were established through
histological findings, ultrasonography, or surrogate
markers such as alanine aminotransferase or GGT levels with different CVD endpoints (nonfatal CVD
events, deaths from CVD, revascularization procedures,
and all-cause mortality), but the maximum length of
these studies was 7 to 8 years. In our study, FLI was
HEPATOLOGY, July 2011
independently associated with CVD mortality after a
15-year observation period.
These data, therefore, support the view and join the
growing body of evidence suggesting that CVD may
be the major cause of death among individuals with
NAFLD. In our opinion, because FLI was also independently associated with cancer mortality regardless
of hepatic-related mortality, we should not forgot that
these data also suggest a link between NAFLD and
mortality due to malignancies, which was also reported
in diabetic patients of the Olmsted County study (but
only as a trend).21
In our opinion, it is important to emphasize that in
our study, FLI was more reliable than the glucose tolerance status (based on the OGTT procedure), metabolic syndrome (based on the Adult Treatment Panel
III definition), and aminotransferases.
Mechanistic Interpretation. The pathogenesis of
diabetes and NAFLD are closely related to insulin
resistance and hyperinsulinemia.22 Consequently, we
wanted to determine whether the association of FLI
with mortality rates was independent of a surrogate
index of insulin resistance, HOMA-IR, which is based
on the product of the fasting plasma insulin concentration and is frequently used in population studies.23
HOMA-IR has been reported to be associated with
CVD mortality in several population studies,24-27 and
we observed that HOMA-IR was also associated with
cancer mortality in the population of the Cremona
study (G. Perseghin, MD, G. Calori, MD, G. Lattuada, PhD, F. Ragogna, PhD, E. Dugnani, PhD, M.
P. Garancini, MD, P. Crosignani, MD, M. Villa, MD,
E. Bosi, MD, G. Ruotolo, MD, L. Piemonti, MD,
unpublished data, 2011). In this set of data, FLI maintained its independent association with hepatic-related
mortality; this suggests a specificity of this index that
goes beyond the potential effects of other metabolic
variables expressed by HOMA-IR. In contrast, FLI did
not retain an independent association with CVD and
cancer mortality rates when HOMA-IR was included
in the analysis. The surrogate marker of insulin resistance appeared to be more important than FLI, and
this suggests the primacy of the systemic insulin-resistant state over the fatty liver, which is the hepatic component of metabolic syndrome. We feel that it is very
difficult to determine the roles of HOMA-IR and FLI;
from a statistical point of view, the correlation between
the two variables is considerable, and from a pathophysiological point of view, it is well known that
NAFLD not only may be a marker of insulin resistance syndrome but also may be involved in its pathogenesis. In particular, it was hypothesized for CVD
HEPATOLOGY, Vol. 54, No. 1, 2011
that the pathogenic process may be mediated through
the systemic release of pro-atherogenic mediators from
the inflamed liver to peripheral tissues. Following this
line of thinking, we noticed that FLI was correlated
with surrogate markers of low-grade inflammation,
such as fibrinogen, high-sensitivity C-reactive protein,
and tumor necrosis factor a soluble receptor II.
Another mechanism tightly linking insulin resistance
to the fatty liver is the recently reported alteration of
hepatic adenosine triphosphate metabolism, which
affects patients with type 2 diabetes and has been associated with the severity of the intrahepatic fat
accumulation.28
Strengths and Limitations. The major strengths of
the present study are as follows: (1) its population,
which includes both males and females; (2) its careful
and homogeneous acquisition of the anthropometric
parameter of interest; and (3) most importantly, its
long follow-up period (15 years).
Its limitations are as follows: (1) its lack of intermediate data points for the parameters of interest during the
15-year observation period; (2) its lack of data about
dietary habits and habitual physical activity, which have
a well-recognized impact on insulin sensitivity and a
fatty liver; and (3) the inferiority of the homeostasis
model assessment versus the glucose clamp technique,
which is the gold standard for the assessment of insulin
sensitivity. Nevertheless, it has been suggested that the
homeostasis model assessment is specifically suited for
large-scale epidemiological studies when only fasting
glucose and insulin concentrations are available.23
In conclusion, FLI as a surrogate marker of NAFLD
is associated with hepatic-related, CVD, and cancer
mortality rates regardless of the diabetic status, fasting
glucose concentration, or metabolic syndrome. This
provides epidemiological support for the hypothesis that
NAFLD is an important and independent factor affecting not only the hepatic prognosis but also the nonhepatic prognosis of affected people. The tight association
of FLI with HOMA-IR makes it difficult for us at this
stage to understand the primacy of NAFLD versus systemic insulin resistance in explaining the strong association of fatty livers with all-cause mortality.
References
1. Marchesini G, Brizi M, Bianchi G, Tomassetti S, Bugianesi E, Lenzi
M, et al. Nonalcoholic fatty liver disease: a feature of the metabolic
syndrome. Diabetes 2001;50:1844-1850.
2. Angulo P. Nonalcoholic fatty liver disease. N Engl J Med 2002;346:
1221-1231.
3. Ruhl CE, Everhart JE. Epidemiology of nonalcoholic fatty liver. Clin
Liver Dis 2004;8:501-519.
CALORI ET AL.
151
4. Sattar N, McConnachie A, Ford I, Gaw A, Cleland SJ, Forouhi NG,
et al. Serial metabolic measurements and conversion to type 2 diabetes
in the West of Scotland Coronary Prevention Study: specific elevations
in alanine aminotransferase and triglycerides suggest hepatic fat accumulation as a potential contributing factor. Diabetes 2007;56:984-991.
5. Perseghin G. The role of non alcoholic fatty liver disease in cardiovascular disease. Dig Dis 2010;28:210-213.
6. Targher G, Day CP, Bonora E. Risk of cardiovascular disease in patients
with nonalcoholic fatty liver disease. N Engl J Med 2010;363:1341-1350.
7. Hamaguchi M, Kojima T, Takeda N, Nagata C, Takeda J, Sarui H,
et al. Nonalcoholic fatty liver disease is a novel predictor of cardiovascular disease. World J Gastroenterol 2007;13:1579-1584.
8. Targher G, Bertolini L, Poli F, Rodella S, Scala L, Tessari R, et al.
Nonalcoholic fatty liver disease and risk of future cardiovascular events
among type 2 diabetic patients. Diabetes 2005;54:3541-3546.
9. Targher G, Bertolini L, Rodella S, Tessari R, Zenari L, Lippi G, et al.
Nonalcoholic fatty liver disease is independently associated with an
increased incidence of cardiovascular events in type 2 diabetic patients.
Diabetes Care 2007;30:2119-2121.
10. Haring R, Wallaschofski H, Nauck M, Dörr M, Baumeister SE, Völzke
H. Ultrasonographic hepatic steatosis increases prediction of mortality risk
from elevated serum gamma-glutamyl-transpeptidase levels. HEPATOLOGY
2009;50:1403-1411.
11. Bedogni G, Bellentani S, Miglioli L, Masutti F, Passalacqua M, Castiglione A, et al. The fatty liver index: a simple and accurate predictor of
hepatic steatosis in the general population. BMC Gastroenterol 2006;6:33.
12. Gastaldelli A, Kozakova M, Højlund K, Flyvbjerg A, Favuzzi A, Mitrakou A, et al. Fatty liver is associated with insulin resistance, risk of coronary heart disease, and early atherosclerosis in a large European
population. HEPATOLOGY 2009;49:1537-1544.
13. Garancini MP, Calori G, Manara E, Izzo A, Ebbli E, Galli L, et al. An
Italian population-based study of the prevalence of diabetes: some
methodological aspects. Diabete Metab 1993;19:116-120.
14. Garancini MP, Calori G, Ruotolo G, Manara E, Izzo A, Ebbli E, et al.
Prevalence of NIDDM and impaired glucose tolerance in Italy: a
OGTT-based population study. Diabetologia 1995;38:306-313.
15. Arnlov J, Ingelsson E, Sundstrom J, Lind L. Impact of body mass index
and the metabolic syndrome on the risk of cardiovascular disease and
death in middle-aged men. Circulation 2010;121:230-236.
16. Sundstrom J, Riserus U, Byberg L, Zethelius B, Lithell H, Lind L.
Clinical value of the metabolic syndrome for long term prediction of
total and cardiovascular mortality: prospective, population based cohort
study. BMJ 2010;332:878-882.
17. Perseghin G, Caumo A, Caloni M, Testolin G, Luzi L. Incorporation
of the fasting plasma FFA concentration into QUICKI improves its
association with insulin sensitivity in nonobese individuals. J Clin
Endocrinol Metab 2001;86:4776-4781.
18. Piemonti L, Calori G, Mercalli A, Lattuada G, Monti P, Garancini MP,
et al. Fasting plasma leptin, monocyte chemoattracting protein-1
(MCP-1/CCL2) and a-tumor necrosis factor receptor 2 (a-TNFR-2)
concentrations in glucose tolerant and intolerant women. Impact on
cardiovascular mortality. Diabetes Care 2003;26:2883-2889.
19. D’Angelo A, Ruotolo G, Garancini P, Sampietro F, Mazzola G, Calori
G. Lipoprotein(a), fibrinogen and vascular mortality in an elderly
northern Italian population. Haematologica 2006;91:1613-1620.
20. Adams LA, Lymp JF, St Sauver J, Sanderson SO, Lindor KD, Feldstein
A, et al. The natural history of nonalcoholic fatty liver disease: a population-based cohort study. Gastroenterology 2005;129:113-121.
21. Adams LA, Harmsen S, St Sauver JL, Charatcharoenwitthaya P, Enders
FB, Therneau T, et al. Nonalcoholic fatty liver disease increases risk of
death among patients with diabetes: a community-based cohort study.
Am J Gastroenterol 2010;105:1567-1573.
22. Marchesini G, Brizi M, Morselli-Labate AM, Bianchi G, Bugianesi E,
McCullough AJ, et al. Association of nonalcoholic fatty liver disease
with insulin resistance. Am J Med 1999;107:450-455.
23. Bonora E, Targher G, Alberiche M, Bonadonna RC, Saggiani F, Zenere
MB, et al. Homeostasis model assessment closely mirrors the glucose
152
CALORI ET AL.
clamp technique in the assessment of insulin sensitivity. Diabetes Care
2000;23:57-63.
24. Bonora E, Kiechl S, Willeit J, Oberhollenzer F, Egger G, Meigs JB,
et al. Insulin resistance as estimated by homeostasis model assessment
predicts incident symptomatic cardiovascular disease in Caucasian subjects from the general population: the Bruneck study. Diabetes Care
2007;30:318-324.
25. Hedblad B, Nilsson P, Engström G, Berglund G, Janzon L. Insulin
resistance in non-diabetic subjects is associated with increased incidence of myocardial infarction and death. Diabet Med 2002;19:
470-475.
HEPATOLOGY, July 2011
26. Hanley AJG, Williams K, Stern MP, Haffner SM. Homeostasis model
assessment of insulin resistance in relation to the incidence of cardiovascular disease: the San Antonio Heart Study. Diabetes Care 2002;25:
1177-1184.
27. Nakamura K, Sakurai M, Miura K, Morikawa Y, Ishizaki M, Yoshita
K, et al. Homeostasis model assessment of insulin resistance and the
risk of cardiovascular events in middle-aged non-diabetic Japanese men.
Diabetologia 2010;53:1894-1902.
28. Schmid AI, Szendroedi J, Chmelik M, Krssák M, Moser E, Roden M.
Liver ATP synthesis is lower and relates to insulin sensitivity in patients
with type 2 diabetes. Diabetes Care 2011;34:448-453.