Int. J. Epidemiol. Advance Access published May 20, 2015 International Journal of Epidemiology, 2015, 1–10 doi: 10.1093/ije/dyv086 Original article
Original article
Urinary cadmium and mortality from all causes, cancer and cardiovascular disease in the general population: systematic review and meta-analysis of cohort studies Downloaded from http://ije.oxfordjournals.org/ at University of Manitoba on June 10, 2015
Susanna C Larsson* and Alicja Wolk Unit of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden *Corresponding author. Unit of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, PO Box 210, SE-171 77, Stockholm, Sweden. E-mail: [email protected]. Accepted 27 April 2015
Abstract Background: Cadmium is a toxic heavy metal distributed in the environment. We conducted a systematic review and meta-analysis to examine the association between urinary cadmium concentration and mortality from all causes, cancer and cardiovascular disease (CVD) in the general population. Methods: Studies were identified by searching PubMed and Embase (to 30 March 2015) and the reference lists of retrieved articles. We included prospective studies that reported hazard ratios (HR) with 95% confidence intervals (CI) for the association between urinary cadmium concentration and all-cause, cancer or CVD mortality. A random-effects model was used to combine study-specific results. Results: Nine cohort studies, including 5600 deaths from all causes, 1332 deaths from cancer and 1715 deaths from CVD, were eligible for inclusion in the meta-analysis. The overall HRs for the highest vs lowest category of urinary cadmium were1.44 (95% CI, 1.25–1.64; I2 ¼ 40.5%) for all-cause mortality (six studies), 1.39 (95% CI, 0.96–1.99; I2 ¼ 75.9%) for cancer mortality (four studies) and 1.57 (95% CI, 1.27–1.95; I2 ¼ 34.0%) for CVD mortality (five studies). In an analysis restricted to six cohort studies conducted in populations with a mean urinary cadmium concentration of 1 mg/g creatinine, the HRs were 1.38 (95% CI, 1.17–1.63; I2 ¼ 48.3%) for all-cause mortality, 1.56 (95% CI, 0.98–2.47; I2 ¼ 81.0%) for cancer mortality and 1.50 (95% CI, 1.18–1.91; I2 ¼ 38.2%) for CVD mortality. Conclusions: Even at low-level exposure, cadmium appears to be associated with increased mortality. Further large prospective studies of cadmium exposure and mortality are warranted. Key words: Cadmium, cancer, meta-analysis mortality, prospective studies
C The Author 2015; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association V
1
2
International Journal of Epidemiology, 2015, Vol. 0, No. 0
Key Messages • High cadmium exposure was associated with increased mortality, with the strongest and most consistent association
with CVD mortality. • There was a suggestive positive association between urinary cadmium concentration and cancer mortality, but the
number of studies was few and results were inconsistent. • The positive association between urinary cadmium concentration and mortality was observed in populations with
low-level cadmium exposure.
Introduction
Methods Search strategy and inclusion criteria Studies were identified by searching PubMed and Embase (1966 to 30 March 2015), without restrictions, using the search terms cadmiumANDmortality.We also searched the reference lists of retrieved articles to find additional studies. Eligible for inclusion in this meta-analysis were cohort studies that reported hazard ratios (HR) or relative risk estimates with 95% confidence interval (CI) for the association of urinary cadmium concentrations with all-cause,
Data extraction and quality assessment The data extracted included the first author’s last name, year of publication, country, duration of follow-up, sex and age of participants, number of deaths, cut-offs for the highest vs lowest category of urinary cadmium concentration, variables adjusted for in the multivariable model, and the most fully adjusted HRs with 95% CIs for the highest vs lowest category of cadmium concentrations. One of the authors (S.C.L.) extracted the data from each study. We used the Newcastle-Ottawa Scale (NOS) for quality assessment of the studies.20 We used pre-defined criteria: selection (population representativeness); comparability (adjustment for confounders); and outcome (ascertainment of outcome and follow-up). For the adjustment for confounders, studies were awarded one point if adjustment was made for age and sex and two points if further adjustments were made for smoking and body mass index or body weight. The NOS ranged from 0 to 9, with 9 points reflecting the highest study quality.
Statistical analysis Study-specific HRs were combined using a random-effects model, which considers both within- and between-study variability.21 When results were presented separately for men and women, we combined the sex-specific HRs, using a random-effects model and included the overall estimate in the meta-analysis. We conducted a dose-response metaanalysis of urinary cadmium concentrations and all-cause mortality using the same method as described in a previous dose-response meta-analysis.22 Statistical heterogeneity among studies was assessed with the Q and I2 statistics.23 We conducted a sensitivity analysis excluding cohort studies of participants living in cadmium-polluted areas and cohorts of highly exposed participants. We performed stratified analyses by sex and age range of the study
Downloaded from http://ije.oxfordjournals.org/ at University of Manitoba on June 10, 2015
Cadmium is a heavy metal widely distributed in the environment. It accumulates in the body, especially in the kidneys, and has an elimination half-time of 10–30 years.1 The general population is exposed to cadmium mainly through inhalation of cigarette smoke and intake of food.2 Cadmium is primarily present in foods that are otherwise considered to be healthy, such as whole-grain foods, potatoes and vegetables. Those foods contribute to 60–70% of total cadmium intake in the European and US populations.3,4 The presence of cadmium in whole grains and vegetables may counteract the potential health-beneficial substances in those foods, such dietary fibre, essential nutrients, antioxidants and phytochemicals. Besides being a carcinogen,5 cadmium exposure is associated with kidney disease,6,7 osteoporosis and bone fractures,8,9 atherosclerosis10–12 and cardiovascular disease (CVD).13–15 Epidemiological studies on low-level cadmium exposure in relation to risk of cancer and CVD are limited and have generally been based on small sample sizes, a case-control or cross-sectional design, or dietary cadmium intake as a measure of cadmium exposure.14–19 To further examine the health effects of cadmium exposure, we conducted a systematic review and metaanalysis to summarize available evidence from prospective studies assessing the associations of urinary cadmium concentration with all-cause, cancer and CVD mortality in the general population.
cancer or CVD mortality. If results for the same mortality endpoint were reported in more than one article, we included the study with the most comprehensive data.
International Journal of Epidemiology, 2015, Vol. 0, No. 0
3
Records identified by searching PubMed (n=673) and Embase (n=1115) Duplicate records excluded (n=535) Records remaining after removal of duplicates (n=1253) Records excluded based on title or abstract (n=1236) Records remaining after title and abstract review (n=17)
Articles included in meta-analysis (n=9) • All-cause mortality: 6 • Cancer mortality: 4 • Cardiovascular disease mortality: 5 Figure 1. Flow-chart of study selection.
population, country, duration of follow-up, and study quality, and assessed heterogeneity across subgroups through meta-regression. Publication bias was evaluated using Egger’s test.24 Stata (version 12.0, StataCorp, College Station, TX) was used for the statistical analyses.
Results Literature search Our literature search identified 1253 non-duplicate records of which 17 remained after title and abstract review (Figure 1). Among the records obtained for full-text review, we excluded four because the exposure was not urinary cadmium25–27 or no HR was reported.28 We further excluded four studies29–32 due to duplicate publications from the same study population and same follow-up period.
Characteristics of the included studies Table 1 shows the characteristics of the nine included studies.33–41 Results were presented for all-cause mortality
in six studies (including 5600 deaths), for cancer mortality in four studies (1332 deaths) and for CVD mortality in five studies (1715 deaths). All studies included both men and women. Five studies were conducted in the USA, one in Belgium and three in Japan. The geometric mean urine cadmium concentration (in mg/g creatinine) ranged from about 0.3 to 1.0 in the US and Belgian populations, with higher concentrations in women than in men. In the Japanese studies, urine cadmium concentration (in mg/g creatinine) ranged from 1.8 in men and 2.4 in women from non-polluted areas to about 4–6 in men and 7–8 in women from a cadmium-polluted area. The studies by Nakagawa et al.33 and Li et al.36 were based on the same Japanese population but the mortality end-points were different. Likewise, the studies by Menke et al.35 and Lin et al.38 were based on the same US population. Menke et al.35 reported results for all-cause, cancer and CVD mortality during follow-up from 1988–94 through 2000. Lin et al.38 reported results for cancer mortality only and the follow-up was from 1988–94 through 2006. We included results for all-cause and CVD mortality from the publication by Menke et al.35 and results for cancer from Lin et al.38
Downloaded from http://ije.oxfordjournals.org/ at University of Manitoba on June 10, 2015
Records excluded (n=8): • Exposure not urinary cadmium (n=3) • No hazard ratios reported (n=1) • Duplicate publications from the same cohort and same follow-up period (n=4)
Japan, NA
Belgium, CadmiBel
Nakagawa, 200633
Nawrot, 200834
women, 20
4.8
USA, NHANES Men and
Tellez-Plaza,
(1999–2004)
Men and 22.0 women, 50
Japan, NA
Li, 201136
201237
Men and 9.0 women, 20
Menke, 200935 USA, NHANES III (1988–94)
Men and 20.3 women, 20
Men and 15.0 women, 50
Country, study Sex, age (years) Followname up (years)
First author, year
8989
3119
13 958
944
3119
Subjects
7
9
8
7
Study quality
CVD
524 all causes ,191 8
334 cancer, 267 CVD
1690 all causes, 769 CVD
201 all causes, 54 cancer, 88 CVD
948 all causes
No. of deaths
–
1.68 (1.09–2.58) [men]; 1.14 (0.78–1.66) [women]
(1.71–6.53) [women] 1.20 (1.04–1.39)
1.32 (0.61–2.84) [men]; 3.11
All-cause
1.1 (0.7–1.9) [men]; 0.8 (0.4–1.6) [women] –
–
1.8 (1.0–3.1) [men]; 2.4 (1.1–5.1) [women] 1.74 (1.07–2.83)
1.33 (0.69–2.56) [men]; 0.82 (0.47–1.42) [women]
1.43 (1.08–1.89)
–
1.07 (0.85–1.34)
–
CVD
Cancer
HR (95% CI) of mortality
> 0.57 vs 0.14 (0.28) 1.52 (1.00–2.29)
20 vs
Original article
Urinary cadmium and mortality from all causes, cancer and cardiovascular disease in the general population: systematic review and meta-analysis of cohort studies Downloaded from http://ije.oxfordjournals.org/ at University of Manitoba on June 10, 2015
Susanna C Larsson* and Alicja Wolk Unit of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden *Corresponding author. Unit of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, PO Box 210, SE-171 77, Stockholm, Sweden. E-mail: [email protected]. Accepted 27 April 2015
Abstract Background: Cadmium is a toxic heavy metal distributed in the environment. We conducted a systematic review and meta-analysis to examine the association between urinary cadmium concentration and mortality from all causes, cancer and cardiovascular disease (CVD) in the general population. Methods: Studies were identified by searching PubMed and Embase (to 30 March 2015) and the reference lists of retrieved articles. We included prospective studies that reported hazard ratios (HR) with 95% confidence intervals (CI) for the association between urinary cadmium concentration and all-cause, cancer or CVD mortality. A random-effects model was used to combine study-specific results. Results: Nine cohort studies, including 5600 deaths from all causes, 1332 deaths from cancer and 1715 deaths from CVD, were eligible for inclusion in the meta-analysis. The overall HRs for the highest vs lowest category of urinary cadmium were1.44 (95% CI, 1.25–1.64; I2 ¼ 40.5%) for all-cause mortality (six studies), 1.39 (95% CI, 0.96–1.99; I2 ¼ 75.9%) for cancer mortality (four studies) and 1.57 (95% CI, 1.27–1.95; I2 ¼ 34.0%) for CVD mortality (five studies). In an analysis restricted to six cohort studies conducted in populations with a mean urinary cadmium concentration of 1 mg/g creatinine, the HRs were 1.38 (95% CI, 1.17–1.63; I2 ¼ 48.3%) for all-cause mortality, 1.56 (95% CI, 0.98–2.47; I2 ¼ 81.0%) for cancer mortality and 1.50 (95% CI, 1.18–1.91; I2 ¼ 38.2%) for CVD mortality. Conclusions: Even at low-level exposure, cadmium appears to be associated with increased mortality. Further large prospective studies of cadmium exposure and mortality are warranted. Key words: Cadmium, cancer, meta-analysis mortality, prospective studies
C The Author 2015; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association V
1
2
International Journal of Epidemiology, 2015, Vol. 0, No. 0
Key Messages • High cadmium exposure was associated with increased mortality, with the strongest and most consistent association
with CVD mortality. • There was a suggestive positive association between urinary cadmium concentration and cancer mortality, but the
number of studies was few and results were inconsistent. • The positive association between urinary cadmium concentration and mortality was observed in populations with
low-level cadmium exposure.
Introduction
Methods Search strategy and inclusion criteria Studies were identified by searching PubMed and Embase (1966 to 30 March 2015), without restrictions, using the search terms cadmiumANDmortality.We also searched the reference lists of retrieved articles to find additional studies. Eligible for inclusion in this meta-analysis were cohort studies that reported hazard ratios (HR) or relative risk estimates with 95% confidence interval (CI) for the association of urinary cadmium concentrations with all-cause,
Data extraction and quality assessment The data extracted included the first author’s last name, year of publication, country, duration of follow-up, sex and age of participants, number of deaths, cut-offs for the highest vs lowest category of urinary cadmium concentration, variables adjusted for in the multivariable model, and the most fully adjusted HRs with 95% CIs for the highest vs lowest category of cadmium concentrations. One of the authors (S.C.L.) extracted the data from each study. We used the Newcastle-Ottawa Scale (NOS) for quality assessment of the studies.20 We used pre-defined criteria: selection (population representativeness); comparability (adjustment for confounders); and outcome (ascertainment of outcome and follow-up). For the adjustment for confounders, studies were awarded one point if adjustment was made for age and sex and two points if further adjustments were made for smoking and body mass index or body weight. The NOS ranged from 0 to 9, with 9 points reflecting the highest study quality.
Statistical analysis Study-specific HRs were combined using a random-effects model, which considers both within- and between-study variability.21 When results were presented separately for men and women, we combined the sex-specific HRs, using a random-effects model and included the overall estimate in the meta-analysis. We conducted a dose-response metaanalysis of urinary cadmium concentrations and all-cause mortality using the same method as described in a previous dose-response meta-analysis.22 Statistical heterogeneity among studies was assessed with the Q and I2 statistics.23 We conducted a sensitivity analysis excluding cohort studies of participants living in cadmium-polluted areas and cohorts of highly exposed participants. We performed stratified analyses by sex and age range of the study
Downloaded from http://ije.oxfordjournals.org/ at University of Manitoba on June 10, 2015
Cadmium is a heavy metal widely distributed in the environment. It accumulates in the body, especially in the kidneys, and has an elimination half-time of 10–30 years.1 The general population is exposed to cadmium mainly through inhalation of cigarette smoke and intake of food.2 Cadmium is primarily present in foods that are otherwise considered to be healthy, such as whole-grain foods, potatoes and vegetables. Those foods contribute to 60–70% of total cadmium intake in the European and US populations.3,4 The presence of cadmium in whole grains and vegetables may counteract the potential health-beneficial substances in those foods, such dietary fibre, essential nutrients, antioxidants and phytochemicals. Besides being a carcinogen,5 cadmium exposure is associated with kidney disease,6,7 osteoporosis and bone fractures,8,9 atherosclerosis10–12 and cardiovascular disease (CVD).13–15 Epidemiological studies on low-level cadmium exposure in relation to risk of cancer and CVD are limited and have generally been based on small sample sizes, a case-control or cross-sectional design, or dietary cadmium intake as a measure of cadmium exposure.14–19 To further examine the health effects of cadmium exposure, we conducted a systematic review and metaanalysis to summarize available evidence from prospective studies assessing the associations of urinary cadmium concentration with all-cause, cancer and CVD mortality in the general population.
cancer or CVD mortality. If results for the same mortality endpoint were reported in more than one article, we included the study with the most comprehensive data.
International Journal of Epidemiology, 2015, Vol. 0, No. 0
3
Records identified by searching PubMed (n=673) and Embase (n=1115) Duplicate records excluded (n=535) Records remaining after removal of duplicates (n=1253) Records excluded based on title or abstract (n=1236) Records remaining after title and abstract review (n=17)
Articles included in meta-analysis (n=9) • All-cause mortality: 6 • Cancer mortality: 4 • Cardiovascular disease mortality: 5 Figure 1. Flow-chart of study selection.
population, country, duration of follow-up, and study quality, and assessed heterogeneity across subgroups through meta-regression. Publication bias was evaluated using Egger’s test.24 Stata (version 12.0, StataCorp, College Station, TX) was used for the statistical analyses.
Results Literature search Our literature search identified 1253 non-duplicate records of which 17 remained after title and abstract review (Figure 1). Among the records obtained for full-text review, we excluded four because the exposure was not urinary cadmium25–27 or no HR was reported.28 We further excluded four studies29–32 due to duplicate publications from the same study population and same follow-up period.
Characteristics of the included studies Table 1 shows the characteristics of the nine included studies.33–41 Results were presented for all-cause mortality
in six studies (including 5600 deaths), for cancer mortality in four studies (1332 deaths) and for CVD mortality in five studies (1715 deaths). All studies included both men and women. Five studies were conducted in the USA, one in Belgium and three in Japan. The geometric mean urine cadmium concentration (in mg/g creatinine) ranged from about 0.3 to 1.0 in the US and Belgian populations, with higher concentrations in women than in men. In the Japanese studies, urine cadmium concentration (in mg/g creatinine) ranged from 1.8 in men and 2.4 in women from non-polluted areas to about 4–6 in men and 7–8 in women from a cadmium-polluted area. The studies by Nakagawa et al.33 and Li et al.36 were based on the same Japanese population but the mortality end-points were different. Likewise, the studies by Menke et al.35 and Lin et al.38 were based on the same US population. Menke et al.35 reported results for all-cause, cancer and CVD mortality during follow-up from 1988–94 through 2000. Lin et al.38 reported results for cancer mortality only and the follow-up was from 1988–94 through 2006. We included results for all-cause and CVD mortality from the publication by Menke et al.35 and results for cancer from Lin et al.38
Downloaded from http://ije.oxfordjournals.org/ at University of Manitoba on June 10, 2015
Records excluded (n=8): • Exposure not urinary cadmium (n=3) • No hazard ratios reported (n=1) • Duplicate publications from the same cohort and same follow-up period (n=4)
Japan, NA
Belgium, CadmiBel
Nakagawa, 200633
Nawrot, 200834
women, 20
4.8
USA, NHANES Men and
Tellez-Plaza,
(1999–2004)
Men and 22.0 women, 50
Japan, NA
Li, 201136
201237
Men and 9.0 women, 20
Menke, 200935 USA, NHANES III (1988–94)
Men and 20.3 women, 20
Men and 15.0 women, 50
Country, study Sex, age (years) Followname up (years)
First author, year
8989
3119
13 958
944
3119
Subjects
7
9
8
7
Study quality
CVD
524 all causes ,191 8
334 cancer, 267 CVD
1690 all causes, 769 CVD
201 all causes, 54 cancer, 88 CVD
948 all causes
No. of deaths
–
1.68 (1.09–2.58) [men]; 1.14 (0.78–1.66) [women]
(1.71–6.53) [women] 1.20 (1.04–1.39)
1.32 (0.61–2.84) [men]; 3.11
All-cause
1.1 (0.7–1.9) [men]; 0.8 (0.4–1.6) [women] –
–
1.8 (1.0–3.1) [men]; 2.4 (1.1–5.1) [women] 1.74 (1.07–2.83)
1.33 (0.69–2.56) [men]; 0.82 (0.47–1.42) [women]
1.43 (1.08–1.89)
–
1.07 (0.85–1.34)
–
CVD
Cancer
HR (95% CI) of mortality
> 0.57 vs 0.14 (0.28) 1.52 (1.00–2.29)
20 vs
