This is the first population-based study with exposure biomarkers and a prospective design assessing the potential associations between risk of developing end-stage renal disease and low-level exposure to the metals Cd, Pb, and Hg. It indicates that the Pb concentration in blood (erythrocytes) is a predictor of development of end-stage renal disease in the general population. Causality is, however, not demonstrated, though one possibility is that Pb decreases the kidney’s ability to resist other factors. In addition, Ery-Cd tended to be related to an increased risk of end-stage renal disease, but confounding by Pb and Hg could partly explain this. In contrast, Ery-Hg showed a negative association. In addition, gender-specific analyses showed that men carried almost all of the Ery-Pb and Ery-Cd associated risks.
One strength of this study is the prospective design, with population-based blood samples collected several years before diagnosis of end-stage renal disease. In addition, end-stage renal disease is a well-defined, hard renal end-point, and the Swedish Renal Registry has been validated, showing that > 95% of the patients who started active end-stage renal disease treatment were reported to the SRR. Thirdly, as is shown by results for reference materials, the quality of our metal analyses was very good.
One weakness of this study is that neither renal function (e.g., estimated glomerular filtration rate) nor urinary albumin were available at baseline. Thus, risk estimates were not adjusted for any impairment of kidney function already existing before the sampling. At normal kidney function glomerular filtration rate is above 90 ml/min/1.73 m2. Renal replacement therapy is generally begun at 5–10 ml/min/1.73 m2. The rate of decline in renal function in a general healthy population older than 30 years is approximately 0.8 ml/min/1.73 m2/year ; meanwhile, patients with pre-existing mild to moderate kidney disease (chronic kidney disease 3–4) may lose 2 to 13 ml/min/1.73 m2/year . Thus, several of the cases probably had reduced renal function already at the time of sampling.
Another weakness is the relatively few cases of end-stage renal disease, which limits the ability to determine whether risk estimates are homogeneous between the cohorts. From a risk-assessment perspective, an additional third weakness is that the Ery-Pb has decreased over time . Therefore, the measured erythrocyte concentrations neither reflect a lifetime mean, nor the level at time of development of end-stage renal disease. Participation rates in the cohorts were between 45-77%. It is unlikely that any over- or under- representation would affect the risk estimates.
The referents were drawn from the general population, thus having normal values for Pb, Cd, and Hg. In an international perspective, the loads of Pb and Cd in Sweden have been low, while the load of Hg is relatively high [11, 34]. The levels of Pb and Cd in erythrocytes are usually about twice as high as those found in whole blood (variation in erythrocyte volume fraction, of course, affect this, but only to a negligible extent), while for Hg, the measured concentrations are closer to those in blood, but depend on the rates between methylmercury and inorganic mercury. For comparability with studies using urinary Cd, it can be noted that 50–60 year-old women from northern Sweden had a median Cd concentration in urine of 0.189 (range 0.03-0.84) μg/L, and in whole blood 0.24 (range: 0.08-2.59) μg/L (corresponding to ca 0.49 μg/L in erythrocytes) .
Associations between Cd and Pb and end-stage renal disease have been found in earlier studies using less adequate methods. One previous Swedish study combining occupational and ecological exposure assessment indicated increased risk of developing end-stage renal disease after Cd exposure , while another ecological study in Japan did not show any association between mortality from renal failure and Cd levels in local brown rice . These studies are weakened by their lack of individual biomarker data. Cross-sectional studies have found higher levels of Ery-Cd and Ery-Pb in hemodialysis patients than in healthy referents (Taiwan: ; U.S.: ; Pakistan: ; China: ). In addition, progression of renal failure was associated with blood-Pb concentration (China: [21, 22]) and a slower decline in renal function have been observed after lead chelation therapy (Taiwan: ). In most of these countries, the Pb exposure has been higher than in Sweden. On the other hand, in another Swedish study, there was no association between a history of occupational Pb exposure, as assessed by a job-exposure matrix, and risk of end-stage renal disease .
For less-severe kidney effects, several studies indicate associations with biomarkers of Cd  and Pb  exposure. A cross-sectional study in Taiwan showed significant associations between blood-Pb and risk of renal dysfunction and hyperuricemia . Likewise, studies of U.S. adults displayed associations between Cd and Pb in blood on the one hand, and both albuminuria and reduced estimated glomerular filtration rate, on the other , for blood-Pb also in adolescents .
The evidence for the development of nephropathy at high and prolonged Pb exposure was described as ‘anecdotic’ in a relatively recent review, while data on the relationship between Pb and end-stage renal disease was considered scarce . However, some of the most recent cross-sectional studies were not included in the review. Thus, taking all the recent studies and our present results into account, such an association is likely, even at fairly low-level exposure.
Information on Hg is scarce, but not in conflict with our data, although subjects with occupational exposure to elemental Hg vapor have been shown to display albuminuria . Also, high exposure to inorganic Hg may cause severe membranous nephropathy . However, such high-level exposure to Hg is rare in Sweden. In the general population, the major exposure is to elemental Hg from dental amalgam and methyl-Hg through fish consumption, in both cases at relatively low exposures, not expected to be sufficient to cause renal effects.
It has been claimed that Cd and Pb may cause a general worsening of renal disease . Our results tend to support this, since the patterns were similar for Ery-Pb and Ery-Cd in several primary causes of end-stage renal disease. However, an exception was cases with diabetic nephropathy. Despite the fact that previous studies suggest that Cd may induce diabetes [43, 44], and aggravate diabetic renal complications , we did not find a difference in Ery-Cd between cases with diabetic nephropathy and referents (only 19 cases).
The results contain two unexpected findings: first a gender difference for all three metals, and second, a negative association between Ery-Hg and the risk of developing end-stage renal disease. The difference in risk estimates for men and women could be due to a gender difference in susceptibility or a difference in exposure biomarkers reliability between men and women. This should be further evaluated. The possibility of confounding by unobserved covariates should be considered for both Ery-Pb and Ery-Hg. A truly protective effect of Hg appears unlikely. Instead, it can be speculated that there may be confounding by a healthy lifestyle, including fish consumption and with a low burden of other risk factors  such as low socio-economic status [38, 46, 47]. An alternative, and even more speculative interpretation, is a protective effect of long-chain omega-3 polyunsaturated fatty acids (PUFAs) from fish. PUFAs correlate to Ery-Hg  and have been suggested in treatment of kidney disease, i.e., IgA-nefritis [49, 50].
If confounding by healthy life-style is a possible explanation of the results for Hg, such confounding must be considered also for Pb. Ery-Hg and Ery-Pb were positively correlated, and it has previously been observed that Ery-Pb is associated with fish consumption (; possibly due to co-variation with leafy vegetables, a source of dietary Pb). Therefore, the association between end-stage renal disease and Ery-Pb is not likely to be due to confounding by a healthy life-style, since the risk estimate for Ery-Pb then would rather decrease than increase. Significant confounding by occupational factors is also unlikely because occupational exposure could only explain a minor fraction of the cases in this population.
As previously mentioned, several of the cases probably had reduced renal function already at baseline. We therefore performed subgroup analyses within quartile limits of time between sampling and end-stage renal disease (see Additional file 1). The impact of time from sampling to diagnosis only had a limited effect on the odds ratio for Ery-Pb, but for Ery-Cd the odds ratio was lower among individuals with a long time period from sampling to diagnosis. No consistent pattern was seen for Ery-Hg. These results do not indicate that the results for Ery-Pb and Ery-Hg reflect reverse causation, while this may be the case for Ery-Cd. A previously published study showing an association between bone-Pb, which has a very slow turnover, and end-stage renal disease  is also in accordance with our finding on Pb.
Both the present and other studies show an increased risk of developing end-stage renal disease in smokers [38, 46, 47]. Smoking is, however, also an important source of Cd. When both smoking and Cd were included in a multiple analysis, both were statistically significant, thus the possibility remains that other smoking-related factors than Cd may cause the association. The number of never-smokers in the present study is too small to allow a separate analysis of these individuals.
Although it is still uncertain if Pb is truly a cause of end-stage renal disease, this first low-level study using prospectively collected samples for exposure assessment shows associations that give reasons for concern.