Introduction

Cadmium (Cd) is a toxic element ubiquitous in the environment, and long-term exposures to this element through the general environment might elevate the risk of renal tubular dysfunction (International Programme on Chemical Safety 1992a, 1992b). In the practice of epidemiological studies on health effects of Cd exposure among general populations, β2-microglobulin (β2-MG) has been most popularly employed as a marker of renal tubular dysfunction (Kawada 1995), and attention has also been paid to other markers of tubular dysfunction, such as α1-microglobulin (α1-MG) (Tohyama et al. 1986; Kawada 1995) and N-acetyl-β-D-glucosaminidase (NAG) (Nogawa et al. 1986; Kawada et al. 1992; Stengel et al. 1999). It has been a common practice (World Health Organization 1996; American Conference of Governmental Industrial Hygienists 2003) to correct urinary marker levels in terms of either concentration of creatinine [CR (or cr)] or specific gravity [SG (or sg)] (e.g., β2-MGcr or β2-MGsg).

Experiences among Cd-polluted areas have shown that the Cd-induced tubular dysfunction would be more evident after exposures for many years and therefore among elderly populations (International Programme on Chemical Safety 1992a, 1992b). Previous analyses (Ikeda et al. 2005; Moriguchi et al. 2005) have shown that both urinary CR and SG decrease substantially as a function of age above 30 years, and that the decrease is more marked for CR than for SG. It is the purpose of the present analysis to examine if Cd and tubular dysfunction marker levels in urine show age-dependent changes among women in Cd-nonpolluted areas in Japan, and if such trends would be further modified by the correction for CR or SG. The possible implication of correction-induced changes in evaluation of Cd exposure-associated tubular dysfunction is also discussed.

Materials and methods

Populations surveyed

Three databases were employed in combination for present analyses, as elsewhere described in detail (Moriguchi et al. 2005). In short, three databases, one established in 2000–2001 on 10,883 adult women (aged mostly from 40 to 60 years; Ezaki et al. 2003) the second one established in 2002 on 1,482 adult women (mostly at the ages of 35–60 years; Tsukahara et al. 2003), and an additional collection of data on 1,274 women volunteers (mostly of ≧60 years of age; Ikeda et al. 2005) were combined for the present analyses. All participants provided informed consents. Data on Cd, α1-MG, β2-MG, NAG, CR and SG in urine and data on social habits were sorted from the combined databases. After adjustment of the duplicate registration, selection for never-smoking women gave 11,090 cases in total.

Urinalysis for Cd, tubular function markers, CR and SG

The methods of urinalyses and the programs of quality assurance were as previously detailed (Ezaki et al. 2003; Moriguchi et al. 2003). In essence, Cd was measured by graphite furnace atomic absorption spectrometry, α1-MG and β2-MG by RIA, NAG by ELISA, CR by colorimetry and SG by refractometry, respectively. The quality of urinalysis for Cd was validated by the 31st and 32nd Round Robin external quality assurance programs (organized by German Society of Occupational Medicine and Environmental Medicine; Schaller et al. 2002).

Statistical analysis

Urinary analyte levels (e.g., Cd in urine) were expressed in some instances after correction for CR concentration (Jackson 1966) (expressed, e.g., as Cdcr), and also after correction for a urine SG of 1.016 (Buchwald 1964; Rainsford and Lloyd Davies 1965) (e.g., Cdsg), in addition to the values as observed (e.g., Cdob). In statistical analysis, SG was expressed in terms of factor G, which was defined by the equation that factor G = (SG−1.000)×1,000 (Levine and Fahy 1945) for practical convenience.

A log-normal distribution was assumed for Cd, α1-MG, β2-MG and NAG (Ikeda et al. 1995, 2000), and the distribution was expressed in terms of a geometric mean (GM) and a geometric standard deviation (GSD), whereas the age, CR and SG distributed almost normally (Moriguchi et al. 2005), and the distribution was expressed in terms of an arithmetic mean (AM) and an arithmetic standard deviation (ASD). Accordingly, regression analyses between the age and urinary analytes were conducted after logarithmic conversion of Cd, α1-MG, β2-MG and NAG.

Results

Urinary analyte levels in the group studied

GMs and GSDs for Cd, α1-MG, β2-MG and NAG are summarized in Table 1, together with AM and ASD for age, CR and SG. The GM values for Cd as observed was 1.10 μg/l, and it was 1.32 μg/g cr when corrected for CR.

Table 1 Basic parameters of the group studied
Full size table

In agreement with the estimation that subjects were from non-polluted areas, the GM values for α1-MG, β2-MG and NAG were, e.g., 2.6 mg/g cr, 121 μg/g cr and 4.1 units/g cr, respectively, indicating that the participants did not show tubular dysfunction on a group basis.

Age-dependent increases in Cd, and three tubular dysfunction markers

The relation of Cd (in μg/l or μg/g CR) with age (in years) was examined after logarithmic conversion of Cd values (Table 2). The slopes were positive (i.e., >0) and correlation coefficient was statistically significant (P<0.01) irrespective of correction for urine density. The relation of log Cdcrwith age is presented in Fig. 1A together the regression line for visual understanding. The observation as a whole suggested that Cd most probably increased steadily throughout the age ranges studied.

Table 2 Age-dependent increases in Cd, α1-MG, β2-MG and NAG in urine. The figures are parameters of a regression line in which Y = α+βX, where X is age (in years) and Y is the parameter given in the table after logarithmic conversion (e.g., log Cd-Uob)
Full size table
Fig. 1
figure 1

Age-dependent changes in A Cdob, B α1-MGsg, C β2-MGcr and D NAGob. Each dot represents one case (11,090 cases in A, B and C, and 2,115 cases in D), and the line is drawn with an assumption of linear relation. For the regression equations, see Tables 1, 2, 3

Full size image

Similar analyses for the age-dependency were conducted with three tubular dysfunction function markers; i.e., α1-MG, β2-MG and NAG (Table 2, Fig. 1B–D). The results showed that these parameters with positive slopes and significant (P<0.01) correlation coefficients (irrespective of correction) also increased as a function of the age throughout life. The correlation coefficients for β2-MG (0.031–0.279), however, tended to be smaller than the corresponding values for α1-MG (0.078–0.318) and NAG (0.063–0.454), as examined by Wilcoxon’s signed rank test for paired samples (P≒0.10).

The age-dependent decreases in CR and SG (being more marked in CR) was confirmed in the present study (Table 2) as observed in previous analyses (Moriguchi et al. 2005), and similarly age-dependent increase was observed for Cd and three tubular dysfunction markers in the present analyses (Table 2). In order to evaluate the two changes of the different direction in combination, the analyte levels at two separate ages of 30 and 80 years were estimated taking advantage of the regression equations given in Table 2. The results of the calculation are summarized in Table 3, in which I and II show the estimated level at the ages of 30 and 80 years, respectively.

Table 3 Ratio of estimates at the ages of 80 years over those at 30 years: Cd, α1-MG, β2-MG and NAG
Full size table

It is clear from Table 3 that the ratios for the three tubular dysfunction markers were generally greater than 1, and that the lower limits of the 95% confidence range were all >1 except for NAGob irrespective of correction for urine density. It was also possible to note that the ratios were larger for Cd than for three dysfunction markers, i.e., 3.16–6.31 in cases of Cd, and it was 1.12–3.16 for three dysfunction markers (P≒0.10 for comparison of Cd with α1-MG, β2-MG or NAG by Wilcoxon’s signed rank test and P≒0.01 for comparison of Cd with a combination of α1-MG, β2-MG or NAG by Mann–Whitney’s U-test for unpaired samples).

In a further analysis to compare whether the ratios for CR-corrected values were larger than that for SG-corrected values, the relative ratio was calculated taking the ratio for observed (ob) values as 1 [i.e., the relative ratio = (the ratio for CR-corrected or SG-corrected values)/(the ratio for OB values)], and the relative ratios were compared between CR-corrected values and SG-corrected values. Comparison by Wilcoxon’s signed rank test showed that the 80/30 years old ratios for CR-corrected values were significantly (P≒0.01) larger than the ratios for SG-corrected values.

Discussion

In the present analysis, only women were studied, because they are potentially at higher risk than men with regard to health effects due to Cd exposure (Sartor et al. 1992). In addition, the smoking rate among the Japanese population is substantially lower in women (about 16% for those in their 20s, with a gradual age-dependent decrease to 3% at 70 and above) than that in men (58–59% in their 20s to 40s, with a decrease to 29% at ≥70 years; Ministry of Health, Labour and Welfare 2003), indicating better opportunity of having never-smokers among women than men; never-smokers were selected in the present analysis because smoking is a known confounder in estimation of environmental exposure to Cd (Ikeda 1992). In the present analysis, urine samples were collected only once. Nevertheless, high reproducibility in analysis results after repeated urine sampling among populations with no apparent Cd-induced health effects has been confirmed for Cd, α1-MG and β2-MG (Ikeda et al. 2005) as well as for NAG (Stengel et al. 1999).

It appears that the age-dependent changes in urinary Cd levels have been studied only on limited occasions. According to Sartor et al. (1992), who studied men and women (2,327 subjects in total, and in a age range of about 20–80 years) in Belgium, Cd in 24-h urine samples increased from 0.34 μg/day at the age of slightly above 20 years to reach a peak of 1.12 μg/day in men and 0.87 μg/day in women at the age of around 60 years, followed by decreases to 0.96 μg/day (men) or 0.67 μg/day (women) at the ages of above 80 years. It was not possible, however, to detect such bi-phasic changes in the present study, probably because a linear regression was considered. Although smaller correlation coefficients for log β2-MG compared with those for log α1-MG or log NAG might be related to this possibility of bi-phasic relation, a quadratic regression between age and log β2-MG (with or without correction for CR or SG) did not show the presence of a local maximum (data not shown). The lack of decrease in Cd in urine at higher age in the present analysis may be associated with the fact that dietary exposure to Cd was substantially higher in the late 1960s (e.g., 50–95 μg/day) in Japan compared with the level at present (25–35 μg/day) even among general populations in areas where no environmental pollution with Cd was known (Ikeda et al. 2004). In this connection, it should be added that the general population exposure to Cd has been almost exclusively via foods (>99.5%), with essentially no contribution from atmospheric air (Ikeda 1992).

Reports on the relation of tubular dysfunction marker levels with age appear to be very limited. In an early report, Nomiyama et al. (1983) observed that both β2-MG and NAG increased age-dependently among the elderly populations (60s–90s) in an area with no known Cd pollution in Japan. In addition, they observed that such increases were however not accompanied by other changes of kidney function markers such as creatinine clearance or tubular re-absorption of phosphorus.

Age-dependent increase in Cd and tubular function markers may be further amplified when the levels were corrected for CR or SG (Table 3), because both CR and SG decrease as a function of age. Although it has been a common practice in Cd epidemiology (World Health Organization 1996; American Conference of Governmental Industrial Hygienists 2003) to correct urinary analyte concentration by CR concentration; Berlin et al. (1985), for example, did not find any advantage of CR correction over the use of observed values, although Mason et al. (1998) observed that CR correction might reduce intra-individual variation when multiple samples were collected from the same individuals.

Assuming that the problem associated with different urine density will be overcome by the determination of a large number of samples, and thus findings with the observed values can be considered as most reliable, the correction with CR concentration will induce amplification by two times and that with factor G (or SG) by 1.4 times in evaluating the changes in Cd as well as tubular dysfunction markers (Table 3). Thus, care should be practiced in evaluating these marker levels especially among the populations with ages in wide ranges.