Introduction

Cadmium (Cd) is a ubiquitous hazardous metal known to induce renal tubular dysfunction after long-term exposure even at low dose (International Programme on Chemical Safety 1992). In occupational health, both Cd in blood and urine (Cd-B and Cd-U, respectively) have been evaluated as markers of exposure (e.g., Cd-B and Cd-U by American Conference of Governmental Industrial Hygienists 2009 and Cd-U by Deutsche Forschungsgemeinschaft 2009). There is a general understanding that Cd-B is an indicator of recent exposure to Cd, whereas Cd-U reflects Cd exposure over a long term (e.g., Hassler et al. 1983; Lauwerys and Hoet 2001). In environmental health, it appears to be a tradition to use Cd-Ucr [i.e., Cd divided by creatinine (CR) in urine] as a Cd exposure marker among general populations. Nevertheless, it is known that CR [and to a lesser extent urine specific gravity (SG) also] will decrease as a function of advancement in age (Moriguchi et al. 2003, 2005).

The present study was initiated to investigate whether Cd-B and Cd-U (typically Cd-Ucr) correlate with each other irrespective of age among general populations for whom the exposure sources are almost exclusively foods (Ikeda et al. 2000, 2004) and therefore the exposure is basically a matter of everyday, and which one of Cd-B or Cd-U correlates more closely with renal tubular dysfunction markers such as α1-microglobulin (α1-MG), β2-microglobulin (β2-MG) and N-acetyl-β-d-glucosaminidase (NAG) (e.g., Moriguchi et al. 2004, 2009).

Materials and methods

Data and sample collection, and ethical issues

Data on a Cd exposure marker of Cd-U (e.g., as Cd-Uob) and three Cd effect markers of α1-MG-U, β2-MG-U and NAG-U on 1,197 adult women in six prefectures in Japan were cited from Tsukahara et al. (2003); the sampling and analyses were conducted in 2002. In addition, blood and urine samples were collected from 206 adult women in two separate prefectures in Japan in 2008 and analyzed for Cd exposure and effect markers similarly. Cd in blood levels (Cd-B) for these women, 1,403 in total, was cited from Ikeda et al. (2010) so that the data on five items of Cd-B, Cd-U, α1-MG-U, β2-MG-U and NAG-U for 1,403 adult women in eight prefectures (Fig. 1) were made available for the present analysis. Of the 1,403 cases, 1,367 cases (97.4%) reported on the smoking habits when asked by self-administered questionnaires on life style (Tsukahara et al. 2003; Ikeda et al. 2010); 1,123, 201 and 43 cases were never- (82.1%), current- (14.7%) and ex-smokers (0.2%). None of the current smokers consumed ≧20 cigarettes/day.

Fig. 1
figure 1

The locations of the eight prefectures where blood samples were collected. The eight prefectures of blood sample collection are shown with shades. Note that Japan consists of 47 prefectures in total

Full size image

The study protocol was approved by the Ethics Committee in Kyoto Industrial Health Association. Each of women provided informed consent in writing.

Blood and urine analyses

Cd in blood was analyzed by inductively coupled plasma mass spectrometry (ICP-MS;ELEMENT2, Thermo Fisher Scientific, Bremen, Germany) after wet ashing of the blood sample by heating in the presence of extra-pure nitric acid; performance of the analytical method including external and internal quality assurance was sufficient as previously described (Ikeda et al. 2010). The quantification limit was 0.1 μg/l blood. When the method was applied to urinalysis, correction was made after US Environmental Protection Agency (1990) as 111Cd in urine was interfered with 95Mo16O in ICP-MS assay; there was no need of the correction in case of blood assay because Mo in blood was low enough.

Urinalysis for Cd was by graphite furnace atomic absorption spectrometry (GFAAS) after light wet ashing with nitric acid. Namely, 1.5 ml urine sample was mixed with 3.5 ml 9% extra-pure nitric acid in a acid-washed tube and the tube was sealed, and the mixture in the tube was heated at 100°C for 30 min. The digest was introduced to a GFAAS (Hitachi Z-8270) to measure Cd at 228.8 nm as previously described (Ezaki et al. 2003). The quantification limit was 0.5 μg/l urine (Ezaki et al. 2003). Urinalyses for α1-MG-U, β2-MG-U, NAG-U, creatinine (CR) and urine specific gravity (SG) were by conventional methods as previously described (Ezaki et al. 2003; Moriguchi et al. 2003; Tsukahara et al. 2003).

The analyte levels in urine were expressed as observed (e.g., Cd-Uob; in μg/l) and after correction for CR (in g creatinine/l urine) in urine (e.g., Cd-Ucr in μg/g cr; Jackson 1966) or for SG (1.016 as a standard; e.g., Cd-Usg; Rainsford and Lloyd Davies 1965). SG was expressed in terms of factor G, which was defined as factor G = (SG-1.000) × 1,000.

Statistical analysis

Cd-B, Cd-U, α1-MG-U, β2-MG-U and NAG-U distributed log normally. Accordingly, the analyte concentrations were expressed in terms of geometric means (GMs) and geometric standard deviations (GMs). A normal distribution was assumed for age, CR and SG (in terms of factor G). In some cases, the analyte levels were originally presented in terms of AM ± ASD (e.g., Arai et al. 1994) and were converted to GM (GSD) by use of the moment method (Sugita and Tsuchiya 1995).

Possible age dependency and other correlations were examined by simple (SRA) and multiple regression analysis (MRA). ANOVA followed by multiple comparison (Sceffé) was also employed. Difference between paired data was examined by paired t-test (parametric) and Wilcoxon test (non-parametric). Difference between slopes of two regression lines and two correlation coefficients was examined after Ichihara (1995).

Results

Populations surveyed

Basic parameters for the 1,403 women surveyed are summarized in Table 1. Age, CR and SG distributed in wide ranges, but the AM values were 44.3 years for age, 1.07 g/l for CR and 18.1 for SG (in terms of factor G). GM Cd-Uob was less than 1 μg/l to suggest low levels of Cd exposure. Correspondingly, the GM values for α1-MG-Uob, β2-MG-Uob and NAG-Uob were 1.92 mg/l, 91 μg/l and 2.8 units/l, respectively.

Table 1 Basic parameters for the population studied
Full size table

Correlation between the results by GFAAS and that by ICP-MS

As two separate methods were applied in the analyses, i.e., GFAAS for Cd-U and ICP-MS for Cd-B, the compatibility in the results was examined. For this purpose, 20 urine samples containing Cd at 0.3–10.8 μg/l were selected and subjected to the analyses by the two methods. The results showed that there existed a close agreement between the pairs of the data. The regression equation was such that Y = 0.08 + 1.036X, r = 0.991, p < 0.01, where X was Cd-U (μg/l) by ICP-MS and Y was Cd-U (μg/l) by GFAAS. The intercept was essentially 0, and the slope was very close to 1.

Correlation between Cd-B and Cd-U

When regression analyses were conducted taking Cd-B as an independent variable and one of Cd-U (i.e., Cd-Uob, Cd-Ucr and Cd-Usg) as a dependent variable, the two variables correlated significantly (p < 0.01) irrespective of urine density correction, i.e., r was 0.462, 0.563 and 0.551 for Cd-Uob, Cd-Ucr and Cd-Usg, respectively. The lower 95% limits for slopes were >0.53 in all cases. Thus, it was evident that there existed a close correlation between Cd-B and Cd-U, and Cd-U increased as a function of increasing Cd-B.

Changes in Cd-B and Cd-U in relation to age

When a total of 1,403 cases (i.e., Japan as a whole) were subjected to the analyses to examine if Cd-B (like Cd-U; Moriguchi et al. 2003) would increase in association with age, it was found that all of Cd-B and Cd-U (as Cd-Uob, Cd-Ucr and Cd-Usg) correlated significantly (p < 0.01) with age. The lower 95% limits for the slopes were also >0 in all cases indicating that both Cd-B and Cd-U increased as age advanced (the top half in Table 2). Nevertheless, the correlation coefficient for Cd-B, 0.195, was significantly (p < 0.05) smaller than the coefficients for Cd-U, e.g., 0.237 for Cd-Uob. Furthermore, the slope for Cd-B (0.015) was significantly smaller than the slope for Cd-Uob (0.026, p < 0.05) and for Cd-Usg (0.026, p < 0.01).

Table 2 Lines of regression for age with Cd-B and Cd-U
Full size table

To examine if the findings with 1,403 cases (i.e., Japan as a whole) could be reproduced even on a prefectural basis, the 483 cases from a single prefecture of Kyoto were subjected to the analyses. Essentially the same results were obtained in the sense that the slopes were positive (i.e., >0) both for Cd-B and for Cd-U, and that the correlation coefficient was largest for Cd-Ucr, as shown in the bottom half in Table 2.

Age-dependent change in the Cd-U over Cd-B ratio

As both Cd-B and Cd-U increased in accordance with age, comparison was made to know which one of Cd-B and Cd-U tended to increase more sharply as a function of age. For this purpose, the ratio of Cd-U (as Cd-Uob, Cd-Ucr or Cd-Usg) over Cd-B was calculated for each of the 1,403 cases, and the regression analyses were conducted taking age as an independent variable and one of the three ratios as a dependent variable (Table 3).

Table 3 Increase in Cd-U over Cd-B as a function of age
Full size table

The analyses showed that the lower 95% limits of the slopes were all >0 (i.e., positive), indicating that the ratio would increase in accordance with age. Further analyses disclosed that the ratio would reach 1 at the age of 51, 46 and 59 years in case of Cd-Uob over Cd-B, Cd-Ucr over Cd-B and Cd-Usg over Cd-B ratios, respectively. The case of Cd-Ucr over Cd-B ratio against age is depicted in Fig. 2 for visual understanding. The observation as a whole was taken to suggest that Cd-U would increase more sharply than Cd-B especially in the latter half of life.

Fig. 2
figure 2

Age-dependent increase in the Cd-Ucr over Cd-B ratio. The ratio of GM Cd-Ucr over GM Cd-B was spotted against age (n = 1,403). The thick line in the middle is a calculated regression line and two dotted curves on both sides show the 95% confidence range. Each dot represents a group of subjects. Note that the slope is >0, and that the regression line crosses the ratio = 1 line in the middle of the age range of 40–50 years

Full size image

A separate analysis taking Cd-B as an independent variable and the Cd-Ucr over Cd-B ratio as a dependent variable revealed that the slope was −0.174 (the 95% range; −0.217 to −0.131). The 95% upper limit of the slope was <0 (i.e., −0.084 for both cases) even when Cd-Ucr was replaced with Cd-Uob or Cd-Usg. The observation was taken to suggest that the ratio would decrease as a function of increasing Cd-B among women with low Cd exposure (the highest Cd-B studied, 6.9 μg/l; Table 1).

Correlation of Cd-B and Cd-U with three effect markers of α1-MG-U, β2-MG-U and NAG-U

Correlation analysis was conducted between two exposure markers of Cd-B and Cd-U (the top line in Table 4). It was observed that the two exposure markers significantly (p < 0.01) correlated with each other irrespective of urine density correction, as discussed above. The correlation of NAG-Uob with Cd-Uob is depicted in Fig. 3.

Table 4 Correlation coefficients between exposure markers and effect markers
Full size table
Fig. 3
figure 3

Correlation of NAG-Uob with Cd-Uob. The meaning of the three lines and dots in the figure is as in Fig. 2

Full size image

Further analyses with the three tubular dysfunction markers (i.e., α1-MG-U, β2-MG-U and NAG-U) (the bottom three lines in Table 4) revealed that Cd-U always gave greater correlation coefficients than Cd-B, irrespective of urine density correction. Statistical evaluation by Wilcoxon test showed that the coefficients with Cd-U were significantly (p < 0.01 by both tests) greater than the coefficients with Cd-B. Coefficients with Cd-B were all smaller than 0.1.

Some of the coefficients were statistically significant even though the coefficients were as small as <0.1 (e.g., p < 0.01 for r = 0.078), because the number of cases was as large as 1,403. Nevertheless, it should be worth noting that Cd-B failed to show significant correlation with β2-MG-U (p > 0.10), irrespective of correction of β2-MG-U for urine density.

To detect possible effects of exposure intensity on the correlation, a total of 1,403 cases were classified into tertiles in terms of either Cd-B, Cd-Uob or Cd-Ucr (n = 467–468 per group). When coefficients of correlation between the exposure markers and three effect markers were examined, no exposure intensity-dependent changes in correlation coefficients were detected (data not shown).

Discussion

The present analysis of 1,403 women cases collected all over Japan made it clear that both of two Cd exposure markers of Cd-B and Cd-U increased as a function of age (Table 2), and that the two markers correlated significantly (p < 0.01) with each other (Table 3). The Cd-U over Cd-B ratio increased as a function of age. Fourthly, Cd-U correlated significantly (p < 0.01) with three effect markers of α1-MG-U, β2-MG-U and NAG-U more closely than Cd-B did (Table 4). Cd will be first in blood (Cd-B) when absorbed and will be accumulated in the kidney to give damage to tubules when the dose is high enough, while a part of Cd in the kidney will be excreted into urine (Cd-U) (International Programme on Chemical Safety 1992). Thus, it is plausible to expect that Cd-U rather than Cd-B correlates better with tubular dysfunction markers such as α1-MG, β2-MG and NAG.

With regard to instrumental analysis for Cd-B, a previous study on the comparative evaluation of Cd-B measurement results by ICP-MS (SPQ 9000 Plasma Quadrupole MS) and by GFAAS, it was observed that ICP-MS results were 12–19% lower than GFAAS results (Zhang et al. 1997). The reason for better agreement in the present study remains unknown. As for the evaluation of effect marker values, there are proposals to take β2-MGcr = 300 μg/g cr (e.g., Bernard 2008) or 1,000 μg/g cr (e.g., Ezaki et al. 2003). The distribution of β2-MGcr among the present population studied was such that the 95%-tile was 312 μg/g cr. The proportion of cases with >300 μg β2-MGcr/g cr was 5.7%. Nevertheless, none of the cases had Cd-Ucr in excess of 2 μg/g cr (Moriguchi et al. 2010). It should be added that in cases of patients with Itai–itai disease and suspected subjects, β2-MGcr was as high as >50,000 μg/g cr (e.g., Ikeda et al. 2003). To the knowledge of the authors, there is no proposal of cut-off values for α1-MG or NAG.

This study group has previously pointed out that α1-MG-U is possibly better than traditional β2-MG-U as a marker of Cd-induced effects on tubular function (Moriguchi et al. 2004). In a subsequent study, the group suggested that NAG should be more sensitive than α1-MG-U and β2-MG-U as a marker of health effects among general populations (Moriguchi et al. 2009). It was also made clear that the traditional method of correction for urine density as a ratio over creatinine (Jackson 1966) may induce bias especially among aged populations as creatinine levels in urine will decrease substantially among old people (Moriguchi et al. 2003, 2005).

In the present study, the relationship with tubular dysfunction markers was compared between Cd-B and Cd-U. It was concluded that Cd-U should be a better marker of exposure than Cd-B. The choice of urine rather than blood carries substantial implication in practice. Whereas urine sampling is not invasive and therefore easy to apply, blood sampling is inevitably associated with several risks both on phlebotomists in charge (Jones and Pether 1993; Pungpapong et al. 1999; Rapparini et al. 2007) and blood sample donors (Berry and Wallis 1977; Kimura 2008; Okamoto et al. 2009). Such risks need not be considered when urine (rather than blood) is employed for exposure monitoring.

With regard to comparative evaluation of Cd-B and Cd-U, literatures reporting Cd-B and Cd-U of the same populations were rather limited. Thus, nine articles (Ghezzi et al. 1985; Hassler et al. 1983; Jakubowski et al. 1987; Arai et al. 1994; Higashikawa et al. 2000; Shimbo et al. 2000; Trzcinka-Ochocka et al. 2002; Uetani et al. 2006; US Centers for Disease Control and Prevention 2009) were available (Table 5) as those in which both Cd-B and Cd-U (either one or more of Cd-Uob, Cd-Ucr or Cd-Usg) were reported for a given population. Five reported on exposed populations and six on non-exposed populations; Jakubowski et al. (1987) and Arai et al. (1994) reported on the non-exposed population (in addition to the exposed) as the controls to the exposed workers. Most of the articles reported Cd-B and Cd-U as GM values; Arai et al. (1994) however reported the results as AM ± ASD and the values were converted to GM (GSD) by use of the moment method (Sugita and Tsuchiya 1995) for uniformity of the data understanding. The nine reports as a whole offered Cd-B and Cd-U on 39 groups of occupationally non-exposed subjects and 19 groups of Cd-exposed workers, although the number of subjects in each group were various from a small of 5 to a large of 4,525 cases.

Table 5 Cd-B and Cd-U reported in literature
Full size table

With regard to Cd-U, data were most abundant for Cd-Ucr. Accordingly, the analysis was conducted primarily on the Cd-B and Cd-Ucr pairs. Genders studied were biased, namely, both men and women were studied in the non-exposed groups, whereas mostly men were studied in the exposed groups.

A regression analysis with all groups (i.e., 58 groups ignoring the difference in the number of subjects in groups) showed that there was a significant correlation (p < 0.01) between Cd-B and Cd-U irrespective of urine density correction. The correlation coefficient with Cd-Ucr (n = 54) was as high as 0.812 (p < 0.01), possibly because the GM Cd-B among the exposed groups was up to 49 μg/l. When cases were divided into the exposed and non-exposed, the correlation coefficient with Cd-Ucr for the non-exposed (n = 38) went down to 0.642 (that was nevertheless statistically significant; p < 0.01), being greater than the present analysis results (the top line in Table 4). Although the slope for the exposed groups appeared to be steeper (1.12) than that for the non-exposed (0.53), the difference was statistically insignificant (p > 0.05).

The ratio of Cd-Uob over Cd-B decreased in the non-exposed groups as a function of increasing Cd-B (the slope, −0.516 with the 95% range of −0.753 to −0.278), whereas the ratio did not vary irrespective of Cd-B in the exposed group (the 95% range for slope; −0.234 to 0.094). The finding on the non-exposed group was in agreement with the present observation on the 1,403 non-exposed women in the sense that the slope was negative and the 95% range for the two analyses overlapped with each other. The findings as a whole may be on the line with the understanding that Cd-U will not increase rapidly when Cd exposure is moderate (e.g., up to Cd-B of 4 μg/l) as Cd absorbed will be stored in the kidneys (International Programme on Chemical Safety 1992), whereas both Cd-U and Cd-B will increase when Cd exposure is rather intense (e.g., up to Cd-B of 50 μg/l) as in the case of Cd exposure in workplaces in the past (e.g., Hassler et al. 1983; Trzcinka-Ochocka et al. 2002).

Information on the effects of aging on the Cd-U over Cd-B ratio was very limited. Uetani et al. (2006) presented Cd-B and Cd-U (i.e., Cd-Uob and Cd-Ucr) in non-exposed subjects in three survey areas by age groups (50–59, 60–69 and 70–99 years) and by smoking habits. ANOVA (followed by Scheffé) taking the Cd-Uob over Cd-B ratio or the Cd-Ucr over Cd-B ratio as a dependent variable and age range and smoking habits (non-smoking vs. smoking) as two independent variables failed to show the effects of age on the two ratios. The inability might be due to the fact that the subjects were biased to older people (with no younger groups).

In overall evaluation, it appears prudent to conclude that, although the two exposure markers of Cd-B and Cd-U correlate with each other (as supported by literature survey results), Cd-U (irrespective of correction for urine density) correlates with the three tubular effect markers of α1-MG-U, β2-MG-U and NAG-U more closely than Cd-B. The observation suggests that Cd-U rather than Cd-B should be recommended as an exposure marker of choice in Cd biological monitoring of general populations. This recommendation however refers primarily on cases of long-term exposure. In case of acute, high-dose exposure, Cd-B will be the indicator of choice. In evaluating the results, effects of aging should be taken into account.