Cadmium (Cd) is a non-essential element in the human body, and long-term exposure to sufficiently high Cd levels through food and water consumption, skin contact and inhalation may cause adverse health effects, such as renal dysfunction and osteoporosis [1–4]. Cd and its compounds are considered to be human carcinogens (primarily cancer resulting from inhalation exposures) [5], and Cd exposure is associated with breast cancer development in females [2]. Cd and its compounds are released to all elements of the environment due to a number of human activities, such as mining, smelting, industrial activities, waste disposal, application of fertilizer and pesticides and vehicle exhaust [6–8]. Cd pollution in soil, water and air, as well as different foods, has been frequently reported in China in recent years, which imposes increasing health risks to the public [8, 9]. To protect people from the adverse effects of Cd exposure, it is important to evaluate the reference level, which is the lowest concentration of urinary Cd (UCd) with a low probability of Cd-induced adverse health effects.

The benchmark dose (BMD) represents an estimate of the dose corresponding to a specified level of increased response (the benchmark response, BMR). It has been suggested that the no observed adverse effect level (NOAEL) can be replaced with the lower 95 % confidence limit of BMD [10, 11]. Compared with the NOAEL, the BMD method is not constrained as one of the experimental doses and makes better use of the dose–response information. This method more appropriately reflects the sample size and is regarded as a better approach than NOAEL to estimate the reference point for a continuous outcome variable [12].

Previous studies have examined the reference level for Cd-induced kidney effects using the BMD method and have currently displayed relatively large variations in critical UCd levels [13–22]. The Japanese population data from previous studies suggested that the benchmark dose lower confidence limit (BMDL) of UCd for renal effects ranged from 2.5 μg/g cr to 10.3 μg/g cr in males and 1.4 μg/g cr to 11.4 μg/g cr in females [13–16]. Another study reported that the BMDLs of UCd were between 0.5 μg/g cr to 1.2 μg/g cr in a Swedish population [17], whereas some other researchers reported BMDLs ranging from 0.44 μg/g cr to 12.18 μg/g cr in Chinese populations [18–21]. The BMDL values may have been significantly affected by many factors, such as the sample population and Cd exposure levels, thereby necessitating further studies to identify the possible factors affecting the BMD evaluation.

Itai-Itai disease primarily results from cadmium (Cd) exposure and is known as one of the four major pollution diseases of Japan. Considering that Cd pollution is more serious in certain areas of China than in Japan, environmental epidemiological studies on Cd pollution in China clearly warrant more attention. However, there is still a lack of information on the reference level of Cd exposure for the adverse health effects in the general Chinese population. The limited studies on the Chinese population used a much smaller sample size than the sample sizes of the Japanese population reported in other studies (374 to 790 participants vs. 1270 to 3103 participants) [14–17, 19–21].

In the present study, a large sample survey of 6103 participants from five Cd-polluted provinces in China was performed to identify the reference level of UCd for renal dysfunction, with urinary β₂-microglobulin (Uβ₂-MG) serving as the renal effect biomarker. The BMD values of UCd for renal dysfunction in the population of each study province were evaluated individually to study the effect of gender and the sample population on the BMD. Furthermore, the overall BMD of UCd for renal dysfunction in the Chinese population was assessed by combining the five data sets from all 6103 subjects.

The BMD of UCd for renal dysfunction in the population of five different Chinese provinces was assessed individually using Uβ₂-MG as the effect indicator, and the results showed that the estimated BMD value was significantly affected by the participants’ geographic region and gender. For the first time, the BMD of UCd for kidney dysfunction in the Chinese population was further evaluated by combining the five data sets from all 6103 subjects. The overall BMDLs of UCd for Uβ₂-MG with an excess risk of 10 % was 2.00 μg/g cr in males and 1.69 μg/g cr in females of the total sample population. These values were markedly lower than the reference level of 5 μg/g cr for Cd-related kidney effects recommended by the World Health Organization (WHO) [29].

Reports on individuals living in Cd-polluted and non-Cd-polluted areas have shown a close relationship between UCd excretion and the total body burden of Cd or lifetime Cd intake [13]. Therefore, UCd is considered to be a useful indicator of the internal dose of Cd exposure. Uβ₂-MG was chosen as an indicator for Cd-induced renal effects [13, 14, 18]. The Uβ₂-MG level of 1000 μg/g cr, which is from GB/T 17221–1998 in China, was used as the cut-off value. The threshold level for Uβ₂-MG that is associated with the conversion from reversible to irreversible renal damage was determined to be 1000 μg/g cr of Uβ₂-MG excretion [32]. Kobayshi et al. employed 84 % and 95 % of the upper limit values of Uβ₂-MG of the non-smoking population as the cut-off values [11]. Usually, the 95 % upper limit values, which were 994 μg/g cr for men and 784 μg/g cr for women in their study, were used. When 95 % of the upper limit value or 1000 μg/g cr were used as a cut-off value, the value was the level over which the renal damage may be irreversible [13].

It is difficult to arrive at a conclusion on the relationship between the mean UCd and Uβ₂-MG levels and the corresponding prevalence of the elevated UCd and Uβ₂-MG levels by geographic regions. In some study areas, e.g., Hubei and Guangdong, the high UCd levels in the sample population corresponded well with the high Uβ₂-MG levels as well as the high prevalence of the elevated UCd and Uβ₂-MG levels (Table 1). Notably, the results observed in some other areas are different. The UCd level and the corresponding prevalence of elevated UCd levels in the subjects from Guizhou, which had more serious environmental [7] and food (e.g., rice) Cd pollution (detailed information will be reported in another study), were significantly higher than that observed in the other study areas (Table 1). However, the Uβ₂-MG levels of the subjects in the Guizhou province were the lowest, despite their high Cd exposure levels. The corresponding prevalence of the elevated Uβ₂-MG levels was also lower than that observed in most other study areas. Previous studies also showed that the Uβ₂-MG levels of a sample population in a slightly exposed area were even higher than that observed in another highly exposed area (699 vs. 507 μg/g cr), whereas the UCd levels of the subjects in the former were significantly lower (1.4 vs. 7.9 μg/g cr) [33]. These results may be due to large differences of the ethnic group and lifestyle of the study population from different geographic regions. Even so, the Chi-squared linear trend test (p < 0.05) demonstrated the positive correlation between the prevalence of renal dysfunction and the UCd levels in the sample population from each geographic region. A positive correlation has also been reported in some previous studies of the Chinese and Japanese populations [34, 35], which indicates that Uβ₂-MG may be an ideal biomarker for Cd-induced renal effects.

Thus far, the critical UCd levels for kidney effects have displayed large variations. A study based on the Swedish population showed that the BMDLs of UCd related to renal dysfunction ranged from 0.5 μg/g cr to 1.2 μg/g cr [17]. The BMDLs observed in Japanese samples varied from 1.4 μg/g cr to 11.4 μg/g cr [13–16]. The BMDLs of 2.13–4.85 μg/g cr were obtained based on two sets of data from occupational epidemiology in China [19]. The BMDLs of UCd for the renal effects were between 0.44 μg/g cr and 2.62 μg/g cr in another Chinese population from Cd-exposed areas [21]. These findings show that the BMDLs for the Cd-induced adverse health effects may be affected by many factors, such as race, effect biomarkers, sampling differences and BMD analysis methods. In the current study, the higher BMDLs of UCd for Uβ₂-MG were observed in both males and females from the Guizhou province compared to the other provinces (Table 3). The selection of the sample populations affected the final BMDL values when a similar number of subjects from each province and the same race, effect biomarker, sampling and analytical procedures and BMD methods were considered. The present results showed that the geographic region selection significantly affected the BMDL evaluation. Because there was little difference between the prevalence of Uβ₂-MG in the males and females of the Hubei (38.8 %–40.3 %) and Guangdong (40.3 %–41.3 %) provinces, the BMD values were quite similar (1.00 to 1.13 μg/g cr vs. 1.10 to 1.33 μg/g cr). The frequency that exceeded the standard of Uβ₂-MG in the sample populations of Gansu, Guizhou and Guangdong ranged from 19.5 % to 31.6 %, which were significantly lower than 38.8 % to 42.7 % in the other two provinces (Table 1). The corresponding BMDLs were higher than those observed in the Hubei and Guangdong (1.37 to 3.66 μg/g cr vs. 1.00 to 1.33 μg/g cr). Previous studies reported that the prevalence of Uβ₂-MG in females from Cd-polluted areas [13] and non-polluted areas [30] was 30.4 % and 12.9 %, respectively, and the corresponding BMD values of UCd were also quite different (1.5 μg/g cr vs. 3.3 μg/g cr) [13, 30]. Notably, the slightly exposed region (Hubei) showed the lowest BMDs for both males and females, whereas the highest values were observed in Guizhou. These results may be partially due to significantly higher Cd-exposure levels in Guizhou. Subjects who are living in highly polluted areas may become somewhat more resistant to Cd toxicity than those in slightly polluted areas. In addition, large differences of the ethnic groups as well as different lifestyles may also affect Cd’s toxic effect. More than 30 % individuals of the total population in Guizhou belong to minority ethnic groups, while this value in Hubei is only approximately 4 % (i.e., most of local residents in Hubei are the Chinese Han nationality).

In addition, the BMDLs of UCd for renal dysfunction were related to gender in addition to the sample population. Previous findings on the effects of gender on BMD values were different. In some studies, the BMDLs for Cd-induced renal dysfunction in females were higher than those in males [13, 16, 36]. However, other studies reported that the BMDLs were lower in females [14, 15], which indicated that the female subjects appear to be somewhat more sensitive to Cd damage than the male subjects. In the present study, the BMDLs of UCd for Uβ₂-MG in females were lower than those of the males from each province, with the exception of the Yunnan province (Table 3 and 4). The overall BMDLs of the total population from all five provinces were also slightly lower in females (Table 3). Therefore, these results provide additional support to the latter findings.

The data obtained from all 6103 subjects from the five study areas were used to evaluate the reference level of Cd-induced kidney effects. This was performed to increase the statistical and biological confidence in the calculated BMD. The present investigation on the BMD has a greater statistical power than many previous studies of this type, particularly those for Chinese populations. The number of subjects in the present study was as high as 6103, much larger than the number (N = 374–790) of Chinese subjects in previous studies [18–21]. The sample size was also relatively larger than the Japanese populations in the previous studies of Kobayashi et al. (N = 2778) [13], Suwazono et al. (N = 1270) [15] and Suwazono et al. (N = 3103) [16]. Although the number of subjects in a previous BMD study of the Japanese population reached 6032 [36], these subjects came from different surveys that were completed separately in 1982 and 1998–2000 [34, 36]. In the present study, all participants were in the same survey and the same time frame. Furthermore, the exposure range was relatively broader than that of the previous studies [13, 30]. As reported in a previous study based on the population in non-polluted areas, the frequency of subjects with UCd concentrations above 7 μg/g cr was approximately 8 % [13]. The frequency in another study based on the population in polluted areas was 19 % [30]. In the current study, the frequency of subjects with UCd concentrations above 6 μg/g cr reached 40 %, based on the survey in the polluted provinces, which indicated much broader Cd-exposure levels in these subjects. In the present study, the estimated BMDL of UCd for Uβ₂-MG with BMRs set at 10 % was 2.00 μg/g cr in males and 1.69 μg/g cr in females. The BMDLs obtained in the present study were significantly lower than other reports on the BMDLs of 3.99–12.18 μg/g cr in the Chinese population [18] and 2.4–11.4 μg/g cr in the Japanese population [13, 16, 32]. The present results were similar to those obtained in Swedish females (0.5–1.2 μg/g cr) [17], the Japanese population (1.4–2.6 μg/g cr) [15], the Chinese population that was also exposed to arsenic (0.9–1.2 μg/g cr) (Hong et al., [37]), and another Chinese population (0.44–0.53 μg/g cr) [21].

Previous studies have reported that the mortality risk significantly increases among the subjects with UCd values higher than 3 μg/g cr after age adjustment [38]. Cd toxicity has already been evaluated by several international bodies, including the International Agency for Research on Cancer, the Joint Food and Agriculture Organization (FAO)/WHO Expert Committee on Food Additives (JECFA) (2003, 2006, 2005, 2010) [39, 40], the European Commission [41], and the Agency for Toxic Substances and Disease Registry [42]. In 2010, the 73rd JECFA established a provisional tolerable monthly intake (PTMI) of 25 μg/month/kg body weight (bw) based on the findings of previous epidemiological studies [40]. The European Food Safety Authority recommended a considerably lower tolerable weekly intake of 2.5 μg/week/kg bw (i.e., nearly 10.7 μg/month/kg bw) based on an analysis of the relationship between UCd levels and tubular proteinuria [43]. The Scientific Committee on Toxicity, Ecotoxicity and the Environment of EC reported that adverse health effects may occur even at a lower level of 0.5 μg/g cr [44]. Recently, several guidance values for Cd exposure have become available, such as bio-monitoring equivalents (2 μg/g cr) [45] and human bio-monitoring values (1.0 μg/L for adults; 0.5 μg/L for children/adolescents) [46]. These values were significantly lower than the reference level of 5.0 μg/g cr for Cd-induced kidney dysfunction recommended by the WHO [29] and MOHC [47]. In the present study, the estimated threshold values for Cd-induced renal dysfunction were within the range of 1.00 μg/g cr to 3.66 μg/g cr, which provides additional scientific support for identifying a lower biological limit for Cd exposure and a provisional tolerable monthly intake (PTMI).

This study covers a wide range of ages and Cd exposure levels in both men and women. Furthermore, when considering the larger sample population size than that of previous studies [18–21], we believe that this study improved the accuracy of the BMD of UCd for renal dysfunction in the Chinese population. However, this study has limitations that should be considered in future studies. The results of Japanese studies showed that the prevalence of Uβ₂-MG is increased with increasing age in both sexes. The obtained BMDL gradually decreased with increasing age, which indicated that the margin between the threshold level and average excretion level of urinary Cd was small in the older population in Japan. Therefore, it is important to evaluate the threshold level of Cd exposure in the general Chinese population, while accounting for an age effect to prevent Cd’s toxic effects. Furthermore, there are 56 ethnic groups from more than 30 provinces, municipalities and autonomous regions in China. Large differences in the ethnic group, lifestyle, and diet and health history of the study population may also be important factors related to Cd-induced adverse health effects. Further studies are needed to reach a conclusion.

The BMD value of UCd for Uβ₂-MG was related to the participants’ geographic region and gender. The effect of the participants’ geographic region on the BMD estimate may be partially attributable to the large differences in Cd exposure level, ethnic group, lifestyle and diet of the sample population in these study areas. The overall BMDLs of UCd for Uβ₂-MG with an excess risk of 10 % in this large sample survey of 6103 subjects was be 2.00 μg/g cr in males and 1.69 μg/g cr in females, which were significantly lower than the reference level for Cd-related kidney dysfunction suggested by the WHO. These findings provide additional scientific support for reconsidering the threshold exposure value and PTMI for Cd to protect people from Cd-induced adverse health effects.