**1. Introduction**

The kidney is a target organ of external cadmium (Cd), and renal tubular dysfunction is the most prevalent adverse health e ffect induced by Cd exposure. Significant associations between renal tubular dysfunction, indicated by increasing urinary low molecular protein and Cd exposure markers such as urinary Cd, were reported in residents of Cd-polluted areas in Japan [1].

Cd in rice (RCd) is believed to be a good exposure marker in Japan because Tsuchiya and Iwao (1978) [2] reported that Japanese residents took half or two-thirds of Cd from rice in a study conducted in the 1970s. Nogawa et al. (1989) [3] showed a prevalence of renal tubular dysfunction indicated by an increase in urinary β2–microglobulin (β2–MG) ≥1000 μg/g creatinine (g Cr) with increasing residential duration (years) as well as RCd produced in the communities, suggesting that lifetime Cd intake (LCd), defined from both factors of RCd and residential duration, is a useful indicator of external dose to show the dose–response relationship, with health e ffects induced by Cd exposure.

Significant dose–response relationships between LCd and urinary metallothionein in the Cd-polluted Kakehashi River basin [4,5] and between LCd and urinary glucoproteinuria in the Jinzu River basin [6] were reported in residents that were environmentally exposed to Cd in Japan.

In our previous studies, following-up subjects from the Kakehashi River basin who participated in health impact surveys in 1981–1982 for 9 and 15 years, we reported increased mortality for all causes, cardiovascular diseases, and renal diseases in the subjects with Cd-induced renal tubular dysfunction indicated by urinary β2-MG, protein, glucose, amino acids, and retinol-binding protein [7–12]. An investigation of the relationship between the urinary Cd levels and mortality in the 15- and 22-year follow-up studies showed increased mortality for all causes, including renal diseases and heart failure, among subjects with urinary Cd levels ≥10 μg/g Cr in the 1981–1982 survey compared with subjects with Cd levels <3 μg/g Cr [13,14].

Relationships between LCd and mortality have been reported in the inhabitants of the Cd-polluted Jinzu River basin in three studies [15–17], but these relationships have not been demonstrated among residents of the Kakehashi River basin in our previous studies. Therefore, we extended the follow-up period to 35 years and analyzed dose–response relationships between LCd and mortality for causes of deaths to clarify the effect of environmental Cd exposure on mortality risks, particularly for renal diseases.

#### **2. Materials and Methods**

The Kakehashi River basin is one of the Cd-polluted areas in Japan, and it includes 700 ha of rice paddy fields. The pollution was due to mining activity that included a full-scale operation near the Kakehashi River that started in 1930 and continued until mining ceased in 1971. Rice with Cd > 0.4 (ppm) was detected in 12 of 22 villages in this area, and restoration of polluted paddy fields was undertaken from 1977 to 1988.

A total of 2602 subjects ≥50 years of age (1169 men and 1433 women) living in the Cd-polluted Kakehashi River basin, whose residential periods were ascertained in the 1981–1982 health impact survey, were enrolled in the present follow-up study (82% participant rate). LCd was estimated for them from RCd levels produced in their living communities and residential period (days) until 1981–1982 using the following formula: LCd = (RCd × 333.5 g + 34 μg) × residential period in Kakehashi River basin + 50 μg × residential period in nonexposed areas [3] on the assumption of 333.5 g for daily rice intake obtained by duplicated diet method, 34 μg for daily Cd intake from food other than rice from the polluted area [18], and 50 μg for daily Cd intake from nonpolluted areas [19].

Their mean levels of age, LCd, urinary Cd, and urinary β2–MG, examined at the survey in 1981–1982, with test results by *t*-test to compare them between sexes, are shown in Table 1. At this time, urinary Cd and β2-MG were transformed to log10 values for analysis because of the improvement of distribution. Positive rates for urinary Cd and urinary β2–MG are also shown in Table 1. Significantly greater mean values of LCd, urinary Cd, and urinary β2–MG and greater positive rates of urinary Cd and β2–MG were found in women compared with men.

Cd in urine samples was measured by flameless atomic absorption spectrometry after ashing with HNO3, HsSO4, and HClO4 and extraction with ammonium pyrrolidine dithiocarbamate (APDC) and methyl isobutyl ketone (MIBK) [20]. Freeze-dried standard reference material for toxic elements in urine (The National Bureau of Standards, Washington, DC, USA) was used to test the accuracy and precision of the analytical method of urinary Cd. Cd concentration corrected by Cr in urine was used for analysis.


**Table 1.** Levels of external cadmium (Cd) exposure and renal dysfunction examined in the 1981–1982 health impact survey and the follow-up rates and period until 2016.

1 Geometrical mean and geometrical standard deviation, 2 Participants were 1121 men and 1333 women. N: number of subjects, SD: standard deviation.

A 35-year prospective follow-up survey of subjects was conducted from the day of their initial examination at the health impact survey in 1981–1982 until November 2016. The survival status (alive or dead), the date of death, place of residence (still residing or not in the target area), and the date of death, if applicable, were determined for 2527 residents (1149 men and 1378 women) from family registry records of all subjects with the cooperation of the Prefecture Public Health O ffice and City Municipal O ffice. Then, individual causes of death were ascertained from vital statistics by linking them to health survey data, with survival status based on birthday, death day, gender, and address after getting the permission of the Ministry of Health in Japan. The final number of subjects available for mortality analysis was 2496 (1135 men and 1361 women, 96% follow-up rate) and their mean follow-up period (months) are shown in Table 1. Causes of deaths were classified according to the 9th and 10th Revised International Classification of Diseases. The study protocol was approved by the Ethics Committee of Kanazawa Medical University (No-212, 17 June 2014).

Increased mortality risks for all causes and major causes of deaths, including renal diseases and renal failure, associated with a 1-g increase of LCd or a 10-μg/gCr (1 of log10 transformed value) increase of urinary Cd were investigated after adjusting for age using Cox's proportional hazard model. This model is one of regression analysis used for investigating associations between the survival time from the date of the baseline survey to the endpoint of subjects and relevant factors. In addition, to investigate which exposure marker, LCd or urinary Cd, contributed more to increased mortality for all causes and renal diseases, the stepwise elimination method based on Wald was used in Cox's mortality risk analysis. The SPSS (version 21.0) software package for Windows (SPSS Inc., Armonk, NY, USA) was used for statistical analysis.
