**4. Discussion**

In a 35-year follow-up study, an increase in LCd mortality from renal diseases, particularly for renal failure, in women ≥50 years of age, living in the environmentally contaminated Kakehashi River basin in Japan, suggests a dose–response relationship between the external dose of Cd and mortality from renal diseases in women.

Previously, we conducted follow-up studies of residents of the Kakehashi River basin in Japan who had participated in a health impact survey in 1981–1982 and reported the poor life prognosis of subjects associated with an increase in urinary Cd in a 15-year follow-up study [13] and a 22-year follow-up study [14]. In particular, for renal diseases, the mortality ratio was significantly higher in women in the higher urinary Cd group, with a 10- [13] or 20-μg/g Cr cut-o ff level [14] compared with the lower Cd-exposure group, albeit no dose–response relationships were shown between urinary Cd and mortality risks for renal diseases. In the present study, however, a 10-μg/g Cr increase of urinary Cd increased mortality not only for all causes but also for renal diseases.

LCd was used as an external Cd exposure maker to reflex the Cd exposure dose from the environment over a lifetime. In the present study, an increase in LCd was significantly associated with increased mortality from renal diseases in women, albeit no association was found in our previous 22-year follow-up. In women, we detected 26 deaths from renal diseases, which increased the statistical power in the present 35-year follow-up, and 14 deaths from renal diseases in the previous 22-year follow-up. Moreover, LCd was selected from two Cd exposure markers as a factor that contributed to increased mortality from renal diseases, suggesting that LCd showed a better dose–response relationship with mortality from renal diseases during a long time observation compared with urinary Cd. Since urinary Cd shows a sharp increase with increasing external Cd exposure only before the occurrence of renal tubular dysfunction [21,22], LCd might reflex the Cd dose that induces fatal renal diseases more than urinary Cd among subjects with a high prevalence of renal tubular dysfunction.

In the Jinzu River basin, Toyama, in Japan, where itai-itai disease is endemic, an increased mortality risk associated with LCd intake was suggested in residents who participated in the health survey in 1967 [15,16]. Recently, it was reported that an increase of LCd increased the mortality for all causes, renal disease, and the toxic e ffects of Cd (itai-itai disease) in the 26-year follow-up of female participants to the health impact survey in the Jinzu River basin conducted in 1979–1984 [17]. These results from the Jinzu River basin are consistent with our present results, indicating that LCd dose-dependently increased mortality over long time observation, particularly for renal diseases in women, albeit their follow-up period was shorter than ours. As their mean (SD) of LCd was 3.5 (1.9) g for men and 3.0 (1.5) g for women [17] and much higher than our values, 2.7 (1.6) g for men and 2.5 (1.6) g for women, this suggests heavier Cd contamination in rice in the Jinzu River basin. Although their hazard ratio for renal diseases for a 1-g increase of LCd was 1.25 and lower than that (1.49) in our present study of the Kakehashi River basin, direct causes of deaths might be renal failure in some cases diagnosed as the toxic e ffects of Cd (itai-itai disease), suggesting an underestimation of the mortality risk ratio for renal diseases in the Jinzu River basin. Moreover, increased mortality associated with an increase in LCd was also observed for all causes in the Jinzu River basin in women because of the much higher rate of death for specific causes of death to Cd exposure, such as renal diseases and the toxic e ffects of Cd (14.1%), compared with our results in the Kakehashi River basin (3.8%).

The association between LCd and mortality for renal diseases was significant only in women in both Cd-polluted areas, suggesting that women who took Cd from rice for a long time have more risk of dying from renal failure compared with men who took Cd at the same level. Since Cd absorption is higher in women because of lower body iron stores [23,24], we compared urinary Cd, uncorrected by Cr, in each LCd category, but no significant di fference was found in any categories between men and women. However, urinary-uncorrected ß2–MG increased with increasing levels of LCd and was significantly higher in women compared with men when LCd ≥ 4.5 g. These findings sugges<sup>t</sup> that renal tubular dysfunction may aggravate women more than men and cause renal failure deaths in women that are highly exposed to Cd.

In the study by Nogawa et al. [17], mortality was analyzed with three covariates, including age, smoking status at present, and hypertension history, while we performed our analysis with only one covariate of age, which is a limitation of our present study. However, no e ffects of smoking status

or hypertension history were observed on mortality from renal diseases and the toxic e ffects of Cd of residents of the Jinzu River basin, suggesting that the confounding factors of smoking status and hypertension history may be limited in the association between mortality from renal diseases and LCd in the women of our study. In addition, the strength of the present study may be analyzing the effects of urinary Cd and LCd and internal and external Cd exposure markers on mortality at the same time in the same population. We showed that an increase in urinary Cd also increased mortality from renal diseases in women, as well as an increase in LCd, surely indicating the fatal renal toxic e ffects of environmental Cd exposure on women.

In Cd-polluted areas other than Japan, studies to investigate the dose relationship between mortality and external Cd exposure are limited. Only in Belgium, significantly higher lung cancer risk (incidence of fatal cancer) was reported for a doubling of Cd concentration in the soil of residential areas in a prospective population-based study from 1985 to 2004, which targeted the Flemish population who participated in the Cd in Belgium Study (CadmiBel) [25]. In almost all studies to investigate mortality risks for all causes [26], for cancer [27,28], and for cardiovascular diseases [29,30], internal Cd exposure markers such as urinary Cd or blood Cd were used as exposure markers because the exposure level is within background levels in their targeted population and it was di fficult to estimate LCd.
