General characteristics of study participants. Table 1 shows the general characteristics of the study participants. We studied a total of 298 study participants, consisting of 36.6% males and 63.4% females. The composition of males and females in each survey area are as follows: 28 males (37.8%) and 46 females (62.2%) in the LEAMM area, 27 males (39.7%) and 41 females (60.3%) in the HEAMM area, 43 males (35.5%) and 78 females (64.6%) in the refinery area, and 11 males (31.4%) and 24 females (68.6%) in the control area. Accordingly, there were more females than males in every area. The average age of the participants was 70.3 years, which did not differ between the exposure and control areas. The average period of residence of all participants was 33.0 years, and in the LEAMM and HEAMM areas were 42.0 years and 39.0 years, respectively, which were longer than those of residents in the refinery and control areas. According to the survey for smoking and alcohol consumption, non-smokers (72.48% overall) and non-drinkers (49.65% overall) accounted for the largest groups in all areas. The residents in the LEAMM and refinery areas used water purifiers and tap water as their main drinking water sources, whereas those in the HEAMM and control areas used ground water in addition to water purifiers and tap water. Overall, 31.54% of residents consumed at least half of their rice from locally grown sources, and this proportion was highest in the HEAMM area at 64.71%.
Comparison between heavy metal concentrations of study participants and general population group. The geometric mean concentrations (95% CI) of blood Pb, Cd, and urinary Cd in all exposure areas were 2.12 (1.99, 2.25) μg/dL, 1.89 (1.75, 2.04) μg/L, and 2.11 (1.90, 2.35) μg/L, respectively.
By area, the blood Pb concentrations (95% CI) in the LEAMM, HEAMM, and refinery areas were 2.66 (2.52, 2.81), 1.77 (1.55, 2.02), and 2.04 (1.85, 2.25) μg/dL, respectively, and that in the control area was 1.18 (1.00, 1.40) μg/dL. Hence, concentrations were higher in the exposure areas than in the control area. Concentrations of blood Hg and urinary total As were similar in all areas except the LEAMM area; however, the difference was not statistically significant (Table S1).
The geometric mean concentrations (95% CI) of blood Cd in the LEAMM, HEAMM, and refinery areas were 1.37 (1.22, 1.53), 1.93 (1.61, 2.33), and 2.27 (2.07, 2.50) μg/L, respectively, and that in the control area was 0.89 (0.78, 1.02) μg/L. Thus, the exposure areas showed higher concentrations than the control area, and the refinery area showed the highest blood Cd concentration. The geometric mean concentrations of urinary Cd in the LEAMM, HEAMM, and refinery areas were 2.87 (2.36, 3.49), 2.46 (2.06, 2.93), and 1.61 (1.37, 1.89) μg/L, respectively, and that in the control area was 1.11 (0.85, 1.44) μg/L. Thus, the exposure areas showed higher concentrations than the control area, and the LEAMM showed the highest urinary Cd concentration. The geometric means of urinary Cd adjusted with urinary creatinine values were also significantly higher in the exposure areas than the in control area (Table S1).
Based on data of the 2017 KNHANES and the third KoNEHS, we calculated the geometric means of blood Pb, Hg, and blood and urinary Cd, and compared them with the concentrations of the study participants. The blood Pb concentrations in all exposure areas were higher than the KNHANES geometric mean (1.68 μg/dL). However, this concentration was higher than the KoNEHS geometric mean (1.81 μg/dL) in only the LEAMM and refinery areas, whereas that in the HEAMM area was lower (Fig. 1a). Blood Hg concentrations were higher in the LEAMM and HEAMM areas than the KNHANES and KoNEHS geometric means but lower in the refinery and control areas (Fig. 1b). All areas except the control area showed higher blood Cd concentrations than the KNHANES geometric mean (1.01 μg/L) (Fig. 1c). Urinary Cd concentrations were higher in all surveyed areas than the KoNEHS geometric mean (0.49 μg/L) (Fig. 1d).
ORs for exceeding reference values of the general population group. Table 2 shows a comparison of the ORs of exceeding the reference value (RV95) for Cd with the general population by area.
The ORs (95% CI) for blood Cd in the LEAMM, HEAMM, and refinery areas in Model 1 were higher than the control area by 5.89 times (95% CI: 1.26, 27.63), 23.11 times (95% CI: 4.95, 107.81), and 35.66 times (95% CI: 7.94, 160.26), respectively, and those in Model 2 were higher by 5.42 times (95% CI: 1.14, 25.79), 21.39 times (95% CI: 4.33, 105.65), and 38.29 times (95% CI: 8.24, 177.90), respectively. The OR (95% CI) for urinary Cd in the LEAMM, HEAMM, and refinery areas in Model 1 were higher than the control area by 8.46 times (95% CI: 3.33, 121.50), 6.03 times (95% CI: 2.40, 15.17), and 2.48 times (95% CI: 1.08, 5.71), respectively, and those in Model 2 were higher by 7.08 times (95% CI: 2.73, 18.39), 2.87 times (95% CI: 1.05, 7.89), and 1.82 times (95% CI: 0.75, 4.39), respectively. The ORs of exceeding the reference value for blood and urinary Cd were more significant in Model 1 than in Model 2.
Analysis of correlation between renal function indicators and Cd. We used two-way graphs (scatterplots) to examine the correlation between the renal function indicators and blood/urinary Cd, the results of which are shown in Figures. 2 and 3.
Blood Cd and NAG showed a statistically significant (p = 0.045) positive correlation (r = 0.244) in the HEAMM area, while no statistically significant correlation was observed for the other areas (Fig. 2a–d). Blood Cd and β2-MG levels showed a positive correlation (r = 0.311) in the HEAMM area, but the correlation was weak in the remaining areas (Fig. 2e–h). The HEAMM area showed the highest negative correlation between blood Cd level and eGFR (r = -0.464) and the refinery area showed a lower negative correlation (r = -0.207). However, the LEAMM and control areas showed either an extremely low correlation or no statistically significant correlation (Fig. 2i–l).
Urinary Cd and NAG levels showed a statistically significant positive correlation in all areas (Fig. 3a–d). The control area showed the highest correlation (r = 0.687, p < 0.001), and the HEAMM area showed the lowest correlation (r = 0.250, p = 0.040). The correlation between urinary Cd and NAG levels was higher in the refinery area than in the LEAMM and HEAMM areas. β2-MG and urinary Cd levels showed a statistically significant (p = 0.024) positive correlation (r = 0.206) in the refinery area, and a weak correlation in the remaining areas (Fig. 3e–h). The correlations between urinary Cd level and eGFR were weak or not statistically significant in all areas (Fig. 3i–l).
Analysis of relationships between renal function indicators and heavy metals. To examine the relationships between renal function indicators and heavy metals, we performed a logistic regression analysis on the values greater or less than the reference values of NAG, β2-MG, and eGFR, which are renal function indicators (Table 3). We did not distinguish between the areas. and used sex, age, period of residence, and BMI as the adjustment variables. Then, we investigated the relationship with log-transformed heavy metal concentrations. As the urinary Cd concentration increased, the probability of exceeding the reference value of NAG became 5.27 times higher (OR: 5.27, 95% CI: 2.71, 10.27), and BMI (p < 0.05) showed a significant correlation with NAG level. In addition, as the blood Cd concentration increased, the probability of exceeding the reference value of β2-MG was 2.37 times higher (OR: 2.37, 95% CI: 1.15, 4.90), and age (p < 0.05) showed a significant correlation with β2-MG level. As the concentration of blood Cd increased, the probability of falling below the reference value of eGFR became 4.08 times higher (OR: 4.08, 95% CI: 1.54, 10.77).