Dietary cadmium exposure, risks to human health and mitigation strategies

Abstract Cadmium (Cd) is a toxic and carcinogenic pollutant widely distributed in the environment. Dietary intake is the main source of Cd exposure for the nonsmoking population. Assessment of dietary Cd intake provides a pathway to predict Cd body burden and potential health effects. Kidney has been considered as the most sensitive target of chronic Cd exposure. Because of the proportional relationship between Cd accumulation in kidney and Cd excretion via urine, urinary Cd (UCd) has been used as a biomarker of Cd exposure. Here, we review the dietary Cd intake levels in different countries, summarize the global food Cd concentrations reported in both market basket and field surveys, discuss UCd levels among different populations, and illustrate the associations between dietary Cd intake and UCd levels in population-based studies. Recommendations for the prevention and reduction of Cd exposure through anthropogenic inputs and the food chain are also proposed. This review presents a worldwide overview of Cd exposure status through diet for the general population as well as those living in contaminated areas, and provides evidence for policy makers to protect humans from Cd exposure and related health effects. GRAPHICAL ABSTRACT


Introduction
Cadmium (Cd) is classified as a Group I human carcinogen by the International Agency for Research on Cancer (IARC, 2018). Occupational or high level environmental exposure to Cd may cause high blood pressure, kidney damage, osteoporosis, and increased risk of cancers in the lung, endometrium, bladder, prostate, and breast Cho et al., 2013;Satarug et al., 2010). Numerous studies have also addressed the importance of chronic exposure to Cd as an environmental hazard worldwide. For example, excess cancer mortality was found to be associated with environmental Cd exposure in prospective studies conducted in Japan, the US, and Europe (A˚kesson et al., 2008;Menke et al., 2009;Nawrot et al., 2006), raising concerns about potential adverse health effects from Cd exposure at environmental levels.
For the general population, environmental Cd exposure occurs through multiple pathways, including food ingestion, smoking, inhalation, and drinking water. Dietary intake is the primary route of Cd exposure for the nonsmoking population, accounting for approximately 90% of the total Cd exposure (Satarug et al., 2010). Less than 10% of the total exposure occurs via inhalation of Cd contaminated air or drinking water (Olsson et al., 2002). Cigarette smoking is an additional source of Cd exposure for smokers, since tobacco leaves naturally accumulate high levels of Cd (Satarug et al., 2013).
Following exposure, most of the Cd absorbed by the human body is excreted via feces and urine. Absorption of Cd in the human body is relatively low (5-10%), and about half of the absorbed Cd is accumulated in kidney (Nordberg et al., 2007). As a result of this preferential accumulation, kidney is the most sensitive organ, with damages being reported when Cd concentration in the renal cortex exceeds 200 mg/kg (Nordberg et al., 2007). Since Cd excretion via urine is proportional to Cd accumulation in the kidney cortex, urinary Cd concentration (UCd) has been used as a suitable biomarker of the body Cd burden (Bernard, 2016;Menke et al., 2009). Threshold values of UCd have been established by the European Food Safety Authority (EFSA) and Food and Agriculture Organization/World Health Organization (FAO/WHO) to protect against kidney damage (EFSA, 2011;FAO/WHO, 2010). However, increasing evidence suggests that UCd at levels below the established thresholds may still pose adverse health effects (Gallagher & Meliker, 2010;Satarug et al., 2010). Analysis of data from the Third National Health and Nutrition Examination Survey (NHANES III) (n ¼ 13,663), a cohort representative of the US population, showed that the geometric mean of UCd level was 0.25 and 0.35 lg/g creatinine in men and women, respectively. Each 2-fold increase in UCd was associated with a 26% and 21% higher hazard ratio of cancer mortality among men and women, respectively (Adams et al., 2012). Similarly, a 4.3-fold increase in cancer mortality was observed among 6,558 men from NHANES 1988-1994 who had UCd > 0.48 lg/g creatinine compared to <0.21 lg/g creatinine (Menke et al., 2009).
The studies mentioned above have raised widespread concern about food safety and the associated health risk caused by Cd exposure. Although Cd contamination has been identified as an issue, it is unclear whether the consumption of Cd contaminated food would result in elevated UCd levels and pose adverse health risks in the general population and those living in polluted areas. In this review, we summarize 1) dietary Cd intake levels in different countries; 2) global food Cd concentrations reported in both market basket and field surveys; 3) UCd levels among different populations; 4) associations between dietary Cd intake and UCd levels in populationbased studies; and 5) strategies to reduce human Cd exposure. Knowledge gaps and future research needs are also highlighted.

Tolerable Cd intake
In order to reduce Cd exposure and protect humans from kidney disease, the Joint FAO/WHO Expert Committee on Food Additives (JECFA) established a provisional tolerable weekly intake (PTWI) for Cd of 7 lg/kg body weight (bw)/week in 1989 at the 61 st JECFA meeting (FAO/ WHO, 1989). The value was derived based on the critical renal Cd concentration of 100-200 lg/g wet weight, which corresponds to an UCd threshold of 5-10 lg/g creatinine. However, after including epidemiological studies of environmental Cd exposure, an excess prevalence of renal tubular dysfunction was observed when UCd was lower than 2.5 lg/g creatinine. Given the long half-life of Cd in the human body, the JECFA recommends the use of monthly unit as the minimum unit in dietary Cd intake assessment. Thus, the PTWI was lowered to 25 lg/kg bw/month at the 73 rd JECFA meeting in 2010, which corresponds to an intake of 50 lg/day assuming a body weight of 60 kg (FAO/WHO, 2010). In comparison, the EFSA Panel on Contaminants in the Food Chain recommended a much lower tolerable weekly intake of 2.5 mg/kg bw/week, which corresponds to an intake of 21.4 lg/day for a 60 kg person. The value was proposed with the target of UCd to remain below 1 lg/g creatinine in 95% of the population by age 50 based on paired data of Cd intake and UCd concentrations in a cohort of Swedish women (EFSA, 2009).

Dietary Cd intake in different countries
The tolerable intake levels recommended by JECFA and EFSA provide benchmarks for assessing the dietary Cd exposure in different countries. Total Diet Studies (TDS), duplicate diet analysis, and dietary assessment have been used to estimate dietary exposures among the general population from different countries (Figure 1). In the US, an average dietary Cd intake of 4.6 lg/day was estimated by matching two-days dietary intake data from NHANES 2007-2012 with food Cd concentrations from the Food and Drug Administration (FDA) 's TDS 2006(Kim et al., 2018. For Canadian, the average dietary Cd intake was estimated to be 13.2 lg/day, based on Cd concentration data from foods sold in Canada in 2007(Health Canada, 2007. The lower dietary Cd intake estimation in the US population compared to the Canadian may be related to the Figure 1. Dietary Cd intake levels (lg/day) in the general populations across different countries (Bars represent the mean value, dots represent the 95 th percentile of Cd intake level). The red dash line and solid line indicate the tolerable intake of 2.5 lg per kg body weight per week and 25 lg per kg body weight per month set by EFSA and JECFA, respectively, which are equivalent to 21.4 lg/day and 50 lg/day assuming a body weight of 60 kg. Detailed information of intake values, sampling year, and references is provided in Supplementary material Table S1. reduced Cd concentrations in the US food supply due to the new legislation in certain US states limiting Cd content of phosphate fertilizers (Kim et al., 2018;Roberts, 2014).
For European countries, a report of the EFSA concluded that the average dietary Cd intake in adults were 12.9-19.1 mg/day assuming a body weight of 60 kg (EFSA, 2012). For the highly exposed adults (e.g. the 95 th percentile), dietary Cd intake ranges between 21.2-41.2 mg/day (EFSA, 2012). These estimates were made by matching the consumption data and food Cd concentration data at individual level for adults. Consumption data from the EFSA Comprehensive European Food Consumption Database covering over 67,000 populations in the 22 Member States and food Cd concentrations obtained by the EFSA during 2003-2011 were used for the estimation. Among these European countries, average dietary Cd intake was highest in Spain (19.1 mg/day), followed by Italy (18.9 mg/day), Ireland (16.9 mg/day), and was lowest in Germany (12.9 mg/day).
Similar or higher dietary Cd intakes have been reported in Asian countries. For South Korean, an average dietary Cd intake of 14.5 lg/day was estimated using food consumption data from the Korean Nutrition Survey and food Cd concentration determined by the Korean Food and Drug Administration (Kim & Wolt, 2011). For Chinese, the 6 th TDS conducted during 2016-2019 covering approximately 86% of the population from 24 provinces estimated an average dietary Cd intake of 17.3 lg/day (Zhao et al., 2022a). For Japanese, dietary assessment conducted during [2001][2002][2003][2004][2005] reported an average dietary Cd intake of 26.4 lg/day for women aged 20-74 years (Itoh et al., 2014). In Bangladesh, the mean dietary Cd intake during 2008-2009 was 34.6 lg/day (Al-Rmalli et al., 2012).
As described above, average dietary Cd intakes in most countries (4.6-19.1 lg/day) were below the limits set by EFSA and JECFA, while dietary Cd intakes in Japan and Bangladesh (26.4-34.6 lg/day) were lower than JECFA's limit of 50 lg/day, but higher than the limit of 21.4 lg/day set by EFSA. For consumers with high exposure (e.g. the 95 th percentile), dietary Cd intakes (21.2-41.2 lg/day) were close to or slightly exceeded the limit set by EFSA. Thus, for these consumers, there is essentially no margin for other sources of Cd exposure if the EFSA's tolerable intake level is not to be exceeded.
Furthermore, increasing evidence suggests that chronic Cd exposure at levels lower than the threshold values set by EFSA or JECFA may still increase the risk of cancers in the endometrium, bladder, and breast (A˚kesson et al., 2008;Julin et al., 2012). A study of 30,210 postmenopausal women in the Swedish Mammography Cohort reported a 2.9-fold increase in the endometrial cancer risk for women who consumed over 15 lg/day of Cd through dietary sources than those below 15 lg/day (A˚kesson et al., 2008). In the same cohort, people with dietary Cd intake levels over 16 lg/day had a significantly higher risk of invasive breast cancer than those with Cd intakes <13 lg/ day, with a rate ratio of 1.27 (95% CI, 1.07-1.50) after adjustment for potential covariates (n ¼ 55,987) (Julin et al., 2012). These studies raise a question regarding whether the current guidelines established by EFSA and JECFA can adequately protect humans from health risks caused by dietary Cd intake, since these limits are based on protection against the non-carcinogenic effects of Cd.
To assess the risk of Cd as a human carcinogen, excess lifetime cancer risk (ELCR) has been estimated to present the incremental probability of an individual developing cancer over a lifetime as a result of dietary Cd exposure. Yuan et al. (2014) performed a health risk assessment of dietary Cd intake by adults in 31 provinces of China. They found that even though the mean dietary Cd intake values were lower than the JECFA limit, the carcinogenic risk for the southern adult population was higher than the safety risk level. Thus, the JECFA limit may not be protective against possible carcinogenic effects.

Temporal and geographical variability in dietary Cd intakes
Besides country-wide dietary intake studies at a given time, it is also important to consider how dietary Cd intake changes over time. Generally, dietary Cd intakes show a decreasing trend in developed countries due to stricter controls on Cd emissions to the environment, but appear to be increasing over time in some developing countries (Moon et al., 2012;Song et al., 2017). In Korea, the average dietary Cd intake decreased slightly from 17.0 lg/day in 1998 to 14.8 lg/day in 2007 based on the Korea National Health and Nutrition Examination Survey (Moon et al., 2012). In a large-scale survey during 2010-2011 covering the whole country of Korea, the average dietary Cd intake level was 11.5 lg/day (Lim et al., 2015), showing a 32% decrease over the 13year period between 1998 and 2011. In China, average dietary Cd intake increased from 13.8 to 32.7 lg/day during the period of 1990 to 2009-2013 due to rapid industrialization and urbanization, with an annual increase of 0.95 lg/day (Wu et al., 2018;Wu & Li, 2015). Similarly, in a large-scale survey conducted during 2011-2015, the average dietary Cd exposure of the general Chinese population was 30.6 lg/day (Song et al., 2017). In 2016-2019, the average Cd intake in China had dropped to 17.3 lg/day, representing a 47% decrease compared with that in 2009-2013, which was likely due to the country's efforts at controlling environmental pollution in recent years (Zhao et al., 2022a).
Another important consideration when assessing country-wide dietary Cd intake is geographical variation. Large variation in dietary Cd intake across different regions of the country is expected. In rural areas of northern Guangdong Province, China, which had a history of mining and smelting activities, dietary Cd intake of 272 lg/day for local residents was much higher than the country average (Zhu et al., 2016a). Among 1,381 female farmers living in five districts in Japan, the geometric means of dietary Cd intake in four Cd contaminated areas (17.7-51.2 lg/ day) were much higher than that in the control area (6.99 lg/day) when assuming a constant contribution from rice consumption (50%) (Horiguchi et al., 2004). Ngo et al. (2012) compared individuals living and working in a metal recycling community and an agricultural (reference) community in Vietnam, finding that the mean dietary Cd intake for the adult population in the recycling village was 4 times higher than that of the reference village, with rice contributing 98% to the overall dietary Cd exposure (Ngo et al., 2012). These studies reveal highly elevated Cd exposure for the residents living in contaminated areas and food consumption as the main contributor to elevated Cd exposure.

Cadmium concentrations in food items
In 2011, the JECFA reviewed Cd concentration data in various food items from 19 European countries and 11 other countries (WHO, 2011). In most food categories, national average Cd concentrations ranged between not detected and 0.04 mg/kg. Higher Cd concentrations were reported for shellfish/mollusks (0.01-4.8 mg/kg), coffee, tea and cocoa (0.0001-1.8 mg/kg), vegetables (including dried) (0.006-1.0 mg/kg), meat and poultry offal (0.006-0.5 mg/kg), and spices (0.006-0.2 mg/kg). Shellfish and mollusks are known to have the ability to accumulate higher Cd from the aquatic environment; however, populations with high Cd intake due to ingestion of seafood enriched in Cd often show no significant increase in adverse health outcomes, possibly due to high Zn and Fe in the seafood (McKenzie-Parnell et al., 1988;Wright et al., 2010). Cadmium uptake by cocoa has attracted attention and raised public concern, since chocolate is an important vector of Cd intake for children. In January 2019, the European Union enacted new regulations on the maximum Cd levels in cocoa, threatening the sustainability of cacao production in several cocoa exporting countries, especially South America. Since cocoa and chocolate products that do not comply with these maximum levels are denied access to European markets (Abt & Robin, 2020;Arg€ uello et al., 2019).
In the US, a total of 2,923 samples were selected to represent key foods in the US diet by FDA from 2014-2016. The top 5 foods with the highest mean Cd concentrations (values < LOD are set to zero) included sunflower seeds (0.38 mg/kg), boiled spinach (0.12 mg/kg), potato chips (0.09 mg/kg), leaf lettuce (0.06 mg/kg), and iceberg lettuce (0.05 mg/kg) (Spungen, 2019). In China, food samples from 24 provincial administrative divisions were analyzed in the 6 th China TDS (2016-2019). Foods with the highest mean Cd concentrations include aquatic foods (0.017 mg/kg), legumes (0.017 mg/kg), cereals (0.015 mg/kg), vegetables (0.013 mg/kg), and potatoes (0.011 mg/kg) (Zhao et al., 2022a). In South Korea, Cd concentration in shellfish (0.46 mg/kg) was the highest among Korean foods, followed by seafoods including seaweeds, mollusks and crustaceans (0.12-0.31 mg/kg) (Kim & Wolt, 2008). The average Cd concentrations in rice, cereals, potatoes, vegetables and fruits were about 0.02 mg/kg. However, dietary Cd exposure depends not only on food Cd concentration, but also the amount of the food consumed. When consumption rate was considered, cereals/grains contributed the most to dietary Cd exposure in many countries.

Cadmium concentrations in rice
Rice is the staple food for many Asian populations, contributing about 70% of the daily energy intake (Phuong et al., 1999). Rice Cd concentrations have been reported globally (Table 1). From     a large survey of 32 countries on 6 continents, rice Cd concentrations varied from <0.005 to 3.7 mg/kg in a total of 2,270 polished rice samples, with a global median of 0.02 mg/kg (Shi et al., 2020a). Among different countries surveyed, Mauritius, China, Nepal, Japan, and Pakistan were the top 5 countries having the highest Cd concentrations in market rice, with median Cd concentrations being 0.09, 0.07, 0.05, 0.05, and 0.03 mg/kg, respectively. However, sample density in the survey varied among different regions, which could result in biased conclusions when interpreting these data, especially for countries with small sample sizes. In another survey of 12 countries on 4 continents, a total of 1,147 rice samples from both market and field surveys were analyzed for Cd concentrations . Rice Cd concentrations were relatively high in Bangladesh, Sri Lanka, India, and Japan, with mean values of 0.10, 0.08, 0.08, and 0.06 mg/kg, respectively. Coupled with high per capita rice consumption, a subset of these populations could have Cd intakes exceeding the limit recommended by JECFA . Asia is the major contributor to rice production, with China, India, Indonesia, and Bangladesh together accounting for approximately 70% of the global production (FAOSTAT., 2018). In China, the mean Cd concentration was 0.05 mg/kg (range: <0.001-0.74 mg/kg) in 712 milled rice collected from local agricultural agencies and retail markets covering over 20 provinces during 2005-2008, and 2.2% of samples had Cd concentrations exceeding the maximum permission limit of 0.2 mg/kg set by Ministry of Health of the People's Republic of China (Qian et al., 2010). A large scale survey of 19,786 rice samples collected from supermarkets, local markets, and fields covering 31 provinces of China during 2011-2015 showed a similar average Cd concentration of 0.06 mg/kg (Song et al., 2017). In India and Indonesia, the geometric means of Cd concentration in rice collected from fields and commercial markets in different regions were 0.009-0.02 and 0.02-0.07 mg/kg, respectively (Rivai et al., 1990). In Bangladesh, Cd concentrations were 0.001-0.18 mg/kg in 144 polished rice samples collected from the markets of 16 districts (Shahriar et al., 2020). Italy is the leading rice producer among European countries, accounting for approximately 50% of the total production in Europe (Italian Grain and Feed Report, 2012). A national survey of Italian commercial rice (n ¼ 101) purchased from supermarkets during 2012 showed that the mean Cd concentration in rice ranged from 0.02 to 0.04 mg/kg by producing region . These Cd concentrations are well below the standard of 0.2 mg/kg set by the EFSA.
However, numerous studies conducted in Cd contaminated areas reported high Cd in rice that could threaten human health. In an area contaminated by nearby zinc mining activities in Mae Sot District, Thailand, Cd concentration in 524 rice samples ranged from 0.05 to 7.75 mg/kg, with over 90% of samples having Cd higher than 0.2 mg/kg (Simmons et al., 2005). In a study of 484 rice samples collected from five mining and smelting impacted areas in China, the average Cd concentrations were 0.15-0.19 mg/kg, with 18-41% of rice samples exceeding 0.2 mg/kg (Ke et al., 2015a). In some areas of Hunan, China, a province that hosts major nonferrous metal production, rice Cd concentrations were relatively high (range: 0.005-4.80 mg/kg) (Chen et al., 2018b;Zhu et al., 2016b). Hu et al. (2016) summarized rice Cd concentrations from market basket and field surveys across China. Among rice samples collected from fields, the average Cd concentration was higher in rice from mining areas than those randomly selected and from noncontaminated areas (0.32 vs. 0.08 vs. 0.06 mg/kg) ( Figure S1). The result is similar to market rice, with rice collected from mining areas showing higher Cd than non-contaminated areas (0.19 vs. 0.07 mg/kg) (Hu et al., 2016).
In an extreme case, chronic Cd exposure from consumption of Cd-contaminated rice caused "itai-itai" disease (kidney failure and softening of bones) in residents from the Jinzu River basin in Japan during 1955-1977(Kobayashi et al., 2002. Paddy fields in this area were contaminated with Cd from irrigation waters polluted by mining activities upstream, leading to an average rice Cd concentration of 0.38 mg/kg (range: 0.02-0.95 mg/kg, n ¼ 2446) (Nogawa et al., 2017). These surveys highlight the impact of mining and smelting activities on elevated Cd concentration in rice. Thus, implementing strict controls on emissions from mining and smelting and disposals of wastes is imperative to preventing rice Cd contamination.
Food safety standards have been established for rice Cd content by different countries and organizations. The current FAO/WHO limit for rice Cd concentration is 0.4 mg/kg (Codex Alimentarius Commission, 2006), whereas stricter standards have been set by European Union (European Commission, 2006), EFSA (2009), and China (0.2 mg/kg) (Ministry of Health of the People's Republic of China, 2012) and Australia and New Zealand (0.1 mg/kg) (Australia New Zealand Food Standards, 2015). However, adverse health effects through food consumption is not only determined by rice Cd concentration, but also by the rate of rice consumption. For a typical rice consumption of 250 g/day in Asian countries, a rice Cd concentration at the current FAO/ WHO limit of 0.4 mg/kg would give rise to a Cd intake of 100 mg/day, which is double the tolerable intake limit recommended by JECFA. Thus, for populations with high rates of rice consumption, the FAO/WHO limit for rice Cd concentration maybe not protective from adverse health effects.

Biomarkers of Cd exposure
Cadmium concentrations in blood, urine, hair, or nails have been used as biomarkers of Cd exposure. Blood Cd level is considered to be a useful marker of recent exposure to Cd, which is usually determined in the whole blood. Cadmium concentrations in hair and nails have also been used as a measure of Cd exposure; however, no standard operating procedures are available for these matrices and there is a high possibility of substantial exogenous contamination. Renal injury is the main toxic effect caused by chronic low-dose Cd exposure. Population-based studies showed that kidney Cd burden is proportional to Cd excreted in urine, making UCd a suitable biomarker of long-term Cd exposure (Menke et al., 2009;Vacchi-Suzzi et al., 2016). The major advantages of using urine are its noninvasive collection and accessible in large volumes, which can be used for the monitoring of low levels of chemicals in human investigations, especially in surveys involving a large number of participants. However, variation in UCd exists in spot urine samples. When timed urine excretion or 24-h samples are not collected, UCd concentrations (lg/ L) are typically adjusted by urinary creatinine concentrations (g/L) to correct for variable dilutions among spot samples, and are reported as micrograms of Cd per gram of creatinine (lg/ g creatinine).

Urinary Cd levels in the general population
To protect against kidney damage, the FAO/WHO established an UCd threshold of 5.24 lg/g creatinine, based on the critical concentration of Cd in kidneys (FAO/WHO, 2010). Rather than starting from the critical kidney Cd concentration, EFSA established a threshold value of 1 lg/g creatinine based on the dose-effect relationship between UCd and b2-microglobulin (a marker of tubular impairment) from a meta-analysis of 54 epidemiological studies among environmentally exposed populations (EFSA, 2011).
Urinary Cd levels have been reported worldwide to reflect human body Cd burden. In the US general population, the geometric mean of UCd were 0.28 and 0.40 lg/g creatinine for men and women, respectively, based on NHANES III data (Menke et al., 2009). In European countries, Spain (median: 0.39 lg/g creatinine), French (mean: 0.29 lg/g creatinine), and Sweden (mean: 0.26 lg/g creatinine) showed similar UCd levels, which were all below the threshold value of 1 lg/g creatinine set by EFSA (Frery et al., 2010;Grau-Perez et al., 2017;Olsson et al., 2002).
Higher UCd levels were reported in Asian countries compared to the US and European countries. A cross-sectional study conducted among the general population in eastern China, showed that the geometric mean of UCd were 0.36-0.45 lg/g creatinine (Sun et al., 2016). In Korea, a time-dependent decrease in UCd was identified through literature survey published during 1975-2015, with UCd levels being around or below 1 lg/g creatinine in the years after 2010 (Moon et al., 2016). In Boso Peninsula, Japan, the geometric mean of UCd was 1.3 lg/g creatinine among 547 men and 723 women who lived in an area without any known environmental Cd pollution (Suwazono et al., 2011). These UCd levels are well below the threshold of 5.24 lg/g creatinine established by FAO/WHO, however, the limit was challenged by studies showing that low-level UCd can result in multiple adverse health effects (Gallagher et al., 2008;Gallagher & Meliker, 2010;He et al., 2013). These studies imply no margin of safety between the current environmental Cd exposure levels and the threshold for adverse health effects.

Urinary Cd levels in populations living in polluted areas
Besides the general population, elevated UCd levels have been reported in occupationally exposed workers and people living in polluted areas. In industrial workers of a Belgian company involved with the production of refined Cd, UCd level were 0.4-38.3 lg/g creatinine (Viaene et al., 2000). In Yixing, China, wastewater contaminated by Cd-containing enamels and pigments was discharged into nearby rivers for farmland irrigation during 2002-2012, resulting in higher UCd (range: 0.2-5.4 lg/g creatinine) of the cohorts who consumed local food (Zhao et al., 2017a).
Industrial operations, particularly mining and smelting activities, often lead to higher UCd levels for nearby residents through elevated dietary Cd exposure. Among populations living in a heavily Cd-polluted area due to mining activities in the upper reach of the Kakehashi River, Japan, the geometric mean of UCd were 4.6 and 7.2 lg/g creatinine in men and women, respectively, with the highest UCd level being up to 49.6 and 57.5 lg/g creatinine in 1981-1982 (Nakagawa et al., 2006). In Mae Sot District, Thailand, 45.6% of the 7,697 persons surveyed had UCd levels < 2 lg/g creatinine, 52.1% were between 2-10 lg/g creatinine and 2.3% had Cd concentrations >10 lg/g creatinine due to the consumption of Cd contaminated rice contaminated by the nearby zinc mine that had been actively operated for more than 20 years (Swaddiwudhipong et al., 2007). In five Cd polluted areas caused by mining in China, UCd ranged from 0.05-57.3 lg/g creatinine in 6,103 participants, with 47.8% of males and 50.5% of females having UCd levels higher than 5 lg/g creatinine (Ke et al., 2015b). However, UCd is easily affected by physiological variations in the excretion of creatinine and proteins (Chaumont et al., 2012(Chaumont et al., , 2013, thus, biomonitoring studies using creatinine adjusted UCd concentrations should take into account the factors affecting urinary creatinine concentration.

The association between dietary Cd intake and urinary Cd in population studies
While numerous studies have examined dietary Cd intakes and UCd levels, there is a paucity of direct evidence showing the relationship between diet survey and UCd data in population studies. Here, we summarize studies showing the association between dietary Cd intakes and UCd in cohort studies (Table 2).

Studies on the association between dietary Cd intake and urinary Cd
Studies of populations living in Cd contaminated areas showed strong correlations between dietary Cd intake and UCd. Among 1,815 Cd exposed inhabitants in the Kakehashi River basin, Japan, a strong correlation was found between UCd levels and estimated life-time dietary Cd intake based on the average Cd concentration in rice produced in their village and the period of residence in the polluted area (r ¼ 0.93 and 0.88 for men and women, respectively) (Kido et al., 1992). When both adults living in polluted and non-polluted areas in Japan were studied, significant correlation coefficients were obtained (0.61 and 0.59 in men and women, respectively) (Kobayashi et al., 2005). Similarly, for adults living in polluted and non-polluted areas in Zhejiang, China, life-time dietary Cd intake was positively correlated with UCd (r ¼ 0.69) Wang et al., 2021).
Correlation between dietary Cd intake and UCd was weaker among the general populations. In a nationwide survey conducted between 1991-1998 including 607 nonsmoking Japanese women from 30 survey sites, 24 h duplicates of food intake correlated with spot UCd concentrations (r ¼ 0.39) . Among 57 nonsmoking Swedish women aged 20-50 years, UCd was associated with daily dietary Cd intake which was assessed by Cd in duplicate food portions collected during four consecutive days (r ¼ 0.38) (Julin et al., 2011).
Many other studies found little or no relationship between UCd and dietary Cd intakes. Among 1,002 US women with an average age of 63.4 years from the Women's Health Initiative, dietary Cd did not correlate significantly with spot UCd levels (p-value ¼ 0.14) (Quraishi et al., 2016). Among 1,764 post-menopausal women participated in the Danish Diet Cancer and Health Cohort, there was a weak association between U-Cd and dietary Cd estimates (Vacchi-Suzzi  Gunier et al., 2013Gunier et al., et al., 2015. Among 296 women enrolled in the California Teachers Study, the TDS database coupled with Food Frequency Questionnaires were used to quantify dietary Cd intake. However, dietary Cd was not a significant predictor of UCd (Gunier et al., 2013).

Challenges in associating dietary Cd intake and urinary Cd
The conflicting results regarding the association between dietary Cd intake and UCd levels may be caused by several reasons. Uncertainty may exist in the estimation of dietary Cd intake. In some studies, dietary Cd intakes were estimated by food consumption rate based on Food Frequency Questionnaires, with food Cd concentrations being sourced from literatures, or using average food Cd concentrations at the national/provincial scale. However, food Cd concentration varies substantially by regions (Hu et al., 2016;Yuan et al., 2014). In addition, dietary habit is highly heterogenous with dietary Cd intake estimates varying more than 1000 times (Adams et al., 2012;. Some other studies measured Cd in duplicate food samples (Julin et al., 2011;Shimbo et al., 2000). Although this methodology provides a more accurate estimation of dietary Cd intake, the data reflect current dietary Cd exposure under the assumption that dietary habits of the study participants have not changed dramatically over the past few years. Changes in diet habit with age may be an important consideration in future diet-based assessments of Cd exposure. Thus, there is a high possibility that the correlation would be improved if life-time cumulative Cd intake was used for populations with stable dietary Cd intakes. Single measurement of Cd in urine samples may have weakened the association between dietary Cd estimation and UCd, because UCd levels are likely to vary over time due to the variability in the volume of urine, variation in Cd concentrations in urine samples collected over time, the timing of sample collection, and the frequency of urination (Barr et al., 2005;Wang et al., 2016a). The life course changes in glomerular filtration rate (GFR) could also be a factor that positively influences the renal elimination of Cd, resulting in lifetime changes in UCd with age (Chaumont et al., 2013). Correction for urinary dilution, such as creatinine concentrations may introduce an additional source of uncertainty, since the excretion rate of creatinine differs among different demographic groups, including age, sex, and race/ethnicity (Barr et al., 2005;Hays et al., 2015).
Another reason for the poor association may be due to varying Cd bioavailability in food, which is often used to refine uncertainty in risk assessment (Zhao et al., 2017a;2018). Large variations in Cd relative bioavailability (the fraction of Cd in food that is absorbed into the systemic circulation following oral ingestion) were observed in vegetables (18-78%), wheat (37-68%) and rice (17-57%) collected from Cd-contaminated areas in Yixing, China (Zhao et al., 2017a). Predicted UCd (geometric mean: 4.14 lg/g creatinine) based on a toxicokinetic model using total Cd concentration in rice was 3.5-fold greater than the measured UCd, while when Cd relative bioavailability in rice was incorporated, predicted UCd agreed well with the measured UCd (1.07 vs 1.20 lg/g creatinine) (Zhao et al., 2017a). Furthermore, dietary factors, such as Ca, Fe, and Zn content, may influence food Cd relative bioavailability were not considered for the association between dietary Cd intake and UCd (Zhao et al., 2017b).
Variation in Cd absorption related to nutritional status is another possible explanation for conflicting results. Gender differences exist in Cd metabolism and toxicity, women tend to absorb more Cd from the intestinal tract due to lower body status of nutrients such as Fe and Zn stores (Gallagher et al., 2011;Julin et al., 2011). The "itai-itai" disease, which was mainly found in elderly women who had given birth to several children, could be explained by depleted Fe stores during pregnancy (Akesson et al., 2002). Julin et al. (2011) found that the inclusion of Fe status could improve the correlation between dietary Cd intake and UCd (r ¼ 0.38 vs. r ¼ 0.54).
The lack of association between dietary Cd intake and UCd could also be due to non-dietary Cd exposures, such as smoking, air inhalation, and dust ingestion. All these factors may eventually lead to variability in UCd level that is unrelated to dietary Cd intake.
Besides studies on the empirical relationship between dietary Cd intake and UCd, both elaborate eight-compartment physiologically based toxicokinetics (PBTK) model and a simpler onecompartment toxicokinetic (TK) model have been shown to be a better prediction of UCd based on dietary Cd intakes (Amzal et al., 2009;Julin et al., 2011). However, Cd absorption and metabolism vary widely in populations worldwide due to regional and ethnic differences, many parameters used in the model such as half-life, absorption rate, and ratios of Cd in kidney cortex and UCd need to be determined for specific populations.

Strategies to reduce human Cd exposure
Given the potential health risks of Cd exposure, it is important to take actions to reduce Cd exposure in the general populations and, particularly, in the residents living in contaminated areas. To reduce human Cd exposure, a number of strategies have been proposed by reducing Cd inputs from anthropogenic sources, cleaning up Cd contaminated soils, limiting Cd transfer from soil to crops, and reducing Cd absorption in human body (Figure 3).

Reducing Cd inputs from anthropogenic sources
6.1.1. Reducing mining and industrial emissions Nonferrous metal mining, smelting, and other related industrial operations are major sources of anthropogenic Cd emissions. Mine tailings, mining drainage discharged into surrounding rivers, and emissions to the atmosphere and deposition have given rise to Cd contamination in many locations around the world. A heavily contaminated area has been reported in the village of Shipham, which was built on slag heaps from a former zinc-lead mine near Bristol in southwest England during the 18 th and 19 th centuries. Cadmium concentrations in garden surface soils were up to 800 mg/kg (Strehlow & Barltrop, 1988). The Jinzu River basin in Toyama prefecture, Japan, polluted by Cd that originated from a zinc-mine located upstream, leading to the outbreak of "itai-itai" disease due to consumption of Cd contaminated rice (Kobayashi et al., 2005). To reduce Cd inputs from mining, smelting and other industries, stricter environmental protection legislation, more rigorous monitoring and stronger enforcement of environmental protection laws are required.

Reducing Cd inputs from fertilizers and manures
Besides industrial emissions, applications of phosphate fertilizers are also a major source of Cd input to soils (Verbeeck et al., 2020). In most Western countries, more than 50% of the anthropogenic source of Cd was due to the use of phosphate fertilizers (Alloway, 1995). Elevated Cd concentrations (<0.7 to 42 mg/kg) have been reported in 196 phosphate fertilizers from 12 European countries (Nziguheba & Smolders, 2008), which is probably due to the variation in Cd level in phosphate rocks (Robarge et al., 2004;Roberts, 2014). In a field experiment in the US, continuous applications of phosphate fertilizers at a rate of 175 kg/ha/year for 36 years resulted in a 14-fold increase in soil Cd concentration from 0.07 to 1 mg/kg (Mulla et al., 1980). To produce low Cd phosphate fertilizers, low Cd phosphate rocks and cost-effective technologies to remove Cd in phosphate fertilizers are therefore needed (de Boer et al., 2019;McLaughlin et al., 2021;Van Kauwenbergh, 2010). Optimizing the applications of phosphate fertilizers and improving phosphate utilization efficiency in crops are also ways to reduce Cd inputs into soils.
Some animal manures and organic composts made from manures may contain relatively high Cd (Luo et al., 2009;Yang et al., 2017). In a field experiment, applications of pig manures for 17 years increased soil Cd concentration by 18-fold (Wu et al., 2012). High level of Cd in animal manures is derived from animal feeds and feed additives, thus, stricter regulations of feed additives are needed to reduce Cd in animal manures.

Removal of Cd from soil
Removal of Cd from soils is the most effective and direct method to minimize Cd transfer from soil to crops and reduce human Cd exposure, although this strategy is also the hardest. The remediation methods include phytoremediation and soil washing; both are still under development and have not been practiced in large scale.

Phytoremediation
Phytoremediation has been considered as an attractive and sustainable process for reducing soil Cd concentration. Hyperaccumulators such as Arabidopsis halleri, Noccaea caerulescens, and Sedum plumbizincicola can accumulate high concentrations of Cd in the aboveground parts from soil (Lombi et al., 2000;McGrath et al., 2006;Yang et al., 2004). In a field study in south China, two consecutive crops of Sedum plumbizincicola reduced soil Cd concentration from 0.64 mg/kg to 0.29 mg/kg (Hu et al., 2019), suggesting considerable potential for Cd phytoremediation in soils. The advantage of this approach is that Cd can be removed from soils for good, however, the process of metal removal can be rather slow, during which the land's productivity is lost.

Chemical washing
Ferric chloride (FeCl 3 ) has been selected as a promising chemical for remediation of Cd contaminated paddy soils by means of soil-washing (Makino et al., 2006(Makino et al., , 2008. Different metal salts were used to extract Cd from three Cd-contaminated paddy soils, FeCl 3 extract the highest percentage of Cd from soils (24-66%) with no negative effect on rice yield (Makino et al., 2008). The high efficiency of FeCl 3 was mainly attributable to the low extraction pH caused by proton release from the generation of iron hydroxides, as the solubility of iron hydroxides is low. The formation of Cd-Cl complexes could also enhance Cd extraction from soils by hindering re-adsorption of extracted Cd onto adsorption sites on soil particles. Chemical washing has the potential to remove Cd effectively and quickly from paddy soils that are highly contaminated. However, the use of chemical reagents may disturb soil physical and chemical properties and damage soil fertility.

Reducing Cd availability in soil
Many factors, such as pH, water management, and applications of fertilizers can greatly affect Cd solubility in soil. To reduce Cd transfer from soil to crops for human consumption, a key step is to minimize Cd availability in soil.
6.3.1. Liming to reduce Cd availability in acid soils Liming has been the most common practice to mitigate soil acidification and to reduce soil Cd availability and its uptakes by crops. A two-year field trial conducted in southern China found single applications of limestone at 5.25 ton/ha alone increased soil pH by 0.89-1.15 unit and decreased grain Cd concentrations by 45-62% (Fang et al., 2021). A recent meta-analysis of 35 field studies found lime application reduced grain Cd concentration by 48% and increased rice yield by 12.9% on average, confirming that liming is effective in reducing Cd availability in acid soils and Cd uptake by rice (Liao et al., 2021). While liming has been proposed as an effective practice to reduce Cd availability in acid soil, the effect may vary due to factors such as initial soil pH, soil characteristics and plant species (Chaney et al., 2009;Li et al., 1996).

Fertilizer management
Fertilization is an important practice to increase rice yield. Fertilizers of manganese (Mn), zinc (Zn), and silicon (Si) can be used as the potential approach to reduce Cd accumulation in rice grains (Fahad et al., 2015;Huang et al., 2022;Lv et al., 2019;Shao et al., 2017;Wang et al., 2016b). Applications of Mn oxides or Mn-loaded biochar in field trials, significantly decreased Cd accumulation in rice (Fang et al., 2021). Manganese suppress Cd uptake in rice through competitions for Mn transporter Oryza sativa Natural Resistance-Associated Macrophage Protein 5 (OsNRAMP5), which is constitutively expressed in the roots and located at the distal side of both exodermis and endodermis cells (Sasaki et al., 2012;Yang et al., 2014).

Paddy water management
Paddy water management can change redox potential and pH of soil, and subsequently affect Cd bioavailability in soil. When a paddy field is flooded and the soil is under anaerobic conditions, soil Cd availability decreases greatly due to the formation of insoluble CdS and increase in soil pH (de Livera et al., 2011;Wang et al., 2019). Once the field is drained and the soil becomes aerobic, CdS is oxidized to Cd 2þ , increasing Cd bioavailability in soils and elevated rice grain Cd concentrations . Thus, flooding has been proposed as an effective strategy to reduce Cd accumulation in rice grains (Arao et al., 2009;Sun et al., 2014). However, continuous flooding of paddy soil can increase rice arsenic (As) accumulation, which deserves attention if soils are contaminated with Cd and As simultaneously.

Decreasing rice grain Cd through cultivar screening and genetic engineering
Large variations in grain Cd concentrations have been reported among different rice cultivars (Arao & Ishikawa, 2006). A field survey including 1,763 highly diverse rice accessions collected from the world showed a 41-and 154-fold variation in grain Cd concentration under flooded and non-flooded conditions, respectively (Pinson et al., 2015). Quantitative trait loci (QTL) mapping is a powerful genetic approach to identify the effects of genetic factors controlling Cd accumulation in rice. A number of genes controlling Cd uptake and translocation have been characterized (Ishikawa et al., 2012;Sui et al., 2018;Zhao et al., 2022b), paving the way to develop low accumulating rice cultivars through molecular breeding or genetic engineering. Oryza sativa heavy metal ATPase 3 (OsHMA3) plays an important role in transporting Cd into the vacuoles in root cells (Ueno et al., 2010). Overexpression of OsHMA3 in an Indica cultivar of rice decreased Cd content in brown rice grain by 94-98% with little effects on grain yield or the concentrations of essential trace elements (Lu et al., 2019).

Reducing Cd absorption in the human body
Additional strategies to reduce human Cd exposure can focus on reducing absorption and body retention of Cd at the consumption point. Supplementation of micronutrients or changing diet structure are recommended for people at risk of Cd exposure, since these practices are easily operatable and affordable.
6.5.1. Nutrient supplementation Micronutrients have a major impact on Cd absorption since Cd utilizes the same intestinal transporters for Zn, Fe, and Ca in human body. There is significant correlation between Cd absorption and the expression of the divalent metal transporter DMT1, which transports both Cd and Fe across the intestinal epithelium in a competitive manner (Akesson et al., 2002;Tallkvist et al., 2001). Thus, Cd absorption in the human body can be decreased by high Ca, Fe, and Zn in the diet but increased by Ca, Fe, Zn deficiency. Flanagan et al. (1978) first showed Fe deficiency can increase Cd absorption in both humans and mice models. The situation is particularly significant among women of childbearing age, with increased DMT1 density to optimize micronutrients absorption (Kippler et al., 2012). Therefore, nutrient supplementation may potentially decrease human Cd accumulation and associated Cd induced health risks.
6.5.2. Changing diet structure Vegetables, seafood and fruits are important dietary sources of nutrients, vitamins, and proteins. Vegetables such as tomato has been reported to reduce Cd absorption in rat liver, due to the formation of metal chelating proteins, peptides, phytochelatins by tomato, leading to the enhanced mobilization, chelation and excretion of Cd (Nwokocha et al., 2012). Other food types including garlic, green tea, and grapes may be effective in decreasing Cd absorption and accumulation in human body (Zhai et al., 2015), even though the potential mechanisms are unknown. Furthermore, reducing the consumption of rice containing high levels of Cd could also be a part of the strategy in the prevention of excessive dietary Cd exposure. This is important for consumers who subsist on rice-based diets as they may have a greater Cd absorption than individuals with other staple diets because of the low mineral nutrients in rice.

Conclusions and perspectives
Dietary intake is the main source of Cd exposure for the nonsmoking population. For average consumers, dietary Cd intakes for the general populations in most countries were below the tolerable Cd intake limits set by EFSA and JECFA. Asian populations who consume rice as the staple food generally show higher dietary Cd intake levels, with subsets of the populations in some Asian countries exceeding the EFSA limit. Among multiple food items, shellfish and mollusks are known to have higher Cd accumulation from the aquatic environment, but ingestion of these food show no significant increase in adverse health effects. Furthermore, dietary Cd exposure is depended not only on food Cd concentration, but also the amount of the food consumed. When consumption rate is considered, cereals/grains, especially rice, generally contribute the most to dietary Cd exposure. Global surveys of rice Cd concentration showed that market rice produced by most rice producing countries are below the permissible limits. However, for populations with a high per capital consumption or residents living in contaminated areas, especially in mining and smelting polluted areas, excessive Cd intake from rice consumption could pose a risk to public health. Urinary Cd have been used as biomarkers for long-term Cd exposure. Large variations in UCd have been reported among different populations, with higher UCd levels for those living in mining and smelting polluted areas, which is consistent with elevated dietary Cd intake. However, conflicting results on the association between dietary Cd intake and urinary Cd levels have been documented.
To reduce Cd exposure, a range of management options can be taken by reducing Cd inputs from anthropogenic sources, removing Cd from soil through phytoremediation and chemical washing, limiting Cd transfer from soil to crops through liming, fertilizer management, and water management, decreasing rice grain Cd through breeding or transgenic techniques, and reducing Cd absorption in human body through nutrient supplementation or changing diet structure.
A number of knowledge gaps remain, requiring concerted efforts through further studies. First, long-term monitoring of dietary Cd intake and food Cd levels at large scale is needed to identify the potential health risk of Cd exposure, especially in contaminated areas. Second, surveys of UCd in the general populations at national scale should be performed, especially in developing countries, as most previous studies focused on UCd levels for residents living in Cd contaminated areas. Third, paired dietary Cd intake and UCd data should be collected simultaneously. It is also likely that the correlation between dietary Cd intake and UCd would be improved when taking food Cd relative bioavailability and human nutrition status into account. Fourth, further studies are required to establish quantitative relationships between soil-food-UCd at regional scale, taking into account key factors that influence these relationships. Fifth, previous strategies have been applied singly. To achieve greater effectiveness of intervention strategies, several methods can be used simultaneously to evaluate the effectiveness throughout the whole foodhuman chain in future studies.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported in part by National Natural Science Foundation of China (42107430)