Balkan Endemic Nephropathy (BEN) is a tubulointerstitial kidney disease that progresses slowly over many years. First characterized in the Vratza District, Bulgaria in 1956 1, similar nephropathies were subsequently described in Yugoslavia 23 and in Romania 4. In 1964 this disease was recognized as a new nosological entity and named Balkan Endemic Nephropathy. The final disease stage is characterized by renal failure and shrinkage of both kidneys to the size of walnuts 5. BEN shows familial clustering and develops only in certain areas in the Balkan countries – Bulgaria, Romania, Serbia, Montenegro, Croatia, and Bosnia and Herzegovina. Its spatial distribution shows a mosaic-like pattern, with some villages afflicted for decades, whereas others situated in the same vicinity have remained free of BEN 678. It is still unknown what causes BEN. Although this disease is endemic in rural areas of the Balkan, endemic nephropathies are reported to occur in other regions. For instance, chronic kidney diseases of uncertain etiology, which resemble BEN, have been reported in Sri Lanka (personal communication, Dr. Alturaliya, Sri Lanka). Every now and then, new explanations emerge emphasizing lignites 9, aristolochic acid 10, ochratoxin A 11, metals, and metalloids 12131415. However, these claims are often not substantiated by sufficient evidence or scientific studies. For instance, studies claim that exposures to metals (cadmium and lead) and metalloids (arsenic and selenium) are related to the disease, with selenium being protective 12131415.

An important characteristic of BEN is the late onset of the disease (not before age 50) and thus probably a long latency period and/or a period of being at risk between exposure and the gradually developing clinical manifestation of the disease. To determine exposures before the onset of the disease, we chose to study the offspring of patients with BEN, who are at a higher risk for the disease, and use the offspring of other non-BEN patients as a control group.

Early markers for BEN are a lower creatinine clearance, increased excretion in urine of total protein, albumin, and β₂-microglobulin, and symmetrically decreased kidney measures 161718192021222324252627. We recently reported in a cross-sectional investigation that changes in these markers are associated with being offspring of BEN parents 28. In this work we link clinical markers with the concurrent burden of metals and metalloids.

We investigated four metals and metalloids: lead, arsenic, cadmium, and selenium. Lead, absorbed in blood, is bound extensively to erythrocytes (90–95%) and lead content in whole blood is used routinely as a biomarker of exposure (up to a few months before sampling) 2930. Most arsenic absorbed by the lungs or by the gastro-intestinal tract is excreted in the urine, mainly within 1–2 days. Measurement of urinary arsenic levels is generally accepted as the most reliable indicator of recent exposure 31. Total arsenic in urine, which is measured in this study, includes inorganic arsenic, as well as its metabolites monomethylarsonic acid and dimethylarsinic acid.

Cadmium is a heavy metal which accumulates predominantly in the kidneys (half-life: 15–40 years) and liver and is excreted mainly in urine. In low-level exposures (below 10 μg/g creatinine) cadmium content in urine reflects body burden and is used as a biomarker of cumulative exposure 3032. After absorption, selenium in blood is distributed between red blood cells and plasma/serum. The concentration in plasma or serum is considered to reflect short-term exposure, whereas the selenium content of erythrocytes indicates long-term exposure 32. In a steady state, the Se levels in red blood cells are considered to be higher than those in serum 3334, however, both are significantly correlated and comparably associated with daily intake 353637. Serum selenium is used for assessment of occupational and environmental selenium exposure, including nutritional status, in various studies, for instance, the U.S. National Health and Nutrition Examination Survey 323839.

Based on prior reports in the BEN literature we formulated five hypotheses:

1) Selenium serum concentration is protective against adverse kidney effects.

2) Lead in blood, 3) cadmium in urine, and 4) arsenic in urine are associated with adverse alterations in BEN kidney markers and size.

5) Adult offspring with a parental history of BEN show stronger associations between metals and/or arsenic and kidney variables.

Since the incidence of BEN shows a spatial distribution with a mosaic-like pattern one may consider to simultaneously compare residents from BEN villages and non-BEN villages and specific exposures. By doing so, we could compare combined effects of residence and of specific exposures. This would be advantageous, since we would also control for unknown risk factors acting in BEN villages. However, we need to take into consideration that we do not have BEN patients from non-BEN villages (Figure 1). Thus, a formal testing of interaction terms will not work. Alternately, if we compare the effect of metals in BEN-offspring from BEN villages with the effect of metals in non-BEN offspring from non-BEN villages, we would mix two effects: the specific metal effect and an unknown spatial effect, which would result in confounding. To estimate spatial effects, ecological studies have been conducted and have attributed differences to specific factors 14. We considered that testing of specific risk factors requires a more rigorous comparison, namely of offspring of BEN and non-BEN patients from BEN villages.

Of the 201 participants from the first year, 189 participated in the follow-up (94%). Two were deceased, one moved out of the area, two could not be contacted, and seven decided not to participate further. Of the twelve lost nine were offspring of BEN patients. With the exception of hypertension, potential risk factors show similar distributions in each of the two investigations (Table 1). Of the BEN offspring, 30.4% had hypertension in 2003/04 and 29% in 2004/05. A lower prevalence was detected in control offspring (19.2 and 22.9%, respectively). BEN patients also reported more often a history of kidney stones and pyelonephritis.

Regarding arsenic and cadmium (measured in morning urine) and creatinine (measured in morning urine samples), we found no agreement between the two investigations for creatinine, a moderate agreement for arsenic and a good agreement for cadmium (Spearmen rank correlation and ICC, Table 2). For creatinine in morning urine samples the within-subject variance exceeds the between-subject variance. When we used the directly creatinine-adjusted parameters (division by the creatinine concentrations) the agreement for cadmium and arsenic was reduced, simply because of the high intra-individual variance of creatinine. There was a better agreement for two measurements, namely lead in blood and selenium in serum, measured one year apart.

Regarding the kidney parameters, Spearman rank correlation and ICC were low for total protein and albumin in serum, for creatinine in the 1-hour and 4-hour urine sample, and for the glomerular filtration rate (Table 3). These values did not improve when we directly adjusted for creatinine, since the within-subject variance of creatinine was high. The agreement between the sonographic determinations of kidney dimensions (kidney cortex and kidney length), β₂-microglobulin, and the estimated creatinine clearance (CCR) was sufficient to conclude that they measure some underlying disease processes. In the following analyses we focus on these four parameters to test whether arsenic and metals explain their variance.

We did not find any significant effect of arsenic and metals on kidney cortex width (Table 4). Of the metals and metalloids analyzed, only selenium showed an association with kidney length, β₂-microglobulin, and the estimated CCR in the models presented in Table 4. Lead gained importance only in the control offspring. Similar to selenium, it seems to decrease CCR.

Higher selenium levels are associated with a longer kidney length in the total group and the BEN offspring. When comparing an individual with 70 μg/L with one having 20 μg/L selenium, on the average kidney length would be increased by 1 mm (50 *0.02 mm). This is only a small increase compared to the total kidney length (approximately 120 mm). However, in BEN offspring the respective increase is 2 mm (50 *0.04 mm; Table 4). Selenium seems to decrease CCR in the total group and the control group, but not in BEN offspring. When comparing an individual with 70 μg/L with one having 20 μg/L serum selenium, CCR would be reduced by 4.5 mL/min/m² (50 *0.09 mL/min/m²) in the total group. This is a minimal decrease of CCR (range: 54–87 mL/min/m²). In the control group, lead is associated with a decrease (0.14 mL/min/m²). β₂-microglobulin was log-transformed to achieve a normal distribution. A 50 μg/L increase in selenium would be approximately equivalent to a 1.4% increase in β₂-microglobuline [10^(50*0.003)].

Of elements investigated, cadmium and arsenic were associated neither with kidney dimension nor with function (hypotheses 3 and 4). Lead had a negligible effect on CCR (hypothesis 5). Selenium showed weak but significant associations with three of the four kidney parameters tested, namely kidney length, estimated CCR, and β₂-microglobulin, but not kidney cortex width.

Our study included only BEN villages. Others think that the soil geochemistry in villages affected and not affected by BEN may be an etiologic factor 14. It has been suggested that arsenic, cadmium, and lead (Pb) are elevated in the soil of BEN areas, whereas selenium is deficient 121450. Ecological studies do not take individual exposures into account. Nevertheless, according to these investigations, we would have expected higher levels of arsenic, cadmium, and lead in our participants. To examine whether the levels in our samples were different, we compared these to other reports on metal and metalloid exposures in Bulgaria. Tsolova et al., when studying copper-smelter workers, described urine arsenic concentration of 7.73μg/L in a control group 51. We found approximately 3μg/L in an unexposed population (Table 2). Regarding cadmium, two Bulgarian investigations reported urine levels of 0.8–1.0μg/L 5253; we had urine concentrations of 0.6–0.7μg/L. However, adjusted for creatinine, urinary cadmium levels were higher in our Bulgarian samples (approximately 0.85μg/g creatinine) than levels reported for the United States (0.26μg/g creatinine) 54. Lead blood concentration in our samples was lower (median of 90 μg/L in 2003/4, Table 3) than levels reported for occupational exposures in Bulgaria 5355. Selenium levels in this study (range 36–95 μg/L, medians: 56.9 and 72.1; Table 2) were comparable to levels reported in other studies in Bulgaria (57 and 66.5 μg/L) 3056. However, serum selenium levels seems to be lower in this rural region of Bulgaria compared to other countries 5758. For instance, in the United States, mean selenium levels were 125.7 μg/L. However, in rural regions selenium serum levels are lower; in central France on average 79 μg/L and in sample of rural Serbia on average 38.2 μg/L 5759. These concentrations are within the 95% confidence limit of our study (Table 2). In summary, concentrations of arsenic, cadmium, lead, and selenium in our samples were either comparable to concentrations in samples taken from other subpopulations in Bulgaria. Hence, these comparisons do not support the suggestion that BEN families are exposed to higher levels of arsenic, cadmium, selenium, or lead that could contribute to the occurrence of the disease.

Our study subjects were offspring of BEN patients who are likely to have a higher susceptibility of their kidneys to adverse exposures (hypothesis 5). Therefore, we expected to find that some of the elements would be associated with kidney function in BEN offspring. However, compared to controls, BEN offspring were not more affected (Table 4). For lead (Pb) we found a weak association with creatinine clearance, but only in control offspring. Further, since our investigation is based on offspring who have not yet developed the disease, we believe that this excluded behavioral changes such as avoidance of chemicals, which could have distorted the results. Therefore, since prior studies that reported associations between arsenic/metals and BEN had an ecological design or were based on geochemistry analyses, our negative findings suggest that non-epidemiologic studies may have been misleading.

A limitation of our investigation is the high intra-individual variance of kidney parameters. If our analysis would have been based on a single examination, this could have led to biased results. However, we included repeated measurements that took intra-individual variances into account and provided added confidence in our results. In addition, we focused on more stable indicators and did not consider clinical markers such as creatinine in urine, which showed a large variability over time. A second limitation is that offspring of BEN patients were younger (Table 1) and degenerative kidney disorders often occur in older people. Hence, it is possible, that we underestimated the effect of the metals/metalloids. Thirdly, as outlined in the introduction, there are no BEN patients born in non-BEN villages. Hence a two-way analysis (metals/metalloids and BEN-village) is not feasible. Since many new conclusions on what causes BEN have been presented based on aggregative (village data), we preferred a more rigorous testing of risk factors and focused on the association of metal/metalloids and BEN status within villages who are labeled BEN villages.

We are not aware of another study that investigated the association between kidney parameters and selenium in humans. Nevertheless, a recent study on selenium supplementation found an increased risk for type 2 diabetes 60. Based on other findings, we expected that selenium would have a protective effect (hypothesis 1). However, we found three unexpected findings: Serum selenium was associated with increased β₂-microglobulin, with increased kidney length, and decreased CCR. These may be chance findings. However, the US Agency for Toxic Substances and Disease Registry lists a number of comparable kidney effects in animals following exposure to selenium at levels several hundred times higher than normal human intake 38. Effects include hydropic degeneration in sheep 61, increases in kidney weights in mice and ducks 626364, larger Bowman's capsules and minimal papilla degeneration of the kidneys in rats 6265, and glomerulonephritis and proximal convoluted tubule nephropathy in long-tailed macaques 66.

Recently, Balkan endemic nephropathy has attracted some attention 91011. First, a number of potential new risk factors have been presented. Second, its uniqueness in the Balkan regions has been questioned as similar manifestations were observed in other regions too 10, for instance in Sri Lanka 67. There is a necessity for carefully testing various proposed risk factors and pathological mechanisms. We believe that our study contributes to the evidence that metals and metalloids are no risk factors for BEN. Further studies are necessary to test the claim of other proposed risk factors.

Our results do not support the hypotheses that cadmium, lead, or arsenic pose a risk for the development of Balkan Endemic Nephropathy. Against the assumption in the literature, selenium was not protective, but a risk factor for all three markers: β2-microglobulin, creatinine clearance, and kidney length. Since comparable associations with anatomical and functional markers were observed in animals, future studies are warranted to investigate selenium effects on the kidney in humans.

arsenic (As), atomic absorption spectrometer (AAS), Balkan Endemic Nephropathy (BEN), body surface area (BSA), cadmium (Cd); creatinine clearance rate (CCR), electrodeless discharge lamps (EDL), intra-class correlation coefficient (ICC), lead (Pb), selenium (Se), transverse heated graphite atomizer (THGA)

The author(s) declare that they have no competing interests.

PD and WK designed the study, analyzed the data, and prepared the manuscript. VS and AB conducted the clinical investigations and helped with the preparation of the paper and the clinical interpretation. RG led the measurements of the metals and metalloids and reviewed the manuscript. ST provided toxicological expertise, help with the implementation of the study, data preparation, analyses, and the writing of the manuscript. All authors have read and approved the final manuscript.