Assessment of occupational exposures to multiple metals with urinary porphyrin profiles

Abstract

Chronic occupational exposures to low levels of metal mixtures necessitates biomonitoring of exposed workers. However, a single biomarker (BM) is rarely sufficient to ascertain the exposure of an individual to a complex mixture, with multiparameter analysis of the same sample considered recently as a preferred approach. Porphyrins are formed as intermediates of heme biosynthesis and different metals can exert their effects at different points of this metabolic pathway, leading to changed urinary porphyrins excretion profiles. The aim of this work was to develop a model that could serve to identify, on an individual basis, multiple metal exposure resulting from mining work, by using urinary porphyrin profiles. Urine samples of workers were obtained from a Portuguese mining company and a non-occupationally exposed group was used as control. The levels of uro-, hepta-, hexa-, penta-, copro- and protoporphyrins were determined by HPLC. It was observed that only heptaporphyrin levels in miners were significantly (p<0.05) different from controls. However, when the concentrations of all porphyrins were combined by binary logistic regression, their ability to discriminate between miners and controls was higher than each one of the porphyrins alone, as indicated by a greater curve’ area under a ROC curve. Moreover, when the combined porphyrins were used to calculate the probability of each subject fit in the occupationally exposed group, 83% of 47 individuals were correctly identified with respect to their type of exposure. These results suggest that the integration of the urinary porphyrin profile is a promising tool for the detection of subjects exhibiting biochemical modifications due to occupational exposure to metals in mines.

1. Introduction:

Mining represents one of the most hazardous occupations [1], and has been long recognized as arduous and liable to injury and disease [2]. Biomonitoring of chronic occupational exposures to low levels of metal mixtures in miners as well as other occupations is indispensable for workers’ health, with molecular biomarkers (BM) representing a critical tools in achieving this task [3, 4]. However a single BM is rarely sufficient to ascertain exposure in a single person to a complex mixture. Accordingly, information from multivariate analysis of multiple parameters, measured simultaneously in the same sample is being considered as an promising approach [5]. Porphyrin analysis represents one such approach. Porphyrins are compounds formed as intermediates of heme biosynthesis [6], and it well established that various metals can exert their effects at different points of the porphyrin metabolic pathway, leading to changed urinary porphyrin excretion profiles. Thus, porphyrins have been recognized as promising BMs of metal mixture-induced toxicity [7]. Indeed, previous work in our laboratory revealed that changes in rat urinary porphyrin profiles could predict the magnitude of effects induced by a metal mixture in individual animals [8, 9]. Our recent work has focused on the translation of these observations to human populations exposed to metal mixtures, such as mine workers [3]. The aim of this study was to develop a model that could serve to identify multiple metal exposures resulting from mining work taking advantage of urinary porphyrin profile analysis on an individual basis.

2. Material and Methods

Urine samples from workers in a mining company in Portugal were obtained from the Occupational Health Department under the supervision of a nurse and doctor; The working group (n=29) was composed of male miners aged between 24 and 56 years. All of them were working in the mine for more than one year, A non-occupationally metal exposed group was composed of workers from the same company (also male individuals aged between 22 and 60) was used as a control group. Individuals in the control group were not exposed to metals in their work environment (n=20). All the biological samples were collected on the last day of the week and stored at –80°C until the analysis.

Chromatographic porphyrin analysis was performed by HPLC after sample preparation according to Woods et al, 2009 [8] to obtain porphyrin profiles by determining uro-, hepta-, hexa-, penta-, copro- and protoporphyrins concentrations.

Statistical analysis was performed with the SPSS 16.0 statistical package for Windows (SPSS, Inc., Chicago, IL, USA). Data are expressed as means ± standard deviations (SD). 1) The occupational and non-occupational exposed groups were compared by Mann-Whitney U tests, respecting to their levels of each porphyrin. 2) The ability of urinary porphyrins to detect occupational exposures in mines was evlauted by ROC analysis, which is a statistical tool to evaluate the diagnostic accuracy of BMs, alone or in combination [10]. In this analysis, BMs levels are plotted in a curve under a 1- Specificity (x) vs Sensitivity (y) axis, being the area under the curve directly proportional to the BM(s) diagnostic accuracy Actualy, these plots of 1-Specifity (false positives or 1-true negatives) versus Sensitivity (true positives) is an effective measure of BM(s) accuracy [11]. Thus, an area=100% indicates that all subjects are true positives and true negatives indicating a maximum accuracy of the BM(s); by its turn an area that lies close or under 50% have no information content and indicate that the BM(s) do not have diagnostic utility (Warnock and Peck, 2010)[12]. Here, the levels of each porphyrin alone and the combination all the porphyrins by binary logistic regression, were tested by ROC analysis; the obtained areas were compared to evaluate which BM(s) exhibited the highest diagnostic accuracy. 3) Posteriorly we created a mathematical expression that could serve to determine the odds a new subject under study was exposed to metals in his work environment [13]; The odd values was calculated using levels of urinary porphyrins. This expression was generated by binary logistic regression. Notably, combining porphyrins profiles had a significantly better diagnostic accuracy under the ROC curves for metal exposure.

3. Results

3.1. Urinary porphyrin profiles

Figure 1 shows that a significant (p<0.05) difference and lower urinary levels of heptaporphyrins were noted in miners when compared to controls.

Figures 1 a), b), c), d), e) and f):

uroporphyrin (a), heptaporphyrin (b), hexaporphyrin (c), pentaporphyrin (d), coproporphyrin (e) and protoporphyrin (f) urinary levels in controls (C) and miner (M) population. Data represent the mean ± SD and n=20 and 29 in C and M groups, respectively. All groups were compared by Mann-Whitney U tests. * represents p < 0.05 versus C.

3.2. Capability of urinary porphyrins to detect occupational exposures in mines:

The diagnostic accuracy of each porphyrin alone was tested for their ability detect subjects occupationally exposed in mines (true positives). This approach was deemed unsactisfactory as the areas under the 1- Specificity (x) vs Sensitivity (y) axis were lower than 50%, attesting to high false positives and false negtives, and indicating lack of diagnostic utility of each of the BMs alone (Figure 2a). Alternatively, when all the porphyrins were combined by binary logistic regression, the number of true positives and true negatives increased substantially as reflected by the increase in the area under the ROC curve (area= 0.743; p<0.05) (Figure 2b).

Figures 2 a and 2 b:

The ability of the urinary levels of uroporphyrins, heptaporphyrins, hexaporphyrins, pentaporphyrins, coproporphyrins and protoporphyrins individually (a) and their combination by binary logistic regression (b) to identify subjects with mining related exposures. The curve’s area under the 1- Specificity (x) vs Sensitivity (y) axis are plotted by ROC analysis: areas are directly proportional to the BMs discriminant capabilities. All represented areas reached statistical significance (p<0.05). N=49.

3.3. Detection of mining work related exposure in an individual basis:

To detect the mining work related exposure in each subject, we developed a mathematical expression. Expression 1 represents the odds of a subject to be considered as exposed in a mine (oem), by depicting the concentrations of their urinary porphyrins (uro-, hepta-, hexa-, penta-, copro- and protoporphyrins), by binary logistic regression:

$$ODDs(oem)=26.6Uro-176.3Hepta+484.9Hexa+111.3Penta+5.9Copro+790.2Proto+1.4$$

The use of combined urinary porphyrin levels leads to 83% of correct detections of occupational exposure in miners (Figure 3b). More precisely, 15 of 20 subjects included in the control group were correctly identified as not occupational exposed exhibiting only a 0.15 to 0.45 range odd of being a miner (Figure 3b). Additionally, 4 of the 5 controls were erroneously classified as miners showed a 0.80 to 0.85 range odd of being a miner, while one person had an odd of only 0.5 (Figure 3a). With regard to the miners, 24 of 29 of the workers were correctly identified as miners with odds that ranged from 0.6 to 0.9 with a higher frequency at the odd of 0.8 (Figure 3a). Concerning to the false negatives, 5 miners were incorrectly included in the control group and a 0.20 to 0.45 range odd of being a miner was attributed to these persons (Figure 3a).

Figure 3.

Predicted probability of each subject in the minning group (M) and the native group (Group) after application of Expression 1 to each studied individual. Each symbol (C or M) represents 1 subject (a) and the ODDs cut value for miners group membership is ≥ 0.5. The table (b) represents the counts of subjects included in each group as predicted by the model vs the real group (observed). N=20 and 29 in C and M groups, respectively.

Discussion:

Low-dose chronic exposure to metals may result in insidious poisoning that may manifest clinically at a very late stage. Thus it is critical to monitor mine workers so that subclinical toxicity can be identified at the onset of disease or even at erlier stages [3, 14]. While molecular BMs represent key tools in occupational medicine [15], urinary porphyrin profiles may serve as BMs of exposure to toxic metals in humans [16, 17]. Since various metals can exert their effects at different points of the heme synthesis pathway, we hypothesized that the urinary porphyrin profile might serve as a “fingerprint” that would permit a satisfactorily identification of individuals exposed to metals due to their occupation in mines.

Our results showed that each of the porphyrins alone poorly detected occupational exposures to metals, even on a group basis, with miners and controls exhibiting differences that lacked statistical significance (p>0.05) (Figure 1). Miners are exposed to several metals [3], such as Pb, As, Mn, Hg and Al, which interfere at common, and also at different points, in the heme synthesis sequential pathway [18, 19, 20, 21, 22]. These events may lead to distinct fingerprint in urinary excretion of each of the porphyrins, explaining their poor diagnostic performance when used alone. These results were confirmed by the observation of areas <0.5 under a 1- Specificity (x) vs Sensitivity (y) axis plotted by ROC analysis (Figure 2a). Indeed, BMs with a curve that lies close or under the diagonal reference line (area = 0.5) had no information content, and therefore no diagnostic utility [12]. In contrast, when the urinary porphyrins profile was applied as a whole with integration of all the porphyrins by binary logistic regression, the area under a ROC curve increased the diagnostic utility of the acceptable value (area=0.743) (Figure 2b) [12]. Therefore, the concentrations of all the porphyrins were used to generate an equation to determine the odds of a new subject under analysis be occupationally exposed in a mine (Expression 1). When this equation was tested in the studied subjects, a correct detection of the occupational exposure in the mines was obtained for 83% of the cases (Figure 3b). Our results are satisfactory for a preliminary study with N=49, obtaining 25% of false positives and 17% of false negatives. These findings also lead us to proceed with additional future work, expanding on the sample size, to further establish the proof-of-concept of these alalyses as predicitive tools of metal mixture exposures. Overall, the integration of the urinary porphyrin profile showed superiority in discriminating miners or subjects not occupationally exposed, then any of the porphyrin concentrations alone. The proposed multiparameter approach is promising for the detection of mining work related to multiple metal exposures, and should be further validated in future studies.

Abbreviations:

BM

biomarkers

HPLC

High Performance Liquid Chromatography

ROC

Receiver Operating Characteristics

uro-

uroporphyrins

hepta-

heptaporphyrins

hexa-

hexaporphyrins

penta-

pentaporphyrins

copro-

coproporphyrins

proto-

protoporphyrins

oem

occupationally exposed in a mine