Organophosphate pesticides: Another silent liver hazard?

Environmental and occupational pollutants and their role as potential liver hazards is a largely neglected area of research.1 Organophosphate pesticides (OP) for domestic or agriculture purposes are highly toxic compounds widely used as pesticides owing to their cost effectiveness and for which people got exposed through ingestion of residues on foods, dust and soil and during suicide attempts as well as inhalation from spray drift, especially in cases of occupational exposure in agriculture. Hence, OP poisoning is frequently reported resulting in alarming mortality rates.2 However, the effects of OP on exposed subjects without manifestations of acute toxicity have not been yet investigated in epidemiologically designed studies. In this issue of Liver International, Li et al3 explored the association between urinary dialkylphosphates (DAPs) metabolites of OP [including diethylphosphate (DEP), diethylthiophosphate (DETP), diethyldithiophosphate (DEDTP), dimethylphosphate (DMP), dimethylthiophosphate (DMTP) or dimethyldithiophosphate (DMDTP)], which are eliminated in the urine within 6–24 h after exposure and can be used as a surrogate of OP exposure and biomarkers of liver injury (AST, ALT and ALP and composite fibrosis and steatosis scores) and liver function (bilirubin, total protein and albumin). They used different cohorts; cohort 1 enrolled 5952 participants for the outcomes of ALT, AST, AST/ALT, TBIL, ALB, TP and FIB-4; cohort 2, included 4953 participants for association with ALP. Besides, cohort 3 (5488 participants) and cohort 4 (2639 participants) were evaluated for hepatic steatosis index (HSI) and US fatty liver index (USFLI) score, respectively. The authors found that DMP, DEP and DMTP were significantly associated with AST/ALT ratio and FIB-4 score, as well as that urinary DMTP and DETP concentrations were negatively associated with serum albumin and total protein levels, indicating for the first time a potential toxic effect of ‘long-term’ OP exposure to human liver.3 The source of data for the study was the National Health and Nutrition Examination Survey (NHANES). The NHANES is an ongoing, national, cross-sectional series of surveys with multistage complex sampling strategies, representative of the health and nutritional status of the noninstitutionalized civilian (adults and children) United States population. This program consists of an interview about demographic, socio-economic, dietary and health-related questions, and a health examination that includes medical and dental examination, physiological measurements and laboratory tests. This database is part of the US Centers for Disease Control and Prevention, and therefore we can safely assume that high-quality data were retrieved.4 Notably, the authors performed appealing analyses to study the associations between OP exposure and biochemical liver parameters, which deserve detailed comments. First, the authors used a weighted multivariate linear regression to assess the relationship between OP metabolites and liver enzymes, fibrosis and steatosis scores. This weighted linear regression approach is an extension of the simple linear regression that allows for the heteroscedasticity of the dependent variable by providing a more realistic fit by ‘minimizing’ the effect of the points with greater absolute variation and forcing the regression line to pass closer to the points with smaller variance.5 Thus, this analytic method allowed to study the associations between exposure and outcome without the constraints of the simple multivariate linear regression. In some epidemiological studies conducted in specific contexts, where non-linearities can be anticipated, the assumption of a linear dose–response is misleading. The authors wisely expanded their analyses beyond a first trend test performed after modelling the OP exposure as an ordinal variable and decided to test the non-linearity relation between the exposure and liver markers. Restricted cubic splines are used when the relationship between the outcome and the explanatory variable is not linear (Figure 1). They allow a transformation of the continuous independent variable into a categorized variable, in which the range of values of the independent variable is split up, with ‘knots’ defining the end of one segment and the start of the next.6 Lastly, to estimate the key components contributing to the deleterious effect of OP metabolites on the liver, a weighted quantile sum (WQS) regression was performed. The WQS approach estimates a weighted linear index in which the weights are empirically determined by bootstrap sampling.7, 8 In the context of a study in which analysis is conducted with components that present with a complex correlation pattern, the election of WQS regression over other shrinkage methods such as lasso regression is founded on the limitation of this latter method when predictor variables are highly correlated.9 Beyond the statistical approach used in this study, its findings deserve further comment. Environmental pollutants and pesticides have been involved in causing (and worsening pre-existing) non-alcoholic fatty liver disease, the so-called “toxicant” associated steatohepatitis (TASH) being the typical phenotype of liver injury, as, for instance, has been described in workers of the petrochemical industry majority of whom did not have insulin resistance suggesting that fatty liver was a direct effect of the volatile compounds for which they were exposed to.10 Although Li and coworkers in a recently published article of similar design found that among a mixture of chemical compounds, organophosphate pesticides were associated with fatty liver measured by the liver steatosis indexes, HSI and USFLI (OR: 1.32, 95% CI 1.15–1.51)11, in the present study, however, they did not replicate their previous findings.3 Instead, a novel and intriguing finding of this study is the association of some urinary OP metabolites with parameters of liver fibrosis and liver dysfunction, which suggest that long-term exposure to these pesticides might apparently lead to subclinical liver damage severe enough to affect liver function. The mechanisms for which OP exposure would result in liver fibrosis and hepatic dysfunction are elusive. In rodent models, OP were shown to induce the production of reactive oxygen species and oxidative liver damage.12 Presumably, with continued exposure to these pesticides and persistent necro-inflammatory changes in the liver fibrosis and hepatic dysfunction would ensue but this hypothesis is highly speculative at present time. Indeed, some of the assumptions by Li et al3 must be toned down, as the authors have acknowledged, because of the retrospective nature of the study and its cross-sectional design with urinary metabolites measured only once, which precludes one hand form clearly ascertaining the causality of exposure to OP and disruptions in liver biomarkers and does not reflect the toxic effect of long-term exposure. Therefore, further cohort studies in people exposed to low-level organophosphate pesticides with a prospective design are warranted to confirm the findings presented in this manuscript. In the meantime, is advisable to consider exposure to pesticides in the assessment of transaminase elevation and hepatic fibrosis of indeterminate cause. None. The authors declare that they have no conflict of interest. Not applicable. Not applicable. Not applicable.