Grayscale differential box counting as a measure of complexity of liver texture in common carp (Cyprinus carpio) sub-chronically exposed to perfluorooctanoic acid (PFOA)

Perfluorooctanoic acid (PFOA), a perfluorinated alkylated substance (PFAS), poses a worldwide concern for its wide distribution, bioaccumulation in food webs, long half-life in organisms, and potential toxic, carcinogenic and endocrine disrupting effects on animals. Fish are excellent candidates in aquatic biomonitoring programs and toxicologic testing, frequently focusing on liver due to its pivotal role in the health of the whole organism and its highly sensitivity to contaminants. Histopathology can be useful in evaluating toxicological effects on fish health and texture analysis can represent an objective, replicable diagnostic tool, potentially free from operator-dependent bias. In the present survey, liver histological texture was comparatively assessed in specimens of common carp (Cyprinus carpio) sub-chronically exposed to PFOA. Twenty specimens were exposed to two PFOA dosages (10 exposed to 200 ng l-1, 10 exposed to 2 mg l-1) for 56 days and compared to other 10 unexposed fish. Grayscale differential box counting (fractal dimension and lacunarity) was evaluated on representative pictures taken from liver histological sections. Differential box counting was implemented by converting two-dimensional grayscale images into pseudo three-dimensional information. Hence, fractal dimension and lacunarity acted as a measure of the complexity and of the heterogeneity of the grayscale levels distribution, respectively. Redundancy Analysis (RDA) was performed on the obtained numerical data in order to summarize the part of grayscale differential box counting variation that is explained by the following biometric/experimental variables: PFOA liver concentration, liver mass, proliferating cell nuclear antigen (PCNA) positive nuclei, after removing the effects of fish total length. The t-values of the regression coefficients of liver PFOA concentration and of liver mass with both fractal dimension and lacunarity, and of PCNA positive nuclei with lacunarity, showed values larger than 2, while the t-value of the regression of PCNA positive nuclei with fractal dimension appeared to be close to 2. Considering the selected biometric/experimental variables, liver PFOA concentration correlated with PCNA positive nuclei but did not correlate with liver mass, whereas PCNA positive nuclei correlated with liver mass. Interestingly, fractal dimension contributed better than lacunarity in treatment groups ordination. Recently, fractal analysis has been adopted to estimate the complexity loss associated with pathological changes. In the present survey, contrary as expected, liver texture modification related to liver PFOA concentration increase was associated with a significant complexity increase, related to reversible changes (hydropic degeneration), possibly acting as an initially adaptive strategy, rather than representing mere degeneration, to cope with PFOA challenge. The possible occurrence of a hormetic response should be further investigated