Towards an alternative to nano-QSAR for nanoparticle toxicity ranking in case of small datasets

Abstract

Statistical analysis approaches have been developed to predict the biological response to nanoparticles, especially quantitative structure-activity relationship (QSAR) models. But one major limitation remains the quantitative lack of data to build accurate models. The aim of this study was to investigate if simple alternative mathematical models could rank nanoparticles in a very binary way (i.e. toxic or not) in case of small dataset. We synthesized and characterized 25 nanoparticles from 6 metal (hydr)oxide families with particle size and shape tuning. We assessed their toxicity on RAW 264.7 cells and investigated relationships with both physicochemical and dimensional descriptors. A simple partial least square (PLS) regression analysis allowed ranking nanoparticles with respect to their toxicity, without false-negative results. But this model was not predictive due to the specific response of each family to dimensional parameters variations. A classification tree extracted the same main bulk descriptor as PLS, but interestingly showed the relevance of dimensional descriptors for the second and third node. We thus recommend the development of family-specific models and propose the combination of these two simple methods as pre-screening tools, a compromise to bridge the gap between case-by-case studies (expensive and time-consuming) and sophisticated nano-QSAR models (not suitable for small datasets).

Synthesis of Metarhizium anisopliae–Chitosan Nanoparticles and Their Pathogenicity against Plutella xylostella (Linnaeus)

“…A proper classification/identification of insect pathogenic microorganisms is a basic/initial requirement for their development as biopesticides [24]. The identification of fungi is mainly carried out through the comparison of the internal transcribed spacer fragments (ITS) of DNA [33].…”

“…These properties will allow decreased pesticide doses and achieve greater pest control without repeated treatments [23]. Some circles of the scientific community still think that the use of nanopesticides is not fully safe as chemical entities with unknown physicochemical and toxicological properties are released into the environment [24]. However, the use of quantitative structure-activity relationship/quantitative structure-property relationship (QSAR/QSPR) tools offer viable solutions to address experimental issues observed during pesticide risk assessment [25].…”

Experimental and Computational Nanotoxicology—Complementary Approaches for Nanomaterial Hazard Assessment

“…In addition, nanomaterials represent very structurally diverse groups of chemicals, making it difficult to build a significant dataset of structurally related nanomaterials [152]. To sum up, a large quantity of experimental data of a high quality should be available to develop a reliable nanoQSAR model [150,152,156,[162][163][164][165][166]. Stronger collaborations between experimentalists and modelers should result in more useful datasets and allow QSAR predictive models to improve their potential [154,161,167], as shown in Figure 3.…”

“…Regarding the statistical methods used for QSAR models, different types of machine learning algorithms can be used (Figure 3), including principal component analysis (PCA), multiple linear regression (MLR), partial least squares (PLS) methods, random forests (RFs), support vector machines (SVMs), k-nearest neighbors, Bayesian networks, and artificial neural networks (ANNs) [20,149,166,172,[174][175][176][177][178][179][180][181][182][183][184].…”

Nano-(Q)SAR for Cytotoxicity Prediction of Engineered Nanomaterials

“…Their toxicity was evaluated using RAW 264.7 cells. A model with four chemical composition-related descriptors (metal cation charge, hydration rate, radius of the metallic cation, and Pauling electronegativity) was also built [53].…”