With the rapid development of industry and economy, the emergence of a large number of chemicals has made of risk management more difficult. Traditional risk assessment relies on animal experiments for toxicity testing. However, animal experiments are time-consuming, costly, and unable to meet the practical needs of risk assessment. The increasing maturity of toxicity testing alternative technologies signifies the possibility of rapid, sensitive, and accurate identification of chemical toxicity. This article focuses on the research and applications of alternative toxicity testing by reviewing the background, developments, and current research at home and abroad. It also discusses the progress in alternative testing methods in such areas as cosmetics and food safety risk assessment and explores the problems with the development of alternative testing technologies and risk assessment in China. This review aims to provide a reference for the system construction of cosmetics health risk assessment in China.

Food toxicology plays a crucial role in supporting scientific and technical aspects of food safety risk assessment. However, traditional methods relying on animal testing are becoming increasingly inadequate for identifying and evaluating emerging foods and unknown risks. There is a strong push worldwide towards the development of new approach methodologies (NAMs) based on non-animal testing methods. Policies and regulations related to NAMs are being standardized gradually in the European Union, the United States, and China. Some progress has also been made in applying these methodologies in food toxicology research in China. For instance, within the "Food Toxicology Program" at the National Center for Food Safety Risk Assessment, high-content and high-throughput in vitro hazard identification models employing model organisms like human macrophages, hepatocytes, adipocytes, embryonic stem cells, and zebrafish, as well as Toxicological Thresholds of Concern and quantitative in vitro to in vivo extrapolation based on physiologically-based toxicokinetic models have been established and applied. Nonetheless, new toxicological hazard identification technologies still face challenges such as inadequate elucidation of toxic mechanisms, insufficient collaborative research efforts, and inefficient translation of these findings into practical applications.

Integrated Approaches to Testing and Assessment (IATA) is a toxicological assessment decision-making method that integrates existing information from various sources, including physical and chemical properties, animal testing, and non-animal testing. Through the evaluation and analysis of a series of iterative strategies, it ultimately obtains risk assessment conclusions, thereby providing a basis for risk management decisions regarding chemical substances. The use of IATA is becoming increasingly prevalent in such areas as ocular irritation and genotoxicity. This paper introduces the conceptual connotation of IATA, sorts out the framework elements and sequential processes, explains the commonly used framework construction methods, shares cases of application in various exposure scenarios, and finally envisions future research directions in order to provide better methodological support for the risk assessment of chemical substances.

With the explosive growth of knowledge and technologies in biological disciplines represented by omics techniques, which promise broad applications in the field of toxicology. New methods of alternative toxicology to non-animal testing based on transcriptomics are entering the regulatory phase. However, given the limitations of a single omics technique in risk assessment and the fact that the transition from scientific research to regulatory specification has not yet been realized, risk assessment will be helped to be accomplished more accurately by integrating multiple new approach methodologies (NAMs) and different information sources. This review introduces genomics, proteomics, protein post-translational modification genomics and metabonomics, which are currently the focus of attention both at home and abroad, and their prospective applications in chemical risk assessment. 

Chemical pollution-induced damage to ecosystem function has been a global challenge. The latest "Kunming-Montreal Global Biodiversity Framework" proposed reducing pollution risks to levels harmless to biodiversity and function, placing higher demands on chemical risk management at the ecosystem level. Conventional ecotoxicity tests have focused on single species, only to neglect genetic diversity protection and simplify species interactions. Here, we proposed using environmental RNA (eRNA) and metatranscriptomic analysis to establish a multi-species, multi-biological level chemical pollution ecological risk assessment approach in exposed communities. We reviewed the current status and trends of eRNA in chemical pollution risk assessment and proposed a strategy for bioeffect testing from molecules to communities based on eRNA, constructing ecological risk assessment models for different protection goals. Finally, we summarized the theoretical and technical challenges facing eRNA-based toxicity testing and outlined the future applications of eRNA in capturing real ecological effects of chemical pollution in the field.

New approach methodologies (NAMs) based next generation risk assessment for toxicity testing has quietly emerged globally. As an important component of NAMs, quantitative in vitro to in vivo extrapolation (QIVIVE) has come to be one of the significant technologies for chemical toxicity testing and risk assessment, and has gained rapid development and wide applications in the field of toxicology. The implementation of QIVIVE technology relies on two vital branches of computational toxicology: physiologically based toxicokinetics and physiologically based toxicodynamics. The system integrates population research, in vivo/in vitro testing, and computational toxicology research to quantitatively study the nonlinear quantitative relationship between exposure doses and physiological response, providing an efficient and accurate means of toxicity testing for health risk assessment of chemicals. At the same time, utilizing increasingly mature multi omics technologies to obtain high-throughput and 
effective toxicology information, and drawing a signal network of toxicity occurrence and development in the context of big data can greatly improve the efficiency and range of toxicology research.

With the emergence of high-throughput technology and massive toxicology data, toxicology research has entered the era of big data. How to efficiently integrate existingtoxicological data, clarify the toxic effects of chemicals, and use these patterns to providenew information, in order to achieve efficient prediction of the toxicity of new chemicalsubstances, is one of the cutting-edge issues in toxicology. In view of the high cost, low throughput and difficulty in revealing the mechanism information of traditional chemical toxicity testing methods, high throughput prediction models are urgently needed. Machine learning methods have been applied to toxicity testing, such as supervised learning models, unsupervised learning models, deep learning models, reinforcement learning models, and transfer learning models. Chemical characteristic data commonly used in machine learning models include chemical structure data, text data, toxicological genome data and image data. There is huge potential for applying machine learning to toxicity testing and machine learning methods have made some progress. However, current research focuses on the processing of data and development of  models, which has failed to produce a widely used and accepted method. In addition, the prediction accuracy of machine learning models is not only dependent on algorithms, but also affected by data quality, and the mutual promotion and development of algorithms and data quality remains a big challenge. In short, data 
processing and model construction in the field of toxicology require interdisciplinary cooperation and technological innovation. With the increasing perfection of toxicology databases and the continuous optimization of various model algorithms, the toxicity prediction of new chemicals based on machine learning models will become increasingly efficient and accurate, playing an important role in ensuring human health and environmental safety.

As a critical quantitative step in risk assessment, dose-response assessment is intended to identify a safe dose level based on experimental dose-response data. Compared with the traditional approach (no observed adverse effect level, NOAEL), the benchmark dose (BMD) methodology not only has several important advantages, but also greatly facilitates the 3R principle, e.g., reduction, replacement, and refinement. In this review, after a brief introduction to the NOAEL approach and BMD methodology, three important features of the BMD methodology that are closely related to its ability to facilitate animal use reduction are discussed, including less dependence on study design, better quantification of uncertainty in point of departure （POD）, and capability to combine additional information. Then, quantitative approximations are performed to study the extent to which the BMD methodology can reduce the animal use through better utilization of dose-response data and data integration. As the BMD methodology is being increasingly accepted as the preferred dose-response modeling approach, it will have a long-term impact on chemical risk assessment and facilitate the 3R principle.

The quality of toxicological data is critical to toxicity assessment and risk prediction. Integrating multiple streams of evidence and enhancing the quality of toxicological data are indispensable for more efficient and precise health risk assessment. There are several commonly used toxicology data quality evaluation systems abroad, which include Klimisch rating system, ToxRTool evaluation tool, the SciRAP evaluation tool and the integrated risk information system tool. The TRAM reliability evaluation tool is developed in China for toxicology data in food safety risk assessment.  These evaluation tools have different backgrounds and applicability of evaluation standards. Each evaluation system also has its advantages and disadvantages.

OBJECTIVE  To explore the applicablity of ′BlueScreen HC′(BSHC), a high throughput genotoxicity screening system based on human growth arrest and DNA damage inducible 45α(GADD45α) gene, in detecting the genotoxicity of migrants mixtures from food contact materials (FCM). METHODS  The 2000 bp sequence upstream of the open reading frame of human GADD45α gene was used as the promoter to construct the lentiviral plasmid pEZX-LvPG04, which was double labeled by purinamycin and Gausluciferase (Gluc), and the lentiviral plasmid was infected with human lymphoblastocyte TK6 to obtain a stable transmutation cell line TK6-Gluc.  Methyl methylate (MMS) at concentrations of 0, 1.56, 3.13, 6.25, 12.5, 25.0 and 50.0 mg·L^-1 was selected as the genotoxin without liver S9, cyclophosphamide (CTX) 0, 0.78, 1.56, 3.13, 6.25, 12.5, 25.0 mg·L^-1 was selected as the pre-genotoxin with liver S9, and dimethyl sulfoxide (DMSO) 0, 0.35, 0.69, 1.38, 2.75, 5.5 and 11.0 g·L^-1 was selected as the non-genotoxin. The constructed BSHC was verified with the above known genetic positive and negative substance respectively. Polybutyleneadipate-co-terephthalate (MS/PBAT) was tested using 4% (V/V) acetic acid, and 10%, 20%, 50% and 95% (V/V) ethanol as food simulants at 40 ℃ for 24 hours to obtain 5 multi-component migrants of MS/PBAT that were obtained by using DMSO as a solvent. TK6-Gluc cells were treated with 5 multi-component migrants of MS/PBAT at concentrations of 0, 0.38, 0.76, 1.53, 3.05, 6.10 and 12.20 g·L^-1 with or without liver S9. Cells were treated without liver S9 for 48 h. Cells treated with liver S9-mix were incubated for 3 h at a final concentration of 1% (V/V) liver S9 before being washed and re-suspended in fresh recovery media for another 45 h. After exposure, the cell viability was detected using the CCK-8 cell activity kit, and the Gluc Luminescence in the medium was detected with Secrete-PairTM Gaussia Luciferase Assay Kit. In addition, the mutagenicity on Salmonella typhimurium TA98 and TA100 was detected by micro-fluctuation Ames test with 5 multi-component migrants of MS/PBAT at concentrations of 3.05 and 12.20 g·L^-1. The in vitro mammalian cell chromosome aberration test was performed on CHL cells with 5 multi-component migrants of MS/PBAT at concentrations of 3.05 and 12.20 g·L^-1 to detect the chromosomal aberration. The results of genotoxicity were compared with those of BSHC. RESULTS  The lowest effect centration (LEC; <80% relative cell viability) and the coytotoxicity (<30% relative cell viability) was defined. A positive genotoxicity result threshold was determined at 1.8-fold relative induction. For the liver S9 protocol, the same process was followed, and the decision threshold derived was 1.5-fold relative Gluc induction. It is considered as genetic substance only when a positive genotoxicity result was reached and there was no cytotoxicity. Compared with the vehicle control group, no genotoxicity was observed at all concentration of DMSO by BSHC. MMS 12.5, 25.0 and 50.0 mg·L^-1 produced genotoxicity without liver S9 while CTX 6.25, 12.5 and 25.0 mg·L^-1 produced genotoxicity with liver S9. Significant cell growth inhibition was observed in 95% ethanol migrants of MS/PBAT at concentrations of 6.10 and 12.20 g·L^-1, and in 50% ethanol migrants of MS/PBAT at a concentration of 12.20 g·L^-1 without liver S9. No cytotoxicity with a relative cell viability below 30% was observed in any of the treatment groups, and no high expression of Gluc was observed. Therefore, none of the 5 multi-component migrants produced genotoxicity without liver S9. Significant cell growth inhibition was observed in 95% ethanol migrants of MS/PBAT at a concentration of 12.20 g·L^-1, and in 4% acetic acid migrants of MS/PBAT at concentrations of 6.10 and 12.20 g·L^-1 with liver S9. No cytotoxicity with a relative cell viability below 30% was observed in any of the treatment groups. No high expression of Gluc was observed. Therefore, none of the 5 multi-component migrants produced genotoxicity with liver S9. In the micro fluctuation Ames test, when 5 multi-component migrants of MS/PBAT were treated with concentrations of 3.05 and 12.20 g·L^-1 on TA98 and TA100 strains, there was no significant difference in the number of mutagenic positive wells compared with DMSO control group with or without liver S9,  indicating that no mutagenic effect was produced.  When CHL cells were treated with 5 multi-component migrants of MS/PBAT at concentration of 3.05 and 12.20 g·L^-1, compared with DMSO control group, there was no significant difference in chromosome aberration rate of CHL cells with or without liver S9. CONCLUSION  BSHC based on GADD45α gene has been established, which can be used for in vitro genotoxicity evaluation of migrants mixtures of FCM, but further exploration of its minimum effective concentrations is still needed, and more types of mixtures need to be applied for further validation.