Classifying chemicals into toxicity categories based on LC50 values- Pros and cons

It is a usual practice to determine LC50 value in acute toxicity studies conducted in aquatic organisms as an initial step to assess the toxicity of chemicals. In regulatory toxicity studies, normally conducted in GLP (Good Laboratory Practice) certified facilities, acute toxicity of chemicals is evaluated in fish, crustacea, and or alga following the methods given in OECD Guidelines. The chemicals are classified into different toxicity categories based on the LC50/EC50 determined from the acute toxicity studies. For calculating LC50 in acute toxicity tests, the methods given in the OECD (2019) Guidelines are Probit or Logit Analysis (Litchfield & Wilcoxon method and Probit Analysis), Spearman-Karber method, the binomial method, the moving average method, and the graphical method. LC50 is the concentration of a substance that causes 50 % mortality in a batch of test organisms (eg. fish). In acute toxicity studies with laboratory animals like rats, mice, rabbits, etc, instead of LC50, the terminology LD50 is used. The procedure for the calculation of both LC50 and LD50 is same. In this article, LC50 and LD50 are written interchangeably. It means if 100 fish are exposed to LC50, theoretically 50 fish would die. In fact, the inventor of LC50 (Trevan, 1927) defined LC50 as the median lethal concentration. Like any other median value, the LC50 is not affected by extreme values of either side. Unfortunately, Trevan was ruthlessly misquoted by the animal ethicists, as they believed that he was responsible for killing millions of animals for determining the median lethal concentration. According to Rowan (1983), the median lethal concentration in animals varies considerably among the species and is affected by environmental factors. Trevan proposed median lethal concentration (LD50) in frogs and rodents for biological standardization of digitalis extract, insulin, and diphtheria toxin when he was working at Wellcome Research Labs, Beckenham (Pillai et al., 2021a). Trevan never promoted sacrificing more animals to determine median lethal concentration. He was aware of the fact that the determination of median lethal concentration is affected by several factors. The 'characteristic' of a dose-response curve proposed by Trevan is species and test substance-specific. However, after Trevan, LD50s were determined in acute toxicity studies to evaluate the effect of a substance, not for the biological standardization of drugs. His intention was to establish a numerical quality control standard to assess batch-to-batch variation, if any, of the therapeutic products of the Wellcome Research Labs. Based on the LC50/EC50 values determined in aquatic toxicity studies, the chemicals are classified into a hazard category. For example, according to United Nations Global Harmonized System (GHS), if the 96 LC50 of a chemical to fish is ≤ 1 mg l-1 , this chemical is classified into hazard category I (GHS, 2019).Though several methods are prescribed in OECD (2019) Guidelines, if the mortality data are adequate, Probit Analysis of Finney (1978) and Litchfield and Wilcoxon (1949) method may be preferred to determine LC50 as these methods provide additional valuable information on the concentration-mortality relationship. If the lowest mortality obtained is close to 16% and the highest mortality is close to 84%, most of the above-mentioned methods would result in a more or less similar LC50value (Pillai et al., 2021a).Calculation of LC50 manually by the Litchfield and Wilcoxon method is somewhat easier, but Probit Analysis is a bit cumbersome. Commercial statistical software is available for the calculation of LC50 by both the above methods. But, using the software without understanding the underlying concepts of the statistical methods has certain disadvantages. Researchers also present the toxicity of a substance in terms of LC10, LC90, etc. Since the variability of these estimates is large, their biological relevance is limited. Concentration-mortality curve in the 16-84% mortality range is linear, hence the LC50 determined from this concentration-mortality curve is reliable. The method of Litchfield and Wilcoxon (1949), uses the 16-84% mortality range for calculating LC50. This method does not consider mortality below 16 and above 84% for the LC50 calculation. But Probit Analysis by Finney (1978) considers all mortality values (excluding 0 and 100 % mortality) for the calculation of LC50. Researchers in academic institutions use LC50 values to compare the toxicity of the test substances - the lower the LC50, the substance is more toxic, and vice-versa (Islam et al., 2021).Toxicity grading of substances solely based on LC50 is inappropriate. Recently, the appropriate use of LC50 values for the GHS classification of chemicals has been questioned (Pillai et al., 2021a). LC50s vary in a wide range from one species to the other (Geyer et al., 1993) and many times are irreproducible within the same species (Peres and Pihan, 1991), as the physico-chemical parameters of dilution water play a crucial role in LC50 experiments. Hrovat et al. (2009) reported significant variability of fish LC50 test results for 44 compounds. A consistent LC50 could not be obtained in more than 750 tests conducted on fathead minnows with 644 chemicals (Mc Carty, 2012). It is a statutory requirement for the United Nations GHS that the environmental hazards should be mentioned on the labels of chemicals for distribution. The European Chemicals Agency (ECHA, 2017) uses fish LC50 for the environmental classification of a chemical according to the GHS of Classification, Labelling and Packaging of Chemicals (Paparella et al., 2021). The major disadvantage of such labelling is that the LC50 value alone does not provide information on the toxicity profile of chemicals. Showing a similar LC50 does not mean that the toxicity profile of the chemicals is same. It is important to consider the slopes of the concentration-mortality curve when comparing the LC50s of the chemicals. The slope which reflects the concentration-mortality relationship provides a better understanding of the causality between a toxicant and response (Tsatsakis et al., 2018). In Probit Analysis, parallel regression lines of mortality probits on log concentrations indicate that the mode of action of chemicals on test organisms is similar (Finney, 1978). If the regression lines are not parallel, it is a clear indication that the chemicals possess different modes of action on that particular organism. Also, it is important to present LC50 with 95% confidence limits. If the 95% confidence limits of LC50s of the chemicals are distinctly separate, LC50s can be considered different from each other. The LC50s cannot be considered different from each other if the 95% confidence limits of the LC50s overlap. Chemicals with similar LC50 values may manifest toxicity differently. Similarly, chemicals with different LC50 values may manifest similar toxicity effects; hence, the classification of chemicals into various groups based on LC50 values may not have much relevance (Pillai et al., 2021b). Ethical conduct of fish toxicity studies and euthanizing of exposed fish are emphasized in the OECD (2019) and CCSEA (Committee for Control and Supervision of Experiments on Animals) Guidelines (CPCSEA, 2021). Earlier the fish toxicity studies were conducted with 10 fish exposed to each test concentration, but the revised OECD (2019) Guideline recommends a minimum number of 7 fish for each test concentration. The probable mortality data that can be obtained in an acute test where 7 numbers of fish are exposed to each test concentration are (number of fish died/total number of fish exposed) 0/7, 1/7, 2/7, 3/7, 4/7, 5/7, 6/7, or 7/7. For calculating LC50 values by the methods of Litchfield and Wilcoxon (1949) and Finney (1978), 0 and 100% mortality are not used, since no probit values can be assigned for 0 and 100% mortality. The remaining 6 numbers of mortality data are adequate for calculating a reliable LC50 value, if the mortality data spreads over all phases of the concentration-mortality curve, particularly covering 16-84% mortality region. If the mortality data does not spread over all the phases of the concentration-mortality curve in a concentration-dependent manner, the confidence limits of LC50 could be exploded (Pillai et al., 2021b). Estimation of LD50 in rodents by the methods of Litchfield and Wilcoxon (1949) and Finney (1978) is discouraged by US Consumer Product Safety Commission, US Environmental Protection Agency, US Food and Drug Administration, National Toxicology Program, and OECD, due to ethical reasons and poor reproducibility of LD50 values. But, classical methods are used to determine LC50 values in environmental toxicity studies, especially with aquatic organisms. It is more biologically relevant to interpret LC50 in terms of the slope of concentration-mortality curve and confidence interval of LC50. My association with Dr. R.C. Dalela and Journal of Environmental Biology began in the early 1980s when he was working at D.A.V College Muzaffarnagar. His research work and enthusiasm for bringing up the Journal of Environmental Biology to an international standard fascinated me. I realized from his research work that he was a committed environmentalist. I had an opportunity to majorly organize two national conferences of the Academy of Environmental Biology. He always occupied the front row in the conferences listening to all scientific presentations keenly. He had taken a lot of hardships to bring the journal to this sustainable level with a WOS Impact Factor of 0.70. I remember as it had happened yesterday, my meeting with him at D.A.V. College, Muzaffarnagar, at JRF, Vapi, Marathwada Ambedkar University, Aurangabad, and in Chennai. He was an excellent teacher, a great scientist, a mentor to several researchers, and self-disciplined.