Many in vitro methods are being developed, evaluated, and validated for inclusion in an integrated testing approach for the prediction of acute systemic toxicity. The assembly of the endpoints from these assays into a predictive test battery/scheme, and the prospective validation of the test scheme, will be the next challenges. Walum, et al. (2005) discussed a number of in vitro screening assays and their possible integrated use for decision making but noted that the concepts of prospective validation for this purpose need to be developed. ECVAM established a Task Force on Integrated Testing Strategies in 1999 for the purpose of providing suggestions on the design and implementation of integrated strategies (Gennari, et al., 2004).
A high-throughput version of the 3T3 NRU cytotoxicity assay was evaluated for compatibility with the ICCVAM-validated prediction model (PM) for this assay (King & Jones, 2003). The purpose of this industry study was to use the NRU assay as one component of a high-throughput test strategy for eliminating the most toxic substances and prioritizing the remaining ones. Two prediction models generated with the high-throughput NRU assay fit reasonably well within the confidence limits of the ICCVAM PM, indicating that the in-house NRU assay can “be applied for estimating acute rodent oral toxicity in the manner outlined in the NIEHS reports.”
The ACuteTox program has developed an integrated approach with defined research programs for the development of relevant models and assays. However, exactly how the final endpoints will be selected and used in a test scheme still needs to be determined. For additional information on current approaches to using integrated testing strategies for toxicity testing please refer to Emerging Technologies: Integrated Testing Strategies & Risk Assessment.
Test batteries may be required for addressing some of the endpoints within an integrated testing scheme. For example, simple cytotoxicity assays using HepG2 cells have not been sufficient for identifying hepatotoxic drugs. However, “a panel of pre-lethal mechanistic cellular assays” was found to be more specific in detecting drug toxicity (Xu, et al., 2004). This approach for in vitro drug screening used high content screening (HCS) assays – automated high-throughput assays that simultaneously assess multiple mechanistic endpoints. Cellular parameters affected by drugs included cell number, nuclear area, mitochondrial membrane potential, membrane permeability, and intracellular calcium (O’Brien, et al., 2006). The test battery showed 80% sensitivity and 90% specificity in predicting human toxicity potential. The researchers concluded “that human hepatotoxicity is highly concordant with in vitro cytotoxicity in this novel model and as detected by HCS.”
In a subsequent study, O’Brien, et al. (2007) observed HCS assay results with 87% sensitivity and more than 90% specificity. These researchers have found the HCS assays that evaluate multiple parameters to be “more predictive because they cover a wider spectrum of effects,” and they therefore “address some of the limitations of traditional in vitro methods.”
A (Q)SAR model based on in vitro cytotoxicity data has been developed for use in predicting acute toxicity in rodents. In vitro (cytotoxicity) and in vivo (rat and mouse intravenous and oral LD50 values) data from ZEBET’s Registry of Cytotoxicity were reassessed for mechanism-based chemical subclasses (Freidig, et al., 2007). The predictions from this model tend to overestimate toxicity for irritants, meaning that substances with predicted no or low toxicity could be eliminated from additional in vivo testing. Therefore, application of this (Q)SAR model “could result in the elimination of numerous unnecessary acute in vivo toxicity tests for nontoxic compounds and focus in vivo testing on cases with potentially higher risks for humans.” The proposed (Q)SAR model is considered to have “almost completely fulfilled” the principles for validation of (Q)SAR models. The type of in vitro data for the model is cytotoxicity, which occurs by various nonspecific effects, so that the fifth validation principle, “a mechanistic interpretation of the relation,” may not be fully met at this time.
Additional topics in emerging research, methods, and policies for acute systemic toxicity testing will be provided on AltTox.org in the future; please check back again.