Ecotoxicology – NOEC and LOEC
Ecotoxicity tests are experiments performed to evaluate the toxicity of a test substance in order to predict the effect on natural populations. There are many different types of test, but the general principle is that organisms are exposed to different concentrations or doses of a test substance, with at least one control group that is not exposed but is otherwise treated identically. A range of measures may be collected, covering outcomes such as survival, growth and reproduction.
NOEC and LOEC
Historically, OECD guidance recommended summarising environmental risk using the no observed effect concentration (NOEC) and the lowest observed effect concentration (LOEC).
The LOEC is the lowest tested concentration that is significantly different from control. The NOEC is the tested concentration immediately below the LOEC which, when compared with the control, has no statistically significant effect (p < 0.05) within a given exposure time.
Key Takeaways
- NOEC (No Observed Effect Concentration) and LOEC (Lowest Observed Effect Concentration) are conventional measures in environmental risk assessment that rely on specific test concentrations.
- These measures can be problematic because they imply a “no effect” level when it’s merely the highest tested dose without statistically significant effects, lack estimates of variability or uncertainty, and fail to capture the full concentration-response relationship.
- Regulatory guidance is increasingly favoring regression-based procedures, which model the entire concentration-response curve to derive effective concentration values (e.g., ECx) with confidence intervals.
Note that NOEC/LOEC refer to tested concentrations. If the tested concentrations are widely spaced, the reported LOEC and NOEC may differ substantially from what would have been reported had the tested concentrations been more closely spaced.
NOEC limitations
- Name has potential to mislead. NOEC is not equivalent to a true “no effect concentration”. It is simply the highest concentration used in a particular test that did not show a statistically significant effect versus control over the test period.
- No estimate of variability / uncertainty. NOEC/LOEC can give a false impression of certainty because they do not provide confidence intervals or other uncertainty measures.
- No information on the concentration–response relationship. They reduce the full dataset to a single threshold-like label and do not describe how response changes with concentration.
- Not the most efficient use of data (and test organisms). Focusing on hypothesis tests at individual concentrations can waste information that could be used to characterise the full response curve.
The OECD have not fully rejected reporting NOEC/LOEC, but in more recent guidance they have encouraged moving away from these measures as the main summary parameters, recommending regression-based procedures instead.
Regression-based procedures
What do we mean by a “regression-based procedure”? This involves fitting a mathematical model to measurements at each concentration to estimate the concentration–response relationship.
Once the concentration–response relationship has been estimated, we can calculate effective concentration values (ECx). ECx is the concentration that causes a response that is x% of the maximum. Sometimes specific acronyms are used, for example LD50, the dose that is lethal in 50% of a sample.
Regression-based procedures resolve many of the major disadvantages of NOEC/LOEC. Modelling allows confidence intervals to be calculated around ECx estimates, and can support estimation in scenarios that would require a new experiment under a NOEC/LOEC framework (for example, when the LOEC is the lowest tested concentration and the NOEC is undefined).
Regression-based procedures are not without drawbacks. While the OECD has suggested phasing out NOEC/LOEC estimation, for now these measures remain in the guidance, but with a reduced role.
A more detailed review of recent EFSA guidance on this topic will be the subject of a future blog post.
