Taking a realistic approach to risk assessment

Not all highly potent compounds require extreme containment. In this article, Jeff Pavlovich, senior process safety engineer at Cambrex Charles City, examines the intricacies of risk assessment and the importance of a realistic approach.

There is a growing demand for capacity for the manufacture of highly potent active pharmaceutical ingredients (HPAPIs). This is partly due to the growing number of molecules in drug development pipelines that are thought of as highly potent, but it also reflects the fact that some of these drugs are now reaching patent expiry, and are now subject to generic competition.

HPAPIs need to be carefully handled in a contained environment and before a synthesis is begun, a full assessment of the potential hazards of manufacturing and handling of the products must be carried out. As well as the HPAPI itself, every reagent and intermediate requires assessment, and for a contract development and manufacturing company (CDMO) or plant manager, there is a careful interplay of ensuring the appropriate potency strategy for the safety of operators and the local environment, and the avoidance of overlaying excessive operating costs to customers by over-specifying the containment required.

Defining potency

If a compound has an eight-hour time-weighted average occupational exposure limit (OEL) of 10 µg/m³ or less, it is deemed ‘potent’. In the absence of a formal definition of ‘highly potent’, however, different risk assessors may define the hazards posed by an individual compound differently. Some take a very conservative view, and may decide that most potent molecules are in fact highly potent; whereas others will deem very few to be highly potent.

It is important to remember that potency is not the same as toxicity. A drug may be both highly potent and toxic, but this is not necessarily the case. A highly potent drug is one for which only a very small dose is required to give a therapeutic effect, but by no means is it a given that it will also be highly toxic. Both have an impact on the way it will need to be handled in the manufacturing facility.

The first step for any risk assessment is to gather known information about the safety properties of the molecule, including the OEL. A large amount of safety data will already have been compiled for any molecule that is heading into large scale manufacture. As well as pre-clinical toxicology and animal studies, this may also include insights gleaned from Phase I clinical trials. This is used to inform the OELs and Occupational Exposure Bands (OEBs) determined by risk assessors, and then to select the appropriate containment and personal protective equipment, as well as the engineering strategies that will be applied.

However, toxicity data from preclinical and clinical work do not directly translate into the process of determining an OEL: they are designed to discover at what level the drug should be dosed to humans, bearing in mind its therapeutic effect and the side-effects it may cause. There is a huge difference between orally imbibed or intravenous exposure, and the route by which operators might be exposed in the facility, which is far more likely to be inhalation or topical contamination.

It does, however, give insight to what effects might occur with both acute and chronic exposure. Acute problems might include respiratory or lachrymatory problems, whereas chronic effects could be that the compound is carcinogenic, mutagenic or a sensitizer on prolonged exposure.

The spectrum of effects

The dose–response curve gives an insight into effects. At the no observed effect level (NOEL), a compound has no biological effect. Next is the no adverse event level (NOAEL), which is often mentioned in risk assessments. Continuing up the curve, there is the low effect level (LOEL), and the low adverse effect level (LOAEL), then well-tolerated doses. After the maximum tolerated dose has been reached, any higher doses are likely to prove hazardous. In animal models, ED₅₀ represents the dose at which half the test subjects will experience an effect, whereas at TD₅₀ half will experience toxicity, and LD₅₀, half of test subjects will experience a lethal dose.

When calculating a molecule’s OEL, various uncertainty factors will be included to compensate for the fact that not everything is known about the compound. Components that can affect it include the duration of the study, inter-subject variation, the severity of the effect, and factors such as bioavailability, bioaccumulation and pharmacokinetics.

Some risk assessors will also include modifying factors. These might be the slope of the dose–response curve, the clinical significance of a critical effect and whether it is reversible, and whether this is relevant to operators within the plant.

With so many uncertainty and modifying factors that might be included, there is little wonder that there is so much variability from one risk assessor to the next — it is not unheard of for the OEL from one assessor being 100 million times greater than the OEL established by another.

Looking at the real world

Getting around this variability in risk assessments involves the application of real world context, notably a consideration of what level of exposure risk is acceptable for an operator. A starting point of one-in-a-thousand, is probably more true than one-in-a-million.

Taking a more realistic view can have a huge impact on cost, and this is why the use of categorisation of molecules into OEBs is more useful to inform containment requirements. A compound in OEB1, the lowest level, is non-toxic, and an OEL of 500 µg/m³ is appropriate. Moving up, an OEB2 compound will be more hazardous, and so the OEL has to be lower; and similarly, OEB3 chemicals have greater hazards. The highest banding, OEB4, represents those molecules where the hazards are extreme, and containment must be at the forefront of engineering design.

It is at the OEB3 level where significant savings can be made by taking a real-world view. A theoretical OEL is based on an eight-hour exposure, but if the operator could only ever be exposed for a few minutes, then containment and protection requirements are not so great. It may be that containment within a normal plant will already be acceptable; otherwise relatively low-cost options such as HEPA filters or additional soft-sided isolators could be sufficient. Surrogate testing should be used to prove that this approach will be successful.

By taking a realistic approach, it is clear that some compounds that might be considered highly potent do not, in reality, require extreme containment. Isolators should be reserved for those projects for which they are truly necessary. HPAPI capacity can easily be maximised and costs minimised by not using an isolator where it is not needed.