One approach I have seen was to use a correlation between OEL and data available as the API is developed. In the order of least-to-most reliable reference point the OEL could be calculated from:

1.       NOAEL (Animal) (pre-Phase I preliminary tox data)

2.       NOAEL (Human) (Phase I data)

3.       LOAEL (Animal) (pre-Phase II 6-mo tox data)

4.       LOAEL (Human) (Phase I and/or II data)

5.       TDmin (Human) (Phase III data)

These values would be adjusted by appropriate factors (Body Weight, species-to-species Uncertainty Factors, adjustments for pharmacokinetics, Breathing Rate, bioavailability).

Not all data is available at all stages of development. There is nothing special about HP-APIs except that the exposure numbers get really small. Below 1 =ECg/m^3 all work needs to be done in isolators (lab scale) or in flexible containment (pilot plant/production scale).

As usually, engineering controls are established first to get the exposure below OEL and PPE is added on top to protect workers in case of a breach of containment.

Slawomir Janicki, Ph.D.