The article ‘Have legislative changes curbed use of ’legal highs’?’ (The Pharmaceutical
Journal 2018;301:7920) concerning novel psychoactive substances (NPSs) highlighted the huge number of these drugs that are now available, with more appearing each week. One problem is that many of these NPSs have very limited or no data on which to base assessments of their toxicity.
I have recently found, using a quantitative toxicity–toxicity relationship (QTTR) approach, that the human fatal dose (HFD) of psychoactive drugs correlates very well with their rat and mouse intravenous LD50 (amount of an ingested substance that kills 50% of the test sample) values. For a diverse series of 18 psychoactive drugs taken from Gable
, the human–rat correlation had r2 = 0.823, s = 0.473, while the human–mouse correlation had r2 = 0.756, s = 0.541 (r2 is coefficient of determination; s is standard error of prediction (log units)). For three external test drugs, predicted HFD values were all within one standard deviation of the observed values.
Two software packages, TerraBase and ACD/Labs, predict rodent LD50 values, and these also correlate well with HFD values (human–rat r2 = 0.654, s = 0.644; human–mouse r2 = 0.678, s = 0.621). While these correlations are not as good as those using experimentally-measured rodent LD50 values, they nevertheless yielded predicted HFD values within one standard deviation of the observed values.
Hence, the use of measured or calculated rodent LD50 values can rapidly give an indication of the likely HFD of a new psychoactive drug, provided that it is of the same chemical and activity class as one or more of the 18 drugs used to develop the QTTR. A paper describing this work has been accepted for publication in the International Journal of Quantitative Structure-Property Relationships, and will be published in early 2019.
John Dearden, emeritus professor of medicinal chemistry, School of Pharmacy and Biomolecular Sciences, Liverpool John Moores University