gms | German Medical Science

Objective: In pharmacology, a drug candidate’s therapeutic potential is typically expressed as its ability to suppress cell growth. Different methods in assessing the cell phenotype and calculating the drug effect have been established. However, terminology is not standardised and inconsistencies in drug response outcome have been reported. It is still unclear whether and to what extent the choice of data post-processing methods is responsible for that.

Methods: We exposed a collection of nine in vitro models of glioblastoma to a library of 231 clinical drugs. Cell viability was modelled by a four-parameter Hill equation. The therapeutic potential of the drugs was determined using three established criteria: 1) the drug concentration needed for 50% growth inhibition (GI50), 2) the drug concentration for which the drug starts to show an effect, also known as point-of-departure (PoD), and 3) area under the curve (AUC). The curves were fitted to experimental data using least-squares on non-linearly transformed data, which allowed for reliable computation of confidence intervals (CIs). PoD was defined as the drug concentration at which the CI of the effect ceases to overlap with the CI when no drug was present.

Results: An effect was detected on 36% of drugs when relying on GI50 and on 27% when using PoD. For the area under the curve (AUC) a threshold of 9.5 or 10, depending on whether GI50 or PoD was used as the “ground truth”, could be set to discriminate between the drugs with and without an effect. GI50, PoD, and AUC were highly correlated (R² > 0.94, p < 10^-30 for all three). The exact ranking of substances by different criteria varied somewhat, but the group of the top 20 substances according to one criterion always included 17-19 top candidates according to another.

Conclusion: The three methods were largely in accord, at least for the top 20 ranked substances. The differences arise when a substance shows an effect at low concentrations, but plateaus afterwards, so that GI50 is never reached, or when the variance in the effect is so large at the beginning, so that PoD is never reached, despite achieving GI50. Regarding reported inconsistencies we hypothesise that unclear/inconsistent terminology plays a role. PoD can be defined in different ways and IC50, which describes the point half-way between the curve’s upper and lower plateaus, is occasionally used instead of GI50. It remains to be seen whether the observed effects can translate into clinical practice.

Figure 1 [Fig. 1], Figure 2 [Fig. 2]