EGFR inhibitor data of lung cancer cells

Dear DepMap team,

I am analyzing the responses of lung cancer cells to EGFR small molecules. Specifically, many of these lung cancer cells appear to have AUC values close to 1 (for example, Gefitinib which was designed to treat lung cancer). According to a previous Q&A post Interpretation of Drug sensitivity AUC and Logfold change value, this means that most of these lung cancer cells are LESS sensitive to gefitinib. However, EGFR mutations cannot explain this discrepancy as most cells do not have them.

Similarly, most cells show log2FC of gefitinib (at the highest dose of 10uM) around 0.

Given the history of this compound, I found this observation counter-intuitive - either:
(A) I have misunderstood the definitions of AUC and log2FC of the PRISM repurposing secondary screen;
(B) The drug dose range used in this screen is well tolerated by this panel of lung cancer cells; or
(C) Other technical reasons?

Would be great if the team can help clarify my understanding of the compound screen data:

  1. Drug Sensitivity AUC: Values close to 1 mean cells are less sensitive to a compound?
  2. log2FC is a measure of cell viability between treated and controls?

Thanks much

Hi Jason,

Apologies for the very (!) late response. Yes your interpretation is correct for both LFC and AUC values. I am assuming the problem is likely due to other technical issues. Here are some potential culprits:

  • The dose or assay conditions may not be relevant for the expected response.
  • The compound batch used in PRISM Secondary set may not be high enough quality (stale).
  • Drug response data may have too much noise to let you detect the EGFR mutation association
  • EGFR mutation annotations may be imperfect.

I quickly checked for the other instances of gefitinib on the portal and saw GDSC2 data has strong correlation (-0.45) with EGFR expression (similarly PRISM AUC’s has -0.4):

Hope this helps.