A dose-response curve plots the fraction of cells surviving against drug concentration, usually on a logarithmic concentration axis. Two curves are needed to judge selectivity: one for the tumor cell line and one for a representative normal cell type, such as bone marrow progenitors. The horizontal separation between the two curves is the selectivity window. The therapeutic index condenses that separation into a number. If the dose that kills 50% of normal cells is written \(TD_{50}\) and the dose that kills 50% of tumor cells is written \(ED_{50}\), then \(TI = \frac{TD_{50}}{ED_{50}}\). A drug with \(TD_{50} = 40\) units and \(ED_{50} = 4\) units has \(TI = 10\), meaning normal cells need ten times the concentration to reach the same kill level. A second drug with \(TD_{50} = 12\) and \(ED_{50} = 4\) has \(TI = 3\). Both drugs kill tumor cells at the same concentration, but the first has a wider window and is therefore safer to push to an effective dose. The curves also reveal steepness: a steep normal-tissue curve means small dose increases cause large toxicity jumps, which narrows the usable range even when the ratio looks acceptable.
Selective Killing: How Chemotherapy Harms Cancer Cells More Than Normal Cells
The Selectivity Problem: Why Preferential Killing Is Possible but Imperfect
Reading Dose-Response Curves to Compare Two Drugs
2 / 2
Look at the two pairs of curves. In each pair, the left curve is tumor cells and the right curve is normal cells, and the horizontal gap between them is the selectivity window. Drug A has a wide gap: normal cells need about ten times the concentration to reach the same kill level, so its therapeutic index is ten. Drug B has a narrow gap, with a therapeutic index of about three. Both drugs kill tumor cells equally well at the same dose, but only Drug A lets you push the dose high enough to be effective without unacceptable normal-tissue damage. Notice also how steep the normal curve is in Drug B: a small dose increase causes a large jump in normal-cell kill, which shrinks the usable range even further.
0:00 / 0:00