The transient build-up of DNA supercoiling during the translocation of replication forks threatens genome stability and is controlled by DNA topoisomerases (TOPs). This crucial process has been exploited with TOP poisons for cancer chemotherapy. However, pinpointing cellular determinants of the best clinical response to TOP poisons still remains enigmatic. Here, we present an integrated approach and demonstrate that endogenous and exogenous expression of the oncofetal high-mobility group AT-hook 2 (HMGA2) protein exhibited broad protection against the formation of hydroxyurea-induced DNA breaks in various cancer cells, thus corroborating our previously proposed model in which HMGA2 functions as a replication fork chaperone that forms a protective DNA scaffold at or close to stalled replication forks. We now further demonstrate that high levels of HMGA2 also protected cancer cells against DNA breaks triggered by the clinically important TOP1 poison irinotecan. This protection is most likely due to the recently identified DNA supercoil constraining function of HMGA2 in combination with exclusion of TOP1 from binding to supercoiled substrate DNA. In contrast, low to moderate HMGA2 protein levels surprisingly potentiated the formation of irinotecan-induced genotoxic covalent TOP1-DNA cleavage complexes. Our data from cell-based and several in vitro assays indicate that, mechanistically, this potentiating role involves enhanced drug-target interactions mediated by HMGA2 in ternary complexes with supercoiled DNA. Subtelomeric regions were found to be extraordinarily vulnerable to these genotoxic challenges induced by TOP1 poisoning, pointing at strong DNA topological barriers located at human telomeres. These findings were corroborated by an increased irinotecan sensitivity of patient-derived xenografts of colorectal cancers exhibiting low to moderate HMGA2 levels. Collectively, we uncovered a therapeutically important control mechanism of transient changes in chromosomal DNA topology that ultimately leads to enhanced human subtelomere stability.