E23285 Background: Centrosome amplification (CA) and aberrant mitotic figures are well recognized as beacons of malignancy. However, it has recently been brought to light that mitotic indices are much lower in patient tumors than in continuously cultured cells, which may explain the clinical failure of many mitosis-targeted drugs that exhibited preclinical efficacy. We investigated the concordance between clinical tumor samples and cultured cancer cells by evaluating mitotic activity and CA. Methods: Using indirect immunofluorescence confocal imaging, we compared the number of mitotic figures (phosphohistone H3 staining) and CA (γ-tubulin staining), in formalin-fixed paraffin-embedded clinical tumor samples of breast, pancreas and bladder (n = 20 for each cancer type), immortalized cancer cell lines (MDA-MB-231-breast, MIA PaCa-pancreas and T24-bladder), tumor xenograft (derived from MDA-MB-231) and patient-derived cell line (breast). Results: We found while < 5% cells were mitotically active in clinical tumors, 5-15% of cultured cells were mitotically active. By contrast, CA ranged from 35-50% in tumor samples and xenografts as opposed to only 5-15% in continuous cell cultures. Patient-derived tumor cells displayed the highest CA (~50%), consonant with human tumors, albeit at low passage numbers only. Mechanistically, we found that hypoxia can induce CA; thus intratumoral hypoxia may explain the discordance between the extent of CA in patient tumors and cultured cancer cells. These findings were validated in silico by probing the publicly-available Kao microarray (n = 327) dataset using Gene Set Enrichment Analysis. Results suggested that high hypoxic breast tumors are associated with (a) high expression of genes related to CA and chromosomal instability (CIN25 gene signature), and (b) worse distant metastasis-free survival. Conclusions: These results bring to light the previously unrecognized discordance in mitotic frequency and CA between various cancer model systems and highlight the importance of utilizing low-passage-number cell lines, preferably patient-derived, in the study of CA to more accurately represent cellular phenotypes in patient tumors.