Gemtuzumab ozogamicin (GO) is a novel targeted therapy with promising clinical activity in acute myeloid leukaemia (AML). Consisting of a monoclonal CD33 antibody coupled to the toxin calicheamicin, GO enters the cell after binding to the CD33 antigen and releases the toxin intracellularly, resulting in site-specific DNA double-strand breaks (dsb). The more precise cellular and molecular mechanisms behind the clinical response, and resistance to GO remain however incompletely known. With the final aim to identify potential applicable biomarkers related to clinical outcome of GO treatment, we have investigated in vitro effects of GO and of calicheamicin alone, on primary cells isolated from AML patients and on AML cell lines. Cell incubations with GO or calicheamicin at clinically relevant concentrations (10 to1000 and 0.3 to 30 ng/ml, respectively) during 24 to 72 h revealed clear dose and time dependent effects by both drugs on CD33-positive primary cells isolated from an AML patient at diagnosis and on HL-60 cells. A maximum decrease in cell viability of 60% and 80%, respectively, as compared to untreated cells, was noted at 72 h at the highest drug concentrations used (MTT assay). In contrast, GO-induced toxicity was clearly lower in cells obtained from an AML patient in clinically resistant relapse and only minute, even at the highest drug concentrations, in KG1a AML cells. Expression of CD33 (as examined by FACS) appeared to influence the response to GO administered at low to intermediate, but not at higher doses. Thus, GO at 100 ng/ml did not affect viability of CD33-negative HL60 cells, while this dose induced a more than 60% survival reduction amongst CD33-positive HL60 cells. At a higher GO concentration (1000 ng/ml), however, viability was markedly decreased in both cell populations. In contrast, cytotoxicity elicited by calicheamicin appeared independent of CD33-expression at all concentrations tested. These findings support the present notion, that the clinical effect of GO administered to AML patients at sufficiently high doses is not clearly linked to the degree of CD33-positivity of the individual leukemic cell population. Since the active part of GO, calicheamicin, causes site specific DNA dsbs, which may induce apoptosis through the intrinsic apoptotic pathway, we examined activation of the mitochondrial pathway after GO adminstration. GO and calicheamicin were both found to cause mitochondrial depolarization followed by caspase-3 activation in a dose- and time-dependent way in GO-sensitive HL60 cells, and to some extent activation of caspase-3 in the GO-sensitive primary AML patient cells. In contrast, GO failed to induce any of these events in the KG1a cells. When exposed to another DNA damage-inducing cytotoxic substance, etoposide, also KG1a cells showed reduced viability and activation of caspase-3. DNA damage-induced apoptosis has been linked to a sustained activation of the stress-activated protein kinases (SAPKs) such as p38. We observed an increased level of phosphorylated p38 in GO-sensitive HL60 cells as well as in sensitive primary AML cells upon GO-treatment. In contrast, in GO-resistant KG1a cells this GO-induced activation of p38 was not observed, suggesting that activation of p38 is an essential part of GO response. In conclusion, our results from GO-treated primary AML cells and AML cell lines pinpoint the importance, not only of CD33-expression but also of SAPK activation with subsequent initiation of mitochondrial depolarization and caspase-3 activation, as molecular determinants for clinical GO responsiveness.