While previous studies demonstrated the importance of drug efflux by P-glycoprotein (Pgp) for resistance to gemtuzumab ozogamicin (GO; Mylotarg®), an anti-AML immunoconjugate between an anti-CD33 antibody (hP67.6) and a toxic calicheamicin-γ1 derivative, the role of CD33 expression for GO efficacy remained controversial. Using CFC assays performed on 77 primary AML specimens, we herein found that GO, but not unconjugated hP67.6 or a calicheamicin-γ1 immunoconjugate employing hCTM01 (an antibody targeting an endocytic epithelial tumor antigen [MUC1]), reduced CFC growth in a dose-dependent manner. The level of CD33 expression on AML blasts, as determined by flow cytometry, was associated with the extent of CFC reduction by GO at both 1 ng/mL (r=0.3467, n=46, p<0.05) and 10 ng/mL (r=0.4129, n=54; p<0.002); i.e., the higher CD33, the more pronounced the GO-induced CFC growth inhibition. Conversely, there was an inverse relationship between Pgp activity and the extent of CFC growth inhibition by GO (e.g., at 10 ng/mL GO: r=−0.3426, n=77; p<0.005). The Pgp inhibitor, cyclosporine A (CSA), significantly enhanced the effect of GO on CFC growth, with greater effects found in samples with higher Pgp activity (e.g., at 10 ng/mL GO: r=0.3375, n=39, p<0.05). Given the differentiation stage-dependent expression of Pgp and CD33, we next determined the ability of myeloid growth factors to modulate these factors and impact GO efficacy. CD33 expression was increased by 34±7%, 48±13%, and 111±18% in 8 primary AML samples treated with G-CSF, GM-CSF, or 4 growth factors [SCF/IL-3/G-CSF/GM-CSF] (all p<0.01), respectively, after 2 days of culture compared to cytokine-free controls; a similar effect of 4 growth factors was also found on FACS-isolated subpopulations of CD33neg/low cells. Conversely, growth factors decreased Pgp activity, as measured in a set of 9 samples, in which G-CSF/GM-CSF for 2 days decreased Pgp activity from 1.89±0.35 to 1.51±0.19 (p<0.01). Finally, we compared GO toxicity against CFC kept for 2 days in SCF/IL-3 alone vs. SCF/IL-3/G-CSF/GM-CSF. In 15 primary AML samples with adequate CFC growth, the addition of GM-CSF/G-CSF enhanced GO toxicity in 5/5 samples with limited GO susceptibility, as defined by <40% CFC reduction in medium containing SCF/IL-3 only (51±5% vs 14±13%, p=0.0079). This effect was only apparent in the absence of CSA, suggesting the importance of drug efflux for GO resistance in these cases and indicating that growth factors can overcome this resistance. In conclusion, these data support the importance not only of Pgp but also of quantitative CD33 expression for GO efficacy against primary AML cells. Importantly, priming with myeloid growth factors may modulate these factors and enhance GO efficacy against CFC with limited GO susceptibility, suggesting a potential role as chemosensitizer for GO-based anti-AML therapies.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author