Relationship between CD33 Expression, P-Glycoprotein-Mediated Drug Efflux, and Clinical Outcome in Patients Treated in Phase II Trials with Gemtuzumab Ozogamicin Monotherapy.

Background: The anti-acute myeloid leukemia (AML) immunoconjugate, gemtuzumab ozogamicin (GO; Mylotarg™), contains a humanized anti-CD33 antibody (hP67.6) to facilitate uptake of the toxic calicheamicin-γ₁ derivative in CD33-positive AML cells. This putative mechanism implies a critical role for the intracellular accumulation of the toxic moiety for GO-induced cytotoxicity. Indeed, drug efflux by P-glycoprotein (Pgp) mediates resistance to GO and correlates with clinical outcome after GO monotherapy. Furthermore, recent in vitro data obtained in human myeloid cell lines have unequivocally demonstrated a quantitative relationship between CD33 expression and GO-induced cytotoxicity. In light of these findings, we have now re-examined the significance of CD33 expression levels on AML blasts and relationship with Pgp activity for clinical outcome of patients treated with GO monotherapy.

Methods: Pre-treatment bone marrow samples from patients enrolled in multicenter phase II protocols evaluating the safety and efficacy of GO monotherapy (generally 2 doses of 9 mg/m² 14 days apart) were used for analysis. Relative CD33 expression was quantified by flowcytometry immunophenotyping using the hP67.6 antibody, and linear fluorescence values used for calculations. Pgp function was cytofluorometrically determined by efflux of the fluorescent dye, DiOC₂. Results are presented as mean values and 95% confidence intervals. Unpaired t-tests, Pearson correlations, and logistic regression models were used for statistical analysis.

Results: Patients achieving a complete remission (CR) or CR with incomplete recovery of platelet counts (CRp) had statistically significantly higher mean CD33 expression levels (71.20 [57.20–85.19], n=69) compared to non-responders (54.44 [48.38–60.51], n=203; p=0.01). There was an inverse relationship between CD33 expression and Pgp efflux (r=−0.23) and this contributed to responders having a statistically significantly lower mean Pgp efflux (1.40 [1.28–1.52], n=57) compared to non-responders (1.83 [1.72–1.95], n=173; p<0.0001). The addition of Pgp statistically significantly improved a logistic regression model containing only CD33 (p<.0001), whereas the addition of CD33 did not lead to a statistically significantly improved logistic regression model containing only Pgp (p=0.14).

Conclusion: These data indicate that CD33 expression levels as well as Pgp efflux are associated with clinical outcome of patients treated with GO monotherapy, and that knowledge of Pgp provides important information regarding the probability of achieving a response, even after consideration of CD33 expression level. The inverse relationship between CD33 abundance and Pgp efflux is consistent with the notion of maturation-stage dependent expression of these proteins, and offers the rationale for the use of cell-differentiation-promoting agents, for example cytokines, in combination with GO to enhance GO-induced cytotoxicity and possibly improve clinical outcome of patients undergoing GO-containing AML therapy.