P-glycoprotein (P-170) and CD34 expression in adult acute myeloid leukemia (AML)

We investigated the prognosis value of CD34 and P-170 expression in blast cells of adult patients affected by de novo acute myeloid leukemia (AML). CD34 antigen was analyzed by indirect immunofluorescence (IFI) and alkaline phosphatase-labeled streptavidin biotine (AP-LSAB) in 62 patients (median age: 51 years). P-170 expression was determined by AP-LSAB in 51 cases using JSB1 and C219 monoclonal antibodies. All patients were treated with conventional chemotherapy induction regimen. Follow-up was from 6 to 79 months. Complete remission (CR) rate was not statistically different between CD34+ and CD34- patients (67 vs. 84%, p = 0.2). The duration of CR and survival were not influenced by CD34 expression. Karyotype abnormalities were more frequent among MDR+ patients (65 vs. 21%, p < 0.01). CR rate was statistically lower in MDR+ patients as compared to MDR- patients (63 vs. 96%, p = 0.01). Median disease-free survival (DFS) was shorter for MDR+ patients but the difference was not significant (5 vs. 10 months, p = 0.09). Patients who were positive for both parameters CD34 and P-170, had a poor prognosis with a 50 vs. 100% CR rate for CD34/P-170 negative patients, (p = 0.002), a lower median DFS (3 vs. 12 months, p = 0.01) and overall survival (OS) (3 vs. 14.5 months, p = 0.01). Results of cytogenetic analysis did not influence CR rate but the relapse rate was higher, although not significant, for the patients with unfavorable karyotype (63 vs. 33%). The seven CD34+/MDR+ patients with poor prognosis karyotype had a statistically lower CR rate, median DFS and OS than the 7 CD34-/MDR- patients with normal or favorable karyotype (CR: 29% vs. 100%, p = 0.02), (DFS: 3 vs. > 12 months, p = 0.01), (OS: 4 vs. > 12 months, p = 0.02). Our data indicate that P-170 but not CD34 expression is predictive for a lower CR rate. The identification of a bad prognosis subgroup of CD34+/MDR+ AML patients (and especially those with poor prognosis karyotype) has to be confirmed on larger series using uniform methodology.