P-Glycoprotein Overexpresion and Internal Tandem Duplications of FLT3 Are Characteristic of Discrete Populations of Elderly AML Patients.

Poor outcome in elderly patients with AML has been linked to an increased incidence of p-glycoprotein (pgp) expression, poor risk cytogenetics and secondary AML. Internal tandem duplication of the FLT3 gene is prognostic for poor outcome in younger patients with AML, but has not been analysed alongside other poor risk variables in the elderly. The ongoing AML14 trial for patients of 60 years or older is conducted by the UK National Cancer Research Institute Haematological Oncology Study Group, and is sponsored by the Leukaemia Research Fund. Patients could enter randomisations to intensive (daunorubicin- and ara-C-based) or non-intensive chemotherapy regimens. Using flow cytometry, we measured functional and phenotypic p-glycoprotein as well as MRP, LRP, bcl-2 and bax in CD45-gated blasts from 274 AML 14 patients. FLT3 ITDs were measured by PCR in the same cohort.

We found that pgp protein and function were associated with a lower presenting white blood count (median WBC 27.0 × 10⁹/l in non-functional cases, 6.0 × 10⁹/l in functional cases, p<0.0001; p=0.0002 for protein). Conversely, FLT3 ITDs were associated with a high presenting white count (wildtype median WBC 18.2 × 10⁹/l, ITD median 58.5 × 10⁹/l). Moreover FLT3 ITDs were strongly associated with the absence of functional p-glycoprotein (P=0.0005): 18% patients had FLT3 ITDs, 45% patients had functional pgp, but only 4% patients had both. There was also a suggestion that pgp is associated with poor risk cytogenetics (P=0.02 for pgp protein expression, P=0.01 for functional pgp).

Of the 274 patients analysed, 235 had intensive chemotherapy and 39 non-intensive treatment. Of the factors investigated, only pgp was associated with lower remission rates in univariate analysis adjusted for type of chemotherapy (functional pgp, p=0.01 for trend, protein p<0.0001), largely due to increases in the incidence of resistant disease (overall p=0.0008 for functional pgp, p<0.0001 for protein). In multivariate analysis, allowing for the effects of age, cytogenetics, WBC and performance status, only pgp protein proved significant in predicting CR or resistant disease.

For overall survival, none of the variables proved significant in univariate or multivariate analyses adjusted for part of trial (intensive or non-intensive). However, this finding should be construed as lack of evidence of effect, rather than lack of effect: the confidence intervals for the hazard ratio for pgp protein range from 0.92 to 1.62, so the results are still consistent with a moderate adverse effect of pgp protein. Among the cohort, 154 patients were randomised to receive the pgp inhibitor PSC-833 (n=62) or not (n=92). PSC-833 has been previously reported as conferring worse overall survival, mainly arising from early deaths. There is evidence that higher levels of pgp protein mitigate the adverse effects of PSC 833 (p=0.005 for heterogeneity) in remission rates. In conclusion, pgp over-expression was associated with a low white cell count and with increased levels of resistant disease in elderly patients with AML. We suggest that pgp positive (slow-growing) and growth-factor receptor mutated (fast-growing) AML clones form distinct subsets which may merit distinct therapeutic approaches in future trials.