HFE Gene Mutation Status Predicts Response to Gemtuzumab Ozogamicin in AML

BACKROUND: The protein defective in type 1 hereditary hemochromatosis, called HFE, is similar to MHC class I-type proteins, associates with beta2-microglobulin and is implicated in membranous protein recycling. Gemtuzumab ozogamicin (GO), a monoclonal antibody directed against CD33 linked to a cytotoxic agent, has been used with controversial effects in acute myeloid leukemia (AML). Internalization of GO is required for its anti-leukemic effect. Therefore, we hypothesized that H63D or C282Y HFE gene mutations may impair GO activity by preventing its internalization.

METHODS: Wild type, C282Y and H63D HFE leukemic cells and primary cell were used to assess effects of GO alone or in combination with cytarabine on cell proliferation and apoptosis. Flow cytometry, confocal and AMNIS stream analysis were used to evaluate GO internalization. HFE mutations were analyzed by PCR analysis on DNA from patients included in clinical studies. Post-hoc subgroup analysis was performed to assess in vivo the role of HFE status on GO efficacy and toxicity among the 280 patients of the ALFA-0701 study as a study cohort (patients aged 50-70 years; GO 3mg/m² on days 1, 4, and 7 of chemotherapy and on day 1 of the first and second induction ; total dose 15 mg/m² ) and then on the GOELAMS-LAM 2006 IR study (patients aged 18-60 years; GO 6 mg/m² on day 4 of chemotherapy during the induction and the first consolidation; total dose 12 mg/m²) and UK NCRI AML17 study (patients aged 18-81 years; GO 3 vs 6 mg/m² on day 1 of chemotherapy but not during consolidation) as validating cohorts.

RESULTS: GO induced cell death by apoptosis in AML cell lines and primary cells in a dose-dependent manner and synergistically in combination with cytarabine. However, the IC5O of GO was significantly higher in HFE mutated cells (125 vs 10 ng/mL p<0.001). To further understand this phenomenon, the CD33 internalization was analyzed upon GO or anti-CD33 antibody exposure. In line with the cytotoxic effect, CD33 was significantly less internalized in HFE mutant cells (17.4% vs 65.1% after 1 hour, p<0.01).

HFE mutations were screened in 242 of the 280 ALFA-0701 patients with DNA available. There were 155 non-mutated patients (64%), 68 (28%) heterozygous for H63D, 11 (5%) heterozygous for C282Y, and 8 (3%) homozygous for H63D, which is consistent with the prevalence of the various mutations in the French population. Median age was 62 years (50-71) and the M/F ratio was 0.5, equally distributed among the different groups. No significant difference was observed with respect to the diagnosis of various hematological parameters, including white blood cell count, blasts number, cytogenetic subgroups, molecular mutation incidences (FLT3, NPM1, CEBPA, IDH, DNMT3A).

In the ALFA-0701 study cohort, HFE wild-type (WT) patients had a higher overall survival (OS) when treated in the GO arm (median, not reached vs 19.5 months, p=0.0193). In contrast, OS was similar among patients with heterozygotes HFE mutations treated in the GO and the control arm (median, 19.9 vs 21.9, p=0.9675).

In confirmatory cohorts, GO treatment led to a trend to increased OS in the GOELAMS-LAM 2006 IR cohort only in HFE WT patients (4-year OS with GO 62% vs 48%, p=0.08) but not in mutated patients (4-year OS with GO 73% vs 64%, p=0.45). Furthermore, in this cohort in the FLT3 WT patients subgroup, GO further improve OS in WT patients (4-year OS with GO 72% vs 50%, p=0.017), but not in patients with heterozygous HFE mutation (4-year OS with GO 80% vs 65%, p=0.23) In the UK NCRI AML17 cohort, which used GO only during induction, 245 patients were randomized between GO at 3mg/m² and 6mg/m² and evaluated for HFE status. Overall there was no effect of GO dose on outcomes, and no evidence of either any heterogeneity by HFE, nor any subgroup, which showed a differential effect of GO dose.

CONCLUSIONS: Future studies should focus on optimising the fractionated schedule for GO at the 3mg/m². Fractionated high doses (3x3mg/m2) during induction and single dose during consolidation seems to be the best schedule. Most importantly, the effect of GO treatment differed between HFE WT and heterozygote mutated AML patients. GO only increased OS among HFE WT patients. This is likely related to impaired internalization of the CD33 target. Our data suggest that HFE status should be used as a companion test to predict outcome of AML treated with GO.