Distribution of the relative FLT3/TKD mutant level in 115 cases. 

Context in source publication

Context 1

... and therapeutic advances in recent years. 1 A classic example is Fms-like tyrosine kinase 3 (FLT3), a class III TK receptor expressed on the surface of normal hematopoietic progenitors and blast cells from most cases of acute myeloid leukemia (AML), 2 which is constitutively activated in many cases of AML due to the presence of internal tandem duplications (ITDs) within the juxtamembrane (JM) domain. 3-5 FLT3/ITDs occur in approximately 25% of cases of AML in younger adults, and in nonacute promyelocytic leukemia (non-APL) AML they predict for increased relapse rate from complete remission (CR) and reduced survival. 3-7 Subsequent to the identification of FLT3/ITDs in AML, it was noted that primary AML blasts from a number of patients had constitutively phosphorylated FLT3 in the absence of an FLT3/ ITD, 8 raising the possibility of other mechanisms of aberrant FLT3 activation. Shortly afterward, 2 groups independently reported activating mutations affecting codons 835 and 836 in the FLT3 second TK domain (TKD). 9,10 Presence of these mutations has been confirmed by others, and the overall incidence in AML was reported to be approximately 7%. 11-15 A number of different mutations have been reported. Most are point mutations such as substitution of asparate residue 835 with a tyrosine (D835Y, the most frequent mutation), histidine, valine, or glutamate; other alterations include small deletions ( ⌬ 835, ⌬ 836) and insertions. All mutations initially described disrupted an Eco RV restriction enzyme cutting site that facilitated a simple screening technique using digestion of polymerase chain reaction (PCR) products. Subse- quently, however, a number of other point mutations have been reported at codons 839, 841, and 842 16-18 as well as a 2 amino acid insertion between codons 840 and 841, 19 all of which occur outside of the Eco RV restriction digest site. The clinical significance of FLT3/TKD mutations in AML is unclear, and different groups have reported either no significant impact of an FLT3/TKD mutation on overall survival (OS) 9,12,13,15 or an adverse prognostic impact. 11 The differences in outcome in these studies may partly be due to the low incidence of the mutations, such that relatively small numbers of mutant-positive patients have been investi- gated, as well as heterogeneity in both treatment protocols and characteristics of the patients included. A recent meta-analysis of the above studies has suggested that the outcome of FLT3/TKD mutations may be similar to that of FLT3/ITDs. 20 This has important implications for the application of risk-adapted treatment in these patients, particularly in view of the debate regarding optimal therapy for FLT3/ITD ϩ patients 6,21 and the current interest in molecularly targeted therapy with FLT3 inhibitors. 22 To address this issue, we have screened a large cohort of young adult non-APL AML patients with known FLT3/ITD status, treated in the United Kingdom Medical Research Council (MRC) AML10 and AML12 trials, for FLT3/TKD mutations using denaturing high-performance liquid chromatography (dHPLC). We have examined the impact of mutations on clinical outcome and compared this with the impact of FLT3/ITDs. In the total cohort of 1107 patients, 127 (11%) were FLT3/TKD mutant-positive (FLT3/TKD ϩ ) by WAVE screening, of which 122 had a pattern consistent with a single mutant and 5 had evidence of 2 mutants. Mutations were initially determined by direct sequencing; however, it soon became apparent that the WAVE chromatogram pattern was specific to the underlying mutation and, therefore, most were identified by WAVE pattern and confirmed by mutation-specific restriction enzyme digest (Table 2; Figure 1). All mutants were identified and confirmed except those in 7 patients in whom the heteroduplex peak was very small, suggesting that the mutant present was only a small percentage of the total FLT3 alleles (Figure 1A). Overall, 13 different mutations were detected (Table 3). D835Y was the most common, accounting for 50% of mutant-positive patients; D835H, D835V, D835E, and ⌬ 836 were all found at a similar frequency of 6% to 12% of positive patients. Two novel mutations were detected, each in a single case: (1) a 3 bp insertion that replaced aspartate codon 835 with glycine and proline and (2) a 16 bp tandem duplication plus 10 bp deletion within the kinase domain that replaced codons 835 and 836 (DI) with VIPT. Five of the mutations detected (4%) occurred outside the Eco RV restriction digest site. The relative mutant level as a percentage of total FLT3 alleles was quantified in 115 samples using labeled PCR products and size separation after mutation-specific or Eco RV restriction enzyme digestion if required. Quantification was not performed in 12 cases—either because the mutation had not been identified (7 cases) or due to lack of material—and these were scored by visual estimation of the chromatogram. The distribution of the relative mutant level is shown in Figure 2. Mutants were arbitrarily classified as “higher” level, with greater than the median level of 25%, or “lower” level when less than or equal to this. Of the total cohort of 127 mutant-positive cases, a higher level of mutant was present in 62 (49%) (ie, 6% of all patients in the cohort), and the median mutant level in this group was 43% (range, 26% to 93%). Of note, only 2 patients showed evidence of biallelic mutation or loss of WT alleles, with a 88% and 93% mutant level, respectively. A lower level of mutant was present in 65 cases (51%) (ie, 6% of all patients), with a median level of 6% (range, 1% to 25%). In the 5 cases with 2 mutants, 4 of them had 1 major mutant of higher level plus a minor mutant of very low level (approximately 2% of total), and in the remaining case both mutants were lower level (less than 25% of total). Eco RV digestion and dHPLC analysis were both performed in 247 patients, including all 127 with an abnormal WAVE pattern: 120 were WT by both techniques, but 13 (10%) that were mutant by dHPLC were WT by Eco RV digestion. Four of these 13 cases had a mutant of higher level, but the mutations were outside of the Eco RV cutting site. Nine cases had a lower-level mutant by dHPLC. Of these, 5 were identified, all of which disrupted the Eco RV cutting site; 3 were quantified and found to be 3%, 3%, and 1% of total FLT3 alleles, respectively. Consequently, approximately one third of the mutants that were missed by Eco RV digestion were because the mutation did not change this cutting site, and the rest were below the detection level of the Eco RV technique. Details of the presenting features of the 1107 patients in the cohort are given in Table 1. The presence of a mutation was not related to sex or age. Mutations were found in all French-American-British (FAB) subtypes except M7. There was some evidence that FLT3/ TKD mutations were less frequent in patients with secondary AML compared with de novo disease ( P ϭ .02). The presence of an FLT3/TKD mutation correlated with a high presenting white blood cell count (WBC): median WBC, 38.3 ϫ 10 9 /L in mutant-positive patients and 23.0 ϫ 10 9 /L in WT patients ( P ϭ .006). This correlation was restricted to higher-level TKD mutants: median WBC, 49.0 ϫ 10 9 /L in higher-level mutants and 24.4 ϫ 10 9 /L in lower- level mutants ( P ϭ .002). Information on karyotype was available in 882 patients (Table 4). There was a high incidence of FLT3/TKD mutations in patients with inv(16) (24%, P ϭ .005) and a low incidence in cases with adverse cytogenetics (3%, P ϭ .008). There were a number of features associated with FLT3/TKD ϩ cases that differed markedly from those of FLT3/ITD ϩ cases. FLT3/TKD mutations were uncommon in patients with secondary AML, whereas in our cohort the incidence of FLT3/ITDs did not differ significantly between de novo and secondary AML, 23 although others have reported that FLT3/ITDs are less frequent in secondary AML. 11,12,26 Unlike FLT3/ITDs, which are particularly frequent in patients with normal cytogenetics, the only group in which FLT3/TKD mutations were more common were cases with inv(16). Of the 1107 non-APL AML patients, 723 (65%) had WT FLT3, 257 (23%) were FLT3/ITD ϩ only, 100 (9%) FLT3/TKD ϩ only, and 27 (2%) had both mutations. Outcome data were available on the whole cohort with a median follow-up of 7.9 years (range, 6 to 193 months). Of the 127 FLT3/TKD ϩ patients, 14 received an autograft and 17 an allograft. The CR rate for all 1107 patients was 85%, and the presence of an FLT3/TKD mutation did not influence the rate of CR, ID, or RD. In univariate analysis, patients with an FLT3/TKD mutation had a reduced CIR, improved RFS, and improved OS at 5 years (Table 5). Outcome in a significant proportion of the FLT3/TKD mutant- negative cases would be influenced by the presence of an FLT3/ ITD; therefore, the impact in the cohort of both types of mutation was examined. The CIR at 10 years was 68% for FLT3/ITD TKD patients, 62% for FLT3/ITD ϩ TKD Ϫ patients, 46% for FLT3/ ITD Ϫ TKD Ϫ patients, and 37% for FLT3/TKD ϩ ITD Ϫ patients (Figure 3A). OS at 10 years was 37%, 24%, 36%, and 51%, respectively (Figure 3B). A direct comparison of FLT3/TKD ϩ and FLT3/ITD ϩ cases demonstrated a highly significant difference in CIR (OR, 0.45; 95% confidence interval [CI], 0.33 to 0.61; P Ͻ .001) and OS (OR, 0.53; 95% CI, 0.41 to 0.69; P Ͻ .001). There was no evidence that the impact of an FLT3/TKD mutation on OS differed according to FLT3/ITD status (test for interaction, P ϭ .8), although there were only 27 patients with both types of mutation (Figure 3C). Kaplan-Meier curves for OS in the 3 cytogenetic risk groups stratified by FLT3/TKD status are shown in Figure 4A. When the cohort was stratified according to cytogenetic risk group, there was no heterogeneity or trend between the groups in the effect of an FLT3/TKD mutation on OS ( P ϭ .8, Figure 4B). Similarly, stratification according to specific cytogenetic ...