IDH1 and IDH2-mutant AML
IDH1 and
IDH2 mutations are reported at a frequency of 7–14% and 8–19%, respectively, in AML [
16,
79]. Mutations in these genes typically occur in the conserved arginine residues (IDH1
R132, IDH2
R140, and IDH2
R172) of the catalytic domain of isocitrate dehydrogenase. The prognostic implications of
IDH1/IDH2 mutations are not entirely clear. Some reports have suggested that
IDH2 mutations are associated with better outcomes while
IDH1 mutations confer worse outcomes [
80,
81], though there is significant heterogeneity in the prognostic impact of co-mutations such as
NPM1 [
16,
81].
In a phase 1/2 study, patients with R/R AML harboring IDH2
R140 or IDH2
R172 mutations (mean age 67 years) were treated with ENA [
82,
83]. Patients with IDH2
R140 mutations had 2-HG reductions greater than 90% regardless of response, while 2-HG levels in those with IDH2
R172 mutations correlated with response (82.0% reduction from baseline if CR, 44.3% reduction from baseline if non-CR response, and 38.4% reduction from baseline if no response). CR/CRi/CR with incomplete platelet recovery (CRp) was observed in 29% of patients. Median OS for all patients was 8.8 months but extended to 22.9 months in patients achieving CR. While reduction in mutant
IDH2 VAF was not required for response, follow-up data demonstrated that clearance of
IDH2-mutated clones was associated with 100% CR [
83]. Furthermore, co-occurring mutations in
NRAS or MAPK pathway were suggested to contribute to treatment resistance [
84]. In a phase 3 randomized study from the BEAT AML Master trial, older patients with
IDH2-mutant R/R AML were randomized to ENA or conventional care [
85]; patients in the ENA arm had a doubling of event-free survival (EFS) and significant improvements in CR/CRi/CRp rates and hematologic response. The presence of
DNMT3A co-mutations has been shown to be associated with CR, and no deleterious effects of RAS signaling pathway co-mutations were observed; however, the presence of ≥ 4 co-mutations decreased the overall response rates (ORR) significantly (27.3% compared to 47.1% with < 4 co-mutations) [
86]. Thus, although ENA only has an FDA-approval label for R/R AML with
IDH2 mutations, the NCCN guidelines provide a recommendation to consider frontline ENA use for patients older than 60 years who are not candidates for intensive remission induction [
45].
In a phase 1 study, DiNardo et al. assessed the use of IVO in
IDH1-mutant R/R AML [
87]. The median age of patients was 67 years, and CR/CRp rates were 30.4% with a median DOR of 8.2 months. Median OS was 8.8 months with an 18-month OS rate of 50.1% in patients with CR/CRp. It was noted that patients with a lower co-mutational burden had improved CR/CRp rates, but no specific predictive co-mutations were identified. As a follow-up to this study, 34 patients with ND AML and
IDH1 mutations (median age 76.5 years) received IVO in the frontline setting [
88]. The composite CR (CRc) rate was 42.4% with over 60% of patients maintaining CRc at 1 year. Patients who received prior HMA therapy for antecedent hematologic disorder achieved CRc approximately half as frequently as those without prior HMA. Receptor tyrosine kinase (RTK) pathway mutations were observed in 36.8% of patients who did not achieve CRc compared to no patients who achieved CRc.
IDH1 mutant clone clearance was reported in 64.3% of patients who achieved CRc and was not observed in any patients without CRc. These two studies ultimately led to the approval of IVO in both R/R AML with
IDH1 mutations and ND AML with
IDH1 mutations in patients who are ineligible for standard chemotherapy.
On December 1, 2022, the FDA approved another IDH1 inhibitor, OLU [
89], for use in R/R AML based on a phase 1/2 trial of patients with
IDH1-mutant R/R AML who were naïve to IDH1 inhibitors [
90]. Patients with a median age of 71 years were treated with OLU until progression. Notably, CR/CRh rates were 35% and were achieved at a median of 1.9 months. Responses appeared durable with a median DOR of 25.9 months in patients who achieved CR/CRh. Prior VEN exposure did not appear to decrease response efficacy. Survival data will require further maturation. Another phase 1/2 study examining OLU with or without AZA in ND and R/R AML demonstrated similar CR/CRh rates in R/R AML patients treated with monotherapy while responses were surprisingly worse when combined with AZA [
91]. While the role of OLU is not clear considering experience with using IVO, one compelling scenario could be in the setting of IVO resistance through
IDH1 mutations, though this is solely based on preclinical data [
92] with no current data on response rates after prior IVO exposure.
Recent studies have evaluated the use of IVO or ENA in combination with other frontline therapies in ND AML. A phase 1 study assessed the use of IVO or ENA in combination with 7 + 3 (or bioequivalent dose of idarubicin) in patients with
IDH1/IDH2 mutations [
93]. Patients received IVO or ENA throughout induction, consolidation, and maintenance, though IVO or ENA were discontinued in patients who underwent alloHSCT. CR/CRi/CRp rates were 72% for IVO and 63% for ENA at the end of induction, which is slightly better compared to historical controls with
IDH1/2 mutations [
94]. In an updated analysis [
95], the authors reported CR/CRi/CRp rates of 78.3% in the IVO subgroup and 73.6% in the ENA subgroup. Responses for the sAML subgroup were improved if there was no prior HMA exposure, consistent with previously reported data [
87]. Co-mutations did not impact response rates in the IVO cohort, but in the ENA cohort, co-mutations with
ASXL1, NRAS, U2AF1, and
TP53 were associated with worse response rates, while
DNMT3A co-mutations were associated with marginally improved CR/CRi/CRp. When assayed by digital polymerase chain reaction (dPCR), 39% of patients treated with IVO cleared
IDH1-mutant clones and 23% of patients treated with ENA cleared
IDH2-mutant clones. Approximately half of the patients receiving IVO or ENA ultimately proceeded to alloHSCT. The use of IVO or ENA in combination with induction therapy was tolerable and ultimately did not significantly impact the time to recovery of the absolute neutrophil count or platelet count. An unanswered question that will require further investigation is the role of maintenance IVO or ENA after alloHSCT, particularly in patients who are unable to clear their mutant clone prior to transplantation. In patients eligible for intensive chemotherapy, the improved response rates compared to historical controls with the addition of IVO or ENA [
93,
95] provides a compelling argument for this practice. However, in the absence of randomized head-to-head comparison, this combination is neither FDA-approved nor recommended by the NCCN guidelines [
45].
The majority of
IDH1/2 mutations have been shown to be exquisitely sensitive to the combination of AZA/VEN with response rates similar to or higher than those achieved with standard induction. Therefore, it is reasonable to consider low-intensity therapy in this patient population without compromising outcomes. In a pooled analysis of from VIALE-A [
48] and a phase 1b HMA/VEN study [
70], the patients with
IDH1/2 mutations achieved CR/CRi rates of 79% with AZA/VEN compared with 11% with AZA alone [
96]. Median DOR was 29.5 months and 9.5 months, respectively, and median OS was 24.5 months and 6.2 months, respectively. CR/CRi and OS were relatively better with
IDH2 mutations compared to
IDH1.
The combination of AZA with IVO has recently emerged as an alternative non-intensive treatment option [
97]. The AGILE study was a phase 3 trial randomizing patients with ND
IDH1-mutant AML ineligible for intensive chemotherapy to receive AZA/IVO or AZA/placebo [
98]. At a median follow-up of 12.4 months, EFS was significantly longer in the AZA/IVO group compared to AZA/placebo with an estimated 12-month EFS of 37% and 12%, respectively. As a secondary endpoint, the median OS was 24 months for AZA/IVO and 7.9 months for AZA/placebo. CR/CRp rates were 53% with AZA/IVO compared to 18% with AZA/placebo, and DOR was longer for AZA/IVO compared to AZA/placebo (22.1 months versus 9.2 months). Patients with RTK pathway mutations (
FLT3, KIT, NRAS, KRAS, PTPN11) and
TP53 mutations were more likely to respond to AZA/IVO, and follow-up data suggest that relapse appears to preferentially occur with the acquisition of secondary high-risk mutations, independent of
IDH1 [
99]. The findings of this study led to the recent FDA approval of AZA/IVO for the frontline treatment of patients with ND
IDH1-mutant AML. While the rate of differentiation syndrome in these patients approaches 20%, the rate of cytopenias compared to AZA/VEN is significantly less. Therefore, when thinking about the various options for patients with
IDH1 mutations, toxicity, quality of life, and sequencing of treatment should be considered.
The combination of AZA/ENA has also been studied in patients with R/R [
100] and ND AML with
IDH2 mutations ineligible for intensive chemotherapy [
100,
101]. In the phase 2 analysis of patients with ND AML, patients were randomized to AZA/ENA or AZA monotherapy [
101]. The median age of patients was 75 years with CR/CRi/CRp rates of 63% with the AZA/ENA group compared to 30% with AZA alone; similar ORR were seen regardless of R140 or R172 mutations. Both ORR and CR were more durable with AZA/ENA compared to AZA (24.1 months versus 9.9 months and not reached versus 12.7 months, respectively). In a 2-year post hoc analysis, median EFS with AZA/ENA was 15.7 months compared to 11.9 months with AZA alone and OS was 22 months with AZA/ENA compared to 18.6 months with AZA alone; while the survival differences were not statistically significant, this study was not powered to detect significant differences in survival outcomes.
A major question remains as to whether therapies such as AZA/IVO or AZA/ENA would outperform AZA/VEN for ND
IDH1/2-mutant AML. The lack of survival advantage with AZA/ENA compared to AZA monotherapy [
101] would suggest that AZA/VEN is superior in
IDH2 mutations with the caveat that the median OS of patients with AZA in this study was significantly longer than reported for patients with
IDH2 mutations in the AZA/VEN studies [
96]. For patients with
IDH1 mutations, CR rates appear to be better with AZA/VEN [
96] than with AZA/IVO [
97], though median OS was essentially the same. At present, the widespread availability of AZA/VEN and its ability to bridge to alloHSCT favors its use in the frontline setting for patients with
IDH1/2 mutations who are ineligible for intensive chemotherapy, thereby preserving IVO, OLU, or ENA in the case of R/R disease. However, AZA/IVO could be considered in patients who are at high risk of complications with the myelosuppression of AZA/VEN. While ENA and IVO monotherapy are both NCCN-recommended options for frontline therapy [
45], only IVO is approved in this setting, and these should only be considered in patients with a very poor performance status. An area for future study will be triplet therapies. In an exploratory study of patients with treatment-naïve and R/R AML with
IDH1 mutations [
102], patients who received AZA/VEN/IVO had a CRc rate of 85–100% depending on dose intensity; however, this compares similarly to CRc rates of 67–100% with IVO/VEN alone. Nevertheless, the use of a triplet regimen improved MRD to 86% from 25% with doublet therapy. Studies combining different permutations of ENA, OLU, or IVO with HMA and/or VEN are currently underway (NCT04092179, NCT03471260, NCT04774393, NCT02719574) as are studies of these agents in the maintenance setting (NCT05010772, NCT03728335, NCT03564821, NCT03515512, NCT04522895).
FLT3-mutant AML
FLT3 encodes a type 3 RTK (FMS-like tyrosine kinase 3) and is widely expressed on AML blasts [
103]. Mutations in
FLT3 are seen in about 25–32% of cases of ND AML with 25% harboring ITDs and 7–10% harboring TKD mutations [
104,
105]. Previous reports prior to the era of
FLT3-targeting TKIs (henceforth referred to as FLT3i) have suggested that AR of
FLT3-ITD and the presence of
NPM1 co-mutations variably affect outcomes [
106‐
108]. Nevertheless, recent data suggest that relapse risk is higher in patients with
FLT3-ITD AML irrespective of AR or presence of
NPM1 mutation, and these patients should be considered for alloHSCT in first remission (CR1) if eligible [
109,
110]; this is reflected in the updated ELN recommendations [
24]. FLT3i can be divided into Type I and Type II inhibitors [
111], which are active against both ITD and TKD mutations or ITD only, respectively. The first approved FLT3i in AML was the Type I staurosporine-derived inhibitor MIDO [
112,
113] with early reports of its synergy with chemotherapy in patients with ND
FLT3-mutant AML [
114]. Newer and more selective FLT3i, GILT (Type I) and quizartinib (QUIZ) (Type II), have demonstrated promising responses in patients with R/R AML [
115‐
117].
The RATIFY trial [
118] was a randomized, placebo-controlled phase 3 trial investigating the addition of MIDO to standard induction chemotherapy and high-dose cytarabine (HiDAC) consolidation in adult patients under the age of 60 with ND AML and
FLT3 mutations (TKD or ITD). CR rates were similar between both groups, though median EFS and OS were significantly improved with MIDO (8.2 months versus 3 months and 74.7 months versus 25.6 months, respectively). OS was durable with 51.4% of patients in the MIDO group surviving at 4 years. More patients underwent alloHSCT in CR1 with MIDO compared to placebo (28% versus 23%, respectively); notably, follow-up studies after the publication of the RATIFY trial demonstrated deeper molecular remission with the addition of FLT3i to induction therapy [
119,
120], perhaps explaining in part the durable differences in OS despite similar CR and EFS rates.
Recent findings from a phase 2 study have also established the efficacy of adding to MIDO to induction, HiDAC consolidation, and maintenance in patients up to the age of 70 with ND
FLT3-mutant AML [
121,
122]. It should be noted that at present, MIDO is not currently approved as monotherapy and therefore is not recommended for post-consolidation maintenance given minimal benefit demonstrated after alloHSCT [
36,
118,
121‐
123]. Preliminary data from a phase 1 study (NCT02236013) evaluating GILT in combination with 7 + 3, consolidation, and maintenance in ND AML [
124] noted a median OS of 35.8 months with CRc achieved by 81.8% of all patients. AlloHSCT was performed in 30.4% of all patients. These data have led to ongoing clinical trials of GILT versus MIDO in addition to induction chemotherapy and consolidation (NCT04027309, NCT03836209).
GILT is the only FDA-approved FLT3i for use in R/R AML with
FLT3 mutations. The ADMIRAL study was a phase 3 randomized control trial of patients with R/R AML and
FLT3-mutations who received GILT or salvage chemotherapy [
117]. Similar rates of prior FLT3i exposure were noted in both arms, and approximately 20% of patients in either group had previously undergone alloHSCT. Median OS for patients receiving GILT was 9.3 months versus 5.6 months for those receiving salvage chemotherapy. [
125]. CRc rates were 54.3% with GILT and 24.8% with chemotherapy, and the median DOR was 11 months in the GILT group. Median OS for the
FLT3-ITD and
FLT3-TKD groups that received GILT were 9.3 months and 8 months, respectively. An important aspect of this study was the efficacy in both
FLT3-ITD and
FLT3-TKD populations, as the latter has been demonstrated to confer secondary resistance to type II FLT3i [
126].
Given the increasing use of low-intensity regimens in AML, pooled data from VIALE-A [
48] and a phase 1b HMA/VEN study [
70] showed that patients with
FLT3-ITD had a CR/CRi rate of 63% with AZA/VEN and a median OS of 9.9 months, while those with
FLT3-TKD had a CR/CRi rate of 77% and a median OS of 19.2 months [
127]. Of patients with
FLT3 mutations, approximately 36% had
NPM1 mutations in each of the AZA/VEN and AZA groups. Of those with concurrent
FLT3-ITD and
NPM1 mutation, AZA/VEN conferred a CR/CRi rate of 70% and a median OS of 9.1 months; patients with
FLT3-ITD and wild-type
NPM1 had a median OS of 10.6 months. This study has two important takeaways for older patients with mutated
FLT3. First, the rate of
FLT3 mutations in this population was lower, and the patients were older than typical
FLT3-driven AML seen in younger patients, possibly suggesting different disease kinetics and biology. Second, while CR/CRi rates were worse for patients with
FLT3-ITD mutations with wild-type
NPM1 compared to those with mutated
NPM1, the overall survival did not differ significantly, suggesting that
NPM1 status has an unclear prognostic value for patients treated with HMA/VEN.
Trials combining GILT [
128] or MIDO [
129] with HMA have not yielded encouraging results to date, though early data suggest that AZA/sorafenib (SORA) may be effective in patients with R/R AML and
FLT3-ITD [
130]. Data demonstrating the efficacy and tolerability of DAC/SORA in patients with R/R AML with
FLT3-ITD [
131] have led to NCCN recommendations for the use of AZA/SORA or DAC/SORA as low-intensity therapy in elderly patients with
FLT3-ITD AML or in R/R AML with
FLT3-ITD [
45], though it does not carry FDA approval for these indications. Despite the modest benefit with HMA, there seems to be synergy between FLT3i and VEN [
132‐
134]. A phase 1b study for VEN/GILT enrolled patients with
FLT3-wild-type or
FLT3-mutant (dose escalation) and
FLT3-mutant (dose expansion) R/R AML [
135]. The median age of patients was 63 years, 31% of whom had received prior alloHSCT and 16% of whom received prior VEN. No patients had previously received GILT, though 64% of patients with
FLT3 mutations had received other prior FLT3i. Patients with
FLT3-ITD had CR/CRi/CRp rates of 43%, while those with
FLT3-TKD has rates of 33%, and response rates were slightly better in those who were FLT3i-naïve. Median OS was 10 months for all
FLT3-mutated patients, though there was a significant improvement in those who had undergone alloHSCT after VEN/GILT (not reached) compared to those who did not receive alloHSCT (6.3 months).
Furthermore, a phase 2 trial evaluated the use of triplet therapy (DAC/VEN/FLT3i) in older patients with ND
FLT3-mutant AML and all adult patients with R/R
FLT3-mutant AML. In ND AML, the CRc rate was 92% with high rates of 91% MRD negativity in responders by PCR. In patients with R/R AML, CRc rates were 63% with MRD negativity by PCR in all patients who responded. At a median follow-up of 14.5 months, the median OS was not reached in ND patients (2-year OS estimated at 80%); the median OS in R/R patients was 6.8 months. Approximately one-third of patients underwent alloHSCT in either group. These results compare favorably to other reports of FLT3i/HMA in the ND setting [
128,
131,
136], though CRc rates appear to be higher with VEN/GILT in patients with R/R AML [
135].
Although not approved, two other FLT3i deserve mention given recent reports of their efficacy in AML. QUIZ is a second-generation type I FLT3i that can achieve significant marrow remissions in R/R
FLT3-mutant AML [
116,
137‐
139], though survival advantage was minimal compared to salvage chemotherapy in the phase 3 QuANTUM-R study [
138]. Due to these underwhelming results and concerns about cardiotoxicity and increased myelosuppression compared to other FLT3i, QUIZ has not been approved in the USA or Europe, though it is approved for use in Japan. In the frontline setting, the phase 3 QuANTUM-FIRST (NCT02668653) trial [
140] enrolled patients up to age 75 with ND AML and
FLT3-ITD and randomized them to QUIZ or placebo in addition to induction therapy with 7 + 3. Patients who achieved CR/CRi received up to 4 cycles of HiDAC with QUIZ or placebo and/or alloHSCT followed by up to 3 years of maintenance therapy with QUIZ or placebo. CR/CRi rates were 71.6% and 64.9% in the QUIZ and placebo arms, respectively, with DOR of 38.6 months and 12.4 months, respectively. Median OS and RFS were 31.9 months versus 15.1 months and 39.3 months versus 13.6 months in the QUIZ and placebo arms, respectively. AlloHSCT was performed in CR1 at similar rates between both arms; when censored for alloHSCT, OS trended toward a benefit with QUIZ over placebo. Moreover, an updated report from the study found that QUIZ conferred a deeper molecular remission compared to the placebo arm, perhaps underscoring the durability of benefit [
141]. Although RATIFY [
118] had already demonstrated a benefit to the addition of MIDO to induction chemotherapy, the QuANTUM-FIRST study is unique in that it evaluates the addition of an FLT3i for the higher-risk
FLT3-ITD mutation.
Lastly, a type I FLT3i emerging in clinical discussion is crenolanib (CREN). Long-term data were recently reported regarding the use of CREN in combination with 7 + 3 in adult patients with
FLT-mutant ND AML [
142]. CREN maintenance was offered up to 1 year after HiDAC or alloHSCT. The median age of patients enrolled was 57 years, 34% of which were over the age of 60 years.
FLT3 mutations were 75% ITD, 18% TKD, and 7% both ITD and TKD. CR/CRi rates above 80% were reported across several subgroups including those with
FLT3-ITD mutations or concomitant
FLT3/DNMT3A/NPM1 mutations. MRD-negative CR/CRi was achieved in 94% of evaluable patients, and 50% of patients underwent alloHSCT. Median OS has not been reached at a median follow-up of 45 months. Furthermore, translational studies found that no
FLT3 mutant clones were found at relapse in patients who completed protocol therapy.
In considering the role of FLT3i in ND
FLT3-mutant AML, intensive induction chemotherapy plus MIDO remains a standard of care for eligible patients. However, the formal release of data from QuANTUM-FIRST is awaiting, and ongoing trials will assess other frontline combinations with QUIZ (NCT04209725, NCT04047641), CREN (NCT03258931), and GILT (NCT04027309, NCT03836209), including head-to-head comparisons against MIDO. If QUIZ is approved for ND AML, it should be emphasized that its use would be limited to patients with
FLT3-ITD, while those with TKD mutations should still receive MIDO. For patients who are ineligible for intensive chemotherapy, AZA/VEN is effective for those with
FLT3-TKD mutations; unfortunately, better frontline options for those with
FLT3-ITD are currently limited. Nonetheless, GILT remains a very active FLT3i in the relapsed setting, and early data from doublet and triplet FLT3i combinations are encouraging in ND and R/R AML [
143]. Several trials exploring triplet combinations with DAC/VEN/QUIZ (NCT03661307) and AZA/VEN/GILT (NCT04140487) are currently enrolling with results highly anticipated.