Venetoclax and hypomethylating agents in FLT3-mutated acute myeloid leukemia

1 INTRODUCTION

The FMS-like tyrosine kinase 3 (FLT3) receptor belongs to the receptor tyrosine kinase family, and plays a crucial role in regulating hematopoiesis.^{1, 2} FLT3 gene mutations are ones of the most prevalent genetic abnormalities in AML occurring in 25-35% of cases,^{1, 3, 4} and are enriched among cases of AML with normal cytogenetics and t(6;9),^{5, 6} FLT3 mutations commonly manifest as either internal tandem duplication (ITD) or tyrosine kinase domain (TKD) point mutation. FLT3 mutated (FLT3m) AML represents a high-risk leukemia associated with increased risk of relapse and decreased overall survival (OS).^{5, 7} This is indeed the case with patients harboring ITD mutations at a high allelic ratio, particularly in the absence of NPM1 mutation.^{4, 8-11}

The standard therapy for newly diagnosed FLT3m AML entails induction with intensive chemotherapy regimens in physically-fit patients, and nowadays, this is most commonly given in combination with the FLT3-inhibitor, midostaurin.⁴ Consolidation with allogeneic hematopoietic cell transplantation (HCT) is routinely recommended for patients with FLT3-ITDm AML whenever feasible. Other FLT3-inhibitors have been tested in the relapsed/refractory (r/r) setting in FLT3m AML, and have produced modest single-agent responses with short-lived remissions.^{12, 13} Therefore, the search for additional effective therapies is warranted in FLT3m AML, particularly for newly diagnosed frail patients as well as for relapsed/refractory (r/r) patients.

Venetoclax (VEN) is a selective BCL-2 inhibitor that has shown high response rate when combined with hypomethylating agents (HMAs) in treatment-naïve frail patients with AML with CR/CRi rate exceeding 70% and low treatment-related mortality.¹⁴ Encouraging response to VEN-HMA was also observed in r/r AML, and a subset of patients with advanced AML were able to proceed to allogenic HCT afterwards.¹⁵ The excellent activity of VEN-HMA was observed across various high-risk cytogenetics and mutations in AML including FLT3 mutations.^14-17 Nonetheless, data specifically focused on response rate for VEN-HMA among FLT3m AML, including the impact of AML setting (de novo vs r/r), the type of FLT3 mutation, the ITD allelic ratio and concurrent mutations are scarce. In the largest report of VEN-HMA treatment in de novo AML, DiNardo and colleagues reported a CR/CRi rate of 72% in newly diagnosed patients with FLT3m AML (n = 10 as ITD, 7 as TKD, and 1 as both ITD+ TKD).¹⁴ Interestingly, enrichment or gain of FLT3 mutated clone has been reported in cases of AML with primary and acquired resistance to venetoclax-based therapy.¹⁷

Considering the gap in knowledge on how factors related to FLT3 mutation affect response to VEN-HMA therapy, we sought to identify factors that potentially could predict response and durability of remission among this AML subset.

2 METHODS

We conducted a retrospective analysis that included AML patients who were treated at our institution with VEN-HMA between March 2016 and March 2020. We identified cases with FLT3 mutations utilizing a City of Hope next generation sequencing panel or a FLT3 PCR assay. Data on FLT3 type, allelic ratio, and associated somatic mutations were collected. The study was approved by the COH Institutional Review Board. Response was primary endpoint and defined as either complete remission (CR) or CR with incomplete count recovery (CRi). CR was defined as Bone marrow blasts <5%; absence of circulating blasts; absence of extramedullary disease; ANC ≥1000/mL; platelet count >100 000/mL. CRi was defined as all CR criteria except for residual neutropenia (<1000/mL) or thrombocytopenia (<100 000/mL).¹⁸ AML genetic risk-stratification was classified based on the 2017 European LeukemiaNet (ELN) criteria.¹⁸ Minimal residual disease was evaluated using multiparameter color flow cytometry done at a reference laboratory (University of Washington).¹⁹ Bone marrow restaging was performed 1-2 cycles after initial treatment and as clinically indicated thereafter. Secondary endpoint overall survival (OS) was defined as the time interval from start of VEN-HMA treatment to death or censored if alive at last follow-up. LFS among patients with CR/CRi was defined as the time interval from date of response to relapse or death, whichever occurred first, and was censored at last follow-up if still leukemia-free.

In univariate analysis, descriptive statistics were used to summarize the covariates such as demographics, clinical variables, or genetic risks (mutations) for both newly diagnosed and r/r cohorts and for response and non-response groups. The association between response to VEN-HMA or disease setting (newly vs R/R) and these covariates were assessed by the Pearson Chi-square or Fisher exact test for categorical variables (Tables 1-3). A logistic multivariable regression model was applied for covariates showing a significant association at 0.1 level in univariate analysis. Odds ratios and 95% CI before and after adjustment for other covariates are listed for each of the three covariates in Table 4. Associations between OS/LFS and each baseline covariate or genetic feature were assessed by Log-rank test. Median LFS and OS, hazard ratio and its 95% CI were summarized at each level of each covariate. All tests were two-sided and a P value of .05 or less was considered statistically significant. All analyses were performed by SAS 9.4 (SAS Institute, Cary, NC, USA) and R 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria).

3 RESULTS

We identified 50 patients with FLT3m AML treated with VEN-HMA. The median age was 66 (range; 18-82) years and 56% were male. The majority of patients had de novo AML (74%). Seventeen (34%) patients were treated with VEN-HMA as frontline therapy for newly diagnosed AML, while the majority (n = 33; 66%) were treated for r/r AML. Among the r/r AML cohort, the median number of prior lines of therapies was 2 (range:1-7), with 39% having received prior HMA treatment and 27% were recipients of allogeneic HCT (Table 1). Nineteen (58%) patients with r/r AML received prior FLT3-based tyrosine-kinase inhibitors (TKI) therapy either during induction or at the time of r/r disease before starting VEN-HMA (two patients received 2 different TKIs), including sorafenib (n = 5), midostaurin (n = 10), gilteritinib (n = 4), quizartinib (n = 1), and crenolanib (n = 1).

The AML stratification according to the 2017 ELN combined cytogenetic-molecular risk was favorable, intermediate, and high-risk in 10%, 32% and 58%, respectively, at the time of initiating VEN-HMA. FLT3 mutation was ITD in 70% (n = 35) of the cases and TKD in the rest. Within the FLT-ITD cohort, the allelic ratio was high (≥0.5) in 69% of cases with a median of 0.63 (range; 0.05-16.41) (Table 1). Additional mutations beside FLT3 were observed in the vast majority of cases (92%), and most common concurrent mutations were DNMT3A (28%), NPM1 (26%), RUNX1 (24%), WT1 (22%) and IDH (14%). All IDH1/2 (n = 7, P = .081) and WT1 (n = 11, P = .01) mutations were observed in r/r AML cases while all TET2 mutations (n = 5, P = .002) were observed in newly diagnosed cases (Table 2) Decitabine in combination with VEN was used in the majority of cases (92%), and the schedule was 10-day course in 59% for the first cycle.

Compared to newly diagnosed AML patients, r/r patients in this cohort were younger (median: 58 vs 75, P < .001) and had more intermediate- and less high-risk genetics (P = .02). No significant difference in the frequency of secondary AML (P = .69), FLT3 mutation type or allelic ratio (P = .11), and the type of the administered HMA or the duration of first cycle of decitabine (P = .73) between newly diagnosed and r/r patients (Table 1).

The CR/CRi rate for the whole cohort was 60% (n = 30), including 14 CR (28%) and 16 CRi (32%). The median number of cycles to achieve CR/CRi was 1.5 (1–4) cycles. There were additional 4 (8%) patients who achieved morphologic leukemia-free state (MLFS), all among cases with r/r AML. Twenty-five patients who achieved CR/CRi had MRD assessment using multiparameter flow cytometry, of which 18 (72%) attained MRD-negativity during VEN-HMA therapy. FLT-3 mutational burden was not routinely reassessed after treatment. The median numbers of administered VEN-HMA cycles for responders and non-responders were 3 (range; 1-35) and 2 (range; 1-12), respectively.

The 4- and 8-weeks treatment-related early mortality in the absence of disease progression was 0% and 2% (n = 1, newly diagnosed AML patient died at day 48 at outside hospital of unknown cause). During cycle 1 of the VEN-HMA, 16 (32%) patients developed neutropenic fever, 3 (6%) patients developed blood stream bacterial infections, 2 (4%) patients developed invasive fungal infection, 2 (4%) patients developed grade ≥ 3 bleeding (one was pre-existed condition), and no patients developed clinical TLS.

The CR/CRi rate was higher among newly diagnosed FLT3m AML patients compared to r/r patients (94% vs 42%, P = .0005) (Table 3). MRD-negativity was achieved in 82% and 64% among newly diagnosed (n = 11) and r/r (n = 14) patients, respectively, who achieved CR/CRi and had MRD assessment. Among r/r patients, CR/CRi was 37% in patients with prior exposure to FLT3-TKI and 50% for patients who were naïve for FLT3-based TKI therapy. Five patients (4 r/r and 1 newly diagnosed) had FLT3-based TKI combined with VEN-HMA (sorafenib in 4 and gilteritinib in 1). Among these 5 cases, 3 patients who achieved morphological CR/CRi with VEN-HMA, sorafenib was added in cycle 2 or 3 due to persistent MRD (2 subsequently achieved MRD- and 1 remained MRD+) while the other 2 had TKI added to cycle 1 of VEN-HMA and both did not respond (1 achieved MLFS).

Additionally, CR/CRi rate was higher in patients who were HMA-naïve (P = .02) and patients younger than ≤60 years old (P = .01), but no difference in response was observed according to patient sex (P = .49), whether AML was secondary or de novo (P = .24), the reception of prior allogeneic HCT (P = .72), the ELN cytogenetic-molecular risk (P = .16), the type of FLT3 mutation and allelic ratio for ITD cases (P = .52), and the type of HMA and the number of days of first cycle of decitabine (P = .66) (Table 3).

In multivariate analysis including age, prior HMA exposure and AML disease setting, only having r/r AML independently predicted lower CR/CRi in response to VEN-HMA (OR: 0.08; 95%CI: 0.00-0.60, P = .03) (Table 4). When response rate to VEN-HMA was analyzed separately for newly diagnosed and r/r FLT3m AML cohorts, none of the studied factors influenced response with the exception of trend toward lower response in secondary AML compared to de novo AML in r/r cases (11% vs 54%, P = .098) (Table S1).

Among concurrent mutations, IDH mutations resulted into lower CR/CRi (P = .01) while ASXL1 and TET2 mutations showed a non-significant association toward higher CR/CRi (P = .07, both). When adjusted to AML setting, none of IDH, ASXL1, and TET2 mutations independently predicted response (P = .10, .12, and .83, respectively) while AML setting remained an independent predictor for response (P = .002, .001, .003, respectively).

With a median follow up of 238 days, the median OS was 338 days. For patients who achieved CR/CRi or MLFS, the median LFS was 323 days. Univariate factors for LFS and OS are shown in Tables S2 and S3. No statistically significant relationships were found except an association of 10 days of decitabine during first cycle with inferior LFS (6.61; 95%:1.71-25.5, P = .01), although the 10-days cohort of responders was more enriched with r/r cases (63% vs 42%) and high-risk ELN risk (58% vs 33%) compared to 5-days course cohort.

Eleven (22%) patients (8 with r/r AML and 3 with de novo AML) underwent allogeneic HCT post VEN-HMA. Of these, 9 were responders, 1 had MLFS, and 1 patient had refractory disease and underwent transplant with active AML.

4 DISCUSSION

In this report focusing exclusively on FLT3m AML treated with VEN-HMA, we have shown encouraging results with a CR/CRi rate as high as 94% for older treatment-naïve patients (median age = 75 years). Consistent with previous reports, the early treatment-related mortality with VEN-HMA in this older population is extremely low.¹⁴ While response rate to intensive induction regimens is high in newly diagnosed FLT3m AML, a large proportion of these patients are ineligible for standard regimens either due to older age or pre-existing co-morbidities. Therefore, our results along with other published studies suggest VEN-HMA therapy as potentially the preferred choice to induce unfit patients with FLT3m AML considering its high response rate and the low toxicity.¹⁴

As expected, we observed a lower response to VEN-HMA in the r/r setting of FLT3m AML compared to newly diagnosed patients. Nonetheless, around half of these patients still achieved CR/CRi or MLFS with this treatment, and this is notwithstanding the fact that many of these patients had had prior exposure to FLT3-inhibitors (58%), HMA therapy (39%) or failed allogeneic HCT (27%). Moreover, responses to VEN-HMA was meaningful in r/r FLT3m AML setting as a subset of patients successfully underwent potentially curative allogeneic HCT after achieving remission. Similarly to de novo AML, early treatment-related mortality was low in r/r AML patients despite the fact that this cohort included older patients as well as patients who were previously treated with multiple lines of therapies and included recipients of allogeneic HCT.

While the type of FLT3 mutation, allelic ratio and concurrent mutations are known to influence outcomes of FLT3m AML treated with conventional chemotherapies,^8-11 we did not observe such associations between outcomes (remission rate, LFS or OS) and these FLT3 mutation-related factors in patients treated with VEN-HMA. This may relate to the fact that the mechanism of action of VEN is different from standard chemotherapy, and is mediated through inhibition of BCL2 proteins which are overexpressed in leukemia stem cells irrespective of their mutation profile.²⁰ Therefore, once our results are confirmed in larger study, the combination may represent a rational frontline choice for high-risk FLT3m cases, namely those harboring ITD mutation, with high allelic ratio and in the absence of NPM1 mutation, factors that predict adverse outcomes with conventional chemotherapy. Interestingly, we found that patients with concurrent IDH mutation were less responsive to VEN-HMA in presence of FLT3 mutations. This is unexpected given the recognized favorable correlation between IDH-mutations and response as well as the durability of remissions when treated with VEN-HMA.^{14, 17} Nonetheless, all cases with concurrent IDH-mutations in this cohort had r/r AML, which in of itself was an independent predictor for lower response in our analysis.

Our study is limited by the inherent flaws of such retrospective analyses that introduces selection bias as well as lack of uniformity regarding the treatment regimen as well as disease and patient related factors. A small number of patients had a FLT3 TKIs added to initial therapy (n = 2) or after response to the first cycle (n = 3). Although we report the largest experience focused on VEN-HMA outcome in FLT3m AML, we recognize that our study is relatively small to allow adequate power for sub-analysis for various FLT3 mutation related variables, and that it included cases of both treatment-naive and r/r FLT3m AML.

In conclusion, VEN-HMA therapy has marked activity in FLT3m AML regardless of mutation type, allele burden or concurrent mutations. VEN-HMA can serve as a promising backbone in FLT3m AML where it could be combined with FLT3-inhibitors in newly diagnosed and r/r setting to further improve outcome, and there are ongoing prospective studies investigating this concept (NCT04140487, NCT03625505, NCT03661307). It would appear reasonable to propose a randomized study comparing the activity of a second generation FLT3 inhibitor in combination with either VEN-HMA or standard 7 + 3 induction in newly diagnosed fit FLT3m AML patients given the high CR rate in treatment-naïve patients. As the toxicity profile is expected to be more favorable with VEN-HMA treatment, high-risk older patients with FLT3m AML may have a smoother transition to a curative allogeneic HCT consolidation with a more robust performance status and less co-morbidities.

AUTHOR CONTRIBUTIONS

IA, JZ, GM and VP designed research; all authors collected, assembled, analyzed, and interpreted data, JZ performed statistical analysis, IA and VP wrote the manuscript; and all authors approved the final version.

CONFLICT OF INTERESTS

I.A. has served on advisory boards with AbbVie, KiTE pharmaceuticals, Agios, consultant for Autolus and Amgen, speaker for Jazz Pharmaceuticals. A. Salhotra has served as consultant for Kadmon Corporation and has received research funding from Celgene. RN has served on advisory boards with Merck and Celgene and has research collaboration with Jazz Pharmaceuticals. A. Stein serves on the speaker bureau for Stemline, Amgen and Celgene, and on advisory boards for Stemline and Amgen. G.M. is a member of the speakers' bureau for AbbVie and Novartis, and serves on advisory board with Janssen Pharm. VP has served on the advisory boards for AbbVie and Jazz Pharmaceuticals and is member of speakers' bureau for Jazz, Amgen, Novartis and AbbVie. The remaining authors have no relevant conflict of interest to declare.