NUP98 rearranged acute myeloid leukemia: An initial experience

INTRODUCTION

Acute myeloid leukemia (AML) is an oligoclonal, molecularly heterogeneous disease with several recurrent genetic abnormalities that form the basis of therapeutic decision-making according to recent mandates and guidelines. The European Leukemia NET (ELN) risk stratification in 2022^[1] has added a few more of these into the risk stratification to aid the treating clinicians in adequate triaging for transplants. One such is Nucleoporin 98 (NUP98) rearranged leukemia, which has been categorized as an adverse risk as per the ELN 2022.^[1]

NUP98 is a nucleoporin protein encoded by the NUP98 gene, located on chromosome 11p15.^[2] It is a component of the nucleopore complex, and its rearrangement with other partner genes has been implicated in myeloid malignancies.^[3] The first fusion to be identified was NUP98/HOXA9, the discovery of which dates back to 1996.^[4] This was discovered in AML belonging to French American British (FAB) classifications M2 and M4.^[4] NUP98 rearrangement has been reported in 4–10% cases of pediatric cases and in ~2–4% cases of adult AML.^[5] Over the past two decades, close to 40 different partners have been identified in AML cases, as well as in therapy-related myelodysplasia.^[6] The fusion transcript formed comprises the N terminal of NUP98 and the C-terminal of the fusion partner, the most common being NSD1 in ~68% of cases, followed by Lysine Demethylase 5A (KDM5A) in 20% of cases.^[7] These fusion partners consist mainly of homeodomain proteins that have been shown to play a role in transcriptional/epigenetic regulation.^[7]

NUP98 AML cases have distinct clinical profiles and pathological features, including morphology and immunophenotype.^[8] In the TCGA dataset of AML, irrespective of age, structural variants involving fusions of NUP98 have been reported in five cases of AML only.^[9] Other fusions have been reported in solid organ malignancies, including breast, uterine, lung, bladder, and prostate cancers.^[10] Other collaborative groups have reported real-world outcomes of this subtype of AML, and all studies conclusively showed an inferior outcome in this subgroup, irrespective of age.^[6,9,11-18]

These rearrangements are detected in next-generation sequencing (NGS)-based assays, typically RNA sequencing, and are not part of the initial AML multiplex polymerase chain reaction (PCR) panels. They are cryptic translocations that can also be detected on karyotyping; however, false-negative results may occur. Hence, NGS is the mainstay of detection, underscoring the need to profile all AML cases at baseline using extended panel-based testing.^[19]

The reported outcome across all age groups is poor, with a complete remission rate of 50% after one course of induction and an overall survival (OS) rate of only 25–35%. The rate of relapse rate is as high as 64–68%. Allogeneic stem cell transplantation is the mainstay treatment; however, it has not shown remarkable outcomes.^[6]

In the current precision era, some studies have shown the activity of dasatinib in cases with moderate responses, and some other molecules are under development.^[20] However, this remains elusive in the Indian context. Dasatinib has been shown to act synergistically with bcl2 inhibitors navitoclax and venetoclax along with targeted FLT3 inhibition (wherever indicated).^[19] The obvious attributable causes are economic and social restraints; however, the main cause is under-detection, as the current AML diagnostic battery includes limited PCR and karyotype, and many patients are not profiled on NGS.

This is a single-center experience of NUP98 rearranged AML, depicting the clinical, pathological, and genomics of this subgroup, along with clinical outcomes based on the therapy offered.

METHODS

RESULTS

A total of 16 patients with NUP98 rearranged AML were included in the study, of which 14 were adults and two were pediatric. The median age of the adult AML patients was 27 years (range: 21–34 years). There was a clear male predilection, with a male-to-female ratio of 3.2:1. All patients presented with hyperleukocytosis with a mean white blood cells count of 110,000 cells/cu mm (range: 88,000– 210,000 cells/cu mm). The median bone marrow blast percentage was 70% (44–90%).

According to the FAB classification, the majority of cases were M5 (7/16, 44%), followed by M1 and M4 in 5 (31%) and 4 (25%) cases, respectively. All cases were classified as adverse based on the 2022 ELN Risk Stratification. No specific pattern was observed in the flow cytometric immunophenotype, suggesting NUP98 rearrangement. Karyotyping was requested in only five of the 16 patients, and all these patients had a normal karyotype.

Among the 16 patients, all had NSD1 as their partner. The fusion breakpoints were not heterogeneous and the same fusion transcript was identified in all 16 patients. Twelve (75%) patients harbored concurrent FLT3-ITD mutations and 56% had a pathogenic missense or truncating variant in the WT1 gene. Other common concomitant genomic alterations included TET2 (25%), NRAS (19%), NF1 (13%), DNMT3A (13%), and RAD21 (6%) in descending order. The co-mutational landscape is depicted in Figures 1 and 2, which depict the OncoPrint of the cohort as well as the strength of association of mutations with NUP98 in a circos plot.

Close

Figure 1:

Oncoprint showing the landscape of co-occurring genomic alterations with Nuceoporin98 rearrangement in the study cohort. Each grey box represents an individual patient.

Export to PPT

Close

Figure 2:

Circos plot showing strength association of co-mutations with Nuceoporin98.

Export to PPT

The median time to detection of NUP98 rearrangement was 3.3 days and hence, induction was based on genomic findings. Both pediatric cases were treated with azacitidine, venetoclax, and FLT3 directed therapy (Azacytidine-7 days and venetoclax-14 days, FLT3:21 days) for induction, and among the adult patients, two patients were treated with fludarabine, arabinofuranosyl cytidine, granulocyte colony-stimulating factor (FLAG) + venetoclax (venetoclax for 14 days), with the addition of midostaurin on day 8. Nine patients received azacytidine + venetoclax (14 days) + midostaurin (and two were treated with intensive induction chemotherapy. One adult patient was treated with decitabine (10 days), gilteritinib, venetoclax (14 days), and dasatinib (21 days). The use of intensive chemotherapy was restricted due to the performance status in these patients and the presence of concomitant pneumonia in five of the nine patients who received azacytidine and venetoclax therapy. The doses and dosages are as follows: Venetoclax (orally, once daily, for 28 days, target dose 400 mg), decitabine (20 mg/m² intravenously on days 1–5), and azacytidine (75 mg/m², subcutaneous).

None of the patients achieved CR. Molecular measurable residual disease (MRD) was performed in five patients and was positive (>0.1%) in three patients and low in two patients (<0.1%). The patient who was administered decitabine, gilteritinib, venetoclax, and dasatinib showed marrow in remission on day 14 of therapy. She was awaiting haploidentical transplantation. None of the other patients were subjected to hematopoietic stem cell transplantation because CR was not achieved during induction.

DISCUSSION

This is a single-center experience of NUP98 rearranged AML in India, possibly the largest reported so far from the peninsula.

NUP98 rearranged AML has been described to follow an aggressive and tumultuous disease course with non-achievement of CR, even with intensive chemotherapy. The prevalence of NUP98 has been reported in real-world studies in 4–10% of cases of childhood AML and in 2–3% cases of adults.^[5] More than 40 fusion partners^[6] are known, with NSD1 being the most common in both adult and pediatric cases, demonstrating a male predilection and normal karyotype.^[22] All of these reported findings also concord with the current study and depict no geographic or ethnic heterogeneity. However, the sample size may be limited in the present study to draw definitive conclusions with respect to population-based differences.

Pathologically, these cases are expected to present with high leukocyte and blast counts, as reported by Sun YJ and colleagues in their series of 14 patients.^[5] The median counts of total leukocytes and blasts were 43,800 cells/cu mm and 72%, respectively. The reported FAB classifications, which are common with NSD1 partners, are usually M5, M4, and M1. Cases of KDM5A partners have been reported to show megakaryoblastic differentiation.^{[5-6,9,11-18,22]}

With respect to FLT3-ITD, 75% of patients in this study harbored a concomitant FLT3-ITD, whereas the reported frequency in contemporary literature is 80%.^[22] Knockout studies in mouse models to study this strength of association showed that the penetrance of NUP98/NSD1 alone is relatively low, which is enhanced by FLT3-ITD to initiate leukemogenesis.^[23] The additional loss of WT1 which has been reported in 33–55% cases of NUP98 AML cases also contributes to the leukemogenic drive in these cases.^[20-22] Other co-mutations have also been reported at a similar frequency in other studies.

NGS-based reports have a turnaround time of typically 7–17 days internationally^[24] hence, induction therapy choice is mostly based on immunophenotype and a basic single gene PCR-based testing, which is performed in ~90% of cases, as a practice worldwide. However, in our setting, the turnaround time was <5 days, and the choice of induction therapy was based on genomic findings. This is the reason for the varied induction regimens used in the study cohort. Intensive care was not offered due to poor performance status or concomitant pneumonia, as previously described. Induction was switched to hypomethylating agents combined with venetoclax and FLT3 directed therapy in patients with a positive ITD mutation. It has been shown in a real-world series, intensive chemotherapy also results in failure to achieve CR in 80% of cases, and the choice of transplant is a therapeutic dilemma.^[25] The reason for achieving CR and a stormy disease course has been attributed to upregulation of HOXA/B genes, which are locked in a transcriptionally active state.^[19] This could be a future therapeutic target in these cases by employing menin inhibitors, as described in various translational studies.

Since none of the patients in our cohort who underwent induction were amenable to transplant, the outcomes of transplant were not described. However, as per the literature, event-free survival and OS have been reported as 17% and 36%, respectively, for NUP98/NSD1 fusion-positive AML.^[6,22,23]

With restrictions on sample size and limited data on transplants, the power of this study may be limited. However, it elucidates an important aspect, particularly for the Indian scenario of this particular AML subtype. The detection of aggressive high-risk AML before induction therapy using fast in-house NGS reduces the unnecessary financial burden of empirical intensive chemotherapy for the patient. This is particularly important in economically restrained setups; hence, accurate and prompt detection is an unmet need. This can be achieved through the decentralization of laboratory processes and policy changes, including better insurance coverage.

CONCLUSION

This is a single-center experience from India and the largest reported so far, underscoring the need for urgent therapeutic advances. The co-mutational landscape is also important for therapeutic decision-making and transplant triaging. Hence, this also underscores the need for upfront NGS-based testing to detect this fusion as it portends an adverse prognosis and also to decipher the disease biology affected by concurrent genomic alterations. Translational research holds promise for the same, and the use of menin and SRC kinase inhibitors through controlled trials may help in understanding their efficacy.