What's new?

Our study was a comprehensive, retrospective, 20‑year analysis of clinical outcomes of therapy‑related acute myeloid leukemia (t‑AML), a high‑risk disease triggered by previous radiotherapy and / or chemotherapy implemented for the treatment of preceding solid tumors or hematological neoplasms. Molecular biology testing for 17p13 deletion and TP53 mutation was also conducted, reporting 4 novel TP53 gene mutations and noting a high frequency of cytogenetic abnormalities within t‑AML. Our work outlines that t‑AML treatment strategy should consist in performing allogenic hematopoietic cell transplantation as soon as possible.

Introduction

Therapy‑related acute myeloid leukemia (t‑AML) is a hematological neoplasm occurring as a late complication of DNA‑damaging therapy administered for prior malignancies.^1,2 It accounts for 20% of all AML cases and is usually associated with a poor prognosis.^3-5 Unique clinical features distinguish t‑AML from primary AML. Mutagenic damage characterizing this subtype is usually triggered by exposure to alkylating agents, topoisomerase II inhibitors, radiotherapy, antimetabolites, or antitubuline agents. Therefore, patients undergoing chemotherapy have a 4.7‑fold higher risk for t‑AML than the general population.⁶ Adverse cytogenetic abnormalities and a high frequency of TP53, DNMT3A, FLT3, NPM1, and NRAS gene mutations are often observed in t‑AML,⁷ influencing its aggressive biology.⁸ Disease management strategy in t‑AML should be adjusted to a patient’s medical condition, taking into account the cumulative toxicity of prior therapies. While allogenic hematopoietic cell transplantation (alloHCT) seems to be the most appropriate treatment approach in t‑AML, data concerning the effects of such therapy are limited.^9,10 Even though the number of t‑AML cases is increasing worldwide, mostly as a result of a growing number of cancer survivors,^3,6 comprehensive reports of treatment outcomes remain sparse in the literature.¹¹ Hence, this study aimed to describe the clinical characteristics and treatment outcomes of patients with t‑AML, focusing primarily on the genetic profile of the disease and treatment complications. In our opinion, understanding the impact of the genetic landscape of t‑AML on the clinical outcomes remains essential for any potential improvements in the treatment strategy.

Patients and methods

We retrospectively analyzed medical records of 723 consecutive patients with AML treated at the Department of Hematology and Bone Marrow Transplantation (Poznań, Poland) to select individuals with the following characteristics: 1) a diagnosis of t‑AML according the 2016 World Health Organization criteria,¹ 2) age greater than or equal to 18 years, 3) hospitalization between January 2000 and July 2021. The clinical characteristics recorded in this group included age, sex, molecular and cytogenetic data, type of treatment, treatment complications (toxicity, graft versus host disease [GvHD], infectious complications), treatment response, and details on the primary malignancy (date of diagnosis, type of cytotoxic therapy, latency period). To define the genetic risk of t‑AML, we applied and evaluated the 2017 European Leukemia Net (ELN) criteria.^1,12

Testing for the TP53 DNA sequence variant (exons 2–11, Sanger sequencing, reference sequences: NM_000546.6, NP_000537.3) and 17p13 deletion (fluorescence in situ hybridization) was performed using retrospectively‑collected bone marrow (BM) samples. TP53 molecular analysis was always conducted in the case of negative 17p13 deletion results. The remaining genetic and cytogenetic tests were performed at the time of t‑AML diagnosis in accordance with the applicable guidelines.

The study was approved by the local bioethical committee (1040/19) and conducted in accordance with the Declaration of Helsinki. The patients provided their written informed consent to participate in the study.

Results

Cytogenetic and molecular characteristics of therapy‑related acute myeloid leukemia

Cytogenetic abnormalities were observed in 82.9% of the patients with t‑AML, the most common ones being a complex karyotype (CK) and 17p13 deletion, with a frequency of 26.8% and 26.7%, respectively. Furthermore, FLT3‑ITD and TP53 mutations were respectively found in 15.4% and 12.5% of cases.

A total of 5 TP53 DNA sequence variants, namely, c.711G>A p.(Met237Ile), c.704A>G p.(Asn235Ser), c.375G>C p.(Thr125=), c.733G>C p.(Gly245Arg), and c.989T>C p.(Leu330Pro) were detected in 4 patients with t‑AML (Figure 1, Table 3). Importantly, only 1 variant, c.711G>A p.(Met237Ile), was previously described in AML according to the Catalogue of Somatic Mutations In Cancer (COSMIC) database.¹³ In 75% of the patients with t‑AML, a TP53 DNA sequence variant co‑occurred with CK (Table 3). Additionally, based on the analysis of TP53 mutation and the presence of 17p13 deletion, the majority of the patients with t‑AML (60.9%) were categorized into the adverse genetic risk group, according to the 2017 ELN risk stratification criteria¹² (Table 4).

Table 3. Clinical characteristics of patients with therapy‑related acute myeloid leukemia with detected TP53 DNA sequence variants and 17p13 deletion

ID	TP53 DNA sequence variant	17p13 deletion (% of cells)	Primary malignancy	Cytotoxic therapy	Latency time, y	Cytogenetics	Molecular markers	2017 ELN risk category12	t‑AML treatment	Age, y	Sex	OS, mo	Alive
a With all‑trans retinoic acid and arsenic trioxide Abbreviations: CLL, chronic lymphocytic leukemia; NHL, non‑Hodgkin lymphoma; OS, overall survival; PML‑RARA, promyelocytic leukemia / retinoic acid receptor alpha; others, see Tables 1 and 2
5	c.711G>A p.(Met237Ile)	Negative	CLL	CTH	5	Complex karyotype	No	Adverse	CTH	58	Woman	2	No
40	c.704A>G p.(Asn235Ser)	Negative	Multiple myeloma	CTH, autoHCT	9	Normal karyotype	No	Adverse	CTH, alloHCT	66	Woman	13	No
49	c.989T>C p.(Leu330Pro)	Negative	Breast cancer	CTH	3	Complex karyotype	No	Adverse	CTH	34	Woman	4	No
4	c.375G>C p.(Thr125=) c.733G>C p.(Gly245Arg)	Negative	CLL	CTH	3	Complex karyotype	No	Adverse	Palliative	59	Man	1	No
37	Negative	Positive (16)	Breast cancer	CTH, RTH	14	Normal karyotype	No	Adverse	CTH	63	Woman	5	No
1	Negative	Positive (6)	Testicular cancer	CTH	5	Unanalyzable	PMR‑RARA, FLT3‑ITD	Adverse	Palliative	26	Man	1	No
44	Not tested	Positive (12)	NHL	CTH, autoHCT	13	Complex karyotype	No	Adverse	CTH, alloHCT	51	Woman	140	No
34	Negative	Positive (6)	Prostate cancer	RTH	N/A	8 trisomy	Not tested	Adverse	Palliative	72	Man	3	No
18	Not tested	Positive (8)	Hodgkin lymphoma	CTH, autoHCT	6	Unanalyzable	Not tested	Adverse	CTH, alloHCT	51	Man	5	No
33	Not tested	Positive (20)	Breast cancer	CTH, RTH	2	Inversion (16)	Not tested	Adverse	CTH, alloHCT	39	Woman	151	Yes
2	Not tested	Positive (10)	Breast cancer	RTH	5	Complex karyotype	Not tested	Adverse	CTH, alloHCT	55	Woman	47	Yes
38	Negative	Positive (6)	Breast cancer	CTH, RTH	5	Unanalyzable	No	Adverse	CTH, alloHCT	46	Woman	9	No
50	Negative	Positive (6)	Breast cancer	CTH	5	t(15;17), del(7)	PMR‑RARA	Adverse	CTHa	55	Woman	61	Yes
13	Not tested	Positive (12)	Ovarian cancer	CTH	5	Unanalyzable	Not tested	Adverse	Palliative	58	Woman	4	No
55	Not tested	Positive (49)	Multiple myeloma	CTH, autoHCT	10	Complex karyotype	No	Adverse	CTH, alloHCT	63	Woman	24	Yes
7	Negative	Positive (7)	Hodgkin lymphoma	CTH, RTH	2	Unanalyzable	PML‑RARA	Adverse	CTHa	27	Woman	52	Yes

Table 4. Cytogenetic and molecular characteristics of therapy‑related acute myeloid leukemia

Cytogenetic or molecular marker	n / N (%)
a Defined as ≥3 cytogenetic abnormalities b Number of patients analyzed using FISH c Defined by 2017 ELN12 as ≥3 chromosome abnormalities in the absence of the WHO‑designated recurring translocations or inversions d Defined as ≥2 autosomal monosomies or 1 autosomal monosomy and 1 structural abnormality Abbreviations: C‑KIT, KIT proto‑oncogene, receptor tyrosine kinase; FISH, fluorescence in situ hybridization; n, number of positive results; N, number of tested patients; NPM1, nucleophosmin 1; WHO, World Health Organization; others, see Tables 2 and 3
Cytogenetic assessment
Metaphases not analyzable	9/50 (18)
Metaphases analyzable	41/50 (82)
Normal karyotype	7/41 (17.1)
Cytogenetic abnormalities	34/41 (82.9)
Complex karyotypea	11/41 (26.8)
Deletion of chromosome 17p13	12/45 (26.7); 45b
Complex karyotypec	10/41 (24.4)
t(v;11q23.3); KMT2A rearranged	8/41 (19.5); 13b
t(8;21); RUNX1‑RUNX1T1	7/41 (17.1); 17b
Deletion of chromosome 5	5/41 (12.2)
t(15;17); PML‑RARA	3/41 (7.3); 10b
Deletion of chromosome 7	3/41 (7.3)
t(9;11); KMT2A‑MLLT3	3/41 (7.3); 13b
Monosomal karyotyped	2/41 (4.9)
inv(16) or t(16;16); CBFB‑MYH11	2/41 (4.9); 9b
t(9;22); BCR‑ABL1	1/41 (2.4); 12b
DNA sequence variants
FLT3‑ITD (cDNA)	4/26 (15.4)
TP53	4/32 (12.5)
FLT3‑TKD (D835) (cDNA)	2/26 (7.7)
NPM1	1/22 (4.5)
C‑KIT	1/1
2017 ELN genetic risk stratification12
Favorable	4/46 (8.7)
Intermediate	14/46 (30.4)
Adverse	28/46 (60.9)

In the t‑AML‑MRC subgroup with cytogenetics and molecular biology analysis performed (n = 14), 84.6% and 15.4% of the patients were classified into adverse and intermediate risk categories, respectively, according to the 2017 ELN criteria.¹² 17p13 deletion occurred in 21.4% of the t‑AML‑MRC patients (3/14), while the frequency of FLT3‑ITD and TP53 mutation was 40% and 20%, respectively.

Discussion

In this work, we present a comprehensive analysis of t‑AML, which holds a significant value as compared with registry studies which, while performed in larger cohorts of patients with t‑AML, often lack detailed treatment data.

The Kaplan–Meier survival analysis revealed longer OS in the t‑AML patients undergoing alloHCT and intensive chemotherapy, those who achieved CR after induction therapy, the individuals younger than 64 years, and those with t‑APL. In turn, worse OS was observed in the patients with CK t‑AML. Among the alloHCT recipients, we noted a tendency toward longer OS when the procedure was performed during CR and with the use of MAC.

In the univariable Cox proportional hazard regression model, t‑AML treatment with alloHCT, intensive chemotherapy, CR after induction therapy, age under 64 years, and female sex were the factors associated with better OS. On the other hand, the presence of CK was linked to worse OS. In the multivariable analyses, treatment with alloHCT remained an independent prognostic factor for better OS. In turn, higher neutrophil count at the diagnosis was an independent prognosticator of an unfavorable outcome.

The frequency of t‑AML in our report was 8.1%, and was comparable to previously reported numbers.^3,5,14 The majority of t‑AML cases were preceded by ST, which corresponded to a higher ST incidence worldwide in comparison with HN.^3,6,14 Moreover, the most common primary malignancies in t‑AML were breast cancer and lymphomas, which is in line with other available reports.^3,16 High incidence of breast and gynecological tumors as the primary malignancy reflected the female predominance in t‑AML. The growing use of adjuvant therapy for the treatment of early‑stage breast cancer, the most frequent malignancy affecting women in Poland between 2000 and 2021, could be the factor contributing to the increase in the number of t‑AML cases.¹⁷ On the other hand, the large proportion of lymphomas as the primary malignancy corresponded to the large number of autologous hematopoietic cell transplant procedures performed to reduce lymphoproliferation and mitigate its leukemogenic effect. We observed a tendency toward better OS rates in the cases of t‑AML preceded by HN, which might have reflected the active hemato‑oncological surveillance allowing for prompt diagnosis and faster initiation of treatment. The median latency time in t‑AML was shorter when the disease had been preceded by ST than HN, which is in accordance with previous reports.^11,18 Higher neutrophil count at the diagnosis represented an unfavorable predictor of survival, which is consistent with an earlier study.¹⁹

Moreover, the type and dose of cytotoxic therapy influences the latency time in t‑AML.^3,20,21 We noted the shortest latency period after prior radiotherapy, which is known to evoke BM environment damage disturbing hematopoiesis.^3,22 We reported the shortest 2‑year OS after prior combined cytotoxic therapy, which probably reflects the highest cumulative dose. Interestingly, despite the known effect of cytotoxic chemotherapy on the development of therapy‑related myeloid neoplasms (t‑MNs), some data suggested an association between the BRCA DNA sequence variants and an increased risk of de novo leukemia development. This mechanism involved inactivation of the error‑free repair process for double‑stranded DNA within the BRCA1 / BRCA2 protein pathway in breast cancer survivors.²³ Some studies report that modern ST therapies using poly (ADP‑ribose) polymerase inhibitors (PARPi; eg, olaparib) have been associated with a higher risk of t‑MNs.²⁵ On the other hand, there is an ongoing phase 2 clinical trial based on the use of PARPi for the treatment of t‑AML (NCT03953898).

Even though 17p13 deletion was detected in all t‑APL patients in our study (with a lack of TP53 mutation), their treatment using ATO / ATRA remained highly efficient. Interestingly, ATO inactivates TP53 functions through the 26S proteasome pathway,²⁵ and has been found to amplify wild‑type TP53 functions and provoke upregulation of its target genes, thus promoting apoptosis.²⁶ Interestingly, t‑AML cases after successful APL treatment were reported.²⁷

Genomic heterogeneity of t‑AML is associated with previous cytotoxic exposure, age of the patients, and the presence of clonal hematopoiesis prior to toxic exposure.²⁸ It should be noted that, over the period between 2000 and 2021, the diagnostic strategy in t‑AML has evolved, with immunoflow cytometry, cytogenetic, and molecular biology examinations introduced in our center in 2008. Cytogenetic abnormalities (82.9%) and CK (26.8%) were detected in proportions of patients comparable with those reported in previous reports on t‑AML,^{3,4,7,19,29-32}with overrepresentation of 11q23 translocations as well as adverse cytogenetics, including complex and monosomal karyotypes, and with underrepresentation of intermediate‑risk karyotypes (P <⁠0.001).

Moreover, we found a similar frequency of FLT3‑ITD mutations (15.4%) but lower TP53 (12.5%) and NPM1 (4.5%) mutation frequency, as compared with other authors.^3,7,33,34 Surprisingly, we detected 17p13 deletion more often (26.7%) than other studies.³ Nevertheless, the real number of molecular abnormalities could have only been determined through a comprehensive retrospective molecular analysis.

After the retrospective molecular analysis of 17p13 deletion and TP53 mutation, the 2017 ELN risk stratification category¹² changed to adverse in 10 out of 46 cases, with the vast majority of the t‑AML patients (60.9%) being classified as adverse‑risk. Overall, 17p13 deletion or TP53 mutation was detected in 32.6% of the t‑AML patients in our study. Therefore, in the case of a clinical diagnosis of t‑AML, we strongly emphasize the need for testing for those abnormalities. When analyzing the results, in the absence of t‑AML–specific genetic stratification tool, we recommend adhering to the 2017 ELN criteria.¹² Interestingly, c.704A>G p.(Asn235Ser), c.375G>C p.(Thr125=), c.733 G>C p.(Gly245Arg), and c.989T>C p.(Leu330Pro) TP53 DNA sequence variants detected in our study represented novel mutations in AML according to the COSMIC database.¹³

Targeted therapies were implemented in 2 patients with FLT3-positive t‑AML. Midostaurin was added to the induction chemotherapy (resulting in CR), while gilteritinib was implemented in refractory t‑AML (CR was not achieved). The number of t‑AML patients receiving targeted therapy was insufficient to conclude on the effectiveness of such treatment in our study.

There are several ongoing clinical trials on the use of molecular targeted therapies in ultra‑high‑risk TP53-mutant AML patients.³⁵ Eprenetapopt (APR‑246) is a promising molecule destabilizing the individual TP53 point mutation, and the results of a phase 3 study on the effectiveness of this agent in combination with azacitidine (NCT03745716) are pending publication.³⁶

Among the patients with t‑AML‑MRC, we reported a higher frequency of adverse risk category according to the 2017 ELN classification¹² and a higher frequency of FLT3‑ITD mutation than in the individuals with t‑AML. Furthermore, in line with other reports, we observed a tendency toward worse treatment outcomes among the t‑AML‑MRC patients,³⁷ which may be explained by the presence of additional, distinct secondary‑type mutations.³⁸

AlloHCT in t‑AML significantly extended OS rates as compared with standard chemotherapy, which is in line with previous reports.^39,40 An anti‑leukemic effect of alloHCT in t‑AML was pre‑eminently visible when the procedure was performed during CR, which confirmed observations of other groups.^41,42 However, we reported higher OS and PFS rates after alloHCT than other authors,^11,41 which indicated that not all t‑AML cases were characterized by poor prognosis.¹¹ Older t‑AML patients undergoing alloHCT, with a higher comorbidity index score, represent preferable candidates for RIC regimens, which decrease toxicities related to the procedure.^11,43 In our report, RIC was performed in the vast majority of the patients with t‑AML, which was a major difference as compared with other studies.^41,42,44 Moreover, we observed a tendency toward longer OS rates after MAC (limited number of patients), which was consistent with the European Society for Blood and Marrow Transplantation data.⁴⁵ Hence, the treatment strategy in t‑AML should be based on performing alloHCT as soon as possible in the patients in CR, and with the use of intensive conditioning regimens. On the other hand, the use of alloHCT should be questioned in t‑AML patients who are in an active disease stage.

The main cause of death in our study was t‑AML progression, which corresponds to the findings of other authors.⁴¹ Nonetheless, prolonged depletion of hematopoietic reserves resulting from previous cytotoxic treatment predisposed the patients to severe treatment‑related complications.¹³ In the t‑AML patients undergoing intensive chemotherapy, hepatotoxicity, renal toxicity, and cardiotoxicity were the most frequent causes of death. At the same time, after alloHCT, organ toxicity rates were higher, which reflected cumulative toxicity resulting from previous cytotoxic therapies. Organ toxicity resulting from alloHCT is mostly triggered by the use of MAC and calcineurin inhibitors, infections, and specific organ complications.^31,44 Furthermore, primary malignancy progression was the cause of death in every tenth case of t‑AML.

We demonstrated a high frequency of infectious complications after intensive chemotherapy in t‑AML. However, it should be noted that prophylaxis and treatment strategies against fungal, bacterial, and viral infections evolved within the study period. After intensive chemotherapy, during neutropenia, gram‑positive bacteria (GPB) BSIs were the most common, which is in agreement with the results of other authors.⁴⁶ Gram‑negative bacteria (GNB) in neutropenia were detected more frequently after alloHCT, which is in line with the recently observed shift toward GNB predominance after alloHCT.⁴⁷ However, after intensive chemotherapy, GPB BSIs during neutropenia were more common. The t‑AML patients after alloHCT developed drug‑induced T cell dysfunction, which corresponds to a higher frequency of viral reactivations. Furthermore, prolonged neutropenia in t‑AML patients increases the risk of developing IFI.⁴⁷ In our study, IFI was confirmed in 9.5% of the t‑AML cases after intensive chemotherapy, and in 4.3% alloHCT recipients, with the numbers being notably lower than in other studies.^48,49 As infections represented the most common non–relapse‑related cause of death, improvements in their prophylaxis and treatment remain essential to improve t‑AML outcomes.^3,50

We have identified several limitations of our study: 1) comprehensive information on the type and dose of cytotoxic agents was not registered, 2) t‑AML therapeutic strategies and known treatment complications changed over the 2 decades during which the study was conducted, 3) only a limited number of patients were subjected to a comprehensive molecular characterization within the study period.

To summarize, our study reveals that 1) not every t‑AML subtype is associated with a poor prognosis, 2) the treatment strategy in t‑AML should be based on performing alloHCT as soon as possible, 3) t‑AML is characterized by a high frequency of CK, predisposing to inferior OS, 4) molecular testing for the TP53 DNA sequence variant or 17p13 deletion allows for proper t‑AML stratification using the 2017 ELN criteria¹² and creates an opportunity for therapeutic approaches targeting the TP53 mutation, 5) further improvements in the management of toxicity- and infection‑related t‑AML complications remain essential to improve its outcomes.