Allogeneic Hematopoietic Stem Cell Transplantation in Therapy Related Acute Leukemia

Abstract

Therapy related acute leukemia consists of a unique subset of acute leukemia with an increased frequency of high risk cytogenetic and molecular abnormalities, dismal response to therapy, higher relapse rates and poor overall survival. Therapy related acute myeloid leukemia (t-AML) is a better defined disease entity than therapy related acute lymphoid leukemia (t-ALL). However, in recent times, t-ALL is also being increasingly recognized and extensively studied. Therapy related acute myeloid leukemia is usually classified together with therapy related myelodysplastic syndrome. However, the management of these two diseases maybe different regarding needs of induction chemotherapy and eligibility for upfront allogeneic hematopoietic stem cell transplantation (Allo HSCT). There is also evidence regarding differences in prognosis and outcomes between these two entities. Allo HSCT offers a potential for cure in t-AML and t-ALL. However, existing literature on the same is confounded by inclusion of t-MDS and secondary acute leukemias. Here we review the current evidence on the outcomes and predictors of outcomes of Allo HSCT in the management of therapy related acute leukemias. We also shed light into the under-representation of therapy related leukemias in clinical trials. This stresses the need for prospective trials incorporating measurable residual disease monitoring and sequential next generation sequencing based genomic data for accurate prognostication and management of therapy related acute leukemia.

Background

Therapy-related acute leukemia is an uncommon, distinct subset of acute leukemia that occurs after exposure to cytotoxic chemotherapy and/or radiation for a malignant or non-malignant disorder [1]. The incidence of therapy-related acute leukemias is increasing given the improved survival of patients who are treated for prior malignancies, and increased life expectancy of the population in general [2]. In view of the unique pathogenesis, clinical course and poor prognosis, therapy-related acute myeloid leukemia (t-AML), along with therapy-related myelodysplastic syndrome (t-MDS), have been separately classified as a distinct entity termed “Therapy-related myeloid neoplasms (t-MN)” in the World Health Organization classification of Myeloid Neoplasms [3]. Therapy-related acute myeloid leukemia (t-AML) is the well-described entity among therapy-related acute leukemias. Therapy-related acute lymphoblastic leukemia (t-ALL) is less commonly reported in comparison to t-AML, consisting of 10–15% of therapy-related acute leukemias [4]. However, the entity is being increasingly recognized and it’s prevalence is reported to be up to 10% of all acute lymphoblastic leukemias [5].

Studies in t-ALL and t-AML are confounded by the inclusion of “secondary leukemia” which can be defined as the occurrence of acute leukemia in patients with prior malignancy, independent of exposure to cytotoxic therapy [6, 7]. Secondary acute leukemia may also include genetic syndromes known to predispose to acute leukemia and, (in the case of AML) progression of other hematological malignancies like MDS or myeloproliferative diseases. Available data suggests that patients with therapy-related acute leukemia (both t-ALL and t-AML) have an increased frequency of adverse cytogenetic abnormalities, conferring a poor prognosis (median survival: 5–16 months) [8–10].

The literature on allogeneic hematopoietic stem cell transplantation (Allo HSCT) in therapy-related acute leukemia is relatively sparse. This is primarily due to the rarity of the disease but may also be due to patient attrition owing to comorbidities, poor performance status (due to prior cytotoxic therapy), inability to attain remission status and relapse. Here we review the current evidence on the use and outcomes of Allo HSCT in patients with t-AML and t-ALL.

Allo HSCT in t-AML

Therapy-related AML was first described in patients who had prior exposure to the alkylating agent melphalan [11]. Subsequently, t-AML was described in patients after exposure to radiotherapy or topoisomerase inhibitors [12]. Although t-AML has been classified together with t-MDS and both these entities share similar cytogenetic and mutation profiles; the biological characteristics, and outcomes of these two diseases are notably different [13]. Patients with t-AML developing from t-MDS were shown to have poorer outcomes after Allo HSCT when compared to patients with t-MDS [14].

Patients with t-AML have a higher prevalence of abnormal karyotype and an increased frequency of adverse cytogenetic abnormalities (like Chromosome 5 and/or 7 abnormalities, complex karyotype etc.) [15]. The mutation panel in t-AML shows a distinct pattern with increased incidence of certain mutations like TP53, and less frequent incidence of mutations which are otherwise common to denovo AML (like FLT3 and NPM1) [16, 17]. This suggests a dissimilar mechanism of leukemogenesis in t-AML when compared to denovo AML and hence the possibility of different outcomes after Allo HSCT. Existing literature on Allo HSCT in patients with t-AML have uniformly shown poorer prognosis in patients with high-risk cytogenetics [9, 15]. The CIBMTR data on 868 patients of t-MDS/AML showed that irrespective of Allo HSCT, high-risk karyotypes had a negative impact on the outcomes [18]. Data from the European Society of Blood and Marrow transplantation (EBMT) showed that the presence of any abnormal karyotype, except favorable-risk cytogenetics, was a predictor of poor prognosis after Allo HSCT in patients with t-MDS/AML [19]. In a study on Allo HSCT in patients with morphological remission of t-AML at our center, any abnormal karyotype or high-risk chromosomal abnormalities like deletion of Chromosome 7 or 11q23 rearrangement was predictive of poor outcomes [20]. Thus, presence of poor-risk cytogenetic abnormalities predicts outcomes even after Allo HSCT in patients with t-AML.

Another concern in patients of t-AML undergoing Allo HSCT is the increased non-relapse mortality (NRM) in this subset of patients. The higher NRM in t-AML may be attributable to additional unfavorable factors like poor organ tolerance of treatment, chronic immunosuppression from previous therapy, poor hematopoietic reserve, and transfusion refractoriness [21, 22]. The French Society of Bone Marrow Transplantation has reported a 2-year NRM of 49% after Allo HSCT in t-MDS/AML [23]. Anderson et al. [24] described a 5-year NRM of 44.3% after Allo HSCT in secondary AML (AML after MDS, and t-AML) and further observed that a shorter duration from diagnosis to transplant was associated with lower NRM. In their multicenter study, Kayser et al. reported a 4-year NRM of 38.5% [25] while data from the CIBMTR showed an NRM of 49% in 5 years after Allo HSCT in patients of t-AML [18]. NRM after Allo HSCT is decreasing over time due to improvements in modern supportive care, changes in GVHD prophylactic regimens, and improvement in diagnosis and treatment of both GVHD and graft failure.[26] Data from the EBMT demonstrated 3-year NRM to be 37% after Allo HSCT in t-MDS/AML, but also revealed continuous improvement in NRM with each year of the study [19]. Analogous serial improvement of NRM with each passing year was also demonstrated by Chang et al. [27] in their study on Allo HSCT in patients with MDS/MDS-AML. Our center reported a 2-year NRM after Allo HSCT in t-AML of 34.5% with patients undergoing Allo HSCT after 2010 having better outcomes when compared to patients whose Allo HSCT occurred before 2010 [20].

The poor survival outcomes in patients with t-AML due to intolerance to intensive induction regimens and high relapse rates, have led to an attempt at different consolidation treatment strategies, including Allo HSCT. Initial data from Fred Hutchison Cancer Centre showed a 5-year overall survival (OS) of 24.4% after Allo HSCT in t-AML [24]. The French Society of Bone Marrow Transplantation reported a 2-year OS of 30% while the CIBMTR cohort had a 5-year OS of 22% after Allo HSCT in t-MDS/AML [23]. Both these early studies had significant proportion of patients (91.3% and 66% respectively) who were not in morphological remission / chemo refractory at time of transplantation. Fianchi et al. [28] reported a median OS of 58.8 months after Allo HSCT in t-MDS/AML which was significantly better than the survival in patients consolidated with other modalities The EBMT group initially reported a 35% probability of 3-year disease-free survival after Allo HSCT in patients with t-MDS/AML transplanted from HLA-identical siblings [29]. The follow-up study from EBMT in t-MDS/AML reported a 3-year event-free and overall survival rates of 33% and 35%, respectively [19]. Data from CIBMTR and EBMT registries had a significant proportion of patients of t-AML who were not in remission (26% and 38% patients respectively) at the time of Allo HSCT [18, 19]. Almost all studies in Allo HSCT report that morphologically active disease or inadequate disease control at the time of transplant is a risk factor for poor outcomes [27]. This can partly explain the poor overall survival of the patients in these studies. There is a paucity of multicenter registry data regarding survival outcomes of patients with t-AML undergoing Allo HSCT with disease in morphological remission. Our 20-year experience of patients with t-AML transplanted in morphological remission showed a 2-year OS of 44.1% [20]. The paucity of literature on measurable residual disease (MRD) monitoring in patients with t-AML prevents us from making definite conclusions on the depth of response needed for favorable outcomes after Allo HSCT. Available literature on Allo HSCT in t-AML and outcomes are summarized in Table 1.

Table 1.

Outcomes of therapy related acute myeloid leukemia undergoing allogeneic hematopoietic stem cell transplant

References	Inclusion criteria	Total patients included	Allo HSCT in t-AML (n)	NRM	Outcomes after Allo HSCT
Anderson et al. [24]	s-AML (t-AML/MDS-AML)	46	17	5-year NRM—44.3% (t-AML)	5-year OS = 7.8% (t-AML)
Yakoub-Agha et al. [23]	t-MDS/AML	70	39	2-year NRM—49% (t-MDS/AML)	2-year OS = 30% (t-MDS/AML)
De Witte et al. [29]	MDS/MDS-AML/t-MDS-AML	885	67 (t-MDS-AML)	NA	3-year OS = 35% (t-MDS-AML)
Chang et al. [27]	sMDS/AML (developing from MDS)	257	80	NA	5 years RFS = 19% (t-MDS-AML)
Kroger et al. [19]	t-MDS/AML	461	308	3-year NRM—37% (t-MDS/AML)	3-year OS = 35%(t-MDS/AML)
Litzow et al. [18]	t-MDS/AML	868	545	5-year NRM—49% (t-MDS/AML)	3-year OS = 27% (t-MDS/AML)
Kayser et al. [25]	AML	2853	NA	4-year NRM—38.5% (t-AML)	4-year OS = 42.6% (t-AML)
Granfeldt et al. [21]	AML (denovo/ sAML/t-AML)	3055	12	NA	NA
Fianchi et al. [28]	t-MDS/AML	277	42 (t-MDS/AML)	NA	Median OS 58.8 months (t-MDS/AML)
Nampoothiri et al. [20]	t-AML	52	52	2-year NRM—34.5% (t-AML)	2-year OS = 44.1% (t-AML)

Allo HSCT allogeneic hematopoietic stem cell transplantation, t-AML therapy related acute myeloid leukemia, NRM non-relapse mortality, t-MDS therapy related myelodysplastic syndrome, OS overall survival, t-MDS-AML AML developing from t-MDS, s-AML secondary acute myeloid leukemia, t-MDS/AML tMDS or t-AML

Although the role of cytogenetics and disease control at the time of Allo HSCT has been established to some degree, the significance of other predictive factors is varied across different studies. Younger age was associated with better outcomes in some studies [18, 19] and other studies showed that good performance status at the time of HSCT correlated with comparable response rates and survival as denovo AML [20]. Transplantation with less than well-matched donors was associated with poor outcomes in patients with t-AML, as suggested by Nampoothiri et al. as well as data from the CIBMTR [18, 20]. The French Society of Bone Marrow Transplantation reported negative recipient CMV serology, and female gender was associated with a better outcome in 70 patients of t-MDS/AML after Allo HSCT [23] Nampoothiri et al. [20] reported improved outcomes in patients using non-CSA/MTX GVHD prophylactic regimens. Although there is data to suggest that modern GVHD prophylactic regimens incorporating a more frequent usage of drugs like anti-thymocyte globulin and /or post-transplant cyclophosphamide results in improved GVHD free relapse-free survival [30], there is a paucity of literature on the prognostic impact of GVHD prophylactic regimes on outcomes after Allo HSCT in t-AML.

Allo HSCT offers a chance of cure in a subset of patients with t-AML. However, the existing literature on Allo HSCT in t-AML is a patchwork of studies with additional inclusion of t-MDS, secondary AML and t-AML undergoing HSCT with the disease not in morphological remission. Based on available literature we would recommend Allo HSCT for eligible and fit patients with t-AML after intense induction regimens and in morphological remission. Published studies on Allo HSCT in t-MDS/AML are mostly retrospective in nature with a lack of measurable residual disease (MRD) monitoring and next-generation sequencing (NGS) genomic data. Moreover, patients with t-AML are grossly under-represented in clinical trials and thus, trial generated data cannot be extrapolated into the management of patient with t-AML [31]. There is a need for prospective trials incorporating NGS and MRD monitoring data to distinguish between actual therapy-induced AML and cases occurring as stochastic events or due to an inherited genetic predisposition. Such trials will delineate the benefit of Allo HSCT in patients with t-AML and accurately enumerate the prognostic factors associated with favorable outcomes.

Allo HSCT in t-ALL

Therapy-related acute lymphoblastic leukemia is a less common and increasingly recognized subgroup of therapy-related acute leukemia. The incidence of the same was thought to be around 1.2–2.6% of all ALL patients [32, 33], but recent literature estimates the figure to be around 10% [5]. There is a paucity of literature on genetic mechanisms, clinical characteristics and treatment outcomes of t-ALL. Moreover, most of the existing literature on t-ALL is confounded by the inclusion of patients with ALL occurring after prior malignancy, without exposure to any cytotoxic chemo/radiotherapy (secondary ALL) [6]. There is also some ambiguity in the definition of t-ALL; and given the absence of genomic data in t-ALL, there is an ongoing debate over the actual existence of a distinct t-ALL entity [34]. However, available epidemiological studies strongly suggests that t-ALL is a separate entity with characteristic cytogenetic / molecular abnormalities and dismal prognosis [35].

Similar to t-AML, there is a higher incidence of poor-risk cytogenetic abnormalities in patients with t-ALL [10, 36]. It has been postulated that certain characteristic cytogenetic abnormalities (MLL gene rearrangements, monosomal karyotype, complex karyotype, deletion/monosomy 7, and deletion/monosomy 5) may be seen more frequently in patients with ALL in whom disease occurs post therapy, and hence having a worse prognosis. Apart from prognostic impact, this hypothesis is supported by a shorter latent period between exposure to cytotoxic therapy and diagnosis of t-ALL in patients with MLL gene rearrangements [37]. Certain other cytogenetic abnormalities (normal karyotype, BCR-ABL gene translocation) appear in same frequency in patients with t-ALL and patients with denovo ALL [37]. Thus, it is postulated that patients with this group of cytogenetic abnormalities may have an incidental de novo ALL, unrelated to the therapy for prior malignancy [38]. However, when compared to patients with de novo Ph+ ALL, the incidence of additional cytogenetic abnormalities was found to be higher in patients with Ph+ve t-ALL, suggesting the effect (and adverse prognostic impact) of prior therapy even in these patients [39]. There is literature to support the correlation of MLL gene rearrangement in patients with therapy-related acute leukemia (both t-AML and t-ALL) with previous exposure to topoisomerase inhibitors [10, 40, 41].

The literature on Allo HSCT in t-ALL is sparse. Available data suggest that outcomes are comparable between t-ALL and de-novo ALL after Allo HSCT, especially with disease in the first complete remission (CR1) [37, 39]. There is a lack of prospective data on the outcomes of allogeneic HSCT in patients with t-ALL. Retrospective data on Allo HSCT in t-ALL are also scarce, with fewer than 140 reported cases overall (Table 2). Many of these studies are further confounded by the inclusion of patients having secondary ALL. Vasudevan et al. [42] reported a 2-year RFS and OS of 47.1%, and 51.8% after Allo HSCT in t-ALL but without a comparison to de novo ALL. Aldoss et al. [39] had similar outcomes in their subgroup of patients with t-ALL undergoing allogeneic HSCT. Their series, which is the largest published data set on Allo-HSCT in t-ALL, demonstrated a comparable NRM and OS in patients undergoing HSCT for t-ALL and de novo ALL, with neither cytogenetics nor prior cytotoxic therapy predicting survival. This was different from the recent multicenter study by Saygin et al. who demonstrated that although patients of t-ALL who underwent Allo HSCT had better outcomes than patients (with t-ALL) who did not, overall outcomes were inferior to de-novo ALL [10]. They also showed that Allo HSCT was the only predictive factor for a favorable outcome in patients with t-ALL.

Table 2.

Outcomes after allogeneic hematopoietic stem cell transplant in patients with therapy related ALL

Study	t-ALL (N)	Most frequent cytogenetic abnormality	Allo HSCT n (%)	NRM	Outcomes after Allo HSCT (2-year OS)
Shivakumar et al. [44]	89b	11q23 rearrangement	14 (15.73)	NA	NA
Abdulwahab et al. [5]	23	11q23 rearrangement	5 (21.7)	20% (2 years)	60%
Ganzel et al. [34]	23	t (9; 22)	8 (34.8)a	NA	NA
Kelleher et al. [45]	16	Hyperdiploidy (33%)	2 (12.5)	NA	NA
Aldoss et al. [39]	93	t (9; 22)	49 (53.7)	28.5% (unknown time)	53.4%
Saygin et al. [10]	116	t (9; 22)	42 (36.2)	NA	NA (median OS 46 months)
Vasudevan et al. [42]	18	11q23 rearrangement	18 (100)	33.3% (2 years)	51.8%

t-ALL therapy related acute lymphoblastic leukemia, Allo HSCT allogeneic hematopoietic stem cell transplant, NRM non relapse mortality, OS overall survival, NA not available

^aBoth secondary ALL(s-ALL) and t-ALL combined

^bIncludes both own patients and that collected from literature

Unlike t-MDS/AML there is a lack of large registry (CIBMTR/EBMT) studies on outcomes and outcome predictors of t-ALL after Allo HSCT. Hence potential predictors of successful therapy, specific to patients of t-ALL undergoing Allo HSCT, remain largely unknown. Despite these gaps in available literature, we would recommend Allo HSCT for eligible and fit patients with t-ALL in morphological remission. Prospective studies with NGS data and systematic MRD testing, may delineate prognostic parameters and help predict which patients will potentially benefit from allogeneic HSCT. This is more relevant in the era of evolving targeted treatments for ALL like Bispecific T cell engagers (BiTEs), antibody–drug conjugates, and chimeric antigen receptor T (CAR-T) cells.

Conclusions

Therapy-related acute leukemias are increasingly recognized disease entities with characteristic, clinical, pathological, cytogenetic and molecular abnormalities. They are grossly under-represented in clinical trials, possibly due to the poor performance status and outcomes of these patients. Only 1.5% of all AML patients included in EORTC-GIMEMA trials had secondary AML(s-AML) although the number of patients with s-AML registered during the same period was at least 6% [43]. Inside the MRC AML-10, 11, and 12 trials, patients with t-AML accounted for only 2.3% of all AML patients [9], emphasizing the underrepresentation of the disease entity in clinical trials. As a consequence of this discrepancy, the optimal treatment of t-AML and t-ALL is largely unknown. Allogeneic hematopoietic stem cell transplantation offers a chance of long-term survival for a proportion of patients with therapy-related acute leukemias but the predictors for improved survival, especially in the era of genomic monitoring panels and targeted cellular therapies, need to be well elucidated.

Funding

No funding source.

Compliance with Ethical Standards

Conflict of interest

The authors declare that they have no conflict of interest to disclose.