Background
Since 1982, patients with 20–30 % bone marrow blasts have been considered to have myelodysplastic syndromes with refractory anaemia and excess blasts in transformation (MDS-RAEB-t) according to the French-American-British (FAB) classification [
1]. When the World Health Organization (WHO) classification came into effect in 2001, these patients were considered to have acute myeloid leukaemia (AML) with a low bone marrow blast count (hereafter AML20–30; Additional file
1: Table S1) [
1,
2]. This new classification (updated in 2008 [
3]) was driven by novel insights from several studies that identified that bone marrow blast count had more prognostic weight than was originally perceived and that MDS-RAEB-t patients had similar outcomes to patients with AML and more than 30 % bone marrow blasts (hereafter AML30+), partly owing to the fact that MDS-RAEB-t commonly transformed into AML [
4‐
12].
Although the sum of available data led the WHO to conclude that AML20–30 (formerly MDS-RAEB-t) and AML30+ were essentially the same disease 15 years ago, several relevant groups do not appear to consider the scientific evidence to be strong enough: (i) The National Comprehensive Cancer Network (NCCN) endorses both FAB and WHO classification systems, allowing MDS-RAEB-t to be diagnosed and treated as either MDS or AML [
13,
14]; (ii) many large phase III randomised clinical trials still retain MDS-RAEB-t as an MDS sub-entity [
15]; and (iii) while the division of the category MDS-RAEB into RAEB-I and RAEB-II by WHO was validated and generally accepted to add significant prognostic value [
16‐
18], scientific debate regarding the abandonment of the sub-entity MDS-RAEB-t by WHO remains between members of the MDS Study Group [
19], the WHO Myeloid Disease Writing and Clinical Advisory Committees [
20], and between other renowned experts in the field [
21,
22]. Therefore, uncertainty prevails regarding the diagnosis, prognosis, and optimal treatment timing and strategy for patients with AML20–30.
Azacitidine was approved for the treatment of patients with MDS and AML20–30 in 2004 by the Food and Drug Agency (FDA) and in 2008 by the European Medicines Agency (EMA). Although the patient population included in clinical trial protocols resulting in drug approval included up to 38 % of patients with AML30+ (CALGB-protocols 8921 and 9221) [
23], both the FDA and the EMA restricted approval of azacitidine to AML20–30, and a further large randomised clinical trial was required to prove the efficacy of azacitidine in AML30+. This artificial distinction between AML20–30 and AML30+ was made by the regulatory agencies and reflects neither the former FAB classification of MDS-RAEB-t (which could also be diagnosed if bone marrow blasts were <20 % with the presence of >4 % peripheral blood blasts or Auer rods) nor the current WHO classification of MDS and AML. Azacitidine treatment for patients with AML30+ was off-label until very recently (30 October 2015), when the EMA extended the indication for azacitidine to include this patient subgroup. Approval was mainly based on the results of a phase III randomised trial performed exclusively in AML30+ patients [
24]. While bone marrow blast percentage has retrospectively been analysed in smaller patient cohorts for its potential as a prognostic factor for AML patients treated with azacitidine front-line [
25,
26], no in-depth analysis of patient baseline characteristics, treatment characteristics and outcomes exist. To date, no clinical trial has directly compared the efficacy of azacitidine in AML patients with 20–30 % vs >30 % bone marrow blasts, a gap we aimed to bridge.
In this study, we provide the first comparison of baseline characteristics and outcomes of patients with MDS and AML treated with azacitidine front-line, with the intention to (i) provide further insight into the efficacy of azacitidine in the subgroup of AML patients for whom this drug was initially approved (i.e. AML20–30) and to assess whether patients with AML30+ benefit from azacitidine treatment in a similar way to patients with AML20–30; (ii) evaluate the potential prognostic relevance of the presence of MDS-related features (MRF) in AML and (iii) assess the outcomes of patients with MDS and AML as classified by both the WHO and FAB systems, in order to clarify whether elderly, intensive chemotherapy (IC)-ineligible patients with AML20–30 (formerly RAEB-t) should be regarded and treated as having MDS or AML. To answer these questions, this analysis focuses on the differences in morphological features (blast count in peripheral blood and/or bone marrow and presence of dysplasia) between the WHO and FAB classifications.
Discussion
To date, no clinical trial has been performed that specifically assessed the efficacy of azacitidine in the AML patient subgroup for which the drug was initially approved, namely AML20–30 (formerly MDS-RAEB-t). Depending on the classification system used to distinguish MDS from AML, AML20–30 was either grouped together with high-risk MDS (FAB classification) or AML (WHO classification). The only available clinical trial data specific to the AML20–30 subgroup of patients treated with azacitidine is based on sub-analyses of the AZA-MDS-001 trial and the CALGB protocol 8421, which had low AML20–30 patient numbers (
n = 53 and
n = 24) [
23,
36]. In addition, one retrospective analysis from the Dutch named patient programme published data on 38 AML20–30 patients treated with azacitidine, but 13 % of these had relapsed/refractory AML [
37]. We therefore present data from the largest cohort of patients with AML20–30 treated with azacitidine front-line (
n = 79) to date. Outcomes were encouraging with an ORR of 34 % and a median OS of 13.1 months, especially taking the advanced age of this cohort into consideration (median age 77 years with 60 % >75 years). In comparison, patients with AML20–30 included within former clinical trials were younger (median ages were 65, 70 and 72 years in the CALGB-8421 protocol, AZA-MDS-001 trial and the Dutch named patient programme, respectively) [
23,
36,
37]. As a side note, the median OS of 24.5 months obtained in patients with AML20–30 in the AZA-MDS-001 trial is exceptionally long [
36], and thus far, no other group has been able to find similarly long OS times in elderly AML patients, no matter which treatment was investigated, and no matter which bone marrow blast count was used as cut-off (Table
7 [
24,
38‐
48]). Median OS of patients receiving conventional care regimen in the AZA-MDS-001 trial was also extraordinarily high (16.0, 17.0, 14.2 and 13.4 months for all conventional care regimen combined, low-dose cytarabine, IC and best supportive care (BSC), respectively) [
36], indicating that patient selection may have favoured improved survival. The lack of clinical trials allowing direct comparison of the efficacy of azacitidine in AML20–30 vs AML30+ is likely due to the requirements imposed by the registration agencies. A randomised trial performed exclusively in AML30+ [
24] was requested in order to widen the registration indication of azacitidine to include AML30+, thus eliminating the possibility of a direct comparison of the efficacy of azacitidine in AML20–30 vs AML30+. Our data show the first direct comparison of these two patient groups and reveal similar baseline and treatment characteristics, ORR and OS for patients with AML20–30 (
n = 79) vs AML30+ (
n = 111) treated with azacitidine front-line, respectively. We further confirm the efficacy of azacitidine in the subset of patients with AML30+, with a median OS of 10.9 months observed in our cohort (
n = 111), which was similar to that observed in the recently published phase III clinical trial AML-001 (10.4 months;
n = 241) [
24]. Thus, patients with AML30+ seem to derive similar clinical benefit from azacitidine in terms of OS prolongation, as patients with AML20–30.
Table 7
Elderly AML front-line treatment options and median OS times
Untreated | 3367 | ≥65 | 77 | n.g. | 2 | Retrosp. | |
HU+/-ATRA | 99a
| >60 | 74 | 1 | ~3 | III | |
HD-LEN | 33 | >60 | 71 | 30 | 4 | II | |
LD-AraC+/-ATRA | 103a
| >60 | 74 | 18 | <5 | III | |
CFA | 112 | >60 | 71 | 46 | 9.4 | II | |
CFA + LD-AraC | 54 | >60 | 71 | 63 | 11.4 | II | |
CFA + LD-AraC/DAC | 60 | >60 | 70 | 58 | 12.7 | II | |
CFA + LD-AraC/DAC | 118 | >60 | 68 | 60 | 11.1 | II | |
Allo-SCT | 46 | ≥65 | 67 | n.g. | 22 | Retrosp. | |
Intensive CTX | 1856 | ≥65 | 74 | n.g. | 6 | Retrosp. | |
Intensive CTX (3 + 7) | 416 | >65 | 67 | 57 | 12 | III | |
BSC ↔ allo-SCT | 352 | ≥60 | n.g. | n.g. | 9.0 | Retrosp. | |
BSC ↔ allo-SCT | 5480 | ≥65 | 78 | n.g. | 3.0 | Retrosp. | |
DAC (DACO-16) | 238 | >65 | 74 | 18 | 7.7 | III | |
AZA (AZA-AML-001) | 241 | >65 | 75 | 28 | 10.4 | III | |
AZA-AAR | 193 | >17 | 77 | 18 | 12.6 | Retrosp. | |
AZA + LEN | 42 | >60 | 74 | 28 | 15.9 | I/II | |
In AML patients, the presence of MRF has been shown to be associated with adverse clinical outcome [
49‐
51]. Although previous studies reporting on azacitidine in the front-line setting included patients with MRF, outcomes were not reported separately. We demonstrate for the first time that the presence of MRF has no adverse effect on OS of elderly patients treated with azacitidine front-line (Fig.
2a‐
c). This seems to be of clinical relevance in light of adverse outcomes observed with front-line IC in patients with secondary AML compared with de novo AML (6.8 vs 14.8 months;
p < 0.05 [
52]; 8.6 vs 23.0 months;
p < 0.001 [
53]). Another trial performed in patients with AML-MRF reported a median OS of only 14 months in young patients (<60 years) treated with an IC regimen [
54]. The median OS of 13.2 months observed in our elderly AML-MRF patients (median age 77 years) indicates that these patients benefit from treatment with azacitidine (Fig.
2c).
We also present the first direct comparison of baseline factors, treatment-related factors and outcomes of patients with MDS-RAEB-I, MDS-RAEB-II, AML20–30 and AML30+ treated with azacitidine front-line (Tables
1,
2,
3,
4 and
5). Our data indicate that patients with AML20–30 have comparable baseline, treatment and response characteristics to patients with MDS-RAEB-II or AML30+ but behave more like AML30+ than MDS-RAEB-II with respect to OS (Table
4; Fig.
1a). This implies that the WHO reclassification of patients with 20–30 % bone marrow blasts from MDS-RAEB-t to AML seems appropriate in patients receiving azacitidine as front-line agent. Near identical observations have been made in patients treated with decitabine in a pooled sub-analysis of two clinical trials, which demonstrated significantly shorter median OS in patients with AML20–30 compared with patients with higher-risk MDS (9.0 vs 16.6 months;
p = 0.021), respectively [
34].
The effect of time from diagnosis to treatment (TDT) on overall survival of patients with MDS and AML remains obscure. In high-risk MDS patients including RAEB-t (treated with chemotherapy, azacitidine, dectiabine, lenalidomide or others), median TDT varied from 4.8 months [
55] to >1 year [
15], whereas separate analyses of RAEB-t and RA/RARS by others revealed a significantly shorter median TDT for RAEB-t (7.3 vs. 18.3 months,
p = 0.021) [
34], and others found a significantly shorter TDT for those MDS patients that eventually transformed to AML (10.8 months) as compared with those who did not (20.8 months) [
35]. These reports do not allow conclusions as to why shorter TDT have been observed in patients with higher-risk MDS and RAEB-t. One might speculate, however, that these patients are perceived as being in more dire need of treatment (due to higher bone marrow blasts, worse cytopenias and/or adverse cytogenetics) and are more likely to receive treatment soon after initial diagnosis. This is backed up by our own observations, which show a progressively shorter time from initial cytopenias (4.2, 2.8, 1.6 and 0.9 months) as well as initial diagnosis (3.0, 1.6, 0.6 and 0.5 months) to azacitidine treatment start for patients with RAEB-I, RAEB-II, AML20–30 and AML30+, respectively (Additional file
3: Table S3). Whether earlier treatment initiation in higher-risk MDS and AML20–30 translates into earlier response, longer response duration, or possibly results in a survival advantage, remains unknown at this time point. Even in the event that a correlation between TDT and outcome could be shown, it would still need to be clarified, whether this might also reflect a more aggressive underlying biology and kinetics of the disease.
There are not many studies that address the topic of TDT in AML, but those that did were all performed exclusively in patients treated with intensive chemotherapy approaches, with (partially) controversial results [
53,
56‐
59]. Most results however, do indicate that longer TDT is associated with worse prognosis, i.e. lower response rates and shorter OS [
56‐
59], and it was concluded that initiating therapy as soon as possible after diagnosis might be a potential strategy to improve OS in AML patients [
59]. Whether this can be translated to patients treated with hypomethylating agents remains to be shown.
In this report, we have shown that patients with AML20–30 treated with azacitidine front-line should be regarded as ‘true AML’. In line with the above [
56‐
59], we believe that treatment should thus be initiated without delay.
For decades, it has remained controversial whether AML20–30 (formerly MDS-RAEB-t) potentially follows a more benign disease trajectory than AML30+ and whether consideration should be given to possible retention of MDS-RAEB-t as a separate disease entity distinct from MDS and AML [
7,
16‐
19,
22,
31,
33,
34,
60‐
69]. The fact that the term ‘RAEB-t’ is still used in very recent reports on the efficacy of decitabine in patients with RAEB-t [
34,
70] shows the actuality of this issue. In an effort to clarify this matter for elderly patients treated with azacitidine front-line, we reclassified our cohort according to the FAB classification. Patients with MDS-RAEB-t had significantly worse OS than patients with MDS-RAEB, but similar survival to patients with AML30+ (Table
6; Fig.
1c). We thus show that the FAB disease category MDS-RAEB-t does not adequately distinguish risk categories or behave as an entity distinct from both MDS and AML with regard to patient outcome. In contrast, and as mentioned above, the WHO classification of MDS and AML adequately distinguished patient categories with distinct outcomes (13.1 vs 18.9 months for AML20–30 vs MDS-RAEB-II,
p = 0.010; Table
4 and Fig.
1a). We conclude that the elimination of the FAB category RAEB-t, and its incorporation within the WHO categories MDS-RAEB-II and AML, seems justified in elderly AML patients treated with azacitidine front-line.
In our cohort, the distinction between MDS-RAEB-I or MDS-RAEB-II could not separate groups with differing OS (23.7 vs 18.9 months,
p = 0.302; Table
4 and Fig.
1a). Of note, the IPSS [
8] gave more weight to the bone marrow blast threshold between 10 and 11 % (+1.0 additional score points for 11–20 %) than to the threshold between 20 and 21 % (+0.5 additional score points). Similarly, the revised IPSS [
71] conceded 1 additional score point for patients with ≥11 % bone marrow blasts but gave no further weight to bone marrow blast percentages >11 % (0 additional score points). Thus, while we could not confirm the capability of the WHO classification of MDS-RAEB-I and MDS-RAEB-II for distinguishing differing risk categories, we could confirm the weighting for bone marrow blasts chosen by the IWG for the prognosis of MDS.
Acknowledgements
The authors would like to acknowledge and thank Lucinda Huxley from FireKite, an Ashfield company, part of UDG Healthcare plc, for proofreading as an English native speaker, as well as editing of the references and figures to the required journal format. She had no influence on planning the study, interpreting the data, the writing or content of the manuscript or the decision to submit. Editorial assistance was funded by Celgene.
The AAR is a Registry of the AGMT Study Group which served as the responsible sponsor and holds the full and exclusive rights to data. Financial support for the AGMT was provided by Celgene. Celgene had no role in study design, data collection, data analysis, data interpretation, writing of the manuscript or the decision to submit the manuscript for publication.
Competing interests
LP has been a consultant for Celgene, Bristol-Myers Squibb, and Novartis and reported receiving honoraria from Celgene, Bristol-Myers Squibb, Novartis and AOP Orphan Pharmaceuticals. SB has been a consultant for Celgene and Novartis, a member on the Board of Directors or advisory committees for Celgene and Novartis, and reported receiving research funding from Celgene and honoraria from AOP Orphan Pharmaceuticals, Celgene, Mundipharma and Novartis. RS has been a consultant and a member on the Board of Directors or advisory committees for Celgene and reported receiving research funding and honoraria from Celgene, Teva (Ratiopharm) and Novartis. MG has been a consultant for Mundipharma and reports receiving honoraria from Mundipharma and Pfizer and research funding from Pfizer. HS reported receiving research funding from Celgene and has been an advisory board member for Celgene. SMS has been a member of an advisory board for Celgene. AZ reports receiving honoraria from Celgene. MP has been a consultant for Celgene and Novartis and reported receiving honoraria received from Celgene, Novartis and Janssen-Cilag. AL has been a consultant for Celgene. KG has been a member of advisory boards for Celgene. DG is a Board member of the Austrian Society of Haematology and Oncology. WRS has been a consultant for Celgene. RG reported receiving honoraria from Bristol-Myers-Squibb, Cephalon, Amgen, Eisai, Mundipharma, Merck, Janssen-Cilag, Genentech, Novartis, AstraZeneca, Boehringer Ingelheim, Pfizer, Roche and Sanofi Aventis; research funding from Cephalon, Celgene, Amgen, Mundipharma, Genentech, Pfizer, GSK and Ratiopharm; and has been a consultant for Bristol-Myers-Squibb, Cephalon and Celgene. The other authors declare that they have no competing interests.
Authors’ contributions
LP wrote the manuscript, provided substantial contributions to the acquisition and interpretation of data, edited the manuscript, provided intellectual input and contributed patients; SB, RS, MG, HS, KS, JT, BH, SMS, AZ, AP, MP, EMA, AL, KG, DV, DG, WRS, SH, IMR, JA and RG contributed patients, provided substantial contributions to the acquisition and interpretation of data, edited the manuscript and provided intellectual input. All authors reviewed and approved the final manuscript.