Azacitidine improves outcome in higher-risk MDS patients with chromosome 7 abnormalities: a retrospective comparison of GESMD and GFM registries

Azacitidine improves outcome in higher-risk MDS patients

with chromosome 7 abnormalities: a retrospective comparison

of GESMD and GFM registries

Marı´a Dı´ez-Campelo,1,2

Jose I. Lorenzo,3Raphael Itzykson,4 Silvia M. Rojas,1Ce´line Berthon,5 Elisa Lun˜o,6Odile Beyne-Rauzy,7 Jaime Perez-Oteyza,8Norbert Vey,9 Joan Bargay,10Sophie Park,11 Teresa Cedena,12

Dominique Bordessoule,13

Juan A. Mun˜oz,14_{Emmanuel Gyan,}15 Esperanza Such,3Sorin Visanica,16 Fe´lix Lo´pez-Cadenas,1

Ste´phane de Botton,17 Jesu´s M. Herna´ndez-Rivas,1,2

Shanti Ame,18_{Aspasia Stamatoullas,}19 Jacques Delaunay,20_{Celia Salanoubat,}21 Franc¸oise Isnard,22Romain Guieze,23 Joan Pe´rez Guallar,24Llorenc Badiella,24 Guillermo Sanz,*,3Consuelo Can˜izo*,1,2 and Pierre Fenaux *,4

1_{Haematology, Hospital Universitario de} Sala-manca, SalaSala-manca, Spain,2_{Institute of} Biomedi-cal Research of Salamanca (IBSAL), Salamanca, Spain,3Haematology Department, Hospital Universitario y Polite´cnico La Fe, Valencia, Spain,4Service d’Hematologie, St Louis, assis-tance publique hoˆpitaux de Paris and Paris Diderot University,Paris, France,5Maladies du Sang, Hopital Huriez, CHRU de Lille, Lille, France,6_{Haematology, Hospital Universitario} Central de Asturias, Oviedo, Spain,7 Haematol-ogy, CHU Purpan, Toulouse, France,8 Haematol-ogy, Hospital Sanchinarro Norte, Madrid, Spain, 9_{Haematology department, Institut Paoli} Cal-mettes, Marseille, France,10_{Hospital Sont} Llat-zer, Palma de Mallorca, Spain,11_Haematology, Clinique Universitaire d´He´matologie, CHU de Grenoble, Grenoble, France,12Hospital 12 de Octubre, Madrid, Spain,13Unite´de Recherche Clinique, Service d’He´matologie Clinique et The´rapie Cellulaire, Limoges, France,14Hospital Puerta del Mar, Cadiz, Spain,15Service d 'héma-tologie et the´rapie cellulaire, CHRU de Tours, Team 3 UMR CNRS 7292, Universite´ Franc¸ois

Summary

Treatment with azacitidine (AZA) has been suggested to be of benefit for higher-risk myelodysplastic syndrome (HR-MDS) patients with chromo-some 7 abnormalities (Abn 7). This retrospective study of 235 HR-MDS

patients with Abn 7 treated with AZA (n= 115) versus best supportive care

(BSC; n = 120), assessed AZA treatment as a time-varying variable in

mul-tivariable analysis. A Cox Regression model with time-interaction terms of overall survival (OS) at different time points confirmed that, while chro-mosome 7 cytogenetic categories (complex karyotype [CK] versus non-CK) and International Prognostic Scoring System risk (high versus intermediate-2) retained poor prognosis over time, AZA treatment had a favourable impact on OS during the first 3 years of treatment compared to BSC (Hazard ratio [HR] 0
5 P < 0
001 at 1 year, 0
7 P = 0
019 at 2 years; 0
73

P = 0
029 at 3 years). This benefit was present in all chromosome 7

cate-gories, but tended to be greater in patients with CK (risk reduction of 82%, 68% and 53% at 1, 3 and 6 months in CK patients; 79% at 1 month

in non-CK patients, P < 0
05 for all). AZA also significantly improved

pro-gression-free survival (P< 0
01). This study confirms a time-dependent

benefit of AZA on outcome in patients with HR-MDS and cytogenetic abnormalities involving chromosome 7, especially for those with CK.

Keywords: azacitidine, chromosome 7 abnormalities, time-dependent

analysis, high risk MDS.

Rabelais, Tours, France,16_{He´matologie–} Hoˆpital Bon Secours, Metz-Thionville, Thionville, France,17_{Hematologie, Institut Gustave Roussy,} Villejuif, France,18Haematology Department, Hoˆpital Civil, Strasbourg, France,19Haematology Department, Centre Henri Becquerel, Rouen, France,20Haematology Department, CHU de Nantes, Nantes, France,21H^opital de Corbeil, Corbeil, France,22Department of Haematology, Assistance Publique-Hopitaux de Paris (AP-HP) Saint-Antoine, Universite Pierre et Marie Curie, Paris, France,23_{Hematologie, CHU Estaing,} Clermont-Ferrand, France and24_{Department of} application statistics, Universidad Autonoma de Barcelona, Barcelona, Spain

Received 29 October 2017; accepted for publication 29 January 2018

Correspondence: Mar
ıa D
ıez Campelo, Haematology department, University hospital o Salamanca, Paseo de San Vicente 82-182, 37007 Salamanca, Spain.

E-mail: mdiezcampelo@usal.es *Equally contributed as senior authors.

Myelodysplastic syndromes (MDS) are clonal haematopoietic stem cell disorders characterized by morphological dysplasia in the bone marrow, peripheral cytopenia and an increased risk of evolution into acute myeloid leukaemia (AML). Although the only curative approach for these patients is allogeneic haematopoietic stem cell transplantation (HSCT) (Santini et al, 2010; Greenberg et al, 2011), hypomethylating agents have recently been shown to modify the disease course(Gattermann et al, 2007; Fenaux et al, 2009; Stone, 2009).

Prognostication of patients with MDS has been largely based on cytogenetic features, first in the International Prog-nostic Scoring System (IPSS) score(Greenberg et al, 1997), which distinguished three cytogenetic risk categories, in which chromosome 7 abnormalities (Abn 7) were included in the high-risk category. More recently, in the revised IPSS classification (IPSS-R), data from larger series allowed better cytogenetic stratification, considering 17 specific chromoso-mal abnorchromoso-malities classified in five cytogenetic categories with clear prognostic impact(Sole et al, 2005; Greenberg et al, 2012; Schanz et al, 2012). In particular, this large series confirmed previous findings from the Spanish Grupo Espa~nol de S
ındromes Mielodispl
asicos (GESMD) coopera-tive group showing that patients with isolated monosomy 7 or isolated del(7q) have different prognosis(Cordoba et al, 2012).

Several relatively small reports have suggested that MDS carrying Abn 7, mainly isolated monosomy 7, may

specifically benefit from treatment with azacitidine (AZA) (Fenaux et al, 2009; Ravandi et al, 2009). This is particularly relevant due to the poor outcome of these patients with con-ventional chemotherapy (Fenaux et al, 2009; Ravandi et al, 2009; (Knipp et al, 2007) however; those results not con-firmed, especially for patients with complex karyotypes (Itzykson et al, 2011).

In the current study, we analysed the impact of treatment with AZA on outcomes [response rate, overall survival (OS) and progression-free survival (PFS)] in a large series of higher-risk MDS patients with Abn 7 receiving AZA or best supportive care (BSC).

Patients and methods Inclusion criteria

Inclusion criteria in this retrospective study were: (i) higher-risk MDS, according to the World Health Organization [WHO] 2008 criteria) (Swerdlow et al, 2008) with high or intermediate 2 (int-2) IPSS risk (Greenberg et al, 1997), (ii) prospectively enrolled in the French and Spanish MDS reg-istries (between 2005 and 2011) and with complete follow-up data available (iii) presence of an Abn 7 (with or without other chromosomal abnormalities). Patients who received intensive AML-type chemotherapy or allogeneic HSCT were excluded from this analysis. Eligible patients were further stratified into four subgroups according to the cytogenetic

abnormalities present: (i) monosomy 7, isolated or with another single abnormality (non-complex -7), (ii) del(7q), isolated or with another single abnormality (non-complex del(7q), (iii) 7p- isolated or with another single abnormality (non-complex 7p-), and (iv) complex karyotype (CK) i.e. at least 3 cytogenetic abnormalities including any Abn 7. Two groups were established for each cytogenetic category accord-ing to the treatment approach received: (i) The AZA group

(defined as patients who received at least one cycle of AZA in a clinical trial, compassionate use programme or following European Medicines Agency approval for higher-risk MDS) and (ii) The supportive care group (BSC, patients receiving only transfusions and antibiotics). Treatment decision (AZA or BSC) was made by the treating physician and was based on his judgment and drug availability at the time of treatment.

Table I. Baseline disease characteristics.

Characteristic

AZA n= 115

BSC

n= 120 P value

Age at diagnosis (years)

Mean 67
4 67
4 NS Median 69
7 73
8 <60 years, n (%) 29 (25%) 27 (23%) NS ≥60 years, n (%) 86 (75%) 92 (77%) Follow-up (months) Median (95% CI) 47
5 (24
19–122
9) 59
8 (15
5–NR) NS

Time to AZA onset (months)

Median (range) 2
_{03 (0–66
2) –} Sex Male 64 (55
7%) 80 (66
7%) NS Female 51 (44
3%) 40 (33
3%) WHO 2008 RARS/RCUD/RCMD/RAEB-1/others 40 (35%) 57 (52%) <0
015 RAEB-2/AML 75 (65%) 53 (48%) Unknown 10 Cytogenetics at diagnosis

Monosomy 7 (/+ other abn) 26 (22
8%) 29 (24%) NS

del (7q-) (/+ other abn) 16 (14%) 22 (18
3%)

del (7p-) 2 (1
8%) 2 (1
7%) Complex karyotype 70 (61
4%) 67 (56%) Complex karyotype No 44 (38
6%) 54 (45%) NS Yes 70 (61
4%) 66 (55%) IPSS* Int-2 52 (45%) 67 (56%) NS High 63 (55%) 53 (44%) Type of MDS N= 78 De novo 74 (64%) 70 (90%) <0
001 Secondary 41 (36%) 8 (10%) Haemoglobin level (g/l)* Median (range) 91 (61_–140) 89 (42_–150) NS Platelets (9109/l)* Median (range) 58 (5_–970) 70 (4_–794) NS % of PB blasts* Median (range) 1 (0–29) 0 (0–22) NS % of BM blasts* Median (range) 13 (0–29) 10 (0–29) NS

95% CI, 95% confidence interval; abn, abnormality; AML, acute myeloid leukaemia; AZA, azacitidine; BM, bone marrow;BSC, best supportive care; CMML, chronic myelomonocytic leukaemia; IPSS, international prognostic score system; MDS-U, myelodysplastic syndrome-unclassified; NR, not reached; NS, not significant; PB, peripheral blood; RAEB, refractory anaemia with excess blasts; RARS, refractory anaemia with ring sideroblasts; RCMD, refractory cytopenia with multilineage dysplasia; RCUD, refractory cytopenia with unilineage dysplasia; WHO, World Health Organization.

*At Date of AZA onset for AZA patients and diagnosis for supportive care patients. Numbers in bold are referred to statistical significant p values and HR.

Response to treatment was evaluated after 4–6 cycles of AZA and response was defined according to the International Working Group criteria 2006 for MDS (Cheson et al, 2006): complete remission (CR), partial remission (PR), stable dis-ease with haematological improvement (HI), stable disdis-ease without haematological improvement (SD) and progressive disease (PD). Red blood cell (RBC) transfusion indepen-dence, defined as 8 weeks or more without RBC transfusions, was also evaluated.

The different survival outcomes were defined as time from diagnosis of higher risk MDS (diagnosis) to the corresponding event: death for OS, death or PD (blast increase) for PFS.

Patients were included after informed consent in accor-dance with the Declaration of Helsinki. The institutional review board of the GESMD and Groupe Francophone des My
elodysplasies (GFM) approved the study.

Statistical Methods

The statistical analysis was performed using R v3.1.2 (Simon & Makuch, 1984; Luis Meira-Machado, 2011; R Development Core Team 2011; Therneau, 2014). For all statistical tests a nominal significance level of 5% (P< 0
05) was applied. No adjustment for multiple tests was performed. All data spread-sheets, analysis codes and outputs were electronically stored and archived.

The relationship between OS and potential explanatory variables (age, karyotype, IPSS, WHO, de novo MDS and treatment) was examined by means of bivariate analyses. Sur-vival functions for each explanatory variable and each group were obtained using the Kaplan-Meier estimator. A collinear-ity study excluded significant associations between treatment and potential risk factors including age, gender, karyotype, IPSS, WHO, karyotype and counts.

To apply a multivariate Cox proportional hazards regres-sion model, the hypothesis of proportional risks was evalu-ated inspecting the risk functions Schoenfeld residuals. The hypothesis of proportional risks between the patients treated with AZA and BSC was rejected. Thus, the effect of AZA treatment on OS had a time varying pattern, with a positive early effect that was reduced with longer follow-up. To cir-cumvent this methodological limitation of standard Cox models, multivariate analysis of OS was performed including interaction terms between treatment and time. In order to explore and represent the relationship between time and risk, the Aalen’s nonparametric estimator was also used and, to quantify the AZA treatment’s time-varying effects, a Cox model including relevant explanatory variables and time interaction terms was used. This model allowed estimating the benefit of AZA over BSC at different time points (full information in Appendix S1, Figures S1–S4). The potential interaction between AZA treatment and karyotype was also

Fig 1. Response patterns to azacitidine according to karyotype (P= ns, chi-square test). CK, complex karyotype; Marrow CR w HI, bone marrow complete remission with haematological improvement; Marrow CR w/o, bone marrow complete remission without haematological improvement; SD, stable disease; SD HI, stable disease with haematological improvement. [Colour figure can be viewed at wileyonlinelibrary.com]

explored. The same methodological approach was followed for the secondary outcome variable, PFS.

Results

Baseline patient characteristics

Main baseline characteristics for the patients are shown in Table I. Two hundred and thirty-five patients with Abn 7 were analysed, including 115 treated with AZA and 120 patients who received BSC (the control group).

Seventy-four (64%) of AZA patients had de novo MDS and 41 (36%) had therapy-related (secondary MDS), com-pared to 70 (90%) and 8 patients (10%) in the BSC group (P= 0
0001). According to WHO 2008 classification (Swerdlow et al, 2008), 65% in the AZA group and 48% in the BSC group had refractory anaemia with excess of blasts type 2 (RAEB-2) or secondary AML (AML with <30% of blasts) (P= 0
015). The AZA and BSC groups were well balanced for other baseline parameters, including age, gen-der, cytogenetic risk category and IPSS risk. In the AZA group, 55% of the patients were IPSS high-risk and 45% int-2-risk and 61% had CK, 23% non-complex -7, 14% complex del(7q), and only 2 patients (1
8%) had non-complex 7p-. Median follow-up time from diagnosis was 47
5 months (95% confidence interval [CI]: 24
2–122
9) in the AZA group and 59
8 months (95% CI: 15
5–not reached) in the BSC group (P= not significant [ns]). Med-ian time from diagnosis to AZA treatment was 2 months (range 0–66
2) (Table I).

Ninety-two patients (80%) received AZA according to the conventional 7 days every 28 days schedule whereas 20% received 5-day cycles. The median number of AZA cycles received was 5 (range, 1–32). Information on transfusion support at AZA onset was available in only 100 of 115 patients in the AZA group; 74% were RBC (57%) and/or platelet (18%) transfusion dependent (TD).

Outcome with AZA treatment

Twelve AZA patients were not evaluable for response accord-ing to IWG 2006 criteria because insufficient data was recov-ered. In the 103 patients evaluable for response to AZA, the overall response rate (ORR) was 37
9% (39/103), including 14
6% CR and 23
3% SD with HI. Among AZA non-respon-ders (62
1%), 27
1% had SD without HI, 23
3% PD, and 11
7% (n = 12) had early death (8, infection; 1, bleeding, 3, unknown cause). Eleven of 67 RBC-TD patients (16%) became RBC transfusion independent. Regarding the cytoge-netics, the ORR was 38
1% in patients with CK, 32% in patients with non-complex -7 and 46
2% in patients with non-complex del(7q) (P= ns for complex versus non-com-plex, chi-square test, Fig 1). Median response duration was 13
1 months (95% CI 8
8–16
4) and duration of response was significantly higher for non-CK patients as compared to

CK patients [15
7 months (95% CI 12
2–18
2) vs. 9
4 months (95% CI 8
2–10
8), P = 0
0022].

Interestingly, ORR was 37
5% in “de novo” and 38
4% in secondary MDS, respectively (P = ns).

Comparison between AZA treatment and BSC

Overall survival. Median OS was not statistically significantly different between the 3 categories: median OS 11
4 months for CK, 17
8 for non-complex -7 and 21
3 months for non-com-plex del(7q) patients, (P = 0
1 and P = 0
07; Fig 2). This result suggests a specific improved outcome, especially for patients with CK receiving AZA (median OS of CK patients treated with BSC was 5
5 months). No OS differences were also detected when comparing patients with de novo versus ondary MDS (14
6 vs. 11
1 months, for de novo versus sec-ondary MDS, respectively, P = 0
17). In univariate analysis, AZA treatment gave a borderline survival advantage over BSC by Kaplan-Meier estimators of the survival curves, with a med-ian of 14 vs. 8
5 months (log-rank test P = 0
063).

Time 0 250 500 750 1000 1250 1500 Estimated Sur viv al Probability 0·0 0·2 0·4 0·6 0·8 1·0 Karyotype 7− 7q− CK AZA 7q−: median OS 21·3 months 7−: median OS 17·8 months CK: median OS 11·4 months P = 0·076 vs. 7q− P = 0·145 vs. 7− Time 0 250 500 750 1000 1250 1500 Estimated Sur viv al Probability 0·0 0 ·2 0·4 0 ·6 0·8 1 ·0 Karyotype 7− 7q− CK BSC 7q−: median OS 19 months 7−: median OS 13·6 months CK: median OS 5·5 months P < 0·001 vs. 7q− P = 0·005 vs. 7− (A) (B)

Fig 2. OS with azacitidine treatment compared with BSC. When comparing each group of treatments independently, it was confirmed that CK patients treated with BSC had a significantly lower OS (me OS 5
5 of months) than those with -7 (13
6 months, P = 0
0054) and in del(7q) (19 months, P< 0
001). Nevertheless, median OS in the AZA-treated group was not statistically significantly different between the 3 subsets [median OS 11
4 months for CK versus 17
8 and 21
3 months for -7 and del(7q) patients, respectively (P = 0
1 and P= 0
07). This suggests a specific improved outcome, particu-larly for patients with CK receiving AZA. AZA, azacitidine; BSC, best supportive care; CK, complex karyotype; OS, overall survival.

The OS benefit with AZA was more pronounced during the first months of treatment and decreased over time (Fig 3A), with HRs of 0
1, 0
3, 0
5 and 0
8 at 1, 3, 6 and 12 months after treatment onset (P< 0
0001 for the first 6 months and P = 0
1 at 12 months). AZA improved OS in all chromosome 7 categories, 50% of risk reduction among CK patients, 45% among non-complex -7 and 30% among non-complex del(7q) at 6 months after treatment. Neverthe-less, we observed a trend towards a greater risk benefit among CK patients (with the lowest HR and where differ-ences between AZA and BSC retained statistical significance after 1 year after treatment) as compared to non-CK (with slightly higher HR and significant difference only during the first 6 months after treatment onset, Fig 3B).

Multivariable analysis of OS at different time points is detailed in Tables II and III. AZA treatment had a significantly favourable impact on OS during the first 3 years of treatment as compared to BSC, confirming the trend observed in uni-variable analysis. This benefit was confirmed for the 3 cytoge-netic categories with no differences between them (P= 0
925). Nevertheless, the benefit of AZA treatment, compared to BSC, decreased over time, with an increase in HR from 0
3 at 6 months, 0
5 at 1 year and 0
7 at 2 and 3 years.

PFS. Univariate analysis demonstrated significantly better PFS in the AZA group compared to BSC (Figure S5). Multi-variate analysis confirmed an increase in PFS among patients receiving AZA (Table IV) and again it was noted when time-interaction terms were included in the final model: treatment

with AZA significantly decreased the risk of death or pro-gression (HR 0
01, P < 0
01). As for OS analysis, the benefit of AZA treatment decreased after the first year of treatment, from a HR of 0
4 (P < 0
001) and 0
6 (P = 0
004) at 6 and 12 months, to a HR of 0
79 (P = 0
107) at 24 months, respectively (Table IV and Table SI). The benefit of AZA treatment in terms of PFS was similar in the 3 cytogenetic categories (P= 0
794).

Discussion

In this large retrospective cohort of higher-risk MDS patients with Abn 7, we demonstrated a clear advantage in OS and PFS for patients treated with AZA compared with those receiving BSC, especially in patients with CK.

Information suggesting a particular efficacy of AZA in patients with Abn 7 was based on low patient numbers (Fenaux et al, 2009; Itzykson et al, 2011; Komrokji et al, 2013). Additionally, apart from the AZA-001 trial (Fenaux et al, 2009), there are no data comparing long-term outcomes after AZA and BSC. In the AZA-001 randomized clinical trial, AZA showed an OS benefit over conventional care in higher-risk MDS patients with Abn 7 (Fenaux et al, 2009), but only 30 patients in this category had received AZA. Several studies demonstrated that chemotherapy not a reliable option for patients with higher-risk MDS and AML with CK because of the lower complete response rates (46% as compared to 70% for those with no CK) and the poorer OS (median of 4 months as compared to 18 months for those with normal karyotype)

Time (A) (B) Start 1 month 3 months 6 months 1 Year HR 0·020 0·168 0·336 0·520 0·811 95% CI (0·002,0·183) (0·067,0·422) (0·193,0·587) (0·345,0·785) (0·527,1·251) P−value <0·0001 <0·0001 <0·0001 <0·0001 0·1031 0·1 0 0·5 1 1·5 HR AZA versus BSC Time Start 1 month 3 months 6 months 1 Year 0 0·5 1 1·5 2 HR AZA versus BSC Non−CK CK

Fig 3. (A) OS improvement with azacitidine treatment compared with BSC at different time points. (B) OS improvement with azacitidine treatment compared with BSC at different time points according to cytogenetics. 95% CI, 95% confidence interval; AZA, azacitidine; BSC, best supportive care; CK, complex karyotype; HR, hazard ratio; OS, overall survival.

(Knipp et al, 2007). In the AZA-001 trial, only 25 patients were candidates for induction and no significant differences were observed due to the low number of patients included, and is why the present study focused on BSC patients.

On average, our patients had higher risk features than pre-vious series of higher-risk MDS patients: Cytogenetic were poor-risk in all cases [compared with 47% and 28% in the GFM study (Itzykson et al, 2011) and AZA-001 trial (Fenaux et al, 2009), respectively] and 36% of our patients had sec-ondary MDS (versus 26% and 0% in the GFM study and AZA-001 trial). In addition, 65% of the patients in our AZA group had>10% bone marrow blasts versus 48% of the BSC group and 36% had secondary MDS (versus 10% in the BSC group). With those poor risk factors, the ORR in our series (37
8%) was somewhat lower than in the GFM retrospective study (Itzykson et al, 2011) (43%) and the prospective AZA-001 clinical trial (49%) (Fenaux & Ades, 2009; Fenaux et al, 2009) while OS was only slightly shorter in our cohort (me-dian, 14 months for AZA recipients) than in other unselected series of higher risk MDS patients (median, 13–24 months) (Fenaux & Ades, 2009; Fenaux et al, 2009; Itzykson et al, 2011; Bernal et al, 2015; Dinmohamed et al, 2015). Thirty six percent of our patient population had secondary MDS, but their ORR was similar to that of de novo ORR (38
5%

and 37
5%, P = ns, respectively), in agreement with previ-ously published data (Muller-Thomas et al, 2014). Thus, it appears that the effectiveness of AZA regarding the response rate in secondary MDS with Abn 7 is comparable to that seen in de novo MDS with these chromosomal abnormalities. We found no clear differences in ORR in AZA treatment between the three different categories defined in this report (non-complex -7, non-complex del 7q and CK involving chromosome 7), although there was a trend towards better response in non-complex del(7q) patients (46
1% ORR). This absence of clear differences in ORR between patients with non-complex and CK had been previously reported (Itzykson et al, 2011). For OS, a trend towards better out-come was seen in del(7q) patients (median OS of 19 months vs. 13
6 for -7 and 11
4 months for CK, respectively). Of note, median OS was 13
1 and 13
3 months in patients with -7/del(7q) in treated patients, respectively, in the AZA-001 clinical trial and GESMD retrospective studies (Fenaux & Ades, 2009). It is difficult to differentiate response and survival among patients with CK not involving Abn 7 because only a small number of patients with CK had no involvement of Chr 7. The conclusion of the recently pub-lished larger study that analysed karyotype and AZA treat-ment confirmed our results: baseline cytogenetic findings

Table II. Multivariate analysis of OS over time according to proportional hazards modelling. 6-month OS HR [95% CI] P 1-year OS HR [95% CI] P 2-year OS HR [95% CI] P 3-year OS HR [95% CI] P Age 1
02 [1
00;1
04] 0
066 1
02 [1
00;1
03] 0
032 1
01 [1
00;1
02] 0
095 1
01 [1
00;1
02] 0
075 Gender

Male Ref. 0
723 Ref. 0
371 Ref. 0
917 Ref. 0
822

Female 1
09 [0
66;1
80] 0
84 [0
58;1
22] 0
98 [0
72;1
34] 0
97 [0
72;1
3] MDS type

De novo Ref. 0
₃₀₉ Ref. 0
₆₄₇ Ref. 0
₉₄₈ Ref. 0
₈₃₉

Secondary 0
86 [0
44;1
69] 1
07 [0
68;1
68] 0
95 [0
65;1
40] 1
11 [0
77;1
58] Unknown 1
_{45 [0
82;2
55]} 1
_{23 [0
80;1
90]} 1
_{02 [0
70;1
49]} 1
_{07 [0
75;1
53]} WHO 2008

AML/RAEB-2 Ref. 0
069 Ref. 0
091 Ref. 0
101 Ref. 0
115

Others 0
55 [0
30;1
00] 0
66 [0
44;1
00] 0
70 [0
50;0
98] 0
72 [0
52;0
98] Unknown 1
25 [0
64;2
44] 1
09 [0
65;1
83] 0
95 [0
60;1
49] 0
93 [0
60;1
44] Karyotype

-7 Ref. 0
018 Ref. <0
001 Ref. <0
001 Ref. <0
001

del(7q) 0
_{48 [0
15;1
52]} 0
_{44 [0
20;0
98]} 0
_{76 [0
44;1
32]} 0
_{80 [0
48;1
31]} del(7p) 1
23 [0
16;9
50] 1
48 [0
35;6
28] 1
14 [0
27;4
75] 0
90 [0
22;3
75] Complex 1
85 [0
96;3
56] 1
87 [1
18;2
96] 2
07 [1
39;3
06] 2
02 [1
40;2
92] IPSS

Int-2 Ref. 0
001 Ref. <0
001 Ref. <0
001 Ref. <0
001

High 2
36 [1
40;3
98] 1
93 [1
34;2
77] 1
82 [1
35;2
47] 1
73 [1
30;2
31] Treatment

BSC Ref. <0
001 Ref. <0
001 Ref. 0
019 Ref. 0
029

AZA 0
31 [0
18;0
55] 0
51 [0
36;0
74] 0
70 [0
52;0
94] 0
73 [0
55;0
97]

95% CI, 95% confidence interval; AML, acute myeloid leukaemia; AZA, azacitidine; BSC, best supportive care; HR, hazard ratio; IPSS, interna-tional prognostic score system; MDS, myelodysplastic syndrome; ns, not significant; OS, overall survival; RAEB-2, refractory anaemia with excess blasts type 2; Ref. , reference; WHO, World Health Organization.

poorly predicted response to AZA but were strong predictors for OS (Sebert et al, 2017).

In the GFM study, patients with CK showed shorter response duration than non-CK patients (Itzykson et al, 2011). Our study also confirmed shorter response duration in those patient subsets compared to non-CK patients (9
4 vs. 15
7 months, respectively).

In the present series of MDS with chromosome 7 rearrange-ments, AZA improved OS compared to BSC. However, the improvement was only borderline in the univariate analysis, except in the first months of treatment. Multivariate analysis using time interaction effects better demonstrated this OS advantage, which however decreased during the second and third year of treatment, with a reduction in the risk of death of 70% in the first 6 months, 50% at 1 year and 30% at 2 and 3 years compared to BSC. This feature probably reflects the fact that responses induced by AZA are generally transient, with most patients losing their response after 6 months to 2 years, and confirming that AZA is not a curative approach.

We also analysed the interaction between karyotype and treatment and found no effect over the risk’s function, mean-ing that efficacy was similar in all subtypes. Nevertheless, a trend towards a larger benefit of AZA treatment over BSC dur-ing the first year was observed for CK, compared to non-CK patients. AZA-treated CK patients had a lower HR and more durable impact on OS compared to non-CK patients (risk

reduction of 82%, 68% and 53% at 1, 3 and 6 months in CK patients). Finally, PFS differences between AZA and BSC were similar to those found for OS, with a benefit over BSC during the first year that diminished with longer follow-up.

In the AZA-001 randomized clinical trial (Fenaux et al, 2009), the benefit of AZA over conventional care treatment (including BSC) in patients with Abn 7 was only seen when the karyotype was not complex (median OS of 24
5 vs. 8
1 for AZA versus conventional care treatment patients, respec-tively) while in CK patients the median OS was 5
3 vs. 3
9 months for AZA versus conventional care treatment patients). Our study, with a higher number of patients, demonstrated an OS benefit in patients with CK (median OS of 11
4 months vs. 5
5 months for BSC patients). A Dutch study (Dinmohamed et al, 2015) also found a global survival benefit with AZA treatment for patients with Abn 7 (median OS 21
4 months with Aza vs. 3
9 months for BSC; P= 0
01).

To summarize, the current study shows a survival benefit of AZA treatment over BSC in higher-risk MDS patients with Abn 7, especially for those with CK. However, the survival improvement in survival remains modest. New treatment Table III. Multivariable Cox Regression with time-interaction terms

for OS.

Cox model coefficient (standard error) HR Age 0
02 (0
01)* 1
020 Karyotype -7 Ref. 7q- -0
33 (0
29) 7p- -13
08 (447
1) Complex Karyotype 0
57 (0
22)* 1
7682 IPSS

Int-2 (Ref.) Ref.

High 0
51 (0
24)* 1
6652 Treatment BSC Ref. AZA -3
92 (1
14)** 0
0198 AZA*log(t) 0
63 (0
20)* 1
8776 Gender ns MDS type ns WHO 2008 ns Karyotype ns AIC 1216
01 1216
01 R2 _{0
27 0
27} Maximum R2 1
00 1
00 Events, n 147 147 Observed, n 172 172 Missing data 2 2

Proportional hazard test 0
₀₀ 0
₀₀ *P < 0
05, **P < 0
01.

Table IV. Multivariate Cox Regression with time-interaction terms for PFS. Cox model coefficient (standard error) HR Age ns Karyotype -7 Ref. 7q- ns 7p- ns Complex Karyotype 0
59 (0
22)** 1
8039 IPSS

Int-2 (Ref.) Ref.

High 0
57 (0
25)* 1
7682 Treatment BSC Ref. AZA -4
01 (1
12)** 0
0181 AZA*log(t) 0
65 (0
21)** 1
9155 Gender ns MDS type ns WHO 2008 ns Karyotype ns R2 _{0
29 0
29} Max. R2 1
00 1
00 Num. Events 152 152 Num. Observ. 172 172

AIC, Akaike information criterion; AZA, azacitidine; BSC, best sup-portive care; HR, hazard ratio; Int-2, intermediate-2; IPSS, interna-tional prognostic score system; MDS, myelodysplastic syndrome; ns, not significant; OS, overall survival; PFS, progression-free survival; Ref. , reference; WHO, World Health Organization.

strategies are needed for non-responders, and to delay relapse in responders.

Acknowledgments

The authors would like to thank all sub investigators of the GFM and GESMD for data collection, department of applica-tion statistics, Universidad Aut
onoma de Barcelona for statis-tical analysis and Celgene Spain for partially support the study. Authors from Servei d’Estad
ıstica Aplicada-UAB acknowledge support from the grant MTM2015-69493-R from the Spanish Ministry of Economy and Competitiveness, and financial support from FIS 2017, PI17/01741 and Con-tratos R
ıo Hortega CM17/00171.

Disclosures

Dr. D
ıez Campelo, Dr Itzykson, Dr Gyan, Dr de Botton, Dr Stamatoullas and Dr Sanz received honoraria from Celgene: Dr. D
ıez Campelo, Dr de Botton and Dr Sanz received hono-raria from Novartis; Dr de Botton receives honohono-raria from Agios, Pzifer, Servier and Pierre Fabre; Dr Guieze receives honoraria from Jannsen-Cilag and Abbie; Dr Sanz receives honoraria from Amgem and Roche.

Dr. D
ıez Campelo, Dr de Botton, P
erez Gallar, Dr Badiella, Dr Sanz and Dr Ca~nizo received compensation as consultants for Celgene: Dr. D
ıez Campelo, Dr Park, Dr de Botton, Dr Ame, Dr Delaunay and Dr Ca~nizo received com-pensation as consultants for Novartis; Dr. P
erez-Oteiza and Dr Sanz received compensation as consultants for Jannsen-Cilag Company; Dr de Botton received compensation as a consultant for Agios, Pzifer, Servier and Pierre Fabre; Dr Ame received compensation as a consultant for BMS; Dr Stamatoullas received compensation as a consultant for Takeda; Dr Sanz received compensation as a consultant for Amgem and Abbie; Dr Ca~nizo received compensation as a consultant for Astex.

Dr. D
ıez Campelo and Dr Ca~nizo receives compensation as a speakers’ bureau for Celgene and Novartis; Dr. P
erez-Oteiza and Dr Gyan, receives compensation as a speakers’ bureau for Janssen-Cilag Company.

The institutions of Dr. D
ıez Campelo, Dr P
erez-Oteiza, Dr Park, Dr Sanz and Dr Ca~nizo receive s research funding from Celgene Company; The institutions of Dr. D
ıez Campelo, Dr

Park and Dr Ca~nizo receive research funding from Novartis; The institutions of Dr. D
ıez Campelo, Dr Itzykson and Dr Ca~nizo receive compensation as research funding from Jans-sen; The institution of Dr. Park receives research funding from Pzifer; The institution of Dr. Gyan research funding from Mundipharma; The institution of Dr. Gyan receives research funding from Amgem; The institution of Dr. Gyan receives research funding from Roche.

Dr. Lorenzo, Dr Rojas, Dr Berthon, Dr Lu~no, Dr Beyne-Rauzy, Dr Vey, Dr Bargay, Dr Cedena, Dr Mu~noz, Dr Such, Dr Visanica, Dr L
opez-Cadenas, Dr Hernandez-Rivas, Dr Sal-anoubat, Dr Isnard and Dr Fenaux declared no conflict of interest.

Author contributions

Dr. D
ıez Campelo and Dr Ca~nizo conceptualised and designed the study. All authors participated in the collection and assembly of the data. Dr. D
ıez Campelo, Dr Fenaux, Dr Sanz, Dr Lorenzo, Dr Itkinson, Dr Badiella, Joan Gallar and Dr Ca~nizo reviewed the data analysis and interpretation. All authors gave final approval the manuscript and agree to be accountable for all aspects of the work.

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Appendix S1. Overall Survival analysis.

Fig S1. Univariate analysis for OS, impact of age, sex, karyotype and IPSS on OS.

Fig S2. Univariate analysis for OS, impact of treatment on OS.

Fig S3. Demonstration of the proportional risk property of the covariable treatment on survival by two methods. (A) Method 1, plotting the curve log(-log(Survival probability) versus log(survival time). (B) Method 2, Sconfeld0s residuals plot along with the parameter estimation over time.

Fig S4. Cumulative coefficient0s estimations for each covariable with 95% confident intervals.

Fig S5. ONLINE: PFS improvement with Aza treatment compared with BSC at different time points.

Table SI. PFS multivariate analysis over time according to proportional hazards modelling.

References

Bernal, T., Martinez-Camblor, P., Sanchez-Garcia, J., de Paz, R., Luno, E., Nomdedeu, B., Arda-naz, M.T., Pedro, C., Amigo, M.L., Xicoy, B., del Canizo, C., Tormo, M., Bargay, J., Val-carcel, D., Brunet, S., Benlloch, L. & Sanz, G. (2015) Effectiveness of azacitidine in unselected

high-risk myelodysplastic syndromes: results

from the Spanish registry. Leukemia, 29, 1875–

1881.

Cheson, B.D., Greenberg, P.L., Bennett, J.M., Lowenberg, B., Wijermans, P.W., Nimer, S.D., Pinto, A., Beran, M., de Witte, T.M., Stone, R.M., Mittelman, M., Sanz, G.F., Gore, S.D., Schiffer, C.A. & Kantarjian, H. (2006) Clinical application and proposal for modification of the International Working Group (IWG) response

criteria in myelodysplasia. Blood, 108, 419–425.

Cordoba, I., Gonzalez-Porras, J.R., Nomdedeu, B., Luno, E., de Paz, R., Such, E., Tormo, M., Valle-spi, T., Collado, R., Xicoy, B., Andreu, R.,

Munoz, J.A., Sole, F., Cervera, J. & del Canizo, C. (2012) Better prognosis for patients with del (7q) than for patients with monosomy 7 in myelodysplastic syndrome. Cancer, 118, 127– 133.

Dinmohamed, A.G., van Norden, Y., Visser, O., Posthuma, E.F., Huijgens, P.C., Sonneveld, P., van de Loosdrecht, A.A. & Jongen-Lavrencic, M. (2015) Effectiveness of azacitidine for the treat-ment of higher-risk myelodysplastic syndromes in daily practice: results from the Dutch

population-based PHAROS MDS registry.

Leu-kemia, 30, 1795–1796.

Fenaux, P. & Ades, L. (2009) Review of azacitidine trials in Intermediate-2-and High-risk myelodys-plastic syndromes. Leukemia Research, 33, S7– S11.

Fenaux, P., Mufti, G.J., Hellstrom-Lindberg, E., Santini, V., Finelli, C., Giagounidis, A., Schoch, R., Gattermann, N., Sanz, G., List, A., Gore, S.D., Seymour, J.F., Bennett, J.M., Byrd, J., Backstrom, J., Zimmerman, L., McKenzie, D., Beach, C. & Silverman, L.R. (2009) Efficacy of azacitidine compared with that of conventional care regimens in the treatment of higher-risk

myelodysplastic syndromes: a randomised,

open-label, phase III study. The Lancet Oncology,

10, 223–232.

Gattermann, N., K€undgen, A. & Germing, U.

(2007) Treatment of patients with high-risk myelodysplastic syndromes. Cancer Treatment

Reviews, 33, S64–S68.

Greenberg, P., Cox, C., Lebeau, M.M., Fenaux, P., Morel, P., Sanz, G., Sanz, M., Vallespi, T., Ham-blin, T., Oscier, D., Ohyashiki, K., Toyama, K., Aul, C., Mufti, G. & Bennett, J. (1997) Interna-tional scoring system for evaluating prognosis in myelodysplastic syndromes. Blood, 89, 2079– 2088.

Greenberg, P.L., Attar, E., Bennett, J.M., Bloomfield, C.D., de Castro, C.M., Deeg, H.J., Foran, J.M., Gaensler, K., Garcia-Manero, G., Gore, S.D., Head, D., Komrokji, R., Maness, L.J., Millenson, M., Nimer, S.D., O’Donnell, M.R., Schroeder, M.A., Shami, P.J., Stone, R.M., Thompson, J.E. & Westervelt, P. (2011) NCCN Clinical Practice Guidelines in Oncology: myelodysplastic syn-dromes. Journal of the National Comprehensive

Cancer Network, 9, 30–56.

Greenberg, P.L., Tuechler, H., Schanz, J., Sanz, G., Garcia-Manero, G., Sole, F., Bennett, J.M., Bowen, D., Fenaux, P., Dreyfus, F., Kantarjian, H., Kuendgen, A., Levis, A., Malcovati, L., Caz-zola, M., Cermak, J., Fonatsch, C., le Beau, M.M., Slovak, M.L., Krieger, O., Luebbert, M., Maciejewski, J., Magalhaes, S.M., Miyazaki, Y., Pfeilstocker, M., Sekeres, M., Sperr, W.R., Stau-der, R., Tauro, S., Valent, P., Vallespi, T., van de Loosdrecht, A.A., Germing, U. & Haase, D. (2012) Revised international prognostic scoring system for myelodysplastic syndromes. Blood,

120, 2454–2465.

Itzykson, R., Thepot, S., Quesnel, B., Dreyfus, F., Beyne-Rauzy, O., Turlure, P., Vey, N., Recher, C., Dartigeas, C., Legros, L., Delaunay, J., Sala-noubat, C., Visanica, S., Stamatoullas, A., Isnard, F., Marfaing-Koka, A., de Botton, S., Chel-ghoum, Y., Taksin, A.L., Plantier, I., Ame, S., Boehrer, S., Gardin, C., Beach, C.L., Ades, L. &

Fenaux, P. (2011) Prognostic factors for

response and overall survival in 282 patients

with higher-risk myelodysplastic syndromes

treated with azacitidine. Blood, 117, 403–411. Knipp, S., Hildebrand, B., Kundgen, A.,

Giagouni-dis, A., Kobbe, G., Haas, R., Aul, C., Gatter-mann, N. & Germing, U. (2007) Intensive chemotherapy is not recommended for patients

aged >60 years who have myelodysplastic

syn-dromes or acute myeloid leukemia with

high-risk karyotypes. Cancer, 110, 345–352.

Komrokji, R. S., Sekeres, M. A., Prebet, T., Clu-zeau, T., Santini, V., Sanna, A., Al Ali, N. H., Hobson, S., Eclache, V., List, A., Fenaux, P. & Ades, L. (2013) Impact of cytogenetics and cyto-genetic response on outcome in Myelodysplastic Syndromes (MDS) treated with Azacitidine (AZA). A collaborative study in 878 patients. Blood, 122, 389–389.

Luis Meira-Machado, J. R. P. (2011) Analyzing survival data from an illness-death model. Jour-nal of Statistical Software, 38, 1–18.

Muller-Thomas, C., Rudelius, M., Rondak, I.C., Haferlach, T., Schanz, J., Huberle, C., Schmidt, B., Blaser, R., Kremer, M., Peschel, C., Germing, U., Platzbecker, U. & Gotze, K. (2014) Response to azacitidine is independent of p53 expression in higher-risk myelodysplastic syndromes and secondary acute myeloid leukemia.

Haematolog-ica, 99, e179–e181.

R Development Core Team. (2011) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna. http://www.R-project.org

Ravandi, F., Issa, J.P., Garcia-Manero, G., O’Brien, S., Pierce, S., Shan, J., Borthakur, G., Verstovsek, S., Faderl, S., Cortes, J. & Kantarjian, H. (2009) Superior outcome with hypomethylating therapy in patients with acute myeloid leukemia and high-risk myelodysplastic syndrome and chro-mosome 5 and 7 abnormalities. Cancer, 115,

5746–5751.

Santini, V., Alessandrino, P.E., Angelucci, E., Bar-osi, G., Billio, A., di Maio, M., Finelli, C.,

Locatelli, F., Marchetti, M., Morra, E., Musto, P., Visani, G. & Tura, S. (2010) Clinical man-agement of myelodysplastic syndromes: update of SIE, SIES, GITMO practice guidelines. Leuke-mia Research, 34, 1576–1588.

Schanz, J., Tuchler, H., Sole, F., Mallo, M., Luno, E., Cervera, J., Granada, I., Hildebrandt, B., Slo-vak, M.L., Ohyashiki, K., Steidl, C., Fonatsch, C., Pfeilstocker, M., Nosslinger, T., Valent, P., Giagounidis, A., Aul, C., Lubbert, M., Stauder, R., Krieger, O., Garcia-Manero, G., Faderl, S., Pierce, S., le Beau, M.M., Bennett, J.M., Green-berg, P., Germing, U. & Haase, D. (2012) New comprehensive cytogenetic scoring system for primary myelodysplastic syndromes (MDS) and oligoblastic acute myeloid leukemia after MDS derived from an international database merge.

Journal of Clinical Oncology, 30, 820–829.

Sebert, M., Komrokji, R.S., Sekeres, M.A., Prebet, T., Cluzeau, T., Santini, V., Gyan, E., Sanna, A., Ali, N.H., Hobson, S., Eclache, V., List, A., Fenaux, P. & Ades, L. (2017) Impact of baseline cytogenetic findings and cytogenetic response on outcome of high-risk myelodysplastic syndromes and low blast count AML treated with azaci-tidine. Leukemia Research, 63, 72–77.

Simon, R. & Makuch, R.W. (1984) A non-para-metric graphical representation of the relation-ship between survival and the occurrence of an event: application to responder versus

non-responder bias. Statistics in Medicine, 3, 35–44.

Sole, F., Luno, E., Sanzo, C., Espinet, B., Sanz, G.F., Cervera, J., Calasanz, M.J., Cigudosa, J.C., Milla, F., Ribera, J.M., Bureo, E., Marquez, M.L., Arranz, E. & Florensa, L. (2005) Identifi-cation of novel cytogenetic markers with prog-nostic significance in a series of 968 patients with primary myelodysplastic syndromes.

Hae-matologica, 90, 1168–1178.

Stone, R.M. (2009) How I treat patients with myelodysplastic syndromes. Blood, 113, 6296– 6303.

Swerdlow, S. H., Campo, E., Harris, N.L., Jaffe, E.S., Pileri, S.A., Stein, H., Thiele, J. &

Vardi-man, J.W. (2008) WHO Classification of

Tumours of Haematopoietic and Lymphoid Tis-sues, 4th edn. International Agency for Research on Cancer Press, Lyon, France.

Therneau, T. (2014) A Package for Survival Analy-sis in S. R package version 2.37-7. http://CRAN. R-project.org/package=survival.