A natural history of lower-risk myelodysplastic syndromes with ring sideroblasts: an analysis of the MDS-CAN registry

Abstract

Patients with lower-risk (LR) myelodysplastic syndromes (MDS) with ring sideroblasts (RS) have better prognosis than those without RS, but how they fare over time is not fully understood. This study’s objective was to assess the natural history of LR MDS with RS ≥5% using MDS-CAN registry individual data. Kaplan–Meier estimates and generalized linear mixed models were used to describe time-to-event outcomes and continuous outcomes, respectively. One hundred and thirty-eight patients were enrolled; median times from diagnosis to enrollment and follow-up were 6.6 and 39.6 months, respectively. Within 5 years of enrollment, 65% of patients had ≥1 red blood cell transfusion dependence episode. Within 5 years of diagnosis, 59% developed iron overload, 38% received iron chelation therapy, 14% progressed to acute myeloid leukemia, and 42% died. Patients exhibited inferior health-related quality of life trends. These first real-world data in LR MDS-RS in Canada indicate a high level of morbidity and mortality over a 5-year period. Clinical Trial Registration: ClinicalTrials.gov Identifier: NCT02537990

Keywords:

• Lower-risk myelodysplastic syndromes
• ring sideroblasts
• natural history
• overall survival
• Canadian MDS registry

Introduction

Myelodysplastic syndromes (MDS) comprise a diverse group of clonal hematopoietic malignancies characterized by morphologic dysplasia, progressive bone marrow failure, ineffective hematopoiesis, peripheral blood cytopenias, and cytogenetic and molecular abnormalities [1–3]. MDS is associated with significant morbidity and high mortality with variable risk of progression to acute myeloid leukemia (AML) [2,4]. Anemia is the predominant cytopenia and a known cardiovascular risk factor in MDS [5]. The estimated incidence of MDS is ∼15.5 new cases per 100 000 persons per year in Canada (based on 5900 new cases per year for a population of 38 million) [6].

The International Prognostic Scoring System (IPSS) and, more recently, the revised IPSS (IPSS-R) are used to predict prognosis in MDS. Based on these scoring systems, IPSS low/intermediate risk and IPSS-R scores of ≤3.5 are considered lower-risk (LR) MDS [7]. Patients with LR MDS have inferior overall survival (OS) and leukemia-free survival compared with the general population, with mean life-year loss estimated at 6.3 years [8]. Best supportive care such as red blood cell transfusions (RBCTs), erythropoiesis-stimulating agents (ESAs), platelet transfusions, and hematopoietic growth factors are focused on the improvement of cytopenias and are the cornerstones of therapy for LR MDS (non-del5q) patients [9].

MDS with ring sideroblasts (RS) (MDS-RS) is a subtype of MDS characterized by <5% bone marrow blasts and ≥15% RS (or ≥5% if cells carry a mutation in the SF3B1 gene) [10]. RS are formed by a pathologic accumulation of iron in the mitochondria [11]. Heterozygous loss of function mutations in the SF3B1 gene occurs in 80% of patients with MDS-RS with single-lineage dysplasia and 40% of patients with MDS-RS with multilineage dysplasia, with the SF3B1 mutant allele burden correlating with the percentage of bone-marrow RS [12]. These mutations may result in dysregulation of mitochondrial iron homeostasis resulting in the formation of RS and ineffective hematopoiesis [10,11]. A small series focused on MDS patients with RS from Germany, Italy, and the U.S.A. suggested that patients with MDS-RS are generally stratified into LR categories using the IPSS and IPSS-R [12–14]. Despite understanding the clinical presentations of patients with LR MDS and RS, it is unclear how the disease progresses over time, how patients are treated, how transfusion dependence evolves, and how patient quality of life changes over time. Therefore, the objective of this analysis was to assess the natural history in patients with LR MDS-RS using individual patient data (IPD) from the Canadian MDS (MDS-CAN) registry.

Methods

Study design, setting, and participants

This study was a non-comparative, prospective cohort analysis of the natural history of LR MDS-RS. IPD were obtained from the MDS-CAN registry (NCT02537990), which was initiated in August 2005 and is ongoing in 15 tertiary care centers across 8 of 10 provinces in Canada. Patients aged ≥18 years who were diagnosed with MDS, myeloproliferative neoplasms (MPN), chronic myelomonocytic leukemia (CMML), and low blast count AML (20–30% blasts) as defined by the World Health Organization (WHO) 2008 classification, were enrolled in the registry. Due to an absence of routinely available testing for SF3B1 at the time of patient enrollment, and the possibility of the presence of RS in other subtypes such as CMML, MDS/MPN, and MDS with excess blasts (MDS-EB), some of the included patients may not be assigned to MDS-RS as per the current classification [10]. Patients whose diagnostic bone marrow assessment occurred ≥2 years prior to signed consent, with AML and bone marrow blast ≥31% at the time of consent, or who previously received allogeneic stem cell transplantation were not eligible for enrollment.

Ethics

All patients provided written informed consent at enrollment and all 15 participating medical centers received Institutional Review Board approval for the registry study.

Data collection

Detailed demographic and disease characteristics, as well as retrospective diagnostic characteristics, were identified at enrollment. Retrospective diagnostic information including MDS diagnosis per the WHO 2008 classification, RS, bone marrow blasts, IPSS, and IPSS-R were captured. Treatments received [e.g. iron chelation therapy (ICT), ESAs, hypomethylating agents, and lenalidomide] and RBCTs were recorded from enrollment through end of follow-up. Dates of AML progression, death, and last contact were also captured by the registry. Laboratory tests (e.g. serum ferritin, hemoglobin) and health-related quality of life (HRQoL) outcomes (i.e. the EuroQol 5 dimensions [EQ-5D]) and Rockwood frailty were measured every 6 to 12 months starting at enrollment [15–17]. The MDS-CAN registry utilized an EQ-5D with 3 levels (EQ-5D-3L) instrument until December 2018, and the EQ-5D with 5 levels (EQ-5D-5L) instrument thereafter. Both instruments captured patient-reported responses to five attributes (mobility, usual activities, self-care, pain/discomfort, anxiety/depression) and overall health as measured on a visual analog scale.

Outcomes

Several time-to-event outcomes were used to characterize the natural history of LR MDS-RS. Clinical outcomes included OS, AML progression, time to disease risk (per IPSS-R) progression, time to iron overload, and time to first episode of RBCT dependence. IPSS-R progression was defined as an increase to the next IPSS-R category from a baseline IPSS-R of very low, low, or intermediate. Iron overload was defined as (a) first receipt of ICT or (b) serum ferritin ≥1000 ng/mL RBCT dependence was defined as two consecutive post-enrollment 8-week periods during which a patient received ≥1 RBCT. All outcomes were censored at date of last contact, and all event-free times except that of RBCT dependence were measured in years from diagnosis. Since detailed RBCTs were captured starting at enrollment, time to RBCT dependence was restricted to patients who were transfusion independent at diagnosis. Among patients who died during follow-up, cause of death was reported. In addition to these clinical outcomes, the EQ-5D-3L was used to capture HRQoL from enrollment through 5 years follow-up. Detailed definitions of all endpoints are summarized in Table S1 in the Supplementary Material (Appendix 1). As this was a descriptive study, missing outcome data were not imputed.

Statistical models

Time-to-event outcomes were summarized using Kaplan–Meier curves from the index date (typically the date of diagnosis) through 5 years follow-up. Kaplan–Meier estimates were used to derive cumulative probabilities of events at 1 to 5 years follow-up. The 95% confidence intervals (CIs) for all estimates were derived using Greenwood’s formula [18].

All components of the EQ-5D-3L index were estimated using generalized linear mixed models with specified random patient-level intercepts. Time trends were modeled using restricted cubic splines. Each of the five attributes of the EQ-5D-3L instrument were modeled using cumulative probit distributions. All EQ-5D-3L outcomes were fit using Bayesian methods. Point estimates for these outcomes were derived using the mean of the posterior distribution and 95% credible intervals (Crls) were derived using the 2.5th and 97.5th percentiles of the posterior distribution. A more detailed summary is given in the Supplementary Material (Appendix 2).

All analyses were conducted in R software [19]. Generalized linear mixed models were fit in R using the brms package [20].

Reporting

A Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement—which consists of a checklist of 22 items to improve the reporting, critical appraisal, and interpretation of observational research [21,22]—is reported in Table S2 in the Supplementary Material (Appendix 3).

Results

Study population

Between 1 January 2008 and 31 December 2019, a total of 1168 patients were enrolled in the MDS-CAN registry. Of this population, 138 patients diagnosed with LR MDS, defined as very low, low, or intermediate risk as per the IPSS-R, and RS ≥5% at diagnosis were included in this study (Figure 1). The median time from diagnosis to registry enrollment was 6.6 [inter-quartile range (IQR): 2.1–18.9] months, and median follow-up time from diagnosis was 39.6 (IQR: 24.7–79.1) months. At diagnosis, 92.8% (n = 128) of patients were transfusion independent, 63.8% were male, and the median age was 74 (IQR: 66–79) years. A total of 23.9% (n = 33), 48.6% (n = 67), and 27.5% (n = 38) of patients had IPSS-R scores of very low, low, and intermediate, respectively. Overall, 97.8% (n = 135) of this cohort had <5% blasts, and 74.6% (n = 103) had ≥15% RS. Additionally, a total of 36.2% (n = 50) of patients had initiated ESAs prior to registry enrollment (Table 1).

Figure 1. Cohort attrition diagram. Analysis of IPSS progression excluded 14 patients with <2 IPSS records and patients whose baseline IPSS was high. Analysis of HRQoL excluded 24 patients without complete data on at least one EQ-5D-3L assessment. int-2: Intermediate-2; IPSS: International Prognostic Scoring System; IPSS-R: Revised IPSS; LR MDS: lower-risk myelodysplastic syndromes; EQ-5D-3L: EuroQoL 5 dimensions, 3 levels; HRQoL: health-related quality of life; MDS-CAN: MDS Canadian registry; n: number of cases; N: sample size; RS: ring sideroblasts.

Table 1. Demographic and clinical characteristics at diagnosis.

Characteristic	Primary analysis	Sensitivity analysis
	LR MDS with RS ≥ 5%	LR MDS with RS ≥ 15% and exclusion of CMML diagnoses
Sample size, N	138	99
Time to enrollment, median (IQR), months	6.6 (2.1–18.9)	6.6 (2.1–18.9)
Transfusion dependence, n (%)
Yes	10 (7.2)	6 (6.1)
No	128 (92.8)	93 (93.9)
Blasts, n (%)
Absent or <3	88 (63.8)	61 (61.6)
3–4	47 (34.1)	36 (36.4)
5–9	2 (1.4)	1 (1.0)
≥10	1 (0.7)	1 (1.0)
Sideroblasts, n (%)
Sideroblasts <15	35 (25.4)	0 (0.0)
Sideroblasts ≥15	103 (74.6)	99 (100.0)
IPSS at diagnosis, n (%)
Low	92 (66.7)	72 (72.7)
Int-1	46 (33.3)	27 (27.3)
IPSS-R at diagnosis, n (%)
Very low	33 (23.9)	25 (25.3)
Low	67 (48.6)	47 (47.5)
Intermediate	38 (27.5)	27 (27.3)
WHO classification at diagnosis, n (%)
MDS-RS-MLD	23 (16.7)	23 (23.2)
MDS/MPN-RS-T	8 (5.8)	7 (7.1)
MDS-RS-SLD	56 (40.6)	53 (53.5)
MDS-MLD	21 (15.2)	8 (8.1)
MDS-SLD	10 (7.2)	3 (3.0)
CMML-0/CMML-1	8 (5.8)	0 (0.0)
MDS-U	4 (2.9)	0 (0.0)
MDS-5q	3 (2.2)	1 (1.0)
MDS-EB1	2 (1.4)	1 (1.0)
MDS/MPN	1 (0.7)	1 (1.0)
MDS-T	1 (0.7)	1 (1.0)
MDS/MPN-T	1 (0.7)	1 (1.0)
Sex, n (%)
Female	49 (35.5)	40 (40.4)
Male	88 (63.8)	58 (58.6)
Missing	1 (0.7)	1 (1.0)
Age at diagnosis, years
Mean (SD)	71.8 (10.8)	71.2 (11.0)
Median (IQR)	74 (66.0–79.0)	72.0 (65.5–78.5)
Diagnosis type, n (%)
Primary	125 (90.6)	91 (91.9)
Secondary	13 (9.4)	8 (8.1)
ECOG PS
0	55 (39.9)	41 (41.4)
1	67 (48.6)	45 (45.5)
2	15 (10.9)	12 (12.1)
3	1 (0.7)	1 (1.0)
Initiated ESAs* prior to enrollment
Yes	50 (36.2)	39 (39.4)
No	88 (63.8)	60 (60.6)

*Epoetin alfa (EPREX^®) or darbepoetin alfa (Aranesp^®).

CMML: chronic myelomonocytic leukemia; ECOG PS: Eastern Cooperative Oncology Group performance status; Int-1: Intermediate-1; IPSS: International Prognostic Scoring System; IPSS-R: Revised IPSS; IQR: inter-quartile range; LR MDS: lower-risk myelodysplastic syndromes; MDS-5q: MDS with isolated del(5q); MDS-EB1: MDS with excess blasts; MDS-MLD: MDS with multilineage dysplasia; MDS/MPN-RS-T: MDS/MPN with ring sideroblasts and thrombocytosis; MDS/MPN-T: MDS/MPN with thrombocytosis; MDS-RS-MLD: MDS with ring sideroblasts and multilineage dysplasia; MDS-RS-SLD: MDS with ring sideroblasts and single-lineage dysplasia; MDS-SLD: MDS with single-lineage dysplasia; MDS-T: MDS with thrombocytosis; MDS-U: MDS unclassifiable; MPN: myeloproliferative neoplasm; n: number of observations; N: sample size; RS: ring sideroblasts; SD: standard deviation; WHO: World Health Organization.

Overall survival and disease progression

Overall, 67 patients died after study enrollment, 63 patients were alive and still participating in the registry, and 8 patients withdrew from the study. Among patients who died after study enrollment, 49 (73%) had died within 5 years of diagnosis. The Kaplan–Meier estimate of median OS from diagnosis was 6.7 (95% CI: 4.6–8.3) years (Figure 2A). The cumulative mortality rate was 8% (95% CI: 3–13%) 1 year after diagnosis, and 42% (95% CI: 32–51%) 5 years after diagnosis (Table 2). Cause of death was reported for 41 of 49 patients who died within 5 years of diagnosis. Of these deaths, 76% (n = 31) were attributable to MDS-related causes and 24% (n = 10) were due to other causes (i.e. cardiovascular complication, other cancers). The leading causes of MDS-related deaths were classified as disease progression (n = 11), AML (n = 9), infection (n = 7), and bleeding (n = 4, Table 3).

Figure 2. Kaplan–Meier curves for time-to-event outcomes for (A) OS, (B) AML progression, (C) IPSS-R progression, (D) cardiovascular complication, (E) iron overload (serum ferritin ≥1000 ng/mL), (F) iron overload (receipt of ICT), and (G) RBCT dependence for the primary cohort (Panel 1) and after excluding patients with ring sideroblasts <15% or CMML diagnosis (Panel 2). All time to event outcomes except for RBCT dependence were indexed to the date of diagnosis. RBCTs were recorded systematically only after enrollment, so RBCT dependence was indexed to the date of enrollment. OS was censored at the date of last follow-up. All other outcomes were censored at the earliest of the date of last follow-up or date of death. Since ascertainment of RBCT dependence required a minimum of 16 weeks follow-up, patients who were lost to follow-up in <16 months of enrollment were censored rather than excluded from the analysis. AML: acute myeloid leukemia; ICT: iron chelation therapy; IPSS-R: Revised International Prognostic Scoring System; MDS-RS+: myelodysplastic syndromes with RS; OS: overall survival; RBCT: red blood cell transfusion; RS: ring sideroblasts.

Table 2. Kaplan–Meier estimates of cumulative probabilities of events at 1 to 5 years follow-up.

	Mortality		AML progression		IPSS-R progression		Cardiovascular complication		Iron overload Serum ferritin ≥ 1000 ng/mL		Iron overload ICT		RBCT dependence
Follow-up (years)	p (95% CI)	n	p (95% CI)	n	p (95% CI)	N	p (95% CI)	n	p (95% CI)	n	p (95% CI)	n	p (95% CI)	n
Primary analysis of patients with LR MDS with RS ≥5%
1	0.08 (0.03–0.13)	123	0.04 (0.00–0.07)	120	0.04 (0.00–0.07)	119	0.03 (0.00–0.06)	122	0.21 (0.14–0.27)	99	0.08 (0.03–0.13)	112	0.24 (0.16–0.31)	93
2	0.16 (0.09–0.22)	105	0.06 (0.02–0.11)	103	0.11 (0.05–0.16)	97	0.06 (0.02–0.11)	102	0.34 (0.25–0.41)	72	0.15 (0.08–0.21)	89	0.39 (0.29–0.47)	68
3	0.32 (0.23–0.40)	72	0.08 (0.03–0.12)	71	0.17 (0.10–0.24)	66	0.11 (0.05–0.16)	72	0.42 (0.32–0.50)	49	0.21 (0.13–0.28)	59	0.44 (0.34–0.52)	50
4	0.36 (0.27–0.45)	58	0.10 (0.04–0.16)	56	0.22 (0.13–0.30)	51	0.12 (0.05–0.18)	57	0.50 (0.39–0.59)	36	0.27 (0.17–0.35)	44	0.58 (0.47–0.67)	34
5	0.42 (0.32–0.51)	49	0.14 (0.06–0.21)	49	0.25 (0.15–0.33)	44	0.12 (0.05–0.18)	48	0.59 (0.47–0.68)	26	0.38 (0.26–0.48)	33	0.65 (0.53–0.73)	25
Analysis results after excluding patients with RS <15% or CMML diagnosis
1	0.03 (0.00–0.06)	93	0.00 (0.00–0.00)	93	0.01 (0.00–0.03)	92	0.02 (0.00–0.05)	92	0.15 (0.08–0.22)	78	0.10 (0.04–0.16)	83	0.18 (0.09–0.25)	73
2	0.10 (0.03–0.15)	79	0.02 (0.00–0.06)	78	0.06 (0.01–0.10)	75	0.07 (0.01–0.12)	76	0.32 (0.21–0.41)	53	0.17 (0.09–0.25)	65	0.33 (0.23–0.43)	52
3	0.27 (0.17–0.36)	55	0.02 (0.00–0.06)	55	0.15 (0.06–0.22)	49	0.09 (0.03–0.15)	55	0.42 (0.30–0.52)	37	0.25 (0.16–0.34)	44	0.39 (0.27–0.49)	38
4	0.30 (0.19–0.39)	44	0.06 (0.00–0.12)	43	0.19 (0.09–0.27)	38	0.11 (0.04–0.18)	43	0.49 (0.36–0.59)	27	0.33 (0.21–0.43)	32	0.56 (0.43–0.66)	25
5	0.36 (0.24–0.46)	37	0.10 (0.02–0.18)	37	0.23 (0.11–0.33)	32	0.11 (0.04–0.18)	36	0.61 (0.47–0.71)	19	0.44 (0.30–0.56)	24	0.62 (0.48–0.72)	19

AML: acute myeloid leukemia; CI: confidence interval; CMML: chronic myelomonocytic leukemia; ICT: iron chelation therapy; IPSS-R: Revised International Prognostic Scoring System; LR MDS: lower-risk myelodysplastic syndromes; n: number of patients at risk of event; p: probability of event (cumulative); RBCT: red blood cell transfusion; RS: ring sideroblasts.

Table 3. Distribution of causes of death in patients who died within 5 years of diagnosis.

	Primary analysis of patients with LR MDS with RS ≥5%		Analysis results after excluding patients with RS <15% or CMML diagnosis
Cause of death	All deaths, n (%)	Deaths with a known cause, n (%)	All deaths, n (%)	Deaths with a known cause, n (%)
All deaths within 5 years of diagnosis	49	41	29	23
Deaths related to MDS	31 (63)	31 (76)	16 (55)	16 (70)
Disease progression	11 (22)	11 (27)	6 (21)	6 (26)
AML	9 (18)	9 (22)	5 (17)	5 (22)
Infection	7 (14)	7 (17)	4 (14)	4 (17)
Bleeding	4 (8)	4 (10)	1 (3)	1 (4)
Other causes of death	10 (20)	10 (24)	7 (24)	7 (30)
Cardiac	6 (12)	6 (14)	4 (14)	4 (17)
Another cancer	1 (2)	1 (2)	0 (0)	0 (0)
Post-allogenic stem cell transplantation complications	1 (2)	1 (2)	1 (3)	1 (4)
Other, not elsewhere classified	2 (4)	2 (5)	2 (7)	2 (9)
Unknown	8 (16)	NA	6 (21)	NA

AML: acute myeloid leukemia; CMML: chronic myelomonocytic leukemia; LR MDS: lower-risk MDS; MDS: myelodysplastic syndromes; NA: not applicable; RS: ring sideroblasts.

Events of interest included increased IPSS risk score, development of AML, and cardiovascular complication. In this cohort, 14% (95% CI: 6–21%) of patients developed AML within 5 years of diagnosis. Over the same follow-up period, 12% (95% CI: 5–18%) of patients died or were hospitalized due to cardiovascular complication during follow-up, and 25% (95% CI: 15–33%) of patients experienced an increase in their IPSS-R disease risk (Table 2). The median time-to-event was not reached for any of these endpoints (Figure 2B–D).

Red blood cell transfusion dependence and iron overload

For this population of patients with LR MDS-RS, the cumulative probability of RBCT dependence was 24% (95% CI: 16–31%) within 1 year of diagnosis and 65% (95% CI: 53–73%) within 5 years of diagnosis (Table 2). The median time from diagnosis to RBCT dependence was 3.4 (95% CI: 2.9–4.4) years (Figure 2G). For this cohort, 59% (95% CI: 47–68%) had elevated serum ferritin within 5 years of follow-up (Figure 2E; Table 2). The cumulative probability of receiving ICTs for treatment of iron overload was 38% (95% CI: 26–48%) within 5 years of follow-up (Figure 2F; Table 2).

Health-related quality of life

Results for HRQoL, per the EQ-5D-3L instrument, are summarized graphically in Figure 3. Overall, the composite utility index worsened over time, falling from a baseline of 0.830 (95% CrI: 0.810–0.847) to 0.782 (95% CrI: 0.746–0.815) at 5 years follow-up. Approximately 35–50% of patients reported some problems with mobility or usual activities, moderate pain, or moderate anxiety or depression at enrollment. In contrast, 90% of this cohort reported having no problems with self-care at baseline. Consistent with results for the composite utility index, the proportion of patients reporting some problems increased over follow-up for each component of the EQ-5D-3L. Across domains, the largest decrement in HRQoL was observed for pain/discomfort, with the proportion of patients reporting moderate pain/discomfort increasing from 41% (95% CrI: 36–47%) at baseline to 57% (95% CrI: 49–66%) at 5 years follow-up. Detailed numeric results for HRQoL are reported in Table S3 in the Supplementary Material (Appendix 4).

Figure 3. Predicted time trends in the mean of the EQ-5D-3L (A) utiliity index and EQ-5D-3L item response probabilities ((B) usual activites, (C) pain/discomfort, (D) mobility, (E) self-care, (F) anxiety/depression). The analysis sample for health-related quality of life outcomes included 120 patients. CI: confidence interval; EQ-5D-3L: EuroQoL 5 dimensions, 3 levels.

Discussion

The objective of this study was to characterize the natural history of LR MDS-RS using the MDS-CAN database, focusing on survival, transfusion dependence, iron overload, and HRQoL. The MDS-CAN is a prospective clinical registry which has been ongoing since 2005 and represents a mature and rich ‘real-world’ data source on the treatments and clinical outcomes in patients with MDS.

For the cohort examined in this study, the median OS was 6.7 years and the cumulative probability of developing AML within 5 years of diagnosis was 14% (the median time to AML progression was not reached during follow-up). The median OS for this population exceeded estimates for LR MDS patients in the European MDS registry (4.8 years), and for patients with refractory anemia and RS (RARS) in the Dutch MDS registry (4.6 years) and in the US National Cancer Database (NCDB, 2.8 years) [23–25]. The longer OS estimated in this study compared with prior estimates for RARS from the US and the Netherlands is likely due in part to differences in age at diagnosis, which was slightly higher in the US study [median age: 76 (NCDB) versus 74 (MDS-CAN) years], and the absence of IPSS-R risk category in both the US and Dutch cohorts, which defined RARS according to WHO 2008 criteria exclusively.

The 5-year cumulative probability of AML progression estimated for the MDS-CAN cohort (14%), in which approximately two-thirds had bone marrow blasts of 0–2%, was lower than for MDS patients from the Düsseldorf Registry with medullary blasts of 0–2% (19%) and 3–4% (31%) [26]. The cumulative probability of IPSS progression was similar to AML progression for the cohort investigated in this study. This study also documented a consistent worsening of health-related utility throughout follow-up and a notable increase in reporting of moderate pain or discomfort. Increased pain for this population could reflect either both disease progression or an increase in the prevalence of ageing-related comorbidities that manifest in pain [27–29].

A key strength of this study was the use of multiple endpoints to characterize the burden of illness in LR MDS-RS. In addition to OS, AML-free survival, and IPSS progression, cardiovascular complications, RBCT dependence, iron overload, and HRQoL were examined. The cumulative probability of death due to cardiac complication was 12% within 5 years of enrollment. However, this represents a rather extreme episode of cardiac impairment that likely understates the true risk of cardiovascular comorbidity in this population. Among patients with a diagnosis of LR MDS in the Surveillance, Epidemiology, and End Results (SEER) database of the National Cancer Institute, 29.9% of the known causes of death were attributed to cardiovascular disease within 5 years of diagnosis [30]. Overall, 43% (n = 52) of the MDS-CAN cohort considered in the present study had a cardiovascular-related death, a documented cardiovascular comorbidity, or was hospitalized for a cardiovascular-related complaint within 3 months of a scheduled 6-month assessment (data not shown), thus suggesting a higher cardiovascular burden than implied by cardiac-related deaths.

RBCTs were commonly used as supportive therapy. The cumulative probability of developing RBCT dependence within 5 years of diagnosis was 65%, with a median time from diagnosis to dependence of 3.4 years. The high incidence of RBCT dependence in this population corresponded with high levels of iron overload over the same follow-up period. Notably, the 5-year cumulative probability of ≥1 episode of serum ferritin ≥1000 ng/mL and of receiving ICT were 59% and 38%, respectively.

We recognize several potential limitations associated with this study. Despite using a large national registry to inform the analysis, only 138 patients were identified with LR MDS and RS ≥5%. This may limit the precision and external generalizability of results, particularly over longer follow-up times. Despite this, most Kaplan–Meier estimates were based on risk sets of >30 patients even after 5 years of follow-up. In addition, patients were enrolled in tertiary care centers, which may bias the study cohort in unknown ways. For example, if patients were selected based on their ability to attend regular appointments and adhere to treatments, then population-level clinical outcomes may be more favorable in this registry than for the general population of patients with LR MDS-RS. Alternatively, patients with better performance status, fewer comorbidities, or less need for transfusion may not be referred to tertiary care and would therefore not be captured in this study. We included four patients with RS <15% which, in the absence of SF3B1 testing available, may not technically be assigned to MDS-RS in the current classification [10]. A subgroup analysis conducted in patients with RS ≥15% and with WHO diagnoses that excluded CMML (Tables 1–3, Figure 2, and Table S4 in the Supplementary Material, Appendix 5) produced slightly lower cumulative mortality and AML progression rates through 5 years follow-up OS (Table 2 and Figure 2). There were also some detailed demographic and diagnostic variables measured at diagnosis missing from the dataset (particularly SF3B1 testing, which may provide greater prognostic value than RS [31]), thereby limiting our ability to conduct multivariable adjustment and stratification of natural disease endpoints. Another limitation is that reporting of diagnostic information was not reviewed by central pathology. However, all patients were followed by MDS experts at academic centers, and pathology reviews would have been done when needed to confirm diagnosis. Despite these limitations, this is the first study to comprehensively evaluate the burden of illness in LR MDS-RS using real-world data in Canada.

In conclusion, this study used MDS-CAN registry data to determine manifestations of MDS progression in patients with LR MDS-RS. Almost half of this cohort died within 5 years of MDS diagnosis, ∼65% became RBCT dependent, and 60% developed iron overload. Consistent with findings for LR MDS in general, most deaths of LR MDS-RS were MDS-related, although a sizeable minority died of unrelated causes. These results indicate a high level of morbidity and mortality in MDS-RS over a 5-year period. Further studies that incorporate SF3B1 testing may be required to better understand heterogeneity in the natural history of disease in this population and to confirm the results observed in this study.

Author contributions

R.B., C.W., C.C., P.S., and D.T. conceived the study, were involved in the conduct of the study; R.B., L.C., L.M., K.W.L.Y., M.G., N.Z., A.S., H.A.L., G.C., V.B., L.B., D.K., E.S-H., N.F., T.N., M-M.K., J.S., R.D., A.P., A.T., M.S., and A.M. performed data acquisition; P.S. analyzed the data; R.B., L.C., L.M., K.W.L.Y., M.G., N.Z., A.S., H.A.L., G.C., V.B., L.B., D.K., E.S-H., N.F., T.N., M-M.K., J.S., R.D., C.W., C.C., P.S., and D.T. interpreted the data. All authors reviewed and approved the final version of the manuscript and are accountable for all aspects of the research in ensuring that questions related to the accuracy or integrity of the work are appropriately investigated and resolved.

Acknowledgments

The authors received editorial support in the preparation of this manuscript from Saba Choudhary, PhD, of Excerpta Medica, funded by Bristol Myers Squibb. The authors would like to thank all the patients and their families, nurses, study personnel, and investigators who participated in the MDS-CAN Registry and made this study possible, the clinical study teams who participated in the study, and the protocol managers for this study. The authors would also like to thank Celgene Inc., a Bristol-Myers Squibb Company, Canada for unrestricted funds that help support the registry operational costs.

Disclosure statement

R.B. received research support from BMS/Celgene and Takeda; and received honoraria and served on the advisory board for BMS, Taiho Oncology, and Takeda. K.W.L.Y. was a consultant for Astex, BMS/Celgene, F. Hoffmann La Roche, Novartis, Otsuka, Paladin, Pfizer, Shattuck Labs, Taiho Oncology, and Takeda; received research funding from Astex, Forma Therapeutics, F. Hoffmann La Roche, Genentech, Geron, Janssen, Jazz Pharmaceuticals, Medimmune, Novartis, Onconova, and Tolero; and received honoraria from AbbVie and Novartis. M.G. received research support (institutional clinical trial contract) from AbbVie, Amgen, BMS/Celgene, Geron, Gilead, Janssen, and Novartis and honoraria from BMS/Celgene; was an expert witness for Novartis and Taiho Oncology; and served on the advisory board for AbbVie, Amgen, BMS/Celgene, Jazz Pharmaceuticals, Novartis, Paladin, Pfizer, and Taiho Oncology. N.Z. received honoraria and served on the advisory board for BMS and Taiho Oncology. A.S. received honoraria and served on the advisory board for AbbVie, BMS, and Janssen; and received research funding from BMS. H.A.L. served on the advisory board and received honoraria and research funding from AbbVie, Alexion, AstraZeneca, BMS/Celgene, Janssen, Novartis, Taiho Oncology. V.B. received research support from the Canadian Institutes of Health Research and the Leukemia & Lymphoma Society of Canada and royalties or licenses from BioGene; was a consultant for AbbVie, AstraZeneca, and Janssen; and received honoraria from AbbVie, AstraZeneca, Janssen, Medicom, and Oncology Education. L.B. served on the advisory board and received honoraria from AbbVie, Alexion, Amgen, Astellas, Astex, BMS/Celgene, Janssen, Jazz Pharmaceuticals, Novartis, Otsuka, Paladin, Pfizer, Roche, and Treadwell; was a consultant and expert witness for AbbVie, Novartis, and Pfizer; and received travel/conference support from Novartis and Pfizer. E.S-H. served on the advisory board for BMS and Taiho Oncology. T.N. served on the advisory board and received speaker fees from BMS/Celgene. M-M.K. served on the advisory board for Celgene and Taiho Oncology; and received speaker fees from BMS. J.S. received research support from AbbVie, Astellas, BMS/Celgene, Novartis, Pfizer, and Taiho Oncology; served on the advisory board for AbbVie, Amgen, Astellas, BMS/Celgene, Jazz Pharmaceuticals, Novartis, Paladin, Pfizer, Taiho Oncology, and Teva; and is/was an investigator in clinical trials for AbbVie, Astellas, BMS/Celgene, Janssen, and Novartis. C.W. and D.T. are paid employees of BMS; D.T. is also a stockholder/shareholder of BMS. C.C. is a paid employee and a stockholder/shareholder of EVERSANA™, which was contracted by Celgene, a Bristol Myers Squibb Company, to work on this project. P.S. was a paid employee of EVERSANA™ at the time the study was conducted. G.C., L.C., L.M., D.K., N.F., R.D., A.P., A.T., M.S., and A.M. have no disclosures to report.

Data availability statement

The data that support the findings of this study are available in the supplementary material of this article and the data that support the findings of this study are available from the corresponding author upon reasonable request.

Additional information

Funding

This study was funded by Bristol Myers Squibb.