Impact of Transfusion Dependence on Clinical and Economic Burden in Patients with Lower-Risk Myelodysplastic Syndromes: A 28-Year Retrospective Study

Introduction

Myelodysplastic syndromes (MDS) comprise a diverse group of clonal hematopoietic stem-cell malignancies that predominate in the elderly and are characterized by ineffective hematopoiesis, leading to cytopenia and morphologic dysplasia in the bone marrow, and a higher risk of acute myeloid leukemia (AML) or early mortality.¹

Erythropoiesis-stimulating agents (ESAs), with or without granulocyte colony-stimulating factor, are generally the first-line drugs used for treating anemia in patients with lower-risk MDS (LR-MDS).² ESAs are effective at inducing and sustaining erythroid responses, reducing red blood cell (RBC) transfusion requirements, and prolonging the time-to-first-RBC transfusion, although they may be effective only for those with endogenous erythropoietin (EPO) levels <200 U/L.^3,4 Furthermore, ESA efficacy and duration of response is limited, with only 1/3 patients responding over a 6–18-month duration.^2,4–6 Treatment options after ESA failure are limited, and most patients with LR-MDS eventually require regular RBC transfusions as part of their supportive care,² although this provides only transient relief from anemia and may lead to iron overload, with the potential for cardiac, hepatic, and endocrine complications.^2,7

RBC transfusion-dependence (RBC-TD) is associated with a decreased overall survival (OS) and a higher chance of AML progression in patients with MDS, with more prominent negative impacts on survival and leukemic evolution observed in patients with LR-MDS.^8–11 In addition, real-world reports indicate that RBC-TD is a major driver of the economic and healthcare resource utilization (HCRU) burden in MDS. In a retrospective analysis conducted by Goldberg et al, which included 512 patients aged ≥65 years with newly diagnosed MDS, cumulative 3-year mean Medicare costs were US$49,156.¹² In that study, transfused patients had greater use of hospital inpatient and outpatient services and incurred significantly higher mean costs than non-transfused patients (US$88,824 vs US$29,519, p<0.001).¹² It should be noted, however, that as this study was published more than a decade ago, the current costs would most likely be higher. In another retrospective analysis conducted by DeZern et al that analyzed cost patterns of MDS care during RBC-TD and RBC transfusion-independence (RBC-TI) over a 5-year period, cost increases in patients who were TD were proportional to transfusion frequency, and achieving TI yielded economic benefits.¹³

Although providing valuable insight into the burden of RBC-TD MDS, the outcomes examined in most studies to date have been based on studies with relatively short observation periods (from a few to around 10 years) that may not have been long enough to fully capture the burden of RBC-TD.^12–17 Analysis of outcomes for patients in the combined International Working Group for Prognosis in MDS patient database (N = 7012), for example, showed that the median survival for patients is 5.3 years for low and 8.8 years for very low Revised International Prognostic Scoring System (IPSS-R) risk categories, with the median time for 25% of patients to develop AML being 10.8 years for those with low IPSS-R risk and not being reached for patients with very low IPSS-R risk.¹⁸ It is therefore necessary to conduct a study with an extended observation period even longer than 10 years in order to identify the real-world transfusion burden in LR-MDS patients.

We conducted a 28-year retrospective observational study to determine the burden of RBC-TD in terms of HCRU, medical cost, and clinical impact in South Korean patients with LR-MDS.

Methods

Study Design and Participants

We conducted a retrospective observational study of adult patients newly diagnosed with MDS who were initiating treatment, using the electronic medical record and expense database of the Samsung Medical Center (SMC), Seoul, South Korea, between September 1, 1994 and September 30, 2022 (observation period). The index date was defined as the date of the first documented diagnosis of MDS in the SMC’s electronic medical record database between September 1, 1994 and April 1, 2022 (baseline period). After the index date, the earliest laboratory result before initiating treatment was defined as the baseline laboratory value.

This study was conducted in accordance with the ethical principles that have their origin in the Declaration of Helsinki. The protocol, amendments, and patient informed consent received appropriate approval by the Institutional Review Board (IRB) of the SMC (Institutional Review Board number 2022–09-090-001), as required by local law prior to initiation of the study at the site. The study is registered to the Korean Clinical Research Information Service (CRIS), which is partnered with the International Clinical Trials Registry Platform (ICTRP). Registration number is KCT0010062.

This study did not require informed consent from the patients as the data were anonymized during the process of data extraction from the electronic medical records. The requirement for consent was waived by the IRB.

Patients were included in the study if they were at least 18 years old and had a documented diagnosis of MDS (International Classification of Diseases 10 code of D46.XX) during the baseline period at the SMC and had medical records of MDS treatment and HCRU data during the observation period. Patients were excluded if they had been diagnosed with AML (a documented diagnosis or presence of ≥20% blasts in the bone marrow or peripheral blood) within a month after the index date. Patients were scored and categorized into LR-MDS and higher-risk MDS (HR-MDS) at baseline using the Revised International Prognostic Scoring System (IPSS-R).¹⁸ In this study, patients with very low, low, or intermediate risks were classified as LR-MDS, and those with high or very high risks were classified as HR-MDS.

RBC-TD was defined as the occurrence of any 16-week period without a consecutive transfusion-free 56 days, requiring RBC transfusions of ≥2 units per 8 weeks. Patients who experienced an RBC-TD event at least once during the observation period were categorized as TD, while others were categorized as non-transfusion-dependent (NTD).

Outcomes

The primary endpoint was the economic and HCRU burden of TD in patients with LR-MDS. HCRU and medical costs were estimated and compared between TD and NTD subgroups per 1000 person-years. The HCRU indices comprised:

• the number of outpatient visits
• the number of hospitalizations
• the number of emergency room visits
• the length of hospital stay
• the units of RBC transfused.

The medical costs comprised:

• total medical costs
• outpatient costs
• hospitalization costs
• emergency room costs.

Cost data were adjusted to 2022 values according to the South Korean consumer price index and converted to US dollars using the 2022 currency exchange rate as US$1 = South Korean won 1205.25.

The secondary endpoints were clinical characteristics, including ESA treatment response and clinical outcomes. ESA treatment response evaluations included primary resistance, time to ESA discontinuation, ≥8-week or ≥16-week RBC transfusion-independence achievement during consecutive ESA treatments, and attainment of erythroid response.¹⁹ The clinical outcomes that were evaluated comprised median OS, acute myeloid leukemia-free survival (LFS), and progression-free survival (PFS). OS and LFS were estimated from the index date of MDS diagnosis to the date of documented death due to any cause, and the date of either documented death or the diagnosis of AML, respectively. PFS was defined as the time between the index date of MDS diagnosis and the date of a documented death, a diagnosis of AML, or an HR-MDS progression. Additionally, OS and LFS from the date of first TD event were also estimated in TD patients with LR-MDS to approximate the impact of TD on prognosis.

Statistical Analysis

Baseline characteristics, treatment pattern, and clinical outcomes were reported with observed rates with interquartile ranges (IQRs). In patients with LR-MDS, those who were TD and NTD were compared to evaluate baseline characteristics, HCRU, medical cost, and clinical outcomes. As appropriate, Wilcoxon’s rank-sum test was used for continuous variables, and chi-squared or Fisher’s exact test for categorical variables. OS, LFS, and PFS were analyzed using Kaplan–Meier methods and Log rank test between TD and NTD patients with LR-MDS from the index day of MDS diagnosis. Statistical significance was assessed using a 2-sided p-value <0.05.

Results

Patient Characteristics

A total of 926 adults with a documented diagnosis of MDS were initially identified. Following exclusion of 95 patients who were diagnosed with AML within a month after the initial diagnosis of MDS, there remained 831 all-risk MDS patients in the overall study population. Baseline IPSS-R scores were available for 543 patients and resulted in the classification of 349 patients as having LR-MDS at baseline. Among patients with LR-MDS, 29.5% (103/349) were TD, and 70.5% (246/349) were NTD (Figure 1). No significant differences in median age or sex were observed between the two groups, but TD patients exhibited lower median baseline hemoglobin level (8.7 g/dL vs 9.4 g/dL, p = 0.004) and higher median baseline EPO level (290.5 U/L vs 112 U/L, p = 0.002) compared with NTD patients. Furthermore, only 32.9% of TD patients showed baseline EPO levels ≤200 U/L, in contrast to 62.4% of NTD patients (p < 0.0001) (Table 1).

Table 1 Patient Demographics and Baseline Disease Characteristics of Study Population with LR-MDS

Figure 1 Patient enrollment flow chart. Abbreviations: AML, acute myeloid leukemia; HCRU, healthcare resource utilization; HR-MDS, higher-risk MDS; ICD, International Classification of Diseases; IPSS-R, revised international prognostic scoring system; LR-MDS, lower-risk MDS; MDS, myelodysplastic syndrome; NTD, non-transfusion-dependent; TD, transfusion-dependent.

Among the 831 all-risk MDS patients identified, 194 (23.3%) were TD and 637 (76.7%) were NTD. Patient characteristics for the all-risk MDS population, stratified by transfusion dependence, are provided in Supplementary Table 1.

The median duration of follow-up was 116.5 months for TD LR-MDS and 61.1 months for NTD LR-MDS patients, respectively (p <0.0001).

HCRU and Medical Costs by RBC Transfusion Burden

For the primary endpoint, the transfusion burden in terms of HCRU and medical costs was higher for patients who were TD compared with those who were NTD across the assessed criteria in patients with LR-MDS (n = 349) (Figure 2) and also when evaluating the all-risk MDS population (n = 831) (Supplementary Figure 1). The differences between TD and NTD patients in all HCRU and medical cost criteria tended to be greater for patients with LR-MDS than for the all-risk MDS patient population. Total medical costs were more than twice as high for TD patients than for NTD patients (13,501,635 US dollars per 1000 patient-years vs 6,086,585 US dollars per 1000 patient-years, respectively). Among patients with LR-MDS (n = 101) who were TD, HCRU and costs per 1000 person-years increased after the first RBC-TD event occurred (Supplementary Figure 2 and Supplementary Table 2).

Figure 2 Healthcare resource utilization and medical costs for patients with LR-MDS. (A) HCRUa of patients with LR-MDS by RBC transfusion burden, (B) medical costsb,c in patients with LR-MDS by RBC transfusion burden. aHCRU data were accessible from 1994. bCost data were accessible from 2001. cAll cost data were adjusted to 2022 values and converted to USD using the 2022 currency exchange rate of USD1 = KRW1205.25. Abbreviations: HCRU, healthcare resource utilization; KRW, South Korean won; LR-MDS, lower-risk myelodysplastic syndromes; NTD, non-transfusion-dependent; RBC, red blood cell; TD, transfusion-dependent; USD, US dollars.

ESA Treatment Pattern and Outcomes in TD Patients with LR-MDS

Among the 103 LR-MDS patients who were TD, 76 (73.8%) had baseline EPO level data, of whom, 51 (67.1%) were theoretically eligible for ESA treatment as their EPO levels were ≤500 U/L (Table 1). Among these patients, only 20 (39.2%) initiated their first ESA treatment after developing RBC-TD, with darbepoetin alfa being the only ESA used.

ESA treatment pattern and outcomes for the 20 patients receiving ESA treatment after TD development are shown in Table 2. The median (IQR) baseline EPO level for the ESA-treated patients was 283.0 (106.6–300.0) U/L, and 85% of patients (17/20) showed primary resistance during the first 8 weeks of ESA use. While the median (IQR) time from first ESA administration to discontinuation was 23.5 (9.2–42.5) months, median (IQR) cumulative duration of ESA treatment was 6.8 (3.1–11.9) months. Overall, during 24 weeks of ESA treatment, 8 out of 20 TD patients undergoing their first ESA treatment (40%) achieved ≥8-week RBC-TI, with only 3 (15%) maintaining ≥16-week RBC-TI. After Week 24 of treatment, only half of ≥8-week TI responders (4 out of 8) maintained the response, and no additional patients achieved their first RBC-TI ≥8-week between Weeks 25 and 48. Only 25% (5 out of 20) achieved an erythroid response within the first 24 weeks of ESA treatment, and just one of these five maintained the erythroid response between Weeks 25 and 48. Overall, 95% (19 out of 20) discontinued ESA treatment without re-initiation. Reasons for discontinuation were not available.

Table 2 ESA Treatment Pattern and Outcomes in Patients with LR-MDS After RBC-TD Development

OS, LFS, and PFS in Patients with RBC-TD vs Those without TD

TD patients with LR-MDS exhibited significantly shorter median [IQR] OS (58.4 [22.8–161.5] vs 103.1 [29.9–213.4] months, p = 0.014) and LFS (52.7 [17.1–116.0] vs 102.7 [29.8–213.4] months, p = 0.003) than did NTD patients (Figure 3 and Table 3), although those differences did not reach significance in all-risk MDS patients (Supplementary Table 3). Additionally, in TD patients with LR-MDS, the median (IQR) OS and LFS estimated from the first RBC-TD event were 33.8 (12.9–157.6) and 33.6 months (11.5–157.5), respectively (Supplementary Figure 3). In terms of PFS, including progression to HR-MDS as well as to AML and death, TD patients demonstrated significantly shorter median [IQR] PFS than NTD patients (35.5 [16.0–89.3] vs 74.8 [24.2–213.4] months, p = 0.0003) (Figure 3 and Table 3).

Table 3 Survival Outcomes for Patients with TD vs NTD LR-MDS

Figure 3 Kaplan–Meier estimated survival outcomes for patients with TD vs NTD LR-MDS. (A) overall survival, (B) leukemia-free survival, and (C) progression-free survival. Abbreviations: IQR, interquartile range; LFS, leukemia-free survival; LR-MDS, lower-risk myelodysplastic syndromes; MDS, myelodysplastic syndrome; NTD, non-transfusion-dependent; OS, overall survival; PFS, progression-free survival; TD, transfusion-dependent.

Discussion and Conclusions

This 28-year retrospective study demonstrates the substantial economic and HCRU burden associated with LR-MDS in South Korea. The findings indicate markedly greater burden for patients who are TD compared with those who are NTD (Figure 2), with both HCRU and medical costs per 1000 patient-years increasing dramatically after the first RBC-TD event (Supplementary Figure 2 and Supplementary Table 2). These findings are consistent with previous reports that also showed TD to be a major factor contributing to both economic and HCRU burden.^12–16 Interestingly, the differences between TD and NTD patients were more prominent among patients with LR-MDS than among the all-risk MDS population. For example, the number of outpatient visits per 1000 patient-years among TD patients was nearly 2.0-fold higher than among NTD patients in the LR-MDS population, vs only 1.5-fold greater in the all-risk MDS population (Figure 2 and Supplementary Figure 1). Similarly, outpatient care costs per 1000 patient-years were 2.4-fold higher among TD patients than among NTD patients in the LR-MDS population vs 1.8-fold in the all-risk MDS population (Figure 2 and Supplementary Figure 1). The findings suggest that while RBC-TD remains a significant contributor to the economic and HCRU burden across the entire MDS population, its impact is relatively greater in patients with LR-MDS patients compared with those with higher-risk MDS. Given the chronic and progressively deteriorating prognosis of LR-MDS, RBC-TD serves as an important early indicator of increased economic burden and HCRU. This highlights its significance not only for individual patients but also for healthcare payers, such as public organizations including South Korea’s National Health Insurance Service (NHIS), which shoulder these costs.

Our study indicates that although ESAs are the primary treatment to ameliorate anemia in patients with LR-MDS, their real-world clinical efficacy is limited. Among TD LR-MDS patients who had baseline EPO level data, 67.1% (51 out of 76) showed EPO levels >200 U/L and 32.9% had EPO levels >500 U/L (25 out of 76) (Table 1). While treatment guidelines recommend ESAs for patients with EPO ≤500 U/L,²⁰ response rates drop significantly when baseline EPO level exceeds 200 U/L.^3,4 Furthermore, among patients who initiated ESA treatment after developing TD, 85% showed primary resistance to treatment. Despite this resistance, the median time to discontinuation was 23.5 months. It is also notable that the median cumulative duration of ESA treatment was only 6.8 months, which is shorter than the median time to discontinuation (Table 2). These findings may reflect the absence of alternative anti-anemic therapies and the limited efficacy of ESAs in managing RBC-TD. Regarding sustainable efficacy, this study demonstrated that 40% of TD patients achieved 8-week RBC-TI within the first 24 weeks of ESA treatment, and only half of them maintained the response in the subsequent 24 weeks (Table 2). Overall, these results suggest that, because of the lack of viable alternative options to mitigate RBC-TD, ESAs have been used even for the ESA-resistant patients with LR-MDS, despite their inconsistent efficacy.

In our study, RBC-TD emerges as a significant predictor of shorter survival and higher risk of disease progression in patients with LR-MDS (Figure 3), which is consistent with the literature.^{8,10,16,17,21} The shorter life expectancy among TD patients may be due to RBC transfusion-associated risks, including iron overload, higher chances of infection, and presumably, RBC autoimmunization following the formation of RBC allo-antibodies after repeated RBC transfusions.^7,22,23 Furthermore, it is notable that the differences in survival and disease progression between TD and NTD patients were only significant in LR-MDS patients but not in the all-risk MDS patients (Figure 3 and Supplementary Table 3), which suggests RBC-TD is an indicator of poorer prognosis in LR-MDS rather than in HR-MDS. Overall, our findings emphasize the importance of reducing transfusion burden in TD LR-MDS patients. Despite being categorized as “lower-risk”, RBC-TD may indicate greater chances of poorer prognosis, higher healthcare cost, and increased HCRU.

In a recent large-scale claims database study, achieving RBC-TI was associated with improved survival and reduced progression in TD LR-MDS patients.²⁴ However, if ESA treatment continues to be preferred first-line therapy, the potential for improvements in survival and disease progression may remain suboptimal due to its limited clinical efficacy, as demonstrated in this study and supported by previous literature.^4–6 In this regard, two novel agents – luspatercept (an erythroid maturation agent) and imetelstat (a telomerase inhibitor) – have recently been approved in the US for treatment of RBC-TD in patients with LR-MDS.^25,26 In placebo-controlled Phase 3 trials, both agents significantly reduced the severity of anemia and enabled RBC-TI for at least 8 weeks compared to placebo in patients with LR-MDS unresponsive to, or ineligible for ESA treatment.^27,28

One key strength of our study is its longitudinal real-world observation period, spanning 28 years. To our knowledge, no prior research has examined the long-term impact of RBC-TD in LR-MDS patients over such an extended timeframe. This allowed us to demonstrate the influence of RBC-TD on clinical outcomes and economic burdens over two decades.

However, the study has notable limitations. Data were sourced from electronic medical records at a single institution, which may have limited the capture of outcomes, such as death or AML progression for patients who transferred to other facilities. Additionally, unmeasured confounding factors and potential biases may have influenced the results. Although the long observation period is a strength of the study, it also results in some challenges regarding changes that may have occurred during the observation period that were not captured. These include changes in transfusion thresholds, access to therapies and overall medical costs. Finally, it is recognized that the magnitude of the differences of higher costs/HCRU and worse OS/LFS/PFS in TD vs NTD may be confounded due to contributing factors, including those mentioned above, and would benefit from covariate adjustment, which was not conducted.

In conclusion, leveraging a 28-year single-center database, this study provides critical insights into the economic burden, HCRU, and clinical outcomes associated with RBC-TD in LR-MDS. Key findings indicate that RBC-TD correlates with higher economic burden, increased HCRU, rapid leukemic progression, and reduced life expectancy. Although ESA treatments remain the standard first-line therapy, their limitations underscore the need for alternative treatments to effectively mitigate RBC-TD and improve patient outcomes in LR-MDS. Novel agents such as luspatercept and imetelstat hold promise, but their long-term real-world efficacy requires further investigation.