Characteristics, treatment and outcomes of myelodysplastic syndrome in two Finnish hospital districts

Data access

This was a retrospective study of patients diagnosed with MDS or Chronic Myelomonocytic Leukemia (CMML) in two hospital districts: HDSF and PHD in Finland from January 1^st 2010 to December 31^st 2019. Findata, the social and health data authority in Finland, approved this study (data permit number THL/2767/14.02.00/2020). According to the Finnish Act on Secondary use of Health and Social data (552/2019) no patient consent was required.

Patient inclusion criteria/Cohort formation

Patients were included if they were at least 18 years of age at the date of the first recorded MDS or CMML diagnosis code and had a diagnosis code for MDS (D46.*) or CMML (C93.1) recorded during 1.1.2010-31.12.2019 (HDSF) or 1.1.2012-31.12.2019 (PHD). Patients recorded with MDS or CMML prior to the beginning of the inclusion period of the corresponding hospital district were excluded. At PHD diagnoses were available starting from 2010, thus the inclusion period at PHD started at 2012 in order to have 2-year washout period to prevent the inclusion of non-incident patients. In order to prevent inclusion of patients misdiagnosed with MDS during the diagnosis process whose diagnosis was corrected later to another haematological condition, we decided on the following criteria: 1) The diagnosis code had to be marked at least three times at the electronic health records within 6 months of the first diagnosis, and 2) if the patient received a diagnosis code for D47 (other neoplasms of uncertain behaviour of lymphoid, hematopoietic and related tissue), C90 (multiple myeloma), C92.1 (chronic myeloid leukemia), C81-C88 (lymphomas) or D45 (polycythemia vera) and MDS was no longer recorded after that, the patient was excluded 3) if the patient died within 6 months of first MDS diagnosis, they were included regardless of criteria 1) and 2). The date of the first record of the MDS or CMML diagnosis code was defined as the index. The length of follow-up was defined as time from index until death, progression to AML (first record of C92.0), or end of study (31.12.2019).

Subgroup formation

Patient age and sex along with all ICD-10 diagnosis codes registered during the health care contacts, medical treatments with time stamp, generic names and ATC-codes, procedure codes and records of laboratory measurements were obtained from the HDSF and PHD data lakes. Patients were stratified into 2 subgroups: 1) patients treated at the specialty health care (SHC) and 2) patients who were untreated and only received supportive care or who were only diagnosed at SHC and possibly treated at primary health are (PHC). As the PHC data was not available for this study, further assessing the latter patient group was not feasible. Two mutually distinct subgroups were distinguished from the patients treated at SHC, namely patients who were detected receiving allogeneic haemotopoietic stem cell transplant (HSCT) and patients receiving azacitidine treatment. Note that some of the patients in the HSCT subgroups also received azacitidine, however, this was used either as an induction therapy/treatment for the transplant or as maintenance/relapse treatment after the transplant. Only transplants and azacitidine treatments prior to potential progression to AML were considered. In order to detect the stem cell transplants, three criteria were applied: 1) procedure recorded at the data lake (WW302, WW304 or WW306), 2) text specifier recorded alongside the Z94.8 diagnosis code (“Other transplanted organ and tissue status”) mentioning the exact date of the transplant, or 3) immunosuppressive medication combined with a prior HLA-testing. Immunosuppressive medications included were: ciclosporin, everolimus, tacrolimus, mycophenolic acid and methotrexate. Date of the first record of the immunosuppressive medication was used as the date of the transplant as the immunosuppressive treatment is typically initiated day before or the day of transplant.

Patient characteristics & treatments

Number and proportion of patients per sex and age group in 10-year intervals were reported. Number and proportion of patients with either a malignancy or chemotherapy before MDS or CMML diagnosis were reported. For a previous malignancy it was required that a patient had at least three records of the same cancer diagnosis code (at two-character level i.e., ‘C’ followed by two numbers) on separate days. A previous chemotherapy was defined as any medical treatment with an ATC-code starting with ‘L01’ (excluding hydroxycarbamide; ATC-code L01XX05) at least 3 months before the index. Charlson comorbidity index was calculated for each patient from the ICD-10 diagnosis codes recorded prior to index (5). The number and proportion of patients with CCI 0, 1-2, 3-4 and 5+ were reported.

Mortality and time to event analyses

Dates of death were obtained from the HDSF and PHD data lakes and causes of death were obtained from the causes of Death (Statistics Finland) register up to the end of study period. Overall survival was defined as time from index until death (event) or end of study (31.12.2019; censoring event). MDS or CMML mortality was defined as any death with MDS or CMML recorded as immediate or main cause of death. Other mortality was defined as death without records of MDS or CMML in either main or immediate causes of death. Date of progression to AML was defined as the first day patient being recorded with C92.0. Overall survival was estimated via Kaplan-Meier fits while progression rate to AML, MDS or CMML mortality and other mortality were assessed simultaneously via Aalen-Johannsen fits, which is an extension of the Kaplan-Meier method taking into account the likely competing risks setting between multiple, alternative outcomes. Event free survival (EFS) was defined as time from index until composite event of either progression to AML or death (any cause).

Health care resource utilization (HCRU)

All hospitalizations and outpatient contacts of the patients at the specialty health care of HDSF and PHD were obtained. Outpatient contacts per patient year (PPY) and continuous hospitalizations, as well as total number of inpatient days (i.e., days spent at the ward) PPY were calculated. Furthermore, prices of the contacts were joined based on the specialty of the contact. Prices were obtained from Mäklin et al (6). These prices were then used to calculate costs PPY. Here ‘patient’ years correspond to the length of follow-up. Contacts recorded with MDS or CMML diagnosis were defined as MDS or CMML specific HCRU. Confidence intervals were obtained using bootstrapping with 10 000 bootstrap samples.

Statistical analysis

All data were delivered to and analysed in the Findata secure data analysis environment. Only aggregated results files not containing any patient-level data were pulled from the analysis environment for further reporting. All statistical analyses were run using R: A language and software for statistical computing, version 4.0.3. Time to event analyses were run using the survival R-package, A Package for Survival Analysis in R (7).

Patient characteristics

In total 565 patients fulfilling the inclusion criteria, 325 from HDSF and 232 from PHD. As in Finland the diagnosis of MDS is always made in the specialised healthcare setting, this likely represents all MDS patients from these two districts. The median age of the patients was 77 years (interquartile ratio (IQR): 68-83 years), included 223 females (39.5%) and had a median follow-up time of 17 months (IQR: 6-39 months) (Table 1). Incidence of MDS in our cohort was 6,06 per 100 000 persons. Our cohort included 13 patients with diagnosis specifically for CMML, but as the number of these patients was so small, they were not analysed separately. 52 patients (9%) had a previous cancer diagnosis, most prominently prostate cancer (2,8%), breast cancer (1,1%) and other haematological malignancies. 70 (12,4%) patients had previous cancer diagnosis or chemotherapy, indicating secondary or treatment related MDS. Other frequent comorbidities (% during the whole follow-up) included essential hypertension (37,9%), atrial fibrillation (25,3%), heart failure (20,7%) and chronic ischaemic heart disease (19,3%). Charlson comorbidity index (CCI) which classifies and summarises prognostic diagnoses into an integer score (8) where scores of 0-1 are considered low comorbidity burden, and 2 or more is considered increasingly higher. In our data, 36% of patients has CCI score >2.

141 patients (25%) initially diagnosed with MDS at specialised healthcare (SHC), received no further treatment at specialised healthcare (Table 1., “Not treated at SHC”). Most of these patients likely received treatment at primary care setting, of which we have no data. These patients include low risk, non-symptomatic patients who are usually treated with watchful observation or receive supportive care.

Treatment of MDS patients at specialty health care

Only a handful of medications are used to treat MDS in Europe and worldwide (9). HSCT is the only potentially curative treatment for MDS patients, but many MDS patients are not suitable candidates for it, due to old age, comorbidities, poor performance status or lack of appropriate donors. We identified 26 (5%) patients who received HSCT in our cohort. However, patients in need of HSCTs at PHD are sent to either to HDSF or Helsinki for the procedure. As we have no data from Helsinki, we may be missing some HSCT patients from PHD. On average, HSCT receiving patients were younger and had less comorbidities (Table 1).

The treatment options for MDS patients unfit for HSCT are limited to supportive care (e.g., blood transfusions, EPO), the hypomethylating agent azacitidine, and lenalidomide for patients with chromosome 5q deletion. Azacitidine was given to 95 (22%) patients in our study, of which 19 were given azacitidine in association with or after HSCT. All patients who received azacitidine before HSCT received it within a year before the transplant and most received it as an induction, i.e., 6 cycles pre transplant. 72 (17%) patients then received azacitidine independently of HSCT (Table 2). HSCT group was youngest with median age 63.3 years (IQR: 60.9-65.5). Azacitidine receiving group was older at median index 72.6 years (IQR: 66.6-77.4) and median initiation of Azacitidine at 73.1 years (IQR: 67.2-77.8).

Red blood cell transfusions (RBCT) were given to 380 (67%) patients. MDS patients often develop dependency on blood transfusions (BT), and indeed, patients received 7 units per patient year (PPY), which increased to 16 units PPY for patients receiving azacitidine, posing a significant burden to both patients and healthcare. Median minimum hemoglobin value within one week pre first RBCT was 82 g/l (IQR: 76-90). Platelets were given to 123 patients (22%), with a frequency of 2.6 units PPY (Table 3).

Outcomes

Median survival of all MDS patients was 26.9 months (95% CI: 23.6-35.1) (Figure 1A). Patients whose disease required treatment at specialised care had a poorer survival than those treated with watchful observation or supportive care at primary care setting (Figure 1B). Median survival for patients treated at specialised care was 22.5 months (95% CI: 17.8-24.8), whereas for patients who are treated at primary care / not treated, have a median survival of 79.7 months (95% CI: 49.7 – not reached), supporting the notion that these patients are indeed low/ medium risk patients. Survival of patients who undergo HSCT is significantly improved, with 73% of patients alive at 24 months (95% CI: 49-87; median survival not reached during study period). As expected, overall survival was also strongly dependent on age at index, with 86,5% (95% CI:76,96-97,21) of under 60-year-olds alive at 24 months, whereas only 43,1% (95% CI: 36,48-50,87) of over 80-year-olds were alive at 24 months (Figure 1C).

• Download figure
• Open in new tab

• Download figure
• Open in new tab

Figure 1.

A) OS of all MDS patients, both specialised care treated and those who were treated in primary care. Dashed line represents median survival B) OS of MDS patients who received treatment at specialty healthcare (dashed line) and who did not (solid line). C) Overall survival of MDS patients in different age groups. D) Competing risk model of all MDS patients to AML transformation (red line), MDS caused mortality (black line) and other cause mortality (grey line). Blue line represents proportion of live, not-AML-transformed patients.

A major risk for MDS patients is progression into AML. 60 MDS patients (10%) in our study progressed to AML (Figure 1D). These patients had a poor outcome, with median OS of just 3.9 months (95% CI: 2.1-5.3). MDS was the underlying or immediate cause of death for 17% of the cohort. The large proportion of other (non-MDS) causes of death is indicative of the fragile nature of these patients (Figure 1D). For the “other mortality” group the most common causes of death were pneumonia (29 patients, 13% of other mortality), chronic ischemic heart disease (24 patients, 11% of other mortality), heart failure (13 patients, 6% of other mortality) and myocardial infarction (11 patients, 5% of other mortality). Of note, 25 patients who did not have a record of AML diagnosis during their lifetime, had AML recorded as a cause of death.

For the subgroup initiating azacitidine treatment median OS is 22.6 months (95% CI: 17.3-25.2) (Figure 2A), close to that of all specialised care treated patients (Figure 1B). However, azacitidine treated patients had a higher risk of developing AML, and proportionally higher risk of dying of MDS (Figure 2B).

• Download figure
• Open in new tab

Figure 2.

A) OS for azacitidine treated patients from the beginning of azacitidine treatment. B) Time to event analysis of azacitidine treated patients from the beginning of azacitidine treatment. AML= transformation to acute myeloid leukemia, MDS-mortality = MDS specific mortality where MDS is an immediate or underlying cause of death, Other mortality = cause of death other than MDS.

HCRU

The cost of care for MDS patients in Finland is substantial. Specialised healthcare resource utilization of MDS patients was calculated per patient year and divided to inpatient and outpatient HCRU, as well as MDS-specific (visits where MDS is recorded as diagnosis) and other cause costs. The economic burden of MDS treatment consists of both hospitalizations and outpatient visits, and is higher for azacitidine treated patients (Figures 3 and 4). On average, SHC treated MDS patients spent 42 days (95% CI: 37-48 days) in hospital per year (inpatient and outpatient days combined), whereas for azacitidine treated patients the number of days in hospital per year is 87 (95% CI: 73-103 days). For all SHC treated MDS patients approximately one third of total per patient year costs were MDS specific (total cost PPY 17 563€, 95% CI: 15 488-19 898€, MDS specific costs 4 704€, 95% CI: 3 747-5 846€), whereas for azacitidine treated patients the costs were more strongly driven by MDS, with approximately half of costs being MDS specific (total costs for MDS patients per patient year 33 094€, 95% CI: 27 592-39 875, MDS specific costs 18 824€ 95% CI: 14 060-24 424€).

• Download figure
• Open in new tab

Figure 3.

Hospitalizations (panel A), inpatient days (i.e. days at ward; panel B) and outpatient contacts (panel C) per patient year according MDS specific and other health care use in all MDS patients treated at specialty health care and AZA treated MDS patients.

• Download figure
• Open in new tab

Figure 4.

HCRU related MDS/CMML specific costs and other costs per patient year in all MDS patients treated at specialty health care and in AZA treated MDS patients. Total costs are reported in panel A while the costs originating from hospitalizations and outpatient contacts are reported separately in panels B and C, respectively.