One hundred and sixty patients entered the study from March 12, 2007 to May 14, 2008. Twelve subjects were withdrawn due to major violation of inclusion criteria: 4 for lack of significant cytopenias, 3 for history of other neoplastic disease, and 5 because of excessive time span between MDS diagnosis and the baseline visit.
The remaining 148 subjects (males 83) were of mean age 72.0 ± 10.7). Thirty-nine (26.4%) were TD (35 subjects (23.6%) required ≥2 RBC units/month before enrolment) and were comparable for gender, age, education, marital status and home care. One hundred and fifteen cases (77.7%) had Hb <12 g/dL, of which 82 (55.4%) with Hb <10 g/dL. illustrates clinical features at diagnosis. Charlson's comorbidity index was >1 in 33 subjects (22.3%). Cytogenetic information, thus IPSS score, was unavailable in 14 cases (9.5%) who continued the study because of a bone marrow blast count <10% and a single cytopenia.
Clinical features at diagnosis
Patient disposition and therapies
Eighty-nine patients (60.1%) completed the study. Thirteen (8.8%) died during the study (4 TF and 9 TD), while 17 (11.5%) discontinued the study for medical reasons, and 21 (14.2%) were lost to follow-up.
During the study, 93 (62.8%) patients received ESAs (mainly epoetin alfa), 18 (12.2%) received azacitidine and 3 (2.0%) received lenalidomide.
Changes in hematological parameters during the study
Fifty-two patients (35.1%) received transfusions during the study. Within three months from diagnosis, 75% of all previously TD patients and 9% of all previously TF patients were transfused. Overall mean Hb values increased by 0.37 g/dL (95%CI 0.10-0.64, p=0.0075), from 10.32 (median 9.8, IQR 8.8-11.8) at baseline to 10.69 (median 10.7, IQR 9.1-12.0) at end of study.
QOL instrument validity and scores
Cronbach’s alpha, calculated at each visit, varied for QOL-E domains between 0.54 and 0.85 with most values ranging from 0.6 to 0.8; for the EQ-5D dimensions it ranged from 0.7 to 0.8. Most Spearman’s correlation coefficients between each QOL-E domain score and VAS measures (the three LASA and the EQ-5D VAS) ranged from 0.4 to 0.65, and Pearson’s correlation coefficients between each LASA and the EQ-5D VAS were mostly in the 0.55-0.65 range.
The scores of all scales at baseline are reported in . Fatigue was not prevalent; however, average energy levels were low and physical function was poor. TD patients had a significantly poorer QOL than TF patients in QOL-E physical, functional, and MDS-specific, in LASA general, and EQ-5D VAS health state scores. There was no difference between the mean EQ-5D summary index in TD and TF patients.
In general, changes in QOL in time did not achieve statistical significance once adjusted for the multiplicity of comparisons. However, there was significant worsening in the QOL-E MDS-specific domain () with a mean change in score of -7.4 (95%CI, -12.9 to -1.9, p=0.0024) after 12 months, persisting at the end of the study. LASA energy scores increased significantly only during the first 6 months (), the mean change being +8.7 (95% CI, +1.1 to +16.3, p=0.017), and decreased thereafter.
A. QOL-E score patient (Physical, Functional, Social, Fatigue, MDS-specific) in time; B. QOL LASA score patient (energy, activity, overall) in time.
Very limited changes over time were observed in the EQ-5D summary index. Similarly, the ECOG PS through time did not change in most patients.
Predictive factors of quality of life
Univariate and multivariate analyses of the effects of candidate predictors of QOL, including: age, gender, marital status, education, family/home care, Charlson’s comorbidity index, Hb levels, platelet count, neutrophil count, transfusions and time from baseline (disease duration), were performed for each QOL scale ( and ).
Predictive factors of QOL-E scores
Predictive factors of LASA and EQ-5D VAS
At univariate analysis, better QOL-E physical, functional, social and fatigue scores, LASA and EQ-5D VAS scores were associated with younger age, Charlson’s index ≤1 (except LASA energy), higher Hb levels and transfusion independence. QOL-E social scores were also associated with marital status (married subjects had worse scores) and inversely with duration of disease. QOL-E MDS-specific scores were associated with Charlson’s index, Hb levels and transfusion independence and inversely with time from diagnosis.
At multivariate analysis, Hb concentration was the most important independent predictor, higher values being significantly associated with better QOL scores. After adjustment for other factors, transfusions were still significantly associated with worse QOL-E physical, QOL-E functional, QOL-E social, QOL-E MDS-specific, LASA activity and EQ-5D VAS scores. With the exception of the LASA energy score, Charlson’s comorbidity index was a significant independent predictor of all QOL indexes, multiple comorbidity being associated with a poorer QOL. The effect of age achieved statistical significance only for the QOL-E physical score (younger age predicted better scores). Marriage was still significantly associated with worse social scores, while male gendere was significantly associated with better fatigue scores in multivariate analysis.
No independent predictive effect was found for thrombocytopenia or neutropenia. There was a significant worsening in QOL-E physical, social and MDS-specific scores in time.
Physician’s assessment of QOL compared to patients’ self-assessment
Physician- and patient-assessed QOL-E scores were significantly correlated with most kappa correlation coefficients ranging from 0.2 to 0.6. Exceptions were observed at baseline for the item “Does inability to travel disturb your daily life?”, at baseline and after 12 months for the item “Does depending on hospital, physicians and nurse staff disturb your daily life?”, and for an item regarding sexual life after 6 and 12 months.
Relevant differences between physician- and patient-assessed mean scores were observed for 3 QOL-E items: item 1 (a general statement of health state perception); “Has climbing stairs been limited by your health state over the last week?”; “Has lowering yourself been limited by your health state over the last week?”. The physicians’ score rating was always higher than the patients’ own scores.
Correlations between patient and physician LASA scores were initially weak (Pearson’s coefficients 0.31-0.33) and increased considerably at later visits (0.53-0.74). Physicians’ ratings for LASA energy were significantly higher than the patients’ scores. Physicians’ scores for LASA activity and LASA general were significantly higher at 6 and 12 months.
The relationship between patient-assessed QOL-E physical and ECOG PS is shown in , and . There were significant mismatches since about half of patients with ECOG PS = 0 (judged as being fully active and with excellent functioning) experienced poor QOL-E physical scores.
Figure 2 A. Distribution of QOL-E physical scores in patients (n=94) with ECOG PS = 0 at baseline; B. Distribution of QOL-E physical scores in patients (n=44) with ECOG PS = 1 at baseline; C. Distribution of QOL-E physical scores in patients (n=10) with ECOG PS (more ...)