- © 2003 by American Society of Clinical Oncology
Quality of Life in Patients With Newly Diagnosed Chronic Phase Chronic Myeloid Leukemia on Imatinib Versus Interferon Alfa Plus Low-Dose Cytarabine: Results From the IRIS Study
- Elizabeth A. Hahn,
- G. Alastair Glendenning,
- Mark V. Sorensen,
- Stacie A. Hudgens,
- Brian J. Druker,
- Francois Guilhot,
- Richard A. Larson,
- Stephen G. O’Brien,
- Deborah G. Dobrez,
- Martee L. Hensley,
- David Cella and
- on behalf of the IRIS Investigators
- From the Center on Outcomes, Research and Education, Evanston Northwestern Healthcare, Evanston; The University of Chicago, Chicago, IL; Novartis Pharmaceuticals Corporation, East Hanover, NJ; Oregon Health Sciences University, Portland, OR; Centre Hospitalier Universitaire de Poitiers, Poitiers, France; The University of Newcastle, Newcastle, United Kingdom; and Novartis AG, Basel, Switzerland.
- Address reprint requests to Elizabeth A. Hahn, Center on Outcomes, Research and Education, 1001 University Pl, Ste 100, Evanston, IL 60201; email: e-hahn{at}northwestern.edu.
Abstract
Purpose: Quality of life (QOL) outcomes in patients with chronic myeloid leukemia (CML) were evaluated in an international phase III study.
Patients and Methods: Newly diagnosed patients with chronic phase CML were randomly assigned to imatinib or interferon alfa plus subcutaneous low-dose cytarabine (IFN+LDAC). Cross-over to the other treatment was permitted because of intolerance or lack of efficacy. Patients completed cancer-specific QOL (Functional Assessment of Cancer Therapy–Biologic Response Modifiers) and utility (Euro QoL-5D) questionnaires at baseline and during treatment (n = 1,049). The primary QOL end point was the Trial Outcome Index (TOI; a measure of physical function and well-being). Secondary end points included social and family well-being (SFWB), emotional well-being (EWB), and the utility score. Primary analyses were intention to treat with secondary analyses accounting for cross-over.
Results: Patients receiving IFN+LDAC experienced a large decline in the TOI, whereas those receiving imatinib maintained their baseline level. Treatment differences at each visit were significant (P < .001) and clinically relevant in favor of imatinib. Mean SFWB, EWB, and utility scores were also significantly better for those patients taking imatinib. Patients who crossed over to imatinib experienced a large increase in TOI; significant (P < .001) differences were observed between patients who did and did not cross over in favor of imatinib.
Conclusion: Imatinib offers clear QOL advantages compared with IFN+LDAC as first-line treatment of chronic phase CML. In addition, patients who cross over to imatinib from IFN+LDAC experience a significant improvement in QOL compared with patients who continue to take IFN+LDAC.
UNTIL RECENTLY, the only treatment choices for chronic myelogenous leukemia (CML) were stem-cell transplantation, which is limited to a small proportion of patients, or hydroxyurea-based or interferon alfa (IFNα)–based regimens.1,2 Treatment with IFNα has a deleterious effect on the patient’s quality of life (QOL), and is associated with physical toxicities such as fevers, chills, flu-like symptoms, hypotension, and fatigue as well as neurocognitive toxicities such as depression, impaired memory, and inability to concentrate.1,3–,7 Hydroxyurea-based treatment is well tolerated and has few side effects compared with IFNα, but is of limited efficacy, with no effect on disease progression or survival.8,9 Imatinib mesylate (Glivec, Gleevec; Novartis, Basel, Switzerland) is a new oral targeted therapy.10 In clinical trials,11–,13 imatinib has demonstrated a high level of efficacy in CML patients and is associated with significantly fewer toxicities, which is likely to translate into QOL benefits. The objective of this study was to compare the impact of QOL in patients receiving either imatinib or IFNα plus low-dose cytarabine (IFN+LDAC) during an international phase III study. This article reports the QOL results during the first 12 months of treatment.
PATIENTS AND METHODS
Eligibility
Patients were eligible if they were between 18 and 70 years of age and had chronic phase, Philadelphia chromosome-positive CML diagnosed within 6 months of study entry. They must have been previously untreated for CML with the exception of hydroxyurea and/or anagrelide. Of the 1,106 randomly assigned patients, 1,049 (95%) participated in the disease-specific QOL assessments. Preference-based QOL assessments were also conducted as part of a resource utilization substudy among 865 patients (78%).
Study Design and Treatments
The International Randomized IFN versus STI571 (IRIS) Study is a prospective, multicenter, open-label, phase III, randomized, cross-over design study, conducted in an outpatient setting. Patients were randomly assigned to receive either oral STI571 (imatinib) 400 mg daily or subcutaneous IFNα+LDAC. Patients were allowed to cross over to the other treatment arm if they experienced loss of response, increase in WBC, or intolerance of treatment; if treatment failed to achieve a major cytogenetic response at 12 months; or if treatment failed to achieve a complete hematologic response at 6 months. The primary end point of this trial was the duration of progression-free survival. Secondary end points were the rate and duration of complete hematologic response and major cytogenetic response, safety, tolerability, molecular remission, pharmacogenomics, pharmacokinetics, and QOL. The study was conducted in accordance with the Declaration of Helsinki and the study protocol was reviewed by the ethics committees or institutional review boards of all participating centers. All patients gave written informed consent according to institutional regulations. Additional details of the study and clinical results have been published elsewhere.14
QOL Assessments
Functional Assessment of Cancer Therapy–Biologic Response Modifiers (FACT-BRM).
Patient-rated disease-specific QOL was measured with the FACT-BRM,7,15 which is part of the Functional Assessment of Chronic Illness Therapy (FACIT) measurement system.16,17 Patients use a five-point scale (0 = not at all, 4 = very much) to rate their physical, functional, social and family, and emotional well-being as well as treatment-specific QOL concerns. The FACT-BRM is a valid 40-question instrument that has been translated into many languages and used in other oncology trials.6,15,18,19 Patients in the IRIS Study completed the FACT-BRM at baseline and months 1, 2, 3, 4, 5, 6, 9, and 12 in their preferred language (Dutch, English, French, German, Italian, Norwegian, Spanish, or Swedish). Raw scores were calculated by summing the item responses in each QOL domain such that a higher score indicates better QOL.17 A measure of physical function and well-being (the Trial Outcome Index [TOI]) was created as the primary QOL end point using the physical, functional, and treatment-specific subscales (27 items; score range, 0 to 108). This approach is often recommended for pharmacologic interventions because the TOI provides the most sensitive summary end point.17 Social and family well-being (SFWB; seven items; range, 0 to 28) and emotional well-being (EWB; six items; range, 0 to 24), which together with the TOI comprise the 40-item FACT-BRM, were evaluated as secondary QOL end points. Rasch measurement models were used to evaluate the psychometric properties of the FACT-BRM.20–,22 Measurement models create a ruler by which to measure a latent trait QOL. Just like a ruler used to measure a person’s height, the QOL ruler should function the same regardless of the patient’s cultural background or the timing of the measurement. Overall, measurement properties remained stable across languages and time.23 Nearly all baseline questionnaires were completed before or on the same day as random assignment to treatment. A small number were completed after random assignment to treatment, but before the start of treatment (n = 30). Mean baseline QOL scores did not differ according to the timing of the assessment (before, same day as, or after random assignment to treatment).
EuroQoL-5D.
A different measure of QOL that reflects the patient’s value of her or his current health state was assessed at baseline and months 3, 6, 9, and 12 with the EuroQol-5D (EQ-5D).24,25 The instrument has five dimensions (mobility, self-care, usual activities, pain or discomfort, and anxiety or depression) and one overall rating of current health. A summary utility score was created as a secondary QOL end point on the basis of responses to the five dimensions (range, 100 to −100).26,27 A score of 100 denotes current health that is valued to be equivalent to full health, a score of 0 denotes current health that is valued to be equivalent to death, and a score less than 0 denotes current health that is valued to be worse than death.
Statistical Analyses
The primary analyses were intention to treat (ITT); that is, data were analyzed according to the original treatment arm assignment regardless of cross-over. Secondary analyses accounted for cross-over. To evaluate the influence of missing data, patients were classified into three groups: those who discontinued treatment before month 12 (n = 182), those who completed a FACT-BRM questionnaire at month 12 (n = 756), and those who did not complete a questionnaire at month 12 (n = 111). The proportions in each category were compared across treatment arms using a χ2 test. Logistic regression models and other strategies were used to evaluate whether data were missing at random.28–,30
A multivariate mixed-effects growth curve model using all available data was chosen for the longitudinal analyses, with an unstructured covariance pattern to account for the correlations within patients.31–,33 The model describes the rate of change in QOL, allowing for the possibility that change may not be constant over time. It also takes account of between-patient variability by incorporating each patient’s rate of change into the model. A pattern-mixture technique was implemented to assist in the analysis of missing data that should not be ignored.33–,35 When data are missing, it is often because of reasons related to the patient’s QOL; for example, discontinuation of treatment because of toxicities. Separate longitudinal models for patient groups were created (dropouts v those remaining on-study) and a weighted average was derived for each treatment group. Patient age, sex, previous treatment with hydroxyurea (yes v no) and baseline Sokal-Hasford risk stratification scores36,37 were evaluated to determine whether they had any independent effect on the QOL scores. A likelihood ratio test was used to assess their contributions. Growth curve models were also used to evaluate cross-over effects by including a time-dependent covariate (cross-over, yes v no). Model estimates of the treatment arm means were plotted for 12 months. Sensitivity analyses were performed to ensure that treatment arm effects were consistent across different analytic methods and to evaluate the range of possible treatment arm differences.
Each patient’s average QOL score during treatment was calculated for the TOI, SFWB, and EWB, and compared across treatment arms using analysis of covariance adjusting for baseline. Treatment arm differences were also expressed as effect sizes; that is, mean difference divided by the baseline SD. The effect size provides a standardized value for the size of treatment differences. Experience with the FACIT instruments has shown that an effect size of 0.33 or greater is clinically meaningful.17 Predefined criteria were used for clinically relevant QOL change38 and individual TOI change scores at months 1, 3, 6, 9, and 12 were classified into three categories: clinically relevant improvement (increase of 5 or more from baseline), clinically relevant decline (decrease of 5 or more from baseline), or no change. The change proportions were compared across treatment arms using the Mantel-Haenszel χ2 test for an ordinal response. Results were summarized on the basis of the ITT approach and were also restricted to first-line treatment (up to the point of cross-over) or second-line treatment (after cross-over).
The summary utility score from the EQ-5D was evaluated using ordinary least squares regression models to estimate the effects of treatment on the mean utility score. Separate analyses were performed at each assessment, and treatment effects were adjusted for patient age, sex, and ethnicity. Analyses were performed using the ITT approach and also were restricted to first-line treatment. Similar analyses were conducted using logistic regression models to determine the influence of treatment on the likelihood that a patient would report that her or his current health was equivalent to perfect health (utility score = 100 v other). All analyses were performed with SAS (Version 8.0, SAS Institute, Cary, NC) or STATA (Version 7.0, STATA Software, College Station, TX) and there were no adjustments for multiple comparisons.
RESULTS
Patient Characteristics
Patients in the two treatment arms were comparable in terms of most sociodemographic and clinical characteristics (Table 1⇓). Median time from diagnosis to random assignment to treatment was 2.1 months in the imatinib arm and 1.7 months in the IFN+LDAC arm (P < .05), and the majority of patients had the highest functioning performance status rating. There were comparable proportions in the Sokal-Hasford risk stratification groups, although this information was unavailable for approximately 30% of patients in each arm.
Baseline Characteristics for Quality of Life Study Population
Missing Data
The baseline FACT-BRM was completed by 95% of patients in both the imatinib and IFN+LDAC treatment arms. If all randomly assigned patients had completed a baseline questionnaire, it is likely that the randomization process would have ensured even greater treatment arm comparability of baseline QOL scores than that observed (Figs 1 to 4⇓⇓⇓⇓). At month 12, an assessment was completed by 434 patients (82%) randomly assigned to receive imatinib and 322 (62%) randomly assigned to receive IFN+LDAC; excluding patients who died (n = 2) or discontinued study treatment (n = 180) before month 12, these proportions increased to 88% and 86%, respectively. More patients discontinued treatment in the IFN+LDAC arm compared with the imatinib arm (28% v 7%, respectively; P < .001). Within and between treatment arms, baseline demographic and clinical characteristics were comparable for those who discontinued treatment and those who did not, although those who discontinued treatment in the IFN+LDAC arm were slightly younger (P = .109) and those who remained enrolled onto the study were younger in the imatinib arm compared with the IFN+LDAC arm (P = .028). For this reason, age was included in all of the multivariate models. Lower TOI scores were associated with a higher probability of subsequent missing data; that is, patients with poorer QOL were less likely to complete the questionnaires. In addition, patients who discontinued treatment exhibited a different pattern of change over time than those who remained enrolled onto the study. Overall, 315 patients in the imatinib arm (59%) and 198 in the IFN+LDAC arm (38%) completed all nine scheduled assessments. The analysis strategies developed for this study helped to compensate for intermittent missing data as well as missing data caused by dropout.
Estimated mean Trial Outcome Index scores by treatment arm, adjusted for missing data (intent-to-treat approach). P values are for difference in treatment arm means at each scheduled administration of the Functional Assessment of Cancer Therapy-Biologic Response Modifiers. IFN+LDAC, interferon alfa plus subcutaneous low-dose cytarabine.
Estimated mean social or family well-being scores by treatment arm, adjusted for missing data (intent-to-treat approach). P values are for difference in treatment arm means at each scheduled administration of the Functional Assessment of Cancer Therapy-Biologic Response Modifiers. IFN+LDAC, interferon alfa plus subcutaneous low-dose cytarabine.
Estimated mean emotional well-being scores by treatment arm, adjusted for missing data (intent-to-treat approach). P values are for difference in treatment arm means at each scheduled administration of the Functional Assessment of Cancer Therapy-Biologic Response Modifiers. IFN+LDAC, interferon alfa plus subcutaneous low-dose cytarabine.
Estimated mean Trial Outcome Index scores by treatment arm and cross-over status. P values are for interferon alfa plus subcutaneous low-dose cytarabine (IFN+LDAC) after cross-over and imatinib (top row), IFN+LDAC first-line treatment (bottom row). Results not shown for imatinib cross-overs (n = 7) or IFN+LDAC cross-overs before month 3 (n = 3).
Effect of Treatment on the FACT-BRM Scores: ITT Approach
After 1 month of treatment, there was a large decline in the primary QOL end point (TOI) for patients in the IFN+LDAC arm, which persisted through month 2 (Fig 1⇑). When the ITT approach was used, the mean TOI increased after month 2 but never recovered to baseline levels. Mean TOI scores for patients in the imatinib arm changed very little over time, although there was a trend of lower TOI scores among those who discontinued treatment (n = 37). There were large differences in mean scores between treatment arms from months 1 through 12 (P < .001). Exploratory analyses of the three subscales that comprise the TOI showed similar patterns. The average TOI score during treatment was 84.3 in the imatinib arm and 67.1 in the IFN+LDAC arm (P < .001; effect size = 1.05). To better understand the meaning of the TOI scores, the average response to each item during treatment is shown in Fig 5⇓. Patients taking imatinib reported better daily functioning and well-being, less fatigue, and milder emotional or cognitive complaints compared with patients in the IFN+LDAC arm. Physical side effects such as fever, chills, and nausea were not bothersome for imatinib patients compared with IFN+LDAC patients. The extent of pain, sweating, and feeling ill was comparable in the two treatment arms and was only experienced “a little bit” in both arms.
Average response to items on the Trial Outcome Index (TOI) during treatment (intention to treat approach).‡ IFN + LDAC, interferon alfa plus subcutaneous low-dose cytarabine.
Independent of treatment effects, patient sex was associated with TOI scores; specifically, the average score for men was approximately 4 points higher (better) than the score for women. Age was included in all models, but was not significantly associated with TOI scores. Baseline Sokal-Hasford scores were highly associated with TOI scores, but were excluded from further analyses because of a large amount of missing data. Previous hydroxyurea treatment usage was marginally associated with better TOI.
The pattern of change in SFWB was similar to that observed in the primary QOL end point; that is, mean scores changed little for patients in the imatinib arm, whereas there was a large decline for patients in the IFN+LDAC arm (Fig 2⇑). Although most of the differences between treatment arms at months 1 through 12 were statistically significant (P < .05), they were small. The average SFWB score during treatment was 22.7 for patients in the imatinib arm and 21.6 for patients in the IFN+LDAC arm (P = .005; effect size = 0.23). The pattern of change was very different for EWB, which increased dramatically for patients in the imatinib arm and only slightly for patients in the IFN+LDAC arm (Fig 3⇑). The differences between treatment arms at months 1 through 12 were statistically significant (P < .05) and clinically important. The average EWB score during treatment was 19.3 for patients in the imatinib arm and 17.4 for patients in the IFN+LDAC arm (P < .001; effect size = 0.43). Age was not associated with either of these end points. The average SFWB score was 1.5 points higher for men compared with women (P < .001), but there were no sex differences in EWB. Results of sensitivity analyses supported the findings of the primary analyses.
Effect of Treatment on the FACT-BRM Scores: Accounting for Cross-Over
To determine whether the gradual increase in the QOL end points for patients in the IFN+LDAC arm (ITT approach) was the result of crossing over to imatinib, cross-over status was incorporated into the multivariate models. Estimated mean scores for the TOI are shown in Fig 4⇑ and are arranged separately by cross-over status and treatment arm. These results show that patients who crossed over to imatinib experienced a large increase in TOI scores, whereas those who continued to receive IFN+LDAC did not improve. There were statistically significant (P < .001) differences between patients who continued to receive IFN+LDAC and those who crossed over to imatinib. In addition, mean TOI scores at month 12 were comparable for patients randomly assigned to the imatinib arm and those who crossed over from IFN+LDAC to imatinib (P = .088). Similar patterns were observed for SFWB and EWB.
Clinically Important Changes in the TOI
According to available data at baseline and months 1, 3, 6, 9, and 12, 52% to 73% of patients in the IFN+LDAC arm experienced a clinically relevant decrease of 5 or more points from baseline in their TOI scores, compared with only 22% to 29% of patients in the imatinib arm (ITT approach; P < .001; Table 2⇓, last two columns). The proportions of patients with clinically relevant improvement in TOI scores were 9% to 25% for IFN+LDAC and 29% to 43% for imatinib. The proportions of patients with clinically relevant decline were approximately 70% at all visits for first-line treatment with IFN+LDAC (Table 2⇓).
Clinically Relevant Changes From Baseline in the Trial Outcome Index (TOI)
Effect of Treatment on the EQ-5D Preference-Based QOL Scores
Mean utility scores were significantly higher at each assessment after baseline for patients in the imatinib arm compared with the IFN+LDAC arm for first-line treatment and the ITT approach (P < .001; Table 3⇓). In addition, patients in the imatinib arm were significantly more likely to report that their current health was equivalent to perfect health (P < .001; results not shown).
Estimated Mean EQ-5D Utility Scores by Treatment Arm
DISCUSSION
This is the largest prospective evaluation of QOL during treatment for chronic phase CML and the first published report that is based on internationally validated QOL questionnaires. During the first 12 months of treatment, patients receiving IFN+LDAC experienced a large decline in the primary QOL end point, whereas imatinib preserved physical function and well-being and was not associated with the common adverse physical, emotional, and cognitive toxicities linked with IFN. The difference between treatment arms (17.2 points) was not only statistically significant, it was also three times the magnitude of a clinically meaningful difference (5 points), making this one of the largest treatment group differences in QOL reported in cancer trials to date (effect size = 1.05). Imatinib was superior to IFN+LDAC on the basis of the ITT approach. In addition, patients who crossed over to imatinib (second-line treatment) after IFN+LDAC reported higher QOL than those who continued to receive IFN+LDAC, and increased their QOL at 12 months to a level comparable with patients in the arm randomly assigned to receive imatinib. SFWB, EWB, and preference-based QOL (utility scores) were also significantly better in the imatinib-treated group compared with those treated with IFN+LDAC. Results of sensitivity analyses supported the findings of the primary analyses. Treatment arm comparisons of individual patient changes in the TOI also supported the primary findings. Clinically significant decline was reported by a higher proportion of patients receiving IFN+LDAC than in patients receiving imatinib. In addition, the proportion of patients receiving imatinib who reported clinically significant improvement on the FACT-BRM TOI was two to three times as large as the proportion of patients with improvement in the IFN+LDAC arm (Table 2⇑).
Although QOL is increasingly recognized as a valued end point of cancer care,39–,42 prospective evaluation of QOL during or after treatment for CML is rare. The limited available information on QOL has been obtained from retrospective evaluations of bone marrow transplant survivors,43 non–cancer-specific questionnaires,44 or quality-adjusted life-year estimates that are based on health state utilities obtained from previous research.45–,48 QOL information from the IRIS Study is therefore unique and highly advantageous because it was prospectively collected during a randomized treatment study, cancer-specific QOL concerns and patient-reported valuations of their current health state were gathered, and patients in 14 countries participated. Because all of the major health-related QOL domains were assessed, the results from this study can provide clinicians, researchers, and patients with a greater understanding of the CML patient’s experience during the first 12 months of treatment. Specifically, patients receiving imatinib reported better daily functioning and well-being compared with patients receiving IFN+LDAC. In addition, patients receiving imatinib reported less fatigue, milder emotional or cognitive complaints, better social functioning, more positive emotional well-being, and a greater tendency to consider their current health to be equivalent to a state of perfect health.
Perhaps as important as the findings regarding treatment differences are the patterns of change observed in the different QOL domains. The initial decrease in mean TOI scores for the patients in the IFN+LDAC treatment arm, followed by partial recovery, is consistent with the pattern observed in a recent randomized phase III trial in patients with advanced renal cell carcinoma.6 In that study, both treatment arms received IFN and both showed a sharp decrease in the TOI. This indicates that the TOI is a reliable indicator of physical function and well-being because it provided consistent results for the experiences of two different disease populations who both received IFN. Whereas physical function and well-being, and social and family well-being were both preserved throughout imatinib treatment, emotional well-being actually increased. Patients felt better able to cope with their illness, less worried, and less sad. Improvement in emotional well-being during cancer treatment is not uncommon.49–,52 It may be related to knowledge about a favorable clinical response, the experience of having few treatment toxicities, or even hearing good publicity about their treatment. Emotional well-being sometimes improves even when toxicities and burden of disease increase. These findings make it all the more important to measure multiple QOL dimensions to fully capture the patient’s treatment experience.
Among patients alive and enrolled onto the study, the FACT-BRM questionnaire was completed by 95% at baseline and 88% at month 12, which is only slightly lower than cooperative group standards of 100% and 90%, respectively.53,54 Among all patients randomly assigned to treatment, however, only 82% in the imatinib arm and 62% in the IFN+LDAC arm completed a QOL questionnaire at month 12, and smaller proportions completed all nine scheduled assessments through month 12, primarily because more patients dropped out of the IFN+LDAC arm. The analysis strategies that were implemented overcame some of the bias resulting from missing data by using all available assessments and by weighting the estimates of treatment arm effects by the proportions of patients who discontinued treatment and those who remained enrolled onto the study. Although the results should be interpreted with some caution because complete data were not available for all patients, given that less-healthy patients were more likely than others to fail to provide follow-up QOL data and there were more treatment discontinuations in the IFN+LDAC arm, the reported differences may actually underestimate the gap between treatment arms.
In conclusion, imatinib offers clear QOL advantages compared with IFN+LDAC as first-line treatment of chronic phase CML. In addition, patients who cross over to imatinib experience a significant improvement in their QOL and function better than those who continue to receive IFN+LDAC.
APPENDIX
The following investigators and institutions participating in this study.
The Study Management Committee: B.J. Druker, Portland, OR; F. Guilhot, Poitiers, France; R.A. Larson, Chicago, IL; and S.G. O’Brien, Newcastle, United Kingdom.
The Protocol Working Group advised on the protocol, trial conduct, and publication policy and comprised M. Baccarani, Bologna; G. Saglio, Orbassano, Italy; F. Cervantes, Barcelona, Spain; J. Cornelissen, Rotterdam, the Netherlands; T. Fischer, Mainz; A. Hochhaus, Mannheim, Germany; T. Hughes, Adelaide; K. Taylor, Brisbane, Australia; K. Lechner, Wien, Austria; J.L. Nielsen, Aarhus C, Denmark; J. Reiffers, Pessac; P. Rousselot, Paris, France; J. Shepherd, Vancouver, British Columbia, Canada; B. Simonsson, Uppsala, Sweden; A. Gratwohl, Basel, Switzerland; J.M. Goldman, London, United Kingdom; H. Kantarjain, Houston, TX; and G. Verhoef, Leuven, Belgium.
In addition to the authors, the following investigators participated in the IRIS Study. Australia: S. Durrant, Brisbane; A. Schwarer, J. Seymour, A. Grigg, Melbourne; D. Joske, Perth; D. Ma, C. Arthur, K. Bradstock, D. Joshua, Sydney. Belgium: A. Louwagie, Brugge; P. Martiat, N. Straetmans, Bruxelles; A. Bosly, Yvoir. Canada: C. Shustik, D.-C. Roy, Montreal; J. Lipton, Toronto; D. Forrest, Halifax; I. Walker, Hamilton; M. Rubinger, Winnipeg; I. Bence-Bruckler, Ottawa; D. Stewart, Calgary; M. Kovacs, London; A.R. Turner, Edmonton. Denmark: H. Birgens, Herlev; O. Bjerrum, Copenhagen. France: T. Facon, Lille; J-L. Harousseau, Nantes; M. Tulliez, Créteil; A. Guerci, Vandoeuvre-les-Nancy; D. Blaise, Marseille; F. Maloisel, Strasbourg; M. Michallet, Lyon. Germany: D. Hossfeld, Hamburg; R. Mertelsmann, Freiburg; R. Andreesen, Regensburg; C. Nerl, C. Peschel, München; M. Freund, Rostock; N. Gattermann, Düsseldorf; K. Hoeffken, Jena; G. Ehninger, Dresden; M. Deininger, Leipzig; O. Ottmann, Frankfurt; S. Fruehauf, Heidelberg; A. Neubauer, Marburg; P. Le Coutre, Berlin; W. Aulitzky, Stuttgart. Italy: R. Fanin, Udine; G. Rosti, G. Martinelli, Bologna; F. Mandelli, Roma; E. Morra, Milano; A. Carella, Genova; M. Lazzarino, Pavia; M. Petrini, Pisa; P. Rossi Ferrini, Firenze; F. Nobile, Reggio Calabria; V. Liso, Bari; F. Ferrara, Napoli; V. Rizzoli, Parma; G. Fioritoni, Pescara. Netherlands: G. Ossenkoppele, Amsterdam. New Zealand: P. Browett, Auckland. Norway: T. Gedde-Dahl, J.-M. Tangen, Oslo; I. Dahl, Tromso. Spain: J. Odriozola, J.L. Steegman, J. Diaz, M.N. Fernández, Madrid; J.C. Hernández Boluda, Valencia; C. Cañizo, Salamanca; A. Sureda, Barcelona; A. Granena, Llobregat. Sweden: L. Stenke, C. Paul, Stockholm; M. Bjoreman, Orebro; C. Malm, Linköping; H. Wadenvik, Göteborg; P-G. Nilsson, Lund; I. Turesson, Malmo. Switzerland: U. Hess, Sankt Gallen; M. Solenthaler, Bern. United Kingdom: N. Russel, Nottingham; G. Mufti, J. Cavenagh, London; R.E. Clark, Liverpool; A.R. Green, Cambridge; T.L. Holyoake, Glasgow; G.S. Lucas, Manchester; G. Smith, Leeds; D.W. Milligan, Birmingham; S.J. Rule, Plymouth; A.K. Burnett, Cardiff. United States: R. Moroose, Orlando; L. Kalman, Miami, FL; M. Wetzler, Buffalo; J. Gabrilove, E. Berman, R. Silver, New York; S. Graziano, Syracuse, NY; J. Bearden, Spartanburg, SC; R. Brown, St Louis, MO; M. Lobell, Tucson, AZ; S. Cataland, Columbus; M. Kalaycio, Cleveland; P. Kuebler, Columbus; H. Gross, Dayton, OH; I. Rabinowitz, Albuquerque, NM; B. Meisenberg, Baltimore, MD; K. Thompson, Montgomery; P. Emanuel, Birmingham, AL; P. Cobb, Billings, MT; R. Bhatia, Duarte; D. Irwin, Berkeley; A. Bashey, La Jolla, CA; S. Dakhil, Wichita, KS; B. Issell, Honolulu, HI; S. Pavletic, Omaha, NE; E. Layhe, East Lansing; M. Shurafa, Detroit, MI; P. Butera, Providence, RI; J. Glass, Shreveport; R. Veith, New Orleans, LA; J. Moore, Durham; B. Powell, Winston-Salem; S. Limentani, Charlotte; T. Shea, Chapel Hill, NC; B. Grant, Burlington, VT; H. Niell, D. Strickland, Memphis; H. Burris, S. Cooper, Nashville, TN; R. Herzig, D. Stevens, Louisville, KY; B. Peterson, Minneapolis, MN; D. Stirewalt, Seattle, WA; W. Samlowski, Salt Lake City, UT; T. Seay, Atlanta, GA; L. Akard, Beech Grove, IN; G. Smith, Farmington, NM; P. Becker, Worcester, MA; S. DeVine, M. Tallman, Chicago, IL; R. Hart, Milwaukee, WI; J. Wade, Decatur, GA; M. Brunvand, Denver, CO; R. Shadduck, M. Agha, Pittsburgh, PA; H. Safah, New Orleans, LA; M. Rubenstein, Campbell, CA; R.Collins, H. Holmes, Dallas, TX; A. Keller, Tulsa, OK; R. Stone, Boston, MA; A. Pecora, Hacksensack, NJ.
Acknowledgments
We thank coinvestigators, medical, nursing, and research staff, trial monitors, data managers, and programmers for their contributions. We also thank Jane Brandman for comments on the manuscript, and Jennifer Beaumont for statistical assistance. We thank all patients who participated in this trial and allowed us to assess their quality of life.
Footnotes
-
Supported by Novartis Pharma.
-
Presented in part at the 7th Congress of the European Haematology Association (Florence, Italy, June 2002) and at the 2002 Annual Meeting of the American Society of Hematology (Philadelphia, PA, December 2002).
-
E.H., M.S,, S.H., B.D., F.G., R.L., S.O., D.D., and D.C. are consultants to Novartis Pharma; G.G. and M.H. are employees of Novartis Pharma.
- Received December 26, 2002.
- Accepted February 27, 2003.
