Assessment of Quality of Life in the Supportive Care Setting of the Big Lung Trial in Non–Small-Cell Lung Cancer
http://www.100md.com
《临床肿瘤学》
the Clinical Trials Research Unit, University of Leeds
New Cross Hospital, Wolverhampton
Cancer Research United Kingdom and University College London Cancer Trials Centre
Medical Research Council Clinical Trials Unit
St Bartholomew's Hospital
The Middlesex Hospital, London
Stobhill National Health Service Trust, Glasgow
Glenfield Hospital, Leicester
York Health Economics Consortium, University of York, United Kingdom
ABSTRACT
PURPOSE: The Big Lung Trial (BLT) was a large, pragmatic trial to evaluate the addition of chemotherapy to primary treatment (ie, surgery, radical radiotherapy, or supportive care) in non–small-cell lung cancer (NSCLC). In the supportive care group, there was a small but significant survival benefit in patients treated with chemotherapy compared with supportive care alone (no chemotherapy). A substudy was undertaken to evaluate the quality of life (QoL) implications of the treatment options. QoL was assessed using European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaires C30 (QLQ-C30) and LC17, and daily diary cards.
PATIENTS AND METHODS: EORTC QLQ-C30 and LC17 were collected at 0, 6 to 8, 12, 18, and 24 weeks. Diary cards were completed during the first 12 weeks of the study. The primary end point was global QoL at 12 weeks.
RESULTS: A total of 273 patients were randomly assigned: 138 to no chemotherapy and 135 to chemotherapy. There was no evidence of a large detrimental effect on QoL of chemotherapy. No statistically significant differences in global QoL or physical/emotional functioning, fatigue and dyspnea, and pain were detected at 12 weeks. Higher rates of palliative radiotherapy in the no chemotherapy arm may have lessened differences in QoL. Global QoL, role functioning, fatigue, appetite loss, and constipation were prognostic indicators of survival at 12 weeks.
CONCLUSION: There were no important adverse effects of chemotherapy on QoL.
INTRODUCTION
Advanced non–small-cell lung cancer (NSCLC) is associated with a poor prognosis. The median survival of untreated patients with stage IV disease is approximately 4 months.1 It is also associated with multiple symptoms,2 and the survival advantage offered by interventions such as chemotherapy is modest.1 In this context, the impact of treatment on quality of life (QoL) and symptoms becomes of major importance for the individual patient and the development of management policies. The NSCLC Collaborative Group meta-analysis revealed that little attention was given to these issues in the large number of chemotherapy-based trials included in the analysis.1
The Big Lung Trial (BLT) was established with the main aim of confirming the survival advantage seen in the NSCLC Collaborative Group meta-analysis when chemotherapy was added to surgery, radical radiotherapy, or supportive care. At the start of the BLT, chemotherapy was being administered to only a minority of patients with advanced NSCLC, and therefore there was an opportunity to randomly assign patients to best supportive care with chemotherapy (chemotherapy group) or without chemotherapy (no chemotherapy group), provided both the physician and patient had doubts about the efficacy of chemotherapy. Thus, the results of the meta-analysis could be re-examined, but with attention focused on the QoL of participants. In the supportive care group, chemotherapy-related improvement for median survival of approximately 8 weeks and an increase of survival at 1 year from 20% to 28% was confirmed, compared with the no chemotherapy arm.3 A total of 725 patients were recruited to the supportive care setting, and this report relates to the subgroup of these patients in whom QoL was measured. European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire C30 (QLQ-C30)4 and the lung cancer–specific module LC17, together with a daily diary card, were used to provide information on the changes in QoL during treatment.
PATIENTS AND METHODS
Eligibility
Patients were eligible for the main trial if they fulfilled local criteria for diagnosis of NSCLC, were to be treated with supportive care, were fit enough to receive chemotherapy, and had no clear contraindications for chemotherapy (eg, inadequate renal function). Patients were ineligible if they had concurrent malignancy or a history of malignancy other than nonmelanomatous skin cancer within the last 3 years. In addition, patients participating in the QoL substudy had to be able to start chemotherapy within 1 week of random assignment; have consented to the QoL assessments; be able to read/complete the assessments; have a life expectancy of at least 8 weeks; and have completed baseline questionnaires before random assignment.
Treatment and Collection of QoL Data
Patients were randomly assigned to receive supportive care with or without chemotherapy. Choice of chemotherapy was made before random assignment. Clinicians could choose any one of four regimens: cisplatin 80 mg/m2 and vindesine 3 mg/m2 days 1 and 8; mitomycin 6 mg/m2, ifosfamide 3 g/m2, and cisplatin 50 mg/m2; mitomycin 6 mg/m2, vinblastine 6 mg/m2, and cisplatin 50 mg/m2; or vinorelbine 30 mg/m2 days 1 and 8 and cisplatin 80 mg/m2. Each regimen was administered for up to three cycles, once every 3 weeks.
The self-administered EORTC QLQ-C304 and the lung cancer supplementary questionnaire (LC17) were completed in the clinic at baseline (before random assignment) and at 6 to 8, 12, 18, and 24 weeks after random assignment. The QLQ-LC17 was under development during the trial and has since been validated (but not yet published) and reduced to 14 items. All 17 items were used in this trial (see Appendix 1). Both questionnaires asked patients to rate their QoL during the last week.
A diary card pack, containing three 4-week diary cards (to be completed daily and to cover the period of chemotherapy for those patients allocated to chemotherapy) was given to all patients at random assignment. The diaries were based on the Medical Research Council daily diary card5 (Appendix 2).
Trial Design
The BLT-QoL study was a multicenter, prospective, randomized study. The Cancer Research UK, University College London Cancer Trials Centre (London, United Kingdom), and the Medical Research Council Clinical Trials Unit carried out random assignment by telephone. Random assignments were minimized by center, WHO performance status, sex, and histology.
A survey of participating clinicians highlighted those elements of QoL that the clinicians believed were most important to patients. On the basis of this survey, the primary end point for the BLT-QoL study was defined as the difference in global QoL at 12 weeks. Clinicians indicated that only large differences in global QoL would be of clinical interest. Sample size was therefore based on a large difference between the two groups in global QoL at 12 weeks. Using methods described by King6 and Osoba et al,7 this translated into the detection of an effect size of approximately 0.4 (mean difference between groups divided by common standard deviation). To detect a standardized effect size of 0.4 (two-sided 5% significance level, 80% power), 200 patients were required. An effect size of 0.5 would require 128 patients. The sample size was set at 300 to allow for noncompliance. In practice, the main trial was closed on the planned closure date, at which time 273 patients had entered the BLT-QoL study. However, compliance at 12 weeks was 71%; therefore, the required sample size was achieved. Clinicians also highlighted emotional and physical functioning, fatigue, dyspnea, and pain at 12 weeks for additional study. All other domains and symptoms were classified as secondary end points.
Statistics
The EORTC QLQ-C30 was scored according to EORTC guidelines. Higher scores indicated better functioning, whereas higher scores for symptom scales indicated increased severity. LC17 was scored according to the QLQ-LC14 manual (three multi-item scales and seven single-symptom scores); the three additional questions were summarized separately.
Primary and highlighted end points were compared using multilevel repeated-measures modeling accounting for data at all time points (assuming missing data at random [MAR] and allowing for time, treatment, treatment-time interaction, and adjusting for baseline QoL [all fixed effects], patient, and patient-time [random effects]). Minimization factors, age, and marital status were added as covariates in the above. Given that inclusion of covariates made little difference to interpretation, differences between treatment estimates are presented from the multilevel model adjusted for baseline QoL only. Pattern mixture multilevel model was also fitted by categorizing patients into strata based on clinical information, stage (< 4, 4), age (< 65, 65), and WHO score at 12 weeks (< 2, 2; assuming data are missing not at random [MNAR]).8 Pattern mixture models for bivariate (baseline and 12 week) data were also fitted using a variety of restrictions reflecting the missing data pattern ranging from complete case missing variable restriction (MAR) to Brown's protective restriction (MNAR).9 To ease clinical interpretation, treatment estimates from a logistic regression model investigating the proportion of patients improving for each of the primary and highlighted end points at 12 weeks, adjusted by baseline score, are also presented (assuming MAR).10-12 As based with King,6 improvement in individual patients and a large difference between groups was defined as +10 points for global QoL, +25 points for physical functioning, +7 points for emotional functioning, and +20 points for dyspnea, fatigue, and pain.
The analysis approach used summarized data in different ways to encompass as much information as possible, and allowed for differing assumptions about the missing data (ignorable and not ignorable); therefore, some testing of the sensitivity of the analysis to the missing data was undertaken.13
Means and 95% CIs, adjusted for baseline, were calculated for all other domains of the QoL questionnaires. No statistical testing was carried out on these data. The primary end point analyses were carried out at a 5% significance level; all other analyses were carried out at 1% significance level.
An additional analysis was performed to validate results from Montazeri et al,14 which showed that pretreatment global QoL was a prognostic factor for survival at 3 months, and to assess further other baseline QoL factors in a hypothesis-generating exercise. Univariate relationships were investigated using a log-rank test. QoL scales/symptoms identified at the 10% level were then considered in a multivariate model. A forward and backward stepwise Cox proportional hazards model was performed and 5% significance was used as the criteria for inclusion. Known prognostic factors in NSCLC (age, sex, performance status, histology, and stage of disease) and treatment group were also included in the multivariate model to assess the additional impact of the QoL. Survival time was calculated from date of the baseline questionnaire to either date of death (all causes) for those who died or 12 weeks for those who lived beyond 12 weeks. For those not observed to 12 weeks, survival time was the date last seen alive. Proportional hazards were evaluated by analysis of the hazard function plots.
Daily diary cards were descriptively compared between groups by examining plots of the percentage of patients reporting moderate/severe symptoms each day from random assignment. Moderate or severe symptoms were defined as grade 3 or 4 nausea, vomiting, tiredness, and breathlessness; grade 4 or 5 mood, overall condition, appetite, and activity; and grade 3, 4, or 5 difficulty in swallowing.
Lines to indicate approximate timing of chemotherapy were added to aid interpretation. Missing data were not taken into account in these summaries, and hence results should be interpreted with caution.
Compliance with questionnaires was calculated as the number of completed questionnaires divided by the total number of questionnaires expected according to the protocol (excluding deaths). Calculations were based on a period of 2 weeks around the expected date of completion (1 week for the 6- to 8-week questionnaire). All statistical analyses were performed with SAS software, version 8.2 (SAS Institute, Cary NC).15
Ethics
The Northern and Yorkshire Multicenter Research Ethics Committee and Local Research Ethics Committees approved the protocol. An independent Data Monitoring and Ethics Committee advised on the conduct of the study.
RESULTS
Patient Characteristics
The BLT main trial started recruiting in November 1995; the QoL study started in March 1998 (once funding was obtained). Between March 1998 and November 2001, 273 patients from 33 (32 in the United Kingdom, one from Australia) of 62 centers in the supportive care part of the main trial, and 46 clinicians entered the QoL substudy. One hundred thirty-eight patients were randomly assigned to the no chemotherapy group and 135 patients were randomly assigned to the chemotherapy group, representing 38% of all patients in the supportive care part of the main BLT, given that center participation was not compulsory (58% during the QoL recruiting period). Treatment groups were well balanced with respect to patient baseline characteristics (Table 1). The baseline characteristics were also similar for those patients in the supportive care part of the main trial but not in the QoL study.
Baseline QoL score across all scales, and for all but three symptoms (constipation, diarrhea, and financial difficulties) from EORTC QLQ-C30, were better for the chemotherapy group compared with the no chemotherapy group (Table 2). No notable differences were apparent for LC17 (Table 2).
Treatment Details
Of the patients receiving chemotherapy, 30 (22%) received mitomycin, ifosfamide, and cisplatin; 63 (47%) received mitomycin, vinblastine, ifosfamide, and cisplatin; 42 (31%) received vinorelbine and cisplatin; and none received the cisplatin and vindesine regimen. Sixty-four percent of patients received three cycles, 12% received two cycles, 18% received one cycle, and 7% received none. Grade 3/4 toxicity was suffered by 30% of patients; these patients mainly experienced nausea/vomiting, neutropenic fever, hematologic, neurologic, and renal toxicities. Similar numbers of patients in both arms received nonthoracic radiotherapy (1%); however, 50% of the no chemotherapy group compared with 21% of the chemotherapy group received thoracic radiotherapy during the first 24-week period (47% v 6% during the first 8 weeks from random assignment).
Compliance
Compliance with EORTC questionnaires was high at baseline but decreased at follow-up, although 60% of patients completed questionnaires at week 24 (Table 3). There were few missing items on completed questionnaires. Compliance at 12 weeks was better; because as this time point was of primary interest patient response was more intensively pursued. Questionnaires at both baseline and 12 weeks were completed by 136 patients (50%). Reasons for noncompletion of questionnaires are listed in Table 4 and were balanced across groups.
QoL Results
Primary and highlighted outcomes. In keeping with the main trial results, the 1-year survival rate for the QoL patients increased in the chemotherapy group to 29% from 21% in the no chemotherapy group. There were no statistically significant differences between treatment groups for the QoL primary or highlighted end points. The various multilevel and pattern mixture analyses had CIs that included the possibility of a large positive impact (clinically important) on global QoL in favor of chemotherapy, and the possibility of a large detrimental effect was only apparent using Brown's restriction. At 12 weeks large positive or negative effects of chemotherapy on fatigue and dyspnea were ruled out in all analyses, but results for emotional functioning did not rule out the possibility of large effects in either direction. For physical functioning, some of the analyses assuming MNAR indicated a large positive effect for chemotherapy, but all ruled out large negative effects. For pain, large negative effects were ruled out and the possibility of large positive effects was apparent regardless of assumptions made about missing data. Figures 1A to 1C summarize the differences in adjusted mean scores at 12 weeks from various models with differing missing data assumptions.
Similar percentages of patients in each treatment group showed large improvements in the primary and highlighted end points at 12 weeks (P > .1; Table 5). Primary and highlighted outcome measures at 6 to 8, 18, and 24 weeks showed no statistical differences between groups (1% significance level), apart from pain at weeks 6 to 8, which was borderline significantly better for the chemotherapy group (95% CI for the difference in scores, –19.4 to –3.5; P = .01).
Secondary end points. Means, adjusted for baseline, for all other QoL domains/symptoms at all time points were similar between the groups except for hair loss (as expected), which was worse in chemotherapy group up to 24 weeks; peripheral neuropathy, which was worse in the chemotherapy group at all times; and dysphagia, which was worse for the no chemotherapy group at 6 to 8 weeks.
Daily Diary Cards
Plots of daily diary cards were remarkably similar for the two groups except for activity and breathlessness. Patients in the no chemotherapy group consistently reported more problems with breathlessness throughout the study than those in the chemotherapy group (Fig 2). At the start of their chemotherapy cycles, patients were less active in the chemotherapy group than in the no chemotherapy group (Fig 3). There were few nausea problems in the chemotherapy group (Fig 4A). Patterns reflecting chemotherapy cycles only emerge if the threshold is changed to patients reporting nausea of grade 2 or higher (Fig 4B).
Survival and QoL
At 12 weeks, 185 of 239 patients who completed a baseline questionnaire were still alive. Univariate analysis identified 15 scales and symptoms as important pretreatment prognostic factors for survival at 12 weeks (global QoL, physical, role, emotional, cognitive and social functioning, fatigue, nausea/vomiting, pain, dyspnea, appetite loss, constipation, dysphagia, hoarseness, and pain in chest [all P < .1]). In the multivariate model, only global QoL (hazard ratio [HR] = 0.98; P = .009), role functioning (HR = 0.99; P = .026), fatigue (HR = 0.98; P = .013), appetite loss (HR = 1.01; P = .023), and constipation (HR = 1.02; P = .0003) remained significant prognostic factors for survival. Survival was longer for those patients with better global QoL and role functioning scores, for those with less appetite loss, and for those with less constipation. However, survival was longer for those patients with more initial fatigue.
DISCUSSION
The BLT-QoL substudy indicates that there is the possibility of a large positive impact of chemotherapy on global QoL at 12 weeks. The possibility of a large detrimental effect is only apparent if we assume MNAR. The study indicates that there are no large detrimental effects of chemotherapy on physical functioning, fatigue, dyspnea, and pain at 12 weeks. There is evidence bordering on statistical significance that chemotherapy reduces pain over all time points; in particular, at 6 to 8 weeks. However results for emotional functioning are inconclusive and do not rule out the potential for large effects in either direction at 12 weeks.
Current evidence on the QoL of NSCLC patients receiving cisplatin-based chemotherapy is limited. Of eight trials included in the NSCLC meta-analysis, only two measured QoL, and these were not reported because of problems with compliance and data collection. Subsequent studies have suggested that patients receiving chemotherapy report a better QoL than those not receiving chemotherapy,16-18 but these studies used small, unbalanced samples of patients, nonstandard questionnaires, unadjusted multiple tests, or inappropriate analysis (not taking account of repeated measures, baseline variables, or attrition).
In a trial of 351 patients, Cullen et al16 reported on 84 patients (32 no chemotherapy, 52 chemotherapy) with data at 6 weeks using a questionnaire based on the EORTC QLQ-LC13 questionnaire.19 A borderline significant difference in total score, in favor of chemotherapy, after adjustment for patient characteristics was found (P = .06). Details about individual toxicities and symptoms have not yet been published.
Thongprasert et al17 treated 172 patients with one of two platinum-containing regimens or no chemotherapy. QoL was recorded at 2 and 6 months for patients receiving no chemotherapy, and at the start of the second course of chemotherapy and 2 months after chemotherapy for patients receiving chemotherapy. They used a modified Functional Living Index–Cancer20 and a modified QoL index,21 and found that QoL scores improved for the chemotherapy arm but not in the supportive care arm. However, interpretation of the unvalidated modified scales is difficult.
Helsing et al18 used the EORTC QLQ-C30, LC13, and a daily diary card. Only 25 patients were assessable at 12 weeks. They constructed nonstandard summary indices of physical and psychosocial functioning. They found a less favorable outcome at 20 weeks in patients who did not receive chemotherapy in the physical (P = .025) and social domains (P = .0022) and in global QoL (P = .02). These effects were not evident at 12 weeks. No attempt was made to adjust analyses for multiple testing or patient characteristics. No a priori hypothesis was identified. The results from the diary cards were not presented.
One reason why the results of the BLT QoL study may appear contradictory to the previous studies is that for the main supportive care arm of the BLT, we did not restrict inclusion to patients with advanced disease; the only restriction was that the proposed treatment was supportive care. There are therefore eight patients with tumors of stage I or II, and 58 patients with tumors of stage IIIa disease. Thongprasert et al17 and Helsing18 included stage IIIb and IV patients only; Cullen16 did not specify staging in their publication but included patients with unresectable NSCLC who were unsuitable for radical radiotherapy. The differences in QoL between the no chemotherapy and chemotherapy groups within our study may well be expected to be smaller.
There were few problems with nausea and vomiting. This finding may relate to the relatively mild nontoxic chemotherapies used, effectiveness of antiemetics, and because radiotherapy was given to many patients in the no chemotherapy group. More importantly, there was less pain in the chemotherapy group over time, and significantly less pain in the chemotherapy group at 6 to 8 weeks. The significantly higher usage of thoracic radiotherapy in the no chemotherapy group probably explains the higher rate of dysphagia at the 6- to 8-week time point in the no chemotherapy arm, and may explain why there was less difference in QoL and symptoms overall between the two groups.
The assessment of QoL data in the palliative care setting is difficult. Until recently there has been no consensus on instruments; good compliance rates are difficult to achieve, and analysis and interpretation of nonrandom missing data are complex processes. The EORTC QLQ-C30 questionnaire was validated recently and recommended for use in the palliative care setting.22 Compliance of questionnaires at 12 weeks was an improvement on previous studies, although it was still low.17,18 However, missing data was still a problem in this study. The analysis approach attempted to assess the impact of missing data on the results.13
Montazeri et al14 found that physical and role functioning and global QoL scores at diagnosis were independent prognostic factors for survival at 3 months, and that global QoL was also significant in a multivariate analysis after accounting for age, sex, extent of disease, performance status, and weight loss. We have confirmed that global QoL and role functioning are important prognostic indicators of survival. In addition, we have highlighted appetite loss and constipation as important adverse factors for survival; these factors should be studied further. Fatigue was identified but the result would seem counterintuitive: patients who suffered more fatigue were more likely to survive. This result highlights the danger of prognostic analysis on relatively small numbers of patients, and the need for large confirmatory studies.23
Our study represents the most detailed investigation of QoL in patients with NSCLC. The study used standard questionnaires, a priori hypotheses, adequate predefined sample size, and a variety of analyses to verify the sensitivity of conclusions. The results from this study and previous studies show that overall, there are no large detrimental effects of using chemotherapy in patients with poor prognosis NSCLC.
There is now incontrovertible evidence that chemotherapy prolongs survival in patients with advanced NSCLC; indeed, chemotherapy is recommended in the National Guidelines in the United Kingdom. There has been reluctance to treat patients with advanced NSCLC. In the United States, for example, 78% of those older than age 65 did not receive chemotherapy.24 Presumed adverse effects on QoL often have been presented as an argument for not offering chemotherapy. This study does not indicate any important adverse effect of chemotherapy on QoL, and there is no longer a basis for failure to offer chemotherapy to patients in the supportive care setting.
Acknowledgment
We thank P. Shevlin, M. Stead, K. Poulter, Pierre Fabre Oncology, and National Health Service Cancer Research and Development program.
NOTES
Presented in part at the BCRM Clinical Trials Showcase Meeting, June 30, 2002, Glasgow; 38th Annual Meeting of the American Society of Clinical Oncology May 18-21, 2002, Orlando, FL; and BLT Participants' Meeting, April 24, 2002, Leeds, United Kingdom.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Non-Small Cell Lung Cancer Collaborative Group: Chemotherapy in non-small cell lung cancer: A meta analysis using updated data on individual patients from 52 randomised clinical trials. BMJ 311:899-909, 1995
Hopwood P, Stephens RJ: Symptoms at presentation for treatment in patients with lung cancer: Implications for the evaluation of palliative treatment. Br J Cancer 71:633-636, 1995
Spero S, Rudd RM, Souhami RL, et al: Chemotherapy versus supportive care in advanced non–small cell lung cancer: Improved survival without detriment to quality of life. Thorax 59:828-836, 2004
Aaronson NK, Ahmedzai S, Bergman B, et al: The European Organisation for Research and Treatment of Cancer QLQ-C30: A quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 85:365-376, 1993
Fayers PM, Bleehen NM, Girling DJ, et al: Assessment of quality of life in small-cell lung cancer using a daily diary card developed by the Medical Research Council Lung Cancer Working Party. Br J Cancer 64:299-306, 1991
King MT: The interpretation of scores from the EORTC quality of life questionnaire QLQ-C30. Qual Life Res 5:555-567, 1996
Osoba D, Rodrigues G, Myles J, et al: Interpreting the significance of changes in health-related quality-of-life scores. J Clin Oncol 16:139-144, 1998
Fairclough DL, Peterson HF, Cella D, et al: Comparison of several model-based methods for analysing incomplete quality of life data in cancer clinical trials. Stat Med 17:781-796, 1998
Fairclough DL: Design and Analysis of Quality of Life Studies in Clinical Trials. New York, Chapman & Hall, 2002, pp 153-167
Moore MJ, Osoba D, Murphy K, et al: Use of palliative endpoints to evaluate the effects of mitoxantrone and low-dose prednisone in patients with hormonally resistant prostate cancer. J Clin Oncol 12:689-694, 1994
Tannock IF, Osoba D, Stockler MR, et al: Chemotherapy with mitoxantrone plus prednisone or prednisone alone for symptomatic hormone-resistant prostate cancer: A Canadian randomized trial with palliative end points. J Clin Oncol 14:1756-1764, 1996
Osoba D, Tannock IF, Ernst DS, et al: Health-related quality of life in men with metastatic prostate cancer treated with prednisone alone or mitoxantrone and prednisone. J Clin Oncol 17:1654-1663, 1999
Curran D, Molenberghs G, Fayers PM, et al: Incomplete quality of life data in randomized trials: Missing forms. Stat Med 17:697-709, 1998
Montazeri A, Milroy R, Hole D, et al: Quality of life in lung cancer patients as an important prognostic factor. Lung Cancer 31:233-240, 2001
SAS Institute: SAS/STAT User's Guide, Version 8. Cary, NC, SAS Institute, 1999
Cullen MH, Billingham LJ, Woodroffe EM, et al: Mitomycin, ifosfamide, and cisplatin in unresectable non–small cell lung cancer: Effects on survival and quality of life. J Clin Oncol 17:3188-3194, 1999
Thongprasert S, Sanguanmitra P, Juthapan W, et al: Relationship between quality of life and clinical outcomes in advanced non-small cell lung cancer: Best supportive care (BSC) versus BSC plus chemotherapy. Lung Cancer 24:17-24, 1999
Helsing M, Bergman B, Thaning L, et al: Quality of life and survival in patients with advanced non-small cell lung cancer receiving supportive care plus chemotherapy with carboplatin and etoposide or supportive care only: A multicentre randomised phase III trial. Eur J Cancer 34:1036-1044, 1998
Bergman B, Aaronson NK, Ahmedzai S, et al: The EORTC QLQ-LC13: A modular supplement to the EORTC core quality of life questionnaire (QLQ-C30) for use in lung cancer clinical trials. Eur J Cancer 30a:635-642, 1994
Schipper H, Levitt M: Measuring quality of life: Risks and benefits. Cancer Treat Rep 69:1115-1123, 1985
Spitzer WO, Dobson JA, Hall J, et al: Measuring the quality of life of cancer patients: A concise QL-index for use by physicians. J Chronic Dis 34:585-597, 1981
Str?mgren AS, Groenvold M, Pedersen L, et al: Symptomatology of cancer patients in palliative care: Content validation of self-assessment questionnaires against medical records. Eur J Cancer 38:788-794, 2002
Altman DG, Lyman GH: Methodological challenges in the evaluation of prognostic factors in breast cancer. Breast Cancer Res Treat 52:289-303, 1998
Earle CC, Vendetti LN, Neumann PJ, et al: Who gets chemotherapy for metastatic lung cancer? Chest 117:1239-1246, 2000(J. Brown, H. Thorpe, V. N)
New Cross Hospital, Wolverhampton
Cancer Research United Kingdom and University College London Cancer Trials Centre
Medical Research Council Clinical Trials Unit
St Bartholomew's Hospital
The Middlesex Hospital, London
Stobhill National Health Service Trust, Glasgow
Glenfield Hospital, Leicester
York Health Economics Consortium, University of York, United Kingdom
ABSTRACT
PURPOSE: The Big Lung Trial (BLT) was a large, pragmatic trial to evaluate the addition of chemotherapy to primary treatment (ie, surgery, radical radiotherapy, or supportive care) in non–small-cell lung cancer (NSCLC). In the supportive care group, there was a small but significant survival benefit in patients treated with chemotherapy compared with supportive care alone (no chemotherapy). A substudy was undertaken to evaluate the quality of life (QoL) implications of the treatment options. QoL was assessed using European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaires C30 (QLQ-C30) and LC17, and daily diary cards.
PATIENTS AND METHODS: EORTC QLQ-C30 and LC17 were collected at 0, 6 to 8, 12, 18, and 24 weeks. Diary cards were completed during the first 12 weeks of the study. The primary end point was global QoL at 12 weeks.
RESULTS: A total of 273 patients were randomly assigned: 138 to no chemotherapy and 135 to chemotherapy. There was no evidence of a large detrimental effect on QoL of chemotherapy. No statistically significant differences in global QoL or physical/emotional functioning, fatigue and dyspnea, and pain were detected at 12 weeks. Higher rates of palliative radiotherapy in the no chemotherapy arm may have lessened differences in QoL. Global QoL, role functioning, fatigue, appetite loss, and constipation were prognostic indicators of survival at 12 weeks.
CONCLUSION: There were no important adverse effects of chemotherapy on QoL.
INTRODUCTION
Advanced non–small-cell lung cancer (NSCLC) is associated with a poor prognosis. The median survival of untreated patients with stage IV disease is approximately 4 months.1 It is also associated with multiple symptoms,2 and the survival advantage offered by interventions such as chemotherapy is modest.1 In this context, the impact of treatment on quality of life (QoL) and symptoms becomes of major importance for the individual patient and the development of management policies. The NSCLC Collaborative Group meta-analysis revealed that little attention was given to these issues in the large number of chemotherapy-based trials included in the analysis.1
The Big Lung Trial (BLT) was established with the main aim of confirming the survival advantage seen in the NSCLC Collaborative Group meta-analysis when chemotherapy was added to surgery, radical radiotherapy, or supportive care. At the start of the BLT, chemotherapy was being administered to only a minority of patients with advanced NSCLC, and therefore there was an opportunity to randomly assign patients to best supportive care with chemotherapy (chemotherapy group) or without chemotherapy (no chemotherapy group), provided both the physician and patient had doubts about the efficacy of chemotherapy. Thus, the results of the meta-analysis could be re-examined, but with attention focused on the QoL of participants. In the supportive care group, chemotherapy-related improvement for median survival of approximately 8 weeks and an increase of survival at 1 year from 20% to 28% was confirmed, compared with the no chemotherapy arm.3 A total of 725 patients were recruited to the supportive care setting, and this report relates to the subgroup of these patients in whom QoL was measured. European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire C30 (QLQ-C30)4 and the lung cancer–specific module LC17, together with a daily diary card, were used to provide information on the changes in QoL during treatment.
PATIENTS AND METHODS
Eligibility
Patients were eligible for the main trial if they fulfilled local criteria for diagnosis of NSCLC, were to be treated with supportive care, were fit enough to receive chemotherapy, and had no clear contraindications for chemotherapy (eg, inadequate renal function). Patients were ineligible if they had concurrent malignancy or a history of malignancy other than nonmelanomatous skin cancer within the last 3 years. In addition, patients participating in the QoL substudy had to be able to start chemotherapy within 1 week of random assignment; have consented to the QoL assessments; be able to read/complete the assessments; have a life expectancy of at least 8 weeks; and have completed baseline questionnaires before random assignment.
Treatment and Collection of QoL Data
Patients were randomly assigned to receive supportive care with or without chemotherapy. Choice of chemotherapy was made before random assignment. Clinicians could choose any one of four regimens: cisplatin 80 mg/m2 and vindesine 3 mg/m2 days 1 and 8; mitomycin 6 mg/m2, ifosfamide 3 g/m2, and cisplatin 50 mg/m2; mitomycin 6 mg/m2, vinblastine 6 mg/m2, and cisplatin 50 mg/m2; or vinorelbine 30 mg/m2 days 1 and 8 and cisplatin 80 mg/m2. Each regimen was administered for up to three cycles, once every 3 weeks.
The self-administered EORTC QLQ-C304 and the lung cancer supplementary questionnaire (LC17) were completed in the clinic at baseline (before random assignment) and at 6 to 8, 12, 18, and 24 weeks after random assignment. The QLQ-LC17 was under development during the trial and has since been validated (but not yet published) and reduced to 14 items. All 17 items were used in this trial (see Appendix 1). Both questionnaires asked patients to rate their QoL during the last week.
A diary card pack, containing three 4-week diary cards (to be completed daily and to cover the period of chemotherapy for those patients allocated to chemotherapy) was given to all patients at random assignment. The diaries were based on the Medical Research Council daily diary card5 (Appendix 2).
Trial Design
The BLT-QoL study was a multicenter, prospective, randomized study. The Cancer Research UK, University College London Cancer Trials Centre (London, United Kingdom), and the Medical Research Council Clinical Trials Unit carried out random assignment by telephone. Random assignments were minimized by center, WHO performance status, sex, and histology.
A survey of participating clinicians highlighted those elements of QoL that the clinicians believed were most important to patients. On the basis of this survey, the primary end point for the BLT-QoL study was defined as the difference in global QoL at 12 weeks. Clinicians indicated that only large differences in global QoL would be of clinical interest. Sample size was therefore based on a large difference between the two groups in global QoL at 12 weeks. Using methods described by King6 and Osoba et al,7 this translated into the detection of an effect size of approximately 0.4 (mean difference between groups divided by common standard deviation). To detect a standardized effect size of 0.4 (two-sided 5% significance level, 80% power), 200 patients were required. An effect size of 0.5 would require 128 patients. The sample size was set at 300 to allow for noncompliance. In practice, the main trial was closed on the planned closure date, at which time 273 patients had entered the BLT-QoL study. However, compliance at 12 weeks was 71%; therefore, the required sample size was achieved. Clinicians also highlighted emotional and physical functioning, fatigue, dyspnea, and pain at 12 weeks for additional study. All other domains and symptoms were classified as secondary end points.
Statistics
The EORTC QLQ-C30 was scored according to EORTC guidelines. Higher scores indicated better functioning, whereas higher scores for symptom scales indicated increased severity. LC17 was scored according to the QLQ-LC14 manual (three multi-item scales and seven single-symptom scores); the three additional questions were summarized separately.
Primary and highlighted end points were compared using multilevel repeated-measures modeling accounting for data at all time points (assuming missing data at random [MAR] and allowing for time, treatment, treatment-time interaction, and adjusting for baseline QoL [all fixed effects], patient, and patient-time [random effects]). Minimization factors, age, and marital status were added as covariates in the above. Given that inclusion of covariates made little difference to interpretation, differences between treatment estimates are presented from the multilevel model adjusted for baseline QoL only. Pattern mixture multilevel model was also fitted by categorizing patients into strata based on clinical information, stage (< 4, 4), age (< 65, 65), and WHO score at 12 weeks (< 2, 2; assuming data are missing not at random [MNAR]).8 Pattern mixture models for bivariate (baseline and 12 week) data were also fitted using a variety of restrictions reflecting the missing data pattern ranging from complete case missing variable restriction (MAR) to Brown's protective restriction (MNAR).9 To ease clinical interpretation, treatment estimates from a logistic regression model investigating the proportion of patients improving for each of the primary and highlighted end points at 12 weeks, adjusted by baseline score, are also presented (assuming MAR).10-12 As based with King,6 improvement in individual patients and a large difference between groups was defined as +10 points for global QoL, +25 points for physical functioning, +7 points for emotional functioning, and +20 points for dyspnea, fatigue, and pain.
The analysis approach used summarized data in different ways to encompass as much information as possible, and allowed for differing assumptions about the missing data (ignorable and not ignorable); therefore, some testing of the sensitivity of the analysis to the missing data was undertaken.13
Means and 95% CIs, adjusted for baseline, were calculated for all other domains of the QoL questionnaires. No statistical testing was carried out on these data. The primary end point analyses were carried out at a 5% significance level; all other analyses were carried out at 1% significance level.
An additional analysis was performed to validate results from Montazeri et al,14 which showed that pretreatment global QoL was a prognostic factor for survival at 3 months, and to assess further other baseline QoL factors in a hypothesis-generating exercise. Univariate relationships were investigated using a log-rank test. QoL scales/symptoms identified at the 10% level were then considered in a multivariate model. A forward and backward stepwise Cox proportional hazards model was performed and 5% significance was used as the criteria for inclusion. Known prognostic factors in NSCLC (age, sex, performance status, histology, and stage of disease) and treatment group were also included in the multivariate model to assess the additional impact of the QoL. Survival time was calculated from date of the baseline questionnaire to either date of death (all causes) for those who died or 12 weeks for those who lived beyond 12 weeks. For those not observed to 12 weeks, survival time was the date last seen alive. Proportional hazards were evaluated by analysis of the hazard function plots.
Daily diary cards were descriptively compared between groups by examining plots of the percentage of patients reporting moderate/severe symptoms each day from random assignment. Moderate or severe symptoms were defined as grade 3 or 4 nausea, vomiting, tiredness, and breathlessness; grade 4 or 5 mood, overall condition, appetite, and activity; and grade 3, 4, or 5 difficulty in swallowing.
Lines to indicate approximate timing of chemotherapy were added to aid interpretation. Missing data were not taken into account in these summaries, and hence results should be interpreted with caution.
Compliance with questionnaires was calculated as the number of completed questionnaires divided by the total number of questionnaires expected according to the protocol (excluding deaths). Calculations were based on a period of 2 weeks around the expected date of completion (1 week for the 6- to 8-week questionnaire). All statistical analyses were performed with SAS software, version 8.2 (SAS Institute, Cary NC).15
Ethics
The Northern and Yorkshire Multicenter Research Ethics Committee and Local Research Ethics Committees approved the protocol. An independent Data Monitoring and Ethics Committee advised on the conduct of the study.
RESULTS
Patient Characteristics
The BLT main trial started recruiting in November 1995; the QoL study started in March 1998 (once funding was obtained). Between March 1998 and November 2001, 273 patients from 33 (32 in the United Kingdom, one from Australia) of 62 centers in the supportive care part of the main trial, and 46 clinicians entered the QoL substudy. One hundred thirty-eight patients were randomly assigned to the no chemotherapy group and 135 patients were randomly assigned to the chemotherapy group, representing 38% of all patients in the supportive care part of the main BLT, given that center participation was not compulsory (58% during the QoL recruiting period). Treatment groups were well balanced with respect to patient baseline characteristics (Table 1). The baseline characteristics were also similar for those patients in the supportive care part of the main trial but not in the QoL study.
Baseline QoL score across all scales, and for all but three symptoms (constipation, diarrhea, and financial difficulties) from EORTC QLQ-C30, were better for the chemotherapy group compared with the no chemotherapy group (Table 2). No notable differences were apparent for LC17 (Table 2).
Treatment Details
Of the patients receiving chemotherapy, 30 (22%) received mitomycin, ifosfamide, and cisplatin; 63 (47%) received mitomycin, vinblastine, ifosfamide, and cisplatin; 42 (31%) received vinorelbine and cisplatin; and none received the cisplatin and vindesine regimen. Sixty-four percent of patients received three cycles, 12% received two cycles, 18% received one cycle, and 7% received none. Grade 3/4 toxicity was suffered by 30% of patients; these patients mainly experienced nausea/vomiting, neutropenic fever, hematologic, neurologic, and renal toxicities. Similar numbers of patients in both arms received nonthoracic radiotherapy (1%); however, 50% of the no chemotherapy group compared with 21% of the chemotherapy group received thoracic radiotherapy during the first 24-week period (47% v 6% during the first 8 weeks from random assignment).
Compliance
Compliance with EORTC questionnaires was high at baseline but decreased at follow-up, although 60% of patients completed questionnaires at week 24 (Table 3). There were few missing items on completed questionnaires. Compliance at 12 weeks was better; because as this time point was of primary interest patient response was more intensively pursued. Questionnaires at both baseline and 12 weeks were completed by 136 patients (50%). Reasons for noncompletion of questionnaires are listed in Table 4 and were balanced across groups.
QoL Results
Primary and highlighted outcomes. In keeping with the main trial results, the 1-year survival rate for the QoL patients increased in the chemotherapy group to 29% from 21% in the no chemotherapy group. There were no statistically significant differences between treatment groups for the QoL primary or highlighted end points. The various multilevel and pattern mixture analyses had CIs that included the possibility of a large positive impact (clinically important) on global QoL in favor of chemotherapy, and the possibility of a large detrimental effect was only apparent using Brown's restriction. At 12 weeks large positive or negative effects of chemotherapy on fatigue and dyspnea were ruled out in all analyses, but results for emotional functioning did not rule out the possibility of large effects in either direction. For physical functioning, some of the analyses assuming MNAR indicated a large positive effect for chemotherapy, but all ruled out large negative effects. For pain, large negative effects were ruled out and the possibility of large positive effects was apparent regardless of assumptions made about missing data. Figures 1A to 1C summarize the differences in adjusted mean scores at 12 weeks from various models with differing missing data assumptions.
Similar percentages of patients in each treatment group showed large improvements in the primary and highlighted end points at 12 weeks (P > .1; Table 5). Primary and highlighted outcome measures at 6 to 8, 18, and 24 weeks showed no statistical differences between groups (1% significance level), apart from pain at weeks 6 to 8, which was borderline significantly better for the chemotherapy group (95% CI for the difference in scores, –19.4 to –3.5; P = .01).
Secondary end points. Means, adjusted for baseline, for all other QoL domains/symptoms at all time points were similar between the groups except for hair loss (as expected), which was worse in chemotherapy group up to 24 weeks; peripheral neuropathy, which was worse in the chemotherapy group at all times; and dysphagia, which was worse for the no chemotherapy group at 6 to 8 weeks.
Daily Diary Cards
Plots of daily diary cards were remarkably similar for the two groups except for activity and breathlessness. Patients in the no chemotherapy group consistently reported more problems with breathlessness throughout the study than those in the chemotherapy group (Fig 2). At the start of their chemotherapy cycles, patients were less active in the chemotherapy group than in the no chemotherapy group (Fig 3). There were few nausea problems in the chemotherapy group (Fig 4A). Patterns reflecting chemotherapy cycles only emerge if the threshold is changed to patients reporting nausea of grade 2 or higher (Fig 4B).
Survival and QoL
At 12 weeks, 185 of 239 patients who completed a baseline questionnaire were still alive. Univariate analysis identified 15 scales and symptoms as important pretreatment prognostic factors for survival at 12 weeks (global QoL, physical, role, emotional, cognitive and social functioning, fatigue, nausea/vomiting, pain, dyspnea, appetite loss, constipation, dysphagia, hoarseness, and pain in chest [all P < .1]). In the multivariate model, only global QoL (hazard ratio [HR] = 0.98; P = .009), role functioning (HR = 0.99; P = .026), fatigue (HR = 0.98; P = .013), appetite loss (HR = 1.01; P = .023), and constipation (HR = 1.02; P = .0003) remained significant prognostic factors for survival. Survival was longer for those patients with better global QoL and role functioning scores, for those with less appetite loss, and for those with less constipation. However, survival was longer for those patients with more initial fatigue.
DISCUSSION
The BLT-QoL substudy indicates that there is the possibility of a large positive impact of chemotherapy on global QoL at 12 weeks. The possibility of a large detrimental effect is only apparent if we assume MNAR. The study indicates that there are no large detrimental effects of chemotherapy on physical functioning, fatigue, dyspnea, and pain at 12 weeks. There is evidence bordering on statistical significance that chemotherapy reduces pain over all time points; in particular, at 6 to 8 weeks. However results for emotional functioning are inconclusive and do not rule out the potential for large effects in either direction at 12 weeks.
Current evidence on the QoL of NSCLC patients receiving cisplatin-based chemotherapy is limited. Of eight trials included in the NSCLC meta-analysis, only two measured QoL, and these were not reported because of problems with compliance and data collection. Subsequent studies have suggested that patients receiving chemotherapy report a better QoL than those not receiving chemotherapy,16-18 but these studies used small, unbalanced samples of patients, nonstandard questionnaires, unadjusted multiple tests, or inappropriate analysis (not taking account of repeated measures, baseline variables, or attrition).
In a trial of 351 patients, Cullen et al16 reported on 84 patients (32 no chemotherapy, 52 chemotherapy) with data at 6 weeks using a questionnaire based on the EORTC QLQ-LC13 questionnaire.19 A borderline significant difference in total score, in favor of chemotherapy, after adjustment for patient characteristics was found (P = .06). Details about individual toxicities and symptoms have not yet been published.
Thongprasert et al17 treated 172 patients with one of two platinum-containing regimens or no chemotherapy. QoL was recorded at 2 and 6 months for patients receiving no chemotherapy, and at the start of the second course of chemotherapy and 2 months after chemotherapy for patients receiving chemotherapy. They used a modified Functional Living Index–Cancer20 and a modified QoL index,21 and found that QoL scores improved for the chemotherapy arm but not in the supportive care arm. However, interpretation of the unvalidated modified scales is difficult.
Helsing et al18 used the EORTC QLQ-C30, LC13, and a daily diary card. Only 25 patients were assessable at 12 weeks. They constructed nonstandard summary indices of physical and psychosocial functioning. They found a less favorable outcome at 20 weeks in patients who did not receive chemotherapy in the physical (P = .025) and social domains (P = .0022) and in global QoL (P = .02). These effects were not evident at 12 weeks. No attempt was made to adjust analyses for multiple testing or patient characteristics. No a priori hypothesis was identified. The results from the diary cards were not presented.
One reason why the results of the BLT QoL study may appear contradictory to the previous studies is that for the main supportive care arm of the BLT, we did not restrict inclusion to patients with advanced disease; the only restriction was that the proposed treatment was supportive care. There are therefore eight patients with tumors of stage I or II, and 58 patients with tumors of stage IIIa disease. Thongprasert et al17 and Helsing18 included stage IIIb and IV patients only; Cullen16 did not specify staging in their publication but included patients with unresectable NSCLC who were unsuitable for radical radiotherapy. The differences in QoL between the no chemotherapy and chemotherapy groups within our study may well be expected to be smaller.
There were few problems with nausea and vomiting. This finding may relate to the relatively mild nontoxic chemotherapies used, effectiveness of antiemetics, and because radiotherapy was given to many patients in the no chemotherapy group. More importantly, there was less pain in the chemotherapy group over time, and significantly less pain in the chemotherapy group at 6 to 8 weeks. The significantly higher usage of thoracic radiotherapy in the no chemotherapy group probably explains the higher rate of dysphagia at the 6- to 8-week time point in the no chemotherapy arm, and may explain why there was less difference in QoL and symptoms overall between the two groups.
The assessment of QoL data in the palliative care setting is difficult. Until recently there has been no consensus on instruments; good compliance rates are difficult to achieve, and analysis and interpretation of nonrandom missing data are complex processes. The EORTC QLQ-C30 questionnaire was validated recently and recommended for use in the palliative care setting.22 Compliance of questionnaires at 12 weeks was an improvement on previous studies, although it was still low.17,18 However, missing data was still a problem in this study. The analysis approach attempted to assess the impact of missing data on the results.13
Montazeri et al14 found that physical and role functioning and global QoL scores at diagnosis were independent prognostic factors for survival at 3 months, and that global QoL was also significant in a multivariate analysis after accounting for age, sex, extent of disease, performance status, and weight loss. We have confirmed that global QoL and role functioning are important prognostic indicators of survival. In addition, we have highlighted appetite loss and constipation as important adverse factors for survival; these factors should be studied further. Fatigue was identified but the result would seem counterintuitive: patients who suffered more fatigue were more likely to survive. This result highlights the danger of prognostic analysis on relatively small numbers of patients, and the need for large confirmatory studies.23
Our study represents the most detailed investigation of QoL in patients with NSCLC. The study used standard questionnaires, a priori hypotheses, adequate predefined sample size, and a variety of analyses to verify the sensitivity of conclusions. The results from this study and previous studies show that overall, there are no large detrimental effects of using chemotherapy in patients with poor prognosis NSCLC.
There is now incontrovertible evidence that chemotherapy prolongs survival in patients with advanced NSCLC; indeed, chemotherapy is recommended in the National Guidelines in the United Kingdom. There has been reluctance to treat patients with advanced NSCLC. In the United States, for example, 78% of those older than age 65 did not receive chemotherapy.24 Presumed adverse effects on QoL often have been presented as an argument for not offering chemotherapy. This study does not indicate any important adverse effect of chemotherapy on QoL, and there is no longer a basis for failure to offer chemotherapy to patients in the supportive care setting.
Acknowledgment
We thank P. Shevlin, M. Stead, K. Poulter, Pierre Fabre Oncology, and National Health Service Cancer Research and Development program.
NOTES
Presented in part at the BCRM Clinical Trials Showcase Meeting, June 30, 2002, Glasgow; 38th Annual Meeting of the American Society of Clinical Oncology May 18-21, 2002, Orlando, FL; and BLT Participants' Meeting, April 24, 2002, Leeds, United Kingdom.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Non-Small Cell Lung Cancer Collaborative Group: Chemotherapy in non-small cell lung cancer: A meta analysis using updated data on individual patients from 52 randomised clinical trials. BMJ 311:899-909, 1995
Hopwood P, Stephens RJ: Symptoms at presentation for treatment in patients with lung cancer: Implications for the evaluation of palliative treatment. Br J Cancer 71:633-636, 1995
Spero S, Rudd RM, Souhami RL, et al: Chemotherapy versus supportive care in advanced non–small cell lung cancer: Improved survival without detriment to quality of life. Thorax 59:828-836, 2004
Aaronson NK, Ahmedzai S, Bergman B, et al: The European Organisation for Research and Treatment of Cancer QLQ-C30: A quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 85:365-376, 1993
Fayers PM, Bleehen NM, Girling DJ, et al: Assessment of quality of life in small-cell lung cancer using a daily diary card developed by the Medical Research Council Lung Cancer Working Party. Br J Cancer 64:299-306, 1991
King MT: The interpretation of scores from the EORTC quality of life questionnaire QLQ-C30. Qual Life Res 5:555-567, 1996
Osoba D, Rodrigues G, Myles J, et al: Interpreting the significance of changes in health-related quality-of-life scores. J Clin Oncol 16:139-144, 1998
Fairclough DL, Peterson HF, Cella D, et al: Comparison of several model-based methods for analysing incomplete quality of life data in cancer clinical trials. Stat Med 17:781-796, 1998
Fairclough DL: Design and Analysis of Quality of Life Studies in Clinical Trials. New York, Chapman & Hall, 2002, pp 153-167
Moore MJ, Osoba D, Murphy K, et al: Use of palliative endpoints to evaluate the effects of mitoxantrone and low-dose prednisone in patients with hormonally resistant prostate cancer. J Clin Oncol 12:689-694, 1994
Tannock IF, Osoba D, Stockler MR, et al: Chemotherapy with mitoxantrone plus prednisone or prednisone alone for symptomatic hormone-resistant prostate cancer: A Canadian randomized trial with palliative end points. J Clin Oncol 14:1756-1764, 1996
Osoba D, Tannock IF, Ernst DS, et al: Health-related quality of life in men with metastatic prostate cancer treated with prednisone alone or mitoxantrone and prednisone. J Clin Oncol 17:1654-1663, 1999
Curran D, Molenberghs G, Fayers PM, et al: Incomplete quality of life data in randomized trials: Missing forms. Stat Med 17:697-709, 1998
Montazeri A, Milroy R, Hole D, et al: Quality of life in lung cancer patients as an important prognostic factor. Lung Cancer 31:233-240, 2001
SAS Institute: SAS/STAT User's Guide, Version 8. Cary, NC, SAS Institute, 1999
Cullen MH, Billingham LJ, Woodroffe EM, et al: Mitomycin, ifosfamide, and cisplatin in unresectable non–small cell lung cancer: Effects on survival and quality of life. J Clin Oncol 17:3188-3194, 1999
Thongprasert S, Sanguanmitra P, Juthapan W, et al: Relationship between quality of life and clinical outcomes in advanced non-small cell lung cancer: Best supportive care (BSC) versus BSC plus chemotherapy. Lung Cancer 24:17-24, 1999
Helsing M, Bergman B, Thaning L, et al: Quality of life and survival in patients with advanced non-small cell lung cancer receiving supportive care plus chemotherapy with carboplatin and etoposide or supportive care only: A multicentre randomised phase III trial. Eur J Cancer 34:1036-1044, 1998
Bergman B, Aaronson NK, Ahmedzai S, et al: The EORTC QLQ-LC13: A modular supplement to the EORTC core quality of life questionnaire (QLQ-C30) for use in lung cancer clinical trials. Eur J Cancer 30a:635-642, 1994
Schipper H, Levitt M: Measuring quality of life: Risks and benefits. Cancer Treat Rep 69:1115-1123, 1985
Spitzer WO, Dobson JA, Hall J, et al: Measuring the quality of life of cancer patients: A concise QL-index for use by physicians. J Chronic Dis 34:585-597, 1981
Str?mgren AS, Groenvold M, Pedersen L, et al: Symptomatology of cancer patients in palliative care: Content validation of self-assessment questionnaires against medical records. Eur J Cancer 38:788-794, 2002
Altman DG, Lyman GH: Methodological challenges in the evaluation of prognostic factors in breast cancer. Breast Cancer Res Treat 52:289-303, 1998
Earle CC, Vendetti LN, Neumann PJ, et al: Who gets chemotherapy for metastatic lung cancer? Chest 117:1239-1246, 2000(J. Brown, H. Thorpe, V. N)