Japanese Journal of Clinical Oncology 33:470-476 (2003)
© 2003 Foundation for Promotion of Cancer Research
Influence of Clinical Parameters on Quality of Life During Chemotherapy in Patients with Advanced Non-small Cell Lung Cancer: Application of a General Linear Model
1 Department of Epidemiological and Clinical Research Information Management, Kyoto University Graduate School of Medicine, Kyoto, 2 Department of Pulmonary Medicine, Saitama Cancer Center, Saitama, 3 Division of Pulmonary Medicine, Tokai University School of Medicine, Kanagawa, 4 Department of Respiratory Disease, Musashigaoka Hospital, Kumamoto, 5 Division of Respiratory Medicine, Tokyo Metropolitan Komagome Hospital, Tokyo, 6 Department of Internal Medicine, Mitoyo General Hospital, Kagawa and 7 School of Health Sciences and Nursing, University of Tokyo, Tokyo, Japan
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Objective: The aim of this study was to determine the relative influence of physician-assessed clinical parameters, including non-hematological adverse events and performance status, on quality of life (QOL) during chemotherapy.
Methods: QOL questionnaires consisting of four domains (functional, physical, mental and psychosocial) were self-administered every week during chemotherapy by patients with advanced non-small cell lung cancer in two phase III clinical trials; 377 patients who completed the questionnaires at baseline and at least once during the first course of therapy were analyzed. A general linear model was applied, where the four domains and the clinical parameters (nausea/vomiting, anorexia, diarrhea, fever, peripheral neuropathy and performance status) were used as the response and explanatory variables, respectively. In this model, the multi-dimensional and longitudinal aspects of QOL data were taken into account.
Results: All four domains were significantly affected by the occurrence of nausea/vomiting, anorexia and diarrhea. No influence of peripheral neuropathy on the domains was detected. Performance status was significantly related to the domains (except the psychosocial domain).
Conclusion: This study revealed, by examination of multi-dimensional repeated QOL data, that clinical parameters had significant effects on QOL in patients undergoing chemotherapy. Our findings suggest that supportive care to control non-hematological adverse events, especially gastrointestinal, could maintain overall QOL in cancer patients in an earlier phase of chemotherapy.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
In the late 1980s and early 1990s, active discussion took place about the definition of quality of life (QOL) and the methodology for assessing QOL in cancer patients (1–7). Today, the consensus view holds that QOL is a subjective, multi-dimensional construct of both positive and negative domains (3,4,6,8), which should, at a minimum, include functional, physical, mental and social well-being (9). These domains are thought to be conceptually related to one another at a moderate correlation (1,2). A number of well-validated cancer-specific instruments for QOL assessment have been developed, such as the European Organization for Research and Treatment of Cancer (EORTC) core questionnaire QLQ-C30 (2) and the Functional Assessment of Cancer Therapy Scale – General (FACT-G) (10). In addition, the QOL Questionnaire for Cancer Patients Treated with Anticancer Drugs (QOL-ACD) was developed by the Japanese Quality of Life Research Group as a generic questionnaire based on a multi-dimensional construct (11). Subsequently, QOL has been investigated in a longitudinal fashion in clinical trials of cancer therapy using those questionnaires as an important supplementary endpoint to traditional endpoints such as survival, tumor response and toxicity.
Several studies have investigated the relationship between patient-reported QOL and physician-assessed clinical parameters, such as toxicity and performance status, owing to concern about patients’ tolerance of chemotherapy (12–17). These studies, however, were not necessarily adequate to capture the entire effect of clinical parameters on QOL, because of the following factors: (1) they did not analyze longitudinal data, even though a longitudinal analysis is expected to provide added efficiency (statistical power) in estimating the associations between clinical parameters and QOL relative to a cross-sectional analysis (18); (2) they did not examine multi-domain QOL data despite the multi-dimensional aspect of QOL; or (3) they evaluated the effect of one of the clinical parameters on QOL without adjusting for other parameters, thereby preventing an evaluation of their relative effects.
This study investigated multi-dimensional QOL data weekly during chemotherapy using a general linear model, in order to determine the relative influence of physician-assessed clinical parameters on QOL in patients with advanced non-small cell lung cancer (NSCLC).
|
PATIENTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This was an ancillary study for two clinical trials (trials A and B), which were conducted by, respectively, the West and East CPT-11 Lung Cancer Study Groups in Japan to compare three chemotherapy regimens using common endpoints and treatment/evaluation schedules. Previously untreated patients with stage IIIB/IV NSCLC and an ECOG performance status (PS) of 0–2 were eligible for these trials. The primary endpoint of these trials was overall survival. Toxicity, response rate and time to disease progression were secondary endpoints and QOL was also investigated in the trials. In trial A (July 1995 to January 1998), 398 patients were randomized to receive one of the following regimens: (1) cisplatin (80 mg/m2 on day 1) and irinotecan (60 mg/m2 on days 1, 8 and 15), (2) cisplatin (80 mg/m2 on day 1) and vindesine (3 mg/m2 on days 1, 8 and 15) or (3) irinotecan alone (100 mg/m2 on days 1, 8 and 15), with all schedules repeated every 4 weeks. In trial B (June 1995 to October 1997), 210 patients were randomized to receive either cisplatin and irinotecan or cisplatin plus vindesine by the same schedules as in trial A.
QOL Assessment and Evaluation of Clinical Parameters
In both trials, QOL was measured by the QOL-ACD. This questionnaire has been previously validated for patients with advanced NSCLC (19). The QOL questionnaires were self-administered by patients prior to treatment, every week during treatment (before infusion or just before days 1, 8, 15 and 22 of each course) and for 1 month after the last administration and monthly thereafter. We analyzed the data observed at baseline and in the first course (weeks 1–4) in patients who completed at least two QOL questionnaires (i.e. one before treatment plus one or more during the first course), in order to avoid the possible influence of disease progression on QOL and because we expected that a considerable amount of missing data would occur after the first course.
The QOL-ACD is a 22-item questionnaire covering four domains evaluating functional well-being (items 1–6), physical well-being (items 7–11), mental well-being (items 12–16) and psychosocial well-being (items 17–21). A face scale (item 22) was also used to evaluate global QOL (Fig. 1). The validation study found that two items (item 6 owing to its low test–retest reliability and item 16 owing to its poor convergent validity) should be excluded from QOL evaluation of NSCLC patients, at least in the inpatient setting (19). The score for each domain was obtained by calculating the average of the items (except items 6 and 16) completed by the patient and then applying a linear transformation to obtain an average score ranging from 0 to 100. For all items and domains, higher scores represent better QOL.
|
Non-hematological adverse events (nausea/vomiting, anorexia, diarrhea, fever and peripheral neuropathy) and the ECOG PS were the clinical parameters used for analysis. The World Health Organization (WHO) criteria (20) were used to grade adverse events, except diarrhea, which was assessed according to the ECOG common toxicity criteria (21). The most severe adverse events and the worst PS experienced within each week were computed for each patient.
Statistical Analysis
Mean QOL scores for the four domains were calculated at baseline and in each week during the first course of chemotherapy. Frequencies of patients who experienced adverse events in each week during the first course were also calculated. In addition, QOL scores obtained during the second course of chemotherapy were analyzed to examine whether the same longitudinal profiles of QOL were observed as during the first course. Regarding the influence of clinical parameters on QOL, multivariate longitudinal data from baseline through the first course were analyzed using a general linear model, with a covariance structure designed to account for correlations both among domains and across time (18,22,23). For estimating the relative effects of the clinical parameters on the four domains, changes in the scores for the four domains in the questionnaire from baseline were used as response variables and the severity of non-hematological adverse events and the changes in PS from baseline were used as explanatory variables.
Additional general linear model analyses stratified by gender (male/female), age (<64 or 
64 years) and treatment arm (cisplatin + irinotecan/cisplatin + vindesine/irinotecan alone) were performed to examine whether quantitative or qualitative differences existed in the relationship of the clinical parameters with the four domains in such subgroups.
Although some data were expected to be missed even in the first course, all available QOL data from each of the weeks 1–4 were used in these analyses, under the assumption that data missed were missed at random and therefore ignorable (24,25). In other words, no specific analytical techniques were applied to deal with missing data, such as imputation-based approaches (24,25). The frequencies of adverse events and deterioration of PS in week 4 (last week of the first course), however, were calculated to verify the assumption for the missing data.
These analyses were performed using SAS for Windows release 8.02 (SAS Institute, Cary, NC) and the general linear model analyses were carried out using the SAS Mixed procedure.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Clinical Characteristics
Of 583 eligible patients randomized into the phase III trials, 377 who completed the questionnaire before treatment and at least once during the first course were analyzed. Their baseline clinical characteristics and treatment outcomes are summarized in Table 1. The median age was 61 years (range: 35–75 years) and 78.5% of the patients were men. The patients had stage IV (61.4%) or stage IIIB (38.4%) disease, but a good PS (ECOG PS was 0–1 in 96.9%). These 377 patients had almost the same clinical characteristics at baseline as those patients who were excluded from the analysis. Higher tumor response rate, longer median survival time and better treatment compliance, however, were observed in the 377 analyzed patients compared with the excluded patients.
|
QOL and Clinical Parameters
The mean scores for the four domains calculated using the QOL data observed at baseline and at each week during the first course of chemotherapy are shown in Fig. 2. Marked changes in QOL scores were observed over time, especially for the physical and functional domains. The maximum decrease in score for each domain was observed at week 1 (within a week before day 8 of the first course). The compliance of patients with QOL assessment is also shown in Fig. 2. Overall, compliance decreased from 96.0% in week 1 to 84.6% in week 4. During the second course of chemotherapy, the same profiles were found for all four domains as in the first course. In other words, highly similar mean QOL scores were observed for each of the four domains in the corresponding weeks of the two courses (i.e. weeks 1 and 5, 2 and 6, 3 and 7, 4 and 8). For example, the mean QOL scores for the physical domain in each of weeks 1–8 were 49.9, 62.7, 66.9, 72.1, 52.6, 62.3, 65.9 and 71.2, respectively. However, this finding must be viewed with caution because the compliance during the second course declined considerably from 81.4% in week 5 to 64.5% in week 8.
|
The clinical parameters of the patients during the first course of chemotherapy are summarized in Table 2. More severe nausea/vomiting and anorexia were experienced in week 1, while diarrhea and fever were seen more frequently in week 2. A small number of patients experienced peripheral neuropathy during the first course. More severe deterioration of PS was observed in weeks 1 and 2, while a noticeable number of patients experienced improved PS in week 4. The longitudinal changes in QOL scores could be divided into two phases of deterioration and recovery that appeared to correspond to the occurrence of adverse events and changes in PS. There were no large differences in the clinical parameter profiles between patients who were analyzed and those who were excluded from analysis. For example, of the 58 patients excluded from analysis in week 4, 1.7, 13.8, 1.7, 20.7, 5.2 and 10.4% experienced nausea/vomiting, anorexia, diarrhea, fever, peripheral neuropathy and deterioration of PS, respectively. Grade 2 and 3 (and 4) toxicities observed were nausea/vomiting (grade 2, 0%; grade 3; 1.7%), anorexia (grade 2, 5.2%; grade 3, 1.7%); diarrhea (grade 2, 0%; grades 3 and 4, 1.7%) and fever (grade 2, 3.5%; grade 3, 0%). Grade 2 and 3 peripheral neuropathy did not occur in these 58 patients. Changes in more than one PS score were observed in 5.2% of patients.
|
Multivariate Longitudinal Data Analysis
The general linear model was used to evaluate the influence of clinical parameters on all four domains simultaneously. In the analysis, a negative regression coefficient indicated the occurrence of more severe adverse events related to the deterioration of QOL. Table 3 shows coefficients with P values that were estimated from the general linear model. Nausea/vomiting and anorexia had a significant influence on all four domains, especially the physical one. Diarrhea had a significant influence not only on the physical but also on the mental and psychosocial domains. However, influence of fever and peripheral neuropathy on the domains was not detected. PS was significantly related to the mental as well as functional and physical domains.
|
Based on the results of subgroup analyses, neither qualitative nor noticeable quantitative differences were observed. In other words, similar relationships between clinical parameters and QOL domains were also observed in the gender, age and treatment arm subgroups.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
In earlier studies, cross-sectional or one-dimensional QOL data were analyzed to investigate influences of chemotherapy-induced adverse events on QOL in different types of cancer (12–17). However, in order to capture the entire effect of chemotherapy-induced adverse events on QOL, it is important to assess QOL by using a multi-dimensional instrument (12). Moreover, repeated QOL assessments over time during chemotherapy may be vital, especially when anti-cancer drugs are administered on a weekly schedule. Thus, QOL data should be both longitudinal and multi-dimensional. In this study, we analyzed weekly-collected multi-dimensional QOL data during chemotherapy to determine the relative influence of physician-assessed clinical parameters on QOL in patients with advanced NSCLC. A general linear model, in which the inter-domain and time-series correlations were handled, was employed for the analysis.
Our study demonstrated that nausea/vomiting, anorexia and diarrhea had a greater influence on the physical domain of the questionnaire than other domains. Interestingly, these clinical parameters had a considerable influence over the mental and psychosocial as well as physical and functional domains. Moreover, the relationship between the severity of adverse events and deterioration in the four domains appeared to hold when subgroups were analyzed by gender, age and treatment arm. Therefore, these data suggest that supportive treatment for non-hematological adverse events, especially gastrointestinal, could provide improvement/maintenance of not only physical and functional well-being, but also overall QOL, including mental and psychosocial well-being during chemotherapy in this study population.
Our analysis did not detect a significant influence of fever and peripheral neuropathy on any of the four domains. This may be due to lower frequencies of these adverse events compared with others. Because peripheral neuropathy, in particular, is a chronic toxicity of chemotherapy, its influence on QOL may only be captured if relevant data are collected with good compliance in the second course and thereafter in future studies.
It is necessary to discuss the limitations of this study with regard to missing QOL data commonly observed in the context of advanced stage disease (26,27). First, 206 patients who did not complete the QOL questionnaires were excluded from the analysis. Whereas the baseline clinical characteristics of patients analyzed were very similar to those of the excluded patients, treatment outcomes, such as tumor response, survival time and treatment compliance were rather different, in favor of the subset of analyzed patients. Although we used the data gathered prior to the second course of chemotherapy to avoid possible impacts of disease progression or early termination of treatment on the associations between clinical parameters and QOL, we admit that the results from our analysis may not easily be generalized. Second, a noticeable number of patients (4, 9, 12 and 15% in weeks 1, 2, 3 and 4, respectively) did not fill out the QOL questionnaire, even during the first course of chemotherapy. Similar frequencies of adverse events and a similar deterioration in PS, however, were observed between the patients who filled out the questionnaires and those who did not. This result could support the validity of our findings, because the missing QOL data observed during the first course was probably unrelated to the clinical deterioration and therefore its impacts on our analysis may be ignored. Finally, we did not analyze QOL data obtained in the second course of treatment and thereafter because of the high percentage of missing assessments (for example, 35% at the end of the second course). Taking into account that the current standard chemotherapy has a tendency to be repeated for more than two courses, the influence of chemotherapy on QOL in the second course or later phases of treatment should be examined in future studies where the best efforts (better staff training and more support of clinical research coordinators) are made to improve data collection.
In conclusion, we analyzed multi-dimensional repeated QOL data and revealed significant influences of nausea/vomiting, anorexia and diarrhea on mental and psychosocial, as well as physical and functional, domains of patients with advanced NSCLC undergoing chemotherapy. These findings suggest that supportive care to control these non-hematological adverse events could help maintain overall QOL during chemotherapy in cancer patients.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
We thank the West and East CPT-11 Lung Cancer Study Groups (Dr Hisanobu Niitani, Dr Masahiro Fukuoka, Dr Keiichi Nagao and other investigators) for providing us with valuable data. We also thank Akira Sahashi, Fumiyasu Fukuda and Tomoji Miyazaki for skillful data management during this study. This paper could not have been written without their helpful contributions.
|
FOOTNOTES |
|---|
+ For reprints and all correspondence: Satoshi Morita, Department of Epidemiological and Clinical Research Information Management, Kyoto University Graduate School of Medicine, Yoshidakonoe-cho, Sakyo-ku, Kyoto 606-8501, Japan. E-mail: smorita{at}pbh.med.kyoto-u.ac.jp
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
1 Aaronson NK, Bullinger M, Ahmedzai S. A modular approach to quality-of-life assessment in cancer clinical trials. Recent Results Cancer Res 1988;111:231–49.[Medline]
2 Aaronson NK, Ahmedzai S, Bergman B, Bullinger M, Cull A, Duez NJ, et al. The European Organization 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 1993;85:365–76.
3 Osoba D. Lessons learned from measuring health-related quality of life in oncology. J Clin Oncol 1994;12:608–16.[Abstract]
4 Moinpour CM, Feigl P, Metch B, Hayden KA, Meyskens FL, Crowley J. Quality of life end points in cancer clinical trials: reviews and recommendations. J Natl Cancer Inst 1989;81:485–95.
5 Gotay CC, Korn EL, McCabe MS, Moore TD, Cheson BD. Quality-of-life assessment in cancer treatment protocols: research issues in protocol development. J Natl Cancer Inst 1992;84:575–9.
6 Nayfield SG, Ganz PA, Moinpour CM, Cella DF, Hailey BJ. Report from a National Cancer Institute (USA) Workshop on quality of life assessment in cancer clinical trials. Qual Life Res 1992;1:203–10.[CrossRef][Medline]
7 Machin D. Assessment of quality of life in clinical trials of the British Medical Research Council. J Natl Cancer Inst Monogr 1996;20:97–102.
8 Ganz PA. Quality of life and the patient with cancer: individual and policy implications. Cancer 1994;74(Suppl 4):1445–52.[CrossRef][Web of Science][Medline]
9 Rowland JH. Outcome assessment: cancer-specific quality-of-life measures – beyond the research setting. In: Educational Book, ASCO 33rd Annual Meeting, Denver, CO, May 17–20, 1997;342–9.
10 Cella DF, Tulsky DS, Gray G, Sarafian B, Linn E, Bonomi A, et al. The functional assessment of cancer therapy scale: development and validation of the general measure. J Clin Oncol 1993;11:570–9.
11 Kurihara M, Shimizu H, Tsuboi K, Kobayashi K, Murakami M, Eguchi K, et al. Development of quality of life questionnaire in Japan: quality of life assessment of cancer patients receiving chemotherapy. Psycho-Oncology 1999;8:355–63.[CrossRef][Medline]
12 Osoba D, Zee B, Warr D, Kaizer L, Latreille J, Pater J. Quality of life studies in chemotherapy-induced emesis. Oncology 1996;53(Suppl 1):92–5.
13 Lindley CM, Hirsch JD, O’Neill CV, Transau MC, Gilbert CS, Osterhaus JT. Quality of life consequences of chemotherapy-induced emesis. Qual Life Res 1992;1:331–40.[CrossRef][Medline]
14 O’Brien BJ, Rusthoven J, Rocchi A, Latreille J, Fine S, Vandenberg T, et al. Impact of chemotherapy-associated nausea and vomiting on patients’ functional status and on costs: survey of five Canadian centres. Can Med Assoc J 1993;143:296–302.
15 Luoma ML, Hakamies-Blomqvist L, Sjostrom J, Mouridsen H, Pluzanska A, Malmstrom P, et al. Physical performance, toxicity and quality of life as assessed by the physician and the patient. Acta Oncol 2002;41:44–9.[CrossRef][Web of Science][Medline]
16 Harper-Wynne C, English J, Meyer L, Bower M, Archer C, Sinnett HD, et al. Randomized trial to compare the efficacy and toxicity of cyclophosphamide, methotrexate and 5-fluorouracil (CMF) with methotrexate mitoxantrone (MM) in advanced carcinoma of the breast. Br J Cancer 1999;81:316–22.[CrossRef][Web of Science][Medline]
17 Tchekmedyian NS, Hickman M, Siau J, Greco A, Aisner J. Treatment of cancer anorexia with megestrol acetate: impact on quality of life. Oncology 1990;4:185–92.[Medline]
18 Diggle PJ, Heagerty P, Liang KY, Zeger SL. Analysis of Longitudinal Data, 2nd ed. Oxford: Oxford, Oxford University Press 2002:22–80.
19 Matsumoto T, Ohashi Y, Morita S, Kobayashi K, Shibuya M, Yamaji Y, et al. QOL questionnaire for cancer patients treated with anticancer drugs (QOL-ACD): validity and reliability in Japanese patients with advanced non-small cell lung cancer. Qual Life Res 2002;11:483–93.[CrossRef][Web of Science][Medline]
20 World Health Organization. WHO Handbook for Reporting Results of Cancer Treatment. WHO Offset Publication No. 48. Geneva: WHO 1979.
21 Oken MM, Creech RH, Tormey DC, Horton J, Davis TE, McFadden ET, et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol 1982;5:649–55.[Web of Science][Medline]
22 Galecki AT. General class of covariance structures for two or more repeated factors in longitudinal data analysis. Commun Statist 1994;23:3105–19.
23 Littell R, Milliken G, Stroup W, Wolfinger RD. SAS System for Mixed Models. Cary, SC: SAS Institute 1996;87–134.
24 Staquet MJ, Hays RD, Fayers PM. Quality of Life Assessment in Clinical Trials. Oxford: Oxford University Press 1998;227–80.
25 Curran D, Molenberghs G, Fayers PM, Machin D. Incomplete quality of life data in randomized trials: missing forms. Stat Med 1998;17:697–709.[CrossRef][Web of Science][Medline]
26 Bernhard BJ, Cella DF, Coates AS, Fallowfield L, Ganz PA, Moinpour CM, et al. Missing quality of life data in cancer clinical trials: serious problems and challenges. Stat Med 1998;17:517–32.[CrossRef][Web of Science][Medline]
27 Moinpour CM, Lyons B, Grevstad PK, Lovato LC, Crowley J, Czaplicki K, et al. Quality of life in advanced non-small cell lung cancer: results of a Southwest Oncology Group randomized trial. Qual Life Res 2002;11:115–26.[CrossRef][Web of Science][Medline]
Received January 5, 2003; accepted August 12, 2003
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. J. Gralla Quality-of-Life Considerations in Patients with Advanced Lung Cancer: Effect of Topotecan on Symptom Palliation and Quality of Life Oncologist, December 1, 2004; 9(suppl_6): 14 - 24. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||