Quality of Life Outcomes From a Randomized Phase III Trial of Cisplatin With or Without Topotecan in Advanced Carcinoma of the Cervix: A Gyn
http://www.100md.com
《临床肿瘤学》
the Division of Gynecologic Oncology, Chao Family Comprehensive Cancer Center, University of California Irvine Medical Center, Orange, CA
Gynecologic Oncology Group Statistical and Data Center, Roswell Park Cancer Institute, Buffalo, NY
Department of Psychiatry and Behavioral Science, Institute for Health Services Research and Policy Studies, Northwestern University
Center on Outcomes, Research and Education, Evanston Northwestern Healthcare, Evanston, IL
Mayo Clinic College of Medicine, Rochester, MN
ABSTRACT
PURPOSE: To prospectively assess the impact of treatment with cisplatin alone or in combination with topotecan (CT) on quality of life (QOL) in patients with advanced or recurrent cervical cancer, and to explore the prognostic value of baseline QOL scores.
PATIENTS AND METHODS: Patients entered on Gynecologic Oncology Group (GOG) Protocol 179 were expected to complete QOL assessments at four time points using Functional Assessment of Cancer Therapy–General (FACT-G), Cervix subscale (Cx subscale), FACT/GOG–Neurotoxicity subscale (NTX subscale), Brief Pain Inventory (BPI), and UNISCALE (UNI). Adjusting for patient age, baseline scores, and effects of time, we longitudinally examined treatment effect on QOL during and after chemotherapy.
RESULTS: Among patients randomly allocated to receive cisplatin (n = 146) or CT (n = 147), there were no statistically significant differences in QOL up to 9 months after randomization despite more hematologic toxicity in the combination arm. QOL assessments were completed at rates of 98%, 85%, 68%, and 59%, respectively, for the four time points, with similar rates and reasons for nonparticipation between regimens. Baseline FACT-G (P = .0016) and BPI (P = .0001) scores were significantly associated with patient age; older patients had better QOL and less pain. Baseline UNI was positively correlated with FACT-G (r = 0.66; P < .001) and Cx subscale (r = 0.29; P < .001), and negatively related to BPI (r = –0.41; P < .0001). Baseline FACT-Cx (FACT-G + Cx subscale) was associated with survival.
CONCLUSION: Despite increased toxicity, CT did not significantly reduce patient QOL when compared with cisplatin alone. Patient-reported QOL measures may be an important prognostic tool in advanced cervix cancer.
INTRODUCTION
Despite effective screening programs, cervical carcinoma is second only to breast cancer in malignant neoplastic diseases among women worldwide, with almost one half million new cases annually.1,2 Advances in primary treatment have improved local control and survival; however, recurrent and/or advanced disease is still common, such that identifying effective systemic therapies is of paramount importance.2,3
Currently, cisplatin is the most widely used cytotoxic chemotherapeutic agent for treatment of this disease. When administered intravenously at a dose of 50 mg/m2, multicenter trials have reported response rates of approximately 20%, with median survival rates of approximately 7 to 9 months.3,4 Because randomized trials of other therapies have failed to show a survival advantage, and careful comparisons of single-agent cisplatin to best supportive care are lacking, palliative care alone remains another alternative.2-4 Topotecan, a topoisomerase I inhibitor, has shown significant activity in recurrent cervical cancer as both a single agent5 and in combination with cisplatin.6 On the basis of this activity and its tolerable toxicity profile, the Gynecologic Oncology Group (GOG) selected the combination of cisplatin and topotecan (CT) for a phase III study in this patient population.7,8
The GOG has shown that response rate and progression-free survival (PFS) of cisplatin can be improved by increasing the dose or by adding ifosfamide or paclitaxel.4 Although it confers no survival advantage, the addition of paclitaxel has been advocated as standard treatment for recurrent cervical cancer9 because of its dramatic improvement in response rate, to almost double that seen with cisplatin alone. Importantly, despite increased toxicity from the addition of paclitaxel, quality-of-life (QOL) measurements did not show any significant negative impact when this combination was compared with cisplatin (McQuellon RP et al, submitted for publication).9
When chemotherapeutic agents are combined and there is potential for increased toxicity, QOL measures become critical end points. Patient-reported QOL outcomes allow us to assess the impact of increased toxicity in the presence of improved response rates or survival, and to estimate the effect of combination therapies on disease-related symptoms such as pain and suffering.
The primary clinical objective of GOG Protocol 179 was to determine if CT improves survival compared with cisplatin in the treatment of advanced cervical cancer. The study’s secondary end point, which is the subject of this report, was to compare the impact of these treatments on quality of life.
PATIENTS AND METHODS
Eligibility
Consenting patients with measurable advanced (stage IVB) recurrent or persistent cervical carcinoma who were unsuitable candidates for curative treatment with surgery and/or radiotherapy, and who met all other eligibility criteria, were entered onto GOG Protocol 179.7,8 Consistent with federal, state, and local regulations, patients also provided written informed consent to participate in the QOL component of the treatment study before random assignment to treatment.
Chemotherapy
Random assignment of treatment was to either single-agent cisplatin 50 mg/m2 intravenously on day 1 repeated every 21 days (cisplatin), or to topotecan 0.75 mg/m2 intravenously on days 1 to 3 plus cisplatin 50 mg/m2 on day 1 repeated every 21 days (CT).
QOL
All living patients were expected to complete a 67-question QOL assessment at four time points: at baseline (before random assignment to treatment), just before the second and fifth chemotherapy cycles, and 9 months after random assignment to treatment. Assessments were to be completed as scheduled even if treatment was terminated. If treatment was delayed, the corresponding QOL assessment was to be completed as near to the scheduled time point as possible, but not more than 2 weeks later. The reason for a missed assessment was to be documented. Reasons for a missed assessment included patient death, patient refusal, illness, institutional error, loss of contact with patient, incomplete questionnaire, or other reason.
Every effort was made to minimize patient burden by using easy to complete yet comprehensive questionnaires to capture information related to overall QOL, neurotoxicity-related symptoms, and pain experience during the course of treatment. Assessment instruments included Functional Assessment of Cancer Therapy–Cervix subscale (FACT-Cx), FACT/GOG–Neurotoxicity subscale, Brief Pain Inventory (BPI), and UNISCALE (UNI).
FACT-Cx. The FACT-Cx is the FACT-G plus the Cx subscale (McQuellon RP et al, submitted for publication).10 The FACT-G (version 4) is a 27-item self-reporting QOL measure developed and validated among cancer patients for use in clinical trials.11 It consists of four subscales measuring physical well-being, functional well-being, social/family well-being, and emotional well-being. Each subscale produces a score that can be aggregated into one total score. The Cx subscale contains 15 additional items developed by cervical cancer patients and clinicians (McQuellon RP et al, submitted for publication).
The Fact-Cx can be self-administered or used in an interview format, and is easily completed in 10 to 15 minutes. Patients are asked to score how they feel today and during the previous 7 days, with a higher score indicating better QOL. Cronbach’s for each subscale has been reported as follows: physical well-being (.82), functional well-being (.80), social/family well-being (.69), emotional well-being (.74), and total FACT-G (.89).
NTX subscale. Patient-reported neurotoxicity symptoms and concerns were assessed with the NTX subscale of the FACT/GOG-NTX, version 4. The FACT/GOG-NTX, a 38-item self-reporting questionnaire consisting of two components, a general measure of QOL (FACT-G), and an 11-item NTX subscale. The NTX subscale has demonstrated sensitivity to meaningful clinical distinctions and change over time.12
Consistent with the Functional Assessment of Chronic Illness Therapy (FACIT) measurement system, a 5-point scale from 0 to 4 was applied to the Fact-G, the Cx subscale, and the NTX subscale, where 0 = not at all, 1 = a little bit, 2 = somewhat, 3 = quite a bit, and 4 = very much. The range of possible scores is 0 to 108 for the FACT-G, 0 to 60 for the Cx subscale, and 0 to 44 for the NTX subscale, respectively.
BPI. The BPI consists of 14 questions designed to assess pain related to cancer and other diseases.13 It measures pain intensity (five-item sensory dimension), and interference of pain in the patient’s life (seven-item reactive dimension). Each item is scaled from 0 to 10, where 0 = no pain/interference and 10 = pain as bad as you can imagine/complete interference. Two items regarding medications and their effect were not included here because they were not relevant to this study. The BPI has demonstrated both reliability and validity across cultures and languages, and has been used to study the effectiveness of pain treatment.14 Because it may be a result of disease progression, pain was measured separately to assess the differential effects of the treatment arms. Given that the BPI used in this study consisted of 12 questions each scaled from 0 to 10, we calculated the total possible score as 0 to 120.
UNI. The UNI is a single-item visual analog scale that asks the patient to place an "x" on a 0 to 10 scale representing overall QOL.15 A recent study of advanced colorectal cancer patients examining the utility of four QOL instruments, each with varying numbers of items, provided support for UNI because it is easy to complete and it correlates with longer, more comprehensive measures.16
Statistical Considerations
To assess QOL, scores for each subscale in the FACT-G, the Cx and NTX subscales and BPI were computed if more than 50% of subscale items were answered. A subscale score Si with Ni items was calculated as
where ij is equal to 1 when the jth item has a valid response; otherwise it is equal to 0 and sij is the score of the jth item. The total FACT-G score is the sum of the subscale scores and is considered valid only if at least 80% of the items are completed. Higher scores are associated with better QOL (FACT-G, Cx subscale, and UNI) and worse adverse effects (NTX subscale and BPI).
Subsequent to random assignment, patients who did not participate in the QOL component of the study were still included in the intent-to-treat QOL analysis. The sample size of 133 patients for each regimen was calculated based on the primary study objective. Assuming 80% of the patients in each arm (106 patients) would complete the FACT-G questionnaire at the second time point, and a correlation between the repeated measures was as high as 0.8, for a two-tailed test and 5% type I error, the sample size provided a 90% power to detect a difference of 7.0 units in the FACT-G, which was considered clinically significant.17 The final assessment was scheduled for completion 9 months after random assignment or approximately 5 months after completion of protocol treatment. Given that this time point is relatively distant from the date of diagnosis, and this is an aggressive disease, it is not unexpected that compliance would decrease despite every effort. Thus, the comparison of treatment difference was considered exploratory at this time point. In this study, compliance was defined as the number of valid QOL assessments as a proportion of those expected at each time point. Other than for patients who died or refused, assessments that were anticipated but absent were considered noncompliance. To compare compliance between the two treatment arms, the generalized estimating equations method with unstructured working correlation matrix was applied. Compliance rates were assumed to follow binomial probability distribution, and logistic link function was chosen to relate to the linear predictor. The interaction of treatment and assessment time was examined for constant treatment effect on compliance across time.
The a priori hypotheses of the current study were that the addition of topotecan to cisplatin would prolong survival of women with advanced or recurrent cervical carcinoma, but would be associated with a corresponding decrement in QOL because of the increased toxicity associated with this combination chemotherapy regimen. The FACT-G, Cx subscale, NTX subscale, and BPI scores were studied separately because the important conceptual distinction among them might be lost if they were included together. Because of the multiplicity of outcome measures, each measure was tested at 0.0125 to control the overall type I error at the 5% level.
The general linear model was applied to examine whether the QOL scores were different between the two regimens at baseline, adjusting for patient age at entry. To examine whether the follow-up QOL scores (including FACT-G, Cx subscale, NTX subscale, and BPI) were affected by treatment, a mixed linear model with restricted maximum likelihood and unstructured covariance was applied and adjusted by time effect, patient age at entry, and baseline scores. To explore the association between survival and baseline QOL score, Cox proportional hazards models were fitted, in which ties in the failure times were handled with the approximate likelihood as described by Efron.18
RESULTS
Patient Characteristics
Between June 1999 and September 2002, 293 eligible women enrolled onto GOG Protocol 179 were randomly allocated to receive either cisplatin (n = 146) or CT (n = 147; Table 1). Patient characteristics were well balanced between the two regimens, with a median age of 48 (cisplatin) and 46 years (CT). Interestingly, although cervical cancer is more common among ethnic minorities,1,2 almost three fourths of the patients were white.
Chemotherapy
An increased response rate and an improved median progression-free survival and overall survival among patients in the CT arm were statistically significant. Detailed treatment results are presented elsewhere.7,8
QOL Compliance
After random assignment, three patients (cisplatin, one patient; CT, two patients) refused to participate in the QOL component; thus, 145 patients in each arm were included in the analyses. Of them, 143 patients (99%) in the cisplatin arm versus 141 patients (97%) in the CT arm provided a valid baseline QOL assessment. The compliance rate for valid/expected questionnaires decreased to 86% and 84% at the second time point, to 64% and 72% at the third time point, and to 56% and 62% at the fourth time point for the cisplatin and CT arms, respectively (Table 2). A fitted generalized estimating equation model indicates there was no significant difference between treatment arms in the compliance rate across time points.
Sixty-four patients completed all four assessments, 229 missed at least one assessment, and only six patients did not complete any assessments. The most common reason for missing assessments was death; 164 patients (56%) died during the course of assessment. The reasons for noncompliance are summarized in Table 3. Although there is some distinction at the third assessment point, when the major reasons for noncompliance are institutional error in the cisplatin arm and illness in the CT arm, overall they are similarly distributed in the two regimens.
Relationship Between QOL and Missing Assessments
Among 223 patients who missed at least one assessment (excluding those who did not complete any), 17 intermittently missed assessments and 206 missed all subsequent assessments. The latter group, labeled dropouts (and which included patients who died), was separated into three subgroups according to the time point at which they dropped out. Their QOL scores over time were plotted together with the scores of patients who completed all scheduled assessments to explore whether missing assessments were associated with worsening QOL before dropout. Figure 1 showed that patients with better QOL (relative to the FACT-G and Cx subscale) and worse adverse effects (relative to the NTX subscale and BPI) were more likely to complete subsequent QOL assessments. For dropouts in both treatment arms, the analysis of variance for the first two time points and a two-sample t test for the third time point showed that patients who reported worse QOL (relative to the FACT-G and Cx subscale; P < .01) and severe pain symptoms (P < .001) tended to drop out or die earlier in the study.
QOL End Points
There was no statistical difference between arms in baseline QOL (Table 4; Fig 2). Baseline FACT-G scores (P = .0016) and BPI (P = .0001) scores were significantly associated with patient age at study entry. Older women tended to report better general QOL (2.7 points higher) and fewer pain symptoms (7.7 points lower) for every 10 years of age.
The mixed linear models were fitted for the subsequent QOL scores (including FACT-G, Cx subscale, NTX subscale, BPI, and UNI). After adjusting for baseline scores and age at entry, there was no statistical evidence suggesting that reported QOL and adverse effects scores changed over time differently across regimens (meaning the interaction of time and treatment was not statistically significant). Both arms presented nearly stable FACT-G and Cx subscale scores during the assessment period; however, there was an increasing trend for the NTX score (P < .001) and a declining trend for the BPI score (P = .03) for time effect. The increment in NTX scores did not occur until 9 months after random assignment for both arms. Follow-up QOL scores were significantly associated (P < .001) with baseline scores, such that if a patient initially had a poor QOL or adverse effects score, subsequent assessments were similar. In addition, neurologic (P < .001) and pain symptoms (P < .001) also had a significantly negative effect on QOL, such that patients with worse neurotoxicity or pain symptoms tended to report lower QOL.
In addition, the UNI did not differ between treatment regimens (Table 4) but was significantly correlated with the FACT-G (Pearson correlation coefficient, 0.65; P < .001), Cx subscale (r = 0.28; P < .001), and BPI (r = –0.40; P < .0001) at baseline.
Prognostic Value of QOL (FACT-Cx)
The association between baseline QOL score (FACT-Cx) and survival was explored by fitting a Cox proportional hazards model that adjusted for treatment effect, age, and performance status at baseline. The baseline FACT-Cx as a continuous variable was not found to be associated with PFS, but was significantly associated with overall survival (for every 10 units higher of FACT-Cx score, hazard ratio [HR], 0.91; 95% CI, 0.86 to 0.97; P = .002). To further explore the impact of QOL on survival, patients were divided into four subgroups according to the quartiles of the FACT-Cx score (Fig 3). The estimated hazard of death for patients in the highest QOL quartile was 47% lower than for patients in the lowest quartile (HR, 0.53; 95% CI, 0.36 to 0.78; P = .001), 40% lower than for patients in the second lowest quartile (HR, 0.60; 95% CI, 0.41 to 0.89; P = .001), and 40% lower than patients in the third lowest quartile (HR, 0.60; 95% CI, 0.40 to 0.88; P = .001). There were no significant differences among the lower three quartiles.
DISCUSSION
Historically, treatment trials, including those conducted by the GOG, relied on traditional clinical outcomes to determine the level of activity attributable to a given therapeutic approach or to measure differences between one or more therapies. The psychosocial implication of disease and treatment is a relatively recent and evolving field of study, such that the current trial represents only the second GOG study (and the first to be published) in cervical carcinoma to include QOL measures in a large-scale, randomized chemotherapy trial. If some investigators were previously unconvinced of the value to be gained from collecting psychosocial data, the results of this study clearly demonstrate that (in conjunction with standard indicators including response, survival, and physician-rated toxicity) self-reported QOL assessments offer a unique perspective that is equally germane to discussions of treatment cost versus benefit.
The primary objectives of GOG Protocol 1797,8 were to compare the effect of combination CT versus cisplatin on survival, response, and toxicity in patients with advanced cervical cancer. In designing the study, there was speculation that topotecan could prolong survival, but that its addition to cisplatin created the potential for increased adverse effects. This question of benefit (increased length of life) versus cost (compromised quality of life) was instrumental in the development of this prospective study in which patients reported serially on the impact and interference of disease and treatment in this setting. As reported by Long et al,7,8 the addition of topotecan to cisplatin resulted in a statistically significant improvement in median progression-free survival and median overall survival at a cost of substantially more toxicity. Our (secondary) report demonstrates that the more toxic combination regimen (CT) produced no significant reduction in QOL, and this effect remained stable over time. Not only were these data useful in assessing these two regimens in terms of the nature, severity, and longevity of disruptions to patients’ QOL, in addition, the degree of that impairment at baseline was found to have prognostic significance. This dramatic association with survival is illustrated in Figure 3.
Although the completion rate of QOL questionnaires was high at baseline, attrition (mostly because of patient death as a result of cervical cancer) resulted in progressively lower rates of completion throughout the study (Table 2). However, it is important to note that rates and reasons for noncompliance were similar between treatment arms and at each time point (Table 3). This negates the potential for bias that could be induced by missing data because the incidence of noncompliance is similar in both arms and across time points. Thus, the interpretation of QOL data in this trial is more reliable than if the missing data were unbalanced. Moreover, the QOL assessment used in the current trial was reliable in terms of sensitivity to detect clinically significant differences in QOL over time.
To further minimize inconvenience to patients and staff, and to evaluate the sensitivity of a simplistic QOL instrument, Spitzer’s single-item UNI15 was used. This tool has been advanced by some investigators to alleviate the problem of missing data that sometimes occurs with longer questionnaires.19 Conversely, it has been argued that single-item instruments are more a measure of emotional than physical health.20 In this study, the UNI performed similarly to the total score of the multi-item (FACT-Cx) instrument. Future studies using both multi-item and single-item instruments could provide additional opportunity for comparison of outcome and compliance rates.
It has been shown that the use of CT results in increased survival in advanced and recurrent cervical cancer when compared with cisplatin. Equally important, despite the increase in toxicity with CT, there were no between-regimen differences in QOL scores during the treatment period or up to 9 months after randomization; taken together, these results clearly favor CT compared with cisplatin in this patient population. The prospective inclusion of reliable and disease-specific assessment instruments as used in this study, along with a positive commitment from practitioners and a carefully constructed statistical design, has allowed us to demonstrate the applicability of QOL results to the evaluation of current and future therapeutic studies.
Appendix
The following Gynecologic Oncology Group institutions participated in this study: University of Alabama at Birmingham, Duke University Medical Center, Abington Memorial Hospital, Walter Reed Army Medical Center, Wayne State University, University of Minnesota Medical School, University of Mississippi Medical Center, Colorado Gynecologic Oncology Group, PC, University of California at Los Angeles, University of Washington, University of Pennsylvania Cancer Center, Milton S. Hershey Medical Center, University of Cincinnati, University of North Carolina School of Medicine, University of Iowa Hospitals and Clinics, University of Texas Southwestern Medical Center at Dallas, University of Indiana Medical Center, Wake Forest University School of Medicine, Albany Medical College, University of California Medical Center at Irvine, Tufts-New England Medical Center, Rush-Presbyterian-St. Luke’s Medical Center, SUNY Downstate Medical Center, University of Kentucky, Community Clinical Oncology Program, The Cleveland Clinic Foundation, State University of New York at Stony Brook, Washington University School of Medicine, Cooper Hospital/University Medical Center, Columbus Cancer Council, University of Massachusetts Medical School, Fox Chase Cancer Center, Medical University of South Carolina, Women’s Cancer Center, University of Oklahoma, University of Virginia, University of Chicago, Tacoma General Hospital, Thomas Jefferson University Hospital, Mayo Clinic, Case Western Reserve University, Tampa Bay Cancer Consortium, North Shore University Hospital, and Ellis Fischel Cancer Center.
Authors' Disclosures of Potential Conflicts of Interest
The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Consultant/Advisory Role: Bradley J. Monk, GlaxoSmithKline; Harry J. Long III, GlaxoSmithKline. Stock Ownership: Harry J. Long III, Genentech, GlaxoSmithKline, Pfizer. Honoraria: Bradley J. Monk, GlaxoSmithKline; Harry J. Long III, GlaxoSmithKline. Research Funding: Bradley J. Monk, GlaxoSmithKline. For a detailed description of these categories, or for more information about ASCO’s conflict of interest policy, please refer to the Author Disclosure Declaration and Disclosures of Potential Conflicts of Interest found in Information for Contributors in the front of each issue.
Acknowledgment
We thank Caron Modeas for editorial assistance in preparing this article.
NOTES
Supported by National Cancer Institute grants to the Gynecologic Oncology Group Administrative Office (CA 27469), the Gynecologic Oncology Group Statistical and Data Center (CA 37517), and Bradley J. Monk, Principal Investigator (CA 87558-04).
Presented in abstract form at the Annual Meeting of the Society of Gynecologic Oncologists, San Diego, CA, 2004.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Franco EL, Schlecht NF, Saslow D: The epidemiology of cervical cancer. Cancer J 9:348-359, 2003
DiSaia PJ, Creasman WT: Invasive cervical cancer. In: Clinical Gynecologic Oncology (ed 6). St. Louis, MO, Mosby-Year Book, 2002, pp 58-111
Im SS, Monk BJ: New developments in the treatment of invasive cervical cancer. Obstet Gynecol Clin North Am 29:659-672, 2002
Thigpen T: The role of chemotherapy in the management of carcinoma of the cervix. Cancer J 9:425-432, 2003
Fiorica JV: The role of topotecan in the treatment of advanced cervical cancer. Gynecol Oncol 90:S16-S21, 2003
Fiorica J, Holloway R, Ndubisi B, et al: Phase II trial of topotecan and cisplatin in persistent or recurrent squamous and nonsquamous carcinomas of the cervix. Gynecol Oncol 85:89-94, 2002
Long HJ III, Bundy BN, Grendys EC Jr, et al: Randomized phase III trial of cisplatin with or without topotecan in carcinoma of the uterine cervix: A Gynecologic Oncology Group Study. J Clin Oncol 23:10.1200/JCO.2005.10.021
Long HJ, Bundy BN, Grendys EC, et al: Randomized phase III trial of cisplatin (P) vs cisplatin plus topotecan (T) vs MVAC in stage IVB, recurrent or persistent carcinoma of the uterine cervix: A Gynecologic Oncology Group study. Gynecol Oncol 92:397, 2004 (abstr 9)
Moore DH, Blessing JA, McQuellon RP, et al: Phase III study of cisplatin with or without paclitaxel in stage IVB, recurrent or persistent squamous cell carcinoma of the cervix: A Gynecologic Oncology Group study. J Clin Oncol 22:3113-3119, 2004
FACIT: Functional Assessment of Chronic Illness Therapy. http://www.facit.org
Cella DF, Tulsky DS, Gray G, et al: The Functional Assessment of Cancer Therapy scale: Development and validation of the general measure. J Clin Oncol 11:570-579, 1993
Calhoun EA, Welshman EE, Chang CH, et al: Psychometric evaluation of the Functional Assessment of Cancer Therapy/Gynecologic Oncology Group-Neurotoxicity (Fact/GOG-Ntx) questionnaire for patients receiving systemic chemotherapy. Int J Gynecol Cancer 13:741-748, 2003
Cleeland CS, Ryan KM: Pain assessment: Global use of the Brief Pain Inventory. Ann Acad Med Singapore 23:129-138, 1994
Payne R, Mathias SD, Pasta DJ, et al: Quality of life and cancer pain: Satisfaction and side effects with transdermal fentanyl versus oral morphine. J Clin Oncol 16:1588-1593, 1998
Spitzer W, Dobson AJ, 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
Sloan JA, Loprinzi CL, Kuross SA, et al: Randomized comparison of four tools measuring overall quality of life in patients with advanced cancer. J Clin Oncol 16:3662-3670, 1998
Cella D, Hahn E, Dineen K: Meaningful change in cancer-specific quality of life scores: Differences between improvement and worsening. Qual Life Res 11:207-221, 2002
Efron B: The efficiency of Cox's likelihood function for censored data. J Am Stat Assoc 72:557-565, 1977
Hobday TJ, Kugler JW, Mahoney MR, et al: Efficacy and quality of life data are related in a phase II trial of oral chemotherapy in previously untreated patients with metastatic colorectal carcinoma. J Clin Oncol 20:4574-4580, 2002
Cella D: What do global quality-of-life questions really measure Insights from Hobday et al and the "Do Something" rule. J Clin Oncol 21:3178-3179, 2003(Bradley J. Monk, Helen Q.)
Gynecologic Oncology Group Statistical and Data Center, Roswell Park Cancer Institute, Buffalo, NY
Department of Psychiatry and Behavioral Science, Institute for Health Services Research and Policy Studies, Northwestern University
Center on Outcomes, Research and Education, Evanston Northwestern Healthcare, Evanston, IL
Mayo Clinic College of Medicine, Rochester, MN
ABSTRACT
PURPOSE: To prospectively assess the impact of treatment with cisplatin alone or in combination with topotecan (CT) on quality of life (QOL) in patients with advanced or recurrent cervical cancer, and to explore the prognostic value of baseline QOL scores.
PATIENTS AND METHODS: Patients entered on Gynecologic Oncology Group (GOG) Protocol 179 were expected to complete QOL assessments at four time points using Functional Assessment of Cancer Therapy–General (FACT-G), Cervix subscale (Cx subscale), FACT/GOG–Neurotoxicity subscale (NTX subscale), Brief Pain Inventory (BPI), and UNISCALE (UNI). Adjusting for patient age, baseline scores, and effects of time, we longitudinally examined treatment effect on QOL during and after chemotherapy.
RESULTS: Among patients randomly allocated to receive cisplatin (n = 146) or CT (n = 147), there were no statistically significant differences in QOL up to 9 months after randomization despite more hematologic toxicity in the combination arm. QOL assessments were completed at rates of 98%, 85%, 68%, and 59%, respectively, for the four time points, with similar rates and reasons for nonparticipation between regimens. Baseline FACT-G (P = .0016) and BPI (P = .0001) scores were significantly associated with patient age; older patients had better QOL and less pain. Baseline UNI was positively correlated with FACT-G (r = 0.66; P < .001) and Cx subscale (r = 0.29; P < .001), and negatively related to BPI (r = –0.41; P < .0001). Baseline FACT-Cx (FACT-G + Cx subscale) was associated with survival.
CONCLUSION: Despite increased toxicity, CT did not significantly reduce patient QOL when compared with cisplatin alone. Patient-reported QOL measures may be an important prognostic tool in advanced cervix cancer.
INTRODUCTION
Despite effective screening programs, cervical carcinoma is second only to breast cancer in malignant neoplastic diseases among women worldwide, with almost one half million new cases annually.1,2 Advances in primary treatment have improved local control and survival; however, recurrent and/or advanced disease is still common, such that identifying effective systemic therapies is of paramount importance.2,3
Currently, cisplatin is the most widely used cytotoxic chemotherapeutic agent for treatment of this disease. When administered intravenously at a dose of 50 mg/m2, multicenter trials have reported response rates of approximately 20%, with median survival rates of approximately 7 to 9 months.3,4 Because randomized trials of other therapies have failed to show a survival advantage, and careful comparisons of single-agent cisplatin to best supportive care are lacking, palliative care alone remains another alternative.2-4 Topotecan, a topoisomerase I inhibitor, has shown significant activity in recurrent cervical cancer as both a single agent5 and in combination with cisplatin.6 On the basis of this activity and its tolerable toxicity profile, the Gynecologic Oncology Group (GOG) selected the combination of cisplatin and topotecan (CT) for a phase III study in this patient population.7,8
The GOG has shown that response rate and progression-free survival (PFS) of cisplatin can be improved by increasing the dose or by adding ifosfamide or paclitaxel.4 Although it confers no survival advantage, the addition of paclitaxel has been advocated as standard treatment for recurrent cervical cancer9 because of its dramatic improvement in response rate, to almost double that seen with cisplatin alone. Importantly, despite increased toxicity from the addition of paclitaxel, quality-of-life (QOL) measurements did not show any significant negative impact when this combination was compared with cisplatin (McQuellon RP et al, submitted for publication).9
When chemotherapeutic agents are combined and there is potential for increased toxicity, QOL measures become critical end points. Patient-reported QOL outcomes allow us to assess the impact of increased toxicity in the presence of improved response rates or survival, and to estimate the effect of combination therapies on disease-related symptoms such as pain and suffering.
The primary clinical objective of GOG Protocol 179 was to determine if CT improves survival compared with cisplatin in the treatment of advanced cervical cancer. The study’s secondary end point, which is the subject of this report, was to compare the impact of these treatments on quality of life.
PATIENTS AND METHODS
Eligibility
Consenting patients with measurable advanced (stage IVB) recurrent or persistent cervical carcinoma who were unsuitable candidates for curative treatment with surgery and/or radiotherapy, and who met all other eligibility criteria, were entered onto GOG Protocol 179.7,8 Consistent with federal, state, and local regulations, patients also provided written informed consent to participate in the QOL component of the treatment study before random assignment to treatment.
Chemotherapy
Random assignment of treatment was to either single-agent cisplatin 50 mg/m2 intravenously on day 1 repeated every 21 days (cisplatin), or to topotecan 0.75 mg/m2 intravenously on days 1 to 3 plus cisplatin 50 mg/m2 on day 1 repeated every 21 days (CT).
QOL
All living patients were expected to complete a 67-question QOL assessment at four time points: at baseline (before random assignment to treatment), just before the second and fifth chemotherapy cycles, and 9 months after random assignment to treatment. Assessments were to be completed as scheduled even if treatment was terminated. If treatment was delayed, the corresponding QOL assessment was to be completed as near to the scheduled time point as possible, but not more than 2 weeks later. The reason for a missed assessment was to be documented. Reasons for a missed assessment included patient death, patient refusal, illness, institutional error, loss of contact with patient, incomplete questionnaire, or other reason.
Every effort was made to minimize patient burden by using easy to complete yet comprehensive questionnaires to capture information related to overall QOL, neurotoxicity-related symptoms, and pain experience during the course of treatment. Assessment instruments included Functional Assessment of Cancer Therapy–Cervix subscale (FACT-Cx), FACT/GOG–Neurotoxicity subscale, Brief Pain Inventory (BPI), and UNISCALE (UNI).
FACT-Cx. The FACT-Cx is the FACT-G plus the Cx subscale (McQuellon RP et al, submitted for publication).10 The FACT-G (version 4) is a 27-item self-reporting QOL measure developed and validated among cancer patients for use in clinical trials.11 It consists of four subscales measuring physical well-being, functional well-being, social/family well-being, and emotional well-being. Each subscale produces a score that can be aggregated into one total score. The Cx subscale contains 15 additional items developed by cervical cancer patients and clinicians (McQuellon RP et al, submitted for publication).
The Fact-Cx can be self-administered or used in an interview format, and is easily completed in 10 to 15 minutes. Patients are asked to score how they feel today and during the previous 7 days, with a higher score indicating better QOL. Cronbach’s for each subscale has been reported as follows: physical well-being (.82), functional well-being (.80), social/family well-being (.69), emotional well-being (.74), and total FACT-G (.89).
NTX subscale. Patient-reported neurotoxicity symptoms and concerns were assessed with the NTX subscale of the FACT/GOG-NTX, version 4. The FACT/GOG-NTX, a 38-item self-reporting questionnaire consisting of two components, a general measure of QOL (FACT-G), and an 11-item NTX subscale. The NTX subscale has demonstrated sensitivity to meaningful clinical distinctions and change over time.12
Consistent with the Functional Assessment of Chronic Illness Therapy (FACIT) measurement system, a 5-point scale from 0 to 4 was applied to the Fact-G, the Cx subscale, and the NTX subscale, where 0 = not at all, 1 = a little bit, 2 = somewhat, 3 = quite a bit, and 4 = very much. The range of possible scores is 0 to 108 for the FACT-G, 0 to 60 for the Cx subscale, and 0 to 44 for the NTX subscale, respectively.
BPI. The BPI consists of 14 questions designed to assess pain related to cancer and other diseases.13 It measures pain intensity (five-item sensory dimension), and interference of pain in the patient’s life (seven-item reactive dimension). Each item is scaled from 0 to 10, where 0 = no pain/interference and 10 = pain as bad as you can imagine/complete interference. Two items regarding medications and their effect were not included here because they were not relevant to this study. The BPI has demonstrated both reliability and validity across cultures and languages, and has been used to study the effectiveness of pain treatment.14 Because it may be a result of disease progression, pain was measured separately to assess the differential effects of the treatment arms. Given that the BPI used in this study consisted of 12 questions each scaled from 0 to 10, we calculated the total possible score as 0 to 120.
UNI. The UNI is a single-item visual analog scale that asks the patient to place an "x" on a 0 to 10 scale representing overall QOL.15 A recent study of advanced colorectal cancer patients examining the utility of four QOL instruments, each with varying numbers of items, provided support for UNI because it is easy to complete and it correlates with longer, more comprehensive measures.16
Statistical Considerations
To assess QOL, scores for each subscale in the FACT-G, the Cx and NTX subscales and BPI were computed if more than 50% of subscale items were answered. A subscale score Si with Ni items was calculated as
where ij is equal to 1 when the jth item has a valid response; otherwise it is equal to 0 and sij is the score of the jth item. The total FACT-G score is the sum of the subscale scores and is considered valid only if at least 80% of the items are completed. Higher scores are associated with better QOL (FACT-G, Cx subscale, and UNI) and worse adverse effects (NTX subscale and BPI).
Subsequent to random assignment, patients who did not participate in the QOL component of the study were still included in the intent-to-treat QOL analysis. The sample size of 133 patients for each regimen was calculated based on the primary study objective. Assuming 80% of the patients in each arm (106 patients) would complete the FACT-G questionnaire at the second time point, and a correlation between the repeated measures was as high as 0.8, for a two-tailed test and 5% type I error, the sample size provided a 90% power to detect a difference of 7.0 units in the FACT-G, which was considered clinically significant.17 The final assessment was scheduled for completion 9 months after random assignment or approximately 5 months after completion of protocol treatment. Given that this time point is relatively distant from the date of diagnosis, and this is an aggressive disease, it is not unexpected that compliance would decrease despite every effort. Thus, the comparison of treatment difference was considered exploratory at this time point. In this study, compliance was defined as the number of valid QOL assessments as a proportion of those expected at each time point. Other than for patients who died or refused, assessments that were anticipated but absent were considered noncompliance. To compare compliance between the two treatment arms, the generalized estimating equations method with unstructured working correlation matrix was applied. Compliance rates were assumed to follow binomial probability distribution, and logistic link function was chosen to relate to the linear predictor. The interaction of treatment and assessment time was examined for constant treatment effect on compliance across time.
The a priori hypotheses of the current study were that the addition of topotecan to cisplatin would prolong survival of women with advanced or recurrent cervical carcinoma, but would be associated with a corresponding decrement in QOL because of the increased toxicity associated with this combination chemotherapy regimen. The FACT-G, Cx subscale, NTX subscale, and BPI scores were studied separately because the important conceptual distinction among them might be lost if they were included together. Because of the multiplicity of outcome measures, each measure was tested at 0.0125 to control the overall type I error at the 5% level.
The general linear model was applied to examine whether the QOL scores were different between the two regimens at baseline, adjusting for patient age at entry. To examine whether the follow-up QOL scores (including FACT-G, Cx subscale, NTX subscale, and BPI) were affected by treatment, a mixed linear model with restricted maximum likelihood and unstructured covariance was applied and adjusted by time effect, patient age at entry, and baseline scores. To explore the association between survival and baseline QOL score, Cox proportional hazards models were fitted, in which ties in the failure times were handled with the approximate likelihood as described by Efron.18
RESULTS
Patient Characteristics
Between June 1999 and September 2002, 293 eligible women enrolled onto GOG Protocol 179 were randomly allocated to receive either cisplatin (n = 146) or CT (n = 147; Table 1). Patient characteristics were well balanced between the two regimens, with a median age of 48 (cisplatin) and 46 years (CT). Interestingly, although cervical cancer is more common among ethnic minorities,1,2 almost three fourths of the patients were white.
Chemotherapy
An increased response rate and an improved median progression-free survival and overall survival among patients in the CT arm were statistically significant. Detailed treatment results are presented elsewhere.7,8
QOL Compliance
After random assignment, three patients (cisplatin, one patient; CT, two patients) refused to participate in the QOL component; thus, 145 patients in each arm were included in the analyses. Of them, 143 patients (99%) in the cisplatin arm versus 141 patients (97%) in the CT arm provided a valid baseline QOL assessment. The compliance rate for valid/expected questionnaires decreased to 86% and 84% at the second time point, to 64% and 72% at the third time point, and to 56% and 62% at the fourth time point for the cisplatin and CT arms, respectively (Table 2). A fitted generalized estimating equation model indicates there was no significant difference between treatment arms in the compliance rate across time points.
Sixty-four patients completed all four assessments, 229 missed at least one assessment, and only six patients did not complete any assessments. The most common reason for missing assessments was death; 164 patients (56%) died during the course of assessment. The reasons for noncompliance are summarized in Table 3. Although there is some distinction at the third assessment point, when the major reasons for noncompliance are institutional error in the cisplatin arm and illness in the CT arm, overall they are similarly distributed in the two regimens.
Relationship Between QOL and Missing Assessments
Among 223 patients who missed at least one assessment (excluding those who did not complete any), 17 intermittently missed assessments and 206 missed all subsequent assessments. The latter group, labeled dropouts (and which included patients who died), was separated into three subgroups according to the time point at which they dropped out. Their QOL scores over time were plotted together with the scores of patients who completed all scheduled assessments to explore whether missing assessments were associated with worsening QOL before dropout. Figure 1 showed that patients with better QOL (relative to the FACT-G and Cx subscale) and worse adverse effects (relative to the NTX subscale and BPI) were more likely to complete subsequent QOL assessments. For dropouts in both treatment arms, the analysis of variance for the first two time points and a two-sample t test for the third time point showed that patients who reported worse QOL (relative to the FACT-G and Cx subscale; P < .01) and severe pain symptoms (P < .001) tended to drop out or die earlier in the study.
QOL End Points
There was no statistical difference between arms in baseline QOL (Table 4; Fig 2). Baseline FACT-G scores (P = .0016) and BPI (P = .0001) scores were significantly associated with patient age at study entry. Older women tended to report better general QOL (2.7 points higher) and fewer pain symptoms (7.7 points lower) for every 10 years of age.
The mixed linear models were fitted for the subsequent QOL scores (including FACT-G, Cx subscale, NTX subscale, BPI, and UNI). After adjusting for baseline scores and age at entry, there was no statistical evidence suggesting that reported QOL and adverse effects scores changed over time differently across regimens (meaning the interaction of time and treatment was not statistically significant). Both arms presented nearly stable FACT-G and Cx subscale scores during the assessment period; however, there was an increasing trend for the NTX score (P < .001) and a declining trend for the BPI score (P = .03) for time effect. The increment in NTX scores did not occur until 9 months after random assignment for both arms. Follow-up QOL scores were significantly associated (P < .001) with baseline scores, such that if a patient initially had a poor QOL or adverse effects score, subsequent assessments were similar. In addition, neurologic (P < .001) and pain symptoms (P < .001) also had a significantly negative effect on QOL, such that patients with worse neurotoxicity or pain symptoms tended to report lower QOL.
In addition, the UNI did not differ between treatment regimens (Table 4) but was significantly correlated with the FACT-G (Pearson correlation coefficient, 0.65; P < .001), Cx subscale (r = 0.28; P < .001), and BPI (r = –0.40; P < .0001) at baseline.
Prognostic Value of QOL (FACT-Cx)
The association between baseline QOL score (FACT-Cx) and survival was explored by fitting a Cox proportional hazards model that adjusted for treatment effect, age, and performance status at baseline. The baseline FACT-Cx as a continuous variable was not found to be associated with PFS, but was significantly associated with overall survival (for every 10 units higher of FACT-Cx score, hazard ratio [HR], 0.91; 95% CI, 0.86 to 0.97; P = .002). To further explore the impact of QOL on survival, patients were divided into four subgroups according to the quartiles of the FACT-Cx score (Fig 3). The estimated hazard of death for patients in the highest QOL quartile was 47% lower than for patients in the lowest quartile (HR, 0.53; 95% CI, 0.36 to 0.78; P = .001), 40% lower than for patients in the second lowest quartile (HR, 0.60; 95% CI, 0.41 to 0.89; P = .001), and 40% lower than patients in the third lowest quartile (HR, 0.60; 95% CI, 0.40 to 0.88; P = .001). There were no significant differences among the lower three quartiles.
DISCUSSION
Historically, treatment trials, including those conducted by the GOG, relied on traditional clinical outcomes to determine the level of activity attributable to a given therapeutic approach or to measure differences between one or more therapies. The psychosocial implication of disease and treatment is a relatively recent and evolving field of study, such that the current trial represents only the second GOG study (and the first to be published) in cervical carcinoma to include QOL measures in a large-scale, randomized chemotherapy trial. If some investigators were previously unconvinced of the value to be gained from collecting psychosocial data, the results of this study clearly demonstrate that (in conjunction with standard indicators including response, survival, and physician-rated toxicity) self-reported QOL assessments offer a unique perspective that is equally germane to discussions of treatment cost versus benefit.
The primary objectives of GOG Protocol 1797,8 were to compare the effect of combination CT versus cisplatin on survival, response, and toxicity in patients with advanced cervical cancer. In designing the study, there was speculation that topotecan could prolong survival, but that its addition to cisplatin created the potential for increased adverse effects. This question of benefit (increased length of life) versus cost (compromised quality of life) was instrumental in the development of this prospective study in which patients reported serially on the impact and interference of disease and treatment in this setting. As reported by Long et al,7,8 the addition of topotecan to cisplatin resulted in a statistically significant improvement in median progression-free survival and median overall survival at a cost of substantially more toxicity. Our (secondary) report demonstrates that the more toxic combination regimen (CT) produced no significant reduction in QOL, and this effect remained stable over time. Not only were these data useful in assessing these two regimens in terms of the nature, severity, and longevity of disruptions to patients’ QOL, in addition, the degree of that impairment at baseline was found to have prognostic significance. This dramatic association with survival is illustrated in Figure 3.
Although the completion rate of QOL questionnaires was high at baseline, attrition (mostly because of patient death as a result of cervical cancer) resulted in progressively lower rates of completion throughout the study (Table 2). However, it is important to note that rates and reasons for noncompliance were similar between treatment arms and at each time point (Table 3). This negates the potential for bias that could be induced by missing data because the incidence of noncompliance is similar in both arms and across time points. Thus, the interpretation of QOL data in this trial is more reliable than if the missing data were unbalanced. Moreover, the QOL assessment used in the current trial was reliable in terms of sensitivity to detect clinically significant differences in QOL over time.
To further minimize inconvenience to patients and staff, and to evaluate the sensitivity of a simplistic QOL instrument, Spitzer’s single-item UNI15 was used. This tool has been advanced by some investigators to alleviate the problem of missing data that sometimes occurs with longer questionnaires.19 Conversely, it has been argued that single-item instruments are more a measure of emotional than physical health.20 In this study, the UNI performed similarly to the total score of the multi-item (FACT-Cx) instrument. Future studies using both multi-item and single-item instruments could provide additional opportunity for comparison of outcome and compliance rates.
It has been shown that the use of CT results in increased survival in advanced and recurrent cervical cancer when compared with cisplatin. Equally important, despite the increase in toxicity with CT, there were no between-regimen differences in QOL scores during the treatment period or up to 9 months after randomization; taken together, these results clearly favor CT compared with cisplatin in this patient population. The prospective inclusion of reliable and disease-specific assessment instruments as used in this study, along with a positive commitment from practitioners and a carefully constructed statistical design, has allowed us to demonstrate the applicability of QOL results to the evaluation of current and future therapeutic studies.
Appendix
The following Gynecologic Oncology Group institutions participated in this study: University of Alabama at Birmingham, Duke University Medical Center, Abington Memorial Hospital, Walter Reed Army Medical Center, Wayne State University, University of Minnesota Medical School, University of Mississippi Medical Center, Colorado Gynecologic Oncology Group, PC, University of California at Los Angeles, University of Washington, University of Pennsylvania Cancer Center, Milton S. Hershey Medical Center, University of Cincinnati, University of North Carolina School of Medicine, University of Iowa Hospitals and Clinics, University of Texas Southwestern Medical Center at Dallas, University of Indiana Medical Center, Wake Forest University School of Medicine, Albany Medical College, University of California Medical Center at Irvine, Tufts-New England Medical Center, Rush-Presbyterian-St. Luke’s Medical Center, SUNY Downstate Medical Center, University of Kentucky, Community Clinical Oncology Program, The Cleveland Clinic Foundation, State University of New York at Stony Brook, Washington University School of Medicine, Cooper Hospital/University Medical Center, Columbus Cancer Council, University of Massachusetts Medical School, Fox Chase Cancer Center, Medical University of South Carolina, Women’s Cancer Center, University of Oklahoma, University of Virginia, University of Chicago, Tacoma General Hospital, Thomas Jefferson University Hospital, Mayo Clinic, Case Western Reserve University, Tampa Bay Cancer Consortium, North Shore University Hospital, and Ellis Fischel Cancer Center.
Authors' Disclosures of Potential Conflicts of Interest
The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Consultant/Advisory Role: Bradley J. Monk, GlaxoSmithKline; Harry J. Long III, GlaxoSmithKline. Stock Ownership: Harry J. Long III, Genentech, GlaxoSmithKline, Pfizer. Honoraria: Bradley J. Monk, GlaxoSmithKline; Harry J. Long III, GlaxoSmithKline. Research Funding: Bradley J. Monk, GlaxoSmithKline. For a detailed description of these categories, or for more information about ASCO’s conflict of interest policy, please refer to the Author Disclosure Declaration and Disclosures of Potential Conflicts of Interest found in Information for Contributors in the front of each issue.
Acknowledgment
We thank Caron Modeas for editorial assistance in preparing this article.
NOTES
Supported by National Cancer Institute grants to the Gynecologic Oncology Group Administrative Office (CA 27469), the Gynecologic Oncology Group Statistical and Data Center (CA 37517), and Bradley J. Monk, Principal Investigator (CA 87558-04).
Presented in abstract form at the Annual Meeting of the Society of Gynecologic Oncologists, San Diego, CA, 2004.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Franco EL, Schlecht NF, Saslow D: The epidemiology of cervical cancer. Cancer J 9:348-359, 2003
DiSaia PJ, Creasman WT: Invasive cervical cancer. In: Clinical Gynecologic Oncology (ed 6). St. Louis, MO, Mosby-Year Book, 2002, pp 58-111
Im SS, Monk BJ: New developments in the treatment of invasive cervical cancer. Obstet Gynecol Clin North Am 29:659-672, 2002
Thigpen T: The role of chemotherapy in the management of carcinoma of the cervix. Cancer J 9:425-432, 2003
Fiorica JV: The role of topotecan in the treatment of advanced cervical cancer. Gynecol Oncol 90:S16-S21, 2003
Fiorica J, Holloway R, Ndubisi B, et al: Phase II trial of topotecan and cisplatin in persistent or recurrent squamous and nonsquamous carcinomas of the cervix. Gynecol Oncol 85:89-94, 2002
Long HJ III, Bundy BN, Grendys EC Jr, et al: Randomized phase III trial of cisplatin with or without topotecan in carcinoma of the uterine cervix: A Gynecologic Oncology Group Study. J Clin Oncol 23:10.1200/JCO.2005.10.021
Long HJ, Bundy BN, Grendys EC, et al: Randomized phase III trial of cisplatin (P) vs cisplatin plus topotecan (T) vs MVAC in stage IVB, recurrent or persistent carcinoma of the uterine cervix: A Gynecologic Oncology Group study. Gynecol Oncol 92:397, 2004 (abstr 9)
Moore DH, Blessing JA, McQuellon RP, et al: Phase III study of cisplatin with or without paclitaxel in stage IVB, recurrent or persistent squamous cell carcinoma of the cervix: A Gynecologic Oncology Group study. J Clin Oncol 22:3113-3119, 2004
FACIT: Functional Assessment of Chronic Illness Therapy. http://www.facit.org
Cella DF, Tulsky DS, Gray G, et al: The Functional Assessment of Cancer Therapy scale: Development and validation of the general measure. J Clin Oncol 11:570-579, 1993
Calhoun EA, Welshman EE, Chang CH, et al: Psychometric evaluation of the Functional Assessment of Cancer Therapy/Gynecologic Oncology Group-Neurotoxicity (Fact/GOG-Ntx) questionnaire for patients receiving systemic chemotherapy. Int J Gynecol Cancer 13:741-748, 2003
Cleeland CS, Ryan KM: Pain assessment: Global use of the Brief Pain Inventory. Ann Acad Med Singapore 23:129-138, 1994
Payne R, Mathias SD, Pasta DJ, et al: Quality of life and cancer pain: Satisfaction and side effects with transdermal fentanyl versus oral morphine. J Clin Oncol 16:1588-1593, 1998
Spitzer W, Dobson AJ, 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
Sloan JA, Loprinzi CL, Kuross SA, et al: Randomized comparison of four tools measuring overall quality of life in patients with advanced cancer. J Clin Oncol 16:3662-3670, 1998
Cella D, Hahn E, Dineen K: Meaningful change in cancer-specific quality of life scores: Differences between improvement and worsening. Qual Life Res 11:207-221, 2002
Efron B: The efficiency of Cox's likelihood function for censored data. J Am Stat Assoc 72:557-565, 1977
Hobday TJ, Kugler JW, Mahoney MR, et al: Efficacy and quality of life data are related in a phase II trial of oral chemotherapy in previously untreated patients with metastatic colorectal carcinoma. J Clin Oncol 20:4574-4580, 2002
Cella D: What do global quality-of-life questions really measure Insights from Hobday et al and the "Do Something" rule. J Clin Oncol 21:3178-3179, 2003(Bradley J. Monk, Helen Q.)