Original article
What is a clinically meaningful change on the Functional Assessment of Cancer Therapy–Lung (FACT-L) Questionnaire?: Results from Eastern Cooperative Oncology Group (ECOG) Study 5592

https://doi.org/10.1016/S0895-4356(01)00477-2Get rights and content

Abstract

To assess the impact of disease and treatment on patients with advanced non-small cell lung cancer (NSCLC), we set out to determine a clinically meaningful change (CMC) on the Lung Cancer Subscale (LCS) and the Trial Outcome Index (TOI) of the Functional Assessment of Cancer Therapy-Lung (FACT-L) questionnaire. We used data from Eastern Cooperative Oncology Group study 5592 (E5592), a randomized trial comparing three chemotherapeutic regimens in 599 advanced NSCLC patients. Patients completed the FACT-L at baseline (pretreatment), 6 weeks, 12 weeks, and 6 months. Comparing across baseline performance status (0 vs. 1), prior weight loss (<5% vs. ⩾5%), and primary disease symptoms (⩽1 vs. >1), LCS and TOI score differences ranged from 2.4 to 3.6 and 6.5 to 9.2, respectively (all Ps < .001). Mean improvement in LCS score from baseline to 12 weeks was 2.4 points in patients who had responded to treatment versus 0.0 points in patients who had progressive disease. Twelve-week LCS change scores for patients progressing early were 3.1 points worse than those of patients progressing later (mean = −1.2 vs.1.9, respectively). Similarly, the average TOI change score from baseline to 12 weeks was −6.1 for patients who had progressive disease versus −0.8 points for patients who had responded to treatment. Twelve-week TOI change scores for patients progressing early (mean = −8.1) were 5.7 points worse than those of patients progressing later (mean = −8.1 vs. −2.4, respectively). Analyses assuming nonrandom missing data resulted in slightly larger differences. Clinically relevant change scores were estimated as two to three points for the LCS and five to seven points for the TOI, setting upper limits for minimal CMCs. These values were comparable to suggested distribution-based criteria of a minimally important difference. These results support use of a two to three point change in the LCS and five to six point change on the TOI of the FACT-L as a CMC, and offer practical direction for inclusion of important patient-based endpoints in lung cancer clinical trials.

Introduction

Prior to 1990, clinical trial reports rarely included formal measurement of quality of life (QoL) [1]. However, over the past decade a number of instruments were developed to measure various aspects of QoL and have been incorporated into cancer clinical trials. General and cancer-specific instruments have been used. General instruments like the SF-36 [2] and the Sickness Impact Profile [3] have rarely been used in lung cancer studies. More use has been made of disease-specific instruments. These instruments can be divided into more general disease-specific questionnaires like the European Organization for the Research and Treatment of Cancer's Core Quality of Life Questionnaire (EORTC QLQ-C30) [4] and the Functional Assessment of Cancer Therapy-Lung (FACT-L) [5], and more exclusively symptom-focused questionnaires like the Lung Cancer Symptom Scale [6], the Rotterdam Symptom Check List [7], the Memorial Symptom Assessment Scale [8] and the Symptom Distress Scale [9]. The general, disease-specific measures assess the health-relevant domains of QoL (i.e., physical, functional, social, and emotional function) in addition to the specific concerns of a disease or its treatment. To accomplish this, both the FACT and EORTC measurement systems add disease-targeted subscales that are designed to tap specific concerns of lung cancer patients, such as disease symptoms, that are not captured by the more generic “core” questionnaire.

Regulatory agencies such as the U.S. Food and Drug Administration have expressed interest in QoL as a primary endpoint in cancer clinical trials [10]. Several cancer drug submissions have included QoL data, and in some circumstance these data have been influential in approval 11, 12. Regulatory agency submissions have not typically focused on formal QoL assessment scores, although formal assessment is often included in the registration trial. Instead, more “clinician-friendly,” symptom-focused indicators of “clinical benefit” have tended to be the focus of the evaluation of therapeutic effectiveness. This is not because clinical benefit approaches have superior reliability and/or validity (in fact, they introduce additional error); rather, they are more resonant to the clinical evaluator. This raises the question of how scores on multi-item questionnaires can be made more clinically interpretable. One way to do this is to investigate the meaningfulness of cross-sectional differences and longitudinal change scores on a commonly used QoL questionnaire by anchoring differences and changes to clinically familiar events.

In oncology, there are several clinically meaningful indicators. These include the performance status rating (PSR), long-accepted as a global rating of patient activity level and highly predictive of survival in advanced lung cancer 13, 14, 15, 16, the response of measurable tumor to treatment, time to progression of disease, and survival. All of these indicators are available in the database of Eastern Cooperative Oncology Group study 5592 (E5592), a randomized phase III clinical trial in advanced (stage IIIb and IV) non-small cell lung cancer (NSCLC). Also available in E5592 are QoL data collected using the Functional Assessment of Cancer Therapy–Lung (FACT-L) Questionnaire. We therefore set out to establish clinically meaningful differences and changes by anchoring scores on the more symptom-focused components of the FACT-L to indicators such as PSR, tumor response, time to progression, and survival time.

Additionally, distribution-based measures offer a way to identify clinically meaningful change that can provide further corroboration when coupled with difference scores tied to clinical anchors. Distribution-based measures rely on the statistical distributions of QoL scores in a given study [17]. These may include reliance on the standard deviation, standard error of measurement, or both [18]. These approaches have been used to determine clinically meaningful differences in QoL scores 18, 19.

Section snippets

Patients and procedure

In E5592, 599 patients with NSCLC were randomly assigned to receive one of three chemotherapeutic regimens: (1) cisplatin 75 mg/m2 IV over 1 hr plus etoposide 100 mg/m2 IV × 3 (days 1, 2, and 3); (2) cisplatin 75 mg/m2 IV over 1 hr plus standard dose paclitaxel (135 mg/m2 IV) over 24 hr, or (3) cisplatin 75 mg/m2 IV over 1 hr plus higher dose paclitaxel (250 mg/m2 IV) over 24 hr with granulocyte colony stimulating factor (g-csf) support (5 μg/kg/day). Eligible patients were newly diagnosed with

Baseline summary statistics

A total of 573 assessments were obtained at baseline in E5592. Of the original total of 599 patients randomized to treatment, 26 were excluded from the final analysis. Many of these exclusions (19%) were due to withdrawn consent. Baseline descriptive and clinical characteristics are displayed in Table 2. The majority of patients were male, Caucasian, and ambulatory, but with some limitation due to disease symptoms (ECOG PSR of 1). The average age of patients was 60.6 years.

Weight loss

We found baseline

Discussion

Our analyses suggest that a two- to three-point difference on the LCS and a five- to seven-point difference on the TOI approximate clinically meaningful changes. These conclusions are drawn from many sources including baseline differences in clinical indicators, changes in LCS and TOI scores over time, comparisons with distribution-based criteria of what constitutes a meaningful change, and sensitivity analyses that estimate QoL parameters under selective assumptions about patterns of missing

Acknowledgements

This work was supported in part by grants from the National Institutes of Health (CA 23318, CA66636, CA21115, CA17145, and CA49957) and AstraZeneca Pharmaceuticals. The contents of this manuscript are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute.

References (30)

  • J.C. de Haes et al.

    Measuring psychological and physical distress in cancer patientsstructure and application of the Rotterdam Symptom Checklist

    Br J Cancer

    (1990)
  • R. McCorkle et al.

    Development of a symptom distress scale

    Cancer Nurs

    (1978)
  • J.R. Johnson et al.

    Food and Drug Administration requirements for approval of new anticancer drugs

    Cancer Treat Rep

    (1985)
  • M.J. Moore et al.

    Use of palliative end points to evaluate the effects of mitoxantrone and low-dose prednisone in patients with hormonally resistant prostate cancer

    J Clin Oncol

    (1994)
  • N. Thatcher et al.

    Improving quality of life in patients with non-small cell lung cancerresearch experience with gemcitabine

    Eur J Cancer

    (1997)
  • Cited by (321)

    View all citing articles on Scopus
    View full text