Fax: (847) 570-8033
Measuring the side effects of taxane therapy in oncology
The Functional Assessment of Cancer Therapy–taxane (FACT-taxane)
Article first published online: 20 JUN 2003
Copyright © 2003 American Cancer Society
Volume 98, Issue 4, pages 822–831, 15 August 2003
How to Cite
Cella, D., Peterman, A., Hudgens, S., Webster, K. and Socinski, M. A. (2003), Measuring the side effects of taxane therapy in oncology. Cancer, 98: 822–831. doi: 10.1002/cncr.11578
- Issue published online: 1 AUG 2003
- Article first published online: 20 JUN 2003
- Manuscript Accepted: 9 MAY 2003
- Manuscript Revised: 21 APR 2003
- Manuscript Received: 5 FEB 2003
- Bristol Myers-Squibb
- Aventis Pharmaceuticals
- quality of life (QOL);
- taxane therapy;
- peripheral neuropathy
Cancer chemotherapy with some of the taxane class of agents can be associated with significant neurotoxicity, arthralgias, myalgias, and skin changes that may offset the therapeutic benefits of taxane use.
The authors developed and tested a set of questions to assess these important side effects of taxane therapy from the patient's perspective. The current study evaluated the taxane subscale of the Functional Assessment of Chronic Illness Therapy (FACIT) measurement system. Reliability, validity, and responsiveness to expected change were evaluated in the context of an ongoing clinical trial comparing four cycles of carboplatin plus paclitaxel with a strategy of carboplatin plus paclitaxel until disease progression in patients with advanced nonsmall cell lung carcinoma (NSCLC).
The 16-item Taxane subscale score and the 11-item peripheral neuropathy subset both demonstrated excellent internal consistency and concurrent validity, and the scores worsened as one would predict during a 12-week treatment course of taxane therapy. Results of the psychometric analyses supported the use of this subscale for measuring the unwanted adverse consequences of effective cancer therapies. Measuring the patient perception of treatment side effects also allowed a preliminary exploration of the relative quality of life (QOL) impact of symptom relief and treatment toxicity. The results indicated that toxicity and symptom improvement may make relatively equivalent contributions to total QOL as measured by the summary score from a multidimensional QOL instrument, the Functional Assessment of Cancer Therapy–General. However, symptom status and improvement appear to play a stronger role than taxane toxicity in patients' global rating of their QOL.
Future research might examine this question of competing benefits as a potential aid to decision-making regarding the administration of toxic therapies in the setting of advanced disease. Cancer 2003;98:822–31. © 2003 American Cancer Society.
Taxanes are a class of chemotherapy agents that promote the polymerization of tubulin into highly stable, intracellular microtubules. These microtubules cause cell death by interfering with normal cell division.1, 2 The first taxane developed and tested in the field of oncology was paclitaxel.3–5 The introduction of paclitaxel to the armamentarium of oncologic therapy in the early 1990s had a great impact on the standard of care in lung carcinoma, breast carcinoma, ovarian carcinoma, and other solid tumors. Subsequent to this development, another taxane, docetaxel,6–8 was introduced. These agents have become a significant component of cancer care in the treatment of both early-stage disease8 and advanced disease.9–12 The incremental advantage of their efficacy relative to previously standard therapy appears to be clear; survival time is increased with the use of taxane therapy in carcinoma of the lung, breast, and ovary. As a result, taxanes have been widely used over the past decade in the management of these solid tumors. For example, the combination of paclitaxel and carboplatin has become the most commonly used regimen in patients with advanced nonsmall cell lung carcinoma (NSCLC).13
Unfortunately, taxane therapy is associated with side effects such as peripheral neuropathy, myelosuppression, arthralgias, myalgias, and skin reactions that may adversely affect patient-reported quality of life (QOL). Peripheral neuropathy in particular can be severe. This toxicity is cumulative across the course of therapy, can be a dose-limiting toxicity, and may lead to dose reduction or cessation of therapy.14 Thus, we continue to be in a treatment era in which agents known to improve clinical endpoints such as response rate, time to treatment failure, and overall survival also are associated with life-altering, and sometimes permanent, concerns stemming from neuropathic drug effects, pain, and disability. If one is to gain a sense of the overall benefit of therapy, the acute and long-term toxicities must be balanced against the efficacy of treatment.
To obtain a better understanding of these side effects, particularly those that affect patient QOL, we attempted to develop and validate an instrument with which to measure the side effects associated with taxane treatments. This information then can be factored into decision-making regarding the value of these chemotherapy agents relative to other options.
MATERIALS AND METHODS
Data were obtained from 230 patients with Joint Commission on Cancer Staging System (JCCSS) Stage IIIB/IV NSCLC who were participating in a randomized Phase III clinical trial comparing limited-duration therapy (four cycles) with continuous therapy. Study procedures and results of this trial have been reported elsewhere.15 Briefly, subjects were randomized to one of two treatment arms: a defined duration (four cycles) of carboplatin and paclitaxel or continuous treatment with the same schedule of carboplatin and paclitaxel until disease progression. Once disease progression occurred, all subjects were eligible to receive second-line weekly paclitaxel treatment. Doses and schedules were reported previously in the study by Socinski et al.15
Patient-reported QOL was assessed at baseline (pretreatment), prior to the third treatment cycle (Week 6), prior to the fifth treatment cycle (Week 12), and at the 26-week postbaseline follow-up. The questionnaire was self-administered at baseline. A central research coordinator conducted follow-up assessments by telephone, usually employing a Computer-Assisted Telephone Interview (CATI) procedure.
Subjects completed the QOL assessments at baseline and at Weeks 6, 12, and 26 unless disease progression occurred or the subject withdrew from the study. A total of 230 patients were enrolled in the clinical trial. The questionnaire completion rate (number completed/number expected) was 97.8% at baseline, 95.1% at Week 6, and 85.2% at Week 12 (Table 1). Because there also were patient deaths and dropouts from the treatment study, 143 of the 230 enrolled patients (62%) completed all of the first 3 scheduled QOL assessments (at baseline, at 6 weeks, and at 12 weeks) and were used for what will be referred to as the “complete case” analyses. The majority of surviving patients were either off the study or off all treatment at the time of the 26-week evaluation. Because the current study is a report of a treatment toxicity scale, only data from the baseline through Week 12 are presented.
|Baseline||6 weeks||12 weeks|
|Reason for discontinuation (illness-related)|
|Withdrew from study||2||5||8|
|Total QOL received||218||194||150|
|% QOL compliance||97.8%||95.1%||85.2%|
|Reason for noncompliance|
|Change Tx plan||—||—||1|
|Unable to reach||—||—||1|
The Functional Assessment of Cancer Therapy–Taxane (FACT-Taxane) is a self-report instrument that was developed to measure the health-related QOL of patients receiving taxane-containing chemotherapy. The FACT-Taxane is comprised of the FACT-General16 (FACT-G) plus a 16-item Taxane subscale.
The FACT-G forms the core of the Functional Assessment of Chronic Illness Therapy (FACIT) measurement system and is comprised of subscales assessing Physical Well-Being (PWB), Social/Family Well-Being (SFWB), Emotional Well-Being (EWB), and Functional Well-Being (FWB). The Taxane subscale combines the previously validated 11-item Neurotoxicity subscale17 and 5 additional questions assessing symptoms related to arthralgia, myalgia, and skin discoloration (Table 2).
|Physical Well-Being (PWB)|
|I have a lack of energy|
|I have nausea|
|Because of my physical condition, I have trouble meeting the needs of my family|
|I have pain|
|I am bothered by side effects of treatment|
|I feel ill|
|I am forced to spend time in bed|
|Social Well-Being (SWB)|
|I feel close to my friends|
|I get emotional support from my family|
|I get support from my friends|
|My family has accepted my illness|
|I am satisfied with family communication about my illness|
|I feel close to my partner (or the person who is my main support)|
|I am satisfied with my sex life|
|Emotional Well-Being (EWB)|
|I feel sad|
|I am satisfied with how I am coping with my illness|
|I am losing hope in the fight against my illness|
|I feel nervous|
|I worry about dying|
|I worry that my condition will get worse|
|Functional Well-Being (FWB)|
|I am able to work (include work at home)|
|My work (include work at home) is fulfilling|
|I am able to enjoy life|
|I have accepted my illness|
|I am sleeping well|
|I am enjoying the things I usually do for fun|
|I am content with the quality of my life right now|
|Taxane subscale (Tax)|
|Neurotoxicity (Ntx) component|
|I have numbness or tingling in my hands|
|I have numbness or tingling in my feet|
|I feel discomfort in my hands|
|I feel discomfort in my feet|
|I have joint pain or muscle cramps|
|I feel weak all over|
|I have trouble hearing|
|I get a ringing or buzzing in my ears|
|I have trouble buttoning buttons|
|I have trouble feeling the shape of small objects when they are in my hand|
|I have trouble walking|
|I feel bloated|
|My hands are swollen|
|My legs or feet are swollen|
|I have pain in my fingertips|
|I am bothered by the way my hands or nails look|
The FACT-G yields a total score, as well as individual subscale scores, with higher scores reflecting better QOL. The Taxane subscale provides an 11-item score from a Neurotoxicity subscale that was codeveloped with the Gynecologic Oncology Group (FACT/GOG-Ntx) and a 16-item Taxane subscale score. The previously validated Lung Cancer Subscale18 (LCS) also was administered at each assessment time. Results from the LCS and the Trial Outcome Index-Lung (TOI-L) are reported herein to facilitate a comparison of the symptom improvement and treatment toxicity experienced by the trial participants. The TOI-L is a 21-item aggregation of the FACT-G PWB, FWB, and LCS scores, and often is used as a single clinical trial outcome index.
All available data were used for the calculation of internal consistency reliability estimates and examination of the validity of known groups at baseline. The longitudinal analyses (e.g., mean change in Taxane subscale scores during chemotherapy) were conducted using data only from complete cases.
The reliability of the subscale aggregations was assessed by calculation of the Cronbach coefficient α, a widely used indicator of internal consistency that ranges from 0.0–1.0. For this specific objective, all available data were used for α calculations. All analyses were performed using SAS software (version 8.20; SAS Institute Inc., Cary, NC).
Using a linear model for the analysis of longitudinal data, we compared change over time on all subscales. A linear test of trend also was used to test the amount of change over time within each specific subscale. Effect size estimates for sensitivity to change from baseline were calculated by subtracting time-dependent means from baseline and dividing them by the standard deviation of the sample score at baseline. Baseline differences in Karnofsky performance status rating were compared using independent sample Student t tests. Effect size estimates for group differences concerning KPS at baseline were calculated by subtracting group-level means and dividing by the pooled within-group standard deviation. For comparisons of mean differences, effect sizes of approximately 0.50 (range, 0.20–0.80) generally are regarded as being moderate in size.19
Average change scores over time were calculated by subtracting mean scores at subsequent assessments from baseline for the disease- and treatment-specific subscales (LCS, Neurotoxicity, and Taxane) and the QOL scales (FACT-G, Global QOL). Correlation coefficients (r) as well as associated effect sizes between the disease- and treatment-specific and QOL subscales were calculated. The effect size was calculated as the correlation coefficient squared divided by one minus the squared correlation coefficient. For correlation coefficients, effect sizes of approximately 0.30 (range, 0.10–0.50) generally are regarded as being moderate in size.19
Clinically meaningful differences with regard to the Neurotoxicity and Taxane subscales were computed using the distribution-based standard error of measurement (SEM) for all available data at baseline. The SEM was calculated using the following formula: . Clinically meaningful differences with regard to the LCS were presented previously by Cella et al.20 Based on the study by Wyrwich et al.,21 patients then were classified based on a change of at least 1.0 SEM on the Neurotoxicity and Taxane subscales as either having “no emerging neurotoxicity” or having “emerging neurotoxicity.” In addition, patients were classified on the LCS as having “symptomatic improvement” or “no symptomatic improvement.” Independent sample Student t tests were used to test the mean changes in overall QOL scores according to these defined changes in symptoms and neurotoxicity over time.
Table 3 summarizes the sociodemographic and disease characteristics of the 230 subjects enrolled in the study and the 143 subjects in the complete case sample. The majority of patients were male, white, and had a diagnosis of Stage IV disease. Only subjects with a KPS > 70 were enrolled. At baseline, patients were roughly evenly split between those with a KPS of 70–80 (47.4%) and those with a KPS of 90–100 (52.6%).
|Enrolled patients (n = 230)||Complete case sample (n = 143)|
|Age (yrs) (mean, SD)||62.1 (10.3)||62.1 (10.3)|
|Male||144 (62.6%)||80 (55.9%)|
|Female||86 (37.4%)||63 (44.1%)|
|White||168 (73.0%)||105 (73.4%)|
|Nonwhite||62 (27.0%)||38 (26.6%)|
|Stage III||31 (13.5%)||19 (13.3%)|
|Stage IV||199 (86.5%)||124 (86.7%)|
|70–80||109 (47.4%)||57 (39.9%)|
|90–100||121 (52.6%)||86 (60.1%)|
|Arm A||114 (49.6%)||73 (51.1%)|
|Arm B||116 (50.4%)||70 (48.9%)|
Table 4 presents the Cronbach α coefficients for each of the QOL subscales using all available data at baseline, at 6weeks, and at 12 weeks. The Cronbach α is an indicator of internal consistency, or the extent to which a set of questions reflects a single issue or problem. As a general rule, coefficients > 0.70 are considered to offer sufficient support for adding questions together into one score. The Cronbach α coefficients for the 11-item Neurotoxicity subscale ranged from 0.82–0.86 and that for the 16-item Taxane subscale ranged from 0.84–0.88. For comparison, α coefficients in this sample for the previously validated LCS and TOI-L18 ranged from .054–0.70 and 0.85–0.88, respectively.
|Baseline (n = 188)a||5 weeks (n = 196)a||11 weeks (n = 171)a|
|Physical Well-Being (7 items)||0.77||0.80||0.75|
|Social/Family Well-Being (7 items)||0.73||0.70||0.77|
|Emotional Well-Being (6 items)||0.80||0.76||0.81|
|Functional Well-Being (7 items)||0.82||0.82||0.83|
|FACT-G (27 items)||0.86||0.87||0.88|
|Lung Cancer subscale (7 items)||0.61||0.64||0.64|
|Trial Outcomes Index-Lung (21 items)||0.85||0.88||0.88|
|Neurotoxicity (11 items)||0.86||0.82||0.82|
|Taxane (16 items)||0.88||0.84||0.85|
Differences between KPS groups at baseline were examined. Because the Taxane subscale is not targeted to detect changes in patient performance status, we would expect no major baseline (prechemotherapy) differences between KPS groups using this scale. However, significant KPS group differences should be observed on the PWB, FWB, and LCS subscales, as well as the aggregated TOI-L, because KPS is based on the functional and physical status of the patient. The findings in Table 5 support these predictions, with the exception that there were no significant differences between groups with regard to the PWB subscale (Table 5).
|Baseline KPS group|
|KPS of 70–80 (n = 109)a||KPS of 90–100 (n = 121)a||Total (n = 230)||P value||ESb|
|Scale||Mean (SD)||Mean (SD)||Mean (SD)|
|Physical Well-Being||19.9 (5.2)||20.5 (5.8)||20.2 (5.5)||0.418||0.12|
|Social/Family Well-Being||22.8 (4.9)||22.7 (5.0)||22.7 (4.9)||0.904||0.02|
|Emotional Well-Being||16.2 (5.4)||16.7 (5.0)||16.4 (5.2)||0.541||0.09|
|Functional Well-Being||14.8 (6.2)||17.2 (6.1)||16.1 (6.2)||0.007||0.39|
|FACT-G total||73.8 (14.3)||77.1 (15.4)||75.7 (15.0)||0.135||0.22|
|FACT-Lung and Total|
|Lung Cancer subscale||18.1 (4.9)||19.6 (4.5)||18.9 (4.7)||0.033||0.31|
|Trial Outcome Index-Lung||52.9 (12.5)||57.3 (13.0)||55.4 (13.0)||0.019||0.34|
|FACT-Lung||91.9 (16.4)||96.6 (18.0)||94.6 (17.4)||0.064||0.27|
|FACT-Taxane and Totalc|
|Neurotoxicity||36.8 (7.1)||35.1 (9.1)||35.8 (8.3)||0.159||0.20|
|Taxane||55.0 (8.7)||53.4 (11.2)||54.1 (10.2)||0.270||0.16|
|FACT-Taxane||128.7 (19.8)||130.5 (22.3)||129.7 (21.2)||0.577||0.08|
Responsiveness to expected change
A key feature of the current study is that all trial participants were chemotherapy-naïve at the time of the baseline assessment, and then received a median of four courses of carboplatin/paclitaxel. Thus, all subjects had had significant taxane exposure by the 6-week and 12-week QOL assessments. Analyses examining predictable deterioration over time with regard to the Taxane and Neurotoxicity subscales were therefore performed on the pooled sample. Because subjects received 4 courses of carboplatin plus paclitaxel, average Taxane subscales scores were expected to decline (get worse) over the 12-week treatment period. Cumulative toxicity should be reflected in a score decrease from baseline to 6 weeks and a larger decrease from baseline to 12 weeks. Table 6 presents the average scores for all QOL scales at baseline, at 6 weeks, and at 12 weeks and Figures 1–2 depict the changes graphically. Significant overall linear time effects were observed (P value < 0.05) for all scales, indicating responsiveness to expected change over time. It is noteworthy that the largest effect sizes were noted for the Neurotoxicity and Taxane subscales; the magnitude of the change (decline [d]) over time exceeded the standards set by Cohen19 for the magnitude of effect size: d = 0.37–0.40 (small to moderate effect) at 6 weeks and d = 0.91–1.01 (large effect) at 12 weeks for the Taxane subscale. It also is interesting to note the significant improvement in lung carcinoma symptoms and EWB across the treatment period, whereas all other (sub)scales were found to demonstrate a significant decrease (Table 6).
|Baseline, Mean (SD)||6 weeks,||12 weeks,||Test for linear trend,||Summary|
|Mean (SD)||[ES]a||Mean (SD)||[ES]a||F||P value|
|Physical Well-Being (7 items)||21.7 (5.4)||20.1 (5.3)||[0.29]||18.3 (5.1)||[0.64]||47.78||0.000||Declined moderately|
|Social/Family Well-Being (7 items)||23.0 (5.0)||22.4 (5.0)||[0.12]||21.9 (4.7)||[0.23]||8.99||0.003||Declined slightly|
|Emotional Well-Being (6 items)||16.2 (5.2)||18.4 (4.4)||[0.43]||18.0 (4.6)||[0.34]||15.28||0.000||Improved slightly|
|Functional Well-Being (7 items)||17.7 (6.0)||16.4 (5.8)||[0.22]||14.9 (6.1)||[0.46]||32.84||0.000||Declined slightly|
|FACT-G total (27 items)||78.6 (14.9)||77.3 (14.9)||[0.09]||73.0 (15.0)||[0.38]||22.12||0.000||Declined slightly|
|Lung Cancer subscale (7 items)||19.0 (4.6)||20.3 (4.5)||[0.30]||20.1 (4.4)||[0.25]||9.20||0.003||Improved slightly|
|Trial Outcome Index-Lung (21 items)||58.3 (12.8)||56.8 (13.0)||[0.12]||53.3 (12.9)||[0.39]||23.76||0.000||Declined slightly|
|Neurotoxicity (11 items)||38.0 (7.0)||35.2 (7.4)||[0.40]||31.0 (8.8)||[1.01]||80.45||0.000||Declined substantially|
|Taxane (16 items)||56.6 (9.2)||53.2 (9.1)||[0.37]||48.3 (10.6)||[0.91]||71.38||0.000||Declined substantially|
Relation between Symptom Change, Treatment Toxicity, and QOL
For the next set of analyses, we defined overall QOL in two ways: “Total” QOL (defined as the FACT-G Total score) and “Global” QOL (defined as the response to the single global question on the FACT-G: “I am content with the quality of my life right now.”). We examined each overall QOL score in two ways. First, we compared the magnitude of the correlations of the two QOL change scores with the Neurotoxicity and Taxane toxicity subscale scores versus those of the QOL change scores with the other FACT-L subscale change scores. Our purpose was to assess the relative impact that disease symptoms and functioning have on QOL appraisal compared with that of treatment toxicity. These results can be found in Table 7. At the 6-week assessment, there was no noteworthy difference between the correlations of toxicity with QOL versus disease symptom and functioning with QOL. At 12 weeks, similar results were found for the correlations with the FACT-G, but not for global QOL. Rather, the LCS remained significantly correlated (r = 0.33; P < 0.05) with global QOL but the Taxane toxicity subscale was not found to be significantly correlated with global QOL (r = 0.06; P value not significant [NS]).
|Assessment time and scale||Change score,||Correlation with QOL|
|Mean (SD)||Total FACT-G, r (ES)a||Global QOL, r (ES)a|
|Baseline to 6 weeks|
|FACT-G||−1.3 (13.0)||1.0||0.44bc (0.24)|
|Global QOL||−0.2 (1.3)||0.44bc (0.24)||1.0|
|Lung Cancer subscale||1.4 (4.9)||0.37b (0.16)||0.21b (0.04)|
|Neurotoxicity subscale||−2.8 (7.6)||0.40b (0.19)||0.15 (0.02)|
|Taxane subscale||−3.4 (9.8)||0.40b (0.19)||0.17b (0.03)|
|Baseline to 12 weeks|
|FACT-G||−5.6 (14.3)||1.0||0.47bc (0.28)|
|Global QOL||−0.3 (1.4)||0.47bc (0.28)||1.0|
|Lung Cancer subscale||1.2 (4.6)||0.40b (0.19)||0.33b (0.12)|
|Neurotoxicity subscale||−7.0 (9.4)||0.30b (0.10)||0.06 (0.00)|
|Taxane subscale||−8.3 (11.8)||0.32b (0.11)||0.06 (0.00)|
Second, we divided the sample into those patients who reported symptomatic benefit at either 6 weeks or 12 weeks versus those who did not report symptomatic benefit. We also divided the sample into those who reported significant emerging neurotoxicity at 6 weeks or 12 weeks versus those who did not. Significant symptomatic benefit in lung carcinoma symptoms was defined as an improvement of at least 3 points in the LCS score from the time of baseline to follow-up.20 Significant emerging toxicity was defined as a decrease of at least 1 SEM calculated from the baseline data.21 Because 1 SEM on the Neurotoxicity and Taxane subscales were calculated to be 2.8 points and 3.2 points, respectively, we also considered a 3-point decline to be significant. These clinically distinct groups were compared with one another to appraise the extent to which improved cancer symptoms versus worsened treatment side effects contributed to differences in patients' perceptions of their QOL.
At both 6 weeks and 12 weeks, patients who reported a significant improvement in lung carcinoma symptoms also reported an improvement in their total QOL as measured by the FACT-G; those with an improvement in symptoms had improved QOL, whereas those without improvement had decreased QOL scores (Table 8). The difference between the 2 groups with regard to global QOL was not statistically significant at 6 weeks, but was found to be significant at 12 weeks (Table 8).
|QOL scale||Lung carcinoma subscale improved?|
|Baseline to 6 weeks||Baseline to 12 weeks|
|No (n = 89) Mean (SD)||Yes (n = 54) Mean (SD)||P value||No (n = 86) Mean (SD)||Yes (n = 57) Mean (SD)||P value|
|FACT-G change||−3.7 (13.6)||2.6 (10.9)||0.004||−9.1 (13.5)||−0.4 (13.8)||0.000|
|Global QOL change||−0.3 (1.4)||0.0 (1.2)||0.234||−0.6 (1.3)||0.2 (1.5)||0.001|
The results comparing those with emerging toxicity versus those with no emerging toxicity are similar for the Neurotoxicity and the Taxane subscales (Table 9). FACT-G change scores from baseline to 6 weeks and from baseline to 12 weeks differed significantly between the groups with and without neurotoxicity. In fact, the small sample of patients who reported no emerging toxicity had relatively large improvements in their QOL. No significant association was observed with the global QOL question, suggesting that treatment toxicity may not play a large role in the global evaluation of QOL.
|QOL scale||Emerging Ntx|
|Baseline to 6 weeks||Baseline to 12 weeks|
|No (n = 19)||Yes (n = 124)||P value||No (n = 11)||Yes (n = 132)||P value|
|FACT-G change||9.7 (9.7)||−3.0 (12.6)||< 0.0001||3.1 (9.5)||−6.3 (14.4)||0.035|
|Global QOL change||0.2 (1.4)||−0.3 (1.3)||0.145||−0.3 (1.3)||−0.3 (1.4)||1.000|
|Baseline to 6 weeks||Baseline to 12 weeks|
|No (n = 19)||Yes (n = 124)||P value||No (n = 11)||Yes (n = 132)||P value|
|FACT-G change||9.6 (10.5)||−3.0 (12.6)||< 0.0001||5.5 (9.4)||−6.5 (14.2)||0.007|
|Global QOL change||0.1 (1.6)||−0.3 (1.3)||0.275||−0.3 (1.3)||−0.3 (1.4)||1.000|
The primary purpose of the current study was to evaluate the performance of the FACT-Taxane scale in a longitudinal clinical trial of carboplatin/paclitaxel therapy in patients with advanced NSCLC. The scale demonstrated excellent internal consistency reliability at each of the three assessment times, indicating that a total score can be aggregated without significant loss of information. In addition, known group validity and responsiveness to change over time were demonstrated by capitalizing on two key features of the clinical trial from which these data were obtained. That is, trial participants were chemotherapy-naïve at the initiation of the trial and then received four cycles of a taxane-containing chemotherapy regimen. Scores on the Neurotoxicity and Taxane subscales did not appear to differ between the KPS groups at baseline, as would be expected prior to the receipt of chemotherapy. However, significant score decreases reflected the emergence of measurable neurotoxicity, observed by the 6-week assessment. Larger decreases were reported by the 12-week assessment, indicating the cumulative treatment toxicity observed after 4 cycles of this chemotherapy. Responsiveness to this expected change over the course of chemotherapy is a key feature of a scale purporting to measure treatment side effects. The magnitude of the effect over this time frame suggests the impact is considerable to the individual patient.
During chemotherapy for advanced cancer, the possibility of discontinuation requires continuous clinical judgment weighing benefit against harm. These judgments typically weigh symptom reduction, tumor response, and survival against the decrements in QOL brought about by tumor progression, fitness of the patient, and treatment toxicity. Although this is never an easy decision, it may be particularly difficult when the toxicities are not life-threatening but can significantly impair a patient's QOL (e.g., peripheral neuropathy). The Taxane subscale allows an assessment of the patient's perception of the impact of QOL-relevant treatment side effects. This may provide additional information that is useful for maintaining the delicate balance between the benefits and drawbacks of treatment when cure is not possible.
Using the patient-reported data concerning lung carcinoma symptoms and treatment toxicity in the current trial, we conducted a preliminary evaluation of the relative impact of symptom relief and treatment toxicity on overall QOL. The results indicate that toxicity and symptom improvement may make relatively equivalent contributions to total QOL, as measured by the FACT-G Total score. However, with regard to the patients' global rating of QOL, symptom status and improvement appear to play a stronger role than taxane toxicity. There are important differences between the summation of a set of questions meant to reflect multidimensional QOL and a single global rating of QOL. If one is concerned with the global evaluation a patient makes regarding life quality, these data suggest disease control plays a greater role than treatment toxicity. This has important implications for the value one places on each of these treatment effects when evaluating the efficacy of new therapies in oncology. However, the results should be interpreted cautiously because of the small number of subjects who did not experience toxicity by the 6-week assessment. Further study of these trade-offs between treatment benefit and treatment toxicity could address the direct evaluation of patient preferences22 to supplement assessments of health states such as the one tested herein. It would be particularly relevant to obtain preferences for anticancer therapy that juxtapose clinical benefits with drug toxicity. The preliminary data from the current study suggest that whereas advancing toxicity is measurable from the patient perspective, its effect on overall QOL may not be comparable to numerically equivalent changes in disease symptoms.
The generalizability of the results should also be considered in the context of the strategy chosen for the longitudinal data analysis. Missing data is a common problem when QOL is assessed in clinical trials and the optimal choice of analytic strategy is based on the reasons for the missing data (e.g., whether the data is missing at random) and the specific study aims.23, 24 For this evaluation of the psychometric properties and potential utility of the scale, only complete case data were used.
This is typically a less desirable alternative to more advanced modeling strategies for data that are not missing at random, as is the case with the current trial data. However, we did not intend to draw inferences regarding the severity of toxicity over time on therapy, but rather to evaluate the scale's validity and potential utility as an indicator of toxicity experienced by subjects receiving active therapy. In that regard, the > 85% compliance rate of expected evaluations supports the use of a complete case analysis. If the missing data are disproportionately from patients who are in poorer health, including those with more taxane toxicity, the omission of their data would tend to constrict or compress the variability and amplitude of the results reported herein. This likely produces a conservative estimate of the strength of the correlation between toxicity, general health status, and self-reported treatment side effects.
The Taxane subscale of the FACIT measurement system possesses strong psychometric properties, including excellent internal consistency reliability, known group validity, and responsiveness to change over time. It also appears to provide an important adjunct to currently available scales that typically measure general QOL and disease-specific symptoms. Given the widely recognized significance of patient perceptions in the evaluation of treatment impact, this type of scale fills a critical need. Measuring both disease symptoms and treatment toxicity likely has significant utility for maintaining a balance between the benefits and costs of treatment. This may be particularly true in the setting of advanced disease, in which cure is not possible and maximizing QOL is an important treatment goal.
- 8Dose-dense doxorubicin, docetaxel, and granulocyte colony-stimulating factor support with or without tamoxifen as preoperative therapy in patients with operable carcinoma of the breast: a randomized controlled open phase IIb study. J Clin Oncol. 2001; 19: 3506–3515., , , et al.
- 17Validity and sensitivity of the Functional Assessment of Cancer Therapy/Gynecologic Oncology Group–Neurotoxicity (FACT/GOG-Ntx). Proc Am Soc Clin Oncol. 2000; 19: 446a., , , et al.
- 19Statistical power analysis for the behavioral sciences, 2nd edition. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 1988..
- 22Measuring health state preferences and utilities: rating scale, time trade off and standard gamble techniques. In: SpilkerB, editor. Quality of life and pharmacoeconomics in clinical trials, 2nd edition. Philadelphia: Lippincott-Raven, 1996: 253–265., .