Characteristics of Advanced Cancer Patients With Cancer-Related Fatigue Enrolled In Clinical Trials and Patients Referred to Outpatient Palliative Care Clinics

Abstract

Objectives

The primary aim of this study was to compare the characteristics of two groups of patients with advanced cancer and moderate to severe fatigue: patients in cancer-related fatigue clinical trials (CCTs) and patients at an outpatient palliative care clinic (OPC).

Methods

We retrospectively reviewed the records of 337 patients who were enrolled in one of five CCTs for advanced cancer patients at The University of Texas M. D. Anderson Cancer Center as well as records of 1896 consecutive patients who were referred to our OPC from January 2003 through December 2010. Patients with fatigue scores of ≥4/10 (measured by the Edmonton Symptom Assessment System [ESAS]) were eligible (1252 OPC patients and 337 CCT patients). Patient characteristics, ESAS scores and survival times were compared using Chi-square tests, Wilcoxon rank sum tests and the Kaplan-Meier method.

Results

Compared with the CCT patients, OPC patients were more likely to be older (58 vs. 59 years; P=0.009) and male (38% vs. 52%; P<0.001). The most common primary cancer type was breast cancer (22%) in the CCT patients and lung cancer (23%) in the OPC patients (P<0.001). The median ESAS scores in the OPC and CCT, respectively, were 6 and 4 for pain (P<0.001), 7 and 7 for fatigue (P=0.525), 3 and 2 for depression (P=0.004), 3 and 2 for anxiety (P<0.001), 3 and 2 for dyspnea (P<0.001), and 43 and 32 for the symptom distress score (P<0.001). The median overall survival times were 17.9 months (95% confidence interval [CI], 13.5–22.3 months) in the CCT group and 3.8 months (95% CI, 3.5–4.1 months) in the OPC group (P<0.001).

Conclusion

Baseline characteristics and overall survival times significantly differed between patients enrolled in the CCT and OPC groups. Therefore, we conclude that the results of CCTs cannot be generalized to patients being treated in OPCs.

Introduction

Cancer-related fatigue (CRF) is one of the most common and chronic symptoms among patients with advanced cancer.¹ Systematic reviews have identified approximately 31 published randomized, controlled, clinical trials for the treatment of CRF using pharmacologic agents; these agents had mixed results in terms of effectiveness.^2,3 The heterogeneity of the treatments tested and mixed results mean that very limited information is available to physicians trying to decide between various clinical treatment options for CRF. The National Comprehensive Cancer Network believes that clinical trials offer the best way to determine the efficacy of treatments for CRF,⁴ so the limitations of established evidence for the treatment of severe CRF justifies the need for more CRF clinical trials (CCTs).

Because the majority of fatigue clinical trials have been conducted in cancer survivors, the results are not applicable to patients with advanced cancer receiving palliative care. Although some CCTs have been conducted in those with advanced cancer, one of the limitations in interpreting the results of these trials is the lack of knowledge of whether the characteristics of patients enrolled in the CCT differ from those of patients in outpatient palliative care clinics (OPCs). If these groups are not the same, the safety and benefits demonstrated in CCTs might not be reproducible in the outpatient setting.

Previous studies of clinical treatment trials for various cancers have found that the characteristics of clinical trial participants are not similar enough to those of the target populations to ensure the generalizability of the results.^5-9 This difference might be the result of barriers or promoters of accrual to the clinical trials, lack of diverse inclusion criteria, cultural factors, and discrepancy between health providers’ attitudes.¹⁰

At The University of Texas M. D. Anderson Cancer Center, we have conducted five CRF pharmacologic treatment trials^11-13 in patients with advanced cancer. The preliminary results of these trials have shown a promising effect of each tested agent on clinically significant CRF. The primary aim of the current study was to determine whether the characteristics and Edmonton Symptom Assessment System (ESAS) scores of enrolled CCT patients were similar to those of patients with advanced cancer seen at our institution’s OPCs. The results of this study were expected to provide preliminary evidence of whether results from CCTs could be generalized to OPC patients. We hypothesized that the characteristics of CCT participants significantly differ from those of OPC patients.

Methods

Patient Selection

A total of 337 patients who participated in five CCTs at our institution and 1252 patients treated at our OPC during the same period were included in the current study. The Institutional Review Board at the M. D. Anderson Cancer Center approved the study protocol and granted waivers of the requirements for informed consent and authorization.

CCT Cohort

All 337 participants in the five CRF clinical trials (three using methylphenidate, one using donepezil, and one using a testosterone replacement)that were performed at M. D. Anderson Cancer Center from January 2003 through December 2010 were included in the current study. Eligibility criteria for inclusion in the CCT group were a diagnosis of advanced cancer (defined as locally recurrent or metastatic cancer), ESAS fatigue score of ≥4 during the previous 24 hours, and a Mini-Mental State Examination (MMSE) score in the normal range (≥24/30). The study design and patient recruitment methods used for those trials were described previously.^10-12 For the current analysis, we accessed the original baseline data from these prospective clinical trials and extracted data related to the participants’ demographic and clinical characteristics and their ESAS and MMSE scores.

OPC Cohort

We reviewed patient records for 1896 consecutive patients with advanced cancer who presented at the M. D. Anderson Cancer Center outpatient Supportive Care Center for an initial consultation. Eligibility criteria for inclusion in the OPC group were a diagnosis of advanced cancer (locally recurrent or metastatic cancer), an ESAS fatigue score of ≥4, and a Memorial Delirium Assessment Scale (MDAS) score of ≤7 (out of 30) or an MMSE score of ≥23. A total of 644 consecutive patients did not meet these eligibility criteria, leaving 1252 patients in the OPC cohort.

Dropouts

We performed a subgroup analysis of the CCT and OPC patients’ characteristics by dividing the two cohorts according to whether they remained active or dropped out of treatment (Fig. 1). We hypothesized that patients who dropped out of clinical trials had higher levels of symptom distress than those who completed clinical trials and, therefore, were more similar to patients at the OPC than to patients who had follow-up visits. Dropout for trial participants was defined as not completing a trial for any reason. Dropout for palliative care outpatients was defined arbitrarily as being lost to follow-up or failing to appear for a scheduled follow-up visit within one month.

Fig. 1.

Study flowchart of data collection.

Assessment Tools

ESAS

We previously designed the ESAS to assess the severity of 10 symptoms commonly experienced by cancer patients during the 24 hours before the assessment: pain, fatigue, nausea, depression, anxiety, drowsiness, dyspnea, anorexia, sleep disturbance, and lack of a feeling of well-being. The severity of each symptom is rated on a numerical scale of 0 to 10 (0 = no symptom, 10 = worst possible severity). The ESAS is both valid and reliable in the assessment of the intensity of symptoms in cancer patients.^14,15 We defined the ESAS symptom distress score (SDS) as the sum of all the items except sleep disturbance and used this score to indicate symptom burden.

MMSE

The MMSE is one of the tools most widely used by health care providers to assess cognitive function.¹⁶ The MMSE is easy to administer and requires less time than other mental status tests or neuropsychological test batteries.¹⁷

MDAS

The MDAS is a structured, 10-item, clinician-rated scale. Each item is scored on a scale of 0 to 3, and the possible total score ranges from 0 to 30; a higher score indicates more severe delirium in medically ill patients. This tool was originally tested in a heterogeneous population of patients with and without cancer. The MDAS has been validated and used for the diagnosis of delirium in cancer patients.¹⁷ For our current study, we defined delirium as an MDAS score of ≥7.

Overall Survival

Overall survival (OS) time also was compared between the CCT and OPC groups. The OS time was defined as the time from the date of trial enrollment to death for CCT patients and from the date of the initial consultation to death for OPC patients.

Death data were obtained from the M. D. Anderson Tumor Registry. In this registry, data about deaths, including dates, are continuously updated as new information becomes available. The database information is reconciled with death certificate information, which is collected from a monthly search of the records of all deaths from the Bureaus of Vital Statistics in Texas and adjacent states. Follow-up of inactive patients is conducted through annual phone calls or letters.

Statistical Analysis

Descriptive statistics (means, medians, frequencies, and percentages) were used to summarize age, sex, race, and primary tumor site in the two groups. A Chi-square test was used to compare categorical variables (sex and primary tumor site) between the two groups. A Wilcoxon rank sum test was used to compare age, ESAS symptom scores, and SDSs in the two groups. The OS time was estimated using the Kaplan-Meier method, and survival curves were compared using the log-rank test.

Statistical significance was declared for P-values of <0.05. All analyses were performed using the SPSS statistical package 15.0 (SPSS Inc., Chicago, IL) or SAS version 9.2 (SAS Institute Inc., Cary, NC).

Results

Patient Demographic and Clinical Characteristics

Patient characteristics are shown in Table 1. In the OPC group (n=1252), the median age was 59 years. Six hundred forty-seven patients were male (52%), and 920 were Caucasian (74%). The most common cancer types were lung cancer (n=285, 23%) and gastrointestinal tract cancer (n=246, 20%).

Table 1.

Baseline Demographic and Clinical Characteristics

Characteristic		OPC Cohort (N=1252) n (%)	CCT Cohort (N=337) n (%)	P-value a
Age, yrs median (SD)		59 (13)	58 (12)	0.009
Sex	Male	647 (52%)	128 (38%)	<0.001
	Female	605 (48%)	209 (62%)
Caucasian race		920 (74%)	236 (70%)	0.206
Primary cancer site				<0.001
	Lung	285 (23%)	55 (16%)
	Gastrointestinal	246 (20%)	41 (12%)
	Genitourinary	136 (11%)	27 (8%)
	Breast	103 (8%)	72 (22%)
	Gynecologic	102 (8%)	21 (6%)
	Head and neck	70 (6%)	21 (6%)
	Hematologic	53 (4%)	23 (7%)
	Other	257 (20%)	77 (23%)

CCT = patients in cancer-related fatigue clinical trials; OPC = patients at an outpatient palliative care clinic.

^aWilcoxon rank sum test for age and Chi-square test for sex, race, and primary cancer site.

In the CCT group (n=337), the median age was 58 years. Two hundred nine patients were male (62%) and 236 were Caucasian (70%). The OPC and CCT patients significantly differed in age (P=0.009), sex (P<0.001), and primary cancer site (P<0.001) but not race (P=0.206).

ESAS Scores and SDS

Table 2 shows the median values (and quartile 1-3 ranges) for the 10 ESAS symptoms and SDS in the OPC and CCT groups. Symptom intensity was significantly lower in CCT patients than in OPC patients for every ESAS item except fatigue (P=0.525) and as measured by SDS. Although the median values were the same for appetite and well-being, mean intensity scores were lower for measures in the CCT group.

Table 2.

Baseline ESAS Symptom Scores

ESAS item	OPC Cohort (N=1252)		CCT Cohort (N=337)		P-value a
	Median (Q1-Q3)	Mean (SD)	Median (Q1-Q3)	Mean (SD)
Pain	6 (3-8)	5.5 (2.9)	4 (1-6)	3.8 (2.8)	<0.001
Fatigue	7 (5-8)	6.8 (1.8)	7 (5-8)	6.9 (1.9)	0.525
Nausea	1 (0-4)	2.4 (2.8)	0 (0-3)	1.6 (2.3)	<0.001
Depression	3 (0-5)	3.3 (3.0)	2 (0-5)	2.7 (2.5)	0.004
Anxiety	3 (1-6)	3.6 (3.0)	2 (0-5)	2.8 (2.7)	<0.001
Drowsiness	5 (3-7)	4.9 (3.1)	4 (1-6)	4 (3.1)	<0.001
Appetite	4 (2-7)	4.2 (3.0)	4 (1-6)	3.7 (3.0)	<0.001
Well-being	5 (3-7)	5.1 (2.7)	5 (3-6)	4.4 (2.5)	0.015
Dyspnea	3 (0-5)	3.3 (3.0)	2 (0-5)	2.6 (2.9)	<0.001
Sleep	5 (2-7)	4.8 (2.9)	4 (2-6)	4.1 (2.7)	<0.001
ESAS SDS	43 (32-54)	43.8 (15.6)	34 (25-45)	36 (14.7)	<0.001

ESAS = Edmonton Symptom Assessment System; Q = quartile; SD = standard deviation; SDS = symptom distress score.

^aWilcoxon rank sum test.

Characteristics of Dropouts

Of the 1252 OPC patients who were included in the current study, 1216 patients (97%) appeared for follow-up visits within one month of the initial visit; the other 36 (3%) dropped out (i.e., were lost to follow-up). Of the 337 CCT patients, 206 (61%) completed a trial; the other 131 (39%) dropped out (Fig. 1).

Table 3 shows that there were no differences in patient characteristics between the CCT patients who completed trials and CCT patients who dropped out except for a higher percentage of Caucasians in the dropout subgroup (76% vs. 61%; P=0.004).

Table 3.

Characteristics of Clinical Trial Evaluable Patients and Clinical Trial Dropouts

Characteristic		Clinical Trial Completers (N=206)	Clinical Trial Dropouts (N=131)	P-valuea
Age, yrs median (SD)		58 (11.87)	59 (11.87)	0.902
Sex, n (%)	Male	81 (39%)	47 (36%)	0.302
	Female	125 (61%)	84 (64%)
Race, n (%)	Caucasian	156 (76%)	80 (61%)	0.004
Primary cancer site, n(%)	Gastrointestinal	21(10%)	20(15%)	0.296
	Lung	34(17%)	21(16%)
	Breast	43(21%)	29(22%)
	Genitourinary	19(9%)	8(6%)
	Gynecologic	11(5%)	10(8%)
	Head and neck	17(8%)	4(3%)
	Hematologic	17(8%)	6(5%)
	Other	44(22%)	33(25%)
ESAS item, median (Q1-Q3)
	Pain	4 (2-6)	4 (1-6)	0.587
	Fatigue	7 (5-8)	7 (6-8)	0.832
	Nausea	0 (0-3)	0 (0-3)	0.831
	Depression	2 (0-5)	3 (0-5)	0.744
	Anxiety	2 (0-5)	2 (0-5)	0.952
	Drowsiness	4 (1-6)	4 (1-6)	0.887
	Appetite	3 (1-5)	4 (1-7)	0.218
	Well-being	5 (2-6)	5 (3-6)	0.576
	Dyspnea	2 (0-5)	3 (0-5)	0.434
	Sleep	4 (2-6)	4 (2-6)	0.295
	SDS	34 (24-45)	36 (25-46)	0.627

ESAS = Edmonton Symptom Assessment System; Q = quartile; SD = standard deviation; SDS = symptom distress score.

^aWilcoxon rank sum test for age and Chi-square test for sex, race, and primary cancer site.

When we compared the 1216 OPC patients with the 131 CCT dropouts to determine if these groups were similar, we found that the OPC group without dropouts had significantly fewer women (48% vs. 64%, P=0.001), non-Caucasians (27% vs. 39%, P=0.003), and breast cancer patients (8% vs. 22%, P<0.001) than the CCT dropout group did. Age did not significantly differ between the two groups (P=0.469). Patients in the CCT dropout subgroup had significantly lower symptom intensity (pain [P<0.001], nausea [P=0.002], depression [P=0.03], anxiety [P=0.003], drowsiness [P=0.001], lack of well-being [P=0.004], sleep disturbance [P=0.001], and SDS [P<0.001]) than OPC patients without dropouts, but the two groups did not significantly differ in appetite (P=0.493) or dyspnea (P=0.065), indicating that these two groups were not similar, as we had hypothesized.

OS Time

The median OS time was significantly longer for the 337 CCT patients (17.9 months; 95% confidence interval [CI], 13.5–22.3 months) than for the 1252 OPC patients (3.8 months; 95% CI, 3.5–4.1 months; P<0.001) (Fig. 2). The difference in median OS time remained significant in the subgroup analysis: OS time was 13.2 months in the CCT dropout subgroup (95% CI, 7.5–18.9 months) but only 4.0 months in the OPC subgroup without dropout (95% CI, 3.7–4.3 months; P<0.001). In addition, the median OS time was statistically only marginally different between the CCT dropout and non-dropout subgroups (13.2 months vs. 21.6 months; P=0.041), in contrast to the statistically strong disparity between the OPC dropout and non-dropout subgroups (1.4months vs. 4.5 months; P<0.001).

Fig. 2.

Overall survival times of patients in the OPC and CCT cohorts.

Discussion

Patients with CRF and advanced cancer who were admitted to one of five CCTs at our institution had lower symptom intensity and longer OS times than did those who were seen at our OPC. The findings persisted in the subgroup analyses. These findings raise the important issue of whether the results of CCTs can be generalized to ambulatory palliative care patients with fatigue. To our knowledge, this is the first study to investigate this issue of generalizability of fatigue clinical trials in patients with advanced cancer.

The difference between the OPC and CCT cohorts was large enough to have considerable clinical relevance. For example, a difference in pain intensity of two points on the ESAS (Table 2) is considered clinically relevant in pain research studies.¹⁸ In addition, the median intensities of depression and anxiety were both 3 in the OPC group (Table 2) and both 2 in the CCT cohort. Since threshold value for anxiety and depression as a clinically relevant finding is ≥2,¹⁹ the frequency of clinical depression in the OPC cohort was likely considerably higher than in the CCT cohort, in which at least half the patients had lower levels of anxiety and depression than is required for the diagnosis of mood disorders.

It is of interest that the median intensity of fatigue was not different between the two groups. However, fatigue is a multidimensional symptom,²⁰ and our findings strongly suggest that pain, depression, and anxiety were much stronger contributors to the intensity of fatigue expression in the OPC group than in the CCT group. This result suggests that fatigue in the OPC cohort might be more likely to improve by successful analgesic or mood interventions than fatigue in the CCT group would. On the other hand, an intervention such as exercise or an anti-inflammatory medication, based on the symptoms of fatigue, could have much more impact in the CCT group than in the OPC group because of the former’s lower levels of contributors such as pain and mood disorders.

More research is necessary to better characterize the impact of different dimensions on the expression of fatigue. Ultimately, the goal is to develop personalized multimodal interventions based on multidimensional profiles. Similar results have been reported for studies that compared cancer patients who received standard cancer treatment with therapeutic trial participants.^21-24 Elting et al.²⁵ performed a large retrospective cohort study to compare patients who received standard cancer treatments (n=19,340) with patients who participated in cancer treatment clinical trials (n=6321) and reported that clinical trial participants had comparatively better prognostic baseline factors and longer survival times than nonparticipants did.²⁵ Similarly, Peppercorn²¹ et al. found in a systematic review that patients in cancer treatment trials had better performance status and prognostic factors than nonparticipants.²¹ The issue of whether results from clinical trials can be applied to the general patient population is not limited to oncology: differences in patient characteristics have been reported for other chronic diseases, such as rheumatoid arthritis²⁶ and depression.²⁷

Our findings suggest that patients with advanced cancer admitted to CCTs are at an earlier time point in the trajectory of their disease than patients referred to OPCs. It is, therefore, possible that patients in the standard clinical setting have more frequent and more severe side effects, more drug interactions, or less response to treatment than those in CCTs.

Our findings also suggest that the results of CCTs need to be confirmed in “real world,” open-label studies of patients treated with these agents in a clinical setting to better understand the agents’ therapeutic efficacy and side effect profiles. Eligibility criteria for CCTs also could be altered to increase participation of patients with other symptoms such as pain, sleep disturbance, nausea, anxiety, depression, and overall symptom burden, in addition to fatigue.

Previous research suggested that industry-funded trials select patients who are healthier and more likely to respond to investigational agents.^28-30 However, the five fatigue clinical trials used in the present study were not funded by industry, and, therefore, it is likely that differences we found between CCT and OPC patients were the result of the strict criteria for eligibility, as well a lack of participation by very ill patients. Our finding of higher overall symptom distress in OPC patients than in CCT patients suggests that the mechanisms of fatigue and likelihood of response differ between the two groups. More research is needed to better characterize and understand this important issue.

The main strength of this study is that it is the first attempt to investigate the generalizabilty of CCTs to OPC using a large sample and validated assessment tools. Limitations are the retrospective nature of the study and the use of the ESAS item for fatigue to assess what is a multidimensional symptom. However, in clinical settings, the ESAS is the most established tool for symptom assessment in palliative care.^31,32 Nevertheless, our findings emphasize the need for a multidimensional tool to assess fatigue in both clinical and research settings.

In conclusion, we found that patients admitted to CCTs had lower intensities of symptom distress and longer survival times than OPC patients. For the results of CCTs to be generalizable, future efforts must minimize the differences between patients with advanced cancer and CRF admitted to CCTs and palliative care patients with fatigue seen in ambulatory clinical settings.