A placebo-controlled double blind trial of etanercept for the cancer anorexia/weight loss syndrome

Results from N00C1 from the North Central Cancer Treatment Group




Tumor necrosis factor-α (TNF-α) is a putative mediator of the cancer anorexia/weight loss syndrome. The current study was designed to determine whether etanercept (a dimeric fusion protein consisting of the extracellular ligand-binding portion of the human 75-kilodalton TNF receptor linked to the Fc portion of human immunoglobulin [Ig] G1) could palliate this syndrome.


A total of 63 evaluable patients were randomly assigned to receive either etanercept at a dose of 25 mg subcutaneously twice weekly versus a comparably administered placebo. All patients had an incurable malignancy, acknowledged loss of weight and/or appetite as a concern, and reported a weight loss of >2.27 kg over 2 months and/or a daily intake of <20 calories/kg body weight.


Over time, weight gain was found to be minimal in both treatment arms; no patient gained ≥10% of their baseline weight. Previously validated appetite questionnaires revealed negligible improvements in both treatment arms. The median survival was also comparable (175 days vs 148 days in etanercept-treated and placebo-exposed patients, respectively; P = .82). Finally, preliminary data regarding adverse events demonstrated that patients treated with etanercept had higher rates of neurotoxicity (29% vs 0%) but lower rates of anemia (0% vs 19%) and thrombocytopenia (0% vs 14%). Infection rates were negligible in both groups. Genotyping for TNF-α-238 and TNF-α-308 polymorphisms revealed no clinical significance for these genotypes, except for a preliminary association between presence of the −238 G/A genotype and relatively less favorable survival.


Etanercept, as prescribed in the current trial, does not appear to palliate the cancer anorexia/weight loss syndrome in patients with advanced disease. Cancer 2007. © 2007 American Cancer Society.

Cancer patients with anorexia and weight loss tend to die sooner than those who maintain their appetite and weight.1 Numerous studies have sought to understand this prognostic effect and, as a result of such efforts, the cytokine tumor necrosis factor-α (TNF-α) has emerged as a key putative mediator of the cancer anorexia/weight loss syndrome.2

Several lines of investigation point to the pivotal role of TNF-α in this syndrome and to the potential for gleaning therapeutic and palliative value as a result of its inhibition. First, Llovera et al. found that when Lewis lung cancer was implanted into TNF-α knockout mice deficient in this cytokine, the rates of protein degradation were lower compared with those observed in wild-type mice with implanted tumors.3 Similarly, the ubiquitin-proteasome system was activated to a lesser degree, an observation that is of significance when one considers that this system appears to be responsible for approximately 80% of muscle wasting in cancer patients.4 These findings, coupled with evidence that links TNF, NF-κ-B, and the essential muscle transcription factor MyoD, provide a clear mechanistic pathway with which to explain how this inflammatory cytokine might give rise to muscle wasting and weight loss.2, 5 Second, Torelli et al. tested a dimeric, pegylated 55-kilodalton (kD) TNF-α receptor construct that acted as a TNF-α inhibitor.6 Tumor-bearing mice that received the receptor construct ate better and gained more weight compared with those that received only vehicle. These compelling preclinical data raise the possibility that these same favorable effects might be observed among anorexic cancer patients who continue to lose weight. Third, TNF-α inhibitors are now clinically available. For example, etanercept, a dimeric fusion protein, binds to TNF-α in vivo and thereby blocks its interaction with cell surface receptors. This agent has been extensively studied in the clinical setting and represents a standard therapy for patients with rheumatoid arthritis.7 Thus, preclinical data and the timely availability of TNF-α inhibitors provided the rationale for and the feasibility of testing TNF-α inhibition in a clinical setting for the treatment of the cancer anorexia/weight loss syndrome.

Therefore, the North Central Cancer Treatment Group (NCCTG) conducted a placebo-controlled study with etanercept in patients with the cancer anorexia/weight loss syndrome. The purpose of the current study was to explore proof-of-concept that TNF-α inhibition provides a means to palliate the cancer anorexia/weight loss syndrome.



All sites within the North Central Cancer Treatment Group were provided the opportunity to enroll patients on this trial. The Institutional Review Boards at each site approved the study protocol. All patients provided informed, written consent prior to enrollment.

Eligibility Criteria

The following patient eligibility criteria were to be met at the time of study enrollment: 1) age ≥18 years; 2) histologic or cytologic proof of cancer other than brain cancer; 3) life expectancy of >3 months based on the judgment of the enrolling physician; 4) an Eastern Cooperative Oncology Group performance score of 2 or better; 5) ability to administer subcutaneous medication twice weekly; 6) alert and mentally competent; 7) history of weight loss of ≥2.27 kg over the preceding 2 months (documentation in the medical record was not necessary) and/or an estimated caloric intake of <20 calories per kilogram of body weight per day; 8) treating oncologist believed that weight gain would be beneficial to the patient; 9) patient perceived weight loss as a problem; and 10) incurable malignancy. Patients were allowed to begin or continue with chemotherapy and/or radiotherapy.

In addition, a patient was considered ineligible in the event of any 1 of the following: 1) currently receiving enteral tube feedings or parenteral nutrition; 2) clinical evidence of ascites; 3) current (within the past month) or planned treatment with adrenal corticosteroids (inhaled or optical steroids were acceptable; the short-term use of dexamethasone around the time of chemotherapy also was acceptable); 4) mechanical obstruction of the alimentary canal, malabsorption, or intractable vomiting; 5) pregnant, nursing, or of childbearing potential and unwilling to employ contraception; 6) poorly controlled congestive heart failure or hypertension; 7) use of etanercept or infliximab within the preceding month; and 8) pacemaker, implantable defibrillator, or internalized metal stent.

Stratification and Randomization

Patients were stratified based on the following parameters: 1) primary malignancy (lung vs gastrointestinal vs other); 2) severity of weight loss within the preceding 2 months (<10 pounds vs ≥10 pounds); 3) planned chemotherapy (yes vs no); 4) patient age (<50 years vs ≥50 years); 5) male versus female; 6) treating oncologist plans to use megestrol acetate or other progestational agent (yes vs no); and 7) prognostic index (good vs bad vs unsure).8 Stratification utilized a minimization algorithm that balanced the marginal distributions.

Subsequently, in a double blind manner, patients were randomly assigned to 1 of the following treatment arms: 1) etanercept at a dose of 25 mg subcutaneously twice weekly for a possible a total of 24 weeks or 2) identical placebo administered subcutaneously twice weekly for a possible total of 24 weeks. This dose of etanercept was chosen because of its proven efficacy in other disease settings such as rheumatoid arthritis.7 All patients were instructed on the administration of a subcutaneous therapy and were evaluated by nursing staff at each site before being allowed to self-administer subcutaneous injections. Patients were encouraged to begin treatment/placebo as soon as possible but were required to start within 2 weeks of registration.

It was strongly suggested that megestrol acetate or other progestational agents not be prescribed during the first month of study enrollment unless this intention had been declared at the time of study enrollment. In the event that they had been prescribed during the first month, this information was to be included in the study record. Other interventions, such as nutritional counseling, were to be recommended as per the discretion of the treating oncologist.

Blood Draw for Genotyping

Prior to receiving any study agent, patients were to have a baseline blood draw for the genotyping of selected functional TNF-α polymorphisms. Buffy coats were prepared and sent on ice directly to the Mayo Clinic in Rochester, Minnesota. Once received, they were directly frozen in −70°C until the time of DNA isolation and genotyping. Concomitant assessment of serum concentrations of TNF-α was not performed because previous studies from our group and others showed this cytokine to be undetectable in serum.9


A history and physical examination that included weighing the patient in the oncologist's office was performed at baseline and then on a monthly basis. Assessment of adverse events occurred during these visits as well. The National Cancer Institute Common Toxicity Criteria (CTC; version 2) were used for this purpose, and patient-reported adverse events were assessed using a patient-completed questionnaire.

Patients were to complete the NCCTG Anorexia/Weight Loss questionnaire at baseline, weekly for 4 weeks, and then monthly for as long as they were receiving the agent/placebo. Other questionnaires were completed at the same frequency, and these consisted of the Uniscale global quality of life scale and the Functional Assessment Anorexia/Cachexia (FAACT) questionnaire (version 4).10, 11

Patients were to continue to administer injections twice weekly as long as the treatment was well tolerated and both the patient and treating oncologist thought it beneficial for up to a total of 24 weeks. These stopping points were implemented in part because of the lack of drug availability beyond 24 weeks and in part because of prior studies from our group that demonstrated very high patient dropout rates early during the treatment period.12, 13

Laboratory Methods

Genomic DNA was extracted from buffy coat preparations using the EASY DNA Kit (Invitrogen, Carlsbad, Calif.) according to the manufacturer's instructions. TNF-α-238 A/G and TNF-α-308 G/A alleles were detected using polymerase chain reaction (PCR) restriction fragment length polymorphism assays.


One hundred nanograms of genomic DNA were amplified in a total volume of 50 μL containing 0.2 μM of each primer, 0.5 units of platinum Taq DNA polymerase (Invitrogen), 200 μM of each dNTP, and PCR reaction buffer (Invitrogen) for 1 cycle at 94°C for 5 minutes followed by 35 cycles of 94°C for 1 minute, 60°C for 1 minute, and 72°C for 1 minute plus 72°C for 5 minutes and then 4°C. The PCR products were digested at 37°C with BamH1 and with NcoI (all from New England Biolabs, Ipswich, Mass.), followed by 4% agarose gel electrophoresis to detect TNF-α-238 A/G and TNF-α-308 G/A alleles, respectively. Each PCR run included a “blank” to which no DNA had been added to ensure that there had been no sample contamination.

Statistical Analyses

The primary objective of the current study was to compare nonfluid weight gain between the 2 study arms. Patients who gained ≥10% of their baseline weight at any point in the study were deemed a success, and a Fisher exact test was used to compare the percentage of successful patients between groups. This 10% increase in weight from baseline was chosen because other successful agents, such as megestrol acetate, have been proven to achieve this degree of weight gain.12, 13 The sample size reported herein provided 81% power to detect a 24% difference in the percentage of patients who gained ≥10% of their baseline weight at any point while on the study with a 2-tailed alternative and a 5% Type I error rate. Similar analyses were also undertaken for lesser degrees of weight gain in an exploratory manner. Patients who manifested edema at any time point after study entry were censored for weight assessment from that point onward. Changes in appetite were assessed in the same manner based on questionnaire data, as were the adverse event data. Kaplan-Meier survival curves were constructed for each patient group, and comparisons were made with the log-rank test. Correlations between changes in weight and genotype were explored. A Cox proportional hazards model was constructed to explore the prognostic effect of variables, including genotype.


Baseline and Descriptive Data

A total of 66 patients were recruited between 2003 and 2005. The study was terminated early because of poor accrual, and less than half the intended accrual goal was reached. Three patients withdrew from the study before receiving any study drug, and thus a total of 63 evaluable patients are the focus of this report.

The patients treated with etanercept (33 patients) and the placebo-exposed patients (30 patients) were comparable at baseline (Table 1). The median age in the former group was 64 years (range, 42–84 years) and was 68 years (range, 29–89 years) in the latter group. Men comprised 52% of the etanercept group and 47% of the placebo group. Otherwise, the groups were similar based on anticipated use of appetite stimulants, anticipated use of chemotherapy, site of primary cancer, and degree of weight loss at baseline.

Table 1. Baseline Characteristics*
CharacteristicNo. of patients receiving etanercept n = 33No. of patients receiving placebo n = 30P
  • *

    Percentages may not always add up to 100% because of rounding.

Median age (range), y64 (42–84)68 (29–89).87
 Male (%)17 (52)14 (47).70
 Female (%)16 (48)16 (53)
Anticipated concurrent chemotherapy
 Yes (%)27 (82)27 (90).35
 No (%)6 (18)3 (10)
Primary tumor site
 Lung (%)11 (33)16 (53).26
 Gastrointestinal (%)9 (27)5 (17)
 Other (%)13 (39)9 (30)
Anticipated megestrol acetate
 Yes (%)5 (15%)3 (10%).54
 No (%)28 (85%)27 (90%)
Weight loss at entry, kg
 <4.5 (%)14 (42%)12 (40%).84
 ≥4.5 (%)19 (58%)18 (60%)


Reasons for withdrawing from therapy in the etanercept and placebo arms were as follows: patient declined further therapy and/or developed adverse events in 32% and 40%, respectively; patient developed disease progression or developed other medical problems in 12% and 17%, respectively; and patient died on study in 13% and 23%, respectively. The remaining patients in the cohorts withdrew from treatment for a variety of other reasons that are less easily categorized.

Of note, the initiation of other potential appetite stimulants was well balanced between the study arms. Four of the etanercept-treated patients and 3 patients who were given placebo were prescribed a progestational agent during the first month of therapy. In addition, 7 patients treated with etanercept and 11 placebo-exposed patients were prescribed corticosteroids.


No patient in either treatment arm achieved the primary endpoint of gaining ≥10% of their baseline weight. If weight gain was categorized by increments of a smaller percentage, 27% of the etanercept-treated patients and 3% of the placebo-exposed patients gained 0–4% of their baseline weight, and 17% and 9%, respectively, gained 5% to 9% of their baseline weight. Nonetheless, there was no statistically significant difference in weight gain noted in 1 group compared with the other (Fig. 1).

Figure 1.

Symmetry of plots that demonstrate the percentage of change in weight from baseline indicates that 1 treatment arm did not perform better than the other in terms of weight gain.

In an exploratory manner, we also reported weight analyses at Weeks 4 and 8. At 4 weeks, the median change in weight from baseline in the patients treated with etanercept (26 patients) was 0.3kg (range, −6.3–6.4 kg), and in the patients who received placebo (25 patients), this figure was −1.4kg (range, −13.1–6.4 kg) (P = .13). At 8 weeks, the median change in weight from baseline in the etanercept-treated group (17 patients) was 0.4 kg (range, −8.0–4.3 kg), and was −0.4 kg in the placebo-exposed group (15 patients) (range, −8.0–6.1 kg) (P = .35). Although dropout rates were high for the reasons discussed earlier, such rates appear roughly equivalent at the 4-week and 8-week timepoints. Again, these exploratory analyses suggest no statistically significant improvement in weight with the administration of etanercept.

Anorexia and Quality of Life

The NCCTG Anorexia/Weight Loss Questionnaire and the FAACT questionnaire found no significant differences in appetite over time in the 2 treatment groups (Table 2). For example, evaluating baseline and 1-month responses to the question “What effect, if any, do you feel the study medications have had on your food intake?” demonstrated that 6 patients treated with etanercept described their appetite as “the same,” and 10 placebo-exposed patients provided a similar response. Four etanercept-treated patients described their appetite as “reduced” at 1 month in contrast to none of the placebo-exposed patients. The only item found to demonstrate a slight trend in improvement centered around nausea and vomiting, and this questionnaire response ironically favored placebo as the palliative agent; 68% of the patients treated with etanercept described no vomiting at 1 month, in contrast to 94% of patients given a placebo (P = .052). Of parenthetical note, the global quality of life scale demonstrated no clinically or statistically significant differences between groups over time except for a suggestion that the etanercept-treated patients reported a stronger sense of mental well-being at 1 month. However, in the context of the other negative observations, this finding was viewed as spurious.

Table 2. Percentage of Patients Reporting a Best Follow-up Response to Select Appetite Questions*
QuestionResponse% Patients treated with etanercept n = 32% Placebo-exposed patients n = 28P
  • *

    Not all percentages add up to 100% because of rounding.

How would you compare your appetite now to what it was before your present illness?Increased912 
The same1823.87
Slightly reduced2718 
Moderately reduced2324 
Markedly reduced2324 
What effect, if any, do you feel the study medications have had on your food intake?I eat less180 
I eat the same2762.11
I eat slightly more2719 
I eat moderately more2319 
I eat considerably more00 
I eat very much more50 
What is your current food intake in comparison to before your present illness?Increased146 
The same929.42
Slightly reduced2318 
Moderately reduced2723 
Markedly reduced2723 
How is your appetite now in comparison to before you started the study medication?Reduced90 
The same3256.31
Increased only slightly2719 
Increased moderately2519 
Increased considerably56 
Increased very much00 


Overall survival was not found to be statistically significantly different between the 2 groups. The median survival among etanercept-treated patients was 175 days and was 148 days in patients who were given a placebo (P = .82) (Fig. 2).

Figure 2.

The median survivals of those patients treated with etanercept (solid line) and placebo-exposed patients (dotted line) were 175 days and 148 days, respectively (P = .82).

TNF Polymorphisms

Genetic polymorphisms of the TNF-α gene at positions −238 and −308 were assessed in 54 patients. The lack of feasibility in shipping a blood sample accounted for the absence of a sample in the other patients. The TNF-α-238 G/A and G/G genotypes were observed in 4 patients and 50 patients, respectively. The TNF-α-308 G/G, A/A, and G/A genotypes were observed in 35 patients, 2 patients, and 17 patients, respectively.

Hardy-Weinberg equilibrium was tested for the observed frequencies of the TNF genotypes described earlier, and none of the marker frequencies achieved a statistically significant difference from what would be expected based on the allelic frequencies observed in our sample.

Clinical correlative studies demonstrated no association between any of these genotypes and patient-reported weight loss at baseline. Moreover, over time, none of these genotypes was found to be predictive of weight loss or weight gain. It is interesting to note that the TNF-α-238 G/A genotype appeared to predict a shorter survival compared with the G/G genotype (85 days vs 203 days, respectively [P = .01]) (Fig. 3). However, the small number of patients in this analysis invites caution when interpreting this observation.

Figure 3.

The tumor necrosis factor-α (TNF-α)-238 G/A genotype appeared to be predictive of a shorter survival (solid line) compared with patients with the TNF-α – G/G genotype (dotted line): 85 days versus 203 days, respectively (P = .01). However, the small number of patients in this analysis invites caution when interpreting this observation.

Adverse Events

Maximum adverse events were for the most part not significantly different between the 2 groups with the exception of anemia, which was milder among patients treated with etanercept, and thrombocytopenia, which was also found to be milder among etanercept-treated patients (Table 3). Another notable finding is that neuromotor adverse events appeared to be slightly more pronounced among patients who received etanercept. Etanercept was not associated with higher infection rates. Patient-reported adverse events, as ascertained by questionnaire completion, did not contribute other information beyond that reported in Table 3 and demonstrated no statistically significant differences between the study arms (data not shown.)

Table 3. Maximum Adverse Events
Adverse event and grade*% Patients treated with etanercept N = 28% Placebo-exposed patients N = 21P
  • *

    Refers to National Cancer Institute Common Toxicity Criteria (version 2.0). If grade is not listed, then no events of this grade occurred.

  • Some patients dropped out prior to an assessment of adverse events, thus explaining the diminishing sample size.

Abdominal pain
Bone pain


To our knowledge, the current study represents 1 of the first clinical trials to test an inhibitor of TNF to treat the cancer anorexia/weight loss syndrome in patients with solid tumor malignancies. This study found that the endpoints of weight, appetite, and survival were not improved with the administration of etanercept, a TNF-α inhibitor. Although the current study was relatively limited in its sample size, it nonetheless suggests that further study of this syndrome with this agent administered at this dose is unlikely to yield positive findings.

Do the results of this study indicate that TNF-α is not a mediator of the cancer anorexia/weight loss syndrome? Despite the negative clinical findings reported herein, it is still plausible that this cytokine is a mediator of this syndrome. First, an extensive literature, as reviewed earlier, points to the role of TNF-α in mediating weight loss and appetite in the setting of cancer. It would be unwise to invoke the results of only 1 clinical trial to negate the compelling findings from a large body of preclinical literature. One possible explanation for the negative findings of the current trial may hinge on the fact that multiple other cytokines work in concert to mediate this syndrome.14 Disruption of 1 component of this interplay may not be sufficient for a net positive clinical effect. Second, along these same lines, the dose of etanercept tested in the current trial deserves some comment. In the setting of the cancer anorexia/weight loss syndrome, it is plausible that a much higher degree of TNF-α inhibition must be instituted to lead to clinical manifestations of efficacy. The fact that we observed that 27% of etanercept-treated and 3% of placebo-exposed patients gained 0% to 4% of their baseline weight might suggest that with a much larger dose and sample size, this study might have detected a statistically significant improvement in weight with this agent. For now, our only conclusion is that the dose of etanercept prescribed in this trial does not appear to yield a clinically notable palliative effect in patients with the cancer anorexia/weight loss syndrome.

Third, it is important to point out that this study closed early because of poor accrual. Perhaps the use of a subcutaneous injection that required self-administration or the inclusion of a placebo arm contributed to this poor accrual. Despite the speculation regarding the reasons for poor accrual, we must acknowledge that a larger sample size might have allowed for the detection of more subtle differences between the study arms to demonstrate some preliminary benefits of TNF blockade in palliating the cancer anorexia/weight loss syndrome. Overall, although this study did not demonstrate major differences in outcomes between the study arms, one could make an argument in favor of better characterizing the role of TNF blockade in patients with the cancer anorexia/weight loss syndrome.

It is interesting to note that the current study provided 2 other observations that are worthy of further investigation. First, etanercept appeared to result in lower rates of anemia and thrombocytopenia. Several other investigators have described the role of this cytokine in bone marrow disorders and, recently, Boula et al. observed that TNF inhibition with a monoclonal antibody improved the hematopoietic supporting capacity of bone marrow stromal cells.15 The effects of etanercept on anemia and thrombocytopenia are plausible in the context of similar reports. Secondly, and along these same lines, this study observed that the TNF-α-238 G/A genotype is associated with a relatively poorer outcome. The less favorable effects of this genotype have been observed in other clinical settings. Pastor et al. reported that this genotype is associated with a higher risk of alcoholic cirrhosis,16 and Achrol et al. observed that this genotype appears to increase the risk of hemorrhage in patients with arteriovenous malformations of the brain.17 Thus, although the TNF-α-238 G/A genotype occurred in only a small number of patients and although its prognostic effect requires confirmation, this preliminary observation appears plausible in the context of earlier studies.