点击显示 收起
1 From the Royal Infirmary of Edinburgh, Edinburgh, United Kingdom (KCF), and the Ross Products Division, Abbott Laboratories, Columbus, OH (ACV and DSH)
2 Supported in part by Abbott Laboratories. 3 Reprints not available. Address correspondence to KC Fearon, Department of Clinical and Surgical Sciences (Surgery), University of Edinburgh, The Royal Infirmary, 51 Little France Crescent, Edinburgh, EH 16 4SA United Kingdom. E-mail: k.fearon{at}ed.ac.uk.
ABSTRACT
Background: Cancer cachexia is a multifactorial syndrome that is poorly defined.
Objective: Our objective was to evaluate whether a 3-factor profile incorporating weight loss (10%), low food intake (1500 kcal/d), and systemic inflammation (C-reactive protein 10 mg/L) might relate better to the adverse functional aspects of cachexia and to a patient's overall prognosis than will weight loss alone.
Design: One hundred seventy weight-losing (5%) patients with advanced pancreatic cancer were screened for nutritional status, functional status, performance score, health status, and quality of life. Patients were followed for a minimum of 6 mo, and survival was noted. Patients were characterized by using the individual factors, 2 factors, or all 3 factors.
Results: Weight loss alone did not define a population that differed in functional aspects of self-reported quality of life or health status and differed only in objective factors of physical function. The 3-factor profile identified both reduced subjective and objective function. In the overall population, the 3 factors, 2 factors, and individual profile factors (except weight loss) all carried adverse prognostic significance (P < 0.01). Subgroup analysis showed that the 3-factor profile carried adverse prognostic significance in localized (hazard ratio: 4.9; P < 0.001) but not in metastatic disease.
Conclusions: Weight loss alone does not identify the full effect of cachexia on physical function and is not a prognostic variable. The 3-factor profile (weight loss, reduced food intake, and systemic inflammation) identifies patients with both adverse function and prognosis. Shortened survival applies particularly to cachectic patients with localized disease, thereby reinforcing the need for early intervention.
Key Words: Systemic inflammation • food intake • cachexia • prognosis
INTRODUCTION
Cachexia is a clinical syndrome that is difficult to define (1). Patients with advanced cachexia are characterized by anorexia, early satiety, severe weight loss, weakness, anemia, and edema (2). In early forms of cachexia, these features occur to a variable extent and may change in severity during the course of a patient's illness. The complex, multifactorial origin of cachexia precludes a uniform pathophysiologic profile. These issues have hindered clinical studies both at a mechanistic level and for targeting therapeutic intervention.
In relation to the approval of novel therapeutics for cachexia, regulatory authorities suggest it is important not only to show efficacy for improved nutritional status such as lean body mass (LBM) but also functional status such as performance status. Ongoing weight loss has been the main criterion used to enter patients into either mechanistic studies or therapeutic trials. However, it is not clear to what extent weight loss alone is associated with adverse functional status. Poor physical function in cachexia may relate to many factors, including loss of body mass, reduced substrate supply (food intake), or reduced volitional effort (fatigue or depression); all of which have been related, at least in part, to the effects of systemic inflammation (3, 4). The purpose of the present study was to evaluate in a homogeneous cohort of patients with cancer the role of weight loss, low food intake, and the presence of systemic inflammation in a multiple-factor profile of cachexia which aimed to reflect patients' adverse function and survival duration. The potential influence of these cachexia-related factors on function and prognosis in patients with different stages of disease was also evaluated.
SUBJECTS AND METHODS
Subjects
The patient population was originally recruited to a multicenter randomized controlled trial (n = 200) of 2 different oral nutritional supplements and was reported previously (5). Median survival from study enrollment for all patients was 130 d, and no significant difference was seen between the treatment groups (experimental: 142 d; control: 128 d). Thus, for the purposes of survival analysis the treatment of patients during the follow-up period can be considered uniform. Patients with unresectable pancreatic cancer were selected specifically because these patients usually experience severe progressive weight loss. Patients were included if they had lost 5% of their preillness stable weight during the previous 6 mo, had a Karnofsky Performance Score (KPS) 60, and had a life expectancy > 2 mo. Patients were excluded if they had undergone surgery, endoscopic stenting, radiotherapy, or chemotherapy during the previous 4 wk; had other active medical conditions (major gastrointestinal disease, chronic renal failure, uncontrolled diabetes, and HIV); a body mass index (in kg/m2) > 30; or received medication that could profoundly modulate metabolism or weight. Patients were allowed to receive any cancer therapy felt relevant during the period of follow-up (eg, palliative radiotherapy for bone metastasis), but in general they did not receive systemic therapy. The ethics committees of the participating centers approved the protocol, and written informed consent was obtained from all patients. Procedures followed were in accordance with the International Committee for Harmonization, Good Clinical Practice, and the Helsinki Declaration.
To be used for the present analysis (n = 170), a subject was required to have data for the 3 factors used in the profile of cachexia. These factors included baseline percentage of weight loss from usual preillness weight, baseline average daily caloric intake, and baseline C-reactive protein (CRP).
Stage of disease
Fifty-eight percent of the patients had histologic proof of diagnosis, and the remainder were known to have pancreatic cancer on the basis of unequivocal clinical (operative) or radiologic [computed tomography (CT) scan] findings. All patients were staged by CT scan as having localized (stage II), locally advanced (stage III), or metastatic (stage IV) disease.
Weight and body composition
Height, preillness stable weight, and duration of weight loss were self-reported. Patients were weighed without shoes and wearing light clothing on spring balance scales. Body composition was assessed by using a Xitron Hydra multiple frequency bioelectrical impedance analyzer (Xitron Technologies, San Diego, CA), as described previously (6).
Grip strength
Patients' grip strength was assessed by using a hand-held spring-loaded dynamometer.
Performance status
Performance status was assessed by the attending clinician by using KPS (7). Patients were considered to have a level of performance status that would affect significantly on their activity if their KPS score was <80.
Dietary intake
Three-day diet diaries completed before assessments at baseline were used to assess the patients' dietary intakes (8, 9). Patients were considered to have reduced food intake that would affect significantly on their energy balance if their energy intake was 1500 kcal/d.
Quality-of-life assessment
Quality of life was measured at baseline, by using 2 self-administered questionnaires: 1) EuroQol EQ-5D, a generic quality-of-life measure that provides a single index score (EQ-5Dindex) and the respondent's assessment of his or her overall heath state [EQ-5DVAS (visual analogue scale)] (10) and 2) the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 (11). For the purposes of the present study only the fatigue, dyspnea, and physical function components of the EORTC QLQ-C30 measure were included because these are most likely to reflect the patients' overall function and might also relate to their nutritional status (eg, LBM).
Systemic inflammation
Systemic inflammation was assessed by assaying serum CRP concentration with the use of an immunoturbidimetric assay (Abbott TDX, Abbott Laboratories, Maidenhead, United Kingdom). Systemic inflammation was present if CRP was 10 mg/L.
CA19-9 measurement
The cancer-specific antigen CA19-9 was assayed in serum by immunoassay (Bayer HealthCare, Tarrytown, NY).
Statistical analysis
The rationale for the development of the multifactor profiles was as follows: all patients enrolled in the study had lost >5% of their preillness stable weight and were therefore considered to have some degree of cachexia. Conventional definitions of clinically significant weight loss usually use >10% as a cutoff (12). Recent data have suggested that in weight-losing patients with advanced pancreatic cancer resting energy expenditure is 1400 kcal/d and total energy expenditure is 1700 kcal/d (13). An intake of <1500 kcal/d was therefore deemed to be at a level that would seriously compromise energy balance and would likely result in reduced physical activity along with continued weight loss. Finally, a CRP > 10 mg/L was used repeatedly to define the presence or absence of systemic inflammation and was shown to be a robust marker of shortened survival in patients with advanced pancreatic cancer (14). The following multifactor profile of cachexia was therefore constructed: weight loss 10% from preillness usual weight, plus food intake 1500 kcal/d, plus a CRP concentration >10 mg/L.
Analyses were done to compare categories for the 3-factor profile, 2-factor profile, and degree of weight loss alone. The variables of age, total caloric intake, and the EQ-5DVAS were analyzed with 2-sample t test, all remaining variables were analyzed with the Wilcoxon's rank-sum test. A 3-factor analysis of variance model was used to examine the effects of the factors that make up the 3-factor cachexia profile on outcome variables related to body composition and functional status. In general, no significant interactions were detected, but the power of this analysis was limited by the small number of patients in each of the 8 combinations. There are 8 possible combinations among the levels of the 3 factors comprising the cachexia profile. In general, the combination corresponding to the cachexia definition had an arithmetic mean that was the worst case for the outcome variable analyzed.
Survival analysis
Survival time in days from the starting point of study baseline was examined. There were 15 censored observations as follows: 1 subject with an unknown date of death, 1 subject lost to follow-up, and 13 subjects still alive at 6 mo after the study. For each cachexia profile, a Cox proportional hazards regression model was fit to all variables of interest and then a stepwise model was fit. In the stepwise model, significance levels for the explanatory variables were 0.20 to enter the model, and 0.10 to remain in the model. To be used in the survival analysis, a subject needed data for all variables being considered for the model (n = 148). In addition, both the model with all variables and the stepwise model were fit separately to the subset of subjects that were classified as having localized disease (stage II or III) or having metastatic disease (stage IV).
RESULTS
The characteristics of the patients included in the analysis of cachexia severity (n = 170) and in the survival analysis (n = 148) are shown in Table 1. Patients had advanced disease with 46% showing evidence of distant metastases (stage IV). The values for performance status, health status, and functional aspects of quality of life reflect a debilitated older group of patients with significant impairment of physical function and global health status.
View this table:
TABLE 1. . Characteristics of patients in the analysis of cachexia severity and in the survival analysis1
Differences in patients' nutritional and functional characteristics based on a profile of cachexia severity by using weight loss alone are shown in Table 2. Body composition, functional aspects of quality of life, and inflammatory status were not significantly different in those with weight loss 10% than in those with weight loss < 10%. KPS and grip strength were significantly lower in those with weight loss 10% than in those with weight loss < 10%.
View this table:
TABLE 2. . Nutritional and functional characteristics of the patients (n = 170) according to weight loss (10%)1
Differences in the patients' nutritional and functional characteristics on the basis of the 3-factor cachexia profile (weight loss, food intake, and inflammatory status) are shown in Table 3. With weight loss 10%, food intake 1500 kcal/d, and CRP 10 mg/L, 22% (37 of 170) of patients met the cachexia profile definition. All quality-of-life function variables, health status, KPS, grip strength, and LBM were significantly reduced.
View this table:
TABLE 3. . Nutritional and functional characteristics of the patients (n = 170) according to the multifactor profile of cachexia (weight loss 10%, food intake 1500 kcal/d, C-reactive protein 10 mg/L)1
Differences in the patients' nutritional and functional characteristics on the basis of having met 2 of the 3 factors in the 3-factor cachexia profile are shown in Table 4. The profile definition was met by 60% (102 of 170) of patients. All quality-of-life function variables were reduced as was grip strength, KPS, and health status but not LBM.
View this table:
TABLE 4. . Nutritional and functional characteristics of the patients (n = 170) according to whether or not they met 2 of 3 factors of the multifactor profile of cachexia (weight loss 10%, food intake 1500 kcal/d, C-reactive protein 10 mg/L)1
To test the prognostic significance of the 3-factor cachexia profile (and its component subparts), a Cox proportional hazards regression model was constructed by using known prognostic variables (stage of disease, log CA19-9, KPS, and age), the cachexia profile (met all 3 factors present, met 2 of 3 factors, or met individual factors of the profile), and the other variables used as markers of patients' physical function and quality of life (EORTC QLQ-C30 scores for physical function, dyspnea, and fatigue; health status; LBM; grip strength). When variables were entered into the stepwise model for the 3-factor cachexia profile model (Table 5), log CA19-9, LBM, health status, and the 3-factor profile itself were prognostic. When the cachexia profile model in which 2 of the 3 factors were met was used, the results were similar in that log CA19-9, LBM, health status, and the variable met 2 of 3 factors in the cachexia profile itself all carried prognostic value. When the individual factors in the 3-factor cachexia profile were included, log CA19-9, KPS, LBM, health status, food intake, and CRP (but not weight loss) carried prognostic value.
View this table:
TABLE 5. . Survival analysis (n = 148) with the stepwise Cox proportional hazards model1
To control as closely as possible for the influence of stage of disease, stepwise models were constructed by using only those patients with localized disease (stages II and III, n = 80) or metastatic disease (stage IV, n = 68). When the 3-factor cachexia profile was included in the model for patients with localized disease, log CA19-9 [hazard ratio (HR): 1.35; P = 0.019] and the profile itself (HR: 4.94; P < 0.001) were prognostic. When the cachexia profile model in which 2 of the 3 factors were met was use included, log CA19-9 (HR: 1.31; P = 0.026) and the profile itself (HR: 2.40; P < 0.001) were prognostic. When the individual factors were used, log CA19-9 (HR: 1.43; P = 0.005), food intake factor (HR: 2.00; P = 0.01), and CRP (HR: 3.74; P < 0.001) carried prognostic significance. For patients with metastatic disease, log CA19-9 (HR: 1.47; P = 0.007) and dyspnea (HR: 1.01; P = 0.025) were prognostic when either the 3-factor cachexia profile or the met 2 of 3 factors in the profile were included in the model. When the individual factors were used in the model, only log CA19-9 (HR: 1.43; P = 0.01) and CRP (HR: 1.83; P = 0.022) carried prognostic significance.
DISCUSSION
In the present study weight loss alone (10%) did not define a population that differed for self-reported functional aspects of quality of life. A weight loss 10% did, however, define a group of patients with different physician-reported performance scores and grip strengths. In marked contrast, a multifactor cachexia profile that was based on weight loss, food intake, and inflammatory status defined a cachectic population that had both a lower subjective and objective functional ability (Table 3). Although this finding does not prove a cause-and-effect relation between such variables and the development or effects of cachexia, it does suggest that food intake and systemic inflammation may be important factors that affect patients' function and may therefore be worthy targets for specific intervention in the multimodal management of the cachexia syndrome (15)
Previous studies have suggested a strong link between systemic inflammation and fatigue or depression in advanced cancer (4, 16). Thus, the inclusion of a marker of systemic inflammation (ie, CRP) in a cachexia stratification system could account for patients with objective loss of function also perceiving themselves to have reduced function. Alternatively, it might be that a stage of cachexia characterized by weight loss alone is associated with objective impairment of function and that a generally more advanced stage of cachexia (characterized by weight loss, systemic inflammation, and reduced food intake) results in the patient having both an objective and subjective reduction in functional ability. The use of CRP as a generic marker for systemic inflammation in the present study does not preclude the testing of alternative markers (which may be even more predictive) in future studies.
In the present study (which included only patients with pancreatic cancer), weight loss alone was not identified as an independent prognostic variable, and this finding agrees with previous studies from our own (14) and other groups (17). The lack of an association may possibly relate to the confounding effect of edema and ascites on body weight measurement or the almost uniform development of weight loss (18) and shortened survival (14) in pancreatic cancer. In contrast, studies in patients with other types of cancer have suggested weight loss to be a negative prognostic indicator (19). However, whether weight loss in such patients is an independent prognostic indicator or is more related to the ability of patients without weight loss to complete chemotherapy (20) and thus survive longer is not known.
Pancreatic cancer is difficult to stage accurately. In particular, it is difficult to distinguish on a CT scan a localized tumor from peritumor pancreatitis or the presence or absence of lymph node metastases. Further refinement of staging may result from more routine use of positron emission tomography or CT scanning. However, within these limitations, in the present study stage of disease was not selected as a significant prognostic variable in the 3-factor model or subcomponents thereof. These findings emphasize the cachexia phenotype as a valid therapeutic target not only for function (described earlier) but also prognosis. Interestingly, for patients with localized disease, the 3-factor cachexia profile carried strong prognostic value. In contrast, for patients with metastatic disease it was CRP (known to increase with tumor progression) and CA19-9 (a potential marker of tumor bulk) that individually were significant prognostic indicators. Thus, although a multidimensional view of cachexia would appear to be important for prognosis in localized disease, tumor-related factors (CA19-9 and CRP) appear to dominate prognosis in patients with metastatic disease. These findings emphasize the changing problems faced by patients with cancer as their disease advances. It is of substantial interest that cachexia may be a more meaningful therapeutic target (for altering prognosis) when patients have an earlier stage of disease.
Previous studies in the treatment of malnourished or high-risk patients with artificial nutritional support have suggested that it is those patients with the poorest nutritional status and food intake that respond best (21). In reaching a definition of cachexia for trial purposes it is important to balance the inclusion of those who will survive long enough to benefit from the intervention (moderate to good prognosis) compared with identifying those who may respond most to the intervention (poor nutritional or functional status). Use of a multifactor profile of cachexia to stratify patients entering therapy trials might be one method to help resolve this question prospectively.
In the present study when all 3 factors were used to characterize cachexia, 22% of the population was identified with both objective and subjective loss of functional ability. However, when the profile was modified so that 2 of the 3 factors were required, 60% of the population was identified as cachectic with no loss of the ability to discriminate a population that had reduced subjective and objective functional capacity (Table 4). Thus, the cachexia profile in which 2 of 3 factors were met would appear to be the more inclusive, and therefore more useful, at least for use as an entry criterion for clinical trials. It must be recognized, however, that the documentation of food intake by a 3-d diet diary may be difficult in routine clinical practice and that an alternative method may be required to allow deployment of the cachexia profile outside a research environment.
The present study was conducted with a uniform cohort of patients who all had advanced pancreatic cancer, and this is one of the main strengths of the study. It could be argued, however, that the particular factors used to characterize cachexia in more detail might not be applicable to different types of malignancy. Pancreatic cancer is thought to be a useful paradigm for cancer cachexia because of the patients' almost uniform pattern of weight loss and shortened duration of survival (18). Thus, the results may be applicable to other groups, but this would require prospective validation.
The patients in the present study were not considered suitable to receive systemic chemotherapy either because of advanced stage of disease, being too ill, or the balance of benefit compared with toxicity being felt to be in favor of the latter. Some centers would routinely consider palliative chemotherapy for patients with pancreatic cancer, which has been documented to provide a 1-mo survival advantage and a limited improvement in quality of life (22). Given the relatively limited effect of chemotherapy on outcome and the recognized interaction between chemotherapy and nutritional status, it would seem likely that the predictive profile identified in the present population would be applicable to a population so treated, particularly those with more advanced disease. However, this would also require prospective validation.
In conclusion, cachexia is a multidimensional, multifactorial syndrome and is not synonymous with weight loss alone. A profiling system for cachexia needs to reflect such complexity, especially when considering the physical function sequelae for patients with cachexia. Inclusion of factors such as energy intake and the presence of systemic inflammation in addition to weight loss appears to provide a platform to further understanding of the important therapeutic targets for patients. Moreover, such a profile may be of practical value in determining who should be included in future clinical trials.
ACKNOWLEDGMENTS
We thank the Cancer Cachexia Study Group for their help in collecting the original data set.
KCF contributed to the conception and design of the study, provision of patients, data analysis and interpretation, manuscript writing, and final approval of the manuscript. ACV contributed to the conception and design of the study, administrative support, collection and assembly of data, data analysis and interpretation, manuscript writing, and final approval of the manuscript. DSH contributed to the collection and assembly of the data, data analysis and interpretation, manuscript writing, and final approval of the manuscript. ACV and DSH are employees of Ross Products Division, Abbott Laboratories. KCF has received lecture fees and research funding from Abbott Laboratories.
REFERENCES