- Email: [email protected]
Resting energy expenditure in patients with solid tumors undergoing anticancer therapy
Nutrition 22 (2006) 609 – 615 www.elsevier.com/locate/nut
Applied nutritional investigation
Resting energy expenditure in patients with solid tumors undergoing anticancer therapy Marina M. Reeves, Ph.D.,a,* Diana Battistutta, Ph.D.,a Sandra Capra, Ph.D.,b Judy Bauer, Ph.D.,c and Peter S. W. Davies, Ph.D.d a
Centre for Health Research, School of Public Health, Queensland University of Technology, Brisbane, Queensland, Australia b School of Health Sciences, University of Newcastle, Callaghan, New South Wales, Australia c The Wesley Research Institute, Auchenflower, Brisbane, Queensland, Australia d Children’s Nutrition Research Centre, Royal Children’s Hospital, University of Queensland, Herston, Brisbane, Queensland, Australia Manuscript received April 29, 2005; accepted March 17, 2006.
Abstract
Objective: Few studies have investigated the resting energy expenditure (REE) of, or determined the individual predictive accuracy of prediction equations in, cancer patients undergoing anticancer therapy. This study compared the measured REE of patients with cancer undergoing anticancer therapy with (1) healthy subjects and (2) REE estimated from commonly used prediction methods. Methods: Resting energy expenditure was measured in 18 cancer patients and 17 healthy subjects by using indirect calorimetry under standard conditions and was estimated from seven prediction methods. Fat-free mass (FFM) was measured by bioelectrical impedance analysis. Data were analyzed with regression modeling to adjust REE for FFM. Agreement between measured and predicted REE values was analyzed using the Bland-Altman approach. Results: There was no significant difference in FFM-adjusted REE between cancer patients and healthy subjects (mean difference 10%). Limits of agreement were wide for all prediction methods in estimating REE as much as 40% below and up to 30% above measured REE. Conclusions: REE in cancer patients undergoing anticancer therapies does not appear to be as high as commonly thought. None of the prediction equations examined were acceptable for predicting REE of individual cancer patients or healthy subjects. © 2006 Elsevier Inc. All rights reserved.
Keywords:
Resting energy expenditure; Prediction equations; Anticancer therapy; Indirect calorimetry; Energy requirement; Nutritional support
Introduction Weight loss and malnutrition commonly occur in patients with cancer [1,2], which may in part be a result of metabolic alterations caused by the tumor [3]. As such, it is generally believed that energy expenditure and, hence, energy requirements are increased in cancer patients. Minimizing weight loss and maintenance of weight are important goals for the nutritional care of patients with cancer. The ability to accurately determine the energy requirement of patients is essential for the provision of optimal nutritional support. * Corresponding author: Tel.: ⫹61-7-3258-2322; fax: ⫹61-7-32582310. E-mail address: [email protected] (M.M. Reeves). 0899-9007/06/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.nut.2006.03.006
Several studies have investigated the resting energy expenditure (REE) of cancer patients compared with noncancer subjects. However, only a few studies have used appropriate statistical analyses to adjust for body composition differences between groups and have found conflicting results. No significant difference was observed between REE of cancer patients with mixed tumor sites and that of non-cancer controls when adjusted for fat-free mass (FFM) [4 – 6], whereas studies of patients with lung cancer appear to consistently show high REE when compared with healthy control subjects [7,8]. These studies have isolated the effect on REE of the tumor itself and excluded patients who have commenced anticancer therapy. In a hospital setting, however, most patients who require nutritional support are undergoing some form of anticancer therapy.
610
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
The most appropriate method for determining patients’ energy requirements is by measurement of energy expenditure, most commonly by indirect calorimetry. These methods are rarely available and are not practical in a clinical setting due to the expense and time involved and the need for a trained technician. As such, prediction equations derived from healthy populations are commonly used to estimate patients’ REE and energy requirements in a clinical setting. Several studies have compared measured REE with predicted REE from the Harris-Benedict equations [9]. These studies have usually used the equations as standards for defining hypermetabolism, with variable results [10 –12]. To be of use in a clinical setting and assist in determining patients’ energy requirements, the individual predictive accuracy of the equations is of most interest. Only one pilot study has investigated the individual predictive accuracy of a number of prediction equations in patients with pancreatic cancer [13]. The aim of this study therefore was to compare the measured REE of patients with cancer undergoing anticancer therapy with that of healthy control subjects and with estimated REE to determine the individual predictive accuracy of commonly used prediction equations. Other data from this study have been presented elsewhere [14].
Materials and methods Subjects Patients who were ⬎18 y of age with histologically proved solid tumors were recruited from consecutive new patients attending a private radiation oncology center over a 6-mo period. Patients with solid tumors of the breast, prostate, or brain were excluded from the study because the limited literature suggests little effect of the tumor on energy expenditure in these patients [15–17]. Patients who had undergone surgery within the month before the study, had severe endocrine abnormalities (e.g., hypothyroidism, hyperthyroidism), or were treated with high-dose steroid medications were also excluded. These criteria were based on excluding conditions that have an independent effect on energy expenditure. No exclusion criteria based on current anticancer treatment were included; however, details of treatment at the time of the study were recorded. Healthy subjects were recruited from a purposive volunteer sample of individuals from the affiliated institutions. Healthy subjects were in self-reported good health and were group matched to cancer patients based on gender, age (⫾10 y), height (⫾10 cm), and weight (⫾5 kg) as surrogates for matching the FFM between the groups. Healthy subjects did not have a history of cancer or severe endocrine abnormalities, had not undergone surgery within 1 mo of the study, and were not treated with high-dose steroid medication. Because health status was self-reported and not based on a
full medical assessment, we cannot completely exclude the influence of other potential health factors on energy expenditure. The ethics committees of the relevant medical and tertiary institutions approved the conduct of this study. Voluntary signed informed consent was obtained from each subject before commencement of the study. Measurement protocol The measurement protocol and methods for measuring REE have been described in detail elsewhere [14]. In brief, measurements of REE were conducted under outpatient conditions, after an overnight fast (ⱖ12 h), and between 0700 and 0900 h after a 30-min rest period. REE was measured by breath-by-breath respiratory gas exchange with the VMax 229 (SensorMedics, Yorba Linda, CA, USA) indirect calorimetry device using a mouthpiece and nose clip. Measurements were identical for cancer patients and healthy subjects. Measurements were ceased once a steady-state period defined as three-minute period during which oxygen consumption per unit time, respiratory quotient, and minute ventilation changed by ⱕ10% was achieved [18]. Respiratory quotient was calculated as the ratio of carbon dioxide production to oxygen consumption per unit time. Oxygen consumption per unit time and carbon dioxide production were converted to REE by using the abbreviated Weir equation [19]. Height without shoes was measured to the nearest 0.5 cm using a stadiometer (KaWe, Asperg, Germany). Foot-tofoot bioelectrical impedance analysis (Model 300GS, Tanita Inc., Tokyo, Japan) was used to measure weight and body composition to the nearest 0.1 kg without shoes or heavy clothing. The proprietary algorithm for calculating FFM was used for healthy subjects. For cancer patients, the equation of Isenring et al. [20] was used to calculate total body water from impedance measured by the foot-to-foot bioelectrical impedance analysis, from which FFM was estimated assuming a constant hydration level of 73.2% [21]. This equation has been found to be acceptable for measuring body composition of groups of cancer patients (as is used in this study) but not for individual patients [22]. Nutritional status was measured with the Subjective Global Assessment, which categorizes people as well nourished (scale A), at risk of malnutrition or moderately malnourished (scale B), or severely malnourished (scale C) [23]. Prediction of REE Resting energy expenditure was estimated from a range of prediction equations [9,24 –29]. A ratio method, based on a value of 20 kcal/kg, was also calculated to compare with REE, although common recommendations for cancer patients are for a value of 30 to 35 kcal/kg to estimate total daily energy expenditure [30,31]. An additional prediction method for the group of cancer patients was included based
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
611
on recommendations for use of an injury factor for cancer of ⱖ1.3 in combination with the Harris-Benedict equations [32,33]. An adjusted body weight was used in the prediction of REE from the Harris-Benedict equations for patients with a body mass index ⬎29 kg/m2 [34,35]. This accounts for the increased proportion of body weight as low metabolically active adipose tissue and skeletal muscle in obese people [36].
ment at the individual level and assess whether the bias between predicted and measured REEs was consistent across the entire range of REE measurements. Pearson’s correlation coefficients were used to assess whether there were any trends in the magnitude of the bias with increasing REE measurement. The proportion of cancer patients with predicted REE within clinically acceptable limits (⫾10%) of measured REE was also calculated.
Statistical analysis
Results
Data were analyzed with SPSS 11.0.1 for Windows (SPSS Inc., Chicago, IL, USA). Continuous variables were normally distributed and are presented as mean ⫾ standard deviation or standard error, as stated. Categorical variables are presented as counts (percentages). A clinically significant difference of 30% between measured REE in cancer patients and healthy subjects was defined a priori, based on common recommendations that energy expenditure in cancer patients is 30% higher than in healthy subjects [32,33]. For comparison of measured to predicted REE, a clinically acceptable discrepancy of 10% was defined a priori based on the assumption that measured REE would fall within ⫾10% of predicted REE [9]. Of the two power calculations, the sample for the paired analysis was larger. As such a minimum sample of 11 subjects per group was required to detect these clinically meaningful differences, assuming a standard deviation of 10%, with 90% power at the 95% significance level (two-tailed). Characteristics of cancer patients and healthy control subjects were compared by independent sample t tests for continuous variables and Fisher’s exact tests for categorical variables. Unadjusted REE between cancer patients and healthy subjects was assessed by independent sample t tests. A general linear modeling approach was taken for the regression analysis to adjust the association between measured REE and health status for differences in FFM between the two groups. Results are expressed as adjusted means ⫾ standard error. Analysis of REE data for body composition differences between groups was based on appropriate methods reported by other investigators [37,38]. Weight loss was also considered as a potential confounding variable; however, preliminary analysis indicated that weight loss was not associated with REE (adjusted for FFM) and therefore was not included in the final model. Tumor type and surgery were considered as potential effectmodifying variables through stratified analyses; however, the small sample precluded meaningful analyses. Measured REE was compared with REE predicted by the different prediction methods using the Bland-Altman approach [39]. Paired t tests were used to first assess agreement between the measured and predicted REEs at the group (average) level. Mean bias, limits of agreement (⫾2 standard deviations), and plot of bias against the average of measured and predicted REEs were used to describe agree-
Eighty-three eligible patients attended the radiation oncology center over the 6-mo period, 19 of whom consented to participate in the study. There was no significant difference between cancer patients who consented to participate and the total eligible pool with respect to gender, age, and tumor site (data not shown). One patient became ill after consent and was unable to undertake further data collection. As such, results are based on data from 18 cancer patients. Cancer patients were categorized into three groups according to tumor site: lung, which included non–small cell lung cancer and small cell lung cancer (n ⫽ 8); gastrointestinal tract (GIT), which included any tumors located in the GIT from the mouth to the anus (n ⫽ 7); and other, which included bladder, cervical, and testicular tumors (n ⫽ 3). Data on tumor stage were available for a limited number of patients. Three patients were diagnosed with pulmonary metastases, two patients were treated for tumor recurrence, and five patients had undergone surgical resection of the tumor. All but one patient had commenced radiotherapy treatment and 50% were undergoing concurrent chemotherapy at the time of the study. Seventeen healthy subjects participated in the study. Characteristics of subjects are presented in Table 1. There was no significant difference in gender, age, weight, height, body mass index, or FFM between groups. The nutritional status of cancer patients and healthy subjects differed significantly (Fisher’s exact test, P ⬍ 0.001), with the largest proportion of cancer patients categorized as moderately malnourished compared with healthy subjects who were well nourished. Although not statistically significant, the two groups differed in the amount of weight lost over the 6 mo before the study. All healthy subjects had stable weight, except two who had lost ⱖ5% body weight. Five cancer patients had stable weight or had gained weight over the previous 6 mo, six had lost ⬍5% body weight, and four had lost ⱖ5% body weight. Two cancer patients did not achieve steady state during REE measurement and one cancer patient was unable to have FFM assessed due to the presence of a pacemaker. Regression analyses are therefore based on data from 15 cancer patients and 17 healthy subjects. The average duration of REE measurements was a median of 9.5 min (range 5–23). Mean ⫾ standard deviation respiratory quotients for
612
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
Table 1 Subjects’ characteristics and REE*
Male:female Age (y) Height (cm) Weight (kg) BMI (kg/m2) FFM (kg) Body fat (%) Weight loss over previous 6 mo (%) SGA Well nourished (scale A) Moderately malnourished (scale B) Severely malnourished (scale C) Unadjusted REE (kJ/d) FFM-adjusted REE (kJ/d)
Cancer patients (n ⫽ 18)
Healthy subjects (n ⫽ 17)
11:7 65 ⫾ 13 167 ⫾ 14 80.2 ⫾ 19.8 28.4 ⫾ 5.0 51.9 ⫾ 10.9 33.7 ⫾ 10.5 2.2 ⫾ 4.6
10:7 60 ⫾ 11 169 ⫾ 9 75.0 ⫾ 13.0 26.3 ⫾ 4.1 52.2 ⫾ 11.3 30.5 ⫾ 8.8 0.9 ⫾ 2.6
4 (22%) 12 (67%)
17 (100%) 0 (0%)
2 (11%)
0 (0%)
6660 ⫾ 376† 6595 ⫾ 276†
5979 ⫾ 303‡ 6024 ⫾ 259‡
BMI, body mass index; FFM, fat-free mass; SGA, Subjective Global Assessment; REE, resting energy expenditure * Data are mean ⫾ standard error. † n ⫽ 15. ‡ n ⫽ 17.
the REE measurements were 0.70 ⫾ 0.09 for cancer patients and 0.72 ⫾ 0.08 for healthy subjects. Unadjusted and FFM-adjusted REEs for cancer patients and healthy subjects are presented in Table 1. There was no statistically or clinically significant difference between cancer patients and healthy subjects for unadjusted REE or FFM-adjusted REE. Adjusted REE was 9.5% (571 kJ) higher in cancer patients than in healthy subjects. Regression lines are shown in Figure 1. Further analyses indicated that FFM-adjusted REE differed significantly between tumor sites. Adjusted REE was similar between patients with lung cancer (mean ⫾ standard error, 6825 ⫾ 306 kJ) and those with GIT cancer (6584 ⫾ 331 kJ) but was considerably lower in those with “other” cancers (3774 ⫾ 736 kJ). FFM-adjusted REE excluding the three “other” patients increased the difference between cancer patients (lung and GIT) and healthy subjects to 12%, but the difference remained clinically and statistically insignificant. FFM-adjusted REE also differed between patients who had surgical removal of the tumor (7575 ⫾ 514 kJ) and those with the tumor in situ (6326 ⫾ 310 kJ). The sample of subjects who had undergone surgical removal was small (n ⫽ 4, one was excluded because steady state was not reached) and therefore did not permit further investigation of this group. The difference in FFM-adjusted REE between cancer patients and healthy subjects was further decreased (mean difference 5%) when the four subjects with surgical removal of the tumor were excluded from analysis. There was no difference in FFM-adjusted REE between patients who were receiving chemotherapy and those who were not. Measured REE in cancer patients and healthy subjects
was also compared with predicted REE from a number of commonly used prediction equations, to determine the individual predictive accuracy of the various methods. Predicted REE, mean bias, and limits of agreement for the cancer patients are listed in Table 2. In terms of mean bias, all prediction methods with the exception of the HarrisBenedict equation, with an injury factor of 1.3, were within clinically acceptable limits of ⫾10% of measured REE. The smallest mean bias was observed with the equations by Wang et al. [29], based on FFM. However, when data were analyzed for accuracy for individuals, each prediction method consistently produced wide limits of agreement (⬎20%), indicating potentially poor prediction for individual cancer patients. The Harris-Benedict (alone), the equations by Owen et al. [26,27] and Mifflin et al. [25], and the 20-kcal/kg method predicted REE within clinically acceptable limits for just over 50% of the sample of cancer patients. The remaining prediction methods failed to estimate REE within clinically acceptable limits for ⬎66% of individual cancer patients. For the group of cancer patients, a significant trend was observed in the spread of bias, with increasing REE values between measured REE and REE predicted by the equations of Cunningham et al. [24] (r ⫽ ⫺0.503, P ⫽ 0.056) and Wang et al. [29] (r ⫽ ⫺0.507, P ⫽ 0.054). Although not statistically significant, the bias observed with REE predicted from the equation of Schofield [28] showed a tendency toward underestimation of measured REE with increasing REE values (r ⫽ ⫺0.328, P ⫽ 0.215). No patterns were observed between the bias and mean REE value for any of the other prediction methods. For the group of healthy subjects, mean bias was within clinically acceptable limits for all prediction equations. However, the limits of agreement were wide (⫺30% to 40%), indicating poor prediction for individual healthy subjects.
Fig. 1. Plot and regression lines of FFM versus REE for cancer patients and healthy subjects. The interaction between REE and FFM was not significant (F1,31 ⫽ 0.336, P ⫽ 0.567). FFM, fat-free mass; REE, resting energy expenditure.
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
613
Table 2 Predicted REE, mean bias, and limits of agreement for difference between predicted and measured REE* of cancer patients (n ⫽ 16)
Harris-Benedict Harris-Benedict ⫻ 1.3 Schofield Owen Mifflin Cunningham§ Wang§ 20 kcal/kg
Predicted REE (kJ/d)
Bias (kJ/d)
Limits of agreement (⫾2 SD)†
Proportion within ⫾10%‡
6328 ⫾ 1368 8226 ⫾ 1778 6373 ⫾ 1131 6800 ⫾ 1214 6315 ⫾ 1427 6383 ⫾ 950 6516 ⫾ 946 6885 ⫾ 1635
⫺303 ⫾ 798 1595 ⫾ 980 ⫺258 ⫾ 904 169 ⫾ 955 ⫺315 ⫾ 882 ⫺276 ⫾ 1099 ⫺144 ⫾ 1099 254 ⫾ 1301
⫺28–20% 6–54% ⫺31–23% ⫺26–31% ⫺31–22% ⫺41–32% ⫺39–34% ⫺35–43%
50.0% 18.8% 31.3% 56.3% 56.3% 20.0% 26.7% 56.3%
REE, resting energy expenditure * Measured REE ⫽ 6631 ⫾ 1411 kJ/d. Data are mean ⫾ standard deviation. † Percentage of measured REE. ‡ Clinically acceptable limit (⫾10% of measured). § n ⫽ 15, measured REE ⫽ 6660 ⫾ 1455 kJ/d.
Discussion An understanding of whether metabolism and energy expenditure are altered in cancer patients is necessary for determining appropriate nutritional support recommendations for these patients. Most studies that have investigated energy expenditure in cancer patients have been conducted from a “purist” viewpoint. This is, they have attempted to isolate the effect of the tumor itself on energy expenditure. However, many cancer patients who require nutritional support are often undergoing anticancer therapies. Thus, this study was conducted from a “clinical practice” viewpoint. In addition, due to the limited ability for practitioners to measure patients’ energy expenditure in a clinical setting, there is a demand for easy tools for estimating individual patient requirements. Therefore, this study compared the measured REE of patients with cancer undergoing anticancer therapy with that of healthy control subjects and with REE estimated by a number of commonly used prediction methods. The sample of cancer patients recruited was heterogeneous in nature in terms of tumor site, stage, and type, current treatment, and previous weight loss. The group of healthy subjects was closely matched to the group of cancer patients in terms of FFM. Although the sampling frame for cancer patients and healthy subjects differed, it is primarily FFM that influences REE, our outcome measurement, and not other characteristics associated with a person’s environment. Recruitment of healthy subjects was based on characteristics matching cancer patients and self-reported health and exclusions were based only on the specified criteria. However, because ⬎50% of cancer patients were ⬎60 y of age, the likelihood of the presence of some health conditions in the control subjects is increased. In this study, when measured REE was adjusted for FFM, there was no difference between healthy subjects and cancer patients when defined by our a priori clinically acceptable limit of 30%. This definition of meaningful difference was based on common recommendations that energy expenditure in cancer patients is ⱖ30% higher than in
healthy counterparts (injury factor 1.3) [32,33]. This study found only a 10% difference in REE between the two groups, but the sample was not powered to detect this difference. Similar results have been observed in other studies. Staal-van den Brekel et al. [8] found a 9% higher FFMadjusted REE in patients with non–small cell lung cancer and a 17% higher FFM-adjusted REE in patients with small cell lung cancer compared with healthy control subjects. Jatoi et al. [7] observed an adjusted REE in patients with non–small cell lung cancer that was 10% to 12% higher than in healthy controls matched by gender, age, and body mass index. Our sample consisted of patients with solid tumors of different sites. Our data indicated that the “other” cancers, which included tumors of the bladder, cervix, and testes, in three patients appeared to have little effect on REE in this study. However, excluding these subjects from analysis had minimal effect on the results. Cancer patients also differed with respect to tumor stage and type (primary, recurrent, or metastatic). Studies have suggested that REE is not affected by differences in tumor stage [8], whereas tumor recurrence has shown slight increases in REE relative to initial primary diagnosis [40] and presence of metastatic disease has shown no to a small effect on REE in some studies [41,42] and a significant increase in REE in others [43]. It should be noted that several of these studies did not analyze energy expenditure data appropriately, so these results should be interpreted with caution. Because this study included patients who had commenced anticancer therapies, patients were included who had undergone surgical removal of the tumor. It has been suggested that metabolic alterations in patients with cancer are related to cytokine release from the tumor [3]. After surgical removal of the tumor and recovery from surgery, it would be expected that tumor effects on REE would no longer persist. Luketich et al. [44] and Arbeit et al. [41] observed decreases in REE after curative tumor resection, indicating that the tumor-bearing state affected energy me-
614
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
tabolism. In our study, four patients who had undergone tumor resection had an unexpectedly higher FFM-adjusted REE than did those with the tumor in situ. These results can only be speculative and are subject to confounding because there may be other characteristics of these patients, such as tumor type or stage, that influence REE. However, these four patients underwent surgery 7 to 11 wk previously, so the influence of surgery on energy expenditure would have diminished [45]. Percentage of initial body weight loss over the previous 6 mo was not associated with REE in this study. Hansell et al. [5] found that the weight-losing state rather than the tumor-bearing state affected REE, but other investigators have not confirmed this result [6]. These conflicting results may be due to differences in the definition of weight losing or absolute amount of weight loss between studies. In our study only two patients (11%) had lost ⬎5% of initial body weight compared with 43% of patients with ⬎10% weight loss [5] and 56% with ⬎5% weight loss [6]. As a result of the small amount of weight loss observed in our study, we cannot make any judgments on the effect of weight loss on REE in patients with cancer. To be of use in a clinical setting, prediction equations should estimate REE within a reasonable degree of accuracy for individuals. Only one study has previously investigated the individual predictive accuracy of prediction equations in cancer patients and found that at the group level the prediction methods acceptably estimated REE of pancreatic cancer patients but were not as accurate for predicting REE for individual patients [13]. We observed similar results in our study. No prediction method estimated REE within acceptable limits for the full range of individual cancer patients. For this sample of cancer patients, the best combinations of smallest bias, narrowest limits of agreement, and largest proportion within clinically acceptable limits were observed with the HarrisBenedict (alone), Owen, and Mifflin equations. Automatic application of an injury factor of 1.3 with the HarrisBenedict equations consistently overestimated REE in this study. The results observed in the sample of healthy subjects were consistent with those previously reported [46,47]. In summary, this study found an overall 10% increase in REE in cancer patients receiving anticancer therapies compared with healthy control subjects, which is well below common recommendations of a 30% increase in REE in cancer patients. Further studies with larger samples are required to determine whether this 10% difference is significant. Current prediction methods do not predict individuals’ REE within clinically acceptable limits. Our results call into question automatic application of “injury factors” when predicting energy requirements for cancer patients from the Harris-Benedict equations. In a clinical setting, to ensure appropriate nutritional management of patients with cancer, intake and patient outcomes, such as weight and nutritional status, should be monitored regularly to determine whether patients’ energy requirements are being met.
Acknowledgements The authors thank the cancer patients and healthy subjects for willingly participating in the study, the doctors and nursing staff from Wesley Cancer Care Centre for assistance with patient recruitment, and staff at the Queensland Respiratory Laboratory for assistance with data collection.
References [1] Kern KA, Norton JA. Cancer cachexia. JPEN 1988;12:286 –98. [2] DeWys WD, Begg C, Lavin PT, Band PR, Bennett JM, Bertino JR, Cohen MH, et al. Prognostic effect of weight loss prior to chemotherapy in cancer patients. Am J Med 1980;69:491–7. [3] Fearon KCH, Barber MD, Moses AGW. The cancer cachexia syndrome. Surg Oncol Clin North Am 2001;10:109 –26. [4] Lindmark L, Bennegard K, Eden E, Ekman L, Schersten T, Svaninger G, Lundholm K. Resting energy expenditure in malnourished patients with and without cancer. Gastroenterology 1984;87:402– 8. [5] Hansell DT, Davies JW, Burns HJ. The relationship between resting energy expenditure and weight loss in benign and malignant disease. Ann Surg 1986;203:240 –5. [6] Fredrix EW, Soeters PB, Rouflart MJ, von Meyenfeldt MF, Saris WH. Resting energy expenditure in patients with newly detected gastric and colorectal cancers. Am J Clin Nutr 1991;53:1318 –22. [7] Jatoi A, Daly BD, Hughes VA, Dallal GE, Kehayias J, Roubenoff R. Do patients with nonmetastatic non–small cell lung cancer demonstrate altered resting energy expenditure? Ann Thorac Surg 2001;72: 348 –51. [8] Staal-van den Brekel AJ, Schols AM, Dentener MA, ten Velde GP, Buurman WA, Wouters EF. Metabolism in patients with small cell lung carcinoma compared with patients with non–small cell lung carcinoma and healthy controls. Thorax 1997;52:338 – 41. [9] Harris JA, Benedict FG. A biometric study of basal metabolism in man. Washington, DC: Carnegie Institution of Washington; 1919. [10] Bosaeus I, Daneryd P, Svanberg E, Lundholm K. Dietary intake and resting energy expenditure in relation to weight loss in unselected cancer patients. Int J Cancer 2001;93:380 –3. [11] Dempsey DT, Feurer ID, Knox LS, Crosby LO, Buzby GP, Mullen JL. Energy expenditure in malnourished gastrointestinal cancer patients. Cancer 1984;53:1265–73. [12] Knox LS, Crosby LO, Feurer ID, Buzby GP, Miller CL, Mullen JL. Energy expenditure in malnourished cancer patients. Ann Surg 1983; 197:152– 62. [13] Bauer J, Reeves MM, Capra S. The agreement between measured and predicted resting energy expenditure in patients with pancreatic cancer - a pilot study. JOP. J Pancreas (Online) 2004;5:32– 40. [14] Reeves MM, Capra S, Bauer J, Davies PSW, Battistutta D. Clinical accuracy of the MedGem™indirect calorimeter for measuring resting energy expenditure in cancer patients. Eur J Clin Nutr 2005;59: 603–10. [15] Demark-Wahnefried W, Peterson BL, Winer EP, Marks L, Aziz N, Marcom PK, et al. Changes in weight, body composition, and factors influencing energy balance among premenopausal breast cancer patients receiving adjuvant chemotherapy. J Clin Oncol 2001;19: 2381–9. [16] Leenders KL. PET: blood flow and oxygen consumption in brain tumors. J Neurooncol 1994;22:269 –73. [17] Platz EA. Energy imbalance and prostate cancer. J Nutr 2002;132: 3471S– 81. [18] Reeves MM, Davies PSW, Bauer J, Battistutta D. Reducing the time period of steady state does not affect the accuracy of energy expenditure measurements by indirect calorimetry. J Appl Physiol 2004; 97:130 – 4.
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615 [19] Weir JBDeV. New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol 1949;109:1–9. [20] Isenring E, Bauer J, Capra S, Davies PSW. Evaluation of foot-to-foot bioelectrical impedance analysis for the prediction of total body water in oncology outpatients receiving radiotherapy. Eur J Clin Nutr 2004; 58:46 –51. [21] Hoffer EC, Meador CK, Simpson DC. Correlation of whole-body impedance with total body water volume. J Appl Physiol 1969;27: 531– 4. [22] Bauer J. New insights into medical nutrition therapy for cancer cachexia. Centre for Health Research. Brisbane: Queensland University of Technology; 2003. [23] Detsky AS, McLaughlin JR, Baker JP, Johnston N, Whittaker S, Mendelson RA, Jeejeebhoy KN. What is subjective global assessment of nutritional status? JPEN 1987;11:8 –13. [24] Cunningham JJ. Body composition as a determinant of energy expenditure: a synthetic review and a proposed general prediction equation. Am J Clin Nutr 1991;54:963–9. [25] Mifflin MD, St Jeor ST, Hill LA, Scott BJ, Daugherty SA, Koh YO. A new predictive equation for resting energy expenditure in healthy individuals. Am J Clin Nutr 1990;51:241–7. [26] Owen OE, Holup JL, D’Alessio DA, Craig ES, Polansky M, Smalley KJ, et al. A reappraisal of the caloric requirements of men. Am J Clin Nutr 1987;46:875– 85. [27] Owen OE, Kavle E, Owen RS, Polansky M, Caprio S, Mozzoli MA, et al. A reappraisal of caloric requirements in healthy women. Am J Clin Nutr 1986;44:1–19. [28] Schofield WN. Predicting basal metabolic rate, new standards and review of previous work. Hum Nutr Clin Nutr 1985;39:5– 41. [29] Wang Z, Heshka S, Gallagher D, Boozer CN, Kotler DP, Heymsfield SB. Resting energy expenditure–fat-free mass relationship: new insights provided by body composition modeling. Am J Physiol Endocrinol Metab 2000;279:E539 – 45. [30] Bloch AS. Cancer. In: Matarese LE, Gottschlich MM, editors. Contemporary nutrition support practice: a clinical guide. Philadelphia: WB Saunders; 1998. [31] A.S.P.E.N. Board of Directors and the Clinical Guidelines Task Force. Guidelines for the use of parenteral and enteral nutrition in adult and pediatric patients. JPEN 2002;26:1–138. [32] Alpers D, Stenson W, Bier D. Manual of nutritional therapeutics. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2002. [33] Dietitians’ pocket book. Perth: Curtin University of Technology; 1999.
615
[34] Barak N, Wall-Alonso E, Sitrin MD. Evaluation of stress factors and body weight adjustments currently used to estimate energy expenditure in hospitalized patients. JPEN 2002;26:231– 8. [35] Glynn CC, Greene GW, Winkler MF, Albina JE. Predictive versus measured energy expenditure using limits-of-agreement analysis in hospitalized, obese patients. JPEN 1999;23:147–54. [36] Heymsfield SB. Heat and life: the ongoing scientific odyssey. JPEN 2002;26:319 –32. [37] Davies PSW, Cole TJ. The adjustment of measures of energy expenditure for body weight and body composition. Int J Body Comp Res 2003;1:45–50. [38] Ravussin E, Bogardus C. Relationship of genetics, age, and physical fitness to daily energy expenditure and fuel utilization. Am J Clin Nutr 1989;49:968 –75. [39] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1: 307–10. [40] Fredrix EW, Soeters PB, Wouters EF, Deerenberg IM, von Meyenfeldt MF, Saris WH. Effect of different tumor types on resting energy expenditure. Cancer Res 1991;51:6138 – 41. [41] Arbeit JM, Lees DE, Corsey R, Brennan MF. Resting energy expenditure in controls and cancer patients with localized and diffuse disease. Ann Surg 1984;199:292– 8. [42] Hansell DT, Davies JW, Burns HJ. Effects of hepatic metastases on resting energy expenditure in patients with colorectal cancer. Br J Surg 1986;73:659 – 62. [43] Macfie J, Burkinshaw L, Oxby C, Holmfield JH, Hill GL. The effect of gastrointestinal malignancy on resting metabolic expenditure. Br J Surg 1982;69:443– 6. [44] Luketich JD, Mullen JL, Feurer ID, Sternlieb J, Fried RC. Ablation of abnormal energy expenditure by curative tumor resection. Arch Surg 1990;125:337– 41. [45] Duke JH, Jorgensen SB, Broell JR, Long CL, Kinney JM. Contribution of protein to caloric expenditure following injury. Surgery 1970; 68:168 –74. [46] Taaffe DR, Thompson J, Butterfield G, Marcus R. Accuracy of equations to predict basal metabolic rate in older women. J Am Diet Assoc 1995;95:1387–92. [47] Siervo M, Boschi V, Falconi C. Which REE prediction equation should we use in normal-weight, overweight and obese women? Clin Nutr 2003;22:193–204.
Applied nutritional investigation
Resting energy expenditure in patients with solid tumors undergoing anticancer therapy Marina M. Reeves, Ph.D.,a,* Diana Battistutta, Ph.D.,a Sandra Capra, Ph.D.,b Judy Bauer, Ph.D.,c and Peter S. W. Davies, Ph.D.d a
Centre for Health Research, School of Public Health, Queensland University of Technology, Brisbane, Queensland, Australia b School of Health Sciences, University of Newcastle, Callaghan, New South Wales, Australia c The Wesley Research Institute, Auchenflower, Brisbane, Queensland, Australia d Children’s Nutrition Research Centre, Royal Children’s Hospital, University of Queensland, Herston, Brisbane, Queensland, Australia Manuscript received April 29, 2005; accepted March 17, 2006.
Abstract
Objective: Few studies have investigated the resting energy expenditure (REE) of, or determined the individual predictive accuracy of prediction equations in, cancer patients undergoing anticancer therapy. This study compared the measured REE of patients with cancer undergoing anticancer therapy with (1) healthy subjects and (2) REE estimated from commonly used prediction methods. Methods: Resting energy expenditure was measured in 18 cancer patients and 17 healthy subjects by using indirect calorimetry under standard conditions and was estimated from seven prediction methods. Fat-free mass (FFM) was measured by bioelectrical impedance analysis. Data were analyzed with regression modeling to adjust REE for FFM. Agreement between measured and predicted REE values was analyzed using the Bland-Altman approach. Results: There was no significant difference in FFM-adjusted REE between cancer patients and healthy subjects (mean difference 10%). Limits of agreement were wide for all prediction methods in estimating REE as much as 40% below and up to 30% above measured REE. Conclusions: REE in cancer patients undergoing anticancer therapies does not appear to be as high as commonly thought. None of the prediction equations examined were acceptable for predicting REE of individual cancer patients or healthy subjects. © 2006 Elsevier Inc. All rights reserved.
Keywords:
Resting energy expenditure; Prediction equations; Anticancer therapy; Indirect calorimetry; Energy requirement; Nutritional support
Introduction Weight loss and malnutrition commonly occur in patients with cancer [1,2], which may in part be a result of metabolic alterations caused by the tumor [3]. As such, it is generally believed that energy expenditure and, hence, energy requirements are increased in cancer patients. Minimizing weight loss and maintenance of weight are important goals for the nutritional care of patients with cancer. The ability to accurately determine the energy requirement of patients is essential for the provision of optimal nutritional support. * Corresponding author: Tel.: ⫹61-7-3258-2322; fax: ⫹61-7-32582310. E-mail address: [email protected] (M.M. Reeves). 0899-9007/06/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.nut.2006.03.006
Several studies have investigated the resting energy expenditure (REE) of cancer patients compared with noncancer subjects. However, only a few studies have used appropriate statistical analyses to adjust for body composition differences between groups and have found conflicting results. No significant difference was observed between REE of cancer patients with mixed tumor sites and that of non-cancer controls when adjusted for fat-free mass (FFM) [4 – 6], whereas studies of patients with lung cancer appear to consistently show high REE when compared with healthy control subjects [7,8]. These studies have isolated the effect on REE of the tumor itself and excluded patients who have commenced anticancer therapy. In a hospital setting, however, most patients who require nutritional support are undergoing some form of anticancer therapy.
610
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
The most appropriate method for determining patients’ energy requirements is by measurement of energy expenditure, most commonly by indirect calorimetry. These methods are rarely available and are not practical in a clinical setting due to the expense and time involved and the need for a trained technician. As such, prediction equations derived from healthy populations are commonly used to estimate patients’ REE and energy requirements in a clinical setting. Several studies have compared measured REE with predicted REE from the Harris-Benedict equations [9]. These studies have usually used the equations as standards for defining hypermetabolism, with variable results [10 –12]. To be of use in a clinical setting and assist in determining patients’ energy requirements, the individual predictive accuracy of the equations is of most interest. Only one pilot study has investigated the individual predictive accuracy of a number of prediction equations in patients with pancreatic cancer [13]. The aim of this study therefore was to compare the measured REE of patients with cancer undergoing anticancer therapy with that of healthy control subjects and with estimated REE to determine the individual predictive accuracy of commonly used prediction equations. Other data from this study have been presented elsewhere [14].
Materials and methods Subjects Patients who were ⬎18 y of age with histologically proved solid tumors were recruited from consecutive new patients attending a private radiation oncology center over a 6-mo period. Patients with solid tumors of the breast, prostate, or brain were excluded from the study because the limited literature suggests little effect of the tumor on energy expenditure in these patients [15–17]. Patients who had undergone surgery within the month before the study, had severe endocrine abnormalities (e.g., hypothyroidism, hyperthyroidism), or were treated with high-dose steroid medications were also excluded. These criteria were based on excluding conditions that have an independent effect on energy expenditure. No exclusion criteria based on current anticancer treatment were included; however, details of treatment at the time of the study were recorded. Healthy subjects were recruited from a purposive volunteer sample of individuals from the affiliated institutions. Healthy subjects were in self-reported good health and were group matched to cancer patients based on gender, age (⫾10 y), height (⫾10 cm), and weight (⫾5 kg) as surrogates for matching the FFM between the groups. Healthy subjects did not have a history of cancer or severe endocrine abnormalities, had not undergone surgery within 1 mo of the study, and were not treated with high-dose steroid medication. Because health status was self-reported and not based on a
full medical assessment, we cannot completely exclude the influence of other potential health factors on energy expenditure. The ethics committees of the relevant medical and tertiary institutions approved the conduct of this study. Voluntary signed informed consent was obtained from each subject before commencement of the study. Measurement protocol The measurement protocol and methods for measuring REE have been described in detail elsewhere [14]. In brief, measurements of REE were conducted under outpatient conditions, after an overnight fast (ⱖ12 h), and between 0700 and 0900 h after a 30-min rest period. REE was measured by breath-by-breath respiratory gas exchange with the VMax 229 (SensorMedics, Yorba Linda, CA, USA) indirect calorimetry device using a mouthpiece and nose clip. Measurements were identical for cancer patients and healthy subjects. Measurements were ceased once a steady-state period defined as three-minute period during which oxygen consumption per unit time, respiratory quotient, and minute ventilation changed by ⱕ10% was achieved [18]. Respiratory quotient was calculated as the ratio of carbon dioxide production to oxygen consumption per unit time. Oxygen consumption per unit time and carbon dioxide production were converted to REE by using the abbreviated Weir equation [19]. Height without shoes was measured to the nearest 0.5 cm using a stadiometer (KaWe, Asperg, Germany). Foot-tofoot bioelectrical impedance analysis (Model 300GS, Tanita Inc., Tokyo, Japan) was used to measure weight and body composition to the nearest 0.1 kg without shoes or heavy clothing. The proprietary algorithm for calculating FFM was used for healthy subjects. For cancer patients, the equation of Isenring et al. [20] was used to calculate total body water from impedance measured by the foot-to-foot bioelectrical impedance analysis, from which FFM was estimated assuming a constant hydration level of 73.2% [21]. This equation has been found to be acceptable for measuring body composition of groups of cancer patients (as is used in this study) but not for individual patients [22]. Nutritional status was measured with the Subjective Global Assessment, which categorizes people as well nourished (scale A), at risk of malnutrition or moderately malnourished (scale B), or severely malnourished (scale C) [23]. Prediction of REE Resting energy expenditure was estimated from a range of prediction equations [9,24 –29]. A ratio method, based on a value of 20 kcal/kg, was also calculated to compare with REE, although common recommendations for cancer patients are for a value of 30 to 35 kcal/kg to estimate total daily energy expenditure [30,31]. An additional prediction method for the group of cancer patients was included based
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
611
on recommendations for use of an injury factor for cancer of ⱖ1.3 in combination with the Harris-Benedict equations [32,33]. An adjusted body weight was used in the prediction of REE from the Harris-Benedict equations for patients with a body mass index ⬎29 kg/m2 [34,35]. This accounts for the increased proportion of body weight as low metabolically active adipose tissue and skeletal muscle in obese people [36].
ment at the individual level and assess whether the bias between predicted and measured REEs was consistent across the entire range of REE measurements. Pearson’s correlation coefficients were used to assess whether there were any trends in the magnitude of the bias with increasing REE measurement. The proportion of cancer patients with predicted REE within clinically acceptable limits (⫾10%) of measured REE was also calculated.
Statistical analysis
Results
Data were analyzed with SPSS 11.0.1 for Windows (SPSS Inc., Chicago, IL, USA). Continuous variables were normally distributed and are presented as mean ⫾ standard deviation or standard error, as stated. Categorical variables are presented as counts (percentages). A clinically significant difference of 30% between measured REE in cancer patients and healthy subjects was defined a priori, based on common recommendations that energy expenditure in cancer patients is 30% higher than in healthy subjects [32,33]. For comparison of measured to predicted REE, a clinically acceptable discrepancy of 10% was defined a priori based on the assumption that measured REE would fall within ⫾10% of predicted REE [9]. Of the two power calculations, the sample for the paired analysis was larger. As such a minimum sample of 11 subjects per group was required to detect these clinically meaningful differences, assuming a standard deviation of 10%, with 90% power at the 95% significance level (two-tailed). Characteristics of cancer patients and healthy control subjects were compared by independent sample t tests for continuous variables and Fisher’s exact tests for categorical variables. Unadjusted REE between cancer patients and healthy subjects was assessed by independent sample t tests. A general linear modeling approach was taken for the regression analysis to adjust the association between measured REE and health status for differences in FFM between the two groups. Results are expressed as adjusted means ⫾ standard error. Analysis of REE data for body composition differences between groups was based on appropriate methods reported by other investigators [37,38]. Weight loss was also considered as a potential confounding variable; however, preliminary analysis indicated that weight loss was not associated with REE (adjusted for FFM) and therefore was not included in the final model. Tumor type and surgery were considered as potential effectmodifying variables through stratified analyses; however, the small sample precluded meaningful analyses. Measured REE was compared with REE predicted by the different prediction methods using the Bland-Altman approach [39]. Paired t tests were used to first assess agreement between the measured and predicted REEs at the group (average) level. Mean bias, limits of agreement (⫾2 standard deviations), and plot of bias against the average of measured and predicted REEs were used to describe agree-
Eighty-three eligible patients attended the radiation oncology center over the 6-mo period, 19 of whom consented to participate in the study. There was no significant difference between cancer patients who consented to participate and the total eligible pool with respect to gender, age, and tumor site (data not shown). One patient became ill after consent and was unable to undertake further data collection. As such, results are based on data from 18 cancer patients. Cancer patients were categorized into three groups according to tumor site: lung, which included non–small cell lung cancer and small cell lung cancer (n ⫽ 8); gastrointestinal tract (GIT), which included any tumors located in the GIT from the mouth to the anus (n ⫽ 7); and other, which included bladder, cervical, and testicular tumors (n ⫽ 3). Data on tumor stage were available for a limited number of patients. Three patients were diagnosed with pulmonary metastases, two patients were treated for tumor recurrence, and five patients had undergone surgical resection of the tumor. All but one patient had commenced radiotherapy treatment and 50% were undergoing concurrent chemotherapy at the time of the study. Seventeen healthy subjects participated in the study. Characteristics of subjects are presented in Table 1. There was no significant difference in gender, age, weight, height, body mass index, or FFM between groups. The nutritional status of cancer patients and healthy subjects differed significantly (Fisher’s exact test, P ⬍ 0.001), with the largest proportion of cancer patients categorized as moderately malnourished compared with healthy subjects who were well nourished. Although not statistically significant, the two groups differed in the amount of weight lost over the 6 mo before the study. All healthy subjects had stable weight, except two who had lost ⱖ5% body weight. Five cancer patients had stable weight or had gained weight over the previous 6 mo, six had lost ⬍5% body weight, and four had lost ⱖ5% body weight. Two cancer patients did not achieve steady state during REE measurement and one cancer patient was unable to have FFM assessed due to the presence of a pacemaker. Regression analyses are therefore based on data from 15 cancer patients and 17 healthy subjects. The average duration of REE measurements was a median of 9.5 min (range 5–23). Mean ⫾ standard deviation respiratory quotients for
612
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
Table 1 Subjects’ characteristics and REE*
Male:female Age (y) Height (cm) Weight (kg) BMI (kg/m2) FFM (kg) Body fat (%) Weight loss over previous 6 mo (%) SGA Well nourished (scale A) Moderately malnourished (scale B) Severely malnourished (scale C) Unadjusted REE (kJ/d) FFM-adjusted REE (kJ/d)
Cancer patients (n ⫽ 18)
Healthy subjects (n ⫽ 17)
11:7 65 ⫾ 13 167 ⫾ 14 80.2 ⫾ 19.8 28.4 ⫾ 5.0 51.9 ⫾ 10.9 33.7 ⫾ 10.5 2.2 ⫾ 4.6
10:7 60 ⫾ 11 169 ⫾ 9 75.0 ⫾ 13.0 26.3 ⫾ 4.1 52.2 ⫾ 11.3 30.5 ⫾ 8.8 0.9 ⫾ 2.6
4 (22%) 12 (67%)
17 (100%) 0 (0%)
2 (11%)
0 (0%)
6660 ⫾ 376† 6595 ⫾ 276†
5979 ⫾ 303‡ 6024 ⫾ 259‡
BMI, body mass index; FFM, fat-free mass; SGA, Subjective Global Assessment; REE, resting energy expenditure * Data are mean ⫾ standard error. † n ⫽ 15. ‡ n ⫽ 17.
the REE measurements were 0.70 ⫾ 0.09 for cancer patients and 0.72 ⫾ 0.08 for healthy subjects. Unadjusted and FFM-adjusted REEs for cancer patients and healthy subjects are presented in Table 1. There was no statistically or clinically significant difference between cancer patients and healthy subjects for unadjusted REE or FFM-adjusted REE. Adjusted REE was 9.5% (571 kJ) higher in cancer patients than in healthy subjects. Regression lines are shown in Figure 1. Further analyses indicated that FFM-adjusted REE differed significantly between tumor sites. Adjusted REE was similar between patients with lung cancer (mean ⫾ standard error, 6825 ⫾ 306 kJ) and those with GIT cancer (6584 ⫾ 331 kJ) but was considerably lower in those with “other” cancers (3774 ⫾ 736 kJ). FFM-adjusted REE excluding the three “other” patients increased the difference between cancer patients (lung and GIT) and healthy subjects to 12%, but the difference remained clinically and statistically insignificant. FFM-adjusted REE also differed between patients who had surgical removal of the tumor (7575 ⫾ 514 kJ) and those with the tumor in situ (6326 ⫾ 310 kJ). The sample of subjects who had undergone surgical removal was small (n ⫽ 4, one was excluded because steady state was not reached) and therefore did not permit further investigation of this group. The difference in FFM-adjusted REE between cancer patients and healthy subjects was further decreased (mean difference 5%) when the four subjects with surgical removal of the tumor were excluded from analysis. There was no difference in FFM-adjusted REE between patients who were receiving chemotherapy and those who were not. Measured REE in cancer patients and healthy subjects
was also compared with predicted REE from a number of commonly used prediction equations, to determine the individual predictive accuracy of the various methods. Predicted REE, mean bias, and limits of agreement for the cancer patients are listed in Table 2. In terms of mean bias, all prediction methods with the exception of the HarrisBenedict equation, with an injury factor of 1.3, were within clinically acceptable limits of ⫾10% of measured REE. The smallest mean bias was observed with the equations by Wang et al. [29], based on FFM. However, when data were analyzed for accuracy for individuals, each prediction method consistently produced wide limits of agreement (⬎20%), indicating potentially poor prediction for individual cancer patients. The Harris-Benedict (alone), the equations by Owen et al. [26,27] and Mifflin et al. [25], and the 20-kcal/kg method predicted REE within clinically acceptable limits for just over 50% of the sample of cancer patients. The remaining prediction methods failed to estimate REE within clinically acceptable limits for ⬎66% of individual cancer patients. For the group of cancer patients, a significant trend was observed in the spread of bias, with increasing REE values between measured REE and REE predicted by the equations of Cunningham et al. [24] (r ⫽ ⫺0.503, P ⫽ 0.056) and Wang et al. [29] (r ⫽ ⫺0.507, P ⫽ 0.054). Although not statistically significant, the bias observed with REE predicted from the equation of Schofield [28] showed a tendency toward underestimation of measured REE with increasing REE values (r ⫽ ⫺0.328, P ⫽ 0.215). No patterns were observed between the bias and mean REE value for any of the other prediction methods. For the group of healthy subjects, mean bias was within clinically acceptable limits for all prediction equations. However, the limits of agreement were wide (⫺30% to 40%), indicating poor prediction for individual healthy subjects.
Fig. 1. Plot and regression lines of FFM versus REE for cancer patients and healthy subjects. The interaction between REE and FFM was not significant (F1,31 ⫽ 0.336, P ⫽ 0.567). FFM, fat-free mass; REE, resting energy expenditure.
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
613
Table 2 Predicted REE, mean bias, and limits of agreement for difference between predicted and measured REE* of cancer patients (n ⫽ 16)
Harris-Benedict Harris-Benedict ⫻ 1.3 Schofield Owen Mifflin Cunningham§ Wang§ 20 kcal/kg
Predicted REE (kJ/d)
Bias (kJ/d)
Limits of agreement (⫾2 SD)†
Proportion within ⫾10%‡
6328 ⫾ 1368 8226 ⫾ 1778 6373 ⫾ 1131 6800 ⫾ 1214 6315 ⫾ 1427 6383 ⫾ 950 6516 ⫾ 946 6885 ⫾ 1635
⫺303 ⫾ 798 1595 ⫾ 980 ⫺258 ⫾ 904 169 ⫾ 955 ⫺315 ⫾ 882 ⫺276 ⫾ 1099 ⫺144 ⫾ 1099 254 ⫾ 1301
⫺28–20% 6–54% ⫺31–23% ⫺26–31% ⫺31–22% ⫺41–32% ⫺39–34% ⫺35–43%
50.0% 18.8% 31.3% 56.3% 56.3% 20.0% 26.7% 56.3%
REE, resting energy expenditure * Measured REE ⫽ 6631 ⫾ 1411 kJ/d. Data are mean ⫾ standard deviation. † Percentage of measured REE. ‡ Clinically acceptable limit (⫾10% of measured). § n ⫽ 15, measured REE ⫽ 6660 ⫾ 1455 kJ/d.
Discussion An understanding of whether metabolism and energy expenditure are altered in cancer patients is necessary for determining appropriate nutritional support recommendations for these patients. Most studies that have investigated energy expenditure in cancer patients have been conducted from a “purist” viewpoint. This is, they have attempted to isolate the effect of the tumor itself on energy expenditure. However, many cancer patients who require nutritional support are often undergoing anticancer therapies. Thus, this study was conducted from a “clinical practice” viewpoint. In addition, due to the limited ability for practitioners to measure patients’ energy expenditure in a clinical setting, there is a demand for easy tools for estimating individual patient requirements. Therefore, this study compared the measured REE of patients with cancer undergoing anticancer therapy with that of healthy control subjects and with REE estimated by a number of commonly used prediction methods. The sample of cancer patients recruited was heterogeneous in nature in terms of tumor site, stage, and type, current treatment, and previous weight loss. The group of healthy subjects was closely matched to the group of cancer patients in terms of FFM. Although the sampling frame for cancer patients and healthy subjects differed, it is primarily FFM that influences REE, our outcome measurement, and not other characteristics associated with a person’s environment. Recruitment of healthy subjects was based on characteristics matching cancer patients and self-reported health and exclusions were based only on the specified criteria. However, because ⬎50% of cancer patients were ⬎60 y of age, the likelihood of the presence of some health conditions in the control subjects is increased. In this study, when measured REE was adjusted for FFM, there was no difference between healthy subjects and cancer patients when defined by our a priori clinically acceptable limit of 30%. This definition of meaningful difference was based on common recommendations that energy expenditure in cancer patients is ⱖ30% higher than in
healthy counterparts (injury factor 1.3) [32,33]. This study found only a 10% difference in REE between the two groups, but the sample was not powered to detect this difference. Similar results have been observed in other studies. Staal-van den Brekel et al. [8] found a 9% higher FFMadjusted REE in patients with non–small cell lung cancer and a 17% higher FFM-adjusted REE in patients with small cell lung cancer compared with healthy control subjects. Jatoi et al. [7] observed an adjusted REE in patients with non–small cell lung cancer that was 10% to 12% higher than in healthy controls matched by gender, age, and body mass index. Our sample consisted of patients with solid tumors of different sites. Our data indicated that the “other” cancers, which included tumors of the bladder, cervix, and testes, in three patients appeared to have little effect on REE in this study. However, excluding these subjects from analysis had minimal effect on the results. Cancer patients also differed with respect to tumor stage and type (primary, recurrent, or metastatic). Studies have suggested that REE is not affected by differences in tumor stage [8], whereas tumor recurrence has shown slight increases in REE relative to initial primary diagnosis [40] and presence of metastatic disease has shown no to a small effect on REE in some studies [41,42] and a significant increase in REE in others [43]. It should be noted that several of these studies did not analyze energy expenditure data appropriately, so these results should be interpreted with caution. Because this study included patients who had commenced anticancer therapies, patients were included who had undergone surgical removal of the tumor. It has been suggested that metabolic alterations in patients with cancer are related to cytokine release from the tumor [3]. After surgical removal of the tumor and recovery from surgery, it would be expected that tumor effects on REE would no longer persist. Luketich et al. [44] and Arbeit et al. [41] observed decreases in REE after curative tumor resection, indicating that the tumor-bearing state affected energy me-
614
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615
tabolism. In our study, four patients who had undergone tumor resection had an unexpectedly higher FFM-adjusted REE than did those with the tumor in situ. These results can only be speculative and are subject to confounding because there may be other characteristics of these patients, such as tumor type or stage, that influence REE. However, these four patients underwent surgery 7 to 11 wk previously, so the influence of surgery on energy expenditure would have diminished [45]. Percentage of initial body weight loss over the previous 6 mo was not associated with REE in this study. Hansell et al. [5] found that the weight-losing state rather than the tumor-bearing state affected REE, but other investigators have not confirmed this result [6]. These conflicting results may be due to differences in the definition of weight losing or absolute amount of weight loss between studies. In our study only two patients (11%) had lost ⬎5% of initial body weight compared with 43% of patients with ⬎10% weight loss [5] and 56% with ⬎5% weight loss [6]. As a result of the small amount of weight loss observed in our study, we cannot make any judgments on the effect of weight loss on REE in patients with cancer. To be of use in a clinical setting, prediction equations should estimate REE within a reasonable degree of accuracy for individuals. Only one study has previously investigated the individual predictive accuracy of prediction equations in cancer patients and found that at the group level the prediction methods acceptably estimated REE of pancreatic cancer patients but were not as accurate for predicting REE for individual patients [13]. We observed similar results in our study. No prediction method estimated REE within acceptable limits for the full range of individual cancer patients. For this sample of cancer patients, the best combinations of smallest bias, narrowest limits of agreement, and largest proportion within clinically acceptable limits were observed with the HarrisBenedict (alone), Owen, and Mifflin equations. Automatic application of an injury factor of 1.3 with the HarrisBenedict equations consistently overestimated REE in this study. The results observed in the sample of healthy subjects were consistent with those previously reported [46,47]. In summary, this study found an overall 10% increase in REE in cancer patients receiving anticancer therapies compared with healthy control subjects, which is well below common recommendations of a 30% increase in REE in cancer patients. Further studies with larger samples are required to determine whether this 10% difference is significant. Current prediction methods do not predict individuals’ REE within clinically acceptable limits. Our results call into question automatic application of “injury factors” when predicting energy requirements for cancer patients from the Harris-Benedict equations. In a clinical setting, to ensure appropriate nutritional management of patients with cancer, intake and patient outcomes, such as weight and nutritional status, should be monitored regularly to determine whether patients’ energy requirements are being met.
Acknowledgements The authors thank the cancer patients and healthy subjects for willingly participating in the study, the doctors and nursing staff from Wesley Cancer Care Centre for assistance with patient recruitment, and staff at the Queensland Respiratory Laboratory for assistance with data collection.
References [1] Kern KA, Norton JA. Cancer cachexia. JPEN 1988;12:286 –98. [2] DeWys WD, Begg C, Lavin PT, Band PR, Bennett JM, Bertino JR, Cohen MH, et al. Prognostic effect of weight loss prior to chemotherapy in cancer patients. Am J Med 1980;69:491–7. [3] Fearon KCH, Barber MD, Moses AGW. The cancer cachexia syndrome. Surg Oncol Clin North Am 2001;10:109 –26. [4] Lindmark L, Bennegard K, Eden E, Ekman L, Schersten T, Svaninger G, Lundholm K. Resting energy expenditure in malnourished patients with and without cancer. Gastroenterology 1984;87:402– 8. [5] Hansell DT, Davies JW, Burns HJ. The relationship between resting energy expenditure and weight loss in benign and malignant disease. Ann Surg 1986;203:240 –5. [6] Fredrix EW, Soeters PB, Rouflart MJ, von Meyenfeldt MF, Saris WH. Resting energy expenditure in patients with newly detected gastric and colorectal cancers. Am J Clin Nutr 1991;53:1318 –22. [7] Jatoi A, Daly BD, Hughes VA, Dallal GE, Kehayias J, Roubenoff R. Do patients with nonmetastatic non–small cell lung cancer demonstrate altered resting energy expenditure? Ann Thorac Surg 2001;72: 348 –51. [8] Staal-van den Brekel AJ, Schols AM, Dentener MA, ten Velde GP, Buurman WA, Wouters EF. Metabolism in patients with small cell lung carcinoma compared with patients with non–small cell lung carcinoma and healthy controls. Thorax 1997;52:338 – 41. [9] Harris JA, Benedict FG. A biometric study of basal metabolism in man. Washington, DC: Carnegie Institution of Washington; 1919. [10] Bosaeus I, Daneryd P, Svanberg E, Lundholm K. Dietary intake and resting energy expenditure in relation to weight loss in unselected cancer patients. Int J Cancer 2001;93:380 –3. [11] Dempsey DT, Feurer ID, Knox LS, Crosby LO, Buzby GP, Mullen JL. Energy expenditure in malnourished gastrointestinal cancer patients. Cancer 1984;53:1265–73. [12] Knox LS, Crosby LO, Feurer ID, Buzby GP, Miller CL, Mullen JL. Energy expenditure in malnourished cancer patients. Ann Surg 1983; 197:152– 62. [13] Bauer J, Reeves MM, Capra S. The agreement between measured and predicted resting energy expenditure in patients with pancreatic cancer - a pilot study. JOP. J Pancreas (Online) 2004;5:32– 40. [14] Reeves MM, Capra S, Bauer J, Davies PSW, Battistutta D. Clinical accuracy of the MedGem™indirect calorimeter for measuring resting energy expenditure in cancer patients. Eur J Clin Nutr 2005;59: 603–10. [15] Demark-Wahnefried W, Peterson BL, Winer EP, Marks L, Aziz N, Marcom PK, et al. Changes in weight, body composition, and factors influencing energy balance among premenopausal breast cancer patients receiving adjuvant chemotherapy. J Clin Oncol 2001;19: 2381–9. [16] Leenders KL. PET: blood flow and oxygen consumption in brain tumors. J Neurooncol 1994;22:269 –73. [17] Platz EA. Energy imbalance and prostate cancer. J Nutr 2002;132: 3471S– 81. [18] Reeves MM, Davies PSW, Bauer J, Battistutta D. Reducing the time period of steady state does not affect the accuracy of energy expenditure measurements by indirect calorimetry. J Appl Physiol 2004; 97:130 – 4.
M. M. Reeves et al. / Nutrition 22 (2006) 609 – 615 [19] Weir JBDeV. New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol 1949;109:1–9. [20] Isenring E, Bauer J, Capra S, Davies PSW. Evaluation of foot-to-foot bioelectrical impedance analysis for the prediction of total body water in oncology outpatients receiving radiotherapy. Eur J Clin Nutr 2004; 58:46 –51. [21] Hoffer EC, Meador CK, Simpson DC. Correlation of whole-body impedance with total body water volume. J Appl Physiol 1969;27: 531– 4. [22] Bauer J. New insights into medical nutrition therapy for cancer cachexia. Centre for Health Research. Brisbane: Queensland University of Technology; 2003. [23] Detsky AS, McLaughlin JR, Baker JP, Johnston N, Whittaker S, Mendelson RA, Jeejeebhoy KN. What is subjective global assessment of nutritional status? JPEN 1987;11:8 –13. [24] Cunningham JJ. Body composition as a determinant of energy expenditure: a synthetic review and a proposed general prediction equation. Am J Clin Nutr 1991;54:963–9. [25] Mifflin MD, St Jeor ST, Hill LA, Scott BJ, Daugherty SA, Koh YO. A new predictive equation for resting energy expenditure in healthy individuals. Am J Clin Nutr 1990;51:241–7. [26] Owen OE, Holup JL, D’Alessio DA, Craig ES, Polansky M, Smalley KJ, et al. A reappraisal of the caloric requirements of men. Am J Clin Nutr 1987;46:875– 85. [27] Owen OE, Kavle E, Owen RS, Polansky M, Caprio S, Mozzoli MA, et al. A reappraisal of caloric requirements in healthy women. Am J Clin Nutr 1986;44:1–19. [28] Schofield WN. Predicting basal metabolic rate, new standards and review of previous work. Hum Nutr Clin Nutr 1985;39:5– 41. [29] Wang Z, Heshka S, Gallagher D, Boozer CN, Kotler DP, Heymsfield SB. Resting energy expenditure–fat-free mass relationship: new insights provided by body composition modeling. Am J Physiol Endocrinol Metab 2000;279:E539 – 45. [30] Bloch AS. Cancer. In: Matarese LE, Gottschlich MM, editors. Contemporary nutrition support practice: a clinical guide. Philadelphia: WB Saunders; 1998. [31] A.S.P.E.N. Board of Directors and the Clinical Guidelines Task Force. Guidelines for the use of parenteral and enteral nutrition in adult and pediatric patients. JPEN 2002;26:1–138. [32] Alpers D, Stenson W, Bier D. Manual of nutritional therapeutics. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2002. [33] Dietitians’ pocket book. Perth: Curtin University of Technology; 1999.
615
[34] Barak N, Wall-Alonso E, Sitrin MD. Evaluation of stress factors and body weight adjustments currently used to estimate energy expenditure in hospitalized patients. JPEN 2002;26:231– 8. [35] Glynn CC, Greene GW, Winkler MF, Albina JE. Predictive versus measured energy expenditure using limits-of-agreement analysis in hospitalized, obese patients. JPEN 1999;23:147–54. [36] Heymsfield SB. Heat and life: the ongoing scientific odyssey. JPEN 2002;26:319 –32. [37] Davies PSW, Cole TJ. The adjustment of measures of energy expenditure for body weight and body composition. Int J Body Comp Res 2003;1:45–50. [38] Ravussin E, Bogardus C. Relationship of genetics, age, and physical fitness to daily energy expenditure and fuel utilization. Am J Clin Nutr 1989;49:968 –75. [39] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1: 307–10. [40] Fredrix EW, Soeters PB, Wouters EF, Deerenberg IM, von Meyenfeldt MF, Saris WH. Effect of different tumor types on resting energy expenditure. Cancer Res 1991;51:6138 – 41. [41] Arbeit JM, Lees DE, Corsey R, Brennan MF. Resting energy expenditure in controls and cancer patients with localized and diffuse disease. Ann Surg 1984;199:292– 8. [42] Hansell DT, Davies JW, Burns HJ. Effects of hepatic metastases on resting energy expenditure in patients with colorectal cancer. Br J Surg 1986;73:659 – 62. [43] Macfie J, Burkinshaw L, Oxby C, Holmfield JH, Hill GL. The effect of gastrointestinal malignancy on resting metabolic expenditure. Br J Surg 1982;69:443– 6. [44] Luketich JD, Mullen JL, Feurer ID, Sternlieb J, Fried RC. Ablation of abnormal energy expenditure by curative tumor resection. Arch Surg 1990;125:337– 41. [45] Duke JH, Jorgensen SB, Broell JR, Long CL, Kinney JM. Contribution of protein to caloric expenditure following injury. Surgery 1970; 68:168 –74. [46] Taaffe DR, Thompson J, Butterfield G, Marcus R. Accuracy of equations to predict basal metabolic rate in older women. J Am Diet Assoc 1995;95:1387–92. [47] Siervo M, Boschi V, Falconi C. Which REE prediction equation should we use in normal-weight, overweight and obese women? Clin Nutr 2003;22:193–204.