Assessment of nutrition intervention for patients with unresectable pancreatic cancer in a fish oil supplement trial – does it make a difference?
Wendy Louise Davidson BSc Qld, GradDipNutDiet QUT
Centre for Health Research School of Public Health Queensland University of Technology
submitted for Master in Applied Science (Research)
2003
ii
Keywords
Nutrition, carcinoma of the pancreas, pancreatic neoplasms, weight loss, cachexia, quality of life, dietetics, prognosis, survival
iii
Abstract
Severe and progressive weight loss is a feature of pancreatic cancer but it has been unclear whether dietetic intervention can improve patient outcomes. This study is a post hoc analysis of the extensive data available from the multicentre BH80 Cancer Cachexia trial to examine how patients with unresectable pancreatic cancer respond to intensive dietetic intervention. The BH80 study compared an n-3 fatty acid enriched oral supplement with an isonitrogenous, isocaloric supplement given over an eight week period. Additional qualitative data was also collected and examined for the patients recruited by the Australian sites.
The aims were to determine whether achieving weight stabilisation is an appropriate goal of nutrition intervention for people with unresectable pancreatic cancer; to identify determinants of weight stabilisation; and to describe nutrition-related features of patients recruited by the Australian sites, prior to and during intensive nutrition intervention.
Data from 107 patients for whom weight change data over an eight week nutrition intervention period was available, was divided into weight losing (> 1kg lost) and weight stable (≤ 1 kg lost) patients. Group survival duration (Kaplan Meier log rank test) and global quality of life (EORTC QLQ-C30 global health status/quality of life) were compared. Variables including energy intake, BMI, presence of pain, nausea and vomiting, and appetite loss, C reactive protein and stage of disease were also compared to determine predictors of weight stability using logistic regression analysis.
This study demonstrates that weight stabilisation is not only achievable for some patients in the short term, but it is also associated with improved outcomes. Those patients who were able to stabilise their weight after eight weeks of oral nutrition support lived longer from baseline and reported better quality of life than those who continued to lose weight. Weight stabilisation is therefore a reasonable and worthwhile goal for this patient group.
iv
Patients who halted their weight loss had greater energy and protein intakes than those who continued to lose weight. They also had lower BMI’s and were less likely to have pain, appetite loss or an acute phase protein response at baseline. Female gender and the absence of nausea or vomiting at baseline, were independently associated with an increased likelihood of weight stabilisation.
This study also confirms reports that patients with unresectable pancreatic cancer experience a wide range of symptoms that impact negatively on their nutritional status. Common issues include loss of appetite, pain, nausea and vomiting, taste alterations, pancreatic insufficiency, fatigue and depression. Nutrition intervention needs to be individualised and intensive to adjust to disease progression and to address the patient specific barriers to intake. Further research is recommended such as randomised studies of oral nutrition support versus standard care with the goal of weight stabilisation, methods for predicting survival time and whether optimal management of nausea and vomiting improves outcomes in patients with advanced cancer.
Table of Contents Keywords
ii
Abstract
iii
Table of Contents
v
List of Tables
ix
List of Figures
x
List of Abbreviations
xii
Acknowledgments
xv
1 1.1
INTRODUCTION Overall aim
1 1
1.2 Specific aims 1.2.1. Aim 1 1.2.2. Aim 2 1.2.3. Aim 3
3 3 3 4
2
5
LITERATURE REVIEW
2.1 Pancreatic cancer 2.1.1 Incidence 2.1.2. Risk factors 2.1.3. Symptoms and features 2.1.4. Prognostic features 2.1.5. Treatment
5 5 6 7 10 11
2.2 Cancer associated weight loss 2.2.1. Prevalence of weight loss in cancer 2.2.2. Effects of weight loss on patient outcomes 2.2.3. Metabolic factors affecting weight loss 2.2.4. Resting energy expenditure 2.2.5. Acute phase protein response 2.2.6. Proinflammatory cytokines 2.2.7. Catabolic factors 2.2.8. Pancreatic insufficiency
12 12 14 15 17 19 20 21 21
2.3 Causes of reduced dietary intake 2.3.1. Appetite and taste changes 2.3.2. Gastrointestinal symptoms 2.3.3. Pain and depression
22 26 29 31
2.4 Effect of nutrition support 2.4.1. Studies of nutritional support in cancer 2.4.2. Research on oral nutrition support
32 32 35
2.5 Evidence-based goals for nutrition support 2.5.1. Current dietetic practice in oncology 2.5.2. Increased survival time 2.5.3. Improved quality of life
37 37 40 40
vi 2.5.4. 2.5.5. 2.5.6. 2.6
3
Nutritional requirements Body weight & lean body mass Functional improvements Summary
METHODS – AIMS 1 &2
43 44 45 46
47
3.1 BH80 trial 3.1.1. BH80 study design 3.1.2. Subjects 3.1.3. Ethical issues 3.1.4. Anthropometry 3.1.5. Nutrient intake 3.1.6 Biochemical analysis 3.1.7. Quality of Life 3.1.8. Appetite & Satiety 3.1.9. Functional status 3.1.10. Survival duration
47 47 49 50 50 51 52 52 53 54 54
3.2 3.2.1 3.2.2 3.2.3 3.2.4
Preparation of data for secondary analysis Rationale for pooling subjects Creation of weight losing or weight stable groups Inclusion criteria Confirming normal distribution
54 54 56 56 57
3.3 3.3.1 3.3.2
Data analysis Outcomes of dietetic intervention Determinants of weight stabilisation
57 57 59
4
RESULTS – AIMS 1 & 2
67
4.1 4.1.1 4.1.2 4.1.3
Description of patients excluded from the study Reasons for excluding patients from the study Comparison of included & excluded patients Description of included patients
67 67 68 70
4.2 4.2.1. 4.2.2.
Outcomes of nutrition intervention Survival Duration Quality of Life
71 71 73
4.3 4.3.1. 4.3.2.
Determinants of weight stabilisation Comparison of the WL and WS groups Determinants of weight stability - multivariate analysis
73 75 77
4.4
Summary
79
5
METHODS - AIM 3
81
5.1
Comparison of qualitative and quantitative methods
81
5.2
Subjects
84
5.3 5.3.1. 5.3.2. 5.3.3.
Data from the BH80 study Accuracy of dietary recording methods Nutrient intake Anthropometry
84 85 86 87
vii 5.3.4. 5.3.5. 5.3.6. 5.3.7. 5.4 5.4.1. 5.4.2. 5.4.3. 5.4.4.
6
Quality of life questionnaire Appetite Functional status Statistical methods Sub-protocol Ethical issues Structured Interview Recent Food Intake Report Semi-structured Interview (Narrative)
RESULTS – AIM 3
88 88 88 88 89 89 90 90 90
93
6.1 6.1.1.
Description of Australian participants Comparison with BH80 participants.
93 94
6.2
Body weight changes
95
6.3 6.3.1. 6.3.2 6.3.3.
Changes in dietary intake Energy intake. Macronutrients Protein intake
98 98 103 103
6.4 6.4.1. 6.4.2.
Nutrition supplements Supplement intake The role of supplements
105 105 106
6.5 6.5.1. 6.5.2. 6.5.3.
Changes in appetite and taste perceptions Appetite changes Early satiety Taste changes
107 107 108 109
6.6 6.6.1. 6.6.2. 6.6.3. 6.6.4. 6.6.5.
Symptoms and quality of life Pain Nausea & vomiting Asthenia Constipation/diarrhoea Xerostomia
110 111 111 111 112 112
6.7
Summary
113
6.8 6.8.1. 6.8.2.
Case Studies Case 1 - Mrs A Case 2 - Mr B
113 113 115
7
DISCUSSION
119
7.1
Study design issues
119
7.2 7.2.1. 7.2.2. 7.2.3.
Outcomes for weight stabilisation Survival duration Quality of life Summary - Aim 1
121 122 123 123
7.3 7.3.1. 7.3.2. 7.3.3.
Determinants of weight stabilisation Independent determinants Determinants significant by bivariate analysis Characteristics which were not predictive
124 125 128 134
viii 7.3.4.
Summary - Aim 2
137
7.4 7.4.1. 7.4.2. 7.4.3. 7.4.4. 7.4.5.
Additional nutrition related factors Bodyweight changes Nutrient intake Taste & appetite changes Symptoms and quality of life Summary - Aim 3
139 139 140 141 142 143
7.5 7.5.1. 7.5.2.
Implications for nutrition intervention Appropriate level of nutrition intervention Individualised and intensive intervention
143 144 146
8
CONCLUSION
APPENDICES
151 155
Appendix 1: BH80 study sites
155
Appendix 2: BH80 Inclusion/Exclusion Criteria
156
Appendix 3: BH80 patient information and consent
158
Appendix 4: EORTC QLQ-C30 (version 3)
162
Appendix 5: Karnofsky Performance Status Scale
164
Appendix 6: Staging for Pancreatic Cancer
165
Appendix 7: Sub-protocol patient information and consent (ver 2)
166
Appendix 8: Identification coding of patients from the Australian sites
169
Appendix 9: T=0 structured interview (ver 1)
170
Appendix 10: Recent food intake report (ver 1)
172
Appendix 11: T=9/10 semi-structured interview/narrative
173
Appendix 12: Notation used in transcription of interviews
175
Appendix 13: Published paper (in press) - Clinical Nutrition
176
Appendix 14: Oral presentation – DAA 21st National Conference
185
Appendix 15: Poster presentation – DAA 20th National Conference
186
Appendix 16: Descriptive statistics for study participants including data to test for normality187 Appendix 17: Splett’s cascade model of outcome measures
REFERENCES
188
189
ix
List of Tables Table 2.1 Prevalence of symptoms reported by patients with pancreatic cancer referred to a palliative care service Table 2.2
Categories of degree of weight loss
8 14
Table 2.3 Summary of studies describing metabolic changes in pancreatic and other gastrointestinal cancers. 18 Table 2.4 Summary of cross-sectional studies describing the dietary protein and energy intake of patients with cancer, as well as baseline intakes from intervention studies. 24 Table 2.5 Factors proposed as contributing to cancer anorexia
26
Table 2.6 Summary of studies of oral nutrition support in cancer
33
Table 2.7 Summary of nutrition-intervention studies in pancreatic cancer
36
Table 2.8 Recommendations regarding energy requirements for cancer patients.
42
Table 2.9 Recommendations regarding protein requirements for cancer patients.
44
Table 3.1 Outcome variables of the BH80 study.
47
Table 3.2 Major components of BH80 study supplements
48
Table 3.3 Schedule of data collection in BH80 study
49
Table 3.4 Description of variables used for assessing outcomes for weight stabilisation in patients with unresectable pancreatic cancer receiving nutrition intervention over eight weeks (Aim 1) 59 Table 3.5 Description of baseline variables used to identify determinants of weight stability in patients with unresectable pancreatic cancer receiving nutrition intervention over eight weeks (Aim 2) 61 Table 3.6 Description of Week 8 variables used to identify determinants of weight stability in patients with unresectable pancreatic cancer receiving nutrition intervention over eight weeks (Aim 2) 62 Table 3.7 Plasma EPA as a percentage of total phospholipids, with and without fish oil or EPA supplementation.
65
Table 4.1 Comparison of baseline characteristics of patients with unresectable pancreatic cancer who were included in analysis with those who were excluded (continuous variables). 68 Table 4.2 Comparison of baseline characteristics of patients with unresectable pancreatic cancer who were included in analysis with those who were excluded (categorical variables). 69 Table 4.3 Comparison of baseline characteristics of 107 weight losing (WL) and weight stable (WS) patients with unresectable pancreatic cancer receiving eight weeks of nutrition intervention. 74 Table 4.4 Comparison of self-reported pre-illness body mass index (BMI) for weight losing (WL) and weight stable (WS) pancreatic cancer patients 75 Table 4.5 Comparison of plasma eicosapentanoic acid (EPA) distributions for weight losing and weight stable pancreatic cancer patients following eight weeks of nutrition intervention 76
x
Table 4.6 Comparison of characteristics of weight losing (WL) and weight stable (WS) pancreatic cancer patients at baseline and week 8 76 Table 4.7 Comparison of energy intake distributions at week 8 for weight losing and weight stable pancreatic cancer patients 77 Table 4.8 The effect of different variables on the likelihood of patients with pancreatic cancer being weight stable following eight weeks of nutrition intervention (n=83) – logistic regression analysis. 78 Table 5.1 Comparison of quantitative and qualitative research methods
82
Table 6.1 Comparison of baseline characteristics of patients with unresectable pancreatic cancer from the Australian sites versus the combined sites. 95 Table 6.2 Changes in body weight (kg) over time, of patients with unresectable pancreatic cancer recruited into the BH80 study from Australian sites 97 Table 6.3. Macronutrient contributions to total energy for patients with unresectable pancreatic cancer recruited into the BH80 study from Australian sites. 102 Table 6.4 Total protein intake for patients with unresectable pancreatic cancer recruited into the BH80 study from Australian sites 104 Table 6.5 Self-reported lack of appetite scores for patients with unresectable pancreatic cancer at the Australian sites immediately prior to baseline, week 4 and week 8 assessments over eight weeks of nutrition intervention 108
List of Figures Figure 1.1
Model representing the three aims of this study
2
Figure 2.1
Frequency and degree of weight loss in patients with advanced cancer
13
Figure 2.2
Otterey’s algorithm of optimal nutritional intervention
38
Figure 3.1
Model representing the first two aims of this study of nutrition intervention in patients with unresectable pancreatic cancer
55
Figure 4.1
Reasons for exclusion of some BH80 participants from secondary analysis
67
Figure 4.2
Model representing the first aim of this study of nutrition intervention in patients with unresectable pancreatic cancer 71
Figure 4.3
Comparison of survival time from baseline for weight losing (n=44) and weight stable pancreatic cancer patients – Kaplan Meier survival plot 72
Figure 4.4
Model representing the second aim of this study of nutrition intervention in patients with unresectable pancreatic cancer 73
Figure 5.1
Model representing Aim 3
81
Figure 6.1
Percentage of pre-illness body weight lost by each Australian patient with unresectable pancreatic cancer until final weight measurement
96
xi Figure 6.2
Change in energy intake between baseline and week 4 for patients with unresectable pancreatic cancer recruited into the BH80 study from Australian sites. 99
Figure 6.3
Energy intake at baseline (T0) and week 4 (T4) for patients with unresectable pancreatic cancer recruited into the BH80 study from Australian sites who survived less than ten weeks from baseline (Group 1) 100
Figure 6.4
Energy intake at baseline, week 4 and week 8 for patients with unresectable pancreatic cancer recruited into the BH80 study from Australian sites who survived more than ten weeks from baseline (Group 2). 101
Figure 7.1
Model representing Aims 1 & 2 of this study of nutrition intervention in unresectable pancreatic cancer
Figure 7.2
138
Clinical pathway for medical nutrition therapy in unresectable pancreatic cancer 148
xii
List of Abbreviations
APPR
Acute phase-protein response
BIA
Bioelectrical impedance analysis
BMI
Body Mass Index
BW
Body weight
CI
Confidence interval
CRP
C reactive protein
CT
Chemotherapy
DV
Dependent variable
ECOG
Eastern Cooperative Oncology Group
EI/BMR
Ratio of energy intake to basal metabolic rate
EORTC
European Organisation for Research and Treatment of Cancer
EORTC QLQ-C30
EORTC quality of life questionnaire
EPA
Eicosapentaenoic acid
EQ5D
EuroQol quality of life questionnaire
Global QoL
EORTC QLQ-C30 Global health status/QoL scale
IL-1
Interleukin-1
IL-6
Interleukin-6
INFγ
Interferon gamma
IV
Independent variable
KPS
Karnofsky Performance Status
LAS
Linear analogue scale
LBM
Lean body mass
LMF
Lipid mobilising factor
ONS
Oral nutrition support
OR
Odds ratio
PAH
Princess Alexandra Hospital
PC
Pancreatic cancer
PG-SGA
Patient Generated – Subjective Global Assessment
PIF
Proteolysis inducing factor
xiii Pt
Patient
QoL
Quality of life
RCT
Randomised control trial
REE
Resting energy expenditure
RT
Radiotherapy
T0
Baseline data collection time point
T4
Week 4 data collection time point
T8
Week 8 data collection time point
TBW
Total body water
TNF
Tumour necrosis factor
TWH
The Wesley Hospital
WL
Weight losing (loss of >1kg over eight weeks)
WS
Weight stable (loss of ≤ 1kg over eight weeks)
xiv
The work contained in this thesis has not been previously submitted for a degree or diploma at any other higher education institution. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due reference is made. Signed: _______________________________
Dated: ________________________________
xv
Acknowledgments To the people with pancreatic cancer and their carers who gave so generously of their time and to the Cancer Cachexia Study Group which made access to this data possible. To my supervisors, Dr Susan Ash and Prof Sandra Capra, whose enthusiasm, insight and dedication to the advancement of nutrition and dietetics are inspirational. To Dr Diana Battistutta for her valuable statistical advice and to the members of the Nutrition Practice Research Group. To my husband, Gary, my family, Jacob, Elliott, Dylan, Tegan and Paavi, and to my parents, Barbara and Collin Davidson, for their love and support throughout the process.
1 Introduction Severe and progressive weight loss is a feature of adenocarcinoma of the pancreas and this wasting process is distressing to the patient and their family. It is unclear whether dietetic intervention can improve patient outcomes. There are no known cures currently available for patients with unresectable pancreatic cancer and expected survival time is short.
Medical management focuses on palliative measures, such as the reduction of pain, to improve the patient’s quality of life. Nutrition therapy for advanced cachexia has been considered by some doctors to be futile – an ineffective and possibly unethical use of resources. Is dietary intervention appropriate in the palliative care setting for patients who are losing weight, especially if we have no reliable method for helping patients to restore that lost weight?
A rare opportunity was available through the multicentre BH80 Cancer Cachexia trial, to examine how patients with unresectable pancreatic cancer respond to intensive dietetic intervention. The trial was controlled to determine the effect of adding fish oil to a protein and energy-dense oral supplement, therefore both arms of the study received a form of dietary intervention. All patients were initially losing weight, as a criteria for entrance to the trial, and were expected to experience ongoing weight loss - the usual disease process.
This study is a post hoc analysis of the extensive data available from the BH80 trial, pooling both arms of the trial to analyse outcomes for this large, relatively homogeneous group of patients with an advanced cancer. More detailed data has also been collected and examined for the patients recruited by the Australian sites.
1.1.
Overall aim
To determine appropriate goals for effective nutrition intervention for weight losing patients with unresectable pancreatic cancer, which would assist dietitian nutritionists in providing appropriate clinical services (Fig 1.1).
2
Introduction
Figure 1.1
Model representing the three aims of this study
Aim 1 QoL Weight change Survival
Aim 2 ONS
Symptoms T0
Energy intake T0 Energy intake T8
BMI T0
Weight change
PS T0
Stage of disease
Aim 3 pain, n&v, app,
Patient’s
taste, depression
perceptions
ONS
Symptoms T0
QoL E, Pr, F,
Intake T0
CHO
E, Pr, F, BMI T0
Intake T8
beyond T8
change
CHO food texture
Survival
PS T0
Stage of disease
Weight
Psychosocial factors
Key to symbols: independent variable
intervention
dependent variable
ONS – oral nutrition support, QoL – quality of life, BMI – body mass index, PS – performance status, n&v – nausea & vomiting, app – appetite, E – energy, Pr – protein, F – fat, CHO – carbohydrate, T0 – baseline, T8 – week 8.
Introduction
1.2.
1.1.1
3
Specific aims
Aim 1
To determine whether achieving weight stabilisation is an appropriate goal of nutrition intervention for people with unresectable pancreatic cancer.
Hypotheses
a)
Patients who lose no more than 1kg over eight weeks of nutrition intervention will have increased survival time compared to those who lose more than 1kg.
b)
Patients who lose no more than 1kg over eight weeks of nutrition intervention will have better quality of life than those who lose more than 1kg.
If these hypotheses are supported then weight stabilisation could be considered a measure of “effective nutrition intervention” and therefore an appropriate goal for medical nutrition therapy for this patient group.
1.1.2
Aim 2
To identify determinants of weight stabilisation in patients with unresectable pancreatic cancer.
Hypotheses
a)
Patients who would be likely to benefit from nutrition intervention, (ie. weight stabilisation is feasible), can be identified using a variety of features on presentation – higher BMI, greater energy intake, better performance status, less pain, less loss of appetite and less nausea and vomiting.
b)
Patients who would be likely to benefit from nutrition intervention, (ie. weight stabilisation is feasible), will be those who can achieve greater energy intake by week eight.
4
Introduction
1.1.3
Aim 3
To describe nutrition-related features of patients with unresectable pancreatic cancer recruited by the Australian sites, prior to and during intensive nutrition intervention.
a)
To describe changes to body weight, dietary intake, taste perceptions and appetite and the role of dietary supplements.
b)
To explore patients’ perceptions regarding changes in weight, appetite and quality of life through case studies.
This section uses both quantitative and qualitative data in order to provide a clearer picture of the illness as experienced by the patient, to assist in providing patientfocussed dietetic care.
The methods and results for Aims 1 and 2 (Chapters 3 & 4) will be considered separately from those for Aim 3 (Chapters 5 & 6).
2 Literature Review 2.1
Pancreatic cancer
Pancreatic cancer is a particularly challenging disease for dietitian nutritionists because of its severe impact on nutritional status. The pancreas is the site of production of digestive enzymes and disease often spreads to adjacent parts of the gastrointestinal tract. Weight loss is severe and patients have a range of nutrition related symptoms.
Most cases of pancreatic cancer are not suitable for potentially curative surgery (Lillemoe, Yeo, & Cameron, 2000), therefore the focus of care is palliation – to minimise symptoms and maximise quality of life.
2.1.1
Incidence
Pancreatic cancer is the fourth leading cause of cancer mortality in the USA with more than 28,000 deaths/year (Greenlee, Hill-Harmon, Murray, & Thun, 2001). In Australia it is the fourth most common cause of cancer death for women and fifth most common cause for men. 1788 Australians (916 men and 872 women) were diagnosed as having pancreatic cancer in 1999 (Australian Institute of Health and Welfare, 2002).
Exocrine tumours account for 95% pancreatic cancers, the majority of which are adenocarcinomas. The remaining 5% are islet cell tumours, most of which are benign.
Crude incidence rates have been increasing for both men and women in Australia. For both sexes the crude rates were 7.3 cases diagnosed per 100,000 population in 1983 and 9.4 per 100,000 in 1999 (Australian Institute of Health and Welfare, 2002). This increase is likely to be due largely to the ageing population because the age adjusted rates for Australians have been relatively stable.
6
Literature Review
2.1.2
Risk factors
The incidence of pancreatic cancer rises steadily with age, with 84% of Australian cases in 1999 occurring in people over 60 years of age. The disease is uncommon in people younger than 40 years (Australian Institute of Health and Welfare, 2002; Greenlee et al., 2001; Lillemoe et al., 2000). Men are more likely to develop cancer of the pancreas than women. Lillemoe has reported a narrowing of gender differences in incidence rates in the USA over recent years (Lillemoe et al., 2000). Australian age standardised data (Australian Institute of Health and Welfare, 2002) show a slight increase in incidence for females in the 1980’s but this appears to have steadied in the 1990’s.
Inherited genetic alterations may be a factor in 5 – 10% cases of pancreatic cancer (Howe & Conlon, 1997). Occupational exposure to certain chemicals may also increase the risk (Hoppin et al., 2000).
Chronic pancreatitis is associated with an increased risk of pancreatic cancer, however this may be due to a failure to adjust for common risk factors such as smoking. Evidence linking diabetes mellitus and pancreatic cancer is also inconsistent (Lillemoe et al., 2000; Silverman et al., 1999). Diabetes may develop as an early symptom of pancreatic cancer rather than being a causal factor.
The best established risk factor for pancreatic cancer is cigarette smoking with about 30% cases estimated to be due to smoking. This increases the risk of developing pancreatic cancer with odds ratios for current smokers between 1.3 and 5.5 (Howe & Burch, 1996; Silverman et al., 1994). Between 1989 and 1995, the male incidence rate for smoking-related cancers in Australia fell by an average of 0.9% per year, while the rate for females rose by 1.3% per year [Australian Institute of Health and Welfare, 1998 #289].
The evidence for other dietary or lifestyle risk factors is less clear. Although increased risk of pancreatic cancer has been associated with consumption of alcohol and coffee, neither were associated with an increased risk based on food frequency questionnaires from the Health Professionals Follow-Up Study and the Nurses
Literature Review
7
Health Study (Michaud, Giovannucci, Willett, Colditz, & Fuchs, 2001a). No association with coffee consumption was found in a case-control study of 436 cases (Silverman et al., 1998). Folate appears to be protective (Kim, 1999; StolzenbergSolomon et al., 1999).
Positive relationships have been shown between body mass index and pancreatic cancer (Michaud et al., 2001b) (Silverman et al., 1998). In the study by Michaud, obese individuals were 72% more likely to be diagnosed with pancreatic cancer than those who were neither obese nor overweight, however physical activity had a protective effect for overweight and obese people. These findings have led to suggestions that chronic hyperinsulinaemia may be involved (Kaaks & Lukanova, 2001).
2.1.3
Symptoms and features
Most patients with pancreatic cancer experience multiple symptoms (Krech & Walsh, 1991; Lillemoe, 1998; Ottery, 1996a). Krech et al reported an average of 11 symptoms per patient in their study of 39 people with pancreatic cancer referred to a palliative care service at the Cleveland Clinic (Krech & Walsh, 1991). The percentage of patients reporting various symptoms is listed in Table 2.1. The following physical features were also present; 44% patients were cachectic on presentation, 33% had serum albumin less than 3.5mg/dL, palpable abdominal mass in 36%, ascites 23%, jaundice 21% and metastasis to at least one major organ was documented in 64% of patients.
Pain, being easily fatigued, and anorexia are commonly found in advanced cancer. They were consistently among the ten most prevalent and clinically important symptoms in a study of the prevalence and severity of symptoms in 1000 patients regardless of tumour site (Donnelly, Walsh, & Rybicki, 1995).
Signs and symptoms that accompany pancreatic cancer are particularly likely to affect nutrient intake or utilisation due to the location of the tumour. Many symptoms relate to compression or invasion of related physical structures, eg. approximately 50% of people with pancreatic cancer will experience jaundice during the course of
8
Literature Review
the disease. This is usually resolved by biliary bypass surgery or endoscopic insertion of a biliary stent. Duodenal obstruction with nausea and vomiting, is usually a late manifestation of a tumour in the head of the pancreas. Liver metastases and retroperitoneal implants are the most common sites of distant spread, therefore ascites and oedema frequently occur in the final weeks of life (Sauter & Coleman, 1999).
Table 2.1
Prevalence of symptoms reported by patients with pancreatic cancer referred to a palliative care service (n=39). Symptom
Percentage of patients
pain
82%
anorexia
64%
early satiety
62%
xerostomia
54%
sleep problems
54%
weight loss
51%
easily fatigued
46%
weakness
41%
nausea
41%
constipation
41%
depression
39%
dyspepsia
36%
vomiting
31%
hoarse
26%
taste changes
23%
bloating
23%
belching
23%
dyspnoea
21%
dizzy
21%
oedema
21%
cough
18%
diarrhoea
15%
hiccoughs
15%
itch
15%
dysphagia
3%
Source: Krech 1991(Krech & Walsh, 1991)
Literature Review
9
Pancreatic cancer may also cause early satiety - a sense of feeling too full to continue eating that occurs much sooner than the individual would normally expect (Gallagher-Allred, 1989). Delayed gastric emptying, with no mechanical obstruction, is also common in advanced cancer (Ottery, 1996a) and could contribute to early satiety.
Pancreatic insufficiency is reported to be a common problem (Ottery, 1996a), however there is little objective evidence in the literature. Reports on the prevalence of diabetes conflict, ranging from 15-20% (Sauter & Coleman, 1999) to 70-80% of patients (Permert et al., 1993).
Anecdotal reports of taste alterations such as a metallic taste to food as well as an aversion to the smell of food during preparation (Ottery, 1996a), and aversions to “proteinaceous foods such as beef and pork” (Gallagher-Allred, 1989) can be found. There are no reported studies however investigating in detail the taste changes that occur in pancreatic cancer.
Loss of weight has been reported in more detail than the other features of unresectable pancreatic cancer with general agreement that weight loss is common, severe and progressive. Wigmore et al observed changes in the nutritional status of 20 patients with unresectable pancreatic cancer from diagnosis to death (Wigmore, Plester, Richardson, & Fearon, 1997b). Eighty-five percent of patients had lost weight at the time of diagnosis, and all had lost weight as death approached, with median BMI dropping from 24.9 kg/m2 premorbidly to 20.7 kg/m2 at diagnosis and 17.7 kg/m2 near death. Patients had a median weight loss of 14.2% at diagnosis (1020% IQR) increasing to 24.5% near death (11.5-29.7% IQR).
Marked and progressive weight loss was also demonstrated in a study of 102 patients with unresectable pancreatic cancer (Falconer et al., 1995). Median weight loss was 11% at diagnosis and 20% within six weeks of death.
In summary, some features of pancreatic cancer, such as the marked and progressive weight loss, have been described in detail, but for others there is little agreement or evidence in the literature.
10 Literature Review
2.1.4
Prognostic features
Unfortunately the nonspecific nature of early symptoms, including anorexia, weight loss, abdominal discomfort and nausea, may contribute to a delay in diagnosis and therefore make curative treatment options less likely (Lillemoe et al., 2000).
The prognosis for people with pancreatic cancer is known to be poor, but it is difficult even for experienced clinicians to predict individual survival times. The ability to better predict survival would assist in clinical decision making.
The five year relative survival rates for pancreatic cancer over the past three decades (adjusted for normal life expectancy) have been 3-4% (Greenlee et al., 2001) with fewer than 20% of patients surviving one year from diagnosis (Lillemoe, 1998). Advanced age, male gender, liver metastases and large tumour diameter have been reported as unfavourable prognostic factors (Lillemoe et al., 2000; van den Bosch et al., 1994).
Weight loss has been shown to be a negative prognostic indicator for a range of cancers, however differences in survival duration for pancreatic cancer patients with and without weight loss in two large studies (Andreyev, Norman, Oates, & Cunningham, 1998; DeWys et al., 1980) did not reach statistical significance. This may reflect the very short expected survival times for this disease (median survival times of 12 vs 14 weeks in DeWys’ study) and the confounding effect of oedema and ascites on body weight measurement.
In a study of 102 patients with unresectable pancreatic cancer, percentage weight loss at diagnosis was inversely related to survival duration (Falconer et al., 1995). Weight loss however did not have independent prognostic value after adjusting for stage of disease and presence of an acute phase protein response (APPR). The presence of an APPR, defined as elevated C reactive protein (CRP) levels, was demonstrated to be a useful independent prognostic indicator. Median survival for patients with an APPR was 66 days compared to 222 days for those without an APPR.
Literature Review
11
Length of survival for a group of 25 pancreatic cancer patients who presented to a palliative care service within one month of diagnosis, was analysed according to the presence of a range of reported symptoms. Survival time was unaffected by all symptoms except dyspnoea. Those with poorer performance status had a reduced length of survival (Krech & Walsh, 1991). The importance of the ability to perform activities of daily living, also sometimes referred to as functional status, is discussed in more detail in section 2.5.6.
A prognostic score has been developed from pooled data from 1020 patients with unresectable pancreatic cancer to assist in determining the most appropriate treatment – endoscopic management for those who are expected to have shorter survival and palliative surgery for those with better prognostic scores (Terwee et al., 2000). Prognosis was found to be poorer for those with metastases (Stage IV disease), pain or weight loss at diagnosis, while the presence of jaundice was a relatively good prognostic sign. Older men faired worse than women and younger men. The prognostic score however is derived from only gender, age (for men) and the presence of metastases, along with a hospital-specific six month mortality rate.
2.1.5
Treatment
Tumour resection with curative intent may be possible for only 10-20% of patients (Cooperman, Kini, Snady, Bruckner, & Chamberlain, 2000). In addition to the patient being fit for surgery, there must be no vascular invasion or metastases and the tumour needs to be small in size. Surgical intervention usually takes the form of a Whipple procedure in which all or part of the pancreas, part of the stomach, duodenum, gallbladder and part of the common bile duct are removed. This complicated surgery can have significant side effects including delayed gastric emptying, and a lengthy recovery period.
Resection is not an option for the majority of patients, and, because the prognosis for unresectable pancreatic cancer is so poor, medical care focuses on the management of symptoms to improve quality of life (Alter, 1996; Ellison, Chevlen, Still, & Dubagunta, 2002).
12 Literature Review
Non-curative treatment is directed primarily at relieving obstructive jaundice, duodenal obstruction and pain, as well as the management of individual issues such as nausea and steatorrhoea. Obstruction of the common bile duct or duodenum frequently occur and so palliative endoscopic or open surgical procedures are often required. In a study of the natural history of unresectable pancreatic cancer, 19 out of the 20 patients underwent either palliative bypass procedures or endoscopic stenting (Wigmore et al., 1997b).
Treatment strategies for unresectable cancer vary considerably, with palliative chemotherapy and/or radiotherapy offered at some centres and not others due to lack of consistent evidence that benefits to the patient in terms of palliation or survival time outweigh side effects, patient burden and cost. A number of randomised clinical trials to determine the optimal use of drugs such as gemcitabine are underway (Neoptolemos et al., 2003).
In some centres, patients with locally advanced tumours, previously considered unresectable, are now undergoing preoperative, adjuvant chemoradiotherapy (Cooperman, 2001; Wanebo et al., 2000). Other areas of research into the treatment of pancreatic cancer include gene therapy, immunotherapy, gut hormone therapy and the use of anti-angiogenesis factors.
2.2
Cancer associated weight loss
2.2.1
Prevalence of weight loss in cancer
Unintentional weight loss occurs commonly in cancer but more frequently with some tumour types than others, making it inappropriate to pool such groups in nutrition support studies.
An analysis of more than 3000 patients with cancers unsuitable for curative surgery or radiation, who participated in chemotherapy trials by the Eastern Cooperative Oncology Group (DeWys et al., 1980) demonstrated the differences in frequency of pretreatment weight loss for different cancer types (Figure 2.1).
Literature Review
13
The highest frequency of weight loss occurred in patients with pancreatic or gastric cancers, with approximately 60% of patients having lost at least 5% of their body weight prior to commencing chemotherapy. These results are supported by a study of 1555 patients with locally advanced or metastatic gastrointestinal cancer preparing for chemotherapy in which 70% of patients with cancer of the oesophagus, stomach or pancreas, but only a third of the patients with colorectal cancer, had weight loss (Andreyev et al., 1998). Figures for prevalence of weight loss at diagnosis of unresectable pancreatic cancer have ranged from 51 – 85% (Krech & Walsh, 1991; Wigmore et al., 1997b).
Figure 2.1
Frequency and degree of weight loss in patients with advanced cancer
This figure is not available online. Please consult the hardcopy thesis available at the QUT Library.
Source: (DeWys et al., 1980) Weight lost over previous six months (two months for pancreatic cancer) NHL, non-Hodgkin’s lymphoma; Leukaemia, acute nonlymphocytic leukaemia
14 Literature Review
Definitions for “weight loss” vary in these studies, but it is clear that gastrointestinal cancers in general, and pancreatic cancer in particular, are associated with considerable weight loss.
2.2.2 Effects of weight loss on patient outcomes
Weight loss is not only a common feature of cancer, it can be substantial, and for at least 20% of cancer patients, death is attributed to the effects of malnutrition (Langer, Hoffman, & Ottery, 2001). Studies involving weight losing cancer patients typically deal with median weight losses of 15-20% (O'Gorman, McMillan, & McArdle, 1998; Tchekmedyian et al., 1992). In Wigmore et al’s descriptive study of patients with unresectable pancreatic cancer, 15% of patients had lost more than 20% of their usual body weight at diagnosis and 60% had done so just prior to death (Wigmore et al., 1997b).
Blackburn classified unintentional weight loss of 15-20% as severe, even if it occurs over a six month period (Blackburn, Bistrian, Maini, & al, 1977). Blackburn’s categories, published 25 years ago are still widely referred to in nutrition support literature (Bloch, 1993; Langer et al., 2001; McCallum & Polisena, 2000). The definitions for significant and severe weight loss are shown in Table 2.2. This degree of loss, some of which will be from lean tissue stores, would be expected to impair physiological functions such as respiratory function.
Table 2.2
Categories of degree of weight loss
This table is not available online. Please consult the hardcopy thesis available at the QUT Library.
Source: (Blackburn et al., 1977)
Literature Review
15
Many studies have demonstrated that malnourished patients have poorer health outcomes;
reduced survival time (Andreyev et al., 1998; DeWys et al., 1980; Persson, Sjoden, & Glimelius, 1999)
reduced tolerance of, and response to chemotherapy (Andreyev et al., 1998; DeWys et al., 1980; Ottery, 2000)
poorer surgical outcomes (van Bokhorst de van der Scheuren et al., 1997; Windsor & Hill, 1988)
impaired immune response and increased risk of infection (Chandra, 1995; Lesourd, 1997)
delayed wound healing (Demling & De Santi, 1998)
increased health care costs (Ottery, 1996b)
reduced functional status, impaired muscle function and strength (DeWys et al., 1980; Lopes, Russell, Whitwell, & Jeejeebhoy, 1982; O'Gorman et al., 1998; Ollenschlager, Thomas, Konkol, Diehl, & Roth, 1992)
decreased quality of life (Larsson, Akerlind, Permerth, & Hornqvist, 1995; O'Gorman et al., 1998; Ovesen, Hannibal, & Mortensen, 1993b)
Weight loss has been shown to be a negative prognostic indicator for cancers in general, although for pancreatic cancer the association is less clear as was discussed in section 2.1.4. Similarly, in a study of patients about to receive chemotherapy for advanced cancer, decreasing weight was correlated with decreasing performance status, except for patients with pancreatic and gastric cancer (DeWys et al., 1980).
Weight loss with resulting malnutrition is never desirable, however weight gain may not be beneficial in all situations. For example, weight gain is thought to be an adverse prognostic factor for women with breast cancer receiving adjuvant therapy, possibly as the result of increases in circulating hormone levels from increased body fat stores (Nixon, 1996). There is no evidence however that tumours of nonreproductive organs such as the pancreas, respond in this way.
2.2.3 Metabolic factors affecting weight loss
16 Literature Review
Cancer associated weight loss appears to involve not only reduced nutrient intake but also increased requirements and barriers to the utilisation of nutrients. Nutrition support studies (Chlebowski, Palomares, Lillington, & Grosvenor, 1996; Evans et al., 1987; Ovesen, Allingstrup, Hannibal, Mortensen, & Hansen, 1993a) suggest that increasing intake alone does not reverse this process. The field of altered metabolism in cancer is an active one with in vitro, animal and human studies continuing to expand knowledge of the multiple processes that appear to be involved (Barber, Ross, Voss, Tisdale, & Fearon, 1999b; Haslett, 1998; Inui, 1999; Langer et al., 2001; Loprinzi et al., 1999; McMillan et al., 1999).
Cancer associated weight loss differs from starvation in that the normal adaptive mechanisms to conserve body protein do not function (Brennan, 1977). Adipose tissue is lost to a greater degree in starvation with fat stores utilised to replace glucose as the principal energy source. This spares lean body tissue, minimising loss of skeletal muscle. In contrast, the wasting process of cancer involves the loss of both fat and lean body mass. Skeletal muscle in particular, is lost, with relative conservation of visceral protein stores (Tisdale, 1997a).
Anaerobic glycolysis by the tumour results in increased lactate levels in the blood. This is converted back to glucose in the liver but results in a net loss of ATP. This Cori cycle accounts for 20% of glucose turnover in the healthy person but can be 50% in cancer (Tisdale, 1997b). Loss of muscle mass is increased as a result of activation of the ATP-ubiquitin-dependent proteolytic pathway (Tisdale, 2002). Several mediators have been proposed including cytokines (tumour necrosis factor and interleukin-6) and proteolysis inducing factor (PIF).
Alterations in metabolic rate (Wigmore, Plester, Ross, & Fearon, 1997c), proinflammatory cytokines (Falconer, Fearon, Plester, Ross, & Carter, 1994; Karayiannakis et al., 2001) and catabolic factors such as PIF and lipid mobilising factor (LMF) (Hirai, Hussey, Barber, Price, & Tisdale, 1998; Todorov et al., 1996; Wigmore et al., 2000b) have been reported in weight losing pancreatic cancer patients.
Literature Review
17
Cancer associated weight loss is often referred to in the literature as “cachexia”. While there have been many studies investigating cachexia, it remains a poorly defined syndrome with the result that the specific causes of the substantial weight loss of the cachectic patients referred to in studies is unclear. For example, while weight loss occurring in a group of patients with acute myelogenous leukaemia undergoing chemotherapy was determined to be primarily due to inadequate oral intake (Ollenschlager, Konkol, & Modder, 1988) this may not be the only reason for weight loss in hypermetabolic patients with pancreatic cancer receiving no active treatment. Unfortunately, in studies of weight loss in cancer, the patient groups are often not homogeneous for cancer type or presence of cachexia, making results difficult to interpret.
In 1977 Brennan defined cancer cachexia as “the syndrome of emaciation, debilitation and malnutrition that accompanies cancer” (Brennan, 1977). More recent descriptions incorporate the expanding knowledge of the cachectic process - “cancer cachexia is a complex, multifactorial syndrome that results from a reduction in food intake, a variety of metabolic abnormalities (including hypermetabolism) or more often a combination of the two” (Fearon & Moses, 2002). Cachexia has also been described in a number of chronic diseases such as HIV-AIDS, renal and liver failure, and arthritis, as well as in relation to the aging process. There are, however, no diagnostic criteria for cancer cachexia.
Some of the metabolic abnormalities and nutrition-related symptoms that occur in weight losing cancer patients are discussed below. Issues that relate specifically to chemotherapy, radiotherapy or surgical interventions have not been included.
2.2.4 Resting energy expenditure
Studies of resting energy expenditure (REE) in weight-losing cancer patients have failed to show consistent alterations. Both increases and decreases have been demonstrated which may reflect the many different causes of weight loss in these patients. Neither advanced stage of disease nor tumour bulk has been consistently associated with hypermetabolism although successful resection has been shown to ameliorate it.
Literature Review Table 2.3
18
Summary of studies describing metabolic changes in pancreatic and other gastrointestinal cancers.
Author, Year, Country
Patient Group
Parameters
Results
Falconer 1994 UK
21 PC pts (mean wt loss 18%) 16 controls
CRP (>10mg/L = APPR) REE (indirect calorimetry) BCM, FFM (BIA), weight
Higher REE* for pts than controls (25.9 vs 19.4 kcal/kg BW/d) Higher REE* for pts with APPR than without (28.7 vs 23.8 kcal/kg BW/d) * whether as a proportion of BW, BCM or FFM
Falconer 1995 UK
102 PC pts (median wt loss 11%)
CRP (>10mg/L = APPR) survival duration
Shorter median survival for pts with APPR than without (66d vs 222d)
Wigmore 1995 UK
16 PC pts (median wt loss 17%) 17 controls
CRP, REE (indirect calorimetry) FFM (BIA), weight
Higher REE for pts than controls (25.6 vs 19.2 kcal/kg BW/d). Also significant as a proportion of FFM. Mean CRP for pts was 51 mg/L – not detectable in controls
Wigmore 1997 UK
35 PC pts
CRP (≥10mg/L = APPR) REE (indirect calorimetry) predicted REE (Schofield eqn)
Measured REE greater than predicted REE Higher REE for pts with APPR than without (26.6 vs 23.3 kcal/kg BW/d)
O’Gorman 1998 UK
119 GI cancer pts Grp1 = wt loss >5% (n=97) (median wt loss 17%) Grp2 = steady wt (n=22)
CRP (>10mg/L = APPR) albumin, weight
More pts with an APPR in Grp1 than Grp2 (71% vs 27%) CRP greater in Grp1 than Grp2 (29 vs 70 yrs M,F
BMI pre-illness
continuous
kg/m2
from measured Ht & reported pre-
BMI pre-illness
categorical
≤24, 25-29, ≥30
illness Wt from measured Ht & reported pre-
BMI T0
continuous
kg/m2 kg/m2
illness Wt from measured Wt & Ht
BMI T0
categorical
150 kJ/kg/d but so did 10% patients in the WL group. If the T8 results accurately represent intake over the study period, then those individuals would appear to be hypermetabolic. A closer examination of food intake and assessments of absorptive capacity and vomiting would be needed to clarify this.
It is possible that the increase in nutrient intake would have been greater, or may have occurred in a greater number of patients, if the intensive nutrition intervention had not been in the setting of a research trial. The purpose of the BH80 study was to determine whether the fish oil containing supplement was more effective than the control supplement in improving weight, therefore the focus was on achieving a minimum of 1.5 cans of supplement per day. In a non-research setting, a dietitian would be able to utilise a wider range of supplements and foods eg. the addition of protein powder to savoury foods such as soup, to adapt to individual taste
130
Discussion
preferences. Patients with poor appetite, however, may be more motivated to take a specialised research product and less inclined to persevere with “ordinary” foods.
This study cannot clarify whether providing oral supplements leads to weight stability, however, it does demonstrate that many weight losing patients with unresectable pancreatic cancer were able to be assisted to increase oral intake and minimise their weight loss, whereas those who were unable to increase their intake continued to lose weight.
Measurement of dietary intake
One of the difficulties in determining whether nutrition intervention is effective for weight losing cancer patients, is the estimation of nutrient intake. The act of recording food intake has been shown to affect intake even in healthy well-motivated subjects (Rebro, Patterson, Kristal, & Cheney, 1998). Under-reporting of food intake has frequently been observed in studies of healthy, weight stable populations (Black et al., 1991; Livingstone et al., 1990; Martin et al., 1996).
The ratio of EI/BMR, a tool for identifying under-reporting, was developed using data from weight stable populations (Goldberg et al., 1991). It is based on the concept that ratios of energy intake to basal metabolic rate can be estimated below which it is highly unlikely that an individual could maintain energy balance and therefore at that level it is likely that under-reporting has occurred. It is not useful for detection of under-reporting in this situation because every study participant was losing weight on study entry. Potential alterations to metabolic rate due to cancer would further complicate the situation.
Over-reporting may actually be more of an issue for this group of patients many of whom were struggling with early satiety and nausea, while being encouraged to increase intake to meet study goals. Another limitation of the use of food diaries may be the day to day variation in intake that occurs for patients with fluctuating symptoms such as pain and nausea. Recording of food intake can be further complicated by malabsorption or episodes of vomiting, reducing the availability of nutrients.
Discussion
131
Compliance to the dietary prescription is a challenge in the presence of the many symptoms experienced by patients with advanced cancer, and recording food intake is an added burden. There was more data missing for energy intake at T8 than any of the other variables examined (12% vs less than 4% missing for other variables). In some cases diaries had been kept but contained insufficient detail to be of use. This missing data would be expected to bias the results towards higher food intakes assuming that sicker patients were the ones who were less likely to complete the food records and would be likely to consume less.
Tracers such as deuterated water can be included to confirm intake of supplements, but unfortunately no objective marker is available for overall nutrient intake. The double labelled water technique allows the comparison of reported energy intake with expenditure but is not practical for general clinical use. Patient-recorded food diaries, despite their limitations, provide a balance between the need to gather sufficient information and the minimisation of patient burden in the palliative care setting. The individual therapist/client relationship formed in this study was felt to improve the accuracy of the nutrient intake data. Patients were reminded by phone when the recordings were due to commence and any unclear or unusual diary entries were cross-checked with patients at each data collection point.
Appetite
A greater proportion of the WS group reported no loss of appetite at baseline than the WL group (41% vs 18%). After adjusting for the other variables in logistic regression, however, it did not reach statistical significance as an independent determinant of weight stability. Interrelationships would be expected between appetite, food intake and symptoms such as nausea and pain. Loss of appetite has been shown to be a strong negative prognostic indicator in advanced lung and colorectal cancer (Sloan et al., 2001) and hospice patients with various cancer types (Reuben, Mor, & Hiris, 1988). Krech et al, however, were unable to show any relationship between appetite and survival time in unresectable pancreatic cancer (Krech & Walsh, 1991).
132
Discussion
Appetite, as with all symptoms, is subjective and therefore difficult to measure. The measure of appetite used in this study was a simple one, taken from the EORTC QLQ-C30 questionnaire and related to the week prior to commencement of the study. It cannot be used as a surrogate for food intake because appetite is not the only factor involved in determining the amount eaten. Some patients find it less difficult than others to override a loss of appetite, at least initially, and are able to eat more than they feel like eating (DeWys, 1977).
Acute phase protein response
The WL group contained a higher proportion of patients with raised CRP levels (41% vs 21% for WS), supporting the findings by Falconer et al that an acute phase response is associated with hypermetabolism (Falconer et al., 1994) and weight loss (Falconer et al., 1995). Lack of APPR, however, was not found to be an independent predictor of weight stability.
The majority of patients in this study did not have elevated CRP levels at baseline despite the fact that they had all lost at least 5% of body weight on entry to the study. This highlights the fact that the increased energy requirement that occurs during an APPR is only one of a range of factors likely to be responsible for weight loss in unresectable pancreatic cancer (Inui, 2002; Tisdale, 2002). Further studies examining the effects of EPA-fortified supplements on patients demonstrating an APPR or the presence of cachectic factors such as PIF may provide information on the effect of EPA on these metabolic changes.
Body mass index
The majority of patients included in this study (72%) were overweight or obese prior to weight loss from their cancer, and none had been underweight. This supports previous findings that BMI may be a risk factor for pancreatic cancer (Michaud et al., 2001b; Silverman et al., 1998). Pre-illness and baseline BMI values but did not differ for those who were or were not included in analysis so this observation reflects the entire BH80 study group.
Discussion
133
Mean BMI values, both pre-illness and at baseline, were greater for patients in the WL group than the WS group – 28.0 vs 26.1 kg/m2 pre-illness and 23.4 vs 21.4 kg/m2 at baseline. Baseline BMI produced a high odds ratio in logistic regression analysis but did not reach statistical significance (P = 0.058). The odds ratio (likelihood of being in the WS group) for those who commenced the study with a BMI in the healthy weight range (20-25 kg/m2) was 4.8 (95% CI 1.0-24.0) compared to those who were overweight or obese (> 25 kg/m2).
Pancreatic cancer patients with measurable levels of PIF in the urine have been reported to have had greater pre-illness weight than those who were PIF negative (76 vs 65 kg, p=0.009) (Wigmore et al., 2000b). They also went on to have greater weight loss, leading the authors to suggest that production of PIF by the tumour might be related to body mass. This may explain why overweight/obese was more common in the WL group in our study.
These results suggest that being overweight or obese may be a disadvantage, not only as a risk factor for development of pancreatic cancer, but also in terms of poorer outcomes during the course of the disease. Further investigations would be needed to confirm this, determine the mechanisms involved and then to identify intervention strategies.
Pain
Pain has been identified as a one of the symptoms frequently occurring in cancer that can limit nutrient intake (Feuz & Rapin, 1994; Gallagher-Allred, 1989; Ottery, 1995). The treatment of pain can also lead to further nutrition impact symptoms such as nausea and constipation.
At baseline, pain was experienced by the majority of patients as has been reported in other studies (Krech & Walsh, 1991; Lillemoe, 1998; Ottery, 1996a). It was significantly more common amongst the patients who continued to lose weight (84%), than those who stabilised their weight loss (67%) but lack of pain was not independently predictive of weight stability.
134
Discussion
The measure of pain used was a simple all-or-nothing scale, derived from the QoL questionnaire as for the other symptom scales, and does not reflect whether pain was well managed during the study. Effective pain relief measures have been shown to be associated with prolonged survival in advanced pancreatic cancer (Lillemoe et al., 2000), possibly, at least in part, by facilitating adequate food and fluid intake.
7.3.3
Characteristics which were not predictive
Some of the features that were not significantly different between the weight groups, are discussed below.
Stage of disease
The lack of significant difference in disease staging between the groups supports the proposal that patients in the WL group did not simply have more extensive disease than those in the WS group. The four level staging was provided by medical officers at enrolment and does not account for any progression that may have occurred over the eight week study period, however not even a trend was apparent.
There was a trend towards more advanced disease, however, in those patients who were excluded from secondary analysis compared to those who were included (p = 0.082) (Table 4.2). Thirty percent of included patients had Stage IV disease compared to 46% of excluded patients and this is supported by the fact that death within two months of enrolment was one of the main reasons for exclusion. The presence of metastases (Stage IV disease), particularly liver metastases, has been reported as having prognostic value (Falconer et al., 1995; Terwee et al., 2000; van den Bosch et al., 1994). These same studies acknowledge difficulties in staging pancreatic cancer, especially in distinguishing between Stages II and III. Nevertheless there was clearly no difference between the WL and WS groups in the proportion of Stage IV disease (Table 4.16).
Randomisation & plasma EPA
Discussion
135
Two methods were used for examining the relationship between consumption of EPA and weight stability. The simplest was whether patients had been randomised to either the control or experimental arm of the BH80 study but because of known compliance problems in both arms, plasma EPA was included as a more direct measure of the effect of EPA.
There was no difference in the proportion of patients from each arm of the BH80 trial in the two weight groups (Table 4.12). This supports the decision to pool the two randomisation groups and consider them as one large group, all of whom had received ONS. The number of patients who were able to reach plasma EPA levels ≥3% by T8 was also not significantly different for the two groups (p= 0.133), although there did appear to be a trend towards a greater proportion in the WS group (40% vs 24% for the WL group) (Table 4.24).
A plasma level of 3% EPA is well above levels found in unsupplemented people (Barber et al., 1999b; Sinclair, O'Dea, & Johnson, 1994; Wigmore et al., 2000a; Zuijdgeest-Van Leeuwen et al., 2002) and is approximately the 25th percentile for studies involving supplementation with 2g EPA per day (Barber et al., 1999b; Wigmore et al., 1996); a dose not achieved by the majority of patients in this study. Higher levels of EPA may be capable of producing significant increases in weight and LBM, however strategies for improving compliance would be required for median plasma EPA levels to reach the 5% achieved in the pilot study for the BH80 trial (Barber et al., 1999b).
EPA may later be found to provide beneficial outcomes for specific subgroups of cancer patients or at higher plasma levels, but the BH80 study was unable to show any benefit over and above that of intensive nutrition intervention which included a protein and energy dense supplement (Fearon et al., 2003).
136
Discussion
Functional status
Measures of performance status such as The Eastern Cooperative Oncology Group performance status (ECOG-PS) and the Karnofsky Performance Status scale (KPS) have been shown to have prognostic value (Krech & Walsh, 1991; Reuben et al., 1988; Sloan et al., 2001). The fact that KPS scores for patients who were excluded from this secondary analysis were significantly lower than those who were included (p
