2807710353

Nutritional Status and Body Composition

in Patients with Chronic Liver Disease

Angela Maiy Madden

Royal Free Hospital

i'

i i r

Submitted for the degree of Doctor of Philosophy

Royal Free Hospital School of Medicine

University of London

March 1998 ProQuest Number: 10015918

All rights reserved

INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest.

ProQuest 10015918

Published by ProQuest LLC(2016). Copyright of the Dissertation is held by the Author.

All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code. Microform Edition © ProQuest LLC.

ProQuest LLC 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 i6 i S iK-i k r V y= ^ à fL^fi^.* i LW wB i iJ ti—

^ '. • ; - -/ 1 i ^ **» ABSTRACT

Patients with chronic liver disease are often malnourished. Clinical assessment of nutritional status is difficult in this patient population because of the confounding effects of abnormalities of hydration status and protein metabolism. To overcome these difficulties, a global method of assessment incorporating objective and subjective data was developed. Its reproducibility and internal validity were confirmed and it was used to assess the prevalence of malnutrition in 215 patients with cirrhosis. The relationship between global nutritional status and outcome was examined in 271 liver transplant candidates. Pre-transplant malnutrition was significantly associated with a reduced survival 90 days after surgery.

Many factors contribute to impaired nutritional status in patients with cirrhosis. The potential effects of hypermetabolism were examined by measuring resting energy expenditure in 1 0 0 patients with cirrhosis and a group of comparable healthy volunteers. Resting energy expenditure was significantly higher in the patients when adjusted for fluid retention and fat-free mass, determined using anthropometry.

Further nutrition research in patients with chronic liver disease is hampered by an absence of validated methods of accurately assessing body composition. Such data are required, not only to fully validate clinical assessment techniques, but also to quantify metabolically active tissue and fat compartments which are essential prerequisites in energy expenditure, protein turnover and nutrition intervention studies. The difficulties of assessing body composition in chronic liver disease may be overcome by using a four-compartment model in which total body water, mineral, fat and protein are assessed. This model was applied to 20 patients with cirrhosis who underwent underwater weighing, deuterium dilution and dual-energy X-ray absorptiometry studies. The results were compared with bedside methods of assessments and significant differences found. The hydration fraction of fat-free mass varied between 71.7-83.0%. Although this model of body composition assessment is a research tool, it has shown the limitations of other techniques and provided data upon which further studies can be based. I i :) i fhU CONTENTS

Abstract 2 List of tables 7 List of figures 10 Abbreviations 11 Acknowledgements 13 Introduction 16

Chapter 1 The assessment of nutritional status in patients with chronic liver disease 1.1 Outline 23 1.2 Introduction 24 1.3 Nutrient intake 25 1.4 Anthropometry 26 1.5 Biochemical variables 29 1.6 Immune function 30 1.7 Muscle function 31 1.8 Clinical nutritional status 32 1.9 Other assessment methods 33 1.10 Summary 34 1.11 Study 1; A new global assessment method for evaluating nutritional status in patients with cirrhosis 1.11.1 Aims 35 1.11.2 Patients, methods and results 35 1.11.3 Discussion 42

Chapter 2 The prevalence of malnutrition in patients with chronic liver disease 2.1 Outline 53 2.2 Introduction 53 2.3 Nutrient intake 54 2.4 Anthropometry 54 4

2.5 Muscle function ; . - . , 5 5 2.6 Clinical nutritional status 55 2.7 Prevalence of malnutrition and severity of liver disease 55 2.8 Prevalence of malnutrition and aetiology of liver disease 55 2.9 Summary 56 2.10 Study 2: A survey of the prevalence of malnutrition in patients with cirrhosis

2 .1 0 . 1 Aim 56

2 .1 0 . 2 Patients 57 2.10.3 Assessment of nutritional status 57 2.10.4 Statistical analysis 59 2.10.5 Results 59

2 .1 0 . 6 Discussion 64

Chapter 3 The relationship between malnutrition and adverse clinical outcome in patients with chronic liver disease 3.1 Outline 85 3.2 Introduction 85

3.3 Nutritional status and outcome in liver disease 8 6 3.4 Nutritional status and prognostic indices for use in patients with

cirrhosis 8 8 3.5 Nutritional status and outcome in patients with liver disease undergoing abdominal surgery 89 3.6 Nutritional status and clinical outcome in patients undergoing liver transplantation 90 3.7 Summary 93 3.8 Defining clinical outcome 94 3.9 Study 3: An investigation into the association between pre-operative nutritional status and liver transplant survival 3.9.1 Aim 95 3.9.2 Patients 95 3.9.3 Nutritional assessment 96 3.9.4 Transplant outcome 98 5

3.9.5 Statistical analysis 98 3.9.6 Results 99 3.9.7 Discussion 100

Chapter 4 Factors contributing to malnutrition in cirrhosis: The role of hypermetabolism 4.1 Outline 117 4.2 Introduction 117 4.3 Nutrient intake 118 4.4 Nutrient loss 125 4.5 Nutrient utilization 128 4.6 Summary 133 4.7 Resting energy expenditure in cirrhosis 134 4.8 Study 4: Assessment of resting energy expenditure in an unselected group of patients with cirrhosis 4.8.1 Aim 140 4.8.2 Patients and volunteers 140 4.8.3 Assessment of nutritional status 141 4.8.4 Preparation for REE measurement 142 4.8.5 Indirect calorimetry 143 4.8.6 Predicted resting energy expenditure 145 4.8.7 Statistical analysis 146 4.8.8 Results 146 4.8.9 Discussion 153

Chapter 5 The assessment of body composition in patients with cirrhosis 5.1 Outline 189 5.2 Introduction 190 5.3 Densitometry 191 5.4 Dilution techniques 193 5.5 Total body potassium 196 5.6 Neutron activation 197 5.7 Dual energy x-ray densitometry 199 6 5.8 Anthropometry 202 5.9 Bioelectrical impedance analysis 207 5.10 Imaging techniques 209 5.11 Near infrared interactance 211 5.12 Urinary metabolite excretion 212 5.13 Single body composition techniques in cirrhosis 214 5.14 Multi-compartment models of body composition assessment 214 5.15 The four-compartment model 216 5.16 Study 5: The measurement of gastrointestinal gas 5.16.1 Aims 218 5.16.2Patients and methods 218

5.16.3 Results 2 2 2 5.16.4 Discussion 223

Study 6 : An evaluation of sampling times for dilution studies in patients with cirrhosis 5.17.1 Aim 227 5.17.2 Patients and methods 227 5.17.3 Statistical analysis 230 5.17.4 Results 231 5.17.5 Discussion 232 5.18 Study 7: Assessment of body composition in patients with cirrhosis using a four-compartment model 5.18.1 Aims 235 5.18.2 Patients and methods 235 5.18.3 Statistical analysis 240 5.18.4 Results 241 5.18.5 Discussion 242

Conclusions 271 Future Work 272 Appendix 275 References 278 LIST OF TABLES

1.1 Subjective Global Assessment data collection sheet 45 1.2 Assessment of nutritional status in chronic liver disease using composite methods 46 1.3 Demographic details of the patients in Study 1 47 1.4 Repeatability coefficients of anthropometric variables 48 1.5 Interobserver agreement and relationship with modified SGA in Study 1 Phase 1 49 1.6 Agreement in nutritional category in Study 1 Phase 2 50 1.7 Inter observer agreement and relationship with RFH GA in Study 1 Phase 3 51

2.1 Body weight in chronic liver disease 67

2.2 Triceps skinfold measurements in chronic liver disease 6 8 2.3 Mid-arm muscle circumference in chronic liver disease 69 2.4 Nutritional status in chronic liver disease using clinical methods 70 2.5 Nutritional status and severity of chronic liver disease 71 2.6 Nutritional status and aetiology of chronic liver disease 72 2.7 Aetiology and severity of liver disease in Study 2 73 2.8 Gastrointestinal symptoms in patients with cirrhosis 74 2.9 Anthropometric measurements in patients with cirrhosis 75 2.10 Global nutritional status and anthropometric variables 76 2.11 Severity of liver disease and nutritional variables 77 2.12 Aetiology of liver disease and nutritional variables 78 2.13 Grip-strength and global nutritional status and severity of liver disease 79

3.1 Nutritional variables and outcome in chronic liver disease 107 3.2 Nutritional variables and outcome in patients with chronic liver disease undergoing surgery 108 3.3 Nutritional variables and outcome in liver transplantation 109 8

3.4 Aetiology of patients in Study 3 110 3.5 Pre-transplant nutritional status 110

3.6 Survival after liver transplantation 1 1 1 3.7 Post-transplant survival: Objective pre-transplant

nutritional variables 1 1 2 3.8 Post-transplant survival: Pre-transplant nutrient intake 113 3.9 Post-transplant survival: Pre-transplant weight change 114

4.1 Nutrient intake in patients with chronic liver disease 161 4.2 Demography of patients with cirrhosis in REE studies 163 4.3 Procedural details of REE studies in cirrhosis 165 4.4 REE in patients with cirrhosis and healthy volunteers 168 4.5 Critical evaluation of REE studies in cirrhosis 169 4.6 Aetiology and severity of liver disease in Study 4 170 4.7 Demographic and anthropometric data in cirrhotic patients and healthy volunteers in Study 4 171 4.8 Demographic and anthropometric data in patients with and without clinical evidence of fluid retention & volunteers 172 4.9 REE and RQ in patients and volunteers 173 4.10 REE and RQ in patients with and without clinical evidence of fluid retention and volunteers 174 4.11 REE and RQ in non-smoking patients and volunteers 175 4.12 REE and RQ in patients and volunteers who smoked 176 4.13 Anthropometric data and REE in male and female patients 177 4.14 Anthropometric data and REE in male and female volunteers 178 4.15 Demographic and anthropometric data in patients: Effect of disease severity 179 4.16 REE and RQ in patients: Effect of disease severity 180 4.17 Demographic and anthropometric data in patients: Effect of nutritional status 181 4.18 REE and RQ in patients: Effect of nutritional status 182 4.19 Measured and predicted REE in patients with cirrhosis 183 4.20 Measured and predicted REE in healthy volunteers 184 9 5.1 Dilution techniques in chronic liver disease 247 5.2 Total body potassium in chronic liver disease 248 5.3 DEXA assessments in chronic liver disease 249 5.4 Anthropometric assessments in chronic liver disease 250 5.5 Comparative BIA studies in chronic liver disease 251 5.6 BIA studies of paracentesis in patients with cirrhosis 252 5.7 24-hour urinary creatinine excretion in cirrhosis 253 5.8 The four-compartment model of body composition 254 5.9 Subjective assessment of flatulence and abdominal bloating with non-absorbable disaccharides 255 5.10 Patients in Study 5, Objective Assessment 1 256 5.11 GI tract gas estimated by CT scanning 257 5.12 Patients in Study 5, Objective Assessment 2: GI tract gas 258 5.13 Published studies quantifying gastrointestinal gas 259

5.14 Demography, anthropometry and clinical data in Study 6 260 5.15 Deuterium emichment in saliva, serum and ascites after oral administration 261 5.16 Demographic and anthropometric data in patients with cirrhosis undergoing body composition assessment 262 5.17 Bias in total body fat assessed using the four-compartment model and single assessment techniques 263 5.18 Bias in FFM assessed using the four-compartment model and single assessment techniques 264 5.19 Comparison of total body fat using the four-compartment model as a reference against single assessment techniques 265 10 LIST OF FIGURES i Inscribed model of a sheep’s liver, 1900-1600 BC 22

1.1 Algorithm for categorizing nutritional status in cirrhosis 52

2.1 Assessment of muscle function using hand held dynamometer 80 2.2 Nutritional status and adequacy of nutrient intake 81 2.3 Nutritional status and dietary restrictions 82 2.4 Nutritional status and oral nutritional supplements 83 2.5 Adequacy of nutrient intake and severity of liver disease 84

3.1 Global nutritional status in liver transplant patients 115 3.2 Survival after liver transplantation 116

4.1 Measured and predicted REE: Harris-Benedict formulae in patients with cirrhosis 185 4.2 Measured and predicted REE: Harris-Benedict formulae in healthy volunteers 186 4.3 Measured and predicted REE: Schofield formulae in patients with cirrhosis 187 4.4 Measured and predicted REE: Schofield formulae in healthy volunteers 188

5.1 Abdominal CT scan showing gastrointestinal tract gas 266 5.2 Deuterium emichment of saliva, serum and ascitic fluid 267 5.3 Two views of the underwater weighing tank 268 5.4 Calculation of body density in underwater weighing 269 5.5 Hydration fraction of FFM in patients with cirrhosis 270

Appendix RFH GA of nutritional status in patients with cirrhosis 277 11 ABBREVIATIONS USED THROUGHOUT THE THESIS

BCM Body cell mass BIA Bioelectrical impedance analysis BMI Body mass index DEXA Dual-energy X-ray absorptiometry ECW Extracellular water FFM Fat-free mass ICW Intracellular water MAC Mid-arm circumference MAMC Mid-arm muscle circumference RQ Respiratory quotient REE Resting energy expenditure RFH GA Royal Free Hospital global assessment SGA Subjective global assessment TBK Total body potassium TBW Total body water TSF Triceps skinfold thickness

STATISTICAL ANALYSIS

All statistical tests were considered significant at the 5% level. 12

PUGH’S SCORE

The severity of liver disease in patients with cirrhosis is categorized according to Pugh’s modification of Child’s grading system {Pugh 1973} using the sum of the scores from five variables:

Variable Score

1 2 3

Serum albumin >35 28-35 <28 (g/i) Serum bilirubin <34 34-51 >51 (/jmol/l)

Prothrombin time <17 17-19 > 2 0 (s) Ascites none mild moderate +

Encephalopathy none I-II III + (grade)

On the basis of their Pugh’s score, patients are grouped into two or three categories:

Pugh’s score (5-15)

Child’s grade A 5-6 Child’s grade B 7-9 Child’s grade C 10-15

Compensated liver disease 5-7 Decompensated liver disease 8-15 13 ACKNOWLEDGEMENTS

I am greatly indebted to Dr Marsha Morgan for giving me the opportunity to pursue my chosen field of research. She has supervised my work with her own inimitable style and taught me a tremendous amount about many things. In addition, I am grateful to her for providing financial support throughout my studies. I would also like to thank Dr Jim Owen and Professor Richard Bruckdorfer for the roles they played on my supervisory committee.

The British Dietetic Association General and Educational Trust, the British Association for Parenteral and Enteral Nutrition, Scientific Hospital Supplies and Abbott Laboratories Limited contributed additional financial assistance and I am grateful to them.

I would like to express my sincere thanks to the numerous people who have given me practical help in the seven studies which comprise this thesis. Dr Richard Morris, Department of Primary Care and Public Health, provided immense support with statistical advice throughout; Clare Soulsby, Senior Dietitian, was a pleasure and inspiration to work with on the clinical assessments in Study 1; Karen Smith, Dr Bob Dick, and Dr Graham Dodge, Department of Radiology, gave invaluable help in the gas evaluations in Study 5; Dr Harriet Gordon, Department of

Medicine, enthusiastically and expertly supervised sample collection in Study 6 ; and Nigel Fuller and Graham Jennings, Dunn Clinical Nutrition Unit, taught me

an enormous amount during Studies 6 and 7 and I enjoyed their company immensely. I have also benefitted from the personal support, humour, expertise and enthusiasm of all these generous people.

The staff at the Dunn Clinical Nutrition Centre in Cambridge made my patients and I feel especially welcome. I acknowledge the privilege of working there and am extremely grateful to Dr Marinos Elia for facilitating this with such kindness and the expertise which he contributed to this work. Dr Susan Jebb provided the initial spark that led to the body composition study becoming reality and I am grateful to her. 14

From the Royal Free Hospital, I would also like to thank Carol Bateman, Nick Bradley, Jonathan Marshall and Justine Smith for their immeasurable companionship and support throughout the last four years. I am grateful to Professor Neil McIntyre, Dr Andrew Burroughs and Professor Geoff Dushieko for allowing me to study their patients and to other staff in the Department of Medicine, the Medical Library and Department of Medical Illustrations whose assistance has been appreciated.

I would like to express my sincere thanks to the many patients and healthy subjects who have taken part in my studies so willingly and with such good grace. They have been an immense pleasure to work with and provided far more than a raison d’être.

Outside work, I have been fortunate to receive unfailing support from so many people. My friends, particularly from the London Mountaineering Club, have listened, laughed, contributed ideas, acted as healthy controls and, when things were difficult, taken me to the hills. Thanks to all of them.

Finally, without the enduring support of my family I would never have started this work, let alone come this far. Thanks to you all. 15

For my mother and father.

"... but the patients rescued us in turn and gave meaning to what we did, or tried to do." Thomas E Starzl^®^ 16 INTRODUCTION

HISTORICAL PERSPECTIVE

The liver The liver has been recognised as an organ of great importance to the health and functioning of the body by many diverse cultures for thousands of years^^^’"^®’^^^. One of its earliest portrayals is a 4000 year old clay model found at Sippar, in what is now Iraq, which depicts the different hepatic lobes (Figure i). During the era in which this model was made, the shape of a liver removed from sacrificed animals was used to predict the future^^^ and the inscriptions on the model are thought to be omens.

A relationship between the liver and nutrition was recognised some two thousand years later in ancient Greece where Herophilos [280 BC] described the liver as the nourishing force in the body^^^. Galen [129-199 AD], a Greek medical authority who practised and taught in Rome, perceived the importance of the liver in the absorption and digestion of food. He described a "cooking process whereby the liver transforms the nutrients and charges them with a spirit that governs nutrition, growth and reproduction"^^^’^®^’"^^^. Today, with the benefit of current knowledge of anatomy, physiology and biochemistry, this concept remains basically correct. Indeed, Galen’s overall work achieved such prominence at the time that it was adopted throughout Europe and Arabia and his ideas remained largely unchallenged until the 15th century^^’^^^.

Leonardo da Vinci [1452-1519 AD]|disagreed with some of Galen’s theories and concluded from anatomical studies that the vena cava did not arise in the liver, an organ to which he attributed no specific role^°^. Paracelsus [1493-1541 AD] was also opposed to much of Galen’s work, and suggested an alternative alchemistic approach in which the body converted food into its own substance according to need^^^.

Dissection studies during the 17th century helped to define the hepatic anatomy Introduction 17

and in 1650 Olaf Rudbeck [1639-1702] discovered that fluid in the lacteals drained into the thoracic duct, thus establishing a pathway for nutrients to enter the blood stream without passing through the liver^^^. Four years later, Francis Glisson [1598- 1671] published "Anatomia Hepatis", describing the hepatic circulation, including the portal vein^^^.

Food shortages in post-revolutionary France provoked studies which led to further important discoveries relating to the liver and nutrition. François Magendie [1783- 1855], Professor of Medicine at the Collège de France in Paris, was appointed to lead an investigation into the nutritional value of various food extracts and subsequently demonstrated the need for adequate dietary protein^^^. Magendie's work was continued by his pupil, Claude Bernard [1813-1878], who discovered the glycogenic function of the liver, thus establishing a functional link with nutrition^^^’^^^.

Nutrition The first recorded use of the word "nutrefyde" was by Andrew Boorde [1490-1550?] in describing the value of bread in his book "Dietary of Health". However, the term "nutrition" is a relatively recent addition to the English vocabulary, being rarely used before the 20th century^^^. Prior to this, the word "diet", derived from the Latin "diaeta", meaning way of life, was in more general use and originally implied a broader sense of health and hygiene than today’s exclusively food-related understanding.

Early dietary guidelines prescribed for hospitalised patients, for example those recorded in St Bartholomew’s Hospital Journal in 1687, were intended to provide an adequate intake rather than for therapeutic purposes^^^. For almost two centuries, these remained little influenced by contemporaneous scientific discoveries^^^. The organisation skills and logical approach of Florence Nightingale [1829-1910] helped achieve significant improvements in the provision of hospital diets^^"* although these advances were not based upon systematic scientific enquiry 439 ' 511 . Introduction 18

However, the scientifically based achievements of Justus Leibig [1803-1873], Max von Pettenkofer [1818-1901], Carl Voit [1831-1908], Max Rubner [1854-1932] and WO Atwater [1844-1907] in calorimetry and protein metabolism^^® paved the way for nutrition to be described as a science of the 20th century^^^. Towards the end of this century, interests have developed beyond the pure exploration of nutritional physiology and biochemistry towards the application of this knowledge in health and disease. It is only in the last two decades, however, that the challenge of its application has been examined in the field of hepatology where much nutritional practice is still empirical rather than scientifically based.

So, what are the reasons for such a dilatory interest in nutrition and liver disease? The lack of a recognised association between the two is unlikely, having been appreciated in ancient Greece. The perception of nutritional factors as simplistic and of limited consequence may contribute in a discipline where patients are often critically ill and in urgent need of sophisticated medical and surgical care. The development of dietetics from a nursing and food-orientated, rather than scientific, background^^ undoubtedly does not help. However, the major contributory cause must be the difficulty with which nutritional investigations are carried out in this patient population. Indeed, many fundamental issues, for example the investigation of body composition and energy expenditure, upon which observations of therapeutic intervention are based, have not been successfully undertaken in patients with liver disease, even though such studies have been undertaken in healthy individuals for more than 80 years^^^.

Body composition The concept of what composes the human body was debated in ancient Greece some 2500 years ago, although the theories postulated at that time bear no relation to our present understanding of body composition. However, even then, the components of the body were considered in relation to nutrition and changes occurring in health and disease^^’"^^^. Anaxagoras of Clazomenae [500-428 BC] developed a cosmological theory based on his views of physiology and nutrition and recognised how body tissues could be built out of and replenished from bread and Introduction 19

other foods"^^^.

Plato [427-347 BC] described the body as comprising four elements - earth, air, fire and water - and wrote that when each "is produced in order ... health as a rule results; but if in the reverse order, disease"^^. Galen also described four elements or qualities, defining these as hot, cold, wet and dry, and related them to the four bodily humours of blood, phlegm, yellow bile and black bile"^^^. Interestingly, in his treatise "The Art of Medicine", Galen draws together the three notions of the liver, nutrition and body composition, describing the diagnosis of "morbid states of the liver, from deficiency or excess of the humours ... and also by those things which undergo change in the process of distribution, nutrition or the evacuation of waste substances

At the end of the twentieth century while remarkable advances have been made in the individual fields of the liver, nutrition and body composition, relatively little work has been undertaken to draw these threads together. The aims of this thesis are to achieve this by providing a scientific basis for future work which will ultimately improve patient care.

CURRENT PERSPECTIVE

Today, malnutrition is frequently associated with chronic liver disease^^^. While evidence of this relationship is increasing, much of the research upon which it is established has been based on fundamental principles which have not been validated in this patient population. The main difficulties that have precluded such validation include the presence of fluid retention^® and abnormalities of protein^^^ and mineral metabolism^®^ which violate the assumptions on which much nutrition theory and practice is based.

For example, the presence of gross ascites and oedema will render the calculation of body mass index meaningless if the patient’s actual body weight is used. If, in Introduction 20

turn, the body mass index is then used to define whether the patient is malnourished and thus at risk, subsequent conclusions and decision making will be incorrect.

A review and critical evaluation of both clinical and research methods of assessing nutritional status in patients with chronic liver disease is thus clearly needed. Criteria for undertaking clinical assessments are addressed in Chapter 1 and the design and evaluation of a reproducible assessment scheme, for use in this patient population, reported.

Development of such methodology allows the true prevalence of malnutrition and the extent of associated problems to be determined reliably in this patient population. In Chapter 2, published prevalence data are examined and the devised scheme used to survey the nutritional status of 215 patients with cirrhosis.

A reproducible method for identifying malnutrition in patients with chronic liver disease also facilitates investigation of the effects of malnutrition on outcome, establishing whether the presence of impaired nutritional status is associated with detrimental consequences or just an incidental finding in this condition. This issue is addressed in Chapter 3 which includes an investigation of the association between malnutrition and survival after liver transplantation.

While the identification of malnutrition has been described as the first step in its treatment'*^, the causes contributing to its presence in patients with chronic liver disease must also be addressed. The potential causes of malnutrition include impaired nutrient intake, increased nutrient losses and impaired utilization and these are considered in Chapter 4. At present, the data available are of variable quality and, in some cases, limited, primarily because of the innate difficulties associated with fluid retention and abnormalities of protein and mineral metabolism. To investigate the contribution of possible hypermetabolism to malnutrition, resting energy expenditure was measured in a group of patients with cirrhosis and comparable group of healthy volunteers. Introduction 21

Measurements of resting energy expenditure can be expressed per individual or relative to body weight or metabolically active tissue. The latter can be estimated using a number of bedside techniques, including anthropometry, but for research purposes a more accurate and precise method of evaluation is required. Accurate quantification of body composition would also facilitate other investigations, for example protein turnover and nutrition intervention studies, comparisons with bedside body composition techniques and further validation of clinical methods of assessment.

In Chapter 5, the validity of individual body composition techniques and the difficulties of using them in patients with chronic liver disease are reviewed. While few individual methods are considered satisfactory, multicompartment models which employ a number of different techniques offer a potential solution to the difficulties associated with assessing this patient population. To this end, an investigation of body composition using underwater weighing, dual-energy X-ray absorptiometry estimates of body mineral and total body water derived from

deuterium dilution studies was undertaken in 2 0 patients with cirrhosis.

This thesis, therefore, sets forth to establish a reproducible and valid tool for assessing clinical nutrition status in patients with chronic liver disease; to use this tool to determine the prevalence of malnutrition and then observe its association with clinical outcome variables; to consider the causes of nutritional impairment, in particular hypermetabolism; and, in order to provide supporting data for all of these, to undertake an investigation of body composition using a multicompartment model. 22

PO» 4 3

g

1 *

R

3^ '

i %Ca V

Figure i Inscribed model o f a sheep’s liver, 1900-1600 BC. British Museum, London 1998 23 CHAPTER 1

THE ASSESSMENT OF NUTRITIONAL STATUS IN PATIENTS WITH CHRONIC LIVER DISEASE

1.1 Outline The assessment of nutritional status in patients with chronic liver disease is difficult because inherent abnormalities of hydration status and protein metabolism preclude the use of many of the objective assessment techniques available. Evaluation of recent nutritional intake and anthropometric variables provide some of the more useful data available. However, even these are limited because nutrient requirements have not been clearly defined in this patient population and skinfold compressibility is altered in the presence of fluid retention. Subjective assessments of nutritional status have proved reproducible in general medical and surgical patients and a modified scheme has been developed to facilitate assessments in patients with chronic liver disease. This is based on subjective assessments alone and does not take into account anthropometric variables which would provide an estimate of body composition; concern has been expressed about its lack of objectivity. Only one small study has been undertaken to evaluate the reproducibility of the modified scheme while its external validity has not been examined.

In an initial study, 26 patients with cirrhosis were subjectively assessed by two observers using the modified scheme. Good agreement was found between the two observers’ categorization of nutritional status and moderate agreement between nutritional categories and objective nutritional variables. However, in order to obtain more meaningful assessments, three objective measures of nutritional status with good interobserver agreement - body mass index (BMI), relative triceps skinfold thickness (TSF) and relative mid-arm muscle circumference (MAMC) - were combined with subjective assessments of nutritional status in the Royal Free Hospital Global Assessment Scheme (RFH GA).

This was evaluated in a second phase by two observers who assessed 150 patients Chapter 1 Assessment of nutritional status 24 with cirrhosis using a single set of data. Improved agreement was observed between the two observers’ categorization of nutritional status and between nutritional categories and both BMI and relative MAMC but less so between nutritional categories and relative TSF.

In a third phase, relative TSF was removed from the RFH GA which was then used to assess a further 50 patients with cirrhosis with data collected independently by two observers. Good agreement was again found between observers and with BMI and relative MAMC. This scheme provides a reproducible and clinically useful picture of global nutritional status while incorporating objective variables. Further validation of this assessment scheme is required, including comparison with reference assessments of body composition and investigation of its association with clinical outcome.

1.2 Introduction Assessment of nutritional status is required to (i) identify individuals who are at risk from the complications of malnutrition^^ and who might benefit from nutritional support, and (ii) to monitor the effects of such support, other treatments and disease progression.

The association between impaired nutritional status and poor clinical outcome has been recognised for many decades^^’^"^^’^^^’^^^’^^^. Early studies of nutritional status focused primarily on deficiencies of specific vitamins^'^’^^^’'^^^’^^^. However, increasing awareness of the difficulty of evaluating vitamin status^^^’^^^ and the publication of surveys highlighting the substantial prevalence of malnutrition in hospitalized patients^^’'^®’^^^, led to a re-focusing of interest on protein-calorie malnutrition.

The main difficulty in assessing malnutrition is the absence of a single reliable method for its identification and classification. Further potential problems include the interpretation of results in the present of injury or disease. Consequently, nutritional assessment requires the collection of a combination of data on nutrient intake, physical measurements and biochemical and functional evaluations that are Chapter 1 Assessment of nutritional status 25 valid in the patient population under investigation"^. At present, most developmental studies and evaluations of nutritional assessment techniques, whether single methods or composite models, have been undertaken in general medical and surgical populations^^’^^’^^’^^"^’^^^ with relatively less data from patients with specific diseases, for example those with hepatic^^^ or renal disorders^^"^"^^^, where interpretational difficulties may be greatest.

The nutritional assessment of patients with chronic liver disease has been undertaken using a range of different techniques including the collection of data on nutrient intake 67,348 ,361,367,396 ,474^ anthropometry 88 ,235,309 ,348 ,355,361,367,401,437^ biochemical variables^^"^^^, immune function^^'^^\ muscle function^^^'^^^, and clinical appearance^^^’^^^’^^^’^^^.

1.3 Nutrient intake A patient’s recent nutrient intake is usually evaluated in the clinical setting by taking a diet history, involving face to face questioning of the patient by an observer trained in the technique^. While the diet history was originally developed to obtain information for the basis of therapeutic dietary modification^^^, the data acquired are also used to judge, usually subjectively, the adequacy of nutrient intake by comparison with estimated requirements. More sophisticated methods of estimating nutrient intake, for example using weighed dietary records^^^, influence habitual food intake^^^ and are often impractical in the clinical situation^. With the exception of patients with overt encephalopathy and some individuals abusing alcohol, where data may be umeliable^^^, there are no methodological reasons why the information obtained from a diet history taken from an individual with chronic liver disease should not be valid.

For nutritional assessment purposes, intake data should be compared with the patients’ nutrient requirements so that its relative adequacy can be judged. In patients with chronic liver disease, information on nutritional requirements is limited. While there is good evidence that protein requirements are increased^^^’^^^’^^^ the data on resting energy expenditure (REE) are less Chapter 1 Assessment of nutritional status 26

Estimates of energy requirements can be used to assess the adequacy of energy intake. However, standard formulae for predicting REE^°^ provide mean estimates of up to 13% below measured values in patients with chronic liver disease^^'^’'^^^’'^^^. Consideration of increased energy expenditure is, therefore, necessary when assessing the adequacy of j energy intake in this patient population.

1.4 Anthropometry The status of body fat and muscle stores can be estimated using non-invasive anthropometric measurements which are quick and simple to perform. The most frequently used variables include body mass index^^ (BMI), skinfold thicknesses^, and mid-arm muscle circumference^^^ (MAMC).

Quetelet"^^ first observed the relationship between weight and height in healthy adults in 1869. The ratio:

weight (kg) / height^ (m^) was devised by Keys and colleagues^^^ and termed body mass index (BMI). Adequately nourished, non-obese individuals ideally have a BMI between 20 and 25 kg.m"^ The calculation of BMI is used as a routine screening tool in the assessment of nutritional status and individuals with a BMI of less than 20 are considered at risk from malnutrition^^^.

Values of BMI may be less useful in patients with chronic liver disease where accurate measures of weight are difficult to obtain in the presence of fluid retention or, in children, an enlarged liver or spleen. However, an estimate of dry weight, that is body weight minus a correction for retained fluid based of the patient's clinical history, previously documented weights and published guidelines^"^^, may facilitate the calculation of a more accurate BMI. The accuracy and reproducibility of this method of deriving BMI have not been tested and while it may be inferior to actual measurements in patients without fluid retention. Chapter 1 Assessment of nutritional status 27 corrected BMI provides a simple screening tool for use in this patient population.

Measurements of skinfold thicknesses have been used to assess subcutaneous fat in the evaluation of obesity and undernutrition for over fifty years^’^"^^’^^ and these can be compared against reference data obtained from large populations of healthy volunteers^^’^^^’^^. Measurements made at the triceps site are those most frequently undertaken and are considered the most useful and accurate for predicting body fat"^^^. From the triceps skinfold thickness (TSF) and mid-arm circumference (MAC), an estimate of MAMC can be calculated^^^:

MAMC = MAC - (TSF X 0.3142).

Published standards, stratified for age and sex, are also available for comparison of measurements of MAC and MAMC^^’^^^’^^. Estimates below the 5th percentile of the appropriate population band suggest nutritional depletion^^^.

In patients with chronic liver disease the presence of subcutaneous oedema may confound measurement of skinfold thicknesses. In practice, this is difficult to evaluate, because of both the inherent errors associated with remeasurement^^^ and because changes in subcutaneous fluid retention occurring over time may be accompanied by changes in fat stores. Skinfold thicknesses measured at the triceps and biceps sites are less likely to be affected by oedema than those on the lower limbs, although even the triceps site may be affected if the patient remains supine for long periods.

Laven^^^ examined the compressibility of skinfolds in young children with oedema and found a significantly greater increase in compression between 5 and 60 seconds after the calipers were applied compared to measurements in healthy children. Becque and colleagues^^ used skinfold calipers with an electronic readout to assess the dynamics of skinfold measurements and concluded that initial, rapid compressibility represented the expression of interstitial water from the skinfold while the subsequent and smaller compression was due to the deformation of the Chapter 1 Assessment of nutritional status 28 tissue by the pressure of the calipers. It may be suggested that a later reading of the caliper might be more accurate in oedematous individuals but further studies are required to confirm this. Measurements made of oedematous skinfold thicknesses will result in an overestimate of fat stores and thus true depletion of fat stores may not be recognised. Conversely, measurements of MAMC calculated from oedematous skinfold thicknesses may be underestimated and the true depletion of muscle stores exaggerated.

Skinfold measurements can also be used to estimate total body fat using prediction formulae based on comparisons with underwater weighing^"^^’^^^’^^^’^^^. These formulae have not been validated in patients with chronic liver disease but in the absence of such validation the method of Durnin and Womersley^"^^ is probably the one best suited for use in these patients. This uses the logarithmic sum of skinfolds measured at the triceps, biceps, subscapular and suprailiac sites, and avoids measurements on the lower limbs and waist circumferences which are included in other models^^^’^^^’^^^ and more likely to be adversely affected by fluid retention. However, even the Durnin method may produce inaccurate results in patients with gross ascites as the measurement of skinfolds at the suprailiac site may be difficult or even precluded by abdominal distension.

Nevertheless, in spite of these difficulties, anthropometric assessment has been widely used to estimate nutritional status in patients with chronic liver Loguerico and colleagues^^^ found no significant intra- or interobserver differences between repeat measurements of skinfold thicknesses and MAC in 80 patients with cirrhosis, although individuals with fluid retention were specifically excluded from the study. However, Camilo and colleagues^^ included ascitic patients in their study comparing TSF with estimates of body fat and corrected arm muscle area with body cell mass (BCM) determined by multiple dilution studies. They found significant correlations between both sets of variables and concluded that this validated the use of these measurements in the assessment of nutritional status in this patient population. It is difficult to appraise this conclusion as the study has only been published in Chapter 1 Assessment of nutritional status 29 abstract form.

Anthropometry may play a useful role in the assessment of nutritional status in this patient population providing that the difficulties inherent in the use of skinfold thicknesses, namely their potential for over-estimating fat stores and under­ estimating muscle stores, are appreciated.

1.5 Biochemical variables Depletion of body protein stores is generally accompanied by a fall in the concentration of circulating proteins, including albumin, pre-albumin, transferrin and retinol binding protein^^'^ "^^^. Measurement of the circulating concentrations of these proteins is used to estimate protein stores'^. However, plasma proteins are synthesized by the liver and thus concentrations fall in the presence of decompensated liver disease^^\ Protein synthesis is also directly inhibited by alcohof^^^. As a result, depleted concentrations of plasma proteins cannot be used as a marker of malnutrition in patients with chronic liver disease or those actively abusing alcohol.

The creatinine height index, which compares the urinary excretion of a patient with that of a healthy individual of the same height, has been proposed as a sensitive measure of protein status^^. Creatinine is a breakdown product of creatine, which is synthesized from methionine, glycine and arginine in the liver, and plays a subsidiary role in the production of ATP^"^. Creatine is stored in skeletal muscle and when muscle stores are depleted, creatine breakdown, and hence creatinine excretion, is reduced^^\ Patients with cirrhosis excrete low levels of urinary creatinine'^^^’'^^® which may reflect depletion of skeletal muscle and / or impaired hepatic production of creatine.

Pirlich and colleagues'^^ compared estimates of muscle mass derived using 24-hour urinary creatinine excretion and anthropometry in 102 patients with cirrhosis. Although significant differences were observed between estimates of muscle mass determined by creatinine excretion and anthropometry, the authors still concluded Chapter 1 Assessment of nutritional status 30 that creatinine excretion could be used to predict muscle mass in this patient population, unless renal function was impaired. They also concluded that creatinine excretion was independent of hepatic function although the comparisons of urinary creatinine were made between subgroups of patients who were neither matched for age or sex. These conclusions are premature.

There is insufficient evidence to recommend the use of the creatinine height index as a marker for depletion of muscle stores in patients with chronic liver disease.

1.6 Immune function The immune response is adversely affected by malnutrition as the proportion and absolute number of T cells are reduced^"^ and the function of T helper cells is impaired^^. Starvation is also associated with deficiencies of fibronectin and secretory immunoglobulin while concentrations of serum immunoglobulin G may increase in response to severe undernutrition^^.

Immunocompetence can be tested by assessing delayed cutaneous hypersensitivity which involves the intradermal injection of a number of antigens, for example Candida or streptokinase-streptodornase, and subsequent measurement of the diameter of any indurations occurring at the site of injection within 48 hours^^. Indurations exceeding 5mm constitute a positive test"^; negative responses to more than one of five antigens suggest the patient is anergic and, in some circumstances, may be indicative of malnutrition^^.

Testing of delayed cutaneous hypersensitivity has been undertaken in patients with chronic liver disease^^'^^'^^^'^^'^^^'"^°^ and while high levels of anergy have been observed^^’^^’"^^^ the association with other markers of nutritional status was poor. Patients with chronic liver diseases have impaired reticuloendothelial and granulocyte function, reduced concentrations of complement and impaired cell- mediated immunity'*^^. Thus, there is no value in using immune function to assess nutritional status in this patient population. Chapter 1 Assessment of nutritional status 31

1.7 Muscle function Malnourished patients have depleted muscle protein stores^^^ and while anthropometric measurements can be used to quantify these directly, independent functional assessments of muscle strength can also be undertaken. Assessment of the muscular response of the adductor pollicis to electrical stimulation^"^^ has shown that malnutrition is associated with increased muscle fatiguability which can be reversed by nutritional support^^\

Lafleur and colleagues^^^ used the same electrical stimulation test^"^^ to assess adductor pollicis function in 74 patients with cirrhosis. They found no significant differences between the muscle function of patients with cirrhosis who were well nourished and healthy controls, or between malnourished patients with cirrhosis and malnourished patients with other gastrointestinal disorders. They concluded that muscle function is not affected by cirrhosis per se and thus its measurement may be useful in assessing nutritional status in this patient population. However, they did not test the independence of the relationship between nutritional status and muscle function from the severity of liver disease.

Tarter and colleagues^^^ reached a different conclusion in their study of concentric and eccentric muscle force measured using a Kinetic Communicator^"^. They observed significant differences in isokinetic muscle strength between patients with cirrhosis and healthy volunteers which were independent of body weight and lean body mass (LBM) determined by bioelectrical impedance analysis (BIA). They concluded that the observed impairment of muscle function was a reflection of cirrhosis per se.

Muscle function can also be assessed using hand-grip dynamometry. Assessments of grip-strength in healthy volunteers correlate significantly with forearm muscle circumference and MAMC^^^ and in surgical patients grip-strength has been used to predict post-operative complications^^^’^^^. While hand-grip dynamometry is non- invasive and simple to perform, it cannot be used in severely ill or unconscious patients or those with local pathology^. Chapter 1 Assessment of nutritional status 32

With these exceptions, dynamometry can be undertaken in patients with chronic liver disease although assessments may be difficult to interpret in the presence of either acute or chronic alcoholic myopathy^^^ "^^^. Dynamometry has been used in liver transplant candidates^^^ where pre-operative hand-grip strength was predictive of post-operative complications. Mendenhall and colleagues^^^ also reported a significant relationship between survival and hand grip strength in 77 patients with alcoholic hepatitis.

While using muscle function tests to assess nutritional status is appealing because of the simplicity of the measurements, the limited but conflicting data published to date suggest further research is required.

1.8 Clinical nutritional status Nutritional status can also be assessed clinically by taking a routine history which includes details of weight changes, symptoms likely to influence food intake and recent illnesses and infections, and by a physical examination where evidence of muscle wasting, loss of subcutaneous fat, oedema and signs of vitamin deficiency, for example cheilosis and glossitis, are noted (Table 1.1)^^’^^^. On the basis of these data, but without recourse to a scoring system, the observer categorizes the patient as adequately nourished or mild/moderately or severely malnourished. This method of subjective global assessment (SGA) was first formally tested by Baker and colleagues^\ who reported an interobserver agreement of 81% in nutritional category in 59 general surgical patients. The nutritional categories were significantly correlated with several objective measures of nutritional status including serum albumin and body weight expressed as a percentage of ideal weight. Subsequent studies in large numbers of general medical and surgical patients have confirmed the reproducibility of this method^^^’^^^.

Nutritional status has been assessed clinically in patients with chronic liver disease in a number of studies^^^’^^^’^^^ but was first formally evaluated by Hasse and colleagues^^^ who modified the original SGA method^^ for use in liver transplant candidates. The method was modified to include clinical data specifically relevant Chapter 1 Assessment of nutritional status 33 to patients with liver disease, for example, encephalopathy, ascites and gastrointestinal bleeding. The interobserver agreement in a series of 20 patients was 80%, with muscle wasting and loss of subcutaneous fat the strongest contributors to the final category of nutritional status. This method of assessment has not, however, been validated against any external variables nor has the interobserver agreement been tested by other workers.

Naveau and colleagues^^^ cast doubt on the reliability of using purely subjective clinical data to assess nutritional status in patients with chronic liver disease as they observed a poor correlation between clinical assessments, based on patients’ general appearance, and nutritional status determined using an anthropometric score. Further studies are required to confirm the value of subjective assessment in this patient population.

1.9 Other assessment methods The prognostic nutrition index (PNI), which was based on stepwise linear regression analysis of a number of nutritional variables, provides a global but objective assessment of nutritional status in surgical patients^^’^^"^ and an estimate of the risk of post-operative complications:

PNI = 158 -(ALB X 16.6)-(TFNx 0.20)-(TSFx 0.78)-(DCHx 5.8) where ALB is serum albumin (g/100 ml), TFN serum transferrin (g/100 ml), TSF (mm) and DCH delayed cutaneous hypersensitivity to any of three antigens, where 0 = non-reactive; 1 = 0-5 mm induration; and 2 = >5 mm induration. Patients with an index exceeding 60 are at severe risk of post-operative complications.

Although the index has proved useful in accurately predicting risk of post-operative complications in general surgical patients^^ it is not predictive in liver transplant candidates'^. This is not surprising as three of the four index variables - serum albumin, serum transferrin and delayed cutaneous hypersensitivity - are not useful markers of nutritional status in this patient population^^^'^^^. Chapter 1 Assessment of nutritional status 34

However, Higashiguchi and colleagues^^^ devised a predictive scoring system - the prognostic nutrition index for surgery (PNIS) - specifically for use in patients with hepatic dysfunction. They assessed 102 patients, of whom 34% had cirrhosis, 61% obstructive jaundice and 5% chronic hepatitis, before elective hepatobihary or pancreatic surgery and compared post-operative complications with pre-operative variables using stepwise linear regression. The most predictive variables were identified as percentage ideal body weight (IBW), percentage weight change (WC), percentage TSF (%TSF) and hepaplastintest (HP), a commercial test of haemostasis. Their relationship to risk of complications was described as:

PNIS = (IBWx 0.046)-(WC x 0.147) + (%TSF x 0.01) + (HP x 0.051).

Individuals scoring less than five are predicted to be at high risk of post-operative complications while those with a score above ten are classified as having a good prognosis.

The index proved reliable when tested in a further cohort of similar patients. More studies are required in patients with parenchymal liver disease, for example liver transplant candidates, to evaluate this index further.

Nutritional status has also been assessed in patients with chronic liver disease using composite methods of assessment^’^^^’^^, in which patients have been categorized on the basis of anthropometric, biochemical and immune function variables (Table 1.2). While the theory of combining different aspects of nutritional assessment is good, addressing, as it does, the absence of a single reliable assessment method, each of the contributing techniques must be valid if the composite method is to be of value. The inclusion of biochemical and immune function variables in the composite methods used to date, preclude their recommendation for use in this patient population.

1.10 Summary In patients with chronic liver disease, there are inherent difficulties, predominantly Chapter 1 Assessment of nutritional status 35 relating to fluid retention and the hepatic synthesis of protein, which preclude assessment of nutritional status by many of the standard methods valid in other patient populations. While evaluation of recent nutrient intake and anthropometric measurements may prove useful, consideration must be given to the limitations associated with ill-defined | energy requirements and the effects of fluid retention on tissue compressibility. Good interobserver agreement in the clinical evaluation of nutritional status suggests that this method may also be useful, either alone or in conjunction with other objective measurements, but requires external validation. With these provisos, it is recommended that assessment of nutritional status in patients with chronic liver disease in the clinical setting can best be achieved using a combination of nutrient intake data, anthropometric measurements and clinical evaluation. Further studies are required to confirm this.

1.11 STUDY 1; THE DEVELOPMENT OF A NEW GLOBAL NUTRITIONAL ASSESSMENT METHOD FOR USE IN PATIENTS WITH CHRONIC LIVER DISEASE

1.11.1 Aims (i) To assess the reproducibility of a modified SGA scheme^^^ and to check its convergent validity against anthropometric data; (ii) to use the modified SGA as a basis for devising a new global assessment technique incorporating anthropometric variables, and (iii) to assess the reproducibility and validity of the newly devised method.

The study was undertaken in three separate phases over a 20 month period; each phase was completed and the data analyzed before proceeding to the next phase.

1.11.2 Patients, Method and Results Phase 1 Reproducibility and validity of the modified SGA Patients and methods: The study population comprised 26 patients, 14 men and 12 women of mean (range) age, 50.4 (34-66) years, with cirrhosis of varying aetiology and severity (Table 1.3). The aetiology of the liver injury was determined using Chapter 1 Assessment of nutritional status 36

historical, clinical, laboratory and histological variables, whilst the functional severity of the liver injury was assessed using Pugh’s modification of Child’s grading system"^^.

Patients were assessed independently by two observers experienced in the nutritional management of patients with chronic liver disease. The two assessments were undertaken within 4 days of each other using the modified SGA format as published^^^; the data were separately recorded. Nineteen patients were assessed as inpatients, seven as outpatients.

Clinical information: Clinical information was obtained directly from the patients and from their medical records. Gastrointestinal symptoms which might influence nutrient intake, such as anorexia, nausea, vomiting, difficulty in mastication, dysphagia, indigestion or food-related abdominal pain were recorded and graded, in relation to their degree, frequency and duration, as absent, mild, moderate or severe. Bowel habits, and any recent changes in stool frequency, colour or consistency were recorded. A history and details of infections, renal dysfunction, hepatic encephalopathy and gastrointestinal bleeding were sought and detailed. Pre-illness weight and weight range in the preceding 6 months were recorded and the overall weight change during this period calculated. Finally, patients were questioned about recent levels of activity and fatigue to determine whether their capacity for work and muscle endurance was reduced; dysfunction, when present, was further characterized in relation to its severity and duration.

Nutrient intake: The patient’s appetite was assessed as good, fair or poor and any factors likely to interfere with food intake, for example, early satiety or taste changes were noted. Recent dietary intake was assessed using an established diet history method^ supplemented, where necessary, with additional information from relatives, nursing staff and food record sheets. Details of dietary restrictions and oral, enteral or parenteral nutritional support were recorded. The data obtained were not intended to provide a quantitative evaluation of intake but an idea of the overall adequacy of the diet in relation to estimated requirements assessed using Chapter 1 Assessment of nutritional status 37 the Harris-Benedict equations^®®. Energy intake was classified as adequate, if it met estimated requirements, inadequate, if it failed to meet estimated requirements but exceeded 500 kcal/day, or negligible if it provided less than 500 kcal daily.

Physical status: A subjective evaluation of subcutaneous fat stores - good, fair, poor - and the presence of muscle wasting and fluid retention - none, mild to moderate, severe - was made by visual inspection. Height was measured to the nearest centimetre with patients standing in socks or bare feet, using a wall mounted or free-standing measure (Seca, Hamburg, Germany). Patients were weighed to the nearest 0.1 kg, wearing night-wear or light indoor clothing but without shoes, using a weighing scale (Seca, Hamburg, Germany) or a seat balance scale (Marsden W/M Group, London, UK). No adjustment was made for fluid retention. Ideal body weight was calculated from the patients’ height based on a BMI of 22.5 kg.m'^

529

Global assessment: Each dietitian categorized the patients using the data they had themselves collected as: (1) well-nourished; (2) moderately malnourished (or suspected of being so); (3) severely malnourished.

Anthropometric validation: Once the patients had been classified using the modified SGA, they were assessed anthropometrically and the data separately recorded. An estimated dry weight was determined by deducting a weight for ascites and/or oedema based on clinical assessment, previously documented weights and published guidelines'^. BMI was calculated from the estimated dry weight.

MAC and TSF were measured^^’^"^^’^"^^ using Holtain/Tarmer-Whitehouse skinfold calipers and a steel tape measure (Holtain Ltd, Crymych, Dyfed, UK). Measurements were undertaken on the non-dominant side of the body using the writing hand to define dominance unless anatomical abnormalities or the siting of intravenous infusions precluded measurement on that side. The mid-point of the Chapter 1 Assessment of nutritional status 38 upper arm was marked between the acromion process of the scapula and the olecranon process while the patient held their forearm horizontally with the palm of the hand facing upwards. MAC was then measured in a horizontal plane at the site of the mid-arm mark with the arm hanging at full extension with the palm facing inwards. Care was taken not to distort the skin surface while the measurement was made.

With the patient’s arm remaining in the same position, a skinfold was formed over the belly of the triceps muscle and parallel to the humerus, using the thumb and index finger of the observer’s non-dominant hand. Care was taken not to include underlying muscle and TSF was measured at the mid-arm mark while continuing to hold the skinfold. Three readings were taken to the nearest 0.2 mm, 2 or 3 seconds after applying the calipers and a mean value calculated. MAMC was also calculated^^^.

Measurements of MAMC and TSF were compared with published standards for the appropriate sex and age^^ and expressed both in relation to the 5th percentile and as a percentage of the 50th percentile, described as the relative measurements.

Intraobserver repeatability of anthropometric assessment was evaluated by remeasurement of variables within 5 days of each other in five patients.

Statistical analysis: Statistical analysis was undertaken using the Minitab software package (1994, State College, PA, USA). Repeatability coefficients'^^ were calculated for anthropometric variables for each of the observers and expressed as a percentage of the mean of the repeated values. Interobserver agreement in anthropometric measurements was evaluated using the intraclass correlation coefficient, rho. Comparison of interobserver global categorization of nutrition was undertaken using the kappa statistic, a measure of interobserver reliability that controls for the agreement expected by chance. Rho and kappa values between 0.41 and 0.60 were considered to indicate a moderate association and above 0.60 to indicate a substantial relationship^^^. Spearman rank correlations were used to Chapter 1 Assessment of nutritional status 39

examine the relationships between contributing variables, including anthropometric measurements and| energy intake, and global categorization of nutrition. Correlation coefficients above 0.39 were considered to indicate a significant association for the sample of 26 patients at the 5% levef^^.

Results: There was agreement between observers in the nutritional status of 21 (81%) of the 26 patients using the modified SGA (kappa = 0.67). Where there was discordance, it was by a single category and without bias.

The intraobserver repeatability coefficients for anthropometric variables ranged from zero to 8.0% (Table 1.4).

All anthropometric variables showed excellent interobserver agreement with intraclass correlation coefficients exceeding 0.8 (Table 1.5). There was moderate agreement between evaluation of energy intake (rho = 0.42).

Correlation between the SGA and BMI, relative TSF and relative MAMC yielded r values of -0.30, -0.48 and -0.53 respectively (Table 1.5).

Thus, the modified SGA method of assessment was reproducible between two observers, and significantly associated with relative TSF and relative MAMC. However, it did not take explicitly into account anthropometric measurements which are considered useful in providing quantitative estimates of body composition. Concerns have been raised about the use of subjective assessment alone in this patient population as it fails to identify as malnourished some individuals with depleted anthropometric measurements^^^.

Therefore, in order to develop an improved assessment scheme, three objective variables - BMI, relative TSF and relative MAMC - were incorporated. These variables related significantly to the modified SGA and all had high interobserver agreement. Chapter 1 Assessment of nutritional status 40

Fhas^ 2 Derivation of New Global Assessment Tool Patients and methods: An assessment scheme was devised to categorize nutritional status in patients with chronic liver disease on a semi-structured basis determined by a combination of subjective data and the three objective variables.

Data were collected on gastrointestinal and other clinical symptoms, nutrient intake and physical status. Height, weight, MAC and TSF were measured, dry weight estimated and, from these, BMI, MAMC, relative MAMC and relative TSF were calculated. All the findings and measurements, together with derived variables and demographic and clinical information, were collected as described above but documented on a single data collection form.

Patients were then categorized using the BMI, relative TSF, relative MAMC and nutrient intake data on an algorithmic basis as (Figure 1.1): (1) well-nourished; (2) moderately malnourished (or suspected of being so); (3) severely malnourished. A subjective override was incorporated, based on the other data collected, to allow factors, such as profound recent weight loss, severe steatorrhoea or aggressive nutritional intervention which may effect nutritional status, to be taken into account. This override allowed the assessor to change nutritional class by one category only.

In the first assessment of this phase, one observer assessed a group of patients and documented all relevant data on anonymized collection sheets without recording the derived nutritional categorization. The second observer then reviewed the data sheets and categorized the patients’ nutritional status on the basis of the recorded information only; no access was allowed to the patients or their medical records.

The study population comprised 150 patients (87 men: 63 women; mean age 50.7 [26-75] years) with cirrhosis of varying aetiology and severity (Table 1.3). One hundred and nine (73%) patients were assessed as inpatients, 41 as outpatients. Chapter 1 Assessment of nutritional status 41

Comparisons were made between global categorization of nutritional status by the two observers using the kappa statistic, Spearman rank correlations were used to examine the relationship between contributing variables and global categorization.

Results: There was agreement in the categorization of nutritional status in 145 (97%) of patients using the devised scheme yielding a kappa value of 0.94 for the whole group. Where discordance occurred it was by a single category; four patients were categorized as more malnourished by the first observer and one by the second. The degree of agreement varied little in subgroups divided on the basis of gender, aetiology or severity of liver disease (kappa = 0.88-1.00) (Table 1.6). The relationships between the global categorization and BMI, relative TSF and relative MAMC yielded r values of -0.69, -0.47 and -0.68 respectively.

Compared to the modified SGA, the new Royal Free Hospital global assessment scheme (RFH GA) provided an improved interobserver categorization of nutritional status which was more closely associated with BMI and relative MAMC. Based on these results, the assessment scheme was modified by excluding relative TSF from the algorithm for the next phase.

Phase 3 Reproducibility and validity of the devised global assessment RFH GA method Patients and methods: The study population comprised 50 patients, 34 men and 16 women of mean (range) age 50.1 (27-70) years with cirrhosis of varying aetiology and severity assessed as previously (Table 1.3). All were inpatients at the Royal Free Hospital at the time of assessment.

All patients were assessed by two observers working independently without exchange of objective or subjective data other than the name and location of the patients in the study. Assessments were undertaken within 7 days of each other. The data collected were used to determine nutritional status using the modified algorithm of the RFH GA (Appendix). Chapter 1 Assessment of nutritional status 42

The data were analysed as described in Phase 1. Where the subjective override had been employed to change the final category of nutritional status, the factors influencing the decision were examined using unpaired Student’s t-tests and Chi- squared analyses.

Results: There was agreement in the categorization of nutritional status in 44 (88%) of the 50 patients (kappa = 0.79). Where discordance occurred it was by a single category with four patients categorized as more malnourished by the first observer and two by the second.

Excellent interobserver agreement was observed for BMI and relative MAMC (rho = 0.93 and 0.90). The correlation between global categorization and BMI and relative MAMC yielded r values of -0.78 and -0.69 respectively (Table 1.7).

The subjective override was employed in 9/50 and 8/50 of the patients by the two observers respectively. In only one patient was the override: used by both observers. Thirteen patients were categorized as more malnourished than previously while four were categorized as less malnourished. Patients categorized as more malnourished had a greater mean weight loss and reported more symptoms than those who were categorized as less malnourished although in neither instance was the difference significant. As a consequence of using the override, agreement in categorization was achieved in 11 patients which had previously been discordant, while in the remaining five patients the override changed agreement in categorization to discordance.

1.11.3 Discussion Hasse’s SGA method, developed specifically for liver transplant candidates, has been used to assess the nutritional status of large numbers ( > 500) of patients^^^. However, its reliability has been tested in just 20 individuals, 12 of whom had cirrhosis, with a outcome of 80% agreement and a Cronbach coefficient a-test, examining internal validity, of 0.707 This evaluation was undertaken by the authors who devised the scheme and not repeated until now. Chapter 1 Assessment of nutritional status 43

The results from Phase 1 of the present study show that this method is reproducible between two observers and that the global categories of nutritional status are moderately associated with anthropometric variables. However, the merit of using only subjective measures of assessment in patients with liver disease has been questioned by Naveau and colleagues^^^ who reported that a clinical evaluation of nutritional status in 636 patients with alcoholic liver disease underestimated malnutrition as defined by anthropometry in 51% of patients. Mid-arm muscle area and mid-arm fat area are significantly associated with the survival of patients with cirrhosis^^^ and thus the inclusion of anthropometric variables in the global assessment has a rational basis.

In Phase 2 of the present study, the combination of specific objective and subjective variables in the semi-structured framework of the RFH GA increased the agreement of nutritional categorization between the two observers when categorization was based on a single set of data. This was despite the fact that only one observer interviewed the patients and collected the data while the second investigator simply used the documented data to categorize the patients. While BMI and MAMC were closely associated with global categorization, there was less, albeit still significant, association with relative TSF and, therefore, for subsequent assessments this variable was removed from the categorization scheme.

The results from Phase 3 showed the RFH GA gave good interobserver agreement. The possibility that improved agreement occurred as a result of a training effect over the duration of the investigation must be considered. Data were systematically collected in the three phases over a 20 month period. No drift was found in agreement between observers in the early stages of each phase compared to the later stages. It is, therefore, considered unlikely that a training effect alone would be responsible for the improved agreement observed.

It is likely that the proposed RFH GA method of assessing patients with liver disease will be more closely associated with the anthropometric variables than Hasse’s SGA method because these variables were included in the RFH GA Chapter 1 Assessment of nutritional status 44 scheme.

The techniques employed in the proposed scheme are simple to perform and the equipment required are readily available. Calculation of BMI and comparison of anthropometric data with standards can be undertaken at the bedside to provide immediate results. The subjective assessment of the adequacy of 1 energy intake in relation to estimated energy requirements may also be undertaken with a minimum of equipment and calculation but does require a reasonable degree of skill and experience which should be within the ability of all trained dietitians.

Study conclusions: The RFH GA for assessing nutritional status in patients with chronic liver disease has been developed which is reproducible and provides a clinically useful picture of global nutritional status. However, further validation studies are required which should include verification by other workers, comparison with assessments of body composition using reference methods, and examination of the predictive value of the scheme in relation to clinical outcome. 45

Table 1.1

Data collection sheet for assessment of Subjective Global Assessment after Baker et al^^’.

Select appropriate category with a checkmark or enter numerical value where # indicated ; by

A HISTORY 1 Weight change Overall loss in past 6 months: amount = # ...... kg; % loss = #...... Change in past 2 weeks increase, no change, decrease. 2 Dietary intake change (relative to normal) no change change duration = #...... weeks Type: ...... suboptimal solid diet, ...... full liquid diet hypocaloiic liquids, ...... starvation 3 Gastrointestinal symptoms for (>2 weeks) none, nausea, vomiting, diarrhoea, anorexia 4 Functional capacity no dysfunction (e.g. full capacity) dysfunction...... duration = #...... weeks Type: ...... working suboptimally ambulatory bedridden 5 Disease and its relationship to nutritional requirements Primary diagnosis (specify)...... Metabolic demand (stress): no stress, low stress, moderate stress, high stress.

B PHYSICAL (for each trait specify: 0 = normal, 1+ = mild, 2+ = moderate, 3+ = severe) #...... loss of subcutaneous fat (triceps, chest) #...... muscle wasting (quadriceps, deltoids) #...... ankle oedema #...... sacral oedema #...... ascites

C SGA RATING (select one) A = well nourished, B = moderately malnourished (or suspected of being), C = severely malnourished. Table 1.2

The assessment of nutritional status in patients with chronic liver disease using composite methods.

First author Population Cirrhosis Global categories of Definition of categories Year (n) (%) nutritional status

MendenhalP'’® 1984 alcoholic hepatitis 64 Normally nourished No abnormality detected (284) Kwashiorkor ■i Serum albumin, i transferrin, lymphocytopaenia, i DCH

Marasmus i LBM from CHI, i % body weight, DCH

Kwashiorkor - marasmus Depletion of variables from both types

Nutritionally abnormal Depletion presence but not diagnostic

Abad* 1987 chronic liver disease 100 Adequately nourished ) All categories empirically defined on the basis of (90) Moderately malnourished ) TSF, MAMC, serum albumin;

Severely malnourished ) "greatest weight" given to MAMC

Lautz^ 1992 chronic liver disease 100 Normally nourished No abnormality detected (123) Kwashiorkor X Serum albumin, "normal" body composition

Marasmus ■i BCM, i TSF, i body weight

Kwashiorkor-marasmus Depletion of variables from both types

CHI - creatinine height index; DCH - delayed cutaneous hypersensitivity.

Os Table 1.3

Demographic details of the patients with cirrhosis included in the three phases of the study.

Variable Phase 1 Phase 2 Phase 3

(n=26) (n=150) (n=50)

Mean (range) age (yr) 50.4 (34-66) 50.7 (26-75) 50.1 (27-70)

Male : female 14:12 87:63 34:16

Aetiology of cirrhosis (%) alcohol 61 59 68

post-viral 23 20 18

biliary 8 9 6

autoimmune 0 4 4

others 8 8 4

Functional s e v e r i t y (%) Grade A 8 31 8

Grade B 42 32 22

Grade C 50 37 70 48

Table 1.4

Repeatability coefficients of anthropometric variables measured by the two observers

(percentage of mean repeated values).

Variable Observer 1 Observer 2

Height (m) 0.004 (0.2%) 0

Actual weight (kg) 0.214 (0.3%) 1.017 (1.7%)

BMP (kg.m ") 0.134 (0.6%) 0.200 (0.9%)

TSF (mm) 0.869 (8.0%) 0.949 (6.1%)

MAC (cm) 0.363 (1.4%) 0.214 (0.8%)

MAMC (cm) 0.422 (1.9%) 0.300 (1.4%)

“BMI calculated from actual weight with correction for fluid retention. 49

Table 1.5

Interobserver agreement and relationship with modified SGA for anthropometric variables and nutrient intake in 26 patients with cirrhosis in Phase 1.

Variable Intraclass Spearman rank

correlation between correlation

two observers with SGA category'’

(rho) (r)

Height 0.96 +0.08

Actual weight 1.00 -0.13

BMP 0.97 -0.30

MAC 0.89 -0.71

TSF 0.80 -0.58

MAMC 0.88 -0.57

Relative TSF 0.86 -0.48

Relative MAMC 0 88 -0.53

Energy intake 0.42 +0.31

^BMI calculated from actual weight without correction for fluid retention

'’Mean value from two observers. 50

Table 1.6

Agreement in categorization of nutritional status using the RFH GA in 150 patients with cirrhosis in Phase 2.

Variable Population subgroups Interobserver

agreement

(kappa)

Gender Male (n=87) 0.96

Female (n=63) 0.92

Aetiology of liver disease Alcohol (n=89) 0.95

HBV/HCV (n=30) 1.00

PBC/PSC (n=14) 0.88

Severity'’"'^ of liver disease Decompensated (n=88) 0.98

Compensated (n=62) O j#

Total population (n=150) 0.94

HBV/HCV - hepatitis B or C virus; PBC - primary biliary cirrhosis; PSC - primary sclerosing cholangitis. 51

Table 1.7

Interobserver agreement and relationship with RFH GA for anthropometric variables and nutrient intake in 50 patients in Phase 3.

Variable Intraclass Spearman rank

correlation between correlation with

two observers RFH GA category*"

(rho) (r)

Height 0.96 +0.09

Weight 0.99 -0.47

Estimated dry weight 0 9 6 -0.60

Weight loss 0.26 -0.34

BMP 0.93 -0.78

MAC 0.92 -0.76

TSF 0.72 -0.50

MAMC 0.94 -0.64

Relative TSF 0.74 -0.46

Relative MAMC 0.90 -0.69

1 Energy intake 0.40 +0.46

®BMI calculated from estimated dry weight

^Mean value from two observers BMI

>20 <20

MAMC± TSF M AMC±TSF «5^ > 5th % ile < 5 th % lie > 5th % ile < 5 th % ile

Dietary Intake Dietary Intake Dietary Intake Dietary Intake

Adequate Inadequate Negligible Adequate Inadequate Negligible Adequate Inadequate Negligible Adequate Inadequate Negligible

Adequately Moderately Malnourished Severely Malnourished Nourished or Suspected of Being

Subjective Override

Figure 1.1 Algorithm used in Study 1 Phase 2 for the categorization of nutritional status in patients with cirrhosis to 53 CHAPTER 2

THE PREVALENCE OF MALNUTRITION IN PATIENTS WITH CHRONIC LIVER DISEASE

2.1 Outline The prevalence of malnutrition in patients with chronic liver disease varies with the population studied and the methods and comparative references used. There is evidence, from a number of published studies, of anthropometric depletion in up to 55% of patients with up to 80% globally or clinically assessed as malnourished. These studies also showed that increasing severity of liver disease is associated with a higher prevalence of malnutrition but reported less evidence of an association between aetiology of liver disease and malnutrition. In order to quantify the prevalence of malnutrition in patients with chronic liver disease at the Royal Free Hospital, 215 patients with cirrhosis were assessed globally using a reproducible method that incorporated objective and subjective variables. Evidence of malnutrition was substantial with 41% of patients having an inadequate or negligible intake, 51% exhibiting anthropometric variables indicative of depletion and 57% assessed globally as malnourished. A higher prevalence of malnutrition was observed in patients with the most severely decompensated liver disease. The results of this study confirm previous findings and indicate that a considerable number of patients with cirrhosis are at risk from complications associated with malnutrition.

2.2 Introduction The prevalence of malnutrition in patients with chronic liver disease is variously quoted as being between 10 and 100% 2 5 5 ,3 2 2 ,3 7 7 figures reflect the diversity of the patients examined and the methods of assessment used in the studies p u b l i s h e d t o d a t e 1.10-54,67,88,115,134,211,235,282,287,309,348,355,361,365,376,390,392,393,401,419,420,433,456,474,

487,495,527,559,565

The patients studied have varied from de-socialised alcoholics^^^’^^^ to liver transplant candidates^®’^^’"^^^’'^^^ and the assessment techniques used have included Chapter 2 Prevalence of malnutrition 54

evaluation of individual vitamin status^^'^^, and global assessments using both objective^^^ and subjective tecbniques^^^. Porayko and colleagues'^^ warn about the difficulties of defining the degree of nutritional depletion in this patient population but acknowledge that heightened awareness of potential nutritional problems is important if they are to be identified and treated.

It is with this in mind that a number of studies have been examined to try to provide more focused data on the prevalence of malnutrition in patients with chronic liver disease. Of the studies surveyed, only those using valid assessment techniques and appropriate comparisons, have been included. Specifically, assessment of vitamin status, biochemical variables and immune response have been excluded because of the limited contribution these make to the overall assessment of nutritional status in patients with chronic liver disease'^^’'^^^.

2.3 Nutrient intake The adequacy of nutrient intake in patients with chronic liver disease has been compared with data from healthy volunteers in a number of However, few provide data on the number of patients with an inadequate intake and, where this has been reported, comparisons have been made with recommended intakes for healthy populations which is probably inappropriate^^^’^^^’^^^.

2.4 Anthropometry The measurement of height, weight, MAC and skinfold thicknesses and the calculation of derived variables, BMI and MAMC have been undertaken in patients with chronic liver disease in many studies (Tables 2.1-2.3). However the reference values against which comparisons have been made vary, making inter-study comparisons difficult. Where body weight or BMI have been investigated, between 2 to 30% of patients have been identified as below the reference value, while the respective proportion of patients with depleted measurements for TSF and MAMC were 7 to 55% and 9 to 50% respectively. Chapter 2 Prevalence of malnutrition 55

2.5 Muscle function In the studies^^^’^^^’^^®’^^^ where muscle function has been investigated in patients with chronic liver disease, the prevalence of impaired results was not reported although evidence of impairment was found^^^’^^°.

2.6 Clinical nutritional status Nutritional status in patients with chronic liver disease has been evaluated by clinical means alone in eight (Table 2.4) but in only one of these^^^ has the method used been described in detail and its reproducibility examined^^^. Comparisons are difficult because the categorization varied between studies and was incompletely described in the majority. However, with these limitations, between 21 and 88% of patients assessed were categorized as adequately nourished, and thus between 12 and 79% must be considered to be malnourished (Table 2.4). Overall, between 19 and 59% of patients were categorized as mildly malnourished, 21 to 43% as moderately malnourished and 10 to 26% as severely malnourished.

2.7 Prevalence of malnutrition and severity of liver disease No differences were observed between the nutritional status of patients with pre­ cirrhotic liver disease and those with cirrhosis in one study^^^; in another^^^, patients with pre-cirrhotic primary biliary cirrhosis (stages 1-111) had a significantly lower prevalence of depleted anthropometric variables than patients whose disease had progressed to cirrhosis (stage IV).

Results from studies of patients with cirrhosis show that worsening liver function is associated with an increasing prevalence and / or greater degree of malnutrition assessed anthropometrically and clinically^'^^^'^^^'^^^'^^'^^^ (Table 2.5).

2.8 Prevalence of malnutrition and aetiology of liver disease The results of comparisons of nutritional variables between patients with liver disease of differing aetiologies are conflicting (Table 2.6). While differences have been observed in some studies^^’^^’^^^’^^^’^^^, generally with greatest depletion found Chapter 2 Prevalence of malnutrition 56

in patients with primary biliary cirrhosis, this is not universaP’"^^^. However, in only one of these studies^^^, was account taken of the confounding effects of the severity of the liver disease. No significant differences were found on stepwise regression analysis between nutritional abnormalities in different aetiological groups in male patients but, among women, greatest depletion of fat stores were observed in those with alcoholic liver disease while those with hepatitis B chronic liver disease showed greatest depletion of muscle stores (Table 2.6).

2.9 Summary The prevalence of malnutrition in patients with chronic liver disease varies greatly and is dependent on the population studied, the methods of assessment used and the references against which comparisons of variables are made. Studies where nutritional status has been assessed by methods that are considered reasonably useful in this patient population, such as anthropometric and clinical variables, reveal prevalence rates of between 10 and 80%. Although the prevalence of malnutrition increases with increasing severity of liver disease^^’^^^, there are less data on the relationship between the prevalence of malnutrition and the aetiology of the liver disease.

2.10 STUDY 2: A SURVEY OF THE PREVALENCE OF MALNUTRITION IN PATIENTS WITH CIRRHOSIS

2.10.1 Aim To survey the nutritional status of a large group of unselected patient with cirrhosis using a reproducible assessment tool.

2.10.2 Patients Between January 1995 and July 1997, 215 patients with cirrhosis were nutritionally assessed. The population comprised 129 men and 86 women of mean (range) age 51.0 (26-76) years with liver disease of varying aetiology and severity. The aetiology of the liver injury was diagnosed on the basis of historical, clinical. Chapter 2 Prevalence of malnutrition 57 laboratory, radiological and histological variables whilst the functional severity of the liver disease was assessed using Pugh’s modification of Child’s grading scheme"^^ (Table 2.7). One hundred and sixty four patients were assessed as inpatients and 51 as out-patients.

2.10.3 Assessment of nutritional status All patients were nutritionally assessed by one observer using a global assessment scheme incorporating objective and subjective data. The following were collected:

2.10.3.1 Clinical information was obtained directly from the patients and from their medical records. Gastrointestinal symptoms which might influence nutrient intake, such as nausea, vomiting, indigestion or food-related abdominal pain were recorded and graded, in relation to their degree, frequency and duration, as absent (0), mild (1), moderate (2) or severe (3). A symptom score was then calculated for each patient by adding the individual scores. Bowel habits, and any recent changes in stool frequency, colour or consistency were recorded. A history and details of any infection, renal dysfunction, hepatic encephalopathy and gastrointestinal bleeding were sought. Pre-illness weight and weight range in the preceding 6 months were recorded and the overall weight change in this period calculated. These data were checked, where possible, by reference to the clinical notes. Finally, patients were questioned about their usual physical activity, endurance and fatigue.

2.10.3.2 Nutrient intake: The patients’ appetite was assessed as good, fair or poor and any factors likely to interfere with food intake, for example, anorexia, difficulty in mastication, taste changes, dysphagia or early satiety or taste changes, noted. Recent dietary intake was assessed using an established diet history method^ supplemented, where necessary, with additional information from relatives, nursing staff and food record sheets. The data obtained were not intended to provide a quantitative evaluation of intake but to give an idea of overall adequacy of the diet in relation to estimated requirements. These were assessed using Schofield’s modification'^^^ of the Harris-Benedict equations^®^ using Chapter 2 Prevalence of malnutrition 58 the patients’ clinical condition as the basis of any adjustments felt necessary in the absence of specific correction factors for cirrhosis^^^. Energy intake was categorized as adequate, if perceived to meet estimated requirements, as inadequate, if it failed to meet estimated requirements but exceeded 500 kcal/day, or negligible if estimated to provide 500 kcal daily or less. Details of dietary restrictions and oral, enteral or parenteral nutritional support were also recorded.

2.10.33 Physical status: A subjective evaluation of subcutaneous fat stores and the presence of muscle wasting and fluid retention was made by visual inspection.

2.10.3.4 Anthropometric measurements: Height was measured to the nearest centimetre with patients standing in socks or bare feet, using a wall mounted or free-standing measure (Seca, Hamburg, Germany). Patients were weighed to the nearest 0.1 kg, wearing night-wear or light indoor clothing and without shoes, using a weighing scale (Seca, Hamburg, Germany) or a seat balance scale (Marsden W/M Group, London, UK).

An estimated dry weight was determined by deducting a weight for ascites and/or oedema based on clinical assessment and previously documented weights using published guidelines'^. BMI was calculated from the estimated dry weight.

MAC and skinfold thicknesses at the triceps, biceps, subscapular and suprailiac sites were measured using standardized techniques^"^^’^"^^ and MAMC calculated^^^. MAMC and TSF were compared with published standards^^, and expressed as a percentage of the 50th percentile for the appropriate sex and age category. Percentage body fat was also calculated from the logarithmic sum of the skinfold thicknesses at the four sites^'^^.

2.10.3.5 Muscle function: Muscle function was evaluated using a hand held dynamometer to measure grip-strength (TKK 5101, Takei Scientific Instruments Co Ltd, Tokyo, Japan) (Figure 2.1). Patients were instructed to hold the dynamometer Chapter 2 Prevalence of malnutrition 59

in their non-dominant hand and compress its two bars together as hard as possible without stressing their abdominal or jaw muscles. Patients undertook the exercise with either an extended or bent elbow and were allowed to stand or sit down depending on their preference. Three measurements were made; the patient released the dynamometer and flexed their fingers between repeat measurements; the highest of the three readings was used. Comparisons were made against reference values^^^.

2.10.3.6 Global assessment: Nutritional status was categorized, using the algorithm in Appendix, as: (1) well-nourished; (2) moderately malnourished (or suspected of being so); (3) severely malnourished.

2.10.4 Statistical analysis Statistical analysis was performed using the Minitab software package (1994, State College, PA, USA). The demographic and anthropometric data were analysed using descriptive statistics. Comparisons of demographic and anthropometric variables and symptom scores were made between patient subgroups using unpaired Student’s t-tests. Chi-squared tests and oneway analysis of variance; grip- strength data were not normally distributed and were compared using Mann- Whitney U and Kruskal-Wallis tests. Two way analysis of variance was carried out on anthropometric variables, to test whether global nutritional status was related to these independently of the severity of liver disease. These analyses were undertaken on logarithmic values for grip-strength. Data are expressed as mean ± ISD and median (range) values.

2.10.5 Results 2.10.5.1 Clinical data: Of the 215 patients studied, 134 (62%) had alcoholic cirrhosis, 41 (19%) post-viral cirrhosis and 21 (10%) biliary cirrhosis. Overall, 65 (30%) were Child’s A, 63 (29%) Child’s B and 87 (41%) Child’s C ^. However, the proportion of individuals in each aetiological group with decompensated liver Chapter 2 Prevalence of malnutrition 60

disease varied (Table 2.7).

Ninety nine (46%) patients reported no gastrointestinal symptoms likely to influence nutritional intake. Indigestion was the most commonly reported symptom, experienced by 60 (28%) patients (Table 2.8). The majority of patients described the severity of their symptoms as mild with only 1-3% of individuals reporting severe manifestations. The mean symptom score for the total patient population was 1.3 ± 1.8.

Bowel frequency varied between motions passed every third day to eight times per day with a mean daily frequency of 2.1 in the total patient population. Ninety one (42%) patients described their stools as "normal", compared to 57 (26%) "semi­ solid", 47 (22%) "soft" and ten (5%) each as "liquid" or "hard".

No evidence of hepatic encephalopathy was found in 125 (58%) patients, while 89 (41%) showed signs of grade I II encephalopathy and the remaining two patients were categorized as grade III. Data on infections, renal dysfunction and gastrointestinal bleeding were used in the global assessment of nutritional status but were not documented in a structured form that allowed the prevalence in the surveyed population to be quantified. The weight changes in the 6-month period preceding the assessment varied from a gain of 4 kg/month to a loss of 7 kg/month, with a mean overall loss of 0.8 kg/month.

2.10.5.2 Nutrient intake: Energy intake was considered to be adequate in 126 (59%) patients, inadequate in 77 (36%) and negligible in 12 (5%). One hundred and twenty (56%) patients were following restricted diets, the most common being low sodium (91 [42%]) and diabetic diets (31 [14%]) with 12 (6%) patients following both restrictions. There was no significant difference in the adequacy of the i energy intake between patients who were following restricted diets and those who were not (restricted diets: adequate intake 65 [54%], inadequate 49 [41%], negligible 6 [5%]; unrestricted diets: 61 [64%], 28 [30%], 6 [6%]; ns). Chapter 2 Prevalence of malnutrition 61

At the time of assessment, 63 (29%) patients were taking oral nutritional supplements, for example, fortified milkshakes, soups and/or juices, while a single patient received nasogastric feeding; none were being fed parenterally. Patients taking oral supplements were more likely to have an inadequate intake, including their supplements, than those who were unsupplemented (supplemented: adequate intake 21 [33%], inadequate 37 [59%], negligible 5 [8%]; unsupplemented: 105 [69%], 40 [26%], 7 [5%]; p< 0.0001).

2.10.5.3 Physical status: Ascites was observed in a total of 109 (51%) patients, and was classified as mild in 51 (24%), moderate in 28 (13%) and severe in 30 (14%) individuals. Although muscle wasting and loss of subcutaneous fat were subjectively assessed by visual inspection and used in the global assessment of nutritional status, it was not documented in a structured form that allowed the prevalence in the surveyed population to be quantified.

2.10.5.4 Anthropometric measurements: The range of anthropometric measurements observed in the population studied was large, with, as expected, differences between variables in the men and women (Table 2.9). There was evidence of malnutrition in a considerable number of patients with 47 (22%) patients having a BMI below 20 kg.m"^ and 92 (43%) and 37 (17%) patients respectively having a MAMC or TSF of less than the 5th percentile for the appropriate reference group. One hundred and ten (51%) patients had at least one of these three measurements in the depleted range while 32 (15%) and 17 (8%) had two and three depleted measurements respectively. Significantly more women than men had TSF measurements below the 5th percentile (24 [28%] vs 13 [10%], p< 0.001) while significantly more men than women had depleted MAMC measurements (69 [53%] vs 23 [27%], p< 0.0001). Overall, the prevalence of depleted MAMC was greater than TSF.

2.10.5.5 Muscle function: The median grip-strength for the whole population was 23.1 (5.2-54.6) kg force or 63.1 (17.3-123.7) % of published reference values^^^. Only six (5%) men and three (3%) women had grip-strengths equal to or greater Chapter 2 Prevalence of malnutrition 62

than the lower limits of the reference values.

2.10.5.6 Global categorization: Ninety two patients (43%) were categorized as well nourished, 85 (39%) as moderately malnourished, or suspected of being so, and 38 (18%) as severely malnourished. The subjective override was used in 16 patients, with nutritional status categorized as more malnourished in eight patients and less malnourished in eight. Of the 123 patients who were globally assessed as malnourished, 99 (80%) had at least one depleted anthropometric variable while the remaining 24 patients had adequate measurements of BMI, TSF and MAMC.

There were no significant differences in age or sex between the patients in the three categories of nutritional status. Well nourished patients had significantly lower symptom scores and were significantly less likely to report nausea than those who were malnourished (symptoms score: well nourished 0.8 ± 1.4; moderately malnourished 1.6 ± 1.7; severely malnourished 2.1 ± 2.4, p< 0.0001; nausea: 13%; 26%; 39%, p<0.05). Well nourished patients were significantly more likely to have an adequate energy intake and to follow an umestricted, unsupplemented diet than malnourished patients (adequate intake: well nourished 95% moderately malnourished 40%; severely malnourished 13%, p< 0.0001 umestricted diet: 55%; 34%; 39%, p<0.05; no oral supplements: 91%; 59% 47%, p < 0.0001)(Figures 2.2 - 2.4). Measurements of BMI, TSF and MAMC were significantly more depleted and relative grip-strength significantly lower in patients with impaired nutritional status than those who were well nourished (Table 2.10).

2.10.5.7 Seventy of liver disease: Patients in the three Child’s categories of functional severity"^^ were matched for age and sex (Table 2.11). Mean symptom scores and reports of abdominal pain increased significantly with deteriorating liver function (symptom score: Child’s Grade A 0.9 ± 1.5; Grade B 1.2 ± 1.6; Grade C 1.7 ± 2.0, p < 0.05; abdominal pain: 15%; 19%; 41%, p< 0.005). There was also a significant association between the severity of liver disease and the adequacy of 1 energy intake (Figure 2.5), with fewer patients in Grades A and B following restricted diets or taking supplements than those in Grade C (restricted diets: Chapter 2 Prevalence of malnutrition 63

28%; 60%; 74%, p< 0.0001; supplements; 11%; ;44%; 64%, p< 0.0001).

Increasing severity of liver disease was also significantly associated with diminishing anthropometric and grip-strength measurements and with a deterioration in global nutritional status (p<0.0001); 72% of patients with Child’s Grade C were identified as malnourished (Table 2.11).

2.10.5.8 Aetiology of liver disease: Patients in the aetiological subgroups were not matched for age, sex, or severity of liver disease (Table 2.7). The patients with biliary cirrhosis comprised significantly more women and older patients than other groups.

Men with biliary cirrhosis reported significantly more nausea and indigestion than men in the other aetiological groups (nausea score: biliary 1.6 ± 1.1; other 0.2 ± 0.5, p< 0.0001; indigestion: 0.8 ± 1.1; 0.3 ± 0.6, p< 0.0001). However, there were no significant differences between their energy intake, dietary restrictions followed or supplements taken; anthropometric variables, grip-strength and global nutritional status was not statistically different between the groups.

In the women, there were no significant differences in the number of symptoms reported, energy intake or dietary restrictions between the three main aetiological groups although women with alcohol-related cirrhosis were more likely to take oral supplements than women in the other groups (38% vs 14%, p < 0.05). Women with biliary cirrhosis had a number of significantly more depleted anthropometric variables than other women (Table 2.12). However, no significant differences were found in grip-strength or global nutritional status between the groups.

2.10.5.9 Inter-dependence of severity of liver disease and nutritional status: Twoway analysis of variance showed that global nutritional status continued to be related, independently of the severity of liver disease, to weight (p< 0.0001), dry weight (p< 0.0001), BMI (p< 0.0001), MAC (p< 0.0001), MAMC (p< 0.0001), TSF (p < 0.0001), relative MAMC (p< 0.0001), relative TSF (p< 0.0001), grip-strength Chapter 2 Prevalence of malnutrition 64

(p< 0.0001) and relative grip-strength (p<0.0001)(Table 2.13). However, it also showed that the effect of severity of liver disease, independently of global nutritional status, was less marked on weight (p< 0.0001), dry weight (ns), BMI (ns), MAC (ns), MAMC (ns), TSF (ns), relative MAMC (ns), relative TSF (ns), grip- strength (ns) and relative grip-strength (p < 0.05).

2.10.6 Discussion In this survey, more than half of the patients with cirrhosis were categorized as malnourished on global assessment. The prevalence of depleted BMI, TSF and MAMC measurements was similar. However, 20% of the patients considered globally malnourished had anthropometric measurements in the acceptable range. If these variables alone were used to identify individuals at nutritional risk, some patients would not be recognised thus emphasizing the importance of undertaking a composite nutritional assessment.

In the present study, the significant association observed between severity of liver disease and impaired nutritional status confirms previous findings^'^^^'^^^'^^^'^^'^^^. It could be surmised that this relationship arises because sicker patients experience worse symptoms and, as a result of an inadequate intake, become more malnourished. This is supported in the present study, by significant associations found between the severity of liver disease and both the symptom score and energy intake and between nutritional status and both the symptom score and energy intake. At present, however, a causal relationship remains conjecture and a more sophisticated, probably prospective, study of larger numbers of patients is required to address this hypothesis.

In the present study, no significant relationships were found between the aetiology of liver disease and global nutritional status when data from men and women were analysed together, also confirming previous findings^^’"^^^. However, when data from male and female patients were analysed separately, women with biliary cirrhosis were significantly more depleted than women with alcohol-related or viral disease. This concurs with results from the Italian Multicentre Co-operative Project^^^ where Chapter 2 Prevalence of malnutrition 65 aetiology was found to have a significant effect on the nutritional status of female but not male patients; the exact nature of the effect was not reported. The association observed between aetiology and malnutrition in the present study is weaker than that observed between severity of liver disease and malnutrition. However, it could be similarly speculated that an association between the aetiology of liver disease and nutritional status relates to the symptoms more commonly found in particular diseases than others, for example, impaired nutrient intake in alcoholic liver disease^^’^^^ and steatorrhoea in biliary cirrhosis^°°. However, this cannot be confirmed by the results from the present study, although patients with biliary cirrhosis, who were more depleted, also had significantly higher symptom scores.

One of the principal aims of identifying individuals who are at nutritional risk is to determine where nutritional support is required; consideration must therefore be given to the relationship between nutritional status and the use of nutritional supplements. In the present study, approximately one third of patients were taking oral nutritional supplements while only one received nasogastric feeding. The total number of patients receiving nutritional support is less than the 42% who were considered to have an inadequate or negligible energy intake and considerably less than the 57% of individuals categorized as malnourished. It appears that the provision of nutritional support is inadequate to address the apparent deficit but this criticism must be tempered. Many of the assessments undertaken in hospital inpatients were carried out shortly after admission and, on some occasions, before the patients had been clinically assessed and dietary treatment initiated.

In the present study, comparisons were made between patients’ grip-strength measurements and reference values published by Webb and colleagues^^^. These values, which are described by the authors as the lower acceptable limits below which the risks of post-operative complications increase, are higher than those published by Klidjian and colleagues^^^ but were selected for use because of a reported greater sensitivity and specificity for identifying patients at risk^^^. However, as only nine (4%) patients had grip-strength measurements equal to or Chapter 2 Prevalence of malnutrition 66 greater than the reference values these could not be used as cut off values to divide the patient population. Therefore, grip-strength measurements were expressed as a percentage of the age- and sex-matched reference value.

Comparisons between the relative grip-strength and both global nutritional status and functional severity of liver disease showed equally significant associations. Both relationships were independent of the other, suggesting that measurement of grip-strength is useful in identifying impaired nutritional status in this patient population. This has not been previously reported^^^’^^®’^^^. The relationships between other objective variables, including BMI and MAMC, and both global nutritional status and severity of liver disease, were, however, only independently related to global nutritional status and not to severity of liver disease, suggesting that these are more useful indicators of nutritional status than grip-strength.

Study conclusions: The results from this study confirm a high prevalence of malnutrition in a large group of unselected patients with cirrhosis. The significant associations found between severity of liver disease, reported gastrointestinal symptoms, energy intake and nutritional status may represent a causative link. The prevalence and degree of nutritional impairment were highest in patients with the worst liver function and in women with biliary cirrhosis. Table 2.1

Body weight in patients with chronic liver disease.

First author Population Cirrhosis Fluid Evidence of Prevalence of

Year (%) retention depletion depletion (%)

Neville^^° 1968 34 alcoholics 9 minimal in 1% "underweight" 9

Bollef" 1973 27 alcoholic liver disease NA NA < reference^"^ 24

0 ’Keefe"°' 1980 66 chronic liver disease 100 NA <80% reference^''" 21

Mills'"' 1983 32 alcoholic liver disease 38 NA <80% reference^"*" 3

Merli'"" 1987 70 chronic liver disease 100 nil <90% reference^"*" 18

Lolli'°" 1992 200 chronic liver disease 100 nil <90% reference'"" 18

Wood""" 1992 39 alcoholics 41 NA <90% reference'*" 22

Italian MCP"'" 1996 1402 chronic liver disease 100 nil <90% reference^"' 5

Thuluvath"^’ 1994 132 chronic liver disease NA minimal in 26% <80% reference"*^ 2

Caregaro** 1996 120 chronic liver disease 100 65% <90% reference'^* 30

NA - not available.

On < 1 Table 2.2

Triceps skinfold measurements in patients with chronic liver disease.

First author Population Cirrhosis Evidence of depletion Prevalence of Year (%) depletion (%)

O’Keefe'®' 1980 66 chronic liver disease 100 < 60% reference^'® 21

Mills^®' 1983 32 alcoholic liver disease 38 < 60% reference^'® 9

Jhangiani^^® 1986 8 alcoholic liver disease 100 < 5“^ percentile'^® 38

Lautz^®^ 1992 123 chronic liver disease 100 < 50% reference® 18

LollT®" 1992 200 chronic liver disease 100 < 5th percentile'^^ 7

Wood^®^ 1992 39 alcoholics 41 < 90% reference^'® 55

Akerman'® 1993 104 transplant candidates NA < 5"^ percentile^'® 35

Italian 1994 1402 chronic liver disease 100 < 5"^ percentile MAFA'^^ 16

Thuluvath^^^ 1994 132 chronic liver disease NA < 5"^ percentile'’^ 12

Wicks"'" 1995 80 primary biliary cirrhosis 43 < 5"^ percentile"® 47

Caregaro®* 1996 120 chronic liver disease 100 < 5"^ percentile ± MAMC'” 30

NA - not available; MAFA - mid-arm fat area; “Men - 12.5mm; women - 16.5mm.

ooON Table 2.3

Mid-arm muscle circumference in patients with chronic liver disease.

First author Population Cirrhosis Evidence of depletion Prevalence of

Year (%) depletion (%)

Mills'^^ 1983 32 alcoholic liver disease 38 < 80% reference^'*" 13

Jhangiani^^° 1986 8 alcoholic liver disease 100 < 5^^ percentile 50

Merli^^^ 1987 70 chronic liver disease 100 < cut off values^'" 21

LollT°" 1992 200 chronic liver disease 100 < 5*^ reference^^^ 33

Wood'"' 1992 39 alcoholics 41 < 90% reference^'*" 22

Akerman^° 1993 104 transplant candidates NA < 5**^ percentile^"" 45

Italian MCP^^' 1994 1402 chronic liver disease 100 < 5^ percentile MAMA'^^ 23

Thuluvath'^^ 1994 132 chronic liver disease NA < 5^ percentile^^^ 32

Wicks"" 1995 80 primary biliary cirrhosis 43 < 5^ percentile^" 9

Caregaro®® 1996 120 chronic liver disease 100 < 5*^ percentile ± 34

NA - not available; MAMA - mid-arm muscle area.

'O

Nutritional status of patients with chronic liver disease assessed using clinical methods.

First author Population Cirrhosis Global categories of nutritional status Year (%) and prevalence (%)

Adequate Mild Moderate Severe nutrition® malnutrition^ malnutrition malnutrition

Patek'''® 1948 124 chronic liver disease 100 27 73

Morgan^ 1981 55 alcoholic liver disease 33 71 <- 29 20 primary biliary cirrhosis 70 60 <- 40 17 chronic active hepatitis 53 12

Hasse^ 1992 570 transplant candidates NA 26 < r- 61 13

Nielsen^®^ 1993 37 alcoholic liver disease 100 22 35 43

Italian MCP^"" 1994 1402 chronic liver disease 100 69 21 10

PikuP 1994 68 transplant candidates NA 21 19 34 26

Loguerico^ 1996 184 chronic liver disease 100 86 14

RiccP 1997 436 transplant candidates NA 28 59 - -13'- - -

NA - not available; ®also defined as well nourished, normally nourished or excellent nutritional status. Includes overweight patients who had no signs of nutritional depletion; '’also defined as slightly malnourished or fair nutritional status; 'defined as poor nutritional status. <1 o Table 2.5

Nutritional status and the severity of chronic liver disease.

First author Population Cirrhosis Categorization Nutritional variable Mean values, by severity of liver disease Year (%) of liver disease

Mendenhalf* 1984 361 ALD 59 mild, moderate Dry weight Mild: 67.7kg moderate: 63.3kg severe: 61.1kg* and severe^"*® TSF Mild: 8.9mm moderate: 8.5mm severe: 7.5mm** MAMC Mild: 24.6cm moderate: 23.4cm* severe: 23.1cm**

Abad^ 1987 125 CLD 100 Child’s grade"”® Adequate nutrition Child A 22%” Child B: 66% Child C: 12% Moderate malnutrition Child A 0% Child B: 27% Child C: 73% Severe malnutrition Child A 0% Child B: 0% Child C: 100%

Merli^” 1987 70 CLD 100 Liver damage""® Patients with ^ -I Mild: 24% moderate: 31% severe: 37% anthropometric variables

Crawford^^^ 1994 57 CLD 100 Child’s grade” ® Body fat Child A: 97% Child B: 71% Child C: 65%” (% of predicted)

Italian MCP""" 1994 1402 CLD 100 Child’s grade” ® Adequate nutrition Child A 81% Child B: 61% Child C: 42% Moderate malnutrition Child A 15% Child B: 26% Child C: 32% Severe malnutrition Child A 4% Child B: 13% Child C: 26%

Loguerico^°® 1996 184 CLD 100 Liver damage” ^ Well-nourished Child A 57% Child B & C: 59% Malnourished Child A 8% Child B & C: 26% Obese Child A 35% Child B & C: 15%

*p<0.01 cf mild; **p<0.05 cf mild; ^p<0.001 Chi-squared test; ^p<0.05 cf Child B. Table 2.6

Nutritional status and the aetiology of chronic liver disease.

First author Population Cirrhosis Prevalence of depletion (%) Year (%) ALDPBCCAHCVD CC Evidence of depletion ALDPBC CAHCVD CC

Morgan^*’ 1981 55 20 17 -- 33:70:53:-:- Clinically malnourished 29 40 12 - -

Italian MCP“ ’ 1994 445 - - 357 396 100 MAFA <5th-ile‘” M i l ; F 42 - M 9; F 35 M 9; F 15 MAMA <5th-ile'” M 38; F 5 M 32; F 10 M 29; F 3

Pikul"^ 1994 13 17 20 NA No differences observed

Thuluvath^^^ 1994 51 45 28 NA Weight <80% reference"*^ 2 4 0 TSF <5th-ile reference^” 10 16 25 - - MAMC <5th-ile reference*^ 35 40 38

Caregaro** 1996 77 43 --- 100 No differences observed

Reference variable Comparison with reference values (%)

Dicecco^^ 1989 - 20 24 -- NA Ideal weight*^ 105 115

TSF^^ - 45 71 -- MAMC'^ - 56 76 --

Akerman^” 1993 . 21 34 -. NA Ideal weight^® 102 118

Ts f i7« - 68 75 -- M AM C’® - 86 91 --

ALD - alcoholic liver disease; PBC - primary biliary cirrhosis; CAH - chronic active hepatitis; CVD - chronic viral disease; CC - cryptogenic cirrhosis; NA - not available; MAFA - mid-arm fat area; MAMA - mid-arm muscle area.

o to 73

Table 2.7

Aetiology and functional severity of liver disease in 215 patients with cirrhosis undergoing assessment o f nutritional status.

Data expressed as number (%).

Aetiology of cirrhosis Subjects Child’s grade""^

(n) A B C

Alcohol 134 (62) 44 (33) 31 (23) 59 (44)

Post-viral 41 (19) 14 (34) 11 (27) 16 (39)

Biliary (PBC & PSC) 21 (10) 5(24) 8 (38) 8 (38)

Autoimmune CAH 7(3 ) 0 (0 ) 5 (71) 2 (29)

Cryptogenic 6(3) 0 (0 ) 5 (83) 1 (17)

Other 6 (3 ) 2 (33) 3 (50) 1 (17)

Total 215 (100) 65 (30) 63 (29) 87 (41)

PBC - primary biliary cirrhosis; PSC - primary sclerosing cholangitis; CAH - chronic active hepatitis. 74

Table 2.8

Frequency and severity of gastrointestinal symptoms likely to influence nutrient intake in

215 patients with cirrhosis.

Symptoms Patients Patients reporting symptom

symptom-free n(% )

n(%) Mild Moderate Severe

Nausea 166 (77) 25 (12) 18 (8) 6 (3 )

Vomiting 188 (87) 15 (7) 6(3) 6 (3)

Indigestion 155 (72) 44 (20) 14(7) 2(%

Food-related abdominal pain 157 (73) 39 (18) 14 (7) 5(% 75

Table 2.9

Anthropometric measurements in 215 patients with cirrhosis using appropriate standards^^.

Data expressed as mean ± ISD, except where shown.

Variable All patients Men Women (n=215) (n=129) (n=86)

Height 1.67 1.72 1.60" (m) ± 9.6 ± 7.6 ±7.5

Weight 71.8 77.2 63.6" (kg) ± 16.2 ± 16.3 ± 12.0

Dry weight 67.2 72.2 59.6" (kg) ± 15.7 ± 15.8 ± 12.0

BMI 23.9 24.3 2T3 (kg.m'^) ± 4.7 ± 5.0 ± 4.3

TSF 14.1 12.8 15.9* (mm) ± 6.9 ± 7.0 ± 6.2

MAMC 22.2 23.4 20.5" (cm) ± 3.2 ± 3.1 2.7

Relative TSF 94.0 113.1 65.7" (%) ± 56.4 ± 62.8 ± 26.2

Relative MAMC 87.0 83.4 92.3" (%) ± 12.4 ± 11.1 ± 12.3

TSF <5‘^%ile 37 13 24* n(%) (17%) (10%) (30%)

MAMC <5“'%ile 92 69 23* n(%) (43%) (53%) (27%)

Body fat 21.6 18.1 26.9" (%) ±7.8 ± 6.4 ± 6.8

Significance of difference between men and women; *p<0.005; **p<0.0001. 76

Table 2.10

Relationship between global nutritional status and anthropometric variables in 215 patients with cirrhosis. Data expressed as mean ± ISD, except for grip-strength, median (range).

Variable Well Moderately Severely Statistical nourished malnourished malnourished difference® (n=92) (n=85) (n=38)

Age 52.5 49.0 51.7 ns (yr) ± 10.7 ± 9.7 ± 12.6 Male: female 56:36 51:34 22:16 ns (%) (61:39) (60:40) (58:42)

Height 1.68 1.67 1.67 ns (m) ±0.09 ±0.10 ± 0.10

Weight 79.3 69.8 58.2 p<0.0001 (kg) ± 15.2 ± 14.3 ± 11.9

Dry weight 77.0 64.0 50.7 p<0.0001 (kg) ± 14.2 ± 12.0 ± 8.0

Estimated fluid 2.2 5.8 7.7 p<0.0001 retained (kg) ±4.2 ± 6.6 ± 7.4

Weight change -0.2 -1.1 -1.8 p<0.0001 (kg) ± 1.2 ± 1.5 ± 1.7 BMI 27.2 22.9 18.1 p<0.0001 (kg.m^) ± 3.7 ± 3.4 ± 2.1

MAC 29.8 25.7 20.8 p<0.0001 (cm) ± 3.6 ± 3.1 ± 2.3

TSF 17.1 13.4 8.0 p<0.0001 (mm) ± 6.9 ± 5.7 ± 4.6

MAMC 24.4 21.5 18.4 p<0.0001 (cm) ± 2.6 ± 2.4 ± 1.8

Relative TSF 116.2 90.2 49.7 p<0.0001 (%) ± 61.0 ± 48.2 ± 26.1

Relative MAMC 95.7 84.2 71.8 p<0.0001 (%) ± 9.6 ± 8.4 ± 7 .7

Body fat 25.1 21.0 14.4 p<0.0001 (%) ± 7.2 ± 6.5 ± 6.8

Grip-strength 26.5 21.9 17.9 p<0.0001 (kg.f) (9.5-54.6) (5.2-44.0) (8.3-43.1)

Relative grip- 71.8 56.9 51.0 p<0.0001 strength (%) (25.1-123.7) (17.3-108.9) (30.7-95.8)

“Oneway ANOVA, except for grip-strength where Kmskal-Wallis tests used. 77

Table 2.11

Relationship between severity of liver disease and nutritional variables in 215 patients with cirrhosis. Data expressed as mean ± ISD, except for grip-strength, median (range).

Variable Child’s Child’s Child’s Statistical grade A grade B grade C difference® (n=65) (n=63) (n=87)

Age 51.3 52.8 49.4 ns (yr) ± 11.0 ± 11.8 ± 9.5

Male: female 39:26 37:26 53:34 ns (%) (60:40) (59:41) (61:39)

Height 1.66 1.68 1.68 ns (m) ±0.10 ±0.10 ± 0.09

Weight 71.3 70.0 73.5 ns (kg) ± 16.2 ± 15.4 ± 16.6

Dry weight 71.2 66.5 64.7 p<0.05 (kg) ± 16.1 ± 15.1 ± 15.4

Estimated retained fluid 0.1 3.5 8.7 p<0.0001 (kg) ±0.6 ±4.4 ±7.0

Weight change 0 -1.0 -1.3 p<0.0001 (kg) ± 1.1 ± 1.6 ± 1.6

BMI 25.7 23.5 22.8 p<0.005 (kg.m'^) ±4.6 ±4.7 ±4.5

MAC 28X5 26.4 25.3 p<0.0001 (cm) ± 4.4 ±4.3 ± 4.4

TSF 16.1 14.3 12.3 p<0.005 (mm) ±7.3 ±7.1 ± 5.9

MAMC 23.5 21.9 21.5 p<0.0001 (cm) ±3.2 ±2.9 ± 3.3

Relative TSF 108.8 94.0 83.5 p<0.05 (%) ± 61.9 ± 57.6 ±48.9

Relative MAMC 91.5 86.4 83.9 p<0.005 (%) ± 11.8 ± 11.6 ± 12.5

Body fat 23.9 21.0 20.4 p<0.05 (%) ± 8.3 ± 7.8 ± 7.2

Grip-strength 26.0 21.8 20.3 p<0.05 (kg.f) (8.2-54.6) (10.8-39.0) (5.2-47.6)

Relative grip-strength 72.0 633 55.4 pO.OOOl (%) (25.1-121.3) (32.6-108.9) (17.3-123.7)

Global nutrition, (n): adequate nutrition 44 24 24 ) moderate malnutrition 14 32 39 ) pO.OOOl severe malnutrition 7 7 24 )

Oneway ANOVA, except for grip-strength where Kruskal-Wallis tests used and sex and global nutrition where Chi-squared test used. 78

Table 2.12

Relationship between aetiology and nutritional variables in 76 women with cirrhosis.

Data expressed as mean ± ISD, except for grip-strength, median (range).

Variable Alcoholic Viral Biliary Statistical cirrhosis cirrhosis cirrhosis difference® (n=48) (n=12) (n=16)

Age 52.2 49.6 60.3 p<0.05 (yr) ± 10.4 ± 11.0 ± 11.2

Height 1.60 1.58 1.58 ns (m) ±0.07 ±0.11 ±0.07

Weight 64.6 63.6 57.9 ns (kg) ± 12.5 ± 9.1 ± 10.0

Dry weight 59.3 61.1 55.9 ns (kg) ± 12.8 ± 8.0 ± 10.3

Estimated fluid retained 5.3 2.6 2.0 ns (kg) ± 7.4 ±3.9 ± 2.9

Weight change -0.7 -0.4 -1.1 ns (kg) ± 1.7 ±0.5 ± 1.5

BMI 23.1 24.6 22.2 ns (kg.m-^) ±4.5 ± 3.3 ± 3 .2

MAC 25.3 27.9 23.5 p<0.05 (cm) ±4.6 ±2.5 ± 3.2

TSF 15.4 19.5 14.2 ns (mm) ± 6.7 ±4.2 ± 5.7

MAMC 20.5 21.7 19.0 p<0.05 (cm) ±2.8 ± 1.9 ± 1.9

Relative TSF 63.3 81.0 59.3 ns (%) ±28.1 ± 18.0 ±23.4

Relative MAMC 92.6 97.8 84.9 p<0.05 (%) ± 13.1 ± 8.8 ± 8.4

Body fat 27.0 31.3 23.0 p<0.005 (%) ± 6.8 ±4.3 ± 6.0

Grip-strength 15.1 17.4 15.7 ns (kg.f) (5.2-33.9) (13.4-26.0) (8.2-29.4)

Relative grip-strength 54.4 5&8 56.9 ns (%) (17.3-113.0) (48.6-92.9) (38.0-108.9)

Global nutrition, (%): adequate nutrition 44 67 25 ) moderate malnutrition 33 33 50 ) ns severe malnutrition 23 0 50 )

“Oneway ANOVA, except for grip-strength where Kruskal-Wallis tests used and global nutrition where Chi- squared test used. 79

Table 2.13

Relationship between relative grip-strength (%) and both global nutritional status and severity of liver disease"*"^. Data expressed as mean ± ISD (number of patients).

Well Moderately Severely

nourished malnourished malnourished

(n=92) (n=85) (n=38)

Child’s A 74.5 ± 20.6 63.4 ± 18.2 66.5 ± 21.2

(n=65) (44) (14) (7)

Child’s B 66.7 ± 16.3 62.7 ± 15.2 49.0 ± 9.7

(n=63) (24) (32) (7)

Child’s C 69.8 ± 22.0 57.3 ± 16.1 49.9 ± 14.9

(n=87) (24) (39) (24)

Twoway ANOVA: Relative grip-strength and global nutritional status: p<0.0001

Relative grip-strength and severity of liver disease: p<0.05. 80

Figure 2.1 Assessment o f muscle function using hand held dynamometer. 100% y

□ Neglible intake TW&i Inadequate intake 60% j □ Adequate intake

4094 ^ t 1 3W%t 2M4j

10?4 f ! 0% ^ ______Well Moderately Severely nourished malnourished malnourished (n=92) (n=85) (11= 3 8 ) Global nutritional category

Figure 2.2 Nutritional status and adequacy of energy intake in 215 patients with cirrhosis. Significance of difference between groups, p<0.0001. 00 100% r I 90% I 80% I 70%& 60%^ ^ Unrestricted diet £3 Dietary restrictions

40% , 30?4 j 20% 4

10% I 0% i Well Moderately Severely nourished malnourished malnourished (n=92) (n=85) (n=38) Global nutritional status

Figure 2.3 Nutritional status and dietary restrictions in 215 patients with cirrhosis. Significance of difference between groups, p<0.05. 00K) 100% y

80% j

60% T E3 No supplements □ Supplements 50% 4 Ïo ! 1 40% 4

10% -f

0% m a Well Moderately Severely nourished malnourished malnourished (n=92) (n=85) (n=38) Global nutritional status

Figure 2.4 Nutritional status and oral nutritional supplements in 215 patients with cirrhosis. Significance of difference between groups, p<0.0001. U)00 100%

90%

80%

70%

60% □ Negligible intake El Inadequate intake c 50% E3 Adequate intake

40%

30%

20%

10% m #

0% Child's A (n=65) Child's B (n=63) Child's C (n=87) Severitv' of liver disease

Figure 2.5 Adequacy of energy intake and severity of liver disease in 215 patients with cirrhosis. Significance of the difference between groups, p<0.001. 85 CHAPTERS

THE RELATIONSHIP BETWEEN MALNUTRITION AND ADVERSE CLINICAL OUTCOME IN PATIENTS WITH CHRONIC LIVER DISEASE.

3.1 Outline Poor nutritional status has been associated with increased morbidity and mortality, in a variety of patients, including those with chronic liver disease. In spite of the difficulties of evaluating nutritional status in this patient population, it has been shown that patients with chronic liver disease who undergo surgery are more at risk from complications arising from poor nutrition and have less chance of survival. It is not surprising that, although liver transplantation is now an established treatment for patients with end-stage chronic liver disease and for some forms of acute liver failure, post-transplant outcome is adversely influenced by pre-transplant nutritional status.

In order to investigate this association more fully, the survival of 271 transplanted patients who had been nutritionally assessed prior to surgery using a global method, incorporating objective and subjective variables, was examined. A significant relationship was found between pre-transplant nutritional status and survival at 90 days after surgery. None of the individual objective or subjective pre-transplant nutritional variables were significantly associated with survival. This study confirms that malnourished patients have a worse survival after liver transplantation and highlights the predictive value of global nutritional assessment.

3.2 Introduction A close association between malnutrition and morbidity and mortality has been recognised for many Early work centred on the relationship between impaired nutritional status and poor outcome after surgery. Studley^^^, for example, reported a higher mortality in patients with greater pre-operative weight loss who underwent surgery for chronic gastric ulceration. However, it has only been in the last 25 years, with the publication of results from surveys of the Chapter 3 Malnutrition and clinical outcome 86 nutritional status of hospitalized patients^^’"^®’^^^’^^^’^^"^ that the prevalence of malnutrition in both surgical and medical hospital patients has been more fully documented. Subsequent studies have confirmed that patients who are poorly nourished are more at risk than well-nourished individuals^^’^"^^ and have shown that this may be exhibited as impaired muscle function^^’^^^, reduced respiratory^^’^^’^^^ and cardiac function^^^’^^^ or an increased susceptibility to infection'^^’^^ and an increase in mortality^^’^^^.

3.3 Nutritional status and clinical outcome in chronic liver disease. Patients with chronic liver disease who are more malnourished have a poor outcome^’^^^’^^^’"^®^ (Table 3.1). O’Keefe and colleagues'^®^ assessed the nutritional status and immunocompetence of 156 patients with liver disease, of whom 42% had cirrhosis, and reported that significantly more patients who were fat depleted, as defined by a low TSF, failed to react to antigen skin tests than patients whose fat stores were considered adequate. They did not, however, suggest a causative link between the two variables but suggested that immunocompetence might be improved in this patient population by re-feeding. However, it is more likely that impaired immunocompetence is a feature of liver disease per se and is not directly related to nutritional status'^^^.

Mendenhall and colleagues^^^ evaluated the nutritional status of 352 patients with alcoholic hepatitis at presentation and during the following year. A protein calorie malnutrition score was assigned at initial assessment on the basis of a numerical score derived from the patients’ percentage of ideal body weight, TSF, MAMC, creatinine height index, serum albumin and transferrin levels, total lymphocyte count and delayed cutaneous hypersensitivity. A significant relationship was found between the 1 year mortality rates and the degree of malnutrition at presentation. Rates of 14%, 43% and 76% were observed respectively in patients with mild, moderate and severe malnutrition.

Abad and colleagues^ prospectively studied 90 hospitalized patients with cirrhosis separated into three groups on the basis of their nutritional status determined by Chapter 3 Malnutrition and clinical outcome 87

anthropometric measurements and serum albumin concentrations. At the end of an unspecified period, a significant stepwise increase in mortality was observed from 10% in patients who were acceptably nourished, to 19% in the moderately malnourished and 43% in the severely malnourished. The prevalence of ascites and infection was also significantly higher in the malnourished patients than in those with acceptable nutritional status, although the relationship to disease severity was not taken into account. The methodology by which ascites, infections and severe infections were defined was not described.

Lautz and colleagues^^^ assessed 123 patients with cirrhosis and categorized them as either normally nourished or malnourished on the basis of their body weight, TSF, BCM determined by BIA, and plasma albumin concentrations. The incidence of resistant ascites in the 6 months prior to assessment and the mortality rate, after a mean follow-up period of 845 days, were significantly higher in the malnourished group (44% vs 24%, p<0.05; 35% vs 16%, p<0.05). However, there was also an association between the degree of malnutrition and the severity of the liver injury but confounding effects were not examined; this raises the question that the adverse outcomes observed may have been influenced by the severity of illness rather than, or in addition to, nutritional status.

Merli and colleagues^^^, undertook a prospective investigation of the role of malnutrition as a risk factor for mortality in patients with cirrhosis. In this multicentre study, undertaken at 16 centres throughout Italy, 1053 patients with cirrhosis were nutritionally assessed using clinical and anthropometric variables. Patients were followed for five years and survival data collected. The depletion of muscle or fat stores, defined by a measurement of the mid-arm muscle or fat area of less than the 5th percentile of an age- and sex-matched population^^^, was associated with a higher risk of mortality (relative risk [95% confidence intervals]: 1.79 [1.47-2.20] and 1.35 [1.02-1.70] respectively). Impaired nutritional status, assessed on the basis of nutritional history and physical examination, was also found to be associated with an increased risk of mortality, with relative risks of 1.78 [1.42- 2.25] and 2.42 [1.81-3.24] associated with the categories moderate and severe Chapter 3 Malnutrition and clinical outcome 88 malnutrition. However, multivariate analysis did not identify any of the nutritional variables as independent risk factors. The authors suggest that the presence of other factors, for example, ascites and oesophageal varices may obscure the importance of nutritional status, particularly in those patients with more severely impaired liver function.

From these studies it can be concluded that an impaired nutritional status in patients with chronic hver disease is associated with an increased mortality although severity of liver disease is a major confounding factor. Further studies are required to confirm a relationship between nutritional status and morbidity. Although the association with mortality may not be independent, it is of interest because treatment addressing impaired nutritional status may have the potential for improving clinical outcome.

3.4 Nutritional status and prognostic indices for use in patients with cirrhosis The prognostic significance of nutritional status was identified in patients with cirrhosis undergoing surgery for portal hypertension and incorporated into the index proposed by Child and colleagues^®^. The index proposed three categories of severity of cirrhosis. A, B and C, into which patients could be assigned on the basis of five variables comprising clinically assessed nutritional status, encephalopathy, ascites and serum albumin and bilirubin concentrations. The index was validated by comparing the post-operative mortality in 128 patients who underwent end-to-side portal decompression for recurrent variceal haemorrhage. None of the patients in category-A, who were the least severely ill, died after surgery, while the mortality in categories-B and -C was 9% and 53% respectively. These figures cannot be directly interpreted in relation to nutritional status as separate data for each of the five variables were not published.

Child’s index^^^ has been superseded by a modified version proposed by Pugh and colleagues who replaced nutritional status with prothrombin time"^^. The rationale for doing so was not discussed and, indeed, the sole tangential reference to nutrition is the reporting of the death of one patient from anorexia nervosa. It has Chapter 3 Malnutrition and clinical outcome 89

been suggested that the reason for the replacement of nutritional status was that the objective measurement of prothrombin time was statistically more powerful than the subjective assessment of nutrition'^^

Subsequently, Christensen and colleagues undertook a study of 415 patients with cirrhosis who were followed for up to 12 years^®^ and found a significant association between nutritional status and prognosis and incorporated a nutritional variable into their prognostic index. However, the index is mathematically more complex than that of Pugh et al"^^ and consequently less frequently used. A number of other prognostic indices of varying complexity have been devised^’^’'^®^’'^^^ but none incorporate nutritional variables. Orrego and colleagues'^®^ derived the combined clinical and laboratory index from eleven predictive variables which included anorexia, defined on the basis of a patient reporting a lack of appetite and, as a result, eating less than normal. Although patients with anorexia were significantly more likely to die within 1 year than patients with a good appetite (53% vs 27%, p< 0.001), anorexia was not considered sufficiently statistically powerful enough to be included in the variables comprising the final index.

3.5 Nutritional status and clinical outcome in patients with liver disease undergoing abdominal surgery. The relationship between nutritional status and clinical outcome following surgery has not been extensively studied in patients with chronic liver disease^®^’^^^’^®^’"^^^ (Table 3.2).

Garrison and colleagues^^^ retrospectively reviewed 100 patients with cirrhosis who had undergone major abdominal surgery. Significantly more patients identified in the medical notes as malnourished died during the post-operative period than those identified as well nourished and the rate of post-operative complications was higher in the malnourished survivors.

Similar results were reported by Halliday and colleagues^®^ who prospectively assessed the nutritional status of 32 patients undergoing surgery for malignant Chapter 3 Malnutrition and clinical outcome 90

hepatobiliary and pancreatic disease. The patients who died had a significantly greater mean pre-operative weight loss, a lower mean total body potassium (TBK) and a smaller mean MAC.

There are few data on the association between impaired nutritional status and post­ operative complications but those available support a relationship. Further studies, using optimum methods of assessing nutritional status, are required in this patient population.

3.6 Nutritional status and clinical outcome in patients undergoing liver transplantation. Patients selected for liver transplantation are often malnourished^^’^*^'^’^^’^^^’^^^’^^^. It is difficult to compare the prevalence of malnutrition in different studies as the assessments undertaken have been based on different variables including anthropometric, laboratory, immunological and global factors. Hasse^^^, for example, found that of 570 patients evaluated by SGA prior to transplantation, 148 (26%) were well nourished, 348 (61%) were mildly to moderately malnourished while 74 (13%) were severely malnourished^^\ Other investigators^^’^^ reported similar findings with a high prevalence of malnutrition in liver transplant candidates from all aetiological groups.

The relationship between impaired nutritional status and post-transplant outcome has been investigated in a number of studies in both adults and some of which have been published in only abstract form^’^°’^°^’^^^ (Table 3.3). In order to examine the relationship between nutritional status and clinical outcome, appropriate nutritional assessment methods are required. However, because of the confounding effects of liver dysfunction, including fluid retention^^ and abnormalities of protein metabolism^^\ many assessment techniques used in other populations are not valid in patients with chronic liver disease. Some studies, which are discussed below, have used questionable methodology for assessing nutritional status^®’"^^^. Chapter 3 Malnutrition and clinical outcome 91

Shaw and colleagues"^^^ observed a significant correlation in 160 adults undergoing liver transplantation, between the malnutrition score, based on the degree of muscle wasting, and 6 month survival. Block and colleagues^^, from the same transplant centre, reported a significant relationship between pre-transplant BMI in 81 patients without ascites and length of stay in the intensive care unit and post­ operative survival to initial discharge. Similarly, Hasse and colleagues^^^ found that severely malnourished transplant candidates, identified by global assessment, had a significantly longer post-operative stay in the intensive care unit than better nourished patients.

Harrison and colleagues^^^ examined the relationship between post-transplant outcome variables with pre-operative MAMC in 95 patients undergoing elective liver transplantation. In a multiple regression analysis, significant inverse relationships were found between the pre-transplant MAMC and the number of post-operative infective episodes. However, no significant relationships were observed by Bathgate and colleagues^ who compared the length of hospital stay and survival at 30 days after surgery in 79 transplant candidates who were categorized on the basis of pre-operative anthropometric measurements.

Pikul and colleagues"^^^ retrospectively reviewed 68 adults who had been nutritionally assessed using the SGA method of Detsky et aP^^ before undergoing liver transplantation and found that patients categorized as moderately and severely malnourished, were ventilated for significantly longer post-operatively and spent significantly longer in the intensive care unit and in hospital than adequately nourished patients. A significantly higher mortality rate was also reported in the moderately and severely malnourished groups before discharge from hospital.

A significant relationship was identified between pre-transplant nutritional status and outcome in a multicentre study of 436 patients transplanted for cholestatic liver disease"^^^. On univariate analysis, nutritional status and 12 other variables, including age. Child’s grade and renal failure, were found to be statistically predictive of blood loss, length of intensive care stay and early post-transplant Chapter 3 Malnutrition and clinical outcome 92

complications. No variables were significantly associated with complications occurring more than 30 days after surgery although nutritional status came closest to statistical significance (p = 0.075). However, nutritional status was not included in the predictive model developed from the highest ranking predictive factors in the multivariate analysis which were age, renal failure. Child’s grade and UNOS status (United Network for Organ Sharing^"^^).

Some criticisms may be made about this study in relation to the nutritional data. The methodology for assessing nutritional status was not described or referenced although it was stated that information was collected on the National Institutes of Health Liver Transplant Database forms at all the three centres, Baylor University Medical Centre (Dallas, Texas), University of Pittsburgh (Pittsburgh, Pennsylvania) and Mayo Clinic (Rochester, Minnesota) between 1985 and 1994. Detailed and published pre-transplant nutrition assessment procedures were in practice at that time at Baylor (Jeanette Hasse, personal communication, 1992) and the Mayo Clinic (Sara DiCecco, personal communication, 1992) while no formal procedures for such assessment were in place at Pittsburgh (Diana Cabelof, personal communication, 1992). It is likely that the nutritional data collected from the three centres differ; it may be speculated that these differences, when combined, reduce the predictive power of nutrition status in relation to outcome. Indeed, a greater proportion of patients at Pittsburgh were categorized as having an excellent nutritional status (50%) compared with Baylor (20%) and the Mayo Clinic (16%). Therefore, this paper makes a limited contribution to the study of the association between nutritional status and post-transplant outcome.

Selberg and colleagues"^^^ pre-operatively assessed nutritional status in 150 patients undergoing liver transplantation, using anthropometry, creatinine height index and BIA and then monitored outcome for a mean of 48 months. Survival was lower in patients with a lower BCM, although the difference was not significant. The merit of using single frequency BIA to assess BCM in a patient population of whom 61% have ascites and 41% oedema, is questionable^^’^^^. No association was reported between survival and other nutritional variables. Chapter 3 Malnutrition and clinical outcome 93

Moukarzel and co-workers^^^ studied 102 children undergoing liver transplantation for end-stage liver disease and concluded that the height Z score was an important prognostic factor in the evaluation of post-transplant outcome. The height Z score is calculated from:

Z = individual patient’s height - mean height for age standard deviation height for age where the mean and standard deviations for height are derived from a reference population. The resulting score represents growth and thus reflects the nutritional status of children"^^^.

Weight, expressed as a standard deviation score, was identified by Rodeck and colleagues"^^^ as one of six independent significant predictors of mortality in a study of 118 children undergoing liver transplantation. Shepherd and colleagues'*^^ examined the nutritional profiles - using clinical, biochemical and body composition methods - in 37 children awaiting liver transplantation and reported that the post­ transplant survival was significantly greater in those who were relatively well- nourished compared with those who were undernourished (88% vs 38%, p < 0.003). Beath and colleagues^^ studied the outcome in 25 babies transplanted in the first year of life and identified three infants whose stay in the intensive care unit was prolonged because of dependency on mechanical ventilation as a consequence of severe malnutrition. They also reported a non-significant trend towards a longer hospital stay in those patients who were malnourished on the basis of anthropometric measurements.

3.7 Summary From these data it can be concluded that, despite some difficulty in defining nutritional status in patients with chronic liver disease and the confounding effects of the severity of liver disease, that malnutrition is associated with an adverse clinical outcome in a variety of different patient populations with liver disease. Further confirmatory studies are needed. Chapter 3 Malnutrition and clinical outcome 94

3.8 Defining clinical outcome In order to examine the relationship between impaired nutritional status and clinical outcome it is necessary, first, to review the outcome measures that can be investigated. The term "outcome" has been defined as the result, effect or product of an event or situation^\ When used in the clinical situation, outcome carries an implication of either advantage to the patient, realisation of the aims of treatment or of benefit to society, or conversely and negatively, of detriment or harm to the patient or failure of treatment. The concept of clinical outcome is highly dependent on the area being examined and, using some non-nutritional examples, may vary from the number of patients being reinfected with hepatitis C virus after liver transplantation^^ to the rate of abstinence from alcohol in patients undergoing detoxification^^.

In the discipline of nutrition, a positive outcome may be associated with improved mortality rates, reduced morbidity, an improvement in the quality of life or decreased treatment costs^^^’^^^’^^^’'^^^. Outcome measures must reflect these benefits in an unambiguous and quantifiable way and ideally should be evaluated in a defined population or in response to a specific disease or treatment event^"^^.

The field of liver transplantation provides such an arena in that a discreet, albeit heterogeneous, sample of patients can be nutritionally examined before surgery and followed subsequently. Obvious outcome measures in this patient population include survival at timed end-points, length of hospital stay, length of intensive care stay, ventilation time, episodes of infection, treatment costs and the perceived quality of life of the patients. The advantages and disadvantages of collecting these data have been considered and survival at 30, 90 and 365 days after transplantation was considered to be the most practical, precise and informative (Brian Davidson, personal communication, 1994).

The collection of data on ventilation times and length of stay have been rejected on the grounds that these outcome measures were frequently influenced by non- clinical issues, for example, the individual practice of clinicians, the requirement of Chapter 3 Malnutrition and clinical outcome 95

intensive care unit beds for other patients, the distance the study patient lived from the hospital and the presence or absence of family support at home. Death is obviously unambiguous and not open to the variable interpretation which may occur with defining episodes of infection or other complications. Few patients are lost to follow-up after transplantation making it relatively easy to collect information about their continued survival for up to a year after surgery. However, death is also the ultimate negative outcome of transplantation and, therefore, unless it has a strong association with nutritional status this may not be identified without undertaking a multivariate analysis to consider all other factors including patients’ age, underlying disease and donor graft characteristics. The undertaking of such a study was beyond the scope of this research although all clinical data have been retained for possible future review.

3.9 STUDY 3: THE RELATIONSHIP BETWEEN PRE-OPERATIVE NUTRITIONAL STATUS AND SURVIVAL FOLLOWING ORTHOTOPIC LIVER TRANSPLANTATION

3.9.1 Aim To examine the relationship between pre-operative nutritional status and survival in patients undergoing orthotopic liver transplantation.

3.9.2 Patients Between October 1988 and July 1997, 428 orthotopic liver transplants were carried out at the Royal Free Hospital, London. Nutritional assessment was undertaken in 300 (70%) of these individuals, a median (range) of 37 (1-569) days prior to transplantation. Data from 29 patients who underwent re-transplantation within 1 year of their first operation were excluded from subsequent analysis, giving a main population of 271 individuals. Within this population, 110 patients with cirrhosis in whom assessments were undertaken s30 days before surgery were identified and are subsequently referred to as the subgroup. Chapter 3 Malnutrition and clinical outcome 96

The main patient population of 271 individuals comprised 166 men and 105 women of mean (range) age 46.3 (5-68) years while the subgroup comprised 76 men and 34 women of mean age 48.0 (9-68) years. Post viral and biliary liver disease were the most common reasons for transplantation in both the main patient population and the subgroup (Table 3.4). Cirrhosis, where present, was diagnosed before transplantation on the basis of historical, clinical, laboratory, radiological and histological findings and was confirmed after surgery by histological examination of the explanted liver.

3.9.3 Nutritional Assessment Pre-transplant nutritional status was assessed in a standardized manner in all patients by a single observer in 266 (98%) patients. The remaining five patients were assessed by the same observer in collaboration with the paediatric and the intensive care unit dietitians. Patients were assessed either as outpatients or during hospitalization for their pre-transplant evaluation or for management of clinical problems. All patients were assessed globally with data collected on clinical history, physical evaluation and anthropometric measurements. The methodology was developed and refined over the 8M year period of data collection but overall, the same pattern of assessment was used throughout.

3.9.3.1 Clinical history: Data on the patient’s clinical history were obtained from their medical notes, nursing records and by interview. Because some patients did not speak English, additional information was obtained with the assistance of interpreters or English speaking relatives.

3.9.3.2 Nutrient intake: Information was collected on appetite, food intake, dietary restrictions and supplementation in 253 (95%) patients. jEnergy intake was then subjectively categorized as adequate, inadequate or negligible in comparison with estimated requirements"^^^.

3.8.3.3 Physical status: A subjective evaluation of fat and muscle stores and evidence of fluid retention was made by visual inspection. Muscle function was Chapter 3 Malnutrition and clinical outcome 97

evaluated in 119 (44%) patients assessed after October 1993 using a hand held dynamometer to measure grip-strength (TKK5101, Takei Scientific Instruments Co Ltd, Tokyo, Japan). Patients held the dynamometer in their non-dominant hand and compressed its two bars together as hard as possible. Three measurements were made and the highest reading was then compared against reference values^^^.

A subjective estimation was made of changes in body weight in the 6 months prior to interview in 210 (77%) patients, excluding, where possible, differences that resulted from fluctuations in fluid retention. Changes were expressed as kg / month.

3.93.4 Anthropometric measurements: Height was measured to the nearest centimetre with patients standing in socks or bare feet, using a wall mounted or free-standing measure (Seca, Hamburg, Germany). In a small number of bed- bound patients, an estimate of height was made by measuring their length in the supine position using a pocket stadiometer (Holtain Ltd, Crymych, Dyfed, UK) with the assistance of another member of staff. All patients were weighed to the nearest 0.1 kg, wearing night-wear or light indoor clothing and without shoes, using a weighing scale (Seca, Hamburg, Germany) or a seat balance scale (Marsden W/M Group, London, UK).

Height and weight measurements were recorded in 259 (96%) and 263 (97%) patients in the main population and 103 (94%) and 106 (96%) patients in the subgroup respectively. An estimate of dry weight was made using clinical history of fluid retention, previously documented weight records and, after 1992, published guidelines'^. BMI was calculated from estimated dry weight. MAC and TSF were measured using standardized techniques^"^^’^"^^ and MAMC was calculated^^^. Measurements were undertaken in 254 (94%) patients in the main study and 100 (91%) patients in the subgroup respectively. Relative TSF and MAMC were calculated as a percentage of the 50th percentile reading for the appropriate age and sex group from published standard data^^. Eight children and adolescents below the age of 18 years were excluded from the comparison with standard Chapter 3 Malnutrition and clinical outcome 98

figures.

3.Ç.3.5 Global assessment: Patients were categorized subjectively as either adequately nourished or as mildly, moderately or severely malnourished and the proportion in each category calculated. The main patient population was divided into six cohorts with each, except the final cohort, comprising 50 consecutively assessed patients. The proportion of patients in each nutritional category was calculated in each cohort in order to observe any trends in categorization with time.

3.9.4 Transplant outcome Survival data were collected from the clinical records, the computerised Royal Free Hospital Patient Administrative System and the records held by the Liver Transplant Co-ordinators. The proportion of patients alive at 30, 90 and 365 days after transplantation in each nutrition category was then calculated.

3.9.5 Statistical analysis Statistical analyses were performed using the Minitab software package (1994, State College, PA, USA). The association between pre-transplant global nutritional status and patient survival at 30, 90 and 365 days after transplantation was tested using a Chi-squared analysis. The relationships between the objective pre­ transplant variables, such as height, weight, BMI, relative TSF, relative MAMC and grip-strength, and post-transplant survival were examined using unpaired Student’s t-tests. The relationships between subjective pre-transplant variables, such as energy intake and estimated weight changes, and post-transplant survival were examined using a Chi-squared analysis and unpaired Student’s t-tests. Analyses were undertaken in the main patient population, the subgroup and, in both of these, in male and female patients separately. Analyses were repeated in the main patient population after excluding the data from 11 patients with a BMI above 30 to remove any potential confounding effects of obesity. Normally distributed data are expressed as mean values ± 1 standard deviation; skewed data are expressed as median (range) values. Chapter 3 Malnutrition and clinical outcome 99

3.9.6 Results The proportion of patients categorized into the four nutritional status groups was similar in both the main patient population and the subgroup (Table 3.5). Over the year period during which the data were collected, no obvious trends or changes were observed in the proportion of patients in each nutritional status category (Figure 3.1).

During the study period, 71 patients from the main patient population and 32 patients from the subgroup died. Death occurred a median (range) of 67 (0-345) days after surgery in both populations.

In the main population, a significantly higher proportion of patients with an adequate pre-transplant nutritional status were alive 90 days after transplantation compared with those who were malnourished (p < 0.005)(Table 3.6 and Figure 3.2). A similar trend was observed in the subgroup but this was not, however, statistically significant (Table 3.6). No statistical difference was found in survival at either 30 or 365 days after transplantation in relation to pre-transplant nutritional status in either the main patient population or the subgroup. When male and female patients from both the main population and the subgroup were considered separately, similar trends were observed between pre-transplant nutritional status and survival but were only significant for the male patients of the main population at 90 days (p<0.05).

No significant differences were found between the objective pre-transplant variables in patients alive or dead at the three time points in the main patient population (Table 3.7), the subgroup, or male or female patients in either population.

No significant differences were found between the subjective pre-transplant variables,! energy intake and estimated weight change, in patients alive or dead at 30, 90 or 365 days in the main patient population, the subgroup, or in either group when male and female patients were considered separately (Tables 3.8 and 3.9). Chapter 3 Malnutrition and clinical outcome 100

No changes were observed in the statistical significance between any of the variables considered and survival at 30, 90 or 365 days after surgery in any population when the 11 obese individuals were excluded from the analyses.

3.9.7 Discussion This study confirms an association between pre-operative nutritional status and survival after liver transplantation, indicating that patients who are malnourished before undergoing surgery have an increased mortality within the first 90 post­ operative days. One year after transplantation the association between survival and pre-operative nutritional status was no longer significant. The significant association within the first 90 days was, however, only found in relation to global assessment of nutritional status and not in relation to any of the individual objective or subjective pre-transplant variables.

3.9.7.1 Methodology: In the current study, small changes in methodology occurred as the assessment process was developed over the 8% years of data collection but the effects of this were minimized by the continuity of one observer, alone or in collaboration, undertaking the assessments in all patients. An examination of the proportions of patients categorized in each of the successive cohorts of patients revealed no trends to change classification criteria over the duration of the study. Although this is a crude method of evaluating the consistency of assessment over time, because it depends on the stability of the patient population, no alternative procedures, including periodic reclassification, were considered viable.

The methodology used to assess nutritional status has been previously discussed and, therefore, consideration in this chapter will be limited to the difference in methodology and characteristics of the population specific to the present study. The collection of clinical and nutritional information from patients undergoing a pre-transplant nutritional assessment is perhaps facilitated by the patient’s and their family’s heightened awareness about their condition and, in the light of previous Chapter 3 Malnutrition and clinical outcome 101 advice, their nutrient intake. However, because of their anxiety to be accepted onto the transplant programme, some individuals presented an over optimistic report of their nutritional well-being, making a true assessment more difficult to obtain and relying more heavily on anthropometric data and clinical impressions.

Many of the patients assessed in the current study had travelled a long distance, including from overseas, to the Royal Free Hospital and a considerable number did not speak English. Although the interpreters or family members who assisted were generally able to provide a broad outline describing illness and food intake, some of the subtleties of questioning, for example about non-fluid related weight loss or the precise quantification of nutrient intake, were lost in translation. In addition, some individuals including British patients referred from other hospitals, had been given inappropriate dietary advice, including instructions to restrict protein and fat intakes. While it might be expected that such discrepancies in treatment should not influence assessment of nutritional status, their occurrence in a pre-transplant population was notably more frequent than that encountered in previous studies where the majority of patients were solely managed at the Royal Free Hospital.

Assessment of nutritional status was undertaken, at least for the first 6 years of the data collection, for clinical rather than research purposes. All patients who were being considered as potential transplant candidates were, therefore, assessed as fully as possible within the limitations of their clinical condition. As a result, the main patient population in the current study includes some patients who were very ill, some of whom were on the intensive care unit when assessed. This is in contrast to the earlier studies in this thesis where assessments were undertaken, primarily at least, for research purposes and, as a result, were not carried out on individuals who were very ill or where an additional interview and measurements were deemed to be an unacceptable burden to the patient. Ideally, the severity of liver disease should have been described using Pugh’s grading system in patients with cirrhosis"^^ but, unfortunately, insufficient data were collected from the earlier patients assessed to facilitate this. However, if this data were available the Chapter 3 Malnutrition and clinical outcome 102

confounding effects of the severity of liver disease on the relationship between nutritional status and survival could be examined.

3.9.7.2 Categorization of nutritional status: Four nutritional categories were used in the current study, compared with the three in Study 1:

Study 1 Well nourished, moderately malnourished (or suspected of being so), severely malnourished; Study 3 Adequately nourished, mildly malnourished, moderately malnourished, severely malnourished.

When data collection started in 1988, no papers addressing the assessment of nutritional status in patients with liver disease had been published. As assessments were undertaken for clinical purposes, it was considered that four categories would be the most useful to identify individuals in need of nutritional intervention. However, with the subsequent publication of work by Hasse^^^’^^^ it seemed appropriate to adopt her three categories because these had at least undergone limited validation and their use would facilitate comparisons between populations. In addition, it is more difficult to achieve acceptable reproducibility in a scheme with four categories than in one with three and it seemed justifiable to compromise the possible clinical advantages of the former in order to develop a method that can be reproduced. The apparent loss of sensitivity arising from the use of three rather than four categories is overridden in clinical care when the contributing objective variables are systematically measured and used for monitoring purposes.

Despite the advantages of using three categories to evaluate nutritional status, four categories were retained in the current study. The justification for doing this lies with the unacceptability of re-categorizing patients when their clinical outcome was known. If re-categorization was undertaken simply to review the prevalence of nutritional impairment within a liver transplant population, the result might be considered a little imprecise but, generally, acceptable. However, having become Chapter 3 Malnutrition and clinical outcome 103 familiar with the patients undergoing transplantation, often through months of clinical nutritional support, and being particularly aware of the deaths that occurred in malnourished individuals, it would be impossible to re-categorize the nutritional status of these individuals without bias. Consequently, it was decided to complete the current study using the four categories as they had been originally designated rather than compromise the scientific basis of the study by re-categorization.

3.9.7.3 Time between assessment and transplantation: Another cause for criticism of the study is the length of time between assessment of nutritional status and transplantation. In the main patient population, the median time between assessment and surgery was 37 days, ranging from 1 to 569 days. Obviously, patients’ nutritional status may change considerably during a period of five weeks and, generally, in this patient population, deterioration is more likely than improvement in the absence of a re-feeding programme. The length of time between the patient’s acceptance on the transplant list and the transplant taking place also varied from one day to more than one year. In the relatively well patients, nutritional assessment was undertaken during their pre-transplant evaluation, which, if they were accepted as being suitable for transplantation, immediately preceded their name being added to the waiting list. If these patients remained well and did not require further hospitalization during the waiting period, no further nutritional assessment would be undertaken. However, in sicker patients or those encountering a medical complication, in whom a change in nutritional status would be more likely, additional dietetic input was given with a concomitant updating of their nutritional assessment. Although this method cannot be described as flawless, the assessments undertaken are considered to reflect realistically the nutritional status of patients undergoing transplantation. No studies have been undertaken, to date, to assess the change of nutritional status with time in either pre-transplant patients or a population with chronic liver disease. Such work might justify the inclusion of all the patients assessed within the study.

In order to circumvent the potential difficulty of extended times between Chapter 3 Malnutrition and clinical outcome 104

assessment and transplantation, a more homogeneous subgroup of patients was selected who had undergone nutritional assessment not more than 30 days before transplantation. Only patients transplanted for cirrhosis were included in order to eliminate the confounding effects of individuals with acute liver failure. In the resulting subgroup, the proportion of patients in each of the nutritional categories was very similar to that in the main patient population with similar objective pre­ transplant variables. However, despite these similarities the association between pre-transplant nutritional status and post-transplant survival did not reach statistical significance, although at 30 and 90 days after surgery, a similar trend to that in the main patient population was observed. The lack of significance may be due to the smaller number of the patients in the subgroup rather than any innate difference in the relationship between pre-transplant nutritional status and survival in this patient population.

3.9.7.4 Objective and subjective variables: While finding a significant association between pre-transplant nutritional status and post-transplant survival was not unexpected, the absence of such a relationship between any of the contributing objective and subjective pre-transplant variables and survival was less anticipated. Examination of the data in Table 3.7 failed to reveal even general trends between the objective pre-transplant variables from patients alive and those who had died within 30, 90 and 365 days of transplantation. It could be predicted that any effect of a poor pre-transplant nutritional status would have been dissipated by 365 days after transplantation, but the absence of an association at 30 and 90 days requires consideration. Large standard deviations were observed for some of the variables, for example in Table 3.7 the standard deviations of the pre-transplant TSF considered at 30 days are 45% and 54% of the respective means. However these large ranges do not account for the lack of difference as higher mean values are not always found in the patients who were alive.

The combining of anthropometric data from male and female patients may obscure the true picture of muscle and fat stores as it is well recognised that there is a sex Chapter 3 Malnutrition and clinical outcome 105

difference in the relative depletion of these stores in patients with cirrhosis^^’^^^. However, separating the data for the sexes did not clarify the position or result in significant differences between the patients who were alive and those who were not. Similarly, the effects of obesity may have obscured the beneficial effects of adequate but not excessive body fat and muscle stores. Re-analysis of the data after exclusion of obese individuals did not affect the results; previously studies have reported that post-transplant morbidity and mortality rates are not increased in obese patients^^’^^.

While each of the individual objective and subjective variables contributes to the overall nutritional status of the individual, no single variable is powerful enough to predict survival after transplantation. This is plausible considering the complexity of the transplant procedure, post-operative immunosuppression and the multitude of other non-nutritional factors that may influence outcome. The absence of a predictive role for any single individual objective or subjective pre-transplant nutritional variable adds weight to value of the global assessment and underlines the importance of using a reproducible method for assessment.

3.9.7.5 Implications of nutritional status in a wider transplant context: Since the first liver transplant was performed 25 years ago^^°, this procedure has become established treatment for end-stage chronic liver disease and some forms of acute liver failure^^^. Greatly improved survival rates have been achieved over this time period, reflecting, to a large extent, better patient selection, developments in surgical techniques and improvements in donor organ management, peri-operative patient care and immunosuppressive therapy^"^^. In view of the life saving nature of liver transplantation, there is no justification for withholding surgery from severely malnourished patients despite their increased risk of post-operative death. Seventeen of the 26 severely malnourished patients who underwent liver transplantation in the current study were alive one year after surgery, some of whom have subsequently returned to work and enjoy what can be subjectively called a good quality of life. Chapter 3 Malnutrition and clinical outcome 106

The financial implications of increased morbidity need consideration but further evidence is required first that this is associated with an impaired pre-transplant nutritional status'^®®. Rather than providing evidence that severely malnourished patients should be excluded from liver transplantation, this study has identified their vulnerability and must lead to further questions about how they should be actively managed pre- and post-operatively in terms of nutritional support.

Study conclusions: The results from this study show that impaired pre-transplant nutritional status is associated with increased post-operative mortality. These data provide the basis for future investigations of the relationship between nutritional status and post-transplant morbidity and the effects of pre-transplant nutritional support on both morbidity and mortality. Table 3.1

Association between nutritional variables and clinical outcome in patients with chronic liver disease.

First author Patients Diagnosis Nutritional Outcome

Year (n) variables

O’Keefe'®^ 1980 156 Liver disease (42% cirrhosis) i TSF i Response to antigen skin tests

MendenhalT^^ 1986 352 Alcoholic hepatitis i Malnutrition score t Mortality (1 year)

Abad^ 1987 90 Cirrhosis i Anthropometry / albumin t Mortality (? period)

Lautz^^^ 1992 123 Cirrhosis i BW / TSF / BCM (BIA) / SA t Mortality (2.3 years)

t Resistant ascites

Merli""" 1996 1053 Cirrhosis i Clinical nutrition t Mortality (5 years)

i Anthropometry

BW - body weight; BCM (BIA) - BCM estimated by BIA; SA - semm albumin. Table 3.2

Association between nutritional variables and clinical outcome in patients with chronic liver disease undergoing surgery.

First author Patients Diagnosis Surgery Nutritional Outcome Year (n) variables

Child^°^ 1964 128 Cirrhosis Portal decompression for variceal -I Nutrition® t Mortality (? period) haemorrhage

Pitt""" 1981 155 Hepatobiliary disease Cholecystectomy -I Serum T Mortality (? period) Hepaticoj ejunostomy albumin Cholangioj ejunostomy

Gairison^®^ 1984 100 Cirrhosis Biliary tract procedures Malnutrition - t Mortality (? period) Cholecystectomy as clinically t Complications Duodenal ulcer repair reported Colectomy

Halliday"“ 1988 32 Malignant hepatobiliary disease Liver resection t Weight loss t Mortality (30 days) Biliary enteric bypass i TBK / t Serious post-operative Prosthesis insertion MAMC complications Laparotomy

®No further details given.

o oo Table 3.3 Association between nutritional variables and clinical outcome in patients undergoing liver transplantation.

First author Patients Nutritional variables Outcome measures Year (n)

Shaw"*®^ 1985 160 t Malnutrition score T Mortality (6 months)

Block"» 1989 81 i BMI t Mortality (to discharge) T LOS ICU

Hasse^^^ 1991 107 i Global nutritional status t LOS ICU

Harrison^»® 1993 95 i MAMC T Post-operative infective episodes

PikuP"" 1994 68 •I Nutritional status (SGA) t Ventilation t LOS ICU / LOS t Mortality (to discharge)

RiccP"» 1997 436 ■i Nutritional status t Blood loss t LOS ICU T Early post-operative complications

Selberg'*®^ 1997 150 i Anthropometry / CHI / BCM (BIA) No significant association

Bathgate^" 1997 79 i MAMC ± i TSF LOS ICU / LOS unchanged

MoukarzeP^^ 1990 102P i Height Z score T LOS t Surgical / infective complications t Mortality (1-4 years)

Shepherd'*»^ 1991 37P ■i Nutritional status t Mortality (1 year)

Beath^® 1993 25P t Malnutrition t LOS ICU

Rodeck'*®^ 1996 118P 'I' Weight Z score t Mortality (to 8 years ?)

P - paediatric patients. CHI - creatinine height index; BCM (BLAl) - BCM estimated by BIA; LOS ICU - length of stay in intensive care unit; LOS - length of hospital stay.

o 110

Table 3.4

Aetiology of liver disease in patients assessed before liver transplantation.

Data expressed as number (%).

Aetiology of Main population Subgroup

liver disease (n=271) (n=110)

Post-viral (HBV + HCV) 85 (31%) 40 (36%)

Biliary (PEG + PSC) 80 (30%) 36 (33%)

Alcohol 52 (19%) 19 (17%)

Other 54 (20%) 15 (14%)

HBV / HCV - hepatitis B and C virus;

PEC / PSC - primary biliary cirrhosis and primary sclerosing cholangitis.

Subgroup comprises patients with cirrhosis who underwent nutritional assessment within 30 days of transplantation.

Table 3.5

Categorization of patients by pre-transplant nutritional status.

Data expressed as number (%).

Nutritional status Main population Subgroup

(n=271) (n-110)

Adequately nourished 84 (31%) 31 (28%)

Mildly malnourished 89 (33%) 40 (37%)

Moderately malnourished 72 (26%) 29 (26%)

Severely malnourished 26 (10%) 10 (9%) Table 3.6

Survival after liver transplantation: The relationship with pre-transplant nutritional status in two patient populations.

Data expressed as patients alive/all patients (%).

Pre-transplant nutritional status Post-transplant survival

Mam population Subgroup

30 days 90 days* 365 days 30 days 90 days 365 days (n=271) (n=267) (n=248) (n-110) (n=110) (n=108)

Adequately nourished 80/84 76/81 55/74 30/31 29/31 22/30 (95%) (94%) (74%) (97%) (94%) (73%)

Mildly malnourished 81/89 67/89 58/83 34/40 32/40 27/39 (91%) (75%) (70%) (85%) (80%) (69%)

Moderately malnourished 65/72 56/71 47/67 28/29 25/29 21/29 (90%) (79%) (70%) (97%) (86%) (72%)

Severely malnourished 23/26 21/26 17/24 9/10 8/10 8/10 (88%) (81%) (71%) (90%) (80%) (80%)

‘Significance of difference between survival in adequately nourished vs all malnourished patients: p<0.005. Analyses of the differences at all other time points in both populations were not significant. Table 3.7

Comparison of objective pre-transplant nutritional variables in patients alive or dead at 30, 90 and 365 days after transplantation in the main population of 271 patients.

Data expressed as mean ± ISD.

30 days 60 days 365 days

alive (n=249) dead (n=22) alive (n=220) dead (n=47) alive (n=177) dead (n=71)

Height (m) 1.68 ± 0.09 1.72 + 0.10 1.68 ± 0.09 1.69 ±0.11 1.68 ±0.09 1.69 ±0.11

Weight (kg) 65.2 ± 13.3 67.5 ± 14.6 65.4 ± 13.2 64.5 ± 14.0 64.6 ± 12.9 65.1 ± 13.2

BMI (kg.m'^) 23.0 ± 3.9 23.1 ± 4.1 23.1 ± 3.9 22.6 ± 3.7 22.9 ± 3.9 22.7 ± 3.4

TSF (mm) 12.1 ± 5.5 11.4 ± 6.2 12.2 ± 5.7 11.2 ± 5.1 12.0 ± 5.9 11.8 ± 5.0

MAMC (cm) 21.8 ± 3.2 22.5 ± 2.6 21.8 ± 3.2 21.8 ± 2.9 21.6 ± 2.9 21.8 ± 3.1

Relative TSF (%) 82.7 ± 44.0 79.9 ± 50.1 83.3 ± 44.7 75.9 ± 43.7 81.4 ± 45.3 83.1 ± 42.6

Relative MAMC (%) 84.7 ± 12.2 85.5 ± 10.8 84.7 ± 12.5 84.8 ± 9.9 84.2 ± 11.8 84.1 ± 12.1

Grip-strength (kg.f) 25.2 ± 9.9 37.0 ± 9.6 25.2 ± 9.9 25.4 ± 11.0 24.8 ± 9.4 23.7 ± 9.7

Relative grip-strength (%) 62.5 ± 20.5 82.7 ± 21.2 62.4 ± 20.9 64.6 ± 18.0 61.2 ± 19.1 57.3 ± 19.7

No significant difference detected between variables of patients alive or dead at any time interval. Table 3.8

Survival after liver transplantation: The relationship with subjectively evaluated pre-transplant energy intake in two patient populations.

Data expressed as patients alive/all patients (%).

Pre-transplant energy intake Post-transplant survival

Main population Subgroup

30 days 90 days 365 days 30 days 90 days 365 days

(n=258) (n=254) (n=234) (n=104) (n=104) (n=102)

Adequate intake 134/148 125/145 91/128 45/50 41/50 34/48

(91%) (86%) (71%) (90%) (82%) (71%)

Inadequate intake 95/101 82/100 69/97 48/52 42/52 36/52

(94%) (82%) (71%) (92%) (81%) (69%)

Negligible intake 7/9 7/9 7/9 2/2 2/2 2/0

(78%) (78%) (78%) (100%) (100%) (100%)

No significant difference detected at any time interval. 114

Table 3.9

Comparison of subjectively evaluated weight change before transplantation in patients alive

30, 90 and 365 days after transplantation with those who had died by these time points from

a total population of 210 patients.

Data expressed as mean ± ISD.

30 days 90 days 365 days

alive dead alive dead alive dead

(n=195) (n=15) (n=179) (n=29) (n=140) (n=51)

Weight change® -0.87 -0.89 -0.86 -0.91 -0.79 -1.04

±1.30 +0.95 ±1.33 ±0.90 ± 1.21 ±1.36

“Weight change ejqjressed in kg/month.

No significant difference detected at any time interval. □ SevereK m alnourished

□ M oderately m alnourished

ü M ildly m alnourished

E3 A dequately nourished

Cohort Cohort Cohort Cohort Cohort Cohort 1 2 3 4 5 6 (n=50) (n=50) (n=50) (n=50) (n=50) (n=21)

Figure 3.1 Categorization of global nutritional status in consecutive cohorts of patients undergoing liver transplantation. 100 r * 90 -

80

70 □ A dequately nourished 60 M M ildly m alnourished □ M oderately m alnourished 50 ^ □ Severely m alnourished

40

30

20

10

0 4— 30 90 365

T im e post-transplantation (days)

Figure 3.2 Survival after liver transplantation in 271 patients. * Significance of the difference between adequately nourished patients and all malnourished patients, p<0.005.

ON 117 CHAPTER 4

FACTORS CONTRIBUTING TO MALNUTRITION IN CIRRHOSIS: THE ROLE OF HYPERMETABOLISM

4.1 Outline Several factors are thought to contribute to the malnutrition observed in patients with cirrhosis. There is some evidence that nutrient intake is reduced in these patients, particularly if they have ascites, follow restricted diets or are hospitalized, but, at present, few quantitative data are available. Nutrient losses may be increased in some patients with cirrhosis, particularly those with decompensated disease, but again, information is limited. Nutrient utilization may also be increased in this patient population and, while evidence is substantial in the case of protein requirements, data supporting an increase energy requirements are less conclusive.

REE was measured in 100 patients with cirrhosis and 41 comparable healthy volunteers using indirect calorimetry. Although no differences were found between absolute values for REE in patients and volunteers, significantly higher values were observed in the patients when the results were adjusted for fluid retention, body weight and fat-free mass (FFM). The possibility that increased REE may contribute to the malnutrition observed in this patient population requires further investigation.

4.2 Introduction The prevalence of malnutrition in patients with chronic liver disease is high and this is associated with an adverse clinical outcome. Elucidation of the factors important in the genesis of malnutrition in this patient population is necessary to provide a basis for treating malnutrition.

Nutrient "balance" can be described as a state of equilibrium in which nutrient intake is equivalent to the nutrients lost and those utilized'^^^: Chapter 4 Factors contributing to malnutrition 118

Nutrient intake = nutrient loss + nutrient utilization.

This equilibrium may be disrupted when nutrient intake is reduced or because nutrient loss or utilization are increased, or as a result of a combination of changes in these factors. In some clinical situations a change in a single factor may be responsible for initiating nutritional impairment, but subsequent physiological and pathological processes result in interactions which influence and potentially exacerbate impairment. For example, in starvation, an inadequate nutrient intake obviously leads to malnutrition but this may be associated with alterations in nutrient loss and utilization as adaptive changes in energy expenditure develop and as complications such as infections arise^^. In other disease states, deterioration in nutritional status may be multifactorial, for example in cancer where changes in nutrient intake, nutrient absorption and energy expenditure coexist and their effects may be compounded by treatment with surgery, radiotherapy or chemotherapy^^®.

In view of the interaction between this triumvirate of nutrient intake, loss and utilization it is unlikely that any single factor will be responsible for the nutritional impairment observed in patients with cirrhosis. However, examination of these factors may reveal the magnitude of their effects, if any, and help to determine areas where nutritional intervention could be most effectively directed. While acknowledging the importance of considering aspects of multiple nutrients in this patient population, the following discourse concentrates on energy metabolism and, to a lesser extent, protein metabolism.

4.3 Nutrient intake Early investigations of the relationship between nutrient intake and nutritional status in patients with chronic liver disease focused on individuals with alcohol- related liver and the role impaired nutritional status was thought to play in its pathogenesis^^^’"^^^. However, although there is undoubtedly a close relationship between nutrition and the development of liver disease^^’^^^, the consensus is that the association is facilitative but not causaP^^’^^^’"^^^. Chapter 4 Factors contributing to malnutrition 119

The assessment of nutrient intake is difficult^^’^^’^^’^^^, which, in patients with chronic liver disease, combined with the heterogenicity of patients studied and the different standards used for comparison^^’^^’^^^ may explain the inconsistencies in the findings published to date (Table 4.1).

Most studies have been undertaken in patients with alcoholic liver although some data are available from patients with chronic liver disease of other aetiologies^^^’^^^’^^^’^^^. The severity of liver disease of the patients studied has also varied^^’^^^.

In the majority of studies some variation of the dietary recall or diet history method has been used to obtain information on nutrient which has the advantage that data are collected in a relatively straightforward manner and analyzed without recourse to complex procedures. However, these techniques provide data of varying accuracy^^; indeed, in one study the data obtained from patients with alcoholic liver disease were considered too unreliable to be of value^^^. Nutrient intake during hospitalization has been evaluated semi-quantitatively by observation in alcoholic patients^^^ and in malnourished individuals with cirrhosis^^. However, in only one study was the nutritional intake of patients with chronic liver disease investigated using weighed food records together with chemical analysis of duplicate portions^^^.

Intake data in patients with chronic liver disease have been compared with those in healthy volunteers^^^’^^^’"^^"^’^^^, other hospitalized patients^^^’^^®’"^^^, published recommended intake values and measured energy expenditure^^^. D ata have been expressed in absolute or relative to body weight 293,361,367,495 ideal body weight^^.

Not surprisingly, the intake data recorded in these studies vary widely (Table 4.1). Cabré and colleagues^^, for example, found that severely malnourished patients with cirrhosis consumed only 60% of the calories provided by a standard hospital diet; by contrast, Noël-Jorand and coworkers^^^ reported that alcoholic patients. Chapter 4 Factors contributing to malnutrition 120

57% of whom had cirrhosis, consumed, excluding alcohol, 148% of the calories eaten by healthy individuals.

Even within studies, data from patients with liver disease of different aetiologies vary considerably. For example, Morgan^^^ assessed the intake of patients with alcoholic liver disease, primary biliary cirrhosis and chronic active hepatitis and found that, compared with control subjects, their mean energy intakes were 77%, 93% and 104% respectively. Levine and Morgan^^^ found that while the energy intake of patients with chronic liver disease was approximately 91% of that of control patients, intakes differed between patients with alcoholic and non-alcoholic liver disease, their respective energy intakes being 83% and 99% of the control values. However, when energy intake was expressed relative to body weight, rather than in absolute terms, the discrepancy between patients and controls and between the patient subgroups was no longer apparent. Notwithstanding these differences in expression of results, nutrient intake must also be considered in relation to the nutritional status of the individual and, in alcoholic patients, thought given to the contribution of alcohol intake.

It is not, therefore, possible to make a broad statement about nutrient intake in patients with chronic liver disease. Undoubtedly, some individuals have a grossly inadequate diet and this is likely to contribute to malnutrition if prolonged. However, other patients take a seemingly adequate intake and yet still develop malnutrition. Consideration of the specific factors that influence nutrient intake in this patient population is required.

4.3.1 Anorexia Anorexia, or loss of appetite, is a common finding in patients with liver disease^^^’^^’"^®^’"^^^; the prevalence increases with increasing disease severity^^. While the presence of this symptom is widely recognised, little is known about its causes in liver disease. It has been suggested that these may be associated with changes in energy expenditure or circulating concentrations of hormones, including glucagon and cholecystokinin, and nutrients including glucose and free fatty acids^^. Chapter 4 Factors contributing to malnutrition 121

In a study of 273 malnourished men with alcoholic hepatitis, Mendenhall and colleagues^^ reported that the patients with the most severe liver disease and the greatest malnutrition had the most severe anorexia and ate least well. Although they did not quantify the relationship, it is likely that anorexia contributes to impaired nutrient intake in this patient population.

4.3.2 Taste perception and food preferences Disordered gustatory acuity has been reported in individuals with liver disease^^’^^^’^^^’^^^’^*^*^’^^^ manifest as changes, usually a reduction, in taste acuity for detecting and recognising some or all of the basic tastes of bitter, salt, sweet and sour. Aversions to fatty foods, meat and fried dishes and a preference for citrus fruits have also been reported^^’^^^’^^'^’^^^, although in none of these studies was gustatory acuity tested formally. However, the effects of alterations in taste acuity and food preferences on nutrient intake have not been examined in patients with chronic liver disease, although it is reasonable to suppose that they may be interrelated^^^’^^’^^.

4.3.3 Nausea and vomiting Patients with chronic liver disease, particularly alcohol-related disease, frequently report nausea and, to a lesser extent, vomiting^’"^^^. Although these symptoms are described as contributing to impaired nutritional intake^^^'^^^'^^^, this association has not been examined. However, it is likely that such a causal relationship exists.

4.3.4 Ascites Ascites is a serious complication of chronic liver disease associated with portal hypertension and impaired liver cell function^^. Patients with ascites are often malnourished"^^’^^^’^^^’^^^’^^^’^^^’^®^’^^^. Nutrient intake maybe reduced in these patients because of the pressure effects of the accumulated fluid in the abdomen"^^’^^^’^^®’^^^.

Wicks and colleagues^^^ used a standard diet history technique to assess energy intake in 80 patients with primary biliary cirrhosis, 39% of whom had ascites, and compared the data obtained with published reference values^^^. The respective Chapter 4 Factors contributing to malnutrition 122

mean energy intakes, expressed as a percentage of the corresponding estimated average requirements^^^, in patients without ascites, with mild ascites and with moderate ascites were 97%, 92% and 80%. These results show that the presence of ascites impairs nutrient intake but this conclusion should be interpreted with caution as no statistical comparisons were made between the intakes of the subgroups or between the demographic or clinical profiles of the patients.

Franco and colleagues^^® reported that patients with gross ascites found eating difficult because of abdominal tension and early satiety. They also observed that "normal" eating was resumed 3 to 5 days after the insertion of a peritoneovenous or LeVeen shunt, a valved tube connecting the abdominal cavity to one of the jugular veins to allow ascitic fluid to re-enter the systemic circulation^^®; they did not, however, assess dietary intake quantitatively.

Intake was evaluated by Blendis and colleagues'^^ in 17 patients before and a mean of 14 months after the insertion of a LeVeen shunt and a significant increase in mean energy intake was reported following relief of ascites. Gauthier et al^^® showed that reduction of ascitic volume is accompanied by a subjective improvement in appetite but did not correlate these findings with data on nutrient intake.

Thus, the presence of ascites appears to be associated with a reduction in dietary intake. Additional work is required to confirm and further explore this relationship.

4.3.5 Dietary restrictions While it is widely recognised that therapeutic dietary restrictions may be accompanied by a reduction in the adequacy of nutrient intake^^®''^^®, few studies have been undertaken in which this association, or its effects on nutritional status, have been quantified. Where the benefits of dietary restrictions have been questioned in hepatobiliary disease^^®’^^®’^^^’^®^, it is for therapeutic rather than nutritional reasons. Chapter 4 Factors contributing to malnutrition 123

Davidson^^^ reported a subjective improvement in nutritional status in 30 patients with cirrhosis whose sodium intake was restricted to 9 mmol/day for up to 16 months as treatment for ascites. While the patients’ intake was not quantified, considerable efforts were made by staff and patients to maximise the quantity of food and supplements consumed. Thus dietary restrictions are not inevitably accompanied by a reduction in nutrient intake.

Wicks and colleagues^^^ assessed the effects of dietary restriction on nutrient intake in 42 patients with primary biliary cirrhosis using a diet history technique. Five

patients following low sodium diets consumed a mean of 1 0 1 % of their estimated requirements while 17 patients restricting fat and seven restricting both fat and sodium had mean intakes of 93% and 81% respectively. However, no statistical comparisons were made between nutrient intake in these patients or between their demographic or clinical profiles. Thus, while these results show that dietary restrictions may have a deleterious effect on nutrition intake, they must be interpreted with caution.

Soulsby and colleagues^®^ observed the effect of two different levels of sodium restriction on dietary intake over a 6 -week period in six patients with cirrhosis who had mild to moderate ascites. Intakes of energy and protein, assessed by 7-day weighed dietary records, were found to be lower on the restrictive 40 mmol sodium diet than on the more liberal "no added salt" regimen. In addition, the stricter regime was associated with a reduction in body weight and mid-arm muscle circumference. While the number of patients studied is small, the findings demonstrate a relationship between nutrient intake and the degree of therapeutic dietary restriction in this patient population.

Thus the use of dietary restrictions may result in a reduction in overall food intake. Further studies are required.

4.3.6 Iatrogenic fasting Hospitalized patients are routinely fasted before general anaesthesia and a number Chapter 4 Factors contributing to malnutrition 124

of investigative and therapeutic procedures^^^’^^^. In a recent audit of clinical practice in the United Kingdom^^ it was found that surgical patients were fasted pre-operatively for between 3 and 29 hours with a mean fasting time of 11 hours; the length of the fast depended on the procedure undertaken and the centre concerned. The effect of preoperative fasting on nutritional status may be limited, because few patients undergo surgery often enough for this to significantly impair their food intake. However, some patients may be fasted repeatedly before clinical investigations and the cumulative effect on nutrient intake may be considerable.

Liver transplant candidates, for example, undergo ultrasound and computed tomographic examination of the abdomen, coeliac angiography, upper gastrointestinal endoscopy, endoscopic retrograde cholangio-pancreatography and, often, measurement of hepatic wedge pressures. Patients are fasted for a minimum

of 6 hours before each of these procedures which, in effect, means abstaining from oral or enteral nutrient intake from the previous night for morning appointments or after an early breakfast for investigations arranged for the afternoon (Linda Selves, personal communication, 1997). It is not uncommon for procedures to be postponed or, in some cases, for additional or repeat tests to be undertaken. In some hospitalized individuals it is, therefore, possible for nutrient intake to be proscribed for a substantial part of their waking hours. While some motivated and relatively well patients may be able to increase their food intake when permitted, thus making up for the earlier deficits, this is unlikely to happen in all cases.

The extent of fasting and its effect on total nutrient intake have not been studied in patients with liver disease. The effects of eating breakfast have been investigated in healthy populations and those who ate breakfast were generally found to have higher daily intakes of energy, protein and micronutrients than those who did While caution must be used in extrapolating this evidence to the clinical situation, it is likely that iatrogenic fasting has a detrimental effect on nutritional status in hospitalized patients with liver disease.

From the evidence presented, there seems little doubt that nutrient intake is Chapter 4 Factors contributing to malnutrition 125

reduced in some patients with chronic liver disease. While this is not a universal finding, there are some patients who appear to be particularly at risk including those who have decompensated disease and ascites, those following restricted diets and those undergoing serial clinical investigations. Nutrient intake has not been directly correlated with nutritional status in patients with chronic liver disease although it is reasonable to assume that there is a causal relationship between them.

4.4 Nutrient loss Patients with cirrhosis may lose nutrients either because of malabsorption^^^ or because of impaired utilization^^^. Excess loss may also occur in relation to the complications of the disease such as gastrointestinal bleeding^^ or else as a result of treatment, for example the loss of protein which occurs following repeated paracentesis for ascites^^^.

4.4.1 Malabsorption While malabsorption is considered under the section of nutrient loss, it is acknowledged that it may also be regarded as a failure of nutrient intake. Malabsorption of fat is common in patients with chronic liver disease; faecal fat excretion is increased in 50-60% of adults with cirrhosis^^^’^^^’"^^. One of the major causes of fat malabsorption is a reduction in the intraluminal concentration of bile salts resulting from the impairment of bile acid uptake, synthesis and secretion^^^.

However, malabsorption may also arise from the direct effect of alcohol on the intestinal mucosa resulting in the depletion of brush border enzymes and impairment of sodium, water, glucose and amino acid absorption^^^.

The presence of liver disease does not affect the absorptive capacity of the small bowel per se^^^. However, patients with liver disease may develop gastrointestinal disorders which themselves cause malabsorption, for example primary biliary cirrhosis and coeliac disease^®^. Chapter 4 Factors contributing to malnutrition 126

The consequence of malabsorption is to increase intestinal losses of nutrients. The relationship between malabsorption and nutritional status has not yet been studied in patients with chronic liver disease.

4.4.2 Impairment of protein metabolism The liver plays an important role in the metabolism of protein by processing dietary and endogenous amino acids, regulating their supply to the rest of the body and converting ammonia into urea for excretion. In cirrhosis, liver glycogen stores are depleted^^^’^^^ and this is associated with reduced fasting splanchnic blood glucose concentrations'^^^. In order to maintain the level of circulating glucose, the rate of hepatic gluconeogenesis is increased^^ which results in additional losses of amino acids with potential depletion of protein stores.

The studies on whole body protein synthesis and breakdown in cirrhotic patients have produced conflicting results, mainly because of the confounding effects of variables such as the patient’s nutritional status and recent protein intake, and the fact that tracers used may perform differently in patients with liver disease than in healthy volunteers. Consequently, whole body protein synthesis rates have been reported as increased^^^’^^^’^^^, unchanged^^®’^^^’"^^ or reduced^^^ '^^^

There is some evidence that nitrogen loss in breath and faeces may be increased in patients with chronic liver disease. Fasting expiratory ammonia concentrations are generally higher in patients with cirrhosis than healthy volunteers, particularly in those with encephalopathy who also have higher blood ammonia concentrations^^®. The concentration of ammonia in the breath of patients with cirrhosis appears relatively low at 7-16 ptg/1^^®. However, if the tidal volume is 500 ml and respiratory rate 13 breaths per minute"^^, approximately 0.05-0.12 g of metabolized nitrogen could be lost in breath daily or the equivalent of 9-23 g protein per month.

Faecal nitrogen concentrations are generally not increased in patients with cirrhosis. However, faecal nitrogen losses are increased in those taking lactulose Chapter 4 Factors contributing to malnutrition 127

and other non-absorbable disaccharides for the treatment of hepatic encephalopathy as these drugs stimulate the incorporation of nitrogen into bacterial protein within the intestinal lumen^^^’^^^. While treatment with non-absorbable disaccharides increases faecal nitrogen, it reduces urinary nitrogen excretion by reducing blood ammonia and may also reduce expiratory losses^^^. Thus the overall effect of non-absorbable disaccharides on nitrogen balance may not be significant.

4.4.3 Paracentesis Current treatment of ascites includes bed rest, sodium restriction, diuretics and therapeutic paracentesis^^. The removal of large volumes of ascitic fluid must be accompanied by intravenous replacement of protein with salt-poor albumin to prevent haemodynamic disturbances and renal failure^^^’"^^^. The protein content of ascitic fluid varies between 5-60 g/1 and although only patients with malignant or infected ascites have concentrations approaching 60 g/ 1, up to 30% of patients with uncomplicated ascites have protein concentrations in excess of 30 g/1^^. A typical treatment regimen involves the replacement of 40 g albumin for every 4-6 1 of ascites removed^^^ and patients may, therefore, have a net protein loss of up to 50 g protein per litre of ascites removed. Some patients undergo repeated large volume paracentesis and if the protein losses are not adequately replaced, protein stores may diminish.

Bertrân and colleagues^^ investigated the nutritional effects of paracentesis in 17 patients with cirrhosis by measuring TSF and MAMC before and 2 days after a single paracentesis. They concluded that the procedure had no long-term effects on nutritional status because they observed no changes in the measurements over this time period. However, it is extremely unlikely that changes in anthropometric variables would be detected over such a short time period^^^. Thus while relieving ascites by repeated paracentesis is clearly beneficial to the patient and allows them to increase their food intake, a gradual depletion of protein stores may also occur unless adequate replacement is given. Chapter 4 Factors contributing to malnutrition 128

4.4.4 Gastrointestinal bleeding Gastrointestinal bleeding, another complication of chronic liver disease, is associated with decreased hepatic function and an increase in mortality of between 3% and 36% in the first two days following a bleeding episode^^’^^^. Bleeding usually occurs from oesophageal or gastric varices which develop as a consequence of portal hypertension. Acute variceal haemorrhage is a medical emergency and priority is given to correcting hypovolaemic shock by replacing blood volume with whole blood, plasma proteins and colloid expanders^^; once the patient has been resuscitated attempts are made to stop the bleeding with drugs which alter portal haemodynamics and by local approaches on the varices themselves. In general, patients are undertransfused as increasing the circulating volume excessively might result in further variceal haemorrhage^^^’"^^. As each litre of blood contains 205 g of protein^^^ repeated haemorrhage and undertransfusion might have nutritional consequences.

Nutrient losses are increased in patients with cirrhosis and, as with reductions in nutrient intake, these losses are more evident in individuals with decompensated liver disease. The mechanisms by which these losses occur are not, in all cases, fully understood nor is the extent to which they affect nutritional status known. Consequently there is a need for further studies.

4.5 Nutrient utilization The body’s requirements for energy and protein must be met first and foremost. Indeed, the demand for energy takes precedence over all other requirements^^® and is intimately related to the utilization and requirements for protein^^^’^^^.

4.5.1 Energy utilization The energy requirements of an individual are determined by their total energy expenditure which comprises basal energy expenditure, dietary-induced thermogenesis and the energy cost of physical activity.

Basal energy expenditure, or basal metabolic rate, describes the energy costs of the Chapter 4 Factors contributing to malnutrition 129

essential metabolic processes required to maintain life. Its definition is generally

accepted as the energy expended by an individual at least 1 0 hours after the last meal, resting awake in a lying position, at normal body and ambient temperature without physical or psychological stress^'^^'^"'^^^. However, because measurements of energy expenditure under truly basal conditions are difficult to achieve, REE is usually measured instead. REE has been variously and inconsistently defined^^’"^®^’"^^^ but in this thesis the term will be used to describe the energy expended by an individual under conditions as close as possible to those described for basal energy expenditure, acknowledging that the measurement may be higher than those at true basal levels. REE accounts for between 50% and 80% of total

energy expenditure in the majority of individuals i 40,i89 ,417,452,560

Dietary-induced thermogenesis comprises obligatory and adaptive thermogenesis. Obligatory thermogenesis, is the increase in energy expenditure observed in response to the intake of food. It was first described by Lavoisier (1743-1794) and later termed the "specific dynamic action" of food by Rubner in 1902^^®. Obligatory thermogenesis comprises the energy costs of digestion, absorption and assimilation of nutrients. Adaptive thermogenesis was originally described as "luxus consumption" by Grafe in 1911 and represents energy expenditure following food ingestion in excess of obligatory thermogenesis^^^; adaptive thermogenesis is thought to play a role in weight control by facilitating the utilization of excessive caloric intake^^®. Dietary-induced thermogenesis accounts for approximately 6 % of energy expenditure"^^^ although in some individuals adaptive thermogenesis alone may account for up to 15% of totaP®.

The energy expended in physical activity, sometimes referred to as the thermic effect of exercise, is the most variable component of total energy expenditure. The energy expended is dependent on the nature of the activity, the intensity with which it is carried out and the size and weight of the individual. Physical activity accounts for up to approximately 40% of total energy expenditure in most healthy individuals^^. Chapter 4 Factors contributing to malnutrition 130

4.5.2 Total energy expenditure The estimation of total energy expenditure is difficult, especially in free-living individuals^^^. Measurements can be made using a permanently sited calorimetry chamber using either direct or indirect calorimetry"^^’^*^^’^^^’^^^ but the equipment is often inaccessible and the technique complex. Because of the limited availability of the equipment and the necessity of confining the individual within the calorimeter, few studies have been undertaken on diseased patients rather than healthy volunteers. Portable devices, for example, the Douglas Bag, can be use to collect expired respiratory gases, facilitating the measurement of energy expenditure by indirect methods^^^. However, although this equipment permits much greater freedom of movement, the necessity of wearing a nose clip and mouthpiece may lead to inaccurate measurements^^^ and, by preventing oral intake, precludes assessment of energy expenditure over long periods.

Total energy expenditure can also be estimated in free-living individuals using isotope dilution techniques to estimate carbon dioxide production^^^. This is undertaken using the differential disappearance of deuterium and oxygen-18 from doubly-labelled water^^^’"^^^ or carbon-14 from a sodium bicarbonate infusion^^^. Although the doubly-labelled water method is sometimes considered a reference method for estimating total energy expenditure in free living conditions, its accuracy is dependent on a number of assumptions, including the subject’s body water turnover and mean respiratory quotient over the measurement period^^^. In approximately 60 studies in which total energy expenditure determined using doubly-labelled water and calorimetry were compared, a mean standard deviation of 8 % was observed thus yielding 95% confidence intervals between the two methods of 84 to 116% (Marinos Elia, personal communication, 1998). The high cost of the doubly-labelled isotope is also a disadvantage although the method is generally considered acceptable to subjects. A limited number of studies have been undertaken using the doubly-labelled water or bicarbonate-urea methods to estimate total energy expenditure in patient populations^^^ "^^^ '^^, but this technique offers potential for future clinical studies. Chapter 4 Factors contributing to malnutrition 131

Non-calorimetric methods for assessing total energy expenditure have also been developed, for example the Socially Acceptable Monitoring Instrument, which utilizes the relationship between energy expenditure and heart rate^°. Activity diaries can also be used to estimate total energy expenditure by adding the energy costs of individual activities calculated from either standard tables or previously measured data on energy expenditure^. Physical activity has been assessed using a satellite-based global positioning system which can detect the distance and speed travelled by an individual carrying a tracking device"^^^ and although an initial study looks interesting it is doubtful that the technique could be applied to a clinical situation. The errors associated with non-calorimetric methods of assessing energy expenditure, particularly in obese individuals, those at rest and smokers, are unacceptable and limit their

4.5.3 Total energy expenditure in cirrhosis The measurement of total energy expenditure in patients with cirrhosis and its comparison with suitable reference standards would give some indication of whether energy requirements are increased in this patient population.

Only three studies have been published, to date, on total energy expenditure in patients with cirrhosis^^’^^^’^^^. The results from one, published in abstract form only^\ estimated total energy expenditure in three patients with alcoholic cirrhosis and ascites using doubly-labelled water and activity diaries. The mean total energy expenditure determined using doubly-labelled water was 1388 kcal/day while estimates derived using the diaries ranged from 0.9% less to 2.1% more.

Nielsen and colleagues^^^, from the same department, calculated total energy expenditure in 37 stable, ambulatory patients with alcoholic cirrhosis over one 24- hour period using a modification of the activity diary method. The patients kept records of their activity every 15 minutes throughout a 24-hour period and their basal metabolic rate was calculated from standard prediction equations^^^. Activity factors for sleeping (0.9), lying awake in bed ( 1 .2 ), sitting (1.3), walking at 3 km/h (2.5) and training on a bicycle at 15 km/h (7.0)^"^^ were then used to calculate Chapter 4 Factors contributing to malnutrition 132 scores for the various activities performed during the observation period. Total energy expenditure was calculated from the cumulative 24-hour score.

The authors verified their method using the preliminary work published by Martinsen^\ described above, and by comparing oxygen uptake during exercise, which they found almost identical in patients with cirrhosis and healthy volunteers.

The median (range) total energy expenditure was 98% (57-294%) of calculated energy intake determined by dietary recall. Although the method used to determine total energy expenditure in this study had been verified in only three patients, no information was given about its accuracy in the 37 patients studied nor are comparative data available to assess whether their total energy expenditure was elevated in comparison to healthy individuals.

Verboeket-van de Venue and colleagues^^^ measured total energy expenditure by indirect calorimetry over a 44 hour period in four men and four women with cirrhosis and ten healthy men using a respiration chamber. There were no significant differences in the 24-hour total energy expenditure or in the 24-hour sleeping metabolic rate between the two groups. However, total energy expenditure was significantly lower in the patients than the volunteers when expressed as a multiple of the sleeping metabolic rate. This implies that the energy expended during waking hours was lower in the patients; this may reflect decreased activity- or dietary-related energy expenditure.

No firm conclusions can be drawn from these studies on total energy expenditure in patients with cirrhosis. Thus, its contribution to nutritional status is unknown at present.

4.5A Protein utilization Accurate balance studies have been undertaken to quantify the protein intake required to achieve nitrogen balance in a number of studies^^^’^^^’^^^. Swart and colleagues^^^ observed the effects of three isocaloric diets providing 40, 60 or 80 g Chapter 4 Factors contributing to malnutrition 133 protein on the nitrogen balance of eight stable patients with cirrhosis over a total period of 30 days. All patients exhibited negative nitrogen balance while on the 40 g protein diet; four remained in negative balance on 60 g protein per day, while all achieved a positive nitrogen balance on the 80 g protein diet. The authors concluded that a mean minimum daily intake of 0.75 g protein / kg was required to reach nitrogen balance, representing a safe protein intake of 1 . 2 g/kg/day.

Nielsen and colleagues^^^ undertook a linear regression analysis to determine the protein requirements of 37 patients with cirrhosis who took part in nitrogen balance studies while eating spontaneously. They concluded that the mean protein intake required for nitrogen balance was 0.83 g/kg/day and that an intake of 1.2 g/kg/day was required to ensure that all patients achieved nitrogen balance.

In a second study by the same group of workers^^^, nitrogen balance studies were undertaken in 15 malnourished patients with cirrhosis over a period of 38 days during which their protein intake was increased from 1.0 to 1.8 g/kg/day. The mean protein intake required for nitrogen balance was again calculated by linear regression and determined as 0.81 g/kg/day, with a recommended protein intake of 1.3 g/kg/day. These findings are supported by the results from a third study from the same group^^^ in which a long term re-feeding programme was undertaken. The authors concluded from all three studies that patients with cirrhosis have an increased requirement for protein rather than a decreased utilization.

Thus in patients with cirrhosis, protein requirements of a minimum of 0.8 g/kg/day is required to maintain nitrogen balance. Intakes of 1.0-1.3 g/kg/day are recommended in patients with well-compensated disease but of 1.5 g/kg/day for those who are decompensated or malnourished^^^’"^^^. These figures are substantially higher than the 0 . 6 g/kg/day recommended for healthy people^^^.

4.6 Summary The malnutrition observed in patients with cirrhosis is multifactorial, with limited Chapter 4 Factors contributing to malnutrition 134 evidence suggesting that nutrient intake is impaired and nutrient losses increased in some patients, particularly those with decompensated disease. While there is good evidence that protein requirements are increased in this patient population, there is limited and inconclusive evidence of increased requirements for energy. An examination of REE in patients with cirrhosis may provide useful information.

4.7 RESTING ENERGY EXPENDITURE IN PATIENTS WITH CIRRHOSIS 4.7.1 Measurement of REE REE can be estimated by indirect calorimetry, that is, by measuring oxygen consumption and carbon dioxide production and making a number of assumptions about the oxidation of carbohydrate, fat and protein and production of heat from chemical energy^^^. Measurements are most easily undertaken using a purpose- made calorimeter, or metabolic cart. This comprises a system for collecting expired air, either as a flow-through canopy, mouthpiece or mask, in-built analysers for determining volumes and concentrations of expired gases and an integral computer for data calculation. Flow-though models draw room air through the canopy at a measured rate enabling the subject to breathe in a relatively unrestricted way. Oxygen and carbon dioxide concentrations are measured in diluted expired air and room air and their respective uptake and production calculated; from these, energy expenditure and fuel utilization are computed. Calorimeters with mouthpiece or mask systems for collecting expired gas are available but may be less accurate than the canopy versions^^^. The 24-hour REE can be measured with an accuracy to within 5% using a flow-through indirect calorimeter^^.

4.7.2 REE in cirrhosis There are a large number of studies in which REE has been measured in patients with chronic liver disease, using indirect calorimetry, either as a primary objective or as part of other metabolic 291,294,295,340,354,376,378,379,380,392,408,427,457,463,478,487,488,489,526,538,546,551,563

The quality of these studies and the conclusions that can be drawn from them vary considerably. Of the 40 studies reported, 36 have been published as full papers Chapter 4 Factors contributing to malnutrition 135

and four as abstracts^^^’^^"^’^^^’^^^. Eight studies did not include comparative data from control while in a further eight, data from patients with cirrhosis were presented together with, and indistinguishable from, data from patients with other liver disorders including alcoholic hepatitis, hepatocellular carcinoma and "hepatic

The methodology in six studies was either inadequately described, for example omitting data on the fasting status of the subjects, or inappropriate, for example, using an assumed respiratory quotient (RQ) of 0.75 to calculate energy expenditure or simultaneously infusing intravenous glucose while measurements were being made^^^'^^^'"^^^' ^63,487,563

However, the remaining 22 studies provided adequately controlled data and sufficient detail to allow evaluation (Tables 4.2 and 4.3). In these studies, REE has been expressed in different units of body composition; there is no consensus as to which units provide the most useful comparison of energy expenditure or how body composition should be determined. The studies are, therefore, presented according to the units in which the REE is expressed but discussed critically as a whole.

4.7.2.1 REE in absolute values. Twelve studies have been undertaken in which REE was measured in patients with cirrhosis and healthy volunteers and the results expressed in absolute terms; no significant differences were found in ten of these Levine and colleagues^^^ observed a significantly higher REE (+13%) in 17 patients with cirrhosis compared to 17 healthy volunteers matched for age, sex and weight. Müller et af^^ found a significantly lower REE in ten patients with compensated cirrhosis (-14%) compared with ten unmatched healthy volunteers but no difference in REE in patients with moderately decompensated disease. Overall, therefore, there are probably no significant changes in REE in cirrhotic patients when data are expressed in absolute terms.

4.7.22 REE/kg body weight. No significant differences were observed in Chapter 4 Factors contributing to malnutrition 136

REE in patients with cirrhosis and healthy volunteers in six of seven studies in which results were expressed in relation to body However, varying proportions of the patients studied had ascites (Table 4.2). Green and colleagues^®^ observed a significantly higher REE/kg in seven cirrhotic patients with primary biliary cirrhosis without ascites, (+27%) compared to age- and sex- matched healthy volunteers. Thus in the majority of studies, no increase in REE was observed in patients with cirrhosis when expressed relative to body weight.

4.7,23 REE/m^. Comparisons of REE expressed relative to body surface area have been made in patients with cirrhosis and healthy volunteers in ten In all of the studies, body surface area was determined using the formula of Du Bois and Du Bois^^^:

BSA = X X 71.84 where BSA is body surface area (cm^), W is weight (kg) and H is height (cm). The determination of body surface area by this method has not been validated in patients with ascites.

No significant differences in REE/m^ were found between patients and volunteers in eight of these

REE/m^ was significantly higher in patients with cirrhosis than in healthy volunteers in two studies^^'^^\ Campillo et al^^ observed a significant increase in mean REE/m^ (+13%) in ten patients with cirrhosis, 40% of whom had ascites, compared with seven, unmatched healthy volunteers. W atanabe and colleagues^^^ also observed a significant increase in mean REE/m^ (+17%) in four patients with cirrhosis compared with three healthy volunteers who were on average 9 years younger; the prevalence of ascites was not reported. Thus overall, there are probably no significant changes in REE/m^ in patients with cirrhosis.

4.7.2.4 REE/kg LBM. REE expressed relative to LBM has been evaluated Chapter 4 Factors contributing to malnutrition 137

in patients with cirrhosis in two studies^^^’^"^^. Waluga and colleagues used BIA to estimate LBM in 15 patients with cirrhosis, some of whom had minimal ascites, and 20 age- and sex-matched healthy controls^"^^. No significant difference was found in R EE/kg LBM between the two groups. However, Müller and colleagues^^^ observed a significantly higher REE/kg LBM ( + 54%), estimated using 24 hour

urinary creatinine excretion, in 1 0 patients with cirrhosis, 50% of whom had ascites, compared to 10 unmatched healthy volunteers. As neither BIA or 24-hour urinary creatinine excretion are likely to produce valid assessments of LBM in patient with chronic liver disease^^’"^^ no conclusions can be drawn from these studies.

4.7.2.5 REE/kg FFM. Comparisons of REE/kg FFM in patients with cirrhosis and healthy volunteers have been undertaken in five studies^^'^^"^'^^^'^^^'^^^. No significant differences were found between patients and volunteers in three studies in which FFM was estimated either by dilution of oxygen-18^^^ or by extrapolation from skinfold data®^’^^"^. However, in two studies^^^’^^^ in which data on FFM were obtained using BIA, significantly higher REE/kg FFM values ( + 33% and +36%) were reported in the cirrhotic patients. In both of these studies, the patients and volunteers were age- but not sex-matched. As BIA is unlikely to produce valid estimates of FFM in patients with cirrhosis, no conclusions can be drawn from the results from these two studies. Thus overall, no significant difference has been observed in REE/m^ between patients with cirrhosis and healthy volunteers.

4.7.2.Ô REE/kg BCM. A comparison of REE/kg BCM in patients with cirrhosis and healthy volunteers has been made in only one study^^^ where BCM was estimated using BIA. REE/kg BCM was significantly higher (+33%), in 10 patients with cirrhosis, 50% of whom had ascites, than in 10 imperfectly matched healthy volunteers. The use of BIA to determine BCM in this patient population is probably not valid^®'^, thus limiting the conclusions that can be drawn from this study.

4.7.2.7 REE/g urinary creatinine. REE/g urinary creatinine (UC) has been Chapter 4 Factors contributing to malnutrition 138 measured in patients with cirrhosis in two studies^^’"^^^. Schneeweiss and colleagues"^^^ found significantly higher REE/g UC ( + 65%) in 22 patients with cirrhosis, of whom 27% had ascites, compared with 2 0 unmatched healthy volunteers. Chang et al^^ also reported a significantly higher REE/g UC ( + 45%) in 16 patients with cirrhosis, 59% of whom had ascites, compared to eight unmatched healthy controls. The use of urinary creatinine excretion as a measure of metabolically active tissue is limited because it reflects muscle mass which accounts for only 18% of basal metabolic rate^^. However, in patients with cirrhosis it may be even less useful because of the confounding effects of hepatic impairment on creatinine excretion"^^ which limit the conclusions that can be drawn from these studies.

Heymsfield and colleagues^^ reviewed the studies of REE in cirrhosis undertaken up to 1990 and made a number of methodological recommendations to facilitate better comparisons between patients and healthy volunteers. These were made in relation to (i) subject preparation, (ii) measurement techniques and (iii) the assessment of body composition.

4.7.2.8 Subject preparation. It was recommended that dietary intake should be held constant for several days or even weeks before the assessment of REE, and that the procedure should be undertaken after an overnight fast. Of the 22 studies reviewed, the subjects in all but two, where data were not included^^^’^^^, were reported to have received a reasonably adequate dietary intake prior to assessment.

The patients in all 22 studies had fasted overnight for a minimum of 8 hours before assessment.

Interestingly, however, no specific recommendation was made about alcohol intake. Alcohol abuse per se is associated with an increase in REE but these changes reduce after 5 to 7 days of abstinence^^^’^^"^’^^^. Thus patients with alcoholic liver disease should be abstinent from alcohol for at least 5 days before measurement of REE. Patients with alcoholic cirrhosis were included in 19 of the 22 studies; in 15, patients were reported to have been abstinent from alcohol for a minimum Chapter 4 Factors contributing to malnutrition 139

of 14 days. No such information was reported in the four remaining

4.7.2.9 Measurement techniques. It was recommended that a ventilated canopy should be used in preference to a mouthpiece or mask for collecting expired gas in order to minimise subject discomfort. A mouthpiece was used in four of the 2 2 studies^^’^^’^^’^^ and may have introduced errors in energy expenditure measurements^^^. However, as the same technique was used in both patients and volunteers, errors, if they arise, should be consistent between the two groups. A significant difference was found between REE in patients with cirrhosis and volunteers in only one of the four studies where a mouthpiece was used^^.

4.7.2.10 Assessment of body composition. It was recommended that REE should be expressed against a measure of metabolically active tissue. Dilution techniques, using labelled water and bromide, and potassium counting have been suggested as suitable^^ although their use has not yet been validated in this patient population.

Body composition was assessed in only eight of the 22 However, none of the methods used, which included 24-hour urinary creatinine, BIA, oxygen-18 dilution and anthropometry, have been validated against reference methods in this patient population^^’^®^’^^®’"^^ and thus their use may be questioned.

4.7.3 Summary of REE studies in cirrhosis. In 14 of the 22 controlled studies undertaken to date no differences were observed between REE in patients with cirrhosis and healthy volunteers, regardless of how the results were expressed (Table 4.4). REE was significantly increased in cirrhotic patients in eight one of which^^^ also showed a decrease in REE in some patient subgroups.

These studies can, however, be criticized for poor matching of comparative populations, small numbers of subjects, inadequate descriptions of the patients with Chapter 4 Factors contributing to malnutrition 140

regard to the severity of liver disease, ascites and alcohol intake and to inappropriate assessment of body composition which, combined, limit the overall

conclusions that can be drawn. Indeed, only § 1x^ 8 ,3 5 4 ,376,378,478,546 studies evaluated critically met all of the procedural recommendations (Table 4.5) and, of these, REE was increased in cirrhotic patients in just three studies^^’^^^’"^^^ in which estimates of body composition had been derived using BIA and 24-hour urinary creatinine excretion. Thus there is very little evidence to show that REE is increased in this patient population.

It is, therefore, necessary to undertake a controlled study of REE in patients with cirrhosis in which procedural recommendations are followed and the problems encountered in published studies avoided.

4.8 STUDY 4: ASSESSMENT OF RESTING ENERGY EXPENDITURE IN AN UNSELECTED GROUP OF PATIENTS WITH CIRRHOSIS 4.8.1 Aim (i) To assess REE in patients with cirrhosis and comparable healthy volunteers. (ii) To compare measured REE values with those derived using prediction formulae.

4.8.2 Patients and Volunteers The patient population comprised 56 men and 44 women of mean (range) age 49.3 (26-71) years with cirrhosis of varying aetiology and severity. The aetiology of the liver injury was determined using historical, clinical, laboratory, radiological and histological variables whilst the functional severity of the liver disease was assessed using Pugh’s modification of Child’s grading scheme"^^ (Table 4.6). Fifty nine patients were assessed as in-patients while the remaining 41 were assessed as out­ patients. Fifty-six patients smoked a mean of 15 (1-40) cigarettes per day. All 72 patients with alcohol-related cirrhosis had been abstinent from alcohol for periods ranging from 5 days to more than one year; the majority had been abstinent for Chapter 4 Factors contributing to malnutrition 141 one month or more.

A total of 43 healthy volunteers were screened for inclusion in the reference population. These were recruited from hospital staff, friends and family of staff and individuals who answered newspaper or poster advertisements. All subjects were in good health and none gave a history or had clinical evidence of liver disease. Alcohol intake in the weeks preceding the measurement of REE was not quantified but none of the volunteers gave a history of alcohol abuse or reported exceeding the recommended guidelines for alcohol intake^^^. One individual was subsequently excluded from the study because he had abnormal liver function test results. A second individual failed to attend for the assessment of REE. The resulting population of 41 healthy volunteers comprised 20 men and 21 women of mean (range) age 48.9 (22-79) years. Nine of the healthy volunteers smoked, eight of them smoking a mean of 10 cigarettes (< 1-30) per day while the ninth smoked approximately 4 g of pipe tobacco daily.

4.8.3 Assessment of nutritional status All patients and volunteers underwent assessment of their nutritional status within 7 days of the metabolic assessment. Nutritional status was assessed globally in all individuals using a reproducible method which included an evaluation of their clinical history, nutrient intake, physical status and anthropometric measurements. Patients were categorized as well nourished, moderately malnourished (or suspected of being so) or severely malnourished according to the algorithm in Appendix.

Recent nutrient intake was assessed using a standard diet history technique^.

Height was measured to the nearest centimetre with patients standing in socks or bare feet, using a wall mounted or free-standing measure (Seca, Hamburg, Germany). Patients were weighed to the nearest 0.1 kg, wearing night-wear or light indoor clothing and without shoes, using a weighing scale (Seca, Hamburg, Germany) or a seat balance scale (Marsden W/M Group, London, UK). Chapter 4 Factors contributing to malnutrition 142

In patients with fluid retention, an estimated dry weight was determined from body weight by subtracting a weight for fluid based on clinical assessment and previously documented weights using published guidelines'^.

The estimated dry weight of patients with fluid retention and the actual body weight of patients with no clinical evidence of fluid retention and of healthy volunteers, collectively referred to as the corrected weight, was then used to calculate BML

MAC and skinfold thicknesses at the triceps, biceps, subscapular and suprailiac sites were measured using standardized techniques^'^^’^'^^ and MAMC was calculated^^^. MAMC and TSF were compared with published standards^^, and expressed as a percentage of the 50th percentile for the appropriate sex and age category.

Percentage fat was calculated from the sum of four skinfold measurements using the method of Durnin and Womersley^"^^ and used to calculate an estimated FFM from both the actual body weight and the corrected weight, subsequently referred to as FFM and FFM'"""'"''^.

Thus, data on weight and FFM were collected or calculated in all subjects, while in patients with fluid retention an estimated dry weight was also noted and the FFM‘^°"'®‘^‘^‘* calculated. In order to facilitate comparisons between the patient group as a whole, the term corrected weight is used and refers to the actual weight of patients without fluid retention and the estimated dry weight of those with. Similarly, FFM^°""^^*^^ refers to the FFM of patients without fluid retention and FFM calculated from estimated dry weight of those with.

4.8.4 Preparation for REE measurement A 24-hour urine sample was collected immediately prior to the assessment of REE for determination of urinary urea which was measured using a standard procedure^^^'^^^. Nursing staff supervised the collection of urine in hospitalized Chapter 4 Factors contributing to malnutrition 143 patients. Patients were asked to abstain from smoking, eating or drinking all beverages except water for 10 hours before their assessment. Prescribed medication was given on the morning of the assessment but these and all drugs taken within the preceding week were recorded.

All measurements of REE were undertaken between 08.00 and 11.00 hours. All subjects attended lightly clothed or wearing nightwear and had been asked to evacuate their bladder and bowel beforehand. Body weight was re-checked using a digital weighing scale (Seca, Hamburg, Germany) and blood pressure and pulse measured using an automated digital blood pressure meter (Takeda Medical, Tokyo, Japan). The length of time since the last intake of nutrients was noted along with details of what was consumed. Subjects lay supine on a padded couch in a quiet room in thermal equilibrium between 22 and 26° C, with their head and shoulders at an angle of approximately 30° to the horizontal; they rested for approximately 15 to 30 minutes or until their pulse, if raised initially, reached steady state.

4.8.5 Indirect calorimetry REE was assessed by indirect calorimetry using a flow-through, canopy calorimeter (2900 Metabolic Measurement Cart, SensorMedics, Bilthoven, Netherlands). The calorimeter was calibrated both volumetrically and analytically at the beginning of each assessment using a 3 litre syringe and two standard gas mixtures; the concentrations of the gases, confirmed by chemical analysis, were 26% oxygen, 0 % carbon dioxide and 16% oxygen, 4% carbon dioxide respectively, both balanced with nitrogen.

After the rest period, the calorimeter canopy was placed in position over the subject’s head and neck and a cellophane collar used to prevent excessive gas leakage. The patients were asked to relax, to stay awake but to remain still and avoid talking. Air was drawn through the system at a variable but measured rate between 18 and 50 1/minute and a sample extracted from the exhaust stream each minute for the duration of the assessment. The air sample was passed through a Chapter 4 Factors contributing to malnutrition 144 sample-conditioning system to equalize sample gas and calibration gas with ambient conditions and was then analyzed for oxygen using a zirconium analyser and for carbon dioxide using an infrared absorption analyser. Measurements of the oxygen and carbon dioxide concentrations in the expired air and the gas flow rate were recorded every minute; gas concentrations of the inspired air were automatically checked every 10 minutes. Data were collected until energy expenditure readings stabilized for at least 15 minutes; the minimum data collection period was 30 minutes.

The expired gas concentrations and gas flow rates were automatically entered into a programme in the integral computer together with estimated values of 24-hour urinary nitrogen excretion calculated from the urinary urea excretion:

1 mmol urea = 28 mg nitrogen TUN = (UUNx 1.17) + 0.7 if UUN ^15g/24 hours TUN = (UUN -k 4.0) if UUN >15g/24 hours where TUN is total urinary nitrogen and UUN is urinary urea nitrogen 412

Energy expenditure and non-protein RQ were then calculated for each 1 minute period using standard formulae^^’^^^:

Energy expenditure (kcal) = (1.0548 - [0.0504 x FEO2 J) x VE (1 + [0.082 X PJ)

where FEO 2 is concentration of O 2 in expired air, VE is ventilatory rate and P is fraction of total dietary energy from protein;

RQ (non-protein) = (VCO 2 - [4.8 x TUN]) \(V02 - [5.9X TUN])

where VCO 2 is carbon dioxide production and VO 2 is oxygen uptake. The data Chapter 4 Factors contributing to malnutrition 145 collected for each one minute period from 5 to 30 minutes or until the end of data collection were averaged for each subject and the 24-hour REE extrapolated.

The reported accuracy of the calorimeter is ± 0.1% for analysis of oxygen concentrations (0-99%), ± 0.2% for carbon dioxide concentrations (0-10%) and ± 2% for flow rate (6-600 l.m'^)^^. The accuracy of determining the RQ has not been reported but its validation against combustion of methanol is described as having "little error"^^. The reproducibility of the measurements was evaluated in three patients and three volunteers over one 5-day period.

4.8.6 Predicted REE Predicted REE was calculated for all subjects using the formulae proposed by Harris and Benedict^^^ and Schofield"^^^:

Harris-Benedict Men: FREE = 66.47 + 13.75W + 5.00H - 6.76A Women: FREE = 655.10 + 9.56W + 1.85H - 4.68A

Schofield Men (18-30years): BMR = 15.0W + 690 Men (30-60 years): BMR = 11.4W + 870 Men (60+ years): BMR = 11.7W + 585 Women (18-30 years): BMR = 14.8W + 485 Women (30-60 years): BMR = 8.1W + 842 Women (60+ years): BMR - 9.0W + 656 where FREE is predicted resting energy expenditure (kcal), BMR is basal metabolic rate (kcal), W is actual weight (kg), H is height (cm) and A is age (years). In addition, predicted values for REE were calculated from both the Harris-Benedict and Schofield formulae substituting the estimated dry weight for actual weight in the patients with fluid retention. Chapter 4 Factors contributing to malnutrition 146

4.8.7 Statistical analysis Statistical analysis was performed using the Minitab software package (1994, State College, PA, USA). Repeatability coefficients'^^ were calculated for REE and RQ for patients and volunteers and expressed as a percentage of the mean of the repeated values. Comparisons of demographic and anthropometric data and energy expenditure were undertaken between patients and volunteers, (inter-population differences), using unpaired Student’s t-tests and Chi-squared analyses. These techniques were also used to make comparisons between subgroups of patients (intra-population differences) without clinically detectable fluid retention and patients with non-alcoholic cirrhosis and the volunteers. Intra-population differences in demographic, anthropometric and REE variables were analysed in subgroups of all subjects divided on the basis of sex and smoking behaviour and in subgroups of patients divided on the basis of severity of liver disease and nutritional status using unpaired Student’s t-tests. Chi-squared analyses and oneway analysis of variance. The General Linear Model facility on the Minitab package was also used to compare inter- and intra-population differences in REE adjusted for the confounding effects of variables including age, sex, smoking. The relationship between REE and age was examined using Pearson correlations. The relationships between measured and predicted values of REE were examined using unpaired Student’s t-tests and modified Bland and Altman plots'*^.

4.8.8 Results 4.8.8.1 Interpopulation comparisons Demography and anthropometry The patients with cirrhosis and the healthy volunteers were matched for age and sex (Table 4.7). There were no significant differences between the patients and volunteers in mean values for height, weight, corrected weight, EMI, FFM or, FFM^°""^^^^^ (Table 4.7). Female patients had a significantly lower corrected body weight than body weight in the healthy volunteers.

No significant differences were found in age, sex, height, weight, BMI or FFM between the 49 patients who had no clinical evidence of fluid retention and the Chapter 4 Factors contributing to malnutrition 147 volunteers (Table 4.8).

Similarly, there were no significant differences between these variables in the subgroup of 51 patients who had clinically detectable fluid retention and the healthy volunteers although the patients had a significantly lower mean corrected weight (Table 4.8).

No differences were observed in measured variables between patients with non­ alcoholic cirrhosis and those with alcoholic cirrhosis or the healthy volunteers.

There was no significant difference in smoking behaviour between the male patients and volunteers but significantly more female patients than volunteers smoked (26 [59%] vs 2 [10%], p< 0.0005). No significant differences were found in demographic and anthropometric variables between the non-smoking patients and volunteers, either in the total group or in the women separately. However, the non-smoking male volunteers were significantly taller than their patient counterparts (mean height ± ISD, 1.75 ± 0.05 vs 1.70 ± 0.06 m, p<0.05). No significant differences were found in demography or anthropometry between patients and volunteers who smoked, either in the total population or in men and women separately. However, far smaller numbers of volunteers smoked than patients (9 [22%] vs 56 [56%], p< 0.001).

Resting energy expenditure The repeatability coefficients for REE and non-protein RQ were 2.1% and 1.3% in patients and 3.0% and 1.9% in volunteers.

No significant difference was observed between the mean REE expressed in absolute terms in the patients with cirrhosis and the healthy volunteers (Table 4.9). However, the mean REE/kg body weight was significantly greater in the patients than the volunteers whether using actual or corrected body weight (Table 4.9). The mean REE/kg FFM was also significantly greater in the patients but only when corrections for fluid retention were applied (Table 4.9). When the mean values of Chapter 4 Factors contributing to malnutrition 148

REE/kg expressed relative to corrected weight and were adjusted for the effects of age, sex and smoking, those in the patients remained significantly higher (R EE/kg corrected weight 24.3 ± 0.4 vs 22.0 ± 0.6, p < 0.005; R EE/kg ^^^con.ctcà 3^ 7 ^ Q 4 29.0 ± 0.7, p < 0.005).

Data for men and women, when considered separately, mirrored changes observed in the total population although the differences between values in patients and volunteers were generally less in men.

No significant differences were observed between REE expressed in absolute terms in either patients with or without clinically detectable fluid retention and healthy volunteers. However, mean values were significantly higher in patients without fluid retention compared to the volunteers when adjusted for body weight and FFM (Table 4.10). Similarly, mean values were significantly higher in patients with fluid retention compared to healthy volunteers when adjusted for fluid, body weight and FFM.

No significant differences were found in REE between patients with non-alcoholic cirrhosis and those with alcoholic cirrhosis, regardless of how the results were expressed.

When the subjects who did not smoke were considered separately (Table 4.11), similar observations were made to those in the whole groups with, generally, higher measurements of REE observed in the patients compared with the volunteers. These differences were significant when expressed relative to body weight corrected for fluid retention.

Comparison of REE between patients and volunteers who smoked revealed no significant differences regardless of how the data were expressed. In addition, there was no general trend for the patients who smoked to have higher values than the corresponding volunteers (Table 4.12). Chapter 4 Factors contributing to malnutrition 149

Respiratory quotient Significantly lower RQ values were observed in the total patient population, the patients without clinical evidence of fluid retention and the patient who did not smoke compared with the volunteers (Tables 4.9-4.11). These differences remained significant after mean values were adjusted for the effects of age, sex and smoking (0.79 ± 0.01 vs 0.81 ± 0.01, p<0.05). No significant differences were observed between the RQ in the patients with non-alcoholic cirrhosis and those with alcoholic cirrhosis or healthy volunteers.

4.8.8.2 Intrapopulation differences Fluid retention REE was not significantly different between patients with clinical evidence of fluid retention and those without when expressed in absolute terms but was significantly lower in patients with fluid retention when expressed relative to body weight (Table 4.10). However, when adjustments were made for fluid weight, no significant differences were found between the two subgroups of patients.

Age In the total patient population, advancing age was significantly correlated with a decrease in REE, whether this was expressed in absolute terms (r =-0.227; p<0.05), or in relation to body weight (r = -0.423; p< 0.0001), corrected weight (r = - 0.338; p< 0.005), FFM (r = -0.324; p< 0.005), or FFM'""'"''"'' (r = -0.238; p<0.05). Similar associations were found when male and female patients were considered separately. However, there was no significant correlation in the main volunteer population between age and REE whether expressed in absolute terms (r = -0.131), or relative to body weight (r=-0.181), or FFM (r = -0.103) nor were any significant correlations observed when male and female volunteers were considered separately.

Sex Significant differences were observed between men and women in both the patients and the volunteers in relation to anthropometric variables, with the men in both populations being significantly taller, heavier and having a greater FFM than the Chapter 4 Factors contributing to malnutrition 150 women (Tables 4.13 and 4.14). These differences were also observed when both populations were divided by smoking status, except for weight in the smoking volunteers (Tables 4.13 and 4.14). Male patients and volunteers also exhibited a significantly higher absolute REE than women. However, when energy expenditure was expressed relative to FFM, it was significantly lower in male patients than in women (Tables 4.13), and remained so when mean values were adjusted for the effects of age, smoking, severity of liver disease and nutritional status (28.6 ± 0.6 vs 31.8 ± 0.6, p< 0.001). A significantly higher RQ was observed in the male patients, including the subgroup of non-smokers, but not in the volunteers (Tables 4.13 and 4.14). Mean RQ values remained significantly higher in men when adjusted for the effects of age, smoking, severity of liver disease and nutritional status (0.80 ± 0.01 vs 0.78 ± 0.01, p<0.05).

Smoking status Overall, no significant differences or trends were observed between subjects who smoked and those who did not in either the patients or volunteers, either as whole populations, or as subgroups of men or women, with relation to demographic or anthropometric data, energy expenditure or RQ. The effects of smoking on REE and RQ remained insignificant in the patients after mean values were adjusted for the effects of age, sex, severity of liver disease and nutritional status.

Severity of liver disease There were no significant differences in age, sex, smoking status, height, weight, BMI, FFM or FFM^°^^^^^^ between the patients with compensated liver disease and those with decompensated disease (Table 4.15). However, the patients with compensated liver disease had a significantly higher mean corrected weight than those with decompensated disease in both the whole patient population and the female patients but not in the men (Table 4.15). The effect of the severity of liver disease on REE varied depending on how the latter was expressed. In the total patient population and male and female patients separately, there were no significant differences in REE between patients with compensated and decompensated liver disease when expressed in absolute terms or relative to body Chapter 4 Factors contributing to malnutrition 151 weight or However, when expressed relative to corrected body weight, the R E E was lower in the compensated patients but higher when expressed relative to FFM in comparison to those with decompensated liver disease. These differences were significant in the total patient population (Table 4.16), and remained so when mean values were adjusted for the effects of age, sex, smoking and nutritional status. The female patients with compensated liver disease had a significantly lower RQ than those with decompensated liver disease but this was not observed in the total patient population or the male patients (Table 4.16).

Nutritional status Fifty-three patients were categorized as being well nourished, 33 as moderately malnourished, or suspected of being so, and 14 as severely malnourished. In the total patient population, as expected, significant differences were found in weight, corrected weight, FFM, and BMI across the nutritional categories with higher values observed in the well nourished patients for all variables (Table 4.17).

A significantly higher mean REE was observed in the well nourished patients compared to less well nourished patients when data were expressed in absolute terms, and remained so when mean values were adjusted for the effects of age, sex, smoking and severity of liver disease. However, the mean REE was significantly lower in the well nourished patients when expressed relative to body weight, corrected weight or FFM^"""^^^^^, and also remained so when adjusted for covariate effects. The differences in REE/kg FFM and RQ were not significant across the nutritional categories (Table 4.18).

4.8.S.3 Comparison with prediction formulae Harris-Benedict Formulae^^^ In the total patient population, predicted values underestimated measured daily REE by a mean of 121 kcal (7%) with 95% limits of agreement between 548 kcal (33%) less to 308 (19%) more (Table 4.19 Figure 4.1). The predicted REE calculated using corrected rather than actual body weight was significantly lower than m easured values in the patients (1476 ± 245 vs 1646 ± 340 kcal, p< 0.0001) Chapter 4 Factors contributing to malnutrition 152 and was lower than, but not significantly different from values calculated using actual weight.

There were no significant differences between the predicted REE and measured values in the healthy volunteers or any of the subgroups (Table 4.20). In volunteers, predicted values underestimated measured values by a mean of 53 kcal (3%) with 95% limits of agreement between 395 kcal (25%) less to 289 (18%) more than measured values (Figure 4.2),

Schofield Formulae‘^^^ The mean values for predicted REE calculated from the formulae of Schofield were closer to the measured values than those calculated from the Harris-Benedict formulae but were also lower than measured values in the total patient population and all the subgroups. The differences were significant in the female patients and female patients who smoked (Table 4.19). In the total patient population, predicted values were a mean of 84 kcal (5%) lower than measured values with 95% limits of agreement between 548 kcal (33%) less and 380 (23%) more (Figure 4.3). When actual body weight was replaced by corrected weight in the Schofield formulae, the difference between predicted REE and measured values became significant in the total patient population (1519 ± 250 vs 1646 ± 340 kcal, p < 0.005).

There were no significant differences between the predicted energy expenditure and measured values in the healthy volunteers or any of the subgroups (Table 4.20). Predicted values were a mean of 45 kcal (3%) lower than measured values with estimates varying from 409 kcal (26%) less to 357 (23%) more (Figure 4.4).

4.8.8A Summary of results i REE was increased in patients with cirrhosis when corrections for fluid retention were applied and when the data were expressed relative to body weight and FFM. ii Generally, these differences were also observed in subgroups determined by gender, aetiology of liver disease and smoking behaviour. Chapter 4 Factors contributing to malnutrition 153 iii Fluid retention did not appear to influence REE when corrections were made for the additional weight of ascites and / or oedema. iv Increasing age was associated with decreasing REE in patients but not in volunteers. V REE was higher in women than men when expressed relative to FFM. vi REE may be increased in decompensated compared with compensated patients. vii REE is increased in malnourished patients, when expressed, after correction for fluid retention, relative to body weight and FFM. viii Patients with cirrhosis had significantly lower RQ than healthy volunteers. ix The prediction formulae proposed by Harris-Benedict^^^ and Schofield"^^^, using actual or corrected weight, underestimated REE in patients with cirrhosis.

4.8.9 Discussion The results from this study show that REE is increased in patients with cirrhosis, although dependent on the way data are expressed, and that RQ is decreased in this patient population. This confirms previous findings of both increased

and decreased 98, 291 ,376,408,478,526 includes a larger number of patients than reported previously and has been undertaken in accordance with procedural recommendations. REE has not been expressed relative to corrected weight or FFM'^°"'^^^^^ in any published studies.

Increases in REE in cirrhosis are attributable to changes which occur in carbohydrate and fat metabolism in this patient population. Such patients have depleted liver glycogen stores^^^ which lead to reduced splanchnic glucose production‘s®^; in order to maintain circulating glucose concentrations hepatic gluconeogenesis is increased^^'^^^, a process which requires energy^^^ and hence may lead to increases in REE. These observations are supported by the presence of increased circulating concentrations of free fatty acids^®^, indicating increased lipolysis, which also explains the observed decrease in RQ. Chapter 4 Factors contributing to malnutrition 154

Procedural recommendations In the present study, the procedural recommendations of Heymsfield and colleagues^^^ have been followed as far as practical in a clinical setting. All subjects consumed an adequate diet in the days preceding assessment, verified by diet history, and all fasted for a minimum of 10 hours before measurements were made. In addition, all patients with alcoholic liver disease had abstained from alcohol for a minimum of 5 days prior to investigation.

REE was measured using a standard flow-through calorimeter with a canopy to collect expired gases. Apart from a small number of patients who had undergone previous REE assessments, all the volunteers and a large majority of the patients were unfamiliar with the procedure thus minimising any potential differences between the two groups.

FFM was calculated from skinfold anthropometry in order to provide a measure of metabolically active tissue against which REE could be referenced. While the use of anthropometry has been validated in healthy individuals^"^^, it has not in patients with liver disease. In spite of this limitation, anthropometry was selected as the most useful and practical method available for assessing a large number of subjects in a clinical setting. A fuller discussion of body composition is presented in the subsequent chapter.

Fluid retention Although it is not possible to quantify accurately the fluid retention of an individual patient, the method used to do so in this study is reproducible between two observers (Table 1.7). Account needs to be taken of the contribution made by ascites and oedema to total body weight as, at present, there is no evidence to confirm that ascitic fluid itself is metabolically active. This does not imply that the presence of ascites is unaccompanied by some additional costs of energy expenditure, and indeed, although there are no published reports, it is quite likely that total energy expenditure increases simply because of the associated weight gain. Chapter 4 Factors contributing to malnutrition 155

In the present study, the potentially confounding effects of fluid retention were minimized by making comparisons between the patients who had no clinical evidence of fluid retention and the healthy volunteers who were demographically and anthropometrically similar. REE expressed relative to body weight and FFM were still found to be significantly higher in the patients without fluid retention than the healthy volunteers. Of the eight studies where significantly increased values of REE were reported in patients with cirrhosis, only one was undertaken in patients without ascites^^®; in four others, between 21% and 75% of patients had ascites^^'^^'^^'"^^ while no information about fluid retention was given in the remaining three^^^'^^^'^^\

Dolz and colleagues^^^ measured REE in 10 cirrhotic patients using indirect calorimetry before and after abdominal paracentesis. Following the procedure, the mean weight loss was 17 kg while REE, measured a mean of 11 days later, was significantly lower. The authors comment that the changes in REE in six patients were, however, within the limits of repeatability of the method. It is also possible that patients were more relaxed during the second assessment because of increased familiarity with the calorimetry mouthpiece^^^ and that changes may have occurred in other variables. In spite of these methodological problems, the results show that the presence of ascites is associated with increased REE.

A reduction in energy expenditure with advancing age has been previously reported in healthy populations^^^ but this decrease is likely to reflect a decline in physical activity and reduction of However, in the present study, a significant decline in REE was observed in patients with cirrhosis with advancing age but not in the healthy volunteers regardless of how REE was expressed. The reason for this difference between the groups has not been determined.

Sex In a number of studies of healthy individuals, a significantly higher REE has been reported in men than women whether expressed in absolute terms or relative to Chapter 4 Factors contributing to malnutrition 156

metabolically active tissue^^’^^’^^^. However, this finding is not universaP^^’"^^^’^^^. In the present study, mean REE was higher in men than women, in both study populations, when expressed in absolute terms, but no significant difference was observed when data were expressed relative to body weight. Significantly lower mean REE values were observed in male compared to female patients when data were expressed relative to FFM; a similar difference in the volunteers was not significant. This observation may be explained by the relatively small contribution made to REE by skeletal muscle (18%) compared to other organs comprising FFM, including the liver, brain and heart^^. Proportionally lower musculature in women means their FFM includes relatively more of the "metabolically active" organs.

Variations in energy expenditure have been observed in women at different times of the menstrual cycle^^’^^"^ but the majority of women in this study were either anovulatory or post-menopausal so it is unlikely these variations are responsible for the observations made^^^.

Smoking Previous studies investigating REE in patients with cirrhosis have not taken account of the smoking status. Interest in the relationship between smoking and energy expenditure was based initially on the effect of smoking on body weight"^^^ and the weight gain frequently associated with the cessation of regular smoking^^^. Although there is evidence that smoking is associated with a short-term increase in metabolic rate"^^’^% there is little evidence that this increase continues for more than 30 minutes after smoking stops^^^’^"^^.

In the present study, all subjects were asked to abstain from smoking from the midnight before they underwent REE assessment and, although this was not verified beyond asking the subjects when they had last smoked, all were accompanied by the investigator for at least 30 minutes before measurements were made and none smoked during this time or during the subsequent measurement of REE. Chapter 4 Factors contributing to malnutrition 157

The results from the present study confirm that, overall, there were no significant differences or trends in REE between the subjects who smoked or those who did not in either patients or volunteers. It would seem reasonable, therefore, to recommend for future studies, that providing subjects abstain from smoking for a verifiable period of 30 minutes before assessment, it is unnecessary to match patients and volunteers for smoking status.

Seventy of liver disease In the present study, the differences in REE observed between the patients with compensated and decompensated liver disease are unclear as the decompensated patients have significantly higher mean REE values when expressed relative to corrected weight but significantly lower values relative to uncorrected FFM. This probably reflects the higher prevalence of fluid retention in the decompensated patients. In patients with fluid retention, FFM, before correction for fluid weight, is obviously increased by the additional fluid. Therefore, values of REE expressed relative to FFM are likely to be underestimated and the significance of the comparison to be misleading. However, the mean REE/kg FFM^°"'^'^^^^ was not significantly raised in the patients with cirrhosis.

Müller and colleagues^^^ reported higher values for R E E/kg FFM, estimated by BIA, in patients with Child’s grade B cirrhosis compared to those with Child’s grade A. However, these differences were not significant. Further studies with optimum assessments of body composition are required to clarify the relationship between REE and severity of liver disease.

Nutritional status In the present study, REE was significantly higher in patients whose nutritional status was impaired whether data were expressed relative to body weight, corrected weight or FFM. However, although nutritional status and disease severity are closely associated^^’^^^, the effect of nutritional status on REE was independent of disease severity. The observation of increased REE with impaired nutritional status may result, at least in part, from loss of skeletal muscle and body fat, which Chapter 4 Factors contributing to malnutrition 158 make a relatively small contribution to REE compared with vital organs^^. However, if decompensated liver disease is associated with increased REE, which is suggested but not confirmed in the present study, it is tempting to hypothesise that nutritional status deteriorates because of an increase in REE secondary to a decline in liver function. A long-term, prospective study of REE in patients with cirrhosis, accompanied by appropriate body composition measurements is needed to explore this theory.

Prediction formulae The significant underestimation of measured REE obtained using the Harris- Benedict^®^ and Schofield"^^^ prediction formulae in patients with cirrhosis indicates that their use should be questioned in this patient population. The individual variation observed between measured and predicted values in the patients and volunteers, exceeded ± 10% which has been suggested as including "almost all normal subjects"^^. However, neither the Harris-Benedict nor the Schofield equations were developed to predict REE in sick individuals although both have been widely used in clinical Even in healthy populations, where these equations have been used to predict REE, the results have both under-^^^’"^^ and overestimated^values determined by indirect calorimetry by mean values of up to 15% with individual values exceeding 25%^^^.

Comparisons of predicted and measured values of REE also vary considerably in diseased and injured populations leading to a number of recommendations that the use of these formulae is inappropriate in patients with cancer^^^’^^^, human immunodeficiency virus^^^, respiratory disease^^^’"^^^, anorexia nervosa^^^ and pancreatitis^^^ and in those requiring ventilation on an intensive care unit^^’^^^’^^°.

Comparisons of REE predicted by the Harris Benedict equations^^^ and measured values determined by indirect calorimetry have been undertaken in published studies of patients with chronic liver There was general agreement that mean predicted values underestimated mean measured values by up to 13%, with individual differences in a third of patients exceeding Chapter 4 Factors contributing to malnutrition 159

In only one study, involving children with cirrhosis or chronic hepatitis^^^, no significant differences were observed between measured REE and values predicted using the Schofield formulae"^^^.

It can be concluded from these data, and those from the present study, that the Harris-Benedict formulae are inaccurate for estimating REE in patients with cirrhosis. There is less evidence on which the validity of the Schofield formulae can be assessed in this patient population. In a clinical situation where an estimate of REE is required in patients with cirrhosis but no facilities available to undertake measurements, the values predicted from the Schofield formulae should be increased by 2% for men and 6% for women. However, even if these adjustments are undertaken, it must be recognised that, as in healthy individuals, the actual REE of a substantial number of patients will differ from the adjusted predicted values. In addition, no guidelines can be given, at present, about the additional "stress factors" that should be added to REE to provide an estimate of energy requirements in this patient population^^®. To facilitate this, an investigation of total energy expenditure as well as REE is required.

Study conclusions The results from the present study, with its limitation of assessing body composition using anthropometry, show that REE is increased in patients with cirrhosis. The mean increase, of 16% above control data in some patient subgroups, is substantial and, if it contributes to an increased total energy expenditure, is likely to have a detrimental effect on nutritional status. This is supported by the observation that patients who are more malnourished have a greater REE than those who are better nourished. The discrepancies observed between predicted and measured values of REE in patients with cirrhosis preclude the use of the Harris-Benedict equations in this patient population. Modifications to the Schofield formulae may, however, prove useful. Further investigations of REE, accompanied by optimum body composition analysis, would help clarify the position and may facilitate the proposal of more suitable prediction equations. Whether hypermetabolism contributes to malnutrition in patients with cirrhosis can only be addressed when total energy Chapter 4 Factors contributing to malnutrition 160 expenditure is assessed in this patient population and appropriate comparisons made. Table 4.1 Studies of nutrient intake in patients with chronic liver disease. PTO for legend.

First author Population Cirrhosis Method Reference for Energy intake’ Energy intake*’ Comment' Year (%) comparison absolute / relative (% reference) (kcal) (kcal/kg) absolute/relative

Leevy^*^ 1965 172 alcoholics 40 diet history NA -- Intakes varied greatly

Neville^^ 1968 34 alcoholics 9 diet history non-liver patients - 72 Intakes varied greatly

Patek'^ 1975 304 alcoholics 64 diet history NA 1530 - Alcohol = 55% kcal

Huif^ 1981 53 alcoholics NA diet history NA 2029 - Alcohol = 35% kcal 58 alcoholics NA observation 2661

Morgan^^ 1981 55 ALD 33 ) ) 23.3 77* Alcohol = 53% kcal 20 PEC 70 ) diet history ) healthy controls 27.6 93 17 CAH 53 ) ) 31.4 104

Simko*’* 1982 62 ALD NA diet history non-liver patients 28.1 101 Alcohol = 35% kcal

Bunout^ 1983 10 AH + ALC NA diet history NA 17.3'' - Alcohol = 14% kcal

Mills^‘1 1983 23 ALD 38 diet history PRV”° 2452 99

Mendenhall’^* 1984 102 AH + ALC 100 diet history p R y l ’ PlO 1827 71

Cabré’’ 1990 19 ALC 100 observation hospital diet 1320 60"

Nielsen’” 1993 37 ALC 100 diet history estimated energy 1722 102 expenditure

Noël-Jordan” ® 1994 72 alcoholics 57 diet history healthy controls 2842 148 Alcohol = 35% kcal

Thuluvath” ’ 1994 11 PBC NA NA healthy controls 1497 100

Wicks"" 1995 34 PBC 100 diet history PRY*” - 90

Levine™ 1996 20 CLD 50 3d weighed intake non-liver patients 2056 33.5 91 100

Sarin*’^ 1997 67 ALD 73 ) diet history ) healthy controls 1547 83 Alcohol = 46% kcal 44 CLD 100 ) ) 1495 80

ON Table 4.1 Legend

Data expressed as mean values.

Energy intake from food only, alcohol excluded

Statistical analyses were not undertaken in all studies

Alcohol expressed as % of total energy intake

Data expressed as kcal/kg ideal body weight

ALD Alcoholic liver disease

PBC Primary biliary cirrhosis

CAH Chronic active hepatitis

AH Alcoholic hepatitis

ALC Alcoholic cirrhosis

CLD Chronic liver disease

NA Not available

PRV Published reference values

Difference from controls; p<0.01

Difference from food provided: p<0.0001

K)ON Table 4.2 Demographic details of subjects included in controlled studies of REE in patients with cirrhosis. Data expressed as mean values.

First author Subjects Sex Age Weight Height Alcoholic Abstinence Ascites cirrhosis from alcohol Year (n) (maleifemale) (year) (kg) (m) (%) (days) (%)

Owen'*'’* P 8 P 6:2 P 44.0 P 68.7 P 1.70 P 100 p NAP 25 1983 V 10 V 9:1 V 28.0 V 72.9 V 1.76 V - V - V -

Mullen"'" P 6 P 5:1 P 40.8 P 72.4 P 1.71 P NA p >30 P 0 1986 V 6 V 5:1 V 36.7 VNA VNAV - V - V -

Campillo*" p 30 p 21:9 p 54.0 P 61.9 P 1.65 p 20 p >42 p 63 1989 V 10 V 6:4 V 44.7 V 65.0 V 1.70 V - V - V -

Eriksson""" p 8 p 2:6 p 53.0 p 65.0 p 1.69 p 75 p NA p 63 1989 V 8 V matched V matched V 72.0 V 1.74 V - V - V -

Levine"*" p 17 p 9:8 p 51.1 p 67.9 p NA p 59 p >150 p NA 1989 V 17 V 9:8 V 52.6 V 65.0 V NAV - V - V -

Merli""* p 25 p 16:9 p 57.1 p 63.3 p 1.63 p 48 p >15 p 0 1990 V 10 V 4:6 V 57.0 V 64.4 V 1.62 V - V - V -

Schneeweiss""* p 22 p 17:5 p 48.9 p 65.8 p 1.73 p 100 p >60 p 27 1990 V 20 V 11:9 V 26.2 V 68.8 V 1.74 V - V - V -

Green"'"’ p 7 p 0:7 p 57.0 p 51.7 p 1.61 p 0 p NA p 0 1991 V 7 V 0:7 V 55.0 V 61.1 V 1.66 V - V - V -

Müller"'* p 10 p 6:4 p 38.2 p 63.0 p 1.71 p 100 p mean:365 p 50 1991 V 10 V 8:2 V 33.2 V 72.9 V 1.78 V - V - V -

Vermeij""* p 10 p 5 : 5 p 48.4 p 62.6 p 1.67 p 40 p >14 p 60 1991 V 50 V NA V 35.0 V 72.0 V 1.77 V - V - V -

Campillo*" p 10 p 5 : 5 p 50.0 p 65.3 p 1.63 p 100 p >14 p 40 1992 V 7 V 5:2 V 41.6 V 65.8 V 1.76 V - V - V -

P - patients; V - volunteers; NA - not available. ONu> Table 4.2 continued:

Demographic details of subjects included in controlled studies of REE in patients with cirrhosis. Data expressed as mean values.

First author Subjects Sex Age Weight Height Alcoholic Abstinence Ascites cirrhosis from alcohol Year (n) (male female) (year) (kg) (m) (%) (days) (%)

Müller^^ P 25 P NA P 40.1 P 65.5 P 1.71 P 40 P NAPNA 1992 V 10 V NA V 32.2 V 72.6 V 1.78 V - V - V -

Thome” ® P 8 P 7:1 P 51.6 P 77.0 P 1.75 P 75 P >21 P 38 1992 V 8 V 7:1 V 52.0 V 76.0 V 1.76 V - V - V -

Campillo*^ p 24 p 17:7 p 50.6 p 63.2 p NA p 100 p yes p 33 1993 V 13 V 9:4 V 29.9 V 63.8 VNAV - V - V -

Kmszynska^” p 8 p NA p 54.0 p 71.0 p 1.67 p NA p >30 p 0 1993 V 8 V NA V 49.0 V 73.0 V 1.74 V - V - V -

Kruszynska*’'* p 8 p 4:4 p 54.0 p 71.0 p 1.66 p 100 p >60 p 0 1993 V 6 V 3:3 V 49.0 V 73.0 V 1.75 V - V - V -

Campillo*® p 13 p 11:2 p 51.7 p 61.3 p NA p 100 p >90 p 31 1994 V 7 V 5:2 V 29.7 V 63.1 V NA V - V - V -

Kmszynska^” p 8 p 4:4 p 49.0 p 70.0 p 1.67 p 100 p >60 p 0 1994 V 7 V 4:3 V 46.0 V 75.0 V 1.70 V - V - V -

Martines^® p 7 p 7:0 p 55.0 p 74.0 p 1.73 p 71 p NA p 0 1994 V 6 V 6:0 V 35.0 V 72.0 V 1.75 V - V - V -

Waluga®"® p 15 p 10:5 p 37.7 p 72.5 p 1.70 p 0 p NA p none/mild 1996 V 20 V 13:7 V 31.9 V 70.4 V 1.75 V - V - V -

Watanabe” ^ p 4 p 3:1 p 56.0 p NA p NA p 0 p NA p NA 1996 V 3 V 2:1 V 47.0 V NAV NAV - V - V -

Chang** p 16 p 13:3 p 50.0 p 60.0 p NA p 56 p >90 p 75 1997 V 8 V 6:2 V 51.0 V 65.0 V NA V - V - V -

P - patients; V - volunteers; NA - not available. Table 4.3

Procedural details of controlled studies of REE in patients with cirrhosis.

First author Pre-test Pre-test Pre-test Measurement Measurement LBM FFM BCM dietary fasting rest technique duration assessment assessment assessment Year intake (hours) (minutes) (minutes)

Owen'"’* >1800kcal overnight 30 canopy + NA no no no 1983 10-12 Douglas bags

Mullen’’’* 35 kcal/kg overnight NA canopy + NA no oxygen-18 no 1986 Ig protein/kg gas analysers

Campillo*’ 2175 kcal overnight 20 mouthpiece + 4 x 3 no 4SF no 1989 12-15 Douglas bags

Eriksson’” regular diet overnight yes canopy + 60 no no no 1989 12-14 gas analysers

Levine^’ NA overnight NA open circuit 40 no no no 1989 calorimetry

MerlP” 30-35 kcal/kg overnight 30 canopy + until 5 min o f no 4SF no 1990 calorimeter (S) steady state

Schneeweiss*’* >1800 kcal overnight yes canopy + 15 no no no 1990 10-12 calorimeter (B)

Green^** standard overnight NA canopy + 30 no no no 1991 hospital diet calorimeter (D)

Muller^’* weight overnight 30 canopy + 40 24 hour UC BIA BIA 1991 maintaining 10-14 calorimeter (D)

Vermeij” * NA overnight 30 canopy + 30 no no no 1991 8+ calorimeter (D)

Campillo*^ "well balanced overnight 20 mouthpiece + 3 x 3 no no no 1992 and adequate" Douglas bags

PTO for legend

O n LA Table 4.3 continued: Procedural details of controlled studies of REE in patients with cirrhosis.

First author Pre-test Pre-test Pre-test Measurement Measurement LBM FFM BCM dietary fasting rest technique duration assessment assessment assessment Year intake (hours) (minutes) (minutes)

Müller^^ weight overnight 30 canopy + 60+ no BIABIA 1992 maintaining calorimetry (D)

Thorne"^ >200g overnight yes canopy + 60 no no no 1992 carbohydrate 12-14 gas analysers

Campillo*^ 2070kcal overnight 30 mouthpiece + 4 x 3 no no no 1993 Douglas bags

Kruszynska^” >200g overnight NA canopy + 30 no no no 1993 carbohydrate calorimetry (D)

Kruszynska^^" >200g overnight NA canopy + 30 no no no 1993 carbohydrate calorimetry (D)

Campillo** >2000kcal overnight 30 mouthpiece + 4 x 3 no no no 1994 Douglas bags

Kruszynska^^ >200g overnight NA canopy + 15 no no no 1994 carbohydrate 14 calorimetry (D)

Marlines^'’ >250g overnight NA canopy + 30 no no no 1994 carbohydrate calorimetry (M)

Waluga^^® 300g overnight 30 canopy + 30 BIA no no 1996 carbohydrate calorimetry (M)

Watanabe” ^ regular / overnight NA canopy + 30 no no no 1996 balanced diet calorimetry (D)

Chang®* 45-50% overnight 30 canopy + 45 no no no 1997 carbohydrate 13 calorimetry (D)

PTO for legend

O s OS Table 4.3 Legend

NA Not available

Douglas bag Intermittent collection of samples of expired gas

Gas analysers Continuous measurement and CO^ concentrations in expired gas

Open circuit No addition details available

Calorimeter Integrated system of expired gas collection, O 2 and CO^ analysis and calculation of energy expenditure: (S) manufactured by SensorMedics;

(B) manufactured by Beckman; (D) manufactured by Deltatrac; (M) manufactured by Medical Graphics Corporation.

UC Urinary creatinine excretion

4SF Four skinfold sites^'^^

<1ON 168

Table 4.4

REE in patients with cirrhosis expressed as a percentage of the mean value in healthy

volunteers.

First author REE REE/kg REE/m" REE/kg REE/kg REE/kg REE/ Year body body lean body FFM body g urinary weight surface mass cell creatinine area mass

Owen'*"® 1983 99 106 105 -- - -

Mullen^’® 1986 ns ns -- ns --

Campillo®’ 1989 99 104 104 - 95 --

Eriksson*^” 1989 86 98 -----

Levine^®’ 1989 113* 110 -----

Merli 1990"’" male: 97 102 -- 103 - -

female: 100 104 100

Schneeweiss"’®1990 107 112 108 --- 165*

Green^”® 1991 107 127* 116 ----

Müller"’® 1991 89 106 102 154* 133* 133* -

Vermeij’"® 1991 93 107 -----

Campillo®" 1992 -- 113* ----

Muller 1992""® A 86* 90 - - I l l --

B 90 115 136*

Thome’"® 1992 99 98 -----

Campillo 1993 ®" +L - - 104 ----

-L 100

Kruszynska""" 1993 97 100 -----

Kruszynska"’" 1993 101 104 -----

Campillo®® 1994 ------

Kruszynska""" 1994 106 113 -----

Martines""® 1994 104 101 -----

Waluga’"® 1996 96 95 95 90 ---

Watanabe”* 1996 -- 117* ----

Chang®® 1997 106 115 - - - - 145*

A/B Severity of liver disease'*''®; +/-L patients exhibiting postprandial lipogenesis (+) or not (-); ns not significantly different from volunteers; ‘different from volunteers p<0.05. 169

Table 4.5

Critical evaluation of the controlled studies of REE in patients with cirrhosis. A score of one is awarded for each procedural recommendation met.

First author Subject Methodology Body Total Year preparation (canopy) composition score analysis Pre-test Pre-test Abstinenc intake fasting e adéquat overnight confirmed (0-5) e >5 days

Owen""* 1983 1 1 0 1 0 3

Mullen^"" 1986 1 1 1 1 1 5

Campillo®^ 1989 1 1 1 1 4

Eriksson^” 1989 1 1 1 0 3

Levine^’ 1989 1 1 1 0 3

MeriP” 1990 1 1 1 1 1 5

Schneeweiss^’^ 990 1 1 1 1 1 5

Green“ “ 1991 1 1 NA = 1* 1 0 4

Muller*^ 1991 1 1 1 1 1 5

Vermeij^^* 1991 1 1 1 0 3

Campillo®^ 1992 1 1 1 0 3

M ülleP” 1992 1 1 1 1 4

Thome” ® 1992 1 1 1 1 0 4

Campillo®^ 1993 1 1 1 0 3

Kraszynska^” 1993 1 1 1 1 0 4

Kruszynska^’''1993 1 1 1 1 0 4

Campillo®® 1994 1 1 1 0 3

Kruszynska’”1994 1 1 1 1 0 4

Martines’'*® 1994 1 1 0 1 0 3

Waluga®"® 1996 1 1 N A = r 1 1 5

Watanabe®®' 1996 1 1 N A = 1* 1 0 4

Chang®® 1997 1 1 1 1 1 5

*No alcoholic patients included in study. 170

Table 4.6

Aetiology and functional severity of liver disease in 100 patients undergoing assessment of

REE.

Data expressed as number.

Aetiology of cirrhosis Subjects Child’s grade"''® (n) A B C

Alcohol 72 25 17 30

Post-viral 14 5 3 6

Biliary (PBC & PSC) 5 1 4 0

Autoimmune CAH 4 0 2 2

Cryptogenic 2 0 2 0

Other 3 1 1 1

Total 100 32 29 39

PBC - primary biliary cirrhosis; PSC - primary sclerosing cholangitis; CAH - chronic active hepatitis. Table 4.7

Demographic and anthropometric data in patients with cirrhosis and healthy volunteers undergoing assessment of REE.

Data expressed as mean ± ISD.

Variable Total population Men Women

P(n=100) V (n=41) p value P (n=56) V (n=20) p value P (n=44) V (n=21) p value

Age 49.3 48.9 ns 47.8 48.5 ns 51.1 49.1 ns (year) ±10.6 ±13.6 ±10.3 ±14.8 ±10.9 ±12.7

Sex 56M:44F 20M:21F ns 56M 20M _ 44F 21F (n)

Smokers :non- smokers 56:44 9:32 pO.OOOl 30:26 7:13 ns 26:18 2:19 p<0.0005 (n)

Height 1.67 1.69 ns 1.73 1.74 ns 1.60 1.64 ns (m) ±0.09 ±0.08 ±0.07 ±0.06 ±0.07 ±0.07

Weight 72.6 ns 78.8 ns 64.7 ns (kg) ±15.4 73.1 ±15.2 77.9 ±11.6 68.6 ±13.6 ±11.4 ±14.1 Corrected weight 68.3 ns 74.8 ns 60.2 <0.05 (kg) ±15.4 ±14.5 ±12.3

BMI 25.9 25.6 ns 26.5 25.6 ns 25.2 25.6 ns (kg.m-2) ±4.8 ±4.2 ±5.3 ±3.5 ±4.0 ±4.9

FFM 56.1 ns 63.6 ns 46.6 ns (kg) ±11.6 54.2 ±9.2 62.6 ±6.1 46.3 ±11.6 ±8.8 ±7.8 p’pjyjoonected 52.8 ns 60.3 ns 43.2 ns (kg) ±11.3 ±8.6 ±5.9

P - patients with cirrhosis; V - healthy volunteers. o 172

Table 4.8

Comparison of demographic and anthropometric data in patients with cirrhosis with and without clinical evidence of fluid retention and healthy volunteers.

Data expressed as mean ± ISD.

Variable Patients -fluid Patients +fluid Healthy volunteers (n=49) (n=51) (n=41)

Age (year) 49.3 ± 11.1 49.3 ± 10.2 48.9 ± 13.6

Sex (n) 29M;20F 27M:24F 20M.-21F

Height (m) 1.67 ± 0.09 1.68 ± 0.10 1.69 ± 0.08

Weight (kg) 73.4 ± 15.6 71.7 ± 15.3 73.1 ± 13.6 Corrected weight (kg) 65.1 + 14.9*

BMI (kg.m'^) 25.8 ± 4.2 26.0 ± 4.8 25.6 ± 4.2

FFM (kg) 57.9 ± 12.2 54.2 ± 10.8 54.2 ± 11.6 ppj^ooirected 51.3 ± 11.7

‘Different from healthy volunteers, p<0.01; -fluid no clinical evidence of fluid retention; ffluid clinical evidence of fluid retention. Table 4.9

REE and RQ in patients with cirrhosis and healthy volunteers.

Data expressed as mean ± ISD.

Variable Total population Men Women

P (n=100) V (n=41) p value P (n=56) V (n=20) p value P (n=44) V (n=21) p value

REE 1646 1579 ns 1808 1753 ns 1439 1414 ns (kcal) ±340 ±304 ±308 ±268 ±260 ±240

REE/weight 23.0 p<0.05 23.4 ns 22.5 p<0.05 (kcal/kg) ±3.8 21.7 ±3.9 22.6 ±3.5 20.9 ±2.9 ±2.8 ±2.8 REE/corrected weight 24.5 p<0.0001 24.6 p<0.05 24.3 p<0.0005 (kcal/kg) ±4.0 ±4.2 ±3.9

REE/FFM 29.7 ns 28.6 ns 31.0 ns (kcal/kg) ±4.4 28.8 ±4.2 28.1 ±4.5 29.5 ±3.7 ±2.9 ±4.3 REE/FFM"°""'^"^ 31.6 p<0.0005 30.1 p<0.05 33.4 p<0.005 (kcal/kg) ±4.6 ±4.3 ±4.4

RQ 0.79 0.81 p<0.05 0.80 0.82 ns 0.77 0.80 ns ±0.05 ±0.05 ±0.04 ±0.05 ±0.06 ±0.05

P - patients with cirrhosis; V - healthy volunteers. <1U) Table 4.10

Comparison of REE and RQ in patients with cirrhosis with and without clinical evidence of fluid retention and healthy volunteers.

Data expressed as mean ± ISD.

Variable Patients -fluid Patients +fluid Healthy volunteers Difference; Difference: Difference: (P-f)(n=49) (P+f)(n=51) (V)(n=41) P-f vs V P+f vs V P+f vs P-f

REE 1687 ± 352 1606 ±327 1579 ± 304 ns ns ns (kcal)

REE/weight 22.2 ± 3.7 ns p<0.05 (kcal/kg) 23.8 + 3.7 21.7 + 2.9 p<0.005 REE/corrected wt 25.1 ± 4.3 p<0.0001 ns (kcal/kg)

REE/FFM 28.1 ± 4.1 ns p<0.0005 (kcal/kg) 31.4 ± 4.2 28.8 ± 3.7 p<0.005 REE/FFM"°"="^"'^ 31.7 ± 5.0 p<0.005 ns (kcal/kg)

Respiratory 0.78 ± 0.05 0.79 ± 0.05 0.81 ± 0.05 p<0.05 ns ns quotient

+fluid - clinical evidence of fluid retention; -fluid - no clinical evidence of fluid retention. Table 4.11 REE and RQ in non-smoking patients with cirrhosis and healthy volunteers. Data expressed as mean ± ISD.

Variable Non-smoking subjects (all) Non-smoking men Non-smoking women

P (n=44) V (n=32) p value P (n=26) V (n=13) p value P (n=18) V (n=19) p value

REE 1628 1541 ns 1790 1740 ns 1393 1405 ns (kcal) ±349 ±311 ±310 ±296 ±262 ±245

REE/weight 22.6 ns 23.4 ns 21.5 ns (kcal/kg) ±4.1 21.2 ±4.2 21.6 ±3.7 20.9 ±2.3 ±1.4 ±2.8 REE/corrected weight 24.0 p<0.0005 24.7 p<0.005 23.1 p<0.05 (kcal/kg) ±4.1 ±4.4 ±3.6

REE/FFM 29.2 ns 28.9 p<0.05 29.7 ns (kcal/kg) ±4.5 28.6 ±3.9 27.0 ±5.3 29.7 ±3.8 ±1.8 ±4.4 REE/FFM‘=°"“‘®‘* 31.0 p<0.05 30.4 p<0.005 31.9 ns (kcal/kg) ±4.3 ±4.3 ±4.3

RQ 0.78 0.81 p<0.05 0.80 0.83 ns 0.77 0.80 ns ±0.05 ±0.05 ±0.04 ±0.05 ±0.05 ±0.05

P - patients with cirrhosis; V - healthy volunteers. LA Table 4.12

REE and RQ in patients with cirrhosis and healthy volunteers who smoked.

Data expressed as mean ± ISD.

Variable Smoking subjects (all) Smoking men Smoking women

P (n=56) V (n=9) p value P (n=30) V (n=7) p value P (n=26) V (n=2) p value

REE 1660 1716 ns 1823 1798 ns 1471 1429 ns (kcal) ±336 ±246 ±311 ±206 ±258 ±139

REE/weight 23.3 ns 213 ns 213 ns (kcal/kg) ±3.5 23.6 ±3.8 23.8 ±3.1 23.0 ±3.9 ±4.4 ±0.1 REE/corrected weight 24.8 ns 24.5 ns 25.1 ns (kcal/kg) ±4.0 ±4.1 ±3.9

REE/FFM 30.1 ns 28.4 ns 31.9 ns (kcal/kg) ±4.4 29.5 ±4.4 28.9 ±3.6 31.5 ±3.6 ±3.9 ±1.9 REE/FFM“°™““^‘^ 32.0 ns 29.9 ns 34.4 ns (kcal/kg) ±4.8 ±4.4 ±4.2

Respiratory 0.79 0.81 ns 0.80 0.80 ns 0.78 0.84 p<0.0001 quotient ±0.06 ±0.05 ±0.05 ±0.06 ±0.06 ±0

P - patients with cirrhosis; V - healthy volunteers. <1 Os Table 4.13

Comparison of anthropometric data and REE in male and female patients with cirrhosis. Data expressed as mean ± ISD.

Variable All patients Non-smoking patients Smoking patients

Men Women P Men Women P Men Women P (n=56) (n=44) value (n=26) (n=18) value (n=30) (n=26) value

Height 1.73 1.60 p<0.0001 1.70 1.59 p<0.0001 1.75 1.61 p<0.0001 (m) ±0.07 ±0.07 ±0.06 ±0.08 ±0.07 ±0.07

Weight 78.8 64.7 p<0.0001 78.2 65.8 p<0.01 79.3 63.9 p<0.0001 (kg) ±15.2 ±11.6 ±17.2 ±12.3 ±13.4 ±11.3

Corrected weight 74.8 60.2 p<0.0001 74.0 61.2 p<0.01 75.4 59.5 p<0.0001 (kg) ±14.5 ±12.3 ±15.8 ±13.4 ±13.7 ±11.6

FFM 63.6 46.6 pO.OOOl 62.6 47.3 p<0.0001 64.4 46.0 p<0.0001 (kg) ±9.2 ±6.1 ±10.9 ±6.9 ±7.5 ±5.6

ppjyjcoirected 60.3 43.2 p<0.0001 59.2 43.8 p<0.0001 61.2 42.8 p<0.0001 (kg) ±8.6 ±5.9 ±10.0 ±6.2 ±7.2 ±5.7

REE 1808 1439 p<0.0001 1790 1393 p<0.0001 1823 1471 p<0.0001 (kcal) ±308 ±260 ±310 ±262 ±311 ±258

REE/FFM 28.6 31.0 p<0.01 28.9 29.7 ns 28.4 31.9 p<0.005 (kcal/kg) ±4.2 ±4.5 ±3.9 ±5.3 ±4.4 ±3.6

REE / FFM"°"""^'^ 30.1 33.4 p<0.0005 30.4 31.9 ns 29.9 34.4 p<0.0005 (kcal/kg) ±4.3 ±4.4 ±4.3 ±4.3 ±4.4 ±4.2

RQ 0.80 0.77 p<0.05 0.80 0.77 p<0.05 0.80 0.78 ns ±0.04 ±0.06 ±0.04 ±0.05 ±0.05 ±0.06

Variables not shown; age, BMI and REE/kg: no significant differences found between male and female patients.

'-a Table 4.14

Comparison of anthropometric data and REE in male and female volunteers.

Data expressed as mean ± ISD.

Variable All volunteers Non-smoking volunteers Smoking volunteers

Men Women P Men Women P Men Women P (n=20) (n=21) value (n=13) (n=19) value (n=7) (n=2) value

Height 1.74 1.64 p<0.0001 1.75 1.62 p<0.0001 1.77 1.64 p<0.05 (m) ±0.06 ±0.07 ±0.05 ±0.05 ±0.07 ±0.01

Weight 77.9 68.6 p<0.05 80.4 67.9 p<0.05 76.8 62.3 ns (kg) ±11.4 ±14.1 ±12.6 ±13.6 ±9.5 ±5.8

FFM 62.6 46.3 p<0.0001 64.3 45.2 p<0.0001 62.7 45.4 p<0.0005 (kg) ±8.8 ±7.8 ±9.2 ±6.0 ±6.1 ±1.6

REE 1753 1414 p<0.001 1740 1405 p<0.005 1798 1429 ns (kcal) ±268 ±240 ±296 ±245 ±206 ±139

REE/FFM 28.1 29.5 ns 27.0 29.7 p<0.05 28.9 31.5 ns (kcal/kg) ±2.9 ±4.3 ±1.8 ±4.4 ±3.9 ±1.9

RQ 0.82 0.80 ns 0.83 0.80 ns 0.80 0.84 ns ±0.05 ±0.05 ±0.05 ±0.05 ±0.06 ±0

Variables not shown; age, BMI and REE/kg; no significant differences found between male and female volunteers.

oo Table 4.15

Demographic and anthropometric data in patients with cirrhosis undergoing assessment of REE: Effect of disease severity.

Data expressed as mean ± ISD.

Variable All patients Men Women

C (n=43) D (n=57) p value C (n=26) D (n=30) p value C (n=17) D (n=27) p value

Age 50.3 48.4 ns 49.9 46.0 ns 51.1 51.2 ns (year) ±11.4 ±10.1 ±11.7 ±8.7 ±11.2 ±10.9

Sex 26M:17F 30M-.27F ns 26M 30M . 17F 27F (n)

Smokersmon-smokers 22:21 34:23 ns 14:12 16:14 ns 8:9 18:9 ns (n)

Height 1.66 1.68 ns 1.71 1.74 ns 1.59 1.61 ns (m) ±0.10 ±0.09 ±0.07 ±0.07 ±0.08 ±0.07

Weight 73.7 71.7 ns 77.8 79.7 ns 67.5 62.9 ns (kg) ±16.3 ±14.8 ±16.5 ±14.1 ±14.1 ±9.7

Corrected weight (kg) 73.0 64.8 pO.Ol 77.1 72.7 ns 66.8 56.0 p<0.05 ±16.0 ±13.9 ±16.0 ±13.1 ±14.4 ±8.6

BMI 26.6 25.4 ns 26.5 26.5 ns 26.6 24.3 ns (kg.m-^) ±5.1 ±4.5 ±5.6 ±5.2 ±4.5 ±3.4

FFM 55.1 56.8 ns 61.5 65.4 ns 45.3 47.3 ns (kg) ±11.7 ±11.7 ±9.4 ±8.8 ±7.1 ±5.4 ppj^conected 54.5 51.4 ns 60.9 59.7 ns 44.8 42.2 ns (kg) ±11.3 ±11.2 ±8.8 ±8.5 ±6.9 ±5.0

C - patients with compensated liver disease"*^®; D - patients with decompensated liver disease''^®. Table 4.16

REE and RQ in patients with cirrhosis: Effect of disease severity.

Data expressed as mean ± ISD.

Variable All patients Men Women

C (n=43) D (n=57) p value C (n=26) D (n=30) p value C (n=17) D (n=27) p value

REE 1687 1614 ns 1822 1796 ns 1481 1413 ns (kcal) ±357 ±327 ±336 ±287 ±287 ±243

REE/weight 23.2 22.8 ns 23.8 22.9 ns 22.3 22.7 ns (kcal/kg) ±3.5 ±4.0 ±3.6 ±4.3 ±3.2 ±3.7

REE/corrected weight 23.4 25.2 p<0.05 24.0 25.1 ns 22.5 25.4 p<0.05 (kcal/kg) ±3.4 ±4.3 ±3.5 ±4.7 ±3.1 ±4.0

REE/FFM 30.9 28.8 p<0.05 29.7 27.7 ns 32.7 30.0 p<0.05 (kcal/kg) ±4.1 ±4.5 ±3.8 ±4.3 ±4.1 ±4.5

REE/FFM"°"="'=® 31.2 31.8 ns 29.9 30.3 ns 33.1 33.5 ns (kcal/kg) ±3.9 ±5.1 ±3.6 ±4.9 ±3.6 ±4.9

RQ 0.78 0.79 ns 0.80 0.80 ns 0.75 0.79 p<0.05 ±0.05 ±0.05 ±0.05 ±0.04 ±0.05 ±0.06

C - patients with compensated liver disease'"'®; D - patients with decompensated liver disease'' oo o Table 4.17

Demographic and anthropometric data in 100 patients with cirrhosis; Effect of nutritional status.

Data expressed as mean ± ISD.

Variable Well nourished Moderately malnourished Severely malnourished p value* (or suspected of being) (n=53) (n=33) (n=14)

Age 49.6 ± 10.5 49.8 + 10.8 46.8 ± 11.2 ns (year)

Height 1.67 ± 0.08 1.66 ± 0.11 1.68 ± 0.10 ns (m)

Weight 79.5 + 14.2 67.1 ± 12.4 59.2 ± 12.2 p<0.0001 (kg)

Corrected weight (kg) 77.6 ± 13.0 60.5 ± 10.1 51.9 ± 8.5 p<0.0001

FFM 59.3 ± 11.6 53.4 ± 11.1 50.2 ± 9.8 p<0.001 (kg)

pp^yjooneoted 57.8 ± 10.4 48.3 ± 9.8 44.2 + 8.6 p<0.0001 (kg)

BMI 28.4 ± 4.4 24.2 ± 3.3 20.8 ± 3.1 p<0.0001 (kg.m'2)

'Oneway ANOVA Table 4.18

REE and RQ in 100 patients with cirrhosis: Effect of nutritional status.

Data expressed as mean ± ISD.

Variable Well nourished Moderately malnourished Severely malnourished p value* (or suspected of being) (n=53) (n=33) (n=14)

REE 1766 ± 329 1495 ± 295 1545 ± 328 p<0.005 (kcal)

REE/weight 22.4 ± 2.9 22.6 ±4.1 26.3 ± 4.2 p<0.005 (kcal/kg)

REE/corrected weight 22.9 ± 3.0 24.8 ± 3.9 29.6 ± 3.2 p<0.0001 (kcal/kg)

REE/FFM 30.1 ± 4.3 28.5 ±4.9 30.9 ± 3.6 ns (kcal/kg)

REE/FFM"°"""^"' 30.8 ± 4.2 31.3 ± 5.0 35.0 ± 3.9 p<0.01 (kcal/kg)

RQ 0.79 ± 0.05 0.78 ± 0.04 0.80 ± 0.07 ns

‘Oneway ANOVA

K)o o Table 4.19 Comparison between measured REE and estimated values predicted by the Harris-Benedict^°* and Schofield''*^ equations in patients with cirrhosis. Data expressed as mean ± ISD.

Population Number Measured REE Predicted REE (Harris-Benedict) Predicted REE (Schofield)

(kcal) (kcal) % of MREE p value (kcal) % of MREE p value (MREE vs FREE) (MREE vs FREE)

All patients 100 1646 ± 340 1525 ±252 94.3 ± 12.7 p<0.01 1562 ±256 96.7 ± 14.3 ns

Men 56 1808 ± 308 1684 ±208 94.5 ± 12.2 p<0.05 1737 ± 191 97.8 ± 13.8 ns

Women 44 1439 ± 260 1324 ± 128 94.0 ± 13.4 p<0.05 1340 ± 117 95.4 ± 14.9 p<0.05

All non-smokers 44 1628 ± 349 1516 ±260 95.0 ± 13.8 ns 1555 ±255 97.7 ± 15.7 ns

Male non-smokers 26 1790 ± 310 1651 ±242 93.4 ± 12.8 ns 1702 ± 219 96.7 ± 14.7 ns

Female non-smokers 18 1393 ±262 1321 ± 128 97.2 ±15.2 ns 1342 ± 115 99.1 ± 17.5 ns

All smokers 56 1660 ± 336 1533 ±247 93.8 ± 11.9 p<0.05 1568 ±258 96.0 ± 13.1 ns

Male smokers 30 1823 ± 311 1713 ± 171 95.5 ± 11.8 ns 1767 ± 161 98.7 ± 13.2 ns

Female smokers 26 1471 ± 258 1325 ± 131 91.8 ± 11.9 p<0.05 1338 ± 120 92.8 ± 12.5 p<0.05

Compensated patients 43 1687 ± 357 1531 ±251 92.3 ±11.7 P<0.05 1564 ±246 94.7 ± 14.3 ns

Decompensated patients 57 1614 ± 327 1522 ±254 95.8 ± 13.4 ns 1560 ±265 98.3 ± 14.2 ns

MREE - measured resting energy expenditure; FREE - predicted resting energy e^enditure.

oU) o Table 4.20

Comparison between measured REE and estimated values predicted by the Harris-Benedict^®^ and Schofield"**^ equations in healthy volunteers.

Data expressed as mean ± ISD.

Population Number Measured REE Predicted REE (Harris-Benedict) Predicted REE (Schofield)

(kcal) (kcal) % of MREE p value (kcal) % of MREE p value

(MREE vs FREE) (MREE vs FREE)

All volunteers 41 1579 ±304 1526 ±236 98.0 ± 12.1 ns 1534 ±241 98.5 ± 12.5 ns

Men 20 1753 ±268 1672 ± 212 96.2 ± 10.3 ns 1705 ± 202 98.5 ± 12.7 ns

Women 21 1414 ± 240 1387 ± 164 99.7 ± 13.6 ns 1371 ± 143 98.5 ± 12.7 ns

Non-smoking volunteers 32 1541 ± 311 1507 ±236 99.2 ±11.6 ns 1516 ±250 99.8 ± 12.5 ns

Male non-smokers 13 1740 ± 296 1704 ± 202 98.9 ± 6.7 ns 1747 ± 193 101.8 ± 11.2 ns

Female non-smokers 19 1405 ± 245 1372 ± 148 99.5 ± 14.2 ns 1359 ± 134 98.5 ± 13.4 ns

Smoking volunteers 9 1716 ± 246 1593 ±235 93.6 ± 13.2 ns 1596 ±211 93.8 ± 12.4 ns

Male smokers 7 1778 ± 226 1611 ±234 91.3 ± 14.2 ns 1628 ±210 92.3 ± 13.9 ns

Female smokers 2 1499 ±238 1529 ± 317 101.7 ± 5.0 ns 1483 ± 241 99.0 ± 0.4 ns

MREE - measured resting energy expenditure; FREE - predicted resting energy expenditure.

oo4^ 85

« 1000

”2 800 4-

600 • • 400

200

c -200 -

5 -400

1000 1200 1400 1600 1800 2000 2200 Predicted REE (kcal)

Figure 4.1 Harris Benedict"®^ prediction formulae for REE: Comparison of measured - predicted against predicted values for REE in 100 patients with cirrhosis. 186

400 -

Predicted REE (kcal)

Figure 4.2 Harris Benedict“^^ prediction formulae for REE: Comparison of measured - predicted against predicted values for REE in 41 healthy volunteers. 187

800 j

s 600 .aÎ 1 400 CL

200 - ia 1 0 - ë s -200 -

£ -400 - ê Q -600 - 1000 1200 1400 1600 1800 2000 2200 Predicted REE (kcal)

Figure 43 Schofield"^^” prediction formulae for REE: Comparison of measured - predicted against predicted values for REE in 100 patients with cirrhosis. 188

500 -r s â 400 ? 300 - 1.1 • # Q. 200 - # e

i3 100 i B 0 - u -100

-200 - s e 1 -300 - -400 - 1000 1200 1400 1600 1800 2000 2200 Predicted REE (kcal)

Figure 4.4 Schofield^^“ prediction formulae for REE: Comparison of measured - predicted against predicted values for REE in 41 healthy volunteers. 189 CHAPTERS

THE ASSESSMENT OF BODY COMPOSITION IN PATIENTS WITH CIRRHOSIS

5.1 Outline The assessment of body composition in patients with cirrhosis is difficult because of the innate characteristics of the condition, most significantly fluid retention, which contravene the assumptions upon which most single assessment techniques are based. Consequently, results derived using single techniques are unvalidated and probably inaccurate. Data obtained using a number of different single techniques can be combined to provide more reliable estimates of body composition, providing that each contributing component is not dependent on violable assumptions. The most practical of such multi-compartment models is based on assessments of body density, total body water and body mineral obtained using underwater weighing, deuterium dilution and dual energy X-ray absorptiometry (DEXA) scanning.

Before applying this model to patients with cirrhosis, two preliminary studies were undertaken. In order to investigate potential errors in underwater weighing which might arise from the flatogenic effects of commonly prescribed non-absorbable disaccharides, volumes of gastrointestinal gas were evaluated by computed tomography. No evidence of increased gas volumes were found with this medication. The equilibration of deuterium in body fluid was examined in six patients with cirrhosis, three of whom had gross ascites, in order to determine the optimum time for collecting post-dose samples in dilution studies. In patients without clinically detectable fluid retention, this was determined as 6 hours after oral administration.

The multi-compartment model was then used to assess body composition in 20 patients with cirrhosis, none of whom had clinically evident fluid retention. Chapter 5 Body composition in cirrhosis 190

Significant differences were found between estimates of body fat and FFM derived using the model and considered reference data, and those obtained using each of the single techniques and skinfold anthropometry, confirming that the single techniques are inaccurate in this patient population. Mean values for the hydration fraction and density of FFM differed significantly from published values in healthy volunteers (0.764 vs 0.738 and 1.097 vs 1.102 kg/1). Whilst because of its practical limitations this multi-compartment model must be considered a research tool, its use has been demonstrated as feasible in patients with cirrhosis and thus provides the first reference method of assessing body composition in this population.

5.2 Introduction Accurate body composition data are needed in cirrhotic patients if currently available bedside assessment methods, for example skinfold anthropometry, are to be validated. They are also essential if accurate assessments of energy expenditure per unit of metabolically active tissue are to be made^^ and if nutritional therapy is to be evaluated^^^'^^

Attempts to determine the composition of the human body were first made in the 1940’s^^ Subsequently, a variety of techniques have been developed to measure various body compartments or subcompartments. Some of these techniques provide "direct" estimates of body compartments, although assumptions are often made in the derivation of the data which, while valid in the healthy populations in which the techniques were developed, may not be valid in diseased populations. Other methods provide "indirect" estimates of body compartments calculated on the basis of certain assumptions, and these require validation against the more direct techniques. The terms "direct" and "indirect" may alternatively be expressed as "reference" and "bedside" but these categorizations are less helpful than the description of techniques in terms of their precision, accuracy and applicability to the population under investigation. Chapter 5 Body composition in cirrhosis 191

5.3 Densitometry Evaluation of whole body density can be used to provide estimates of body fat and FFM. Density measurements are undertaken by immersing the subject in water and measuring either the volume of water displaced^^^ or the weight of the subject underwater^^"^, or by measuring air displacement in a plethysmographic chamber^^^'^^^. Purpose-built and permanently sited equipment are required for both methods. In addition, facilities for measuring the residual air in the respiratory tract are essential if an accurate assessment is required and this can be undertaken relatively simply using a helium dilution technique^'^. The volume of gas within the gastrointestinal tract must also be taken into account and for this, in most studies, a correction of 100 ml is applied^^’^®^’^^^.

Practical considerations limit the determination of density by submersion in water as this cannot be undertaken in individuals who are severely ill, debilitated, injured or at the extremes of age. However, the plethysmographic method of measuring body density simply requires the subject to sit, minimally clothed, in an enclosed chamber and thus is more practical except for individuals who require constant clinical monitoring.

The technique of assessing body composition using densitometry is based on the assumption that the fat and FFM compartments are distinct and that each has a known and constant density of 0.9 kg/1 and 1.1 kg/1 respectively. The proportion of fat in the body can be calculated directly using either the formula devised by Siri^^^:

- 4 5 / X 700; or that proposed by Brozek^^:

% /hr = - 4^42/ X 7 # Chapter 5 Body composition in cirrhosis 192

where d = body density.

The main theoretical limitation of the method is that physiological and pathological variations in density of either fat or FFM compartments will lead to errors in the calculation of body fat. Variations in the hydration state of FFM, the density of bone mineral or volume of gastrointestinal gas will compromise the accuracy and precision of the method. However, for many years, densitometry has been considered a reference method against which other techniques are validated^^^.

Errors of 3 to 4% in the estimates of body fat may occur due to variation in the density of fat and FFM^°^'"^^^. In studies where simultaneous measurements of respiratory tract volume are not made, or when relatively large volumes of gastrointestinal tract gas are present, additional errors of 1 to 2% may arise. The precision of repeated measurements represent errors of less than 1%^^.

Densitometry in chronic liver disease No studies have been published to date on the use of densitometry in patients with chronic liver disease. Its value as a two compartment method, that is in assessing fat and FFM using the Siri or Brozek equations, in this patient population is questionable because of the confounding effects of fluid retention^^ and abnormalities of mineral metabolism^®^ which invalidate the assumptions of constant densities upon which these equations depend.

However, densitometry may play a role in a multicompartment approach to assessing body composition and this is discussed below.

The assumption that the volume of gastrointestinal gas is approximately 100 ml may not be tiue in some patients with cirrhosis who are frequently prescribed non­ absorbable disaccharides, lactulose and lactitol, for the management of portal systemic encephalopathy^^^. These agents exert their beneficial effect by increasing Chapter 5 Body composition in cirrhosis 193 stool frequency and weight and by reducing intestinal ammonia production and absorption^®®. However, one of the side effects associated with their use is the production of excess flatus^®^. A quantitative assessment of gas within the gastrointestinal tract is required in patients prescribed these drugs if accurate measurements of body density are to be made by underwater weighing.

5.4 Dilution techniques Isotope dilution studies can be undertaken to evaluate the size of body fluid compartments using the principle that the quantity of an isotope tracer introduced into an unknown volume is the same before and after mixing in that volume^^^. Thus, when the initial volume and concentration of the tracer and its final concentration are known, the final volume in which it is dispersed can be calculated. The tracer must be harmless to the subject undergoing assessment, remain predominantly within the specified body volume, reach a plateau concentration on mixing before significant quantities are excreted or metabolized and be measured in samples of body fluid with relative ease^^^.

Antipyrine was used by early investigators to assess total body water (TBW)^®^ but results from subsequent studies questioned its reliability as it was found to be metabolized, selectively bound to protein and influenced by smoking behaviour'^^^’^^^. TBW can be more reliably estimated by dilution of deuterium

(2r )476, oxygen-18 or tritium (^H)'^®^, the first two of which are stable isotopes while the third is radioactive.

Extracellular water (ECW), from which intracellular water (ICW) can be calculated by difference from TBW, can be estimated by dilution of stable or radioactive bromide^^"^’^^^.

Estimates of LBM and BCM can be made from TBW and ECW using the formulae: Chapter 5 Body composition in cirrhosis 194

LBM = TBW/ 0.732 ^lo

BCM = (TBW - ECW) / 0.72

These formulae, derived from studies undertaken in healthy individuals, are based on the assumptions that TBW comprises 73.2% of LBM and that ICW makes up 72% of BCM.

The proportion of water in LBM, 73.2%, is a mean value derived from a number of studies of difference mammals, excluding man"^^^, and is equal to that determined by chemical analysis of one cadaver by Widdowson and colleagues^^^. An alternative figure of 72% has also, but less frequently, been used^^^’"^^^.

Variations in hydration state from these standard figures will invalidate the assumptions made and result in inaccurate estimations of LBM and BCM.

The errors arising from these techniques are dependent on the tracer, the analytical precision of the method used to measure tracer concentrations and variations in subjects’ hydration status. The precision for measuring deuterium concentrations using infrared absorption is less than 2.5%^^’^^"^.

Dilution techniques in chronic liver disease Body water spaces have been assessed in patients with cirrhosis using a variety of tracers including antipyrine, deuterium, oxygen-18, tritium and bromide-82 (Table ^^^6414415,231,3m,320,321,363,376,404,445,^^^^^ With the exccptionof data derived from antipyrine studies, mean values for TBW as a percentage of body weight in patients without ascites (56-66%) and for those with ascites (58-70%), are higher than those recorded by Moore and colleagues for healthy individuals (48-59%)^^^. However, although the mean values for ECW, expressed as a percentage of TBW, in patients without ascites (41-45%) are of a similar range to those recorded by Moore and colleagues for healthy individuals (41-47%), the values are higher in Chapter 5 Body composition in cirrhosis 195 patients with ascites (46-68%).

Data on body water spaces have been extrapolated to other body compartments in patients with cirrhosis in a number of studies^^®’^^^’^^^’^^^. Mullen and colleagues^^^ derived FFM from TBW but did not give details of the calculation. However, by back calculation, a hydration fraction of 70.2% appears to have been used in both patients with cirrhosis and healthy controls. This is less than the standard hydration fraction of 73.2% proposed for healthy individuals"^^^, and thus difficult to justify, particularly as it is not referenced.

In two studies by McCullough and co-workers^^^'^^\ FFM and BCM were derived from TBW and ECW respectively using, in both cases, an assumed figure of 73.25% for the hydration fraction of FFM, even though eight out of 21 patients in one study^^® had ascites. This figure is inappropriate as body water compartments were obviously expanded in some patients"^^®.

Verboeket-van der Venne and colleagues^^^ similarly derived FFM from TBW using a hydration fraction of 73% in eight patients with cirrhosis. Even though the patients had no evidence of fluid retention, this derivation can also be criticised as no studies have confirmed that this figure is valid in patients with cirrhosis.

Interestingly, only one of these studies^^^, was undertaken with the primary aim of investigating body composition, with the others focusing on protein metabolism^^^’^^^ or energy expenditure^^^. While the accuracy of the body composition data derived in these studies must be questioned, it is a concern that flawed methodology is being used as a basis for investigating wider nutritional issues in this patient population.

The optimum times for collecting post-dose samples have not been systematically examined in patients with chronic liver disease. The equilibration of tracers Chapter 5 Body composition in cirrhosis 196

throughout body fluids is known to be prolonged in individuals with fluid retention^^^’^^^’'^^’'^ ’'^^^, but post-dose samples have been collected at varying times after tracer dosing in patients with chronic liver disease (Table 5.1). Further work is required to establish the optimum times at which this should be undertaken.

5.5 Total body potassium Potassium is the major intracellular cation in lean tissue but is found in much smaller quantities in adipose tissue. Thus, measurement of TBK can facilitate estimation of or BCM^^^. This can be undertaken by measuring naturally occurring radioactive using a sensitive and well shielded counter; patients are required to lie still within an enclosed machine for up to 60 minutes^k The availability of sufficiently sensitive counters is limited and, while the method is safe, it is impractical for severely ill, confused or claustrophobic individuals or those at the extremes of age.

Calculation of LBM from TBK is based on the assumption that this tissue has a constant potassium content. Published values vary and among the most commonly used are those determined by Forbes and colleagues:

men: 66 mmol /kg; women: 60 mmol /kg men: 68.1 mmol /kg; women: 64.2 mmol /k g

However, these assumptions may not be true in individuals with potassium-wasting conditions, such as hyperaldosteronism, bulimia or chronic liver disease, or in those taking loop diuretics^^.

The error of potassium counting has been reported as 5%^^ which is due to the differences in detectors and geometrical variations of the subjects. However, when data are extrapolated to lean tissue, further errors will arise from variations in potassium content of lean body mass. Chapter 5 Body composition in cirrhosis 197

TBK in chronic liver disease TBK has been measured in patients with chronic liver disease in a number of studies (Table although there is evidence in this patient population that disturbances of potassium metabolism may invalidate the assumptions used for calculation^^ "^^^. In addition, while excessive adipose tissue is known to have a shielding effect which may result in inaccurately low estimates of potassium"^^® potentially similar effects of ascites have not been investigated.

Data on TBK, where published, have been expressed in absolute terms"^^’^^^ or as where BCM (kg) = 0.00833 mmol K^ Limited evidence from one study^^^ suggests that TBK is depleted in patients with cirrhosis and becomes increasingly so with deteriorating liver function.

Comparisons have been made between BCM or FFM derived from TBK and estimates determined using DEXA^, BIA, anthropometry and 24-hour urinary creatinine excretion^^^’^^^’^^®’^^^’"^^. However, these comparative techniques are indirect rather than reference methods of assessing body composition. In addition, most of the statistical analyses undertaken have inappropriately used correlations'^^. The bias between methods was quantified in only two studies^^’^^^; in one, there were significant differences between data derived from TBK and DEXA^, and in the other, while the differences between TBK and BIA-derived data were insignificant, the standard deviation of the mean difference was 28%^^.

It can be concluded from the limited information available that the use of TBK in the assessment of body composition in this patient population is inappropriate.

5.6 Neutron activation In vivo neutron activation analysis facilitates elemental analysis by measuring the type and frequency of radiation emitted from the body by different elements after irradiation with fast neutrons^^. This is the only method, other than chemical Chapter 5 Body composition in cirrhosis 198

analysis, of determining the elemental composition of the body but involves exposing the subject to a dose of up to 6 mSv of radiation^^ which is equivalent to the radiation exposure from 300 chest X-rays"^^^. Total body nitrogen can be determined using prompt gamma neutron activation, calcium, sodium and phosphorus by delayed gamma neutron activation, and total body carbon by inelastic scattering"^^^. The facilities required to undertake neutron activation are restricted to a few centres world-wide and this, combined with the relatively high dose of radioactivity and cost, limits the practical availability of this method.

The accuracy of estimates of body composition varies with the element assessed, the phantoms used for comparison and the facilities where studies are undertaken. Values for assessing nitrogen have been reported to range from 2 to 10%^.

Neutron activation in chronic liver disease No assumptions are made in the direct quantification of body elements by neutron activation which might be violated in diseased populations. Therefore, this method should provide useful data in patients with chronic liver disease. To date, two such studies"^^’"^^ have been undertaken in this patient population.

Blendis and colleagues'^^ measured total body nitrogen (TBN) in seven patients with cirrhosis and ascites. Data were expressed as a nitrogen index in which the measured total body nitrogen was compared with values predicted from the height and arm span using published tables^^^. The mean total body nitrogen for the group was 1.54 ± 0.10 kg with a low nitrogen index of 74 ± 4%, suggesting protein depletion. TBK was also assessed in these patients but no comparisons were made between the two measurements although a TBN/TBK ratio was calculated. The patients underwent repeat neutron activation studies 12 months after insertion of a Le Veen shunt to relieve ascites and, in three patients, an increase in mean total body nitrogen from 1.65 ± 0.04 kg to 1.78 ± 0.07 kg was observed. Chapter 5 Body composition in cirrhosis 199

Prijatmoko and co-workers'*^^ also measured total body nitrogen in 38 patients with cirrhosis. Total body protein was calculated from the total body nitrogen using the formula^*^^:

total body protein = total body nitrogen x 6.25 and data were also expressed as a nitrogen index^^®,

nitrogen index = measured total body nitrogen predicted total body nitrogen.

No quantitative data were reported for either total body nitrogen or protein but a significantly lower nitrogen index was observed in the patients compared with matched healthy volunteers.

The extrapolation of total body nitrogen data to total body protein using a factor of 6.25 used by Prijatmoko et al'*'*^ may be confounded in chronic liver disease because of abnormalities of nitrogen metabolism in this patient population^^®.

5.7 DEXA DEXA measures the differential attenuation by various tissues of X-rays which are beamed at the body at two different energy levels, thus distinguishing between bone mineral, fat and fat-free soft tissue^^. The procedure is relatively easy and quick to perform, involving the subject lying supine on the scanner table for 5-20 minutes during which they are accessible should they require clinical attention. A whole body DEXA scan provides a radiation dose of less than 0.005 mSv or the equivalent of one day’s background radioactivity in the United Kingdom^\ The scanners, although expensive, are becoming more accessible as they are extensively used for screening for metabolic bone disease^^®. Chapter 5 Body composition in cirrhosis 200

However, there are a number of disadvantages to using DEXA to assess body compartments, including the practical issues of small scanner tables, incompatibility of data from different machines which precludes comparisons, and the need to reanalyse data as updated software becomes available. Theoretical problems include the different attenuation effects of tissue of varying hydration status which may result in incorrect estimates of lean tissue^^^’^^^. However, changes in hydration do not appear to reduce the ability of DEXA to measure bone mineral^^^’^^^. Measurements may, however, be affected by the depth of the tissue measured^^^; errors increase with tissue thickness^^.

The short-term precision of DEXA estimates of whole body mineral is approximately and for body fat and LEM, approximately 1.5%^^.

DEXA in chronic liver disease DEXA and its fore-runner, dual photon absorptiometry, an earlier dual energy projection method, have been used predominantly to assess bone mineral density in patients with chronic liver Few studies have been undertaken to investigate body compartments^^'^^^"*^^'^^'^^^ (Table 5.3).

Meys and co-workers^^^ compared changes in bone loss using DEXA in 31 patients following orthotopic liver transplantation with 33 patients with chronic liver disease comparable for age, sex and menopausal status. They also prospectively assessed 16 patients before and after transplantation. All patients underwent a whole body DEXA scan and total body bone mineral and fat were assessed. No significant difference was found in total body bone mineral between the post-transplant patients and those with chronic liver disease although a significant decrease of 3.5% was observed in the 16 patients who were re-assessed after the first post­ operative year. Total body fat was significantly higher in the post-transplant patients than those with chronic liver disease but no data on body fat were provided from the prospective study. Chapter 5 Body composition in cirrhosis 201

The effects of liver transplantation on body composition were also examined by Hussaini and colleagues^^ who undertook whole body DEXA scans on 35 patients before and after surgery. A significant increase of 78% was observed in total body fat, accompanied by a significant decrease in FFM of 6% between 12 to 24 months after transplantation.

Bramley and colleagues^^ used DEXA to assess total body fat in 53 patients with cirrhosis, 43% of whom had significant fluid retention. Comparisons were then made with measurements derived from skinfold anthropometry, BIA and TBK. Significant bias was observed between results determined using DEXA and both anthropometry and TBK. The bias between DEXA and BIA was not significant but had 95% limits of agreement of 12.4 kg.

Oldroyd and colleagues'^®^ used DEXA to assess FFM and total body fat in the same group of cirrhotic patients used by Bramley and co-workers^^ but comparisons with results obtained using other single assessment techniques were not undertaken.

Wicks and colleagues^^^ found a significant correlation between estimates of body fat made using DEXA and TSF in a group of 21 patients with primary biliary cirrhosis. However, the bias between the estimates was not reported.

Prijatmoko and colleagues'^^ undertook whole body DEXA scans in 38 men with cirrhosis as part of an investigation of body composition which included assessment using skinfold anthropometry, BIA, neutron activation analysis and deuterium dilution. The difference in total body bone mineral between patients and matched controls was not significant, nor were there any significant changes in total body bone mineral with increasing severity of liver disease. Total body fat measured by DEXA was significantly greater in patients with Child’s grades A and B compared to the healthy controls or Child’s C patients and this difference was also detected by anthropometry and BIA. No significant differences were found between body Chapter 5 Body composition in cirrhosis 202 fat or FFM measured using DEXA, anthropometry or BIA in any group.

Neither the effects of fluid retention on attenuation nor the shielding effects of ascites have been examined. It can be concluded, therefore, that although DEXA- derived estimates of total body fat and FFM are comparable with those determined using other predictive methods of assessment, DEXA remains unvalidated in the absence of comparison with a reference method of evaluation in this patient population.

5.8 Anthropometry Anthropometry, the measurement of body dimensions, circumferences and skinfold thicknesses, can be used to estimate muscle mass and total body fat using prediction These formulae have been developed by comparing anthropometric measurements from large numbers of healthy subjects with estimates of their body composition obtained using reference techniques, for example, densitometry.

Collection of anthropometric data is quick, straightforward, and safe, requiring only simple equipment, such as skinfold calipers, a reliable tape measure, weighing scales and height measure. The observer must follow strict anatomical guidelines to maximise the accuracy of the measurements obtained but, even when these are followed, both intra- and interobserver variation may be significant^^^’^^^.

The extrapolation of body composition data from anthropometric measurements is based on the assumption that measurements from the subject being assessed will follow the same predictive pattern as that of the population in which the formulae were derived. This assumption may not always be true, due to anatomical differences, even in healthy individuals from the same population, but it is particularly vulnerable in diseased or injured patients or individuals from ethnic or age groups other than that of the reference population. Chapter 5 Body composition in cirrhosis 203

Errors of estimates of body fat derived from anthropometry range between 3 and

9 <^236,307

Anthropometry in chronic liver disease No predictive anthropometric formulae have been developed or validated in patients with chronic liver disease. Even so, anthropometry has been extensively used to assess body composition in this patient population^^'^^'^'^^'^^'^^^'^^^'^^'^^^'^^"^' 434,445,487 5.4) and the following predictive formulae have been used:

(i) Dumin and Womersley^'^^, developed regression equations from densitometric and anthropometric data in 481 healthy adults, in which body density (D) is calculated from the logarithmic sum (S) of skinfolds measured at the triceps, biceps, subscapular and suprailiac sites; percentage body fat is then calculated from body density using the Siri equation"*^^. A total of 180 different equations were developed for men and women, for each 10 year age band between 17-72 years using a combination of measurements from the four skinfold sites. Examples of equations for individuals aged 40-49 years using skinfold measurements at all four sites are:

men D = 1.1620 -0.0700 x S women D = 1.1333 - 0.0612 x S.

From these equations, the standard error in density estimations is 0.0082 for men and 0.0107 for women or, using mean density values for both subgroups, the equivalent of 4% of body fat in men and 5% in women^"^^.

(ii) Brook^^, developed similar equations from estimates of TBW and anthropometric data in 23 children aged 1-11 years, making an assumption that the hydration fraction of FFM is 73%: Chapter 5 Body composition in cirrhosis 204

boys D = 1.1690- 0.0788 x S girls D = 1.2063 - 0.0999 x S.

(iii) Bruce and colleague^^, developed regression equations to predict body fat (BF) from weight (W), height (H) and age (A) based on assessments of TBK and TBW in 476 adults:

men BF = (0.43 x W) - (0.044 x H) + (0.096 x A) - 14 women BF = (0.61 x W) - (0.23 x H) + (0.04 xA)- 15.

The standard error of the estimates of body fat were 3.8 kg for men and 3.0 kg for women.

(iv) Heymsfield and colleagues^^, developed an equation to predict total body muscle mass (MM) from estimates of mid-arm muscle area (MAMA) based on urinary creatinine excretion in 26 healthy and 42 undernourished adults, making an assumption that 1 g of creatinine is equivalent to 20 kg of muscle:

MM = H x (0.0264 + [0.0029 x AMA]).

The 95% confidence intervals of the regression equation between mid-arm muscle area and creatinine excretion were wide, with a mid-arm muscle area of 42 cm^ predicting a urinary creatinine excretion between 7.1 to 9.3 mg/cm. In a subject of 180 cm tall this predicts a muscle mass of between 25.6 and 33.5 kg.

Although the validity of these formulae has not been determined in patients with chronic liver disease, some general consideration can be given to their acceptability for use in this patient population.

The formulae of Durnin and Womersley^"^^ are associated with smaller errors than Chapter 5 Body composition in cirrhosis 205

those of Bruce et and Heymsfield et aP^ but depend on the robustness of the Siri equation"^^^ in determining body fat from density in a healthy population. The assumptions made by Siri may not be correct in some patients with chronic liver disease but are probably superior to that used in the formulae of Brook^^ where the hydration fraction is assumed to be 73% as values vary between 75-79% in healthy children aged 1-11 years^®^; similarly, the assumed relationship between urinary creatinine excretion and muscle mass used by Heymsfield et aP^ is weak^^ and was derived from a relatively small number of subjects.

The Durnin and Womersley formulae^"^^ do not use skinfold measurements on the lower limbs and waist circumferences which are included in other predictive anthropometric models^^^’^^^’^^^, and which may be adversely affected by fluid retention. However, even in the Durnin and Womersley formulae^^^, accurate measurement of suprailiac skinfolds may be difficult or precluded by abdominal distension in patients with gross ascites.

Comparisons have been made between estimates of body fat determined using anthropometry and other assessment techniques in four studies in patients with chronic liver disease^^’^^^’^^^’"^^.

Bramley and co-workers^^ estimated body fat in 53 patients with cirrhosis using anthropometry and DEXA. They found significant differences between the results obtained using the two methods.

Similarly, in the multiple technique assessment of 38 men with cirrhosis undertaken by Prijatmoko and colleagues'^^, estimates of FFM derived using anthropometry significantly underestimated those derived using combined data from neutron activation, deuterium dilution and DEXA.

The body fat estimates made by Crawford and colleagues^^^, using the formulae of Chapter 5 Body composition in cirrhosis 206

Bruce^^, in 57 patients with cirrhosis were higher than those determined using a combination of TBK and deuterium dilution. The authors did not suggest that the Bruce formula was appropriate in this patient population but used the data calculated from it to facilitate comparisons with other methods. Neither TBK or TBW have been validated in this patient population; the presence of fluid retention will give the impression of an expanded FFM and thus lead to an underestimation of fat mass. This may explain the discrepancy but, even so, the results are inconclusive as to the value of anthropometry in patients with chronic liver disease.

Madden and Morgan^^^ estimated body fat in 60 patients with cirrhosis using anthropometry and BIA. Mean values derived using anthropometry were significantly lower than those from BIA in the group as a whole (83%) and the subgroup with fluid retention (75%). Individual anthropometric values varied from 18% less to 10% more than those derived using BIA. This study illustrates the discrepancies between the results from these two methods, but as both are predictive techniques, the study does not provide validation of either.

Anthropometric derivations of muscle mass have been compared with other techniques in just one study. Pirlich and colleagues'^^, estimated muscle mass in 102 patients with cirrhosis using the anthropometric formula of Heymsfield et aP^ and 24-hour urinary creatinine excretion based on the assumption that 1 g of creatinine represents 18.5 kg of muscle^^\ The anthropometric formula of Heymsfield et aP^ was derived from comparisons of creatinine excretion but using a different assumption that 1 g of urinary creatinine represents 20 kg of muscle. The authors found muscle mass determined by 24-hour creatinine excretion was significantly greater than anthropometric values, which is the opposite of what would be expected on the basis of these discrepant assumptions alone. While the role of creatinine excretion is discussed below, no conclusions can be drawn from this study on the merit of using anthropometry in this population. Chapter 5 Body composition in cirrhosis 207

5.9 BIA The difference in electrical conductivity of fat and FFM is the principle upon which BIA is based and used to quantify these compartments^^^’'^^^. A small current is passed between transmitting and receiving electrodes placed on the skin at anatomically defined positions and the resistance and reactance to the current between the electrodes is measured by a portable machine. Equations have been devised, based on impedance data from populations of healthy individuals, to predict the water volume of the subject under investigation and this, together with information on height, weight, age and sex, is used to calculate Measurements can be made with a current at a single frequency, usually 50 kHz, or, using more sophisticated equipment, at a range of frequencies between 1-1000 kHz. The advantage of the latter is that at lower frequencies the electrical current only passes through the extracellular fluid while at higher frequencies it also penetrates the cell membranes and thus passes through both ICW and ECW, theoretically facilitating measurement of both"^^^.

BIA is practical, quick and safe to use, requiring the subject to lie supine and still while measurements are made via electrodes placed at specific sites. The observer error with BIA is less than that with anthropometry^^^ and generally less than 4% for estimating 50 1 of TBW^^.

Assessment of body compartments by BIA is based on the assumption that measures of impedance in the subject under investigation will predict body water in the same way as in the reference population. However, prediction equations have been developed in different populations and, because of physiological variation in hydration status, are influenced by age and racial factors^^'^^^'^'^^. A further assumption is made when FFM is calculated from BIA-derived estimates of body water and thus errors arise in individuals who do not fit the population- specific prediction equations^^^’^®. Chapter 5 Body composition in cirrhosis 208

The short-term precision of this technique has been reported as less than 2%^^^ while errors of 2.7% in body fat were determined in comparison with densitometry^^^.

BIA in chronic liver disease In recent years, several body composition studies have been undertaken in patients with chronic liver disease using BIA. While some have used the method simply to provide body composition data as part of wider investigations^^^’^^^’^^®’^^^’'^^^’'^^^’'^^^’^'^^, others have undertaken comparisons with different 477.569 Qj. BIA to assess changes in hydration with paracentesis^^’^®'^’^^^’^^®’'^^'^’'^^^’

566.570

In comparative studies (Table 5.5), data on TBW, ECW, BCM, fat and FFM have been derived using single frequency BIA and compared with estimates derived using various other techniques including anthropometry, DEXA, TBK, dilution studies and combinations of these methods. Although no comparisons have been made between BIA and a reference method of assessment in patients with chronic liver disease, the overall conclusion from the investigations undertaken so far is that body composition estimates derived using single frequency BIA are inaccurate in this patient population, particularly in patients with fluid retention.

Similarly, results from seven out of eight studies where the effects of paracentesis on TBW were investigated using single frequency BIA (Table 5.6), showed that this method could not accurately detect the volume of fluid removed. This is probably because the trunk accounts for approximately 8% of whole body impedance^ and thus the presence of ascites has relatively little effect on total measurements. Schloerb and colleagues'^^^, however, found a good correlation between BIA- estimated volumes of ascites removed in two patients, one of whom was studied twice, when the resistivity of the drained ascitic fluid was separately calibrated in vitro. The authors concluded that as the ascitic fluid from each patient assessed Chapter 5 Body composition in cirrhosis 209 would require separate calibration, the same results could be achieved more easily by weighing the patient before and after paracentesis.

Few studies on multifrequency BIA in patients with cirrhosis have been published to date. Borghi and colleagues^^ measured TBW and ECW in 34 patients, none of whom had clinical or ultrasonic evidence of fluid retention, using deuterium and bromide dilution studies and multifrequency BIA at 1, 5 and 100 kHz. Although the sampling time in the dilution studies was 3.5 hours after dosing, which may not have been long enough for equilibrium of the tracers, good agreement was found between the results from both methods for TBW and ECW.

Abergel and colleagues^ also used multifrequency BIA to assess TBW and ECW in 22 patients with cirrhosis, ten of whom had ascites, and compared the results with values obtained using oxygen-18 and bromide. Sampling times in the dilution study were not given. Values for both TBW and ECW did not differ significantly when assessed using BIA and dilution studies in patients without ascites. Data were not reported for patients with ascites although BIA underestimated the volume of ascitic fluid removed by paracentesis. The authors concluded that multifrequency BIA was a reliable and reproducible method for assessing fluid volumes in patients with cirrhosis only in the absence of ascites.

These initial results look interesting but multifrequency BIA requires further investigation, including comparison with reference methods of assessments, before it can be validated for use in patients with chronic liver disease.

5.10 Imaging techniques Estimates of body compartments can be made from imaging techniques including computed tomography (CT), magnetic resonance imaging (MRI) and ultrasonography Such methods are able to quantify adipose and muscle tissue and the volume of some individual organs within the body. Visual Chapter 5 Body composition in cirrhosis 210

images, which differentiate tissues of different densities, are detected in response to irradiating a section of the body with X-rays (CT) or exposing it to radiofrequency waves in the presence of a magnetic field (MRI) or ultrasound waves (US). In CT and MRI, the area representing the tissue or organ of interest can then be measured on a visual "slice" taken through the body and, the volume calculated from the distance between consecutive slices. The sum of estimates of fat thickness derived using US at specified sites can be used to predict total body fat in a similar way to that by caliper measured skinfold thicknesses.

The scanners required to undertake such assessments are usually located within hospital radiology departments and their accessibility for body composition studies is dependent on their clinical usage, costs and availability of operating expertise. The subject undergoing assessment is required to lie still and supine for periods of up to one hour and, for CT scans, is exposed to a relatively high dose of radiation of approximately 8 mSv for an abdominal scan or the equivalent of 400 chest X- rays"^^^. MRI and US scans do not use ionising radiation.

The coefficient of variation for measuring whole body fat is less than 1% for CT scans and between 5 and 1% for MRI scans^\ while that for the sum of ultrasound measurements is less than 2%^^.

Imaging techniques in chronic liver disease Only one study has been published in which body composition studies have been undertaken using imaging techniques in patients with chronic liver disease. Chiba and colleagues^^^ used ultrasound and anthropometry to assess fat and muscle stores in 10 children with advanced liver disease. Measurements of mid-arm muscle area determined anthropometrically were significantly greater than those derived using ultrasonography, even though the measurements correlated significantly. The authors concluded that ultrasound is an accurate method for assessing fat and protein stores. Further studies using a reference assessment Chapter 5 Body composition in cirrhosis 211 technique are required.

Campbell and co-workers^\ undertook serial measurements of muscle thickness on comatose patients with multiorgan failure. They concluded that serial measurements over the biceps, anterior forearm and anterior thigh may be a satisfactory method of monitoring muscle wasting in the presence of oedema. However, while this is useful for clinical purposes it does not validate the quantification of FFM in this patient population.

5.11 Near infrared Interactance The assessment technique of near infrared interactance (NIRI) involves the measurement of the optical density of reflected radiation from a beam of near infrared radiation directed at the body^^®. The reflected radiation is influenced by the characteristics of the underlying tissue. Prediction equations, which take account of the subject’s age, sex, weight, height and level of physical activity, are then used to estimate body fat.

The equipment required is portable and the technique is quick, safe and easy to undertake, making it practical for use in clinical situations.

The disadvantages of the technique include its limited validation against reference methods of assessment and its reliance upon prediction equations for estimating body composition. In particular, considerable weight is given in these equations to the physical activity of the subject^^^. The penetration of the near infrared beam is restricted to a depth of approximately Icm^^ thus limiting the accuracy of estimates made in obese individuals.

Comparisons of body fat have shown that values determined using NIRI were a mean of 2 ± 8% less than those derived by densitometry^® although, in some individuals, the discrepancy may be as high as Chapter 5 Body composition in cirrhosis 212

Near infrared interactance in chronic liver disease

There are no published studies on the use of near infrared interactance in patients with chronic liver disease.

5.12 Urinary metabolite excretion Twenty-four-hour urinary creatinine and 3-methyl histidine excretion can be used to estimate skeletal muscle mass^^^’^^^’^^^. Creatinine is a breakdown product of creatine which plays a subsidiary role in the production of ATP^"^; 98% of creatine is located in the muscle. 3-methyl histidine is produced during the catabolism of myofibrillar proteins in skeletal muscle. Both creatinine and 3-methyl histidine are exclusively excreted in the urine. Creatinine excretion is considered to be a more reliable predictor of FFM in healthy individuals than 3-methyl histidine^®.

Complete, accurate urine collections are required for assessment and three consecutive 24-hour collections are recommended to achieve a representative sample^^^. Analysis of urinary creatinine can be undertaken using the Jaffé reaction^^ while urinary 3-methyl histidine concentrations is determined using ion- exchange chromatography^^^.

The disadvantages of using urinary metabolites to estimate muscle mass include the practical difficulties of obtaining accurate urine collections and the variation in excretion observed with changes in dietary intake and chnical conditions such as renal impairment and sepsis.

The precision of analysing urinary creatinine concentrations is 1 to 2% but overall errors of estimating FFM from 24-hour urinary creatinine excretion are between 3 to 20%^^

Urinary creatinine excretion in chronic liver disease Urinary creatinine excretion has been used in a large number of studies to estimate Chapter 5 Body composition in cirrhosis 213 muscle or LEM or to provide a measure of nutritional status in patients with chronic liver disease^'^'^°^'^i'^^'^'^'^'^^'3^'3"3,427,434,474,478,^^^ validity of using this technique in this patient population has been questioned^^’^^^’"^^^.

Creatine is produced in the liver^"^ and thus its manufacture may be impaired in hepatic dysfunction^^. Zoli and colleagues^^^ measured urinary creatinine excretion in patients with cirrhosis and observed concentrations within the predicted range in those who were not malnourished while those who were depleted showed reduced urinary concentrations. However, no data were presented on the severity of liver disease in either group.

Reduced serum creatinine concentrations have been reported in patients with cirrhosis^^’^^^ but this is not always associated with a reduction in urinary excretion^^^ and may reflect changes in renal tubular excretion of creatinine rather than decreased synthesis^^^’^^^. Low urinary excretion of creatinine has been reported in patients with cirrhosis'^®^’'^^’'^^^’'^^^ but its association with muscle mass or LBM has not been investigated independently of hepatic function.

Pirlich and colleagues'^^, reported no difference in 24-hour urinary creatinine excretion in relation to the severity of liver disease and concluded that urinary creatinine excretion is independent of hepatic function and thus, in the absence of renal dysfunction, provides a good measure of muscle mass. However, although there were no significant differences in age, height or weight between patients in each of the Child’s grades, there was a decrease in mean age and increases in mean height and the proportion of men in each group with increasing severity of liver disease. It is likely that these factors contributed to urinary creatinine excretion in the patients with worse liver function but this was not examined statistically. Thus the independence of urinary creatinine excretion from hepatic function has not been proven. Chapter 5 Body composition in cirrhosis 214

Comparisons made between body composition data derived from creatinine excretion and those obtained using anthropometry, BIA and TBK have limited value because none of these methods have been validated in patients with cirrhosis (Table 5.7).

Urinary 3-methyl histidine excretion in chronic liver disease Estimates of urinary 3-methyl histidine excretion have been undertaken in patients with cirrhosis in the investigation of protein metabolism^^’^^^’'^^^’'^^'^’^^^’^^^ but none was extrapolated to quantify body compartments. The validity of this method has not been established in this patient population. It is likely to be less useful in pathological states where muscle proteolysis is altered"^^^ and thus in chronic liver disease, where there are alterations of protein metabolism^^^’^^® it would be predicted to have limited value. Further studies are required to confirm this.

5.13 Single body composition techniques in patients with chronic liver disease The assessment techniques described above have been variously validated in healthy populations but not systematically investigated in patients with chronic liver disease. With the exception of neutron activation, all the techniques considered are limited by the inherent characteristics of liver disease which contravene the assumptions upon which they are based. As a consequence, the results obtained are subject to varying and unknown degrees of error. The lack of validation of body composition techniques has been recognised in this patient population^^^’"^^^ but has not yet resulted in the development of a method that can overcome these problems. The solution to this problem may lie, therefore, not in the use of single techniques but in combinations of them.

5.14 Multi-compartment models of body composition assessment All of the assessment techniques described above, with the exception of neutron activation and DEXA, are based on a two-compartment model, that is, the compartment of interest against the rest of the body. In densitometry, for example. Chapter 5 Body composition in cirrhosis 215 these compartments are fat and FFM, while in dilution studies, TBW and dry matter. The combining of different techniques facilitates superior estimates of body composition by eliminating assumptions used in two-compartment models. However, each contributing technique must be valid for measuring the compartments concerned and resulting data must be combined appropriately.

The first three-compartment model was proposed by Siri"^^^ and combined densitometry data with estimates of TBW in order to reduce the errors caused by variation in subjects’ hydration status. The model thus comprised fat, water and protein-plus-mineral. The development of DEXA, providing assessments of total body mineral, facilitated the division of Siri’s third compartment and led to the proposal of a four-compartment model by Heymsfield and colleagues^^^.

In this model, the body is divided on a molecular leveP^ into fat, water, mineral and protein thus enabling these major body compartments to be quantified independently of abnormalities of hydration and mineralisation. The model can be applied using two different methods, one by undertaking assessments of body density, total body water and total body mineral using densitometry, isotope dilution and and the other, by assessing total body water and both total body nitrogen and calcium using neutron activation analysis^^^. The former requires less radiation exposure.

The improved accuracy with which multi-compartment models assess body composition is not compromised by a reduction in precision as the errors associated with each technique are not cumulative. The precision with which body fat is assessed by the four-compartment model, using densitometry, deuterium dilution and DEXA, is between 0.55 and 0.75

Multi-compartment models in chronic liver disease There are only two reports of body composition assessment by multi-compartment Chapter 5 Body composition in cirrhosis 216 model in this patient population, Oldroyd and colleagues'^^^ proposed a model for use in cirrhotic patients using DEXA and TBK to quantify body fat, ECW, extracellular solids and BCM. However, although this is an improvement on some single techniques, its authors advise caution in its application as the calculation of BCM from TBK may not be valid.

Prijatmoko and colleagues'^^, undertook assessments on 38 men with alcoholic cirrhosis using a multi-compartment model based on neutron activation, deuterium dilution and DEXA. However, the model applied was not that of Heymsfield and the calculations undertaken were neither referenced nor explained. In addition, post-dose samples were collected 2 hours after deuterium dosing in all patients, including those with ascites, and the values of TBW obtained must, therefore, be challenged. Consequently, it is not possible to evaluate multi-compartment modelling in this study.

Heymsfield’s four-compartment modeP^^ is applicable to patients with chronic liver disease as it overcomes the obstacles of aberrant hydration and mineralisation. However, at present, no reports have been published of such assessments in this patient population. Of the two methods described, that using densitometry, deuterium dilution and DEXA, proposed by Heymsfield and modified by Fuller and colleagues^^^, is, therefore, considered in more detail.

5.15 The four-compartment model This model, which facilitates the determination of TBW, mineral, protein and fat, is based on the density of a body being equal to the sum of the mass of its components divided by the sum of the component volumes, thus: body density = + mas/^^

volume^^^^ + + volumeP^°^^^^ + volume^^^. Chapter 5 Body composition in cirrhosis 217

By measuring total body density, water and mineral using underwater weighing, deuterium dilution and DEXA respectively, and by making a number of assumptions, described below, the unknown components can be calculated using simple algebraic equations which are defined in Table 5.8. The assumptions used are considered robust even in pathological conditions, such as cirrhosis. The results obtained, however, are untested because no superior reference method of assessment is available.

Assumptions Density of water at 36° C = 0.99371 kg/1 Total body mineral mass = ash x 1.2741 kg Total body mineral density = 3.0375 kg/1 Osseous : non-osseous mineral = 0.8191 : 0.1809 Fat^+P% = (2.7474 / density^+P) - 2.0503 where fat^'^P% is the percentage of fat in the combined "fat and protein", densit/'^P is the combined density of "fat and protein" Density of fat = 0.9007 kg/1 Density of protein = 1.3400 kg/1

Limitations Glycogen is not measured separately in this model and, because its density, 1.5 kg/1, is similar to that of protein, it will be included in the protein compartment. Glycogen comprises approximately 0.5 to 0.7 kg in a fasting, healthy 70 kg individual^, but in cirrhosis, both liver and skeletal muscle stores of glycogen stores are depleted by approximately 30%^^^'"*°^. The error in overestimating protein will thus be approximately 0.4 to 0.5 kg.

Similarly, nucleic acids, with a density of 1.30 kg/1, are not separately measured and will also be included in the protein compartment. As they comprise approximately 0.2 kg in a 70 kg man, they represent a further but small Chapter 5 Body composition in cirrhosis 218 overestimation of body protein.

The ratio of osseous to non-osseous mineral is assumed to be constant. However, in the four-compartment model, changes of up to 50% in osseous mineral will result in errors in body fat of less than 0.5% of body weight, while changes in non- osseous mineral, which comprises only 18% of total mineral mass, will be even less^^.

While these limitations exist, they are small, and it is likely that this model potentially provides the most accurate method available for assessing body composition in patients with chronic liver disease. Two practical issue, however, must be addressed before such studies can be undertaken in this patient population. Firstly, to assess the gastrointestinal gas volume of cirrhotic patients taking lactulose or lactitol so that density measurements can be undertaken with confidence, and secondly, to establish the optimum time after oral deuterium dosing for collecting post-dose samples, ensuring equal distribution throughout TBW in patients with fluid retention.

5.16 STUDY 5: THE MEASUREMENT OF GASTROINTESTINAL GAS

5.16.1 Aims (i) To investigate the prevalence of gastrointestinal side-effects of patients with cirrhosis taking non-absorbable disaccharides; (ii) to assess the volume of gastrointestinal gas in these patients using computerized tomographic scanning.

5.16.2 Patients and methods Subjective assessment The study population comprised 14 patients with cirrhosis, nine of whom, six men and three women, of mean (range) age 52.6 (25-70) years, were on long-term Chapter 5 Body composition in cirrhosis 219 treatment with non-absorbable disaccharides (lactulose or lactitol); the remaining patients, three men and two women, of mean age 48.6 (35-65) years, were not taking this medication (Table 5.9).

Seven of the treated patients had compensated liver disease (Pugh’s score ^7)"^^, while the remaining two had stable but decompensated cirrhosis (Pugh’s score ^8). All five untreated patients had compensated cirrhosis. None of the patients had clinically detectable ascites.

All patients were asked to keep symptom diaries in which they recorded their subjective assessment of flatulence and abdominal bloating at defined intervals over one 6-month period. Patients assigned a symptom score denoting the presence or absence of these symptoms for each day during the periods of record keeping. A symptom score of zero denoted the absence of the symptom while scores of one, two and three indicated mild, moderate and severe symptoms respectively. Scores were recorded on a mean of 106 (81-131) days by treated patients and on 96 (76- 131) days by those who were untreated. All patients were monitored every two to three weeks in the outpatients department and their symptom diaries and medication reviewed.

A mean daily score was calculated for each patient for both flatulence and abdominal bloating and mean scores compared between the treatment and non- treatment groups.

Objective assessment 1 The study population comprised six men with alcoholic cirrhosis of mean (range) age, 56.5 (46-66) years who had been abstinent from alcohol for a mean of 20.8 (4- 43) months. Five of the patients had compensated liver disease (Pugh’s score ^7), while the remaining patient had stable but decompensated cirrhosis (Pugh’s score 9); none of the patients had clinically detectable ascites. All were taking a Chapter 5 Body composition in cirrhosis 220 prescribed daily dose of either lactulose (20-35 ml) or lactitol (10-40 g) for management of portal systemic encephalopathy (Table 5.10).

The height and weight of each patient was recorded.

All patients underwent a spiral CT scan of the abdomen, as a routine screening procedure for hepato-cellular carcinoma. This was undertaken using an IGE Hi Speed Advantage CT Scanner (IGE Medical Systems, Milwaukee, USA). Patients were scanned from the xiphisternum to the iliac crest, using a slice thickness of 10 mm with image reconstruction at 10 mm intervals (1:1 pitch) (Figure 5.1). The ten slices immediately above the iliac crest were selected for assessment of gas volume by first highlighting all the pixels within the range of -200 to -2000 Hounsfield units to measure the total area of gas in mm^ in each slice. The volume of gas was then calculated by multiplying the sum of the total area of gas in all ten slices by 10, the slice thickness being 10 mm.

The measurement procedure was straightforward and undertaken by a senior radiographer taking approximately 50 minutes per patient analysed.

Abdominal spiral CT scans, undertaken for clinical purposes in six male control patients of mean age, 58.8 (40-72) years with no history or clinical evidence of liver or colonic disease, were selected from archived records for comparison with the study group (Table 5.10). The retrieved films were processed identically and gas volumes calculated as above. The medical notes of each control patient were reviewed to determine their height, weight and diagnosis, and to confirm, with the assistance of records from the Department of Pharmacy, that they were not taking a non-absorbable disaccharide.

The gas volume determined in both patient groups represented only that present in the gastrointestinal tract lying 10 cm above the iliac crest, subsequently referred Chapter 5 Body composition in cirrhosis 221 to as the middle abdomen; gas in the gastrointestinal tract outside this region was not quantified. Comparisons were made between the ages, height, weight and volume of gastrointestinal gas in the two groups using Mann-Whitney U-tests.

Objective assessment 2 The study population comprised one male and one female patient with cirrhosis aged 35 and 57 years respectively. Cirrhosis was secondary to alcohol in the male patient but of unknown aetiological origin in the woman. Both patients had Pugh’s scores of 8 and no significant ascites; both were also taking 30 ml lactulose daily.

Both patients underwent an abdominal and pelvic spiral CT scan as part of the clinical investigation of their disease. Patients were scanned from the xiphisternum to the bifurcation of the aorta using the same scanner and method described above. The volume of gas was calculated as before, but in order to provide an estimation of total gastrointestinal gas, all slices, rather than just the 10 above the iliac crest, were included. These were arbitrarily subdivided into (i) the upper abdomen - between the diaphragm and 10 cm above the iliac crest; (ii) the middle abdomen - between 10 cm above the iliac crest and the iliac crest; and (iii) the lower abdomen - between the iliac crest and the bifurcation of the aorta.

Abdominal and pelvic spiral CT scans, undertaken for clinical purposes, in one male and one female control patient aged 40 and 57 years respectively with no history or clinical evidence of liver or colonic disease, were selected from archived records for comparison with the study patients and the volume of gastrointestinal gas determined as described.

The total volume of gastrointestinal gas determined in the two patients with cirrhosis and the two control patients were compared. Chapter 5 Body composition in cirrhosis 222

5.16.3 Results Subjective assessment All patients remained well throughout the study period with a mean change in Pugh score of less than one in both groups.

The median score for flatulence was higher in patients with cirrhosis who were taking a non-absorbable disaccharide than in those who were not (0.61 [0-1.49] vs 0.23 [0.02-1.02]) but median scores for abdominal bloating were similar in both groups (0.01 [0-2.05] vs 0.03 [0-1.11]) (Table 5.9). No relationship was observed between the medication dose and symptom scores.

Objective assessment 1 The two patient groups were well matched with no significant differences between their age, height or weight. The median volume of gastrointestinal gas in the middle abdomen was not significantly different in the cirrhotic or control patients (81.5 [15-152] vs 57.0 [19-332] ml)(Table 5.11).

Objective assessment 2 The age, height and weight of the patients with cirrhosis and the control patients were comparable (Table 5.12). The total volume of gas within the gastrointestinal tract was 376 and 87 ml in the patients with cirrhosis prescribed lactulose compared to 366 and 190 ml in the control patients.

The proportion of gas located in the upper, middle and lower regions of the abdomen varied in all four patients. In both patients with cirrhosis prescribed lactulose, the smallest proportion of gas was located in the upper region with 55 and 36% located in the middle and 28 and 63% in the lower region (Table 5.12). By comparison, one control patient had the highest proportion of gas in the upper region while in the other, the highest proportion was located in the lower region. Chapter 5 Body composition in cirrhosis 223

5.16.4 Discussion Twenty to 25% of patients treated with non-absorbable disaccharides experience flatulence and abdominal pain^^^. In the subjective assessment of the present study, the median symptom score for flatulence in patients with cirrhosis taking a non­ absorbable disaccharide was higher than in the control patients and the three patients who reported the highest flatulence scores were all taking this medication.

Flatulence and abdominal pain have been described as often resolving after several days of continued administration of non-absorbable disaccharides^^^; the patients taking this medication in the present study had been doing so on a long-term basis. Resolution of symptoms could either suggest that production of gastrointestinal gas from non-absorbable disaccharides decreases with continuation of treatment or that patients become habituated to its effects. Clarification of this point is not possible from a subjective assessment of symptoms.

The results from the subjective assessment are suggestive of an increase in gastrointestinal gas with non-absorbable disaccharides but not conclusive. Therefore, quantification of gas volume was considered essential.

Previous studies quantifying gastrointestinal gas Volumetric assessment of gastrointestinal gas was first undertaken to investigate abdominal pain and discomfort in sailors and airmen subjected to explosive decompression^^^. Although volumes of egested flatus have been measured since the beginning of the 20th century^^^, one of the first assessments of gastrointestinal gas volume"^^ was undertaken by plethysmography in 1947 using a manometer situated in the stomach of a subject sitting in an airtight chamber. The pressure change arising from a maximum expiratory effort with the glottis open was then measured in six healthy men (Table 5.13). Subsequent work has shown that the volume determined, 960 ml, was an overestimation and that the methodology used was flawed because the subjects were unable to maintain a constant intrathoracic Chapter 5 Body composition in cirrhosis 224 pressure^^^.

Bedell and colleagues^^ also used a plethysmographic method in an airtight chamber of approximately 600 1 volume but improved the technique by taking measurements when the subject had stopped breathing at the end of a normal expiration rather than at the end of a full expiration thus avoiding the cooling and condensing of respiratory gas. They reported a mean abdominal gas volume of 115 ml in 13 healthy volunteers and 47 patients with lung disease.

Greenwald and co-workers^°^ reported a similar mean gas volume of 111 ml in healthy volunteers assessed using a displacement volumeter filled with partially degassed water and situated within a large altitude chamber. The changes in abdominal pressure were measured by manometry, at the end of expiration, in response to atmospheric decompression.

Gastrointestinal gas was also measured and qualitatively analysed in 11 healthy young men in a study by Levitt using an argon flush technique^^^. The mean volume determined was 90 (30-200) ml. Although the technique is considered to provide accurate results, the practical aspects of flushing argon via a nasogastric tube into the duodenum, collecting gases passed per rectum and then analysing them precludes its use in most investigative situations (Michael Levitt, personal communication, 1995).

Gastrointestinal gas has also been evaluated using radiological techniques. Keys and Brozek^^^ undertook assessments by measuring the visible areas of gas on an abdominal X-ray following a barium meal. Their estimated mean volume of 28 ml is probably an underestimate because coils of intestine may obscure the detection of gas in a plain abdominal X-ray (Robert Dick, personal communication, 1996).

Chami and colleagues^^ used a plain abdominal X-ray to determine bowel gas area Chapter 5 Body composition in cirrhosis 225 and showed that the positioning of the patient affected the result with a significantly greater mean area observed in the supine than the erect position. They postulated that the flattening of gas bubbles in the supine position caused a reduction in depth and thus an increase in cross-sectional area. They speculated that a three-dimensional technique, for example spiral CT, could provide a more accurate method of quantification.

The present study In the first objective assessment, similar volumes of gastrointestinal gas were observed in the middle abdomen of patients with cirrhosis taking non-absorbable disaccharides and untreated control patients. The wide range of volumes observed in both patient groups reflects values reported previously in healthy individuals^^’^^^. However, although these results confirm that the volume of gas within the middle region of the abdomen is not significantly increased with non-absorbable disaccharides, the results cannot be extrapolated to reflect total volume.

It could be speculated that if more gas is being produced within the gastrointestinal tract, it is likely to be located in the lower abdomen as it travels towards the rectum for egestion, although it must be noted that the lower abdomen is not synonymous with the lower gastrointestinal tract. It could also be argued that gas will rise because it is less dense and will collect in the splenic and hepatic flexures if the subjects is erect, but be more equally distributed throughout the abdomen if supine.

The results from the second objective assessment showed no obvious pattern of gas distribution in the three regions of the abdomen and, overall, variable but comparable volumes in the two patients taking lactulose and the two who were not. In particular, the results provide no evidence to suggest that the volume or proportion of volume of gas in the upper and lower abdomen in the patients taking lactulose were excessive. The volumes determined in all four patients were within Chapter 5 Body composition in cirrhosis 226 the range quantified previously in healthy adults using plethysmography^^; in one patient with cirrhosis taking lactulose and one control patient the gas volume exceeded the range determined using the argon washout technique^^^. However, the number of patients studied was small, limiting the conclusions that can be drawn.

All the CT scans examined in the present study had been undertaken for clinical purposes. The dose of radioactivity from an abdominal and pelvic spiral CT, 12 mSv (John Wood, personal communication, 1995) or the equivalent of 600 chest X-rays"^^^, precludes its use for research purposes alone. Ideally, the second objective study should have included a greater number of patients taking lactulose or lactitol and comparisons should have been made against healthy volunteers. However, the patients with cirrhosis studied were the only two out of approximately 400 with cirrhosis who were screened and found to have undergone abdominal and pelvic spiral CT and to be taking long-term non-absorbable disaccharides. It would be unethical to expose other patients or healthy volunteers to the radiation doses associated with spiral CT simply for research purposes.

The control patients, in both objective assessments, were selected from archived records to match the demography, height and weight of the patients taking lactulose or lactitol and care was taken to avoid individuals with gastrointestinal pathology. It is considered that the gas volumes determined from these patients are likely to reflect that in healthy individuals.

Study conclusions Patients with cirrhosis on long-term medication with non-absorbable disaccharides have more gastrointestinal symptoms than similar but untreated patients. Objective measurements of gastrointestinal gas volume by spiral CT were comparable in patients taking lactulose and lactitol and untreated non-cirrhotic patients. This suggests that the use of a standard correction value for gastrointestinal gas in Chapter 5 Body composition in cirrhosis 227 underwater weighing calculations will not be associated with greater errors in patients with cirrhosis on this medication than in other subjects.

5.17 STUDY 6: AN EVALUATION OF SAMPLING TIMES FOR DILUTION STUDIES IN PATIENTS WITH CIRRHOSIS

5.17.1 Aim To determine the distribution of deuterium in serum, saliva and ascitic fluid in patients with cirrhosis in order to determine optimum sampling times when determining total body water in this patient population.

5.17.2 Patients and methods The patient population comprised six men of mean (range) age 53.8 (37-68) years, with alcoholic cirrhosis (Table 5.14). Three patients had no clinical evidence of fluid retention while the remaining three had gross ascites. At the time of study all the patients were clinically stable with no history in the preceding month of gastrointestinal bleeding, infection or severe encephalopathy. Four patients had been abstinent from alcohol for between 1 and 25 months at the start of the study while two patients continued to consume alcohol until the day before the study.

Written informed consent was obtained from all patients following explanation of the procedures, permission for which had been granted by the Ethical Practices Sub-Committee of the Royal Free Hospital.

Anthropometry: Height and weight were measured using standardized procedures and, in the three patients without clinically detectable fluid retention, BMI calculated. In the three patients with ascites, BMI was calculated from a dry weight estimated from previously documented weights and clinical assessment^^. MAC and skinfold thicknesses at the triceps, biceps, subscapular and suprailiac Chapter 5 Body composition in cirrhosis 228

sites were measured using standardized procedures^"^^’^"^^. MAMC^^^ and percentage body fat^"^^ were then calculated.

Dilution studies: After an overnight fast, baseline samples of saliva and serum were collected from all patients and, in addition, ascitic fluid was sampled from the three patients with ascites. Saliva was collected, before patients cleaned their teeth, using small balls of cotton wool which were chewed by the patient and expectorated into a small, tightly-lidded specimen jar. The cotton wool was transferred to a 10 ml syringe and squeezed by the plunger to yield a saliva sample. Blood was drawn from indwelling intravenous cannulas specifically sited for the study and kept open for the duration of the investigation by flushing with 5 ml of sterile water. A 5 ml blood sample was withdrawn from the cannula and discarded before collecting each study specimen in Monovette serum gel tubes (Sarstedt, Germany). These were centrifuged at 3500 rpm for 15 minutes and the serum separated. Samples of ascitic fluid were obtained by transabdominal needle aspiration, centrifuged at 3500 rpm for 15 minutes and the supernatant separated. All samples of saliva, serum and ascitic fluid were stored in tightly lidded plastic tubes at -20° C until analysis.

After obtaining baseline samples, an accurately weighed dose of deuterium oxide (0.40 g ^H20 / kg body weight) was administered by mouth. The dosing solution was prepared by adding 400 g deuterium oxide (99.9 atom %) (Sigma Chemical Company, Poole, UK) to 523 g deionised water. After dosing, the patients drank approximately 50 ml of tap water to minimise deuterium losses from the mouth. Further samples of saliva, blood and, where applicable, ascitic fluid were collected at 2 hourly intervals until 12 hours after dosing. Patients were permitted to eat and drink as required but took nothing orally in the 30 minutes before each sampling period in order to minimise contamination of saliva. A record of the patients’ fluid intake and output was kept throughout.

Sample deuterium enrichment: The deuterium enrichment of all samples and Chapter 5 Body composition in cirrhosis 229 diluted dosing solutions was measured using Fourier transform infrared spectroscopy.

A standard solution of deuterium was prepared by accurately weighing 250 g deionised water and adding 250 ptg deuterium oxide. The concentration, checked by mass spectroscopy, was 988 ppm deuterium.

The infrared spectrometer, (ATI Mattson Genesis Series FTIR, ATI, Cambridge,

Uk )247 calibrated by scanning the standard solution of deuterium against a background of deionised water and measuring absorption at a wavelength of 2500 cm'k The absorption of post-dose samples, scanned against a background of the corresponding pre-dose sample, was then measured. The apparent deuterium enrichment of each sample was calculated by comparison with the absorption of the standard solution:

D 2 O enrichment - absorption^^^^^^ x 988 absorption^^^^^^^^ where 988 was the enrichment of the standard solution. However, unlike the standard solution, the samples of saliva, serum and ascitic fluid did not contain 100% water. Their absorption was measured through a physical distance of fluid within the spectrometer cuvette, and thus a correction was required to adjust for the solid matter within a given volume of fluid. Published data are available for saliva and serum^^^’^^^, but it was necessary to determine an appropriate value for ascitic fluid by measuring its density and hydration fraction.

Sample density: A newly calibrated 1 ml pipette was used to measure repeated aliquots of deionised water and all samples of ascitic fluid which were weighed to 0.0001 g. The relative density of ascitic fluid was determined by comparison of its weights with those of water and a mean value determined for each patient. The Chapter 5 Body composition in cirrhosis 230

true density of ascitic fluid was calculated by multiplying the relative density by 0.99823, the relative density of water at 20°

Sample hydration fraction: One millilitre aliquots of all samples of ascitic fluid were accurately weighed, freeze dried to constant weight (Modulyo 4K, Edwards, Crawley, UK) and re-weighed. The mean hydration fraction of ascitic fluid was then calculated from the weight lost from each sample.

Calculations: Published values of 0.994 and 0.945 g/ml^^^’^^^ were used to adjust for the water content of saliva and serum respectively. For ascitic fluid, a value was calculated from the relative density (RD) and hydration fraction (HP);

CF (g/ml) = RDx 0.99823 x HF.

The apparent deuterium enrichment of each sample was then divided by the appropriate correction factor to give a true enrichment.

The intraobserver repeatability of the sample absorption was evaluated by re­ measurement of five samples.

5.17.3 Statistical analysis A repeatability coefficient'^^ was calculated for sample absorption and expressed as a percentage of the mean of the repeated values. To examine the distribution of deuterium throughout the different body fluids, comparisons were made between enrichment by calculating the difference, expressed as a percentage of the mean enrichment, for each pair of fluids in each patient separately. A difference of less than 2% between samples was considered desirable, indicating that deuterium enrichment was equivalent in the fluids measured. To observe the changes in deuterium enrichment, the values in each fluid was plotted graphically against time for each patient (Figure 5.2)^^. The optimum time for collecting post-dose samples Chapter 5 Body composition in cirrhosis 231 was considered to be when deuterium enrichment reached a plateau and differed by less than 2% in the fluids measured.

5.17.4 Results Anthropometric variables showed a wide variation in body size and nutritional status among the six patients studied (Table 5.14).

The intraobserver repeatability coefficient for sample absorption was 0.00581 or 0.9% of the mean values.

The methodological variations in repeated estimates of relative density and hydration fraction of ascitic fluid were between 0.06 - 0.90% and 0.10 - 1.40% of the respective standard deviations of the means. The inter- and intrapatient variation did not exceed the methodological variation for either relative density or hydration fraction and, thus, respective mean values of 1.0096 and 0.9738 g/g were determined and, from these, a correction factor of 0.9814 calculated for ascitic fluid.

Pre-dose samples of saliva and serum were obtained from all patients. Six post­ dose samples of saliva and serum were obtained from all patients with the exception of a 2 hour saliva sample from patient 4. Pre-dose and six post-dose samples of ascitic fluid were obtained from patients 4, 5 and 6 with the exception of the 4 and 10 hour samples from patient 5.

In the three patients with no clinical evidence of fluid retention, the smallest difference between deuterium enrichment in saliva and serum was observed at 6 hours after dosing and ranged from 0 to 0.6% (Table 5.15).

In patients with gross ascites, the differences observed between deuterium enrichment of saliva, serum and ascitic fluid were considerable; the smallest mean Chapter 5 Body composition in cirrhosis 232

(range) difference between saliva and serum was 5.0 (3.1-7.4) % and was observed 2 hours after dosing, between saliva and ascitic fluid was 1.8 (1.1-2.4) % 10 hours after dosing, and between serum and ascitic fluid was 5.0 (2.6-S.4) % 8 hours after dosing (Table 5.15). At no time point was the difference in deuterium emichment less than 2% in all pairs of fluids sampled in all three patients.

The increase in deuterium emichment of saliva and serum after oral administration followed a similar pattern in all patients studied (Figure 5.2) with a rapid increase in the first two hours succeeded by much smaller changes over the subsequent 10 hours. However, the increase in emichment of ascitic fluid was slower, reaching maximum values at 4, 6 and 12 hours after dosing in patients 4, 5 and 6 respectively, and was accompanied by small decreases in saliva and serum emichment after 2 hours followed by a plateau from 6 hours to the end of the study.

5.17.5 Discussion In healthy individuals, equilibration of deuterium in total body water has been reported to occur 2 to 3 hours after oral administration^^"^. Deuterium dilution has been used to assess TEW in a number of studies of patients with but with few published data to support sampling times.

Previous deuterium dilution studies in cirrhosis In an early study, O’Meara and colleagues'^^ used a falling drop technique to estimate deuterium emichment in serum and ascitic fluid at hourly intervals up to 6 hours after dosing in nine patients with cirrhosis. Although these patients were described as fluid retaining, the degree of fluid retention was not reported. Neither was a correction made for the difference in water content of the fluids. The ratio of deuterium emichment in ascitic fluid to that in serum increased from 79% at 2 hours to 98% at 6 hours and the authors concluded that between 4.5 and 6.0 hours Chapter 5 Body composition in cirrhosis 233

are required for the distribution of deuterium to stabilise in patients with cirrhosis who are retaining fluid.

Moore and colleagues^^^, used similar sampling times to assess TBW using deuterium dilution in three women with cirrhosis and ascites but sampled only serum and did not report changes in deuterium emichment with time.

In two more recent studies^^"^’^^^, single urine samples, collected 4 to 6 hours after an oral deuterium dose, were used to estimate TBW in patients with end-stage liver disease, some of whom had gross ascites. However, no additional data on isotope equilibration were reported.

Prijatmoko and colleagues'^^ undertook deuterium dilution studies in 38 patients with cirrhosis, an unknown proportion of whom had ascites. They justified their collection of blood samples 2 hours after oral dosing on the basis of preliminary experiments in which deuterium equilibration occurred within 2 hours of dosing in patients with ascites. These data were not reported but appear contrary to the results from the current study.

Borghi and colleagues^^ also quoted preliminary studies where deuterium concentrations reached equilibrium 2.5 hours after oral administration in five cirrhotic patients with no clinical evidence of fluid retention. However, no data were published to support this finding.

Zillikens and colleagues^^^ investigated deuterium equilibration in eight cirrhotic patients with variable amounts of ascites and sampled plasma at 2.5, 3.0, 3.5, 4.0, 4.5 and 5.0 hours after oral dosing. They also sampled ascitic fluid in five patients between 4.5 and 7.0 hours after dosing. Plasma deuterium concentrations had stabilised at 2.5 hours after dosing. The single samples of ascitic fluid, taken at least two hours after this, showed deuterium concentrations comparable to those Chapter 5 Body composition in cirrhosis 234

of plasma collected at the same time. However, ascitic fluid was not sampled at 2.5 hours after dosing, therefore, equilibration can not be confirmed at this time.

The present study Data show that the time required for oral deuterium to equilibrate throughout TBW in patients with cirrhosis exceeds the 2 to 3 hours recommended for healthy individuals^^'^. Indeed the optimum sampling time even in patients without evidence of fluid retention is 6 hours after dosing. At this time the deuterium enrichment of saliva and serum is stable with minimal differences between the two fluids.

However, in patients with ascites, the differences in deuterium emichment of saliva, serum and ascitic fluid were considerable even at 12 hours after dosing indicating that equilibration had not occurred. Although collecting post-dose samples after 6 hours may be expected to provide a better evaluation of deuterium distribution, the differences in enrichment observed between saliva, serum and ascitic fluid were still too great to suggest that deuterium was distributed equally throughout TBW. Thus, for example, estimates of TBW based on the saliva, serum and ascitic fluid data at 10 hours after dosing in patient 6 would provide estimates of total body water of 61.9, 57.3 and 63.4 kg respectively. Clearly, these results are unacceptable.

Ascitic fluid originates from both hepatic and intestinal lymph^^ and, in patients with between 6 and 17 1 of ascites, approximately 120 1 of water are exchanged daily between the ascitic fluid and central fluid compartments^®^. This represents a mean daily water flux of 13 times the ascitic volume with an observed range of 6.5 to 17.2. It may be possible to use mathematical modelling of the deuterium emichment data to provide reliable estimates of TBW in this subgroup of patients. However, further studies continued over longer periods of time and possibly using additional tracers, for example methylene blue^^®, to estimate ascitic volume, are required. Chapter 5 Body composition in cirrhosis 235

Study conclusions Deuterium emichment of saliva and serum is comparable and stable approximately 6 hours after oral dosing in patients with cirrhosis without clinically detectable fluid retention and this is the optimum time at which post-dose samples should be collected. However, in patients with ascites, comparable levels of deuterium emichment do not occur in saliva, serum and ascitic fluid even 12 hours after dosing and, therefore, meaningful estimates of TBW cannot be determined using a simple dilution model in this patient population.

5.18 STUDY 7: ASSESSMENT OF BODY COMPOSITION IN PATIENTS WITH CIRRHOSIS USING A FOUR-COMPARTMENT MODEL

5.18.1 Aims To assess body composition in patients with cirrhosis using a four-compartment model and to compare the results obtained with those determined using single assessment techniques.

5.18.2 Patients and methods Ten men and ten women with biopsy-proven alcoholic cirrhosis (mean [range] age 48.6 [26-67] years) were recruited for the study (Table 5.16). All were clinically stable and none had actively abused alcohol or suffered a gastrointestinal bleed within the preceding 3 months. No patient gave a history of major gastrointestinal surgery, overt encephalopathy, gross ascites or any significant anatomical abnormality, for example limb amputation. All were moderately mobile and considered physically fit enough to undergo the procedures required. Approval for the study was granted by the Ethical Practices Sub-Committee of the Royal Free Hospital and The Ethical Committee of the Dunn Clinical Nutrition Centre, and all patients gave their informed written consent. Chapter 5 Body composition in cirrhosis 236

Study outline Body composition was assessed in all patients using the four-compartment model of Fuller et al^^^ in which data from underwater weighing, deuterium dilution and DEXA assessments of whole body mineral were combined to provide the best estimates of body compartments currently available. Estimates were also made using single techniques using data obtained independently from underwater weighing, deuterium dilution, two DEXA scans and skinfold anthropometry. All investigations on a single patient were undertaken within a 24 hour period. Initial evaluation, deuterium dosing and anthropometric evaluation were undertaken early in the morning at the Royal Free Hospital (RFH) in London before transfer of patients to the Dunn Clinical Nutrition Centre (DCNC) in Cambridge for underwater weighing and one DEXA scan (DEXA^). A second DEXA scan was undertaken at the Royal Free Hospital (DEXA^) during either the late afternoon before or the early morning after the main study day.

Patients attended for assessment between 07.00 and 09.00 hours after fasting for a minimum of 8 hours. A small snack of 30 g biscuits and 250 ml orange juice, providing a total of 200 kcal and 3 g protein, was given one hour after starting the study; patients then remained nil by mouth until completion of the underwater weighing and DEXA scan when a sandwich lunch was provided. Food and fluid intake were monitored but umestricted until the collection of the last samples for the dilution studies.

Underwater weighing Body volume and density was assessed by the technique of Akers and Buskirk^^, using a helium-dilution technique to measure lung volume^^'^. Patients, wearing a close-fitting swimming costume, sat on the seat of a weighing frame with integral foot rest and handles, within a plastic tank containing neck-deep water at 32-36° C (Figure 5.3). The weighing frame was suspended on three load cells connected to a visual display. A weight belt, bearing one or two 1 kg lead weights, was Chapter 5 Body composition in cirrhosis 237 attached to the waist of all female patients and male patients with good stores of adipose tissue to assist submergence. The patients were able to submerge totally by leaning forward without being restrained, while retaining control of their position. After familiarization with the submerging procedure, a nose clip and mouth-piece were positioned, the latter connected via tubing, which was initially open to room air, to a spirometer (PK Morgan, Gillingham, UK) pre-filled with measured concentrations of air, oxygen and helium. Patients then submerged totally and breathed out holding their breath for a count of five before slowly sitting up in the water and continuing to breathe through the mouthpiece. The underwater weight was recorded as the water calmed towards the end of a count of five and, at the exact moment of weighing, the connection from the patient’s breathing tube was switched from room air to the spirometer facilitating, by dilution of the helium, measurement of respiratory volume. The patient continued to breathe through the spirometer system until the helium concentration stabilized, denoting equilibrium. During the period of equilibrium, carbon dioxide was removed from expired gas by passage through calcium oxide, while the total volume of gas within the system was maintained by the controlled addition of oxygen.

A tare weight was obtained for the swimming costume, nose clip and, where appropriate, the weight belt. The dead space of the spirometer and connection hoses were separately measured by helium dilution and the temperature of the tank water and spirometer and barometric pressure were also recorded.

Body density, calculated from the underwater weighing data, was used to calculate body fat and FFM in both the four-compartment model of Fuller^^^ and the two- compartment model of Siri"^^^.

Deuterium dilution Samples of saliva were collected from all patients using small balls of cotton wool which were chewed and expectorated into a small, tightly-lidded specimen jar. The Chapter 5 Body composition in cirrhosis 238

saliva was extracted by squeezing the cotton wool in a 10 ml syringe and then stored in small, tightly-lidded plastic tubes at -20° C until analysis.

After obtaining baseline samples, an accurately weighed dose of deuterium oxide (0.40 g ^H20 / kg body weight) was administered by mouth followed by 150 ml of tap water. The oral solution was prepared by adding 400 g of deuterium oxide (99.9 atom %) (Sigma Chemical Company, Poole, UK) to 523 g of distilled deionised water. Further samples of saliva were taken from all patients at 6 hours after dosing ensuring that the mouth had remained empty for at least 30 minutes before collection.

The deuterium emichment of saliva and diluted dosing solutions was measured using Fourier transform infrared spectroscopy (ATI Mattson Genesis Series FTIR, ATI, Cambridge, UK)^^. The machine was calibrated by scanning a standard solution of deuterium against a background of distilled, deionised water and absorption at 2500 cm'^ was measured. The absorption of post-dose samples, scanned against a background of the corresponding pre-dose sample, was then measured. The deuterium emichment of the post-dose samples was calculated by comparing its absorption with that of the standard solution and, from this, TBW calculated.

TBW was used to calculate body fat and FFM using the four-compartment model of Fuller^^^ and the two-compartment of McCance and Widdowson^^^.

DEXA scanning Patients underwent whole body DEXA scanning while lying supine on the scanning table wearing a cotton gown. DBXA^ was undertaken using a Hologic QDR- 1000/W scanner and analysed using the Enhanced Whole Body V5.61 programme, while DEXA^ was undertaken on a Hologic QDR-4500W scanner with Whole Body V810a:5 analysis (Hologic, Waltham MA, USA). Chapter 5 Body composition in cirrhosis 239

DEXA data provided estimates of total body mineral, which were used to calculate body fat and FFM in the four-compartment model of Fuller^^^, and also provided direct estimates of body fat.

Anthropometric assessment Height and weight were measured to the nearest 0.5 cm and 0.1 kg using standardized procedures and equipment with patients wearing a swimming costume (Seca, Hamburg, Germany). Skinfold thicknesses were measured at the triceps, biceps, subscapular and suprailiac sites by a single observer using standardized procedures^"^^’^"^^.

Percentage body fat was calculated from the logarithmic sum of the four skinfold thicknesses using the formulae of Durnin and Womersley^"^^.

Calculations and assumptions: (i) Four-compartment model a) Body density was calculated from the patient’s body weight and volume, where volume is calculated from underwater weight and corrections made for respiratory and gastrointestinal tract gas at body temperature and pressure and temperature and density of the displaced water (Figure 5.4). b) TBW was calculated using the formula: TBW = [(Q X d)/CJ X 1.04/0.994 where is the deuterium enrichment of the dosing solution, d is the weight

of the dose given, C 2 is the deuterium emichment of the 6-hour sample, 1.04 is a correction for losses of deuterium outside the TBW space and 0.994 is a correction for the water content of saliva^^^. c) Total body mineral was derived directly from the whole body DEXA scan.

The data on total body density, water and mineral were then entered into the equations described in Table 5.8, together with body weight, and from these, body fat (%), FFM (kg) and density and hydration fraction of FFM were calculated. Chapter 5 Body composition in cirrhosis 240

Calculations and assumptions (ii) Single techniques Body fat (BF) was calculated as a percentage from each of the four single assessment techniques using the following assumptions and formulae: a) Underwater weighing (provides body density data): BF % = ([4.95/body density] - 4.5) x Assuming that the densities of fat and FFM are 0.9 and 1.1 kg/1 respectively"^^^. b) Deuterium dilution (provides TBW): BF % = (body weight - [TBW/0.732]) x 100 / body weight; Assuming that FFM = 73.2% water"^^°. c) DEXA (provides total body fat directly): BF (%) was calculated using the Enhanced Whole Body V5.61 and Whole Body V810a:5 analysis programmes. Assuming that the ratio of low to high energy attenuation by fat and FFM is constant and not influenced by changes in hydration of FFM^^; d) Skinfold anthropometry (provides skinfold thicknesses): D = c - (m X log skinfold sum), where D is body density and c and m are age and sex dependent variables; BF (%) = ([4.95/D] - 4.5) x 100^^^ Assuming that body density is proportional to the logarithmic sum of skinfold thicknesses at the triceps, biceps, subscapular and suprailiac sites^"^^ and the densities of fat and FFM are 0.9 and 1.1 kg/1 respectively"^^^.

For each of the single techniques, FFM was calculated from body fat (%) and body weight:

FFM (kg) = body weight - (body weight x BF %)/100.

5.18.3 Statistical analysis Statistical analysis was performed using Microsoft Excel 5 software package (1993, Chapter 5 Body composition in cirrhosis 241

Redmond, WA, USA). The bias and 95% (± 2SD) limits of agreement between results from the four-compartment model and each of the single assessment techniques were calculated"^^. Values for bias for FFM are equal and opposite to those for total body fat, and the limits of agreement are equal for both. Comparisons were made between the results obtained using the four-compartment model and each of the single assessment techniques using paired Student’s t-tests.

5.18.4 Results A complete set of data was obtained using the four-compartment model and four single assessment techniques in all 20 patients.

Bias was observed between data derived using the four-compartment model and the four single assessment techniques in both men and women and, generally, represented underestimations of body fat and overestimations of FFM using the single techniques (Tables 5.17 and 5.18). In men, estimates of body fat obtained using underwater weighing as a single technique were closest to those derived using the four-compartment model; anthropometry provided the least accurate estimates, with a mean underestimation of almost 10%. In women, DEXA^ assessments provided the closest estimates of body fat whilst DEXA^ provided the least accurate with a mean overestimation of approximately 4%.

In consequence, body fat measured using the different methods varied significantly from estimates derived from the four-compartment model (Table 5.19). Substitution of values at the extremes of the ranges observed for either density or hydration fraction of FFM by the assumed and commonly used values, 1.1 kg/1 and 0.732^^^, resulted in potential errors in estimating total body fat or FFM of up to 4.6kg and 5.4kg respectively, that is a 23% error in the fat estimation and about 10% in FFM in the patients assessed.

The mean density of FFM, derived from the four-compartment model, was 1.0991 Chapter 5 Body composition in cirrhosis 242

± 0,0073 kg/1 and 1.0941 ± 0.0090 kg/1 for men and women respectively, with no significant difference between the sexes. The overall mean was thus 1.0966 ± 0.0084 kg/1 with a range of 1.0754 - 1.1096 kg/1. Published values^^^ of density of FFM in healthy individuals, determined using the same four-compartment model and same equipment, were 1.1024 ± 0.0078 kg/1 in men and 1.1003 ± 0.0066 kg/1 in women with an overall range of 1.0795-1.1110 kg/1. In comparison with these, the values derived in the present study were a mean of 0.3% and 0.6% lower in men and women with 95% limits of agreement of 1.3% and 1.7% respectively.

The hydration fraction of FFM, also derived from the four-compartment model was 0.7563 ± 0.0263 for men and 0.7726 ± 0.0270 for women, with no significant difference between them. Thus the mean hydration fraction for all subjects was 0.7644 ± 0.0273 ranging between 0.7172 and 0.8297 (Figure 5.5). Published values^^^ of hydration fraction of FFM in healthy individuals, again determined using the same model and equipment, were 0.7332 ± 0.0219 in men and 0.7449 ± 0.0192 in women with an overall range of 0.6941-0.7837. In comparison with these, the values determined in the present study were a mean of 3.1% and 3.7% higher in men and women respectively with 95% limits of agreement of 7.2% in both.

Substitution of 30 and 400 ml for the assumed 100 ml volume of gastrointestinal tract gas in the calculation of body density yielded a mean increase of 0.9% in whole body density and a mean decrease of 4.0% fat respectively.

5.18.5 Discussion The Fuller four-compartment model^^^ is considered a suitable reference method for assessing body composition, particularly in individuals where the constancy of composition of FFM is questionable, as is the case in cirrhosis. A superior, reference method of assessment, against which the results from the four- compartment model could be evaluated, does not seem to be available and thus it is not possible to challenge the results obtained using the model in the present Chapter 5 Body composition in cirrhosis 243 study. The theory of the model is, however, based on assumptions that are considered robust even in pathological extremes and, therefore, the model is confidently assumed to be correct, providing reference data against which other methods can be compared.

The results from this study show that estimates of total body fat and FFM determined by single body composition techniques other than underwater weighing differ significantly from those derived using the four-compartment model.

Bias between single techniques and four-compartment model The bias between results and 95% limits of agreement for both body fat and FFM derived from the single techniques and the four compartment model exceeded published values^^^ obtained from healthy individuals using the same methodology and equipment. None of the single techniques provide estimates which confidently predict body fat to within 4% of the values determined by the four-compartment model. The use of these single techniques in patients with cirrhosis, must therefore be questioned. In particular, the limitations of DEXA and anthropometry, which are becoming increasingly used to provide estimates of body composition in clinical situations, should be recognized.

The values of bias could be used as correction factors to adjust estimates of fat or FFM derived using the single techniques. However, the limits of agreement of the bias are also large, casting doubt on the usefulness of the corrected data. This may be of sufficient accuracy in some projects, for example clinical screening procedures, but not in others such as monitoring studies where greater sensitivity is required.

Standard values for density and hydration fraction of FFM The substitution of standard values for density and hydration fraction of FFM at the extremes of the ranges determined in the present study resulted in Chapter 5 Body composition in cirrhosis 244 unacceptable errors; discrepancies of 0.01 kg/1 in density and 0.01 in hydration fraction represent potential errors of 5% and 1% fat respectively. Thus, in patients with cirrhosis in whom there was no clinical evidence of fluid retention, standard values cannot be used to derive useful estimates of body composition. It would be expected that even greater errors would arise in patients with cirrhosis and ascites and thus the derived estimates of body composition would be meaningless.

The values for the density of FFM calculated in the present study were lower than published values^^^ and represent increased hydration or, to a lesser extent, decreased mineralisation in FFM in the patients with cirrhosis, both of which have been reported previously^^’^*^®. The hydration fraction of FFM in the present study was greater than in healthy volunteers^^^ indicating that patients with cirrhosis with no clinical evidence of fluid retention may still be overhydrated^^^.

Limitations of the results The four-compartment model divides the body into fat, water, mineral and protein but does not quantify other components, including glycogen, nucleic acids and amino acids. These are included in the protein estimates because their densities are closer to that of protein than fat. However, in patients with cirrhosis, glycogen stores are depleted^^^’^^^, and thus overestimations of protein associated with glycogen in this patient population will be less than in healthy individuals. Synthesis of nucleic acids is inhibited by alcohoF® and malnutrition^^^ and abnormalities of amino acids are observed in patients with liver disease^^°. These changes have not been quantified in patients with cirrhosis, but it is unlikely that they will have a major effect on estimates of protein as these components comprise less than 1 kg of the total body mass in a 70 kg man.

The accuracy of the four-compartment model is dependent on the accurate measurement of whole body density^^^. In the present study, errors in density measurements of 0.01 kg/1 will result in errors of approximately 3% fat. This Chapter 5 Body composition in cirrhosis 245

emphasises the need for careful measurement not only of underwater weight but also of the respiratory tract volume at the moment of weighing and the use of a realistic correction volume for gastrointestinal tract gas. The assessment of body density by plethysmography, for example using the Bod Pod Body Composition System (Life Measurement Instruments, Concord, CA, USA), would reduce errors arising from inaccurate correction for gastrointestinal tract gas; this is because body density is derived directly from body volume in this method rather than determined from values for displaced water as in the present study^^^’^^^.

Practical issues This complex, multiple technique assessment of body composition was carried out in patients with cirrhosis utilizing facilities at two centres 60 miles apart. All of the patients studied were physically fit enough to travel and take part in the investigations, although some were less robust than individuals who might be considered typical volunteers for this type of study. However, by briefing the patients thoroughly before the study, approaching each of the assessment procedures steadily and incorporating rest periods and food breaks into the protocol, all investigations and data collection were successfully completed for each patient.

Although the assessments undertaken were time consuming and labour intensive, the patients taking part were not exposed to large doses of radioactivity^^^. Even so, this series of investigations is a research tool and is unsuitable for routine use in the clinical situation. By comparison, in terms of simplicity, ease of use and cost, a number of bedside techniques are superior. However, the results obtained using these techniques compare unfavourably with those derived from the multicompartment model and must be interpreted accordingly.

Study conclusions A four-compartment body composition model can be applied to patients with Chapter 5 Body composition in cirrhosis 246 cirrhosis without evidence of fluid retention using underwater weighing, deuterium dilution and DEXA assessments to provide estimates of fat, water, mineral and protein. Estimates of body fat and FFM obtained using deuterium dilution and DEXA, as single assessment techniques, and skinfold anthropometry differ significantly from those derived using the four-compartment model thus indicating that they cannot be used to provide accurate body composition in this patient population. Table 5.1 Measurement of body water in patients with chronic liver disease using dilution techniques. Data e^qjressed as mean ± ISD.

First author Patients Cirrhosis Tracer Sample time Compartment TBW (% of weight) ECW (% of TBW) Year (n) (%) (hr post-dose) assessed No ascites Ascites No ascites Ascites

O’Meara‘S 1957 8 100 'H P 2-6 TBW - 66.5 ± 5.6 --

Moore^® 1963 3 100 % o 4-18 TBW - 63.7 ± 8.0 - ^Br 4-18 ECW 68.0 ± 13.2

Schober'*’’ 1979 12 100 'HP NA TBW - 67.4 ± 9.8 - ^Br NA ECW 48.6 ± 8.0

Vitale"' 1985 11 100 'H P 24-48 TBW - 63.6 ±1 1 .6 .

Mullen” ® 1986 6 100 Hj'^O 0-6 TBW 61.7 ± 6.1 --

Loguerico^®^ 1988 23 100 antipyrine 24 TBW 72.7 91.0 .

McCullough^“ 1991 21 100 Hz'% 0-6 TBW 60.8 ± 2.1 69.7 + 1.2 Br 0-6 ECW 45.2 ± 0.4 46.3 ± 0.9

McCullough^^' 1992 6 100 H,'% 0-4 TBW 66.1 ± 1.6 - - Br 0-4 ECW 42.3 ± 2.5

Zillikens"’ 1992 27 100 % o 2.5-5 TBW 56.4 ± 5.7 58.6 ± 6.2 -.

Crawford'" 1993 15 80 'H P 6.0 TBW 60.9 66.8 -.

Prijatmoko"® 1993 38 100 'H P 2+ TBW 60.5 - 67.4*

Crawford"® 1994 48 100 'H P 6.0 TBW - -

Holt^ 1994 41 NA 'H P 4-6 TBW - -

Verboeket-van Venne®^’ 1995 8 100 'H P 9-11.5 TBW - -

Borghi®® 1996 34 100 'HP 3.5 TBW 56.6 - - Br ECW 41.5

Schloerb®” 1996 16 NA 'HP NA TBW -- -- Br ECW

AbergeP 1997 12 100 H^'^O NA TBW 53.1 - - Br ECW 46.8

‘Ascites status not given, mean data given by severity of liver disease (Child’s grades A - Cf*^; NA - not available. Table 5.2 Measurement of TBK in patients with chronic liver disease. Data expressed as mean ± ISD.

First author Patients Cirrhosis TBK % of Reference^^ Assumptions made*’ Year (n) (%) (mmol)

Blendis'*® 1986 7 100 3075 ± 203 --

162 Lautz^ 1992 30 100 -- BCM = TBK X 0.00833

Muller^’® 1992 20 100 - - S FFM H T B K /68.1 162 $ FFM = TBK / 64.2

Mtillef"° 1992 25 100 - - S FFM = TBK / 68.1 162 ? FFM 5 TBK / 64.2

Muller’*' 1992 20 100 - - S FFM = TBK 7 68.1 162 9 FFM = TBK / 64.2

Bramley^^ 1993 53 100 - - S FFM = TBK / 60.8 366 9 FFM = TBK / 58.1

162 Crawford"'* 1993 15 100 2515 ± 731 - BCM = TBK X 0.00833

162 Oldroyd'*'*^ 1993 55 100 <3A° 3085 ± 588 BCM = TBK X 0.00833 dA+ 2821 + 816 - ?A° 2185 ± 372 9A+ 2101 ± 348

162 Crawford"^ 1994 57 100 A“ 2346 + 541 89.3 ± 16.8 BCM = TBK X 0.00833 B 2597 ± 713 87.5 + 7.8 C 2021 ± 397 72.6 ± 12.6

Holt^" 1994 62 NA - - NA

70 Pirlich'”'* 1996 26 100 - - BCM = TBK X 0.0108

A" No ascites; ascites; “Severity of liver disease"^®; *’BCM in kg, FFM in kg, TBK in mmol.

4^to oo Table 5.3

Assessments of body composition in patients with chronic liver disease using DEXA.

Data expressed as mean ± ISD.

First author Patients Number Total body Total body fat FFM Comparisons with other methods Year bone mineral* (%) (%) of assessing body composition

Bramley®’ 1993 cirrhosis 53 NA NA NA BIA, SF, TBK

Oldroyd'’”^ 1993 cirrhosis 55 dA° NA 6A° 23 dA° 77 4 compartment model based on dA+ NA dA" 19 dA+ 81 DEXA, TBK $A° NA 9A° 35 9A° 65 5A^ NA ?A" 34 9A" 66

Piijatmoko'’*® 1993 cirrhosis 38 A’’ 0.99 ± 0.04 A 26.2 ± 7.2 A 77.5 ± 1.7 BIA, SF B 0.99 ± 0.05 B 27.1 ± 6.9 B 76.8 ± 1.8 multicompartment model based on C 1.09+0.15 C 21.0 ± 5.8 C 83.7 ± 1.7 DEXA, IVNAA, TBW ("H^O)

Meys^®^ 1994 CLD 33 2.10 ± 0.54 22.0 ± 6.3 NA none post-transplant 31 2.13 ± 0.48 26.8 ± 8.8 NA

Wicks®®^ 1995 PBC 80 NA NA NA SF

Feller’®^ 1997 ESLD 24 85.4 ± 14.4 NA NA none

CLD chronic liver disease; PBC primary biliary cirrhosis; ESLD end-stage liver disease; SF skinfold anthropometry; deuterium dilution; IVNAA in vivo neutron activation analysis. Total body bone mineral expressed in different units (Prijatmoko kg/mVm; Meys kgHA; Feller % matched controls); ‘’Severity of liver disease'*'’®. to VO Table 5.4 Assessments of body composition in patients with chronic liver disease using anthropometry.

Data expressed as mean ± ISD.

First author Patients Male:Female Total body fat Muscle mass Prediction Comparison with other methods Year (n) (%) (%) formulae of assessing body composition

Loguerico^®'' 1988 23 cirrhosis 15:8 25 ± 10 NA Dumin^''^ none

Campillo*^ 1989 30 cirrhosis 21:9 20 ± 2 NA Dumin^'*^ none

Merlf'" 1990 25 cirrhosis 16:9 26 ± 9 N A Dumin^''^ none

Campillo*^ 1992 10 cirrhosis 5:5 23 ± 2 29 ± 2 Dumin^''^ Heymsrield^^'* none

Bramley” 1993 53 cirrhosis 20:33 NA NA Dumin^''^ BIA, DEXA

Prijatmoko'*''^ 1993 38 cirrhosis A“ 17:0 25 ± 6 NA Dumin*''^ BIA, DEXA, IVNAA, B 15:0 11 ± 1 C 6:0 17 ± 6

Campillo*® 1994 13 cirrhosis 11:2 20 ± 3 29 ± 2 Dumin^"^ Heymsfield^^'* none

Crawford^ 1994 57 cirrhosis 23:34 25 NA Bruce®^ TBK, 'H P

Madden^^^ 1994 60 cirrhosis 40:20 20 ± 8 NA Dumin^'*^ BIA

Campillo®" 1995 26 cirrhosis 21:5 16 ± 6 NA Dumin^'*^ none

Ksiazyk^^^ 1996 11 cirrhosis ± CH 6:5 NA N A Dumin^'*^ Brook^® none

Pirlich''^'' 1996 102 cirrhosis A “ 6:13 NA 30 ± 9 Heymsfield^^'* 24UC B 22:26 3 1 + 9 C 22:13 29 ± 8

Selberg''^^ 1997 75 transplant candidates 37:38 NA 29 ± 8 Heymsfield^^" none

“Severity of liver disease'*"^; CH - chronic hepatitis; IVNAA - in v iv o neutron activation analysis; - deuterium dilution; 24UC - 24-hour urine creatinine. NJLn O Table 5.5

Body composition studies undertaken in patients with chronic liver disease using BIA: Comparisons with other techniques.

First author Patients Ascites Comparative methods Compartments Conclusions Year (%) assessed

McCullough^^° 1991 21 cirrhosis 38 TBW Correlation good TBW in non-ascitic patients, poor with ascites Br ECW Correlation poor in all groups

Lautz^^ 1992 30 cirrhosis NA TBK BCM BIA overestimated TBK data if BCM < 18kg

Bramley^^ 1993 53 cirrhosis 43 DEXA fat Correlation good in women and men without ascites

Prijatmoko"'^^ 1993 38 cirrhosis NA AN, DEXA fat BIA data comparable with AN and DEXA data DEXA, IVNAA FFM BIA underestimated combmed DEXA + IVNAA data TBW BIA underestimated data

Holt^^' 1994 41 CLD "most" AN FFM 23% } TBK FFM 28% } standard deviation of mean differences between methods ^ 0 FFM 14% }

Madden^^^ 1994 60 cirrhosis 50 AN fat BIA overestimated AN data

Cabré"" 1995 55 cirrhosis 58 AN TBW BIA did not detect differences between groups observed by AN

Schloerb''"" 1996 15 cirrhosis NA TBW Poor correlation with Tl^O data

Pirlich""'' 1996 26 cirrhosis NA TBK BCM Good correlation with TBK data reported (data not shown)

- oxygen-18 dilution; Br - bromide dilution; AN - anthropometry; IVNAA - neutron activation analysis; Tl^O - deuterium dilution.

K5LA Table 5.6

Body composition studies undertaken in patients with cirrhosis using BIA; Measurement of changes in TBW with paracentesis.

Data expressed as mean ± ISD.

First author Patients BIA equipment Volume of ascites removed Change in TBW Year by paracentesis (1) detected by BIA (1)

Guglielmi^®'' 1991 10 SF BIA-101 (RJL) 7.9 ± 3.8 1.9 ± 1.0

McCullough^^° 1991 8 SF BIA-104 (RJL) 6.9 ± 1.4 2.1 ± 0.4*

Lautz^*^ 1992 5 SF BIA-101 (RJL) 4.0 1.1

Zillikens^^° 1992 12 SF BIA-101 (RJL) 9.0 ± 3.4 4.6 ± 2.6*

Wu^^® 1994 3 BIA-106 (RJL) NA "poorly correlated"

Cabré"" 1995 13 BIA-109 (RJL) 5.6 ± 0.8 2.0 "no association detected"

Panella"'" 1995 6 SF BIA-101 (RJL) 8.3 ± 5.0 "no change detected"

Schloerb''"" 1996 3 SF BIA-IOIQ (RJL) 10.0 "correlation good" r=0.99

RJL - RJL Systems, Detroit, MI; NA - not available; ‘difference from volume of ascites removed, p<0.01.

to N)LA 253

Table 5.7

Body composition studies undertaken in patients with cirrhosis using 24-hour urinary

creatinine excretion. Data expressed as mean ± ISD.

First author Patients Urinary creatinine Assumptions made Year excretion (mg/24hr)

ZolF"' 1982 30 cirrhosis 1.125 ±0.06 1 g creat = 20 kg muscle 199

Shanbhogue"®’ 1987 10 ESLD 1.900 ±0.60 NA

Takabatake^^ 1988 7 FH & 1 PBC 0.858 NA

John^^^ 1989 20 AH 1.036 NA

Schneeweiss*’* 1990 22 CLD 1.059 ± 0.42 NA

Miiller^’® 1991 10 cirrhosis NANA

Lautz^^ 1992 123 cirrhosis NA <3 23 mg creat = 1 kg LBM ? 18 mg creat = 1 kg LBM

LollP°" 1992 200 cirrhosis NA LBM kg H 7.38 + (0.02908mg creat) 163

Müller^^® 1992 123 cirrhosis NA d 1 mg creat = 23 mg muscle mass * ? 1 mg creat = 1 8 mg muscle mass

Pierrugues"^^ 1992 40 cirrhosis 0.862 NA

Nielsen^®^ 1993 37 cirrhosis NA LBM kg = 7.38 + (0.02908mg creat) 163

Caregaro®^ 1994 56 cirrhosis 1.081 NA

Chen'°* 1994 57 HCC 1.243 NA

Selberg""" 1994 14 cirrhosis 1.195 ±0.50 1 g creat = 18.5 kg muscle =

Campillo^ 1995 26 cirrhosis 0.864 ±0.15 NA

Nielsen^^ 1995 15 cirrhosis 0.780 ± 0.03 LBM kg = 7.38 + (0.02908mg creat) 163

221 Pirlich^^'' 1996 102 cirrhosis 1.090 ±0.44 1 g creat = 18.5 kg muscle

Chang®^ 1997 16 cirrhosis 0.877 ± 0.08 NA

Sarin"’^ 1997 52 CLD 51% predicted NA

Selberg'*®^ 1997 75 OLT 91% predicted NA

ESLD - end-stage liver disease; FH - fulminant hepatitis; PBC - primary biliary cirrhosis; AH - alcoholic hepatitis; CLD - chronic liver disease; HCC - hepatocellular carcinoma; OLT - orthotopic liver transplant candidates; NA - not available; creat - creatinine. ®As described in the paper, but not referenced. Table 5.8

The four-compartment model o f body composition^*^. To calculate body compartments, start at the top row, left hand column and move to the right.

Continue on lower row, moving left to right and finish at bottom right with hydration fraction.

Variable Body mass Body volume Body density Water mass Water volume Mineral ash Mineral mass Mineral volume Protein + fat (kg) (1) (kg/1) (kg) (1) mass (kg) (kg) (1) density (kg/1)

Example; Patient 07 72.20 71.53 1.0094 35.34 35.12 1.8537 2.3618 0.7775 0.9682 Male 47 years

Identifier A BC DE FG HJ

Source of measured: measured: calculated: measured: calculated: measured: calculated: calculated: calculated: values scales underwater C = A/B deuterium E = D X 0.99371 DEXA G = F X 1.2741 H = G / 3.0375 J = (A-ID+Gl) weighing dilution (B-[E+H])

Protein + fat Fat mass FFM Fat Protein + fat Protein mass Protein volume Fat-fiee Hydration mass (kg) (kg) (kg) (%) volume (1) (kg) (1) density (kg/1) fraction

34.50 27.16 45.04 37.62 35.63 7.34 5.48 1.089 0.785

K LM NP RS TU

calculated: calculated: calculated: calculated: calculated: calculated: calculated: calculated: calculated: K = A-(D+G) L = ([2.7474/J] - M = A - L N = L X 100/A P = K / J, or R = K - L S=P-(L/0.9007) T = (D+G+R) U = D /M 2.0503) X K P = B-(E+H) (E+H+S)

Assumptions and constants are listed on page 217. to LA 255

Table 5.9

Subjective assessment: Symptom scores for flatulence and abdominal bloating in patients with cirrhosis in relation to treatment with non-absorbable disaccharides

Patient Age Sex Aetiology Lactulose / Flatulence® Abdominal of lactitol bloating® (yr) cirrhosis (ml or g/day)

TREATED

1 25 M alcohol 5-20g 0.41 0

2 40 M alcohol 15-30ml 0.42 0.01

3 51 M alcohol 20ml 0 0

4 52 M alcohol 30g 0.90 0.66

5 62 M PSC 45-50g 0.61 0.07

6 63 M alcohol 40g 1.27 0

7 50 F alcohol 25ml 1.49 2.05

8 60 F alcohol 30g 1.31 0.22

9 70 F alcohol 10ml 0.12 0

Median 52 6M;3F - - 0.61 0.02 (range) (25-70) (0-1.49) (0-2.05)

UNTREATED

10 35 M alcohol - 1.02 0.03

11 60 M alcohol - 0.02 0

12 65 M alcohol - 0.15 0

13 36 F alcohol - 0.23 1.11

14 47 F alcohol - 0.34 0.25

Median 47 3M:2F - - 0.23 0.03 (range) (35-65) (0.02-1.02) (0-1.11)

^Mean daily symptom score: 0 = absent; 1 = mild; 2 = moderate; 3 = severe. PSC - Primary sclerosing cholangitis 256

Table 5.10

Objective assessment 1: Age, diagnosis and indication for spiral CT scanning in 6 patients with cirrhosis prescribed non-absorbable disaccharides and 6 control patients.

Patient Age Diagnosis Reason for Lactulose /

(yr) CT scan lactitol

(ml or g/day)

TREATED

1 46 cirrhosis HCC screen 20g

2 53 cirrhosis HCC screen 20g

3 54 cirrhosis HCC screen 35ml

4 58 cirrhosis HCC screen 20ml

5 62 cirrhosis HCC screen lOg

6 66 cirrhosis HCC screen 40g

Mean 56.5 - - 28ml / 23g

UNTREATED

7 40 NAD abdomen teratoma 3 years ago nil routine folow-up

8 45 chronic renal abdominal pain nil failure

9 65 NAD abdomen gastric carcinoma 3 years ago nil routine follow-up

10 65 NAD abdomen haematuria nil

11 66 NAD abdomen weight loss nil altered bowel habit

12 72 nodes in left prostatic carcinoma nil iliac region routine screening

Mean 58.8 - - nil

HCC - Hepatocellular carcinoma; NAD - Nothing abnormal detected. Table 5.11

Objective assessment 1: Area of gas in the gastrointestinal tract in the middle abdomen in six men with cirrhosis taking non-absorbable disaccharides and six male control patients.

Measurements made in 10 consecutive slices above the iliac crest by spiral CT.

Distance above iliac crest Patients with cirrhosis (mm^) Control patients (mm^) (mm) 1 2 3 4 5 6 7 8 9 10 11 12

100 643 1180 54 321 46 1458 498 661 1908 822 3568 85

90 774 1029 341 309 33 1127 601 485 1609 579 4692 10

80 1040 1083 877 1141 165 1061 439 313 2115 1185 4447 444

70 1087 812 1147 1077 151 992 356 0 2855 1397 3816 507

60 1047 245 685 1280 118 1299 280 40 2705 908 3214 603

50 1050 478 959 1935 57 1866 320 32 2775 500 3006 615

40 1015 418 969 1735 203 2036 457 150 2462 110 2836 443

30 825 296 566 1731 281 1968 1610 195 2064 25 2987 605

20 747 418 154 1348 274 1814 703 0 1968 20 2494 713

10 768 1346 214 837 209 1608 592 1 1730 15 2093 711

Individual total area (mm^) 8996 7305 5966 11714 1537 15229 5856 1877 22191 5561 33153 4736

Individual volume (ml) 90 73 60 117 15 152 58 19 222 56 332 47

Group median volume (ml)“ 81.5 (15-152) 57.0 (19-332)

^Individual volume (ml) = Sum of individual total areas (mm^) x 10mm / 1000. to <1LA Table 5.12

Objective assessment 2: Demography, height, weight and volumes of gastrointestinal gas determined in two patients with cirrhosis prescribed lactulose and two control patients.

e e g tain Age Sex HeightPatient WeightDiagnosis Reason for Gastrointestinal tract gas volume CT scan ml (% of total) (yr) (m) (kg) upper middle lower total volume abdomen abdomen abdomen

Patients with cirrhosis prescribed lactulose;

1.65 67.4 alcoholic HCC screen 206 107 376 cirrhosis (17%) (55%) (28%) ( 100%)

1.77 69.7 cryptogenic HCC screen cirrhosis ( 1%) (36%) (63%) ( 100%)

Control patients not prescribed lactulose or lactitol:

1.69 68.9 orchidectomy routine 230 366 for teratoma follow-up (21%) (16%) (63%) ( 100%)

1.69 65.2 hysterectomy routine 100 190 ovanan cancer follow-up (53%) (41%) (6%) ( 100%)

Upper abdomen - >10cm above iliac crest; middle abdomen - O-lOcm above iliac crest; lower abdomen - below iliac crest. toLn oc Table 5.13

Summary of published studies where gastrointestinal gas has been quantified.

First author Subjects Health Age Sex Mean Minimum Maximum Method

Year status range volume volume volume

(n) (yr) (M:F) (ml) (ml) (ml)

Blair'*" 1947 6 healthy NA 6:0 960 NA NA plethys­

mography

Keys""" 1953 21 healthy "young" 21:0 28 NA 49 X-ray

Bedell"^ 1956 13“ healthy 20-43 11:2 115 0 506 plethys­ mography 47" lung disease 15-76 35:12 116 0 537

Greenwald"°' 1969 18 healthy 18-28 18:0 111 NA NA water

displacement

Levitt""" 1971 11 healthy "young" 11:0 90 30 200 argon

washout

NA - not available; “32 observations; '’48 observations.

Lh Table 5.14

Demographic, clinical and anthropometric data from six male patients with alcoholic cirrhosis undergoing deuterium dilution studies.

Age Pugh’s Height Weight Ascites* Estimated BMI Body fat®

(y) score""® (m) (kg) (kg) Dry Weight (kg.m'^) (%)

1 53 8 1.75 93.1 0 93.1 30.4 25

2 37 5 1.83 91.6 0 91.6 27.4 25

3 53 6 1.75 64.4 0 64.4 21.0 19

4 50 14 1.55 54.9 10 44.9 18.7 10

5 68 12 1.69 63.5 11.5 52.0 18.2 8

6 58 9 1.59 94.8 18 76.8 30.4 22

Mean ± ISD 53.2 ± 10.1 9 ± 3.5 1.69 ±0.11 77.1 ± 18.0 - 70.5 ± 20.2 24.4 ± 5.7 18.2+7.5

“Estimated weight. ^Determined from skinfold anthropometry^"^.

K) ON o Table 5.15

Difference in deuterium enrichment between pairs of body fluids sampled at 2 hourly intervals after oral administration of deuterium in six patients with cirrhosis. Data expressed as % of mean values.

Patient Post-dose (hr)^ 2 4 6 8 10 12 Fluids compared -I

1 Saliva & serum 6.5 3.8 0.6 4.2 4.2 0.9

2 Saliva & serum 0.8 0.4 0.4 4.8 0 1.3

3 Saliva & serum 8.8 4.2 0 1.2 1.6 2.6

Saliva & serum NA 3.1 5.5 5.8 6.6 4.4 4 Saliva & ascitic fluid NA 0.6 4.5 3.2 1.1 1.2

Serum & ascitic fluid 10.9 2.6 0.9 2.6 5.5 3.2

Saliva & semm 6.8 7.4 8.4 11.5 10.9 10.1 5 Saliva & ascitic fluid 5.4 NA 7.9 7.5 NA 6.6

Serum & ascitic fluid 12.3 NA 0.5 4.0 NA 3.5

Saliva & serum 5.4 4.4 8.2 9.7 7.7 7.2 6 Saliva & ascitic fluid 66.3 26.7 8.9 1.3 2.4 2.4

Serum & ascitic fluid 71.1 31.0 17.1 8.4 10.1 9.6

K)0\ 262

Table 5.16

Demographic and anthropometric data in 10 men and 10 women with alcoholic cirrhosis undergoing body composition assessment.

Patients Sex Age Pugh’s Height Weight BMl (yr) score (m) (kg) (kg.m’^) (5-15)

01 M 67 7 1.86 74.49 21.5

02 M 54 6 1.75 63.70 20.8

03 M 54 8 1.76 95.20 30.7

04 M 56 7 1.75 83.90 27.4

05 M 38 5 1.82 86.20 26.0

06 M 48 5 1.64 89.70 33.4

07 M 47 8 1.78 72.20 22.2

08 M 28 7 1.73 60.10 20.1

09 M 43 5 1.64 82.26 30.6

10 M 49 5 1.79 70.80 22.1

Mean (male) - 48.4 6.3 1.75 77.9 25.5

11 F 51 6 1.52 68.50 29.6

12 F 48 8 1.70 82.20 28.4

13 F 49 9 1.55 63.10 26.3

14 F 56 5 1.44 38.30 18.5

15 F 51 5 1.47 64.20 29.7

16 F 52 8 1.60 61.80 24.1

17 F 53 7 1.70 81.14 28.1

18 F 40 6 1.62 88.50 33.7

19 F 42 6 1.61 67.80 26.2

20 F 46 5 1.68 60.90 21.6

Mean (female) - 48.8 6.5 1.59 67.6 26.6 T tiW v J.IT

Comparison of total body fat (%) using estimates derived from the four-compartment model as a reference against those derived using single assessment techniques in 10 male and 10 female patients with cirrhosis and, for comparison, published data from 28 healthy volunteers.

Data expressed as bias (4-CM - alternative) and 95% (±2SD) limits of agreement as a percentage of body weight. Values for bias for FFM as a percentage of body weight are equal and opposite to those for total body fat (%), and the limits of agreement are equal for both"*^.

Patients with cirrhosis Healthy volunteers'*^

4-compartment model 4-compartment model 4-compartment model (10 men) (10 women) (16 men + 1 2 women)

UWW -0.25 + 4.04 -1.45 ± 5.01 +0.75 ± 3.78

"H20 +2.33 ± 5.24 +3.67 + 4.98 +1.70 + 3.91

DEXA^ +5.89 + 4.88 +4.03 ± 6.78 +1.36 + 4.95

DEXA*^ +3.29 + 5.23 -0.21 + 5.06 NA

4SF +9.56 + 10.77 +3.26 + 11.32 +0.09 ± 5.42

4-CM four-compartment model; UWW underwater weighing; deuterium dilution; 4SF anthropometry.

to asU) Table 5.18

Comparison of FFM (kg) using estimates derived from the four-compartment model as a reference against those derived using single assessment techniques in 10 male and 10 female patients with cirrhosis and, for comparison, published data from 28 healthy volunteers.

Data e7q)ressed as bias (4-CM - alternative) and 95% (±2SD) limits of agreement in kg. Values for bias for total body fat (kg) are equal and opposite to those for FFM (kg), and the limits of agreement are equal for both''®.

Patients with cirrhosis Healthy volunteers'®^

4-compartment model 4-compartment model 4-compartment model (10 men) (10 women) (16 men + 1 2 women)

UWW +0.10 ± 3.18 +0.90 + 3.12 -0.69 + 3.19

2H20 -1.67 ± 3.86 -2.36 + 2.91 -1.37 + 2.97

DEXA° -4.42 ± 3.08 -2.66 + 4.07 -1.16 ± 4.07

DEXA^ +2.37 + 3.60 +0.19 + 2.87 NA

4SF -7.31 ± 7.96 -1.93 + 6.75 -0.20 + 3.88

4-CM four-compartment model; UWW underwater weighing; deuterium dilution; 4SF anthropometry.

to 4:^ON 265

Table 5.19

Total body fat assessed using the four-compartment model and single assessment techniques in 10 male and 10 female patients with cirrhosis.

Data expressed as %.

Patients 4-CM UWW DEXA° DEXA^ 4SF

01 â 20.80 20.26 20.19 14.1 14.7 12

02 s 34.54 34.15 32.83 25.2 27.2 19

03 $ 29.98 28.26 30.75 27.8 30.6 28

04 s 23.50 20.53 25.05 19.2 21.3 16

05 s 28.50 29.19 24.94 22.0 25.5 24

06 s 37.34 39.13 33.52 32.3 36.3 25

07 s 37.62 40.39 33.13 31.8 32.8 21

08 s 9.36 11.45 3.37 3.8 6.8 7

09 3 43.51 41.77 43.22 40.1 42.8 29

10 3 31.47 33.99 26.35 21.4 25.7 20

Mean ±SD â 29.66 29.91 27.34* 23.77*** 26.37** 20.10*** ±9.84 ±9.97 ±10.56 ±10.24 ±10.41 ±6.97

11 $ 41.63 41.07 41.69 38.8 42.1 39

12 ? 38.82 39.03 36.68 33.3 37.7 35

13 9 34.36 34.01 30.72 27.0 34.5 29

14 9 32.53 39.32 23.53 22.7 28.2 22

15 9 45.59 45.62 43.27 43.1 45.2 32

16 9 26.24 27.15 23.89 24.6 27.6 26

17 9 43.39 44.80 39.59 37.6 41.8 40

18 9 35.20 40.36 29.10 30.5 35.2 37

19 9 34.63 34.36 31.84 32.1 36.9 34

20 9 23.20 24.38 18.62 25.6 28.5 29

Mean ±SD 9 35.56 37.01 31.89** 31.53** 35.77 32.30 ±7.15 ±7.03 ±8.35 ±6.75 ±6.23 ±5.81

4-CM four-compartment model; UWW underwater weighing; deuterium dilution; 4SF anthropometry. Significance of difference between 4-CM and single techniques: p<0.05; *p<0.005; **p<0.0005. 266

.«ï,

kV 120 mA 280-^ Smart mA 280 Large 1 n rimm/1 2- I

Figure 5.1 Abdominal CT scan showing gastrointestinal tract gas. Gas is indicated by black areas labelled G. 267

No clinically detected Gross ascites fluid retention

8001

600-

400

200 Patient 1 Patient 4

E C l 800 C m 600 E

400 mc E 3 2 00 - Patient 2 Patient 5 5B 0) 0 ------1— I— 1 Q

800

600 ^ ^^ O'^ '

400-

200 Patient 3 Patient 6

0 2 4 6 8 10 12 0 2 4 6 8 10 12

Time after dose (hours)

—• — Saliva '"O—■ Serum “ ■— Ascitic fluid

Figure 5.2 Deuterium enrichment in saliva, serum and ascites following oral

administration in six patients with cirrhosis, three without clinically

detected fluid restriction and three with gross ascites 268

m m

Figure 5.3 Two views o f the undeiwater weighing tank 269 Figure 5.4

Calculation of body density in underwater weighing.

Body density = [bodv weight’" x densitv of water)

body volume - (residual volume x density of water)

where:

body volume weight" - weight"

and

residual volume = (RT gas + GIT gas) x BTPS

RT gas is the gas within the respiratory tract, determined thus:

RT gas = ([spirometer volume + dead space] x H) - dead space

where the dead space is the volume in the breathing tube between the spirometer and the patient and H is the change in helium concentration after equilibration in the respiratory tract:

H = [He concentration""^'^' - He concentration

He concentration^’"®^

GIT (gastrointestinal tract) gas is assumed to be 100 ml

BTPS is a factor to correct for the effects of temperature and pressure on the gases within the body^^^. 0.84 1

0.82 1

0.8 t

0.78 I ■I £ 0.76 I

I£ 0.74 ■ * Standard hydration fraction I 0.732 0.72 4

0.7 4

j

0.68 I

123456789 10 11 12 13 14 15 16 17 18 19 20

M e n Women

Figure 5.5 Hydration of fat-free mass in 10 men and 10 women with cirrhosis, <1tsj 271

CONCLUSIONS

Global nutritional status can be assessed in patients with chronic liver disease using the RFH G A scheme which was developed to take into account both subjective and objective nutritional variables. The scheme is practical to use in a clinical situation and has good interobserver reliability.

The prevalence of malnutrition in patients with cirrhosis at the Royal Free Hospital, assessed using the RFH GA scheme, was 57%. A higher prevalence of malnutrition was observed in patients with the most severely decompensated liver disease and malnutrition was associated with a greater prevalence of gastrointestinal symptoms and inadequate nutritional intake. Objective variables, including relative MAMC and relative grip-strength, were significantly associated with nutritional status, independently of the severity of liver disease.

The nutritional status of patients assessed using the devised scheme was significantly associated with survival in a large number of patients undergoing orthotopic liver transplantation. However, none of the individual contributing variables, objective or subjective, were significantly associated with survival when considered alone.

Increased resting energy expenditure was observed in patients with cirrhosis when adjusted for body weight and, when adjusted for fluid retention, for FFM. The Harris Benedict^®^ and Schofield"^^^ formulae used for predicting energy expenditure provided estimates which were significantly lower than measured values.

In patients with cirrhosis without clinical evidence of fluid retention, estimates of body fat and FFM obtained using a reference four-compartment model of body composition, utilizing underwater weighing, deuterium dilution and DEXA assessments of mineral, differed significantly from estimates derived using bedside techniques and deuterium dilution and DEXA alone. Therefore, such techniques should not be used t3 provide accurate body composition data in this patient population. The reference model provides a realistic alternative which can be used in research situations. 272

FUTURE WORK

The results from the studies presented in this thesis, have answered some of the questions addressed, but have also highlighted other issues that require investigation. A number of these issues are interrelated and some, arising from work in the earlier chapters, can be re-considered in the light of results from Chapter 5.

The nutritional assessment scheme devised in Chapter 1 has been shown to be reproducible and clinically useful. In Chapter 3, it was also shown to have predictive validity for survival in patients undergoing liver transplantation. Comparisons are required between categorization of nutritional status using the scheme and estimates of body composition derived using the four-compartment model. So far, insufficient patients have been assessed, most of whom were well nourished, to undertake a useful comparison but this should be carried out in the future.

The confirmation of an association between global nutritional status and post­ transplant survival in Chapter 3 suggests that there may also be a relationship between global nutritional status and morbidity variables. This could be undertaken simply by collating the additional variables from documented data. It would also be interesting to examine the association between body compartments and post-transplant survival and morbidity, but there may be practical difficulties in undertaking such assessments in this patient population.

The results from Chapter 4 show that REE, adjusted for body weight, fluid retention and FFM, is significantly increased in patients with cirrhosis. However in this study, FFM was estimated using skinfold anthropometry and, in Chapter 5, this has been shown to differ significantly from data provided by the reference four- compartment method. Comparisons of REE adjusted for FFM determined using the reference method are thus required. In order to facilitate this, REE was measured, in addition to body composition assessments, in the patients in Study 7. Control data has not yet been collected, but when it is available, REE adjusted for Future work 273 reference measures of FFM will be compared between groups.

The results from the studies in Chapter 5 show that the four-compartment model can be undertaken in patients with cirrhosis, providing that they have no clinically detectable fluid retention. Whilst this is the first time that a reference method of assessing body composition has been undertaken in this patient population, the exclusion of patients with clinically evident fluid retention is a concern. This eliminates from investigation not only a substantial number of patients but also precludes the study of some of the most vulnerable and nutritionally impaired individuals"^^^, who may be of particular interest with regard to nutritional intervention.

The accurate measurement of total body water in patients with fluid retention would overcome this problem, and this may be possible using mathematical modelling of the deuterium dilution data. A preliminary model can been devised considering three water compartments - gastrointestinal tract, ascitic fluid and remaining total body water - and using a non-linear least squared modelling technique to solve equations which describe the flow of deuterium between the compartments (Les Bluck, personal communication, 1997). The additional measurement of ascitic fluid volume using a different tracer, for example methylene blue^^^, and the collection of post-dose samples at time points earlier than 2 hours might also help resolve this problem.

The measurement of ECW in this patient population remains an unsolved problem. The dosing solution given to patients in Study 6 contained, in addition to deuterium, bromide which was given with the intention of using its dilution to determine ECW. Bromide enrichment was assayed in serum and ascitic fluid using an automated spectrophotometric fluorescein method. The repeatability of the results was poor, not considered to provide useful data and thus not presented in Study 6. Flowever, it was apparent even from these data, that bromide enrichment was still increasing in some patients when the final samples were collected at 12 hours after dosing. This compares with equilibration times of up to 12 hours after Future work 274 oral dosing in healthy volunteers. Once a reliable bromide assay has been established, further studies are required to determine the optimum post-dose sampling time in this patient population, which may exceed 12 hours, and thus to estimate ECW which, like deuterium, may require mathematical modelling in fluid retaining patients.

The patients who underwent body composition assessment using the four- compartment model in Study 7 were also given an oral dose of bromide and additional blood samples taken at 11 hours after dosing. These remain in storage at -20°C. Depending on the optimum equilibration time in non-fluid retaining patients and the development of an appropriate model, the samples may be used to estimate ECW in this patient population and, by difference from TBW, to calculate ICW.

In addition to the four-compartment model and single assessments techniques described, the patients in Study 7 were also assessed using multifrequency bioelectrical impedance analysis, both whole body and segmental, and near infrared interactance. Analyses of these data are yet to be undertaken and will provide comparison of two more bedside techniques of assessments with the reference four- compartment model.

In addition to this further work arising from the studies presented in the thesis, the development of tools for assessing nutritional status and providing reference body composition data in patients with cirrhosis will facilitate a variety of other projects. These could range from clinically based nutrition intervention studies to the exploration of metabolism including investigations of the effects of nutrient intake and exercise on total energy expenditure. 275

Appendix

RFH Global Assessment Data Collection Form

Patient Identification Age: .years Sex: male / female

CLINICAL INFORMATION Diagnosis

Anorexia absent mild / moderate severe Nausea absent mild / moderate severe Vomiting absent mild / moderate severe Difficulty in mastication absent mild / moderate severe Dysphagia absent mild / moderate severe Indigestion absent mild / moderate severe Food-related abdominal pain absent mild / moderate severe Bowel frequency ■daily Stool consistency / colour

Infections yes / no. Dates...... Renal dysfunction absent / mild / moderate / severe Hepatic encephalopathy absent / mild / moderate / severe GI tract bleeding absent / mild / moderate / severe Weight loss ...... kg, % Physical activity working as usual ...... working suboptimally ...... ambulatory...... bedridden ...... Fatigue absent / mild / moderate / severe

NUTRIENT INTAKE Appetite good / fair / poor Early satiety absent / mild / moderate / severe Taste changes absent / mild / moderate / severe Recent dietary intake

Estimated requirements •kcal Dietary restrictions Nutritional supplements oral / enteral / parenteral Nutrient intake adequate / inadequate / negligible 276

RFH Global Assessment Data Collection Form - continued

PHYSICAL STATUS Subcutaneous fat stores good / fair / poor Muscle wasting absent / mild / moderate / severe Peripheral oedema absent / mild / moderate / severe Ascites absent / mild / moderate / severe

ANTHROPOMETRIC MEASUREMENTS Height m Weight...... kg Estimated dry weight kg BMI (dry weight) kg/m^ MAC ...... cm TSF mm MAMC ...... cm MAMC 5th% ile^® ...... cm MAMC in relation to 5th% ile above / below

SUMMARY BMI: >20 MAMC: >5*%ile Nutrient intake: adequate <20 <5*%ile inadequate negligible

Refer to algorithm

Interim nutritional category: Well nourished Moderately malnourished (or suspected of being) Severely malnourished

Subjective override: None / decrease / increase

Global nutritional category: Well nourished Moderately malnourished (or suspected of being) Severely malnourished BMI

MAMC MAMC

> 5th % ile < 5 th % ile k 5th 7o ile < 5 th % ile

Dietary Intake Dietary Intake Dietary Intake Dietary Intake

I V V V V V V V V V Adequate Inadequate Negligible Adequate Inadequate Negligible Adequate Inadequate Negligible Adequate Inadequate Negligible

1 IV V ______y J 7 Adequately Moderately Malnourished Severely Malnourished Nourished or Suspected of Being

Subjective Override

Appendix RFH GA for assessing nutritional status in patients with cirrhosis. Patients are categorized in relation to their BMI, relative MAMC and nutrient intake into one of three categories - well nourished, moderately malnourished (or suspected of being so) or severely malnourished. A subjective override based on such factors as profound recent weight loss or aggressive nutrition intervention, can be used to modify the classification by up to one category. K) o 278

REFERENCES

1 Abad A, Cabre E, Gonzalez-Huix F, Gine JJ, Dolz C, Xiol X, Gassull MA. Influence of the nutritional status in the prognosis and clinical outcome of hospitalized patients with liver cirrhosis. Preliminary report. Journal of Clinical Nutrition and Gastroenterology 1987;2:63-68.

2 Abdelhadi M, Eriksson SA, Ljusk Eriksson S, Ericzon BG, Nordenstrom J. Bone mineral status in end-stage liver disease and the effect of liver transplantation. Scandinavian Journal of Gastroenterology 1995;30:1210-1215.

3 Abergel A, Bonny C, Ritz P, Boire JY, Beaufrère B, Bommelaer G. Nutritional evaluation by multifrequence bioimpedance analysis in patients with alcoholic cirrhosis: Comparison to dilutional methods. Hepatology 1997;26:208A(Abstract 319).

4 Abu Khaled M, McCutcheon MJ, Reddy S, Pearman PL, Hunter GR, Weinsier RL. Electrical impedance in assessing human body composition: The BIA method. American Journal of Clinical Nutrition 1988;47:789-792.

5 Acheson KJ, Campbell IT, Edholm OG, Miller DS, Stock MJ. The measurement of daily energy expenditure - An evaluation of some techniques. American Journal of Clinical Nutrition 1980;33:1155-1164.

6 Acheson KJ, Schütz Y, Bessard T, Anantharaman K, Flatt JP, Jéquier JP. Glycogen storage capacity and de novo lipogenesis during massive carbohydrate overfeeding in man. American Journal of Clinical Nutrition 1988;48:240-247.

7 AchordJL. Malnutrition and the role of nutritional support in alcoholic liver disease. American Journal of Gastroenterology 1987;82:1-7.

8 Adler M, Van Laethem J, Gilbert A, Gelin M, Bourgeois N, Vereerstraeten P, Cremer M. Factors influencing survival at one year in patients with nonbiliary hepatic parenchymal cirrhosis. Digestive Diseases and Sciences 1990;35:1-5.

9 Adler M, Verset D, Bouhdid H, Bourgeois N, Gulbis B, Le Moine O, Van de Stadt J, Gelin M, Thiry P. Prognostic evaluation of patients with parenchymal cirrhosis. Proposal of a new simple score. Journal of Hepatology 1997;26:642-649.

10 Akerman PA, Jenkins RL, Bistrian BR. Preoperative nutrition assessment in liver transplantation. Nutrition 1993;9:350-356. 279

11 Akers R, Buskirk ER. An underwater weighing system utilizing "force cube" transducers. Journal of Applied Physiology 1969;26:649-652.

12 Altman DG. Practical statistics for medical research. London: Chapman and Hall 1991:431- 432.

13 American Diabetes Association Inc and American Dietetic Association. A guide for professionals: Effective application of exchange lists for meal planning. New York: American Diabetes Association hic and American Dietetic Association 1977.

14 Amundsen LR. Isometric muscle strength testing with fixed-load cells. In: Amundsen L, ed. Muscle strength test testing: Instrumented and noninstrumented systems. New York: Churchill Livingstone 1990:89-112.

15 Anonymous. Bioelectrical impedance analysis in body composition measurement: National Institutes of Health technology assessment conference statement 1994 Dec 12-14. American Journal of Clinical Nutrition 1996;64:524S-532S.

16 Arciero PJ, Goran Ml, Poehlman ET. Resting metabolic rate is lower in women than in men. Journal of Applied Physiology 1993;75:2514-2520.

17 Arora NS, Rochester DF. Respiratory muscle strength and maximal voluntary ventilation in undernourished patients. American Review of Respiratory Disease 1982;126:5-8.

18 Arroyo V, Ginès P, Jimenez W, Rodés J. Ascites, renal failure, and electrolyte disorders in cirrhosis. Pathogenesis, diagnosis and treatment. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;1:429-470.

19 Ashwell M, Cole TJ, Dixon AK. Obesity: New insight into the anthropometric classification of fat distribution shown by computed tomography. British Medical Journal 1985;290:1692- 1694.

20 Baker JA, Humphrey SJE, Wolff HS. Socially acceptable monitoring instruments (SAMI). Journal of Physiology 1967;188:4P-5P.

21 Baker JP, Detsky AS, Wesson DE, Wolman SL, Stewart S, Whitewell J, Langer B, Jeejeebhoy KN. Nutritional assessment. A comparison of clinical judgment and objective measurements. New England Journal of Medicine 1982;306:969-972. 280

22 Balogh M, Kahn HA, Medalie JH. Random repeat 24-hour dietary recalls. American Journal of Clinical Nutrition 1971;24:304-310.

23 Bastow MD, Rawlings J, Allison SP. Benefits of supplementary tube feeding after fractured neck of femur: A randomised controlled trial. British Medical Journal 1983;287:1589-1592.

24 Bathgate AJ, Hynd P, Garden OJ, Hayes PC. Nutritional status and outcome following liver transplantation. Hepatology 1997;26:348A(Abstract 879).

25 Baumgartner RN, Chumlea WC, Roche AF. Estimation of body composition from bioelectric impedance of body segments. American Journal of Clinical Nutrition 1989;50:221-226.

26 Beath SV, Brook GD, Kelly DA, Cash AJ, McMaster P, Mayer AD, Buckels JAC. Successful liver transplantation in babies under 1 year. British Medical Journal 1993;307:825-828.

27 Beaton GH, Milner J, Corey P, McGuire V, Cousins M, Stewart E, de Ramos M, Hewitt D, Grambsch PV, Kassim N, Little JA. Sources of variance in 24-hour dietary recall data: Implications for nutrition study design and interpretation. American Journal of Clinical Nutrition 1979;32:2546-2559.

28 Becque MD, Katch VL, Moffatt RJ. Time course of skin-plus-fat compression in males and females. Human Biology 1986;58:33-42.

29 Bedell GN, Marshall R, DuBois AB, Harris JH. Measurement of the volume of gas in the gastro­ intestinal tract. Values in normal subjects and ambulatory patients. Journal of Clinical Investigation 1956;35:336-345.

30 Beddoe AH, Hill GL. Clinical measurement of body composition using in vivo neutron activation analysis. Journal of Parenteral and Enteral Nutrition 1985;9:504-520.

31 Behnke AR Jr, Feen BG, Welham WC. The specific gravity of healthy men. Journal of the American Medical Association 1942;118:495-498.

32 Benotti PN, Bistrian B. Metabolic and nutritional aspects of weaning from mechanical ventilation. Critical Care Medicine 1989;17:181-185. 281

33 Bertrân X, Femândez-BafïaresF, Planas R, Cabré E, Morillas R, De Léon R, Guilera M, Quer JC, Llovet JM, DoménechE, Gassull MA. Repercusiôn sobre el estado nutricional energético- proteico de la paracentesis total asociada a la infusion de albùmina o dextrano-70 en el tratamiento de la ascitis a tension en la cirrosis hepâtica. Revista Espanola de Enfermedades Digestivas 1991;79:320-323.

34 Bingham SA. The dietary assessment of individuals: Methods, accuracy, new techniques and recommendations. Nutrition Abstracts and Reviews 1987;57:705-742.

35 Bisdee JT, James WPT, Shaw MA. Changes in energy expenditure during the menstrual cycle. British Journal of Nutrition 1989;61:187-199.

36 Bishop CW, Bowen PE, Ritchey SJ. Norms for nutritional assessment of American adults by upper arm anthropometry. American Journal of Clinical Nutrition 1981;34:2530-2539.

37 Bistrian BR, Blackburn GL, Hallowell E, Meddle R. Protein status of general surgical patients. Journal of the American Medical Association 1974;230:858-860.

38 Bistrian BR, Blackburn GL, Scrimshaw NS, Flatt JP. Cellular immunity in semistarved states in hospitalized adults. American Journal of Clinical Nutrition 1975;28:1148-1155.

39 Bistrian BR, Blackburn GL, Sherman M, Scrimshaw NS. Therapeutic index of nutritional depletion in hospitalized patients. Surgery Gynecology and Obstetrics 1975;141:512-516.

40 Bistrian BR, Blackburn GL, Vitale J, Cochran D, Naylor J. Prevalence of malnutrition in general medical patients. Journal of the American Medical Association 1976;235:1567-1570.

41 Bistrian BR, Sherman M, Blackburn GL, Marshall R, Shaw C. Cellular immunity in adult marasmus. Archives of Internal Medicine 1977; 137:1408-1411.

42 Black D. Obesity. A report of the Royal College of Physicians. Journal of the Royal College of Physicians of London 1983;17:5-65.

43 Blackburn GL, Bistrian BR, Maini BS, Schlamm HT, Smith MF. Nutritional and metabolic assessment of the hospitalized patient. Journal of Parenteral and Enteral Nutrition 1977;1:11-

22 .

44 Blackburn GL, Thornton PA. Nutritional assessment of the hospitalized patient. Medical Clinics of North America 1979;63:1103-1115. 282

45 Blair HA, Dem RJ, Bates PL. The measurement of volume of gas in the digestive tract. American Journal of Physiology 1947;149:688-707.

46 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-310.

47 Bland M. An introduction to medical statistics. 2nd edn. Oxford: Oxford University Press 1995:197.

48 Blaxter KL, Brockway JM, Boyne AW. A new method for estimating the heat production of animals. Quarterly Journal of Experimental Physiology and Cognate Medical Sciences 1972;57:60-72.

49 Blendis LM, Harrison JE, Russell DM, Miller C, Taylor BR, Greig PD, Langer B. Effects of peritoneovenous shunting on body composition. Gastroenterology 1986;90:127-134.

50 Block GD, Warty V, Sanghvi A, Van Thiel DH. Effects of nutrition status on liver transplantation (OLTx) outcome. Hepatology 1989;10:657(Abstract 356).

51 Blue LS, Hasse JM, Levy MF, Jennings LW. Effect of obesity on clinical outcomes in liver transplantation. Journal of the American Dietetic Association 1993;93S:A49(Abstract).

52 Boddy K, King PC, Hume R, Weyers E. The relation of total body potassium to height, weight and age in normal adults. Journal of Clinical Pathology 1972;25:512-517.

53 Boileau RA, Horstman DH, Buskirk ER, Mendez J. The usefulness of urinary creatinine excretion in estimating body composition. Medicine and Science in Sports 1972;4:85-90.

54 Bollet AJ, Owens S. Evaluation of nutritional status of selected hospitalized patients. American Journal of Clinical Nutrition 1973;26:931-938.

55 Bonnell M. The growth of clinical nutrition. Journal of the American Dietetic Association 1974;64:624-629.

56 Borghi A, Bedogni G, Rocchi E, Seven S, Farina F, Battistini N. Multi-frequency bioelectric impedance measurements for predicting body water compartments in patients with non-ascitic liver cirrhosis. British Journal of Nutrition 1996;76:325-332. 283

57 Bramley P, Oldroyd B, Stewart S, Simpson M, Truscott J, Losowsky M, Smith M. Body composition analysis in liver cirrhosis. The measurement of body fat by dual energy X-ray absorptiometry in comparison to skinfold anthropometry, bioelectrical impedance and total body potassium. Basic Life Sciences 1993;60:211-214.

58 Braunfeld MY, Chan S, Pregler J, Neelakanta G, Sopher MJ, Busuttil RW, Csete M. Liver transplantation in the morbidly obese. Journal of Clinical Anesthesia 1996;8:585-590.

59 Brook COD. Determination of body composition of children from skinfold measurements. Archives of Disease in Childhood 1971;46:182-184.

60 Brooke-Wavell K, Jones PR, Norgan NG, Hardman AE. Evaluation of near infra-red interactance for assessment of subcutaneous and total body fat. European Journal of Clinical Nutrition 1995;49:57-65.

61 Brown L, ed. The new shorter Oxford English dictionary on historical principles. Oxford: Clarendon Press 1993;2:2037.

62 Browne EG. Arabian Medicine. London: Cambridge University Press 1921:1-32.

63 Brozek J, Grande F, Anderson JT, Keys A. Densitometric analysis of body composition: Revision of some quantitative assumptions. Annals of the New York Academy of Sciences 1963;110:113-140.

64 Brozek J, Keys A. Evaluation of leanness-fatness in man: A survey of methods. Nutrition Abstracts and Reviews 1950;20:247-256.

65 Bruce Â, Andersson M, Arvidsson B, Isaksson B. Body composition. Prediction of normal body potassium, body water and body fat in adults on the basis of body height, body weight and age. Scandinavian Journal of Clinical and Laboratory Investigation 1980;40:461-473.

66 Bugianesi E, KaUian SC, Marchesini G, McCullough AJ. Increased contribution of gluconeogenesis to basal and stimulated glucose production in cirrhosis. Hepatology 1997;26:487A( Abstract 1433).

67 Bunout D, Gattas V, Iturriaga H, Pérez C, Pereda T, Ugarte G. Nutritional status of alcoholic patients: Its possible relationship to alcoholic liver damage. American Journal of Clinical Nutrition 1983;38:469-473.

68 Burch RE, Sackin DA, Ursick JA, Jetton MM, Sullivan JF. Decreased taste and smell acuity in cirrhosis. Archives of Internal Medicine 1978;138:743-746. 284

69 Burke BS. The dietary history as a tool in research. Journal of the American Dietetic Association 1947;23:1041-1046.

70 Burmeister W. Human body composition as related to surface area. European Journal of Pediatrics 1980;135:147-151.

71 Burroughs AK, Bosch J. Clinical manifestations and management of bleeding episodes in cirrhotics. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;1:408-425.

72 Burroughs AK, Mezzanote G, Phillips A, McCormick PA, McIntyre N. Cirrhotics with variceal hemorrhage. The importance of the time interval between admission and the start of analysis for survival and rebleeding rates. Hepatology 1989;9:801-807.

73 Bursztein S, Elwyn DH, Askanazi J, Kinney JM. Energy metabolism, indirect calorimetry, and nutrition. Baltimore: Williams & Wilkins 1989.

74 Buskirk ER. Underwater weighing and density. A review of procedures. In: Brozek J, Henschel A, eds. Techniques for measuring body composition. Washington DC: National Academy of Sciences, National Research Council 1961:90-116.

75 Buzby GP, Mullen JL, Matthews DC, Hobbs CL, Rosato EF. Prognostic nutritional index in gastrointestinal surgery. American Journal of Surgery 1980;139:160-166.

76 Cabré E, de Leon R, Planas R, Bertrân X, Domènech E, Gassull MA. Fiabilidad del analisis de impedancia bioeléctrica como método de monitorizaciôn nutricional en la cirrosis con ascitis. Gastroenterologia y Hepatologia 1995;18:359-365.

77 Cabre E, Gonzalez-Huix F, Abad-Lacruz A, Esteve M, Acero D, Femandez-Bahares F, Xiol X, Gassull MA. Effect of total enteral nutrition on the short-term outcome of severely malnourished cirrhotics. A randomized controlled trial. Gastroenterology 1990;98:715-720.

78 Cahill GF Jr. Starvation: Some biological aspects. In: Kinney JM, Jeejeebhoy KN, Hill GH, Owen OE, eds. Nutrition and metabolism in patient care. Philadelphia: WB Saunders Company 1988:193-204.

7 9 Cahill KM. Platonic concepts of hepatology. Archives of Internal Medicine 1963 ; 111:819-822.

80 Camilo ME, Siqueira MC, Bivar L, Pinto Correira J. Body composition assessment in cirrhosis with ascites. Clinical Nutrition 1983;Sl:072(Abstract). 285

81 Campbell IT, Watt T, Withers D, England R, Sukumar S, Keegan MA, Faragher B, Martin DF. Muscle thickness measured with ultrasound may be an indicator of lean tissue wasting in multiple organ failure in the presence of edema. American Journal of Clinical Nutrition 1995;62:533-539.

82 Campillo B, Bories PN, Devanlay M, Sommer F, Wirquin E, Fouet P. The thermogenic and metabolic effects of food in liver cirrhosis: Consequences on the storage of nutrients and the hormonal counterregulatory response. Metabolism 1992;41:476-482.

83 Campillo B, Bories PN, Fouet P. Postprandial de novo lipogenesis in alcoholic liver cirrhosis: Relationship with fuel homeostasis and nutritional status. European Journal of Clinical Nutrition 1993;47:640-647.

84 Campillo B, Bories PN, Leluan M, Pomin B, Devanlay M, Fouet P. Short-term changes in energy metabolism after 1 month of a regular oral diet in severely malnourished cirrhotic patients. Metabolism 1995;44:765-770.

85 Campillo B, Bories P, Pomin B, Devanlay M, Linsker S, Guillemin A, Wirquin E, Fouet P. Dépenses énergétiques et utilisation des nutriments chez le cirrhotique à jeun et au repos. Influence de l’hépatite alcoolique et du score de gravité de la maladie. Gastroentérologie Clinique et Biologique 1989;13:544-550.

86 Campillo B, Sommer F, Bories PN, Wirquin E, Devanlay M, Pomin B, Fouet P. Nutritional and metabolic consequences of basal hyperinsulinemia in alcoholic liver cirrhosis: Relationship with postprandial changes in erythrocyte insulin-receptor affinity. Nutrition 1994;10:532-537.

87 Cannon PR, Wissler RW, Woohidge RL, Benditt EP. The relationship of protein deficiency to surgical infection. Annals of Surgery 1944;120:514-525.

88 Caregaro L, Alberino F, Amodio P, Merkel C, Bolognesi M, Angeli P, Gatta A. Malnutrition in alcoholic and vims-related cirrhosis. American Journal of Clinical Nutrition 1996;63:602-609.

89 Caregaro L, Menon F, Angeli P, Amodio P, Merkel C, Bortoluzzi A, Alberino F, Gatta A. Limitations of semm creatinine level and creatinine clearance as filtration markers in cirrhosis. Archives of Internal Medicine 1994;154:201-205.

90 Carter EA, Wands JR. Ethanol-induced inhibition of liver cell function: 1. Effect of ethanol on hormone stimulated hepatocyte DNA synthesis and the role of ethanol metabolism. Alcoholism Clinical and Experimental Research 1988;12:555-562. 286

91 Casati A, Colombo S, Leggieri C, Muttini S, Capocasa T, Gallioli G. Measured versus calculated energy expenditure in pressure support ventilated ICU patients. Minerva Anestesiologica 1996;62:165-170.

92 Chami TN, Schuster MM, Bohlman ME, Pulliam TJ, Kamal N, Whitehead WE. A simple radiologic method to estimate the quantity of bowel gas. American Journal of Gastroenterology 1991;86:599-602.

93 Chandra RK. Immunocompetencein undemutrition. Journal of Pediatrics

94 Chandra RK. Rosette-forming T lymphocytes and cell-mediated immunity in malnutrition. British Medical Journal 1974;3:608-609.

95 Chandra RK. Nutritional deficiency and susceptibility to infection. Bulletin of the World Health Organisation 1979;57:167-177.

96 Chandra RK. Serum thymic hormone activity in protein-energy malnutrition. Clinical and Experimental Immunology 1979;38:228-230.

97 Chandra RK. Immunocompetence as a functional index of nutritional status. British Medical Bulletin 1981;37:89-94.

98 Chang WK, Chao YC, Tang HS, Lang HE, Hsu CT. Effects of extra-carbohydrate supplementation in the late evening on energy expenditure and substrate oxidation in patients with liver cirrhosis. Journal of Parenteral and Enteral Nutrition 1997;21:96-99.

99 Chapman A. Current theory and practice: A study of pre-operative fasting. Nursing Standard 1996;10:33-36.

100 Chapman RW, Angus PW. Effect of liver disease on the gastrointestinal tract. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;2:1240-1245.

101 Chen WJ, Chung YC. Energy expenditure in patients with hepatocellular carcinoma. Cancer 1994;73:590-595.

102 Chiba T, Lloyd DA, Bowen A, Condon-Meyers A. Ultrasonography as a method of nutritional assessment. Journal of Parenteral and Enteral Nutrition 1989;13:529-534. 287

103 Child CG 111, Turcotte JO. Surgery and portal hypertension. In: Child CG III, ed. The liver and portal hypertension. Philadelphia: WB Saunders Company 1964; 1-85.

104 Chin SE, Shepherd RW, Thomas BJ, Cleghom GJ, Patrick MK, Wilcox JA, Ong TH, Lynch SV, Strong R. The nature of malnutrition in children with end-stage liver disease awaiting orthotopic liver transplantation. American Journal of Clinical Nutrition 1992;56:164-168.

105 Christensen E, Schlichting P, Kragh Andersen P, Fauerholdt L, Schou G, Vestergaard Pedersen B, Juhl E, Poulsen H, Tygstrup N, Copenhagen Study Group for Liver Diseases. Updating prognosis and therapeutic effect evaluation in cirrhosis with Cox’ s multiple regression model for time-dependent variables. Scandinavian Journal of Gastroenterology 1986;21:163-174.

106 Cohn SH, Palmer HE. Recent advances in whole body counting: A review. International Journal of Nuclear Medicine and Biology 1974;1:155-165.

107 Cohn SH, Vaswani AN, Yasumura S, Yuen K, Ellis KJ. Assessment of cellular mass and lean body mass by noninvasive nuclear techniques. Journal of Laboratory and Clinical Medicine 1985;105:305-311.

108 Compston J. The effect of liver disease on bone. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;2:1263-1272.

109 Conn HO. Adverse reactions and side effects of lactulose and related agents. In: Conn HO, Bircher J, eds. Hepatic encephalopathy. Syndromes and therapies. Bloomington: Medi-Ed Press 1994:299-310.

110 Conway JM, Norris KH, Bodwell CE. A new approach for the estimation of body composition: Infrared interactance. American Journal of Clinical Nutrition 1984;40:1123-1130.

111 Coss-Bu JA, Jefferson LS, Walding D, Yadin D, O’Brian Smith E, Klish WJ. Resting energy expenditure in children in a pediatric intensive care unit: Comparison of Harris-Benedict and Talbot predictions with indirect calorimetry values. American Journal of Clinical Nutrition 1998;67:74-80.

112 Coward WA, Prentice AM. Isotope method for the measurement of carbon dioxide production rate in man. American Journal of Clinical Nutrition 1985;41:659-661.

113 Crawford DHG, Cuneo RC, Shepherd RW. Pathogenesis and assessment of malnutrition in liver disease. Journal of Gastroenterology and Hepatology 1993;8:89-94. 288

114 Crawford DHG, Halliday JW, Cooksley WOE, Murphy TL, Golding SD, Wallace JD, Cuneo RC, Lynch SV, Strong RJ, Powell LW. Distribution of body water in patients with cirrhosis: The effect of liver transplantation. Hepatology 1993;17:1016-1021.

115 Crawford DHG, Shepherd RW, Halliday JW, Cooksley GWGE, Golding SD, Cheng WSC, Powell LW. Body composition in nonalcoholic cirrhosis: The effects of disease etiology and severity on nutritional compartments. Gastroenterology 1994;106:1611-1617.

116 Crosbie OM, Curry M, Freaney R, McKenna M, Hegarty JE. Bone loss in chronic liver disease and following liver transplantation. Hepatology 1997;26:347A(Abstract 876).

117 Cuddüiy A, Whelton MJ. Infertility in chronic active hepatitis. Hepatology 1987;7:1146(Abstract 496).

118 Daly JM, Heymsfield SB, Head CA, Harvey LP, Nixon DW, Katzeff H, Grossman GD. Human energy requirements: Overestimation by widely used prediction equation. American Journal of Clinical Nutrition 1985;42:1170-1174.

119 Daly JM, Lieberman MD, Goldfine J, Shou J, Weintraub F, Rosato EF, Lavin P. Enteral nutrition with supplemental arginine, RNA, and omega-3 fatty acids in patients after operation: Immunologic, metabolic, and clinical outcome. Surgery 1992;112:56-67.

120 Daly JM, Thom AK. Neoplastic diseases. In: Kinney JM, Jeejeebhoy KN, Hill GH, Owen OE, eds. Nutrition and metabolism in patient care. Philadelphia: WB Saunders Company 1988:567- 587.

121 Daniel PM, Pratt OE, Spargo E. The metabolic homeostatic role of muscle and its function as a store of protein. Lancet 1977;2:446-448.

122 Davidson CS. Cirrhosis of the liver treated with prolonged sodium restrictions. Journal of the American Medical Association 1955;159:1257-1261.

123 Davidson S, Passmore R, Brock JF, Truswell AS. Human nutrition and dietetics. 6th edn. Edinburgh: Churchill Livingstone 1975:15-42.

124 DeakinV. Assessment of nutritional status of athletes. In: Burke L, Deakin V, eds. Clinical sports nutrition. Sydney: McGraw-Hill Book Company 1994:38-73.

125 Deems RO, Friedman LS, Friedman MI, Munoz SJ, Deems DA, Maddrey WC. Relationship between biochemical tests and dietary intake in patients with liver disease. Journal of Clinical Gastroenterology 1994;18:304-308. 289

126 Deems RO, Friedman MI, Friedman LS, Munoz SJ, Maddrey WC. Chemosensory function, food preferences and appetite in human liver disease. Appetite 1993;20:209-216.

127 Dempster P, Aitkens S. A new air displacement method for the determination of human body composition. Medicine and Science in Sports and Exercise 1995;27:1692-1697.

128 Department of Health. Dietary reference values for food energy and nutrients for the United Kingdom. Report of the Panel on Dietary Reference Values of the Committee on Medical Aspects of Food Policy. Report of health and social subjects, 41. London: Her Majesty’s Stationary Office 1991.

129 Department of Health. Sensible drinking. The report of an inter-departmental working group. London: Her Majesty’s Stationary Office 1995.

130 Department of Health and Social Security. Recommended daily amounts of food energy and nutrients for groups of people in the United Kingdom. Report by the Committee on Medical Aspects of Food Policy. Report of health and social subjects, 15. London: Her Majesty’s Stationary Office 1979.

131 Detsky AS, McLaughlin JR, Baker JP, Johnston N, Whittaker S, Mendelson RA, Jeej eebhoy KN. What is subjective global assessment of nutritional status ? Journal of Parenteral and Enteral Nutrition 1987;11:8-13.

132 Deurenberg P, van der Kooij K, Evers P, Hulshof T. Assessment of body composition by bio electrical impedance in a population aged >60 y. American Journal of Clinical Nutrition 1990;51:3-6.

133 Deurenberg P, Weststrate JA, Seidell JC. Body mass index as a measure of fatness: Age- and sex-specific prediction formulas. British Journal of Nutrition 1991 ;65:105-114.

134 DiCecco SR, Wieners EJ, Wiesner RH, Southom PA, Plevak DJ, Krom RAF. Assessment of nutritional status of patients with end-stage liver disease undergoing liver transplantation. Mayo Clinic Proceedings 1989;64:95-102.

135 Dickerson RN, White KG, Curcillo PG 11, King SA, Mullen JL. Resting energy expenditure of patients with gynecologic malignancies. Journal of the American College of Nutrition 1995;14:448-454.

136 Doekel RC, Zwillich CW, Scoggin CH, Kryger M, Weil JV. Clinical semi-starvation: Depression of hypoxic ventilatory response. New England Journal of Medicine 1976;295:358- 361. 290

137 Dolz C, Raurich JM, Ibanez J, Obrador A, Marsé P, Gaya J. Ascites increases the resting energy expenditure in liver cirrhosis. Gastroenterology 1991;100:738-744.

138 Douglas CG. A method for determining the total respiratory exchange in man. Journal of Physiology 1911;42:xvii-xviii.

139 Du Bois D, Du Bois EF. Clinical calorimetry. A formula to estimate the approximate surface area if height and weight be known. Archives of Internal Medicine 1916;17:863-871.

140 Dumin JVGA, Blake EC, Brockway JM, Drury EA. The food intake and energy expenditure of elderly women living alone. British Journal of Nutrition 1961;15:499-506.

141 Dumin JVGA, Passmore R. Energy work and leisure. London: Heinemann 1967.

142 Dumin JVGA, Rahaman MM. The assessment of the amount of fat in the human body from measurements of skinfold thickness. British Journal of Nutrition 1967;21:681-689.

143 Dumin JVGA, Womersley J. Body fat assessed from total body density and its estimation from skinfold thickness: Measurements on 481 men and women aged 16 to 72 years. British Journal of Nutrition 1974;32:77-97.

144 Eastell R, Dickson ER, Hodgson SF, Wiesner RH, Porayko MK, WahnerHW, Cedel SL, Riggs BL, Krom RAF. Rates of vertebral bone loss before and after liver transplantation in women with primary biliary cirrhosis. Hepatology 1991;14:296-300.

145 Eckhoff DE, Pirsch JD, D'Alessandro AM, Knechtle SJ, Young CJ, Geffner SR, Belzer FO, Kalayoglu M. Pretransplant status and patient survival following liver transplantation. Transplantation 1995;60:920-925.

146 Edwards DAW. Observations on the distribution of subcutaneous fat. Clinical Science 1950;9:259-270.

147 Edwards RHT. Physiological analysis of skeletal muscle weakness and fatigue. Clinical Science and Molecular Medicine 1978;54:463-470.

148 Efron G. Abdominal wound dismption. Zawcei 1965;1:1287-1291.

149 Eghtesad B, Bronsther O, Irish W, Casavilla A, Abu-Elmagd K, Van Thiel D, Tzakis A, Fung JJ, Starzl TE. Disease gravity and urgency of need as guidelines for liver allocation. Hepatology 1994;20:56S-62S. 291

150 Elia M. Artificial nutrition support. Medicine International \990-,^T3391-3396.

151 Elia M. Body composition analysis: An evaluationof 2 component models, multicomponent models and bedside techniques. Clinical Nutrition 1992;11:114-117.

152 Elia M, Fuller NJ, Murgatroyd PR. Measurement of bicarbonate turnover in humans: Applicability to estimation of energy expenditure. American Journal of Physiology 1992;263:E676-687.

153 Elia M, Parkinson SA, Diaz E. Evaluation of near infra-red interactance as a method for predicting body composition. European Journal of Clinical Nutrition 1990;44:113-121.

154 Enia G, Sicuso C, Alati G, Zoccali C. Subjective global assessment of nutrition in dialysis patients. Nephrology, Dialysis, Transplantation 1993;8:1094-1098.

155 Eriksson LS, Thome A, Wahren J. Diet-induced thermogenesis in patients with liver cirrhosis. Clinical Physiology 1989;9:131-141.

156 Fanelli MT, Kuczmarski RJ. Ultrasound as an approach to assessing body composition. American Journal of Clinical Nutrition 1984;39:703-709.

157 Fearon KCH, Richardson RA, Hannan J, Cowan S, Watson W, Shenkin A, Garden OJ. Bioelectiical impedance analysis in the measurement of the body composition of surgical patients. British Journal of Surgery 1992;79:421-423.

158 Feller RB, McDoinald JA, Sherbon K, McCaughan GW. Evidence of continuing bone recovery at a mean of 7 years after liver transplantation. Hepatology 1997;26:347A(Abstract 875).

159 FerranniniE. The theoreticalbases of indirect calorimetry: a review. 1988;37:287- 301.

160 Ferraro R, Lillioja S, Fontvieille AM, Rising R, Bogardus C, Ravussin E. Lower sedentary metabolic rate in women compared with men. Journal of Clinical Investigation 1992;90:780- 784.

161 Flatt JP. On the maximal possible rate of ketogenesis. Diabetes 1972;21:50-53.

162 Forbes GB. Human body composition - growth, aging, nutrition, and activity. New York: Springer Verlag 1987:4-100. 292

163 Forbes GB, Bruining GJ. Urinary creatinine excretion and lean body mass. American Journal of Clinical Nutrition 1976;29:1359-1366.

164 Forbes GB, Gallup J, Hursh JB. Estimation of total body fat from potassium-40 content. Science 1961;133:101-102.

165 Forse RA. Comparison of gas exchange measurements with a mouthpiece, face mask, and ventilated canopy. Journal of Parenteral and Enteral Nutrition 1993;17:388-391.

166 Foster JH. History of liver surgery. Archives of Surgery \99\,\26..

167 Fowles FE. Smithsonian Physical Tables. 8th edn. Washington DC: Smithsonian Institution 1934.

168 Fox RA, Scheuer PJ, Sherlock S, Sharma O, James DG. Delayed hypersensitivity in primary biliary cirrhosis. Gut 1968;9:729(Abstract).

169 Franco D, Belghiti J, Cortesse A, Boucquey BM, Charra M, Lacaine F, Bismuth H. Nutrition et immunité du cirrhotique alcoolique. Gastroentérologie Clinique et Biologique 1981;5:839- 846.

170 Franco D, Charra M, Jeambrun P, Belghiti J, Cortesse A, Sossler C, Bismuth H. Nutrition and immunity after peritoneovenous drainage of intractable ascites in cirrhotic patients. American Journal of Surgery 1983;146:652-657.

171 Fredrix EWHM, Soeters PB, Deerenberg IM, Rester ADM, von Meyenfeldt MF, Satis WHM. Resting and sleeping energy expenditure in the elderly. European Journal of Clinical Nutrition 1990;44:741-747.

172 Friedl KE, DeLuca JP, Marchitelli LJ, Vogel JA. Reliability of body-fat estimations from a four-compartment model by using density, body water, and bone mineral measurements. American Journal of Clinical Nutrition 1992;55:764-770.

173 Friedman Al, Macewan G. Small bowel absorption in portal cirrhosis with ascites. American Journal of Gastroenterology 1963;39:114-122.

174 Friedman MI, Preti G, Deems RO, Friedman LS, Munoz SJ, Maddrey WC. Limonene in expired lung air of patients with liver disease. Digestive Diseases and Sciences 1994;39:1672- 1676. 293

175 Fries JA. Intestinal gases of man. American Journal of Physiology 1906;16:468-474.

176 Frisancho AR. Triceps skin fold and upper arm muscle size norms for assessment of nutritional status. American Journal of Clinical Nutrition 1974;27:1052-1057.

177 Frisancho AR. New norms of upper limb fat and muscle areas for assessment of nutritional status. American Journal of Clinical Nutrition 1981;34:2540-2545.

178 Frisancho AR. New standards of weight and body composition by frame size and height for assessment of nutritional status of adults and the elderly. American Journal of Clinical Nutrition 1984;40:808-819.

179 Frost GS, Corish C. Reproducibility of upper-arm anthropometry in subjects of differing body mass. Journal of Human Nutrition and Dietetics 1989;2:403-406.

180 Fukagawa NK, Bandini LG, Young JB. Effect of age on body composition and resting metabolic rate. American Journal of Physiology 1990;259:E233-238.

181 Fuller NJ, Jebb SA, Goldberg GR, Pullicino E, Adams C, Cole TJ, Elia M. Inter-observer variability in the measurement of body composition. European Journal of Clinical Nutrition 1991;45:43-49.

182 Fuller NJ, Jebb SA, Laskey MA, Coward WA, Elia M. Four-component model for the assessment of body composition in humans: Comparison with alternative methods, and evaluation of the density and hydration of fat-free mass. Clinical Science 1992;82:687-693.

183 Gabuzda GJ. Cirrhosis, ascites, and edema. Gastroenterology \910,5^\5^6-553.

184 Gamble JL Jr, Robertson JS, Hannigan CA, Foster CG, Farr LE. Chloride, bromide, sodium, and sucrose spaces in man. Journal of Clinical Investigation 1953;32:483-489.

185 Gam SM, Nolan P, Jr. A tank to measure body volume by water displacement (BOVOTA). Annals of the New York Academy of Sciences 1963;110:91-95.

186 Garrett-Laster M, Russell RM, Jacques PF. Impairment of taste and olfaction in patients with cirrhosis: The role of vitamin A. Human Nutrition: C/z«/ca/Aw/r/Y/ow 1984;38C:203-214.

187 Garrison RN, Cryer HM, Howard DA, Polk HC. Clarification of risk factors for abdominal operations in patients with hepatic cirrhosis. Annals of Surgery 1984;199:648-655. 294

188 Garrow JS. Weight penalties. British MedicalJournal \919-,2:\\1\-\\12.

189 Garrow JS. Resting metabolic rate as a determinant of energy expenditure in man. In: Garrow JS, Halliday D, eds. Substrate and energy metabolism. London: John Libbey 1985:102-107.

190 Gauthier A, Levy VG, Quinton A, Michel H, Rueff B, Descos L, Durbec JP, Fermanian J, Lancrenon S, the ENT AC group. Salt or no salt in the treatment of cirrhotic ascites: A randomised study. Gwi 1986;27:705-709.

191 Gerok W, Gross V. Secretory proteins; synthesis, secretion and function. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;1:174-198.

192 Gibson SA, O’Sullivan KR. Breakfast cereal consumption patterns and nutrient intakes of British schoolchildren. Journal of the Royal Society of Health 1995;115:366-370.

193 Ginés P, Arroyo V, Quintero E, Planas R, Bory F, Cabrera J, Rimola A, Viver J, Camps J, Jimenez W, Mastai R, Gaya J, Rodés J. Comparison of paracentesis and diuretics in the treatment of cirrhosis with tense ascites. Results of a randomized study. Gastroenterology 1987;93:234-241.

194 Gnaedinger RH, Reineke EP, Pearson AM, Van Huss WD, Wessel JA, Montoye HJ. Determination of body density by air displacement, helium dilution, and underwater weighing. Annals of the New York Academy of Sciences 1963;110:96-108.

195 Going SB, Massett MP, Hall MC, Bare LA, Root PA, Williams DP, Lohman TG. Detection of small changes in body composition by dual-energy X-ray absorptiometry. American Journal of Clinical Nutrition 1993;57:845-850.

196 Gottdiener JS, Gross HA, Henry WL, Borer JS, Ebert MH. Effects of self-induced starvation on cardiac size and function in anorexia nervosa. Circulation 1978;58:425-433.

197 Graham DY, Smith JL. The course of patients after variceal haemorrhage. Gastroenterology 1981;80:800-809.

198 Gray GE, Gray LK. Validity of anthropometric norms used in the assessment of hospitalized patients. Journal of Parenteral and Enteral Nutrition 1979;3:366-368.

199 GraystoneJE. Creatinine excretion during growth. In: Cheek DB, ed. Human growth: Body composition, cell growth, energy and intelligence. Philadelphia: Lea and Febiger 1968:192. 295

200 Green JH, Bramley PN, Losowsky MS. Are patients with primary biliary cirrhosis hypermetabolic ? A comparison between patients before and after liver transplantation and controls. Hepatology 1991;14:464-472.

201 Greenwald AJ, Allen TH, Bancroft RW. Abdominal gas volume at altitude and at ground level. Journal of Applied Physiology 1969;26:177-181.

202 Grimble G. Fibre, fermentation, flora, and flatus. Gut 1989;30:6-13.

203 Guggenheim KY. Paracelsus and the science of nutrition in the renaissance. On the occasion of the 500th anniversary of his birth. Journal of Nutrition 1993;123:1189-1194.

204 Guglielmi FW, Contento F, Laddaga L, Panella C, Francavilla A. Bioelectrical impedance analysis: Experience with male patients with cirrhosis. Hepatology 1991 ; 13:892-895.

205 Haas LF. François Magendie (1783-1855). Journal of Neurology, Neurosurgery and Psychiatry 1994;57:692.

206 Halliday AW, Benjamin IS, Blumgart LH. Nutritional risk factors in major hepatobiliary surgery. Journal of Parenteral and Enteral Nutrition 1988;12:43-48.

207 Halpem SL, Glenn MB, Goodhart RS. New height-weight tables. Journal of the American Medical Association 1960,173:1576.

208 Harris JA, Benedict TG. Biometric studies of basal metabolism in man. Washington, DC: Carnegie Institute of Washington, Publication No 279 1919.

209 Harrison JD, McKieman J, Miki C, Neuberger J and McMaster P. The effect of recipient nutrition on outcome in elective orthotopic liver transplantation (OLT). Hepatology 1993;18:333A (Abstract).

210 Harrison JE, McNeill KG, Strauss AL. A nitrogen index -total body protein normalized for body size - for diagnosis of protein status in health and disease. Nutrition Research 1984;4:209- 224.

211 Hasse JM. Nutrition considerations in liver transplantation. Topics in Clinical Nutrition 1992;7:24-33.

212 Hasse JM, Redus AM. Nutritional status and nutritional support in orthotopic liver transplant (OLT) patients are associated with length of stay. Journal of the American Dietetic Association 1991;91S:A20(Abstract). 296

213 Hasse J, Strong S, Gorman MA, Liepa G. Subjective global assessment; Alternative nutrition- assessment technique for liver transplant candidates. Nutrition 1993;9:339-343.

214 Hawkins FG, Léon M, Lopez MB, Valero MA, Larrodera L, Garcia-Garcial, Loinaz C, Moreno Gonzalez E. Bone loss and turnover in patients with liver transplantation. Hepato- Gastroenterology 1994;41:158-161.

215 Heaton KW. Bile salts. In: Wright R, Millward-Sadler GH, Alberti KGMM, Karran S, eds. Liver and biliary disease. Pathophysiology, diagnosis, management. London: Baillière Tindall WB Saunders Company 1985:277-299.

216 Hébuteme X, Hastier P, Péroux JL, Zeboudj N, Delmont JP, Rampai P. Resting energy expenditure in patients with alcoholic chronic pancreatitis. Digestive Diseases and Sciences 1996;41:533-539.

217 Hehir DJ, Jenkins RL, Bistrian BR, Blackburn GL. Nutrition in patients undergoing orthotopic liver transplant. Journal of Parenteral and Enteral Nutrition 1985;9:695-700.

218 Henkin RI, Smith FR. Hyposmia in acute viral hepatitis. Lancet 1971;1:823-826.

219 Henry CJK. Influence of body composition on protein and energy requirements: Some new insights. In: Davies PSW, Cole TJ, eds. Body composition techniques in health and disease. Cambridge: Cambridge University Press 1995:85-99.

220 Herrmann R, McIntyre N. Amino-acid metabolism, urea production and pH regulation. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;1:157-174.

221 Heymsfield SB, Arteaga C, McManus C, Smith J, Moffitt S. Measurement of muscle mass in humans: Validity of the 24-hour urinary creatinine method. American Journal of Clinical Nutrition 1983;37:478-494.

222 Heymsfield SB, Bethel RA, Ansley JD, Gibbs DM, Felner JM, Nutter DO. Cardiac abnormalities in cachectic patients before and during nutritional repletion. American Heart Journal 1978;95:584-594.

223 Heymsfield SB, Lichtman S, Baumgartner RN, Wang J, Kamen Y, Aliprantis A, Person RN Jr. Body composition of humans: Comparison of two improved four-compartment models that differ in expense, technical complexity, and radiation exposure. American Journal of Clinical Nutrition 1990;52:52-58. 297

224 Heymsfield SB, McManus C, Smith J, Stevens V, Nixon DW. Anthropometric measurement of muscle mass: Revised equations for calculating bone-free arm muscle area. American Journal of Clinical Nutrition 1982;36:680-690.

225 Heymsfield SB, Waki M, Reinus J. Are patients with chronic liver disease hypermetabolic ? Hepatology 1990;11:502-505.

226 Higashiguchi T, Yokoi H, Noguchi T, Kawarada Y, Mizumoto R, Hasselgren PO. The preoperative nutritional assessment of surgical patients with hepatic dysfunction. Japanese Journal of Surgery 1995;25:113-118.

227 Hill GL, Blackett RL, Pickford 1, Burkinshaw L, Young GA, Warren JV, Schorah CJ, Morgan DB. Malnutrition in surgical patients. An unrecognised problem. Zawcet 1977;1:689-692.

228 Himes JH, Roche AF, Webb P. Fat areas as estimates of total body fat. American Journal of Clinical Nutrition 1980;33:2093-2100.

229 Hirsch S, de ObaldiaN, PetermannM, Rojo P, Barrientos C, IturriagaH, Bunout D. Subjective global assessment of nutritional status: Further validation. Nutrition 1991;7:35-37.

230 Hodgson SF, Dickson ER, Wahner HW, Johnson KA, Mann KG, Riggs BL. Bone loss and reduced osteoblast function in primary biliary cirrhosis. Annals of Internal Medicine 1985;103:855-860.

231 Holt TL, Cui C, Thomas BJ, Ward LC, Quirk PC, Crawford D, Shepherd RW. Clinical applicability of bioelectric impedance to measure body composition in health and disease. Nutrition 1994;10:221-224.

232 Horber FF, Thomi F, Casez JP, Fonteille J, Jaeger P. Impact of hydration status on body composition measured by dual energy X-ray absorptiometry in normal volunteers and patients on haemodialysis. British Journal of Radiology 1992;65:895-900.

233 Hurt RD, Higgins JA, Nelson RA, Morse RM, Dickson ER. Nutritional status of a group of alcoholics before and after admission to an alcoholism treatment unit. American Journal of Clinical Nutrition 1981;34:386-392.

234 Hussaini SH, Soo S, Stewart SP, Oldroyd B, Roman F, Smith MA, O'Grady JG, Losowsky MS. Steroid dose after liver transplantation is a risk factor for loss of lean body mass (LBM). Hepatology 1996;24:525A( Abstract 1596). 298

235 Italian Multicentre Cooperative Project on nutrition in liver cirrhosis. Nutritional status in cirrhosis. Journal of Hepatology 1994;21:317-325.

236 Jackson AS, Pollock ML. Prediction accuracy of body density, lean body weight, and total body volume equations. Medicine and Science in Sports 1977;9:197-201.

237 Jackson AS, Pollock ML. Generalized equations for predicting body density of men. British Journal of Nutrition 1978;40:497-504.

238 Jackson AS, Pollock ML, Ward A. Generalized equations for predicting body density of women. Medicine and Science in Sports and Exercise 1980;12:175-182.

239 Jacob RA. Assessment of human vitamin C status. Journal of Nutrition

240 Jebb SA, Elia M. Assessment of changes in total body water in patients undergoing renal dialysis using bioelectrical impedance analysis. Clinical Nutrition 1991;10:81-84.

241 Jebb SA, Elia M. Techniques for the measurement of body composition: A practical guide. International Journal of Obesity 1993;17:611-621.

242 Jebb SA, Elia M. Multi-compartment models for the assessment of body composition in health and disease. In: Davies PSW, Cole TJ, eds. Body composition techniques in health and disease. Cambridge: Cambridge University Press 1995:240-254.

243 Jebb SA, Goldberg GR, Elia M. DXA measurements of fat and bone mineral density in relation to depth and adiposity. Basic Life Sciences 1993;60:115-119.

244 Jeejeebhoy KN. Cause of hypoalbuminaemia in patients with gastrointestinal and cardiac disease. Lancet 1962;1:343-348.

245 Jeejeebhoy KN. The functional basis of assessment, hi: Kinney JM, Jeej eebhoy KN, Hill GL, Owen OE, eds. Nutrition and metabohsm in patient care. Philadelphia: WB Saunders Company 1988:739-751.

246 Jelliffe DB. The assessment of the nutritional status of the community (With special reference to field surveys in developing regions of the world). WHO Monograph 53. Geneva: WHO 1966.

247 Jennings G, Bluck LJC, Chowings C, Podesta D, Elia M. Evaluation of an infra-red method for the determination of total body water in a clinical context. Clinical Nutrition 1995;14:S2:53- 54(Abstract P74). 299

248 Jensen MD. Research techniques for body composition assessment. Journal of the American Dietetic Association 1992;92:454-460.

249 Jensen MD, Braun JS, Vetter RJ, Marsh HM. Measurement of body potassium with a whole- body counter: Relationship between lean body mass and resting energy expenditure. Mayo Clinic Proceedings 1988;63:864-868.

250 Jhangiani SS, Agarwal N, Holmes R, Cayten CG, Pitchumoni CS. Energy expenditure in chronic alcoholics with and without liver disease. American Journal of Clinical Nutrition 1986;44:323-329.

251 John WJ, Phillips R, Ott L, Adams LJ, McClain CJ. Resting energy expenditure in patients with alcoholic hepatitis. Journal of Parenteral and Enteral Nutrition 1989;13:124-127.

252 Johnson MW, Washburn WK, Freeman RE, FitzMaurice SE, Dienstag J. Basgoz N, Jenkins RL, Cosimi AB. Hepatitis C viral infection in liver transplantation. Archives of Surgery 1996;131:284-291.

253 Joint FAOAVHO/UN Expert consultation. Energy and protein requirements. Technical Report Series 724. Geneva: WHO 1985.

254 Jones PRM, Norgan NG. A simple system for the determination of human body density by underwater weighing. Journal of Physiology 1974;239:71P-73P.

255 Kalman DR, Saltzman JR. Nutrition status predicts survival in cirrhosis. Nutrition Reviews 1996;54:217-219.

256 Keys A, Brozek J. Body fat in adult man. Physiological Reviews \9Sy,?>?)\lA5-?)25.

257 Keys AJ, Brozek J, Henschel O, Michelson O, Taylor HL. In: The biology of human starvation. Minneapolis, MN: University of Minnesota Press 1950:140-160.

258 Keys A, FidanzaF, Karvonen MJ, KimuraN, Taylor HL. Indices of relative weight and obesity. Journal of Chronic Diseases 1972;25:329-343.

259 Keys A, Taylor HL, Grande F. Basal metabolism and age of adult man. Metabolism 1973;22:579-587.

260 Kinney JM. Energy metabolism: Heat, fuel and life. In: Kinney JM, JeejeebhoyKN, Hill GH, Owen OE, eds. Nutrition and metabolism in patient care. Philadelphia: WB Saunders Company 1988:3-34. 300

261 Klausen B, Toubro S, Astrup A. Age and sex effects on energy e?q)enditure. American Journal of Clinical Nutrition 1997;65:895-907.

262 Klidjian AM, Archer TJ, Foster KJ, Karran SJ. Detection of dangerous malnutrition. Journal of Parenteral and Enteral Nutrition 1982;6:119-121.

263 Klidjian AM, Foster KJ, Kammerling RM, Cooper A, Karran SJ. Relation of anthropometric and dynamometric variables to serious postoperative complications. British Medical Journal 1980;281:899-901.

264 Klijnsma MP, Cameron ML, Bums TP, McGuigan SM. Out-patient alcohol detoxification. Outcome after 2 months. Alcohol and Alcoholism 1995;30:669-673.

265 Knell AJ. Liver function and failure: The evolution of liver physiology. Journal of the Royal College of Physicians of London 1980;14:205-208.

266 Knox LS, Crosby LO, Feurer ID, Buzby GP, Miller CL, Mullen JL. Energy expenditure in malnourished cancer patients. Annals of Surgery 1983;197:152-162.

267 Kondrap J, Müller MJ. Energy and protein requirements of patients with chronic liver disease. Journal of Hepatology 1997;27:239-247.

268 Kondmp J, Nielsen K, Juul A. Effect of long-term refeeding on protein metabolism in patients with cirrhosis of the liver. British Journal of Nutrition 1997;77:197-212.

269 Krahn DD, Rock C, Dechert RE, Naim KK, Hasse SA. Changes in resting energy expenditure and body composition in anorexia nervosa patients during refeeding. Journal of the American Dietetic Association 1993;93:434-438.

270 Kravetz D, Bosch J, Arderiu M, PizcuetaMP, Rodés J. Hemodynamic effects of blood volume restitution following a hemorrhage in rats with portal hypertension due to cirrhosis of the liver: Influence of the extent of portal systemic shunting. Hepatology 1989;9:808-814.

271 Kmszynska Y, Williams N, Perry M, Home P. The relationship between insulin sensitivity and skeletal muscle enzyme activities in hepatic cirrhosis. Hepatology 1988;8:1615-1619.

272 Kruszynska YT, Harry DS, Fryer LG, McIntyre N. Lipid metabolism and substrate oxidation during intravenous fructose administration in cirrhosis. Metabolism 1994;43:1171-1181. 301

273 Kruszynska YT, Meyer-Alber A, Darakhshan F, Home PD, McIntyre N. Metabolic handling of orally administered glucose in cirrhosis. Journal of Clinical Investigation 1993;91:1057-1066.

274 Kruszynska YT, Meyer-Alber A, Wollen N, McIntyre N. Energy expenditure and substrate metabolism after oral fructose in patients with cirrhosis. Journal of Hepatology 1993;19:241- 251.

275 Ksiazyk J, Lyszkowska M, Kierkus J. Energy metabolism in portal hypertension in children. Nutrition 1996;12:469-474.

276 Kudsk KA, Croce MA, Fabian TC, Minard G, Tolley EA, Foret HA, Kuhl MR, Brown RO. Enteral versus parenteral feeding. Effects on septic morbidity after blunt and penetrating abdominal trauma. Annals of Surgery 1992;215:503-511.

277 Lafleur M. Lab elle F, Lemoyne M. Nutritional assessment in chronic liver disease patients: A comparison between skeletal muscle function and standard nutritional parameters. Nutrition Research 1996;16:545-553.

278 Lake JR. Changing indications for liver transplantation. Gastroenterology Clinics of North America 1993;22:213-229.

279 Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-174.

280 Laskey MA. Dual-energy X-ray absorptiometry and body composition. Nutrition \996',\2 A^- 51.

281 Laskey MA, Lyttle KD, Flaxman ME, Barber RW. The influence of tissue depth and composition on the performance of the Lunar dual energy X-ray absorptiometer whole-body scanning mode. European Journal of Clinical Nutrition 1992;46:39-45.

282 Lautz HU, Selberg O, Korber J, Bürger M, Muller MJ. Protein-caloiie malnutrition in liver cirrhosis. Clinical Investigator 1992;70:478-486.

283 Laven GT. Skin-fold compressibility. A measure of body composition and a source of error. American Journal of Clinical Nutrition 1983;37:727(Abstract).

284 Law DK, Dudiick SJ, Abdou Nl. Immunocompetence of patients with protein-caloiie malnutrition. The effects of nutrition repletion. Annals of Internal Medicine 1973;79:545-550. 302

285 Lean MEJ, Han TS, Deurenberg P. Predicting body composition by densitometry from simple anthropometric measurements. American Journal of Clinical Nutrition 1996;63:4-14.

286 Ledesma Castana F, Casafont Morencos F, Martin Ramos L, Sanchez Hernandez E, Ordonez Gonzalez J, Pons Romero F. Inflnencia del estado nutricional sobre las caracteristicasdel liquido ascitico en cirrosis hepatica alcoholica. Nutricion Hospitalaria 1996;11:238-244.

287 Leevy CM, Baker H, Ten Hove W, Frank O, Cherrick GR. B-Complex vitamins in liver disease of the alcoholic. American Journal of Clinical Nutrition 1965;16:339-346.

288 Lentner C, ed. In: Geigy Scientific Tables. Units of measurements, body fluids, composition. Basle: Ciba-Geigy Ltd 1981;1:115

289 Lentner C, ed. In: Geigy Scientific Tables. Physical chemistry, composition of blood, hematology, somatometric data. Basle: Ciba-Geigy Ltd 1984;3:78.

290 LeveenHH, Christoudias G, Moon IP, Luft R, Falk G, Grosberg S. Peritoneo-venous shunting for ascites. Annals of Surgery 1974;180:580-590.

291 Levine JA, Harris M, Morgan MY. The hypermetabolic state associated with chronic alcohol abuse. Hepatology 1994;20:318A(Abstract).

292 Levine JA, Morgan MA. Measurement of energy expenditure in man: A review of available methods. Journal of Nutritional Medicine 1990;1:325-343.

293 Levine JA, Morgan MY. Weighed dietary intakes in patients with chronic liver disease. Nutrition 1996;12:430-435.

294 Levine JA, Stanger LC, Morgan MY. Resting and post-prandial energy expenditure and fuel utilisation in alcoholic patients and control subjects. Hepatology 1988;8:1334(Abstract).

295 Levine JA, Stanger LC, Morgan MY. Resting and post-prandial energy expenditure and fuel utilisation in patients with compensated cirrhosis and control subjects. Clinical Nutrition 1989;8S:54(Abstract).

296 Levitt MD. Volume and composition of human intestinal gas determined by means of an intestinal washout technic. New England Journal of Medicine 1971;284:1394-1398.

297 Levitt MD, Bond JH, Levitt DG. Gastrointestinal gas. In: Johnson LR, ed. Physiology of the gastrointestinal tract. New York: Raven Press 1981:1301-1302. 303

298 Lieber CS, Jones DP, DeCarli LM. Effects of prolonged ethanol intake: Production of fatty liver despite adequate diets. Journal of Clinical Investigation 1965;44:1009-1021.

299 Lieberman FL, Reynolds TB. Plasma volume in cirrhosis of the liver: Its relation to portal hypertension, ascites and renal failure. Journal of Clinical Investigation 1967;46:1297-1308.

300 Lieberthal MM. The pharmacology of lactulose. In: Conn HO, Bircher J, eds. Hepatic Encephalopathy: Syndromes and therapies. Bloomington: Medi-Ed Press 1994:165-194.

301 Lifson N, Gordon GB, McClintock R. Measurement of total carbon dioxide production by means of Dj'^O. Journal of Applied Physiology 1955;7:704-710.

302 Llach J, Rimola A, NavasaM, Ginès P, Salmeron JM, Ginès A, Arroyo V, Rodés J. Incidence and predictive factors of first episode of spontaneous bacterial peritonitis in cirrhosis with ascites: Relevance of ascitic fluid protein concentration. Hepatology 1992;16:724-727.

303 Logan RFA, Ferguson A, Finlayson NDC, Weir DG. Primary biliary cirrhosis and coeliac disease. An association ? Zawcet 1978;1:230-233.

304 Loguerico C, Gentile S, Spinelli P, Orlando C, Del Vecchio Blanco C, Coltorti M. Studio nutrizionale del cirrotico. Confronte fra plicometria cutanea ed acqua corporea. Rivista Italiana di Nutrizione Parenterale ed Enterale 1988;6:23-27.

305 Loguerico C, Sava E, Marmo R, Del Vecchio Blanco C, Coltorti M. Malnutrition in cirrhotic patients: Anthropometric measurements as a method of assessing nutritional status. British Journal of Clinical Practice 1990;44:98-101.

306 Loguerico C, Sava E, Sicolo P, Castellano L, Narciso O. Nutritional status and survival of patients with liver cirrhosis: Anthropometric evaluation. Minerva Gastroenterologica e Dietologica 1996;42:57-60.

307 Lohman TG. Skinfolds and body density and their relation to body fatness: A review. Human Biology 1981;53:181-225.

308 Lohman TG. Advances in body composition assessment. Champaign: Human Kinetics Publishers 1992;3:7-23.

309 Lolli R, Marchesini G, Blanchi G, Fabbri A, Bugianesi E, Zoli M, Pisi E. Anthropometric assessment of the nutritional status of patients with liver cirrhosis in an Italian population. Italian Journal of Gastroenterology 1992;24:429-435. 304

310 Long CL, Schaffel N, Geiger JN, Schiller WR, Blakemore WS. Metabolic response to injury and illness; Estimation of energy and protein needs from indirect calorimetry and nitrogen balance. Journal of Parenteral and Enteral Nutrition 1979;3:452-456.

311 Lopes J, Russell DMcR, Whitwell J, Jeejeebhoy KN. Skeletal muscle function in malnutrition. American Journal of Clinical Nutrition 1982;36:602-610.

312 Lukaski HC. Methods for the assessment of human body composition: Traditional and new. American Journal of Clinical Nutrition 1987;46:537-556.

313 Lukaski HC, Bolonchuk WW, Hall CB, Siders WA. Validation of tetrapolar bioelectrical impedance method to assess human body composition. Journal of Applied Physiology 1986;60:1327-1332.

314 Lukaski HC, Johnson PE. A simple, inexpensive method of deterniining total body water using a tracer dose of D^O and infrared absorption of biological fluids. American Journal Clinical Nutrition 1985;41:363-370.

315 Lukaski HC, Johnson PE, Bolonchuk WW, Lykken 01. Assessment of fat-free mass using bio electrical impedance measurements of the human body. American Journal of Clinical Nutrition 1985;41:810-817.

316 Lukaski HC, Mendez J. Relationship between fat-free weight and urinary 3 -methylhistidine excretion in man. Metabolism 1980;29:758-761.

317 McCafferty M, Stone KA, Cashman LK, Camel S. Correlation of preoperative hand grip strength postoperative complications m liver transplant patients. Journal of the American Dietetic Association 1993;93S:A33(Abstract).

318 McCanceRA, Widdowson EM. A method of breaking down the body weights of living persons into terms of extracellular fluid, cell mass and fat, and some applications of it to physiology and medicine. Proceedings of the Royal Society of London 1951;138:115-130.

319 McCrory MA, Gomez TD, Bemauer EM, Mole PA. Evaluation of a new air displacement plethysmograph for measuring human body composition. Medicine and Science in Sports and Exercise 1995;27:1686-1691.

320 McCullough AJ, Mullen KD, KaUian SC. Measurement of total body and extracellular water in cirrhotic patients with and without ascites. Hepatology 1991;14:1102-1 111. 305

321 McCullough AJ, Mullen KD, Kalhan SC. Body cell mass and leucine metabolism in cirrhosis. Gastroenterology 1992;102:1325-1333.

322 McCullough AJ, Tavill AS. Disordered energy and protein metabolism in liver disease. Seminars in Liver Disease 1991;11:265-277.

323 McDermott VGM, Chapman ME, Gillespie I. Sedation and patient monitoring in vascular and interventional radiology. British Journal of Radiology 1993;66:667-671.

324 McIntyre N. Symptoms and signs of liver disease. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;1:273-290.

325 McIntyre N, Rosalki S. Biochemical investigations in the management of the liver. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodés J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;1:293-309.

326 MacKay EM, MacKay LL. The concentration of urea in the blood of normal individuals. Journal of Clinical Investigation 1927;4:295-306.

327 MacLean AR, Renwick C. Audit of pre-operative starvation. Anaesthesia 1993,AS:\64-166.

328 McNurlan MA, Tomkins AM, Garlick PJ. The effect of starvation on the rate of protein synthesis in rat liver and small intestine. Biochemical Journal 1979;178:373-379.

329 McWhirter JP, Pennington CR. Incidence and recognition of malnutrition in hospital. British Medical Journal 1994;308:945-948.

330 Madden A. The role of low fat diets in the management of gall-bladder disease. Journal of Human Nutrition and Dietetics 1992;5:267-273.

331 Madden AM, Bradbury W, Morgan MY. Taste perception in cirrhosis: Its relationship to circulating micronutrients and food preferences. Hepatology 1997;26:40-48.

332 Madden AM, Morgan MY. A comparison of skinfold anthropometry and bioelectrical impedance analysis for measuring percentage body fat in patients with cirrhosis. Journal of Hepatology 1994;21:878-883. 306

333 Maffeis C, Pinelli L, Zaffanello M, Schena F, lacumin P, Schütz Y. Daily energy expenditure in free-living conditions in obese and non-obese children: Comparison of doubly labelled water method and heart-rate monitoring. International Journal of Obesity and Related Metabolic Disorders 1995;19:671-677.

334 Marchesini G, Zoli M, Angiolini A, Dondi C, Bianchi FB, Pisi E. Muscle protein breakdown in liver cirrhosis and the role of altered carbohydrate metabolism. Hepatology 1981;1:294-299.

335 Marchesini G, Zoli M, Dondi C, Bianchi G, Cirulli M, Pisi E. Anticatabolic effect of branched- chain ammo acid-enriched solution in patients with liver cirrhosis. Hepatology 1982;2:420-425.

336 Marin GA, Clark ML, Senior JR. Studies of malabsorption occurring in patients with Laennec’s cirrhosis. Gastroenterology 1969;56:727-736.

337 Marks BL, Perkins KA, Metz KF, Epstein LH, Robertson RJ, Goss FL, Sexton JE. Effects of smoking status on content of caloric intake and energy expenditure. International Journal of Eating Disorders 1991;10:441-449.

338 MarrJW. Individual dietary surveys: Purposes and methods. World Review of Nutrition and Dietetics 1971;13:105-164.

339 Martin F, Peters TJ. Alcoholic muscle disease. Alcohol and Alcoholism 19S5;20:125-136.

340 Martines D, Martmes V, Pasini M, Cocco G, Lora L, Vamier M, Venier GB, Sammartano G, Naccarato R. Carbohydrate-induced thermogenesis in liver cirrhosis: Glucose vs fructose. Nutrition 1994;10:521-526.

341 Martinsen L, Kondrup J, Stilling B, Dossing H, Nielsen K. Energy expenditure measured with the doubly labelled water method in patients with alcoholic liver cirrhosis and ascites. Clinical Nutrition 1990;9:S:38(Abstract 37).

342 Mattes RD. Sensory influences on food intake and utilization in humans. Human Nutrition: Applied Nutrition 1987;41A:77-95.

343 Mattes-Kulig DA, Henkin RJ. Energy and nutrient consumption of patients with dysgeusia. Journal of the American Dietetic Association 1985;85:822-826.

344 Maxwell JD, Hodt P, Taylor WH. Portable chair for testing isometric muscle strength. Lancet 1984;1:18-19. 307

345 Mazess RB, Barden HS, Bisek JP, Hanson J. Dual-energy X-ray absorptiometry for total-body and regional bone-mineral and soft-tissue composition. American Journal of Clinical Nutrition 1990;51:1106-1112.

346 Meakins JL, Pietsch JB, Bubenick O, Kelly R, Rode H, Gordon J, MacLean LD. Delayed hypersensitivity: Indicator of acquired failure of host defences in sepsis and trauma. Annals of Surgery 1977;186:241-249.

347 Mendenhall CL. Protein-calorie malnutrition in alcoholic liver disease. In: Watson RR, Watzl B, eds. Nutrition and alcohol. Boca Raton: CRC Press 1992:363-384.

348 Mendenhall CL, Anderson S, WeesnerRE, Goldberg SJ, CrolicKA. Protein-calorie malnutrition associated with alcoholic hepatitis. American Journal of Medicine 1984;76:211-222.

349 Mendenhall CL, Moritz TE, Roselle GA, Morgan TR, Nemchausky BA, Tamburro CH, Schiff ER, McClain CJ, Marsano LS, Allen JI, Samanta A, WeesnerRE, Henderson W, Gartside P, Chen TS, French SW, Chedid A, Veterans Affairs Cooperative Study Group 275. A study of oral nutritional support with oxandrolone in malnourished patients with alcoholic hepatitis: Results of a Department of Veterans Affairs Cooperative Study. Hepatology 1993;17:564-576.

350 Mendenhall CL, Moritz TE, Roselle GA, Morgan TR, Nemchausky BA, Tamburro CH, Schiff ER, McClain CJ, Marsano LS, Allen JI, Samanta A, WeesnerRE, Henderson WG, Chen TS, French SW, Chedid A, VA Cooperative Study Group. Protein energy malnutrition in severe alcoholic hepatitis: Diagnosis and response to treatment. Journal of Parenteral and Enteral Nutrition 1995;19:258-265.

351 Mendenhall CL, Tosch T, Weesner RE, Garcia-Pont P, Goldberg SJ, Kieman T, Seeff LB, Sorrell M, Tamburro C, Zetterman R, Chedid A, Chen T, Rabin L. VA cooperative study on alcoholic hepatitis II : Prognostic significance of protein-calorie malnutrition. American Journal of Clinical Nutrition 1986;43:213-218.

352 Mendez J, Keys A. Density and composition of mammalian muscle. Metabolism Clinical and Experimental 1960;9:184-188.

353 Merli M, Riggio O, Dally L, Policentrica Italiana Nutrizione Cirrosi. Does malnutrition affect survival in cirrhosis ? Hepatology 1996;23:1041-1046.

354 Merli M, Riggio O, Romiti A, Ariosto F, Mango L, Pinto G, Savioli M, CapocacciaL. Basal energy production rate and substrate use in stable cirrhotic patients. Hepatology 1990;12:106-

112. 308

355 Merli M, Romiti A, Riggio O, Capocaccia L. Optimal nutritional indexes in chronic liver disease. Journal of Parenteral and Enteral Nutrition 1987;11;130S-134S.

356 Metropolitan height and weight tables. Statistical Bulletin of the Metropolitan Life Insurance Company (New York) 1983;64:3-9.

357 Meys E, Fontanges E, Fourcade N, Thomasson A, Pouyet M, Delmas PD. Bone loss after orthotopic liver transplantation. American Journal of Medicine 1994;97:445-450.

358 Mezey E. Alcoholic liver disease: Roles of alcohol and malnutrition. American Journal of Clinical Nutrition 1980;33:2709-2718.

359 Milani A, Ciammella AM, Degen C, Siciliano M, Rossi L. Ascites dynamics in cirrhosis. Proposal and validation of a methylene blue dilution test. Journal of Hepatology 1992;16:369- 375.

360 Millikan WJ, Henderson JM, Galloway JR, Warren WD, Matthews DE, McGhee A, Kutner MH. In vivo measurement of leucine metabolism with stable isotopes in normal subjects and in those with cirrhosis fed conventional and branched-chain amino acid-enriched diets. Surgery 1985;98:405-412.

361 Mills PR, Shenkin A, Anthony RS, McLelland AS, Main ANH, MacSween RNM, Russell RI. Assessment of nutritional status and in vivo immune responses in alcoholic liver disease. American Journal of Clinical Nutrition 1983;38:849-859.

362 Miines Walker R. Francis Glisson and his capsule. Annals of the Royal College of Surgeons of England 1966;38:71-91.

363 Moore FD, Olesen KH, McMurrey JD, Parker HV, Ball MR, Boyden CM. The body cell mass and its supporting environment. Body composition in health and disease. Philadelphia: WB Saunders Company 1963.

364 Moore K, Wilkinson S, Williams R. Ascites and renal dysfunction in liver disease. In: Millward-Sadler GH, Wright R, Arthur MJP, eds. Wright’s liver and biliary disease. London: WB Saunders Company 1992;2:1346-1371.

365 Morgan AG, Kelleher J, Walker BE, Losowsky MS. Nutrition in cryptogenic cirrhosis and chronic aggressive hepatitis. Gut 1976;17:113-118.

366 Morgan DB, Burkinshaw L. Estimation of non-fat body tissues from measurements of skinfold thickness, total body potassium and total body nitrogen. Clinical Science 1983;65:407-414. 309

367 Morgan MY. Enteral nutrition in chronic liver disease. Acta Chirurgica Scandinavica Supplement 1981;507:81-90.

368 Morgan MY. The treatment of chronic hepatic encephalopathy. Hepato-Gastroenterology 1991;38:377-387.

369 Morgan TR, Moritz TE, Mendenhall CL, Haas R, VA Cooperative Study Group. Protein consumption and hepatic encephalopathy in alcoholic hepatitis. Journal of the American College of Nutrition 1995;14:152-158.

370 Morris CD. Effect of dietary sodium restriction on overall nutrient intake. American Journal Clinical Nutrition 1997;65S:687S-691S.

371 Morrison WL, Bouchier lAD, Gibson JNA, Rennie MJ. Skeletal muscle and whole-body protein turnover in cirrhosis. Clinical Science 1990;78:613-619.

372 Moukarzel AA, Najm 1, Vargas J, McDiarmid SV, Busuttil RW, Ament ME. Effect of nutritional status on outcome of orthotopic liver transplantation in pediatric patients. Transplantation Proceedings 1990;22:1560-1563.

373 Mueller KJ, Crosby LO, Oberlander JL, Mullen JL. Estimation of fecal nitrogen in patents with liver disease. Journal of Parenteral and Enteral Nutrition 1983;7:266-269.

374 Mullen JL, Buzby GP, Waldman MT, Gertner MH, Hobbs CL, Rosato EF. Prediction of operative morbidity and mortality by preoperative nutritional assessment. Surgical Forum 1979;30:80-82.

375 Mullen JL, Gertner MH, Buzby GP, Goodhart GL, Rosato EF. Implications of malnutrition in the surgical patient. Archives of Surgery 1979;114:121-125.

376 Mullen KD, Denne SC, McCullough AJ, Savin SM, Bruno D, Tavill AS, Kalhan SC. Leucine metabolism in stable cirrhosis. Hepatology 1986;6:622-630.

377 Müller MJ. Malnutrition in cirrhosis. Journal of Hepatology \995,13.

378 Müller MJ, Fenk A, Lautz HU, Selberg O, Canzler H, Balks HJ, Von Zur Mühlen, Schmidt E, Schmidt FW. Energy expenditure and substrate metabolism in ethanol-induced liver cirrhosis. American Journal of Physiology 1991;260:E338-E344. 310

379 Müller MJ. Lautz HU, Plogmaim B, Burger M, Korber J, Schmidt FW. Energy expenditure and substrate oxidation in patients with cirrhosis: The impact of cause, clinical staging and nutritional state. Hepatology 1992;15:782-794.

380 Müller MJ, Willmann O, Fenk A, Rieger A, Selberg O, Canzler H, Von Zur Mühlen A, Schmidt FW. Resting energy expenditure and the thermic effect of adrenaline in patients with liver cirrhosis. Clinical Science 1992;83:191-198.

381 Müller MJ, Willmann O, Rieger A, Fenk A, Selberg O, Lautz HU, Bürger M, Balks HJ, Von Zur Mühlen A, Schmidt FW. Mechanism of insulin resistance associated with liver cirrhosis. Gastroenterology 1992;102:2033-2041.

382 Murgatroyd PR, Shetty PS, Prentice AM. Techniques for the measurement of human energy expenditure: A practical guide. International Journal of Obesity and Related Metabolic Disorders 1993;17:549-568.

383 Murphy MD, Ireton-Jones CS, Hilman BC, Gorman MA, Liepa GU. Resting energy expenditures measured by indirect calorimetry are higher in preadolescent children with cystic fibrosis than expenditures calculated from prediction equations. Journal of the American Dietetic Association 1995;95:30-33.

384 Naber THJ, Schermer T, de Bree A, Nusteling K, Eggink L, Kruimel JW, Bakkeren J, van Heereveld H, Katan MB. Prevalence of malnutrition in nonsurgical hospitalized patients and its association with disease complications. American Journal of Clinical Nutrition 1997;66:1232- 1239.

3 85 National Health and Medical Research Council. Standard height-weight tables for Australians. Canberra: Australian Institute of Anatomy 1957.

386 Naveau S, Belda E, Borotto E, Genuist F, Chaput JC. Comparison of clinical judgment and anthropometric parameters for evaluating nutritional status in patients with alcoholic liver disease. Journal of Hepatology 1995;23:234-235(Letter).

387 Nelson WP 111, Rosenbaum JD, Strauss MB. Hyponatremia in hepatic cirrhosis following paracentesis. Journal of Clinical Investigation 1951;30:738-744.

388 Nettleton JA, Hegsted DM. Protein-energy interrelations during dietary restriction: Effects on tissue nitrogen and protein turnover. Nutrition and Metabolism 1975;18:31-40. 311

389 Neumann U, Ziegenhom J. Kinetic enzymatic method for the determination of urea in serum with the LKB 8600 Reaction Rate Analyser. Scandinavian Journal of Clinical and Laboratory Investigations 1977;37;Supp 147;90(Abstract 97).

3 90 Neville JN, Eagles JA, Samson G, Olson RE. Nutritional status of alcoholics. American Journal of Clinical Nutrition 1968;21:1329-1340.

391 Nicklas TA, Bao W, Webber LS, B erenson OS. Breakfast consumption affects adequacy of total daily intake in children. Journal of the American Dietetic Association 1993;93:886-891.

392 Nielsen K, Kondrup J, Martinsen L, Dossing H, Larsson B, Stilling B, Jensen MG. Long-term oral refeeding of patients with cirrhosis of the liver. British Journal of Nutrition 1995;74:557- 567.

3 93 Nielsen K, Kondrup J, Martinsen L, Stilling B, Wikman B. Nutritional assessment and adequacy of dietary intake in hospitalized patients with alcoholic liver cirrhosis. British Journal of Nutrition 1993;69:665-679.

394 Nightingale F. Notes on Nursing: What it is and what it is not (1859). London: Gerald Duclcworth and Company Ltd 1952.

395 Nilsson LH. Liver glycogen content in man in the postabsorptive state. Scandinavian Journal of Clinical and Laboratory Investigation 1973;32:317-323.

396 Noel-Jorand MC, Bras J. A comparison of nutritional profiles of patients with alcohol-related pancreatitis and cirrhosis. Alcohol and Alcoholism 1994;29:65-74.

397 Norgan NG, Ferro-Luzzi A. Weight-height indices as estimators of fatness in men. Human Nutrition: Clinical Nutrition 1982;36:363-372.

398 Novick WM, Nusbaum M, Stein TP. The energy costs of surgery as measured by the doubly labelled water (^H^'^O) method. Surgery 1988;103:99-106.

399 Ofman J, Koretz RL. Clinical economics review: Nutritional support. Alimentary Pharmacology and Therapeutics 1997;11:453-471.

400 O’Grady JG. Clinical economics review: Ciyqx \i2û^sp\disAs!àon. Alimentary Pharmacology and Therapeutics 1997;11:445-451. 312

401 O’Keefe SJD, El-Zayadi AR, Carraher TE, Davis M, Williams R. Malnutrition and immuno- incompetence in patients with liver disease. Lancet 1980;2:615-617.

402 Oldroyd B, Bramley PN, Stewart SP, Simpson M, Truscott JG, Losowsky MS, Smith MA. A fom-compartment model to determine body composition in liver cirrhosis. Basic Life Sciences 1993;60:221-224.

403 Oliviero R, Merli M. Measurement of resting energy expenditure in patients with liver disease. Journal of Parenteral and Enteral Nutrition 1991;15:694-695.

404 O’Meara MP, Birkenfeld LW, Gotch FA, Edelman IS. The equilibration of radiosodium (Na^"), radiopotassium (K'*^), and deuterium oxide (DjO) in hydropic human subjects. Journal of Clinical Investigation 1957;36:784-792.

405 Orrego H, Israel Y, Blake JE, Medline A. Assessment of prognostic factors in alcoholic liver disease: Toward a global quantitative expression of severity. Hepatology 1983;3:896-905.

406 Owen OE. Regulation of energy and metabolism. In: Kinney JM, JeejeebhoyKN, Hill GH, Owen OE, eds. Nutrition and metabolism in patient care. Philadelphia: WB Saunders Company 1988:35-39.

407 Owen OE, Reichle FA, Mozzoli MA, Kreulen T, Patel MS, Eifenbein IB, Golsorkhi M, Chang KHY, Rao NS, Sue HS, Boden G. Hepatic, gut, and renal substrate flux rates in patients with hepatic cirrhosis. Journal of Clinical Investigation 1981;68:240-252.

408 Owen OE, Trapp VE, Reichard GA Jr, Mozzoli MA, Moctezuma J, Paul P, Skutches CL, Boden G. Nature and quantity of fuels consumed in patients with alcoholic cirrhosis. Journal of Clinical Investigation 1983;72:1821-1832.

409 Pace N, Kline L, Schachman HK, Harfenist M. Studies on body composition. IV Use of radioactive hydrogen for measurement in vivo of total body water. Journal of Biological Chemistry 1947;168:459-469.

410 Pace N, Rathbum EN. Studies on body composition. Ill The body water and chemically combined nitrogen content in relation to fat content. Journal of Biological Chemistry 1945;158:685-691.

411 Pacy PJ, Preedy VR, Peters TJ, Read M, Halliday D. The effect of chronic alcohol ingestion on whole body and muscle protein synthesis. A stable isotope study. Alcohol and Alcoholism 1991;26:505-513. 313

412 Palmo A, Ombra L, Massarenti P, De Nicola A, Gavassa L, Veglia F, Balzola F. Proposai of a new formula to calculate total urinary nitrogen from urea urinary nitrogen. Clinical Nutrition 1989;8:45-47.

413 Panella C, Guglielmi FW, Francavilla A. Nutritional status in cirrhosis: Role of bioelectric impedance analysis. Italian Journal of Gastroenterology 1993;25:402-407.

414 Panella C, Guglielmi FW, Mastronuzzi T, Francavilla A. Whole-body and segmental bio electrical parameters in chronic liver disease: Effect of gender and disease stages. Hepatology 1995;21:352-358.

415 Pannemans DL, Bouten CV, Westerterp KR. 24 h energy expenditure during a standardized activity protocol in young and elderly men. European Journal of Clinical Nutrition 1995;49:49- 56.

416 Passmore R, Eastwood MA. Davidson and Passmore human nutrition and dietetics. Edinburgh: Churchill Livingstone 1986:3-28.

417 Passmore R, Thomson JG, Wamock GM. A balance sheet of the estimation of energy intake and energy expenditure as measured by indirect calorimetry using the Kofranyi-Michaelis calorimeter. British Journal of Nutrition 1952;6:253-264.

418 Patek AJ Jr. Alcohol, malnutrition, and alcoholic cirrhosis. American Journal of Clinical Nutrition 1979;32:1304-1312.

419 Patek AJ Jr, Post J, Ratnoff OD, Mankin H, Hillman RW. Dietary treatment of cirrhosis of the liver. Results in one hundred and twenty-four patients observed during a ten year period. Journal of the American Medical Association 1948;138:543-549.

420 Patek AJ, Toth IG, Saunders MG, Castro GAM, Engel JJ. Alcohol and dietary factors in cirrhosis. Archives of Internal Medicine 1975;135:1053-1057.

421 Paton NIJ, Elia M, Jebb SA, Jennings G, Macallan DC, Griffin GE. Total energy expenditure and physical activity measured with the bicarbonate-urea method in patients with human immunodeficiency virus infection. Clinical Science 1996;91:241-245.

422 Perez GO, Oster JR. Altered potassium metabolism in liver disease. In: Epstein M, ed. The kidney in liver disease. 2nd edn. New York: Elsevier Biomedical 1983:183-201.

423 Perkins KA. Metabolic effects of cigarette smoking. Journal of Applied Physiology 1992;72:401-409. 314

424 Perkins KA, Epstein LH, Stiller RL, Marks BL, Jacob RG. Acute effects of nicotine on resting metabolic rate in cigarette smokers. American Journal of Clinical Nutrition 1989;50:545-550.

425 Petrides AS, Luzi L, Reuben A, Riely C, DeFronzo RA. Effect of insulin and plasma amino acid concentration on leucine metabolism in cirrhosis. Hepatology 1991;14:432-441.

426 Pett LB, Ogilvie OF. The Canadian weight-height survey. Human Biology 1956;28:177-188.

427 Pierrugues R, Blanc P, Daures JP, Michel H. Relationship of resting energy expenditure with liver function and nutritional status inpatients with alcoholic cirrhosis. Nutrition 1992;8:22-25.

428 Piers LS, Diffey B, Soares MJ, Frandsen SL, McCormack LM, Lutschini MJ, O’Dea K. The validity of predicting the basal metabolism of young Australian men and women. European Journal of Clinical Nutrition 1997;51:333-337.

429 Pierson RN Jr, Wang J, Heymsfield SB, Dilmanian FA, Weber DA. High precision in-vivo neutron activation analysis: A new era for compartmental analysis in body composition. Basic Life Sciences 1990;55:317-325.

430 Pierson RN Jr, Wang J, Thornton JC, Van Itallie TB, Colt EWD. Body potassium by four-pi '*°K counting: An anthropometric correction. American Journal of Physiology 1984;246:F234- F239.

431 Pignon JP, Poynard T, Naveau S, Marteau P, Zourabichvili O, Chaput JC. Analyse multidimensionelle selon le modèle de Cox de la survie de patients atteints de cirrhose alcoolique. Gastroentérologie Clinique et Biologique 1986;10:461-467.

432 Pike RL, Brown ML. Nutrition: An integrated approach. 2nd edn. New York: John Wiley and Sons Inc 1975.

433 Pikul J, Sharpe MD, Lowndes R, Ghent CN. Degree of preoperative malnutrition is predictive of postoperative morbidity and mortality in liver transplant recipients. Transplantation 1994;57:469-472.

434 Pirlich M, Selberg O, Boker K, Schwarze M, Müller MJ. The creatinine approach to estimate skeletal muscle mass in patients with cirrhosis. Hepatology 1996;24:1422-1427.

435 Pitt HA, Cameron JL, Postier RG, Gadacz TR. Factors affecting mortality in biliary tract surgery. American Journal of Surgery 1981;141:66-71. 315

436 Pittman JG, Decker JW. Acute and chronic myopathy associated with alcoholism. Neurology 1971;21:293-296.

437 Plauth M, Merli M, Kondrup J, Weimann A, Ferenci P, Müller MJ. ESPEN guidelines for nutrition in liver disease and transplantation. Clinical Nutrition 1997;16:43-55.

438 Plevak DJ, DiCecco SR, Wiesner RH, Porayko MK, Wahlstrom HE, Janzow DJ, Hammel KD, O’Keefe SJ. Nutritional support for liver transplantation: Identifying caloric and protein requirements. Mayo Clinic Proceedings \ 99A,69:225-130.

439 Polgar S, Thomas SA. Introduction to research in the health sciences. 3rd edn. Melbourne: Churchill Livingstone 1995:3-17.

440 Popper H. Vienna and the liver. Gastroenterology 1984;86:982-986.

441 Porayko MK, DiCecco S, O’Keefe SJD. Impact of malnutrition and its therapy on liver transplantation. Seminars in Liver Disease 1991;11:305-314.

442 Portmann B, Popper H, Neuberger J, Williams R. Sequential and diagnostic features in primary biliary cirrhosis based on serial histologic study in 209 patients. Gastroenterology 1985;88:1777- 1790.

443 Post J, Rose JV, Shore SM. Intravenous use of salt-poor human albumin: Effects in 34 patients with decompensated hepatic cirrhosis. Archives of Internal Medicine 1951;87:775-788.

444 Prentice TC, Siri W, Joiner EE. Quantitative studies of ascitic fluid circulation with tritium- labeled water. American Journal of Medicine 1952;13:668-673.

445 Prijatmoko D, Strauss B JG, Lambert JR, Sievert W, Stroud DB, Wahlqvist ML, Katz B, Colman J, Jones P, Korman MG. Early detection of protein depletion in alcoholic cirrhosis: Role of body composition analysis. Gastroenterology 1993;105:1839-1845.

446 Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Transection of the oesophagus for bleeding oesophageal varices. British Journal of Surgery 1973;60:646-649.

447 Pullicino E, Copperstone C, Luzi L, McNeill G, Elia M. Relationship between anthropometric indices of body fat distribution and basal energy metabolism in healthy Maltese women. Acta Diabetologica 1996;33:198-204.

448 Pullicino E, Coward A, Elia M. Total energy expenditure in intravenously fed patients measured by the doubly labeled water technique. Metabolism Clinical and Experimental 1993;42:58-64. 316

449 Quetelet LAJ. Physique Sociale; Ou, essai sur le diveloppement des facultis de rhomme. Brussels: Muquardt C 1869;2:92.

450 Randecker GA, Smiciklas-Wiight H, McKenzie JM, Shannon BM, Mitchell DC, Becker DJ, KieseUiorst K. The dietary intake of children with fDDM. Diabetes Care 1996;19:1370-1374.

451 Rathmacher JA, Flakoll PJ, Nissen SL. A compartmental model of 3 -methylhistidine metabolism in humans. American Journal of Physiology 1995;269(32):E192-E198.

452 Ravussin E, Bogardus C. Relationship of genetics, age and physical fitness to daily energy expenditure and fuel utilization. American Journal of Clinical Nutrition 1989;49:968-975.

453 Ravussin E, Lillioja S, Anderson TE, Christin L, Bogardus C. Determinants of 24-hour energy expenditure in man: Methods and results using a respiratory chamber. Journal of Clinical Investigation 1986;78:1568-1578.

454 Rennie MJ, Millward DJ, Halliday D, Lundholm K. A dissenting view of the significance of 3-methylhistidine excretion in cirrhotic patients. Clinical Science 1983;64:243-246(Letter).

455 Rhoads JE, Alexander CE. Nutritional problems of surgical patients. Annals of the New York Academy of Sciences 1955;63:268-275.

456 Ricci P, ThemeauTM, Malinchoc M, Benson JT, Petz JL, Klintmalm GB, Crippin JS, Wiesner RH, Steers JL, Rakela J, Starzl TE, Dickson ER. A prognostic model for the outcome of liver transplantation in patients with cholestatic liver disease. Hepatology 1997;25:672-677.

457 Riggio O, Merli M, Romiti A, Pinto G, Panella R, Attili AF, CapocacciaL. Early postprandial energy expenditure and macronutrient use after a mixed meal in cirrhotic patients. Journal of Parenteral and Enteral Nutrition 1992;16:445-450.

458 Rimola A, Soto R, Bory F, Arroyo V, Piera C, Rodés J. Reticuloendothelial system phagocytic activity in cirrhosis and its relation to bacterial infections and prognosis. Hepatology 1984;4:53- 58.

459 Rimola A. Infections in liver disease. In: McIntyre N, Benhamou JP, Bircher J, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford: Oxford University Press 1991;2:1272-1284.

460 Rinehart JF, Greenberg LD, Baker F, Mettier SR, Bruckman F, Choy F. Metabolism of vitamin C in rheumatoid arthritis. Archives of Internal Medicine 1938;61:537-551. 317

461 Rodeck B, Melter M, Kardorff R, Hoyer PF, Ringe B, Burdelski M, Oldhafer KJ, Pichlmayr R, Brodehl J. Liver transplantation in children with chronic end stage liver disease. Transplantation 1996;62:1071-1076.

462 Roehrig CS, Lee JA. Cost analysis in nutrition-intervention outcomes studies. Nutrition 1996;12:558-559.

463 Romijn JA, Endert E, Sauerwein HP. Glucose and fat metabolism during short-term starvation in cirrhosis. Gastroenterology 1991;100:731-737.

464 Romiti A, Merli M, Martorano M, Pairilli G, Martino F, Riggio O, Truscelli A, CapocacciaL, Budillon G. Malabsorption and nutritional abnormalities in patients with liver cirrhosis. Italian Journal of Gastroenterology 1990;22:118-123.

465 Ross R, Léger L, Morris D, De Guise J, Guardo R. Quantification of adipose tissue by MRl: Relationship to anthropometric variables. Journal of Applied Physiology 1992;72:787-795.

466 Rothschild MA, Oratz M, Schreiber SS. Alcohol, amino acids, and albumin synthesis. Gastroenterology 1974;67:1200-1213.

467 Royal College of Radiologists. Making the best use of a department of clinical radiology. Guidelines for doctors. 3rd edn. London: Royal College of Radiologists 1995:12.

468 Rubin E, Lieber CS. Alcohol-induced hepatic injury in nonalcoholic volunteers. New England Journal of Medicine 1968;278:869-876.

469 Runyon BA. Treatment of patients with cirrhosis and ascites. Seminars in Liver Disease 1997;17:249-260.

470 Russell RI, Williamson JM, Goldberg A, Wares E. Ascorbic-acid levels in leucocytes of patients with gastrointestinal haemorrhage. Lancet 1968;2:603-606.

471 Ruxton CH, O’Sullivan KR, Kirk TR, Belton NR, Holmes RA. The contribution of breakfast to the diets of a sample of 136 primary-schoolchildren in Edinburgh. British Journal of Nutrition 1996;75:419-431.

472 Ryan CF, Road JD, Buckley PA, Ross C, Whittaker JS. Energy balance in stable malnourished patients with chronic obstructive pulmonary disease. Chest 1993;103:1038-1044. 318

473 Salusky IB, Fine RN, Nelson P, Blumenkrantz MJ, Kopple JD. Nutritional status of children undergoing continuous ambulatory peritoneal dialysis. American Journal of Clinical Nutrition 1983;38:599-611.

474 Sarin SK, Dhingra N, Bansal A, Malhotra S, Guptan RC. Dietary and nutritional abnormalities in alcoholic liver disease: A comparison with chronic alcoholics without liver disease. American Journal of Gastroenterology 1997;92:777-783.

475 Schlichting P, Christensen E, Andersen PK, Fauerholdt L, Juhl E, Poulsen H, Tygstrup N, the Copenhagen Study Group for Liver Diseases. Prognostic factors in cirrhosis identified by Cox’s regression model. Hepatology 1983;3:889-895.

476 Schloerb PR, Friis-Hansen B J, Edelman IS, Solomon AK, Moore FD. The measurement of total body water in the human subject by deuterium oxide dilution: With a consideration of the dynamics of deuterium dilution. Journal of Clinical Investigation 1950;29:1296-1310.

477 Schloerb PR, Forster J, Delcore R, Kindscher JD. Bio electrical impedance in the clinical evaluation of liver disease. American Journal of Clinical Nutrition 1996;64S:510S-514S.

478 Schneeweiss B, Graninger W, Ferenci P, Eichinger S, Grimm G, Schneider B, Laggner AN, Lenz K, Kleinberger G. Energy metabolism in patients with acute and chronic liver disease. Hepatology 1990;11:387-393.

479 Schober O, Mariss P, Schmidt FW, Hundeshagen H. Total body water, extracellular water, plasma volume and total body potassium in cirrhosis of the liver. Klinische Wochenschrift 1979;57:757-761.

480 Schoeller DA, van Santen E. Measurement of energy expenditure in humans by doubly labeled water method. Journal of Applied Physiology 1982;53:955-959.

481 Schoeller DA, van Santen E, Peterson DW, Dietz W, Jaspan J, Klein PD. Total body water measurements in humans with '*0 and labeled water. American Journal of Clinical Nutrition 1980;33:2686-2693.

482 Schofield WN. Predicting basal metabolic rate, new standards and review of previous work. Human Nutrition: Clinical Nutrition 1985;39C;S1:5-41.

483 Schütz Y, Chambaz A. Could a satellite-based navigation system (GPS) be used to assess the physical activity of individuals on earth ? European Journal of Clinical Nutrition 1997;51:338- 339. 319

484 Seaton A, Seaton D, Leitch AG. Crofton and Douglas’s respiratory diseases. 4th edn. Oxford: Blackwell Scientific Publications 1989:28-75.

485 Segal KR, Burastero S, Chun A, Coronel P, Pierson RN Jr, Wang J. Estimation of extracellular and total body water by multiple-frequency bioelectrical-impedance measurement. American Journal of Clinical Nutrition 1991;54:26-29.

486 Segal KR, Gutin B, Presta E, Wang J, Van Itallie TB. Estimation of human body composition by electrical impedance methods: A comparative study. Journal of Applied Physiology 1985;58:1565-1571.

487 Selberg O, Bottcher J, Tusch G, Pichlmayr R, Henkel E, Müller MJ. Identification of high- and low-risk patients before liver transplantation: A prospective cohort study of nutritional and metabolic parameters in 150 patients. Hepatology 1997;25:652-657.

488 Selberg O, Radoch E, Walter GF, Müller MJ. Skeletal muscle glycogen content in patients with cirrhosis. Hepatology 1994;20:135-141.

489 Shanbhogue RLK, Bistrian BR, Jenkins RL, Jones C, Benotti P, Blackburn GL. Resting energy expenditure in patients with end-stage liver disease and in normal population. Journal of Parenteral and Enteral Nutrition 1987;11:305-308.

490 Shanbhogue RLK, Bistrian BR, Lakshman K, Crosby L, Swenson S, Wagner D, Jenkins RL, Blackburn GL. Whole body leucine, phenylalanine, and tyrosine kinetics m end-stage liver disease before and after hepatic transplantation. Metabolism 1987;36:1047-1053.

491 Shaw BW Jr, Wood RP, Gordon RD, Iwatsuki S, Gillquist WP, Starzl TE. Influence of selected patient variables and operative blood loss on six-month survival following liver transplantation. Seminars in Liver Disease 1985;5:385-393.

492 Shepherd RW, Chin SE, Cleghom GJ, Patrick M, Ong TH, Lynch SV, Balderson G, Strong R. Malnutrition in children with chronic liver disease accepted for liver transplantation: Clinical profile and effect on outcome. Journal of Paediatrics and Child Health 1991;27:295-299.

493 Shetty PS, Watrasiewicz KE, Jung RT, James WPT. Rapid-tumover transport proteins: An index of subclinical protein-energy malnutrition. Lancet 1979;2:230-232.

494 Shiomi S, Kuroki T, Masaki K, Takeda T, Nishiguchi S, Nakajima S, Seki S, Kobayashi K, Okamura T, Ochi H. Osteopenia in primary biliary cirrhosis and cirrhosis of the liver in women, evaluated by dual-energy X-ray absorptiometry. Journal of Gastroenterology 1994;29:605-609. 320

495 Simko V, Connell AM, Banks B. Nutritional status in alcoholics with and without liver disease. American Journal of Clinical Nutrition 1982;35:197-203.

496 Simopoulos AP. Opening address. Nutrition and fitness from the first Olympiad in 776 BC to 393 AD and the concept of positive health. American Journal of Clinical Nutrition 1989;49:S5:921-926.

497 Singer PN. Galen - selected works. Oxford: Oxford University Press 1997:202-295.

498 Siri W. Body composition from fluid spaces and density: Analysis of methods. In: Brozek J, Henschel A, eds. Techniques for measuring body composition. Washington DC: National Academy of Sciences, National Research Council 1961:223-224.

499 Siri WE. The gross composition of the body. In: Lawrence TH, Tobias CA, eds. Advances in biological and medical physics. New York: Academic Press 1956;4:239-280.

500 Smith FR, Henkin RJ, Dell RB. Disordered gustatory acuity in liver disease. Gastroenterology 1976;70:568-571.

501 Snellen JW, Chang KS, Smith W. Technical description and performance characteristics of a human whole-body calorimeter. Medical and Biological Engineering and Computing 1983 ;21:9-

20 .

502 Snyder WS, Cook MJ, Nasset ES, KarhausenLR, Howells GP, Tipton IH. Report of the task group on reference man. Oxford: Pergamon Press 1975:289.

503 Soberman R, Brodie BB, Levy BB, Axelrod J, Hollander V, Steele JM. The use of antipyrene in the measurement of total body water in man. Journal of Biological Chemistry 1949; 179:31- 42.

504 Soulsby CT. Dietary treatment of hepatic encephalopathy: A survey of current practice. Proceedings of the Nutrition Society 1997;56:275A(Abstract).

505 Soulsby CT, Madden AM, Morgan MY. The effect of dietary sodium restriction on energy and protein intake m patients with cirrhosis. Hepatology 1997;26:382A(Abstract 1013).

506 Spinnler G, Jéquier E, Favre R, Dolivo M, Vannotti A. Human calorimeter with a new type of gradient layer. Journal of Applied Physiology 1973;35:158-165. 321

507 Stamford BA, Matter S, Fell RD, Papanek P. Effects of smoking cessation on weight gain, metabolic rate, caloric consumption, and blood lipids. American Journal of Clinical Nutrition 1986;43:486-494.

508 Stanley MM, Belknap S, Biliack SA, Hartz CE, Honk JH. Ascitic fluid albumin and water flows in patients with alcoholic cirrhosis: Effects of peritoneovenous shunting. Journal of Laboratory and Clinical Medicine 1994;123:206-217.

509 Starzl TE. The puzzle people - memoirs of a transplant surgeon. Pittsburgh: University of Pittsburgh Press 1992:339.

510 Starzl TE, Marchioro TL, Von Kaulla KN, Hermann G, Brittain RS, Waddell WR. Homotransplantation of the liver in humans. Surgery Gynecology and Obstetrics 1963;117:659- 676.

511 Strachey L. Eminent Victorians. London: Penguin Books 1986:111-161.

512 Studley HO. Percentage of weight loss: A basic indicator of surgical risk in patients with chronic peptic ulcer. Journal of the American Medical Association 1936;106:458-460.

513 Stumiolo GC, DTnca R, Parisi G, Giocomazzi F, Montino MC, D ’Odorico A, Soranzo P, Naccarato R. Taste alterations in liver cirrhosis: Are they related to zinc deficiency ? Journal of Trace Elements and Electrolytes in Health and Disease 1992;6:15-19.

514 Suter PM, Jéquier E, Schütz Y. Effect of ethanol on energy expenditure. American Journal of Physiology 1994;226:R1204-R1212.

515 Suter PM, Schütz Y, Jequier E. The effect of ethanol on fat storage in healthy subjects. New England Journal of Medicine 1992;326:983-987.

516 Siittmann U, Ockenga J, Hoogestraat L, Selberg O, Schedel I, Deicher H, Muller MJ. Resting energy expenditure and weight loss in human immunodeficiency virus-infected patients. Metabolism Clinical and Experimental 1993;42:1173-1179.

517 Swan PB. The liver forms, stores and secretes glucose (Claude Bernard, 1860). Journal of Nutrition 1997;127(S5):1019S-1020S.

518 Swart GR, van den Berg JWO, van Vuure JK, Rietveld T, Wattimena DL, Frenkel M. Minimum protein requirements in liver cirrhosis determined by nitrogen balance measurements at three levels of protein intake. Clinical Nutrition 1989;8:329-336. 322

519 Swart GR, van den Berg JWO, Wattimena JLD, Rietveld T, van Vuure JK, Frenkel M. Elevated protein requirements in cirrhosis of the liver investigatedby whole body protein turnover studies. Clinical Science 1988;75:101-107.

520 Swinamer DL, Grace MG, Hamilton SM, Jones RL, Roberts P, King EG. Predictive equation for assessing energy expenditure in mechanically ventilated critically ill patients. Critical Care Medicine 1990;18:657-661.

521 Szeluga DJ, Stuart RK, Utermohlen V, Santos GW. Nutritional assessment by isotope dilution analysis of body composition. American Journal of Clinical Nutrition 1984;40:847-854.

522 Takabatake T, Ohta H, Ishida Y, Hara H, Ushiogi Y, Hattori N. Low serum creatinine levels in severe hepatic disease. Archives of Internal Medicine 1988;148:1313-1315.

523 Tanis AA, van den Berg JWO, Rietveld T, Wattimena JLD, Swart GR. Depletion of liver glycogen in cirrhosis using a ’^C-emiched diet and breath samples. In: Record C, A1 Mardini H, eds. Advances in hepatic encephalopathy and metabolism in liver disease. Newcastle upon Tyne: Medical Faculty, University of Newcastle upon Tyne 1997:39-44.

524 Tanner JM, Whitehouse RH. The Harpenden skinfold caliper. American Journal of Physical Anthropology 1955;13:743-746.

525 Tarter RE, Panzak G, Switala J, Lu S, Simkevitz H, Van Thiel D. Isokinetic muscle strength and its association with neuropsychological capacity in cirrhotic alcoholics. Alcohol Clinical and Experimental Research 1997;21:191-196.

526 Thome A, Johansson U, Wahren J, Eriksson S. Thermogenic response to intravenous nutrition in patients with cirrhosis. Journal of Hepatology 1992;16:145-152.

527 Thuluvath PJ, Triger DR. Evaluation of nutritional status by using anthropometry in adults with alcoholic and nonalcoholic liver disease. American Journal of Clinical Nutrition 1994;60:269- 273.

528 Todhunter EN. Some aspects of the history of dietetics. World Review of Nutrition and Dietetics 1973;18:1-46.

529 Todorovic VE, Micklewright A. A pocket guide to clinical nutrition. 2nd edn. Birmingham: British Dietetic Association 1997:2.9. 323

530 Tmlson MF, McCann MB. Comparison of dietary survey methods. Journal of the American Dietetic Association 1959;35:672-676.

531 Tschegg E, Sigmund A, Veitl V, Schmid P, Irsigler K. An isothermic, gradient-free, whole- body calorimeter for long-term investigation of energy balance in man. Metabolism 1979;28:764-770.

532 Turner D. The estimation of the patient’s home dietary intake. Journal of the American Dietetic Association 1940;16:875-881.

533 Tweedie FJ, Long RC. Abdominal wound disruption. Surgery Gynecology and Obstetrics 1954;99:41-47.

534 Tzankoff SP, Norris AH. Effect of muscle mass decrease on age-relatedBMR changes. Journal of Applied Physiology 1977;43:1001-1006.

535 Underwood BA. Methods for assessment of vitamin A status. Journal of Nutrition 1990;120:1459-1463.

536 Van Thiel DH, Lipsitz HD, Porter LE, Schade RR, Gottlieb GP, Graham TO. Gastrointestinal and hepatic manifestations of chronic alcoholism. Gastroenterology 1981;81:594-615.

537 Verboeket-van de Venne WPHG, Westerterp KR, van Hoek B, Swart GR. Energy e?qpenditure and substrate metabolism in patients with cirrhosis of the liver: Effects of the pattern of food intake. Gut 1995;36:110-116.

538 Vermeij CG, FeenstraBWA, Oomen AMFA, De Graaf EJR, Zillikens MC, Swart GR, Bruining HA. Assessment of energy expenditure by indirect calorimetry in healthy subjects and patients with liver cirrhosis. Journal of Parenteral Enteral Nutrition 1991;15:421-425.

539 Vestal RE, Norris AH, Tobin JD, Cohen BH, Shock NW, Andres R. Antipyrine metabolism in man: Influence of age, alcohol, caffeine and smoking. Clinical Pharmacology and Therapeutics 1975;18:425-432.

540 Virgin F, Maiani G, Zahoor ZH, Ciarapica D, Raguzzini A, Ferro-Luzzi A. Relationship between fat-free mass and urinary excretion of creatinine and 3 -methylhistidine in adult humans. Journal of Applied Physiology 1994;76:1946-1950. 324

541 Vitale GC, Neill CD, Fenwick MK, Stewart WW, Cushieri A. Body composition in the cirrhotic patient with ascites: Assessment of total exchangeable sodium and potassium with simultaneous serum electrolyte determination. American Surgeon 1985;51:675-681.

542 Von Meyenfeldt MF, Meijerink WJHJ, Ronflait MMJ, Buil-Maassen MTHJ, Soeters PB. Perioperative nutritional support: A randomised clinical trial. Clinical Nutrition 1992;11:180- 186.

543 Voss AC, Gallagher-Allred C. The importance of outcomes research in documenting the benefit of nutrition intervention. Nutrition 1996;12:59-60.

544 Walker JB. Creatine: Biosynthesis, regulation and function. Advances in Enzymology 1979;50:177-242.

545 Walker SP, Grantham-McGregor S, Powell C, Fletcher P, Himes JH. Bioelectrical impedance, anthropometry and body composition in stunted and non-stunted children. European Journal of Clinical Nutrition 1990;44:763-768.

546 Waluga M, Zahorska-Markiewicz B, Janusz M, Slabiak Z, Chelmicka A. Resting energy expenditure in patients with cirrhosis of the liver measured by indirect calorimetry, anthropometry and bioelectrical impedance analysis. Experientia 1996;52:591-596.

547 Wang ZM, Pierson RN Jr, Heymsfreld SB. The five-level model: A new approach to organizing body-composition research. American Journal of Clinical Nutrition 1992;56:19-28.

548 Warwick PM, Barnes J. Energy e^qrenditure in free-living smokers and nonsmokers: Comparisons between factorial, intake balance, and doubly labeled water measures. American Journal of Clinical Nutrition 1996;63:15-21.

549 Warwick PM, Chappie RS, Thomson ES. The effect of smoking two cigarettes on resting metabolic rate with and without food. International Journal of Obesity 1987;11:229-237.

550 Watanabe A, Kuwabara Y, Murata H, Kobashi K. Clinical significance of the expiratory ammonia concentration. In: Record C, A1 Mardini H, eds. Advances in hepatic encephalopathy and metabolism in liver disease. Newcastle upon Tyne: Medical Faculty, University of Newcastle upon Tyne 1997:133-137.

551 Watanabe A, WakabayashiH, Kuwabara Y. Nutrient-induced thermogenesis and protein-sparing effect by rapid infusion of a branched chain-emiched amino acid solution to cirrhotic patients. Journal of Medicine 1996;27:176-182. 325

552 Webb AR, Newman LA, Taylor M, Keogh JB. Hand grip dynamometry as a predictor of postoperative complications. Reappraisal using age standardized grip strengths. Journal of Parenteral and Enteral Nutrition 1989;13:30-33.

553 Webb P. Energy expenditure and fat-free mass in men and women. American Journal of Clinical Nutrition 1981;34:1816-1826.

554 Webb P. 24-hour energy expenditure and the menstrual cycle. American Journal of Clinical Nutrition 1986;44:614-619.

555 Weber FL, Jr. The effect of lactulose on urea metabolism and nitrogen excretion in cirrhotic patients. Gastroenterology 1979;77:518-523.

556 Weber FL Jr, Bagby BS, Licate L, Kelsen SG. Effects of branched-chain amino acids on nitrogen metabolism in patients with cirrhosis. Hepatology 1990; 11:942-950.

557 Weir JB. New methods for calculating metabolic rate with special reference to protein metabolism. Journal of Physiology 1949;109:1-9.

558 WestreichM. Liver disease in the Talmud. Journal of Clinical Gastroenterology 1990;12:57-62.

559 Wicks C, Bray GP, Williams R. Nutritional assessment in primary biliary cirrhosis: The effect of disease severity. Clinical Nutrition 1995;14:29-34.

560 Widdowson EM, Edholm OG, McCanceRA. The food intake and energy expenditure of cadets in training. British Journal of Nutrition 1954;8:147-55.

561 Widdowson EM, McCance RA, Spray CM. The chemical composition of the human body. Clinical Science 1951;10:113-125.

562 Williamson DF, Madans J, Anda RF, Kleinman JC, Giovino G A, Byers T. Smoking cessation and severity of weight gain in a national cohort. New England Journal of Medicine 1991;324:739-745.

563 Wilson SR, Amoroso P, Baker A, Ponte JC, Ball C. Resting energy expenditure in children with liver disease. Proceedings of the Nutrition Society 1991;50:157A(Abstract).

564 Witte CL, Witte MH, Cole WR, Chung YC, Bleisch VR, Dumont AE. Dual origin of ascites in hepatic cirrhosis. Surgery, Gynecology and Obstetrics 1969;129:1027-1033. 326

565 Wood B, Nicholls KM, Breen KJ. Nutritional status in alcoholism. Journal of Human Nutrition and Dietetics 1992;5:275-285.

566 Wu TJ, Huang JJ, Lin CY. Effects of fluid retention on the measurement of body composition using bioelectric impedance. Journal of the Formosan Medical Association 1994;93:939-943.

567 YamadaKM, Olden K. Fibronectins - Adhesive glycoproteins of cell surface and blood. Nature 1978;275:179-184.

568 Young VR, Munro HN. N'-methylhistidine (3-methylhistidine) and muscle protein turnover: An overview. Federation Proceedings 1978;37:2291-2300.

569 Zillikens MC, van den Berg JWO, Wilson JHP, Rietveld T, Swart GR. The validity of bio electrical impedance analysis in estimating total body water in patients with cirrhosis. Journal of Hepatology 1992;16:58-65.

570 Zillikens MC, van den Berg JWO, Wilson JHP, Swart GR. Whole-body and segmental bioelectrical-impedanceanalysis in patients with cirrhosis of the liver: Changes after treatment of ascites. American Journal of Clinical Nutrition 1992;55:621-625.

571 Zoli M, Marchesini G, Dondi C, Blanchi GP, Pisi E. Myofibrillar protein catabolic rates in cirrhotic patients with and without muscle wasting. Clinical Science 1982;62:683-686. (‘tv’; h.1 ,/! »i lii i Jowiuil of Hepatology 1994; 21; 878-883 Copyright © Jounml of Hepatology 1994 ^ Pmted in Denmark. All rights reserved r r* 1 Journal of Hepatology Munksgaard. Copenhagen ISSN 0J68-8278

A comparison of skinfold anthropometry and bioelectrical impedance analysis for measuring percentage body fat in patients with cirrhosis

Angela M. Madden and Marsha Y. Morgan University Department of Medicine, Royal Free Hospital and School of Medicine, London, UK (Received 27 January 1994)

Percentage body fat was determined by skinfold anthropometry and bioelectrical impedance analysis in 60 patients with cirrhosis, 30 of whom had overt fluid retention, and in 60 age- and sex-matched healthy volunteers. In the control population the mean±lSD percentage body fat assessed using bioelectrical impedance analysis, 22.6±6.5%, and skinfold anthropometry, 22.5±6.9%, were comparable. However, there was considerable variation in individual values such that measurements made using bioelectrical impedance analysis could be from 9% less to 8% more than the corresponding anthropometric values. In patients with cirrhosis the mean percentage body fat assessed using bioelectrical impedance analysis, 24.4±8.9%, was significantly greater than the value obtained using skinfold anthropometry 20.3±8.4% (p<0.01) for the whole group and for the subgroup of patients with fluid retention, 24.5±8.5% cf. 18.4±7.1% (p<0.005); estimates of body fat in the subgroup of patients without overt fluid retention were comparable between methods, 24.4±9.4% cf. 22.1 ±9.2%. In the whole patient group individual measurements made using bioelectrical impedance analysis could be from 10% less to 18% more than the corresponding anthropometric values; similar degrees of variation in individual values were observed in the two patient subgroups.. Bioelectrical impedance analysis should not be used interchangeably with skinfold anthropometry for assessment of percentage body fat in patients with cirrhosis, irrespective of their state of hydration. © Journal of Hepatology.

Key words: Bioelectrical impedance analysis; Body composition; Cirrhosis; Healthy volunteers; Skinfold anthropometry

The measurement of body composition in individuals rent. Regression equations are devised, using the subject’s with chronic liver disease is difficult and few reliable height, weight, age and resistance and reactance to a methods of measurement have been validated in this pa­ known current, which are then used to calculate the sub­ tient group (1). Skinfold anthropometry has been used to ject’s total body water, fat free mass, and, by derivation, evaluate nutritional status in patients with cirrhosis (2-5) fat mass and percentage body fat. The technique is easy but measurements are subject to intra-observer variation to perform and reproducible, and is said to be less prone (6) and can be significantly affected by skin oedema (7); to intra-observer variation than skinfold anthropometry in addition, the subsequent calculation of percentage body (10,11). fat relies on certain assumptions validated in healthy " Bioelectrical impedance analysis provides valid assess­ populations which may not apply in disease populations. ments of body composition in healthy individuals in spe­ In a number of recent studies (8,9) body composition cific population groups (12-17) and in disease states data have been obtained in patients with cirrhosis by use where hydration is constant (18,19). However, it is of little of bioelectrical impedance analysis. This is an indirect value for assessing body composition in patients whose method of assessing body composition based on the resist­ state of hydration is variable (20), for example in those ance and reactance of the body to a small electrical cur­ with cardiac and pulmonary disease (21) or renal disease

Correspondence to: Dr M. Y. Morgan, University Department of Medicine, Royal Free Hospital, Pond St, London NW3 2QG, UK. BODY FAT MEASUREMENTS IN CIRRHOSIS 879 requiring dialysis (22,23). Patients with chronic liver dis­ intake, posture or exercise within 6 h of assessment, or ease also tend to retain fluid (7,24), so this technique may phase of the menstrual cycle. not provide valid data on body composition in this patient The mid-arm circumference was measured and skinfold group. Indeed, it has already been shown that bioelectrical measurements were made at four sites: triceps, biceps, impedance analysis is of little value for detecting fluid subscapular and suprailiac areas, as described by Dur- changes in patients with cirrhosis and ascites (24-27) and nin & Womersley (30). All measurements were made on of limited value for assessing body cell mass (28). No data the non-dominant side of the body, using the writing hand are available currently, however, on its value for assessing to define dominance. All skinfold measurements were percentage body fat in this patient population. made using the same Holtain/Tanner-Whitehouse skinfold In this study, comparison was made between measure­ calipers and steel tape measure (Holtain Ltd, Crymych, ments of percentage body fat, obtained using bioelectrical Dyfed, UK). Measurements were taken to the nearest 3.0 impedance analysis and skinfold anthropometry, from a mm, and a mean result was calculated from three or four large group of patients with cinhosis and in a carefully readings. Percentage body fat was calculated using the matched control group of healthy volunteers. method of Durnin & Womersley (30). Bioelectrical impedance was measured using a BIAlOl S impedance analyser (RJL Systems, Mt Clemens, Michi­ Patients and Methods gan, USA). Standard procedures were followed with the The patient population comprised 60 patients (40 men subject lying supine on a flat surface with the legs not and 20 women) of mean (range) age 50.4 (19-72) years touching at the thighs and the arms not touching the with cirrhosis of varying aetiology and severity. The aeti­ torso. ECG electrodes (Silver Mactrodes, Marquette Elec­ ology of the liver injury was diagnosed on the basis of tronics Inc, Florida, USA) were attached to the skin sur­ historical, clinical, laboratory and histological variables, face of the right dorsal side of the body proximal to the whilst the functional severity of the liver injury was as­ second metacarpo-phalangeal joint, medial to the distal sessed using Pugh’s modification of Child’s grading system prominences of the radius and the ulna, proximal to the (29) (Table 1). Thirty (50%) of these individuals had clin­ third metatarso-phalangeal joint and between the medial ically detectable ascites and/or ankle oedema. The and lateral malleoli of the ankle. The electrodes were then matched reference population comprised 60 healthy Vo­ connected to the analyser and the resistance and reactance lunteers (40 men and 20 women) of mean (range) age 50.1 to an 800 n k current at 50 kQ recorded. The analyser was (22-71) years (Table 2). calibrated daily using a standard resistor of 500 Q. Sub­ Body composition measurements were undertaken on jects were weighed to the nearest 0.5 kg, wearing light in­ all subjects using skinfold anthropometry and bioelec­ door clothing and without shoes, using a balance scale trical impedance analysis. Both techniques were under­ (Rent a Scale, London, UK) or weighing scale (Seca, taken on individual subjects within one 15-min period by Hamburg, Germany). Height was measured to the nearest the same experienced observer (AMM). Controls were not 0.5 cm with subjects wearing socks or bare feet, using a exercised for ambient temperature, recent food or fluid wall mounted measure or pocket stadiometer (Holtain Ltd, Crymych, Dyfed, UK). Percentage body fat was cal­ culated using RJL software (RJL Systems, Mt Clemens, TABLE 1 Michigan, USA) based on published regression equations Details of the aetiology and functional severity of the liver injury in (31). the 60 patients with cirrhosis under study Repeatability was assessed by remeasurement of Aetiology of cirrhosis Child’s grade Total A B C TABLE 2 11 12 Alcohol 6 29 Characteristics of the patients with cirrhosis and the matched healthy Hepatitis C 2 4 3 9 volunteers Primary sclerosing cholangitis 2 3 0 5 Cryptogenic 0 4 1 5 Healthy volunteers Patients with cirrhosis Primary biliary cirrhosis 1 2 1 4 (n=60) (n=60) a-I-Antitrypsin deficiency 0 3 0 3 Sex ratio (m:f) 40:20 40:20 Hepatitis B 1 1 0 2 Age (years) 50.1 (22-71) 50.4 (19-72) Autoimmune CAH 0 1 0 1 Height (m) 1.68 (1.54-1.91) 1.67 (1.53-1.90) Methotrexate 0 1 0 1 Weight (kg) 69.6 (48.6-110.4) 70.3 (44.7-101.3) Wilson’s disease 0 1 0 1 BMI (kg m"^ 24.7 (17.9-31.6) 25.2 (18.9-39.6) Total 12 (20%) 31 (52%) 17 (28%) 60 Values expressed as mean (range). CAH=Chronic active hepatitis. BMI=Body mass index. A. M. MADDEN and M. Y. MORGAN variables in four healthy volunteers and in four patients with cirrhosis between 24 and 72 h after the initial assess­ i 10 - ment. Data were analysed using the method suggested by I Mean Bland & Altman (32). Repeatability coefficients were cal­ culated for each method separately in each study group. I - 10- -2SD Comparative data for percentage body fat estimates were - 20 - displayed on scatter plots and the mean and standard de­ is viations of the difference calculated for each study popu­ i| * I 0 5 10 IS 20 25 30 35 40 45 lation. The limits of agreement, defined as the mean±2 Average peicentage tiody tat by sklnfold anthropometry and bioelectrical impedance standard deviations of the difference, and their 95% con­ (BFAtBF,)/2 fidence intervals were then calculated. Fig. 2. Degree of agreement between measurements of percentage body fat made in 60 healthy volunteers using skinfold anthropometry (BFa) and bioelectrical impedance analysis (BFi). Differences be­ tween the measurements (BFa-BFi) were plotted against the mean Results of the two readings (BFa+BFi)/2 for individual data sets, as sug­ gested by Bland & Altman (32). The closer the relationship between The repeatability coefficient for the measurement of the difference and the mean of the difference, the greater the agree­ percentage body fat by skinfold anthropometry was 1.7% ment between the two measurements. for both healthy volunteers and patients with cirrhosis. The repeatability coefficient for measurements made by bioelectrical impedance analysis was 3.2% for healthy vol­ 20-1 unteers and 1.7% for patients with cirrhosis (Fig. 1). 10- In the healthy volunteers, the mean±SD percentage body fat, assessed using anthropometry, 22.5±6.9% and Mean bioelectrical impedance analysis, 22.6±6.5% were com­ parable. However, individual comparative values varied widely (Fig. 2). The meant SEM of the difference between Mean-2SD the measurement was 0.52±0.55%, while the 95% limits of agreement were -7.98±0.95% and +9.0210.95%. ■30

In the patients with cirrhosis, the meant ISD percent­ Average percentage body (at by sidntold anthropometry and bioelectrical Impedance age body fat assessed using bioelectrical impedance analy­ (BFa+BF|)/2 sis, 24.418.9%, was significantly greater than the value Fig. 3. Degree of agreement between measurements of percentage obtained using skinfold anthropometry, 20.318.4% body fat made in 60 patients with cirrhosis, with (0) and without (•) fluid retention, using skinfold anthropometry (BF^) and bioelectrical (p<0.01). Individual comparative values varied widely impedance analysis (BFJ. Differences between the measurements (Fig. 3). The meanlSEM of the difference between the (BFa-BFi) were plotted against the mean of the two readings (BFa+BFi)/2 for individual data sets as suggested by Bland & Alt­ man (32). The closer the relationship between the difference and the mean of the difference, the greater the agreement between the two Healthy Volunteers (n=4) measurements. 40- SF 40- BIA

5 30- ' 30-

^ 20- 20- measurements was -4.210.92%, while the 95% limits of ^ 10- 10- agreement were -18.3811.59% and +9.9811.59%.

0- — 1------1 0- Reacting 1 Reading 2 Reading 1 Reading 2 In patients with cirrhosis with fluid retention the meanll SD percentage body fat assessed using bioelec­ Patients with Cirrhosis (n=4)

SF BtA trical impedance analysis, 24.518.5% was significantly 40- 40-1 greater than the corresponding anthropometric value, 3 30- 30- 18.417.1% (p<0.005) whilst values in the patients without i 20- ' — ; overt fluid retention were comparable, 24.419.4% cf 5® 10- 10-

0- — 1 0- 22.119.2%. The meanlSEM of the difference between Reading 1 Reading 2 Reading 1 Reading 2 measurements in the patients with cirrhosis with fluid re­ Fig. 1, Repeat measurements of percentage body fat undertaken in tention was -6.111.4% while the limits of agreement four healthy volunteers and four patients with cirrhosis using skinfold anthropometry (SF) and bioelectrical impedance analysis (BIA). All were -21.311.7% and +9.111.7%. In the patients with readings were undertaken by the same observer over one 72-h period. cirrhosis without overt fluid retention the meanlSEM of BODY FAT MEASUREMENTS IN CIRRHOSIS 881 the difference between measurements was -2,2 ±1.1% and In the patients with cirrhosis the mean percentage body the limits of agreement -14.3±1.4% and +9.9±1.4%. fat, determined anthropometrically was lower than in the healthy volunteers, especially in the subgroup of patients with overt fluid retention. This confirms the results of pre­ Discussion vious studies showing depletion of fat stores in patients Skinfold anthropometry is an indirect, prediction tech­ with cirrhosis, especially in those with decompensated dis­ nique which can be used to measure body composition in ease (2-5,36). In contrast, the results of impedance ana­ man. In healthy individuals skinfold anthropometry pro­ lysis suggested that, if anything, fat stores were increased vides estimates of percentage body fat which compare in the patients with cirrhosis irrespective of the functional favourably with direct measurements made using total severity of their liver disease. There was, therefore, con­ body densitometry (30). However, little or no validation siderable variation in comparative values for percentage has been undertaken of the accuracy of skinfold anthro­ body fat determined by the two techniques, and hence pometry in patients with chronic liver disease. Neverthe­ httle agreement in the results. Thus, values determined by less, the technique is routinely used to obtain body com­ impedance analysis could be from 10% less to 18% more position data in patient with cirrhosis (2,3,5) and has than corresponding anthropometric values. Even greater proved useful for monitoring body composition, in this divergence in the results was observed in the patients with patient population, over time (4). overt fluid retention. Bioelectrical impedance analysis is also an indirect, pre­ There are no other studies with which the results of the diction technique for obtaining body composition data in present study can be compared directly, although bio­ man. In healthy individuals this technique provides esti­ electrical impedance analysis has been used to obtain data mates of percentage body fat which compare favourably on other aspects of body composition in patients with with data obtained from total body densitometry (12,14, chronic liver disease (24-28). Thus, McCullough and co­ 33). Bioelectrical impedance analysis has not, however, workers (24), measured total body water and extracellular been validated for measurement of this body composition waters in 26 patients with cirrhosis using H2[*^0] isotope variable in patients with chronic liver disease. It is, never­ dilution and corrected bromide space. The results were theless, being used increasingly to obtain body composi­ then compared with estimates of total body water and tion data in this patient population (8,9). extracellular water determined, in the same individuals, Neither skinfold anthropometry nor bioelectrical impe­ using bioelectrical impedance analysis. In the 18 patients dance analysis provides unequivocally correct measure­ with cirrhosis without ascites there was an excellent cor­ ments of pereentage body fat. The anthropometric tech­ relation between H 2[** 0] dilution and impedance measure­ nique is well established and is therefore widely used in clin­ ments of total body water. However, in the eight patients ical practice; the place of bioelectrical impedance analysis is with cirrhosis with ascites the correlation between the two yet to be decided. If it could be shown that there were close measurements was poor. The correlation between bromide agreement between the results of the two techniques, then space and impedance measurements of extracellular water they could be used interchangeably or the newer technique was poor in both patient sub-groups. Impedance measure­ might even be used to replace the established technique, es­ ments were repeated in the patients with ascites after re­ pecially if it had other advantages. moval of, on average, 6.9 1 of ascitic fluid. Extrapolation In the present study the degree of agreement between from the pre- and post-drainage impedance measurements the results of body fat measurements obtained with the indicated that total body water had been reduced by, on two techniques was assessed using the method of Bland & average, only 2.1 1. These workers concluded, therefore, Altman (32). In healthy volunteers mean values for per­ that bioelectrical impedance analysis is not an accurate centage body fat were comparable between methods con­ technique for estimating total body water and extracellu­ firming the results of previous studies (10,34). However, lar water in patients with cirrhosis. Other workers have there was considerable variation in comparative values, on" also shown that impedance measurements significantly an individual basis, such that estimates of percentage underestimate total body water losses in patients with cir­ body fat made using impedance analysis could be from rhosis following paracentesis, and have similarly con­ 9% less to 8% more than corresponding anthropometric cluded that the technique is unsuitable for monitoring values. These findings support the recommendations fluid changes in this patient population (25-27). made previously (12,14,35) that bioelectrical impedance Lautz and colleagues (28) used measurements of total analysis is best used to obtain body composition data, in body potassium and bioelectrical impedance analysis to healthy volunteers, in the context of population surveys obtain data on body cell mass in 30 patients with cir­ or large epidemiological studies. rhosis. Overall there was close correlation between values 882 A. M. MADDEN and M. Y. MORGAN for body cell mass derived from the two techniques, except Acknowledgements in patients who were significantly malnourished. In this patient subgroup the impedance data significantly over­ The authors wish to thank Rosemary Mallalieu and estimated the potassium-derived data. These authors con­ Carol Bateman for assistance in recruiting the subjects. cluded that bioelectrical impedance analysis could be used Dr Richard Morris for statistical advice, and the patients to obtain preliminary data on body composition in pa­ and volunteers who took part in the study. tients with cirrhosis, but that it might be necessary to con­ firm values obtained in malnourished patients by use of more complex techniques. References It would, therefore, appear that bioelectrical impedance analysis is of limited value for providing body composi­ 1. Heymsfield SB, Waki M, Reinus J. Are patients with chronic hver disease hypermetabolic? Hepatology 1990; 11: 502-5. tion data in patients with chronic liver disease. This may, 2. Camilo ME, Siqueira MC, Bivar L, Pinto Correia J. Body com­ however, reflect the inappropriateness of the regression position assessment in cirrhosis with ascites. Clin Nutr 1983; equations currently used for calculating composition vari­ Suppl, 0.72 (abstract). 3. Loguercio C, Sava E, Marmo R, Del Vecchio Blanco C, Coltorti ables, as these have invariably been derived in healthy in­ M. Malnutrition in cirrhotic patients: anthropometric measure­ dividuals. It might be possible to devise more appropriate ments as a method of assessing nutritional status. Br J Clin Pract regression equations using data derived from comparative 1990; 44: 98-101. 4. Sokol RJ, Stall C. Anthropometric evaluation of children with studies of more direct methods for evaluating body com­ chronic liver disease. Am J Clin Nutr 1990; 52: 203-8. position. For example, multiple isotope dilution tech­ 5. Merli M, Riggio 0, Capocaccia L. (Italian Multicentre Co­ niques and in vivo neutron activation analysis. However, operative Project on Nutrition in Liver Cirrhosis). Nutritional status in liver cirrhosis. J Hepatol 1994; in press. it is likely that any such regression equations would only 6. Frost GS, Corish C. Reproducibility of upper-arm anthro­ be of value in the specific populations for which they were pometry in subjects of differing body mass. J Hum Nutr Diet devised. Multichannel frequency devices are now avail­ 1989; 2: 403-6. 7. Loguercio C, Gentile S, Spinelli P, Orlando C. Del Vecchio able and it is possible that they may provide more valid Blanco C, Coltorti M. Studio nutrizionale del cirrotico. Con- data, in this patient population, but this has yet to be fronto fra plicometria cutanea ed acqua corporea. Riv It Nutr assessed. Parent Ent 1988; 6: 23-30. 8. Muller MJ, Fenk A, Lautz HU, Selzberg 0, Canzler H, Balks HI, It is our opinion, that bioelectrical impedance analysis et al. Energy expenditure and substrate metabolism in ethanol- should not be used in this patient population and that, of induced liver cirrhosis. Am J Physiol 1991; 260: E338-44. currently available bedside techniques for assessing body 9. Muller MJ, Lautz HU, Plogmann B, Bürger M, Korber J, Schmidt FW. Energy expenditure and substrate oxidation in pa­ fat stores, skinfold anthropometry remains the most use­ tients with cirrhosis: the impact of cause, clinical staging and ful. However, caution is still required in its application as nutritional state. Hepatology 1992; 15: 782-94. measurements are subject to both intra- and inter-ob- 10. Kushner RF, Haas A. Estimation of lean body mass by bioelec­ trical impedance analysis compared to skinfold anthropometry. server error (6), although this can be minimised if the Eur J Clin Nutr 1988; 42:101-6. method is standardised and the observer experienced 11. Heitman BL. Evaluation of body fat estimated from body mass (3,6). Measurement of the suprailiac skinfold is difficult index, skinfolds and impedance. A comparative study. Eur J Clin Nutr 1990; 44: 831-7. and may be painful for subjects with tense ascites and is 12. Lukaski HC, Johnson PE, Bolonchuk WW, Lykken GI. Assess­ impractical in recumbent patients. The presence of subcu­ ment of fat-free mass using bioelectrical impedance measure­ taneous oedema may affect skinfold reading (7), and for ments of the human body. Am J Clin Nutr 1985; 41: 810-7. 13. Segal KR, Gutin B, Presta E, Wang J, Van Itallie TB. Estimation this reason use of the four-site formula recommended by of human body composition by electrical impedance methods: a Durnin & Womersley (30) is more useful than the seven- comparative study. J Appl Physiol 1985; 58:1565-71. site formula suggested by Jackson & Pollock (37), which 14. Lukaski HC, Bolonchuk WW, Hall CB, Siders WA. Validation of tetrapolar bioelectrical impedance method to assess human included a lower-limb site. Finally, it should be remember­ body composition. J Appl Physiol 1986; 60: 1327-32. ed that the calculation of percentage.body fat from skin­ 15. Abu Khaled M, McCutcheon MJ, Reddy S, Pearman PL, Hunter fold data relies on certain assumptions validated in GR, Weinsier RL. Electrical impedance in assessing human body compbsitiOûT the BIA method. Am J Clin Nutr 1988; 47: 789- healthy individuals which may not apply to patients with 92. cirrhosis. 16. Guo S, Roche AF, Houtkooper L. Fat-free mass in children and There is a paucity of body composition data in patients young adults predicted from bioelectrical impedance and anthro­ pometric variables. Am J Clin Nutr 1989; 50: 435-43. with chronic liver disease. Studies are needed using direct 17. Deurenberg P, van dcr Kooij K, Evers P, Hulshof T. Assessment methodologies such as isotope dilution and ‘in vivo neu­ of body composition by bioelectrical impedance in a population tron activation. When these data are available, the value aged >60 years. Am J Clin Nutr 1990; 51: 3-6. 18. Kushner RF, Schoeller DA. Estimation of total body water by of the various indirect techniques in current use can be bioelectrical impedance analysis. Am J Clin Nutr 1986; 44:417- more fairly assessed. 24. BODY FAT MEASUREMENTS IN CIRRHOSIS 883

19. Fredrix EWHM, Saris WHM, Soeters PB, Wouters EFM, Kester caloric malnutrition in liver cirrhosis. Clin Invest 1992; 70: 478- ADM, von Meyenfeldt FM, et al. Estimation of body compo­ 86 . sition by bioelectrical impedance in cancer patients. Eur J Clin 29. Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, Nutr 1990; 44; 749-52. Williams R. Transection of the oesophagus for bleeding oeso­ 20. Gray DS. Changes in bioelectrical impedance during fasting. Am phageal varices. Br J Surg 1973; 60: 646-9. J Clin Nutr 1988; 48:1184-7. 30. Durnin JVGA, Womersley J. Body fat assessed from total body 21. Katch FI, Soloman RT, Shayevitz M, Shayevitz B. Validity of density and its estimation from skinfold thickness: measurements bioelectrical impedance to estimate body composition in cardiac on 481 men and women aged from 16 to 72 years. Br J Nutr and pulmonary patients. Am J CUn Nutr 1986; 43:972-3. 1974; 32: 77-97. 22. Jebb SA, Elia M. Assessment of changes in total body water in 31. Kushner RF, Schoeller DA, Fjcld CR, Danford L. Is the impe­ patients undergoing renal dialysis using bioelectrical analysis. dance index (ht^/R) significant in predicting total body water? Proc Nutr Soc 1991; 50:153A. Am J Clin Nutr 1992; 56: 835-9. 23. de Lorenzo A, Barra PFA, Sasso OF, Battistini NC, Deurenberg 32. Bland JM, Altman DG. Statistical methods for assessing agree­ P. Body impedance measurement during dialysis. Eur J Clin Nutr ment between two methods of clinical measurement. Lancet 1986; 1991; 45: 321-5. i: 307-10. 24. McCullough AJ, Mullen KD, Kalhan SC. Measurement of total 33. Lukaski HC, Bolonchuk WW Theory and validation of the tetra­ body water and extracellular water in cirrhotic patients with and polar bioelectrical impedance method to assess human body without ascites. Hepatology 1991; 14:1102-11. composition. In; Ellis KJ, Yasumura S, Morgan WD, eds. In Vivo 25. Guglielmi FW, Contento F, Laddaga L, Panella C, Francavilla Body Composition Studies. London: The Institute of Physical A. Bioelectrical impedance analysis: experience with male pa­ Sciences in Medicine, 1987; 410-4. tients with cirrhosis. Hepatology 1991; 13: 892-5. 34. Deurenberg P, van Der Kooy K. Bioelectrical impedance or 26. Zillikens MC, van den Berg JWO, Wilson JHP, Swart GR. anthropometry? Eur J Clin Nutr 1989; 43: 503-4. Wholebody and segmental bioelectrical-impedance analysis in 35. Deurenberg P, Smit HE, Kusters CSL. Is the bioelectrical impe­ patients with cirrhosis of the liver: changes after treatment with dance method suitable for epidemiological field studies? Eur J ascites. Am J Clin Nutr 1992; 55: 621-5. Clin Nutr 1989; 43: 647-54. 27. Cabré E, de Leon R, Planas R, Doménech E, Guilera M, Bertrân 36. Morgan MY. Nutrition and the liver. In: Millward-Sadler GH, X, et al. Body impedance analysis (BIA) is not a reliable method Wright R, Arthur MJP, eds. Wright’s Liver and Biliary Disease, for assessing body composition in cirrhotics with ascites. Clin 3rd Edn. London: WB Saunders Co Ltd, 1992; 1:106-73. Nutr 1992; US; 99 (abstract). 37. Jackson AS, Pollock ML, Generalized equations for predicting 28. Lautz HU, Selberg 0, Korber J, Bürger M, Müller MJ. Protein- body density of men. Br J Nutr 1978; 40:497-504. !■ : 11V il'j* iftif-.r't V i ■ REVIEW ARTICLE Nutrition Vol. 13, No. 1, 1997 ROYAL FhZmAL

I f i ‘ ,1 r \ ] \iiA \'AAA) The Potential Role of Dual-Energy X-ray Absorptiometry in the Assessment of Body Composition in Cirrhotic Patients

ANGELA M. MADDEN, BSc, AND MARSHA Y. MORGAN, MD, FRCP

. From the University Department o f Medicine, The Royal Free Hospital and School of Medicine, London, United Kingdom

Date accepted: 11 April 1996

ABSTRACT Very little information is available on body composition in patients with cirrhosis. Difficulties arise in studying these patients because they tend to retain fluid and this results in changes in tissue density and in the hydration fraction of fat- free mass. As the classic body composition techniques rely on the assumption that these variables remain constant, use of these methods will result in either under- or overestimates of body composition variables. Use of multicomponent models, employing two or more measurement techniques, will obviate the need for some of the assumptions inherent in the use of sin^e techniques, thereby increasing the accuracy of the assessments without loss of precisioiL Dual-energy x-ray absorptio­ metry can be used to measure total body bone mineral, fat, and fat-free soft tissue mass. In healthy individuds excellent agreement is observed between data obtained using this technique and data obtained from the more established reference methods. However, the degree to which the absorptiometry measurements of soft tissue are sensitive to changes in hydration is not known. Thus, in order to assess this method of body composition analysis in patients with chronic liver disease, a multicomponent model must be devised which incorporates the absorptiometry technique and allows cross-validation of the individual component measures. ©Elsevier Science Inc. 1997. Nutrition 1997; 13:40-45

Key words: body composition methodology, cirrhosis, dual energy X-ray absorptiometry, niulti-component model

INTRODUCTION need to be accurately defined so that variables such as energy Patients with cirrhosis are frequently malnourished;this expenditure and protein synthesis can be correctly referenced.''’ is adequately documented and its importance in terms of out­ Finally, “bedside” techniques need to be validated so that their come well recognized.^"’ The changes in body composition that use in the clinical setting can be optimized.'^ occur in this population are, however, poorly researched and, In this article, body composition variables will be defined, consequently, inadequately documented. There are a number body composition moidels will be described, the use of dual­ of reasons for this, not least was the fact that many of the energy x-ray absorptiometry for measurement of body composi­ assumptions made in the assessment of body composition in tion will be outlined, its use in patients with cirrhosis will be healthy individuals cannot be made in those who have abnor­ discussed, and proposals for future studies will be delineated. malities of fluid homeostasis and compartmentalization,® im­ paired protein and urea synthesis,®*'® and defective bone model­ BODY COMPOSITION TERMINOLOGY ing and remineralization. “ A number of terms are used in the description of body composi­ Accurate body composition data are required for a nuinber tion in man (Table I). Body fat mass is defined as the quantity of reasons. First, to allow definition of the pattern of tis^e of pure fat in the body, whereas adipose tissue is defined as'^ loss, which might provide insight into the disease process and fat mass togetha^with its supporting cellular and extracellular information on outcome and prognosis; it might also affect structures; adipose tissue in healthy individuals comprises sqiprox- approaches to nutritional therapy, and its monitoring and evalu­ imately 83% fat, 2% protein, and 15% water.'® Lean body mass ation.'’*'^ Second, metabolic^y active tissue compartments is defined as the part of the body that is free of adipose tissue;

Correspondence to: M. Y. Morgan, MD, FRCP, University Department of Medicine, Royal Free Hospital and School of Medicine, Pond Street, Hampstead, London NW3 2QG, United Kingdom.

Nutrition 13:40-45,1997 ©Elsevier Science Inc. 1997 0899-9007/97/$17.00 Printed in the USA. All rights reserved. ELSEVIER PH: 80899-9007(96)00321-8 DXA AND BODY COMPOSITION IN CIRRHOSIS 41

TABLE I. In applying this technique it is assumed that the densities of fat and fat-free mass are constant at 0.9 and 1.1 kg/L, respec­ BODY COMPOSITION TERMINOLOGY tively, that there is no variability in the water content of the fat-free mass, which remains at 72-74%, and likewise that Term Definition there is no variation in bone mineral density. In the presence of fluid retention and osteoporosis, both of which are commonly Body fat mass Quantity of pure fat in the body observed in patients with cirrhosis,® ** the density of the fat- Adipose tissue mass Fat plus supporting cellular and free mass will be reduced, resulting in an underestimate of the extracellular tissues (pure fat 83%, size of this , compartment. Indeed, such variation in hydration protein 2%, water 15%) and bone mineral density could, theoretically, introduce errors Lean body mass* Nonadipose tissue body mass (body cell of 3-4% when predicting percentage body fat in this population mass plus extracellular water and using this technique.’®’’^ extracellular solids) Potassium is the major cation in lean tissue so that measure­ Fat-free mass* Lean body mass plus nonfat component ment of total body potassium can be used to estimate lean body of adipose tissue mass.’*’’* Total body potassium can be estimated by measuring Body cell mass Cellular components of body including emission radiation from the naturally occurring isotope '**K water (lean body mass minus using a sensitive, well-shielded, whole-body counter. extracellular water and extracellular In using this technique it is assumed that fat is potassium- solids) free and that the potassium content of lean tissue is constant Extracellular solids Total body bone mineral (skeleton) plus at 60 mmol/kg for women and 66 mmol/kg for men.” How­ fascia and cartilage (skeleton 85%, ever, the presence of cirrhosis may be complicated by the devel­ other 15%) opment of hypokalemia and a reduction in total body potas­ sium.®® Thus, use of total body potassium to assess lean body mass in patients with cirrhosis may result in underestimation * Often used interchangeably, but incorrectly. of this body component. i Measurement of total body water (TBW), by isotope dilu­ tion, is used to predict lean body mass based on the assumption it thus comprises body cell mass plus extracellular water and that this compartment contains a relatively fixed proportion of extracellular solids. Fat-free mass, a tom that is sometimes used water, whereas fat is anhydrous. However, adipose tissue has interchangeably, althou^ incorrectly, with lean body mass, con­ been shown to contain approximately 15% water by weight,”’®* sists of lean body mass together with the nonfat componaits of which makes assessment of TBW unsuitable as a measure of adipose tissue. Body cell mass is the term used to define the body composition if used alone. If, however, extracellular water cellular components of the body including intracellular watCT. The (ECW) is separately assessed, then intracellular water (IC!W) extracellular solids consist of total body bone mineral, that is, can be calculated as the difference between TBW and ECW. the-skeleton, which in healthy individuals comprises 85% of this ICW is then used to predict body cell mass (BCM) by assuming compartment, plus fascia and cartilage.** that the hydration fraction of BCM is 72%:®* BODY COMPOSITION MEASUREMENT TECHNIQUES TBW - ECW = ICW Attempts to determine the composition of the human body began during the 1940s.*’ Subsequently, a variety of techniques ICW BCM = have been introduced that preferentially measure certain body 0.72 compartments or subcompartments.**’*®’*® Some of these tech­ niques provide direct measurements of body compartments, In patients with cirrhosis there may be significant variation in while others provide only indirect information, which requires the hydration fraction of the fat-free mass. Under these circum­ cross-vahdation against the more direct techniques. However, stances use of the classic water dilution equations wül result the distinction between measurement techniques as direct or in overestimation of the size of this body compartment. indirect is arbitrary and unhelpful, and techniques are best de­ Thus, it is clear that a number of difficulties arise in using scribed in terms of their precision, accuracy, and applicability. the two-component model for assessing body comjposition, re­ The majority of the body composition techniques currently em­ flecting uncertainties over the validity of the assumptions made. ployed are based on a two-component model in which the In consequence, attempts have been made to combine the use body is assumed to be divided into fat and fat-free tissues.’* In of two or more techniques so that the components of the fat-free applying this model to body composition analysis a number of mass can be measured more directly and with greater certainty. assumptions need to be made, which, while valid in healthy These attençts have resulted in the development of, first, fluee-®’ individuals, may not be valid in disease states. This is best and, later, four-component models®®'®* of body composition. exemplified by consideration of three direct body composition techniques: densitometry, total body potassium, and total body MULTICOMPONENT MODELS “ water. The three-component model of water, fat, and protein/min­ densitometry is used to assess body fat mass. Density is eral is based bnineasurements obtained from both densitometry measured as the mass of the subject in air divided by the volume and water dilution.®’ This model overcomes some of the uncer­ of body tissue; volume is usually assessed using an underwater tainties over the hydration fraction of fat-free mass but does weighting technique.’*’” The proportion of fat in the body is assume a constant ratio of protein to mineral. The four-compo­ then calculated from the formula:’® nent model of water, fat, protein, and mineral incorporates di­ rect measurement of total body mineral content using dual- fat = 100 energy x-ray absorptiometry or in vivo neutron activation, as [f-H- well as measurements of body density and water.®®'®* This model removes the need for some,of the assumptions inherent where d = body density. in the two- and three-component models regarding the densities 42 DXA AND BODY COMPOSITION IN CIRRHOSIS of the different components of the fat-free mass and their rela­ be used as reference methods, particularly in disease states tive proportions. However, the densities of fat and protein are associated with altered hydration or bone density, and can be assumed to be constant, as are both the density of total body used to validate the individual component measurements and mineral and the ratio of osseous to non-osseous mineral. various “bedside” techniques available for body composition On theoretical grounds, the use of three- and four-compo­ analysis. nent models has clear advantages because the errors associated with variations in the proportions of water, protein, and mineral DUAL-ENERGY X-RAY ABSORPTIpMETRY are fewer (Table II). Thus, overhydration does not compromise The recent development of dual-energy projection methods the prediction of fat-free mass using the three- or four-compo­ for estimating body composition represents an important tech­ nent model, but does compromise the predictions obtained by nological advance.®’’"'® The initial technique of dual-photon ab­ densitometry and water dilution. Errors may arise using the sorptiometry (DPA) was used primarily to assess regional bone three-component model when the proportion of protein to min­ mineral density and subsequently total body bone mineral den­ eral varies and when using the four-component model when sity and content. The later technique of dual-energy x-ray ab­ the ratio of osseous to non-osseous mineral alters (Table II). sorptiometry (DXA) can also be used to determine regional Theoretically, use of more than one technique for body com­ and whole-body fat and fat-free soft tissue mass. position analysis might result in a compounding of the errors DXA data are obtained by scanning the body with x-rays at associated with use of the individual techniques. However, two different energy levels and then measuring dieir differential analysis of the propogation of errors shows that the precision attenuation to allow separation of bone from soft tissue; the of the estimates is not compromised by the use of multiple ratio of low to high energy attenuation in soft tissues allows techniques.^’^® separation of fat from fat-free mass. More complex body composition models could be devel­ There are a number of advantages to using DXA for body oped by including other more minor components of the fat-free composition analysis, (i ) The technique is relatively quick and mass, such as glycogen, by subdividing body water into extra- simple to perform, thus ensuring patient cooperation; scanning and intracellular components, and by subdividing mineral into times vary but generally do not exceed 30 min for a single osseous and non-osseous components. whole-body compositional analysis. (2) The radiation dose is In practical terms, the use of multicomponent models, and low at 2-5 fiS\. (3) The reproducibility of the measurements hence the use of several measurement techniques, increases the obtained is excellent, with agreement to within 1% for fat in time and cost of assessments. Nevertheless, these models can healthy adidts. (4) The short-term precision errors are small.

TABLE n.

THEORETICAL ERRORS IN THE ESTIMATE OF PERCENTAGE BODY FAT IN RELATION TO CHANGES IN THE COMPONENTS. OF FAT-FREE MASS (ADAPTED FROM ELIA®®)

Two-Component Model Three- Four- Component Component Component of Fat-Free Mass (% Change) Densitometry Water Dilution Model Model

Protein +20 -2.6 +3.25 0 0 0 0 0 0 0 -20 +2.78 -3.48 0 0

Water +40 . +8.54 -6.92 0 0 +20 +4.69 -3.80 0 0 0 0 0 0 0 . Osseous mineral +25 -2.52 +0.89 -0.64 +0.24 0 0 0 0. 0 -25 +2.56 -0.90 +0.65 -0.24 -50 +5.17 -0.82 +1.35 +0.48 Non-osseous mineral +20 +0.47 +0.16 -0.13 -0.19 0 0 0 0 0 -20 -0.47 +0.16 +0.13 +0.19

Assumptions: '—

Absolute amount of fat = 25% body weight Fat-free masis = 71.95% water,®^ 22.24% protein, 5.82% mineral Protein: mineral = 0.224: 0.0582®® Osseous: non-osseous mineral = 0.8191: 0.1809®® Densities: fat-free mass 1.1 kg/L protein 1.34 kg/L mineral 3.0375 kg& osseous mineral 2.982 kg/L non-osseous mineral 3317 kg/L DXA AND BODY COMPOSITION IN CIRRHOSIS 43 amounting to 0.8% for total body bone mineral and 1.5% for in 21 patients with primary biliary cirrhosis, 4 (19%) of whom lean body mass.^ (5) Data can be analyzed to provide informa­ had ascites. They found a significant correlation between estimates tion on whole-body composition or else to provide information of percentage body fat made using DXA and measurements of on the composition of body segments, muscle mass, and the triceps skinfold thickness (r = 0.68) and mean arm circumference distribution of fat between trunk and limbs.'” . (r = 0.55). No otha: body composition variables were assessed There are also a number of disadvantages to the use of DXA and no other comparisons were made; therefore, this study does for body composition analysis.'*^ (i) The scanning area of the not validate the use of DXA in this patient population. rnachine is relatively small, which may preclude its use in obese Bramley and colleagues*’ used DXA to assess total body or very tali subjects; indeed, the accuracy with which individual fat in 53 patients with cirrhosis, 23 (43%) of whom had sig­ segments may be distinguished declines as body size increases. nificant fluid retention. Comparisons were then made with mea­ (2) Differences exist in both the hardware and softw.are used surements of total body fat derived from skinfold anthropome­ by the various manufacturers with the result that data obtained try, bioelectrical impedance analysis, and total body potassium. with,different machines cannot be used interchangeably.'^^’^ (3) There was a highly significant correlation between tire estimates The algorithms used for analysis are continually being revised, of total body fat made using DXA and skinfold anthropometry resulting in repeated software upgrading and the need for fre­ (r = 0.93 ), even in the presence of ascites, but poor agreement quent data reanalysis. (4) A variety of other technical limita­ between tire DXA estimates and those made using the other tions exist, such as beam hardening, edge differentiation, and methods. the choice of material for calibration, which need to be further These authors, in an apparently separate study,®’ used DXA investigated. (5) Alteration in the hydration status of soft tis­ to obtain estimates of total body fat, total body lean tissue mass, sues can alter attenuation of the dual-energy source, resulting in and total body bone mineral content. These measures were then an incorrect estimate of the calculated lean tissue per pixel:'*^ '*^ combined with assessments of body cell mass from total body conversely, changes in hydration status do not affect the esti- potassium measurements to derive a four-component model miate of totol body bone mineral,although these measure­ comprising total body fat, extracellular water, extracellular sol­ ments may be affected by tissue thickness.'^® (6) There is evi­ ids, and body ceU mass. This model was then used to compare dence from in vitro studies that DXA measurements may under­ body composition variables in 54 healthy controls and 55 pa­ estimate absolute fat mass and that the error increases with the tients with cirrhosis, 23 (42%) of whom had ascites. Mean body proportion of fat.^’ , cell mass was reduced in the cirrhotic patients independently of In healthy adult Caucasians, excellent agreement is observed the presence of fluid retention. Extracellular water was signifi- between (/) measurements of percentage body fat made by cantiy increased in patients with ascites while extracellular sol­ DXA and both densitometry,and water dilution®^*^®’^ (2) ids were increased selectively in male cirrhotic patients with measurenients of fat-free soft tissue mass made by DXA and ascites. However, the authors caution against the suitability of either densitometry or water dilution^*’^ and (3) measurements this model because of doubts over the validity of the measure­ of lean body mass made by. DXA and However, DXA ments of total body potassium and, by extrapolation, the esti­ cannot yet be considered a “gold standard” method for body mates of body cell mass in their patients with cirrhosis. composition analysis for a number of reasons. (J) The degree to which measurements of soft tissue are sensitive to changes in PROPOSALS FOR FUTURE STUDIES USING DXA IN PATIENTS hydration is unknown. (2) The degree to which body thickness WITH CIRRHOSIS affects measurements of soft tissue mass is uncertain. (3) Evi­ dence froih in vitro studies that DXA measurements may under­ In order to assess body composition in patients with cirrhosis estimate fat mass,'*’ needs to be further evaluated. a model is required that either takes into account, or else can At present it is probably not advisable to rely on body composi­ be applied independently of, changes in hydration and bone tion data provided by DXA alone, particularly in sick or injured mineralization. The four-component model devised by Fuller individuals. DXA measurements could, however, be incorporated and colleagues*® enables measurements of fat, water, protein, into a multicomponent model such as that devised by FuUa: and and bone mineral to be made independently of these variables. coUeagues;®^ These workers combined measurements from densi­ The data are compiled using DXA to measure total body bone tometry, deuterium dilution, and DXA to provide a four-compo­ mineral, underwater weighing to determine body density, and nent model of water, fat, protein, and mineral, which they then deuterium dilution to determine total body water. This model evaluated in healthy adults. They found tihat the detaminations could be modified for use in patients with chronic liver disease of fat and fat-free mass obtained using DXA agreed slightly less by including a measure of extracellular water using a bromide well with values from the four-conçonait model than the values dilution technique.®* In their study of healthy volunteers. Fuller obtained from densitometry and wata^ dilution. Nevertheless, and colleagues*® found that DXA was the best predictor of DXA was the best predictor of mean body composition obtained mean body composition of the reference methods used and that by four separate two-component reference methods combining the it provided a suitable alternative to their four-component model use of densitometry, deuterium dilution, DXA, and total body for assessment of body composition in healthy individuals. potassium. Indeed, DXA provided a suitable alternative to their If body composition were assessed in patients with cirrhosis, four-component model for assessment of. body composition., in using a modification of the model devised by Fuller and col­ these hedthy individuals. leagues,*® then similar evaluations of the individual component techniquesV'and'Of.DXA in p^cular, could be undertaken. USE OF DUAL-ENERGY PROJECTION TECHNIQUES IN PATIENTS WITH CIRRHOSIS SUMMARY DPA has, been used to assess bone mineral density in the Patients with cirrhosis retain fluid and, hence, classic body lumbar spine of patients with primary ttliary cirrhosis.^’” Detailed composition assessment techrtiques, which rely on an assump­ serial measurements were made in one study to evaluate the effects tion of constant hydration, cannot be used in this patient popula­ of both disease progression and orthotopic liver transplantation;®^ tion. It is hypothesized that dual-energy x-ray absorptiometry, no other body composition variables woe assessed. used singly or as part of a multicomponent model, might facili­ Wicks and colleagues®* used DXA to assess body composition tate such assessments in these patients. ,44 DXA AND BODY COMPOSITION IN CIRRHOSIS

REFERENCES

1. Morgan MY. Enteral nutrition in chronic liver disease. Acta Chir 26. Pierson RN, Jr, Wang J, Thornton JC, Van Itallie TB, Colt EWD. Scand 19Sl;507(suppl.):81 Body potassium by four-pi ^K counting: an anthropometric correc­ 2. DiCecco SR, Wieners EJ, Wiesner RH, et al. Assessment of nutri­ tion. Am J Physiol 1984; 246:F234 tional status of patients wiA end-stage liver disease undergoing 27. Garrow JS. New approaches to body composition. Am J Clin Nutr liver transplantation. Mayo CHn Proc ,1989; 64:95 1982; 35:1152 ' ' . 3. Italian Multicentre Cooperative Project. Nutritional status in cirrho- 28. Perez GO, Osker JR. Altered potassium in liver disease. In: Epstein sis. J Hepatol 1994;21:317 M, ed. The kidney in Over disease. New York: Elsevier Biomedical, 4. Shaw BW, Jr, Wood RP, Gordon RD, et al; Influence, of selected 1983:183 variables and operative blood loss on six-month survival following 29. Wang J, Pierson RN, Jr. Disparate hydration of adipose and lean liver transplantation. Sem Liver Dis 1985;5:385 tissue require a new model for body water distribution in man. J 5. Adler M, Gavaler JS, Duquesnoy R, et d. Relationship between Nutr 1976; 106:1687 the diagnosis, preoperative evaluation and prognosis ^ r or- 30. Jensen MD, Martin ML, Kanaley JA. Fat associated water: implica­ thotopic liver transplantation. Ann Surg 1988;208;196 tions for body composition assessment. Qin Res 1991; 39:205A 6. Halliday AW, Benjamin IS, Blumgard LH. Nutritional risk factors 31. Cohn SH, Vaswani AN, Yasumura S, Yuen K, Ellis KJ. Assessment in major hepatobiliary surgery. JPEN 1988; 12:43 of cellular mass and lean body mass by non-invasive nuclear tech­ 7. Moukarzel AA, Najm L Vargas J, et al. Effect of nutritional status niques. J Lab Clin Med 1985; 105:305 on outcome of orthotopic liver transplantation in pediatric patients.. 32. Siri WE. Body composition from fluid spaces and density: analysis Transplant Proc 1990;22:1560 of methods. In: Brozek J, Henschel A, eds. Techmquesfor measur­ 8. Arroyo V, Ginès P, Jimenez W, Rodés J. Ascites, renal failure, ing body composirfon. Washington DC: National Academy of Sci­ and electrolyte disorders in cirrhosis. Pathogenesis, diagnosis and ences, National Research Council, 1961:223 treatment In: McIntyre N, Benharaou J-P, Bircher J, Rizzetto M, 33. Lohman TG. Research progress in validation of laboratory methods Rodés J, eds.Oxford textbook ofcliniccd hepatology. Oxford: GUP, of assessing body composition. Med Sci Sports Exerc 1984; 16:596 1991:429 34. Heymsfield SB, Lichtman S, Baumgartner RN, et al. Body compo­ 9. Herrmann R, McIntyre N. Amino-acid metabolism, urea production sition of humans: comparison of two improved four-compartment and pH regulation. In: McIntyre N, Benhamou J-P, Bircher J, Riz­ models that differ in expense, technical complexly, and radiation zetto M, Rodés J, eds. Oxford textbook of clinical hepatology. exposure. Am J Clin Nutr 1990;52:52 Oxford: GUP, 1991:157 35. Fuller NJ, Jebb SA, Laskey MA, Coward WA, Elia M. Four- 10. Gerok W, Gross V. Secretory proteins; synthesis, secretion, and component model for the assessment of body composition in hu­ function. In: McIntyre N, Benhmou J-P, Bircher J, Rizzetto M, mans: comparison with alternative methods, and evaluation of the Rodés J, eds.Qjford textbook cf clinical hepatology. Oxford: GUP, density and hydration of fat-free mass. Clin Sci 1992;82:687 .1991:174 36. Elia M. Body composition analysis: an evaluation of 2 component 11. Compstori J, The effect of liver disease on bone. In: McIntyre N, models, multicomponent models and bedside techniques. Clin Nutr Benhamou J-P, Bircher J, Rizzetto M, Rodés J, eds. Oxfordtextbook 1992; 11:114 o/c/irticflj/lepofology. Oxford: GUP, 1991:1263 37. Brozek J, Grande F, Anderson T, Keys A. Densitometric analysis 12. McCullough AJ, Tavill AS. Disordered energy and protein metabo­ of body composition: revisions of some quantitative assumptions. lism in liver disease. Sem Liver Dis 1991; 11:265 . Ann NY Acad Sci 1963; 110:113 13. Porayko MK, DiCecco S, O’Keefe SJD. Impact of m’alnutrition 38. Fuller NJ, Jebb SA, Goldberg GR, et al. Inter-observer variability in , and its therapy on liver transplantation. Sem Liver Dis 1991 ; 11:305 the measurement of body composition. Eur J Clin Nutr 1991;45:43 14. Heymsfield SB, Waki M, Reinus J. Are patients with chronic liver 39. Mazess RB, Peppier WW, Gibbons M. Total body composition by disease hypermetabolic? Hepatology 1990; 11:502 dual-photon ("^Gd) absorptiometry. Am J Clin Nutr 1984;40:834 15. Madden AM, Morgan MY. A comparison of skinfold anthropome- 40. Mazess RB, Barden .HS, Bisek JP, Hanson J. Dual-energy x-ray try and bioelectrical impedance analysis for measuring percentage absorptiometry for total-body and regional bone-mineral and soft- body fat in patients with cirrhosis. J Hepatol 1994; 21:878 tissue composition. Am J CUn Nutr 1990;5i:l 106 16. Jensen MD. Research techniques for body composition assessment. 41. Fuller NJ, Laskey MA, Elia M. Assessment of the composition of J Am Diet Assoc 1992;92:454 major body regions by dual-energy X-ray absorptiometry (DEXA) 17. Behnke AR, Feen BG, Welham WC. The specific gravity of healthy widi special reference to limb muscle mass. Clin Phys 1992; 12:253 men. JAMA 1942; 118:495 42. Roubenoff R, Kehayias JJ, Dawson-Hughes B, Heymsfield SB. Use 18. Lukaski HC. Methods for the assessment of human body composi­ of dual-energy x-ray absorptiometry in body-composition studies: tion: traditional and new. Am J Clin Nutr 1987; 46:537 not yet a “gold standard.” Am J Clin Nutr 1993;58:589 19. Bramley P, Oldroyd B, Stewart S, et al. Body composition analysis 43. Tothill P, Avenell A, Reid DM. Precision and accuracy of whole- in liver cirrhosis. The measurement of body fat by dual energy x- body bone mineral: comparisons between Hologic, Lunar and Nor­ ray absorptiometry in comparison to skinfold anthropometry, bio­ land dual;cnergy X-ray absorptiometers. Br J Radiol 1994;67:1210 electrical impedance and total body potassium. In: Ellis KJ, East- 44. Tothill P, Avenell A, Love J, Reid DM. Comparisons between man JD, eds.Human body composition. New York: Plenum Press, Hologic, Lunar and Norland dual-energy X-ray absorptiometers 1993:211 and other techniques used for whole-body soft tissue measurements. 20. Keys A, Brozek J Body fat in adult man. Physiol Rev 1953; 33:245 Eur J Clin Nutr 1994;48:781 21. Jones PRM, Norgan NG. A simple system for the determination 45. Horber FF, Thomi F, Casez JP, Fonteille J, Jaeger P. Impact of of humari body density by underwater weighing. J Physiol hydration status on body composition as measured by dual energy 1974; 239:7 IP X-ray absorptiometry in normal volunteers and patients on haemo- 22. Buskirk ER. Underwater weight and body, density. A reviewLof dialysis. Br J Radiol 1992;65:895 procedures. In: Brozek J, Henschel A, eds. Techniques for measur- 46. Brunton JA, Bayley HS, Atkinson SA. Body composition analysis' ing body composition. Washington DC: National Academy of Sci­ by dual energy x-ray absorptiometry compared to chemical analysis ences, National Research Council, 1961:90 of fat, lean and bone mass in small piglets. In: Ellis KV, Eastman 23. Siri WE. The gross composition of the body. In: Lawrence TH, JD, eds. Human body compositbn. New York: Plenum Press, Tobias CA, eds. Advances in biological and medical physics. Vol 1993:157 4. New York: Academic Press, 1956:239 47. Going SB, Massett MP, Hall MC, et al. Detection of small changes 24. Lohman TG. Skinfolds and body density and their relation to body in body composition by dual-energy x-ray absorptiometry. Am J fatness: a review. Hum Biol 1981;53:181 Clin Nutr 1993;57:845 25. Jensen MD, Braun JS, Vetter RJ, Marsh HM. Measurement of body 48.Laskey MA, Lyttle KD, Flaxman ME, Barber RW. The influence potassium with a whole-body counter relationship between lean of tissue depth and composition on the performance of the Lunar body mass and resting energy expenditure. Mayo Clin Proc dual-energy X-ray absorptiometer whole-body scanning mode. Eur 1988;63:864 J Clin Nutr 1992;46:39 DXA AND BODY COMPOSITION IN CIRRHOSIS 45

49. Jebb SA, Goldberg GR, Jennings G, Elia M. Dual-energy X-ray 55. Slosman DO, Casez J-P, Pichard C, et al. Assessment of whole- absorptiometry measurements of body composition: effects of depth body composition with dual-energy x-ray absorptiometry. Radiol­ and tissue thickness, including comparisons with direct analysis. ogy 1992; 185:593 Clin Sci 1995; 88:319 56. Hodgson SF, Dickson ER, Wahncr HW, et al. Bone loss and re­ 50. Dalsky GP, Rraemer W, Zetterland AE, et al. A comparison of duced osteoblast function in primary biliary cirrhosis. Ann Intern methods to assess body composition. Med Sci Sports Exerc Med 1985; 103:855 1990;22:S112 57 Eastell R, Dickson ER, Hodgson SF, et al. Rates of vertebral bone 51. Van Loan MD, Mayclin PL. Body composition assessment dual- loss before and after Hver transplantation in women with primary energy X-ray absorptiometry (DEXA) compared to reference meth­ bihary cirrhosis. Hepatology 1991; 14:296 ods. Eur J Clin Nutr 1992; 46:125 58 Wicks C, Bray GP, WilHams R. Nutritional assessment in primary 52. Johansson AG, Forslung A, Sjodin A, et al. Deteraiination of body bihary cirrhosis: the effect of disease severity. CUn Nutr composition—a comparison of dual-energy x-ray absorptiometry 1995; 14:29 and hydrodensitometry. Am J Clin Nutr 1993;57:323 59. Oldroyd B, Bramley PN, Stewart SP, et al. A four-compartment 53. Pritchard JE, Nowson CA, Strauss BJ, et al. Evaluation of dual model to determine body composition in Hver cirrhosis. In: ElHs energy X-ray absorptiometry as a method of measurement of body KJ, Eastman JD, eds.Human body composition. New York: Plenum fat. Eur J Clin Nutr 1993;47:216 Press, 1993:221 54. Wellens R, Chumlea WC, Guo S, et al. Body composition in white 60. Trapp SA, Bell EF. An improved spectrophotometric bromide assay adults by dual-energy x-ray absorptiometry, densitometry and total for the estimation of extracellular water volume. CHn C!him Acta body water. Am J Clin Nutr 1994;59:547 1989; 181:207 ' 'HY ... V,' i : i iJ\ I

V i r ■- Taste Perception in CirrhpsiSl ItSilRelationship to Circulating Micronutrients and Food Preferences

A ngela M. Madden,* W ayne Bradbury,^ and M arsha Y. M o r g a n *

Impairment of gustatory acuity may influence nutrient in­ in patients with dry mouths, e.g., those with primary biliary take and hence nutritional status. The aim of this study was cirrhosis and associated Sjogren’s syndrome.*** Second, pa­ to evaluate gustatory acuity in patients with cirrhosis and its tients and healthy volunteers were riot matched for certain relationship to circulating concentrations of micronutrients, factors known to affect gustatory function, such as smoking and food preferences. Gustatory evaluation was undertaken, behavior**’*^ or denture wearing.*^ using a rinsing technique, in 75 cirrhotic patients and 75 Zinc and vitamin A are both important in maintaining taste comparable healthy volunteers. Circulating concentrations of integrity, although their exact roles have not been eluci­ magnesium, zinc, vitamin A, and a- and ]3-carotene were mea­ dated.*'* Although assessment of zinc status in humans is sured, and food preferences were assessed by questionnaire. difficult, zinc depletion is commonly reported in individuals The cirrhotic patients showed impaired gustatory function with cirrhosis.*^ *^ However, no conclusive association has with significantly higher (less sensitive) median thresholds been reported between zinc depletion and impaired gustatory for detection of salt, sweet, and sour and for recognition of function in patients with liver disease.'*'^'^ *® *^ Indicators of bitter, salt, sweet, and sour, together with a higher overall vitamin A status are also reduced in this patient popula­ median gustatory score (P < .0001). Mean circulating concen­ tion,^**’^* and there is evidence that vitamin A supplementa­ trations of magnesium, zinc, vitamin A, and a~ and ^-carotene tion may improve gustatory function in cirrhotic patients were significantly lower in the patient population. Serum mag­ who are vitamin A-depleted.® Magnesium deficiency has nesium was significantly negatively associated with detection been documented in patients with cirrhosis,^^ and although of salt (P = .02) and gustatory score (P = .02). Patients’ it has not previously been associated with taste integrity, subjective assessment of taste acuity did not correspond with the anorexia commonly observed in cirrhotic patients may objective measurements. Overall, no difference» were ob­ resolve after magnesium supplementation.^^ served in food preferences between the two groups, nor was There is little information on the relationship between im­ any association found between food preferences and gustatory paired taste perception and alterations in food selection and acuity. Patients with cirrhosis have impaired gustatory acuity eating habits, although it is reasonable to suppose that they that is associated with hypomagnesemia but apparently does may be interrelated. ’^'* Aversions to fatty foods, meat, and not affect food selection. (H epatology 1997;26:4048.) fried dishes and a preference for citrus fruits have been re­ ported in patients with liver disease,although gustatory acuity was not formally tested in any of these studies. Patients with cirrhosis are often malnourished*'^; this mal- The aims of this study were: 1) to evaluate taste thresholds nourishment may, at least in part, reflect a reduction in food in patients with cirrhosis and in a group of healthy volunteers intake. Many such patients are anoreic, and although this is comparable in age, sex, smoking behavior, and denture wear­ usually ascribed to generalized ill health, the unpalatability ing; 2) to measure their circulating concentrations of magne­ of prescribed diets, the pressure effects of ascites, and the sium, zinc, vitamin A, and a- and ^-carotene; 3) to investi­ possibility that it also represents alterations in taste acuity, gate their food preferences by questionnaire; and 4) to and hence food preferences, needs to be considered. observe the interrelationships, if any, among these variables. Disorders in gustatory acuity have been reported in indi­ viduals with liver disease'*'® manifest as changes, usually a PATIENTS AND METHODS reduction, in taste acuity for detection and recognition of A total of 87 cirrhotic patients were screened for the study. The some or all of the basic tastes of bitter, salt, sweet, and sour. etiology of liver disease was determined on the basis of historical, However, the studies undertaken to date may be criticized clinical, laboratory, endoscopic, and radiological variables. The for a number of reasons. First, the dropping technique*used functional severity of the liver injury was determined using Pugh’s to evaluate gustatory acuity^ may overestimate impairment modification of the Child’s grading system.^' Patients were excluded from the study if they were unwilling or considered too sick to participate, if they did not speak English fluently, or if they were currently “nil by mouth.’’ Patients with hepatitis B- or C-related From the 'University Department of Medicine and the ^Department of Chemical cirrhosis who were positive for hepatitis B surface antigen or anti­ Patholog)’, The Royal Free Hospital and School of Medicine. London. UK. body to hepatitis C virus were excluded on the grounds that their Received August 6, 1996; accepted April 9,1997 saliva might be infectious, and therefore the investigation might Scientific Hospital Supplies. Liverpool, England, funded the preparation of test constitute a potential hazard to staff. Patients with alcohol-related solutions and analysis of plasma vitamin A and a - and /J-carotene concentration. Address reprint requests to: Marsha Y. Morgan, F.R.C.P., University Department of liver disease were excluded if they were abusing alcohol within to Medicine, Ros'al Free Medicine. Pond Street, London NW3 2QG, UK. 1 month of the study. Patients were also excluded if they were taking Copyright © 1997 by the American Association for the Study of Liver Diseases. drugs known to affect gustatory perception, e.g., penicillamine, or 0270-9139/97/2601-0007S3.00/0 if they had a history of significant ear, nose, and throat disease.

40 lit TA 1^01 'X .V Vol. lb. No. \. m i MADDEN, BHADBL'RV. AND MORGAN 41

TadI-E 1. Details of the Etiology and Functional Severity of Liver Injury bottles and medicine cups containing the solutions were randomly in 75 Cirrhotic Patients Studied assigned a letter for identification and labeled accordingly. The cups were arranged in ascending order of concentration on trays to be Child’s Grade* presented to the subject by the investigator (Table 2). Subjects The tasting procedure was carried out using the method described Etiology of C i r r h o s i s ( n ) A ( n ) C by Barale et al.^® Subjects rinsed their mouths with filtered tap water at room temperature at the beginning of testing and again between Alcohol 5 9 ( 7 9 % ) 3 3 1 3 1 3 tastings. A 20-mL sample of the lowest concentration of the first Primaiy biliary test solution was presented to the subject, who was instructed to cirrhosis 4 ( 5 % ) I 3 0 rinse the fluid around the entire oral cavity and cover all parts of Ciyptogcnic 4 ( 5 % ) 3 1 0 the tongue before expectorating. The subject was then asked if the Primary^ sclerosing sample tasted different from water and, if so, to describe the taste cholangitis 3 ( 4 % ) 2 1 0 experienced, Solutions of increasing concentration were presented Autoimmune chronic in turn until the subject was able to correctly identify the taste active hepatitis 3 ( 4 % ) ■ 0 2 1 present. Open questions were asked, and a wide variety of terms a,-Antitrypsin were accepted to describe the tastes; for example, the terms acid, deficiency 0 1 0 1 ( 1 % ) sharp, tart, citrus, and vinegar were all considered acceptable to Schistosomiasis 0 1 0 1 ( 1 % ) describe the sour taste of hydrochloric acid. The test procedure was Total 75 (100%) 39 (52%) 22 (29%) 14 (IÇ repeated if there was any doubt about the answers given. * Functional severity of liver injury assessed using Pugh’s modification The detection threshold was defined as the lowest concentration of the Child's grading system.A, Pugh’s score 5-6; B, Pugh’s score 7-9; C, at which the subject perceived a difference in the taste of the test Push’s score 10-15. solution from water. The recognition threshold was defined as the lowest concentration at which the subject could correctly identify the taste present. When the detection and recognition thresholds The resulting patient population of 75 individuals comprised 43 for the first taste had been determined, the subject thoroughly men and 32 women of median age, 48 (range, 27-77) years. The rinsed their mouth with water before the procedure was repeated majority of these patients had alcohol-related cirrhosis, and approxi­ using the second test solution. The order in which the solutions mately half had well-compensated disease (Table 1); 39 (52%) were, presented was the same for all subjects. smoked cigarettes, and 23 (31%) wore dentures. Median detection and recognition thresholds were determined in A total of 85 healthy volunteers were screened for inclusion in the healthy volunteers, for each taste, to be used as reference points the reference population. These were recruited from hospital staff, to devise an overall gustatory score for each subject. A score of 0 family and friends of patients and staff, and subjects who answered was allocated to thresholds that equaled or were below the reference newspaper advertisements and radio appeals. No subjects gave a medians, while a score of 1 was allocated to thresholds above the history or had clinical evidence of either alcohol abuse or chronic reference medians. A gustatory score was calculated for each indi­ liver disease. However, 9 individuals were excluded because the vidual from the sum of the scores for the four detection and four results of one or more biochemical tests of liver function were recognition thresholds. A score of 0 indicated excellent taste percep­ outside the reference range; another subject was excluded because tion, while a score of 8 indicated severe impairment of gustatory insufficient blood was obtained for analysis. The resulting reference acuity. population of 75 healthy volunteers comprised 35 men and 40 The gustatory evaluation was repeated in a subgroup of 10 cir­ women of median age 52 (range, 25-82) years. None gave a history rhotic patients a mean of 141 (range, 68-217) days after initial of significant ear, nose, or throat disease; 29 (39%) smoked ciga­ testing and in 5 healthy volunteers after a mean of 207 (range, 176- rettes; and 13 (17%) wore dentures. 231) days. Details of all medications, both prescribed and over the counter, Olfactory Perception. Olfactory perception was assessed in a sub­ taken within the preceding 7 days and qualitative details of current group of 32 patients comprising 17 men and 15 women of median dietary restrictions and supplements were obtained from all sub­ age 52 (range, 27-77) years, 44% of whom were smokers and 38% jects. of whom wore dentures, and in 49 healthy volunteers comprising Patients and healthy volunteers were tested at the Royal Free Hospital between 7:45 am and 7:30 pm. All subjects were asked to abstain from eating and drinking, smoking, and dental hygiene Table 2. Concentrations of Test Solutions Used to Assess Taste activities for at least 1 hour before testing. The room temperature Perception in Healthy Volunteers and Patients With Cirrhosis

in the test areas ranged from 20"C to 26“C, with a mean temperature Solution S Solution D for volunteer testing of 21.8% and for patient testing of 23. TC. Solution A Sodium Solution C Hydrochloric Auditory and visual stimulation were minimized in the test areas. Urea Chloride Sucrose Acid Concentrated sterile solutions of urea (Analar grade; BDH, Lut- (mmol/L) (mmol/L) (mmol/L) (mmol/L) teru'orth, England), sodium chloride (BP grade; Evans, Leatherhead, Al I 81 1 Cl 1 D1 0.5 England), sucrose (Analar grade; BDH), and hydrochloric acid .(An­ A2 50 32 10 Cl 10 D2 1 alar grade; BDH), representing the four basic tastes of bitter, salt, A3 100 B3 20 C3 20 D3 '2 sweet, and sour, were provided by the Royal Free Hospital Pharmacy A4 200 ''"■'B4 30 C4 30 D4 3 Department. These solutions were stored at room temperature in A5 300 B5 40 C5 40 D5 5 sealed glass bottles and had shelf lives of 3 to 18 months. Test A6 400 B6 50 C6 50 D6 10 solutions were prepared from the concentrated solutions by addi­ A7 500 B7 60 C7 60 D7 30 tion of distilled water under clean, but not sterile, conditions; 14 A8 600 B8 80 C8 80 D8 50 concentrations were prepared of each solution (Table 2), and these A9 800 B9 100 C9 100 D9 70 were then stored at 4°C for a maximum of 1 month. The solutions AID 1,000 BIO 200 CIO 200 DIO 100 were checked for concentration and bacterial contamination before All 1,500 Bll 300 Cll 300 Dll 200 use. A12 2,000 B12 400 C12 400 D12 300 On the day of testing, 20-mL aliquots of each solution were A13 2,500 B13 600 C13 600 D13 400 poured into separate plastic medicine cups (Henleys Medical Sup­ A14 3,000 B14 800 C14 800 D14 500 plies, Welwyn Garden City, England) at room temperature. The 42 MADDEN, BRADBURY. AND MORGAN H e p a t o i x x .v J u l y 1 ^ 9 7

22 m en and 27 w om en of m edian age 42 (range, 25-82) years, 49% T a b l e 3. Taste Threshold Concentrations and Gusutory Scores in 75 of w hom w ere sm okers and 16% of w hom wore dentures. Healthy Volunteers and 75 Cirrhotic Patients > Oils of lemon, peppermint, and eucalyptus were used for testing; Healthy Cirrhotic Significance of a single concentration of each was placed in a small, individually Thresholds and Volunteers Patients Iniergroup numbered bottle; bottles were presented to each subject in the same G usutory Score (mmol/L) (mmol/L) Difference order. Subjects were instructed to remove the lid and smell each one, first through one nostril while holding the other one closed, B i t t e r and then through the other. They were asked if they could detect D e t e c t i o n 50 179] 100 (128) NS the smell and if so to identify it; a score of 1 was awarded for each ( 1 - 8 0 0 ) ( 1 - 1 , 5 0 0 ) smell detected, and a score of 1 given for each smell correctly R ecognition 1 0 0 ( 2 2 0 1 2 0 0 ( 3 4 9 ) P = . 0 0 7 identified. An olfactory score was determined from the sum of these ( 1 - 3 . 0 0 0 ) ( 1 - 3 , 0 0 0 ) scores for each subject. A score of 0 indicated severely impaired Salt olfactory perception, while a score of 6 indicated an excellent sense D e t e c t i o n 1 0 ( 1 1 ] 1 0 ( 1 6 ) P = . 0 0 4 of smell. ( 1 - 6 0 ) ( 1 - 5 0 ) 5 0 ( 7 2 ) Laboratory Analysis. Blood samples were collected in serum gel Recognition 30 (60) P = . 0 0 0 9 tubes (Sarstedt, Germany) for immediate analysis of serum albumin, ( 1 - 8 0 0 ) ( 1 - 8 0 0 ) bilirubin, alkaline phosphatase, aspartate transaminase, alanine S w e e t transferase, and y-glutamyl transferase, and in coagulation tubes Detection 10 (12) 2 0 ( 1 6 ) P = . 0 4 (Sarstedt, Germany) for measurement of the prothrombin time. A ( 1 - 6 0 ) ( 1 - 8 0 ) separate sample was spun and the serum separated and stored at R ecognition 3 0 ( 3 5 1 3 0 [ 6 3 1 P = . 0 0 6 -20°C for subsequent analysis of magnesium by a colorimetric (1-300) (1-800) method using xylidyl blue^’ and zinc by atomic absorption spectros­ S o u r copy.^ Further samples were collected in lithium heparin tubes D e t e c t i o n 0 . 5 ( 0 . 6 ) 0 . 5 ( 0 . 7 1 P = . 0 0 3 (Sarstedt, Germany) using aluminum foil to protect the sample from ( 0 . 5 - 5 ) ( 0 . 5 - 2 ) light, and the separated plasma stored at -20°C for subsequent R ecognition 1 . 0 ( 1 . 6 ) 1 . 0 ( 5 1 ) P = . 0 3 analysis of vitamin A and a- and (5-carotene using the liquid chro­ ( 0 . 5 - 3 0 ) ( 0 . 5 - 2 0 0 ) matographic method of Thumham et al.^^ G ustatory Score (0-8) 2 ( 2 . 2 ) 4 ( 3 . 6 ) P < . 0 0 0 1 The reference ranges for the micronutrients were determined ( 0 - 8 ) ( 0 - 7 ) from the mean ± 2 SD of the values in the healthy volunteers. N O TE. Values expressed as m edian [m ean] (range).

Questionnaires. questionnaire A was devised and piloted by the A bbreviation: NS; not significant. authors to investigate the subjects’ subjective opinion of their senses of taste and smell and their appetites. Specific questions were asked about their choice of food tastes, food preferences, and use of condi­ ments and flavoring agents. healthy volunteers (4 [range, 0-7) vs. 2 [range, 0-8]; P < A General Health Questionnaire,’^ which uses closed questions .0001), indicating overall impairment of taste perception. on present physical, mental, and emotional status to assess general There were no significant differences in median threshold well-being, was completed by all subjects. The potential range of levels or median gustatory scores between patients with com­ scores was from 0 to 28, with lower scores indicating a greater pensated (Pugh’s score ^7) and those with decompensated sense of well-being. (Pugh’s score ^8) liver disease^^ or between patients with Statistical Analysis. Statistical analysis was performed using the Minitab software package (1991, State College, PA). The taste alcohol-related and those with nonalcoholic liver disease. thresholds and gustatory and olfactory scores were compared be­ Age had a significant effect on gustatory acuity in healthy tween groups and subgroups using the Mann-Whitney U test, while volunteers; younger individuals had superior taste percep­ the unpaired Student’s t test was used to compare magnesium, zinc, tion, but there was no significant effect of age on median vitamin A, and a- and ^-carotene concentrations between groups. threshold levels or on median gustatory score in the patient tests were undertaken to analyze the food preference question­ population. naire between groups and subgroups, and Kruskal-Wallis tests were Sex differences in gustatory acuity were also observed used to compare the distribution of variables by taste perception among healthy volunteers, with women having superior taste category from the questionnaire. The Spearman ranked correlation perception; however, this was not observed in the patients test was used to analyze the relationships between pairs of variables in which at least one variable was ordinal, while a Pearson correla­ with cirrhosis. tion was used when both variables were continuous. No significant overall differences were observed in median threshold levels or in median gustatory scores in relation to RESULTS denture wearing or smoking in either healthy volunteers or patients with cirrhosis. The patient and reference populations were well matched Olfactory Perception. In healthy volunteers, the median ol­ for age and sex distribution and did not differ significantly factory score was significantly higher in younger individuals, in numbers of smokers and denture wearers. as defined by the group median age (6 [range, 3-6[ vs. 5 Concentrations of undiluted solutions were within an ac­ [range, 3^% P = .0009); in women (6 (range, 3-6] vs. 5 ceptable limit of ±10%, and all were sterile on microbiologi­ [range, 3-6] ; P = .01); and in non-denture wearers (6 [range, cal testing. 3-6] vs. 5 [range, 3-6] ; P = .03); smoking had no significant The repeatability coefficient for the gustatory score was effect on the median olfactory score in this population sub­ 1.7% for healthy volunteers and 2.0% for cirrhotic patients. group. Age, sex, and smoking status had no significant effect The median detection and recognition thresholds for each on median olfactory scores in patients with cirrhosis. Patients of the four taste modalities, bitter, salt, sweet, and sour, were who wore dentures had a significantly lower median olfactory consistently higher in the patient population than in the score than individuals who did not (3 [range, 2-6] vs. 5 healthy volunteers (Table 3). In patients with cirrhosis, the [range, 3-6]; P = .04). median gustatory score was significantly higher than in The subgroup of cirrhotic patients who underwent olfac- lorv testing was significantly older and included more den- Magnesium i.i \urc wearers than the corresponding reference population of 5 0.9 healthy volunteers. However, when adjustments were made lor these factors, the median olfactor\'score remained signifi­ I 0. cantly lower in younger cirrhotic patients than in younger 0.5 0* healthy volunteers, as defined by group median age (4.5

1 range, 3-6] vs. 6 [range, 3-61: P = .002). Zinc 15 In healthy volunteers, a significant inverse relationship was • • A *...... 13 . à^A^àkà...... o b s e r v e d between the olfactor\- score and the gustatory score 11 - (P = .019), indicating that a better sense of smell was associ­ 9 - \ - A ...... a t e d with a better sense of taste. In cirrhotic patients there 7 . was a similar but not significant relationship. 5 Restricted Diets. Significantly more cirrhotic patients than healthy volunteers were following restricted diets (37 [49%] Vitamin A 5.0 /A vs. 11 [15%]; P < .0001); 22 (29%) patients were limiting 4.0 . . . J ^ jk A...... - . . ^ A 0A ...... sodium intake. In cirrhotic patients, there was a general but 3.0 i • • aAaÀ^â^iA'A’ ...... *...... # ..... ; : ; : ; : : ; : : : : ; : : : insignificant trend toward higher taste thresholds in those 2.0

who were following restricted diets. 1.0 Micronutrients. The mean serum magnesium concentration 0.0 was significantly lower in cirrhotic patients than in healthy a-Carotene volunteers (Fig. 1); 9% of patients and 1% of healthy volun­ 0.40 1

teers had serum magnesium concentrations below the lower S 0.30 reference limit of 0.6 mmol/L. The mean serum zinc concen­ I ...... - 0.20 trations were also significantly lower in patients with cirrho­ sis; 32% of patients and 3% of healthy volunteers had serum 0.10

zinc concentrations below the lower reference limit of 0.00 8.1 pmol/L. Similarly, the mean plasma vitamin A concentra­ tions were significantly reduced in the patient population; 3-Carotene 2.0^ 72% of patients and 1% of healthy volunteers had plasma 1.6 vitamin A concentrations below the lower reference limit of 1.2

1.4 /iinol/L, while 20% of patients but none of the healthy 0.8

volunteers had plasma vitamin A concentrations below 0.35 0.4 pmol/L, a value considered to indicate deficiency.^^ In cir­ 0.0 rhotic patients, the mean plasma a- and ^-carotene concen­ Healthy Cirrhotic trations were also significantly lower than in healthy volun­ volunteers patients (n=75)* (n=75) teers; 16% of patients and 8% of healthy volunteers had plasma a-carotene concentrations below the lower reference Fig. 1. Circulating concentrations of magnesium, zinc, vitamin A, and limit of 0.01 pmol/L. Circulating concentrations of all five a - and ;9-caroiene in healthy volunteers and cirrhotic patients. A. Healthy nutrients were lower in patients with decompensated disease volunteers (n = 75); #, compensated liver disease (n = 51), Pugh's score (Pugh’s score >8)"^; the difference was significant for all but 5-7’'; 0 , decompensated li\er disease (n = 24), Pugh's score 8-15.’' "Data magnesium (Fig. 1). on vitamin A and a - and ;0-carotene from 74 healthy volunteers. Shaded areas represent reference ranges defined as means ± 2 SD of values from In healthy volunteers, there were no relationships between healthy volunteers. The single interrupted horizontal line in the vitamin A circulating magnesium, zinc, vitamin A, or a- and ;6-carotene panel represents the value that defines deficiency.” concentrations and taste thresholds or gustatory or olfactory scores. In cirrhotic patients, significant negative associations were observed between serum magnesium concentrations reported that they enjoyed their food (56 [75%] vs. 72 [96%]; and detection thresholds for salt (P = .02) and overall gusta­ P < .001). tory scores (P = .02). Healthy volunteers who described their sense of taste as Medication. Twenty-five patients (33%) were taking spiro­ “good” were significantly better at detecting salt than those nolactone at the time of the study; these indinduals had a who described their sense of taste as “bad” (1 [range, 1- significantly lower mean serum magnesium concentration 601 vs. 10 [range, 1-401 mmol/L; P = .008). There was no than their untreated counterparts (0.66 ± 0.08 vs. 0.72 ± significant association between subjective and objective as­ 0.08 mmol/L; P < .005), but no significant differences were sessments of taste acuity in cirrhotic patients. observed in median detection and recognition thresholds or Significantly fewer cirrhotic patients reported that they had median gustatory scores between the two groups. No other a “good” sense of smell (50 [67%] vs. 72 [96%]; P < .001). relationships were observed between ingested medication, There was no significant association between subjective and taste thresholds, and circulating micronutrient concentra­ objective assessments of olfactory acuity in either the patient tions. or volunteer groups. Questionnaire Responses. There was no significant differ­ Overall, cirrhotic patients reported significantly greater use ence between healthy volunteers’ and patients’ subjective as­ of vinegar as a savor)' condiment than healthy volunteers (57 sessment of their sense of taste. Significantly more cirrhotic [76%] vs. 45 [60%]; P < .05); there were no other significant patients described their appetites as “bad” (9 [12%] vs. 2 differences in the use of savory condiments between the two [3%]; P < .05), and significantly fewer cirrhotic patients groups. As expected, significantly fewer patients who were 44 MADDEN, BR.ADBLRV. AND MORGAN HriPATOi.oGY July 1997 following sak-restricied diets reported using salt than pa­ cirrhotic patients have been undertaken,"*’® with results that tients who were following unrestricted diets (7 [23%] vs. 38 are qualitatively similar to those in the present study. How­ [84%1; P < .0001), although their use of other condiments ever, these findings are difficult to evaluate and place in did not differ significantly. No overall trends were observed context. In four studies,^"' for example, gustatory acuity was among healthy volunteers or patients between the taste assessed using a dropping technique^ in which a single drop thresholds observed and the reported use of condiments and of tastant solution and two drops of distilled water are placed flavorings. in random order on the tongue and the subject is asked There were few differences in the food selections made by which, if any, of the drops taste different than water. This healthy volunteers and cirrhotic patients presented with two technique may result in underestimation of gustatory acuity food choices, although significantly fewer cirrhotic patients in individuals, for example, with primary biliary cirrhosis chose potato crisps (salt) in preference to pickled gherkins and Sjogren’s syndrome because of inadequate circulation of (sour) (P < .001). Significantly more patients following salt- the test solutions within the oral cavity. In addition, while restricted diets selected grapefruit (sour) in preference to in most of these studies controls were exercised for some peanuts (salt) than patients who were not following salt- factors known to influence gustatory acuity, for example, age restricted diets (77% vs. 42%; P < .005); there were no and sex,"^"® they were not exercised for denture wearing and other significant differences in food choices between the two smoking. patient subgroups. Smith et ah'* evaluated gustatory acuity in 22 patients with No overall trends were obsen ed between taste thresholds acute viral hepatitis and 16 patients with chronic liver dis­ and food choices in either subject group. However, a number ease, 14 of whom had cirrhosis. Comparisons were made of significant differences were observed among subjects se­ with 47 healthy volunteers of comparable age and sex. The lecting sour-tasting foods, e.g., lemon drops in preference to mean detection thresholds for bitter, salt, sweet, and sour, barley sugar. Healthy volunteers who selected lemon drops and the mean recognition thresholds for bitter, salt, and sour had a significantly lower median threshold for recognizing were significantly higher in both patient groups. Burch et sour taste than those who selected barley sugar (0.5 [range, al.^ observed significantly higher mean detection thresholds G.5-5.0] vs. 1 [range, 0.5-30] mmol/L; P = .03); cirrhotic for bitter and salt and a significantly higher mean recognition patients who selected lemon drops had a significantly lower threshold for bitter in eight cirrhotic patients compared with threshold for detecting sweet taste than those who selected 13 healthy volunteers of comparable age. barley sugar (10 [range, 1-50] vs. 20 [range, 1-80] mmol/L; Garrett-Laster et al.® investigated gustatory function in 37 P = .04). cirrhotic patients with vitamin A deficiency and 21 healthy The responses given to open questions regarding favorite volunteers of comparable age and sex. They found signifi­ food and beverage choices also showed few differences be­ cantly impaired gustatory function in the patient population tween the two groups, although significantly more healthy manifest as increased detection thresholds for bitter, salt, volunteers than patients chose non-meat pasta- or rice-based sweet, and sour and increased recognition thresholds for bit­ dishes as their favorite meals (12 [16%1 vs. 2 [3%1; P < .01). ter and salt. Similar findings were reported by Stumiolo et Healthy volunteers had significantly better median well­ al.,^ who investigated gustatory acuity in 60 patients with being scores on the General Health Questionnaires than cirrhosis and 30 healthy volunteers of similar sex distribution cirrhotic patients (0 [range, 0-13] vs. 5 [range, 0-25]; whose ages were not reported. The patients had significantly P < .00001). There were no significant relationships between higher detection thresholds for salt, sweet, and sour. the General Health Questionnaire score and taste thresholds, Deems et al.® assessed chemosensory function in 51 pa­ gustatory scores, or olfactor)' scores in either population. tients with chronic liver disease, of whom an unknown but Patients with decompensated cirrhosis had a significantly worse median well-being score on the General Health Ques­ undoubtedly small number had cirrhosis, and 51 healthy tionnaire and were also significantly less likely to report en­ matched controls. All subjects tasted 12 different food items joyment of food (7 [range, 1-17] vs. 3 [range, 0-25]; P = and rated their pleasantness on a nine-point scale. Overall, .0016 and 11 [46%] vs. 40 [78%]; P < .01). patients rated the tastes of the food items as less pleasant than controls did, significantly so for bitter lemon soda and DISCUSSION plain yogurt. There was also some evidence of taste distortion The results of this study show that overall, patients with in that patients attributed different predominant taste quali­ cirrhosis have considerably impaired gustatory acuity. They ties to the foods sampled; items were reported as bitter signif­ also have impaired olfactory acuity, but the relationship be­ icantly more frequendy and as sweet significantly less fre­ tween the acuity of the two sensory modalities was not signif­ quently. icant. Significantly lower circulating concentrations *df mag­ In the present study, median detection and recognition nesium, zinc, vitamin A, and a- and /5-carotene were thresholds in cirrhotic patients were generally lower'than observed in patients with cirrhosis, and hypomagnesemia mean values in previous studies.'^’® This finding may reflect was associated with impairment of gustatory function. Al­ the use of a rinsing procedure^ rather than a dropping tech­ though many patients reported that their appetites were poor nique,’ although the thresholds observed in our healthy vol­ and that they did not enjoy their food, their subjective assess­ unteers were comparable to values reported previously. ment of their ability to perceive taste was not reliable. Gener­ The effects of age and sex on taste acuity observed in ally, patients’ food selections and expressed taste preferences healthy volunteers were not observed in the patients with were similar to those of their healthy counterparts, sug­ cirrhosis in our study. This may be because the impairment gesting that impairment of gustatory acuity is not necessarily of gustatory function associated with the presence of liver associated with any qualitative change in dietary intake. disease obscured the impairment of gustatory acuity nor­ A small number of other studies of gustatory acuity in mally associated with increasing age and male gender. This 1 \\il. 2n. No. !. I'Hi; VIADDnN. BRADBURY, AND MORGAN 45

finding reinforces the importance of close matching of con­ the electrical potential of the cell membrane by controlling trol and patient groups for studies of this kind. sodium and potassium flux.^® Its effect on gustatoiy activity In previous studies it has been reported that smoking" in cirrhotic patients may be mediated by an effect on the cell and denture wearing affect gustatory function in healthy vol­ membranes of taste buds. In support of this possibility, it unteers,'^ and in the present study controls were exercised has been shown in rats that deprivation of magnesium results for these variables. However, we found no significant effects in a decreased voluntary intake of sodium chloride solution, of either smoking or denture wearing on taste acuity in suggesting an association between magnesium status and healthy volunteers and little if any effect in patients. There­ taste preference.''* In healthy young adults,a reduction fore, it is possible that stratification of subjects by smoking in the ability to discriminate sodium chloride concentrations and denture wearing may not be necessary in future studies, correctly has been shown to correlate with hypomagnesuria. although additional information is needed to support this. Although no conclusion can be drawn about the mechanism, A simple qualitative evaluation of olfactory function was these results, combined with the outcome of the current undertaken in the present study, using easily identifiable study, indicate that there may be a relationship between mag­ oils of lemon, peppermint, and eucalyptus, to exclude major nesium status and gustatory function. impairment of this sensory modality. Among healthy volun­ It could be suggested that the relationship between magne­ teers, women and younger individuals showed superior olfac­ sium status and gustatory function is a consequence of treat­ tory acuity, as previously documented.^^^^ In the subgroup of ment with spironolactone because patients taking this the patients who underwent assessment of olfactory function, diuretic had significantly lower serum magnesium concentra­ significant impairment was observed in younger individuals. tions than their untreated counterparts. Indeed, there is a Impaired olfactory acuity has been reported previously in single report of reversible loss of taste and reduced ability to patients with cirrhosis^ and in patients with viral hepatitis. discriminate salt from sugar in an 82-year-old man treated Burch et al.^ evaluated olfactory acuity in eight patients with with this drug,''’ although no information was given on circu­ cirrhosis and 13 healthy volunteers using a forced-choice, lating magnesium concentrations. However, in the present three-stimulus, sniff technique to assess the olfactory thresh­ study no significant difference was observed between gusta­ olds for pyridine, thiophene, and nitrobenzene.^* They ob­ tory function in patients taking spironolactone and those served significantly higher detection thresholds in cirrhotic who were not. Therefore, it is unlikely that the association patients for all three test substances. Henkin and Smith^^ between serum magnesium concentrations and gustatory observed similar, but reversible, impairment in 19 patients function is a consequence of treatment with this drug. with acute viral hepatitis. In the present study, the mean serum zinc concentration Deems et al.® assessed the olfactory function of 51 patients in the patient group was significantly lower than that in with chronic liver disease, a small percentage of whom had healthy volunteers. This finding is in agreement with previ­ cirrhosis, and 51 healthy matched controls. All siibjecis were ous reports in patients with chronic liver disease.'*’ asked to smell nine different items including grapefruit, dish­ However, assessment of zinc status in humans is difficult^^ washing liquid, coffee, and smoke flavoring and to rate them because circulating concentrations of zinc respond to non- on a nine-point scale for pleasantness. Overall, patients rated nutritional changes. Serum zinc concentration is not there­ the odors as less pleasant; the difference was significant for fore a reliable indicator of zinc status. A reduction in serum coffee and smoke flavoring. concentration is unlikely to be observed until the body’s In the present study, a significant relationship was ob­ capacity for maintaining zinc homeostasis, for example by served between impairment of olfactory and gustatory acuity reducing excretion or slowing growth, has been exhausted. in healthy volunteers but not in cirrhotic patients. This dis­ An association between zinc depletion and impaired gusta­ crepancy has not been reported previously.^^^^^ tory acuity has been observed in animals*^ and in humans In the present study, the mean serum magnesium concen­ in association with a number of disease states.Zinc is tration was significantly lower in the patient population than required for the synthesis and activity of the salivary protein in healthy volunteers. Although serum magnesium, which gustin, which plays a role at taste bud receptor sites.^® How­ constitutes only 1% of total body magnesium, does not al­ ever, no association has been observed between taste thresh­ ways reflect total body magnesium stores,^® it is frequently olds and serum zinc concentrations in healthy volunteers.’’ used to assess magnesium status.” "'^' Comparison of total In the present study, no association was observed between serum magnesium with non-protein-bound (physiologically serum zinc concentrations and taste acuity in either healthy active) serum magnesium indicates that the former is sensi­ volunteers or cirrhotic patients, although the mean serum tive to the detection of magnesium deficiency but overesti­ zinc concentration in the patient population was significantly mates its occurrence.^' Therefore, Rude^ has suggested that reduced; this is in agreement with the findings in other stud­ deficiency be defined as a serum concentration of <0.7 ies in patients with liver disease.However, two double­ mmol/L; in the present study, the lower limit of the reference blind studies'^'” undertaken to observe the effects of long­ range was 0.6 mmol/L, and deficient values were observed term zinc supplementation on gustatory function in patients in 9% of cirrhotic patients. with cirrhosis have reported improvement in gustatory func­ In previous studies, patients with cirrhosis have been re­ tion after supplementation, which was accompanied in one ported as having either low or normal serum magnesium by an increase in serum zinc concentrations.' concentrations.In our patients with cirrhosis, there Plasma vitamin A concentrations were significantly lower was an association between hypomagnesemia and impaired in our cirrhotic patients than in the comparable healthy vol­ gustatory function that has not been reported previously in unteers, confirming the results of previous studies in patients humans. Magnesium plays an important role in many enzyme with chronic liver disease.^*’^' *°'*^ However, circulating con­ systems within the body; it is a cofactor for the membrane centrations of vitamin A only correlate with vitamin A status Na^,K^-adenosine triphosphatase system and thus influences when body reserves become critically depleted or consider- 4(1 MADOHN. HRAPFU RY. AND MORGAN Hi r\ioi oi.i July DW7

ably overfilled." Traditionally, concentrations below 0.7 liver disease, 50% of whom had primar\- biliary cirrhosis, ' jumol/L are taken to indicate low vitamin A status, whereas and 42 matched healthy volunteers. levels below 0.35 /jmol/L may be associated with clinical In the present study, significantly more patients than symptoms and are taken to indicate deficiency." Vitamin A healthy volunteers reported “bad” appetites. However, aver­ concentrations below 0.35 /imol/L were obser\'ed in 20% of sions to meat and fried foods were not apparent. Compari­ our patient population but in none of our healthy volunteers. sons between studies are difficult because of differences in An association has been reported in animals^'^° and in the stmcture of the questionnaires used, the food items se­ humans'*®" '” between gustatory function and circulating lected for consideration, and the nationality of the individu­ concentrations of vitamin A, although results are conflict- als under study. However, our patients with cirrhosis made ij.^g -»7./3./4 relationship between vitamin A and gustator)' few food choices that were different from those of healthy perception has not been fully determined but is thought to volunteers. The exceptions to this were the greater prefer­ be associated with the role of vitamin A in the synthesis of ence, particularly among patients on salt-restricted diets, for nonspecific mucopolysaccharides in the epithelial cells of the gherkins (sour) and grapefruit (sour) over potato crisps taste buds, an accumulation of which may impair the access (salt). Friedman et al.^® also noted a preference for sour taste of taste molecules to the taste bud receptors. in a group of eight patients with liver disease whose intake In the present study, no association was found between of fruit Juice, the bulk of which was citrus based, was signifi­ gustatory function and circulating concentrations of vitamin cantly greater than that of matched controls. A in either of our study populations. Conflicting results have In view of the paucity of data and the conflicting results been reported in other studies in patients with liver disease, on food preferences published to date, further studies are most likely reflecting the variation in vitamin A status in the needed to investigate the association between gustatory func­ subjects studied and the limited value of measuring circulat­ tion, food selection, and nutrient intake in patients with ing concentrations of this vitamin to assess vitamin A status; cirrhosis. If impairment of gustatory function is not associ­ thus a positive association between gustatory acuity and ated with alteration in food selection, as in the present study, plasma vitamin A concentration has been reported in patients it may be that any adverse affects on nutrient intake and with cirrhosis® and obstructive jaundice,^^ but no such asso­ thus on nutritional status are mediated through quantitative ciation was observed in patients with acute viral hepatitis rather than qualitative changes in diet. Indeed, it has been or chronic liver disease."* Nevertheless, Garrett-Laster et al.® shown that patients with dysgeusia have reduced energy in­ showed that supplementation with vitamin A resulted in both take'^ and also that nutritional status in elderly patients im­ an increase in serum vitamin A concentrations and an im­ proves in response to flavor enhancement of a “normal” hos­ provement in gustatory function in vitamin A-depleted pa­ pital diet.®® tients with cirrhosis. However, even if impairment of gustatory function does A number of other factors may affect gustatory acuity in not influence nutrient intake, it may affect two other aspects patients with cirrhosis. Stumiolo et al.^ concluded that the of patient care. First, approximately one third of our patients taste impairment observed in their patients with liver disease were following sodium-restricted diets, and impairment of was a feature of hepatic disease per se and not of zinc defi­ taste acuity may compromise their compliance and hence ciency. It has also been suggested that the presence of volatile impede resolution of fluid retention.®^ Second, general well­ fatty acids,'® mercaptans,^^ and hydrogen sulphide”® in the being and quality of life may be adversely affected by a reduc­ breath of patients with decompensated liver disease might tion in perceived pleasantness of food secondary to impair­ affect taste bud receptors or chemically interact with certain ment of gustatory function. Indeed, in the present study a tasiants. However, in the present study there was no signifi­ negative association was found between General Health cant relationship between taste thresholds and the degree of Questionnaire scores and patients’ reported enjoNinent of hepatic decompensation. . food. Therefore, if the impairment of gustatory acuity proves There are few published studies detailing food preferences amenable to treatment with appropriate micronutrient sup­ and eating behavior in patients with liver disease that include plementation, or if the effects of taste impairment can be details of food selection rather than nutrient intake alone.® ameliorated by suitable advice on flavor enhancement, then It is difficult to evaluate the impact of alterations in gustator)' the possible detrimental effects on nutrient intake, dietary function on nutrient intake without knowing the extent to compliance, and quality of life may be alleviated. which taste changes influence food selection. Henkin and Smith”^ investigated 19 patients with acute hepatitis and re­ Acknowledgment: The authors thank Dr. Richard Morris, ported that 90% had an aversion to meat anti 74% had an Department of Primary Care and Population Sciences, The aversion to fried foods. Deems et al.® used a dietary behavior Royal Free Hospital School of Medicine, for statistical advice; questionnaire to assess food and taste preferences in 88 pa­ Rosemary Mallalieu for practical assistance; the Pharmacy tients with liver disease, 13 (15%) of whom had cirrhosis, Department at the Royal Free Hospital for preparation of the and in a well-matched control group of healthy individuals. concentrated testing solutions and for providing facilities for They reported that the patients gained less pleasure from the preparation of the test solutions; the Department of Clini­ eating meat, fried and predominantly bitter-tasting foods cal Chemistry at the City Hospital Birmingham for their pro­ than their healthy counterparts. Patients with cirrhosis re­ ficient service in the analysis of plasma vitamin A an(l a- ported a lower preference for fried chicken, fried fish sand­ and j^-carotene concentrations; Professor Archie Young, The wiches, tuna, spinach, mustard, and bittersweet chocolate University Department of Geriatric Medicine, The Royal Free than healthy volunteers. However, in a second study from Hospital School of Medicine, for allowing us to approach the same group,no differences were found in the frequency volunteers on his register; and all the cirrhotic patients and of intake of different food groups between 42 patients with healthy volunteers for their participation. iwiottKA \'ol. 26. No. 1, ly y ] MADDEN. BRADPE'RY. AND MORGAN 47

REFERENCES w ith sodium l-azo-2-hydroxy-3-(2,4-dim ethylcarboxanilido)-naphtha- lene-l'-(2-hydroxybenzene-5-sulfonate). Anal Chem 1956:28:202-205. 1. DiCccco SR, W ieners EJ. W iesner RH. Southom PA, Plevak Dj, Krom 30. Butrim ovitz GP, Purdy WC. The determ ination of zinc in blood plasm a R.AF. A ssessm ent ol nutritional status of patients w ith end-stage liver by atom ic absorption spectrom etry. Anal C him Acta 1977;94:63-73. disease undergoing liver transplantation. M ayo Clin Proc 1989;64:95- 31. Thurnham D l, Sm ith E, Flora PS. C oncurrent liquid-chrom atographic 102. assay of retinol, a-tocopherol, /9-carotene, a-carotene, lycopcnc and P- 2. Basse JM . N utrition considerations in liver transplantation. T op C lin cryptoxanthin in plasm a, w ith tocopherol acetate as internal standard. Nutr 1992;7:24-33. Clin Chem 1988;34:377-381. 3. Italian M ulticentre C ooperative Project. N utritional status in cirrhosis. 32. G oldberg D. The G eneral H ealth Q uestionnaire (G H Q 28). Institute of J Hepatol 1994;21:317-325. Psychiatry. W indsor: NFER-NELSO N , 1981. 4. Sm ith FR, H enkin Rl, D ell RB. Disordered gustatory acuity in liver 33. Underwood BA. M ethods of assessm ent of vitam in A status. J Nutr disease. G astroenterology 1976;70:568-571. 1990; 120S: 1459-1463. 5. Burch RE, Sackin DA, llrsick JA. Jetton M M , Sullivan JF. D ecreased 34. Chauhan J, Hawrysh ZJ, G ee M , Donald EA, Basu TK. Age related taste and sm ell acuity in cirrhosis. Arch Intern M ed 1978; 138:743-746. olfactory and taste changes and interrelationships betw een taste and 6. Garrett-Laster M , Russell RM , Jacques PF. Im pairm ent of taste and nutrition. J Am D iet A ssoc 1987;87:1543-1550. olfaction in patients w ith cirrhosis: the role of vitam in A. H um N utr 35. Ship JA, W eiffenbach JM . Age, gender, m edical treatm ent and m edica­ Clin Nutr 1984;38C :203-214. tion effects on sm ell identification. J G erontol 1993;48:M 26-M 32. 7. Stum iolo G C , D ’lnca R, Parisi G , G iocom azzi F, M onti no M C , D’O do- 36. H enkin Rl, Bartter EJ. Studies of olfactory thresholds in norm al m an rico A, Soranzo P, et al. Taste alterations in liver cirrhosis: are they and in patients w ith adrenal cortical insufficiency. T he role of adrenal related to zinc deficiency? J Trace Elem Electrolytes H ealth D is 1992; con ical steroids and of serum sodiu m concentration. J C lin Invest 1966; 6 : 1 5 - 1 9 . 45:1631-1639. 8. D eem s RO, Friedm an M l, Friedm an LS, M unoz SJ, M addrey W C. C he­ 37. H enkin Rl, Schechter PJ, H oye R, M attem CFT. Idiopathic hypogeusia m osensory function, food preferences and appetite in hum an liver dis­ w ith dysgeusia, hyposm ia, and dysosm ia. A new syndrom e. JAM A 1971; ease. A ppetite 1993;20:209-216. 217:434-440. 9. H enkin R l, G ill JR, Banter FC. Studies on taste thresholds in norm al 38. M cLean RM. M agnesium and its therapeutic uses: a review. Am J M ed m an and in patients w ith adrenal cortical insufficiency: the role of 1994;96:63-76. adrenal cortical steroids and of scrum sodium concentration. J C lin 39. Reinhart RA. M agnesium m etabolism . A review w ith special reference Invest 1963;42:727-735. to the relationship betw een intracellular content and serum levels. Arch 10. A larcdn-Segovia D, D iaz-Jouanen E, Fishbein E. Features of Sjogren’s Intern M ed 1988;148:2415-2420. syndrom e in prim ary biliary cirrhosis. Ann Intern M ed 1973;79:31-36. 40. Rude RK. M agnesium m etabolism and deficiency. Endocrinol M etab 11. K aplan A R , G Ian ville EV, Fischer R. C um ulative effect of age and sm ok ­ Clin North Am 1993;22:377-395. ing on taste sensitivity in m ales and fem ales. J G erontol 1965;20:334- 41. Zaloga GP, W ilkens R, Tourville J, W ood D, Klyme DM . A sim ple 3 3 7 . m ethod for determ ining physiologically active calcium and m agnesium 12. Lelicvre G , Le Floch JP, Perlem uter L, Peynegre R. Le gout chez le concentrations in critically ill patients. Crit Care M ed 1987; 15:813- sujet sain. Influence de la consom m ation d’alcool et de tabac. Ann 8 1 6 . O tolaryngol C hir C ervicofac 1989;106:541-546. 42. W allach S, Cahill LN, Rogan FH, Jones HC. Plasm a and erythrocyte 13. H enkin R l, C hristiansen RL. Taste thresholds in patients w ith dentures. m agnesium in health and disease. J Lab C lin M ed 1962;59:195-210. J Am Dent A ssoc 1967;75:118-120. 43. Rosner F, Gorfien PC. Erythrocyte and plasm a zinc and m agnesium 14. Russell RM , C ox M E, Solom ons N . Zinc and the special senses. Ann levels in health and disease. J Lab C lin M ed 1968;72:213-219. Intern M ed 1983;99:227-239. 44. Cohen M l, M cNamara H, Finberg L. Serum m agnesium in children 15. Sullivan JF, Lankford HG. Zinc m etabolism and chronic alcoholism . w ith cirrhosis. J Pcdiatr 1970;76:453-455. Am J Clin N utr 1965;17:57-63. 45. R occhi E, Borella P, Borghi A , Paolillo F, Pradelli M , Farina F, Casal- 16. H alstedJA , H ackley B, Rudzki C, Sm ith JC Jr. Plasm a zinc concentra­ grandi G . Zinc and m agnesium in liver cirrhosis. Eur J C lin Invest 1994; tions in liver diseases. C om parison w ith norm al controls and certain 24:149-155. other chronic diseases. G astroenterology 1968;54:1098-1105. 46. Tordoff M G, Salt intake in rats fed diets deficient in calcium , iron, 17. Sullivan JF, H eaney RP. Zinc m etabolism in alcoholic liver disease. A m m agnesium , phosphorus, potassium or all m inerals. A ppetite 1992; 18: J Clin N utr 1970;23:170-177. 2 9 - 4 1 . 18. W eism ann K, C hristensen E, Dreyer V. Zinc supplem entation in alco­ 47. Suzuki T, Ishida H, Suzuki H, K ongo T, Kobayashi K, Yoshinaga J, holic cirrhosis. A double-blind clinical trial. A cta M ed Scand 1979; 205: Igaw a S. M inerals in plasm a and urine and gustatory function tests for 3 6 1 - 3 6 6 . salt in a group of healthy young m en. J N utr Sci V itam inol 1987;33: 19. Scholm erich J, Langenberger G , Steiert B, Lôhle E, W okalet H , Tauber 2 8 9 - 2 9 7 . R, K ottgen E, G erok W . Long term zinc substitution in patients w ith 48. Suzuki T, Ishida H, Suzuki H , K ongo T, K obayashi K, O hba T, Shidoji zinc deficiency in liver cirrhosis— results of a placebo controlled trial Y. M inerals and vitam in-A status in relation to gustatory functions for [A bstract]. G astroenterology 1986:90:1765. salt in healthy fem ale college students. J N utr Sci V itam inol 1988; 34: 20. Leevy CM , T hom pson A, Baker H. Vitam ins and liver injury. A m J C lin 2 0 9 - 2 2 2 . Nutr 1970;23:493-498. 49. C lee M D, Burrow L. Taste and sm ell in disease [Letter]. N Engl J M ed 21. Russell RM , M orrison SA, Sm ith FR, Oaks EV, Carney EA. V itam in-A 1983; 309:1062. reversal of abnorm al dark adaptation in cirrhosis. Study of effects on 50. Vallee BL, W acker W EC, Bartholom ay AF, H och FL. Zinc m etabolism the plasm a retinol transport system . Ann Intern M ed 1978;88:622-626. in hepatic dysfunction. Ann Intern M ed 1959;50:1077-1091. 22. Stuizm an FL, A raatuzio DS. Blood serum m agnesium in portal cirrhosis 51. G rungreiff K, Abichi K, K luge M , Presser HJ, Frankc D, Kleine FD, and diabetes m ellitus. J Lab Clin M ed 1953;41:215-219. K lauck S, et al. C linical studies on zinc in chronic liver diseases. Z 23. Lim P, Jacobs E. M agnesium deficiency in liver cirrhosis. Q J M ed 1972; G astroenterol 1988;26:409-415. 41:291-300. 52. K ingJC . A ssessm ent of zinc status. J N utr 1990;120S;1474-1479. 24. M attes RD. Sensory influences on food intake and utilization in hum ans. 53. M cConnell SD, H enkin Rl. Altered preference for sodium chloride, H um Nutr Appl Nutr 1987;41A:77-95. anorexia, ancf changes in plasm a and urinary zinc in rats fed a zinc- 25. H enkin R l, Sm ith FR. Hj’posm ia in acute \iral hepatitis. Lancet 1971; deficient diet. J N utr 1974;104:1108-1114. 1 : 8 2 3 - 8 2 6 . 54. Prasad AS, H aktedJA , N adim i M . Syndrom e of iron deficiency anem ia, 26. Friedm an M l, Preti G, D eem s RO, Friedman LS, M unoz SJ, M addrey hepatosplenom cgaly, hypogonadism , dwarfism and geophagia. Am J W C . L im onene in expired lung air of patients with liver disease. D ig M ed 1961;31:532-546. D is Sci 1994;39:1672-1676. 55. H enkin Rl, Patten BM , Re PK, Bronzert DA. A syndrom e of acute zinc 27. Pugh RNH , M urray-Lyon IM , Dawson JL, Pietroni M C, W illiam s R. loss. Cerebellar dysfunction, m ental changes, anorexia and taste and Transection of the oesophagus for bleeding oesophageal varies. Br J sm ell dysfunction. A rch N eurol 1975;32:745-751. Surg 1973;60:646-649, 56. M ahajan SK, Prasad AS, Lam bujon J, Abbasi AA, Briggs W A, M cDonald 28. Barale K, A ker SN . M artinsen CS. Prim ary taste thresholds in children FD . Im provem ent of urem ic hypogeusia by zinc: a double-blind study. w ith leukaem ia undergoing m arrow transplantation. JPEN J Patenter A m J Clin Nutr 1980;33:1517-1521. Enteral N utr 1982;6:287-290. 57. H am bidge KM , Silverm an A. Pica w ith rapid im provem ent after dietary 29. M ann CK, Y oe JH. Spectrophotom etric determ ination of m agnesium zinc supplem entation. Arch D is C hild 1973;48:567-568. 4 8 m a ;)L'l :-. i;; M^LR'i. A' HirA’oiotA J liIv 1*^47

58. H enkin Rl. D rug-induced taste and sm ell disorders. Drug Safetv 1944; 71. Saubcrlich HE. Vitam in m etabolism and requirem ents. Som e aspects 11:318-377. reviewed. S Afr M ed J 1975;49:2235-2244. 59. Bales CW . Steinm an LC, Freeland-G ravesJH , Stone JM.Young RK. The 72. Ishida H. Takahasi H. Suzuki H. K ongo T. Suzuki T. Shidoji Y. Yoon effect of age on plasm a zinc uptake and taste acuity. Am j C lin Nutr KH. Interrelationship of som e selected nutritional param eters relevant 1986:44:664-669. to taste for salt in a group o f college-aged w om en. J N utr Sci V itam inol 60. Sm ith JC. Brown ED. W hite SC, Finkelstein JD. Plasm a \itam in A and 1985;31:585-598. zinc concentrations in patients w ith alcoholic cirrhosis. Lancet 1975: 73. C asper RC. K irschner B. Sandstead H H , Jacob R.A. D avis JM . An evalua­ 1:1251-1252. tion of trace m etals, vitam ins and taste function in anorexia nenosa. 61. C arney EA. R ussell RM . C orrelation of dark adaptation test results w ith Am J Clin Nutr 1980;33:1801-1808. serum vitam in A levels in diseased adults, j N utr 1980:110:552-557. 74. Palin D, U nderw ood BA. D enning CR. The effect of oral zinc supplem en­ 62. H erlong HF. Russell RM , M addrey W C. Vitam in A and zinc therapy in tation on plasm a levels of vitam in A and retinol-binding protein in prim ary biliar\- cirrhosis. Hepatology 1981;1:348-351. cystic fibrosis. A m J C lin N utr 1979;32:1253-1259. 63. M unoz SJ. H eubi JE, Balistreri W F, M addrey W C. V itam in E deficiency 75. Im am ine T, O kuno M , M oriwaki H, N inom iya M. Nishiwaki S, Nom a in prim ary biliary cirrhosis: gastrointestinal m alabsorption, frequency A , M uto Y. Plasm a retinol transport system and taste acuity in patients and relationship to other lipid-soluble vitam ins, Hepatology 1989:9: w ith obstructive jaundice. G astroenterol Jpn 1990;25:206-211. 525-531. 76. C hen S, M ahadevan V, Zieve L. V olatile fatty acids in the breath 64. W ard RJ, Peters TJ. T he antioxidant status of patients w ith either alco­ of patients w ith cirrhosis o f the liver. J Lab C lin M ed 1970; 75:622- hol-induced liver dam age or m yopathy. A lcohol A lcohol 1992; 27:359- 6 2 7 . . 3 6 5 . 77. C hen S, Zieve L, M ahadevan V. M ercaptans and dim ethyl sulfide in the 65. Patek AJ. H aig C. T he occurrence of abnorm al dark adaptation and its breath of patients w ith cirrhosis of the liver. Effect of feeding m ethio­ relation to vitam in A m etabolism in patients w ith cirrhosis of the liver. nine. J Lab C lin M ed 1970;75:628-635. J Clin Invest 1939;18:609-616. 78. D eem s RO, Friedm an LS, Friedm an M l, M unoz SJ, D eem s DA, M addrey 66. Solom ons NW , Russell RM. The interaction of vitam in A and zinc: im plications for hum an nutrition. A m J Clin N utr 1980;33:2031-2040. W C . R elationship betw een biochem ical tests and dietary intake in pa­ 67. Scholm erich J, Lôhle E, K ottgen E, G erok W . Zinc and vitam in A defi­ tients w ith liver disease. J C lin G astroenterol 1994; 18:304-308 ciency in liver cirrhosis. H epatogastroenterology 1983;30:119-125. 79. M attes-K ulig DA, H enkin Rl. Energy and nutrient consum ption of pa­ 68. Bernard R .\, H alpem BP, Kare M R. Effect of vitam in A deficiency on tients w ith dysgeusia. J A m Diet A ssoc 1985;85:822-826. taste. Proc Soc Exp Biol M ed 1961;108:784-786. 80. Schiffm an SS, W arw ick ZS. Effect of flavour enhancem ent of foods for 69. Bernard RA, H alpem BP, Kare M R. The reversible effect of vitam in A the elderly and nutritional status: food intake, biochem ical indices and deficiency on taste [A bstract]. Fed Proc 1962;21:362. anthropom etric m easures. Physiol Behav 1993:53:395-402. 70. Bernard RA, H alpem BP. Taste changes in vitam in A deficiency. J Gen 81. A rroyo V. G ines S, Salo J. A European survey on the treatm ent of ascites Phvsiol 1968;52:444-464. in cirrhosis. J H epatol 1994;21:667-672.