Body weight and composition endpoints in cancer cachexia clinical trials: Systematic Review 4 of the cachexia endpoints series

Abstract Significant variation exists in the outcomes used in cancer cachexia trials, including measures of body composition, which are often selected as primary or secondary endpoints. To date, there has been no review of the most commonly selected measures or their potential sensitivity to detect changes resulting from the interventions being examined. The aim of this systematic review is to assess the frequency and diversity of body composition measures that have been used in cancer cachexia trials. MEDLINE, Embase and Cochrane Library databases were systematically searched between January 1990 and June 2021. Eligible trials examined adults (≥18 years) who had received an intervention aiming to treat or attenuate the effects of cancer cachexia for >14 days. Trials were also of a prospective controlled design and included body weight or at least one anthropometric, bioelectrical or radiological endpoint pertaining to body composition, irrespective of the modality of intervention (e.g., pharmacological, nutritional, physical exercise and behavioural) or comparator. Trials with a sample size of <40 patients were excluded. Data extraction used Covidence software, and reporting followed the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidance. This review was prospectively registered (PROSPERO: CRD42022276710). A total of 84 clinical trials, comprising 13 016 patients, were eligible for inclusion. Non‐small‐cell lung cancer and pancreatic cancer were studied most frequently. The majority of trial interventions were pharmacological (52%) or nutritional (34%) in nature. The most frequently reported endpoints were assessments of body weight (68 trials, n = 11 561) followed by bioimpedance analysis (BIA)‐based estimates (23 trials, n = 3140). Sixteen trials (n = 3052) included dual‐energy X‐ray absorptiometry (DEXA)‐based endpoints, and computed tomography (CT) body composition was included in eight trials (n = 841). Discrepancies were evident when comparing the efficacy of interventions using BIA‐based estimates of lean tissue mass against radiological assessment modalities. Body weight, BIA and DEXA‐based endpoints have been most frequently used in cancer cachexia trials. Although the optimal endpoints cannot be determined from this review, body weight, alongside measurements from radiological body composition analysis, would seem appropriate. The choice of radiological modality is likely to be dependent on the trial setting, population and intervention in question. CT and magnetic resonance imaging, which have the ability to accurately discriminate tissue types, are likely to be more sensitive and provide greater detail. Endpoints are of particular importance when aligned with the intervention's mechanism of action and/or intended patient benefit.


Introduction
Cancer cachexia is a complex multifactorial syndrome characterized by loss of muscle and body fat. 1 These changes are strongly associated with poorer quality of life, increased morbidity and worse survival. 2The 2011 consensus definition for cancer cachexia provided diagnostic and staging criteria that have been instrumental in aiding cachexia trial design. 1 At present, there is no similar consensus regarding endpoints, and significant variations remain amongst the clinical assessments used in cancer cachexia trials.
A comprehensive patient assessment of cachexia would consider changes in body composition, dietary intake, biomarkers of the pathophysiological drivers of cachexia, physical function and quality of life, and the influence on associated oncological outcomes.Depending on the mechanism of a given clinical trial intervention, particular weighting may be assigned to chosen measures within this broad range.Selected endpoints must be both sensitive enough to detect change and specific enough not to be readily influenced by other conditions or treatments.Furthermore, it is imperative that they convey clinical relevance.
Endpoints pertaining to body weight and composition are amongst the most frequently reported in cachexia trials and will be the focus of this review.Anthropometric measurements and electric bioimpedance analysis (BIA) are simple modalities that, although inexpensive and non-invasive, are prone to confounders and provide finite levels of detail.Dual-energy X-ray absorptiometry (DEXA) is widely available and can provide estimates of regional/whole-body fat or lean tissue mass.However, DEXA is unable to discriminate between different types of 'lean tissue' (e.g., skeletal muscle vs. organs) or anatomical locations (e.g., visceral vs. subcutaneous adipose tissue). 3While cachexia research has traditionally focused on the loss of muscle, it is now known that adipose tissue also plays an important role in cachexia pathophysiology, and different mechanisms underpin the loss of each tissue type. 4As such, the ability of modalities to distinguish between body tissue compartments is of increasing relevance.Computed tomography (CT) and magnetic reso-nance imaging (MRI) scans are considered the 'gold-standard' assessment modalities for body composition owing to their specificity in discriminating tissue identities and their precision. 5Comparison of the two has shown high levels of agreement in assessments of muscle quantity and quality 6 ; however, CT has been more frequently utilized in cachexia research owing to its more widespread use in routine clinical practice.
At present, it is not known what the best endpoints for cancer cachexia trials are.This may have resulted in sub-optimal clinical trial design, which could have in turn hindered the development of effective therapies.An appraisal of the endpoints currently used would seem like a logical starting point.The aim of this systematic review is to assess the frequency and diversity of measures that have been used to assess body weight and body composition in cancer cachexia trials.

Methods
This systematic review is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 7The review protocol was prospectively registered at the International Prospective Register of Systematic Reviews: PROSPERO (CRD42022276710). 8his review will address assessments of body weight, alongside anthropometric, bioelectrical or radiological endpoints pertaining to body composition.It is one of a series of six that will comprehensively evaluate the endpoints examined in cancer cachexia trials.Given the breadth of outcome measures in the literature, these were categorized broadly under the following domains: physical function, 9 quality of life, appetite and dietary intake, body weight and composition, oncological outcomes and biomarkers.

Search strategy
A systematic search of MEDLINE (Ovid), Embase (Ovid) and Cochrane Central Register of Controlled Trials databases was conducted by a senior research librarian (University of Oslo).All published studies from 1 January 1990 to 2 June 2021 were eligible.Search results were synthesized and managed using the web-based systematic review software 'Covidence' (Veritas Health Innovations, Melbourne, Australia), and duplicates were removed.A detailed search strategy is outlined in Appendix A.

Study eligibility criteria
Prospective clinical trials that considered an intervention aiming to treat or attenuate the effects of cachexia in adult patients (≥18 years) with cancer were considered for eligibility.Inclusion was irrespective of the site of primary malignancy, modality of intervention (e.g., pharmacological, nutritional and physical exercise) or choice of comparator.Articles were excluded if they studied fewer than 40 patients and/or if the intervention lasted <14 days.Studies in which patients underwent surgery during the assessment period were excluded.All included full-text articles were written in the English language.

Data selection and extraction
The titles and abstracts of the identified studies were independently reviewed by three authors (OD, TSS and BJAL).Those selected were subsequently subject to full-text review (LRB and MSS).In instances of discrepancies between reviewers regarding an article's inclusion, consensus was reached through consultation between reviewers or with the wider authorship group.A pre-defined data extraction table was developed and pilot-tested before relevant data points were extracted independently by the lead authors (LRB and MSS).

Relevant outcome measures
Endpoints considered by this review were those pertaining to assessments of body weight and other modalities that aim to assess changes in body composition.These shall be categorized as anthropometric (e.g., body weight, circumference or skinfold measurements), bioelectrical (e.g., BIA) or radiological (e.g., DEXA, CT or MRI) measure of body composition.

Assessment of methodology and risk of bias
The methodological quality of each study was independently assessed by four reviewers (JS, JM, OD and BJAL) using the modified Downs and Black checklist. 10This tool assesses several criteria including study design, internal and external validity, and reporting standards.

Data analysis
Study characteristics, patient details and disease demographics were reported descriptively.The aim of this review was to describe the body weight and composition outcomes used, rather than estimate treatment effects.As such, quantitative meta-analysis was not performed.Furthermore, the heterogenous nature of the trials and interventions studied made meta-analysis of treatment effects on each endpoint impractical.Analyses and visualization were conducted using RStudio Version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria) with packages including maps and tidyverse.

Results
Overall, 8166 studies were identified following systematic searches of MEDLINE (Ovid), Embase (Ovid) and Cochrane Central Register of Controlled Trials databases (Appendix A).Following the removal of duplicates (n = 2191), further screening of the title and abstract for 5975 studies was performed.Of these, 5606 articles were excluded and 369 were retrieved for full-text review.Following detailed screening against the chosen inclusion and exclusion criteria, 84 clinical trials were eligible for inclusion.The PRISMA flow chart is detailed in Figure 1.

Study characteristics
Between 1990 and 2021, a total of 84 prospective clinical trials (n = 13 016 participants) included body weight or measure(s) of body composition as an endpoint.While numerous primary tumour sites were considered, pancreatic cancer (n = 11 trials) and non-small-cell lung cancer (n = 10 trials) were the most frequently studied.Cohorts ranged in size, with the largest cohort being 979 patients studied by the ROMANA 1 and 2 trials. 11Pharmacological interventions (n = 43 trials) were most evaluated, followed by nutritional (n = 28 trials), multi-modal (n = 9 trials) and exercise-based modalities (n = 4 trials).The key characteristics of the included trials are detailed in Table 1.
Figure 2 depicts the geographical distribution of the included cancer cachexia clinical trials.For multicentre or multinational trials (n = 12 studies), coordinates for the institution of the corresponding author were used.It was noted that limited research has been conducted in Eastern Europe, Africa or South America.

Temporal trends in body composition endpoint selection
The relative use of body weight and body composition assessments over time is depicted in Figure 3.The proportion of trials that included assessments of body weight or body mass index (BMI) as an endpoint measure did not vary particularly over the time frame considered.Other anthropometric measures (e.g., skinfold thickness or arm circumference) have been less utilized in recent years with only three trials in the last decade reporting these endpoints.BIA has been used with relative consistency during the last 20 years, whereas DEXA-based estimates of body composition were included in only two trials before 2010, when its use increased.The reporting of CT body composition analysis in clinical trials is more contemporary with only eight trials, all conducted within the last decade, having included this assessment modality.

Body weight and other anthropometric endpoints
Seventy-five trials (n = 12 056 participants) measured body weight or other anthropometric endpoints pertaining to body composition (Table 1 and Appendix B).Assessments of body weight (68 trials, n = 11 561 participants) were utilized in pharmacological (n = 37), nutritional (n = 22), multi-modal (n = 8) and exercise-based (n = 1) clinical trials.A body weight assessment was selected as the (co-)primary endpoint in 32 (47.1%) of these trials and was a secondary/exploratory outcome for the other 36 (52.9%).Analyses were based on absolute change in body weight in 48 trials (70.6%) and percentage change from baseline in 16 trials (23.5%) with 4 trials considering both (5.9%).The ACT-ONE trial analysed the rate (slope) of absolute and percentage weight change. 62ody weight was handled as an ordinal variable by three trials 19,32,54 where comparison was drawn between proportions of weight-gaining, weight-stable and weight-losing participants.Over one third of the trials that considered body weight (38.2%, n = 4735) noted significant differences between trial groups (Figure 4 and Table 2).BMI was reported as an endpoint for 13 trials, most commonly in addition to body weight (6/13 trials).The majority of studies that chose BMI as an endpoint employed a nutritional intervention (61.5%).It was the primary endpoint in only one of these trials (7.7%).Fourteen trials included anthropometric measures of the arm as endpoints (n = 1901 participants).Eight of these evaluated pharmacological interventions (57.1%), five were nutritional (35.7%) and one was multi-modal (7.1%).Measurements of mid-arm circumference, and other derived upper arm measures such as muscle and fat areas, were selected as endpoints in 12 trials (n = 1324, Appendix B).Skinfold thickness was also used commonly as an endpoint       (11 trials, n = 1727 participants).All of these 11 trials measured the triceps skinfold, and some also considered skinfold thickness at the biceps, subscapular skinfold and other sites.Triceps skinfold thickness was the primary endpoint for one trial, where arm anthropometry was used as a secondary/exploratory outcome for all other trials.All studies that included arm-based anthropometric measurements also included body weight as an endpoint.Of these, 9 (64.3%)identified no statistically significant difference between groups using any selected outcome measure.Two trials 17,42 identified statistically significant improvements in triceps skinfold thickness but no corresponding change in body weight.Conversely, McMillan et al.'s trial led to in-creased body weight (5.1 kg median difference between trial arms) and mid-arm circumference measurement (1 cm median difference between trial arms), but no change in skinfold thickness measurements. 28

Bioelectrical body composition endpoints
Endpoints based on assessment with BIA were used in 23 trials (n = 3140 participants, Table 1 and Appendix C).The interventions tested were commonly nutritional (11 trials, 47.8%) or pharmacological (9 trials, 39.1%).BIA-based endpoints were selected as a (co-)primary endpoint in 6 trials (26.1%)Fourteen studies (60.9%) that used BIA did not detect statistically significant differences between trial groups with any of their selected endpoints (including those not BIA-based) (Table 2 and Figure 5).In two trials, improvements were identified in BIA estimates of LBM/FFM that were congruent with increased body weight. 75,80A statistically significant increase in body weight was identified in two trials of nutritional counselling 35,54 but there were no accompanying changes in BIA estimates of LBM.Two of the seven trials (28.6%) that estimated FM using BIA demonstrated a statistically significant increase with their intervention that was congruent with body weight gain. 46,63

Radiological body composition endpoints
DEXA was used in 16 trials (n = 3052) with LBM (n = 2957 participants) and FM (n = 2162) being the most frequently reported endpoints (Table 1 and Appendix C).Appendicular lean mass was used as an alternative endpoint in two trials 81,87 and alongside whole-body LBM in another two trials 58,65 (Figure 6).Pharmacological interventions were used for most trials that used DEXA (68.8%).DEXA-based measures were used as the primary endpoint for 50% (n = 8) of these trials.
Five trials included in this review compared the effects of anamorelin against placebo (n = 1929 participants).Of these, four evaluated LBM using DEXA, and all identified a statistically significant increase compared to placebo with congruent increases in overall body weight. 11,58,63,73Of note, Takayama ). 63nly eight trials included endpoints based on CT body composition (n = 841, Table 1 and Appendix C), with two of these (25%) considering it a (co-)primary endpoint.All but one 43 of these had relatively small sample sizes (≤125 patients).Four measured the cross-sectional area of skeletal muscle (n = 346) at the third (L3) 68,69,94 or fourth/fifth lumbar vertebral level. 72One of these studies reported the L3 cross-sectional area of muscle as normalized for height, termed skeletal muscle index (SMI). 68Others included the L3 cross-sectional area of psoas major 78 or derived estimates of LBM (kg) based on L3 muscularity. 43,48CT estimates of adipose tissue were also reported by two trials. 72,81Six of the included trials noted no significant differences for any of their selected endpoints.
The five-arm phase III randomized controlled trial (RCT) (n = 332) by Mantovani et al. 43 identified improved LBM using DEXA (mean difference: 2.1 kg) and CT estimates (mean difference: 2.6 kg) in one of the trial arms but detected no difference using BIA (mean difference: 1.2 kg, P = 0.609).Similarly, Madeddu et al. 48found improvements in LBM across both trial arms based on DEXA and CT estimates, but not with BIA.

Discussion
This systematic review summarizes the frequency and diversity of endpoints examining body weight and composition in cancer cachexia clinical trials.It is one of six systematic reviews being undertaken, with others considering physical function, 9 quality of life, appetite and dietary intake, biomarkers and oncology/survival endpoints.Assessments of body weight were the most commonly reported endpoint, used by over 80% of the included trials.Other anthropometric measures, such as skinfold thickness measurements and arm circumference, were less frequently used, especially in more contemporary trials.BIA-based estimates were often included but have been largely superseded by DEXA, especially in larger trials, and more recently by CT body composition analyses.
Body weight is the simplest and most widely available assessment that can indicate alterations to body composition and has long been regarded as a central tenet of cachexia.Included trials have used it to create various endpoints (e.g., absolute/percentage differences from baseline or comparison between weight-stable, weight-gaining and weight-losing groups).Further study is likely required to establish consen-sus on which of these specific body weight endpoints is most informative.While BMI is likely a helpful baseline descriptor, given the relevance of obesity as a prognostic variable in patients with cachexia and/or sarcopenia, [95][96][97] the use of BMI as  an endpoint instead of, or alongside, body weight does not add value.Other simple anthropometric measures, such as triceps skinfold thickness or other arm anthropometry, were featured in earlier cachexia trials but have been used less in recent years.All of these methods pose practical advantages, such as low cost, widespread availability and ease of measurement, but provide limited information when compared with radiological methods.Estimates of body composition using BIA have been featured as chosen endpoints for a number of the included trials.BIA measures the electrical properties of tissues (resistance and reactance), and these values can be used in equations to approximate FM/FFM/LBM.Previous studies have shown that BIA is prone to overestimation for lean mass and underestimation of FM 98 and has poor agreement with comparable DEXA 99 and CT-based measures. 100Indeed, some of the included studies identified significant differences between trial groups on DEXA and/or CT estimates of body composition that were not evident using BIA-based estimates. 43,48,63However, it must be noted that BIA does have several practical advantages.It uses equipment that is portable and non-invasive, making it perhaps a more attractive option for assessments in frail, incurable cachexia populations or in community-based studies where assessments may take place in non-clinical locations.As such, BIA may remain an appropriate option for particular trial settings despite its limitations.
DEXA has been widely used for assessment of lean and adipose tissue in cachexia trials to date, including amongst many of the larger studies included in this review. 11,36,43It does not routinely feature during the clinical staging pathway for patients with cancer but is an attractive research imaging adjunct owing to its modest cost, low radiation dose and short imaging time.While DEXA estimates of LBM have shown excellent correlation with both MRI and CT, 101 there is some evidence that measures of adipose tissue using DEXA can tend towards underestimation. 102It also lacks the specificity to assess changes in individual muscle groups and is unable to identify myosteatosis, which appears to be an important prognostic feature on CT 103 and MRI body composition analyses 6 via observational studies.
CT body composition analysis was used in eight contemporary cancer cachexia trials (2010 onwards).Within the broader literature, a large number of observational studies have evaluated the impact of low muscle quantity 104 or radiodensity 103 as per CT body composition analysis, with consistently adverse prognostication noted for patients with cancer.While only one of the included trials identified significant differences between groups using CT-based endpoints, it must be noted that similar negative findings were seen on other concurrent endpoints (including body weight, DEXA and BIA) in all but one study. 69This is therefore likely reflective of the efficacy of the interventions used, rather than the sensitivity of the endpoint measured.Furthermore, some of the trials that in-cluded CT-based assessments used muscle groups or vertebral levels that have not yet been adequately validated to the same standard as the use of L3 cross-sectional area. 3Other assessments, such as radiodensity or volumetric body composition analysis, remain unexplored in the cachexia trial setting.CT is a very pragmatic imaging modality for some patient groups, owing to its routine use in the clinical staging and follow-up of cancer patients.However, when research imaging requirements are in excess of clinical need, the issue of associated ionizing radiation must be duly considered.Furthermore, trial timepoints may not align with the clinical pathway.Balancing this alongside finite resources and the need to limit patient burden can be challenging.Despite the dose reductions achieved with improved technology and focused scanning, 105 the necessary radiation exposure from CT still often exceeds that of DEXA. 106As such, when clinical imaging is not available, other assessment modalities such as MRI may be more appropriate.
Non-significant findings may result from an intervention that lacks efficacy, a trial that is inadequately powered relative to the chosen outcome or a measurement that lacks the precision to detect a true effect.With this in mind, limited inferences may be drawn regarding how a trial's significant or non-significant results may reflect on the selected endpoint(s), particularly if these are secondary endpoints that were not featured during sample size calculations.When comparing methods of body composition assessment, it is useful to consider several key measurement characteristics: reliability, validity, responsiveness to change, minimally important clinical difference and sensibility.
Reliability refers to the consistency and repeatability of a measurement when applied under similar conditions.One could consider intra-rater reliability (agreement of measures by a single evaluator on different occasions) or inter-rater reliability (agreement between different evaluators).Body weight would be expected to have excellent inter-rater reliability, owing to ease of measurement.However, achieving intra-rater reliability is dependent on consistent patient factors (e.g., clothing or fasting and hydration status) and use of the calibrated instruments.While good intra-rater and inter-rater reliability has been demonstrated for BIA, this also requires adherence to strict standardization of measurement conditions, 107 which may be difficult to achieve in real-world settings.DEXA and CT assessments of body composition may previously have been subject to lower levels of reliability, owing to their need for manual segmentation of anatomical features or regions of interest, but technological and software advancements have led to more reliable measurements now being obtained. 108Furthermore, CT performs well during tests of precision; the ability of a measurement technique to reproduce results when performed in an identical manner. 109alidity refers to a method's accuracy in assessing what it is intended to measure.In the case of cachexia, researchers are likely to be interested in changes in quantities of muscle and/or fat.This may present an obvious limitation with body weight, as it cannot inform us regarding the alterations to body composition that have led to any change in weight.This limits its use in isolation.Furthermore, the potential for fluid accumulations (e.g., ascites/peripheral oedema/hydration status) to influence body weight, BIA or even DEXA could lead to these modalities providing less valid assessments. 110While most studies of CT body composition extrapolate single-slice measurements to estimate whole-body composition, it is not known whether wasting occurs uniformly throughout the body. 109Analyses conducted over larger regions of interest may yet improve the validity of CT body composition. 111esponsiveness to change is a measure's ability to detect meaningful differences over time and is crucial for monitoring responses to trial interventions.Multiple factors can influence these parameters, and in the setting of cachexia clinical trials, it can be challenging to assess efficacy independent of confounders.While single axial slice CT (e.g., L3) and whole-body measurements are known to be highly correlated, 3 this may not hold true when assessing changes over time. 109 minimal clinically important difference (MCID) describes the smallest change that could be considered clinically significant.Such a metric may be determined with consideration of how changes in body composition endpoints relate to other outcomes.For example, what change in muscle mass is required to influence quality of life or survival?As has been observed with muscle mass and function, 112 the relationships between endpoints may be non-linear, and this must be acknowledged when considering responsiveness to change.The minimum body weight change that is considered clinically important for an individual's health is more commonly studied in the field of obesity than in cachexia.Semaglutide (glucagon-like peptide-1 [GLP-1] receptor agonist) was granted Food and Drug Administration (FDA) approval based on >5% weight loss being regarded as clinically meaningful. 113A >5% weight gain may be considered an equivalent MCID for treatment of cachexia, yet such precedent has not been set thus far.
Measurements should have sensibility (or interpretability) so they can be understood with ease.Body weight is meaningful and can be easily interpreted by clinicians, researchers and even patients.Conversely, bioelectrical estimates of body composition are less well known, and as such, the sensibility of these is limited.With improved consensus regarding effective assessment methods and endpoints, it should be anticipated that relevant stakeholders will become more informed regarding their chosen measurement techniques and how their findings relate to patients.
It is evident that no single assessment method currently fulfils all requirements.Rather, researchers should choose appropriate endpoints to align with their study aims and the cohort in question.What represents the 'gold-standard' measure would be dependent on the choice of intervention and its underlying mechanism.Assessment of body weight alongside dietary intake may be reasonable when assessing an appetite stimulant and has the added advantage of having regulatory approval in the obesity arena.Similarly, assessing DEXA or CT-based body composition would be sensible when trialling an exercise intervention aimed to improve lean mass.The practicalities, including cost and participant burden, mean that assessing these in clinical trials may be aspirational, despite a clear need.
A large volume of data has been compiled through each of the six reviews undertaken within this series.While there was a need to give a detailed appraisal within each of these, further work is ongoing to examine the relationships between these parameters.The findings presented are likely to have even greater value when considered in the context of other endpoints.For example, how do improvements in lean mass relate to physical function?The group aspires towards achieving a wider consensus alongside the identification and prioritization of areas for future research.Key strengths of this review include the broad search criteria and the robust methodological approach and appraisal process.However, the eligibility criteria could be considered a limitation, as balance was sought between the need to find trials of sufficient quality against having to appraise an impractical number of manuscripts.Although a specific time period was defined for the purposes of this work, it is accepted that trials published before 1990 may have yielded additional data, though this would pre-date the use of most endpoints considered by this review.The sample size cut-off was felt to be appropriate, as trials with <40 participants were expected to be insufficiently powered to assess changes in the endpoints being assessed.Furthermore, the minimum intervention time of 14 days was selected as interventions conducted for a shorter duration were felt unlikely to influence the disease course of cachexia.It should be acknowledged, however, that these restrictions may have precluded the inclusion of some informative trials.While the focus of this review was prospective, interventional trials, it is also acknowledged that many high-quality studies of other designs may yet inform the process of establishing a consensus regarding the optimum endpoints for cachexia trials, and future review of these would also be informative.
Endpoints should be intrinsically linked to inclusion criteria and vice versa.Changes in body composition are a key feature of cancer cachexia and, therefore, by definition, often feature as a baseline descriptive criterion (and, thus, outcome measure) in cachexia intervention trials.The Global Leadership Initiative on Malnutrition (GLIM) consensus criteria 114 were not designed to be used for the diagnosis of cachexia; however, they are intended to complement the existing cachexia literature, acknowledging that all patients with cachexia would meet their diagnosis of malnutrition.This group agreed that a body composition-based phenotypical criterion (weight loss, low BMI or reduced muscle mass) and an aetiological criterion (reduced food intake/assimilation or inflammation) are required for a diagnosis of malnutrition.Their recommendation for methods of estimating low muscle mass was for DEXA or 'corresponding standards using other body composition methods like BIA, CT or MRI'.The group also stated that anthropometric measures, such as arm muscle circumference, may be used as an alternative when radiological imaging is unavailable.Similarly, broad assessment methodology was proposed by the European consensus definition for sarcopenia, 5 with DEXA, BIA, CT and MRI all listed as options for evaluating 'muscle quantity or quality'.As shown in this present review, the range of modalities within the guidance is reflective of the heterogeneity within the existing literature.

Conclusions
Based on the findings presented herein, the use of body weight alongside a radiological modality for body composition analysis would seem like suitable endpoints for cancer cachexia trials.Thus far, body weight has been reported in a variety of ways, and further consensus is required regarding the specific body weight endpoint that should be used for future trials.The choice of radiological modality is likely to be dependent on the trial setting, population and intervention in question.When available, CT imaging is a well-validated and often pragmatic option that provides good levels of detail regarding body composition.Through ongoing exploration of this, and other assessment methods such as MRI, 110 further evidence is likely to emerge that will help standardize the appraisal of body composition.Endpoint heterogeneity in cancer cachexia clinical trials has greatly contributed to the lack of approved treatments by regulatory authorities. 115oreover, discrepancies between clinicians, regulatory industries and patients' perspectives regarding the most clinically relevant endpoints in cancer cachexia remain challenging.It is vital that consensus is achieved to ensure reporting consistency and maximize the efficacy of upcoming trials aiming to counteract the devastating effects of cancer cachexia.

Figure 2
Figure 2 Geographical distribution of included cancer cachexia trials.

Figure 3
Figure 3 Temporal trends in relative use of body weight and body composition assessments.BIA, bioimpedance analysis; BMI, body mass index; CT, computed tomography; DEXA, dual-energy X-ray absorptiometry.

Figure 4
Figure 4 Summary of anthropometric measures of body composition by intervention modality.BMI, body mass index.

Figure 6 A
Figure 6 A network diagram illustrating combinations of reported bioelectrical/radiological endpoints.Endpoints pertaining to muscle are highlighted in green.Endpoints pertaining to fat are highlighted in orange.Green connecting lines highlight the use of two measures of muscle in the same trial.Similarly, orange lines highlight the use of two measures of fat, while grey lines indicate combinations of muscle and fat measures.The size of the nodes reflects the number of studies that have reported the endpoint.The thickness of the connecting line reflects the number of studies reporting each pair of measures.BIA, bioimpedance analysis; CT, computed tomography; DEXA, dual-energy X-ray absorptiometry; LBM, lean body mass.

Table 1
Key characteristics of included clinical trials a Kardinal et al. b Pottel et al. b Garcia et al.

Table 1
(continued) b Uster et al. b Xie et al. c Cereda et al. b Laviano et al. b Obling et al.

Table 1
a Qiu et al. b Hunter et al. b Kutz et al. b Note: Sample sizes are reported as per 'intention to treat'.Abbreviations: BIA, bioimpedance analysis; BMI, body mass index; CT, computed tomography; DEXA, dual-energy X-ray absorptiometry; EPA, eicosapentaenoic acid; LBM, lean body mass; NSCLC, non-small-cell lung cancer; PEG, percutaneous endoscopic gastrostomy; RCT, randomized controlled trial.a Endpoints that are bold underlined had a statistically significant difference between groups.b Endpoint expressed as change in absolute value from baseline.c Endpoint expressed as percentage change from baseline.

Table 2
Utilization of body composition endpoints Documentation on the literature search for 'What is the optimal endpoint to evaluate effect of interventions aiming to treat cancer cachexia?'All searches were done on 2 June 2021 by Gunn Kleven, senior librarian at the Library of Medicine and Science, University of Oslo.