Clinical and biological characterization of skeletal muscle tissue biopsies of surgical cancer patients

Abstract

Background

Researchers increasingly use intraoperative muscle biopsy to investigate mechanisms of skeletal muscle atrophy in patients with cancer. Muscles have been assessed for morphological, cellular, and biochemical features. The aim of this study was to conduct a state‐of‐the‐science review of this literature and, secondly, to evaluate clinical and biological variation in biopsies of rectus abdominis (RA) muscle from a cohort of patients with malignancies.

Methods

Literature was searched for reports on muscle biopsies from patients with a cancer diagnosis. Quality of reports and risk of bias were assessed. Data abstracted included patient characteristics and diagnoses, sample size, tissue collection and biobanking procedures, and results. A cohort of cancer patients (n = 190, 88% gastrointestinal malignancies), who underwent open abdominal surgery as part of their clinical care, consented to RA biopsy from the site of incision. Computed tomography (CT) scans were used to quantify total abdominal muscle and RA cross‐sectional areas and radiodensity. Biopsies were assessed for muscle fibre area (μm²), fibre types, myosin heavy chain isoforms, and expression of genes selected for their involvement in catabolic pathways of muscle.

Results

Muscle biopsy occurred in 59 studies (total N = 1585 participants). RA was biopsied intraoperatively in 40 studies (67%), followed by quadriceps (26%; percutaneous biopsy) and other muscles (7%). Cancer site and stage, % of male participants, and age were highly variable between studies. Details regarding patient medical history and biopsy procedures were frequently absent. Lack of description of the population(s) sampled and low sample size contributed to low quality and risk of bias. Weight‐losing cases were compared with weight stable cancer or healthy controls without considering a measure of muscle mass in 21 out of 44 studies. In the cohort of patients providing biopsy for this study, 78% of patients had preoperative CT scans and a high proportion (64%) met published criteria for sarcopenia. Fibre type distribution in RA was type I (46% ± 13), hybrid type I/IIA (1% ± 1), type IIA (36% ± 10), hybrid type IIA/D (15% ± 14), and type IID (2% ± 5). Sexual dimorphism was prominent in RA CT cross‐sectional area, mean fibre cross‐sectional area, and in expression of genes associated with muscle growth, apoptosis, and inflammation (P < 0.05). Medical history revealed multiple co‐morbid conditions and medications.

Conclusions

Continued collaboration between researchers and cancer surgeons enables a more complete understanding of mechanisms of cancer‐associated muscle atrophy. Standardization of biobanking practices, tissue manipulation, patient characterization, and classification will enhance the consistency, reliability, and comparability of future studies.

Introduction

Several radiologically defined features of skeletal muscle have been associated with clinical outcomes in patients with cancer. Reduced muscle mass (i.e. sarcopenia), loss of muscle mass over time, and reduced muscle radiodensity are related to mortality, shorter progression—free survival, chemotherapy toxicity, and complications of cancer surger.1, 2, 3, 4 In light of the associations between muscle and outcomes, researchers are increasingly investigating the pathophysiology of muscle abnormalities5, 6, 7 and attempting to relate the findings to the much broader base of knowledge that exists from research in animal models. Muscle may be obtained from cancer patients by percutaneous biopsy as well as intraoperatively during cancer surgery. Clinical data aligned with the biopsy provides a comprehensive approach to understand cancer cachexia from the vantage point of muscle wasting. Evaluation of human muscle contributes significantly to the understanding of molecular mechanisms in a variety of primary pathologies of skeletal muscle.8, 9

Biopsy and tissue manipulation techniques can induce changes in the muscle that alter enzyme activity, metabolite concentrations, and protein metabolism.10, 11, 12 Also, patient characteristics such as age, sex, cancer type, co‐morbidities, and medications (including chemotherapy) taken at the time of biopsy collection are known factors that influence muscle metabolism.13, 14, 15, 16, 17 These methodological issues pose limitations in the reliability, interpretation, and comparability of the findings on muscle biopsies in patients with cancer. Therefore, our first aim was to conduct a state‐of‐the‐science review of the literature on muscle biopsy in cancer patients. This type of review retains many features of a systematic review except that studies are not excluded on the basis of a quality assessment and thus presents a broader search of the literature. An associated aim was to provide recommendations of components to consider when evaluating and reporting results of muscle biopsies from cancer patients.

The second aim of this study was to evaluate sources of variation in the muscle biopsy material to better understand the risk of sampling bias, to determine variance and effect size to enable sample size calculations, and to determine the possible consequences of sexual dimorphism and age as confounders using a relatively well‐powered sample (n = 190). Our research group has experience in the radiological characterization of muscle2, 18, 19 and skeletal muscle morphology, cell biology, and biochemistry.7, 20, 21, 22 Our collaborative effort with hepatopancreatobiliary cancer surgeons has enabled muscle biobanking and exploration of muscle biology within large populations. We have published studies on muscle expression of mRNA, microRNA, and alternative splice variants,20, 21, 23 alongside specific and precise measures of muscle mass, radiodensity, and muscle loss.

Materials and methods

Literature review

A state‐of‐the‐science review24 is a broad search of the literature that includes all studies in a particular area. Our review protocol follows the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses25 guidelines to reduce bias (Figure 1). Articles indexed in SCOPUS from 1 January 1900 to 16 August 2018 were queried to capture reports on skeletal muscle biopsies from cancer patients. Search terms included adult humans, malignant disease [(cancer) OR (neoplasm) OR (carcinoma) OR (tumor) OR (malignant) OR (metastasis)], skeletal muscle [(skeletal muscle) OR (muscle mass) OR (lean body mass) OR (rectus abdominis) OR (cachexia) OR terms for other specific muscle], and biopsy. Review articles and studies on experimental models, laboratory animals, non‐cancer populations, or those not employing muscle biopsies were excluded. Bibliographies of identified articles were hand searched to find additional relevant publications. There were no exclusion criteria regarding number of patients and type of study (retrospective, prospective, or cross sectional). Data were extracted from the result sections, tables, and figures of each article. As we did not aggregate the data, no additional data were contributed from the investigators.

Figure 1.

Flow chart of search. PRISMA diagram for the identification, screening, eligibility, and inclusion of papers (1 January 1990–16 August 2018) from SCOPUS. All articles included investigated cancer, skeletal muscle, and muscle biopsies. Excluded records: review articles and ongoing clinical trials.

Two reviewers independently assessed each of the included studies, and disagreements were resolved by consensus. A score for study quality was given using assessment tools provided by the National Heart, Lung and Blood Institute (NIH—U.S. Department of Health & Human Services) for cross‐sectional, cohort, case‐control, randomized control trials and before–after studies. The Newcastle–Ottawa scale modified for cross‐sectional studies26 was used to give a bias score based on the (i) representativeness, (ii) size, and (iii) non‐respondent report.

Rectus abdominis biological characterization

Subjects and acquisition of muscle samples

The study was approved by the Health Research Ethics Board of Alberta‐Cancer. Patients undergoing elective abdominal surgery were consecutively approached to participate in tumour and tissue banking at a hepatopancreatobiliary surgical service in Alberta, Canada. Three per cent of approached patients declined participation. Patients provided written informed consent for muscle biopsy and tissue banking. Release of n = 190 samples from the bank for analysis, as well as patient information (demographic, clinical, and operative data) from medical records, was performed under the auspices of Protocol ETH‐21709: The Molecular Profile of Cancer Cachexia. Patients consent freely to muscle biopsy from the site of incision at the time of surgery, as this entails little if any incremental discomfort or risk, as the surgery is inherently invasive. All patients were either diagnosed as having cancer or were suspected of having cancer due to their symptoms and radiological assessments such as computed tomography (CT) imaging.

The study cohort and conditions for acquisition of muscle samples have been described previously.23 Briefly, rectus abdominis (0.5–3 g) samples were collected during open abdominal surgery scheduled as part of their clinical care. Upper abdominal transverse incision was performed, and muscle biopsy was obtained at opening by sharp dissection, without the use of electrocautery.

Computed tomography image analysis

Digital axial CT scans performed preoperatively and used to plan surgery were used to quantify skeletal muscle cross‐sectional area (CSA, cm²) as in our prior work.18, 27 Measures with CT have excellent precision (precision error values of ~1.5%).28 Briefly, images at the 3rd lumbar vertebra (L3) were analysed for total L3‐CSA within a specified Hounsfield unit (HU) range (−29 to +150) using Slice‐O‐Matic software (v.4.3, Tomovision, Magog, Canada). Muscle area was normalized for stature and reported as skeletal muscle index (SMI, cm²/m²). Mean radiodensity (HU) was also reported. Adipose tissue CSA at L3 was calculated in a HU range of −150 to −50 and −190 to −30, for visceral and subcutaneous adipose tissue, respectively.28 The distribution of SMI of the patients providing biopsy for this study was compared with a previously described large cohort of oncology patients (n = 1473) to confirm that the population sampled is representative of muscle mass distribution and mean values for our population (Figure 2). Sarcopenia was classified according to previously reported19, 29 sex‐specific and body mass index (BMI)‐specific criteria: for BMI <30 kg/m², SMI <52.3 cm²/m² for men and <38.6 cm²/m² for women, and for BMI ≥30 kg/m², SMI <54.3 cm²/m² for men and <46.6 cm²/m² for women.

Figure 2.

Figure represents overlap of L3 SMI distributions for male (A) and female (B) patients of current cancer population (small, light gray distribution) and a cancer cohort with solid tumours of gastrointestinal tract and lung (big, dark gray distribution).1, 2 (A) L3 SMI mean ± standard deviation values are 50.8 ± 8.3 and 51.5 ± 8.9 cm²/m² for the current population and Martin et al. gastrointestinal and lung cancer cohort, respectively. (B) L3 SMI mean values are 39.8 ± 6 and 41.3 ± 7 cm²/m² for the current population and Martin et al. gastrointestinal and lung cancer cohort, respectively. L3 SMI mean values for healthy 30‐year‐old kidney donor candidates (dotted line) are placed at 60.9 and 47.7 cm²/m² for men and women, respectively.15

Processing of muscle biopsy

From each biopsy, several analysis were performed, each with specific preparation procedures. In the operating room, visible adipose and connective tissue was removed from the biopsy and it was cut into two pieces: one piece to be used for analysis of gene expression, and myosin heavy chain (MyHC) by electrophoresis was immediately frozen in liquid nitrogen in the operating room prior to being transported to the lab for storage in liquid nitrogen until analysis. The other piece of the biopsy to be used for microscopy was transported on ice to the laboratory within 20 to 30 min. For morphological preservation, isopentane (2‐methylbutane, C₅H₁₂) was cooled at −160°C in liquid nitrogen for 20 min or until the appearance of a thick frozen layer at the bottom of the container. A piece of muscle was oriented for transverse section and delicately placed on aluminum foil. Tissue was submerged in isopentane for 20 s, and aluminum foil was turned upside down to allow full exposure of the muscle section. After submersion, tissue was wrapped and left in liquid nitrogen for 5 min. Information about surgery date, time, and sample reception was documented.

Immunofluorescence: fibre types, laminin/dystrophin, and nuclear stain

Muscle serial sections (10 μm) were cryosectioned (cryostat Leica model CM300) transversely at −22°C and stored at −80°C until staining. MyHC I, IID, and IIA were determined as previously described.30 Primary and secondary antibodies are described in Supporting Information, Table S1. After the secondary antibody application, a nuclear stain (4',6‐diamidino‐2‐phenylindole) was added for 2 min and washed. Slides (Apex^TM superior adhesive slides, Leica biosystems) were mounted, covered, and let dry for 12 h. Images for tissue sections were acquired using a 20X/0.85 oil lens with a spinning disk confocal microscope (Quorum Wave FX Spinning Disc Confocal System–Quorum technologies). Individual Z‐stacked images were assembled to create a composite image of a whole tissue cross section. Tissue images were capture and analysed with Volocity 6.3 software (PerkinElmer, Waltham, MA, USA). A software script was established to identify muscle fibres types (I, I/IIA, IIA, IIA/D, and D) using intensity of the MyHC stains and quantified automatically by the software. Mean muscle fibre area (μm²) was calculated by the detection of membrane (laminin/dystrophin antibody) fluorescence of muscle fibres in a cross section. Percentage of fibres with centralized nuclei was manually assessed by selecting muscle fibres with mispositioned nuclei (clearly separated from sarcolemma, equidistant, or not) in a tissue cross section.

Electrophoretic analysis of myosin heavy chain isoform content

Semi‐quantitative MyHC isoform analyses were completed on frozen rectus abdominis using western blotting as previously described.30, 31, 32 All three of the adult MyHC isoforms (I, IIA, and IID) were clearly visible on all gels and reliably quantified in at least triplicate by integrated densitometry (Syngene ChemiGenius, GeneTools, Syngene).

Triglyceride content analysis

A piece of biopsy (50 mg) was ground using a frozen pestle and mortar without letting the tissue thaw. Ground tissue was homogenized in a 1.6 mL calcium chloride (CaCl2; 0.025%) solution with glass beads (0.5 mm diameter; FastPrep ®‐24, MP Biomedicals, Santa Ana, CA, USA) in 20 s intervals for 1 min. Samples were placed on ice for 15 s between each homogenization interval. A modified Folch method was used to extract lipids using chloroform/methanol (2:1, vol/vol) as previously described.33, 34 The triglyceride (TG) fraction was isolated on G‐plates and the TG band was identified and scraped. An internal standard C15:0 (10.2 mg/100 mL hexane) was added, followed by saponification and methylation. Samples were analysed using gas liquid chromatography (flame ionization detector) on a Varian 3900 (Varian Instruments, Georgetown, ON, Canada). Quantity of fatty acids within the TG fraction was calculated by comparison with the known concentration of the internal standard and sum of all fatty acids was reported as total TG.

Gene expression: microarray

Microarray was conducted as previously described.23 The data have been deposited in the U.S. National Center for Biotechnology Information Gene Expression Omnibus25 and are accessible through GEO series accession number GSE41726.

Statistical analysis

Statistical analyses were conducted in IBM® SPSS ® software, version 24. A test for normal distribution was applied to the continuous variables. Descriptive statistics were reported as mean ± standard deviation. Comparisons between groups were conducted using independent t‐test or Mann–Whitney U according to the variable normal distribution and χ² test for categorical variables. Statistical significance was considered at P values less than 0.05 (two‐sided).

Results

Literature review

A total of 59 articles reporting analysis of skeletal muscle in cancer populations were reviewed. The Preferred Reporting Items for Systematic Reviews and Meta‐Analyses25 flow diagram of our search strategy is shown in Figure 1.

Study quality and design

Table 1 includes all of the extracted data as well as scores for sampling bias (Newcastle–Ottawa scale) and study quality assessments (NIH). In general, the study quality rated as low for the majority of studies (Table 1). Applying the Newcastle–Ottawa criteria for sampling bias revealed the majority of studies had a high risk of sampling bias with 58% of studies lacking representativeness, 96% lacking sample size justification, and no study mentioned non‐respondent rate (% of population approached who declined participation). Muscles biopsied were rectus abdominis (n = 40), quadriceps (n = 20), tibialis anterior (n = 1), gastrocnemius (n = 1), pectoralis major (n = 1), sternocleidomastoid (n = 1), serratus anterior (n = 1), diaphragm (n = 1), and latissimus dorsi (n = 1), and in seven studies, more than one muscle was collected. Four studies reported evaluation of rectus abdominis from cancer patients and quadriceps for non‐cancer controls, and four studies reported biopsied muscle from two or three different muscles.

Table 1.

Original articles reporting muscle biopsy collection in patients with cancer: assessment of bias and study quality, population characteristics weight loss, or cancer cachexia classification criteria

Author	Biasc	Qualityd	Muscle	Cancer site	Cancer Stage	Cancer population		Control group		Patient weight loss or cachexia criteria
						n (% male)	Age (years) mean ± SD	n (% male)	Age (years) mean ± SD
Acharyya 200535	1/3	3/12	RA	Gastric	NR	27 (NR)	NR	14 (NR)	NR	N/A
Agustsson 201136	1/3	3/12	RA	Pancreas Other GI	NR	Pancreas 13 (30) Other GI 8 (37)	Pancreas: 70 ± 2 Other: 68 ± 3	Benign: 8 (37) Pancreatitis: 8 (63)	Benign: 53 ± 4 Pancreatitis 52 ± 3	NR
Aversa 201637	1/3	6/12	RA	Colorectal pancreas gastric	1–4	All: 29 (59) WS = 14 WL = 15	68 ± 10.7	11 (63)	63 ± 13.2	5% WL (6 months)
Bonetto 201338	1/3	3/12	RA	Gastric	1–4	16 (NR)	64 ± 11	6 (NR)	62 ± 17.4	>5% WL
Bossola 200639	1/3	5/12	RA	Gastric	1–4	16 (50)	60.8 ± 11.2	5 (60)	65.6 ± 7.5	WL mild: 0–5%. WL moderate: 6–10%. WL severe: >10%.
Bossola 200140	1/3	4/12	RA	Gastric	NR	20 (55)	61 ± 79.6	10 (60)	62 ± 45.8	WL mild: 0–5%. Moderate 6–10%. Severe: >10%.
Bossola 200341	1/3	5/12	RA	Gastric	NR	23 (61)	59.5 ± 16.1	14 (64)	61.2 ± 12.3	>10% WL
Busquets 200742	0/3	3/12	RA	Esophageal gastric pancreas	1–4	16 (NR)	66 ± 10	11 (NR)	66 ± 10.2	>5% WL (1 month)
DeJong 200543	0/3	4/12	RA	Pancreas	1–4	16 (63)	66 ± 8	11 (81)	67 ± 13.2	N/A
D'Orlando 201444	1/3	6/12	RA	Gastric	1–4	38 (66)	68.1 ± 11.6	12 (58)	64.2 ± 11.6	>5% WL (6 months)
Eley 200845	1/3	3/12	RA	Esophageal gastric	1–4	15 (87)	66 (49–83)a	9 (10)	56 (41–86)a	N
Johns 201722	2/3	9/12	RA	Esophageal gastric lung and other	1–4	134 (51)	65 ± 13	N/A	N/A	WL >5% >10% >15% and SMI with any degree of WL (>2%)
Johns 201446	0/3	5/12	RA	Upper GI pancreas	NR	41 (73)	65 ± 12.8	N/A	N/A	>5% WL (6 months) and low muscularity with 2% WL
Khal 200547	0/3	1/12	RA	Pancreas colorectal	NR	All: 18 (67) WS = 5 (60) WL = 13 (69)	WS: 79.8 ± 2.2 WL: 70.6 ± 8.2	10 (80)	69.6 ± 7.3	WL moderate: 1–11%. WL severe: >11%.
Lundholm 197648	1/3	3/12	RA	Esophageal gastric pancreas colorectal kidney and others	NR	43 (44)	♂: 62 ± 13.1 ♀: 63 ± 9.7	55 (51)	56 ± 14.8	N/A
Marzetti 201749	1/3	5/12	RA	Gastric	1–4	All: 18 (94) WS = 9 (100) WL = 9 (89)	WS: 70.6 ± 8.63 WL: 66.8 ± 12.5	9 (88)	57.4 ± 15.9	>5% WL (6 months)
Narasimhan 201721	2/3	8/12	RA	Pancreas colorectal	1–4	22 (41)	64.9 ± 10	20 (45)	63.6 ± 7.9	>5% WL (6 months) or BMI of <20 with WL >2% and sarcopenia
Narasimhan 201820	1/3	5/12	RA	Pancreas colorectal	1–4	All: 40 (43) WS = 19 (47) WL = 21 (40)	WS: 64 ± 8 WL: 66 ± 11	N/A	N/A	WL >5% >10% >15% and sarcopenic (SMI) with any degree of WL (>2%)
Noguchi 199850	0/3	3/12	RA	Esophageal gastric colorectal	1–4	10 (90)	56 (50 to 63)a	N/A	N/A	N/A
Pessina 201051	1/3	6/12	RA	Gastric	1–3	30 (57)	63.8 ± 2.8	8 (62)	64.2 ± 2.6	N/A
Prokopchuk 201652	0/3	4/12	RA	Pancreas	1–4	All: 25 (32) NC = 13 (38) CC = 12 (25)	NC: 67 (36–87) CC: 70 (52–83)a	Benign = 15 (80) Pancreatitis = 9 (45)	Benign: 67 (32–73) Pancreatitis: 49.5 (40–75)a	>10% WL (6 months)
Ramage 201853	1/3	3/12	RA	Esophageal gastric pancreas	1–4	32 (81)	64.5 (43–83)	N/A	N/A	>5% WL of pre‐illness
Rhoads 200954	1/3	6/12	RA	Gastric	1–4	All: 14 (57) WS = 6 (66) WL = 8 (50)	64.2 ± 3.8	10 (60)	63.9 ± 2.8	NR
Schmitt 200755	0/3	2/12	RA	Pancreas	2, 4	All: 16 (63) NC = 8 (37) CC = 8 (88)	NC: 62 ± 8.5 CC: 53 ± 11.3	N/A	N/A	>10% WL (6 months)
Skorokhod 201256	0/3	1/12	RA	Pancreas	2–4	All: 23 (61) WS = 13 (69) WL = 10 (50)	WS: 66 (51–69) WL: 65 (57–74)	N/A	N/A	>10% WL of pre‐illness
Smith 201057	0/3	4/12	RA	Gastric	1–4	15 (67)	66 ± 11.6	15 (80)	57 ± 19.3	>5% WL
Stephens 201158	0/3	2/12	RA	Esophageal gastric pancreas rectal	2–4	19 (58)	67 ± 10	6 (33)	53 ± 8	>10% WL (6 months)
Stephens 201559	0/3	3/12	RA	Esophageal gastric pancreas and other	1–4	All: 92 (72) NC = 41 (82) CC = 51(63)	All: 65 ± 10 NC: 68 ± 9 CC: 63 ± 9	15 (53)	56 ± 17	>5% WL
Stretch 201323	0/3	4/12	RA	Liver bile duct GI tract pancreas and other	NR	134 (51)	♂: 59 ± 13 ♀: 63 ± 13	N/A	N/A	N/A
Sun 201260	0/3	5/12	RA	Gastric	1–4	102 (71)	62.13 ± 6.54	29 (72)	61.8 ± 6.4	>10% WL
Taskin 201461	0/3	1/12	RA	Colorectal pancreas gastric and other	NR	All: 14 (50) NC = 8 (37) CC = 6 (66)	NC: 68 ± 5 CC: 70 ± 15	5 (40)	77 ± 5	>10% WL (6 months) weight stable <5%
Williams 199962	0/3	2/12	RA	Colorectal	NR	6 (66)	67 (53–76)a	6 (83)	54 (22–92)a	N/A
Zeiderman 199163	0/3	5/12	RA	Esophageal gastric colorectal pancreas	NR	30 (70)	Hospital diet: 67 ± 9.5 3 days intervention: 72 ± 3.2 7 days intervention: 67 ± 6.3	N/A	N/A	>5 kg WL (3 months)
Zampieri 201064	0/3	3/12	RA, QF	Colorectal	NR	14 (36)	65.1 ± 10.3	Myopathy: 13 (38) Healthy = 19 (NR)	Myopathy: 64.3 ± 6.3 Healthy: 30.1 ± 13.3	N/A
Zampieri 200965	0/3	1/12	RA, QF	Colorectal	2–3	10 (30)	65.1 ± 10.3	10 (NR)	22.7 ± 2.6	N/A
Zampieri 201066	1/3	3/12	RA, QF	Colorectal	2–3	11 (36)	65.1 ± 10.3	7 (0)	44.5 ± 18.3	N/A
Aversa 201267	1/3	3/12	RA, SA	NSCLC gastric	1–4	39 (74)	Lung: 66 ± 9 Gastric: 65 ± 10	10 (50)	Abdominal: 63 ± 10 Thoracic: 65 ± 12	NR
MacDonald 201568	0/3	2/12	RA, QF	Esophageal gastric	1–4	All: 14 (57) WS = 6 (66) WL = 8 (50)	WS: 62.5 (57.0–70.3)b WL: 63.4 (61.5–66.3)b	7 (42)	52.1 (51.5–53.1)b	>5% WL
Shaw 199169	0/3	6/14	RA, SCM	Colorectal pancreas head & neck thyroid and other	NR	All: 43 (42) WS = 25 (48) WL = 18 (66)	WS: 61 ± 20 WL: 64 ± 12.7	18 (33)	57 ± 16.9	>15% WL of pre‐illness
Stephens 201070	1/3	3/12	RA, VL, DIAPH	Esophageal gastric pancreas	NR	18 (66) WL	67 ± 8.4	3 (66)	45 ± 3.4	>5% WL
Brzeszczynska 201671	0/3	2/12	QF	Esophageal gastric pancreas	2–3	All: 28 (75) NC = 18 (72) CC = 10 (80)	NC: 67 ± 10.5 CC: 65 ± 8.1	Middle age 20 (60) Elderly: 21 (52)	Middle age: 61 ± 7 Elderly: 79 ± 3.6	>5% WL of pre‐illness
Ebhardt 201772	0/3	1/12	QF	Esophageal gastric pancreas	NR	All: 19 (79) NC = 14 (85) CC = 5 (60)	Non‐CC: 66.3 ± 10.2 CC: 64 ± 4.1	Non‐sarcopenic 10 (60) Sarcopenic 8 (50)	Non‐sarcopenic: 77.4 ± 2.3 Sarcopenic: 80.3 ± 3.9	>5% WL of pre‐illness
Gallagher 201273	1/3	7/14	QF	Esophageal gastric pancreas	1–3	12 (83)	65	6 (66)	58	NR
Christensen 201674	N/A	13/14	VL	Testicular germ cell	NR	8 (100)	33.4 ± 7.5	Control = 9 (100) Ref = 13 (100)	Control: 37.8 ± 7.6 Reference group: 32.1 ± 6.3	N/A
Christensen 201475	N/A	13/14	VL	Testicular germ cell	NR	15 (100)	Intervention: 34.4 ± 7.6 Control: 35.8 ± 8.9	19 (100)	31.5 ± 6.0	N/A
Lamboley 201776	1/3	3/12	VL	Prostate	2	8 (100)	68 ± 5.6	14 (100)	71 ± 3.7	N/A
Nilsen 201677	N/A	9/14	VL	Prostate	NR	12 (100)	67 ± 7	11 (100)	64 ± 6	N/A
Op den Kamp 201578	0/3	6/12	VL	NSCLC	3–4	All: 26 (65) Pre‐CC = 10 (80) CC = 16 (56)	Pre‐CC: 62.4 ± 10.4 CC: 59.8 ± 8.2	22 (59)	61.4 ± 7.0	>5% WL (6 months)
Op den Kamp 201279	0/3	3/12	VL	NSCLC	1–3	16 (93)	65.9 ± 7.5	10 (70)	63.7 ± 5.6	10% WL (6 months)
Op den Kamp 201380	0/3	5/12	VL	NSCLC	3–4	All: 26 (65) Pre‐CC = 10 (80) CC = 16 (56)	Pre‐CC: 62.4 ± 10.4 CC: 59.8 ± 8.2	22 (59)	61.4 ± 7.02	5% WL (6 months) 2% WL with BMI <20 or sarcopenia
Phillips 201381	0/3	4/14	VL	Colorectal	Early	8 (50)	62.5 ± 23.4	8 (50)	70.7 ± 4.5	N/A
Puig‐Vilanova 201482	1/3	3/12	VL	Lung	1–4	10 (100)	65 ± 9	Healthy = 10 (100) COPD = 16 (100)	65 ± 11 64 ± 9	Fat free mass index: <18.5 kg/m2
Weber 200783	0/3	3/12	VL	Gastric pancreas leukemia	NR	17 (53)	52.5 ± 6.5	27 (52)	57.9 ± 12.4	>10% WL (6 months)
Weber 200984	0/3	2/12	VL	GI tract (not defined)	NR	19 (52)	58 ± 9	19 (53)	56 ± 7	>10% WL (6 months)
Williams 201285	0/3	5/12	VL	Colorectal	Early	13 (46)	66 ± 10.8	8 (50)	71 ± 5.6	N/A
Banduseela 200786	N/A	N/A	TA	NSCLC	NR	1 (100)	63	6 (50)	Healthy: 49 ± 7 Myopathy: 60 ± 18	NR
Higuchi 200087	N/A	N/A	Gastroc	Gastric	NR	1 (100)	54	N/A	N/A	N/A
Jagoe 200288	0/3	1/12	LD	Lung	3–4	36 (75)	64.1 ± 9	10 (40)	51.3 ± 15.1	Any % WL (6 months)
Bohlen 201889	0/3	4/12	PM	Breast	1–4	14 (0)	56.5 ± 17.2	6 (0)	44.2 ± 7.4	N/A

Values reported as mean ± standard deviation (SD) unless indicated otherwise. BMI, body mass index; DIAPH, diaphragm; Gastroc, gastrocnemius; GI, gastrointestinal; LD, latissimus dorsi; N/A, not applicable; NC, non‐cachexia; NIH‐NHLBI, National Heart Lung and Blood Institute; NSCLC, non‐small cell lung carcinoma; NR, not reported; PM, pectoralis major; QF, quadriceps femoris; RA, rectus abdominis; TA, tibialis anterior; SA, serratus anterior; SCM, sternocleidomastoid; SMI, skeletal muscle index; VA, vastus lateralis; WL, weight loss; WS, weight stable.

^aMedian (range).

^bMedian (interquartile range).

^cModified Newcastle–Ottawa scale.

^dQuality assessment score—high score means high quality.

Gastrointestinal cancers were the most common diagnoses; 31/59 studies included patients of exclusively one cancer type: colorectal, pancreatic, gastric, breast, or prostate. Inclusion of patients with two or more cancer types was reported in 27/59 studies. Cancer stage or presence of metastasis was described in 39/59 studies. Combined data from two or more cancer stages were reported in 38/59 studies.

The majority of studies were cross sectional (Supporting Information, Table S2). For investigation of patients with cancer cachexia, weight loss was considered as the main reference for classification. In 36 studies, weight loss was graded with varying cut points (e.g. 5%, 10%, or 15%). Time frame of weight loss was not specified in 16 of these studies (Table 1). Percentage weight loss ranged from 5% to 22% in weight‐losing groups (Supporting Information, Table S2). Measures of body composition were included in 25 studies; however, these measures were used to assess muscle mass or rate of muscle wasting over time in only seven studies (Supporting Information, Table S2).

Total sample size in each study was generally limited (mean, n = 26; median n = 18; and range 1–134). Seventy‐six per cent of studies included n ≤ 30 cancer patients; 48/59 studies included a non‐cancer control group, sample size ranging from n = 3 to 41. Fifty‐two studies included men and women, 5 studies only men, 1 study only women, and 2 studies did not specify the sex of their patients. For those studies including both sexes, 50 had an imbalance between treatment groups in the % of male and female patients, and only 3 studies matched the number of male and female participants. When reporting the results, almost all of the studies (98%) presented aggregate data from men and women.

When a non‐cancer control group was employed in the study, the majority of studies included control groups that went under surgical procedures (i.e. cholelithiasis and cholecystitis, ovarian cyst, inguinal hernia, laparocele, abdominal aorta aneurysm, hemangioma of liver, gallstones, and chronic pancreatitis) or healthy volunteers (Supporting Information, Table S2). No study defined the criteria used to select healthy volunteers. Table 1 highlights the features of the cancer groups compared with control groups. More than 54% of the studies included cancer patients with an average age of ≥65 years, and for studies involving non‐cancer patients as controls, 26% included patients with an average age of ≥65 years.

Most (33/59) reports failed to mention co‐morbidities as a component of their exclusion criteria or patient's demographics. Commonly excluded diagnoses were diabetes, chronic obstructive pulmonary disease, liver failure, renal failure, chronic hepatitis, autoimmune diseases, and inflammatory bowel disease. Use of medications (e.g. corticosteroids, anabolic/catabolic agents, and/or beta blockers) was described in 17 studies as clinical characteristics or exclusion criteria. Prior exposure to antineoplastic drugs was reported in 14/59 studies. Inclusion of patients naïve to chemotherapy or radiotherapy was stated in 6/59 studies, two studies acknowledged the inclusion of some patients with one or fewer cycles of chemotherapy that concluded 4 weeks previous to biopsy collection.

Technical considerations

Biobanking protocol and tissue manipulation

Abdominal and thoracic muscle biopsies were collected during a surgical procedure in 43 studies, with collection at the start of surgery being explicitly stated in 31 studies (Table 2). Presence or absence of tissue cauterization was specified in 29/43 studies. Percutaneous procedure (needle biopsy) was the main method for collection of muscles of the lower limb (n = 19 studies), open muscle biopsy technique was reported in one study, and in one study, the collection method was unspecified. For both surgical and percutaneous biopsies, removal of blood traces and/or fat/fibrotic tissue after collection was mentioned in 7/59 studies (Table 2).

Table 2.

Biopsy collection and handling procedures across the studies

Author	Biopsy collection (collected in start or end of surgery)	Cauterized	Blood traces, fat, or connective tissue removed	Sample handling and storage conditions
Acharyya 200535	NR	NR	NR	NR
Agustsson 201136	Initial phase of surgery	NR	NR	Incubated in vitro
Aversa 201637	Initial phase of surgery	No	NR	Immediately frozen, stored at −80°C
Aversa 201267	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
Banduseela 200786	Percutaneous biopsy (local anaesthesia)	N/A	Yes (fat, connective tissue)	Immediately frozen, stored at −80°C
Bohlen 201889	NR	N/A	NR	Stored in RNA stabilization solution at −4°C overnight and then stored at −80°C
Bonetto 201338	Initial phase of surgery	No	NR	Immediately frozen, stored at −80°C
Bossola 200639	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
Bossola 200140	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
Bossola 200341	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
Brzeszczynska 201671	Initial phase of surgery	No	Yes (blood)	Immediately frozen, stored at −80°C
Busquets 200742	Initial phase of surgery	No	NR	Immediately frozen, stored at −80°C
Christensen 201674	Percutaneous biopsy (local anaesthesia)	N/A	NR	Immediately frozen, stored at −80°C
Christensen 201475	Percutaneous biopsy (local anaesthesia)	N/A	NR	Immediately frozen, stored at −80°C
DeJong 200543	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
D'Orlando 201444	Initial phase of surgery	No	NR	Immediately frozen, stored at −80°C
Ebhardt 201772	Percutaneous biopsy (local anaesthesia)	N/A	Yes (blood)	Immediately frozen, stored at −80°C
Eley 200845	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
Gallagher 201273	Percutaneous biopsy (local anaesthesia)	N/A	Yes (blood)	Immediately frozen, stored at −80°C
Higuchi 200087	NR	N/A	NR	NR
Jagoe 200288	Initial phase of surgery	No	NR	Immediately frozen, stored at −80°C
Johns 201722	Initial phase of surgery	No	NR	Immediately frozen, stored in liquid nitrogen
Johns 201446	Initial phase of surgery	NR	Yes (blood)	Immediately frozen, stored at −80°C
Khal 200547	NR	No	NR	Immediately frozen, stored at −70°C
Lamboley 201776	Percutaneous biopsy (local anaesthesia)	N/A	Yes (blood)	Immediately and stored in liquid nitrogen
Lundholm 197648	Initial phase of surgery	NR	NR	Muscle fibre isolation on fresh tissue
MacDonald 201568	Initial phase of surgery and percutaneous biopsy (local anaesthesia)	NR	NR	Immediately frozen in liquid nitrogen, storage temperature NR
Marzetti 201749	Initial phase of surgery	No	NR	Immediately frozen, stored at −80°C
Narasimhan 201721	Initial phase of surgery	No	NR	Immediately frozen, stored at −80°C
Narasimhan 201820	Initial phase of surgery	No	NR	Immediately frozen, stored at −80°C
Nilsen 201677	Percutaneous biopsy (local anaesthesia)	N/A	Yes (fat)	Frozen by immersion in isopentane, stored at −80°C
Noguchi 199850	Initial phase of surgery	NR	NR	Immediately frozen in situ, stored at −70°C
Op den Kamp 201578	Percutaneous biopsy (local anaesthesia)	N/A	NR	Immediately frozen, stored at −70°C
Op den Kamp 201258	Percutaneous biopsy (local anaesthesia)	N/A	NR	Immediately frozen, stored at −80°C
Op den Kamp 201380	Percutaneous biopsy (local anaesthesia)	N/A	NR	Frozen by immersion in isopentane, stored in −80°C
Pessina 201051	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
Phillips 201381	Percutaneous biopsy (local anaesthesia)	N/A	NR	Immunoblotting in fresh tissue
Prokopchuk 201652	NR	NR	NR	Immediately frozen and stored at −80°C
Puig‐Vilanova 201482	Open muscle biopsy technique	N/A	NR	NR
Ramage 201853	NR	NR	NR	Immediately frozen, stored at −80°C
Rhoads 200954	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
Schmitt 200755	NR	NR	NR	Immediately frozen in liquid nitrogen, storage temperature NR
Shaw 199169	NR	NR	NR	Snap‐frozen in liquid nitrogen, thawed after 48 h
Skorokhod 201256	Initial phase of surgery	NR	NR	Immediately frozen in liquid nitrogen, storage temperature NR
Smith 201157	Initial phase of surgery	No	NR	Immediately frozen and stored at −80°C
Stephens 201158	Initial phase of surgery	NR	NR	Fixation for microscopy
Stephens 201070	Rectus abdominis–NR Quadriceps–percutaneous biopsy (local anaesthesia)	NR	NR	Immediately frozen
Stephens 201559	Initial phase of surgery	NR	Yes (blood)	Immediately frozen, stored at −80°C
Stretch 201323	Initial phase of surgery	No	NR	Immediately frozen and stored in liquid nitrogen
Sun 201260	Initial phase of surgery	NR	NR	Immediately frozen and stored at −80°C
Taskin 201461	NR	NR	NR	Transferred to lab on ice cold buffer, stored at −20°C
Weber 200783	Percutaneous biopsy (local anaesthesia)	N/A	NR	Immediately frozen, stored at −80°C
Weber 200984	Percutaneous biopsy (local anaesthesia)	N/A	NR	Immediately frozen, stored at −70°C
Williams 201285	Percutaneous biopsy (local anaesthesia)	N/A	NR	NR
Williams 199962	Initial phase of surgery	No	NR	Immediately frozen, stored at −70°C
Zampieri 201066	Rectus abdominis–NR Quadriceps–Percutaneous biopsy (local anaesthesia)	NR	NR	Immediately frozen and stored in liquid nitrogen
Zampieri 200965	Rectus abdominis–NR Quadriceps–Percutaneous biopsy (local anaesthesia)	NR	NR	Immediately frozen and stored in liquid nitrogen
Zampieri 201064	Rectus abdominis–NR Quadriceps–Percutaneous biopsy (local anaesthesia)	NR	NR	Immediately frozen and stored in liquid nitrogen
Zeiderman 199163	Initial phase of surgery	NR	NR	Incubation

N/A: not applicable.

Information provided on biopsy manipulation was limited and mainly focused on freezing and storage procedures. In 43/59 studies, immediate freezing in liquid nitrogen was reported. In only one study was it explicitly stated that freezing was done in the operating room vs. a laboratory facility. The most common temperatures for sample storage were between −70°C and −80°C; storage details were not mentioned in 11/59 studies. Details on time between biopsy and transportation to laboratory facilities and waiting periods were not reported in any study.

Rectus abdominis biological characterization

Study population

Demographics and clinical data from 190 patients are provided in Table 3. Nearly all patients (97%) who were approached consented to intraoperative biopsy, as this entails little, if any, incremental discomfort as the surgery is inherently invasive. Therefore, there was no selection bias inherent in the cohort. Typical of hepato‐pancreatic‐biliary case load, 88% of cancers were gastrointestinal, with the largest proportions being colorectal and pancreatic cancer. Surgical procedures included hepatectomy, liver metastasectomy, pancreatectomy, Whipple procedure, bile duct resection, cholecystectomy, colectomy, and gastrectomy. Metastasis was present in 50% of the patients. Most of the patients were naïve to chemotherapeutic agents, 23% had exposure to chemotherapy within 2 to 4 weeks prior to the surgical procedure. The majority of patients were classified as overweight. Diabetes type II and hypertension were the most common co‐morbidities. Most commonly used medications reported among the population were analgesics, anti‐inflammatory, statins, glucose‐lowering drugs, anti‐hypertensives, anti‐reflux, and thyroid hormone replacement (Table 4).

Table 3.

Patient characteristics

	Male (n = 122)	Female (n = 68)	P values
Age, mean years ± SD (Min–Max)	61 ± 12 (19–87)	65 ± 12 (21–87)	0.049
Tumour type, % (n)			0.006
Colorectal	45 (55)	26 (18)
Pancreas	23 (28)	31 (21)
Other gastrointestinala	25 (31)	22 (15)
Otherb	6 (8)	20 (14)
Presence of metastasis, % (n)	56 (68)	40 (27)	0.03
Chemotherapy exposure within 4 weeks prior to muscle biopsy, % (n)	23 (28)	22 (15)	0.9
Patients with weight lost, % (n)	56 (14)c	55 (11)d	0.9
Sarcopenia, % (n)	60 (61)e	50 (23)f	0.2
BMI (kg/m2), mean ± SD	27 ± 5	28 ± 7	0.7
BMI classification, % (n)			0.1
Underweight	1 (1)	1 (1)
Normal	26 (32)	26 (18)
Overweight	39 (48)	28 (19)
Obesity I	17 (21)	6 (4)
Obesity II	5 (6)	10 (7)
Obesity III	2 (2)	4 (3)
Missing BMI	10 (12)	24 (16)
Co‐morbidities, % (n)
Diabetes type II	12 (15)	18 (12)	0.3
Hypertension	24 (29)	29 (20)	0.4
Cardiovascular disease	15 (18)	7 (5)	0.1
Dyslipidemia	7 (9)	7 (5)	0.9
History of smoking habit, % (n)	28 (34)	24 (16)	0.3
Computed tomography, body composition analysis, mean ± SD
Subcutaneous adipose tissue (cm2)	166.4 ± 91.5g	251.1 ± 134.4h	<0.001
Visceral adipose tissue (cm2)	174.8 ± 105.1g	111.9 ± 65.7h	<0.001
Muscle biopsy triglyceride content (μg/mg), mean ± SD	13.2 ± 14.8i	29.5 ± 21.7j	<0.001

Differences between men and women were analysed by independent t‐test (continuous variables) and χ² test (categorical variables). BMI, body mass index.

^aOther gastrointestinal: stomach, small intestine, liver, intrahepatic bile duct, gallbladder, biliary tract, and appendix.

^bOther: adrenal gland, skin, kidney, mesothelium, lymphoma, melanoma, chronic lymphocytic leukemia, prostate, ovary, uterus, head, and neck.

^cPatients with weight loss information: n = 25.

^dPatients with weight loss information: n = 20.

^ePatients with sarcopenia information: n = 102.

^fPatients with sarcopenia information: n = 46.

^gCT adipose tissue information: n = 98.

^hCT adipose tissue information: n = 44.

ⁱPatients with muscle biopsy triglyceride content: n = 69.

^jPatients with muscle biopsy triglyceride content: n = 19.

Table 4.

Most common medications prescribed and potential effects on skeletal muscle

Class of drug	Medication	% (n)	Common use	Possible implications to skeletal muscle
Cyclooxygenase inhibitors	Aspirin and acetaminophen	15 (29)	Pain, fever, inflammation, and prevention of cardiovascular disease	Influence muscle prostaglandin synthesis, muscle protein metabolism, and cellular processes regulating muscle protein synthesis90, 91, 92, 93
HMG‐CoA reductase inhibitors	Rosuvastatin, simvastatin, and atorvastatin	13 (24)	Lipid lowering	Association with myalgia and related symptoms. Associated to mitochondrial oxidative stress94, 95
Biguanide	Metformin	8 (16)	Type 2 diabetes, suppressor of hepatic gluconeogenesis	Mitochondrial dysfunction in skeletal muscle. Sensitizes muscle to insulin; increases glucose disposal in skeletal muscle95, 96, 97, 98
Proton pump inhibitors	Omeprazole and pantoprazole	8 (16)	Gastroesophageal reflux and erosive esophagitis	Concomitant administration with atorvastatin and dexamethasone is associated to increase risk of myopathy99
Hormones	Levothyroxine	7 (13)	Thyroid hormone (T4) deficiency	Influences myogenesis, associated with sarcopenia and myopathy15, 100
Angiotensin converting enzyme inhibitor	Ramipril	7 (13)	Hypertension and congestive heart failure	Associated with larger muscle cross sectional area and muscle remodeling, associated with cancer cachexia99, 100, 101, 102, 103, 104
Thiazide diuretic	Hydrochloro‐thiazide	6 (12)	Hypertension and diuretic by reducing sodium reabsorption	None reported or reviewed
Calcium channel blockers	Amlodipine	5 (9)	Hypertension and calcium channel blocker	None reported and reviewed105
Opioid	Oxycodone	3 (5)	Pain	Hypogonadism and testosterone depletion in men106
Alpha‐adrenergic blocker	Tamsulosin	3 (5)	Muscle relaxer of prostate and bladder	None reported or reviewed
Xanthine oxidase inhibitor	Allopurinol	3 (5)	Gout prevention and decrease blood uric acid levels	Prevents skeletal muscle atrophy107
Anticoagulant	Warfarin	3 (5)	Anticoagulant	None reported or reviewed

Percentage of patients prescribed this medication out of a total of 190 patients who had a medical history available with information provided on current medication use.

Computed tomography image analysis

Muscle L3‐CSA, SMI, and muscle radiodensity of rectus abdominis and total muscle are shown in Table 5. Sarcopenia was present in 56% of the patients, 60% (n = 97) of men and 49% (n = 42) of women. Weight history was available for 45 patients. Fifty‐six per cent of patients experienced weight loss (11 ± 12% in 5 ± 12 months), and 60% of weight‐losing patients were sarcopenic. Out of 44% (n = 20) weight stable patients, 70% were sarcopenic.

Table 5.

Computed tomography defined muscle composition at L3 for rectus abdominis and total skeletal muscle in cancer patients, stratified by sex and age decade

Sex	Age stratum	N	Rectus abdominis	Total lumbar muscle	Lumbar skeletal muscle index	Rectus abdominis	Total lumbar muscle
			L3‐CSA (cm2)		cm2/m2	Radiodensity	(Hounsfield units)
Male	<50	17	15.9 ± 3.8 (9.8–23.4)	188.7 ± 29.1 (123.6–238.2)	58.2 ± 8.9 (42.8–73.3)	36.2 ± 12.3 (7.6–54.8)	39.6 ± 10.5 (15.4–55.3)
	50–60	34	13.6 ± 3.9 (6.6–24.5)	156.2 ± 27.5 (107.2–228.9)	50.6 ± 8.2 (37.1–66.5)	30.9 ± 12.2 (4.4–50.0)	36.5 ± 8.9 (13.8–50.5)
	60–70	23	13.3 ± 3.3 (5.7–19.4)	158.4 ± 20.7 (109.0–192.5)	50.8 ± 6.6 (36.4–60.8)	28.0 ± 12.3 (−10.8–44.3)	33.8 ± 10.1 (7.1–54.4)
	70–80	23	11.5 ± 2.6 (6.0–17.6)	141.4 ± 23.0 (94.6–187.2)	46.6 ± 6.0 (35.6–59.1)	20.0 ± 11.3 (−2.0–44.6)	28.9 ± 7.7 (10.0–42.6)
	>80	4	9.8 ± 4.2 (6.2–15.2)	139.0 ± 16.4 (122.8–160.9)	46.1 ± 7.1 (40.1–56.3)	21.5 ± 8.3 (12.3–30.4)	27.5 ± 3.0 (24.8–31.5)
Female	<50	3	9.3 ± 3.2 (5.9–12.2)	114.9 ± 14.8 (97.8–124.4)	43.8 ± 1.6 (42.9–45.7)	32.0 ± 5.7 (26.6–38.0)	45.1 ± 5.3 (40.5–50.9)
	50–60	11	7.0 ± 2.4 (3.8–10.9)	101.5 ± 16.8 (67.5–125.4)	38.3 ± 6.8 (23.9–46.4)	22.7 ± 13 (4.2–41.1)	35.4 ± 7.6 (20.9–46.1)
	60–70	15	8.7 ± 3.7 (2.8–16.9)	102 ± 16.6 (66.2–122.7)	39.2 ± 7.0 (27.7–52.8)	19.1 ± 10.3 (2.5–39.1)	29.0 ± 7.1 (18.2–39.6)
	70–80	16	6.7 ± 2.3 (1.4–10.9)	101.0 ± 13.8 (79.0–127.3)	40.5 ± 4.8 (33.8–49.7)	13.1 ± 10.0 (−7.7–30.9)	28.9 ± 7.0 (15.0–38.9)
	>80	3	7.7 ± 3.1 (4.2–10.0)	92.8 ± 14.8 (77.9–107.5)	41.1 ± 8.1 (32.9–49.1)	12.2 ± 19.8 (−10.1–27.6)	22.9 ± 4.1 (18.2−25.3)
Total male		101	13.6 ± 3.8 (5.7–24.5)	158.2 ± 29 (94.6–238.2)	50.8 ± 8.3 (35.6–73.3)	28.2 ± 12.9 (−10.8–54.8)	34.3 ± 9.7 (7.1–55.3)
Total female		48	7.6 ± 2.9 (1.4–16.9)	101.7 ± 15.4 (66.2–127.3)	39.8 ± 6 (23.9–52.8)	18.2 ± 12 (−10.1–41.1)	31 ± 8.3 (15–50.9)

Values reported in mean ± SD (range). CSA, cross‐seccional area; L3, 3rd Lumbar vertebra.

Sex differences

In light of the fact that most of the papers in the literature review included samples of mixed sex of varying proportions, we examined all of the biopsy features for sex differences. Sexual dimorphism was prominent in L3‐CSA total lumbar muscle and RA, muscle radiodensity of RA and total muscle (Table 5), mean fibre CSA (Table 6), and in expression of genes associated with muscle growth, apoptosis, and inflammation (Table 7). Proportions of fibre types using both quantification methods, MyHC isoforms and individual fibre types, were not different between male and female patients (Table 6).

Table 6.

Rectus abdominis myosin heavy chain content and mean muscle fibre area of cancer patients

	All	Male	Female	P value
A. MyHc content by electrophoresisa (% ± SD N = 40 M/n = 8 F)
MyHC I (%)	39.3 ± 11.1	39.1 ± 10.3	40.6 ± 15.6	0.73
MyHC IIA (%)	38.4 ± 11.1	37.5 ± 10.0	42.6 ± 15.7	0.24
MyHC IID (%)	22.3 ± 8.9	23.4 ± 8.6	16.8 ± 9.1	0.06
B. MyHc content by immunohistochemistrya (% ± SD N = 20 M/n = 10 F)
MyHC isoforms (%)
MyHC type I	47.1 ± 13.0	47.0 ± 12.6	47.3 ± 14.6	0.91
MyHC type IIA	51.8 ± 13.4	52.4 ± 12.6	50.5 ± 15.6	0.53
MyHC type IID	16.7 ± 14.3	19.2 ± 13.7	11.8 ± 15.1	0.19
All Hybridsb	15.5 ± 13.5	18.5 ± 13.5	9.6 ± 12.2	0.08
Individual fibre types (%)
Fibre type I	46.4 ± 12.9	48.9 ± 9.4	46.2 ± 14.2	0.32
Fibre type I/IIA	0.7 ± 1.0	0.6 ± 0.9	1.2 ± 1.6	0.15
Fibre type IIA	36.1 ± 9.5	35.7 ± 9.4	40.7 ± 9.6	0.71
Fibre type IIA/D	15.0 ± 13.7	13.1 ± 12.4	8.5 ± 12.9	0.39
Fibre type IID	1.8 ± 4.6	1.7 ± 3.7	3.4 ± 7.3	0.32
C. Mean muscle fibre area (μm2) (% ± SD N = 20 M/n = 10 F)
All fibres	3236 ± 1390	3784 ± 1285	2139 ± 854	<0.05
MyHC isoforms (μm2)
MyHC type I	2323 ± 944	2591 ± 970	1786 ± 635	<0.05
MyHC type IIA	4009 ± 1937	4848 ± 1725	2331 ± 1054	<0.05
MyHC type IID	4026 ± 2060	4722 ± 1895	2461 ± 1546	<0.05
Individual fibre types (μm2)
Fibre type I	2325 ± 941	2591 ± 970	1795 ± 633	<0.05
Fibre type I/IIA	2253 ± 1209	2726.6 ± 1181	1306 ± 528	<0.05
Fibre type IIA	3940 ± 1970	4760 ± 1820	2299 ± 1012	<0.05
Fibre type IIA/D	4012 ± 2055	4833.5 ± 1841	2266 ± 1268	<0.05
Fibre type IID	5243 ± 2407	5323 ± 2553	4729 ± 1524	0.75

MyHC: myosin heavy chain.

^aThere were no differences in age, BMI, metastasis, chemotherapy exposure, co‐morbidities, nor smoking history between men and women.

^bAll hybrids refer to fibres of mixed myosin heavy chain isoforms MyHC type I/IIA and MyHC type I.

Table 7.

Skeletal muscle gene expression for genes associated with cancer cachexia in cancer patients^a

Biological function	Gene symbol	Gene name	Agilent transcript ID [Refseq RNA ID]	Female (n = 64)	Male (n = 69)	P value
Atrophy	FOXO1	Forkhead box O1	A_24_P22079	1.53 ± 1.04	1.11 ± 0.68	0.005
Autophagy	BECN1	Beclin 1	A_23_P433071 [NM_003766]	0.91 ± 0.27	1.03 ± 0.3	0.05
			A_23_P89410 [NM_003766]	1.00 ± 0.27	1.11 ± 0.33	0.05
	CTSL2	Cathepsin L2	A_23_P146456 [NM_001333]	1.31 ± 0.57	0.99 ± 0.44	<0.0001
Apoptosis	CASP8	Caspase 8	A_23_P209389 [NM_033355]	0.97 ± 0.32	1.09 ± 0.38	0.08
	CASP9	Caspase 9	A_23_P97309 [NM_001229]	0.95 ± 0.19	1.06 ± 0.25	0.008
			A_24_P111342 [NM_001229]	0.97 ± 0.22	1.08 ± 0.31	0.03
Muscle growth	AKT1	V‐Akt murine thymoma viral oncogene homolog 1	A_23_P2960 [NM_005163]	1.23 ± 0.52	1.04 ± 0.35	0.03
	DMD	Dystrophin	A_24_P342388 [NM_004019]	1.34 ± 0.67	0.94 ± 0.29	<0.0001
			A_24_P185854 [NM_004010]	1.11 ± 0.27	0.94 ± 0.23	<0.0001
			A_24_P34186 [NM_004010]	1.19 ± 0.55	0.97 ± 0.39	0.01
			A_32_P199796 [NM_004023]	1.27 ± 0.66	0.98 ± 0.42	0.005
	MSTN	Myostatin	A_23_P165727 [NM_005259]	1.71 ± 2.43	2.74 ± 3.74	0.02
	PAX7	Paired box 7	A_23_P126225 [NM_013945]	0.99 ± 0.49	1.08 ± 0.39	0.05
			A_23_P500985 [NM_013945]	0.96 ± 0.45	1.03 ± 0.33	0.09
	PPARGC1A	Peroxisome proliferator‐activated receptor gamma, coactivator 1 alpha	A_24_P303052 [NM_013261]	1.22 ± 0.77	1.00 ± 0.51	0.07
	SMAD3	SMAD family member 3	A_23_P48936 [NM_005902]	1.14 ± 0.42	1.00 ± 0.28	0.07
	TGFB1	Transforming growth factor, beta 1	A_24_P79054 [NM_000660]	1.42 ± 1.47	1.06 ± 0.54	0.01
Inflammation	JAK1	Janus kinase 1	A_24_P410678 [NM_002227]	0.92 ± 0.37	1.15 ± 0.43	0.001
	JAK2	Janus kinase 2	A_23_P123608 [NM_004972]	1.21 ± 0.48	1.06 ± 0.45	0.03
	JAK3	Janus kinase 3	A_23_P329112 [NM_000215]	1.03 ± 0.46	1.19 ± 0.57	0.09
	STAT3	Signal transducer and activator of transcription 3	A_23_P107206[NM_213662]	1.21 ± 1.02	0.53 ± 0.35	0.02
	STAT5A	Signal transducer and activator of transcription 5A	A_23_P207367 [NM_003152]	1.12 ± 1.01	0.32 ± 0.34	0.03
			A_24_P173088 [NM_003152]	1.19 ± 1.00	0.47 ± 0.45	0.005
	TNF	Tumor necrosis factor	A_24_P50759 [NM_000594]	0.99 ± 0.35	1.15 ± 0.44	0.03

Values (unitless) reported as mean ± standard deviation.

^aCancer type (0.003) and metastasis presence (0.002) were different between men and women. There were no differences in age, BMI, chemotherapy exposure, co‐morbidities, nor smoking history between men and women.

For centralized nuclei assessment, the mean % of fibres with centralized nuclei was 12 ± 9% (4 to 36%) and 10 ± 8% (3 to 27%) in men and women, respectively. No differences were found between men and women (p=0.39) with a combined mean value of 11 ± 8%.

Rectus abdominis: proportion of fibre types and muscle fibre area

Electrophoretic analysis of MyHC isoforms confirmed MyHC I and MyCH IIA to be present at similar proportions, while MyHC IID was less abundant (Table 6A). MyHC type IIA was the most abundant isotype, followed by MyHC type I and IID (Table 6B). In addition, 15.5% of the fibres were identified as hybrids, which is the sum of MyHC type I/IIA and IIA/D. For individual fibre types, type I fibres comprised the greatest proportion (46.4%) followed by fibre type IIA (36.1%) and hybrid type IIA/D (15%). Presence of fibre type IID, as well as hybrid type I/IIA, was minimal (1.8% and 0.7%). Mean muscle fibre area (μm²) was calculated by the detection of membrane (laminin/dystrophin antibody) fluorescence on 1069 ± 771 muscle fibres per biopsy (Table 6C). Mean muscle fibre area was determined in total and per fibre type, which includes collective results of MyHC isoforms and individual fibre types (Table 6C). Mean fibre area of MyHC type I was smaller than MyHC type IIA and IID. For individual fibre types, type I and type I/IIA were smaller compared with type IIA, IIA/D, and IID. Type IID had the largest mean fibre area compared with the other individual fibre types.

Age effects

Comparison of older (74 ± 4 years, n = 13) and younger (50 ± 6 years, n = 13) men revealed no differences between groups with respect to mean muscle fibre area (total, individual fibre types, or MyHC isoforms), % of individual fibre types, or % of MyHC isoforms. Age effect was evaluated in men (n = 26) by comparing mean values of a younger group vs. an older group. No significant differences were found in relation to % of MyHC isoform content.

Skeletal muscle gene expression for genes associated with cancer cachexia

Differences in genes encoding proteins commonly explored in cancer‐muscle wasting are summarized in Table 7 (also see Supporting Information, Table S3). Atrophy, autophagy, apoptosis, muscle growth, and inflammation genes were selected based on reviewed literature on muscle atrophy in cancer.22, 40, 41, 47, 50, 70, 87, 88, 108 Sexual dimorphism exists in pathways related to skeletal muscle anabolism and catabolism illustrating the need for caution when generalizing results from only one sex or discussing results from a mixed group of cancer patients.

Discussion

There is a perceived need to understand the human biology of cancer‐associated muscle atrophy and to frame it in the context of our larger understanding of experimental findings.6, 22, 109, 110, 111 The emergent literature on human muscle biopsies has been generated with that intent but has a number of substantial limitations within the study design as well as procedures for collection and preparation of the biopsy material. At the same time, there is substantial opportunity for collaboration between cancer surgeons and researchers to obtain intraoperative biopsies with a high rate of patient consent and the additional capability to describe the muscles of these patients with precise radiological metrics. Agreement to a set of standardized procedures and reporting will enhance the consistency, reliability, and comparability of future research in this area. Evaluation of human rectus abdominis muscle presents the expected variation in several measures that may be of interest for emerging studies in this area.

Study quality and design

The quality of the studies reporting on biopsy material to characterize varying features of muscle biology was uniformly low. Quality assessment tools revealed several inconsistencies in sample selection strategies, study design, data collection, and analysis in the existing literature. Bias assessment of sample selection exposed a clear absence of sample representativeness in 59% of studies and lack of sample size justification in 96% of studies. In 75% of the studies reviewed, samples from a relatively small number of participants (n = ≤30) were evaluated without accounting for age or sex variation.

The majority of published studies use weight loss (vs. weight stability) to define cachexia. This approach is limited by not accounting for the characteristics of muscle (muscle mass or change in muscle over time), which are the clinically relevant features related to cancer outcomes. Indeed, weight stable patients may well be losing muscle over time112 and they can also be profoundly sarcopenic.2, 27 Weight loss was the most commonly used criteria for cancer cachexia assessment; however, application of this measure alone poses major concerns in misclassification and unintended exclusion of cachectic patients. Many studies were published prior to the widespread use of CT images to quantify muscle, as well as prior to the publication of the international cachexia consensus, which defines muscle mass as a diagnostic criterion for cachexia.35, 36, 39, 40, 41, 42, 45, 47, 48, 50, 54, 55, 57, 63, 64, 65, 66, 69, 70, 83, 84, 86, 87, 88, 108 The premise of using weight loss when muscle is being evaluated is erroneous. Muscle wasting can be experienced by patients with less than 5% weight loss.112 Also, the arbitrary selection of weight loss percentage and time frame in different studies complicates the comparison of results between studies. In the cohort of patients we evaluated, 70% of weight stable patients and 60% of weight‐losing patients were sarcopenic. Therefore, assessment of muscle mass is essential, and this can be easily achieved through the secondary analysis of CT images used to plan the surgery.18, 19, 29

Some authors reported mortality‐defined cutpoints to define sarcopenia according to age and sex of a reference population27, 113 and these have been secondarily used by other authors.114 Caution should be used in applying these cutpoints to define sarcopenia in patients undergoing muscle biopsy, and these may not necessarily reflect the population from which biopsies are evaluated.114 Here, we suggest to use CT to quantify muscle features for the overall population from which the biopsy sampling is done. In this way, patients providing biopsy for our study are clearly representative of the entire L3 SMI distribution of our regional population (Alberta, Canada) (Figure 2). This representation eliminates the possibility of sampling bias. It also allows each patients' SMI to be ranked within the population distribution overall as well as compared with values available for healthy young individuals.115

Age and sex differences exist at the level of muscle function, biochemistry/metabolism, and mass.14, 17, 116 The majority of studies reported combined data from both sexes without acknowledging sexual dimorphisms. Age was generally not accounted for. In the first 40 years of life, muscle mass is relatively stable in both men and women, and then it begins to decline; however, the rate of loss is slower in women than in men.62 In our sample, differences between men and women were observed for muscle fibre area, SMI, and muscle radiodensity. Sexual dimorphism in gene expression was not limited to a particular pathway or function but was identified in growth (AKT1, FOXO1, MSTN, PAX7, and TGFα1), apoptosis (CASP9), and inflammation (TNF and STAT3). In relation to the age effect, we did not find any significant differences in mean muscle fibre area and proportion of fibre types when comparing young vs. old male cancer patients; this could be potentially explained by the narrow age range in our study. Differences between young (18 to 48 years) and older (66 to 99 years) participants117 have been reported for fibre type distribution in rectus abdominis and vastus lateralis. Therefore, age differences and sexual dimorphism must be acknowledged when comparing, reporting, and interpreting muscle characteristics.

Here, we present many characteristics of human rectus abdominis muscle. We obtained a detailed analysis of its radiological features, for the first time. Our analysis of fibre type is multidimensional and confirms the mixed fibre distribution of the rectus abdominis. A prior study in cancer patients with upper gastrointestinal malignancies reported mean values of 48% and 55% for MyHC type I and IIa, respectively46. Muscle gene expression and TG content levels as presented here are new information about rectus abdominis. Future work on rectus abdominis can be usefully planned, using this base of information. The majority of evidence to date (Table 1) on muscle from cancer patients is coming from rectus abdominis. Due to the unique characteristics of each muscle type, we suggest that future researchers identify candidate muscles for intensive research using the principle that the muscle(s) most often transected in cancer surgeries would be the greatest resource. This can be decided in function of the common surgical approaches. Thus, over time, a large base of evidence may be obtained from latissimus dorsi, serratus anterior, or intercostal muscle (e.g.) from thoracic cancer surgeries.

A key component of case‐control studies is to provide details of the control group relative to the research question. However, this is rarely done in the literature that we reviewed.20, 21 Detailed clinical characterization of non‐cancer controls is usually missing, and assumption of a healthier status of the control group when compared with cancer patients is common. In many cases, the comparator group is a non‐cancer surgical patient population; however, there is no documentation provided around diagnosis or medications. Presumably, healthy volunteers could have underlying co‐morbid conditions or be taking medications that impact skeletal muscle. Co‐morbidities and use of medications were not generally mentioned either for patients undergoing non‐cancer surgery or ‘healthy' volunteers recruited outside the clinical setting. Approximately 60% of people diagnosed with malignancy are 65 years and older.13 Prevalence of co‐morbidity in cancer population ranges from 30% to 50% depending on type of cancer19 and a patient with history of cancer has on average three co‐morbidities.118, 119 Diabetes and hypertension were the most common conditions in our patient population, but cardiovascular disorders and mental health problems are also prevalent in the cancer population.13, 19 These chronic conditions and medications taken to control them can independently affect muscle physiology15, 106, 120, 121, 122, 123, 124, 125, 126, 127, 128 (Table 4). COX inhibitors, statins, biguanides, proton pump inhibitors, and thyroid hormones were the most common medications prescribed in our patient population apart from those prescribed during cancer treatment. These classes of drugs have known effects on muscle protein synthesis90, 91, 92, 129 and catabolism,130, 131, 132, 133 atrophy pathways,134 insulin sensitivity,96 and mitochondria function.97 Therefore, it is important that for both the cancer group and ‘control' groups have a detailed medical history that captures diagnosis of other conditions and medications. In addition to drugs prescribed for management of co‐morbid conditions, antineoplastic treatment previous to tissue biopsy is also a relevant event that may impact interpretation of results as the long‐lasting effects in the muscle are unknown.135

Technical considerations

We suggest recommendations for minimum procedures to follow in biobanking practices, tissue manipulation, and patient characterization to enhance the consistency, reliability, and comparability of future research (Table 8). Acknowledgement of differences between muscle groups is essential when comparing and interpreting results. RA is commonly collected in patients with gastrointestinal disease due to its practicality in relation to the surgical incision while maintaining patient burden to the essential minimum. Its broad extension in the abdominal area enables for collection of muscle tissue from a variety of locations136; however, no one has demonstrated how homogeneous the RA is in relation to the biopsy site. On the other hand, quadriceps or tibialis anterior are collected in healthy volunteers serving as controls as there is no justification for surgical intervention. Importantly, physiological variations between muscle groups exist,137, 138 which strongly suggest that studies collecting different muscles must avoid comparing or combining data of more than one muscle.

Table 8.

Summary of recommendations for muscle biopsy processing and population characterization

(A) Biobanking protocols and tissue manipulation · For intraoperative muscle biopsies, collect at the start of the surgical procedure and avoid cauterization. · Avoid or report the use of foreign substances (e.g. use of saline‐moistened gauzes). · Report waiting periods between surgical/needle removal, transportation to other facilities, and freezing; include the use or not of crushed ice during the waiting process. · Report any removal of blood traces or unrelated tissue from the muscle biopsy. · If muscle is ‘immediately frozen,' clarify the location, time, and other relevant details (e.g. RNA stabilizer solution) of this action after the surgical removal. · Sample storage recommended ≤−70°C; however, the temperature selection will depend on the molecules of interest and/or experimental techniques.

(B) Cancer population characterization · Clearly state the patient selection method and possible limitations. · Report information on metastatic status or tumour classification. · Report co‐morbidities and medications. · Report past or current exposure of antineoplastic treatments.

(C) Inclusion of control groups · Provide a clear characterization of the control group. · Report co‐morbidities and medications. · Match age and sex with study population. Provide justification for case‐matching criteria. · Collect same muscle in control and study populations.

(D) Classification and results · Avoid mixing the results of two or more muscle groups or comparing one muscle group with a different muscle group (e.g. rectus abdominis vs. quadriceps). · Acknowledge sexual dimorphism in skeletal muscle by reporting results based in men and women, include mean and standard deviation. · Classification of cancer cachexia should include both, body composition analysis (muscle mass values) and weight loss.

Most researchers did not report on surgical procedures and muscle biopsy collection, transport, and processing of the samples, each of which can impact on the morphological and molecular profile of the biopsy.10, 139, 140 Collecting abdominal muscle biopsies at the start of the surgical procedure and avoidance of electrocautery is strongly recommended to reduce variations associated with the surgical trauma, variable duration of surgery, and intraoperative effect of anaesthetics.10, 11, 141, 142, 143, 144 Skeletal muscle collected at the start and end of a surgery expresses differences in genes associated with inflammation, growth differentiation, and transcription factors.142 For percutaneous biopsies, the Bergstrom protocol is a well‐developed method with several adjustments to improve the quality of the muscle biopsies.145, 146 Procedures followed after biopsy collection must also be detailed as sample preservation and storage impacts on muscle integrity and potentially interpretation of the results. Lastly, the numbers of medical conditions and drugs taken by patients in this sample are important and all of these and their different combinations may have an impact on specific aspects of muscle biology. As much as possible, we recommend to annotate the presence of co‐morbidities and medications in patients consenting to biopsy.

Overall, the literature review reveals a high risk of sampling bias and poorly characterized patient populations. These features make reliable comparison between studies and aggregation of data challenging. Muscle biopsy preparation and biobanking practices are also variable between studies. Data from an unbiased sample of 190 patients present a variety of measures of interest on rectus abdominis to provide a point of reference for researchers exploring biological characteristics of this muscle. Continued collaboration between researchers and cancer surgeons would enable a more complete understanding of mechanisms of cancer‐associated muscle atrophy.

Author contributions

Ms. Anoveros‐Barrera and Mr. Bhullar, who each contributed equally to data analyses, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. A.A. and A.S.B. contributed to conceptualization, design, analysis, writing, and interpretation. C.S. contributed to the gene array data analysis and interpretation. N.E. contributed with data collection and analysis. A.R.D. contributed with CT image analysis and experimental optimization. K.J.B.M. contributed to experimental optimization and image analysis. D.B., T.M., R.G.K., and O.F.B. contributed in patient recruitment, biopsy, and clinical data collection. S.D., R.J.S., and C.T.P. contributed interpretation and editing. V.C.M. and V.E.B contributed to conceptualization, design, analysis, interpretation, and editing. All authors of this research paper have approved the final version submitted.

Funding

Funding for this project is from the Canadian Institutes of Health Research (CIHR) operating grants awarded to V.C.M. and V.E.B. A.A. is a recipient of the Doctoral Fellowship from the Consejo Nacional de Ciencia y Tecnología (CONACyT) of Mexico. A.S.B. is the holder of the Alberta Innovates Technology Futures (AITF) award at University of Alberta, Canada. Images for immunofluorescence experiments were conducted at the facilities of the Cell Imaging Centre, Faculty of Medicine and Dentistry, University of Alberta, Canada.

Conflict of interest

No authors declare a conflict of interest.

Supporting information

Table S1. Antibody information used for immunofluorescence experiments: muscle fiber types, laminin/dystrophin and nuclear stain.

Table S2. Complete extraction table of the reviewed articles in relevance of muscle biopsy collection in cancer patients

Table S3. Skeletal muscle gene expression for genes associated with cancer cachexia in cancer patients

Anoveros‐Barrera A., Bhullar A. S., Stretch C., Esfandiari N., Dunichand‐Hoedl A. R., Martins K. J. B., Bigam D., Khadaroo R. G., McMullen T., Bathe O. F., Damaraju S., Skipworth R. J., Putman C. T., Baracos V. E., and Mazurak V. C. (2019) Clinical and biological characterization of skeletal muscle tissue biopsies of surgical cancer patients, Journal of Cachexia, Sarcopenia and Muscle, 10, 1356–1377. 10.1002/jcsm.12466.