Skeletal muscle omics signatures in cancer cachexia: perspectives and opportunities

Abstract

Cachexia is a life-threatening complication of cancer that occurs in up to 80% of patients with advanced cancer. Cachexia reflects the systemic consequences of cancer and prominently features unintended weight loss and skeletal muscle wasting. Cachexia impairs cancer treatment tolerance, lowers quality of life, and contributes to cancer-related mortality. Effective treatments for cancer cachexia are lacking despite decades of research. High-throughput omics technologies are increasingly implemented in many fields including cancer cachexia to stimulate discovery of disease biology and inform therapy choice. In this paper, we present selected applications of omics technologies as tools to study skeletal muscle alterations in cancer cachexia. We discuss how comprehensive, omics-derived molecular profiles were used to discern muscle loss in cancer cachexia compared with other muscle-wasting conditions, to distinguish cancer cachexia from treatment-related muscle alterations, and to reveal severity-specific mechanisms during the progression of cancer cachexia from early toward severe disease.

Cancer-associated cachexia is a complex disorder featuring weight loss, skeletal muscle atrophy with or without fat depletion, and progressive deterioration of physical function (1,2). It is driven by a combination of systemic inflammation, protein hypercatabolism, and disrupted energy balance from tumor-induced changes in endocrine, immune, nervous, musculoskeletal, and cardiac systems among others (3-5). Although multiple organ systems associate with cancer cachexia, skeletal muscle wasting is a hallmark feature of the disorder (3). Despite the multifactorial nature of cancer cachexia, routine diagnosis in the clinic is based on weight loss and/or body composition, such as more than 5% weight loss in 6 months or more than 2% weight loss plus body mass index (BMI) less than 20 kg/m². Stages of cachexia severity and progression were proposed by international consensus to include early or precachexia, cachexia, and refractory cachexia that reflects severe disease unlikely to respond to treatment (2). Mechanisms before cachexia appears are relevant for early detection and prognostication (6,7), and mechanisms driving disease progression are potentially important for devising therapies to slow decline toward severe disease (8).

Molecular maps of disease mechanisms are enabled by numerous omics technologies (9-12). These omics technologies probe mechanisms in an unbiased manner at distinct levels of organization from genome, exome, and epigenome to the transcriptome, proteome, and metabolome (13,14). Each offers unique but complementary information on cell or tissue state. Whole-genome sequencing or whole-exome sequencing by high-throughput DNA sequencing can sequence all genomic regions or only protein-coding exons, respectively, for detection of disease-linked genomic variants (15,16). Parallel transcriptome analysis by RNA sequencing can reveal how RNA species are impacted by variations in the DNA sequence, such as overexpression, gene fusions, or splicing events caused by somatic variants in cancer (17). Proteomes and metabolomes of biological samples on the other hand are interrogated by mass spectrometry or nuclear magnetic resonance spectroscopy and considered a closer reflection of disease phenotypes in vivo (18,19). Large-scale, comprehensive molecular profiles recently achieved by integrating these multiple omics platforms have led to new discoveries of disease biology and potential information for clinical decision making, as shown in the proteogenomics of lung and pancreatic cancer by the Clinical Proteomic Tumor Analysis Consortium at the US National Cancer Institute (9,20,21).

Similar to other disciplines, omics approaches are being used to study cancer cachexia at a greater scale. The emergence of these approaches are reflected in recent scientific literature. Search of the PubMed database showed steadily increasing output for cancer cachexia in relation to transcriptome, proteome, metabolome, and microbiome since 2016, with noted rise in 2021. This trend is likely to be amplified further in the near term when factoring the systemic, multifactorial nature of cancer cachexia. Multiple organ systems may need to be mapped accordingly by the different omics platforms. Furthermore, primary malignancy, cancer progression, severity of inflammation, nutritional status, and anticancer therapies influence cancer cachexia incidence and progression (2,3,22,23). Omics-based discovery of biomarkers and treatment targets should consider not only different tissue types but these factors as well. Additional noted approaches include omics profiling to distinguish skeletal muscle alterations in cancer cachexia from other muscle-wasting conditions (24) and from cancer treatments that independently compromise muscle function (25). To this end, there is a great need for multi-omics, system-wide, multitimepoint analyses that integrate this information to more comprehensively characterize cancer cachexia. This would help shape clinical interventions by enabling possible early intervention and treatment strategies based on cancer type and treatment regimen.

In this paper, we appraise investigations that used high-throughput omics analysis of skeletal muscle in cancer cachexia relative to 1) age and cancer treatment–related muscle alterations and 2) cancer cachexia severity and progression. The aim was to objectively examine by way of high-throughput technologies the molecular alterations in skeletal muscle during cancer cachexia and recognize potential confounding influences from cancer treatment, age, and disease state. Finally, to better address interactions of biological entities across multiple omics layers, we performed a secondary analysis to examine proteome-wide changes in mouse skeletal muscle during the progression from moderate to severe cancer cachexia. This progression offers a unique time frame relevant for potential slowing of cachexia. Taken together, we offer new avenues of exploration and implications for biomarkers and treatment targets.

Skeletal muscle omics signatures in cancer cachexia

Experimental murine models of cancer cachexia

Many experimental murine models of cancer cachexia entail subcutaneous or intramuscularly injected tumor homografts of carcinogen-induced tumors engrafted into immunocompetent syngeneic hosts. Widely used models discussed herein include Lewis lung carcinoma (LLC) and C26 colorectal carcinoma (C26). LLC is a commercially available cell line that was established from the lung of a C57BL mouse bearing a tumor resulting from implantation of primary, spontaneously occurring LLC (26). LLC-engrafted tumors induce rapid weight loss and anorexia 2 weeks postimplant when tumor burden is approximately 6% of animal body weight (27,28). C26 model is a murine colon carcinoma cell line derived from a tumor induced by rectal application of N-nitroso-N-methylurethane in BALB/c mice. Similar to LLC tumors, C26 tumors account for approximately 6% of body weight at 2 weeks postinjection and induce substantial carcass weight loss (29). Extent of body weight reduction, induction of a systemic cachexia-like response, and composition of the tumors depend on the site of tumor engraftment (30). These experimental models share similar limitations to other syngeneic and xenograft tumor models as the initiation and progression of cancer does not reflect the natural tumor progression in humans. Similarly, the progression of cancer cachexia is more rapid in these models. The need for additional experimental models that more closely reflect disease evolution in humans is discussed below.

Cancer cachexia muscle loss vs other types of muscle loss

Muscle atrophy is a key feature in cancer cachexia. Systemic wasting muscle atrophy models include cachexia, fasting, aging, and pharmaceutical induced (eg, dexamethasone administration) (24,31,32). Increasing accessibility of omics platforms has provided an avenue to globally examine whether proteogenomic changes are similar across these different types of muscle atrophy or if changes are specific to the atrophy stimuli. One of the first studies to pursue this line of inquiry was a landmark study from 2004 using microarray-based technology to compare muscle transcript profiles in 4 catabolic states including cancer cachexia, fasting, diabetes, and kidney disease (33). Among the groundbreaking findings were commonly induced atrogenes involved in protein degradation (atrogin-1, muscle RING-finger protein-1 [MuRF-1]) and downregulation of genes associated with energy production (33,34). Building on this seminal work, a recent deep multi-omics study interrogated skeletal muscle response to systemic wasting atrophy models, which included dexamethasone, cancer cachexia, and aging-induced muscle atrophy. Mice were used to simulate dexamethasone- and cancer-induced muscle atrophy and were 6-month-old male C57BL/6J mice, and aging-induce atrophy was assessed in 24-month-old mice of the same background and sex. To generate dexamethasone-induced muscle atrophy, mice were treated with 20 mg/kg per day of dexamethasone for 14 days via intraperitoneal injection. For cancer-induced atrophy, 5 x 10⁵ LLC cells were injected into the right and left flank of mice and tumors allowed to grow for approximately 3 weeks before analysis. RNA sequencing and quantitative proteomics detected more than 15 000 unique messenger RNAs (mRNAs) and more than 5000 proteins, respectively, for each muscle atrophy model (24). Although these approaches have been used on many occasions before, an advantage of this specific study was adherence to a standardized workflow for all 3 muscle atrophy models. This enabled direct comparison of the transcriptome or proteome between aging, dexamethasone, and cancer cachexia (in addition to their individual analysis) because the same analytical methods were applied to each muscle-wasting condition.

Specifically, 124 genes were uniquely upregulated and 131 genes downregulated by cancer cachexia. In contrast, 263 and 942 genes were induced by aging and dexamethasone, respectively, and 9 and 203 transcripts were downregulated. Gene set enrichment analysis showed terms associated with immunity and inflammation were upregulated in all atrophy models, highlighting an overall inflammatory state at the transcript level in these 3 muscle-wasting conditions. Genes categorized as “response to lipoprotein particle” were only upregulated in cancer cachexia, which may indicate biologically significant involvement of dysregulated lipid metabolism in cachexic skeletal muscle. Genes associated with nicotinamide adenine dinucleotide dehydrogenase complex assembly were commonly downregulated across muscle atrophy models, whereas genes associated with cytochrome complex assembly were specifically downregulated in cachexia. These downregulated categories suggest multifaceted mitochondrial structural and/or functional alteration to be shared in transcriptional signatures in muscle atrophy because of aging, corticosteroids, and cancer. It was noteworthy that overall, no individual transcript was regulated in the 3 muscle-wasting conditions. Only a limited number of differentially regulated transcripts were shared in any 2 muscle-wasting types (range of 5-34 transcripts). Therefore, transcriptome-wide changes in skeletal muscle gene expression appear to contain considerable specificity to a given atrophic condition.

Like RNA sequencing, enrichment analysis of the individual muscle-wasting proteomes revealed mitochondrial gene expression as a common signature, adding additional support for altered mitochondrial function as a shared feature in atrophy associated with age, dexamethasone, and cancer cachexia. The proteome analysis also showed overall limited overlap in regulated proteins from these muscle atrophy models. Proteins reported to increase and decrease in the 3 muscle-wasting conditions are listed in Figure 1, A [extracted from (24)]. This list of 16 differentially regulated proteins were input to ShinyGO 0.76 for gene ontology biological process analysis (http://bioinformatics.sdstate.edu/go/) (Figure 1, B). Relationships were evident between the represented terms that mostly pertained to coagulation, hemostasis, and fibrin formation (Figure 1, B), suggesting immunity and inflammation as unifying protein-level features of muscle atrophy from aging, corticosteroids, and cancer cachexia. When only aging and cancer cachexia were compared, 5 upregulated proteins and 2 downregulated proteins were shared [Figure 1, C (24)]. These results suggest distinct protein signatures of muscle wasting in aging vs cancer in mice. Validation of these findings in human samples may imply need of different treatments for age- and cancer-related muscle loss, which potentially co-occur in elderly cancer patients.

Figure 1.

Proteome signatures in aging, dexamethasone, and cancer-associated muscle wasting. Findings shown in panels A-D were reanalyzed from proteomics datasets in Hunt et al. [(24); Supplementary Table S3] and ShinyGO 0.76 enrichment tool (bioinformatics.sdstate.edu/go/) using the gene ontology biological process database. A) TMT-based quantitative proteomic analysis was performed on skeletal muscle from 3 different mouse models of muscle atrophy including aging (sarcopenia), dexamethasone treatment, and cancer cachexia induced by ectopic injection with Lewis lung carcinoma cells. Proteins differentially regulated up and down in the 3 muscle-wasting conditions listed. Graphic made with BioRender.com. B) Pathways (top) and networks (bottom) were derived from proteins regulated in the 3 types of muscle wasting (listed in A). Coagulation, hemostasis, fibrin formation, and wound healing are represented. C) Only 5 proteins increased in both aging and cancer cachexia. Age- and cancer-related muscle wasting in mice may be molecularly distinct. D) Twenty-two upregulated proteins shared in dexamethasone and cancer cachexia were analyzed in silico to obtain enriched pathways (bottom) and networks (top). E) Cancer may alter iron metabolism in skeletal muscle and reduce availability for iron-dependent enzymes of mitochondrial energy-yielding pathways, leading to energy stress and muscle atrophy in cachexia. FDR = false discovery rate; Neg = negative; OXPHOS = oxidative phosphorylation; Pos = positive; Reg. = regulation; TCA = tricarboxylic acid; TMT = tandem mass tag.

When only dexamethasone and cancer cachexia were compared, 22 upregulated and 12 downregulated proteins were shared [not shown; extracted from table S3 in (24)]. We analyzed the 22 upregulated proteins in silico for functional enrichments. Gene ontology of the 22 shared upregulated proteins enriched iron transport and homeostasis, complement pathway, coagulation and hemostasis, and fatty acid oxidation (Figure 1, D), highlighting iron metabolism in addition to inflammation and lipid metabolism in skeletal muscle challenged by tumor load and dexamethasone. Implication of iron metabolism is intriguing based on recent evidence detailing biologically significant modulation of cancer cachexia phenotypes by iron (35). Iron is an essential mineral needed for iron-dependent mitochondrial enzymes involved in energy production. It was reported that weight-losing pancreatic cancer patients were iron deficient, and induced iron deficiency caused myocyte atrophy in vitro (by genetically blocking iron import and use in myotubes) and muscle loss in vivo (by blood removal and low iron diet in mice) (35). Further, muscle mitochondria of cachexic mice were iron deficient, and provision of supplemental iron improved mitochondrial function and rescued cancer-induced muscle loss in vitro in myotube cultures and in vivo in cachexic mice (35). Iron-deficient cancer patients given supplemental iron also improved muscle function (35). Based on these findings, a working model was proposed in which the tumor alters iron metabolism in skeletal muscle, reducing iron availability in mitochondria for iron-dependent enzymes of the tricarboxylic acid (TCA) cycle and oxidative phosphorylation (OXPHOS), leading to restricted adenosine triphosphate (ATP) provision, energy stress, and subsequent muscle wasting in cachexia (Figure 1, E). Because iron provision mitigated cancer cachexia through a mitochondria-muscle axis, modulation of muscle mitochondrial energy metabolism by other targeted means might also be beneficial. These observations may be especially relevant to muscle health of cancer patients for tumor-induced cachexia and because dexamethasone is a common component of induction chemotherapy. Muscle dysfunction arising from tumor load (ie, cancer cachexia) and concurrent cancer-related treatments such as dexamethasone and chemotherapy may need to be recognized and managed accordingly by patients and their treating oncologist.

Cancer cachexia vs cancer treatment effects

Cancer treatments, such as chemotherapies, can also cause muscle wasting, termed chemotherapy-mediated cachexia (36). This may be related to the drug’s anticancer mechanism or to toxic side effect(s) of the drug, which may not have been evaluated (alone or in combination) for their muscle-wasting effects (37). Muscle wasting due to cancer therapies (in addition to cancer) places a great burden on cancer patients, and severe muscle loss is associated with poor prognosis (38). High-throughput omics approaches are proving effective in determining underlying mechanisms for cancer-mediated cachexia and chemotherapy-mediated cachexia.

One omics study compared muscles from a C26 cancer–induced cachexia model with muscles from a chemotherapy-induced muscle-wasting model by proteomics (25). Briefly, for the cancer-induced cachexia model, 8-week-old CD2F1 male mice were injected subcutaneously with 1 x 10⁶ C26 cells. For the chemotherapy-induced wasting model, 8-week-old CD2F1 male mice were treated via intraperitoneal injection with Folfiri cocktail (comprised of 5-fluoruracil, leucovorin, and irinotecan) twice weekly for 5 weeks. Control mice were given an equal volume of vehicle. Tumor hosts and animals treated with chemotherapy were sacrificed when muscle loss was comparable. Of the 269 statistically significantly altered proteins in the C26 cancer cachexia muscles, 235 were downregulated and 34 were upregulated. Of the 386 statistically significantly altered proteins in the Folfiri chemotherapy group, 345 were downregulated and 41 were upregulated. A comparative analysis revealed that 240 proteins were altered in both experimental groups (Figure 2), most of which were similarly downregulated. This analysis identified 2 main subsets of proteins related to metabolic pathways (47% commonality: 44% in C26 cancer cachexia model; 39% in Folfiri chemotherapy cachexia model) and structural proteins (21% commonality: 22% in C26 cancer cachexia model; 24% in Folfiri chemotherapy cachexia model). Pathway analysis revealed that mitochondrial function, calcium signaling, oxidative phosphorylation, beta oxidation and fatty acid metabolism, and the Krebs cycle were severely impacted in both experimental groups (Figure 2). Both proteomic signatures show inhibition of pathways regulating ATP synthesis, nucleotide and fatty acid metabolism, reactive oxygen species scavenging, and muscle and heart function. However, C26 cancer cachectic muscles showed a significant upregulation of acute phase proteins and pro-inflammatory proteins that were not found in chemotherapy cachectic muscles (Figure 2) (25).

Figure 2.

Skeletal muscle alterations in cancer cachexia and chemotherapy-associated cachexia. Findings shown were reanalyzed from skeletal muscle proteome datasets in Barreto et al. [(25); Supplementary Table S4] and ShinyGO 0.76 enrichment tool (bioinformatics.sdstate.edu/go/) using the gene ontology biological process database. 240 differentially expressed proteins (DEP) in skeletal muscle were shared between cancer cachexia and chemotherapy-induced cachexia induced by Folfiri treatment (center). This exceeded the number of DEP unique to each condition alone. Similar alterations may be occurring in skeletal muscle under tumor load and chemotherapy, including mitochondrial metabolism and muscle function (center). Inflammation and wound healing pathways were well represented in cancer cachexia (left) but not in chemotherapy cachexia, implying inflammation drives cancer cachexia but is not a major aspect of chemotherapy-associated muscle dysfunction. Graphic in top panel made with BioRender.com. ATP = adenosine triphosphate; FDR = false discovery rate; Neg = negative; Pos = positive; proc. = process; reg. = regulation.

A different omics study, using the same experimental models as described above, compared C26 cancer–induced cachexia model vs Folfiri chemotherapy-induced cachexia model by metabolomics and similarly found some overlapping metabolic signatures whereas others differed (39). Cancer-mediated and chemotherapy-mediated cachexia models showed decreased circulating levels of glucose and decreased liver glucose and glycogen, indicating an increased systemic glucose demand due to cachexia. Interestingly, each model showed unique changes in beta-oxidation (more severely impaired in chemotherapy-induced cachexia) and Krebs cycle metabolism, which was intermediate dependent. Finally, cancer-mediated cachexia showed an increase in inflammation and low-density lipoprotein markers (40). Together, these data indicate that although there is overlap between affected proteins and metabolites in cancer-mediated muscle wasting and chemotherapy-mediated muscle wasting, key differences also exist. Based on these findings, it appears that inflammatory signaling may be unique to cancer-mediated cachexia. Utilization of high-throughput omics methods is well positioned to identify mechanisms that play a critical role in the development and progression of various forms of cachexia.

Cancer cachexia severity and progression

Omics-based discovery in cancer cachexia severity and progression is most often pursued in experimental models (Table 1). An early investigation adopting this approach used array-based technology to examine muscle transcripts in C26 tumor–bearing mice with moderate and severe cachexia, defined as 15% and 20% weight loss, respectively (41). Weight loss is a hallmark feature of the disorder and used in clinical diagnosis, thus, weight loss–based categories in experimental models allow study of severity mechanisms based on a clinically relevant feature. Major downregulated pathways in moderate and severe cachexia were suggestive of altered muscle structural integrity and regenerative function (eg, striated muscle contraction, extracellular matrix receptor interaction, myogenesis), dampened muscle anabolism (insulin signaling), and diminished mitochondrial oxidative metabolism (peroxisome proliferator-activated receptor-gamma coactivator–1α, pyruvate metabolism, citrate cycle) (41). Top pathways activated in both moderate and severe cachexia related to proteasome and inflammatory cytokine signaling, especially signal transduction through the interleukin-6 (IL-6), Janus kinase, and signal transducer and activator of transcription–3 axis (41). This signaling pathway, which leads to increased hepatic production of acute phase proteins in the systemic inflammatory response, appeared to also be active in skeletal muscle during moderate and severe cancer cachexia. These observations led to the proposal that in addition to liver, skeletal muscle is a major site of secreted acute phase protein production by IL-6 and signal transducer and activator of transcription–3 signaling, occurring at the expense of contractile protein synthesis and leading to skeletal muscle atrophy in cancer cachexia irrespective of severity (41). This study provided a nice early example of high-throughput analysis guiding and refining a mechanism of cancer cachexia in relation to severity of the condition.

Table 1.

Skeletal muscle transcriptome, proteome, and metabolome in relation to cancer cachexia progression^a

Technology	Approach	Findings	Reference
RNA sequencing	Lewis lung carcinoma mouse model Gastrocnemius muscle analyzed at 0, 1, 2, 3, and 4 wk after LLC injection.	Muscle atrophied at 4 wk vs 0 wk (PBS control). Most robust alterations in transcriptome were evident at 4 wk concurrent with muscle atrophy. Pathway analysis pointed to altered maintenance of actin cytoskeleton and mitochondrial OXPHOS at 4 wk vs PBS. OXPHOS was identified in top-10 pathways at 4 wk vs PBS and in 4 wk vs 2 wk but not in the other group comparisons.	Blackwell et al. (42)
Gene microarrays	C26 tumor–bearing mouse model Quadriceps muscle analyzed in moderate and severe cachexia (10% and 15% weight loss).	Complement coagulation cascades, proteasome, and MAPK, IL-6, and STAT3 pathways among top upregulated pathways in moderate and severe. Muscle contraction, PGC-1α, ECM receptor interaction, myogenesis, insulin signaling, pyruvate metabolism, and TCA cycle were among top downregulated pathways in moderate and severe. Transcript profiles of moderate vs severe not directly compared.	Bonetto et al. (41)
MS-based proteomics	Quadriceps biopsy from 4 patient groups: healthy elderly control, cancer weight stable, cancer cachexia, age-related sarcopenia	Differentially regulated proteins in cancer cachexia (n = 4) vs all other groups were input to STRING database for potential protein interaction networks, revealing 3 major clusters: F0 complex, electron transport chain, and muscle fiber contraction, suggesting altered mitochondrial energy metabolism and muscle structural composition.	Ebhardt et al. (47)
MS-based proteomics	C26 tumor–bearing mouse model Gastrocnemius muscle analyzed in moderate and severe cachexia (10% and 20% weight loss) vs weight-stable mice.	In moderate cachexia, pathways enriched from differentially regulated proteins pertained to ribosomes, complement coagulation cascades, fibrin formation, and integrin and MAPK signaling, suggesting altered translational control, acute phase response, and fibrosis. These pathways remained enriched in severe; however, TCA cycle and OXPHOS pathways were noted additions to severe not reflected in moderate, and they were enriched from decreased proteins.	Khamoui et al. (8)
NMR metabolomics	Walker-256 rat model gastrocnemius analyzed at 4 timepoints: day 0 (control), 9 days (initial), 14 days (intermediate), and 21 days (advanced) after tumor cell injection.	Metabolite changes in advanced (vs control): ↑ BCAA, phenylalanine, tyrosine ↑ Carnosine (also ↑ vs intermediate) ↑ Cytidine ↑ Pyridoxine (B6 metabolism) ↑ 4-pyridoxate (also ↑ vs initial, intermediate) ↑ N-methylhydantoin Metabolite changes in intermediate (vs control): ↑ Anserine ↑ sn-glycero-3-phosphocholine Metabolites were not altered in initial (vs control).	Chiocchetti et al. (43)

^a BCAA = branched-chain amino acid; ECM = extracellular matrix; IL-6 = interleukin-6; LLC = Lewis lung carcinoma; MAPK = mitogen-activated protein kinase; MS = mass spectrometry; NMR = nuclear magnetic resonance; OXPHOS = oxidative phosphorylation; PGC-1 α = peroxisome proliferator-activated receptor-gamma coactivator-1α; PBS = phosphate-buffered saline; STAT3 = signal transducer and activator of transcription-3; TCA = tricarboxylic acid.

More recently, RNA sequencing was used to study alterations in the cachexic muscle transcriptome using a time-course approach (42). An advantage of RNA sequencing by next-generation sequencing platforms is that it allows for novel, de novo discovery of RNA alterations on a transcriptome-wide scale, whereas array-based technologies as in the aforementioned study probe expression of a large but fixed number of genes. Here, skeletal muscle was collected at 1 week intervals up to 4 weeks after subcutaneous injection with LLC cells compared with sham saline-injection control mice. This design permitted examination of events before and after cachexia-associated muscle wasting occurred. Muscle atrophy was present at 4 weeks compared with control mice. The most robust transcriptome alterations occurred at the same 4-week timepoint coincident with muscle wasting. Prominent annotations based on ingenuity pathway analysis were altered maintenance of actin cytoskeleton and mitochondrial OXPHOS (42). Further, OXPHOS was identified as a top-10 pathway not only at 4 weeks vs 0 weeks (sham injection saline control) but also in 4 weeks vs 2 weeks (30), suggesting disturbance in mitochondrial oxidative metabolism to be prominent around or at the time muscle wasting appeared.

Relating transcript expression with proteins and metabolites by proteomic and metabolomic profiling enables evaluation of downstream impact. Proteomic analysis of skeletal muscle was recently performed in C26 tumor–bearing mice with moderate and severe cancer cachexia, defined as 10% and 20% weight loss, respectively (8). In moderate cachexia, pathways enriched from differentially regulated proteins included ribosomes, complement coagulation cascades, fibrin formation, and integrin- and mitogen-activated protein kinase signaling, suggesting altered translational control, acute phase response, and possible fibrosis (8). When increased and decreased proteins were analyzed separately, complement and coagulation cascades were the prominent pathways enriched from increased proteins based on P value (8), reflective of an inflammatory disease. Ribosome was the noted pathway enriched from decreased proteins in moderate cachexia (8), implicating ribosomal protein loss. Many of these same pathways remained enriched in severe cachexia; however, noted additions to severe cachexia not reflected in moderate cachexia were terms related to mitochondrial energy metabolism. Specifically, TCA cycle and OXPHOS were prominently enriched based on the P value from the decreased proteins alone (26), suggesting mitochondrial bioenergetic limitation in severe cachexia. Further, metabolomic analysis in a rat model of cancer cachexia showed the most robust skeletal muscle metabolite change during advanced tumor-bearing corresponding to severe cachexia (43). Among the modifications were increased concentrations of branched chain amino acids and other amino acids capable of anaplerotic entry into the TCA cycle (43), suggesting muscle protein breakdown to supply amino acids for energy provision and attempted alleviation of energy stress in the severe state. Together, these proteome and metabolome findings suggest mitochondrial metabolic dysregulation and energy stress to distinguish severe cancer cachexia.

In a proteome analysis of C26 tumor–bearing mice, we note the observation of muscle ribosomal protein loss shared across cancer cachexia severity (8). This finding is notable because ribosome biogenesis is strongly implicated in control of skeletal muscle mass (44). Ribosomal protein loss due to increased degradation and/or restricted biogenesis would be expected to limit translational capacity and rates of protein synthesis, causing net protein loss and potential contribution to cachexia-related muscle atrophy (45,46). Ribosome-targeting strategies in skeletal muscle that manipulate ribosome biogenesis and/or degradation may be a potential avenue to protect muscle mass in early and severe cachexia.

Patient-derived samples are the ideal biological material to study cancer cachexia biomarkers and treatment targets in skeletal muscle. Samples from this patient group are difficult to obtain, however, and this is reflected in the relatively larger representation in the literature by experiments to date. Nevertheless, an early small-scale study of patient samples from quadriceps muscle biopsy was recently reported (47). Here, proteome analysis was performed on quadriceps muscle biopsies from 4 patient groups: healthy elderly control, age-related sarcopenia, cancer weight–stable, and cancer cachexia. Aside from direct study of patient muscle samples, this design is advantageous because it allows true cachexia mechanisms to be discriminated from aging- and cancer-related effects that co-occur and are confounding in this population. Proteins identified as differentially regulated in cancer cachexia compared with all other groups were input to the STRING database to identify potential protein interaction networks. This analysis revealed 3 major clusters derived from 16 proteins: F0 complex (Atp5a1, Atp5f1, and Atp5h), electron transport chain (Etfb, Aldh4a1, Uqcrfs1, Sdhb, Cox5a, Cox5b), and muscle fiber contraction (Myoz2, Myh8, Myh4, Myl6b, Myl3, Myl2, Tnnt1) (47). Although the sample size was small and findings exploratory in nature (48), the authors proposed the possible existence of a skeletal muscle protein level signature reflecting muscle structure and energy status in human cancer cachexia that, pending additional testing in a larger cohort, could be an objective indicator in skeletal muscle tissue to aid clinical evaluation of cancer cachexia. A cachexia signature of this nature is consistent with the above-mentioned experiments showing transcriptome, proteome, and metabolome profiles of muscle energy metabolism disturbance especially in advanced cachexia.

Secondary analysis of muscle proteome—progression from moderate to severe cachexia

Experimental studies based on degree of weight loss have relevance to the clinic. Differences in survival were reported at several weight-loss categories in cancer patients including weight loss of 6% to 10.9% and at least 15% (49). We performed a secondary analysis of recently published skeletal muscle proteome data from C26 tumor-bearing mice with moderate (10% weight loss) and severe cancer cachexia (20% weight loss) (8). We aimed to expand that analysis by focusing on skeletal muscle proteins differentially regulated between moderate and severe groups (Figure 3, A). This paired comparison reflects alterations in the muscle proteome transitioning from a moderate to severe state. Clues for slowing advancement toward severe cachexia refractory to clinical treatment may be revealed by this comparison. This departs from prior examination of changes during early and late stages of cachexia separately, without directly comparing them by global molecular profiling (50). Differentially expressed proteins were first sorted into separate lists containing increased and decreased proteins (Figure 3, A). Increased and decreased proteins were then separately analyzed in silico to determine interaction patterns, annotations, and enriched pathways.

Figure 3.

Skeletal muscle proteome analysis may offer clues to slow progression from moderate to severe cancer cachexia. A) Skeletal muscle was collected from C26 tumor–bearing mice with moderate and severe cancer cachexia (10% and 20% weight loss, respectively) (8), and TMT-based quantitative proteomics performed. We did a secondary analysis of this previously published proteome data set focusing on direct comparison of moderate and severe groups. Increased and decreased proteins (P < .05) were separately analyzed by in silico tools. Analysis of increased proteins is shown here. B) Potential interaction patterns among increased proteins visualized by STRING. C) Reactome pathways enriched from increased proteins by KOBAS-i. D) Working hypothesis in which inflammation contributes to skeletal muscle fibrosis associated with type VI collagen during transition to severe cachexia. E) Therapeutic strategies may need to address multiple components concurrently to slow progression to severe disease. Based on annotations from increased muscle proteins, these may include inflammation, fibrosis, and contractile protein loss. APC/C = Anaphase-promoting complex/cyclosome; CDC6 = cell division cycle 6; DEP = differentially expressed proteins; ECM = extracellular matrix; ORC = origin of replication; UCH = Ub C-terminal Hydrolase; WL = weight loss.

Predicted interactions among increased proteins were visualized by the STRING database (v11.5) (Figure 3, B). Prominent clusters comprised acute phase response and inflammation, extracellular matrix, striated muscle structure, and development and ubiquitin-proteasome system (Figure 3, B). To determine enriched pathways, increased proteins were imported into KOBAS-i and analyzed by the gene-list enrichment function (Figure 3, C). Pathways were consistent with major clusters visualized in STRING; extracellular matrix/integrin/laminin, inflammation/neutrophil degranulation/innate immune system, muscle contraction, and proteasome were shared between them (Figure 3, C). Increased acute phase proteins in muscle and blood are well-established observations in cancer cachexia, indicating muscle-localized and systemic inflammation (41,51). Inflammatory cytokine signaling leading to protein degradation and skeletal muscle atrophy in cancer cachexia is also well documented (52). These annotations reinforce the prominence of inflammation in cancer cachexia.

Of the increased acute phase reactants, 4 belong to the serine protease inhibitor (Serpin) superfamily including Serpina1b, Serpina1d, Serpina3k, and Serpina3n. Increased Serpina3n was independently detected in other muscle proteome experiments and suggested as a cancer cachexia biomarker (24,40,53,54). Biological significance of increased Serpina3n to cancer cachexia is unclear but may be compensatory protective. For instance, transgenic overexpression of Serpina3n protected against muscle defects in dystrophic mice (55). Serine protease inhibition, a main function of Serpina3n, was also shown to reduce proteolysis and slow muscle fiber size loss in disuse atrophy (56). Interestingly, increased Serpina3n was reported to coincide with the development of lung fibrosis in mouse models (57). Knockdown of Serpina3n by adeno-associated virus serotype 9 in the same models alleviated fibrosis (57). Possible fibrosis may be prominent in moderate to severe cachexia, as reflected by in silico identification of acute phase response and inflammation, extracellular matrix, and integrin signaling (Figure 3, B and C). Fibrosis is a well-known terminal event of injury-induced inflammation in muscle (58), occurs in various skeletal muscle disease, and was previously reported in pancreatic cancer patients with cachexia (59). Given these findings, a potential hypothesis is that elevated Serpina3n production initially attempts to protect muscle against cachexia-inducing tumor load but is maladaptive when prolonged, accumulating fibrous proteins such as type VI collagen (consisting of α1, α2, α3 chains) at the expense of contractile proteins and leading to muscle atrophy (Figure 3, D). Based on the in silico analysis of increased proteins, strategies to slow cachexia-related muscle dysfunction may need to consider several mechanisms concurrently (Figure 3, E).

Next, interaction patterns were visualized for decreased proteins; this yielded 1 predominant cluster composed of mitochondrial proteins (Figure 4, A). The 58 decreased proteins were referenced against the MitoCarta 3.0 database, an expertly curated inventory of mammalian mitochondrial proteins and pathways (60). Nearly 60% of the decreased proteins were matched in MitoCarta 3.0 (34 mitochondrial proteins of 58 decreased proteins overall) (Figure 4, B). Of the 34 proteins, 90% were located in the matrix and mitochondrial inner membrane (Figure 4, B). Sublocalization to these specific compartments are reflected in the top-10 enriched pathways identified by KOBAS-i, which all pertained to mitochondrial energy metabolism and, in particular, TCA cycle and OXPHOS (Figure 4, C). Closer inspection showed many of the 34 mitochondrial proteins functioning at the TCA cycle and respiratory chain as expected (Figure 4, D). Of those localized to the respiratory chain, the vast majority resided at mitochondrial complex I (Figure 4, D), the largest respiratory complex and major site of electron input, accounting for approximately 40% of proton motive force (61,62). Two proteins, Ndufs3 and Ndufs8 (Figure 4, D), are core subunits required for complex I catalytic activity and energy transduction (63). In prior analysis of mitochondrial oxygen flux in permeabilized muscle fibers from C26 tumor–bearing mice, respiratory limitation at complex I occurred in severe but not moderate cachexia [Figure 4, E (64)]. Together, these findings suggest prominent complex I deficiency progressing from moderate to severe cancer cachexia. An intriguing possibility emerges in which improvement of TCA cycle flux and/or mitochondrial complex I activity could alleviate energy stress and slow progression to severe cancer cachexia (Figure 4, F).

Figure 4.

Mitochondrial energy metabolism limitation during decline from moderate to severe cancer cachexia. Differentially regulated proteins that decreased in severe vs moderate cachexia (58 in total) were analyzed in silico. A) Interaction networks among decreased proteins were performed by STRING showing a cluster composed of mitochondrial proteins. B) The 58 decreased proteins were referenced against the MitoCarta 3.0 database of which 34 were matched. Of 34, 31 (approximately 90%) mitochondrial proteins localized to the matrix and mitochondrial inner membrane (MIM). C) Reactome pathways enriched from decreased proteins by KOBAS-i show prominence of energy-yielding pathways in mitochondria. D) Inspection of individual mitochondrial proteins indicated localization primarily to the TCA cycle and complex I of the respiratory chain. Two proteins denoted in red font are core subunits of complex I. E) Previously reported in situ mitochondrial respiration in permeabilized skeletal myofiber bundles showed decreased complex I supported respiration in severe but not moderate cachexia (64). When considered together with decreased mitochondrial proteins in panel D, complex I deficiency may associate with transition from moderate to severe cachexia. F) Based on in silico analysis of decreased proteins, a possible therapeutic strategy alleviating energy stress through TCA cycle flux and/or mitochondrial complex I might slow progression from moderate to severe cancer cachexia. ATP = adenosine triphosphate; CI = complex I; MOM = mitochondrial outer membrane; TCA = tricarboxylic acid; UCP = uncoupling proteins; WL = weight loss.

Underexplored areas, future opportunities, and conclusions

A need exists for experimental models of cachexia that recapitulate molecular profiles of human cancers and parallel cancer-induced wasting in humans (65). Experimental models with limited genetic and phenotypic features of human disease might partly contribute to attrition of therapeutics in development (66). Expanding the pool of experimental models for cachexia was a noted aim of the 2020 Cancer Grand Challenges (https://cancergrandchallenges.org/challenges/cachexia), an initiative founded by Cancer Research UK and the National Cancer Institute, which seeks to overcome obstacles in cancer by supporting diverse scientific teams. Experimental models in development will benefit from the comprehensive molecular characterization offered by the various omics technologies. Recently, several genetically engineered mouse models successfully demonstrated molecular alterations of human tumors and the hallmarks of cachexia. An inducible, genetic model of non-small cell lung cancer was developed by intranasal administration of Cre to induce deficiency of Lkb1 in mice carrying Kras-mutated tumors (67). Mice carrying Kras-mutated tumors developed metastatic lung tumors spontaneously and showed weight loss, anorexia, systemic inflammation, and progressive depletion of skeletal muscle and white adipose tissue (67). In addition, a genetic model of pancreatic cancer was developed by tamoxifen-inducible Cre that inactivated Pten in Kras mutants (KPP mice) (68). The KPP mice spontaneously developed tumors and showed a gene expression profile by RNA sequencing in skeletal muscle similar to pancreatic ductal adenocarcinoma patients (68). Muscle and fat wasting occurred over an extended period of time, resembling the progressive wasting experienced by pancreatic ductal adenocarcinoma patients (68). Together, these animal models of increasing relevance to human cancer cachexia may improve confidence in translation of findings that may have been limited in older experimental models.

Beyond experimental models, uncovering the genetic predisposition for cancer cachexia remains ongoing. Features of cachexia such as body weight, BMI, lean mass, adiposity, and strength are heritable traits. Knowing the extent to which genetic variation predicts susceptibility to cachexia could aid in early identification of predisposed individuals. Prior work approached this by analyzing single nucleotide polymorphisms (SNPs) associated with weight loss, low BMI, computed tomography–derived muscularity, and survival (69-73). SNPs are single-base substitutions in germline DNA that represent common genetic variations in the population, with minor allele frequencies greater than 5%. Individual SNPs at any given gene locus by themselves are unlikely to explain risk for complex diseases like cachexia, because of small individual effects of most SNPs, but may become more informative for disease risk through their combined effects (74). In cancer cachexia, SNPs in LEPR, ACVR2B, TNF, ACE, IL-10, VDR, AKT1, and SELP genes showed statistically significant associations with weight loss and/or low muscularity by computed tomography (69-73). These genes are known to encode proteins regulating muscle metabolism and cell growth (LEPR, ACVR2B, VDR, AKT1), inflammation (TNF, IL-10), and muscle performance (ACE), suggesting possible genomic biomarkers of muscle-wasting risk.

The individual SNPs selected for genotyping in these studies were chosen through reviews of the literature (69-73). Identification of unexpected or rare variants is not offered through this approach. Discovery of low-frequency variants along with common variations could be enabled by whole-genome sequencing and large-scale, genome-wide association studies (GWAS) (75). Search of the GWAS catalog for cachexia and muscle wasting did not yield results (www.ebi.ac.uk/gwas/home; accessed December 23, 2022), indicating infrequent application of GWAS to cachexia compared with other complex diseases. Several studies examined the genetic architecture of lean body mass and strength (76-79), however, only a single GWAS of cancer cachexia was located by literature search (80). Here, more than 400 000 SNPs were genotyped by array-based technology and tested for associations with delta BMI in gastrointestinal cancer (80). A single SNP, in the myogenic gene DOCK1, was associated with delta BMI in gastrointestinal cancer cases (80). To expand these lines of inquiry, whole-genome sequencing data in large-scale GWAS approaches might be an opportunity to better capture all risk variants associated with cachexia-related traits and explore use of polygenic risk scores.

Despite recent advances in our understanding of cancer cachexia, its underlying mechanisms remain incompletely understood. There is a specific need for sufficiently powered human muscle biopsy–based studies in which multiple muscle types are sampled and comprehensively characterized. Analyzed biopsy specimens from cachexic cancer patients are less widespread because of challenges with invasive sample procurement but are needed to bridge preclinical findings toward clinical use. Examining 1 tissue type by a single technology is typical in this setting. For instance, the serum proteome was analyzed in patients with pancreatic cancer cachexia to identify potential protein biomarkers in circulation (81). Proteome-wide examination of biomarkers in blood is attractive for translation into routine clinical use given the relative ease of collecting blood compared with tissue biopsies. Blood proteins associated with clinical cachexia parameters including weight loss, skeletal muscle index, and skeletal muscle density were identified (81). Specifically, inflammation-related proteins associated with each of the 3 clinical parameters (81), adding to existing evidence for a prominent role of inflammatory status in cancer cachexia. Another study profiled muscle micro-RNAs (miRNAs) in patients with cancer cachexia using RNA sequencing (82). They reported 8 differentially expressed miRNAs that were upregulated and not previously implicated in cancer cachexia (82). Of these miRNAs, 190 potential mRNA interactors were identified, and pathways enriched from these mRNAs involved muscle development and inflammation (82). Continued leverage of different omics platforms to study patient samples, like skeletal muscle in particular, will stimulate further discovery of cachexia mechanisms.

Liquid biopsies have received considerable attention as a potential clinical tool for cancer diagnosis, treatment selection, and disease monitoring (83,84). Liquid biopsy methods involve analysis of circulating tumor cells (CTCs), circulating tumor DNA, tumor-educated platelets, cancer-associated fibroblasts, proteins, RNAs, or extracellular vesicles in patient blood samples (or other fluids such as cerebrospinal fluid, saliva, ascites, urine, and pleural effusion) (84). Materials enriched from liquid biopsies can be analyzed by high-throughput sequencing technologies for comprehensive molecular profiling (83,84). Because of the routine and minimally invasive nature of blood collection, use of liquid biopsies for cancer cachexia surveillance is an interesting avenue of exploration. This possibility was reviewed and discussed with a focus on circulating miRNAs as the analyte of choice (85,86). Relevance of other liquid biopsy material such as CTCs to cancer cachexia is not extensively studied. CTCs shed from primary tumors enter circulation and are believed to function as triggers of metastasis (87), which can co-occur with and exacerbate cachexia (88). CTCs were also positively associated with C-reactive protein and IL-6 (89), 2 inflammatory molecules consistently elevated in cancer cachexia. Further, CTCs inversely associated with BMI (89), suggesting high CTC count in weight-losing cancer patients. Questions to consider are whether CTCs directly contribute to cancer cachexia, if CTCs can be used for surveillance, and if particular molecular features of CTCs relate to cachexia.

Integrating omics profiles with clinical data to predict risk of developing cachexia is another avenue of exploration. Molecular profiles from high-throughput technologies are already used in this way for prognostic purposes in oncology. In gastric cancer, array-based tumor gene expression analysis was used to develop a prognostic risk score based on a 6-gene signature that predicts risk of relapse after surgery (90). An extension of this approach was done in diffuse, large B-cell lymphoma, where whole-exome sequencing and transcriptome sequencing was performed on approximately 1000 patient tumors (91). A prognostic model that integrated clinical variables with key oncogenic mutations identified by sequencing outperformed existing risk assessment methods for predicting survival (91). Similar applications have not been widely investigated in cancer cachexia. Prognostic indication is mostly limited to clinical variables. For instance, in lung cancer patients, disease stage, serum albumin, systemic inflammation, anemia, and surgical treatment were independent risk factors for developing cachexia (92). The same factors and sex and Karnofksy performance status were prognostic for survival in lung cancer patients with existing cachexia (92). Nomograms were developed based on this information to screen risk of cachexia in patients and survival rates once cachexia occurred (92). Extending these approaches by incorporating molecular profiles from high-throughput technologies to improve prediction of risk for cachexia or guide treatment choice may be worth further study.

A major challenge is to map and integrate the large amount of data produced by multi-omics interrogation of relevant organ systems by age, sex, and stage of cachexia severity. Similarly complex projects are already underway in oncology. The pan-cancer analysis of whole genomes published in 2020 used whole-genome sequencing to analyze nearly 3000 genomes from 38 tumor types (11). The project aimed to develop a resource documenting genetic variation in different cancers and was made possible by data sharing between the International Cancer Genome Consortium and The Cancer Genome Atlas. Other disciplines have initiated similarly ambitious projects such as the Kidney Precision Medicine Project funded by the National Institute of Diabetes and Digestive and Kidney Disease that aim to develop a molecular atlas of healthy and diseased adult kidneys (93). Projects of this scale for cancer cachexia would be a large undertaking requiring substantial investment but having access to a multi-omics catalogue of tissue- and severity-specific alterations in cancer cachexia could be a valuable resource to advance research in the field and help identify targets and biomarkers that predict risk of cachexia and inform treatment choice. Shared resources of this nature available to the research community would promote discovery and hopefully guide advancement of those discoveries into improved patient care.

Overall, advancements in sequencing technologies, reduction in costs, and lower barrier to entry will encourage further application of these high-throughput methods toward discovery biology in cancer cachexia. Our analysis of selected applications (see above) in conjunction with other high-throughput omics analyses (see above) in skeletal muscle discovered that inflammatory response, muscle structural integrity, and mitochondrial energy metabolism were recurring signatures in several experimental models and some early clinical investigation. Such information sheds an important light on possible therapeutic targets. High-throughput omics approaches are a critical first step, however, validation of pathways and targets are critical. Next steps might rigorously evaluate the clinical relevance of these signatures, such as aiding the diagnosis of cancer cachexia, predicting disease trajectory, or the ability to evaluate treatment response. Looking forward, continued leverage of these methods are likely to play an important role in unraveling the complexities of cancer cachexia biology, potential development of routine clinical tools incorporating molecular information, and progress toward effective treatment.

Data availability

Protein identification data sets analyzed for the secondary analysis are available in the Figshare open data repository and can be accessed at https://doi.org/10.6084/m9.figshare.14883183.

Author contributions

Gwendolyn A Thomas, PhD (Conceptualization; Writing—Review & Editing), Andy V Khamoui, PhD (Conceptualization; Formal Analysis; Investigation; Methodology; Project Administration; Visualization; Writing—Original Draft; Writing—Review & Editing), Linda Anne Gilmore, PhD, RD (Conceptualization; Project Administration; Writing—Original Draft; Writing—Review & Editing), Traci L Pary, PhD (Writing—Original Draft; Writing—Review & Editing).

Funding

This work was supported in part by the National Institutes of Health Transdisciplinary Research in Energetics and Cancer (TREC) Training Workshop R25CA203650 (PI: Melinda Irwin).

Conflicts of interest

The authors declared no conflict of interest relative to this work.