Change in skeletal muscle mass during systemic cancer treatment: a systematic review and meta-analysis Authors: Lukas Svendsen, Sandra Jensen, Stine Hansen, Victor Sørensen, Christoffer Johansen, Charlotte Suetta, Helle Pappot, Casper Simonsen, Lars Hermann Tang, Susanne Oksbjerg Dalton, Gunn Ammitzbøll, Bolette Skjødt Rafn Published: 2026 Introduction With increasing cancer survival, clinical priorities have shifted from tumor control alone toward also preserving physical function and quality of life during treatment and survivorship [1]. Across cancer types, skeletal muscle mass (SMM) has emerged as a key determinant of these outcomes and a potent factor influencing treatment tolerance and prognosis [2, 3]. Low SMM, often termed sarcopenia [4], is present in approximately one-third of patients with cancer [5, 6] and has consistently demonstrated clinical relevance. Low SMM at time of diagnosis is associated with shorter overall survival [7–10], shorter progression-free survival [11], higher treatment-related toxicity [12], postoperative complications [13], and reduced quality of life and depression [14]. Collectively, these findings demonstrate that low SMM is an important prognostic biomarker that could be incorporated into clinical evaluations and research [15]. However, while the baseline SMM could be clinically useful, particularly in settings where longitudinal imaging is unavailable, emerging evidence shows that changes in SMM during systemic therapy offer additional and often stronger prognostic information across cancer types [16–19]. In a multicohort study including patients with advanced non-small cell lung cancer (n = 1,791), less SMM decline was associated with a 26–54% lower mortality risk (Hazard Ratio [HR]: 0.46–0.74) [20]. Similarly, SMM loss during treatment was associated with poorer overall survival as in colorectal (n = 67; ≥ 9% SMM loss, HR: 4.47, 95% confidence interval [CI]: 2.21–9.05) [21], biliary tract (n = 524; SMM loss; HR: 2.58, 95% CI: 1.86–3.58) [22] and pancreatic cancer (n = 127; ≥ 7.9% SMM loss; HR: 4.02, 95% CI: 1.87–8.97) [23]. Collectively, these findings suggest that while baseline SMM is informative, the trajectory of SMM during treatment captures prognostic information that is not evident from a single time point. Despite this, existing studies are heterogeneous with respect to cancer populations, treatment regimens, measurement methods, and timing of assessments, which complicates direct comparison and limits the ability to derive a single, generalizable estimate of treatment-associated SMM change. The body of literature describing treatment-related SMM loss therefore remains insufficiently characterized in terms of its magnitude, timing, and variability across clinical contexts. This study aimed to systematically describe changes in SMM during systemic cancer treatment supported by quantitative pooled estimates and to explore if changes vary across cancer types and treatment modalities by conducting a systematic review and meta-analysis of SMM changes during chemotherapy and/or immunotherapy (± targeted therapy). Material and methods This systematic review and meta-analysis was reported according to the Cochrane Handbook [24] and to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [25] (Table S1). The protocol was pre-registered at PROSPERO (CRD42022308388). Search strategy An initial systematic literature search was performed in PubMed, Embase, and Web of Science on May 17, 2023. A subsequent updated search was conducted on April 11, 2025, to ensure identification of all relevant studies. The search combined controlled vocabulary terms and free-text keywords organized into three concept blocks: cancer, muscle mass or sarcopenia, and chemotherapy or immunotherapy. The complete search strategies for all databases are provided in Table S2. Eligibility criteria Studies were eligible if they were published in peer-reviewed journals, included adults with cancer receiving chemotherapy or immunotherapy between baseline and follow-up assessments of SMM or prevalence of low SMM, and reported absolute values at both timepoints. To be included in the meta-analysis, studies also had to provide corresponding measures of variability (e.g. standard deviation [SD] or range). Combination regimens with targeted agents were accepted; studies in which ≥ 10% of patients received targeted therapy were classified as chemotherapy + targeted therapy or immunotherapy + targeted therapy. Eligible studies were required to quantify SMM using an objective method – computed tomography (CT), magnetic resonance imaging (MRI), dual-energy X-ray absorptiometry (DXA), or bioelectrical impedance analysis (BIA). Studies involving surgery or radiotherapy between SMM assessments were excluded to isolate the effects of systemic therapy. Studies not obtainable in English or a Scandinavian language, or those involving structured physical exercise or nutritional interventions, were excluded. Complete inclusion and exclusion criteria are provided in Table S3. Selection of studies The selection of studies was performed using Covidence software. After duplicate removal, two independent reviewers (LS, SH, SJ, VS, GA, and BSR) screened titles and abstracts, followed by full-text assessment according to eligibility criteria. Disagreements were resolved through discussion or consultation with a third reviewer until consensus was reached. Interrater reliability between reviewers was assessed using Cohen’s kappa coefficient (κ), which ranges from 0 to 1, with values of 0.01–0.20 indicating slight, 0.21–0.40 fair, 0.41–0.60 moderate, 0.61–0.80 substantial, and 0.81–1.00 almost perfect agreement. Data extraction Data were extracted using a standardized Microsoft© Excel spreadsheet by two independent reviewers (LS, SH, SJ, VS, GA, and BSR). Disagreements were resolved through discussion or consultation with a third reviewer. When essential data were missing, two contact attempts by email were made to obtain additional information. The following data were extracted: author, year, study design, country, patient setting, sample size, sex, age, cancer type and stage, treatment regimen and number of cycles, SMM assessment method and anatomical site (e.g. skeletal muscle index [SMI] and area [SMA], pectoralis muscle area [PEMA], psoas muscle index [PMI] and area [PMA], and lumbar muscle volume [LMV]), time between assessments, outcome type and cut-off values, baseline and follow-up SMM, correlation coefficient between baseline and follow-up, low SMM prevalence, corresponding p-values and confidence intervals, data source, and funding. Risk of bias Risk of bias was assessed independently by at least two reviewers using the National Heart, Lung, and Blood Institute (NHLBI) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies [26]. Individual items were rated as ‘yes’, ‘no’, or ‘other’ (not reported, not applicable, or not determinable). An overall study rating was then assigned based on the proportion of criteria rated ‘yes’: ≥ 75% indicates good quality, 50–74% fair quality, and < 50% poor quality [27]. We tailored the risk-of-bias assessment to the specific objective of this review, which addressed within-patient changes in SMM rather than exposure–outcome associations. Therefore, items related to confounding (of the exposure-outcome association) were considered not applicable in terms of risk of bias relevant to the data we extracted for this review. Further, sample size justification was considered not applicable due to the retrospective population-based study designs, and loss to follow-up was not relevant due to availability of two measurements being an inclusion criterion for this review. However, we recognize that the study designs included in this review combined with inclusion criteria of two available SMM measurements carry considerable risk of bias, which we highlight in the discussion section. Statistics Random-effects meta-analyses were conducted using the restricted maximum-likelihood (REML) estimator with Hartung–Knapp adjustment. A random-effects model was chosen because substantial clinical and methodological heterogeneity was expected across cancer types, treatment regimens, and body composition assessment methods. Continuous outcomes were synthesized as standardized mean changes (SMC) with 95% confidence intervals (CI), relative to the baseline SD. Effect sizes were interpreted according to Cohen’s thresholds (0.2 = small, 0.5 = moderate, 0.8 = large). Because the standard error (SE) depends on the correlation between baseline and follow-up values, we calculated the mean correlation coefficient from studies that reported it and imputed this value to studies without available data. To assess robustness, sensitivity analyses were performed using correlation coefficients of 0.1, 0.5, and 0.9. Studies reporting medians with interquartile ranges (IQR) or ranges, means, and SDs were estimated using the method described by Wan and colleagues [28]. Between-study variance (τ2) was estimated using REML. Statistical heterogeneity was further quantified using I2, with thresholds of 0–40% (might not be important), 30–60% (may represent moderate heterogeneity), 50–90% (may represent substantial heterogeneity), and 75–100% (considerable heterogeneity). Potential sources of heterogeneity were explored through predefined subgroup analyses (cancer type, treatment type, treatment setting, assessment tool, and sex) followed by sensitivity analyses (estimated means, non-small trials, imputed correlation coefficient, and study design). Prediction intervals were calculated to illustrate the expected range of true effects in future comparable studies. Publication bias was assessed by visual inspection of funnel plots and with Egger’s test. Percentage change in SMM was calculated from reported baseline and follow-up values when not explicitly provided in the original studies, using the formula: percentage change = ((follow-up − baseline)/baseline) × 100. All analyses were conducted in R (version 4.4.1; RStudio version 2025.09.0, Posit Software) using the meta package. Results The search yielded 13,349 records. After removal of duplicates and exclusion of ineligible studies, 78 studies were included (Figure 1). Study characteristics are summarized in Table 1 and Table S4. Seventy-eight studies (n = 10,502; 52% male; median age 64 years, IQR: 34–77) were included. Most studies (n = 65, 83%) were retrospective cohorts, and publications spanned from 2007 to 2025, with 58 (74%) published between 2019 and 2025. Interrater reliability was moderate for title and abstract screening (Cohen’s κ = 0.43) and substantial for full-text screening (Cohen’s κ = 0.79). Most studies were of good methodological quality (median rating 91%) (Table S5). Table 1Study characteristics.Author (year)/countryStudy designN (male/ female)Age (± SD) or min-maxaCancer typeCancer stage (TNM unless otherwise specified)SettingPlanned treatment regimenCycle(s) Pancreatic cancer Chemotherapy Shimura (2023) [51]/JapanRetro75 (40/35)67 (± 8)Pancreatic cancerUICC 8th stage:I: 17, II: 34, III: 7, IV: 16NeoGemcitabine + S-11 Jin (2021) [52]/ ChinaRetro119 (59/60)60 (± 8)Pancreatic cancerNRNeoNab-paclitaxel + gemcitabine; gemcitabine-based; FOLFIRINOX (multi-agent)NR Griffin (2019) [53]/ IrelandRetro78 (37/41)64 (± 8)Pancreatic cancerTumor stage:I: 6, II:2, III:17, IV: 0NeoFOLFIRINOX; nab-paclitaxel + gemcitabine; gemcitabine monotherapy; gemcitabine + platinumNR Lee (2024) [40]/ South KoreaRetro456 (272/184)61 (± 10)Pancreatic cancerMetastaticPalliativeFOLFIRINOX or gemcitabine + nab-paclitaxel (multi-agent)NR Rollins (2016) [54]/ UKRetro98 (55/44)65 (± 9)Pancreatic cancerLocally advanced: 60, Metastatic: 38PalliativeGemcitabine-basedNR Aberle (2025) [55]/NetherlandsRetro52 (31/21)64 (± 13)Pancreatic cancerLocally advanced: 27, Metastatic: 25NRFOLFIRINOX (multi-agent)4 Davis (2025) [56]/ USARetro103 (58/45)68 (± 11)Pancreatic cancerI:5, II:35, III:15, IV:43NRSingle-agent irinotecan/gemcitabine/oxaliplatin/paclitaxel; combinations incl. gemcitabine + paclitaxel ± irinotecan/oxaliplatinNR Uemura (2020) [23]/JapanRetro69 (38/31)63 (38–74)Pancreatic cancerIVNRFOLFIRINOX (multi-agent)Every 2 weeks Lee (2019) [41]/ South KoreaRetro57 (32/25)61 (38–78)Pancreatic cancer(first line)II:12, III:11, IV:34NRFOLFIRINOX (multi-agent)4 (3–6) Urological cancer Chemotherapy Miyake (2018) [57]/ JapanRetro14 (12/2)73 (64–77)Advanced urothelial cancerClinical TII:8, III:3, IV:3NeoGemcitabine + cisplatin/carboplatin (GC/GCa, platinum-based)3 MacDonald (2024)/Canada [58]Retro70 (59/11)65 (± 8)Muscle-invasive bladder cancerClinical T cT1:9, cT2:50, cT3:6, cT4:5NeoGemcitabine + cisplatin (97%) or gemcitabine + carboplatin (3%) (platinum-based)2 Rimar (2018) [59]/ USARetro26 (19/7)67 (40–82)Muscle-invasive bladder cancerClinical TT2:18, T3:8, N1:8NeoMVAC; gemcitabine + cisplatin; gemcitabine + carboplatin (platinum-based)3–5 Lyon (2019) [60]/ USARetro183 (155/28)65 (57–72)Muscle-invasive bladder cancerClinical TII:98, III:54, IV:18NeoGemcitabine + cisplatin (majority); MVAC or related regimens (platinum-based)4 (1–4) Takai (2021) [38]/ JapanRetro44 (44/0)37 (19–80)Testicular cancerII:10, III:9, IV:18AdjuvantBEP/EP; VIP/TIP/VeIP; GEMOX; other cisplatin-based regimens4 (1–14) Mitsui (2019) [61]/ JapanRetro50 (50/0)34 (16–67)Testicular cancerClinical S0:1, I: 17, II :17, III: 8NRNR2–4 Semerad (2022) [62]/ CzhechiaPro30 (30/0)37 (22–60)Testicular cancerI: 7, II: 9, III: 7NRBEP (bleomycin, etoposide, cisplatin) (platinum-based)3 (2–4) Buxton (2024) [63]/ USARetro182 (182/0)31 (26–39)Testicular cancerStage 1–IS:33; II–IIC:74; III–IIIC:71; Unknown:4NRBEP, EP, VIP, or related cisplatin-based regimens3 Lung cancer and pleural mesothelioma Chemotherapy Goncalves (2018) [37]/USARetro88 (42/46)65 (55–71)Non-Small Cell Lung cancerI:12, II:15, III:58, IV:3NeoTaxane- or gemcitabine-based;N = 7, 8% received bevacizumab2–6 Stene (2015) [64]/NorwayPro35 (18/17)67 (± 7)Non-Small Cell Lung cancerIIIB: 6IV: 29PalliativeCarboplatin; vinorelbine; gemcitabine (platinum-based)1–3 Kazemi-Bajestani (2019) [65]/CanadaPro50 (24/26)65 (± 8)Non-Small Cell Lung cancerIVPalliativeCarboplatin doublets: with vinorelbine, gemcitabine, paclitaxel, or pemetrexed (platinum-based)1–4 Nattenmüller (2017) [66]/GermanyRetro200 (130/70)62 (± 10)Non-Small Cell Lung cancerUICCI:3, II:10, III: 43, IV: 144NRCarboplatin/cisplatin with gemcitabine, vinorelbine, etoposide, pemetrexed, or others (platinum-based)1–8 Kidd (2024) [67]/UKRetro111 (91/20)69 (63–72)Pleural mesotheliomaI:45, II:22, III:12, IV:19NRCisplatin or carboplatin with pemetrexed (platinum-based)NR Kakinuma (2018) [35]/JapanRetro44 (31/13)67 (± 8)Non-Small Cell Lung cancerIVNR(Only chemotherapy cohort) Carboplatin or cisplatin with pemetrexed, gemcitabine, paclitaxel/nab-paclitaxelNRImmunotherapy Khan (2023) [68]/AustraliaRetro97 (55/42)68 (± 10)Non-Small Cell Lung cancerIII:15IV:81PalliativeImmune checkpoint inhibitorsNRChemotherapy + immunotherapy Chaunzwa (2024) [20]/USARetro1,791 (913/878)65 (26–84)Non-Small Cell Lung cancerAdvanced or metastaticPalliativeChemotherapy (SOC); chemo-immunotherapy; or immunotherapy monotherapyNRChemotherapy + targeted therapy Cortellini (2018) [69]/ItalyRetro81 (53/28)68 (39–90)Non-Small Cell Lung cancerIVNRPlatinum doublets (pemetrexed, gemcitabine, paclitaxel + bevacizumab); or single agents (carboplatin, docetaxel, vinorelbine) (platinum-based)NR Gastric, esophagogastric and esophageal cancers Chemotherapy Sato (2024) [70]/JapanPro50 (39/11)64 (38–75)Locally advanced gastric cancercStage (14th)II:1, III:27, IV:22NeoDocetaxel + cisplatin + S-1 (DCS, platinum-based)2 Juez (2024) [71]/SpainRetro61 (34/27)68 (± 9)Locally advanced gastric cancerAJCCI:10, II:30, III:21NeoFLOT (docetaxel, oxaliplatin, leucovorin, 5-FU)4 Li (2024) [17]/ChinaRetro345 (109/236)61 (± 12)Advanced gastric cancerII:189, III:156NeoSOX (S-1 + oxaliplatin), XELOX (capecitabine + oxaliplatin), or FOLFOX (5-FU + leucovorin + oxaliplatin)4–6 Mirkin (2017) [72]/USARetro36 (13/23)65 (NR)Advanced gastric cancerNRNeoEpirubicin + cisplatin + 5-FU (epirubicin-based)NR Matsuura (2020) [73]/JapanRetro41 (28/13)72 (48–82)Advanced gastric cancerII:9, III:25, IV:7NeoS-1 + cisplatin; S-1 + docetaxel + cisplatin; or S-1 + oxaliplatin (platinum-based)2 (1–4) Horii (2022) [74]/JapanRetro38 (28/10)64 (44–78)Advanced gastric cancerClinical stage (stage J)II: 18, III: 13: IV:7NeoDCS (S-1 + cisplatin + docetaxel), DS (S-1 + docetaxel), XP (capecitabine + cisplatin), SP (S-1 + cisplatin, platinum-based)2 Sugiyama (2018) [16]/JapanRetro118 (69/48)64 (27–84)Advanced gastric cancerMetastaticPalliativeFluoropyrimidine + cisplatin; oxaliplatin (platinum-based)NR Park (2020) [39]/KoreaRetro111 (80/31)65 (31–87)Advanced gastric cancerIVPalliativeS-1 + cisplatin; XP; FOLFOX/XELOX; S-1 or capecitabine alone;N = 11, 9.9% received trastuzumab + XPNR Palmela (2017) [75]/PortugalRetro47 (32/15)68 (± 10)Esophagogastric cancerIII:5, IV:42NeoECF, EOF, EOX, ECX, XELOX, FOLFOX, capecitabine, or DCF2 den Boer (2020) [76]/ NRRetro199 (158/41)66 (28–80)Esophagogastric cancerClinical T T1:1, T2:62, T3:107, T4: 28NeoECX, CX, or related platinum–fluoropyrimidine regimens1–4 Rinninella (2021) [77]/ItalyRetro26 (18/8)63 (± 11)Esophagogastric cancerPathologic stage:0: 0, 1: 4, 2: 5, 3: 6, 4: 9, missing: 2NeoFLOT (perioperative regimen)4 Fujihata (2021) [78]/JapanRetro99 (89/10)68 (61–72)Esophagogastric cancerPathologic stage 0:3, I:12, II:41, III:43Neo5-FU + cisplatin (FP) or docetaxel + cisplatin + 5-FU (DCF, platinum-based)1–2 Dijksterhuis (2019) [79]/NetherlandsRetro88 (66/22)63 (56–69)Esophagogastric cancerMetastaticPalliativeCapecitabine + oxaliplatin (CAPOX, platinum-based)1–6 Hacker (2022) [80]/GermanyPro509 (387/122)< 65: 375≥ 65: 134Esophagogastric cancerMetastaticPalliativePlatinum–fluoropyrimidine chemotherapy (platinum-based)NR Awad (2012) [81]/ UKRetro47 (34/17)63 (± 12)Esophagogastric cancerT3 N0/1: 23, T2 N0/1: 10, T3 N2/3: 5, T1 N0: 5, T4 N1: 1, Tis N0: 1, No residual tumor: 2NeoEpirubicin + cisplatin + 5-FU; cisplatin + 5-FU; capecitabine + cisplatin; epirubicin + oxaliplatin1–4 Onishi (2024) [82]/ JapanRetro215 (178/37)67 (40–81)Esophageal cancerDCF: II:11, III:58; CF: II:80, III:66NeoDocetaxel + cisplatin + 5-FU (DCF) or cisplatin + 5-FU (CF, platinum-based)2–3 Harada (2025) [83]/JapanRetro69 (53/16)73 (4)Esophageal cancerClinical stage IB, II, III, or IV without distant organ metastasisNeoCisplatin + 5-FU (FP), FOLFOX, or DCF (platinum-based)2–4 Yip (2014) [84]/ UKRetro35 (30/5)63 (34–78)Esophageal cancerII:10, III:23, IV:2.NeoECF/ECX (epirubicin, cisplatin, 5-FU/capecitabine) (platinum-based)3 (1–6) Miyata (2017) [85]/ JapanRetro94 (76/18)64 (± 9)Esophageal cancerI:5, II:24, III:54, IV:11NeoAdriamycin + cisplatin + 5-FU (ACF) or docetaxel + cisplatin + 5-FU (DCF, platinum-based)2 (1–3) Ishida (2019) [86]/JapanRetro165 (144/21)65 (NR)Esophageal cancerClinical T I+II:14; III+IV:29NeoDCF vs. ACF (platinum-based)2Chemotherapy + immunotherapy Zhao (2024) [87]/ChinaRetro85 (69/16)67 (59–71)Esophageal cancerII:37, III:40, IV:8NeoPD-1 inhibitor (camrelizumab) + platinum + paclitaxel (platinum-based)2–4 Ying (2025) [88]/ChinaRetro83 (81/2)68 (49–87)Esophageal cancerLymph node metastasis: 98%Distant metastasis: 37%NRPD-1 inhibitor + chemotherapyNR Ovarian cancer Chemotherapy Wood (2023) [89]/USARetro174 (0/174)64 (± 10)Ovarian cancerII:1, III:108, IV:65NeoNRNR Ubachs (2020) [90]/NetherlandsRetro212 (0/212)61 (± 8)Ovarian cancerFIGOIII:212NeoCarboplatin + paclitaxel.2 Yoshino (2020) [91]/JapanRetro60 (0/60)64 (43–81)Ovarian cancerFIGOIII:36, IV:24NeoCarboplatin + paclitaxel/docetaxel/irinotecan (platinum-based)4 (2–6) Del Grande (2021) [92]/SwitzerlandRetro25 (0/25)65 (± 11)Ovarian cancerFIGOI: 1, II:3, III:45, IV:20.NeoPlatinum-basedNR Van der Zanden (2021) [93]/NetherlandsRetro111 (0/111)77 (74–79)Ovarian cancerFIGOIII:73, IV:38NeoPlatinum-based2 Studies including multiple cohorts with multiple cancers Chemotherapy Toama (2022) [94]/ USARetro474 (161/313)61 (53–68)Breast cancer (n = 192)Lymphoma (n = 184)Sarcoma (n = 98)I:49, II:73, III:84, IV:234, Missing: 34NRAnthracycline-based chemotherapyNRImmunotherapy Loosen (2021) [19]/GermanyPro88 (48/42)67 (34–87)Lung: 39.8%Malignant melanoma: 15.9%Urothelial cancer: 13.6%GI cancer: 13.6%Head and neck cancer: 8.0%Others: 9.1%UICC III: 7%, UICC IV: 93%NRNivolumab, pembrolizumab, nivolumab + ipilimumab, or othersNRChemotherapy + immunotherapy Roeland (2021) [95]/USAPro38 (20/18)62 (± 2)Gastrointestinal: 71%Lung: 13%Gynecologic: 8%Head and neck: 8%Other: 5%MetastaticNRNRNRChemotherapy + targeted therapy Oflazoglu (2020) [96]/TurkeyPro276 (122/154)57 (± 11)Breast: 33.7%Colorectal: 26.8%Pancreaticobiliary: 10.9%Urological: 8.7%Gastroesophageal: 7.6%Lung: 3.6%Head and neck: 2.5%Others: 6.2%Metastasis status:No: 229Yes: 47NRSite-specific regimens: breast (AC + paclitaxel ± trastuzumab), colorectal (XELOX, FOLFOX ± bevacizumab/panitumumab, capecitabine), pancreaticobiliary (cisplatin-based, gemcitabine, FOLFIRINOX), urological (platinum-based), gastroesophageal (cisplatin-based, XELOX), lung (cisplatin-based, carboplatin + paclitaxel), head & neck (cisplatin-based), othersNR Melanoma Immunotherapy Daly (2017) [97]/IrelandRetro84 (52/32)54 (43–66)MelanomaM1a: 9, M1b: 9 , M1c: 66NRIpilimumab (CTLA-4 inhibitor)4 Liver cancer Immunotherapy Chen (2025) [98]/ChinaRetro85 (68/17)62 (± 12)Intermediate and advanced liver cancerChild-Pugh score:A: 48, B: 31, C: 6NRPD-1 inhibitors (sintilimab, tislelizumab, camrelizumab, pembrolizumab) or PD-L1 inhibitors (atezolizumab, durvalumab)NRImmunotherapy + targeted therapy Shigefuku (2024) [99]/Japan (ATZ-BEV)Retro56 (45/11)74 (69–80)Advanced liver cancerChild–Pugh score:5: 52, 6: 37, 7: 7, 8: 1NRAtezolizumab + bevacizumabNR Colorectal cancer Chemotherapy Okuno (2019) [100]/USARetro169 (97/72)56 (± 12)Colorectal cancerNRNeoOxaliplatin-based; irinotecan-based; multiple regimens (platinum-based)6 (2–24)Chemotherapy + targeted therapy Nozawa (2021) [101]/JapanRetro98 (58/40)65 (28–88)Colorectal cancerIVNeo ConversionPalliativeFOLFOX; CAPOX; SOX; FOLFIRI; FOLFOXIRI; IRIS (platinum- or irinotecan-based ± targeted)NR Palle (2016) [102]/DenmarkPro18 (10/8)67 (± 6)Colorectal cancerPatients with tumor stage T3–4, N0–N1 and/or V0–V1.AdjuvantCapecitabine; capecitabine + oxaliplatin; capecitabine + oxaliplatin + bevacizumab (platinum-based)1–8 Huemer (2019) [103]/AustriaRetro10 (6/4)65 (42–81)Colorectal cancerMetastaticNRTAS-102 (trifluridine/tipiracil); regorafenibNR Blauwhoff-Buskermolen (2016) [21]/NetherlandsPro67 (42/25)66 (± 11)Colorectal cancerMetastaticPalliativeCAPOX ± bevacizumab; FU + oxaliplatin ± bevacizumab; capecitabine + irinotecan; irinotecan monotherapy; capecitabine ± bevacizumab (platinum-based)NR Gallois 2021 [104]/FrancePro149 (82/67)70 (NR)Colorectal cancerMetastaticNR5-FU-based regimens: oxaliplatin-based; irinotecan-based; single agent; doublet; triplet; ± bevacizumab; ± cetuximab/panitumumab (platinum/irinotecan-based)NR Breast cancer Chemotherapy Jang (2022) [105] (AC-T)/South KoreaRetro214 (0/214)53 (± 11)Breast cancerI:3, II:153, III:51NeoAnthracycline–cyclophosphamide → taxane (AC-T)6 Campbell (2007) [106]/CanadaPro10 (0/10)47 (± 6)Breast cancerI: 2, II-IIIA: 8AdjuvantCEF or AC (anthracycline-based)5 Jung (2020) [107]/South KoreaPro37 (0/37)51 (± 9)Breast cancerI:13, II:22, III:2AdjuvantAC or TC (docetaxel + cyclophosphamide)NRChemotherapy + immunotherapy (+ targeted therapy) Camilleri (2024) [108]/FranceRetro111 (0/111)60 (12)Breast CancerMetastaticPalliativeNot otherwise specifiedNRChemotherapy + targeted therapy Zhang (2024) [109]/ChinaRetro43 (0/43)51 (± 10)Breast cancerII-IIINeoTaxanes and anthracyclines. HER2-positive received targeted therapy with trastuzumab and pertuzumab.6–8 Rossi (2023) [110]/ItalyRetro52 (0/52)37 (± 5)Breast cancerNRNeoEpirubicin + cyclophosphamide (EC) → paclitaxel; some with carboplatin + paclitaxel; HER2+ with trastuzumab4 Karaca (2024) [111]/TurkeyRetro226 (0/226)50 (± 12)Breast cancerII–IIINeoAnthracycline–cyclophosphamide → paclitaxel/docetaxel; HER2+ with trastuzumab/pertuzumabNR Amitani (2022) [18]/JapanRetro141 (0/141)53 (± 10)Breast cancerII:89, III:52NeoFEC/EC → taxane (docetaxel or paclitaxel); HER2+ with trastuzumab4 Lee (2021) [112]/ South KoreaRetro246 (0/246)48 (42–54)Breast cancerI:9, II:123, III:114NeoAC ± paclitaxel; subset with trastuzumabNR Rossi (2020) [113]/ ItalyRetro101 (0/21)56 (± 11)Breast cancerNRNeoEC → paclitaxel; subset with pertuzumab + trastuzumab4–6 Mazzuca (2018) [114]/ ItalyRetro21 (0/21)54 (39–72)Breast cancerI:7, II:11, III:3AdjuvantFEC/EC; EC + taxane; ~43% with trastuzumab4 Lymphoma Chemotherapy + targeted therapy Xiao (2016) [115]/ USARetro342 (331/11)63 (± 11)Diffuse large B-cell lymphomaClinical stage I+II:145; III+IV:195, 2: missingNRCHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) ± rituximab (R-CHOP)NRTNM: Tumor, Node, and Metastasis; N: number of participants at baseline; SD: standard deviation; NR: not reported; Neo: neoadjuvant treatment; Adjuvant: adjuvant chemotherapy; Retro: Retrospective cohort; Pro: prospective cohort.arounded to the nearest whole number. Study characteristics. TNM: Tumor, Node, and Metastasis; N: number of participants at baseline; SD: standard deviation; NR: not reported; Neo: neoadjuvant treatment; Adjuvant: adjuvant chemotherapy; Retro: Retrospective cohort; Pro: prospective cohort. rounded to the nearest whole number. PRISMA flowchart. Across cancer types, systemic treatment with chemotherapy and/or immunotherapy (± targeted therapy) was associated with a decline in SMM (59 studies; n = 6,373; SMC = –0.24, 95% CI: –0.29 to –0.20; I2 = 92%), corresponding to an unweighted median of −5% (IQR: −7 to −2) over a median interval of 90 (IQR: 71–129) days. A total of 30 (38%) studies did not report the interval between SMM assessments. The 95% prediction interval (–0.56 to 0.08) indicated between-study variability in the magnitude and direction of effects (Figure 2 and Table 2). Most studies (n = 70, 89%) assessed SMM using CT imaging, predominantly at the third lumbar vertebra (L3) (n = 63, 73%). The SMI was the most common outcome measure, reported in 51 (65%) studies. Assessment methods and outcomes are presented in Table 3. No evidence of publication bias was identified by Egger’s test (p = 0.229) or visual inspection of the funnel plot (Figure S1). Among the five studies [28–32] reporting the correlation coefficient between repeated measurements, the mean correlation was 0.88. Sensitivity analyses using assumed correlations of 0.1, 0.5, and 0.9 yielded nearly identical pooled estimates, with only slight increases in SE. Table 2Meta-analyses of the change in skeletal muscle mass during systemic cancer treatment.Meta-analyses N SMC (95% CI)Comparisons I 2 P Primary analysis6,373−0.24 (−0.29 to −0.20)59 [16–19, 21, 23, 35, 37–41, 53, 54, 56–59, 61–63, 65–68, 71,, 73–83, 85–87, 89–91, 93–95, 97, 98, 100, 104–111, 114]92%- Subgroup analysesa Treatment type Chemotherapy5,169−0.27 (−0.32 to −0.22)45 [16, 17, 23, 35, 37–41, 53, 54, 56–59, 61–63, 65–67, 71, 73–83, 85, 86, 89–91, 93, 94, 100, 105–107]92%- IO287−0.15 (−0.30 to 0.01)4 [19, 68, 97, 98]63%- Chemo-IO234−0.05 (−0.13 to 0.04)3 [87, 95, 108]0%- Chemo-TT683−0.23 (−0.38 to −0.08)7 [18, 21, 104, 109–111, 114]91%-Treatment setting Neoadjuvant3,321−0.22 (−0.28 to −0.17)32 [17, 18, 23, 37, 53, 56–59, 71, 73–78, 81–83, 85–87, 89–91, 93, 100, 105, 109–111]90%- Adjuvant102−0.14 (−0.61 to 0.33)4 [38, 106, 107, 114]91%- Palliative1,393−0.31 (−0.47 to −0.16)10 (21, 39–41, 54, 65, 68, 79, 80, 108)95%- Not reported1,557−0.25 (−0.34 to −0.17)13 [16, 19, 35, 61–63, 66, 67, 94, 95, 97, 98, 104]86%-Assessment tool CT SMI (cm/m2)4,670−0.23 (−0.29 to −0.18)39 [16–19, 23, 35, 38–40, 53, 54, 56–59, 63, 66–68, 71, 75–80, 82, 83, 87, 89, 90, 93, 98, 100, 104, 105, 108, 114]92%- CT SMA (cm2)424−0.25 (−0.42 to −0.08)8 [21, 41, 61, 65, 81, 91, 95, 97]89%- CT PMI (cm/m2)718−0.27 (−0.49 to −0.05)4 [73, 74, 86, 94]93%- CT PMA (cm2)269−0.40 (−0.66 to −0.15)2 [109, 111]0% BIA (kg)154−0.09 (−0.49 to 0.29)3 [62, 85, 107]67%- Sensitivity analyses Estimated meansb3,968−0.22 (−0.27 to −0.17)40 [16, 18, 21, 23, 35, 38, 53, 54, 56, 58, 59, 62, 65, 66, 68, 75–77, 79–83, 85, 86, 89, 90, 95, 97, 98, 100, 104–111]88%-Non-small trialsc2,286−0.24 (−0.32 to −0.16)20 [16–18, 39, 40, 63, 66, 76, 80, 82, 86, 89, 90, 93, 94, 100, 104, 105, 108, 111)96%-Correlation coefficient estimationd6,373−0.25 (−0.29 to −0.20)59 [16–19, 21, 23, 35, 37–41, 53, 54, 56–59, 61–63, 65–68, 71, 73–83, 85–87, 89–91, 93–95, 97, 98, 100, 104–111, 114]72%-Prospective study design917−0.18 (−0.27 to −0.09)9 [19, 21, 62, 65, 80, 95, 104, 106, 107]61%-By sex (male)438−0.21 (−0.42 to −0.01)8 [16, 21, 39, 55, 79, 101, 102, 104]92%By sex (female)263−0.40 (−0.69 to −0.12)8 [16, 21, 39, 55, 79, 101, 102, 104]93% Funnel plot asymmetry 0.229N: number with complete data; SMC: standardized mean change; CI: confidence interval; I2, heterogeneity; Chemo: chemotherapy; TT: targeted therapy; IO: immunotherapy; Multiple: studies combining multiple diagnoses; CT: computed tomography; CM: centimeter; SMI: Skeletal Muscle Index; SMA: Skeletal Muscle Area; PMI: Psoas (or Pectoralis) Muscle Index; PMA: Psoas (or Pectoralis) Muscle Area; BIA: bioimpedance analysis; Kg.: kilograms.aOnly subgroups with ≥ 2 studies are presented.bEstimated means: Sensitivity analysis excluding studies where means and SDs were derived from medians/IQRs using Wan et al.’s method.cNon-small studies: Sensitivity analysis excluding studies with n ≤ 100.dCorrelation coefficient estimation: Sensitivity analysis with imputed correlation coefficient (r = 0.5). Meta-analyses of the change in skeletal muscle mass during systemic cancer treatment. N: number with complete data; SMC: standardized mean change; CI: confidence interval; I2, heterogeneity; Chemo: chemotherapy; TT: targeted therapy; IO: immunotherapy; Multiple: studies combining multiple diagnoses; CT: computed tomography; CM: centimeter; SMI: Skeletal Muscle Index; SMA: Skeletal Muscle Area; PMI: Psoas (or Pectoralis) Muscle Index; PMA: Psoas (or Pectoralis) Muscle Area; BIA: bioimpedance analysis; Kg.: kilograms. Only subgroups with ≥ 2 studies are presented. Estimated means: Sensitivity analysis excluding studies where means and SDs were derived from medians/IQRs using Wan et al.’s method. Non-small studies: Sensitivity analysis excluding studies with n ≤ 100. Correlation coefficient estimation: Sensitivity analysis with imputed correlation coefficient (r = 0.5). Table 3Changes in skeletal muscle mass during systemic cancer treatment for studies reporting continuous pre- and post-treatment skeletal muscle mass.Author (year)*Assessment methodOutcome measureBody segmentDays between measurementsPre-treatment muscle massPost-treatment muscle massChange mean or medianChange % P Mean or MedianSD or min–maxMean or medianSD or min–max Pancreatic cancer Chemotherapy Shimura (2023) [51]CTSMIL3NRM: 45.70F: 35.70M: 9.70F: 5.70M: 41.10F: 34.90M: 8.80F: 6.10NR−10.07%**−2.24%**M: < 0.01F: 0.153 Griffin (2019) [53]CTSMIL312845.608.7042.309.30NR−7.24%**< 0.01 Lee (2024) [40]CTSMIL36043.1039.10–49.9040.1035.90–45.00NR−6.96%**< 0.01 Rollins (2016) [54]CTSMIL36042.208.6039.808.00NR−5.69%**0.060 Aberle (2025) [55]CTSMIL390M: 49.10F: 39.20M: 8.80F: 4.70M: 45.90F: 35.50M: 8.70F: 4.20M: −2.20F: −3.50−6.52%**−9.44%**M: 0.002F: 0.001 Davis (2025) [56]CTSMIL371.5046.607.6546.007.94−0.90−1.29%**NR Uemura (2020) [23]CTSMIL37140.207.3036.306.30NR−7.90%NR Lee (2019) [41]CTSMML360100.4020.3086.4020.20NR−13.94%**< 0.001 Urological cancer Chemotherapy Miyake (2018) [57]CTSMIL3NR53.7048.80–58.0052.8048.80–55.20NR−1.68%**0.016 MacDonald (2024) [58]CTSMIL36952.4010.8050.1010.10−2.2 (± 3.2)−4.39%**< 0.001 Rimar (2018) [59]CTSMIL311049.10NR44.50NRNR−6.40%< 0.01 Lyon (2019) [60]CTSMIL313950.70NR48.60NRNR−4.14%**NR Takai (2021) [38]CTSMIL318251.6028.60–70.6045.6032.40–60.60NR−11.63%**NR Mitsui (2018) [61]CTSMAL321150.2076.30–206.90140.5073.40–200.70NR−8.50%NR Semerad (2022) [62]BIASMM kgWhole bodyNR32.674.5931.344.65NR−1.33%**0.005 Buxton (2024) [63]CTSMIL311457.5052.50–62.9055.0049.30–60.70−3.60(−6.50 to 0.40)−6.10%< 0.001 Lung cancer and pleural mesothelioma Chemotherapy Goncalves (2018) [37]CTLMVT10–L1NR296.50249.00–369.00283.00241.00–337.00NR−4.39%NR Stene (2015) [64]CTSMAL388121.9030.80117.40NRNR−3.69%**< 0.01 Kazemi−Bajestani (2019) [65]CTSMAL3112130.5036.00120.9029.70NR−8.90%< 0.01 Nattenmüller (2017) [66]CTSMIL2–L312945.708.7044.308.60NR−3.06%**< 0.01 Kidd (2024) [67]CTSMIL3NR50.5044.00–57.8047.0042.00–57.00NR−6.93%**< 0.01 Kakinuma (2018) [35]CTSMIL313244.807.3040.406.60NR−9.82%**NRImmunotherapy Khan (2023) [68]CTSMIL3NR43.808.5043.709.40−0.10−0.23%**NR Gastric, esophagogastric and esophageal cancers Chemotherapy Sato (2024) [70]CTSMIL3NR47.90NR44.10NRNR−3.40%NR Juez (2024) [71]CTSMIL3NR46.6940.10–55.2045.5239.00–51.00NR−2.57%< 0.01 Li (2024) [17]CTSMIL3NR40.4032.10–44.2039.6030.50–43.10−1,3−2.60%< 0.01 Matsuura (2020) [73]CTPMIL3NR4.771.114.501.20NR−5.93%< 0.01 Horii (2022) [74]CTPMIUmbilicus level606.57*3.84–9.746.21*3.30–8.89NR−5.48%**< 0.01 Sugiyama (2018) [16]CTSMIL342939.008.0236.408.05NR−7.00%< 0.01 Park (2020) [39]CTSMIL3NR40.709.0035.308.30NR−11.30%< 0.01 Palmela (2017) [75]CTSMIL386.4048.209.6045.309.50NR−6.02%**< 0.01 Boer (2020) [76]CTSMIL310551.8710.3149.199.71NR−5.17%**< 0.01 Rinninella (2021) [77]CTSMIL395.5048.749.7646.529.80NR−4.55%**< 0.01 Fujihata (2021) [78]CTSMIL3NR40.6636.3–46.6139.3335.70–46.13NR−1.87%NR Dijksterhuis (2019) [79]CTSMIL379.0046.909.9044.4010.00NR−5.33%**< 0.01 Hacker (2022) [80]CTSMIL38461.629.4459.2NRNR−3.93%**NR Awad (2012) [81]CTSMAL3107140.0031.70130.5028.00NR−6.79%**< 0.01 Onishi (2024) [82](DCF treatment)CTSMIL3NR41.507.6040.307.20NR−2.40%NR Onishi (2024) [82](CF treatment)CTSMIL3NR40.408.3038.407.50NR−4.40%NR Harada (2025) [83]CTSMIL3NR43.107.8040.907.60NR−5.10%**< 0.01 Miyata (2017) [85]BIASMMWhole body7725.004.8024.904.80NR−0.40%**NR Ishida (2019) [86]CTPMIL3NR7.172.016.971.86NR−2.79%**< 0.01Chemotherapy + immunotherapy Zhao (2024) [87]CTSMIL3NR45.1042.25–49.7044.8042.17–48.83−0.12−0.67%**0.146 Ying (2025) [88]CTSMIL3NR69.20NR65.40NRNR−5.50%**NR Ovarian cancer Chemotherapy Wood (2023) [89]CTSMIL49338.307.9037.807.90NR−1.31%**NR Ubachs (2020) [90]CTSMIL36039.605.4038.105.00NR−5.90%NR Yoshino (2020) [91]CTSMAL3NR87.2052.50–129.0083.4059.20–122.00NR−4.36%**0.019 Del Grande (2021) [92]CTSMIL3NR48.008.9045.0068.10NR−6.25%**0.052 Van der Zanden (2021) [93]CTSMIL36639.1036.30–43.1037.2034.70–40.50NR−6.00%0.001 Studies including multiple diagnoses Chemotherapy Toama (2022) [94]CTPMIT2–T3NR5.804.90–7.705.204.40–6.40NR−10.50%NRImmunotherapy Loosen (2021) [19]CTSMIL38476.7946.00–124.9074.0244.20–113.50NR−3.61%**NRChemotherapy + immunotherapy Roeland (2021) [95]CTSMANR90132.1035.90132.8036.30NR+0.53%**0.648 Melanoma Immunotherapy Daly (2017) [97]CTSMAL3146151.3037.20144.2037.30NR−4.69%**NR Liver cancer Immunotherapy Chen (2025) [98]CTSMIL39042.847.8741.638.11−1.21 (± 3.72)−2.82%**NR Immunotherapy + targeted therapy Shigefuku 2024 [99]CTPMIL32135.004.10–6.304.91NRNR−1.20%0.06 Colorectal cancer Chemotherapy Okuno (2019) [100]CTSMIL3NR51.2010.6050.6010.70NR−1.17%**0.033Chemotherapy + targeted therapy Nozawa (2021) [101] (Conversion)CTSMIL3127M: 38.8F: 30.208.106.10M: 42.20F: 32.256.205.75NR+9.40%NR Nozawa (2021) [101] (NACT)CTSMIL375M: 41.70F: 34.007.306.60M: 41.20F: 30.606.606.10NR−5.90%NR Nozawa (2021) [101] (Palliation)CTSMIL3118M: 39.90F: 28.007.104.90M: 38.50F: 26.605.006.70NR−3.70%NR Palle (2016) [102]BIASMMWhole body27.60M: 37.40F: 25.202.502.70M: 37.40F: 25.002.803.20NR0.00%**0.9440.156 Blauwhoff-Buskermolen (2016) [21]CTSMAL378138.6032.10131.9031.70NR−6.10%< 0.01 Gallois (2020) [104]CTSMIL36041.008.8039.208.090NR−4.39%**NR Breast cancer Chemotherapy Jang (2022) [105] (AC-T)CTSMIL316142.405.4042.405.90−0.220.00%**0.83 Campbell (2007) [106]DXASMMWhole body10543.304.2043.504.50NR+0.46%**0.65 Jung (2020) [107]BIASMMWhole bodyNR39.414.8939.425.15NR+0.03%**0.187Chemotherapy + immunotherapy (+ targeted therapy) Camilleri (2024) [108]CTSMIL318240.806.4040.406.40NR−0.98%**0.09Chemotherapy + targeted therapy Zhang (2024) [109]CTPEMATH1215226.096.8023.665.98NR−9,31%**< 0.00 Rossi (2023) [110]MRIPMAT4–T51589.702.608.702.20−1.41−10.31%**< 0.01 Karaca (2024) [111]CTPMA (mm2)L3NR502.80118.00454.30115.10NR−9.65%**< 0.01 Amitani (2022) [18]CTSMIL3NR46.507.6046.308.00NR−0.43%**NR Rossi (2020) [113]MRIPMAPectoralis (sternal angle)166.808.12NR7.03NRNR−13.33%**< 0.01 Mazzuca (2018) [114]CTSMIL3NR39.2031.60–52.9039.2031.60–52.90NR0.00%**NR Lymphoma Chemotherapy + targeted therapy Xiao (2016) [115]CTSMAL3NR173.6NR168.8NR−4.8−2.8%NR*Table 3 includes only studies reporting mean or median skeletal muscle mass values. Studies reporting solely the prevalence of low SMM are presented in Table S14.** Percentage change was calculated by the present authors. SD: standard deviation; CT: computerized tomography; MRI: magnetic resonance imaging; DXA: dual-energy X-ray absorptiometry; BIA: bioelectrical impedance analysis; SMI: skeletal muscle index, cm2/m2; SMM: skeletal muscle mass, kg; SMA: skeletal muscle area, cm2; PEMA: pectoralis muscle area, cm2; PMI: psoas muscle index, cm2/m2; PMA: psoas muscle area, cm2; LMV: lumbar muscle volume, cm3; SMI inclination: (SMI-change/SMI)/duration; M: male. F: female. NR: not reported. Changes in skeletal muscle mass during systemic cancer treatment for studies reporting continuous pre- and post-treatment skeletal muscle mass. Table 3 includes only studies reporting mean or median skeletal muscle mass values. Studies reporting solely the prevalence of low SMM are presented in Table S14. Percentage change was calculated by the present authors. SD: standard deviation; CT: computerized tomography; MRI: magnetic resonance imaging; DXA: dual-energy X-ray absorptiometry; BIA: bioelectrical impedance analysis; SMI: skeletal muscle index, cm2/m2; SMM: skeletal muscle mass, kg; SMA: skeletal muscle area, cm2; PEMA: pectoralis muscle area, cm2; PMI: psoas muscle index, cm2/m2; PMA: psoas muscle area, cm2; LMV: lumbar muscle volume, cm3; SMI inclination: (SMI-change/SMI)/duration; M: male. F: female. NR: not reported. Meta-analyses of the change in skeletal muscle mass during systemic cancer treatment. In exploratory subgroup analyses by treatment type, the largest decline was observed in chemotherapy (45 studies; n = 5,169; SMC: –0.27, 95% CI: –0.32 to –0.22; I2 = 92%), followed by chemotherapy + targeted therapy (7 studies; n = 683; SMC: –0.23, 95% CI: –0.38 to –0.08; I2 = 91%) and immunotherapy (4 studies; n = 287; SMC: –0.15, 95% CI: –0.30 to 0.01; I2 = 63%). No loss of SMM with combined chemotherapy and immunotherapy was observed. By treatment setting, the largest decline was observed for palliative treatment (10 studies; n = 1,393; SMC: –0.31, 95% CI: –0.47 to –0.16; I2 = 95%). The magnitude of SMM loss varied across cancer types, but differences were not statistically significant (p = 0.193). Still, overlapping confidence intervals together with high heterogeneity limit direct comparisons between individual groups. The largest and most consistent reductions were observed among patients with pancreatic (6 studies; n = 828; SMC = –0.41, 95% CI: –0.63 to –0.19; I2 = 94%), urological (7 studies; n = 401; SMC = –0.30, 95% CI: –0.42 to –0.18; I2 = 66%) and lung cancer (6 studies; n = 552; SMC = –0.30, 95% CI: –0.54 to –0.06; I2 = 94%), corresponding to unweighted mean declines of –8%, –6%, and –5%, respectively. No significant change in SMM was observed in breast cancer, colorectal cancer, or studies including multiple cancer types. Overall heterogeneity remained considerable, but exploratory sensitivity analyses did not alter the pooled estimates, except for the sex-stratified subgroup analysis. Males (n = 438) had smaller SMM losses (SMC: –0.21, 95% CI: –0.42 to –0.01; I2 = 92%) than females (n = 263; SMC: –0.40, 95% CI: –0.69 to –0.12; I2 = 93%). All subgroup analyses are presented in Table 2 and Tables S6–S13. Thirty-one studies (n = 5,376 patients) reported the prevalence or percentage of low SMM during treatment (Table S14). In total, 20 distinct definitions of low SMM were identified. The SMI measured by CT at the L3 was the dominant criterion used in 28 (90%) studies, typically expressed with sex-specific cut-offs between < 52–55 cm2/m2 for males and < 38–41 cm2/m2 for females. In 26 (84%) studies, the prevalence of low SMM increased during treatment, with the mean percentage rising from 43% to 51% (Table S14). Excluded studies on full-text screening are presented in Table S15. Discussion Findings from our comprehensive systematic review and meta-analysis indicate that declines in SMM are frequently observed during systemic cancer treatment, particularly during chemotherapy (± targeted therapy), although effect sizes were generally small per Cohen’s thresholds. Among the 59 studies included in the meta-analysis, 78% showed a decline in SMM. However, substantial between-study heterogeneity suggests that the magnitude of change varies considerably across clinical contexts. The wasting of SMM may have several possible mechanistic pathways. Cancer itself may directly and indirectly suppress muscle protein synthesis while increasing protein catabolism [29]. Tumor growth and systemic disease can create a chronic catabolic environment characterized by pro-inflammatory signaling, hormonal imbalance, altered metabolism, and energy deficit [3, 29]. Pro-inflammatory cytokines such as interleukin-6, interleukin-1β, tumor necrosis factor-α, and transforming growth factor-β promote proteolysis and inhibit muscle protein synthesis through activation of the ubiquitin, proteasome system and suppression of mTOR signaling [3, 30]. Chronic inflammation may also interact with hormonal changes such as reduced testosterone and insulin-like growth factor-1, to further impair anabolism [3, 30]. In a large cohort of patients with colorectal cancer (n = 2,470), the coexistence of low SMM and systemic inflammation more than doubled the risk of cancer-specific mortality (HR: 2.43, 95% CI: 1.79–3.29) [31]. These findings support the concept that cancer-related systemic inflammation may contribute to SMM loss, although the relative contribution of disease- versus treatment-related mechanisms cannot be disentangled in the included studies. The magnitude of SMM loss differed markedly by treatment type. Across (n = 45) cohorts with patients receiving chemotherapy (n = 5,169), the pooled SMC was –0.27 (95% CI:–0.32 to –0.22), with consistent direction of change across studies, albeit considerable heterogeneity. These findings align closely with the review by Jang et al. [32], who synthesized 15 studies (n = 2,662) and reported a mean absolute reduction of 2.72 cm2/m2 (95% CI: 1.77–3.67) in SMI during chemotherapy treatment [32]. Our study extends these observations across a broader range of cancer types, treatment settings and sample sizes. Across other treatment modalities, SMM loss was evident but varied in magnitude. Estimates for immunotherapy and combined treatment should be interpreted with caution due to the limited number of studies. The observed treatment differences likely reflect distinct biological and metabolic mechanisms. Cytotoxic chemotherapy, particularly platinum-based regimens, induces direct myotoxicity through mitochondrial dysfunction, oxidative stress, and activation of catabolic transcription factors, while concurrently suppressing anabolic pathways and increasing myostatin expression [3]. These molecular effects are further compounded by systemic inflammation, toxicities, and treatment-related symptoms such as nausea, fatigue and pain, which collectively can diminish nutrient intake and physical activity, and thus reinforcing a cycle of muscle disuse and atrophy [3, 33]. Immunotherapy could influence SMM though similar pathways as cytotoxic regimens through symptom burden, cytokine-driven inflammation, and physical inactivity rather than direct cellular injury [34]. Nonetheless, biological pathways underlying muscle depletion during immunotherapy remain incompletely understood and should be addressed through robust mechanistic and prospective longitudinal studies. Our findings did not indicate a clear additional effect of targeted agents on SMM loss when administered in combination with chemotherapy. This interpretation aligns with Kakinuma et al. [35], in a study of patients with advanced non-small cell lung cancer (NSCLC) (n = 65). In their cohort, chemotherapy led to a 10% decline in SMI (from 44.8 to 40.4 cm2/m2), whereas targeted therapy treatment did not reduce SMI [35]. Sex differences in SMM loss may partly explain these findings. In our study, six of eight studies (75%) in chemotherapy + targeted therapy subgroups included breast cancer cohorts. In Jang et al. [32] (males, n = 823; females, n = 352), absolute declines in SMI were 1.6 times greater in males than in females (–4.52 vs. –2.86 cm2/m2) in sex-stratified sub-analyses. By contrast, in our review, eight sex-stratified studies showed larger relative losses in females (SMC: –0.40; n = 263) than in males (SMC: –0.21; n = 438). However, these findings are based on a limited number of studies and should be interpreted cautiously. Thus, further elucidation of absolute and relative changes stratified by sex and treatment type is warranted to conclude whether sex-based differences exist. Across treatment settings, the greatest SMM loss was observed in patients in palliative treatment settings with SMC –0.31 (95% CI: –0.47 to –0.16). In agreement, a longitudinal cohort of (n = 3,075) community-dwelling older adults aged 70–79 years. Williams and colleagues [36] found that among the (n = 515) adults who developed cancer, the loss in SMM was most pronounced among people with metastatic disease, indicating that cancer stage and treatment setting amplify age-related SMM wasting [36]. Still, among the seven studies showing the most pronounced declines (SMC from –0.69 to –0.50), neoadjuvant [23, 37], adjuvant [38], palliative [39–41], and unclassified treatment [35] settings were present. This distribution suggests that SMM loss may occur across different treatment settings, although the magnitude of change varies substantially between studies. Clinical implications of SMM loss Reduced overall and progression-free survival could, in part, be explained by reduced treatment tolerance in patients with SMM wasting, potentially reflecting a pharmacokinetic mismatch driven by current dosing practices. Most cytotoxic agents are dosed by body surface area, which does not account for lean versus fat mass distribution [1, 42]. Because anticancer drugs distribute primarily into metabolically active tissues, patients with low SMM and high body surface area may experience a higher relative dose per unit of lean tissue, predisposing them to toxicity and dose reductions [1, 42]. Conversely, individuals with preserved SMM have greater drug clearance and fewer adverse effects [43]. In the recent phase II Randomized LEANOX Trial, Assenat et al. [44] found that using an lean body mass-based oxaliplatin dose significantly reduced peripheral neurotoxicity and improved quality of life without affecting relapse-free and overall survival [44]. This indicates that SMM loss may contribute directly to treatment-related toxicity. In healthy individuals around age 75, muscle strength declines by roughly 3–4% per year in men and 2–3% per year in women [46]. Studies evaluating both strength and SMM within the same cohorts further indicate that strength decreases at a rate two to five times greater than SMM [46]. Importantly, loss of muscle strength and power is a more consistent predictor of disability and mortality than loss of SMM in older adults [46, 47]. Still, prospective studies directly linking cancer-related SMM and strength loss to clinical outcomes remain scarce. Physical exercise remains the most potent non-pharmacological strategy to preserve SMM and strength [45], and progressive resistance training can provide an anabolic stimulus in patients with cancer. A meta-analysis of 34 randomized trials demonstrated a mean gain of 0.85 kg (95% CI: 0.26–1.43) in lean body mass compared with controls [4, 48]. Yet, most studies have excluded older, malnourished, or patients with low physiological fitness – the individuals most vulnerable to muscle wasting but also those with the greatest potential for relative improvements in physical function and clinical outcomes [33]. Current evidence suggests that patients should aim for a protein intake of approximately 1.5 g/kg/day, or 15–20% of total caloric intake, to mitigate treatment-related SMM loss [49]. However, high-quality trials are required to confirm feasibility and efficacy of such multimodal interventions [33, 45]. Strengths and limitations To our knowledge, this meta-analysis represents the largest and most comprehensive synthesis to date of systemic treatment-related SMM loss across cancer types. By excluding studies involving surgery or radiotherapy, we provide a clearer picture of chemotherapy and immunotherapy-related changes in SMM. Most studies (76%) were eligible for quantitative synthesis, which enhanced statistical power and generalizability. No evidence of publication bias was detected. Nonetheless, substantial heterogeneity persisted, likely reflecting variation in cancer types and treatment regimens. Accordingly, the pooled estimate should be interpreted as a summary of heterogeneous findings rather than a single generalizable effect. We recognize that observed changes in SMM are most likely influenced by factors such as disease progression, treatment-related toxicity, and nutritional status, and should not be interpreted as independent of these processes. In particular, the inclusion criteria of two available SMM assessments, which preclude a loss to follow-up evaluation, introduce selection and survivorship bias in retrospective studies. Accordingly, the risk-of-bias assessment focuses on the validity of estimating within-patient change while acknowledging the inherent limitations of observational study designs, including residual confounding and selection bias. In addition, variation in outcome definitions (e.g. SMI, SMA, and absolute SMM) represents a further source of heterogeneity, as these measures are not directly equivalent despite standardization. Because analyses relied on study-level summary estimates, planned stratification by treatment agent, age, disease stage or number of cycles were not possible. Furthermore, 38% of studies did not report the interval between SMM assessments, limiting precise interpretation of the rate and timing of SMM loss. Finally, varying definitions of low SMM, primarily lacking functional measures [1], precluded evaluation of sarcopenia as this is defined by combined assessments of muscle strength and SMM [50]. Conclusion In summary, declines in SMM are frequently observed during systemic cancer treatment. However, substantial heterogeneity across studies indicates that the magnitude of change varies across clinical contexts, and findings should not be interpreted as a uniform effect. The extent varies across cancer types and treatment modalities, reflecting the interplay of biological, treatment-related, and behavioral factors. Integrating automated body composition analysis into routine imaging could enable early detection of clinically meaningful SMM wasting, inform treatment planning, and facilitate timely preventative interventions. Supplementary Material