Introduction

Glioblastoma (GBM) is the most common malignant primary tumor of the central nervous system (CNS). Standard-of-care-treatment includes supramaximal surgical resection followed by concurrent chemoradiotherapy and adjuvant temozolomide1. Despite this multimodal approach, prognosis remains poor, with a median survival < 15 months2.

The 2021 WHO CNS tumor classification (new classification) has updated the diagnostic criteria for GBM. These include molecular diagnosis with genetic parameters: telomerase reverse transcriptase promoter (TERTp) mutation, epidermal growth factor receptor (EGFR) gene amplification, and a combined gain of chromosome 7 and loss of chromosome 10 (7p+/10q-). Consequently, isocitrate dehydrogenase (IDH)-wild type diffuse astrocytic tumors, previously designated as WHO grade 2 or 3 without certain histological features, are now categorized as molecular GBM (mGBM)3.

Sarcopenia is defined as a progressive and generalized loss of skeletal muscle mass and is associated with poor prognosis across various cancers4,5,6,7,8. Sarcopenia can be quantified using computed tomography-based measurement of muscle composition at the third lumbar vertebra level, a method known to precisely reflect overall body composition9. Leitner et al. report that temporal muscle thickness (TMT) is strongly correlated with lumbar skeletal muscle mass, suggesting its potential as a surrogate marker for sarcopenia10,11. TMT can be easily measured using routine brain magnetic resonance imaging (MRI), enabling sarcopenia evaluation in patients with a brain tumor, without the need for additional imaging. Its prognostic significance is reported across various tumor types6,12.

Several research reports that TMT is a significant prognostic factor for survival in patients with GBM4,7,12,13,14,15. However, these findings are based on the 2016 WHO CNS tumor classification and therefore include only histological GBM (hGBM). Recent studies suggest that hGBM and mGBM may differ in epidemiological characteristics and median overall survival (OS)16,17.

To our knowledge, no study on the prognostic value and age-dependent prognostic difference of TMT has been investigated based on the new classification, considering molecular alteration.

Therefore, this study aims to evaluate the prognostic relevance of TMT in predicting survival among patients with GBM classified according to the new classification, incorporating molecular features and clinical characteristics. Moreover, we investigated whether the prognostic value of TMT differs across age groups.

Results

Patient characteristics

Overall, 285 patients (157 males, 55.1%) met the inclusion criteria. Twenty-nine of 285 patients (11.3%) were diagnosed with mGBM, while 256 patients (88.7%) were diagnosed with hGBM. Table 1 presents the corresponding data. The mean age of all patients was 58.8 years (SD = 14), and the average KPS was 84.8. mMGMT mutation was detected in 121 of 285 patients (42.5%). TERTp mutation was present in 83 of 285 promoters (29.1%). PTEN mutation was found in 27 of 285 patients (9.5%). EGFR gene amplification was observed in 95 of 285 patients (33.3%). No patients exhibited the 7p+/10q- genetic profile. Gross total resection (GTR) was performed in 203 of the 285 patients (71.2%). The mean TMT for all patients was 8.59 mm, with males averaging 8.92 mm and females averaging 8.18 mm. The difference in mean TMT between males and females was statistically significant (p < 0.001). Based on the TMT cut-off values (7.46 and 6.94 mm for males and females, respectively), 209 of 285 patients (73.3%) were classified as non-sarcopenic, while 76 patients (26.7%) were classified as sarcopenic. The mean TMT were 9.37 mm (range 7.26–15.1 mm) and 6.32 mm (range 3.28–7.24 mm) in the non-sarcopenic and sarcopenic groups, respectively. The median OS for all patients was 775 days (interquartile range [IQR]: 581–982 days). Patients in the non-sarcopenic group had a median OS of 826 days (IQR: 631–1046 days), whereas those in the sarcopenic group had a median OS of 625 days [IQR: 466–787 days]. The sarcopenic group demonstrated a significantly poorer prognosis compared to the non-sarcopenic group (HR 1.42 [95% CI, 1.06–1.90]; p = 0.02).

Table 1 Clinicopathologic characteristics.
Full size table

Univariable Cox regression analysis

In the univariable Cox regression analysis (Table 2), TMT was significantly associated with OS (HR 1.48 [95% CI, 1.11–1.97]; p = 0.007). Among clinical variables, female sex, age, and KPS were significantly associated with OS (HR, 0.70 [95% CI, 0.54–0.90]; p < 0.001; HR, 1.02 [95% CI, 1.01–1.03]; p = 0.006; HR, 0.98 [95% CI, 0.97–0.99]; p < 0.001, respectively). EOR was not significantly associated with OS (HR, 0.90 [95% CI, 0.68–1.19]; p = 0.458). Among the molecular variables, only mMGMT status showed a significant association with OS (HR, 0.60 [95% CI, 0.46–0.78]; p < 0.001). PTEN status, EGFR gene amplification, and TERTp status were not significantly associated with OS (HR, 0.98 [95% CI, 0.62–1.53]; p = 0.918; HR, 0.82 [95% CI, 0.63–1.07]; p = 0.149; HR 1.11 [95% CI, 0.82–1.49]; p = 0.481, respectively).

Table 2 Univariable and multivariable Cox regression analyses for identifying factors associated with OS before PSM.
Full size table

Multivariable Cox regression analysis

EOR, PTEN mutation, EGFR gene amplification, and TERTp mutation were not significantly associated with OS in the univariable Cox regression analysis. However, as these variables are previously established prognostic factors, they were included in the multivariable Cox regression analysis. The multivariable Cox analysis including TMT demonstrated that TMT, age, KPS, and mMGMT status were significant predictors for OS (HR, 1.42 [95% CI, 1.06–1.90]; p = 0.02; HR, 1.02 [95% CI, 1.01–1.03]; p = 0.001; HR, 0.98 [95% CI, 0.97–0.99]; p = 0.017; HR, 0.47 [95% CI, 0.35–0.63]; p = 0.001, respectively). Sex was not significantly associated with OS (HR, 0.82 [95% CI, 0.63–1.09]; p = 0.172). Similarly, EOR, PTEN mutation, EGFR gene amplification, and TERTp mutation were not significantly associated with OS (HR, 0.85 [95% CI, 0.64–1.14]; p = 0.287; HR, 1.01 [95% CI, 0.64–1.59]; p = 0.974; HR, 0.86 [95% CI, 0.64–1.15]; p = 0.3; HR, 1.02 [95% CI, 0.76–1.38]; p = 0.877, respectively). Kaplan–Meier survival curves show that patients in the sarcopenic group had significantly shorter OS compared to those in the non-sarcopenic group (p = 0.006) (Fig. 1a).

Fig. 1
figure 1

Kaplan–Meier curves for overall survival based on TMT status (sarcopenia vs. non-sarcopenia). (a) Before and (b) after PSM. P-values were calculated using the log-rank test. TMT temporal muscle thickness, PSM propensity score matching.

Full size image

Propensity score matching analysis

PSM was performed using clinical variables considered likely to be associated with prognostic outcomes. In the resulting age- and sex-matched cohort of 152 patients, TMT was significantly association with OS (HR, 1.57 [95% CI, 1.09–2.28]; p = 0.014). Age, KPS, and mMGMT status were also significantly associated with OS (HR, 1.02 [95% CI, 1.01–1.04]; p = 0.001; HR, 0.98 [95% CI, 0.97–0.99]; p = 0.001; HR, 0.40 [95% CI, 0.26–0.61]; p = 0.001, respectively) (Table 3). The Kaplan–Meier survival curves show that patients in the sarcopenic group had significantly shorter OS compared to those in the non-sarcopenic group (p = 0.018) (Fig. 1b).

Table 3 Univariable and multivariable Cox regression analyses for identifying factors associated with OS after PSM.
Full size table

Age-dependent analysis of the prognostic value of Temporal muscle thickness

The inclusion of TMT in the risk difference plot demonstrated an enhanced prognostic value for OS, which became more pronounced when stratified by age (Fig. 2).

Fig. 2
figure 2

Risk difference plot for OS before and after including TMT across age groups. An age-dependent increase in the prognostic value of TMT is observed. TMT temporal muscle thickness, OS overall survival.

Full size image

Subgroup analysis: comparison between the molecular and histological GBM groups

The mean TMT was 8.06 mm (SD = 1.86 mm) in the mGBM group and 8.65 mm (SD = 1.93 mm) in the hGBM group. No statistically significant difference exists in TMT between the two groups (p = 0.722).

Univariable Cox regression analysis showed that TMT was significantly associated with OS in the hGBM group (HR, 1.88 [95% CI, 1.11–3.17]; p = 0.016). In contrast, TMT was not significantly associated with OS in the mGBM group (HR, 1.15 [95% CI, 0.87–1.5]; p = 0.330).

The median OS was 26 months [IQR: 11–41 months] in the mGBM group and 21 months [IQR: 20–27 months] in the hGBM group. No statistically significant difference exists in OS between the two groups (p = 0.163, log-rank test).

Current literature review

We reviewed current literature4,12,14,15,18,19 investigating the relevance of TMT and GBM survival. The findings were synthesized in a summary table (Table 4).

Table 4 Summary of current literature on Temporal muscle thickness as a prognostic marker in glioblastoma.
Full size table

Discussion

Prior to the release of the 2021 WHO classification of CNS tumors, the diagnosis of GBM was primarily based on histological evaluation20. The new classification incorporates molecular information and histological features for an integrated diagnosis3. Afterward, Lee et al. report that mGBM show a significantly longer survival than hGBM17. Ramos-Fresnedo et al. reveal significantly longer progression-free survival in mGBM compared with hGBM21. These findings highlight the prognostic distinction between mGBM and h GBM.

Several current literature reveal an association between TMT and OS in patients with GBM, based on the 2016 WHO classification of CNS tumors4,12,14,15,18,19. These studies have consistently reported that reduced TMT predicts poorer survival in GBM, with sex-specific cut-offs showing prognostic value for OS and progression-free survival (PFS) and support TMT as a simple, non-invasive surrogate marker of frailty and sarcopenia that may aid in treatment stratification. However, because the 2016 WHO classification of CNS tumors did not define mGBM3, these studies excluded mGBM from their analyses. Consequently, evidence regarding the relationship between TMT and OS in hGBM and mGBM remains limited, and no previous study has specifically investigated this association. Therefore, this study aims to evaluate the prognostic significance of TMT in patients with GBM diagnosed based on the updated 2021 WHO classification. Riccardo et al. reveal the cut-off value without considering sex21. However, their study did observe a relationship between TMT and sex. Accordingly, we compared TMT between male and female patients and found a statistic significant difference (p < 0.001). Therefore, sex-specific cut-off values for males and females were established in this study.

Based on the new classification, TMT was identified as a significant prognostic factor for OS in patients with GBM according to multivariable Cox regression analysis. A correlation between TMT and patient survival was observed, with the sarcopenic group demonstrating a significantly shorter median OS (625 days) compared to the non-sarcopenic group (826 days) (HR, 1.42 [95% CI, 1.06–1.90]; p = 0.02). In addition, univariable and multivariable Cox regression analyses reveal a significant association between TMT and OS (HR, 1.48 [95% CI, 1.11–1.97]; p = 0.007; HR, 1.42 [95% CI, 1.06–1.90]; p = 0.02, respectively). Furthermore, through PSM, we confirmed that TMT was significantly associated with OS regardless of sex or age. TMT influenced OS independent of established prognostic factors. The prognostic value of TMT became more pronounced with increasing patient age. This finding suggests that evaluating TMT as a marker of muscle strength may contribute more effectively to patient management strategies than relying on age alone. Accordingly, identifying a low TMT value may assist clinicians in managing patients, particularly those of advanced age. However, no clear survival difference was observed between the mGBM and hGBM groups; although the median OS was 5 months shorter in the hGBM group, the difference was not statistically significant (p = 0.163, log-rank test), consistent with findings from a previous study21. Additionally, no significant association between TMT and OS was found in the mGBM group (p = 0.698). These findings may be attributed to the relatively small sample size of the mGBM group (n = 29), compared with that of the hGBM group (n = 256). Some results were contrary to those of previous reports in the literature. Specifically, PTEN mutation, EGFR gene amplification, and TERTp mutation—established prognostic factors—were not significantly associated with OS (p = 0.918, p = 0.149, and p = 0.481, respectively)22,23. This discrepancy may be explained by differences in the study population compared with that of previous studies. In this study, the rates of positive molecular alterations were relatively low, with 9.5%, 33.3%, and 29.1% of patients testing positive for PTEN mutation, EGFR gene amplification, and TERTp mutation, respectively. However, age (HR, 1.02 [95% CI, 1.01–1.03]; p = 0.001), KPS (HR, 0.98 [95% CI, 0.97–0.99]; p = 0.001), and mMGMT (HR, 0.47 [95% CI, 0.35–0.63]; p = 0.001) were confirmed as significant prognostic factor of OS, consistent with that of previous reports22,24,25. EOR was not significantly associated with OS in univariable analysis; however, it demonstrated a significant association in the multivariable analysis with TMT after PSM (HR, 0.65 [95% CI, 0.43–0.98]; p = 0.038), aligning with findings from a previous study22. From a comprehensive perspective, our findings suggest that even in the molecular era, extracranial imaging features such as TMT may serve as independent and clinically meaningful prognostic indicators, consistent with that of the study by Pasqualetti et al., which reveals the prognostic relevance of the internal carotid artery calcium score in patients with GBM26.

This study has some limitations. First, as the revised classification system was only recently introduced, a substantial number of patients were excluded owing to insufficient molecular information, resulting in a small sample size of the mGBM group. This limitation could be addressed in future studies with larger sample sizes. Second, owing to the retrospective nature of the analysis, it was not possible to correlate TMT with actual patient muscle strength or objective functional or frailty indices (e.g., comorbidity scores). This limitation may be mitigated in future research through segmentation and signal intensity analysis using a deep learning-based system27,28. Third, the study may be subject to survivorship bias, as noted by Pasqualetti et al., since it includes only patients who underwent surgical treatment, which may limit the generalizability of the findings to the broader glioblastoma population29. Fourth, owing to the heterogeneity of additional treatment— such as completion of the Stupp protocol, use of bevacizumab, corticosteroids, or salvage therapies—beyond EOR, these variables may influence both sarcopenia and OS, potentially serving as unmeasured confounders that could affect the observed associations. Fifth, owing to limited sample size and data availability, external validation or internal bootstrapping techniques were not performed, resulting in no validation cohort. This limitation may be addressed in future studies employing larger, independent cohorts. Lastly, instead of using standardized reference points for skeletal muscle mass reflecting sarcopenia in the general population, the groups were divided based on the lower quartile of TMT values within the study cohort.

Conclusions

This study demonstrates that TMT serves as an independent prognostic marker for OS in patients with GBM, with its prognostic significance increasing with patient age. From a broader perspective, the findings suggest that even in the current era of molecular genetics, imaging-derived features from extra-axial or extracranial structures may represent important, independent, and potentially more direct prognostic indicators. These findings suggest that clinicians may incorporate TMT assessment into treatment strategy, particularly for elderly patients exhibiting sarcopenia.

Methods

Study design

This retrospective study was approved by the Institutional Review Board (IRB No. No. 2212-077-1385) of Seoul National University Hospital, with a waiver of written informed consent owing to its retrospective design. All analyses were conducted in accordance with the ethical standards of the Institutional Review Board and Declaration of Helsinki (1975, as revised in 2008).

Study population

Consecutive patients with GBM were identified at a single tertiary hospital between April 2009 and September 2022. Patients were included if they had a newly diagnosed IDH-wild type GBM according to the new classification, were pathologically confirmed based on craniotomy or stereotactic biopsy, had accessible baseline thin slice T1-weighted MRI (T1-WI), and had accessible data on survival status and/or death date. Patients were excluded if they had a prior history of brain surgery or radiation therapy for any medical reason; missing genetic information on IDH mutation, 1p19q co-deletion, TERTp mutation, EGFR gene amplification, or 7p+/10q- (Fig. 3). Age, sex, Karnofsky Performance Scale (KPS), and extent of resection (EOR) were obtained from electronic medical records. The date of death was also collected, and OS was defined as the time from pathological diagnosis to death. Molecular genetic alterations, including IDH mutation, 1p19q co-deletion, TERTp mutation, EGFR gene amplification, 7p+/10q-, O6-methylguanine-DNA methyltransferase (mMGMT) promoter methylation status, and phosphatase and tensin homolog (PTEN) mutation status, were collected. All patients were classified into two groups based on the 2021 WHO CNS classification: hGBM, defined based on characteristic histological presence of microvascular proliferation or necrosis, and mGBM, defined as IDH-wildtype astrocytomas lacking histological features of GBM but harboring specific genetic alterations, including TERTp mutation, EGFR amplification, or 7p+/10q-17. Detailed information on tissue diagnosis and genetic analysis is provided in the Supplementary material.

Fig. 3
figure 3

Flowchart of patient inclusion and exclusion criteria. *Molecular genetic alterations include IDH mutation, 1p/19q co-deletion, TERTp mutation, EGFR gene amplification, and 7p+/10q- status. IDH isocitrate dehydrogenase, TERTp telomerase reverse transcriptase promoter, EGFR epidermal growth factor receptor, 7p+/10q- gain of chromosome 7 and loss of chromosome 10.

Full size image

Temporal muscle thickness assessment

TMT was measured on 1 mm-thickness axial slices of contrast-enhanced T1-WI images from preoperative navigation MRI (Fig. 4). The imaging plane was aligned parallel to the anterior commissure-posterior commissure line. Maximal thickness was measured perpendicular to the long axis of the temporal muscle, using the orbital roof and Sylvian fissure as anatomical landmarks, consistent with a previous study18. TMT was measured separately on the left and right sides, and the mean value was calculated, following the previously established method18. The first quartile value was used as the cut-off point, consistent with a prior study13, and sex-specific cut-off values were applied.

Fig. 4
figure 4

Example of TMT measurements using 3D thin-Sect. (1 mm-thickness) contrast-enhanced T1-weighted MRI. (a–c) Axial images demonstrating MRI alignment for TMT acquisition. (d) A 55-year-old female with sarcopenia (left), a 76-year-old male without sarcopenia (right). TMT temporal muscle thickness, MRI magnetic resonance imaging.

Full size image

Statistical analysis

The first quartile TMT value was determined separately for males and females, and all patients were categorized into two groups: sarcopenic and non-sarcopenic. Univariable Cox regression analyses were conducted to identify predictors of OS, including TMT, clinical characteristics (age, sex, and KPS), and molecular features (IDH mutation, mMGMT, TERTp mutation, EGFR gene amplification, and PTEN mutation status). All variables that were statistically significant in the univariable analysis, and those previously established as significant predictors despite not reaching significance in the current dataset, were included in the consecutive multivariable Cox regression analysis to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for predicting OS. Survival curves were estimated using the Kaplan–Meier method and compared with the log-rank test. Propensity score matching (PSM) was conducted to control potential confounders and balance baseline variables (age and sex) between the two groups. The age-dependent prognostic difference before and after including TMT was assessed. Subgroup analyses were conducted for mGBM and hGBM groups. Univariable Cox regression analyses were conducted to investigate the prognostic significance of TMT on OS within each subgroup. The log-rank test was used to investigate the survival difference between the mGBM and hGBM groups. A p-value < 0.05 was considered statistically significant. All Statistical analyses were conducted using SPSS (version 17.0; SPSS Inc., Chicago, IL, USA), MedCalc (version 17.9; MedCalc Software, Mariakerke, Belgium), SAS (version 9.3; SAS Institute, Cary, NC, USA), and R (version 3.5.2; R Foundation for Statistical Computing, Vienna, Austria; http://www.r-project.org).

Current literature review

We also reviewed the current literature on the prognostic role of TMT in GBM patients. A search of PubMed and Embase was to identify relevant studies published up to past five years. Eligible studies included retrospective and prospective cohorts that evaluated TMT measured on cranial MRI or CT and its association with OS and/or progression-free survival (PFS). Key study characteristics such as sample size, imaging modality, measurement method, cut-off values, outcomes, and results of multivariable analyses were extracted and summarized.