Microstructural MRI combined with metabolic MRI to evaluate the progression-free survival of patients with glioblastoma at the early follow-up after tumor treating fields: a pilot study Authors: Jue Lu, XinLi Zhang, Xiaotong Guo, Qian Qin, Jiaqi Chen, Peng Sun, Jing Wang Published: 2026 ABSTRACT BackgroundNon-invasive assessment of treatment response in Tumor Treating Fields (TTFields) and exploration of progression-free survival (PFS) by imaging techniques remains a major challenge.ObjectivesThis exploratory pilot study investigated the feasibility of using time-dependent diffusion MRI (td-dMRI) combined with chemical exchange saturation transfer (CEST) MRI to assess the TTFields treatment efficacy and explore associations with PFS in postoperative glioblastoma (GBM) patients.MethodsThis study employed td-dMRI and CEST techniques to scan patients before and 1–6 months after receiving TTFields or temozolomide treatment between May 2023 and November 2024. Microstructural metrics, including Diameter, intracellular volume fraction (Vin), Cellularity index and extracellular diffusivity (Dex), were calculated using the IMPULSED method. Multiple metabolic parameters were derived through magnetization transfer ratio asymmetry analysis and pH-weighted analysis. Longitudinal changes in these parameters were analyzed using linear mixed-effects models (LMM), with sensitivity analysis conducted via the Mann-Whitney U test. Spearman correlation analysis assessed relationships between parameters and PFS.ResultsTen patients were included (age range 29–74 years; 8 men). The TTFields group showed a trend toward greater early reduction in cellularity index at first follow‑up. LMM analysis showed different dynamic evolution trends in Diameter (F = 5.042, P = 0.035) and pH-weighted values (F = 5.291, P = 0.055) between the progression and non-progression groups. At the third month post-treatment, the Diameter in the progression group increased from baseline, while it decreased in the non-progression group (estimated value = 3.17, 95% CI: 0.29–6.06, P = 0.036). Sensitivity analyses supported these descriptive between‑group differences. Correlation analysis showed that at the third month post-treatment, the change in diameter was associated with the pH-weighted change (r = 0.697, P = 0.025). During the early treatment phase, both the change in diameter and the pH-weighted change relative to baseline showed negative associations with patient PFS (r = -0.636, P = 0.048; r = -0.697, P = 0.025).ConclusionsThis exploratory pilot study suggested that combining td-dMRI with CEST MRI may help describe microstructural and metabolic changes that could be associated with TTFields efficacy during early treatment. Changes in Diameter and pH-weighted values, both in trend and magnitude, showed correlations with patient PFS. These observations support further exploratory investigations in larger cohorts to clarify the role of multiparametric MRI in evaluating early TTFields‑associated changes and prognosis.Critical relevance statementEarly assessment of the efficacy of TTFields for GBM is key to individualized therapy, but challenges remain. This exploratory pilot study combined microstructural and metabolic MRI to describe early longitudinal changes linked to TTFields response. The preliminary findings provided initial insights that may inform future larger-scale studies aimed at better understanding treatment response in GBM patients.Clinical trial numberNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12880-026-02482-1. Introduction Gliomas are the most common primary central nervous system malignancies, with glioblastoma (GBM) being the most malignant and having the worst prognosis. Despite aggressive and multimodal treatments, patients have a median overall survival (OS) of approximately 15–18 months, with a 5-year survival rate of less than 10% [1–3]. Once GBM recurs, the median OS is only 24–44 weeks [4]. Thus, current treatments have shown little improvement in progression-free survival (PFS) and OS, necessitating the investigation of novel therapies. Tumor Treating Fields (TTFields) have demonstrated a significant increase in PFS and OS (6.7 vs. 4.0 months and 20.9 vs. 16.0 months, respectively, for TTFields with temozolomide (TMZ) versus TMZ alone) in a randomized clinical trial, and were approved by the FDA for newly diagnosed GBM in 2015 [5–7]. TTFields deliver low-intensity, medium-frequency electric fields via a transducer array on the scalp, targeting GBM’s proliferative nature to induce mitotic arrest and apoptosis in rapidly dividing cells [8–11]. Additionally, TTFields may enhance antitumor immune responses and influence the tumor microenvironment (TME) through metabolic changes [12–14]. However, precise quantification of the treatment response is challenging due to TTFields’ complex effects. The Response Assessment in Neuro-Oncology (RANO) criteria, commonly used in GBM trials, are insufficient for explaining TTFields’ mechanisms or predicting early responses. The specificity of T1 enhancement is inadequate for distinguishing histopathological regions, and enhancing lesions can indicate both tumor remnants and treatment effects [15–18]. Nonenhancing lesions with high T2 signal may represent invasive tumors or edema [19, 20]. Therefore, reliable imaging biomarkers for TTFields treatment response are needed for better disease management. Recent advances in time-dependent diffusion MRI (td-dMRI) have demonstrated unique advantages in depicting cellular microstructures and distinguishing pathological tissues [21–23]. Traditional td-dMRI is defined as the variation of apparent diffusion coefficient (ADC) with diffusion time, requiring observation of changes in t-space while maintaining constant q-space. Subsequently, the Imaging Microstructural Parameters Using Limited Spectrally Edited Diffusion (IMPULSED) method is developed, followed the MR cytometry approach [24, 25]. This method combines both pulsed gradient spin echo (PGSE) and oscillating gradient spin echo (OGSE) sequences to acquire data with multiple b-values and multiple effective diffusion times, thereby integrating information from t-space and q-space. By fitting this richer dataset to a biophysical model, the method can estimate specific microstructural parameters, such as Diameter, intracellular volume fraction (Vin), Cellularity index, and extracellular diffusivity (Dex) [26–28]. Recently, clinical applications of td-dMRI based on IMPULSED method have yielded promising results in the grading, molecular typing, and differential diagnosis of gliomas, though its performance in evaluating postoperative glioma treatment is not fully confirmed [29–31]. Chemical exchange saturation transfer (CEST) MRI has advanced significantly in monitoring glioma treatment response by noninvasively tracking metabolic alterations in target regions [32, 33]. By leveraging pH-dependent proton exchange rates, CEST enables precise quantification of GBM’s acidic microenvironment, capturing Warburg Effect-induced lactate accumulation [34]. Amide proton transfer (APT), a key subclass of CEST, reflects elevated protein/peptide concentrations in brain tumors by saturating amide protons, aiding in tumor grading and identifying pseudoprogression. The integration of APT-derived protein metabolism profiles, pH-sensitive microenvironment analysis, and td-dMRI’s microstructural parameters provides complementary insights for comprehensive tumor characterization. Given the complex therapeutic mechanism of TTFields and the intricate tumor microenvironment of glioblastoma, single-modality imaging may not fully capture treatment response. We hypothesize that combining parameters reflecting cellular structure with those reflecting microenvironmental metabolic status may provide complementary information to potentially support a more comprehensive assessment of early changes induced by TTFields and to explore potential associations between early multiparametric MRI changes and patient progression-free survival (PFS) in this pilot exploratory cohort. Materials and methods Patients Ten patients were prospectively recruited between May 2023 and September 2024. Inclusion criteria were: (1) age > 18 years; (2) diagnosis of supratentorial GBM according to the 2016 WHO classification criteria for tumors of the central nervous system; (3) completion of TMZ-based standard concurrent radiochemotherapy; (4) Karnofsky performance score > 70. Exclusion criteria included: (1) early progression of GBM after completion of concurrent radiochemotherapy (imaging examination required to rule out pseudoprogression); (2) receipt of any other cytotoxic or biological anti-tumor therapy. Adequate hematologic, hepatic, and renal function were required for temozolomide chemotherapy.At the end of radiochemotherapy, patients were randomized at a 2:1 ratio to receive standard maintenance TMZ chemotherapy (150–200 mg/m²/day for 5 days every 28 days for 6 cycles) with or without TTFields. Patients were randomly assigned to groups without stratification. TTFields were continued until patients experienced adverse clinical reactions or MRI showed early progression. This is a prospective exploratory pilot study. The sample size (n = 10) was primarily determined based on feasibility, considering the novelty of the combined MRI protocol and the challenges in recruiting patients for the relatively new and expensive TTFields therapy. The primary objective of this study was to assess the technical feasibility of acquiring and analyzing combined td-dMRI and CEST data in this patient population and to explore preliminary trends and associations, thereby laying the groundwork for future hypothesis-driven studies of larger scale. Follow-up All patients received conventional MRI, td-dMRI, and CEST baseline scans within 2 weeks before treatment, followed by the first scan within a month after treatment and monthly regular follow-ups. Routine MRI included T1-weighted images before and after Gd-enhancement, T2-weighted images, and fluid-attenuated inversion recovery (FLAIR) images. Patients were assessed clinically by neurological examination and the Karnofsky Performance Score at baseline and every 4 weeks during treatment. Two experienced radiologists obtained all images, and tumor treatment response or progression was determined based on MRI and clinical assessments, with a third radiologist acting as an arbiter in case of disagreement. According to RANO criteria [16], PFS was defined as the time interval between initiation of treatment and tumor progression with clinical deterioration and MRI findings consistent with progressive disease. This study adhered to the declaration of Helsinki and received approval from the Institutional Review Board of our institution (Wuhan Union Hospital, NO. UHCT22438). Written informed consent was obtained from all patients in this study. Data acquisition All patients underwent scanning using a 3.0 Tesla magnetic resonance scanner (Philips Ingenia CX; Philips, Best, Netherlands) with a maximum gradient of 80 mT/m, slew rate of 200 mT/m/s, and a 32-channel head coil. td-dMRI was acquired using a combination of OGSE and PGSE sequences. CEST sequence based on 2D at 64 unevenly sampled saturation offsets was applied to the maximum cross-sectional areas of the tumors. Table 1 summarized the key parameters of td-dMRI, CEST, and conventional MRI. Details of td-dMRI and CEST sequence are given in the Appendix S1. Table 1Scan sequence parametersParametertd-dMRICEST MRIContrast-Enhanced T1wT2-FLAIRKey Sequence Parametersoscillating frequencies:0 Hz, 25 Hz, 50 HzSaturation Power: 0.9 µT--Effective diffusion time:15.9ms(0 Hz), 12ms(25 Hz), 6ms(50 Hz)Saturation Duration: 3000 ms--b-values/ Saturation Offsets0, 250, 500, 1000, 1500 s/mm² (0 Hz);0, 250, 500, 800, 1200s/mm² (25 Hz);0, 100, 200, 300 s/mm² (50 Hz)61 offsets (0 to ± 30 ppm), S₀ at 100 ppm--Repetition Time3000 ms5000 ms7.9 ms4800 msEcho Time106 ms14 ms3.5 ms313 msField of View220 × 220 mm²200 × 200 mm²250 × 199 × 170 mm³250 × 250 × 150 mm³ Table 1Scan sequence parametersParametertd-dMRICEST MRIContrast-Enhanced T1wT2-FLAIRKey Sequence Parametersoscillating frequencies:0 Hz, 25 Hz, 50 HzSaturation Power: 0.9 µT--Effective diffusion time:15.9ms(0 Hz), 12ms(25 Hz), 6ms(50 Hz)Saturation Duration: 3000 ms--b-values/ Saturation Offsets0, 250, 500, 1000, 1500 s/mm² (0 Hz);0, 250, 500, 800, 1200s/mm² (25 Hz);0, 100, 200, 300 s/mm² (50 Hz)61 offsets (0 to ± 30 ppm), S₀ at 100 ppm--Repetition Time3000 ms5000 ms7.9 ms4800 msEcho Time106 ms14 ms3.5 ms313 msField of View220 × 220 mm²200 × 200 mm²250 × 199 × 170 mm³250 × 250 × 150 mm³ Scan sequence parameters 0, 250, 500, 1000, 1500 s/mm² (0 Hz); 0, 250, 500, 800, 1200s/mm² (25 Hz); 0, 100, 200, 300 s/mm² (50 Hz) Image analysis Tumor masks for each case were manually delineated by two radiologists (Radiologist 1 with 3 years of experience and Radiologist 2 with 5 years of experience) on td-dMRI and CEST images using ITK-SNAP (version 4.2; www.itksnap.org). The solid tumor was defined as either the contrast-enhanced region on T1-weighted images or the hyperintense region on T2-FLAIR images when contrast enhancement was not detected. Tumor masks excluded areas of internal necrosis. When the initial mask delineations by two radiologists were inconsistent, they would undergo a joint review under the supervision of an experienced senior physician to reach consensus through discussion. In this study, microstructural parameters were derived using the IMPULSED method, which follows the MR cytometry framework as defined in the Introduction. Specifically, data acquisition simultaneously recorded OGSE and PGSE sequences to sample diffusion signals at multiple effective diffusion time points. These signals were subsequently fitted to a non-exchange two-compartment model to estimate microstructural parameters [35–37]. Quantitative analysis was performed using a custom MATLAB script (Version 2024b; MathWorks). Based on a previous study [38], custom MATLAB code (version 2024b; MathWorks) was used for quantitative image analysis of CEST. Details of the theory of the IMPULSED method and the principle of CEST are summarized in Appendix S1. The microstructural parameters, pH-weighted values and APTw values were evaluated by two radiology trainees who were blinded to the findings of the other radiologist. Statistical analysis All statistical analyses were performed using IBM SPSS Statistics (v26.0, IBM Corp., USA) and GraphPad Prism v9.0.0.121 (GraphPad Software, San Diego, CA, USA). Variables following a normal distribution are presented as mean ± standard deviation (mean ± SD). Variables not following a normal distribution are described using median combined with interquartile range (median, IQR). Survival analyses were conducted using Cox proportional hazard models and Kaplan-Meier analyses, with median survival and Log-Rank p-values reported to assess differences in efficacy between the two treatment groups. Descriptive statistics are presented as means. The intraclass correlation coefficient (ICC) was used to assess inter-reader agreement. A linear mixed-effects model (LMM) was employed to analyze the longitudinal trends in microstructural and metabolic parameters between the progression and non-progression groups across different treatment phases. Thus, we investigated the interaction effect between time as a within-subject factor (different treatment phases) and treatment outcome as a between-subject factor (tumor progression and non-progression groups) on the dependent variables (microstructural and metabolic parameters). Sensitivity analysis was performed using Mann-Whitney U test as a supplementary explanation for the LMM analysis. Specifically, it was used to compare differences between baseline and the first month post-treatment, baseline and the third month post-treatment, baseline and the sixth months post-treatment for participants who were diagnosed with tumor progression and for those who were diagnosed with non-progression. Spearman correlation analysis was employed to assess associations among parameters from different scanning sequences. P value < 0.05 was considered statistically significant. Result Patient demographic In this pilot prospective exploratory study, Ten adult patients with histologically confirmed glioblastoma after surgical resection were prospectively recruited over 24 months. Demographic and clinical characteristics were summarized in Table 2. A large male predominance was observed (80%), with a mean age of 51.1 years (range: 29–74 years). All patients underwent maximal tumor resection followed by a standard course of radiochemotherapy. Patients were randomly assigned to two groups: the TTFields group (receiving TTFields and TMZ chemotherapy, n = 7) and the TMZ group (receiving TMZ chemotherapy only, n = 3). In the TTFields subcohort, the median PFS was 13.6 months (range: 4.4–22.3 months) at the data cutoff on October 11, 2024. In the TMZ subcohort, the median PFS was 15.6 months (range: 11.8–19.5 months). Notably, patient 7 withdrew from the treatment group after rapid disease progression at the fourth month of follow-up and did not participate in subsequent imaging. Table 2Patient characteristics, groups and outcomesIDAgeGenderDiagnosisIDH1/2 MutationTreatmentClinical Diagnosis165MGBM-TTFields + TMZProgress249MGBM+TTFields + TMZProgress340FGBM-TTFields + TMZNon- Progress444MGBM-TTFields + TMZNon- Progress542MGBM-TTFields + TMZNon- Progress629MGBM-TTFields + TMZNon- Progress755MGBM+TTFields + TMZProgress873MGBM-TMZProgress940FGBM+TMZNon- Progress1074MGBM-TMZNon- ProgressGBM, Glioblastoma; TTFields, Tumor treating fields; TMZ, temozolomide Table 2Patient characteristics, groups and outcomesIDAgeGenderDiagnosisIDH1/2 MutationTreatmentClinical Diagnosis165MGBM-TTFields + TMZProgress249MGBM+TTFields + TMZProgress340FGBM-TTFields + TMZNon- Progress444MGBM-TTFields + TMZNon- Progress542MGBM-TTFields + TMZNon- Progress629MGBM-TTFields + TMZNon- Progress755MGBM+TTFields + TMZProgress873MGBM-TMZProgress940FGBM+TMZNon- Progress1074MGBM-TMZNon- ProgressGBM, Glioblastoma; TTFields, Tumor treating fields; TMZ, temozolomide Patient characteristics, groups and outcomes GBM, Glioblastoma; TTFields, Tumor treating fields; TMZ, temozolomide Based on imaging and clinical assessment, four individuals were diagnosed with tumor progression (progression group), three of whom were in the TTFields group. Six individuals were diagnosed with tumor response to treatment (non-progression group), four of whom were in the TTFields group. Comparison of microstructural characterization At each treatment time point, no significant differences were observed in microstructural parameters between the TTFields group and the TMZ group (all P > 0.05). The microstructural parameter values for patients in the two treatment groups are presented in Table S1. There were no significant changes in Diameter and Dex in either group before and after treatment (Fig. 1A-B). During the follow-up period, Vin in TTFields group and cellularity index in both treatment groups initially decreased, with a trend toward a greater decrease in the TTFields group (Fig. 1C-D). Subsequently, cellularity index in both groups began to increase. The relative apparent diffusion coefficient (ΔADC) in the TTFields group showed a slight decrease after the first treatment, followed by a gradual increase, with ΔADC2 exhibiting a more pronounced trend (Fig. 1E-F). In contrast, no significant trend was observed for ΔADC in the TMZ group. ADC remained relatively unchanged in both treatment groups at different follow-up periods (Fig. 1G-I). Fig. 1Box plots show comparisons of td-dMRI microstructural parameters for different treatment groups at different follow-up times, including (A) Diameter, (B) extracellular diffusivity (Dex), (C) intracellular volume fraction (Vin), (D) cellularity index, (E-F) Relative ADC (∆ADC1 and ∆ADC2), (G)apparent diffusion coefficient at 0 Hz (D0Hz), (H) apparent diffusion coefficient at 25 Hz (D25Hz) and (I) apparent diffusion coefficient at 50 Hz (D50Hz) Box plots show comparisons of td-dMRI microstructural parameters for different treatment groups at different follow-up times, including (A) Diameter, (B) extracellular diffusivity (Dex), (C) intracellular volume fraction (Vin), (D) cellularity index, (E-F) Relative ADC (∆ADC1 and ∆ADC2), (G)apparent diffusion coefficient at 0 Hz (D0Hz), (H) apparent diffusion coefficient at 25 Hz (D25Hz) and (I) apparent diffusion coefficient at 50 Hz (D50Hz) Figure 2 showed td-dMRI-based microstructural maps of two representative patients from the tumor progression and non-progression groups. The complete parameter changes for the patient in the aforementioned progression group are shown in Figures S2-S3. The initial tumor mask delineations by two radiologists demonstrated good consistency, with an intraclass correlation coefficient (ICC) of 0.93 (95% CI: 0.89–0.96). At all follow-up time points, the Diameter in the progression group was slightly higher than that in the non-progression group, with the difference increasing between 3 and 6 months post-treatment (Figure S4B). Similarly, Dex in the progression group increased slightly over time, whereas the non-progression group exhibited a more pronounced decrease (Figure S4B). Both the progression and non-progression groups exhibited transient decreases in Vin, cellularity index, and ΔADC2 post-treatment (Figure S4C-D, F). Following treatment, cellularity index gradually increased in the progression group while decreasing in the non-progression group. ΔADC1 and ADC values at different oscillation frequencies were slightly higher in the progression group compared to the non-progression group (Figure S4E, G-I). Detailed microstructural parameter values for patients in both groups are provided in Table S3. Fig. 2Typical comparison images. The figures show td-dMRI-based microstructural parameter images and CEST-based metabolic parameter images of two patients in the TTFields group at different time intervals. Subfigure (A) shows the image associated with the patient (patient1 in Table 2) whose final outcome was progression with a PFS of 13.7 months. Subfigure (B) shows the image associated with a patient (patient5 in Table 2) with a final outcome of no progression and a PFS of 17.7 months. The solid arrow indicates the direction of tumor progression Typical comparison images. The figures show td-dMRI-based microstructural parameter images and CEST-based metabolic parameter images of two patients in the TTFields group at different time intervals. Subfigure (A) shows the image associated with the patient (patient1 in Table 2) whose final outcome was progression with a PFS of 13.7 months. Subfigure (B) shows the image associated with a patient (patient5 in Table 2) with a final outcome of no progression and a PFS of 17.7 months. The solid arrow indicates the direction of tumor progression The linear mixed-effects model revealed a significant group-by-time interaction for Diameter relative to baseline between the two groups (F = 5.042, P = 0.035), indicating a potential difference in longitudinal trends between the progression and non-progression group. Parameter estimation indicated that the difference in Diameter between the progression group and the non-progression group was significant at the third-month post-treatment compared to baseline (estimated value = 3.17, 95% CI: 0.29–6.06, P = 0.036). In terms of effect size, the Cohen’s d for the early interaction was approximately 0.85. Figure 3 showed the longitudinal changes in Diameter for all patients. In contrast, under the current sample size, no statistically significant differences in longitudinal trends between the progressive and non-progressive groups were observed for changes relative to baseline in other microstructural parameters, including Vin, cellularity index, Dex, and ADC values at different oscillation frequencies. Fig. 3Longitudinal changes in Diameter for all patients. (A) The line chart illustrated the dynamic evolution of Diameter in the progression group and non-progression group. (B) The line chart showed the difference the change in Diameter relative to baseline (∆Diameter) between the progression group and non-progression group. Time interval 1 referred to the time between the first month post-treatment and baseline. Time interval 2 referred to the time between the third month after treatment and baseline. Time interval 3 referred to the time between the sixth month after treatment and baseline Longitudinal changes in Diameter for all patients. (A) The line chart illustrated the dynamic evolution of Diameter in the progression group and non-progression group. (B) The line chart showed the difference the change in Diameter relative to baseline (∆Diameter) between the progression group and non-progression group. Time interval 1 referred to the time between the first month post-treatment and baseline. Time interval 2 referred to the time between the third month after treatment and baseline. Time interval 3 referred to the time between the sixth month after treatment and baseline Comparison of metabolic information No significant changes in pH-weighted and APTw values were observed across time in both treatment groups (Figure S1). Complete metabolic parameter values for both treatment groups were presented in Table S2. Figure 2 showed CEST-based metabolic maps of two representative patients from the tumor progression and non-progression groups at different follow-up times. Interestingly, the high-value regions on pH-weighted and APTw mappings showed a potential correlation with the actual sites of tumor progression, providing a research hypothesis for future localization studies. At the third and the sixth month after treatment, both pH-weighted and APTw values remained higher in the progressive group than in the non-progressive group. The specific metabolic parameter values for patients in the progression group and non-progression group were presented in Table S4. The linear mixed-effects model found that although pH-weighted values and APTw values between the progression group and non-progression group showed a trend toward difference at various treatment time points, these differences did not reach statistical significance in this sample (F = 5.291, P = 0.055 and F = 5.779, P = 0.065). Sensitivity analysis Summarized results of the Mann-Whitney U test that were used to compare differences in three parameters (Diameter, pH-weighted and APTw values) between baseline and the third month post-treatment and between baseline and the sixth month post-treatment for the progression group and the non-progression group, analyzed separately, were shown in Table 3. Results showed that Diameter in the progression group increased from baseline at the third month after treatment (1.8158, 0.8396­1.9739), whereas it decreased in the non-progression group (-2.2172, -3.9497 - -1.2881). The difference between groups was statistically significant in this exploratory analysis (P = 0.033). At the third month after treatment, the pH-weighted values in the progression group increased from baseline (0.0381, 0.0245­0.1019), while those in the non-progression group showed little change from baseline (0.0001, -0.0084–0.0027) (P = 0.011). There was no significant difference in the change in APTw values relative to baseline between the two groups at either the third or sixth month post-treatment (P = 0.088, P = 0.302). Table 3Summary of results of sensitivity analysisParameterTime pointGroupP valueProgression groupNon-progression grouDiameter11.8158(0.8396, 1.9739)-2.2172(-3.9497, -1.2881)0.0332-0.0560(-0.2281, 0.1019)-0.0665(-0.2210, -0.0387)0.606pH-weighted10.0381(0.0245, 0.0399)0.0001(-0.0084, 0.0027)0.01120.0237(0.0122, 0.0313)-0.0039(-0.0082, 0.0064)0.071APTw10.0293(0.0188, 0.0330)0.0032(-0.0010, 0.0165)0.08820.0091(0.0038, 0.0242)0.0032(-0.0024, 0.0205)0.302Note.—These results were used to compare differences in microstructural and metabolic parameters—between baseline and the third month post-treatment (time point 1), and between baseline and the sixth month post-treatment (time point 2)—for the tumor progression group and non-progression group, respectively Table 3Summary of results of sensitivity analysisParameterTime pointGroupP valueProgression groupNon-progression grouDiameter11.8158(0.8396, 1.9739)-2.2172(-3.9497, -1.2881)0.0332-0.0560(-0.2281, 0.1019)-0.0665(-0.2210, -0.0387)0.606pH-weighted10.0381(0.0245, 0.0399)0.0001(-0.0084, 0.0027)0.01120.0237(0.0122, 0.0313)-0.0039(-0.0082, 0.0064)0.071APTw10.0293(0.0188, 0.0330)0.0032(-0.0010, 0.0165)0.08820.0091(0.0038, 0.0242)0.0032(-0.0024, 0.0205)0.302Note.—These results were used to compare differences in microstructural and metabolic parameters—between baseline and the third month post-treatment (time point 1), and between baseline and the sixth month post-treatment (time point 2)—for the tumor progression group and non-progression group, respectively Summary of results of sensitivity analysis Note.—These results were used to compare differences in microstructural and metabolic parameters—between baseline and the third month post-treatment (time point 1), and between baseline and the sixth month post-treatment (time point 2)—for the tumor progression group and non-progression group, respectively PFS correlated with td-dMRI-derived parameters and metabolic information based on CEST At the third month post-treatment, the change in Diameter relative to baseline showed a positive correlation with the change in pH-weighted value relative to baseline in this pilot cohort (r = 0.697, P = 0.025) (Fig. 4A). Fig. 4Results of Spearman correlation analysis. (A) The scatter plot showed the correlation between the change in Diameter relative to baseline (∆Diameter) and the change in pH-weighted value relative to baseline (∆pH-weighted) at the third month post-treatment. (B-C) The scatter plots displayed the correlation between progression-free survival and ∆Diameter, as well as ∆pH-weighted, at the third month post-treatment Results of Spearman correlation analysis. (A) The scatter plot showed the correlation between the change in Diameter relative to baseline (∆Diameter) and the change in pH-weighted value relative to baseline (∆pH-weighted) at the third month post-treatment. (B-C) The scatter plots displayed the correlation between progression-free survival and ∆Diameter, as well as ∆pH-weighted, at the third month post-treatment A larger change in Diameter relative to baseline was associated with a shorter survival period at the third month after treatment (r = -0.636, P = 0.048) (Fig. 4B). At the same time, the change in pH-weighted value relative to baseline was negatively correlated with PFS (r = -0.697, P = 0.025) (Fig. 4C). Discussion TTFields is a promising therapeutic approach with various anti-tumor mechanisms and survival benefits. However, research evaluating its treatment response remains limited. Therefore, we designed a prospective exploratory study to explore the potential research utility of td-dMRI in conjunction with CEST for examining the treatment mechanisms of TTFields and exploring early longitudinal changes associated with PFS in postoperative GBM patients. In this pilot cohort, our results suggest that patients receiving TTFields in combination with TMZ showed a more pronounced reduction trend in Vin and cellularity index during the early phase of treatment than those receiving TMZ alone. The change in Diameter relative to baseline exhibited different longitudinal trends between the progression group and the non-progression group. At the third month post-treatment, the progression group showed a trend toward increased Diameter relative to baseline, whereas the non-progression group showed a decrease. Concurrently, the change in pH-weighted value relative to baseline also differed between groups in this pilot cohort. In the early post-treatment period, changes in both parameters were correlated, with both the change in Diameter relative to baseline and the change in pH-weighted value relative to baseline showing a negative correlation with PFS. These exploratory preliminary findings suggested the need for further investigation of these non-invasive imaging metrics to explore their potential for evaluating the TTFields therapeutic efficacy in postoperative GBM patients. TTFields, first included in the National Comprehensive Cancer Network (NCCN) guidelines in 2018 for treating recurrent glioblastoma patients, can disrupt tumor cell mitosis leading to increased apoptosis and autophagy [39, 40]. It can also disrupt tumor cell motility and migration, activate the immune system, trigger downstream anti-tumor immune responses, and downregulate the expression of key genes related to DNA damage repair [41–43]. Therefore, TTFields can directly alter tumor cells and modify the tumor microenvironment in which they reside. Despite progress in combining TTFields with radiotherapy and immunotherapy, using imaging techniques to interpret therapeutic mechanisms and assess efficacy remains challenging [44, 45]. Most previous studies focused on animal models, lacking human imaging evidence. Thus, this study employed td-dMRI based on IMPULSED method to explore potential microstructural changes associated with TTFields treatment in humans at a cellular level. Early in the treatment period, Vin and cellularity index were observationally reduced in the T1-enhanced region, consistent with post-treatment changes resulting from TTFields’ primary anticancer mechanism: inhibition of cell proliferation through disruption of cell mitosis. However, metabolic indices did not show significant changes after treatment, possibly reflecting that cellular changes may precede changes in TME. The effect of TTFields on the TME requires further investigation to fully understand the mechanism of action under immune activation. In general, cell size is a key distinction between cancer cells and normal brain cells. Despite its heterogeneity, glioma cells have a significantly larger diameter, averaging 15–20 μm, while most brain cells are glial cells, which typically measure around 5 μm [27, 46]. Our study found that cells in the tumor progression group maintained a diameter of approximately 15 μm, consistent with glioma cell size, whereas cells in the non-progression group were generally smaller and showed a trend toward reduction over time. Our preliminary data suggested a potential difference in the dynamic evolution of Diameter changes relative to baseline between the two groups, with the progression group exhibiting increased Diameter compared to baseline. This aligns with the pathological alteration of enlarged local cell diameter caused by tumor cell infiltration. During follow-up, the tumor progression group exhibited mild increases in Vin and cellularity index. We hypothesize this stems partly from tumor cell proliferation and infiltration, and partly from the proliferation of various immune cells within the inflammatory tumor microenvironment. These immune cells, particularly T cells and macrophages, play a pivotal role in promoting pro-tumor inflammation, which accelerates tumor progression and metastatic spread [28, 47, 48]. Dex, as a measure of water diffusivity in the extracellular space, may be influenced by factors such as cell size and cell density within the tumor microenvironment. Interestingly, we observed descriptive trends higher Dex values in the tumor progression group compared to the non-progression group. Given the current limited discussion on Dex, we hypothesize this phenomenon may relate to the lower overall structural density of tumor cells in the progression group, while their cell diameters remain within the typical range for tumor cells. Therefore, a comprehensive understanding of tumor microstructure is crucial for accurate therapeutic efficacy assessment. Additionally, we observed a trend that the ∆ADC values were higher in the progression group than non-progression group, with ∆ADC2 values exhibiting an upward trend. ∆ADC1 and ∆ADC2 reflected the degree of cellular dispersion limitation at low and high oscillation frequencies, respectively. Elevated values indicate increased small cell structures within the region. Zhu et al. found that the ∆ADC (the ratio of ADC values at high oscillation frequencies to those at 0 Hz) in the true progression area of the tumor was significantly higher than in the pseudo-progression area via histogram analysis [30], which is similar to our observations. By dynamically monitoring changes in the complex tumor microenvironment, td-dMRI may contribute to a deeper understanding of the pathophysiological mechanisms in the postoperative lesion area of glioblastoma in future research. Metabolic imaging may provide complementary information for early response assessment in patients with GBM after radiotherapy. Several studies have shown that changes in CEST parameters significantly precede morphological changes, such as tumor volume reduction, after radiotherapy [49–51]. APT imaging and pH-weighted MRI indirectly reflect cell proliferation status by detecting acid-base changes in the tumor microenvironment. When tumor cells proliferate vigorously, glycolysis (Warburg Effect) is enhanced, generating excessive lactic acid accumulation and leading to a decrease in extracellular pH [32, 52]. This appears consistent with our observation: although no statistically significant differences were observed due to the small sample size, both the pH-weighted values and APTw values measured at different treatment time points showed a moderate trend of difference between the progression group and the non-progression group. Sensitivity analysis provided supportive evidence for the exploratory discriminatory potential of pH-weighted change. In general, tumor pseudoprogression occurs mostly within 6 months after concurrent radiotherapy, and distinguishing pseudoprogression from true progression using T1-enhanced images is difficult [19, 53, 54]. In this study, the microstructural and metabolic parameters observed across the progression group and the non-progression group during the early treatment phase (3–6 months) showed descriptive longitudinal evolution trends. This may provide a preliminary reference basis for future studies aimed at early treatment adjustments. Notably, areas with high values in pH-weighted, APTw, and Diameter images appear to correlate with areas of tumor progression in the early stages of treatment (Figs. 2 and 3). The visual representation of these parameters as maps may support clinical interpretation in future applications, potentially aiding quick image comparison and clinical decision-making. From the perspective of tumor microstructure, regions falling within the typical tumor cell diameter range often exhibit stronger cellular proliferation tendencies. Metabolically, proliferative activity of tumor cells directly impacts the acid-base balance of their microenvironment. An acidic microenvironment not only drives tumor invasion and drug resistance but also activates pro-proliferative signaling pathways (e.g., HIF-1α) [55, 56]. Similarly, we observed a positive correlation between diameter changes measured by td-dMRI and pH-weighted values derived from CEST, suggesting a potential association between tumor microstructural features and metabolic activity. Currently, td-dMRI technology has a well-established research foundation in preoperative grading and molecular subtyping prediction for brain gliomas [29, 30, 57]. To our knowledge, this pilot exploratory study is among the first to explore the application of IMPULSED-derived microstructural parameters for treatment efficacy assessment, and to examine the relationship between these microstructural parameters and CEST-based metabolic parameters. The current study has several limitations. First, the small sample size (n = 10) and single-center nature significantly limit the generalizability of these preliminary findings. The primary reason was that TTFields therapy remains in the clinical trial phase in our institution, with treatment costs prohibitive for most families. Additionally, some enrolled patients declined to return to our hospital for follow-up examinations due to geographical constraints, opting instead for local hospitals for follow-up care, resulting in partial data loss for these enrolled patients. However, our preliminary analysis indicated that early changes in Diameter exhibited a substantial effect size (Cohen’s d = 0.85) between the progression and non-progression groups. Based on this effect size for a power analysis (set at α = 0.05, power = 0.8), we estimate that future similar studies would require approximately 18–22 patients per group, totaling about 40–50 patients. Therefore, we recommend conducting larger-scale multicenter studies in the future to further investigate the potential research value of the proposed microenvironment parameters and metabolic imaging metrics. Second, the time between the end of postoperative radiotherapy and the start of treatment varied for each patient in this study. Ideally, the patient enrollment time should be more comprehensively standardized to improve consistency. Third, this study may be subject to confounding effects from adjuvant treatment regimens. Due to variations in postoperative concurrent chemoradiotherapy received by patients, these therapies themselves may influence the tumor microenvironment and metabolic status. Future investigations should be conducted in more homogeneous treatment cohorts. Fourth, this study lacks histopathological or molecular-level validation. The biological significance of current imaging parameters remains largely descriptive and inferential, and their precise correspondence with tumor microstructure and metabolic activity requires confirmation through pathologically controlled studies. In our exploratory study, seven patients treated with the combination of TTFields and TMZ had a prolonged median PFS (approximately 13.6 months), consistent with Stupp et al., but shorter than the 15.6 months for those treated with TMZ alone [5]. This difference may be attributed to variations in preoperative tumor volume, molecular subtypes, and the dose of postoperative concurrent radiotherapy. Additionally, tumor progression in patient 7 in the fourth month may have a greater impact on the median PFS in this small cohort. Furthermore, the study focused only on short-term patient benefit as measured by PFS and did not address long-term survival, as many patients had not died by the time of this study. Therefore, future studies should focus on long-term survival endpoints and include a larger patient cohort. Conclusion In conclusion, this pilot study explored the combination of td-dMRI and CEST MRI to investigate imaging changes associated with TTFields treatment. The tumor progression group and non-progression group exhibited moderately different dynamic evolution trends in Diameter and pH-weighted values, and the magnitude of change in these metrics correlated with patient PFS. Furthermore, areas of high values in pH-weighted, APTw, and Diameter maps may suggest potential directions for future localization studies. These preliminary findings suggested that combining td-dMRI and CEST may reveal associations with early TTFields treatment outcome, warranting validation in larger multicenter cohorts. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1