Prediction of Survival in Surgically Treated Glioblastoma Multiforme Utilizing DTI and Contrast-Enhanced MRI

Abstract

Introduction Glioblastoma multiforme (GBM) is a highly malignant brain tumor with poor prognosis, despite maximal safe resection and chemoradiotherapy. Predicting survival outcomes is crucial for optimizing treatment strategies. While conventional magnetic resonance imaging (MRI) reveals tumor characteristics, advanced sequences like diffusion tensor imaging (DTI) and susceptibility-weighted imaging (SWI) may enhance prognostic accuracy.

Materials and Methods A retrospective study reviewed 45 newly diagnosed GBM patients treated with maximal safe resection, adjuvant radiotherapy, and temozolomide between 2016 and 2022. Preoperative MRI data, including conventional sequences, DTI, and SWI, were analyzed. Radiological parameters—tumor volume, edema:tumor volume ratio, necrosis, enhancement, fractional anisotropy (FA), and microhemorrhage—were assessed for survival prediction. Kaplan–Meier survival analysis and Cox regression evaluated their prognostic significance.

Results Among the 45 patients, the median follow-up was 58.3 weeks, with 30 deaths reported. Significant differences were observed in FA, edema:tumor volume ratio, necrosis, enhancement, and microhemorrhage between survivors and nonsurvivors. Higher edema:tumor volume ratio (cutoff ≥ 0.905, area under the curve [AUC] = 0.895) and higher FA (cutoff ≥ 0.655, AUC = 0.8) correlated with mortality and survival, respectively. Multivariate Cox regression identified edema:tumor volume ratio (hazard ratio [HR] = 7.4, p < 0.05) and microhemorrhage > 25% (HR = 45.9, p < 0.05) as independent predictors of mortality.

Conclusion Tumor-related edema and necrosis significantly influence mortality, with edema:tumor volume ratio emerging as a stronger predictor than individual tumor or edema volumes. FA values reflect tumor aggressiveness, correlating with survival. Incorporating advanced imaging parameters like DTI and SWI alongside conventional MRI enhances prognostic precision in GBM management.

 

MissingLinkText

Keywords

DTI - edema:tumor volume ratio - glioblastoma multiforme - microhemorrhage - survival

Introduction

Glioblastoma multiforme (GBM) is the most commonly diagnosed primary malignant tumor in adults that is almost always fatal.[1] The maximal safe resection of the tumor combined with concurrent chemoradiotherapy as well as adjuvant chemotherapy with temozolomide (TMZ) is the standard of care.[2] [3] Despite this, the prognosis of GBM remains unfavorable, with a median survival period of approximately 12 to 14 months.[3] [4] Nonetheless, patient's survival and response to chemotherapy vary.

Early prediction of survival outcomes in patients with GBM is a crucial part of optimizing treatment selection and individualization. This has prompted a quest for prognostic variables in GBM, such as clinical and genetic profiles, as well as magnetic resonance imaging (MRI) biomarkers.[5] [6] Genetic profiles of the tumor can be known only after the surgery. As a noninvasive and conventional preoperative examination for GBM, MRI can reveal fine tumor properties such as size, shape, and heterogeneity, as well as a full, macroscopically displayed picture of the entire tumor.[7]

To predict survival in GBM, radiomics has recently been used extensively to extract and analyze quantitative imaging data such as textural heterogeneity, intensity distributions, spatial correlations, and shape descriptors.[6] [7] [8] [9] [10] [11] [12] However, each of these analyses made use of a particular set of features that were obtained out of multiparametric imaging sequences. The implementation of these advanced sequences is not a common practice across all centers, as it necessitates supplementary expertise and financial resources. Additionally, higher-order feature extraction is time-consuming and involves many postprocessing procedures.[13]

Few conventional MRI characteristics have been studied such as contrast enhancement, multifocality, tumor location, edema, and cysts that are potentially associated with survival outcomes in GBM.[14] [15] [16] [17] [18] [19] [20] Therefore, in this study, we aimed to evaluate whether various sequences of MRI in addition to the conventional contrast MRI such as DTI, diffusion-weighted imaging (DWI), and susceptibility-weighted imaging (SWI) can predict the survival outcome in patients with GBM.

 

MissingLinkText

Materials and Methods

Patient Population and Study Design

A total of 45 adult patients who were newly diagnosed with GBM and underwent surgical treatment at our institution between January 2016 and March 2022 were reviewed retrospectively. The study was approved by the Institution's Ethical Committee, and it was performed in accordance with the Declaration of Helsinki. The written consent was not required due to the retrospective nature of the study. The study participants were included if they met the following criteria: (1) aged > 18 years, (2) presurgical MR scans available (including T1-weighted, T2-weighted, FLAIR, SWI, DTI, and postcontrast images), (3) pathologically confirmed GBM based on the World Health Organization (2016) histological grading system.[21] Patients with unavailability of preoperative MRI, recurrent brain tumors, and a history of previous adjuvant treatment were excluded. Two independent radiologists, each with an experience of more than 12 years, who were blinded, assessed the radiological parameters. All patients underwent maximal safe resection with postoperative focal radiotherapy (RT) with adjuvant TMZ therapy as per the standard protocol. The overall follow-up duration of the study was 24 months.

 

MissingLinkText

MRI Acquisition

Preoperative MRI examinations were performed in all individuals using a 3.0-Tesla MR system (Discovery MR750, GE Healthcare Systems, Milwaukee, Wisconsin, United States) equipped with a 32-channel high-resolution receiver head coil. Following MRI sequences were obtained.

1. Conventional sequences: Routine sequences included were axial T1-weighted, T2-weighted, fluid attenuation inversion recovery (FLAIR), and contrast-enhanced T1-weighted spin-echo sequences. Fast spin echo (FSE) T2-weighted images were obtained using the following MR parameters: TR-6235, TE-111, 5-mm section thickness, 1.5-mm intersection gap, 240 × 240-mm field of view (FOV), matrix 320 × 224, echo train length 36, and NEX 3.5. FSE T1-weighted imaging was performed using the following parameters: 600/24, 5-mm section thickness, 1.5-mm intersection gap, and 240 × 240-mm FOV, NEX 2, and echo train length 8. Postcontrast T1 images were acquired after the administration of 0.2 mmol of gadodiamide (Omniscan-GE, Cork, Ireland) per kilogram of body weight. FLAIR sequence was run using the following scan parameters: TR = 9,500 ms; TE = 89 ms; TI 2,250 ms, interslice gap = 1.5 mm; NEX 1, echo train length 18, matrix 320 × 224, FOV 240 × 240 mm. All the sequences in the study were analyzed in AW server 3.2 (Advantage Workstation, GE).

2. Tumor volume and edema/tumor volume ratio ([Fig. 1]): The volume of the tumor was assessed using T1 and T2 images. The tumor was traced throughout all slices occupied by it and was delineated and punched out manually using brush tools and volume was calculated using AW server workstation. The same was done in FLAIR images for perilesional vasogenic edema, which was traced all around the tumor, and the cumulative volume of edema across different sections was calculated. Subsequently, the ratio of edema/tumor volume was calculated manually. In this study, tumor volume was calculated not just by contrast enhancement but by correlating in T1, T2, and areas showing diffusion restriction. For example, tumor component not showing contrast enhancement but showing diffusion restriction was included within the tumor volume. Although postcontrast study was predominantly used for calculating viable tumor tissue volume, nonenhancing component of tumor, if any, was correlated with T1- and T2-weighted images as well as DWI and were included in the tumor volume. Sequences utilized for tumor volume calculation was tailored different for different cases.

3. Necrosis: In our study, necrosis was defined not just based on internal nonenhancing areas but based on homogenous internal T2/FLAIR hyperintense areas within the tumor. The area of necrosis was calculated from T2 and postcontrast T1 images. Visual scoring was done based on the percentage of tumor that had undergone necrosis.

Fig. 1 Tumor and edema volume calculation.

Grading of Necrosis

Grade	Necrosis
0	No necrosis
1	Limited to a small focal area or less than 25%
2	Necrotic area was 25 to 50% of the tumoral area
3	Necrotic area was more than 50% of the tumoral area

• 4. Enhancement: Postcontrast T1-weighted imaging with subtraction images was analyzed to detect enhancement and grade of enhancement in a manner similar to necrosis. In our study, grading of enhancement was done with the aim to account for the viable tissue within the tumor and try to correlate with survival. Even though not all parts of tumor enhances, our aim was just to assess if there is any relationship with this enhancing elements and survival, as an individual variable, independent of other confounding factors.

Grading of Enhancement

Grade	Enhancement
0	No enhancement
1	Mild enhancement: focal area or less than 25%
2	Moderate enhancement: 25 to 50% of the tumoral area
3	Marked enhancement: avidly enhancing tumor with more than 50% of the area of tumor enhancing

• 5. DWI and calculation of absolute apparent diffusion coefficient (ADC) value ([Fig. 2]): A DWI sequence was performed in three orthogonal directions with three different b-values (0, 800, and 1000 s/mm²) with the following parameters: section thickness 5 mm; FOV 26 × 26 cm; matrix 192 × 128; spacing between slices 1.5 mm; repetition time/echo time 5703/69.5 ms. From the ADC maps, region of interest (ROI) was kept excluding blood vessels and areas of hemorrhagic, cystic, and/or necrotic areas, and absolute ADC in 10–6 mm²/s was recorded.

• 7. Calculation of fractional anisotropy (FA) from diffusion tensor imaging (DTI) ([Fig. 3]): The diffusion MRI acquisition was performed using a single-shot spin echo-echo planar imaging sequence with gradients applied in 6 noncollinear directions with a b-value of 1,500 s/mm². The following parameters were used for acquisition: b = 0, 800, and 1000, TR = 5,316 ms, TE 114ms, and FOV 24 × 24 cm, matrix 128 × 140, slice thickness 5 mm, and spacing 1.5 mm. FA maps were generated by postprocessing, which was used as a major workhorse for the calculation of tumoral FA. ROI of similar size was kept within the tumor and reference ROI was placed in the contralateral brain matter. Then, relative FA was calculated by dividing tumoral FA by its contralateral normal white matter value. Three ROIs for FA calculation were copied and pasted upon diffusion-weighted image for comparison. Areas showing higher diffusion restriction indicating high cellularity were chosen, based on the assumption that these areas will show more white matter tract infiltration resulting in FA alterations. To avoid the inhomogeneities within the tumor, three ROIs were placed, and mean FA was included in the study.

• 8. SWI for grading of microhemorrhage ([Fig. 4]): Axial SWI data were acquired using the following imaging parameters: matrix 288 × 288; slice thickness of 1.6 mm; TR 23 ms, TE 25 ms; and flip angle of 15 degrees. The magnitude and phase data in SWI were brought together. Imaging data were postprocessed to highlight any signal intensity drop caused by susceptibility effects following acquisition. The first step was to remove phase shifts generated by static magnetic field heterogeneities from the images. After that, a phase mask was constructed from the MR phase images and multiplied by magnitude data to produce an enhanced contrast magnitude image that is particularly sensitive to slow venous blood and other sources of susceptibility effects such as hemorrhage, calcium, and iron storage.

Fig. 2 Calculation of apparent diffusion coefficient (ADC).

Fig. 3 Calculation of relative fractional anisotropy (FA).

Fig. 4 Microhemorrhage/intratumoral susceptibility signal (ITSS) occupying > 50% of tumor grade 3 microhemorrhage.

Intratumoral susceptibility signals (ITSSs) in previous studies were based on their exact number within the tumor; however, since our study comprised of glioblastoma, which was more prone to a higher degree of microhemorrhage and counting of ITSS was impractical, we devised a rational classification system as follows.

Grading of Microhemorrhage

Grading	Microhemorrhage
0	No microhemorrhage
1	ITSSs occupying less than 25% of tumor
2	ITSSs involving 25 to 50% of tumor
3	ITSSs involving > 50% of tumor

 

MissingLinkText

Statistical Analysis

We have used SPSS-26 software for data analysis. Categorical variables were presented as proportions. Continuous variables were presented as median and interquartile range values. Shapiro–Wilk test was used to verify the normal distribution of all variables. Nonparametric tests, such as the Mann–Whitney U test, and parametric tests, such as the independent sample t-test, were applied for the continuous variables and the chi-square test for the categorical variables. Pearson's correlation coefficient was used to measure the association between mortality with radiological variables with a significance test for the correlation coefficient. Receiver operating characteristic (ROC) curve analysis was used to determine the optimum cutoff points of significantly correlated variables with the mortality/survival and area under the curve (AUC) with a 95% confidence interval (CI) with a significance value for each variable. The cutoff point was determined using the index of union method. Kaplan–Mayer survival analysis was done. Based on ROC cutoff values, the radiological parameters were dichotomized and the log-rank test was conducted to determine the significance of the difference in median survival time between those two groups. We ran an univariate and multivariate COX regression to identify the hazard ratio (HR). A two-tailed p-value of < 0.05 was considered statistically significant.

 

MissingLinkText

 

MissingLinkText

Results

Study Participants

A total of 45 patients were included in this study. There were 27 males and 18 females among the study participants. The mean age was 31.4 ± 19.2 years. The mean time of follow-up since diagnosis was 58.3 ± 27 weeks. Fifteen patients were alive and 30 were dead at the last follow-up. [Table 1] includes the demographic characteristics of study participants.

Table 1 Demographic and radiological profile of the GBM patients

Characteristics	Values
Mean age	31.42 ± 19.2 y
Sex
Male  Female	27 (60%) 18 (40%)
Status
Live  Death	15 (33.33%) 30 (66.67%)
Mean follow-up since diagnosis	58.3 ± 27 wk
Mean ADC	1036 ± 204.23 * 10−6 mm2/s
Mean FA	0.517 ± 0.122
Mean edema:tumor volume ratio	1.51 ± 0.59
Mean volume of tumor	635.28 ± 1229.4 cm3
Mean volume of perilesional edema	640.19 ± 1166.53 cm3
Necrosis (%)
0  1  2  3	16 (35.56) 13 (28.89) 12 (26.67) 4 (8.89)
Enhancement (%)
0  1  2  3	4 (8.89) 13 (28.89) 22 (48.89) 6 (13.33)
Microhemorrhage (%)
0  1  2  3	11 (24.44) 17 (37.78) 10 (22.22) 7 (15.56)

Abbreviations: ADC, apparent diffusion coefficient; FA, fractional anisotropy; GBM, glioblastoma multiforme.

 

MissingLinkText

Comparison of Radiological Parameters between Survived and Dead Patients [Table 2]

Among all those parameters, there were significant differences among edema:tumor volume ratio, FA, along with the presence of necrosis, microhemorrhage, and enhancement, between those who survived and those who did not. The values of edema:tumor volume ratio was higher among the patients who did not survive. The value of FA was lower among the patients who died. The patients who survived had no necrosis. The majority of patients who did not survive had necrosis between 25 and 50%. There was no statistically significant difference between age, sex, and mean follow-up period of the survived and dead patients. [Table 2] demonstrates the comparison between the radiological parameters between alive and dead patients.

Table 2 Comparison of radiological parameters between survived and dead

Radiological parameters	p-Value (Shapiro–Wilk test)	Live (n = 15)	Dead (n = 30)	p-Value
1. Age	0.06	20.5 ± 12.5	36.7 ± 20.9	0.106
2. Volume of tumor	0.001	1212 (0.87–2490)	18.56 (15.92–31.06)	0.626
3. Volume of perilesional edema	0.001	1008 (0.64–2349)	39.34 (21.56–64.32)	0.770
4. Edema:tumor volume ratio	0.06	0.8 ± 0.1	1.6 ± 0.6	0.007[*]
5. ADC	0.25	1154.4 ± 222.7	1021.9 ± 188.2	0.183
6. FA	0.001	0.68 (0.55–0.8)	0.49 (0.41–0.63)	0.04[*]
7. Sex				0.759
Male  Female		66.67% 33.33%	58.33% 41.67%
8. Necrosis				0.038[*]
0  1  2  3		53.33% 46.67%	26.67% 20% 40% 13.33%
9. Enhancement				0.038[*]
0			13.33%
1		46.67%	20%
2		53.33%	46.67%
3			20%
10. Microhemorrhage				0.012[*]
0  1  2  3		60% 40%	6.67% 36.67% 33.33% 23.33%
11. Mean follow-up (in weeks)		59.3 ± 16	24 ± 45	0.1

Abbreviations: ADC, apparent diffusion coefficient; FA, fractional anisotropy.

Note: “n” denotes the number of sample taken into consideration for analysis of that particular parameter.

^* p-value less than 0.05 considered significant.

 

MissingLinkText

Correlation of Radiological Parameters with Mortality

Out of all radiological parameters, the edema:tumor volume ratio showed a positive and significant correlation with mortality in GBM patients. FA value was found to be negatively and significantly correlated with mortality. In other words, a lower value of FA was associated with higher mortality. [Table 3] demonstrates the correlation of various radiological parameters with mortality in GBM patients.

Table 3 Correlation coefficients of radiological parameters with mortality

Dependent variable	Parameters	Correlation coefficients	p-Value
Mortality	1. Age	0.324	0.106
	2. Edema:tumor ratio	0.517	0.007[*]
	3. ADC	−0.269	0.183
	4. Volume of tumor	−0.098	0.635
	5. Volume of perilesional edema	−0.059	0.776
	6. FA	−0.410	0.037[*]

Abbreviations: ADC, apparent diffusion coefficient; FA, fractional anisotropy.

^* p-value less than 0.05 considered significant.

 

MissingLinkText

Cutoff Values of Each Radiological Parameter in Predicting Mortality or Survival

ROC curves of the radiological parameters were plotted to determine the specific cutoff values predicting the mortality/survival of GBM patients. The significantly associated variables were edema:tumor volume ratio and FA ([Table 4]). Except for FA, the rest of the parameters predicted mortality. For FA, ROC curve predicted the survival of the patients. The ROC curve of edema:tumor volume ratio (AUC = 0.895, 95% CI: 0.768–1) suggested that the best cutoff point predicting the mortality was ≥ 0.905 units with a sensitivity of 90.5% and specificity of 80%. The ROC curve of FA (AUC = 0.8, 95% CI: 0.63–0.97) suggested that the best cutoff point was ≥ 0.655 units predicting the survival of GBM patients with a sensitivity of 80% and specificity of 81%. In [Fig. 5], we showed the Kaplan–Meier survival plot based on the ROC cutoff. In that we can see that patients with edema:tumor volume ratio equal or more than the ROC cutoff had shorter survival and patients with FA values less than the ROC cutoff values had shorter survival.

Table 4 Cutoff values of radiological parameters to predict mortality/ survival

Variable	Prediction	Cutoff	Sensitivity	Specificity	AUC	p-Value
1. Edema:tumor volume	Mortality	0.905	90.5%	80%	0.895 (0.768–1)	0.007[*]
2. FA	Survival	0.655	80%	81%	0.8 (0.63–0.97)	0.04[*]

Abbreviations: AUC, area under the curve; FA, fractional anisotropy.

^* p-value less than 0.05 considered significant.

Fig. 5 Kaplan–Meier curves of survival analysis (A, edema:tumor volume ratio; B fractional anisotropy [FA] value) (blue line – values < receiver operating characteristic [ROC] cutoff, red line – values ≥ ROC cutoff).

 

MissingLinkText

Univariate and Multivariate Analysis

[Table 5] demonstrates the result of univariate and multivariate Cox regression. In univariate analysis, microhemorrhage was found to be statistically significant. The hazard of death in patients with microhemorrhage of more than 25% was 2.919 (95% CI: 1.172–7.27) times higher than in patients with microhemorrhage of less than 25%.

Table 5 Univariate and multivariate Cox regression

Parameters	Univariate			Multivariate
	Unadjusted hazard ratio	95% CI	p-Value	Adjusted hazard ratio	95% CI	p-Value
Age	1.01	0.99–1.03	0.36	1.06	0.98–1.14	0.11
Sex (female)	1.05	0.43–2.56	0.91	3.47	0.54–22.07	0.19
Volume of tumor	1	0.99–1	0.26	1.01	1–1.01	0.04[*]
Necrosis (> 25%)	2.12	0.87–5.17	0.1	0.56	0.55–5.69	0.62
Enhancement (> 25%)	2.18	0.89–5.32	0.09	0.01	0–0.9	0.04[*]
Volume of perilesional edema	1	0.99–1	0.25	0.99	0.90–0.99	0.02[*]
Edema:tumor volume ratio	1.93	0.98–3.78	0.06	7.41	1.04–52.98	0.04[*]
Microhemorrhage (> 25%)	2.92	1.17–7.27	0.02[*]	45.94	3.01–703.23	0.01[*]
ADC	0.99	0.99–1.01	0.19	1.01	0.99–1.001	0.52
FA	1.04	0.99–1.09	0.08	1.06	0.96–1.17	0.24

Abbreviations: ADC, apparent diffusion coefficient; CI, confidence interval; FA, fractional anisotropy.

^* p-value less than 0.05 considered significant.

In multivariate COX regression, the volume of tumor, enhancement, the volume of perilesional edema, edema:tumor volume ratio, and microhemorrhage were found to be significant. Patients with microhemorrhage of more than 25% had 45.9 times significantly more likelihood of dying because of GBM as compared with patients with microhemorrhage of less than 25%.

 

MissingLinkText

 

MissingLinkText

Discussion

Previous survival analysis studies for patients with GBM have tested various factors, including the extent of resection, tumor grade, method of adjuvant RT, radiation dose, age, Karnofsky Performance Status scale, and conventional MRI features.[15] [22] [23] [24] [25] [26] [27] [28] In this study, data from various MRI sequences were evaluated to assess the predictive value of survival for patients with GBM who were treated with maximal safe resection, radiation therapy, and chemotherapy.

Several factors influence ADC levels in brain tissues. Water motion in the interstitium is the primary source of elevated ADC values. Normal brain tissues are replaced by tumor cells in most glioma-affected locations and water diffusion in gliomas may be primarily influenced by tumor cellularity. The presence of many tumor cells in normal brain tissue is expected to limit water transport, resulting in lower ADC values in tumors with a greater histologic cellular composition. Thus, detecting the locations with the lowest ADC values may be beneficial in evaluating gliomas as it corresponds to the more aggressive behavior of the tumor. Sugahara et al discovered that increasing tumor cellularity was associated with decreased ADC values.[29] Among the GBM patients, the mean value of ADC was found to be (0.82 ± 0.13) 10⁻³ mm²/s according to Kono et al,[32] and (0.83 ± 0.14) 10⁻³ mm²/s according to Higano et al.[30] In this study, the mean ADC was 1 ± 0.2 × 10⁻³ mm²/s for the GBM patients. In a few studies, it was also shown that there was a significantly shorter median survival time for patients with low mean normalized ADC (nADC) values than high nADC values. The observation that low ADC is associated with shorter survival is because of higher cellularity and hence larger tumor burden.[29] [31] [32] [33] In this study, the mean value of ADC was higher among the live patients than those who died but it was not statistically significant. The ADC values were found to be negatively correlated with mortality.

FA is a common DTI measurement. Gliomas regularly have lower FA values than normal white matter. An anisotropy loss owing to tumor infiltration can explain the lower FA value of cerebral glioma than normal white matter. Due to tumor invasion, white matter anisotropy is reduced; and the more the aggressive tumor, the more will be the tumor invasion and less will be their FA values. In a previous study done by Sinha et al, the mean tumoral relative FA was 0.23 ± 0.07 for high-grade glioma (HGG).[34] In our study, the mean FA for all those HGGs was 0.517 ± 0.122, which was slightly higher than previous studies. The median value of FA was lower among the patients who died and the difference with FA of the patients who survived was statistically significant. This was suggestive of the presence of more aggressive behavior of tumors among those patients. Therefore, the values of FA were found to be significantly negatively correlated with mortality.

Glioma-related edema is a key factor in glioma-related morbidity and death. Glioma cells cause peritumoral brain edema by an active mechanism that provides a favorable environment for peritumoral invasive cells, implying that glioma-related peritumoral brain edema is determined by tumor cell invasiveness.[35] Studies have shown that peritumoral edema (PTE) plays a vital role in the symptoms of GBM patients, which is the main cause of neurological impairment.[36] PTE on preoperative MRI can serve as an independent prognostic factor in addition to postoperative Karnofsky Performance Scale score, age, and type of tumor resection. Patients with major edema have significantly shorter overall survival (OS) compared with patients with minor edema.[37] [38] [39] A retrospective study by Liu et al revealed that preoperative PTE was an independent prognostic factor for decreased survival in GBM patients.[40] Mummareddy et al[41] performed segmental volume analysis of preoperative MRI in 210 GBM patients and found that elevated PTE level was associated with decreased survival. In this study along with the volume of perilesional edema, we have calculated the edema:tumor volume ratio. There was no significant difference in the volume of tumor and volume of perilesional edema among live and dead patients. However, the edema:tumor volume ratio was significantly higher among the dead patients. The ratio was found to be significantly positively correlated with mortality. In multivariate COX analysis, the volume of tumor along with perilesional edema volume and edema:tumor volume ratio were found to be significant. However, the HR of the first two parameters were almost equal to 1 but the HR of the edema:tumor volume ratio was 7.4. With each unit increase in the edema:tumor volume ratio, the chances of dying from GBM significantly increased by 640.99%. These findings indicate that this may be a better predictor for mortality than individual volume of tumor and edema volume.

More aggressive tumors will have higher necrosis and less survival. Henker et al prospectively evaluated preoperative MRI images from patients harboring a primary supratentorial GBM and illustrated that preoperatively measured necrosis volume was one of the most important radiological features of GBM with a strong influence on OS.[42] In our study, we found that all the patients with grade 2 and grade 3 necrosis died. Note that 40 and 13% of the patients who died had grade 2 and grade 3 necrosis, respectively [Table 2]. This difference was statistically significant, which implies that high-grade necrosis was associated with high mortality.

Theoretically, when the grade of tumor increases, more will be the neoangiogenesis, and most of these vessels are friable and are prone to rupture fast leading to microhemorrhages. So, the patients who died will have more microhemorrhage. In our data we have found similar results. Mortality was strongly related to >25% microhemorrhage ([Table 5]) whereas no patients who lived had grade 2 or 3 microhemorrhage [Table 2]. This difference was statistically significant. It was found to be a significant parameter both in univariate and multivariate cox regression. This suggests that microhemorrhage can predict the survival of GBM patients. Like microhemorrhage, enhancement of the tumor was also found to be a significant predictive factor for survival of GBM patients. The person who died had tumor of higher grade enhancement than the persons who lived [Table 5]. In multivariate cox regression, it was a significant factor.

Several limitations of this study should be considered. First, the patients were enrolled from a single institution, and second, the data were analyzed retrospectively. We had a limited sample size. The results of this study should be validated in a large prospective cohort.

The standardization and integration of radiological biomarkers, including FA and the edema-to-tumor volume ratio, into clinical procedures are crucial for improving their effectiveness in prognostic models. Standardizing FA values necessitates the implementation of similar MRI acquisition protocols, preprocessing techniques, and consistent placement of ROIs to reduce inter- and intraobserver variability. Multicenter investigations could further corroborate the thresholds, considering discrepancies in scanner technology and software. The edema-to-tumor volume ratio can be enhanced through automated volumetric analysis employing sophisticated segmentation techniques, promoting consistency and simplifying implementation. Machine learning models may integrate these metrics with other recognized prognostic factors to enhance predictions and assist in treatment planning. These biomarkers may be integrated into current workflows by incorporating them into clinical imaging platforms or decision-support systems. Radiology reports may incorporate these data as standard elements, offering clinicians relevant information to inform therapy decisions and monitoring techniques.

 

MissingLinkText

Conclusion

In conclusion, this study examines the potential value of the radiological parameters in predicting the mortality and survival of GBM patients. The development of new radiological characteristics that significantly predict survival in surgically treated GBM cases is of considerable clinical importance. These indicators can enhance patient classification, facilitating customized decisions about adjuvant medicines and increased monitoring for high-risk people. Integrating these prognostic markers into clinical procedures can optimize GBM management, potentially enhancing survival outcomes and improving resource allocation. This method highlights the significance of precision medicine in tackling the difficulties posed by aggressive cancers such as GBM. The edema:tumor volume ratio along with FA can significantly predict the mortality/survival with good sensitivity and specificity. A significantly shorter life span was seen in GBM with higher edema:tumor volume ratio.

 

MissingLinkText

 

MissingLinkText

Conflict of Interest

None declared.

Note

Presented at Organization – NSICON 2023 (71st Annual Conference of Neurological Society of India); Place – Bhubaneswar; Date: December 17, 2023.

Authors' Contributions

A.T. contributed to the conceptualization and design of the study, defining the intellectual content, conducting the literature search, performing clinical and experimental studies, data acquisition, data analysis, statistical analysis, manuscript preparation, manuscript editing, manuscript review, and is a guarantor of the study.

M.S. contributed to the literature search, performing experimental studies, data analysis, statistical analysis, manuscript preparation, manuscript editing, manuscript review, and is a guarantor of the study.

S.K. contributed to the conceptualization and design of the study, defining the intellectual content, performing clinical studies, data acquisition, data analysis, manuscript preparation, manuscript editing, manuscript review, and is a guarantor of the study.

A.K.D. contributed to the conceptualization and design of the study, defining the intellectual content, performing clinical studies, data acquisition, manuscript editing, and manuscript review.

S.K.S contributed to the conceptualization and design of the study, performing clinical studies, data acquisition, data analysis, manuscript preparation, manuscript editing, manuscript review, and is a guarantor of the study.

• References

• 1 Louis DN, Perry A, Reifenberger G. et al. The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol 2016; 131 (06) 803-820
• 2 Nam JY, de Groot JF. Treatment of glioblastoma. J Oncol Pract 2017; 13 (10) 629-638
• 3 Paolillo M, Boselli C, Schinelli S. Glioblastoma under siege: an overview of current therapeutic strategies. Brain Sci 2018; 8 (01) 15
• 4 Dolecek TA, Propp JM, Stroup NE, Kruchko C. CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in 2005-2009. Neuro-oncol 2012; 14 (Suppl 5, Suppl 5): v1-v49
• 5 Hu LS, Ning S, Eschbacher JM. et al. Radiogenomics to characterize regional genetic heterogeneity in glioblastoma. Neuro-oncol 2017; 19 (01) 128-137
• 6 Bae S, Choi YS, Ahn SS. et al. Radiomic MRI phenotyping of glioblastoma: improving survival prediction. Radiology 2018; 289 (03) 797-806
• 7 Wang B, Zhang S, Wu X. et al. Multiple survival outcome prediction of glioblastoma patients based on multiparametric MRI. Front Oncol 2021; 11: 778627
• 8 Ammari S, Sallé de Chou R, Balleyguier C. et al. A predictive clinical-radiomics nomogram for survival prediction of glioblastoma using MRI. Diagnostics (Basel) 2021; 11 (11) 2043
• 9 Pak E, Choi KS, Choi SH. et al. Prediction of prognosis in glioblastoma using radiomics features of dynamic contrast-enhanced MRI. Korean J Radiol 2021; 22 (09) 1514-1524
• 10 Lao J, Chen Y, Li ZC. et al. A deep learning-based radiomics model for prediction of survival in glioblastoma multiforme. Sci Rep 2017; 7 (01) 10353
• 11 Tixier F, Um H, Bermudez D. et al. Preoperative MRI-radiomics features improve prediction of survival in glioblastoma patients over MGMT methylation status alone. Oncotarget 2019; 10 (06) 660-672
• 12 Stringfield O, Arrington JA, Johnston SK. et al. Multiparameter MRI predictors of long-term survival in glioblastoma multiforme. Tomography 2019; 5 (01) 135-144
• 13 Priya S, Agarwal A, Ward C, Locke T, Monga V, Bathla G. Survival prediction in glioblastoma on post-contrast magnetic resonance imaging using filtration based first-order texture analysis: comparison of multiple machine learning models. Neuroradiol J 2021; 34 (04) 355-362
• 14 Hammoud MA, Sawaya R, Shi W, Thall PF, Leeds NE. Prognostic significance of preoperative MRI scans in glioblastoma multiforme. J Neurooncol 1996; 27 (01) 65-73
• 15 Lacroix M, Abi-Said D, Fourney DR. et al. A multivariate analysis of 416 patients with glioblastoma multiforme: prognosis, extent of resection, and survival. J Neurosurg 2001; 95 (02) 190-198
• 16 Pope WB, Sayre J, Perlina A, Villablanca JP, Mischel PS, Cloughesy TF. MR imaging correlates of survival in patients with high-grade gliomas. AJNR Am J Neuroradiol 2005; 26 (10) 2466-2474
• 17 Jafri NF, Clarke JL, Weinberg V, Barani IJ, Cha S. Relationship of glioblastoma multiforme to the subventricular zone is associated with survival. Neuro-oncol 2013; 15 (01) 91-96
• 18 Kappadakunnel M, Eskin A, Dong J. et al. Stem cell associated gene expression in glioblastoma multiforme: relationship to survival and the subventricular zone. J Neurooncol 2010; 96 (03) 359-367
• 19 Maldaun MV, Suki D, Lang FF. et al. Cystic glioblastoma multiforme: survival outcomes in 22 cases. J Neurosurg 2004; 100 (01) 61-67
• 20 Utsuki S, Oka H, Suzuki S. et al. Pathological and clinical features of cystic and noncystic glioblastomas. Brain Tumor Pathol 2006; 23 (01) 29-34
• 21 Louis DN, Ohgaki H, Wiestler OD, Cavenee WK. WHO Classification of Tumours of the Central Nervous System. 4th ed.. Lyon: IARC Press; 2016
• 22 Simpson JR, Horton J, Scott C. et al. Influence of location and extent of surgical resection on survival of patients with glioblastoma multiforme: results of three consecutive Radiation Therapy Oncology Group (RTOG) clinical trials. Int J Radiat Oncol Biol Phys 1993; 26 (02) 239-244
• 23 Kreth FW, Berlis A, Spiropoulou V. et al. The role of tumor resection in the treatment of glioblastoma multiforme in adults. Cancer 1999; 86 (10) 2117-2123
• 24 Kowalczuk A, Macdonald RL, Amidei C. et al. Quantitative imaging study of extent of surgical resection and prognosis of malignant astrocytomas. Neurosurgery 1997; 41 (05) 1028-1036 , discussion 1036–1038
• 25 Albert FK, Forsting M, Sartor K, Adams HP, Kunze S. Early postoperative magnetic resonance imaging after resection of malignant glioma: objective evaluation of residual tumor and its influence on regrowth and prognosis. Neurosurgery 1994; 34 (01) 45-60 , discussion 60–61
• 26 Shrieve DC, Alexander III E, Black PM. et al. Treatment of patients with primary glioblastoma multiforme with standard postoperative radiotherapy and radiosurgical boost: prognostic factors and long-term outcome. J Neurosurg 1999; 90 (01) 72-77
• 27 Lote K, Egeland T, Hager B. et al. Survival, prognostic factors, and therapeutic efficacy in low-grade glioma: a retrospective study in 379 patients. J Clin Oncol 1997; 15 (09) 3129-3140
• 28 Larson DA, Prados M, Lamborn KR. et al. Phase II study of high central dose Gamma Knife radiosurgery and marimastat in patients with recurrent malignant glioma. Int J Radiat Oncol Biol Phys 2002; 54 (05) 1397-1404
• 29 Sugahara T, Korogi Y, Kochi M. et al. Usefulness of diffusion-weighted MRI with echo-planar technique in the evaluation of cellularity in gliomas. J Magn Reson Imaging 1999; 9 (01) 53-60
• 30 Higano S, Yun X, Kumabe T. et al. Malignant astrocytic tumors: clinical importance of apparent diffusion coefficient in prediction of grade and prognosis. Radiology 2006; 241 (03) 839-846
• 31 Tien RD, Felsberg GJ, Friedman H, Brown M, MacFall J. MR imaging of high-grade cerebral gliomas: value of diffusion-weighted echoplanar pulse sequences. AJR Am J Roentgenol 1994; 162 (03) 671-677
• 32 Kono K, Inoue Y, Nakayama K. et al. The role of diffusion-weighted imaging in patients with brain tumors. AJNR Am J Neuroradiol 2001; 22 (06) 1081-1088
• 33 Guo AC, Cummings TJ, Dash RC, Provenzale JM. Lymphomas and high-grade astrocytomas: comparison of water diffusibility and histologic characteristics. Radiology 2002; 224 (01) 177-183
• 34 Sinha S, Bastin ME, Whittle IR, Wardlaw JM. Diffusion tensor MR imaging of high-grade cerebral gliomas. AJNR Am J Neuroradiol 2002; 23 (04) 520-527
• 35 Lin Z-X. Glioma-related edema: new insight into molecular mechanisms and their clinical implications. Chin J Cancer 2013; 32 (01) 49-52
• 36 Dubinski D, Hattingen E, Senft C. et al. Controversial roles for dexamethasone in glioblastoma - opportunities for novel vascular targeting therapies. J Cereb Blood Flow Metab 2019; 39 (08) 1460-1468
• 37 Wu C-X, Lin G-S, Lin Z-X. et al. Peritumoral edema on magnetic resonance imaging predicts a poor clinical outcome in malignant glioma. Oncol Lett 2015; 10 (05) 2769-2776
• 38 Leroy HA, Delmaire C, Le Rhun E, Drumez E, Lejeune JP, Reyns N. High-field intraoperative MRI in glioma surgery: a prospective study with volumetric analysis of extent of resection and functional outcome. Neurochirurgie 2018; 64 (03) 155-160
• 39 Zhao M, Guo LL, Huang N. et al. Quantitative analysis of permeability for glioma grading using dynamic contrast-enhanced magnetic resonance imaging. Oncol Lett 2017; 14 (05) 5418-5426
• 40 Liu SY, Mei WZ, Lin ZX. Pre-operative peritumoral edema and survival rate in glioblastoma multiforme. Onkologie 2013; 36 (11) 679-684
• 41 Mummareddy N, Salwi SR, Ganesh Kumar N. et al. Prognostic relevance of CSF and peri-tumoral edema volumes in glioblastoma. J Clin Neurosci 2021; 84: 1-7
• 42 Henker C, Kriesen T, Glass Ä, Schneider B, Piek J. Volumetric quantification of glioblastoma: experiences with different measurement techniques and impact on survival. J Neurooncol 2017; 135 (02) 391-402

Publication History

Article published online:
04 June 2025