Assessing radiological response of IDH-mutated glioma to vorasidenib by tumor diameter slopes: a series of 8 cases Authors: Emmanuel Mandonnet Published: 2026 ABSTRACT The INDIGO trial demonstrated improved progression-free survival under vorasidenib in IDH-mutated grade 2 glioma patients. However, the pattern of vorasidenib effects remains unknown. We thus performed a retrospective analysis of tumor kinetics in eight patients enrolled in the INDIGO trial. Tumor diameters (computed from FLAIR volumetry) and their slopes were individually estimated, before and under vorasidenib. Three patients showed no change in slope. Three patients exhibited a marked slope decrease, and two demonstrated slope inversion. Tumor diameter slopes allows individual assessment of radiological response, providing clinically relevant insights that cannot be captured by the cohort-level analysis performed in the INDIGO trial. Introduction The recent results of the INDIGO trial brought some hope to improve survival of IDH-mutated patients, by demonstrating improved progression-free survival after a follow-up of 14.2 months [5]. The recent update [1] not only confirmed the progression-free survival difference between the two arms after 6 additional months of follow-up (58.8% vs 26.2% at 24 months) but also provided some insights into 6-months volumic tumor growth rates (–1.3% vs 14.4%). However, this analysis was performed at the cohort-level, thus masking individual heterogeneities in vorasidenib effect. In a first attempt to address this gap in knowledge, we retrospectively analyzed individual evolutions of tumor diameter in a small series of 8 patients included in the INDIGO trial. We demonstrated that, indeed, three distinct patterns of evolution were observed (progression, stabilization, response) and we discuss how these new data should help to inform patients about expected benefits of vorasidenib. Methods Patients We retrospectively analyzed a series of eight patients who underwent awake surgery (between 2016 and 2021) in our institution and were then included (between March 2021 and January 2022) in the INDIGO trial, testing vorasidenib versus placebo in grade 2 IDH-mutated patients. Seven of the eight patients were assigned to the experimental arm. The only patient affected to the placebo group underwent the crossover in March 2023. Imaging All patients had an MRI follow-up as per protocol recommendations. In the present series, all patients underwent every 3 months a 3 Teslas examination with 3D-Flair sequences. For each patient, volumetry on FLAIR sequences was manually performed by the author, using Carestream PACS software. Tumors were segmented slice by sclice, on axial slices (3 mm thick, no gap). The volumes were then converted into a diameter (D = 3√(2 V)). This volume-to-diameter conversion was motivated by the mathematical rationale and clinical validation that the curve of diameter evolution is linear with time [4], for both spontaneous evolution [3] and under chemotherapy [8, 9]. For the period before vorasidenib, the first postoperative and last available MRIs were selected, allowing to simply compute the slope (in mm/year) as (Dlast-Dfirst)/time period, where the diameters are expressed in mm and the time period in years. For the period under vorasidenib, we selected yearly time points and estimated the slope of each patient by linear regression of the diameter against time. Results The Table 1 summarizes the characteristics of the present series. Median age at surgery was 42.5 years (range 32–54) and initial symptoms were generalized seizures in 3 patients and partial seizures in 5 patients. Median preoperative volume was 28 cc (range 20–125) and median extent of resection was 98% (range 92–100). Table 1Series characteristicsInitial symptomsTumor location1p19 codelAge (years)Preop volume (cc)Postop volume (cc)EOR (%)Delay surgery-vora (months)Pre-vora volume (cc)Slope pre-vora (mm/year)Slope on-vora (mm/year)FU (months)Status1PSFrontalNo532601005219,31,5058On-vora2PSInsularNo321256,59531325,46,110Redo3GSInsularNo34453,293102210−1,346On-vora4PSTemporalNo4125,301003011,84,873Redo5PSInsularNo44201,59233122,8−1,647On-vora6GSFrontalNo543001003011,532,436Redo7GSFrontalYes4532,501003217,170,345On-vora8PSInsularNo41251964717,43,7−0,647On-voraPS partial seizure, GS generalized seizure, EOR extent of resection, FU follow-up Table 1Series characteristicsInitial symptomsTumor location1p19 codelAge (years)Preop volume (cc)Postop volume (cc)EOR (%)Delay surgery-vora (months)Pre-vora volume (cc)Slope pre-vora (mm/year)Slope on-vora (mm/year)FU (months)Status1PSFrontalNo532601005219,31,5058On-vora2PSInsularNo321256,59531325,46,110Redo3GSInsularNo34453,293102210−1,346On-vora4PSTemporalNo4125,301003011,84,873Redo5PSInsularNo44201,59233122,8−1,647On-vora6GSFrontalNo543001003011,532,436Redo7GSFrontalYes4532,501003217,170,345On-vora8PSInsularNo41251964717,43,7−0,647On-voraPS partial seizure, GS generalized seizure, EOR extent of resection, FU follow-up Series characteristics PS partial seizure, GS generalized seizure, EOR extent of resection, FU follow-up The median follow-up time between surgery and treatment initiation by vorasidenib was 31.5 months (range 10–51). At that time, median pre-vora volume was 17.25 cc (range 11.5–32). The median slope of tumor diameter pre-vora evolution was 4.25 mm/year (range 1.5–10). Under vorasidenib, the median slope of tumor diameter was 0.15 mm/year (range −1.6 – 7). The Fig. 1 gives the individual curves for each of the eight cases. In three cases (# 2,4,6; in red on the Fig. 1), the slope was unchanged. These three patients were reoperated on, after respectively 10, 3, and 36 months of vorasidenib treatment. The cases 2 & 4 have been extensively studied in a previous publication [10]. The Fig. 2 shows the imaging of case 6. The immuno-chemical examination of the redo surgery of this patient found some scant IDH-stained cells.Fig. 1Diameter evolution before and under vorasidenib treatment. Vorasidenib was started at time 0, except for the purple case (patient # 8), in whom vorasidenib was started after cross-over. No change in slopes was observed for the cases highlighted in red (non-responders; cases # 2, 4, 6), while inversion of slope was observed for the cases highlighted in green (responders; cases # 3,5). Tumor diameter stabilized in the cases in yellow and purple (cases # 1, 7, 8)Fig. 2Longitudinal FLAIR imaging in a non-responder (case # 6), illustrative of an unchanged growth slope of tumor diameter under vorasidenib Diameter evolution before and under vorasidenib treatment. Vorasidenib was started at time 0, except for the purple case (patient # 8), in whom vorasidenib was started after cross-over. No change in slopes was observed for the cases highlighted in red (non-responders; cases # 2, 4, 6), while inversion of slope was observed for the cases highlighted in green (responders; cases # 3,5). Tumor diameter stabilized in the cases in yellow and purple (cases # 1, 7, 8) Longitudinal FLAIR imaging in a non-responder (case # 6), illustrative of an unchanged growth slope of tumor diameter under vorasidenib In two cases (# 3,5; in green on the Fig. 1), the tumor volume underwent a decrease, evidenced as a slope inversion. The Fig. 3 shows the imaging of case # 3, illustrating the tumor shrinkage at the level of the anterior perforated substance residue.Fig. 3Longitudinal FLAIR imaging in a responder (case # 3), illustrative of a slope inversion. Note the significant decrease of FLAIR hypersignal at the level of the anterior perforated substance residue, but the slight increase at the level of the superior temporal gyrus Longitudinal FLAIR imaging in a responder (case # 3), illustrative of a slope inversion. Note the significant decrease of FLAIR hypersignal at the level of the anterior perforated substance residue, but the slight increase at the level of the superior temporal gyrus Finally, in three cases (# 1,7,8; in yellow and purple on the Fig. 1), the tumor volume remained stable under vorasidenib. The Fig. 4 shows the imaging of case 1, illustrating the tumor stabilization under vorasidenib.Fig. 4Longitudinal FLAIR imaging in patient # 1, illustrative of a stabilization under vorasidenib Longitudinal FLAIR imaging in patient # 1, illustrative of a stabilization under vorasidenib The median follow-up under vorasidenib was 45.5 months (range 3–56). At last follow-up, the three progressive patients had been reoperated while the 5 others were still on vorasidenib. Discussion In this case series, we analyzed the individual radiological response under vorasidenib in 8 patients included in the INDIGO trial [5]. Our methodology allowed to identify distinct patterns of vorasidenib effects, suggestive of failure, response, and stabilization, respectively. Our approach contrasts with the group-level methodology used in the recent update of the INDIGO trial estimating the volumetric evolution in the experimental and placebo groups [1]. These authors applied a linear mixed model on the log-transformed volume, a methodology which is disputable, for several reasons. First, it assumes an exponential growth of the volume, an hypothesis that has been repeatedly shown to be unmet in diffuse low-grade glioma [3, 4], at least for spontaneous growth. Indeed, it is rather expected that the diameter (D = 3√ (2xV)) is growing linearly with time [2]. Hence the volume evolution should follow a third order polynomial law. To address this question, it would be very interesting that Cloughesy & al. [1] leverage their unique large longitudinal dataset of tumor volumes to improve our knowledge about the mathematical law followed by the volume/diameter individual curves of patients under vorasidenib. Second, and whatever the choice of the mathematical law fitting the tumor size evolution, performing cohort-level analysis, while relevant for detecting a global difference between the two treatment arms, does not speak to the clinicians, nor to the patients. With this regard, it is much more informative to provide patient-level analysis. Such a complementary patient-level analysis is given in the Fig. 3 of Cloughesy & al., but only for best tumor volume change. Here, by estimating the slope of diameter evolution individually, our results reveals a more detailed pattern of vorasidenib effects, which cannot be captured by the average 6-months volumic tumor growth rates value of −1.3%, nor by the histograms of best volume change. Our individual-level analyses revealed that some patients (3/8 = 37.5%) were non-responders (unchanged tumor diameter slope), thus prompting to stop their treatment. Redo surgery in these three cases confirmed tumor recurrence. Some others (3/8 = 37.5%) demonstrated a stabilization. It is worth mentioning that the stabilization was defined at the patient-level, that is by a decrease in the diameter slope under vorasidenib compared to before treatment. Such a decrease suggests treatment efficacy and these three patients were still on vorasidenib at last follow-up. It would have been informative that Cloughesy & al. estimated the pre-treatment growth rates and performed pre- and post-treatment intra-patient comparisons. This is all the more important that growth rates constitute the strongest prognostic factor of these tumors [6, 7], hence a posteriori verification of balance between the two groups would have been welcome. Finally, two patients (2/8 = 25%) experienced an inversion of the diameter slope. Interestingly, slice-by-slice visual analysis of these 2 patients revealed spatial heterogeneity in evolution, with areas of FLAIR hypersignal regression and areas of extension (see the increase in the frontal operculum for patient # 3 & 5 on Fig. 5 and in the temporal operculum for patient # 3 on Fig. 3), raising questions about dissociated responses (existence of resistant clones) and/or the nature of the hyper-flair zones under vorasidenib treatement (inflammatory glial reaction?).Fig. 5Two illustrative cases of dissociated responses. The green arrows point to areas showing tumor shrinkage under vorasidenib, while the red arrows point to areas showing tumor enlargement Two illustrative cases of dissociated responses. The green arrows point to areas showing tumor shrinkage under vorasidenib, while the red arrows point to areas showing tumor enlargement The main limitation of the present work is the limited number of cases included in our series, precluding reliable generalization of the results. Another limitation comes from the fact that tumor volumes were segmented by a single rater, but this weakness was mitigated by the fact that for any given image, the volume is computed only once all slices are segmented, so that the rater cannot be biased during contouring. Moreover, although any FLAIR hypersignal was defined as tumoral, some areas might rather relate to postoperative gliosis. Thus, pre-treatment slopes might reflect active gliosis rather than oncogenic process. Under this hypothesis, the stabilization under vora would be interpreted as the spontaneous end of the gliotic scaring process, combined with a vora-induced inhibition of recurrence. Conclusion Complex statistical methodological framework, that necessarily comes with any rigorous trial, should not obscure the information given to patients, so that they can make an informed choice among the different therapeutic options offered to them. To this end, patient-level analysis of diameter evolution should be the method of choice to analyze vorasidenib effects on tumor kinetics, as previously demonstrated for standard chemotherapy [8, 9]. Moreover, thorough review of tumor topography revealed in responders some spatial heterogeneity of response, with areas of significant decreasing size contrasting with other areas of increasing size. More data are needed to unveil the significance of this spatial heterogeneity.