Evaluation of isocitrate dehydrogenase mutation in 2021 world health organization classification grade 3 and 4 glioma adult-type diffuse gliomas with 18F-fluoromisonidazole PET

Purpose This study aimed to investigate the uptake characteristics of 18F-fluoromisonidazole (FMISO), in mutant-type isocitrate dehydrogenase (IDH-mutant, grade 3 and 4) and wild-type IDH (IDH-wildtype, grade 4) 2021 WHO classification adult-type diffuse gliomas. Materials and methods Patients with grade 3 and 4 adult-type diffuse gliomas (n = 35) were included in this prospective study. After registering 18F-FMISO PET and MR images, standardized uptake value (SUV) and apparent diffusion coefficient (ADC) were evaluated in hyperintense areas on fluid-attenuated inversion recovery (FLAIR) imaging (HIA), and in contrast-enhanced tumors (CET) by manually placing 3D volumes of interest. Relative SUVmax (rSUVmax) and SUVmean (rSUVmean), 10th percentile of ADC (ADC10pct), mean ADC (ADCmean) were measured in HIA and CET, respectively. Results rSUVmean in HIA and rSUVmean in CET were significantly higher in IDH-wildtype than in IDH-mutant (P = 0.0496 and 0.03, respectively). The combination of FMISO rSUVmean in HIA and ADC10pct in CET, that of rSUVmax and ADC10pct in CET, that of rSUVmean in HIA and ADCmean in CET, were able to differentiate IDH-mutant from IDH-wildtype (AUC 0.80). When confined to astrocytic tumors except for oligodendroglioma, rSUVmax, rSUVmean in HIA and rSUVmean in CET were higher for IDH-wildtype than for IDH-mutant, but not significantly (P = 0.23, 0.13 and 0.14, respectively). The combination of FMISO rSUVmean in HIA and ADC10pct in CET was able to differentiate IDH-mutant (AUC 0.81). Conclusion PET using 18F-FMISO and ADC might provide a valuable tool for differentiating between IDH mutation status of 2021 WHO classification grade 3 and 4 adult-type diffuse gliomas. Supplementary Information The online version contains supplementary material available at 10.1007/s11604-023-01450-x.


Introduction
The radiotracer 18 F-fluoromisonidazole (FMISO) accumulates in hypoxic viable cells after reduction reactions in the absence of oxygen.PET using 18 F-FMISO allows the detection of hypoxia associated with the rapid depletion of nutrients that occurs with the abnormal proliferation of tumor cells seen in glioma [1,2].Hypoxia is associated with resistance to radiotherapy and chemotherapy in gliomas, and is related to the outcomes of glioma therapies [3].Despite the importance of clarifying the extent of hypoxia in gliomas, common imaging modalities cannot clearly identify hypoxia in gliomas.
Isocitrate dehydrogenase (IDH) mutation is known to affect the prognosis of patients with glioma [4][5][6], and the knowledge of IDH mutation has been incorporated into 2021 WHO classification of brain tumors [7].Prediction of IDH mutation by imaging would facilitate the optimization of therapeutic strategies for gliomas.Previous reports have demonstrated that 2-hydroxyglutarate (2-HG), which accumulates in IDH-mutated gliomas, can be detected by magnetic resonance spectroscopy [8].On the other hand, PET probes have been reported to potentially allow prediction of IDH mutation status.Recent studies have found significant associations between 18 F-fluoro-ethyl-tyrosine ( 18 F-FET) PET results and IDH mutation status [9][10][11].Another recent paper investigated the association of 3'-deoxy-3'- 18 F-fluorothymidine ( 18 F-FLT) PET and 18 F-FMISO PET, as well as relative cerebral blood volume in 31 patients with glioblastoma [12].
As 18 F-FMISO PET is known to be useful in differentiating glioma grades [13], we hypothesized that there might be some association between IDH mutation status and 18 F-FMISO uptake in glioma.The present study aimed to investigate the characteristics of 18 F-FMISO uptake by 2021 WHO classification grade 3 and 4 glioma in terms of IDH mutation status.

Patients
The institutional ethics committee approved this prospective study.Patients who were suspected intracranial brain lesions were enrolled in this study between September 2015 and March 2018, and written informed consent was obtained from each patient.In cases where the patient could not provide a signature, another family member provided informed consent instead.Tumors were included or excluded according to 2021 WHO classifications [7]. Figure 1 shows the inclusion and exclusion criteria.Note that patients who were histopathologically diagnosed with other brain tumors were excluded from our study (n = 6) (*).Five of them were histopathologically diagnosed as schwannoma 2, metastatic tumor 1, central neurocytoma 1, ependymoma 1.One patient was clinically diagnosed as pilocytic astrocytoma without surgery at the time of PET scan, but was recently operated due to tumor volume increase, and histopathologically diagnosed as pilocytic astrocytoma because the specimens were not in suitably good condition, so those four patients were excluded from the study (n = 4).IDH testing was performed with an immunohistochemistry assay.

PET protocol
Static PET images of the brain were acquired 4 h after intravenous injection of 350-550 MBq of 18

Definition of volume of interest (VOI)
Two VOIs were manually placed by a board-certified radiologist with 8 years of experience in neuroradiology, using ITK-SNAP software (https:// www.itksn ap.org) [14] and approved by another board-certified radiologist with 22 years of experience in neuroradiology.(a) Hyperintense areas on FLAIR imaging (HIA) were defined as areas of hyperintensity around and inside the tumor on FLAIR imaging.(b) Contrast-enhanced tumors (CET) were defined as areas of tumor enhancement on CE T1WI.Areas of central hypointensity on CE T1WI were considered to represent regions of central necrosis and were excluded from among VOIs of CET.Hyperintense areas on NE T1WI were considered to represent hemorrhagic lesions and were removed from among the VOIs of CET.Representative VOIs are shown in Fig. 2. Cerebellar cortical VOIs were created for reference using segmented cerebellar cortices.

Data analysis
Relative standardized uptake value (SUV) of 18 F-FMISO PET images was calculated as follows: SUV max , SUV mean in HIA, and SUV mean in CET were divided by the SUV mean of cerebellar cortical VOIs, respectively.This resulted in rSUV max and rSUV mean in HIA, and rSUV mean in CET.
We measured ADC 10pct in HIA, ADC mean in HIA, ADC 10pct in CET, and ADC mean in CET using the ADC map of MRI images [15,16].MANGO software (Research Imaging Institute, UTHSCSA, http:// ric.uthsc sa.edu/ mango/) was used for these measurements.
We performed the following analyses to determine IDH mutation status of glioma.

IDH-mutant vs. IDH-wildtype
We compared rSUV max and rSUV mean in HIA, rSUV mean in CET, ADC 10pct and ADC mean in HIA, and ADC 10pct and ADC mean in CET between IDH-mutant and IDH-wildtype in terms of following: A) comparisons between IDH-mutant and IDH-wildtype in all patients (n = 35); B) in patients in astrocytic tumors (n = 31).

Statistical analysis
We applied the Mann-Whitney U test for measured values that did not follow a normal distribution.
We also performed logistic regression analysis with the above-mentioned rSUV and ADC, followed by receiver operating characteristic (ROC) curve analysis.Areas under the curve (AUCs) were calculated with optimal cutoff, sensitivity, and specificity in terms of following: (A) determination of IDH mutation status in all patients; (B) determination of IDH mutation status in patients in astrocytic gliomas; AUCs were compared with DeLong test.
All statistical analyses were performed using JMP version 15 software (SAS Institute, Cary, NC, United States).Values of P < 0.05 were considered significant.

Patients
A total of 35 patients were included in this study.The demographic characteristics of patients are shown in Table 1 and Fig. 1. rSUV and ADC in IDH-mutant and IDH-wildtype are shown in Table 2.No registration error was observed.Representative images are shown in Fig. 3.

Comparisons between IDH-mutant and IDH-wildtype in all patients (n = 35)
rSUV mean in HIA and rSUV mean in CET were significantly higher for IDH-wildtype than for IDH-mutant (P = 0.0496 and 0.03, respectively) (Fig. 4).rSUV max were higher for  IDH-wildtype than for IDH-mutant but not significantly (P = 0.06).ADC 10pct in HIA and ADC mean in HIA were lower for IDH-wildtype than for IDH-mutant but not significantly (P = 0.24 and 0.16, respectively).No significant differences were found in ADC 10pct in CET or ADC mean in CET (P = 0.98 and 0.54, respectively).

ROC curve analysis
ROC curve analysis was performed using logistic regression analysis with rSUV and ADC to determine glioma IDH mutations status in all patients (n = 35) (Fig. 6a) ) in all patients.The rSUV mean in HIA (b), and rSUV mean in CET (c) are significantly higher for IDH-wildtype than for IDHmutant.The rSUV max (a) tend to be higher for IDH-wildtype than for IDH-mutant, but not significantly.The ADC 10pct in HIA (d) and ADC mean in HIA (f) tend to be lower for IDH-wildtype than for IDHmutant, but not significantly.No significant differences were found in ADC 10pct in CET (e) and ADC mean in CET (g).Asterisks (*) represent statistically significant differences Fig. 5 Comparison between IDH-mutant (n = 9) and IDH-wildtype (n = 22) in astrocytic tumors.The rSUV max (a), rSUV mean in HIA (b) and rSUV mean in CET (c) were higher for IDH-wildtype than for IDHmutant, but not significantly.The ADC 10pct in HIA (d) and ADC mean in HIA (f) were lower for IDH-wildtype than for IDH-mutant but not significantly.No significant differences were found in ADC 10pct in CET (e) or ADC mean in CET (g) and in astrocytic tumors (n = 31) (Fig. 6b).AUCs of all parameters are shown in Supplemental Tables 1 and 2.

Discussion
We were able to differentiate IDH mutation status using rSUV mean in HIA and ADC 10pct in CET, with AUC of 0.80 in all patients, and AUC of 0.81 in the patients with astrocytic tumors, according to 2021 WHO classification.Obviously, FMISO does not directly reflect IDH mutation status.A recent study revealed that multidrug-resistant protein 1 (MRP1) inhibitors increase 18 F-FMISO accumulation in hypoxic cells.This suggests that 18 F-FMISO PET imaging is affected by MRP1 inhibitors independent of the state of hypoxia [17].FMISO uptake is known to reflect hypoxic circumstances in tissues, and increased tumor aggressiveness induces greater hypoxia inside the tumor [18][19][20][21].IDH mutation is considered to offer a strong predictor of less-aggressive glioma [22][23][24], and FMISO would thus indirectly reflect IDH mutation status.
In the present study, no significant difference in ADC was apparent between IDH-wildtype and IDH-mutant.However, diagnostic performance increased after combining ADC with SUV.Our study demonstrated that rSUV mean in HIA and ADC 10pct in CET can differentiate IDH mutation status with high diagnostic ability in grade 3, 4 gliomas by 2021 WHO classification (AUC = 0.80, in all patients; AUC = 0.81, in astrocytic tumors, respectively).In the literature, detection of IDH mutation status was possible in grade 2 and 3 glioma of 2016 WHO classification, using the ratio of ADC mean to ADC of normalappearing white matter (AUC 0.83) [25], ADC ratio (AUC 0.95) [26], minimum ADC (ADC min ) (AUC 0.87), and relative ADC min (AUC 0.84) [27].Meanwhile, in grade 3, 4 gliomas, IDH mutation status was able to be differentiated by ADC mean (AUC 0.71), ADC 10pct (AUC 0.71) from histogram study [28], and ratio of ADC min to normal white matter (AUC = 0.70) in grade 4 glioma [29].Previous study also showed that, FMISO tumor-blood SUV ratio (TBR) could differentiate IDH-mutant type from IDH-wildtype in grade 3, 4 gliomas (AUC = 0.78 in all patients, AUC = 0.76 in astrocytic tumors) [30].Our results were comparable to the those of previous studies.
IDH mutation status has some association with other PET tracers.The ratios of the SUV max of the tumors to the SUV mean of the contralateral cortex (T/N ratios) of FLT-PET/CT can be used to determine the IDH mutation status with an AUC of 0.911; The T/N ratios of 11 C-methionine (MET) methionine can be used to determine the IDH mutation status with an AUC of 0.727 [31].Based on 2016 WHO classification, the differences in mean 18 F-FLT tumor-normal tissue ratio (TNR) and 18 F-FMISO TBR were significant between GBM and other glioma subtypes (P < 0.001); and regarding the comparison between Gd-T1WI volumes and 18 F-FLT MTVs or 18 F-FMISO MTVs, previous study identified significant differences between IDH-wildtype and IDHmutant or 1p19q-codeletion (P < 0.01) [32].The percentage difference between the standard biological tumor volume (BTV) on standard summation images and BTV on early summation images could differentiate IDH mutation status with an AUC of 0.83 using 18 F-FET PET [10].Time-to-peak value in a dynamic 18 F-FET PET study also showed good diagnostic performance for IDH mutation status in gliomas (AUC 0.75) [33].While 3D-VOIs were manually created for HIA and CET in our study, as in a previous study [34], 2-dimensional regions of interest (2D-ROI) were used in most studies examining 18 F-FMISO PET.Our study registered images from PET and MRI, and 3D-segmented cerebellar cortices were also used as references to calculate relative SUV, although 2D-ROIs were used as reference for 18 F-FMISO uptakes in previous articles [18,34,35].3D-VOIs for HIA may underestimate rSUV mean of 18 F-FMISO uptake because 3D-VOIs for HIA are larger than those for CET.However, rSUV max can compensate such underestimation in HIA, since physiological 18 F-FMISO uptake is low and homogeneous in the brain parenchyma compared with other amino acid PET tracers.
Some limitations need to be acknowledged in this study.First, the number of patients enrolled was small (n = 35).More patients need to be enrolled to confirm the present results.Second, biodistribution of 18 F-FMISO has not been evaluated in this study. 18F-FMISO is relative lipophilic and diffuses through cell membranes, and mild uptake of 18 F-FMISO is seen in normal tissue [36].In addition, the previous study had performed dynamic 18 F-FMISO and dynamic 15 O-H 2 O PET in brain tumors to measure tumor hypoxia and perfusion, and increased 18 F-FMISO tumor retention at late scan time was found predominantly in glioblastoma, but not found in meningiomas, which lacks the blood brain barrier (BBB) [35].Their data suggested that late 18 F-FMISO PET images obtained 4 h after the injection provide a spatial description of hypoxia in brain tumors that is independent of BBB disruption and tumor perfusion.Third, dynamic susceptibility contrast perfusion weighted imaging (DSC-PWI), which is beneficial in differentiation IDH-wildtype glioma and IDH-mutant glioma, was not used in this study.IDH mutation status of grade 2 and 3 gliomas of 2016 WHO classification can be differentiated by relative CBV max with an AUC of 0.82 [27].IDH mutation status in glioblastoma could be differentiated by rCBV mean with AUC of 0.886 [37].IDH mutation status could be differentiated using Visually AcceSAble Rembrandt Images (VASARI) MRI feature set in grade 2 and 3 gliomas, grade 2 glioma only, and grade 3 glioma only with AUCs of 0.78, 0.83 and 0.87, respectively [38].

Conclusions
In conclusion, PET using 18 F-FMISO and ADC might provide a valuable tool for differentiating IDH mutation status of 2021 WHO classification grade 3 and 4 adult-type diffuse glioma.
need to obtain permission from the copyright holder.To view a copy of this licence, visit http:// creat iveco mmons.org/ licen ses/ by/4.0/.

Fig. 4
Fig. 4 Comparison between IDH-mutant (n = 13) and IDH-wildtype (n = 22) in all patients.The rSUV mean in HIA (b), and rSUV mean in CET (c) are significantly higher for IDH-wildtype than for IDHmutant.The rSUV max (a) tend to be higher for IDH-wildtype than for IDH-mutant, but not significantly.The ADC 10pct in HIA (d) and

Fig. 6
Fig. 6 ROC analysis was performed for 2 groups.The highest three AUCs were shown for each group.a Prediction of IDH mutation status in all patients (n = 35): Model 1A, rSUV mean in HIA and ADC 10pct in CET (AUC, 0.80); Model 1B, rSUV max and ADC 10pct in CET (AUC, 0.80); Model 1C, rSUV mean in HIA and ADC mean in CET

Table 1
Demographic characteristics of patients