Prognostic value of subventricular zone involvement in relation to tumor volumes defined by fused MRI and O-(2-[¹⁸F]fluoroethyl)-L-tyrosine (FET) PET imaging in glioblastoma multiforme

Glioblastoma multiforme (GBM) is the most aggressive malignant primary central nervous system tumor. While the majority of GBMs have similar pre-treatment magnetic resonance imaging (MRI) characteristics, subgroups exist with distinct clinical behaviors, genetic alterations, and outcomes. According to grading prognostic assessment (GPA) scoring, patients with newly diagnosed GBM qualifying for chemoradiotherapy have a two-year overall survival (OS) of between 5 and 35%, but 5–10% of GBM patients experience long-term survival [1]. Identifying prognostic groups who would benefit from different, personalized treatment remains challenging.

Age, Karnofsky performance status (KPS), and extent of surgery are all prognostic in GBM [2, 3], and more recently prognostic biomarkers have been described including O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation [4, 5], isocitrate dehydrogenase 1 or 2 gene mutations [6, 7], and subventricular zone (SVZ) involvement [8]. However, the relationship between pre-irradiation MRI contrast enhancement-based tumor volume and clinical outcome remains controversial [9, 10].

Therefore, accurately predicting tumor behavior in individual patients based on imaging parameters remains challenging, especially when molecular-genetic factors are not available. Imaging may be especially important given that mutations show intratumoral heterogeneity in non-operable, sub-totally operated, or MGMT promoter status-undefined patients [11‐13].

SVZ infiltration defined by MRI is known to be associated with treatment outcomes and progression and is thought to arise from neural stem cells [14, 15]. Extensive peritumoral edema on imaging may also be associated with survival [16, 17], since edema defined by MRI-T2 sequences may represent a mixture of neoplastic cells as well as vasogenic edema [18]. However, imaging parameters that more accurately define prognosis are still urgently needed to individualize treatment.

Positron-emission tomography/CT (PET) using O-(2-[¹⁸F]fluoroethyl)-L-tyrosine (FET) has been widely used for static and dynamic imaging in patients with brain tumors [19, 20]. Dynamic FET PET is helpful for defining aggression in WHO III astrocytomas [21] and low-grade gliomas (LGGs) [22, 23]. Moreover, WHO I-II gliomas show increased uptake kinetics compared to WHO III-IV high-grade gliomas (HGGs) [24, 25]. Dynamic acquisition more accurately differentiates LGGs from HGGs than standard static scans (20–40 min post-injection (p.i.)), mainly due to the characteristic high FET uptake in HGGs in the initial phase [26]. However, many institutions do not have routine access to dynamic PET imaging techniques. When dynamic PET cannot be performed, FET PET acquisition at a few selected time points may be a cost- and time-effective alternative as demonstrated using relatively early and very late time points (20–40 min p.i. and 70–90 min p.i.) [27]. However, experience with dtp FET and other amino acid PET tracers in patients with gliomas remains limited. Biological tumor volume defined by dtp FET PET correlates with progression site [28], but to the best of our knowledge, the prognostic impact of dtp FET PET parameters in GBM patients has yet to be determined. We hypothesized that dtp FET PET imaging in combination with SVZ infiltration would accurately select subgroups of patients with different chemoradiotherapy outcomes.

Here we show that pre-irradiation tumor volumes have a prognostic impact in GBM. Of the analyzed volumes, fused dtp FET PET for T1-Gad-based volume without the tumor bed was the most powerful predictor and may therefore be of value for radiation treatment planning. SVZ involvement, KPS performance status, and age but not the tumor-to-brain uptake ratios or FET kinetics measured by dtp PET/CT were prognostic in univariate analysis. However, the presence or absence of SVZ was associated with higher PET/MR tumor volumes, so this association was no longer significant in multivariate analysis.

Tumors are known to differ in shape and size when defined 10 and 60 min post FET injection and corresponded with the site of recurrence [28]. The current study shows that dtp FET PET parameters does not provides additional information as a prognostic imaging biomarker, although we note that FET PET performed after irradiation treatment response has previously been shown to be a marker of both PFS and OS [9].

Current PET tracers provide additional prognostic value in GBM [9, 35‐38]. However, amino acids and FET specifically are most commonly used for PET due to low uptake into inflammatory tissues, high stability, and longer half-life of 18F-FET [39].

Tumor volumes are vulnerable to the FET PET acquisition method. Tumor-to-brain ratios of 1.6 or greater determine the FET tumor volume and depend on the time of measurement, spatial resolution of the PET scans, and image processing [20]. The most commonly used method represents a summation of dynamic PET scans and a single static scan 20–40 min post FET injection. Pre-irradiation tumor volumes defined on static PET have been shown to be prognostic [9]. However, in high-grade gliomas, there can be increased tracer uptake at earlier time frames [26, 40], so tumor volume might be underestimated in the standard 20–40-min scan frequently taken in static PET-based radiation treatment planning [26]. However, in our group, 25% of patients had different uptake kinetics that may underestimate tumor volumes when categorized only on early (5–15 min p.i.) acquisition. The largest study to date on the topic reported a correlation between PET volumes and OS based on dynamic PET results [38]. However, dynamic FET PET is more time-consuming and costly, requiring 40–50 min of scanning time [20], which may be too long to patients to tolerate the thermoplastic mask.

Grosu et al. [41] reported that gross tumor volumes were not significantly different when measured by L-[methyl-¹¹C]-methionine (MET) PET or FET PET by static acquisition. However, FET PET-based volumes depend on the time of uptake measurement, which may have limited this comparison. Moreover, the different uptake kinetics (also known as the time-activity curve (TAC)) is a feature of FET not observed with MET [20].

Uptake kinetics have been shown to be prognostic in more aggressive low-grade [22] and WHO III [21] gliomas. Further, in a study of selected patients, uptake kinetics had an impact on prognosis [23]. We could not confirm this finding here, perhaps due to the smaller group size, non-selected cohort, or different method of PET acquisition.

With respect to the impact of SVZ invasion on GBM prognosis, our results are consistent with other studies [8, 14, 33]. However, to our knowledge, this is the first report that FET-PET-detected infiltration of the surrounding brain is larger in SVZ areas then in other locations. These large FET uptake volumes may, to some extent, explain worse outcomes for patients with SVZ invasion, which is typically explained in pre-clinical studies by the “neural stem cell niche” concept. Moreover, tumor not defined as infiltrating the SVZ on MRI may actually extend into this area when defined by dtp FET PET. Further, the frequency of SVZ involvement not shown on MRI but present in dtp FET PET is unknown. Here, this frequency was not significantly increased, but this was simply due to the larger volumes defined by dtp FET PET. It could be of value to target SVZ areas with modified or higher than routine radiation doses in future research studies.

The maximal safe resection of contrast-enhancing tumors is the mainstay of treatment for newly diagnosed GBM. Although extensively studied, the prognostic value of partial resection remains controversial, but the benefit of gross total resection in association with survival has been established [42]. This, however, was not seen in our study. Other factors such as MGMT methylation have a substantial impact on prognosis and, in combination with the small number of events, could impact on the lack of effect of gross total removal on survival. Here, 24 patients underwent gross total resection as defined by MRI; interestingly, all 35 patients had pathological uptake values in the surrounding tumor bed. This suggests that when the aim is to remove a contrast enhanced portion of a tumor, partial rather than gross tumor removal is the actual result. A recent large retrospective analysis showed that the additional removal of a significant portion of the FLAIR-abnormal region was associated with better survival [43]. Further, PET-based tumor removal may prolong survival in patients with high-grade gliomas [10]. Our study supports the concept that pre-radiation tumor volumes, especially when defined by PET rather than contrast-enhanced tumor removal, influences prognosis [9, 44]. The prognostic value of PET-based volume was also recently reported in a case of re-irradiation [45]. Nevertheless, performance status post-surgery remains the most important clinical marker of treatment outcome.

The dtp FET PET methodology used here and that published by Lohmann et al. [27] are different (20–40 p.i. and 70–90 p.i. in [27] and 5–15 p.i. and 50–60 p.i. here). The frequently recommended acquisition (20–40 p.i.) is not one of the two time points, which might have influenced the results (the threshold of 1.6 of the background has been validated on 20–40 p.i. static images). However, this might explain why 25% of GBMs showed decreased uptake (the maximal uptake could have been obtained in the period 20–40 min p.i., and thus missed by the choice of timepoints).

The main limitations of this study are the post hoc analysis of SVZ infiltration, manual measurement of tumor volumes that may influence results, and the relatively small number of patients. The lack of known MGMT promoter methylation status may also be regarded as limiting, but this was not standard care during recruitment. However, it has been shown that the FET PET volumes are independent of MGMT methylation status [38]. This study is also strengthened by its prospective nature, no pre-selection of patients, and confirmation of the exact date of death.

FET PET-based tumor volumes may be useful for predicting a worse prognosis in glioblastoma patients. Although the presence of SVZ infiltration is linked to higher PET/MRI-based tumor volumes, the independent value of dtp FET PET parameters and SVZ infiltration as prognostic markers pre-irradiation has not been confirmed.