Glioblastoma is the most common malignant brain tumor in adults. Standard treatment includes maximal safe resection followed by radiotherapy with concomitant and adjuvant temozolomide. Despite advances in oncological therapies, glioblastoma remains associated with tremendously poor prognosis – median overall survival (OS) ranges from 14 to 18 months and median progression-free survival (PFS) is approximately 10 to 12 months (1–4).

Early diagnosis of treatment failure to initiate salvage therapy remains controversial due to limited therapeutic options and the low accuracy of MRI in reliably determining the true extent of disease. Moreover, conventional MRI sequences often fail to distinguish post-treatment changes - such as radionecrosis and pseudoprogression - and actual tumor progression, which may lead to inappropriate treatment decisions.

Posttreatment changes on conventional MRI vary depending on the type of previous treatment. Radiotherapy-induced changes are typically located within areas that received the highest radiation dose – the most often near resection cavity or tumor bed. These post-radiotherapy changes may include Swiss-like and bubble-like patterns, oedema and mass effect. On T2-weighted images, radiation-induced alternations are characterized by low signal intensity with a centrally increased signal corresponding to necrosis. Additional findings may include enhancement of the white matter and cortex (5). Changes after treatment with bevacizumab with or without lomustine are commonly manifested by the reduction in contrast enhancement and oedema (6, 7). Response assessment criteria based on MRI findings of adult gliomas have been described in RANO 2.0 (8).

In recent years, PET imaging has gained prominence for evaluating treatment response (9–11). The most commonly used PET tracer in oncology is ¹⁸F-fluorodeoxyglucose ([¹⁸F] FGD). However, its diagnostic utility in brain tumors is limited due to high physiological glucose uptake in normal brain tissue (12, 13). Gliomas are characterized by the overexpression of L-amino acid transporters compared to normal brain cells (14). It enables the use of amino acid tracers such as ¹¹ C-methionine ([¹¹C] MET), ¹⁸ F-dihydroxyphenylalanine ([¹⁸ F] F-DOPA) and ¹⁸ F-fluoroethyl-L-tyrosine ([¹⁸ F] FET) in neuro-oncology.

These tracers have the unique ability to cross blood-brain barrier (BBB) and to visualize tumor extent beyond the areas of contrast enhancement seen on MRI or even beyond FLAIR (15, 16). Compared to conventional MRI, PET offers higher sensitivity and specificity for detecting neoplastic infiltration, superior assessment of metabolic response to treatment, greater accuracy in differentiating tumor progression from radiation-induced changes and response to treatment (9, 17, 18). The recently published ESTRO/EANO recommendations on reirradiation of glioblastoma endorse the addition of amino acid PET to assess recurrence after primary treatment (19). FET-PET demonstrates excellent sensitivity of 0.82 and specificity of 0.76 in the diagnosis of primary brain tumors (20). FET is characterized by high in vivo stability and easy synthesis (21). Its concentration is significantly increased in neoplastic lesions compared to healthy or infected tissues (22, 23). The high intracellular accumulation of FET is attributed to elevated LAT1–3 expression on glioma cells as well as its molecular weight of approximately 220 g/mol (24, 25). Another advantage of FET is its half-life of 110min, which makes is suitable for clinical use (26). FET uptake and kinetics in glioma are generally stable, but tumor-to-background ratio (TBR) might slightly decrease due to dexamethasone administration (25). General recommendations for the use of PET in glioma patients have been summarized by the PET/RANO working group and in EANM/EANO/RANO/SNMMI guidelines (27, 28). More recently, PET RANO 1.0 criteria have been introduced to standardize treatment response assessment based on PET imaging (29). In response to growing interest in amino acid PET tracers in neuro-oncology, PET RANO BM 1.0 criteria were also published to evaluate the metabolic response of brain metastases (30).

For radiation oncologists, accurate diagnosis of recurrence is particularly crucial and requires advanced imaging. Recently published recommendations on the use of FET-PET focus only on treatment assessment after first-line therapy and do not address other challenges such as post-treatment changes, selection criteria for retreatment and response assessment after re-irradiation. Therefore, we aimed to summarize the available evidence on the use of FET-PET in clinical scenarios following initial treatment: in patients with residual tumor, patients with enlarging lesions, patients with new, small areas of contrast enhancement and after retreatment.

The authors conducted a literature search of published journal articles written in English between January and May 2025 in PubMed. Keywords such as:” glioblastoma”, “primary brain tumors”, “FET”, “PET”, “recurrent glioblastoma”, “treatment-related changes”, “PET/RANO” and” RANO” were used. Retrospective studies, prospective studies and systematic reviews were analyzed. A workflow of the literature selection process is illustrated in Figure 1 .

The optimal management and prognosis of patients with residual, stable tumor on MRI after radiochemotherapy remain unclear. The association between prognostic factors such as OS and PFS and both static (e.g. TBR_mean/max – tumor-to-brain ratio; MTV- metabolic tumor volume; BTV - biological tumor volume) and dynamic (TTP – time to peak; TAC – time-activity curve) FET-PET parameters after glioblastoma treatment has been evaluated in five prospective studies.

In a study by Galldiks et al., 25 patients with glioblastoma were evaluated with MRI and FET-PET at different timepoints following standard treatment – after surgery, 7–10 days after radiochemotherapy and 6–8 weeks later. The results showed that a ≥ 10% decrease in both TBR_max and TBR_mean on FET-PET done 7–10 days after radiochemotherapy was a favorable prognostic factor for PFS (TBR_max 9.3 vs. 4.7 months; p= 0.002; TBR_mean 10.3 vs. 5.1 months p <0.001) and OS (TBR_max 15.4 vs. 8.5 months; p = 0.001; TBR_mean 16.1 vs. 9.3 months, p<0.001). FET-PET performed 6–8 weeks later had a lower predictive value for TBR_max/mean but a reduction in T_VOL1.6 (tumor volume with TBR > 1.6) was associated with improved PFS (9.3 vs. 5.1 months; p = 0.002). MRI-based tumor volume changes were not associated with survival (31).

Another study by Piroth et al. analyzed static and dynamic FET-PET parameters in 25 glioblastoma patients at the same timepoints. They found that a ≥ 10% decrease in TBR_max on FET-PET after radiochemotherapy (7–10 days) was associated with longer median PFS (9.3 vs 4.7 months; p =0.002) and OS (18.0 vs 8.5 months; p <0.01). Similarly, a ≥5% reduction in TBR_mean (cutoff 5%) was associated with longer PFS (10.3 vs 5.1 months) and OS (22.8 vs 9.3 months), with p < 0.001 for both. However, changes in TTP and the slope of the TAC were not associated with survival outcomes (32).

A study by Ceccon et al. included 41 glioma patients (90% diagnosed with WHO grade 4 glioblastoma) who received surgery, radiochemotherapy and adjuvant temozolomide. FET-PET performed after two cycles of adjuvant temozolomide showed that a reduction in TBR_max and MTV compared to baseline was associated with longer OS (24 vs. 12 months; p =0.032, and 29 vs. 12 months; p = 0.005) and PFS (both 11 vs. 8 months; p = 0.031 and 0.007, respectively). No significant correlations were observed between MRI findings and OS or PFS (33).

A prospective study by Suchorska et al. assessed both static and dynamic FET-PET in 79 patients with newly diagnosed glioblastoma treated with surgery, radiochemotherapy with temozolomide and adjuvant temozolomide. FET-PET imaging was performed before and after surgery, 4–6 weeks after radiochemotherapy and after four cycles of adjuvant temozolomide. A smaller BTV (below 9.5cm³, sensitivity 64%, specificity 70%) before radiochemotherapy was identified as a prognostic factor. Median OS and PFS for BTV below 9.5 cm³ were 17.5 and 8.8 months (p<0,002), respectively, compared to 10.7 and 3.9. months for BTV above 9.5cm³ (p <0.08). Additionally, patients with initially increased TACs had longer OS (29.7 vs 12.5 months; p <0.02, HR 2.1) and longer PFS (11.9 vs 5.8 months; p < 0.05, HR 1.8) (34).

Finally, a prospective study by Harat et al. evaluated the use of a simultaneous integrated boost planned on FET-PET in the postoperative treatment of 17 patients with newly diagnosed glioblstoma. The study demonstrated that FET-PET-based treatment response assessment enables accurate evaluation of biological response. In 81% of analyzed cases, there was discordance between PET and MRI-based RANO criteria for response assessment (35). In summary, FET-PET may be considered as a reliable tool in early treatment response evaluation and survival prediction. Abovementioned studies have been summarized in Table 1 .

After radiochemotherapy a progression of contrast enhancement on MRI might represent both tumor progression and radionecrosis.

Radionecrosis is a local tissue response to radiotherapy and typically develops within regions exposed to the highest dose of radiation. The risk of radionecrosis increases with higher total dose, larger fraction size and concurrent chemotherapy. Treatment-related necrosis can occur up to 30% of patients. In contrast to pseudoprogression, radionecrosis generally develops more than three months after treatment, although it can also manifest years or even decades later. On MRI images, radionecrosis can be characterized by contrast enhancement, which may be accompanied by clinical symptoms (34–36). Symptomatic radionecrosis can be managed with steroids, bevacizumab and surgery (36, 37). While biopsy can distinguish necrosis from recurrence, it carries surgical risks and might be limited due to sampling errors that do not represent the full pathological profile of the tumor (38–40). FET-PET might help to distinguish post-treatment necrotic tissue from tumor progression. However, most studies are retrospective and include relatively small patient cohorts with heterogenous histopathology.

A systematic review analyzing three studies on the use of FET-PET for the differential diagnosis of radionecrosis and glioma recurrence demonstrated higher diagnostic accuracy compared to FDG-PET (specificity 78–95% versus 70–88% for FDG and sensitivity 82–91% for FET 70–84% for FDG) (41).

The diagnostic accuracy, sensitivity and specificity of MRI were reported as 93.75%, 96%, and 85.7%, respectively. When both MR and FET-PET were evaluated, these values increased to 96.87%, 100%, and 85.7%, respectively. TBR_max cut-off (threshold) of 2.09 provided sensitivity of 100% and specificity of 72% for distinguishing recurrence from radionecrosis. A TBR_mean cut-off of 1.517 yielded a slightly lower specificity but higher sensitivity – 89% and 86% (42).

Another study showed that similar TBR_max cut-off 2.07 (measured 30–40 min. postinjection) resulted in sensitivity and specificity of 80% and 84.6%, respectively (43).

In contrast, a study by Vidmar reported higher value TBR_max cut-off - 3.03 (sensitivity and specificity of 77% and 82%) (44). The discrepancies in TBR tresholds might be attributed to small, heterogenous study populations, the lack of histopathological examination, and differences in PET protocols and interpretation.

A retrospective study by Bashir et al. involving 146 glioblastoma patients showed that FET-PET parameters measured at least 6 months post-treatment were significantly higher in recurrent glioblastoma than in post-treatment changes (TBR_max 3.2 vs 1.6; TBR_mean 2.0 vs 1.6; and BTV 14.8 cm³ vs. 0.01 cm³; p < 0.0001). Optimal thresholds were identified as 2.0 for TBR_max and 1.8 for TBR_mean and 0.55 cm³ for BTV to accurately differentiate tumor progression from treatment-related changes. Increasing TBR_max (HR 1.328, 95% CI: 1.116–1.582; p = 0.001) and BTV during follow-up (HR 1.303, 95% CI: 1.179–1.439; p < 0.0001) were associated with shorter OS (45).

Another retrospective study by Werner et al. evaluated 48 patients with high grade glioma and suspicious MRI findings. FET-PET demonstrated significantly superior diagnostic performance (threshold for both TBR_max and TBR_mean: 1.95; accuracy: 83%; p < 0.001) compared to apparent diffusion coefficient (ADC) values (threshold:1.09 × 10⁻³ mm²/s; accuracy, 69%; p = 0.13). The optimal TTP cut-off of 32.5 min was optimal for the differentiation (accuracy 72%; sensitivity 80%; specificity 69%; p< 0.01) and for slope - 0.32 SUV/h (change of SUV per hour) with an accuracy of 74%, sensitivity of 70% and specificity of 75% (p= 0.02). Combining static FET-PET parameters with ADC increased diagnostic accuracy to 89%, while combining both static and dynamic FET-PET parameters yielded the highest accuracy (93%). TBR_max and TBR_mean values <1.95 were associated with longer OS (p = 0.01) (46).

Celli et al. retrospectively analyzed 45 glioma patients (26 with WHO grade 4 glioblastoma) and found similar sensitivity (86.2%) and specificity (81.31%) for FET-PET in the differential diagnosis performed at least 12 weeks after radiotherapy. The optimal cut-offs for recurrence were consistent with previous findings (TBR_max ≥ 2.1, SUV_max ≥ 3.5, and TTP ≤ 29 min). However, no FET-PET parameters significantly impacted overall survival (47).

Maurer et al. conducted a retrospective study of 147 glioma patients (67 with WHO grade 4 glioblastoma). The diagnostic performance of TBR_max and TBR_mean thresholds of 1.95 was slightly lower (sensitivity 70%; specificity 71%; accuracy 70%; for TBRmax and sensitivity 56%; specificity 79%; accuracy 62%; for TBRmean). The optimal cut-off for slope was lower – 0.2SUV/h (sensitivity 54%; specificity 86%; accuracy 63%);. Nonetheless, TBR_max > 1.95 combined with a slope < 0.2 SUV/h achieved a sensitivity of 86%, a specificity of 67%, and an accuracy of 81% in detecting progression (48).

In summary, static (TBR_max/mean < 2) and dynamic (TTP >30min) FET-PET parameters can differentiate necrosis from tumor progression. Diagnostic accuracy can be further improved by combining FET-PET with MRI-derived parameters, such as ADC. However, these improvements do not appear to translate into significantly longer OS.

Management of asymptomatic patients with new, small, multicentric lesions remains challenging, as such MRI findings might indicate either tumor progression or pseudoprogression.

The pathogenesis of pseudoprogression remains unclear, and no universally accepted definition currently exists. However, it is often described as at least 25% increase in tumor size according to Macdonald criteria, followed by either partial response or stable disease lasting for at least six months after radiochemotherapy (49).

Pseudoprogression is more commonly observed in patients with MGMT promoter methylation and in those treated with radiochemotherapy including temozolomide, compared to radiotherapy alone (50, 51).

Similar to necrosis, pseudoprogression can mimic tumor progression on MRI. Histopathological examination may reveal residual, stable tumor tissue within areas of pseudoprogession and necrosis, potentially leading to a wrong diagnosis of progression (39, 52–55).

Studies of FET-PET in patients with suspected MRI findings showed promising results.

A retrospective study by Galldiks et al. included 22 glioblastoma patients with suspicious MRI findings within the first 12 weeks after radiochemotherapy. In patients with pseudoprogression, FET uptake was significantly lower than in those with true progression (TBR_max 1.9 ± 0.4 vs. 2.8 ± 0.5, TBR_mean 1.8 ± 0.2 vs. 2.3 ± 0.3; p<0.001). TTP was shorter in progression than in pseudoprogression (mean TTP 26 ± 10 vs. 35 ± 9 min, P = 0.05). TACs type II (uptake peaking at a mid-point; >20–40 min) or III (early uptake peak ≤20 min followed by a constant decline) were more frequently observed in progression (p=0.04). The optimal TBR_max cut-off for identifying pseudoprogression was 2.3 (sensitivity 100%, specificity 91%, accuracy 96%, p<0.001) (56).

Another study by Kebir et al. reported comparable results in 26 glioblastoma patients with MRI changes observed at least 12 weeks post-radiotherapy. TBR_max and TBR_mean were significantly higher in patients with progression compared to those with late pseudoprogression (TBR_max 2.4 ± 0.1 vs. 1.5± 0.2, P =0.003; TBR_mean 2.1 ± 0.1 vs. 1.5 ± 0.2, p =0.012). The optimal cut-off for both parameters was 1.9 (TBR_max: sensitivity 84%, specificity 86%, accuracy 85%, p =0.015; TBR_mean: sensitivity 74%, specificity 86%, accuracy 77%, p= 0.023). The authors suggested to diagnosing late progression when TBR_max exceeds 2.4 and late pseudoprogression when it is below 1.0; values between 1.0 and 2.4 should be interpreted with caution. TTP was significantly shorter in progression (mean TTP 25 ± 2 vs. 40 ± 2 min, p< 0.001). TACs type II or III were more frequently observed in progression, with a sensitivity of 84%, specificity of 100%, and an accuracy of 89% (p< 0.001) (57).

A retrospective study by Werner et al. analyzed 23 glioblastoma patients with progressive MRI findings after chemoradiation with lomustine and temozolomide. FET-PET were consistent with previous studies and significantly contributed to diagnosing pseudoprogression. Lower TBR_mean/max values were observed in pseudoprogression compared to progression (TBR_mean 1.9 ± 0.2 vs. 2.1 ± 0.2; P = 0.023 and TBR_max 2.8 ± 0.6 vs. 3.2 ± 0.5; P = 0.045), while TTP was higher in pseudoprogression (36.6 ± 8.3 vs. 24.8 ± 9.4 minutes; p = 0.005). The optimal cut-off values were: TBR_mean 1.95 (sensitivity 82%; specificity 92%; accuracy 87%; P = 0.029), TBR_max 2.85 (sensitivity 64%; specificity 92%; accuracy 78%; P = 0.046) and TTP 35 minutes (sensitivity, 64%; specificity, 83%; accuracy 74 P = 0.010). Combining TBR_mean with TTP improved the specificity and positive predictive value to 100% for detecting pseudoprogression (sensitivity 55%; accuracy 78%; P = 0.005) (58).

In contrast, a study by Mihovilovic et al. reported higher FET-PET parameters in 36 glioblastoma patients with suspected recurrence. The optimal tresholds for differentiating progression from late pseudoprogression were: TBR_max of 3.52 (sensitivity 89%, specificity 75%, area under the curve (AUC) 0.87 ± 0.07; p=0.0020) and TBR_mean of 2.98 (sensitivity 82%, specificity 87.5%, AUC 0.84 ± 0.08, p=0.004) (59).

In conclusion, current evidence supports the utility of FET-PET in distinguishing progression from pseudoprogression. Static and dynamic parameters appear similar in necrosis and pseudoprogression. However, the benefit of FET-PET lies in identifying patients requiring immediate therapeutic intervention due to progression, rather than in distinguishing between begin post-treatment changes – such as pseudoprogression and radionecrosis - that can be monitored. Some studies did not clearly differentiate between necrosis and pseudoprogression, focusing instead on distinguishing recurrence from general treatment-related changes. A retrospective study by Lohmeier et al. involving 42 patients with low- and high-grade gliomas found that a TBRmax threshold of 2 provided a sensitivity of 81% and specificity of 60% in differentiating glioma recurrence from post-treatment changes (60). In a retrospective study by Puranik et al., TBR_max cutoff value of 2.5 in patients with grade 3 and 4 gliomas yielded a sensitivity of 91.6% and a specificity of 76.9% in distinguishing tumor recurrence from post-treatment changes (61). Verger et al., in a study of 31 low- and high-grade glioma patients reported that a TBR_max of 2.61 yielded a sensitivity of 80%, and a specificity of 86% in distinguishing progression from treatment-related changes (62).

These varying results may stem from methological differences between centers, including how background activity is defined (63). A retrospective study comparing five methods for differentiating pseudoprogression from true progression more than 12 weeks post-radiotherapy found similar diagnostic performance across approaches, with AUCs ranging from 0.80 to 0.88 (64). Another study demonstrated variability in background activity measurement depending on the region-of-interest method (2D-ROI vs. VOI 3cm diameter vs. crescent-shaped VOI), with the crescent-shaped VOI yielding the most consistent results (65). Most of these studies were conducted before the establishment of the PET RANO 1.0 criteria or EANM/EANO/RANO/SNMMI guidelines, which recommended using a crescent-shaped VOI in the contralateral frontal lobe including healthy white and gray matter for background activity assessment (28, 29).

Most patients with glioblastoma experience tumor recurrence shortly after initial treatment. However, there is scarce evidence regarding the use of FET-PET in the context of reirradiation.

A systematic review demonstrated that amino acid PET, including FET-PET, has superior prognostic value compared to MRI using RANO criteria in predicting OS in patients treated with bevacizumab for recurrent glioma. FET-PET predicted 9-month OS with a sensitivity of 76% (95% CI 60–87) and specificity of 71% (95% CI 53–83), whereas MRI yielded a sensitivity of 32% (95% CI 19–48) and specificity of 82% (95% CI 66–92) (66).

Several studies have showed that changes in static and dynamic FET-PET parameters can predict prognosis. However, the results are inconsistent. This variability is likely due to small patient populations and retrospective nature of most studies.

In a retrospective study by Fleischmann et al., 72 patients with recurrent high-grade glioma underwent FET-PET prior to re-irradiation with or without bevacizumab (at least 6 months after the initial radiotherapy). A shorter TTP_min prior to treatment was associated with shorter post-recurrence survival (PRS): 6 months for TTP_min <12,5min, 7 months for TTP_min 12.5–25 min and 11 months for TTP_min >25 min (p=0.027). Early TBR_max was not predictive of PRS (67).

Another retrospective study by Niyazi et al. included 56 patients with recurrent malignant glioma who underwent FET-PET before re-irradiation with or without bevacizumab. The study reported a significant decrease in median TBR_max (3.3 vs 2.6, p <0.001) and BTV (13.7 cc vs 7.3 cc, p = 0.006) after re-irradiation, but without significant impact on PFS or survival. TBR_mean values also did not correlate with survival. However, patients with pretherapeutic decreasing FET kinetics had worse survival compared to those with other kinetics (p = 0.01) (68).

A phase I clinical trial evaluated the prognostic value of FET-PET in 31 patients undergoing re-irradiation for high grade glioma. Baseline BTV and MRI volume were both prognostic for OS (HR = 1.3 p < 0.01 and HR = 1.3 p < 0.01, respectively). However, changes in BTV and TBR_max were not correlated with survival (69).

In a prospective study by Galldiks et al., 21 glioblastoma patients with recurrent glioblastoma undergoing bevacizumad with lomustine therapy (without re-irradiation) were monitored using FET-PET. A reduction in TBRmax > 27% and TBR _mean > 17% predicted improved OS > 9 months (sensitivity 92%, specificity 63% for both; p = 0.036 and p=0.02, respectively) Additionally, an absolute MTV below 5 ml at follow-up was associated with improved OS (12 vs. 6 months, sensitivity 85%; specificity, 88%; p < 0.001). MRI-based response assessment did not predict OS (7).

The role of FET-PET in the assessing response to re-irradiation remains limited due to the lack of significant correlation reported in current studies.

In a phase I trial, Moller et al. used FET-PET to monitor results of stereotactic radiotherapy for recurrent glioblastoma. While pre-treatment TBR_max and BTV were associated with better outcomes, changes in this parameter did not predict survival. Similar findings were reported by Niyazi in patients undergoing re-irradiation with bevacizumab. These findings indicate the need for further studies. Recent data suggest that early-phase FET-PET imaging may reflect the most aggressive tumor regions (70, 71). High uptake in early acquisition is more commonly observed in IDH-wildtype gliomas, and longer TTP appears to have a favorable prognostic impact (72). A retrospective study in patients treated with TTFields (tumor treating fields) in recurrent tumors demonstrated the value FET-PET monitoring. Increased uptake was observed in time of further progression compared to baseline (TBR_max 3.5 ± 0.6, range 2.5–4.4; TBR_mean 2.7 ± 0.7, range 2.0–4.0). In patients treated with TTFields as a maintenance therapy without progression, FET-PET showed either reduced metabolic activity or no increased uptake at follow-up (73).

Abovementioned studies have been summarized in Table 2 .

Growing evidence supports the efficacy of FET-PET in accurate selection for reirradiation. Based on current data, FET-PET can reliably distinguish post-treatment changes from true tumor progression and may enhance target definition by identifying areas of tumor infiltration beyond contrast enhancement on MRI. Most studies suggest that TBR_mean and TBR_max above 2.0 should be considered indicative of progression or metabolically active disease following primary radiotherapy. In the PET RANO classification, stable disease corresponds to stable uptake after treatment. However, it may still reflect a metabolically active tumor. Careful evaluation is crucial, and optimal management for this subgroup remains unclear. Future research should explore whether additional therapies for metabolically stable glioblastoma could lead to improved outcomes. The utility of FET PET in monitoring patients after re-irradiation also warrants further investigation.