18 F-FDG PET standard uptake values of the normal pons in children: establishing a reference value for diffuse intrinsic pontine glioma

Background Positron emission tomography (PET) scanning with [18 F]fluorodeoxyglucose (18 F-FDG) is a useful diagnostic and prediction tool in brain tumors, but its value in childhood diffuse intrinsic pontine glioma (DIPG) is still unclear. For interpretation of 18 F-FDG PET results in DIPG, uptake values of the normal pons of children of increasing ages are mandatory. The aim of this study was to determine 18 F-FDG standard uptake value ratios (SUVr) of the normal pons and to compare these to those of DIPG. Methods We studied 36 subjects with a normal, non-affected pons (aged 5 to 23 years) and 6 patients with DIPG (aged 4 to 17 years) who underwent 18 F-FDG PET scanning. Magnetic resonance imaging (MRI) was co-registered to define the regions of interest. SUVr and SUVrmax for the pons/cerebellum (SUVrp/c) and the pons/occipital lobe (SUVrp/o) were calculated. Independent-samples t tests and Mann–Whitney U tests were used to compare the mean SUVr and Pearson’s test for correlations. Results For the normal pons, mean SUVrp/c and SUVrp/o were 0.65 (±0.054) and 0.51 (±0.056), respectively. No significant correlations were found between the SUVr of the normal pons and sex, age, nor pontine volume. A modest but statistically significant correlation was found between SUVr and post-injection time acquisition timing. For DIPG, mean SUVrp/c and SUVrp/o were 0.74 (±0.20) and 0.65 (±0.30), respectively, while mean SUVrp(max)/c and SUVrp(max)/o were 1.95 (±0.48) and 1.81 (±0.20), respectively. Conclusion The SUVr of the unaffected pons are strikingly constant between children, irrespective of sex and age, and can therefore be well used as a reference value for 18 F-FDG PET studies in DIPG.


Background
Positron emission tomography (PET) scanning with [ 18 F] fluorodeoxyglucose ( 18 F-FDG) provides information on glucose metabolism. 18 F-FDG PET positively correlates with an increasing WHO grade in astrocytomas [1]. In high-grade glioma (HGG), 18 F-FDG PET is an indicator of response to therapy and is used for PET-guided planning of stereotactic brain biopsy [2][3][4][5]. In the past few years, 18 F-FDG PET studies have been introduced in diffuse intrinsic pontine glioma (DIPG) [6][7][8][9][10], a fatal disease that almost exclusively occurs in children [11]. Interestingly, 18 F-FDG metabolism in the majority of the DIPG was lower than that in the non-affected occipital lobe, but increased 18 F-FDG uptake correlated with decreased overall survival [10]. However, reference values of 18 F-FDG uptake in the normal pons of children of increasing age are mandatory to know what increased uptake is in the pons, and these data are lacking. Therefore, the aim of this study was to calculate the standard uptake value ratios (SUVr) for the pons/ cerebellum (SUVr p/c ) and for the pons/occipital lobe (SUVr p/o ) in subjects with a normal pons and to investigate the influence of age, pontine size, and post-injection interval on the SUVr. The SUVr of the normal pons were then compared to the SUVr and SUVrmax of DIPG.

Subjects
To study the 18 F-FDG uptake of the normal pons, a retrospective cohort was used. Thirty-six children and adolescents aged 6 to 23 years who underwent 18 F-FDG PET scans for epilepsy surgery planning in the period of 2002 until 2012 were included. All controls had focal epilepsy and were in a non-ictal state at the moment of scanning. We inventoried the anti-epileptic agents used at the day of scanning. We excluded scans that revealed spaceoccupying lesions anywhere in the brain or epilepsyinduced changes in the pons, occipital lobe, and cerebellum and scans that did not meet the criteria as described under 'Scanning procedure'. The affected population consisted of six children with a newly diagnosed DIPG, based on criteria as described elsewhere from VU University Medical Center (VUmc), Amsterdam, the Netherlands, who underwent an 18 F-FDG PET scan at diagnosis [11]. The study was approved by the institutional review board of VUmc.

Scanning procedure
Scans of controls and DIPG patients were performed using an ECAT EXACT HR + PET scanner (Siemens/CTI, Knoxville, TN, USA), as previously described [12]. Patients and controls fasted for at least 4 h before the PET scan. Fifteen minutes before injection, they were positioned in a quiet, darkened room, with their eyes closed and no noise. After injection of 185 MBq 18 F-FDG (mean 187.2 MBq ±5.6), subjects remained in the quiet, darkened room for 35 min followed by a 10-min 2D transmission scan, acquired using retractable rotating 68 Ge sources, used for attenuation correction purposes. Approximately 45 min post-injection, a static 3D emission scan of 15 min was acquired. All emission scans were reconstructed using ordered subset expectation maximization (OSEM, 4 iterations, 16 subsets) with a Hanning filter with a cutoff at 0.5 times the Nyquist frequency and included the usual corrections for normalization, decay, dead time, attenuation, scatter, and randoms [13]. During reconstruction, a zoom factor of 2.123 and a matrix of 256 × 256 were used, resulting in voxel sizes of 1.2 × 1.2 × 2.4 mm 3 . All subjects underwent structural magnetic resonance imaging (MRI) T1-T2 for diagnostic purposes. PET characteristics are summarized in Table 1.

Image analysis
Each patient's T1-weighted MR image was co-registered to their 18 F-FDG PET using VINCI software (Max Planck Institute, Cologne, Germany) and subsequently used to manually define the regions of interest (ROIs) of the pons, occipital lobe, and cerebellum in normal subjects ( Figure 1). For DIPG, the ROI was defined as the hypointense pontine lesion on T1 MRI, independent of contrast enhancement. The ROIs were projected on the PET, and the mean uptake (becquerel per cubic centimeter) was calculated for the entire defined ROI. Next, the SUV ratios were calculated by dividing the activity (becquerel per cubic centimeter) of the pons by the reference regions. Control group reference

Baseline characteristics
Baseline characteristics are summarized in Table 1.  Figure S1). Figure 2 shows the SUV ratios of the normal pons.
Pontine SUV ratios in relation to pontine volume, sex, and age The average volume of the normal pons was 10 cm 3 (±1.4). The pontine volume linearly increased with age (regression coefficient 0.17, r = 0.51, p = 0.001; Figure 3a). There was no significant correlation between SUVr p/o (r = 0.18, p = 0.28; Figure 3b) and pontine volume nor SUVr p/c (r = −0.13, p = 0.45) and pontine volume. Furthermore, SUV ratios were found to be age independent, with r values of −0.17 (p = 0.324) and 0.18 (p = 0.305) for SUVr p/c and SUVr p/o (Figure 3c), respectively. We also found no significant difference between male and female subjects for SUVr p/c (p = 0.86) nor SUVr p/o (p = 0.98).

Pontine FDG SUV ratios as a function of post-injection uptake time
To determine whether uptake time influenced the 18 F-FDG uptake, we investigated the correlation between the SUV ratios and the post-injection uptake time in the control group (Figure 4). A modest positive correlation was found with both SUVr p/c (r = 0.37, p = 0.034) and SUVr p/o (r = 0.43, p = 0.012) and increasing postinjection time. The regression coefficients were small (0.0011/min and 0.0015/min, respectively).

F-FDG uptake in the normal pons versus DIPG
The average DIPG volume on MRI was 27 cm 3

Discussion
In an era where numerous drug trials in DIPG are ongoing or will be initiated shortly, it is essential to develop tools to predict disease evolution and to monitor response to therapy [14]. 18 F-FDG PET has the potential to be such a tool. However, the interpretation of 18 F-FDG PET results in DIPG is hampered by a lack of data on normal pontine glucose metabolism in children. We show in this study that 18 F-FDG SUV ratios of the normal pons versus those of the cerebellum and occipital lobe are very consistent in between controls, independent of sex, age, and pontine volume, and are therefore suitable as a reference value for 18 F-FDG PET studies in DIPG. Not only the pons of controls but also the pons infiltrated by tumor often showed lower 18 F-FDG uptake than the cerebellum and occipital lobe, a phenomenon that has been reported before [10]. Moreover, the mean SUVr of DIPG were not significantly higher than those of the normal pons, but this is probably due to the small DIPG sample size as the standard deviations were high. One may therefore question the role of 18 F-FDG PET in DIPG; however, the mean SUVrmax clearly increased in DIPG compared to the normal pons. Indeed, a recent study showed a significant correlation between increased 18 F-FDG tumor uptake and decreased survival in patients with this disease [10]. This correlation might be even stronger when considering that a SUVr p/o in DIPG between 0.5 and 1.0 already reflects increased 18 F-FDG uptake in comparison with the normal pons. This consideration is not taken into account in studies using semi-quantitative measurements that lead to classification as 'hypo/iso/hypermetabolic' compared to other brain areas [6][7][8][9][10]. An explanation for the limited 18 F-FDG uptake in DIPG compared to supratentorial HGG is that DIPGs are heterogeneous tumors with a mixed histologic tumor grade, as local uptake of the tracer is related to the presence of anaplastic features [11,15,16]. Calculating the SUVrmax, reflecting the highest local uptake in the tumor, is helpful in those tumors with heterogeneous 18 F-FDG uptake. Other explanations of the limited uptake are the frequently observed integrity of the blood-brain barrier in DIPG and the presence of white matter in the pontine region, which has low glucose metabolism [17].
We further investigated whether the time between injection and PET scanning had an influence on the 18 F-FDG uptake in the pons of controls compared to other brain areas. Indeed, SUVr p/c and SUVr p/o were positively correlated with increasing post-injection time. This suggests a delayed uptake of this tracer in the pons compared to the cerebellum and occipital lobe. However, the SUVr regression coefficients were small, and therefore, the influence of the uptake interval in clinical practice is negligible.
The main advantage of SUV ratios is that the possible errors in the measurement of weight or transcription and dose administered are minimized by the ratio between the two SUV measurements [18]. This applies especially for pediatric cancer, with low patient numbers and therefore often multi-national multi-center trials. In this study, we showed that SUV ratios of the normal pons are independent of sex, pontine volume, and age, although we had an under-representation of the youngest children (<5 years) in the control group. Although SUV ratios may give useful information in serial measurements, they have their limitations. In situations in which the 18 F-FDG uptake of the reference tissue varies, changes in SUV ratios can be misleading. For example, this may be the case when patients use steroids, which influence the glucose metabolism of the brain [19]. A methodological issue in this study was the use of epilepsy patients as controls, as 18 F-FDG PET data of healthy children could not be obtained due to ethical reasons regarding radiation exposure. We, however, do not expect significant changes in glucose metabolism of the pons due to epilepsy as all our subjects were in an inter-ictal state, which is not associated with changed glucose metabolism [20]. Furthermore, several anti-epileptic drugs including phenobarbital, phenytoin, benzodiazepines, and valproic acid have been associated with hypometabolism of the brain and especially the cerebellum and may therefore overestimate the SUVr p/c . Of these drugs, only valproic acid and clobazam were used in this study by, respectively, 3 and 4 out of 37 controls [21,22]. The lack of variance in between controls of both SUVr p/c and SUVr p/o presumes that the use of anti-epileptic drugs has not influenced our results significantly. In addition, the use of the cerebellum as a reference in epileptic patients in 18 F-FDG PET studies is not uncommon [23,24]. Future 18 F-FDG PET studies in DIPG may now compare SUVr and SUVrmax in DIPG to the here reported mean SUV ratios of the normal pons. By comparing SUV ratios to the normal pons, smaller increases in glucose metabolism can be detected in comparison with semi-quantitative measurements, as DIPGs often show lower glucose metabolism than the reference brain tissue (occipital lobe). In this way, the sensitivity and applicability of 18 F-FDG PET as a predictive and response monitoring tool for patients with DIPG can be increased.

Conclusion
We established a reference SUVr for 18 F-FDG uptake in the normal pons. SUV ratios are very consistent in between controls and independent of pontine volume, sex, or age. Not only was the 18 F-FDG uptake in the normal pons low compared to that in the reference brain areas, but also the uptake in DIPG was often lower than that in the occipital and cerebellar tissues. We encourage a study in controls to validate our results and propose that future 18 F-FDG PET trials in DIPG calculate SUV and SUV(max) ratios in order to relate these to the here reported mean SUV ratios of the normal pons. Smaller changes in the tumor's glucose metabolism can be detected in this way, which may have prognostic relevance for the patient.