The results of this study showed that DSC (Gd, T/N 2.0) of 0.77 and JI (Gd, T/N 2.0) of 0.7 were the highest among the derived CTVs. Thus, the overlap with the Gd-enhanced region when using T/N 2.0 in 11C-MET PET was larger than that using other threshold values. However, the DSC and JI values of 0.77 and 0.7 are still not sufficiently high to indicate a good match, and some patients have values of < 0.5 (DSC: 3 cases, JI: 7 cases), which indicates a weak correlation between the Gd-enhanced and 11C-MET accumulation regions. In mapping of correlation coefficients and p-values of JI (Gd, T/N 2.0), there were 3 cases with higher JI and 9 cases with lower JI with r < 0.3, but 7 cases with higher JI and 1 case with lower JI with r > 0.3. Cases with high agreement between Gd-enhanced MRI and 11C-MET PET had a high correlation of pixel values, which indicates that the Gd-enhanced and 11C-MET regions are similar. However, this trend was absent in about half of the cases, which suggests that 11C-MET PET and Gd-enhanced MRI do not necessarily reflect similar phenomena.
On the other hand, the mean value of DSC using T2 was lower than 0.4 and JI was lower than 0.3 as shown in Fig. 1-(C, D). Mapping of correlation coefficients with 11C-MET accumulation showed that they are not as high in the T2 signal as in the Gd distribution. For example, correlation coefficients were < 0.3 in 8/10 patients with a lower JI, and > 0.3 in only 1/10 with a higher JI. Based on these results, we consider that the association between T2-weighted imaging and 11C-MET PET is considerably lower than with Gd-contrast imaging.
In this study, the treatment plan was for recurrent brain tumors and all patients had a history of surgery and RT. In these cases, it is particularly important to reduce the dose to normal brain tissue to avoid further brain damage. Radiation-induced brain damage includes brain dysfunction and brain necrosis as late adverse events, and damage of the brain is closely associated with the dose and volume of irradiation. For example, Milano et al. found that V12Gy (volume of brain tissue receiving 12 Gy) of 5, 10, and 15 cm3 or greater correlated with a 10, 15, and 20% risk of symptomatic brain necrosis in single stereotactic radiosurgery (SRS) [30]. Sayan et al. showed that the risk of cerebral necrosis in SRS increased significantly with increasing V100%, V50%, V12Gy, and V10Gy, indicating that dose-volume considerations are important [31]. In re-irradiation, the risk of brain damage is of particular concern. Chang et al. found a higher risk of reduced learning and memory function at 4 months with SRS plus whole brain RT, compared to SRS alone [32], and Trifiletti et al. reported that in SRS for brainstem metastases, the probability of neurotoxicity is higher in patients who previously received whole brain RT [33]. Brain necrosis was observed in 64% of patients who were re-irradiated at > 12 months after surgery for recurrent GBM [34].
In our simulation study, there was a large variation in brain dose depending on CTV settings. If CTV was determined using the Gd-enhanced region, D50% for the brain was 3.5 Gy, but this value was 6.2 Gy using T/N 1.3 of 11C-MET, 4.2 Gy using T/N 2.0, and 5.1 Gy using P-E. Similarly, D33% was 5.4 Gy using the Gd-enhanced region, and 9.8 Gy, 6.5 Gy, and 7.9 Gy using the three other respective settings. Thus, in RT planning using 11C-MET findings, the normal brain dose is higher than that using Gd enhanced findings and can be about twice as high depending on the setting of the tumor boundary using 11C-MET. Therefore, since the normal brain dose should be as low as possible, contouring of the target should be very carefully performed if 11C-MET PET is used for re-irradiation planning. On the other hand, D50% and D33% using 11C-MET findings were very low compared to taht using T2 high region, especially brain dose using T/N 2.0 and P-E were less than half. This finding mean treatment using 11C-MET PET may cause less brain damage than MRI T2 high region based treatment.
It is also important to administer a sufficient irradiation dose to the target for local control. One purpose of our simulation study was to evaluate the dose that would have been administered to the real tumor if the 11C-MET accumulation region had been used in a real treatment planning setting. V95% of CTV (Gd) was 93.7% using the Gd-enhanced region, 53.5% using T/N 1.3 of 11C-MET PET, 81.2% using T/N 2.0, and 59.8% using P-E. Even if the tumor contour was determined using T/N 2.0, which had the highest correlation with Gd, the V95% of CTV (Gd) is only about 80%. This is not sufficient for tumor control. The simulation results using other parameters showed even lower coverage of the CTV. On the other hand, the T2 high-signal area large because it included the area of edema, and as shown in Fig. 5, it encompassed the CTV established using 11C-MET PET, therefore the V90% and V95% values of the CTVs hardly changed.
Grosu et al. found mean target volumes of 11 cm3 for the Gd-enhanced region and 19 cm3 for the 11C-MET region, with an overlap volume of 6 cm3 at a T/N ratio of 1.7. 11C-MET accumulation extended beyond the Gd-enhanced region in 29/39 cases (74%) [27]. In the current study, the respective mean target volumes were 16 cm3 and 19 cm3, with overlap of 13 cm3. The 11C-MET region was larger than the Gd-enhanced region in 16/20 cases (80%) using a tumor boundary at a T/N ratio of 2.0. These results show similar trends to those in Grosu et al. Pirotte et al. found that the 11C-MET PET volume did not match the MRI volume and improved the tumor volume delineation in 88% of low-grade glioma and 78% of high-grade glioma cases compared with surgical findings [35]. In studies using RT, Navarria et al. showed that the 11C-MET region was within the FLAIR high signal region plus 1 cm in all cases at a T/N ratio of 1.5, and that the 11C-MET region coincided with sites of recurrence [36]. Lee et al. reported recurrence in 2 of 14 patients and suggested that the 11C-MET area had a high risk for recurrence using a T/N ratio of 1.5 [37].
Based on the above findings, it seems to be reasonable to use 11C-MET PET in treatment planning, but previous studies have used different tumor boundary settings and an optimal threshold value for the tumor boundary has yet to be established. However, despite the many unresolved issues, some physicians have started to use 11C-MET PET for treatment planning. Grosu et al. found an increased mean survival period from 5 to 9 months when 11C-MET PET or 123I-α-methyltyrosine single-photon emission CT was added in treatment planning that was previously performed using contrast-enhanced MRI alone [38]. Miwa et al. reported 6-month and 1-year survival rates of 71.4% and 38.1% for recurrent lesions treated with tumor contouring with a T/N ratio of 1.3 and SBRT of 25 Gy/5 fr to 35 Gy/7 fr [39].
There are several limitations in this study. This is a retrospective study. Also, the content of this study is a simulation study, which speculates on the possible effects on the tumor and brain if the tumor were defined by 11C-MET PET or by MRI. In actual treatment, 5 of 20 patients received re-irradiation and 6 of 20 patients received surgery. Among the 5 re-irradiation cases, Gd-enhanced tumor was reduced in 3 patients, but remaining 2 patients showed recurrence. The form of recurrence was a marginal recurrence, and the site of recurrence was lateral to any of the CTVs used in this study. Therefore, this study does not validate whether 11C-MET PET or MRI is more accurate. The purpose of this study is to raise the issue of the ambiguity of the current use of 11C-MET PET and MRI in medical practice. Prospective studies are needed to establish more accurate contouring techniques in the future.
The results of this study go some way to addressing this fundamental issue and may help to establish the significance and effects of use of 11C-MET PET for contouring in RT planning.