Driven by the observation that patients who had received protons seemed subject to contrast enhancing imaging changes that differed from the typical Ps seen after photons, a retrospective analysis of post-treatment imaging changes following both radiation modalities was performed.
Based on our analysis, 24.6% of patients who received proton therapy for glioma had imaging changes unique to treatment with protons and different from usual pseudoprogression (Ps) seen after photon therapy, which we term Proton Pseudoprogression (ProPs). We establish criteria which can be used to define ProPs as different from photon related Ps. ProPs are typically small and round or oval in shape, appear later than photon Ps approximately 2 cm deep to the radiation target, and have a high affinity for white matter.
Current RANO guidelines for HGG suggest that Ps should be considered within the first three months after chemoradiation therapy, and any new lesion for HGG or LGG patients qualifies as progressive disease . The 14 patients in our study who exhibited PRoPs would inaccurately be considered as having progressive disease under these criteria, and our ProPs criteria addresses the unique changes that occur following proton therapy. These changes did not have the radiologic hallmarks of tissue necrosis, and most patients resolved without clinical symptoms or treatment.
Methylation was seen in the two patients whose pathological report tested MGMT promotor status. IDH1 mutation status in LGG has been shown to decrease Ps but increase Ps rates in HGG [15, 16]. In our study, 12 of the 14 patients with ProPs had IDH mutant tumors. An association between IDH mutated tumors and ProPs may likely be due to the longer survival of patients with IDH mutated tumors. Codeletion of 1p/19q has been linked with lower risk of Ps , and five patients with the codeletion also had ProPs.
The specificity to proton radiation and the locality of ProPs can possibly be explained by the increased RBE of protons at the end range and beam angle selection. The exit beam of protons targeted roughly 2 cm from the observable tumor boundary and when beams are superimposed, may lead to the “fluffy” enhancements outside the resection cavity in a predictable location. While a constant RBE of 1.1 is commonly used when planning treatment with protons, the true proton RBE likely varies with linear energy transfer (LET), dose, physiological and biological factors, as well as endpoint . At the distal edge and distal falloff of a spread-out Bragg peak (SOBP), the RBE are reported to be considerably higher than 1.1 [18, 19]. Variation in RBE as compared to the clinically utilized value may lead to dose distribution that differs from the treatment plan and to higher depositions of energy in the locations where ProPs occurs, further supported by the relationship between higher Ps incidence and higher radiation dose [2, 7].
A predictive model of localizing late contrast-enhancing brain lesions by Bahn et al. added clinical evidence for RBE that increases with linear energy transfer, suggesting the non-random appearance of proton-induced enhancements and arguing for consideration of increased RBE with LET during treatment planning . Another study of 34 pediatric ependymoma patients saw a correlation of changes following proton therapy on MR images with increased LET, providing clinical evidence of variable proton biological effectiveness . In a model of six glioma patients following proton therapy, a correlation with dose and LET to the spatial distribution of late treatment-induced MR changes adds to the evidence of dose-response modeling for proton therapy . Also implicated in Bahn and Eulitz’s models are the localization of lesions near the ventricle system, as decreased tissue repair capacity and vascular supply may make these areas more vulnerable [20, 21].
While pseudoprogression following photons occurs 3–6 months following therapy, radiation necrosis occurs 6 months to several years post-treatment, often progressing without treatment, exhibiting mass effect, accompanied by clinical worsening [23, 24]. While the mean time to ProPs was about 15 months and as late 27 months, no mass effect was seen, the lesions were small, and patients were generally asymptomatic or had mild symptoms. Further, radiation necrosis typically exhibits T1 hypointensity and DWI positivity, which was not seen in the ProPs enhancements, nor was the classic “soap bubble” enhancing appearance [13, 25]. Perfusion imaging was seen to be normal or slightly increased in the imaging of ProPs. In recurrent tumor, perfusion measured by relative cerebral blood volume (rCBV) is typically elevated, while the reduced blood flow of necrotic brain in radiation necrosis has lower rCBV values. MR spectroscopy is another modality used to differentiate these processes, with tumor recurrence characterized by high choline/creatine and choline/N-acetylaspartate ratios, and necrosis with elevated lactate and lipid peaks . The enhancements seen following proton therapy in our cohort showed mildly elevated choline levels.
Possible limitations of this study include selection bias of the two treatment cohorts. Similar proportions of grade III and II tumors were used in each cohort, but the proton cohort had a higher percentage of IDH mutated tumors and MGMT methylated tumors (Table 1). Future studies should confirm our findings in an independent cohort and investigate outcomes in the context of ProPs.
In conclusion, proton radiation therapy can induce a pattern of Ps which manifests differently than Ps following photon therapy. Patients present with ProPs significantly later than the 3 months defined by RANO criteria and would inaccurately be considered tumor recurrence by the guidelines. The appearance of ProPs is often patchy, located in white matter, opposite the target beam entry, usually ~ 2 cm from the resection cavity. Recognizing ProPs can help prevent unnecessary advanced imaging, treatment for mistaken tumor progression, and inaccurate enrollment in clinical trials. Ongoing trials that allow use of proton radiation will need to distinguish ProPs from active tumor progression to avoid inaccurate interpretations.