Metastatic brain tumors are the most common brain tumor in adults and the frequency of brain metastasis is increasing with up to 200,000 new cases every year [2]. External beam radiation therapy, in particular SRS, is considered part of first line therapy for intracranial metastases [21]. The efficacy of SRS in patients with intracranial metastases has been shown to have control rates of 70–90% [22]. One of the most common problems of SRS for both primary brain gliomas and intracranial metastases is correctly identifying progressive reactive changes from radiation injury. Early true progression is difficult to distinguish from reactive changes (pseudoprogression) in the short term and irreversible injury (radiation necrosis) at latter time points [23]. Radiation necrosis is difficult to distinguish from tumor recurrence by both clinical presentation and imaging studies, and can be seen in up to 25% of patients after SRS [24]. Both recurrent tumor and radiation necrosis demonstrate increased FLAIR signal and disruption of the blood brain barrier resulting in contrast enhancement [3, 13]. The ability to accurately identify true progression from therapy-related changes is critical as it enables appropriate therapeutic intervention. Even with MRI techniques such as perfusion[25] and spectroscopy[26], differentiation between radiation necrosis and active metastatic brain lesion is difficult and brain biopsy remains the gold standard [3].
FDG, while widely used, has discordant results in its ability to differentiate recurrent brain metastasis from radiation necrosis, possible due to different thresholds used in each study and elevated background brain parenchymal uptake [27]. Amino acid PET agents such as [F-18]-fluroethyltyrosine ([F-18]-FET) and [C-11]-MET [28, 29] have been used with some success as a means to differentiate progressive metastatic disease from radiation necrosis. [F-18]-FET TBR values of have been shown to accurately identify recurrent metastases with metastatic uptake being significantly higher than that of radiation necrosis [30]. Additionally, dynamic FET PET imaging has been shown to improve accuracy in distinguishing recurrent disease with characteristic time-activity curves [31]. None of these most commonly used amino acid PET radiopharmaceuticals used for intracranial metastatic evaluation are yet FDA approved and thus have limited application in research studies in the United States. Fluciclovine, on the other hand, is FDA approved for evaluation of biochemically recurrent prostate cancer and has orphan drug status for evaluation of brain gliomas. Several other extra-prostatic malignancies including breast [32], renal [33], colon and lung (unpublished personal experience) have also been shown to have increased fluciclovine uptake. Our goal in this study was to evaluate the ability of fluciclovine to distinguish progressive metastatic lesions from radiation necrosis.
In this small sample set, all lesions, including both recurrent disease and radiation necrosis, demonstrated progressive post-contrast enhancement on prior standard of care MRI studies. There was overall good correlation between follow up MRI findings and pathology results when available. It should be noted that there was a single lesion that was initially suggestive of recurrent disease on short term follow-up with progressive increase in size and enhancement on subsequent follow up MRI at 2 and 4 months. Conversely, there was low fluciclovine uptake in this lesion (SUVmax of 0.8 at 5 min increasing to 1.3 at 55 min) suggestive of radiation necrosis, and radiation necrosis was confirmed upon surgical resection and final pathology.
It is important to note that fluciclovine uptake in the recurrent disease was relatively stable over the 55 minutes of imaging (Fig. 3). Conversely, fluciclovine PET uptake in radiation necrosis showed mild progressively increased uptake for the duration of the uptake scan resulting in lower accuracy at the 55 minute time point. These observations suggest that there may be differing time-activity curves between lesions with recurrent disease and radiation necrosis which may further help distinguish them from each other, although further investigation is needed. Moreover, it appears that optimal timing of image acquisition to distinguish radiation necrosis from recurrent disease for fluciclovine is at early time points ( up to 30 min) as progressive fluciclovine uptake in radiation necrosis lesions at later times points may confound discrimination. It is important to note that although fluciclovine uptake in radiation necrosis was lower compared to that of recurrent disease, it remained greater compared to contralateral normal brain parenchyma. This is possibly due to fluciclovine accumulation in inflammatory processes which while less than with FDG, is still present [34]. Lastly, the overlap of fluciclovine uptake also likely reflects the heterogeneity of the treated lesions with coexistent viable tumor and radiation related changes which are typically seen on histological examination [35].
There are several limitations of our study. One limitation is the small patient population in both the recurrent disease and radiation necrosis groups, with a total of 8 patients having 15 lesions. Of these lesions, 11 met criteria for recurrent disease and only 4 were in radiation necrosis. However, despite such a small number of patients and lesions, we were able to achieve statistical significance in fluciclovine uptake between two groups with an optimal SUVmax threshold of ≥ 1.3. Secondly, pathological confirmation was not available for all of the patients and lesions were categorized based on follow up MRI findings which is not ideal. However, in all instances in histological verification was available, fluciclovine findings were consistent with pathology (Figure 2), even when MRI suggested otherwise. Additionally, this study included intra-cranial metastatic disease from 4 different primaries and no patients had fluciclovine PET prior to SRS to demonstrate fluciclovine uptake in areas of viable disease. Moreover, this study was not powered for evaluation of intracranial metastasis from any one individual primary malignancy. Finally, it is not known if there is an optimal temporal point after radiation therapy to discriminate between the two etiologies.
Further investigation with larger data sets is needed to confirm these preliminary findings and to further establish optimal PET imaging parameters. Specific questions that will be evaluated include optimal timing of fluciclovine PET for evaluation of brain metastasis in post radiation patients, evaluation of fluciclovine PET parameters to simultaneous MRI techniques. The observed difference in background brain fluciclovine uptake between patients with recurrent disease and radiation necrosis is unable to be adequately explained and is believed to be an artifact from the small sample size and will be further evaluated on a planned study with a larger sample size. If fluciclovine PET is found to have a high accuracy in distinguishing recurrent disease from radiation necrosis, this may help guide biopsy and obviate the current need for serial MRI evaluation after treatment, saving both time and money.