PRE-GKS Annual Hemorrhage Rates
Total observation period was 84.0 patient-years. A total of 75 hemorrhage episodes in 57 patients were counted. Since the date of diagnosis was the same as the date of the first hemorrhage, the initial hemorrhage was not considered as an episode in the calculation. After excluding initial hemorrhage episode, 18 episodes of recurrent hemorrhage happened. The AHR of all CCMs was 21.4% (Fig. 1A).
When calculating the AHR for brainstem lesions, of a total of 18 episodes, 9 episodes occurred in the brainstem. Total observation period was 33.1 patient-years. The AHR of brainstem CMs was 27.2% (Fig. 1A).
POST-GKS Annual Hemorrhage Rates
A total of 22 hemorrhage episodes occurring in 16 patients were observed during the follow-up period after GKS. Six episodes were recurrent. Each of the four patients had hemorrhage twice. One patient had hemorrhage three times. The other 11 patients had a single episode.
Figure 1B shows trend of the timing of the first 16 hemorrhages after GKS. After 15 years since performing GKS, the first hemorrhage was not observed. If recurrent episode was considered, it was observed even after 15 years (Fig. 1C). Fifteen episodes of asymptomatic hemorrhage were also observed.
Total observation period was 1,084 patient-years. The AHR of all CCMs was 2.0%. During the first two years after GKS, six episodes occurred in which the AHR (< 2 years) was 3.8% (6 hemorrhages/158 patient-years). Between 2 years and 10 years after GKS, 9 episodes occurred. The AHR at < 10 years was 1.4% (9 hemorrhages/ (632-5) patient-years). Two lesions located in the brainstem were censored at 7 and 8 years, respectively, after GKS due to the need for secondary GKS. Ten years after GKS, the AHR was 2.3% (7 hemorrhages/299 patient-years) (Fig. 1A).
When calculating the AHR for brainstem lesions, 14 of a total of 22 hemorrhage episodes occurred in the brainstem (n = 26). Total observation period was 369 patient-years. The AHR of brainstem CMs was 3.8%. During the first two years after GKS, three episodes of hemorrhage occurred in which the AHR (< 2 years) was 5.8% (6 hemorrhages/52 patient-years). Between 2 years and 10 years after GKS, 7 episodes occurred. The AHR at < 10 years was 3.4% (7 episodes/ (208-5) patient-years). Ten years after GKS, the AHR was 3.5% (4 episodes/114 patient-years) (Fig. 1A).
Neurological and Seizure Outcomes
Thirty-five (44.3%) of 79 patients had an initial presentation of focal neurologic deficit. Many cases of neurological symptoms appeared together in combination. At the last clinical follow-up, 17 (48.6%) patients had fully improved symptoms and 9 patients (25.7%) had partially improved symptoms. A total of 74.3% of patients showed a favorable neurologic outcome (Table 1). There were no cases of mortality related to CCM.
Thirteen (17%) patients presented with seizure. Six of them had sporadic episodes and seven showed epilepsy. One of them had received surgical removal of CCM at 10 years after GKS. Among these sporadic cases, five (83%) of six patients had a favorable outcome. For epilepsy cases, three (43%) of seven patients had a favorable outcome (Table 2). Overall, eight (61.5%) of 13 patients achieved a favorable outcome. Locations of these lesions were comprised of frontal cortex (n = 6), temporal cortex (n = 6), and parietal cortex (n = 1).
Adverse Radiation Effect
After GKS, a total of 16 perilesional edemas in 15 (19%) patients were observed. In three lesions, perilesional edema was accompanied by hemorrhage. All edemas occurred within a mean of 1.1 years. Thirteen of 16 perilesional edemas occurred in the supratentorial location (frontal, n = 8; temporal, n = 3; occipital, n = 1; insula, n = 1; cerebellum, n = 2; midbrain, n = 1). Among a total of 43 supratentorial lesions, the mean dose for lesions with perilesional edema was 18.3 Gy and the mean GTV was 1.82 cm3. The mean dose for lesions without perilesional edema was 19.4Gy and the mean GTV was 0.91 cm3. When performing Student's t-test, volume (p = 0.019) showed a significant difference rather than dose (p = 0.224).
Five (6.3%) of these 15 patients presented symptoms. All these symptomatic AREs occurred in non-brainstem location. Four patients recovered from their symptoms. One patient developed a permanent facial nerve palsy. Perilesional edema had resolved within a mean of one year.
Changes of Lesions in the Follow-Up MRI
Of a total of 96 lesions, 78 (81.3%) decreased in size (Fig. 2), 10 (10.4%) remained stable, and 8 (8.3%) increased in size. Particularly, 28 (35.9%) of 78 lesions in the 'Decrease' group were found to have completely obliterated cores in the last follow-up MRI.
At the time of GKS, there were 34 (35.4%) type I lesions, 58 (60.4%) type II lesions, and four (4.2%) type III lesions. At the last follow-up MRI, there were only two (2.1%) type I lesions, 63 (65.6%) type II lesions, and 31 (32.3%) type III lesions. For the 16 lesions that experienced hemorrhage after GKS, at the time of GKS, there were 9 lesions of type I, 7 lesions of type II, and no lesion of type III. In 13 patients with seizure, all lesions were type II at the time of GKS. Ten lesions were type II and 3 lesions were type III at the last follow-up MRI.
Of the 96 lesions, a total of 24 (25%) were found to have a DVA near the lesion. DVAs were commonly found in deep-seated locations (17/24, 71%) such as brainstem (n = 8), cerebellar peduncles (n = 4), basal ganglia (n = 3), and thalamus (n = 2). Others were located in relatively superficial areas, including frontal lobe (n = 2), temporal lobe (n = 1), occipital lobe (n = 1), and cerebellar cortex (n = 3). Only two (8.7%) of these DVA related lesions showed a hemorrhage episode.
Risk Factors of Hemorrhage
Different variables such as age, sex, location (brainstem vs non-brainstem), history of hemorrhage before GKS, DVA, MRI type based on the Zabramski’s classification (Type I or not), and GTV were analyzed to find out the statistically significant risk factors influencing on hemorrhage episode. Due to the dependency of dose on location, dose was not included as a factor in the analysis. In Cox-regression analysis, Previous hemorrhage history (HR 8.38, 95% CI 1.07–65.88; p = 0.043), Brainstem location (HR 3.10, 95% CI 1.26–7.64; p = 0.014) were statistically significant. Age (p = 0.917), Sex (p = 0.825), DVA (p = 0.386), MRI type (p = 0.590), GTV (p = 0.783) showed no significant relationship with hemorrhage episode.