Infratentorial cerebral hemorrhage due to a direct carotid–cavernous fistula (CCF) is very rare. To our knowledge, only four such cases have been reported. Cerebellar hemorrhage due to a direct CCF has not been reported. We describe a 63-year-old female who presented with reduced consciousness 3 days after undergoing a maxillectomy for maxillary cancer. Computed tomography showed a cerebellar hemorrhage. Magnetic resonance angiography showed a left-sided direct CCF draining into the left petrosal and cerebellar veins through the left superior petrosal sinus (SPS). Her previous surgery had sacrificed the pterygoid plexus and facial vein. Increased blood flow and reduced drainage could have led to increased venous pressure in infratentorial veins, including the petrosal and cerebellar veins. The cavernous sinus has several drainage routes, but the SPS is one of the most important routes for infratentorial venous drainage. Stenosis or absence of the posterior segment of the SPS can also result in increased pressure in the cerebellar and pontine veins. We emphasize that a direct CCF with cortical venous reflux should be precisely evaluated to determine the hemodynamic status and venous drainage from the cavernous sinus.
Cavernous Sinus ; Cerebral Hemorrhage ; Consciousness ; Constriction, Pathologic ; Drainage ; Female ; Fistula* ; Hemodynamics ; Hemorrhage* ; Humans ; Magnetic Resonance Angiography ; Middle Aged ; Veins ; Venous Pressure

The DCCT syudy report (in 1993) showed that long-term maintenance of near normoglycemia markedly delays the onset and/or progression of microangiopathic complications of type 1 diabetes. Now, intensive insulin therapy has been considered to be a standard treatment of type 1 diabetes. However, it brings about serious hypoglycemia three times as frequently as conventional therapy with split-dose insulin mixtures. If IDDM patients could predict their blood glucose levels, it would be beneficial to prevention of hypoglycemia and to strict glycemic control. We intended to research whether insulin-treated diabetic children in a diabetic summer camp could predict their blood glucose levels. The number of subjects was 28. They were elementary and junior high school children 9 to 16 years of age. They predicted their blood glucose levels before every meal and bedtime, and then monitored their blood glucose levels. Their HbA₁c was 7.6±0.3%, total daily insulin dosage 36.0±3.2U/day, and infection times 3.3±0.2/ day.
Measured blood glucose levels and predicted blood glucose levels were not distributed normally, but their natural logarithms (Ln (mBG), Ln (pBG)) were. Ln (pBG) (5.01±0.02) was significantly higher than Ln (mBG) (4.92±0.03) (p<0.01)
A positive correlation was found between Ln (mBG) (x) and Ln (pBG) (y) (y=0.359 x+3.239, r=0.495). Events of which Ln (pBG) was within±20% and±30% of Ln (mBG) were 124 (31.5%) and 175 (44.5%), respectively, of the total 393 events. As for the relationship between Ln (mBG) and Ln (pBG) at each time, a change of Ln (mBG) was significantly larger than that of Ln (pBG).
These results showed that prediction of blood glucose levels was difficult for type 1 iabetic children, especially when blood glucose levels were extremely high or low and when they fluctuated sharply.