Hypertensive Cerebral Hemorrhage. Demonstration of the Source of Bleeding

There have been many studies of hypertensive cerebral hemorrhage and the associated microvascular abnormalities. The actual primary site of bleeding has probably never been identified. In the present study, a 2- × l-cm hypertensive thalamic hemorrhage was studied in its entirety using continuous 8-μm-thick serial sections and stained with Mallory's phosphotungstic acid hematoxylin stain. The site of bleeding was focal rupture of a thin-walled elongated aneurysmal dilatation of a small artery normally 180 μm in diameter. The dilatation was 5-mm long and 600 μm in diameter. At the proximal end of the dilatation, the feeding artery showed a typical hypertensive lipohyalinotic lesion. There was only 1 site of bleeding. It was concluded that the aneurysmal dilatation was a preformed intracerebral saccular aneurysm of unusual shape and size. Rupture occurred during the period of hypertension induced by the infusion of pitressin. The presence of the lipohyalinotic lesion is considered a coincidence. This is probably the first demonstration of the primary site of arterial rupture in a hypertensive hemorrhage. The source of bleeding in hypertensive intracerebral hemorrhage has been studied over the years by many pathologists beginning with Charcot and Bouchard in 1868 (1). The most thorough investigations were by Matuoka in 1939 and 1952 (2, 3). It is probably correct to say, however, that despite the many studies of hypertensive brain hemorrhage, the actual primary site of bleeding has never been demonstrated. One of the main obstacles is the blood clot itself, which when fixed in formalin preparatory to making microscopic studies, becomes hardened to a degree that precludes a methodical microscopic study of the artery or arteries suspected of harboring the bleeding site. Another complicating factor is the development of secondary hemorrhages, “bleeding globes,” as a result of the mass effect of the main hemorrhage stretching and disrupting the surrounding arteries, causing secondary bleeding, in which case it may not be possible to distinguish the primary bleeding site from the secondary sites (4). In the pathological study reported here, a thalamic hemorrhage 2 cm in diameter was sectioned serially in its entirety revealing a single site of bleeding. This is probably the first demonstration of the vascular process at the site of an intra-cerebral hemorrhage. The hemorrhage occurred in a patient known to be hypertensive, but the unusual circumstances in which the hemorrhage occurred might require that the term “classical” be held in reserve. The patient suffered severe third degree burns complicated by duodenal bleeding and hypotension (shock). Treatment with intra-arterial pitressin resulted in severe hypertension (230/110 mm Hg), at which time the thalamic hemorrhage occurred. This raised the question whether induced hypertension adds any additional features to the more natural course. Since the pathological changes at a site of bleeding are unknown in any case, the question cannot be answered. A hypertensive 72-yr-old man suffered third degree burns to 35% of his total body surface, involving face, chest, and arms. In the Emergency Department he was cooperative but vomiting. The blood pressure was 150/84 mm Hg and the pulse 104/min. There was a history of a stroke involving the right side of the body 3 yr before, with almost full recovery. Emergency treatment consisted of fluid replacement, antibiotics, steroid, silver nitrate dressings, and nasogastric suction. On hospital day (HD) 2, gastric bleeding associated with hypotension required transfusion of 7 units of blood. Bleeding continued and on HD 5, coeliac angiography demonstrated a site of bleeding in the duodenum. An intra-arterial infusion of pitressin, 0.2 units per ml, was begun. Six hours later the patient became stuporous, reacting only to painful stimuli. During the infusion the blood pressure ranged from 190/90 to 230/110 mm Hg and remained in the range of 170 to 190/90 mm Hg for 4 days. Renal and pulmonary complications failed to respond to therapy and the patient died 14 days after the onset of stupor. The cerebrospinal fluid was xanthochromic. A general pathologic examination showed thermal burns, acute bronchopneumonia, and 3 gastric ulcers, 1 with perforation. The heart weighed 360 g. The left ventricular wall measured 18 mm in thickness. The brain weighed 1,400 g. There were no surface abnormalities. Coronal sectioning at l cm intervals revealed a subacute hemorrhage in the mid-portion of the right thalamus measuring 2.0 cm in length (anterior-posterior) and 1.0 cm wide (horizontal). The hemorrhage bordered on the third ventricle but did not enter it grossly. The color was consistent with an age of 2 wk. In addition to the hemorrhage there were 4 old lacunar infarcts in the basal ganglia and 1 in the basis pontis. The blood in the thalamic hemorrhage was not unduly firm and serial sectioning was undertaken. The 2 cut surfaces of the divided thalamus were placed in apposition and embedded in paraffin. The block was serially sectioned at 8-μm thickness, retaining the partial sections from the apposed surfaces. All sections were stained, using the phosphotungstic-acid hematoxylin method of Mallory. (A single stain was used since the use of more than 1 stain sometimes results in interruption of the continuity of the numbered slides). Microscopic examination of the entire region of the hemorrhage revealed only 1 site of bleeding. The hemorrhage arose from a dilated, elongated, aneurysmal segment of a small artery normally about 180 μm in diameter. The dilated segment was 5 mm in length and 600 μm in diameter. Its wall was thin and poorly demarcated, consisting of a few loosely arranged fibrous strands. By following the aneurysmal outline in the serial sections it was established that the fibrillar material constituted a thin retaining wall enclosing the contents of the lumen everywhere except at the site of a “break” 1 mm wide, through which the intravascular contents were in continuity with the extravasated blood of the thalamic hemorrhage (Fig. 1A–C). This was the only site of extravasation. Fibrillar strands lying free in the vicinity of the “break” were probably the free edges of the gap in the wall. The appearance suggested physical disruption at the site. A: Region of the site of bleeding showing aneurysmal dilatation (AD) of artery (AD), the site of rupture (SR), the hypertensive lipohyalinotic lesion on the afferent artery (LH), the parent feeding artery running along one side (FA), and the area of hemorrhage (H). The plane of microscopic sectioning fortuitously included virtually all of the principal structures in the same section. The entirety of the involved vasculature was encompassed in 40 serial sections. B: Higher power view of the aneurysmal dilatation and the site of rupture. C: Same view as in (B) with the main structures traced out in ink. The site of rupture lies at the intersection of the arrows The proximal end of the aneurysmal segment was continuous, with a classical hypertensive lipohyalinotic lesion involving the afferent feeding artery. The arterial wall was thickened as the normal compact laminar arrangement was replaced by loosely arranged bundles of fibers separated by fluid-filled spaces. Small deposits of fibrin were present in the wall. The lumen was preserved but narrowed. Lipohyalinosis is known by several names, including hyalinosis, plasmatic vascular destruction, hypertensive fibrinoid necrosis, fibrinoid necrosis, etc (5). There was no evidence of dissection of the wall. Immediately proximal to this lipohyalinotic lesion, the small artery (approximately 200-μm diameter) joined the parent artery of which it was a branch, and whose wall was normal except for increased thickness consistent with arterial hypertension. A sketch of the arterial arrangement is presented in Figure 2. When traced distally, the aneurysmal segment resumed the normal structure of a small cerebral artery with a slightly thickened wall. Panoramic diagram of the involved structures. Abbreviations: Aneurysmal dilatation (AD), rupture (R), lipohyalinosis (L), and body of hemorrhage (H) A review of the literature revealed no description of a hypertensive arterial lesion of similar appearance. Nor was one found in this author's material relating to hypertensive small vessel disease (5). Four microvascular lesions or processes are potential sources of hypertensive intracerebral hemorrhage: 1) miliary aneurysms associated with lipohyalinosis; 2) miliary saccular aneurysms; 3) asymmetric fusiform aneurysms; and 4) “bleeding globes” (5). The elongated aneurysmal dilatation found in the present case does not readily fall in any of these categories. One description of the arterial lesions associated with lacunar infarcts noted that a hypertensive lipohyalinotic process may extend along an artery for a distance of up to 10 times the diameter of the involved artery, but a tubular aneurysmal dilatation of such a segment was not mentioned (6). The vascular lesion is open to at least 2 interpretations. First, the elongated arterial dilatation may be a miliary saccular or fusiform aneurysm, an abnormality found with some frequency on the cerebral arteries of older patients both hypertensive and normotensive (7). To be tubular and as large as in the present case would be most unusual. Such aneurysms are suspected of rupturing and causing cerebral hemorrhages, although there has never been a clinico-pathological confirmation. The present dilatation may have been more rounded originally and become flattened by the extravasated blood. Such an aneurysm would have been present before the present illness and ruptured during the elevated blood pressure induced by the pitressin infusion. A second possibility is that the lipohyalinotic process lying just proximal to the aneurysmal dilatation extended distally, damaging and weakening the wall of the dilated segment. Against this formulation is the complete absence of even traces of residual muscularis and elastica in the aneurysmal wall. In this case the presence of the lipohyalinotic lesion would be a coincidence. The fact that there was only 1 cerebral hemorrhage and 1 aneurysmal dilatation supports the suggestion that the arterial lesion predated the infusion. Although the patient lived for 14 days after the onset of the hemorrhage, secondary changes seemed not to have altered the original post-rupture state to an important degree. There was no inflammatory response. The remnants of the aneurysmal wall at the site of “rupture” appeared to be preserved. Platelet masses lay within the lumen and immediately outside the site of bleeding. The walls of the hemorrhage showed abundant hemosiderin-filled macrophages. Examination of the thalamic tissue in the rest of the block showed none of the 4 arterial changes referred to above. Small arteries had thickened walls reflecting chronic arterial hypertension. There was no sign of a previous hemorrhage or of lacunar infarction. The changes of amyloid angiopathy were not seen. Hypertension was no doubt a factor in production of the lesion giving rise to the hemorrhage. The patient was known to have hypertension. His blood pressure in the Emergency Department was 150/84 mm Hg despite extensive burns. He probably had had a pure motor hemiplegia 2 yr before. Pathological examination of the brain showed old lacunar infarcts in the basal ganglia and the pons. The heart weight at autopsy was 360 g and the left ventricle was 18-mm thick, both consistent with hypertension. The morphology of the hemorrhage provided some hints concerning the dynamics involved. The direction taken by the escaping blood must have depended on where on the circumference of the aneurysmal dilatation the “break” occurred. The jet of escaping blood formed a rounded oblong pool with its long axis at a right angle to the bleeding artery. As portrayed in Figures 1 and 2, the hemorrhage appeared to be restricted along one side by a long stretch of the parent artery, an indication that a neighboring arterial system may determine the shape of a hemorrhage and direct it towards the more distal territory of an artery. The exact location, shape, and orientation of hypertensive cerebral hemorrhages have not been studied. These points are raised in case they are relevant in the interpretation of hemorrhages displayed by CT and MRI imaging. The diffusion of blood in gray matter would not be the same as in white matter. In summary, the site of bleeding in a hypertensive cerebral hemorrhage has been identified. The arterial lesion from which the hemorrhage arose was an elongated thin-walled aneurysm 5-mm long. Such aneurysms were found in 14 of 15 hypertensive brains in which special postmortem injections of barium sulfate were made (7). In the present case it must be inferred that rupture was related to the severe hypertension induced by the arterial infusion of pitressin. Intracerebral hemorrhage occurring in the natural course of high blood pressure could arise from such an aneurysm, rupture of a lipohyalinotic arterial lesion, or another as yet unidentified process.