Hemorrhagic transformation of brain infarct

Hemorrhagic transformation (HT) of a brain infarct is a major concern in stroke management. Bleeding within the infarcted area is a spontaneous event, with a peak occurrence during the first week after stroke. ^[1] However, its frequency seems to be higher in patients treated with anticoagulant therapy, ^[2] especially in the context of large-sized ischemic lesions of embolic origin. ^[3-8]

The fear that HT could worsen the clinical course has generated a debate between proponents of early anticoagulation ^[5,9] and those who suggest starting such therapy only after the 48-hour CT shows no evidence of large infarcts or mass effect. ^[8,10] The issue of HT is rendered particularly pressing by the current renewed interest in thrombolytic treatment of ischemic stroke. Thrombolytics are capable of recanalizing the occluded artery, but this benefit must be weighed against an increased risk of HT, which presents the clinician with a new dilemma. Several lines of evidence indicate that the earlier the treatment is given, the higher the potential efficacy of any pharmacologic approach to ischemic stroke, ^[11] and some have suggested that drug administration should not be delayed any longer than 6 hours. ^[12] Consequently, the identification of factors that are able to predict eventual HT at an early stage might lead to a better understanding of whether thrombolytics are actually responsible for the event. However, shortly after stroke onset it is difficult to identify patients likely to develop HT, i.e., those with a large infarct due to cerebral embolism, since the actual extent of the ischemic damage is usually not yet apparent at CT.

Whether HT really does lead to a deterioration in acute ischemic stroke is also a matter of controversy. Pessin et al. ^[13] suggested that HT is only a marker of a large lesion and does not exert, per se, a negative influence on the clinical outcome. However, they derived this provocative hypothesis from a small case series that needs to be verified in a larger sample to allow for an adjustment for concomitant prognostic factors.

The present work is based on the registry of a stroke unit and focuses on a consecutive series of patients hospitalized and evaluated by a neurologist shortly after an acute ischemic stroke in the anterior circulation. All the patients underwent CT within 5 hours of onset and were systematically assessed for the occurrence of HT in the first week after stroke by means of a repeat CT or autopsy. We attempted to identify the predictors of HT among the initial clinical and radiologic findings and investigated the influence of HT on the clinical evolution, using multivariate analysis to adjust for potential confounders and for the possible effect of anticoagulant therapy.

Methods.

All patients admitted to the emergency unit of the First University Hospital of Rome from January 1987 to May 1989, following a first-ever ischemic supratentorial stroke and referred to our Stroke Unit within 5 hours of the event, were the subjects for the present study.

On admission, the neurologist on duty performed a quantitative neurological examination by administering the Canadian Neurological Scale (CNS), ^[14] the total score of which ranges from 1.5 (maximum deficit) to 10 (absence of deficit). Patients or proxies were interviewed to obtain information on medical history, with special attention to the presence of risk factors for stroke such as atrial fibrillation, other potentially embolic cardiopathies, ^[15] hypertension, diabetes, and cigarette smoking. Patients received the medical therapies required by the concomitant diseases. In addition, when indicated, we also administered glycerol, 500 ml bid, or mannitol, 150 ml four to six times per day; acetylsalicylic acid, 300 mg uid; subcutaneous heparin, 5,000 IU tid; intravenous heparin, initial bolus of 10,000 IU followed by continuous infusion adjusted to raise the activated partial thromboplastin time to 1.5 to 2 times the basal values; oral anticoagulants, according to a target prothrombin time of 1.5 to 2.5 times the control values All treatments were started within 24 hours of stroke onset

After the neurologic examination, all patients underwent a high-resolution plain CT (Siemens-Somatom CR, Erlangen, Germany). Axial orbito-meatal sections of a thickness of 2 mm at 4-mm intervals from the occipital foramen to the sellar region and of 8 mm at 10-mm intervals for the remaining brain were used to detect early focal hypodensity ^[16-17] involving the lentiform nucleus or the cortex, or both; mass effect; and hyperdense middle cerebral artery (MCA) sign. ^[18] The occurrence of HT was investigated on a repeat CT performed within the first week of stroke. HT was defined as any degree of hyperdensity within the area of low attenuation. This was distinguished as petechial when characterized by scattered or spotty distribution and as parenchymal when it appeared as a small or massive hematoma. ^[2] In addition, we assessed the site and size of the qualifying infarct, presence of mass effect, and possible persistence of hyperdense MCA sign. Based on standard templates, ^[19] the site of the lesion was defined as (a) subcortical, when involving the MCA deep branch or internal borderzone territories; (b) cortical, when involving the MCA superficial branch territories; (c) cortico-subcortical, with partial to total involvement of the MCA deep and superficial territories. The size of infarcts was quantified as (a) small, when involving less than one-half a lobe or when CT were permanently negative; (b) medium, when between one-half and one lobe was affected; (c) large, when the lesion involved more than one lobe. ^[20] Mass effect was graded as (a) slight, when only a sulcal effacement or compression of ventricles without dislocation was present; (b) moderate, when we observed a partial ventricular midline shift; and (c) severe, when we observed a total ventricular midline shift. The size of lesion, mass effect, and occurrence of HT in patients who died before the second CT examination were determined by autopsy.

Follow-up duration was 30 days. The CNS was administered daily for the first 10 days or until coma or death At. the end of follow-up, we calculated the case fatality rate, distinguishing between cerebral, cardiac, and other causes of death (pulmonary embolism, sepsis) and graded the disability in survivors by means of the Barthel Index Scale. ^[21]

Results.

Our series comprised 150 consecutive patients (mean age, 66.1 plus minus 7.5 SD; 85 men, 65 women) with a first-ever ischemic supratentorial stroke diagnosed on the basis of the clinical picture and of the CT performed within 5 hours of onset. Repeat CT was performed in 143 patients between day 5 and day 9 after stroke. Seven patients died before the CT and underwent autopsy. HT was observed in 65 patients (43%): 23 (35%) had scattered and 35 (54%) had spotty petechial infarcts, whereas five patients (8%) had a small and two (3%) a massive deep parenchymal hematoma.

Discussion.

In this series of first-ever anterior circulation ischemic strokes, 43% of patients showed HT within the first nine days of the event, which is in keeping with figures reported in previous prospective CT studies. ^[2,6]

In agreement with the literature, patients who developed HT were more severely affected at hospital admission, ^[6] whereas risk factors for stroke were present with similar frequencies in HT and non-HT patients. The two groups of patients had the same spectrum of treatments, which would make antiplatelet agents or anticoagulants less likely to explain HT. This is also true for the subgroup of seven patients with hematoma, only three of whom were given anticoagulants (one with a massive and two with small hematomas). This is in contrast to the reported higher frequency of secondary hematomas in anticoagulated patients. ^[2]

An important feature of our study is that the patients underwent their first CT within 5 hours of stroke. Although traditional spin-echo magnetic resonance is reported to detect secondary bleeding better than CT, ^[23] the latter is at least as accurate in identifying the very early signs of the infarct. ^[24-25] Moreover, the fact that CT is still the most widely used tool both in daily clinical practice and in therapeutic trials, including those with thrombolytic drugs, ^[26-30] guided our choice to look for CT predictors of HT in the hyperacute phase of stroke. Focal hypodensity on CT was selected by logistic regression as the sole independent predictor of HT. Unfortunately, we did not manage to characterize patients who developed parenchymal hematoma, since they did not show differential CT features when compared with patients with petechial HT. In a sample of 93 patients submitted to thrombolytic therapy, Wolpert et al ^[31] recently found HT in 45% of patients exhibiting signs of early ischemia at first CT, as opposed to 30% in their counterpart. However, this difference was not statistically significant, probably due to small numbers.

We can speculate on the pathogenesis of HT in our patients on the basis of these data. Early focal hypodensity is reputed to be a sign of intracellular edema, ^[7,32] and initial mass effect is due to extracellular edema; both indicate that, some hours following stroke onset, patients with subsequent HT already had marked brain edema. Therefore, the most probable mechanism leading to HT was the reopening of the pial collaterals, following a decrease in the compression initially exerted by the brain edema. ^[6] The spotty or scattered petechial pattern, with cortical or cortico-subcortical distribution, present in 89% of HTs in our study supports this hypothesis. The role of MCA reopening, long postulated as the main event leading to HT, ^[33] was minimal in our patients, ^[34] probably accounting, at most, for the remaining 11% of HTs with characteristics of deep parenchymal hematoma. Moreover, the lower frequency of the hyperdense MCA sign at the first CT in the non-HT group suggests that these patients might have had fewer embolic MCA occlusions. ^[35-36] This would be consistent with the lower frequency of large infarcts, ^[35-36] and hence of HTs, shown by this subgroup at the second CT or autopsy.

HT patients more frequently developed early neurologic deterioration compared with non-HT individuals. They also had a worse 30-day outcome, in terms of both case fatality rate and residual disability of survivors. Brain herniation accounted for 63% of deaths among HT patients as opposed to none in the non-HT subjects; this might, in part, account for the higher rate of HT reported in the autopsy series ^[3] than in the CT series. ^[2,4-8] However, apart from the two patients with massive hematoma who deteriorated and died during the first 5 days of stroke, HT did not appear to independently influence either early deterioration or 30-day outcome, once the size of the lesion or the severity of mass effect at the repeat CT and autopsy were considered. ^[13] In other words, we observed that large lesions and mass effect were related to poor outcome irrespective of the presence or absence of HT. The fact that 95% of the patients underwent a repeat CT within a similar time interval from stroke onset, irrespective of their clinical evolution, made our CT findings reliably comparable. Among HT patients, the administration of antithrombotics was not linked to a higher frequency of early deterioration or of poor outcome, indicating that neither the occurrence of HT, as mentioned above, nor the clinical evolution of HT patients were influenced by these drugs.

In conclusion, HT after ischemic stroke is a frequent occurrence that is spontaneous and not determined by the treatments administered, including antithrombotic agents. In the vast majority of cases, HT is likely to be the consequence of the reperfusion of pial collaterals and does not affect outcome, whereas the infrequent cases of deep hematoma are probably determined by the reopening of the occluded artery and are likely to influence outcome when the hematoma is massive. As CT is quite reliable in predicting HT (focal hypodensity detected from the very early hours of stroke onset), hyperacute CT findings can be a guide to quantify the risk of a therapeutic approach with antithrombotic and thrombolytic drugs.

Acknowledgments

We gratefully acknowledge Prof. Cesare Fieschi for his criticism and advice in reviewing the manuscript. We are also indebted to the residents staff for their care and selection of the patients. Special thanks to Mr. Lewis Baker for having revised our English.