Symptomatic intracranial hemorrhage after thrombolysis.

suggested that sulcal effacement without hypodensity corresponds to increased blood flow or cerebral blood volume and is therefore a marker for reperfusion rather than tissue damage. It is therefore highly questionable to analyze those signs together. Furthermore the evidence for a predictive value of EIC in the 3-hour time window is thin at best. Reanalyses of the data from the NINDS trial did not show a higher risk for bleeding complications in patients with EIC [2, 3] . The same is true for the 3-hour European-Australian Acute Stroke Study (ECASS) II population [4] . Still most experts agree that patients with extended hypoattenuation should not be treated. The main problem with EIC is that they are very difficult to read and have a poor sensitivity and interrater agreement [5] . At present we still do not succeed in treating enough patients based on the simple protocol given by the NINDS trial. Adding the uncertainty and the mystery of EIC has seriously impaired the implementation of thrombolysis over the years. The NINDS trial as well as clinical practice demonstrate that treatment based on the sole exclusion of hemorrhage works. Once we achieve a broader use of this simple yet effective protocol, it will be time to improve on this fundament. The second predictor identified by the authors is deviation from the treatment protocol. While this has been shown to be a risk factor in previous studies too [6–8] , we believe that there are two completely different types of deviation that need to be discussed separately. First there are the ‘unintentional’ deviations such as screening mistakes, dosing of the thrombolytic agent and deviations from general handling and treatment algorithms, e.g. blood pressure control. These are the kind of deviations that clearly need to be avoided, and it has been shown that Symptomatic intracranial hemorrhage (sICH) is the most feared complication after systemic thrombolysis for acute ischemic stroke. The respect for this complication has hampered the propagation of this life-saving and disability-reducing therapy and has led to the fact that even a decade after the approval of recombinant tissue plasminogen activator (rtPA) thrombolysis is still widely underused. Singular cases of devastating hemorrhages lead to continuing doubts of safety and efficacy of a – for neurologists unusually aggressive – therapy even after the benefit and safety of this therapy have been repeatedly shown in numerous studies. Therefore establishing predictors for occurrence of thrombolysis-related sICH is an important objective. In this issue of Cerebrovascular Diseases Marti-Fabregas et al. present a study in which they examined 347 patients treated with rtPA in 7 Spanish university hospitals from 1999 to 2004 [1] . They analyzed the frequency of thrombolysis-related bleeding complications and examined the predictive value of different clinical, radiological and laboratory data for occurrence of sICH. With only 8 cases (2.3%), the frequency of sICH was very low in their study. Logistic regression analysis for predictors of sICH yielded two very interesting results worth taking a closer look at. The first predictor – early ischemic changes (EIC) on CT – is a topic of ongoing controversies. While EIC have been an exclusion criterion for some studies, others, most prominently the NINDS trial, only used CT for exclusion of hemorrhage. First of all it has to be noted that different types of EIC have different pathophysiological equivalents. While it is believed that hypoattenuating brain tissue represents irreversible damage, it has recently been Published online: November 15, 2006