Malignant cerebral infarction is a large hemispheric infarction caused by a middle cerebral artery (MCA) or internal carotid artery (ICA) occlusion. This state occurs in between 10 and 15% of all cerebral infarction patients, and has a mortality rate of up to 80%, despite maximum conservative care^7,17,32). Decompressive craniectomy and duraplasty are widely performed to reduce intracranial pressure and improve circulation to reduce the risk of mortality in cerebral infarction patients^3,6,32), and several authors have reported that surgical decompression lowers the mortality rate to between 16 and 42 percent^{4,5,11,13,17,20),} but in most of these studies, infarctions were limited to the MCA territory. On the other hand, Klincer et al. reported surgical results for ICA territory infarctions in 10 of 25 malignant infarction patients that underwent surgical decompression, and concluded that surgical treatment provided no benefit¹⁸⁾.

Since involvement by an ICA territory infarction is wider than that of a MCA territory infarction and cerebral edema is more severe, it is difficult to predict the efficacy of decompressive surgery. In the present study, we studied the usefulness of surgical decompression in patients with an ICA territory infarction and sought to identify prognostic factors.

All 17 patients underwent decompressive craniectomy and duraplasty. Eight patients were men and nine were women. Average patient age was 61.53±11.64 years (range from 53 to 77 years). NIHSS on admission at the department of neurology was 10.71±11.44 (range from 1 to 38) whereas NIHSS immediately before surgery, was 25.53±11.79 (range from 7 to 39). Eight patients (47.05%) showed anisocoria and all patients appeared to have hemiplegia. Eight patients (47.05%) had infarctions in the dominant hemisphere, and nine patients in the non-dominant hemisphere (52.94%).

The mean degree of midline shift in CT scans taken before surgery was 7.47±3.55 mm (range from 2 to 13 mm). Mean time from symptom development to surgery was 43.29±30.41 hours (range from 5 to 96 hours). No surgically related complication occurred. Two patients (11.8%) were moderately disabled within one month after surgery and six (35.2%) presented in a severely disabled state. Nine patients (52.9%) died within two weeks of surgery due to cerebral edema progression (Table 1).

Large hemispheric infarctions due to MCA or ICA occlusion constitute a major cause of severe morbidity and mortality⁴⁰⁾. Cerebral edema develops as an immediate consequence of cerebral infarction and its severity is closely related to degree of cerebral infarction²⁴⁾. Severe, life-threatening brain edema occurs in about 10% of patients with a large hemispheric infarction, and the majority of these patients succumb to an increase in intracranial pressure and subsequent uncal herniation²⁶⁾.

Hacke et al.⁹⁾ defined 'malignant cerebral infarction' as a subtype of stroke, that presents itself clinically with severe hemispheric stroke syndrome and almost always indicates death due to herniation, despite maximum medical treatment for identified brain edema. When this clinical presentation is accompanied by early CT signs of major infarct during the first 12 hours after stroke and a distal internal carotid artery or proximal MCA plus anterior cerebral artery occlusion, massive hemispheric swelling occurs during the subsequent 24-72 hours⁸⁾. Furthermore, most of these patients rapidly deteriorate within 2-4 days^18,32,35). In cases of malignant cerebral infarction, mortality is exceptionally high and severe neurological deficits may develop despite conservative treatment, including intensive care, head elevation, sedation, hyperventilation, osmotherapy, and hypothermia treatment, in an intensive care unit^2,6,9).

Decompressive craniectomy was first reported in the 1950's as a more aggressive treatment for malignant cerebral infarction¹⁹⁾. The number of positive outcomes after decompressive craniectomy in malignant cerebral infarction increased since brain CT became widely available in the 1980's^{3,6,8,17,30,32)}. Forsting et al.⁸⁾ reported that the decompressive craniectomy per se increases leptomeningeal collateral and cerebral perfusion pressures, and not only lowers mortality but also reduces cerebral infarction size and improves prognosis. Furthermore, Bendszus et al.¹⁾ in a perfusion CT study, reported that ischemic changes in areas that were initially unaffected by infarction were reduced after decompressive craniectomy.

Surgical decompression plays a certain role in treating malignant cerebral infarction, as it lowers the mortality rate and improves functional outcomes, but most studies on the subject addressed MCA territory infarction. Acute ICA infarction is recognized to be a rare, but critical disease, the outcomes of patients with a cerebral infarction due to an acutely occluded ICA are poor. In fact, only 2-12% achieve a good outcome, 16-55% succumb to a complication, and 40-69% are severely disabled²³⁾. Trouillas et al.³⁴⁾ in a multivariate regression analysis of 100 patients, concluded that proximal ICA occlusions are associated with a poor outcome. Although the natural history of patients with a cerebral infarction due to an acutely occluded ICA is known to be dismal, its clinical outcomes can take many different forms that are dependent on the collateral circulation. Powers et al.^27,28) considered that the collateral circulation was critical to maintain adequate cerebral perfusion in patients with an ICA infarction, and several studies have shown that adequate collateral circulation may prevent the progression of hemodynamic failure^11,15,37). For these reasons, an ICA infarction and ICA territory infarction should be regarded as different entities.

Although mortality among patients with ICA territory infarction is generally higher than among patients with an isolated MCA territory infarction (Table 3), few have reported associated factors and outcomes of decompressive surgery in patients with an ICA territory infarction. Kilincer et al.¹⁸⁾ reported that three of 10 patients that underwent surgical decompression for an ICA territory infarction died and seven survived with severe disabilities, and concluded that surgical treatment for an ICA territory infarction was unhelpful, except in young patients, or when the infarction affected only the nondominant hemisphere. Walz et al.³⁸⁾ also reported that of eight ICA territory infarction patients, three died, one remained in a severely disabled state, and four remained in a moderately disabled state. Chen et al.⁴⁾ found that the factors associated with higher mortality were an age of ≥60 years, involvement of more than one vascular territory, and signs of clinical herniation before surgery. The authors recommended that patients with an infarction involving more than one vascular territory be viewed as unsuitable for decompressive surgery. Our results are broadly similar to those mentioned above, as nine of our 17 patients died, three remained in a vegetative state, and three remained in a severely disabled state. Furthermore, the mortality rate was 53% and 82% achieved a poor outcome. Unfortunately, the studies by Walz et al.³⁸⁾ and Chen et al.⁴⁾ did not identify any meaningful risk factors of poor outcomes. In the present study, we selected patients with an ICA territory infarction and evaluated prognoses using prognostic factors of malignant cerebral infarction.

Many factors that affect surgical outcome in cases of malignant cerebral infarction have been reported^18,29,39). Kilincer et al.¹⁸⁾ concluded that an advanced age (>60 years), preoperative midline shifts ≥10 mm, Glasgow Coma Scale (GCS) of ≤7, preoperative anisocoria, neurological deterioration within three days of stroke, and ICA territory infarction are factors of a poor prognosis. Park et al.²⁵⁾ reported that diabetes mellitus, a dominant-hemisphere infarct and a low preoperative GCS of ≤7 were poor prognostic factors for severe brain infarction.

Age could be a key prognostic factor of surgical outcome in cases of malignant cerebral infarction. Wijdicks and Diringer⁴¹⁾ examined mortality among 42 MCA infarction patients, and found a mortality rate of 28% for patients ≤45 years and of 90.9% for those ≥45 years. In addition, Harscher et al.¹⁰⁾ reported that the mortality rate could be lowered to 20%, if surgery was performed early in patients under 50 years of age. In the present study, age was found to have a negative effect on outcome.

Many researchers have argued that early surgery might reduce the mortality rate, because delayed surgery could cause ischemic injury in the brain stem and worsen the prognosis^3,9,20,31). Carter et al.³⁾ argued that early surgery has better outcomes than late surgery, but Harscher et al.^10), Walz et al.^38), Holtkamp et al.^13), and Cho et al.⁵⁾ reported no significant correlation between time from onset to surgery and outcomes. In the present study, the timing of surgery was not significantly different in the two groups.

Relations between surgical outcomes and preoperative neurologic conditions are also controversial. Steiger³³⁾ reported that of early neurologic findings, degree of motor paralysis and decreased mentality most importantly determine prognosis. Mattos et al.²²⁾ found that a GCS score of ≤8 before surgery was associated with poorer prognosis, but Walz et al.³⁸⁾ found that the surgical outcome of malignant MCA infarction was not related to initial NIHSS. In the present study, average preoperative NIHSS in group B was significantly lower than in group A (p=0.019). Thus, it seems that a lower preoperative NIHSS might be a positive prognostic factor for a good outcome.

As relations between functional outcomes and dominancy, Harscher et al.¹⁰⁾ reported no significant difference between left and right hemispheres, whereas Wang et al.³⁹⁾ argued that prognosis is poorer in older patients and in patients with a dominant hemisphere infarction. In the present study, no significant difference in dominancy was found between the two groups.

Anisocoria could be a referential factor for determining the merits of surgical decompression. Cho et al.⁵⁾ reported poor functional outcomes in anisocoric patients with a loss of light reflex, and thus, recommended that surgery be scheduled before light reflex is lost. In contrast, Rabinstein et al.²⁹⁾ found that a change in pupil size was not related to prognosis. In the present study, no significant prognostic difference was found with respect to the presence of anisocoria, but we cannot exclude the possibility that dominancy and the presence of anisocoria are not related to prognosis because our sample size was small and both patients in group B had a non-dominant infarction and no anisocoria.

Kilincer et al.¹⁸⁾ studied the relationship between midline shifts in pre-surgical CT scans and prognosis, and reported that a midline shift of ≥10 mm is related to a poor outcome. Huh et al.¹⁴⁾ also reported that prognosis was poorer when the midline shift was ≥11 mm. On the other hand, Rabinstein et al.²⁹⁾ found no relation between midline shift and functional outcomes, and similarly, in the present study, we found no significant difference between midline shifts in our two groups.

Wartenberg⁴⁰⁾ reported that the only valid predictor of a malignant course seems to be an infarction size of ≥50% of the MCA territory by CT and ≥145 mL on diffusion weighted images obtained within 14 hours of stroke onset. In present study we considered broader infarction areas, and in order to evaluate the nature of the relation between degree of cerebral infarction and surgical efficacy, we included in our analysis cerebral infarction fraction before surgery, the ratio of area increase caused by brain herniation after surgery, and midline shift. Cerebral infarction fraction in group A was 33.67±7.53% (range from 24.47 to 47.43), which was significantly smaller than in group B, 23.72±2.53% (21.93, 25.51) (p=0.017). This finding suggests that clinical results are better for smaller infarction fractions for ICA territory infarction. Decompression ratios in group A and B were 3.25±3.01% (range from 0.13 to 10.89) and 1.58±1.62% (0.43, 2.72), respectively, which were not significantly different (p=0.345). Although the decompression ratio was larger in group A, the prognosis was poor. This may be due to broader area of infarction and more severe neurological damage.

Our results show that the results of decompressive surgery for an ICA territory infarction are unsatisfactory, and that preexisting prognostic factors, such as, age, time to surgery, presence of anisocoria, and degree of midline shift are not obviously associated with outcome. Nonetheless preoperative NIHSS and infarction fraction were found to be viable prognostic factors that could be helpful when considering surgery.

However, we confess that our findings are compromised by small cohort size, and thus, we suggest that a substantially larger-scale prospective study be conducted.

We believe that decompressive craniectomy reduces mortality and improves functional outcome in cases of malignant cerebral infarction. However, the mortality rate of ICA territory infarction is high despite maximal medical treatment and surgical decompression and most survivors have severe disabilities. Therefore, when determining surgical treatment in patients with an ICA territory infarction, we recommend, in addition to prognostic factors, such as, preoperative NIHSS, and infarction fraction, that quality of life, and caregiver's burden be carefully considered.