Early CT signs in acute middle cerebral artery infarction

CT is widely used for early evaluation of acute strokes. Most importantly, CT excludes acute hemorrhage or other diseases mimicking ischemia (tumor, subdural hematoma). Therefore, CT is the main imaging examination in patients with brain ischemia and when antithrombotic agents are considered. ^[1,2] Although CT does not usually show much in the first 24 hours after cerebral ischemia, ^[3,4] there are early abnormal findings on CT, such as the hyperdense middle cerebral artery sign (HMCAS) and reduced contrast attenuation of the cerebral parenchyma. HMCAS, first described in 1983, reflects arterial occlusion, ^[5-7] and its value in predicting secondary large infarcts is subject to controversy. ^[8,9] Early parenchymal abnormalities, the attenuation of lentiform nucleus (ALN), loss of the insular ribbon (LIR), or hemispheric sulcus effacement (HSE), occur less frequently. ^[10,11] The size, locations, and degree of acute ischemia could influence the time in which these signs appear. Early parenchymal abnormalities might also predict subsequent infarct extension ^[12] and hemorrhagic transformation. ^[13] Finally, initial CT findings may help to predict response to therapy. ^[14] This should have important implications for selecting subgroups of stroke patients in therapeutic trials. ^[15,16]

The aim of our study was to evaluate the frequency of each early CT sign, its value in predicting the topography of subsequent infarct, and its prognostic value for the clinical state of patients with middle cerebral artery (MCA) territory infarct.

Methods.

Among 562 patients admitted to the Neurology unit at the University Hospital of Besancon between January 1989 and December 1990 and who were also part of the Besancon Stroke Registry, we excluded 77 patients with primary hemorrhage, 88 with only TIA, 96 with lacunar syndromes, 72 with vertebrobasilar symptoms, and 37 with multiple infarcts or a previous stroke on the first CT. Among 192 patients with a first ever stroke clinically involving the MCA territory, 123 were admitted within the first 14 hours and 6 had no second CT. We reviewed the first 100 consecutive patients.

All had at least two CT scans, the first one at admission and a control CT demonstrating the MCA infarct. Severity of the clinical features of MCA ischemia was evaluated with a modified version of the Orgogozo scale routinely used in our unit since 1987. ^[17] It allowed us to grade patients from 0 (no neurologic deficit) to 28 (major neurologic disturbances). The onset of ischemic insult was always precisely known. All patients received IV heparin after the first CT.

All noncontrast CT were performed using a GE CT scanner 9800, with 512 times 512 matrix and 2- or 3-second scan time. The section thickness was 3 mm with 5-mm increments from the foramen magnum to the suprasellar region and 10-mm contiguous slices above. All CTs were reviewed by two independent physicians blinded to the patient's clinical data and subsequent infarct location. These physicians were trained to identify early CT signs. When disagreement was observed, the CTs were reviewed by the same physicians together and a decision was made.

The results of the first CT were classified as either no abnormality visible or early abnormalities. These latter signs were defined as follows. HMCAS was defined as a spontaneous high contrast in the MCA. This definition required that the vessel appear not only brighter than adjacent brain tissue but also brighter than the other intracranial arteries, especially the contralateral MCA Figure 1A, and finally that it not be attributable to calcification Figure 1B. ^[8] ALN was defined as a decrease in density involving the lentiform nucleus area inducing the loss of the precise delineation of this area. ^[10] We classified this pattern in two grades according to the degree of reduced density involving the lentiform nucleus Figure 2, A and B: grade 1, slight attenuation (limits of the lentiform nucleus still visible) and grade 2, moderate attenuation (some limits no longer visible). LIR was defined as decreased precision in delineating the gray-white interface at the lateral margin of the insula. ^[11] This aspect was graded according to its anatomic extension: anterior, posterior, or total insular area Figure 3, A and B. As for ALN, we established grading in the attenuation of the insular ribbon as either a slight decrease in the contrast of the gray and white matters (grade 1) or as a disappearance of contrast between these structures (grade 2) Figure 3, C and D. HSE was defined as decreased contrast density inducing a loss of precise delineation of the gray-white interface in the margins of the cortical sulci corresponding to localized mass effect Figure 4, A and B. Early MCA infarct was defined if marked hypodensity involved the MCA territory. Ipsilateral ventricular compression, generally limited to the frontal horn, or contralateral shift of midline structures were also noticed.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Unenhanced CTs showing (A) a left hyperdense middle cerebral artery sign (HMCAS) (arrow) and (B) a calcification in the right MCA (arrow).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. (A) Attenuation of the lentiform nucleus (ALN) sign (grade 1) in the left hemisphere (between arrows). Note the decrease in contrast of the lentiform nucleus but the respect of its limits. (B) ALN sign (grade 2) with loss of precise delineation of the left lentiform nucleus (arrows). In addition, presence of a left posterior loss of the insular ribbon (LIR) (arrowheads) and edema with sulcus effacement involving the left temporal cortex.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. (A) Total loss of the insular ribbon (LIR) (grade 1) in the right hemisphere. Note the slight decrease in contrast of the gray and white matters, but the limits are still visible (white arrows). In addition, presence of an attenuation of the lentiform nucleus (ALN) sign (grade 1) (black arrow). (B) Total LIR in the left hemisphere. A precise delineation between the gray and white matters is not possible (arrows). (C) Posterior LIR (grade 1) in the right hemisphere (arrows); note the perfect delineation on the anterior part of the LIR. (D) Anterior LIR (grade 1) in the right hemisphere (arrows). Note the beginning of a decrease in density involving the insular ribbon (head arrow).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. (A) Hemispheric sulcus effacement (HSE) limited to the left rolandic cortex (arrows). (B) Extensive HSE involving the left frontoparietal cortex (arrows).

A noncontrast control CT was performed within the first 10 days after stroke. Using templates, ^[18] infarct locations were classified, as follows:

1. Large MCA territory, if total superficial and deep MCA territories or the total superficial MCA territory were involved;

2. Multiple MCA infarct, if partial superficial MCA and deep territories were involved;

3. Deep MCA infarct, if the infarct was limited to the deep MCA territory and corresponded to a large lenticulate infarct (>15 mm);

4. Partial superficial MCA infarct, if the superficial (superior, insular, or inferior) MCA territory was involved.

Extended MCA infarct corresponded to type 1 or 2. Hemorrhagic transformations were located and classified according to hemorrhagic infarct (HI) or intrainfarct hematoma (IIH). ^[19] Presumed infarct etiologies were classified into the following groups: large-artery disease (>50% stenosis in the ipsilateral large artery), potential cardiac source of embolism, other etiologies (dissection, hematologic disorders), and undetermined. At hospital discharge, the outcome was evaluated using the Rankin scale. Statistical analysis was performed using descriptive univariate, bivariate analysis with the chi-square and ANOVA tests. The interobserver reproducibility of the early CT signs was determined using kappa statistics.

Results.

There were 48 men and 52 women with a mean +/- SD age of 70.8 +/- 14.2 years (range 31 to 102). Cardioembolism was the main etiology of infarcts (51.4%). Orgogozo scale, Rankin scale, and patients' outcome are reported in Table 1.

Discussion.

There are only occasional reports of a relationship between the presence of early CT signs and the topography of delayed MCA. ^[12,20-22] In our study, the main criterion was the clinical presentation of an ischemia involving the MCA territory. In this situation, we only excluded patients for whom a second CT was not performed, corresponding to less than 5% of our patients admitted for MCA ischemia. In our series, early CT abnormalities were frequently found in all patients (94%) and those admitted within the first 6 hours (88%). Our results were similar to those previously reported that vary from 31 to 92%, depending on the time the CT was performed (4 to 8 hours). ^{[10,11,20-24]} However, our results may also be explained by patients with lacunar syndromes and small deep infarcts, being clinically excluded. In addition, we excluded patients with clinical evidence corresponding to other hemispheric territory involvement (mainly the posterior cerebral territory). Information is scarce concerning patients with a clinical picture mimicking MCA territory involvement but actually with an infarct involving other hemispheric territories. ^[25,26] No such pattern was observed in our series either. Nevertheless, it would also be possible to identify early CT signs of ischemia in these other territories. ^[27,28] A prospective study including all patients with hemispheric (cortical and deep) ischemia should be done, thus allowing an assessment of the value and specificity of early CT signs.

HMCAS is an indirect sign of ischemia that corresponds to an arterial occlusion if its criteria are strictly defined. ^[29] The appearance of unilateral MCA must be more dense than that of its counterpart, covering several millimeters within the first or second segment of the MCA, and must not be caused by calcification; ^[30] the specificity of HMCAS is then very good, ranging from 85 to 100% in most series. ^{[9,20,23,24,31]} However, some patients with clinical MCA ischemia do not exhibit HMCAS despite an angiographically proven MCA occlusion, illustrating a lower sensitivity that varies from 27 to 69%. ^[22-24,29] Our very high interobserver agreement can probably be explained by our using strict criteria that could be well reproducible, although we never performed angiography to evaluate MCA occlusion. In our series, HMCAS was present in fewer patients (22%) than in other series (26 to 50%). ^{[9,20,23,24,29,31]} Our standard CT procedure did not include specific scan sections (less than 5 mm in thickness) on the perisylvian fissure, which may provide a higher positivity rate. On the other hand, although HMCAS usually occurs very early in the timecourse of MCA occlusion, the MCA occlusion is often a transient phenomenon, as suggested by serial CT studies. ^[22] Finally, the value of isolated HMCAS as a predictor of outcome remains controversial. ^[9,31-33] This variation may depend on whether or not HMCAS is associated with other early CT signs, as demonstrated in our study in which HMCAS was never isolated. ^[22]

Early CT parenchymatous signs are defined as a slight decrease in density leading to the loss of precise delineation between the gray and white matter. ^[10,11] They correspond to an early increase in the water component in brain cells, already present 1 hour after the onset of cerebral arterial occlusion. ^[34,35] Moreover, CT attenuation is linearly proportional to the water volume in the brain parenchyma. For each 1% change in water content, a change of 2.6 Hounsfield units on CT measurement occurs. ^[36] The visualization of infarction is obvious when the hypodensity corresponds to about 20 to 30 Hounsfield units. ^[37] The early decrease in density is probably related to infarction in progress and, theoretically, could predict irreversible brain damage. ^[38] It may also depend on the location of the arterial occlusion and the duration of ischemia, especially in more sensitive regions accounting for the location of the early parenchymatous signs.

Tomura et al. ^[10] first described the ALN sign as slight hypodensity recognized even as early as 1 hour after the onset of ischemia. The deep MCA territory seems extremely sensitive to ischemia, because lenticulostriate arteries supply end zone territories and account for the ALN sign. In our series, ALN was well correlated to a subsequent deep MCA infarct (positive predictive value 90% and negative predictive value 98%).

Truwit et al. ^[11] described LIR as a hypodensity involving the insular region. The blood supply to the insular region is mainly provided by the claustral arteries arising from the M₂ segment of MCA, which also supply the extreme capsule, the claustrum, and the external capsule. ^[39] In MCA occlusion distal to the lenticulostriate arteries, this territory would be the farthest from the potential collateral flow rising to the anterior or posterior cerebral artery. The insular region would, therefore, become a watershed arterial zone. ^[11,40] However, paradoxically, this region would have a multiple arterial supply. ^[41] In our series, LIR had the lowest specificity (76%) and positive predictive value (81%); this could probably be explained by the difficulty in carrying out an accurate analysis. Indeed, a clear identification of the insular region requires a very good CT section and proper positioning of the patient. In addition, the analysis of the insular ribbon should include both the appearance (contrast attenuation) and extension in the anteroposterior axis. In our study, a partial LIR was related to subsequent partial MCA infarct involving either superior or inferior MCA branches, depending on whether the LIR involvement was anterior or posterior. The proposed mechanism of this selective phenomenon is occlusion of only one MCA branch. ^[42,43]

Reports are rare of HSE as an early ischemic sign in isolation, ^[10,11,13] as in our series, where it was found in 17% of the patients, corresponding to previous report. ^[22] This sign had a high specificity and positive predictive value (100%) in subsequent cortical MCA involvement. A grading according to its extension may determine further extension of infarct. Although midline shift was seldom found in our series, it was never observed in patients with early CT signs and was always present in those with early infarcts. This sign could correspond to a more intense MCA infarct and may rule out potential aggressive therapy. ^[14,24] Normal CTs were rare in our series (6%), and a subsequent partial superficial MCA infarct and cortical atrophy might have masked early ischemic edema. In addition, these CTs were performed earliest.

Finally, the location, the extension, and the intensity of ischemia could influence the time at which early CT signs would first be detected. For this reason, we found a significant difference between CT timing and the grade of ALN and LIR. In addition, early infarcts, already visible in 7% of our patients, may also be due to a more delayed CT examination (mean 7.8 hours). Another explanation might be less efficacy in the collateral blood supply inducing a more severe ischemia ^[24,42-44] Furthermore, cardioembolic infarcts might also produce much more tissue damage than thrombotic infarcts because the embolus is larger. ^[19] We found a potential cause of cardioembolism in 51.4% of our patients.

Our results confirmed a clear association between both the number and location of early CT signs and the subsequent extension of infarcts. ^[13,22] The three parenchymatous signs alone and associated with HMCAS were significantly correlated with a subsequent extended MCA infarct. However, only one early parenchymatous sign was associated with a limited MCA infarct: ALN with deep territory and HSE with partial cortical territory. Initially, abnormal CT largely underestimates the size and volume of the infarction in progress. ^[24] Therefore, we believe that the number of early parenchymatous CT signs may be a more reliable predictive factor for the subsequent extended infarction. In our series, the number of early CT signs was well correlated to the severity of both initial clinical intensity and outcome; this may also be explained by the intensity of ischemia and a delayed large infarction. ^{[12,14,24,45]} Clinical analysis should be associated with CT evaluation in the first hours after ischemic stroke.

On the whole, hemorrhagic transformation seems to be related to both the intensity of ischemia and the extension of delayed infarct rather than to the presence of early CT signs. ^[23,46] However, two studies demonstrated a close relationship between HI and early CT signs. ^[13,47] We did not find an obvious correlation between early CT signs and hemorrhagic transformation, except for IIH. This contradiction probably depends on the type of classification used for secondary bleeding. ^[19]

Diagnosing early CT signs is often relatively difficult, as the CT findings may be quite subtle. ^[10,21] This difficulty is reflected in the significant interobserver variability in interpreting the initial CTs. ^[24,48] A positive rate of CT signs varies according to the type of study (retrospective or prospective) and the type of analysis performed (blinded or not to clinical data). ^{[21,23,29,31]} Nevertheless, the identification of early CT signs should be improved by experience and the improvement of CT technology. In addition, it requires extensive training to achieve good interobserver analysis. That early CT findings may help to predict infarct extension could have important implications in therapeutic management, and it might allow the selection of subgroups of stroke patients less likely to achieve a satisfactory outcome. ^[14,23]

In summary, early CT signs of ischemia are frequently present, even during the first few hours of ischemic stroke. Early CT signs may allow the prediction of further infarct locations. Moreover, the number and type of early CT signs may also indicate the degree of severity of ischemia. CT provides a simple mean of evaluating early prognosis of MCA infarcts.

Acknowledgments

We thank Dr. Marc Hommel and Dr. Julien Bogousslavsky for their helpful comments.