Commentary on the Article of C. Woertgen, O. W. Ullrich, R. D. Rothoerl, A. Brawanski: Comparison of the Claassen and Fisher CT Classification Scale to Predict Ischemia after Aneurysmatic SAH

We read with great interest the article by Woertgen et al. [3] comparing the well-established Fisher scale [2] and our recently proposed CT rating scale [3] for the prediction of delayed cerebral ischemia (DCI) and infarction from DCI. These authors compared both scales based on a retrospective analysis of a large cohort of SAH patients. This is an important attempt since any newly proposed scale should be evaluated in an independent sample of patients. Their analysis concluded that while both scales were correlated with infarction after SAH, the newly proposed CT rating scale did not provide additional information when compared with the Fisher scale.

In Fisher's original paper, the risk of DCI or infarction did not increase between grade 3 and 4, nor did it in our analysis [3]. Using our modified scale, the risk of DCI or infarction progressively increased with higher scores, indicating that biventricular blood adds independent predictive value when considered separately from thick cisternal blood. In fact, the same was true in Woertgen et al.'s [1] dataset as well, although they also found that Fisher 4 patients had a higher DCI risk than Fisher 3 patients. The Fisher scale when used correctly does not incorporate the extra information about IVH and this is a weakness. We used a multivariate model as a tool to identify the most predictive components for our scale, but the result is a simple scale not a complex model. We do not think that creating multivariate models based on these scales determines how useful CT rating scales are. The point is that this new scale did show in Woertgen's as well as in our original dataset an incremental increase of the risk of developing DCI with each grade on the CT rating scale, which indicates that this may be very useful in predicting DCI.

After reviewing Woertgen et al.'s methodology, however, we have identified a number of issues that may limit a direct comparison of their results with our study. Inclusion criteria differed between the two studies. Woertgen et al. comprised a highly selected subgroup of SAH patients by excluding 39 % of their cohort (185 of 477 patients). They excluded patients without angiographic evidence of an aneurysm (N = 88), patients with infarction on the admission or post-op scan (N = 44), patients with coil embolization of the aneurysm (N = 9), and they do not sufficiently explain why 14 patients were excluded who were operated under deep hypothermia. This represents nearly 39 % of all SAH cases from their series, compared to 8 % from our consecutive series. We feel that our less stringent criteria provides a more clinically useful and widely applicable scale, as early, rapid identification of risk is important in all cases of acute SAH.

Perhaps more importantly, the main outcome variables in the two studies differed quite significantly. Woertgen et al. defined DCI as any new cerebral infarction on the CT, without excluding other causes for infarction except for ictal infarction seen on the admission scan (N = 44) and infarction seen on the post operative scan (N = 16). In contrast, we defined DCI as an otherwise unexplained 1) clinical deterioration (i. e., a new focal deficit, decrease in level of consciousness, or both) or 2) a new infarct on CT that was not visible on the admission or immediate postoperative scan, or both. Because of our study design, we were able to prospectively evaluate, in detail, all potential causes of ischemia, and thus were able to attribute DCI to cerebral vasospasm, as opposed to the more narrow definition of DCI (“new cerebral infarction on the CT scan”) employed by Woertgen et al. We share the authors concern that neurologic deterioration is difficult to detect in poor-grade patients, however, because of the known insensitivity of CT scanning to detect acute cerebral infarction, and the prospective nature of our study, we would expect any retrospective study relying only on CT criteria for DCI to exhibit discrepant findings.

We also analyzed infarction alone, but limited this to infarction presumably due to vasospasm and did not include other causes of infarction, such as herniation. Not surprisingly, our frequencies of “infarction due to DCI” were substantially lower than “any postoperative infarction”, 12 % (Claassen et al. 2001 [3]) vs. 28.5 % (Woertgen et al. 2003 [1]). We assume that this is mainly a difference of definitions and not a true difference in outcome. Regardless, this difference of definitions makes these studies difficult to compare.

Finally Woertgen et al. do not reveal if CT scans were read by physicians blinded to the clinical course and if there was any validation procedure attempted. In future studies comparing different rating scales ROC curves may be helpful to determine the superior discriminating properties.

In summary we feel that studies evaluating the strength of prognostic factors are important, as acute management strategies are developed and refined. In an effort to replicate our findings, Woertgen et al. [1] compiled a highly selected subset of SAH patients and applied the new CT rating scale to predict any infarction during hospitalization, an approach that differed substantially from our methodology. Unfortunately, we feel these methodological differences invite us to reserve judgement on the validity of this new scale until the findings can be replicated or invalidated in another unselected series of SAH patients.

References

• 1 Woertgen C, Ullrich O W, Rothoerl R D, Brawanski A. Comparison of the Claassen and Fisher CT classification scale to predict ischemia after aneurysmatic SAH?.  Zentralbl Neurochir. 2003;  64 104-108
• 2 Fisher C M, Kistler J P, Davis J M. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning.  Neurosurgery. 1980;  6 1-9
• 3 Claassen J, Bernardini G L, Kreiter K, Bates J, Du Y E, Copeland D, Connolly E S, Mayer S A. Effect of cisternal and ventricular blood on risk of delayed cerebral ischemia after subarachnoid hemorrhage: the Fisher scale revisited.  Stroke. 2001;  32 2012-2020