Computed tomographic angiography criteria in the diagnosis of brain death—comparison of sensitivity and interobserver reliability of different evaluation scales

The percents of agreement between the radiologists were equally high (89–95 %) for all scales. The fact that the observers did not reach perfect agreement could be explained by significantly lower contrast enhancement of cerebral arteries in brain-dead patients compared to normal opacification [14]. However, the analysis of interobserver agreement showed a trend towards lower values of Kendall’s tau-b coefficient using the 4-point scale. Interobserver agreement coefficients, including tau-b, are affected by the prevalence of measures under consideration; therefore, low values of tau-b may be due to rare findings, such as negative CTA results using the 4-point scale, may not reflect low rates of agreement. This is a known paradox explained by Feinstein and Cicchetti [19, 20]. In such situations, the percentage agreement is more reliable indicator of interobserver concordance.

Concerning opacification of individual cerebral vessels as indicators of cerebral circulatory arrest, two groups can be distinguished:

In previous studies, different sets of vessels were considered in the diagnosis of cerebral circulatory arrest [8, 21‐26]. The first evaluation scale was introduced by Dupas et al. in 1998 [8]. They observed 100 % sensitivity and specificity of the pericallosal artery, MCA-M4, ICV, and GCV, and proposed the 7-point system, which excluded arteries of the posterior cerebral circulation, because their sensitivity was lower (85.7 % for BA). Delayed opacification of the BA in clinically brain-dead patients was observed in several studies using catheter angiography [27‐31]. Braun et al. reported this phenomenon in 24 (17 %) of 140 patients meeting clinical criteria of BD; these findings were verified by positive TCD in six cases [31]. Only four previous studies assessed opacification of the BA with CTA in BD [8, 21, 23, 25]. Combes et al. reported a 100 % sensitivity of the BA [25]; however, the other authors observed sensitivities that ranged from 86 to 94 %. These later results are consistent with our finding for the BA, 82.9 % sensitivity. Simultaneously, in all cases, catheter angiography showed a lack of opacification of BA in the 30- to 50-s series. In the study by Welschehold et al., all cases were verified by a positive TCD result [21]. These findings show that opacification of BA can be observed in up to 17 % of brain-dead patients and does not indicate preserved cerebral perfusion, thus does not preclude diagnosis of BD. A high frequency of BA opacification in brain-dead patients can be explained by the protective role of the cerebellar tentorium. The majority of primary brain injuries affect the supratentorial compartment; therefore, increasing intracranial pressure (ICP) primarily stops capillary and venous flow to the cerebral hemispheres. In these injuries, ICP may rise more slowly in the infratentorial region, allowing contrast to enter arteries of posterior cerebral circulation. Cessation of flow in posterior cranial fossa can occur with a delay, providing false negative results of blood flow examinations. One of such cases is presented in Fig. 3.

In regard to the diagnostic value of pericallosal artery, the 100 % sensitivity observed by Dupas et al. was not confirmed in other studies that reported a sensitivity of 57–83 % for this artery [8, 21‐25]. In this study, the pericallosal artery had the lowest level among all the vessels evaluated. The shorter distance from the skull base to the pericallosal artery, in comparison with the MCA-M4, can explain the lower sensitivity of the ACA-A3 for the CTA diagnosis of BD. According to Poisseuille’s equation, cerebral blood flow (CBF) is inversely proportional to the length of the vessel.

Therefore, raised ICP primarily stops the flow in the distal cortical segments of the MCA, whereas the contrasts column continues to reach more proximal pericallosal arteries for some time. One such case is presented in Fig. 4.

A high frequency of opacification of ACA-A3 in brain-dead patients is the principal cause of the low sensitivity of CTA in the 10- and 7-point scales previously reported and confirmed by this study, see Table 4.

In 2006, Leclerc et al. introduced the 4-point scale, which is limited to the most sensitive parameters, the MCA and ICV, and excludes the ACA and GCV [26]. Exclusion of GCV was motivated by the fact that it drains blood partially from the posterior circulation. Therefore, this was a consequence of exclusion of the BA and PCA. In addition, circulation to the basal ganglia is assessed by opacification of the ICVs; therefore, opacification of the GCV, a confluence of the ICVs, does not provide any additional information, despite its high individual sensitivity. This approach was included in the recommendations of the French Society of Neuroradiology. The GCV was included in their earlier, 2007, recommendations, but was excluded from their most recent, 2011, revised recommendations [12, 32]. To summarize, evaluation of CTA in the diagnosis of BD evolved from the 10- and 7-point systems to the more highly sensitive 4-point scale.

It must be noted that respecting the widely accepted concept of the whole brain death, an ancillary test should assess the perfusion of the entire brain including anterior and posterior circulation. This condition is not fulfilled in the 7- and 4-point scales limited to anterior circulation. However, some established ancillary tests do not entirely address this issue either. TCD for the diagnosis of BD consists of assessing blood flow in MCAs and BA, omitting ACAs and PCAs. There are opinions that in planar radionuclide perfusion examination it is difficult to evaluate the perfusion of the brainstem because of overlying parotid gland and other soft tissues [33]. Evoked potentials allow examination of specific areas of interest in the brainstem. Finally, EEG records activity from only the cortical layers immediately beneath the scalp; it does not record from subcortical structures, such as the brainstem or thalamus. It should be considered that ancillary test, even if limited to assessment of the anterior circulation, is always combined with the clinical examination testing brainstem function. Besides, no false positive result of CTA according to 7- or 4-point scale has been reported so far (i.e., CTA confirmed BD, yet the patient survived).

While analyzing a relationship between demographic and clinical features, and the sensitivity of CTA, it was found that craniectomy independently predisposed to false negative results. Preserved intracranial opacification in brain-dead patients with skull defects, particularly craniectomy, was previously reported [22, 23, 30, 31, 34]. Three of such cases are presented in Figs. 3, 4, and 5. This can be explained by regional reduction of ICP caused by a skull defect. Alvarez et al. postulated that intracranial hypertension is nevertheless severe enough to induce cerebral circulatory arrest in these cases [30]. However, the residual cerebral flow at the region of the skull defect is a major limitation of perfusion tests for confirmation of BD. Frampas et al. recommended a minimum observation period of 6 h before CTA in this clinical situation because of the slow increase of ICP [23].

SAH or pseudoSAH appeared to be a second factor that predisposed to false negative CTA results. This finding may be explained by increased baseline density of the subarachnoid space in patients with SAH or pseudoSAH, which can be mistaken for a true vascular opacification. These findings support the observation of Gutierrez et al. that interpretation of CT can be difficult in such circumstances [35].

In many patients, catheter angiography showed more proximal intracranial opacification than did CTA. Dupas et al. observed the same divergence in four of seven cases [8]. A larger volume of contrast was used in CTA than in catheter angiography, but during angiography, contrast was injected intra-arterially, which minimizes the dilutional effect. This discrepancy could have resulted from the sequence in which CTA and catheter angiography were performed. Angiography was performed from 15 min to 3 h after CTA. During this time interval, a slowly rising ICP could further suppress cerebral circulation and stop contrast movement at more proximal levels.

The proposed CTA protocol differs from the standard technique of CTA in the diagnosis of BD consisting of NECT and two post-contrast phases started at 20 and 60 s after injection [36]. The 20- to 60-s protocol was proposed by Dupas et al. in 1998 [8] and was applied in several later studies [21‐26]. However, in clinical practice, particularly in France, it is routinely combined with whole body CT for assessing transplantable organs with a single contrast injection. Such combination requires a larger volume of contrast (120 ml instead of ≈80 ml) and longer injection duration (40 s instead of ≈20 s) in comparison to a standard CTA of the brain.

The technique proposed in the present study is dedicated for assessing cerebral vessels, as whole body CT was not included in the examination. Therefore, a volume of contrast was reduced to 80 ml—a standard for cerebral CTA. A standard flow rate of 4 ml/s was used as well. This gave injection duration of 20 s—twice as shorter in comparison to the 20- to 60-s protocol. While choosing a proper delay of CTA scanning for the diagnosis of BD, one must take into account that in some cases cerebral circulation may be slowed due to intracranial hypertension. A key issue is to cover this delayed vascular opacification with CTA scanning. As perfusion and angiographic studies show, there is a threshold time of approximately 15 s for cerebral circulation to preserve a viability of the brain [28, 37‐41]. With longer circulation time, an opacification of cerebral vessels does not indicate preserved brain perfusion thus does not preclude the diagnosis of BD. Thus, optimal CTA scan for the diagnosis of BD should be performed approximately 15 s after peak concentration of contrast is achieved in the head and neck arteries. This time is 40 s after injection with the injection duration of 20 s used in the present study. It should be noted that the standard 60-s phase is also performed approximately 15 s after peak concentration of contrast in the head and neck arteries, because the standard protocol includes the 40-s injection duration.

Applying the injection duration of 20 s and following the standard protocol with scanning at 60 s after injection would be consistent with assessing vascular opacification over 30 s after peak contrast concentration in the head and neck arteries. This would detect extremely slow cerebral circulation not indicative of preserved perfusion but providing false negative results and decreasing the sensitivity of CTA.

Another difference between the proposed protocol and the 20- to 60-s technique is an abandonment of the early 20-s phase. The early phase is used for assessing of opacification of extracranial arteries as an indicator of sufficient contrast delivery. However, the study by Leclerc et al. showed that extracranial arteries are similarly opacified in the 20-s and in the 60-s phase [26]. They postulated that the scanning protocol could be limited to a single, 60-s phase. Therefore, it was reasonable to assume that scanning with a delay of 40 s will be sufficient to detect opacification of extracranial arteries. The results confirmed these assumptions because all CTA examinations revealed sufficient opacification of the STA. The proposed single-phase protocol can be a feasible alternative to the two-phase technique in the diagnosis of BD, when whole body CT is not simultaneously performed. In comparison to the two-phase protocol, the single-phase method is simpler to perform and requires a lower volume of contrast, which is less harmful for the kidneys. However, the accuracy of both protocols needs to be comparatively evaluated.