Should Patients with Traumatic Brain Injury with Significant Contusions be Treated with Different Neurointensive Care Targets?

In moderate to severe traumatic brain injury (TBI), management is guided toward emergency evacuation of traumatic lesions with mass effect [1] and neurointensive care to reduce the development of secondary brain injury [2‐4]. Traditional neurointensive care aims at keeping systemic and cerebral physiological variables within fixed targets, e.g., intracranial pressure (ICP) below 20–22 mm Hg and cerebral perfusion pressure (CPP) between 60 and 70 mm Hg [2, 3, 5]. Although this approach has led to tremendous improvements in survival and functional recovery [3, 6], fixed treatment targets for all patients may be too crude, considering the heterogeneous nature of the disease [7]. Thus, treatments based on patient-specific demographics, advanced software analysis of neuromonitoring data, and classifications of intracranial injury patterns could be a way toward precision medicine in neurointensive care [2]. One suggested approach in this direction is to adapt CPP management based on the cerebral pressure autoregulatory status in each patient [8‐12]. The cerebral autoregulatory status can be continuously monitored as the correlation coefficient between ICP and arterial blood pressure (ABP) [12‐15], such as the pressure reactivity index (PRx) [14]. The cerebral autoregulatory status differs between patients and changes over time after TBI [12, 16]. Some studies have found that the optimal absolute CPP interval may differ depending on the cerebral autoregulatory status [13, 17]. For example, patients with pressure passive cerebral vessels and impaired cerebral autoregulation may benefit from slightly lower CPP to avoid hyperemia and worsening of brain edema [13]. In addition, PRx often varies with CPP in a U-shaped way, corresponding to the Lassen curve, and the CPP with the lowest PRx (best cerebral pressure autoregulation) has been suggested to be optimal (CPPopt) [10, 11, 18, 19]. Deviations from CPPopt (ΔCPPopt = actual CPP − CPPopt) has been associated with worse brain oxygenation [20, 21], disturbed energy metabolism [22], and unfavorable outcome [10‐12] in observational studies, and CPPopt was shown to be a safe and feasible target in a prospective trial [9].

Furthermore, patients with TBI exhibit a wide array of one or more traumatic intracranial lesion features [1, 7, 23]. This may be of relevance for treatment in neurointensive care because the optimal physiological targets may differ among patients depending on the intracranial injury pattern. In an earlier pilot study, we found that patients with a focal brain injury benefited from lower CPP levels, whereas the opposite was found for those with a diffuse injury if the cerebral pressure autoregulation was disrupted [24]. However, these analyses were limited by a small patient cohort (n = 107), and a crude classification into diffuse vs. focal injury according to the Marshall grade [25] was used. More studies are warranted concerning whether CPP management should be individualized according to the type of brain injury and the significance of cerebral autoregulatory status in different injury types. In particular, high CPP is likely more dangerous in patients with cerebral contusions because they suffer from significant blood–brain barrier disruption [26] and more easily develop brain edema [27]. At the same time, it is possible that the global variables of cerebral pressure autoregulation (PRx and CPPopt) may be insensitive to the pericontusional autoregulatory status and blood flow disturbances that may occur in the presence of a significant focal cerebral contusion [19].

To explore further ways of precision approaches of neurointensive care, we investigated the role of absolute CPP, PRx, and ΔCPPopt in relation to functional outcome for patients with TBI with significant contusions in relation to those with only small or no contusions. We hypothesized that high CPP would have a stronger association with unfavorable outcome in patients with significant contusions and that the global cerebral autoregulatory variables, PRx and CPPopt, would only be weakly associated with outcome in this TBI subtype because of the focal nature of the injury.

In this study, the main findings were that patients with TBI with significant cerebral contusions recovered more favorably if CPP was below 70–80 mm Hg, and even values below 60 mm Hg could overall be tolerated. However, elevated PRx and ΔCPPopt were only weakly associated with outcome in this TBI subtype. On the contrary, in patients without significant contusions (i.e., those with diffuse injury, focal extra-axial hemorrhages, or smaller contusions), PRx and ΔCPPopt were independently associated with outcome. Our findings suggest that PRx and CPPopt, which reflect the global cerebral pressure autoregulation, may be most relevant in patients without large contusions but less valid in patients with predominant focal brain lesions. In the latter group, focus should rather be to avoid high absolute CPP levels. However, these findings are hypothesis generating and need to be validated in prospective trials before implementation into clinical practice.

The association between the cerebral physiological variables and outcome differed between patients with TBI with significant contusions and those with small/no contusions. In the TBI cohort with significant contusions, there was a clear transition toward unfavorable outcome when CPP exceeded 80 mm Hg, as illustrated in the heatmaps. This finding may reflect that these patients were particularly vulnerable to develop detrimental brain edema with high CPP [26], whereas CPP slightly below 60 mm Hg could usually be tolerated [35]. It was clear in these heatmaps that the association between CPP and GOS-E did not change depending on the concurrent PRx. This fact could be explained by the fact that PRx is a global measure of cerebral pressure autoregulation and, thus, may not be sensitive to the regional autoregulatory and blood flow disturbances near the focal lesion [36]. Although PRx and ΔCPPopt were not independently associated with clinical outcome in the multiple logistic regressions, there was a trend toward unfavorable outcome in the PRx/ΔCPPopt plot for these patients when ΔCPPopt was above 10 mm Hg in combination with PRx above 0.25. This combination of PRx/ΔCPPopt likely reflected when the upper limit of autoregulation was exceeded and hyperemia took place in the brain [37]. This finding suggests that the interpretation of PRx/ΔCPPopt combinations still may be of some clinical value in this TBI subtype. Otherwise, a higher %GMT of ΔCPPopt ± 5 mm Hg was only marginally associated with better outcome and did not hold as an independent variable in the multiple regressions in these patients. Altogether, our results indicate that patients with large cerebral contusions exhibit a significant focal injury and that global metrics of the cerebral autoregulatory status, including PRx and CPPopt, may then be less valid. Instead, in this particular subgroup, it seems that elevated CPP should be avoided because it may worsen brain edema. These ideas are consistent with the Lund concept [4]. Although the role of individualized therapy based on global cerebral autoregulatory variables appears less promising in these patients, it is possible that CPP management could still be further individualized by other means, such as focal neuromonitoring of brain oxygenation and energy metabolism (microdialysis) [2]. These methods could aid in determining when hypoxia and energy metabolic decompensation start to occur in the contusional penumbra because of low CPP with focal ischemia [2, 35].

For patients with small/no significant contusions, elevated PRx and CPP deviation from CPPopt were independently associated with a lower rate of favorable outcome. In this group, CPP within 60–70 mm Hg was also associated with favorable outcome, but it was weaker and did not hold true in multiple regression analysis. Thus, the role of global autoregulatory metrics may be of greater validity to further individualize neurointensive care management in this TBI subtype. However, as stated previously, this needs to be corroborated in prospective trials.

A strength of this article was the large patient cohort with available high-frequency physiological monitoring data. We also used a novel approach, developed by our group [17], to study combinations of cerebral perfusion variables, including PRx/CPP and PRx/ΔCPPopt, in relation to outcome.

There were also some limitations. This was a single-center study, and our findings are reflections of the cohort demographics, brain injury patterns, and management, which may limit the external validity. The study design was observational and exploratory. The associations found in the results may not be causal but could reflect potential confounding variables, although this was taken into account to some extent in the multiple regressions. Furthermore, it has been questioned if PRx and CPPopt are reliable in the case of an open EVD or post-DC. Using an EVD system with a certain outflow resistance preserves much of the ICP amplitude when the EVD is open and makes the measurements sensitive for very rapid ICP changes, which is a prerequisite for reliable PRx calculations that were based on 10-s averages of high-resolution data. Several studies support that these measures remain valid in these scenarios [38‐40], and we therefore decided not to exclude these patients. In addition, the EVD was typically closed or only intermittently opened in the majority of patients with this monitoring type. The cutoff for significant contusions at 10 mL was chosen to capture a sufficiently large focal lesion that could have a clinical impact without necessarily requiring surgery. The cutoff was chosen a priori and was similar to the one used in the STITCH trial [31]. Ten milliliters was also deemed as the lower end of what was considered a significant contusion in a recent Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI) study [41]. We considered studying patients with isolated contusions without other concurrent lesions, e.g., aSDH and tSAH, but this would both limit the size of the patient cohort in this study and reduce the external validity of our findings to a minority of all patients with TBI. Although the patients in the significant contusion group also had other lesions, these were typically small. However, in those cases when the concurrent extra-axial lesions were significantly large, they were often evacuated, and the patients then still had their significant contusion injury with leaky vessels, which likely impacted their clinical course. Furthermore, some of the large contusions were evacuated, and these patients were kept in the significant contusions group because the postoperative intracerebral cavity and the pericontusional area were still expected to exhibit leaky vessels. Thus, for all of these reasons, we consider our simple contusion dichotomization justified. In addition, the associations between cerebral physiology and outcome were weak; however, this was expected, taking into account the multitude of factors (e.g., demographic factors, primary brain injury, secondary brain injuries, and rehabilitation) that influence long-term functional outcome [42]. Also, the analyses focused on the most acute phase (first week), although secondary injuries (e.g., due to edema evolution) occasionally occur later. Furthermore, the radiological assessments were performed by only one of the authors (TSW), which potentially may have decreased the accuracy, reliability, and interpretation of the results. However, the volume calculations were automated using the Brainlab software, and the delineation of regions of interests was generally simple, which makes it unlikely that the number of inaccurate measurements would have affected the results significantly. Lastly, the odds ratios of the physiological variables in Table 5 overlapped between the significant contusions and the small/no contusions groups in many cases, which indicates that the differences between these cohorts were relatively small.

In this study, the main findings were that patients with significant cerebral contusions recovered more favorably if CPP was within 60–70 mm Hg, whereas particularly higher values did not appear to be well tolerated. However, elevated PRx and CPP deviation from CPPopt were only weakly associated with outcome in this subgroup. On the contrary, in patients without significant contusions, PRx and ΔCPPopt were independently associated with outcome. Our findings suggest that PRx and CPPopt, which reflect the global cerebral pressure autoregulation, may be of less relevance in patients with predominant large cerebral contusions, whereas these global cerebral autoregulatory variables may be more useful in patients without significant focal intracerebral lesions. Instead, among the patients with TBI with significant contusions, higher CPP, above 80 mm Hg, could be particularly detrimental and should be avoided. However, this was an observational, hypothesis-generating study, and our findings need to be validated in prospective trials before implementation into clinical practice.