The value of intraoperative intracranial pressure monitoring for predicting re-operation using salvage decompressive craniectomy after craniotomy in patients with traumatic mass lesions

The demographic and clinical characteristics of the patients are shown in Table 1. In total, 284 patients underwent craniotomy for mass lesion evacuation and bone flap repositioning over a period of 7 years. Forty-one (14.4 %) patients underwent SDC for refractory intracranial hypertension. The main reason for re-operation using SDC was regional oedema and progressive traumatic haematomas (82.9 %). The mean GCS was 5.2, ranging from 3 to 8. In the radiological findings of preoperative CT scan, midline shift > 5 mm was observed in 219 (77.1 %) patients. Compressed basal cisterns with close proportion (74.3 %) were observed in 211 patients. SDH was the main type of mass lesion, occurring in 122 (43.0 %) patients. The mean initial ICP in all patients was 30.6 ± 8.7 mmHg, ranging from 13 to 54 mmHg. Figure 1 shows that initial ICP was significantly higher in the SDC group than in the non-SDC group (P < 0.001). The mean dura suture ICP was 9.3 ± 3.4 mmHg. We did not find ICP > 16 mmHg in either group after the dura was sutured.

Over 6 months of follow-up, unfavorable outcomes (GOSE 1–4) were confirmed in 193 (68.0 %) patients. Outcomes were not the principal objectives of the study; nevertheless, we found that both initial ICP (OR: 1.231, P < 0.01) and dura suture ICP (OR: 1.187, P = 0.01) were significantly correlated with unfavorable outcomes. Higher GCS meant mild severity and lower risk of unfavorable outcomes (OR: 0.322, P < 0.001). However, we did not find a significant correlation between SDC requirement and poor outcomes.

On screening 18 candidate variables, Table 2 presents potential correlations between SDC requirement and older age, combined lesions, SDH, lower blood pressure and higher initial ICP. By multivariate logistic regression analysis, the independent risk factors are indentified in Table 3. For every 1-mmHg increase in initial ICP, the risk for the need of SDC increased by 10 % [OR: 1.100; 95 % confidence interval (CI): 1.052–1.151]. The Hosmer–Lemeshow test result (P = 0.434) indicated that the regression model was well calibrated. SDC could be explained by this model (Nagelkerke R²) in 31 % cases. Figure 2 shows the ROC curve established by multivariate analysis. The area under the curve was 0.771 (P < 0.001, 95 % CI: 0.678–0.864).

Correct indication is an inevitable critical issue when we discuss the role of craniotomy or primary DC in STBI treatment. We attempted to evaluate if there is any ideal perspicuous boundary between craniotomy and primary DC and if there is, whether we can quantify the criteria by objective factors. Several authors have tried to evaluate the efficacy of primary DC for STBI treatment in animal models and clinical settings [23‐25]. In retrospective studies, other authors established opposing views, indicating that initial DC does not show any significant advantage compared with craniotomy [26, 27]. However, irrespective of the opinion about DC, as the most widely used surgical treatment, craniotomy has become the costar for DC study. One of the limitations of using craniotomy as the initial surgery for STBI due to mass lesions is the probability of re-operation using SDC on account of postoperative deteriorating brain oedema and progressive intracranial hemorrhage.

In this retrospective study, we summarised the characteristics of SDC after craniotomy. The incidence of SDC in our centre was 14.4 %. In the study by Compagnone et al., 11 % patients underwent initial craniotomy and salvage operations [1]. Considering that their subjects included patients undergoing both craniotomy and DC, the actual incidence of re-operation may be higher. Huang et al. reported that 37.5 % patients required re-operation using SDC in the craniotomy group [28]. In our study, the main cause of re-operation was regional oedema and progressive traumatic haematomas (82.9 %) rather than focal neonatal haematomas or remote site haematomas. This explains why we used DC instead of craniotomy for re-operation to decrease refractory intracranial hypertension and evacuate focal lesions.

Our study demonstrated that initial ICP is an independent risk factor of SDC requirement. Intracranial hypertension is the result of various pathophysiological mechanisms and the reason for subsequent intracranial deterioration. Based on ICP monitoring for postoperative and conservative patient care, some neurosurgery guidelines and consensus have been established for rethinking the techniques. Only a few studies have noted the value of intraoperative ICP. Yuan et al. found that initial ICP has a significant prognostic value for intracranial hypertension in diffusion traumatic brain injuries [29]. Kuo et al. concluded that after haematomas evacuation, patients with intraoperative ICP of <17.5 mmHg would have better outcomes [30]. Tsai et al. considered that the ICP threshold for favourable outcomes should be lowered to <14 mmHg [31]. To the best of our knowledge, our study is the first to evaluate the correlation between intraoperative ICP and subsequent salvage re-operation. Use of intraoperative ICP as the main indicator of evaluation was associated with the characteristics of ICP itself. First, ICP can be determined before the final decision of bone flap disposal in the procedure timeline. Second, it is an objective parameter that can directly weigh the clinical event compared with many other demographic and dichotomous imaging variables. Third, it is easy to use. Our data showed that the average initial ICP in the SDC group was almost 10 mmHg higher than that in the non-SDC group (Table 2). According to multivariate regression analysis, initial ICP increments of 10 mmHg would elevate the risk for the need of SDC 1.6 times. Contrary to the results of initial ICP, we did not find any definite correlation between dura suture ICP and SDC requirement. The maximum dura suture ICP did not increase over 16 mmHg in either group. This indicated that through craniotomy, intracranial hypertension was ameliorated; however, the pathologic damage obviously did not cease by the end of surgery. Some studies have indicated that even if the intracranial hypertension is temporarily reversed, ischemic insult and cerebral oedema could occur subsequently [24, 32].

Multivariate regression analysis indicated that higher age, early hypotension and combined lesions were the independent risk factors associated with SDC requirement. With respect to age, the risk for the need of SDC increased by 3.9 % for every 1-year increase in age (OR: 1.039; 95 % CI: 1.002–1.077). This was consistent with the results of previous studies that reported an age-related increase in risk of additional hemorrhage and brain swelling for degeneration of cerebral microstructure [33, 34]. In our study, to ensure competent physiological background and sound extracranial organ function, we did not include patients >65 years. Considering the increased space owing to cerebral atrophy, we believe that DC would be more complicated in geriatrics. In the presence of early hypotension in patients with traumatic mass lesion, the OR of SDC requirement increased 1.52 times (OR: 2.524; 95 % CI: 1.107–5.756). Low systolic blood pressure (<90 mmHg) can directly aggravate cerebral ischemia and lead to diffuse brain swelling [35]. In our centre, all patients with early hypotension received strict blood pressure management recommended by the guidelines [7]. Obviously, this intervention did not completely erase the negative impact of hypotension. We infer that the pre-hospital duration is a potential contributor to hypotension. Samant et al. concluded that the first 6 h hypotension in STBI correlated with poor outcome [36]. However, we can monitor and manage blood pressure only on arrival of the patient. It is impractical to determine the exact length of pre-hospital hypotension. Another significant factor correlating with SDC requirement was combined lesions (OR: 3.329; 95 % CI: 1.199–9.244). We found that disease progression of patients with combined lesions was different from that of patients with simple contusion or SDH. All patients with combined lesions had broader traumatic areas, which established the conditions for diffuse swelling. After a great volume of SDH was evacuated, the majority of these patients showed a sharp decline in ICP over a short time. We infer that the drastic change of intracranial microenvironment would impair the stability of autoregulation. Without the force of extraparenchymal haematomas oppression, the original contusion and oedema would be more easily aggravated in subsequent treatments under impaired autoregulation [37, 38]. As reported in many previous studies, motor scores, midline shift and compressed basal cisterns were the most powerful independent risk factors for predicting the outcome [19, 29]. In this study, we did not find any significant correlation between these factors and SDC requirement. This indicated that traditional clinical severity indicators or CT parameters should not be simply used to predict the risk for the need of SDC after craniotomy.

Our study demonstrates that preoperative data can be used to evaluate the treatment options for STBI. Among the patients with STBI but without intraoperative brain swelling after mass lesion evacuation by craniotomy, the incidence of re-operation using SDC was 14.4 %. We found that initial ICP was an independent risk factor for re-operation. We also found that there was a significant correlation between SDC requirement and older age, early hypotension and combined lesions. Our results present the statistical inference of clinical data. To further validate the conclusions, a randomised, controlled trial should be performed.