Parenteral diclofenac infusion significantly decreases brain-tissue oxygen tension in patients with poor-grade aneurysmal subarachnoid hemorrhage

Prospectively collected data of 29 consecutive aSAH patients with brain multimodal neuromonitoring between September 2010 and March 2012 were retrospectively analyzed. Twenty-one patients receiving parenteral diclofenac infusion during the neuromonitoring time were included (see Additional file 1). Criteria for invasive neuromonitoring were approved by the institutional review board of Innsbruck Medical University (UN3898 285/4.8) as follows: (a) GCS ≤8, (b) poor likelihood for regaining consciousness within the following 48 hours, (c) high likelihood for surviving at least 48 hours, and (d) age older than 18 years. Written informed consent was obtained according to federal regulations.

In general, patient care conformed to guidelines of the American Heart Association [18]. Patients were clinically graded with the Hunt and Hess scale and radiologically assessed with the modified Fisher scale [19] of the first available cerebral computed tomography (CT) scan. Prophylactic, parenteral continuous nimodipine (1 to 2 mg/hour) was routinely given in all patients. A mean flow velocity >200 cm/second of the basal cerebral arteries assessed with transcranial Doppler sonography (TCD, DWL Doppler-Box system; Compumedics, Singen, Germany) was considered as TCD flow-velocity acceleration. Catheter angiography was performed in patients with TCD flow-velocity acceleration. Patients with angiographic vasospasm were treated with intraarterial nimodipine and vasopressors. All patients included in this study had continuous invasive blood pressure monitoring. Fluid therapy (colloids or crystalloids) were used as first-line, and vasopressors (noradrenaline, phenylephrine, or dobutamine) were used as second-line therapy for maintaining MAP and CPP. Whenever a critical CPP <50 mm Hg was reached, vasopressors were started/increased immediately. In case of unexpected blood pressure decreases, the parenteral nimodipine dose was decreased or stopped.

We did not perform wake-up trials on the patients, following our institutional guidelines. Therefore, delayed cerebral infarction (DCI) was defined as any new infarct appearing on cerebral CT scan that was judged by an independent radiologist to be attributable to vasospasm. Pneumonia was defined as radiologic infiltrate and elevated white blood cell count. Start and finish time of any pharmacologic intervention, including dosage, were exactly documented in the electronic patient data-management system (PDMS; Centricity Critical Care 7.0, General Electric Healthcare Company, IL, USA).

For fever treatment, 75 mg diclofenac-sodium diluted in 100 ml normal saline (Ratiopharm, parenteral diclofenac (Diclobene), 75 mg, drug approval number: 1-19719) was administered intravenously. Starting and ending time points of intervention were exactly documented in the PDMS. Interventions with administration of other antipyretics or invasive cooling 4 hours before or after parenteral diclofenac were excluded (details on the inclusion algorithm are shown in Appendix Figure 1).

Monitoring probes were placed into the hemisphere deemed at greatest risk for secondary injury and either tunneled or fixed by using a triple-lumen bolt. CT scan of the brain was used to check the probe location, usually within 24 hours after implantation. P_btO₂ was measured with a Clark-type probe (Licox, Integra, Germany), intracranial pressure (ICP) with an intraparenchymal probe (Neurovent, Raumedic, Germany). Brain-tissue hypoxia was defined as brain-tissue oxygen tension <20 mm Hg, based on previous studies [14, 21] demonstrating increased odds for poor outcome and metabolic distress below this threshold value. Cerebral microdialysis (CMD) was performed by using a 100-kDa-cutoff microdialysis catheter (CMA-71; CMA/MicrodialysisTM, Stockholm, Sweden) at a perfusion rate of 0.3 μl/min. Samples were collected hourly and frozen at -80°C. Outcome was evaluated prospectively 3 months after aSAH with telephone interview by using the modified Rankin Scale (mRS). Poor neurologic outcome was defined as mRS >4. The study nurse was blinded to the monitoring data and their analysis. Complications (pneumonia, vasospasm, DCI) were stated after adjudication of all relevant clinical information in weekly meetings by the study team (RH, BP, RB, MF, AS, and ES).

The PDMS was used to acquire digital data for ICP, MAP, CPP, P_btO₂ , and body temperature every 3 minutes from the monitoring device (Carescape B650; General Electric Company, IL, USA). Monitoring data were averaged to 10-minute mean values. A 10-minute mean average before intervention was considered a baseline value for a hemodynamic parameter, and 1 hour before intervention was considered a baseline value for the CMD dataset. Time-series data were analyzed with a generalized linear model by using a normal distribution and identity link function. The model was extended by generalized estimating equations (GEEs) by using time after intervention as a factor and important parameters (CPP, body temperature) as covariates.

An autoregressive matrix of the first order (AR-1) was used to handle repeated observations within subjects [22]. Logarithmic transformation was applied to meet assumptions of normality. Cut-off levels were calculated by using receiver operating characteristics (ROCs).

Data are given in median and interquartile range (IQR) or in mean ± one standard error of the mean (SEM).

Baseline characteristics are demonstrated in Table 1. Mean age was 55 ± 11 years, and the median Hunt and Hess grade was 4 (3 to 5, IQR). Neuromonitoring was initiated at day 1 (1 to 2, IQR) and maintained for 12 days (8 to 14, IQR); 29% (n = 6) of patients had poor outcomes.

One-hundred twenty-three interventions were analyzed. A median of four interventions (two to eight, IQR) per patient were administered over a 34-minute period (20 to 45, IQR). Mean central venous pressure at baseline was 13 ± 0.2 mm Hg. Ten days (7 to 12 days, IQR) after aSAH was the median time point of intervention, when fever was most common (30% of daily monitoring time; Figure 1A).

Body temperature at baseline (38.3 ± 0.05°C) decreased by 0.8 ± 0.06°C to a minimum value of 37.5 ± 0.05°C within 330 minutes (260 to 470, IQR) after the start of an intervention (P < 0.001; Figure 1B). In 9% (n = 11) of interventions, temperature did not decrease after parenteral diclofenac. We found a significant interaction between body temperature and MAP (P = 0.02), but no interaction between body temperature and P_btO₂.

MAP (baseline, 93 ± 1.2 mm Hg) and CPP (baseline, 85 ± 1.4 mm Hg) decreased by 10% after intervention (P < 0.001; Figure 2). Maximum blood pressure decreases were observed 160 minutes (90 to 320, IQR) after intervention. Colloids were administered in 33% (n = 41) and crystalloids in 5% (n = 7) 29 minutes (20 to 86, IQR) after intervention. Vasopressors were increased within 75 minutes (30 to 155 minutes, IQR) in 26% (n = 32) of interventions. A decrease of CPP <70 mm Hg and <50 mm Hg was observed in 71% (n = 87) and 12% (n = 15) of interventions, respectively. ICP (baseline: 8.6 ± 0.4 mm Hg) did not significantly change after intervention.

P_btO₂ decreased by 13% from a baseline value of 28.1 ± 2.2 mm Hg to 24.5 ± 2.1 mm Hg (P < 0.001; Figure 2), resulting in brain-tissue hypoxia (P_btO₂ <20 mm Hg) in 35% (n = 43) of all interventions and in 38% (n = 8) of patients. This finding remained significant (P < 0.01) after adjusting for CPP and body temperature, indicating an intrinsic effect of diclofenac on P_btO₂. An interaction occurred between P_btO₂ and CPP (P = 0.02), which is shown in Figure 3. Baseline P_btO₂ <30 mm Hg (Figure 4) was associated with brain-tissue hypoxia after parenteral diclofenac (likelihood ratio, 45; specificity, 81%; sensitivity, 88%; P < 0.001; OR 85.1; 95% CI, 13 to 550), with an area under the curve (AUC) of 93% (95% CI, 87% to 99%). Brain-tissue hypoxia was observed 3 hours (1.5 to 3 hours, IQR) after intervention and persisted for 25 minutes (18 to 180 minutes, IQR). A CPP decrease <70 mm Hg was associated with brain-tissue hypoxia (P < 0.01).

The percentage of total monitoring time with P_btO₂ <20 mm Hg (32 ± 9.4% in patients with good outcome versus 66% ± 12% in patients with poor outcome: OR = 1.04; 95% CI, 1.001 to 1.08; P < 0.05), but not the absolute time (48 ± 20 hours versus 56 ± 27 hours; P = 0.07) was independently associated with poor outcome after adjusting for disease severity.

Cerebral microdialysis (CMD) was available in 48% (n = 10) of patients and 28% (n = 34) of interventions. No significant change in brain metabolism was noted after parenteral diclofenac (Figure 5) when compared with baseline values (CMD-lactate, 4 ± 0.3 mM; CMD-pyruvate, 141 ± 7 μM; CMD-LPR, 28 ± 1.5; CMD-glucose, 2 ± 0.2 mM; and CMD-glutamate, 22 ± 8 μM).