CEREBRAL OEDEMA AFTER EXTENSIVE THERMAL INJURY:
PROGNOSTIC SIGNIFICANCE OF EARLY INTRACRANIAL AND CEREBRAL PERFUSION PRESSURES

Gueugniaud P.Y., Jauff ray M., Bertin-Maghit M., Bouchard C., Durand F., Fonrouge J.M., Petit P.

Burn Centre, Edouard Herriot Hospital, University of Lyons 1, Lyons, France

SUMMARY. Massive fluid resuscitation is required after severe burn injury. This leads to extensive oedema which can affect not only the burned skin but also unburned areas and vital organs such as the heart and lungs. Cerebral oedema has never been systematically investigated in burn patients, and burn encephalopathy remains a well-known but unexplained complication. A prospective study was conducted on early intracranial and cerebral perfusion pressures in severely burned patients with a percentage of total burn surface area (T13SA) over 60% and a unit burn score (UBS) of up to 220. Intracranial pressure monitoring was set up as soon as possible with an epidural screw and recorded over a period of five days. Thirty-two patients were successively included (age = 37 ± 14 years; TBSA = 68 ± 19%; UBS 258 ± 73). Markedly abnormal cerebral pressure data were observed, mainly between days 1 and 3 post-burn: intracranial pressure peak values increased to 31.4 ± 10.4 mm Hg on day 2 post-burn. Cerebral perfusion pressures dropped to critical values also on day 2 post-burn (41.0 ± 10.2 min Hg). Eighteen patients were discharged alive (56%). A comparison between survivors and non-survivors showed that non-survivors always presented higher intracranial perfusion pressures and lower cerebral perfusion pressures than survivors. The significance between survivors and non~survivors over the five days of monitoring was calculated by a variance analysis for repeated measures. The intracranial pressure average was slightly above significant levels (p = 0.06). Cerebral perfusion pressure differences were statistically significant as regards both minimum recorded values (p = 0.009) and average values (p = 0.02), whereas mean arterial pressures were not significantly different (p = 0. 12). These data suggest that very severely burned patients suffer cerebral oedema capable of inducing abnormal intracranial pressure. During this initial massive fluid resuscitation, intracranial hypertension and haemodynamic instability may lead to deleterious cerebral hypoperfusion. It is concluded that the use of therapy for cerebral oedema combined with catecholamines may be useful in the limitation of cerebral perfusion after extensive burn injury.

In 1832 Dupuytren first described %ntensive cerebral congestion in the autopsy of patients dying after burn injury. Since then, many cases of neurological sequelae of burns have been reported, especially in children The concept of burn encephalopathy was thus created and defined as the acute onset of change in sensorium or the occurrence of seizures and abnormal neurological signs, or both, at any time after the initial burn injury.` In 1981, Mohnot et al. reported that burn encephalopathy resulted from "complex metabolic, hematological and hemodynamic abnormalities father than from a single metabolic abnormality. If cerebral oedema was once thought to be an inevitable component of burn encephalopathy, it has rarely been described subsequently, except in burned and asphyxiated children." On the other hand, extensive oedema in severely burned patients can affect not only the burned skin but also unburned areas` and certain vital organs, such as the myocardium` and kings.
We therefore decided, after observing some unexplained neurological complications following burn therapy, to include intracranial pressure (ICP) and cerebral perfusion pressure (CPP) monitoring in our routine intensive care management of major burns.

Patients
In accordance with the care protocol applied in our Centre since 1991, and with the approval of our local ethics committee, we monitored early ICP and CPP in all adult patients suffering extensive thermal injury with a percentage of total burn surface area (TBSA) over 60% (including face burns), with a unit burn score (UBS) index (calculated as % TBSA + 3 x % full-thickness burn area) of over 220, and when the delay between the burn injury and the ICP set-up was less than 24 hours.
Patients with prior vital organ dysfunction or associated injuries (polytrauma, intoxication, lung injury, electrocution, etc.) were not analysed, as the aim of the study was to outline the consequences of isolated burn injury. Serious coagulation problems on admission constituted another criterion for exclusion.

Methods
On admission, a double lumen central intravenous line was introduced percutaneously into the superior vena cava and an indwelling arterial femoral catheter was inserted. All patients were intubated and ventilated in order to obtain adequate deep sedation by means of flunitrazepan and phenoperidine (initially 0.4 mglh and 0.8 mg/h, respectively). Fluid resuscitation was achieved according to Evans'formula with 3-4 ml/kg/% T13SA on day 1 and 2 ml/kg/% on days 2 and 3, with half crystalloids (Ringer) and half colloids (alburnin 4%). In the event of pressure collapse or inadequate urine output in spite of this standard fluid filling, i.e. mean arterial pressure (MAP) < 70 mm Hg and diuresis < 0.5 ml/k,-,/h), the haemodynamic profile was analysed by inserting a blood flow-directed balloontipped pulmonary artery catheter and optimized by introducing catecholamines: dobutamine, when the cardiac index (Cl) was < 3 1/min/m', and doparnine, when the systemic vascular resistances indexed (SVRI) were < 600 dyn.sec/cm5.ml. The catecholamine infusion rate was increased if necessary: if a 20 y/kg/min dopamine infusion proved ineffective to maintain sufficient SVRI, we switched to norepinephrine. In other respects, biological values were maintained: haemoglobin  10 g/dl; Pa02  12 kPa; PaC02 :5 6 kPa; pH 7.3; natraemia 130 mmol/1; proteinaemia 50 g/1.
An epidural screw (Hirsch-Sainte Rose type, 1CP kit, Plastimed, France) was set up as soon as possible and when workable in a nonburn area. Extensive disinfection was effected in the trephine hole with 10% povidone-iodine (BetadineR; Asta Medica, M6rignac, France) achieved before the screw was inserted and after it was removed some days later. Samples were also taken daily during a monitoring system flush for bacterial cultures. 1CP and invasive blood pressure were both monitored continuously. CPP values were calculated directly by the monitoring calculator (Space Labs 9030 A, USA) (CPP = MAP - 1CP). This complete management was performed for at least the first five days of evolution.

Therapy protocol
We considered the cerebral pressure range to be norm al when ICP was !~ 20 turn Hg and CPP over 50 turn Hg. An ICP level over 20 turn Hg was recorded as such if after a monitoring system flush and a controlled zero this level was confirmed at least three times over a period of 15 min without any stimulation of the patient.
Abnormal ICP values were treated first by an IV bolus of 3 mg/kg weight thiopentone, repeated when necessary, followed by continuous thiopentone infusion if the IV bolus was not sufficient to maintain normal ICP. A low CPP level was treated either by an initial continuous 'dopamine infusion starting at 5 y/kg/min, increasing 5 7/kg/min every 10 min if necessary, or by increasing the previous catecholarnine infusion rate. In the event of simultaneous high ICP, an IV bolus of thiopentone was injected during the increasing amine infusion rate when MAP reached at least 70 turn Hg.

Data analysis
1CP and CPP average values representing the average of all hourly values recorded each day are reported. Extreme 1CP, MAP and CPP values, which are the highest 1CP value and the lowest MAP and CPP values recorded each day, are also reported. Data are expressed as mean ± SD. Survivor (S) and nonsurvivor (NS) groups are compared. A paired Student's t test is used for daily comparison of the average continuous variables between the two groups: a p value < 0.05 was considered statistically significant. Analysis of variance for repeated measures is used to compare the monitoring of the two groups over the five days.

Results

Thirty-two successive patients (age, 37 ± 14 years; TBSA burns, 68 ± 19%; UBS, 258 ± 73) admitted to our centre between January 1990 and January 1994 were studied. The first ICP measurements were achieved within a mean of 13.3 ± 8 hours. Fifty-six per cent of the patients (18/32) were discharged alive without neurological sequelae. None of the patients died during the first three days; three patients died between days 4 and 5 because of multiple organ failure. Death occurred after day 10 in eleven patients owing to septic complications (eight patients) and/or cardiopulmonary complications (three patients) unrelated to direct cerebral after-effects. No complications arose as a result of the epidural technique, e.g. haemorrhage or infection: no organism grew in any of the samples during this period. These data must not conceal the potential dangers of infection, especially if the trephine hole is drilled in a burned skin area: however, we probably prevented septic complications because of the strict technique employed (respecting dural integrity) and the wide use of local disinfectant. Table I shows no significant difference between S and NS groups by age, TBSA or time before first 1CP recording. Fluid requirements during the first two days were marginally higher in the S group than in the NS group (3.8 vs 3.2 ml/kg/% TBSA on day 1; 2.5 vs 2.2 ml/kg/% T13SA on day 2). Abnormal cerebral pressures values were clearly found throughout the first three days post-injury in all patients (Table II).

The highest pathological values were observed on day 2: 1CP maximum values, 31.4 ± 10.4, 1CP average, 18.1 ± 6.5 mm Hg; minimum CPP, 41.0 ± 10.2, CPP average, 59.0 ± 8.4 mm Hg.
A comparison of these values in S and NS groups is presented in Table III. During the first week postinjury, ICP values were always higher and MAP and CPP values lower in the NS group than in the S group. The difference is statistically significant for minimum CPP (p = 0.009) and CPP average (p = 0.02) (Figs. 2 and 5). The significance is borderline for ICP average (p = 0.06) (Fig.4). ICP peaks and minimum MAP values were not significantly different (respectively 0,11 and 0.12) in either group (Figs. I and 3).
If we consider the daily comparison for each variable, we can identify a significant difference on each day: on day 1, minimum CPP (Fig. 2); on day 2, minimum CPP and minimum MAP (Figs. 2 and 3); on days 3 and 4, ICP average (Fig. 4); on day 5, maximum ICP (Fig. 1), minimum CPP (Fig. 2), and CPP average (Fig. 5).

Fig. 1 - Temporal patterns of maximal inuacianial pressures (ICP) in sin v ivors (S = dashed line) and non-survivors (NS = solid line). Vertical Imrs represent SR A paired Student's t test is used to compare the daily average variables between the two groups: * = p < 0.05. Differences over all days of monitoring are calculated by analysis of variance for ic[)eated measures: p = 0.11. Fig. 2 - Temporal patterns of minimal cerebral perfusion pressures (CPP) in survivors (S = dashed line) and non-survivors (NS = solid line). Vertical bars represent SD. A paired Student's t test is used to compare I lie daily average variables between the two groups: * = p < 0.05; ** = p - 0.02. Differences over all days of monitoring are calculated by analysis of variance for repeated measures: p = 0.009.

Fig. 3 - Temporal patterns of minimal mean arterial pressures (MAP) in survivors (S = dashed line) and non-survivors (NS = solid line). Vertical bars represent SEl. A paired Student's t test is used to compare the daily average variables between the two groups: * = p < 0.05. Differences over all days of monitoring are calculated by analysis of variance for repeated measures: p = 0. 12. Fig. 4 - Temporal patterns of daily intracranial pressure (ICP) average in survivors (S = dashed line) and non-survivoys (NS = solid line). Vertical bars represent SD. A paired Student's t test is used to compare the daily average variables between the two groups: * = p < 0.05. Differences oveT all days of monitoring are calculated by analysis of variance for repeated measures: p = 0.06.

Fig. 5 - Temporal patterns of daily perfusion pressure (CPP) average in survivors (S = dashed line) and non-survivors (NS = solid line). Vertical bars represent S1). A paired Student's t test is used to compare the daily average variables between the two groups: * = p < 0.05. Differences over all days of monitoring are calculated by analysis of variance for repeated measures: p = 0.02.

During this period, all patients received at least one catecholamine to improve the haemodynamic profile. Dopamine was specifically indicated for cerebral hypoperfusion and allowed to increase CPP in seven patients (three in the S group and four in the NS group). Twentyfive out of the 32 patients needed at least one thiopentone IV bolus, and continuous thiopentone infusion was necessary to treat ICP levels over 20 mm Hg in 16 patients (six in the S group, ten in the NS group). In spite of this planned therapy protocol, intracranial pressure was incompletely controlled in patient in the NS group.

Discussion

The first study of neurological complications in burns dates from 1832, when Duptlytren described %ntensive cerebral congestion" in the autopsies of burned children. ifle causes of these neurological disturbances were initial ly attributed to cerebral oedema," demyelination' or "lc)xins",21 and more recently to complex metabolic, hae matological and haemodynamic abnormalities.` The incidence of burn-related encephalopathy has varied from 1 to 85%. Nowadays, numerous aetiologies are discussed: hypoxia, hypovolaemia, acidosis, hyponatraemia, hyperglycaemia, infection, etc.
During the last year in our centre, we have observed several central system dysfunctions after severe burn injury and prolonged intensive care: various neuropsychiatric manifestations, convulsions, and three cases of severe neurological deficit with quadriplegia in which a diagnosis was made of central pontine myelinolysis.` This burn-related complication has long been known? It may be linked to cerebral oedema. 4,22 Moreover, in the first days postinjury and during the necessary initial fluid overload, oedema in severely burned patients is extensive, affecting unburned areas and some vital organs.
In the light of this evidence, we suspected the occurrence of brain hypertension due to cerebral oedema, which could be involved in burn encephalopathy.' In our study, 1CP and CPP abnormalities were observed in the first three days post-injury, particularly on day 2 (respectively 31.4 ± 10.4 and 41 ± 10.2 nim Hg). During this period, 1CP peaks and CPP reductions may be deleterious for the brain. The drops in MAP linked to the usual initial haemodynamic disturbances in critically burned patients lead to a precarious and even life-threatening cerebral perfusion. The average daily values are obviously less meaningful because of the therapies systematically administered in order to correct the abnormal values. When we compare the data according to outcome, abnormal 1CP and CPP values are still recorded in both groups. The study nevertheless shows that survivors' 1CP and CPP values were always nearer the norm than those of non~suirvivors. The fact that all patients received catecholamine infusion and that 78% of them required thiopentone to normalize 1CP could be due to mean pressures being in the upper normal range and sometimes similar in both groups. We realize that the lack of a control group in our study makes these values questionable. An analysis of variance for repeated measures calculated throughout all the days of monitoring shows that CPP differences between the S and NS groups were statistically significant for both minimum recorded values and average values. The 1CP average difference was of borderline significance, while 1CP peak differences and minimum MAP differences were not significant. In other respects, when we compare variables day by day, we find altemalively significant differences for all values, but we note that three variables appear significantly different on the same day, i.e. day 5. At this point in the evolution, which represents the end of the acute post-injury phase, survivors seem to be able to normalize intracranial homeostasis better than nonsurvivors: these results suggest that variations in 1CP and especially CPP could at least reflect survival chances. Nevertheless, if the two groups are not statistically different, one must be cautious because a difference of 14 years between the two groups may well increase the difference in mortality curves. Burn patient prognosis can be improved by increasing fluid filling, according to Miller et al.or by increasing Cl and oxygen delivery and oxygen consumption, according to others.In this study, cerebral oedema was not directly related to fluid overload. The explanation of cerebral oedema occurring at some distance from the burned areas is still unknown: the role of cytokines, which are involved in burn mortality, needs to be discussed,` particularly that of interleukin-6, which may considerably increase between days 2 and 4 post- injury. In conclusion, the ability to maintain cerebral pressures may itself represent a favourable prognostic factor in severely burned patients, even if we cannot affirm that 1CP monitoring and the control of cerebral pressures by specific therapy improve survival.

1. Cobb S., Lindemann E.: Burns - neuropsychiatric observations. Ann. Surg., 117: 814-24, 1943.
2. Hugues L.: Burns with cerebral complications. Med. J. Aust., 2: 122, 1947.
3. Lindsay W.K., Murphy E.G., Birdsell D.C.: Thermal burn encepha lopathy. Can. J. Surg., 8: 165-71, 1965.
4. Emery J.L., Reid D.A.C.: Cerebral oedema and spastic herniplegia following minor burns in young children. Br. J. Surg., 50:53-6,1962.
5. Haynes B.W., Bright R.: Burn coma: a syndrome associated with severe burn wound infection. J. Trauma, 7: 464-75, 1967.
6. Hall K.V.: Encephalopathy complicating burn injuries in children. Scand. J. Plast. Reconstr. Surg., 2: 59-63, 1968.
7. Warlow C.P., Hinton P.: Early neurological disturbances following relatively minor burns in children. Lancet, 2: 978-82, 1969.
8. Rosenbloom C., Kravath R.: Neurological disturbances following minor burns. Lancet, 2: 1423, 1969.
9. Mettler F.A.: Burn encephalopathy as a diagnosis. J. Med. Soc. N. J., 71: 817-23, 1974.
10. Antoon A.Y., Volpe J.J., Crawford J.D.: Burn encephalopathy in children. Pediatrics, 50: 609-16, 1972.
11. Prekop R.: Brain edema in burned children. Acta Chir. Plast., 26: 184-92, 1984.
12. Roth N.: Encephalopathy due to burns. Arch. Neurol. Psych., 45: 980-3, 1941.
13. Bender L.: Burn encephalopathy in children. Arch. Pediatr., 60: 70 87, 1943.
14. Mohnot D., Snead O.C., Benton J.W.: Burn encephalopathy in children. Ann. Neurol., 12: 42-7, 1982.
15. Kay S., Siva Rao G.S., Lord D. et al.: Intracranial pressure monitoring as an aid to resuscitation in the burnt and asphyxiated child: three case reports. Burns, 12: 212-3, 1986.
16. Demling R.H.: Fluid resuscitation after major burns. JAMA, 250: 1438-40, 1983.
17. Cioffi W.G., Demeules J.E., Gamelli R.L.: The effects of burn injury and fluid resuscitation on cardiac function in vitro. J. Trauma, 25: 638-42, 1986.
18. Shoemaker W.C., Vladeck B.C., Bassin R. et al.: Burn pathophysiology in man: 1. Sequential hemodynamic alterations. J. Surg. Res., 14: 64-73. 1973.
19. Kruse F.: Enzephalitis und Amauruse nach Verbrennung. Dtsch. Med. Wochenschr., 54: 1029-41, 1928.
20. Madow L., Alpers B.J.: Brain changes in patients with extensive body burns. Arch. Neurol. Psychiatry, 72: 440-51, 1954.
21. Kaye E.M., Butler E.J.: Neurologic complications of burns in childhood. In: "Burns in children", Carjaval HF., Parks D.H. (Eds.), Chicago, Medical Year Book, 280-6, 1988.
22. Bertin-Maghit M., Gueugniaud P.Y., Bouchard C. et al.: Perturbations neurologiques après réanimation des brûlés graves: à propos d'un cas évoquant une myélinolyse centre-pontine. Réanim. Urg., 5: 49-52, 1996.
23. Globus J.H., Bender M.B.: Disseminated toxic degenerative encephalopathy (disseminated sclerosing demyelination) secondary to extensive and severe burns. J. Nerv. Ment. Dis., 85: 518-29, 1936.
24. Gueugniaud P.Y., Bertin-Maghit M., Pannefier J.C. et al.: Hypertension intracrAmenne et hypoperfusion cerebrate: une complication habituelle des br6l6s graves. Press. Med., 21: 1482, 1992.
25. Miller J.G., Bunting P, Burd D.A.R., Edwards J.D.: Early cardiorespiratory patterns in patients with major burns and pulmonary insufficiency. Burns, 20: 542-6, 1994.
26. Bernard F., Gueugmaud P.Y., Bouchard C. et al.: Etude des paramètres h6modynamiques chez le brW grave pendant les 72 premieres heures. Ann. Fr. Anesth. R6anim., It: 623-8, 1992.
27. Bernard E, Gueugniaud P.Y., Bertin-Maghit M. et al.: Prognostic significance of early cardiac index in severely burnt patients. Burns, 20: 529-31, 1994.
28. Drost A.C., Burleson D.G., Cioffi W.G. et al.: Plasma cytokines following thermal injury and their relationship with patient mortality, burn size and time postburn. J. Trauma, 36: 335-9, 1993.
29. De Bandt J.P., Chollet-Martin S., Hemvan A., et al.: Cytokine response to burn injury: relationship with protein metabolism. J. Trauma, 36: 624-8, 1994.
30. Gueugmaud P.Y., Vilasco B., Pham E., Hirschauer C., Fabreguette A., Bertin-Maghit M., Petit P.: Brûlés graves: état hémodynamique, transport et consommation d'oxygène, cytokines plasmatiques. Ann. Fr. Anesth. R6anirn., 15: 27-31, 1996.