Cerebral fat embolism syndrome (FES): similar cases with different outcomes

Abstract

Fat embolism syndrome (FES) is a rare multisystem, clinical syndrome occurring in 0.9%–2.2% of long-bone fractures. The severity of FES can vary from subclinical with mild respiratory changes and haematological aberrations to a fulminant state characterised by sudden onset of severe respiratory and neurological impairment. Here we present two patients with cerebral FES secondary to femur fracture. Both patients exhibited profound neurological impairment with varied outcomes. Our cases highlight the importance of a high clinical suspicion of FES in patients with long-bone fractures and neurological deterioration. We recommend early plate osteosynthesis to prevent additional emboli in patients with FES and situational placement of intracranial pressure monitoring. Finally, cerebral FES has low mortality even in a patient with tentorial herniation and fixed, dilated pupils.

Background

Fat embolism syndrome (FES) is a rare multisystem, clinical syndrome occurring in 0.9%–2.2% of long-bone fractures.¹ Fat globules released from the bone marrow enter the circulation, causing occlusions. Patients typically present with signs and symptoms within 24–72 hours of the inciting event.² The major criteria for FES, as described by Gurd, are neurological impairment, respiratory insufficiency and petechial rash. Some minor diagnostic criteria include: fever, tachycardia, retinal abnormalities, thrombocytopenia and anaemia.³ However, diagnosis of FES is rarely so straightforward. It requires a high degree of clinical suspicion since its presentation can vary from subclinical with mild respiratory changes and haematological aberrations to a fulminant state characterised by sudden onset of severe respiratory and neurological impairment.⁴ Respiratory involvement typically manifests as tachypnoea, shortness of breath and hypoxaemia. The petechial rash associated with FES is non-palpable and commonly found in the axillae.² The neurological impairment from FES can present as irritability, headache, stupor, convulsions or even coma. Onset of neurological findings is often simultaneous with respiratory impairment.² Focal findings are less common and often difficult to identify in patients with trauma.⁵ Imaging studies assist in the diagnosis of FES, particularly in patients with neurological involvement. Brain MRI is most helpful in identifying FES as a patient’s initial head CT is often completely normal. FES-specific findings are punctate lesions within the white matter on T2-weighted images which are present as early as 4 hours after onset of FES.⁵ The pathognomic ‘starfield pattern’ can often be seen on MRI diffusion-weighted sequences.

Here we present two patients with cerebral FES secondary to femur fracture. Both patients exhibited profound neurological impairment, yet neither patient was found to have a pulmonary or intracardiac shunt. In comparing our patients, we found strikingly similar courses with varied outcomes.

Case presentation

Case 1

A 29-year-old man with no significant medical history presented to our emergency department via emergency medical services as a transfer from an outside hospital after involvement in a high-speed motor vehicle collision positive for loss of consciousness. Vital signs were stable, and Glasgow Coma Scale (GCS) was 15. Pupils were equal, round and reactive. On physical examination, the patient was found to have a left open both bone forearm fracture, left closed midshaft humerus fracture, left closed midshaft femur fracture which had been splinted, and a head laceration. The right upper extremity and bilateral lower extremities were neurovascularly intact. The left upper extremity had no motor function in the hand, had a 1+ radial pulse and was with decreased sensation in median and ulnar nerve distributions concerning for brachial plexus injury. Imaging confirmed midshaft transverse fracture of the left femur, midshaft both bone fractures of the left forearm and midshaft left humerus fracture. CT of left lower extremity demonstrated hypodense material with locules of gas in the left external iliac vein concerning for fat emboli (figure 1). Initial head CT had no acute intracranial abnormalities. Prioritising early fixation of the patient’s femoral fracture, he was scheduled for surgery. However, during preoperative evaluation in presurgical unit for fixation of his femur, the patient was found to be acutely unresponsive. Pupils at this time were equal, round and reactive. He was emergently intubated. Immediate CT scan was negative for any acute pathology. Subsequent MRI of the brain showed findings consistent with fat emboli with diffuse bilateral hypodensity spanning from the basal ganglia to the cortex (figure 2). Otherwise, the MRI was negative for other differential diagnoses, including cerebral contusion, herniation or stroke. He also developed petechial rashes in both axilla consistent with FES (figure 3). His GCS score had fallen to 3T. The following day, he was noted to have non-reactive, enlarged pupils bilaterally. An updated head CT, performed approximately 12 hours later, showed increased intracranial pressure (ICP) and developing uncal herniations without evidence of transforaminal herniations (figure 4). An ICP monitor was placed. The patient was then treated with supportive medical therapy which included administration of hypertonic saline and mannitol while remaining intubated and sedated. ICPs ranged from 7 mm Hg to a maximum of 13 mm Hg as demonstrated in figure 5. With these supportive treatments, his gag reflex returned within the first day. He was noted to continue to have non-reactive pupils with a GCS score of 3T. Over the next 48 hours, his left pupil remained larger than his right. His anisocoria and non-reactive pupils persisted for the next 10 days; after this point, his pupils were equal and slowly reactive. Eleven days following initiation of supportive treatments, he was grimacing periodically and opening his eyes to verbal commands and touch. Five days following this, he was noted to open his eyes spontaneously and follow motor commands in all extremities except the left upper extremity. This was due to the significant trauma to that extremity. Two days later, he had return of his speech revealing that he was fully oriented to person, place and time. A transthoracic echocardiogram (TTE) with bubble study showed no evidence of intracardiac shunting. CT angiogram did not exhibit pulmonary shunting.

Figure 1.

A CT of the left lower extremity demonstrating hypodensities (arrow) most consistent with fat emboli.

Figure 2.

Sagittal MRI demonstrating hyperintensities in the basal ganglia as well as diffusion restrictions throughout the cortical white matter, subcortical white matter and grey/white matter interface indicative of embolic infarctions and fat emboli.

Figure 3.

Axillary petechial rash commonly seen with fat embolism syndrome.

Figure 4.

(A–C) Head CT exhibiting increased intracranial pressure and uncal herniation without transforaminal herniation.

Figure 5.

Intracranial pressure trend of case 1.

Case 2

A 31-year-old man without pertinent medical history presented to our emergency department as a transfer from an outside hospital after involvement in a motorcycle versus vehicle collision. The patient was wearing a helmet, collided with a vehicle at approximately 50 mph, and reported loss of consciousness. On presentation, vital signs were stable, and GCS score was 14 as the patient was slightly lethargic. Pupils were equal, round and reactive. On physical examination, motor and strength were intact bilaterally in upper and lower extremities. No neurological deficits were noted. The patient only endorsed tenderness to palpation over the right thigh. Injuries, confirmed by imaging, included unstable L3 burst fracture, right midshaft transverse femur fracture and right displaced, comminuted ischium pelvic fracture. Initial head CT exhibited no acute intracranial abnormalities. On day 2 of hospital stay, the patient was observed to have acute onset seizure-like activity. The patient had three episodes of eye rolling with tonic stiffening of the arms lasting several minutes each with an associated postictal state. An electroencephalogram indicated bihemispheric dysfunction interpreted to be caused by hypotension, hypoxia or true seizure activity, although the wave activity was atypical for such. A loading dose of levetiracetam was administered. Prioritising early fracture fixation, the patient was scheduled for fixation of his unstable lumbar burst fracture with concomitant intramedullary nailing of the right femur less than 24 hours after arrival to our institution. However, the surgery was postponed when he was found in the preoperative area to have acute onset neurological deficits in the extensor hallucis longus bilaterally (right lower extremity ⅗ strength, left lower extremity ⅘ strength). Neurosurgery was consulted, and surgery for lumbar spine was planned next day. His neurological examination for lower extremity remained stable with numbness but improved motor examination. Also, the patient declined surgery for femur and requested his spine to be fixed first. Few hours later, the patient continued to have seizure-like activity requiring sedation and intubation for airway protection and initiation of supportive medical treatment including hypertonic saline and mannitol administration. On hospital day 3, a brain MRI exhibited multiple small foci of diffusion restriction in the bilateral basal ganglia likely representing diffuse axonal injury (figure 6). Ophthalmology was consulted to perform a dilated pupil examination looking for retinal evidence of fat emboli. The patient was found to have areas of whitening in the macula of both eyes, right worse than left. Most likely these areas represented Purtscher retinopathy, an embolic event often seen after non-ocular trauma, including long-bone fractures. A suboptimal TTE was performed during which a few bubbles were noted in the left atrium and ventricle after contrast injection. These findings are consistent with a right-to-left shunt at the atrial or pulmonary level. However, the degree of shunting was determined to be an unlikely cause of hypoxia. A more optimal transoesophageal echocardiogram, performed later in the hospital stay, exhibited no evidence of intracardiac shunting. By hospital day 5, the patient’s pupils became non-reactive and asymmetric, with the left pupil larger than the right. A head CT showed diffuse cerebral oedema with transtentorial herniation and findings consistent with diffuse fat emboli (figure 7). An MRI performed 12 hours prior to the patient’s acute deterioration exhibited no signs of herniation or other diagnosis, such as contusion (figure 8). GCS score had fallen to 3T. Neurosurgery performed a ventriculostomy and placed an ICP monitor. Despite supportive therapy, the patient’s ICP remained elevated and proved difficult to control. Over the course of his stay, intracranial pressures ranged from 4 mm Hg to 46 mm Hg, averaging at approximately 17 mm Hg. The trend in ICPs is shown in figure 9. Due to the patient’s poor clinical status, spine stabilisation was not performed until hospital day 23. His femur fracture was stabilised with external fixation at bedside early in his hospital course, but definitive fixation was delayed until 3 weeks post FES presentation.

Figure 6.

Brain MRI diffusion weighted image demonstrating diffusion restriction in bilateral basal ganglia (arrows).

Figure 7.

(A–C) Head CT exhibiting diffuse cerebral oedema with transtentorial hernation.

Figure 8.

Brain MRI obtained prior to deterioration without signs of herniation or contusion.

Figure 9.

Intracranial pressure trend of case 2.

Outcome and follow-up

Case 1

Over the span of his 40-day hospital stay, the patient’s neurological status gradually improved, and he was discharged to a rehabilitation facility. No specific therapy was targeted towards his cerebral fat embolism, and all fractures were stabilised with plate osteosynthesis as soon as it was medically safe.

Case 2

The patient’s neurological status failed to improve over the course of his nearly 50-day stay. He was discharged home per the family’s request and died 2 days later of unknown cause.

Discussion

The pathogenesis of cerebral FES is currently unknown. However, it is thought to be due to either fat emboli entering arterial circulation, or the cytotoxic effect caused by the fat globules. In the mechanical theory of FES, large fat emboli are thought to enter the arterial circulation via intracardiac shunts or pulmonary arteriovenous malformations leading to direct brain injury. Intracardiac shunts are present in approximately 20%–25% of the population² and may place a patient at increased risk for FES. However, the syndrome is also seen in patients without intracardiac shunts.⁶ It is hypothesised that microembolic fat globules may traverse normal pulmonary capillaries to enter systemic circulation. In young, healthy patients with large pulmonary reserves, neurological manifestations of the systemic emboli may predominate as seen in the two patients presented above.⁷ An alternative theory of FES pathogenesis postulates this multisystem syndrome is due to an inflammatory response. When fat emboli enter circulation, free fatty acids act as inflammatory mediators resulting in endothelial injury and subsequent cytotoxic oedema.⁸

Here we present two cases that highlight certain important aspects of cerebral FES. Many similarities existed between our two patients. They were healthy, young, muscular males with high-velocity injuries causing closed mid-shaft femur fractures. Both patients arrived as transfers from outside facilities with some delay of initial care. On presentation, they had a GCS score of 15/15, but both deteriorated within 12 hours to a GCS score of 3T. Our two patients had negative head CTs, but similar MRI findings consistent with cerebral FES. Their hospital courses progressed similarly with both developing accelerated cerebral oedema leading to tentorial herniation with a fixed, dilated pupil. With treatment, both patients’ fixed pupil regained function. Even with these similarities, the outcomes of our patients varied in that the patient in case 1 made significant neurological recovery whereas the patient in case 2 died at home from unknown causes. Two unique aspects of our case patients were CT findings of venous fat emboli in case 1 and the severity of cerebral oedema in case 2.

The incidental finding of the fat embolism in the iliac vein on CT imaging of case 1 is rare, however, it carries significant clinical importance. As evidenced by our case, the finding of the fat emboli in the vein heralded the patient’s respiratory and neurological impairment. It has long been known that fat emboli can be detected via intraoperative transoesophageal echocardiogram in patients undergoing intramedullary nailing.⁹ Yet, few case reports have been published with imaging findings consistent with fat emboli outside of intraoperative monitoring. In two cases reported by de Vasconcelos et al and Healy et al, CT scans showed fat emboli in the patient’s inferior vena cava and right common femoral vein, respectively, following orthopaedic trauma. Both patients went on to develop respiratory symptoms without neurological deficits.^{10 11} This, however, is not a rule. In three reported cases, venous fat emboli identified on CT following orthopaedic trauma, none of the patients developed respiratory or neurological symptoms.^12–14 In addition to CT detection, incidental findings of fat emboli during venous Duplex scanning of patients at risk for FES have been reported. Washko et al described ultrasound findings consistent with fat emboli in an asymptomatic patient 12 hours after right total knee arthroplasty.¹⁵ Nadaff et al reported on two patients, both with traumatic femoral fractures, who underwent Duplex scanning to rule out lower extremity deep venous thrombosis. Incidentally, both were noted to have hyperechoic masses consistent with fat emboli. Although neither were showing signs or symptoms of FES, they were deemed high risk, and transferred to the intensive care unit for continuous pulse oximetry and cardiac monitoring.¹⁶ Ultimately, these cases highlight that when venous fat emboli are identified on imaging following trauma, whether it be CT or ultrasound, medical management should adjust in anticipation of possible FES.

The severe and progressive cerebral oedema we observed in case 2 is also a rare finding in FES. In 2013, Kellogg et al identified only 54 reported cases of cerebral FES with clinically significant neurological dysfunction over the span of 30 years of literature. The majority (greater than 50%) of these patients achieved what Kellogg et al deemed a ‘meaningful recovery’.¹⁷ The most common neurological sequelae of FES include memory loss, some cognitive dysfunction, subtle changes in personality and/or focal deficits. The mortality of FES ranges from 5% to 15% but is mostly attributed to pulmonary sequelae.¹⁸ Only a few case reports identify patients with FES who experienced severe cerebral oedema resulting in brain death.^18–20 It is unclear what sets these patients with severe cerebral oedema and poor outcomes apart from others. Optimisation of medical and surgical management before rapid deterioration appears to be of utmost importance.

Currently, the mainstay of management is early recognition, since FES manifests within 24–72 hours of the inciting event,² and supportive therapy which includes oxygenation, ventilation and fluid resuscitation. As hypoxaemia is predominately the earliest sign of FES and has been observed in greater than 75% of patients with FES, early and continuous monitoring via pulse oximetry may be beneficial for high-risk patients.²¹ For patients with neurological impairment causing seizures, adequate antiepileptic therapy and proper management of ICP are also a mainstay of treatment. ICP monitoring is recommended especially in comatose, multitrauma patients to allow for early hyperosmolar therapy when needed.¹⁷ Use of albumin to bind oleic acid, a potent cytotoxic agent leading to lung injury in fat emboli, has been theorised, however with few studies showing its efficacy.²²

The use of corticosteroids in the prevention of FES has been studied since the 1980s. Sen et al reviewed these studies concluding that, although there was a statistically significant decrease in the incidence of FES with corticosteroid use, the studies had significant flaws including a variety of definitions of FES, small sample sizes, differing fracture populations and varying doses and types of corticosteroids used.²³

Early fracture fixation is a cornerstone in management for patients with FES, but definitive surgical fixation is dependent on the clinical stability of the patient, as evidenced by our case patients. Pape et al recommended stratifying patients based on clinical status (stable, unstable, borderline or in extremis) to determine which patients should receive temporary, external fixation, as opposed to definitive fixation.²⁴ Pape et al’s algorithm further emphasises the need for effective medical management in patients with FES to allow for surgical intervention.²⁴ Operative immobilisation, ideally between 12 and 24 hours after injury, has been associated with a fivefold reduction in FES compared with delayed treatment.²⁵ Furthermore, early fixation has been associated with shorter hospital stay and fewer pulmonary complications.²⁵ Which surgical fixation technique is superior, however, remains controversial. In a comparative study by Bosse et al, no difference in rates of acute respiratory distress syndrome was found with intramedullary nailing with reaming versus plate fixation.²⁶ Mellor and Soni found that unreamed nails, blunt reamers and intramedullary lavage caused less disruption of bone marrow and may decrease perpetuation of FES.²⁷ Conversely, Helttula et al concluded that reamed versus unreamed nails have no significant difference in their impact on haemodynamics. Yet, he found unreamed nails increase postoperative oxygen consumption.²⁸ Also, with the advent of the reamer–irrigator–aspirator, Dunn et al showed that a significantly decreased fat emboli burden was detected by transoesophageal echocardiogram when compared with a conventional reamer.²⁵ With these varying findings, no current guideline exists on which type of fixation should be used.

Although FES has a favourable prognosis and a reported mortality less than 15% with supportive therapy alone, there is still a significant lack of knowledge about the condition, particularly isolated cerebral fat embolism. The patients presented in this case report help to delineate the rapid decline possible with cerebral fat emboli. Furthermore, the lack of intracardiac shunt in both of our patients support the hypothesis that minute fat globules can traverse the pulmonary vasculature and cause systemic sequelae. Thus, cerebral FES is an important addition to the differential diagnosis in a patient with significant fractures and neurological impairment within 24–72 hours of injury, despite lack of history of intracardiac shunting. Although these two patients presented very similarly, their outcomes were quite different. We are unsure why case 2 had a worse outcome than case 1. We believe the burden of embolism was more severe with case 2 due to the prolonged coma and lack of early recovery, as well as the associated seizure activity. Optimal supportive management may have assisted in stabilising this patient for surgery earlier and exposed him to a lower embolic burden. Larger studies focusing on the pathophysiology, prevention, management and long-term outcome of FES are needed.

In conclusion, patients with long-bone fractures and neurological deterioration with a normal head CT should be strongly considered for cerebral FES. As evidenced by case 1, the finding of venous fat emboli on imaging may be an early sign of impending FES. Monitoring these patients closely with continuous pulse oximetry and regular neurological assessment may assist in early diagnosis of FES. In patients already diagnosed, we recommend close monitoring of rising cerebral pressure. Both patients discussed above exhibited rapid progression to tentorial herniation within a matter of hours. Early placement of an ICP monitor can facilitate early detection of neurological deterioration attributed to rising pressure; however, this should only be used in situations necessitating treatment of elevated ICP. Due to the controversial findings pertaining to the role of corticosteroids in treating FES, steroids were not administered to either patient per hospital protocol. Once medically stable, we recommend early plate osteosynthesis in patients with established FES to prevent further emboli as there is no current guideline available. Our cases reiterate the fact that supportive therapy is central to the care of a patient with FES. Every attempt should be made to resuscitate patients with FES as literature shows the syndrome has very low mortality. As evidenced by case 1, fixed dilated pupils in the setting of FES does not necessarily signify poor prognosis and still carries a good chance of meaningful recovery.

Learning points.

• Patients with long-bone fractures and neurological deterioration despite a normal head CT should be strongly considered for cerebral fat embolism syndrome (FES).

• Imaging findings of venous fat emboli may be an early sign of impending FES.

• Placement of an intracranial pressure monitor in patients with FES may assist in early detection of neurological deterioration, however is not necessary in routine management.

• Early plate osteosynthesis in patients with medically stable FES may prevent further emboli.

• Supportive care is central to the care of a patient with FES, who despite signs of severe neurological impairment, still carries a good chance of meaningful recovery.