CHAPTER 338 Indications and Techniques for Cranial Decompression after Traumatic Brain Injury

Reduction of death and disability from traumatic brain injury (TBI) is the goal of cranial decompressive surgery. Secondary insults and injury, including brain edema, elevated intracranial pressure (ICP), and decreased cerebral perfusion pressure (CPP), are the leading causes of in-hospital deaths after TBI.¹ The primary goals of treatment of TBI are therefore to control brain edema and maintain adequate blood flow and delivery of oxygen to the injured brain tissue. Adherence to published prehospital, intensive care, and surgical guidelines for the management of severe TBI has resulted in improved survival and outcomes by mitigating the deleterious effects of secondary insults and injury. ICP monitoring guides the use of specific therapies such as drainage of cerebrospinal fluid (CSF), sedation, mild hyperventilation, or hyperosmolar euvolemia with the administration of hypertonic saline or mannitol (or both) for control of ICP and CPP. Although these therapies may provide adequate treatment for many patients, there is a cohort of patients in whom cerebral edema will continue to propagate despite “maximal medical management” and culminate in increased cellular injury and death and ultimately in poorer outcomes. These patients are possible candidates for decompressive surgery to assist in the control of ICP.

Background

The concept of cranial decompression is by no means new and perhaps dates as far back as Neolithic times when trephination is thought to have been practiced. In the modern medical era, decompressive surgery has been used intermittently with variable success since Bergmann first described the technique in 1880, followed by Cushing’s descriptions of subtemporal decompressive craniectomy for control of ICP.¹^–³ Reports of high morbidity and mortality rates were common in the 1960s to 1980s, but a variety of techniques were used, primarily after clinical deterioration.⁴^–¹⁶ With the development of aggressive intensive care management of TBI patients, decompressive surgery has garnered renewed interest in the past decade or so because of improved results.^1,^2,¹⁷^–³¹ The setting in which decompressive craniotomy or craniectomy (DC) is used has evolved tremendously in recent years, most notably with respect to prehospital care, neuroanesthesiology, neuroradiology, neurointensive care, and rehabilitation. Advances in these areas have contributed to a reduction in mortality and improved outcomes for patients with severe TBI.^32,³³

DC has been used in nontrauma settings as well. Uncontrolled ICP after aneurysmal subarachnoid hemorrhage has been treated with DC in a variety of settings, including signs of cerebral edema during aneurysm clipping; increased ICP and epidural, subdural, or intracerebral hematoma after aneurysm surgery; and cerebral edema and elevated ICP without radiologic evidence of cerebral infarction after aneurysm surgery.^34,³⁵ DC has also been reported for the treatment of cerebral edema after cerebral infarction.³⁶^–⁴⁴ Cerebral edema and mass effect after an infarction further compromise CPP, oxygenation, and metabolism and lead to refractory intracranial hypertension and eventually transtentorial herniation, the leading cause of death in these patients.³⁵ In experimental models, DC has been shown to improve CPP, survival, and neurological outcome, as well as reduce the volume of infarction after extensive ischemic stroke.⁴⁵ These beneficial effects have been attributed to increased collateral circulation, reduction in tissue edema, and improvements in oxygenation and energy metabolism in the ischemic penumbra.³⁵ DC for other indications such as brain tumor, meningitis, acute encephalitis, toxoplasmosis, encephalopathy secondary to Reye’s syndrome, subdural empyema, and cerebral venous and dural sinus thrombosis has also been described.^35,⁴⁶^–⁵²

With respect to TBI, progress has certainly been made in our understanding of the pathophysiologic processes of cerebral edema, alterations in blood and oxygen supply and demand, metabolic failure, and cell death. However, clinical trials of drugs and other treatment modalities have failed thus far to show significant class I evidence of benefit, and identification of effective neuroprotective interventions remains elusive. DC has advantages over other neuroprotective therapies in that it has a global action (reduction of ICP and improvement of CPP), is not restricted to a singular physiologic pathway, and is potentially associated with fewer systemic side effects than occur with other medical options.

The rationale for DC is straightforward. The incidence of elevated ICP requiring therapy after severe TBI is 65%, and half of the patients who die of severe TBI die with uncontrolled ICP.^53,⁵⁴ The total time that ICP is elevated greater than 20 mm Hg correlates directly with outcome, and mortality is increased when ICP is not controlled.⁵⁵^–⁵⁹ Therapeutic approaches for decreasing ICP include reduction of the volume of intracranial contents (blood, brain, or CSF), reduction of cerebral metabolic demands, or an increase in cranial volume via DC. DC has been proved to decrease ICP.^19,⁶⁰^–⁶² Decompressive surgery also leads to a rightward shift of the pressure-volume curve and therefore to massive increases in compliance, as well as a decrease in the amplitude of ICP waves and an increase in compensatory reserve.^25,^63,⁶⁴ This results in an increase in cerebral blood flow and subsequently CPP and cerebral microcirculation, which allows rebalancing of cerebral inflow-outflow regulation.⁶¹ Brain tissue oxygenation (PbtO₂) also improves with durotomy at DC.^61,⁶² Because reduction of ICP is associated with improved outcome and DC is associated with reduction of ICP, it stands to reason that DC should also be associated with improved outcome, given an acceptable risk-benefit ratio and appropriate patient selection.

Indications

Most authors agree that patients with bilaterally fixed and dilated pupils, a Glasgow Coma Scale (GCS) score of 3, brainstem injury, and central herniation are universally poor candidates for DC because of the known association of these findings with poor outcomes.^16,⁶⁵ The postresuscitation GCS score, especially the motor score, is one of the most important factors to consider in patient selection.^63,⁶⁶ Care should be taken to exclude possible influences on GCS scores such as intoxication, hypoxia, hypotension, and paralytics or sedatives. Younger patients generally have better outcomes; however, age alone should not be used as an exclusion criterion.^65,^18,⁶⁷ The presence of midline shift on computed tomography (CT) of the brain is highly predictive prognostically. The degree of shift is inversely related to outcome, and elevated ICP is presumed.^68,⁶⁹ Absent or compressed cisterns are also predictive of elevated ICP and a poor outcome.^68,⁷⁰ The decision to proceed to DC must therefore take into account the preoperative CT appearance, clinical factors, and measured neuromonitoring trends.

The timing of decompressive surgery is important prognostically; once ICP becomes unmanageable and signs of brainstem compression are noted, DC may be lifesaving, but at the expense of severe neurological impairment. Early decompression (within 4 hours of injury) results in profound decreases in mortality and improvement in functional outcome at 6 months.¹⁸ Decompressive surgery should be considered very shortly after failure of maximal medical therapy to control ICP. Initial management includes elevation of the head to 30 degrees, use of sedation with or without paralytics, adjustment of ventilation parameters to maintain Paco₂ at 30 to 35 mm Hg, maintenance of normothermia, maintenance of hyperosmolar euvolemia (with the administration of mannitol and hypertonic saline), avoidance of hyperglycemia, support of CPP with volume or pressors, and drainage of CSF. Occasionally, pentobarbital therapy or hypothermia may be used before proceeding to surgery. ICP in the low twenties may be tolerated without proceeding to DC if CPP or PbtO₂ (or both) is judged to be adequate. However, ICP elevations or CPP or PbtO₂ depressions in the face of maximization of the aforementioned parameters are not tolerated for long periods before surgery.

Technique

DC may be performed in a variety of clinical settings encompassing a heterogeneous group of brain-injured patients. It may be performed in conjunction with evacuation of an extra-axial mass lesion; in conjunction with removal of an intraparenchymal hemorrhage; in conjunction with lobectomy; for diffuse brain edema; for penetrating trauma with débridement of bone fragments, foreign material, and necrotic brain tissue; in the face of major open depressed comminuted skull fractures with underlying brain injury; and in various combinations of the preceding scenarios. Basically, however, DC techniques may be considered in terms of two patient populations: those in whom a craniotomy is being performed on an emergency basis for evacuation of a mass lesion or major open cranial injury with the bone flap left out and those who are taken to the operating room expressly for DC for refractory intracranial hypertension. The basic techniques of the craniotomy (extent of the scalp incision and bone opening) and the duraplasty, however, are consistent for both groups.

The decision to perform a bifrontal or a unilateral hemicraniectomy must be made first and is based on the presence, location, and extent of mass lesions (extra-axial or intraparenchymal), penetrating injuries, and midline shift. In cases of diffuse brain edema with no mass lesion or midline shift, nondominant hemisphere unilateral or bifrontal DC may be used. Otherwise, the side with the greater lesion volume or cerebral edema is chosen for unilateral decompression. Occasionally, the nondominant side is selected for unilateral decompression in patients with minimal or nonlateralizing signs.

The comatose status of severely injured patients precludes clearance of the cervical spine from ligamentous instability, so patients are typically left in the neutral position in a cervical collar even if bony cervical spinal column injury has been ruled out. The patient can be placed in the reverse Trendelenburg position for head elevation because the thoracolumbar spine is frequently not yet cleared. The head can be turned to facilitate exposure of the hemicranium by placement of a sandbag or shoulder roll under the ipsilateral shoulder. We use a doughnut rather than a Mayfield headrest to expedite surgery and prevent interference with the craniotomy by the presence of the pins; cranial immobilization may be provided by the assistant during drilling.

After hair clipping extending just across midline and as far posteriorly as possible, the hemicranium is prepared, marked, and injected with 1% lidocaine with epinephrine to facilitate hemostasis before draping. For a unilateral craniotomy, a standard large question mark or reverse question mark incision is used. The skin incision should start 1 cm in front of the tragus at the zygomatic arch and extend posteriorly above the auricle, upward over the parieto-occipital area, and forward to the frontal region to the hairline. The superior limb should approach the midline, and the posterior limb should be sufficiently posterior to allow creation of an adequately sized bone flap. Although the exact dimensions of the bone flap may vary according to the size and shape of the cranium, the scalp exposure should allow access to specific bony landmarks. For example, the inferior exposure at the temporal region must allow the temporal craniectomy to be extended to the floor of the temporal fossa after the bone flap has been removed (Fig. 338-1; see later also). Scalp hemostasis is facilitated with the use of Raney clips. Bovie cautery is then used to divide the temporalis fascia and muscle in line with the scalp incision. The temporalis muscle, which is often quite edematous, may be reflected anteriorly and inferiorly with the cutaneous flap and both secured with fishhooks after protecting the musculocutaneous flap with rolled sponges underneath. An epinephrine-soaked laparotomy sponge is frequently used on the galeal surface of the cutaneous and muscle flaps to aid in hemostasis. For bifrontal DC, a standard Souttar incision is used.

FIGURE 338-1 Drawing demonstrating the extent of the incision and underlying bone resection for decompressive surgery. Note that the temporal bone must be resected down to the level of the middle fossa floor.

The craniotomy itself for unilateral DC must encompass a large enough area to prevent brain herniation and strangulation, typically from just lateral to the superior sagittal sinus, frontally to the midpupillary line, inferiorly to the floor of the temporal fossa, and posteriorly to the parieto-occipital area (Fig. 338-2A). Bifrontal openings should span from the anterior cranial fossa floor to the coronal suture posteriorly and to the temporal fossa floor bilaterally (Fig. 338-2B). The frontal sinus may be entered, and if so, it should be cranialized. Visibility of the mesencephalic cisterns on postoperative CT correlates with the distance from the craniectomy to the base of the cranium,¹⁸ and the size of the bone flap correlates with the degree of reduction of ICP¹⁹ (Fig. 338-2C).

FIGURE 338-2 A, Extent of bone resection necessary for unilateral decompression. A temporal craniectomy to the level of the middle fossa floor must be performed to avoid strangulation of the temporal lobe. B, Extent of bone resection necessary for bifrontal decompression extending across the orbital rims and down to the base of the temporal fossa bilaterally. C, Three-dimensional view of the skull after unilateral decompression.

For patients in whom a large “trauma flap” is turned to evacuate a mass lesion in anticipation of leaving the bone flap out, the decision to do so is made intraoperatively. However, the scalp incision and bone flap must be planned in anticipation of this eventuality. Several factors must be taken into account when making this decision, namely, the preoperative presence of a midline shift out of proportion to the mass lesion and the appearance of the basal cisterns. The intraoperative finding of cerebral herniation out of the craniotomy opening after removal of the mass lesion or lesions is an indication to perform a duraplasty and leave the bone flap out. In contrast, in instances in which evacuation of the mass lesion has resulted in adequate cerebral decompression (as can be seen with ultra-early evacuation of a subdural hematoma, an atrophic brain, or removal of a large intraparenchymal hematoma), the bone flap may be replaced. However, caution should be exercised when considering bone flap replacement because postoperative swelling can be more extensive than expected and a reoperation to remove the bone flap may be required. An attempt to “unherniate” the uncus is made after evacuation of the mass lesion or lesions and before the duraplasty. This may be achieved by gentle elevation with a Penfield dissecting instrument or retraction blade. In cases of intraparenchymal hemorrhage, especially mixed-density contusions, we routinely avoid aggressive débridement of contusions to preserve potentially viable tissue, particularly in the posterior temporal lobe. Duraplasty over such contusions allows preservation of cerebral tissue and edema without compression. Epidural hematoma requires leaving the bone flap out much less often than subdural hematoma does because of the relative lack of underlying brain injury with the former.

We generally allow Pco₂ to rise intraoperatively and observe the brain for several minutes before deciding to replace the bone flap, in addition to taking into account the following: the degree of preoperative midline shift relative to the volume of the mass lesion evacuated, the appearance of the cisterns on the preoperative CT scan, the absolute volume of hematoma removed, the appearance of the hemisphere at surgery (degree of swelling and hemorrhage, pulsatility, appearance of the vasculature), the age of the patient, the mechanism of injury, the presence of other non–central nervous system (CNS) injuries (especially pulmonary), the time from injury to evacuation of the initial lesion, and the extent and correctability of the coagulopathy.

For patients sustaining massive open wounds to the cranium with underlying brain injury, as in crush injuries, major blunt force, blast injuries from terrorist or military actions, or major penetrating wounds from gunshots, shotgun blasts, or sharp objects, explicit attention to the scalp wounds must be taken into account. Frequently, modifications of standard incisions must be made to incorporate lacerations or entrance wounds. The blood supply to the scalp needs to be preserved by maintaining an inferior vascular pedicle, and creation of “islands” of tissue must be strictly avoided. Extensive débridement of necrotic tissue and irrigation of hematoma, contaminated tissue, and foreign material are undertaken. A high-speed drill may be used to create a wide bony opening in anticipation of duraplasty and evolution of postoperative hemispheric edema. If multiple bone fragments are encountered, small ones are discarded and a synthetic implant should be considered at the time of reconstruction.

In cases in which it is determined that the bone flap will be left out after evacuation of a mass lesion or after primary DC, durotomy must be performed; otherwise, the operation is ineffective in reducing ICP.^61,^66,⁷¹ Durotomy is akin to fasciotomy in other body compartments. A variety of dural openings have been described. For unilateral operations we use a “C-shaped” dural opening mirroring the scalp incision, with the dural base along the frontal fossa floor (Fig. 338-3A and B). Some have advocated slit dural openings, stellate openings, or other curvilinear openings. For bifrontal procedures, ligation and division of the anterior sagittal sinus, as well as the falx at its most anterior extent, are advocated to allow maximal relief of constriction on the frontal lobes. Whatever the dural opening technique, it is imperative that the temporal dura be opened to decompress the temporal lobe or lobes.

FIGURE 338-3 A, Curvilinear dural incision beginning over the temporal lobe. B, After durotomy, relaxing incisions in the perimeter of the exposure are made for additional relaxation. C, Dural closure incorporating a generous dural patch to allow outward herniation of the brain.

Duraplasty using autograft or allograft may be performed or the dura left open. Several options for dural graft material are available: fascia lata, pericranium, bovine pericardium, collagen matrix products, and synthetic dural substitutes. We perform duraplasty with a collagen matrix or bovine pericardium dural substitute to provide brain protection and avoid adherence of the galea to the underlying brain tissue because return to surgery will be mandated for survivors (Fig. 338-3C). Duraplasty also restores more normal CSF circulation and may prevent or minimize the development of postoperative hygromas and leakage of CSF from the wound. The duraplasty is constructed such that a wide patch graft is created over the herniating cerebral hemisphere or hemispheres, but the following must be avoided: constriction of underlying brain tissue by dural folds and wrinkles and redundant dura or patch graft material resulting in excess graft tissue volume (remembering that in cases of profound cerebral edema, loss of autoregulation, and poor compliance, even the smallest excess volume in the intracranial compartment can contribute to significant increases in ICP). We have used an additional layer of nonbiologic, nonincorporated, synthetic dural substitute as a subtemporalis barrier at the time of DC, which can be readily identified and removed during reimplantation of the bone flap. This has dramatically decreased the dissection time at reoperation and reduces the associated risks of traction on the underlying brain.

A vascular tunneling technique has been described by Csokay and colleagues to alleviate venous congestion (and therefore evolution of brain edema) after decompressive surgery.⁷² In this technique, small supporting pillars made of hemostatic sponge are placed around the superficial vessels supplying the protruding portion of the brain, thus preventing them from being compressed by the pressure between the dural edge and the brain tissue. In our experience, performing a large craniectomy and durotomy with dural slits along the perimeter of the dural opening perpendicular to the dural incision alleviates the need to perform such a maneuver (Fig. 338-3B).

Partial or anatomic lobectomies have also been performed in conjunction with DC for intractable intracranial hypertension; however, the need for this intervention is significantly lessened when DC is performed early after failure of medical management. If performed, it is often used in conjunction with evacuation of cortical contusions or intraparenchymal hematomas and is limited to the frontal and anterior temporal lobes, preferably on the nondominant side. Lobectomy does not preclude a good outcome.⁷³ However, in our experience, lobectomy is seldom necessary and is reserved for instances of a “pulped” or necrotic lobe with malignant cerebral edema at surgery, pulselessness of the hemisphere, vessel occlusion or infarction (which can be related to a blunt vascular injury), profound coagulopathy in the face of multiple intracranial hemorrhages, or any combination of such conditions. There is a high correlation of lobectomy with poor initial GCS score and emergency surgery in our institution.

In patients with extreme cerebral edema it may be difficult to close the scalp. Undermining the galea and creating barrel stave incisions in the galea will allow additional elasticity of the scalp. Temporalis resection may be used but should be a “last-ditch” effort because absent temporalis can contribute to significant problems.

The bone flap may be stored in an abdominal subcutaneous pocket or in a certified tissue bank. Abdominal storage has been associated with resorption, rhabdomyolysis,⁷⁴ and infection and of course requires an additional incision and procedure at both storage and retrieval. Contaminated flaps should not be stored in the abdominal wall.

A second option is storage in a certified tissue bank. Compliance with extensive federal regulations for tissue storage essentially precludes in-hospital storage, but certain institutions with certified tissue banks may wish to use in-house storage. Bone flaps should be cleared of hematoma and foreign material and be irrigated with antibiotic solution before storage. Flaps are frozen until the time that they are needed for reimplantation. Potential drawbacks with this method include the possibility of the flap being lost or misplaced, freezer malfunction, and detailed administrative requirements and agreements regarding storage services. In centers in which patients may travel to other centers for eventual reimplantation, additional shipping costs for the bone flap will apply.

Notation should be made of the number and size of the bone segments submitted for storage and the contaminated open wounds in contact with bone; pre-plating before storage is not recommended. For penetrating injuries, contaminated open wounds in contact with bone, or comminuted fractures resulting in multiple bone fragments, use of autograft may be foregone in favor of a customized synthetic implant. Use of multiple fragments, especially when small, can lead to later bone resorption and inadequate cerebral protection. It is our custom to replace up to three bone fragments if sizable, but if adequate coverage cannot be achieved or the flap is more fragmented, a customized synthetic implant is used. A synthetic implant may also be required in the case of later bone flap infection or resorption. Several Food and Drug Administration–approved customized synthetic implants are commercially available; they may be made from various materials, including polyetheretherketone (PEEK), porous polyethylene, acrylic, or titanium, and selection is a matter of surgeon preference and experience. We use PEEK because of its fit, facile instrumentation, design and strength characteristics, ability to be resterilized, and ease of dissection if recurrent surgery is needed.

The interval between decompressive surgery and secondary bone flap replacement in the literature varies from 4 weeks up to 12 months, during which time the unprotected brain is exposed to the risk of injury.⁷⁵ We recommend the use of a soft helmet when out of bed until the bone flap is replaced to reduce the risk for secondary brain injury. The helmet is removed when the patient is safely in bed or in a chair to prevent scalp sweating and friction on the incision. Our typical approach to the timing of staged secondary bone replacement is to replace the flap at around 3 months. When patients with severe TBI arrive for their first neurosurgical follow-up after discharge from the hospital or acute inpatient rehabilitation, elective surgery for replacement of the bone flap is scheduled if no contraindications exist, such as ongoing active infection, ongoing advanced-stage decubitus ulcers, or extremely poor nutritional status. This approach allows in-hospital and rehabilitative treatment of associated injuries, resolution of cerebral edema, and reduction in the risk for infection related to intensive care unit interventions or nosocomial infections. At the same time, avoidance of longer intervals before bone replacement reduces the incidence of scalp contraction (which can contribute to wound dehiscence), temporalis atrophy (which can contribute to temporomandibular joint complaints and cosmetic asymmetry), and development of the syndrome of the trephine (which can cause severe headache). Furthermore, some patients do exhibit neurological improvement once the flap is replaced, especially improvement in motor and speech deficits, reduction in seizure frequency, and cognitive improvements. Early replacement before the development of severe encephalomalacia also eliminates the potential dead space under the flap postoperatively, thereby decreasing the postoperative risk for epidural hematoma, pneumocephalus, and fluid collections. Finally, early replacement provides cerebral protection during the phases when patients are most at risk for further cerebral injury, such as after they are beginning to be actively transferred and are starting to walk again but are still likely to experience imbalance, incoordination, and orthostatic hypotension.

Before reimplantation, a CT scan of the brain should be obtained to permit identification of hydrocephalus or subdural or subgaleal fluid collections. In the case of the former, a lumbar drain or ventricular catheter may need to be used at the time of reimplantation surgery to decompress the ventricles and allow the bone flap to be replaced without undue pressure on the underlying brain tissue. If there is fullness related to a fluid collection, it can be drained at surgery to allow sufficient slack for replacement of the bone flap. Finally, the temporalis muscle can be assessed with CT to delineate the degree of atrophy and delimit the borders of the muscle.

Reimplantation surgery requires meticulous wide sterile preparation of the scalp incision. We use a wide hair clip to identify potential problems from lacerations, abrasions, and other injuries and to facilitate preparation and postoperative wound care. Perioperative antibiotics are administered. The previous incision is used, and care is taken to avoid direct trauma to the underlying brain tissue from local anesthetic injection, scalp incision, and dissection of the cutaneous flap away from the dura. Care must be taken to avoid transfer of heat to the underlying brain from monopolar cautery or pressure on the brain from digital dissection or traction on the cutaneous flap. We use sharp dissection with Metzenbaum scissors and bipolar cautery for hemostasis. The cutaneous flap is gently and consistently lifted upward and gradually retracted as the dissection proceeds, and the bone edges are identified and dissected free of scar tissue with a Penfield No. 1 dissecting tool. Removal of previously placed epidural tack-ups is often necessary to achieve anatomic reseating of the bone flap. There is frequently a ridge of scar tissue at the perimeter of the bone flap that may be shrunken back with bipolar cautery (Fig. 338-4A). The border of the temporalis muscle is identified by sharp dissection, and the muscle is dissected off the underlying dura, to which it is frequently quite adherent unless a barrier has been placed as previously described. It is often not necessary to dissect the temporalis entirely because the bone flap may not fill the cranial defect completely if portions of the temporal bone were rongeured and discarded at the original surgery. Central epidural tack-ups are used to aid in postoperative hemostasis.