Spontaneous cerebrospinal fluid fistula secondary to hyper-pneumatized paranasal sinuses and skull base: two case reports

Spontaneous cerebrospinal fluid (CSF) fistulas occur due to various reasons other than well-identified causes such as trauma, neoplasia or infection. Various contributory factors are attributed to formation of spontaneous CSF leaks such as idiopathic intracranial hypertension leading to prominent arachnoid granulations. Further, presence of hyper-pneumatized paranasal sinuses or the skull base weakens the bone and predisposes to development of spontaneous defects and further fistulas. This case report highlights two cases of spontaneous CSF leaks associated with hyper-pneumatized petrous bone and sphenoid sinus. A 26-year-old female patient with history of right rhinorrhea with imaging evidence of bilateral hyper-pneumatized petrous bones and a bony defect in the right petrous bone on computed tomography (CT). Subsequent CT cisternography demonstrated CSF leak extending into the right pneumatized petrous apex cells, Eustachian tube, middle ear cavity, aditus, antrum and mastoid air cells. Pooling of contrast in the right nasal cavity and ethmoid cells was also seen. A 49-year-old female patient with history of right rhinorrhea with features of hyper-pneumatization of sphenoid bone involving right greater wing of sphenoid bone and bilateral pterygoid process with a bony defect in the right greater wing of sphenoid was demonstrated on CT. Corroborative magnetic resonance imaging (MRI) brain Constructive interference in steady state (CISS) sequence revealed a meningoencephalocele. Additionally, a suspicious focal dehiscence was observed in the right cribriform plate CSF pockets herniating into right ethmoid sinus. Hyper-pneumatized petrous bone and paranasal sinuses can be attributed as a risk factor for formation of spontaneous CSF leaks.


Background
Cerebrospinal fluid (CSF) fistulas are classified based on their etiology into congenital and acquired fistulas. The acquired fistulas are further classified as nontraumatic, traumatic and spontaneous [1,2].
Spontaneous CSF leaks may arise from defects in the skull base, in relation to the Tegmen tympani, Tegmen mastoideum, sigmoid sinus and posterior semicircular canal. Leaks in these locations cause indirect CSF rhinorrhoea implying a communication of the subarachnoid space with the middle ear cavity. A constellation of extensive pneumatization, arachnoid pits and empty sella plays a role in the pathogenesis of sphenoid sinus fistulae [3] apart from congenital skull base defects. Arachnoid granulations in relation to the temporal bone are another *Correspondence: Divya Vishwanatha Kini drkinidivya@gmail.com Department of Radiology, JSS Hospital, Ramanuja Road, Agrahara, Mysore 570004, Karnataka, India contributing factor for spontaneous CSF otorrhea and rhinorrhoea [1]. Detection of these defects is best done with computed tomography (CT) and CT Cisternography with an additional role of MRI in assessment of associated parenchymal herniation with utmost accuracy. In this study, we wish to highlight the existence of spontaneous CSF leak as an entity associated with hyper-pneumatized petrous bone and sphenoid sinus.

Case 1
A 26-year-old female patient presented with history of right sided rhinorrhoea for 1 month. There was no history suggestive of meningitis, trauma, visual blurring. Patient was conscious, oriented. No meningeal signs were seen. Fundoscopy was normal. Clinically patient appeared of appropriate built and nourishment for age and gender. Blood work-up revealed no significant abnormality. No previous similar complaints were noted in the past.
Plain CT study showed hyper-pneumatization of the petrous bones with associated soft tissue thickening within (Fig. 1). In addition, a bony defect of ~ 2 mm with adjacent bone scalloping involving the anterolateral aspect of petrous part of right temporal bone ( Fig. 1) was demonstrated. CT cisternography showed contrast opacified CSF extending into the pneumatized petrous apex cells, Eustachian tube, middle ear cavity, aditus, antrum, mastoid air cells (Fig. 2). However, no contrast opacification of cochlea and vestibular apparatus was observed. Pooling of contrast in the right nasal cavity and ethmoidal air cells was also seen (Fig. 2). Corresponding MRI brain revealed features of right mastoiditis extending to involve the pneumatized petrous bone (Fig. 3). Diagnosis of right petrous defect with paradoxical CSF rhinorrhea was hence made.
Follow up: Patient underwent right temporal craniotomy and CSF leak repair. Intra-operatively, the petrous bone was thin with cobble stoning and few areas of defect over the petrous ridge. A defect of 1 × 1 cm was seen toward the lateral aspect close to the anterior petrous ridge. A temporal muscular pedicle flap was used as a covering fascia. Patient tolerated the procedure well and was discharged with no CSF leak. Patient is currently asymptomatic.

Case 2
A 49-year-old female patient presented with history of right sided watery nasal discharge over the last 15 days. There was no history suggestive of meningitis, trauma or visual blurring. Patient had no other comorbidities with no similar prior complaints. Clinically patient appeared of appropriate built and nourishment for age and gender. Blood work-up revealed no significant abnormality.
On HRCT of the temporal bone, hyper-pneumatization of sphenoid bone involving right greater wing of sphenoid bone and bilateral pterygoid processes was noted (Fig. 4). Also, bony defect measuring 1.4 mm noted in the right greater wing of sphenoid bone (lateral to the foramen rotundum) (Fig. 5). Corroborative MRI brain CISS sequence revealed predominant downward herniation of meninges and adjacent minimal brain parenchyma through the above-mentioned defect into the hyperpneumatized right greater wing of sphenoid bone and right pterygoid process (Fig. 6). Follow up: Patient was lost to follow-up and hence information regarding further management could not be provided.

Discussion
The development of skull base defects is caused by the interaction of CSF pulsation with the bone. The anatomical predisposition differs between the anterior, middle and posterior cranial fossa. Site of CSF leaks in temporal bone in various studies are summarized in Table 1 according to the age of the patients [4][5][6][7][8].
The novelty in the first case lies in the fact that none of these cases presented with a defect in the anterior ridge of the petrous bone. Unlike the thin tegmen tympani and tegmen mastoideum, the anterior ridge of the petrous bone is thicker and does not have any congenital dehiscences. Due to the presence of petrous apex hyper-pneumatization, the otherwise thicker anterior ridge could have given way in response to the raised intracranial pressure. The scalloping of the anterior ridge of petrous bone indicated a chronic process in the background of raised CSF pressure. Other factors for spontaneous CSF leak such as arachnoid granulations defects and idiopathic intracranial hypertension were ruled out in this patient [1]. Multiple defects were detected in the right petrous bone in this case. Difference between the previous cases and this case was that despite having a temporal bone defect, the patient presented with rhinorrhea instead of otorrhea. This indicated that the tympanic membrane was intact. The cause of rhinorrhea could be attributed to paradoxical leak into the nasopharynx via the eustachian tube and further into the ethmoid sinus and nasal cavity. It is very important to highlight here the identification of the middle fossa defect, which if left unrepaired may result in persistent leak. Also emphasizes the role of radiologists in accurate localization of site of leak which decides the appropriate surgical strategy [1,[4][5][6][7][8].
The second case highlighted the association of CSF rhinorrhea with presence of meningoencephaloceles and hyper-pneumatized paranasal sinuses. Literature has thrown light on the likely pathogenesis of development of meningoencephaloceles secondary to CSF fistulas. The presence of raised intracranial pressure has been strongly associated with the presence of spontaneous CSF fistulas-they are seen to promote formation of prominent arachnoid granulations [1,2,9]. Owing to the altered CSF dynamics within these, there is a path of reduced resistance created through which the meninges and cerebral parenchyma tend to easily herniate resulting in creation of a spontaneous CSF fistula [1]. However, the current accepted theory is that of a multifactorial process involving the combination of elevated intracranial pressure and anatomical deficiencies (viz. hyper-pneumatized paranasal sinuses) [10,11] aiding in the development of meningoencephaloceles and spontaneous CSF fistulas [1]. High resolution computed tomography (HRCT) of the paranasal sinuses as well as the temporal bones is a baseline imaging modality which helps identify skull base defects irrespective of the presence of CSF leaks [2,11]. However, the detection of dural defects and differentiation of mucus secretions from actual CSF is not possible. To aid in this, use of additional CT Cisternography and MR cisternography is beneficial. CT Cisternography utilizes intra-thecal injection of iodinated contrast in order to opacify the CSF and thus, it accurately demonstrates the site of leak with visualization of active leak through the defect. MR Cisternography (MRC) utilizes the property of high signal intensity of CSF on heavily T2 weighted images in the subarachnoid spaces with the added advantage of its excellent soft tissue resolution thereby helping detect the presence of an associated dural defect with super-added meningoencephaloceles [2,10,11]. However, bony details maybe sub-optimally evaluated on MRI and hence interpretation of the imaging findings must be performed along with CT. Additional modalities that have been proven to aid in the diagnosis are contrast enhanced MRC and Radionuclide cisternography [10]. At our institute, we follow a protocol consisting of HRCT of the paranasal sinuses and temporal bones followed by MR brain with MR Cisternography (CISS sequences-heavily T2 weighted images). This is further followed by a confirmatory CT Cisternography to demonstrate site of active leak with use of dynamic maneuvers facilitating the leak. As demonstrated in these cases, in the presence of normal intracranial pressure, hyper-pneumatization is a significant risk factor for the development of spontaneous CSF fistulas. Such correlation has not been documented in previous studies. Hence, we tend to propose a novel hypothesis of hyper-pneumatization predisposing to formation of spontaneous CSF leaks. As a part of routine reporting protocol, documentation of presence of hyperpneumatization is essential to estimate the risk of development of such fistulas.

Limitation
The second case was lost to follow-up and thus a confirmation regarding the radiological diagnosis by surgical correlation could not be made.

Conclusions
The recognition of spontaneous CSF leaks is vital and is to be made with accurate detection of the site, etiology and severity of the same as poorly managed cases result in deleterious complications. HRCT of the involved region with further CT cisternography and corroborative MRI is an essential and beneficial imaging protocol. Careful scrutiny of the skull base to look for hyper-pneumatized sinuses and anatomical dehiscences should be performed in a meticulous manner.

CSF
Cerebrospinal fluid CT Computed tomography MRI Magnetic resonance imaging CISS Constructive interference in steady state