Imaging was performed during clinically active CSF leak, as reported by the patient. In all cases, high-resolution CT and contrast-enhanced CT Cisternography were performed on the same day.

CT scan was carried out on 128 slice Somatom definition edge, Siemens Health care, Erlangen., Germany. All the patients initially underwent non-contrast high-resolution CT of paranasal sinuses from the level of hard palate superiorly including the frontal sinuses in cranio-caudal direction using slice thickness of 1 mm. Sub-millimetric reconstructions (0.5 mm) in bone and soft tissue kernels were performed.

Prior to starting the procedures pledgets one in each nostril were introduced. After lumbar puncture, 8-10 mL of an iodinated nonionic low-osmolar contrast agent (iopropamide/iohexol, 300 mg/ml iodine) was instilled into thecal sac. The patients were then placed in knee elbow position with dependent flexed head to provoke CSF leak andcontrast enhanced CT scan was performed with collimation similar to non-contrast CT.

The images were evaluated by two radiologists (7- and 15-years’ experience) and discrepancies if any were settled by consensus.

On plain study note was made of site of visible defects, fractures in case of trauma, presence of pneumoencephalous and sinus opacification /extracranial fluid in communication with CSF.

On contrast study an increase in 50% of HU value of the paranasal sinus fluid /extra cranial fluid as compared to non-contrast study [2], a column of contrast extending from subarachnoid space to the extra cranial space (paranasal sinus) or opacification of the nasal pledget was considered positive for leak. Pooling of contrast material at a fracture site with deformity of a local sulcus as described by Fagerlund and Liliequist [5] was considered positive. Additonally, pneumoencephlous in communication with para nasal sinus air was also considered positive for CSF leak.

The demographic data, clinical presentation, treatment details of surgical or conservative treatment were evaluated and tabulated.

RESULTS

We retrospectively evaluated 14 patients who underwent CT Cisternography from the year 2013 to 2016. The medical records, surgical details, investigations and CT Cisternography images were evaluated. Of the 14 patients 9 were females mean age was 45.5 with age ranging from 29 years to 56 years. The procedure was tolerated well in most patients with minor headache controlled by NSAIDS and head rising in most, while the headache was severe in one patient warranting admission for 2 days post procedure. The demographic distribution, imaging and surgical details are as depicted in Tables 1 and 2.

All patients presented with typical history of persistent clear nasal discharge, which could not be sniffed back, and which increased on forward bending. The duration of symptoms ranged from 1 day to one and a half year. 2 of 14 patients one with spontaneous leak and the other with post traumatic leak had signs and symptoms suggestive of meningitis. The cause was non traumatic in 11 of 14 patients, while in others the cause was trauma accidental (n=3) or iatrogenic (n=1). Over all the cribriform plate and sphenoid sinuses were equally involved (n=5).

Of the patients with non-traumatic leak, thecribriform plate (n=5/11) was most commonly involved, the next common site was sphenoid sinus (n=4/10). Left cribriform plate was more frequently involved than left (n=3/5). In all patients with leak fromcribriformplate the defect was clearly appreciated and a continuous column of contrast extending from the subarachnoid space to the extra cranial space (intofrontoethmoidal recess) was noted with opacification of nasal swab (Figure 1). One patient had two sites of leak from both the Fovea Ethmoidalis and Cribriform plate (Figure 2), while in the other two no leak or defect could be appreciated. Of the four cases of sphenoid defect, one patient had defect in the inferolateral wall also known as Sternsberg Canal (Figure 3) and other three in the posterior wall (Figure 4). These leaks were seen as continuous extension of contrast from subarachnoid space to the sinus (Figures 2 and 3). In one patient significant spontaneous penumoencephalus prevented opacification of subarachnoid space. The defect was easily identified on HRCT with subarachnoid air seen communicating with para nasal sinus air (Figure 5). All the patients with cribriform leak or absence of obvious leak on CT Cisternogram were treated conservatively and it subsided spontaneously, while those involving sphenoid sinuses were treated surgically (Table 1).

The post trauma group of the 3 patients comprised of one iatrogenic leak and two post traumatic leaks. One patient presented immediately post trauma and CT Cisterogram was done within a week of trauma. This patient had fracture of anterior and middle cranial fossa and pneumoencephaolus with extra dural extravasation of contrast reaching upto the fracture with deformity of an adjacent sulcus (Figure 6). This patient was treated with fracture fixation and duroplasty. The other patient has past history of road accident and nasal bleed. He presented with one-and-a-half-year history of CSF rhinorrhea. CT Cisternogram showed multiple defects in frontal sinus and roof of sphenoid and ethmoid sinus. Leak was however seen from the roof of ethmoid and sphenoid sinus as increase in density of sinus fluid by

more than 50% (Figure 7). Frontal sinus encephalocele was also noted in this patient (Figure 7). This leak was also surgically treated. We had one patient of iatrogenic cause of CSF rhinorrhea post microsurgery for trigeminal neuralgia. In this patient there was a right occipito-mastoid defect seen with extravasation of contrast into the defect and mastoid aircells, further extension into the middle ear and subsequently into the Eustachian tube and right nasopharynx were seen (Figure 8). In this patient another defect was appreciated in the right lateral wall of sphenoid sinus with adjacent sinus opacification (Figure 8).

DISCUSSION

CSF Rhinorrhea was classified by Oyammainto two groups traumatic (including iatrogenic) and non-traumatic [6]. He further subdivided the latter into nontraumatic with normal pressure and nontraumatic with CSF hypertension. The high-pressure leaks were those in whom the leak was secondary to congenital defects, infections or tumors resulting in raised intracranial tension [6,7]. While the other category that is low pressure leak is now identified as spontaneous leak and is thought to be a manifestation of benign intracranial hypertension considering its association with empty sella [7].

In literature few authors report traumatic CSF leak to be most common [8,9]. However, in our study we found the non-traumatic category to be common, which is consistent with research findings [10,11]. However, this could be because of the fact that plain HRCT is usually done for patients with significant head injury usually immediately after trauma, it usually suffices to detail the site of fractures in such cases and CT Cisternography is not feasible in acute settings.

The commonest site of leak was cribriform plate followed by sphenoid bone which was similar to other findings [8]. This is probably due to the fact that cribriform plate is thinnest and weakest portion of anterior cranial fossa [8,12]. All patients had typical findings of persistent nasal discharge which could not be sniffed back and increased on forward bending. Only two patients presented with meningitis, while hypotensive headache was present in all. Meningitis in the two cases was probably the reason of CSF rhinorrhea. Raised intracranial tension due to meningitis was probably the cause for CSF leak in these patients rather than meningitis being a complication CSF leak.

Patients with CSF rhinorrhea pose a diagnostic challenge for both the otolaryngologists and radiologists. In this era of minimally invasive procedures where small skull base defects are routinely repaired using nasal endoscopes having limited field of view for, accurate localization of the leak is mandatory [3,8]. Also, symptoms in many patients with CSF leak resolve spontaneously or with conservative management [8]. Hence, we see that decision regarding management depends on the location and etiology. Plethora of imaging modalities have been described and discarded in an attempt to find the gold standard. The studies that are available for evaluation of CSF rhinorrhea are radionuclide study, HRCT alone, HRCT with CT cisternography, MRI cisternography with HRCT PNS and recently described CEMR cisternography. Of these the radionuclide study owing to its extremely low specificity is considered obsolete and is reserved only in cases where the diagnosis itself is in question [2,13]. There are few studies which say that HRCT alone suffices [14], but more than one site of leak and more than one defect are not rare. CT Cisternography has overtime undergone a drastic revolution due to the thin slice capabilities of current scanners and use of water-soluble contrast. It combines best of both worlds, that is it provides excellent localization of osseous defect and also shows the actual extra-cranial pooling of contrast. Of late it has lost its appeal with advent of MRI as it is slightly invasive in terms of requirement of lumbar puncture, patient discomfort in the form of post procedural headache, theoretical possibility of opening up of the defects, introducing  infection, lumbar CSF leak and its low sensitivity in detecting the site in inactive leaks [3,15-17]. At present the most recommended modality is MRI Cisternography, HRCT PNS or combination of both which are non-invasive and quick [3,17]. In most of our patients we were able to isolate the site of leak with only mild patient discomfort. Post procedure complication rate was low and only one patient developed significant headache requiring admission.

We found that in patients who presented with recent trauma had multiple fractures on HRCT and pneumocephalus, which could interfere with visualization of leak on MRI. While with Cisternography though we could not see leakage into sinus of leak we could infer the possible site of leak using the distorted sulcal sign. Similarly, in a post-traumatic leak presenting late we found that the site of leak was different from the site of obvious defect or site of sinus opacification. In our only case of iatrogenic CSF leak post otomastoidectomy we additionally found a small defect in lateral wall of sphenoid sinus with adjacent sphenoid sinus opacification. However, on cisternography we found the leak to be into the mastoid air cells and subsequently into the naspharynx via Eustachian tube.

In non-traumatic category we found congenital defects in few patients. Sternberg’s Canal which represents persistent lateral cranio-pharyngeal canal connecting intracranial space and sphenoid sinus due to congenital fusion defect between basi-sphenoid and greater wing of sphenoid was seen in one patient [18]. Additionally, defects in posterior wall were seen in three patients which requires surgical intervention as the sizes of defect were significant. In one patient placement of patient in knee elbow position probably induced spontaneous pneumocephalous, which prevented opacification of intracranial subarachnoid space. However, we could infer the site of leak as the subarachnoid air was seen in communication with sinus air through defect in posterior wall of sphenoid sinus. Other patients in non-traumatic group had leak from cribriform plate (seen as a column of contrast extending from subarachnoid space to extracranial space) and were treated conservatively. Multiple sites of leak from cribriform plate and fovea ethmoidalis were noted in one patient. In two patients no defects or leak could be appreciated probably due to inactivity of the leak during procedure.

Thus, in patients in both traumatic and non-traumatic categories wherein multiple defects, multiple sites of leak are anticipated CT Cisternogram provides concrete evidence regarding site of leak.

The major limitation of our study was lack of MRI for all patients due to which comparative analysis was not possible. Most of the post trauma patients rarely undergo CT Cisterography in acute settings and hence were not included in the study.

CONCLUSION

In conclusion, despite many limitations of CT Cisternography, it still holds importance as a problem-solving tool for detection of exact site of leak in patients where multiple sites of leak or multiple defects are suspected. Additionally, we have demonstrated spectrum of possible imaging findings of CSF leak on CT Cisternography.