, Volume 46, Issue 1, pp 75–80 | Cite as

MRI cisternography with gadolinium-containing contrast medium: its role, advantages and limitations in the investigation of rhinorrhoea

  • K. AydinEmail author
  • K. Guven
  • S. Sencer
  • J. R. Jinkins
  • O. Minareci
Diagnostic Neuroradiology


Our purpose was to evaluate the utility of intrathecal gadopentetate dimeglumine -enhanced magnetic resonance cisternography (GdMRC). We injected 0.5 ml contrast medium into the subarachnoid space via lumbar puncture in 20 patients with suspected cerebrospinal fluid (CSF) rhinorrhoea. MRC showed CSF leakage in 14 patients with rhinorrhoea at the time of the examination, into the ethmoid air cells in nine, the sphenoid sinus in three and the frontal sinus in two cases. In 12 of these the site leakage was confirmed during surgical repair of the fistula. No leakage was observed in four patients with intermittent rhinorrhoea, not present at the time of the examination. GdMRC showed leakage in two patients with negative CT cisternography. GdMRC may prove better than CT cisternography, especially with slow CSF flow. We also showed low-dose GdMRC to be a feasible and relative safe way of confirming the presence of and localising active CSF leaks prior to surgical repair.

Key words

Cerebrospinal fluid rhinorrhoea Magnetic resonance imaging Cisternography Gadopentetate dimeglumine 


Leakage of cerebrospinal fluid (CSF) through a defect in the meninges and skull base into the paranasal sinuses and nose, CSF rhinorrhoea, is classified as traumatic or nontraumatic (i.e., spontaneous) [1]. The majority of cases are traumatic: CSF rhinorrhoea develops in 2% of patients with cranial trauma [2, 3]. Although nontraumatic cases are less common, the likelihood of spontaneous closure is less than that in traumatic cases. If spontaneous closure does not occur, operative repair of the meningeal defect must be performed because of the high risk of recurrent infectious meningitis. Precise localisation of the fistula before surgery increases the rate of successful repair and decreases the duration of the operative procedure.

Accurate localisation of the site of the CSF fistulae is challenging for neuroradiologists as well as for surgeons. High-resolution CT (HRCT) of the paranasal sinuses, CT with intrathecal water-soluble iodinated contrast medium (CTC), and MR cisternography (MRC) without contrast medium are the most commonly used radiological methods [4, 5, 6, 7, 8, 9]. HRCT shows bony defects in the skull base which may or may not be the site of leakage. Congenital or traumatic bony defects in the absence of an accompanying meningeal defect may result in false positive diagnoses. The presence of soft tissue or fluid within the sinuses adjacent to a fracture is not a specific sign of ongoing CSF leakage. Extensive and/or multiple fractures of the walls of the paranasal sinuses adjacent to CSF-containing spaces at the skull base may lead to inconclusive examinations.

Unenhanced T2-weighted images have been used in an attempt to localise the CSF fistula. Demonstration of high signal intensity fluid extending from the subarachnoid space directly into adjacent paranasal sinuses, or herniation of brain into a sinus through a bone defect, have been the principal diagnostic criteria [4, 10, 11, 12, 13]. But, as in HRCT, the criteria are not specific. Pre-existing paranasal sinus inflammation has been reported to lead to false-positive diagnoses of CSF fistula with intrathecal iodine-based contrast-enhanced MRC [14].

The principal test for reliable diagnosis is intrathecal contrast-enhanced CTC. Although the risk of serious complications is minimal, it is not without side-effects. Allergic reactions, and rare intracranial haemorrhage have been reported after intrathecal injection [15, 16]. In many institutions, unenhanced HRCT is performed before CTC, in the perhaps misguided belief that this increases accuracy. However, this procedure also doubles the radiation exposure. Also, even though a head-down position is used, some patients require delayed images, which further increases radiation exposure.

Our purpose is to report our experience of our institution with a newly initiated diagnostic method for diagnosis of CSF fistula: MR cisternography with intrathecal gadolinium-containing contrast medium (GdMRC).

Materials and methods

We studied 20 patients aged 19–56 years (12 men, eight women) with clinically suspected CSF rhinorrhoea and no history of epilepsy, central nervous system neoplasia or neurodegenerative disease. The protocol was approved by the Institutional Ethics Committee. Informed consent was obtained.

There were 11 patients who had cranial trauma preceding their rhinorrhoea; the time between the trauma and imaging ranged from 1 week to 10 months. Nine patients presented with spontaneous CSF rhinorrhoea. Seven patients had intermittent rhinorrhoea and six were “dry” at the time of the GdMRC. Three patients had episodes of bacterial meningitis preceding our study; in two of whom this occurred after negative CTC 4 and 6 months before the GdMRC. The meningitis was in both cases treated successfully, and no patient had signs of meningitis at the time of the intrathecal injection.

MRI was were performed at 1.0 and 1.5 tesla. Coronal and sagittal 3 mm T1-weighted sections with and without fat suppression were obtained prior to the GdMRC. The patients then underwent lumbar puncture using a 21 G needle. After 5 ml CSF was, 0.5 ml undiluted Magnevist™ was injected into subarachnoid space. The contrast medium was not mixed with CSF so that we could follow it fluoroscopically. Following this a small amount of the CSF was reinjected to fill the dead space of the needle and ensure that the whole 0.5 ml of contrast medium was injected. The patients were positioned prone, 30–40 ° head down for 15 min after withdrawal of the needle to maximise accumulation of contrast medium in the basal cisterns and to attempt to provoke its passage into any CSF fistula. T1- and T1-weighted fat-suppressed images with the same parameters as previously were obtained with the patient prone, chin-up. In two patients there was insufficient intracranial contrast medium for analysis, and delayed images were obtained 30 min later.

All patients were observed hourly by a neuroradiologist for 12 h after the examination. Hourly checks were then made up to 24 h on the ward to assess for gross behavioural or cognitive changes, neurological impairment, subjective complaints and vital signs, as well as for more serious adverse effects (seizures or anaphylactoid reactions). Follow-up was also carried out 24  h and 1 and 3 months after the cisternograms.

Coronal HRCT of the paranasal sinuses and skull base and/or temporal bone was performed preceding the GdMRC, to give anatomical bony land marks to the surgeons in the patients with positive GdMRC. Slice thickness was 1–3 mm, field of view 160–180 mm, with a 512–256×512–256 matrix.


No patient developed any evidence of allergic reaction, convulsion, behavioural changes, neurological deficit or functional impairment during follow-up. Vital signs of all patients were stable during the first 12 h of observation, with no change from pre-procedure measurements. Four patients complained of headache within 12 h of the intrathecal injection, which resolved within 24 h with bed rest.

GdMRC indicated the site of CSF leakage in 16 of the 20 patients, into the sphenoid sinus in three, the ethmoid air cells in ten and the frontal sinuses in three (Fig. 1). In one patient leakage was into a nasal cephalocele, through a congenital defect in the cribriform plate (Fig. 2). All delayed GdMRC images were positive (Fig. 3). In the 14 of the patients with positive GdMRC, meningeal tears were confirmed at surgery and successfully repaired. One patient with positive GdMRC refused an operation, and another was referred to another hospital.
Fig. 1a–c

A 41-year-old woman with spontaneous rhinorrhoea. a Coronal high-resolution (HR) CT shows no abnormality; there is no obvious inflammatory change in the ethmoid cells. b, c Coronal contrast-enhanced MRI cisternography (CEMRC): T1- weighted images without and with fat saturation reveal leakage of contrast medium into the nasal cavity on the right ( arrows)

Fig. 2a–e

A 21-year-old man with intermittent spontaneous rhinorrhoea. a, b Coronal and sagittal T1-weighted images reveal a cystic mass in the left nasal cavity, with nodular high-signal soft tissue within it. c Sagittal T2-weighted image shows a septate cyst. d, e Coronal and sagittal GdMRC demonstrates leakage of contrast medium into the nasal cephalocele

Fig. 3a–c

A 27-year-old man with traumatic rhinorrhoea. a Coronal GdMRC acquired immediately after subarachnoid injection shows encephalomalacia in the right gyrus rectus and medial orbital gyrus. There is reflux of contrast medium into the ventricles, but none is visible in the frontal subarachnoid space. b, c Delayed coronal GdMRC without and with fat saturation demonstrates contrast medium in the superficial frontal subarachnoid space and leakage into the ethmoid sinus ( arrows)

The four patients with negative GdMRC have been followed conservatively . All had intermittent leaks and none had active rhinorrhoea at the time of the GdMRC. In one of these patients, T2-weighted images showed an empty sella turcica and a high-signal cyst in the sphenoid sinus, suspicious for CSF leakage (Fig. 4). GdMRC did not show a CSF leak, but confirmed the empty sella turcica.
Fig. 4a, b

A 32-year-old woman with clinically suspected spontaneous rhinorrhoea. a Coronal T2-weighted image shows a partially empty sella and complete opacity of the sphenoid sinus. b There is no evidence of leakage into the sphenoid sinus on coronal GdMRC

Two patients with negative CTC had multiple skull-base fractures. In both the site of the CSF fistula was shown on GdMRC (Fig. 5). HRCT showed a fractures or a congenital bone defect at the site of leakage in all cases.
Fig. 5a, b

A 28-year-old man with suspected rhinorrhoea following a traffic accident. a Coronal CT cisternography reveals broad dehiscence of the cribriform plate. There is no leakage of contrast medium into the paranasal sinuses. b Coronal GdMRC shows leakage through the dehiscence ( arrow)


Di Chiro et al. [17] reported the use of intrathecal gadolinium-containing contrast medium in the investigation of CSF leaks in dogs . Another animal study showed the practicability of GdMRC in the demonstration of the site of surgically induced nasoethmoidal CSF fistulae [18]. GdMRC has been reported to be accurate in preoperative localisation of CSF fistulae in patients with suspected CSF rhinorrhoea [19].

In two published reports, the adverse effects of intraventricular gadopentetate dimeglumine and gadodiamide in animals included behavioural and neurological (focal seizures, ataxia, delayed tremor) and histological changes (loss of oligodendroglia, hypertrophy of astrocytes, formation of eosinophilic granules) [20, 21]. However, these were seen only following intraventricular injection of relatively high doses; they were not seen when the total dose of intraventricular gadolinium-containing contrast medium was less than 15 μl (3.3 μmol/g brain) [22]. A recent animal study with low intrathecal doses of a different compound (Omniscan™) did not reveal effects of this type [22].

In the first prospective human trial, no significant gross neurological, CSF, EEG or MRI changes related to the intrathecal contrast medium were seen on initial examination follow-up [23]. Using a total volume of 0.5–1.0 ml, the estimated intrathecal dose per gram of brain in this trial was much less than in the animal experiments (0.07–0.36 vs 2.5–15 μmol/g brain) [17, 23]. Other isolated clinical reports have shown the anodyne nature of the technique [24].

In addition to the sequences reported in prior studies of GdMTC, we used fat-suppressed T1-weighted sequences. Without any formal analysis, we believe that fat suppression increases the conspicuity of contrast medium leakage due to the saturation of medullary bone fat in the skull base and nasal cavity, which might conceivably be confused with leakage.

MRI without contrast medium has been used for investigation of CSF fistulae. Shetty et al. [25] claimed an accuracy of 89%, but Hegarty et al. [14] reported a relatively high (40%) false-positive rate. The presence of inflammatory soft tissue in the paranasal sinuses increases the rate of false-positive studies; false negatives are also reported [14].

Six of our patients, of whom two had positive GdMRC, have not been operated upon. We therefore cannot calculate the sensitivity and specificity of GdMRC. However, there were no false positives in the operated patients, so that the accuracy of GdMRC in this group was 100%.

CTC is accepted as the most accurate method for investigation of active cranial CSF leakage, but its disadvantages and shortcomings have already been discussed.

GdMRC imaging is better than CTC for demonstration of the soft tissues and brain parenchyma. In one of our patients, GdMRC demonstrated the communication between a nasal cephalocele sac and the cranial cavity and the contents of the cephalocele, which was a dermoid, not heterotopic glial tissue. Had CTC been chosen in this case, MRI would also have been needed to show the contents of the cephalocele and possibly its relationship to the frontal lobes.

There are very few reports of patients with rhinorrhoea having both CTC and GdMRC [26]. Although there were considerable intervals between the two examinations, in two of our patients a positive GdMTC followed a negative CTC. Both had inflammatory changes in the mucosa of the paranasal sinuses adjacent to a bone defect. Mucosal inflammation might have decreased leakage of CSF and caused a negative CTC. The greater viscosity of iodinated contrast medium might limit its subarachnoid distribution and leakage, through the inflamed mucosa, into the sinuses. Iodinated contrast media and bone are both dense on CT, so that a minute amount of leakage might not be detected, while gadolinium-containing contrast media give good contrast on fat-suppressed images.

GdMRC combines the contraindications of both lumbar puncture (e.g., meningitis) and MRI (e.g., metallic foreign bodies, pacemakers). High-signal blood products might cause nondiagnostic or false positive on T1-weighted images in patients with paranasal sinus haemorrhage following trauma, in whom GdMRC may not be the ideal examination. However, many patients do not need to be investigated in the acute/subacute period following trauma, because most traumatic CSF leaks seal spontaneously. The best time for CSF leak analysis by GdMRC is in the chronic phase, when the methaemoglobin has evolved into a nonparamagnetic form.

A final caveat: it must be noted that intrathecal administration of gadolinium-containing contrast media is not currently approved worldwide.


  1. 1.
    Park J, Strelzow V, Friedman W (1983) Current management of cerebrospinal fluid rhinorrhea. Laryngoscope 93: 1924–1300Google Scholar
  2. 2.
    Stafford-Johnson DB, Brennan P, Toland J, O’Dwyer AJ (1996) Magnetic resonance imaging in the evaluation of the cerebrospinal fluid fistula. Clin Radiol 51: 837–841PubMedGoogle Scholar
  3. 3.
    Colquhoun IR (1993) CT cisternography in the investigation of cerebrospinal rhinorrhoea. Clin Radiol 47: 403–408PubMedGoogle Scholar
  4. 4.
    Manelfe C, Cellerier P, Sobel D, Prevost C, Bonafé A (1982) Cerebrospinal fluid rhinorrhea: evaluation with metrizamide cisternography. Am J Roentgenol 138: 471–476Google Scholar
  5. 5.
    Nickaus P, Dutcher PO, Kido DK, Hengerer AS, Nelson CN (1998) New imaging techniques in the diagnosis of cerebrospinal fluid fistula. Laryngoscope 98: 1065–1068Google Scholar
  6. 6.
    Lloyd MNH, Kimber PM, Burrows EH (1994) Posttraumatic cerebrospinal fluid rhinorrhoea: modern high-resolution computed tomography is all that is required for effective demonstration of the site of leakage. Clin Radiol 49: 100–103PubMedGoogle Scholar
  7. 7.
    Ahmadi J, Weiss MH, Segall HD, Schultz DH, Zee CS, Giannotta SL (1985) Evaluation of cerebrospinal fluid rhinorrhea by metrizamide computed tomography. Neurosurgery 16: 54–59PubMedGoogle Scholar
  8. 8.
    Drayer BP, Wilkins RH, Boehnke M, Horton JA, Rosenbaum AE (1977) Cerebrospinal fluid rhinorrhea demonstrated by metrizamide CT cisternography. Am J Roentgenol 129: 149–151Google Scholar
  9. 9.
    Stone JA, Castillo M, Neelon B Mukherji SK (1999) Evaluation of CSF leaks: high-resolution CT compared with contrast-enhanced CT and radionuclide cisternography. AJNR 20: 706–712Google Scholar
  10. 10.
    Eljamel MS, Pidgeon CN, Toland JB, Phillips J, O’Dwyer AJ (1994) MRI cisternography and the localization of CSF fistulae. Br J Neurosurg 8: 433–437PubMedGoogle Scholar
  11. 11.
    El Gammal T, Brooks BS (1994) MR cisternography: initial experience in 41 cases. AJNR 15: 1647–1656Google Scholar
  12. 12.
    Eberhardt KEW, Hollenbach HP, Deimling M, Tomandl BF, Huk WJ (1997) MR cisternography: a new method for the diagnosis of CSF fistulae. Eur Radiol 7: 1485–1491PubMedGoogle Scholar
  13. 13.
    Muftah S, Eljamel M, Christopher N, et al (1994) MRI cisternography, and the localization of CSF fistula. Br J Neurosurg 8: 433–437PubMedGoogle Scholar
  14. 14.
    Hegarty SE, Millar JS (1997) MRI in the localization of CSF fistulae: is it of any value? Clin Radiol 52: 768–770PubMedGoogle Scholar
  15. 15.
    Sand T, Myhr G, Stovner LJ, Dale LG (1990) Side effects after lumbar iohexol myelography: relation to radiological diagnosis, sex and age. Neuroradiology 31: 523–528PubMedGoogle Scholar
  16. 16.
    Van de Kelft E, Bosmans J, Paziel P, Van Vyve PM, Selosse P (1991) Intracerebral hemorrhage after lumbar myelography with iohexol: report of a case and review of the literature. Neurosurgery 28: 570–574PubMedGoogle Scholar
  17. 17.
    Di Chiro G, Knop RH, Girton MR, et al (1985) MR cisternography and myelography with Gd-GTPA in monkeys. Radiology 157: 373–377PubMedGoogle Scholar
  18. 18.
    Jinkins JR, Williams RF, Xiong L (1999) Evaluation of gadopentetate dimeglumine magnetic resonance cisternography in an animal model. Invest Radiol 34: 156–159CrossRefPubMedGoogle Scholar
  19. 19.
    Jinkins JR, Rudwan M, Krumina G, Tali T (2002) Intrathecal gadolinium-enhanced MR cisternography in the evaluation of clinically suspected cerebrospinal fluid rhinorrhea in humans: early experience. Radiology 222: 555–559PubMedGoogle Scholar
  20. 20.
    Ray DE, Cavanagh JB, Nolan CC, Williams SCR (1996) Neurotoxic effects of gadopentetate dimeglumine: behavioral disturbance and morphology after intracerebroventricular injection in rats. AJNR 17: 365–373Google Scholar
  21. 21.
    Ray DE, Holton JL, Nolan CC, Cavanagh JB, Harpur ES (1998) Neurotoxic potential of gadodiamide after injection into the lateral cerebral ventricle of rats. AJNR 19: 1455–1462Google Scholar
  22. 22.
    Skalpe IO, Tang GJ (1997) Magnetic resonance imaging contrast media in the subarachnoid space: a comparison between gadodiamide injection and gadopentetate dimeglumine in an experimental study in pigs. Invest Radiol 32: 140–148PubMedGoogle Scholar
  23. 23.
    Zeng Q, Xiong L, Jinkins JR, Fan Z, Liu Z (1999) Intrathecal gadolinium (gadopentetate dimeglumine)-enhanced MR myelography: a pilot study in human patients. Am J Roentgenol 173: 1109–1115Google Scholar
  24. 24.
    Siebner HR, Grafin von Einsiedel H, Conrad B (1997) Magnetic resonance ventriculography with gadolinium DTPA: report of two cases. Neuroradiology 39: 418–422PubMedGoogle Scholar
  25. 25.
    Shetty PG, Shroff MM, Sahani DV, Kirtane MV(1998) Evaluation of high-resolution CT and MR cisternography in the diagnosis of cerebrospinal fluid fistula. AJNR 19: 633–639Google Scholar
  26. 26.
    Wenzel R, Leppien A (2000) Gadolinium-myelocisternography for cerebrospinal fluid rhinorrhoea. Neuroradiology 42: 874–880PubMedGoogle Scholar

Copyright information

© Springer-Verlag 2004

Authors and Affiliations

  • K. Aydin
    • 1
    • 2
    Email author
  • K. Guven
    • 3
  • S. Sencer
    • 2
  • J. R. Jinkins
    • 4
  • O. Minareci
    • 2
  1. 1.CamlikyoluCamlikyolu, B. Mehmetpasa sokak Yavuz Apartment 10/10Etiler, IstanbulTurkey
  2. 2.Neuroradiology Division, Department of RadiologyIstanbul University Medical SchoolCapa, IstanbulTurkey
  3. 3.Department of RadiologyIstanbul University Medical SchoolCapa, IstanbulTurkey
  4. 4.Department of Radiological SciencesMedical College of Pennsylvania-Hahnemann, Drexel UniversityPhiladelphiaUSA

Personalised recommendations