MRI in perianal ﬁ stulae

MRI has become the method of choice for evaluating perianal ﬁ stulae due to its ability to display the anatomy of the sphincter muscles orthogonally, with good contrast resolution. In this article we give an outline of the classi ﬁ cation of perianal ﬁ stulae and present a pictorial assay of sphincter anatomy and the MRI ﬁ ndings in perianal ﬁ stulae. This study is based on a retrospective analysis of 43 patients with a clinical diagnosis of perianal ﬁ stula. MRI revealed a total of 44 ﬁ stulae in 35 patients; eight patients had only perianal sinuses.


Introduction
Perianal fi stulae commonly occur in middle-aged men. [1] They are thought to be a result of anal gland obstruction, with secondary abscess formation and external rupture of the abscess. [1] They have traditionally been imaged by conventional fistulograms; the procedure involves cannulation of the external opening and injection of a watersoluble contrast into the fi stula. This method has two main disadvantages: First, the primary tract and its extensions do not fi ll with contrast if they are plugged with pus or debris and, second, the sphincter muscle anatomy is not imaged and hence the relation between the tract, the internal/external sphincter, and the levator ani muscle is not revealed. [2] Transrectal ultrasound bett er depicts fi stulae and their relation to the anal sphincter muscles. The operator dependence, limited fi eld of view and absence of a coronal plane of imaging, however, are its disadvantages. [2] CT fi stulography is limited by the fact that att enuation values of the fi stula tract, the areas of fi brosis, and sphincter muscles are similar to each other. [2] Multidetector row CT fi stulography with its isotropic voxels is expected to improve the results from this modality. [2] The role of MR fi stulography in the preoperative evaluation of perianal fi stulae is now well established. [3][4][5] Normal MRI anatomy of the anal sphincter [ Figure 1] An optimal examination utilizes both endoluminal and external phased-array surface coils. [2] However, imaging with an external coil alone also provides good results. [4,6,7] The external anal sphincter (a striated muscle) is clearly visualized on MRI. It is hypointense on T1W, T2W, and fat-suppressed T2W images, and is bordered laterally by the fat in the ischioanal fossa [ Figure 1 a,b and d].
The coronal images depict the levator ani muscle (levator plane), the identifi cation of which is important to distinguish supralevator from infralevator infections. The puborectalis ring is seen as a thickening of the superior fi bers of the external sphincter [ Figure 1d]. The puborectalis further merges with the levator plate superiorly.

Classifi cation of perianal fi stulae
Depending on the location and course of the primary tract, perianal fi stulae have been classifi ed into four types. [8] 1. Intersphincteric (incidence 60-70%): [9] The infection starts from an anal gland and develops in the inter sphincteric plane, lying between the internal and external sphincters, without penetrating the external sphincter. It eventually ruptures onto the skin, thereby creating the fi stula. 2. Transsphincteric (incidence 20-30%): [9] This occurs when the intersphincteric infection penetrates the external sphincter to reach the ischioanal fossa and, eventually, the perianal skin. 3. Suprasphincteric (uncommon): These fi stulae extend superiorly in the intersphincteric plane to reach above

Materials and Methods
The study population comprised 43 patients whose MRI studies (done between October 2002 and December 2008) were evaluated retrospectively. The patients had been referred to the MRI unit for MR fi stulography.
All MRI studies were carried out on a 1.5-

Results
Of the 43 patients in our study, eight (18%) were identifi ed as having a perianal sinus only, with no fi stula extending into the anal canal. The rest of the 35 cases were evaluated for the site of the primary tract and its ramifi cations, the presence/absence of external sphincter involvement, and the location of the internal openings.
Three patients had a primary or recurrent perianal fi stula with associated Crohn's disease [ Figure 2]. Two of these three cases had multiple fi stulae and all three had abscess formation [ Figure 2d and e]. Of the remaining 32 patients without Crohn's disease, 24 had a primary fi stula and, of these, seven had previously undergone perianal abscess drainage. Eight patients had undergone previous fi stula surgery and had presented with a recurrence.

Discussion
MRI imaging of perianal fi stulae relies on the inherent high soft tissue contrast resolution and the multiplanar display of anatomy by this modality. In one of the early studies on MRI fi stulography, Lunniss et al. reported a concordance rate of 86-88% between MRI and surgical fi ndings. [10] Subsequent studies have suggested that MRI is more sensitive than even surgical exploration of the tract. [5,11] MRI is especially useful in patients with fi stulae associated with Crohn's disease and those with recurrent fi stulae, [3] as these entities are associated with branching fi stulous tracts. Missed extensions are the commonest cause of recurrence. [12] T2W images (TSE and fat-suppressed) provide good contrast between the hyperintense fl uid in the tract and the hypointense fi brous wall of the fi stul a, while providing good delineation of the layers of the anal sphincter. [6,13] In our experience, axial T2W fat-suppressed images were the most useful for locating the fi stulous tract. Gadolinium-enhanced T1W images are useful to diff erentiate a fl uid-fi lled tract from an area of infl ammation. [14] The tract wall enhances, whereas the central portion is hypointense. Abscesses are also very well depicted on post-gadolinium images.
The exact location of the primary tract (ischioanal or intersphincteric) is most easily visualized on axial images; the presence of disruption of the external anal sphincter diff erentiates a transsphincteric fi stula from an intersphincteric one. The internal opening of the fi stula is also best seen in this plane.
As mentioned earlier, coronal images depict the levator plane, thereby allowing diff erentiation of supralevator from infralevator infection. A combination of an axial and a longitudinal series (coronal, sagitt al, or radial) will provide all the necessary details. [15] To summarise, evaluation of an enhanced T1W image, in conjunction with a fat-suppressed T2W image, provides most of the details necessary for accurate evaluation of perianal fi stulae.