Imaging of Fistula in Ano1 STATE of the ART Ⅲ

Imaging of Fistula in Ano1 STATE OF THE ART Ⅲ

Steve Halligan, MD, FRCP, FRCR Fistula in ano is a common condition that often recurs Jaap Stoker, MD, PhD despite seemingly adequate surgery, usually because of infection that was missed at surgery. It is now increasingly recognized that preoperative imaging can help identify REVIEWS AND COMMENTARY infection that would have otherwise gone unidentified. In particular, magnetic resonance (MR) imaging findings have been shown to influence surgery and markedly diminish the chance of recurrence; thus, preoperative imaging will become increasingly routine in the future. In this article, the authors describe the pathogenesis, classification, and imaging of fistula in ano, with an emphasis on MR imaging. Most important, the authors describe how the radiologist is well placed to answer the surgical riddles that must be solved for treatment to be effective.

௠ RSNA, 2006

1 From the Department of Specialist Radiology, University College Hospital, Level 2, Podium, 235 Euston Rd, Lon- don NW1 2BU, England (S.H.), and Department of Radiol- ogy, Academic Medical Center, Amsterdam, the Nether- lands (J.S.). Received June 11, 2004; revision requested August 24; revision received October 15; accepted Janu- ary 10, 2005; ﬁnal version accepted March 2; updated September 29. Address correspondence to S.H.

஽ RSNA, 2006

18 Radiology: Volume 239: Number 1—April 2006 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

istula in ano (anal fistula) is a com- der surrounded by two muscular crypts that lie proximal to the valves. mon condition that has a tendency sphincters, the internal and external These glands were first linked to the Fto recur despite seemingly ade- anal sphincters, which are composed of genesis of fistula in ano by Chiari (3), quate surgery. Recurrence is usually smooth and striated muscle, respec- who suggested that they were the due to infection that has escaped surgi- tively (Fig 1). The external sphincter source of infection. Their purpose is un- cal detection and has thus gone un- has posterior attachments to the ano- clear, although they may help lubricate treated. It is now increasingly recog- coccygeal ligament and anterior attach- the anus by secreting mucus into the nized that preoperative imaging, nota- ments to the perineal body and urogen- anal crypts. The origin of the anal bly magnetic resonance (MR) imaging, ital diaphragm (and bulbocavernous glands within the surrounding tissues is can help identify infected tracts and ab- muscle in boys and men) and merges variable. For example, they are present scesses that would otherwise have been proximally with the slinglike puborec- in the subepithelium and may be missed. Preoperative MR imaging find- talis muscle (which defines the anorec- present in the internal sphincter, and ings have been shown to influence sub- tal junction), which itself merges with approximately one- to two-thirds of sequent surgery and markedly diminish the levator plate of the pelvic floor. these glands are deeply sited within the the chance of recurrent disease as a The internal sphincter is the distal intersphincteric space (4) (Fig 1). Most result. Because of this, preoperative im- termination of the circular muscle of the authorities believe that it is infection of aging is likely to become increasingly gut tube. The rectal longitudinal smooth these intersphincteric glands that is the routine in the future, especially in pa- muscle interdigitates between the inter- initiating event in fistula in ano, in a tients with recurrent disease. nal and external sphincters and is process known as the “cryptoglandular This review will detail the pathogen- thought to have no obvious sphincteric hypothesis” (5). Furthermore, lym- esis of fistula in ano, explain how patho- effect; rather, its role is probably to phoid aggregates surround the anal genesis causes the different types of fis- bind the anus together (1). glands, which may partly explain the in- tula encountered, and describe how The intersphincteric space is the creased incidence of anal fistula in these types can be imaged, with the em- surgical plane of dissection between the Crohn disease (6,7). phasis on MR. Most important, we will internal and external sphincters and is It is believed that gland infection re- describe how the radiologist is well most frequently found between the lon- sults in an intersphincteric abscess if the placed to answer the surgical questions gitudinal muscle and external sphincter, draining duct becomes blocked by in- that must be solved for treatment to be where it exists as a sheet of fat contain- fected debris. This abscess may resolve effective. ing loose areolar tissue. The fat-filled by means of spontaneous drainage into ischioanal fossa lies lateral to the the anal canal or may progress to an sphincter complex and is traversed by a acute anorectal abscess, which is a com- Anatomy and Etiology network of fibroelastic connective tissue mon acute surgical emergency and is To fully understand the role of imaging fibers. We prefer the term ischioanal familiar to all general and coloprocto- with regard to fistula in ano, an appreci- fossa to ischiorectal fossa because this logic surgeons (8). Treatment generally ation of its etiology and how the various space predominantly surrounds the consists of incision and drainage of the fistula types are defined by anatomic anus rather than the rectum. However, most fluctuant part of the abscess; how- boundaries is mandatory. the two terms are interchangeable. ever, this procedure does not pay due The anal canal is essentially a cylin- With regard to the lining of the anal attention to the source of infection in canal, the proximal half is characterized the intersphincteric space, with the re- by longitudinal mucosal folds, the anal sult that as many as 87% of patients Essentials columns of Morgagni (2). The distal as- with an acute abscess may subsequently Ⅲ Recurrent fistula in ano is usually pect of each column is linked to its develop a fistula (9). Acute anorectal caused by infection that was neighbor by a small semilunar fold (the abscess and fistula in ano are, there- missed during surgical explora- anal valves), which in turn forms small fore, generally believed to be acute and tion. pockets (the anal sinuses, or crypts of chronic manifestations, respectively, of Ⅲ MR imaging depicts remote foci of Morgagni). The distal undulating limit of the same disease. Because of this, the infection better than any other these valves is the dentate (pectinate) search for intersphincteric infection and modality, including surgical explo- line, which also marks the most distal an anal canal internal opening followed ration. aspect of the anal transitional zone, a Ⅲ MR image–guided surgery helps histologic junction between anal squa- Published online reduce postoperative recurrence mous epithelium and rectal columnar 10.1148/radiol.2391041043 by 75% in patients with complex epithelium. disease. The dentate line lies approximately Radiology 2006; 239:18–33 Ⅲ Anal endosonography is a viable 2 cm proximal to the anal verge and is a Abbreviations: alternative to MR imaging when crucial landmark in fistula in ano be- EUA ϭ examination under anesthesia the latter is not available. cause the anal glands empty into the STIR ϭ short tau inversion recovery

Radiology: Volume 239: Number 1—April 2006 19 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

by treatment of these at the time of more tortuous than others, and by pene- importance, this fistula does not pene- acute manifestation has been advocated trating and involving the muscles of the trate the adjacent external sphincter, in an attempt to reduce the incidence of anal sphincter and surrounding tissues to which forms a relative barrier to spread. subsequent fistula (10). a variable degree. Fistulas may thus be However, some truculent fistulas can Fistula in ano develops when an in- classified according to the route taken by cross the external sphincter and reach the tersphincteric infection is allowed to this “primary tract” that links the internal ischioanal fossa by doing so (Fig 2). This continue unabated. It has a prevalence and external openings. results in a transsphincteric fistula, which of approximately 0.01% and predomi- Furthermore, classification largely composed 30% of cases in the series of nantly affects young adults (11). Fistula determines treatment. In 1934, Milligan Parks et al. Other fistulas may spread up- in ano is commoner in men, who domi- and Morgan (14) stressed the importance ward in the intersphincteric space and nate in all published series, with a male- of the “anorectal ring” (anatomically, the arch over the puborectalis muscle, where to-female ratio of approximately 2:1 puborectalis muscle) and categorized fis- they must cross the levator plate to reach (11). Patients most commonly present tulas as those that entered the anal canal the perianal skin. This type, the supras- with discharge (65%), but local pain above or below this structure, warning phincteric fistula (Fig 2), composed 20% due to inflammation is also frequent that postoperative incontinence was of cases in the series of Parks et al. (12). However, some fistulas may be highly likely if high fistulas were surgically Parks et al (6) also noted a fourth entirely asymptomatic. divided without due attention. This classi- type of fistula in 5% of cases. This type fication was subsequently modified and was characterized by the surprising ab- refined by other authors, but the most sence of intersphincteric infection. In- Fistula Classification comprehensive and practical classifica- stead, the fistula entered the rectum or By definition, a fistula is an abnormal tract tion in use today is that of Parks et al (6), anorectal junction directly (Fig 2). that connects two epithelial surfaces. The who carefully analyzed a consecutive se- Parks et al termed these extrasphinc- anatomic course of an anal fistula will be ries of 400 patients referred to the sur- teric fistulas. Clearly, infection of the dictated by the location of the infected geons of St Mark’s Hospital, London, En- anal glands cannot explain this type of anal gland and the anatomic planes and gland. Parks et al found that they were fistula, and Parks et al stressed that pri- boundaries that surround it. There will able to assign all fistulas into one of four mary rectal or pelvic disease (eg, diver- usually be an internal enteric opening in groups; intersphincteric, transsphinc- ticular disease, rectal Crohn disease, the anal canal at the level of the dentate teric, suprasphincteric, and extrasphinc- carcinoma) should be sought when this line—that is, at the original site of the teric (6). Of importance, most of these type was encountered. It should be duct draining the infected gland. In most groupings could be explained in terms of noted that the Parks et al series inevita- cases this is at the 6-o’clock position, be- the cryptoglandular hypothesis. bly suffered spectrum bias owing to the cause anal glands are more abundant pos- The path of least resistance for fester- specialized nature of St Mark’s Hospital, teriorly (radial positions around the anus ing intersphincteric infection is straight a bias acknowledged by Parks himself, are referenced with respect to a clock down the intersphincteric space, which with the result that complex fistulas face, with 12 o’clock being directly ante- creates an intersphincteric fistula; this were almost certainly overrepresented. rior) (13). The fistula can reach the peri- type of fistula composed 45% of cases in For example, Parks et al did not de- anal skin by a variety of routes, some the series of Parks et al (6) (Fig 2). Of scribe submucosal fistulas, which are

Figure 1 Figure 2

Figure 1: Illustration of anal canal anatomy in the coronal plane. EAS ϭ external anal sphincter, IAS ϭ internal anal sphincter, IS ϭ intersphincteric space, LA ϭ levator ani, LM ϭ longitudinal Figure 2: Illustrations in coronal plane show classiﬁcation of ﬁstula in ano according to Parks et al (6): muscle, PR ϭ puborectalis muscle. intersphincteric, transsphincteric, suprasphincteric, and extrasphincteric.

20 Radiology: Volume 239: Number 1—April 2006 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

very superficial and do not involve the means of surgical incision, usually by ized the importance of the puborectalis sphincter at all. cutting down onto a metal probe that muscle for continence as long ago as While most fistulas probably start as has been inserted into the primary 1934—“Whereas, if this ring be cut, a simple single primary tract, unabated tract. Indeed, James Syme (in 1838) loss of control surely results”—they sug- infection may result in ramifications thought laying open a fistula to be gested that most of the remaining that branch away from this. These sec- straightforward and wondered why the sphincter could be sacrificed (14). How- ondary tracts are generally known as procedure was performed in a theater ever, it is now generally accepted that known as “extensions” (Fig 3). Exten- with “all the pomp and circumstance of functional disability can follow even sions may be intersphincteric, ischio- a great operation” (15). Unfortunately, minimal division of the sphincter. Per- anal, or supralevator (pararectal), and he was mistaken; there are many traps haps most important, injudicious prob- their morphology may suggest tracts or waiting for the unwary. Inappropriate ing during EUA can create new second- abscesses. Exactly when a tract be- incision and injudicious exploration can ary tracts very easily. For example, comes an abscess is not precisely de- all too easily convert a simple fistula into forceful probing in the roof of the ischio- fined, but both terms describe regions a surgical nightmare, with disastrous anal fossa can rupture through the leva- of infection. The ischioanal fossa is the consequences for the patient. The pri- tor plate, causing a supralevator exten- commonest site for an extension (6), mary objectives are to eradicate the sion or even a rupture into the rectum, especially one that arises from the apex tract and drain all associated sites of which would cause an extrasphincteric of a transsphincteric fistula (A in Fig 3). infection while simultaneously preserv- fistula (6,12). Extensions also occur in the horizontal ing anal continence. Identification of extensions at EUA plane and are known as “horseshoes” if It is the right balance between eradi- is central to curing the fistula. It is well there is ramification each side of the cation of infection and preservation of recognized that missed extensions are internal opening (6) (Fig 3). The ana- function that is the art of fistula surgery. the commonest cause of recurrence, tomic description of the path taken by To achieve this, two surgical questions which has reached 25% in some series the primary fistula tract and the location need to be answered preoperatively: (16). Extensions require specific treat- of any associated extension forms its (a) What is the relationship between the ment and inevitably necessitate more “classification.” fistula and the anal sphincter (ie, can the extensive surgery. For example, supral- tract be safely laid open with only a low evator extensions are particularly diffi- risk of postoperative incontinence), and cult both to diagnose (because they are Assessment and Treatment (b) are there any extensions from the pri- high above the anal canal) and to treat Treatment is usually straightforward mary tract that need to be treated to pre- (because the levator plate forms a bar- and involves laying open the fistula by vent recurrence, and, if so, where are they? rier to drainage). Surgeons have traditionally relied The net result is that it can be very Figure 3 on examination of patients who have difficult at EUA to classify the primary been administered a general anesthetic, tract with confidence, and there is am- a procedure referred to as examination ple opportunity to make matters worse. under anesthesia (EUA), to answer Patients with recurrent disease are a these questions. At EUA, the surgeon particular case in point. Such patients attempts to classify the fistula by palpat- are more likely to harbor missed dis- ing it to determine its relationship to the ease but are also difficult to assess. Mul- sphincter. However, anesthesia and tiple failed operations are the rule consequent loss of tone impair precise rather than the exception, with the re- identification of underlying muscles. A sult that digital palpation frequently metal probe is inserted into the external cannot help distinguish between scar- opening and is directed toward the den- ring due to repeated surgery and indu- tate line to find the internal opening. ration due to an underlying extension. This is also not as straightforward as it Furthermore, patients in this group are sounds. For example, the internal open- also more likely to have extensions that ing is frequently not obvious, and the travel several centimeters away from Figure 3: Illustration in coronal plane shows surgeon may need to inject hydrogen the primary tract, frustratingly in al- fistula extensions (secondary tracts): A ϭ exten- peroxide into the external opening most any direction, which further ham- sion into roof of ischioanal fossa, arising from while inspecting the anal canal. pers detection of the extensions. The ϭ apex of a transsphincteric fistula; B supraleva- The height of the internal opening more chronic the fistula, the more com- tor pararectal extension arising from apex of a relative to the anal canal and sphincters plicated the associated extensions tend transsphincteric fistula; C ϭ supralevator exten- is also crucial; the higher the opening, to be. The inevitable result is that these sion originating from intersphincteric plane, D ϭ the more sphincter will be divided. Al- patients become progressively more dif- intersphincteric horseshoe. though Milligan and Morgan (14) real- ficult to treat, with both patient and sur-

Radiology: Volume 239: Number 1—April 2006 21 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

Figure 4 geon becoming ever more exasperated. man and co-workers (18) found fistulog- The key to breaking this loop is accurate raphy to be more useful in that it pro- preoperative assessment. vided helpful information in nearly half of the 27 subjects in their study. It has been suggested that the prime Imaging Fistula in Ano: Fistulography reason why fistulography is generally un- For many years radiologists have at- helpful is that radiologists are not familiar tempted to help answer the surgical with the concepts of fistula pathogenesis questions posed above, with varying de- and anatomy and the relevant surgical grees of success. Contrast material–en- questions (17). One of the most hazard- hanced fistulography was the first mo- ous misinterpretations is to diagnose a dality used. In fistulography, the exter- direct rectal opening merely because nal opening is catheterized with a fine there is contrast material in the rectal cannula, and a water-soluble contrast lumen; usually, the contrast agent has agent is injected gently to define the fis- merely refluxed up from the internal anal tula tract (Fig 4). opening. Such radiology reports only en- Unfortunately, fistulography has two courage the surgeon to look for nonexist- major drawbacks. First, extensions from ent openings and extensions, which can the primary tract may fail to fill with con- result in iatrogenic secondary tracts. trast material if they are plugged with de- Figure 4: Fistulography in a male patient. bris, are very remote, or there is exces- Coronal image shows that it is obvious that there Imaging Fistula in Ano: CT sive contrast material reflux from either are several high extensions (arrows) surrounding the internal or external opening. Second, Computed tomography (CT) may depict the anorectal junction; however, the exact ana- the sphincter muscles themselves are not fistula in ano, especially if rectal and intra- tomic location of these is unclear because the directly imaged, which means that the re- venous contrast material are used, and pelvic floor (ie, levator ani in this case) cannot be lationship between any tract and the initial reports were encouraging (19–21). directly visualized. Definition of extension location sphincter must be guessed. Furthermore, However, fistula depiction is not enough; (supra- or infralevator) is central to surgical an inability to visualize the levator plate fistulas must be classified correctly, and management. means that it can be difficult to decide more recent and better data suggest that whether an extension has a supra- or an CT cannot be used for this purpose with infralevator location. Similarly, the exact sufficient accuracy. This is because the dosonography, with CT probably limited level of the internal opening in the anal CT attenuation of the anal sphincter and to the diagnosis of fistula-associated pel- canal is often impossible to determine pelvic floor is similar to that of the fistula vic abscesses where other imaging is un- with sufficient accuracy to help the sur- itself, unless the latter contains air or con- available or cannot be tolerated. geon. The net result is that fistulographic trast material. This is compounded by the findings are both difficult to interpret and inability to image in the surgically relevant unreliable. coronal plane. Comparative studies are Imaging Fistula in Ano: Anal Very little has been written on fistu- sparse: A study of 25 patients with 17 Endosonography lography for fistula in ano, probably be- fistulas found that CT could be used to Anal endosonography, developed by Clive cause the modality is so fraught with correctly classify only four fistulas, in con- Bartram, FRCP, FRCS, FRCR, was the errors. Kuijpers and Schulpen (17) at- trast to the 14 correct classifications first technique to directly depict the anal tempted to determine its value by retro- achieved with endosonography (22). Po- sphincter complex in detail (23). Simple spectively reviewing fistulographic im- tentially, the disadvantages of CT might modification of a rotating rectal endo- ages in 25 patients. They found that the be overcome by using multi–detector row probe by covering it with a nondeform- internal opening and associated exten- CT fistulography, which offers the possi- able plastic cone allowed the transducer sions were demonstrated and correctly bility of isotropic voxels and multiplanar to be withdrawn through the anal canal interpreted in only four (16%) subjects. imaging, but results are awaited and the itself, situating it very close to the target Moreover, false-positive diagnoses of motivation to perform these studies is structures and thus providing images of rectal openings and supralevator exten- likely limited by the ready availability of high spatial resolution. The technique has sions were made in three (12%) pa- MR imaging. It should be borne in mind attracted considerable attention because tients, which would have resulted in se- that CT is frequently used to search for of its ability to demonstrate the presence rious surgical errors if acted on. The abscesses in the context of Crohn disease, and extent of anal sphincter disruption, authors concluded that fistulography and fistula in ano may be encountered at notably after vaginal delivery (24). Anal was “inaccurate and unreliable,” al- the time of examination. At present, how- endosonography has also been exten- though they admitted prior bias against ever, accurate classification of these fistu- sively used for the preoperative classifica- the technique (17). In contrast, Weis- las is best left to MR imaging or anal en- tion of fistula in ano.

22 Radiology: Volume 239: Number 1—April 2006 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

Figure 5 The examination is simple, rapid, and that cross the external sphincter to reach well tolerated by patients. The patient lies the ischioanal fossa (Fig 6). As would be in the left lateral position or in the prone expected, extensions are revealed as hy- position if female (25). The probe is poechoic fluid collections (Fig 7). gently inserted into the distal rectum and It was initially hoped that anal en- then withdrawn through the anal canal. dosonography would revolutionize preop- The internal sphincter is visualized as a erative fistula classification, a view sup- hypoechoic ring encircling the anal canal, ported by the results of early studies (27). whereas the external sphincter is of However, subsequent work has been in- mixed echogenicity (Fig 5). The inter- conclusive. For example, some investiga- sphincteric space and longitudinal muscle tors (28,29) have found the technique to lie between these and are of mixed echo- be useful, while others (30) have found it genicity and are easily identified by using to be no better than digital rectal exami- modern 10-MHz transducers (26). nation. Furthermore, in direct compari- Endosonography is particularly well sons with MR imaging, anal endosonogra- Figure 5: Intersphincteric fistula. Anal en- suited to identification of the internal phy has been variously found to be supe- dosonogram in transverse plane at mid–anal canal opening, because this opening is usually rior (31), equivalent (32,33), or inferior level in a male patient shows fistula with hypo- positioned right at the probe surface. It is (34,35). Much of this discrepancy is prob- echoic tract located in intersphincteric plane be- important to realize, however, that a tract ably related to operator expertise, since tween external (EAS) and internal (IAS) anal extending up to the anal mucosal surface anal endosonography, similar to fistulog- sphincters. Internal sphincter is markedly hypo- is rarely seen. Although a breach in the raphy, is highly operator dependent. echoic. At surgery, the internal opening was lo- subepithelial layer of the anal canal is oc- However, there are undoubtedly sev- cated at 6 o’clock posteriorly and was correctly predicted from anal endosonographic visualiza- casionally present, it is more common for eral areas where anal endosonography tion because of the radial position of the fistula the position of the internal opening to be has specific disadvantages. For example, within the intersphincteric plane. revealed as a hypoechoic focus in the in- insufficient penetration of the ultrasound tersphincteric space that abuts the inter- beam beyond the external sphincter, es- nal sphincter, often with a small corre- pecially with high-frequency transducers, sponding defect in the internal sphincter limits the ability to resolve ischioanal and (Fig 6). Because intersphincteric fistulas supralevator infections, with the result do not stray beyond the intersphincteric that extensions from the primary tract Figure 6 space, they are usually very well visual- may be missed at endosonography. Also, ized at anal endosonography. Trans- anal endosonography cannot be used to sphincteric fistulas are revealed by tracts reliably distinguish infection from fibro- sis, because both have a hypoechoic ap- pearance (30). This causes particular dif- Figure 7 ficulties in patients with recurrent disease, since infected tracts and fibrotic scars are frequently combined. Attempts have been made to clarify the course of patent tracts by injecting hydrogen peroxide or sonography contrast agents into the external opening during examination (34,35). However, gas formed within the tract as a result may cause acoustic shadowing that mimics an extension. Indeed, this phenomenon can occur with any tract Figure 6: Transsphincteric fistula shown on that contains air, leading, for example, to anal endosonogram in the transverse plane at the intersphincteric fistulas being inadvertently mid–anal canal level in a female patient. In con- classified as transsphincteric (Fig 8). -has penetrated the Anal endosonography is also disad (ء) trast to Figure 5, the fistula external anal sphincter (EAS). The internal open- Figure 7: Anal endosonogram at upper anal vantaged by the inability to image in the ing was correctly predicted at 7-o’clock position. canal level in a male patient shows extensive hy- surgically important coronal plane, so ء Note that the internal sphincter is relatively thinned poechoic horseshoe extension ( ). Because en- that it may be very difficult to distin- here, which is a clue to the site of the internal dosonography is limited to the transverse plane, it guish supra- from infralevator exten- opening, but there is no tract extending to the anal is difficult to determine whether this extension is sions (Fig 7). Some investigators (36) mucosa. infra- or supralevator. have attempted to overcome this disad-

Radiology: Volume 239: Number 1—April 2006 23 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

Figure 8 Figure 9 vantage by employing a three-dimensional acquisition (Fig 9), but this technique remains relatively experimental. There is no doubt that anal endosonography is a valuable technique in the right hands. A recent study (37) in which anal endosonography was compared with digital rectal evaluation and MR imaging in 108 primary tracts found that digital evaluation resulted in correct classification of 61% of fistulas; anal endosonography, 81%; and MR imaging, 90%. Anal endosonography was particularly adept with regard to the site of the internal opening, with correct prediction in 91% of cases compared Figure 8: Transverse anal endosonogram at with 97% for MR imaging (37). How- upper anal canal level in a female patient shows intersphincteric horseshoe extenstion (arrows). ever, there is little doubt that MR imag- Figure 9: Three-dimensional anal endosono- ,(ء) Gas in the fistula causes acoustic shadowing ing is a superior technique overall and is gram (coronal view) in a female patient shows which could be mistaken for transsphincteric transsphincteric tract (arrows) that has been out- now also generally available. Given this, tracts. lined by injecting hydrogen peroxide into the ex- the major role of anal endosonography ternal opening. in fistula disease is probably in the assessment of the degree of sphincter dis- body coil was used; this was thought to ruption in patients who become anally be possibly due to lower spatial resolu- incontinent after surgery for a fistula. tion. The introduction of external Endosonography has high spatial reso- phased-array surface coils increased the Figure 10 lution, so it also has a particular role in signal-to-noise ratio and spatial resolu- patients who potentially have a small tion, to good effect (46,47), and these intersphincteric abscess that might be coils quickly became generally available. difficult to resolve by using standard The best spatial resolution is achieved body- or surface-coil MR imaging. by using a dedicated endoluminal anal coil (48), which may be combined with a surface coil to increase the field of view (49). Imaging Fistula in Ano: MR Imaging It should be noted that these endoluminal In recent years, MR imaging has emerged coils are not the same as a rectal coil but as the leading contender for preoperative are of smaller diameter and are designed classification of fistula in ano. The ability to be placed in the anus. The receiver coil of MR imaging to help not only accurately is generally housed in a plastic cover that Figure 10: Endoanal receiver coil for MR classify tracts but also identify disease allows placement across the anal canal imaging. that otherwise would have been missed (Fig 10). The external diameter generally has had a palpable effect on surgical treat- ranges between 12 and 19 mm, although ment and, ultimately, patient outcome. smaller coils have been used for pediatric scine butylbromide is not licensed for use examinations (50,51). Endoluminal coils in the United States.) Technique are susceptible to motion artifact, but this The choice of coil depends on per- Magnetic field strength does not appear can be reduced by means of careful pa- sonal preference, availability, the pa- to be a critical factor for good results tient preparation. For example, patients tient group studied, and the clinical (38). Although higher field strength should be asked to try and relax the question in each patient. In a study of 10 might be relevant for more subtle differ- sphincter and pelvic floor as much as pos- patients with cryptoglandular fistula entiation of sphincter anatomy and sible, and due attention should be paid to (45), an endoluminal coil was found to tracts, no definite diagnostic benefit has comfort, including support for the coil be superior to a surface coil; in a subse- been demonstrated, to our knowledge. and patient with pads (50). Spasmolytics quent study of 30 patients (52), how- Authors of initial reports (39–43) on such as 20 mg of hyoscine butylbromide ever, a body coil was found to be supe- MR imaging necessarily used the body (Buscopan; Boehringer Ingelheim, In- rior overall because the limited field of coil, with good results. However, au- gelheim, Germany) or 1 mg of glucagon view inevitable with endoluminal imag- thors of some subsequent studies (44– administered intramuscularly may help to ing meant that distant extensions were 45) reported lower accuracy when a reduce motion-induced artifacts. (Hyo- missed, a phenomenon that is especially

24 Radiology: Volume 239: Number 1—April 2006 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

frequent in patients with Crohn disease which may reflect differences in patient provides good contrast between hy- (53). In a third study (49), endoluminal population since the coil used in both perintense fluid in the tract and the and phased-array coils were compared studies was identical. hypointense fibrous wall of the fistula in 20 patients, and the investigators When circumstances allow, it is while simultaneously enabling good found that while the endoluminal coil likely that an optimal examination will discrimination between the several was superior for classification of the pri- be achieved by using a combination of layers of the anal sphincter (50,55). mary tract, extensions were better im- both external and endoluminal coils. Others have used T1-weighted se- aged by using the superior field of view However, it should be borne in mind quences, which must be combined of the external coil. that accuracy with a body or external with intravenous contrast material for These results clearly suggest that a coil alone remains high (41–43, the fistula to be highlighted (43). Fat- large field of view is necessary when- 47,49,52), and lack of an endoluminal suppression techniques are widely ever extensions are suspected—for ex- coil alone is insufficient reason to avoid used with both T2-weighted (56) and ample, in patients with recurrent fistula preoperative MR imaging of fistula in gadolinium-enhanced T1-weighted se- or Crohn disease. The high spatial reso- ano. Indeed, examination with a body quences (the latter may be especially lution of endoluminal coils makes them or phased-array coil has become stan- valuable in patients with Crohn dis- ideal for precise demonstration of the dard practice, not least because endolu- ease, to differentiate between fluid location and height of the internal open- minal coils specifically designed for anal im- and inflammatory tissue that are both ing, and they may have a special role for aging remain relatively unavailable. hyperintense on fat-suppressed T2- demonstrating ano- or rectovaginal fis- weighted and/or short tau inversion- tulas, which are notoriously difficult to MR Sequences recovery [STIR] images). Normal ano- image (54). Endoluminal coils are also Various investigators have adopted rectal structures do not enhance sub- valuable when simultaneous informa- different strategies with respect to the stantially, except for the internal anal tion on the degree of sphincter disrup- MR sequences used to image fistula in sphincter and blood vessels, including tion is needed, which may be the case in ano. All agree that anatomic precision hemorrhoids. Investigators have also patients who have undergone previous is needed so that the course of the successfully employed STIR, a se- surgery. Endoluminal coils are some- fistula with respect to adjacent struc- quence that combines fat suppression times difficult to place owing to anal ste- tures can be judged accurately, and all with high conspicuity of active tracts nosis or local pain as a result of exten- use some method with which infection (41,47,52). Other approaches have in- sive infection. Halligan and Bartram (usually pus) can be highlighted. cluded saline instillation into the ex- (52) found that an endoluminal coil These aims can be achieved by a vari- ternal opening in an attempt to in- could not be placed in 17% of their pa- ety of means. Many investigators use crease tract conspicuity (57) or rectal tients, whereas Stoker and colleagues the rapid and convenient fast spin- contrast medium (58,59). However, (50) failed to place the coil in only 3%, echo T2-weighted sequence, which such measures increase the complex-

Figure 11 Figure 12

Figure 12: Coronal (a) T2-weighted fast spin-echo (2500/70; echo train length, 16; ﬁeld of view, 300 mm; Figure 11: Correct orientation for MR imaging matrix, 256 ϫ 512; section thickness, 4 mm; gap, 0.4 mm) and (b) coronal STIR (4000/42, inversion time of of anal canal. Sagittal T2-weighted scout image 150 msec; echo train length, 16; matrix, 224 ϫ 256; section thickness, 4 mm; gap 0.4 mm; two signals ac- through patient’s midline is used to plan images quired) MR images acquired with external phased-array coil show complex transsphincteric ﬁstula with tract that are truly transverse with respect to anal canal, (short straight arrows) in left ischioanal fossa that extends below ischial bone (I) toward the upper leg (not as shown by white lines. Coronal imaging is then shown). At the ischial tuberosity, bone marrow edema (long straight arrow) is visible on b. Arrowhead ϭ ex- performed at 90° to the transverse plane. ternal opening, curved arrow ϭ small abscess, AS ϭ anal sphincter.

Radiology: Volume 239: Number 1—April 2006 25 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

External Phased-Array 1.5-T MR Protocols for Imaging Perianal Fistula Section Thickness/ Echo Train Field of Intersection Gap Sequence and Plane* TR/TE† Length View (mm) Matrix (mm)‡

T2-weighted fast spin echo without fat suppression Sagittal, coronal, transverse 2500/70 10 300§ 256 ϫ 512 3/0.3 T2-weighted fast spin echo with fat suppression Transverse 2500/70 10 300§ 256 ϫ 512 3/0.3 T1-weighted fast spin echo with fat suppression Transverse# 600/minimal 3 450 256 ϫ 256 4/0.4

Note.—Parameters are typical and were established with EchoSpeed imager (GE Healthcare, Milwaukee, Wis). Field of view should include rectum and perineum; at endoanal MR, volume should encompass entire sensitive region of coil. For all sequences, two signals are acquired and bandwidth is 20.83 kHz. * Transverse and coronal planes are off axis, orthogonal and parallel, respectively, to anal canal. † TR/TE ϭ repetition time (msec)/echo time (msec). ‡ For endoanal MR, transverse section thickness and gap are 2–3 mm and 0.2–0.3 mm, respectively. § For endoanal MR, ﬁeld of view is 160 mm for coronal and sagittal planes and 100 mm for transverse plane. # Additional value of routine use of this contrast-enhanced sequence is not yet determined, but it may help differentiate inﬂammation from abscess.

Figure 13

Figure 13: Transsphincteric fistula in a man with Crohn disease. (a, b) Transverse T2-weighted fast spin-echo MR images (echo train length, 16; field of view, 300 mm; matrix 256 ϫ 512; section thickness, 4 mm; gap, 0.4 mm; two signals acquired) obtained (a) without (2500/70) and (b) with (4000/85) fat saturation and (c) transverse fat-saturated contrast-enhanced T1-weighted fast spin-echo image (see Table for parameters) show two separate fistula tracts (straight and curved arrows) in left posterior ischioanal space, close to the anal sphincter (A). Both tracts show confluent high signal intensity centrally, which represents pus in the tract lumen. On a and b, the surrounding inflammatory tissue (arrowheads) is of low signal intensity (a), which increases with fat-saturation (b) and especially with contrast enhancement (c). Anterior tract (curved arrow) demonstrates more adjacent inflammation (arrowheads) than does posterior tract (straight arrow). ity of the examination in the face of the proximately 45°, straight transverse cases of an internal opening high in the already excellent results achieved with and coronal images will fail to achieve rectum, but this is seldom needed. more standard procedures. Imaging this alignment because of marked par- It is important that the imaged volume protocols are detailed in the Table. tial volume effect. Oblique transverse extend several centimeters above the leva- and coronal planes oriented orthogonal tors and include the whole presacral space, Imaging Planes and parallel, respectively, to the anal both of which are common sites for exten- It is central to success that imaging sphincter are therefore necessary and sions. The entire perineum should also be planes are correctly aligned with re- are most easily planned by using a mid- included. On occasion, tracts may extend spect to the organ of interest, namely line sagittal image (Fig 11). It may be for several centimeters, even leaving the the anal canal. Because the anal canal is necessary to align supplementary exam- pelvis or reaching the legs, and any tract tilted forward from the vertical by ap- inations with the rectal axis in complex visible must be followed to its termination if

26 Radiology: Volume 239: Number 1—April 2006 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

this has not been included on the standard when the fistula is at the 3- or 9-o’clock hyperintense longitudinal structures on image volume (Fig 12). The imaged volume position. Instead of being obtained parallel T2-weighted or STIR images (Fig 12). should encompass the whole sensitive re- to the anal canal, the image planes radiate On contrast-enhanced T1-weighted im- gion of the coil when an endoanal receiver from it like the spokes of a wheel (60). Little ages, active granulation tissue will en- is used. The precise location of the primary research has specifically addressed the ben- hance while fluid in the tract itself re- tract (eg, ischioanal or intersphincteric) is efits of various imaging planes, but in a mains hypointense (Fig 13). Active usually most easily appreciated by using study of 20 patients (60) investigators found tracts are often surrounded by hypoin- transverse images; the radial site of the in- that the combination of a transverse series tense fibrous walls (Fig 13), which can ternal opening is also well seen on images in and a longitudinal series (coronal, sagittal, be relatively thick, especially in patients this plane. Coronal images best depict the radial, or a combination) provided all nec- with recurrent disease and previous levator plate, which helps distinguish supra- essary information for successful interpre- surgery. Occasionally, some hyperin- from infralevator infection. The height of tation. tensity in this fibrous area may be seen, the internal opening may also be best ap- probably reflecting edema. Hyperinten- preciated on coronal images, with the ca- Interpretation sity may also extend beyond the tract veat that the anal canal must be imaged The success of MR imaging for preoper- and its fibrous sleeve, where it repre- along its entire craniocaudal extent. ative classification of fistula in ano is a sents adjacent inflammation (Fig 13). The radial plane (as used for imaging direct result of the sensitivity of MR for The external anal sphincter is clearly menisci of the knee) is seldom used but tracts and abscesses combined with visualized by using MR imaging. It is rela- seems attractive because it has the poten- high anatomic precision and the ability tively hypointense, and its lateral border tial to depict fistulas along their full cranio- to image in surgically relevant planes. contrasts against the fat in the ischioanal caudal extent, something that is only Accurate preoperative classification is fossa, both on STIR (Fig 14) and especially achieved with standard coronal imaging achieved by correctly relating the im- on fast T2-weighted MR studies (Fig 13a). aged fistula to the anal sphincter. Thus, it is relatively easy to determine whether a fistula is contained by the exter- Figure 14 Primary Tract nal sphincter or has extended beyond it. If a Active tracts are filled with pus and fistula remains contained by the external granulation tissue and, thus, appear as sphincter throughout its course, then it is

Figure 15

Figure 14: Intersphincteric fistula in a male patient. Transverse STIR MR image (1500/15; field of view, 375 mm; matrix, 256 ϫ 256; section thickness, 4 mm; gap, 1 mm; four signals acquired) shows that lateral margin of external sphincter (long arrow) contrasts against fat in the Fistula (short arrow) is in the Figure 15: Fistula classified as suprasphincteric on coronal STIR MR images (same parameters as for Fig .(ء) ischioanal fossa intersphincteric space posteriorly at 6 o’clock and 14) in a female patient. (a) Primary tracts in right (long arrow) and left (short arrow) ischioanal fossae are is contained by the external sphincter. There is no shown. (b) Image obtained just posterior to a shows that right-sided primary tract (white arrows) arches over .(to reach a lower internal opening at the dentate line level (black arrow (ء) tract in the ischioanal fossa. puborectalis muscle

Radiology: Volume 239: Number 1—April 2006 27 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

Figure 16 Figure 17 highly likely to be intersphincteric (Fig 14). In contrast, any evidence of a tract in the ischioanal fossa effectively excludes an intersphincteric fistula. However, transsphincteric, suprasphincteric, and extrasphincteric fistulas all share the common feature of a tract that lies beyond the con- fines of the external sphincter. While a transsphincteric fistula will be the commonest cause of a tract in the ischioanal fossa (Fig 13), differentiation between these three fistulas is only possible by locating the internal opening and determining the course between this and the primary tract.

Internal Opening Figure 17: Transsphincteric fistula in a male patient. Transverse STIR MR image (same param- The exact location of the internal opening eters as for Fig 14) shows primary tract (vertical can be difficult to define, whatever the arrow) in right ischioanal fossa, where it can be to(ء) imaging modality used. Two questions clearly seen to penetrate external sphincter need to be answered. What is the radial reach the intersphincteric space. Internal opening site of the internal opening, and what is Figure 16: Fistula classified as extrasphinc- is posterior at 6 o’clock (horizontal arrow), at den- its level? The vast majority of anal fistulas teric in a female patient on coronal T2-weighted tate line level. open into the anal canal at the level of the MR image (4563/150; field of view, 350 mm; ma- dentate line, commensurate with the trix, 256 ϫ 256; section thickness, 6 mm; gap, 0.6 cryptoglandular hypothesis of fistula mm; four signals acquired). Fistula tract (horizon- pathogenesis. Furthermore, most fistulas tal white arrows) is seen in left ischioanal fossa. Figure 18 also enter posteriorly, at the 6-o’clock po- Levator plates (vertical white arrows) are well de- sition. Unfortunately, the dentate line picted bilaterally. Tract penetrates the left levator cannot be identified as a discrete ana- plate, and the internal opening (top horizontal tomic entity, even when endoanal re- white arrow) is into the rectum, above the level of and well above the (ء) ceiver coils are used, but its general posi- the puborectalis muscle tion can be estimated with sufficient pre- dentate line (black arrow). cision for the imaging assessment to be worthwhile. The dentate line lies at approximately the mid–anal canal level. hampers precise identification of the This is generally midway between the su- mid–anal canal level. Nevertheless, with perior border of the puborectalis muscle experience it is possible to estimate the and the most caudal extent of the subcu- exact height of the internal opening with taneous external sphincter. These land- reasonable precision (47). marks define the “surgical” anal canal (as Any tract that penetrates the pelvic distinct from the “anatomic” anal canal, floor above the level of the puborectalis which is shorter and is defined as the muscle is potentially a suprasphincteric canal caudal to the anal valves). The den- or extrasphincteric fistula. The level of Figure 18: Transsphincteric fistula in a male tate level is probably best appreciated on the internal opening distinguishes be- patient. Transverse STIR MR image (same param- coronal views, which allow the craniocau- tween these types of fistula; specifically, eters as for Fig 14) at level of the internal opening dal extent of the puborectalis muscle and the internal opening is anal in supra- shows primary tract (vertical arrow) at 4–5 external sphincter to be appreciated; with sphincteric fistulas (Fig 15) and rectal in o’clock. Unlike Figure 17, the tract cannot be experience, however, its location can be extrasphincteric fistulas (Fig 16). traced right to the anal mucosa, and the adjacent estimated with reasonable precision by Transsphincteric fistulas penetrate internal sphincter (horizontal arrow) appears in- using transverse views. the external sphincter, a feature that can tact. However, an internal opening at 4–5 o’clock It should be noted that in some pa- be easily appreciated on transverse (Fig was reported because this position indicated site tients the puborectalis muscle is rather 17) or coronal views. However, recent of maximal infection in the intersphincteric plane. gracile, unlike the bulky muscle sug- studies (61) in which MR imaging was The internal opening was confirmed at this site during subsequent EUA. Intersphincteric plane is gested in many anatomy texts, and it used have revealed that a transsphinc- well seen in this patient between hyperintense frequently and imperceptibly segues teric tract may cross the sphincter at a internal sphincter and the external sphincter. into the external sphincter, all of which variety of angles. For example, it may

28 Radiology: Volume 239: Number 1—April 2006 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

arch upward as it passes through the ex- traced right to the anal mucosa (Fig 17). one that arises from the apex of a trans- ternal sphincter and thus cross the mus- The radial site is reported with respect to sphincteric tract and extends into the cle at a higher level than would be de- a clock face, with 12 o’clock being directly roof of the ischioanal fossa (Figs 3, 19). duced merely by inspecting the level of anterior. However, like endosonography, The major benefit of MR imaging find- the internal opening. This is important, it is frequently impossible to trace a tract ings is that they can alert the surgeon to because such tracts will require a greater right up to the anal mucosa, especially if extensions that would otherwise be degree of sphincter incision during fistu- an endoanal coil has not been used. In missed. For example, extensions may lotomy, with a corresponding increase in such cases, an intelligent deduction must be several centimeters from the pri- the risk of postoperative incontinence. be made as to where the internal opening mary tract (Fig 20), which makes them MR imaging in the coronal plane is best is likely to be. This is best accomplished difficult to detect during clinical exami- for estimates of the precise angulation of by looking for the area of maximal inter- nation or EUA. This is especially the the tract with respect to the surrounding sphincteric sepsis, since the internal case when extensions are contralateral musculature (61). opening is likely to lie very close to this to the primary tract (Fig 21). It is also The radial site of the internal opening (Fig 18). The intersphincteric space and important to search for supralevator ex- is easy to identify if the fistula tract can be longitudinal layer are often seen as a hy- tensions (Fig 22), since these are not pointense ring between the internal and only difficult to detect but pose specific Figure 19 external sphincters (Fig 18). The internal problems with regard to treatment. sphincter is hyperintense on both T2- Horseshoe extensions spread across weighted fast spin-echo and STIR images, both sides of the internal opening and especially if contrast material has been are recognized on MR images by their used (62). unique configuration (Fig 23); horseshoe extensions may be intersphinc- Extensions teric, ischioanal, or supralevator. Com- The major advantage of MR imaging is plex extensions are especially common the facility with which it can demonstrate in patients with recurrent fistula in ano extensions associated with a primary (Fig 21) or in those who have Crohn tract. Morphologically, extensions fre- disease (Fig 13). quently take the form of complex tract systems, regions of which have often be- Effect of Preoperative MR on Surgery come dilated to create an abscess (al- and Outcome though a precise radiologic distinction be- Over the past few years, imaging, nota- tween abscess and a large tract remains bly MR, has revolutionized the treat- elusive). Extensions appear as hyperin- ment of patients with fistula in ano. This Figure 19: Left-sided transsphincteric tract tense regions on T2-weighted and STIR is because MR can be used to classify (short arrow) in a female patient. Coronal STIR MR images and enhance if intravenous con- fistulas preoperatively with high accu- image (same parameters as for Fig 14) shows trast material is used. Again, collateral racy while also alerting the surgeon to large extension (long arrow) from apex of tract into inflammation can be present to a variable disease that would otherwise have been roof of ipsilateral ischioanal fossa. extent. missed. While there are reports of the The commonest type of extension is technique dating from 1989 (40), it was not until the description by Lunniss and Figure 20 co-workers (41) that the true potential of MR imaging was fully appreciated. Figure 20: Left-sided trans- Lunniss et al imaged 16 patients with sphincteric fistula (short straight cryptoglandular fistula in ano and com- arrows) with internal opening at 6 pared the MR classifications they ob- o’clock (long straight arrow) in a tained with those from subsequent female patient. Transverse STIR EUA. MR imaging proved correct in 14 MR image (same parameters as for (88%) cases, which immediately sug- Fig 14) shows remote extension gested that it was the most accurate (curved arrow) into ipsilateral preoperative assessment yet available. buttock that was unsuspected at However, the remaining two patients, clinical examination but is well in whom MR suggested disease but EUA demonstrated at MR imaging. The yielded normal findings, re-presented extension was found at surgery guided by MR findings. some months later with disease at the site initially indicated on MR images. This led the authors to conclude that

Radiology: Volume 239: Number 1—April 2006 29 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

Figure 21 Figure 22 MR imaging “is the most accurate method for determining the presence and course of anal fistulae” (41). This work was rapidly confirmed by others working in the field and was subsequently elaborated on. Spencer and colleagues (63) inde- pendently classified 37 patients into those with simple or those with complex fistulas on the basis of MR imaging and EUA and found that MR results were the better predictor of outcome, with positive and negative predictive values, respectively, of 73% and 87% for MR and 57% and 64% for EUA. These re- Figure 21: Transsphincteric primary tract sults clearly implied that MR imaging (short arrow) in the right posterior quadrant of a and outcome were closely related and female patient. Transverse STIR MR image (same again raised the possibility that preop- parameters as for Fig 14) shows two left-sided Figure 22: Bilateral supralevator extensions erative MR could help identify features contralateral extensions (long arrows) that were (long arrows) in a female patient. Coronal STIR that cause postoperative recurrence. undetected at EUA until results of patient’s MR MR image (same parameters as for Fig 14) clearly Beets-Tan and colleagues (46) ex- examination were revealed to the surgeon in the show levator plates (short arrows) bilaterally, so tended this hypothesis by investigating operating theater. the therapeutic effect of preoperative MR that it is easy for the radiologist to be confident that imaging; the MR imaging findings in 56 infection extends above them. patients were revealed to the surgeon af- imaging on clinical outcome in patients with ter he or she had completed an initial fistula in ano at initial presentation and EUA. MR imaging provided important ad- found that the scheduled surgical approach ditional information that precipitated fur- changed in 10% of this group. Figure 23 ther surgery in 12 (21%) of 56 patients, Ever since the results of Lunniss et al predominantly in those with recurrent fis- (41) suggested that EUA might be an im- tula or Crohn disease (46). perfect reference standard with which to Buchanan and co-workers (47) hypoth- judge MR imaging, comparative studies esized that the therapeutic influence and, have been plagued by the lack of a genu- thus, beneficial effect of preoperative MR ine reference standard. It is now well rec- imaging would be greatest in patients with ognized that surgical findings at EUA are recurrent fistula, since these patients had often incorrect. In particular, there are the greatest chance of harboring occult in- frequent false-negatives. In a recent com- fection, while such fistulas were also the parative study of endosonography, MR most difficult to evaluate clinically. After an imaging, and EUA in 34 patients with fis- initial EUA, Buchanan et al revealed the tula due to Crohn disease, Schwartz and findings of preoperative MR imaging to the co-workers (32) found that a combination surgeons for 71 patients with recurrent fis- of the results of at least two modalities tulas and left any further surgery to the was necessary to arrive at a correct clas- discretion of the operating surgeon. They sification. Indeed, it is well established found that postoperative recurrence was that many false-negative surgical results only 16% for surgeons who always acted if will only reveal themselves during long- Figure 23: Horseshoe extension (arrows) aris- MR findings suggested that areas of infec- term clinical follow-up, and, at this point ing from intersphincteric fistula in a male patient. tion had been missed, whereas recurrence in time, comparative studies that ignore Transverse STIR MR image (same parameters as was 57% for those surgeons who instead clinical outcome are likely to be seriously for Fig 14) shows that, in this case, the horseshoe always chose to ignore imaging results (47). flawed. practically encircles the anal canal. Furthermore, in the 16 patients who needed further unplanned surgery, MR initially correctly predicted the site of this dis- Imaging for Differential Diagnosis sinus, actinomycosis, tuberculosis, proc- ease in all cases (47). Using a similar ap- Not all cases of perianal sepsis are due to titis, human immunodeficiency virus, lym- proach, Buchanan and colleagues (64) also fistula in ano. For example, acne conglo- phoma, and anal and rectal carcinoma investigated the effect of preoperative MR bata, hidradenitis suppurativa, pilonidal may all cause perianal infection. While

30 Radiology: Volume 239: Number 1—April 2006 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

clinical examination results are often con- (70). Small-bowel imaging may be used seton in place for a few months, after clusive, this is not always the case, and to search for Crohn disease when it is which time the internal opening can be imaging may help with the differential di- suspected, and the possibility of under- closed with a rectal mucosal advance- agnosis. The cardinal feature of fistula in lying pelvic disease should be consid- ment flap. A seton can also be placed at ano is intersphincteric infection, which is ered in any patient with an extrasphinc- EUA when the surgeon is uncertain not generally found in other conditions. teric fistula, whether thought due to about the relationship between the tract Whenever imaging suggests that infection Crohn disease or otherwise. and the sphincter; postoperative imag- is superficial rather than deep seated and ing can then be used to help answer this that there is no sphincteric involvement, question. In patients who undergo novel Which Patients Should Be Imaged? other conditions such as hidradenitis sup- therapies, such as use of fibrin glue, im- purativa should be considered (Fig 24). While fistula in ano is simple to diag- aging may be necessary during instilla- For example, authors of a recent study nose and simple to treat in most pa- tion to be confident that the whole tract (65) of patients with pilonidal sinus and tients, many other patients will benefit has been filled (71). fistula in ano found that MR imaging could from detailed and accurate preopera- MR imaging is not restricted to sur- be used to reliably distinguish between tive investigation. Where there is easy gical assessment. Infliximab, a chimeric the two on the basis of intersphincteric access to MR imaging, it could be ar- monoclonal antibody to human tumor infection and an enteric opening. gued that all patients should undergo necrosis factor–␣, currently has a The possibility of underlying Crohn preoperative imaging. For example, it prominent role for the medical treat- disease should always be considered in has been estimated that the therapeutic ment of Crohn disease, especially in pa- patients who have a particularly com- effect of MR imaging is 10% in patients tients with a chronic fistula (72). How- plex fistula, especially if the history is presenting for the first time with a ever, infliximab therapy is contraindi- relatively short. Indeed, a perianal fis- seemingly simple fistula (64). Where ac- cated if an abscess is present, and MR tula is the presenting condition in 5% of cess to imaging is more restricted, how- imaging may be used to search for this patients (34), and 30%–40% of patients ever, the clinician and radiologist will (70,73). Indeed, MR imaging may be with Crohn disease will experience anal need to select those patients who are used to monitor infliximab therapy, disease at some time (66–68). Specific most likely to benefit. Since there is now since it seems that fistulas may persist scoring systems for perianal Crohn dis- overwhelming evidence that MR imag- in the face of clinical findings that sug- ease have been proposed, such as the ing alters surgical therapy and improves gest remission (74). For example, inves- Perianal Crohn’s Disease Activity Index clinical outcome in patients with recur- tigations (70,73) of MR in patients (69) and a recently described system rent disease, MR should be routine in whose external opening has closed have based solely on MR imaging findings such cases. Patients presenting for the revealed that underlying infection can first time with a fistula that appears still be present, indicating a need for Figure 24 complex at clinical examination should continued therapy. Further studies are also be referred, as should patients with needed to determine whether MR mon- known Crohn disease, since the prepon- itoring improves outcome. derance of complex fistulas is increased. Where MR imaging is unavailable or There are also surgical situations where competent interpretation of the where imaging is likely to be particularly images is not possible, then anal en- beneficial, even when the fistula itself is dosonography is a viable and useful al- simple. For example, the anterior exter- ternative. While recent comparisons nal sphincter is very short in women, have shown that MR imaging outper- and division of this sphincter during fis- forms anal endosonography in all re- tula incision is particularly associated spects, the latter is far superior to sim- with postoperative incontinence, even ple clinical examination, and its perfor- when the fistula itself is simple and has mance in some areas is very close to no extensions. Faced with such a di- that of MR (37). Notably, anal en- lemma, the surgeon may choose to pass dosonography is very adept at depicting a thread (seton) through the tract the internal opening. rather than incise the fistula to provide Figure 24: Hidradenitis suppurativa in a male drainage. The patient can undergo post- patient. Transverse STIR MR image (same param- operative imaging so that the potential Conclusion eters as for Fig 14) shows extensive superficial extent of sphincter division can be as- In those patients with fistula in ano who infection (arrows). Absence of any infection related sessed by visualizing the relationship of have a high likelihood of complex dis- to anal canal and intersphincteric space meant that the seton to the external sphincter. A ease, the evidence that preoperative diagnosis could be confidently made preopera- decision can then be made whether to MR imaging influences the surgical ap- tively by using imaging. progress with fistulotomy or to keep the proach and the extent of exploration

Radiology: Volume 239: Number 1—April 2006 31 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

and improves the ultimate outcome is anorectal fistulae. Lancet 1934;2:1213– Mortensen NM. The role of anal ultrasound now overwhelming. We hope that this 1217. in the management of anal fistulas. Colorec- tal Dis 2002;4:118–122. article will stimulate radiologists to pro- 15. Syme J. On diseases of the rectum. Edin- vide this service to their surgeons in the burgh, Scotland: Adam and Charles Black, 30. Choen S, Burnett S, Bartram CI, Nicholls expectation that this will reduce the in- 1838; 1–2. RJ. Comparison between anal endosonography and digital examination in the evaluation cidence of recurrent fistula in ano and 16. Lilius HG. Fistula-in-ano, an investigation of of anal fistulae. Br J Surg 1991;78:445–447. the misery that this causes. human foetal anal ducts and intramuscular glands and a clinical study of 150 patients. 31. Orsoni P, Barthet M, Portier F, Panuel M, Acta Chir Scand Suppl 1968;383:7–88. Desjeux A, Grimaud JC. Prospective com- References parison of endosonography, magnetic reso- 17. Kuijpers HC, Schulpen T. Fistulography for 1. Lunniss PJ, Phillips RKS. Anatomy and func- nance imaging and surgical findings in ano- fistula-in-ano: is it useful? Dis Colon Rectum tion of the anal longitudinal muscle. Br J Surg rectal fistula and abscess complicating 1985;28:103–104. 1992;79:882–884. Crohn’s disease. Br J Surg 1999;86:360– 18. Weisman RI, Orsay CP, Pearl RK, et al. The 364. 2. Woodburne RT. Essentials of human anat- role of fistulography in fistula-in-ano: report omy. 7th ed. Oxford, England: Oxford Uni- 32. Schwartz DA, Wiersema MJ, Dudiak KM, et of five cases. Dis Colon Rectum 1991;34: versity Press, 1983. al. A comparison of endoscopic ultrasound, 181–184. magnetic resonance imaging, and exam un- 3. Chiari H. Uber die analen divertikel der rec- 19. Guillaumin E, Jeffrey RB, Shea WJ, et al. der anesthesia for evaluation of Crohn’s peri- tumschleimhaut und ihre beziehung zu den Perirectal inflammatory disease: CT find- anal fistulas. Gastroenterology 2001;121: anal fisteln. Wien Med Press 1878;19:1482– ings. Radiology 1986;161:153–157. 1064–1072. 1483. 20. Fishman EK, Wolf EJ, Jones B, Bayless TM, 33. Gustafsson UM, Kahvecioglu B, Astrom G, 4. McColl I. The comparative anatomy and pa- Siegelman SS. CT evaluation of Crohn’s Graf W. Endoanal ultrasound or magnetic thology of anal glands. Ann R Coll Surg Engl disease: effect on patient management. AJR resonance imaging for preoperative assess- 1967;40:36–67. Am J Roentgenol 1987;148:537–540. ment of anal fistula: a comparative study. Colorectal Dis 2001;3:189–197. 5. Parks AG. Pathogenesis and treatment of 21. Yousem DM, Fishman EK, Jones B. Crohn fistula-in-ano. Br Med J 1961;5224:463–469. disease: perirectal and perianal findings at 34. Kruskal JB, Kane RA, Morrin MM. Perox- ide-enhanced anal endosonography: tech- 6. Parks AG, Gordon PH, Hardcastle JD. A CT. Radiology 1988;167:331–334. nique, image interpretation, and clinical ap- classification of fistula-in-ano. Br J Surg 22. Schratter-Sehn AU, Lochs H, Vogelsang H, plications. RadioGraphics 2001;21(Spec 1976;63:1–12. Schurawitzki H, Herold C, Schratter M. En- Issue):S173–S189. doscopic ultrasonography versus computed 7. Langman JM, Rowland R. Density of lym- tomography in the differential diagnosis of 35. Chew SS, Yang JL, Newstead GL, Douglas phoid follicles in the rectum and at the ano- perianorectal complications in Crohn’s dis- PR. Anal fistula: Levovist-enhanced endoanal rectal junction. J Clin Gastroenterol 1992; ease. Endoscopy 1993;25:582–586. ultrasound—a pilot study. Dis Colon Rectum 14:81–84. 2003;46:377–384. 23. Law PJ, Bartram CI. Anal endosonography: 8. Buchan R, Grace RH. Anorectal suppu- technique and normal anatomy. Gastrointest 36. Buchanan GN, Bartram CI, Williams AB, ration: the results of treatment and the fac- Radiol 1989;14:349–535. Halligan S, Cohen CR. Value of hydrogen tors influencing the recurrence rate. Br J peroxide enhancement of three-dimensional Surg 1973;60:537–540. 24. Sultan AH, Kamm MA, Hudson CN, Thomas endoanal ultrasound in fistula-in-ano. Dis J, Bartram CI. Anal sphincter disruption 9. Fucini C. One stage treatment of anal ab- Colon Rectum 2005;48:141–147. during vaginal delivery. N Engl J Med 1993; scesses and fistulae. A clinical appraisal on 329:1905–1911. 37. Buchanan GN, Halligan S, Bartram CI, Wil- the basis of two different classifications. Int J liams AB, Tarroni D, Cohen CRG. Clinical Colorectal Dis 1991;6:12–16. 25. Frudinger A, Bartram CI, Halligan S, Kamm examination, endosonography, and MR im- MA. Examination techniques for anal en- aging in preoperative assessment of fistula in 10. Eisenhammer S. Advance of anorectal sur- dosonography. Abdom Imaging 1998;23: ano: comparison with outcome-based refer- gery with special reference to ambulatory 301–303. treatment. S Afr Med J 1954;28:264–266. ence standard. Radiology 2004;233(3):674– 26. Frudinger A, Halligan S, Bartram CI, Price 681. 11. Sainio P. Fistula-in-ano in a defined popula- AB, Kamm MA, Winter R. Female anal 38. Madsen SM, Myschetzky PS, Heldmann U, tion. Incidence and epidemiological aspects. sphincter: age-related differences in asymp- Rasmussen OO, Thomsen HS. Fistula in ano: Ann Chir Gynaecol 1984;73:219–224. tomatic volunteers with high frequency en- evaluation with low-field magnetic resonance doanal US. Radiology 2002;224:417–423. 12. Practice parameters for treatment of fistula- imaging (0.1 T). Scand J Gastroenterol in-ano: supporting documentation. The Stan- 27. Law PJ, Talbot RW, Bartram CI, et al. Anal 1999;34:1253–1256. dards Practice Task Force. The American endosonography in the evaluation of perianal 39. Fishman-Javitt MC, Lovecchio JL, Javors B, Society of Colon and Rectal Surgeons. Dis sepsis and fistula-in-ano. Br J Surg 1989;76: Naidich JB, McKinley M, Stein HL. The value Colon Rectum 1996;39:1363–1372. 752–755. of MRI in evaluating perirectal and pelvic 13. Kuster GG. Relationship of anal glands and 28. Deen KI, Williams JG, Hutchinson R, Keigh- disease. Magn Reson Imaging 1987;5:371– lymphatics. Dis Colon Rectum 1965;8:329– ley MR, Kumar D. Fistula in ano: endoanal 380. 332. ultrasonographic assessment assists decision 40. Koelbel G, Schmiedl U, Majer MC, et al. making for surgery. Gut 1994;35:391–394. 14. Milligan ET, Morgan CN. Surgical anatomy Diagnosis of fistulae and sinus tracts in pa- of the anal canal with special reference to 29. Lindsey I, Humphreys MM, George BD, tients with Crohn disease: value of MR imag-

32 Radiology: Volume 239: Number 1—April 2006 STATE OF THE ART: IMAGING OF FISTULA IN ANO Halligan and Stoker

ing. AJR Am J Roentgenol 1989;152:999– using endoanal receiver coils. AJR Am J Windsor AC, Ambrose NS. Outcome after 1003. Roentgenol 1997;169:201–206. surgery for perianal fistula: predictive value of MR imaging. AJR Am J Roentgenol 1998; 52. Halligan S, Bartram CI. MR imaging of fistula 41. Lunniss PJ, Armstrong P, Barker PG, et al. 171:403–406. Magnetic resonance imaging of anal fistulae. in ano: are endoanal coils the gold standard? Lancet 1992;340:394–396. AJR Am J Roentgenol 1998;171:407–412. 64. Buchanan GN, Halligan S, Williams AB, et al. Magnetic resonance imaging for primary fis- 42. Barker PG, Lunniss PJ, Armstrong P, 53. Stoker J, Lame´ris JS. MR imaging of perianal tula in ano. Br J Surg 2003;90:877–881. Reznek RH, Cottam K, Phillips RK. Magnetic fistulas using body and endoanal coil. AJR 65. Taylor SA, Halligan S, Bartram CI. Pilonidal resonance imaging of fistula-in-ano: tech- Am J Roentgenol 1999;172:1139–1140. sinus disease: MR imaging distinction from nique, interpretation and accuracy. Clin Ra- 54. Stoker J, Rociu E, Schouten WR, Lame´ris fistula in ano. Radiology 2003;226:662–667. diol 1994;49:7–13. JS. Anovaginal and rectovaginal fistulas: en- 66. Platell C, Mackay J, Collopy B, Fink R, Ryan 43. Spencer JA, Ward J, Beckingham IJ, Adams doluminal sonography versus endoluminal P, Woods R. Anal pathology in patients with C, Ambrose NS. Dynamic contrast-en- MR imaging. AJR Am J Roentgenol 2002; Crohn’s disease. AustNZJSurg 1996;66: hanced MR imaging of perianal fistulas. AJR 178:737–741. 5–9. Am J Roentgenol 1996;167:735–741. 55. Maier AG, Funovics MA, Kreuzer SH, et al. 67. Winter AM, Hanauer SB. Medical manage- 44. Van Beers B, Grandin C, Kartheuser A, et al. Evaluation of perianal sepsis: Comparison of ment of perianal Crohn’s disease. Semin MRI of complicated anal fistulae: comparison anal endosonography and magnetic reso- Gastrointest Dis 1998;9:10–14. nance imaging. J Magn Reson Imaging 2001; with digital examination. J Comput Assist 68. Schwartz DA, Loftus EV Jr, Tremaine WJ, et 14:254–260. Tomogr 1994;18:87–90. al. The natural history of fistulizing Crohn’s 45. Stoker J, Hussain SM, van Kempen D, Elev- 56. Halligan S, Healy JC, Bartram CI. Magnetic disease in Olmsted County, Minnesota. Gas- elt AJ, Lame´ris JS. Endoanal coil in MR im- resonance imaging of fistula-in-ano: STIR or troenterology 2002;122:875–880. aging of anal fistulas. AJR Am J Roentgenol SPIR? Br J Radiol 1998;71:141–145. 69. Sandborn WJ, Feagan BG, Hanauer SB, et 1996;166:360–362. 57. Myhr GE, Myrvold HE, Nilsen G, et al. Peri- al. A review of activity indices and efficacy endpoints for clinical trials of medical ther- 46. Beets-Tan RG, Beets GL, van der Hoop AG, anal fistulas: use of MR imaging for diagnosis. Radiology 1994;191:545–554. apy in adults with Crohn’s disease. Gastro- et al. Preoperative MR imaging of anal enterology 2002;122:512–530. fistulas: does it really help the surgeon? Ra- 58. Mergo PJ, Helmberger T, Cerda JJ, Urrutia 70. van Assche G, Vanbeckevoort D, Bielen D, diology 2001;218:75–84. M, Ros PR. Rectal perflubron: new applica- et al. Magnetic resonance imaging of the ef- tion in MRI of perirectal fistulae. J Comput 47. Buchanan G, Halligan S, Williams A, et al. fects of infliximab on perianal fistulizing Assist Tomogr 1997;21:259–264. Effect of MRI on clinical outcome of recur- Crohn’s disease. Am J Gastroenterol 2003; rent fistula-in-ano. Lancet 2002;360:1661– 59. Sabir N, Sunguntekin U, Erdem E, Nessar M. 98:332–339. 1662. Magnetic resonance imaging with rectal Gd- 71. Buchanan GN, Bartram CI, Phillips RK, et al. DTPA: new tool for the diagnosis of perianal 48. Hussain SM, Stoker J, Schouten WR, Hop Efficacy of fibrin sealant in the management fistula. Int J Colorectal Dis 2000;15:317– WC, Lameris JS. Fistula in ano: endoanal of complex anal fistula: a prospective trial. 322. sonography versus endoanal MR imaging in Dis Colon Rectum 2003;46:1167–1174. classification. Radiology 1996;200:475–481. 60. Stoker J, Jong Tjien Fa VE, Eijkemans MJ, 72. Present DH, Rutgeerts P, Targan S, et al. Schouten WR, Lame´ris JS. Endoanal MR im- Infliximab for the treatment of fistulas in pa- 49. deSouza NM, Gilderdale DJ, Coutts GA, aging of perianal fistulas: the optimal imaging tients with Crohn’s disease. N Engl J Med Puni R, Steiner RE. MRI of fistula-in-ano: a planes. Eur Radiol 1998;8:1212–1216. 1999;340:1398–1405. comparison of endoanal coil with external phased array coil techniques. J Comput As- 61. Buchanan GN, Williams AB, Bartram CI, 73. Bell SJ, Halligan S, Windsor AC, Williams sist Tomogr 1998;22:357–363. Halligan S, Nicholls RJ, Cohen CR. Potential AB, Wiesel P, Kamm A. Response of fistulat- clinical implications of direction of a trans- ing Crohn’s disease to infliximab treatment 50. Stoker J, Rociu E, Zwamborn AW, Schouten sphincteric anal fistula track. Br J Surg 2003; assessed by magnetic resonance imaging. Al- WR, Lame´ris JS. Endoluminal MR imaging of 90:1250–1255. iment Pharmacol Ther 2003;17:387–393. the rectum and anus: technique, applica- 74. van Bodegraven AA, Sloots CE, Felt-Bersma tions, and pitfalls. RadioGraphics 1999;19: 62. deSouza NM, Kmiot WA, Puni R, et al. High RJ, Meuwissen SG. Endosonographic evi- 383–398. resolution magnetic resonance imaging of dence of persistence of Crohn’s disease-as- the anal sphincter using an internal coil. Gut 51. deSouza NM, Gilderdale DJ, MacIver DK, sociated fistulas after infliximab treatment, 1995;37:284–287. Ward HC. High-resolution MR imaging of irrespective of clinical response. Dis Colon the anal sphincter in children: a pilot study 63. Spencer JA, Chapple K, Wilson D, Ward J, Rectum 2002;45:39–46.

Radiology: Volume 239: Number 1—April 2006 33