This retrospective study was approved by the institutional review board of the Affiliated Hospital of Nanjing University of Chinese Medicine, which waived the requirement for informed consent. Between June 2006 and August 2018, ten patients diagnosed with carcinoma arising from anal fistula underwent preoperative pelvic MRI. Age, sex, pathological subtypes, clinical symptoms, duration of anal fistula, and history of surgery were recorded (Table 1). All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and national research committees and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

MRI was performed either on a 1.5-T system (Aera, Siemens AG, Munich, Germany) (n = 9) or 3.0-T system (Verio, Siemens AG, Munich, Germany) (n = 1). Unenhanced MRI of the pelvis was performed in all the ten patients. In five patients, this was followed by dynamic contrast enhanced (DCE) scans. In the remaining five, this was followed by diffusion-weighted imaging (DWI). All patients were examined using a pelvic phased-array coil. The imaging technique used for pelvic MRI at our institution is provided in Table 2.

Carcinoma arising from anal fistula was diagnosed according to Rosser’s criteria: (1) The fistula should usually antedate the carcinoma by at least 10 years; (2) The only tumor present in the rectum or anal canal should be secondary to direct extension from the carcinoma in the fistula; and (3) The internal opening of the fistula should be into the anal canal and not into the tumor itself[12].

Pathological specimens were observed by light microscopy and immunohistochemical analysis. All tumors were confirmed to be carcinomas arising from anal fistula by surgery (n = 7) or biopsy (n = 3). Cases of rectal cancer, malignant signs in the anal mucosa, and the internal opening of fistula into the tumor were excluded before definite diagnosis.

Pelvis MRI images were reviewed and interpreted in consensus by two radiologists with 5 and 20 years of experience, respectively, who were blinded to the histopathologic results. Morphologic features (location, shape, maximum diameter, capsule, and margin); signal characteristics [fat-suppressed T2-weighted MRI (FS-T2WI), DWI, and apparent diffusion coefficient (ADC)]; fistula between the mass and anus; contrast enhancement of mass; signal on FS-T2WI and enhancement of peritumoral areas; and regional lymphadenopathy were assessed on MRI. These features were selected based on the previous literature and our personal experience[9-11].

The capsule was observed as a hypointense area on both T1-weighted magnetic resonance imaging (T1WI) and T2WI. Signal intensity on FS-T2WI was classified as markedly hyperintense, hyperintense, and slightly hyperintense. Marked hyperintensity was defined as signal intensity similar to that of fluid in the bladder. Hyperintensity was defined as signal intensity higher than that of muscle but lower than that of fluid. Slight hyperintensity was defined as signal intensity slightly higher than that of muscle. Fistula between the mass and the anus was defined as a tubular or linear hyperintense communication between the mass and the anus assessed on FS-T2WI.

The enhancement pattern of tumor was assessed on T1-DCE images. We observed the enhancement characteristics of the tumor including the inner edge of the lesion (smooth or rough), enhancement of septum, and unenhanced necrotic area within the lesion. Contrast enhancement of peritumoral areas was also assessed on T1-DCE images. If there was enhancement around the tumor, the signal intensity of this area was observed on FS-T2WI.

Regional lymphadenopathy was defined as perirectal and inguinal lymph nodes > 5 mm in the short axis, morphological change (irregular or round), signal intensity on FS-T2WI, heterogeneous enhancement, or hyperintensity on DWI[13].

A previous report suggested that this condition is slightly more common in men, with the average age at diagnosis being 55 years[11]. Our results were consistent with this report, given that most of our patients were male (8/10), and most of the patients were middle-aged or elderly (9/10; age range: 50-75 years).

There are three main pathological subtypes of carcinoma arising from an anal fistula—mucinous adenocarcinoma, adenocarcinoma, and squamous cell carcinoma[16-19]. Thomas et al[15] stated that the histology of adenocarcinoma was the most common (~59%); however, mucinous adenocarcinoma was the most common pathological type in our study (8/10). The possible reason is that previous studies considered only two major types, namely, adenocarcinoma and squamous cell carcinoma, and mucinous adenocarcinoma was classified as adenocarcinoma.

Early symptoms of carcinoma arising from anal fistula are similar to those of anal fistula or perianal abscess, usually without diarrhea or obstruction, which may help to distinguish from rectal cancer. Digital examination may reveal only an area of induration on the side where the fistula is situated. Six patients completely fulfilled the above criteria in our study, while in the remaining four patients, the anal fistula has been present < 10 years. To ascertain whether the standard “the fistula should usually antedate the carcinoma by at least 10 years” is still decisive, it is necessary to analyze a larger group of patients.

Abdominoperineal resection (APR) is the most common surgical treatment for this rare malignancy, and the effect of radiotherapy and chemotherapy is still controversial[15,17]. Prognosis of carcinoma arising from anal fistula is usually poor[20].

The masses could be unilateral or grow across the midline and typically surrounded the anal canal. The size of the tumor lesion could be relatively large. The mean of the maximum lesion diameter was 5.68 cm in our patients, and even exceeded 10 cm in two casas. In our study, the typical features of mucinous adenocarcinoma with respect to shape were multiloculated or cauliflower-like with internal septa. Hama et al[10] hypothesized that the mucin pools produced by a mucinous adenocarcinoma without thick fibrous capsule may reflect that cancer cells have invaded the surrounding tissues before the fibrotic reaction. However, the perianal fistulas or abscess whose mucin was secreted by the anal gland had a thick fibrous capsule. We speculated that the thin capsule could still be seen in a mucinous adenocarcinoma, and the focal boundary of tumors was unclear, which could reflect the invasive nature. Adenocarcinoma lesions appeared ill-defined and had no capsule. Because of the small number of adenocarcinoma cases, the relevant characteristics could not be effectively summarized.

Mucinous adenocarcinoma mostly presented with markedly hyperintense signals on T2-weighted MR images, slightly hyperintense signals on DWI, and hyperintense signals on ADC map, which represented the signal of mucin pool. Some lesions of mucinous adenocarcinoma contained solid components, showing focal slightly hyperintense signal on FS-T2WI, hyperintense signals on DWI, and hypointense signals on ADC map. Adenocarcinoma showed heterogeneous high or slightly high signal on FS-T2WI, high signal on DWI, and low signal on ADC map.

Hama et al[10] reported that a fistula between the mass and the anus is a characteristic finding of mucinous adenocarcinoma arising from an anal fistula. Our results are consistent with this conclusion because eight patients had a fistula between the mass and the anus, and two patients showed the masses surrounding the anal canal and appearing to be connected to the canal. Because of the history of anal fistula in all patients, whether the fistulas between the mass and the anus were caused by the tumor or the previous anal fistula remains to be further explored. At present, we are inclined to the latter possibility.

The lesions of mucinous adenocarcinoma showed progressively mesh-like contrast enhancement, septal enhancement, and rough inner margin. Although the number of patients with post-contrast MRI scan was small, the results were basically consistent with those reported in the literature[10]. Meanwhile, we considered that the progressively enhanced pattern was similar to that of the fiber, which was indicative of the existence of an internal septum and thin capsule. Adenocarcinomas presented with persistent heterogeneous enhancement.

Contrast enhancement of the peritumoral areas was noted in three of our cases. Hama et al[10] believed that this could be a reliable finding reflecting the invasion of cancer cells. We found it challenging to distinguish contrast enhancement from enhancement of active inflammation. Considering the unclear boundary and the slightly hyperintense signal on FS-T2WI of the enhanced region, we suspected that the tumor invasion and inflammation were coexisting conditions. We recommend MRI re-examination after remission of inflammation.

Perirectal or inguinal lymphadenopathy was noted in seven of ten patients in our study. Unfortunately, pathologic confirmation could not be obtained. Iesalnieks et al[21] found that approximately 50% appeared inguinal lymph node involvement and about 10% presented with distant organ metastasis. Therefore, we believe that the signs of lymph nodes are helpful in the diagnosis of carcinoma arising from anal fistula.

Perianal abscess is often accompanied with anal fistula; therefore, it is considered as the main differential diagnosis of mucinous adenocarcinoma arising from the anal fistula. Perianal abscess could be unilocular or multilocular and is typically associated with active inflammation. It presents higher signal intensity and lower signal intensity than mucinous adenocarcinoma on DWI and ADC maps. Essentially, the abscess usually demonstrates ring-enhancement without containing solid components.

The fistula between the mass and the anus and the history of long-standing chronic anal fistula could be helpful to distinguish between mucinous adenocarcinoma arising from anal fistula and malignant transformation of cystic lesions in the pelvic cavity, such as a tailgut cyst, teratoma, or dermoid cyst[10].

Our study has some limitations. First, this was a retrospective study of patients from a single institution. Second, carcinoma arising from anal fistula is a rare diagnosis, and the sample size is therefore relatively small. It remains to be further analyzed whether imaging features of lesions could be completely summarized. Third, the included patients had a relatively long time span (between June 2006 and August 2018), and the corresponding MRI techniques were therefore not consistent.

Magnetic resonance imaging (MRI) is the standard investigation for suspected perianal diseases. This is important to early diagnose carcinoma arising from anal fistula, even if the incidence is low.

To describe and summarize the MRI findings of carcinoma arising from anal fistula.

To summarize the MRI manifestations of carcinoma arising from anal fistula to help improve the ability to diagnose this entity.

A retrospective study was performed on ten patients diagnosed with carcinoma arising from anal fistula and confirmed by pathological diagnosis between June 2006 and August 2018. All patients underwent preoperative pelvic MRI. Five patients underwent enhanced MRI scans. Rosser’s criteria were used for diagnosing carcinoma arising from anal fistula. Morphologic features, signal characteristics, fistula between the mass and the anus, contrast enhancement of mass, signal and enhancement of peritumoral areas, and regional lymphadenopathy were assessed.

Most patients (90%) were older than 50 years. There were eight mucinous adenocarcinomas and two adenocarcinomas. The maximum diameter of the tumors ranged from 3.4 cm to 12.4 cm (median: 4.15 cm; mean: 5.68 cm). Eight patients had a fistula between the mass and the anus. Perirectal or inguinal lymphadenopathy was frequent (7/10).

Most lesions of mucinous adenocarcinoma were multiloculated and cauliflower-like, with a thin capsule and focally unclear boundary. They were markedly hyperintense on fat-suppressed T2WI, slightly hyperintense with focal hyperintense on diffusion-weighted imaging (DWI), and hyperintense with focal hypointensity on apparent diffusion coefficient (ADC) map, with progressive mesh-like contrast enhancement.

Adenocarcinomas had an infiltrative margin without a capsule and appeared heterogeneously hyperintense or slightly hyperintense on fat-suppressed T2WI, hyperintense on DWI, and hypointense on ADC map, with persistent heterogeneous enhancement.

A negative biopsy does not rule out the diagnosis of cancer, clinicians should be highly suspicious of cancer in patients with chronic perineal fistulas whose symptoms change, and a repeat biopsy should be recommended. The definitive diagnosis of cancer arising from an anal fistula is always challenging. Although the number of cases was inadequate owing to the rarity of the disease, we believe that several characteristic MRI findings could contribute to accurate and timely diagnosis of carcinoma arising from anal fistula in selection of puncture route and screening of high-risk groups.

Earlier and better identification and monitoring of high-risk groups of carcinoma arising from anal fistula are required.