High-resolution direct magnetic resonance imaging fistulography with hydrogen peroxide for diagnosing anorectal fistula: A preliminary retrospective study

Abstract

BACKGROUND

Fistula-in-ano is an abnormal tunnel formation linking the anal canal with the perineum and perianal skin. Multiple imagining methods are available to evaluate it, among which magnetic resonance imaging (MRI) is the most advanced noninvasive preoperative method. However, it is limited in its visualization function.

AIM

To investigate the use of intraluminal MRI for perianal fistulas via a novel direct MRI fistulography method.

METHODS

We mixed 3% hydrogen peroxide (HP) with gadolinium for HPMRI fistulography, retrospectively analyzing 60 cases of complex/recurrent fistula-in-ano using physical examination, trans-perineal ultrasonography (TPUS), low-spatial-resolution MRI, and high-resolution direct HPMRI fistulography. We assessed detection rates of fistula tracks, internal openings, their relationship with anal sphincters, and perianal abscesses using statistical analyses, including interobserver agreement (Kappa statistic), and compared results with intraoperative findings.

RESULTS

Surgical confirmation in 60 cases showed that high-resolution direct HPMRI fistulography provided superior detection rates for internal openings (153) and fistula tracks (162) compared to physical exams, TPUS, and low-spatial-resolution MRI (Z > 5.7, P < 0.05). The effectiveness of physical examination and TPUS was also inferior to that of our method for detecting perianal abscesses (54) (Z = 6.773, 3.694, P < 0.05), whereas that of low-spatial-resolution MRI was not significantly different (Z = 1.851, P = 0.06). High-resolution direct HPMRI fistulography also achieved the highest interobserver agreement (Kappa: 0.89, 0.85, and 0.80), while low-spatial-resolution MRI showed moderate agreement (Kappa: 0.78, 0.74, and 0.69). TPUS and physical examination had lower agreement (Kappa range: 0.33-0.63).

CONCLUSION

High-resolution direct HPMRI fistulography enhances the visualization of recurrent and complex fistula-in-ano, including branched fistulas, allowing for precise planning and improved surgical outcomes.

Core Tip: High-resolution direct hydrogen peroxide (HP)-enhanced magnetic resonance imaging (MRI) fistulography offers superior visualization of complex and recurrent fistula-in-ano. It enhances diagnostic accuracy for internal openings, fistula tracks, and perianal abscesses, surpassing physical examination, trans-perineal ultrasound, and low-resolution MRI. With high sensitivity, specificity, and interobserver reliability, HPMRI aids precise surgical planning, reducing recurrence rates. This technique represents a valuable advancement in the preoperative assessment of anorectal fistulas, combining enhanced soft-tissue contrast with detailed anatomical visualization to address diagnostic challenges in intricate cases effectively. Future standardization and training could improve accessibility and further validate its clinical utility.

INTRODUCTION

Fistula-in-ano, a chronic perianal tissue infection characterized by the formation of an abnormal tunnel connecting the anal canal with the perineum and perianal skin, is a prevalent disease of the anorectal canal[1,2]. Crohn’s disease, pelvic infections, tuberculosis, diverticulitis, birth trauma, pelvic malignancies, and radiotherapy are factors that can contribute to the development of this disease[3]. Fistula-in-ano is typically more common in younger adults and has a male-to-female ratio of 2:1[4]. The main symptoms include discharge near the perianal fistula and localized pain[5]. In addition to Parks and Marks’s classifications, we use the term “recurrent fistula” in this paper, which is defined as a fistula that develops after the failure of a previous operation[6-8]. Several imaging techniques are available for evaluating fistula-in-ano, including X-ray fistulography, computed tomography (CT), ultrasound, and magnetic resonance imaging (MRI) of which ultrasound and MRI are the most common[9,10]. Recent studies have highlighted the effectiveness of three-dimensional endoanal ultrasonography and hydrogen peroxide (HP)-enhanced contrast ultrasonography for evaluating fistula-in-ano owing to their high sensitivity and specificity. However, these two methods cannot always detect highly restricted (suprasphincteric), subcutaneous, horseshoe-type, or small additional branching lesions[11]. MRI offers excellent soft-tissue contrast, facilitating direct visualization and quantification of internal openings, fistula tracks, and abscess cavities[12-14]. However, despite being the most advanced non-invasive preoperative assessment method for fistula-in-ano, MRI has limitations related to visualizing the connection between the internal opening and fistula track[15,16].

Here, we determined the feasibility of performing intraluminal MRI of perianal fistulas via a novel direct MRI fistulography method. Additionally, we determined the diagnostic value of this method for recurrent and complex fistula-in-ano via a preliminary study.

MATERIALS AND METHODS

Patient data

We conducted a retrospective analysis of 60 patients with fistula-in-ano admitted to Bozhou Hospital of Traditional Chinese Medicine who underwent surgery between October 2021 and July 2024. All patients had a confirmed history of fistula-in-ano and/or exhibited clinical symptoms of a perianal abscess. All patients were scheduled for a follow-up MRI examination and subsequent surgical intervention for treatment and clinical and pathological confirmation. The inclusion criteria were as follows: (1) Presence of perianal swelling and pain, purulence, pruritus ani, or discharge on admission, as well as a diagnosis of recurrent or complex fistula-in-ano confirmed through digital rectal examination and intraoperative visual confirmation; (2) No contraindications for undergoing MRI examination; and (3) No contraindications for undergoing surgery. The exclusion criteria were as follows: (1) Patients aged < 18 years; (2) Pregnant patients; (3) Patients with mental illnesses hindering their ability to cooperate during examinations; and (4) Patients with fistula-in-ano without external openings or whose external openings have closed. Ultimately, 60 patients with a mean age of 38.5 years (range: 20-65 years) were enrolled in this study. Table 1 summarizes the clinical characteristics of the patients. All patients underwent physical and imaging examinations, and the major fistulas were evaluated based on the classification system developed by Parks et al[6].

Table 1.

Clinical characteristics of patients with fistula-in-ano (n = 60)

Clinical characteristic	Cases, n	Percentage (%)
Age (years)
Median	38.5 (20-65)
Distribution
≤ 45	45	75
> 45	15	25
Sex
Male	57	95
Female	3	5
Symptoms
Discharge in the vicinity of perianal fistula	60	100
Localized pain	54	90
History of treatment
None	24	38.09
Medication	9	14.28
Previous perianal abscess surgery	18	28.57
Previous fistula-in-ano surgery	12	19.04

Procedures

All patients underwent a clinical subspecialty physical examination, including visual perianal examination, perianal area and fistula palpation, and proctoscopy. Furthermore, all patients underwent preoperative ultrasound subspecialty examinations (LOGIQ 5 color Doppler ultrasound with a high-frequency linear array superficial transducer and a frequency of 12 MHz, General Electric Company) and pelvic MRI plain scans (8-channel body coil, Philips Ingenia 1.5T, the Netherlands).

Immediately before patient examination, a 5 mL HP-gadoterate meglumine mixture was prepared at a ratio of two drops of gadoterate meglumine injection per 1 mL of 3% HP. The patients were positioned for examination in the left lateral recumbent posture with their legs flexed. The skin around the external opening of the fistula was disinfected, and the solution was aspirated into a 5 mL disposable syringe. A needleless disposable catheter was connected to the external opening, through which the solution was subsequently injected into the fistula. To prevent any leakage of the contrast medium from the external opening resulting from the injection pressure, the catheter was firmly secured to the external opening. Approximately 1-5 mL of the solution was injected. Subsequently, the external opening was sealed with a gauze swab and a transparent medical dressing. The examination was performed using a 1.5T MRI system (Philips Ingenia), equipped with a specialized 8-channel body coil positioned at the sacroiliac joint level. The patient was positioned head-first in the apparatus, lying supine, with the center of the magnetic field aligned at the level of the pubic symphysis.

First, a Turbo spin echo (TSE) T2WI scan was conducted in the sagittal orientation, aligned with the long axis of the body, to help establish the relative anatomical position of the anal canal findings. Second, sagittal scans were performed in three different positions: Oblique coronal, parallel to the anal canal, perpendicular to the anal canal, and oblique axial orientation. The image sequences, matrix sizes, fields of view, and other technical parameters are summarized in Table 2. In the low-spatial-resolution MRI protocol, we did not routinely perform gadolinium-enhanced MRI or diffusion-weighted imaging for patients with fistula-in-ano. Both MRIs were conducted on the same day.

Table 2.

Magnetic resonance imaging protocol.

Sequences	TR (millisecond)	TE (millisecond)	Slice thickness (mm)	Slice spacing (mm)	FOV (mm)	NSA	Voxel size (mm3)
Low-spatial-resolution MRI protocol
Sagittal T2WI TSE	3000	100	7	1.5	250 × 350	1	0.87 × 1.16 × 7
Oblique axial T2WI TSE	3361	80	6	1	300 × 400	1	1.1 × 1.39 × 6
Oblique coronal T2WI TSE	3000	100	4	0.4	250 × 250	1	0.87 × 1.16 × 4
Oblique axial/coronal T1WI-TSE	520	10	6	1	300 × 400	1	1 × 1.15 × 6
High-resolution direct HPMRI fistulography protocol
Oblique axial/coronal 3D-T2WI-FSE	2000	244	2	-1	200 × 200	2	1 × 1 × 2
Oblique axial/coronal T1WI-mDIXON-TFE	6.4	2.1	3	-1.5	200 × 200	1	1 × 1 × 3
Oblique axial/coronal T1WI-mDIXON-TFE (postcontrast)	6.4	2.1	3	-1.5	200 × 200	1	1 × 1 × 3

FOV: Field of view; HP: Hydrogen peroxide; MRI: Magnetic resonance imaging; TE: Echo time; TR: Repetition time; TSE: Turbo spin echo; NSA: Number of signal averages.

Image analysis

The imaging results were analyzed by two physicians (associate attending physician or higher title) in the radiology and ultrasound departments, respectively. All disagreements were resolved through consensus. The primary observational indicators included the internal and external openings of the fistula and abscess locations, fistula morphology and tract, and their connection to the anus, rectum, anal sphincter, and levator ani muscle. Each patient underwent surgery conducted by the same surgeon with > 15 years of experience in subspecialty surgery, who was blinded to the imaging results. Throughout the procedure, the surgeon injected methylene blue dye into the fistula and closely observed the location of the internal opening. All imaging results were compared with intraoperative observations.

Statistical analysis

SPSS 22.0 (IBM Corp., Armonk, NY, United States) and MedCalc 20.0 (MedCalc Software, Ostend, Belgium) were utilized for data processing and analysis. Specifically, MedCalc was used to construct receiver operating characteristic (ROC) curves and calculate key metrics such as the area under the curve, sensitivity, specificity, and positive and negative predictive values of the four different tests for diagnosing the internal opening, fistula track, and perianal abscess. The diagnostic performance of the four methods was compared using the DeLong test. The significance level (a) was set to 0.05. Kappa statistics were used to evaluate interobserver agreement across four diagnostic modalities.

RESULTS

Comparison of four different methods for fistula-in-ano with postoperative pathological findings

Table 3 shows the number of different types of fistula correctly identified using four methods, namely physical examination, trans-perineal ultrasonography (TPUS), low-spatial-resolution MRI, and high-resolution direct HPMRI fistulography, as compared to those identified during surgery, which serves as the gold standard. Table 4 shows the sensitivity, specificity, positive predictive value, and negative predictive value for detecting fistula tracks, internal openings, and perianal abscesses compared to those of surgical findings. The interobserver agreement varied significantly across the diagnostic modalities. High-resolution direct HPMRI fistulography demonstrated the highest degree of agreement for all observational indicators, with Kappa values of 0.89, 0.85, and 0.80 for internal openings, fistula tracks, and perianal abscesses, respectively. Low-spatial-resolution MRI showed moderate agreement, with Kappa values of 0.78, 0.74, and 0.69. TPUS and physical examination demonstrated a lower degree of agreement, with Kappa values ranging from 0.33 to 0.63 (Table 5).

Table 3.

Methods for recurrent and complex fistula-in-ano in 60 patients

Type of fistula	Physical examination	Conventional ultrasound (TPUS)	Low-spatial-resolution MRI	HPMRI fistulography	Surgery
Intersphincteric	27	63	78	84	90
Transphincteric	9	36	42	48	54
Extrasphincteric	0	0	0	9	9
Suprasphincteric	0	0	0	9	9
Internal opening	72	99	120	147	153
Perianal abscess	18	36	45	54	54

HP: Hydrogen peroxide; MRI: Magnetic resonance imaging; TPUS: Trans-perineal ultrasonography.

Table 4.

Comparison of findings from different methods used for fistula-in-ano

Observational indicator	Evaluation methods	True positive	False positive	False negative	Sensitivity	Specificity	Positive predictive value	Negative predictive value	Surgical outcome
Internal opening	Physical examination	72	36	81	47.06 (72/153)	63.64 (63/99)	66.67 (72/108)	43.75 (63/144)	153
	TPUS	99	27	54	64.71 (99/153)	72.73 (72/99)	78.57 (99/126)	57.14 (72/126)	153
	Low-spatial-resolution MRI	120	12	33	78.43 (120/153)	87.88 (87/99)	90.91 (120/132)	72.50 (87/120)	153
	High-resolution direct HPMRI fistulography	147	9	6	96.08 (147/153)	90.91 (90/99)	94.23 (147/156)	90.91 (90/96)	153
Fistula track	Physical examination	36	27	126	22.22 (36/162)	66.67 (54/81)	57.14 (36/63)	30.00 (54/180)	162
	TPUS	99	18	63	61.11 (99/162)	77.78 (63/81)	84.62 (99/117)	50.00 (63/126)	162
	Low-spatial-resolution MRI	120	9	42	74.07 (120/162)	88.89 (72/81)	93.02 (120/129)	63.16 (72/114)	162
	High-resolution direct HPMRI fistulography	150	6	12	92.59 (150/162)	92.59 (75/81)	96.15 (150/156)	86.21 (75/87)	162
Abscess	Physical examination	18	36	36	33.33 (18/54)	76.47 (117/153)	33.33 (18/54)	76.47 (117/153)	54
	TPUS	136	27	18	66.67 (36/54)	82.35 (126/153)	57.14 (36/63)	87.50 (126/144)	54
	Low-spatial-resolution MRI	45	9	9	83.33 (45/54)	94.12 (144/153)	83.33 (45/54)	94.12 (144/153)	54
	High-resolution direct HPMRI fistulography	54	6	0	100 (54/54)	96.08 (147/153)	90.00 (54/60)	100 (147/147)	54

HP: Hydrogen peroxide; MRI: Magnetic resonance imaging; TPUS: Trans-perineal ultrasonography.

Table 5.

Interobserver agreement across diagnostic modalities for detecting fistula-in-ano features

Observational indicator	Diagnostic modality	Kappa value	95%CI	Interpretation
Internal openings	Physical examination	0.48	0.40-0.56	Fair agreement
	TPUS	0.63	0.55-0.71	Moderate agreement
	Low-spatial-resolution MRI	0.78	0.71-0.85	Substantial agreement
	High-resolution HPMRI	0.89	0.83-0.95	Almost perfect agreement
Fistula tracks	Physical examination	0.41	0.34-0.48	Fair agreement
	TPUS	0.57	0.50-0.64	Moderate agreement
	Low-spatial-resolution MRI	0.74	0.68-0.80	Substantial agreement
	High-resolution HPMRI	0.85	0.79-0.91	Almost perfect agreement
Perianal abscess	Physical examination	0.33	0.25-0.41	Fair agreement
	TPUS	0.45	0.37-0.53	Moderate agreement
	Low-spatial-resolution MRI	0.69	0.62-0.76	Substantial agreement
	High-resolution HPMRI	0.8	0.74-0.86	Almost perfect agreement

HP: Hydrogen peroxide; MRI: Magnetic resonance imaging; TPUS: Trans-perineal ultrasonography.

ROC curve analysis

ROC curve analysis revealed that the effectiveness of the physical examination, TPUS, and low-spatial-resolution MRI was lower than that of high-resolution direct HPMRI fistulography for detecting internal openings (Z = 5.741, 4.142, 2.038, all P < 0.05) and for detecting fistula tracks (Z = 6.314, 3.658, 1.971, all P < 0.05). The effectiveness of physical examination and TPUS was inferior to that of high-resolution direct HPMRI fistulography for detecting perianal abscesses (Z = 6.773, 3.694, P < 0.05), whereas no statistically significant difference was observed between the effectiveness of low-spatial-resolution MRI and that of high-resolution direct HPMRI fistulography for detecting perianal abscesses (Z = 1.851, P = 0.06; Figure 1).

Figure 1.

Receiver operating characteristic curves of high-resolution direct hydrogen peroxide-enhanced magnetic resonance imaging fistulography, trans-perineal ultrasonography, low-spatial-resolution magnetic resonance imaging, and physical examination for detecting internal opening, fistula track, and perianal abscess, respectively.

DISCUSSION

MRI provides excellent soft-tissue resolution, multiplanar scanning, and 3D imaging, offering clear visualization of perianal anatomy[17]. It reliably identifies fistula tracks, sphincters, and internal openings, with reported sensitivities and specificities exceeding 97%[18]. Buchanan et al[19] noted a 10% higher diagnostic accuracy and three-fold lower recurrence rates compared to endoanal ultrasound (EAUS). Previous studies have shown that MRI reduces recurrence rates by up to 75% compared to other methods[20]. Moreover, the direct magnetic resonance angiography of fistula-in-ano via a contrast agent mixed with saline or HP significantly improves the diagnostic efficiency of fistula track evaluation[21]. High-resolution HPMRI further improves diagnostic accuracy, outperforming 3D-EAUS in detecting complex fistula branches and surpassing contrast-enhanced MRI in dynamic visualization of fistula-sphincter relationships. These advantages make HPMRI the preferred tool for evaluating complex or recurrent fistulas, ensuring precise surgical planning and consistent interobserver reliability. Our analysis demonstrated that high-resolution HPMRI fistulography exhibits excellent sensitivity and specificity, with detection rates of 96.08% and 92.59% for internal openings and fistula tracks, respectively, and specificities of 90.91% and 96.08%. Although some internal openings may be missed due to limited physician experience in image interpretation, the overall performance of HPMRI is significantly superior to that of other methods (Figure 2A-D). Specifically, we used the mDIXON sequence on a Philips 1.5T MRI system to acquire T1WI water images (slice thickness: 3 mm, slice spacing: -1.5 mm, voxel size: 1 mm × 1 mm × 3 mm) and 3D-T2WI sequences (slice thickness: 2 mm, slice spacing: -1 mm, voxel size: 1 mm × 1 mm × 2 mm), achieving high-resolution imaging. Compared to low-resolution MRI, high-resolution HPMRI significantly improves the signal-to-noise ratio, generates smaller voxels, and delivers higher spatial resolution. Considering that the adult anal canal has an approximate length of 2.5 cm and that the typical fistula-in-ano is relatively short, thinner slices and more images are ideal for visualizing fistula tracks and potential branches (Figure 2E). Other studies have reported that MRI sequences with slice thickness and spacing typically range between 3-4 mm and 0.3-1 mm, respectively[22]. Enhanced MRI sequences, such as FS-T1-3D-VIBE (slice thickness: 1.3 mm, slice spacing: 0.4 mm), compared to FS-T1-TSE (slice thickness: 4 mm, slice spacing: 0.4 mm), are reported to offer superior diagnostic potential for evaluating the location of internal openings, their distance from the anal verge, the clarity of the internal opening, and the classification of fistula tracks and complex fistulas[23]. Selecting shorter TR and TE values (e.g., in the mDIXON-TFE sequence for T1WI) effectively reduces motion artifacts and enhances soft-tissue contrast. Additionally, using thin slices (1 mm) and optimized signal-to-noise ratio settings (e.g., 3D-T2WI-FSE sequences) further improves the visualization of internal openings and fistula tracks (Figure 2F). In addition to superior detection efficiency, high-resolution direct HPMRI fistulography demonstrated the highest degree of interobserver agreement among the four diagnostic modalities analyzed, with Kappa values of 0.89, 0.85, and 0.80 for internal openings, fistula tracks, and perianal abscesses, respectively. Low-spatial-resolution MRI showed moderate agreement (Kappa: 0.78, 0.74, 0.69), while TPUS and physical examination demonstrated a lower degree of agreement (Kappa: 0.33-0.63). These results reinforce the diagnostic reliability of HPMRI, particularly in complex or branching fistulas, enabling consistent and reproducible imaging interpretation to aid surgical planning. In one case of complex fistula-in-ano, two external openings were observed. When a mixed contrast agent was injected into one, bubbles emerged from the other, suggesting communication between the two external openings. In another case of complex fistula-in-ano, three external openings were observed, one of which was occluded. When mixed contrast agent was injected into one of the external openings, MRI indicated that the double-track response initially meandered toward the anal canal then reversed, extending toward the closed external openings. Thus, these fistula-in-ano cases had complex branching with dynamics that could not be effectively analyzed via ultrasound. In addition, the ability to ascertain fistula tracks may be slightly compromised when fistulography cannot be conducted because of the closure of some external openings. The efficacy of physical examination and TPUS in detecting perianal abscess was lower than that of high-resolution direct HPMRI fistulography (Z = 6.773, 3.694, P < 0.05). However, the difference between low-spatial-resolution MRI and high-resolution direct HPMRI fistulography for detecting perianal abscesses was not statistically significant. This may be because the perianal abscesses were generally more extensive and easier to observe by imaging. In addition, imaging examination had the highest sensitivity (100%) and specificity (96.08%); this may be attributed to the oxidizing properties of HP, which fills the abscess cavity with gas that creates a strong contrast with the surrounding tissue, facilitating clear observation. The clinical implementation of high-resolution HPMRI fistulography faces several potential barriers. First, its technical complexity and strict reliance on operator expertise pose challenges, particularly for less experienced imaging centers. To address them, standardized training programs should be developed, including comprehensive imaging manuals, case-based learning resources, and guidelines for interpreting dynamic fistula track relationships. Second, the higher upfront costs associated with high-resolution MRI systems may limit accessibility in resource-constrained settings. However, these costs can be offset by the improved diagnostic accuracy, reduced recurrence rates, and fewer repeat surgeries. Future multicenter studies are needed to evaluate the cost-effectiveness of HPMRI, considering both clinical and economic outcomes. Despite our promising results, this study has some limitations: (1) This study is a retrospective analysis, which may have introduced bias in the data and consequently affect the clinical efficacy of high-resolution direct HPMRI fistulography across different patient populations; (2) No further evaluation was conducted on the differences in the diagnostic value of 3D sequence scanning vs 2D scanning for anal fistula; and (3) No further comparison was conducted for the differences in the diagnostic value of HP-enhanced 3D-EAUS vs HPMRI.

Figure 2.

Male patient. A and B: Oblique axial (A) and oblique coronal (B) T1WI-mDIXON-TFE water image scans of the anal canal, indicating no obvious abnormal signals in the anal canal periphery; C and D: Oblique axial (C) and oblique coronal (D) T1WI-mDIXON-TFE water image high-resolution magnetic resonance imaging (HPMRI) fistulography scans of the anal canal showing a “double-track” high signal (low and high signal in the center and edges, respectively) between the sphincters, and an internal opening at seven o’clock in the lithotomy position (arrowheads); E: Sagittal 3D-T2WI-FSE high-resolution direct HPMRI fistulography, indicating a high intersphincteric fistula track; F: Oblique axial 3D-T2WI-FSE scan of the anal canal showing a small patch of slightly increased signal in the lithotomy position at approximately seven o’clock.

CONCLUSION

In this study, we present a novel high-resolution direct HPMRI fistulography technique that has proven effective in elucidating the relationship between the fistula track and sphincters. The proposed technique also allows for the visualization of internal openings, branches, and communication between different branches, thus allowing for precise surgical planning and more effective surgical treatment. We believe that high-resolution direct HPMRI fistulography has substantial potential for evaluating fistula-in-ano and should be applied in clinical settings to develop experience.

Data sharing statement

The data that support the findings of this study are available from the corresponding author upon reasonable request at 56469225@qq.com.