Antibiotics and azathioprine for the treatment of perianal fistulas in Crohn's disease


Dr W. Reinisch, Clinic Internal Medicine IV, AKH, Waehringer Guertel 18–20, A-1090 Vienna, Austria.


Background : Antibiotics and thiopurines have been employed in the management of fistulizing Crohn's disease, although evidence of their efficacy is rare.

Aim : To evaluate, in a prospective, open-label study, the influence of antibiotics and azathioprine on the clinical outcome of perianal fistulas in patients with Crohn's disease.

Methods : Fifty-two patients entered the study, starting with an 8-week regimen of ciprofloxacin (500–1000 mg/day) and/or metronidazole (1000–1500 mg/day). Seventeen patients had already received daily azathioprine (2–2.5 mg/kg) at enrolment, whereas in 14 patients azathioprine was initiated after 8 weeks of antibiotic treatment. Outcome was evaluated by Fistula Drainage Assessment and the Perianal Disease Activity Index at weeks 8 and 20.

Results : Overall, 26 patients (50%) responded to antibiotic treatment, with complete healing in 25% of patients at week 8. The Perianal Disease Activity Index decreased significantly from 8.4 ± 2.9 to 6.0 ± 4.0 (P < 0.0001). At week 20, the outcome was assessed in 49 patients (94%), 29 of whom (59%) had received azathioprine. Response was noted in 17 of the 49 patients (35%), with complete healing in nine patients (18%). Patients who received azathioprine were more likely to achieve a response (48%) than those without immunosuppression (15%) (P = 0.03). The Perianal Disease Activity Index was closely associated with treatment response and perianal disease activity.

Conclusion : Antibiotics are useful to induce a short-term response in perianal Crohn's disease, and may provide a bridging strategy to azathioprine, which seems to be essential for the maintenance of fistula improvement.


Crohn's disease is a chronic inflammatory disorder of the alimentary tract, often complicated by the development of intestinal strictures and/or the formation of fistulas. Fistulas characterize a sub-group of patients with so-defined penetrating Crohn's disease,1 which may be classified as internal (e.g. entero-enteric, entero-vesical or recto-vaginal) or external (entero-cutaneous or perianal) tracts. More than one-third of patients develop fistulas during their lifetime, with the majority suffering from perianal disease.2–4 Patients with colonic manifestation of Crohn's disease have the highest incidence of perianal fistulas, especially those with rectal involvement.5, 6 The treatment of fistulas is based on the severity of symptoms, the anatomic location, the number and complexity of tracts and the presence of stricturing intestinal disease. Although various medical therapies have been employed in the management of this complication, surgical intervention is frequently required.7

Standard drugs for the treatment of active Crohn's disease, such as corticosteroids and aminosalicylates, are not effective in healing perianal fistulas.8 Uncontrolled studies have suggested that antibiotics (e.g. metronidazole and ciprofloxacin) may be of short-term benefit in the closure of fistulas.9–13 Azathioprine and 6-mercaptopurine have shown favourable effects on fistula healing in Crohn's disease, although it takes 3–8 months to reach full efficacy.14, 15 Other immunosuppressive agents, such as ciclosporin and tacrolimus (FK506), have also shown considerable response rates, but their use is limited by significant toxicity or fistula recurrence when the route of administration is switched from intravenous to oral ciclosporin or when the drug is discontinued.16–19

The recent introduction of infliximab, an anti-tumour necrosis factor monoclonal antibody, has opened up a new dimension for the treatment of fistulizing Crohn's disease. Sixty-eight per cent of patients treated with three infusions (5 mg/kg) of infliximab over a 6-week period experienced significant improvement of their external fistulas, with complete fistula closure in 55%.20 However, due to issues of cost, long-term safety and efficacy, infliximab is currently used only as a second-line drug for perianal Crohn's disease.

The current therapeutic approach for the initial medical management of external fistulas is antibiotic treatment, often combined with immunomodulatory agents.8, 21, 22 However, only a few data have substantiated the use of these traditional pharmacological agents or regimens in the treatment of entero-cutaneous fistulas. We have therefore performed an open-label study in order to evaluate antibiotic and immunosuppressive therapy for the treatment of perianal fistulas in Crohn's disease.

Patients and methods


Consecutive patients with perianal Crohn's disease, who attended our out-patient clinic between July 1999 and February 2002, were prospectively enrolled in this study and followed up during treatment. The diagnosis of Crohn's disease was based on clinical, radiological, endoscopic and histopathological findings.23 Eligible patients were 18–65 years of age and had single or multiple, actively draining perianal fistulas of at least 3 months' duration.

Before study entry, physical examination, routine laboratory analyses and assessment of the severity of disease according to the Perianal Disease Activity Index (PDAI) were performed.24 The PDAI evaluates five categories affected by fistulas (Table 1).

Table 1.  Perianal Disease Activity Index (PDAI)
Categories affected by fistulasScore
  1. Items 1–3 are rated by patients. Items 4 and 5 depend on the physician's assessment. The PDAI score ranges from 0 to 20, with a higher score indicating more severe disease.

1. Discharge
  No discharge0
  Minimal mucous discharge1
  Moderate mucous or purulent discharge2
  Substantial discharge3
  Gross faecal soiling4
2. Pain/restriction of activities
  No activity restriction0
  Mild discomfort, no restriction1
  Moderate discomfort, some limitation of activities2
  Marked discomfort, marked limitation3
  Severe pain, severe limitation4
3. Restriction of sexual activity
  No restriction of sexual activity0
  Slight restriction of sexual activity1
  Moderate limitation of sexual activity2
  Marked limitation of sexual activity3
  Unable to engage in sexual activity4
4. Type of perianal disease
  No perianal disease/skin tags0
  Anal fissure or mucosal tear1
  < 3 perianal fistulas2
  ≥ 3 perianal fistulas3
  Anal sphincter ulceration or fistulas with significant undermining of skin4
5. Degree of induration
  No induration0
  Minimal induration1
  Moderate induration2
  Substantial induration3
  Gross fluctuance/abscess4

Fistulas were assessed by clinical examination and, if required, by rectal endoscopic ultrasound and/or magnetic resonance imaging. Having obtained imaging results, fistulas were classified according to the criteria of Parks et al. as superficial, inter-sphincteric, trans-sphincteric, supra-sphincteric or extra-sphincteric.25 Fistulas with many openings, those with internal openings above the dentate line, those with horseshoe tracts and those with high blind extensions were rated as complex.26 Patients with perianal disease complicated by strictures or abscess formation and those with intestinal stoma were excluded.

Medical treatment

All patients enrolled were treated with ciprofloxacin (500–1000 mg/day) and/or metronidazole (1000–1500 mg/day) for 8 weeks, which should not have been administered for more than 7 days within 12 weeks before study entry. The antibiotic dosage was adapted to the patient's body weight, with administration of higher doses for a body weight of ≥ 55 kg. Patients were allowed to receive concomitant therapy with aminosalicylates at stable doses throughout the study and corticosteroids for the control of flares (if not administered for more than 8 weeks during antibiotic treatment). Concomitant therapy with azathioprine was allowed, but had to be given for at least 12 weeks and at a stable dose for more than 6 weeks before enrolment. Patients with a history of incompatibility or allergy to ciprofloxacin and metronidazole were excluded from the study. Patients treated with ciclosporin, methotrexate or infliximab, as well as investigational immunosuppressive agents, within 6 months before study entry were excluded. After cessation of antibiotic treatment, azathioprine was recommended to all patients without immunosuppressive therapy.

Evaluation of efficacy

The primary aim of the study was to evaluate the response in patients receiving azathioprine vs. those without at week 20. Response was defined, according to the Fistula Drainage Assessment, as at least a 50% reduction from baseline in the number of draining fistulas.20 Secondary end-points were complete response (defined, according to the Fistula Drainage Assessment,20 as fistula closure or the absence of any draining fistulas despite gentle finger compression) at week 20, the number of patients with response and complete response to antibiotic treatment at week 8, leucocyte and C-reactive protein counts before and after antibiotic therapy and changes in PDAI at weeks 8 and 20 compared with baseline.


In order to evaluate the possible late onset of azathioprine efficacy and to check for sustained treatment response, fistula activity was again assessed at least 32 weeks after study entry and at the patient's previous visit to the out-patient department.

Statistical analysis

Unless otherwise stated, all results are expressed as the mean ± standard deviation. Comparison of PDAI scores, leucocyte counts and C-reactive protein levels before and during treatment was analysed by the paired exact Wilcoxon signed rank test. For the detection of differences between response rates in patients receiving different types of medication, Fisher's exact test was used. Differences were considered to be significant if P < 0.05. Multivariate logistic regression analysis was performed by SAS (SAS/STAT User's Guide 1999, Version 8, SAS Institute Inc., Cary, NC, USA) in order to assess the simultaneous effects of smoking, azathioprine administration and duration of fistulizing disease on treatment response at week 20.


Patient characteristics

Fifty-two patients were included in the study. The patient characteristics are shown in Table 2.

Table 2.  Patient characteristics
 Patients (n = 52)
  • *

     Values are expressed as median (range).

Sex (female/male)25/27
Age (years)*39 (22–63)
Smokers32 (62%)
Duration of Crohn's disease (years)*11 (2–35)
Disease location
 Ileo-colonic39 (75%)
 Colonic13 (25%)
Intestinal resections26 (50%)
Duration of fistulizing disease (years)* 5 (1–28)
Type of fistula
 Superficial 2
 Supra-sphincteric 2
 Extra-sphincteric 1
 Complex fistulas 7
 Unclassified 9
History of perianal surgery32 (62%)

Fistula classification according to the criteria of Park et al. was performed by endoscopic ultrasound in 22 patients, magnetic resonance imaging in 13 and by both methods in eight. In nine patients, no specific type of fistula was defined because the evaluation of perianal fistulas was based on clinical examination alone.

Seventeen of the 52 patients were receiving concomitant treatment with azathioprine at study entry, 13 due to chronic active disease and four due to severe post-operative endoscopic recurrence of Crohn's disease. The daily azathioprine dose was 2–2.5 mg/kg (median, 150 mg/day; range, 100–200 mg/day). Patients already receiving azathioprine before antibiotic therapy had a longer history of perianal disease (median duration, 8 years; range, 2–14 years) than those without immunosuppression (median duration, 4 years; range, 1–28 years). Baseline leucocyte counts were significantly lower in patients with concomitant azathioprine treatment [with azathioprine, (7.3 ± 2.7) × 109/L; without azathioprine, (9.9 ± 3.4) × 109/L; P < 0.01].

Seven of the 52 patients were receiving prednisolone (median dose, 50 mg/day; range, 12.5–50 mg/day) at study entry because of a flare of Crohn's disease, and 23 were receiving concomitant therapy with 5-aminosalicylates at a dose of 3 g/day (range, 1.5–4.5 g/day).

Antibiotic treatment

All patients enrolled completed the antibiotic treatment for 8 weeks. Thirty-nine patients (75%) received ciprofloxacin at a median dose of 1000 mg/day (range, 500–1000 mg/day). Four patients (8%) with a previously known lack of response or allergic reaction to ciprofloxacin were primarily treated with metronidazole (1000 mg/day; range, 1000–1500 mg/day). Nine patients (17%) started with a combination of the two drugs due to previous inefficacy of ciprofloxacin administration only.

Overall, 26 of the 52 patients (50%) responded to antibiotic treatment at week 8, with complete response occurring in 25%. PDAI scores decreased significantly from 8.4 ± 2.9 at baseline to 6.0 ± 4.0 at week 8 (P < 0.0001). Leucocyte counts [baseline, (9.1 ± 3.4) × 109/L; week 8, (8.8 ± 4.0) × 109/L; P = 0.7] and C-reactive protein levels (baseline, 2.7 ± 3.6 mg/dL; week 8, 2.2 ± 2.9 mg/dL; P = 0.3) did not change during antibiotic treatment.

In patients with concomitant prednisolone treatment (n = 7), five responded to antibiotic therapy. The response rates to antibiotic regimens are shown in Table 3.

Table 3.  Response to antibiotic treatment at week 8
Patients treated with antibiotics (n = 52)ResponseComplete responseNo response
Without azathioprine at baseline (n = 35)19 (54%)10 (29%)16 (46%)
 Ciprofloxacin (n = 26)12 (46%)7 (27%)14 (54%)
 Metronidazole (n = 2)2 (100%)00
 Ciprofloxacin + metronidazole (n = 7)5 (71%)3 (43%)2 (29%)
With azathioprine at baseline (n = 17)7 (41%)3 (18%)10 (59%)
 Ciprofloxacin (n = 13)5 (38%)2 (15%)8 (62%)
 Metronidazole (n = 2)002 (100%)
 Ciprofloxacin + metronidazole (n = 2)2 (100%)1 (50%)0

Treatment response did not differ between patients receiving antibiotic therapy only and those with concomitant azathioprine administration (P = 0.5).

The side-effects observed were diarrhoea (n = 2), nausea (n = 1), muscle pain (n = 1) and skin erythema (n = 1) in patients receiving ciprofloxacin, and metallic taste and nausea (n = 3) in patients treated with metronidazole. No adverse events occurred that required the discontinuation of antibiotic therapy.

Immunosuppressive treatment

Forty-nine of the 52 patients (94%) completed the protocol for the evaluation of the primary end-point at week 20; two patients without azathioprine and one with azathioprine did not adhere to the protocol. After 8 weeks of antibiotic treatment, azathioprine therapy was initiated in another 14 patients. All patients with concomitant azathioprine therapy at study entry continued with this drug, except for one who stopped azathioprine because of nausea and vomiting. Thus, 29 of the 49 patients (59%) had received azathioprine at a dose of 2–2.5 mg/kg (median, 150 mg/day; range, 100–200 mg/day) for at least 3 months by week 20.

Overall, 17 of the 49 patients (35%) showed response to treatment, with complete response in nine (18%). Figure 1 displays the treatment response rates for patients with or without immunosuppressive therapy at week 20. All patients with complete response at week 8 maintained fistula healing. In the one subject who discontinued azathioprine therapy after week 8, fistula activity recurred. Patients receiving azathioprine (n = 29) were more likely to respond (48%) than those who refused immunosuppressive therapy after antibiotic treatment (15%), as shown in Figure 2 (P = 0.03).

Figure 1.

Response rates for patients with or without azathioprine administration after antibiotic treatment (week 8) and at week 20. Two patients without azathioprine and one with azathioprine did not adhere to the protocol after antibiotic treatment. One subject, who experienced fistula improvement with antibiotic and immunosuppressive therapy, stopped azathioprine because of nausea and vomiting after week 8 and had recurrence of fistula activity at week 20.

Figure 2.

Response rates of fistula treatment at week 20 expressed as a percentage of patients (n = 49) within the specified groups. The percentage of patients with response encompasses those with complete response. Treatment response was significantly increased for patients with azathioprine therapy compared with those without azathioprine (P = 0.03).

Logistic regression analysis revealed that azathioprine administration (P = 0.01) and the duration of fistulizing disease (P = 0.04) were significant parameters for treatment response (Figure 3), whereas smoking had no influence (P = 0.3).

Figure 3.

Estimated probability of treatment response assessed for immunosuppressive therapy and duration of fistulizing disease. Treatment response was significantly associated with azathioprine administration (P = 0.01) and duration of perianal Crohn's disease (P = 0.04).

The PDAI was assessed in 44 of the 49 patients, and was closely associated with treatment response. PDAI scores were 7.4 ± 3.1 for non-responders, 2.8 ± 2.4 for patients with response (PDAI no response vs. response: P = 0.0001) and 0.8 ± 1.0 for patients with complete response (PDAI no response vs. complete response: P = 0.0001). Figure 4 displays the differences between the baseline PDAI scores and the scores at 8 and 20 weeks post-treatment for no response, response and complete response.

Figure 4.

Differences between Perianal Disease Activity Index (PDAI) scores for no response, response and complete response compared with baseline scores. Open bars represent differences between PDAI scores after antibiotic treatment at week 8 and baseline; shaded bars represent differences between PDAI scores at week 20 and baseline. *P ≤ 0.0001 vs. no response.

Long-term follow-up

The treatment response at week 32 (after at least 6 months of azathioprine therapy if initiated after antibiotic treatment) was assessed in 47 patients (90%). All patients with complete response at week 20 (n = 9) maintained fistula healing at week 32 and during the following 28.1 ± 9.1 months. Patients on azathioprine continued immunosuppressive treatment and, in three of the eight patients with response but not complete cessation of drainage at week 20, fistula healing was noted at week 32. For the three patients who had experienced fistula healing with ciprofloxacin treatment only, azathioprine therapy was initiated after week 32 due to chronic active Crohn's disease.


This study reports, for the first time, the clinical outcome of perianal fistulas in patients with Crohn's disease treated with antibiotics and azathioprine. Ciprofloxacin and/or metronidazole administration for 8 weeks led to fistula improvement in 50% of patients; however, long-term response, as assessed at week 20, depended on immunosuppressive therapy with azathioprine. PDAI scores were closely associated with treatment response as evaluated by the Fistula Drainage Assessment.

Perianal fistulas in Crohn's disease rarely heal without treatment, which comprises surgical and medical therapies.7, 27 Although some patients only have mild symptoms or are asymptomatic without a need for intervention, perianal Crohn's disease may cause severe morbidity, resulting in disabling and impaired quality of life. Surgery is usually limited to simple fistula tracts and drainage of abscesses due to the risk of incontinence caused by more aggressive interventions. Medical therapy represents the mainstay in the management of perianal fistulas, although clinical practice is mainly based on small uncontrolled trials or case series.

Metronidazole was the first antibiotic used in fistulizing Crohn's disease because of its antimicrobial action, particularly against anaerobic organisms. Uncontrolled studies report the closure of fistulas in 34–83% of cases during metronidazole administration, but with a high incidence of adverse events.9–11 Because of the relatively favourable side-effect profile of ciprofloxacin, clinical practice has switched to the use of this fluoroquinolone antibiotic which covers a broad spectrum of bacteria, including Gram-negative aerobic organisms. However, only anecdotal or preliminary reports of its efficacy in Crohn's disease can be found.12, 13, 28

In our study, 50% of patients treated with ciprofloxacin and/or metronidazole for 8 weeks showed improvement of fistula activity, as evaluated by the Fistula Drainage Assessment and expressed by decreased PDAI scores. Most patients were treated with ciprofloxacin only, with a response in 46%. In nine patients receiving metronidazole or a combination of the two antibiotics, the short-term response was up to 78%. Side-effects were mild to moderate and did not require the discontinuation of antibiotic therapy. However, 12 weeks after the cessation of antibiotic therapy, only three patients (15%) maintained a response without concomitant immunosuppressive treatment.

The high rate of symptomatic recurrence of perianal fistulas after the cessation of antibiotic treatment is in line with previous reports.10, 13 A longer duration of antibiotic administration is limited by enhanced side-effects. Paraesthesias develop in up to 50% of patients receiving metronidazole for more than 6 months.10 Studies on the long-term safety of ciprofloxacin administration for Crohn's disease are not available. Therefore, antibiotics may only serve as a bridging therapy, leading to a response which needs to be maintained by alternative treatments, such as immunosuppressants.

Azathioprine and 6-mercaptopurine have been reported to heal fistulas in about 54% of patients, although the studies within this meta-analysis were primarily concerned with the treatment of active inflammatory Crohn's disease rather than perianal fistulas.29 So far, a controlled trial of azathioprine or 6-mercaptopurine for the treatment of fistulizing Crohn's disease has not been carried out. As the use of these anti-metabolites is limited because of their slow onset of action, it might be reasonable to combine antibiotic and immunosuppressive therapy for fistula treatment.

In our study, patients already receiving azathioprine before the initiation of antibiotic therapy had suffered from perianal Crohn's disease for a longer period of time than those without immunosuppression. However, no difference in the response to antibiotic treatment was found between these two groups. In contrast, 12 weeks after antibiotic treatment, response was significantly associated with azathioprine administration. At week 20, 48% of patients receiving azathioprine showed a response compared with only 15% without immunosuppressive therapy. Logistic regression analysis revealed that azathioprine administration and the duration of perianal disease were independent parameters for treatment response. Repeated assessment of treatment response after 32 weeks, performed in order to consider the possible late onset of azathioprine efficacy, revealed some additional benefit of immunosuppressive therapy.

The currently used activity indices for Crohn's disease fail to adequately reflect the morbidity or impaired quality of life experienced by patients with perianal disease.30 Therefore, we recorded the PDAI which, together with the Fistula Drainage Assessment, has been suggested as the gold standard for fistula activity measurement.31 In our study, PDAI scores reflected the treatment response (as assessed by the Fistula Drainage Assessment) well and effectively differentiated between patients with response, complete response and no response. In a few patients, the PDAI scores were unexpectedly high (up to 7 points) despite decreased fistula activity as assessed by the physician, or were markedly decreased (minus 9 points) during treatment without objective improvement. In these individuals, items of subjective rating, such as restricted activities (including sexual relations), were responsible for the discrepancy between the PDAI scores and clinical findings. Therefore, other restrictions of sexual activity, such as intestinal symptoms, fatigue or partner problems, must be considered and excluded before this item is reported in the index. An awareness of a patient's psychological problems caused by perianal disease is essential to prevent misinterpretation of the clinical situation.

Economic aspects also play an increasing role in the application of new therapeutic regimens. The results of a cost–benefit model of the medical management of perianal fistulas in Crohn's disease have recently indicated that infliximab and a combination of 6-mercaptopurine and metronidazole show similar effects,32 although only a few data are available to support the latter strategy. For the first time, the results of a prospective, open-label study, reported herein, provide evidence for the recommendations and guidelines given for the treatment of perianal fistulas,8, 21, 22, 26 and may support the clinician in the management of this frequent complication of Crohn's disease. However, a randomized controlled trial comparing the efficacy of antibiotics plus azathioprine vs. azathioprine alone for the treatment of perianal fistulas is now warranted, before evaluating the weight of thiopurines against infliximab in this indication.

In conclusion, our data suggest that, once it has been decided to use medical treatment for the management of perianal fistulas in Crohn's disease, antibiotic therapy should be initiated. Rapid reduction of fistula drainage by antibiotics improves the patient's general clinical condition and may provide a bridging strategy to azathioprine treatment, which seems to be essential for the maintenance of fistula improvement.