PDF
Citation
Print
INTRODUCTION
Crohn’s disease (CD) is a chronic inflammatory bowel disease that widely affects the gastrointestinal tract with a progressive disease course for its following transition in location and behavior [1]. Perianal involvement is one of the most common manifestations in CD with different types including ulceration, fistula or abscess, stricture according to the Cardiff classification [2], while perianal fistulizing CD (fistulas/abscesses) represents a common but aggressive phenotype with severely impaired quality of life [3-5]. Management of perianal fistulizing CD encompasses medical and surgical interventions. Local surgical interventions such as drainage seton or fistulotomy with the assistance of symptomatic antibiotic and immunomodulator/biologic application are preferred options according to different fistula anatomy and response [5].
Estimates of the incidence of perianal fistulizing CD in U.S. and European cohorts with 5-year follow-up ranged from 11.6% to 17% [6]. Prolonged follow-up has also revealed that the incidence cumulates, for example, from 15.8% to 29.5% at 10 years after diagnosis to over 28.3% at 20 years [6-8]. In contrast, it is believed that perianal CD is more common in East Asia with a prevalence of up to 58.8% in newly diagnosed CD [9,10], and a recent nationwide study from Korea presented a 40.0% incidence of perianal fistulizing CD at 5 years after diagnosis [11]. This significance remained in South Asian CD patients compared with Caucasians in the United States [12], suggesting differences in inherited susceptibility and clinical manifestations. In addition, the incidence of CD in Asia, and particularly in China, is increasing along with the resulting disease burden during dramatic changes in socioeconomic status and lifestyle [9,13].
Current knowledge about the risk of perianal lesions in CD awaits improvement. Numerous factors have been reported to associate with the risk of perianal lesions by several cohorts, including male sex, younger age at CD diagnosis, different disease manifestations, and the luminal behavior of CD like colonic disease location and internal penetrating behavior [6]; however, the power of these associations was limited by small number of studies with heterogeneity. Furthermore, the detailed impact of the luminal behavior of CD, with its natural tendency toward transition, remains unknown. Research into the transitional nature of CD may be the key to improving its management and prognosis [14,15]. Previously the association of behavior progression with the risk of CD-related surgery was demonstrated in a large-scale Chinese CD cohort [16]. To go one step further, we evaluated the incidence and clinical characteristics of perianal fistulizing CD and tested its association with luminal behavior progression in this long-term follow-up Chinese population.
METHODS
1. Subject Recruitment and Follow-up
Hospitalized CD patients were consecutively enrolled from January 2000 to December 2018. Details of the definitions and methodologies used to recruit and follow-up subjects in this cohort have been published previously [16] and would be described below. CD was diagnosed by gastroenterologists based on clinical, imaging, endoscopic, and histologic evidence according to the contemporary Chinese and European consensus [17,18]. Follow-up data were collected from medical records if the patients revisited as expected. To reduce recall bias, patients were asked to revisit if the disease remained active or if they presented with CD-related complications (including perianal lesions; all definition details were provided in “Data Collection and Variable Definition”), otherwise, the relevant information was allowed to be obtained by contacting the patients or their relatives by telephone or e-mail biennially or triennially. The most recent follow-up was completed in December 2020. Patients who refused or were lost to followup were excluded; only patients with an unchanged primary diagnosis of CD and a minimum follow-up of 2 years were included in the present study (Supplementary Fig. 1).
2. Data Collection and Variable Definition
Information on demographics (age at diagnosis, sex) and exposure to environmental risk (smoking status) was collected from patients. Smoking status was defined as current (self-reported regular and ongoing smoking; all these subjects reported duration of more than 1 year), former (previously regular smoking but currently quit), and never. Clinical disease activity was determined using the Harvey-Bradshaw Index, with a score of < 5 as remission, 5–7 as mild activity, 8–16 as moderate activity, and ≥ 16 as severe activity [19]. The Montreal classification was applied as the criteria for coding of age at diagnosis (A1, ≤ 16 years; A2, 17–40 years; and A3, > 40 years), disease location (L1, small bowel; L2, colon; L3, small bowel and colon; and L4, upper gastrointestinal tract), and behavior (B1, nonstricturing/nonpenetrating; B2, stricturing; and B3, penetrating; perianal disease was classified independently) [20]. Disease location and behavior were reviewed at the end of follow-up, with disease behavior or location transition additionally classified according to different B3 or L2/3, respectively, at diagnosis and follow-up. Endoscopic activity of CD in patients without bowel surgery was determined based on evaluation and cutoffs of Simple Endoscopic Score for Crohn’s Disease [21] at the enrollment, with cutoffs applied to define activity groups as the result: inactive 0–2; mild 3–6; moderate 7–15; and severe > 16. Perianal fistulizing lesions (fistula, abscess) were diagnosed based on physical examination with confirmation by ultrasound, magnetic resonance imaging and/or endoscopy at CD diagnosis or follow-up according to the Cardiff classification [2]. Surgical treatment of drainage setons, fistulectomy and fistulotomy was documented. Refractory perianal lesion was defined as perianal symptoms exacerbating at any time after related effective therapy. In addition, extraintestinal manifestations [17] at diagnosis, medical therapy (systematic corticosteroids, immunomodulators, and biologics), severe CD-related complications (massive gastrointestinal bleeding, intestinal obstruction or perforation), and CD-related surgery (i.e., the gastrointestinal tract resection or enterostomy) during the follow-up were recorded and included in the statistical analysis.
3. Statistical Analysis
Descriptive analyses were presented as median and interquartile range (IQR) for continuous variables and proportions for categorical variables. The Pearson chi-square test or the Fisher exact test was used to compare categorical variables when appropriate. Because our focus was on the transitional behavior of the disease, the Kaplan-Meier survival curve with log-rank test was applied in the univariate analysis for incidence during follow-up only. Additionally, a multivariate Cox regression model of incidence with proportional hazard assumption was assessed and summarized as hazard ratios (HRs) with 95% confidence intervals (CIs). Other multivariate analyses for perianal characteristics were conducted in logistic regression model and presented as odds ratios (ORs) with 95% CI. All multivariable models used stepwise forward selection of potentially more significant factors with a filter of likelihood ratio 0.10 to prevent overfitting. All statistical significance was defined as P<0.05. Statistical analyses were performed using SPSS 26.0 (IBM Corp., Armonk, NY, USA).
RESULTS
1. Characteristic of CD and Perianal Fistulizing Lesions
A total of 504 CD patients with a median age of 30.0 years (IQR, 20.9–42.3 years) were included in this study (Supplementary Fig. 1), of whom 349 (69.2%) were male. Perianal fistulizing lesions occurred in 136 (27.0%) patients after a median follow-up of 102 months (IQR, 68–142 months), and data for detailed demographic and clinical characteristics were summarized with perianal fistulizing lesions (Table 1). The Montreal age classification of A3 consisted of 144 patients (28.6%). Among 321 (63.7%) patients with any colonic involvement (L2/L3) at diagnosis, the disease location of isolated colon (L2) and ileum-colon (L3) were 132 (26.2%) and 189 (37.5%), respectively. Penetrating behavior (B3) was present in 121 (24.0%) patients at diagnosis. The majority of disease activity at diagnosis was classified as moderate in 230 patients (45.6%), while 88 patients (17.5%) were classified as severe. At follow-up, a trend toward CD progression was indicated as the number of patients with colonic involvement increased to 348 (69.0%), including 256 for L3 (50.8%). Penetrating behaviors also increased to 207 of B3 (41.1%). Medical therapy was commonly applied as corticosteroid for 380 (75.4%), immunomodulator for 280 (55.6%) and biologic for 105 (28.3%) patients. Extraintestinal manifestations were reported in approximately half of the patients (n = 249, 49.4%), severe complications of CD were reported in 286 patients (56.7%), and CD-related surgery was performed in 211 patients (48.9%).
Characteristics of perianal lesions were summarized in Table 2. The number of patients with perianal fistulizing lesions at diagnosis was 95 (18.8%) and notably, an additional 41 patients (8.1%) developed perianal fistulizing lesions during the follow-up period. Among all 136 (27.0%) patients with perianal fistulizing lesions, there were 100 (19.8%) cases of fistula and 81 (16.1%) cases of perianal abscess. A total of 104 (76.5%) cases received perianal surgical treatment, including 17 (12.5%) cases of drainage seton only, 67 (49.3%) cases of fistulotomy, and 37 (27.2%) cases of fistulectomy. Refractory perianal lesions were present in 26.5% of the total perianal fistulizing lesions (36/136) and 29.8% of the cases that received perianal surgical treatment (31/104).
2. Factors Associated with Perianal Fistulizing Lesions
Differences and potential associations between perianal fistulizing lesions and CD characteristics were evaluated in univariate analyses (Table 1). Perianal fistulizing lesions were significantly associated with sex (P<0.001), the Montreal classification of age (P<0.001), location at diagnosis (P=0.001) and follow-up (P<0.001), behavior at diagnosis and follow-up (both P<0.001), disease activity (P=0.010), extraintestinal manifestation (P=0.049), biological therapy (P<0.001), severe complication of obstruction (P=0.012), and CD-related surgery (P<0.001). Additionally, all medical and surgical treatments prior to perianal fistulizing lesions were summarized and compared: 24.3% of cases (33/136, P<0.001) used corticosteroids, 14.0% (19/117, P < 0.001) used immunomodulators, 7.4% (10/136, P = 0.028) used biologics, and 5.1% (7/136, P<0.001) underwent surgery. Bi-level analysis for perianal fistulizing lesions showed significant increases in the percentages of location classification of L2/L3 (77.2% vs. 58.7% at diagnosis, P<0.001; 83.8% vs. 63.6% at follow-up, P<0.001), behavior classification of B3 (37.5% vs. 19.0% at diagnosis, P<0.001; 63.2% vs. 32.9% at follow-up, P<0.001), and disease activity of severe (26.5% vs. 14.1%, P=0.001). Perianal fistulizing lesions were significantly associated with a decrease in age classification of A3 (14.0% vs. 34.0%, P<0.001). There was no significant difference of perianal fistulizing lesions between L2 and L3 at diagnosis (P=0.244) or follow-up (P=0.208).
Endoscopic activity was assessed and compared in 266 patients without CD-related surgery, 95 of whom presented with perianal fistulizing lesions. No statistical difference of activity was identified between groups with or without perianal fistulizing lesions (P=0.316) (Supplementary Table 1).
3. Transition of CD and Its Association with Perianal Fistulizing Lesions
Transition of CD occurred in 120 (23.8%) for behavior (B transition) and 87 (17.3%) for location (L transition). The median time to transition was 26.5 months (IQR, 9.0–68.3 months) for behavior and 34.5 months (IQR, 12.4–76.6 months) for location after diagnosis. Of 231 patients initially classified as B1, only 136 patients (58.9%) remained at the end of follow-up, while 34 (14.7%) and 61 (26.4%) of them progressed to B2 and B3, respectively. In addition, 25 out of 152 patients (16.4%) progressed from B2 to B3. Thus, all CD behavior with its transitional trend could be classified into 3 categories: non-B3 (297, 58.9%), progressed to B3 (86, 17.1%), and persistent B3 (121, 24.0%). A similar classification could be applied to CD location. As perianal lesions were proven to be associated with colonic involvement, location trends could be classified as non-L2/3 (144, 28.6%), progressed to L3 (39, 7.7%), and persistent L2/3 (321, 63.7%). The rate of perianal fistulizing lesions was significantly different when compared within both transition groups (both P<0.001) (Fig. 1).
Kaplan-Meier curves were plotted for 41 perianal lesions after diagnosis. The median time to onset was 36 months (IQR, 10.0–83.5 months). The time-to-event plot showed that the cumulative percentage of perianal fistulizing CD was 2.7% at 1 year, 7.2% at 5 years, 10.9% at 10 years, and 16.6% at 20 years after diagnosis (Fig. 2). Plotting the incidence according to different CD behavior transitions revealed a significant difference for B3 at the end of follow-up (log-rank P<0.001), but not at diagnosis (log-rank P=0.472) (Fig. 3A and B). Furthermore, pairwise log-rank tests indicated significant differences for each of the 2 categories of B3 trends (P<0.001 for non-B3 with progressed to B3, P=0.012 for non-B3 with persistent B3, and P=0.036 for persistent B3 with progressed to B3) (Fig. 3C). Besides, there were significant associations of incidence for both of CD location at diagnosis and follow-up (log-rank P=0.005 and 0.013, respectively) (Supplementary Fig. 2A and B); however, pairwise log-rank tests showed significance only for non-L2/3 with persistent L2/3 (P=0.004) but not for others (Supplementary Fig. 2C).
4. Multivariate Regression Analysis for Perianal Fistulizing Lesions
A multivariate Cox regression model was constructed for the incidence of perianal fistulizing lesions using forward selection of possible significant factors discussed above, including variables describing transition of CD location and behavior. The final model identified significantly associated risk factors for each B3 trend (HR, 9.90; 95% CI, 4.60–21.33; P<0.001 for progressed to B3 and HR, 4.72; 95% CI, 1.91–11.67; P=0.001 for persistent B3), and extraintestinal manifestation (HR, 2.03; 95% CI, 1.04–3.93; P=0.037) as well. Risk was also suggested by intensified disease activity compared to remission, but pairwise differences were not statistically significant (HR 2.11, P=0.506 for mild; HR 6.29, P=0.073 for moderate; HR 5.59, P=0.104 for severe). Receiving CD-related surgery (HR, 0.08; 95% CI, 0.03–0.20; P<0.001) and immunomodulator therapy (HR, 0.42; 95%, 0.21–0.84; P=0.014) were associated with a lower risk of perianal fistulizing lesions after diagnosis (Table 3).
Multivariate analyses were also performed to determine factors associated with perianal treatment. Since surgical treatment was common for most perianal fistulizing CD in our cohort, only second-line surgical treatment of fistulotomy and fistulectomy was included in the analysis. Logistic regression of perianal fistulotomy/fistulectomy found a significant association with B3 trends of progressed to B3 (OR, 0.16; 95% CI, 0.05–0.54; P=0.003) and disease activity of mild at diagnosis (OR, 8.96; 95% CI, 1.20–66.72; P=0.032) (Table 4). For refractory perianal fistulizing lesions, multivariate regression model identified the only significant association with B3 trends of persistent B3 (OR, 3.91; 95% CI, 1.36–11.23; P=0.011), predicting a worse prognosis for perianal CD (Table 5).
DISCUSSION
In this long-term CD cohort, the association of perianal fistulizing lesions with disease behavior and its transition was analyzed. There are 3 important findings as follows. First, we found a prevalence of perianal fistulizing lesions of 27.0% at the end of follow-up in this long-term Chinese CD cohort. A large proportion of the perianal fistulizing cases occurred before the diagnosis of CD, and the new lesions comprised 30.1% of the perianal cases with a 10-year incidence of 10.9%. Second, there was a significant impact of behavior transition to penetrating behavior (B3) on the higher incidence of perianal fistulizing CD. Finally, statistics of surgical treatment and refractoriness of perianal fistulizing lesions were presented, indicating that the refractory lesion was independently associated with persistent penetrating CD luminal behavior.
The prevalence and the percentage of new lesions in our study were in parallel with the report of the HKIBDR cohort from territory-wide Hong Kong (28.8% and 27.6%, respectively), which has a similar ethnic setting and also an overwhelming prevalence of perianal abscess and fistula among all types of perianal lesions [22]. In comparison to a high prevalence of 58.8% from an earlier cohort from Guangdong, China, their patients were recruited from the most developed region of China with a relatively small sample size [10]. In another recent cohort study from Japan, 48.2% prevalence of total perianal lesions was reported, with a higher proportion of perianal fistula (28.8%) but less abscess (14.7%) [23]. Through genetic differences from country to country may also contribute to our relatively low prevalence among East Asia, this pattern of prevalence was in line with the growth of socioeconomic status, and therefore vigilance should be maintained in monitoring CD perianal lesions in China.
We analyzed the impact of CD location and behavior on perianal fistulizing lesions. Multiple previous studies reported that both types of colonic involvement including isolated colon (L2) [24,25] or extended to ileocolonic (L3) [11,23,24] could be associated with perianal fistulizing lesions, suggesting a major behavioral difference between colonic and ileal CD. However, most of the associations were univariate, and they were not reliable for each of the L2/L3 across the studies. Correspondingly, in our cohort, we described the association for both L2 and L3 with perianal fistulizing lesions, which failed the test for multivariate significance. Instead, the association with penetrating behavior (B3) was consistent between our result and that of other cohorts [11,22,24,26,27]. Further pathophysiologic studies are required to explain this prevalence, as luminal and perianal fistulizing behavior may not share a similar mechanism [27,28].
We signified the prevalence of perianal fistulizing lesions not only for B3 at follow-up but also the disease behavior transition towards B3. Even though the behavior of persistent B3 significantly increased the cumulative percentage of perianal fistulizing lesions, the behavior transition from B1/2 to B3 provided an additional significant increment on the Kaplan-Meier plot. This effect remained to be significant with a considerable HR of 6.47 (95% CI, 3.04–13.8) in multivariate Cox regression. It is widely accepted that CD has a progressive nature from the initial inflammation to subsequent intestinal stricture and/or penetration. Our previous study of this cohort has detailed the portrait of the CD behavioral transition with correlated factors [16]. Few studies have linked the progression of CD behavior to the manifestation of perianal disease activity at this time. In a French cohort study aiming at the behavioral evolution of CD, a remarkable 70% of 1,199 CD patients developed a penetrating behavior 20 years after diagnosis, while perianal fistulizing lesions accounted for more than 50% of them [26]. Another survival analysis from a New Zealand cohort proposed that perianal disease was a predictor of change in CD behavior, with a significant HR of 1 [62.29]. Our result was the first to conclude a solid perianal lesion risk for disease behavior transition towards B3, signifying it as a marker of uncontrolled CD pathogenesis with relapsing courses and wide involvement.
CD-related surgery was observed to independently reduce the risk of perianal fistulizing lesions in our multivariate Cox regression analysis. Surgery for luminal disease is considered a major event for CD patients and is usually the endpoint of many CD research studies, as the decision for surgery is mainly based on severe disease activity with likely complications; however, the discussion on postoperative presentation of perianal CD in real-world cohorts has been very limited due to high patient heterogeneity and decreasing sample size with improving CD management over time. Surgery attenuates CD manifestation and assists in inducing long-lasting remission by alternating the inflammatory response via diversion of fecal stream and excision of the affected bowel with high penetrability for microbiota [30,31]. An early study in 1,981 reported spontaneous healing of perianal lesions after resection and restoration of continuity [32]. The result from our cohort supported the protective effect of bowel surgery in perianal lesions, but further recruitment and follow-up of CD-related surgery cases would help to demonstrate the detailed effect of surgery.
Specific inflammatory bowel disease medical therapy with immunomodulators and biologics has been recommended in practice guidelines for the treatment of perianal fistulizing lesions [5,33]. Currently, there was low-quality evidence supporting the efficacy of immunomodulator by trials with limited sample size and sparse data in meta-analyses [5,34]; in contrast, more reliable evidence for the efficacy of infliximab in the induction and maintenance of perianal fistulizing CD was concluded by specific randomized controlled trials [35,36], although evidence was still not supportive for other biologics [37]. Our Cox regression model indicated a lower risk of perianal fistulizing lesions in patients receiving immunomodulators but not in those receiving biologics. The clinical application of biologics for the treatment of moderate to severe CD in China could be traced back to 2006 for infliximab as the dominant option for almost a decade before the emergence of adalimumab and other options, which was reflected in our cohort. However, primary non-response and secondary loss of response to anti-TNFα were occasional [38]; meanwhile, patients may discontinue biologics or take non-standard therapy due to poor compliance or limited personal income, as biologics were relatively expensive in early China. Another Danish nationwide real-world cohort reported a similar risk of perianal fistula development with biologic exposure, suggesting limited long-term disease-modifying capabilities of biologics [39]. Combining these observations, the detailed real-world effect of immunomodulators and biologics in perianal CD could be further explored in randomized controlled settings with expanded sample size, specified treatment regimens, and endpoints.
Perianal surgery was eventually performed one or more times in the majority of patients with perianal fistulizing in our cohort. The type of perianal surgical treatment is determined by specialists with reference to the manifestation [5,40]; therefore, fistulotomy and fistulectomy were selected for subgroup analysis in our case because they were second-line options with local involvement of greater extent. However, perianal surgery could be delayed in many instances, according to the specialists, due to active inflammation and especially proctitis, in order to achieve better healing. Patients were usually referred for surgical treatment after optimization of disease activity with medical therapy, which also introduced selection bias [41]. Furthermore, there is currently no well-established guideline for the diagnosis and surgical treatment of for perianal CD, and patients may undergo perianal surgery prior to CD diagnosis. All of these factors pose challenges for the evaluation of perianal surgery in this cohort study.
Meanwhile, refractory perianal lesions could be considered as a proxy for the severity of perianal CD, as it is in parallel with loss of response and impaired patient quality of life. Our multivariate analysis demonstrated the significant association for the behavior of persistent B3 but not progressed to B3. Studies tracking CD recurrence and repeated surgery have identified penetrating disease behavior as an important independent risk factor with a reported HR of up to 6.98 [42,43], but currently none have reached significance for outcome of perianal management between different studies. A multicenter Swedish cohort reviewed 119 patients with a final healing rate of 52% and stated that the risk of perianal recurrence was associated with longer disease duration, proctitis, and different surgical procedures [41]. Another multicenter study in Korea with a postoperative recurrence rate of 42% in 51 patients concluded that the only significant risk factor was escalation of medical therapy [44]. Combined with the fact that the current activity of CD cannot be reliably predicted by previous behavior [26], the duration of penetrating luminal behavior may be the key factor of perianal influence. It may be an indication that active management of the luminal activity and active monitoring of behavior transition may improve the prognosis of perianal lesions. Our cohort previously reported a relatively longer interval from diagnosis to behavior progression [16], therefore further follow-up for patients after transition may provide more explanation by analyzing the duration of behavioral influence.
There were limitations to our study as well. This is a single-center study. The sample size of perianal CD was limited, and the studied population may differ from the general Chinese CD patients nationwide. Only inpatients of this top referral medical center were enrolled, which introduced selection bias toward a higher percentage of moderate to severe CD patients. Data of CD endoscopic activity was only identified in a portion of patients without CD-related surgery. Currently, there was no general consensus about Endoscopic Score for Crohn’s Disease cutoffs, and endoscopic review was not available.
Our work was the first to summarize long-term follow-up data on perianal fistulizing CD and its association with luminal behavior and transition from a hospitalized Chinese cohort. These findings may shed light on the transitional nature of CD and mark the high-risk luminal behavior patients for more intensive follow-up and management to achieve a better quality of life in real-world clinical practice. Future research may focus on continuous data collection of CD patients with penetrating behavior transition and different treatment to improve the understanding of the pathogenesis and prognosis.
XML Download