Smoking Is Not an Independent Risk Factor for Surgery in Patients with Crohn’s Disease on Biologic Therapy

Inflammatory bowel disease (IBD) is a lifelong disease characterized by chronic inflammation of the gastrointestinal tract. It is divided into two different disease phenotypes: Crohn’s disease (CD) and ulcerative colitis (UC) [1]. Despite similarities in symptomatology, these two diseases may differ in histology, location, and extent of involvement. The pathogenesis of IBD has long been studied, and the leading hypothesis posits a multifactorial model of etiology including a complex interplay between genetic susceptibility, environmental triggers, microbiota dysbiosis, and dysregulation of the immune system [1, 2]. Several environmental elements seem to interact with the aforementioned factors, playing either a protective or a harmful role. Diet, for example, can be on either end of the spectrum [3]. A high-fiber diet has been shown to be inversely correlated with either CD or UC [3], while a diet high in fats has been linked with an increased risk of developing both CD and UC [2]. Furthermore, increased stress levels, a sedentary lifestyle, and decreased sleep have been associated with IBD [3]. Although all these different factors have similar patterns of association with both CD and UC, smoking has a distinct interaction with these two diseases. In specific, it has been associated with a detrimental effect on the development and course of CD [4, 5] compared to a possible beneficial effect on the progression of UC [4]. In fact, many studies have confirmed these associations. Upon examining the association between smoking and IBD in a group of 95 adults with CD and UC, it was shown that smokers with CD had more severe disease, hospitalizations, surgeries, as well as increased risk of recurrence post-surgery [6]. Further, smoking was shown to complicate the course of disease in CD patients with more frequent disease flare-up, higher rates of recurrence after surgery, and an increased chance of requiring a second surgery [5]. Interestingly, smoking cessation was shown to be beneficial in CD patients that also smoke [5]. In addition, it has been suggested that smokers have reduced response to anti-tumor necrosis factor (TNF) therapy and may have lower concentrations of the drug in their blood when compared to nonsmokers [7, 12] and that smokers have higher levels of anti-TNF antibodies than nonsmokers [7]. This association between smoking and clinical outcomes and response to biologic therapy has, however, been recently challenged, and several studies have shown little or no effect of smoking on disease outcomes or response to treatment in the era of biologics [13, 15]. Indeed, a large multicenter study has confirmed this notion that smoking status has a limited effect on the prevalence of perianal disease and need for surgery in patients with CD [16]. It is worth noting that in the aforementioned study, the use of biologics and immunosuppressive agents was more prevalent in smokers. However, the results of this study were not replicable when the analysis was performed on a larger population by the same research group [17]. Therefore, whether, as has been suggested, the use of biologics neutralizes the detrimental effect on the natural course of CD [15], remains unanswered, mainly because of the paucity of studies that have specifically looked at the timing and type of CD-related surgeries in relationship to tobacco use (current, former, or never) and biologic exposure. This retrospective study aims to investigate the relationship between smoking and the course of disease in CD patients who received biologic therapy at a tertiary care medical center in Lebanon.

This is a retrospective study performed at the American University of Beirut Medical Center using the electronic health records system. The study was conducted after Institutional Review Board approval. Patients included were adults known to have moderate to severe CD receiving biologic therapy for their disease. Patients who were less than 18 or did not receive any biologic therapy were excluded. Similarly, patients who were lost to follow up with insufficient data in their medical records were excluded. Demographic variables and smoking status were collected. Variables pertaining to disease characteristics as well as therapy plan and management were obtained, namely, years since diagnosis, family history of IBD, current and prior biologics, use of immunomodulators, and number of hospitalizations, infusions, and surgeries if applicable. Details concerning surgeries were reviewed, including type of surgery (abdominal, perianal, or both) and their temporal relationship (before or after) to initiation of biologics.

For descriptive and inferential statistics, SPSS for windows version 25 was used (IBM, SPSS Inc., Chicago, IL, USA). Mean age, mean years since diagnosis, and mean number of hospitalizations, transfusions, and surgeries were obtained along with their standard deviations. Recategorization of the smoking status was done to include it as a binary variable in further analysis; a group of current or previous smokers and another group of nonsmokers. For the bivariate analysis, since the analysis only included categorical variables, a χ² analysis was performed, and for cases where χ² analysis was not applicable, Fisher’s exact test was performed. Survival curves of time until surgery between nonsmokers and smokers were compared using Kaplan-Meier estimates with log-rank test over the follow-up time (follow-up start time was the year of starting biologics until last clinical follow-up). In addition, multivariable regressions models (binary logistic regression and Cox regression) were done to delineate the relation between smoking and risk of surgery and perianal surgery in our study sample.

A total of 251 patients were included, and patient characteristics are shown in Table 1. More than two-thirds were males (70.1%), with a mean age of 36.0 ± 15.0. Most were current or nonsmokers (44.2% and 43.8%, respectively), while 11.6% were former smokers. Patients receiving anti-TNF therapy (infliximab or adalimumab) constituted more than two-thirds of our population, while 25.9% of patients received ustekinumab. Around two-thirds of patients received only one biologic. A total of 97 patients (38.6%) had prior surgeries related to their disease, almost half of which had abdominal surgeries only, while the rest had perianal surgeries. When evaluating the effect of smoking on disease course, no significant difference was observed in terms of hospitalizations, transfusions, and surgeries done between the group of ever-smokers (current or previous) and the group of nonsmokers (shown in Table 2). Figure 1 shows the Kaplan-Meier survival curves for CD-related surgery in the two study cohorts. However, among patients who had surgery, a significant difference was noted between ever-smokers and nonsmokers in terms of the type of surgery; the former group was more likely to have perianal surgery done than the latter (χ² [1] = 5.471, p = 0.019). This conclusion held true for patients who had all their surgeries performed prior to starting biologics as well, but not to those who had surgeries after (χ² [1] = 7.065, p = 0.008 versus χ² [1] = 0.210, p = 0.647) (shown in Table 2).

To better understand the actual effect of smoking on surgery, multivariable regressions were done. First, the odds of having any type of surgery among the whole sample were higher in patients with longer disease duration (odds ratio [OR] = 1.048, 95% CI = 1.007, 1.089) and in patients who received more than one biologic (OR = 2.31, 95% CI = 1.163, 4.593) (shown in Table 3). Cox regression was performed on the whole sample to check factors affecting having surgery after starting biologics while accounting for time at risk (shown in Table 4). The only significant variable was receiving more than one biologic therapy (HR = 3.589, 95% CI = 1.733, 7.429). When the regression analysis was performed on patients who underwent all their surgeries prior to starting biologic therapy, smoking was observed as a significant independent predictor for perianal surgery (OR = 10.58, 95% CI = 1.950, 57.388) (shown in Table 5). This effect of smoking was, however, not observed among patients who had their surgeries after starting biologic therapy (shown in Table 6).

Smoking is a known risk factor for the development of CD and for worse clinical outcomes among patients who have the disease. Smoking has been associated with increased frequency of disease flares [1, 4], increased need for a first- or second-time surgery [1], and earlier age at which the first surgery was required [5]. Further, several studies have shown that smoking is a risk factor for recurrence of the disease after surgery and for recurrence of fistula [6]. In one study, smokers had nearly double the chance of recurrence compared to nonsmokers [7]. Therefore, in the literature, there is ample evidence regarding the deleterious effects of smoking in the general population of patients with CD. However, to the best of our knowledge, this study is the first of its kind to assess whether the impact of smoking on CD outcomes remains significant among a specific subset of patients with CD, namely, patients with moderate to severe disease who receive biologic therapy. Interestingly, our study shows that among this subset of patients, there was no significant difference in the number of Crohn’s-related hospitalizations, need for transfusions, and need for any Crohn’s-related surgery between ever-smokers (former and current smokers) and never-smokers. This suggests that the untoward effect of smoking on disease-related outcomes in CD becomes less significant once biologics are started. The only factor that showed a significant difference between the two groups was the type of surgery, with perianal surgeries being more common among ever-smokers. Smoking status was not associated with the need for perianal surgery in the whole study cohort. However, in patients who only underwent Crohn’s-related surgery prior to starting biologics, ever-smokers were significantly more likely to require perianal surgery compared to never-smokers (OR = 10.577, 95% CI = 1.950, 57.388).

Our study also found that among the population of patients with CD taking biologics, only the number of years since diagnosis, and the use of more than one biologic over the course of the disease were associated with an increased probability of undergoing any Crohn’s-related surgery. In this context, we found no significant correlation between the use of anti-TNF agents and any Crohn’s-related surgery (p value = 0.129) (shown in Table 3), indicating that there was no significant difference between the use of anti-TNF versus other biologic agents when it comes to surgery-free survival. With the limited number of head-to-head randomized controlled trials comparing different biologics in CD, meta-analyses of existing trials and real-world observational studies, such as this study, can provide some insight regarding the presence or absence of superiority between the available agents [9]. However, one factor that our study did not include in the analysis was the duration of taking the medications, which could play an important differentiating role. In fact, a recent Swedish register-based cohort study showed a significantly higher risk of surgery in patients who stopped anti-TNF therapy before 12 months when compared to those who received anti-TNF drugs for more than 12 months [10].

In addition, among the general population of our study, we found that the only factor that was significantly correlated with an increased probability of perianal surgery was requiring the use of immunomodulators, especially methotrexate. This is probably because patients with perianal disease are more likely to be treated with combination of biologic agent (especially anti-TNFs) plus an immunomodulator. These results are in-line with those of a large study that included several gastroenterology centers across Australia and New Zealand, which also found that patients who required immunosuppressive medications were significantly more likely to have perianal disease [11]. However, our study did not find a significant correlation between years since diagnosis and the risk for perianal surgery specifically, although this factor was correlated with the overall risk for any Crohn’s-related surgery. This comes in contrast to previous studies that have concluded a positive correlation between younger age at diagnosis and the incidence of perianal disease [12, 18]. The difference could be due to the fact that one of these studies included pediatric patients only [12], while our study, although did not have an age restriction, had a mean age (±SD) of participants of 36.0 (±15 years).

Our study has several limitations. These include the retrospective nature of the design and the fact that participants’ information was collected from medical records, which do not always specify the chronological sequence of events. This indicates that this study was not able to stratify the factors and outcomes chronologically, so that we could not always ascertain that the measured outcomes occurred after the risk factors that they are being crossed against in the analyses. We tried to ameliorate this limitation by classifying surgeries as done before or after the initiation of biologics. Another problem arises from the fact that several patients included in the study had medical charts dating before the introduction of the electronic health records system “EPIC” at our center, which means that a significant amount of data had to be collected from handwritten charts that are inherently difficult to interpret. Finally, smoking status is included in the medical charts as a qualitative measure rather than a quantitative one, and the exact time of smoking cessation is generally not available. Accordingly, we could not specify clear cutoffs for the classification of participants as smokers, former smokers, or nonsmokers as well as the number of cigarettes per day, duration of smoking history, and time since smoking cessation.

In conclusion, our study shows that smoking is a risk factor for perianal surgery in patients with CD who are naïve to biologic therapy. Smoking is, however, not a significant risk factor for surgery in CD patients receiving biologic therapy, and such therapy appears to overcome the detrimental effect of smoking on the course of CD. This notwithstanding, given the detrimental effects of smoking on health in general, all patients with CD should be encouraged to stop smoking independent of therapy.

This study was approved by the Institutional Review Board of the American University of Beirut (IRB ID: BIO-2021-0134). No informed consent was needed due to the retrospective nature of the study, and as approved by the Institutional Review Board of the America University of Beirut (IRB ID: BIO-2021-0134).

The authors have no conflicts of interest to declare.

No funding was needed or acquired for this study.

Ala I. Sharara: study idea, design, and supervision; interpretation of data; review of literature; drafting of the manuscript; and Guarantor of the study. Saleem M. Halablab, Christian Sadaka, Hassan Slika, Ayman Alrazim, Nour Adra, and Manar Shmais: data collection and analysis; review of literature; and drafting of the manuscript. Wissam Ghusn: study idea, literature review, and drafting and reviewing the manuscript. Ayman Alrazim: data analysis, regulatory administration, and review of the manuscript.

All data were collected and analyzed as presented in this paper. The data are available upon request from the corresponding author to whom further inquiries regarding the data may be directed.

Saleem M. Halablab and Ayman Alrazim contributed equally to this work and should both be considered first authors.