Earlier Anti-Tumor Necrosis Factor Therapy of Crohn’s Disease Correlates with Slower Progression of Bowel Damage

Abstract

Introduction

Crohn’s disease (CD) follows a relapsing and remitting course incurring cumulative bowel damage over time. The question of whether or not the timing of the initiating biologic therapy affects long-term disease progression remains unanswered. Herein, we calculated rates of change in the Lémann index—which quantifies accumulated bowel damage—as a function of the time between the disease onset and initiation of biologic therapy. We aimed to explore the impact of the earlier introduction of biologics on the rate of progression of long-term cumulative bowel damage.

Methods

Medical records of CD patients treated during 2009–2014 at The Mount Sinai Hospital were queried. Inclusion criteria were two comprehensive assessments allowing calculation of the index at t₁ and t₂: two time-points ≥ 1 year apart. Patients with biologics introduced before or within 3 months at inclusion (t₁) were defined as Bio-pre-t₁ and those who did not as Bio-post-t₁. The rate of disease progression was calculated as the change in the index per year during t₁–t₂.

Results

A total of 88 patients were studied: 58 Bio-pre-t₁ and 30 Bio-post-t₁. Among the 58 Bio-pre-t₁ cases, damage progressed in 29 (50%), regressed in 20 (34.5%), and stabilized in 9 (15.5%). Median time to initiation of biologics among patients whose index improved was nominally shorter compared to that in patients whose index progressed (8 vs. 15 years). Earlier introduction of biologics tended to correlate with the slower rate of progression (ρ = 0.241; p = 0.069).

Conclusions

Earlier introduction of biologics tended to correlate with the slower progression of bowel damage in CD, reflected by the reduced rate of Lémann index progression.

Introduction

Crohn’s disease (CD) follows a relapsing and remitting course that typically results in progressive bowel damage [1–3]. Although CD presents in several clinical patterns such as inflammatory, penetrating, and stricturing with or without perianal involvement, the disease behavior, in general, evolves as a result of cumulative bowel damage (CBD) [2, 4, 5]. Despite apparent symptom resolution, an active disease process often continues, accruing irreversible damage over time [6]. Advanced body-imaging techniques such as magnetic resonance enterography (MRE) and computed tomography enterography (CTE) allow detailed assessment of structural changes incurred as a result of the ongoing disease process (e.g., mucosal enhancement, bowel wall thickness) [7]. The Lémann index (Li) is a recently introduced objective quantitative measure of CBD based on small bowel imaging, upper endoscopy, perianal assessment, and colonoscopy [8, 9]. It has been used to identify risk factors for CBD and to assess the impact of various therapeutic strategies on the long-term outcome [10, 11]. The index has been used extensively for evaluating CBD at particular time points, but generally not with longitudinal follow-up [12].

Newer therapeutic goals for CD demand control of the disease process beyond symptomatic relief—to slow, arrest, or reverse the cumulating damage. For the use of biologics, “treat-to-target” and “top-down regimens” are increasingly favored [13, 14], in addition to the combination regimen comprising immunomodulators and biologics, proposed in the landmark SONIC study [15]. Studies have suggested advantages of biologic therapy and “treat-to-target”, which are reported to stabilize CBD [16] and promote mucosal healing [4, 15–17]. Moreover, the need to monitor CD using a dynamic model that incorporates time as a dimension in addition to the established three dimensions: age of onset, anatomic location, and disease behavior, has been proposed for over a decade [4, 17].

A pivotal question remaining is when in the course of CD is it optimal to initiate anti-TNF therapy. What is yet to be elucidated, for example, is whether or not the timing of starting biologics, tabulated on a continuous scale, affects the rate of progression of CBD over a measured time interval. In this study, using data from a tertiary IBD care center, we conducted a longitudinal study exploring the association between the timing of introducing anti-TNF therapy and the course of CD using a new quantitative metric—the rate of progression of CBD—by serial assessment of the Lémann index.

Methods

Patient Selection

This retrospective longitudinal study was conducted at a tertiary care Inflammatory Bowel Disease (IBD) Center, The Mount Sinai Hospital (MSH). The institutional database was queried to identify CD patients who underwent treatment during 2009–2014. Based on the selection criteria, electronic health records were used to extract clinical information. Inclusion criteria were: (1) age ≥ 18 years at time of inclusion; (2) a confirmed diagnosis of CD based on the ICD-9 codes, 555.0, 555.1, 555.2, and 555.9 and chart review; (3) availability of all required test results, including two MREs within 140 days of each Li assessments at least 1 year apart as recommended by the protocol [8]; and (4) first biologic treatment before or within 3 months of t₁. Biologic treatment was defined as a full induction and at the least one maintenance dose of an anti-TNF drug with or without immunomodulators. The MSH institutional review boards approved the study protocol. No patient consent was required.

Data Collection

Using the electronic health records, we collected data on patient demographics including gender, age, and smoking status at onset (non-smoker, smoker, and ex-smoker); and clinical characteristics including diagnosis, Montreal classification for age of diagnosis; disease location; and behavior; endoscopy results; surgical history; disease characteristics; disease duration; and treatment history (Table S1).

Lémann Index of Cumulative Bowel Damage (Li) Calculation Protocol

The Lémann index was calculated for each patient at two different time points (t₁ and t₂) at least a year apart. At each of the two comprehensive assessments, surgical history, endoscopy data, and MRI enterography and/or pelvic MRI results were tabulated according to the standard published protocol required for calculation of Li [8]. Upper endoscopy was assessed for patients with upper tract (esophagus, stomach, and duodenum) involvement. Colonoscopy was also used to confirm the colorectal and perianal lesion reported on the MRE. Pelvic MRI or physical exam reports were used for pelvic involvement. All investigations were completed within 140 days for each time point.

Disease manifestations including fistulae (p₁, p₂, and p₃), strictures (s₁, s₂, and s₃), ulcers (p1), abscesses (p₃), wall thickening and segmental enhancement (s₁ and s₂), mural stratification (s₂), and surgical resection (r) were graded according to the scale used to calculate computerized Lémann index. Li was calculated using Microsoft Excel-based macro published in Gastroenterology [8] (Figure-S1). The range of Li is 0–100; the higher the score, the greater the damage. The Li calculation protocol was established after multiple meetings, which included institutional radiologists and gastroenterologists. Two radiologists (MC and MW) were specifically trained to re-assess MRE for the Li scoring. Following these training sessions, interobserver concordance was assessed in 10 randomly selected patients using a two-way mixed-model method for absolute agreement. Inter-reader inter-class coefficient between the two radiologists was found to be excellent (97.7; 95% CI 91.2–99.4, p < 0.001). Two investigators together calculated the Lémann index (HP and JP). Intra-rater reliability of calculating the Lémann index was done for another set of 10 randomly selected patients and was found to be excellent (inter-class coefficient 99.4; 95% CI 97.8–99.9; p < 0.001).

Delta Lémann Index (ΔLi)

The point change in the value of Li from t₁ to t₂ was defined as ΔLi [ΔLi = Lit₂ – Lit₁]. Patients whose index increased over time (ΔLi > 0) were defined to have “damage progression”; those with unchanged Li (ΔLi = 0) as “damage arrest or stabilization”; and those with index decreased (ΔLi < 0), “damage regression.”

Rate of Progression of the Lémann Index

The rate of disease progression was defined as the change in the Li over time in years and was expressed as a continuous variable. It was calculated by dividing the point change in Li from t₁ to t₂ (ΔLi) by the time in years between t₁ and t₂, i.e., rate = ΔLi/time in years. We plotted the rates of changes in Lémann index as a function of the elapsed time between onset of CD symptoms and initiation of biologic therapy.

Definitions

Biologic therapy

A full induction, i.e., 5 mg/kg IV infliximab at week 0, 2, and 6 or subcutaneous adalimumab 160 mg at week 0 and 80 mg at week 2 or subcutaneous certolizumab 400 mg at week 0, 2, and 4, plus ≥ 1 maintenance dose according to the standard protocol of each FDA-approved biologic therapy for CD including infliximab, adalimumab, and certolizumab with or without immunomodulators.

Time of disease onset (t_o)

Each patient chart was searched for the age of the earliest onset of CD-related symptoms signs including but not limited to abdominal pain, diarrhea, growth retardation, or other systemic symptoms: e.g., weight loss, anemia, extra-intestinal manifestations including joints (arthritis), eye (uveitis, iritis, episcleritis), skin (erythema nodosum, pyoderma gangrenosum), primary sclerosing cholangitis, or signs of malabsorption or malnutrition (e.g., vitamin B12 deficiency, hypocalcaemia, hypoalbuminemia), and then consequently received confirmed diagnosis of CD.

Time of diagnosis (t_d) was established as the time when the diagnosis of CD was confirmed based on at least two of the following: diagnostic ICD-9 codes, diagnostic tests, or patient chart notes recorded by care providers.

Time of initiating biologics (t_b) was noted as the time when a patient completed a full induction and the first maintenance dose.

Time of the first Li assessment (t₁) was recorded on the day of the first MRE. All other related investigations within 140 days of the tests were used to calculate the index according to the protocol.

Time of the follow-up Li assessment (t₂) was recorded as the day of the second MRE. All other related investigations within 140 days of the tests were used to calculate the index according to the Li protocol.

Bio-pre-t₁ cohort

Patients introduced to anti-TNF before or within 3 months at inclusion (t₁).

Bio-post-t₁ cohort

Patients introduced to anti-TNF never or after 3 months of inclusion time (t₁).

Study Design and Statistical Analyses

The exposure of interest was the time interval from earliest symptomatic onset of CD to initiation of biologic therapy. The primary outcome of interest was the rate of progression of Li over time. Our analysis of interest was the correlation, if any, between (a) the time interval to initiation of anti-TNF therapy (t_o to t_b) and (b) the rate of progression of Li over the measured interval (t₁–t₂). The time interval, t₁–t₂, was determined by the dates of the two Li assessments at least 1 year apart. We studied this relationship for both the Bio-pre-t₁ and the Bio-post-t₁ cohorts. Three months was used as a cutoff for Bio-pre-t₁ group, as this time interval is reasonably proximal to the beginning of t₁ to assume stability of Li during that time period. Time longer than that would allow too much change in Li from time t₁ to have occurred spontaneously or in response to other variables.

Continuous variables were compared using Student’s t test for normally distributed variables. Nonparametric test Wilcoxon–Mann–Whitney (for two independent samples) or Kruskal–Wallis (three or more samples) was used for non-normally distributed variables. Categorical variables were compared using Chi-square or Fisher’s exact test statistics. An intrinsic feature of the “rate of progression” of Li is the direction of the change in Li, over time. Therefore, we also performed univariate logistic regression to identify significant predictors of increase in Li. Next, multivariable analysis was performed using a priori determined variables and predictors with p < 0.1 in the univariate method. A priori variables included gender, age, and smoking status at diagnosis, and Li at t₁. Correlation analysis was performed using Pearson for normally distributed and Spearman’s rho test for skewed data. Sensitivity analysis was performed using inter-class coefficient (ICC) for inter-reader agreement in the MRE reports and intra-investigator agreement calculation of Li. All tests were two-sided and considered significant for a p-value less than 0.05. Statistical analyses were performed using IBM SPSS Statistical Software for Mac OS version 21.0 (IBM Corp., Armonk, NY, USA).

Results

A total of 230 patient charts were queried, and 88 patients (38%) met the study inclusion criteria. Of them, 58 (66%) patients received biologics before or within 3 months of Li assessment, while 30 (34%) either did not receive biologics at all or received them later than 3 months of t₁ (Fig. 1).

Fig. 1.

CONSORT diagram depicting patient selection for the study. MRE MRI enterography; t₁ time point 1 when Li was assessed at inclusion; Bio-pre-t₁: biologics before or within 3 months of inclusion, i.e., t₁ Bio-post-t₁ no biologics or initiating biologics after 3 months of inclusion, i.e., t₁

Study Population

Patient and disease characteristics are described in Table 1. For the entire cohort, the median age of onset was 20 years: 32 years at t₁ and 35 years at t₂. Describing the age at diagnosis according to the Montréal classification, the majority of patients were A2 (51%), followed by A1 (41%) and A3 (8%). During the study period, the most advanced Montreal classification of disease behavior was B1 in 6% patients (non-stricturing and non-penetrating), B2 in 39% (stricturing only), and B3 in 56% (penetrating). Over one-third of the patients had perianal CD (p) (Table 1). At the time of diagnosis, 23% patients were active smokers, 8% ex-smokers, and 69% had never smoked tobacco.

Table 1.

Demographic, clinical, and disease characteristics

Characteristics	Total N = 88 N (%)	Bio-pre-t1 n = 58 n (%)a	Bio-post-t1 n = 30 n (%)a	p Value
Age, years, median (min–max)
Age at diagnosis	20 (2–60)	19 (2–60)	21 (2–42)	0.982
Age at t1	32 (10–72)	35 (19–72)	31(10–64)	0.315
Age at t2	35 (19–74)	38 (20–74)	34 (19–68)	0.451
Gender				0.472
Female	43 (48.9)	29 (50.0)	14 (46.7)
Male	45 (51.1)	29 (50.0)	16 (53.3)
Smoking status				0.305
Never	61 (69.3)	37 (63.8)	24 (80.0)
Current	20 (22.7)	15 (25.99)	5 (16.7)
Former	7 (8.0)	6 (10.3)	1 (3.3)
Montréal classification: age of onset				0.836
A1 [≤ 17 years]	36 (40.9)	25 (43.2)	11 (36.7)
A2 [18–40 years]	45 (51.1)	28 (48.3)	17 (56.7)
A3 [> 40 years]	7 (8.0)	5 (8.6)	2 (6.7)
Montréal classification: behaviorb				0.246
B1 [inflammatory]	5 (5.7)	5 (8.6)	0
B2 [stricturing]	34 (38.6)	23 (39.7)	11 (36.7)
B3 [penetrating]	49 (55.7)	30 (51.7)	19 (63.3)
P [perianal/anal]	21 (35.2)	19 (32.8)	12 (40.0)	0.500
Montréal classification: locationb				0.433
L1 [ileal]	26 (29.5)	15 (25.9)	11 (36.7)
L2 [colonic]	10 (11.4)	6 (10.3)	4 (13.3)
L3 [ileocolonic]	52 (59.1)	37 (63.8)	15 (50.0)
L4 [upper GI tract]	14 (15.9)	12 (20.7)	2 (6.7)	0.077
Disease duration at t1				0.952
[< 2 years]	13 (13.8)	8 (13.8)	5 (16.7)
[3–10 years]	28 (31.8)	19 (32.8)	9 (30.0)
[11–20 years]	22 (15.0)	14 (24.1)	8 (26.7)
[21–30 years]	15 (17.0)	11 (19.0)	4 (13.3)
[> 30 years]	10 (11.4)	6 (10.3)	4 (13.3)
Duration between t1–t2, years	2.1 (1–6.6)	2.0 (1–5.5)	2.3 (1–6.6)	0.829
Disease duration at t2, years	16.0 (1–49)	16.0 (3–49)	14.0 (1–41)	0.498
Pre-t1 medications
Immunomodulators	43 (48.9)	27 (46.6)	16 (53.3)	0.546
Systemic corticosteroids	39 (44.3)	25 (43.1)	14 (46.7)	0.750
Post-t1 medications
Immunomodulators	44 (50.0)	30 (51.7)	14 (46.7)	0.653
Systemic corticosteroids	21 (23.9)	15 (25.9)	6 (20.0)	0.541
Pre-t1 surgery
No	33 (37.5)	19 (32.6)	14 (46.7)	0.201
Yes	55 (62.5)	39 (67.2)	16 (53.3)
Upper GI tract	2 (2.3)	2 (3.5)	0	0.546
Small bowel/ileocolic	35 (39.8)	23 (39.7)	12 (40.0)	0.826
Colo-rectal	18 (20.5)	14 (24.1)	4 (13.3)	0.366
Post-t1 surgery
None	68 (77.3)	46(79.3)	22 (73.3)	0.526
Yes	20 (22.7)	12 (20.7)	8 (26.7)
Upper GI tract	0	0	0	NA
Small bowel/ileocolic	11 (12.5)	5 (8.6)	6 (20.0)
Colorectal	9 (10.2)	7 (12.1)	2 (6.7)	0.712

Continuous variables are presented as median (range) and categorical as n (%)

t₁–t₂: Time duration between the two assessments, onset–t₁: time from onset to the first assessment

^aTotal percentage may not add up to 100 because of rounding the decimals

^bNumbers may not add up to 100, as the disease presentation classification is not mutually exclusive for behavior and location

At the baseline Li assessment (t₁), the median disease duration was 12 years (range 0–46), with 13% patients diagnosed within previous 2 years, 28% between 3 and 10 years, 22% for 11–20 years, 15% for 21–30 years, and 10% for over three decades. At the second Li assessment (t₂), median disease duration was 15 years. Half of the cohort had been treated with immunomodulators before t₁, and 44% with systemic corticosteroids. After t₁, 50% patients received immunomodulators, while only 19% received systemic corticosteroids.

Surgical Resections (Table 1)

Prior to the first Li assessment (t₁), over half of the patient cohort had already undergone at least one surgical resection. During the time between the first and the second Li assessment (t₁–t₂), 20 patients underwent surgical resection. There were no significant difference in the proportions of patients undergoing surgery, both before t₁ and after t₁, between the Bio-pre-t₁ and the Bio-post-t₁ cohorts.

Lémann Index of Cumulative Bowel Damage (Li) for the Entire Cohort

The median Li at t₁ was 11.1 (IQR 4.9–23.7), which increased to 14.2 (IQR 5.4–30.8) at t₂ (Table 2). During the time interval between t₁ and t₂, Li increased (damage progressed) in 47 (53.4%), remained unchanged (damage arrested or stabilized) in 15 (17.0%), and decreased (suggesting damage regression) in 26 patients (29.5%).

Table 2.

Lémann index and disease progression trajectories

Lémann index (Li) Parameter	Total N = 88	Bio-pre-t1 n = 58	Bio-post-t1 n = 30
At t1
Mean (SD)	15.1 (13.5)	16.06 (14.0)	13.2 (12.6)
Median (min–max)	11.1 (0–73.3)	13.5 (0.3–73.3)	10.0 (0–51.5)
At t2
Mean (SD)	19.1 (14.2)	20.1 (16.6)	17.3 (16.8)
Median (min–max)	14.2 (0–73.8)	14.5 (0–73.8)	12.5 (0–67.5)
ΔLi [Lit2–Lit1]
Mean (SD)	4.0 (8.6)	4.0 (9.2)	4.0 (7.6)
Median (min–max)	0.6 (− 13.9 to 37.8)	0.3 (− 13.9 to 37.8)	0.7 (− 9.6 to 22.6)
The rate of Li [ΔLi/time in years from t1 to t2]
Mean (SD)	2.7 (6.1)	2.5 (6.3)	3.2 (5.7)
Median (min–max)	0.2 (− 3.0 to 37.8)	0.1 (− 3.0 to 37.8)	0.3 (− 2.9 to 22.6)

Li, the Lémann index; t₁, time point one of the Lémann index assessment; t₂, time point two when the Lémann index was assessed; ΔLi change in the Li over the two time points t₁ and t₂; rate of Li: the change in Li per year for the proposed duration of study, i.e., t₁–t₂

Older age at introducing biologics tended to correlate with the higher score of Li at t₁ (r = 0.249, p = 0.059), but not at t₂ (r = 0.161, p = 0.83). Disease duration correlated with Li at both time assessments: t₁ (r = 0.442, p = 0.001) and t₂ (r = 0.426; p < 0.001). Box-plots depicting increase in the Li correlating with disease duration are presented in Figs. 2 and 3 for t₁ and t₂, respectively.

Fig. 2.

The Lémann index by disease duration at t₁; N = 88; coefficient, Kruskal–Wallis statistic = 13.79, p = 0.008)

Fig. 3.

The Lémann index by disease duration at t₂; n = 88; Kruskal–Wallis test statistic = 15.99; p = 0.003

Lémann Index by the Time of Initiating Biologics

In the 58 patients in the Bio-pre-t₁ cohort that received biologics before or within 3 months of t₁, damage progressed in 29 (50%), stabilized in 9 (15.5%), and regressed in 20 (34.5%). In the Bio-post-t₁ group, the damage progressed in 18 (60%), arrested in 6 (20%), and reversed in 6 (20%).

Delta Lémann Index (ΔLi) and Time to Initiation of Anti-TNF

The median time from disease onset to the introduction of biologics was 9.5 years (range 0–42 years). Of the total patients, 31% received biologics within 5 years of disease onset, 19.0% between 6 and 10 years, 25.9% between 11 and 20 years, and 24.1% after two decades of the disease onset.

Of the 58 patients in the Bio-pre-t₁ group, half (n = 29) experienced a decrease in Li, i.e. ΔLi ≤ 0, during the follow-up time. Median time to initiation of biologics in patients with ΔLi ≤ 0 was almost half of that in patients with ΔLi > 0 (p = 0.058; Fig. 4).

Fig. 4.

Box-plots depicting distribution of the timing of initiating biologics among patients who received the treatment before or within 3 months of t₁ (n = 58) stratified by outcome of interest: damage arrest/regression; Wilcoxon–Mann–Whitney U test, p = 0.058

Univariate logistic regression analysis showed earlier introduction of biologics was associated with damage regression or arrest (ΔLi ≤ 0) (p = 0.043; Table 3). Multivariable analysis adjusting for potential confounders including age of onset, gender, and Li at t₁ confirmed that earlier initiation of biologic therapy remained a significant predictor of bowel damage stabilization or regression (p = 0.03; Table 3).

Table 3.

Logistic regression analyses for bowel damage regression or arrest among patients introduced to anti-TNF before or within 3 months of t₁, n = 58

Characteristics	OR (95% CI)	p Value	AOR (95% CI)	p Value
Delay initiating biologics, each 10 years from onset	0.09 (0.09–0.10)	0.043	0.09 (0.09–0.10)	0.026
Montréal age, years
A1 [≤ 17]	Reference		Reference
A2 [18–40]	0.94 (0.32–2.77)	0.909	0.47 (0.12–1.79)	0.266
A3 [> 40]	4.33 (0.42–44.43)	0.217	2.42 (0.19–31.1)	0.497
Gender
Female	Reference		Reference
Male	1.52 (0.54–4.26)	0.432	1.94 (0.54–6.90)	0.308
Smoking at diagnosis
Never smoker	Reference		Reference
Active smoker	0.20 (0.04–1.03)	0.054	1.45 (0.32–6.49)	0.631
Former smoker	0.66 (0.11–4.04)	0.650	0.49 (0.05–4.77)	0.538
CD duration, years
From diagnosis to t1	0.97 (0.92–1.01)	0.965	–	–
Time from anti-TNF to t1, years	0.87 (0.73–1.03)	0.101	0.86 (0.69–1.06)	0.151
Time from t1 to t2, years	1.03 (0.98–1.09)	0.204
Lémann index at t1	1.00 (0.96–1.04)	0.868	1.04 (0.99–1.09)	0.173

t₁ Time point 1 of the Lémann index assessment, OR odds ratio, AOR adjusted odds ratio, CI confidence interval, Bio initiation of biologics

Rate of Progression of Li and Time of Initiating Anti-TNF Therapy

The overall rate of change of Li (ΔLi/t₂–t₁ years) ranged from −3 to +37.8 Li per year. On a continuous scale, earlier introduction of biologics was associated with either the decreased rate of progression or increased rate of improvement as reflected by the rate of change in Li (p = 0.069; Fig. 5). By contrast, in the Bio-post-t₁ group that either received biologics after 3 months of t₁ or never, no association was observed between the rate of change of Li and time from disease onset (t_o) to either t₁ or t₂. (p = 0.934; Fig. 6).

Fig. 5.

Correlation between the rate of progression of the Lémann index of cumulative bowel damage on the Y-axis and the timing of initiating anti-TNF Therapy on X-axis among patients who received biologics prior of within 3 months of t₁: Bio-pre-t₁, n = 58; correlation coefficient Spearman’s ρ = 0.241; p = 0.069

Fig. 6.

Correlation between the rate of progression of the Lémann index of cumulative bowel damage on the Y-axis and the timing of initiating anti-TNF Therapy on X-axis among patients who received biologics either after 3 months of t₁ or did not receive any biologics during the study period: Bio-post-t₁, n = 30; correlation coefficient Spearman’s ρ = −0.024; p = 0.934

Outlier analysis was performed as a sub-analysis. The outlier in this series with the rate of progression greater than 35 Li per year (Fig. 5) had undergone two resections culminating in short bowel syndrome and also developed perianal disease, all within 14 months of follow-up between t₁ and t₂. The association between the timing of introduction and the rate of progression became significant after excluding the outlier (correlation coefficient, ρ = 0.297; p = 0.025).

Discussion

The use of the Lémann index of cumulative bowel damage in CD has become increasingly common in recent years [10, 18–21]. Moreover, we are not the first investigators to note that Li decreases with anti-TNF therapy [16], but our study has incorporated two innovations. First, we quantified the disease trajectory as the rate of progression of Li and utilized this measure as the primary outcome of interest. Second, we correlated this outcome measure with the time from presumed disease onset to the first introduction of anti-TNF therapy, measuring this interval as a continuous variable. Using this onset-to-treatment interval as the independent variable, and rate of progression of Li as the outcome measure, we found that the earlier initiation of biologics was tended to correlate with a slower rate of progression of Li (p = 0.069; Fig. 5). The association of worsening Li with the longer time interval between the disease onset and initiation of biologics remained significant after application of a multivariable logistic regression analysis, which showed that later introduction of anti-TNF by 10 years decreased the odds of disease regression or stabilization by 91% (Table 3).

In the EXTEND trial, Rutgeerts et al. [22] reported that adalimumab induces and maintains remission. Our study furthermore indicates the impact of earlier anti-TNF treatment on the rate of CD progression. While association is not the same as causation, this observation would be consistent with the hypothesis that early introduction of anti-TNF therapy could have an ameliorative effect on structural damage over the long term. These results are specifically important in the context of the newer therapeutic goals of CD shifting from symptomatic remission to transmural healing. An experimental animal study demonstrated improved outcomes with early introduction of anti-TNF [23]. Other clinical studies have demonstrated that introducing biologics within 2 years of disease onset contributes to improved disease control, but they used outcome measure as clinical indices [24, 25]. Our findings are in support of the emerging treatment paradigm that using the “therapeutic window of opportunity” in early CD may promote healing more effectively, thus preventing further irreversible bowel damage progression culminating into the natural disease course.

In our study, the overall Li, as expected, tended to increase with longer disease duration, an observation that is similar to what was conceptualized at the inception of the Lémann index [8, 9]. Similarly, other authors have shown the association between the longer duration and the higher Lémann index [26]. What came to us as a surprise, however, was our finding that the Li decreased or remained stable in about half of the cohort, representing arrest or even regression of structural damage. Nonetheless, these findings are like many others [2, 11, 18, 27]. For example, Cosnes et al. [2] followed an inception cohort of patients over 5–10 years and found that Li increased in 57%, remained unchanged in 10%, and decreased in 33%, data not unlike our own.

When adjusted for the Li at t₁, gender, smoking status at diagnosis, and disease duration, the multivariable analyses showed no significant impact of age at disease diagnosis on disease progression. These findings differ from a recent study in Korean patients that showed a favorable effect of older age of diagnosis on disease prognosis. However, the outcome measures in this study were remission at the last follow-up and number of surgeries, rather than a quantitative indicator measuring cumulative bowel damage [28].

The retrospective design of this study carries certain limitations. The p value for our primary outcome was only borderline significant, which may be attributed to the smaller sample size. There may be a risk of confounding by indication, as treating physicians may tend to treat more severe disease more aggressively earlier than later. One might expect this confounding, however, to militate against rather than in favor of the apparent benefit of early biologic therapy. There is also a potential for referral bias, since all patients were seen at a tertiary care center. Moreover, on account of the colinearity of disease duration and time interval between the disease onset and introduction of biologic, we did not include the disease duration in our multivariable model; instead, in order to control for preexisting cumulative disease damage, our model used the Li at t₁. Finally, our analysis did not include information on the type of payer, which might have influenced decision making for the use of biologics. Nonetheless, our study reflects a long and nearly complete follow-up of confirmed cases with mostly long-standing disease starting from the time of onset.

Conclusion

This study uses the rate of progression (or regression) of cumulative bowel damage as the primary outcome measure to assess an association with the timing of initiating anti-TNF therapy in the course of Crohn’s disease. We found that earlier introduction of biologics tended to correlate with a slower rate of progression (or with stabilization or even regression of cumulative bowel damage. Besides the therapeutic implications of this finding, we suggest that the rate of progression of the Lémann index of cumulative bowel damage may be a useful outcome measure for future studies of the clinical course of Crohn’s disease.