Evaluation of the modified Van Assche index for assessing response to anti-TNF therapy with MRI in perianal fistulizing Crohn's disease

Journal Pre-proof Evaluation of the modified Van Assche index for assessing response to anti-TNF therapy with MRI in perianal fistulizing Crohn's disease

Kyra L. van Rijn, Corine A. Lansdorp, Jeroen A.W. Tielbeek, C. Yung Nio, Christianne J. Buskens, Geert R.A.M. D'Haens, Mark Löwenberg, Jaap Stoker PII:

S0899-7071(19)30182-2

DOI:

https://doi.org/10.1016/j.clinimag.2019.10.007

Reference:

JCT 8744

To appear in:

Clinical Imaging

Received date:

4 July 2019

Revised date:

4 September 2019

Accepted date:

3 October 2019

Please cite this article as: K.L. van Rijn, C.A. Lansdorp, J.A.W. Tielbeek, et al., Evaluation of the modified Van Assche index for assessing response to anti-TNF therapy with MRI in perianal fistulizing Crohn's disease, Clinical Imaging(2019), https://doi.org/10.1016/ j.clinimag.2019.10.007

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2019 Published by Elsevier.

Journal Pre-proof

1

Evaluation of the modified Van Assche index for assessing response to anti-TNF therapy with MRI in perianal fistulizing Crohn’s disease Kyra L van Rijn MD1, Corine A Lansdorp MD2, Jeroen AW Tielbeek MD PhD1, C Yung Nio MD1, Christianne J Buskens MD PhD3, Geert RAM D’Haens MD PhD4, Mark Löwenberg MD PhD4, Jaap Stoker MD PhD1

Amsterdam UMC, location AMC/University of Amsterdam, Department of Radiology and

of

1

Amsterdam UMC, location AMC/University of Amsterdam, Department of Anesthesiology,

Meibergdreef 9, Amsterdam, The Netherlands

Meibergdreef 9, The Netherlands

Amsterdam UMC, location AMC/University of Amsterdam, Department of

na

4

re

Amsterdam UMC, location AMC/University of Amsterdam, Department of Surgery,

lP

3

-p

2

ro

Nuclear Medicine, Meibergdreef 9, Amsterdam, The Netherlands

Jo ur

Gastroenterology and Hepatology, Meibergdreef 9, Amsterdam, The Netherlands

Corresponding author K.L. van Rijn, MD

Amsterdam UMC, location AMC/University of Amsterdam Department of Radiology and Nuclear Medicine Meibergdreef 9 1105 AZ Amsterdam The Netherlands Tel: 020-5663657 E-mail: [email protected]

Journal Pre-proof

2

Disclosures: Kyra L van Rijn has nothing to disclose. Corine A Lansdorp has nothing to disclose. Jeroen AW Tielbeek has nothing to disclose. C Yung Nio has nothing to disclose. Christianne J Buskens has nothing to disclose. Geert RAM D’Haens has served as advisor for Abbvie, Ablynx, Allergan, Amakem, Amgen, AM Pharma, Arena Pharmaceuticals, AstraZeneca, Avaxia, Biogen, Bristol Meiers Squibb, Boerhinger Ingelheim, Celgene/Receptos, Celltrion, Cosmo, Covidien/Medtronics, Ferring, DrFALK Pharma, Eli Lilly,

of

Engene, Galapagos, Genentech/Roche, Gilead, Glaxo Smith Kline, Hospira/Pfizer, Immunic,

ro

Johnson and Johnson, Lycera, Medimetrics, Millenium/Takeda, Mitsubishi Pharma, Merck Sharp Dome, Mundipharma, Nextbiotics, Novonordisk, Otsuka, Pfizer/Hospira, Photopill,

-p

Prometheus laboratories/Nestle, Progenity, Protagonist, Robarts Clinical Trials, Salix,

re

Samsung Bioepis, Sandoz, Seres/Nestle, Setpoint, Shire, Teva, Tigenix, Tillotts, Topivert,

lP

Versant and Vifor; received speaker fees from Abbvie, Biogen, Ferring, Johnson and Johnson,

na

Merck Sharp Dome, Mundipharma, Norgine, Pfizer, Samsung Bioepis, Shire, Millenium/Takeda, Tillotts and Vifor. Mark Löwenberg has served as speaker and/or

Jo ur

principal investigator for: Abbvie, Celgene, Covidien, Dr. Falk, Ferring Pharmaceuticals, Gilead, GlaxoSmithKline, Janssen-Cilag, Merck Sharp & Dohme, Pfizer, Protagonist therapeutics, Receptos, Takeda, Tillotts, Tramedico. He has received research grants from AbbVie, Merck Sharp & Dohme, Achmea healthcare and ZonMW. Jaap Stoker is a research consultant for Robarts Clinical Trials and has a research agreement with Takeda.

Journal Pre-proof

3

ABSTRACT Background Structured evaluation of magnetic resonance imaging (MRI) is important to guide clinical decisions of perianal fistulas in Crohn’s disease (CD) patients. Purpose To evaluate the, recently developed, modified Van Assche index to assess clinical responses

of

to anti-tumor necrosis factor (TNF) therapy in patients with perianal fistulizing CD.

ro

Methods

A search of medical records identified patients with fistulizing perianal CD who underwent

-p

baseline and follow-up MRI while receiving anti-TNF treatment. Patients were divided into

re

clinical responders and non-responders based on physician’s assessment. MRI-scans were

lP

scored using the original and modified Van Assche index and scores between baseline and

Results

na

follow-up were compared within clinical responders and non-responders.

Jo ur

Thirty cases were included (48% female, median age 27 years). Clinical responders (n=16) had a median modified Van Assche score of 9.6 (IQR5.8-12.7) at baseline and 5.8 (IQR3.58.5) at follow-up (p=0.008). In non-responders (n=14), corresponding scores were 7.7 (IQR5.8-13.5) and 8.2 (IQR5.8-11.5) (p=0.624). In clinical responders, 6/16 showed no drop in modified Van Assche score at follow-up. Scores obtained with the original Van Assche index dropped between baseline and follow-up in clinical responders (13.0 vs. 9.6, p=0.011), whereas no decrease was observed in non-responders (11.5 vs. 11.5, p=0.324). Conclusions While the modified Van Assche index overall decreases significantly in patients with perianal fistulas responding to anti-TNF treatment, one third of responders had unaltered scores at

Journal Pre-proof

4

follow-up. Also, outcomes were comparable to the original Van Assche index. Further optimization of the modified Van Assche index is needed before application in larger studies.

Jo ur

na

lP

re

-p

ro

of

Key words: Anti-TNF therapy; Crohn disease; Magnetic Resonance Imaging; Perianal fistula

Journal Pre-proof

5

1. INTRODUCTION Perianal fistulas are seen in up to one-third of Crohn’s disease (CD) patients.[1,2] Symptoms range from painless fistula discharge to abscess formation and pelvic sepsis, and most patients require medical and/or surgical interventions at some point during their disease course.[3] The most powerful therapeutic agents to treat perianal fistulas are tumor necrosis factor (TNF) antagonists. Most data has been obtained with infliximab and adalimumab, with

of

durable fistula responses in up to 43% of cases.[4] Evaluation of treatment responses is

ro

crucial in order to guide clinical decisions and as an objective endpoint in clinical trials.[3,5]

-p

Magnetic Resonance Imaging (MRI) is an important tool to evaluate perianal fistulas in CD

re

patients and it represents the recommended imaging technique by the European Crohn’s and Colitis Organization and European Society of Gastrointestinal and Abdominal

lP

Radiology.[5,6] In order to evaluate perianal CD in a structured way using MRI, Van Assche et

na

al. developed an index in 2003 (from here on referred to as the original Van Assche index).[7] Previous studies showed that the index is able to capture imaging responses in

Jo ur

clinical responders to medical therapy, but it has several limitations, such as insensitivity in patients with reduced fistula caliber at follow-up MRI.[7–12] Recently, Samaan et al. modified this index to increase objective reflection of clinical responses and in order to improve interrater and intra-rater reliability (Table 1 shows the items of both indices). They evaluated the items of the original Van Assche index, as well as new items that, by expert consensus, were found relevant to evaluate perianal fistula activity.[13] Items of the original index, modified items from the original items and new items were evaluated by four radiologists. After this, nine items with the best intra- and interrater reliability were selected. These were evaluated using a mixed effects model to select the items that best represent disease severity. A total of 5 items was selected which

Journal Pre-proof

6

created the modified Van Assche index. The modified Van Assche index has the potential to serve as a more reliable and consistent MRI activity index as compared to the original Van Assche index. However, the modified Van Assche index has only been evaluated for its operating characteristics and not for assessing responses to therapy in a clinical setting yet. The aim of the present study was to retrospectively evaluate the performance of the

Jo ur

na

lP

re

-p

ro

of

modified Van Assche index in measuring response to anti-TNF therapy in a clinical setting.

Journal Pre-proof

7

2. MATERIAL AND METHODS 2.1 Patient selection This retrospective study was performed at a single tertiary IBD referral center. The institutional review board approved the study and waived informed consent. A search in the database of electronic patient records of *BLIND*, was performed. Eligible patients with perianal CD receiving anti-TNF therapy that underwent at least one pelvic MRI examination

of

were identified between January 1st 2008 and July 1st 2018. This period was chosen for the

ro

availability of electronic medical data and consistency of the MRI protocol (e.g. the use of

-p

contrast-enhancement).

re

Only patients with a confirmed diagnosis of perianal CD[14] that started anti-TNF treatment (infliximab or adalimumab) and had a baseline and follow-up MRI were included. If patients

lP

were already using an anti-TNF agent, but the dose was increased or treatment interval

na

decreased, they were also deemed eligible. Patients in which the interval between the first MRI and the start/intensification of anti-TNF treatment was longer than 12 months were

Jo ur

excluded, as well as patients in whom the follow-up MRI was performed more than 18 months later. Patients were excluded when MRI examinations missed sequences that were necessary to evaluate items of the modified Van Assche index (e.g. post-contrast sequences), if clinical response could not be assessed from medical charts or in case no medical notes were available at the time of the follow-up MRI scan (maximum 1 month before or after) and if patients had undergone CD-related surgery between MRI examinations (except for fistula drainage and/or seton placement, which was allowed). 2.2 MR Imaging MRI’s were performed on 1.5T (Magnetom Avanto, Siemens Healthcare, Erlangen, Germany) or 3.0T (Intera, Philips, Best, The Netherlands) systems according to the local fistula protocol

Journal Pre-proof

8

consisting of T2-weighted Turbo Spin Echo sequences in coronal, sagittal and axial planes, a T2-weighted sequence wit fat suppression in the axial plane and a T1-weighted fatsuppressed sequence in the axial plane after administration of a gadolinium-based contrast agent. Coronal and axial planes were angled perpendicular and rectangular, respectively, to the anal canal. No oral and/or rectal contrast preparation was used. Patients received antiperistaltic medication (butylscopolamine bromide, Buscopan; Boehringer-Ingelheim,

of

Ingelheim, Germany) to prevent artefacts caused by bowel motion.

ro

2.3 Data collection 2.3.1. Clinical parameters

-p

The following baseline characteristics were retrieved from medical records: age, sex, age at

re

diagnosis, smoking status, disease location, prior medical treatment, prior CD-related

lP

surgery and presence of a defunctioning ostomy. Furthermore, treatment notes regarding

follow-up were collected.

Jo ur

2.3.2. Clinical response

na

type of anti-TNF therapy, concomitant medication use and CD-related surgery during MRI

Clinical responses to anti-TNF therapy were determined using medical records at the time of the follow-up MRI. Subjects were divided in clinical responders and non-responders to antiTNF therapy. Response was defined as absent or reduced fistula drainage[15] and/or a reduction in external openings. Non-response was defined as unchanged fistula drainage[15] and/or number of external fistula openings, discontinuation of anti-TNF treatment due to inefficacy and/or need for CD-related surgical interventions. In clinical responders, maintenance of response was assessed 12 months after the follow-up MRI using these same criteria. Clinical response to anti-TNF therapy was assessed by two independent research fellows [BLIND and BLIND] and all clinical responses were evaluated by a gastroenterologist

Journal Pre-proof

9

who is specialized in inflammatory bowel disease [BLIND]; they were all blinded to the MRI outcomes. 2.3.3. MRI assessment The pre- and post-treatment MRI’s were scored by two independent, experienced abdominal radiologists [BLIND and BLIND].[16] A third experienced abdominal radiologist [BLIND] reassessed the items that the first two observers disagreed on. The third radiologist

of

determined the final score of these discrepant reads by choosing between the two scores of

ro

the previous radiologists. Items without disagreement between the first two observers were not evaluated by the third radiologist. The three radiologists were blinded for clinical

-p

information and all MRI exams were reviewed in a random order with no information

re

regarding the timing (pre- or post-treatment) of the MRI. The third radiologist was blinded

lP

for which radiologist proposed what score.

na

All nine items that were used in the development of the modified Van Assche index were assessed[13]: ‘number of fistula tracts’, ‘location’, ‘extension’, ‘hyperintensity on T2-

Jo ur

weighted images’, ‘rectal wall involvement’, ‘presence of a recto/anovaginal tract’, ‘inflammatory mass’, ‘hyperintensity on post-contrast T1-weighted images’, ‘dominant feature of primary tract and extension’. For scoring of rectal wall involvement, the options ‘thickened’ (1) and ‘increased signal intensity’ (2) were merged into ‘thickened and/or increased signal intensity’ (2) since the local MRI protocol did not include rectal distension and the scoring of only thickened rectal wall was thought to be unreliable. Table 1 shows the items of the original and the modified Van Assche index. After assessment of the individual items, the total modified Van Assche index was calculated as proposed by Samaan et al.. The items ‘extension’, ‘collections’ and ‘rectal wall involvement’ of the original Van Assche index were derived from the modified scoring items after which the original Van Assche index was

Journal Pre-proof

10

calculated. Additionally, the quality of the scan was scored as: ‘poor (non-diagnostic)’, ‘adequate’ or ‘good’ and other evident findings were noted. 2.4 Statistical analysis Statistical analyses were performed using IBM SPSS Statistics for Windows version 25.0 (IBM Corp., Armonk, N.Y. USA) and RStudio (RStudio Inc. Boston, MA, USA). The level of significance was determined at P<0.05. Descriptive statistics were used to present the

of

baseline and treatment characteristics. The MRI scores of the separate items were

ro

presented as numbers and frequencies. Continuous variables were reported as medians with their interquartile range (IQR). The changes in modified and original Van Assche index scores

-p

were compared within clinical responders and non-responders using the non-parametric

re

Wilcoxon signed-rank test. Interobserver variability was reflected using weighted Kappa

lP

statistics for ordinal variables and intraclass correlation coefficients (ICC) with a two-way

na

random model for continuous variables. Kappa and ICC values were interpreted according to Landis and Koch[17], where <0.0 reflects ‘poor’, 0.0-0.20 ‘slight’, 0.41-0.6 ‘moderate’, 0.61-

Jo ur

0.8 ‘substantial’ and >0.81 ‘almost perfect’ agreement.

Journal Pre-proof

11

3. RESULTS The disposition of patients is shown in Figure 1. The electronic database search yielded 116 initial cases of perianal CD patients who received treatment with infliximab or adalimumab and had at least two pelvic MRI’s. Out of 116 patients, 25 patients were eligible for inclusion, four of these patients had more than two pelvic MRI’s with different treatment events between MRI’s (i.e. switch to another anti-TNF agent or dose intensification). These patients

of

were handled as separate cases (three patients accounted for two cases each and one

ro

patient accounted for three cases), this resulted in a total of 30 cases.

-p

3.1 Study population

re

Patient characteristics and demographics are summarized in table 2. The patient cohort consisted of 12 females (48%) and the median age was 27 years (IQR 20-37 years). Seven out

lP

of 25 patients (28%) were active smokers. The median age at diagnosis was 23 years (IQR 16-

na

30 years). The bowel disease location, in addition to the perianal disease, was in the ileum in 2/25 patients (8%), 7/25 (28%) had colonic disease, 14/25 (56%) had ileocolonic CD and in

tract.

Jo ur

2/25 (8%) patients the disease was localized to the upper gastrointestinal and ileocolonic

Since one patient can represent multiple cases, with changed treatment courses over time, prior treatment is described for the 30 cases instead of 25 patients. Fifteen cases were previously treated with anti-TNF (50%), 6 with methotrexate (20%), 22 with thiopurines (73%) and 12 with corticosteroids (40%). In 5 cases, a diverting ostomy was present at the time of the MRI (17%). 3.2 Treatment responses Sixteen cases were clinical responders and 14 were classified as non-responders to anti-TNF therapy. The anti-TNF treatment between MRI scans consisted of starting infliximab in 21

Journal Pre-proof

12

cases (70%), starting adalimumab in three (10%), intensification of infliximab in three (10%) and intensification of adalimumab in three cases (10%). Thirteen cases received no concomitant medication during anti-TNF therapy, six received co-treatment with methotrexate (20%), 11 with thiopurines (37%) and three received co-treatment with corticosteroids (10%). Fourteen cases underwent drainage and/or seton placement and one underwent an ostomy in between two MRI’s, of which eight in the response group and six in

of

the non-response group.

ro

3.3. MRI findings

In total, 56 pelvic MRI scans were analyzed. Four MRI scans functioned as both pre- and

-p

post-therapy MRI for separate cases. The quality was classified as good for 46 MRI scans and

re

adequate for 10 scans, none were classified as poor image quality. The median interval

lP

between baseline MRI and treatment was 2.5 months (IQR 1-4 months) and 8 months

3.3.1. Anatomical features

na

between treatment and follow-up MRI (IQR 4-12 months).

Jo ur

Using definitions according to Samaan et al.[13], the clinical responder group showed five single, unbranched fistula tracts (31%), four single, branched fistula tracts (25%) and six multiple fistula tracts (38%), and in one case no fistula tract was observed (6%). In the clinical non-responder group, five single, unbranched fistula tracts (36%), three single, branched fistula tracts (21%) and six multiple fistula tracts (43%) were seen. In the responder group seven tracts (44%) were localized intersphincteric and nine (66%) transsphincteric. In the non-responder group this was nine (64%) and five (36%), respectively. A rectovaginal tract was present in one responder and in one non-responder (6% and 7%). 3.3.2. Modified Van Assche index and clinical response

Journal Pre-proof

13

Frequencies of scoring items in the responder group and non-responder group are shown in table 3. The median modified Van Assche index in responders was 9.6 (IQR 5.8-12.7) at baseline and 5.8 (IQR 3.5-8.5) at follow-up. This absolute decrease of 3.8 points was statistically significant (p=0.008). In the non-responder group, the median score was 7.7 (IQR 5.8-13.5) at baseline and 8.2 (IQR 5.8-11.5) at follow-up (p=0.624). Ten out of 16 clinical responders (62%) showed a decreased modified Van Assche index and 6/16 (38%) an equal

of

or increased score at follow-up (figure 2). In the non-responder group, this was 6/14 (43%)

ro

and 8/14 (57%), respectively. Figures 3 and 4 show MRI examples of clinical responders with and without a decreased modified Van Assche score at follow-up.

-p

3.3.3. Modified Van Assche and maintained response

re

Of the sixteen responders, six (38%) showed a clinically durable response 12 months after

lP

the MRI and ten patients (63%) had a relapse. The median modified Van Assche score before

na

treatment was 5.4 (IQR 3.5-10.7) for patients with a durable response and 10.9 (IQR 8.813.5) who relapsed. At the follow-up MRI, corresponding scores were 1.75 (IQR 0.0-5.8) and

Jo ur

7.1 (IQR 5.0-10.3), respectively.

3.3.4. Original versus modified Van Assche index The median original Van Assche index score was 13.0 (IQR 10.0-15.8) at baseline and 9.5 (IQR 5.3-13.0) at follow-up in the responder group (p=0.011). Corresponding scores were 11.5 (IQR 10.8-16.5) at baseline and 11.5 (IQR 9.8-15.5) at follow-up in the non-responder group (p=0.325). 3.4. Interobserver agreement The interobserver variability for 56 MRI scans is shown in table 4. For the modified Van Assche index, the items ‘extension’, ‘T2-weighted signal intensity’ and ‘inflammatory mass’ showed substantial agreement. For the item ‘rectal wall involvement’ the agreement was

Journal Pre-proof

14

moderate and ‘predominant feature’ had fair interrater agreement. The total modified Van Assche index had a substantial interrater reliability. Figure 5 shows three examples of interobserver differences. The original Van Assche index showed moderated agreement for the items ‘location’, ‘extension’, ‘collections’ and ‘rectal wall involvement’. Substantial agreement was seen for ‘T2-weighted signal intensity’ and almost perfect agreement for ‘number of tracts’.

of

The total original Van Assche index showed ‘substantial’ agreement.

ro

4. DISCUSSION

-p

This is the first study that evaluates the modified Van Assche index in relation to clinical

re

outcomes after anti-TNF treatment. We showed that scores between baseline and follow-up decreased in clinical responders, however, in one third of responders scores did not drop

lP

and we found no benefits over the original Van Assche index. Thus, further optimization of

na

the score is necessary before clinical implementation. Evaluation of any MRI based score is challenging because no gold standard exists to

Jo ur

determine if a fistula has responded or healed. Previous studies that evaluated the use of the original Van Assche index used different definitions and strategies to determine fistula response, remission and healing.[7,9] In this study, we used a combination of factors to determine clinical response, including fistula drainage, presence of external openings and adjustment of treatment regimens. The modified Van Assche score decreased significantly in patients with clinical responses. While this suggests that the modified Van Assche index might be an objective tool to measure treatment responses in daily clinical practice and for clinical trials, in a large proportion of patients the change in score did not match the clinical response.

Journal Pre-proof

15

In approximately one third of responders the modified Van Assche score did not drop at the follow-up MRI and in 43% of the non-responders, the score did drop at follow-up. This mismatch can partly be explained by an insensitivity of the index to detect specific changes such as the extent of inflammatory tissue within fistula tracts, a limitation that was also seen with the original Van Assche index.[9,18] This might be improved by addition of an item that measures fistula caliber or volume. Another explanation could be that this index does not

of

accurately reflect the internal tissue healing process and lag behind clinical response, in line

ro

with previous observations.[7,9,10,18] One could argue what the additive value will be of such an MRI score, if its outcome matches clinical response (which is easier and cheaper to

-p

perform). But, on the other hand is it known that clinical indices can be discordant with

re

underlying fistula activity, indicating that more objective tools are required to monitor

lP

patients.[5,7,9,18,19] Interestingly, in our cohort the baseline and follow-up MRI scores of

na

patients relapsing were relatively high compared to patients with a durable response. This indicates that the MRI score might be able to uncover a clinically relevant disease severity

Jo ur

that is overlooked when only using clinical evaluation, and can be used to determine prognosis of durable responses. A prospective study using sequential MRI scans in a larger number of patients should confirm this hypothesis and establish clinically relevant cut-off values.

The modified Van Assche index was created to improve the original Van Assche index. Existing items were reviewed, after which some were adjusted and some were discarded. Also, several items were added that, proposedly, are more subjective to change to monitor treatment outcomes. Moreover, standardized definitions for these items were added. In the present study, we were unable to establish a clear benefit of the modified Van Assche index compared to the original Van Assche index. Both indices seemed to respond similarly: 10 out

Journal Pre-proof

16

of 16 clinical responders showed a decrease in total scores at follow-up, of which eight cases overlapped between the two indices (i.e. clinical responders that had a decreased score at follow-up). Additional items of the modified Van Assche index made it more complex compared to the original Van Assche index, resulting in a more time-consuming instrument requiring more experienced and trained radiologists. The items of the modified Van Assche index have the potential to be more sensitive to change, but careful reconsideration of the

of

definitions and items is indicated before this MRI index should be used in daily practice or

ro

for clinical trials.

Using the modified Van Assche index, we observed fair to substantial agreement

-p

between the two observers for the different items. This was slightly higher than in the study

re

by Samaan et al.[13], where it ranged from fair to moderate for these individual items.[13]

lP

For both the original and the modified Van Assche index we found an overall substantial

na

interobserver agreement (ICC 0.71 and 0.74), slightly higher than in the previous study by Samaan et al. (ICC 0.66 and 0.67)[13]. In the original study, developing the Van Assche index

Jo ur

in 2003 [7], the interobserver agreement for the total index and for most items were almost perfect with kappa values up to 1.00. However, the same radiologists that developed the score, also reviewed all the MRI-scans in this study. For our observers, the assigned definitions were sometimes unclear. For example, observers measured different sizes of a fluid collection by using different angles and by in- or excluding the inflammatory tissue rim. Furthermore, determination of hyper-intensity depended on the MRI viewing settings and the vessel that was used to compare it with. Further improvement of inter-observer agreement could be accomplished by revision of definitions, training of radiologists and simplification of scoring items by combining categories. Furthermore, we found that the item rectal wall involvement, as proposed by the creators of the modified Van Assche index

Journal Pre-proof

17

to be divided into three categories, does not correspond with clinical practice, as rectal distention commonly is not part of the protocol for fistula MRI’s. For future studies, we suggest to divide this item into two categories: i.e. ‘normal’ and ‘thickened and/or increased signal intensity’. There are several limitations of our study. First, it is a relatively small sample size which limits the conclusions that can be drawn from the study. Second, due to the

of

retrospective nature, assessments of clinical responses are not based on structured patient

ro

reported outcome measures combined with biochemical inflammatory parameters at predefined time points. Third, fourteen patients (six non-responders and eight responders)

-p

had a seton placed between the MRI scans. These patients were not excluded because the

re

definition of response was not solely based on closure of perianal fistula tracts and the aim

lP

of our study was not to evaluate efficacy of anti-TNF therapy, but to evaluate the modified

na

Van Assche index in a clinical setting. However, it would be interesting to exclude patients with setons in future studies in order to investigate response to anti-TNF therapy based on

Jo ur

closure of the external opening. Fourth, clinical response was divided into two categories: non-responders and responders. Subdividing responders into partial and complete responders might give more insight into the true clinical value of the modified Van Assche index. However, given the retrospective nature of this study and the relatively small sample size, a subdivision as such was deemed inappropriate. Fifth, only two MRI scans (at baseline and follow-up) were assessed for each case, and the time-window between the first and follow-up MRI scan varied considerably (IQR 4-12 months). Future studies should therefore include sequential MRI scans for individual patients using longer follow-up periods, to gain knowledge on optimal timing of MRI scans.

Journal Pre-proof

18

In conclusion, this study evaluated the use of the modified Van Assche index for the assessment of treatment responses in perianal fistulas. Overall, there was a significant decrease in total scores in clinical responders to anti-TNF therapy, but in one third of responders scores did not drop. Also, no clear added value was found compared to the original Van Assche index. An advantage might lie in the prediction of durable responses, but further optimization of the modified Van Assche index should be realized before application

of

in a prospective study using sequential MRI scans and consideration of implementation in

-p

ro

daily practice and clinical trials.

re

5. Acknowledgements

lP

Funding: This research did not receive any specific grant from funding agencies in the

Jo ur

na

public, commercial, or not-for-profit sectors.

Journal Pre-proof

19

6. References [1]

Ng SC, Shi HY, Hamidi N, Underwood FE, Tang W, Benchimol EI, et al. Worldwide incidence and prevalence of inflammatory bowel disease in the 21st century: a systematic review of population-based studies. Lancet 2017;390:2769–78. doi:10.1016/S0140-6736(17)32448-0.

[2]

Schwartz DA, Loftus E V., Tremaine WJ, Panaccione R, Harmsen WS, Zinsmeister AR, et

of

al. The natural history of fistulizing Crohn’s disease in Olmsted County, Minnesota.

Gecse KB, Bemelman W, Kamm MA, Stoker J, Khanna R, Ng SC, et al. A global

-p

[3]

ro

Gastroenterology 2002;122:875–80. doi:10.1053/gast.2002.32362.

re

consensus on the classification, diagnosis and multidisciplinary treatment of perianal fistulising Crohn’s disease. Gut 2014;63:1381–92. doi:10.1136/gutjnl-2013-306709. Lee MJ, Parker CE, Taylor SR, Guizzetti L, Feagan BG, Lobo AJ, et al. Efficacy of Medical

lP

[4]

na

Therapies for Fistulizing Crohn’s Disease: Systematic Review and Meta-analysis. Clin Gastroenterol Hepatol 2018;16:1879–92. doi:10.1016/j.cgh.2018.01.030. Panés J, Rimola J. Perianal fistulizing Crohn’s disease: Pathogenesis, diagnosis and

Jo ur

[5]

therapy. Nat Rev Gastroenterol Hepatol 2017;14:652–64. doi:10.1038/nrgastro.2017.104. [6]

Maaser C, Sturm A, Vavricka SR, Kucharzik T, Fiorino G, Annese V, et al. ECCO-ESGAR Guideline for Diagnostic Assessment in IBD Part 1: Initial diagnosis, monitoring of known IBD, detection of complications. J Crohn’s Colitis 2019;13:144-164K. doi:10.1093/ecco-jcc/jjy113.

[7]

Van Assche G, Vanbeckevoort D, Bielen D, Coremans G, Aerden I, Noman M, et al. Magnetic resonance imaging of the effects of infliximab on perianal fistulizing Crohn’s disease. Am J Gastroenterol 2003;98:332–9. doi:10.1016/S0002-9270(02)05909-9.

Journal Pre-proof [8]

20

Horsthuis K, Ziech MLW, Bipat S, Spijkerboer AM, de Bruine-Dobben AC, Hommes DW, et al. Evaluation of an MRI-based score of disease activity in perianal fistulizing Crohn’s disease. Clin Imaging 2011;35:360–5. doi:10.1016/j.clinimag.2010.09.003.

[9]

Ng SC, Plamondon S, Gupta A, Burling D, Swatton A, Vaizey CJ, et al. Prospective evaluation of anti-tumor necrosis factor therapy guided by magnetic resonance imaging for crohn’s perineal fistulas. Am J Gastroenterol 2009;104:2973–86.

Karmiris K, Bielen D, Vanbeckevoort D, Vermeire S, Coremans G, Rutgeerts P, et al.

ro

[10]

of

doi:10.1038/ajg.2009.509.

Long-Term Monitoring of Infliximab Therapy for Perianal Fistulizing Crohn’s Disease by

Tougeron D, Savoye G, Savoye-Collet C, Koning E, Michot F, Lerebours E. Predicting

lP

[11]

re

doi:10.1016/j.cgh.2010.10.022.

-p

Using Magnetic Resonance Imaging. Clin Gastroenterol Hepatol 2011;9:130-136.e1.

na

factors of fistula healing and clinical remission after infliximab-based combined therapy for perianal fistulizing crohn’s disease. Dig Dis Sci 2009;54:1746–52.

[12]

Jo ur

doi:10.1007/s10620-008-0545-y.

Savoye-Collet C, Savoye G, Koning E, Dacher JN, Lerebours E. Fistulizing perianal Crohn’s disease: Contrast-enhanced magnetic resonance imaging assessment at 1 year on maintenance anti-TNF-alpha therapy. Inflamm Bowel Dis 2011;17:1751–8. doi:10.1002/ibd.21568.

[13]

Samaan MA, Puylaert CAJ, Levesque BG, Zou GY, Stitt L, Taylor SA, et al. The development of a magnetic resonance imaging index for fistulising Crohn’s disease. Aliment Pharmacol Ther 2017;46:516–28. doi:10.1111/apt.14190.

[14]

Gomollón F, Dignass A, Annese V, Tilg H, Van Assche G, Lindsay JO, et al. 3rd European evidence-based consensus on the diagnosis and management of Crohn’s disease

Journal Pre-proof

21

2016: Part 1: Diagnosis and medical management. J Crohn’s Colitis 2017;11:3–25. doi:10.1093/ecco-jcc/jjw168. [15]

Present DH, Rutgeerts P, Targan S, Hanauer SB, Mayer L, Van Hogezand RA, et al. Infliximab for the treatment of fistulas in patients with Crohn’s disease. N Engl J Med 1999:1398–405.

[16]

Bankier AA, Levine D, Halpern EF, Kressel HY. Consensus Interpretation in Imaging

of

Research: Is There a Better Way? Radiology 2010;257:14–7.

[17]

ro

doi:10.1148/radiol.10100252.

Landis JR, Koch GG. The Measurement of Observer Agreement for Categorical Data.

Tozer P, Ng SC, Siddiqui MR, Plamondon S, Burling D, Gupta A, et al. Long-term MRI-

re

[18]

-p

Biometrics 1977;33:159–74. doi:10.1109/ICDMA.2010.328.

lP

guided combined anti-TNF-α and thiopurine therapy for crohn’s perianal fistulas.

Sturm A, Maaser C, Calabrese E, Annese V, Fiorino G, Kucharzik T, et al. ECCO-ESGAR Guideline for Diagnostic Assessment in IBD Part 2: IBD scores and general principles

Jo ur

[19]

na

Inflamm Bowel Dis 2012;18:1825–34. doi:10.1002/ibd.21940.

and technical aspects. J Crohn’s Colitis 2019;13:273–84. doi:10.1093/ecco-jcc/jjy114.

Journal Pre-proof

22

Tables Table 1. The items included in the original[7] and modified Van Assche index[13]

of

ro

-p

re

Jo ur

Hyperintensity on T2weighted images Absent Mild Pronounced

lP

Location Extra- or intersphincteric Transsphincteric Suprasphincteric Extension Infralevatoric Supralevatoric

Modified Van Assche index Score Item Extension 0 Absent 1 Infralevatoric 2 Horseshoe 3 configuration Supralevatoric Hyperintensity on T2-weighted 1 images Absent 2 Mild 3 Pronounced Rectal wall involvement 1 Normal 2 Thickened Increased signal intensity Inflammatory mass Absent 0 Diffuse 4 Focal 8 Collection-small Collection-medium Collection-large Dominant feature of tract and extensions 0 Fibrous 4 Granulation tissue Fluid/pus

na

Original Van Assche index Item Number of fistula tracts None Single, unbranched Single, branched Multiple

Collections (cavities > 3 mm diameter) Absent Present Rectal wall involvement Normal Thickened Original Van Assche index

0 2 0-22

Modified Van Assche index

Score Weight 0 1 2

Score

1.5

0-4.5

0 1 2

2.3

0-4.6

0 1 2

1.0

0-2

1.2

0-6

1.2

0-2.4

3

0 1 2 3 4 5

0 1 2

0-19.5

Journal Pre-proof Table 2. Patient characteristics Patient characteristic Sex, female / male* Age in years* Active smokers* Age at diagnosis CD in years* Disease location* Ileal Colonic Ileocolonic Upper GI and ileocolonic

N (%) or median [IQR] 12 (48%) / 13 (52%) 27 [20-37] 7 (28%) 23 [16-30] 2 (8%) 7 (28%) 14 (56%) 2 (8%)

Jo ur

na

lP

re

-p

ro

of

Previous medical treatment# Anti-TNF 15 (50%) Methotrexate 6 (20%) Thiopurines 22 (73%) Steroids 12 (40%) Previous fistula surgical treatment# 23 (77%) Drainage and/or seton 3 (10%) Fistulotomy 1 (3%) Advancement flap Diverting ostomy at time of first 5 (17%) MRI# CD = Crohn’s disease. *descriptives for the 25 patients. #descriptives for the 30 cases.

23

Journal Pre-proof

24

Table 3. Outcomes of modified Van Assche index in clinical responders versus nonresponders

10 (63%) 2 (13%) 1 (6%) 3 (19%)

7 (50%) 1 (7%) 0 (0%) 6 (43%)

6 (43%) 1 (7%) 0 (0%) 7 (50%)

1 (6%) 4 (25%) 11 (69%)

3 (19%) 5 (31%) 8 (50%)

0 (0%) 3 (21%) 11 (79%)

0 (0%) 5 (36%) 9 (64%)

14 (88%) 2 (12%)

15 (94%) 1 (6%)

9 (64%) 5 (37%)

10 (71%) 4 (29%)

13 (81%) 1 (6%) 2 (13%) 0 (0%) 0 (0%) 0 (0%)

9 (64%) 1 (7%) 0 (0%) 0 (0%) 1 (7%) 3 (21%)

7 (50%) 4 (29%) 1 (7%) 0 (0%) 1 7%) 2 (14%)

1 (6%) 12 (75%) 3 (19%)

3 (19%) 14 (81%) 0 (0%)

0 (0%) 12 (86%) 2 (14%)

0 (0%) 14 (100%) 0 (0%)

9.6 [5.8-12.7]

5.8 [3.5-8.5]*

7.7 [5.8-13.5]

8.2 [5.8-11.5]*

ro

-p re

lP

8 (50%) 2 (12%) 0 (0%) 0 (0%) 3 (19%) 3 (19%)

of

7 (44%) 1 (6%) 3 (19%) 5 (32%)

Jo ur

Frequency (%) Extension Absent Infralevatoric Horseshoe Supralevatoric Hyperintensity T2 Absent Mild Pronounced Rectal wall Normal Thickened and increased SI Inflammatory mass Absent Diffuse Focal Collection small Collection medium Collection large Dominant feature Fibrous Granulation tissue Fluid/pus Total modified Van Assche index (median [IQR])

Non-responder N=14 Baseline Follow-up

na

Item

Responder N=16 Baseline Follow-up

*difference between the modified Van Assche index at baseline and follow-up: p=0.008 for responders, p=0.624 for non-responders.

Journal Pre-proof Table 4. Interobserver agreement for the original and modified Van Assche index Modified Van Assche index Extension T2-weighted signal intensity Rectal wall involvement Inflammatory mass Predominant feature Modified Van Assche index

Interrater reliability κ or ICC (95%CI) 0.65 (0.46-0.85) 0.65 (0.47-0.83) 0.56 (0.27-0.85) 0.61 (0.41-0.82) 0.40 (0.06-0.74) 0.74 (0.60-0.84)

Jo ur

na

lP

re

-p

ro

of

Original Van Assche index Number of fistula tracts 0.81 (0.69-0.93) Location 0.50 (0.28-0.72) Extension 0.56 (0.33-0.79) T2-weighted signal intensity 0.65 (0.47-0.83) Collections 0.53 (0.26-0.78) Rectal wall involvement 0.56 (0.27-0.85) Original Van Assche index 0.71 (0.55-0.82) ICC = Intraclass Correlation Coefficient, CI = confidence interval

25

Journal Pre-proof

26

2 (13%) 6 (38%) 2 (13%) 6 (38%)

0 (0%) 5 (36%) 3 (21%) 6 (43%)

0 (0%) 4 (29%) 5 (36%) 5 (36%)

7 (44%)

9 (56%)

9 (64%)

6 (43%)

9 (56%) 0 (0%)

7 (44%) 0 (0%)

5 (36%) 0 (0%)

8 (57%) 0 (0%)

8 (57%) 6 (43%)

7 (50%) 7 (50%)

3 (19%) 5 (31%) 8 (50%)

0 (0%) 3 (21%) 11 (79%)

0 (0%) 5 (36%) 9 (64%)

10 (63%) 6 (38%)

15 (94%) 1 (6%)

10 (71%) 4 (29%)

10 (71%) 4 (29%)

14 (88%) 2 (13%)

15 (94%) 1 (6%)

9 (64%) 5 (36%)

10 (71%) 4 (29%)

13.0 [10.015.8]

9.5 [5.313.0]*

11.5 [10.816.5]

11.5 [9.815.5]*

-p

ro

of

1 (6%) 5 (31%) 4 (25%) 6 (38%)

1 (6%) 4 (25%) 11 (69%)

13 (82%) 3 (19%)

re

11 (69%) 5 (31%)

Jo ur

Frequency (%) Number of fistula tracts None Single, unbranched Single, branched Multiple Location Extra- or intersphincteric Transsphincteric Suprasphincteric Extension Infralevatoric Supralevatoric Hyperintensity on T2weighted images Absent Mild Pronounced Collections (cavities > 3 mm diameter) Absent Present Rectal wall involvement Normal Thickened Total modified Van Assche index (median [IQR])

Non-responder N=14 Baseline Follow-up

lP

Item

Responder N=16 Baseline Follow-up

na

Supplementary Table - Outcomes of original Van Assche index in clinical responders versus non-responders

*difference between the original Van Assche index at baseline and follow-up: p=0.011 for responders, p=0.325 for non-responders.

Journal Pre-proof

27

Figure legends Figure 1. Patient flowchart

Figure 2. Modified Van Assche index per case in clinical responders and non-responders

Figure 3. Baseline and follow-up MRI of a clinical responder with a decreased modified Van

of

Assche index at follow-up. Coronal T2-weighted sequences (A and B) and axial T2-weighted

ro

sequences with fat suppression (C and D) of a 32-year old man at baseline (A and C) and

-p

follow-up (B and D). The baseline MRI scans show a collection that decreased on the posttreatment MRI scans after 11 months. Moreover, the extent and intensity of inflammatory

re

tissue decreased. This resulted in a decrease from 17 to 5 on the modified Van Assche index.

na

lP

The patient has a seton in situ at both time-points (arrow on A and B)

Figure 4. Baseline and follow-up MRI of a clinical responder without a decreased modified

Jo ur

Van Assche index at follow-up. Coronal T2-weighted sequences (A and B) and axial T2weighted sequences with fat suppression (C and D) of a 27-year old man at baseline (A and C) and follow-up (B and D). The baseline MRI scans show a hyperintense fistula tract that decreases in volume on the post-treatment MRI after 16 months. This was not captured by either the original or modified Van Assche index since there is no item that reflects volume changes, the total modified Van Assche index was 10 at baseline and 10 at follow-up.

Figure 5. MRI examples of interobserver differences in three different patients. A: Axial T1weighted post-contrast sequence in a 56-year old woman, one observer scored this (arrow) as a medium collection and the other observer interpreted this as part of a tract and scored

Journal Pre-proof

28

absent inflammatory masses (4 points versus 0 points, respectively, on the modified Van Assche index) B: Axial T2-weighted sequence with fat suppression of a 29-year old man, observers disagreed on this being a focal (2 points) or diffuse (3 points) infiltrate (arrow). C: Axial T1-weighted post-contrast sequence in a 36-year old woman, observers disagreed on

Jo ur

na

lP

re

-p

ro

of

the predominant feature: fluid (2 points) or granulation tissue (1 point)(arrow).

Journal Pre-proof

29

Highlights manuscript ‘Evaluation of the modified Van Assche index for assessing response to anti-TNF therapy with MRI in perianal fistulizing Crohn’s disease’ -

The modified Van Assche index decreases significantly in patients responding to anti-TNF therapy

-

In one third of responders the modified Van Assche index did not drop between baseline and follow-up No clear added value of the modified over the original Van Assche index was found

-

Further optimization of the modified Van Assche index should be realized

Jo ur

na

lP

re

-p

ro

of

-

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5