Study design

This was a retrospective observational study.

Setting

Our inflammatory bowel disease (IBD) service manages a cohort of local and tertiary referrals. Patients receiving biologics are entered onto the IBD audit database.11 Details on the clinical management were available on our electronic clinical information system.

Patient population

We included all patients who initiated IFX between January 2007 and July 2016. We excluded patients who were lost to follow-up within 3 months of initiation, had ulcerative colitis, indeterminate colitis or pouchitis, received less than three or episodic infusions.

Patients initiating IFX received induction dose of 5 mg/kg at weeks 0, 2, 6 and 8-weekly thereafter.

Intervention—TDM service

TDM was introduced to our institution in September 2013 using the lab of Sandwell and West Birmingham NHS Trust. Patients who started IFX before introduction of TDM (pre-TDM era), underwent TDM testing after its introduction.

TDM was undertaken prior to infusion at week 14 and week 50, at diagnosis or suspicion of SLR and to investigate SAEs or guide treatment discontinuation. Additionally, TDM was requested at the discretion of clinicians at any point for non specific reasons (‘routine testing’). It is likely that at the start of TDM service there was uncertainty about its application and routine testing may have been for reassurance that clinical benefits reflected adequate dosing and/or there may have been an intention to adjust therapy that was not pursued.

Application of TDM results to guide therapy in SLR

Dose escalation was offered for undetectable drug trough levels without ATI with evidence of active disease (raised CRP, calprotectin or evidence on imaging or endoscopy). Patients with undetectable trough levels with ATI were switched to different anti-TNF, whereas treatment was changed to a non anti-TNF biologic when the trough levels were therapeutic and there was active disease. The above represented a guide rather than a strict protocol. The final decision was at the discretion of clinicians.

Infliximab trough levels and ATI

Infliximab levels were measured by an ELISA assay as per Barlow et al. 12 Serum was added to a TNF-a coated plate and a detector was added to detect the IFX/TNFa complex. The reaction was detected with a fluorescent substrate.12 The cost was £60 per assay.

TDM Measurements and terminology

Infliximab trough levels were expressed in μg/ml. ATI were reported qualitatively as ‘positive’ when ATI were and ‘negative’ when ATI were not detected. The ATI assay is not drug tolerant and would only detect ATI in the absence of detectable infliximab. Trough levels below 1 µg/mL are classified as ‘undetectable’ and equal to or above 1 µg/mL as therapeutic.

Patients who had TDM at least once are reported as ‘TDM tested’ and those who never had TDM as ‘TDM never-tested’. Undetectable trough level at least once is captured as ‘undetectable levels episode’, ATI positive at least once as ‘ATI episode’ and dose intensification at least once as ‘dose intensification episode’.

Data extraction and definitions

Demographic (age, sex) and disease-specific (Montreal Classification (MC),13 duration) data were extracted.

Disease activity was defined according to physician’s global assessment (PGA) as active or inactive and was recorded from the consultation letters. PNR was defined at week 14 according to PGA for luminal disease as complete absence of clinical improvement and for perianal CD as failure to reduce the number of draining fistulas by 50%.14 SLR was defined according to PGA as worsening of symptoms after an initial response to IFX and raised CRP, calprotectin and/or endoscopic activity.

Infliximab trough level was the level measured on a sample drawn just prior to dosing. SAE was defined as any acute or delayed infusion reaction necessitating infliximab discontinuation.15

Outcome measures

Our primary outcome measures were: The clinicians’ response to each TDM requested and treatment discontinuation because of SLR or SAE.16 Secondary outcome was the need for surgery (small bowel resection, colectomy or stricturoplasty) after the initiation of infliximab. Outcomes were compared between patients whose treatment was initiated before (pre-TDM era) or after the implementation of TDM (post-TDM era) and also between TDM tested and TDM never tested patients.

An additional secondary outcome was the proportion of clinical remission episodes 6 months after every TDM result.

Statistical analyses

Data were analysed with the Statistical Package for Social Sciences (SPSS V.17) and Prism Graphpad V.7. Continuous variables were expressed as median (IQR) and compared using the Mann-Whitney U test. Categorical variables were expressed as absolute numbers n (percentage of population) and compared using the χ² test. Multivariate logistic regression was used to identify factors associated with drug discontinuation due to SLR/SAE and surgery. In the multivariate analysis (MVA) were entered all the variables that were significantly associated with our outcome on univariate analysis (UVA) (p value threshold 0.05). Kaplan-Meier curves and the log rank test compared the IFX discontinuation free and surgery-free survival between groups. Analyses were two-tailed and p<0.05 were considered significant.

Ethical considerations

TDM in our practice constitutes a new clinical service (‘TDM service’) and this study was done as service evaluation. The care provided to patients was part of standard care. Therefore, formal ethics approval was not required. We received approval from the Research and Development (R&D) department of London North West Healthcare NHS Trust (SE16/042) as per Trust Policy.

Population

Total 517 patients were identified on our database. From these, 226 were excluded for various reasons (figure 1). Total 291 patients were included.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Study population. IFX, infliximab; UC, ulcerative colitis.

The median (IQR) patient age was 35.0 (21.0) years with disease duration 89.95 (157.2) months prior to IFX. The MC and IFX indications are described on table 1. The duration of follow-up was 44.7 (40.3) months.

Total 161 (55.3%) patients initiated IFX in the pre-TDM era and 238 (81.8%) were TDM tested. From these, 95 (39.9%) had an undetectable levels episode and 76 (31.9%) an ATI episode. Total 199 (83.6%) had at least one routine TDM.

IFX was discontinued due to SLR or SAE in 109 patients (37.5%) and 39/109 (35.8%) had at least one routine TDM. Of these 39, 18 (46%) had undetectable trough levels, 5 (13%) levels of 1–3 µg/mL, 16 (41%) levels of 3.1–10 µg/mL and 17 (44%) had ATI prior to discontinuation. Surgery was required in 44 patients (15.1%). Total 62 patients (21.3%) had their dose intensified.

Effect of TDM results on clinical management

Total 672 TDM tests were requested for 238 patients. The median (IQR) IFX trough level was 3.4 (5.4) μg/mL. IFX trough levels were undetectable in 154 (22.9%) samples, between 1 and 3 µg/mL in 161 (24%), between 3.1 and 10 µg/mL in 356 (53%) and above 10 µg/mL in 1 (0.1%). Total 118 (17.6%) TDM samples were ATI positive. The time from IFX initiation to first TDM was 12.5 (30.2) months and the first and subsequent trough levels were 3.7 (5.3) μg/ml and 3.2 (5.5) μg/mL, respectively.

Total 469 (69.8%) routine TDM assays were undertaken. The second most common indication was SLR in 180 (26.8%) (table 2).

Most TDMs (526/672; 78.3%) were not followed by change in patient management. Total 58 (8.6 %) TDMs were followed by dose intensification, 40 (6.0%) by switch to alternative anti-TNF, 12 (1.8%) to non-anti-TNF biologic, 10 (1.5%) by addition of thiopurine, 9 (1.3%) by dose de-escalation and 17 (2.5%) by biologic discontinuation. Total 78 (32.8%) TDM tested patients had management modifications as a result of TDM. When TDM indication was PNR or SLR, change in management occurred after 54.6% (100/183) of levels compared with 9.4% (46/489) for other indications (p<0.01).

Total 411 (61.3%) assays were followed by clinical remission after 6 months. When TDM was requested for PNR or SLR, the 6 months remission rates were similar irrespective of whether change in patient management followed (49 (52.1%)) or not (36 (45.0%)) (p=0.35). The median (IQR) trough level followed by remission 6 months after testing was 4.3 (5.4) μg/mL compared with 2.3 (4.7) μg/mL for assays followed by active disease (p<0.01).

Initiation of IFX in the pre-TDM or post-TDM era: patient characteristics and outcomes

The age of patients initiated on IFX in the pre-TDM era was 37.0 (19.0) compared with 33.0 (23.0) years for patients of the post-TDM era (p<0.01). The duration of follow-up was 62.3 (36.5) months for the former and 25.7 (14.5) for the latter (p<0.01). Perianal and penetrative disease were more prevalent among pre-TDM era patients. Previous exposure to biologic was noted in 7 (4.3%) patients of the pre-TDM era and 17 (13.1%) of the post-TDM era (p<0.01). 111 (68.9%) patients of the pre-TDM era had TDM compared with 127 (97.7%) of the post-TDM era (p<0.01). More patients of the post-TDM era had an ATI episode (32.3% vs 21.1%, p=0.03).

IFX was discontinued due to SLR or SAE in 69 (42.9%) patients of the pre-TDM era compared with 40 (30.8%) of the post-TDM era (p=0.03). The median (IQR) time to discontinuation was 48.0 (40.1) and 19.4 (16.8) months, respectively (p<0.01). The rate of intestinal surgery after initiation of IFX and time to surgery did not differ between groups (online supplementary table 1).

Disease outcomes and patient characteristics for TDM tested and TDM never tested patients

There was no difference in the age, disease duration and MC between groups. The duration of follow-up was 37.4 (31.6) months for the TDM tested and 67.0 (53.6) for the TDM never tested. More patients of the former group (24 (10.1%)) compared with the latter (0 (0%)) received IFX as second line biologic (p=0.02) (online supplementary table 2).

IFX was discontinued due to SLR or SAE in 69 (29.0%) TDM tested patients and 40 (75.5%) TDM never tested (p<0.01) and the discontinuation free survival was longer for the former (32.9 (35.1) months) compared with the later (19.9 (22.8) months) (p<0.01) (figure 2A). Less TDM tested patients (29; 12.2%) required surgery compared with TDM never tested (15; 28.3%) (p<0.01) (table 3). Time to surgery did not differ between groups (online supplementary table 2, figure 2B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Survival analysis of the time to drug discontinuation due to SLR or SAE (A) and time to surgery for IBD (B). The dotted line and the solid line represent the ‘TDM never tested’ and the ‘TDM tested’ group, respectively. IBD, inflammatory bowel disease; IFX, infliximab; SAE, serious adverse event; SLR, secondary loss of response; TDM, therapeutic drug monitoring.

Factors associated with IFX discontinuation due to SLR or SAE

The UVA of factors associated with discontinuation of IFX due to SLR or SAE identified that the overall duration of follow-up, previous exposure to biologic, initiating treatment in the post-TDM era and being TDM never tested were associated with higher rates of discontinuation due to SLR or SAE (online supplementary table 3).

MVA (OR (95% CI), p value) showed that being TDM never tested (8.0 (3.8, 17.0), <0.01) and having previously been exposed to a biologic (4.4 (1.8, 10.7), <0.01) were associated with IFX discontinuation (table 3).

Factors associated with intestinal surgery after IFX initiation

The UVA of factors associated with surgery identified that the overall duration of follow-up, being TDM never tested and previous exposure to biologic, are associated with higher rates of surgery after initiation of IFX (online supplementary table 4).

MVA showed that being TDM never tested (2.5, (1.1, 5.7), 0.03) and previous exposure to biologic (4.3 (1.6, 11.8), <0.01) are associated with surgery after initiation of IFX (table 4).

Key results and interpretation

Strengths

Our study has several strengths. Patients were identified through a prospectively built database capturing consecutive patients. Therefore, we report on a large number of patients and TDMs. Second, all tests were undertaken by the same laboratory with identical technique. Last, although the overall duration of follow-up was greater for TDM tested patients and patients of the pre-TDM era, this was not a significant prognostic factor in the MVA, therefore our results do not reflect lead time bias.

Limitations

This is a retrospective study and data collected on clinical outcomes rely record keeping. We used the PGA as a measure of clinical remission. This may be subject to interpretation bias. Second, our results may be subject to tertiary centre bias. Although the majority of patients represent our local area, our outpatient clinic includes tertiary patients who have more complex disease. Although there were no significant differences in the MC among groups, this may be inadequate to describe disease complexity which is better reflected by the level of intestinal damage (Lemann classification).28 Last, we use the endpoints of drug discontinuation due to SLR or SAE and surgery; however, we acknowledge the fact that it is not uncommon to stop IFX for ongoing abdominal pain, without formal assessment of activity. Similarly surgery may not reflect active inflammation or stricturing disease.