Tumour necrosis factor inhibitors in Ulcerative colitis: real-world data on Therapeutic drug monitoring and evaluation of current treatment targets (STRIDE II)

Abstract

Background

The benefit of therapeutic drug monitoring (TDM) and implementation of recommendations from the Selection of Therapeutic Targets in Inflammatory Bowel Disease (IBD, STRIDE) are discussed in the IBD community. We report real-world data in ulcerative colitis patients receiving first-line tumour necrosis factor inhibitor (TNFi) treatment followed by TDM, and assess how implementation of the STRIDE II recommendations might affect clinical practice.

Methods

Adult, biologically naïve UC patients starting TNFi between 2014 and 2021 at Oslo University Hospital were included in a medical chart review study, and data were collected at three and twelve months after the start of treatment. Target serum drug levels were defined as ≥7.5 mg/L for adalimumab and ≥5 mg/L for infliximab.

Results

Of 141 included patients, 36% were in clinical and biochemical (combined) remission after twelve months. Among 102 treatment persistent patients, 54% were in combined remission after twelve months. Target drug level at three months was associated with clinical remission at twelve months (OR = 2.97, 95% CI [1.24–7.12]) and biochemical remission at twelve months (OR = 2.64, 95% CI [1.03–6.77]). In total, 56% of recorded dosage adjustments were related only to serum drug levels.

Conclusions

Combined remission rates at twelve months for treatment persistent patients suggest that 46% should have been considered for a change of treatment according to the STRIDE II recommendations. A majority of dosage adjustments were made proactively. Target drug level at three months was associated with remission at twelve months and supports the use of proactive TDM.

Introduction

Ulcerative colitis (UC) is a chronic inflammatory disease of the colon. Treatment with biological agents is a cornerstone of modern treatment of moderate to severe UC, and these drugs are prescribed to an increasing number of patients at earlier stages of the disease [1]. Treatment goals and surrogates to monitor treatment response are rapidly evolving, and in 2021, the Selection of Therapeutic Targets in Inflammatory Bowel Disease (STRIDE) initiative proposed updated and more comprehensive treatment targets including pre-specified timeframes (short-term, intermediate and long-term targets) [2]. Implementation of more ambitious targets might constitute a major change in the approach to patients with UC.

Tumour necrosis factor inhibitors (TNFi) are at present first-line choice of biologic therapy for UC patients in many countries. The ability of TNFi to induce and maintain remission has been well documented in randomized controlled trials (RCTs) [3, 4], as have their safety profiles [4, 5]. Reported rates of clinical remission after six to twelve months of treatment vary from 32% to 77% [6–9], safety and persistence data display similar discrepancies. As clinical outcome may depend on differences in clinical practice, publication of real-world data including evaluation of clinical decision-making and dosage adjustments are vital to elucidate the actual and potential effectiveness of TNFi treatment. Furthermore, the advent of therapeutic drug monitoring (TDM) has provided additional potential for difference in practice between hospitals and countries. The benefit of TDM and the ideal timing of monitoring are currently under debate [10], while the association between sufficient drug levels and positive outcomes has been established [11–13].

The aim of the present study was to report real-world data on the effectiveness and safety of TNFi treatment in UC patients and explore whether proactive TDM is associated with remission outcomes. Furthermore, we aimed to assess whether our clinical practice was in accordance with the STRIDE II recommendations.

Materials and methods

Study design

In this single-centre, retrospective medical chart review study, all biologically naïve adults (≥18 years) with UC who initiated treatment with infliximab (IFX) or adalimumab (ADA) between 01.01.2014 and 31.12.2020 at the Department of Gastroenterology, Oslo University Hospital (OUH), were eligible for inclusion. Exclusion criteria were previous major inflammatory bowel disease (IBD)-related surgery, uncertain diagnosis, primary sclerosing cholangitis (PSC) or IBD treatment at another hospital during the follow-up period.

Patients were identified by search in the electronic patient journal (EPJ) of the hospital using the ICD-10-code K51 and one of the following ATC-codes: 4AB02 (IFX) or 4AB04 (ADA). Identification was followed by a manual EPJ-review to verify eligibility. Data were recorded for the first twelve months of TNFi treatment or until treatment discontinuation.

Data collection and outcome definitions

Patient demographics, disease characteristics and previous treatment were recorded at the time of treatment induction. Clinical scores and laboratory results including serum drug concentrations were collected from EPJ recordings generated at outpatient visits at three months (±30 days) and twelve months (±90 days) after the start of treatment. Data on adverse events and dosage adjustments were recorded by complete chart evaluation during the follow-up period.

Study definitions

Disease activity and adverse events

Clinical remission was defined as a 6-point Mayo score ≤1 [14]. Biochemical remission was defined as CRP <5 mg/L and Fecal Calprotectin (FC) <250 µg/g. Combined remission was defined as clinical and biochemical remission.

Discontinuation of TNFi therapy during follow-up, or addition of another biologic agent during follow-up was classified as non-remission. Adverse events were defined in accordance with the International Conference on Harmonisation (ICH) Good Clinical Practice guidelines [15].

STRIDE II definitions

The STRIDE II recommendations suggest clinical remission as an intermediate treatment target, defining this as Patient-Reported Outcome (PRO2) = 0 or Partial Mayo Score <3. Normalisation of CRP and FC (100–250 µg/g) is also recommended as an intermediate ­target. We consider the targets from STRIDE II to ­correspond with our study definitions of clinical and biochemical remission.

Endpoints

The primary endpoint was the proportion of patients in clinical, biochemical and combined remission twelve months after start of TNFi therapy. Secondary endpoints were i) the proportion of patients in clinical, biochemical and combined remission after three months and ii) adverse events. Furthermore, we explored associations between i) a number of dosage adjustments and remission outcomes and ii) serum drug levels and remission outcomes.

Measurement of drug levels

Serum IFX and ADA levels were measured at every infusion visit or scheduled outpatient clinic visit every three months after induction treatment, respectively. Serum for IFX levels was collected at trough before every infusion. Serum for ADA levels was collected during outpatient clinic visits, though not at a fixed time point during treatment intervals as the visits were not scheduled according to the timing of ADA injections. Drug level- and anti-drug antibody (ADAb) measurements were performed at the Department of Medical Biochemistry, OUH, Radiumhospitalet, Norway, using the 3-step immunofluorometric method on the AutoDELIFA^® immunoassay platform. ADAb analyses were routinely performed by the laboratory if serum drug levels were <1 mg/L for both IFX and ADA. According to the clinical practice at our department, physicians applied a proactive TDM strategy in the follow-up of patients. No standardised algorithm for dosage adjustments was applied, adjustments were subject to clinical evaluation of the attending physician based on clinical scores and biochemistry, including serum drug levels. The guidelines at our department recommended serum drug levels for patients in remission between 3–8 mg/L for IFX and 5–12 mg/L for ADA in accordance with available data at the onset of our TDM practice. The American Gastroenterological Association (AGA) suggested target serum drug levels of ≥5 mg/L for IFX and ≥7.5 mg/L for ADA in 2017 [16]. Analyses of serum drug levels in this paper applied the AGA serum drug level recommendations.

Recording of dosage adjustments

To examine the impact of the availability of serum drug concentrations on clinical decision-making, the motivation behind the change in dosage was recorded from the EPJ. Adjustments of drug dosage were defined as intended changes of the interval between treatments or dose per infusion/injection recorded in the EPJ as a response to i) serum drug levels, ii) clinical presentation of the patient and/or biochemical markers or iii) both i) and ii). To record a change in dosage for IFX patients, we required the dosage change to persist for a minimum of two adjacent infusion visits.

Statistical analysis

Continuous data are presented as median (range/interquartile range). Categorical data are described with counts and percentages. Univariate logistic regression was applied to explore associations between serum drug levels and remission outcomes and to investigate whether any baseline characteristics were associated with outcomes to identify potential confounders. The Wilcoxon rank-sum test was applied to compare groups regarding continuous variables with skewed distribution. p values <0.05 were considered statistically significant. The software program Stata version 17.0 was used for statistical analyses.

Ethics

This study was conducted as a quality assurance study of IBD care at OUH, and as such an informed consent was waived. Publication was approved by the head of the gastroenterology department in accordance with hospital publishing guidelines concerning quality assurance studies. The study was approved by the institutional review board at OUH (personvern@oslo-universitetssykehus.no), case number 2014/7822 and adhered to the Declaration of Helsinki.

Results

Study population

A total of 210 patients were identified as eligible in the hospital EPJ search. A total of 69 patients were excluded, of whom 16 had undergone major IBD-related surgery, 14 initiated treatment at another hospital, twelve had indeterminate IBD, seven had received golimumab, six had PSC comorbidity, five had changed treatment hospital, four were not TNFi-naïve, and the remaining five exclusions were single occurrences of unavailable chart, period of treatment at another hospital, pause in treatment because of surgery, age <18 years at baseline and patient refusal to participate in hospital register. After exclusion, 141 patients were included in the final dataset. Baseline patient characteristics at the start of treatment are summarized in Table 1.

Table 1.

Baseline characteristics.

	All patients (n = 141)
Sex – male, n (%)	85 (60.3)
Age (years) at time of diagnosis, median (range)	25 (10–73)
Age (years) at baseline, median (range)	32 (18–74)
Disease duration (months) at baseline, median (range)	44 (0–409)
Ulcerative colitis extent, n (%) E1 E2 E3	7 (5.0) 33 (23.4) 101 (71.6)
Endoscopic Mayo score at most recent colonoscopy at baseline, n (%) 0 1 2 3	2 (1.4) 8 (5.7) 83 (58.9) 48 (34.0)
Biologic agent, n (%) Adalimumab Infliximab	46 (32.6) 95 (67.4)
Immunomodulator, n (%) Azathioprine Methotrexate Puri-Nethol None	79 (56) 20 (14.2) 1 (0.7) 41 (29.1)
Fecal Calprotectin (µg/g) at baseline*, median (IQR) Missing n = 45	2140 (720–6000)
CRP (mg/dL) at baseline, median (IQR)	3.8 (1.5–10.6)

*Median days from baseline fecal calprotectin until the start of treatment was 9 (range 2–21).

Achievement and maintenance of remission

Percentages [95% CI] of patients in remission at three and twelve months according to the different definitions are given in Table 2. Baseline and clinical characteristics were individually applied as independent variables in univariate logistic regression analyses with remission outcomes as dependent variables. As none yielded a p-value of <0.1, no multivariate regression was performed, and no baseline demographic or clinical characteristic was associated with remission outcomes. The percentages of combined remission at twelve months were 39.5% (95% CI [29.7 − 50.3]) and 28.6% (95% CI [16.0 − 45.6]) for IFX and ADA, respectively. No statistically significant differences in achievement of remission were observed between the two TNFi for any of the remission definitions. In total, 39 (27.7%) patients discontinued TNFi treatment or added a second biologic agent during the follow-up period of one year. These patients were classified as non-remission. Median time to discontinuation was 169 days (52–220), and 11 of 39 discontinuations occurred within the first three months. Percentages of persistent patients in remission at three and twelve months are listed in Table 3. Most frequent cause for discontinuation of treatment was primary non-response (Table 4). Eight patients (IFX = 6, ADA = 2) discontinued TNFi treatment in the presence of ADAb development and undetectable drug levels (<1 mg/L). ADAb titres were high (> 120 µg/L for IFX and >80 µg/L for ADA) in three of these patients.

Table 2.

Remission rates at 3 and 12 months, n = 141.

	3 months	12 months
Clinical Remission	64.5 [55.1, 73.0] Missing n = 31	57.6 [49.2, 65.7] Missing n = 4
Biochemical Remission	40.7 [31.0, 51.1] Missing n = 50	41.6 [33.2, 50.5] Missing n = 16
Combined Remission	34.2 [24.3, 45.7] Missing n = 65	36.4 [28.2, 45.4] Missing n = 20

Data presented as % [95% CI].

Table 3.

Remission rates at 3 and 12 months in persistent patients, n = 102.

	3 months	12 months
Clinical Remission	71.2 [62.0, 79.8] Missing n = 31	80.6 [71.5, 87.3] Missing n = 4
Biochemical Remission	46.3 [35.5, 57.3] Missing n = 50	60.5 [49.7, 70.3] Missing n = 16
Combined Remission	40.0 [28.7, 52.5] Missing n = 65	53.7 [42.7, 64.3] Missing n = 20

Data presented as % [95% CI].

Table 4.

Causes for non-persistence of TNFi within the first year of treatment (n = 39).

Cause for non-persistence	n (%)
Primary non-response*, without ADAb**	14 (36)
Primary non-response, with ADAb	1 (3)
Loss of response, without ADAb	4 (10)
Loss of response, with ADAb	2 (5)
Side effect, without ADAb	8 (20)
Side effect, with ADAb	5 (13)
Addition of second biological agent	5 (13)

*Non-response within 3 months after treatment start.

**Anti-drug antibody.

Adverse events

A total of 19 serious adverse events were recorded in 18 patients. Of these, infections (n = 8), infusion reactions (n = 4), and exacerbations of colitis (n = 2) were the most common. The remaining events were single occurrences of azathioprine toxicity, methotrexate-induced hepatitis, squamous cell carcinoma, deep vein thrombosis and an assumed non-drug related neurological event that presented with upper extremity paralysis.

Drug levels and dosage adjustments

When applying the cut-off values of AGA, 73.2% (n = 82) and 83.3% (n = 85) of the patients reached a target drug level at three and twelve months, respectively. The percentages of patients with a target drug level at twelve months for the respective TNFi are shown in Figure 1. The difference in achievement of target drug levels between TNFi did not reach statistical significance at three (p = 0.8) or twelve (p = 0.08) months.

Figure 1.

Percentages of patients with a serum drug level above AGA recommendations at 12 months. Serum drug levels were measured routinely at infusion and planned outpatient visits. Target drug levels according to the AGA recommendations were ≥7.5 mg/L for adalimumab (ADA) and ≥5 mg/L for infliximab (IFX). The figure shows the percentages of patients at target drug level according to TNFi agent after 12 months of treatment (ADA vs IFX, p = 0.08).

Serum drug level above cut-off at three months was associated with clinical remission at twelve months (OR= 2.97, 95% CI [1.24–7.12]), biochemical remission at twelve months (OR= 2.64, 95% CI [1.03–6.77]), but not with combined remission (OR= 1.84, 95% CI [0.72–4.76]). For patients in clinical remission and non-remission (including TNFi non-persistent patients and patients started on a second biologic agent) at twelve months, median IFX trough levels at three months were 9.1 mg/L (IQR 5.4–17.3) and 6.3 mg/L (IQR 2.4–13.9), respectively (p = 0.1). For patients treated with ADA, the corresponding serum drug levels at three months were 12.2 mg/L (IQR 8.5–17.7) and 7.9 mg/L (IQR 5.3–15.8) (p = 0.04). The corresponding serum levels for patients in biochemical remission and non-remission were 12.9 mg/L (IQR 10.2–18.1) and 7.9 mg/L (5.1–15.8) (p = 0.049) for the ADA group. There was no significant difference in serum drug levels at three months between patients on IFX that did or did not achieve the biochemical remission outcome at twelve months, nor for the combined remission outcome in either group. When including only TNFi persistent patients at twelve months, there were no associations between drug levels at three months and clinical remission (OR= 2.03, 95% CI [0.60–6.84]), biochemical remission (OR= 1.89, 95% CI [0.64–5.62]) or combined remission (OR= 1.15, 95% CI [0.38–3.48]) at twelve months. A serum drug level above cut-off at twelve months was not associated with any remission outcomes at twelve months. To examine if the lack of association was due to the chosen cut-off, a set of arbitrarily chosen serum levels within an adequate clinical range was applied to the regression analyses (for ADA: 5.0, 6.5, 9.0, 10.0 and for IFX: 3.0, 5.0, 9.0). These additional analyses did not reveal any association (data not shown).

In total, 125 dosage adjustments were recorded. Of these, 70 (56%) were classified as adjustments related only to serum drug levels, 39 (31%) were related to clinical factors only, and 16 (13%) were related to both clinical factors and drug levels. Figure 2 shows percentages of patients with dosage adjustments stratified by TNFi. There were significantly more adjustments recorded for patients receiving IFX (p = 0.02). We found no associations between a number of dosage adjustments and biochemical, clinical or combined remission.

Figure 2.

Percentages of patients according to number of dosage adjustments by treatment. The number of adjustments of dose or interval between treatments was collected from the patient chart. The figure shows that 54% of patients that received adalimumab (ADA) had no dosage change, while 72% of patients on infliximab (IFX) underwent one or more changes of their dosage (ADA vs IFX, p = 0.02).

Discussion

In this single centre, retrospective study following 141 UC patients during the first twelve months after start of TNFi treatment, the rates of clinical, biochemical and combined remission were 58%, 42% and 36% at twelve months, respectively. There were few serious adverse events. Patients were followed according to a proactive TDM strategy and target serum drug levels were reached in 83% of the patients at twelve months. Notably, 56% of all dosage adjustments were in response to serum drug levels and in absence of clinical rationale for adjustments. Target serum drug level at three months was associated with clinical and biochemical remission at twelve months. The STRIDE II criteria for clinical and biochemical remission were not met for one-fifth and two-fifths of the drug persistent patients, respectively.

Our data revealed statistically significant associations between target drug level at three months and clinical and biochemical remission at twelve months. This is in accordance with the PANTS study where drug concentrations at week 14 were independently associated with non-remission at week 52 for patients with Crohn’s disease starting TNFi therapy [17]. To the extent of our knowledge, the present association is a novel finding in a UC cohort. However, several studies have shown an association between high serum drug levels and remission outcomes at a given time point [13, 18], which was not demonstrated in our cohort. A possible explanation might be that a subset of patients with persistent active disease received increasing doses of TNFi in an attempt to reach remission. This might result in higher serum drug levels in this group of patients, which masks an association between serum drug levels and remission in the group of responders. Our data could not demonstrate a significant association between target serum drug level at three months and remission outcome at twelve months when considering only treatment persistent patients. This could conceivably be due to a small sample size at twelve months, and a similar study of a larger cohort would be interesting to investigate this further. However, the association between non-achievement of target serum drug level at three months and treatment failure defined as drug discontinuation, the addition of a second biologic agent or non-remission at twelve months, highlights a potential for proactive TDM to improve outcomes in TNFi treatment.

In our cohort, only six percent of the patients discontinued TNFi in combination with ADAb formation. A 2021 systematic review and meta-analysis found that ADAb-development occurred in 28% of patients on IFX monotherapy and 16% in patients on concomitant treatment with an immunomodulator [19]. In the same review, the corresponding numbers for ADA were about eight and two percent. Associations between ADAb-presence and unfavourable clinical outcomes as well as increased incidence of adverse events have previously been reported [20]. The NORDRUM-B study found that a lower percentage of patients on IFX therapy followed with a proactive TDM strategy developed clinically significant levels of ADAb than patients followed with standard therapy without trough level considerations (9% vs 15%) [21]. Tournier and colleagues observed 54 IBD patients on IFX treatment, of whom 21 discontinued treatment during the 24 months follow-up due to immunogenic drug failure [22]. As the rates of remission in our cohort is comparable to other real-world cohorts, our study does not unequivocally favour proactive TDM. Although ADAbs were not measured routinely in the presence of a serum drug level > 1 mg/L, we consider the observed rates of ADAb formation leading to discontinuation to be low. We also observed that the majority of dosage adjustments were in response to low serum drug levels, resulting in a high proportion of patients with serum levels at target both after three and twelve months. Furthermore, the number of dosage adjustments was significantly higher for IFX than for ADA patients. This may be attributed to higher frequencies of visits in the IFX group than the ADA group due to frequent visits for infusions and thereby more frequent measurements of serum drug levels and tighter follow-up. As the achievement of target drug levels tended towards being significantly higher in the IFX group, the potential for TDM to elicit beneficial results might be higher for patients treated with IFX. The role of proactive TDM in TNFi treatment warrants further investigation, yet we consider our data to strengthen the hypothesis that a proactive effort to ensure target drug levels yields higher probabilities of achieving remission and avoiding drug failure due to anti-drug antibodies in a clinical setting.

When the current real-world data are evaluated in the context of the STRIDE II recommendations, an interesting discrepancy between the recommendations and real-world practice is revealed. In STRIDE II, clinical remission is an intermediate treatment target and change of treatment should be considered if this is not achieved. As the present data demonstrate, 28% and 19% of TNFi persistent patients did not meet the criterion of clinical remission after three and twelve months, respectively. Furthermore, in STRIDE II, improvement of FC to a pre-specified threshold is also recommended as an intermediate target, which is expected within about twelve weeks. Combining clinical remission and normalization of FC is equivalent to our definition of combined remission. Concurrently as many as 60% and 46% of the persistent patients in our cohort did not achieve this combined target after three and twelve months and were thus overdue for consideration for change of treatment when applying the 250 µg/g FC threshold. This threshold is subject to some uncertainty in the recommendations, and 150 µg/g is proposed as one alternative. Applying a 150 µg/g threshold in our data did not significantly change the proportion of patients achieving the STRIDE II remission criteria. The discrepancies between the observed practice and the STRIDE II recommendations might also be interpreted as a need for a more specific and reliable marker of inflammation to aid physicians in decisions regarding treatment. Nevertheless, according to our data, more strict adherence to the STRIDE II guidelines in real-world practice will require a marked shift in the treatment approach, and it might be questioned whether this could lead to overtreatment, excessive treatment changes and increased costs.

Strengths

Patients in the present study had a tight follow-up including measurement of serum drug levels, laboratory evaluations and frequent clinical assessments in an experienced and specialised IBD clinic. Patient care was provided by a dedicated and small group of physicians and nurses over time, thereby reducing potential heterogeneity in treatment strategies. TNFi was first-line biologic therapy for the entirety of the inclusion period, thereby ensuring that the study population remained comparable as TNFi-naïve.

Limitations

Data were in part collected from hospital EPJ, implying a risk of information bias in individual physician recording as well as investigator interpretation of variables. As the clinical follow-up of patients did not align completely with our study design with the collection of data at three and twelve months, there are missing values, especially at three months. Some of our reported confidence intervals display low precision, suggesting that the number of included patients could ideally have been higher. As this is a retrospective observational study where all patients were followed according to the same regime, the design did not allow for a control group, which would have been preferable to assess the potential benefit of TDM practice versus traditional follow-up.

Conclusion

This real-world study of UC patients on TNFi treatment supports the use of a proactive TDM strategy as the target drug level at three months was associated with remission outcomes at twelve months, and the rates of discontinuation due to anti-drug antibodies were low. The present data also suggest that adherence to the STRIDE II recommendations requires a major shift in the care of UC patients as a large proportion of patients that currently may be considered adequately treated, then would be considered for treatment change.

Funding Statement

The study was funded by the South-Eastern Norway Regional Health Authority.

Author contributions

JAL: Methodology, data collection, statistical analysis, interpretations of data, writing original draft, editing of manuscript. AWM: Methodology, supervision, interpretation of data and discussion of findings, editing of manuscript. MLH: Methodology, interpretation of data and discussion of findings, editing of manuscript. KA: Identification of eligible participants, interpretation of data, editing of manuscript. MCS: Supervision of statistical methodology, editing of manuscript. NB: Analysis of serum drug levels/anti-drug antibodies, methodology, editing of manuscript. DW: Analysis of serum drug levels/anti-drug antibodies, methodology, editing of manuscript. All authors have read and approved the manuscript.

Disclosure statement

JAL, MCS, DJW and NB report no conflicts of interest. AWM reports investigator initiated research grants from Takeda. MLH reports speaker fees from Galapagos, Ferring, BMS, Janssen, AbbVie, Meda, Tillotts and Takeda; advisory boards for Takeda, Galapagos, BMS and AbbVie; investigator initiated research grants from Takeda, Pfizer, Tillotts and Ferring. KA reports consultant fees from Takeda.

Data availability statement

The data that support the findings of this study are available from the corresponding author, JAL, upon reasonable request.