Influence of Pharmacokinetic-Pharmacodynamic Failure Mechanisms on Outcomes in Biologic Therapy Sequencing in Inflammatory Bowel Disease: A Retrospective Cohort Study

Abstract

Background

Increasing therapeutic options for inflammatory bowel disease calls for tools to aid choice of sequencing. We investigated if pharmacokinetic (PK) and pharmacodynamic (PD) failure mechanisms prompting therapy change influenced subsequent outcomes when switching to a different biologic drug class.

Methods

Retrospective single-center cohort study including patients treated first with tumor necrosis factor (TNF) inhibitors, followed by vedolizumab, and then ustekinumab. Clinical and objective disease remission were conventionally classified by validated indices. PK-PD failure was defined according to maintenance drug concentrations (PK below thresholds and PD above thresholds): infliximab 8.0 µg/mL, adalimumab 12.0 µg/mL, golimumab 1.4 µg/mL, vedolizumab 15 µg/mL. Primary treatment failure despite dose intensification was ascribed to PD. Primary endpoints were steroid-free treatment persistence and 1-year remission.

Results

The study included 112 patients switching from TNF inhibitors to vedolizumab (infliximab n = 61, adalimumab n = 32, golimumab n = 16, certolizumab pegol n = 3) and 31 subsequently to ustekinumab. Treatment persistence on vedolizumab did not differ between patients discontinuing TNF inhibitors due to PK (n = 28, 31%) or PD (n = 63, 69%) (mean 989 days [95% confidence interval: 554–1424] vs. 951 [659–1242], p = 0.93). One-year steroid-free clinical and objective remission rates on VDZ were also comparable between PK-PD groups (29% vs. 35%, p = 0.63 and 35% vs. 43%, p = 0.48, respectively). Findings for UST were similar. Sensitivity analyses with exclusion of primary non-responders and multivariate analyses correcting for potential confounders supported findings.

Conclusion

PK-PD failure mechanisms do not appear to influence subsequent treatment outcomes when switching to biologics with different modes of action. Sequencing may rather rely on aspects such as efficacy, safety, and costs.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40259-025-00730-4.

Key Points

Sequencing of advanced biologic therapies at treatment failure in patients with inflammatory bowel disease pose an unmet challenge to clinicians.

This study found that the underlying pharmacokinetic (PK) or pharmacodynamic (PD) failure mechanisms do not appear to influence subsequent treatment outcomes when switching between biologics with different modes of action.

Therapeutic drug monitoring has proven useful to identify patients with PK treatment failure likely to benefit from switching within biologic drug class before switching out of class. However, once switching out of class, the prognosis seems unrelated to PK-PD failure mechanisms prompting therapy change.

Introduction

The medical arsenal has undergone rapid expansion, fundamentally transforming the treatment of inflammatory bowel disease (IBD). Presently, advanced therapy is used as a collective term for biologics, including tumor necrosis factor (TNF) inhibitors, cell migration inhibitors, IL-12/23 and selective IL-23 inhibitors, as well as small molecules with diverse actions (JAK inhibitors, S1P modulators). Despite advancements, a significant subset of patients requires switching between agents due to treatment ineffectiveness or side effects [1]. Consequently, the optimal selection and sequencing of therapies are paramount to minimize periods of active disease, associated disability and quality-of-life impairment, and disease progression potentially resulting in irreversible bowel damage and surgery [2–4]. Moreover, each subsequent introduction of an advanced treatment typically diminishes treatment effectiveness [3, 4].

Therapeutic drug monitoring comprises measurements of drug concentrations and anti-drug antibody (Ab) formation and represents a readily applicable and widely available clinical tool to aid in sequencing of biologics [5]. It enables the classification of patients experiencing treatment failure due to underlying pharmacokinetic (PK) or pharmacodynamic (PD) mechanisms [6, 7]. This information can be used to rationally guide treatment choices, such as dose optimization or switching within or outside the drug class.

The objective of this study was to investigate whether subsequent treatment effectiveness after switching to a new biologic drug class was influenced by underlying PK-PD mechanisms prompting therapy change. Inspired by existing data, we hypothesized that patients experiencing PD failure, characterized by adequate drug exposure, would have superior outcomes when switching out of class because different disease pathways, theoretically, need targeting here; whereas  the principal disease pathway is already targeted in PK failure which is instead caused by insufficient drug levels to eliminate ongoing inflammatory activity due to sub-therapeutic exposure with or without anti-drug Ab formation [8–10].

Methods

Study Design and Patients

This was a retrospective cohort study at a tertiary Danish IBD center based on systematically collected, prospectively registered, pre-specified and protocolized, clinical, biochemical, objective, and therapeutic drug monitoring (TDM) data, obtained as part of routine clinical practice as previously outlined [11]. In brief, disease activity scores and biochemical analyses were done at every visit for intravenously administered biologics (4–8 weeks) and every 3 months for subcutaneously administered biologics, clinical consultations and fecal calprotectin every 6 months, and objective disease evaluations (e.g., endoscopy and magnetic resonance imaging) and TDM once a year. The national IBD treatment regulations at time of study dictated TNF inhibitors to be used as the first biologic drug class, followed by vedolizumab (cell-migration inhibitor) as the second-line drug class, and anti-IL12/23 therapy with ustekinumab as the third-line drug class. This research included all patients being switched from TNF inhibitors used as first-line biologic drug class to vedolizumab as second-line biologic drug class due to objectively confirmed active luminal disease until January 2021, with end of follow-up in January 2024. Data were collected from 2022 to 2024. Patients with acute severe ulcerative colitis, pouch, no luminal disease activity, and alternative sequency orders were excluded.

Approvals

Approval was obtained from the Ethics Committee (H-16020384) and Danish Data Protection Agency (2012-58-0004). The study was conducted in accordance with the Helsinki Declaration.

Endpoints

The primary objectives were to compare steroid-free treatment persistence as well as 1-year steroid-free clinical and objective remission rates on vedolizumab, and subsequently ustekinumab, among patients who had discontinued previous biologics due to PK or PD treatment failure mechanisms [6, 7].

Disease Activity

Clinical disease activity was routinely registered at all patient visits, and remission defined as Harvey-Bradshaw Index (HBI) < 5, Simple Clinical Colitis Activity Index (SCCAI) ≤ 2, or partial Mayo score (pMayo) ≤ 2 with no individual subscore > 1 and no rectal bleeding. Objective disease evaluations were routinely carried out once a year or at treatment failure. Objective disease assessments were done by endoscopy in two-thirds of the patients, and remission was defined as Simple Endoscopic Score for Crohn’s disease (SES-CD) ≤ 2 or Mayo endoscopic score ≤ 1. In the remaining patients, alternative objective measures were used such as fecal calprotectin (< 200 µg/g) or clinical imaging showing no signs of ongoing inflammatory activity. Primary treatment failure was defined as inadequate effect of induction therapy despite dose intensification followed by discontinuation of therapy, and secondary failure as loss of treatment effect during the maintenance phase resulting in discontinuation of therapy. Treatment failures were defined by the treating physician and supported by clinical disease activity scorings and objective disease activity evaluations. Patients receiving steroids at the time of biologic treatment initiation were routinely tapered according to local guidelines within 8 weeks for ulcerative colitis and 10 weeks for Crohn’s disease, and patients were classified as having treatment failure if systemic steroid treatment was reinitiated at any later timepoints during therapy.

Therapeutic Drug Monitoring

Following local guidelines, trough levels of TNF inhibitors (limit of detection (LOD) 0.65 µg/mL for all agents) and presence of anti-drug antibodies (Abs) were routinely measured once a year and at time of loss of response by functional reporter gene assays, plus additional measurements of anti-drug antibodies by a highly sensitive, drug-tolerant enzyme immune assay when drug concentrations were > LOD to minimize the risk of false negative measurements (Wieslab, Sweden) [12]. Clinical actions based on these results were defined by the treating physician. Trough vedolizumab concentrations (LOD 0.2 µg/mL) and presence of anti-vedolizumab Abs were measured as part of this study in biobanked samples routinely collected prior to each infusion and every 3 months during subcutaneous treatment from patients with loss of response using immunofluorometric assays (Department of Medical Biochemistry, Oslo University Hospital, Oslo, Norway), and results had not been available to the treating physicians [13].

Drug concentrations below thresholds associated with clinical-endoscopic remission were considered PK failure caused by either immunogenicity or non-immune mechanisms according to detection of neutralizing anti-drug Abs, whereas drug concentrations above thresholds represented PD failure. The primary analysis utilized following thresholds based on the literature: infliximab ≥ 8.0 µg/mL PD failure versus < 8.0 µg/ml PK failure [14, 15], adalimumab ≥ 12.0 µg/mL PD versus < 12.0 µg/mL PK [14, 15], golimumab ≥ 1.4 µg/mL PD versus < 1.4 µg/mL PK [16], and vedolizumab ≥ 15.0 µg/mL PD and < 15.0 µg/mL PK [13, 17]. Primary non-responders had routinely been dose optimized prior to discontinuation of therapy and were classified as having PD failure [6, 14, 18]. Sensitivity analysis using other thresholds to define PK or PD failure comprised (a) infliximab 5.0 µg/mL and adalimumab 7.5 µg/mL [19]; (b) median drug concentrations in the cohort (excluding patients with immune-mediated PK treatment failure caused by anti-drug Abs): infliximab 9.6 µg/mL, adalimumab 8.5 µg/mL, golimumab 2,6 µg/mL, vedolizumab 13.8 µg/mL.

Statistics

Descriptive data were presented as percentages or means with standard deviations (SDs) or standard errors of the mean (SEMs). Fisher’s exact test or Chi square was applied to compare discrete variables. Treatment persistence was analyzed by Kaplan Meyer statistics including Log Rank Test and multivariate Cox regression (Enter method), which included disease (Crohn’s disease vs. ulcerative colitis), disease duration, smoking, body mass index (BMI), total number of TNF inhibitors used, combined with either reason for stopping (primary non-response vs. loss of response vs. remission vs. adverse event) or mechanisms for failure (PK vs. PD). These variables were also contained in multivariate logistic regression analysis of 1 year remission outcomes. Data were analyzed using IBM SPSS 29. Two-sided p values < 0.05 were statistically significant.

Results

Study Population

In all, 131 patients received vedolizumab in the study period. As detailed in Fig. 1, a total of 19 patients were excluded (vedolizumab as prophylaxis while in remission n = 10, alternative sequencing order n = 5, fistulizing Crohn’s disease without luminal disease involvement n = 3, or pouchitis n = 1). The study population thus comprised 112 patients (Crohn’s disease n = 38, 34%; ulcerative colitis n = 74, 66%) initially treated with TNF inhibitors for active luminal disease followed by vedolizumab (interval between agents mean 303 days; SEM 58) (Fig. 1). A subset of 31 patients (Crohn’s disease n = 17, 55%; ulcerative colitis n = 14, 45%) subsequently switched from vedolizumab to ustekinumab (interval mean 455; SEM 110). Patient characteristics are detailed in Table 1.

Fig. 1.

Study population. The study population comprised patients with ongoing luminal active inflammatory bowel disease (IBD) initially treated with tumor necrosis factor (TNF) inhibitors followed by vedolizumab and with a subset subsequently treated with ustekinumab. Numbers in parentheses are percentages. btw between, IBD inflammatory bowel disease, PD pharmacodynamic, PK pharmacokinetic. ^&One TNF inhibitor utilized before switching to vedolizumab n = 62 (55%), two TNF inhibitors n = 44 (39%), three TNF inhibitors n = 4 (4%), four TNF inhibitors n = 2 (2%). *Infliximab n = 5, adalimumab n = 4, golimumab n = 1. **Infliximab n = 7, adalimumab n = 8; golimumab n = 3. *** Infliximab n = 9, adalimumab n = 2, golimumab n = 5. ^¤Missing n = 6 (adalimumab n = 3, golimumab n = 3)

Table 1.

Characteristics of the study population at vedolizumab initiation

Study population, no. (%)	112
Ulcerative colitis	74	(66)
Crohn’s disease	38	(34)
Female, no. (%)	56	(50)
Body mass index, mean (SEM)	25	(0.5)
Age at diagnosis, years, mean (SEM)	31	(1.3)
Disease duration, years, mean (SEM)	10	(0.8)
Smoking, no. (%)	16	(14)
Previous intraabdominal surgery, no. (%)	18	(16)
Ulcerative colitis Montreal classification, no. (%)
E1: proctitis	14	(19)
E2: left sided	21	(28)
E3: extensive	39	(53)
Crohn’s disease Montreal classification, no. (%)
Age at diagnosis
A1: < 17 years	5	(13)
A2: 17–40 years	25	(66)
A3: > 40 years	8	(21)
Location
L1: terminal ileum	4	(11)
L2: colon	9	(24)
L3: ileocolic	23	(61)
L3+L4: ileocolic + upper gastrointestinal	1	(3)
L4: isolated upper gastrointestinal	1	(3)
Behaviora
B1: Non-stricturing, non-penetrating	18	(47)
B2: Stricturing	14	(37)
B3: Penetrating	6	(16)
Extraintestinal manifestations, no. (%)	33	(30)
Concomitant conventional immunosuppressives, no. (%)	18	(16)
Disease activity VDZ initiation, score mean (SEM)
SCCAI (n = 53, 72%)	7	(0.5)
pMayo (n = 21, 28%)	6	(0.5)
HBI	10	(0.8)
Objective disease assessment, days before starting VDZ, mean (SEM)b	28	(3.7)
Biochemical parameters at VDZ initiation, mean (SEM)
Hemoglobin, mmol/L	8.3	(0.1)
Albumin, g/L	40	(0.4)
C-reactive protein, mg/L	9	(1.3)
Steriod bridging, no. (%)	34	(30)
Extra week 10 infusion, no. (%)	58	(52)
Number of infusions 1st year, mean (SD)c	7	(3)
Intensification of maintenance during 1st year, no. (%)	18	(16)

HBI Havey-Bradshaw Index, No. number, pMayo partial Mayo score, SCCAI simple clinical colitis activity index, SEM standard error of the mean, SD standard deviation, VDZ vedolizumab

^aPerianal involvement B1 n = 3, B2 n = 3, and B3 n = 5

^bEndoscopy (n = 103, 93%), fecal calprotectin (n = 90, 80%), bowel ultrasound (n = 14, 13%), magnetic resonance imaging (n = 12, 11%), capsule endoscopy (n = 3, 3%), and/or computed tomography (n = 1, 2%)

^cSeven patients switched from intravenous to subcutaneous VDZ within first year and without standardized timing of switching

Most patients initially had TNF inhibitor therapy with infliximab (n = 106, 95%) (Fig. 1). Among those in whom more than one TNF inhibitor was used (n = 50, 45%), two-thirds received adalimumab (n = 34, 68%) and one-third golimumab (n = 13, 36%). Few patients (n = 6, 5%) had ≥ 3 TNF inhibitors. The last TNF inhibitor before switching to vedolizumab was infliximab in 61 patients (55%), adalimumab in 32 (29%), golimumab in 16 (14%), and certolizumab pegol in three (3%). At this time, mean age was 38 years (SEM 1.3), disease duration 8 years (SEM 0.7), and treatment duration of last TNF inhibitor 370 days (SEM 57.2).

Therapeutic Drug Monitoring

Reasons for switching from TNF inhibitors to vedolizumab were predominantly primary (n = 47, 42%) or secondary treatment failure (n = 50, 45%) (Fig. 1). Likewise, the majority of patients subsequently switching from vedolizumab to ustekinumab had primary (n = 7, 23%) or secondary failure (n = 19, 61%). Biologic drug trough levels in individual patients stopping therapy due to secondary treatment failure are shown in Fig. 2 (infliximab mean 9.1 µg/mL, SEM 1.8; adalimumab 7.2 µg/mL, SEM 1.4; golimumab 2.3 µg/mL, SEM 0.7; vedolizumab 16.4 µg/mL, SD 2.4). In all, TNF inhibitor failure due to PD occurred in 63 patients and PK failure in 28. During vedolizumab therapy, 15 patients discontinued treatment due to PD failure and 11 because of PK failure. Neutralizing anti-drug Abs at the time of treatment failure were detected in ten patients (infliximab n = 5, adalimumab n = 3, golimumab n = 1, vedolizumab n = 1).

Fig. 2.

Drug levels in patients with treatment failure. Biologic drug concentrations at time of secondary treatment failure. Black bold lines mark the primary thresholds applied to differentiate pharmacokinetic (i.e., below thresholds) and pharmacodynamic failure mechanisms (i.e., above thresholds). Gray lines are mean with standard error of the mean in the study population

Treatment Persistence According to PK-PD Reasons for Switching

Vedolizumab

Steriod-free treatment persistence on vedolizumab did not differ significantly between patients having discontinued TNF inhibitors due to PK or PD (mean 989 days, 95% confidence interval (CI) [554–1424] vs. 951 [659–1242], p = 0.93) (Fig. 3). Sensitivity analyses of alternative thresholds to define PK or PD revealed similar results (Online Supplementary Material Figure). Multivariate Cox regression revealed a significantly shorter vedolizumab treatment duration in patients having discontinued TNF inhibitors due to adverse events (RR 0.19 [0.04–0.82], p = 0.03) but was not influenced by other variables including PK-PD mechanisms for failure.

Fig. 3.

Vedolizumab treatment duration. Steroid-free treatment duration on vedolizumab according to pharmacokinetic (PK) or pharmacodynamic (PD) reasons for switching out of biologic drug class from tumor necrosis factor (TNF) inhibitors to vedolizumab. Thresholds applied was infliximab 8.0 µg/mL, adalimumab 12.0 µg/mL, and golimumab 1.4 µg/mL. P value not significant

Ustekinumab

Among patients subsequently switching from vedolizumab to ustekinumab, treatment persistence was neither different between patients having initially discontinued TNF inhibitors due to PK or PD (1553 [997–2110]) vs. 1200 [762–1639], p = 0.42) (Fig. 4A), nor later discontinued vedolizumab due to PK or PD (1305 [832–1778] vs. 1323 [785–1862], p = 0.43) (Fig. 4B). Sensitivity analyses of alternative thresholds to define PK or PD both for TNF inhibitor and vedolizumab failure were non-significant and comparable to the above (data not shown). Multivariate Cox regression showed no associations between ustekinumab treatment duration and variables related to TNF inhibitor or vedolizumab therapy, including PK-PD mechanisms for failure.

Fig. 4.

Ustekinumab treatment duration. Steriod-free treatment duration on ustekinumab according to pharmacokinetic (PK) or pharmacodynamic (PD) reasons for switching out of biologic drug class from (A) tumor necrosis factor (TNF) inhibitors and (B) vedolizumab. P values not significant

Supplementary Analyses

Explorative analyses of vedolizumab and ustekinumab treatment persistence with exclusion of primary non-responders were consistent with the non-significant findings above (data not shown). Explorative analyses of potential influence of PK-PD failure mechanisms of all TNF inhibitors used prior to switching to vedolizumab and ustekinumab were also similar (data not shown).

One-Year Treatment Outcomes According to PK-PD Reasons for Switching

Vedolizumab

One-year steroid-free clinical (29% vs. 35%, p = 0.63), objective (35% vs. 43%, p = 0.48), or clinical-objective remission rates (18% vs. 29%, p = 0.31) on vedolizumab were similar between patients having discontinued TNF inhibitors due to underlying PK or PD mechanisms (Fig. 5A).

Fig. 5.

One-year remission outcomes. One-year steroid-free clinical, objective, and combined clinical-objective remission on (A) vedolizumab according to pharmacokinetic (PK) or pharmacodynamic (PD) failure mechanisms on tumor necrosis factor (TNF) inhibitors; and ustekinumab according to PK or PD failure mechanisms on (B) initial TNF inhibitor therapy and (C) later vedolizumab therapy. Clinical remission was defied as Harvey-Bradshaw Index < 5, Simple Clinical Colitis Activity Index ≤ 2, or partial Mayo score ≤ 2 with no individual subscore > 1 and no rectal bleeding. Objective remission was defined by endoscopy (Simple Endoscopic Score for Crohn’s disease ≤ 2 or Mayo endoscopic score ≤ 1), fecal calprotectin (< 200 µg/g), or clinical imaging showing no signs of ongoing inflammatory activity

Ustekinumab

One-year clinical (55% vs. 57%, p = 1.00), objective (60% vs. 50%, p = 0.70), or clinical-objective remission rates (50% vs. 43%, p = 1.00) on ustekinumab were similar between patients having initially discontinued TNF inhibitors due to PK or PD mechanisms (Fig. 5B). Remission rates were also similar between patients having subsequently discontinued vedolizumab due to PK or PD (clinical remission 55% vs. 53%, p = 1.00; objective remission 55% vs. 53%, p = 0.70; clinical-objective remission 46% vs. 43%, p = 1.00) (Fig. 5C).

Supplementary Analyses

Sensitivity analyses using alternative thresholds to differentiate PK or PD failure mechanisms on TNF inhibitors or vedolizumab, respectively, revealed similar non-significant findings to the above (data not shown). Explorative analyses of vedolizumab and ustekinumab treatment outcomes with exclusion of primary non-responders were also similar (data not shown). Explorative analyses of potential influence of PK-PD failure mechanisms of all TNF inhibitors used prior to switching to vedolizumab and ustekinumab also showed similar results (data not shown). Logistic regression as detailed in Sect. 2, Methods, was non-significant except with lower likelihood for Crohn’s disease of clinical-objective remission on vedolizumab (RR 0.15 [0.04–0.65, p = 0.01) and ustekinumab (RR 0.09 [0.01–0.89], p = 0.04).

Discussion

With the advent of multiple, diversely acting, advanced medical treatment options for IBD, appropriate sequencing of biologics has become important. This study is the first to investigate whether PK-PD failure mechanisms influence treatment outcomes during subsequent biologic sequencing between different drug classes from TNF inhibitors to a cell migration inhibitor (vedolizumab) followed by an IL12/23-inhibitor (ustekinumab). Our results indicate that treatment persistence and clinical-objective outcomes following the switch are not impacted by underlying PK-PD mechanisms prompting the change in therapy. These findings were consistent across multiple sensitivity and exploratory analyses suggesting robustness. Our data come from a real-life cohort with a low risk of selection bias since all patients, per national guidelines, were sequenced from TNF inhibitors to cell migration inhibitors (i.e., vedolizumab) and finally to IL12/23-inhibitors (i.e., ustekinumab). Furthermore, clinical, objective, and TDM data were systematically and prospectively collected according to local protocols [11]. However, our study did not address other orders of drug class sequencing. As treatment persistence may have been biased by limited treatment options available, 1-year steroid-free remission rates being similar across PK-PD failure mechanisms strongly support our observations. In keeping with the literature, we observed relatively low 1-year steroid-free remission rates (clinical 26%, objective 41%) indicating that vedolizumab may be less effective when used as a second-line biologic drug class following TNF inhibitors for treatment of IBD [20].

Despite these insights, our findings remain exploratory due to study limitations necessitating larger confirmatory studies. The sample size was small, particularly for patients treated with ustekinumab. Although data were prospectively registered, they were retrospectively collected. Patients with primary treatment failure had been dose-optimized prior to discontinuation of therapies and were therefore classified as having PD failure as drug levels were considered adequate to neutralize ongoing inflammation; this approach can be questioned [6, 14, 18, 21]. Notably, exploratory analyses excluding primary non-responders consistently revealed similar results [22]. Another limitation was lack of multiple TDM samples up to secondary treatment failure and discontinuation of therapy. Also, measurements of anti-TNF drug and anti-drug Abs (but not of vedolizumab) had in some cases been available to the treating physician which may have introduced bias; however, we had no specific guidelines for handling TDM results and any clinical consequences were according to the treating physician.

Currently, there is no gold standard method to differentiate PK or PD failure on any biologic agent. Thus, classification relies on drug concentrations and the thresholds applied for this distinction. In the primary analysis, we utilized commonly accepted thresholds reported in the literature [13–17]. However, these are not universally well defined and data are limited for vedolizumab. We addressed this limitation by performing several sensitivity analyses using both higher and lower thresholds, carefully selected based on the literature and the actual drug concentrations measured in our cohort. All analyses consistently showed no trends indicating impact of PK-PD failure mechanisms on subsequent biologic treatment outcomes, which was also backed up by multivariate analyses to rule out potential confounders. Previous estimates suggest that approximately one-third of patients with loss of response to TNF inhibitors have an underlying PD etiology, and our results align with this, supporting the applied thresholds [14]. Importantly, we used validated assays with clinically validated thresholds [13, 23–25].

Presumed PD failure is usually managed by switching out of drug classes to target another, theoretically, more relevant disease pathway. PK failure is, on the other hand, typically handled by optimizing the same agent at increased dosing in the absence of anti-drug Abs, and then switching within the same drug class assuming that it is preferable to exhaust all options within a given drug class with proven efficacy for the individual patient before changing to a new one, and because it is believed that the predominant disease pathway is being targeted in patients with PK failure [6, 7, 14, 26]. Our observations suggest that switching out of drug class at PK failure can be a viable and effective alternative with outcomes comparable to patients switching out of class due to PD failure. This may be relevant when seeking a more attractive efficacy or safety profile from a drug with a different mode of action, for patients prone to developing antibodies against multiple TNF inhibitors, and those already on a highly intensified regimens [27]. Factors such as cost and cost-effectiveness, preferred route of administration, and a need to target extraintestinal manifestations or concomitant immunoinflammatory disorders may also justify switching out of class at presumed PK failure. Although this approach is sometimes used clinically, this study is the first to provide scientific support of it.

Conclusion

TDM has proven useful to identify patients with PK treatment failure likely to benefit from switching within biologic drug class before switching out of class. However, once switching to a new biologic drug class, our findings indicate that the prognosis is unrelated to underlying PK-PD failure mechanisms prompting change in therapy. In particular, our study does not support inferior outcomes in patients with PK failure caused by insufficient drug exposure as compared to those with adequate drug exposure and PD failure caused by alternative disease pathways dominating the inflammatory phenotype. Thus, choice of a new biologic drug class during biologic sequencing at treatment failure should instead rely on aspects such as efficacy, safety, and costs.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Funding

Open access funding provided by Odense University Hospital. Measurements of vedolizumab and antibodies against vedolizumab were supported by an Investigator Initiated Research grant from Takeda Pharma A/S, a member of the Takeda group of companies (IISR-2019-103039).

Conflict of interest

CS: Lectures for Takeda, MSD and Janssen-Cilag; research grant from Takeda. JB: advisory board and/or education from Abbvie, Takeda, Janssen, Bristol Myers and MSD. RDL, PNP, JH: none.

Author contributions

Study concept and design: CS. Acquisition of data: CS + RDL + PØP + JH. Analysis of data: CS. Interpretation of data: All authors. Drafting of the manuscript: CS. Critical revision of the manuscript: All authors. Final approval: All authors.

Data availability

Data are available in the article. Raw data are not publicly available.

Ethics approval

Ethics approval was received from the Ethics Committee (H-16020384) and Danish Data Protection Agency (2012-58-0004).

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Code availability

Not applicable.