The impact of biological interventions on health-related quality of life in adults with Crohn's disease: a systematic review with meta-analysis

Summary

Background

Health-related quality of life (HRQoL) is one of the main issues in Crohn's disease (CD). We systematically assessed the effects of a biological treatment on HRQoL in adults with CD.

Methods

We searched MEDLINE, EMBASE, and CENTRAL from inception to 26 Jul 2024 for randomised controlled trials (RCTs) investigating the effects of the biological agents approved for CD on HRQoL-related outcomes. Changes in IBDQ (Inflammatory Bowel Disease Questonnaire), SF-36 (The 36-Item Short Form Survey) and EQ-5D scores were evaluated using random-effects meta-analyses to determine if they met a clinically meaningful improvement (CMI). Cochrane's Risk of Bias tool 2 and Grading of Recommendations Assessment, Development, and Evaluation (GRADE) were applied to assess the quality and certainty of findings. PROSPERO registration: CRD42024573408.

Findings

Twenty-six RCTs involving 9013 participants were included in analysis. A clinically meaningful IBDQ score improvement was observed for infliximab (mean difference (MD) of 16.09 (95% CI: −0.13 to 32.31; 4 studies; 1162 participants, very low-certainty evidence)) and upadacitinib (MD 20.65, 95% CI 7.04–34.25; I² = 42.8%, P = 0.17; 3 studies; 1523 participants; high-certainty evidence). No significant IBDQ improvement was observed for adalimumab, certolizumab pegol, ustekinumab, or natalizumab compared to placebo (low-to high-certainty evidence). Due to considerable heterogeneity, data on vedolizumab and risankizumab could not be pooled.

Interpretation

Methodological issues in HRQoL measurement, including the need for larger sample sizes, standardized reporting, and uniform participant characteristics, contribute to the low methodological quality of current evidence on the impact of biological agents on HRQoL in CD. There is a large unmet need to investigate the association between clinical outcomes and HRQoL outcomes more thoroughly.

Funding

None.

Research in context.

Evidence before this study

Previous systematic reviews investigating HRQoL (health-related quality of life) outcomes were limited in: patient population (mainly included inflammatory bowel disease patients as summa of Crohn's disease and ulcerative colitis patients), treatment (analyses of infliximab, adalimumab, certolizumab, and natalizumab); outcome measures (IBDQ (Inflammatory Bowel Disease Questonnaire), SF-36 (The 36-Item Short Form Survey), and databases (MEDLINE. EMBASE) included. Results were shown descriptively and in terms of statistical significance, not of clinical importance. A statistically significant improvement in HRQoL outcomes were described for infliximab, adalimumab and certolizumab compared to placebo and the opposite results exists for the natalizumab in previous reviews.

Added value of this study

This study showed that six out of eight approved biological treatments for CD failed to achieve a CMI in HRQoL. The review highlighted a discrepancy between clinically efficacy and HRQoL outcome results and discussed it in detail in terms of methodology involved and statistical issues observed in included studies.

Implications of all the available evidence

Our findings showed a discrepancy between the objectively proven effectiveness and HRQoL in biological treatment. Improving the study design of trials to investigate HRQoL outcomes, including participants’ withdrawal data and standardised reporting is essential. This work represent a call to action to harmonize HRQoL data in this field as a cornerstone for future medical practice and policy-sharing decisions.

Introduction

Crohn's disease (CD) is a relapsing-remitting transmural inflammatory bowel disease (IBD). Fatigue, abdominal pain, prolonged diarrhea, weight loss, and fever, with or without gross bleeding, are the hallmarks of CD.¹ It is estimated that IBD affects between one in 200 and one in 300 people in high-income countries. In middle and low-income countries, prevalence is lower, but the rapidly increasing incidence in these countries leads to an ever-growing burden of health care costs soon.^2,3 To clarify, between 1990 and 2019, the projected number of cases of IBD worldwide increased by 47.45%, from 3.32 million (95% UI 2.90–3.79) in 1990 to 4.90 million cases (4.35–5.50) in 2019.⁴ Additionally, indirect costs associated with CD-related work disability accounted for 28% of the total societal cost of CD in the US and up to 64%–69% of the total societal cost in Europe.⁵ The total societal cost is estimated to be equivalent to more than 15 billion euros in Europe and the US.^6,7

The main goal of medical management is to stop the underlying inflammatory process and, therefore, influence patients’ long-term outcomes. Biologic interventions can achieve mucosal healing and deep clinical remission, resulting in a reduced need for surgery, lower hospital admission rates, and increased steroid-sparing.8, 9, 10, 11, 12 Consequently, a downward trend in age-standardized death as well as disability-adjusted life years (DALYs), which represent the sum of years of life lost and years of healthy life lost due to a disease or health condition, was observed worldwide.^4,13,14

CD negatively impacts the health-related quality of life (HRQoL) of patients. It is significantly lower in people with CD compared to the general population.^6,15 By modifying the disease course, biologicals may have a substantial effect on HRQoL and therefore would be beneficial to be introduced earlier in the treatment.¹⁶

Patient-reported outcome measures (PROMs) are instruments (e.g., IBDQ, SF-36, EQ-5D) used to assess patient-reported outcomes (PROs), which reflect patients' perceptions of their health, quality of life (QoL), or functional status in relation to healthcare interventions.17, 18, 19, 20, 21 Furthermore, QoL is often examined as a secondary or ancillary outcome in clinical trials, which typically results in poor data quality and diminishes its perceived importance in patients' lives, with only a few exceptions in recent studies. Consequently, PROMs like the Inflammatory Bowel Disease Questionnaire (IBDQ), The 36-Item Short Form Survey (SF-36) or EQ-5D are frequently used to measure QoL but rarely serve as long-term efficacy outcome measures in CD studies.²² The IBDQ comprises 32 questions covering bowel symptoms (e.g., abdominal pain), systemic symptoms (e.g., fatigue), social function (e.g., work attendance), and psychic function (e.g., depression). A total IBDQ score can range from 32 (very poor HRQoL) to 224 (perfect HRQoL).²³ The SF-36 is a 36-item questionnaire divided to psychical component summary (PCS) and mental component summary (MCS) subscores that assessed restrictions in physical, social, and role activities due to health problems, restrictions to role activities due to emotional problems, bodily pain, general mental health, vitality and general health perceptions.²⁰ The EQ-5D consists of five questions related to the five dimensions: the subject's ability to move, self-care, daily activities, pain or discomfort, and psychological condition.¹⁹ From the patient's perspective, a meaningful change in the IBDQ is defined by the clinically meaningful improvement (CMI), which corresponds to an increase of ≥16 points from baseline. An increase of 5 point in every SF-36 subscores, PCS and MCS, is considered a CMI for SF-36, and a decrease of at least 0.3 points is a CMI for EQ-5D. The lower the EQ-5D score change, the more preferable it is for the patients in terms of QoL.24, 25, 26

A Selecting Therapeutic Targets in Inflammatory Bowel Disease (STRIDE) Initiative acknowledged restoration of QoL and reduction in disability as formal long-term treatment targets, irrespective of other objective markers of inflammation. Indeed, they recommend to consider changing treatment if this target has not been achieved.²⁷ A review that deals with this issue in detail, accurately, and comprehensively is urgently needed as a starting point for changes in this research area. Therefore, this study aimed to systematically assess the effects of the biologic treatment on HRQoL in people with CD.

Methods

This systematic review with meta-analyses adhered to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines.²⁸

Search strategy and selection criteria

The databases MEDLINE, EMBASE, CENTRAL, and ClinicalTrials.gov (www.clinicaltrials.gov), World Health Organization International Clinical Trials Registry Platform (WHO ICTRP; https://trialsearch.who.int/), were searched from inception to 26 July 2024 for randomized controlled trials (RCTs) investigating the effects of the biological agents, approved for CD, on HRQoL-related outcomes. The search strategy is available for all databases in the Supplementary Material. Two types of analysis were performed. Random-effects meta-analyses were used to evaluate the IBDQ and SF-36 score changes based on a score changes and CMI. A qualitative analysis was conducted to assess the quality and certainty of the findings, using the Cochrane's Risk of Bias tool 2 and Grading of Recommendations Assessment, Development, and Evaluation (GRADE).²⁹ All studies identified by the literature search were independently screened for eligibility by two review authors (MSB and VG) based on the inclusion criteria:

• 1.Type of studies–RCTs assessing the impact of biologicals on HRQoL in people with CD irrespective of publication status, language, blinding procedure or design type of studies (induction and maintenance).
• 2.Type of participants–adults with CD as defined by a combination of clinical, biochemical, radiological, endoscopic, and histological criteria³⁰
• 3.Types of interventions–biologicals in doses specifically approved by European Medicine Agency (EMA) and Food and Drug Administration (FDA) for the treatment of CD have been considered for evaluation (e.g., infliximab, adalimumab, natalizumab, vedolizumab). The interventions were compared to placebo or an active comparator (e.g., corticosteroids, azathioprine, 6-mercaptopurine, or methotrexate).
• 4.Type of outcomes–change in HRQoL outcome (IBDQ, SF-36, and EQ-5D assessment tool).

If the HRQoL outcome was planned to be assessed, we included this study in our evaluation, regardless of available results. We obtained full-text reports when studies appeared to satisfy the inclusion criteria based on the title and abstract screening, or when information was insufficient to allow for a decision. Any disagreements were resolved by consensus or by referring to other review authors (AJK, AB, VVP). The number of studies identified, excluded and included, was reported according to the PRISMA flow diagram (Fig. 1).²⁸ We used data that have been made available through Vivli for studies GAINT,³¹ CLASSIC I,³² CLASSIC II,³³ CHARM,³⁴ ENACT-1,³⁵ ENCORE,³⁵ data based on the YODA project #2018-3743 for studies ACCENT I,³⁶ ACCENT II,³⁷ CERTIFI³⁸; IM-UNITI³⁹; UNITI-1³⁹; UNITI-2,³⁹ and through www.ClinicalStudyDataRequest.com for studies PRECISE 1,⁴⁰ PRECISE 2,⁴¹ GEMINI 2,⁴² GEMINI 3.⁴³ Those three independent platforms allowed us to include unpublished source data for our analyses when the published results were insufficient or partially described. Data extraction was carried out independently by three authors (MSB, VG, AJK). The following data were extracted from the original reports: title, authors, journal, year, publication status, study information (design, risk of bias items, power calculation, type and dose of a medication, administration intervals, comparator, inclusion and exclusion criteria, total number screened and randomized, baseline characteristics (age, sex, race, disease activity), length of follow-up, withdrawals, and loss to follow-up, and primary and secondary outcomes). Any discrepancies between published versions were highlighted and resolved via discussion and involving other authors if necessary.

Fig. 1.

PRISMA flow diagram of search results.

Statistics

For continuous outcomes, we calculated the mean difference (MD) and corresponding 95% confidence interval (95% CI) to analyze the effect of different biologicals on HRQoL. We consider a MD based on IBDQ, PCS and MCS of SF-36, EQ-5D score changes from baseline in the intention-to-treat population from the original studies. We assessed the distribution of data by calculating the highest possible value minus the observed mean, and dividing this by the SD.⁴⁴ The heterogeneity was assessed across the included studies using the visual inspection of the forest plot and Chi² test. The level of statistical significance was set to 0.05. The degree of heterogeneity, as a percentage of total variation across trials that results from heterogeneity rather than chance, was described using the I² statistic. The I² statistic was interpreted as follows: 0%–50% might not be important; 50%–75% may represent substantial heterogeneity; 75%–100% represents considerable heterogeneity. In a case of considerable heterogeneity, we discussed possible sources. We used a random-effects model to pool data, regardless of the presence of heterogeneity. The reporting bias was assessed by comparing the outcomes listed in study protocols to those reported in the published studies. If protocols were not available, we compared the outcomes specified in the methods section of the published report to those reported in the results section of the manuscript. Location and language bias have been addressed by searching multiple databases, including non-English language journals. Further information on risk of bias assessment can be found in the Supplementary Material, Supplemental Figure S1.

Data from individual trials were amalgamated for meta-analysis when the interventions, patient populations, and outcomes were sufficiently similar (determined by consensus). If a considerable degree of heterogeneity is detected, data have not been pooled for meta-analysis. We conducted all meta-analyses using a random-effects model with the Hartung-Knapp-Sidik-Jonkman (HKSJ) adjustment to account for uncertainty in the estimation of between-study variance and to ensure more accurate confidence intervals. The between-study heterogeneity was estimated using the REML (Restricted Maximum Likelihood) method.⁴⁵ We explored potential explanations for heterogeneity using a sensitivity analysis, excluding any obvious outliers upon visual inspection of the forest plot. The subgroup analyses were performed to investigate substantial heterogeneity. Subgroup analyses included: different doses of the biological drugs, and induction vs. maintenance studies.

We used the GRADE profiler software (http://gradepro.org/), RevMan Web and “meta” package in R (version 4.4.3.), to analyze data. Evidence from RCTs were evaluated considering (1) risk of bias, (2) indirectness of evidence, (3) inconsistency (i.e. unexplained heterogeneity), (4) imprecision of effect estimates and (5) publication bias.

Sensitivity analysis

We conducted available case analysis for missing continuous outcomes without making assumptions about participants with missing data. No studies available only in abstract form were included in this review, as we obtained complete data for all included studies from both published and unpublished sources. All studies included in this review were sponsored by pharmaceutical companies. Sensitivity analyses excluding abstract-only and sponsored studies were not performed. In cases of substantial heterogeneity, we performed subgroup analyses where possible (different doses of the biological drugs and induction vs. maintenance studies) and described results in the text.

The initial protocol was registered via Cochrane platform.⁴⁶ A subpart of this Cochrane project was registered in PROSPERO (https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=573408) and results are presented here. The differences between Cochrane protocol and PROSPERO project are: we initially planned to include data concerning adverse outcomes from non-RCTs and to include studies assessing all recognised biological interventions for the treatment of CD, irrespective of FDA or EMA approval status. We excluded data from all non-RCTs as including observational data may have introduced unnecessary bias and we limited our evaluation to biological interventions that are EMA or FDA approved for the indication of CD. We did not include biologicals that are still being clinically tested as our main goal was to give a comprehensive and reliable overview of the most frequently used HRQoL outcome.

Ethics

Ethical approval and informed consent of participants were not required for this meta-analysis.

Role of the funding source

There was no funding source for this study.

Results

The search strategy conducted on 26 July 2024 resulted in 4673 records. After screening titles and abstracts, 1477 duplicates and 3005 records were excluded based on the criteria of indication (not CD), study design (e.g., open label), intervention (not biologicals), or outcome (not investigated HRQoL). The final search resulted in 191 full-text studies assessed for eligibility. We have excluded 22 studies, two studies are ongoing, and four are awaiting classification (Supplementary Table S1: Characteristics of excluded, ongoing, and awaiting studies). Finally, 26 studies (77 records) were included in the analysis. All included studies are described in detail and assessed against the risk of bias (Supplementary Material: Assessment of risk of bias in included studies). The search results are presented in the PRISMA flow diagram (Fig. 1).

All included studies were multicenter, randomized, double-blind, placebo-controlled clinical trials conducted between June 1995 and Jan 2022, except two studies.^47,48 The FORTIFY and ENDURE studies are still ongoing, considering the open-label extension part, but data on the HRQoL assessment have been published.^48,49 In all included studies, sample size estimation was done for primary outcomes defined in those RCTs. Assessment of HRQoL was not the primary outcome in any of the included studies. Biological agents analyzed based on data from included studies in this review were:

• 1.anti-TNF agents: IFX, adalimumab, and certolizumab pegol
• 2.anti-interleukin 12/23 ustekinumab
• 3.anti-integrin α4β1/α4β7 natalizumab
• 4.anti-integrin α4β7 vedolizumab
• 5.anti-interleukin 23 risankizumab
• 6.selective JAK inhibitor upadacitinib.

Herein, we presented results on IBDQ score changes across different biological interventions (also see Table 1, Supplementary Table S2). Additionally, results involving SF-36 and EQ-5D assessment tool are shown in the Supplementary Material and Supplemental Figures S2–S8.

Table 1.

Characteristics of included studies.

Study title	Type of the study	Time of follow-up	Participants randomized (No)	Intervention	Sponsor	Age, y, mean (SD)	No. Male/Total No. (%)	Baseline CDAI Mean (SD) placebo/intervention	HRQoL assessment tool	Baseline IBDQ Mean (SD)
SONIC study47 NCT00094458	Induction	30 weeks	508	1. infliximab 5 mg/kg at week 0, 2, 6, 14, and 22 plus placebo (n = 169); 2. Azathioprine 2.5 mg plus placebo at week 0, 2, 6, 14, and 22 (n = 170)	Centocor, Inc. sponsorship	37.1 (11.9)a	326 (38.2)	313.1 (62.0)	IBDQ	122 (44–205)b
Lichtenstein 200250	Induction	4 weeks	108	1. infliximab 5 mg/kg (n = 27); 2. infliximab 10 mg/kg (n = 28); 3. infliximab 20 mg/kg (n = 28); 4. placebo (n = 25)	Centocor, Inc. sponsorship	38.5 (11)/39.3 (10.6)	15 (69)/13 (46)	288 (54)/318 (59)	IBDQ	128 (29)/116 (23)
ACCENT I36 NCT00207662	Maintenance	54 weeks	573, but only 2-weeks responders (n = 335) were assessed for primary endpoints and quality of life scores	1. infliximab 5 mg/kg at week 2 and 6, then every 8 weeks (n = 113); 2. infliximab 5 mg/kg at week 2 and 6, followed by 10 mg/kg every 8 weeks (n = 112); 3. Placebo at week 2 and 6, then every 8 weeks (n = 110)	Centocor, Inc. sponsorship	37 (12)a	130 (38.8)	304 (51)	IBDQ, SF-36	130 (21)d
ACCENT II37 NCT00207766	Maintenance	54 weeks	195	1. infliximab 5 mg/kg at weeks 14, 22, 30, 38, and 46 (n = 96); 2. Placebo at week at weeks 14, 22, 30, 38, and 46 (n = 99)	Centocor, Inc. sponsorship	36 (29–46)/37 (28–47)b	48 (48)/53 (55)	189 (96)/193 (109)	IBDQ, SF-36	164 (132–196)/158 (124–192)
CHARM34 NCT00077779	Maintenance	60 weeks	778	1. Adalimumab 40 mg every other week (n = 260); 2. Adalimumab 40 mg weekly (n = 257); 3. Placebo group (n = 261)	Abbott Laboratories, Abbott Park, IL, USA	37.1 (11.9)	326 (38.2)	313.1 (62)	IBDQ	122 (44–205)c
GAINT31 NCT00105300	Induction	4 weeks	325	1. Adalimumab 160 mg at week 0 and 80 mg at week 2 (n = 159); 2. Placebo at week 0 and week 2 (n = 166)	Abbott Laboratories, Abbott Park, IL, USA	37 (12)/39 (12)	65 (39)/50 (31)	313 (66)/313 (58)	IBDQ	124 (28)/120 (27)
CLASSIC-I32 NCT00055523	Induction	4 weeks	299	1. Adalimumab 40 mg at week 0 and 20 mg at week 2 (n = 74); 2. Adalimumab 80 mg at week 0 and 40 mg at week 2 (n = 75); 3. Adalimumab 160 mg at week 0 and 80 mg at week 2 (n = 76): 4. Placebo at weeks 0 and 2 (n = 74)	Abbott Laboratories, Abbott Park, IL, USA	37 (13)/39 (11)	37 (50)/36 (47)	296 (60)/295 (52)	IBDQ	131 (52–200)/127 (37–192)b
CLASSIC-II33 NCT00055497	Maintenance	56 weeks	55	1. Adalimumab 40 mg every other week (n = 19); 2. Adalimumab 40 mg weekly (n = 18); 3. Placebo group (n = 18)	Abbott Laboratories, Abbott Park, IL, USA	36 (13)/38 (10)	6 (33.3)/9 (50)	107 (62)/88 (50)	IBDQ	191 (138–224)/200 (138–216)b
PRECISE 140 NCT00152490	Maintenance	26 weeks	660	1. Certolizumab pegol 400 mg at weeks 0, 2, and 4 and then every 4 weeks up to week 24 (n = 331); 2. Placebo at weeks 0, 2, and 4 and then every 4 weeks up to week 24 (n = 329)	UCB Pharma, the German Federal Ministry of Education and Research Competence Network Inflammatory Bowel Disease, the National Center for Research Resources	38 (12)/37 (12)	131 (409/157 (47)	297 (62)/300 (64)	IBDQ, SF-36	unknown
PRECISE 241 NCT00152425	Maintenance	26 weeks	428	1. Certolizumab pegol 400 mg at weeks 0, 2, 4 and then every 4 up to week 24 weeks (n = 216); 2. Placebo at weeks 0, 2, 4 and then every 4 weeks up to week 24 (n = 212)	UCB Pharma, the German Federal Ministry of Education and Research Competence Network Inflammatory Bowel Disease	38 (12)/38 (11)	109 (52)/92 (53)	301 (62)/306 (61)	IBDQ, SF-36	122.6 (28.44)a
CERTIFI38 NCT00771667	Induction and maintenance phase	Induction: 8 weeks and maintenance 36 weeks	526	Induction phase 1. Ustekinumab 1 mg/kg at week 0 (n = 131); 2. Ustekinumab 3 mg/kg at week 0 (n = 132); 3. Ustekinumab 6 mg/kg at week 0 (n = 131); 4. Placebo at weeks 0, 2, and 4 and then every 4 weeks up to week 24 (n = 132) Maintenance phase with a follow-up of 36 weeks 1. Ustekinumab 90 mg (n = 72); 2. Placebo (n = 73)	Janssen Research and Development	39.5 (13.1)/38.8 (12.6)	64 (48.5)/153 (38.8)	312.4 (64.2)/327.7 (64.6)	IBDQ, SF-36	116a, b
IM-UNITI39 NCT01369355	Maintenance	44 weeks	397	1. Ustekinumab 90 mg every 8 weeks (n = 132); 2. Ustekinumab 90 mg every 12 weeks (n = 132); 3. Placebo through week 40 (n = 133)	Janssen Research and Development	39.5 (12.7)/38.6 (13.7)	59 (44.4)/114 (43.18)	319 (60.8)/313.1 (58)	IBDQ, SF-36	163.6 (31.76) / 170.5 (29.33)
UNITI-139 NCT01369329	Induction	8 weeks	741	1. Ustekinumab 130 mg at week 0 (n = 245); 2. Weight-range based ustekinumab doses approximating ustekinumab 6 mg/kg at week 0 (n = 249); 3. Placebo at week 0 (n = 247)	Janssen Research and Development	37.3 (11.8)/37.3 (12.5)	118 (57.8)/199 (40.28)	319 (59.7)/327.6 (62)	IBDQ, SF-36	120.0 (29.27)/118.2 (26.64)
UNITI-239 NCT01369342	Induction	8 weeks	628	1. Ustekinumab 130 mg at week 0 (n = 209); 2. Weight-range based ustekinumab doses approximating ustekinumab 6 mg/kg at week 0 (n = 209); 3. Placebo at week 0 (n = 210)	Janssen Research and Development	40.2 (13.1)/38.4 (13.1)	99 (47.1)/194 (46.41)	319.1 (60.8)/313.1 (58)	IBDQ, SF-36	122.7 (31.32)/122.8 (31.62)
ENACT-135 NCT00032786	Induction	12 weeks	905	1. Natalizumab 300 mg at weeks 0, 4, and 8 (n = 724); 2. Placebo at weeks 0, 4, and 8 (n = 181)	Elan Pharmaceuticals, San Diego, Calif, Biogen Ide,c, Cambridge, Mass, the main investigator Dr. Sandborn was awarded a grant from the National Institutes of Health	39 (14)/38 (12)	73 (40)/311 (43)	303 (65)/302 (60)	IBDQ, SF-36	unknown
ENACT-251 NCT00032799	Maintenance	60 weeks	339	1. Natalizumab 300 mg every four weeks from week 12 through 56 (n = 168): 2. Placebo every four weeks from week 12 through 56 (n = 171)	Elan Pharmaceuticals, San Diego, Calif, Biogen Ide,c, Cambridge, Mass, the main investigator Dr. Sandborn was awarded a grant from the National Institutes of Health	37 (12)/37 (13)	59 (35)/77 (46)	118 (57)/105 (54)	IBDQ	unknown
ENCORE study35 NCT00078611	Induction	12 weeks	509	1. Natalizumab 300 mg at weeks 0, 4, and 8 (n = 259); 2. Placebo at weeks 0, 4, and 8 (n = 250)	Elan Pharmaceuticals, San Diego, Calif, Biogen Ide,c, Cambridge, Mass	37.7/38.1	102 (41)/105 (41)	299.5 (63.19)/303.9 (64.8)	IBDQ, SF-36	122.5 (28.44)/123.6 (31.06)
GEMINI 242 NCT00783692	Induction and maintenance	52 weeks	461	Induction phase: 1. Vedolizumab 300 mg at week 0 and week 2 (n = 220); 2. Placebo group (n = 148) Maintenance phase: 1. Vedolizumab every 8 weeks up to 52 weeks (n = 154); 2. Vedolizumab every 4 weeks up to 52 weeks (n = 154); 3. Placebo (n = 153)	Millennium Pharmaceuticals (Takeda Pharmaceuticals)	38.6 (13.2)/36.3 (11.6)	69 (46.6)/105 (47.7)	325 (78)/327 (71)	IBDQ, SF-36, EQ-5D	unknown
GEMINI 343 NCT01224171	Induction	10 weeks	416	1. Vedolizumab 300 mg at weeks 0, 2, 6 (n = 209; 158 of whom were anti-TNF failure); 2. Placebo at weeks 0, 2, 6 (n = 207; 157 of whom were anti-TNF failure)	Millennium Pharmaceuticals (Takeda Pharmaceuticals)	34.8 (19–77)/36.9 (20–69)b	89 (43)/91 (44)	301.3 (55)/313.9 (53.2)	IBDQ, SF-36, EQ-5D	unknown
Feagan 201752 NCT02031276	Induction	12 weeks	121	1. Risankizumab 200 mg at weeks 0, 4, and 8 (n = 41); 2. Risankizumab 600 mg at weeks 0, 4, and 8 (n = 41); 3. Placebo at weeks 0, 4, and 8 (n = 39)	Boehringer Ingelheim (Ingelheim, Germany), AbbVie	36 (14)/40 (13)	16 (41)/16 (39)	295 (237–386)/298 (259–330)b	IBDQ	119.3 (34.9)/111.5 (27.8)
ADVANCE48 NCT03105128	Induction	12 weeks	931	1. Risankizumab 600 mg (n = 373); 2. Risankizumab 1200 mg (n = 372); 3. Placebo (n = 186)	Boehringer Ingelheim (Ingelheim, Germany), AbbVie	37.1 (13.4)/38.3 (13.3)	88 (50)/189 (56)	319.2 (59.4)/311.2 (62.4)	IBDQ, SF-36	122.7 (32.5)/119.7 (31.3)
MOTIVATE48 NCT03104413	Induction	12 weeks	569	1. Risankizumab 600 mg (n = 191); 2. Risankizumab 1200 mg (n = 191); 3. Placebo (n = 187)	Boehringer Ingelheim (Ingelheim, Germany), AbbVie	39.3 (13.5)/40.2 (13.6)	99 (53)/92 (48)	319.6 (69.8)/310.7 (63.6)	IBDQ, SF-36	115.0 (31.9)/119.4 (28.7)
FORTIFY48 NCT03105102	Maintenance	52 weeks	542	1. Risankizumab 180 mg (n = 179); 2. Risankizumab 360 mg (n = 179); 3. Placebo (n = 184)	Boehringer Ingelheim (Ingelheim, Germany), AbbVie	unknown	unknown	133.6 (80.6)/132.8 (75.8)	IBDQ, SF-36	unknown
U-EXCEL49 NCT03345849	Induction	12 weeks	526	1. upadacitinib 45 mg (n = 350); 2. placebo (n = 176)	AbbVie	39.3 (13.6)/39.7 (13.7)	94 (53.4)/189 (54)	293.9 (85.4)/292.4 (81.3)	IBDQ	unknown
U-EXCEED49 NCT03345836	Induction	12 weeks	495	1. upadacitinib 45 mg (n = 324); 2. placebo (n = 171)	AbbVie	37.5 (12.1)/38.4 (13.7)	96 (56.1)/169 (52.2)	308.1 (84.3)/306.6 (89.4)	IBDQ	unknown
U-ENDURE49 NCT03345823	Maintenance	52 weeks	502	1. upadacitinib 15 mg (n = 169); 2. upadacitinib 30 mg (n = 168); 3. placebo (n = 165)	AbbVie	38.1 (13)/38.1 (13.5)	88 (53.3)/102 (60.4)	308.4 (82.3)/300.8 (90.8)	IBDQ	unknown

Abbreviations: n, number; CDAI, Crohn Disease Activity Index; SF-36, The 36-Item Short Form Survey; IBDQ, Inflammatory Bowel Disease Questionnaire; HRQoL, Health-related quality of life.

^aAll patients.

^bMedian (range).

^cCDAI ≥150 (number of patients).

^dResponders.

IBDQ

Infliximab versus placebo

The efficacy of IFX was investigated in 1162 adults with moderately to severely active CD. Demographic and clinical characteristics of participants (age, sex, race, baseline CDAI score, and concomitant therapy) were equally represented between groups in all included studies.^36,37,47,53 Three treatment arms were investigated (IFX vs. azathioprine vs. IFX + azathioprine) in the SONIC study.⁴⁷ We analyzed data only for the IFX vs azathioprine group. Infliximab was compared with placebo in four studies (ACCENT I³⁶; ACCENT II,³⁷ Lichtenstein 2002⁵³; SONIC⁴⁷) in doses of 5 mg/kg, 10 mg/kg, and 20 mg/kg. Lichtenstein 2002⁵³ and SONIC⁴⁷ studies were induction studies with a follow-up of 4 and 30 weeks, and ACCENT I³⁶ and ACCENT II³⁷ were maintenance studies with a follow-up of 54 weeks.

A random-effects meta-analysis using the Hartung-Knapp-Sidik-Jonkman method found that IFX results in a CMI in QoL as measured by the IBDQ score, with a pooled MD of 16.09 (95% CI: −0.13 to 32.31; 4 studies; 1162 participants, very low-certainty evidence, Fig. 2). Although the observed between-study heterogeneity was considerable (I² = 78%, P = 0.003), the effect estimates were consistent across all included studies. Therefore, we did not consider observed heterogeneity as a factor that contributes to very serious inconsistency. The observed heterogeneity may be explained by different treatment dosage schedules. In the Lichtenstein 2002⁵³ study, treatment consisted only of a single IFX or placebo infusion, whilst in SONIC⁴⁷ and ACCENT I³⁶ participants were treated with repeated IFX or placebo infusions.

Fig. 2.

Forest plot illustrating the mean differences of IBDQ score change between infliximab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: IBDQ, Inflammatory Bowel Disease Questionnaire; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; p, p-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

A subgroup meta-analysis was conducted to explore whether the phase of treatment (induction vs. maintenance) modified the effect of infliximab on quality of life outcomes. The pooled MD in the maintenance subgroup (ACCENT I³⁶; ACCENT II³⁷) was 12.78 (95% CI: −37.97 to 63.52), with moderate heterogeneity (I² = 40.5%, P = 0.19). In contrast, the induction subgroup (Lichtenstein 2002 and SONIC) showed a pooled MD of 19.71 (95% CI: −120.01 to 159.43), with high heterogeneity (I² = 91.1%, P < 0.001). The test for subgroup differences did not indicate a statistically significant difference between induction and maintenance phases (χ² = 0.35, df = 1, P = 0.553), suggesting that the treatment phase was not a significant modifier of effect size. However, high variability within the induction subgroup should be taken into account when interpreting the results. An additional analysis was conducted to assess whether the dose of infliximab (5 mg/kg vs. 10 mg/kg) could explain differences in treatment effect. The pooled MD for the 5 mg/kg group (ACCENT II and SONIC) was 9.00 (95% CI: 9.00–9.00), with no observed heterogeneity (I² = 0%, P = 1.00). The 10 mg/kg group (ACCENT I) showed a MD of 17.00 (95% CI: 8.05–25.95). The test for subgroup differences again showed no statistically significant difference between dose groups (χ² = 3.07, df = 1, P = 0.080), although a numerically larger effect size was observed in the 10 mg/kg group. These results should be interpreted with caution, as both subgroup analyses are based on a limited number of studies, and in the case of the dose comparison, only one study contributed to the 10 mg/kg group.

Adalimumab versus placebo

Adalimumab was compared to placebo in 1457 patients with moderate to severe CD. Demographic and clinical characteristics of patients were similar comparing responders and nonresponders at week 4 in the CHARM study³⁴ and comparing adalimumab-treated groups with the placebo group in GAINT, CLASSIC I, and CLASSIC II studies.31, 32, 33

Adalimumab 20 mg, 40 mg, 80 mg and 160 mg was investigated in four studies. GAINT³¹ and CLASSIC I³² studies were designed as induction trials, with a follow-up of 4 weeks, and CHARM⁵⁴ ad CLASSIC II³³ were maintenance trials with a follow-up of 56 weeks.

Adalimumab results in no CMI in QoL in patients with moderate to severe CD, as measured by IBDQ score change (MD 10.59, 95% CI 2.68–18.50; I² = 25%; P = 0.02; 4 studies; 790 participants; high-certainty evidence; Fig. 3).

Fig. 3.

Forest plot illustrating the mean differences of IBDQ score change between adalimumab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: IBDQ, Inflammatory Bowel Disease Questionnaire; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Certolizumab pegol versus placebo

Certolizumab pegol was evaluated in 1088 patients with moderately to severely active CD in two RCTs (PRECISE 1,⁴⁰ PRECISE 2⁴¹). The baseline characteristics were similar between certolizumab pegol and placebo groups. Certolizumab pegol was investigated in a dosage schedule of 400 mg at weeks 0, 2, 4, and then every 4 weeks up to week 24 in a maintenance schedule with a follow-up of 26 weeks.

Certolizumab pegol results in no CMI in QoL, as measured by change in IBDQ score (MD 4.27, 95% CI −16.81 to 25.35; I² = 0%; P = 0.24; 2 studies; 620 participants; low-certainty evidence, Fig. 4).

Fig. 4.

Forest plot illustrating the mean differences of IBDQ score change between certolizumab pegol and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: IBDQ, Inflammatory Bowel Disease Questionnaire; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Ustekinumab versus placebo

The effect of ustekinumab was studied in 2292 participants with moderate to severe CD in four RCTs (CERTIFI³⁸; IM-UNITI³⁹; UNITI-1³⁹; UNITI-2³⁹). Baseline demographic and clinical characteristics of patients were similar across active treatment and placebo groups, except for the CDAI score in the CERTIFI³⁸ induction phase of the study (for placebo-group CDAI score (means ± SD) was 312.4 ± 64.2 and for the ustekinumab-treated group 327.7 ± 64.6). In CERTIFI³⁸; UNITI-1³⁹; UNITI-2,³⁹ patients were required to have met the criteria for primary or secondary non-response to TNF-α inhibitors or to have had antiTNF-α unacceptable side effects.

Ustekinumab 6 mg/kg based on three induction studies with a follow-up of 6–8 weeks (CERTIFI³⁹; UNITI-1³⁹; UNITI-2³⁹) and ustekinumab 90 mg every 8 weeks from one maintenance study, IM-UNITI,³⁹ with a follow-up of 44 weeks were analysed. The CERTIFI study³⁹ included induction and maintenance phases, but data on HRQoL outcomes were presented differently. We analyzed the results of the induction phase as they satisfied definition of the primary outcome for this review. The results of maintenance phase of CERTIFI study³⁹ is presented in Supplementary Material.

Ustekinumab results in no CMI in QoL as measured by IBDQ scores (MD 13.89, 95% CI 6.26–21.51; I² = 58%, P = 0.07; 4 studies; 1417 participants; moderate-certainty evidence, Fig. 5). While statistical heterogeneity was substantial, the direction of effect was consistent across studies. A subgroup meta-analysis was conducted to examine whether the phase of treatment (induction vs. maintenance) modified the effect of ustekinumab on quality of life outcomes. The pooled MD among induction studies (CERTIFI, UNITI 1, UNITI 2) was 14.43 (95% CI: 0.99–27.87), with substantial heterogeneity (I² = 71%, P = 0.032). The maintenance study (IM-UNITI) showed a MD of 11.60 (95% CI: 2.47–20.73). The test for subgroup differences was not statistically significant (χ² = 0.25, df = 1, P = 0.614), suggesting that the treatment phase did not significantly modify the treatment effect. However, these results should be interpreted with caution, as only one study contributed to the maintenance subgroup, limiting the strength of conclusions.

Fig. 5.

Forest plot illustrating the mean differences of IBDQ score change between ustekinumab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: IBDQ, Inflammatory Bowel Disease Questionnaire; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Natalizumab versus placebo

The effect of natalizumab was investigated in 1753 participants with active CD. Baseline characteristics were similarly represented between the treatment and placebo groups in the ENACT-1³⁵ and ENCORE study.³⁵ In the ENACT-2 study,⁵¹ more male patients (77, 46%) were present in the natalizumab group compared to placebo group (59, 35%) and the baseline CDAI score (mean (SD)) was higher in the natalizumab than placebo group (118 (57) vs 105 (54), P < 0.05)). Natalizumab was studied in the dose of 300 mg every four weeks in two induction studies with a follow-up of 12 weeks (ENACT-1³⁵; ENCORE³⁵) and in one maintenance study with a follow-up of 60 weeks (ENACT-2⁵¹).

The evidence suggests that natalizumab results in no CMI in QoL measured by IBDQ score change (MD 11.78, 95% CI −2.24 to 25.79; I² = 62%, P = 0.07; 3 studies, 1752 participants; low-certainty evidence, Fig. 6). Although the point estimate favors natalizumab, the confidence interval includes no effect, and moderate heterogeneity was present.

Fig. 6.

Forest plot illustrating the mean differences of IBDQ score change between natalizumab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: IBDQ, Inflammatory Bowel Disease Questionnaire; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

A subgroup meta-analysis was conducted to examine whether the phase of treatment (induction vs. maintenance) influenced the effect of natalizumab on QoL outcomes. The pooled MD in the induction subgroup (ENACT-1 and ENCORE) was 9.33 (95% CI: −21.02 to 39.68), with low heterogeneity (I² = 26.3%, P = 0.244). The maintenance subgroup (ENACT-2) showed a MD of 18.40 (95% CI: 10.51–26.29). The test for subgroup differences was not statistically significant (χ² = 3.75, df = 1, P = 0.053), indicating that there is no conclusive evidence of a difference in treatment effect between induction and maintenance phases. However, the numerical difference and borderline P-value suggest that this finding may warrant further investigation. As only one study contributed to the maintenance subgroup, and substantial heterogeneity was observed overall (I² = 62.4%), these results should be interpreted with caution.

Vedolizumab versus placebo

The effect of vedolizumab 300 mg was evaluated in 1245 participants with active CD. GEMINI 2⁴² included patients who were nonresponders or had unacceptable side effects from conventional immunosuppressive agents or TNF antagonists. Patients included in the GEMINI 3⁴³ had been anti-TNF failures. The demographic characteristics were similar between the study groups. GEMINI 2⁴² consisted of an induction and maintenance phase with a follow-up of 52 weeks and GEMINI 3⁴³ was an induction study with a follow-up of 10 weeks. Only three studies were analyzed and testing funnel plot asymmetry was not done, although considerable heterogeneity was detected (I² = 99%, P < 0.00001). Included studies were available in the manuscript and Supplementary Material form and we considered them of sufficient methodological quality. Hence, a meta-analysis was not performed. The results are presented descriptively.

The mean (SD) IBDQ score change in the vedolizumab group ranged from 23.1 (2.28) to 50.7 (3.88) points and in the placebo group from 15.4 (2.62) to 35.5 (3.81) points. Effect estimates were consistent across studies, indicating higher IBDQ score changes in the vedolizumab group. Subgroup analysis was not performed, as the maintenance study (GEMINI II maintenance) is a continuation of its corresponding induction study, and it is the only study within that category. Therefore, the subgroup comparison between induction and maintenance was not feasible.

Risankizumab versus placebo

The effect of risankizumab was investigated in 2163 participants with active CD.^48,52 Demographic and clinical patient characteristics were similar between groups, except for exposures to conventional and biological agents before to study enrolment. In Feagan 2017,⁵² patients could have had previous treatment with conventional and one or more TNF antagonists or vedolizumab, but not ustekinumab and any other biologics. The ADVANCE trial enrolled patients with demonstrated intolerance or inadequate response to conventional therapies or/and biologic therapy, whereas enrolment into the MOTIVATE trial was restricted to only patients with previous bio-failure.⁴⁸ Risankizumab 600 mg was studied in three induction studies with a follow-up of 12 weeks each (Feagan 2017, ADVANCE, and MOTIVATE).^48,52 In a maintenance trial, FORTIFY, risankizumab 180 mg was studied with a follow-up of 52 weeks.⁴⁸

Considerable heterogeneity was detected (I² = 80%, P = 0.002) and data were not pooled for meta-analyses. The mean (SD) IBDQ score change in the risankizumab group ranged from 34.7 (29.9) to 60.9 (25.12) and in the placebo group from 7.3 (35.2) to 56.4 (25) points. Effect estimates were consistent across three induction studies (Feagan 2017, ADVANCE, MOTIVATE^48,52), indicating higher IBDQ score changes in the risankizumab group.

Subgroup analysis indicated that the induction versus maintenance phase significantly modified the treatment effect (test for subgroup differences: χ² = 7.29, P = 0.007). The pooled effect was larger in the induction subgroup (MD 18.7, 95% CI 1.74–35.66; I² = 55.1%) compared to the maintenance subgroup (MD 4.5, 95% CI −2.32 to 11.32; 1 study, Fig. 7).

Fig. 7.

Forest plot illustrating the mean differences of IBDQ score change between risankizumab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: IBDQ, Inflammatory Bowel Disease Questionnaire; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Upadacitinib versus placebo

Upadacitinib treatment was investigated in 1523 participants with CD. The demographic and clinical characteristics of the patients at baseline were well-balanced across groups and representative of the overall global population with moderate-to-severe CD.⁴⁹ In the U-EXCEL⁴⁹ study, participants were required to have had a history of failure of one or more conventional or biologics, whilst the U-EXCEED⁴⁹ study included participants who had a history of one or more biologics failures. Upadacitinib 45 mg given as extended-release tablets was studied in two induction trials with a follow-up of 12 weeks each (U-EXCEL, U-EXCEED). In one maintenance study (U-ENDURE), upadacitinib extended-release tablets 15 mg and 30 mg were studied with a follow-up of 52 weeks.⁴⁹

Upadacitinib results in a CMI in QoL assessed by IBDQ score change (MD 20.65, 95% CI 7.04–34.25; I² = 42.8%, P = 0.17; 3 studies; 1523 participants; high-certainty evidence, Fig. 8).

Fig. 8.

Forest plot illustrating the mean differences of IBDQ score change between upadacitinib and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: IBDQ, Inflammatory Bowel Disease Questionnaire; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

TNF-α antagonists (infliximab, adalimumab and Certolizmab pegol) versus placebo

We aimed to pool data from four studies on IFX (ACCENT I³⁶; ACCENT II,³⁷ Lichtenstein 2002,⁵³ and SONIC⁴⁷), four studies on adalimumab (CHARM,³⁴ GAINT,³¹ CLASSIC I³²; CLASSIC II studies³³) and two studies on certolizumab pegol (PRECISE 1⁴⁰; PRECISE 2⁴¹) to investigate an effect of TNF-ɑ antagonists versus control treatment on IBDQ score change. A substantial heterogeneity between studies was detected (I² = 67%, P = 0.002). Effect estimates were inconsistent across studies. The mean (SD) IBDQ score change in the TNF-ɑ antagonist group ranged from 16 (29) to 54.69 (36.69) points and in the placebo group from 5 (6.48) to 46.8 (34.84) points.

The evidence suggests that TNF-α antagonists result in no CMI in QoL as measured by IBDQ score change (MD 11.28, 95% CI 5.45–17.10; I² = 67%, P = 0.0018; 10 studies; 2377 participants; low-certainty evidence, Fig. 9).

Fig. 9.

Forest plot illustrating the mean differences of IBDQ score change between tumour necrosis factor-α antagonists (infliximab, adalimumab and certolizumab pegol) and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: IBDQ, Inflammatory Bowel Disease Questionnaire; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Although the subgroup comparison did not reach statistical significance (I² = 25%, P = 0.08), a trend toward greater effect was observed in induction trials (CLASSIC I,³² GAINT,³¹ Lichtenstein 2002,⁵³ SONIC⁴⁷), where a CMI was achieved (MD 16.37, 95% CI 1.34–31.40; I² = 74.5%, P = 0.008; 4 studies; 895 participants).

In contrast, a smaller and more heterogeneous improvement was found in maintenance trials (CHARM,³⁴ ACCENT I,³⁶ ACCENT II,³⁷ CLASSIC II,³³ PRECISE 1,⁴⁰ PRECISE 2⁴¹), with an MD of 7.39 (95% CI 1.90–12.89; I² = 27.0%, P = 0.23; 6 studies; 1482 participants), not exceeding the threshold for clinical relevance.

Discussion

The evidence suggests infliximab and upadacitinib result in CMI in QoL assessed by IBDQ score change (very low and high-certainty evidence). No CMIs were observed for adalimumab, certolizumab pegol, ustekinumab, and natalizumab treatment (low to high certainty). Considerable heterogeneity prevented us from pooling data in meta-analyse on vedolizumab and risankizumab treatment.

In all included RCTs, a statistically significant difference is achieved for at least one HRQoL result, but a clinically meaningful significance is not achieved in many included analyses.55, 56, 57, 58 Reaching a statistical significance is only a precondition for achieving a clinically meaningful difference. From the patient perspective, a CMI in the HRQoL is a crucial point and treatment goal that should be achieved.⁵⁹ Clinicians should be careful when interpreting results in terms of statistical significance. Improvement in QoL in CD patients encompasses achieving clinical remission of disease, but van der Have and colleagues found that disease activity contributed to only 37% of HRQoL. According to this observation and that all included RCTs have not been powered to detect a clinically meaningful difference in HRQoL outcomes, the fact that a CMI was not achieved in most of our analyses should not be simply interpreted as there is no significant effect of biologicals on HRQoL. Few possible explanations could address our findings: shortcomings in study design considering HRQoL outcomes as secondary, a heterogeneous participant population, and differences in inclusion criteria. For instance, the lack of CMI with adalimumab and certolizumab pegol may reflect suboptimal patient selection, with these agents potentially ineffective in certain CD phenotypes or stages. Study design factors, including small sample sizes and inconsistent HRQoL measures, may have limited the ability to detect subtle improvements. Alternatively, the intrinsic limitations of these agents in enhancing HRQoL, despite clinical efficacy in disease control, suggest a need for further exploration of their broader impact on patient well-being. Sandborn et al. and Panaccione et al. investigated the association between defined clinical and endoscopic remission and patients' well-being.^60,61 Their results are based on partially interpreted IBDQ (only well-being is evaluated), in a relatively small subgroup of patients treated with adalimumab and/or upadacitinib, but they recognized the need for a more accurate examination of the association between clinical effectiveness and HRQoL. In all included studies, sample size was estimated and reported for primary outcomes and not for secondary, such as HRQoL outcomes. This issue may lead to an increase in the probability of type II error, namely, to observe no effect when it exists.⁵⁹ A more appropriate outcome measure for HRQoL (not object—questionnaire, but person –participant in the study, e.g., the number of participants experiencing a CMI) should be investigated in RCTs as a primary outcome. We agree that it should be a reliable, with face and construct validity, responsive, and feasible in clinical practice.²⁷ A similar inclusion and exclusion criteria across included trials contributed to the methodological and clinical homogeneity, but we observed substantial heterogeneity in the analyses of ustekinumab, natalizumab, vedolizumab, and risankizumab treatment. Prior use of TNF-α inhibitors and the type of the study (induction vs maintenance), and other specific factors regarding study design (timing of forced corticosteroid-taper, different scoring methodologies, presence of open-label escape) may be contributing factors to observed clinical heterogeneity.⁶² Lack of data prevents us from making additional, more detailed analyses and drawing more accurate conclusions. Among the various methodological differences, we emphasize the one concerning the type of study and the role of the time lag in the definition of the types of studies.⁶² From the first RCT investigating infliximab⁵³ to the present day, a clear distinction between induction and maintenance studies has disappeared, and the definition of induction and maintenance studies has acquired a few new modalities. In addition to the induction-only and maintenance-only studies, there are induction and maintenance studies designed as a single one and induction and responder maintenance studies, also incorporated in a single study.⁶² Definition of the same HRQoL outcome (e.g., IBDQ score change) across different types of RCTs introduces methodological differences as the baseline time-point differs if we compare, for example, induction-only study data and those from induction and responder maintenance studies. This was a reason why induction and maintenance phases of GEMINI 2 study⁴² were not included concomitantly in the analyses. Additionally, when evaluating HRQoL results from included studies, we noticed a heterogeneity considering data reporting statistics (e.g. only median or mean change, without standard deviation, only percentages/no absolute values presented) accompanied with a variety of interpretation of these data in the main text of studies (e.g., described and emphasized results partially). The average reader can hardly notice and understand those small differences in communication that may impact results and conclusions. The differences in reported results pose obstacles to comparing effects on QoL between biologicals in a network meta-analysis. There arises the need to harmonize and internationally agree on the method of presenting data on the QoL (e.g., always show absolute values, along with measures of dispersion; in addition to the IBDQ questionnaire score change analysis, the percentage of people who have experienced CMI should also be reported). This area is just emerging for research and our results with the STRIDE-II Initiative may be a cornerstone for future orientation in this field.²⁷

All studies included in this review were RCTs sponsored by industry. Publication bias, including writing the manuscript or involving the main investigator who was funded by industry, should not be ignored. In each included study, investigators acknowledged their relationship with pharmaceutical industry and its support. We did not detect any RCT that is not the pharmaceutically sponsored, as the financial burden is too high for investigation with no financial interest. This type of bias should always be appropriately announced, as it cannot be avoided.

We acknowledge limitations of our review: we included only biologicals that are already approved for the indication of CD. The approval process lasts years, and far more drugs are being tested than receive approval which would lead to many useless results. Given that this work is primarily intended to serve clinical practice, it justifies our decision. We analyzed results based on SF-36 and EQ-5D, in addition to IBDQ. The IBDQ results are more representative and reflect HRQoL in all included RCTs. Contrary, SF-36 and EQ-5D are reported in addition to IBDQ or did not report at all. Furthermore, we acknowledge and highlight that it was not possible to assess the sources of heterogeneity, including meta-regression modeling as well as publication bias due to the low number of included studies. The same reason limited us to accurately estimate the between-study heterogeneity, leading to imprecision in effect estimates. Consequently, we cautiously interpret our results: we are unable to conclude that there is no effect, though we did not find evidence of an effect in many analyses.

Three systematic reviews dealing with the HRQoL in CD related to biological treatment have already been published in 2006, 2009 and 2020.63, 64, 65 All three are descriptive, narratively summarizing the original data from the included studies, and analyzing the IFX, adalimumab, certolizumab pegol, natalizumab, filgotinib, upadacitinib, tofacitinib and apilimod mesylate based on statistical significance. The most prominent differences comparing those reviews with this are: we meta-analysed data from RCTs based on CMI. Using unpublished data (in addition to published) allowed us the most comprehensive analyses ever done in this field. As CMI observed in HRQoL scores presume statistically significant results, our results are consistent with results based on published reviews.63, 64, 65 The American College of Gastroenterology (ACG), American Gastroenterological Association (AGA), National Institute for Health and Care Excellence (NICE), and British Society of Gastroenterology (BSG) all make no mention of guidance involving specifically the impact of biological treatment on HRQoL.55, 56, 57^,66 European Crohn's and Colitis Organization (ECCO) guidelines explicitly state their aim to assess HRQoL related to anti-TNF-a, ustekinumab, and vedolizumab, but the data were partial.⁵⁸ Our results, as an objectively proof, fully support selected treatment targets, gaps in knowledge and areas for future research specified by SPIRIT and STRIDE-II Initiative, considering HRQoL and PRO in CD.^27,67

To our knowledge, this is the most up-to-date systematic review that comprehensively evaluates existing evidence on HRQoL outcome (IBDQ, SF-36, and EQ-5D) from all over the world and including all approved biologic treatments for CD. These findings highlight the need to integrate HRQoL outcomes into treatment guidelines, particularly for biologics like infliximab and upadacitinib, whose benefits extend beyond clinical remission to significantly enhance patient well-being. Clinical trial designs must evolve by incorporating larger, more diverse cohorts and standardized HRQoL measures, thereby increasing the sensitivity to detect meaningful patient-reported improvements. Additionally, regulatory bodies should prioritize HRQoL in drug approval processes, ensuring that biologic treatments are evaluated not only for clinical efficacy but for their broader impact on patients' lives, ultimately fostering a more patient-centered approach to therapy.

A clearer and more applicable indicator of HRQoL, which more reliably shows an association between clinically proven effectiveness and CMI in HRQoL, should be considered. Clinicians should be cautious when interpreting results of HRQoL in terms of statistical and clinically meaningful significance.

Contributors

Mirjana Stanić Benić (MSB) drafted the protocol, obtained copies of trials, selected which trials to include, extracted data from trials, entered data into RevMan, carried out the analysis, interpreted the analysis, drafted the final review, and updated the review following repeated searches. Vanja Giljača (VG) drafted the protocol, selected which trials to include, extracted data from trials, carried out the analysis, interpreted the analysis, drafted the final review, and updated the review following repeated searches. Andrej Belančić (AB) carried out the analysis, interpreted the analysis, drafted the final review, and updated the review following repeated searches. Antonija Jeličić Kadić (AJK) carried out the analysis, interpreted the analysis, drafted the final review, and updated the review following repeated searches. Vera Vlahović-Palčevski (VVP) interpreted the analysis, drafted and approved the final version. MSB, VG and AB had accessed and verified the underlying data.

All authors read and approved the final version of the manuscript.

Data sharing statement

The information analysed during the current study are available from the corresponding author on reasonable request.

Declaration of interests

The authors have no competing interests that might be perceived to influence the results and discussion reported in this manuscript. Authors Mirjana Stanić Benić, Vanja Giljača, Andrej Belančić, Antonia Jeličić Kadić, and Vera Vlahović-Palčevski declare that they have no conflict of interest, including specific financial interests and relationships and affiliations relevant to the subject of this manuscript.

Appendix A. Supplementary data

Supplementary Figure S1.

Risk of bias in included studies. Abbreviations: +, low risk of bias; ? unclear risk of bias; -, high risk of bias.

Supplementary Figure S2.

Forest plot illustrating the mean differences of SF-36 Mental Component Summary score change between infliximab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: SF-36, The 36-Item Short Form Survey; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Supplementary Figure S3.

Forest plot illustrating the mean differences of SF-36 Physical Component Summary score change between certolizumab pegol and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: SF-36, The 36-Item Short Form Survey; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Supplementary Figure S4.

Forest plot illustrating the mean differences of SF-36 Mental Component Summary score change between certolizumab pegol and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: SF-36, The 36-Item Short Form Survey; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Supplementary Figure S5.

Forest plot illustrating the mean differences of SF-36 Physical Component Summary score change between ustekinumab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: SF-36, The 36-Item Short Form Survey; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Supplementary Figure S6.

Forest plot illustrating the mean differences of SF-36 Mental Component Summary score change between ustekinumab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: SF-36, The 36-Item Short Form Survey; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Supplementary Figure S7.

Forest plot illustrating the mean differences of SF-36 Physical Component Summary score change between risankizumab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: SF-36, The 36-Item Short Form Survey; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.

Supplementary Figure S8.

Forest plot illustrating the mean differences of SF-36 Mental Component Summary score change between risankizumab and placebo groups across included studies. Squares represent individual study estimates of the mean difference, and horizontal lines (whiskers) indicate the 95% confidence intervals. The size of each square (box plot) reflects the weight of the corresponding study in the meta-analysis. The diamond at the bottom represents the overall mean difference and its 95% CI calculated using a random-effects model. Abbreviations: SF-36, The 36-Item Short Form Survey; CI, confidence interval; I², heterogeneity statistic; MD, mean difference; N, number of participants; P, P-value; SD, standard deviation; t, t-statistic for overall effect; τ², between-study variance.