To Örebro University

BACKGROUND: 16S rRNA gene sequencing is widely used for bacterial community profiling in both clinical and research contexts. The expanding availability of library preparation protocols and bioinformatic pipelines increases analytical flexibility but may also introduce method-dependent biases that affect inferred microbial composition and relative abundance estimates. The relative impact of library preparation protocol, amplicon region, and bioinformatic pipeline on species-level taxonomic inference and compositional agreement remains insufficiently characterised. We therefore compared the Illumina 16S Metagenomic Sequencing Library Preparation protocol (V3-V4) and the Zymo Quick-16S Plus NGS Library Prep Kit (V1-V2 and V3-V4) in combination with two bioinformatic pipelines, nf-core/ampliseq and TRANA. Performance was assessed using defined microbial community standards and human faecal and colonic biopsy samples.

RESULTS: Pipeline choice was the dominant driver of variation in inferred community composition, exceeding the effects of amplicon regions and library preparation protocols. Genus-level profiles were broadly concordant across methods. Species-level resolution and agreement with expected community composition differed systematically between pipelines, with TRANA demonstrating lower Bray-Curtis dissimilarities to expected compositions than nf-core/ampliseq. Amplicon region had a secondary, pipeline-dependent effect, while protocol differences were minor. In clinical samples, inter-individual biological variation exceeded technical variation.

CONCLUSIONS: Bioinformatic processing substantially influenced species-level inference in short-read 16S sequencing, highlighting the importance of pipeline selection for microbiome study design and cross-study comparability.

Shifting to a plant-based diet naturally alters protein source choices. In many countries, protein from yellow pea is widely used as a main ingredient in meat alternatives. Still, its biological effects, especially regarding gastrointestinal health, remain incompletely understood. The aim of our study was to investigate how a weekly increase in the intake of a well-characterized pea protein isolate affects surrogate markers of health, fecal short-chain fatty acids and gut microbiota composition in healthy individuals. Male and female adults (N = 29) participated in this exploratory intervention study. A 4-week pre-intervention period for questionnaires and fecal samples collection was followed by a 4-week supplementation. Participants consumed isolated pea protein in weekly increasing amounts, starting from 0.25 g per kg body mass per day in week 5 to 1.00 g per kg body mass per day in week 8. Questionnaire data, fecal samples as well as fasting blood and 24 h urine samples were collected weekly. Data from biological samples and questionnaires confirmed a healthy study population and compliance. Fecal calprotectin levels significantly increased only in a subset of participants, which was accompanied by higher fecal water cytotoxicity in vitro. Short-chain fatty acids mainly rose in those subjects with stable calprotectin levels. Relative abundances of Limosilactobacillus frumenti, Odoribacter splanchnicus and Lactobacillus crispatus increased significantly in the total population during the intervention while the relative abundance of Bifidobacterium longum and Bifidobacterium catenulatum decreased. Our results indicate that an increased intake of pea protein isolate affects the growth of certain beneficial bacterial strains and differentially influences markers related to gut inflammation in healthy individuals.

Background: Inter-individual differences in drug clearance are common in inflammatory bowel disease (IBD) patients treated with biologics. Associations between inflammatory markers, clinical features, and drug levels may indicate that the inflammatory immune response influences the pharmacokinetics of biologics. We assessed the impact of inflammatory and immunity-related proteins on drug levels in IBD.

Methods: Serum samples from 144 IBD patients initiating biologics in a prospective multicentre cohort were analysed using AFIAS-3 (Boditech Med Inc. Korea). The primary outcome was drug levels after the end of induction treatment (post-induction), analysed as continuous variables or categorised into high or low groups based on the median. Correlations were assessed using the Pearson’s correlation coefficients with false discovery rate (FDR) adjustment. Predictive capacity of clinical and protein data was evaluated with regularised linear regression models. Analyses were stratified by treatment (infliximab, adalimumab, vedolizumab) and diagnosis (Crohn’s disease (CD), ulcerative colitis (UC)).

Results: Patient demographics and clinical characteristics are provided in Table 1. Median (interquartile range, IQR) post-induction infliximab levels were 5.3 (2.2-9.5) μg/ml in patients with CD (n=34) and 3.8 (1.7-5.6) μg/ml in UC (n=33). The corresponding levels were 12.5 (7.0-16.1) μg/ml (n=36) and 8.8 (5.7-12.7) μg/ml (n=13) for adalimumab, 13.9 (6.4-24.4) μg/ml (n=19) and 17.7 (9.8-22.6) μg/ml (n=9) for vedolizumab. Principal component analyses revealed potential baseline protein profile differences between patients with high vs low post-induction levels for infliximab (p=0.09) and adalimumab (p=0.07) in CD, but no differences for vedolizumab (p=0.78) or UC. Individual protein analyses indicated that patients with higher drug levels post-induction seemed to display lower baseline estimates of inflammatory proteins compared to those with lower drug levels. However, statistical significance after correction for multiple comparisons was only observed for CDCP1 in adalimumab-treated CD patients (PFDR=0.04) (Figure 1). Additionally, baseline Flt31 (r=-0.57, PFDR=0.03) and Wisp1 (r=-0.61, PFDR=0.02) correlated with post-induction s-adali-mumab levels in CD, while Mcp3 (r=-0.89, PFDR=0.04) and Frgamma (r=-0.87, PFDR=0.05) correlated in UC. Integrating baseline protein data in prediction models for post-induction drug levels did not improve their accuracy compared to models based on only clinical and biochemical variables.

Conclusion: Inflammatory protein levels may influence post-induction drug levels in IBD, particularly in adalimumab treated CD patients. This insight could aid in optimising personalised treatment.

BACKGROUND AND AIMS: Biomarkers are needed to identify individuals at elevated risk of inflammatory bowel disease (IBD). This study aims to identify protein signatures predictive of IBD.

METHODS: Using large population-based cohorts (n≥180,000), blood samples were obtained from individuals who later in life were diagnosed with IBD and compared to age and sex-matched controls, free from IBD during follow-up. 178 proteins were measured on Olink platforms. We used machine learning methods to identify protein signatures of preclinical disease in the discovery cohort (n=312). Their performance was validated in an external preclinical cohort (n=222) and assessed in an inception cohort (n=144) and a preclinical twin cohort (n=102).

RESULTS: In the discovery cohort, a signature of 29 proteins differentiated preclinical Crohn's disease (CD) cases from controls, with an area under the curve (AUC) of 0.85. Its performance was confirmed in the preclinical validation (AUC=0.87) and the inception cohort (AUC=1.0). In preclinical samples, downregulated (but not upregulated) proteins related to gut barrier integrity and macrophage functionality correlated with time to diagnosis of CD. The preclinical ulcerative colitis (UC) signature had a significant, albeit lower, predictive ability in the discovery (AUC=0.77), validation (AUC=0.67), and inception cohorts (AUC=0.95). The preclinical signature for CD demonstrated an AUC of 0.89 when comparing twins with preclinical CD to matched external healthy twins, but its predictive ability was lower (AUC=0.58; P=.04) when comparing them with their healthy twin siblings, i.e., when accounting for genetic and shared environmental factors.

CONCLUSION: We identified protein signatures for predicting a future diagnosis of CD and UC, validated across independent cohorts. In the context of CD, the signature offers potential for early prediction.

Background: We aimed to identify protein signatures predictive of a future diagnosis of inflammatory bowel disease (IBD).

Methods: We conducted a case-control study, nested within large population-based cohorts with biorepositories. Samples were obtained from individuals who later in life were diagnosed with IBD (preclinical cases) and compared with age and sex-matched individuals who remained free from IBD during follow-up (controls). Using proximity extension assays (Olink, Uppsala), we measured 176 proteins. We applied regularized logistic regression to identify protein signatures of preclinical disease in serum from the discovery cohort (n=312). Their performance was validated in an external preclinical cohort (n=222). The biological relevance of identified proteins was further assessed in an inception cohort (n=144). Finally, we used an IBD twin cohort (n=327) to examine the impact of genetic and shared environmental factors on identified proteins.

Results: We identified 34 proteins associated with preclinical Crohn’s disease (CD) in the discovery cohort (Pfalse discovery rate <0.10), with 9 confirmed in the validation cohort (Pfalse discovery rate <0.05). For preclinical ulcerative colitis (UC), 45 proteins were identified and 12 validated (Fig. 1A-B). In the discovery cohort, a signature of 29 proteins differentiated preclinical CD cases from controls with an AUC of 0.85 (Fig. 1G). Its performance was confirmed when applied to the preclinical validation cohort (AUC=0.84, Fig. 1H). Moreover, the signature had excellent capacity to differentiate newly diagnosed CD from healthy controls in the inception cohort (AUC = 0.99, Fig. 1I). The preclinical UC signature had a significant, but albeit lower, predictive capacity in the discovery (AUC=0.77), validation (AUC=0.67) and inception cohort (AUC=0.90, Fig. 1G-I).15 of 17 proteins associated with preclinical IBD demonstrated significantly higher intra-pair correlation coefficients in healthy monozygotic- compared to dizygotic twin pairs, indicating an influence from genetic factors on the regulation of these protein markers. The preclinical signature for CD demonstrated an AUC of 0.87 when comparing twins with preclinical CD (n=10) to matched external healthy twins. However, its predictive capacity was lower when comparing preclinical CD twins with their healthy twin siblings (AUC=0.58), i.e., when accounting for genetic and shared environmental factors. The difference in AUC estimates in the twin cohort was not significant (P=0.07).

The present study combined a supervised machine learning framework with an unsupervised method, finite mixture modeling, to identify prognostically meaningful subgroups of diverse chronic pain patients undergoing interdisciplinary treatment. Questionnaire data collected at pre-treatment and 1-year follow up from 11,995 patients from the Swedish Quality Registry for Pain Rehabilitation were used. Indicators measuring pain characteristics, psychological aspects, and social functioning and general health status were used to form subgroups, and pain interference at follow-up was used for the selection and the performance evaluation of models. A nested cross-validation procedure was used for determining the number of classes (inner cross-validation) and the prediction accuracy of the selected model among unseen cases (outer cross-validation). A four-class solution was identified as the optimal model. Identified subgroups were separable on indicators, predictive of long-term outcomes, and related to background characteristics. Results are discussed in relation to previous clustering attempts of patients with diverse chronic pain conditions. Our analytical approach, as the first to combine mixture modeling with supervised, targeted learning, provides a promising framework that can be further extended and optimized for improving accurate prognosis in pain treatment and identifying clinically meaningful subgroups among chronic pain patients.

Sepsis is a time dependent condition. Screening tools based on clinical parameters have been shown to increase the identification of sepsis. The aim of current study was to evaluate the additional predictive value of immunological molecular markers to our previously developed prehospital screening tools. This is a prospective cohort study of 551 adult patients with suspected infection in the ambulance setting of Stockholm, Sweden between 2017 and 2018. Initially, 74 molecules and 15 genes related to inflammation were evaluated in a screening cohort of 46 patients with outcome sepsis and 50 patients with outcome infection no sepsis. Next, 12 selected molecules, as potentially synergistic predictors, were evaluated in combination with our previously developed screening tools based on clinical parameters in a prediction cohort (n = 455). Seven different algorithms with nested cross-validation were used in the machine learning of the prediction models. Model performances were compared using posterior distributions of average area under the receiver operating characteristic (ROC) curve (AUC) and difference in AUCs. Model variable importance was assessed by permutation of variable values, scoring loss of classification as metric and with model-specific weights when applicable. When comparing the screening tools with and without added molecular variables, and their interactions, the molecules per se did not increase the predictive values. Prediction models based on the molecular variables alone showed a performance in terms of AUCs between 0.65 and 0.70. Among the molecular variables, IL-1Ra, IL-17A, CCL19, CX3CL1 and TNF were significantly higher in septic patients compared to the infection non-sepsis group. Combing immunological molecular markers with clinical parameters did not increase the predictive values of the screening tools, most likely due to the high multicollinearity of temperature and some of the markers. A group of sepsis patients was consistently miss-classified in our prediction models, due to milder symptoms as well as lower expression levels of the investigated immune mediators. This indicates a need of stratifying septic patients with a priori knowledge of certain clinical and molecular parameters in order to improve prediction for early sepsis diagnosis.Trial registration: NCT03249597. Registered 15 August 2017.