P0817 Large-scale clustering of longitudinal faecal calprotectin and C-reactive protein profiles in Scottish and Danish inflammatory bowel disease
Bibliographic record
Abstract
Abstract Background Despite major advances in our understanding of the molecular underpinning of IBD, we remain challenged in deciphering the observed disease heterogeneity. Attempts to characterise longitudinal disease behaviour have either been restricted to symptom-based profiling - as exemplified in the IBSEN cohorts - or reliance on stochastic endpoints of disease “progression” such as hospitalisation, surgery and fibrosis. Here we present a novel method to examine IBD disease behaviour by modelling longitudinal inflammatory patterns in two large, well-characterised cohorts. Methods We conducted a retrospective study in two population cohorts, the Lothian IBD Registry (LIBDR),1 based in Scotland, and IBD patients described in national Danish registries.2,3 Using latent class mixed models,4 we independently clustered subjects by their FC and CRP profiles across 7 and 5 year post-diagnosis for the Lothian and Danish data respectively. The number of assumed clusters for the Danish models were dictated by the number chosen for the LIBDR data. Montreal classification was available for the LIBDR, and prescribing data were available for both cohorts. Results 1036 LIBDR subjects (544 CD, 492 UC/IBDU) and 7880 Danish subjects (3931 CD, 3949 UC) were included in the FC analysis. In the LIBDR, 10545 (median 9 per subject, IQR 6–13) FC observations were available, with 67986 observations available in the Danish registry (7, 4-11). For the CRP analysis, 1838 (805 CD, 1033 UC/IBDU) LIBDR subjects and 10041 (4415 CD, 5626) Danish subjects were included, with 49364 (median 20, IQR 10-36) and 241689 (19, 9-32) CRP observations respectively. When modelling the FC data, we found eight clusters (Figure 1), with the major clusters (FC1-4,7,8) replicating in the Danish data (Figure 2). The longitudinal trajectories characterising the clusters appear to reflect commonly reported clinical behaviours (e.g rapid remitters, delayed remitters, relapsing remitters and non-remitters). No association was found between ileal vs a colonic disease and cluster assignment. The use and timing of advanced therapies differed by cluster with rapid remitters more likely to receive therapy earlier in the disease course. However, prescribing trends described only a proportion of cluster assignments. When modelling the CRP data, we again found eight clusters, although there was broadly poor agreement between FC and CRP clusters. The follow-up required for reliable cluster assignments depended on the shape of the trajectories. Conclusion Distinct patterns of inflammatory behaviour over time are evident in patients with IBD. These data pave the way for a deeper understanding of disease heterogeneity in IBD and enhanced patient stratification in the clinic. References 1.Jones GR, Lyons M, Plevris N, et al. IBD prevalence in Lothian, Scotland, derived by capture–recapture methodology. Gut. 2019;68(11):1953-1960. doi:10.1136/gutjnl-2019-318936 2.Arendt JFH, Hansen AT, Ladefoged SA, Sørensen HT, Pedersen L, Adelborg K. Existing data sources in clinical epidemiology: Laboratory information system databases in Denmark. Clin Epidemiol. 2020;Volume 12:469-475. doi:10.2147/CLEP.S245060 3.Vestergaard MV, Allin KH, Poulsen GJ, Lee JC, Jess T. Characterizing the pre-clinical phase of inflammatory bowel disease. Cell Rep Med. 2023;4(11):101263. doi:10.1016/j.xcrm.2023.101263 4.Proust-Lima C, Philipps V, Liquet B. Estimation of extended mixed models using latent classes and latent processes: The R package lcmm. J Stat Softw. 2017;78(2):1-56. doi:10.18637/jss.v078.i02
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.006 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".