Short noncoding RNAs as predictive biomarkers for the development from inflammatory bowel disease unclassified to Crohn's disease or ulcerative colitis

Abstract

Numerous pathogenic processes are mediated by short noncoding RNAs (sncRNA). Twenty percent of inflammatory bowel disease (IBD) patients are labelled as IBD unclassified (IBDU) at disease onset. Most IBDU patients are reclassified as Crohn's disease (CD) or ulcerative colitis (UC) within few years. Since the therapeutic methods for CD and UC differ, biomarkers that can forecast the categorization of IBDU into CD or UC are highly desired. Here, we investigated whether sncRNAs can predict CD or UC among IBDU patients. 35 IBDU patients who were initially diagnosed with IBDU were included in this retrospective investigation; of them, 12, 15, and 8 were reclassified into CD (IBDU-CD), UC (IBDU-UC), or remained as IBDU (IBDU-IBDU), respectively. Eight IBD patients, were included as references. SncRNA profiling on RNA from mucosal biopsies were performed using Affymetrix miRNA 4.0 array. Selected probe sets were validated using RT-qPCR. Among all patients and only adults, 306 and 499 probe sets respectively were differentially expressed between IBDU-CD and IBDU-UC. Six of the probe sets were evaluated by RT-qPCR, of which miR-182-5p, miR-451a and ENSG00000239080 (snoU13) together with age and sex resulted in an AUC of 78.6% (95% CI: 60-97) in discriminating IBDU-CD from IBDU-UC. Based on the three sncRNAs profile it is possible to predict if IBDU patients within 3 years will be reclassified as CD or UC. We showed that the expression profile of IBDU patients differ from that of definite CD or UC, suggesting that a subgroup of IBDU patients may compose a third unique IBD subtype.

Fig 1. Traditional IBD diagnosis and classification.

Patient cohort selection and workflow of the study are highlighted in yellow boxes.

Fig 2. Differential probe set expression analysis between IBDU-CD and IBDU-UC samples.

(A) A summary plot showing probe sets highly expressed or lowly expressed in IBDU-CD compared to IBDU-UC represented as blue and red dots, respectively. Probe sets with -1> log FC <1 is highlighted as larger dots and respective names are indicated. (B) Heatmap showing microarray-based expression of 6 probe sets. Age group, duration of IBD, tissue location, sex and disease type are shown as colored bars at the top of the heatmap. Expression levels are normalized to row means. (C) PCA plot showing IBDU-CD and IBDU-UC clustered under unsupervised clustering based the expression levels of the 6 probe sets. (D) PCA plot with IBDU-CD, IBDU-UC and IBDU-IBDU samples based the expression levels of the 6 probe sets. Different symbols represent different age groups among the samples.

Fig 3. Differential probe set expression analysis between adult IBDU-CD and IBDU-UC samples.

(A) Summary plot showing probe sets highly expressed or lowly expressed in IBDU-CD adults compared to IBDU-UC adults represented as blue and red dots respectively. Probe sets with -1> log FC <2 is highlighted as larger dots and respective names are indicated. (B) Heatmap showing microarray-based expression of the 10 probe sets. Duration of IBD, tissue location, sex and disease type are shown as colored bars at the top of the heatmap. Expression levels are normalized to row means. (C) PCA plot showing IBDU-CD and IBDU-UC clustered under unsupervised clustering based the expression levels of the 10 probe sets. (D) PCA plot with IBDU-CD, IBDU-UC and IBDU-IBDU samples based the expression levels of the 10 probe sets.

Fig 4. Expression levels determined by RT-qPCR.

Box plots showing RT-qPCR based relative expression levels of miR-182-5p, miR-451a, and ENSG00000239080 (SnoU13) in all IBDU-CD and IBDU-UC samples (A, B and C) and among only adult IBDU-CD and IBDU-UC samples (D, E and F).

Fig 5. A model based on the short noncoding RNA’s expression, and the age and sex of the patient for prediction of CD and UC development from IBDU.

(A) PCA plot showing samples clustered based on the model into IBDU-CD (green) and IBDU-UC (pink). IBDU-IBDU samples shown in blue color show a tendency towards IBDU-CD or IBDU-UC. (B) ROC plot showing prediction based on 3 individual candidate’s expression and the model based on the relative expression values of miR-182-5p, miR-451a, and ENSG00000239080 (SnoU13) along with the age at diagnosis and sex of the patients.