Multi-Omics Integration Links NID2 to Colorectal Cancer

Depression travels with colorectal cancer far more often than chance would predict, yet separating cause from consequence has been hard, since a cancer diagnosis itself can drag mood down. A 2026 iScience study used multi-omics integration to break that loop, pairing plasma protein quantitative trait loci with genome-wide association data for colorectal precancer and cancer. Across a 2,857-patient colonoscopy cohort and 37,597 NHANES participants, the team chased the depression-cancer signal down to one basement membrane protein, nidogen-2 (NID2).

How Multi-Omics Integration Pinpointed a Shared Protein

The analysis stacks several data types that rarely sit in one paper. Two-sample Mendelian randomization tested whether genetically predicted depression actually pushes colorectal risk, rather than the reverse. Bayesian colocalization then scanned 2,940 plasma proteins from the UK Biobank Pharma Proteomics Project, asking which protein-coding loci share a causal variant with precancer and cancer risk. Building on the colocalization framework described by Giambartolomei and colleagues (2014), the authors layered tissue evidence from CPTAC proteomics and TCGA transcriptomics on top of the genetic signal. This cross-omics stacking is what moves the work past a simple association toward a mechanism you can name.

Key Findings

• A dose-response that holds up: Higher depression severity tracked higher colorectal risk in both cohorts, with major depression carrying 2.3-fold odds of colorectal cancer after adjustment for BMI, diabetes, smoking, and diet.

• Mendelian randomization points one direction: Genetically predicted depression associated with raised precancer and cancer risk, and leave-one-out tests showed no single variant drove the result.

• Colocalization crowned NID2: Among 2,940 proteins, nidogen-2 showed the strongest shared genetic signal with colorectal lesions (posterior probability 0.992), ahead of CHRDL2, UBAC1, MAP4K5, and PDE5A.

• One protein, two readouts, one direction: NID2 fell in major depression plasma (0.49 vs 3.33 ng/mL), dropped stepwise from normal mucosa to precancer to tumor, and separated cancer from normal tissue in CPTAC with an AUC of 0.840.

Figure 1. Participant selection across the two cohorts that anchor the analysis. Panel A traces the prospective Renmin Hospital colonoscopy cohort, starting from 2,913 enrolled patients, removing 56 for prior high-grade neoplasia, serious comorbidity, recent major stressors, or substance and psychiatric history, and leaving 2,857 eligible participants split by depression severity into non-depression (n=2,524), minimal (n=218), minor (n=86), and major (n=29). Panel B follows the NHANES 2005 to 2023 sample from 88,429 down to 37,597 after sequential exclusions for age and missing covariates, grouped by the same four depression categories. Adapted from Huang et al. (2026), iScience.

Mendelian Randomization and the Case for Causation

Reverse causality is the usual objection here, because the stress of a cancer diagnosis can itself produce depressive symptoms. The authors sidestep it cleverly by anchoring much of the work in colorectal precancer, lesions that are typically found at routine colonoscopy and carry little of that diagnostic dread. In the discovery set, inverse-variance weighted MR linked depression liability to precancer (beta 0.24, p 0.004), and a separate replication cohort reproduced the direction. The causal estimate for frank cancer rested on a single instrument in the discovery stage, so it is the weaker leg of the argument and deserves a larger instrument set before anyone over-reads it.

From pQTL Signal to a Tissue-Level Biomarker

A genetic hit means little without protein-level confirmation, and this is where the layering pays off. Plasma protein profiling by ELISA showed NID2 roughly seven times lower in major depression than in matched controls, while immunohistochemistry across 94 colorectal specimens traced a clean decline from healthy mucosa through precancer to invasive tumor. CPTAC proteomics agreed, and a receiver operating characteristic analysis put the diagnostic AUC at 0.840 with 79% sensitivity. The ELISA arm rests on ten samples per group, so the effect sizes are striking but want a larger replication before NID2 reads as clinic-ready.

Why Single-Sample Multi-Omics Changes the Read

This study leans on three molecular layers that were generated by separate consortia, then stitched together statistically, and the seams show: pQTL data, tissue proteomics, and transcriptomics each came from different people. Pulling proteomic and metabolomic readouts from one aliquot, the way the Omni-MS workflow at Dalton does, sidesteps the cross-cohort batch effects that force this kind of study to lean so heavily on colocalization priors. When the protein signal and the expression signal come from the same blood draw, a candidate like NID2 needs far less statistical scaffolding to believe.

Frequently Asked Questions

What is multi-omics integration in biomarker research?

Multi-omics integration combines data from different molecular layers, such as the genome, the plasma proteome, and the transcriptome, into one analysis. The goal is to find signals that agree across layers, which makes a candidate biomarker more believable than any single measurement alone.

How does NID2 connect depression to colorectal cancer?

Nidogen-2 is a basement membrane protein that sits low in the plasma of people with major depression and falls further as colorectal tissue moves from normal to precancer to tumor. The study positions it as a shared mediator, though the exact mechanism still needs functional testing.

What samples does a multi-omics integration study need?

It draws on large genetic association datasets, tissue proteomics, and modest clinical sample sets for validation, here plasma and biopsy tissue. Public consortia like UK Biobank and CPTAC supply most of the scale, while targeted ELISA and immunohistochemistry confirm the protein in patients.

Conclusion

What's solid is the convergence: genetics, plasma proteomics, and tissue staining all point to NID2 sitting lower along the depression-to-cancer path. Less settled is causation at the protein level, because no functional experiment yet shows that restoring NID2 changes tumor behavior. For groups building risk-stratification panels, a depression-aware protein like this one is worth a slot on the discovery plate.

Citation

Huang, L., Wang, J., Tian, H., Wang, W., Nie, Z., Cui, X., Zhang, L., Jin, Y., Ding, Y., Liu, Z., Yu, H., & Wu, L. (2026). Integrative omics identify NID2 as a therapeutic target linking depression and colorectal cancer in humans. iScience, 29(6), 116003. https://doi.org/10.1016/j.isci.2026.116003

Note

This blog post summarizes findings from the above-cited research. Figures are adapted from the original publication. For full details, please refer to the source article.

By Seungjun Yeo, CEO at Dalton Bioanalytics. Specializing in multi-omics mass spectrometry for drug discovery and biomarker research.