Does the inflammation marker CRP actually cause disease?
People born with gene variants that raise their C-reactive protein were not at higher risk of almost any of 32 diseases linked to high CRP, suggesting CRP is mostly a marker rather than a cause, with a surprising lower risk of schizophrenia.
Source
Investigating the Causal Relationship of C-Reactive Protein with 32 Complex Somatic and Psychiatric Outcomes: A Large-Scale Cross-Consortium Mendelian Randomization Study
Study at a glance
- Design
- Mendelian randomisation — Two genetic risk scores for CRP level (4 SNPs in the CRP gene; 18 genome-wide significant CRP SNPs) were applied to published GWAS summary statistics for 32 outcomes using inverse-variance-weighted estimation, with heterogeneity tests and stepwise SNP removal to handle pleiotropy, a Bonferroni threshold, and individual-level replication for schizophrenia.
- N
- No single N: outcome datasets ranged from 1,566 to 184,305 participants of European ancestry; the individual-level schizophrenia follow-up used 25,629 cases and 30,976 controls.
- Population
- Participants of European ancestry in large genome-wide association consortia for 32 autoimmune, cardiovascular, metabolic, neurodegenerative and psychiatric traits.
- Outcome
- Estimated causal effect of genetically raised CRP on risk of each of 32 diseases or traits (e.g., coronary artery disease, type 2 diabetes, Crohn disease, depression, schizophrenia, blood pressure).
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
The researchers used Mendelian randomisation, treating inherited gene variants that nudge CRP levels as a natural experiment. They built two genetic scores for CRP, one from 4 variants in the CRP gene and one from 18 variants across the genome, and applied them to published genetic association results for 32 conditions ranging from heart disease and diabetes to depression and Alzheimer disease. They tested for pleiotropy (variants affecting disease through routes other than CRP), used a strict multiple-testing threshold, and re-checked schizophrenia using individual-level genetic data.
What they found
For most outcomes genetically higher CRP showed no causal effect. Several nominal signals (including coronary artery disease, Crohn disease and blood pressure) disappeared once pleiotropic variants such as those in IL6R were removed. The one robust finding went in an unexpected direction: higher genetically predicted CRP was linked to lower schizophrenia risk (odds ratio 0.86 per 10% higher CRP), confirmed in individual-level data (odds ratio 0.96 per standard deviation of the score), whereas bipolar disorder showed a nominal increase (odds ratio 1.21) that did not survive correction.
The limits
What it doesn't show
The genetic scores explain only about 2% and 5% of variation in CRP, so small causal effects could still be missed, and the heterogeneity-based removal of variants is an imperfect fix for pleiotropy. All data came from people of European ancestry, results rest largely on summary statistics, and the schizophrenia effect is small and its mechanism speculative. A null MR result speaks to lifelong, modest genetic differences in CRP, not necessarily to short-term large rises during illness or to drugs that lower CRP.
Key terms
- Mendelian randomisation
- Using gene variants that affect an exposure as natural instruments to estimate whether that exposure causes a disease, because genes are allocated at conception and are largely free from confounding and reverse causation.
- C-reactive protein (CRP)
- A liver protein whose blood level rises with inflammation; widely used as a marker of disease risk.
- Instrumental variable
- A variable that affects the outcome only through the exposure of interest; here, CRP-raising gene variants.
- Pleiotropy
- When one gene variant affects several traits, so it might influence disease through a pathway other than the exposure, violating a key MR assumption.
- Reverse causation
- When the disease raises the supposed risk factor rather than the other way round, e.g. illness driving up CRP.
- Bonferroni correction
- Dividing the significance threshold by the number of tests (here 0.05 divided by 32) to limit false positives from multiple comparisons.
Flashcards
0 of 11 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What is the main advantage of Mendelian randomisation over an ordinary observational study of CRP?
Common questions
If high CRP predicts heart disease, how can it not cause it?
CRP could rise because of the underlying disease process or shared risk factors like obesity and smoking. MR avoids these by using genes, and people genetically prone to higher CRP did not have more heart disease once pleiotropic variants were removed.
Does this mean inflammation doesn't matter for these diseases?
No. It suggests CRP itself is not the causal molecule; other parts of the inflammatory pathway, such as IL-6 signalling, may still play causal roles.
Should we raise CRP to prevent schizophrenia?
No. The protective effect is small, needs replication and a mechanism, and it contrasts with observational studies showing raised CRP in patients, which may reflect reverse causation.
More on Cardiovascular