rs3791033 — KDM4A

KDM4A — When Your Epigenome Shapes Your Opioid Response

Pain and the medications used to treat it are profoundly personal. The same opioid dose that barely touches one person's pain may overwhelm another's reward circuits. Part of that individual variation is genetic — and some of it traces back to the chromatin architecture of pain-relevant genes themselves. rs3791033 is an intronic variant¹¹ intronic variant
A change within a non-coding region of a gene, located within intron 8 of KDM4A at position c.1163+387 on chromosome 1 in KDM4A, a histone demethylase that regulates which genes are switched on or off in the brain's pain and reward systems. The C allele was identified as one of only two genome-wide significant variants for problematic opioid prescription use in one of the largest studies ever conducted on opioid genetics.

The Mechanism

KDM4A (lysine demethylase 4A²² lysine demethylase 4A
Also known as JMJD2A or JHDM3A; the gene encodes a JmjC-domain Jumonji family demethylase, OMIM 609764) removes trimethyl marks from histone H3 at lysines 9 and 36 (H3K9me3 and H3K36me3). These marks are repressive — regions of chromatin bearing them are condensed and transcriptionally silent. By erasing these marks, KDM4A opens chromatin at specific gene promoters and allows transcription to proceed. At pain-relevant loci and in dopaminergic reward circuits — where H3K9 methylation status tightly controls gene expression — variation in KDM4A activity can shift the transcriptional landscape of pain processing and opioid reward.

rs3791033 lies in an intron, so it does not alter the KDM4A protein sequence. Its effect is likely regulatory: altering splicing efficiency, intron retention kinetics, or binding of regulatory RNA-binding proteins, any of which could modulate KDM4A expression levels in relevant tissues. The mechanism by which this specific intronic change influences opioid use behavior remains under investigation.

The Evidence

Sanchez-Roige et al. (2021)³³ Sanchez-Roige et al. (2021)
PMID 34728798, Molecular Psychiatry, N=132,113 European-ancestry 23andMe participants, ~21% cases conducted a GWAS of "problematic opioid prescription use" — defined as using prescribed opioids not as prescribed — as a proxy phenotype for opioid use disorder risk. Among 132,113 participants, only two loci reached genome-wide significance (p ≤ 5×10⁻⁸): rs3791033 near KDM4A and rs640561 near LRRIQ3. The genetic correlation between this phenotype and clinically diagnosed opioid use disorder was rg = 0.64–0.80, validating the proxy approach. Additional genetic correlations were found with alcohol dependence (rg = 0.74), chronic pain, depression, and insomnia — consistent with the known comorbidity structure of opioid misuse.

The study did not report specific odds ratios or beta coefficients for the lead variants in the abstract, which is a limitation. The finding represents discovery-level evidence: genome-wide significant in a large, well-powered study, but requiring replication in independent clinical cohorts with direct OUD diagnosis and functional follow-up.

Contextualizing the biology, Holen et al. (2023)⁴⁴ Holen et al. (2023)
PMID 37252880, Addiction Biology, ~15,756 OUD cases and ~600,000 psychiatric controls identified 14 novel OUD loci using a cross-disorder conditional FDR approach, including DRD2, FURIN, and MHC region — establishing shared genetic architecture between OUD and major psychiatric disorders. KDM4A's epigenetic regulatory role connects this variant to the broader landscape of transcriptional dysregulation in addiction neuroscience.

Practical Actions

The C allele of rs3791033 signals higher individual susceptibility to problematic opioid use — using opioids in ways that deviate from prescribed instructions, which is an early behavioral marker for developing opioid use disorder. This does not mean opioid medications should never be used; rather, it suggests heightened vigilance when opioids are prescribed for pain management.

CC homozygotes carry two copies of the risk allele and face the greatest concern. Discussing genetic risk explicitly with treating physicians allows for proactive steps: careful dosing titration, shorter prescription durations, preference for non-opioid first-line analgesics where clinically appropriate, and early monitoring for signs of dose escalation or dependence. CT heterozygotes carry an intermediate signal worth noting, particularly for anticipated longer-term opioid exposure (post-surgical recovery, chronic pain management).

Non-opioid pharmacological alternatives include NSAIDs, COX-2 inhibitors, anticonvulsants (gabapentin, pregabalin), SNRIs (duloxetine for neuropathic pain), tricyclic antidepressants for neuropathic and central sensitization pain, and topical agents (lidocaine patches, capsaicin). Interventional options — nerve blocks, spinal cord stimulation — are worth earlier consideration in surgical and chronic pain contexts for individuals with elevated genetic opioid risk.

Interactions

rs3791033 has been studied as an independent locus. The GWAS that identified it found a second significant hit at rs640561 near LRRIQ3, though these are on different chromosomes and are not in linkage disequilibrium. The broader opioid use disorder genetic landscape overlaps with OPRM1 rs1799971 (the A118G variant, the best-studied opioid receptor variant, which alters mu-opioid receptor binding affinity) — though rs3791033 and OPRM1 act through distinct mechanisms (epigenetic regulation vs. receptor function) and both may contribute independently to overall opioid response phenotype.