LDN and the Immune System: What the Science Actually Says

Yoon Hang Kim MD

Board-Certified in Preventive Medicine | Integrative & Functional Medicine Physician

About Dr. Kim

Dr. Yoon Hang Kim is a board-certified physician with more than 20 years of experience in integrative and functional medicine. He completed his integrative medicine fellowship at the University of Arizona under Dr. Andrew Weil and holds training in preventive medicine and medical acupuncture. He specializes in low-dose naltrexone (LDN), autoimmune conditions, chronic pain, integrative oncology, fibromyalgia, chronic fatigue syndrome, mast cell activation syndrome (MCAS), and mold toxicity. He is the author of 8 books and more than 25 peer-reviewed articles.

Professional: www.yoonhangkim.com | Clinical: www.directintegrativecare.com

Important: This article is for educational purposes only and is not medical advice. It does not create a physician–client relationship. Low-dose naltrexone is used off-label. Do not start, stop, or change any medication without talking to your own physician.

Introduction

If you have spent any time in integrative medicine circles, you have probably heard some version of the following: “LDN boosts endorphins, which restores immune balance.” It is a tidy explanation. It fits on a slide. And it is not quite right.

Low-dose naltrexone has become one of the most widely discussed off-label medications in functional and integrative medicine, and for understandable reasons. The clinical experience many of us have had with it — across autoimmune conditions, chronic pain, and complex inflammatory presentations — is often genuinely impressive. But the popular narrative about how LDN works has gotten ahead of what the science actually supports, and in some cases has drifted into territory that does not hold up under scrutiny.

This matters because our credibility as clinicians depends on distinguishing between a plausible biological hypothesis, a measured biomarker change, symptom relief, and demonstrated disease control. LDN has evidence at the first three levels in selected settings. At the fourth, the picture is considerably less certain — and a 2026 multicenter Crohn’s disease trial that stopped early for futility is a reminder that we cannot afford to be casual about that distinction.

This conclusion is not mine alone. A 2026 narrative review in Advances in Therapy (Gouda, Aitcheson, and Steadman) that examined the human LDN literature from 1989 to 2026 found that most of the evidence consists of case reports and small feasibility studies relying heavily on subjective outcomes, and concluded that current evidence does not support routine clinical use.

It helps to keep a simple hierarchy in mind as you read. Randomized placebo-controlled trials with clinical outcomes sit at the top. Below them come human biological signals (a biomarker that moves), then animal disease models, then cellular and mechanistic experiments, and finally clinical observation and prescriber experience. Evidence does not automatically travel upward: a finding in cell culture does not become a finding in people just because the mechanism is elegant.

This article walks through what the foundational research actually showed, the three proposed mechanisms of action, and what controlled clinical trials have — and have not — demonstrated. The goal is not to dismiss LDN but to equip clinicians and informed clients with a more accurate framework for understanding it.

The Origin Story: What the Early Experiments Actually Showed

The history of LDN is often told as a single narrative, but it is really three intertwined threads that have been blended together over time.

The first thread belongs to Ian Zagon and Patricia McLaughlin at Penn State, whose landmark experiments in the early 1980s involved naltrexone injections in neuroblastoma-bearing mice — not, as is sometimes claimed, beta-endorphin injections into rats demonstrating immune restoration. Zagon’s pivotal 1983 study in Science showed that naltrexone had dose-dependent effects on tumor growth: a dose that blocked morphine analgesia for only about four to six hours a day inhibited tumor development. A companion Cancer Letters paper that same year reported 69% tumor take, a 70% delay in tumor appearance, and a 60% increase in median survival.

The key insight was that the duration of opioid blockade mattered — intermittent blockade produced different biological effects than sustained blockade. But tumor growth inhibition in a mouse model is not, by itself, proof of improved immune function in humans.

A 1987 study complicated the popular “immune boosting” narrative. When Zagon tested LDN in athymic nude mice — animals with marked thymus-dependent T-cell immunodeficiency — the antitumor effects persisted. That showed the effect did not require intact thymus-dependent T-cell immunity and strengthened the case for a direct growth-regulatory component. It did not exclude immune contributions altogether, since nude mice retain innate immunity and other immune functions.

The second thread involves Met-enkephalin, a different opioid peptide that Zagon would later call “opioid growth factor” (OGF). This is not the same molecule as beta-endorphin. The growth-regulatory actions Zagon and colleagues studied are attributed to a nonclassical receptor, OGFr, which is distinct from the classical mu, delta, and kappa opioid receptors. Met-enkephalin itself can also act on classical opioid receptors, so the two signaling systems should not be treated as synonymous — and conflating either of them with beta-endorphin makes the mechanism sound more settled than it is.

The third thread is Bernard Bihari’s clinical use of low-dose naltrexone in people with HIV/AIDS during the 1980s. Bihari’s observations were clinically provocative, but they should not be treated as modern controlled evidence of immune restoration or antiviral efficacy.

Three Proposed Mechanisms

Mechanism 1: Transient opioid blockade and the “endorphin rebound”

Naltrexone antagonizes classical opioid receptors, with greatest affinity for the mu-opioid receptor. At the low doses used in LDN, the proposed sequence goes like this: LDN temporarily reduces endogenous opioid signaling, the body mounts compensatory changes in opioid peptide production or receptor expression, and after the antagonist activity wears off, you get enhanced endogenous opioid signaling.

This is where the “endorphin rebound” explanation comes from. It is an elegant hypothesis, but it requires several qualifications.

First, does LDN reliably increase beta-endorphin? Not consistently. A small, uncontrolled study in primary-progressive MS reported increased beta-endorphin in peripheral blood mononuclear cells during treatment — a limited human biological signal, not definitive proof of mediation. More importantly, a 2021 mouse study in eNeuro that specifically tested the hypothalamic POMC/beta-endorphin hypothesis found that LDN did not increase plasma beta-endorphin, POMC mRNA, or opioid-receptor sensitivity in the system examined.

That negative result does not exclude peripheral opioid effects or OGF-related mechanisms. But it does undermine the practice of presenting beta-endorphin elevation as the established explanation for everything LDN does.

Second, the claim that “LDN blocks receptors for exactly four hours” is an oversimplification. Naltrexone has an approximately four-hour plasma half-life, but its active metabolite, 6-beta-naltrexol, has an approximately thirteen-hour half-life. A plasma half-life is not a direct measurement of receptor blockade duration in each tissue or each person. Human receptor-occupancy data come largely from conventional naltrexone doses; occupancy over time at low doses has not been adequately characterized, so the precise timing that the “rebound” story depends on is itself an assumption.

Mechanism 2: The OGF–OGFr axis and lymphocyte proliferation

This is arguably the most coherent connection between Zagon’s animal work and the proposed autoimmune applications. OGF (Met-enkephalin) acts as a brake on cellular proliferation through OGFr. The proposed LDN mechanism is that intermittent blockade leads to compensatory enhancement of the OGF–OGFr system, which then produces stronger growth restraint when the blockade subsides.

Here the details matter. In cell culture, it is OGF itself that suppresses proliferation of activated T and B lymphocytes; sustained naltrexone exposure in those same systems did not directly suppress proliferation. The LDN finding comes from living animals: in EAE mice, intermittent low-dose naltrexone reduced expansion of peripheral T- and B-cell populations, which investigators interpreted as indirect modulation through the endogenous OGF–OGFr pathway. That is a reasonable laboratory rationale for limiting autoreactive cell expansion without conventional cytotoxic immunosuppression. It is not evidence that LDN selectively suppresses pathogenic lymphocytes while preserving protective immunity in people.

Animal studies in experimental autoimmune encephalomyelitis (EAE) — the standard mouse model for MS — have shown delayed onset and reduced disease expression, improvement in neurological deficits, and restoration of reduced circulating OGF concentrations. The cytokine findings were more nuanced than a simple “all inflammation goes down” pattern: one experiment found lower IFN-gamma, TNF-alpha, and IL-10, but higher IL-6. Immune modulation is context-dependent, not a one-directional switch.

These studies support a mechanistic hypothesis. They do not establish prevention of relapses, new MRI lesions, or disability progression in human MS.

Mechanism 3: TLR4, microglia, and innate inflammation

A third hypothesis does not require an increase in beta-endorphin at all. Experimental studies — particularly with the opioid-inactive (+)-naltrexone stereoisomer — show modulation of selected TLR4-related signaling pathways in microglia and other cells. Even here the effect is narrower than often described: detailed pharmacology suggests a biased effect on one downstream branch (the TRIF–IRF3 pathway) rather than global TLR4 inhibition, and (+)-naltrexone’s TLR4 potency is modest enough that chemists have pursued far more potent derivatives. Whether ordinary oral (−)-naltrexone at LDN exposures produces enough TLR4 modulation in humans to explain any therapeutic effect remains unestablished.

The stereochemistry research is instructive. Classical opioid receptors preferentially recognize one stereoisomer of opioid antagonists. Experimental positive stereoisomers can affect TLR4-related pathways without the same classical opioid-receptor antagonism, which helps researchers tease apart opioid from nonopioid mechanisms.

Two cautions follow. Findings involving experimental (+)-naltrexone or (+)-naloxone cannot automatically be treated as proof of how routine oral LDN works. And ordinary compounded LDN should not be described as a mixture in which one isomer “boosts endorphins” and another independently “blocks inflammation.” The stereoisomer experiments are mechanistic probes, not a description of a two-component formulation.

The central translational gap remains whether routine oral LDN achieves sufficient exposure at the relevant tissue targets to produce clinically meaningful anti-inflammatory effects.

What the Clinical Trials Actually Show

Crohn’s disease: early promise, then a negative trial

Crohn’s disease is an especially informative test case because trials can assess objective inflammation through endoscopy rather than relying solely on symptom scores.

An early 2011 randomized trial of 34 participants over 12 weeks showed clinical and endoscopic response favoring LDN, though clinical remission was not statistically significant. A 2018 Cochrane review covering two small trials (46 adults and children total) concluded that the evidence was low quality and insufficient for firm conclusions.

Then came the 2026 multicenter trial. Forty-one participants were randomized across seven Dutch hospitals to LDN or placebo for 12 weeks. The trial stopped early for futility. Endoscopic remission occurred in 5.6% of the LDN group versus 18.8% in the placebo group. Clinical remission also did not favor LDN. Fatigue improved — but fatigue improvement is not the same as suppression of intestinal inflammation.

The small size and early termination limit precision, but this trial substantially weakens the argument that LDN is an established anti-inflammatory treatment for active Crohn’s disease.

Multiple sclerosis: symptom studies, not disease modification

The MS clinical literature is small and largely symptom-focused. An uncontrolled 40-person primary-progressive MS pilot found improved spasticity and increased PBMC beta-endorphin, but did not establish overall clinical efficacy and did not test relapse, MRI, or disability endpoints. Of the randomized studies, a placebo-controlled crossover trial with 60 completers reported improvement in mental-health quality-of-life measures, while a separate 96-person randomized study found no significant quality-of-life difference.

Notably absent from this body of evidence: any demonstration that LDN reduces relapses, prevents new MRI lesions, or slows disability progression. The promising OGF/EAE animal findings remain incompletely translated to human disease-modifying outcomes.

Fibromyalgia: interesting signals, disappointing trials

Early small crossover studies suggested benefit, but larger and more recent parallel-group trials have been less encouraging. An eight-person uncontrolled biomarker study reported reductions in several circulating inflammatory mediators during treatment, but without a placebo group it cannot show that the cytokine changes caused symptom improvement.

A 2024 randomized trial of 99 women (Bruun et al., Lancet Rheumatology) found no superiority over placebo for the primary pain outcome after 12 weeks: a between-group difference of −0.34 points on a 0 to 10 scale (95% CI −0.95 to 0.27).

A common defense of that result was that the dose might have been too high, or 12 weeks too short. The 2026 INNOVA trial addressed both. It randomized 98 women to the more commonly prescribed dose or placebo, with follow-up to 12 months. At three months, pain fell by 0.33 points with LDN and 0.64 with placebo (adjusted difference 0.49, p = 0.236), and LDN showed no meaningful advantage on pain or key secondary outcomes through one year.

Meta-analyses have reached conflicting conclusions — a 2025 review reported a pain benefit, while another 2025 analysis found no significant difference from placebo — largely because they pool a handful of small, heterogeneous trials, several of them crossover designs. The adequately sized parallel-group trials have been the least favorable.

Fibromyalgia findings are also not direct evidence that LDN treats systemic autoimmune disease. Pain processing, neuroinflammation, and peripheral autoimmune activity are related but distinct questions.

Cancer and historical HIV

The early cancer animal studies remain mechanistically interesting, but human oncology evidence is very limited. The clinical cancer literature consists largely of case reports and small uncontrolled series, often involving multiple simultaneous interventions, and does not establish LDN as an effective anticancer treatment. Bihari’s HIV work, while historically important, should not be treated as modern controlled evidence of immune restoration.

A Clinically Defensible Framing

Language matters. The way we describe LDN to clients and colleagues shapes expectations, affects treatment decisions, and reflects on our professional credibility.

For clinician or client education, a defensible description — one that stays within the mechanistic and clinical evidence — sounds something like this:

Low-dose naltrexone is an off-label opioid antagonist being investigated for effects on endogenous opioid growth-regulatory pathways and innate inflammatory signaling. Laboratory studies suggest effects on lymphocyte proliferation and microglial activation, but its mechanism in humans is not fully established, and clinical efficacy varies by condition.

What to avoid saying:

• “LDN automatically boosts a weak immune system and suppresses an overactive one.” There is no evidence that any single medication acts as a perfectly calibrated immune thermostat.
• “LDN reliably raises endorphins by a fixed percentage.” The beta-endorphin data are inconsistent, and a 2021 study specifically failed to confirm this hypothesis.
• “LDN has been proven to restore immune balance.” Immunomodulation is biologically plausible. “Proven” and “balance” each overstate what controlled trials have demonstrated.
• “LDN can replace disease-modifying treatment for autoimmune disease.” Controlled clinical trials have not established LDN as a disease-modifying therapy that prevents relapse, structural organ damage, or disability progression in autoimmune disease.

The most important conceptual distinction is between biological plausibility, biomarker changes, symptom improvement, and disease control. Clinical experience, however compelling, is not the same as controlled evidence of disease modification. We owe our clients the honesty of that distinction.

The Bottom Line

LDN is neither a biologically implausible therapy nor an established disease-modifying treatment. It has a legitimate preclinical literature — particularly the OGF–OGFr growth-regulation pathway and the TLR4/microglial hypothesis — and intriguing clinical signals. But mechanistic plausibility has substantially outpaced confirmatory human evidence.

Many clinicians, myself included, have watched individual clients improve substantially on LDN. Those observations matter: they generate hypotheses and may help identify who responds. But on their own they cannot establish average treatment efficacy or mechanism, because natural disease fluctuation, regression to the mean, placebo and context effects, concurrent treatments, and selective recall can all produce real-looking improvement. That is exactly why randomized trials exist.

The field has a credibility problem when the popular explanation for how LDN works outpaces the evidence. Telling people that LDN “restores immune balance by boosting endorphins” is not supported by the weight of current data. A 2021 animal study failed to confirm the beta-endorphin hypothesis. Two randomized fibromyalgia trials — one in 2024 and one with a year of follow-up in 2026 — did not beat placebo on pain. A 2026 Crohn’s trial stopped early for futility. These are not reasons to abandon LDN, but they are reasons to stop talking about it as though the science is settled.

What we actually need is straightforward: larger, well-designed randomized controlled trials with objective disease-activity endpoints. We need to move beyond symptom questionnaires and into endoscopic outcomes, MRI lesion counts, biomarker panels, and disability measures. We need to identify which people are most likely to benefit, at what doses, and for how long.

Until then, I will continue prescribing LDN where the clinical context supports it — thoughtfully, with informed consent, and without promising more than the evidence can deliver. That is what intellectual honesty looks like in integrative medicine. It is what our clients deserve, and it is what will ultimately give LDN the rigorous evaluation it has earned.

Work with Dr. Kim

Dr. Kim sees a limited number of clients through a membership-based, insurance-free telemedicine practice, which keeps care focused and unhurried for people with complex conditions. His approach is integrative medicine built on a foundation of good conventional care: he works alongside your existing physicians and specialists, not as a replacement for them.

Membership is available to residents of Texas, Illinois, Missouri, Georgia, and Florida, with visit-by-visit guest consultations available in Iowa. To learn more or schedule a discovery call, visit www.directintegrativecare.com.

References

• Gouda AH, Aitcheson NEC, Steadman KJ. Low-dose naltrexone: what is the evidence? A narrative review. Advances in Therapy. Published April 30, 2026.
• Toljan K, Vrooman B. Low-dose naltrexone (LDN): review of therapeutic utilization. Medical Sciences. 2018;6(4):82. PMC6313374
• Zagon IS, McLaughlin PJ. Naltrexone modulates tumor response in mice with neuroblastoma. Science. 1983. doi:10.1126/science.6867737
• Zagon IS, McLaughlin PJ. Naltrexone and neuroblastoma tumor growth in mice. Cancer Letters. 1983. PMID 6640516
• Zagon IS, McLaughlin PJ. Low-dose naltrexone and neuroblastoma in athymic nude mice. Journal of the National Cancer Institute. 1987;78(1):141. PMID 3025501
• Cuitavi J, Hipólito L, Canals M. Reported benefits of low-dose naltrexone appear to be independent of the endogenous opioid system involving proopiomelanocortin neurons and β-endorphin. eNeuro. 2021;8(3). doi:10.1523/ENEURO.0087-21.2021
• Low-dose naltrexone for induction of remission in Crohn’s disease. Cochrane Database of Systematic Reviews. 2018. PMC6494424
• Multicenter randomized placebo-controlled trial of low-dose naltrexone in active Crohn’s disease. 2026. PMID 42701992
• Bruun KD, Christensen R, Amris K, et al. Naltrexone 6 mg once daily versus placebo in women with fibromyalgia: a randomised, double-blind, placebo-controlled trial. Lancet Rheumatology. 2024;6:e31–e39. doi:10.1016/S2665-9913(23)00278-3
• Rodríguez-Freire C, et al. Efficacy of low-dose naltrexone in women with fibromyalgia syndrome: a 12-month randomised, double-blind, placebo-controlled single-centre clinical trial (INNOVA study). European Journal of Pain. 2026;30(6):e70321. doi:10.1002/ejp.70321
• Efficacy and safety of low-dose naltrexone in fibromyalgia: a systematic review and meta-analysis. Annals of Medicine and Surgery. 2025;87(5):2928–2935. PMID 40337423

Yoon Hang Kim MD is the author of MCAS: Epidemic in Plain Sight and LDN Primer, both available on Amazon, and the founder of the LDN Support Group.

Professional: www.yoonhangkim.com | Clinical: www.directintegrativecare.com