LL-37 runs high in oral cancer and low in severe sleep apnea

Saliva levels of LL-37 may help separate oral cancer from less serious changes in the mouth. LL-37 is a small immune protein the body makes in white blood cells and the cells that line surfaces, and it can kill microbes and call in other immune cells ¹. Oral squamous cell carcinoma is the most common type of mouth cancer, and oral potentially malignant disorders are spots that can turn into cancer but are not cancer yet ⁴. The study compared saliva from 80 people in four groups: oral cancer, tobacco-associated oral potentially malignant disorders, non-tobacco-associated oral potentially malignant disorders, and healthy controls ⁴. Researchers used an enzyme-linked immunosorbent assay, or ELISA, a test that measures how much of a protein is present in a fluid ⁴. Salivary LL-37 rose step by step from the healthy group to the cancer group, a result that was statistically significant (p < 0.001), meaning the difference was unlikely to be due to chance ⁴. The test identified cancer cases with 100% sensitivity and 90% specificity, meaning it caught every cancer case but flagged one in ten healthy people as positive ⁴. The limitation is that 80 people is a small sample, and the saliva was measured once, so the study cannot show whether LL-37 drives the cancer or appears after the disease begins.

Blood levels of LL-37 fell as sleep apnea became more severe. Obstructive sleep apnea is a condition where breathing pauses during sleep. The study was a clinical observational study with 58 people who had obstructive sleep apnea and 20 healthy controls ⁵. The team used the apnea-hypopnea index, or AHI, a count of breathing pauses per hour, to split participants into two groups: AHI below 15 and AHI at or above 15; this was the cut point used in the study ⁵. Plasma LL-37 was higher in the AHI below 15 group (75.3 ± 29.3 nanograms per liter) than in the AHI at or above 15 group (47.1 ± 18.3 nanograms per liter) ⁵. The authors also reported that plasma LL-37 was significantly reduced in severe sleep apnea compared with healthy controls ⁵. The study found negative correlations between plasma LL-37 and age, the percentage of sleep spent in the lightest stage, arousal index (how often sleep is interrupted), AHI, and oxygen desaturation index (how often blood oxygen drops), and positive correlations with deep sleep time, REM sleep time, and average and minimum oxygen saturation ⁵. In plain terms, people who spent more time in light sleep and had more oxygen drops had lower plasma LL-37 ⁵. A multiple regression analysis (a statistical method that checks which factor matters while holding the others constant) found that the percentage of light sleep was the independent factor most associated with plasma LL-37 ⁵. The limitation is that this was a cross-sectional comparison, so it cannot show that poor sleep lowers LL-37 or that low LL-37 worsens sleep.

The third study looked at LL-37 in rheumatoid arthritis, and its clearest result was a shift in cholesterol risk. Rheumatoid arthritis is a long-term autoimmune disease that causes joint inflammation. The study was a case-control design with 120 people who had rheumatoid arthritis and 60 healthy controls ⁶. Researchers measured LL-37, toll-like receptor 9, a protein that detects DNA and can push inflammation forward, and several lipid risk indices ⁶. The arthritis group had higher disease activity and higher inflammatory markers such as high-sensitivity C-reactive protein and erythrocyte sedimentation rate (p < 0.001) ⁶. Total cholesterol and triglycerides were not significantly different, but the atherogenic index of plasma, a way of estimating cholesterol particle risk, was higher in arthritis patients (0.306 vs. 0.20, p = 0.004), and the atherogenic coefficient was also higher (p < 0.001) ⁶. This suggests a shift toward small dense LDL cholesterol, which is often linked to higher cardiovascular risk ⁶. The limitation is that the published summary does not state how LL-37 itself differed between groups, so the role of LL-37 in this cholesterol pattern is not yet clear from this source.

These new readings sit next to older biomarker studies that mostly pointed the other way. In 2009, researchers found higher LL-37 in the gum fluid of people with chronic periodontitis, a serious gum infection ⁷. In 2023, a case-control study reported that LL-37 could help predict periodontitis, but smoking clouded the picture, as smoking was linked to lower LL-37 even when periodontitis was present ⁸. The new 2026 oral cancer study also found higher LL-37, this time in saliva across the spectrum from healthy tissue to cancer ⁴. The new 2026 sleep apnea study found lower plasma LL-37 in more severe disease, which runs against the older infection and inflammation pattern ⁵. The 2026 arthritis study is less clear because the summary does not report how LL-37 changed ⁶. Most of the data lean toward higher LL-37 in active inflammation and early cancer, while the sleep apnea finding is a clear exception that may reflect chronic oxygen loss.

Taken together, the studies show LL-37 is not a single-purpose marker. It reflects what tissue or fluid is being sampled and what disease process is active. Saliva may capture local changes in the mouth, where LL-37 rises with injury and cancer ⁴. Blood may capture systemic effects of repeated oxygen loss during sleep, where LL-37 falls ⁵. This is why a one-size reading would be misleading. The value so far is in tracking LL-37 within a specific disease context, not in using it as a general health score.

The studies cannot say whether LL-37 causes the changes or simply tags along with them. Saliva and blood were compared at one point, not over time, which limits any causal claim ⁴. The arthritis summary leaves the direction of LL-37 open ⁶. The new studies did not test LL-37 alongside other immune-active or repair peptides such as the KPV research summary or the BPC-157 research summary, so no combination lessons can be drawn. Larger studies would need to follow people before disease, through diagnosis, and after treatment to see whether LL-37 can guide screening or care. Until then, these are early biomarker clues, not a test a person should seek out.