Low-Level Light Therapy for Dry Eye Disease in Thyroid Eye Disease: A Randomized, Double-Masked, Sham-Controlled Trial

Dry eye disease (DED) affects the majority of patients with thyroid eye disease (TED) and frequently persists despite maximal conservative therapy. Low-level light therapy (LLLT), or photobiomodulation, is a non-invasive treatment with anti-inflammatory and pro-reparative properties whose value in autoimmune, chronically driven ocular surface disease is unknown. In this prospective, randomized, double-masked, sham-controlled trial, eyes of patients with TED-associated DED that remained symptomatic despite intensive preservative-free lubrication, bedtime ophthalmic ointment and daily periocular hygiene were allocated 2:1 to periocular LLLT (630 nm LED mask, 15 min per session, applied over closed eyelids) or to sham treatment (identical procedure with light emission reduced to <30% of therapeutic output). Background ocular surface therapy was maintained unchanged in both arms. Assessments were performed at baseline (T0), weekly during treatment (T1–T3) and at one month (T4) and two months (T5) after the last session. The primary outcome was the Ocular Surface Disease Index (OSDI); secondary outcomes were non-invasive break-up time (NIBUT), fluorescein break-up time (FBUT), Schirmer I test, tear osmolarity, upper- and lower-eyelid meibography, corneal fluorescein staining and lipid-layer grade. The primary OSDI assessment was the change from baseline to T3, the end-of-treatment visit. Additional follow-up assessments were performed at T4 and T5. Forty-six eyes of 46 patients were analyzed (30 LLLT, 16 sham); baseline disease was active in 14/30 (46.7%) and 6/16 (37.5%) eyes, respectively. At T3, adjusted OSDI was lower with LLLT than with sham (between-group difference −7.49 points; 95% CI −11.66 to −3.32; p < 0.001). The difference persisted at T4 and T5 (−11.97 and −16.86 points, respectively; both Holm-adjusted p < 0.001). Tear osmolarity was lower with LLLT at T4 (−10.91 mOsm/L; 95% CI −18.96 to −2.87; p = 0.009) and T5 (−13.81 mOsm/L; 95% CI −22.55 to −5.06; p = 0.004). No multiplicity-adjusted between-group differences were demonstrated for NIBUT, FBUT, Schirmer I test or meibography. The adjusted probability of corneal staining was numerically lower with LLLT (T4: 15.7% vs. 34.7%; risk difference −19.0 percentage points; p = 0.100). In exploratory analyses, the treatment-by-visit-by-activity interaction was significant (p = 0.0159), with OSDI differences of −21.45 and −25.90 points in active disease versus −5.50 and −10.09 points in inactive disease at T4 and T5. CAS and OSDI were positively correlated at baseline (Spearman ρ = 0.547; Holm-adjusted p < 0.001) and at T5 (Spearman ρ = 0.428; Holm-adjusted p = 0.003). No treatment-related adverse events occurred. In conclusion, in TED-associated DED refractory to maximal conservative therapy, LLLT produced a significant and progressively increasing reduction in symptom burden together with an objective reduction in tear hyperosmolarity. Effects on break-up time, aqueous secretion, gland morphology, and lipid layer grade were not demonstrable within two months. The pattern and pace of response differ from those observed with the same technology after cataract surgery and are consistent with a chronic, systemically driven disease in which the ocular surface insult is continuously renewed.