This was a prospective, randomized clinical study. A total of 80 mild or moderate-to-severe TAO patients (80 eyes) with dry eye syndrome were selected from the Ophthalmology Department of the Ninth People’s Hospital Affiliated to the Medical College of Shanghai Jiao Tong University from May to October 2020. The disease stages of all subjects were inactive. The severely affected eye in each patient was considered in the study. The inclusion criteria were as follows: ① Diagnosed as mild or moderate-to-severe TAO and the disease stage was inactive.; ② Diagnosed as dry eye syndrome. The exclusion criteria were as follows: ① Suffering from other ophthalmic diseases affecting the ocular surface; ② Suffering from autoimmune diseases, such as systemic lupus erythematosus, rheumatoid arthritis, and Sjögren’s syndrome; ③ Previous ocular trauma or surgery; ④ Extremely severe TAO; ⑤ Allergic to drug ingredients. The content of this study was approved by the medical ethics committee of our hospital (approval No. sh9h-2019-t343-2, approved on March 18th, 2020) and registered in the Chinse clinical trial registry (Registration no. chictr2000033874, approved on 15/06/2020). Informed consent was signed by all subjects.

TAO was diagnosed according to the Bartley standard [22]. If eyelid retraction was detected, one of the following three signs could be diagnosed as TAO: ① Exophthalmos; ② History of thyroid dysfunction; ③ Extraocular muscle involvement. If eyelid retraction was not found, all the three conditions should be fulfilled: ① History of thyroid dysfunction; ② Combined with exophthalmos, extraocular muscle involvement, or visual dysfunction; ③ Excluding eyelid retraction, exophthalmos, and eye movement disorder caused by other eye diseases.

Mild TAO was defined as patients who could not meet the diagnostic criteria of moderate-to-severe TAO. Moderate-to-severe TAO was defined if a patient had no visual impairment but at least two of the following manifestations: ① Eyelid retraction: upper lid margin to reflex distance (MRD-1) was > 5 mm or a lower lid margin to reflex distance (MRD-2) was > 6 mm. ② Moderate-to-severe soft tissue damage (one of the following: moderate-to-severe eyelid swelling or moderate-to-severe conjunctival congestion); ③ At least 3 mm over the normal value of proptosis (Hertel exophthalmos of any eye ≥ 17 mm); ④ Stable or intermittent diplopia.

Extremely severe TAO was defined as patients with compressive optic neuropathy (DON) and/or severe exposure keratopathy.

The following was the judgment criteria for active and inactive TAO [23]. The clinical activity score (CAS) was used to evaluate active and inactive TAO. The scale comprised seven items, including eyelid edema, eyelid congestion, conjunctival congestion, conjunctival edema, lacrimal caruncle redness and swelling, spontaneous retrobulbar pain, and pain during gaze or eye movement. Each of the items was scored 1 point, < 3 points as inactive period, and ≥ 3 points as the active period.

TAO with dry eye syndrome was defined as fulfilling any of the following items based on the diagnosis of TAO [24]: ① having one of the subjective symptoms (visual fatigue, discomfort, foreign body, dryness, burning, and visual acuity fluctuation) and Schirmer I test (ST) ≤ 10 mm; ② One of the subjective symptoms and break-up time (BUT) ≤ 10 s; ③ Positive fluorescein staining of keratoconjunctiva.

0.1% sodium hyaluronate eye drop: Sodium hyaluronate of 2 million to 3 million Daltons is the main component. The concentration is 0.1% and it does not contain preservatives. Inherent viscosity is 2.4–3.2m3/kg.

Vitamin A palmitate eye gel: Vitamin A palmitate of 524.87 Daltons is the main component. The concentration is 0.8 mg/g and it contains preservatives. Inherent viscosity is 3.2–4.2m3/kg.

In the pre-experiment, 10 patients who applied Vitamin A palmitate eye gel got an effective rate of 60% in improving dry eye while 10 who applied 0.1% sodium hyaluronate eye drop got 30%. We let the inspection level α = 0.049, inspection efficiency 1- β = 0.80. Considering the possible loss of follow-up rate as 20%, the total sample size of this study was 80 and 40 in each group.

In this study, 80 patients (80 eyes) were randomly divided into Group A and Group B by the envelope method. Vitamin A palmitate eye gel was applied in Group A three times/day, one drop each time. Group B was given 0.1% sodium hyaluronate eye drop three times/day, one drop each time. The effect was evaluated 1 month after drug administration. During the treatment, other topical eye drugs were not allowed.

The following data of subjects were collected at baseline and after 1 month of treatment. In consideration of the actual situation, clinical data were collected three days before or after one month was allowed. Generally, gel or eye drops were not used on the day of measurement, and the measurement time was at the same time of the day at baseline or post treatment.

The inverted end of the long tear detection filter paper was placed about 5 mm into the outer 1/3rd of the subject's lower eyelid conjunctival sac without ocular surface anesthesia, and the subject was asked to close his eyes gently. After 5 min, the filter paper was taken down, and the ST value was recorded.

Corneal topography (Keratograph 5 M, type 77,000, Germany) was applied. After the subject placed his head in the specified area, the BUT function option of the software, OCULUS Keratograph, was selected. The subject was instructed to look forward without blinking. The value of BUT was obtained after the examination.

The range of fluorescein staining was observed under the cobalt blue light of the slit lamp microscope after placing the 2% sodium fluorescein staining test paper into the subject’s conjunctival sac of the lower eyelid. The cornea was divided into four quadrants: supratemporal, infratemporal, supranasal, and infranasal. The score was evaluated in every quadrant. The scoring criteria were as follows: 0 for no coloring, 1 for scattered punctate coloring, 2 for diffuse punctate coloring, and 3 for massive coloring. The sum of the scores in each quadrant constituted the corneal FL score with a total of 0–12 points.

The subjects were required to answer 12 questions, and each question was scored 0–4 points [25]. Nonetheless, the subject could refuse to answer some of the questions. The total score was multiplied by 25 and then divided by the number of questions answered to obtain the final score.

The effective rate was if any of the following three conditions were satisfied. ① The value of ST was improved by 2 mm after treatment; ② The value of BUT was improved by 2 s after treatment; ③ The FL was decreased by 1 point after treatment [26].

The statistician generated a random sequence, and the clinician who made the evaluation was blind to the grouping. The data were statistically analyzed using SPSS 24.0. The continuous data obeying the normal distribution were expressed as mean ± standard deviation, and non-normal data were represented as median (interquartile range). The paired sample t-test or Wilcoxon signed-rank test was used for the comparison between the two groups of paired samples, and the independent sample t-test or Mann–Whitney U test was used for the comparison between the two groups of independent samples. Categorical data were described as the number of cases (percentage), and the chi-square test was used for intergroup comparison. P < 0.05 indicated statistically significant difference.