Patients aged 18-65 years fulfilling the 1984 modified AS New York criteria (17) were eligible for inclusion. Patients should present with active disease, defined by a Bath AS Disease Activity Index (BASDAI) (18) of ≥4 on a 0-10 cm visual analog scale (VAS) or an Ankylosing Spondylitis Disease Activity Score using the C-reactive protein level (ASDAS-CRP) (19,20) of ≥1.3. Other defined inclusion criteria were: i) NSAIDs washout period of at least 5 days prior to randomization; ii) disease-modifying antirheumatic drugs washout period of at least 4 weeks prior to randomization; iii) corticosteroids washout period of at least 4 weeks prior to randomization; and iv) biological agents washout period of at least 3 months prior to randomization. Sulfasalazine was permitted if the patient was taking a stable dose for 3 months prior to study entry. Patients with peripheral joint involvement were also included. Adequate contraception throughout the trial was required in women of childbearing age.

Patients were excluded if they had peptic ulcer, unstable cardiac diseases, abnormal hepatic or renal function, hematologic diseases, psychosis, or malignancy. Patients who were allergic to LX were also excluded.

The trial was a randomized open-label study conducted in the Department of Rheumatology of The Third Affiliated Hospital of Sun Yat-Sen University, (Guangzhou, China). The protocol was approved by the Ethics Committee of the Third Affiliated Hospital of Sun Yat-Sen University in accordance with the Declaration of Helsinki principles and the guidelines for Good Clinical Practice (21). The study was registered on 10 January 2019 at www.ClinicalTrials.gov (identifier NCT03800797). Informed consents were obtained from all patients.

Clinical visits were performed at baseline (week 0), week 2, and week 4, or the time of discontinuing treatment by the same investigator. The demographic and disease characteristics were recorded at baseline. Every visit should record clinical and laboratory variables, evaluate efficacy outcomes and monitor adverse events.

Patients were randomized 1:1 to either the LX-P group or the LX-T group for 4 weeks using a computer-generated schedule. Treatment assignment was open to the investigators and patients. Each patient qualifying for treatment was dispensed the corresponding drug. LX-P (Lead Chemical Co., Ltd.) contained 50 mg LX per patch, while LX-T (Daiichi Sankyo Co., Ltd.) contained 60 mg LX. Patients in the LX-P group applied LX-P 100 mg onto the pain spot of vertebra once daily starting at night. Patients in the LX-T group took LX-T 60 mg 3 times daily following meals. The doses for LX-P and LX-T were selected according to a previous clinical trial involving LX-P in knee osteoarthritis (16). Dose adjustments were not allowed during the study period. Usually, an optimal effect of an NSAID is reached no later than after 1-2 weeks (11), so the treatment duration was 4 weeks.

At the screening visit, concomitant therapies with gastrointestinal protective drugs, including misoprostol or proton pump inhibitors, were stopped if there was no history of gastroduodenal ulcers.

The primary efficacy endpoint was the percentage of patients reaching Assessment in Ankylosing Spondylitis 20% (ASAS20) response (22) at weeks 2 and 4, which was defined as an improvement of ≥20% and net improvement of ≥1 unit (0-10-cm VAS) from baseline in ≥3 of the following 4 domains, and absence of worsening in any domain: Patient’s global assessment of disease activity (PTGA), pain assessment, Bath Ankylosing Spondylitis Functional Index (BASFI) (23) and clinical inflammation, determined by 2 morning stiffness-associated scores on the BASDAI. The secondary efficacy endpoint was the percentage of patients achieving ASAS5/6 response at weeks 2 and 4, defined as 20% improvement in 5 of the following 6 domains: PTGA, pain assessment, BASFI, inflammation, spinal mobility and CRP. Other efficacy measures included mean changes from baseline to week 4 for ASDAS-CRP, PTGA, total and nocturnal pain scores, clinical inflammation, BASDAI and BASFI.

Safety evaluations, including adverse events, physical examination and clinical laboratory tests, were monitored by the investigators throughout the study.

Statistical analysis was performed using SPSS 20.0 software (SPSS, Chicago, IL, USA). Baseline demographic and disease characteristic variables were summarized using descriptive statistics. All efficacy endpoint analyses were conducted in the intention-to-treat (ITT) population, which consisted of all randomized patients who took at least 1 dose of treatment. We used the last observation carried forward technique for patients who did not adhere to the study protocol.

Using the ITT population, the ASAS20 and ASAS5/6 response rates were analyzed by Pearson χ2 test. Patients were also stratified based on the presence or absence of peripheral arthritis to determine the effect of peripheral joint involvement on treatment response. Changes of efficacy endpoints from baseline to week 4 were analyzed using the Student’s t-test or Wilcoxon-Mann-Whitney test depending on the Shapiro-Wilk normality test in the ITT population. The safety assessment were also conducted on all randomized patients. The incidence of adverse events was compared between the LX-P and LX-T groups using Fisher’s exact test. For all analyses, statistical analysis was performed at α=0.05. P<0.05 was considered to indicate a statistically significant difference.

From May 2015 to December 2015, 82 patients were screened, and 70 eligible patients were randomized into the 4-week trial, with 35 in the LX-P group and 35 in the LX-T group (Fig. 1). In total, 6 of 70 patients withdrew from the study following randomization, 3 from the LX-P group and 3 from the LX-T group, resulting in an overall dropout rate of 8.6%. The reasons for dropping out included adverse events, insufficient efficacy and loss to follow up.

At baseline, 48 (68.6%) patients were male, and 22 (31.4%) had peripheral arthritis. The baseline demographic and disease characteristics were similar between LX-P and LX-T groups (Table I).

The primary efficacy endpoints of ASAS20 response at weeks 2 and 4 in the ITT population are shown in Fig. 2A. No significant differences between the LX-P and LX-T groups in the proportion of patients achieving ASAS20 response at week 2 [13/35 (37.1%) for LX-P group vs. 17/35 (48.6%) for LX-T group; P=0.334] and week 4 [19/35 (54.3%) for LX-P group vs. 26/35 (74.3%) for LX-T group; P=0.081] were observed. In addition, no statistically significant differences for ASAS5/6 response at weeks 2 and 4 between the two groups were observed [week 2, 7/35 (20.0%) vs. 9/35 (25.7%); P=0.569; week 4, 13/35 (37.1%) vs. 16/35 (45.7%); P=0.467; Fig. 2B]. Changes in efficacy endpoints, including ASDAS-CRP, PTGA, total and nocturnal pain score, clinical inflammation, BASDAI and BASFI, from baseline to week 4 were compared between the LX-P and LX-T groups. There were also no significant differences between the two groups observed (all P>0.05; Table II).

In a subgroup analysis of patients without peripheral arthritis, an increase in the proportion of patients achieving ASAS20 response was observed in the LX-P group (Fig. 3). This suggested that patients without peripheral arthritis in the LX-P group were more likely to achieve ASAS20 response compared with those with peripheral joints involvement.

The incidence of any adverse events described during the study was similar (P=0.710) between the LX-P and LX-T groups: 3/35 (8.6%) patients in the LX-P group and 5/35 (14.3%) in the LX-T group (Table III). All adverse events were mild to moderate in intensity; no serious adverse events or mortalities were observed in either group. The most frequent adverse events for patients receiving LX-P were skin irritation or rashes, while the most common adverse event for patients receiving LX-T was gastrointestinal disorders.