2.1 Participants and recruitment
This study was conducted in the Department of Oral Mucosa, Peking University Stomatological Hospital (tertiary health care) from June 2019 to January 2022, including patients who visited the hospital with complaints of RAU.
Inclusion criteria
(1) The duration of RAU was longer than 1 year; (2) In the past six months, ulcers uninterrupted or occured twice a month or more; (3) classified as minor RAU[9]; (4) Age range of 18-80 years old; (5) Did not participate in any drug trial within 3 months before inclusion.
Exclusion Criteria
(1) Diagnosed as Behcet's disease; (2) Systemic disease background: anemia, immunodeficiency disease, autoimmune diseases, malignant tumors, severe cardiovascular and cerebrovascular diseases; (3) Used analgesics within 24 hours, antibiotics or anti-inflammatory drugs within 1 month, traditional Chinese medicine, glucocorticoids or immunosuppressive agents within 3 months; (4) Long-term diarrhea due to various reasons; (5) Mental disorders and poor compliance, hard to cooperate; (6) Female patients in pregnancy, lactation or planning pregnancy during the study.
2.2 Interventions
According to previous reports, the onset time of TGP capsules was expected to be 12 – 24 weeks[14], so the drug was used for 24 weeks in this clinical trial, and the observation of efficacy and safety lasted for 36 weeks.
The intervention group took TGP capsules (trade name: Pavlin, produced by Ningbo Lihua Pharmaceutical Co., Ltd., H20055058, 0.3 g/capsule). Since there was no classical, widely-accepted systemic treatment regimen for RAU, we used placebo capsules (produced by Ningbo Lihua Co., Ltd.) as control. All subjects received TGP or placebo 0.6g three times a day. If the patient developed frequent or loose stool or abdominal pain, the dose could be reduced to 0.6g two times a day tentatively. On ethical considerations, at the occurrence of ulcers, all subjects were required to use mouthwash and topical medication, that is, Kangfuxin Liquid (Kunming Sino Pharmaceutical Co., Ltd., GYZ53020054, 50 ml-2 bottles/box) and Tong Ren Tang Oral Ulcer Powder (Tong Ren Tang Pharmaceutical Factory, Beijing Tong Ren Tang Co., Ltd., GYZ11020184, 3 g/bottle) 3 times per day.
2.3 Outcome
We determined the outcome indicators referred to the assessment of disease severity in recurrent aphthous stomatitis stated by Tappuni[15] and RAU efficacy evaluation criteria by Chinese Stomatology Association[16].
We divided this trial into five periods: baseline (-4-0 weeks), 0-4 weeks, 5-12 weeks, 13-24 weeks and 25-36 weeks. The RAU severity one month before entering the trial recalled by patients at initial diagnosis was used to calculate for each outcome at the baseline, and daily ulcer records were used for the other periods outcomes.
2.3.1 Primary outcomes
Monthly ulcer-free interval (Interval, days): a total of ulcer-free days in a month.
n= the duration of the observation period, days. I= the total of ulcer-free days in the observation period, days.
2.3.2 Secondary Outcomes
Monthly number of ulcers (Number, n): the sum of the number of oral ulcers per day in a month.
n= the duration of the observation period, days. Ni= number of ulcers on day i in the observation period, n.
Monthly Area of ulcers (Area, cm2): the sum of all the oral ulcers’ area per day in a month.
n= the duration of the observation period, days. Ai= all the oral ulcers’ area on day i in the observation period, cm2.
Since the ulcers’ description need to be recorded by the patients, we divided the size of the ulcers into 6 grades in order to improve the operability, and then we saw ulcers as circular, used the median of each grade as the diameter, and calculated the corresponding area. (See table 1 for details)
Table 1. Ulcer area grading.
Level
|
Description
|
Diameter (cm)
|
Area (cm2)
|
1
|
The point of a needle
|
≈0.1
|
0.01
|
2
|
Millet
|
0.1<d≤0.2
|
0.02
|
3
|
Rice
|
0.2<d≤0.3
|
0.05
|
4
|
Mung bean
|
0.3<d≤0.5
|
0.13
|
5
|
Soybean
|
0.5<d<1.0
|
0.44
|
6
|
Broad bean
|
≈1
|
0.79
|
Pain: assessed by visual analogue scoring (VAS). A line segment was divided into 10 parts, in which 0-10 represents the pain intensity of the patients, 0 means no pain, and 10 means the highest value of pain. It was scored by the patients according to their own symptoms. This result was accessed by average VAS in days with ulcers in a month.
n= the duration of the observation period, days. VASi= VAS on day i in the observation period. nulcers= days with ulcers in the observation period.
2.4 Follow-up
2.4.1 Follow-up Procedures
Patients performed blood routine and blood biochemistry examinations at initial diagnosis and the end of medication.
Patients were requried to take the drug and record the medication, adverse reactions and the severity of ulcers by diary or app (see figure 1 for details) every day. The description of stool referred to Bristol Stool Form Scale[17]. The previous clinical observation found that it was difficult for patients to strictly distinguish stool characteristics. For improving the operability, we simplified it into 5 categories: nuts-shaped, sausage-shaped but lumpy, sausage-shaped, pasty, and watery.
The researchers made a follow-up visit with the subjects every 2 weeks (in the form of telephone or video call) to ensure that the patient made app records or dairy. The patients completed 4 return visits at 4 weeks ± 5 days, 12 weeks ± 7 days, 24 weeks ± 10 days and 36 weeks ± 14 days. At each visit, clinicians measured the ulcer size using a caliper and compared it with the patient's description, and if the description was inconsistent, the patient was corrected.
2.4.2 Follow-up quality control
During treatment, if the subjects (1) used less than 50% or more than 100% of the drug; (2) stopped using or use less drugs more than 60 times; (3) took other traditional Chinese medicine, glucocorticoids or immunosuppressive agents during the trial, the patients were required to withdraw from the trial.
2.5 Sample size calculation
According to the previous study[18], after 6 months of TGP treatment for RAU, the monthly ulcer-free interval of patients was 12.80 ± 6.97 days in the TGP gruop and 8.08 ± 6.18 days in the control group. Considering α=0.05, β=0.20 and 20% dropout Rate, 37.88 patients should be included in each group.
2.6 Random allocation sequence generation and concealment
The subjects in group A and group B were assigned in a 1:1 ratio. Block randomization was used in this study and the block length was 4. The random sequence was generated by a professional statistician using SAS 9.4 software. An independent pharmacist placed the TGP capsules in boxes belonged to group A or B, and placebo to another. The allocation sequence and blinding code were placed in opaque envelopes, stored in a safe.
2.7 Data entry and quality control
We employed the intention-to-treat (ITT) analysis. The data for all subjects, including those who stopped the trial or were not able to be followed up, were considered in the analysis. And we did not make any additions to the missing data.
An electronic data management model was used in this trial. The records filled in by patients in the app were directly uploaded to statisticians. In addition, for the data in the diary, clinicians and a third-party personnel back-to-back entered them into the Epidata 3.1 database, and the inconsistents were checked and corrected with the original records under the supervision.
2.8 Blind Method
The placebo and TGP capsules used the same production pipeline and were identical in terms of color, volume, weight, odor, taste and packaging. Throughout the trial, the subjects, researchers and data-entry personnel only knew the serial number. After completion of the clinical trial and data collection, the statistician opened the envelope to obtain the serial number belonging to group A or B for statistical analysis. After the results were obtained, the unblinding was completed to get which one was the drug group.
2.9 Statistical Methods
Comparisons were made between two groups, with baseline data, and before and after stopping medication (i.e., 25-36 weeks versus 13-24 weeks). If the numerical variables conformed to normal distribution, we used mean ± standard deviation to describe, and T test to compare the differences between two groups. Otherwise, we used median (interquartile range) to describe, and Wilcoxon non-parametric test to compare the differences between two groups. For categorical variables, differences between two groups were compared using the chi-square test or Fisher's exact chi-square test. For intra group comparison, Bonferroni correction method was used to correct the P value, i.e. P adjust. STATA 13.0 software was used for data analysis. P < 0.05 was considered statistically significant.