A systematic literature retrieval was performed in different databases including PubMed, EMBASE, and MEDLINE using the search terms (((((triamcinolone) OR corticosteroids) OR steroids)) AND ((((((randomised controlled trials) OR controlled clinical trials) OR controlled clinical trial) OR randomised controlled trial) OR clinical trial) OR clinical trials)) AND ((((hypertrophic) AND ((((scar) OR scars) OR scarred) OR scarring))) OR keloid). Full articles and abstract are all included. The search was performed in April 2016.

Two reviewers assessed the eligible trials independently. Only randomized controlled trial (RCT) and controlled trial (CT) on patients with pathological confirmed keloid were included. Studies that compared the regime of TAC with a non-steroid-based treatment modality were included.

The primary outcomes included (1) reduction in scar height, thickness, size, vascularity, pliability, and pigmentation and (2) overall scar improvement obtained from patient self-assessment and observer assessment. Patient self-assessment was graded by patients with a 5-point scale: no improvement; poor, up to 25% improvement; fair, 26–50%; good, 51–75% improvement; and excellent, 76–100% improvement. Observer assessment was graded by the observer using a scale that was the same as patient self-assessment. Over 50% improvement was regarded as effective. The secondary outcome was improvement in erythema, pain, and itch. Adverse events included hypopigmentation, telangiectasia, and skin atrophy.

Two reviewers extracted data independently. Study characteristics (author, year of publication, country, number of patients, intervention and control, follow-up period, primary and secondary outcomes) and participant characteristics (age and sex) were summarized. The numbers of patients with >50% improvement, >50% reduction in size, or any specific adverse event were extracted from both arms of intervention. The values of Vancouver Scar Scale including scar pigmentation, vascularity, pliability, and height were extracted from both groups. The values of scar thickness before and after intervention were also extracted.

The risk of bias was evaluated according to Cochrane handbook and analyzed by Review Manager 5.3. The parameters included random sequence generation, allocation concealment, blinding, incomplete outcome data, and selective reporting.

For dichotomous variables, inverse variance risk ratio (RR) and corresponding 95% confidence intervals (CI) were calculated to evaluate the effect of intervention. For continuous data comparisons, weighted mean difference (WMD) and 95% CI were calculated. Heterogeneity between studies was assessed using p value of I² statistics. p value of I² statistics smaller than 0.1 was classified as indication of substantial heterogeneity. Where heterogeneity was observed, random effects model was used. In contrast, fixed effect model was adopted. For TAC treatment, different dosages of TAC were used. A subgroup analysis was performed according to the dosage of TAC. All the analysis was carried out using STATA version 12.0 (StataCorp). Unless otherwise specified, p value below 0.05 was denoted as statistical significance.

A total of 113 studies were identified by searching Pubmed, EMBASE, and Medline. After excluding 101 irrelevant studies, 12 studies were selected for further evaluation and all the studies were available in full paper (Figure 1A). Four studies were further excluded due to duplicate publication or no data on defined primary and secondary outcomes available. The study of Darougheh et al. (15) has duplicate data with Asilian et al. (16); therefore, it was excluded for meta-analysis. No data on reduction in scar size or overall scar improvement could be extracted from the studies of Layton et al. (17), Sproat et al. (18), and Kelemen et al. (19). Finally, eight studies were included in this meta-analysis (11, 16, 20–25). The characteristics of these eight studies were shown in Table 1.

The risk of bias of eight studies included in this meta-analysis was summarized in Figure 1B. Since the studies of Yosipovitch et al. (20) and Tan et al. (11) were not RCT, they were judged as high risk for random sequence generation. Four studies (22–25) provided the method used to generate random sequence and were judged to be at low risk of random sequence generation. Two studies (16, 21) did not report the method of generating random sequence and we judged them at unclear risk. Allocation concealment was not reported in all of the eight studies (11, 16, 20–25), and we judged all trials as unclear for allocation concealment. Two single-blinded trials (16, 24) were judged as high risk for blinding of participants and personnel. Complete information on the blinding processes was not shown by the left six studies (11, 20–23, 25) and we judged them at unclear risk for performance bias. Two studies (16, 25) employed blinded observer to assess outcomes and were judged at low risk of detection bias. The left six trials (11, 20–24) were judged as unclear risk due to no report on blinding of outcome assessment. Four trials (11, 16, 20, 24) were judged at high risk for incomplete outcome data because the numbers lost to follow up were high and no reasons were given for the losses. No reporting bias or other bias was observed in all eight studies (11, 16, 20–25). In summary, most studies have no detail information on allocation concealment, blinding of participants and personnel, and blinding of outcome assessment.

Tan et al. (11) evaluated the efficacy of TAC in keloid treatment. A total of 20 patients with multiple keloids were recruited and treated by TAC for 12 weeks. Seventeen patients in both TAC group and control group completed the 12-week treatment. A significant difference was observed in >50% reduction in size (RR: 33.00, 95% CI: 2.14–509.33) and improvement in erythema (RR: 21.00, 95% CI: 1.33–332.06), favoring TAC in comparison with control. In terms of improvement in pain and itch, there was no statistically significant difference. These results indicated that TAC treatment significantly reduced the size and improved appearance of keloid.

Tan et al. (11) compared the effectiveness of TAC with silicone gel sheet in keloid treatment. Seventeen patients in both TAC group and silicone gel sheet group completed the 12-week treatment. The >50% reduction in size (RR: 8.00, 95% CI: 2.16–29.57) and improvement in erythema (RR: 10.00, 95% CI: 1.43–69.77) were statistically different in favor of TAC compared to silicone gel sheet. For improvement in pain and itch, no apparent difference was observed. TAC treatment not only resulted in a decrease in size but also normalized symptoms of keloid in comparison with silicone gel sheet.

The efficiency of TAC was compared with 5-FU in two trials (21, 22). In the study of Manuskiatti and Fitzpatrick (21), 10 patients including 6 women and 4 men with age ranging from 25 to 74 were treated by TAC or 5-FU. Over 50% improvement was observed in 10 out of 10 patients in both TAC arm and 5-FU arm. The RR was 1 with the standard error of the log relative risk to be 0 and 95% CI was 1 to 1. In data synthesis, this study was excluded by STATA. Sadeghinia and Sadeghinia (22) compared the effects of TAC and 5-FU using patient self-assessment. Forty patients were randomized into TAC group (20 patients) or 5-FU group (20 patients). The ratio of patients with >50% overall improvement was compared. A significant improvement in favor of 5-fluorouracil compared to those treated with TAC (RR: 2.12, 95% CI: 1.20–3.75). The study of Sadeghinia and Sadeghinia (22) showed that no side effect was detected in both TAC and 5-FU group. In contrast, Manuskiatti and Fitzpatrick (21) reported that skin atrophy, telangiectasia, and hypopigmentation were noted in 50% segments of TAC group but not in 5-FU group. However, the difference in complications between TAC and 5-FU groups was not statistically significant (RR: 11.0, 95% CI: 0.69–175.86).

Three trials compared the effects of TAC alone with the combination of TAC and 5-FU (16, 21, 23). There are two outcomes including patient self-assessment and observer assessment. Over 50% improvement was regarded as effective. There was no significant difference between the two groups (RR: 1.18, 95% CI: 0.83–1.69; Figure 2A). For TAC treatment, there are two different regimes: (A) TAC 10 mg/ml, once weekly, eight sessions; and (B) TAC 20 mg/ml, once every 4 weeks, six sessions. For the combination treatment using TAC and 5-FU, TAC dosage reduced to 1–4 mg/ml together with 45 mg/ml 5-FU. Except Manuskiatti and Fitzpatrick (21) (regime B, 20 mg/ml), all other studies used the same treatment regime (regime A, 10 mg/ml). When we performed subgroup analysis based on the treatment regime, a statistically significant difference in favor of TAC with 5-FU was only observed in the studies using 10 mg/ml TAC (RR: 1.27, 95% CI: 1.06–1.52; Figure 2B) but not in the study using 20 mg/ml TAC (RR: 0.90, 95% CI: 0.69–1.18; Figure 2B). In terms of complications, patients in TAC group have a higher risk to experience skin atrophy and telangiectasia compared with combination treatment using TAC and 5-FU (RR: 3.65, 95% CI: 1.65–8.08; Figure 2C). TAC plus 5-FU showed advantages over TAC alone only when a low dose of TAC was used in TAC monotherapy.

Two trials (24, 25) compared the use of TAC with verapamil. Forty patients (48 scars) were divided into TAC group (22 scars) and verapamil group (26 scars) in the study of Ahuja and Chatterjee (25). In the trial of Margaret Shanthi et al. (24), 54 patients were allocated to TAC arm (27 patients) and verapamil arm (27 patients). In both trials, Vancouver Scar Scale including scar pigmentation, vascularity, pliability, and height was used to measure the treatment effects. For scar pigmentation, no statistical difference between TAC and verapamil was observed at 3 weeks (WMD: −0.10, 95% CI: −0.32–0.12; Figure 3A); when vascularity was used as outcome, there was a statistically significant difference in favor of TAC compared to verapamil at 3 weeks (WMD: −0.22, 95% CI: −0.44 to −0.01; Figure 3B); in terms of scar pliability, significant improvement in favor of TAC was observed after 3 weeks (WMD: −0.39, 95% CI: −0.60 to −0.18; Figure 3C); for scar height, no statistical difference between TAC and verapamil was noticed (WMD: −0.05, 95% CI: −0.24–0.14; Figure 3D). Although Margaret Shanthi et al. (24) reported that complications were seen in both groups, no data were presented. In the trial of Ahuja and Chatterjee (25), the difference in complications including skin atrophy and telangiectasia was not statistically significant between TAC and verapamil groups (RR: 15.26, 95% CI: 0.91–256.58). TAC treatment improved the symptoms of keloid but not reduced the height of keloid compared to verapamil.

The effectiveness of TAC was compared with cryotherapy by Yosipovitch et al. (20). Fourteen patients with at least two keloids aged between 17 and 50 years were enrolled in this study. The WMD in scar thickness was statistically different in favor to TAC (WMD: −3.35, 95% CI: −4.95 to −1.75).

Yosipovitch et al. (20) compared the efficacy between TAC and TAC plus Cryotherapy. The WMD in scar thickness was statistically different in favor to TAC plus cryotherapy (WMD: −4.60, 95% CI: −7.54 to −1.66).