Demographic characteristics of included studies
The literature selection process is shown in Fig. 1. By searching the databases, 418 studies were found and 1 additional record was identified by reviewing the reference list of related studies. By removing duplicate articles, 225 remained. After screening titles and abstracts, 205 records were excluded. Four full-text were excluded as one was case report and three articles were not English. After full-text screening, 16 studies were considered for qualitative assessment and 13 studies were included for quantitative synthesis [15-30].
All patients were older than 18 years. In 13 studies, 5 RCTs compared the efficacy of PDT with topical corticosteroids. One article clearly stated that patients with the reticular type were included and two articles included the erosive type. The remaining information is presented in Table 1 and Appendix table 2–3.
Quality assessment of included studies
The results of the Cochrane Collaboration's risk of bias assessment and Downs-Black Checklist are shown in Appendix Fig. 1 and Appendix Table 4. The included RCT studies had a low or unclear risk of bias and, owing to the wide difference in treatment method between PDT and topical corticosteroids, most did not specify blinding (Appendix Fig. 1). The majority of non-RCT studies were of high quality in five fields: study quality, external validity, study bias, confounding effects, and power of the study (Appendix Table 4).
Data synthesis
Lesion response
Ten publications involving 309 lesions assessed the lesion response (CR and PR) after PDT, the details of which are shown in Table 1.
As the recognition criteria for CR was strict, half of the studies had no CR and the pooled proportion of CR by the random effects model (Appendix Fig. 2 shows the heterogeneity) was 0.08, which indicated that only 8% of the lesions reached a CR. No publication bias existed (Appendix Fig. 3) and the sensitivity analysis showed that the results were relatively robust (Appendix Fig. 4).
The PRs are shown in Fig. 2. As heterogeneity was detected by the Q test (p value < 0.05) and I2 statistic (71%), the random effects model was applied to pool the overall proportion of PR, which was 0.77 (95% CI: 0.65–0.85). The funnel plot indicated that no publication bias existed (Appendix Fig. 5) and the robustness of the results was also validated by sensitivity analysis (Appendix Fig. 6).
To examine the influence of factors on the final therapeutic response of PDT, subgroup analyses were performed.
Light sources: five types of light source were utilised in 10 studies, including diode laser (three trials), xenon lamp (one trial), semiconductor laser (one trial), metal halide lamp (one trial), and light-emitting diode (LED) (four trials). As the standard of CR was so strict that half of the included studies achieved no CR, we only applied subgroup analysis for the PR. The forest plots of the different light sources are shown in Appendix Fig. 7 and Fig. 3A shows the results of the random effects model for the pooled PR. No significant difference (tested by u test, p > 0.05) was detected in the PR among light sources.
PS: three types of PS were discussed in the studies, 5-ALA (four trials), MB (five trials), and chlorin e6 derivative (one trial). The pooled PR of 5-ALA was 0.86 (95% CI: 0.80–0.91), which was more effective than other two and the difference was significant (p < 0.05) compared to MB (Fig. 3B and Appendix Fig. 8).
Administration methods: five trials used topical application and five used gargle administration. Topical application was more effective than the gargle method (p < 0.05) and the pooled PR of topical application reached 0.85 (95% CI: 0.80–0.89) (Fig. 3C and Appendix Fig. 9).
Additionally, two types of lesion locations buccal mucosa and/or lips (BM/L) and tongue and/or gingival mucosa (T/G), were detailed in three studies, including 194 lesions. The OR was calculated and pooled to compare the PR of the two lesion locations and the PDT would be regarded as more efficient in BM/L if the OR was greater than 1. The PDT seemed to be more suitable for BM/L, although this was not statistically significant (pooled OR: 1.75, 95% CI: 0.43–7.05) (Appendix Fig. 10).
Changes of lesions
The variables of lesion size and TH were included to assess the changes of the lesions after PDT.
The lesion size was recorded in six publications before and after PDT and 245 lesions were identified for meta-analysis. The lesion size decreased by 1.53 cm2 (95%: 0.71–2.35) after PDT (Fig. 4A). Heterogeneity existed among the six studies, as the I2 statistic was 85% and the p value of the Q test was lower than 0.01. The sensitivity analysis (Appendix Fig. 11) validated that the pooled estimates were stable. Publication bias could be ignored according to the funnel plot (Appendix Fig. 12).
Through subgroup analysis, lesion size decreased more using a semiconductor laser than using a diode laser (tested by u test, p < 0.05, Fig. 3D) and the lesions located on the BM/L decreased more than those located on the T/F/G (pooled MD: 0.37, 95% CI: 0.05–0.68, Appendix Fig. 13), whereas no statistically significant differences were detected between PS and administration method (Fig. 3E and Fig. 3F).
To determine the TH score, five trials, including 88 lesions, were involved. Owing to the heterogeneity (Fig. 4B), the random effects model was recommended and the TH score decreased by 1.33 (95% CI: 0.56–2.10) after PDT, which was validated for robustness with the sensitivity analysis (Appendix Fig. 14) and publication bias could be ignored (Appendix Fig. 15).
The subgroup analysis indicated that the metal halide lamp performed better than the LED and xenon lamp (Fig. 3G), whereas the PS and administration method showed no significant differences in performance (Fig. 3H and Fig. 3I).
Clinical pain symptom
The clinical pain symptom was measured by VAS and six studies with 88 lesions used VAS to assess the improvement of pain after PDT. The VAS score decreased by 3.82 (95% CI: 2.80–4.85) (Fig. 4C). Heterogeneity existed (p value < 0.05 and I2 = 92%) and the result was robust (Appendix Fig. 16). Publication bias could be ignored according to the funnel plot (Appendix Fig. 17).
Unlike with lesion changes, the subgroup analysis revealed that the efficacy of the diode laser was better than that of the metal halide lamp and LED in relieving pain (u test p < 0.05, Fig. 3J). No differences were observed in the subgroup analysis of PS and administration method (Fig. 3K and Fig. 3L). Owing to the information shortage, the subgroup analysis of sites of OLP lesions was not performed.
Table 1 Characteristics of the studies with lesion response and lesion size changes after PDT.
Author
|
Year
|
Light sources
|
Photosensitizer
|
Lesion types
|
Administration Method
|
Lesion locations
|
Sample size
|
CR
|
PR
|
lesion size of pre-PDT
|
lesion size of post-PDT
|
Mean
|
SD
|
Mean
|
SD
|
Aghahosseini F
|
2006
|
diode laser
|
MB
|
mixed
|
gargle
|
mixed
|
26
|
4
|
12
|
1.8
|
0.7
|
1
|
0.9
|
Sadaksharam J
|
2012
|
xenon lamp
|
MB
|
mixed
|
gargle
|
mixed
|
20
|
0
|
10
|
|
|
|
|
Sobaniec S
|
2013
|
semiconductor laser
|
chlorin e6 derivative
|
mixed
|
topical
|
mixed
|
48
|
14
|
25
|
6
|
4.5
|
2.7
|
2.62
|
|
|
|
|
|
|
BM/L
|
40
|
13
|
22
|
6.6
|
4.63
|
2.8
|
2.81
|
|
|
|
|
|
|
T/G
|
8
|
1
|
3
|
3
|
1.9
|
2.1
|
1.21
|
Prasanna SW
|
2015
|
metal halide lamp
|
MB
|
mixed
|
gargle
|
mixed
|
15
|
0
|
13
|
3
|
1.6
|
0.8
|
1
|
Maloth KN
|
2016
|
LED
|
5-ALA
|
mixed
|
topical
|
mixed
|
10
|
0
|
8
|
2.22
|
0.79
|
1.41
|
0.74
|
Bakhtiari S
|
2017
|
LED
|
MB
|
mixed
|
gargle
|
mixed
|
15
|
0
|
2
|
|
|
|
|
Mostafa D
|
2017
|
diode laser
|
MB
|
mixed
|
gargle
|
mixed
|
19
|
7
|
9
|
|
|
|
|
Sulewska M
|
2017
|
LED
|
5-ALA
|
erosive
|
topical
|
mixed
|
22
|
5
|
11
|
1.49
|
1.45
|
1.37
|
1.78
|
|
BM
|
16
|
4
|
7
|
1.06
|
0.98
|
1.08
|
1.57
|
|
T /G
|
6
|
1
|
4
|
2.63
|
1.93
|
2.13
|
2.24
|
Rakesh N
|
2018
|
diode laser
|
5-ALA
|
erosive
|
topical
|
mixed
|
10
|
0
|
10
|
|
|
|
|
Sulewska M
|
2019
|
LED
|
5-ALA
|
reticular
|
topical
|
mixed
|
124
|
46
|
63
|
3.99
|
3.73
|
1.48
|
1.98
|
|
BM
|
80
|
27
|
44
|
4.58
|
4.01
|
1.67
|
2.04
|
|
T /G
|
44
|
19
|
19
|
2.93
|
2.91
|
1.13
|
1.84
|
PDT: photodynamic therapy; GaAlAs: Gallium-Aluminum-Arsenide; LED: light emitting diode; TB: toluidine blue; 5-ALA: 5 aminolevulinic acid; MB: Methylene Blue; BM/L: buccal mucosa and/or lips; T/G: tongue and/or gingival mucosa; mixed: with different required information or information were not mentioned; CR: complete response; PR: partial response; NR: no response. SD: standard deviation.
Comparison with topical corticosteroids
To obtain an understanding of the efficacy of PDT, five RCT trials with 139 lesions were included to compare PDT with topical corticosteroids. The indicators included PR (recorded in two trials, 68 lesions in total), TH (recorded in four trials, 109 lesions in total), and VAS (recorded in four trials, 109 lesions in total), the details of which are shown in Appendix Table 3. The pooled estimates indicated varied results. The efficacy of PDT was better than that of topical corticosteroids (pooled OR: 6.15, 95% CI: 1.65–22.97) based on the on PR (Appendix Fig. 18). With the TH score (Appendix Fig. 19), the pooled mean difference was 0.62 (95% CI: -0.46–1.71), which indicated that the two treatments had similar efficacy to decrease lesion size. Additionally, the pooled mean difference of VAS (Appendix Fig. 20) was -0.30 (95% CI: -1.99–1.40), which indicated the similar improvement of pain between the two treatments. As the number of trials is insufficient, the funnel plot and sensitive analysis were not performed.
Other factors in PDT
The wavelength of 630–660 nm and energy density of 80–150 J/cm2 were commonly used. The duration of irradiation ranged between 120 s and 600s. The range of dressing time was 5–120 min. The frequency of PDT application ranged from 1–10 times throughout the study period at 1- to 2-week intervals. The details are shown in Table 2.
The majority of patients experienced no discomfort or only minor adverse effects (pain, mild burning sensation) during treatment, which disappeared immediately. Most studies were conducted with a usual follow-up time of 1–12 months. Among all studies, six patients in two studies were reported to have relapsed after PDT. However, most studies did not report a cancerous patient.