Treatment of oral fungal infections using photodynamic therapy: Systematic review and meta‐analysis

Abstract Objectives This systematic review evaluated the evidence for the effectiveness of Photodynamic therapy (PDT) in treating oral fungal infections, as an alternative to conventional antifungal medications. Methods Five randomized control trials (168 participants) comparing the treatment of oral fungal infections using met with our inclusion criteria. Clinical and microbiological improvement was assessed by random‐effects meta‐analysis. Methodological quality assessment and heterogeneity were performed using peer‐reviewed criteria. PROSPERO registration: CRD42017076. Results PDT showed statistically non‐significant increased clinical efficacy (risk ratio (RR) = 1.47 [95% confidence interval (CI), 0.68; 3.17]; three studies, n = 108 participants, I 2 = 50%) and mycological efficacy (mean difference (MD) = 0.54 [95%CI, −0.71; 1.79]; three studies, n = 100; I 2 = 39%) at 30 days, as compared with conventional antifungal therapy. Lack of standardization of treatment parameters and variability in the assessment of outcomes was observed across the studies. All included studies had a moderate to low risk of bias. Conclusions PDT showed comparable effectiveness at treating oral fungal infections, particularly denture stomatitis. The small number of studies in this review, small sample size and variability of methods and outcome measures across studies, highlight the need for more standardized studies with longer follow‐up periods to enable recommendation of PDT as an alternative to conventional antifungal therapy.

. In a recent study, 50% of Candida albicans specimens sampled were resistant to azoles (dos Santos Abrantes et al., 2014) and new drugs to treat fungal infections have not been developed since 2006 (Denning & Bromley, 2015). Alternative therapies are thus required to treat these minimally invasive fungal infections without propagating the rise in fungal antimicrobial resistance (Liang et al., 2016). Recently, the use of photodynamic therapy (PDT) has garnered attention as a potential antifungal treatment modality.
PDT, also referred to as photodynamic antimicrobial chemotherapy (PACT), photoradiation therapy and photochemotherapy, comprises three components: a chemical photosensitizer (PS), the application of light and the presence of oxygen. Briefly, the PS is applied to the target tissue (either topically or systemically). Light of an appropriate wavelength is then used to activate the PS, generating highly reactive oxygen species (ROS), including the singlet oxygen, in the target tissue. This results in cytotoxicity of the target cells and elicits an acute inflammatory response in the surrounding tissues (Konopka & Goslinski, 2007;Saini & Poh, 2013). Thus, PDT is being studied as a treatment modality for a variety of clinical applications, including the treatment of oral fungal infections; however, some recent studies have found PDT to be inferior when compared with antifungal medication in the treatment of specific oral fungal infections (Leite et al., 2015;Maciel et al., 2015). Given this equipoise, our systematic review sought to review current evidence on the use of PDT as a treatment modality for oral fungal infections in humans. In addition, we sought to determine the most effective treatment regimen parameters, light delivery parameters and which type and concentration of photosensitizers are most effective for the treatment of oral fungal infections. Lastly, we wished to determine how the risk factors for oral fungal infections such as smoking and diabetes mellitus, affect treatment outcomes.

| MATERIALS AND METHODS
The protocol of this review was registered with PROSPERO, registration number CRD42017076421 and strictly complied with the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines (Moher et al., 2009). This review received an ethics waiver from the University of Cape Town, Faculty of Health Science Human Research and Ethics Committee as this review relied on only publicly available information (HREF 636/2018). No informed consent was required for this review.

| Research question
This systematic review and meta-analysis were conducted to address the following focused question: "Is photodynamic therapy compared with standard anti-fungal treatment modalities, effective for the treatment of oral fungal infections in humans?" 2.2 | Eligibility criteria

| Search strategy
A comprehensive database search was initially conducted in September 2018 using the following databases: The Cochrane Library, BioMed, SciELO, Scopus, EBSCOhost, PubMed/MEDLINE, ISI Web of Science, Clinicaltrials.gov, ProQuest, and WorldCat. The search strategy has been detailed elsewhere (Table S1; Roomaney et al., 2020).
The results of the search were documented, reported and compared between databases (Table S2; Roomaney et al., 2020). The references were managed with EndNote (EndNote X9, version 9.2, Clarivate Analytics, USA) reference manager. An update on the search was conducted in PubMed on June 30, 2020.

| Quality assessment
Each reviewer conducted an assessment of study quality and the risk of bias of each included study using the risk of bias tools of the Cochrane Collaboration (Higgins & Green, 2011).

| Study selection and data extraction
The search results were collated within an online document and two researchers independently performed title and abstract screening, followed by full-text evaluation and data extraction onto a pre-design form. There was no disagreement between the reviewers on the studies to include.

| Statistical analysis
Quantitative data was assessed using Review Manager (RevMan version 5.3) statistical software and the data were pooled, where appropriate, to conduct a meta-analysis. Pooling of the data was done to assess three outcomes: (1) clinical improvement from baseline, Forest plots were created for the time points of 7, 15 and 30 days respectively. In conducting the meta-analysis, we used the number of participants randomized to each arm, irrespective of withdrawal due to incompletion of treatment or loss to follow-up, that is, intentionto-treat analyses.
The effect size was estimated and reported from continuous variables using mean difference and 95% confidence intervals. The weighting of each study was calculated using the inverse of the variance. A random-effects model was used for analysis (Borenstein et al., 2010). Where the researchers found insufficient data, they conducted a narrative report of the results.
The authors used the Cochrane test (P < 0.1 cut-off for statistical significance) to determine statistical homogeneity and the I 2 test was used to quantify heterogeneity. The I 2 test are interpreted as follows: 0%-40% may not be important; 30%-60% considered moderate heterogeneity; 50%-90% considered substantial heterogeneity; and 75%-100% is considerable heterogeneity (Higgins et al., 2019). Subgroup analyses were planned to assess the following: the effect of different treatment parameters, including a comparison of different light delivery devices, wavelengths, photosensitizers and different treatment regimens i.e. duration of application, frequency of applications and time between applications; the various antifungal medications used; the effect of PDT on different fungal strains; and the effect of comorbidities/predisposing medical conditions such as HIV, diabetes mellitus, and dental prosthesis use. Publication bias was assessed via funnel plots, however, the low number of studies rendered them uninformative.

| Literature search and study selection
The initial search yielded 654 titles ( Figure 1). Two additional articles were found by hand-searching reference lists of relevant articles.
Titles were collated and duplicates were excluded. The remaining 353 titles were evaluated, and 273 studies were excluded based on titles. Subsequent abstract screening resulted in a further 68 being excluded. We were unable to find the full-text for one article (Cadastro & Giovani, 2009) Ribeiro et al., 2012;Simunovic-Soskic et al., 2010). A further seven articles were excluded for not fulfilling the inclusion criteria. Details pertaining to the exclusion criteria are provided in the Supporting Information (Table S3; Roomaney et al., 2020). An updated search was conducted in June 2020 leading to the inclusion of an additional study (Alrabiah et al., 2019). Five full-text studies were included in the review.

| Characteristics of the included studies
The general characteristics of the included studies are presented in Table 1. Four of the randomized control trials included were conducted in Brazil (de Senna et al., 2018;Lopes, 2011;Mima et al., 2012;Scwingel et al., 2012) and one was conducted in Saudi Arabia (Alrabiah et al., 2019;N = 168). These studies comprise between 14 and 54 enrolled participants each.
The description of parameters investigated, and technical characteristics of the PDT used in the included studies are documented in Table 2. Three studies conducted PDT on both dentures and oral mucosa (Alrabiah et al., 2019;de Senna et al., 2018;Mima et al., 2012). One study evaluated the treatment of oral candidiasis in HIV positive patients (Scwingel et al., 2012), while the other four studies specifically evaluated the treatment of denture stomatitis (Alrabiah et al., 2019;de Senna et al., 2018;Lopes, 2011;Mima et al., 2012).
Four studies used lasers as the light source (660 nm wavelength). Twin lasers were used in two studies (Lopes, 2011;Senna, 2012) and a GaAIA (Gallium aluminum arsenide) laser was used in two studies (Alrabiah et al., 2019;Scwingel et al., 2012). The studies investigating lasers utilized methylene blue as the PS. A single study used a hematoporphyrin derivative as a PS, which was activated by an LED light of 440-460 nm wavelength . The power of the LED used was 260 mW, which is significantly higher than that provided by the lasers (100, 40, and 30 mW respectively). Pre-irradiation time, which is the length of time between application of the PS and photoactivation, ranged from 1 to 20 min. The length application of the laser per point was between 10 s and 2 min. The length of application of the LED was 20 min. Treatment sessions varied from one session (Scwingel et al., 2012) to two sessions 1 week apart (Lopes, 2011); to six sessions over 15 days . The largest number of sessions were eight PDT sessions over 4 weeks (Alrabiah et al., 2019;de Senna et al., 2018).
Three studies used nystatin suspension as the comparator.
One advised rinsing with 5 mL of 100,000 IU suspension six times a day for 2 weeks (Lopes, 2011) and the other two studies advised rinsing with the same dosage, four times daily for 2 weeks (Alrabiah et al., 2019;Mima et al., 2012). The study with HIVpositive participants used 100 mg of fluconazole daily for 15 days (Scwingel et al., 2012). In the fifth study, miconazole gel was applied to the affected area three times daily for 4 weeks (de Senna et al., 2018).

| Method of clinical and microbiological assessment
Four studies assessed clinical change using three methods. Two studies used the Budtz-Jørgensen et al. (1988) method before treatment and 48 h after treatment (de Senna et al., 2018;Lopes, 2011). One study ) used Newton's classification (Newton, 1962) and the other study used specified subjective comparisons from clinical baseline to assess clinical changes (Scwingel et al., 2012). This was done at baseline, end of treatment (day 15) and on follow-up (days 30, 60 and 90). Quantification of colony-forming units (CFUs) was used to assess the microbiological success of treatment in three studies (Alrabiah et al., 2019;Lopes, 2011;Mima et al., 2012). The remaining studies made use of semi-quantification of CFU/mL (de Senna et al., 2018;Scwingel et al., 2012). This was either done via visual assessment of the medium turbidity (clear, mild or intense) of cell cultures in test tubes and then scored as low, medium or abundant growth of fungus accordingly (Scwingel et al., 2012). Alternatively, the CFUs were counted and expressed as degrees of density (de Senna et al., 2018).

| Study outcomes
There was no statistically significant difference in clinical effectiveness, that is, a reduction of oral lesions, between PDT and conven- (p = 0.20)) at 30 days, which implies there was merit in pooling the data.

| Risk of bias of included studies
Contact was made with authors to clarify the risk of bias of included studies. All included studies were found to have a moderate to low risk of bias (Table 3)  Authors were contacted to retrieve missing information. Despite finding only five studies meeting with our inclusion criteria, we were able to conduct a meta-analysis to present aggregate data of PDT against conventional therapy.
Although the study designs of the included studies were similar, the studies demonstrated significant variability in their methods. The biggest challenges were the lack of standardization of treatment parameters across studies and inconsistency in the assessment of outcomes. Thus, in conducting the meta-analysis, PDT was used as an umbrella term for any intervention (regardless of parameters) meeting our inclusion criterion. Three studies used the quantification of colony-forming units (CFUs; Alrabiah et al., 2019;Lopes, 2011;Mima et al., 2012) as the outcome measure, as opposed to the semiquantification of CFU's used by the remaining two studies (de Senna et al., 2018;Scwingel et al., 2012). We conducted the planned subgroup analyses where data was available; however, the results were uninformative due to the small sample sizes after pooling of data.
Although every effort has been made to reduce bias within our  (Dovigo et al., 2010). Alrabiah et al. (2019) and Mima et al. (2012) compared fungal species before and after treatment and had similar findings. C. albicans was similarly sensitive to PDT and nystatin (75% and 90% reduction), whereas, C. tropicalis appeared to be significantly more sensitive to nystatin than PDT at 15 days (45% and 50% reduction). A laboratory study by Dovigo et al. (2010) comparing the sensitivity of four fungal species to various PDT parameters found that C. tropicalis required PDT at a greater energy density for inactivation than that required by C. albicans. More research is The few studies in our review precluded an assessment of the effect of local risk factors (such as smoking, nocturnal denture wearing, denture hygiene) and systemic risk factors such as HIV, Diabetes mellitus and immunosuppressive therapies, on oral fungal infection treatment outcomes. Smoking is a risk factor for oral fungal infections and treatment outcomes tend to be inferior in smokers compared to non-smokers (Abduljabbar et al., 2017). The study by de Senna et al. (2018) which included four smokers found that miconazole was more effective at reducing fungal load than PDT. There was, however, only one smoker in the miconazole control group compared to three in the PDT group. Although the statistical significance of this finding was not mentioned, it is important to note that all smokers in the study presented with higher fungal loads at follow-up. One study compared fluconazole treatment and PDT in HIV-positive patients (Scwingel et al., 2012)  Recurrence of fungal infection has been mentioned as a particular concern when using PDT (Lopes, 2011;Mima et al., 2012). This corresponds to a recent case series treating five patients with denture stomatitis (Alves et al., 2018) which found recurrence in all patients at the end of day 45 of follow-up. More studies beyond 30 days will be required to assess if recurrence is a problem with PDT therapy in general or to determine if it is related to the specific treatment parameters used. If recurrence is found to be a problem with the use of PDT, it would be important to assess whether new fungal species have emerged, PDT-resistant species have developed or whether an insufficient reduction of patient risk factors is a possible contributory factor to the recurrence.
The importance of finding alternatives to conventional antimicrobial medication cannot be stressed enough. PDT appears to have potential as a therapy for oral fungal infections. However, the lack of recent human studies begs to question as to why progress into this area has stalled. At present, it is still a relatively costly procedure requiring specialized equipment, not commonly available in general dental offices. However, there is an effort to create a more cost-effective LED light source (Daly et al., 2017;Hempstead et al., 2015) which would make PDT more accessible and provide greater scope to evaluate its impact than is cur-

CONFLICT OF INTEREST
The authors have stated explicitly that there are no conflicts of interest in connection with this article.

AUTHOR CONTRIBUTIONS
IR, HH and ME were jointly responsible for the conceptualization of the study. IR and HH performed the data extraction. IR and ME conducted the meta-analysis. All authors performed editing and approved of the final manuscript.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available in