Abstract
The aim of this study was to determine whether highly active antiretroviral therapy (HAART) is associated with the prevalence of oral lesions in HIV-positive patients. This systematic review and meta-analysis was performed in accordance with the PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses). The search was conducted in seven electronic databases (PubMed, Scopus, SciELO, LILACS, Embase, Web of Science, and OpenGrey), without restriction on publication period or language. Studies that showed the prevalence of oral lesions manifested in adult HIV-positive patients, subjected or not to HAART, were selected. The meta-analysis estimate of relative risk was calculated using the Mantel–Haenszel method and DerSimonian and Laird estimator to determine the variance between studies in the random-effects model. The meta-analysis showed significant results in favour of the group on HAART, with lower prevalence for angular cheilitis, erythematous candidiasis, oral herpes, pseudomembranous candidiasis, Kaposi sarcoma, and oral hairy leukoplakia. The prevalence of oral mucosal hyperpigmentation was higher in patients on HAART. These results suggest that the prevalence of oral lesions in HIV-positive patients is lower for those on HAART, which might occur because of the improvement in immunity provided by the therapy.
Infection with the human immunodeficiency virus (HIV) results in a reduction in CD4+ T lymphocyte counts, which begins gradually and increases progressively . Therefore, individuals with HIV become susceptible to opportunistic infections and to certain neoplastic processes .
Highly active antiretroviral therapy (HAART) has become a frequent treatment for HIV infection . It was introduced in the mid-1990s , and over time has consisted of more than 30 different drugs from six drug classes used in different combinations, each of them presenting advantages and disadvantages . Despite blocking viral replication and providing conditions for immune function recovery, these drugs also reduce the viral resistance to drugs . Thus, this therapy increases CD4+ T lymphocyte counts and reduces the viral load in individuals with HIV . Patients on HAART show improvements in immune reconstitution, but develop clinical outbreaks of infectious diseases .
Prior to the advent of HAART, oral lesions, oral candidiasis, and oral hairy leukoplakia were frequently observed in individuals infected with HIV . The use of HAART has contributed to a reduction in the clinical prevalence of oral hairy leukoplakia and a moderate reduction in the prevalence of necrotizing ulcerative gingivitis, necrotizing ulcerative periodontitis, necrotizing ulcerative stomatitis, oral candidiasis, and Kaposi sarcoma .
Considering the change in the prevalence of oral lesions identified in HIV-positive patients on HAART, the current systematic review and meta-analysis aimed to determine whether this therapeutic option is a factor associated with the prevalence of oral lesions in HIV-positive patients when compared to patients not on HAART .
Materials and methods
Protocol and registration
This systematic review and meta-analysis was performed in accordance with the PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses; http://www.prisma-statement.org ) . The systematic review protocol was registered in the PROSPERO database (number CRD42015026278; http://www.crd.york.ac.uk/PROSPERO ).
Eligibility criteria
This systematic review and meta-analysis was conducted in order to answer the following focused question: Is highly active antiretroviral therapy a factor associated with the prevalence of oral lesions in HIV-positive patients when compared to a control group? The research question was based on the PICO strategy, in which P (population) corresponds to HIV-positive adults, I (intervention) corresponds to HAART, C (comparison) corresponds to patients not on HAART, and O (outcome) corresponds to oral lesions previously described in the literature .
With regard to inclusion criteria, only cross-sectional studies were included. There was no restriction on publication period, language, or status. Studies had to present the prevalence of oral lesions manifested in adult HIV-positive patients subjected or not to HAART.
Studies that did not meet the objectives, were case reports, letters to the editor and/or editorials, literature reviews, indexes, or abstracts, that addressed the manifestation of oral lesions in HIV-positive children subjected or not to HAART, that used the sample itself as the control group, or that included patients on antifungal therapy or on mono- or dual therapy, were excluded.
Information sources
The PubMed, Scopus, SciELO, LILACS, Embase, Web of Science, and OpenGrey databases were used to conduct this review. OpenGrey was used to search the grey literature and avoid any publication bias. The research material was obtained through careful assessment of each article found and by contacting authors by e-mail when necessary. Moreover, a manual search was performed by analyzing the reference list of each of the studies included in this review.
Search strategy
The MeSH (Medical Subject Headings) resource was used to select the key words. The Boolean operators ‘AND’ and ‘OR’ were used to enhance the search strategy through several combinations. The literature search was performed on November 30; 2016. Supplementary Table 1 shows the search strategy used. The results obtained were exported to Mendeley Desktop 1.13.3 software (Mendeley Ltd; London; UK); where duplicates were identified.
Study selection
The data collection was performed in two stages. In the first stage, the titles and abstracts were analyzed systematically by two eligibility reviewers independently (VLA and IFPL). The reviewers were not blinded to the names of authors or journals. Those articles whose title matched the objectives of the study but did not have an abstract available were obtained and analyzed in full. At this stage, studies outside the objective, case reports, letters to the editor and/or editorials, literature reviews, indexes, and abstracts were excluded.
In the second stage, the full texts of the studies meeting the criteria in the first stage were obtained and evaluated against the eligibility criteria. In specific cases, when a potentially eligible study presented incomplete data, the authors were contacted by e-mail in search of more information. When there was no mutual agreement between the two reviewers, a third reviewer (FRM) was involved in making a final decision. Studies rejected at this stage were recorded separately, along with the reasons for their exclusion.
Data collection process
One author collected the information required from the selected articles and a second author verified the information in order to confirm the quality of data extracted. Any disagreement was resolved through discussion with a third author. A fourth author was involved when assistance was required for the final decision. Contact attempts were made with the authors of the selected studies to retrieve missing information.
Data items
After screening, the texts of articles selected were reviewed and data were extracted in a standardized way. The information extracted included data such as author, year of publication, study location, and type of study, as well as the following information for the overall sample: sample size, average age, sex distribution, and time on HAART.
Risk of bias in individual studies
The quality and risk of bias of the studies included were assessed by two reviewers, independently, according to the PRISMA guidelines . This assessment prioritized the clear description of information and was performed in a blinded manner, hiding the names of authors and journals, and avoiding any potential bias and conflict of interest. The critical appraisal tool for cross-sectional studies from the Meta-Analysis of Statistics Assessment and Review Instrument (MAStARI) were used , which is based on nine criteria expressed as the following questions: (1) Was the study based on a random or pseudo-random sample? (2) Were inclusion criteria for the sample clearly defined? (3) Were confounding factors identified and strategies to manage them stated? (4) Were outcomes assessed with objective criteria? (5) Was there sufficient description of the groups for conducting comparisons? (6) Was follow-up carried out over a sufficient time period? (7) Were the outcomes of people who withdrew from the study described and included in the analysis? (8) Were outcomes measured in a reliable way? (9) Was an appropriate statistical analysis used?
Risk of bias was classified as high when the study had up to 49% ‘yes’ responses, moderate when the study had 50–69% ‘yes’ responses, and low when the study had over 70% ‘yes’ responses.
Summary measures and synthesis of results
The meta-analysis was conducted separately for each oral lesion. Relative risks (RRs) were the effect size measure used to estimate the association between HAART and each oral lesion. Publication bias was assessed using funnel plots. The Cochran Q test and the I 2 statistic were used to assess heterogeneity among studies. The null hypothesis of the Cochran Q test is that the variability among effect size estimates is zero. The I 2 statistic measure is the percentage of variability observed that may be attributed to heterogeneity. Values of I 2 higher than 0.5 indicate moderate to high heterogeneity. RRs were pooled using the Mantel–Haenszel method and the random-effects model with DerSimonian and Laird estimator. The Mantel–Haenszel method is more appropriate when the effect size is the RR, since it provides interval estimates with greater precision than those produced by the conventional inverse variance method. All statistical tests were two-sided and significance was defined at P < 0.05. The statistical analyses were performed with R version 3.2.1 (R Foundation for Statistical Computing).
Results
Study selection
Seven databases were searched and 522 studies were identified, of which 211 were duplicates. The titles and abstracts of the 311 non-duplicate studies were analyzed in order to eliminate those that did not match the objective of this review or that were literature reviews, case reports, or editorials and/or letters to the editor. The resulting 31 studies were selected for full manuscript reading and assessment against the remaining eligibility criteria of this review. Finally, only seven studies were eligible for inclusion in the qualitative and quantitative synthesis. The manual search through the reference list of each of the studies included resulted in 284 articles. However, none of these was included in this review. A flowchart of the selection process according to the PRISMA guidelines is presented in Fig. 1 .
Study characteristics
The studies were published from 2001 to 2015 and all of them were in English. They were conducted in seven different countries: UK , Tanzania , Brazil , Thailand , South Africa , Nepal , and India . All of the studies selected were observational. Supplementary Table 2 presents a summary of the studies included.
Risk of bias within studies
None of the studies included fulfilled all methodological quality criteria. However, all of them presented a moderate risk of bias according to the MAStARI checklist . All studies were cross-sectional in design, performed at a single time point, and did not consider follow-up or the outcomes of people who withdrew from the study. Thus, questions 6 and 7 of the checklist did not apply to these studies assessed for the risk of bias. Table 1 presents detailed information on the risk of bias of the studies included.
| Author | Questions a | Risk of bias | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | Q5 | Q6 | Q7 | Q8 | Q9 | ||
| Tappuni and Fleming | No | No | Yes | Yes | Yes | NA | NA | Yes | Yes | Moderate (55.5%) |
| Hamza et al. | No | No | Yes | Yes | Yes | NA | NA | Yes | Yes | Moderate (55.5%) |
| Lourenço and Figueiredo | No | Yes | Yes | Yes | Yes | NA | NA | Yes | Yes | Moderate (66.7%) |
| Nittayananta et al. | No | Yes | No | Yes | Yes | NA | NA | Yes | Yes | Moderate (55.5%) |
| Mthethwa et al. | Yes | Yes | No | Yes | Yes | NA | NA | Yes | Yes | Moderate (66.7%) |
| Naidu et al. | No | Yes | No | Yes | Yes | NA | NA | Yes | Yes | Moderate (55.5%) |
| Patil et al. | No | Yes | Yes | Yes | Yes | NA | NA | Yes | Yes | Moderate (66.7%) |
a Q1: Was the study based on a random or pseudo-random sample? Q2: Were inclusion criteria for the sample clearly defined? Q3: Were confounding factors identified and strategies to manage them stated? Q4: Were outcomes assessed with objective criteria? Q5: Was there sufficient description of the groups for conducting comparisons? Q6: Was follow-up carried out over a sufficient time period? Q7: Were the outcomes of people who withdrew from the study described and included in the analysis? Q8: Were outcomes measured in a reliable way? Q9: Was an appropriate statistical analysis used?
Synthesis of the results
Figure 2 presents the forest plots for the lesions of angular cheilitis, erythematous candidiasis, oral herpes, and pseudomembranous candidiasis. Data on the prevalence of angular cheilitis were available in three studies , involving 469 HIV-positive patients. All studies showed a lower prevalence of lesions for the HAART group. Meta-analysis showed that HAART was significantly associated with a lower prevalence of angular cheilitis (RR 0.53, 95% CI 0.30–0.96). Between-study heterogeneity was low ( I 2 = 0%, P = 0.8029).
