Methodological and analytical challenges in microbiome-HPV association studies.
We read the article by Rosário et al.1 in the Journal of Medical Virology with great interest. The authors conducted a study in women participating in the Regional Cervical Cancer Screening Program from the Northern Region of Portugal to evaluate the association between the composition of the cervicovaginal microbiome (CVM), high-risk human papillomavirus (hrHPV) status, and cytology results. Although the authors demonstrate interesting bacterial associations with women's age and hrHPV genotypes, we have several concerns regarding the study methods, results, and statistical analyses. The authors state that they investigated the CVM in their study, however, they only performed the characterization by polymerase chain reaction (PCR), targeting specifically bacterial microorganisms associated with the cervicovaginal microenvironment. This PCR-targeted approach has low species resolution and covers a minority of the CVM composition compared with traditional microbiome sequencing methods.2-4 Moreover, the commercial kits used in the study have not been properly investigated and validated for CVM profiling, especially in the context of hrHPV infections and cervical diseases, and thus, the associations described in the study might not be entirely accurate and should be interpreted cautiously. Further validation of these PCR-based kits should include assessing mock cervicovaginal community samples and comparison analyses with 16S rRNA gene sequencing of HPV-negative and HPV-positive samples. Species-level identification is extremely important for microbiome-hrHPV association studies due to the presence of diverse Lactobacillus species and the assortment of this complex in specific microbial communities with species-specific dynamics in health and disease.5 For instance, communities with Lactobacillus iners dominance are at higher risk of dysbiosis and hrHPV infection6 whereas communities with Lactobacillus crispatus dominance exhibit a protective role toward the development of hrHPV-induced cervical abnormalities.5 Since these species were not distinguished in the study due to the limited resolution of the methods, it is not adequate to strongly conclude the associations found with the Lactobacillus genus, challenging the interpretation and comparison of these results with current literature. Another concern is the described correlation of Gardnerella vaginalis, Atopobium vaginae, and Mobilincus species with protection against cervical abnormalities.1 Most microbiome-hrHPV studies have reported opposite results.2, 5, 7-9 In fact, these bacterial genera and species are characteristic of a Lactobacillus-depleted community that has a higher risk for hrHPV acquisition, persistence, and cervical disease development.5 The only described association of this community and species with a “health” stage throughout hrHPV infections is viral clearance,10 where it is been hypothesized that a shift in immune responses causes a temporary change in the CVM composition, thereby allowing the growth of these pathogenic species.10, 11 The authors also attributed these associations to women's age, and therefore, it is unclear whether age or the occurrence of cervical abnormalities is the contributing factor in these correlations.5, 12 Even though the study referred to these contradictory results in the discussion, we believe that this inconsistency should be further clarified. An in-depth analysis of the potential contributing factors in this population, such as age, ethnicity, immunity, hrHPV genotyping, bacterial vaginosis (BV), microbiome continuity and dynamics, viral-induced dysbiosis,6 and the results observed in the study is needed to fully understand these associations in the context of the natural history of HPV infections. As mentioned above, these inconsistent results may be explained by the study's targeted bacterial profiling approach with low resolution and by the statistical analyses. In the univariate logistic regression analyses,1 the authors showed results from a total of 72 statistical tests but according to the study's Methods and absence of q values, we assume that the authors did not correct their analyses for multiple comparisons. We have similar comments for the results obtained with the multivariate logistic regression analyses.1 In summary, the CVM has been the subject of increasing research interest due to its potential impact on the risk of cervical abnormalities and various gynecological conditions, including cervical cancer. While research into the CVM and cervical neoplasia is promising, it is essential to address methodological challenges and employ comprehensive, high-resolution multiomics investigations to advance our understanding and potentially translate these findings into improved gynecological care and cervical cancer preventive strategies, before strong conclusions on a limited set of data can be drawn. All authors contributed equally to this work. The authors declare no conflict of interest.
Read more