TY - JOUR
T1 - Artificial intelligence estimates the impact of human papillomavirus types in influencing the risk of cervical dysplasia recurrence
T2 - Progress toward a more personalized approach
AU - Bogani, Giorgio
AU - Ditto, Antonino
AU - Martinelli, Fabio
AU - Signorelli, Mauro
AU - Chiappa, Valentina
AU - Leone Roberti Maggiore, Umberto
AU - Taverna, Francesca
AU - Lombardo, Claudia
AU - Borghi, Chiara
AU - Scaffa, Cono
AU - Lorusso, Domenica
AU - Raspagliesi, Francesco
PY - 2019/3/1
Y1 - 2019/3/1
N2 - The objective of this study was to determine whether the pretreatment human papillomavirus (HPV) genotype might predict the risk of cervical dysplasia persistence/recurrence. Retrospective analysis of prospectively collected data of consecutive 5104 women who underwent the HPV-DNA test were matched with retrospective data of women undergoing either follow-up or medical/surgical treatment(s) for genital HPV-related infection(s). Artificial neuronal network (ANN) analysis was used in order to weight the importance of different HPV genotypes in predicting cervical dysplasia persistence/recurrence. ANN simulates a biological neuronal system from both the structural and functional points of view: like neurons, ANN acquires knowledge through a learning-phase process and allows weighting the importance of covariates, thus establishing how much a variable influences a multifactor phenomenon. Overall, 5104 women were tested for HPV. Among them, 1273 (25%) patients underwent treatment for HPV-related disorders. LASER conization and cervical vaporization were performed in 807 (59%) and 386 (30%) patients, respectively, and secondary cervical conization in 45 (5.5%). ANN technology showed that the most important genotypes predicting cervical dysplasia persistence/recurrence were HPV-16 (normalized importance: 100%), HPV-59 (normalized importance: 51.2%), HPV-52 (normalized importance: 47.7%), HPV-18 (normalized importance: 32.8%) and HPV-45 (normalized importance: 30.2%). The pretreatment diagnosis of all of those genotypes, except HPV-45, correlated with an increased risk of cervical dysplasia persistence/recurrence; the pretreatment diagnosis was also arrived at using standard univariate and multivariable models (P<0.01). Pretreatment positivity for HPV-16, HPV-18, HPV-52 and HPV-59 might correlate with an increased risk of cervical dysplasia persistence/recurrence after treatment. These data might be helpful during patients' counseling and to implement new vaccination programs.
AB - The objective of this study was to determine whether the pretreatment human papillomavirus (HPV) genotype might predict the risk of cervical dysplasia persistence/recurrence. Retrospective analysis of prospectively collected data of consecutive 5104 women who underwent the HPV-DNA test were matched with retrospective data of women undergoing either follow-up or medical/surgical treatment(s) for genital HPV-related infection(s). Artificial neuronal network (ANN) analysis was used in order to weight the importance of different HPV genotypes in predicting cervical dysplasia persistence/recurrence. ANN simulates a biological neuronal system from both the structural and functional points of view: like neurons, ANN acquires knowledge through a learning-phase process and allows weighting the importance of covariates, thus establishing how much a variable influences a multifactor phenomenon. Overall, 5104 women were tested for HPV. Among them, 1273 (25%) patients underwent treatment for HPV-related disorders. LASER conization and cervical vaporization were performed in 807 (59%) and 386 (30%) patients, respectively, and secondary cervical conization in 45 (5.5%). ANN technology showed that the most important genotypes predicting cervical dysplasia persistence/recurrence were HPV-16 (normalized importance: 100%), HPV-59 (normalized importance: 51.2%), HPV-52 (normalized importance: 47.7%), HPV-18 (normalized importance: 32.8%) and HPV-45 (normalized importance: 30.2%). The pretreatment diagnosis of all of those genotypes, except HPV-45, correlated with an increased risk of cervical dysplasia persistence/recurrence; the pretreatment diagnosis was also arrived at using standard univariate and multivariable models (P<0.01). Pretreatment positivity for HPV-16, HPV-18, HPV-52 and HPV-59 might correlate with an increased risk of cervical dysplasia persistence/recurrence after treatment. These data might be helpful during patients' counseling and to implement new vaccination programs.
KW - artificial neuronal network analysis
KW - cervical dysplasia
KW - conization
KW - genotypes
KW - human papillomavirus
UR - http://www.scopus.com/inward/record.url?scp=85060909351&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85060909351&partnerID=8YFLogxK
U2 - 10.1097/CEJ.0000000000000432
DO - 10.1097/CEJ.0000000000000432
M3 - Article
C2 - 29360648
AN - SCOPUS:85060909351
VL - 28
SP - 81
EP - 86
JO - European Journal of Cancer Prevention
JF - European Journal of Cancer Prevention
SN - 0959-8278
IS - 2
ER -