Analysis Of Factors Influencing The Presence Of Occult Cervical Cancer In Patients With HSIL And Establishment Of Nomogram Model

Objective:Among female malignant tumors,cervical cancer has the fourth highest incidence and death rate,which seriously endangers women’s health.Occult cervical cancer refers to invasive cervical cancer that is diagnosed histologically by cervical conization or hysterectomy without showing invasive cancer on both cervical colposcopic biopsy and endocervical curettage histopathology.The occurrence of occult cervical cancer is related to the screening and diagnosis and treatment process of cervical cancer.If not detected and treated in time,the late-stage cervical cancer seriously affects patients’survival prognosis.This study analyzed the factors influencing the presence of occult cervical cancer in patients with HSIL,screened the predictors,and developed a nomogram risk model in order to individually guide the scope of conization surgery,improve the detection rate of occult cervical cancer,achieve early diagnosis and standardized treatment,and thus improve the survival prognosis of patients with cervical cancer.Methods:1.A case-control study was used to collect 1179 patients who underwent colposcopic biopsy for HSIL and underwent conization of cervix or hysterectomy at the Department of Gynecology,Second Hospital of Shanxi Medical University between May2016 and July 2022 based on the inclusion and exclusion criteria,of which 887 patients from July 2018 to July 2022 were used as the training set for building the prediction model;292 patients from May 2016 to June 2018 were used as the validation set for external validation of the prediction model.2.The general,clinical and laboratory blood data of the patients were collected.General data included age,BMI,menopausal status,pregnancy and delivery;clinical data included cervical contactive bleeding,irregular bleeding,leucorrhea anomaly,HPV16/18infection,multiple HPV infections,TCT test results,colposcopic impression,acetowhite changes,mosaic,white gland,punctate vessels,atypical vessels,columnar ectopy,lugol’s staining,type of transformation zone,whether ECC test was performed,ECC results,colposcopic biopsy diagnosis,involvement of glands;laboratory routine blood information included preoperative peripheral blood lymphocyte count(LYMH),neutrophil count(NEUT),neutrophil-to-lymphocyte ratio(NLR),platelet(PLT),platelet-to-lymphocyte ratio(PLR),monocyte count(MONO)and lymphocyte-to-monocyte ratio(LMR).3.Univariate analysis was performed on all clinical variables in training set,and those variables with statistically significant differences were included in the LASSO regression model,and then their screened predictors were enrolled in the multivariable logistic regression analysis.The screened variables from the combined analyses were the independent risk factors for the existence of occult cervical cancer in patients with HSIL,and the nomogram prediction model was established based on the above risk factors.4.We evaluate and validate the nomogram prediction model.ROC curves,calibration curve plots,H-L tests,and DCA curves were used to evaluate the model’s discrimination,calibration and clinical utility,respectively.The model was internally validated using the 1000 bootstrap method,while external validation was performed using the validation set.Results:1.The differences between the training and validation set data were statistically significant(P<0.05)for cervical contactive bleeding,columnar ectopy,type of transformation zone,ECC results,and colposcopic biopsy diagnosis,while the others variables were not statistically significant between the two groups.2.A total of seven predictors were screened in the training set by univariate analysis,LASSO regression screening and logistic regression analysis.They were increasing age(OR=1.049,95%CI=1.023-1.076),multiple HPV infection(OR=0.501,95%CI=0.279-0.899),TCT test suggestive of high grade lesions(OR=1.737,95%CI=1.095-2.757),mosaic(OR=1.919,95%CI=1.091-3.378),punctate vessels(OR=1.678,95%CI=1.018-2.764),colposcopic biopsy diagnosed as CIN3(OR=4.209,95%CI=1.933-9.166),and ECC results≥CIN2(OR=3.959,95%CI=2.358-6.647).3.Based on the above seven predictors,a nomogram prediction model for the existence of occult cervical cancer in HSIL patients was developed.The AUCs of the training and validation sets were 0.819(95%CI:0.780-0.859)and 0.780(95%CI:0.707-0.852),respectively,indicating that the model had good discrimination.The H-L tests for the training and validation sets indicated thatχ~2 was 6.593(P=0.581)and 7.536(P=0.480)respectively,and also the calibration curve plots for both datasets showed that the standard and calibration curves were following the same direction,indicating that the prediction model was well calibrated.Drawing the DCA curves of two datasets respectively showed that the model had good clinical practicability.Conclusion:Increasing age,TCT test results suggestive of high-grade lesions,mosaic,punctate vessels,colposcopic biopsy diagnosis of CIN3,and ECC results≥CIN2 were independent risk factors for the existence of occult cervical cancer in patients with HSIL.Multiple HPV infection is a protective factor for the existence of occult cervical cancer in patients with HSIL.The nomogram prediction model based on the above seven factors had good discrimination,calibration and high clinical utility and provided a basis for early detection of occult cervical cancer.