A Study On The Prediction Of Post-molc Gestational Trophoblastic Neoplasia And Chemotherapy Resistance Using Sonographic Parameters

BackgroundThe incidence of hydatidiform mole is around3.8to13cases per1000pregnancies in Asia, which is7-10times more than the incidence in North America and Europe (0.5-1.84/1000). In China, this incidence is5/1000. Although hydatidiform mole is benign, it, especially complete hydatidiform mole (CHM), possibly develops into gestational trophoblastic neoplasia (GTN). The malignant change arises after15%of CHM, which is2000-fold higher than that develops from miscarriage, termination of pregnancy and full-term delivery.Unfortunately, the features of the high-risk population are still controversial. We are not able to know what kind of patients will develop to malignancy in the future. The only thing we should do for every patient is closer follow-up using serum hCG test and imaging in the following2years, as FIGO (2000) recommended, although more than half cases finally show the spontaneous regression. The purpose of closer follow-up evaluation is to early detect the malignant change and give them prompt treatment, but it might bring life inconvenience and considerable psychological distress to patients, as well as increase of social health expenditure. On the other hand, the non-compliant patients possibly develop to severe stage in the case of lacking professional medical supervision. Therefore, the prognostic factor identification in hydatidiform mole is desirable. It appears useful for subgroup patients in different classes according to the risk of developing to GTN. Thus an individualization of follow-up management will be permitted, paying more attention to the high-risk hydatidiform mole and shortening the monitoring period for low-risk group.Once GTN occures, arising from hydatidiform mole or non-molar pregnancies, other thorny issue comes up. This issue is which chemotherapeutic drugs are the most effective, the least toxic and the most suitable for a patient. GTN is a highly sensitive to chemotherapy and a highly curable malignant tumor, expecially low-risk GTN with nearly100%cure rate. Methotrexate (MTX) single regimen is frequently used as the initial chemotherapy in low-risk GTN due to its low toxicity. However, approximately30%of these patients become resistant to MTX, which is just recognized after several courses, and then these patients require other agents. Switching drug after the development of MTX resistance prolongs the overall duration of chemotherapy and accumulates the side effects in the normal organs. Therefore, there is a need to solve this issue to identify the response to MTX before starting treatment and to avoid giving the less effective regimen to the MTX resistances.The basis of GTN occurrence is considered trophoblastic cells invading uterine myometrial vessels, and then both vessel morphology and blood flows are changed. It is important to evaluate the new blood vessel formation (neovascularity) because it is a common feature of malignancy. GTN vessel evaluation depends on in-vivo technique instead of ex-vivo pathological observation due to lack of post-operational tissue documents. In morden clinical practice, high-resolution ultrasound is one of real-time blood flow imaging methods. With the revolusion of medical equipments, ultrasound has been developing to be competent for showing millimeter-level vessels in Color Doppler Flow Imaging (CDFI), and qualitifing the hemodynamic parameters in Spectral Doppler. Also, it can be used to measure the size, location and echo feature of organs and lesions in the gray scale. Thanks to the merits of safety and cheap, ultrasound is widely used in diagnosis, measurement and treatment monitoring of gestational trophoblastic disease. The routine use of ultrasonography leads to early detection, subsequently, the cure rate increases and the incidence of complications dramatically decreases. Thus, it is universally acknowledged that pelvic ultrasound is the first-line imaging method to gestational trophoblastic disease.In this study, we used sonographic parameters, such as morphology, size and blood supplyment of uterus and ovary, to investigate the methods to predicting post-mole GTN on the stage of CHM evacuation, and predicting MTX resistance before chemotherapy. These methods will be beneficial to determine strategies of stratification management and personalized treatment. This study includes two parts:Part I is to setup a logistic regression model for predicting CHM outcome by evaluating the pre-and post-evacuation sonographic parameters; Part II is to predict the MTX resistance in low-risk GTN patients before the initial chemotherapy by using sonographic parameters, especially hemodynamic index.Part I The Value of Pre-and Post-evacuation Sonographic Parameters in Predicting Post-mole Gestational Trophoblastic Neoplasia from Women with Complete Hydatidiform MoleObjectiveTo setup a logistic regression model, including multiple factors, especially pre-and post-evacuation sonographic parameters, for predicting post-mole GTN from a large cohort of CHM patients. MethodsThe prediction model was trained on the retrospective346cases that were treated between January2005and August2012in Women’s Hospital, School of Medicine, Zhejiang University (training set). The following clinical information was recorded:age, gestational age, pre-evacuation serum hCG level, differentiation on histology, pre-evacuation sonographic features (the length, width and anteroposterior diameter of uterus and cavity mass, ovarian cyst, Color Doppler signals), post-evacuation sonographic features (ovarian cyst, Color Doppler signals) and2-year follow-up outcomes. This prediction model was further tested on patients who have been prospectively recruited since September2012(test set). The software used for statistical analysis was SPSS19.0for Windows and R version3.1.1for Mac.Results1The gestational age was shown to affect on the clinical, histological and sonographic features, but it was not statistically significant between training set and test set, or between the patients developing into pGTN and ones going to spontaneous regression. Thus, this comparative study was based on similar gestational ages when ultrasound exams were done, which means the difference between groups was not induced by gestational age.2The results of univariable logistic regression analysis identified that serum hCG level, the size (length, width, anteroposterior diameter and volume) of uterus and cavity mass, the abnormal abundant Color Doppler signals on the pre-and post-evacuation uterus were the significant predictors for pGTN (p<0.05). However, on the time scale of gestational age, the95%confidence interval of malignant group overlapped with normal group at the certain time points. 3. Multivariable logistic regression analysis indicated that serum hCG level and the abnormal abundant Color Doppler signals on the post-evacuation uterus were significantly associated with malignant outcome with relative risk of9.62and73.78.4The fitted logistic model wasLog (p(malignancy=1)/p(normal=0))=-2.23+4.39e-7×hCG+4.11e-5×pre-uterus volume+2.30×abundant blood in post-uterus.The maximum area under curve was0.846(P<0.05). This model had71.6%sensitivity and87.5%specificity at the cut-off value of0.146. Malignance will be predicted if more than0.146.5. Evaluated on the test set, the performance was80.0%positive predictive value,92.0%negative predictive value,85.7%sensitivity,88.5%specificity,7.45positive likelihood ratio,0.16negative likelihood ratio and87.9%accuracy.ConclusionsThe regression model on the basis of pre-and post-evacuation ultrasound and serum hCG has the potential value to be an easy, cheap and safe method to accurately predict the prognosis of CHM. Part Ⅱ The Value of Uterine Artery Hemodynamic Parameters Before Chemotherapy in Predicting Methotrexate Resistance in Low-risk Gestational Trophoblastic NeoplasiaObjectiveTo investigate the mechanism of uterine artery hemodynamic changes in GTN, and the relationship between uterine artery hemodynamic parameters measured before chemotherapy and the final MTX response, in order to find sonographic predictive parameters.MethodsProspective analysis was carried out in a total of58low-risk GTN patinets treated with MTX between September2012and Match2015in Women’s Hospital, School of Medicine, Zhejiang University. Hemodynamic parameters (PS, ED, TAmax, TAmean, S/D, PI and RI) in uterine artery were checked in every patient by ultrasound. In the case of uterine lesion detected, the patient was measured for additional sonographic parameters, including lesion size and hemodynamic parameters in the lesions’vessels. The relationships between sonographic parameters and MTX response were analyzed.Results1The GTN vascularity in uterus induced the hemodynamic change in uterine artery. Every hemodynamic parameter (ED, TAmax, TAmean, S/D, PI or RI) in uterine artery showed significant positive correlation with one in uterine lesion (p<0.05), and PS closed to significant positive correlation (p=0.057).2The hemodynamic parameters were non-uniform in the GTN uterine lesion. The maximum was1.10-27.52times to the minimum in the different sample points within the same lesion. The hemodynamic parameters in GTN uterine artery indirectly but stably reflect the changes of GTN vascularity in the uterine lesion.3In the group of patients without uterine lesion, neither clinical feature nor sonographic parameters before chemotherapy showed significant difference with MTX response (p>0.05).4In the group of patients with uterine lesion, the results of univariable logistic regression analysis identified that serum hCG level, FIGO score, hemodynamic parameters of uterine artery (the maximal PS, ED, TAmax and TAmean) were the significant predictor for MTX response (p<0.05). Multivariable logistic regression analysis indicated that the maximal PS, TAmax and TAmean of uterine artery were independent predictors to MTX response. Among them, TAmean was most powerful to predict MTX response with0.720AUC. It had75.0%sensitivity and63.2%specificity at the cutoff value of19.16cm/s.Conclusions1The hemodynamic parameters in uterine artery measured before chemotherapy, such as PS, TAmax and TAmean, could be used as an independent factor for predicting MTX response in the low-risk GTN patients.2The hemodynamic parameters in GTN uterine artery change with the vascularity in uterine lesion, which is indirectly but full-scaly reflect the changes of GTN vascularity in the uterine lesion and avoids the tumor vessels heterogeneity.