妊娠滋养细胞肿瘤单药化疗疗效影响因素分析及FIGO 2000预后评分系统简化

Evaluation and simplification of risk factors in FIGO 2000 scoring system for gestational trophoblastic neoplasia: a 19-year retrospective analysis 妊娠滋养细胞肿瘤单药化疗疗效影响因素分析及FIGO 2000预后评分系统简化 1 first-author 1 1 2 1 2 3 2 3 2 3 4 corresp Weiguo LU, lbwg@zju.edu.cn Weiguo LU, lbwg@zju.edu.cn lbwg@zju.edu.cn https://orcid.org/0000-0003-2062-7145 Women’s Reproductive Health Key Research Laboratory of Zhejiang Province, Women’s Hospital, School of Medicine, Zhejiang University, Hangzhou 310006, China Women’s Reproductive Health Key Research Laboratory of Zhejiang Province, Women’s Hospital, School of Medicine, Zhejiang University, Hangzhou 310006, China Department of Gynecologic Oncology, Women’s Hospital, School of Medicine, Zhejiang University, Hangzhou 310006, China Department of Gynecologic Oncology, Women’s Hospital, School of Medicine, Zhejiang University, Hangzhou 310006, China Center for Uterine Cancer Diagnosis and Therapy Research of Zhejiang Province, Hangzhou 310006, China Center for Uterine Cancer Diagnosis and Therapy Research of Zhejiang Province, Hangzhou 310006, China Cancer Center of Zhejiang University, Hangzhou 310058, China Cancer Center of Zhejiang University, Hangzhou 310058, China Objective 2 The International Federation of Gynecology and Obstetrics (FIGO) 2000 scoring system classifies gestational trophoblastic neoplasia (GTN) patients into low- and high-risk groups, so that single- or multi-agent chemotherapy can be administered accordingly. However, a number of FIGO-defined low-risk patients still exhibit resistance to single-agent regimens, and the risk factors currently adopted in the FIGO scoring system possess inequable values for predicting single-agent chemoresistance. The purpose of this study is therefore to evaluate the efficacy of risk factors in predicting single-agent chemoresistance and explore the feasibility of simplifying the FIGO 2000 scoring system for GTN. Methods 2 The clinical data of 578 GTN patients who received chemotherapy between January 2000 and December 2018 were retrospectively reviewed. Univariate and multivariate logistic regression analyses were carried out to identify risk factors associated with single-agent chemoresistance in low-risk GTN patients. Then, simplified models were built and compared with the original FIGO 2000 scoring system. Results 2 Among the eight FIGO risk factors, the univariate and multivariate analyses identified that pretreatment serum human chorionic gonadotropin (hCG) level and interval from antecedent pregnancy were consistently independent predictors for both first-line and subsequent single-agent chemoresistance. The simplified model with two independent factors showed a better performance in predicting single-agent chemoresistance than the model with the other four non-independent factors. However, the addition of other co-factors did improve the efficiency. Overall, simplified models can achieve favorable performance, but the original FIGO 2000 prognostic system still features the highest discrimination. Conclusions 2 Pretreatment serum hCG level and interval from antecedent pregnancy were independent predictors for both first-line and subsequent single-agent chemoresistance, and they had greater weight than other non-independent factors in predicting single-agent chemoresistance. The simplified model composed of certain selected factors is a promising alternative to the original FIGO 2000 prognostic system, and it shows comparable performance. 目的： 2 评估现行国际妇产科联盟（FIGO）2000预后评分系统中各因素的作用，从中筛选出与单药化疗失败相关的独立危险因素，并基于此简化现行的预后评分系统。 创新点： 2 临床数据样本量大，简化模型预测效能优秀，具有一定的临床价值。 方法： 2 回顾性收集浙大妇院过去19年（2000年1月至2018年12月）收治的共578例妊娠滋养细胞肿瘤患者的临床数据，通过单因素和多因素Logistics回归分析筛选出与单药化疗失败相关的独立危险因素，从而在现行的FIGO 2000预后评分系统的基础上建立更简化的预测模型。 结论： 2 先前妊娠距化疗开始间隔时间和治疗前血清人绒毛膜促性腺激素（hCG）水平是妊娠滋养细胞肿瘤患者单药化疗失败的独立危险因素，基于这两项独立危险因素建立的简化模型对于妊娠滋养细胞肿瘤患者单药化疗疗效的预测具有良好效能。 Gestational trophoblastic neoplasia (GTN) Single-agent chemotherapy Chemoresistance Risk factor 妊娠滋养细胞肿瘤 单药化疗 化疗耐药 危险因素 1 Introduction 1 Gestational trophoblastic neoplasia (GTN) is a sort of pregnancy-related, frequently malignant trophoblastic disease, which includes invasive mole, choriocarcinoma, placental site trophoblastic tumor, and epithelioid trophoblastic tumor. GTN is a less common group of gynecological cancers, and it can be mostly cured by chemotherapy alone even with wide metastases, with the effective drugs introduced almost 60 years ago ( Lurain, 2003 Lurain, 2003 ). In 2002, the Cancer Committee of International Federation of Gynecology and Obstetrics (FIGO) combined an anatomic FIGO system with the modified World Health Organization (WHO) scoring system to provide clinicians with a unified FIGO classification for reporting GTN, which allowed the valid comparison of the outcomes from different institutions ( FIGO Oncology Committee, 2002 FIGO Oncology Committee, 2002 ). This prognostic scoring system, which is comprised of eight factors to calculate the risk of developing chemoresistance to single-agent therapies, stratifies GTN patients into two categories. A score of 6 or less indicates a "low risk" of resistance to single-agent chemotherapy with methotrexate or actinomycin-D, whereas a score of 7 or more signifies a "high risk" of resistance to single drug and demands combination chemotherapy ( Kohorn, 2002 Kohorn, 2002 ). Guided by the well-established FIGO 2000 system, the primary remission rates with methotrexate therapies reached 69%‒93% ( Smith et al., 1982 Smith et al., 1982 ; Wong et al., 1985 Wong et al., 1985 ; Gleeson et al., 1993 Gleeson et al., 1993 ; Abrão et al., 2008 Abrão et al., 2008 ; Lertkhachonsuk et al., 2009 Lertkhachonsuk et al., 2009 ; Kang et al., 2010b Kang et al., 2010b ; Uberti et al., 2015 Uberti et al., 2015 ), and the complete remission (CR) rates of salvage actinomycin-D therapies reached 75%‍‒‍92% ( Abrão et al., 2008 Abrão et al., 2008 ; Chalouhi et al., 2009 Chalouhi et al., 2009 ; Lurain et al., 2012 Lurain et al., 2012 ; Prouvot et al., 2018 Prouvot et al., 2018 ). Other alternative single-agent options for low-risk GTN patients in Asia include fluorouracil and etoposide, and their remission rates varied from 67% to 85% based on different protocols ( Sung et al., 1984 Sung et al., 1984 ; Lurain and Elfstrand, 1995 Lurain and Elfstrand, 1995 ; Cyriac et al., 2011 Cyriac et al., 2011 ; Mora et al., 2019 Mora et al., 2019 ). However, there is still a gap between the successful rate of single-agent chemotherapy predicted by the FIGO 2000 prognostic system and the actual outcomes. Furthermore, this system is quite elaborate, because several of the risk factors are relevant to tumor burden and therefore may be repetitive and interrelated with each other. In addition, a wide range of variables feature a growing variability, which may lead to inconvenience for clinical implementation and difficulties in disease management. In this study, we retrospectively reviewed GTN patients treated in a territorial referral center of East China, and analyzed the FIGO risk factors associated with single-agent chemoresistance in low-risk patients. Our aim was to explore any potential improvements to the risk evaluation of single-agent chemoresistance, so that the accuracy of the FIGO prognostic scoring system can be enhanced and it can be made more concise. 2 Materials and methods 1 2.1 Patients and data 2 The data of patients diagnosed with GTN between January 2000 and December 2018 were retrospectively collected from the Medical Record Review System of Women's Hospital, School of Medicine, Zhejiang University, China. Patients with incomplete data of diagnosis or treatments or the presence of placental site trophoblastic tumor or epithelioid trophoblastic tumor were excluded. All patients underwent a comprehensive evaluation prior to first chemotherapy at our hospital and were classified according to the revised FIGO 2000 staging and scoring system. A score of 6 or less was considered as low risk and a score of 7 or more was ranked as high risk ( Ngan et al., 2018 Ngan et al., 2018 ). Considering the retrospective nature of the study, no sample size calculation was performed, and all eligible patients were included in the analysis. The data on demographics and clinical characteristics included patient age, history of previous pregnancy and delivery, history of surgery, physical and laboratory examination, and pathological diagnosis if the patient had undergone surgery. In addition, information on primary, secondary and subsequent treatments, as well as the relevant responses was collected. 2.2 Treatment protocols 2 The treatments were formulated by the stage and risk score. Low-risk and high-risk patients were started with single-agent and multi-agent chemotherapy regimens, respectively. During the studied period of 19 years, several chemotherapy regimens were in‍volved. The single-agent chemotherapy regimens included: (1) 5-d methotrexate, (2) 8-d methotrexate/folinic acid, (3) weekly methotrexate, (4) 5-d actinomycin-D, (5) 8-d 5-fluorouracil, and (6) 5-d etoposide. The multi-agent chemotherapy regimens included: (1) EMA-CO (etoposide, methotrexate, actinomycin-D, cyclophosphamide, vincristine), (2) EMA/EP (etoposide, methotrexate, actinomycin-D/etoposide, cisplatin) or EP/EMA, and (3) TP/TE (paclitaxel, cisplatin/paclitaxel, etoposide). Drug resistance was clinically defined as follows: after two successive chemotherapy cycles, (1) the serum human chorionic gonadotropin (hCG) level was not decreased logarithmically, or remained at a plateau above normal or increased; or (2) imaging tests indicated that the tumor size was enlarged or new lesions emerged ( Song et al., 2004 Song et al., 2004 ). Toxicity was scaled in each cycle conforming to WHO criteria ( Miller et al., 1981 Miller et al., 1981 ). In cases with the presence of drug resistance or unendurable toxicity, the chemotherapy regimens would be changed. For low-risk patients, the initial single-agent regimen would be replaced with an alternative single-agent regimen and the multi-agent chemotherapy would be administrated when secondary single-agent regimens still failed. Additional consolidation cycles were continued for at least two rounds in low-risk patients and three in high-risk patients after serum hCG normalization. A CR was defined as three consecutive weekly normal serum hCG levels ( < 5.3 IU/L) after completing the treatments. 2.3 Statistical analysis 2 The patient characteristics were summarized by descriptive statistics. Continuous data were presented as median and interquartile range (IQR), while categorical data were reported as number and proportion. Univariate and multivariate logistic regression analyses were conducted to identify the values of the risk factors for predicting single-agent chemoresistance in low-risk GTN patients, with a two-side P ≤0.05 set as the statistical significant level. Then, modified risk models based on multivariate logistic regression were built for predicting single-agent chemoresistance, and the receiver operating characteristic (ROC) curves were used to assess the performance. Moreover, multivariate logistic regression using stepwise process was carried out in all patients based on the previous results. A number of simplified models that had excellent classification performance in accordance with the original FIGO system were established, and the ROC curves were used to evaluate the discrimination. All data management and statistical analyses were accomplished by IBM SPSS Statistics (Version 20.0, International Business Machines Corporation, Armonk, NY, USA). 3 Results 1 3.1 Overview of patient information 2 This study included a total of 578 GTN patients, among which 475 were low-risk and 103 were high-risk according to the revised FIGO 2000 staging and scoring system ( Ngan et al., 2018 Ngan et al., 2018 ). The median (IQR) age of enrolled patients was 28 (25, 37) years and the median (IQR) value of pretreatment serum hCG level was 5612 (1201, 2‍3 ‍302) IU/L. Metastases outside the uterus were found in 69.6% of patients. Fig. ‍‍1 shows the flowchart of patient inclusion. All patients were treated with chemotherapy and some of them underwent concomitant surgeries depending on their condition, among which 7.6% of low-risk patients and 18.4% of high-risk patients underwent hysterectomy. The demographic and clinical characteristics of the patients were summarized in Table 1 Table 1 . Among low-risk patients, 58.1% achieved CR with first-line single-agent, an additional 30.7% achieved CR with secondary single-agent, and the remaining 11.2% needed multi-agent to salvage. All of the low-risk patients eventually achieved CR. Data are expressed as median (IQR) or number (percentage). hCG: human chorionic gonadotropin; FIGO: International Federation of Gynecology and Obstetrics; IQR: interquartile range. Characteristics Low-risk ( n =475) High-risk ( n =103) Age (years) 28 (25, 37) 29 (25, 38) Obstetric history Gravidity 2 (1, 3) 2 (2, 3) Parity 0 (0, 1) 1 (0, 1) Pretreatment serum hCG (IU/L) 3996 (910, 15 ‍719) 42 512 (10 ‍000, 178‍ ‍413) FIGO score 2 (1, 3) 8 (7, 10) FIGO stage I 151 (31.8%) 25 (24.3%) II 6 (1.3%) 5 (4.9%) III 318 (66.9%) 66 (64.1%) IV 0 7 (6.8%) Surgery Hysterectomy 36 (7.6%) 19 (18.4%) Uterine lesion resection 5 (1.1%) 9 (8.7%) Pulmonary lobectomy 2 (0.4%) 2 (1.9%) Other 21 (4.4%) 28 (27.2%) 3.2 Risk factor analysis 2 In order to determine the risk factors associated with single-agent chemoresistance, univariate and multivariate analyses were performed in 439 of 475 low-risk patients. The remaining 36 patients underwent hysterectomy, among which 25 received surgery as a primary therapy and 11 received it after first-line or second-line chemotherapy as a salvage treatment. These patients were excluded because advanced age was the main indication of surgery and the operation was influential to subsequent treatment, which would consequently result in a large bias. The details of the single-agent chemotherapy regimens of 439 low-risk patients were shown in Table 2 Table 2 . Six FIGO risk factors were taken into account, including age, antecedent pregnancy type, interval from antecedent pregnancy, pretreatment serum hCG level, largest tumor size including uterus, and number of metastases. The site of metastases was not listed because low-risk patients were confirmed as having no distant metastases other than in the lung, and previous failed chemotherapy was not shown because the included low-risk patients were all scored before first-line chemotherapy. * The course interval usually means the interval from the first day of the last course to the first day of the next course. IV: intravenous injection; IM: intramuscular injection. 2 Agent 2 Usage 2 Course interval * (d) 2 Number of patients Agent used as first-line therapy Agent used as second-line therapy Methotrexate 0.4 mg/(kg·d) IV or IM for 5 d 14 401 2 Methotrexate/folinic acid Methotrexate 1 mg/(kg·d) IM on Days 1, 3, 5, and 7 with folinic acid 0.1 mg/(kg·d) on Days 2, 4, 6, and 8 14 30 1 Methotrexate weekly 30‒50 mg/m 2 IM weekly 7 4 0 Actinomycin-D 10‒12 µg/(kg·d) IV for 5 d 14 2 193 5-Fluorouracil 28‒30 mg/(kg·d) IV for 8 d 21 2 4 Etoposide 100 mg/(m 2 ·d) IV for 5 d 14 0 1 Univariate and multivariate analyses were carried out sequentially in low-risk GTN patients treated with single-agent chemotherapy. The univariate analysis revealed that the interval from antecedent pregnancy and pretreatment serum hCG level were significantly associated with the risk of resistance to first-line single-agent therapy, whereas age, antecedent pregnancy type, largest tumor size, or the number of metastases was not. In multivariate analysis, interval exceeding 12 months (odds ratio (OR)=22.37, P =0.003) and pretreatment serum hCG level between 10‍ ‍000 and 100 ‍000 IU/L (OR=2.79, P =0.001) were detected as strongly related to first-line single-agent chemoresistance. The detailed results were shown in Table 3 Table 3 . hCG: human chorionic gonadotropin; OR: odds ratio; CI: confidence interval. 2 Characteristics 2 Rate of first-line single agent resistance (%) 2 Univariate 2 Multivariate OR (95% CI) P value OR (95% CI) P value 2 Age (years) 0.303 0.506 <40 43.2 (161/373) Reference Reference ≥40 36.4 (24/66) 0.75 (0.44‒1.29) 0.303 0.82 (0.47‒1.46) 0.506 Antecedent pregnancy 0.586 0.575 Hydatidiform mole 41.9 (163/389) Reference Reference Miscarriage 47.5 (19/40) 1.25 (0.65‒2.41) 0.496 1.07 (0.51‒2.25) 0.857 Term 30.0 (3/10) 0.59 (0.15‒2.33) 0.456 0.48 (0.12‒1.97) 0.305 Interval (months) 0.014 0.012 <4 41.4 (163/394) Reference Reference 4‒6 25.9 (7/27) 0.50 (0.20‒1.20) 0.120 0.58 (0.22‒1.54) 0.275 7‒12 60.0 (3/5) 2.13 (0.35‒12.87) 0.412 2.47 (0.40‒15.38) 0.334 >12 92.3 (12/13) 17.01 (2.19‒132.08) 0.007 22.37 (2.79‒179.59) 0.003 2 Pretreatment serum hCG (IU/L) 0.020 0.008 < 1×10 3 37.4 (46/123) Reference Reference 1×10 3 ‒1×10 4 36.8 (70/190) 0.98 (0.61‒1.56) 0.921 1.24 (0.74‒2.08) 0.415 1×10 4 ‒1×10 5 53.7 (66/123) 1.94 (1.17‒3.22) 0.011 2.79 (1.49‒5.25) 0.001 > 1×10 5 100.0 (3/3) 0.999 0.999 Largest tumor size including uterus (cm) 0.463 0.514 <3 41.2 (127/308) Reference Reference 3‒<5 41.9 (44/105) 1.03 (0.66‒1.61) 0.904 0.73 (0.43‒1.26) 0.259 ≥5 53.8 (14/26) 1.66 (0.74‒3.71) 0.215 0.95 (0.38‒2.38) 0.908 Number of metastases 0.748 0.810 0 42.1 (146/347) Reference Reference 1‒4 43.5 (37/85) 1.06 (0.66‒1.71) 0.808 1.08 (0.65‒1.79) 0.770 ≥5 28.6 (2/7) 0.55 (0.11‒2.88) 0.479 0.62 (0.11‒3.35) 0.576 When patients failed to respond to a first-line single-agent, they were referred to a second-line single-agent. Thus, the risks of chemoresistance to both first-line and second-line single agents were analyzed. The interval from antecedent pregnancy, pretreatment serum hCG level, and largest tumor size were found to be statistically significant by univariate analysis. Subsequent multivariate logistic regression analysis consistently recognized an interval exceeding 7 months (OR=9.44, P =0.027) and 12 months (OR=7.02, P =0.014), and pretreatment serum hCG level above 10 ‍000 IU/L (OR=5.58, P =0.003) and 100 ‍000 IU/L (OR=34.88, P =0.009) as the independent predictive factors for single-agent chemoresistance, although the numbers in some categories were relatively small. The obtained data were presented in Table 4 Table 4 . hCG: human chorionic gonadotropin; OR: odds ratio; CI: confidence interval. 2 Characteristics 2 Rate of single-agent resistance (%) 2 Univariate 2 Multivariate OR (95% CI) P value OR (95% CI) P value 2 Age (years) 0.100 0.208 <40 11.5 (43/373) Reference Reference ≥40 4.5 (3/66) 0.37 (0.11‒1.21) 0.100 0.45 (0.13‒1.56) 0.208 2 Antecedent pregnancy 0.555 0.875 Hydatidiform mole 10.0 (39/389) Reference Reference Miscarriage 12.5 (5/40) 1.28 (0.47‒3.46) 0.624 0.86 (0.28‒2.60) 0.788 Term 20.0 (2/10) 2.24 (0.46‒10.94) 0.317 1.43 (0.28‒7.39) 0.673 2 Interval (months) 0.115 0.022 <4 9.6 (38/394) Reference Reference 4‒6 11.1 (3/27) 1.17 (0.34‒4.07) 0.804 1.70 (0.39‒7.37) 0.477 7‒12 40.0 (2/5) 6.25 (1.01‒38.55) 0.049 9.44 (1.29‒69.06) 0.027 >12 23.1 (3/13) 2.81 (0.74‒10.66) 0.129 7.02 (1.48‒33.30) 0.014 2 Pretreatment serum hCG (IU/L) <0.001 0.002 < 1×10 3 6.5 (8/123) Reference Reference 1×10 3 ‒1×10 4 6.3 (12/190) 0.97 (0.38‒2.44) 0.947 1.60 (0.56‒4.60) 0.383 1×10 4 ‒1×10 5 19.5 (24/123) 3.48 (1.50‒8.11) 0.004 5.58 (1.81‒17.26) 0.003 > 1×10 5 66.7 (2/3) 28.75 (2.35‒352.05) 0.009 34.88 (2.41‒504.51) 0.009 2 Largest tumor size including uterus (cm) 0.044 0.972 <3 8.1 (25/308) Reference Reference 3‒<5 15.2 (16/105) 2.04 (1.04‒3.98) 0.038 1.06 (0.47‒2.41) 0.890 ≥5 19.2 (5/26) 2.70 (0.94‒7.76) 0.066 1.16 (0.33‒4.00) 0.817 2 Number of metastases 0.919 0.978 0 11.0 (38/347) Reference Reference 1‒4 9.4 (8/85) 0.84 (0.38‒1.88) 0.680 0.91 (0.39‒2.13) 0.832 ≥5 0 (0/7) 0.999 0.999 3.3 Comparison between the simplified models and the FIGO 2000 scoring system 2 We applied ROC curves on the simplified risk models based on multivariate logistic regression results to measure their capabilities of predicting single-agent chemoresistance in low-risk patients. As shown in Fig. 2 Fig. 2 , the simplified models with selected risk factors revealed discrepant predicting capacities. For instance, the interval from antecedent pregnancy and pretreatment serum hCG level, as two FIGO risk factors, were identified as independent parameters for predicting first-line and sequential single-agent resistance in our series. Thus, Model 2, which contains the two independent risk factors, had a predictive accuracy of 66.3%, identical to the actual outcome. In contrast, Model 1, containing four non-independent risk factors, only had an accuracy of 36.9%, which is much lower than that of Model 2. Moreover, with the risk factors adding into the models one by one, the predictive accuracies of the models were gradually increased, but still lagged far behind the intact FIGO 2000 system. The detail results were listed in Table 5 Table 5 . Given that all low-risk patients had already been deemed to have single-agent success according to the FIGO 2000 system, the sensitivity, specificity, accuracy, and predictive values were calculated with the actual outcomes. Here, the single-agent resistance was set as "positive" and single-agent success was set as "negative." hCG: human chorionic gonadotropin; FIGO: International Federation of Gynecology and Obstetrics. 2 Model 2 Risk factor 2 Number of patients 2 Positive predictive value 2 Negative predictive value 2 Sensitivity 2 Specificity 2 Accuracy Predictive positive Predictive negative Model 1 Age; Antecedent pregnancy; Largest tumor size; Number of metastases 313 126 0.131 0.960 0.891 0.308 0.369 Model 2 Interval; Pretreatment serum hCG 168 271 0.196 0.952 0.717 0.656 0.663 Model 3 Interval; Pretreatment serum hCG; Largest tumor size 168 271 0.196 0.952 0.717 0.656 0.663 Model 4 Interval; Pretreatment serum hCG; Largest tumor size; Antecedent pregnancy 174 265 0.201 0.958 0.761 0.646 0.658 Model 5 Interval; Pretreatment serum hCG; Largest tumor size; Number of metastases; Antecedent pregnancy 153 286 0.216 0.955 0.717 0.695 0.697 Model 6 Age; Interval; Pretreatment serum hCG; Largest tumor size; Number of metastases; Antecedent pregnancy 152 287 0.217 0.955 0.717 0.697 0.699 Original FIGO 2000 0 439 0 0.895 0 1.000 0.895 Then, utilizing stepwise logistic regression, we proceeded to conduct these models on all patients to investigate whether simplification of the original FIGO 2000 system for stratifying patients was feasible. Fig. 3 Fig. 3 delineated the discriminating power of these simplified models using ROC analysis when compared with the original FIGO 2000 system. As listed in Table 6 Table 6 , Model 2 was able to classify 94.8% of patients in accordance with the original FIGO criteria. In contrast, Model 1 consisting of four risk factors with inadequate statistical significance was only able to classify 91.5% of patients. With the numbers of adopted risk factors growing from two to six, the accuracies of Model 2 to Model 6 rose from 94.8% to 97.9%, which gradually approached 100%, but still fell behind the integrated FIGO 2000 system. However, the prediction was not always incorrect for the cases with inconsistent classification from the original FIGO 2000 system. For example, in Model 3, with the variables of interval from antecedent pregnancy, pretreatment serum hCG level, and the largest tumor size, nine patients were changed from low-risk to high-risk, among which four turned out to be cases of actual single-agent chemoresistance. Similarly, in Model 6, only seven patients were classified differently, and three were finally proved to be cases of single-agent chemoresistance. Due to the fact that the original FIGO 2000 system had already been used as "gold standard" for these particular data, the sensitivity, specificity, and accuracy were considered as 100%. Here, the high-risk patients were set as "positive" and the low-risk patients were set as "negative." hCG: human chorionic gonadotropin; FIGO: International Federation of Gynecology and Obstetrics. 2 Model 2 Risk factor 4 Number of patients 2 Sensitivity 2 Specificity 2 Accuracy True positive False positive False negative True negative Model 1 Age; Antecedent pregnancy; Largest tumor size; Number of metastases 90 36 13 439 0.874 0.924 0.915 Model 2 Interval; Pretreatment serum hCG 95 22 8 453 0.922 0.954 0.948 Model 3 Interval; Pretreatment serum hCG; Largest tumor size 93 9 10 466 0.903 0.981 0.967 Model 4 Interval; Pretreatment serum hCG; Largest tumor size; Antecedent pregnancy 95 13 8 462 0.922 0.973 0.964 Model 5 Interval; Pretreatment serum hCG; Largest tumor size; Number of metastases; Antecedent pregnancy 100 15 3 460 0.971 0.968 0.969 Model 6 Age; Interval; Pretreatment serum hCG; Largest tumor size; Number of metastases; Antecedent pregnancy 98 7 5 468 0.951 0.985 0.979 Original FIGO 2000 103 0 0 475 1.000 1.000 1.000 4 Discussion 1 Over the years, there have been a variety of staging and classifying systems to assist clinicians with judging prognosis and selecting treatments for GTN patients; however, no worldwide consensus has been reached on the optimal classification system ( Hancock, 2003 Hancock, 2003 ). Bagshawe (1976) Bagshawe (1976) derived a weighted scoring system with thirteen prognostic factors that influenced the response to chemotherapy, which subsequently formed the basis of the WHO prognostic scoring system. In 2000, a combination of the FIGO staging and modified WHO scoring system was proposed and later widely accepted. The main purpose of classifying patients into low- and high-risk groups is to determine at the time of diagnosis whether a given patient would respond to single-agent chemotherapy or not, so as to avoid potential undertreatment or overtreatment ( Seckl et al., 2010 Seckl et al., 2010 ). Several studies have shown that the eight prognostic factors in the currently adopted FIGO scoring system possess inequable values for predicting single-agent chemoresistance. A case-control designed study in Iran reviewed 168 low-risk GTN patients treated with methotrexate- or actinomycin-D-based therapies as initial and alternative single-agent chemotherapies, and reported a resistance rate of 43% and 19% to initial and sequential single-agent chemotherapies, respectively. An interval more than 4 months from antecedent pregnancy, tumor size larger than 3 cm, serum hCG level higher than 100 ‍000 IU/L, and presence of metastases were identified as independent risk factors for resistance to single-agent chemotherapy ( Mousavi et al., 2015 Mousavi et al., 2015 ). Another retrospective study of 358 low-risk GTN patients in USA indicated a response rate of 81% to an initial 5-d methotrexate regimen and a complete response rate of 94% to sequential single-agent chemotherapy with a 5-d actinomycin-D regimen as an alternative. Clinicopathologic diagnosis of choriocarcinoma, higher pretreatment hCG level, and presence of metastases were found to be significantly associated with resistance to initial methotrexate therapy, while only diagnosis of choriocarcinoma was strongly related to sequential single-agent chemoresistance ( Chapman-Davis et al., 2012 Chapman-Davis et al., 2012 ). In this study, we retrospectively investigated the treatments of 475 low-risk GTN patients in a Chinese population with a remission rate of 58.1% and 88.8% to first-line and sequential single-agent chemotherapies, respectively. By conducting univariate and multivariate logistic regression analyses, the pretreatment serum hCG level and interval from antecedent pregnancy were consistently identified to be significantly independent risk factors for both first-line and sequential single-agent chemoresistance. As expected, pretreatment serum hCG level emerged as the most dominant risk factor, as hCG is an irreplaceable marker in diagnosing GTN owing to the fact that its sensitivity is GTN tumor-specific. It is also an indispensable index for monitoring treatment responses, since it quantitatively reflects the tumor burden. Our results showed that patients with a higher pretreatment serum hCG level carried an obviously higher risk of resistance to both first-line and sequential single-agent therapies, which was in accordance with several previous studies ( Mcgrath et al., 2010 Mcgrath et al., 2010 ; Chapman-Davis et al., 2012 Chapman-Davis et al., 2012 ; Taylor et al., 2013 Taylor et al., 2013 ), even though the precise threshold level for predicting single-agent failure was inconclusive ( Growdon et al., 2010 Growdon et al., 2010 ; Kang et al., 2010a Kang et al., 2010a ; Winter et al., 2016 Winter et al., 2016 ; Hoeijmakers et al., 2020 Hoeijmakers et al., 2020 ; Wu et al., 2020 Wu et al., 2020 ). Another influential element was the interval from the antecedent pregnancy. We observed a remarkably elevated risk of chemo‍resistance to both first-line and secondary single-agent therapies with the growing of time span between antecedent pregnancy and disease diagnosis. As a matter of fact, a long interval usually indicates that the malignancy is more advanced. Postmolar patients with a duration of less than 6 months are more likely to have invasive mole, while those with a duration of more than 12 months are more likely to have choriocarcinoma instead. The longer the interval, the higher the likelihood of choriocarcinoma ( Le et al., 2007 Le et al., 2007 ). Some researchers have indicated that GTN patients with a histological diagnosis of choriocarcinoma had a poorer response to initial methotrexate chemotherapy and alternative single-agent chemotherapy than those with a clinicopathological diagnosis of invasive mole or postmolar GTN ( Lurain and Elfstrand, 1995 Lurain and Elfstrand, 1995 ; Osborne et al., 2011 Osborne et al., 2011 ; Braga et al., 2021 Braga et al., 2021 ). Although the aforementioned studies presented different independent risk factors for predicting single-agent chemoresistance, they concurred that some risk factors indeed influenced prognosis while others lacked significance. This may be because some factors incorporated in the FIGO scoring system are correlated with each other and are not independently prognostic ( Soper et al., 1994 Soper et al., 1994 ). For instance, tumor size is always seen as a concretization of tumor burden and the number of metastases is also a reflection of tumor progression, both of which are linked to the level of serum hCG. Moreover, the duration of tumor evolution before diagnosis was usually related to the type of antecedent pregnancy to a certain extent ( Le et al., 2007 Le et al., 2007 ). Therefore, there would be a probability to simplify the FIGO scoring system. A prior study of 813 GTN patients from the Charing Cross Hospital in London (UK) provided a simplified model with five risk factors including age, antecedent pregnancy type, interval, pretreatment hCG level, and number of metastases, which achieved an accuracy of 99.9% identical to the original FIGO 2000 system in classifying low- and high-risk GTN patients ( Eysbouts et al., 2017 Eysbouts et al., 2017 ). Another study from the Peking Union Medical Hospital (China) proposed a new prognostic system, which comprised five risk factors including antecedent pregnancy type, interval, previous failed chemotherapy, site and number of metastases with weighted score for each, and it showed a significant increase in prognostic efficiency relative to the original FIGO 2000 system ( Jiang et al., 2018 Jiang et al., 2018 ). The discrepancies between the two studies may be associated with the differences in selection criteria for patients, treatment protocols, definitions of outcomes, and statistical methodology. Herein, we attempted to reduce the number of existing factors and evaluate the performance of the simplified models. The simplified model with two independent factors, pretreatment serum hCG level and interval from the antecedent pregnancy, showed a much higher prediction efficiency than the model with the other four non-independent factors. However, the addition of other co-factors did improve the predictive accuracy, and the original FIGO 2000 prognostic system showed the highest accuracy. Furthermore, as the classification of the original FIGO 2000 prognostic system was set as the reference, the simplified model with two independent factors also expressed a much higher discriminative capacity than the model with the other four non-independent factors. Likewise, the simplified models obtained extra improvement in classification accuracy through the addition of any combination of the remaining non-independent factors. Besides, certain simplified models also ameliorated misclassifications in some cases by the original FIGO 2000 system. Thus, our results overall suggested that the independent factors have greater weight in predicting single-agent chemoresistance than other non-independent risk factors. Moreover, simplified models can preserve a comparable performance in classifying patients, but whether they can completely replace the FIGO prognostic system still needs further evaluation using sufficient patient numbers. This study has several limitations that need to be addressed. First, given the rarity of GTN and the difficulty of follow-up, it is only a single-center retrospective study with a small sample size. Second, the patient population was heterogeneous with respect to the different drug regimens, surgical interventions, and failure reasons, which might make the results unspecific or inconsistent. Last but not least, although the ROC analysis of the simplified model displayed reasonable discriminative ability, the internal validation might have potentially overestimated the performance. Therefore, more diverse and larger-scale data are needed to certify the performance of the simplified model by external validation and to determine whether this model provides better prediction of patient outcomes when compared with the currently available system. 5 Conclusions 1 In summary, our study evaluated the individual value of each prognostic factor and identified pretreatment serum hCG level and interval from antecedent pregnancy as independent risk factors for both first-line and sequential single-agent chemoresistance in low-risk GTN patients. These independent risk factors showed greater weight than other non-independent factors in predicting single-agent chemoresistance, but were unable to replace them completely. On the whole, a simplified system composed of partial factors is credible in classifying patients, but whether it can substitute the original FIGO 2000 prognostic system still needs further validation. References Abrão RA , de Andrade JM , Tiezzi DG , et al. , 2008 . Treatment for low-risk gestational trophoblastic disease: comparison of single-agent methotrexate, dactinomycin and combin ation regimens. Gynecol Oncol , 108 ( 1 ): 149 - 153 . https://doi.org/10.1016/j.ygyno.2007.09.006 https://doi.org/10.1016/j.ygyno.2007.09.006 Treatment for low-risk gestational trophoblastic disease: comparison of single-agent methotrexate, dactinomycin and combin Bagshawe KD , 1976 . Risk and prognostic factors in trophoblastic neoplasia . Cancer , 38 ( 3 ): 1373 - 1385 . https://doi.‍org/10.1002/1097-0142(197609)38:‍3 https://doi.‍org/10.1002/1097-0142(197609)38:‍3 < 1373::aid-cncr2820380342 > 3.0.co;2-e Risk and prognostic factors in trophoblastic neoplasia Braga A , Paiva G , Ghorani E , et al. , 2021 . Predictors for single-agent resistance in FIGO score 5 or 6 gestational trophoblastic neoplasia: a multicentre, retrospective, cohort study . Lancet Oncol , 22 ( 8 ): 1188 - 1198 . https://doi.org/10.1016/s1470-2045(21)00262-x https://doi.org/10.1016/s1470-2045(21)00262-x Predictors for single-agent resistance in FIGO score 5 or 6 gestational trophoblastic neoplasia: a multicentre, retrospective, cohort study Chalouhi GE , Golfier F , Soignon P , et al. , 2009 . Methotrexate for 2000 FIGO low-risk gestational trophoblastic neoplasia patients: efficacy and toxicity . Am J Obstet Gynecol , 200 ( 6 ): 643.e1 - 643.e6 . https://doi.org/10.1016/j.ajog.2009.03.011 https://doi.org/10.1016/j.ajog.2009.03.011 Methotrexate for 2000 FIGO low-risk gestational trophoblastic neoplasia patients: efficacy and toxicity Chapman-Davis E , Hoekstra AV , Rademaker AW , et al. , 2012 . Treatment of nonmetastatic and metastatic low-risk gestational trophoblastic neoplasia: factors associated with resistance to single-agent methotrexate chemotherapy . Gynecol Oncol , 125 ( 3 ): 572 - 575 . https://doi.org/10.1016/j.ygyno.2012.03.039 https://doi.org/10.1016/j.ygyno.2012.03.039 Treatment of nonmetastatic and metastatic low-risk gestational trophoblastic neoplasia: factors associated with resistance to single-agent methotrexate chemotherapy Cyriac S , Rajendranath R , Sridevi V , et al. , 2011 . Etoposide, cisplatin-etoposide, methotrexate, actinomycin-D as primary treatment for management of very-high-risk gestational trophoblastic neoplasia . Int J Gynecol Obstet , 115 ( 1 ): 37 - 39 . https://doi.org/10.1016/j.ijgo.2011.04.017 https://doi.org/10.1016/j.ijgo.2011.04.017 Etoposide, cisplatin-etoposide, methotrexate, actinomycin-D as primary treatment for management of very-high-risk gestational trophoblastic neoplasia Eysbouts YK , Ottevanger PB , Massuger LFAG , et al. , 2017 . Can the FIGO 2000 scoring system for gestational trophoblastic neoplasia be simplified? A new retrospective analysis from a nationwide dataset. Ann Oncol , 28 ( 8 ):‍ 1856 - 1861 . https://doi.org/10.1093/annonc/mdx211 https://doi.org/10.1093/annonc/mdx211 Can the FIGO 2000 scoring system for gestational trophoblastic neoplasia be simplified? FIGO Oncology Committee , 2002 . FIGO staging for gestational trophoblastic neoplasia 2000 . Int J Gynaecol Obstet , 77 ( 3 ): 285 - 287 . https://doi.org/10.1016/S0020-7292(02)00063-2 https://doi.org/10.1016/S0020-7292(02)00063-2 FIGO staging for gestational trophoblastic neoplasia 2000 Gleeson NC , Finan MA , Fiorica JV , et al. , 1993 . Nonmetastatic gestational trophoblastic disease . Weekly methotrexate compared with 8-day methotrexate-folinic acid. Eur J Gynaecol Oncol , 14 ( 6 ): 461 - 465 . Nonmetastatic gestational trophoblastic disease Growdon WB , Wolfberg AJ , Goldstein DP , et al. , 2010 . Low-risk gestational trophoblastic neoplasia and methotrexate resistance: predictors of response to treatment with actinomycin D and need for combination chemotherapy . J Reprod Med , 55 ( 7-8 ): 279 - 284 . Low-risk gestational trophoblastic neoplasia and methotrexate resistance: predictors of response to treatment with actinomycin D and need for combination chemotherapy Hancock BW , 2003 . Staging and classification of gestational trophoblastic disease . Best Pract Res Clin Obstet Gynaecol , 17 ( 6 ): 869 - 883 . https://doi.org/10.1016/s1521-6934(03)00073-7 https://doi.org/10.1016/s1521-6934(03)00073-7 Staging and classification of gestational trophoblastic disease Hoeijmakers YM , Sweep FCGJ , Lok CAR , et al. , 2020 . Risk factors for second-line dactinomycin failure after methotrexate treatment for low-risk gestational trophoblastic neoplasia: a retrospective study . BJOG , 127 ( 9 ): 1139 - 1145 . https://doi.org/10.1111/1471-0528.16198 https://doi.org/10.1111/1471-0528.16198 Risk factors for second-line dactinomycin failure after methotrexate treatment for low-risk gestational trophoblastic neoplasia: a retrospective study Jiang F , Wan XR , Xu T , et al. , 2018 . Evaluation and suggestions for improving the FIGO 2000 staging criteria for gestational trophoblastic neoplasia: a ten-year review of 1420 patients . Gynecol Oncol , 149 ( 3 ): 539 - 544 . https://doi.org/10.1016/j.ygyno.2018.04.001 https://doi.org/10.1016/j.ygyno.2018.04.001 Evaluation and suggestions for improving the FIGO 2000 staging criteria for gestational trophoblastic neoplasia: a ten-year review of 1420 patients Kang WD , Kim CH , Cho MK , et al. , 2010a . Serum hCG level and rising world health organization score at second-line chemotherapy (pulse dactinomycin): poor prognostic factors for methotrexate-failed low-risk gestational trophoblastic neoplasia . Int J Gynecol Cancer , 20 ( 8 ): 1424 - 1428 . https://doi.org/10.1111/IGC.0b013e3181f5873e https://doi.org/10.1111/IGC.0b013e3181f5873e Serum hCG level and rising world health organization score at second-line chemotherapy (pulse dactinomycin): poor prognostic factors for methotrexate-failed low-risk gestational trophoblastic neoplasia Kang WD , Choi HS , Kim SM , 2010b . Weekly methotrexate (50 mg/m 2 ) without dose escalation as a primary regimen for low-risk gestational trophoblastic neoplasia . Gynecol Oncol , 117 ( 3 ): 477 - 480 . https://doi.org/10.1016/j.ygyno.2010.02.029 https://doi.org/10.1016/j.ygyno.2010.02.029 Weekly methotrexate (50 mg/m2) without dose escalation as a primary regimen for low-risk gestational trophoblastic neoplasia Kohorn EI , 2002 . Negotiating a staging and risk factor scoring system for gestational trophoblastic neoplasia . A progress report. J Reprod Med , 47 ( 6 ): 445 - 450 . Negotiating a staging and risk factor scoring system for gestational trophoblastic neoplasia Le J , Xie X , Lin ZQ , et al. , 2007 . Gynaecology and Obstetrics , 7nd Ed . People’s Medical Publishing House , Beijing , China (in Chinese) . Gynaecology and Obstetrics Lertkhachonsuk AA , Israngura N , Wilailak S , et al. , 2009 . Actinomycin D versus methotrexate-folinic acid as the treatment of stage I, low-risk gestational trophoblastic neoplasia: a randomized controlled trial . Int J Gynecol Cancer , 19 ( 5 ): 985 - 988 . https://doi.org/10.1111/IGC.0b013e3181a8333d https://doi.org/10.1111/IGC.0b013e3181a8333d Actinomycin D versus methotrexate-folinic acid as the treatment of stage I, low-risk gestational trophoblastic neoplasia: a randomized controlled trial Lurain JR , 2003 . Pharmacotherapy of gestational trophoblastic disease . Expert Opin Pharmaco , 4 ( 11 ): 2005 - 2017 . https://doi.org/10.1517/14656566.4.11.2005 https://doi.org/10.1517/14656566.4.11.2005 Pharmacotherapy of gestational trophoblastic disease Lurain JR , Elfstrand EP , 1995 . Single-agent methotrexate chemo therapy for the treatment of nonmetastatic gestational trophoblastic tumors. Am J Obstet Gynecol , 172 ( 2 Pt 1 ): 574 - 579 . https://doi.org/10.1016/0002-9378(95)90575-8 https://doi.org/10.1016/0002-9378(95)90575-8 Single-agent methotrexate chemo Lurain JR , Chapman-Davis E , Hoekstra AV , et al. , 2012 . Actinomycin D for methotrexate-failed low-risk gestational trophoblastic neoplasia . J Reprod Med , 57 ( 7-8 ): 283 - 287 . Actinomycin D for methotrexate-failed low-risk gestational trophoblastic neoplasia Mcgrath S , Short D , Harvey R , et al. , 2010 . The management and outcome of women with post-hydatidiform mole ‘low-risk’ gestational trophoblastic neoplasia, but hCG levels in excess of 100 000 IU l -1 . Br J Cancer , 102 ( 5 ): 810 - 814 . https://doi.org/10.1038/sj.bjc.6605529 https://doi.org/10.1038/sj.bjc.6605529 The management and outcome of women with post-hydatidiform mole ‘low-risk’ gestational trophoblastic neoplasia, but hCG levels in excess of 100 000 IU l-1 Miller AB , Hoogstraten B , Staquet M , et al. , 1981 . Reporting results of cancer treatment . Cancer , 47 ( 1 ): 207 - 214 . https://doi.org/10.1002/1097-0142(19810101)47:1<207::aid-cncr2820470134>3.0.co;2-6 https://doi.org/10.1002/1097-0142(19810101)47:1<207::aid-cncr2820470134>3.0.co;2-6 Reporting results of cancer treatment Mora PAR , Sun SY , Velarde GC , et al. , 2019 . Can carboplatin or etoposide replace actinomycin-D for second-line treatment of methotrexate resistant low-risk gestational trophoblastic neoplasia? Gynecol Oncol , 153 ( 2 ): 277 - 285 . https://doi.org/10.1016/j.ygyno.2019.03.005 https://doi.org/10.1016/j.ygyno.2019.03.005 Can carboplatin or etoposide replace actinomycin-D for second-line treatment of methotrexate resistant low-risk gestational trophoblastic neoplasia? Mousavi AS , Zamani A , Khorasanizadeh F , et al. , 2015 . Resistance to single-agent chemotherapy and its risk factors in low-risk gestational trophoblastic neoplasms . J Obstet Gynaecol Res , 41 ( 5 ): 776 - 783 . https://doi.org/10.1111/jog.12613 https://doi.org/10.1111/jog.12613 Resistance to single-agent chemotherapy and its risk factors in low-risk gestational trophoblastic neoplasms Ngan HYS , Seckl MJ , Berkowitz RS , et al. , 2018 . Update on the diagnosis and management of gestational trophoblastic disease . Int J Gynaecol Obstet , 143 ( Suppl 2 ): 79 - 85 . https://doi.org/10.1002/ijgo.12615 https://doi.org/10.1002/ijgo.12615 Update on the diagnosis and management of gestational trophoblastic disease Osborne RJ , Filiaci V , Schink JC , et al. , 2011 . Phase III trial of weekly methotrexate or pulsed dactinomycin for low-risk gestational trophoblastic neoplasia: a gynecologic oncology group study . J Clin Oncol , 29 ( 7 ): 825 - 831 . https://doi.org/10.1200/JCO.2010.30.4386 https://doi.org/10.1200/JCO.2010.30.4386 Phase III trial of weekly methotrexate or pulsed dactinomycin for low-risk gestational trophoblastic neoplasia: a gynecologic oncology group study Prouvot C , Golfier F , Massardier J , et al. , 2018 . Efficacy and safety of second-line 5-day dactinomycin in case of metho trexate failure for gestational trophoblastic neoplasia. Int J Gynecol Cancer , 28 ( 5 ): 1038 - 1044 . https://doi.org/10.1097/IGC.0000000000001248 https://doi.org/10.1097/IGC.0000000000001248 Efficacy and safety of second-line 5-day dactinomycin in case of metho Seckl MJ , Sebire NJ , Berkowitz RS , 2010 . Gestational trophoblastic disease . Lancet , 376 ( 9742 ): 717 - 729 . https://doi.org/10.1016/S0140-6736(10)60280-2 https://doi.org/10.1016/S0140-6736(10)60280-2 Gestational trophoblastic disease Smith EB , Weed JC , Tyrey L , et al. , 1982 . Treatment of nonmetastatic gestational trophoblastic disease: results of methotrexate alone versus methotrexate-folinic acid . Am J Obstet Gynecol , 144 ( 1 ): 88 - 92 . https://doi.org/10.1016/0002-9378(82)90400-8 https://doi.org/10.1016/0002-9378(82)90400-8 Treatment of nonmetastatic gestational trophoblastic disease: results of methotrexate alone versus methotrexate-folinic acid Song HZ , Yang XY , Xiang Y , 2004 . Trophoblastic Tumor Diagnosis & Treatment , 2nd Ed . People’s Medical Publishing House , Beijing , China (in Chinese) . Trophoblastic Tumor Diagnosis & Treatment Soper JT , Evans AC , Conaway MR , et al. , 1994 . Evaluation of prognostic factors and staging in gestational trophoblastic tumor . Obstet Gynecol , 84 ( 6 ): 969 - 973 . Evaluation of prognostic factors and staging in gestational trophoblastic tumor Sung HC , Wu PC , Yang HY , 1984 . Reevaluation of 5-fluorouracil as a single therapeutic agent for gestational trophoblastic neoplasms . Am J Obstet Gynecol , 150 ( 1 ): 69 - 75 . https://doi.org/10.1016/S0002-9378(84)80112-X https://doi.org/10.1016/S0002-9378(84)80112-X Reevaluation of 5-fluorouracil as a single therapeutic agent for gestational trophoblastic neoplasms Taylor F , Grew T , Everard J , et al. , 2013 . The outcome of patients with low risk gestational trophoblastic neoplasia treated with single agent intramuscular methotrexate and oral folinic acid . Eur J Cancer , 49 ( 15 ): 3184 - 3190 . https://doi.org/10.1016/j.ejca.2013.06.004 https://doi.org/10.1016/j.ejca.2013.06.004 The outcome of patients with low risk gestational trophoblastic neoplasia treated with single agent intramuscular methotrexate and oral folinic acid Uberti EMH , Fajardo MDC , da Cunha AGV , et al. , 2015 . Treatment of low-risk gestational trophoblastic neoplasia comparing biweekly eight-day methotrexate with folinic acid versus bolus-dose Actinomycin-D, among Brazilian women . Rev Bras Ginecol Obstet , 37 ( 6 ): 258 - 265 . https://doi.org/10.1590/SO100-720320150005366 https://doi.org/10.1590/SO100-720320150005366 Treatment of low-risk gestational trophoblastic neoplasia comparing biweekly eight-day methotrexate with folinic acid versus bolus-dose Actinomycin-D, among Brazilian women Winter MC , Tidy JA , Hills A , et al. , 2016 . Risk adapted single-agent dactinomycin or carboplatin for second-line treatment of methotrexate resistant low-risk gestational trophoblastic neoplasia . Gynecol Oncol , 143 ( 3 ): 565 - 570 . https://doi.org/10.1016/j.ygyno.2016.10.001 https://doi.org/10.1016/j.ygyno.2016.10.001 Risk adapted single-agent dactinomycin or carboplatin for second-line treatment of methotrexate resistant low-risk gestational trophoblastic neoplasia Wong LC , Choo YC , Ma HK , 1985 . Methotrexate with citro vorum factor rescue in gestational trophoblastic disease. Am J Obstet Gynecol , 152 ( 1 ): 59 - 62 . https://doi.org/10.1016/S0002-9378(85)80178-2 https://doi.org/10.1016/S0002-9378(85)80178-2 Methotrexate with citro Wu XD , Qin JL , Shen T , et al. , 2020 . The 16-year experience in treating low-risk gestational trophoblastic neoplasia patients with failed primary methotrexate chemotherapy . J Gynecol Oncol , 31 ( 4 ): e36 . https://doi.org/10.3802/jgo.2020.31.e36 https://doi.org/10.3802/jgo.2020.31.e36 The 16-year experience in treating low-risk gestational trophoblastic neoplasia patients with failed primary methotrexate chemotherapy Yang WENG and Yuanyuan LIU contributed to data interpretation, statistical analysis, and writing the manuscript. Yuanyuan LIU and Xiaodong WU contributed to data extraction and draft preparation. Chitapa BENJOED contributed to revising the manuscript. Yang WENG and Sangsang TANG contributed to statistical analysis. Xiao LI contributed to data interpretation and supervision. Xing XIE and Weiguo LU contributed to conception, reviewing and editing the final manuscript. All authors have read and approved the final manuscript, and therefore, have full access to all the data in the study and take responsibility for the integrity and security of the data. 10.1631/jzus.B2100895 1673-1581（2022）03-0218-12 2021-10-24 2021-12-15 2022-03-15 2022-04-19 2022-02-15 Copyright © Zhejiang University and Springer-Verlag GmbH Germany, part of Springer Nature 2022 2022 浙江大学出版社 Yang WENG, Yuanyuan LIU, Chitapa BENJOED, Xiaodong WU, Sangsang TANG, Xiao LI, Xing XIE, and Weiguo LU declare that they have no conflict of interest. All procedures followed were in accordance with the ethical standards of the Ethics Committee of Women's Hospital, School of Medicine, Zhejiang University (approval number: IRB-20190003-R) and with the Helsinki Declaration of 1975, as revised in 2008 (5). Informed consent was waived due to its nature of retrospective analysis. Compliance with ethics guidelines 浙江大学学报（英文版）（B辑：生物医学和生物技术） 3 23