Treatments and outcomes in high-risk gestational trophoblastic neoplasia: a systematic review and meta-analysis

Abstract

Background:

High-risk gestational trophoblastic neoplasia (GTN) is rare and treated with diverse approaches. Limited published institutional data has yet to be systematically reviewed.

Objectives:

To compile global high-risk GTN (prognostic score ≥7) cohorts to summarize treatments and outcomes by disease characteristics and primary chemotherapy.

Search Strategy:

MEDLINE, Embase, Scopus, ClinicalTrials.gov, and Cochrane were searched through March 2021.

Selection Criteria:

Full-text manuscripts reporting mortality among ≥10 high-risk GTN patients.

Data Collection and Analysis:

Binomial proportions were summed, and random-effects meta-analyses performed.

Main Results:

From 1,137 records, we included 35 studies, representing 20 countries. Among 2,276 unique high-risk GTN patients, 99.7% received chemotherapy, 35.8% surgery, and 4.9% radiation. Mortality was 10.9% (243/2,236; meta-analysis: 10%, 95%CI 7-12%), and likelihood of complete response to primary chemotherapy was 79.7% (1,506/1,890; meta-analysis: 78%, 95%CI 74-83%). Across 24 reporting studies, modern preferred chemotherapy (EMA/CO or EMA/EP) was associated with lower mortality (overall: 8.8% vs. 9.5%; comparative meta-analysis: 8.1% vs. 12.4%, OR=0.42, 95%CI 0.20-0.90, 14 studies) and higher likelihood of complete response (overall: 76.6% vs. 72.8%; comparative meta-analysis: 75.9% vs. 60.7%, OR=2.98, 95%CI 1.06-8.35, 14 studies), though studies focused on non-preferred regimens reported comparable outcomes. Mortality was increased for ultra-high-risk disease (30% vs. 7.5% high-risk; meta-analysis OR=7.44; 95%CI 4.29-12.9) and disease following term delivery (20.8% versus 7.3% following molar pregnancy; meta-analysis OR=2.64; 95%CI 1.10-6.31). Relapse rate estimates ranged 3-6%.

Conclusions:

High-risk GTN is responsive to several chemotherapy regimens, with EMA/CO or EMA/EP associated with improved outcomes. Mortality is increased in patients with ultra-high-risk, relapsed, and post-term pregnancy disease.

Funding:

This study received no direct funding.

Introduction

Gestational trophoblastic neoplasia (GTN) is defined by malignant proliferations of placental trophoblastic cells.¹ GTN is diagnosed most commonly following molar pregnancy, by abnormal trends in human chorionic gonadotropin (hCG) levels during routine screening following uterine evacuation.² Less commonly, gestational choriocarcinomas sometimes present months to years following spontaneous abortions or normal pregnancy.^1,3 The exquisite chemo-sensitivity of these diseases made them foundational in the development of the first chemotherapeutics for solid malignancies in the 1950s.⁴ Methotrexate remains a mainstay of GTN treatment today, leading to cures even in some cases of widespread metastatic disease.^3,5

However, single-agent therapy does not offer a universal cure for GTN. Research from leading referral centers identified high-risk features associated with chemoresistance and mortality,^6–8 which were compiled into scoring systems for standardized risk assessment. The first was published by the World Health Organization (WHO) in 1983,⁹ and subsequently other scores were developed.¹⁰ General consensus was achieved with the publication of guidelines by the International Federation of Gynecology and Obstetrics (FIGO) in 2000.¹¹ These guidelines instituted a new staging and scoring system (the modified WHO Prognostic Scoring System as adopted by FIGO), recently re-affirmed in 2021.¹²

High-risk disease (prognostic score ≥7) is typically treated with multi-agent chemotherapy, with some growing consensus on the preferred first-line regimen (EMA/CO or EMA/EP; etoposide, methotrexate, and actinomycin alternating with either cyclophosphamide and vincristine, or etoposide and cisplatin).^5,13–15 However, due to the rarity of high-risk GTN, variation in treatment approach remains across different centers around the globe. Randomized comparative studies are lacking,¹⁶ and published institutional cohorts to date have yet to be systematically reviewed.

The objectives of our study were to compile global reports of high-risk GTN to (1) summarize treatments and outcomes since publication of the FIGO 2000 prognostic scoring guidelines, and (2) compare outcomes by disease characteristics and primary chemotherapy, particularly to assess for benefit of modern preferred chemotherapy with EMA/CO or EMA/EP.

Methods

Study Selection

We followed standard methodology for systematic review and meta-analysis, according to Cochrane, MOOSE (meta-analysis of Observational Studies in Epidemiology) and PRISMA (Preferred Reporting In Systematic review and Meta-Analysis) guidelines.^17–19 A biomedical librarian (SJK) led the development of the search strategy in coordination with the multiple content experts on the authorship team. The MEDLINE (via PubMed), Embase, Scopus, ClinicalTrials.gov, and Cochrane library databases were queried from inception to March 4^th, 2021. We used a combination of Medical Subject Heading (MeSH) terms and keywords to populate sets related to the themes of ‘gestational trophoblastic neoplasia’ and ‘high-risk,’ then the Boolean term “AND” was applied to identify the intersection in search results. Limits or restrictions on dates and language were not applied during the search process. Due to the breadth of the initial search and high number of eligible papers identified, additional hand searching of reference lists was not performed. The general search strategy and search details and outputs for each database are detailed in Appendix S1. The study protocol was prospectively submitted for registration on the PROSPERO international registry (CRD42021254677).

For inclusion, we required studies to meet the following eligibility criteria: (1) the study was published as a English language full-length manuscript, (2) the study was a retrospective or prospective trial, cohort, or case series (study type), (3) the study included cases of high-risk GTN as determined by the modified WHO Prognostic Scoring System as adopted by FIGO published in 2000 (referred to henceforth as ‘prognostic score’) and not limited to a particular GTN histology, surgical treatment, or specific site of metastatic disease (study population), (4) the study reported on overall mortality among the high-risk GTN cases (study outcome), and (5) the study reported data on ≥10 patients with high-risk GTN, excluding placental site trophoblastic and epithelioid trophoblastic tumors. In instances of multiple studies from the same institution with suspected overlap in represented patients, we selected the study with the most complete data reporting for inclusion.

Study selection was performed within Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia). Search results were first purged of duplicates. Studies titles and abstracts were then screened in duplicate by two independent reviewers (one of BBA or BAD, and one of HEK, TE, ECG, or KAM). Disagreements were resolved via consultation with a third independent reviewer (HAM, RAP, or LJH). Every search result was screened by a gynecologic oncology attending or fellow. Full-text manuscripts were accessed, then reviewed by two independent reviewers (BBA and BAD), with disagreements resolved by consensus discussion with two additional reviewers (RAP, LJH). Given the geographic diversity of the GTN literature and the age of some of the data for included cohorts, we did not contact study authors to obtain unpublished data.

Data extraction and quality assessment

Data were then extracted onto a standardized form by the primary reviewer (BBA) using Microsoft Excel v14.7.7 (Microsoft Corporation, Redmond, WA, USA), with any uncertainties resolved by consensus discussion (HAM, RAP, LJH, BAD). We designed the form based on goals of the study, seeking to maximize what could be summarized and what comparisons could be made, recognizing that every study would not report on every data point.

Data extraction included descriptive data on study characteristics, including the country, city, and facility of origin, years of data, and any cohort restrictions. We next collected descriptive data on patient characteristics including antecedent pregnancy (molar pregnancy, term delivery, abortion/ectopic/unknown), prognostic score (high-risk score 7-11 or 7-12 vs. ultra-high-risk score ≥12 or ≥13), FIGO stage, and metastatic sites (lung, vagina, brain, liver, or other), extracting data specific to high-risk GTN cases whenever reported. We then extracted treatment characteristics for high-risk GTN cases including prior failed single-agent treatment of low-risk disease with subsequent high-risk re-scoring, primary chemotherapy for high-risk disease, and utilization of surgery (excluding dilation and curettage) and radiation therapy.

Lastly, we extracted data on patient outcomes specific to cases of high-risk GTN. Our primary outcome was overall mortality (deaths / cases high-risk GTN), which was required to be reported for study inclusion. We recorded additional outcome data specific to high-risk cases when reported, including overall complete response rate to primary chemotherapy (complete responses / cases high-risk GTN) and relapse risk (relapses after complete response / total complete responses). Mortality and complete responses specific to primary chemotherapy regimen were also extracted when reported. We further described deaths by multiple characteristics when reported, including early (≤1 month from presentation), ultra-high-risk disease, stage IV disease, antecedent pregnancy, and following relapse.

As our outcomes of mortality and complete response (study definition, typically hCG normalization sustained for ≥3 weeks) were objective, assessment would only be biased by limitations to patient follow up. Due to the observational nature of the included studies and the objectiveness of outcomes, existing quality scoring mechanisms were not considered useful and, therefore, not applied. Instead, we summarized the extent a of follow-up, and the completeness of data reporting on key elements by each study.

Data Synthesis

We first summarized treatments and outcomes for high-risk GTN by summing binomial proportions across reporting studies. We then performed meta-analyses overall across all studies and in subgroups by study characteristics, utilizing random effects modeling to minimize the risk of type 1 error and account for the expected heterogeneity in the underlying studies due to significant variation in underlying populations and utilized treatments from studies compiled across numerous countries from around the world.¹⁹ Heterogeneity between studies was assessed with Higgins I². Publication bias was assessed graphically using funnel plots.

We first ran meta-analyses of binomial proportions of utilized treatments and disease outcome. We primarily utilized the score (Wilson) confidence intervals,²⁰ which were recommended by a statistical comparison of methods of confidence interval estimation for binomial proportions.²¹ We utilized the Freeman-Tukey transformation to allow inclusion of studies with no recorded deaths.^22,23 We also trialed exact (Clopper-Pearson) confidence intervals as a sensitivity analyses,²⁴ but found no meaningful difference in estimated confidence intervals (results not presented).

We then performed comparative random-effects meta-analyses of mortality risk and complete response rate by treatments and disease characteristics. We assessed both outcomes comparing modern preferred chemotherapy (EMA/CO or EMA/EP)^5,13–15 to other treatments. We also compared mortality risk by disease characteristics of ultra-high-risk disease (versus high-risk), stage IV disease (versus stages I-III), and antecedent term pregnancy (versus molar pregnancy).

We assessed robustness of results with sensitivity analyses limiting to studies with high quality of data reporting. We first performed analyses limited to studies with <10% of patients lost to follow up, then limiting to studies reporting median or mean length of follow up of at least 18 months, then a combination of the two criteria.

Data synthesis, analysis, and figure generation was conducted with STATA v17.0 (Statacorp, College Station, TX, USA), using the imported ‘metaprop’ package for summaries of binomial proportions,²² and the internal ‘meta esize’ package for two-group comparisons of binary outcomes.

Results

After combining search results from all databases, 2,153 total records were identified, with about half excluded as duplicates (n=1,016). Among the 1,137 remaining results, 936 were deemed ineligible by title and abstract review, and an additional 140 were found ineligible upon full text review. A total of 60 studies met all inclusion criteria, 25 of which were excluded for suspected overlapping patient data, reported from the same or overlapping institutions in overlapping time periods (Table S1). Ultimately, 35 studies were included.^25–59 See Figure S1 for PRISMA flow diagram of study selection.

The 35 included studies (Tables S2–S3) reported on 2,276 unique patients with high-risk GTN, representing 20 countries spanning 5 continents. India (7 studies) and China (5 studies) were the most frequently represented countries by study count. Most studies included ≥10 years of compiled data from one or several centers, though only four reported on any patients prior to 1990, and just two included no patients from the last 15 years. Some studies included patients treated prior to publication of the modern prognostic scoring system in 2000, but all patients were re-scored by these guidelines.

Among the 35 included studies, 17 included patients with low-risk disease, with data specific to the high-risk cohort extracted for meta-analysis. Among the 18 reporting only on high-risk patients, four were limited to cases with metastatic disease (excluding high-risk stage I disease), and one was limited to ultra-high-risk disease. A total of 10 studies reported on patients receiving a specific chemotherapy regimen (3 EMA/CO, 2 EMA/EP, and 5 other), either in the primary or salvage setting.

The characteristics of included patients are summarized in Table S3, Table S4, Table S5, and Figure S2. The lung and pelvis/vagina were the most common sites of metastatic disease spread at presentation reported across included studies. Nearly all patients received chemotherapy (99.7%), with 19 studies reporting ≥80% patients received one of the modern preferred regimens, EMA/CO or EMA/EP, while 5 studies included no patients treated on one of these regimens. Utilization of surgery in treatment of high-risk GTN varied greatly. While 8 studies reported surgery in ≥50% of patients, and another 8 studies reported surgical utilization <10%, with overall utilization in 35.8% of patients. Radiation was less commonly utilized, with only 8 studies reporting use in >10% of patients, and an overall utilization rate of 4.9%.

The quality and completeness of data reporting by included studies is summarized in Table S6. Only 10 studies reported any patients lost to follow-up, with just 4 reporting ≥10% of patients lost, and 40 of 2,276 patients lost to follow-up in total (1.8%). While 12 studies failed to report on follow up time, just 3 of 23 reporting studies reported mean/median follow up <18 months. Primary chemotherapy and associated complete response (CR) rates were reported in 33 and 31 of 35 studies, respectively, with a standard CR definition of hCG normalization for ≥3 weeks specified in 28 studies.

Mortality results are summarized in Table 1. Overall, 243 of 2,236 high-risk GTN patients died, for mortality of 10.9%. The meta-analysis estimate of mortality risk was 10% (95%CI 7-12%; I²=68%; Figure 1). Among 24 studies specifying timing of deaths, 37 of 102 deaths (36.3%) occurred ≤1 month from presentation. Our secondary outcome of CR to primary chemotherapy was observed in 1,506 of 1,890 patients (79.7%) across 31 reporting studies, with meta-analysis estimate of 78% (95%CI 74-83%; I²=80%; Figure 2).

Table 1.

Summary of compiled outcomes across included studies for high-risk gestational trophoblastic neoplasia.

	Reporting Studies	Total N	N with outcome	Percent with outcome	Meta-analysis* Estimate (95%CI)
Complete Response
Overall	31	1890	1506	79.7%	78% (73-83%)
Preferred chemotherapy	24	806	617	76.6%	78% (72-84%)
Non-preferred treatment	20	599	436	72.8%	60% (43-76%)
Relapse after complete response	27	901	37	4.1%	3% (1-5%)
Mortality
Overall	35	2236	243	10.9%	10% (7-12%)
Preferred chemotherapy	24	897	79	8.8%	7% (4-10%)
EMA/CO	22	834	74	8.9%	7% (4-11%)
EMA/EP	5	63	5	7.9%	2% (0-13%)
Non-preferred treatment	20	558	53	9.5%	8% (2-17%)
Early death (<1 month)	24	37	1253	3.0%	1% (0-3%)
High-risk (excluding ultra-high-risk)	9	810	61	7.5%	3% (0-7%)
Ultra-high-risk	9	210	63	30.0%	25% (15-36%)
Stage IV	7	51	16	31.3%	28% (14-43%)
After molar pregnancy	5	137	10	7.3%	4% (1-9%)
After term delivery	5	77	16	20.8%	14% (3-28%)
Prior single-agent chemotherapy for low-risk disease	10	96	9	9.4%	7% (1-14%)
Death after relapse	14	37	10	27.0%	18% (3-37%)

Preferred chemotherapy = EMA/CO (etoposide, methotrexate, and actinomycin D, alternating with cyclophosphamide, and vincristine) or EMA/EP (etoposide, methotrexate, and actinomycin D, alternating with etoposide and cisplatin); score refers to modified WHO Prognostic Scoring System as adopted by FIGO, first published in 2000 FIGO guidelines

^*Random-effects analysis using ‘metaprop’ package in Stata v17.0

Figure 1.

Forest plot for meta-analysis of overall mortality from high-risk gestational trophoblastic neoplasia.

Figure 2.

Forest plot for meta-analysis of complete response to primary multi-agent chemotherapy from high-risk gestational trophoblastic neoplasia.

Use of modern preferred EMA/CO or EMA/EP chemotherapy was reported in 897 patients across 24 studies (93.0% EMA/CO) with 79 deaths (8.8%; meta-analysis estimate 7%, 95%CI 4-10%, I²=47%). Mortality among those receiving non-preferred treatments was 9.5% (53/558 patients; meta-analysis estimate 8%, 95%CI 2-17%, I²=77%). Overall CR rate was 76.6% with EMA/CO or EMA/EP (617/806 patients) versus 72.8% for patients receiving other treatments (436/599).

Across 14 studies reporting ≥1 patient with both preferred and non-preferred treatments, comparative meta-analysis showed EMA/CO or EMA/EP chemotherapy was associated with significantly lower mortality (8.1% vs 12.4%; OR=0.42; 95%CI 0.20-0.90; p=0.03; I²=25%; Figure 3A), and significantly higher likelihood of CR (75.9% vs 60.7%; OR=2.98; 95%CI 1.06-8.35; p=0.04; I²=65%; Figure 3B). When limiting to studies with >1 patient in each arm (Figure S3), preferred regimens had statistically lower mortality but improvement in CR rate did not retain statistical significance. Among 9 studies reporting primarily on non-preferred chemotherapy, outcomes were comparable (85% and 5% respectively, Table S7).

Figure 3.

Forest plot for comparative meta-analyses of (A) overall mortality, and (B) complete response (CR) rate with preferred (EMA/CO or EMA/EP) vs. other chemotherapy in high-risk gestational trophoblastic neoplasia, among the 14 included studies reporting on at least one patient receiving ‘preferred’ EMA/CO or EMA/EP and one patient receiving an alternative ‘other’ regimen. *EMA/EP used as first-line in some or all patients in the indicated study vs. EMA/CO; see Table S3 for details of chemotherapy utilization.

We found considerably higher mortality among patients with ultra-high-risk disease, with 63 of 210 patients reported dead (30%) across 9 reporting studies (meta-analysis estimate 25%; 95%CI 15-36%; I²=47%; Figure S4). Estimated mortality was higher in four studies using a threshold of prognostic score ≥13 (35%; 95%CI 15-45%). In comparative meta-analysis, we estimated a statistically significant difference (32.7% vs 7.5% for ultra-high-risk vs high-risk; OR=7.44; 95%CI 4.29-12.89; p<0.01; I²=14%). Risk of death for stage IV disease was similarly high, with 16 of 51 patients reported dead (31.3%) across 7 studies specifying mortality by stage. We also found high mortality for disease following term delivery, with 16 of 77 patients dying (20.8%) across 5 reporting studies, versus 10 of 137 patients (7.3%) following molar pregnancy (OR=2.64; 95%CI 1.10-6.31; p=0.03; I²=0%). Across 10 studies specifying prior single-agent chemotherapy for initially low-risk disease, we found similar mortality (9.4%, 9/96 patients) to the overall risk.

Across 27 reporting studies, we identified 37 cases of relapse among 901 complete responders, for 4.1% overall risk. Meta-analysis of this outcome estimated 3% risk of relapse (95%CI 1-6%; I²=74%; Figure S5). Among the 14 studies reporting deaths specific to relapse, we estimated 27% risk of death after relapse (10 of 37 patients) and relapse deaths accounted for 10 of 94 total deaths (11%).

Sensitivity analyses were performed limiting to highest quality studies: those with <10% loss to follow-up (31 studies), ≥18 months mean/median length of follow-up (19 studies), and a combination of both criteria (18 studies). Meta-analysis estimates of overall mortality (10%, 8%, and 9%, respectively) and CR rate were (79%, 82%, and 81%, respectively) did not differ significantly from our original results. For comparative meta-analyses of highest quality studies, estimated mortality was significantly lower with EMA/CO or EMA/EP chemotherapy (OR=0.22; 95%CI 0.07-0.68), and no significant difference was found in CR rate, though only 3 studies were included in these analyses. Estimated relapse rate of 3% did not change when limiting to studies with standard definition of CR, though increased to 6% when further limiting to highest quality studies on follow-up.

Our assessment for publication bias with funnel plots can be seen in Figure S6. While we saw no clear evidence of publication bias in our primary outcome of overall mortality, asymmetry in the funnel plot for complete response rate indicates that smaller comparative series favoring chemotherapy regimens besides EMA/CO or EMA/EP may be under-represented in the published literature for that outcome.

Discussion

Main Findings

This study presents a global perspective on treatment and outcomes of high-risk GTN according to the most recent definition of high-risk disease.¹¹ Systematic review and meta-analysis are important tools for the study of rare diseases and rare outcomes.⁶⁰ The only prior systematic review of high-risk GTN, published in 2013,¹⁶ sought randomized trials, formally including only one 1989 study.⁶¹ The authors did summarize some of the existing retrospective data, but without any formal meta-analysis, and including only 6 studies that overlap with data presented here.

We found an overall mortality risk from high-risk GTN of 10%. Mortality is strongly related to the extent and risk of disease at presentation, with considerably higher mortality rates (around 30%) for patients with ultra-high-risk or stage IV disease. Additionally, antecedent term pregnancy, which is included in the prognostic scoring system, was specifically associated with increased mortality risk. We found complete response rate of 80% to primary treatment, and low relapse risk of <5% after complete response. Although high-risk GTN typically presents with widespread metastases, it remains highly treatable and curable in most cases. Importantly, given the lack of prospective comparative trials, we found that primary treatment with modern preferred EMA/CO or EMA/EP chemotherapy was associated with statistically significant improvements in both mortality and complete response rate in comparative analysis.

Strengths and Limitations

The strengths of this study begin with our broad search, with screening of over 1,100 unique results, and inclusion of 35 non-overlapping studies, representing 20 countries across 5 continents. We used a rigorous systematic review methodology, including selection of studies independently in duplicate.¹⁹ Our study had an objective primary outcome, and generally complete data reporting for adequate assessment of our main questions, and our primary outcome lacked evidence for any publication bias in assessment via funnel plot.

The weakness of our study is primarily related to the rarity of high-risk GTN. All identified studies were retrospective and observational, with variation in the length and completeness of follow up, and selection bias in chemotherapy treatment may have influenced observed differences in outcomes. Our comparative analysis by primary chemotherapy was limited by the fact that most centers adopt standard practices, and therefore, many studies only reported on a single treatment regimen. Those including patients treated with different regimens may not represent contemporaneous practices, but rather different eras of evolving standard practice or centralized referral after different or failed treatment locally. Reporting of outcomes by subgroups varied from study to study, limiting our ability to adjust for confounding and to include all identified papers for most analyses. Many included studies come from regional referral centers, with potential for higher concentrations of cases with advanced or higher risk disease, possibly upwardly biasing our estimates for mortality.

While we chose to exclude studies lacking English language manuscripts, this led to exclusion of only 7 total studies, with a total of 307 patients. All included English language abstracts, with full-text manuscripts representing 5 different languages (3 Chinese, 1 each of Czech, French, Polish, and Spanish). None of these studies reported on mortality specific to high-risk GTN in the abstract and it is unclear if any of the studies would have otherwise qualified for inclusion. Additionally, exclusion of non-English works has not been shown to represent a significant bias in reported results.⁶²

Interpretation

While our results do not radically change the overall understanding of high-risk GTN, they do provide a formal and novel summary of the global experience with high-risk GTN in the modern era since the FIGO 2000 guidelines, which validates our current understanding and treatment paradigms for this uncommon disease.

EMA/CO was first described by the Charing Cross group in 1986.⁶³ Over time, this regimen has become the most common and preferred first-line chemotherapy.^5,64 It is now explicitly described in the most recent Royal College of Obstetrics and Gynecology, Society of Gynecologic Oncology, and FIGO guidelines,^12,14,15 and formally recommended by the National Comprehensive Cancer Network,¹³ though no randomized comparative trials evaluating its efficacy against alternatives have been published.⁶⁵ EMA/EP was developed initially as a second line regimen and is most often reserved for first-line in ultra-high-risk GTN due to greater relative toxicity.^66,67 Our results offer supportive evidence for EMA/CO as first-line therapy for high-risk disease, in absence of prospective randomized data.

However, EMA/CO and EMA/EP are costly and toxic regimens,^68,69 requiring one to two nights of hospitalization per cycle,⁷⁰ which contributes to both financial and psychosocial costs for patients.⁷¹ High-risk GTN is responsive to a wide range of multi-agent chemotherapies. Our review identified 9 studies focused on 6 other regimens with generally comparable outcomes.^{25,28,31,32,34,36,45,51,53} The efficacy of multiple different regimens is also an important takeaway from this global data, as EMA/CO or EMA/EP may not be accessible or practically feasible to offer to patients in all cases around the world, given differences in available resources. Future research on the tradeoff between clinical benefits and costs of various chemotherapy regimens is warranted.

Conclusion

We found that high-risk GTN is a rare malignancy, described mostly by observational and retrospective institutional series spanning many years. Despite its high likelihood to present with widely metastatic disease, high-risk GTN remains highly curable, with much of the mortality limited to ultra-high-risk disease. Numerous chemotherapy regimens have been shown to lead to high response rates and durable cures, with modern preferred EMA/CO or EMA/EP having the best observational data to support its choice as first-line therapy. While randomized evidence would be ideal to support the choice of this intensive, costly, and sometimes toxic regimen, international collaboration would likely be required to accrue adequate numbers of patients or an appropriately powered study, which may not be practical or feasible given the nature of this disease.