Metastatic gestational trophoblastic disease: A comparison of prognostic classification systems

GYNECOLOGIC

ONCOLOGY

45, 40-45 (1992)

Metastatic Gestational Trophoblastic Disease: A Comparison of Prognostic Classification Systems JOS~E DUBUC-LISSOIR,~ M.D., Division

of Gynecologicl

Oncology,

SUSAN SWEIZIG, M.D., Department

JOHN B. SCHLAERTH, M.D.,

of Obstetrics and Gynecology, University Los Angeles, California 90024

of Southern

AND C. PAUL MORROW, M.D. California

School

of Medicine,

Received July 12, 1991

mycin D, and chlorambucil (MAC) in 1973 [2], the majority of patients with metastatic gestational trophoblastic disease (MGTD) are cured by chemotherapy. In recent years clinical research in MGTD has focused on the reduction of treatment toxicity by more accurate selection of patients who can be cured with single- versus multiagent therapy and by manipulating the drug treatment schedule. There remains, however, a small group of patients with MGTD who continue to have a potentially poor outlook despite application of the standard MAC regimen or any of the newer combination therapies such as MBP (modified Bagshawe protocol) [3] and VBP (velban, bleomytin, &platinum) [4]. While these combinations have been able to salvage a few MAC failures [.5-91, the overall results have been disappointing and the toxicity substantial [lo-131. The recent introduction of etoposide (VP16) has added another very active agent to the antiMGTD armamentarium with efficacy as either first- or second-line therapy [ 14-161. It is now an established component of various combination therapies [16], including EMA/CO (etoposide, methotrexate, actinomycin D, cyclophosphamide, and oncovin), which employs an innoternate multiagent chemotherapy after failing on MAC. There vative schedule for administering these agents. EMA/CO were8 deathsin the 1Zpatient >12 point WHO category(66.7%). On the basisof this analysiswe recommendthat the WHO scoring appears to be less toxic than other regimens and at the yet resystembe utilized for reporting resultsof treatment for MGTD, same time may be the most active combination but the risk categoriesshouldbe redefined:low, <8 points; me- ported [17,18]. With the introduction of these promising new drug prodium, 8-12 points; high, >12 points. o IKJZ Academic PI~SS, IIX. tocols, the identification of patients at risk of treatment failure and death has become of primary importance since INTRODUCTION prompt treatment with more effective regimens could imSince the introduction of methotrexate in 1956 [l] and prove survival. There are currently three prognostic staging-scoring systems, one or more of which are used in combination chemotherapy with methotrexate, actinothis country and around the world: the FIG0 staging ’ Current address: University of Montreal, Hotel-Dieu, 3840 St.- system, the NIH prognosis classification, and the WHO scoring system. It has been proposed that the WHO scorUrbain Street, Montreal, Quebec, Canada H2W lT8. The recordsof 53 consecutivepatients with metastaticgestational trophoblasticdisease(MGTD) treated at the University of Southern California/Los AngelesCounty Medical Center since 1970were analyzed. Forty-eight were evaluable for this study. Treatment during the study period was basedpredominantly on the NIH good-prognosis-poor-prognosis system, employing single-agenttherapy (methotrexate or actinomycin D) for the goodprognosispatients and methotrexate, actinomycin D, cyclophosphamide(MAC) for the poor-prognosispatients. The overall survival rate was 83.3% The study patients were retrospectively classifiedaccordingto the FIGO, NIH, and WHO systemsto test eachsystem’saccuracy in predicting outcomeand the appropriatenessof single-agentor multiagent chemotherapyas the initial treatment in eachcategory. Noneof the systemsascurrently used is clearly superior to the others. Analysis of the WHO scoring systemshowedthat 21 of the 25 (84.0%) study patients with a point score<8 weretreated primarily with a single-agentregimen. All of 21 of thesepatientsachieveda completesustainedremission although 3 (14.3%) required multiagent chemotherapy. The 4 patientsin this point category whoseinitial therapy was a multidrug regimen were also cured. The 23 patients in the WHO high-risk category(>7) had treatment initiated with combination chemotherapy.There were no deaths among the 11 patients in the 8-12 point group, although 3 (27.3%) were salvagedby al-

40 0090~8258/92

$1.50

Copyright 0 1992 by Academic Press,Inc. All rights of reproduction in any form reserved.

METASTATIC

GESTATIONAL

TABLE 1 FIG0 Staging Systemfor GestationalTrophoblasticTumors Stage I I1 III IV

Description Limited to the uterine corpus Extends to the adnexae, outside the uterus, but limited to the genital structures Extends to the lungs with or without genital tract involvement All other metastatic sites

Source. (FIGO), Annual Report on the Results of Treatment Gynecological Cancer, Vol. 20, 1988.

in

ing system is most suited to identifying the high-risk patient [19]. The purpose of this investigation is twofold: (1) to compare retrospectively the reliability of these classification systems in identifying patients with MGTD who will fail chemotherapy and (2) to evaluate the subcategories of each system with respect to the need for singleor multiagent chemotherapy. MATERIALS

TROPHOBLASTIC

(0.5 mg/day), pulse actinomycin D (1.25 mg/m’ every 14 days), weekly im methotrexate (40 mg/m’), and alternating methotrexate/folinic acid. For most of the study period the standard multiagent chemotherapy consisted of MAC (methotrexate, 0.3 mg/kg; actinomycin D, 810 pg/kg; and cyclophosphamide, 3-5 mg/kg (or chlorambucil)) all given daily for 5 days every 2 weeks. No patient received intrathecal chemotherapy. Radiation therapy was given to all patients with brain metastases (2100-3000 cGy) and to some patients with liver metastases (2000-2400 cGy). One patient with extradural GTD received radiation therapy to the spine and another was given 800 cGy to the bladder to control lifethreatening hemorrhage. Other sites were irradiated in the occasional patient with drug-resistant disease that was not resectable. Several patients underwent surgery to control bleeding or resect drug-resistant disease (one laminectomy, five hysterectomies, five thoracotomies, and six taparotomies. RESULTS

AND METHODS

From January 1970 to March 1991, 53 patients with MGTD were treated at the University of Southern California/Los Angeles County Medical Center. All charts were reviewed for prognostic, treatment, and survival data. The study patients were categorized by the FIGO, NIH, and WHO prognostic scoring or staging systems (Tables l-3). Five patients could not be completely analyzed by the WHO scoring system because their ABO blood group, which might have increased their risk category, was not available. Three were cured and two quit treatment, one responding and one failing treatment. They were excluded from the study. Median follow-up of the 48-patient study group was 39 months with a range from 0 to 147 months. Only 1 patient had less than 6 months follow-up. Two patients responding well to therapy were lost before completion of treatment. During the same time period 657 patients with molar gestation and 180 patients with nonmetastatic gestational trophoblastic disease were managed at this medical center. Pretreatment evaluation included a complete history and physical examination; hematologic, liver, and renal function tests; chest X ray; pelvic ultrasound; and radionuclide or CT scan of the brain and liver. CT scan of the lungs was performed when the routine two-view chest X ray was equivocal for metastases. Serum HCG measurement was performed prior to each chemotherapy treatment cycle. Patients were treated with various protocols during the study period, usually on the basis of their NIH prognostic grouping. Single-agent regimens employed were 5-day iv methotrexate (0.4 mg/kg/day), 5-day iv actinomycin D

41

DISEASE

The study patients have been retrospectively classified according to the FIGO, NIH, and WHO systems. The number of patients in each subgroup, the treatment, and the outcome are listed for each system in Tables 4-6. Altogether there were 8 deaths occuring in the 48 patients. Using the FIG0 staging system (Table 4), 3 of the 48 patients (6.25%) fell into the stage II group while 75% of the patients were in stage III. The observed death rates for stages II, III, and IV were 0, 5.5, and 66.7%. The single stage II patient receiving multiagent chemotherapy TABLE 2 National Institute of Health (NIH) Classificationof Gestational TrophoblasticNeoplasia Group I II

Description Nonmetastatic disease: No evidence of disease outside the uterus Metastatic disease: Any disease outside the uterus A. Good-prognosis metastatic disease 1. Short duration (last pregnancy <4 months) 2. Low pretreatment HCG titers: <100,000 W/24 hr in urine or <40,000 mIU/ml of serum 3. No metastasis to brain or liver 4. No significant prior chemotherapy B. Poor-prognosis metastatic disease 1. Long duration (last pregnancy >4 months) 2. High pretreatment HCG titer: >lOO,COO IU/24 hr in urine or >40,000 mIU/ml of serum 3. Brain or liver metastasis 4. Significant prior chemotherapy 5. Term pregnancy

Source. Hammond et

al.

[2].

42

DUBUC-LISSOIR

ET AL.

TABLE 3 World Health Organization (WHO) Scoring Systemfor GestationalTrophoblasticTumors Prognostic factor Score Age (years) Antecedent pregnancy Interval between end of antecedent pregnancy and start of chemo (months) HCG (mIU/ml) ABO group (female x male)

0

1

<39 Mole

>39 Abortion

<4
Largest tumor (cm) incl uterine Site of metastases Number of metastases Prior chemotherapy

4

7-12 104-105 B AB >5 GI, liver 4-8 Single agent

>12 >l@

Term

4-6 l@-lo4 OxA Ax0 3-5 Spleen, kidney l-4

<3

2

Brain >8 Two or more drugs

Note. Total score: <.5, low risk; 5-7, middle risk; >7, high risk (Ref. [24]).

fell into the WHO high-risk and the NIH poor-prognosis groups. Half the patients in stage III and all the patients in stage IV received multiagent chemotherapy. In the NIH staging system, 22 (45.8%) of the study patients are classified as having good-prognosis MGTD (Table 5). No deaths occurred in this category. Singleagent therapy was changed in 3 patients, 1 because of toxicity and 2 because of rising HCG levels. Two patients received primary multiagent chemotherapy (methotrexate and actinomycin D), and an additional two patients received multiagent chemotherapy because of single-agent resistance. All but 1 of the 26 poor-prognosis patients initially received multiagent chemotherapy. The mortality rate in the poor-prognosis group was 30.7% (8 of 26). Results of the WHO scoring system are presented in Table 6. All 8 deaths were observed in the high-risk category constituting 34.8% of the 23 patients in the group. All patients in this group were treated with multiagent chemotherapy. Only 6 patients qualified for middle-risk status. Four of the twenty-five low/middle-risk patients initially received multiagent chemotherapy (16%). Two patients in the NIH poor-prognosis group on the basis of a pretreatment serum HCG levels >40,000 mIU/ml and

one patient whose reference pregnancy anteceded the diagnosis of trophoblastic disease by more than 4 months fell into the WHO low- or middle-risk group. Eleven of the twenty-three WHO high-risk patients had scores from 8 to 12. All 11 were treated with multiagent chemotherapy (MAC, 10; VPB, 1). Ten patients achieved sustained clinical and HCG level complete remission. The 11th patient had a good level response but, because of the toxicity, discontinued treatment before remission. Twelve patients had a WHO score >12, and eight of the twelve (66.7%) died. Of these 8 patients, 1 died of sepsis (extensive liver metastases treated with radiation therapy and one course of combination chemotherapy) and a second probably died from complications of the disease (extensive lung metastases with “adult respiratory distress sydrome”; received only one course of chemotherapy). The antecedent pregnancy was a molar gestation in all patients fitting into the WHO low- and middle-risk groups as well as the NIH good-prognosis group. All 11 patients with an antecedent-term pregnancy were in the NIH poorprognosis group (by definition) and the WHO high-risk category. Five of these eleven patients scored more than 12 and all five died. Of the remaining WHO high-risk

TABLE 4 Patient Distribution, Treatment, and Outcome Using the FIG0 Staging Systemfor GestationalTrophoblastic Disease

TABLE 5 Patient Distribution, Treatment, and Outcome Usingthe NIH Staging Classificationfor GestationalTrophoblasticDisease

Initial chemotherapy FIG0 stage I” II III IV

Total NW) 3 (6.25) 36 (75.00) 9 (18.75)

a Nonmetastatic.

Single agent

Multiagent

-

2 19 0

Chemotherapy Deaths NC%)

1 17 9

0 2 (5.5) 6 (66.7)

Group

Total NC%) -

I II Good prognosis Poor prognosis ’ Nonmetastatic.

22 (45.8) 26 (54.2)

Single agent Multiagent

Deaths NC%)

-

-

-

20 1

2 25

0 8 (30.7)

METASTATIC

GESTATIONAL

TABLE 6 Patient Distribution, Treatment, and Outcome Using the WHO ScoringSystemfor GestationalTrophoblasticDisease

Risk group Low Middle High

Chemotherapy

Total NC%)

Single agent

Multiagent

Deaths NC%)

19 (39.6) 6 (12.5) 23 (47.9)

16 5 0

3 1 23

0 0 8 (34.8)

11

0

12

8 (66.7)

8-12 points 11 (22.9)

0 >12 points

12 (25.0)

0

patients, 4 had an antecedent abortion and 8 an antecedent molar pregnancy. There were no apparent differences in the age (median, 28 years), gravidity, or parity among the various risk groups. The median interval from the end of the reference pregnancy to the start of chemotherapy for the WHO categories was 2 months for the low- and middle-risk groups, 7 months for the high-risk group scoring 8-12, and greater than 24 months for the high-risk patients with a score >12. Tumor size in the low- and middle-risk WHO categories was generally less than 3 cm; none was greater than 5 cm. In the WHO high-risk group, the tumor size was more than 5 cm in six patients (Table 7). Four of these six patients had scores greater than 12 and three died. All eight patients with more than eight metastases were in the WHO high-risk group and four died. Seven patients had CNS involvement, six with a score greater than 12. Four of the seven are dead of disease. All four patients with liver metastases scored more than 12 and two are dead. Eight patients had metastases to more than one site and six of the eight died. DISCUSSION The purpose of this retrospective study was to compare the various staging systems that are being used to classify gestational trophoblastic disease (GTD) and to determine which is the most accurate in predicting an adverse outcome. Since currently available chemotherapy regimens are capable of curing the great majority of patients with GTD, even those with metastases, clinical research has focused on two areas: identifying single-agent therapy that is therapeutically equivalent to the traditional 5-day methotrexate and actinomycin D regimens but less toxic and less expensive and developing more effective remedies for those cases that cannot be cured by currently available drug regimens. The latter objective would be facilitated

TROPHOBLASTIC

DISEASE

43

by an accurate system to predict which patients will fail conventional therapy. It is not immediately obvious from the results of this study that any one of the three staging systems is clearly superior to the others. For example, the FIG0 system produces the poorest distribution of cases per subgroup since 75% fall into the stage III category. In addition only six of the eight treatment deaths were in the stage IV group. On the other hand, six of nine (66.7%) of the patients in the stage IV group died of disease, a predictive accuracy that substantially exceeds the failure rate in the poor-prognosis category of the NIH system (30.7%) and in the high-risk group of the WHO scoring system (34.8%). The FIG0 stage distribution of this study population does not allow a reasonable indication of which patients need multiagent chemotherapy since only 6.25% of the cases fell into the low-risk group (stage II), and two of the treatment failures occurred in the stage III category. The NIH system has only two subgroups, good and poor prognosis. In our study population the patients were fairly evenly distributed between these categories and all the treatment deaths occurred in the high-risk, i.e., the poor-prognosis, group. However, only 30.7% of the poorprognosis group failed treatment. Thus, its predictive accuracy limits its value in terms of identifying patients who are at great risk of failing standard chemotherapy. In addition, assuming that “high-risk” patients need multiagent chemotherapy, this system of classification would lead to this intensive therapy in a somewhat higher percentage of the MGTD patients than the WHO scoring system. One would expect that the relatively complex WHO scoring system, which takes into account more prognostic factors than the other systems and in a semiquantitative fashion, would easily be more predictive of outcome than the other two systems. However, from a distribution perspective, in this study population only 6 patients (12.5%) fell into the middle-risk category, and none of the lowor middle-risk category patients died of their disease. The high-risk category is very similar to that in the NIH system in that approximately half (47.9%) of the patients fit into that group and about one-third (34.8%) died. Nevertheless, with the WHO scoring system it is possible to divide the high-risk category into subgroups: those patients with a score of 8-12 points and those with a score greater than 12 points. All 8 deaths were observed among the 12 patients in the >12 point subgroup (66.7%), making this category the most specific and at the.same time the most sensitive predictor of death from treatment failure of all three systems. Others have also concluded that the WHO system is superior to the NIH and FIG0 classifications (19-21). Clearly all patients in the >12 point risk category need

44

DUBUC-LISSOIR

ET AL.

TABLE 7 Risk Factors for All Patients in the WHO High-Risk Category (years)

Prior Pw

Interval (months)

18 18 32 29 33 29 33 31 18 28 28

Term Eab Term Mole Term Eab Term Mole Term Term Mole

2 6 3 26 20 2 I 12 18 4 17

Score 8-12 points >5 CNS 3-5 Lung <3 Lung <3 Lung <3 Lung 3-5 Lung 3-5 Lung >5 Lung 3-5 Vag <3 Lung, vag <3 Lung, vag

1 4 6 2 1 4-8 1 3 1 >8 >8

49 15 19 20 28 28 27 22 20 22 28 40

Sab Sab Term Mole Term Term Mole Mole Mole Term Mole Term

24 24 5 18 21 23 68 24 22 21 24 26

Score > 12 points >5 Multi 3-5 Lung <3 Multi” 3-5 Lung >5 Lung, kidney 3-5 Multi >5 CNS, lung 3-5 Multi* >5 Multi 3-5 CNS, lung <3 Lung <3 Lung

4 4 >8 >8 4-8 >8 3 >8 >8 4-8 4-8 >8

Age

Largest tumor (cm)

Metas site

No. Metas.

Prior chemo

Single Single Single

Multi Single Multi Multi

Status

A A A A A A A A A A A A A D A D D D A D D D D

Note. Eab, elective abortion; Sab, spontaneous abortion; A, alive and well; D, dead. a Liver, CNS, lung. ’ Liver, spleen, omentum. ’ Liver, CNS, breast, mastoid, pancreas. d Liver, CNS, lung, small bowel. ’ CNS, lung, abdomen.

aggressive chemotherapy. In view of the fact that twothirds of the patients in this group died, it also comprises a population for whom it is suitable to study potentially more effective chemotherapy. An important subsidiary issue is which if any patients in the 12 point and under category need to be treated up-front with multiagent chemotherapy, as suggested by some authors [17-251. All but 4 of the 25 WHO low/middle-risk patients in our study group received single-agent chemotherapy as initial treatment. Although 6 of these 21 patients failed singleagent therapy, all were salvaged by changing to a different single-agent (3 cases) or multiagent (3 cases) therapy. These results certainly support the continued use of single-agent chemotherapy as the primary treatment of patients with a WHO score of less than 8, as reported by others [21]. All 11 of the high-risk patients in the 8-12 point category were treated initially with multiagent chemotherapy and 3 (27.2%) failed to achieve remission with the first-line drug regimen. All 3 were, however, salvaged by changing to alternate combination drug reg-

imens (2 were MAC failures salvaged by platinum-based chemotherapy). Contrary to the data reported by some authors [22], patients with liver or CNS metastases and patients who failed initial chemotherapy in our study population having a WHO score greater than 8 did not fare worse than patients without these features. However, our case numbers are too small to make conclusions. Time interval greater than 4 months since the antecedent pregnancy, postterm MGTD, and multiple organ metastases were more often associated with treatment failure and death in patients with a WHO score greater than 12, as has been reported elsewhere [23]. We also note that three of four patients in our study group with intra-abdominal metastases are deceased. Considering these results it would seem reasonable to utilize the WHO scoring system to classify patients with GTD in preference to the NIH and the FIG0 systems, provided certain adjustments are made. First, the scoring system should apply only to patients with metastatic dis-

METASTATIC

GESTATIONAL

ease. As it is currently set up, the WHO system is not limited to patients with metastatic disease and a patient with nonmetastatic disease could fall into the high-risk category. Second, the point definition of low, middle, and high risk needs to be changed. We suggest that all cases with a point score less than 8 be considered low risk, those with a score of 8-12 be classified as middle risk, and cases with a score greater than 12 constitute the highrisk category. The initial treatment for the low-risk group should be single-agent chemotherapy. While the middlerisk and high-risk groups both require multiagent chemotherapy, the high-risk designation also will identify nearly all the patients destined to die of their disease and, therefore, those most suited to clinical studies in search of a more effective therapy. Our standard drug regimen for the patient with MGTD at high risk for treatment failure has been MAC. Some MAC failures have been salvaged by platinum-based chemotherapy in our hands as well as others [5,6,9,19]. It has been suggested that EMA/CO may be superior to MAC therapy and is also reported to be less toxic than MAC, chamoca, or chamoma [17,18,25]. We are unable to confirm from our reading of the literature, however, whether this regimen will cure more of the high-risk (WHO score greater than 12) patients than the standard MAC regimen. REFERENCES 1. Li, M. D., Hertz, R., and Spencer, D. B. Effects of methotrexate upon choriocarcinoma and chorioadenoma, Proc. Sot. Exp. Biol. Med. 93, 361 (1956). 2. Hammond, C. B., Borchert, L. G., Tyrey, L., Creasman, W. T., and Parker, R. T. Treatment of metastatic trophoblastic disease: Good and poor prognosis, Am. J. Obstet. gynecol. 115,451 (1973). 3. Surwit, E. A., Suciu, T. N., Schmidt, H. J., and Hammond, C. B. A new combination chemotherapy for resistant trophoblastic disease, Gynecol. Oncol. 8, 110 (1979). 4. Einhorn, L. H., and Donohue, J. cti-Diamminedichloroplatinum, vinblastine and bleomycin combination chemotherapy in disseminated testicular cancer, Ann. Intern. Med. 87, 293 (1977). 5. Surwit, E. A., Alberts, D. S., Chirstian, C. D., and Graham, V. E. Poor prognosis gestational trophoblastic disease: An update, Obstet. Gynecol. 64, 21 (1984). 6. Surwit, E. A., and Hammond, C. B. Treatment of metastatic trophoblastic disease with poor prognosis, Obstet. Gynecol. 55, 565 (1980). Schlaerth, J. B., Morrow, C. P., and De Petrillo, A. D. Sustained remission of choriocarcinoma with &-platinum, vinblastine, and bleomycin after failure of conventional combination drug therapy, Am. J. Obstet. Gynecol. l36, 983 (1980). Weed, J. C., Barnard, D. E., Currie, J. L., Clayton, L. A., and Hammond. C. B. Chemotheranv., with the modified Bagshawe nm-

TROPHOBLASTIC

DISEASE

45

tocol for poor prognosis metastatic trophoblastic disease,Obstet. gynecol. 59, 377 (1982). Berkowitz, R. S., Goldstein, D. P., and Bernstein, M. R. Modified triple chemotherapy in the management of high risk metastatic gestational trophoblastic tumors, Gynecol. Oncol. 19, 173 (1984). 10. Gordon, A. N., Gershenson, D. M., Copeland, L. J., Saul, P. B., Kavanagh, J. J., and Edwards, C. L. High risk metastatic gestational trophoblastic disease, Obstet. Gynecol. 65, 830 (1985). 11. Lurain, J. R., and Brewer, J. I. Treatment of high-risk gestational trophoblastic disease with methotrexate, actinomycin D and cyclophosphamide chemotherapy, Obstet. Gynecol. 65, 830 (1985). 12. Curry, S. L., Blessing, J. A., DiSaia, P. J., Soper, J. T., and Twiggs, L. B. A prospective randomized comparison of methotrexate, dactinomycin and chlorambucil versus methotrexate, dactinomycin, cyclophosphamide, doxorubicin, melphalan, hydroxyurea, and vincristine in “poor prognosis” metastatic gestational trophoblastic disease: A Gynecologic Oncology Group Study, Obstet. Gynecol. 9.

73, 357 (1989).

13. Lurain, J. R., Brewer, J. I., Mazur, M. T., and Torok, E. E. Fatal gestational trophoblastic disease: An analysis of treatment failures. Am. J. Obstet. Gynecol. 144, 391 (1982). 14. Newlands, E. S., and Bagshawe, K. D. Antitumor activity of the epipodophyllin derivative VP-16 (etoposide; NSC 141540) in gestational choriocarcinoma, Eur. J. Cancer 16, 401 (1980). 1.5. Wong, L. C., Choo, Y. C., Sing, B. S., and Ma, H. K. Primary oral etoposide therapy in gestational trophoblastic disease, Cancer 58, 14 (1986). 16. Wong, L. C., Choo, Y. C., and Ma, H. K. Etoposide, methotrexate and bleomycin in drug resistant gestational trophoblastic disease, Gynecol. Oncol. 24, 51 (1986). 17. Newlands, E. S., Bagshawe, K. D., Begent, R. H. J., Rustin, G. J. S., Holden, L., and Dent, J. Developments in chemotherapy for medium and high-risk patients with gestational trophoblastic tumors (1979-1984), Br. J. Obstet. Gynecol. 93, 63 (1986). 18. Bolis, G., Bonazzi, C., Landoni, F., Mangili, F. V., Zanaboni, F., and Mangioni, C. EMA/CO regimen in high risk gestational trophoblastic tumor, Gvnecol. Oncol. 31, 439 (1988). 19. Mortakis, A. E., and Brage, C. A. “Poor prognosis” metastatic gestational trophoblastic disease: The prognostic significance of the scoring system in predicting chemotherapy failures, Obstet. Gynecol. 76, 272 (1990). 20.

21.

22.

23. 24. 25.

Lurain, J. R., Cassanova, L. A., Miller, D. S., and Rademaker, A. W. Prognostic factors in gestational trophoblastic tumors: A proposed new scoring system based on multivariate analysis, Am. J. Obstet. Gynecol. 164,611 (1991). Gordon, A. N., Gershenson, D. M., Copeland, L. J., Stringer, C. A., Morris, M., and Wharton, J. T. High-risk gestational trophoblastic disease: Further stratification into clinical entities, Gynecol. Oncol. 34, 54 (1989). DuBeshter, B., Berkowitz, R. S., Goldstein, D. P., and Bernstein, M. R. Analysis of treatment failure in high-risk metastatic gestational trophoblastic disease, Gynecol. Oncol. 29, 199 (1988). Soper, J. T., Clarke-Pearson, D., and Hammond, C. B. Metastatic gestational trophoblastic disease: Prognostic factors in previously untreated patients, Obstet. Gynecol. 71, 338 (1988). World Health Organization Scientific Group. Gestational trophoblastic disease. Tech. Rep. Ser. 692, Geneva, p. 51 (1983). Bagshawe, K. D. Treatment of high-risk choriocarcinoma, J. Reprod. Med. 29, 613 (1984).