Aust. NZ J Obsret Gyaecol 1996; 3 6 3: 331

Regression Pattern of Beta Human Chorionic Gonadotrophin in Blood after Chemotherapy for Gestational

Trophoblastic Neoplasia A. Burrows,’ M.J. Faddy‘ and S.K. Khm3

Department of Obstetrics and Gynaecology and Department of Mathematics, The Universiry of Queensland and the Royal Women’s Hospital, Brisbane

Summary: Although beta human chorionic gonadotrophin (beta-HCG) is a clinically- useful tumour marker in gestational trophoblastic neoplasia, there is limited information on how serum beta HCG regresses in molar pregnancy and choriocarcinoma during chemotherapy. The study included 41 patients who had been treated by single- and multidrug regimens, and decay curves for regression of beta HCG in the circulation in these patients were fitted to the data. Thirty-three patients achieved a biochemical remission (‘cured’) by first-line chemotherapy; this gave an overall efficacy rate of 80%. The beta HCG decay curves convey useful information of the chemosensitivity of the tumour, and may assist in determining the time required for treatment and earlier change in treatment for the chemoresistant tumour.

It is now conventional practice to use human chorionic gonadotrophin (HCG) as the marker of tumour activity in gestational trophoblastic neoplasia. Within the spectrum of HCG-producing lesions - especially those that originate from a molar pregnancy - some will undergo spontaneous remission whilst others will progress to malignant nonmetastatic or metastatic sequelae and become responsive or refractory to cytotoxic chemotherapy (1). An awareness of this complexity is important in the interpretation of serial HCG patterns in monitoring the course of the disease.

The recognition that the HCG molecule possesses a specific beta-subunit has enhanced the usefulness of the marker. Delfs as early as 1959 (2 ) , first demonstrated the importance of HCG levels in the detection of residual trophoblastic disease following treatment for hydatidiform mole. It is now accepted that HCG levels correlate well with the amount of tumour (3-5) and are, therefore, useful in follow-up of the disease. It has been shown that the regression of HCG is characterized by an exponential decline following molar evacuation (5-10). When the HCG levels plateau or start to increase, the disease is deemed to be persistent and chemotherapy is considered necessary.

1. Registrar. 2. Senior Lecturer in Statistics. 3. Professor. Address for correspondence: Professor S.K. Khm, The University of Queensland, Department of Obstetrics and Gynaecology, Clinical Sciences Building, Royal Brisbane Hospital, Herston, Queensland 4029.

However, the pattern of HCG regression which can predict a good or poor response to chemotherapy in the treatment of persistent disease is less well described. Rotmensch et al (7) reported the disap- pearance of beta HCG in an exponential way in chemosensitive tumours in 19 of 21 patients who required methotrexate for nonmetastatic gestational trophoblastic neoplasia. Bagshawe (5) also reported the beta HCG regression in a logarithmic way; he described an initial increase in HCG values with the first course of chemotherapy, and a subsequent progressive logarithmic fall.

The aim of this study was to follow the weekly rate of fall of serum levels of beta HCG in consecutive patients who had been treated by chemotherapy for gestational trophoblastic neoplasia and develop a mathematical model for the regression pattern. The hypothesis to be tested was that the weekly rate of fall of beta HCG is an indicator of a response to chemo- therapy, and that, using a mathematical model, an earlier change of treatment could be determined for unsatisfactory response. The database for this study were 41 patients who had been assessed and treated in 1 centre in whom the type of molar pregnancy, response to various chemotherapeutic regimens, side- effects of treatment and long-term remission were known.

MATERIALS AND METHODS During the period 1980-1995, 41 consecutive

patients who received chemotherapy for gestational trophoblastic neoplasia at the Royal Women’s Hospital, Brisbane, were studied. The patients were derived from the Queensland Registry for Gestational Trophoblastic Neoplasia which administered the registration and the management of patients who developed malignant sequelae.

332 AUST. A N D N.2. JOURNAL OF OBSTETRICS A N D GYNAECOLOGY

Data on the type of index pregnancy, histological diagnosis, and presencefsites of metastatic disease were collected. The initial beta HCG levels in serum and the subsequent serial levels tested weekly during chemotherapy, were recorded,

According to current convention, weekly quantita- tive sewm beta HCG levels were measured during chemotherapy in 3 established laboratories. The cut- off value for a ‘negative’ result was standardized at 4 U L at each laboratory as this was the limit of sensitivity of the assay. The present protocol allowed chemotherapy to be stopped after 3 consecutive weekly ‘negative’ results. Thereafter, the follow-up required beta HCG levels to be determined every month for 12 months, and the patients were advised on use of effective contraception; the majority used the oral contraceptive pill.

The patients with gestational trophoblastic neo- plasia confirmed by histological examination were considered for cytotoxic chemotherapy because of rising or stationary levels over a 3-week period during follow-up. Some of these patients subsequently required a change of chemotherapy regimen or were advised to have a hysterectomy.

The Registry considered patients with the following criteria to be in the ‘high risk’ group: histologically- confirmed choriocarcinoma, delay greater than 4 months before initiation of treatment, evidence of metastases other than in the lungs, and failed previous chemotherapy. In our experience, this classification allowed a clear definition of the ‘high-risk‘ group. Other patients without any of these criteria were placed in the ‘low-risk’ group. The chemotherapy regimen developed for the ‘low-risk’ patients is as follows: IV methotrexate 300 mg over 30 hours, followed by rescue with folinic acid 15 mg in 4 doses; and that for the ‘high-risk’ patients is the EMACO regimen when the following drugs were included: methotrexate with folinic acid rescue, actinomycin-D, etoposide, vincristine and cyclophosphamide. Both regimens were repeated every 2 weeks.

From the data collected on 41 patients, the following were calculated: mean beta HCG values and mean log-beta HCG values (and standard deviation) for the various subgroups of patients with complete mole, partial mole, and choriocarcinorna; time taken to complete treatments; number of courses of chemo- therapy required; and time to complete treatment according to the initial beta HCG level.

Decay curves were fitted to the beta HCG data from all the patients receiving first-line treatment using a mathematical model; these decay curves were devised to allow for possible multiexponential behaviour, and changing patterns of variation as the treatment progressed.

The change from first-line therapy was made on the basis of an unsatisfactory fall in beta HCG levels over 3 consecutive weeks (a plateau or a rise). These

regression patterns in these patients were related to the mathematically-developed regression curve for patients ‘cured‘ with first-line therapy.

RESULTS Of the 41 patients treated by chemotherapy for

gestational trophoblastic neoplasia, the histological diagnosis on the initial gestational tissues was a complete mole in 33 patients, and a partial mole in 5; there were 3 patients with a choriocarcinoma, based on histological examination of tissues obtained by uterine curettage following an episode of abnormal vaginal bleeding after a term pregnancy.

The range in age of the patients was 17-41 years (mean of 27.7 years). There were no recorded deaths during the study period.

The range of individual initial beta HCG levels was from 53 U L to 480,000 UL. The highest mean tog beta HCG value was obtained in patients with a choriocarcinoma (1 1.58 f 2.01)- by comparison with those with a partial mole (7.787 * 2/04) and a complete mole (7.38 f 2.38).

The mean time taken to complete chemotherapy varied between the 3 subgroups, with patients with a partial mole (14.4 * 10.5 weeks) requiring more time than those with a complete mole (8.6 5.4 weeks). Patients with a choriocarcinoma who were treated by the EMACO regimen required a mean time of 12.7 * 5.3 weeks). In this group of patients, the mean number of treatment cycles required to reach a biochemical remission was 5.3 3.0. It was also found that a high pretreatment level tended to require a longer time to complete treatment (table 1). For levels L10,OOO UL. the time was 14.8 weeks and number of cycles 7.1, whereas for medium levels 1.001-10,OOO U/L, they were 8.3 weeks and 4.0 cycles. and for low levels <l,OOO U/L. they were 7.2 weeks and 4.2 cycles.

’bblc 1. Rcbtioarhip to Prrtreatmcnt Beta HCG Levchr to Tlmc nnd Number d Cycles to Compkte ’lbttmcnt

MCM no. of Retnatmcnt Mean time level in weeks (+ SD) cyclcs (t SD) Low (51,000 lull) 7.2 4.2 (n=20) (* 6.0) (t 2.7) Medium ( I .00 1 - 1 O.Oo0) 8.3 4.0 (n=8) (i 4.8) (t 2.2) High (210.001) 14.8 7.1 (n=13) (i 7.3) (t 2.2)

Figure 1 shows the multiexponential decay of serial beta HCG values for the whole patient population, with the regression curve and the upper and lower 95% limits from the regression analysis. The reducing variation apparent from the data (converging limits, as treatment progresses) was a strong effect (p value <O.oOol). The 4 points which were substantially above the upper 95% limit belonged to a single patient

A. BURROWS ET AL 333

REGRESSION ANALYSIS & 95% UMKS

los ‘-tK , O

I 5 10 15

11me (weeks)

Figure I . This shows the multiexponentid decay of serial beta HCG values in the study, with the regression curve and the upper and lower 95% limits.

PATIENTS REQUIRING A CHANGE IN THERAPY 1 0

1 6 ~ , , , ~ , , l l , , , , , , , l , , . , , , , , ( I

0 5 10 15 20 25

time (weeks)

Figure 2. This shows the beta HCG patterns in 8 individual patients who received a change in therapy, in relation to the 95% regression limits for the study. As indicated by the arrows which denoted the time at which the chemotherapy was changed, the change from first-line treatment occurred several weeks after the individual curve crossed the upper 95% limit.

who had choriocarcinoma following a term pregnancy. She was treated according to the ‘high- risk’ chemotherapy protocol with the EMACO regimen and responded satisfactorily to treatment, and achieved ‘cure’ after 10 weeks of treatment.

Figure 2 shows the serial beta HCG patterns in the 8 individual patients who received a change in therapy

in relation to the 95% regression limit for the study. As indicated by the arrows which denoted the time at which the chemotherapy was changed, the change from first-line treatment occurred several weeks after the individual curve crossed the upper 95% limit. Nevertheless, all those patients whose f i t - l ine treat- ment was altered were ‘cured’of the disease; 2 under- went a hysterectomy, as they had completed their family, and the tumour appeared to be localized to the uterus. In 1 patient, the beta HCG levels fell rapidly to negative; in the other, a further 9 courses of chemo- therapy were required to bring the levels to negative.

Six patients in the ‘low-risk’ group were given an additional actinomycin-D to the standard methotrex- ate/folinic acid regimen because of unsatisfactory decline in beta HCG levels. These patients reached a biochemical remission and were able to retain repro- ductive function but treatment took as long as 25 weeks. All the patients in the study remained free of disease (‘cured’) during the period of observation.

There were no serious side-effects from the chemo- therapy used. Side-effects reported with the EMACO regimen included nausea, vomiting, alopecia, abdomi- nal pain, transient dizziness, mouth ulcers and sore throat. Anaemia with haemoglobin level <lo0 g/dL occurred in 2 patients. With the methotrexate-folinic acid regimen, all patients were free of any side-effects.

DISCUSSION Human chorionic gonadotrophin (HCG), a glyco-

protein hormone, plays an integral role in the follow- up and treatment of trophoblastic neoplasia. Synthe- sized by the terminally differentiated syncytiotroph- oblast, HCG is an ‘ideal’ tumour marker (5). The hormone is produced by both normal and neoplastic trophoblastic tissue. The alpha and beta subunits of HCG are joined noncovalently, and modem mono- clonal assays are able to reliably measure the beta HCG moiety. Therefore, serial beta HCG values can be used to predict the amount of tumour load in an individual patient.

After the successful evacuation of a molar pregnancy by suction aspiration, there is agreement that beta HCG regression curves are predictive in outlining the decay and eventual disappearance of viable trophoblastic tissue (5-7; 9-1 1). When beta HCG levels are plotted against time on semilog paper, it is apparent that there is an exponential decline in levels. If this decline, however, does not progress as expected, then the diagnosis of persistence of gestational trophoblastic neoplasia is made, and chemotherapy is then instituted. Protocols of management vary; however, the general consensus is to start treatment if there is a plateau of beta HCG levels for 3 consecutive weeks (4,5,7,8), if the levels rise over 2 consecutive weeks (7), or if beta HCG is detected more than 4-5 months after evacuation of

334 AUST. AND N.Z. JOURNAL OF OBSTETRICS AND GYNAECOL~GY

a molar pregnancy (5). Single-agent regimens using either methotrexate or actinomycin-D, have changed little since their first use over 30 years ago (6). The use of etoposide as a single agent for treatment of gestational trophoblastic disease has been limited by side-effects (6). However, the drug is used effectively in combination therapy in ‘high-risk’ groups.

Once chemotherapy is started, close follow-up is necessary and weekly beta HCG levels are used to indicate the chemosensitivity of the tumour. There are comparatively little data on how these tumour marker levels behave during treatment. Bagshawe (5) postulated a one-log fall with each treatment cycle early in treatment, but a much smaller net fall towards the latter stages of treatment. Rotmensch et al (7) graphically illustrated an exponential fall in 19 patients who had been treated successfully with methotrexate. Two patients in whom the HCG levels showed a plateau above the 90th percentile, were identified as ‘at risk’ for drug resistance.

It is this question of drug resistance that provoked our study. Our objective was to develop beta HCG regression curves which are associated with satisfactory response to drugs in Australian women who have been treated by chemotherapy, be it single or multidrug. Our data show that in patients with a chemosensitive tumour, the regression curves of beta HCG followed those of patients who underwent spontaneous remission after effective evacuation of the uterus. We found that early detection of unsatisfactory response to primary chemotherapy was possible by relating the serial weekly HCG values to the mathematical model. In the 8 patients who failed to achieve a ‘satisfactory’ decline in beta HCG levels, a change of treatment was made by conventional means - based on ‘eye-balling’ a set of beta HCG levels - and this occurred several weeks after the individual curve crossed the 95% limit in the mathematical regression curve. We believe this mathematical regression curve for interpretation of HCG patterns is sound and less arbitrary, and can be used by clinicians to base their decision to alter treatment.

We also found the beta HCG regression curves derived from the 33 patients who showed a satis- factory response to first-line chemotherapy helpful in predicting the length of time required to complete treatment. Those with initial levels greater than 10,OOO ULrequired the longest time (14.8 weeks) and the most number of cycles (7. I), as expected. Overall, our patients were treated for a mean time of 9.6 weeks (SD=7.03), with a total of 5.3 cycles of chemotherapy. This is in accordance with previous studies. Rotmensch et a1 (7), in a study of 21 patients, reported an average of 6 weeks of treatment and 3 courses of chemotherapy were required. Berkowitz et a1 (1 2), in a series of 22 patients treated with methotrexate, found that remission was achieved after 2.2 courses of

chemotherapy; and Kamariah et a1 (9), in 16 patients, reported an average of 11.7 weeks (range 5- 19 weeks) to achieve remission.

The present study showed that our current treatment protocol which chose the chemotherapy according to ‘low-risk’ and ‘high-risk‘ was effective and relatively free of serious side-effects. No patients suffered a life- threatening complication; only minor side-effects were encountered, even with the multidrug regimen. This compared favourably with other reports. Ayhan (13) described a 16.9% complication rate with methotrexate without folinic acid, which included haematological toxicity (8.4%), hepatic toxicity (2.8%), and even deaths (2.8%, n=2). However, when folinic acid rescue was incorporated in the metho- trexate regimen, Berkowitz et a1 found no haemato- logical toxicity, hepatotoxicity, or alopecia in I63 patients (1 2).

The multiexponential beta HCG regression curve which has been developed from our present data is useful in determining chemosensitivity of the tumour. Of patients showing chemoresistance to first-line treatment this effect is provided by the serial values exceeding the upper 95% confidence limits of the regression curve. This can be used to make earlier change in treatment.

References 1. Lurain JR, Brewer JI, Torok EE. Halbem B. Natural history of

hydatidiform mole after primary evacuation. Am J Obstet Gynecol 1983; 591-59s.

2. Delfs E. Chorionic gonadotrophin determinations with hydati- diform mole and choriocarcinoma. Ann NY Acad Sci 1959: 80 12s.

3. Brewer JI . Textbook of Gynecology. Baltimore. Williams & Wilkins Company Third Edition 1%1; 2 6 643.

4. Goldstein DR Berkowitz RS. Current management of complete and panial molar pregnancy. J Reprod Med 1994: 39: I 39- 146.

5. Bagshawc KD. Choriocarcinoma: A model for tumour markers. Reviews in Oncology 1992; 5: 99-106.

6. Kohorn El. Single-agent chemotherapy for nonmetastatic pesta- tional trophoblastic neoplasia. J Reprod Med 1991: 36: 49-5.5.

7. Rotmensch J. Rosenshein NB. Block BS. Comparison of human chorionic gonadotrophin regression in molar pregnancies and post-molar non-metastatic gestational trophoblastic neoplasia. Gynec Oncol 1988; 29: 82-86.

8. Kohorn El. Evaluation of the criteria u.wd to make the diagnosis of nonmctaqtatic gestational trophoblastic neoplasia. Gynecol

9. Kamariah K Satgunasingam N. Nasn N, Ng K. Hydatidifm mole and ptcvacuation regression p a t t m s of serum beta human chorionic gonadobophin. Med J Malaysia 1993; 48: 4045 .

10. Yedema KA. Verheijen RH. Kenemans Pet al. Identification of patients with persistent trophoblastic disease by means of a normal human chorionic gonadotrophin regression curve. Am J Obstet Gynecol 1993: 787-792.

11. Schlaenh JB. Morrow P. Kletzky OA. Nalick R. D‘Ablaing GA. Prognostic characteristics of Serum human chononic gonado- trophin titer regression following molar pregnancy. Obstet Gynecol 1981; 58: 478-482.

12. Berkowitz RS. Goldstein DP. Bemstein MR. Methotrexate infusion and folinic acid in the primary therapy of nonmetartatic gestational trophoblastic tumors. Gynecol Oncol 1990: 3 6 56-59.

13. Ayhan A, Yuce E, Yapar EG, Kisnisci AH. Effects of prophylactic chemotherapy for postmolar trophoblastic disease in patients with complete hydatidiform mole. Int J Gynecol Obstet 1990: 32: 39-4 1.

Onc01 1993; 48: 139-147.