Use of CA-125 To Predict Survival of Patients With Ovarian Carcinoma Gordon J.S. Rustin, Jeffery N. Gennings, Ann E. Nelstrop, Huguette Covarrubias, Hannah E. Lambert, and Kenneth D. Bagshawe for the North Thames Cooperative Group The prognostic value of serum CA-125 measurements was assessed in 54 patients with advanced ovarian adenocarcinoma. All patients received a minimum of two courses of carboplatin as part of the North Thames Cooperative Group trial. With a minimum follow-up of 6 months, 37 patients (69%) have clinical evidence of progressive disease and 28 have died. The absolute prechemotherapy level of CA-125 was of no value in predicting which patients would develop progressive disease. However, the change in CA-125 levels from before chemotherapy to 1 month later, after one course of carboplatin, could be used to divide patients into different prognostic groups. The best discrimina-

MOST

PATIENTS with advanced epithelial ovarian carcinoma receive several months of aggressive chemotherapy, yet fewer than 20% remain relapse-free for more than 3 years.1-3 Much unnecessary drug toxicity could be avoided if clinicians were aware at an early stage of treatment which patients are likely to benefit from continuing the same therapy, and which should be offered alternative or symptomatic therapy. Because it is so difficult to monitor the response to therapy of ovarian carcinoma a serum tumor marker has great potential. Bast et al4 developed a monoclonal antibody OC125 that detects an antigen determined in peripheral blood, designated CA-125. This serum tumor marker is elevated in more than 75% of patients with epithelial ovarian carcinoma and its level has been shown to fall in most patients whose tumor is responding to therapy. 4"9 CA-125 was assayed both before the first course of carboplatin and 1 month later before the second course. It was then determined whether the change in CA-125 levels between these two time points correlated with progression-free survival. Finally, the question whether there was a critical rate of fall of CA-125 below which prolonged survival was unlikely regardless of continuing therapy was investigated. PATIENTS AND METHODS Patients with International Federation of Gynecology and Obstetrics (FIGO) stage III or IV ovarian adenocarcinoma who were within 1 month of their diagnostic operation were

tion was found by dividing the patients into those who showed a greater than sevenfold decrease in CA125 levels and those who showed a smaller change. Eight of 14 (58%) patients with a greater than sevenfold decrease in CA-125 levels remain diseasefree compared with three of 36 (9%) patients with a lesser fall (P = .0005). The change in CA-125 levels during the first month of chemotherapy may indicate which patients should be offered alternative or symptomatic therapy and which should continue with the currently available toxic chemotherapy. J Clin Oncol 7:1667-1671. @ 1989 by American Society of Clinical Oncology. eligible for entry into the current North Thames Cooperative Group Ovarian trial. The intended therapy was 5 monthly courses of carboplatin 400 mg/m 2 followed by a second-look laparoscopy or laparotomy. Patients with less than 2 cm maximum diameter residual disease were then randomized into two arms, receiving either whole abdominal irradiation or five further courses of carboplatin. Between April 1985 and February 1987 141 patients were entered into the study. The median time from start of treatment to date of analysis was 27 months with minimum of 6 months. It was intended that after obtaining informed consent, serum would be collected and stored at -20 0 C from every patient before each course of chemotherapy. Samples were collected immediately prior to the first and second courses of chemotherapy from only 54 patients. The remaining patients started chemotherapy before a serum sample was collected for storage. Only the 54 patients from whom samples were obtained were studied for this analysis. Three patients had their first sample taken within 14 days postsurgery, 8 patients within 14 to 21 days postsurgery, and 43 patients had their From the Cancer Research Campaign Laboratories,Departmentof Medical Oncology, CharingCross Hospital,and the Department of Radiotherapy and Oncology, Hammersmith Hospital, London; and the Regional Radiotherapy Centre,Mount Vernon Hospital,Northwood Middles, United Kingdom. See Appendix for complete list of participantsin the North Thames Cooperative Group. Submitted February6, 1989; accepted June 26, 1989. Supported in part by grants from the DHSS and by the CancerResearch Campaign. Address reprint requests to Gordon J.S. Rustin, MD, Cancer Research Campaign Laboratories, Department of Medical Oncology, Charing Cross Hospital,Fulham Palace Road, London W6 8RF, United Kingdom. © 1989 by American Society of ClinicalOncology. 0732-183X/89/0711-0019$3.00/0

Journalof Clinical Oncology,Vol 7, No 11 (November), 1989: pp 1667-1671

Downloaded from ascopubs.org by 37.44.207.153 on January 26, 2017 from 037.044.207.153 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

1667

RUSTIN ET AL

1668 sample collected more than 21 days after surgery. Until the time of tumor progression all 54 patients were treated according to protocol except for two patients who refused a second-look operation. To date 37 (69%) have clinical evidence of tumor progression and 28 have died. Samples were assayed in batches using a one-step solidphase enzyme immunoassay kit (Abbott Laboratories Ltd, Maidenhead, Berks, United Kingdom). The persons performing the assays had no access to clinical information on the patients. The relationships between various factors and progressionfree survival were studied by univariate analysis using the Mantel-Cox test from the Biomedical Data Processing (BMDP) program." These factors included the initial level of CA-125, the rate of change in level of CA-125 over the first month of therapy, the extent of residual disease after primary surgery, and the histological grade. The date of progression was defined as the first date when there were clinical signs, or radiological or surgical evidence of tumor progression. Symptoms or changes in tumor marker levels were not used to define progression.

within the normal range appeared to have a slightly better median progression-free survival in comparison to patients with higher levels, but there was no statistical difference by log-rank test (P = .9774) between the progression-free curves (Fig 1). There was a highly significant relationship between the progression-free survival time and the change in CA-125 levels from just before the first course of carboplatin to just prior to the second course of carboplatin 1 month later. The Kaplan-Meier progression-free survival at 2 years of the 10 patients who had better than a 10-fold fall in CA-125 levels was 67% compared with 17% in those who had less than a 10-fold but greater than 50% fall, 9% in those who had a less than 50% but greater than 20% fall, and 0% in those who had less than a 20% fall or rise (Fig 2) (P = .001). Further analyses were carried out by dividing the patients into two groups according to the rate of fall of CA-125, using various cutoffs to find the point that gave the greatest difference between the two groups. Optimal discrimination was produced by comparing the 14 patients who had more than a sevenfold fall in CA-125 levels

RESULTS

Forty-one of the 54 patients (76%) had elevated levels of CA-125 (normal, < 35 U/mL). No relationship was found between initial absolute levels of CA-125 and progression-free survival when analyzed using Spearman's rank correlation coefficient. Patients with initial levels 100

a

-i

35 100 1000 U/ml

L. 5n.

0 c-) 0 D_ C)

p value 40

0.3318

5U)

U,

o U) 5-

20

U)

0

0

3

6

12

18

24

MONTHS Fig 1.

CA-125 initial marker levels (progression-free survival).

Downloaded from ascopubs.org by 37.44.207.153 on January 26, 2017 from 037.044.207.153 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

(10)

1669

PREDICTION OF SURVIVAL BY CA-125

100

-

-BI, L-

0

X-·

80

>1 log fall

(10)

>50%

0. (.. 'p

20

0 0

6

3

12

1B

24

MONTHS Fig 3.

Fall in CA-125 levels caused by one course of JM8 (progression-free survival).

prognosis. Despite adverse prognostic factors such as inoperable stage IV disease, the occasional patient still survives several years after chemotherapy. Such a patient, according to this study, is very likely to have had a greater than sevenfold fall in CA-125 levels following the first course of chemotherapy or to have had an initial level of less than 10 U/mL, and would therefore be likely to benefit from further chemotherapy. Conversely, the majority of patients who derive only a short or no survival benefit from chemotherapy could be considered for alternative or symptomatic therapy if they had a less than sevenfold fall in CA-125 levels. It would be improper to rely only on changes in CA-125 levels and automatically stop therapy in this poor-prognostic group because 9% of patients are still predicted to be progression-free at 2 years. Many studies have shown that responders to chemotherapy survive longer than nonresponders. The speed of response measured clinically or by scans recently has been shown to be a significant predictor of survival with certain cisplatin-based regimens.1 1 Changes in CA-125 levels on therapy also have been shown to correlate well with response to therapy.4 9 The value of these serial

measurements is to demonstrate which patients are responding at an earlier time than is otherwise possible in most patients with ovarian adenocarcinoma. If serum samples are taken too soon after major surgical tumor reduction, any noted fall in CA-125 level by 1 month after starting chemotherapy could be due to the surgery rather than the chemotherapy. However, it is probable that a change in CA-125 levels after surgery is similar to a change after chemotherapy in predicting survival. Unfortunately, we did not have preoperative samples to confirm this. Only two patients with CA-125 levels greater than 10 U/mL had serum samples taken within 14 days of surgery and no patient in the present study had complete macroscopic tumor clearance. With a natural CA- 125 half-life of less than 5 days, it is probable that most of the noted decrease in levels were related to the chemotherapy rather than the surgery. The current study is only of a preliminary nature in view of the relatively small number of patients and the fact that the end point used was progression-free survival and not actual survival. However, in patients with ovarian adenocarcinoma there is a very close correlation between

Downloaded from ascopubs.org by 37.44.207.153 on January 26, 2017 from 037.044.207.153 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

PREDICTION OF SURVIVAL BY CA-125

1671

progression-free survival and actual survival.' Several previous studies have tried to relate CA-125 levels with survival. Most reports suggest that a preoperative high serum CA-125 level correlates poorly with survival, probably because this level can be greatly affected by the success of surgery.6,7,12 Similarly, it is likely that variations in the percentage of patients with bulky residual tumor in different studies accounted for our not finding an inverse correlation between prechemotherapy CA-125 levels and prognosis, unlike three previous studies.6,'13,14 Support for our major findings come from a recent report suggesting that the half-life of CA-125 on chemotherapy is an independent prognostic variable.13 The present study provides the most significant prediction of progression-free survival with two serum samples

over the first month of chemotherapy. A study of a far larger group of patients has been set up by the Medical Research Council to determine whether prognostic factors including CA-125 levels could be used confidently to predict survival and thus lead to changes in patient management. ACKNOWLEDGMENT We thank Ken MacRae for expert statistical advice, Pamela Davis for her secretarial assistance, Alison Newham for help in performing the CA-125 assays, and Drs M. Anderson and J. Pryse-Davies for pathological review. We are also grateful to Bristol Myers who made carboplatin available prior to it being licensed and who funded our data collector and secretary, and to Unilever who subsidized our drug costs and to the DHSS who paid for the assay kits.

APPENDIX Participating members of the North Thames Cooperative Group: Mary Anderson; T. Anderson; Dr R. Ashford; J.M. Baldwin; R.M. Ballard; Professor R. Beard; Dr R.H.J. Begent; Professor R. Berry; Dr P. Blake; Dr K. Bragman; R.M. Burton; Dr S. Chapman; R. Clements; A.P. Condie; Dr C.A. Coulter; Dr S. Dische; J. Dowdell; M. Dunstan; Dr A. Epenetos; R.J. Etheridge; Dr D. Fermont; A. Fisher; J. Fox; A. Fraser; Dr M. Glaser; Dr R. Glyn-Jones; H. Gordon; W. Gregory; Dr E. Grosch; Professor D. Hawkins; Dr N. Howard; M. House; J. Irwin; Dr A. Jelliffe; Dr H. Lambert; F. Loeffler, Dr J. Maher; P. Mason; Dr M.I. Middleton; Dr A. Munro; Dr G. McKenzie; Dr N. McWhinney; Dr E. Newlands; Dr A. Nelstrop; J. O'Sullivan; P. Pallewela; Dr A. Parker; Dr R. Phillips; G. Pinker; D.W. T. Roberts; P. Russell; Dr G. Rustin; Dr M. Saunders; Dr R. Shannon; M. Simanowitz; F. Simpson; Dr B. Southcott; P. Soutter; Dr M. Spittle; Dr S.J. Steele; B. Webb; J. Wingfield; M. Witt; R.W. Worth; J. Wright; and G. Nys.

REFERENCES 1. Lambert HE, Berry RJ: High-dose cis-platin compared with high-dose cyclophosphamide in the management of advanced epithelial ovarian cancer (FIGO stages III and IV): Report from N Thames Cooperative Group. Br Med J 290:889-893, 1985 2. Rustin GJS, Minton M, Southcott B, et al: Surgery, chemotherapy and whole abdominal radiotherapy in the mangagement of advanced ovarian carcinoma. Clin Radiol 38:269-272, 1987 3. Neijt JP, ten Bokkel Huinink WW, van der Burg MEL, et al: (CHAP-5 v CP) in advanced ovarian carcinoma. J Clin Oncol 5:1157-1168, 1987 4. Bast RC Jr, Klug TL, St John E, et al: A radioimmunoassay using a monoclonal antibody to monitor the course of epithelial ovarian cancer. N Engl J Med 309:883-887, 1983 5. Bast RC Jr, Hunter V, Knapp RC: Pros and cons of gynecologic tumor markers. Cancer 60:1984-1992, 1987 6. Vergote IB, Bormer OP, Abeler VM: Evaluation of serum CA-125 levels in the monitoring of ovarian cancer. Am J Obstet Gynecol 157:88-92, 1987 7. Lavin PT, Knapp RC, Malkasian G, et al: CA-125 for the monitoring of ovarian carcinoma during primary therapy. Obstet Gynecol 69:223-227, 1987 8. Canney PA, Moore M, Wilkinson PM, et al: Ovarian cancer antigen CA-125: A prospective clinical assessment of its role as a tumour marker. Br J Cancer 50:765-769, 1984

9. Alvarez RD, To A, Boots LR, et al. CA-125 as a serum marker for poor prognosis in ovarian malignancies. Gynecol Oncol 26:284-289, 1987 10. Timothy AR: Cost versus benefit in non-surgical management of patients with cancer. Br Med J 297:471-472, 1988 11. Lawton FG, Kelly KA, Cassia LJS, et al: Speed of response to platinum-based chemotherapy: Implications for the management of epithelial ovarian cancer. Eur J Clin Oncol 23:1071-1075, 1987 12. Cruickshank DJ, Fullerton WT, Klopper A: The clinical significance of pre-operative serum CA-125 in ovarian cancer. Br J Obstet Gynaecol 94:692-695, 1987 13. van der Burgh MEL, Lammes FB, van Putten WLJ, et al: Ovarian cancer: The prognostic value of the serum half-life of CA-125 during the induction of chemotherapy. Gynecol Oncol 30:307-312, 1988 14. Parker D, Patel K, Aired EJ, et al: CA-125 and survival in ovarian cancer: Preliminary communication. J R Soc Med 81:22, 1988 15. Benedetti J, Yuen K, Young L: Life tables and survival functions, in Dixon WJ (ed): BMDP Statistical Software Manual. Berkeley, CA, University of California Press, 1985, pp 557-595

Downloaded from ascopubs.org by 37.44.207.153 on January 26, 2017 from 037.044.207.153 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.