|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Originally published as JCO Early Release 10.1200/JCO.2007.11.6061 on December 17 2007 © 2008 American Society of Clinical Oncology. Survival After Adjuvant Oophorectomy and Tamoxifen in Operable Breast Cancer in Premenopausal Women
From the Comprehensive Cancer Center and The Center for Biostatistics, The Ohio State University, Columbus, OH; Hospital K, National Cancer Institute, Hanoi; Danang General Hospital, Danang; and Hue Central Hospital, Hue, Socialist Republic of Vietnam; People's Hospital of Haimen City, Haimen, Jiangsu, People's Republic of China Corresponding author: Richard R. Love, MD, The Ohio State University, B402 Starling Loving Hall, 320 W 10th Ave, Columbus, OH 43210; e-mail: Richard.Love{at}osumc.edu
Purpose Worldwide, approximately 750,000 new cases of breast cancer are diagnosed annually in premenopausal women with limited economic resources. Longer-term survival benefits from adjuvant therapies in such women with operable breast cancer are unknown. Patients and Methods From 1993 to 1999, we recruited 709 premenopausal women with operable breast cancer to a multisite randomized clinical trial of adjuvant oophorectomy and tamoxifen for 5 years or observation and this combined hormonal therapy on recurrence. Results With a median follow-up of 7.0 years, disease-free and overall survival were significantly improved with the adjuvant treatment (log-rank P = .0003 and .0002, respectively). Five year disease-free survival (DFS) probabilities of 74% and 61% (95% CI for difference, 7% to 21%) and overall survival (OS) rates of 78% and 71% (95% CI for difference, 1% to 21%) were observed in the adjuvant and observation groups. Ten-year DFS probabilities of 62% and 51% (95% CI for difference, 4% to 22%) and OS probabilities of 70% and 52% (95% CI for difference, 6% to 34%) between adjuvant and observation groups, respectively, were observed. In the subset of estrogen receptor–positive patients, 5-year DFS probabilities were 83% and 61%, and 10-year DFS probabilities were 66% and 47%, while 5-year OS probabilities were 88% and 74%, and 10-year OS probabilities were 82% and 49% in the adjuvant and observation groups, respectively. Conclusion In premenopausal women with operable breast cancer not selected for estrogen receptor status or with estrogen receptor–positive tumors, 5- and 10-year DFS and OS rates are significantly improved following adjuvant oophorectomy and tamoxifen.
While in Western countries, the breast cancer case burden is mostly in postmenopausal women, in resource-poor countries, the distribution is reversed.1 With rising populations, estimates state that within the next few years, almost half (750,000) of all annual new cases will occur in premenopausal women in non-Western poorer countries (projections using data in Parkin et al1). For the majority of these women, absence of medical resources and limited economic resources, combined with competing medical conditions, make systemic cytotoxic chemotherapy treatments unobtainable or impractical. Clinical trials of adjuvant therapies for breast cancer have been done almost exclusively in Western or affluent patient populations. Additional issues concerning frequency of estrogen receptor–positive (ER+) tumors in Asian women, and the uncertain status of combined oophorectomy/tamoxifen therapy, as reviewed in the first publication of results from the currently updated clinical trial, led to the design and conduct of a trial of adjuvant surgical oophorectomy and tamoxifen in ER+ women with operable breast cancer in Vietnam and China.2 In this article we report the mature results of the original study and focus on the differential effects of treatment by ER status.
Details of patient recruitment and entry criteria have been previously published.2 In the current report, we used data from an extended follow-up of patients. Patients were scheduled visits every 3 to 6 months for the first 5 years after random assignment, and annually thereafter. Tamoxifen was provided at each visit for the first 5 years. This extension increased the median follow-up time to 7.0 years, as compared with 3.6 years in the original report.2 For patients without events, median follow-up was 7.9 years (first quartile, 6.6 years; third quartile, 9.5 years). Overall survival (OS) times were calculated from the date of random assignment until the date of death from any cause, and disease-free survival (DFS) time was measured from the date of random assignment until the date of disease recurrence or death before recurrence. For the purposes of determining the disease-free date, the following conventions were used for those patients who died without known recurrence or demonstrable non–disease-related death. If the patient died within 6 months of last being seen and was known to have been disease-free, then that patient was censored at the time of death. If the patient died more than 6 months from her last visit, then that patient was censored for recurrence at her last visit date, rather than her death date. The majority of patients were clinically diagnosed on the basis of fine-needle aspiration biopsy of lesions, all greater than 2 cm. On subsequent pathology review, 34 patients (4.9%) were found to not have evidence of invasive breast cancer.2 In the majority of cases, surgical oophorectomy preceded mastectomy done under the same anesthesia, and thus, pathological surgical staging was not known at the time of random assignment.2 Hormonal receptor assays were done using specialized methods in the laboratories of D.C. Allred (Houston, TX) 2 to 7 years after tissue collection, using tumor paraffin blocks available for two thirds of study participants.2 Study sample size was based on estimates of a 50% to 55% 5-year DFS and a 60% to 65% OS in the observation patients, and on increases of 10% to 12% in disease-free survival and 8% to 9% in overall survival with the adjuvant treatment. A sample size of 700 was estimated to provide 82% to 94% power for the expected 5-year DFS difference and 72% to 82% power for the expected 5-year OS difference. Estimates of survival functions for DFS and OS were generated by the Kaplan-Meier method3 and compared via the log-rank test.4 CIs for estimates of the difference in survival probabilities between treatment groups at a specified time point were calculated.5 A univariate Cox proportional hazards model was fit to estimate a crude treatment effect. Multivariable analyses were conducted using Cox proportional hazards regression, testing an interaction of treatment with ER status. The assumption of proportional hazards was evaluated through tests of Schoenfeld residuals and by the interaction of covariates with the natural logarithm (log) of time.6,7 Alternative transformations of time for these interactions were explored by comparing model likelihoods. A kernel smoothing method utilizing the Nelson-Aalen estimates of the hazard function and an Epanechnikov kernel7 was used to obtain smoothed estimates for the hazard of recurrence in the ER+ subgroup. Bandwidth was determined using the general-purpose formula h = 0.9*min (SD, IQR/1.34)*n(–1/5), where "SD" and "IQR" are the standard deviation and interquartile range of the event time points.8 CIs for the smoothed hazard estimates were also computed as described in Klein and Moeschberger.7 Analyses were carried out in either SAS (version 9.1; SAS Institute, Cary, NC) or Stata (version 9.2; StataCorp, College Station, TX).
From early 1993 through mid-1999, 709 women were entered into this clinical trial at seven institutions, six of them in Vietnam. The current report is based on patient evaluations to September 1, 2006. As of this date, the median follow-up times are 7.3 (IQR: 4.4) and 6.5 (IQR: 4.9) years in adjuvant and observation groups respectively. A total of 16 patients were lost to follow-up before year two (8 adjuvant; 8 observed). Status of patients at enrollment and other critical administrative details have been presented in our previous report.2 Detailed patient and tumor characteristics have also been previously presented; a brief summary of major characteristics is presented in Table 1. 2 Randomized groups were similar for major prognostic factors. Of importance, pathologic tumor size averaged 3.3 cm, and 52% had axillary node-positive disease (mean number of positive nodes was 4.2). Age, proportions of patients younger than 40 years, clinical tumor size, pathologic tumor size, fractions axillary node-positive, and histologic grade were not statistically different between the two randomized groups. Pathologic tumor size was greater in the observation group (3.37 v 3.22 cm) and frequency of ER+ tumors was lower in the adjuvant group 56% v 68%. No patients developed or died of serious postoperative complications after oophorectomy surgery. No patients reported stopping tamoxifen therapy for symptoms.9 Compliance with tamoxifen was considered high because most patients returned to obtain this study treatment drug at 3- to 6-month intervals.
Among the 356 women in the adjuvant treatment group, there were 118 events and 101 deaths, and in the 353 women in the observation group, there were 160 events and 146 deaths. Of the 278 recurrences, 153 were distant, 79 were locoregional, and 46 were of unknown location. Thirty patients died with no evidence of recurrence at the last visit (15 in the adjuvant treatment group and 15 in the observation group). Of these 30, five died of causes unrelated to their diagnosed primary cancer with no sign of recurrence (three adjuvant; two observed), 16 died within 6 months of being seen and known to have been disease-free (seven adjuvant; nine observed), and nine died more than 6 months after being seen and were known to have been disease-free (five adjuvant; four observed). There were 13 patients (10 adjuvant; three observed) for whom no reliable DFS follow-up data were available, including eight (five adjuvant; three observed) who had no OS follow-up. These patients were considered censored at their enrollment date. In intent-to-treat analyses (using all registered cases), a significant difference in DFS and OS was found, favoring the treatment group (log-rank test P values = .0003 and .0002, respectively; Fig 1).
Five-year DFS was 74% in adjuvant, and 61% in observation groups (95% CI for difference, 7% to 21%) and 10-year DFS was 62% and 51%, respectively (95% CI for difference, 4% to 22%). Five-year OS rates were 78% and 71% (95% CI for difference, 1% to 21%) and 10-year rates were 70% and 52%, respectively (95% CI for difference, 6% to 34%). Based on a univariate Cox proportional hazards model, the hazard ratio for adjuvant versus observation groups was 0.65 (95% CI, 0.51 to 0.82) for DFS and 0.62 (95% CI, 0.48 to 0.80) for OS. Among the women with known estrogen receptor (ER) status (n = 470), the adjuvant treatment effect was more pronounced in ER+ women. A Cox proportional hazards model was fit for the DFS outcome using treatment, ER status, and the treatment by ER interaction as the predictors. Testing the proportionality of the effects via the scaled Schoenfeld residuals revealed some evidence of nonproportionality in the overall model (P = .005). Interactions between log time and each covariate were added to the model with the treatment by ER term exhibiting significance. The final model included treatment group, ER status, treatment by ER at baseline (Wald test P = .001) and a time-varying effect for treatment by ER in the natural log of time (Wald test P = .002). Additional prognostic factors (number of positive nodes, age, histologic grade, and pathologic size) did not meaningfully affect the coefficients and were not included. Estimates from the model indicated that the treatment benefit decreased over time for ER+ patients. Using the model estimates and ER+/observation patients as the referent group, the hazard ratio for recurrence for ER+ patients undergoing treatment increased from 0.49 (95% CI, 0.31 to 0.75) at year 3 to 1.10 (95% CI, 0.58 to 2.06) at year 8. Kaplan-Meier survival curves reveal that for ER+ women, 5-, 7-, and 10-year DFS probabilities for adjuvant-treatment patients were 83%, 82%, and 66%, respectively, compared with 62%, 56%, and 50%, respectively, for the observed patients (Fig 2A). For ER+ women, 5-, 7-, and 10-year OS probabilities were 87%, 84%, and 80%, respectively, for adjuvant treatment, and 76%, 65%, 51% in the observation group (Fig 2B). In consideration of the significant ER status by treatment interaction, the smaller effect of treatment for ER– patients was not significant (log-rank test P = .46 and .29 for DFS and OS, respectively). The 95% CI for the hazard ratio comparing treatment and observation among ER patients for DFS was 0.56 to 1.29, which is not narrow enough to rule out a clinically important effect of treatment in this subgroup.
Estimates of the hazard functions for recurrence and 95% point-wise CIs are shown in Figure 3 for ER+ patients. These estimates may be roughly interpreted as the instantaneous incidence rate of recurrence at a given time point. The nonproportionality of the hazard functions detected in the above model is clearly exhibited by the convergence of the two curves. The shape of the curves is also revealing of the pattern of recurrence. An apparent peak in the hazard of recurrence occurs at roughly year 2 in the observation patients. For treated patients, some evidence of an increase in the hazard exists after 6 years, though a fair amount of uncertainty in the estimates at later time points is reflected by the CIs.
In the ER+ subset of patients treated in this study, the 5-year DFS of 83% compares favorably with the 81% figure reported with luteinizing hormone-releasing hormone (LHRH) agonist for 3 years combined with tamoxifen for 5 years, in a similar mixed group of stage I and II operable patients.10 Similarly, 10-year percentage risk reductions (defined as the risk difference divided by the control risk) for recurrence and death reported with polychemotherapy and tamoxifen in premenopausal women with hormonal receptor–positive tumors of 0.53 and 0.50, respectively, are similar to the 10-year percentage risk reductions found in the current study of 0.32 and 0.59, respectively.11 Further in this subset, this combined oophorectomy/tamoxifen therapy is associated with higher 10-year percentage risk reductions than are reported for the tamoxifen-alone meta-analysis (0.27 and 0.27 in node-positive patients).11 That combined adjuvant hormonal therapy is more effective than tamoxifen alone is also supported by the results of the Intergroup 0140 study, in which the 5-year OS was 97.5% with oophorectomy plus tamoxifen, 95.1% for tamoxifen alone (not significant).12 Combined hormonal therapy was found to be more effective than LHRH alone in Eastern Cooperative Oncology Group trial EST 5188, where 9-year OS rates in axillary node–positive patients were 70%, 73%, and 76% for cyclophosphamide, doxorubicin, and fluorouracil (CAF) chemotherapy, CAF + LHRH for 5 years, and CAF + LHRH + tamoxifen treatments for 5 years, respectively.13 The general conclusion that combined hormonal therapy is more effective is also supported by the results of a randomized clinical trial of LHRH alone, tamoxifen alone, or the combination in women with metastatic breast cancer; combination hormonal therapy was associated with statistically better survival.14,15 Collectively these observations support the option of combined surgical oophorectomy and tamoxifen as an effective adjuvant therapy in premenopausal women with hormone receptor–positive tumors. What is unclear is the possible magnitude of greater benefit from newer regimens of chemotherapeutic agents (dose dense and with taxanes), combined with tamoxifen or ovarian ablation or suppression and tamoxifen in those women whose ovarian function remains or returns after chemotherapy. Further, the role of antibody therapy with trastuzumab in those patients whose tumors overexpress this oncogene is not well defined as the only effective option in the subset of women who have ER+ tumors. In another analysis of patients in this study, those with Her-2/neu–positive tumors had outcomes comparable to those with Her-2/neu–negative tumors.16 Of further interest in this trial are the data on DFS in ER+ tumor-bearing patients between years 5 and 10. Consistent with observations from the Early Breast Cancer Trialists' Collaborative Group meta-analysis, there was a significant decrease of 17% in the DFS probability between years 5 and 10.11 The decrease in DFS probabilities between years 0 and 5 (17%) was of similar magnitude. This is most clearly revealed in the recurrence hazard function plotted in Figure 3, which shows an increase in the post–5-year period. This observation is consistent with the pattern for recurrence of disease in hormone receptor–positive tumor-bearing (particularly axillary node–positive) patients reported by Saphner et al from Eastern Cooperative Oncology Group data as well as the bimodal distribution of recurrence suggested by Retsky et al.17,18 It is also remarkably consistent with recent data on untreated (after 5 years) hormone receptor–positive tumor-bearing patients reporting significant benefit from "late" adjuvant hormonal therapy with an aromatase inhibitor.19,20 In that study discussion, the authors made particular note of the pattern of increasing hazard for recurrence over 4 years (years 5 to 9 postdiagnosis) in the placebo-treated study subjects.19 These observations lend support for longer-term or post–5-year hormonal therapies, or perhaps other strategies, in women with such hormone-sensitive tumors. Finally, while determination of primary tumor receptor status followed by adjuvant chemotherapy in women with hormone receptor–negative tumors is clearly associated with improved outcomes in this group of patients, a number of barriers prevent this diagnostic and therapeutic approach in women in resource-poor circumstances. An approach supported by the results of the current study, is to do adjuvant surgical oophorectomy and provide tamoxifen to all women, when the usual biology-grounded strategy is not possible. The absolute 10-year OS benefit of 18% demonstrated here, in the face of remarkable cost efficacy should not be ignored.2 Denial of this less than optimal option is inappropriate when safe and complete course chemotherapy programs cannot be given and accurate tumor hormonal receptor testing cannot be accomplished. In summary, mature results from this trial of adjuvant surgical oophorectomy and tamoxifen conducted primarily in Vietnam demonstrates again that useful clinical research can be done in non-Western countries. The survival of adjuvant-treated patients in this trial, particularly the subset of those with estrogen receptor–positive tumors, supports use of this treatment in many circumstances. Finally, the increase in the hazard function for recurrence in adjuvant-treated hormone receptor–positive patients between years 5 and 10 is consistent with other data and supports investigation of intervention and treatment strategies directed to this hazard.
The author(s) indicated no potential conflicts of interest.
Conception and design: Richard R. Love Financial support: Richard R. Love Administrative support: Richard R. Love Provision of study materials or patients: Richard R. Love, Nguyen Van Dinh, Tran Tu Quy, Nguyen Dinh Tung, Tien-zhen Shen Collection and assembly of data: Richard R. Love, Nguyen Van Dinh, Tran Tu Quy, Nguyen Dieu Linh, Nguyen Dinh Tung, Tien-zhen Shen Data analysis and interpretation: Richard R. Love, Erinn M. Hade, Gregory S. Young, David Jarjoura Manuscript writing: Richard R. Love Final approval of manuscript: Richard R. Love, Nguyen Van Dinh, Tran Tu Quy, Nguyen Dieu Linh, Nguyen Dinh Tung, Tien-zhen Shen, Erinn M. Hade, Gregory S. Young, David Jarjoura
Supported by Grant No. CA64339 from the National Institutes of Health, Bethesda, MD, and by the International Breast Cancer Research Foundation, Madison, WI. The tamoxifen (Nolvadex) for this study was provided at cost by AstraZeneca Pharmaceuticals. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. Parkin DM, Pisani P, Bray F, et al: Estimates of the worldwide mortality from 25 cancers in 1990. Int J Cancer 83:18-29, 1999[Medline] 2. Love RR, Duc NB, Allred DC, et al: Oophorectomy and tamoxifen adjuvant therapy in premenopausal Vietnamese and Chinese women with operable breast cancer. J Clin Oncol 20:2559-2566, 2002 3. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457-481, 1958[CrossRef] 4. Peto R, Peto J: Asymptotically efficient rank invariant test procedures (with discussion). Journal of the Royal Statistical Society A 135:185-206, 1972[CrossRef] 5. Borgan Ø, Liestøl K: A note on confidence intervals and bands for the survival curve based on transformations. Scand J Stat 17:35-41, 1990 6. Schoenfeld D: Partial residuals for the proportional hazards regression model. Biometrika 69:239-241, 1982 7. Klein JP, Moeschberger ML: Survival analysis: Techniques for Censored and Truncated Data (ed 2). New York, NY, Springer-Verlag, 2003 8. Silverman BW: Density estimation for statistics and data analysis. New York, NY, Chapman and Hall, 1986 9. Love RR, Duc NB, Binh NC, et al: Symptoms associated with oophorectomy and tamoxifen in premenopausal Vietnamese women. Breast Cancer Res Treat 58:281-286, 1999[Medline] 10. Jakesz R, Hausmaninger H, Kubista E, et al: Randomized adjuvant trial of tamoxifen and goserelin versus cyclophosphamide, methotrexate, and fluorouracil: Evidence for the superiority of treatment with endocrine blockade in premenopausal patients with hormone-responsive breast cancer—Austrian Breast and Colorectal Cancer Study Group Trial 5. J Clin Oncol 20:4621-4627, 2002 11. Early Breast Cancer Trialists' Collaborative Group: Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: An overview of the randomised trials. Lancet 365:1687-1717, 2005[CrossRef][Medline] 12. Robert H, Wang M, Cella D, et al: Phase III comparison of tamoxifen versus tamoxifen with ovarian ablation in premenopausal women with axillary node negative receptor-positive breast cancer 13. Davidson NE, O'Neill AM, Vukov AM, et al: Chemoendocrine therapy for premenopausal women with axillary lymph node-positive, steroid hormone receptor-positive breast cancer: Results from INT 0101 (E5188). J Clin Oncol 23:5973-5982, 2005 14. Klijn JG, Beex LV, Mauriac L, et al: Combined treatment with buserelin and tamoxifen in premenopausal metastatic breast cancer: A randomised study. J Natl Cancer Inst 92:903-911, 2000 15. Klijn JGM, Blamey RW, Boccardo F, et al: Combined tamoxifen and luteinizing hormone-releasing hormone (LHRH) agonist alone in premenopausal advanced breast cancer: A meta-analysis of four randomized trials. J Clin Oncol 19:343-353, 2001 16. Love RR, Duc NB, Havighurst TC, et al: HER-2/neu over expression and response to oophorectomy plus tamoxifen adjuvant therapy in estrogen receptor-positive premenopausal women with operable breast cancer. J Clin Oncol 21:453-457, 2003 17. Saphner T, Tormey DC, Gray R; Annual hazard rates recurrence for breast cancer after primary therapy. J Clin Oncol 14:2738-2746, 1996 18. Retsky MW, Demicheli R, Swartzendruber DE, et al: Computer simulation of a breast cancer metastasis model. Breast Cancer Res Treat 45:193-202, 1997[CrossRef][Medline] 19. Goss P, Ingle JN, Martino S, et al: A randomized trial of letrozole in postmenopausal women after 5 years of tamoxifen therapy for early stage breast cancer. N Engl J Med 349:1793-1802, 2003 20. Ingle JM, Tu D, Pater JL, et al: Duration of letrozole treatment and outcomes in the placebo-controlled NCIC CTG MA. 17 extended adjuvant therapy trial. Breast Cancer Res Treat 99:295-300, 2006[CrossRef][Medline] Submitted March 12, 2007; accepted October 12, 2007.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2008 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
3 cm. Proc Am Soc Clin Oncol 22:5, 2003 (abstr 16)
