Prognostic Value of Chemotherapy-Induced Amenorrhea in Breast Cancer : a Meta-Analysis

Breast cancer is the most frequently diagnosed cancers and the leading causes of cancer death in women , and approximately one-third of breast cancer cases are expected to occur in women <50 years of age (Siegel et al., 2015). Chemotherapy has been the mainstay of adjuvant therapy for premenopausal women with node-positive breast cancer, and its effect was much higher in younger women than in older women (Yazilitas et al., 2015). The benefit of adjuvant chemotherapy has been supposed that at least part of it is derived from chemotherapyinduced ovarian ablation. Although chemotherapy and ovarian function suppression independently improve the outcome of breast cancer (EBCTCG., 2005), there is controversy about the Prognostic role of chemotherapyinduced amenorrhea (CIA) in breast cancer. Recently, a comprehensive review published in 2006 discussed the incidence of amenorrhea with various variables, its impact on survival, and dilemmas in interpreting conflicting data on CIA (Walshe et al., 2006). CIA causes a variety of climacteric problems, including hot flush, genitourinary dysfunctions, and psychological distress (Leining et al., 2006). So, a comprehensive understanding of the impact of CIA in premenopausal women is critical. Due to many discrepancies across studies including the definition of


Introduction
Breast cancer is the most frequently diagnosed cancers and the leading causes of cancer death in women , and approximately one-third of breast cancer cases are expected to occur in women <50 years of age (Siegel et al., 2015).Chemotherapy has been the mainstay of adjuvant therapy for premenopausal women with node-positive breast cancer, and its effect was much higher in younger women than in older women (Yazilitas et al., 2015).The benefit of adjuvant chemotherapy has been supposed that at least part of it is derived from chemotherapyinduced ovarian ablation.Although chemotherapy and ovarian function suppression independently improve the outcome of breast cancer (EBCTCG., 2005), there is controversy about the Prognostic role of chemotherapyinduced amenorrhea (CIA) in breast cancer.Recently, a comprehensive review published in 2006 discussed the incidence of amenorrhea with various variables, its impact on survival, and dilemmas in interpreting conflicting data on CIA (Walshe et al., 2006).CIA causes a variety of climacteric problems, including hot flush, genitourinary dysfunctions, and psychological distress (Leining et al., 2006).So, a comprehensive understanding of the impact of CIA in premenopausal women is critical.Due to many discrepancies across studies including the definition of

Literature search and selection
We conducted a systematic literature search of PubMed and EMBASE through January 2014 by using the following key words: breast cancer, breast neoplasm, chemotherapy, drug therapy, prognostic, prognosis, and amenorrhea.References from identified studies were also reviewed.The search was limited to English language.The eligible studies should meet the following criteria: i) only original papers evaluating the association between chemotherapy-induced amenorrhea and disease free survival (DFS) and/or overall survival (OS) were selected.ii) hazard ratio (HR) for OS or DFS according to CIA either had to be reported or could be calculated from the paper.iii) for publications reported in several, only the most recent one was considered.Abstracts were excluded because of insufficient data.

Data extraction
The following data from eligible publications was extracted respectively by two reviewers (Zha QB and Ren ZJ) with a standardized data collection form: first author's last name, year of publication, country of origin, sample size, study design, disease stage, chemotherapy regimen, median follow-up, median age, definition of CIA, main end point, adjustments for covariates, the hazard ratios (HRs) and corresponding 95% confidence intervals (CIs).If HRs and 95%CIs were not provided directly, estimated value was obtained indirectly by using the methods described by Tierney et al (2007).Disagreements were resolved by discussion.

Statistical analysis
The primary outcome of our meta-analysis was disease-free survival (DFS), because the sample size for OS was relatively small.When univariate and multivariate analyses of DFS and/or OS were both available, the latter was selected to be combined.The subgroup analyses were performed according to definition of CIA, hormone receptor status, premenopausal status, median follow-up, region, tamoxifen use or not, chemotherapy regimen and study design.Heterogeneity between trials was evaluated by chi-square (x 2 ) test and I-squared (I 2 ) statistic (Higgins JP et al., 2003).Statistical heterogeneity was considered significant if p value less than 0.10 for the x 2 test or I 2 50%.We used a random-effects model to pool HRs.Publication bias was assessed by Begg's funnel plots and Egger's regression test (Begg et al., 1994).Additionally, We also conducted a sensitivity analysis to investigate the influence of a single trial on the overall risk (Tobias, 1999).All reported P values were two-sided and P values less than 0.05 were regarded as statistically significant.Statistical analyses were carried out using STATA 12.0 (Stata Corporation, USA).

Study characteristics
The characteristics of these 11 studies are listed in Table S1.Of these, 4 studies were conducted in North America, 4 in Europe, 2 in Korea, and 1 was a multination study.The sample size of the included studies ranged from 221 to 2341 patients.The percentage of patients becoming amenorrheic ranged from 51% (Budman et al., 1998) to 86% (Colleoni M et al., 2006).According to the definition of CIA, 5 of the 11 eligible studies defined it as cessation of menses for three months, 4 defined it as cessation of menses for six months, 1 defined it as no menses within 15months, and 1 defined it as no menses within one year of treatment.Seven earlier trials did not use tamoxifen as adjuvant hormonal therapy, while others did.All of the included patients were stages Ⅱ-Ⅲ.The median time of follow-up ranged from 4 to 9 years.HRs for DFS and OS could be extracted from 11 and 5 of the studies, respectively.Most of the survival data were adjusted for a wide range of potential confounders including age, tumor size, HR status, node status and treatment regimen.Detailed information from each trial was listed in Table S1.

Subgroup analysis
According to the definition of CIA, 5 studies defined it as cessation of menses for three months, and the pooled HR of these trials for DFS showed no statistical significance (HR =0.74, 95%CI 0.55-0.99).Furthermore, 4 trials (Tormey et al., 1992;Jung et al., 2010;Swain et al., 2010;Park et al., 2012) reported HRs for DFS in the CIA≥6 months subgroup.The pooled HRs favoured the CIA group (HR=0.70,95%CI 0.60-0.81).However, only three in the CIA≥6 months subgroup reported HR for OS , and the pooled HR was 0.72 (95%CI 0.62-0.85),which corresponds to a 28% reduction in the risk of death compared with control group.

Publication bias
Visual assessment of the funnel plot did not provide evidence of publication bias for studies in the DFS outcome.The Begg rank correlation test and Egger linear regression test also indicated no evidence of publication bias (Begg,p=0.66;Egger,p =0.57).

Discussion
There is still a great deal of controversy with regard to the role of CIA in predicting prognosis and selecting chemotherapy regimen in breast cancer patients.To confirm whether CIA can serve as a useful factor in predicting clinical effects of systemic adjuvant chemotherapy, we undertook a systematic review and meta-analysis of the literature.The results of our meta-analysis showed that CIA was correlated with significant reduction in the risk  doi.org/10.7314/APJCP.2015.16.14.5939 Prognostic Value of Chemotherapy-Induced Amenorrhea in Breast Cancer: a Meta-Analysis of recurrence (HR=0.74, 95%CI 0.64-0.86).When the five studies reported the HR of overall survival were pooled, a favorable trend was found (HR=0.69,95%CI 0.52-0.91).However, a significant heterogeneity was noticed (I 2 = 51.6%).When one study by Parulekar et al. was excluded, the heterogeneity dropped.Moreover, the pooled HRs for DFS and OS all favored CIA when it definted as cessation of menses for six months with no heterogeneity (I 2 =0.0%).However, the benefit of CIA on OS was uncertain because of less available trials reporting HRs.
The results of subgroup analysis also indicated that there were significantly gains in DFS and OS in the patients with hormone receptor-positive status, but not in the patients with hormone receptor-negative status.These results suggested that benefits of chemotherapy were partially attributed to an indirect endocrine effect in hormone-sensitive tumors.
When we pooled the HR for overall survival of five studies, a significant heterogeneity was found (I 2 = 51.6%).When one study by Parulekar et al was excluded, the heterogeneity declined.The heterogeneity was probably due to the difference in defining amenorrhea and menopause, duration of amenorrhea, the duration of follow-up, chemotherapeutic regimen and characteristics of patients (age, tumor stage, hormone receptor status, race or country).For instance, when we stratified them according to the definition of CIA, Strong heterogeneity existed in the CIA ≥3 months subgroup analysis for DFS.Lack of standardized data collection including bleeding history and long menstrual cycle may partially explain this heterogeneity.In addition, different detecting methods for hormone receptor may also contribute to the heterogeneity in hormone receptor-positive subgroup (I 2 = 41.3 % for DFS; I 2 = 43.5% for OS).Taking heterogeneity into consideration, we used a random-effects model for more conservative estimates.Nevertheless, there is no conclusive interpretation for the heterogeneity.
Potential limitations of this study should be considered.First, the characteristics of these trials including the defining amenorrhea and menopause, duration of amenorrhea, the duration of follow-up, and characteristics of patients were varied.Second, our study was based on abstracted data and not on individual patient data, which may not provide robust estimation for the HRs and 95%CIs.Third, the quality of eligible studies influences the reliability of our meta-analysis remarkably.Published articles often provide insufficient information to assess the quality of the study.Fourth, publication bias may be unavoidable in this article.However, little evidence of publication bias was observed.Five, although most of studies adjusted for a series of key covariates, we still cannot rule out the possible residual confounding variables.Six, the total number of included studies and the total sample size were relatively small, which may influence the accuracy of meta-analysis to some extent.Seven, most chemotherapy regimens were CMF-based regimens, although, the incidence of amenorrhea with anthracycline-taxane-containing chemotherapy regimens and its impact on survival were similar.Finally, the majority of studies were carried out in western countries, and additional research in other populations is warranted to generalize the findings.
Our study still has important clinical implications.It showed that premenopausal patients developing amenorrhea during chemotherapy had a significantly better DFS and OS compared to those who did not.Moreover, the subgroup analysis demonstrated that CIA was correlated with longer DFS and OS in hormone receptor-positive subgroup.However, the correlation between CIA and OS need to be investigated, because of limited data.These results strongly suggest that endocrine manipulation through ovarian suppression is another mechanism of chemotherapy in addition to direct cytotoxicity in women with hormone-sensitive disease.Ovarian suppression, induced either by surgery, or LH-RH superanalogues therapy, has been shown to be an effective treatment in breast cancer patients (Early Breast Cancer Trialists' Collaborative Group., 1992).On this basis, we can suppose that the better outcome of patients with CIA is related to the combined action of two effective treatments: chemotherapy and ovarian ablation.The hypothesis of a double effect, cytotoxic and endocrine, could explain the greater impact of chemotherapy generally observed in premenopausal women with hormone-sensitive breast cancer (Li Xiu-Juan et al., 2014).This inclusion was indirectly supported by two large randomized prospective trials (Jakesz et al., 2002;Jonat et al., 2002).Because of serious climacteric disturbances such as vasomotor symptoms and urogenital atrophy in young women, physicians should avoid artificially inducing amenorrhea in premenopausal patients without selection (Yu Ke-Da et al., 2014).In conclusion, a comprehensive understanding of the impact of CIA on survival in premenopausal breast cancer patients is critical (Zhao Jianli et al., 2014).In order to draw definite conclusions, multicenter randomized prospective trial of adjuvant chemotherapy should evaluate CIA using exact definitions of CIA and premenopausal status.J Natl Cancer Inst, 90, 1205-11. Campora E, Pronzato P, Amoroso D, et al (1992)

Figure 1 .
Figure 1.The Association between CIA and Disease-free Survival (DFS) of Breast Cancer Stratified by the Definition of CIA.The summary HR and 95% CIs were shown (from the random-effects model)

Figure 2 .
Figure 2. Forest Plot Showing the Combined HR from the Random-effects Model for OS Grouped by the Definition of CIA