Prognostic and Clinicopathological Value of Cyclin E Expression in Breast Cancer: A Meta-Analysis

Background: Multiple studies have analyzed the correlation between Cyclin E and breast cancer prognosis, but the results are controversial. In this study, a meta-analysis was used to summarize the published reports, clarify the predictive value of Cyclin E in breast cancer, and the relationship with clinicopathological characteristics. Methods: A systematic search was retrieved in PubMed, Embase, Web of Science, Cochrane Collaboration Library on the comprehensive search strategy up to 24 August 2020. recurrence-free survival (RFS), overall survival (OS), and breast cancer-specic survival (BCSS) was estimated using pooled hazard ratios (HRs) and risk ratios (RRs) with 95% condence intervals (95% CIs) by univariate and multivariate analysis. Results: Twenty-eight eligible studies met our inclusion criteria. The present meta-analysis indicates that high cyclin E expression is independently associated with poor OS, BCSS, and RFS in female breast cancer patients by univariate and multivariate analysis. Furthermore, higher histological grades, estrogen receptor (ER)-negativity, positive lymph node metastasis, younger patients, and premenopausal status were associated with Cyclin E overexpression. Conclusions: High expression of Cyclin E is associated with poor breast cancer outcomes and relevant to multiple clinical characteristics.


Search strategy
The search strategy included the Mesh terms 'breast cancer' AND 'Cyclin E' AND 'prognosis/survival/prognostic/outcome' with all appropriate subheadings contained.An additional le shows this in more detail [see Additional le 1].Besides, we manually retrieved references in the included study to identify the need for inclusion in the study.

Inclusion and exclusion criteria
Inclusion criteria: (1) full-text publication; (2) Studies on the correlation between Cyclin E and prognosis of breast cancer; (3) The subjects were female breast cancer patients diagnosed by pathology; (4) The HR or RR and its 95% CI or its corresponding survival curve were available for data extraction.Exclusion criteria were summarized as follows: (1) Systematic evaluation, meta-analysis, case reports, letters, and comments were excluded; (2) Non-human studies.; (3) The full text of the literature could not be obtained, or the data report was incomplete, or the HR or RR and its 95% CI could not be extracted; (4) Non-English literature.

Data extraction
The literature selection process is given in Fig. 1.Two independent researchers (Wang YF and Yang QJ) independently screened the literature and extracted the data depending on the inclusion and exclusion criteria.Any differentials in their assessment were resolved by discussion.
Various contents were extracted, including the rst author, number of patients, publication year, age, AJCC stage, histological grade, Detection Method, cut-off value, follow-up period.The Kaplan Meier survival curves were read by Engauge Digitizer version 4.1 (http://digitizer.sourceforge.net/).

Statistical analysis
In this meta-analysis, HR or RR was used to measure the correlation between Cyclin E and the prognostic factors.Meanwhile, we also analyzed the relationship between Cyclin E and the clinicopathological characteristics.RR value and 95% CI were used for assessment.Heterogeneity among studying data was evaluated by χ 2 -based Q-tests and I 2 tests.The random-effect model was used only for high heterogeneity (I 2 > 50% or P < 0.01); otherwise, xed-effect model was used [16].Review Manager, version 5.3 (The Cochrane Collaboration, Copenhagen, Denmark) was used to conduct the meta-analysis.An HR or RR > 1 indicated a worse outcome for the positive group, and a P-value < 0.05 throughout was considered statistically signi cant.Publication bias was assessed by funnel plots [17,18].

Search results
A total of 1244 references were retrieved through systematic literature retrieval, of which 476 were duplicated, two were not available, and one was not in English.Through screening the titles and abstracts, 16 studies were non-clinical, 684 articles were unrelated to the study, 27 studies were guides, reviews, letters, or abstracts.After reading the full text, survival data could not be extracted from 9 articles.One article is updated.Finally, 28 studies were included [11][12][13][14].The selection ow chart is shown in Fig. 1.

Characteristics of eligible studies
The included literature was published between 1996 and 2020 and comprised 9289 patients (sample ranged from 56 to 2032 patients) with a median follow-up time ranging from 3 to 15 years.The studies were conducted in 13 countries, with ages ranging between 20 and 90 years (Table 1).Three methods were used to detect Cyclin E level: Western blotting (3 studies), IHC (20 studies), and RT-PCR (5 studies).The different cut-off values and different antibodies were documented by the studies.(Table 2).The prognostic indicators were: overall survival (OS), recurrence-free survival (RFS), and breast cancer-speci c survival (BCSS).In our meta-analysis, the results of DFS and MFS were integrated into RFS, and the effect of Cyclin E on prognosis was analyzed from univariate and multivariate situations.(Additional le 2).Recurrence-free survival was de ned as time to any type of recurrence or death from any cause.Overall survival was de ned from the beginning of the random assignment to death from any cause.Breast cancer-speci c survival was de ned as time to death from breast cancer.The quality of the eligible literature was assessed using the Newcastle Ottawa Scale (NOS) criteria for cohort studies [43].The scale evaluates articles from three aspects (selectivity, comparability, and exposure).A score of 7 or more is considered to be of high quality (Additional le 3).
Two studies were published on the relation between Cyclin E and breast cancer in 1996 and 1999, respectively [19,44].The latter one was an updated article, but they combined cyclin E with other factors when analyzing survival indicators, so we used the former, and only the 100 patients with stage I-III disease were included in the survival analysis.There were two studies [25,26] based on the same patient cohort.Nevertheless, Kuhling et al.'s paper contained 59 patients who received anti-hormonal therapy, causing the survival data to be different, so both of them were included in the analysis.A similar situation occurred in another three articles [27,30,31], with varying amounts of sample leading to different data.Potemski et al. wrote two articles in 2009 and2006 [36, 45], respectively.The paper of 2009 included larger samples from 1997 to 2001, so the 2006 paper was excluded.However, the relationship between Cyclin E and some clinicopathological features was included in this article, so it was used to analyze the relationship between Cyclin E and clinicopathological characteristics.Since the human genome encodes two E-type cyclins: Cyclin E 1(CCNE1) and Cyclin E 2(CCNE2), CCNE1 is the formerly cyclin E, which is located on chromosome 19q12-q13 (46,47).Therefore, survival data from three articles [33,34,41] on CCNE1 were included in the study.shown in Fig. 2A.The multivariate analysis results were close to statistical signi cance (HR, 2.08, 95%CI, 0.91-4.78,P = 0.08), as shown in Fig. 2B.Heterogeneity was high for the results.Thus, a random-effects model was used.

Relationship between Cyclin E and clinicopathological features
We also investigated the relationship between Cyclin E and clinicopathological characteristics.3), the forest plots were shown in Additional le 4.
Any deletion of the literature in this study will not affect the results of this study, which means that the calculation results of the above random effect are stable and reliable.In the subgroup studies with more than ve included literature, funnel plots were used to evaluate publication bias (Fig. 5).As can be seen from the gures, funnel plots of each study were basically symmetrical, so it can be judged that there was no publication bias in this study.

Discussion
Breast cancer is a complex process involving multiple factors, one of the most critical aspects is the disorder of cell cycle regulation [5].Accurate prediction of prognosis is one of the most critical problems in breast cancer treatment.The prediction of breast cancer is closely related to the disease stage, and early diagnosis of the disease is crucial to the patient's prognosis.Once the risk of recurrence has been determined, it can be used to evaluate the potential effects of endocrine therapy, chemotherapy, or a combination of these treatments.The determination of accurate prognostic indicators will also in uence the patient's individualized treatment.Also, identifying prognostic factors with potential biological signi cance can be used as therapeutic targets, which will be of great relevance to identifying and treating high-risk patients.The cell cycle is divided into several stages arti cially, including G0/G1, S, G2, and M phase.Cyclins play different roles at different cell cycle stages depending on their sequence primitives, expression, and activation forms [48].Cyclins, along with the cyclin-dependent kinase (CDKs) and cyclin-dependent kinase inhibitors (CKIs), constitute the cycle regulation system to ensure the cell cycle orderly [49].The limitation point of transition from the G1 to the S1 phase is crucial for cell proliferation and differentiation [50].The primary function of Cyclin E is to regulate the restriction point "R" at the end of the G1-phase to allow cells to enter the S phase [51,52].The restriction point "R" was found to be the endpoint in the transition of the G1 to S phase of the cell cycle, which plays a crucial role in cell regulation and prevents the excessive and unrestricted proliferation of cells.It was found that overexpression of Cyclin E could accelerate the G1 phase of the cell cycle [53].The destruction of these pathways or the absence or overexpression of speci c factors is closely related to the malignant transformation of cells and cancer development [54].Cyclin E expression levels are elevated signi cantly in many human tumors, manifested by ampli cation of the genome locus at which it is located (19q12-Q13) [55,56], leading to widespread concern about its role in tumors.The over-expressed Cyclin E manifests as different molecular phenotypes in breast cancer and issue in the inactivation of the pRb pathway.A lower molecule form of Cyclin E derived from calpain induces a higher kinase activity [57,58].It has been pointed out that the oncogenic activity of Cyclin E can be inhibited by truncating the protein [59].At present, many studies have assessed that the overexpression of Cyclin E may be a prognostic factor of breast cancer patients.Most of the research shows that Cyclin E can indicate breast cancer prognosis, but other studies could not con rm this nding.The present meta-analysis is an updated one to investigate the correlation between Cyclin E expression and breast cancer prognosis.
In this study, a quantitative and systematic analysis was conducted on the published literature.Through reasonable inclusion and exclusion criteria, available literature was obtained, and relevant data were extracted.The correlation between the high expression of Cyclin E in breast cancer and prognosis and clinicopathological characteristics were analyzed systematically.
As the data are shown above, our meta-analysis results demonstrate that Cyclin E overexpression was statistically signi cantly associated with worse OS, BCSS, and RFS either in univariate or multivariate analysis in the overall population.In contrast to the previous two meta-analyses [60,61].Our results suggest that Cyclin E overexpression is associated with poor RFS, which is different from previous studies.
In the original studies, the authors used hazard ratios and risk ratios to analyze the association between Cyclin E and prognosis in breast cancer patients using univariate and multivariate analyses.RR is a measure of the strength of the association between exposure and outcome, that is, the ratio of morbidity or mortality in the exposed group to that in the non-exposed group, and applies to cohort studies or randomized controlled trials.RR indicates how many times the risk of morbidity or death in the exposed group is that of the non-exposed group.The more signi cant the RR value, the greater the effect of exposure, the stronger the association between exposure and outcome.The hazard ratio (HR), which is the ratio of the risk function of the exposed group to the non-exposed group, is generally obtained from the Cox proportional risk model.Most people think that HR and RR mean the same thing, but HR has a time factor.In other words, RR with time effect is HR.In the survival data, RR takes into account the difference in endpoint events, while HR takes into account the presence of endpoint events and the time taken to reach the endpoint.Therefore, there are differences between the two, and it is necessary to analyze them separately.Besides, in some literature, Cyclin E was a statistically signi cant prognostic indicator in univariate analysis, but not in multivariate analysis, so we also analyzed it from two aspects. .LMW has a high prognostic value in primary breast cancer [13,22,24].However, some immunohistochemical assays lack antibodies to detect the amino terminus that may lead to the differences in the prognostic value of Cyclin E in previous studies.Second, the differences in selected patients, adjuvant therapy resulted in bias in the analysis.Moreover, all samples in the studies were obtained through modi ed radical mastectomy or breast-conserving surgery, and most of the studies clearly stated that no preoperative anti-tumor therapy had been given.Meanwhile, some articles did not mention preoperative treatment, which may affect the detection of Cyclin E. Therefore, the combined results should be interpreted with caution.

Conclusion
In conclusion, our results suggest that high expression of Cyclin E may be a prognostic marker for poor OS, BCSS, and RFS regardless of age, HR status, and axillary lymph node status.Cyclin E overexpression was signi cantly associated with a range of clinicopathological parameters, such as age, lymph node metastasis, ER-negative and menopausal status.This information may help clinicians screen patients for anti-cyclin E therapy.To determine the role of Cyclin E expression in breast cancer, high-quality studies in more extensive, homogeneous populations are needed.

Declarations
Ethics approval and consent to participate

Not applicable
Consent for publication Not Applicable.

Figure 1 Literature
Figure 1

Figure 2 Estimated
Figure 2

Figure 3 Estimated
Figure 3

Figure 4 Estimated
Figure 4

Table 3
Correlation between high Cyclin E level and clinicopathological parameters A total of 13 articles assessed the relationship between Cyclin E and BCSS in breast cancer patients.The result suggested that high Cyclin E expression was a prognostic indicator of poor BCSS both in univariate and multivariate analysis using RR (2.06, 95%CI, 1.48-2.87,P<0.0001; 2.52, 95%CI, 1.70-3.74,p<0.00001) as shown in Fig.2C and 2D.When HR was used, high Cyclin E overexpression was only statistically signi cantly associated with poor BCSS in univariate analysis.The combined HR for BCSS was 2.62, with a 95% CI 1.26-5.42(p = 0.01), as