The prognostic value of baseline circulating tumor cells in patients with metastatic breast cancer:a meta-analysis

Ling-Mi Hou,Fang-Fang Li,Lu-Lan Pu,Jin-Shui Li,Jia-You Liu,Yu Chen,Yu-Qing Zhou,Hong-Wei Yang,Shi-Shan Deng＊,Mao-Shan Chen＊Department of Thyroid and Breast Surgery,Affiliated Hospital of North Sichuan Medical College,Nanchong,Sichuan 67000,P.R.China.Department of operating room,Suining Central Hospital,Suining,Sichuan 69000,China.Department of Anatomy,North Sichuan Medical College,Nanchong,Sichuan 67000,P.R.China.Department of Breast and Thyroid Surgery,Suining Central Hospital,Suining,Sichuan 69000,China

Abstract Objective:The prognostic value of circulating tumor cells (CTCs) in metastatic breast cancer(MBC) patients was contentious.A meta-analysis was conducted to evaluate whether MBC patients’ clinical outcomes could be predicted by CTCs detection.Methods: Relevant published studies were searched through electronic databases from January 1990 to February 2018,among which,those investigated the correlation between CTCs and clinical outcomes of progression-free survival and overall survival in MBC patients were involved.The hazard ratios (HR)and confidence intervals(CI)in the studies were extracted from the study using random or fixed effects model,and the meta-analysis was conducted.The prognostic value of tumor cells in patients with different subtypes was estimated by subgroup analysis.Results: Twenty-one eligible studies enrolling 3,837 patients were appropriate for pooled analysis.Progression-free survival(HR,1.66;95%CI,1.47-1.87;P=0.000)and overall survival(HR,2.51;95% CI,2.13-2.96; P=0.000) were worse in patients with CTCs-positive.Subtypes of hormone receptor (HorR)positive,human epidermal growth factor receptor-2 (HER2) negative and triple negative with presence of CTCs showed a statistically significant worse PFS and OS.However,CTCs detection presented no prognostic value in patients with HorR-negative or HER2-positive subtypes.Conclusion: The enumeration of CTCs at baseline in patients with MBC subtypes of HorR-positive,HER2-negative and triple negative is connected with disease progression and poor survival,but inappropriate for HorR-negative and HER2-positive subtypes.

Keywords:Circulating tumor cells;Metastatic breast cancer;Prognosis;Meta-analysis

Background

Breast cancer is the most widespread cancer,which is also the leading cause of cancer death among female[1].Meanwhile,the tumor that has progressed to metastasis is usually incurable and comprehensive therapies for MBC are palliative[2].In order to protect patients from receiving dispensable and/or noneffective therapy that better prognostic and predictive markers are required.

Circulating tumor cells (CTCs) can be detected in peripheral blood and released by primary tumors,metastases or recurrences[3].The detection of CTCs is a new method to detect metastatic disease earlier,less invasively and more reliably than currently available conventional approaches,such as radiographic evaluation and tumor marker detected in blood [4,5].Various methods have been described for CTCs detection,such as reverse transcription polymerase chain reaction (RT-PCR) and EpCAM-based CellSearch System [6-8].A different set of tumor markers is applied in each method to isolate,enrich and detect the CTCs,therefore the sensitivity and accuracy between methods is substantially different.Currently,the CellSearch System is the only platform that has been authorized by the US Food and Drug Administration for the clinical application of CTCs enumeration.Many studies have reported that CTCs-positive status is associated with adverse clinical outcomes[7,9-11].However,some other studies stand with the opposite attitude [12-17].Furthermore,the prognosis value of CTCs detection is significantly different in patients with diverse immunohistochemical subtypes [9,18-22].Therefore,the relationship between CTCs and clinical outcomes in MBC remains contentious.To assess the prognostic value of CTCs detection at baseline for metastatic breast cancer(MBC)and cancer subtypes in patients,which is the purpose of this meta-analys.

Methods

Search strategy

A systematic search was conducted in the databases listed below and the publication date should be between January 1990 and February 2018:PubMed,ISI Web of knowledge and Cochrane Library databases.The following keywords could be variably combined and adopted for searching,such as “CTC*”,“circulating tumor cell*” and “breast cancer/carcinoma/neoplasm”.

Study inclusion criteria

The following situations were considered as the inclusion criteria:MBC was diagnosed;CTCs were enumerated at baseline by Cell-Search system and the cutoff was 5CTCs/7.5 mL;more than 50 patients were enrolled who accepted systematical therapy;clinical outcomes were progression-free survival (PFS) or/and overall survival (OS) and survival data were stratified by CTC condition (presence/positive and absence/negative);sufficient data were provided for determining an evaluate of hazard ratios (HR) and a 95% confidence intervals (CI).Nevertheless,non-English literatures,meeting summaries,letters,reviews and duplicated reports were excluded.

Data extraction

The data extracted were detailed as follows:family name of the first author,journal and the year of publication,research design,number and age of patients enrolled,the number of ER-negative/positive and HER2-negative/positive,the number of CTCs-positive/negative,statistical analysis method,therapeutic method,follow-up period,survival results,HR and 95% CI.Assuming that HR and 95% CI are not included in the reported data,the log HR and its standard error are estimated using the established log-rank test P values and the quantity of events or survival curves observed in each group [23].The latest reported data should be extracted in case that a study was reported back and forth.All literatures filtered and data extracted were finished by two independent reviewers and disagreements were resolved by means of discussion,which were examined by a third investigators.

Statistical analyses

The log HRs and the standard error were calculated,which were necessary in each study,so as to calculate the pooled HRs by adopting the extracted or calculated HRs and 95% CI for PFS and OS.A fixed or random effects model was applied to carry out the merged HRs,then the Z test was employed to determine the significance of the pooled estimate.Cochran’s Q test and I2 index were appropriated for evaluating the heterogeneity among studies.On the condition that Q test (P＞0.1) and I2≤50%,the fixed effects model would be adopted for pooled studies.Otherwise,the random effects model would be used [24].Meta-regression analysis was an approach to explore the potential factor of heterogeneity analysis.The Egger’s and Begg’s test was applied to evaluate the publication bias.The stability and reliability of the conclusion were determined by sensitivity analysis.Statistical analyses were performed by virtue of software STATA SE12.0.All statistical tests were two-tailed,with a significance ofP＜0.05.

Results

Characteristics of identified studies

890 literatures were searched initially after 338 repeated literatures were excluded.After filtering the titles,key words and abstracts,792 literatures were excluded.Ninety-eight literatures were reviewed in full included in this meta-analysis eventually,in which fifteen studies [9,11,20-22,25,26,29,30,32-34,36-38]were for PFS and eighteen studies [9,11,18-22,24-28,31,33-36]for OS(Fig.1).

3,837 MBC patients (1672 CTCs-positive and 2165 CTCs-negative) were involved in the included studies.The HRs and 95% CIs were directly provided in sixteen studies.The median sample size was 154 patients with a range from 52 to 517.Only metastatic inflammatory breast cancer [28]and HER2-positive and twenty-one studies[9,11,18-22,25-38]were MBC [32]were respectively included in one study.After being enrolled in the study,all patients accepted systemic treatment and the assessment of the disease progression was performed through image examination.By adopting the CellSearch system,CTCs had been detected in peripheral blood in all eligible studies at baseline and the cutoff was 5CTCs/7.5mL.The positive rate of CTCs ranged from 20.00% to 61.25%,with a median of 44.86%.Table 1 summarizes the main features of the included studies.

Figure 1 Flowchart of study selection.

Figure 2A

Figure 2B

Figure 3A

Figure 4B

Figure 4C

Figure 4D

Meta-analysis

There were fifteen studies for PFS and eighteen studies for OS,in which 2932 patients and 3640 patients were respectively enrolled.The merged HR revealed that the presence of CTCs was significantly connected with poor PFS (HR,1.66;95% CI,1.47-1.87;P=0.000;random effects;Fig.2A).The heterogeneity between studies was moderate (P=0.079,I2=36.4%).The merged HR of OS revealed that the presence of CTCs was connected with a remarkably increased risk of death (HR,2.51;95% CI,2.13-2.96;P=0.000;random effects;Fig.2B).The heterogeneity was existed(P=0.015,I2=46.9%).

The meta-regression was performed with the following covariates being considered:publication year,study design and sample size.In a multilateral analysis,the exploratory variables had a little influence on HR estimate (adjustedP＞0.05,Table 2).The patients’ subtypes in each study were not included in the regression analysis,in consideration of various subtypes involved in each study.

Sensitivity analysis showed that the stability of result for OS was high and one study [20]was the inter-study heterogeneity observed for PFS(Supplementary data,Fig.3A,3B).The estimate HR was re-analyzed for PFS after eliminating this study;the HR was slightly greater and more precise for PFS(HR 1.73;95% CI,1.56-1.91;P=0.000;fixed effects model;Supplementary data),and heterogeneity disappeared(P=0.484,I2=0.0%).

Begg’s test and Egger’s test revealed unremarkable publication bias in this meta-analysis for PFS (PBegg=1.0;PEgger=0.11,Supplementary data,Fig.4A,4B)and OS(PBegg=1.0;PEgger=0.63,Supplementary data,Fig.4C,4D)outcomes.

Subgroup analysis

Subtypes were categorized according to the condition of hormone receptors (including estrogen receptor and progesterone receptor,HorR) and human epidermal growth factor receptor-2 (HER2).Subtypes of HorR-positive,HorR-negative,HER2-positive,HER2-negative and triple negative (TN)were included in the subtype analysis for PFS and OS.Survival outcome data for the subtypes had been reported in eight studies [9,18-22,30,33],among which,one study enrolled 85 HorR-positive in 95 patients was regarded as HorR-positive subtype [33]and one study merely enrolled two HER2-positive in 65 patients was considered as HER2-negative subtype [30].The outcomes revealed that CTCs were significantly connected with a poor clinical survival of PFS and OS in HorR-positive,HER2-negative and TN subtypes,instead of HorR-negative and HER2-positve MBC(Supplementary data).Subtypes of HorR-positive/HER2-positive and HorR-negative/HER2-positive were analyzed to assess the influence of HorR status on the prognostic value of CTCs about HER2-positive patients by HorR status.The results revealed that the conclusion of CTCs in HER2-positive patients was not changed by the HorR status (Supplementary data).The summarized results were presented in Table 3.However,HR for PFS was not pooled to explore the influence of HorR status on HER2-positive MBC,as all the three studies reached negative conclusions[20-22].

Discussion

The current analysis is according to a sea of cohort researches,which significantly different from other published meta-analysis [39-41],in which smaller series were considered,such as patients in different cancer stages and different CTCs detection methods and thresholds.The prognostic value of baseline CTCs were determined in 21 studies using the CellSearch system,with a threshold consistent with that of MBC patients.The results support the hypothesis that peripheral blood CTCs is connected with poor prognosis in MBC patients.However,subgroup analysis for MBC patients of different subtypes indicated that the CTCs detection presented no prognostic value in HorR-negative patients and HER2-positive subtypes regardless of the HorR status.The data were similar to a pooled analysis of individual patient information published in 2014,which was lack of results for subtypes[42].

The CellSearch system,as a clinical test method,is currently the only one approved by the US FDA.With this system,the reported range of CTCs detection rates(20%-61%) was somewhat wide,in consideration that the patients’ different molecular characteristics included in each study maybe the main reason.It is necessary to suggest that standardized technical protocols and quality control specification should be adopted in CTCs detection.

As a heterogeneous disease,breast cancer can be categorized into Luminal A,Luminal B HER2-negative,Luminal B HER2-positive,HER2-positive and basal-like molecular subtypes according to the immunohistochemical characteristics of Horr and HER2 status[43].The clinicians gradually accept the concept that the prognosis can be evaluated and the therapy can be guided by molecular classification in breast cancer patients [43].Only two studies [21,22]have assessed the prognostic value of CTCs count at baseline according to subtypes that were categorized into Luminal A,Luminal B HER2-negative,Luminal B HER2-positive and TN,and their results were inconsistent.The subgroup analysis of different subtypes of MBC were conducted,in which the classification was based on the HorR and HER2 status.In the subgroup analysis,CTCs detection in HorR-positive,HER2-negative and TN,can be made a strong prediction for clinical outcomes,but without a significant prognostic value in HorR-negative and HER2-positive subtypes.In addition,the influence of HorR status on HER2-positive patients is explored and negative results are obtained.In theoretical analysis,HER2-positive patients are usually likely to undergo progress and distant metastasis,who have a higher CTCs detection rate than other breast cancer subtypes.However,the results suggest that CTCs presented no prognostic value in HER2-positive MBC patients.Almost all the HER2-positive MBC patients in the included studies had accepted target HER2 therapy.For patients with this subtype,target HER2 therapy might thoroughly kill the tumor cells and suppress the CTCs detection in peripheral blood,so as to reduce the prognostic value of CTCs count[9].On the other hand,these validated the effectiveness of target HER2 therapy in HER2-positive cancer patients at some level.At present,the mechanism of target therapy influencing the prognostic value of CTCs remains unclear,which need to be further researched.In terms of the few studies that had evaluated the prognostic value of CTCs exploration of breast cancer according to molecular subtypes,further confirmation is necessary concerning whether CTCs could serve for breast cancer patients of different molecular subtypes as an accurate prognostic indicator.

All the included studies had considered the standard exceeding five tumor cells per 7.5 mL as CTCs-positive.In fact,the threshold of 5CTCs/7.5mL is more likely to be a somewhat more arbitrary base on statistical criteria between prognosis groups than actual clinical or biological correlation.Moerover,the truncation value of CTCs enumeration with a cutoff of 1CTCs/7.5mL,which can be a competent prognostic factor for her2-positive MBC patients [29].Besides,this provides a signal that an appropriate threshold would be pondered to determine the prognosis of patients with different subtypes of breast cancer,so as to realize the CTCs predictive value.

Immunophenotypic characteristics of CTCs expression,such as HER2,estrogen receptor and progestrone receptor,could be quite other thing to those of the tumor tissue [32].Regardless of the characteristics of primary tumor,therapies that directly targeted the features of tumor cells could improve the prognosis of cancer patients [44,45].Thus,Clinicians can noninvitingly predict and monitor treatment responses through features of CTC in "iquid real-time biopsy" of metastatic disease,and also decide therapeutic regimens following phenotypic changes in breast cancer phenotype[46].

Main limitations of this study were discussed as following.To begin with,there were various therapeutic regimens before and after the blood withdrawal midst the researches embodied in this meta-analysis and each subtype of MBC did not receive a matching therapy.Secondly,the subgroup analysis was applied among different subtypes that were categorized by the HorR and HER2 status,which may lead to some overlaps between subtypes.Thirdly,the meta-analysis was based on the data extracted from published literatures and no individual patient data were obtained.Moreover,some patients may enrolled in more than one study.In addition,the regression,subgroup and sensitivity analysis was used in assess potential sources of bias and the inter-study heterogeneity observed.Sensitivity analysis showed that one study [34]was mostly responsible for the heterogeneity for PFS,probably because some patients in this study were also enrolled in other studies in our analysis.A similar estimated HR was gained and the heterogeneity entirely disappeared after being excluded in this study.

Conclusions

Except HorR-negative subtypes and HER2-positive subtypes,the current results suggest that the peripheral blood CTCs detection can be a strong predictor of poor prognosis in MBC patients regardless of the HorR status.In addition,it’s more rational to set a different threshold for different breast cancer subtypes.Future research may focus on whether the clinical therapies can be guided through the features of tumor cells expressing.