Meta-Analysis of Circulating Endothelial Cells and Circulating Endothelial Progenitor Cells as Prognostic Factors in Lung Cancer

Lung cancer accounts for 12 % of all cancers diagnosed and its high incidence with its low survival rate leads it to be one of the leading causes of cancer death in worldwide (Jemal et al., 2011). Therefore, the studies of predictive and prognostic are in constant research, aiming to customize treatment and optimize the effectiveness of available agents (Aggarval et al., 2010). It is important to evaluate therapy effectiveness and avoid unnecessary side effects at an early stage by predictive tools. Recently, significant progress has been made in both imaging techniques and identification of serological markers which might serve to modify the therapeutic strategy of patients who do not respond. Circulating endothelial cells (CECs) is a population of mature endothelial cells with a quite low number in healthy individuals and the increase in the circulation indicates the presence of vascular endothelium damage (Shantsila et al., 2008). The increase of CECs has been observed in various processes, such as inflammations, cardiovascular diseases, perioperative period, autoimmune diseases and cancer. Endothelial progenitor cells (EPCs)


Introduction
Lung cancer accounts for 12 % of all cancers diagnosed and its high incidence with its low survival rate leads it to be one of the leading causes of cancer death in worldwide (Jemal et al., 2011).Therefore, the studies of predictive and prognostic are in constant research, aiming to customize treatment and optimize the effectiveness of available agents (Aggarval et al., 2010).It is important to evaluate therapy effectiveness and avoid unnecessary side effects at an early stage by predictive tools.Recently, significant progress has been made in both imaging techniques and identification of serological markers which might serve to modify the therapeutic strategy of patients who do not respond.
Circulating endothelial cells (CECs) is a population of mature endothelial cells with a quite low number in healthy individuals and the increase in the circulation indicates the presence of vascular endothelium damage (Shantsila et al., 2008).The increase of CECs has been observed in various processes, such as inflammations, cardiovascular diseases, perioperative period, autoimmune diseases and cancer.Endothelial progenitor cells (EPCs)

Meta-Analysis of Circulating Endothelial Cells and Circulating Endothelial Progenitor Cells as Prognostic Factors in Lung Cancer
Min Yu 1 *, Hai-Tao Men 1 , Zhi-Min Niu 2 , Yu-Xi Zhu 1 , Ben-Xu Tan 1 , Long-Hao Li 1 , Juan Jiang 1 are defined as circulating precursor cells with the ability to differentiate to mature endothelial cells, form functional blood vessels, and therefore, play a role in promoting abnormal vascularization in neoplastic sites (Melero-Martin et al., 2011).These factors, such as age, male gender, smoking, blood pressure (BP) levels and cardiovascular risk factors, have been reported to reduce the number of circulating EPCs (CEPCs) (Pirro et al., 2006;Pirro et al., 2007;Fadini et al., 2008;Pirro et al., 2008;Yue et al., 2010).The contribution of these cells to the vasculature of solid tumours and prognosis of cancer patients is still controversial.
The aim of this study was to investigate whether baseline CECs and CEPCs, before the treatment of chemotherapy, radiotherapy or surgery, have prognostic or predictive roles in patients with lung cancer.

Materials and Methods
Identification and Eligibility of Relevant Studies.We identified all studies, published or not, respectively targeting all endothelial cells or endothelial progenitor cells' markers in patients with lung cancer, by an electronic search using online PubMed (Medline) and EMBASE, with the search strategies based on combinations of "circulating endothelial cells", "circulating endothelial progenitor cells", and "lung cancer".References of retrieved articles were also screened to identify any studies missed by the search strategies.
Last query was updated on June 20, 2014.Candidate articles were identified for the meta-analysis studies based on title and abstract after reading by two independent reviewers (Yu M and Men HT).When cannot be categorized, full-text review was retrieved.Abstract review was restricted to English.Reported data required for meta-analysis were then identified and extracted.Prespecified quality-related inclusion or exclusion criteria were not used and each study had not been weight by a quality score because no such score has received general agreement for meta-analyses of observational studies (Altman et al., 2001).We made an effort to contact investigators by e-mail to get unpublished data regarding CECs, CEPCs and survival.
Definitions and Standardizations.The studies included defined pretreatment CECs and CEPCs using flow cytometry, CellSearch Assay or fluorescent microscopy.The main outcome of our meta-analysis was 1-year PFS and OS.All of these patients were followed up until death or for at least 24 months.We also listed age, sex, clinical stage and different makers to identify CECs and CEPCs.
Data Extraction.Two reviewers (Yu M and Niu ZM) independently extracted data from all primary studies.The primary data were the p value, the Kaplan-Meier survival curves or Odds Ratio (OR) and 95% confidence interval (CI) of survival outcomes.Additional data obtained from the studies included the first author, publication year, origination country, number of patients, types for CECs and CEPCs assessment, the marker(s) used for staining, cut-off value, number of events in each category, and 1-year OS and PFS.Disagreements were resolved by consensus between the two readers.All studies included were retrospective.
Statistical Analyses.Using the median or mean CECs and CEPCs as a cut-off, patients were classified into the ''high CECs/CEPCs'' group and ''low CECs/CEPCs'' group.A study was considered significant when the p value was less than 0.05 in univariate analysis.OR with 95% confidence intervals (CI) synthesized were used to assess the strength of association.For the quantitative aggregation of survival results, we measured the impact of CECs and CEPCs on survival by estimating the OR between the ''high CECs/CEPCs'' and ''low CECs/ CEPCs'' group.The simplest method to get OR and their 95% CI is to find the exacted value from the original article.When the values were not available, the methods described by Parmer et al. (1998) were used to estimate OR.The published data including total number of patients and number of events in each group from articles were extracted, and the calculations were done, presuming that the rate of censored patients was constant during the study follow-up.We read Kaplan-Meier curves (Men HT) by Engauge Digitizer version 4.1 (free software downloaded from http//sourceforge.net).Considering the many sources of heterogeneity between studies and consequently between their individual OR estimates, we calculated the overall OR according to the Der Simonian and Laird's method (DerSimonian et al., 1986).A fixed effect model was used for secondary analysis, when homogeneity was fine (p≤0.10,I 2 ≤50%), and a random effect model was used if not.An observed OR>1 indicated worse outcome for the ''high CECs/CEPCs'' group relative to ''low CECs/ CEPCs'' group and would be considered statistically significant if the 95%CI did not overlap 1, with p<0.05.Forrest plots were used to estimate the effect of high CECs or CEPCs counts on survival outcome, and STATA 12.0 (STATA Corporation, College Station, TX) was used for our analysis (Yu M and Zhu YX).Potential publication bias was evaluated using the Begg's funnel plot.It was considered that there is no publication bias when the p value was more than 0.05 (Begg et al., 1994).

Results
Eligible Studies.Our electronic search algorithm retrieved a total of 340 references (221 from PubMed and 24 from Embase) for CECs/CEPCs (245 for CECs and 95 for CEPCs)and NSCLC.14 reports were finally identified, while one of which was excluded for lacking informative clinical data (Nowak et al., 2010).For all the patients, CECs and CEPCs measurements had been done before any treatment.
Baseline characteristics of the 13 eligible studies are listed in Table 1.Seven reports originated from Asian, five from Europe and one from Oceania.All of the eligible studies were observational retrospective studies.All the patients included for CECs and CEPCs meta-analysis are diagnosed with non-small cell lung cancer (NSCLC) except the patients with small lung cancer (SCLC) from  .2015.16.14.6123Circulating Endothelial Cells and Endothelial Progenitor Cells as Prognostic Factors in Lung Cancer -a Meta-Analysis Bogos K's study.The markers used for CECs staining were commonly factor CD146, CD105 or CD45, whereas only one study stained for CD31 (Yuan et al., 2015).The markers for CEPCs staining were CD34, VEGFR2, VEGFR3, CD133 or KDR in all of our eligible studies.The methods to count the cells include CellTracks, immunomagnetic technique / fluorescent microscopy (IMT/FM) and flow cytometry (FCM).
Survival at 1 years.As between-study heterogeneity was significant (I 2 >50.0%) for both CECs and CEPCs group, random model was used.High baseline CECs levels were associated with worse prognosis of OS regarding to the result within 1 years (OR 1.641, 95% CI 0.967-2.786,p=0.008) (Figure 1), but not PFS (OR 1.168, 95% CI 0.649-2.100,p=0.062) (Figure2).Therefore, the overall mortality was 1.641-fold higher for patients whose baseline CECs levels were above the cut-off, compared with those below it.Begg's test and funnel was used to evaluate publication bias.No significant publication biases were found in results of meta-analyses of CECs levels for 1-year PFS or OS (Figure 3).Consequently, for CEPCs, the patients with high baseline CEPCs levels were associated both with worse prognosis of OS (OR 12.673,p<0.0001)and PFS (OR 4.930, 95% CI 0.931-26.096,p=0.006).Subgroup analyses were conducted according to clinical stage.We analyzed the effect of high CECs counting before treatment on patients with advanced non-small-cell lung cancer (NSCLC).Negative effects of CECs on 1-year PFS were shown still (n=521, RR 0.735, 95% CI 0.377-1.433,p=0.302) and positive effects on OS were displayed (n=521, RR 1.493, 95% CI 0.866-2.576,p=0.005).

Discussion
It has been reported that CECs and CEPCs levels might either increase or decrease after treatment (Schillaci et al., 2009, Stein et al., 2008).In patients achieving complete remission or complete response, the level of CECs and CEPCs can be subsequently reduced (Mancuso et al., 2001), while in patients with incomplete surgery or nonresponding patients (patients with local recurrence or stable/progressive disease) tended to have higher CECs and CEPCs levels (Dome et al., 2006).But the current result of the predictive role of pretreatment CECs level is ambiguous, even within the 7 studies we included.Sanchez Hernandez et al. (2015), Najjar et al. (2015), Wang et al. (2013) reported no correlation between response to treatment and pretreatment CECs levels.While the increase of CECs numbers after the first cycle could be a negative predictive factor.Ilie et al. (2014), Fleitas et al. (2012) demonstrated high baseline level of CECs correlated with poor prognosis.However, Chu' study implies that higher pretreatment CECs level indicates higher response rate and improved PFS (Chu et al., 2012).
According to our meta-analysis, it seems that the pretreatment CECs level is a promising predictor of clinical response and survival in advanced NSCLC.Because CECs is a marker of angiogenesis, it is likely that a high CECs value is associated with a poor prognosis and lower effectiveness of antiangiogenic therapy, which will lead to poor PFS/OS.This also indicates an anti-angiogenic regimen might be more effective against tumors with high CECs values.But it still needs more clinical trials before trying to make any certain conclusions.
Circulating endothelial cells (CECs) have been recognized as a useful biomarker for vascular damage (Kawaishi et al., 2009).It is increased in many benign diseases and various cancers.In our meta-analysis, high counts of CECs seem to be associated only with worse 1-year OS in patients with lung cancer.This might partly due to the lack of survival data and small sample size.Many related studies have shown controversial results.
The two main methods for the quantification of CECs and CEPCs are based on flow cytometry or inmunomagnetic separation (Woywodt et al., 2006;Mancuso et al., 2009).Twelve studies in our analysis have been using flow cytometry for the quantification of CECs and CEPCs, while 1 study has been using inmunomagnetic separation.Those two different methods have both been proved to be effective.
Not long ago, malignant tumors were thought to acquire vasculature only through local vessel angiogenesis, the mechanism by which new capillaries can only arise from pre-existing ones (Sakamori et al., 2012).However, recent evidence suggests that tumor vasculature can also arise through vasculogenesis, a process in which bone marrow derived endothelial   doi.org/10.7314/APJCP.2015.16.14.6123 Circulating Endothelial Cells and Endothelial Progenitor Cells as Prognostic Factors in Lung Cancer -a Meta-Analysis progenitor cells contribute to neovascularization (Davidoff et al., 2001;Bolontrade et al., 2002).It is postulated that circulating endothelial progenitor cells (CEPs) are mobilized from the bone marrow into the circulation by tumor-or ischemia-induced signals (Sakamori et al., 2012).CEPs subsequently migrate through blood flow to sites of tumor neovascularization, where they differentiate into endothelial cells and contribute to angiogenesis (Asahara et al., 1997;Rafii et al., 2002;Gao et al., 2008).The angiogenic cytokine released from CEPCs might be a supportive mechanism to improve neovascularization (Urbich et al., 2005;Yoon et al., 2005).
Our analysis showed a high level of pro-treatment CEPCs correlated with both with worse PFS and OS.Further predictive role of CEPCs can't be decided due to the lack of clinical data.It is noteworthy, that the CEPCs level was found to be a sensitive surrogate marker of the angiogenic activity in murine model (Shaked et al., 2005).A study has also showed a tendency that responders tended to have lower pretreatment CEPCs numbers than those who did not respond (Treat et al., 2005).Jeanine (Roodhart et al., 2010) has also demonstrated CEPCs number can increase after chemotherapy and predicted worse PFS/OS, regardless of tumor type or chemotherapy regime.Adjuvant chemotherapy showed similar kinetics indicated the increase in CEPCs is seemingly unrelated to the presence of a tumor.Dome et al. (2006) has reported a significantly higher incidence of death from NSCLC in patients with high pretreatment CEPCs levels compared with patients with low CEPCs levels, which suggesting that the pretreatment levels of CEPCs correlate with the clinical behavior of human NSCLC.It has also been reported the numbers of CEPCs rapidly decline in rectal cancer patients receiving antiangiogenic treatment (Willett et al., 2005).This assumption corresponds to the ''vessel normalization'' hypothesis described by Jain et al. (2005) (Jain et al., 2005;Stetler-Stevenson et al., 2014) regarding the clinical effects of anti-VEGF therapy.Our meta-analysis along with those studies suggest that those patients with higher pretreatment CEPCs numbers (more tortuous intratumoral capillaries), presumably having less normalized tumor vessels, respond worse, whereas those with lower level of CEPCs respond better.Consequently, patients with high pretreatment CEPCs numbers could be treated with anti-VEGF therapy to lower CEPCs number (normalizing the vasculature) before/with chemotherapy, thus potentially improving therapeutic response, which also indicates that pretreatment CEPCs might be a potential target biomarker for the effectiveness of antiangiogenic therapy.
Between-study heterogeneity was significant in our study (I 2 >50%).Different cut-off value of CECs and CEPCs levels were used in different studies.We tried to reduce the variability by screening the literature using the same standard and dividing studies into subgroups, such as the same sex,same clinical stage, and cut-off values in prognostic meta-analysis.Although the heterogeneity could not be eliminated totally, the heterogeneity had decreased in some subgroups such as patients with advanced lung cancer in clinical stage III-IV (17.7%), which revealed that all the factors have effects on the generation of heterogeneity and cannot be eliminated at the same time.In addition, the limitations still exist in the present detection method.We attempted to minimize publication bias by searching completely, but it was unavoidable that some data was missed for various reasons such as unpublished or ignored studies (Nowak et al., 2010).
In conclusion, this meta-analysis suggested that High counts of CECs seem to be associated with worse 1-year OS in patients with lung cancer, while high level of pretreatment CEPCs correlated with both with worse PFS and OS.These results need be confirmed by more clinical trials.

Figure 1 .
Figure 1.Meta-analysis of the Association Between CECs and OS at 1 Years.Each study is shown by the name of the lead author and the OR with95% CIs

Figure 2 .
Figure 2. Meta-analysis of the Association Between CECs and PFS at 1 Years.Each study is shown by the name of the lead author and the OR with 95% CIs

Figure 4 .
Figure 4. Begg's Funnel Plot Showing the Relation Between OR and 1/SE for the analysis of CECs and PFS.Larger Studies (Those with Larger 1/SE) Show Generally Smaller ORs