The association between copper transporters and the prognosis of cancer patients undergoing chemotherapy: a meta-analysis of literatures and datasets

Copper transporter 1 (CTR1), copper transporter 2 (CTR2), copper-transporting p-type adenosine triphosphatase 1 and 2 (ATP7A and ATP7B) are key mediators of cellular cisplatin, carboplatin and oxaliplatin accumulation. In this meta-analysis, we aimed to evaluate the relation of CTR1, CTR2, ATP7A and ATP7B to overall survival (OS), progression-free survival (PFS), disease-free survival (DFS) and treatment response (TR) of cancer patients who received chemotherapy based on published literatures, the Gene Expression Omnibus (GEO) and the Cancer Genome Atlas (TCGA) datasets. Hazard ratios (HRs) and odds ratios (ORs) were pooled using random-effect models. Subgroup analysis and sensitivity analysis were conducted; heterogeneity and publication bias were assessed. Twelve literatures and eight datasets with 2149 patients were included. Our results suggested that high CTR1 expression was associated with favorable OS, PFS, DFS and TR in cancer patients who underwent chemotherapy with acceptable heterogeneity. The relationship of CTR1 to cancer prognosis remained significant in the subgroup of patients who underwent platinum-based chemotherapy, the patients with ovarian cancer and those with lung cancer. The significance of these relationships was not influenced by geological region of publication, data origin or detection method. However, there was no evidence for relation of CTR2, ATP7A or ATP7B to OS, PFS, DFS or TR. Test of publication bias and sensitivity analysis suggested a robustness of all the summary effect estimates. In conclusion, high CTR1 level predicts prolonged survival and enhanced response to chemotherapy in cancer patients who underwent chemotherapy and CTR1 might be a potential target to circumvent chemotherapy resistance.


INTRODUCTION
Platinum-based chemotherapy remains a first-line treatment of various epithelial malignancies including lung, bladder, head and neck and gynecological cancers for over 30 years. Although over half of the patients respond well to platinum-based chemotherapy, relapse and drugresistance are major limitations for successful treatment of cancers. The platinum drugs exert cytotoxicity mainly through binding DNA to form crosslinks, which interfere normal DNA activities and induce cellular apoptosis [1].
However in resistant cancer cells, decreased cellular drug accumulation impairs the formation of DNA-platinum abducts. Low platinum concentration in tumor tissue is significantly associated with reduced tumor response [2,3]. Therefore, drug accumulation is identified as a key process that governs response to chemotherapy in cancer.
Precedent evidence suggested that platinum drugs entered cancer cells through passive diffusion because the time-linear influx of platinum did not reach saturation at relevant drug concentration, was not inhibited by any structural analogous nor was it dependent on an optimal pH [4]. According to recent studies however, Review www.impactjournals.com/oncotarget platinum accumulation was predominantly determined by transporters, which mediate either active or passive transportation of the drug [5]. Early evidence revealed that cellular cisplatin concentration was associated with cellular copper concentration [6,7]. Studies from several independent laboratories revealed that copper metabolic pathways were involved in influx and efflux of platinum. CTR1 (copper transporter 1), CTR2 (copper transporter 2), ATP7A (copper-transporting p-type adenosine triphosphatase 1) and ATP7B (copper-transporting p-type adenosine triphosphatase 2) are four major copper transporters which mediate homeostasis and copper metabolism. With two conserved copper/cisplatin binding motifs, CTR1 facilitates copper and platinum to traverse through cellular membrane. CTR2 regulates copper/ cisplatin storage by interacting and truncating CTR1 to smaller fragment, which is deprived of the ability to transport copper and cisplatin [8]. ATP7A sequesters intracellular platinum complex distribution and facilitates efflux of platinum complex. The turnover of ATP7B between trans Golgi network and cell membrane through endosomal/lysomal vesicle trafficking regulates cellular copper hemostasis by eliminating excess copper, which is also involved in cellular platinum elimination [9]. In addition to cisplatin, carboplatin and oxaliplatin are also substrates of these copper transporters [10][11][12] ( Figure  1). Preclinical studies demonstrated that CTR1, CTR2, ATP7A and ATP7B expression regulate cellular response to cisplatin [8,[13][14][15]. However, the relationship between these copper transporters and the prognosis of cancer patients who received chemotherapy are inconclusive. Therefore, we conducted a meta-analysis of published literatures as well as datasets to assess the association between the expression of copper transporters and the prognosis in cancer patients who underwent chemotherapy. To obtain more detailed information, subgroup analyses stratified by cancer type, chemotherapy regimen, geological region, data origin and detection method were conducted.

Copper transporters and survival
Eleven studies evaluated the relationship between CTR1 expression and OS of cancer patients who received chemotherapy, and ten studies evaluated PFS/DFS. High CTR1 expression was correlated with a favorable OS and PFS in cancers (for OS, HR = 0.64, 95% CI: 0.50-0.82, P = 0.000; for PFS, HR = 0.65, 95% CI: 0.57-0.75, P = 0.000) (Figure 3 and Figure 4). The heterogeneities across studies included were moderate and acceptable (for OS, I 2 = 47.3%; for PFS, I 2 = 0.0%). However, there was no significant relation of the expression of CTR2, ATP7A and ATP7B to OS or PFS/DFS ( Figure 5-10).

Sensitivity analysis and Publication bias
According to sensitivity analyses, the summary HRs for OS of CTR1 and PFS of CTR2 were influenced by one study in the "Europe" subgroup [32]. All the remaining studies did not influence the summary HRs and ORs. No evidence of publication bias was found in overall metaanalyses using funnel plots and Egger's test (all P > 0.05).

DISCUSSION
In this meta-analysis, we systematically evaluated the relationship between four copper transporters (CTR1, CTR2, ATP7A and ATP7B) that influence cellular platinum accumulation and prognosis of cancer patients who received chemotherapy. Our results showed that high CTR1 expression was significantly associated with a favorable OS, PFS, DFS and TR, suggesting that CTR1 is a potential prognostic factor for survival in cancer patients underwent chemotherapy and a treatment target for overcoming platinum resistance.
Ever since the introduction of chemotherapy, drug response becomes a major determiner of cancer prognosis, which parallels with FIGO stage and residual debulked tumor volume. The cure rate of testicular cancer zoomed from 15% to 90% attributing to a great response to platinum-based chemotherapy [36]. In ovarian cancer patients, different histologic subtypes are characterized by distinct platinum response leading to different prognosis, the better the response the better the survival. Switching on and off of an array of genes such as BRCA1, BRCA2, Akt1, ERCC1 and ABCB1 were identified to modulate sensitivity toward chemotherapy in cancer patients. We demonstrated that CTR1 is one of those genes whose expression is significantly associated with OS, PFS, DFS and chemotherapy response in patients underwent chemotherapy, especially in those who received platinum-    based chemotherapy. CTR1 contributes to the active cellular uptake of cisplatin, carboplatin and oxaliplatin. In our study, most patients received cisplatin-or carboplatinbased chemotherapy, both of which are important substrates of hCTR1. It was reported that hCTR1 was responsible to a 60% to 70% cellular uptake of cisplatin and a 30% to 40% uptake of carboplatin [37,38]. So far, no other chemotherapy agents were identified as CTR1 substrates. Unfortunately, we were not able to analyze the impact of CTR1 on patients underwent non-platinum containing chemotherapy due to insufficient data. We found a significant association between CTR1 expression and tumor prognosis by pooling data from 5 studies and 8 datasets involving 2149 patients. Although all data were retrospectively obtained, multiple data sources greatly decreased the publication bias and efficiently increased the validity of the results. Significant clinical associations of CTR1 expression were constant between all subgroups: cancer type, geographical region, data origin and detection method with acceptable heterogeneities.
In our study, we focused on evaluating the initial copper transporter expression in blood and tissue specimens obtained before or during primary surgery. We found that a high initial hCTR1 expression predicted improved prognosis in ovarian and lung cancer patients. However, pre-clinical evidence indicates that the expression of CTR1 is dynamic. Prompt reaction of CTR1 to acute copper concentration alteration is a wellrecognized mechanism that regulates hCTR1 expression. Excessive copper triggers rapid hCTR1 internalization by endocytosis, which reduces the transportation activity of CTR1 [39,40]. Copper deficiency up-regulates transcription-factor-specific protein 1 (Sp1), which binds to hCTR1 promoter region and enhances hCTR1 expression [41]. Some studies also reported that cisplatin induces rapid hCTR1 degradation through ubiquitinmediated proteolysis [42,43]. According to Jandial et al., administration of proteasome inhibitor bortezomib successfully blocked platinum-induced CTR1 degradation and increased platinum accumulation in peritoneal tumors [44]. Whether hCTR1 expression is subjected to similar dynamic change induced by substrate stimulation in cancer patients and whether this change is associated with tumor prognosis are still undetermined.
In relation to chemotherapy response and cancer prognosis, according to our results, CTR1 is of particular attractiveness to serve as a promising target to circumvent platinum resistance. Since correlation between reduced hCTR1 expression and platinum resistance was exhibited in precedent and the present studies, combination of copper lowering agents and platinum based chemotherapy, which stimulates hCTR1 expression and improve platinum uptake, is a promising strategy to treat hCTR1 deficiency mediated platinum resistance. Ishida S et al. successfully enhanced the cytotoxic effect of cisplatin on human cervical and ovarian cancer cells by up-regulating hCTR1 expression through administration of a copper chelating agent tetrathiomolybdate in vitro [45]. Fu et al. carried out the first clinical trial to explore the clinical value of copper chelating agents. Approximate 20% of the platinum resistant patients responded to platinum after receiving copper chelator [46]. However, this strategy was not tested in clinical trials.
We initially expected diverse prognostic values of CTR2 in distinct cancer types, in patients with different genetic background and in studies applied inconsistent CTR2 detection methods based on existing studies. Though CTR2 shares a substantial similar structure with CTR1, it plays drastically different role in platinum uptake. Unlike CTR1, preclinical studies revealed that high CTR2 level was associated with decreased cellular platinum uptake and poor drug response in ovarian cancer [8,47]. Additionally, copper modulation correlated CTR2 with angiogenesis in clear cell renal cell carcinoma, where lower CTR2 expression predicted shorter OS and DFS [48]. Heterogeneity of genetic background of patients not only influences basic CTR2 level of the population but also alters various gene levels and pathways that CTR2 interacts with. Moreover, CTR2 regulates cellular platinum uptake through a protein-protein interaction with CTR1 [8], the relationship between CTR2 and platinum sensitivity is better revealed post-transcriptionally. When evaluating the relationship of CTR2 and cancer prognosis in a clinical setting, we did not find any evidence for correlation between CTR2 and prognosis in different cancer types, data origins and detection methods but saw a discrepancy in geographic stratification of the America and the Europe subgroups regarding PFS. We found that the detection method, cancer type, methodology, type of samples were consistent in all 5 studies from these two subgroups. One study from the Europe subgroup reported a specific result that was opposite to all other studies [32]. We did not have enough evidence to determine whether the inconsistent result was due to genetic difference or simply small size effect. We believed it was necessary to investigate the issue in a much greater clinical scale and covered as much as detection methods as possible since our results were only obtained on the basis of a series gene-array analyses due to limited data, which might not necessarily exhibit the correlation between CTR2 protein expression and platinum resistance.
The present study is the first meta-analysis regarding the impact of copper transporters on the outcomes of cancer patients who received chemotherapy to our knowledge. We used multiple data sources and performed subgroup analyses stratified by geological regions of publication, data origins and detection methods to accommodate possible biases and provide detailed results that help accelerate the relevant clinical translation. However, all the enrolled studies are retrospective, which provide evidence of Level II-1 or II-2 [49] and limit the reliability of our findings. The relationship between CTR1 and chemotherapy clinical outcomes in tumor patients needs validation by large-scale randomized controlled trials. In addition, most data for meta-analysis were obtained from studies on ovarian cancer and lung cancer, which restricts the application of the conclusions in other malignancies.
To sum up, high CTR1 level is associated with improved survival and TR in cancer patients, especially in patients with ovarian or lung cancer undergoing platinumbased chemotherapy, Prospective randomized studies are needed to verify the clinical value of CTR1in tumor patients.

Study Selection
A two-step study selection was conducted including initial review of titles and abstracts, and a second review of full texts retrieved. The inclusion criteria for identifying eligible studies included: 1) clinical studies measured CTR1, CTR2, ATP7A or ATP7B in solid tumors regardless of detection methods; 2) published prospective and retrospective cohort studies; 3) studies provided information of survival and chemotherapy response. Exclusion criteria included: 1) the patients enrolled did not receive chemotherapy; 2) studies focused on gene polymorphisms or DNA methylation; 3) studies investigated different end-points; 4) studies did not provide sufficient data. Eligible datasets should meet the following criteria: 1) cancer patients received chemotherapy; 2) the sample size should be more than 40; 3) gene array was performed using Affymetrix platform. If multiple studies shared overlapped populations, the study with the largest sample size was included. If multiple studies with overlapped populations used different detection methods, all were included.

Quality assessment and Data extraction
The quality of the included literatures was assessed according to the Newcastle-Ottawa Scale (NOS) for cohort studies [50]. We extracted the following data from each individual eligible study or dataset: first author's/ contributor's name, year of publication, country, sample size, number of cases and controls, tumor type, detection methods and detection results. Two authors extracted the data independently and disagreements were solved by discussion.

Statistical Methods
The aim of our study was to investigate the relationship between the expression of copper transporters and the prognosis of cancer patients who received chemotherapy. The primary outcome measure was overall survival (OS), and secondary measures were progressionfree survival (PFS)/disease-free survival (DFS) and treatment response (TR). The impact of platinumtransporter expression on OS, PFS and DFS was assessed by hazard ratio (HR), odds ratio (OR) in case of TR, and its 95% confidence interval (CI). The HRs and 95% CIs were extracted from eligible studies directly or calculated using the survival data extracted from the reported Kaplan-Meier curves with the software GetDATA Graph Digitizer 2.25 according to the method described by Tierney et al. [51]. An on-line survival analysis software KM-plotter (http:// kmplot.com/analysis/) with the selection of optimal cutoff point and microarray probe set was used to calculate the HRs and 95% CIs using data from GEO and TCGA datasets [52][53][54]. Random-effects models were employed to pool the HRs, ORs and 95%CIs [55]. The heterogeneity across studies was assessed according to I 2 statistic with the method of Higgins et al. [56]. A heterogeneity was considered significant when I 2 ≥ 50%. Subgroup analyses stratified by tumor type, geographic region, data origin and detection method were performed. Sensitivity analyses were performed by excluding one study at a time to evaluate the influence of single studies on summary HRs. Publication bias was assessed by funnel plot and Egger's test [57]. All analyses were conducted using STATA14 (MP-Parallel Edition, College Station, Texas 77845 USA). P values less than 0.05 were considered statistically significant.