TTF-1- and/or CD56-positive Circulating Tumor Cells in patients with small cell lung cancer (SCLC)

The aim of the study was to evaluate the phenotypic CTCs heterogeneity (TTF-1+ and/or CD56+) in SCLC patients and correlate it with the CellSearch. Peripheral blood was obtained from 108 consecutive patients. CTCs were detected by CellSearch and double-immunofluorescence using anti-CD45, anti-TTF-1 and anti-CD56 antibodies. Before chemotherapy TTF-1+/CD45−, CD56+/CD45− and TTF-1+/CD56+ CTCs were detected in 66(61.1%), 55(50.9%) and 46(42.6%) patients, respectively; 60.2% of patients were CellSearch+. Among the 22 patients with 0 CTCs/7.5 ml on CellSearch, TTF-1+/CD45−, CD56+/CD45− and TTF-1+/CD56+ CTCs were detected in 8(36.4%), 6(27.3) and 6(27.3%) patients, respectively; no CK+/EpCAM+ or TTF1+/EpCAM+ CTCs were detected in these patients. One-chemotherapy cycle decreased both the number of positive patients (p < 0.001) and their CTC number (p < 0.001), irrespectively of their phenotype and the detection method. The incidence and number of the different CTC subpopulations on PD, was significantly increased at their baseline levels. Multivariate analysis revealed that the increased number of CTCs at baseline and on PD were significantly associated with decreased PFS (p = 0.048) and OS (p = 0.041), respectively. There is an important CTC heterogeneity in such patients according to the expression of TTF-1 and CD56 which could detect EpCAM− CTC subpopulations and, thus, undetectable by CellSearch. These CTC subpopulations are dynamically correlated with treatment efficacy and disease-progression.

including SCLC 28,29 . TTF-1 is a transcription protein of the Nkx-2 gene family and regulates the early human development, cell growth and morphogenesis of thyroid, lung and restricted areas of diencephalon 30 .
The investigation of the phenotype of CTCs is an interesting topic since accumulating evidence points to functional heterogeneity within the CTC compartment with possible clinical implications. Indeed, it has been previously reported that CTCs in SCLC patients are heterogeneous based on their apoptotic and proliferative phenotype 15 . An important problem concerning the investigation of CTCs is related to the used detection methods. Indeed, the clinically validated CellSearch (CS) platform is based on the immunomagnetic capture of EpCAM-positive CTCs which, subsequently, are stained with an anti-pancytokeratin antibody and DAPI; however, since CTCs undergoing Epithelial-Mesenchymal-Transition (EMT) down-modulate the epithelial markers, including EpCAM, it is obvious that CS fails to capture EpCAM-negative CTCs. Therefore, it is interesting to further evaluate the heterogeneity of CTCs in patients with SCLC using known and well established markers, such as TTF-1 and CD56 which are expressed on neuroendoctrine tumor cells. In addition, the presence of CD56-and/or TTF-1-positive CTCs could offer the opportunity to detect CTCs which are undetectable by CS. The aim of the present study was to investigate the phenotypic heterogeneity of CTCs in SCLC patients using the TTF-1 and CD56 neuroendocrine markers.

Patients and Methods
Patient samples and cytospin preparations. Peripheral blood (20 ml in EDTA and 7.5 ml in CellSearch Save preservative tubes; Raritan, NJ, USA) was obtained from newly-diagnosed patients with SCLC before the initiation of systemic treatment (n = 108), after the administration of one chemotherapy cycle (n = 76) and at the time of disease progression (PD; n = 48). All blood samples were obtained at the middle-of-vein puncture after the first 5 ml were discarded to avoid contamination with epithelial cells from the skin. All patients had histologically confirmed SCLC. Patients with ED-SCLC received up to 6-cycles of etoposide/cisplatin while patients with LD-SCLC were treated with concurrent etoposide/cisplatin and radiotherapy. The study has been approved by the Ethics and Scientific Committees of the University General Hospital of Heraklion and all patients signed a written informed consent. All methods were performed in accordance with the relevant guidelines and regulations.
Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll-Hypaque density gradient (d = 1,077 g/ml; Sigma-Aldrich, GmbH, Germany) centrifugation at 1,800 rpm for 30 min. Centrifugation and cytospins were prepared as previously described 31,32 . In brief, aliquots of 5 × 10 5 PBMCs were cyto-centrifuged at 2,000 rpm for 2 min on glass microscope slides. Cytospins were air dried and stored at − 80 °C until use. Two slides (10 × 10 5 PBMCs) from each patient were analyzed at each time point.
Detection of CTCs using the CS platform. For the enumeration of CTCs, peripheral blood samples maintained at ambient temperature and processed within 72 h using the CS platform (Veridex LCC, Raritan, NJ) according to the manufacturer's instructions. CTC morphology was confirmed in all cases and analysis was performed with the CellTracks Analyser II by experienced biologists (E.P and S.A.). Results are expressed as number of CTCs/7.5 ml blood. Double Immunofluorescence Assay (IF). TTF-1 and/or CD56 were detected by double IF staining using monoclonal antibodies either against TTF-1 (DAKO, Agilent Technologies, Denmark) or CD56 (NCAM) (Thermo Fisher Scientific, Fremont, CA, USA) and anti-CD45 (Common Leukocyte Antigen; Santa Cruz, CA, USA) as previously described 33 . Moreover, cytospins were double stained with the mouse anti-TTF-1 or anti-A45-B/B3 (anti-cytokeratins 8, 18, and 19; Micromet, Munich, Germany) and the mouse anti-EpCAM (Acris Antibodies GmbH, Germany) antibodies. In brief, the incubation period for all primary and secondary antibodies was 1 h. Zenon technology (FITC-conjugated IGg1 antibody; Molecular Probes, Invitrogen, CA, USA) was used for the TTF-1 and A45-B/B5 detection. CD56 was labelled with Alexa 555 or Alexa 488 (Molecular Probes); EpCAM and CD45 were labelled with Alexa 555. The omission of the first antibodies has been used as negative controls. DAPI-antifade (Molecular Probes) was added to each sample for nuclear staining. Slides were analyzed using a fluorescence microscope (Leica DM 2500, Heidelberg, Germany). Results are expressed as number of CTCs/10 6 PBMCs.
Cell Line. The human SCLC cell line, H209 (ATCC ® HTB-172) was obtained from the American Type Culture Collection (Manassas, VA, USA) and used as positive control in IF experiments. H209 cells were cultured in RPMI 1640 (Gibco-BRL Life Technologies, Grand Island, NY), supplemented with 10% fetal bovine serum (Gibco-BRL), 2 mM L-glutamine (Gibco-BRL), 10 mMHepes (Gibco-BRL), 1 mM sodium pyruvate (Gibco-BRL), 1.5 g/L NaHCO 3 (Sigma-Aldrich), 4.5 g/L glucose (Sigma-Aldrich) and 50 mg/ml penicillin/streptomycin (Gibco-BRL). Cells were maintained in a humidified atmosphere of 5% CO 2 , at 37 °C. To determine the sensitivity of the method, H209 cells were spiked in peripheral blood obtained from healthy individuals, and the PBMCs and the corresponding, obtained after Ficoll-Hypaque density centrifugation, and cytospins were prepared as above. All experiments were performed during the logarithmic growth phase of the cell line.
Study design and statistics. This is a prospective, single institution study, investigating the expression of CD56 and TTF-1 in CTCs from consecutive patients with SCLC. There was no specific statistical design since the aims of the study were to characterize the heterogeneity of CTCs and to define the presence of phenotypically different subpopulations of CTCs during front-line treatment. The evaluation of CTCs was done blindly to clinical data. PFS was calculated for all patients from the date of the first treatment administration until treatment discontinuation due to disease progression or death. OS was calculated for all patients from enrolment in the study to the date of death due to any cause. The association of risk factors with PFS and OS was analyzed using the log rank test and the Kaplan-Meier method. Univariate and multivariate Cox proportional hazards regression models with hazard ratios Scientific RepoRts | 7:45351 | DOI: 10.1038/srep45351 (HR) and 95% CIs were used to evaluate the predictive and prognostic relevance of the CTCs. Statistical significance was set at p = 0.05. Statistical analysis was performed using the SPSS v. 20 software (IBM Corp. Armonk, NY, USA).

Detection of CTC subpopulations in patients without detectable CTCs by CS. In 22 patients no
CTCs could be detected by CS (0 CTCs/7.5 ml of blood) (  (Table 3). Table 4 demonstrates that in 6 out of 8 patients with detectable CTCs by IF but not by CS, all the subpopulations of CTCs were present; in addition, IF revealed that these patients did not have detectable TTF-1 + / EpCAM + or CK + /EpCAM + CTCs. Ι n addition, no CK + /EpCAM + CTCs could be detected in the remaining 14 patients without detectable CTCs by CS (data not shown). Similarly, the phenotypically different CTC subpopulations could be detected in patients without detectable CTCs both after one treatment cycle and on PD ( Table 3).

Detection of CTC subpopulations and clinical outcome.
Clinical relapse was observed in 89 (82.4%) patients. The incidence of detection of a high number of CTCs by CS both at baseline and after one treatment cycle was higher in the group of patients who experienced a PD compared to patients with no PD (54.2% vs 6.0%; p = 0.004 and 29% vs 0.0%; p = 0.022); however, using IF this difference could not reach any statistical significance (Supplementary Table S2).The median PFS for the whole group of patients was 6.8 months (95% CI: 6.2-7.5). In patients with and without detectable CTCs by CS at baseline, the median PFS was 6.0 and 7.9 (95% CI: 5.4-6.7 and 5.7-10.1) months, respectively (p = 0.001; Fig. 3a); the median PFS was also significantly shorter in patients with detectable CTCs after one chemotherapy cycle (p = 0.004; Fig. 3b Fig. 3c). In addition, the median OS was significantly different in patients with and without detectable CTCs either after one chemotherapy cycle (p = 0.004; Fig. 3d) or on PD (p = 0.021; Supplementary Table S3).
Univariate and Multivariate Analysis. Univariate analysis revealed that PS (ECOG), disease stage, LDH levels, organ metastases, response to treatment and increased number of CTCs at baseline and after one treatment cycle of were significantly associated with a shorter PFS (Supplementary Table S4). Multivariate analysis showed that increased number of CTCs at baseline could be emerged as independent factor associated with a reduced PFS (HR: 1.9, 95% CI: 0.9-3.9; p = 0.048) (Supplementary Table S4). Similarly, PS (ECOG), disease stage, LDH levels, organ metastases, response to treatment and increased number of CTCs at baseline and at the time of PD were significantly associated with a shorter OS in univariate analysis (Supplementary Table S4). Again, in multivariate analysis only the increased number of CTCs at the time of PD could be emerged as independent prognostic factor associated with a decreased OS (HR: 2.1, 95% CI: 0.9-5.3; p = 0.041) (Supplementary Table S4).

Discussion
The prognostic value of CTCs has been previously reported by several investigators 15,17 . Indeed, Hou et al 15 . reported that CTCs were present in 85% of the patients before chemotherapy and their number (≥ 50 CTCs/7.5 ml The current study, demonstrates for the first time that CTCs from patients with newly diagnosed SCLC could be detected using the TTF-1 and CD56 molecules which are expressed in tumor cells of neuroendocrine origin since both TTF-1 and CD56 are expressed in 89-100% 21,24,27,35 and in (90-99%) 36 , of SCLC cases, respectively. In addition, the detection of TTF-1 + and/or CD56 + CTCs clearly revealed the phenotypic heterogeneity of CTCs in patients with SCLC as well as their changes during the different clinical phases of the disease. Indeed, TTF-1 + /CD45 − , CD56 + /CD45 − and TTF-1 + /CD56 + CTCs could be detected in patients before the initiation of any     systemic treatment, after one-cycle and at the time of PD. However, in almost 10-15% of SCLC cases, no TTF-1 expression can be detected [37][38][39] , indicating the absence of these neuroendocrine markers in some CTC subpopulations. On the contrary, the detection of TTF-1 + /CD56 + CTCs, which represent a minority among the whole population of CTCs since their number was less than that of TTF-1 + or CD56 + , further supports the heterogeneity of CTCs in patients with SCLC. Indeed, previous studies have reported both inter-and intra-patient's CTCs heterogeneity by using different methodologies and tumor types [40][41][42][43] . Whether this phenotype represents a specific step of the malignant cell's differentiation or a particular genetic change is unknown.
The presented results also demonstrate that even one treatment cycle resulted to a significant decrease not only of the number of patients with ≥ 1 CTCs/10 6 PBMCs but also of their absolute number; conversely, on disease progression both the number of patients with detectable CTCs as well as the number of CTCs were significantly increased reaching, practically, their baseline levels. These dynamic changes were observed irrespectively of the phenotype of CTC subpopulation. It is interesting to note that in the majority of patients, chemotherapy could not completely eliminate the different CTC subpopulations suggesting that some of them are chemo-sensitive whereas others seem to be chemo-resistant. Similar findings have been observed in other tumor types [44][45][46][47][48] . Moreover, Hou et al. (2012) observed an increase in CTCs expressing the anti-apoptotic protein Bcl-2 after chemotherapy, thus pointing the usefulness of CTCs stratification and pharmacodynamic monitoring in trials 15 . Additionally, they found that the presence of apoptotic CTCs at baseline is significantly correlated with disease stage and number of metastatic sites 15 , whilst intra-and inter-patient heterogeneity was observed for EMT markers in both CTCs and circulating tumor microemboli (CTMs) 49 . Furthermore, the genetic heterogeneity has been documented by single cell genomic hybridization, on DTCs 50 . These observations further support both the phenotypic and molecular heterogeneity of CTCs/DTCs in SCLC patients [51][52][53] .
The presented results also demonstrated for the first time that in a subgroup of patients, CS failed to detect CTCs despite their detection by IF. This observation should be attributed to the finding that, in these particular patients, IF staining could not reveal the presence of either TTF-1 + /EpCAM + or CK + /EpCAM + CTCs since CS recognizes CK + CTCs which are, previously, captured by an anti-EpCAM antibody. It is well known that epithelial tumor cells migrating through the bloodstream undergo EMT which is characterized, among others, by a loss of expression of epithelial markers 32,54,55 ; therefore, it is reasonable to hypothesize that, in this small subgroup of patients without detectable CTCs by CS, the TTF-1 + /EpCAM − CTCs probably represent CTCs undergoing EMT. This is in accordance with studies reporting that the CTC detection rate was higher using EpCAM-independent enrichment methods 56 . Thus, a combination of both EpCAM-and non-EpCAM-based CTC technologies seems to be needed in order to better assess the presence of CTCs in patients with SCLC and to evaluate their biological and clinical relevance. It should be noted that new and promising methodologies have also been developed for the detection of CTCs based on nanoparticles [57][58][59][60][61][62] ; however, these methods have serious limitations, such as the higher cost and the requirement of special equipment compared to the simple immunofluorescence staining.
Previous studies using the CS platform have shown that the detection of CTCs either before the initiation of systemic treatment or after one chemotherapy cycle is significantly associated with PFS and OS 15,17 . However, a recent study reported that although CTCs have a useful prognostic role at baseline, only the important reduction in the CTC number after one chemotherapy cycle significantly improved prognostic accuracy 63 . Our findings are in agreement with the above reports; however, there was no significant correlation between the different CTC subpopulation and the patients' clinical outcome. There is no clear explanation for this observation; however, we cannot exclude that the expression of TTF-1 and CD56 molecules in CTCs are not related with the invasive potential or the aggressivity of tumor cells in contrast with the more epithelial EpCAM + /CK + CTCs detected by CS. Alternatively, we cannot exclude that the used discrimination cut-off in order to characterize the positive samples could be related to this observation; it should be noted that Hou et al. 15 , used as cut-off the 50 CTC/7.5 ml of blood for their analysis.
In conclusion, the current study demonstrated an important phenotypic heterogeneity of CTCs in patients with SCLC and the different subpopulations of CTCs could be used as potential dynamic biomarkers during the different clinical phases of the disease. Subsequent studies should evaluate the biological relevance of these phenotypically different subpopulations of CTCs and, especially, of EpCAM-negative CTCs as well as whether their specific molecular and/or genetic profile could provide information towards a more individualized treatment.