Journal of Thoracic Oncology:
Circulating Tumor Microemboli Diagnostics for Patients with Non–Small-Cell Lung Cancer
Carlsson, Anders PhD*; Nair, Viswam S. MD, MS†; Luttgen, Madelyn S. BS*; Keu, Khun Visith MD‡; Horng, George MD§; Vasanawala, Minal MD‖¶; Kolatkar, Anand PhD*; Jamali, Mehran MBBS¶; Iagaru, Andrei H. MD¶; Kuschner, Ware MD†#; Loo, Billy W. Jr MD, PhD**; Shrager, Joseph B. MD††‡‡; Bethel, Kelly MD§§; Hoh, Carl K. MD‖‖; Bazhenova, Lyudmila MD¶¶; Nieva, Jorge MD##; Kuhn, Peter PhD*; Gambhir, Sanjiv S. MD, PhD¶***†††
*The Scripps Research Institute, Department of Cell Biology, La Jolla, CA; †Department of Medicine, Stanford University School of Medicine Stanford, CA; ‡Centre Hospitalier de l’Universite de Sherbrooke, Department of Nuclear Medicine and Radiobiology, Sherbrooke, Québec; §The California Pacific Medical Center Research Institute, San Francisco, CA; ‖The VA Palo Alto Health Care System, Section of Nuclear Medicine, Palo Alto, CA; ¶Department of Radiology, Stanford University School of Medicine, Stanford, CA; #The VA Palo Alto Health Care System Section of Pulmonary & Critical Care, Palo Alto, CA; **Department of Radiation Oncology; ††Department of Cardiothoracic Surgery, Stanford University School of Medicine, Stanford, CA; ‡‡The VA Palo Alto Health Care System Section of Cardiothoracic Surgery, Palo Alto, CA; §§Scripps Clinic, Department of Pathology, La Jolla, CA; ‖‖Nuclear Medicine Division, University of San Diego Medical Center, San Diego, CA; ¶¶The Moores Cancer Center, University of San Diego Medical Center, La Jolla, CA; ##The Billings Clinic, Department of Hematology/Oncology, Billings, MT; ***Departments of Bioengineering and †††Materials Science and Engineering, Stanford University School of Medicine, Stanford, CA.
Anders Carlsson and Viswam S Nair are the first coauthors of the study.
Peter Kuhn and Sanjiv S. Gambhir are co-senior authors of the study.
A.C. is supported by the Swedish Research Council, Dnr. 2012–235. V.S.N. is supported by the LUNGevity foundation. K.V.K. was supported by a Faculté de Médecine et des Sciences de la Santé de l’Université de Sherbrooke & SMUS Research Fellow Scholarship. The Kuhn lab is supported by U54CA143906 from the National Cancer Institute and the Gambhir Multimodality Imaging Laboratory and Canary Center at Stanford is supported by the NCI ICMIC P50CA114747, NCI CCNE-TR U54 CA119367, CCNE-T U54 CA151459, and the Canary Foundation. Database support for this project was provided by the Stanford Center for Clinical and Translational Education and Research through NIH/NCRR grant UL1 RR025744. The content published here is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Disclosure: Peter Kuhn, Kelly Bethel, Anand Kolatkar, Madelyn Luttgen, Lyudmila Bazhenova, and Jorge Nieva have an ownership interest in Epic Sciences, which has licensed the HD-CTC technology. All other authors declare no conflicts of interest.
Address for correspondence: Sanjiv Sam Gambhir, MD, PhD, Stanford University School of Medicine, 300 Pasteur Drive, Stanford, CA 94305. E-mail: email@example.com
Introduction: Circulating tumor microemboli (CTM) are potentially important cancer biomarkers, but using them for cancer detection in early-stage disease has been assay limited. We examined CTM test performance using a sensitive detection platform to identify stage I non–small-cell lung cancer (NSCLC) patients undergoing imaging evaluation.
Methods: First, we prospectively enrolled patients during 18F-FDG PET-CT imaging evaluation for lung cancer that underwent routine phlebotomy where CTM and circulating tumor cells (CTCs) were identified in blood using nuclear (DAPI), cytokeratin (CK), and CD45 immune-fluorescent antibodies followed by morphologic identification. Second, CTM and CTC data were integrated with patient (age, gender, smoking, and cancer history) and imaging (tumor diameter, location in lung, and maximum standard uptake value [SUVmax]) data to develop and test multiple logistic regression models using a case-control design in a training and test cohort followed by cross-validation in the entire group.
Results: We examined 104 patients with NSCLC, and the subgroup of 80 with stage I disease, and compared them to 25 patients with benign disease. Clinical and imaging data alone were moderately discriminating for all comers (Area under the Curve [AUC] = 0.77) and by stage I disease only (AUC = 0.77). However, the presence of CTM combined with clinical and imaging data was significantly discriminating for diagnostic accuracy in all NSCLC patients (AUC = 0.88, p value = 0.001) and for stage I patients alone (AUC = 0.87, p value = 0.002).
Conclusion: CTM may add utility for lung cancer diagnosis during imaging evaluation using a sensitive detection platform.
Despite the mortality benefit demonstrated by the National Lung Screening Trial,1 there remain concerns about the cost of delivering care for patients identified with a lung nodule given the 96% false positive rate from the study, and the unnecessary procedures resulting from inaccurately risk-stratified nodules using existing prediction models.2 Blood biomarkers have transformative potential by correctly identifying patients who may benefit from treatments with curative intent rather than from watchful waiting.3
CTCs represent one potential advance with direct clinical application given their noninvasive measurement from blood. Although CTCs have been reported in the literature since the 19th century primarily as pathologic curiosities,4–6 in 2004 the Food and Drug Administration’s approval of CTC enumeration using an immuno-bead antibody capture platform (CellSearch; Janssen Diagnostics, Raritan, NJ) advanced the field significantly.7,8 CellSearch was the first technology to demonstrate clinical utility by standardizing the CTC platform, and prospective, observational data have confirmed that CTC burden is related to therapeutic response and prognosis in multiple types of late-stage cancers.9,10 CTC detection in early-stage disease using CellSearch, however, has been less promising due to poor detection sensitivity resulting from suboptimal epithelial cell adhesion molecule (EpCAM) affinity and tumor cell heterogeneity.11–13
More technically sensitive rare cell platforms exist that enrich CTC populations by both EpCAM-dependent14 and EpCAM-independent techniques,15–17 with the ability to detect a two to three log-fold increase in CTCs for non-metastatic cancers. Our group has previously reported that the “High Definition” Circulating Tumor Cell (HD-CTC) platform, which takes advantage of well-established cell markers (DAPI, CK, and CD45) and rapid, automated fluorescence microscopy to identify CTCs by morphology rather than by EpCAM affinity, may be more sensitive than EpCAM-based detection systems.18
In multiple studies over the past decade, we have found that the HD-CTC platform detects large quantities of putative CTCs in metastatic breast, lung, and prostate cancers,18–21 with prognostic discrimination in non–small-cell lung cancer (NSCLC).21 During these studies, we noted that CTCs were detected in high quantities in several early-stage NSCLCs,22 and that tumor cell aggregates (or circulating tumor “microemboli,” [CTM]) were omnipresent across diverse cancers.23 We further discovered that CTCs and CTM were not always related to a tumor’s size or metabolic activity in stage I cancers or across all stages.24 These observations led us to investigate whether CTCs and CTM are not only detectable in stage I NSCLC, but also if they are complementary to current clinical models of risk2 and could aid as a noninvasive diagnostic.
Current methods for risk-stratifying nodules and identifying lung cancer in patients focus on basic demographic variables (i.e., age, cancer, and smoking history),25 CT imaging26,27 and, for nodules greater than 8 mm in size, FDG PET-CT imaging.25,28 To date, rare cell assays have not been well studied for risk-stratifying larger lung nodules to determine whether CTM could be helpful diagnostic adjuncts. We explore here whether adding CTC data to existing clinical and imaging information could enhance diagnostic accuracy by analyzing patients who underwent evaluation with FDG PET-CT imaging during a lung cancer evaluation in a case-control study design.
CTC analysis was performed in the context of a multicenter, prospective, observational study of CTCs in patients with a lung nodule or mass who underwent 18F-FDG PET-CT imaging for lung cancer evaluation. At or near to the time of PET-CT imaging, patients were enrolled at one of five medical centers: two US tertiary referral academic centers, Stanford and the University of California San Diego (UCSD); two US community hospitals, The Billings Clinic and California Pacific Medical Center (CPMC); and one US veteran’s hospital, the VA Palo Alto Health Care System (PAVAHCS). Three of these medical centers (Stanford, CPMC, PAVAHCS) aimed to investigate localized disease (i.e., tumors less than 4 cm in size), whereas the other two (The Billings Clinic and UCSD) were enrolling patients with more advanced disease who were undergoing medical oncology evaluation. Cases were defined by having NSCLC and controls defined as patients who underwent PET-CT but ultimately were diagnosed with a competing, benign diagnosis. Research at all participating facilities was approved by their respective institutional review board, and all patients were enrolled using informed consent according to the principles embodied in the Declaration of Helsinki.
Patient whole blood was collected into 10 ml Streck Cell Free DNA BCT (Streck, Omaha, NE) through a peripheral, upper extremity vein after discarding the first one milliliter to minimize skin tag contamination. Samples were shipped at ambient temperature and processed at The Scripps Research Institute (TSRI) within 48 hours. Before data analysis and integration of diagnosis with CTC data, the interpretation of the High-Definition Circulating Tumor Cell (HD-CTC) assay was performed without knowledge of the diagnosis in a single-blinded approach. We referred to the PRoBE biomarker guidelines to follow best practices during this study’s execution.29
Patients were analyzed as two separate cohorts for split-sample validation using the first set of consecutively enrolled patients undergoing FDG PET-CT imaging and CTC analysis as the training group and the next set of consecutively enrolled patients undergoing FDG PET-CT imaging who had CTC analysis performed as the test group. Since the HD-CTC test is not specific to lung cancer alone,18,22 any patient who had a competing diagnosis of another non-lung cancer, defined as undergoing evaluation or current treatment for it, was excluded to eliminate diagnostic confounding. All blood was drawn before any treatment for lung cancer and within 90 days of PET-CT. All medical centers used the same enrollment criteria, phlebotomy protocols, blood processing protocols, and clinical extraction parameters. Please note that we have previously published data for a subgroup of patients (n = 50) reported here regarding CTC enumeration and its association with tumor FDG uptake.24
We extracted clinical data including age, gender, ethnicity, cancer history, and smoking status. A patient was defined as a current smoker if they were smoking at the time of enrollment, past smoker if they ever reported smoking and were not smoking at the time of enrollment, and nonsmoker if they never smoked. Patients were followed over time through June 1, 2013, at all centers (median time 12.3 months, interquartile range [IQR] 3.7–16.7 months) and characterized as definitively malignant or benign, unknown, or lost to follow-up. NSCLC was determined by biopsy and/or surgery (n = 102), or clinical grounds (n = 2 patients). Benignity was defined by the extracting physician (V.S.N., G.H., L.B., and J.N.) after reviewing the medical record and included patients who had surgically resected nodules that were not cancer (n = 7), a biopsy yielding an alternative diagnosis (n = 6), nodules that diminished over time with or without noncancer-related treatments (n = 7), or radiographic benign nodules per report (n = 5).
Staging of cancer was extracted from the medical chart and defined by the most recent TNM staging system (American Joint Committee on Cancer [AJCC] v 7.0).30 Imaging data collected included maximum standard uptake value (SUVmax) of the lesion, maximum nodule diameter, and tumor location. For lung region, we analyzed upper and lower lung zones; right middle lobe tumors were classified as lower lung zone tumors. We did not partial volume correct for tumor SUVmax31 to simulate the clinical setting.
CTM Classification and CTC Enumeration
Detailed sample analysis for CTM and CTCs was performed as reported previously.18 The technologist, microscopes, and analysis systems were constant throughout this study. Blood samples underwent hemolysis, centrifugation, re-suspension, and plating onto four custom adhesion slides (Marienfeld, Lauda, Germany), followed by −80°C storage. The amount of blood plated onto slides was guided by a cell counter (Medonic M-Series Hematology Analyzer; Stockholm, Sweden) to approximate 10 million nucleated cells so as not to “underload” or “overload” the slide set, representing 1.5 ml of whole blood on average per sample.
Before analysis, slides were thawed, labeled by immunofluorescence (pan cytokeratin, CD45 and DAPI),18 and imaged by automated fluoroscopy. This was followed by manual validation by a pathology-trained technician (MSL). Morphology along with DAPI (+), CK (+), and CD45(−) intensities were defined for each channel to identify HD-CTCs as previously described (Fig. 1).21 Cells that only partially met criteria were not deemed to be an HD-CTC by the technologist but were recorded as well. This included cells that were smaller than an accepted HD-CTC (“Small” HD-CTC Candidates or SHCs) or dimmer by CK staining than a HD-CTC (“Dim” HD-CTC Candidates or DHCs). Thus, the HD-CTC platform was able to categorize HD-CTC populations and unique “CTC like” candidate cells for analysis as previously described.21
For circulating tumor microemboli (HD-CTM) evaluation, we defined groups of two HD-CTCs or more with touching cytoplasm as an HD-CTM over the same area.23 HD-CTMs were enumerated, the number of cells in HD-CTM were defined by the nuclei within them (using the DAPI counterstain), and HD-CTMs were then dichotomized as either present or not for further analysis.
Summary statistics and frequencies were generated as appropriate. Continuous variables are reported as their median and IQR for both parametric and nonparametric distributions. Differences between patients with NSCLC and benign nodules, as well as stage I disease only and benign lesions, were compared using a Student’s t test, Wilcoxon log-rank test, χ2 test, Fisher’s exact test, or Kruskal–Wallis test as appropriate and were annotated for a p value less than 0.05. Differences in HD-CTCs/CTM for other groups (i.e., time to biopsy, center, or histology) were also analyzed using the above tests as appropriate. For differences by histology, we grouped all non-adenocarcinomas together.
To analyze which HD-CTC assay derived features added value in addition to clinical and FDG PET-CT data, we calibrated several multiple logistic regression models using candidate HD-CTC derived variables. We also employed a Lasso model32 to agnostically select important HD-CTC features. The variables included for modeling were clinically derived (four total): age, sex, smoking, and cancer history; FDG PET-CT derived (three total): SUVmax, maximum lesion diameter, and lesion location; and HD-CTC assay derived (seven total): HD-CTCs, HD-CTM, and HD-CTC candidate cell features. Cell features and enumerated thresholds for determining malignant from benign disease were generated to optimize the accuracy of the HD-CTC test alongside traditional, clinical, and imaging parameters of risk for a solitary pulmonary nodule.2 Statistically significant variables were then carried forward to a test set of patients (n = 41) to confirm their importance in the model, in addition to being analyzed in the entire cohort.
Receiver operating characteristic (ROC) curves were generated to illustrate the performance of each model for distinguishing benign patients from all NSCLC patients or stage I patients only. These are reported in the results as the area under the curve (AUC) with 95% confidence intervals (CIs), where CIs for the ROC curves was calculated using the R package pROC.33 From this, we report model sensitivity and specificity in the training, test, and full data sets. We also report the likelihood ratios (LRs), positive predictive value (PPV), and negative predictive value (NPV) for CTM using a 2 × 2 contingency table.
For determining significant differences between models, the function roc.test () from the pROC package was used with bootstrapping (n = 1000). Results were also validated in all patients using 10-fold cross-validation (CV) to determine model stability.34 For this analysis, we considered CIs that did not overlap between models as a statistically significant result. This analysis was performed for all NSCLC cases and for stage I disease only versus benign cases. Lastly, risk scores for the most significant models were developed using all patients with the coefficients (ß) representing the contribution of the variable (x) to the risk model as follows:
Equation (Uncited)Image Tools
where the probability of cancer is equivalent to:
Equation (Uncited)Image Tools
All analyses were performed using SAS Enterprise Guide (v 4.3; Cary, NC) and R (v 3.0.1).35
A total of 170 patients who underwent phlebotomy for CTC analysis were assessed in this study (Fig. 2). Ultimately, 129 patients were eligible for further analysis after CTC assaying, diagnostic verification, and elimination of confounding cancers (Fig. 2). Of these 129 patients, 104 had a diagnosis of NSCLC and 25 had a benign diagnosis (Table 1). Of the 104 NSCLC patients, 80 had a diagnosis of stage I disease, whereas the remaining 24 had stage II–IV disease. The training cohort (n = 88) consisted of 71 cases (54 stage I patients) and 17 controls whereas the test cohort (n = 41) consisted of 33 cases (26 stage I patients) and 8 controls.
For all comers, median age was 69 years (IQR 11 years), 84% of patients were current or past smokers, and 63% of patients were male (Table 1). Overall lesion size for 104 NSCLCs and 25 benign lesions was 2.2 cm (IQR 1.4 cm) and the SUVmax was 4.5 (IQR 7.0). Eighty of the NSCLCs were stage I, whose predominant histology was adenocarcinoma (68%).
Significant differences existed between benign and malignant groups for age, tumor location, SUVmax, HD-CTC counts, and HD-CTM, whereas the training and test cohorts differed only in gender (Table 1). Training and test cohorts did not differ significantly by stage or histology. Notably, lesion size was not different between diagnostic groups but SUVmax was. Although 69 patients had a biopsy before treatment and 41 patients had a biopsy preceding phlebotomy for HD-CTC analysis (10 within 1 week preceding phlebotomy), there was no association between proximity of biopsy to blood draw for HD-CTC counts (p = 0.46) or HD-CTM (p = 0.62).
CTM and CTC Detection
The amount of whole blood analyzed for HD-CTM and HD-CTC detection was approximately 1.5 ml, and this did not vary by diagnosis or cohort (Table 1). A total of 4291 HD-CTCs were discovered in malignant disease (n = 104) versus 65 in benign disease (n = 25) and HD-CTCs ranged from 0 to 378 in the malignant group and from 0 to 16 in the benign group (Supplemental Table 1 and Supplemental Fig. 1, SDC1, http://links.lww.com/JTO/A622). HD-CTM ranged from 0 to 184 for malignant patients and 0 to 2 for benign lesions. (Supplemental Table 1 and Supplemental Fig. 1, SDC1, http://links.lww.com/JTO/A622).
For stage I tumors only, HD-CTCs ranged from 0 to 297 and HD-CTM ranged from 0 to 104. Of the 104 patients, 52 (50%) had HD-CTM for all NSCLCs, and 39 of 80 patients (49%) had HD-CTM in stage I disease only. There were no differences by histology or stage grouping for HD-CTC counts (p = 0.22 and 0.39, respectively) or HD-CTM (p = 0.22 and 0.65, respectively). No differences by enrollment center for HD-CTCs (p = 0.54) or HD-CTM (p = 0.86) were evident either.
Logistic Regression Models
Clinical variables, imaging features on PET-CT, and HD-CTCs/CTM in blood as individual predictors identified cancer patients with only marginal discrimination (AUC = 0.65–0.70). Inline with the existing literature,2 clinical and imaging data together yielded reasonable accuracy for a diagnosis of any NSCLC versus benign disease, or for stage I versus benign disease across training and test cohorts (Table 2). Notably, age, maximal tumor diameter on CT, and tumor SUVmax had the largest impact on the clinical model for all NSCLC patients and by stage I disease only (Supplemental Table 2, SDC1, http://links.lww.com/JTO/A622).
CTM alone had a positive LR of 10, negative LR of 0.5, PPV of 96%, and NPV of 31% for identifying NSCLC assuming a cancer prevalence documented in this cohort of 81%. Based on the superior diagnostic performance of HD-CTM (AUC = 0.70) over HD-CTC (AUC = 0.65) for diagnosis we assessed whether HD-CTM (present or absent) enhanced performance significantly when compared with the clinical diagnostic model alone (Table 2, Fig. 3). Ten-fold cross-validation confirmed that HD-CTM significantly added information to the clinical model when including all comers, or when restricting the analysis to stage I patients alone or those with stage I tumors smaller than 3 cm (Table 3).
We then proceeded to develop a risk score using the HD-CTM model’s coefficients (Supplemental Table 2, SDC1, http://links.lww.com/JTO/A622) for the relevant variable levels defined in Table 1:
Equation (Uncited)Image Tools
We also plotted individual patient risk to graphically illustrate the utility of this score for diagnostic refinement above and beyond clinical data (Supplemental Fig. 2, SDC1, http://links.lww.com/JTO/A622). Finally, we confirmed that HD-CTM were important features in augmenting diagnosis when using the agnostic Lasso model that selected age, SUVmax, and HD-CTC as the most important features for diagnosis from 14 entry variables in training, test, and all groups (Supplemental Fig. 3, SDC1, http://links.lww.com/JTO/A622).
We demonstrate here the diagnostic utility of an EpCAM-independent CTC platform assay that utilizes 1–2 ml of whole blood at room temperature within 48 hours of phlebotomy in a patient cohort undergoing PET-CT imaging for lung cancer evaluation. Although many investigators have been interested in using CTCs as prognostics in cancer, including NSCLC, we show that CTM may be a viable diagnostic when added to integrated clinical and imaging data in early-stage disease, and further, develop a risk score for diagnosis. To illustrate the utility of this score, we give the hypothetical example of a 71-year-old male former smoker, with no cancer history and a 1.7-cm lower lobe nodule whose FDG PET-CT SUVmax is 2.0 and whose blood reveals HD-CTM. Applying Equation3 and variable coefficients given in Supplemental Table 2 (SDC1, http://links.lww.com/JTO/A622), we see that this patient would increase his pretest probability of cancer from 58% to 89% with the addition of CTM data.
Although Tanaka et al.12 performed an important study using a similar patient cohort, they were unable to find a discriminating model using CTCs, and they did not integrate clinical or imaging data during analysis. Their negative results may be in part due to1) sensitivity limitations of the CellSearch platform compared to the HD-CTC platform—since it is dependent on EpCAM antibody affinity—and/or2) the lack of comparison to standard clinical variables of risk for identifying NSCLC patients, since orthogonally related biomarkers like FDG PET and CTCs24 appear to have additive value in our models.
The most discriminating models in our study included CTM. The rarity of such disease derived cell clusters, ranging from a few cells aggregated together to “mega-clusters,” and recent molecular characterizations describing their EMT phenotype suggest that clusters may be a more “cancerous” subtype of putative CTCs.13,36,37 Our data recapitulated this clinically as CTM were the strongest diagnostic in our models using not only a candidate-driven approach, but when employing a lasso model that agnostically selected CTM as the most important HD-CTC feature along with other clinical data.
All diagnostic tests have limited sensitivity and specificity, whether novel or decades old. Original data published using CellSearch were remarkably discriminatory when using healthy patients and nonmalignant disease controls,38 but even this data suggested nonmalignant disease resulted in more false positive results. Additional recent studies have suggested an increased number of false positive results in patients with competing diagnoses undergoing a work-up for colon and lung cancer.12,39 Furthermore, studies assessing single cell genomics that require more rigor when identifying cells have documented the need to purify putative “CTCs” using CellSearch resulting from up to a 30% false positive rate.40
We also found that, unsurprisingly, the HD-CTC assay will detect CTC-like events in high-risk patients with other competing diagnoses (Supplemental Table 1, SDC1, http://links.lww.com/JTO/A622). This suggests that, although enumeration appears to be clinically useful, even sensitive methods for detecting CTCs will benefit from additional molecular characterization to differentiate circulating epithelial cells (CECs) from CTCs regardless of platform type. Using additional protein biomarkers, next generation sequencing for single mutational or whole genome copy number variation analysis to define cells with pathognomonic hallmarks of cancer is one way to approach this issue that we are currently investigating. Whether inconsequential CECs arise from competing inflammatory lesions other than lung cancer or clinically undetected premalignant lesions remains a question that will be answered over time as the patient cohort matures.
Finally, although we believe consecutive enrollment of patients who underwent PET-CT and CTC analysis at multiple centers across the western United States is a strength of this study, we acknowledge that its observational nature may have led to several biases in the collection of the data that render its findings preliminary. These include (1) an inherent spectrum bias at participating centers; (2) the inclusion of patients who only underwent PET-CT that may have led to identifying larger benign lesions (i.e., higher risk lesions, see Supplement Figures 1 and 2, SDC1, http://links.lww.com/JTO/A622) than one would normally see in an evaluation setting; and (3) an additional bias toward identifying larger benign lesions that had a more expeditious work-up compared to smaller lesions that remain indeterminate (and thusly excluded from our analysis) and are being followed over time.
Putative CTM detected using standard cell markers and cell morphology in the HD-CTC platform were useful for risk-stratifying patients undergoing an evaluation for lung cancer and augmented clinical models alone. Although strict enumeration of this important blood biomarker appears to be clinically useful in the appropriate patient population, special attention is required in high-risk populations since CECs resulting from nonmalignant conditions may alter test performance. Advancing fluid biopsies originally developed for disease monitoring in patients with cancer into the setting of primary diagnosis requires additional validation and refinement of test performance in larger cohorts before prospective trials to establish clinical utility.
The authors thank the Molecular Imaging Program and MultiModality Imaging Laboratory at Stanford; Pragya Tripathi, Julie Loero, and the Nuclear Medicine Staff at the PAVAHCS; Claude James at the UCSD Moores Cancer Center; Steve Wharton and Tricia Montgomery at The Billings Clinic; Benjamin Luna and Jamey Schmidt at the CPMC Research Institute; Daniel Lazar, Thomas Metzner, Rachelle Lamy, Loressa Uson, Julia Li, Luisa Fernandez Altuna, Natalie Felch, Janett Stoehr, Nadia Ebrahim, and Newsha Sahaf at TSRI. Lastly, to our patients who consented for this study, we thank you for advancing our understanding of early-stage lung cancer.
1. Aberle DR, Adams AM, Berg CD, et al.National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011;365:395–409
2. Schultz EM, Sanders GD, Trotter PR, et al. Validation of two models to estimate the probability of malignancy in patients with solitary pulmonary nodules. Thorax. 2008;63:335–341
3. Li XJ, Hayward C, Fong PY, et al. A blood-based proteomic classifier for the molecular characterization of pulmonary nodules. Sci Transl Med. 2013;5:207ra142
4. Asworth T. A case of cancer in which cells similar to those in the tumours were seen in the blood after death. Australian Med J. 1869;14:146–147
5. Takahashi M. An experimental study of metastasis. J Pathol Bacteriol. 1915;20:1–13
6. Romsdahl MD, Chu EW, Hume R, Smith RR. The time of metastasis and release of circulating tumor cells as determined in an experimental system. Cancer. 1961;14:883–888
7. Cristofanilli M, Budd GT, Ellis MJ, et al. Circulating tumor cells, disease progression, and survival in metastatic breast cancer. N Engl J Med. 2004;351:781–791
8. Cristofanilli M, Hayes DF, Budd GT, et al. Circulating tumor cells: a novel prognostic factor for newly diagnosed metastatic breast cancer. J Clin Oncol. 2005;23:1420–1430
9. Lucci A, Hall CS, Lodhi AK, et al. Circulating tumour cells in non-metastatic breast cancer: a prospective study. Lancet Oncol. 2012;13:688–695
10. Krebs MG, Sloane R, Priest L, et al. Evaluation and prognostic significance of circulating tumor cells in patients with non-small-cell lung cancer. J Clin Oncol. 2011;29:1556–1563
11. Tanaka F YK, Hasegawa S. Circulating tumor cells (CTCs) in lung cancer: current status and future perspectives. Lung Cancer: Targets and Therapy. 2010;1:77–84
12. Tanaka F, Yoneda K, Kondo N, et al. Circulating tumor cell as a diagnostic marker in primary lung cancer. Clin Cancer Res. 2009;15:6980–6986
13. Yu M, Bardia A, Wittner BS, et al. Circulating breast tumor cells exhibit dynamic changes in epithelial and mesenchymal composition. Science. 2013;339:580–584
14. Nagrath S, Sequist LV, Maheswaran S, et al. Isolation of rare circulating tumour cells in cancer patients by microchip technology. Nature. 2007;450:1235–1239
15. Hofman V, Bonnetaud C, Ilie MI, et al. Preoperative circulating tumor cell detection using the isolation by size of epithelial tumor cell method for patients with lung cancer is a new prognostic biomarker. Clin Cancer Res. 2011;17:827–835
16. Krivacic RT, Ladanyi A, Curry DN, et al. A rare-cell detector for cancer. Proc Natl Acad Sci U S A. 2004;101:10501–10504
17. Kuhn P, Bethel K. A fluid biopsy as investigating technology for the fluid phase of solid tumors. Phys Biol. 2012;9:010301
18. Marrinucci D, Bethel K, Kolatkar A, et al. Fluid biopsy in patients with metastatic prostate, pancreatic and breast cancers. Phys Biol. 2012;9:016003
19. Marrinucci D, Bethel K, Luttgen M, Bruce RH, Nieva J, Kuhn P. Circulating tumor cells from well-differentiated lung adenocarcinoma retain cytomorphologic features of primary tumor type. Arch Pathol Lab Med. 2009;133:1468–1471
20. Lazar DC, Cho EH, Luttgen MS, et al. Cytometric comparisons between circulating tumor cells from prostate cancer patients and the prostate-tumor-derived LNCaP cell line. Phys Biol. 2012;9:016002
21. Nieva J, Wendel M, Luttgen MS, et al. High-definition imaging of circulating tumor cells and associated cellular events in non-small cell lung cancer patients: a longitudinal analysis. Phys Biol. 2012;9:016004
22. Wendel M, Bazhenova L, Boshuizen R, et al. Fluid biopsy for circulating tumor cell identification in patients with early-and late-stage non-small cell lung cancer: a glimpse into lung cancer biology. Phys Biol. 2012;9:016005
23. Cho EH, Wendel M, Luttgen M, et al. Characterization of circulating tumor cell aggregates identified in patients with epithelial tumors. Phys Biol. 2012;9:016001
24. Nair VS, Keu KV, Luttgen MS, et al. An observational study of circulating tumor cells and (18)F-FDG PET uptake in patients with treatment-naive non-small cell lung cancer. PLoS One. 2013;8:e67733
25. Gould MK, Donington J, Lynch WR, et al. Evaluation of individuals with pulmonary nodules: when is it lung cancer? Diagnosis and management of lung cancer, 3rd ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest. 2013;143(5 Suppl):e93S–120S
26. MacMahon H, Austin JH, Gamsu G, et al.Fleischner Society. Guidelines for management of small pulmonary nodules detected on CT scans: a statement from the Fleischner Society. Radiology. 2005;237:395–400
27. Naidich DP, Bankier AA, MacMahon H, et al. Recommendations for the management of subsolid pulmonary nodules detected at CT: a statement from the Fleischner Society. Radiology. 2013;266:304–317
28. Ettinger DS, Akerley W, Borghaei H, et al.NCCN (National Comprehensive Cancer Network). Non-small cell lung cancer. J Natl Compr Canc Netw. 2012;10:1236–1271
29. Pepe MS, Feng Z, Janes H, Bossuyt PM, Potter JD. Pivotal evaluation of the accuracy of a biomarker used for classification or prediction: standards for study design. J Natl Cancer Inst. 2008;100:1432–1438
30. Rami-Porta R, Ball D, Crowley J, et al.International Staging Committee; Cancer Research and Biostatistics; Observers to the Committee; Participating Institutions. The IASLC Lung Cancer Staging Project: proposals for the revision of the T descriptors in the forthcoming (seventh) edition of the TNM classification for lung cancer. J Thorac Oncol. 2007;2:593–602
31. Shankar LK, Hoffman JM, Bacharach S, et al.National Cancer Institute. Consensus recommendations for the use of 18F-FDG PET as an indicator of therapeutic response in patients in National Cancer Institute Trials. J Nucl Med. 2006;47:1059–1066
32. Tibshirani R. Regression shrinkage and selection via the lasso. J Royal Statist Soc B. 1996;58:267–288
33. Robin X, Turck N, Hainard A, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77
34. Picard R, Cook D. Cross-validation of regression models. J Am Stat Assoc. 1984;79:575–583
35. Ihaka S, Gentleman R. R: a language for data analysis and graphics. J Comput Graph Stat. 1996;5:299–314
36. Liotta LA, Saidel MG, Kleinerman J. The significance of hematogenous tumor cell clumps in the metastatic process. Cancer Res. 1976;36:889–894
37. Brandt B, Junker R, Griwatz C, et al. Isolation of prostate-derived single cells and cell clusters from human peripheral blood. Cancer Res. 1996;56:4556–4561
38. Allard WJ, Matera J, Miller MC, et al. Tumor cells circulate in the peripheral blood of all major carcinomas but not in healthy subjects or patients with nonmalignant diseases. Clin Cancer Res. 2004;10:6897–6904
39. Pantel K, Denève E, Nocca D, et al. Circulating epithelial cells in patients with benign colon diseases. Clin Chem. 2012;58:936–940
40. Ni X, Zhuo M, Su Z, et al. Reproducible copy number variation patterns among single circulating tumor cells of lung cancer patients. Proc Natl Acad Sci U S A. 2013;110:21083–21088
NSCLC; CTC; diagnostic; stage I; lung nodule
Supplemental Digital Content
Copyright © 2014 by the European Lung Cancer Conference and the International Association for the Study of Lung Cancer.
Highlight selected keywords in the article text.