PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of jcoHomeThis ArticleSearchSubmitASCO JCO Homepage
 
J Clin Oncol. 2009 June 10; 27(17): 2787–2792.
Published online 2009 May 4. doi:  10.1200/JCO.2008.19.4233
PMCID: PMC2698017

Connective Tissue-Activating Peptide III: A Novel Blood Biomarker for Early Lung Cancer Detection

Abstract

Purpose

There are no reliable blood biomarkers to detect early lung cancer. We used a novel strategy that allows discovery of differentially present proteins against a complex and variable background.

Methods

Mass spectrometry analyses of paired pulmonary venous-radial arterial blood from 16 lung cancer patients were applied to identify plasma proteins potentially derived from the tumor microenvironment. Two differentially expressed proteins were confirmed in 64 paired venous-arterial blood samples using an immunoassay. Twenty-eight pre- and postsurgical resection peripheral blood samples and two independent, blinded sets of plasma from 149 participants in a lung cancer screening study (49 lung cancers and 100 controls) and 266 participants from the National Heart Lung and Blood Institute Lung Health Study (45 lung cancer and 221 matched controls) determined the accuracy of the two protein markers to detect subclinical lung cancer.

Results

Connective tissue-activating peptide III (CTAP III)/ neutrophil activating protein-2 (NAP-2) and haptoglobin were identified to be significantly higher in venous than in arterial blood. CTAP III/NAP-2 levels decreased after tumor resection (P = .01). In two independent population cohorts, CTAP III/NAP-2 was significantly associated with lung cancer and improved the accuracy of a lung cancer risk prediction model that included age, smoking, lung function (FEV1), and an interaction term between FEV1 and CTAP III/NAP-2 (area under the curve, 0.84; 95% CI, 0.77 to 0.91) compared to CAPIII/NAP-2 alone.

Conclusion

We identified CTAP III/NAP-2 as a novel biomarker to detect preclinical lung cancer. The study underscores the importance of applying blood biomarkers as part of a multimodal lung cancer risk prediction model instead of as stand-alone tests.

INTRODUCTION

Lung cancer is the most common cause of cancer deaths worldwide with more than 1.2 million people dying of the disease annually.1,2 The 5-year survival rate is only 16% for patients diagnosed with advanced disease compared with 70% to 90% that can be achieved when lung cancer is diagnosed and treated at an earlier stage.35 Early detection and treatment of lung cancer is a promising strategy to reduce lung cancer mortality. Technologies such as spiral computed tomography (CT) and autofluorescence bronchosocopy can detect lung cancers down to the submillimeter range.6,7 However, the wide variations in lung cancer risk, even among long-term smokers,8 makes these sensitive technologies neither practical nor cost-effective as screening tools in the general population. Application of a filter to identify smokers at the highest risk for lung cancer may improve the positive predictive value of these screening tools.9,10 A blood-based biomarker is an attractive filter because blood is easily accessible and measurements may be repeated over time.1113 Major impediments in the discovery and validation of biomarkers for detection of preclinical lung cancer have included: measurement of thousands of proteins simultaneously in tens of samples, resulting in false positives; use of analytic methods that do not provide precise and accurate determination of potential tumor specific proteins that are expressed in much lower concentrations than other more abundant proteins resulting in false negatives,14,15,16 and a lack of access to blood samples collected in population-based studies before clinical diagnosis of cancer for validation and replication. Herein, we describe a novel approach to biomarker discovery that used the same subject as his/her own control to identify elevated proteins in the pulmonary venous effluent draining the tumor vascular bed compared to matched systemic arterial blood. This approach allows the differentially present proteins to be identified against a complex and variable background. The analytic issue is reduced to determining what is changed in an individual pre- and post- passage through the affected organ to get around the problem of finding low abundance markers in blood. Two of the most differentially present proteins identified were then validated for early detection of lung cancer using peripheral venous blood samples from two independent population-based studies from ever smokers not known to have lung cancer at the time of the blood collection. Finally, we evaluated the clinical value of these biomarkers to detect lung cancer in the context of a multimodal lung cancer risk model that incorporates demographic, clinical factors, and biomarkers.

METHODS

Study Design

At the time of thoracotomy and resection of the tumor, two blood samples were obtained from the lobar pulmonary vein that received drainage directly from the tumor containing lung segment that should contain the highest concentration of a candidate biomarker; and from the radial artery which represents the systemic circulation (Fig 1). Serum samples from pulmonary venous and systemic blood were fractionated and analyzed using surface-enhanced laser desorption ionization time-of-flight mass spectroscopy (Appendix Fig A1, online only). Levels of two proteins were significantly increased in pulmonary venous compared to systemic blood from the radial artery. These proteins were identified as connective tissue-activating peptide III( CTAP III) and haptoglobin using tandem MS (MS/MS; Appendix Fig A2, online only). Enzyme-linked immunosorbent assay (ELISA) was used to validate that the proteins identified were differentially expressed between pulmonary venous and systemic blood and that their levels changed following surgical removal of the lung tumor. Finally, their concentrations in peripheral venous blood were compared between heavy smokers who did and did not develop lung cancer using blood samples from two independent cohorts: a lung cancer prevention study at the British Columbia Cancer Agency and the National Heart Lung and Blood Institute Lung Health Study (LHS).17 The incremental value of these biomarkers to demographic and clinical factors was evaluated for detection of preclinical lung cancer.

Fig 1.
Overview of the study design. Proteomic profiling using surface-enhanced laser desorption ionization time-of-flight mass spectroscopy (SELDI-TOF MS) was used on matched sera from pulmonary vein draining the tumor and systemic blood from 16 subjects at ...

Study Participants

The demographics, histological cell type, and stage of lung cancer of the 16 patients in the first part (ie, discovery phase) and the 64 patients in the second part of the study (ie, confirmatory phase using ELISA) are provided in Appendix Table A1 (online only). The participants in the validation study were from two separate population-based cohorts. The first cohort comprised of volunteers between 45 to 74 years of age with a smoking history of ≥ 30 pack-years. They were initially screened between 2000 and 2006 for enrollment into one of several National Cancer Institute (NCI)–sponsored lung cancer chemoprevention trials (NIH- NCI contract N01-CN-85188 and NCI grant PO1-CA96964, U01CA96109). At the time of the peripheral venous blood collection, none of the study participants had a clinical diagnosis of lung cancer. Lung cancer was subsequently diagnosed in 49 participants. The median interval from blood collection to the diagnosis of lung cancer was 6 months (interquartile range, 2 to 29 months; Table 1). As controls, blood samples from 100 smokers without lung cancer from the same screening cohorts were randomly selected and used for the biomarker validation study. Twenty-eight of the 49 participants with lung cancer also had blood samples available after surgical resection of their tumor (Appendix Table A1). Archival blood samples from a second population based cohort (LHS)17,18 were also used for validation of the biomarkers and for determining the incremental value of these biomarkers to demographic and clinical factors. The LHS blood samples were collected between 1992 and 1994. We performed a case-control study wherein we identified 45 smokers who died of lung cancer within 5 years of their blood sampling and 221 control smokers without lung cancer who were matched for age, sex, race, smoking status (which was validated by salivary cotinine levels), body mass index (BMI), and lung function (forced expiratory volume in one second [FEV1 ] as percent of predicted). We matched at least five controls for every patient.19 The characteristics of these participants are presented in Table 2. The median interval from blood collection to lung cancer death in this cohort was 39 months (interquartile range, 26 to 49 months). We matched at least 5 controls for every case. Informed consent was obtained from the participants. The study was approved by the research ethics board of the University of British Columbia.

Table 1.
Demographic and Clinical Data of Lung Cancer Chemoprevention Study Participants
Table 2.
Clinical Characteristics of Subjects Who Died From Lung Cancer and Matched Controls From the National Heart Lung and Blood Institute Lung Health Study

Measurement of Proteins Using ELISA

Levels of CTAP III in plasma samples were measured using an ELISA kit against human CTAP III/neutrophil activating protein-2 (NAP-2), the c-terminal 70 amino acid region which is present in all pro-platelet basic protein species (DuoSet, R&D Systems, Minneapolis, MN). Haptoglobin that was found to be differentially expressed in part one of this study and proteins that have been cited in the literature as promising biomarkers for lung cancer such as C-reactive protein (CRP), serum amyloid A (SAA), and alpha-1 antitrypsin were also measured using ELISA kits in accordance with the manufacturer's instructions.1113,2024

Statistical Analysis

A Wilcoxon signed rank test was used to compare protein levels between the pulmonary venous and systemic blood in the same patient undergoing surgery for lung cancer and the protein levels before and after surgery. A Wilcoxon rank sum test compared protein levels between subjects who did and did not develop lung cancer in the Lung Cancer Prevention Study. Multiple logistic regression modeling was employed to describe the relationship between levels of proteins and the risk of lung cancer, adjusted for age, sex, smoking status, and FEV1 % predicted. A stepwise model selection process was used to arrive at a parsimonious model. Receiver operating characteristic (ROC) curves were plotted to evaluate the sensitivity and specificity of the biomarker measurements in predicting lung cancer. For the analysis of the matched nested case control samples from LHS, we used conditional logistic regression to model the instantaneous rates of lung cancer mortality.25 A two-tailed P value less than .05 was considered significant. All analyses were conducted using SAS version 9.1 (SAS Institute, Cary, NC) and R 2.5.1 (http://www.r-project.org/). Continuous variables are expressed as mean with or without standard deviaion unless otherwise indicated.

RESULTS

Differences in CTAP III/NAP2 and Haptoglobin in Pulmonary Venous Versus Systemic Blood of Lung Cancer Patients

On identification of CTAP III and haptoglobin by MS/MS from the initial 16 paired venous-arterial sera (Appendix), we determined the concentrations of these proteins in paired venous-arterial blood samples from 64 patients (Appendix Table A1) using ELISA. CTAP III/NAP-2 levels were significantly higher in the pulmonary venous blood compared to systemic blood (P < .001; Appendix Fig A3A, online only). The median difference in CTAP III/NAP-2 between the pulmonary venous and the systemic blood was 4.93 ug/mL with an interquartile range of 3.42 to 5.74 ug/mL. Haptoglobin levels were also higher in the pulmonary venous compared to systemic blood (P < .008), but the differences were less than those observed with CTAP III/NAP-2 (Appendix Fig A3A). The median difference in haptoglobin levels between the pulmonary venous and systemic blood was 0.32 mg/mL (interquartile range, −0.09 to 0.60 mg/mL).

Correlation of CTAP III/NAP-2 and Haptoglobin in the Plasma of Lung Cancer Patients Before and After Surgical Resection

Twenty-eight subjects confirmed to have lung cancer (Appendix Table A1) had plasma samples collected before and after surgical resection. Levels of CTAP III/NAP-2 decreased significantly after surgery (geometric mean before surgery v after surgery, 3.22v1.40 ug/mL; a reduction of 57%; P = .010). Of these 28 patients, five experienced a recurrence of lung cancer after surgery. In these individuals, though at the time of blood sampling, recurrence was not known, CTAP III/NAP-2 levels failed to decrease significantly (geometric mean before surgery v after surgery, 4.14v3.47 μg/mL; P = .107). In contrast, patients who remained disease free had significant reduction in plasma NAP-2 levels after surgery compared with presurgical levels (presurgical geometric mean v postsurgical mean, 3.05 μg/mL v1.15 μg/mL; P = .002; Appendix Fig A3B). Haptoglobin levels did not change significantly after surgery relative to presurgical levels (P = .46; data not shown).

CTAP III/NAP-2 and Haptoglobin Levels in Subjects Who Did, and Did Not Develop Lung Cancer in a Cancer Prevention Study

Forty-nine subjects participating in the lung cancer prevention cohort were found to have lung cancer in screening studies using low-dose spiral CT and/or autofluorescence bronchoscopy. Of these patients, 47% had stage 0 or IA non–small-cell lung cancer. The clinical characteristics of these subjects and the 100 random controls from the same cohort are presented in Table 1. The median level of CTAP III/NAP-2 in peripheral venous blood was 3.15 μg/mL (interquartile range, 1.44 to 3.92 μg/mL) in subjects who developed lung cancer, while it was 0.59 μg/mL (interquartile range, 0.89 to 3.12 μg/mL) for subjects who were free of lung cancer (P = .004; comparing median levels of CTAP III/NAP-2 between the groups; Appendix, Fig A4, online only). Platelet counts were similar between the two groups (258 ± 89v235 ± 45 giga/L, cancer vcontrols, respectively; P = .092); however, there was a modest correlation between the platelet count and NAP-2 levels (Spearman correlation = 0.3; P = .026). CTAP III/NAP-2 did not vary as a function of the tumor TNM stage (P = .936), histological cell type, or smoking status (Appendix Fig A4). There was no significant change in the levels of CTAP III/NAP-2 up to 30 months before the diagnosis of lung cancer.

Several biomarkers including haptoglobin were measured for comparison. The median haptoglobin level was 1.66 mg/mL (interquartile range, 1.08 to 1.97 mg/mL) for subjects who developed lung cancer, while it was 1.06 mg/mL (interquartile range, 0.86 to 1.48) for subjects who were free of lung cancer (P < .001 for the two group comparisons). Other biomarkers such as CRP, alpha-1 antitrypsin, and SAA were not significantly different between the cancer and noncancer subjects (Appendix Fig A4).

The fitted logistic regression model demonstrates the relationship between CTAP III/NAP-2 and the risk of lung cancer as a function of the subjects' baseline FEV1 (% predicted) in the lung cancer prevention study. The interaction term between CTAP III/NAP-2 and FEV1 (% predicted) on the risk of lung cancer was negative (coefficient, −0.03) and significant (P = .009), which indicated that the effect of CTAP III/NAP-2 on the risk of lung cancer was amplified as FEV1 % of predicted decreased (Fig 2A). Plasma haptoglobin was also associated with increased risk of lung cancer (Appendix Table A2, online only). In a replication study using the National Heart Lung and Blood Institute LHS samples (Table 2), we found that CTAP III/NAP-2 was significantly associated with lung cancer mortality (P = .021) when an interaction term was introduced for FEV1 % predicted (Fig 2B). Similarly, serum haptoglobin was associated with lung cancer mortality (P = .016) when an interaction term was introduced for FEV1 % predicted (data not shown).

Fig 2.
A fitted line showing the relationship between the risk of lung cancer and connective tissue-activating peptide III (CTAP III)/ neutrophil activating protein-2 (NAP-2) as a function of forced expiratory volume in one second adjusted for age, sex, race, ...

The ROC curves were constructed for the clinical factors and biomarkers (Appendix Fig A5, online only). The area under curve (AUC) of the ROC curve for CTAP III/NAP-2 was 0.64 (95% CI, 0.55 to 0.74) while that for haptoglobin was 0.70 (95% CI, 0.61 to 0.79). The AUC for the CRP ROC curve was 0.56 (95% CI, 0.46 to 0.66), and that for SAA was 0.48 (95% CI, 0.38 to 0.58). The AUC of age, smoking status, and FEV1 % predicted combined was 0.80 (95% CI, 0.72 to 0.88). Inclusion of CTAP III/NAP-2 into this model increased the AUC to 0.81 (95% CI, 0.73 to 0.89), while inclusion of haptoglobin increased the AUC to 0.82 (95% CI, 0.74 to 0.90). Simultaneous inclusion of both CTAP III/NAP-2 and haptoglobin plus an interaction term with FEV1 % predicted increased the AUC to 0.84 (95% CI, 0.77 to 0.91;Fig 3). Using a threshold of 2.95 μg/mL for CTAP III/NAP-2, the positive predictive value (PPV) was 50.0% while the negative predictive value (NPV) was 77.4%. When CTAP III/NAP-2, haptoglobin, and other covariates were combined together, the PPV increased to 62.7% and the NPV increased to 88.5%.

Fig 3.
The receiver operating characteristics curve combining clinical factors and biomarkers in the Lung Cancer Prevention Study. Addition of age, sex, and forced expiratory volume in one second (FEV1) to haptoglobin and connective tissue-activating peptide ...

DISCUSSION

Application of several mass spectrometry approaches provided an unbiased discovery approach to identify proteins that are elevated after passage through the tumor microenvironment. This led to the discovery of a novel biomarker CTAP III/NAP-2 and a previously reported biomarker—haptoglobin—as potential biomarkers for detection of lung cancer. Increased levels of CTAP III/NAP-2 in the plasma of smokers who subsequently developed lung cancer were demonstrated in two separate, independent population-based cohorts without a known diagnosis of lung cancer when the blood samples were taken. Elevated blood levels of CTAP III/NAP-2 predated the clinical diagnosis of lung cancer by up to 29 months. Along with clinical characteristics such as age, lung function, and smoking status, CTAP III/NAP-2 and haptoglobin can predict the presence of lung cancer with a PPV of 63% and a NPV of 89%. A prospective study to study the incremental benefit of these blood biomarkers as part of a multimodal lung cancer prediction model to stratify high-risk smokers for lung cancer screening with relatively expensive yet sensitive methods such as low-dose spiral CT and autofluorescence bronchoscopy is currently under investigation in a Canada-wide early lung cancer detection study.

Another important finding is a significant decrease in CTAP III/NAP-2 after curative surgical resection but persistence of elevated levels in those who developed recurrent disease after surgery. The potential utility of this biomarker in postoperative surveillance for microscopic residual disease that would lead to clinical recurrence merits further study. A third important finding is the significant interaction between CTAP III/NAP-2 and FEV1. This interaction would have been missed if the biomarkers were evaluated as a stand-alone lung cancer detection test instead of as part of a multimodal lung cancer risk model. The relationship between the risk of lung cancer and chronic obstructive pulmonary disease has long been recognized.26 While an inflammatory link between the two diseases that share a common etiology has been hypothesized, this is the first study that shows a significant interaction between CTAP III/NAP-2, decline in lung function (FEV1 %) and lung cancer risk.

CTAP III belongs to the subfamily of ELR+ CXC chemokines that are potent promoters of angiogenesis, tumorigenesis, and metastases.27 Most of the work on the role of CXC chemokines in lung cancer has been on CXCL5 (ENA-78), CXCL8 (IL-8), and CXCL1.2830 There is a paucity of information on CXCL7 and lung cancer. Initially, CXCL7 was thought to be expressed only within the megakaryocyte lineage.31,32 Recent repots suggest other cell types such as monocytes, lymphocytes, and neutrophils may produce this chemokine as well.33,34 CXCL7 has heparanase activity.35 The very recent finding that premalignant breast cancer cells transfected with CXCL7 became as invasive as malignant breast cells suggest an important role of this chemokine in the tumor invasion process.36 The ability to detect lung cancer at the preinvasive or early invasive stage is key to success of any early detection program. Of significance is that 47% of the subjects in our validation cohort had stage 0/IA lung cancer. This is the first report of a blood biomarker that can detect stage 0 lung cancer.

The lung is a major site of extrahepatic synthesis of haptoglobin. As a major acute-phase reactant, haptoglobin increases in plasma during inflammation and malignancy such as ovarian cancer.20 Although haptoglobin was significantly elevated in patients with lung cancer compared with subjects without lung cancer, the levels did not change significantly after surgery suggesting that haptoglobin is a less specific indicator of lung cancer. In this study, we did not find an association of other acute phase reactants such as serum amyloid A and CRP1113,2123 with lung cancer probably because we performed the tests in population-based cohorts not known to have lung cancer rather than patients with a clinical diagnosis of lung cancer.

In applying our blood biomarkers to early lung cancer detection, we developed a model similar to the highly successful Framingham model to predict cardiovascular disease risk.37 We combined CTAP III/NAP-2, and haptoglobin along with age, smoking status, and FEV1 in our risk model and found an area under the curve for lung cancer of 84% (Fig 3). Our study thus underscores the importance of applying blood biomarkers not as a stand-alone test but as part of a multimodal risk lung cancer prediction model.

Acknowledgment

We thank H.A. Hare, T. Tam, Sukhinder Khattra, and Yuexin Yi for their technical assistance.

Appendix

Blood Collection and Processing

In the surgical patients, blood (10 mL) was collected simultaneously from both the pulmonary vein (draining the tumor-containing segment) and radial artery in a serum separator tube, clotted for 30 minutes, centrifuged, aliquoted, and flash frozen at −80°C. The paired sera were fractionated by isoelectric point and analyzed by surface-enhanced laser desorption ionization time-of-flight mass spectroscopy (SELDI-TOF-MS) or by Enzyme-linked immunosorbent assay (ELISA). In the lung cancer screening study and Lung Health Study participants, blood from a peripheral vein were collected into K2 EDTA tubes and centrifuged immediately at 4°C. The resultant supernatant plasma was transferred into two cryotubes until assay. All blood samples were processed and stored at −80°C within 2 hours after blood draw.

Serum Fractionation for SELDI-TOF-MS Analysis

Aliquots of serum sample were centrifuged (10 minutes at 4°C, 20,000 × g) to remove insoluble material. Samples (20 μL) were denatured for 1 hour at 4°C with 30 μL of U9 buffer (9 M Urea, 2% 3-[(3-Cholamidopropyl)dimethylammonio]-1-propanesulfonate (3-[3-cholamidopropyl-dimethylammonio]-1-propane-sulfonate), 50 mmol/L Tris-HCl pH 9). Anion exchange chromatography was performed in parallel using a ProteinChip Serum Fractionation kit (Bio-Rad Laboratories, Hercules, CA) with a 96-well filter plate containing dessicated anion exchange resin as per the manufacturer's instructions. After rehydration, washing, and equilibration of the anion exchange resin, denatured serum samples were applied and allowed to bind for 30 minutes at 4°C with shaking. Six fractions were collected from each sample. We used Fraction 1 containing the unbound protein flow-through pooled with a pH 9 elution.

SELDI-TOF-MS Analysis

Weak cation exchange ProteinChip arrays were used to bind the proteins in Fraction 1 using a bioprocessor reservoir as per manufacturer's instructions (CM10 ProteinChip Array Kit, Bio-Rad Laboratories). The limit of detection of the ProteinChip surfaces has been determined to typically be in the low femtomole range with a linear response over 2 to 3 orders of magnitude (Diamond DL, Kimball JR, Krisanaprakornkit S, et al: J Immunol Methods 256:65-76, 2001; Xiao Z, Jiang X, Beckett ML, et al: Protein Expr Purif 19:12-21, 2000). The average percent coefficient of variation (%CV) observed for peaks across the m/z 2,500 to 150,000 range have been shown to be better than 25% and peaks in the m/z 10,000 to 15,000 range exhibited less variation with an average %CV of 20% (Le L, Chi K, Tyldesley S, et al: Clin Chem 51:695-707, 2005). Ten μl of fraction 1 samples was added to 90 μL of low stringency CM10 binding buffer (0.1 M sodium acetate, pH 4.0) to washed and equilibrated CM10 arrays. Binding was conducted with vigorous shaking for 1 hour, followed by three washes with binding buffer and a final water wash. ProteinChip arrays were air dried before the addition of sinapinic acid (SPA) matrix (12.5 mg/mL SPA, 50% v/v ACN, 0.5% v/v TFA). Prepared ProteinChip arrays were analyzed using a SELDI-TOF MS (PBSII, Ciphergen Biosystems, Fremont, CA) externally calibrated using an All-in-One Protein Standard (Ciphergen Biosystems Inc). Spectra were generated using an average of 100 laser shots using laser intensities of 225 or 250. The resulting spectra were externally calibrated, baseline subtracted, and normalized to total ion current using Ciphergen Express software (Ciphergen Biosystems Inc). Peaks were autodetected between a mass range of m/z 2,000 to 200,000. First pass criteria requiring a signal to noise ratio (S/N) higher than 5.0 was used to identify well defined peaks. A second less stringent pass (S/N > 2.5) was employed to improve the ability of the peak detection algorithm to detect low intensity peaks differing between groups. Processed spectra were analyzed using Ciphergen Express data analysis software. Normalized peaks from the pulmonary veins (test, positive group for ROC analysis) were compared to those detected in the systemic blood (radial artery).

Identification of Proteins That Were Increased in Samples From the Pulmonary Lobar Vein Draining the Tumor Compared With Systemic Blood

SELDI-TOF-MS profiles from 16 paired venous-arterial sera were compared. We focused on one peak at approximately m/z 9,320 that was significantly elevated in fraction 1 in pulmonary venous samples compared with systemic samples (Appendix Fig A1A, online only). Spectra were normalized based on total ion current to allow direct comparison of the differences in the level of intensity for the m/z 9,320 peak between spectra. The spectra were overlaid to aid in the visualization of the pronounced differences in levels of this peak between the two sources of blood. Enhanced analysis of this region of the spectra is shown for m/z 9,320 for matched samples from the patient with the highest peak intensity of all samples (Appendix Fig A1B). This patient had a stage II non–small-lung cancer. There was no feature in this patient that could distinguish him from the remaining 15 in terms of cell type or tumor stage. Statistical analysis of peak intensities of the spectra revealed the differences observed at m/z 9,320 in venous serum samples of patients were significantly increased compared to the arterial samples (P = .0023; Appendix Fig A1C).

Protein Isolation

To identify the protein at m/z 9,320, aliquots of serum sample (2 × 100 μL) were denatured with U9 buffer (150 μL each) and bound to Q Ceramic HyperD F anion exchange resin (Pall, New York, NY). Samples were fractionated using pH based elutions as described earlier. Efficiency of protein enrichment in each fraction was monitored by SELDI-TOF-MS analysis using ProteinChip chemistries and binding protocols already described. SELDI analysis was used to identify the fraction containing the highest intensity signal at m/z 9320 for subsequent fractionation using reversed phased C18 resin (RPC PolyBio C18 resin; BioSepra, Cergy, France). Fractions were acidified with TFA (0.1% v/v TFA final concentration) and allowed to batch bind C18 resin for 30 minutes at 4°C in spin columns with end over end rotation. Unbound sample flow through was collected and the sample was fractionated in a stepwise manner using 10% increases in acetonitrile concentration (0% to 100% ACN) with 0.1% v/v TFA. The serum fraction containing sufficiently enriched m/z 9320 peak was evaporated to dryness followed by rehydration in 4× LDS reducing sample buffer (20 μL) and analyzed by Bis-Tris (12% or 4% to 10% polyacrylamide) sodium dodecyl sulfate (SDS)-PAGE with MES SDS running buffer (Invitrogen, Carlsbad, CA). Protein bands were visualized using a colloidal coomassie G-250 stain (Colloidal Blue Staining Kit, Invitrogen). Selected SDS-PAGE bands were excised and trypsin digested. Trypsin digests (1 μL) were spotted onto normal phase SELDI arrays (NP20, Bio-Rad laboratories), dried, washed with water (3 × 10 μL), air dried, and 1 μL of a saturated α-cyano-4-hydroxycinnamic acid solution (50% v/v ACN, 0.5% TFA) was added as matrix. Samples were analyzed using a quadrupole time-of-flight MS (QStar XL, Applied Biosystems/MDS Sciex, Foster City, CA) equipped with a SELDI ionization source (PCI-1000, Ciphergen) running Analyst QS 1.1. Survey scans (m/z 1,000-2,500) were acquired for the purpose of selecting ions for MS/MS analysis. The most intense ions observed by TOF-MS were selected for MS/MS analysis. Product ion MS/MS spectra were acquired by accumulating 100 to 300 scans for each selected peptide using collision induced dissociation. All product ion MS/MS spectra were acquired using a mass range from m/z 100 to an upper range that included the precursor mass selected for MS/MS fragmentation. All spectra were acquired in positive ion mode, and the mass spectrometer was externally mass calibrated using MS/MS fragment ions of human [Glu1]-fibrinopeptide B (Sigma-Aldrich, St Louis, MO).

MS/MS Data Analysis

QStar IDA files were viewed using Analyst QS 1.1 software. A built in Mascot script (1.6b21 ABI – Matrix Science Limited) was used to create the peak lists from all files. QStar data charge states were calculated from the TOF-MS scan and ions with a charge state of +1 used. Spectra were discarded if they contained less than 10 peaks. MS/MS data were centroided but not de-isotoped. These peak lists were then sent to a local Mascot search engine V 2.2 (Matrix Science Limited). Trypsin was selected as the digest enzyme and up to 1 missed cleavage was allowed. Searches of trypsin digests with no reduction and alkylation were performed using propionamide modification of Cys, oxidation of Met, and deamidation of Asn/Gln were used as variable modifications. Searches of all other trypsin digests were performed using carbamidomethyl modification of Cys as a fixed modification with oxidation of Met, and deamidation of Asn/Gln used as variable modifications. Additional parameters used for the search of QStar data include a peptide tolerance of ±0.5 Da and MS/MS tolerance of ±0.3 Da of the monoisoptopic mass and MALDI-QUAD-TOF selected for the instrument. The sequence database searched was the International Human Protein Index (IPI_human, release 3.36, EMBL-EBI) which provides a minimally redundant yet maximally complete set of proteins for humans (one sequence per transcript) containing 69,012 entries.

Identification of the m/z 9,320 Protein

The serum sample chosen for enrichment and identification of the m/z 9,320 peak was from the patient with the highest peak intensity in the venous sample and corresponded to Figure A1B. Anion exchange fraction 1 was prepared from 200 μL of serum and subjected to hydrophobic fractionation using reversed phase resin. The maximum signal intensity of the m/z 9,320 peak was observed by SELDI-TOF-MS analysis, using an NP20 ProteinChip array, in the 40% ACN, 0.1% TFA fraction. Following vacuum concentration to dryness, the reconstituted sample was analyzed by SDS-PAGE (Appendix, Fig A2A, online only) and a colloidal coomassie-stained band was observed with a relative mobility (Mr) of approximately 9 kDa. The intact mass of this band was confirmed to be the m/z 9,320 peak by passively eluting the protein from a portion of the band and confirming its mass by SELDI-TOF-MS analysis on a PBSIIc using an NP20 ProteinChip array. The remainder of the protein band, plus a blank region at the edge of the gel were excised and subjected to in gel trypsin digestion and tandem MS/MS analysis using a QStar XL equipped with SELDI ionization source. A mascot MS/MS query of all four MS/MS spectra from the top four most intense ions (1724.8, 1583.8, 1198.6, and 1070.5 [M+H]+ ) was performed against the IPI Human database with a significance threshold ofP < .01. All 4 MS/MS spectra were assigned to four peptides contained in the C-terminal portion of the mature chain of pro-platelet basic protein (Uniprot P02775). Taking into consideration the observed m/z 9,320 which corresponded to CTAP III, the 85 amino acids truncation of pro-platelet basic protein, there was 46% sequence coverage (Appendix, Fig A2B, online only). Further confirmation of protein identification was provided by manual validation of all MS/MS peptide assignments.

Haptoglobin/HPT/P00738 (m/z 20,996) was also identified by LC-MS/MS in a trypsin digest of a 20 kDa Mr band from fraction 1 of a patient's venous sample who had the highest intensity by similar methods (data not shown).

No other proteins besides CATP III and haptoglobin were identified. Although other peaks (6 more in fraction 1, 3 peaks in fraction 4, and 4 peaks in fraction 6) were significantly (P < .02) altered between systemic and venous blood in all of the fractions examined, we pursued identification of only the two presented here. Many of the peaks (8 peaks) were at an m/z that were very small—less than 3,000 to 4,000 Da which are difficult to isolate by standard gel approaches. Proteins that have very small differences between pulmonary venous and arterial blood are also unlikely to pan out when measured in peripheral venous blood since the proteins will be diluted in several liters of blood. Thus, we first wanted to ensure the validity of our approach as shown here with these two extremely promising proteins; CTAPIII (m/z 9320): mean systemic intensity = 2.617 ± 0.3315; mean venous intensity = 6.607 ± 1.165, n = 16,P = .0023; ROC 0.797. Mean fold increase (comparing individual patients): 2.91 ± 1.85.

Measurement of Proteins Using ELISA

The anti-NAP-2 antibody was obtained from DuoSet (R&D Systems, Minneapolis, MN). It was produced in goats immunized with purified, E coli-derived, recombinant human neutrophil activating peptide 2 (Pillai MM, Iwata M, Awaya N, et al: Blood 107:3520-3526, 2006). NAP-2 specific immunoglobulin G was purified by human NAP-2 affinity chromatography. It is anticipated that this antibody will also detect the precursor proteins such as CTAP III, beta-thromboglobulin and platelet basic protein that shared the same the c-terminal 70 amino acid region (Pillai MM, Iwata M, Awaya N, et al: Blood 107:3520-3526, 2006; Smith C, Damås JK, Otterdal K, et al: J Am Coll Cardiol 48:1591-1599, 2006; Maheshwari A, Christensen RD, Calhoun DA: Cytokine 24:91-102, 2003). To make a more specific antibody for CTAP III, which only has four amino acids more than b-thromboglobulin, would also detect PBP thereby eliminating the possibility of a specific antibody for CTAP III. However, there was no cross-reactivity or interference with the assay by ENA-70 (epithelial-derived neutrophil-activating peptide), ENA-74 (epithelial-derived neutrophil-activating peptide-74), ENA-78 (epithelial-derived neutrophil-activating peptide), GCP-2 (granulocyte chemotactic protein-2), GRO-α (growth regulated oncogene-α), GRO-β (growth regulated oncogene-β), GRO-γ (growth regulated oncogene-γ), IL-8 (interleukin-8), IP-10 (interferon-γ inducible protein 10), MIG (monokine induced by interferon-γ), SDF-α (stromal cell-derived factor-α), and SDF-β (stromal cell-derived factor 1). Consistent with the MS data for detection of a species that was 9,320 Da with coverage for CTAP III, the fold-difference for the ELISA data was similar to the fold difference in peak intensity from MS (eg, approximately three-fold difference). The lower limits of detection were 0.015 ng/mL for CTAP III/NAP-2. The mean CV of the NAP-2 measurements was 3.9%. To validate whether the sample values reported were accurate, we performed several spike/recovery experiments on the plasma samples. The mean recovery rate was 107%, which is considered to be in the good to excellent range. The lower limit of detection for the other proteins were 0.010 ng/mL for CRP (R & D Systems); 3.13 ng/mL for haptoglobin (Immunology Consultants Laboratory, Newberg, OR); 4 ng/mL for SAA (BioSource International, Camarillo, CA); and 7.8 ug/mL for alpha-1 antitrypsin (Immunology Consultants Laboratories, Newberg, OR).

Fig A1.

An external file that holds a picture, illustration, etc.
Object name is zlj9990986310004.jpg

Surface-enhanced laser desorption ionization time-of-flight mass spectroscopy (SELDI-TOF MS) analysis. (A) Overlay trace view of SELDI-TOF MS spectra obtained using venous (top) and systemic (bottom) serum samples from 16 lung cancer patients with a CM10 ProteinChip Array between m/z 8,000 and 14,000. The gray box highlights the enhanced intensity at m/z 9,320 that predominates in the venous group of samples compared to the systemic samples. SELDI spectra were normalized using total ion current normalization (TIC) in this and all subsequent SELDI-TOF MS spectra presented. Instrument settings were optimized to prevent saturation during spectra collection. The matrix noise region (m/z < 1,000 with SPA) saturates and was omitted from analyses (including TIC normalization). (B) Comparison of the intensity of the m/z 9320 region between systemic and venous matched samples from one patient to highlight the differences in the intensity of this peak. The arrow points to m/z 9320. (C) A box and whisker plot of peak intensities for the m/z 9320 peak. Biologic replicates with the average intensity of the m/z9320 peak in systemic or venous serum samples from patients with lung cancer that was found to be significantly increased in venous samples. Mean systemic intensity = 2.617 ± 0.3315, mean venous intensity = 6.607 ± 1.165, n = 16, (P = .002267; ROC, 0.7968). Mean fold increase (comparing individual patients): 2.91 ± 1.85.

Fig A2.

An external file that holds a picture, illustration, etc.
Object name is zlj9990986310005.jpg

Isolation and identification of the protein at m/z 9,320 using a preparative 1D gel followed by MS/MS. (A) SDS-PAGE analysis of the 40% ACN reversed phase fraction from the flow through/pH 9 anion exchange (Fraction 1) of the patient serum sample with the highest intensity for the m/z9,320 peak. The proteins in the gel were stained using colloidal coomassie. The serum sample corresponds to Fig A1B venous blood. The arrow head points to the 9 kDa Mr band on the right hand side of the gel that was excised, trypsin digested and analyzed by MS/MS. (B) A collisionally induced dissociation tandem MS spectrum of a singly charged m/z 1724.8 ion observed in the trypsin digest of the m/z9320 protein. Identity of the m/z 1724.8 ion was assigned to the peptide sequence GKEESLDSDLYAELR from PBPP with a Mascot ion score of 102 and expected value of 1.8 × 10−8. Assigned b- and y-fragment ions have been labeled and graphically represented on the overlaid peptide sequence. A total of 4 ions observed in the trypsin digest of the 9 kDa Mr band were subjected to MS/MS analysis and were assigned by Mascot to CTAP III (observed peptide sequences underlined below) with a total protein score of 263. CTAP III is a truncated form of pro-platelet basic protein with a theoretical mw 9291.74). CTAP III sequence: NLAK GKEESL DSDLYAELRC MCIKTTSGIH PKNIQSLEVI GK GTHCNQVE VIATLK DGR KICLDPDAPRI KKIVQKKLAG DESAD.

Fig A3.

An external file that holds a picture, illustration, etc.
Object name is zlj9990986310006.jpg

Connective tissue-activating peptide III (CTAP III)/ neutrophil activating protein-2 (NAP-2) and haptoglobin concentrations by enzyme-linked immunosorbent assay. (A) Differences between pulmonary venous and systemic blood in 64 patients undergoing thoracotomy for small peripheral lung tumors. A significantly higher level of CTAP III/NAP-2 and haptoglobin was observed in the pulmonary venous blood draining from the tumor (P < .001 and P = .008 respectively). (B) Changes in CTAP III/NAP-2 before and after surgical removal of the tumor in 28 patients. A significantly lower level of CTAP III/NAP-2 was observed after tumor removal only in those who did not have tumor recurrence after surgery.

Fig A4.

An external file that holds a picture, illustration, etc.
Object name is zlj9990986310007.jpg

Biomarker levels in peripheral venous plasma samples of subjects who did and did not develop lung cancer in a lung cancer prevention study. (A) Only levels of and connective tissue-activating peptide III (CTAP III)/ neutrophil activating protein-2 (NAP-2) and haptoglobin were significantly higher in plasma of subjects who developed lung cancer (P = .004 and P < .001 respectively). The levels of acute phase reactants such as C-reactive protein, alpha-1-antitrypsin, and serum amyloid A levels were not significantly different between the two groups. (B) CTAP III/NAP-2 levels between current and former smokers showing no significant difference between the two groups (P = .103).

Fig A5.

An external file that holds a picture, illustration, etc.
Object name is zlj9990986310008.jpg

The receiver operating characteristics (ROC) curves of clinical factors and biomarkers. FEV1%P, forced expiratory volume in one second as percent of predicted; AUC, area under the curve; NAP-2, neutrophil activating protein-2; CRP, C-reactive protein.

Table A1.

Demographic and Clinical Data in Surgical Patients

ParameterStudy
SELDI-MS/MS
Venous-Arterial ELISA
Pre- and Postsurgery ELISA
No.%No.%No.%
Total No.166428
Age, years696961
    SD699
    Range58-7644-8548-75
Men:women10:830:3413:15
Smoking status
    Current91113
    Former65315
    Never smoker300
Smoking, pack-years474752
    SD172122
FEV1 % of predicted717179
    SD151620
Carcinoma
    Squamous cell11620311139
    Adenocarcinoma126736561553
    Small cell3161214
    Other21171114
Stage
    000621
    IA1620311657
    IB84424380
    II738142227
    III1657311
    IV161214

Abbreviations: ELISA, enzyme-lined immunosorbent assay; FEV1, forced expiratory volume in 1 second; SELDI, surface-enhanced laser desorption ionization.

Table A2.

Relationship Between Biomarkers, Clinical Characteristics, and the Risk of Lung Cancer in the British Columbia Cancer Agency Lung Cancer Prevention Study

Variableβ Coefficient*SEP
Haptoglobin, mg/mL1.070.50.031
CTAP III/NAP-2, ng/mL2.951.15.010
Age, years0.050.03.108
FEV1 % predicted0.200.09.026
Current smokers v ex-smokers1.080.25< .001
CTAP III/NAP-2 × FEV1 % predicted, interaction term−0.030.01.009

NOTE. A logistic regression model was used to estimate the odds of lung cancer, adjusted for all of the variables listed in the table.

Abbreviations: CTAP III, connective tissue-activating peptide III; NAP-2, neutrophil activating protein-2; FEV1, forced expiratory volume in 1 second.

*For every 1 unit increase.

Footnotes

Supported in part by the Vancouver Coastal Health Research Institute “In-it-for-Life” grant (J.Y., S.L.), Canadian Institutes of Health Research grant (D.D.S., S.F.P.), Grants No. 1PO1-CA96964 and U01CA96109 from the National Institutes of Health, and National Cancer Institute contract N01-CN-85188 (S.L.), grant No. CA105304 (M.D.S.), and the British Columbia Cancer Agency MDS-Rix Endowment fund (S.L.); the Lung Health Study was sponsored by a N01-HR-46002 contract from the Division of Lung Diseases of the National Heart, Lung, and Blood Institute.

Presented in part the International Lung Cancer Conference, Liverpool, United Kingdom, July 9-12, 2008.

Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a “U” are those for which no compensation was received; those relationships marked with a “C” were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment or Leadership Position: None Consultant or Advisory Role: None Stock Ownership: None Honoraria: None Research Funding: John Yee, Vancouver Coastal Health Research Institute; Don D. Sin, Canadian Institute of Health Research; S.F. Paul Man, Canadian Institute of Health Research; Stephen Lam, BC Cancer Agency MDS-Rix Endowment fund Expert Testimony: None Other Remuneration: None

AUTHOR CONTRIBUTIONS

Conception and design: John Yee, Marianne D. Sadar, Stephen Lam

Financial support: John Yee, Marianne D. Sadar, Stephen Lam

Administrative support: John Yee, Stephen Lam

Provision of study materials or patients: John Yee, Jennifer Kondra, Annette McWilliams, S.F. Paul Man, Stephen Lam

Collection and assembly of data: John Yee, Marianne D. Sadar, Don D. Sin, Michael Kuzyk, Jennifer Kondra, Stephen Lam

Data analysis and interpretation: John Yee, Marianne D. Sadar, Don D. Sin, Michael Kuzyk, Li Xing, Stephen Lam

Manuscript writing: John Yee, Marianne D. Sadar, Don D. Sin, Michael Kuzyk, Stephen Lam

Final approval of manuscript: John Yee, Marianne D. Sadar, Don D. Sin, Michael Kuzyk, Li Xing, Jennifer Kondra, Annette McWilliams, S.F. Paul Man, Stephen Lam

REFERENCES

1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2007. CA Cancer J Clin. 2007;57:43–66. [PubMed]
2. Parkin DM, Bray F, Ferlay J, et al. Global cancer statistics, 2002. CA Cancer J Clin. 2005;55:74–108. [PubMed]
3. Cortese DA, Pairolero P, Bergsralh E, et al. Roentgenographically occult lung cancer: A ten-year experience. J Thorac Cardiovasc Surg. 1983;86:373–380. [PubMed]
4. Rami-Porta R, Ball D, Crowley J, et al. 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. [PubMed]
5. Saito Y, Nagamoto N, Ota S, et al. Results of surgical treatment for roentgenographically occult bronchogenic squamous cell carcinoma. J Thorac Cardiovasc Surg. 1992;104:401–407. [PubMed]
6. Henschke CI, Yankelvitz D, Libby DM, et al. Survival of patients with stage I lung cancer detected on CT screening. N Engl J Med. 2006;355:1763–1771. [PubMed]
7. McWilliams AM, Mayo JR, Ahn MI, et al. Lung cancer screening using multi-slice thin-section computed tomography and autofluorescence bronchoscopy. J Thorac Oncol. 2006;1:61–68. [PubMed]
8. Bach PB, Kattan MW, Thornquist MD, et al. Variations in lung cancer risk among smokers. J Natl Cancer Inst. 2003;95:470–478. [PubMed]
9. McWilliams AM, MacAulay C, Gazdar AF, et al. Innovative molecular and imaging approaches for the detection of lung cancer and its precursor lesions. Oncogene. 2002;21:6949–6959. [PubMed]
10. McWilliams AM, Mayo J, MacDonald S, et al. Lung cancer screening: A different paradigm. Am J Respir Crit Care Med. 2003;168:1167–1173. [PubMed]
11. Gao WM, Kuick R, Orchekowski RP, et al. Distinctive serum protein profiles involving abundant proteins in lung cancer patients based upon antibody microarray analysis. BMC Cancer. 2005;5:110. [PMC free article] [PubMed]
12. Khan N, Cromer CJ, Campa M, et al. Clinical utility of serum amyloid A and macrophage migration inhibitory factor as serum biomarkers for the detection of nonsmall cell lung carcinoma. Cancer. 2004;101:379–384. [PubMed]
13. Patz EF, Jr, Campa MJ, Gottlin EB, et al. Panel of serum biomarkers for the diagnosis of lung cancer. J Clin Oncol. 2007;25:5578–5583. [PubMed]
14. Jacobs JM, Adkins JN, Qian WJ, et al. Utilizing human blood plasma for proteomic biomarker discovery. J Proteome Res. 2005;4:1073–1085. [PubMed]
15. Rifai N, Gillette MA, Carr SA. Protein biomarker discovery and validation: The long and uncertain path to clinical utility. Nat Biotechnol. 2006;24:971–983. [PubMed]
16. Anderson NL, Anderson NG. The human plasma proteome: History, character, and diagnostic prospects. Mol Cell Proteomics. 2002;1:845–867. [PubMed]
17. Anthonisen NR, Skeans MA, Wise RA, et al. The effects of a smoking cessation intervention on 14.5-year mortality: A randomized clinical trial. Ann Intern Med. 2005;142:233–239. [PubMed]
18. Man SF, Connett JE, Anthonisen NR, et al. C-reactive protein and mortality in mild to moderate chronic obstructive pulmonary disease. Thorax. 2006;61:849–853. [PMC free article] [PubMed]
19. Rothman KJ, Greenland S. Modern Epidemiology. Philadelphia, PA: Lippincott-Raven Publishers; 1998.
20. Ahmed N, Barker G, Oliver KT, et al. Proteomic-based identification of haptoglobin-1 precursor as a novel circulating biomarker of ovarian cancer. Br J Cancer. 2004;91:129–140. [PMC free article] [PubMed]
21. Benson MD, Eyanson S, Fineberg NS. Serum amyloid A in carcinoma of the lung. Cancer. 1986;57:1783–1787. [PubMed]
22. Liu DH, Wang XM, Zhang LJ, et al. Serum amyloid A protein: A potential biomarker correlated with clinical stage of lung cancer. Biomed Environ Sci. 2007;20:33–40. [PubMed]
23. Yildiz PB, Shyr Y, Rahman JSM, et al. Diagnostic accuracy of MALDI mass spectrometric analysis of unfractionated serum in lung cancer. J Thorac Oncol. 2007;2:893–901. [PubMed]
24. Yang SY, Xiao XY, Zhang WG, et al. Application of serum SELDI proteomic patterns in diagnosis of lung cancer. BMC Cancer. 2005;5:83. [PMC free article] [PubMed]
25. Prentice RL, Kalbfleisch JD, Peterson AV, Jr, et al. The analysis of failure times in the presence of competing risks. Biometrics. 1978;34:541–554. [PubMed]
26. Wasswa-Kintu S, Gan WQ, Man SF, et al. Relationship between reduced forced expiratory volume in one second and the risk of lung cancer: A systematic review and meta-analysis. Thorax. 2005;60:570–575. [PMC free article] [PubMed]
27. Strieter RM, Burdick MD, Mestas J, et al. Cancer CXC chemokine networks and tumour angiogenesis. Eur J Cancer. 2006;42:768–778. [PubMed]
28. Zhong L, Roybal J, Chaerkady R, et al. Identification of secreted proteins that mediate cell-cell interactions in an in vitro model of the lung cancer microenvironment. Cancer Res. 2008;68:7237–7245. [PMC free article] [PubMed]
29. Arenberg DA, Keane MP, DiGiovine B, et al. Epithelial-neutrophil activating peptide (ENA-78) is an important angiogenic factor in non-small cell lung cancer. J Clin Invest. 1998;102:465–472. [PMC free article] [PubMed]
30. Chen JJ, Yao PL, Yuan A, et al. Up-regulation of tumor interleukin-8 expression by infiltrating macrophages: Its correlation with tumor angiogenesis and patient survival in non-small cell lung cancer. Clin Cancer Res. 2003;9:729–737. [PubMed]
31. Gewirtz AM, Zhang J, Ratajczak J, et al. Chemokine regulation of human megakaryocytopoiesis. Blood. 1995;86:2559–2567. [PubMed]
32. Brandt E, Ludwig A, Petersen F, et al. Platelet-derived CXC chemokines: Old players in new games. Immunol Rev. 2000;177:204–216. [PubMed]
33. Iida N, Haisa M, Igarashi A, et al. Leukocyte-derived growth factor links the PDGF and CXC chemokine families of peptides. FASEB J. 1996;10:1336–1345. [PubMed]
34. Pillai MM, Iwata M, Awaya N, et al. Monocyte-derived CXCL7 peptides in the marrow microenvironment. Blood. 2006;107:3520–3526. [PubMed]
35. Hoogewerf AJ, Leone JW, Reardon IM, et al. CXC chemokines connective tissue activating peptide-III and neutrophil activating peptide-2 are heparin/heparan sulfate-degrading enzyme. J Biol Chem. 1995:3268–3277. [PubMed]
36. Tang Z, Yu M, Miller F, et al. Increased invasion through basement membrane by CXCL7 - transfected breast cells. Am J Surg. 2008;196:690–696. [PubMed]
37. Kannel WB, McGee D, Gordon TA. General cardiovascular risk profile: The Framingham Study. Am J Cardiol. 1976;38:46–51. [PubMed]

Articles from Journal of Clinical Oncology are provided here courtesy of American Society of Clinical Oncology