A combination of platelet features allows detection of early-stage cancer

TL;DR: Multiple platelet features, including platelet count, volume and protein content, were significantly changed in lung and head of pancreas cancer patients, and a cancer type-specific combination of these platelets features can be used to discriminate between patients with early-stage cancer and healthy individuals.

Abstract: Background Detection of early-stage cancer significantly improves patient survival. As platelets play an important role in cancer progression, we aimed to investigate whether platelets can be used for the discovery of early-stage cancer. Methods Patients with lung (n = 86) or head of pancreas (n = 42) cancer were included, as were healthy sex- and age-matched controls (n = 92). Blood was collected before initiation of treatment. Platelet count, volume and activation status were quantified in whole blood. Next, concentrations of vascular endothelial growth factor, platelet-derived growth factor, platelet factor 4, thrombospondin-1 and connective tissue–activating peptide III were measured in both platelets and plasma. Using the results, two multivariable diagnostic models were developed and internally validated. Findings Multiple platelet features, including platelet count, volume and protein content, were significantly changed in lung and head of pancreas cancer patients. However, the pattern of changes differed between both groups. The diagnostic model developed for lung cancer discriminated very well between patients and controls (AUC = 88.7%). Addition of smoking as a variable significantly increased the AUC of the model to 94.5%. The diagnostic model for head of pancreas cancer also performed well (AUC = 82.7%). Both models were internally validated, resulting in optimism-corrected AUC's of 86.8% and 80.8%, respectively. Interpretation In patients with lung or head of pancreas cancer, several platelet characteristics are changed compared to healthy sex- and age-matched controls. A cancer type-specific combination of these platelet features can be used to discriminate between patients with early-stage cancer and healthy individuals.

Summary

1. Introduction

• Detection of cancer in its early stages radically improves the effectiveness of available treatment and overall prognosis of patients [1].
• Up to now, studies searching for biomarkers are mostly based on blood plasma or serum parameters.
• In addition, platelets may become activated systemically or within the tumour, potentially resulting in release of platelet content into the circulation [3].
• Altogether, data from literature suggest that several platelet characteristics are affected in cancer patients.
• These features, either alone or in combination, may be useful tools in the detection of (early stages of) cancer.

2.1. Study design and participants

• This study was performed in accordance with the Declaration of Helsinki and approved by the medical ethical committee of Maastricht University Medical Centerþ.
• Informed consent was obtained from all participants.
• Patients with clinically established and histologically proven untreated primary lung (n Z 86) or head of pancreas cancer (n Z 42, including pancreas head cancer [n Z 28], distal cholangiocarcinoma [n Z 8] and duodenumcarcinoma [n Z 6]), that were eligible for surgical resection, were included between July 2012 and October 2014.
• Exclusion criteria were previous history of cancer, neo-adjuvant chemotherapy or radiotherapy, use of platelet-influencing drugs such as aspirin, blood or platelet transfusion during the previous 14 days, active inflammatory disease, non-healing ulcers or fractures.
• Staging was performed in accordance with the tumour-node-metastasis (TNM) classification (version 7) of the Union for International Cancer Control [9].

2.2. Procedures

• Blood from all patients and healthy individuals was collected.
• In case of cancer patients, sampling occurred within 1 week before initiation of treatment.
• To prevent platelet activation during blood collection and sample preparation, blood was collected as described before [10].
• The procedures for quantification of platelet count, volume and activation status, as well as concentrations of various growth factors and chemokines in platelets and plasma.
• ELISA measurements in all patient and matching control samples were performed using the same assay, simultaneously and in the same institution.

2.3. Statistics

• Statistical analyses were performed using SPSS (version 22; SPSS Inc, Chicago, USA) and R (version 3.2.2, R core development team).
• Data are presented as means with standard errors of the mean unless otherwise indicated.
• Patient data were compared to data from the healthy control groups using the t-test for continuous variables and Pearson’s chi-square test for categorical variables.
• Spearman’s rank correlation was used to test the association between variables.
• P-values less than 0.05 were considered statistically significant.

2.4. Development of diagnostic models

• For both cancer groups, logistic regression was used to estimate diagnostic models.
• Model performance was assessed by Nagelkerke’s R2 statistic and by quantifying discrimination.
• Discrimination is the model’s ability to distinguish between those that have cancer and those who do not.
• 1000 bootstrap samples of the original data were drawn, and similar diagnostic models were estimated using these data.
• The average difference in diagnostic model performance between the bootstrap sample and the original sample provided estimates of optimism in the performance measures.

3. Results

• Blood was collected from 86 lung cancer patients, 42 patients diagnosed with head of pancreas cancer and a total of 92 healthy individuals.
• Importantly, the latter group contained individuals that were matched with respect to gender and age to both groups of cancer patients.
• All cancer patients were treatment naive at the moment of blood sampling.
• Detailed characteristics of the study populations are presented in Table 1.
• In the healthy individuals, both gender and age appeared to be related to some of the platelet characteristics.

3.1. Platelet characteristics in patients with untreated lung cancer

• Platelet counts were not different as compared to controls in patients with early-stage (stage IeII) lung cancer, but significantly increased in case of advanced (stage IIIeIV) lung cancer (Fig. 1A).
• Concentrations of VEGF, PDGF, CTAPIII, PF4 and TSP-1 sented as means with standard errors of the means.
• Strikingly, changes in platelet content did not correspond with concentration changes of the same proteins in PFP (Fig. 1HeL).
• No differences between patients and controls were found (Supplemental Fig. 1A and B).
• This suggests that the increase in PDGF, CTAPIII and PF4 plasma concentrations in patients with lung cancer is mainly due to systemic or intratumoural platelet activation, leading to secretion of their content.

3.2. A combination of platelet features discriminates lung cancer patients from controls

• Data from 172 individuals (86 patients with lung cancer and 86 sex- and age-matched controls) were available for multivariable modelling.
• All measured parameters were entered into the model: platelet count, MPV and concentrations of VEGF, PDGF, PF4, CTAPIII and TSP-1 in platelets and PFP.
• Nagelkerke’s R2 of the diagnostic model was 0.572, indicating that the model fits the data well.
• This demonstrates that the model discriminated very well between lung cancer cases and controls in their study population.

3.3. A platelet-based marker for the presence of head of pancreas cancer

• In patients diagnosed with head of pancreas cancer, several platelet parameters were found to be substantially different from the control group as well.
• VEGF concentration was significantly increased in platelets of patients with early- and late-stage head of pancreas cancer (Fig. 3C), while no differences in platelet PDGF, CTAPIII, PF4 and TSP-1 concentrations were observed (Supplemental Fig. 2AeD).
• Nagelkerke’s R2 of this model was 0.418, which is indicative of good model fit.
• This parsimonious model discriminated well between patients with head of pancreas cancer and controls, as confirmed by the boxplots of predicted probability by the diagnostic model (Fig. 4B).

4. Discussion

• The presence of a tumour in the human body appears to influence several platelet features.
• The elevated PF4 and CTAPIII concentrations in plasma reveal the presence of activated platelets in the circulation of patients with lung or head of pancreas cancer.
• Altogether, their study shows that multiple platelet characteristics are changed in patients with cancer, both in early and later stages of development.
• One important issue in this interesting study is the nature of the control group, that is clearly younger and with a different gender distribution than most of the cancer groups.
• Funding The Netherlands Organisation for Scientific Research (project number 017.008.143, to SS).

Figures

Fig. 2. A combination of platelet features allows discrimination betwee of a final set of six parameters (platelet count, MPV, PF4 and CTAP trations in PFP). (A) Receiver operating characteristic (ROC) curves o and after addition of smoking as a variable (hatched line). (B) Boxplot o platelet parameters of lung cancer patients and controls. PF4, platelet f thrombospondin-1; MPV, mean platelet volume; PFP, platelet-free pl

Fig. 3. Head of pancreas cancer (HoPC) is associated with changes in platelet count, mean platelet volume (MPV), increased platelet VEGF content and platelet activation. Platelet characteristics were measured in healthy controls (C, n Z 42) and in patients with HoPC (subdivided according to TNM-stage: IeII, n Z 29 and IIIeIV, n Z 13). (AeB) Whole blood platelet count and MPV were measured after blood collection. Platelets and platelet-free plasma (PFP) were isolated from whole blood and concentrations of VEGF, PDGF, CTAPIII, PF4 and TSP-1 were determined. Only significantly changed parameters are shown: VEGF concentration in platelets (C), and PDGF, PF4, CTAPIII and TSP-1 concentrations in PFP (DeG). Data are presented as means with standard errors of the means. *p < 0.05; **p < 0.01; ***p < 0.001. TNM, tumour-node-metastasis; PDGF, platelet-derived growth factor; PF4, platelet factor 4; CTAPIII, connective tissueeactivating peptide III; TSP-1, thrombospondin-1; VEGF, vascular endothelial growth factor.

Fig. 4. A combination of platelet parameters allows discrimination between patients with head of pancreas cancer and controls. The prediction model consists of a final set of three parameters (platelet count, MPV and VEGF concentration in platelets). (A) Receiver operating characteristic (ROC) curve of the diagnostic model. (B) Boxplot of the predicted probability using the diagnostic model based on a combination of platelet parameters of patients with head of pancreas cancer and controls. VEGF, vascular endothelial growth factor; MPV, mean platelet volume.

Table 1 Patient demographics and clinical profiles.

Fig. 1. Lung cancer is associated with changes in platelet count, mea Platelet characteristics were measured in healthy controls (C, n Z 86) a TNM-stage (IeII, n Z 39; IIIeIV, n Z 47). (AeB) Whole blood plat Platelets and platelet-free plasma (PFP) were isolated from whole blo were determined in platelets (CeG) and PFP (HeL). Data are pre **p < 0.01; ***p < 0.001. TNM, tumour-node-metastasis; PDGF, connective tissueeactivating peptide III; TSP-1, thrombospondin-1; V

Frequently asked questions

Q1. What are the contributions in "A combination of platelet features allows detection of early-stage cancer" ?

• A submitted manuscript is the version of the article upon submission and before peer-review. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher 's website. The final author version and the galley proof are versions of the publication after peer review. The final published version features the final layout of the paper including the volume, issue and page numbers. 

Q2. What future works have the authors mentioned in the paper "A combination of platelet features allows detection of early-stage cancer" ?

Future research is needed to further investigate the clinical relevance of their findings. Platelets are a new and uncharted source of information, which need to be further explored in blood-based biomarker research.