Circulating Platelets A Novel Liquid Biopsy Source for Cancer Diagnostics and Therapy Stratification

Lee-Ann Tjon-Kon-Fat

Department of Radiation Sciences, Oncology

Umeå 2018

This work is protected by the Swedish Copyright Legislation (Act 1960:729) Dissertation for PhD ISBN: 978-91-7601-838-5 ISSN: 0346-6612 New Series number: 1952 Cover photo: Platelets interacting with MCF-7 breast cancer cells. SEM Image by Cheng Choo Lee, UCEM, Umeå Cover design: Oleg Kovrov and Mariam Shirdel Electronic version available at http://umu.diva-portal.org/ Printed by: UmU Print Service, Umeå University Umeå, Sweden 2018

To my biggest supporters, my parents

i

Table of Contents Abstract iii List of Papers iv Abbreviations v Introduction 1

Liquid Biopsies 1 Tissue versus Liquid Biopsy 2 cfDNA or ctDNA 3 Circulating Tumor Cells (CTCs) 3 Platelets 4 Tumor Educated Platelets 4 Platelet Biogenesis and Function 5 Uptake, Storage and Transfer 6 Platelets and Cancer 6 Prostate Cancer 9 Diagnosis 9 Treatment of PCa 10 The Clinical Problem 11

Aims 13 Overall aim 13 Specific aims 13

Paper I 13 Paper II 13 Paper III 13 Paper IV 13

Materials and Methods 14 Patient Samples 14 Platelet Isolation 14 RNA Extraction and cDNA Synthesis 14 Circulating free DNA Isolation 15 Droplet Digital PCR (ddPCR) 15 RNA-in situ Hybridization (RNA-ISH) 15 RNA-Sequencing Library Preparation 16 Statistics 16 Data analysis 17

ii

Results and Discussion 18 Paper I 18 Paper II 20 Paper III 22 Paper IV 24

Conclusions 26 General Discussion 28

Platelets as a Biomarker source 28 Clinical Importance 29

Acknowledgement 32 References 36

iii

Abstract

As conventional tissue biopsies have several drawbacks, much effort has been directed toward the development of minimal-invasive liquid biopsy platforms for detecting and profiling cancer. Platelets are the second most abundant cells in blood and have very versatile functions both in physiological and pathophysiological conditions. When exposed to tumors and their environment, platelets exchange biomolecules with tumor cells changing the platelets’ RNA profile, resulting in tumor-mediated education of the platelets. Our research group and collaborators have previously shown that platelets sequester material while in circulation and with that ability accumulate cancer specific information. Platelet RNA profiles or detection of tumor-derived biomarkers within them may provide insight into ongoing cancer-related processes in a patient, allowing for implementation of personalized therapy strategies. This thesis evaluates whether circulating platelets could have a potential role (as a liquid biopsy source) in cancer diagnostics, therapy stratification, and monitoring of the disease. Gene expression analysis using digital droplet PCR and RNA-sequencing were the main methods used to address this. Prostate Cancer is the main model used in this thesis but this platform is applicable to other tumor types such as colorectal-, breast-, and lung cancer. We found platelets of cancer patients to contain tumor-derived information enabling selection of biomarker panels discriminating early stage cancer patients from healthy individuals as well as therapy responders from non-responders with high accuracy. The RNA transcript within the platelets was more informative in regards to therapy stratification compared to circulating free DNA of matched patient samples, in which genomic changes were analyzed. Combining both increased the accuracy in predicting therapy outcome. Platelets show usefulness as a novel liquid biopsy source for early detection and individualizing patient therapy decisions (for personalized medicine). The techniques used are promising but large-scale validation is necessary.

iv

List of Papers

1. Platelets Harbor Prostate Cancer Biomarkers and the Ability to Predict

Therapeutic Response to Abiraterone in Castration Resistant Patients. Prostate 2018 Jan;78(1):48-53. doi: 10.1002/pros.23443.

2. Combining Circulating Biomarkers to Enhance Predictability of Therapy

Response. (Manuscript)

3. Detection of Early Stage Prostate Cancer Using Tumor Educated Platelet

Profile Support Vector Machine (SVM) - Classification Algorithms (Manuscript)

4. Swarm Intelligence-Enhanced Detection of Non-Small-Cell Lung Cancer Using Tumor-Educated Platelets. Cancer Cell. 2017 Aug 14;32(2):238-252.e9.

doi: 10.1016/j.ccell.2017.07.004.

Not included in this thesis Platelet RNA in Cancer Diagnostics. Semin Thromb Hemost. 2017 Sep 13.

doi: 10.1055/s-0037-1606182

The original article (Paper I) was reprinted with permission from the publisher. Paper IV did not require permission to be reprinted.

v

Abbreviations

PCa PSA CTCs TEPs MKs cfDNA ctDNA EVs DNA RNA CRPC EDTA ISH RIN GS AR DHT EMT TGF-β MMP PTCI UCAN MHC-I NK KLK2 KLK3 FOLH1 PSMA NPY ADT AS NSCLC SVM PSO FNR FPR TR NR

Prostate Cancer Prostate Specific Antigen Circulating Tumor Cells Tumor Educated Platelets Megakaryocytes Circulating free DNA or cell free DNA Circulating tumor DNA Extracellular Vesicles Deoxyribonucleic Acid Ribonucleic Acid Castration Resistant Prostate Cancer Ethylenediaminetetraacetic acid In Situ Hybridization RNA Integrity number Gleason Score Androgen Receptor Dihydrotestosterone Epithelial-Mesenchymal Transition Tumor Growth Factor- beta Matrix Metalloproteinases Platelet Tumor Cell Interaction Uppsala-Umeå Comprehensive Cancer Consortium Major Histocompatibility Complex-I Natural Killer Kallikrein 2 Kallikrein 3 Folate Hydrolase 1 Prostate Specific Membrane Antigen Neuropeptide Y Androgen Deprivation Therapy Androgen Synthesis Non-Small-Cell Lung Cancer Support Vector Machine Particle Swarm Optimization False Negative Rate False Positive Rate Therapy Responder Non-Responder

1

Introduction

Liquid Biopsies

Liquid biopsies are tests that extract tumor-derived material such as tumor cells and fragments of nucleic acids from bodily fluids in a minimal invasive fashion for detection as diagnostic and prognostic biomarkers. Biomarkers are biological measurements that identify the presence, status, and recurrence of disease. For biomarkers to aid in clinical decision making, they need to be accurate, reproducible, cost effective, and acceptable to patients (preferably non-invasive or minimally invasive). To date, the gold standard of invasive biopsy procedures still holds and no circulating biomarker has been able to replace that. Moreover, despite the widespread use of many available tumor biomarkers, patients are often misdiagnosed. In prostate cancer (PCa), the introduction of the PSA test has increased the incidence of PCa; however, PSA-testing is sub-optimal and its benefit-to-harms ratio is unclear (1, 2). An elevated PSA may not necessarily indicate for having PCa. Noncancerous conditions, including prostatitis and benign prostatic hyperplasia (BPH), can also raise PSA levels, leading to false positive results, overdiagnosis and overtreatment (3). There is no absolute cut-off between normal and abnormal PSA values due to the natural occurrence of PSA in serum and the increase during non-cancerous prostate conditions. Therefore, when increasing the cut off it is highly possible the cancer is missed (i.e, false negatives). Clearly, there is a great interest for biomarkers that detect early tumor recurrence, monitor tumor growth, and predict therapeutic response to improve cancer treatments and survival. The treatment of cancer is shifting from an entity-driven approach to molecular-guided personalized approach (4). In personalized medicine, the unique biology of each patient is matched to the medical care provided. This tailored care system should improve health care and lower costs. Biomarkers with high sensitivity, high positive predictive value, and high specificity are needed to identify patients with low tumor load thereby limiting false negatives and false positives. In addition, these biomarkers should enable early detection and treatment and allow the monitoring of the tumor in

2

real-time. Liquid biopsies can be useful in this and a lot of effort has been put into developing new platforms for biomarker discovery.

Figure 1. Main biosources of blood liquid biopsy.

Tissue versus Liquid Biopsy

Sequencing the genetic profile of tumors will eventually be the standard for determining treatment response and therapy stratification. Conventional tissue acquisition is invasive, convey risk of complications, have significant cost and other limitations that could be overcome by liquid biopsies. In the procurement of prostate tissue, infection is one of the most common complications that can have a large impact on the patient and cost of care (5). For some cancer types, it is difficult to access adequate amount of tissue limiting tumor analysis (6) and preservation methods such as formalin-fixation compromise the integrity of DNA, leading to false results of genetic tests (7). Ultimately, sample bias of tissue biopsies is a major limitation due to tumor heterogeneity, challenging the identification of objective prognostic markers (8). Prostate cancer, for example, is known to be a heterogeneous and multifocal disease arising from multiple, independent clonal expansions (9-12). Individual lesions can also grow together, creating an additional level of heterogeneity (13). The sample volume of a prostate biopsy is rather small, so the chances of missing the prostate tumor are quite high. Between 20-40% of confirmed prostate cancer cases initially had false-negative biopsies (14). Multiple biopsies should be analyzed to compensate for heterogeneity and support clinical treatment

3

decisions (15). However, biopsies only reflect on the genetic profile of the tumor at the time of sampling. Tumors are very dynamic and change their genetic composition, especially after selective pressure of targeted treatment. Stratifying patients for treatment decisions based on archived tissue biopsies would possibly be wrong because it would be constructed on a historical disease profile rather than on the present profile (16, 17). Liquid biopsies are minimally invasive and therefore suitable for repeated analysis enabling longitudinal monitoring of disease. Moreover, the genetic information of all lesions present is covered, representing the altered genetic landscape and allowing treatment adaption accordingly. Furthermore, disease progression and therapy failure/resistance may be detected before clinical or radiographic progression is evident, permitting early detection and improving the chances of survival (18). Among the main sources of liquid biopsy for biomarker detection are circulating free DNA (cfDNA) or circulating tumor DNA (ctDNA), circulating tumor cells (CTCs), extracellular vesicles and more recently platelets (Figure 1).

cfDNA or ctDNA

The presence of circulating free DNA (cfDNA) in blood was first reported by Mandel and Metasis in 1948 (19). Through cellular decay processes such as apoptosis and necrosis by both normal and cancer cells, DNA fragments are released into the circulation. cfDNA refers to the DNA fragments released by cells irrespective from origin, whereas fragments released explicitly by cancer cells is mainly referred to as circulating tumor DNA (ctDNA). This biosource may harbor mutations, CNVs, structural variants and methylation changes, all of which may provide information on tumor load and site of tumor origin (20-23). Advanced cancer patients have an increased cfDNA level in plasma of which more than 10% can be traced back to the tumor genome (24), in early disease stage however, very little ctDNA is present. This is a limitation of the biosource that needs to be overcome by increasing the yield using (new) alternative extraction methods or by analyzing more blood.

Circulating Tumor Cells (CTCs)

Circulating tumor cells are cells released in circulation after detaching from a tumor site to possibly form secondary tumors by seeding to distant sites. CTCs have been detected in many types of cancer as reviewed by Lianidou et al. (25).

4

CTCs are an interesting source of tumor information as each CTC presents a cancer genome and transcriptome where mutations and gene fusions can be found (26-28). However, these cells are extremely rare in patients and even in patients with metastatic lesions their frequency is low, making their detection, enumeration and molecular characterization very challenging (29, 30). Many studies have used CTC enumeration to predict recurrence, PFS, OS and therapy resistance (26, 31-36). Extracellular Vesicles (EVs) EVs are membrane vesicles (ranging from ~30 to several hundred nm in size) shed by cells into the extracellular environment. EVs function as intercellular messengers by either budding the from plasma membrane or released through multi-vesicular bodies. As stable carriers of genetic material and proteins from their cell of origin, EVs can be shuttled between cells leading to many different functional changes in the recipient cells. Tumor- and stroma-derived EVs can therefore facilitate the metastatic process (37, 38). Many studies have found that EVs contain RNA with tumor-specific mutations and DNA with gene amplifications and mutations (39-43), making them interesting potential biomarkers for cancers. However, there is a substantial diversity in isolated vesicles based upon differences in isolation protocols, cell origins, and biological functions, leading to an overwhelming, confusing, and conflicting nomenclature. EVs consist of many subpopulations that can potentially be divided by various characteristics such as morphology, density, cargo, and size. A major ongoing challenge is to define methods to discriminate between these (44).

Platelets

Until very recently, platelets have been a rather untouched source of information for cancer and possibly other diseases. Our group and collaborators have found platelets to be a novel biosource for tumor-derived material sequestered during circulation (45). Therefore, platelets have the potential to accurately predict what type of cancer is present (if any), how a patient will respond to different therapies, and which oncogenic modifications drive the disease (46, 47).

Tumor Educated Platelets

Tumor cells can impact and impose changes in platelet RNA and protein content thereby ‘educating’ the platelets. The fact that platelets are cytoplasmic fragments from megakaryocytes (MKs) with neither a nucleus nor genomic DNA, it was

5

thought that transcriptional processes did not occur within these cells. However, Denis et al. reported the presence of functional spliceosomes and unspliced pre-mRNAs in the cytoplasm of platelets as well as provided strong evidence that during platelet activation splicing in the cytoplasm is a key regulatory event (48, 49). After these unexpected observations were reported, Meshorer and Misteli posited several fundamental questions regarding the organization and regulatory role of pre-mRNA splicing (50). Some theories elucidating how the splicing machinery normally found in the nucleus ends up in the cytoplasm include a specific export mechanism that actively and selectively exports the nuclear components into the cytoplasm of MK or that many splicing components undergo nucleocytoplasmic shuttling in which they cycle between the nucleus and the cytoplasm as part of their normal life cycle (51, 52). Normally, mature mRNA moves to the cytoplasm and incomplete processed pre-mRNA is kept in the nucleus where it is degraded by the cell’s RNA quality-control system (53). However, in platelets, pre-mRNA is present as mentioned previously but recognition mechanisms may possibly be in place that identify RNAs destined for platelets, allowing these pre-mRNAs to evade RNA degradation; though, how this mechanism works has not been fully understood (50). The capacity of platelets to sequester (45, 54) circulating mRNA together with the education by splicing in response to external stimuli provides TEPs with a dynamic mRNA repertoire that can be used for cancer diagnostics. Our group and collaborators have developed a method to detect cancer using TEP RNA analysis (45, 55, 56).

Platelet Biogenesis and Function

Platelets, 2-4μm in diameter, are discoid shaped and anucleated cells generated in the bone marrow and in intravascular sites in the lungs from their precursor cells, the megakaryocytes (MKs), via endomitosis. (57-60). Endomitosis is a process in which several cycles of genome replication take place without cell division. The replicated DNA is retained within the original nucleus instead of being divided into two daughter nuclei, accumulating a DNA content of 4n to 128n in one polylobulated nucleus in MKs before proceeding to the next phase (61). In the second phase of maturation, (61-63) cytoplasmic expansion takes place and contents essential for platelet function such as platelet-specific

6

organelles, granules, proteins and internal membrane systems are formed as the polyploid megakaryocytes further mature (62, 64). Precursor platelets are formed and released by matured MKs through long branching protrusions which then further mature and interconvert from pre-platelets to pro-platelets and vice versa, ultimately budding off as mature platelets while in circulation (65). Platelets are the second most abundant cells in blood and about ~1011 platelets are made daily with a lifespan of 7-10 days (66). Hemostasis and thrombosis, the process to stop and repair bleeding by promoting blood clotting is a role that platelets are mostly known for (67-69). Recently, the complexity and versatility of platelets have received increasing attention as they have been shown to participate in many physiological and pathophysiological processes such as innate and adaptive immunity (59, 66), atherosclerosis (67, 70, 71), angiogenesis (72-74), and tumor metastasis (75-78).

Uptake, Storage and Transfer

The open canalicular system is one way the platelets passively sequester material encountered in circulation (e.g. bacteria, viruses, proteins, antibodies and nucleic acids) (45, 54, 79-84), but platelets also sequester material using receptor-mediated active uptake. This continuous invagination of the outer plasma membrane allows for the storage and transfer of ingested material from plasma to organelles (79, 85). The ingested material can also be processed and displayed on MHC-I complexes as in the case of antigens for naïve T-cell presentation and activation, further implicating the importance of platelets as regulators of the immune system (86). However, platelets also help tumor cells to evade the immune system and promote extravasation by transferring bioactive molecules to tumor cells (87), complicating the functional role of platelets and revealing their ‘Jekyll-and-Hyde’ nature.

Platelets and Cancer

Various steps in cancer development and progression indicate the involvement of platelets. An increased risk of cancer incidence and poor prognosis has been associated with thrombocytosis, a disorder that is the result of the body producing too many platelets, increasing the risk of venous thromboembolism (88-90). The platelet lymphocyte ratio has also been shown to be a very valuable predictor of cancer prognosis (91). Another indication that platelets play a role in cancer initiation and progression is the fact that aspirin use decreases risk of

7

cancer incidence and metastasis (92), while tumor cell-induced platelet aggregation shows a correlation between platelet aggregation and tumor metastasis (93). Furthermore, changes in platelet morphology and internal structures have been used to identify patients with ovarian cancer in which an increased number of platelet mitochondria and shorter microtubules predicted cancer and identified platelet hyperactivity in these patients (94). Platelets play role in metastasis by facilitating dissemination and helping cancer cells grow at distal sites (Figure 2) (95, 96). As part of the tumor microenvironment, an interplay between tumor cells and circulating platelets is evident through tumor growth regulation, angiogenesis, dissemination, and metastasis (Figure 2). Metastasis is enhanced with an increase in platelet count, and hampered by a decreased count, further evidence that platelets play a significant role in this process (97, 98). Platelets facilitate invasion by activating the epithelial- mesenchymal transition (EMT) through the secretion of several growth factors such as transforming growth factor-β (TGF-β) and platelet-derived growth factor at the border of primary tumors (99-101). Through direct release of matrix metalloproteinases (MMP) by platelets, proteolytic activity of MMPs disrupts and digests the extracellular matrix supporting migration of the tumor cell towards blood vessels (102). In addition, cancer cells in the bloodstream can be the result of endothelial cell retraction, induced by platelet-released eicosanoid metabolites such as thromboxane, histamine and serotonin (103-105).

Figure 2. Circulating Platelets and their role in metastases. (1) Produced by MKs within the bone marrow and intravascular sites in the lung, platelets mature while in circulation (2-4). The metastatic process of the tumor cells is promoted by platelets.

8

Tumor cells are in close contact with platelets once in circulation as single cells or as part of a tumor-leucocyte complex, resulting in a cross-talk between them that can be mediated through adhesion receptors such as P-selectin upon tumor-induced platelet activation (106, 107). This facilitates the platelet-tumor cell (PTCI) interaction and protects the tumor cells by ‘coating’, providing a physical shield that hides the tumor cells from immunological detection and natural killer (NK) cell clearance. This shield helps cancer cells evade immune surveillance, increasing survival and enabling the dissemination of cells to distal sites (108-110). Furthermore, NK cell recognition is also prevented by the transfer of MHC-I to tumor cells mediated by PTCI (87) and by foiling the activation of receptors on NK cells by TGF-β release (111), triggering TGF-β dependent EMT-promoted extravasation from the circulation (101, 112). Labelle et al. found that EMT-like transition and metastasis is promoted through a synergy of physical interaction and release of TGF-β by platelets (112). The cell surface proteins on platelets act as anchors, adhering proteins to the vessel wall, which aids rolling and tethering of platelet-tumor cell complexes to the endothelial lining. Through this rolling and tethering, tumor cell adhesion to the endothelium is enhanced and extravasation from the circulation is promoted (109, 113-117) by the release of adenosine triphosphate, which binds P2Y2 on the endothelial cells, opening the trans-endothelial barrier (117). Chemokines, CXCL5, CXCL7, RANTES and MCP-2 are released by platelets upon PTCI, attracting granulocytes and monocytes supporting the extravasation of cancer cells and the formation of a metastatic niche, evidence that together with leucocytes, platelets promote the development of metastasis. Platelets contain many proangiogenic and mitogenic factors that are continuously available to all tissues, including tumors, due to their rapid turnover and high platelet number (118). These factors promote tumor growth at distant sites and stimulate neovascularization in the tumor microenvironment when released upon platelet activation (73, 119-122). Many studies (123-126) show that the interplay between platelets, tumor cells, and the microenvironment contributes to the development of metastasis.

9

Prostate Cancer

Prostate cancer (PCa) is the second most common cancer among men worldwide (127). The incidence of PCa varies in different parts of the world, with more than a 25-fold difference. Differences in screening, diet, environment, and genetic background might explain this variation (127, 128). The introduction of prostate-specific antigen (PSA) has definitely contributed to the reported increase in PCa incidence over the last two decades (1). Under normal conditions, PSA, also known as kallikrein III (KLK3), is produced in the prostate and secreted in the ejaculate to help liquefy the semen (129). When the architecture within the prostate is disrupted due to benign prostate hyperplasia, inflammation, or tumor formation, PSA leaks out of the prostate gland into the surrounding stroma and vasculature thereby elevating the PSA level in the blood. This allows PSA to be used as a diagnostic marker for prostate diseases and as a way to assess the risk of having PCa (130, 131).

Diagnosis

Typically, PCa is first suspected by measuring elevated PSA levels in blood. A serum PSA level from 0 to 3 ng/ml is considered normal and a PSA level above 10 ng/ml indicates a high risk for PCa. As mentioned previously, PSA levels are also elevated in other prostate conditions and so PSA levels may not be cancer specific. To establish the diagnosis, biopsies are taken, which are microscopically examined and scored according to the Gleason Score (GS) system to predict prognosis (132). This scoring system is based on the differentiation pattern of the tumor biopsy on a scale ranging from 1 to 5, where 5 represents the poorest differentiated and most aggressive tumor pattern. The most common and second most common differentiation patterns are summed GS, ranging from 6-10 according to the new grading system proposed by the 2014 International Society of Urologic Pathology (ISUP) consensus and adopted by the WHO 2016, as GS below 6 does not exist in practice (133). To date, the GS is the most informative outcome predictor of PCa. A GS 6 indicates a less aggressive status (low-grade tumors) with good prognosis and a GS >8 indicates a high aggressive status (high-grade tumors) with poor prognosis. An intermediate GS (GS 7) decreases the prognostic value and the outcome is very variable (134).

10

After the initial PCa detection, the spread of the disease is determined by the TNM (Tumor, Node, Metastasis) staging (Table 1). TNM staging indicates whether the disease is local, locally advanced, or advanced. When the tumor cells have not spread outside the prostate gland, the disease is considered local. If the tumor extends outside the fibrous capsule covering the prostate gland but no distant metastases have formed, it is classified as locally advanced. In advanced PCa, in addition to the spread outside the gland, metastasis is found at distant sites. Together, the PSA level, GS score, and TNM stage determine the risk classification (Table 2) for localized prostate cancer in order to determine treatment (135).

Table 1. The TNM staging for, reporting the extent of the primary tumor (T), the status of spreading to lymph nodes (N), and the presence of metastases (M).

Treatment of PCa

The treatment option for local PCa depends on the risk classification group the patient belongs to (Table 2). Surgical prostatectomy and radiation therapy are both treatment options for local PCa with curative intent. Active monitoring, another treatment strategy, closely monitors patients for signs of disease progression, to reduce harm and limit overtreatment but maintain benefits of early intervention. Surgery and radiation therapy are also options for locally advanced PCa but are not likely to eliminate the cancer given as a single treatment, therefor it is usually combined with androgen deprivation therapy (ADT) as most prostatic tumors are androgen dependent (136).

11

Table 2. The Risk classification groups for local PCa. (135)

Advanced and metastatic PCa cannot be cured and the treatment strategy changes from curative to palliative in the form of chemical or surgical castration therapy, also known as ADT. Androgens are important for normal prostate development and function (137, 138). Testosterone, produced by cells in the testis, is the main circulating androgen, and is converted in the prostate by 5-alpha reductase to the more potent androgen dihydrotestosterone (DHT) (139-141). Both testosterone and DHT bind to and activate the ligand-dependent androgen receptor (AR), leading to dimerization and translocation from the cytoplasm to the nucleus and functioning as a transcription factor. Ultimately, gene transcription of AR target genes such as KLK2, KLK3 (encoding PSA), STEAP2 and TMPRSS2 are regulated by androgens (142). In most patients, treatment of advanced and metastatic PCa with ADT lowers androgen levels resulting in tumor size reduction, relief of pain associated with bone metastasis and, decline in PSA levels (143). Unfortunately, most patients receiving this treatment relapse within a few years and become castration resistant i.e., the AR remains activated despite the low levels of testosterone (144-146). This AR reactivation could be due to several mechanisms such as AR amplification, abnormal activity of co-regulators of the AR-complex, AR mutation, AR activation through other signaling pathways, or intratumoral androgen synthesis (147). For castration resistant prostate cancer (CRPC) ADT treatment remains and used in combination with therapies such as chemotherapy, radiotherapy, immunotherapy, and novel androgen directed therapies in an attempt to slow down disease progression (148-153)

The Clinical Problem

The diverse outcomes due to the heterogeneity of PCa can range from long-term symptom free survival to aggressive metastatic disease. The optimal management strategy for PCa remains controversial creating a clinical problem. Overdiagnosis leads to overtreatment exposing men unnecessarily to complications and harmful side effects such as urinary incontinence, erectile dysfunction, impaired bowel function, and infection (154). After 25 years of routine PSA-based testing, the medical community is still divided with regard to its effectiveness and its

12

benefits-to-harm ratio. Two large population based screening trials (ERSCP and PLCO) suggested the net benefits of PSA screening to be non-existent or marginal while the harms are proven and substantial leading to the recommendation against PSA-based screening for prostate cancer by the U.S. Preventive Services Task Force (USPSTF) in 2012 (2). In a recently updated recommendation draft (2017), the USPSTF now suggests to inform patients about the potential benefits and harms of PSA-based screening opting for an individualized decision making (155). The problem continues with tissue biopsies that are taken and given a GS to predict prognosis. In addition to the risks of complications due to the invasive nature of conventional biopsies and the other limitations discussed previously, the majority of men diagnosed for PCa have an intermediate GS (7) and the problem arises on how to approach this group as the prognosis is difficult to predict (134). There is a critical need to develop strategies and techniques that reduce the burdens associated with the diagnosis of cancer and accurately classifying the disease. Benefits could be made by applying a personalized medicine approach, where each patient is treated based on their unique disease profile, and therefore better biomarkers and platforms are needed for detecting and monitoring the tumor status in real-time permitting therapy stratification. Liquid biopsies can be used to detect disease, monitor response to treatment, assess the emergence of drug resistance, and quantify minimal residual disease. The information gathered from liquid biopsies can be used to guide patient care and would ultimately benefit clinical practice through the integration of this into decision making.

13

Aims

Overall aim

This thesis evaluates whether circulating platelets could have a potential role (as a liquid biopsy source) in cancer in terms of diagnostics, therapy stratification and monitoring of the disease.

Specific aims

Paper I • To study whether platelets in men with castration resistant prostate

cancer (CRPC) store tumor-derived transcripts that could be used to predict therapy outcome.

Paper II • To investigate whether the predictability of therapy response enhances

by combining circulating biomarkers (cfDNA and TEPs).

Paper III • To identify early stage PCa and distinguish locally confined cancer (stage

I-II) from cancer that extends beyond the prostate capsule (stage IV-V) using TEP-RNA based classifiers obtained with the thromboSeq platform.

Paper IV • To investigate the potential and origin of spliced RNA profiles from TEPs

for the non-invasive detection of early- and late-stage NSCLC

14

Materials and Methods

The original publications contain detailed descriptions of the material and methods used in each study. All studies were approved by the ethics review board of Umeå University (Umeå, Sweden) and/or the institutions involved in the studies. All participants provided written consent.

Patient Samples

Whole blood was collected in EDTA-coated BD Vacutainers containing the anti-coagulant EDTA. Blood samples from PCa patients were collected within the Uppsala-Umeå Comprehensive Cancer Consortium (UCAN) at the University Hospital of Umeå (Umeå, Sweden). Blood from healthy donor samples were included as controls. Samples of cancer patients, patients with non-cancerous conditions, and asymptomatic individuals were also collected at various institutions.

Platelet Isolation

Platelets isolation was done by standard centrifugation at room temperature within two to 48 hours after whole blood was collected in EDTA-coated BD Vacutainers. EDTA is an anticoagulant that prevents clot formation. The cells were removed by centrifugation for 20 minutes at 120G, after which the platelet rich plasma was moved to a new tube and centrifuged for 10 minutes at 120G. The platelets were isolated from the supernatant by centrifugation at 400G for 20 minutes, followed by washing the platelet pellet in 27 mM EDTA PBS and repeating this centrifugation step. Next, the platelet pellet was resuspended in 100 µl RNAlater® (Ambion) and stored in -80°C. In parallel, platelet depleted

plasma was stored in -80°C.

RNA Extraction and cDNA Synthesis

Total RNA from the platelet fraction was extracted using RNeasy Mini Kit (Qiagen)/mirVana miRNA Isolation Kit (Ambion) according to the manufacturer’s instructions. The equivalent of RNA extracted from 1 mL of whole blood was used for reverse transcription and amplification using the whole transcriptome amplification kit (WTA-2, Sigma-Aldrich) according to the manufacturer’s protocol. The cDNA concentration was then determined for each

15

sample using a Nanodrop ND-1000 spectrophotometer (Nanodrop Technologies) or Qubit Fluorometer (Invitrogen).

Circulating free DNA Isolation

Free circulating DNA was extracted from the plasma fraction using the QIAamp Circulating Nucleic Acid Kit according to the manufacturer’s instructions (Qiagen). The concentration was measured by Qubit (Invitrogen) and the quality assessment was done using the Bioanalyzer 2100 (Agilent).

Droplet Digital PCR (ddPCR)

The ddPCR is a system based on water-oil emulsion technology. About 20,000 droplets are generated in which a single sample is fractionated before PCR amplification occurs in each individual droplet. The absolute transcript number for each sample was determined using the droplet digital PCR (QX-200, Bio-Rad). Biomarker status in the platelet fraction was defined as positive (three or more positive droplets) or negative (less than three positive droplets). Amplified cDNA was used in the PCR analysis with TaqMan assays KLK2, KLK3, FOLH1, ARv7 and NPY from Life Technologies. cfDNA was analyzed for copy number amplification within the AR locus (ARCNA) using Taqman assays (Life Technologies) ARCNA exon1, ARCNA exon8, and CYP1B1 as a copy number neutral (ARCNN) control. AR copy number amplification status was determined by a higher than 1,3 AR genomic copy numbers for the average of the two ARCNA assays compared to the control region.

RNA-in situ Hybridization (RNA-ISH)

ISH for KLK-3 mRNA was performed using the RNAscope 2.0 HD Detection kit-BROWN (Advanced Cell Diagnostics, Inc.) according to the manufacturer’s instructions. RNA-ISH amplifies target-specific signals and suppresses background noise, resulting in a better signal-to-noise ratio than nonspecific hybridization. In brief, cells were collected and fixed on slides by cytospin-centrifugation followed by pretreatment with protease III to permeabilize the cells, allowing access to target RNA. Next, the pre-warmed RNA-specific probes (Hs-KLK3, and negative control probe dapB) were hybridized to the target mRNA and a series of horseradish peroxidase-based amplification steps followed that hybridized the target probes. Staining intensity for platelets was modified by

16

adjusting the incubation time for the fifth amplification step from 30 min to 45 min. Finally, the target RNA was visualized using a standard bright field microscope (Leica DM LB, Leica microsystems, 100x).

RNA-Sequencing Library Preparation

Platelet RNA quality was evaluated using the RNA 6000 Pico chip (Bioanalyzer 2100, Agilent) and were only included for subsequent experiments if a RIN-value of at least seven and/or distinctive rRNA curves were obtained. The RNA concentration was measured with the Qubit Fluorometer (Invitrogen). In order to have sufficient platelet cDNA for RNA-seq library preparation, cDNA synthesis and amplification was done using the SMARTer Ultra Low RNA kit for Illumina Sequencing v3 (Clontech). Indexed libraries were prepared using ~500 pg total RNA as starting material for the SMARTer amplification. The quality of the amplified cDNA was checked using the DNA High Sensitivity chip (Bioanalyzer 2100, Agilent). Sample fragments with a distinct peak spanning 400-10,000 base pair (bp) region were selected for further processing. Next, platelet cDNA was fragmented by sonication (Covaris Inc.) and labeled with single index barcodes for Illumina sequencing using the TruSeq Nano DNA Sample Prep Kit (Illumina). Because of the low platelet cDNA input concentration all bead clean-up steps were done using a 15-minute bead-cDNA binding step and a 10-cycle enrichment PCR. For all other steps the manufacturer’s protocol was followed. Quality and quantity was measured using the DNA 7500 or High Sensitivity chip (Bioanalyzer 2100, Agilent) and high-quality samples with product sizes between 300-500 bp were pooled to a final concentration of 10 nM (12 samples per pool). The concentration was measured again using the High Sensitivity chip (Bioanalyzer 2100, Agilent) and submitted for sequencing on the Illumina HiSeq 2500 platform.

Statistics

For detailed description of statistical methods used in each manuscript, see the respective papers’ statistical paragraphs. In brief, time-to-event was evaluated with Kaplan-Meier and the log rank test was used to determine significance. Univariate COX regression analysis was used for assessments of hazard ratios between biomarkers and risk of event. Forward stepwise conditional logistic regression was used to build multivariate predictive models and to evaluate

17

independence between biomarkers. Chi-squared test was used to evaluate the correlation between digitalized (dichotomized) biomarkers. Significance was defined by two-sided tests with a P-value of 0.05 or less. Analysis of variance (ANOVA) was used to find differentially spliced genes between patient groups and Wilcoxon test was used to identify difference in means. Analyses were performed with the use of SPSS and R-studio.

Data analysis

The raw sequencing data generated encoded in FASTQ-files (FASTQ (raw), Figure 3) underwent quality control analysis using FASTQC. Adapters and low-quality bases were subsequently trimmed off using Trim Galore/Trimmomatic, the trimmed reads (FASTQC (trimmed)) were then aligned to the human reference genome (hg19) using STAR (156). Differentially expressed/spliced genes were identified as described in Best et al. (46). All subsequent statistical and analytical analyses were performed in R-studio. Figure 3 shows a simplified workflow of RNA-seq data processing and analysis.

Figure 3. RNA-sequencing processing and analysis workflow. The steps represent the process the data underwent and the tool used to format data at each.

18

Results and Discussion

Paper I

Platelets Harbor Prostate Cancer Biomarkers and the Ability to Predict Therapeutic Response to Abiraterone in Castration Resistant Patients. Patients benefit most when treatments are tailored to their specific needs and even more so by predicting if and when a therapy is not effective to prevent unnecessary suffering of side effects and to possibly receive a better alternative treatment. In addition, a significant amount in costs of therapy and medical care can be prevented. A personal medicine (PM) approach requires real-time monitoring to determine ongoing changes within the tumor and to tailor treatment based on real-time information (i.e., therapy stratification). Here, we investigated whether platelets in men with CRPC store tumor-derived transcripts as previously been shown for other tumor types (lung and glioma) and whether this information could be used to predict therapy outcome. Two patient cohorts treated for CRPC were selected: one treated with docetaxel as 1st line therapy and one treated with abiraterone. We examined the association between biomarkers and therapy response in patients naïve to treatment and thus not pre-selected for resistance. Platelets from CRPC patients contained PCa specific transcripts of which some (KLK2, KLK3 and FOLH1) might be indirect measurements of tumor burden as patients with detectable levels of these transcripts had significantly higher levels of serum PSA than patients in which these transcripts were not detected. The larger the tumor the higher the PSA level is expected to be as more PSA leaks out into the vasculature. The main finding in this study was a three-gene biomarker panel (KLK3, NPY and FOLH1) that provided significant information on therapy response for the abiraterone cohort. If patients were positive for any of these three biomarkers, they had a shorter time to progression (short PFS) and an increased risk of therapy failure compared to biomarker negative patients. These patients may have benefited from an alternative therapy from the start. Among the non-reponders to abiraterone were patients that had an almost mutually exclusive platelet expression pattern for FOLH1 and NPY that provided independent predictive information before treatement. These could indicate the presence of

19

different/distinct molecular subtypes, both with short PFS yet differing in risk of death and OS. To enable further therapeutic developments, it would be interesting to understand the molecular mechanism underlying abiraterone resistance in regards to the FOLH1 and NPY expression. For the chemotherapy cohort, we found no differences in response, irrespective of the suggested resistance biomarkers status, so switching the non-abiraterone responders to a chemotherapy could potentially be an alternative option. Furthermore, these PCa-derived biomarkers residing in the platelet population were better predictors of therapeutic outcome of CRPC patients treated with abiraterone compared to baseline serum PSA and best PSA response.

20

Paper II

Combining Circulating Biomarkers to Enhance Predictability of Therapy Response. For PCa therapy most advances have been made for CRPC patients, with many new therapies showing good effects in a subpopulation of patients, but so far, no companion diagnostic test is available to personalize therapy stratification. For the implementation of personalized medicine, it is of importance to accurately monitor the disease in real time for therapy stratification. Our previous study (Paper I) enabled TEPs to predict therapy responders vs. non-responders in an abiraterone treatment cohort using a three-gene biomarker panel. Many other studies have looked into AR copy number amplifications (ARCNA) as well as mutations using cfDNA and/or CTCs as treatment induced alterations take place due to selection pressure from AR directed therapies, making the patients non-responsive and ongoing therapy useless. The availability of cfDNA from matched plasma samples (to samples used in Paper I of which platelet RNA was isolated and the three-gene panel was found), allowed us to monitor and compare two different biosources and investigate whether combining these biosources (cfDNA and platelet RNA) enhanced predictability of therapy response. The longitudinal monitoring using both biosources provided information, however the platelet biosource gave a better indication of tumor status and progression, while changes in the ARCNA using cfDNA correlated more with tumor burden but did not aid in patient stratification. The association between ARCNA and the individual genes found in platelets comprising the three-gene panel was looked into. KLK-3 expression level was significantly associated with ARCNA, which is expected as KLK-3 is AR regulated. Interestingly the FOLH1 and NPY behaved differently depending on amplification status of AR, co-expressed in ARCNA and mutually exclusive in ARCNN. Again, as in previous study indicating two distinct subpopulations. The findings are based on a small cohort and need to be validated in a larger cohort but implies that ARCNA status may be important in how the patient respond to therapy. Furthermore, including ARCNA information with the three-gene biomarker panel (3GA panel), improved the accuracy in identifying non-responders to therapy

21

compared to the three-gene panel alone. Implementing this test would have a profound influence on how we select therapies and follow the effect of treatment, especially since changes in the platelet biomarkers could monitor ongoing tumor activity.

22

Paper III

Detection of Early Stage Prostate Cancer and Identification of High Risk Patients Using Tumor Educated Platelet Profiles. The best available biomarker to date for the detection of PCa is serum PSA, however the drawback of overdiagnosis and overtreatment of conditions that may never really affect the patient during their lifetime are high. Improvement could be made in this perspective and in determining the prognosis and risk (classification groups). These are currently based on the PSA level together with the gleason score and TNM-stage, but the (difference in) molecular genetic profile/information is not taken into account. The thromboSeq platform has been used for cancer detection and monitoring of tumor phenotype by interrogating platelet RNA as a liquid biopsy source. This study is based on previous co-publications, and aims to diagnose PCa, especially early stage PCa and distinguish locally confined cancer (stage I-II) from cancer that extends beyond the prostate capsule (stage IV-V) using TEP-RNA based classifiers obtained with the thromboSeq platform. Taking previous raised considerations into account when assembling the cohort and processing the samples, the confounding factors have been limited. Since PCa is a disease restricted to occur in men, bias due to sex is nonexistent and blood storage and processing time is shortened reducing the influence of potential confounding factors. In line with previous publication of the group, the platelet platform allowed the identification of PCa vs. HD with high accuracy and identified genes that were differentially spliced. The differentially spliced profiles were used to generate classifiers that predicted PCa (stage I-V) with 93% accuracy, 97% sensitivity and 80% specificity compared to HD based on 505 genes. Early stage cancer (stage I-II) versus HD could be predicted with 95% accuracy, 100% sensitivity and 90% specificity based on 46 genes (Table 3). We also looked at the proportion of spliced transcripts (the splicing factor) and found higher values in platelets from cancer patients compared to the HD cohort. in line with hyper reactive platelets and increased turnover. It is thought that the disease stage leaves a molecular fingerprint that can both be used to identify disease and classify early stage from late stage disease. When we

23

looked at the prediction model for early stage (I-II) vs. late stage (IV-V), 85% of the late stage cases were accurately identified, however there was high risk of false positives (Tabel 3).

Table 3. Overview Clinical performance of the predictions using the thromboSeq platform and TEP-profile based classifiers.

Together, the thromboSeq platform and TEP-profile based classifiers could potentially be added (in addition to PSA, GS and TMN-stage) as a companion diagnostic test to accurately diagnose PCa. Larger cohorts that represent the broad heterogeneity known to exist in PCa, including new parameters like molecular subtype will be needed to improve risk stage classification, determine prognosis and provide information for treatment options. As previously confirmed, age is a confounding factor that may influence the platelet classification score (Paper IV). So far patient age has not influenced the classification score in any way that alters the test outcome (Paper IV). To this end the cohort will be supplemented with additional samples providing a better matched control cohort thereby limiting age as a possible confounding factor.

24

Paper IV

Swarm Intelligence-Enhanced Detection of Non-Small-Cell Lung Cancer Using Tumor-Educated Platelets. Recent developments in detection of cancer-specific transcriptomic alterations in platelets by our group and collaborators added another possible minimal invasive liquid biopsy source in biomarker research. This led to a major breakthrough with the publication of Best et al. 2015, where they were first to report platelets as a blood-based biosource that could be used for multiclass cancer diagnostics through the molecular interrogation of TEP mRNA via self-learning support vector machine (SVM)-based algorithms. They named this multiplexed biomarker signature detection platform thromboSeq. Questions were raised regarding confounding factors such as age that are discussed and answered in this paper. The aim was to investigate the potential and origin of spliced RNA profiles found in TEPs for the non-invasive detection of early- and late-stage NSCLC. We confirmed that the age of the individual and blood storage time can influence the platelet classification score (p = 0.002 and p = 0.09, respectively). Although not significant, blood storage time could result in a statistically significant contribution in a larger dataset. The sample cohorts for NSCLC and non-cancer were assembled matching age, smoking status, and blood storage time to assess the contribution of potential technical and biological variables and to examine the RNA profiles and RNA processing pathways of platelets. New techniques were used to reduce confounding factors, and one of the most important addition was the use of remove unwanted variation (RUV)-normalization, which reduces the relative intersample variability. In addition, particle- swarm optimization (PSO) driven algorithms were used as PSO driven algorithms are inspired by the social behavior of birds flocking and schools of fish that through self-organization efficiently adapt to their environment. This could be used to enhance the selection process of candidate genes for classification algorithms. With these techniques added to the thromboSeq workflow, the intron-spanning RNA reads were selected for the downstream computational platelet RNA analysis.

25

In this study we demonstrated that PSO enhanced algorithms enabled efficient selection of TEP RNA profiles that could accurately identify early (accuracy, 81%; AUC, 0.89; 95% CI, 0.83-0.95; p<0.001) and late stage (accuracy, 88%; AUC, 0.94; 95% CI, 0.92-0.96; p<0.001) NSCLC without influence of potentially confounding factors i.e., smoking status, patient age, and sample processing. Furthermore, we found that the spliced platelet RNA profiles correlated with a P-selectin profile, an increased alternative splicing, and exon skipping in TEPs in cancer patients. PAGODA gene ontology (GO) analysis was used to functionally annotate the differentially spliced platelet RNA in patients with NSCLC. Gene ontologies related to translation, RNA-binding proteins (RBPs), and intraplatelet signaling were among the most significant biological groups with low splicing score in the NSCLC samples compared with non-cancer samples. Genes related to interplatelet signaling and immune response were the most significantly enriched gene cluster in NSCLC patients compared with the non-cancer cohort. Therefore, we investigated the extent of RBP-motifs in the UTRs within the platelet transcriptome and identified unique RBP signatures in TEP RNA profiles. Robust biomarker selection for blood-based cancer diagnostics, independent of bias introduced by the individual’s age, smoking habits, blood storage time, and inflammatory diseases, is permitted by use of the PSO-driven thromboSeq platform. This platform allows samples to be collected at one site and be shipped to another site within 48h for processing and analysis; however, there is room for improving procedures and methods that stabilize the RNA content during transport and this will be an important area for future research. Ultimately, large scale validation of TEPs for (early) detection of NSCLC and other cancer types is necessary.

26

Conclusions

• Paper I Platelets add information to a highly important matter that needs to be further addressed in order to develop personalized medicine. That is, more research is needed to understand how to use circulating tumor material in a way that enables the right treatment to be given to the right patient in the right sequence. Cancer-derived biomarkers with therapy-predictive information were found in the platelet population, so platelets may be one platform that can address the problem of high costs associated with novel therapies as this platform identifies therapy responders. This approach avoids treating non-responders with expensive and potentially ineffective treatments.

• Paper II The analysis of circulating biomarkers in the platelet population can be used to predict with high accuracy the patients who will benefit from androgen synthesis (AS) directed therapy and can therefore be used in treatment stratification. cfDNA analysis did not improve accuracy of therapy stratification in this AS-therapy naïve cohort but combining biosources may improve accuracy of therapy stratification. With respect to longitudinal monitoring of therapy response, both platelet and cfDNA biosources provide information and combining these could potentially be useful in subtyping specific characterizations of the tumor.

• Paper III Tumor induced changes in the transcriptome of the platelet population may be exploited to diagnose PCa. The thromboSeq platform and TEP-profile based classifiers could potentially serve as a companion diagnostic test to accurately classify risk stage, determine prognosis and correct treatment options, reducing the need for repeated tissue biopsies. Both number of differentially spliced transcripts and proportion of spliced transcripts between groups could be used as transcriptional biomarkers in distinguishing PCa patients from HD samples. In addition, the thromboSeq platform enabled the identification of tumor-derived transcripts within the platelet transcriptome, putting forward the use of this platform for analysis of tumor-derived biomarkers.

27

• Paper IV Particle-swarm optimization (PSO) driven algorithms can be used to identify optimal biomarker gene lists, resulting in a tumor-educated platelet RNA biomarker panel that distinguishes patients with NSCLC from healthy individuals and patients with various non-cancerous inflammatory conditions, without the contribution of potential confounding factors.

28

General Discussion

Cancer treatment is slowly shifting from a one-size fits all concept to individualized precision therapies in an attempt to limit toxicity and side effects. To guide personalized medical decision making, we need prognostic biomarkers that enable us to identify patients who are most likely to respond and benefit from their assigned therapy. Blood-based biomarkers can aid in this search and improve therapy monitoring, so the choice for an alternative treatment can be made as soon as resistance is identified. PCa is one example of potentially many other cancers that would benefit greatly from the use of liquid biopsy-based biomarkers as proper cancer diagnosis and patient stratification would provide better health care outcomes. This thesis supports the findings that platelets of cancer patients contain tumor-derived biomarkers that can be used to develop biomarker panels that can accurately discriminate early stage cancer patients from healthy individuals as well as therapy responders from non-responders. In addition, cfDNA of matched patient samples allowed for the assessment of genomic changes (AR amplifications) and the analysis of RNA transcripts in platelets. Both sources contained biological information about the tumor; however, platelet RNA transcripts provided more information regarding therapy stratification. Combinations of biosources may complement each other and enhance predictability of therapy response.

Platelets as a Biomarker source

Platelets are an interesting source for the detection of disease-related biomarkers that are released into the bloodstream as they are now also known for their many interactions with various cell types and circulating molecules such as interacting and ingesting pathogens encountered in addition to selective uptake capacity of macromolecules in circulation. The support for platelets as a liquid biopsy is provided by studies revealing selective uptake and storage of tumor-derived proteins (79) and uptake of nucleic acids from the tumor (45, 47). Klement et al. demonstrated selective accumulation of angiogenic factors released by tumor cells occur in platelets, and the increase of these factors in the platelets could be used to identify microscopic tumors, indicating that this storage in platelets could accumulate higher levels of

29

biomarkers compared with the level in plasma. Furthermore, as platelets are anucleated and no new RNA molecules can be synthesized, they could be used as a diagnostic biosource for RNA biomarkers. The RNA found within platelets is either transferred from MKs during production or has been taken up from the circulation. mRNA in platelets is conserved either silenced until they are needed (157) or is stored as pre-mRNA containing introns that upon activation are spliced out and form the mature mRNA (49). Using RNA over DNA as biomarker to detect or monitor tumors has the advantage that active ongoing gene expression in living tumor cells is represented, whereas cfDNA is mostly released during apoptosis or by apoptotic cells (158). Which can make one wonder why cfDNA analysis mainly released from dying tumor cells would be clinically relevant and provide information on resistant clones. Unlike DNA, RNA expression is dynamic and constantly changing based on environmental and ‘educational’ cues (e.g., pathogens, wounds, cancer, etc.), flexibly altering their transcriptome and protein translation. In addition, the use of platelets provides a window into the disease process showing biomarker accumulation within the platelet lifespan (7-10 days), whereas the free-circulating biomarkers are cleared rapidly once in circulation (24). Moreover, platelets are the second most abundant cells in blood and easily accessible, unlike ctDNA and CTCs which remain technically challenging to detect. Patients with an early stage disease present very low concentration of CTCs or release low amounts of ctDNA. Since most cancers are heterogeneous, it is debated whether the few CTCs found represent the heterogeneous tumor cell population. EVs, although promising, require a cumbersome isolation methodology. Furthermore, because EVs comprise a diverse range of vesicles, it is not always defined what EV population has been studied by different groups. EVs have the tendency to bind exogenous molecules thereby affecting their purity and these contaminants that bind to the surface membranes of the vesicles are not real parts of their molecular content, complicating molecular profiling.

Clinical Importance

Cancer relevant information can easily be provided by platelets as cancer-induced alterations have been shown to occur. Platelets are easily accessible and a minimally invasive source that can be used to identify new cancer cases and the tumor type, monitor cancer progression, and predict therapy response, leading to the implementation of personalized medicine. Personalized medicine has the

30

potential to tailor therapy with the best response and highest safety margin to ensure better patient care. By providing patients with early diagnoses, risk assessments, and optimal treatments, personalized medicine holds promise for improving health care while also lowering costs.

With the 3GA biomarker panel (cfDNA ARCNA and platelet 3G panel), for example, we can hypothesize the potential benefits of a blood-based biosource test (Figure 4). Generally, when PCa patients are not preselected for therapy the response rate for abiraterone treatment is about 33%, subjecting 670 patients (when taking a patient cohort example of n=1000) to expensive ineffective treatment. With a hypothetical screening of the same 1000 patients using the 3GA profile would result into 333 predicted responders of which 87,5% (FPR=0,125) would actually respond to therapy (TR; n=291) and 42 non-responders (NR). The predicted non-responders (n=666) could then be stratified for alternative treatment options. Interrogating the platelet RNA could for instance select 180 patients expressing FOLH1 (encoding prostate specific membrane antigen (PSMA)) for the newly available PSMA-Lu177 treatment and delivering the therapy via PSMA protein on the tumor cells.

This example on information provided by platelets in combination with cfDNA could therefore provide better care and limiting cost due to the use of ineffective treatments.

Figure 4. Visualization of hypothetical cost-benefit of implementing the 3GA panel (cfDNA ARCNA and platelet 3G panel) for therapy stratification. TR, Therapy Responders; NR, Non-Responders.

1000patientsaretreatedwithAbiraterone

333patientsstratifiedfortreatment

Screeningwith3GAprofile(n=1000)

666patientsexcluded

PredictedResponder PredictedNon-Responder

291patientsrespondtoAbiraterone

42patientfailtherapy(FPR)

NR

Noscreen

330patientsrespondtoAbiraterone

670patientfailtherapy

180patientscouldbetreatedwithPSMA-Lu177

83patientaremissed(FNR)

FOLH1positive(27%#)

Alternativetherapy

TR NR

CRPCpatientswithoutpriorAS-directed

therapy

Mediantimetoprogression(PFS)4.1month |🔼 6month| 10.1month

CabazitaxelRadium223

PARPi

2rd linetherapy- Abirateronevsother

Failed1st linetherapy- Docetaxel

Therapystratification

TR

31

Liquid biopsies may not fully replace conventional tissue biopsies for diagnostic purposes but are promising tools for patient stratification that enables longitudinal monitoring of cancer patients (throughout treatment) and may have significant impact by complementing cancer management and other disease areas, by aiding personalized decision strategies. Each blood-based biomarker source has its advantages and limitations. Some major drawbacks include the restricted amount of tumor-derived information in early stage cancers, complicating the separation of information derived from tumor vs. healthy cells, as well as defining the origin of the tumor. Combining blood-based biosources could help develop a fuller picture of the ongoing disease process, increase sensitivity and enhance clinical possibilities.

32

Acknowledgements

As I reached the end of writing this thesis and getting a glimpse of that dot of light at the end of the tunnel, the biggest obstacle I realize now has been to write the acknowledgements as it comes with this bittersweet feeling that the writing has come to an end, but so has a chapter of 5 years or even more, considering the previous roads that all led to this thesis. During this time, I have learned and experienced so much, professionally and on a more personal level, that I will always cherish. I have been blessed to meet exceptional people that I hold in high regards and consider role models in some ways. I would like to thank everyone that has contributed and helped me in one way or another along this journey. First, I would like to thank my supervisor Jonas Nilsson for giving me the opportunity to develop as a scientist and guiding me through my PhD. Your love for science is inspiring, as is your ability to generate a million ideas a minute. You have always come through for me: besides being a great supervisor you have also been a great friend. Linda Köhn, I was so happy when I heard you were joining the group! Finally, we would be a real group in which the other group members were not my multiple personalities. I’m happy you got over your initial shock sharing an office with me, you being quite reserved, organized and calm vs. me all over the place, and an open book…probably sharing way more than you bargained for. I cannot begin to express my gratitude for all your help during this time, and I’m glad we decided to skip the middle man and communicate directly. You and your wonderful family have shown me another side of life in Northern Sweden: hiking in Skuleskogen, outdoor cooking and, the fun day out snowmobiling. Thanks to you all for taking me on these adventures. Pernilla Wikström, my co-supervisor, Thank you for all your input in the projects and your constructive feedback on the papers. I greatly appreciated this and I especially want to thank you for all the help and support during the time I felt abandoned. Håkan Hedman – even though not officially my supervisor, you have gone above and beyond to support me. Your guidance has been extremely valuable, I appreciate that you stepped in when I needed a temporary supervisor and all of your efforts in keeping me sane and guiding me through that period in my PhD. Anders Widmark – I really appreciate you taking the time to educate me on the clinical side of prostate cancer. The enthusiasm with which you go about the

33

work truly indicates how great you are at what you do. I will always remember the ‘Fiat, Volvo, and Porsche’ comparison to disease progression. Tom Wurdinger – As you mentored both Jonas and me, I somehow feel like this journey has all started with you. Without your enthusiasm and positivity, I would not be where I am today. You have been a crucial part in this and I sincerely thank you for that. I would like to thank the co-authors on all of the papers listed in this thesis for their valuable comments and input. Many thanks to Maria Wing for all the help throughout the years, I cannot imagine not having you around. I am truly grateful to all the past and present members of the Oncology research group. Annika, Yvonne, Kristina, Emma, Gunnar, Anna, Sophia, Micke K, Terese, Samuel, Carl, Sylvia, David, Charlotte, Lotta, Mahmood, Thais, Martin, Åsa, Roger, Calle, Andreas, Daniel, Fredrik, Tommy, James, Cedric, and Beatrice – I would like to thank each and every one of you for making me feel so welcomed and at home. Mikael J, thank you for your kind words and confidence in me completing this PhD. I will always remember you sharing your champagne knowledge with us at my surprise birthday champagne tasting. It was fantastic! Ulrika, the thumbs up from across the office while writing this thesis has helped a lot, even if it was to just put a smile on my face for a split second. It was a great experience joining you to the very exciting hockey game. Camilla, you have been such a BIG support to me throughout these years. In you I found a great colleague, talented and patient therapist but most of all a true friend. I have enjoyed all the time we spent together catching up over a tea, the dinners and going to the forest picking mushrooms and berries. The roomies: Soma, Annika, Anna Chiara, Simona, Olga, Britta, and Rike. Soma, you were the first friend I made in Umeå and joining the lab. You are very kindhearted and I want to thank you for all the support during this time. Annika, thank you for all the nice long chats and listening when I needed to vent. Anna Chiara, how can anyone forget you?! Rough on the edges but the biggest heart ever. I loved the times we spent forever working in the lab. Even after your time in Umeå, we have managed to stay in touch, supporting each other from a distance and I really appreciate you as a friend. Simona you have been such a good friend I am super proud of you and have respect for all the challenges you have overcome so far. Olga!! Thank you so much for all the great times and conversations. Hoping to see you and your adorable son soon. Britta, I love the way you are full of energy and see the positive side to everything. It has been a pleasure sharing the office with you and having office after work Fridays with you, Tommy and Rike. Rike, you joined the group last and became our Lab baby * :). Thank you so much for

34

being a great friend, ready to egg a house if it would make me feel better. Looking forward to visiting you in Hamburg. (no houses were egged btw…)

You all contributed to the great atmosphere here at the lab. THANK YOU!<3 Many thanks to the Prostate Group for all the support, advice and willingness to help when needed, especially Marie, Sophia and Mona S. Neuro-Oncology Research Group in Amsterdam, thank you for hosting me in your lab and guiding me through the techniques for RNA sequencing. The Lipid (Fat) group, it was great getting to know each one of you and sharing the great times in Brazil. Stefan, you have been a good friend and a lot of help with the party plan. Thanks for all the caffeine breaks and listening to my drama due to procrastinating. I would like to thank Clas and Terry; if it was not something you could personally help me with you knew exactly who to direct me to. Also, thank you to all the friendly people at 6M: Ravi, Pramood, Karthik, Stina, Lele, Alexej, Yabing, Kerstin, Kjell, Song Jie, Clara, Gunilla, Marene, Francesca, Sandhya, Dario and Cyrinne; especially the lunch and social (beer) corner friends: Denise, Bethie, Anton, Elin, Erik BY, Johan, Stefan, Fredrick, Ola, Lars, Helena, Robin, Erik D, Hélène, Björn, Mona, Mariam, Zahra, Mattias, Oskari, Christine, Oleg, Valeria, Isil and Xingru. Thank you for all the fun and laughter. Somehow the weirdest things always ended up being discussed. Mariam, thank you for all your help, support and tips during the thesis application and the motivational chats while writing. I was sooo lucky you dropped by the office right before my proof print deadline and now ended up with this amazing cover, thank you! Thanks Cheng Choo ‘Nikki’ Lee, for taking the SEM images used for this cover and to Oleg for the brilliant photoshopped figures for the cover. Thank you to all my friends and family near and far. Leonora, I am so happy you stopped by my office in Amsterdam, exactly 10 years ago! You have also played a major part in this journey with all the advice, encouraging talks, and guidance throughout this time. You are the sister I never had. Thank you for everything! Johan, before meeting you in Boston, I had never heard of Umeå. Who would have thought that this is where I would end up? Thank you both for all the support and great times in Boston and Umeå with your family and the girls. Casey, thanks for your encouraging words and support. Flavia, Edward and Sophia, thank you for all the fun and laughs. It is always great to see you guys when visiting.

35

The support system below sea level: Pamela, Stacey, Lotte, Sjoerd, Ravi, Sylvia, Arpana, Cindy, Elham, Negisti, Shirley, and Maisi. Even though we do not see each other that often, you all have supported me from day one and I am genuinely grateful for that. Bedankt voor alle gezelligheid buiten het werk! All the friends that I met here in Umeå that have become family: Jon, Nadia, Roberto, Babs, Anna, Thais, Simona, Brina, Denise, Valeria, Oleg, Soma, Andrés, Bethie, Rike, Björn, Isil and Xingru. I would not have survived Umeå had it not been for you. The many road trips (Amsterdam, Copenhagen, Lofoten), all the traveling (Southeast Asia, Istanbul, Philippines), dinners, and fun times spent together will always be cherished. Last but not least, my parents. Thank you for your unconditional love and support and for always being there for me. Even while being mad at me for leaving the lab late at night and walking home alone you were there on the phone, making sure I got home safe. I dedicate this thesis to you as this is the result of all the sacrifices you made hoping for a better future for us. To my brother, Ian, joking about how I am in a relationship with a cell in the cell culture room. Somehow you always needed to talk to me urgently when I needed to take care of my cells. Thank you and Jenny for all the support. Oma Cheng, dank voor alle steun en liefde. The studies were supported by the Lions Cancerforskningsfond and Cancerforskningsfond Norrland.

36

References

1. F. Bray, J. Lortet-Tieulent, J. Ferlay, D. Forman, A. Auvinen, Prostate cancer incidence and mortality trends in 37 European countries: an overview. Eur J Cancer 46, 3040-3052 (2010).

2. P. F. Pinsky, P. C. Prorok, B. S. Kramer, Prostate Cancer Screening - A Perspective on the Current State of the Evidence. N Engl J Med 376, 1285-1289 (2017).

3. R. B. Nadler, P. A. Humphrey, D. S. Smith, W. J. Catalona, T. L. Ratliff, Effect of inflammation and benign prostatic hyperplasia on elevated serum prostate specific antigen levels. J Urol 154, 407-413 (1995).

4. S. Roychowdhury et al., Personalized oncology through integrative high-throughput sequencing: a pilot study. Sci Transl Med 3, 111ra121 (2011).

5. A. K. Taylor et al., Targeted antimicrobial prophylaxis using rectal swab cultures in men undergoing transrectal ultrasound guided prostate biopsy is associated with reduced incidence of postoperative infectious complications and cost of care. J Urol 187, 1275-1279 (2012).

6. E. S. Kim et al., Gefitinib versus docetaxel in previously treated non-small-cell lung cancer (INTEREST): a randomised phase III trial. Lancet 372, 1809-1818 (2008).

7. N. Quach, M. F. Goodman, D. Shibata, In vitro mutation artifacts after formalin fixation and error prone translesion synthesis during PCR. BMC Clin Pathol 4, 1 (2004).

8. P. L. Bedard, A. R. Hansen, M. J. Ratain, L. L. Siu, Tumour heterogeneity in the clinic. Nature 501, 355-364 (2013).

9. N. Andor et al., Pan-cancer analysis of the extent and consequences of intratumor heterogeneity. Nat Med 22, 105-113 (2016).

10. P. C. Boutros et al., Spatial genomic heterogeneity within localized, multifocal prostate cancer. Nat Genet 47, 736-745 (2015).

11. C. S. Cooper et al., Analysis of the genetic phylogeny of multifocal prostate cancer identifies multiple independent clonal expansions in neoplastic and morphologically normal prostate tissue. Nat Genet 47, 367-372 (2015).

12. M. Karavitakis, H. U. Ahmed, P. D. Abel, S. Hazell, M. H. Winkler, Tumor focality in prostate cancer: implications for focal therapy. Nat Rev Clin Oncol 8, 48-55 (2011).

13. M. Andreoiu, L. Cheng, Multifocal prostate cancer: biologic, prognostic, and therapeutic implications. Hum Pathol 41, 781-793 (2010).

14. A. R. Patel, J. S. Jones, Optimal biopsy strategies for the diagnosis and staging of prostate cancer. Curr Opin Urol 19, 232-237 (2009).

37

15. K. Cyll et al., Tumour heterogeneity poses a significant challenge to cancer biomarker research. Br J Cancer 117, 367-375 (2017).

16. A. F. Gazdar, Activating and resistance mutations of EGFR in non-small-cell lung cancer: role in clinical response to EGFR tyrosine kinase inhibitors. Oncogene 28 Suppl 1, S24-31 (2009).

17. S. Kobayashi et al., EGFR mutation and resistance of non-small-cell lung cancer to gefitinib. N Engl J Med 352, 786-792 (2005).

18. M. Roschewski, L. M. Staudt, W. H. Wilson, Dynamic monitoring of circulating tumor DNA in non-Hodgkin lymphoma. Blood 127, 3127-3132 (2016).

19. P. Mandel, P. Metais, [Not Available]. C R Seances Soc Biol Fil 142, 241-243 (1948).

20. C. Bettegowda et al., Detection of circulating tumor DNA in early- and late-stage human malignancies. Sci Transl Med 6, 224ra224 (2014).

21. R. Lehmann-Werman et al., Identification of tissue-specific cell death using methylation patterns of circulating DNA. Proc Natl Acad Sci U S A 113, E1826-1834 (2016).

22. J. Luo, L. Shen, D. Zheng, Diagnostic value of circulating free DNA for the detection of EGFR mutation status in NSCLC: a systematic review and meta-analysis. Sci Rep 4, 6269 (2014).

23. T. Reinert et al., Analysis of circulating tumour DNA to monitor disease burden following colorectal cancer surgery. Gut 65, 625-634 (2016).

24. F. Diehl et al., Circulating mutant DNA to assess tumor dynamics. Nat Med 14, 985-990 (2008).

25. E. S. Lianidou, A. Strati, A. Markou, Circulating tumor cells as promising novel biomarkers in solid cancers. Crit Rev Clin Lab Sci 51, 160-171 (2014).

26. E. S. Antonarakis et al., AR-V7 and resistance to enzalutamide and abiraterone in prostate cancer. N Engl J Med 371, 1028-1038 (2014).

27. S. Maheswaran et al., Detection of mutations in EGFR in circulating lung-cancer cells. N Engl J Med 359, 366-377 (2008).

28. S. M. Park et al., Molecular profiling of single circulating tumor cells from lung cancer patients. Proc Natl Acad Sci U S A 113, E8379-E8386 (2016).

29. W. J. Allard 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 10, 6897-6904 (2004).

30. M. C. Miller, G. V. Doyle, L. W. Terstappen, Significance of Circulating Tumor Cells Detected by the CellSearch System in Patients with Metastatic Breast Colorectal and Prostate Cancer. J Oncol 2010, 617421 (2010).

38

31. C. Aggarwal et al., Relationship among circulating tumor cells, CEA and overall survival in patients with metastatic colorectal cancer. Ann Oncol 24, 420-428 (2013).

32. M. Cristofanilli et al., Circulating tumor cells, disease progression, and survival in metastatic breast cancer. N Engl J Med 351, 781-791 (2004).

33. D. C. Danila et al., Circulating tumor cell number and prognosis in progressive castration-resistant prostate cancer. Clin Cancer Res 13, 7053-7058 (2007).

34. J. S. de Bono et al., Circulating tumor cells predict survival benefit from treatment in metastatic castration-resistant prostate cancer. Clin Cancer Res 14, 6302-6309 (2008).

35. U. De Giorgi et al., Circulating tumor cells and [18F]fluorodeoxyglucose positron emission tomography/computed tomography for outcome prediction in metastatic breast cancer. J Clin Oncol 27, 3303-3311 (2009).

36. A. Romanel et al., Plasma AR and abiraterone-resistant prostate cancer. Sci Transl Med 7, 312re310 (2015).

37. B. Costa-Silva et al., Pancreatic cancer exosomes initiate pre-metastatic niche formation in the liver. Nat Cell Biol 17, 816-826 (2015).

38. H. Peinado et al., Melanoma exosomes educate bone marrow progenitor cells toward a pro-metastatic phenotype through MET. Nat Med 18, 883-891 (2012).

39. L. Balaj et al., Tumour microvesicles contain retrotransposon elements and amplified oncogene sequences. Nat Commun 2, 180 (2011).

40. C. Kahlert et al., Identification of double-stranded genomic DNA spanning all chromosomes with mutated KRAS and p53 DNA in the serum exosomes of patients with pancreatic cancer. J Biol Chem 289, 3869-3875 (2014).

41. J. Nilsson et al., Prostate cancer-derived urine exosomes: a novel approach to biomarkers for prostate cancer. Br J Cancer 100, 1603-1607 (2009).

42. J. Skog et al., Glioblastoma microvesicles transport RNA and proteins that promote tumour growth and provide diagnostic biomarkers. Nat Cell Biol 10, 1470-1476 (2008).

43. B. K. Thakur et al., Double-stranded DNA in exosomes: a novel biomarker in cancer detection. Cell Res 24, 766-769 (2014).

44. C. Lasser, S. C. Jang, J. Lotvall, Subpopulations of extracellular vesicles and their therapeutic potential. Mol Aspects Med, (2018).

45. R. J. Nilsson et al., Blood platelets contain tumor-derived RNA biomarkers. Blood 118, 3680-3683 (2011).

46. M. G. Best et al., RNA-Seq of Tumor-Educated Platelets Enables Blood-Based Pan-Cancer, Multiclass, and Molecular Pathway Cancer Diagnostics. Cancer Cell 28, 666-676 (2015).

39

47. R. J. Nilsson et al., Rearranged EML4-ALK fusion transcripts sequester in circulating blood platelets and enable blood-based crizotinib response monitoring in non-small-cell lung cancer. Oncotarget 7, 1066-1075 (2016).

48. G. T. Brown, T. M. McIntyre, Lipopolysaccharide signaling without a nucleus: kinase cascades stimulate platelet shedding of proinflammatory IL-1beta-rich microparticles. J Immunol 186, 5489-5496 (2011).

49. M. M. Denis et al., Escaping the nuclear confines: signal-dependent pre-mRNA splicing in anucleate platelets. Cell 122, 379-391 (2005).

50. E. Meshorer, T. Misteli, Splicing misplaced. Cell 122, 317-318 (2005). 51. J. F. Caceres, G. R. Screaton, A. R. Krainer, A specific subset of SR

proteins shuttles continuously between the nucleus and the cytoplasm. Genes Dev 12, 55-66 (1998).

52. J. Yong, L. Wan, G. Dreyfuss, Why do cells need an assembly machine for RNA-protein complexes? Trends Cell Biol 14, 226-232 (2004).

53. T. H. Jensen, K. Dower, D. Libri, M. Rosbash, Early formation of mRNP: license for export or quality control? Mol Cell 11, 1129-1138 (2003).

54. L. Clancy, L. M. Beaulieu, K. Tanriverdi, J. E. Freedman, The role of RNA uptake in platelet heterogeneity. Thromb Haemost 117, 948-961 (2017).

55. M. G. Best et al., RNA-Seq of Tumor-Educated Platelets Enables Blood-Based Pan-Cancer, Multiclass, and Molecular Pathway Cancer Diagnostics. Cancer Cell, (2015).

56. D. C. Calverley et al., Significant downregulation of platelet gene expression in metastatic lung cancer. Clin Transl Sci 3, 227-232 (2010).

57. T. Junt et al., Dynamic visualization of thrombopoiesis within bone marrow. Science 317, 1767-1770 (2007).

58. R. M. Kaufman, R. Airo, S. Pollack, W. H. Crosby, R. Doberneck, Origin of Pulmonary Megakaryocytes. Blood 25, 767-775 (1965).

59. E. Lefrancais et al., The lung is a site of platelet biogenesis and a reservoir for haematopoietic progenitors. Nature, (2017).

60. H. Schwertz et al., Anucleate platelets generate progeny. Blood 115, 3801-3809 (2010).

61. K. Ravid, J. Lu, J. M. Zimmet, M. R. Jones, Roads to polyploidy: the megakaryocyte example. J Cell Physiol 190, 7-20 (2002).

62. T. T. Odell, Jr., C. W. Jackson, Polyploidy and maturation of rat megakaryocytes. Blood 32, 102-110 (1968).

63. J. Zimmet, K. Ravid, Polyploidy: occurrence in nature, mechanisms, and significance for the megakaryocyte-platelet system. Exp Hematol 28, 3-16 (2000).

40

64. S. Ebbe, Biology of megakaryocytes. Prog Hemost Thromb 3, 211-229 (1976).

65. J. E. Italiano, Jr., P. Lecine, R. A. Shivdasani, J. H. Hartwig, Blood platelets are assembled principally at the ends of proplatelet processes produced by differentiated megakaryocytes. J Cell Biol 147, 1299-1312 (1999).

66. L. A. Harker, The kinetics of platelet production and destruction in man. Clin Haematol 6, 671-693 (1977).

67. B. S. Coller, Foreword - A Brief History of Ideas About Platelets in Health and Disease. Platelets 3rd Edition, xix–xliv (2012).

68. B. S. Coller, S. J. Shattil, The GPIIb/IIIa (integrin alphaIIbbeta3) odyssey: a technology-driven saga of a receptor with twists, turns, and even a bend. Blood 112, 3011-3025 (2008).

69. A. S. Weyrich, S. Lindemann, G. A. Zimmerman, The evolving role of platelets in inflammation. J Thromb Haemost 1, 1897-1905 (2003).

70. C. Li et al., Crosstalk between Platelets and the Immune System: Old Systems with New Discoveries. Adv Hematol 2012, 384685 (2012).

71. Y. Wang et al., Platelets in thrombosis and hemostasis: old topic with new mechanisms. Cardiovasc Hematol Disord Drug Targets 12, 126-132 (2012).

72. M. Chatterjee et al., Distinct platelet packaging, release, and surface expression of proangiogenic and antiangiogenic factors on different platelet stimuli. Blood 117, 3907-3911 (2011).

73. J. E. Italiano, Jr. et al., Angiogenesis is regulated by a novel mechanism: pro- and antiangiogenic proteins are organized into separate platelet alpha granules and differentially released. Blood 111, 1227-1233 (2008).

74. I. Yougbare et al., Maternal anti-platelet beta3 integrins impair angiogenesis and cause intracranial hemorrhage. J Clin Invest 125, 1545-1556 (2015).

75. A. T. Franco, A. Corken, J. Ware, Platelets at the interface of thrombosis, inflammation, and cancer. Blood 126, 582-588 (2015).

76. M. Labelle, S. Begum, R. O. Hynes, Platelets guide the formation of early metastatic niches. Proc Natl Acad Sci U S A 111, E3053-3061 (2014).

77. M. Labelle, R. O. Hynes, The initial hours of metastasis: the importance of cooperative host-tumor cell interactions during hematogenous dissemination. Cancer Discov 2, 1091-1099 (2012).

78. M. Leslie, Cell biology. Beyond clotting: the powers of platelets. Science 328, 562-564 (2010).

79. G. L. Klement et al., Platelets actively sequester angiogenesis regulators. Blood 113, 2835-2842 (2009).

41

80. T. H. Lee et al., Distribution of HIV type 1 (HIV-1) in blood components: detection and significance of high levels of HIV-1 associated with platelets. Transfusion 38, 580-588 (1998).

81. T. H. Lee, R. R. Stromberg, D. Henrard, M. P. Busch, Effect of platelet-associated virus on assays of HIV-1 in plasma. Science 262, 1585-1586 (1993).

82. H. M. Verheul et al., Platelets take up the monoclonal antibody bevacizumab. Clin Cancer Res 13, 5341-5347 (2007).

83. J. G. White, Platelets are covercytes, not phagocytes: uptake of bacteria involves channels of the open canalicular system. Platelets 16, 121-131 (2005).

84. T. Youssefian, A. Drouin, J. M. Masse, J. Guichard, E. M. Cramer, Host defense role of platelets: engulfment of HIV and Staphylococcus aureus occurs in a specific subcellular compartment and is enhanced by platelet activation. Blood 99, 4021-4029 (2002).

85. O. Behnke, Electron microscopic observations on the membrane systems of the rat blood platelet. Anat Rec 158, 121-137 (1967).

86. L. M. Chapman et al., Platelets present antigen in the context of MHC class I. J Immunol 189, 916-923 (2012).

87. T. Placke et al., Platelet-derived MHC class I confers a pseudonormal phenotype to cancer cells that subverts the antitumor reactivity of natural killer immune cells. Cancer Res 72, 440-448 (2012).

88. S. E. Bailey, O. C. Ukoumunne, E. A. Shephard, W. Hamilton, Clinical relevance of thrombocytosis in primary care: a prospective cohort study of cancer incidence using English electronic medical records and cancer registry data. British Journal of General Practice, (2017).

89. P. Ferroni et al., Venous thromboembolism risk prediction in ambulatory cancer patients: clinical significance of neutrophil/lymphocyte ratio and platelet/lymphocyte ratio. Int J Cancer 136, 1234-1240 (2015).

90. R. L. Stone et al., Paraneoplastic thrombocytosis in ovarian cancer. N Engl J Med 366, 610-618 (2012).

91. I. Nikolic et al., Neutrophil-to-lymphocyte and platelet-to-lymphocyte ratio help identify patients with lung cancer, but do not differentiate between lung cancer subtypes. Croat Med J 57, 287-292 (2016).

92. P. M. Rothwell et al., Effect of daily aspirin on risk of cancer metastasis: a study of incident cancers during randomised controlled trials. Lancet 379, 1591-1601 (2012).

93. E. Heinmoller et al., Tumor cell-induced platelet aggregation in vitro by human pancreatic cancer cell lines. Scand J Gastroenterol 30, 1008-1016 (1995).

42

94. R. Wang et al., Electron cryotomography reveals ultrastructure alterations in platelets from patients with ovarian cancer. Proc Natl Acad Sci U S A 112, 14266-14271 (2015).

95. D. Stegner, S. Dutting, B. Nieswandt, Mechanistic explanation for platelet contribution to cancer metastasis. Thromb Res 133 Suppl 2, S149-157 (2014).

96. B. Tesfamariam, Involvement of platelets in tumor cell metastasis. Pharmacol Ther 157, 112-119 (2016).

97. D. Buergy, F. Wenz, C. Groden, M. A. Brockmann, Tumor-platelet interaction in solid tumors. Int J Cancer 130, 2747-2760 (2012).

98. G. J. Gasic, T. B. Gasic, C. C. Stewart, Antimetastatic effects associated with platelet reduction. Proc Natl Acad Sci U S A 61, 46-52 (1968).

99. Y. Jia et al., Activation of platelet protease-activated receptor-1 induces epithelial-mesenchymal transition and chemotaxis of colon cancer cell line SW620. Oncol Rep 33, 2681-2688 (2015).

100. D. Kong et al., Platelet-derived growth factor-D overexpression contributes to epithelial-mesenchymal transition of PC3 prostate cancer cells. Stem Cells 26, 1425-1435 (2008).

101. J. P. Thiery, H. Acloque, R. Y. Huang, M. A. Nieto, Epithelial-mesenchymal transitions in development and disease. Cell 139, 871-890 (2009).

102. M. Rolli, E. Fransvea, J. Pilch, A. Saven, B. Felding-Habermann, Activated integrin alphavbeta3 cooperates with metalloproteinase MMP-9 in regulating migration of metastatic breast cancer cells. Proc Natl Acad Sci U S A 100, 9482-9487 (2003).

103. Y. Q. Chen, B. Liu, D. G. Tang, K. V. Honn, Fatty acid modulation of tumor cell-platelet-vessel wall interaction. Cancer Metastasis Rev 11, 389-409 (1992).

104. V. A. Medina, E. S. Rivera, Histamine receptors and cancer pharmacology. Br J Pharmacol 161, 755-767 (2010).

105. G. Skolnik, U. Bagge, G. Blomqvist, L. Djarv, H. Ahlman, The role of calcium channels and serotonin (5-HT2) receptors for tumour cell lodgement in the liver. Clin Exp Metastasis 7, 169-174 (1989).

106. M. Chen, J. G. Geng, P-selectin mediates adhesion of leukocytes, platelets, and cancer cells in inflammation, thrombosis, and cancer growth and metastasis. Arch Immunol Ther Exp (Warsz) 54, 75-84 (2006).

107. D. G. Menter et al., Tumor cell-platelet interactions in vitro and their relationship to in vivo arrest of hematogenously circulating tumor cells. Clin Exp Metastasis 5, 65-78 (1987).

108. L. Borsig et al., Heparin and cancer revisited: mechanistic connections involving platelets, P-selectin, carcinoma mucins, and tumor metastasis. Proc Natl Acad Sci U S A 98, 3352-3357 (2001).

43

109. L. J. Gay, B. Felding-Habermann, Contribution of platelets to tumour metastasis. Nat Rev Cancer 11, 123-134 (2011).

110. J. S. Palumbo et al., Platelets and fibrin(ogen) increase metastatic potential by impeding natural killer cell-mediated elimination of tumor cells. Blood 105, 178-185 (2005).

111. H. G. Kopp, T. Placke, H. R. Salih, Platelet-derived transforming growth factor-beta down-regulates NKG2D thereby inhibiting natural killer cell antitumor reactivity. Cancer Res 69, 7775-7783 (2009).

112. M. Labelle, S. Begum, R. O. Hynes, Direct signaling between platelets and cancer cells induces an epithelial-mesenchymal-like transition and promotes metastasis. Cancer Cell 20, 576-590 (2011).

113. L. Erpenbeck, M. P. Schon, Deadly allies: the fatal interplay between platelets and metastasizing cancer cells. Blood 115, 3427-3436 (2010).

114. B. A. Kerr, N. P. McCabe, W. Feng, T. V. Byzova, Platelets govern pre-metastatic tumor communication to bone. Oncogene 32, 4319-4324 (2013).

115. A. B. Malik, Pulmonary microembolism. Physiol Rev 63, 1114-1207 (1983).

116. O. J. McCarty, S. A. Mousa, P. F. Bray, K. Konstantopoulos, Immobilized platelets support human colon carcinoma cell tethering, rolling, and firm adhesion under dynamic flow conditions. Blood 96, 1789-1797 (2000).

117. D. Schumacher, B. Strilic, K. K. Sivaraj, N. Wettschureck, S. Offermanns, Platelet-derived nucleotides promote tumor-cell transendothelial migration and metastasis via P2Y2 receptor. Cancer Cell 24, 130-137 (2013).

118. J. E. Peterson et al., Normal ranges of angiogenesis regulatory proteins in human platelets. Am J Hematol 85, 487-493 (2010).

119. A. Boucharaba et al., Platelet-derived lysophosphatidic acid supports the progression of osteolytic bone metastases in breast cancer. J Clin Invest 114, 1714-1725 (2004).

120. K. V. Honn, D. G. Tang, Y. Q. Chen, Platelets and cancer metastasis: more than an epiphenomenon. Semin Thromb Hemost 18, 392-415 (1992).

121. H. S. Kuznetsov et al., Identification of luminal breast cancers that establish a tumor-supportive macroenvironment defined by proangiogenic platelets and bone marrow-derived cells. Cancer Discov 2, 1150-1165 (2012).

122. H. M. Verheul et al., Vascular endothelial growth factor-stimulated endothelial cells promote adhesion and activation of platelets. Blood 96, 4216-4221 (2000).

44

123. A. Brill, O. Dashevsky, J. Rivo, Y. Gozal, D. Varon, Platelet-derived microparticles induce angiogenesis and stimulate post-ischemic revascularization. Cardiovasc Res 67, 30-38 (2005).

124. C. Chiodoni et al., Triggering CD40 on endothelial cells contributes to tumor growth. J Exp Med 203, 2441-2450 (2006).

125. B. Ho-Tin-Noe, T. Goerge, S. M. Cifuni, D. Duerschmied, D. D. Wagner, Platelet granule secretion continuously prevents intratumor hemorrhage. Cancer Res 68, 6851-6858 (2008).

126. J. Kisucka et al., Platelets and platelet adhesion support angiogenesis while preventing excessive hemorrhage. Proc Natl Acad Sci U S A 103, 855-860 (2006).

127. L. A. Torre et al., Global cancer statistics, 2012. CA Cancer J Clin 65, 87-108 (2015).

128. M. M. Center et al., International variation in prostate cancer incidence and mortality rates. Eur Urol 61, 1079-1092 (2012).

129. H. Lilja, J. Oldbring, G. Rannevik, C. B. Laurell, Seminal vesicle-secreted proteins and their reactions during gelation and liquefaction of human semen. J Clin Invest 80, 281-285 (1987).

130. M. K. Brawer, P. H. Lange, Prostate-specific antigen in management of prostatic carcinoma. Urology 33, 11-16 (1989).

131. S. Saini, PSA and beyond: alternative prostate cancer biomarkers. Cell Oncol (Dordr) 39, 97-106 (2016).

132. D. F. Gleason, G. T. Mellinger, Prediction of prognosis for prostatic adenocarcinoma by combined histological grading and clinical staging. J Urol 111, 58-64 (1974).

133. P. A. Humphrey, H. Moch, A. L. Cubilla, T. M. Ulbright, V. E. Reuter, The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs-Part B: Prostate and Bladder Tumours. Eur Urol 70, 106-119 (2016).

134. O. Andren et al., How well does the Gleason score predict prostate cancer death? A 20-year followup of a population based cohort in Sweden. J Urol 175, 1337-1340 (2006).

135. A. Heidenreich et al., EAU guidelines on prostate cancer. part 1: screening, diagnosis, and local treatment with curative intent-update 2013. Eur Urol 65, 124-137 (2014).

136. S. D. Fossa et al., Ten- and 15-yr Prostate Cancer-specific Mortality in Patients with Nonmetastatic Locally Advanced or Aggressive Intermediate Prostate Cancer, Randomized to Lifelong Endocrine Treatment Alone or Combined with Radiotherapy: Final Results of The Scandinavian Prostate Cancer Group-7. Eur Urol 70, 684-691 (2016).

137. S. W. Hayward, G. R. Cunha, The prostate: development and physiology. Radiol Clin North Am 38, 1-14 (2000).

45

138. A. A. Shafi, A. E. Yen, N. L. Weigel, Androgen receptors in hormone-dependent and castration-resistant prostate cancer. Pharmacol Ther 140, 223-238 (2013).

139. D. S. Coffey, J. T. Isaacs, Control of prostate growth. Urology 17, 17-24 (1981).

140. M. Krieg, H. Weisser, S. Tunn, Potential activities of androgen metabolizing enzymes in human prostate. J Steroid Biochem Mol Biol 53, 395-400 (1995).

141. W. D. Steers, 5alpha-reductase activity in the prostate. Urology 58, 17-24; discussion 24 (2001).

142. S. M. Dehm, D. J. Tindall, Molecular regulation of androgen action in prostate cancer. J Cell Biochem 99, 333-344 (2006).

143. N. Sharifi, J. L. Gulley, W. L. Dahut, Androgen deprivation therapy for prostate cancer. JAMA 294, 238-244 (2005).

144. C. D. Chen et al., Molecular determinants of resistance to antiandrogen therapy. Nat Med 10, 33-39 (2004).

145. B. J. Feldman, D. Feldman, The development of androgen-independent prostate cancer. Nat Rev Cancer 1, 34-45 (2001).

146. A. Katzenwadel, P. Wolf, Androgen deprivation of prostate cancer: Leading to a therapeutic dead end. Cancer Lett 367, 12-17 (2015).

147. R. B. Montgomery et al., Maintenance of intratumoral androgens in metastatic prostate cancer: a mechanism for castration-resistant tumor growth. Cancer Res 68, 4447-4454 (2008).

148. J. S. de Bono et al., Abiraterone and increased survival in metastatic prostate cancer. N Engl J Med 364, 1995-2005 (2011).

149. J. S. de Bono et al., Prednisone plus cabazitaxel or mitoxantrone for metastatic castration-resistant prostate cancer progressing after docetaxel treatment: a randomised open-label trial. Lancet 376, 1147-1154 (2010).

150. P. L. Kellokumpu-Lehtinen et al., 2-Weekly versus 3-weekly docetaxel to treat castration-resistant advanced prostate cancer: a randomised, phase 3 trial. Lancet Oncol 14, 117-124 (2013).

151. C. Parker et al., Alpha emitter radium-223 and survival in metastatic prostate cancer. N Engl J Med 369, 213-223 (2013).

152. H. I. Scher et al., Increased survival with enzalutamide in prostate cancer after chemotherapy. N Engl J Med 367, 1187-1197 (2012).

153. I. F. Tannock et al., Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer. N Engl J Med 351, 1502-1512 (2004).

154. R. Chou et al., Screening for prostate cancer: a review of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med 155, 762-771 (2011).

46

155. U. S. P. S. T. Force. Draft recommendation statement: prostate cancer: Screening. (2017).

156. A. Dobin et al., STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15-21 (2013).

157. A. Corduan et al., Dissociation of SERPINE1 mRNA from the translational repressor proteins Ago2 and TIA-1 upon platelet activation. Thromb Haemost 113, 1046-1059 (2015).

158. S. Jahr et al., DNA fragments in the blood plasma of cancer patients: quantitations and evidence for their origin from apoptotic and necrotic cells. Cancer Res 61, 1659-1665 (2001).