Biochimica et Biophysica Acta 1844 (2014) 988–1002

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbapap

Review

Proximal fluid proteomics for the discovery of digestive cancer biomarkers☆ Annarita Farina ⁎ Biomedical Proteomics Research Group, Department of Human Protein Sciences, Geneva University, Geneva CH-1211, Switzerland

a r t i c l e

i n f o

Article history: Received 11 July 2013 Received in revised form 15 September 2013 Accepted 22 October 2013 Available online 1 November 2013 Keywords: Bile Pancreatic juice Cyst fluid Gastric juice Saliva Ascites

a b s t r a c t Most digestive malignancies have asymptomatic course, often progressing to poor outcome stages. Surgical resection usually represents the only potentially curative option but a prior assumption of the malignant nature of the lesion is mandatory to avoid exposing patients to unnecessary risks. Unfortunately, currently available diagnostic tools lack accuracy in many cases, consequently more reliable markers are needed to improve detection of malignant lesions. In this challenging context, fluids surrounding digestive malignancies represent a valuable source for the search of new potential biomarkers and proteomic tools offer the opportunity to achieve this goal. The new field of proximal fluid proteomics is thus emerging in the arena of digestive cancer biomarker discovery. In the present review, the state-of-the-art of proteomic investigations aimed at identifying new cancer biomarkers in fluids surrounding gastrointestinal malignancies is summarized. A comprehensive catalog of proteomic studies in which potential cancer biomarkers from gastrointestinal fluids have been identified and assessed for their diagnostic performances is also provided. This article is part of a Special Issue entitled: Biomarkers: A Proteomic Challenge. © 2013 Elsevier B.V. All rights reserved.

1. Introduction: Unmet clinical needs in the diagnosis of digestive malignancies Digestive malignancies refer to a heterogeneous group of cancers affecting he gastrointestinal tract and associated organs. Among them, liver, stomach, colon-rectus, esophagus and pancreas cancers represent five of the ten leading causes of cancer death, accounting for nearly 3 million of all estimated deaths per year in the world [1]. The reason for such a high mortality is mainly related with the common asymptomatic course of these malignancies, which often results in late detection, at an advanced stage of the disease [2]. Surgical resection usually represents the only potentially curative option but it requires a reliable assumption of the malignant nature of the lesion because oncological digestive surgery is associated with an important morbidity and mortality [3]. Unfortunately, gastroenterologists and digestive surgeons are frequently challenged by clinical situations where the lesion is difficult to detect or to diagnose. Currently available diagnostic tools (e.g., imaging, serum biomarkers) lack accuracy in many cases to differentiate between nonmalignant and malignant lesions. Resulting diagnostic uncertainties expose patients to potential harmful risks. In the next sections, I report on the clinical details of digestive cancers for which proteomic studies of proximal fluids have been reported. The

☆ This article is part of a Special Issue entitled: Biomarkers: A Proteomic Challenge. ⁎ Tel.: +41 22 379 5451; fax: +41 22 379 5502. E-mail address: [email protected]. 1570-9639/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbapap.2013.10.011

limitations of differential diagnosis between malignant and nonmalignant diseases are also discussed.

1.1. Liver cancer Liver cancer represents the fifth most frequently diagnosed cancer and the second most frequent cause of cancer death in men. The highest number of cases is diagnosed in Asia, with China accounting for 50% of all estimated cases [1]. A significant increase in liver cancer incidence and death rates was observed in USA from 2005 to 2009 and an overall 5-year survival rate of 15% has been reported. Early stage diagnosis occurs in only 40% of cases and is associated with a 28% 5-year survival rate [2]. Hepatocellular carcinoma (HCC) mostly arises from liver cirrhosis [4] and accounts for 70% to 85% of the total primary liver cancers [1]. HCC is clinically difficult to distinguish from other hepatic benign masses (e.g., focal nodular hyperplasia, hepatocellular adenoma, regenerative nodule and hemangioma) with which it shares common histologic features [5]. Pre-neoplastic hepatocellular lesions in cirrhotic patients (e.g., dysplastic nodules mimicking small HCC) may also complicate the differential diagnosis [6]. In a recent study involving 638 liver transplanted patients with cirrhosis, ultrasonography (US), computed tomography (CT) and magnetic resonance imaging (MRI) disclosed a sensitivity of 46%, 65%, and 72%, respectively, for the detection of HCCs ranging between 2 cm and 4 cm, and 21%, 40%, and 47%, respectively, for smaller lesions. The current gold standard serum biomarker for HCC, alpha-fetoprotein (AFP), also showed an inadequate sensitivity in detecting HCC (53% at 10 ng/dL cutoff level) [7]. Other existing serum

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

biomarkers (e.g. lens culinaris agglutinin-reactive AFP (AFP-L3), des-γcarboxy prothrombin (DCP)) display suboptimal sensitivity when assessed alone [8]. Their contribution to the current standard AFP is currently being evaluated and their clinical use is not yet recommended [9,10]. 1.2. Pancreas cancer In the USA, pancreas cancer accounts for 44,000 new cases each year and has become the fourth leading cause of cancer mortality with 37,000 new estimated deaths in 2013 [2]. From 2004 to 2008, the incidence and death rates of pancreatic cancer have been increasing by 1.5% and 0.4% per year, respectively [11]. The 5-year survival rate is 6% for all diagnosed patients and 2% for patients diagnosed at an advanced stage [11]. Infiltrating ductal adenocarcinoma (or pancreatic adenocarcinoma (PAC)) represents about 90% of all pancreatic tumors [12]. Among imaging techniques, multi-detector computed tomography (MDCT) and endoscopic ultrasonography (EUS) are widely accepted as the methods of choice for diagnosing and staging pancreatic cancer [13]. A recent retrospective study, conducted on 117 patients, revealed MDCT to have limited performances for detecting pancreatic tumors (93% sensitivity, 72% specificity, 95% PPV, 65% NPV and 90% accuracy). The sensitivity, specificity and diagnostic accuracy of EUS–FNA for differentiating nonmalignant vs. malignant pancreatic solid masses is 83%, 100% and 88%. In patients with indeterminate or negative findings at initial EUS–FNA and a high clinical suspicion for pancreatic cancer, repetition of EUS–FNA is strongly advised [14]. However, these figures have been reported by dedicated endosonographers and may be significantly lower in the community [15]. No specific biomarkers exist for pancreatic cancer and the gold standard serum carbohydrate antigen 19-9 (CA 19-9) has a low specificity and limited clinical utility to differentiate nonmalignant from malignant masses [16,17]. 1.2.1. Pancreatic cystic neoplasms Pancreatic cysts represent approximately 10%–15% of the primary masses of the pancreas and consist in a heterogeneous group of malignant and nonmalignant lesions sharing many common clinical features [18,19]. Although cystic tumors of the pancreas are generally uncommon (2% of all pancreatic tumors), the prevalence of pancreatic cystic lesions is rising due to improved detection related to the increased use of cross-sectional imaging. Recent studies estimated the prevalence of cystic lesions between 2.6% and 44.7% [20]. The differential diagnosis between malignant, premalignant and nonmalignant pancreatic cysts, as well as the identification of the histological type, is a complex and highly relevant clinical problem. Actually, the accuracy of CT and MRI in making a correct diagnosis ranges between 40% and 60%. EUS also shows limited performances to distinguish between cystic tumors (56% sensitivity, 48% specificity) [20]. Finally, fluid cytology shows a specificity of about 100% but a much lower sensitivity for identifying malignancies [20]. 1.3. Malignant biliary stenosis Malignant biliary stenosis may arise due to several of the abovementioned malignancies, as a consequence of the extrinsic compression of intra- or extra-hepatic bile ducts by a tumor affecting an adjacent organ (e.g., pancreas, liver, gallbladder), or it may be caused by an intrinsic tumor of the bile duct (cholangiocarcinoma (CC)) or of the ampulla (ampullary carcinoma (AC)). Pancreatic head adenocarcinoma represents the most common cause of malignant biliary stenosis, followed by biliary tract tumors: gallbladder cancer (GBC) and CC. GBC usually arises in the fundus or neck of the gallbladder [21]. It is a rare malignancy but represents, by itself, 80–95% of all biliary cancers [22]. CC, instead, arises from the biliary epithelium mainly at the bifurcation of the hepatic ducts (60–70%). Neoplasms in the distal common

989

bile duct or peripheral intrahepatic ducts are also possible although less frequent (20%–30% and 5%–10%, respectively) [23]. Both CC and GBC have a very poor prognosis [21,24]. Adenocarcinoma of the ampulla of Vater is the third cause of malignant biliary stenosis. For this malignancy, obstructive jaundice occurs relatively early during the course of the disease and its evolution is slower. Consequently, AC presents a relatively better prognosis, with a 5-year survival rate ranging from 33% to 68%, compared to 6% for PAC and 5% for CC [11,21,24]. Finally, bile duct compression by liver cancer is also possible [25]. Besides malignant causes of bile duct stenosis, nonmalignant obstructions can also develop following inflammation associated with bile ducts injuries (e.g., surgery, trauma, pancreatitis, bile duct stones, primary sclerosing cholangitis (PSC)) [26] and represent about 25% of all biliary stenoses [27]. Differentiating malignant vs. nonmalignant biliary stenoses is clinical challenging. All currently available techniques, including cross-sectional imaging (e.g., EUS, CT, MRCP), endoscopic retrograde cholangiopancreatography (ERCP) and biliary brush or grasp cytology, show suboptimal accuracy. As a consequence, uncertainty about the nature of the stenosis persists in up to 50% examinations [14,28–31]. Furthermore, standard serum biomarkers are, on the one hand, generally unreliable to detect bilio-pancreatic malignancies and, on the other hand, significantly altered by the presence of a biliary obstruction (e.g., CA 19-9) [32]. 1.4. Stomach cancer Stomach cancer accounts for 8% of new cancer cases and 10% of cancer deaths in the world; it represents the third most frequent cause of cancer death in men. Over 70% of new cases occur in developing countries [1]. The incidence of stomach cancer has declined in most parts of the world due to a decreasing prevalence of Helicobacter pylori infection as well as to improved diet, hygiene and food storage practices [1,11]. Nevertheless, this malignancy carries a poor prognosis with an overall 5-year survival rate of 26% (62% in the case of early diagnosis). Accuracy of many common screening tools (including barium-meal X-ray, US, CT and MRI) has shown to be inadequate for detecting both, advanced and early-stage gastric cancers [33]. The differential diagnosis between malignant and nonmalignant gastric ulcers may also be difficult due to common morphological characteristics. Virtual gastroscopy (VG) and ES actually offer the best performances in distinguishing malignant from nonmalignant gastric ulcers, with an almost similar accuracy (92.0% and 88.6%, respectively) [34,35]. Finally, a recent screening conducted on a total of 13,118 participants in Portugal revealed that serum pepsinogen (PG) shows an insufficient performance to detect gastric cancer, with an estimated sensitivity, specificity, positive predictive value, and negative predictive value of 67%, 47%, 2% and 99%, respectively [36]. 1.5. Peritoneal cancer The inner membrane lining the abdominal cavity (peritoneum) can be affected by primary or secondary neoplasms. Primary peritoneal neoplasms are rare and usually arise from the mesothelial cell layer. They include malignant peritoneal mesothelioma, well-differentiated papillary mesothelioma, multicystic mesothelioma, desmoplastic small round cell tumor and peritoneal serous carcinoma [37]. Secondary peritoneal neoplasia is much more common and can occur by direct invasion from contiguous organs or through the seeding of cancer cells via the intraperitoneal fluid [37]. Metastases of the peritoneum are frequent in the presence of advanced ovarian and gastrointestinal (e.g., colorectal, pancreatic, gastric) tumors and represent the leading cause of death in most cases. Peritoneal carcinomatosis has long been considered virtually incurable, with an average life expectancy of 6 months, and chemotherapy protocols have proven to increase median survival by only 4 to 14 months [38,39]. The combination of cytoreductive surgery with hyperthermic intraperitoneal chemotherapy

990

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

symptoms during initial stages, the lack of routine oral examination, the presence of lesions in parts of the oral cavity which normally escape inspection (e.g., floor of the mouth), the lack of widely accepted screening tests which can be easily performed, and the similarity between malignant and nonmalignant or premalignant lesions (e.g., candidiasis, recurrent aphthous stomatitis, fibroma, leukoplakia) [47]. The current gold standard for oral cancer diagnosis is scalpel biopsy, an invasive procedure associated with pain and potential morbidity, whose accuracy is strongly related to sampling procedure, practitioner ability and technical limitations [48]. 2. Pitfalls of cancer biomarkers search in serum and plasma

Fig. 1. Studies aimed at identifying cancer biomarkers in proximal fluids. A) Total number of publications per year since 1957; B) Rise in the number of proteomic and nonproteomic publications in the last 10 years.

(CS/HIPEC) recently showed the potential to offer significant survival prolongation in selected patients, at the price of a considerable risk of morbidity and mortality [39–41]. The diagnosis of peritoneal cancer is strongly hampered by the lack of specific serum biomarkers and the low sensitivity of standard imaging (e.g., ultrasonography, helical CT) to detect peritoneal masses [38]. The pathological evaluation of intraperitoneal ascitic fluid, if present, is also not sensitive enough to detect malignancies. In particular, cytopathological examination of ascitic fluid performed in the most adequate conditions shows a sensitivity for the detection of cancer of only 60% [42,43]. In most cases, a tissue biopsy is needed to confirm the presence of peritoneal metastases. 1.6. Oral cancer Cancers of oral cavity and pharynx accounts for about 41,000 new cases per year in the USA and represent the eight leading cause of estimated new cancers in men [2]. In some developing countries (e.g., Sri Lanka, India, Pakistan, Bangladesh) oral cancers are the most common cancers in men, and may contribute up to 25% of all new cases of cancer; France shows the highest incidence within the European Union (5.5%) [44]. Oral squamous cell carcinoma (OSCC), which arises from the oral mucosa lining, accounts for over 90% of all malignant lesions of oral cavity [45,46]. From 2005 to 2009, the overall incidence of oral cancers was reported to be stable for men and in slow decline for women; nevertheless, cancers of the oropharynx specifically associated with human papillomavirus (HPV) infection appear to be increasing. The 5-year survival rates for oral cancers are 62% for all diagnosed stages and 35% for advanced-stage cancers [2]. Early detection of oral cancers is hampered by several factor causing delays in the diagnosis, such as the absence of

Blood fractions (serum or plasma) represented for years the samples of choice for the search of cancer biomarkers. The reason mainly lies in two aspects: i) blood vessels come into direct contact with all body tissues, thus being potentially able to collect molecules released by tumor cells; ii) blood collection is rapid, easy and minimally invasive, which makes it a perfect sample for both, searching investigations and diagnostic measurements. Unfortunately, the use of serum and plasma for protein biomarker search also involves a number of inconveniences. First, the systemic circulation collects not only proteins released by tumor cells but also those produced by other organs, in response to cancer or any other pathophysiological state (e.g. inflammatory proteins) [49]. Second, the search for potential cancer biomarkers into the bloodstream is hampered by the extreme dilution of molecules arising from the tumor. An ideal marker, in fact, is solely produced at the cancer site, thus being transferred into the bloodstream in very low amounts [50]. Third, the protein composition of serum and plasma is extremely complex and distributed according to a very large dynamic range of concentrations that exceeds 10 orders of magnitude [51]. The ten most abundant proteins (namely: albumin, IgG, transferrin, fibrinogen, complement C3, alpha-1-antitrypsin, IgA, alpha-2-macroglobulin, IgM and haptoglobin) account for about 90% of the total serum protein load and show concentrations in the mg/mL range, while the thousands of remaining proteins are comprised between several μg/mL and a few pg/mL. In this context, proteins released by tissues, including tumor markers currently used in clinical practice (e.g., alpha-fetoprotein, AFP; carcinoembryonic antigen, CEA; gastrin, GAST; chromogranin A, CMGA; calcitonin, CALC), mostly fall in the ng to low pg/mL range [52]. Proteomic analysis, which currently represents one of the most effective approaches to investigate proteomes for identifying proteins differentially expressed in pathological conditions [53], is particularly affected by these issues [54,55]. Despite the enormous technological improvements made in the field, indeed, mass spectrometers available for quantitative comparison of biological samples still show relatively inadequate limits of detection (LOD) and quantification (LOQ). According to Surinova et al. [56], the most sensitive mass spectrometry (MS)based method currently available for protein quantification is the selected reaction monitoring (SRM) performed on a triple quadrupole instrument. Nevertheless, without any prior sample treatment, this method only allows the identification of serum proteins in the μg/mL range, thus showing an inadequate threshold for biomarker detection and quantification. A limit of 25 ng/mL can be reached by depleting abundant proteins and additional protein fractionations can further improve the LOQ to as low as 2.5 ng/mL [56]. Whatever the case, proteins at very low concentration in human serum actually remain excluded from quantification. New hybrid instruments, which combine sensitive linear ion trap with high-resolution accuratemass technology (e.g., LTQ-Orbitrap) succeed in adding only minor identifications to the plasma proteome [57]. A fourth weak point of the search for biomarkers in serum and plasma is the assumed correlation between local release of cancerassociated molecules and their measurable concentration in the blood. In actual fact, this hypothesis underestimates a number of circumstances. The specific mechanisms by which tumor proteins are transported into

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

991

Table 1 Proteomic studies aimed at identifying cancer biomarkers of malignant hepatopancreatobiliary lesions in bilea. Reference

[75] [76]

[78]

[79]

Author (year)

Kristiansen et al. (2004)

Chen et al. (2008)

Zabron et al. (2011)

malignant (n=1)

malignant (n=4)

Koopmann et al. (2004) Sample type (number)

malignant (n=1) cholangiocarcinoma (n=1)

pancreatic cancer (n=4)

cholangiocarcinoma (n=9) [pooled] nonmalignant (n=1)

nonmalignant (n=4)

Proteomic analysis

gallstones (n=9) [pooled] Workflowb

centrifugation at 16,000xg, lipid removal, ultrafiltration (3 kDa) and: i) SDS—

centrifugation at 5,000xg, TCA/acetone albumin

PAGE; ii) Lectin affinity chromatography, SDS—PAGE; iii) Lectin affinity

removal, methanol/chloroform delipidation

benign (n=4) centrifugation at 16,000xg, SDS—PAGE, label free analysis

chromatography, immunoglobulin depletion, SDS—PAGE Mass spectrometer

Micromass Q—TOF API—US

LCQ DECA XP Plus

LTQ linear ion trap

Number of identified proteins

87



>200

Number of differentially





10

galectin—3—binding protein (Mac—2BP)

chymotrypsin—like elastase family

neutrophil gelatinase—associated lipocalin (NGAL)

expressed proteins in malignant conditions Candidate Biomarker

c

member 3B (CEL3B) Sample type (number)

malignant (n=26)

malignant (n=22)

nonmalignant (n=52)

cohort A malignant (n=16)

cholangiocarcinoma (n=22)

biliary tract carcinoma (n=26)

pancreatic adenocarcinoma (n=8), metastatic cancer (n=4),

nonmalignant (n=28)

ampullary carcinoma (n=2), cholangiocarcinoma (n=2)

gallstones (n=28)

benign biliary conditions (n=32)

nonmalignant (n=22)

primary sclerosing cholangitis (n=20)

stone (n=11), chronic pancreatitis (n=5), stricture (n=3), sphinter of Oddy dysfunction (n=2), pancreas divisum (n=1) cohort B Verification analysis

malignant (n=7) pancreatic adenocarcinoma (n=5), gallbladder cancer (n=1), hepatocellular carcinoma (n=1) nonmalignant (n=14) stone (n=6), chronic pancreatitis (n=3), primary sclerosing cholangitis (n=3), leak (n=2) cohort C (sub—cohort of A) malignant (n=13) nonmalignant (n=17) cohort D malignant (n=13) pancreatic cancer (n=13) nonmalignant (n=36) stone (n=17), chronic pancreatitis (n=8), stricture (n=3), sphinter of Oddy dysfunction (n=2), pancreas divisum (n=1), primary sclerosing cholangitis (n=3), leak (n=2) Method

ELISA (equal volume—inferred)

activity assay (equal volume)

ELISA (equal volume—inferred)

Bile — single marker

Mac—2BP



NGAL

malignant vs nonmalignant

malignant vs nonmalignant

0.70 ROC—AUC; 0.69 SE, 0.67 SP [cut—off 853 ng/mL]

cohort A 0.76 ROC—AUC; 0.94 SE, 0.55 SP, 0.60 PPV, 0.92 NPV [cut—off 570 ng/mL] cohort B 1 SE, 0.55 SP [cut—off 570 ng/mL] cohort C 0.74 ROC—AUC; 0.92 SE, 0.47 SP [cut—off 570 ng/mL] cohort D

Diagnostic performances

1 SE, 0.55 SP [cut—off 570 ng/mL]

pancreatic cancer vs nonmalignant cohort D 1 SE, 0.56 SP [cut—off 570 ng/mL] Serum/Plasma — single marker

no significant difference



Gold standard/reference marker

bile CA19—9

serum CEA

malignant vs nonmalignant

malignant vs nonmalignant

0.69 ROC—AUC; 0.65 SE, 0.65 SP [cut—off 36,550 U/mL)

no significant difference —

0.503 ROC—AUC serum CA19—9 malignant vs nonmalignant 0.547 ROC—AUC

Panel of markers

bile Mac—2BP & bile CA19—9

bile CEL3B/ bile amylase ratio

bile NGAL & serum CA19—9

malignant vs nonmalignant

malignant vs nonmalignant

malignant vs nonmalignant

0.75 ROC—AUC

0.877 ROC—AUC; 0.818 SE, 0.893 SP [cut—off 0.065]

cohort C 0.85 SE, 0.82 SP, 0.79 PPV, 0.87 NPV [cut—off 3,000 ng/mL and 125 U/L]

Abbreviations: SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TCA, Trichloroacetic acid; 2-DE, bidimensional electrophoresis; iTRAQ, isobaric tags for relative and absolute quantitation; ROC-AUC, area under the curve on the receiver operator characteristic; SE, sensitivity; SP, specificity; PPV, positive predictive value; NPV, negative predictive value. a Only proteomic studies for which information about the diagnostic potential of identified candidate markers was available are reported. b Classical proteomic steps, such as denaturation, reduction, alkylation and digestion are not indicated. c For additional information about candidate biomarkers, see Supplementary Table 1.

blood, for instance, could lead to discrepancies between the number of molecules locally released from cancer cells and the amount transferred into the bloodstream. Furthermore, even when present in blood, biomarkers could be inaccessible for direct measurement due to: i) potential sequestration into plasma membrane-derived particles; ii) depletion or digestion. Recent evidences support the option of tumor proteins translocation through circulating microvesicles (e.g. oncosomes). Cancer cells emit large quantities of these cargoes and

blood from cancer patients has proven to include increased amounts of microvesicles compared to healthy controls [58,59]. The protein content of oncosomes (e.g., cell-surface receptors, intracellular signaling molecules) therefore, although representing a great opportunity for the detection of new cancer biomarkers, could not be directly accessible for blood screening as a consequence of the vesicular “shield”. A selective and irregular transport is also influencing the concentration of oncosome-associated proteins, which strongly depend on the stimuli

992

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

[81]

[82]

[83]

Lankisch et al. (2011)

Shen et al. (2012)

Farina et al. (2013)

malignant (n=16)

malignant (n=3)

malignant (n=2)

cholangiocarcinoma (n=16) nonmalignant (n=34)

cholangiocarcinoma (n=15) [pooled]

pancreatic cancer (n=1)

nonmalignant (n=3)

choledocolithiasis (n=16)

cholangitis (n=10) [pooled]

cholangiocarcinoma (n=1) nonmalignant (n=2)

primary sclerosing cholangitis (n=18)

chronic pancreatitis (n=1) biliary stones (n=1)

n—butanol/iso—propyl ether precipitation, urea extraction,

Enrichment according to Kristiansen et al.,

centrifugation at 16,000xg, lipid removal, ultrafiltration (3 kDa), iTRAQ labeling, OFFGEL

ultrafiltration (10 kDA), Sephadex—based desalting

immunodepletion of 14 most abundant proteins, 2—DE

electrophoresis

Micro—TOF

Ultraflex II MALDI—TOF/TOF

LTQ Orbitrap XL





200

38 overexpressed

66 overexpressed

spermatogenesis—associated protein 20 (SSP411)

carcinoembryonic antigen—related cell adhesion molecule 6 (CEAM6)

malignant (n=30)

malignant (n=29)

cholangiocarcinoma+cholangitis vs choledocolithiasis 83 peptides cholangiocarcinoma vs cholangitis 90 peptides cholangiocarcinoma+cholangitis vs choledocolithiasis 18 peptide marker candidates cholangiocarcinoma vs cholangitis 22 peptide marker candidate malignant cholangiocarcinoma (n=25) nonmalignant choledocolithiasis (n=14)

cholangiocarcinoma (n=30)

pancreatic cancer (n=23)

hepatocellular carcinoma (n=24)

cholangiocarcinoma (n=4) ampullary adenocarcinoma (n=2)

nonmalignant (n=36) cholangitis (n=13)

primary sclerosing cholangitis (n=18)

nonmalignant (n=12)

healthy (n=23)

chronic pancreatitis (n=8)

liver cirrhosis (n=10)

biliary stones (n=2) other benign (n=2)

statistical model

ELISA (equal volume)

ELISA (equal volume)

selected peptides

CEAM6

cholangiocarcinoma+cholangitis vs choledocolithiasis

malignant vs benign

0.93 ROC—AUC; 0.93 SE, 0.86 SP

0.92 ROC—AUC; 0.93 SE, 0.83 SP, 0.93 PPV, 0.83 NPV [cut—off 67.9 ng/mL)

cholangiocarcinoma vs cholangitis 0.87 ROC—AUC; 0.84 SE, 0.78 SP —

SSP411

CEAM6

cholangiocarcinoma vs cholangitis+healthy

malignant vs benign

0.913 ROC—AUC; 0.90 SE, 0.833 SP [cut—off 0.63]

0.57 ROC—AUC

cholangiocarcinoma vs cholangitis 0.836 ROC—AUC; 0.857 SE, 0.769 SP [cut—off 0.65] hepatocellular carcinoma vs liver cirrhosis 0.735 ROC—AUC; 0.777 SE, 0.60 SP [cut—off 0.43] hepatocellular carcinoma vs liver cirrhosis+healthy 0.709 ROC—AUC; 0.417 SE, 0.848 SP [cut—off 0.63] cholangitis vs healthy 0.582 ROC—AUC; 0.317 SE, 0.870 SP [cut—off 0.63] liver cirrhosis vs healthy 0.552 ROC—AUC; 0.50 SE, 0.696 SP [cut—off 0.39] —









bile CEAM6 & serum CA 19—9 malignant vs benign 0.96 ROC—AUC; 0.97 SE, 0.83 SP, 0.93 PPV, 0.91 NPV [cut—off, 67.9 ng/mL and 157 kU/L] 0.95 ROC—AUC; 0.90 SE, 1 SP, 1 PPV, 0.80 NPV [cut—off, 104.7 ng/mL and 157 kU/L]

received by cancer cells [59]. Finally, the plasma clearance and the action of circulating proteases, such as the proteolytic cascade induced by thrombin in serum, could alter the concentration of cancerreleased molecules or vesicles in the blood [60,61]. 3. Analysis of body fluids surrounding digestive malignancies: advantages, drawbacks and technical feasibility The search for biomarkers in fluids gathering in direct vicinity of tumors has become increasingly popular in recent years [62,63]. Fig. 1 shows the rise in the number of scientific publications aimed at identifying new biomarkers in proximal fluids. Studies exploiting proteomic approaches are also shown, which have recorded a 17-fold increase in the last 10 years, compared to a more modest 2.6-fold increase of non-proteomic investigations in the field. The reason for such a growing interest lies mainly in the fact that the identification of potential biomarkers is hugely favored in fluids surrounding malignancies. Proteins secreted or shed by cancer cells, indeed, are directly released into the bounded fluidic environment, gaining from an accumulation effect which leads to a significant concentration increase [63]. In addition, the correlation between pathological expression and fluid concentration is not subordinate to vessel uptake mechanisms so that the measurement of potential cancer biomarkers is not liable for

transport-induced alteration. The analysis of a local fluid environment also reduces typical biological hitches of circulating body fluids (e.g., plasma and serum), such as the identification of proteins released from non-primary organs and the disappearance of potentially interesting molecules due to plasma clearance [49]. Compared to serum and plasma, therefore, proximal fluids investigation could benefit biomarker discovery in many ways. Nonetheless, the proteomic analysis of such a kind of samples is not free of complications. Non-circulating fluids, in fact, may show huge complexity and high dynamic range of concentrations just like circulating ones [49]. This is mainly the case of plasma-leakage body fluids (e.g., ascites) [64], but not only. Amado et al. recently estimated the abundance of salivary proteins to span as many as 14 orders of magnitude [65]. High-abundant proteins could therefore be present in proximal fluids, with the consequence of preventing the detection of less-abundant ones. Additionally, these fluids may specifically contain high concentration of locally released substances that may interfere with proteomic analysis. A striking example is the bile fluid, whose organic content (e.g., lipids, bilirubin, bile salts) is so disproportionate as to prevent the direct analysis of crude samples [66]. Fluids surrounding gastrointestinal epithelia may also contain elevated levels of highly glycosylated proteins (e.g. mucins), which increase the viscosity of the sample and hamper the identification of disguised (i.e., glycosylated) peptides. Finally, the presence

Proteomic analysis

Table 2 Proteomic studies aimed at identifying cancer biomarkers of malignant pancreatic lesions in pancreatic juicea. Reference

[98]

[100]

[103]

Author (year)

Shirai et al. (2008)

Chen et al. (2010)

Park et al. (2011)

Makawita et al. (2011)

Sample type (number)

malignant (n=54) pancreatic ductal carcinoma (n=54) premalignant (n=33) intraductal papillary mucinous neoplasm (n=33)

malignant (n=3)

malignant (n=1)

malignant (n=2)

pancreatic intraepithelial neoplasia grade 3 (n=3) nonmalignant (n=1)

pancreatic cancer (n=3) [pooled]

pancreatic cancer (n=6) [pooled]

nonmalignant (n=2)

nonmalignant (n=5) [pooled]

nonmalignant (n=31) chronic pancreatitis (n=31)

[104]

chronic pancreatitis (n=3) [pooled] normal (n=3) [pooled]

Workflowb

centrifugation at 10,000 rpm

Mass spectrometer

SELDI–TOF

Q–STAR pulsar II

MALDI–TOF/TOF

LTQ–Orbitrap XL

Number of identified proteins







521 (average)

Number of differentially expressed proteins in malignant conditions

5 peaks

20 overexpressed

26 overexpressed or specifically detected in pancreatic cancer



Candidate Biomarkerc

pancreatic secretory trypsin inhibitor (PSTI)

anterior gradient 2 (AGR2)

lithostathine–1–alpha (REG1α α)

anterior gradient 2 (AGR2) syncollin (SYNC)

iTRAQ labeling

TCA/acetone precipitation, 2–DE

centrifugation at 16,000 rpm, dialysis (3.5 kDa), Sephadex– based desalting, strong cation exchange chromatography

olfactomedin–4 (OLFM4) collagen alpha–1(VI) chain (COL6A1) polymeric immunoglobulin receptor (PIGR) –

premalignant (n=25) pancreatic intraepithelial neoplasia grade 2 (n=6) pancreatic intraepithelial neoplasia grade 3 (n=9) intraductal papillary mucinous neoplasm (n=10) malignant (n=8) pancreatic cancer stage 2 (n=1) pancreatic cancer stage 3 (n=3) pancreatic cancer stage 4 (n=2) pancreatic cancer stage undetermined (n=2) nonmalignant (n=18) sphincter of Oddi dysfunction (n=7) chronic pancreatitis (n=11)

malignant (n=23)

malignant (n=20) pancreatic cancer (n=20)

pancreatic cancer (n=23) nonmalignant (n=17)

nonmalignant (n=20)

chronic pancreatitis (n=8)

normal (n=20)

normal (n=9)

Method

Radioimmunoassay (equal volume–inferred)

ELISA (equal volume)

ELISA (equal volume)

ELISA (equal volume)

Pancreatic juice – single marker

PSTI premalignant vs malignant+nonmalignant 0.48 SE, 0.98 SP, 0.89 PPV, 0.83 NPV [cut–off 25000 ng/mL]

AGR2 pancreatic intraepithelial neoplasia grade 3 vs nonmalignant





REG1 α

malignant vs normal AGR2 0.95 ROC–AUC

0.765 ROC–AUC; 0.67 SE, 0.90 SP 0.765 ROC–AUC; 0.11 SE, 0.95 SP Premalignant vs nonmalignant 0.742 ROC–AUC; 0.40 SE, 0.90 SP 0.742 ROC–AUC; 0.16 SE, 0.95 SP malignant vs nonmalignant 0.729 ROC–AUC; 0.25 SE, 0.90 SP 0.729 ROC–AUC; 0.11 SE, 0.95 SP

Serum/Plasma – single marker

no significant differencies

no significant differencies

malignant vs nonmalignant 0.669 ROC–AUC; 0.826 SE, 0.474 SP [cut–off, 327.97 ng/mL] Diagnostic performances

malignant vs normal 0.771 ROC–AUC; 0.826 SE, 0.818 SP [cut–off, 335.12 ng/mL]

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

Verification analysis

Sample type (number)

SYNC 0.80 ROC–AUC OLFM4 0.89 ROC–AUC COL6A1 0.74 ROC–AUC PIGR 0.86 ROC–AUC

Gold standard/reference marker





serum CA19–9 malignant vs nonmalignant 0.819 ROC–AUC; 0.696 SE, 0.895 SP [cut–off, 37 U/mL] malignant vs normal 0.885 ROC–AUC; 0.696 SE, 1 SP [cut–off, 37 U/mL]

Panel of markers







serum CA19–9 malignant vs nonmalignant 0.97 ROC–AUC

malignant vs normal plasma AGR2 & plasma SYNC & plasma OLFM4 & plasma COL6A1 & plasma PIGR 0.98 ROC–AUC plasma SYNC (or OLFM4, or COL6A1 or PIGR) & plasma

993

CA19–9 0.98 ROC–AUC plasma AGR2 & plasma CA19–9 1 ROC–AUC

994

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

of tumor-derived microvesicles in several non-systemic fluids has been reported (e.g., ascites, pleural fluid) [67,68]. As a consequence, the discovery of some potential biomarkers could be prevented by the inaccessibility of vesicular protein content. To minimize analytical hindrances and improve proteome coverage, sample pre-treatments aimed at reducing complexity and/or removing interfering substances are strongly recommended [49]. The pair wise of multiple purification and sub-fractionation steps has proved particularly effective in the analysis of gastrointestinal fluids. As one of the best examples, I cite the case of Barbhuiya et al. who analyzed bile samples using a multipronged approach consisting of delipidation, desalting, immunodepletion, SDS-PAGE, strong cation exchange chromatography (SCX), OFFGEL electrophoresis and liquid chromatography–tandem MS (LC–MS/MS) [69]. The resulting identifications (n = 2552) represented the most comprehensive catalog of biliary proteins described to date. [70]. This kind of approach, however, has its downside. Extra sample handling may indeed introduce artifacts and is not particularly suitable for quantitative analysis. When different clinical conditions have to be compared, it is therefore more prudent to limit the number of preanalytical steps. Over the years, efforts have been specifically devoted to develop pre-treatments that are compatible with quantitative proteomic of proximal fluids. Inter alia, differential centrifugation has recently been established as a powerful and reliable pre-fractionation method for the relative quantification of proteins in bile from malignant and nonmalignant biliary stenoses [71]. 4. Proximal fluids of digestive malignancies: candidate biomarkers from proteomic studies In recent years numerous proteomic analyses have been conducted to investigate the proteome of gastrointestinal proximal fluids in health and disease conditions. Beyond the purely qualitative studies, which initiated in the early years of proteomics era and exponentially grew up with the improvement of analytical techniques, the major contribution to the identification of candidate biomarkers came from the development of comparative and quantitative methods (e.g., label-based and label-free approaches) allowing to evaluate differences in protein expression between different clinical conditions [72]. In the following sections I will focus the attention on proximal fluids surrounding the above-mentioned digestive malignancies, emphasizing the most relevant results obtained by proteomic analysis specifically aimed at the comparison of malignant and nonmalignant conditions. In particular, all proteomic studies for which information about the diagnostic potential of identified candidate markers was available have been systematically discussed and summarized in Tables 1–4. 4.1. Bile as a source of biomarkers for malignant hepatopancreatobiliary lesions 4.1.1. General information Bile is produced by the liver and continuously released into the intrahepatic bile ducts, through which it reaches the gallbladder where it is stored and concentrated. Food intake causes the gallbladder to contract and deliver bile into the duodenum [73]. The main functions of bile are: i) to help emulsifying fats for digestion and absorption; ii) serve as a route of excretion for xenobiotics (e.g., toxins, drugs) and endobiotics (e.g., cholesterol, bilirubin) [70]. Bile salts, phospholipids and cholesterol are the main organic components of bile, while proteins represent only 5–7% of the biliary constituents by dry weight [70,74].

4.1.2. Proteomic studies Kristiansen et al., in 2004, performed the first large-scale proteomic study of human bile collected in the presence of a malignant stenosis due to CC [75]. Among the identified cancer-associated proteins, the authors selected galectin-3-binding protein (Mac-2BP) for further verifications [76]. Mac-2BP is a glycoprotein originally described as a tumor-secreted antigen which promotes integrin-mediated cell adhesion [77]. The levels of bile Mac-2BP were evaluated by using an enzyme-linked immunosorbent assay (ELISA) in samples from patients with biliary tract carcinoma (n = 26), benign biliary conditions (n = 32), and PSC (n = 20). In the diagnosis of biliary tract malignancies, bile Mac2BP showed an area under the curve (AUC) on the receiver operator characteristic (ROC) of 0.70. The sensitivity and specificity were 69% and 67%, respectively (cut-off value, 853 ng/mL). Finally, the panel composed by bile Mac-2BP and bile CA19-9 corresponded to a ROCAUC of 0.75 [76]. Four years later, Chen et al. identified pancreatic elastase 3B (CEL3B), a chymotrypsin-like endopeptidase, as a potential biomarker for CC, using 2D electrophoresis. Significantly higher CEL3B/amylase ratios were found in patients with CC (n = 22) than in those with gallstones (n = 28). The ROC-AUC was 0.877 and the ratio distinguished malignant from nonmalignant causes of biliary obstruction with a sensitivity of 82% and a specificity of 89% (cut-off value, 0.065) [78]. In 2011, Zabron et al. identified, by the mean of a label-free proteomic approach, neutrophil gelatinaseassociated lipocalin (NGAL) as a new candidate biomarker in bile from patients with malignant pancreatobiliary disease [79]. This protein is known to be involved in apoptosis, innate immunity and renal development and has been found to play a key role in cancer processes [80]. Bile NGAL showed 0.76 ROC-AUC, 94% sensitivity and 55% specificity (cut-off value, 570 ng/mL) in distinguishing malignant (n = 16) from nonmalignant (n = 22) conditions. In addition, the diagnostic performances of a panel composed of bile NGAL and serum CA19-9 resulted in 85% sensitivity and 82% specificity (cut-off values, 3000 ng/mL and 125 U/L, respectively). Finally, the assessment of bile NGAL for the specific diagnosis of PAC in a larger cohort of patients (PAC, n = 13; nonmalignant, n = 36), showed 100% sensitivity and 56% specificity (cut-off value, 570 ng/mL) [79]. The same year Lankisch et al., used capillary electrophoresis MS to identify diseasespecific peptide patterns in patients with CC (n = 16), PSC (n = 18) and choledocolithiasis (CL, n = 16). They further developed two models able to discriminate CC and PSC from CL (0.93 ROC-AUC, 86% specificity, 93% sensitivity) and CC from PSC (0.87 ROC-AUC, 78% specificity, 84% sensitivity), respectively [81]. A few months later, Shen et al. showed that spermatogenesis-associated protein 20 (SSP411), a testis-specific protein that may play a role in fertility regulation, was overexpressed in bile from patients with CC compared to bile from patients with cholangitis. Bile SSP411 performances were not assessed, but serum SSP411 was proven to distinguish malignant from nonmalignant conditions with 0.913 ROC-AUC, 90% sensitivity and 83% specificity [82]. More recently, by the mean of an isobaric-tagging-based proteomic approach, Farina et al. highlighted carcinoembryonic antigen-related cell adhesion molecule 6 (CEAM6) as a biomarker of malignant biliary stenoses [83]. CEAM6 is a membrane protein whose activity has been associated with proliferation, migration, invasion and angiogenesis of cancer cells [84]. In bile samples, CEAM6 was able to discriminate between malignant (n = 30) and nonmalignant stenoses (n = 13) with 0.920 ROC-AUC, 93.1% sensitivity, 83.3% specificity, 93.1% PPV, 83.3% NPV and 90.2% accuracy. A marker panel composed of serum CA19-9 and bile CEAM6

Notes to Table Abbreviations: TCA, Trichloroacetic acid; iTRAQ, isobaric tags for relative and absolute quantitation; 2-DE, bidimensional electrophoresis; ROC-AUC, area under the curve on the receiver operator characteristic; SE, sensitivity; SP, specificity; PPV, positive predictive value; NPV, negative predictive value. a Only proteomic studies for which information about the diagnostic potential of identified candidate markers was available are reported. b Classical proteomic steps, such as denaturation, reduction, alkylation and digestion are not indicated. c For additional information about candidate biomarkers, see Supplementary Table 1.

Table 3 Proteomic studies aimed at identifying cancer biomarkers of malignant gastric lesions in gastric juicea. [118]

[119]

[120]

Hsu et al. (2007 and 2010)

Chang et al. (2007)

Kon et al. (2008)

Wu et al. (2012)

Sample type (number)

malignant (n=29) gastric cancer (n=29) nonmalignant (113) gastric ulcer (n=33) duodenal ulcer (n=33) healty subjects (n=47)

malignant (n=34) gastric cancer (n=34) nonmalignant (182) gastric ulcer (n=38) duodenal ulcer (n=38) asymptomatic subjects (n=106)

malignant (n=19) gastric cancer stage 0 (n=1) gastric cancer stage I (n=4) gastric cancer stage II (n=2) gastric cancer stage III (n=2) gastric cancer stage IV (n=10) nonmalignant (n=36) normal (n=9) antral gastritis (n=9) gastritis (n=6) ulcer (n=4) hiatal hernia (n=3) hyperplastic polyps (n=2) Barrett's esophagus (n=1) fundic scar (n=1) adenomatous polyp (n=1)

malignant (n=3) gastric cancer stage I (n=3) [pooled] gastric cancer stage III (n=3) [pooled] gastric cancer stage IV (n=3) [pooled] nonmalignant (n=1) gastritis (n=3) [pooled]

Workflowb

centrifugation at 10,000xg, 2–DE

centrifugation at 10,000xg

centrifugation at 180xg, centrifugation at 16,100xg

centrifugation at 10,000xg, centrifugation at 10,300xg, purification (2–D clean–up kit), 2–DE

Mass spectrometer Number of identified proteins

MALDI–TOF and LTQ–FT – –

SELDI–TOF – 5 peaks up–regulated: 2187, 2387 and 3572 m/z down–regulated: 2753 and 4132m/z 5 peaks

SELDI–TOF – 106 total 60 up–regulated 46 down–regulated 106 peaks

MALDI–TOF/TOF – 15



malignant (n=24) gastric cancer stage I (n=5) gastric cancer stage II (n=4) gastric cancer stage III (n=2) gastric cancer stage IV (n=12) gastric cancer stage undetermined (n=1) nonmalignant (n=29) normal (n=8) gastritis (n=14) fundic gland polyps (n=2) acute gastric ulcer (n=2) duodenitis (n=2) hiatal hernia (n=1)

malignant (n=43) early gastric cancer (n=22) late gastric cancer (n=21) nonmalignant (n=17) gastritis (n=17)

– 5 peaks gastric cancer vs nonmalignant 0.87 ROC–AUC; 0.35–0.79 SE, 069–0.94 SP

– 106 peaks gastric cancer vs gastritides 0.88 SE, 0.93 SP

immunoblot (equal quantity) early gastric cancers vs gastritis S100A9 0.75 ROC–AUC AAT 0.71 ROC–AUC early vs late gastric cancer S100A9 0.78 ROC–AUC GIF 0.83 ROC–AUC – –

Proteomic analysis

[114] [115]

Author (year)

Number of differentially expressed proteins in malignant conditions

α 1-antitrypsin

Sample type (number)

cohort A malignant (n=22) gastric cancer (n=22) nonmalignant (n=90) gastric ulcer (n=30) duodenal ulcer (n=30) healty subjects (n=30) cohort B malignant (n=19) gastric adenocarcinoma (n=18) gastric lymphoma (n=1) nonmalignant (n=74) functional dyspepsia (n=25) gastric erosion (n=7) gastric ulcer (n=21) duodenal ulcer (n=11) erosive esophagitis (n=6) gastric hyperplastic polyp (n=4) ELISA (equal volume)

Verification analysis

Candidate Biomarkerc

Diagnostic performances

Method Gastric juice – single marker

α 1-antitrypsin malignant vs nonmalignant cohort A 0.963 ROC–AUC; 0.96 SE, 0.92 SP, 0.75 PPV, 0.85 NPV [cut–off 717 ug/dL] 0.96 SE, 0.91 SP, 0.72 PPV, 0.99 NPV [cut–off 618 ug/dL] 0.91 SE, 0.92 SP, 0.74 PPV, 0.98 NPV [cut–off 806 ug/dL] cohort B 0.84 ROC–AUC; 0.74 SE, 0.88 SP, 0.61 PPV, 0.93 NPV [cut–off 0.85 ug/dL]

Serum – single marker Gold standard/reference marker

– –

– –

– –

Panel of markers







protein S100–A9 (S100A9) gastric intrinsic factor (GIF) α1–antitrypsin

995

early gastric cancers vs gastritis gastric juice S100A9 & AAT 0.81 ROC–AUC early vs late gastric cancer gastric juice S100A9 & GIF 0.92 ROC–AUC gastric juice S100A9 & GIF & AAT 0.90 ROC–AUC

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

Reference

996

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

further improved the diagnostic efficiency: 0.960 ROC-AUC, 96.6% sensitivity, 83.3% specificity, 93.3% PPV, 90.9% NPV and 92.7% accuracy [83] (Table 1). Some other overexpressed or cancer-associated proteins have been highlighted in proteomic analyses of bile samples collected from malignant conditions. However, no evaluation of diagnostic performances has been reported in these studies [70,71,75,85–89]. 4.2. Pancreatic juice as a source of biomarkers for malignant pancreatic lesions 4.2.1. General information When partially digested food (i.e., chyme) is expelled by the stomach and enters the duodenum, pancreatic juice is released by the relaxation of the hepatopancreatic sphincter. Pancreatic juice is an alkaline mixture of digestive enzymes and bicarbonate ions secreted by the exocrine acinar cells and epithelial cells lining the pancreatic ducts, respectively. Enzymes in pancreatic juice are active in the digestion of proteins (e.g., proteases, such as trypsin and chymotrypsin), fats (e.g., pancreatic lipase) and sugars (e.g., amylase) [90]. 4.2.2. Proteomic studies The first pioneering quantitative proteomic analyses of pancreatic cancer juice were performed by Chen et al. at the end of the 2000s [91,92]. In these studies, 30 and 27 proteins were detected, which showed abundance changes in pancreatic cancer or pancreatitis vs. normal conditions, respectively [91,92]. The cancer-associated overexpression of one of these proteins, insulin-like growth factor-binding protein 2 (IGFBP-2), was further verified by immunoblot in pancreatic juice and tissue [91]. In the following years, a number of further comparative studies were conducted, which led to propose several pancreatic juice proteins as potential cancer biomarkers (e.g. metalloproteinase 9 (MMP9), protein DJ-1 (DJ-1), α-1B-glycoprotein (A1BG), major vault protein (MVP), anterior gradient 2 (AGR2), 14-3-3 protein sigma(1433S), annexin A4(ANXA4), protein S100-A10 (S100A10), serine proteinase 2 (PRSS2), elastase 3B (ELA3B) and transthyretin (TTR)) [93–97]. Of particular relevance, however, were four studies in which a contextual evaluation of the diagnostic performances of new potential biomarkers was performed to corroborate proteomic findings. In the first study, Shirai et al. analyzed pancreatic juice from patients with intraductal papillary mucinous neoplasm (IPMN, n = 33; PAC, n = 54; chronic pancreatitis, CP, n = 31) by using surface-enhanced laser desorption/ionization time-of-flight (SELDI-TOF) MS. An effective overexpression of a 6240-Da peak was detected in IPMN samples and significantly higher levels of the corresponding protein, pancreatic secretory trypsin inhibitor (PSTI), were confirmed by radioimmunoassay [98]. PSTI, also known as tumor associated trypsin inhibitor (TATI), classically controls pancreas protease activity to prevent pancreatitis. Its levels have been found increased in various malignant diseases (e.g. gynecological, urothelial, digestive) and its activity has been shown to correlate with invasion and migration in bladder cancer cell lines [99]. The diagnostic performances of PSTI in pancreatic juice were proven as follows: 48% sensitivity, 98% specificity, 89% positive predictive value and 83% negative predictive value (cut-off value of 25,000 ng/mL) [98]. In a second study, Chen et al. identified 20 proteins whose expression was 2–10 fold higher in individual samples of pancreatic juice from premalignant pancreatic intraepithelial neoplasia of grade 3 (PanIN3, n = 3) compared to pooled samples from nonmalignant diseases (n = 5) [100]. One of these proteins, anterior gradient-2

(AGR2), was further verified by ELISA in a cohort of 25 premalignant (PanIN2, PanIN3 and IPMN), 8 malignant (PAC stage 2, stage 3, stage 4 and undetermined) and 18 benign specimens (sphincter of Oddi dysfunction and CP). Significantly elevated levels of AGR2 were found in pancreatic juice, but not in serum, of patients with premalignant conditions. The best diagnostic performances of AGR2 were achieved in differentiating PanIN3 from nonmalignant conditions, with 0.765 ROC-AUC, 67% sensitivity and 90% specificity [100]. These results are consistent with those of other studies showing that AGR2 plays a significant role in cancer cell growth and survival, as well as in metastasis development [101,102]. Park et al., in 2011, highlighted 26 upregulated spots in bidimensional electrophoresis (2-DE) profiles of pancreatic juice from cancer patients. The overexpression of three proteins, lithostathine-1-alpha (REG1α), brefeldin A-inhibited guanine nucleotide-exchange protein 2 (BIG2) and peroxiredoxin-6 (PRDX6), was further confirmed in pancreatic cancer tissues by immunohistochemical staining. Among others, the growth factor REG1α proved of particular interest because of its established ability to promote tissue regeneration and accelerate tumor progression. The diagnostic value of serum REG1α was thus evaluated, showing 0.771 ROC-AUC, 82.6% sensitivity and 81.8% specificity in distinguishing between pancreatic cancer and normal patients. Lower performances were instead observed when CP was included in the analysis [103]. Finally, Makawita et al. performed the most comprehensive study in the field, by integrating and comparing the proteomes of six pancreatic cancer cell lines, one normal human pancreatic ductal epithelial cell line and two pools of six pancreatic juice samples from PAC patients. As a result, seven overexpressed proteins were selected for further ELISA verification, of which five (e.g. anterior gradient homolog 2 (AGR2), olfactomedin-4 (OLFM4), syncollin (SYNC), collagen alpha-1(VI) chain (COL6A1) and polymeric immunoglobulin receptor (PIGR)) showed a significant increase in plasma from pancreatic cancer patients (n = 20) vs. normal subjects (n = 20). Interestingly, most of these proteins were previously proven to be involved at different levels in cancer processes (Supplementary Table 1). The panel composed of the five candidate biomarkers proven to be slightly more discriminatory in distinguishing between cancer and normal conditions than CA19-9 alone (0.98 vs. 0.97 ROC-AUC). In addition, each of the proteins (i.e., AGR2) was able to improve the CA19-9 ROC-AUC to 0.98 or 1 [104] (Table 2). 4.3. Pancreatic cyst fluid as a source of biomarkers for malignant pancreatic cysts 4.3.1. General information Cystic lesions, including those arising from the pancreas, consist in a collection of fluid surrounded by normal or tumor tissue. The fluid is continually secreted by the epithelial lining of the cyst and may drain into pancreatic ducts or stagnate within the cyst. From a biochemical point of view, the fluid content strongly varies according to the nature of the cyst (e.g., mucinous, serous) [105]. 4.3.2. Proteomic studies Up to now, only four studies used proteomic technologies to compare fluid from malignant and nonmalignant pancreatic cysts. In the first study, pancreatic cyst fluid samples were analyzed by SELDI-TOF MS. This analysis highlighted differential expression profiles between malignant (pancreatic ductal adenocarcinoma (PDAC)), premalignant (neuroendocrine tumor (NET), IPMN, mucinous cystadenoma (MC))

Notes to Table Abbreviations: 2-DE, bidimensional electrophoresis; ROC-AUC, area under the curve on the receiver operator characteristic; SE, sensitivity; SP, specificity; PPV, positive predictive value; NPV, negative predictive value. a Only proteomic studies for which information about the diagnostic potential of identified candidate markers was available are reported. b Classical proteomic steps, such as denaturation, reduction, alkylation and digestion are not indicated. c For additional information about candidate biomarkers, see Supplementary Table 1.

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

997

Table 4 Proteomic studies aimed at identifying cancer biomarkers of malignant oral lesions in salivaa. Reference

[134]

[137]

[138]

Author (year)

Hu et al. (2008)

Jou et al. (2010)

de Jong et al. (2010)

Sample type (number)

malignant (n=1)

malignant (n=1)

malignant (n=1)

Proteomic analysis

oral squamous cell carcinoma (n=16) [pooled] nonmalignant (n=1)

oral squamous cell carcinoma (n=UN) [pooled] nonmalignant (n=1)

healty subjects (n=16) [pooled]

healty subjects (n=UN) [pooled]

oral squamous cell carcinoma (n=4) nonmalignant (n=1) premalignant lesions (n=4)

Workflowb

C4 RP—HPLC pre—fractionation, capillary RP—HPLC

acetone precipitation, 2—DE

centrifugation at 3,000xg, iTRAQ, IEF, SCX, micro capillary RP—HPLC

Mass spectrometer

QSTAR XL

MALDI—TOF/TOF

LTQ linear ion trap

Number of identified proteins

468 (cancer) and 431 (healty)



855

Number of differentially

52 proteins only found in cancer

15 2—fold increased

198

expressed proteins in

31 2—fold decreased

Verification analysis

malignant conditions Candidate Biomarkerc

Protein S100—A9 (S100A9), profilin, CD59 glycoprotein (CD59), catalase, Galectin—3—binding protein (Mac—2BP)

transferrin

myosin, actin

Sample type (number)

malignant (n=48)

malignant (n=41)

malignant (n=12)

oral squamous cell carcinoma (n=48)

oral squamous cell carcinoma (n=41)

nonmalignant (n=48)

stage T1 (n=17)

healty subjects (n=48)

malignant lesions, oral squamous cell carcinoma (n=12) premalignant (n=12)

stage T2 (n=15) stage T3 (n=4)

premalignant lesions (n=12) nonmalignant (n=1)

stage T4 (n=5)]

healthy (n=UN) [pooled]

nonmalignant (n=30) healty subjects (n=30) Method

ELISA and immunoblot

ELISA (equal quantity)

immunoblot (equal quantity)

Saliva — single marker



transferrin

myosin

T1 vs healty

malignant vs premalignant

0.95 ROC—AUC; 1 SE [cut—off, 0.3 OD] T2 vs healty

0.67 SE, 0.83 SP actin

0.94 ROC—AUC; 0.866 SE [cut—off, 0.3 OD]

malignant vs premalignant

Diagnostic performances

T3/T4 vs healty

1 SE, 0.75 SP

0.91 ROC—AUC; 1 SE [cut—off, 0.3 OD] Serum — single marker



no significant difference



Tissue — single marker















Gold standard/reference marker Panel of markers

saliva MRP14 & PROF & CD59 & CATA & MAC—2BP malignant vs nonmalignant 0.93 ROC—AUC; 0.90 SE, 0.83 SP

Abbreviations: RP-HPLC, reversed-phase high-performance liquid chromatography; 2-DE, bidimensional electrophoresis; iTRAQ, isobaric tags for relative and absolute quantitation; IEF, isoelectrofocusing; SCX, strong cation exchange; ROC-AUC, area under the curve on the receiver operator characteristic; SE, sensitivity; SP, specificity; UN, unknown. a Only proteomic studies for which information about the diagnostic potential of identified candidate markers was available are reported. b Classical proteomic steps, such as denaturation, reduction, alkylation and digestion are not indicated. c For additional information about candidate biomarkers, see Supplementary Table 1. d Validation study of previously discovered proteomic and transcriptomic biomarkers.

and nonmalignant (pseudocyst (PP)) cystic lesions of the pancreas. Marked differences were also detected between premalignant etiologies, IPMN and MC [106]. In a second study, Ke et al. investigated pancreatic cyst fluid samples corresponding to nonmalignant and atypical/suspicious cytological diagnosis. A number of potential protein biomarkers were identified which correlate with CEA (i.e., homologs from amylase, mucins, CEACAMs and S100 families) [105]. In the third study, fluids from cystic lesions (serous cystadenoma (SCA), MCN, NET, IPMN and PP) were analyzed by in-gel tryptic digestion followed by liquid chromatography–tandem mass spectrometry (GeLC–MS/MS). Among identified proteins, two of them, olfactomedin 4 (OLFM-4) and mucin 18 (MUC18), were found specifically associated with cyst fluids from MCN/IPMN and NET, respectively. Immunoblots and immunohistochemistry allowed verifying the differential expression of selected proteins in cyst fluids and tissues [107]. Finally, Mann et al., in 2012, performed glycomic and glycoproteomic profiling experiments in 21 pancreatic cyst fluid samples (MCN, IPMN, SCA, PP) showing that six of the samples exhibited high levels of multiple fucosylation. Subsequent investigations highlighted several candidate glycoproteins that appeared hyperfucosylated (e.g., triacylglycerol lipase, pancreatic α-amylase) [108]. Despite encouraging preliminary results, however, none of the above-mentioned candidate biomarker has been further assessed to evaluate the diagnostic performances in distinguishing pancreatic cyst etiologies.

4.4. Gastric juice as a source of biomarkers for malignant gastric lesions 4.4.1. General information Gastric juice is a fluid produced by gastric mucosal glands and its secretion is regulated by the ingestion of food. It is mainly composed by: i) mucus, containing gel-forming glycoproteins (e.g., mucins); ii) digestive enzymes, such as pepsin and less abundant species (e.g., lipase, cathepsin, gelatinase); iii) concentrated hydrochloric acid; and iv) electrolytes (e.g., sodium, potassium, calcium, phosphate, sulfate, and bicarbonate) [109]. Gastric juice acts to solubilize food particles, digest proteins and protect gut from pathogens and mechanical injuries [109,110]. Due to the hydrochloric acid content, the pH range of gastric juice usually falls between 0.9 and 1.5 [111]. 4.4.2. Proteomic studies The number of proteomic studies aimed at analyzing gastric juice is very limited. To date, about a dozen publications are available in the field and only a minor part of them has been devoted to the search of cancer biomarkers [112]. The first comparative study dates back to 2004, when Lee and coworkers confronted 2-DE proteomic patterns of gastric juice collected from various gastric diseases (e.g., gastric cancers, chronic atrophic gastritis and other diseases) with those of healthy subjects. As a result, α1-antitrypsin was found highly overexpressed in 63% gastric cancers [113]. A few years later, two more publications by Hsu

998

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

[139]

[140]

He et al. (2011)

Jou et al. (2011)

Brinkmann et al. (2011)

malignant (n=28)

malignant (n=47)



oral squamous cell carcinoma (n=28)

[141] d

oral squamous cell carcinoma (n=47)

nonmalignant (n=8)

stages T1+T2 (n=32)

oral leukoplakia (n=8)

stages T3+T4 (n=15) nonmalignant (n=30) healty subjects (n=30)

centrifugation at 1,000xg or 10,000xg

centrifugation at 16,000xg, C8—magnetic beads peptide binding



SELDI—TOF

MALDI—TOF









164 differential peaks in saliva

3 differential peaks in saliva



2919 m/z peak, corresponding to ZNF510 protein (ZNF510)

IL1B protein + SAT1 mRNA +DUSP1 mRNA

108 differential peaks in serum 218 differential peaks in tissues 5818, 4617 and 3884 m/z peaks in saliva 6886 and 4162 m/z peaks in serum

IL1B mRNA + IL8 mRNA + SAT1 mRNA

11366 and 3738 m/z peaks in tissues —

IL1B protein + DUSP1 mRNA —

malignant (n=35) oral squamous cell carcinoma (n=35) stages T1+T2 (n=18) stages T3+T4 (n=17) nonmalignant (n=51) healty subjects (n=51)



ELISA (equal volume)

ELISA and qPCR

5818, 4617 and 3884 m/z peaks

ZNF510

(see [113] for single marker performances)

malignant vs nonmalignant 1 SE, 1 SP

T1+T2 vs healty 0.95 ROC—AUC T3+T4 vs healty 0.98 ROC—AUC

6886 and 4162 m/z peaks



















saliva IL1B protein & SAT1 mRNA & DUSP1 mRNA

malignant vs nonmalignant 0.8782 SE, 0.9286 SP 11366 and 3738 m/z peaks malignant vs nonmalignant 0.9629 SE, 1 SP

malignant vs nonmalignant 0.86 ROC—AUC; 0.89 SE, 0.78 SP saliva IL1B mRNA +&I L8 mRNA +&SAT1 mRNA T1+T2 vs nonmalignant 0.85 ROC—AUC; 0.67 SE, 0.96 SP saliva IL1B protein + DUSP1 mRNA T3+T4 vs nonmalignant 0.88 ROC—AUC; 0.82 SE, 0.84 SP

et al. raised the possibility that α1-antitrypsin could be a valuable biomarker of gastric cancer [114,115]. In the first study, the authors used a 2-DE approach to show that α1-antitrypsin was the principal component of specific protein patterns related to gastric cancer samples [114]. Afterwards, they compared α1-antitrypsin concentrations in gastric juice from gastric cancer (n = 22), gastric ulcer (n = 30), duodenal ulcer (n = 30) and healthy subjects (n = 30). The ROC-AUC for identifying gastric cancer was 0.96, with 96% sensitivity and 92% specificity at a cut-off of 717 μg/dL. The value of gastric juice α1-antitrypsin in differentiating gastric malignancies was further validated in an independent cohort of 93 samples collected from dyspeptic patients by using a non-invasive string test. In such a screening, α1-antitrypsin showed lower but nonetheless good diagnostic performances (0.84 ROC-AUC, 74% sensitivity and 88% specificity) [115]. Overall, these results are in line with previous findings demonstrating that α1-antitrypsin is highly overexpressed in various tumors of different origins and that it is involved in cancer cell migration and invasion [116,117]. SELDI-TOF MS was also used to explore gastric juice for potential cancer biomarkers. Chang et al. identified five differential peptides by comparing gastric juice profiles from more than 200 patients. The five biomarkers showed good diagnostic accuracy with an ROC-AUC of 0.87 and the optimum combination was achieved by combining three of them. The overall sensitivity and specificity for detecting gastric cancer corresponded to 79% and 92%, respectively [118]. Kon et al., in 2008, similarly performed a proteomic profiling of gastric juice from gastric cancer (n = 19) and benign gastritis (n = 36), obtaining a list of 106 differential proteomic features. The validation of such a panel in a

second blinded set of samples showed 88% sensitivity and 93% specificity [119]. Finally, Wu et al., in 2012, highlighted the expression changes of protein S100-A9 (S100A9), gastric intrinsic factor (GIF) and α1-antitrypsin by comparing 2-DE patterns of gastric juice from gastritis and gastric cancer at different stages. Subsequent immunoblot verifications revealed that two panels composed of S100A9/α1-antitrypsin and S100A9/GIF were able to distinguish early stage cancers from gastritis (0.81 ROC-AUC) and monitor gastric cancer prognosis (0.92 ROC-AUC), respectively [120]. The bibliographic search, in this case, did not reveal significant studies supporting the hypothesis of a correlation between GIF protein and tumors. In contrast, S100A9 is a wellknown mediator in acute and chronic inflammation whose overexpression has been reported in both malignant and inflammatory diseases [121] (Table 3). 4.5. Ascites as a source of biomarkers for peritoneal carcinomatosis 4.5.1. General information Ascites refers to the abnormal accumulation of fluid into the peritoneal cavity. The most common cause of intraperitoneal ascites is cirrhosis, which accounts for approximately 80% of all cases [122]. A number of other nonmalignant pathologies may also cause ascites (e.g., heart failure, tuberculosis, pancreatitis, nephrosis, peritonitis) [122,123]. Malignant ascites accounts for about a tenth of all cases and originates from a combination of lymphatic drainage impairment and vascular permeability increase, secondary to the presence of cancer cells in the peritoneal space [123]. The occurrence of malignant ascites is

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

considered as a sign of peritoneal carcinomatosis. The ascitic fluid is basically composed of water, electrolytes and plasma-derived proteins but its composition strongly depends on the disease with which it is associated [122]. 4.5.2. Proteomic studies Ascites proteome has been mainly studied for the search of cancer biomarkers in the presence of ovarian malignancies [124–127]. To date, only two proteomic studies have been reported in the literature which focus on the analysis of malignant ascites caused by digestive cancers. In the first study, Choi et al. analyzed microvesicles derived from human colorectal cancer ascites. Among the 846 proteins that were identified, several might be involved in tumor progression through a variety of mechanisms (disruption of epithelial polarity, migration, invasion, tumor growth, immune modulation, and angiogenesis). These proteins include catenins, galectins, claudins, tetraspanins and cell adhesion molecules [128]. In the second study, Kosanam et al. adopted a proteomic approach that included size-exclusion, ionexchange and lectin-affinity chromatography coupled to LC–MS/MS to analyze ascitic fluid associated to pancreatic cancer. As a result, 816 proteins were identified and 20 putative biomarker candidates were proposed for future validation and identification [129]. In both mentioned studies, the diagnostic potential of the highlighted cancer-related proteins was not investigated. 4.6. Saliva as a source of biomarkers for malignant oral lesions 4.6.1. General information Saliva is secreted by salivary glands in response to various stimulations (e.g., time of day, food, age, gender). The majority of secretion mainly derives from the parotid (20%), submandibular (65–70%) and sublingual glands (7–8%), each of which produces a characteristic fluid (i.e., serous, sero-mucous or mucous) which considerably differs in terms of composition and properties [130,131]. The main components of saliva include: inorganic (e.g., strong and weak ions), organic non-protein (e.g., uric acid, glucose, lactate, amines), protein/ polypeptide (e.g., digestive enzymes, inhibitors of proteinases), hormones (e.g., steroids, non-steroids, peptide and protein hormones) and lipid molecules (e.g., cholesterol, fatty acids) [132]. Saliva is mostly involved in protection, lubrication, digestion, taste, cleaning and maintenance of tooth enamel integrity [131]. 4.6.2. Proteomic studies Over the years, saliva aroused the interest of researchers because of its accessibility and ease of sampling. To date, approximately 10,000 publications are listed in PubMed relating to this fluid. Of these, a little more than a hundred correspond to studies involving proteomic approaches. Recently, Amado et al. summarized the major achievement obtained in the field of salivary proteomics showing that up to 2340 total proteins have been described in the whole saliva to date [65,133]. Among the broad number of oral and systemic diseases for which saliva has been explored to identify potential biomarkers, head/neck squamous cell carcinoma (HNSCC), and more particularly OSCC, have been the most intensely studied [65]. Despite the considerable number of publications in the field, however, the first proteomic study including the assessment of the diagnostic potential of selected candidate biomarkers was made by Hu et al., in 2008. The authors performed a shotgun proteomic profiling of pooled saliva samples from patient with OSCC (n = 16) and healthy subjects (n = 16). Among highlighted candidates, the overexpressed proteins S100A9, profilin, CD59 glycoprotein (CD59), catalase and Mac-2BP were successfully verified by immunoassay [134]. The panel composed of these biomarkers showed the following diagnostic performances in detecting OSCC: 0.93 ROC-AUC, 90% sensitivity and 83% specificity [134]. It is worth noting that all the selected molecules have been previously reported in the literature as being dysregulated in cancer diseases

999

(Supplementary Table 1). However, some of them have rather been described as frequently downregulated (e.g. profilin, catalase) [135,136]. Two years later, Jou and coworkers analyzed saliva from patients with OSCC (n = 41) and healthy subjects (n = 30) by using a 2-DE approach. As a result, salivary transferrin was found able to differentiate between normal and cancer samples as follows: i) stage T1; 0.95 ROCAUC, 100% sensitivity; ii) stage T2; 0.94 ROC-AUC, 86.6% sensitivity; and iii) stages T3/T4; 0.91 ROC-AUC; 100% sensitivity. The measurement of transferrin in plasma samples, instead, showed no significant results [137]. These results agree with a few studies previously reported in the literature (Supplementary Table 1). However, the role of transferrin in cancers, as well as its diagnostic value as a biomarker of malignant conditions, is still unclear. De Jong et al., in the same year, performed a label-based quantitative proteomic analysis to compare saliva samples from malignant OSCC (n = 4) and premalignant oral lesions (n = 4). Among differentially expressed proteins, myosin and actin were proven to distinguish between the two conditions with 67% and 83%, 100% and 75% specificity, respectively [138]. Unfortunately, the antibodies selected for immunoblot verifications were somewhat aspecific, being able to detect a wide range of isoforms (Supplementary Table 1). As a consequence, it was not possible to argue about the involvement of a specific protein in cancer diseases. Finally, three new studies were published in 2011, which aimed at identifying new reliable salivary biomarkers of OSCC. In the first, the authors established three panels of differential protein peaks in saliva, serum and tissue samples, by using a SELDI-TOF approach. These allowed distinguishing between OSCC (n = 28) and oral leukoplakia (OLK) patients (n = 8) with absolute sensitivity and specificity in saliva samples, while lower diagnostic performances were recorded in tissue and serum samples [139]. In the second study, three MS peaks from salivary peptidome were identified, which showed a differential expression in OSCC patients (n = 47) and healthy donors (n = 30). One of these peaks, corresponding to a 24-mer peptide of zinc finger protein 510 (ZNF510), was further proven to predict OSCC progression with 0.95 and 0.98 ROC-AUC for T1 + T2 and T3 + T4 stages, respectively [140]. Lastly, Brinkmann et al. performed a verification analysis to evaluate the diagnostic potential of several OSCC candidate biomarkers obtained from previous proteomic and transcriptomic studies. By combining 3 protein and 6 mRNA markers, the study allowed the establishment of three marker panels showing the following diagnostic performances: 0.86, 0.85 and 0.88 ROC-AUC; 89%, 67% and 82% sensitivity; 78%, 96% and 84% specificity for OSCC total, T1–T2 and T3–T4 stages, respectively [141]. These results were subsequently validated by 2 independent laboratories in 5 independent cohorts of 395 subjects [142] (Table 4). Several other interesting studies have been reported, which explored the salivary proteome highlighting potential biomarkers for oral cancers [65,142–149] but further investigations will be required to determine whether candidate biomarkers have a potential diagnostic value. To conclude, it is interesting to note that saliva specimens have also been used to search for biomarkers of digestive malignancies other than oral tumors. In particular, a pilot study based on SELDI-TOF MS profiling has been reported, which aimed at investigating saliva samples from patients with gastric cancer (n = 22) and healthy donors (n = 18). The combination of four differential MS peaks was proven to discriminate between cancer and normal subjects with a 100% specificity and 95.65% sensitivity [150]. Salivary proteomic biomarkers for breast and lung cancers have also been proposed [151,152]. 5. Translating proximal fluid biomarkers into clinical setting: perspectives Among candidate biomarkers highlighted in the above reviewed studies, some seem to have the potential for being transferred into

1000

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

clinical setting. In clinical laboratories, the measurement of protein biomarkers in digestive fluids such as ascites and pancreatic cyst fluid can be performed on automated immunoanalyzers. For fluids with chemical or physical characteristics differing largely from plasma, such as bile and viscous cyst fluids, measurement can be done in diagnostic laboratories with well-established alternative methods (e.g., commercial ELISA). Translating cancer biomarkers into diagnostic routine, however, currently represents a very critical issue. In recent decades, in fact, despite the exponential rise in publications, a few biomarkers have become widely used in diagnostic routine [153–155]. Failure of a biomarker to pass from the discovery to the clinical phase is usually related to a breakdown in the validation phase [153,156]. This may happen at the pre-analytical, analytical or post-analytical stage [153,155,156]: • At the pre-analytical stage, inaccurate assessments of patient characteristics (e.g., age, medical treatments) may cause an inadequate stratification of the population. Carelessness in sample storage or handling may also cause artifactual alteration of biomarker levels. • At the analytical stage, diagnostic performance may be miscalculated because of a number of samples that is insufficient to ensure statistically significant results. Measurement bias and artifacts could also be introduced if a reliable quantitative method is not used. • Finally, at the post-analytical stage, unsound data interpretation may lead to invalid conclusions. Even if a biomarker passes the validation phase and is approved for diagnostic use, it may still fail entering the diagnostic practice [156]. Clinicians play a key role at this stage. If they do not perceive, in fact, that the biomarker represents a significant improvement over established tools, it may be definitely ignored [157]. In addition, costeffectiveness is nowadays an important parameter to consider. For a successful translational proteomics, a close collaboration between basic and clinical researchers is thus crucial to make clinically significant progresses.

6. Conclusions Proximal fluid proteomics is emerging in the arena of digestive cancer biomarker discovery. In this context, the present review has been conceived to illustrate the state-of-the-art of proteomic investigations aimed at identifying new cancer biomarkers in fluids surrounding gastrointestinal malignancies. In particular, the most relevant proteomic results obtained by comparing fluids surrounding malignant and nonmalignant gastrointestinal lesions have been summarized and a comprehensive catalog of proteomic studies in which potential cancer biomarkers from gastrointestinal fluids have been identified and assessed for their diagnostic performances has been provided. The author believes that the present work could serve as a “Gastrointestinal Fluid Proteomics — Global Positioning System” helping future investigators to navigate the planet of digestive cancer biomarkers. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bbapap.2013.10.011.

Competing interests The author declares that no competing interests exist.

Acknowledgements The author warmly thanks Prof. Jean-Louis Frossard, Prof. Denis Hochstrasser and Dr. Myriam Delhaye for having contributed their expertise to the acquisition of knowledge in the field. Additional special thanks to Dr. Jean-Marc Dumonceau and Dr. Pierre Lescuyer for having constructively supported the elaboration of this manuscript.

References [1] A. Jemal, F. Bray, M.M. Center, J. Ferlay, E. Ward, D. Forman, Global cancer statistics, CA Cancer J. Clin. 61 (2011) 69–90. [2] American Cancer Society, Cancer Facts & Figures 2013, American Cancer Society, Atlanta, 2013. [3] M.A. Goldfarb, T. Baker, An eight-year analysis of surgical morbidity and mortality: data and solutions, Am. Surg. 72 (2006) 1070–1081(discussion 1126-1048). [4] K. Iwamura, Capsule formation in hepatocellular carcinomas arising from liver cirrhosis, Leber Magen Darm 18 (1988) 56–60. [5] Y.S. Guan, Y. Hu, Y. Liu, Multidetector-row computed tomography in the management of hepatocellular carcinoma with transcatheter arterial chemoembolization, J. Gastroenterol. Hepatol. 21 (2006) 941–946. [6] C. Bartolozzi, L. Crocetti, M.C. Della Pina, How to differentiate liver lesions in cirrhosis, JBR-BTR 90 (2007) 475–481. [7] N.C. Yu, V. Chaudhari, S.S. Raman, C. Lassman, M.J. Tong, R.W. Busuttil, D.S.K. Lu, CT and MRI improve detection of hepatocellular carcinoma, compared with ultrasound alone, in patients with cirrhosis, Clin. Gastroenterol. Hepatol. 9 (2011) 161–167. [8] G. Bertino, A. Ardiri, M. Malaguarnera, G. Malaguarnera, N. Bertino, G.S. Calvagno, Hepatocellualar carcinoma serum markers, Semin. Oncol. 39 (2012) 410–433. [9] R. Masuzaki, S.J. Karp, M. Omata, New serum markers of hepatocellular carcinoma, Semin. Oncol. 39 (2012) 434–439. [10] Use of tumor markers in liver, bladder, cervical, and gastric cancers, The National Academy of Clinical Biochemistry Laboratory Medicine Practice Guidelines, vol. 2013, American Association for Clinical Chemistry, 2010. [11] American Cancer Society, Cancer Facts & Figures 2012, American Cancer Society, Atlanta, 2012. [12] S. Cascinu, M. Falconi, V. Valentini, S. Jelic, Pancreatic cancer: ESMO clinical practice guidelines for diagnosis, treatment and follow-up, Ann. Oncol. 21 (Suppl. 5) (2010) v55–v58. [13] O.P. Zakharova, G.G. Karmazanovsky, V.I. Egorov, Pancreatic adenocarcinoma: outstanding problems, World J. Gastrointest. Surg. 4 (2012) 104–113. [14] J.M. Dumonceau, M. Polkowski, A. Larghi, P. Vilmann, M. Giovannini, J.L. Frossard, D. Heresbach, B. Pujol, G. Fernandez-Esparrach, E. Vazquez-Sequeiros, A. Gines, Indications, results, and clinical impact of endoscopic ultrasound (EUS)-guided sampling in gastroenterology: European Society of Gastrointestinal Endoscopy (ESGE) Clinical Guideline, Endoscopy 43 (2011) 897–912. [15] J.M. Dumonceau, T. Koessler, J.E. van Hooft, P. Fockens, Endoscopic ultrasonographyguided fine needle aspiration: relatively low sensitivity in the endosonographer population, World J. Gastroenterol. 18 (2012) 2357–2363. [16] U.K. Ballehaninna, R.S. Chamberlain, The clinical utility of serum CA 19-9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: an evidence based appraisal, J. Gastrointest. Oncol. 3 (2012) 105–119. [17] M.M. Bedi, M.D. Gandhi, G. Jacob, V. Lekha, A. Venugopal, H. Ramesh, CA 19-9 to differentiate benign and malignant masses in chronic pancreatitis: is there any benefit? Indian J. Gastroenterol. 28 (2009) 24–27. [18] J.C. Box, H.O. Douglas, Management of cystic neoplasms of the pancreas, Am. Surg. 66 (2000) 495–501. [19] G.F. Hutchins, P.V. Draganov, Cystic neoplasms of the pancreas: a diagnostic challenge, World J. Gastroenterol. 15 (2009) 48–54. [20] R.S. Kwon, D.M. Simeone, The use of protein-based biomarkers for the diagnosis of cystic tumors of the pancreas, Int. J. Proteomics 2011 (2011) 413646. [21] O.A. Catalano, D.V. Sahani, S.P. Kalva, M.S. Cushing, P.F. Hahn, J.J. Brown, R.R. Edelman, MR imaging of the gallbladder: a pictorial essay, Radiographics 28 (2008) 135–155(quiz 324). [22] E.C. Lazcano-Ponce, J.F. Miquel, N. Munoz, R. Herrero, C. Ferrecio, I.I. Wistuba, P. Alonso de Ruiz, G. Aristi Urista, F. Nervi, Epidemiology and molecular pathology of gallbladder cancer, CA Cancer J. Clin. 51 (2001) 349–364. [23] S.A. Khan, H.C. Thomas, B.R. Davidson, S.D. Taylor-Robinson, Cholangiocarcinoma, Lancet 366 (2005) 1303–1314. [24] F. Mihalache, M. Tantau, B. Diaconu, M. Acalovschi, Survival and quality of life of cholangiocarcinoma patients: a prospective study over a 4 year period, J. Gastrointestin. Liver Dis. 19 (2010) 285–290. [25] L.X. Qin, Z.Y. Tang, Hepatocellular carcinoma with obstructive jaundice: diagnosis, treatment and prognosis, World J. Gastroenterol. 9 (2003) 385–391. [26] G.C. Vitale, M. George, K. McIntyre, G.M. Larson, T.J. Wieman, Endoscopic management of benign and malignant biliary strictures, Am. J. Surg. 171 (1996) 553–557. [27] R. Gupta, G.V. Rao, D.N. Reddy, Benign biliary stricture—should they be dilated or treated surgically? Indian J. Gastroenterol. 25 (2006) 202–205. [28] J.M. Dumonceau, C. Macias Gomez, C. Casco, M. Genevay, M. Marcolongo, M. Bongiovanni, P. Morel, P. Majno, A. Hadengue, Grasp or brush for biliary sampling at endoscopic retrograde cholangiography? A blinded randomized controlled trial, Am. J. Gastroenterol. 103 (2008) 333–340. [29] K. Inui, J. Yoshino, H. Miyoshi, Differential diagnosis and treatment of biliary strictures, Clin. Gastroenterol. Hepatol. 7 (2009) S79–S83. [30] P. Charatcharoenwitthaya, F.B. Enders, K.C. Halling, K.D. Lindor, Utility of serum tumor markers, imaging, and biliary cytology for detecting cholangiocarcinoma in primary sclerosing cholangitis, Hepatology 48 (2008) 1106–1117. [31] J.G. Hall, T.N. Pappas, Current management of biliary strictures, J. Gastrointest. Surg. 8 (2004) 1098–1110. [32] J. Hyman, S.P. Wilczynski, R.E. Schwarz, Extrahepatic bile duct stricture and elevated CA 19-9: malignant or benign? South. Med. J. 96 (2003) 89–92. [33] VV.AA., Management of Gastric Cancer, Intech, Rijeka, Croatia, 2011. [34] C.Y. Chen, Y.T. Kuo, C.H. Lee, T.J. Hsieh, C.M. Jan, T.S. Jaw, W.T. Huang, F.J. Yu, Differentiation between malignant and benign gastric ulcers: CT virtual gastroscopy versus optical gastroendoscopy, Radiology 252 (2009) 410–417.

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002 [35] C.Y. Chen, T.S. Jaw, Y.T. Kuo, J.S. Hsu, G.C. Liu, Differentiation of gastric ulcers with MDCT, Abdom. Imaging 32 (2007) 688–693. [36] R. Lomba-Viana, M. Dinis-Ribeiro, F. Fonseca, A.S. Vieira, M.J. Bento, H. Lomba-Viana, Serum pepsinogen test for early detection of gastric cancer in a European country, Eur. J. Gastroenterol. Hepatol. 24 (2012) 37–41. [37] P.D. McLaughlin, A. Filippone, M.M. Maher, Neoplastic diseases of the peritoneum and mesentery, AJR Am. J. Roentgenol. 200 (2013) W420–W430. [38] C. Pfannenberg, N.F. Schwenzer, B.L. Bruecher, State-of-the-art imaging of peritoneal carcinomatosis, Rofo 184 (2012) 205–213. [39] M. Haslinger, V. Francescutti, K. Attwood, J.A. McCart, M. Fakih, J.M. Kane III, J.J. Skitzki, A contemporary analysis of morbidity and outcomes in cytoreduction/hyperthermic intraperitoneal chemoperfusion, Cancer Med. 2 (2013) 334–342. [40] K.I. Votanopoulos, N.A. Newman, G. Russell, C. Ihemelandu, P. Shen, J.H. Stewart, E.A. Levine, Outcomes of cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC) in patients older than 70 years; survival benefit at considerable morbidity and mortality, Ann. Surg. Oncol. 20 (2013) 3497–3503. [41] K.I. Votanopoulos, K. Swett, A.U. Blackham, C. Ihemelandu, P. Shen, J.H. Stewart, E.A. Levine, Cytoreductive surgery with hyperthermic intraperitoneal chemotherapy in peritoneal carcinomatosis from rectal cancer, Ann. Surg. Oncol. 20 (2013) 1088–1092. [42] R.O.S. Karoo, T.D.R. Lloyd, G. Garcea, H.D. Redway, G.S.R. Robertson, How valuable is ascitic cytology in the detection and management of malignancy? Postgrad. Med. J. 79 (2003) 292–294. [43] R. Jha, H.G. Shrestha, G. Sayami, S.B. Pradhan, Study of effusion cytology in patients with simultaneous malignancy and ascites, Kathmandu Univ. Med. J. (KUMJ) 4 (2006) 483–487. [44] S. Warnakulasuriya, Global epidemiology of oral and oropharyngeal cancer, Oral Oncol. 45 (2009) 309–316. [45] B.W. Neville, T.A. Day, Oral cancer and precancerous lesions, CA Cancer J. Clin. 52 (2002) 195–215. [46] D. Kademani, Oral cancer, Mayo Clin. Proc. 82 (2007) 878–887. [47] VV.AA., Oral Cancer, Intech, Rijeka, Croatia, 2012. [48] R.M. Logan, A.N. Goss, Biopsy of the oral mucosa and use of histopathology services, Aust. Dent. J. 55 (Suppl. 1) (2010) 9–13. [49] Sample Preparation in Biological Mass Spectrometry, 1st ed. Springer, New York, 2011. [50] A.L. D'Souza, J.R. Tseng, K.B. Pauly, S. Guccione, J. Rosenberg, S.S. Gambhir, G.M. Glazer, A strategy for blood biomarker amplification and localization using ultrasound, Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 17152–17157. [51] N.L. Anderson, N.G. Anderson, The human plasma proteome: history, character, and diagnostic prospects, Mol Cell Proteomics 1 (2002) 845–867. [52] R. Schiess, B. Wollscheid, R. Aebersold, Targeted proteomic strategy for clinical biomarker discovery, Mol. Oncol. 3 (2009) 33–44. [53] A. Taguchi, S.M. Hanash, Unleashing the power of proteomics to develop blood-based cancer markers, Clin. Chem. 59 (2013) 119–126. [54] M. Fountoulakis, J.F. Juranville, L. Jiang, D. Avila, D. Roder, P. Jakob, P. Berndt, S. Evers, H. Langen, Depletion of the high-abundance plasma proteins, Amino Acids 27 (2004) 249–259. [55] L. Urbas, P. Brne, B. Gabor, M. Barut, M. Strlic, T.C. Petric, A. Strancar, Depletion of high-abundance proteins from human plasma using a combination of an affinity and pseudo-affinity column, J. Chromatogr. A 1216 (2009) 2689–2694. [56] S. Surinova, R. Schiess, R. Huttenhain, F. Cerciello, B. Wollscheid, R. Aebersold, On the development of plasma protein biomarkers, J. Proteome Res. 10 (2011) 5–16. [57] Z. Cao, S. Yende, J.A. Kellum, R.A. Robinson, Additions to the human plasma proteome via a tandem MARS depletion iTRAQ-based workflow, Int. J. Proteomics 2013 (2013) 654356. [58] N.S. Barteneva, E. Fasler-Kan, M. Bernimoulin, J.N. Stern, E.D. Ponomarev, L. Duckett, I.A. Vorobjev, Circulating microparticles: square the circle, BMC Cell Biol. 14 (2013) 23. [59] T.H. Lee, E. D'Asti, N. Magnus, K. Al-Nedawi, B. Meehan, J. Rak, Microvesicles as mediators of intercellular communication in cancer-the emerging science of cellular ‘debris’, Semin. Immunopathol. 33 (2011) 455–467. [60] P. O'Mullan, D. Craft, J. Yi, C.A. Gelfand, Thrombin induces broad spectrum proteolysis in human serum samples, Clin. Chem. Lab. Med. 47 (2009) 685–693. [61] P. O'Mullan, J. Yi, N. Sengupta, C. Gelfand, Intrinsic preanalytical variability of serum samples is evidenced in peptide MALDI MS spectra, Mol. Cell. Proteomics 4 (2005) S86-S86. [62] P. Gromov, I. Gromova, C.J. Olsen, V. Timmermans-Wielenga, M.L. Talman, R.R. Serizawa, J.M. Moreira, Tumor interstitial fluid — a treasure trove of cancer biomarkers, Biochim. Biophys. Acta 1834 (2013) 2259–2270. [63] P.N. Teng, N.W. Bateman, B.L. Hood, T.P. Conrads, Advances in proximal fluid proteomics for disease biomarker discovery, J. Proteome Res. 9 (2010) 6091–6100. [64] P. McGing, R. O'Kelley, The biochemistry of body fluids, ACBI, Ireland, 2009. [65] F.M. Amado, R.P. Ferreira, R. Vitorino, One decade of salivary proteomics: current approaches and outstanding challenges, Clin. Biochem. 46 (2013) 506–517. [66] V. Thongboonkerd, Proteomics of Human Body Fluids, Humana Press, New Jersey, USA, 2007. [67] M.P. Bard, J.P. Hegmans, A. Hemmes, T.M. Luider, R. Willemsen, L.A. Severijnen, J.P. van Meerbeeck, S.A. Burgers, H.C. Hoogsteden, B.N. Lambrecht, Proteomic analysis of exosomes isolated from human malignant pleural effusions, Am. J. Respir. Cell Mol. Biol. 31 (2004) 114–121. [68] S. Runz, S. Keller, C. Rupp, A. Stoeck, Y. Issa, D. Koensgen, A. Mustea, J. Sehouli, G. Kristiansen, P. Altevogt, Malignant ascites-derived exosomes of ovarian carcinoma patients contain CD24 and EpCAM, Gynecol. Oncol. 107 (2007) 563–571. [69] M.A. Barbhuiya, N.A. Sahasrabuddhe, S.M. Pinto, B. Muthusamy, T.D. Singh, V. Nanjappa, S. Keerthikumar, B. Delanghe, H.C. Harsha, R. Chaerkady, V. Jalaj, S. Gupta, B.R. Shrivastav, P.K. Tiwari, A. Pandey, Comprehensive proteomic analysis of human bile, Proteomics 11 (2011) 4443–4453.

1001

[70] A. Farina, J.M. Dumonceau, P. Lescuyer, Proteomic analysis of human bile and potential applications for cancer diagnosis, Expert Rev. Proteomics 6 (2009) 285–301. [71] N. Lukic, R. Visentin, M. Delhaye, J.L. Frossard, P. Lescuyer, J.M. Dumonceau, A. Farina, An integrated approach for comparative proteomic analysis of human bile reveals overexpressed cancer-associated proteins in malignant biliary stenosis, Biochim. Biophys. Acta 1844 (2014) 1027–1034(this issue). [72] V.J. Patel, K. Thalassinos, S.E. Slade, J.B. Connolly, A. Crombie, J.C. Murrell, J.H. Scrivens, A comparison of labeling and label-free mass spectrometry-based proteomics approaches, J. Proteome Res. 8 (2009) 3752–3759. [73] P.-A. Clavien, J. Baillie, Diseases of the Gallbladder and Bile Ducts: Diagnosis and Treatment, 2nd ed. Blackwell Pub., Malden, Mass., 2006 [74] A. Esteller, Physiology of bile secretion, World J. Gastroenterol. 14 (2008) 5641–5649. [75] T.Z. Kristiansen, J. Bunkenborg, M. Gronborg, H. Molina, P.J. Thuluvath, P. Argani, M.G. Goggins, A. Maitra, A. Pandey, A proteomic analysis of human bile, Mol. Cell. Proteomics 3 (2004) 715–728. [76] J. Koopmann, P.J. Thuluvath, M.L. Zahurak, T.Z. Kristiansen, A. Pandey, R. Schulick, P. Argani, M. Hidalgo, S. Iacobelli, M. Goggins, A. Maitra, Mac-2-binding protein is a diagnostic marker for biliary tract carcinoma, Cancer 101 (2004) 1609–1615. [77] N. Tinari, I. Kuwabara, M.E. Huflejt, P.F. Shen, S. Iacobelli, F.T. Liu, Glycoprotein 90 K/MAC-2BP interacts with galectin-1 and mediates galectin-1-induced cell aggregation, Int. J. Cancer 91 (2001) 167–172. [78] C.Y. Chen, W.L. Tsai, H.C. Wu, M.J. Syu, C.C. Wu, S.C. Shiesh, Diagnostic role of biliary pancreatic elastase for cholangiocarcinoma in patients with cholestasis, Clin. Chim. Acta 390 (2008) 82–89. [79] A.A. Zabron, V.M. Horneffer-van der Sluis, C.A. Wadsworth, F. Laird, M. Gierula, A.V. Thillainayagam, P. Vlavianos, D. Westaby, S.D. Taylor-Robinson, R.J. Edwards, S.A. Khan, Elevated levels of neutrophil gelatinase-associated lipocalin in bile from patients with malignant pancreatobiliary disease, Am. J. Gastroenterol. 106 (2011) 1711–1717. [80] S. Chakraborty, S. Kaur, S. Guha, S.K. Batra, The multifaceted roles of neutrophil gelatinase associated lipocalin (NGAL) in inflammation and cancer, Biochim. Biophys. Acta 1826 (2012) 129–169. [81] T.O. Lankisch, J. Metzger, A.A. Negm, K. Vosskuhl, E. Schiffer, J. Siwy, T.J. Weismuller, A.S. Schneider, K. Thedieck, R. Baumeister, P. Zurbig, E.M. Weissinger, M.P. Manns, H. Mischak, J. Wedemeyer, Bile proteomic profiles differentiate cholangiocarcinoma from primary sclerosing cholangitis and choledocholithiasis, Hepatology 53 (2011) 875–884. [82] J. Shen, W. Wang, J. Wu, B. Feng, W. Chen, M. Wang, J. Tang, F. Wang, F. Cheng, L. Pu, Q. Tang, X. Wang, X. Li, Comparative proteomic profiling of human bile reveals SSP411 as a novel biomarker of cholangiocarcinoma, PLoS One 7 (2012) e47476. [83] A. Farina, J.M. Dumonceau, P. Antinori, I. Annessi-Ramseyer, J.L. Frossard, D.F. Hochstrasser, M. Delhaye, P. Lescuyer, Bile carcinoembryonic cell adhesion molecule 6 (CEAM6) as a biomarker of malignant biliary stenoses, Biochim. Biophys. Acta 1844 (2014) 1019–1026(this issue). [84] T.M. Cheng, Y.M. Murad, C.C. Chang, M.C. Yang, T.N. Baral, A. Cowan, S.H. Tseng, A. Wong, R. Mackenzie, D.B. Shieh, J. Zhang, Single domain antibody against carcinoembryonic antigen-related cell adhesion molecule 6 (CEACAM6) inhibits proliferation, migration, invasion and angiogenesis of pancreatic cancer cells, Eur. J. Cancer (2012)(this issue). [85] L. Reinhard, C. Rupp, H.D. Riedel, T. Ruppert, T. Giese, C. Flechtenmacher, K.H. Weiss, P. Kloeters-Plachky, W. Stremmel, P. Schirmacher, P. Sauer, D.N. Gotthardt, S100A9 is a biliary protein marker of disease activity in primary sclerosing cholangitis, PLoS One 7 (2012) e29821. [86] W. Wang, K.X. Ai, Z. Yuan, X.Y. Huang, H.Z. Zhang, Different expression of S100A8 in malignant and benign gallbladder diseases, Dig. Dis. Sci. 58 (2013) 150–162. [87] S.G. Farid, R.A. Craven, J. Peng, G.K. Bonney, D.N. Perkins, P.J. Selby, K. Rajendra Prasad, R.E. Banks, Shotgun proteomics of human bile in hilar cholangiocarcinoma, Proteomics 11 (2011) 2134–2138. [88] A. Farina, J.M. Dumonceau, M. Delhaye, J.L. Frossard, A. Hadengue, D.F. Hochstrasser, P. Lescuyer, A step further in the analysis of human bile proteome, J. Proteome Res. 10 (2011) 2047–2063. [89] A. Farina, J.M. Dumonceau, J.L. Frossard, A. Hadengue, D.F. Hochstrasser, P. Lescuyer, Proteomic analysis of human bile from malignant biliary stenosis induced by pancreatic cancer, J. Proteome Res. 8 (2009) 159–169. [90] M.E. Smith, D.G. Morton, The Digestive System, Churchill Livingstone, Edinburgh; New York, 2001. [91] R. Chen, S. Pan, E.C. Yi, S. Donohoe, M.P. Bronner, J.D. Potter, D.R. Goodlett, R. Aebersold, T.A. Brentnall, Quantitative proteomic profiling of pancreatic cancer juice, Proteomics 6 (2006) 3871–3879. [92] R. Chen, S. Pan, K. Cooke, K.W. Moyes, M.P. Bronner, D.R. Goodlett, R. Aebersold, T.A. Brentnall, Comparison of pancreas juice proteins from cancer versus pancreatitis using quantitative proteomic analysis, Pancreas 34 (2007) 70–79. [93] L. Zhou, Z. Lu, A. Yang, R. Deng, C. Mai, X. Sang, K.N. Faber, X. Lu, Comparative proteomic analysis of human pancreatic juice: methodological study, Proteomics 7 (2007) 1345–1355. [94] M. Tian, Y.Z. Cui, G.H. Song, M.J. Zong, X.Y. Zhou, Y. Chen, J.X. Han, Proteomic analysis identifies MMP-9, DJ-1 and A1BG as overexpressed proteins in pancreatic juice from pancreatic ductal adenocarcinoma patients, BMC Cancer 8 (2008) 241. [95] B. Sitek, B. Sipos, I. Alkatout, G. Poschmann, C. Stephan, T. Schulenborg, K. Marcus, J. Luttges, D.D. Dittert, G. Baretton, W. Schmiegel, S.A. Hahn, G. Kloppel, H.E. Meyer, K. Stuhler, Analysis of the pancreatic tumor progression by a quantitative proteomic approach and immunhistochemical validation, J. Proteome Res. 8 (2009) 1647–1656. [96] J. Gao, F. Zhu, S. Lv, Z. Li, Z. Ling, Y. Gong, C. Jie, L. Ma, Identification of pancreatic juice proteins as biomarkers of pancreatic cancer, Oncol. Rep. 23 (2010) 1683–1692. [97] S. Lv, J. Gao, F. Zhu, Z. Li, Y. Gong, G. Xu, L. Ma, Transthyretin, identified by proteomics, is overabundant in pancreatic juice from pancreatic carcinoma and originates from pancreatic islets, Diagn. Cytopathol. 39 (2011) 875–881.

1002

A. Farina / Biochimica et Biophysica Acta 1844 (2014) 988–1002

[98] Y. Shirai, K. Sogawa, T. Yamaguchi, K. Sudo, A. Nakagawa, Y. Sakai, T. Ishihara, M. Sunaga, M. Nezu, T. Tomonaga, M. Miyazaki, H. Saisho, F. Nomura, Protein profiling in pancreatic juice for detection of intraductal papillary mucinous neoplasm of the pancreas, Hepatogastroenterology 55 (2008) 1824–1829. [99] T. Marchbank, A. Mahmood, R.J. Playford, Pancreatic secretory trypsin inhibitor causes autocrine-mediated migration and invasion in bladder cancer and phosphorylates the EGF receptor, Akt2 and Akt3, and ERK1 and ERK2, Am. J. Physiol. Renal Physiol. 305 (2013) F382–F389. [100] R. Chen, S. Pan, X. Duan, B.H. Nelson, R.A. Sahota, S. de Rham, R.A. Kozarek, M. McIntosh, T.A. Brentnall, Elevated level of anterior gradient-2 in pancreatic juice from patients with pre-malignant pancreatic neoplasia, Mol. Cancer 9 (2010) 149. [101] K.E. Vanderlaag, S. Hudak, L. Bald, L. Fayadat-Dilman, M. Sathe, J. Grein, M.J. Janatpour, Anterior gradient-2 plays a critical role in breast cancer cell growth and survival by modulating cyclin D1, estrogen receptor-alpha and survivin, Breast Cancer Res. 12 (2010) R32. [102] P. Patel, C. Clarke, D.L. Barraclough, T.A. Jowitt, P.S. Rudland, R. Barraclough, L.Y. Lian, Metastasis-promoting anterior gradient 2 protein has a dimeric thioredoxin fold structure and a role in cell adhesion, J. Mol. Biol. 425 (2013) 929–943. [103] J.Y. Park, S.A. Kim, J.W. Chung, S. Bang, S.W. Park, Y.K. Paik, S.Y. Song, Proteomic analysis of pancreatic juice for the identification of biomarkers of pancreatic cancer, J. Cancer Res. Clin. Oncol. 137 (2011) 1229–1238. [104] S. Makawita, C. Smith, I. Batruch, Y. Zheng, F. Ruckert, R. Grutzmann, C. Pilarsky, S. Gallinger, E.P. Diamandis, Integrated proteomic profiling of cell line conditioned media and pancreatic juice for the identification of pancreatic cancer biomarkers, Mol. Cell. Proteomics 10 (2011)(M111 008599). [105] E. Ke, B.B. Patel, T. Liu, X.M. Li, O. Haluszka, J.P. Hoffman, H. Ehya, N.A. Young, J.C. Watson, D.S. Weinberg, M.T. Nguyen, S.J. Cohen, N.J. Meropol, S. Litwin, J.L. Tokar, A.T. Yeung, Proteomic analyses of pancreatic cyst fluids, Pancreas 38 (2009) e33–e42. [106] C.J. Scarlett, J.S. Samra, A. Xue, R.C. Baxter, R.C. Smith, Classification of pancreatic cystic lesions using SELDI-TOF mass spectrometry, ANZ J. Surg. 77 (2007) 648–653. [107] A. Cuoghi, A. Farina, K. Z'Graggen, J.M. Dumonceau, A. Tomasi, D.F. Hochstrasser, M. Genevay, P. Lescuyer, J.L. Frossard, Role of proteomics to differentiate between benign and potentially malignant pancreatic cysts, J. Proteome Res. 10 (2011) 2664–2670. [108] B.F. Mann, J.A. Goetz, M.G. House, C.M. Schmidt, M.V. Novotny, Glycomic and proteomic profiling of pancreatic cyst fluids identifies hyperfucosylated lactosamines on the N-linked glycans of overexpressed glycoproteins, Mol. Cell. Proteomics 11 (2012)(M111 015792). [109] R. Greger, U. Windhorst, Comprehensive Human Physiology: From Cellular Mechanisms to Integration, vol. (2527)Springer-Verlag, New York, 1996. 2. [110] T.C. Martinsen, K. Bergh, H.L. Waldum, Gastric juice: a barrier against infectious diseases, Basic Clin. Pharmacol. Toxicol. 96 (2005) 94–102. [111] J. Ochei, A. Kolhatkar, Medical Laboratory Science: Theory and Practice, Tata McGrow-Hill, 2000. [112] W. Wu, M.C. Chung, The gastric fluid proteome as a potential source of gastric cancer biomarkers, J. Proteomics 90 (2013) 3–13. [113] K. Lee, M. Kye, J.S. Jang, O.J. Lee, T. Kim, D. Lim, Proteomic analysis revealed a strong association of a high level of alpha1-antitrypsin in gastric juice with gastric cancer, Proteomics 4 (2004) 3343–3352. [114] P.I. Hsu, C.H. Chen, C.S. Hsieh, W.C. Chang, K.H. Lai, G.H. Lo, P.N. Hsu, F.W. Tsay, Y.S. Chen, M. Hsiao, H.C. Chen, P.J. Lu, Alpha1-antitrypsin precursor in gastric juice is a novel biomarker for gastric cancer and ulcer, Clin. Cancer Res. 13 (2007) 876–883. [115] P.I. Hsu, C.H. Chen, M. Hsiao, D.C. Wu, C.Y. Lin, K.H. Lai, P.J. Lu, Diagnosis of gastric malignancy using gastric juice alpha1-antitrypsin, Cancer Epidemiol. Biomarkers Prev. 19 (2010) 405–411. [116] S.M. Ganji, F.R. Jazii, A. Sahebghadam-Lotfi, Structural features, biological functions of the alpha-1 antitrypsin and contribution to esophageal cancer, in: X. Li (Ed.), Squamous Cell Carcinoma, InTech, 2012. [117] Y.H. Chang, S.H. Lee, I.C. Liao, S.H. Huang, H.C. Cheng, P.C. Liao, Secretomic analysis identifies alpha-1 antitrypsin (A1AT) as a required protein in cancer cell migration, invasion, and pericellular fibronectin assembly for facilitating lung colonization of lung adenocarcinoma cells, Mol. Cell. Proteomics 11 (2012) 1320–1339. [118] W.C. Chang, P.I. Hsu, Y.Y. Chen, M. Hsiao, P.J. Lu, C.H. Chen, Observation of peptide differences between cancer and control in gastric juice, Proteomics Clin. Appl. 2 (2008) 55–62. [119] O.L. Kon, T.T. Yip, M.F. Ho, W.H. Chan, W.K. Wong, S.Y. Tan, W.H. Ng, S.Y. Kam, A. Eng, P. Ho, R. Viner, H.S. Ong, M.P. Kumarasinghe, The distinctive gastric fluid proteome in gastric cancer reveals a multi-biomarker diagnostic profile, BMC Med. Genomics 1 (2008) 54. [120] W. Wu, W.C. Juan, C.R. Liang, K.G. Yeoh, J. So, M.C. Chung, S100A9, GIF and AAT as potential combinatorial biomarkers in gastric cancer diagnosis and prognosis, Proteomics Clin. Appl. 6 (2012) 152–162. [121] C. Gebhardt, J. Nemeth, P. Angel, J. Hess, S100A8 and S100A9 in inflammation and cancer, Biochem. Pharmacol. 72 (2006) 1622–1631. [122] E. Kipps, D.S. Tan, S.B. Kaye, Meeting the challenge of ascites in ovarian cancer: new avenues for therapy and research, Nat. Rev. Cancer 13 (2013) 273–282. [123] S.L. Sangisetty, T.J. Miner, Malignant ascites: a review of prognostic factors, pathophysiology and therapeutic measures, World J. Gastrointest. Surg. 4 (2012) 87–95. [124] L. Gortzak-Uzan, A. Ignatchenko, A.I. Evangelou, M. Agochiya, K.A. Brown, P. St Onge, I. Kireeva, G. Schmitt-Ulms, T.J. Brown, J. Murphy, B. Rosen, P. Shaw, I. Jurisica, T. Kislinger, A proteome resource of ovarian cancer ascites: integrated proteomic and bioinformatic analyses to identify putative biomarkers, J. Proteome Res. 7 (2008) 339–351. [125] C. Kuk, V. Kulasingam, C.G. Gunawardana, C.R. Smith, I. Batruch, E.P. Diamandis, Mining the ovarian cancer ascites proteome for potential ovarian cancer biomarkers, Mol. Cell. Proteomics 8 (2009) 661–669. [126] S. Elschenbroich, V. Ignatchenko, B. Clarke, S.E. Kalloger, P.C. Boutros, A.O. Gramolini, P. Shaw, I. Jurisica, T. Kislinger, In-depth proteomics of ovarian cancer ascites:

[127]

[128]

[129] [130]

[131]

[132] [133]

[134]

[135] [136] [137]

[138]

[139]

[140]

[141]

[142]

[143] [144]

[145] [146]

[147]

[148]

[149]

[150] [151]

[152]

[153] [154] [155] [156] [157]

combining shotgun proteomics and selected reaction monitoring mass spectrometry, J. Proteome Res. 10 (2011) 2286–2299. G. Hariprasad, R. Hariprasad, L. Kumar, A. Srinivasan, S. Kola, A. Kaushik, Apolipoprotein A1 as a potential biomarker in the ascitic fluid for the differentiation of advanced ovarian cancers, Biomarkers 18 (2013) 532–541. D.S. Choi, J.O. Park, S.C. Jang, Y.J. Yoon, J.W. Jung, D.Y. Choi, J.W. Kim, J.S. Kang, J. Park, D. Hwang, K.H. Lee, S.H. Park, Y.K. Kim, D.M. Desiderio, K.P. Kim, Y.S. Gho, Proteomic analysis of microvesicles derived from human colorectal cancer ascites, Proteomics 11 (2011) 2745–2751. H. Kosanam, S. Makawita, B. Judd, A. Newman, E.P. Diamandis, Mining the malignant ascites proteome for pancreatic cancer biomarkers, Proteomics 11 (2011) 4551–4558. M. Greabu, M. Battino, M. Mohora, A. Totan, A. Didilescu, T. Spinu, C. Totan, D. Miricescu, R. Radulescu, Saliva—a diagnostic window to the body, both in health and in disease, J. Med. Life 2 (2009) 124–132. V. de Almeida Pdel, A.M. Gregio, M.A. Machado, A.A. de Lima, L.R. Azevedo, Saliva composition and functions: a comprehensive review, J. Contemp. Dent. Pract. 9 (2008) 72–80. S. Chiappin, G. Antonelli, R. Gatti, E.F. De Palo, Saliva specimen: a new laboratory tool for diagnostic and basic investigation, Clin. Chim. Acta 383 (2007) 30–40. S. Bandhakavi, M.D. Stone, G. Onsongo, S.K. Van Riper, T.J. Griffin, A dynamic range compression and three-dimensional peptide fractionation analysis platform expands proteome coverage and the diagnostic potential of whole saliva, J. Proteome Res. 8 (2009) 5590–5600. S. Hu, M. Arellano, P. Boontheung, J. Wang, H. Zhou, J. Jiang, D. Elashoff, R. Wei, J.A. Loo, D.T. Wong, Salivary proteomics for oral cancer biomarker discovery, Clin. Cancer Res. 14 (2008) 6246–6252. T.D. Oberley, L.W. Oberley, Antioxidant enzyme levels in cancer, Histol. Histopathol. 12 (1997) 525–535. Z. Ding, P. Roy, Profilin-1 versus profilin-2: two faces of the same coin? Breast Cancer Res. 15 (2013) 311. Y.J. Jou, C.D. Lin, C.H. Lai, C.H. Chen, J.Y. Kao, S.Y. Chen, M.H. Tsai, S.H. Huang, C.W. Lin, Proteomic identification of salivary transferrin as a biomarker for early detection of oral cancer, Anal. Chim. Acta. 681 (2010) 41–48. E.P. de Jong, H. Xie, G. Onsongo, M.D. Stone, X.B. Chen, J.A. Kooren, E.W. Refsland, R.J. Griffin, F.G. Ondrey, B. Wu, C.T. Le, N.L. Rhodus, J.V. Carlis, T.J. Griffin, Quantitative proteomics reveals myosin and actin as promising saliva biomarkers for distinguishing pre-malignant and malignant oral lesions, PLoS One 5 (2010) e11148. H. He, G. Sun, F. Ping, Y. Cong, A new and preliminary three-dimensional perspective: proteomes of optimization between OSCC and OLK, Artif. Cells Blood Substit. Immobil. Biotechnol. 39 (2011) 26–30. Y.J. Jou, C.D. Lin, C.H. Lai, C.H. Tang, S.H. Huang, M.H. Tsai, S.Y. Chen, J.Y. Kao, C.W. Lin, Salivary zinc finger protein 510 peptide as a novel biomarker for detection of oral squamous cell carcinoma in early stages, Clin. Chim. Acta 412 (2011) 1357–1365. O. Brinkmann, D.A. Kastratovic, M.V. Dimitrijevic, V.S. Konstantinovic, D.B. Jelovac, J. Antic, V.S. Nesic, S.Z. Markovic, Z.R. Martinovic, D. Akin, N. Spielmann, H. Zhou, D.T. Wong, Oral squamous cell carcinoma detection by salivary biomarkers in a Serbian population, Oral Oncol. 47 (2011) 51–55. T. Jarai, G. Maasz, A. Burian, A. Bona, E. Jambor, I. Gerlinger, L. Mark, Mass spectrometry-based salivary proteomics for the discovery of head and neck squamous cell carcinoma, Pathol. Oncol. Res. 18 (2012) 623–628. A. Zhang, H. Sun, P. Wang, X. Wang, Salivary proteomics in biomedical research, Clin. Chim. Acta 415 (2013) 261–265. J.Y. Wu, C. Yi, H.R. Chung, D.J. Wang, W.C. Chang, S.Y. Lee, C.T. Lin, Y.C. Yang, W.C. Yang, Potential biomarkers in saliva for oral squamous cell carcinoma, Oral Oncol. 46 (2010) 226–231. N. Spielmann, D.T. Wong, Saliva: diagnostics and therapeutic perspectives, Oral Dis. 17 (2011) 345–354. M.E. Arellano-Garcia, R. Li, X. Liu, Y. Xie, X. Yan, J.A. Loo, S. Hu, Identification of tetranectin as a potential biomarker for metastatic oral cancer, Int. J. Mol. Sci. 11 (2010) 3106–3121. S. Shintani, H. Hamakawa, Y. Ueyama, M. Hatori, T. Toyoshima, Identification of a truncated cystatin SA-I as a saliva biomarker for oral squamous cell carcinoma using the SELDI ProteinChip platform, Int. J. Oral Maxillofac. Surg. 39 (2010) 68–74. A.M. Contucci, R. Inzitari, S. Agostino, A. Vitali, A. Fiorita, T. Cabras, E. Scarano, I. Messana, Statherin levels in saliva of patients with precancerous and cancerous lesions of the oral cavity: a preliminary report, Oral Dis. 11 (2005) 95–99. A. Vidotto, T. Henrique, L.S. Raposo, J.V. Maniglia, E.H. Tajara, Salivary and serum proteomics in head and neck carcinomas: before and after surgery and radiotherapy, Cancer Biomark. 8 (2010) 95–107. Z.Z. Wu, J.G. Wang, X.L. Zhang, Diagnostic model of saliva protein finger print analysis of patients with gastric cancer, World J. Gastroenterol. 15 (2009) 865–870. H. Xiao, L. Zhang, H. Zhou, J.M. Lee, E.B. Garon, D.T. Wong, Proteomic analysis of human saliva from lung cancer patients using two-dimensional difference gel electrophoresis and mass spectrometry, Mol. Cell. Proteomics 11 (2012)(M111 012112). L. Zhang, H. Xiao, S. Karlan, H. Zhou, J. Gross, D. Elashoff, D. Akin, X. Yan, D. Chia, B. Karlan, D.T. Wong, Discovery and preclinical validation of salivary transcriptomic and proteomic biomarkers for the non-invasive detection of breast cancer, PLoS One 5 (2010) e15573. E.P. Diamandis, Cancer biomarkers: can we turn recent failures into success? J. Natl. Cancer Inst. 102 (2010) 1462–1467. E. Drucker, K. Krapfenbauer, Pitfalls and limitations in translation from biomarker discovery to clinical utility in predictive and personalised medicine, EPMA J. 4 (2013) 7. G. Poste, Bring on the biomarkers, Nature 469 (2011) 156–157. E.P. Diamandis, The failure of protein cancer biomarkers to reach the clinic: why, and what can be done to address the problem? BMC Med. 10 (2012) 87. D.A. Morrow, J.A. de Lemos, Benchmarks for the assessment of novel cardiovascular biomarkers, Circulation 115 (2007) 949–952.

Proximal fluid proteomics for the discovery of digestive cancer biomarkers.

Most digestive malignancies have asymptomatic course, often progressing to poor outcome stages. Surgical resection usually represents the only potenti...
937KB Sizes 0 Downloads 0 Views