Extracellular heat shock proteins and cancer: New perspectives.

Quantitative MS-based proteomics coupled with machine learning (ML) statistics is a novel and promising approach for identification of suitable biomarkers in EVs derived from biological fluids of cancer patients [ 22 , 305 ]. Unlike classical statistics, which identifies only differentially expressed proteins, ML performs simultaneous analysis of all proteins in a sample as well as analyzes relationships of these proteins across different patients, capturing all changes in a protein level. Based on our analysis of HSP profiling, we propose the following scheme for the identification of potential HSP candidates ( Fig. 2 ).Extracellular vesicles derived from urine and NK cells showed to be the most abundant source of HSPs and co-chaperones and, therefore, can be a good source for identifying HSP signatures. Regarding a class of EVs, exosomes can be the most useful source for identification of cancer-specific biomarkers as they are formed by inward budding of endosomal membranes and, thus, reflect the protein composition of plasma membranes of the cells they were derived from [229] . Fig. 2 HSP profiling of exosomes for identification of potential HSP-based biomarker candidates. The quantitative MS-based proteomics coupled with machine learning might be used to detect all changes in the level of expression in a sample and between cohorts of patients for identification of clinically useful HSP- based biomarkers in urine and/or NK- derived exosomes of cancer patients. Fig. 2 HSP profiling of exosomes for identification of potential HSP-based biomarker candidates. The quantitative MS-based proteomics coupled with machine learning might be used to detect all changes in the level of expression in a sample and between cohorts of patients for identification of clinically useful HSP- based biomarkers in urine and/or NK- derived exosomes of cancer patients. ML algorithms can be used to classify patients by the type of cancer in which specific proteins may contribute more or less to the cancer prediction model. Various types of cancers may have specific patterns of upregulated and downregulated HSPs that may distinguish them from healthy individuals and from patients with other non-cancerous conditions or even discriminate one type of cancer from another. Since proteins that constitute HSP networks may behave differently in different types of cancer and control groups, ML may be used to identify critical relationships between HSP networks and whether one network contribute more to the prediction of cancer than the other network. Furthermore, monitoring the changes in the level of HSPs in response to therapies may provide a further clues on how different types of therapies affect HSP release and how this is related to the outcome of the patient, which may provide further improvement towards the more personalized approach in treating cancer patients. It will be also of interest to determine specific threshold level of specific protein so that elevated or reduced level of one or multiple proteins is better at predicting cancer. The use of clinical data is advantageous as it may provide a great insight into the understanding of the role of HSPs in cancer. Use of ML that can analyze big data and find certain patterns opens new possibilities for the discovery of specific signatures of cancer and development of more efficient anti-cancer therapies.

Another class of potential biomarkers abundantly present in circulation are small noncoding microRNAs (miRNAs) [259] . MiRNAs play important role in the regulation of gene expression [ 260 , 261 ]. Specifically, complementary base-pairing of miRNA with target messenger RNA (mRNA) or partial binding of miRNA to mRNA results in mRNA cleavage and translation repression, respectively [ 260 , 262 , 263 ]. Several studies reported altered expression of miRNAs in serum and plasma of cancer patients, making them potentially useful for identification of blood-based biomarkers [ 259 , [264] , [265] , [266] ]. Along this line, several studies have used specific miRNA signatures to identify tissue of origin of metastatic cancer and distinguish patients with different subtypes of cancer. For example, Feraccin and co-workers used tissue samples from primary and metastatic tumors to identify cancer-type specific 47-miRNA signature for predicting primary sites of metastatic cancers [267] . In another study, Yousseff and colleagues have used specific miRNA patterns to distinguish patients with different subtypes of renal cell carcinoma [268] . Furthermore, aberrant expression of certain miRNAs showed to indicate an early event in the development of pancreatic ductal adenocarcinoma (PDAC) precursor lesions, suggesting the use of miRNAs in early diagnosis [269] . In addition to the use of miRNAs in diagnosis, several studies used miRNAs as prognostic biomarkers of cancer. As an example, high expression of members of miR-183 family in tumor and serum showed to associate with poor outcome in patients with lung cancer [270] . Promisingly, it was also observed that miRNAs can be used to predict response to therapy in cancer patients [271] . As an example, highly expressed miR-21 associated with shorter overall survival in PDAC patients treated with gemcitabine [272] . Additionally, potential use of circulating miRNA as biomarkers showed to be particularly relevant for lymphoma patients (reviewed in [259] ). Taking into account that miRNAs may deregulate HSPs and serve as potential predictive and prognostic biomarkers, it is important to identify miRNAs that target HSP networks [273] . Deregulated miRNAs in EVs obtained from clinical samples of cancer patients and their experimentally validated HSP targets are summarized in Table 5 ( Fig. 1 ) [274] . Table 5 Micro-RNAs and their HSP targets in EVs derived from cancer patients. Table 5 MicroRNA Target gene Sample Cancer Refs. ↑ miR-205–5p HSPA8; DNAJA1 Serum EV Ovarian [289] ↑ miR-214–3p HSP90AB1;HSPD1 Serum EV Ovarian [289] ↑ miR-214–5p HSPA2;HSPA14; BAG2 Serum EV Ovarian [289] ↑ miR-495–3p HSPA5; HSPA1B; HSP90AA1; DNAJC21 Plasma EV Lung [290] ↓ miR-411–5p DNAJB9;DNAJC10;ST13 Plasma EV Lung [290] ↓ miR-615–3p HSPA1B;HSPA4;CCT8;HSPA8;HSPA9;CCT1;HSP90AB1;DNAJA2;DNAJC9;DNAJC10; PPIF;CDC37;TRAP1;BAG6;BAG5;CCT7 Plasma EV ccRCC [291] ↑ miR-224–5p HSP90AA1 Serum EV ccRCC [292] ↑ miR-224–3p DNAJB4; DNAJC28 Serum EV ccRCC [292] ↑ miR-375 HSP90AA1;DNAJC8; DNAJC2 Urine EV Prostate [ 293 , 294 ] ↑ Let-7c-5p HSPA4;DNAJC6;DNAJC16; DNAJC28 Urine EV Prostate [294] ↑ Let-7c-3p DNAJB6; BAG4 Urine EV Prostate [294] ↓ miR-196a-5p HSPA4L; TRAP1 Urine EV Prostate [295] ↑ miR-326 HSPA1B Plasma EV Prostate [296] ↑ miR-331–3p HSPA1B; HSPA8;CCT1; HSPD1;TRAP1;BAG6 Plasma EV Prostate [296] ↑ miR-141 HSPA4;DNAJC28;HSPA2;HSP90AA1; DNAJB2 Plasma EV; Serum EV Prostate [ 296 , 297 ] ↑ miR-301a-3p HSPA8;CCT6A;CCT8 Plasma EV Prostate [296] ↑ miR-130b-3p HSPA8;HSP90B1;CCT6A Plasma EV Prostate [296] ↑ miR-130b-5p HSPA1B; HSPA6; HSPA8;BAG2 Plasma EV Prostate [296] ↑ miR-1233–5p HSP90AB1;DNAJC8;DNAJC24; DNAJC10; Serum EV ccRCC [298] ↑ miR-432–5p DNAJB6 Plasma EV Prostate [296] ↑ miR-107 DNAJA1;DNAJC10 Plasma EV Prostate [296] ↑ miR-200c-3p DNAJB9;DNAJC3;DNAJB6;FKBP5;BAG4 Plasma EV Prostate [299] ↑ miR-1290 DNAJB9 Plasma EV Prostate [300] ↑ miR-375 HSP90AA1;DNAJC8; DNAJC2 Plasma EV Prostate [ 299 , 300 ] ↑ miR-155–5p HSPA4L; HSPB11; DNAJC2;DNAJC19; CCT2; DNAJB1; BAG5;CDC37 Plasma EV HL [301] ↑ miR-15a-5p HSPA1A; HSPA1B; HSPA8; HSP90B1; DNAJA1; DNAJC9;DNAJC10;BAG4;CCT6B Serum EV DLBCL [302] ↑ miR-125b-5p HSPA1B;HSPD1;HSPBP1 Serum EV DLBCL [303] ↑ miR-21–3p DNAJC10;ST13 Serum EV DLBCL [304] ↑ miR-21–5p DNAJC16;PPIF; FKBP5;STUB1 Plasma EV HL [301] ↑ Let-7a-5p DNAJC6; DNAJC28; FKBP8 Plasma EV HL [301] ↑ miR-24–3p DNAJC3; DNAJC10; DNAJB12 Plasma EV HL [301] ↑ miR-99a-5p DNAJA3; FKBP5 Serum EV DLBCL [303] ↑ miR-127–3p BAG5 Plasma EV HL [301] ссRCC, Clear cell renal cell carcinoma; HL, Hodgkin lymphoma; DLBCL, Diffuse large B-cell lymphoma. Fig. 1 Deregulated miRNAs targeting HSPs in cancer . Graphical representation of deregulated miRNAs in various types of cancer. MicroRNAs regulate the expression of major members of HSP70 and HSP90 networks ( Table 5 ). Fig. 1 Micro-RNAs and their HSP targets in EVs derived from cancer patients. ссRCC, Clear cell renal cell carcinoma; HL, Hodgkin lymphoma; DLBCL, Diffuse large B-cell lymphoma. Deregulated miRNAs targeting HSPs in cancer . Graphical representation of deregulated miRNAs in various types of cancer. MicroRNAs regulate the expression of major members of HSP70 and HSP90 networks ( Table 5 ). MicroRNAs are protected from degradation by external RNAses when either encapsulated into EVs, bound to argonaute 2 (Ago2) or transported by high-density lipoproteins (HDL) [ 259 , [275] , [276] , [277] ]. Notably, during translational stress HSP90 together with its co-chaperones Cdc37, p23, Aha1 and STIP1/HOP showed to recruit Ago2 to stress granules [278] . Inhibition of HSP90 affects miRNA-gene silencing function of Argonaute [278] . Importantly, HSP90 are required for stable binding of Argonaute to Dicer, enzyme responsible for processing of small interfering RNAs and miRNAs in the cytoplasm for their further loading on RNA-induced silencing complex (RISC) [279] , [280] , [281] . HSP70 also showed to be associated with Ago2 following T-cell receptor (TCR) activation for the regulation of T-helper 17 (Th17) cells via miRNA expression in EL-4 murine lymphoma cells [282] . miR-21 showed to modulate resistance of cholangiocarcinoma cells to HSP90 inhibitors via DNAJB5 [283] . Along this line, miR-361 binds and regulates expression of HSP90α/HSP90AA1, leading to the suppression of epithelial-mesenchymal transition in cervical cancer lines [ 283 , 284 ]. miR-27a downregulates the expression of HSP90, resulting in further degradation of HSP90 client proteins in esophageal squamous cell carcinoma cell lines [285] . Stope and co-workers showed that HSP27/HSPB1 inhibits miR-1 in prostate cancer cells [286] . Interestingly, HSP70 inhibitor triptolide showed to upregulate expression of miR-142–3p , consequently inhibiting proliferation of pancreatic ductal adenocarcinoma cells [287] . Additionally, inhibition of miR-29a induces apoptosis by increasing expression of HSP60 and decreasing HSP90, HSP70, HSP27 and HSP40 levels in breast cancer cells [288] . Conclusively, miRNAs play important role in HSP regulation, therefore, elucidating the effects of miRNAs on HSPs may open new perspectives for diagnosis and treatment of cancer patients.

Zarema Albakova : Conceptualization, Formal analysis, Investigation, Data Curation, Writing – Original draft preparation, Writing - Reviewing & Editing, Visualization. Mohammad Kawsar Sharif Siam & Pradeep Kumar Sacitharan: Writing - Reviewing & Editing. Rustam H. Ziganshin, Dmitriy Y. Ryazantsev, Alexander M. Sapozhnikov : Resources.

Heat shock protein family constitutes a large network of molecular chaperones classified into several families with aberrant expression in cancer. Profiling of HSPs in EVs derived from various biological fluids may serve useful for diagnosis, prediction and treatment of cancer patients. More importantly, understanding relationships between different HSP networks and co-chaperones may form the basis for identification of multiple HSP-based biomarkers. In this review, we summarize mass spectrometry studies of EVs derived from various liquid biopsies and different types of immune cells with particular interest in HSP profiling. In addition, we have emphasized emerging role of circulating miRNAs in regulation of HSPs in the context of cancer. Further elucidating role of HSPs in EVs from proteomic and miRNAs perspectives may provide new opportunities for the discovery of clinically useful biomarkers and novel targets for cancer therapies.

Heat shock proteins (HSPs) are evolutionally conserved and ubiquitously expressed molecular chaperones abundantly present in cancer [1] , [2] , [3] . Mammalian HSPs are divided into families, namely HSP70/ HSPA, HSP90/HSPC, HSP40/DNAJ, HSPB, HSP110/HSPH and chaperonins [4] . Several research groups showed that naturally occurring antitumor agents such as geldanamycin and radicicol exert their action by inhibiting ATPase activity of HSP90 chaperone [5] , [6] , [7] , [8] . This finding has led to the development of various HSP90 inhibitors ( Table 1 ). Subsequently, the ability of HSP70 to promote immune responses stimulated the development of HSP-based cancer vaccines [ 9 , 10 ]. Currently, various HSP-based therapies and several HSP-based biomarkers are assessed in clinical trials ( Table 1 ). Table 1 HSPs in cancer clinical trials. Table 1 Intervention HSPs Type of cancer Study Refs. HSP inhibitors AUY922 (Luminespib) HSP90 • Advanced or metastatic adenocarcinoma of the stomach or gastroesophageal junction Phase Ib [23] • Myeloproliferative Neoplasms Phase II [24] • Advanced solid tumors Phase I [25] • Lymphoma Phase II [26] • Non-small cell lung cancer (NSCLC) PhaseI/II [27] , [28] • Gastrointestinal Stromal Tumor Phase II [29] • Advanced Anaplastic lymphoma kinase (ALK)-positive NSCLC Phase II [30] • NSCLC Phase II [31] • Breast cancer Phase I/II [32] • Multiple myeloma Phase I/II [33] • Metastatic colorectal cancer Phase I [34] HS-201 HSP90 Solid tumor Phase I [35] HS-196 HSP90 Solid tumor Phase I [36] STA-9090 (Ganetespib) HSP90 • SCLC Phase II [37] • Acute Myeloid Leukemia, Acute Lymphoblastic Leukemia and Blast-phase Chronic Myelogenous Leukemia Phase I [38] • Acute Myeloid Leukemia; High Risk Myelodysplastic syndrome Phase I/II [39] • Ocular melanoma Phase II [40] • Prostate cancer Phase II [41] • NSCLC Phase II [42] • Solid tumors Phase I [43] , [44] • Hematologic malignancies Phase I [45] • Ovarian cancer Phase II [46] • Platinum-resistant ovarian cancer Phase I/II [47] • Melanoma Phase II [48] • Metastatic human epidermal growth factor receptor 2-positive breast cancer Phase I [49] • Breast cancer Phase II [50] • Advanced Gastrointestinal Carcinomas; Non-Squamous NSCLC; Urothelial Carcinomas; Sarcomas Phase I [51] • Malignant Peripheral Nerve Sheath Tumors Phase I/II [52] HSP990 HSP90 Advanced solid tumors Phase I [53] , [54] TAS-116 HSP90 Advanced solid tumors Phase I [55] KOS-953 (Tanespimycin) HSP90 Multiple myeloma Phase III [56] CNF2024 (BIIB021) HSP90 B-cell Chronic Lymphocytic Leukemia Phase I [57] NVP-BEP800 HSP90 Acute T Lymphoblastic Leukemia; Acute B Lymphoblastic Leukemia Pilot [58] AT13387 (Onalespib) HSP90 • Prostate cancer Phase I/II [59] • Advanced solid tumor; Recurrent ovarian, fallopian tube, primary peritoneal or triple negative breast cancer Phase I [60] • Solid tumors Phase I [61] • Non-small cell lung cancer Phase I/II [62] • Advanced triple negative breast cancer Phase I [63] • Solid tumors Phase I [64] IPI-504 (Retaspimycin) HSP90 • Non-small cell lung cancer • Metastatic melanoma • Advanced breast cancer Phase II Phase II Phase I/II [65] [66] [67] Alvespimycin HSP90 • Relapsed chronic lymphocytic leukemia; small lymphocytic lymphoma; B-cell prolymphocytic leukemia Phase I [68] • Lymphoma; Small Intestine Cancer; Solid tumors Phase I [69] Debio 0932 HSP90 • Advanced solid tumors; Lymphoma • Non-small cell lung cancer Phase I Phase I [70] [71] IPI-493 HSP90 Hematologic malignancies Phase I [72] PU-H71 HSP90 Solid tumors; Non-Hodgkin's Lymphoma Phase I [73] DS-2248 HSP90 Advanced solid tumors Phase I [74] MPC-3100 HSP90 Refractory or relapsed cancer Phase I [75] CNF1010 HSP90 B-cell positive chronic lymphocytic leukemia Phase I [76] SNX-5422 HSP90 • Refractory solid tumor; Non-Hodgkin's lymphoma Phase I [77] • Hematologic malignancies Phase I [78] • Neuroendocrine tumors Phase I [79] • Lung adenocarcinoma Phase I [80] • Chronic lymphocytic leukemia Phase I [81] , [82] Minnelide HSP70 • Acute Myeloid Leukemia Phase I [83] OGX-427 HSP27 • Prostate cancer; Ovarian cancer; Breast cancer; bladder cancer; NSCLC Phase I [84] , [85] • Bladder cancer Phase I [86] • Castration resistant prostate cancer Phase II [87] HSP-based vaccines HSP70 vaccine HSP70 Chronic Myelogenous Leukemia Phase I [88] HSP70 vaccine HSP70 Breast Neoplasms Phase I/II [89] HSPPC-96 vaccine (Vitespen) HSP90B1 (gp96) • High-grade glioma Phase I [90] • Recurrent or progressive high-grade glioma Phase I/II [91] • Malignant melanoma Phase III [92] • Recurrent glioblastoma Phase II [93] • Lymphoma Phase II [94] • Pancreatic cancer Phase I [95] • Recurrent soft tissue sarcoma Phase II [96] • Liver cancer Phase II/III [97] • Glioblastoma Phase II [98] , [101] • Liver cancer; Pancreatic adenocarcinoma Phase I/II [99] • Gastric carcinoma Phase I [100] HSP110-gp100 chaperone complex vaccine HSP110 Advanced melanoma Phase I [102] HspE7 (SGN-00101) HSP65 * Cervical intraepithelial neoplasia Phase II [103] pNGVL4a-Sig/E7(detox)/HSP70 DNA vaccine HSP70 Cervical cancer Phase I/II [104] Gp100 fused with OVA BiP and HSP70 HSP70 Melanoma Phase I [105] HSP-based prognostic biomarkers HSP110DE9 HSP110 Colorectal cancer Pilot [106] HSP-based predictive biomarkers HSP27 HSP27 Androgen-independent prostate cancer Pilot [107] HSP90 HSP90 Peritoneal carcinomatosis Pilot [108] HSP70 HSP70 Melanoma Phase I [109] ⁎ derived from Mycobacterium bovis Bacillus Calmette-Guerin. HSPs in cancer clinical trials. derived from Mycobacterium bovis Bacillus Calmette-Guerin. Researchers studying the role of HSPs in cancer proposed that malignant cells are “addicted to chaperones”, particularly emphasizing the importance of three well-studied HSPs such as HSP70, HSP27 and HSP90 in tumor development [ 11 , 12 ]. Indeed, overexpression of HSPs provides selective advantage to malignant cells by inhibiting apoptosis, promoting tumor metastasis and regulating immune responses [ 1 , [13] , [14] , [15] , [16] ]. Considering critical role of HSPs in cancer, several studies have proposed HSPs as potential biomarkers for cancer diagnosis [ 17 , 18 ]. Controversially, researchers studying the HSP expression in various types of cancers have observed that HSPs are overexpressed in various types of tumors [1] . Thus, they concluded that HSPs might not be informative as tissue- restricted molecular markers, but rather can be used to identify the degree of differentiation and aggressiveness in some cancers [1] . Recent studies using gene expression have shown that HSPs can be used as prognostic markers to predict clinical outcome of breast cancer patients [19] . Furthermore, the discovery of the role of extracellular vesicles (EVs) in transferring proteomic and genetic information (mRNAs, microRNAs and DNA) and identification of HSPs in EVs have opened new opportunities and challenges for determining clinical biomarkers of cancer [20] . EVs are phospholipid bilayer- encapsulated particles released by many types of cells [21] . Advances in mass spectrometry (MS) provided the opportunity to identify and quantify thousands of EV cargo proteins, allowing for the search of potential biomarkers in different types of cancer (reviewed in [22] ). The main focus of this review is extracellular HSPs and microRNAs regulating HSPs in EVs as potential biomarkers for cancer diagnosis. We discuss expression of various HSP members in EVs derived from different biological fluids and patient samples and their use as biomarkers for cancer diagnosis. We will also provide novel perspective of using HSP profiling of EVs released by immune cells as clinical biomarkers and potential targets for investigational therapies. This review may help to better understand the role of HSP chaperones in EVs and their clinical importance for the identification of new cancer biomarkers and therapeutic targets.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Plasma membrane plays an important role in HSP expression and localization [110] . Several studies demonstrated that heat stress or membrane fluidizers such as benzyl alcohol and heptanol induce changes in membrane fluidity and microdomain reorganization leading to subsequent activation of heat shock genes [111] , [112] , [113] . Noticeably, heat stress resulted in membrane hyperfluidization and membrane rearrangements in leukemia cells [114] . Dempsey and co-workers showed that membrane fluidizers affect the HSPs localization causing decrease in intracellular and increase in surface HSP60 and HSP70 [115] . In addition, HSP70-membrane positive tumors showed to contain globotriaoslyceramide allowing for the anchorage of HSP70 in plasma membrane [116] . HSP70-membrane interaction is also showed to be mediated by HSP70 binding to phosphatidylserine [117] . Since HSP70 released into extracellular space via endolysosomal system, elevated level HSP70 in lysosomes showed to stabilize lysosomal membranes of cancer cells by regulating sphingolipid catabolism [ 110 , [118] , [119] , [120] , [121] , [122] ]. Surface expression of HSPs often indicates highly aggressive tumors [ 14 , 110 , 123 ]. In this regard, many members of HSP family have been found in extracellular space or on the cell membranes- collectively called extracellular HSPs [124] , [125] , [126] , [127] , [128] , [129] , [130] , [131] . High expression of extracellular HSPs showed to correlate with increased cell proliferation, cancer stage and poor clinical outcome suggesting potential use of HSP expression in cancer diagnosis [ 110 , 123 , 132 ]. Studies assessing extracellular HSPs as cancer biomarkers are summarized in Table 2 . A clinical prospective pilot study conducted by Gobbo and colleagues showed that concentration of HSP70-positive exosomes in plasma of lung and breast cancer patients was significantly higher compared to healthy volunteers [133] . Moreover, plasma-derived HSP70-positive exosomes showed to be better in discriminating metastatic from non-metastatic patients than circulating tumor cells (CTC) [133] . Campanella and co-workers showed that HSP60 level in plasma-derived exosomes was significantly higher in patients with colorectal adenocarcinoma than in healthy controls[134]. Elevated expression of HSP70 was found in plasma-derived exosomes of stage IV melanoma patients [135] . Several studies showed that HSPs expression in blood could discriminate between patients with early and late stage of cancer, suggesting the potential use of HSPs in early detection of cancer ( Table 2 ) [136] , [137] , [138] , [139] . Table 2 Selection of studies assessing extracellular HSPs in patient samples. Table 2 HSPs Cancer type Sample Findings Refs HSP70 NSCLC Serum Significantly higher expression of serum HSP70 compared to healthy volunteers. Positive correlation between serum HSP70 expression and tumor volume [131] Elevated level of antibodies to HSP70 in cancer patients than in healthy controls [140] SCLC Serum HSP70 was significantly higher in SCLC patients than in healthy individuals and correlated with poor clinical outcome [141] High level of serum HSP70 compared to controls. Serum HSP70 showed to correlate with stage of the disease [142] Lung cancer High expression of HSP70 in serum increased the risk of developing lung cancer in Japanese males [143] Colorectal cancer High level of soluble HSP70 associated with poor survival [144] Significant increase in serum HSP70 with the stage of the disease. High baseline serum HSP70 associated with poor survival [145] Breast cancer sHSP70 was significantly higher in breast cancer patients than in healthy individuals. HSP70 level of >2.41 ng/ml cutoff was used to predict breast cancer with >90% specificity and sensitivity [146] Breast, colon, colorectal cancers; GIST Significantly higher serum HSP70 level than in normal samples [147] Pancreatic cancer sHSP70 level was significantly higher in patients with pancreatic cancer compared to patients with chronic pancreatitis and healthy volunteers. The sensitivity and specificity for discriminating cancer patients from healthy controls were 74% and 90%, respectively [148] HCC Significantly higher level of serum HSP70 in HCC patients than in control groups. High level of HSP70 was found in cohorts of patients with liver cirrhosis, who eventually developed HCC [149] ESCC Significantly higher level of serum HSP70 autoantibody in patients with ESCC than in colon, gastric cancer patients and healthy controls [150] Head and neck cancer; Lung cancer; Colorectal cancer; Pancreatic cancer; Glioblastoma; Hematologic malignancies Significantly higher serum HSP70 in cancer patients compared to healthy volunteers NSCLC Plasma High level of post-therapeutic circulating HSP70 associated with improved response to radiochemotherapy. [151] AML; MDS; ALL High level of circulating HSP70 associated with significantly shorter overall survival [152] Breast cancer; NSCLC Plasma- derived exosomes HSP70-positive tumor-derived exosomes showed to be better at predicting cancer dissemination than CTC and showed to correlate with the disease status and response to therapy. [ 133 ] Melanoma Statistically higher level of HSP70 in exosomes isolated from melanoma subjects with stage IV melanoma compared to healthy controls [135] Colon cancer; Lower rectal cancer; Gastric cancer; LSCC Patient tissue mHSP70 expression showed to be associated with poor outcome in patients with lower rectal and squamous cell carcinoma of the lungs and positive clinical outcome in patients with colon and gastric cancers [132] Colorectal cancer; AML; Lung cancer; Neuronal tumors; Pancreatic cancer mHSP70 expression was found in leukemic blasts of AML patients and freshly isolated biopsies from patients with colorectal, lung, neuronal and pancreatic tumors. No surface HSP70 expression was found on normal tissues and bone marrow of healthy donors. [153] Melanoma HSP70-membrane positive phenotype was found in melanoma metastases [154] SCCHN Tumor biopsy and serum Significantly higher expression of mHSP70 in tumor biopsy than in reference tissue. Significantly higher expression of soluble HSP70 in SCCHN than in healthy controls. Elevated level of mHSP70 on tumors associated with high soluble HSP70 in patients’ serum. Prior therapy, soluble HSP70 showed to correlate with tumor volume. [155] AML Aspirated bone marrow cells Elevated HSP70-membrane expression associated with poor patient prognosis [156] Breast and pulmonary cancers Urine- derived exosomes Elevated concentration of HSP70-positive exosomes in cancer patients compared to healthy volunteers [157] HSP90 NSCLC Serum No significant difference was found between serum antibodies to HSP90 of cancer patients compared to healthy controls [140] SCLC; LAC; LSCC Serum HSP90AB1/HSP90β was significantly increased in patients compared to healthy controls and significantly associated with grade and stage of cancer [ 158 , 159 ] Melanoma Serum HSP90 significantly higher than in healthy controls. No correlation was found between the level of serum HSP90, patient survival and response to chemotherapy [160] HCC Significantly elevated level of serum HSP90 in cancer patients compared to healthy individuals [161] Colorectal cancer Serum HSP90AA1/HSP90α expression significantly higher in cancer patients compared to healthy volunteers [162] AML Significantly higher serum HSP90AA1/HSP90α expression compared to healthy controls. [163] Ductal and lobular breast tumors No significant association was found between the level of serum HSP90 and the severity of the lesion in patients with ductal and lobular breast tumors [164] NSCLC; SCLC Plasma Plasma HSP90AA1/HSP90α level was significantly higher in cancer patients than in healthy volunteers. Patients with advanced stages had higher plasma HSP90 expression than patients in early stages. Significant difference was found in the level of plasma HSP90 before and after surgery. [138] Liver cancer Level of plasma HSP90AA1/HSP90α could discriminate between patients with liver cancer and control groups, diagnose early-stage liver cancer with >90% sensitivity and specificity [137] Colorectal cancer Significantly elevated level of plasma HSP90AA1/HSP90α in patients with colorectal cancer compared to healthy volunteers. Significantly higher level of plasma HSP90α in late stage than in early stage of cancer [139] Melanoma Plasma-derived exosomes HSP90 isoform was found in exosomes of 70% patients with stage IV melanoma [135] HSPB Gynecologic cancers Serum Significantly higher number of patients had serum antibodies to HSPB1/HSP27 compared to patients with benign lesions [165] Ovarian cancer Significantly elevated level of anti-HSP27/HSPB1 antibodies in patients than in healthy women. Higher level of anti-HSP27 antibodies associated with less advanced stage of cancer [166] Breast cancer HSPB1/HSP27 level was up-regulated in breast cancer patients [167] Significantly higher level of serum HSPB1/HSP27 in breast cancer patients than in healthy volunteers [168] Epithelial ovarian cancer Elevated sHSP27 level in patients with peritoneal metastases. The serum level of HSP27 significantly decreased following chemotherapy in patients with peritoneal metastases. [169] Gynecologic cancers Serum and serum-derived exosomes Significantly higher level of sHSPB5/CRYAB in endometrial cancer patients compared to endometriosis patients. No significant difference in HSPB5, HSPB6/HSP20, HSPB8/HSP22 expression was found in serum-derived exosomes [170] HSP40 SCLC Serum Significantly higher level of autoantibodies to HSP40 in serum was found in lung cancer patients than in healthy controls [171] Chaperonins Ovarian cancer Serum No significant difference in the level of antibodies to HSPD1/HSP60 was found between patients with ovarian cancer and healthy women. Significantly higher concentration of antibodies to HSPD1/HSP60 was found in early stages of cancer than in control groups [172] Colorectal cancer Serum Elevated level of sHSP60 in colorectal cancer patients compared to controls. HSP60 level in serum showed to be more specific for late-stage colorectal cancer [136] Breast cancer; DCIS Serum Autoantibodies to HSP60 were found in patients with early breast cancer and DCIS compared to healthy individuals. [173] Colorectal adenocarcinoma Plasma-derived exosomes The level of HSP60 in exosomes was significantly higher in patients before surgery than in patients after surgery and in healthy controls [134] NSCLC, non-small cell lung cancer; mHSP70, membrane HSP70; sHSP70, serum HSP70; AML, acute myeloid leukemia; MSD, Myelodysplastic syndrome; ALL, Acute lymphoblastic leukemia; SCLC, small cell lung cancer; ESCC, esophageal squamous cell carcinoma; CTC, circulating tumor cells; SCCHN, squamous cell carcinoma of the head and neck; GIST, gastrointestinal stromal tumor; HCC, hepatocellular carcinoma; DCIS, ductal carcinoma in situ; LSCC, Lung squamous cell carcinoma; LAC, adenocarcinoma of the lung Selection of studies assessing extracellular HSPs in patient samples. NSCLC, non-small cell lung cancer; mHSP70, membrane HSP70; sHSP70, serum HSP70; AML, acute myeloid leukemia; MSD, Myelodysplastic syndrome; ALL, Acute lymphoblastic leukemia; SCLC, small cell lung cancer; ESCC, esophageal squamous cell carcinoma; CTC, circulating tumor cells; SCCHN, squamous cell carcinoma of the head and neck; GIST, gastrointestinal stromal tumor; HCC, hepatocellular carcinoma; DCIS, ductal carcinoma in situ; LSCC, Lung squamous cell carcinoma; LAC, adenocarcinoma of the lung