MicroRNAs Regulated by the LPS/TLR2 Immune Axis as Bona Fide Biomarkers for Diagnosis of Acute Leptospirosis

Here, we used miRNAs that are differentially regulated by the LPS/TLR2 immune axis to devise a miRNA-based diagnosis for leptospirosis. The study established the role of the circulating stable miRNAs (miR-21-5p, miR-144-3p, and miR-let-7b-5p) as an early diagnostic marker for leptospirosis. These miRNAs can be used to diagnose acute leptospirosis and also to differentiate leptospiral infection from other bacterial and spirochetal infections, as proved by the use of human clinical samples. Thus, our findings indicate that miRNAs can play a crucial role in the diagnosis of infectious diseases, like leptospirosis, that are generally misdiagnosed.

L eptospirosis is a zoonotic disease caused by pathogenic serovars of spirochetal bacteria belonging to the genus Leptospira (1). Although leptospirosis has long been recognized as an important endemic disease in tropical countries, it is now becoming a more common problem in highly populated urban centers (2). The Leptospirosis Burden Epidemiology Reference Group (LERG) report included a systematic literature review that estimated the overall global annual incidence of endemic and epidemic human leptospirosis at 5 and 14 cases per 100,000 population, respectively (57). Leptospirosis ranges in severity from a mild flu-like illness to hepatorenal impairment, multiorgan failure, and septic shock with pulmonary hemorrhages. Clinically, it is typified by diverse symptoms that include fever, myalgia, headache, severe muscle pain, ocular disorders, meningitis, jaundice, renal failure, and pulmonary hemorrhage, and it is often confounded with other entities such as influenza and dengue fever. In this regard, early and definitive diagnosis of leptospirosis is critical for preventing its severe clinical complications.
We and others have determined that immunity against Leptospira depends on the production of circulating antibodies directed against serovar-specific lipopolysaccharide (LPS) (3,4). Leptospira produces an atypical LPS that differs from LPS of Gramnegative organisms in several biochemical, physical, and biological properties (5). LPS is known to activate macrophages by a mechanism that involves the serum protein LPS-binding protein and a plasma membrane receptor complex comprising at least CD14 and one or more members of the Toll-like receptors (TLRs) (6)(7)(8). Among the 10 TLRs present in human, a large body of evidence supports TLR4 as the predominant receptor mediating LPS activation (9,10). In contrast, it was shown recently that leptospiral LPS activates human cells through CD14 and TLR2 recognition (11,12), whereas both TLR2 and TLR4 contribute to the activation of murine cells (11,12). The recognition of LPS by TLRs results in subsequent recruitment of the intracellular adaptor protein MyD88 through homotypic Toll-interleukin 1 receptor (IL-1R) domains, leading to intracellular signaling through NF-B and mitogen-activated protein kinase (MAPK) activation. The ultimate result of this LPS recognition by TLRs is a proinflammatory cytokine and chemokine response. However, when high levels of leptospiremia occur during infection, innate immune mechanisms eventually trigger tissue-based and systemic responses to infection that lead to severe outcomes, such as a sepsis-like syndrome or organ failure (13)(14)(15). Therefore, it seems probable that mechanisms exist to regulate the host immune response. Although several regulators have been proposed, microRNAs (miRNAs) are recognized as important regulators of immune response (16)(17)(18) and as fine tuners of TLRs (19)(20)(21). miRNAs are 22-nucleotide-long small noncoding regulatory RNA molecules that regulate gene expression posttranscriptionally through complementary base pairing with thousands of messenger RNAs and play important roles in regulating diverse physiological, developmental, and pathophysiological processes (22). Alterations of miRNA expression profiles have been observed in various diseases, including cancer and cardiovascular, neurological, inflammatory, and autoimmune diseases. A growing body of evidence supports the key role of miRNA in the activation of both innate and adaptive immune response (23). Studies have shown differential expression of miRNA following LPS stimulation in various types of immune cells (23,24). Several early and late LPS-responsive miRNAs have been reported (23,24). Thus, identification of an miRNA repertoire responsive to TLR2-mediated LPS signaling will be a valid biomarker for the diagnosis of human leptospirosis. Identification of dysregulated miRNAs in infectious diseases is emerging to help devise novel diagnostics, prevention measures, and therapy (25). Additionally, extracellular miRNAs were identified to be circulating in the blood, thus raising the possibility of finding a connection between specific miRNA levels in serum and various disease states (26)(27)(28). Also, biochemical analyses indicate that miRNAs are resistant to RNase activity, extreme pH, extreme temperatures, extended storage, and large numbers of freeze-thaw cycles (26,29,30), making their isolation and analysis straightforward. Compared to protein-based biomarkers, miRNAs are homogenous and can be easily detected by qPCR, while low abundance and posttranslational modifications of protein markers affect the accuracy of the diagnosis (31). In this regard, the use of miRNAs that regulate LPS-TLR immune axis as novel biomarkers represents a new approach for early diagnosis of leptospirosis.
To that end, in the present study we used in vitro and in vivo models to identify the miRNAs that were differentially expressed during leptospiral LPS stimulation through the TLR2 axis. The identified miRNAs were validated for its efficacy to diagnose early leptospirosis using clinical samples. Additionally, we also predicted (i) the target genes, (ii) the potential functions of the differentially expressed miRNAs by gene ontology (GO) enrichment analysis, and (iii) a protein-protein interaction (PPI) network to identify the hub genes. The present study is the first to identify miRNAs specific for leptospiral LPS stimulation and provide a novel perspective on the involvement of these miRNAs in the pathological mechanism of leptospirosis.

RESULTS
Leptospiral LPS stimulates cell death and cytokine production. We first extracted and purified LPS from pathogenic (Leptospira interrogans serovar Australis strain AHF651, L. interrogans serovar Autumnalis strain N2, L. kirschneri serovar Grippotyphosa strain D22, L. interrogans serovar Pomona strain H3, and Leptospira borgpetersenii serovar Ballum strain BDU51) and nonpathogenic leptospiral strains (Leptospira biflexa serovar Andamana strain CH11). The concentration of the purified leptospiral LPS was between 575 and 625 g/ml, and the LPS was free from protein, DNA, and RNA contaminations. The electrophoretic mobility patterns of leptospiral LPS evidenced apparent molecular masses between 14 and 37 kDa (Fig. 1A). Because leptospiral LPS is recognized by TLR2 in human cells due to differential recognition of the atypical lipid A moiety (11,12), we transiently knocked down TLR2 in THP1 cells. Transient knockdown significantly decreased the levels of TLR2 in THP1 cells with no change in the TLR4 levels ( Fig. 1B and C). Then, we verified the biological activity of leptospiral LPS using control (scrambled [Scr] small interfering RNA [siRNA]) and TLR2 knockdown cells. First, we measured the cytotoxicity of leptospiral LPS by MTT (3-(4,5-dimethyl-2thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide) assay. Increased cytotoxicity was observed in THP1 cells stimulated with pathogenic LPS compared to unstimulated or nonpathogenic LPS stimulation ( Fig. S1A to 1F). Additionally, knocking down TLR2 in THP1 cells protected them against LPS-mediated cell death. Because we observed that LPS extracted from L. interrogans serovar Autumnalis strain N2 caused increased cytotoxicity compared to LPS from other pathogenic serovars, LPS from Autumnalis was used for further analysis. We then measured the mRNA levels of the proinflammatory mediators in control and TLR2 knockdown THP1 cells stimulated with Autumnalis LPS for various time points. In control THP1 cells, mRNA levels of tumor necrosis factor alpha (TNF-␣), NF-B, IL-1␤, and IL-10 increased within 3 h of LPS stimulation. On the other hand, TLR2 knockdown significantly inhibited the elevated cytokine mRNA levels ( Fig. 1D to G). This confirmed that extracted leptospiral LPS activated human cells through TLR2 recognition. Further, to better validate the reactivity of the extracted LPS in vivo, cytokines were measured by extracting mRNA from the whole blood of BALB/c mice at different time intervals (0, 4, 7, 14, and 21 days) after intraperitoneal injection of leptospiral LPS in phosphate-buffered saline (PBS). Consistent with the increase in cytokine mRNA levels in THP1 cells, at day 3 LPS postinjection, there was an increase in mRNA expression of proinflammatory mediators, further validating the potency of the extracted LPS ( Fig. S2A to D).
miRNome miScript microarray analysis of miRNAs. After confirming the ability of leptospiral LPS to prime immune responses through TLR2-dependent signaling, we next asked whether any of the positive or negative regulators of the TLR-signaling cascade could be used as biomarkers for the diagnosis of leptospirosis. Such a TLR regulation is primarily known to be achieved by the activation or repression of a large array of genes and microRNAs (miRNAs). Primarily, we used a human miRNome profile panel to identify LPS-responsive miRNAs that are differentially expressed in Scr siRNA and TLR2 knockdown THP1 cells. Analysis of the miRNA profile of the extracellular media of control and TLR2 knockdown cells with or without LPS treatment showed ϳ18 miRNAs to be upregulated (Ͼ10-fold) in THP1 cells treated with LPS compared to untreated controls (Table S1 and Fig. 2A), whereas knocking down TLR2 normalized the upregulated miRNA levels, indicating that these miRNAs are specific to the TLR2-LPS immune axis and thus can serve as biomarkers (Table S1 and Fig. 2A).
To further validate these miRNAs as authentic biomarkers for early diagnosis of leptospirosis, we adopted a murine model of LPS stimulation. Time-dependent changes in miRNAs expression profiles were analyzed in serum of mice injected with LPS. Compared to control animals, ϳ7, 2, 6, and 5 miRNAs showed Ͼ10-fold upregulation at days 4, 7, 14, and 21, respectively (Table S2 and Fig. 2B). Because we were interested in analyzing the miRNAs that are expressed and circulating in serum during early stages of leptospiral infection, we narrowed our miRNA categories to those that were upregulated day 4 and day 7. Of the miRNAs upregulated at day 4 or day 7, three (miR-21-5p, miR-144-3p, and miR-let-7b-5p) shared similarities with the human miRNome profile of the THP1 cells exposed to leptospiral LPS. miRNAs for early diagnosis of leptospirosis. Because miR-21-5p, miR-144-3p, and miR-let-7b-5p shared similarities between human and mouse miRNome profiles and were upregulated during the early stage of leptospiral LPS administration, we next validated these three miRNAs for their diagnostic potential. Semiquantitative PCR was performed with serum samples to validate the diagnostic efficacy of the miRNAs. The fold changes of circulating miR-21-5p, miR-144-3p, and miR-let-7b-5p in the serum of confirmed cases of leptospirosis were significantly higher (P Ͻ 0.001) than those in healthy controls and persons diagnosed with other febrile illness ( Fig. 3A to C).
Quantifying the levels of serum miRNAs is reliable and reproducible, because the sums of the diagnostic sensitivity and specificity were 93.2% and 88.19%, respectively. The high sensitivity and considerable specificity suggest that the signature miRNAs might be potential candidates for diagnosis of acute leptospirosis.
Functional enrichment analysis of miRNA target genes. Because these miRNAs have diagnostic potential and were modulated by the TLR2 axis, we next asked whether these miRNAs target the genes that are involved in immune regulation. In both DIANA-microT-CDS and TargetScan, totals of 993, 521, and 1,611 genes were predicted to be the target genes of miR-let-7b-5p, miR-21-5p, and miR-144-3p, respectively. Gene ontology enrichment analysis revealed that miR-21-5p, miR-144-3p, and miR-let-7b-5p play important roles in disease signal transduction, signaling by interleukins, regulation in actin cytoskeleton, stress-activated protein kinase signaling cascade, MAPK signaling pathway, FoxO signaling pathways, cellular response to a transforming growth factor ␤ (TGF-␤) stimulus, and small-RNA loading onto the RNA-induced silencing complex (RISC) (Fig. 4A to C). Notably, all these biological processes are interconnected (Fig. 4D to F). Our data imply that the genes involved in the target signaling pathways may serve as potential diagnostic and therapeutic targets for leptospirosis in the future.

DISCUSSION
Monocytes belong to a subset of circulating white blood cells that can further differentiate into macrophages and dendritic cells (32). In vivo and in vitro studies have shown that monocytes and their derivatives function as essential components of the innate immune system that mediate host defense and serve as the first line of resistance to microbial attack (33). This innate immunity is triggered by the recognition of pathogen-associated molecular patterns (PAMPs), like LPS, through pattern recognition receptors (PRR); one such conserved PRR is TLR. Indeed, the pleiotropic potential of monocytes suggests that their cellular functions must be tightly regulated in a manner that is distinct from that of more differentiated myeloid cell types to enable appropriate, context-dependent responses.
MicroRNAs (miRNAs) are a large class of small noncoding RNAs that posttranscriptionally regulate mRNAs and subsequently influence essential cellular functions through modulation of gene expression at the RNA or protein level (34). Several miRNAs in myeloid cells were previously identified to impact immune responses, although results have varied due to different cell types, assay methods, and culture conditions (35,36). Despite differences in cell type, both oligonucleotide-based microarray technology (21,37,38) and next-generation sequencing (NGS) technologies (23) indicated that the classic miRNAs hsa-mir-155, hsa-mir-9, and hsa-mir-146a are upregulated in LPS-treated monocytes, suggesting that these miRNAs are critical for promoting pan-myeloid function in response to LPS (21,23,37,38). In another independent study, miRNA profiles of whole blood in mice exposed to LPS demonstrated a significant alteration of the multiple miRNAs (let-7d, miR15b, miR16, miR25, miR92a, miR103, miR107, and miR451) in a dose-and time-dependent fashion. Additionally, LPS from different Gram-negative bacteria, including Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, Salmonella enterica, and Serratia marcescens, induced the upregulation of similar miRNAs (24). However, in the present study, stimulation of monocytes (THP1 cells) or whole animals with leptospiral LPS did not induce the upregulation of any of these known signature miRNAs. Rather, 18 different miRNAs (Table S1 and Fig. 2A) were upregulated by stimulation of THP1 cells with leptospiral LPS. Additionally, knocking down TLR2 in THP1 cells normalized the expression of these miRNAs.
The difference in the miRNAs that are upregulated by leptospiral LPS stimulation could be explained by differences in the structure of the leptospiral lipid A. Lipid A is the anchor moiety of LPS in the bacterial membrane and is the active component of LPS responsible for its toxic activity and functions. Although the structure of lipid A from Gram-negative bacteria shows some variability between bacterial species (39), all lipid A moieties are known to stimulate the TLR4 complex. However, leptospiral LPS activated human cells through TLR2 instead of TLR4 (12). The leptospiral lipid A moiety structure was recently deciphered, revealing that it possesses some peculiar characteristics compared to lipid A of Gram-negative bacteria (40), including one phosphate residue that is capped with a methyl group, a lack of a 4= phosphate group, and four amide-linked acyl chains, longer than in enterobacterial lipid A, with two secondary unsaturated acyl chains. Additionally, of these 18 miRNAs upregulated by leptospiral LPS, 6 (hsa-miR-185-5p, hsa-let-7b-5p, hsa-miR-28-5p, hsa-miR-376c-3p, hsa-miR-195-5p, and hsa-miR-100-5p) were found to be upregulated in bone marrow-derived mesenchymal stem cells stimulated with TLR2 agonist (PAM3CSK4 cells) (41). This further confirms that the identified miRNAs are upregulated by leptospiral LPS through the TLR2 immune axis, and these miRNAs can be significant signature molecules to differentiate leptospiral infection from other bacterial infections with which it is often confused. Because other spirochetes, like Borrelia and Treponema, do not possess LPS, identification of leptospiral LPS-stimulated miRNAs could differentiate leptospiral infection from other spirochetal infections.
A recent work that examined the microtranscriptome of murine macrophages J774A.1 stimulated with virulent, attenuated, and saprophyte strains of Leptospira showed differential expression of 29 miRNAs. Of these 29 miRNAs, the highest fold change was observed for miR155-5p in cells stimulated with pathogenic L. interrogans (42). We did not see an upregulation of this miR155-5p, which could be due to different factors, including but not limited to the difference in the assay method, the cell type, and the culture conditions. Moreover, our study identified the miRNAs that are regulated by LPS stimulation in an in vivo mouse model. In this regard, our study is the first to validate the efficacy of miRNAs for the diagnosis of leptospirosis.
Early diagnosis of leptospiral infection is critical for preventing further complications, including multiorgan failure and hemorrhages. Although routine clinical methods, including isolation of leptospires, microscopic agglutination test (MAT), enzymelinked immunosorbent assay (ELISA), and PCR, are standard methods for diagnosis of leptospirosis, these techniques are time-consuming and require special expertise that might delay treatment. Also, until now, potential biomarkers have included acute phase proteins, cytokines, and chemokines, but they are not sufficient to distinguish other bacterial infections from leptospiral infection. The miRNAs identified in our study are promising biomarkers for early leptospiral infection. In fact, miRNA-based biomarkers are already being used for diagnosis of tuberculosis (43,44). Because we were interested in developing biomarkers for the early identification of leptospirosis, we characterized the circulating miRNAs in a time-dependent manner (4, 7, 14, and 21 days) in the whole blood of mice after LPS stimulation. The miRNAs miR-21-5p, miR-144-3p, and miR-let 7b-5p shared similarity with the human miRNome profile and were also expressed during the early stage of leptospiral infection (Fig. 2). We validated the diagnostic potential of these three miRNAs using patients' serum samples. The diagnostic sensitivity and specificity were calculated to be 93.2% and 88.19%, respectively (Fig. 3). Despite accumulating evidence of miRNAs circulating in the blood and body fluids, and the origin and functions of these circulating extracellular miRNAs remain poorly understood (45). Most circulating miRNAs are part of larger lipid or lipoprotein complexes (26,27,29,45). Currently, little is known regarding the biological roles of these molecules at distant sites in the body (46). Extracellular miRNAs may be mediators of cell-cell communications (47,48) or show hormone-like effects, leading to widespread responses (30,49). The expression of circulating miRNAs is thought to reflect extrusion of miRNAs from relevant remote tissues or organs or disease processes (30). Indeed, the present study did not show that the miRNAs that are differentially expressed in the circulation are expressed in other tissues. Thus, the source of circulating miRNAs not yet understood, and future studies will be performed to clarify the origin and physiological roles of these circulating miRNAs.
However, attempts were made to identify the physiological role of these miRNAs in cell function using GO analysis. Gene ontology enrichment analysis revealed that miRNAs play important roles in disease signal transduction, signaling by interleukins, the stress-activated protein kinase signaling cascade, the MAPK signaling pathway, FoxO signaling pathways, and cellular response to TGF-␤ stimulus (Fig. 4A to C). These biological processes are in fact, interconnected (Fig. 4D to F). We propose that these miRNAs fine-tune the gene expression of key components in the pathway under the impact of LPS stimulation. However, this hypothesis does not rule out other possible functions for these miRNAs, since dozens of the targets, validated or predicted, will be still outside the LPS-TLR immune axis pathway. Therefore, in light of the increasing knowledge about the role of miRNAs in disease, the promise they hold for therapeutics and diagnostics should not be underestimated (50). Our study provides evidence that miRNAs play a crucial role in the diagnosis of infectious diseases like leptospirosis that are generally misdiagnosed.
The major challenge of this study is that the miRNAs were tested for its diagnostic potential in a very small patient cohort and without a random sampling scheme. Also, we consider the analysis of miRNAs in the human clinical samples biased, as it is just a confirmation of the serum miRNAs by PCR. In order to use these identified miRNAs as the best prognostic markers, our plan is to enroll a large cohort of patients with laboratory-confirmed leptospirosis. Additionally, comparing the serum miRNA profiles of greater numbers of healthy individuals and patients with other confirmed febrile illnesses will validate the prognostic use of these identified miRNAs. This approach is expected to overcome the problem of definitive diagnosis of acute leptospirosis in the near future.
Conclusions. The current work established the relationship between miRNAs and innate immunity and established a detailed profile of circulating miRNAs in THP1 cells and mouse sera after exposure to leptospiral LPS. Further, these findings also established the role of the circulating miRNAs (miR-21-5p, miR-144-3p, and miR-let-7b-5p) as an early diagnostic marker for leptospirosis. These miRNAs will facilitate diagnosis of acute leptospirosis and differentiation of leptospiral infection from other bacterial and spirochetal infection.

MATERIALS AND METHODS
Leptospiral strains and media. The pathogenic leptospiral strains L. interrogans serovar Australis strain AHF651, L. interrogans serovar Autumnalis strain N2, L. kirschneri serovar Grippotyphosa strain D22, L. interrogans serovar Pomona strain H3, and L. borgpetersenii serovar Ballum strain BDU51, isolated from ailing human subjects, and nonpathogenic L. biflexa serovar Andamana strain CH11, isolated from water sources, were obtained from the WHO Reference Centre for Leptospirosis, Regional Medical Research Centre, ICMR, Port Blair, Andaman Islands (India). Strains were maintained by regular subculturing in Ellinghausen-McCullough-Johnson-Harris (EMJH) bovine serum albumin-Tween 80 medium (Difco Laboratories, USA) in a shaking incubator (100 rpm) at 30°C. The pathogenicity of the strains was maintained by passaging the isolates in immunocompromised (cyclophosphamide-treated) BALB/c mice ϳ15 times (51). These pathogenic mouse-adapted challenge strains (MACS) were used for all the experiments.
Cells and cell culture. THP-1 cells obtained from NCCS, Pune, India, were cultured in RPMI medium containing 10% fetal bovine serum (FBS) and 2 mM glutamine (Sigma-Aldrich, St. Louis, MO, USA) and supplemented with 100 U/ml penicillin and 100 g/ml streptomycin at 37°C in a humidified 5% CO 2 atmosphere.
Patients and case definition. In total, 46 serum samples collected during the early phase of illness (0 to 10 days after the onset of disease) were selected from a bank of samples from 376 laboratoryconfirmed cases of leptospirosis (a positive IgM ELISA or MAT titer of Ն1:160 or isolation of leptospires from the blood). A total of 22 seronegative healthy controls matched with respect to age (Ϯ5 years) and sex and 66 patients who were hospitalized with a clinical suspicion of leptospirosis and subsequently diagnosed as having other illness, including dengue (22), malaria (22), and typhoid (22), based on laboratory evidence were also included. Informed written consent was obtained from both case patients and controls before sampling, and the study protocol was approved by the Institutional Ethics Committee (IEC) of Bharathidasan University (DM/2014/101/49) as well as receiving permission from the Directorate of Health Services (reference no. 5796/TV 1/07), Tamil Nadu, India.
Extraction of leptospiral LPS. Leptospires were collected from mid-log-phase cultures by centrifugation at 10,000 ϫ g for 30 min and then washed three times with sterile 1ϫ Dulbecco's phosphatebuffered saline (PBS) (Corning, Manassas, VA, USA) before LPS extraction. Crude LPS was extracted following standard hot phenol-water extraction as described previously (4). The phenol phase that contained the LPS was dialyzed extensively against Milli-Q water (8 changes for 4 days) to remove phenol. Purified LPS was pelleted by ultracentrifugation (120,000 ϫ g for 4 h at 4°C), dissolved in Milli-Q water, and lyophilized. Lyophilized crude material was resuspended in 10 ml of Milli-Q water and treated with DNase and RNase (Sigma-Aldrich, St. Louis, MO, USA), followed by proteinase K digestion (Sigma-Aldrich, St. Louis, MO, USA). LPS was quantified by the phenol-sulfuric acid method using sucrose as a standard (52). Protein was determined by bicinchoninic acid (BCA) kit (Sigma-Aldrich, St. Louis, MO, USA). Total and inorganic phosphorus were determined as described previously. Sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) was performed on a 12% polyacrylamide gel using a discontinuous buffer system. The extracted LPS was mixed with 2ϫ SDS-PAGE sample loading buffer (Bio-Rad, Hercules, CA, USA) and boiled for 5 min before loading. Electrophoresis was carried out in a vertical electrophoretic minicell unit (Bio-Rad, Hercules, CA, USA) for 2 h at 100 V using the Tris-glycine running buffer (25 mM Tris, 192 mM glycine, 0.1% SDS [pH 8.3]) (53). A modified silver staining procedure was performed to detect the LPS bands on SDS-PAGE (4, 54) and documented using a gel documentation system (Bio-Rad, Hercules, CA, USA).

Statistical analysis.
All data were normally distributed and are presented as means and standard errors of the means (SEM). In the case of single-mean comparisons, data were analyzed by Student's t test or, when not normally distributed, a nonparametric Mann-Whitney U test. Differences in means among multiple data sets were analyzed using 1-way analysis of variance (ANOVA) with the Tukey test. P values less than 0.05 were considered significant. The diagnostic potential of the miRNAs was evaluated by calculating the area under the receiver operating characteristic (ROC) curve and AUC. For ROC curves, the cutoff value was defined to maximize the sum of sensitivity and specificity. The data were computed with SigmaPlot 11.0 software or GraphPad Prism 8.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.