Lectin nanoparticle assays for detecting breast cancer-associated glycovariants of cancer antigen 15-3 (CA15-3) in human plasma

RESEARCH ARTICLE Lectin nanoparticle assays for detecting breast cancer-associated glycovariants of cancer antigen 15-3 (CA15-3) in human plasma Joonas Tera ¨va ¨ID 1☯‡ , Leena Tiainen 2☯‡ , Urpo Lamminma ¨ki 1 , Pirkko-Liisa KellokumpuLehtinen 2 , Kim Pettersson 1 , Kamlesh GidwaniID 1 * 1Department of Biochemistry/Biotechnology, University of Turku, Turku, Finland, 2Faculty of Medicine and Health Technology, Tampere University and Department of Oncology, Tampere University Hospital, Tampere, Finland ☯These authors contributed equally to this work. ‡ These authors are co-first authors on this work. *[email protected] Abstract Cancer antigen 15–3 (CA15-3) is widely utilized for monitoring metastatic breast cancer (BC). However, its utility for early detection of breast cancer is severely limited due to poor clinical sensitivity and specificity. The glycosylation of CA15-3 is known to be affected by BC, and therefore it might offer a way to construct CA15-3 glycovariant assays with improved cancer specificity. To this end, we performed lectin-based glycoprofiling of BC-associated CA15-3. CA15-3 expressed by a BC cell line was immobilized on microtitration wells using an anti-CA15-3 antibody. The glycosylation of the immobilized CA15-3 was then detected by using lectins coated onto europium (III)-doped nanoparticles (Eu +3 -NPs) and measuring the time-resolved fluorescence of Eu. Out of multiple lectin-Eu +3 -NP preparations, wheat germ agglutinin (WGA) and macrophage galactose-type lectin (MGL) -Eu 3+ -NPs bound to the BC cell line-dericed CA15-3 glycovariants (CA15-3 Lectin ). To evaluate the clinical performance of these two lectin-based assays, plasma samples from metastatic BC patients (n = 53) and healthy age-matched women (n = 20).Plasma CA15-3 Lectin measurements better distinguished metastatic BC patients from healthy controls than the conventional CA15-3 immunoassay. At 90% specificity, the clinical sensitivity of the assays was 66.0, 67.9 and 81.1% for the conventional CA15-3, CA15-3 MGL and CA15-3 WGA assays, respectively. Baseline CA153 MGL and CA15-3 WGA were correlated to conventional baseline CA15-3 levels (r = 0.68, p<0.001, r = 0.90, p>0.001, respectively). However, very low baseline CA15-3 MGL levels �5 U/mL were common in this metastatic breast cancer patient population.In conclusion, the new CA15-3 Lectin concept could considerably improve the clinical sensitivity of BC detection compared to the conventional CA15-3 immunoassays and should be validated further on a larger series of subjects with different cancer subtypes and stages. PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 1 / 16 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Tera¨va¨J, Tiainen L, Lamminma¨ki U, Kellokumpu-Lehtinen P-L, Pettersson K, Gidwani K (2019) Lectin nanoparticle assays for detecting breast cancer-associated glycovariants of cancer antigen 15-3 (CA15-3) in human plasma. PLoS ONE 14(7): e0219480. https://doi.org/10.1371/ journal.pone.0219480 Editor: Lu-Gang Yu, University of Liverpool, UNITED KINGDOM Received: February 11, 2019 Accepted: June 25, 2019 Published: July 25, 2019 Copyright: ©2019 Tera¨va¨et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: This work was funded by the Pirkanmaa Hospital District Science Center, Seppo Nieminen funds (LT, PLKL) and Jane and Aatos Erkko foundation (JT, KG, KP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Introduction Breast cancer (BC) is the most common cancer type and the second leading cause of cancer death in women worldwide [1]. Cancer antigen 15–3 (CA15-3 also known as MUC1) is shed from tumor cells and is a well-known serological marker for monitoring the clinical course of BC patients. A persistent increase in circulating concentration of this marker may suggest an inadequate response to cancer therapy in patients with metastatic BC. However, it has poor sensitivity, especially at early stages of the disease.[2] CA15-3 can also be elevated in healthy individuals and in patients with benign conditions, and it lacks the specificity needed for cancer screening, diagnosis, staging, and/or sole use in monitoring of post-therapy recurrence [3]. A study on retrospective samples found the sensitivity of the commercial Elecsys CA 15–3 immunoassay to be 7, 11, 39 and 78% on stage I, II, III and IV BC patients, respectively [4]. Recently an ultrasensitive, simple and reliable electrochemical immunosensor was developed to detect the lowest alteration of CA 15–3 and CA125, biomarker of breast and ovarian cancer patients respectively [5,6]. For monitoring metastatic breast cancer, international recommendations for the treatment of metastatic BC only recommend the monitoring of CA15-3 levels for patients with nonevaluable metastases, mainly bone-dominant disease [7,8]. Transient increases in plasma CA15-3 levels are possible without tumor progression [9]. This phenomenon is observed especially in the beginning of chemotherapy due to necrosis and apoptosis of tumor cells. Additionally, consensus about clinically meaningful increase in plasma CA15-3 levels to predict disease progression or clinically meaningful decrease to reflect a treatment response do not exist today. Nevertheless, in general plasma CA15-3 levels correlate with the response to chemotherapy in patients with metastatic breast cancer [10,11]. Protein glycosylation plays an important role in a wide variety of normal and diseaserelated biological processes. The phenomenon of aberrant glycosylation associated with malignant transformation, tumor progression and metastasis is well documented [12] and occurs in essentially all types of human cancers. A large number of altered glycosyl epitopes are classified as tumor-associated carbohydrate antigens. [13,14] Among these, the aberrant expression of Tn and sialyl-Tn antigens, L-fucose and terminal N-acetylglucosamine (GlcNAc) have been widely detected in breast cancers [15,16]. Especially, abnormal O-glycans, such as Tn antigen, are found in over 90% of breast cancers[17]. Overall, changes in glycosylation result in the production of various cancer-associated glycoproteins with cancer-associated glycoforms, which are antigenically distinct from the corresponding molecules of the normal tissue. Taking into account these modifications, the specificity of diagnostic cancer markers can be expected to be improved by using the aberrant glycoforms as targets. CA15-3 is upregulated and aberrantly glycosylated in breast and other carcinomas [18].The CA15-3, derived from a large transmembrane protein Mucin 1 with molecular weight ranging from 500 to 1000 kDa, contains multiple Oand N-linked glycosylation sites. The O-glycans of CA15-3 produced by the normal breast tissue are core 2-based and can be complex, while the O-glycans added to the BC mucin are mainly core 1-based [19]. The resulting truncated glycans carried on BC-associated CA15-3 include Tn and T antigens and their sialylated forms [14]. CA15-3 purified from the culture medium of human BC YMB-S cells contains 3-O-sulfated or 3-sialylated core 1 and extended core 1 glycans. [20] Glycans participate in early stages of tumorigenesis [12] and it has been reported that the expression level of an enzyme responsible for mucin-type glycosylation, N-acetylgalactosaminyltransferase-14, declines with breast cancer progression [21]. Thus, it is reasonable to assume that the cancerous glycovariants of glycoprotein tumor markers appear early and differ throughout the course of the disease. Therefore, glycovariant markers may be useful for early detection as well as for monitoring cancer progression. Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 2 / 16 Competing interests: KP and KG have a patent application (Application WO-2018011474-A1 related to this work "Lectin based diagnostics of cancers." This does not alter our adherence to PLOS ONE policies on sharing data and materials. Various lectins, members of a carbohydrate binding protein family, have previously been used to investigate the differences in glycosylation between soluble glycoproteins expressed by cancerous and benign tissues. A recent study described the use of a 3-sulfated core 1 -specific galectin-4 (Gal-4) to establish an assay exhibiting superior clinical performance compared to the conventional CA15-3 immunoassay for BC detection [22]. Also, C-type lectin receptors (CLR) such as macrophage galactose-type lectin (MGL) have been demonstrated to show increased binding to CA15-3 from lysates of colon cancer tissue compared to the healthy lysed colon tissues of the same patients [23]. The Lens culinaris agglutinin, a lectin found in lentil, in turn binds specifically to hepatocellular carcinoma -associated glycovariant of α-fetoprotein (AFP) and is the only lectin used in a commercial application to detect a biomarker glycovariant [24]. While showing these promising binding specificities, lectins unfortunately tend to have weak binding affinity, which apparently limits their exploitation in practical assay applications. We previously reported a novel lectin-based approach for the detection of cancer-associated glycosylation of CA125, a well-known mucin 16 -derived cancer marker used e.g. for monitoring of epithelial ovarian cancer. The approach, relying on the use of highly fluorescent europium(III)-doped nanoparticles (Eu +3 -NPs) coated with the lectin MGL, enabled highly sensitive detection of CA125 produced by ovarian cancer cell line OVCAR-3. In the clinical evaluation, the resulting optimized assay (CA125 MGL ) showed good discrimination between the samples of epithelial ovarian cancer patients and those with endometriosis, a condition that has decisively hampered the use of CA125 for early detection/screening of ovarian cancer. [25] In addition, we found that the new assay could alarm clinicians much earlier (4–6 months) than the conventional CA125 assay about disease relapse. These results motivated us to explore possibilities of the lectin nanoparticle assay concept for detecting the altered glycosylation of CA15-3 in the blood streams of BC patients. In the present work, we utilized the lectin-Eu +3 -NP approach for the glycoprofiling of CA15-3 with a panel of 28 lectins in order to identify lectins recognizing BC related changes in carbohydrate structures of CA15-3. The discovered promising lectins were then validated with plasma from patients with metastatic BC and healthy female controls. Additionally, we explored new CA15-3 lectin assays in monitoring response of metastatic breast cancer. Materials and methods Clinical samples Plasma samples from 53 metastatic breast cancer patients were analyzed. These patients participated in a first-line chemotherapy trial for metastatic breast cancer (NCT00979641). The samples were analyzed at baseline, after six weeks of chemotherapy treatment, after six months of study treatment and at the final study visit. The trial design and the patient demographics have been published previously [26]. In brief, the patients with metastatic HER2-negative BC were enrolled into the trial, if they had not received previous chemotherapy for the advanced disease. The mean age of the study patients was 58 years (range 32–75). Most of the patients had hormone receptor positive disease (81%) and visceral metastases (79%). The median time between six-month sample and the final plasma sample was 11.8 months (inter quartile range 3.5.-18.9 months). The Ethics Committee of Tampere University Hospital approved the study protocol (R08142M). Clinically meaningful change in CA15-3 levels was defined as 30% similarly as the partial response criterion in the response evaluation criteria in solid tumors [27]. The definition of clinically meaningful change in circulating tumor markers varies around 20– 40% in previous studies [10,28]. Disease progression was defined as investigator-assessed radiological progression according to the RECIST criteria [27]. Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 3 / 16 Control plasma samples were obtained from 20 healthy women participating in a mammography-screening program in Tampere City Breast Clinic. These women voluntarily took part in a breast cancer primary prevention study currently in progress at University of Tampere and as a part of the study, plasma samples were drawn for scientific purposes. The mean age of these healthy controls was 56 years (range 54–67). All participants gave written informed consent (Ethics approval R15023). Reagents CA15-3 isolated from the breast cancer cell line ZR-75-1 (ATCC CRL-1500) (BC-CA15-3), two monoclonal anti-CA15-3 antibodies; Ma552 and Ma695, that specifically recognize a PDTRPAPG region of the protein core and sialylated carbohydrate epitope expressed on the CA15-3 antigen respectively, were provided by Fujirebio Diagnostics (Go¨teborg, Sweden). Streptavidin-coated yellow 96-well plates, wash buffer and red assay buffer were purchased from Kaivogen (Turku, Finland). Europium(III)-doped Fluoro-Max polystyrene nanoparticles (97 nm in diameter) (Eu +3 -NP) were acquired from Seradyn (Indianapolis, IN, USA). A panel of plant lectins with different glycan binding specificities (Table A in S1 Dataset) was obtained from Vector laboratories (Burlingame, CA, USA). The recombinant human lectins were purchased from R&D Systems (Abingdon, United Kingdom). Preparation of lectin-Eu 3+ -NPs The use of Eu +3 -NPs has been described before [29]. The coating of lectins on Eu +3 -NPs was performed essentially as described before [30]. The buffer used for storage of the l ectin coated Eu +3 -NPs was 10 mM Tris-HCl, pH 7.8, supplemented with 0.1% BSA and 0.01% sodium azide at +4˚C, covered from light. The particles were thoroughly vortexed and sonicated before every use to disperse aggregates. Labelling of antibodies with biotin Both solid-phase monoclonal antibodies (Ma552 and Ma695 mAb) were biotinylated with 40-fold molar excess of biotin isothiocyanate, for 4 h at room temperature (RT). The labelled antibodies were separated from the unconjugated biotin by using NAP-5 and NAP-10 gel-filtration columns (GE Healthcare, Schenectady, NY, USA) by using 50 mM Tris–HCl (pH 7.75), containing 150 mM NaCl and 0.5 g/L NaN3. The labelled antibodies were stabilized with 1 g/L BSA (Bioreba, Nyon, Switzerland) and stored at +4˚C. [31] In-house CA15-3 Lectin-NP assay The assay principle is represented in Fig 1. Biotinylated Ma552 or Ma695 mAb (100 ng/30 μl/ well) in buffer solution was incubated for 1 h at RT to immobilize them on streptavidin-coated yellow low-fluorescence microtiter wells. The wells were washed two times with wash buffer and 25 μl of CA15-3 standard/sample (diluted 1:40 in buffer) was added and incubated for 1 h at RT with slow shaking. The immobilized BC-CA15-3 was detected by lectin-Eu 3+ -NPs as a tracer by using time-resolved fluorescence (TRF). Ten million lectin Eu +3 -NPs per well in 25 μl of assay buffer containing additional 6 mM CaCl 2 was added. The wells were incubated for two hours at RT in shaking and washed six times. To detect the lectin-Eu 3+ -NPs bound to BC-CA15-3, the TRF of Eu (λ ex : 340 nm; λ em : 615 nm) was measured for 400 μs after a 400 μs delay using Victor 3 V 1420 Multilabel counter (Wallac, Turku, Finland). Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 4 / 16 Conventional CA15-3 immunoassay CA15-3 concentrations were analysed in plasma samples with a CA15-3 enzyme immunoassay (Fujirebio Diagnostics Inc., Malvern, PA, USA) according to manufacturer’s instructions. Statistical analysis Receiver operating characteristics (ROC) were determined and compared, and the areas under the curve (AUC) values calculated using R version 3.3. [32] with the pROC package [33]. The measured concentrations of each assay (Table B in S1 Dataset) were used as the classifier. The comparison of ROCs was done using the bootstrap method provided in the pROC package. Due to the nonparametric distribution of the CA15-3 levels, medians with the interquartile range (IQR) of the median were reported. CA15-3 levels of healthy controls were compared to CA15-3 levels of metastatic BC patients using the Mann Whitney U-test. Wilcoxon Rank test was used when comparing baseline and week six CA15-3 levels in relation to the treatment response. Spearman’s correlation was used to study the correlation between conventional CA15-3 levels and CA15-3 MGL or CA15-3 WGA levels. The Wilcoxon signed-rank test and Spearman’s correlation analyses were performed using SPSS version 23 statistical software package (SPSS Inc., Chicago, IL, USA). P value of less than 0.05 was considered significant in all statistical tests. Results Screening of lectins for binding to BC-CA15-3 Altogether 28 lectins with various carbohydrate-binding specificities (Table A in S1 Dataset) were tested to investigate the glycosylation patterns of the cancer cell line -derived BC-CA15-3 Fig 1. The principle of the conventional and in-house Eu +3 -NP-based CA15-3 lectin assays. In the conventional CA15-3 immunoassay, the capture and tracer mAbs bind to the protein and glycan epitopes of CA15-3. Alternatively, in the lectin assay, the CA15-3 is captured with mAbs and detected with lectins, which have been coated on the surface of Eu 3+ -NPs. This method allows multivalent binding of the tracer to the glycan moieties of BC-CA15-3. https://doi.org/10.1371/journal.pone.0219480.g001 Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 5 / 16 preparation. Fig 2 shows the signal-to-background ratios obtained with the corresponding lectin-NP tracers using two different monoclonal antibodies (Ma552 and Ma695) for capturing CA15-3. Four of the tested nanoparticle tracers; MGLWGA-, Gal-4-, and DSL-NPs, recognized BC-CA15-3 and the trend was similar for both capture antibodies. WGA exhibited highest signal-to-background ratio followed by MGL, Gal-4 and DSL (Fig 2). WGAand MGL-NPs displayed excellent recovery (93% to 98%) when BC-CA15-3 was spiked into pooled healthy plasma samples whereas Gal-4and DSL-NPs scarcely bound to BC-CA15-3 spiked Fig 2. Lectin NPs binding to BC-associated CA15-3 from cell line ZR-75-1 (ATCC CRL-1500) using the lectin assay principle depicted in Fig 1. The different lectin Eu +3 -NPs used are shown on the x-axis and the y-axis displays the signal to background ratios using either biotinylated Ma695 (bioMa695) or biotinylated Ma552 (bioMa552) as the capture mAb. https://doi.org/10.1371/journal.pone.0219480.g002 Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 6 / 16 similarly in plasma. We selected MGL (here after, CA15-3 MGL ) and WGA (CA15-3 WGA ) for further evaluation using clinical samples. Characteristics of CA15-3 MGL CA15-3 WGA assays The analytical performance of the CA15-3 MGL and CA15-3 WGA assays were preliminarily tested using a BC-CA15-3 in a range of concentrations from 1 to 1000 U/mL. Saturation was not observed at the maximum used BC-CA15-3 of 1000 U/mL. The limit of detection, which was set to be the concentration of BC-CA15-3 required for a signal equivalent to the mean of blank calibrator (n = 20) plus three times the standard deviation, was less than 1 U/ mL. Linear in response was observed at a maximum of 125 U/mL (S1 Fig). No cross-reactivity was observed towards two other glycoprotein cancer markers, CA125 and prostate specific antigen (S2 Fig). Plasma CA15-3, CA15-3 MGL , and CA15-3 WGA concentrations in the study cohort We next studied whether CA15-3 in the plasma of BC patients binds with MGL and WGA similar to CA15-3 of a breast cancer cell line. The baseline EDTA plasma samples from 53 patients with metastatic BC and 20 healthy individuals were measured for CA15-3 MGL and CA15-3 WGA and compared with the conventional CA15-3 immunoassay. To assess the diagnostic value of the tumor markers in metastatic BC, ROC curves were plotted and AUC was calculated. The highest AUC value was achieved with CA15-3 WGA (0.943) followed by CA153 MGL (0.852) while the conventional CA15-3 immunoassay yielded the lowest AUC of 0.827 (Fig 3). At 90% specificity the sensitivities of the assays were 81.1, 67.9 and 66.0% for the CA15-3 WGA , CA15-3 MGL and conventional CA15-3, respectively. The difference in the AUC compared to the conventional assay was significant for CA15-3 WGA (p = 0.007) but not for CA15-3 MGL (p = 0.655). Metastatic BC patients had higher median baseline plasma levels of conventional CA15-3 as well as CA15-3 MGL and CA15-3 WGA levels than the healthy controls (Table 1). Plasma samples were available from 53 metastatic breast cancer patients. However, both baseline and week six samples were available only from 41 patients. Median CA15-3 levels were lower at week six than at baseline for all three CA15-3 assays in the entire study population (p-values 0.007, <0.001, <0.001 for CA15-3, CA15-3 MGL and CA15-3 WGA , respectively). The decline in CA153 levels was more pronounced in responding patients for all CA15-3 assays, especially CA153 MGL (Table 2). For all the three different CA15-3 assays, the responding patients had a significant decrease in all assays of CA15-3 between baseline and week six (p-values 0.003, <0.001 and <0.001 for CA 15–3, CA15-3 MGL and CA15-3 WGA , respectively, Table 2). Baseline conventional CA15-3 and CA15-3 MGL levels correlated to each other (r = 0.68, p<0.001, Fig 4A and 4B). However, almost half of the metastatic BC patients had very low baseline CA15-3 MGL levels (�5 U/Ml, dashed vertical line in Fig 4B). A stronger correlation was observed between conventional CA15-3 and CA15-3 WGA (r = 0.90, p<0.001, Fig 4C and 4D). Additionally, we studied CA15-3 levels at disease progression (Fig 5).We had plasma samples available from 19 patients who had a disease progression at final study visit. A clinically meaningful 30% increase in the final CA15-3 levels was observed in eight patients (42%) with the conventional CA15-3, nine patients (47%) with the CA15-3 MGL and six patients (32%) with the CA15-3 WGA . The patients with rising CA15-3 levels at disease progression were not entirely the same individuals for the different CA15-3 assays. Specifically, five patients had similar increase in final CA15-3 MGL levels and CA15-3 levels. However, four patients with Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 7 / 16 rising CA15-3 MGL levels did not have an increase in conventional CA15-3 levels. Furthermore, a similar increase was observed in four patients in final CA15-3 WGA and conventional CA15-3 levels. However, two patients with rising CA15-3 WGA levels did not have an increase in conventional CA15-3 levels. Additionally, at least 30% decrease in the final CA15-3 levels at disease progression was observed in three patients (16%) with the conventional CA15-3, 3 patients (16%) with CA15-3 MGL and five patients (26%) with CA15-3 WGA . Fig 3. ROC plot displaying the AUC of conventional CA15-3 (green), CA15-3 MGL (purple) and CA15-3 WGA (red) from metastatic breast cancer patients (n = 53) and healthy control (n = 20). The 95% confidence intervals of the ROCs are depicted as shaded areas and displayed numerically in brackets. The color of shadings corresponds to the plotted lines and the overlap of conventional CA15-3 and CA153 MGL is dark green, the overlap of all assays is brown and the overlap of CA15-3 MGL and CA15-3 WGA is dark red. https://doi.org/10.1371/journal.pone.0219480.g003 Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 8 / 16 Discussion The results suggest that the glycovariant specific assays provide advantages over the conventional CA15-3 immunoassay in monitoring of BC patients, and especially for the detection of metastatic disease and its recurrence. CA15-3 is a tumor marker commonly used for monitoring patients with advanced BC. However, the currently employed sandwich immunoassays that target two protein epitopes have moderate clinical sensitivity and specificity. [34] While it has been established that abnormal glycosylation occurs in cancers and there has been investigations into multiple different approaches for their detection [35] the efforts to further develop the CA15-3 based diagnostic assay have been limited. The changes in glycosylation can lead to altered interactions of glycoproteins expressed by the tumor cell with different lectins. The development of glycoprofiling assays for blood-derived products has been made difficult by the fact that cancer specific glycovariants may only exist in small amounts in blood and are therefore problematic to detect. We have previously utilized the lectin-NP -based platform successfully to explore the Table 1. CA15-3 levels by conventional CA15-3, CA15-3 MGL , and CA15-3 WGA assay for healthy controls and for metastatic BC patients at study baseline. n Conventional CA15-3 CA15-3 MGL CA15-3 WGA Healthy controls Median CA15-3 U/mL (IQR) 20 13.3 (7.9–23.1) 2.0 (0.2–3.6) 1.6 (0.5–2.7) Metastastic BC patients, Baseline Median CA15-3 U/mL (IQR) 53 47.4 (18.9–99.9) 4.4 (1.3–16.5) 7.0 (3.1–41.0) p-value a <0.001 0.013 <0.001 Abbreviations: n = number of patients, IQR = interquartile range, BC = breast cancer a Mann-Whitney U-test https://doi.org/10.1371/journal.pone.0219480.t001 Table 2. CA15-3 levels with different assays depending on the best response to the chemotherapy treatment. n Baseline Median CA15-3, U/mL (IQR) Week 6 Median CA15-3 U/mL (IQR) Change, median % (IQR) a p b Declining CA15-3 levels, n (%) c Increase in CA15-3 levels, n (%) d Conventional CA15-3 PR 25 71.1 (29.4–228) 55.4 (28.8–103) -23.8 [-52.7-(-14.0)] 0.003 10 (40.0) 2 (8.0) SD 14 19.2 (12.6–81.4) 25.8 (14.9–71.0) -0.4 (-37.1–60.5) 0.875 4 (28.6) 5 (35.7) PD 2 24.1 (15.0–33.1) 29.2 (17.5–40.9) +20.1 (16.7–23.6) 0.180 0 0 CA15-3 MGL PR 25 6.3 (2.1–45.1) 2.4 (0.9–4.8) -75.0 [-86.4-(-41.0)] <0.001 18 (78.2) 3 (13.0) SD 14 3.2 (1.0–5.4) 2.2 (0.8–3.6) -33.3 (-67.0–33.3) 0.036 7 (53.8) 4 (30.7) PD 2 4.1 (3.2–5.0) 3.0 (2.2–3.9) -17.0 (-56.0–21.9) 0.655 1 (50.0) 0 CA15-3 WGA PR 25 13.2 (5.3–76.5) 8.0 (3.5–33.2) -27.2 [-55.9-(-19.4)] <0.001 12 (48.0) 2 (8.0) SD 14 3.2 (2.4–8.7) 5.0 (2.7–8.2) +22.2 (-25.4–60.0) 0.851 3 (21.4) 7 (50.0) PD 2 4.7 (2.3–7.0) 5.8 (2.3–9.3) +16.4 (0–32.9) 0.317 0 1 (50.0) Abbreviations: n = number of patients, CI = confidence interval, PR = partial response, SD = stable disease, PD = progressive disease a Change in CA15-3 levels from baseline to week six in percentiles, median b Wilcoxon Rank Test c Patients with �30% decline in CA15-3 levels from baseline to week six d Patients with �30% increase in CA15-3 levels from baseline to week six https://doi.org/10.1371/journal.pone.0219480.t002 Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 9 / 16 Kellermayer MS, editor. PLoS One. 2017; 12: e0175323. https://doi.org/10.1371/journal.pone.0175323 PMID: 28414807 37. Nollau P, Wolters-Eisfeld G, Mortezai N, Kurze AK, Klampe B, Debus A, et al. Protein Domain Histochemistry (PDH): Binding of the Carbohydrate Recognition Domain (CRD) of Recombinant Human Glycoreceptor CLEC10A (CD301) to Formalin-Fixed, Paraffin-Embedded Breast Cancer Tissues. J Histochem Cytochem. 2013; 61: 199–205. https://doi.org/10.1369/0022155412474823 PMID: 23275449 38. Kubota Y, Fujioka K, Takekawa M. WGA-based lectin affinity gel electrophoresis: A novel method for the detection of O-GlcNAc-modified proteins. PLoS One. Public Library of Science; 2017; 12: e0180714. https://doi.org/10.1371/journal.pone.0180714 PMID: 28686627 39. Beatson R, Maurstad G, Picco G, Arulappu A, Coleman J, Wandell HH, et al. The breast cancer-associated glycoforms of MUC1, MUC1-Tn and sialyl-Tn, are expressed in COSMC wild-type cells and bind the C-type lectin MGL. PLoS One. Public Library of Science; 2015; 10: e0125994. https://doi.org/10. 1371/journal.pone.0125994 PMID: 25951175 40. Chandrasekaran E V., Xue J, Xia J, Khaja SD, Piskorz CF, Locke RD, et al. Novel interactions of complex carbohydrates with peanut (PNA), Ricinus communis (RCA-I), Sambucus nigra (SNA-I) and wheat germ (WGA) agglutinins as revealed by the binding specificities of these lectins towards mucin core-2 O-linked and N-linked glycans a. Glycoconj J. 2016; 33: 819–836. https://doi.org/10.1007/s10719-0169678-y PMID: 27318477 41. Bird-Lieberman EL, Neves AA, Lao-Sirieix P, O’Donovan M, Novelli M, Lovat LB, et al. Molecular imaging using fluorescent lectins permits rapid endoscopic identification of dysplasia in Barrett’s esophagus. Nat Med. 2012; 18: 315–321. https://doi.org/10.1038/nm.2616 PMID: 22245781 42. Van Poznak C, Somerfield MR, Bast RC, Cristofanilli M, Goetz MP, Gonzalez-Angulo AM, et al. Use of biomarkers to guide decisions on systemic therapy for women with metastatic breast cancer: American Society of Clinical Oncology clinical practice guideline. J Clin Oncol. American Society of Clinical Oncology; 2015; 33: 2695–2704. https://doi.org/10.1200/JCO.2015.61.1459 PMID: 26195705 Breast cancer specific CA15-3 lectin assay PLOS ONE | https://doi.org/10.1371/journal.pone.0219480 July 25, 2019 16 / 16