Table 1.
Summary of potential non-invasive biomarkers for early detection of breast cancer.
| Sources | Types | Biomarkers Measured | Detection Types (Sample Size) | Sensitivity | Specificity | Notes | References |
|---|---|---|---|---|---|---|---|
| Peripheral blood | Circulating carcinoma proteins | serum proteins: CEA, FASL, OPN, VEGFC, VEGFD, HGF. | breast cancer (100) vs. non-breast cancer subjects (110). | 74.7%. | 77.0%. | AUC*: 0.79. | [15] |
| tumor-associated autoantibodies: FRS3, RAC3, HOXD1, GPR157, ZMYM6, EIF3E, CSNK1E, ZNF510, BMX, SF3A1, SOX2 | breast cancer (100) vs. non-breast cancer subjects (110) | 72.2% | 70.8% | AUC: 0.77 | |||
| serum proteins and tumor-associated autoantibodies: FASL, IL6, IL8, OPN, VEGFD, HGF, FRS3, MYOZ2, RAC3GPR157, ZMYM6, EIF3E, CSNK1E, ZNF510, BMXSF3A1, SOX2 | breast cancer (100) vs. non-breast cancer subjects (110) | 81.0% | 78.8% | AUC: 0.89 | |||
| Videssa Breast consisting of serum proteins: AFP, CA19-9, CEA, TNF-α, VEGF-C, ErbB2 (HER2); and tumor-associated autoantibodies: ANXA1, ATF3, ATP6AP1, BAT4 (GPANK1), BDNF, CTBP1, DBT, HOXD1, IGF2BP1, IGFBP2, ErbB2 (HER2) | women aged 25–75 (1145) | 93.0% | 64.0% | negative predictive value: 98.0% | [16] | ||
| Videssa Breast consisting of 11 serum protein biomarkers and 33 tumor-associated autoantibodies | women aged under 50 years (545): Breast cancer (32) vs. benign breast tumor (513) | 87.5% | 83.8% | negative predictive value: 99.1% | [17] | ||
| Circulating tumor cells | CellSearch® system | metastatic breast cancer (55) | 47% | N/A | overall positive agreement for both assays: 73% (CTC ≥ 2) and 69% (CTC ≥ 5) | [18] | |
| AdnaTest® assay | metastatic breast cancer (55) | 53% | N/A | ||||
| Circulating cell-free tumor DNA | ctDNA concentrations (≥0.75%) | colorectal and breast cancer patients (10) vs. healthy subjects (10) | >90.0% | >99.0% | [19] | ||
| ctDNA levels (amplifiable per ml of plasma) with CA15-3 expression level | breast cancer (27) | 96.0% | N/A | [20] | |||
| ctDNA levels with CTC numbers | breast cancer (30) | 97.0% | N/A | ||||
| ctDNA levels with PIK3CA E545K and H1047R mutations | early-stage breast cancer (29) | 93.3% | 100.0% | accuracy: 96.7% | [21] | ||
| SCGB3A1 DNA methylation in cfDNA | breast cancer (108) vs. female asymptomatic controls (103) | 16.8% | 80.0% | accuracy: 53.0% | [22,23] | ||
| methylation patterns of six genes in cfDNA: SFN, P16, hMLH1, HOXD13, PCDHGB7, and RASSF1a | breast cancer (125) vs. healthy subjects (104) | 79.6% | 72.4% | AUC: 0.727 | [24] | ||
| methylation analysis of a panel of 16 genes: 12 novel epigenetic markers: JAK3, RASGRF1, CPXM1, SHF, DNM3, CAV2, HOXA10, B3GNT5, ST3GAL6, DACH1, P2RX3, and chr8:23572595; and four internal control markers: CREM, GLYATL3, ELMOD3, and KLF9 |
breast cancer (87) vs. healthy subjects (80) | 86.2% | 82.7% | [25] | |||
| GRAIL DNA methylation patterns of cfDNA for detecting various types of cancers including breast cancer | training set: 844 | N/A | 99.8% | false-positive rate: <1% | [26] | ||
| validation set: 359 | N/A | 99.3% | false-positive rate: <1% | ||||
| GRAIL DNA methylation patterns of cfDNA for detecting breast cancer | training set: Breast cancer (247) | N/A | N/A | precision: 96% (82/85) | |||
| validation set: Breast cancer (104) | N/A | N/A | precision: 93% (40/43) | ||||
| DELFI fragmentation patterns of cfDNA | breast cancer (54) vs. healthy subjects (245) | 57.0% | 98.0% | [27] | |||
| DELFI fragmentation patterns of cfDNA combined with mutations of cfDNA | breast cancer (54) vs. healthy subjects (245) | 65.0% | 98.0% | ||||
| Circulating miRNAs | a panel of five miRNAs: miR-1246, miR-1307-3p, miR-4634, miR-6861-5p, and miR-6875-5p | breast cancer (1206), non-cancer controls (1343), benign breast diseases (54) | 97.3% | 82.9% | accuracy: 89.7% | [28] | |
| a panel of seven miRNAs: Hsa-miR-126-5p, hsa-miR-144-5p, hsa-miR-144-3p, hsa-miR-301a-3p, hsa-miR- 126-3p, hsa-miR-101-3p, and hsa-miR-664b-5p | triple-negative breast cancer (21) vs. healthy subjects (21) | 83.8% | 74.2% | accuracy: 79.0% AUC: 0.814 |
[29] | ||
| Extracellular vesicles | fibronectin | breast cancer (240) vs. non-cancer individuals (205) | 65.1% | 83.2% | AUC: 0.81 | [30] | |
| developmental endothelial locus-1 protein (Del-1) | breast cancer (100) vs. benign breast tumor (38), noncancerous diseases (58), and healthy subjects (46) | 92.31% | 86.62% | [31] | |||
| Metabolites | seven metabolites in the plasma: Glu, Orn, Thr, Trp, Met-SO, C2, and C3. | training cohort: breast cancer (80) vs. healthy subjects (100) validation cohort: breast cancer (109) vs. healthy subjects (50) |
N/A | N/A | AUC in training cohort: 0.87; AUC in validation cohort: 0.80 |
[32] | |
| three metabolites (8-hydroxy-2ʹ-deoxyguanosine, 1-methylguanosine, 1-methyl adenosine) combined with CA15-3 | malignant breast cancer (120) vs. benign breast disease (47) and healthy subjects (55) | 88.8% | 86.8% | AUC: 0.94 | [33] | ||
| seven metabolites in serum: Dimethyldodecane, galactose, α-glyceryl stearate, methyl stearate, 1(1-methoxycarbonyethyl)-4-(2-methyl-2-trimethylsilyl-oxypropyl) benzene, tetradecane, glucopyranoside | pre-operative breast cancer patients (152) vs. healthy subjects (155) | 96% | 100% | [34] | |||
| four plasma metabolites: L-octanoylcarnitine, 5-oxoproline, hypoxanthine, docosahexaenoic acid | discovery set: Breast cancer (40) vs. healthy subjects (30); validation set: Breast cancer (30) vs. healthy subjects (16) |
N/A | N/A | positive predictive value: 100.0% | [35] | ||
| metabolomics signature in the plasma | breast cancer (91) vs. healthy subjects (20) | up to 100.0% | up to 100.0% | [36] | |||
| Lipids | a panel of five serum free fatty acids: C16:1, C18:3, C18:2, C20:4, C22:6. | breast cancer (140) vs. healthy subjects (202) | 83.3% | 87.1% | AUC: 0.953 | [37] | |
| Multi-analyte tests | CancerSEEK testing of eight circulating proteins: CA-125, CA19-9, CEA, HGF, myeloperoxidase, OPN, prolactin, and TIMP-1; and mutations in 16 genes in cfDNA in the blood: NRAS, HRAS, KRAS, CTNNB1, PIK3CA, FBXW7, APC, EGFR, BRAF, CDKN2A, PTEN, FGFR2, AKT1, TP53, PPP2R1A, and GNAS | patients with one of the eight cancer types including breast cancer (1005) vs. individuals without known cancers (812) | 70.0% | >99% | [38] | ||
| breast cancer (209) vs. healthy subjects (812) | 33.0% | >99.0% | |||||
| CancerSEEK tests for different types of cancers | patients with cancers including breast cancer (96) vs. subjects without cancers (9,815) | 27.1% | 98.9% | positive predictive value: 19.4% | [39] | ||
| CancerSEEK tests for breast cancer | breast cancer (27) | 3.7% | N/A | ||||
| CancerSEEK remodeling with CancerA1DE method | first dataset: 1817 patient blood test records second dataset: 626 patient blood test records |
70.0% | 99.0% | remodeling of the CancerSEEK dataset to improve the sensitivity | [40] | ||
| Other body fluids | Urine | miRNAs (miR-21, miR-34a, miR-125b, miR-155, miR-195, miR-200b, miR-200c, miR-375, miR-451) | breast cancer (24) vs. healthy subjects (24) | 91.7% | 91.7% | AUC: 0.932 | [41] |
| four urinary microRNA types (miR-424, miR-423, miR-660, and let7-i) | breast cancer (69) vs. healthy subjects (40) | 98.6% | 100.0% | [42] | |||
| miRNA-21 and MMP-1 | breast cancer (22) vs. healthy subjects (26) | 95.0% | 79.0% | [43] | |||
| succinic acid and dimethyl-heptanoylcarnitine | breast cancer (31) vs. healthy subjects (29) | 93.5% | 86.2% | [44] | |||
| Breath | a set of VOCs of oxidative stress | breast cancer (51) vs. healthy subjects (42) | 94.1% | 73.8% | [45] | ||
| 3-methylhexane | breast cancer (10) vs. healthy subjects (10) | 100% | 40% | [46] | |||
| decene | 100% | 40% | |||||
| caryophyllene | 100% | 60% | |||||
| naphthalene | 90% | 70% | |||||
| trichlorethylene | 80% | 70% | |||||
| five breath VOCs: 2-propanol, 2,3-dihydro-1-phenyl-4(1H)-quinazolinone,1-phenyl-ethanone, heptanal and isopropyl myristate | breast cancer (51) vs. healthy subjects (42) | 93.8% | 84.6% | [47] | |||
| two commercial electronic noses | breast cancer (48) vs. healthy subjects (45) | N/A | N/A | an average of 95% accuracy | [48] | ||
| BreathLink™ point-of-care breath testing systems | breast cancer (50) vs. non-breast cancer subjects (543) | 82.0% | 77.1% | accuracy: 83% | [49] | ||
| VOCs collected by ultra-clean breath collection balloons | breast cancer (54) vs. non-breast cancer subjects (124) | N/A | N/A | low DF values: Negative predictive value > 99.9%; high DF values: Positive predictive value rising to 100% |
[50] | ||
| NAF | Thomsen–Freidenreich (TF) antigen and its biosynthetic precursor Tn antigen | breast cancer (25) vs. healthy subjects (25) | N/A | N/A | detected in 92% of the cancerous breast NAF samples | [51] | |
| TF, Tn, and age information | breast cancer (83) vs. benign disease (41) | N/A | N/A | AUC: 0.83 | [52] | ||
| combination of TF and uPA | breast cancer (83) vs. benign disease (41) | N/A | N/A | accuracy: 84-92% | [53] | ||
| combination of TF, uPA and PAI-1 | breast cancer (83) vs. benign disease (41) | N/A | N/A | predictive ability reached 100% | |||
| deglycase DJ-1 protein | 136 patients with nipple discharge (benign: 63; malignant: 73) | 75.0% | 85.9% | [54] | |||
| proteomic profiles of NAFs | breast cancer (18) vs. healthy subjects (4) | N/A | N/A | 39 proteins differentially expressed in tumor-bearing vs. disease-free breasts | [55] | ||
| dehydroepiandrosterone (DHEA) concentration | breast cancer (160) vs. healthy subjects (157) | N/A | N/A | higher DHEA concentrations in NAFs were associated with breast cancer | [56] | ||
| Tears | proteomic profiles of tears | breast cancer (10) vs. healthy subjects (10) | ~90.0% | ~90.0% | [57] | ||
| a panel of 20 proteins in tears | breast cancer (50) vs. healthy subjects (50) | ~70.0% | ~70.0% | [58] | |||
| MALDI-TOF-TOF-MS-driven semi-quantitative comparison of tear protein levels | breast cancer (25) vs. healthy subjects (25) | N/A | N/A | more than 20 proteins were differentially expressed in the tears | [59] | ||
| surface-enhanced Raman scattering (SERS) spectra of tear fluid | breast cancer (5) vs. healthy subjects (5) | 92% | 100% | accuracy: 96% | [60] | ||
| Apocrine sweat | 20 sweat markers | breast cancer (70) vs. healthy subjects (53) | 97% | 72% | [61] |
* AUC, area under the receiver operating characteristic (ROC) curve; CTC, circulating tumor cell; cfDNA, cell-free DNA; ctDNA, circulating tumor DNA; DELFI, DNA evaluation of fragments for early interception; MALDI-TOF-TOF-MS, matrix-assisted laser desorption/ionization time-of-flight/time-of-flight mass spectrometry; NAF, nipple aspirate fluid; uPA, urinary plasminogen activator; PAI-1, plasminogen activator inhibitor-1; VOC, volatile organic compound.