Deep Learning Prediction of Pathologic Complete Response in Breast Cancer Using MRI and Other Clinical Data: A Systematic Review.

Breast cancer patients who have pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) are more likely to have better clinical outcomes. The ability to predict which patient will respond to NAC early in the treatment course is important because it could help to minimize unnecessary toxic NAC and to modify regimens mid-treatment to achieve better efficacy. Machine learning (ML) is increasingly being used in radiology and medicine because it can identify relationships amongst complex data elements to inform outcomes without the need to specify such relationships a priori. One of the most popular deep learning methods that applies to medical images is the Convolutional Neural Networks (CNN). In contrast to supervised ML, deep learning CNN can operate on the whole images without requiring radiologists to manually contour the tumor on images. Although there have been many review papers on supervised ML prediction of pCR, review papers on deep learning prediction of pCR are sparse. Deep learning CNN could also incorporate multiple image types, clinical data such as demographics and molecular subtypes, as well as data from multiple treatment time points to predict pCR. The goal of this study is to perform a systematic review of deep learning methods that use whole-breast MRI images without annotation or tumor segmentation to predict pCR in breast cancer.

Convolutional Neural Network of Multiparametric MRI Accurately Detects Axillary Lymph Node Metastasis in Breast Cancer Patients With Pre Neoadjuvant Chemotherapy.

BACKGROUND: Accurate assessment of the axillary lymph nodes (aLNs) in breast cancer patients is essential for prognosis and treatment planning. Current radiological staging of nodal metastasis has poor accuracy. This study aimed to investigate the machine learning convolutional neural networks (CNNs) on multiparametric MRI to detect nodal metastasis with 18FDG-PET as ground truths. MATERIALS AND METHODS: Data were obtained via a retrospective search. Inclusion criteria were patients with bilateral breast MRI and 18FDG-PETand/or CT scans obtained before neoadjuvant chemotherapy. In total, 238 aLNs were obtained from 56 breast cancer patients with 18FDG-PET and/or CT and breast MRI data. Radiologists scored each node based on all MRI as diseased and non-diseased nodes. Five models were built using T1-W MRI, T2-W MRI, DCE MRI, T1-W + T2-W MRI, and DCE + T2-W MRI model. Performance was evaluated using receiver operating curve (ROC) analysis, including area under the curve (AUC). RESULTS: All CNN models yielded similar performance with an accuracy ranging from 86.08% to 88.50% and AUC ranging from 0.804 to 0.882. The CNN model using T1-W MRI performed better than that using T2-W MRI in detecting nodal metastasis. CNN model using combined T1- and T2-W MRI performed the best compared to all other models (accuracy = 88.50%, AUC = 0.882), but similar in AUC to the DCE + T2-W MRI model (accuracy = 88.02%, AUC = 0.880). All CNN models performed better than radiologists in detecting nodal metastasis (accuracy = 65.8%). CONCLUSION: xxxxxx.

Machine learning classification of texture features of MRI breast tumor and peri-tumor of combined pre- and early treatment predicts pathologic complete response.

PURPOSE: This study used machine learning classification of texture features from MRI of breast tumor and peri-tumor at multiple treatment time points in conjunction with molecular subtypes to predict eventual pathological complete response (PCR) to neoadjuvant chemotherapy. MATERIALS AND METHOD: This study employed a subset of patients (N = 166) with PCR data from the I-SPY-1 TRIAL (2002-2006). This cohort consisted of patients with stage 2 or 3 breast cancer that underwent anthracycline-cyclophosphamide and taxane treatment. Magnetic resonance imaging (MRI) was acquired pre-neoadjuvant chemotherapy, early, and mid-treatment. Texture features were extracted from post-contrast-enhanced MRI, pre- and post-contrast subtraction images, and with morphological dilation to include peri-tumoral tissue. Molecular subtypes and Ki67 were also included in the prediction model. Performance of classification models used the receiver operating characteristics curve analysis including area under the curve (AUC). Statistical analysis was done using unpaired two-tailed t-tests. RESULTS: Molecular subtypes alone yielded moderate prediction performance of PCR (AUC = 0.82, p = 0.07). Pre-, early, and mid-treatment data alone yielded moderate performance (AUC = 0.88, 0.72, and 0.78, p = 0.03, 0.13, 0.44, respectively). The combined pre- and early treatment data markedly improved performance (AUC = 0.96, p = 0.0003). Addition of molecular subtypes improved performance slightly for individual time points but substantially for the combined pre- and early treatment (AUC = 0.98, p = 0.0003). The optimal morphological dilation was 3-5 pixels. Subtraction of post- and pre-contrast MRI further improved performance (AUC = 0.98, p = 0.00003). Finally, among the machine-learning algorithms evaluated, the RUSBoosted Tree machine-learning method yielded the highest performance. CONCLUSION: AI-classification of texture features from MRI of breast tumor at multiple treatment time points accurately predicts eventual PCR. Longitudinal changes in texture features and peri-tumoral features further improve PCR prediction performance. Accurate assessment of treatment efficacy early on could minimize unnecessary toxic chemotherapy and enable mid-treatment modification for patients to achieve better clinical outcomes.

Physiological substrates and ontogeny-specific expression of the ubiquitin ligases MARCH1 and MARCH8.

MARCH1 and MARCH8 are ubiquitin ligases that control the expression and trafficking of critical immunoreceptors. Understanding of their function is hampered by three major knowledge gaps: (i) it is unclear which cell types utilize these ligases; (ii) their level of redundancy is unknown; and (iii) most of their putative substrates have been described in cell lines, often overexpressing MARCH1 or MARCH8, and it is unclear which substrates are regulated by either ligase in vivo. Here we address these questions by systematically analyzing the immune cell repertoire of MARCH1- or MARCH8-deficient mice, and applying unbiased proteomic profiling of the plasma membrane of primary cells to identify MARCH1 and MARCH8 substrates. Only CD86 and MHC II were unequivocally identified as immunoreceptors regulated by MARCH1 and MARCH8, but each ligase carried out its function in different tissues. MARCH1 regulated MHC II and CD86 in professional and "atypical" antigen presenting cells of hematopoietic origin, including neutrophils, eosinophils and monocytes. MARCH8 only operated in non-hematopoietic cells, such as thymic and alveolar epithelial cells. Our results establish the tissue-specific functions of MARCH1 and MARCH8 in regulation of immune receptor expression and reveal that the range of cells constitutively endowed with antigen-presentation capacity is wider than generally appreciated.

Convolutional Neural Network Detection of Axillary Lymph Node Metastasis Using Standard Clinical Breast MRI.

BACKGROUND: Axillary lymph node status is important for breast cancer staging and treatment planning as the majority of breast cancer metastasis spreads through the axillary lymph nodes. There is currently no reliable noninvasive imaging method to detect nodal metastasis associated with breast cancer. MATERIALS AND METHODS: Magnetic resonance imaging (MRI) data were those from the peak contrast dynamic image from 1.5 Tesla MRI scanners at the pre-neoadjuvant chemotherapy stage. Data consisted of 66 abnormal nodes from 38 patients and 193 normal nodes from 61 patients. Abnormal nodes were those determined by expert radiologist based on 18Fluorodeoxyglucose positron emission tomography images. Normal nodes were those with negative diagnosis of breast cancer. The convolutional neural network consisted of 5 convolutional layers with filters from 16 to 128. Receiver operating characteristic analysis was performed to evaluate prediction performance. For comparison, an expert radiologist also scored the same nodes as normal or abnormal. RESULTS: The convolutional neural network model yielded a specificity of 79.3% ± 5.1%, sensitivity of 92.1% ± 2.9%, positive predictive value of 76.9% ± 4.0%, negative predictive value of 93.3% ± 1.9%, accuracy of 84.8% ± 2.4%, and receiver operating characteristic area under the curve of 0.91 ± 0.02 for the validation data set. These results compared favorably with scoring by radiologists (accuracy of 78%). CONCLUSION: The results are encouraging and suggest that this approach may prove useful for classifying lymph node status on MRI in clinical settings in patients with breast cancer, although additional studies are needed before routine clinical use can be realized. This approach has the potential to ultimately be a noninvasive alternative to lymph node biopsy.

MRI Volume Changes of Axillary Lymph Nodes as Predictor of Pathologic Complete Responses to Neoadjuvant Chemotherapy in Breast Cancer.

INTRODUCTION: Longitudinal monitoring of breast tumor volume over the course of chemotherapy is informative of pathologic response. This study aims to determine whether axillary lymph node (aLN) volume by magnetic resonance imaging (MRI) could augment the prediction accuracy of treatment response to neoadjuvant chemotherapy (NAC). MATERIALS AND METHODS: Level-2a curated data from the I-SPY-1 TRIAL (2002-2006) were used. Patients had stage 2 or 3 breast cancer. MRI was acquired pre-, during, and post-NAC. A subset with visible aLNs on MRI was identified (N = 132). Prediction of pathologic complete response (PCR) was made using breast tumor volume changes, nodal volume changes, and combined breast tumor and nodal volume changes with sub-stratification with and without large lymph nodes (3 mL or ∼1.79 cm diameter cutoff). Receiver operating characteristic curve analysis was used to quantify prediction performance. RESULTS: The rate of change of aLN and breast tumor volume were informative of pathologic response, with prediction being most informative early in treatment (area under the curve (AUC), 0.57-0.87) compared with later in treatment (AUC, 0.50-0.75). Larger aLN volume was associated with hormone receptor negativity, with the largest nodal volume for triple negative subtypes. Sub-stratification by node size improved predictive performance, with the best predictive model for large nodes having AUC of 0.87. CONCLUSION: aLN MRI offers clinically relevant information and has the potential to predict treatment response to NAC in patients with breast cancer.

Protein profiles distinguish stable and progressive chronic lymphocytic leukemia.

Patients with a stable chronic lymphocytic leukemia (CLL) double their blood lymphocyte count in >5 years, but may develop progressive disease with lymphocytes doubling in <12 months. To identify a protein signature for progressive CLL, whole cell extracts of peripheral blood mononuclear cells from patients with CLL (n=27) were screened using iTRAQ (isobaric tags for relative and absolute quantification) analysis. A total of 84 differentially abundant proteins were identified from patients with stable and progressive CLL. Subsequently, 32 of these proteins were quantified by SRM (selected reaction monitoring) using extracts of purified CD19+ CLL cells from patients (n=50). Hierarchical clustering of these protein profiles showed two clusters of patients that correlated with progressive and stable CLL, providing signatures that should be useful for triaging patients. Some of the proteins in the progressive cluster have not been linked with CLL, for example, glutamate dehydrogenase 1 and transcription intermediary factor 1-beta.

Analysis of Post-Liver Transplant Hepatitis C Virus Recurrence Using Serial Cluster of Differentiation Antibody Microarrays.

BACKGROUND: Hepatitis C virus (HCV) reinfection of the liver allograft after transplantation is universal, with some individuals suffering severe disease recurrence. Predictive markers of recurrent disease severity are urgently needed. In this study, we used a cluster of differentiation (CD) microarray to predict the severity of HCV recurrence after transplantation. METHODS: The CD antibody microarray assays of live leukocytes were performed on peripheral blood taken in the first year after transplantation. The results were grouped into phases defined as; Pre-transplant (day 0), Early (day 3 to week 2), Mid (week 4 to week 10), and Late (week 12 to week 26). Hepatitis C virus severity was based on fibrosis stages in the first 2 years (F0-1 mild and F2-4 severe). RESULTS: Serial blood samples from 16 patients were taken before and after liver transplantation. A total of 98 assays were performed. Follow-up was 3 years or longer. Comparing recurrence severity, significantly greater numbers of CD antigens were differentially expressed on the pretransplant samples compared to any posttransplant timepoints. Five differentially expressed CD antigens before transplantation (CD27 PH, CD182, CD260, CD41, and CD34) were significantly expressed comparing severe to mild recurrence, whereas expression of only CD152 was significant in the late phase after transplantation. No relationship was observed between the donor or recipient interleukin-28B genotypes and HCV recurrence severity. CONCLUSIONS: This study shows that circulating leukocyte CD antigen expression has utility in assessing recurrent HCV disease severity after liver transplantation and serves as a proof of principle. Importantly, pretransplant CD antigen expression is most predictive of disease outcome.

Cell surface phenotype profiles distinguish stable and progressive chronic lymphocytic leukemia.

Chronic lymphocytic leukemia (CLL) is clinically heterogeneous. While some patients have indolent disease for many years, 20-30% will progress and ultimately die of their disease. CLL may be classified by the Rai or Binet staging system, mutational status of the immunoglobulin variable heavy-chain gene (IGVH), ZAP-70 overexpression, cytogenetic abnormalities (13q-, + 12, 11q-, 17p-) and expression of several cell surface antigens (CD38, CD49d) that correlate with risk of disease progression. However, none of these markers identify all cases of CLL at risk. In a recent review, we summarized those CD antigens known to correlate with the prognosis of CLL. The present study has identified surface profiles of CD antigens that distinguish clinically progressive CLL from slow-progressive and stable CLL. Using an extended DotScan(™) CLL antibody microarray (Version 3; 182 CD antibodies), and with refined analysis of purified CD19 + B-cells, the following 27 CD antigens were differentially abundant for progressive CLL: CD11a, CD11b, CD11c, CD18, CD19, CD20 (two epitopes), CD21, CD22, CD23, CD24, CD25, CD38, CD40, CD43, CD45, CD45RA, CD52, CD69, CD81, CD84, CD98, CD102, CD148, CD180, CD196 and CD270. The extensive surface profiles obtained provide disease signatures with an accuracy of 79.2%, a sensitivity of 83.9% and a specificity of 72.5% that could provide the basis for a rapid test to triage patients with CLL according to probability of clinical progression and potential earlier requirement for treatment.

Profiles of surface mosaics on chronic lymphocytic leukemias distinguish stable and progressive subtypes.

PURPOSE: Chronic lymphocytic leukemia (CLL) is a heterogeneous disease, some patients may survive for many years, while 20-30% of patients progress and may die within several years. Currently, there is not a single procedure that enables accurate prognosis and triaging of those patients who need immediate and aggressive treatment. All CLL cells are characterised by the expression of the B-cell antigens CD19, CD20, CD21, CD22 and CD23, with aberrant expression of the T-cell antigen CD5. METHODS: We have developed a CD antibody microarray (DotScan) containing 182 immobilised CD antibodies that has been used to obtain extensive surface profiles of CLL cells obtained from 96 patients. RESULTS: Of these 182 antigens, 27 were significantly differentially expressed between stable, stable-progressive and progressive CLL. Some of these antigens are not expressed on normal B-cells and may be targets for therapeutic antibodies against CLL. Unsupervised hierarchical clustering of the surface profiles from 96 patients showed that those with progressive CLL could be distinguished based solely upon this 'disease signature'. The sensitivity (proportion of actual positives correctly identified) was 67.9%, the specificity (proportion of negatives correctly identified) was 77.5%, and the accuracy was 71.9%. CONCLUSIONS: Considerable effort by a number of research groups has resulted in identification of individual markers for progressive CLL, but their collective use is yet to provide a test that identifies CLL patients at risk. Data presented here provide a basis for development of a simple test using an antibody microarray.

Analysis of human liver disease using a cluster of differentiation (CD) antibody microarray.

BACKGROUND: A CD antibody microarray has been previously developed allowing semi-quantitative identification of greater than 80 CD antigens on circulating leucocytes from peripheral blood samples. This assay, which uses a live cell-capture technique, enables an extensive leucocyte immunophenotype determination in a single analysis and to date this has been used successfully to characterise diseases including human leukaemias and HIV infection. AIMS: To determine CD antigen expression profiles for patients with various liver diseases and to look for preserved disease-specific signatures. METHODS: Three liver disease groups including hepatitis C (HCV) (n = 35), non-alcoholic steatohepatitis (NASH) (n = 21) and alcohol-related liver disease (n = 14) were compared with a normal group (n = 23). Hierarchal Clustering (HCL) and Principal Component Analysis (PCA) of the data revealed distinct binding patterns for patients with and without cirrhosis. RESULTS: Patients with cirrhosis and portal hypertension compared with those without cirrhosis had significantly reduced expression of several markers of T-cell function including CD45, CD8, CD28 and TCR α/ß. Disease prediction algorithms based on the expression data were able to discriminate cirrhotics from non-cirrhotics with 71% overall success, which improved to 77% when only patients with HCV were considered. CONCLUSIONS: These results demonstrate disease-specific consensus patterns of expression of CD antigens for patients with chronic liver disease, suggesting that the CD antibody array is a promising tool in the analysis of human liver disease, and with further refinement may have future research and clinical utility.

Surface profiles for subclassification of chronic lymphocytic leukemia.

Chronic lymphocytic leukemia (CLL) has a variable clinical course. Some patients have stable disease while others progress and require treatment. Levels of several cluster of differentiation (CD) antigens are known to correlate with prognosis and may be used to stratify patients according to risk. In this review, we summarize current information on surface CD antigens found on CLL, their pathological significance and their detection using CD antibody microarrays. The use of extensive immunophenotypic patterns or surface profiles as disease signatures for CLL subclassification, prognosis and patient management is discussed with a focus on triaging patients with CLL with progressive disease.

An extended antibody microarray for surface profiling metastatic melanoma.

An antibody microarray was developed for profiling the surface proteome of melanoma cells, which may facilitate melanoma sub-classification and provide important prognostic information useful in predicting the clinical behavior of the melanoma (e.g., likely sites of metastatic spread), patient outcome and treatment response. Forty-eight antibodies were selected based on their correlation with melanoma development, progression and/or prognosis and printed on nitrocellulose slides. The immobilised antibodies capture live cells expressing corresponding antigens to produce a cell binding dot pattern representing the surface antigen profile (immunophenotype) of the melanoma. Surface antigen signatures were determined for a normal melanocyte and 6 melanoma cell lines and cell suspensions prepared from 10 surgically excised melanoma lymph node metastases. A procedure for obtaining separate surface antigen profiles for melanoma cells and leukocytes from clinical lymph node samples was also developed using anti-CD45 magnetic beads. The capture of live, bead-bound leukocytes on these antibody microarrays provides a significant enhancement of this microarray technology. The antibody microarray will be used to profile panels of surgically excised melanoma lymph node metastases (melanoma and leukocyte fractions) to determine whether the immunophenotypes correlate with clinicopathological characteristics, disease progression and clinical outcome.

Surface antigen profiling of colorectal cancer using antibody microarrays with fluorescence multiplexing.

A procedure is described for the disaggregation of colorectal cancers (CRC) and normal intestinal mucosal tissues to produce suspensions of viable single cells, which are then captured on customized antibody microarrays recognising 122 different surface antigens (DotScan CRC microarray). Cell binding patterns recorded by optical scanning of microarrays provide a surface profile of antigens on the cells. Sub-populations of cells bound on the microarray can be profiled by fluorescence multiplexing using monoclonal antibodies tagged with Quantum Dots or other fluorescent dyes. Surface profiles are presented for 6 CRC cell lines (T84, LIM1215, SW480, HT29, CaCo and SW620) and surgical samples from 40 CRC patients. Statistical analysis revealed significant differences between profiles for CRC samples and mucosal controls. Hierarchical clustering of CRC data identified several disease clusters that showed some correlation with clinico-pathological stage as determined by conventional histopathological analysis. Fluorescence multiplexing using Phycoerythrin- or Alexa Fluor 647-conjugated antibodies was more effective than multiplexing with antibodies labelled with Quantum Dots. This relatively simple method yields a large amount of information for each patient sample and, with further application, should provide disease signatures and enable the identification of patients with good or poor prognosis.

Cladribine and Fludarabine Nucleoside Change the Levels of CD Antigens on B-Lymphoproliferative Disorders.

The purine analogs, fludarabine nucleoside (FdA), and cladribine (CdA) (1 µM, 24 hours), significantly changed the levels of some surface antigens on the human B-cell lines MEC2 and Raji. Changes in the surface proteins were identified using a Cluster of Differentiation (CD) antibody microarray that captures live cells and confirmed by flow cytometry. For Raji cells, CdA up-regulated CD10, CD54, CD80, and CD86, with repression of CD22, while FdA up-regulated CD20, CD54, CD80, CD86 and CD95. For MEC2 cells, CdA up-regulated CD11a, CD20, CD43, CD45, CD52, CD54, CD62L, CD80, CD86, and CD95, but FdA had no effect. Up-regulation of particular CD antigens induced on a B-cell lymphoproliferative disorder by a purine analog could provide targets for therapeutic antibodies with synergistic cell killing.

C-reactive protein, an 'intermediate phenotype' for inflammation: human twin studies reveal heritability, association with blood pressure and the metabolic syndrome, and the influence of common polymorphism at catecholaminergic/beta-adrenergic pathway loci.

BACKGROUND: C-reactive protein (CRP) both reflects and participates in inflammation, and its circulating concentration marks cardiovascular risk. Here we sought to understand the role of heredity in determining CRP secretion. METHODS: CRP, as well as multiple facets of the metabolic syndrome, were measured in a series of 229 twins, both monozygotic (MZ) and dizygotic (DZ), to estimate trait heritability (h2). Single nucleotide polymorphism (SNP) genotyping was done at adrenergic pathway loci. Haplotypes were inferred from genotypes by likelihood methods. Association of CRP with hypertension and the metabolic syndrome was studied in a larger series of 732 individuals, including 79 with hypertension. RESULTS: MZ and DZ twin variance components indicated substantial h2 for CRP, at approximately 56 +/- 7% (P < 0.001). CRP was significantly associated (P < 0.05) with multiple features of the metabolic syndrome in twins, including body mass index (BMI), blood pressure (BP), leptin and lipids. In established hypertension, elevated CRP was associated with increased BP, BMI, insulin, HOMA (index of insulin resistance), leptin, triglycerides and norepinephrine. Twin correlations indicated pleiotropy (shared genetic determination) for CRP with BMI (P = 0.0002), leptin (P < 0.001), triglycerides (P = 0.002) and systolic blood pressure (SBP) (P = 0.042). Approximately 9800 genotypes (43 genetic variants at 17 loci) were scored within catecholaminergic pathways: biosynthetic, receptor and signal transduction. Plasma CRP concentration in twins was predicted by polymorphisms at three loci in physiological series within the catecholamine biosynthetic/beta-adrenergic pathway: TH (tyrosine hydroxylase), ADRB1 (beta1-adrenergic receptor) and ADRB2 (beta2-adrenergic receptor). In the TH promoter, common allelic variation accounted for up to approximately 6.6% of CRP inter-individual variance. At ADRB1, variation at Gly389Arg predicted approximately 2.8% of CRP, while ADRB2 promoter variants T-47C and T-20C also contributed. Particular haplotypes and diplotypes at TH and ADRB1 also predicted CRP, though typically no better than single SNPs alone. Epistasis (gene-by-gene interaction) was demonstrated for particular combinations of TH and ADRB2 alleles, consistent with their actions in a pathway in series. In an illustration of pleiotropy, not only CRP but also plasma triglycerides were predicted by polymorphisms at TH (P = 0.0053) and ADRB2 (P = 0.027). CONCLUSIONS: CRP secretion is substantially heritable in humans, demonstrating pleiotropy (shared genetic determination) with other features of the metabolic syndrome, such as BMI, triglycerides or BP. Multiple, common genetic variants in the catecholaminergic/beta-adrenergic pathway contribute to CRP, and these variants (especially at TH and ADRB2) seem to interact (epistasis) to influence the trait. The results uncover novel pathophysiological links between the adrenergic system and inflammation, and suggest new strategies to probe the role and actions of inflammation within this setting.

Analysis of human leukaemias and lymphomas using extensive immunophenotypes from an antibody microarray.

A novel antibody microarray has been developed that provides an extensive immunophenotype of leukaemia cells. The assay is a solid phase cell-capture technique in which 82 antigens are studied simultaneously. This paper presents the analysis of 733 patients with a variety of leukaemias and lymphomas from peripheral blood and bone marrow. Discriminant Function Analysis of the expression profiles from these 733 patients and 63 normal subjects were clustered and showed high levels of consistency with diagnoses obtained using conventional clinical and laboratory criteria. The overall levels of consensus for classification using the microarray compared with established criteria were 93.9% (495/527 patients) for peripheral blood and 97.6% (201/206 patients) for bone marrow aspirates, showing that the extensive phenotype alone was frequently able to classify the disease when the leukaemic clone was the dominant cell population present. Immunophenotypes for neoplastic cells were distinguishable from normal cells when the leukaemic cell count was at least 5 x 10(9) cells/l in peripheral blood, or 20% of cells obtained from bone marrow aspirates. This technique may be a useful adjunct to flow cytometry and other methods when an extensive phenotype of the leukaemia cell is desired for clinical trials, research and prognostic factor analysis.

Multiplex detection of surface molecules on colorectal cancers.

A technique of fluorescence multiplexing is described for analysis of the plasma membrane proteome of colorectal cancer cells from surgically resected specimens, enabling detection and immunophenotyping when the cancer cells are in the minority. A single-cell suspension was prepared from a colorectal tumour, and the mixed population of cells was captured on a CD antibody microarray. The cancer cells were detected using a fluorescently tagged antibody for carcinoembryonic antigen (CEA-Alexa647) or epithelial cell adhesion marker (EpCAM-Alexa488). Using this multiplexing procedure, dot patterns from colorectal cancers were distinct from those of adjacent normal tissue. Subtraction of the expression levels for each antigen from normal tissue from those for the cancer shows differential expression in the cancer of CD66c, CD15s, CD55, CD45, CD71, CD45RO, CD11b and CEA, in descending order. Cells captured on the same microarray were also labelled with fluorescent CD3-phycoerythrin antibody revealing the presence of tumour-infiltrating lymphocytes. The immunophenotypes of T lymphocytes from the tumour samples showed differential expression of HLA-DR, TCR alpha/beta, CD49d, CD52, CD49e, CD5, CD95, CD28, CD38 and CD71, in descending order. Fluorescence multiplexing of mixed cell populations captured on a single antibody microarray enables expression profiling of multiple sub-populations of cells within a tumour sample.

Screening microarrays of novel monoclonal antibodies for binding to T-, B- and myeloid leukaemia cells.

We have developed a microarray (DotScan) that enables rapid immunophenotyping and classification of leukaemias and lymphomas by measuring the capture of cells by immobilized dots of 82 CD antibodies [Belov, L., de la Vega, O., dos Remedios, C.G., Mulligan, S.P., 2001. Immunophenotyping of leukemia using a cluster of differentiation antibody microarray. Cancer Res. 61, 4483; Belov, L., Huang, P., Barber, N., Mulligan, S.P., Christopherson, R.I., 2003. Identification of repertoires of surface antigens on leukemias using an antibody microarray. Proteomics 3, 2147]. The DotScan technology has been used to investigate the properties of 498 new antibodies submitted to the HLDA8 Workshop. These antibodies have been applied as 10 nl dots to a film of nitrocellulose on a microscope slide to make an HLDA8 microarray. After blocking the remaining nitrocellulose surface, individual arrays were incubated with each of 7 cell types from a human leukaemia cell panel consisting of three cell lines, CCRF-CEM (a T-cell acute lymphocytic leukaemia), MEC-1 (derived from B-cell chronic lymphocytic leukaemia) and HL-60 (a promyelocytic leukaemia), and four leukaemias from patients: a T-cell prolymphocytic leukaemia, a B-cell chronic lymphocytic leukaemia, and two acute myeloid leukaemias. Leukaemia cells were captured by those immobilized antibodies for which they expressed the corresponding surface molecule. Unbound cells were gently washed off, bound cells were fixed to the arrays and dot patterns were recorded using a DotScan array reader and quantified using DotScan data analysis software. The data obtained show the unique expression profiles of the 7 cell types in the leukaemia cell panel obtained with the DotScan microarray, and the differential capture patterns for these 7 cell types screened against the 498 antibodies in the HLDA8 microarray constructed for this study.

Identification of repertoires of surface antigens on leukemias using an antibody microarray.

We have previously described a microarray of cluster of differentiation (CD) antibodies that enables concurrent determination of more than 60 CD antigens on leukocytes. This procedure does not require protein purification or labeling, or a secondary detection system. Whole cells are captured by a microarray of 10 nL antibody dots immobilized on a nitrocellulose film on a microscope slide. Distinct patterns of cell binding are observed for different leukemias or lymphomas. These haematological malignancies arise from precursor cells of T- or B-lymphocytic, or myeloid lineages of hematopoiesis. The dot patterns obtained from patients are distinct from those of peripheral blood leukocytes from normal subjects. This microarray technology has recently undergone a number of refinements. The microarray now contains more CD antibodies, and a scanner for imaging dot patterns and software for data analysis provide an extensive immunophenotype sufficient for diagnosis of common leukemias. The technology is being evaluated for diagnosis of leukemias with parallel use of conventional diagnostic criteria.