Deep Learning-Based H-Score Quantification of Immunohistochemistry-Stained Images.

Immunohistochemistry (IHC) is a well-established and commonly used staining method for clinical diagnosis and biomedical research. In most IHC images, the target protein is conjugated with a specific antibody and stained using diaminobenzidine (DAB), resulting in a brown coloration, whereas hematoxylin serves as a blue counterstain for cell nuclei. The protein expression level is quantified through the H-score, calculated from DAB staining intensity within the target cell region. Traditionally, this process requires evaluation by 2 expert pathologists, which is both time consuming and subjective. To enhance the efficiency and accuracy of this process, we have developed an automatic algorithm for quantifying the H-score of IHC images. To characterize protein expression in specific cell regions, a deep learning model for region recognition was trained based on hematoxylin staining only, achieving pixel accuracy for each class ranging from 0.92 to 0.99. Within the desired area, the algorithm categorizes DAB intensity of each pixel as negative, weak, moderate, or strong staining and calculates the final H-score based on the percentage of each intensity category. Overall, this algorithm takes an IHC image as input and directly outputs the H-score within a few seconds, significantly enhancing the speed of IHC image analysis. This automated tool provides H-score quantification with precision and consistency comparable to experienced pathologists but at a significantly reduced cost during IHC diagnostic workups. It holds significant potential to advance biomedical research reliant on IHC staining for protein expression quantification.

Clinicopathologic Predictors of Clinical Outcomes in Mammary Adenoid Cystic Carcinoma: A Multi-institutional Study.

Adenoid cystic carcinoma (AdCC) is an uncommon type of invasive breast carcinoma with a favorable prognosis. However, some cases are aggressive. The study aims to define the clinicopathologic predictors of outcome. Clinical, radiological, and pathologic variables were recorded for 76 AdCC cases from 11 institutions. The following histologic characteristics were evaluated by the breast pathologist in each respective institution, including Nottingham grade (NG), percentages of various growth patterns (solid, cribriform, trabecular-tubular), percentage of basaloid component, tumor borders (pushing, infiltrative), perineural invasion, lymphovascular invasion, necrosis, and distance from the closest margin. Various grading systems were evaluated, including NG, salivary gland-type grading systems, and a new proposed grading system. The new grading system incorporated the growth pattern (percent solid, percent cribriform), percent basaloid morphology, and mitotic count using the Youden index criterion. All variables were correlated with recurrence-free survival. Nineteen (25%) women developed local and/or distant recurrence. Basaloid morphology (≥25% of the tumor) was identified in 20 (26.3%) cases and a solid growth pattern (using ≥60% cutoff) in 22 (28.9%) cases. In the univariate analysis, the following variables were significantly correlated with worse recurrence-free survival: solid growth pattern, basaloid morphology, lymphovascular invasion, necrosis, perineural invasion, and pN-stage. In the multivariate analysis including basaloid morphology, pN-stage, lymphovascular invasion, and perineural invasion, basaloid morphology was statistically significant, with a hazard ratio of 3.872 (95% CI, 1.077; 13.924; P =.038). The NG and the new grading system both correlated with recurrence-free survival. However, grade 2 had a similar risk as grade 3 in the NG system and a similar risk as grade 1 in the new grading system. For solid growth patterns and basaloid morphology, using a 2-tier system with 1 cutoff was better than a 3-tier system with 2 cutoffs. Basaloid morphology and solid growth pattern have prognostic values for AdCC, with a 2-tier grading system performing better than a 3-tier system.

Interferon-induced transmembrane protein 2 promotes epithelial-mesenchymal transition by activating transforming growth factor-ß1/small mother against decapentaplegic 2 signaling in gastric cancer.

BACKGROUND: Gastric cancer (GC) is one of the most prevalent malignancy around the world. Primary tumor cells are enabled to invade and migrate into adjacent normal tissues to form secondary tumors. Epithelial-mesenchymal transitions (EMT) plays a pivotal role in facilitating tumor progression. Abundant evidence suggested that the transforming growth factor-ß1 (TGF-ß1) triggered the process of EMT. Nonetheless, the precise molecular mechanisms underlying EMT requires further elucidation, and there still lacks effective specific therapeutic target. In our recent research, we demonstrated that the interferon (IFN)-induced transmembrane protein 2 (IFITM2) promoted the growth and metastasis of GC. However, it remains unclear whether IFITM2 involves in TGF-ß1 mediated EMT in GC. METHODS AND RESULTS: In the present research, we investigated the functional role of IFITM2 in EMT process and TGF-ß1 signaling pathway in two GC cell lines. We noticed that silencing IFITM2 can effectively inhibit TGF-ß1 signaling mediated EMT by regulating down stream small mother against decapentaplegic (SMAD) 2/3 and transcription factors.This finding was further determined in both tumor tissues from GC patients and normal tissues adjacent to cancer. Our data demonstrated the key role of IFITM2 in TGF-ß1 signaling and EMT in GC. CONCLUSION: The findings enriched our understanding of the underlying mechanism in EMT during the progression of GC. In addition, IFITM2 would be a potential target for treating GC and other malignant tumors.

Overexpression of GmMYB14 improves high-density yield and drought tolerance of soybean through regulating plant architecture mediated by the brassinosteroid pathway.

MYB transcription factors (TFs) have been reported to regulate the biosynthesis of secondary metabolites, as well as to mediate plant adaption to abiotic stresses, including drought. However, the roles of MYB TFs in regulating plant architecture and yield potential remain poorly understood. Here, we studied the roles of the dehydration-inducible GmMYB14 gene in regulating plant architecture, high-density yield and drought tolerance through the brassinosteroid (BR) pathway in soybean. GmMYB14 was shown to localize to nucleus and has a transactivation activity. Stable GmMYB14-overexpressing (GmMYB14-OX) transgenic soybean plants displayed a semi-dwarfism and compact plant architecture associated with decreased cell size, resulting in a decrease in plant height, internode length, leaf area, leaf petiole length and leaf petiole angle, and improved yield in high density under field conditions. Results of the transcriptome sequencing suggested the involvement of BRs in regulating GmMYB14-OX plant architecture. Indeed, GmMYB14-OX plants showed reduced endogenous BR contents, while exogenous application of brassinolide could partly rescue the phenotype of GmMYB14-OX plants. Furthermore, GmMYB14 was shown to directly bind to the promoter of GmBEN1 and up-regulate its expression, leading to reduced BR content in GmMYB14-OX plants. GmMYB14-OX plants also displayed improved drought tolerance under field conditions. GmBEN1 expression was also up-regulated in the leaves of GmMYB14-OX plants under polyethylene glycol treatment, indicating that the GmBEN1-mediated reduction in BR level under stress also contributed to drought/osmotic stress tolerance of the transgenic plants. Our findings provided a strategy for stably increasing high-density yield and drought tolerance in soybean using a single TF-encoding gene.

Experimental Observation of Phase Transitions in Spatial Photonic Ising Machine.

Statistical spin dynamics plays a key role in understanding the working principle for novel optical Ising machines. Here, we propose the gauge transformation for a spatial photonic Ising machine, where a single spatial phase modulator simultaneously encodes spin configurations and programs interaction strengths. Using gauge transformation, we experimentally evaluate the phase diagram of a high-dimensional spin-glass equilibrium system with 100 fully connected spins. We observe the presence of paramagnetic, ferromagnetic, and spin-glass phases and determine the critical temperature T_{c} and the critical probability p_{c} of the phase transitions, which agree well with the mean-field theory predictions. Thus, the approximation of the mean-field model is experimentally verified in the spatial photonic Ising machine. Furthermore, we discuss the phase transition in parallel with solving combinatorial optimization problems during the cooling process and identify that the spatial photonic Ising machine is robust with sufficient many-spin interactions even when the system is associated with optical aberrations and measurement uncertainty.

Genome-Wide Identification and Characterization of Soybean GmLOR Gene Family and Expression Analysis in Response to Abiotic Stresses.

The LOR (LURP-one related) family genes encode proteins containing a conserved LOR domain. Several members of the LOR family genes are required for defense against Hyaloperonospora parasitica (Hpa) in Arabidopsis. However, there are few reports of LOR genes in response to abiotic stresses in plants. In this study, a genome-wide survey and expression levels in response to abiotic stresses of 36 LOR genes from Glycine max were conducted. The results indicated that the GmLOR gene family was divided into eight subgroups, distributed on 14 chromosomes. A majority of members contained three extremely conservative motifs. There were four pairs of tandem duplicated GmLORs and nineteen pairs of segmental duplicated genes identified, which led to the expansion of the number of GmLOR genes. The expansion patterns of the GmLOR family were mainly segmental duplication. A heatmap of soybean LOR family genes showed that 36 GmLOR genes exhibited various expression patterns in different tissues. The cis-acting elements in promoter regions of GmLORs include abiotic stress-responsive elements, such as dehydration-responsive elements and drought-inducible elements. Real-time quantitative PCR was used to detect the expression level of GmLOR genes, and most of them were expressed in the leaf or root except that GmLOR6 was induced by osmotic and salt stresses. Moreover, GmLOR4/10/14/19 were significantly upregulated after PEG and salt treatments, indicating important roles in the improvement of plant tolerance to abiotic stress. Overall, our study provides a foundation for future investigations of GmLOR gene functions in soybean.

Identification of soybean drought-tolerant genotypes and loci correlated with agronomic traits contributes new candidate genes for breeding.

KEY MESSAGE: Drought tolerance level of 136 soybean genotypes, the correlations among traits were evaluated, and several important drought-tolerant genotypes, traits, SNPs and genes were possibly useful for soybean genetic breeding. Drought is an adverse environmental factor affecting crops growth, development, and yield. Promising genotypes and genes with improved tolerance to drought are probably effective ways to alleviate the situation. In this study, our main task was to determine drought tolerance level of 136 soybean genotypes, the correlations among physiological and agronomic traits under drought, and drought-tolerant single nucleotide polymorphism (SNPs) and genes. In this study, twenty-six varieties were identified as excellent tolerant genotypes to stress among which S14, S93 and S135 with high drought-tolerant index (DTI > 1.3) and yield (Y > 300 kg). Fourteen varieties were identified as drought-sensitive genotypes, such as S25, S45 and S58, with low drought-tolerant index (DTI < 0.5). 422 SNPs and 302 genes correlated with seed number per plant (SNPP), maturity (M), number of seeds per pod (NSPP), node number of main stem (NNMS), Stem diameter (SD) and pull stem (PS) were detected under well-watered and drought conditions by genome-wide association study (GWAS). Among them, we found SNPs (Chr 3:1758920-1958934) between drought-tolerant and sensitive genotypes. 13 genes (Glyma.03G017800, Glyma.03G018000, Glyma.03G018200, Glyma.03G018400, Glyma.03G018500, Glyma.03G018600, Glyma.03G018700, Glyma.03G018800, Glyma.03G018900, Glyma.03G019000, Glyma.03G019100, Glyma.03G019200, Glyma.03G019300) correlated with NNMS were detected. By qRT-PCR, the expression level of Glyma.03G018000 and Glyma.03G018900 in drought-tolerant varieties was significantly increased, but low or no expression in sensitive varieties under drought stress. This study provides important drought-tolerant genotypes, traits, SNPs and potential genes, possibly useful for soybean genetic breeding.

The clinical significance of metastatic breast carcinoma to intramammary lymph node.

The incidence of involved intramammary lymph node (intra-MLN) with breast carcinoma (BC) is rare. Its clinical significance and impact on the clinical decision making is unclear. A total of 113 BC cases with at least one positive intra-MLN were collected from 11 academic institutions. The inclusion criteria were subsequent axillary lymph node dissection, and the availability of information on T-stage, size of node metastasis, extranodal extension status, biomarkers status, and clinical follow-up. Stage 4 cases and/or neo-adjuvant treated patients were excluded. AJCC TN-stage was calculated twice, with and without intra-MLN. Five-year overall survival (OS) and relapse (local and/or distant)-free survival (RFS) were calculated and correlated with the clinicopathologic variables. Excluding intra-MLN, TN-stage correlated with OS (P = .016) but not with RFS (P = .19). However, when intra-MLN was included, TN-stage correlated with both OS (P < .001) and RFS (P = .016). In the multivariate analysis, when intra-MLN was excluded, only radiation therapy (RT) correlated with RFS (HR = 0.19, 95% CI: 0.054-0.66, P = .009). However, when intra-MLN was included in the TN-stage both RT (HR = 0.13, 95% CI: 0.04-0.45, P = .001) and TN-stage 3 (HR = 8.92, 95% CI: 1.47-54, P = .017) correlated with RFS. Tumor multifocality was the only variable correlated with OS when the intra-MLN involvement was excluded. When intra-MLN was included, multifocality became insignificant but TN-stage 3 correlated with OS (HR = 8.59, 95% CI: 1.06-69.71, P = .044). Positive intra-MLN is an independent factor in predicting both RFS and OS.

Diagnostic evaluation and treatment of primary breast osteosarcoma: A case report.

Primary breast osteosarcoma (PBO) is a rare tumor of the breast, with only case reports and small case series published in the literature. Classic imaging findings of sunburst calcifications, a rim of intermediate density, and intense uptake on bone scintigraphy can help make the correct diagnosis preoperatively, allowing for appropriate surgical and chemotherapeutic management. We present the imaging evaluation, treatment course, and follow-up of a case of PBO diagnosed in a 67-year-old patient.

Optical spatial differentiator for a synthetic three-dimensional optical field.

We propose a grating-based spatial differentiator to process a synthetic three-dimensional optical field, where several conventional two-dimensional optical images are stacked at multiple wavelengths. The device simultaneously enables both spatial differentiation and demultiplexing during light diffraction. We show the spatial differentiation resulting from coupling and interference of spatial modes and derive the theoretical condition for spatial differentiation based on spatial coupled-mode theory. We numerically investigate field transformation during diffraction and demonstrate spatial differentiation with image processing of edge detection for a synthetic three-dimensional optical field, where four images are stored at different wavelengths.

The role of skin ulceration in breast carcinoma staging and outcome.

Breast carcinoma with skin ulceration (SU) is considered a locally advanced disease. The purpose of the study is to investigate if SU is an independent adverse factor. Breast carcinoma patients with SU (n=111) were included in the study. A subset (n=38, study cohort) was matched with cases that had no SU (n=38, matched cohort); the survival analyses were compared between these groups. Then, cases (n=80) were staged independent from SU into stage I, II or III. Disease free survival (DFS) and overall survival (OS) were analyzed. Patients with larger tumors tended to present with distant metastases more often than patients with smaller tumors (P=.004). In the matched cases, the 5-year DFS probability was 53% for the study cohort and 58% for the matched cohort; and for OS 75% for the study cohort and 84% for the matched cohort with no statistical significant difference. However, there was a trend towards worse DFS for the patients whose tumors had SU. When the cases were staged based on tumor size and node status (I, II or III), the OS was statistically significant (P=.047) but not the DFS (P=.195). Relatively small tumors with SU had an extent of disease similar to that observed in patients with early stages disease. The survival analysis suggests that SU may not be an adverse factor. However, more cases are needed to further examine this finding.

On-grating graphene surface plasmons enabling spatial differentiation in the terahertz region.

We propose a graphene-on-grating nanostructure to enable second-order spatial differentiation computation in the terahertz (THz) region. The differentiation operation is based on the interference between the direct reflected field and the leakage of two excited surface plasmon polaritons counter-propagating along the graphene sheet. With the spatial coupled-mode theory, we derive that the requirement for the second-order spatial differentiation is the critical coupling condition. We numerically demonstrate such an analog computation with Gaussian beams. It shows that the spatial bandwidth of the proposed differentiator is large enough such that even when the waist radius of the Gaussian beam is as narrow as w0=0.68λ (λ is the free-space wavelength), the accuracy of the differentiator is higher than 95%. The proposed differentiator is ultra-compact, with a thickness less than 0.1λ, and useful for real-time imaging applications in THz security detections.

Induction chemotherapy followed by concurrent chemoradiotherapy for advanced stage oropharyngeal squamous cell carcinoma with HPV and P16 testing.

OBJECTIVE: The objective was to report our experience with advanced stage oropharyngeal squamous cell carcinoma treated sequentially with induction chemotherapy followed by concurrent chemoradiotherapy. METHODS: Retrospective chart review identified 49 eligible patients with advanced stage oropharyngeal squamous cell carcinoma treated with induction chemotherapy followed by concurrent chemoradiotherapy. HPV and p16(INK4A) testing was performed on pathology specimens. Follow-up of over 11 months was required unless a death or treatment failure occurred before that time. RESULTS: Treatment with induction chemotherapy followed by concurrent chemoradiotherapy resulted in 44/48 (90%) complete durable response. One death occurred from pulmonary embolism. Toxicity profiles were comparable to other published data. Average follow-up was 3.9 years. Oncologic failure rates among subgroups showed 5.7% failure for HPV+/p16+ cancer, 9.1% failure for HPV-/p16+ cancer, 100% failure for HPV-/p16- cancer, 0% failure for nonsmokers, and 17.9% failure for smokers. CONCLUSIONS: This study showed favorable outcomes in terms of durable oncologic response and acceptable toxicity profiles. It is notable that 36/49 patients were HPV+/p16+ and 11/49 were HPV-/p16+. Only 2 patients were HPV-/p16-, and both died as a result of oncologic failures. This highlights the importance of obtaining HPV and p16 testing in studies evaluating the efficacy of treatments for oropharyngeal squamous cell carcinoma.

Detailed Image Data Quality and Cleaning Practices for Artificial Intelligence Tools for Breast Cancer.

Standardizing image-data preparation practices to improve accuracy/consistency of AI diagnostic tools.

Combined inhibition of CDK4/6 and AKT is highly effective against the luminal androgen receptor (LAR) subtype of triple negative breast cancer.

Luminal Androgen Receptor (LAR) triple-negative breast cancers (TNBC) express androgen receptors (AR), exhibit high frequency of PIK3CA mutations and intact RB. Herein, we investigated combined blockade of the CDK4/6 and PI3K signaling with palbociclib, alpelisib, and capivasertib, which inhibit CDK4/6, PI3Kα, and AKT1-3, respectively. The combination of palbociclib/capivasertib, but not palbociclib/alpelisib, synergistically inhibited proliferation of MDA-MB-453 and MFM-223 LAR cells [synergy score 7.34 (p = 5.81x10-11) and 4.78 (p = 0.012), respectively]. The AR antagonist enzalutamide was inactive against MDA-MB-453, MFM-223, and CAL148 cells and did not enhance the efficacy of either combination. Palbociclib/capivasertib inhibited growth of LAR patient-derived xenografts more potently than palbociclib/alpelisib. Treatment of LAR cells with palbociclib suppressed phosphorylated-RB and resulted in adaptive phosphorylation/activation of S473 pAKT and AKT substrates GSK3ß, PRAS40, and FoxO3a. Capivasertib blocked palbociclib-induced phosphorylation of AKT substrates more potently than alpelisib. Treatment with PI3Kß inhibitors did not block phosphorylation of AKT substrates, suggesting that PI3Kß did not mediate the adaptive response to CDK4/6 inhibition. Phosphokinase arrays of MDA-MB-453 cells treated with palbociclib showed time-dependent upregulation of PDGFRß, GSK3ß, STAT3, and STAT6. RNA silencing of PDGFRß in palbociclib-treated MDA-MB-453 and MFM-223 cells blocked the upregulation of S473 pAKT, suggesting that the adaptive response to CDK4/6 blockade involves PDGFRß signaling. Finally, treatment with palbociclib and the PDGFR inhibitor CP637451 arrested growth of MDA-MB-453 and MFM-223 cells to the same degree as palbociclib/capivasertib. These findings support testing the combination of CDK4/6 and AKT inhibitors in patients with LAR TNBC, and further investigation of PDGFR antagonists in this breast cancer subtype.

PRMT5 is an actionable therapeutic target in CDK4/6 inhibitor-resistant ER+/RB-deficient breast cancer.

CDK4/6 inhibitors (CDK4/6i) have improved survival of patients with estrogen receptor-positive (ER+) breast cancer. However, patients treated with CDK4/6i eventually develop drug resistance and progress. RB1 loss-of-function alterations confer resistance to CDK4/6i, but the optimal therapy for these patients is unclear. Through a genome-wide CRISPR screen, we identify protein arginine methyltransferase 5 (PRMT5) as a molecular vulnerability in ER+/RB1-knockout breast cancer cells. Inhibition of PRMT5 blocks the G1-to-S transition in the cell cycle independent of RB, leading to growth arrest in RB1-knockout cells. Proteomics analysis uncovers fused in sarcoma (FUS) as a downstream effector of PRMT5. Inhibition of PRMT5 results in dissociation of FUS from RNA polymerase II, leading to hyperphosphorylation of serine 2 in RNA polymerase II, intron retention, and subsequent downregulation of proteins involved in DNA synthesis. Furthermore, treatment with the PRMT5 inhibitor pemrametostat and a selective ER degrader fulvestrant synergistically inhibits growth of ER+/RB-deficient cell-derived and patient-derived xenografts. These findings highlight dual ER and PRMT5 blockade as a potential therapeutic strategy to overcome resistance to CDK4/6i in ER+/RB-deficient breast cancer.

EML4-ALK fusions drive lung adeno-to-squamous transition through JAK-STAT activation.

Human lung adenosquamous cell carcinoma (LUAS), containing both adenomatous and squamous pathologies, exhibits strong cancer plasticity. We find that ALK rearrangement is detectable in 5.1-7.5% of human LUAS, and transgenic expression of EML4-ALK drives lung adenocarcinoma (LUAD) formation initially and squamous transition at late stage. We identify club cells as the main cell-of-origin for squamous transition. Through recapitulating lineage transition in organoid system, we identify JAK-STAT signaling, activated by EML4-ALK phase separation, significantly promotes squamous transition. Integrative study with scRNA-seq and immunostaining identify a plastic cell subpopulation in ALK-rearranged human LUAD showing squamous biomarker expression. Moreover, those relapsed ALK-rearranged LUAD show notable upregulation of squamous biomarkers. Consistently, mouse squamous tumors or LUAD with squamous signature display certain resistance to ALK inhibitor, which can be overcome by combined JAK1/2 inhibitor treatment. This study uncovers strong plasticity of ALK-rearranged tumors in orchestrating phenotypic transition and drug resistance and proposes a potentially effective therapeutic strategy.

Intraoperative evaluation of sentinel lymph nodes in patients with breast cancer treated with systemic neoadjuvant therapy.

AIMS: Touch preparation (TP) and frozen section (FS) are the two methods routinely used in the intraoperative evaluation (IOE) of sentinel lymph nodes (SLNs) to detect metastases in patients with breast cancer. Both methods are extremely sensitive and specific in the primary surgery (non-neoadjuvant systemic therapy (non-NST)) setting. Since NST introduces unique challenges in the IOE of SLNs, the aim was to determine the accuracy of TP and FS in the IOE of SLNs in the NST setting and compare the results with the non-NST setting and to examine factors that contribute to any differences. METHODS: We analysed 871 SLNs from 232 patients (615 SLNs from NST and 256 SLNs from non-NST settings) between 2016 through 2019. RESULTS: In the NST group, TP alone (n=366) had a sensitivity of 45.7% and specificity of 99.7%; FS alone (n=90) had a sensitivity of 83.3% and specificity of 100%. When both TP and FS (n=135) were used, the sensitivity was 80.3% and the specificity was 98.6%.In the non-NST group, TP alone (n=193) had a sensitivity of 66.7% and specificity of 100%; FS alone (n=22) had a sensitivity and specificity of 100%; and combined TP and FS (n=34) had a sensitivity and specificity of 100% and 96%, respectively. CONCLUSIONS: Evaluating SLNs intraoperatively in the NST setting can be challenging secondary to therapy-related changes. In the NST setting, FS has higher sensitivity and specificity compared with TP for the IOE of SLNs and should be the preferred method.

Pleiotrophin drives a prometastatic immune niche in breast cancer.

Metastatic cancer cells adapt to thrive in secondary organs. To investigate metastatic adaptation, we performed transcriptomic analysis of metastatic and non-metastatic murine breast cancer cells. We found that pleiotrophin (PTN), a neurotrophic cytokine, is a metastasis-associated factor that is expressed highly by aggressive breast cancers. Moreover, elevated PTN in plasma correlated significantly with metastasis and reduced survival of breast cancer patients. Mechanistically, we find that PTN activates NF-κB in cancer cells leading to altered cytokine production, subsequent neutrophil recruitment, and an immune suppressive microenvironment. Consequently, inhibition of PTN, pharmacologically or genetically, reduces the accumulation of tumor-associated neutrophils and reverts local immune suppression, resulting in increased T cell activation and attenuated metastasis. Furthermore, inhibition of PTN significantly enhanced the efficacy of immune checkpoint blockade and chemotherapy in reducing metastatic burden in mice. These findings establish PTN as a previously unrecognized driver of a prometastatic immune niche and thus represents a promising therapeutic target for the treatment of metastatic breast cancer.