Prediction of biomarker-disease associations based on graph attention network and text representation.

MOTIVATION: The associations between biomarkers and human diseases play a key role in understanding complex pathology and developing targeted therapies. Wet lab experiments for biomarker discovery are costly, laborious and time-consuming. Computational prediction methods can be used to greatly expedite the identification of candidate biomarkers. RESULTS: Here, we present a novel computational model named GTGenie for predicting the biomarker-disease associations based on graph and text features. In GTGenie, a graph attention network is utilized to characterize diverse similarities of biomarkers and diseases from heterogeneous information resources. Meanwhile, a pretrained BERT-based model is applied to learn the text-based representation of biomarker-disease relation from biomedical literature. The captured graph and text features are then integrated in a bimodal fusion network to model the hybrid entity representation. Finally, inductive matrix completion is adopted to infer the missing entries for reconstructing relation matrix, with which the unknown biomarker-disease associations are predicted. Experimental results on HMDD, HMDAD and LncRNADisease data sets showed that GTGenie can obtain competitive prediction performance with other state-of-the-art methods. AVAILABILITY: The source code of GTGenie and the test data are available at: https://github.com/Wolverinerine/GTGenie.

Alignment-free filtering for cfNA fusion fragments.

MOTIVATION: Cell-free nucleic acid (cfNA) sequencing data require improvements to existing fusion detection methods along multiple axes: high depth of sequencing, low allele fractions, short fragment lengths and specialized barcodes, such as unique molecular identifiers. RESULTS: AF4 was developed to address these challenges. It uses a novel alignment-free kmer-based method to detect candidate fusion fragments with high sensitivity and orders of magnitude faster than existing tools. Candidate fragments are then filtered using a max-cover criterion that significantly reduces spurious matches while retaining authentic fusion fragments. This efficient first stage reduces the data sufficiently that commonly used criteria can process the remaining information, or sophisticated filtering policies that may not scale to the raw reads can be used. AF4 provides both targeted and de novo fusion detection modes. We demonstrate both modes in benchmark simulated and real RNA-seq data as well as clinical and cell-line cfNA data. AVAILABILITY AND IMPLEMENTATION: AF4 is open sourced, licensed under Apache License 2.0, and is available at: https://github.com/grailbio/bio/tree/master/fusion.

Numerous uncharacterized and highly divergent microbes which colonize humans are revealed by circulating cell-free DNA.

Blood circulates throughout the human body and contains molecules drawn from virtually every tissue, including the microbes and viruses which colonize the body. Through massive shotgun sequencing of circulating cell-free DNA from the blood, we identified hundreds of new bacteria and viruses which represent previously unidentified members of the human microbiome. Analyzing cumulative sequence data from 1,351 blood samples collected from 188 patients enabled us to assemble 7,190 contiguous regions (contigs) larger than 1 kbp, of which 3,761 are novel with little or no sequence homology in any existing databases. The vast majority of these novel contigs possess coding sequences, and we have validated their existence both by finding their presence in independent experiments and by performing direct PCR amplification. When their nearest neighbors are located in the tree of life, many of the organisms represent entirely novel taxa, showing that microbial diversity within the human body is substantially broader than previously appreciated.

Simultaneously Monitoring Immune Response and Microbial Infections during Pregnancy through Plasma cfRNA Sequencing.

BACKGROUND: Plasma cell-free RNA (cfRNA) encompasses a broad spectrum of RNA species that can be derived from both human cells and microbes. Because cfRNA is fragmented and of low concentration, it has been challenging to profile its transcriptome using standard RNA-seq methods. METHODS: We assessed several recently developed RNA-seq methods on cfRNA samples. We then analyzed the dynamic changes of both the human transcriptome and the microbiome of plasma during pregnancy from 60 women. RESULTS: cfRNA reflects a well-orchestrated immune modulation during pregnancy: an up-regulation of antiinflammatory genes and an increased abundance of antimicrobial genes. We observed that the plasma microbiome remained relatively stable during pregnancy. The bacteria Ureaplasma shows an increased prevalence and increased abundance at postpartum, which is likely to be associated with postpartum infection. We demonstrated that cfRNA-seq can be used to monitor viral infections. We detected a number of human pathogens in our patients, including an undiagnosed patient with a high load of human parvovirus B19 virus (B19V), which is known to be a potential cause of complications in pregnancy. CONCLUSIONS: Plasma cfRNA-seq demonstrates the potential to simultaneously monitor immune response and microbial infections during pregnancy.

A pilot study on the use of cerebrospinal fluid cell-free DNA in intramedullary spinal ependymoma.

Cerebrospinal fluid (CSF) represents a promising source of cell-free DNA (cfDNA) for tumors of the central nervous system. A CSF-based liquid biopsy may obviate the need for riskier tissue biopsies and serve as a means for monitoring tumor recurrence or response to therapy. Spinal ependymomas most commonly occur in adults, and aggressive resection must be delicately balanced with the risk of injury to adjacent normal tissue. In patients with subtotal resection, recurrence commonly occurs. A CSF-based liquid biopsy matched to the patient's spinal ependymoma mutation profile has potential to be more sensitive then surveillance MRI, but the utility has not been well characterized for tumors of the spinal cord. In this study, we collected matched blood, tumor, and CSF samples from three adult patients with WHO grade II intramedullary spinal ependymoma. We performed whole exome sequencing on matched tumor and normal DNA to design Droplet Digital™ PCR (ddPCR) probes for tumor and wild-type mutations. We then interrogated CSF samples for tumor-derived cfDNA by performing ddPCR on extracted cfDNA. Tumor cfDNA was not reliably detected in the CSF of our cohort. Anatomic sequestration and low grade of intramedullary spinal cord tumors likely limits the role of CSF liquid biopsy.

Nigericin Inhibits Epithelial Ovarian Cancer Metastasis by Suppressing the Cell Cycle and Epithelial-Mesenchymal Transition.

Epithelial ovarian cancer (EOC) has the highest mortality among various types of gynecological malignancies. Most patients die of metastasis and recurrence due to cisplatin resistance. Thus, it is urgent to develop novel therapies to cure this disease. CCK-8 assay showed that nigericin exhibited strong cytotoxicity on A2780 and SKOV3 cell lines. Flow cytometry indicated that nigericin could induce cell cycle arrest at G0/G1 phase and promote cell apoptosis. Boyden chamber assay revealed that nigericin could inhibit migration and invasion in a dose-dependent manner by suppressing epithelial-mesenchymal transition (EMT) in EOC cells. These effects were mediated, at least partly, by the Wnt/ß-catenin signaling pathway. Our results demonstrated that nigericin could inhibit EMT during cell invasion and metastasis through the canonical Wnt/ß-catenin signaling pathway. Nigericin may prove to be a novel therapeutic strategy that is effective in patients with metastatic EOC.

Noninvasive in vivo monitoring of tissue-specific global gene expression in humans.

Circulating cell-free RNA in the blood provides a potential window into the health, phenotype, and developmental programs of a variety of human organs. We used high-throughput methods of RNA analysis such as microarrays and next-generation sequencing to characterize the global landscape circulating RNA in a cohort of human subjects. By focusing on genes whose expression is highly specific to certain tissues, we were able to identify the relative contributions of these tissues to circulating RNA and to monitor changes in tissue development and health. As one application of this approach, we performed a longitudinal study on pregnant women and analyzed their combined cell-free RNA transcriptomes across all three trimesters of pregnancy and after delivery. In addition to the analysis of mRNA, we observed and characterized noncoding species such as long noncoding RNA and circular RNA transcripts whose presence had not been previously observed in human plasma. We demonstrate that it is possible to track specific longitudinal phenotypic changes in both the mother and the fetus and that it is possible to directly measure transcripts from a variety of fetal tissues in the maternal blood sample. We also studied the role of neuron-specific transcripts in the blood of healthy adults and those suffering from the neurodegenerative disorder Alzheimer's disease and showed that disease specific neural transcripts are present at increased levels in the blood of affected individuals. Characterization of the cell-free transcriptome in its entirety may thus provide broad insights into human health and development without the need for invasive tissue sampling.

Tumor DNA in cerebral spinal fluid reflects clinical course in a patient with melanoma leptomeningeal brain metastases.

Cerebral spinal fluid (CSF) from brain tumor patients contains tumor cellular and cell-free DNA (cfDNA), which provides a less-invasive and routinely accessible method to obtain tumor genomic information. In this report, we used droplet digital PCR to test mutant tumor DNA in CSF of a patient to monitor the treatment response of metastatic melanoma leptomeningeal disease (LMD). The primary melanoma was known to have a BRAF (V600E) mutation, and the patient was treated with whole brain radiotherapy and BRAF inhibitors. We collected 9 CSF samples over 6 months. The mutant cfDNA fraction gradually decreased from 53 % (time of diagnosis) to 0 (time of symptom alleviation) over the first 6 time points. Three months after clinical improvement, the patient returned with severe symptoms and the mutant cfDNA was again detected in CSF at high levels. The mutant DNA fraction corresponded well with the patient's clinical response. We used whole exome sequencing to examine the mutation profiles of the LMD tumor DNA in CSF before therapeutic response and after disease relapse, and discovered a canonical cancer mutation PTEN (R130*) at both time points. The cellular and cfDNA revealed similar mutation profiles, suggesting cfDNA is representative of LMD cells. This study demonstrates the potential of using cellular or cfDNA in CSF to monitor treatment response for LMD.

Brain tumor mutations detected in cerebral spinal fluid.

BACKGROUND: Detecting tumor-derived cell-free DNA (cfDNA) in the blood of brain tumor patients is challenging, presumably owing to the blood-brain barrier. Cerebral spinal fluid (CSF) may serve as an alternative "liquid biopsy" of brain tumors by enabling measurement of circulating DNA within CSF to characterize tumor-specific mutations. Many aspects about the characteristics and detectability of tumor mutations in CSF remain undetermined. METHODS: We used digital PCR and targeted amplicon sequencing to quantify tumor mutations in the cfDNA of CSF and plasma collected from 7 patients with solid brain tumors. Also, we applied cancer panel sequencing to globally characterize the somatic mutation profile from the CSF of 1 patient with suspected leptomeningeal disease. RESULTS: We detected tumor mutations in CSF samples from 6 of 7 patients with solid brain tumors. The concentration of the tumor mutant alleles varied widely between patients, from <5 to nearly 3000 copies/mL CSF. We identified 7 somatic mutations from the CSF of a patient with leptomeningeal disease by use of cancer panel sequencing, and the result was concordant with genetic testing on the primary tumor biopsy. CONCLUSIONS: Tumor mutations were detectable in cfDNA from the CSF of patients with different primary and metastatic brain tumors. We designed 2 strategies to characterize tumor mutations in CSF for potential clinical diagnosis: the targeted detection of known driver mutations to monitor brain metastasis and the global characterization of genomic aberrations to direct personalized cancer care.

Clinical Analysis of 32 Cases of Subglottic Benign Airway Stenosis Treated With Montgomery T Silicone Stent.

Objective: To explore the complications of long-term placement of Montgomery T silicone stent (T-tube) in the treatment of subglottic benign airway stenosis (SBAS) and the timing of successful T-tube removal. Methods: We retrospectively collected the clinical data of 32 patients with SBAS who underwent the treatment of T-tube and analyzed their placement and successful removal of the T-tube. Results: There were 22 males and 10 females, aged from 21 to 79 years (60.9 ± 13.7 years). The T-tubes were successfully placed in all 32 patients, and 6 patients (18.8%) with mild stenosis were placed by the intravenous conscious sedation. The longest follow-up period was 60.4 months, and 17 patients (53.1%) had the T-tubes for more than 12 months; 5 patients (15.6%) were changed to the tracheostomy cannula after unplanned removal of the T-tubes for various reasons; the T-tubes were successfully removed in 9 patients (28.1%), and the duration of T-tubes placement was 5.2-22.7 months (12.1 ± 6.3 months), among them anatomical stenosis in 9 patients (100%). Secretion retention was observed in 32 patients (100%), granulation tissue hyperplasia was observed in 9 patients (28.1%), and the normal ventilation was not affected in most patients by bronchoscopic treatment and follow-up; the T-tubes were removed in 3 patients due to severe complications. There was no significant difference in the incidences of secretion retention and granulation tissue hyperplasia between the time point at 1 week, 1 month, 3 months, and 12 months, p > 0.05. In patients with T-tube more than 12 months, the severity of secretion retention at 1 week, 1 month, 3 months, and 12 months was significantly different, p < 0.05, however, there was no significant difference in the severity of granulation tissue hyperplasia, p > 0.05. Conclusions: T-tube is safe and effective in the treatment of SBAS. The severity of secretion retention increased in patients with long-term placement of the T-tube. For patients with mild stenosis and anatomical stenosis, the T-tube removal can be attempted at about 1 year of follow-up.

High-performance binary protein interaction screening in a microfluidic format.

The standard procedure to increase microfluidic chip performance is to grow the number of parallel test systems on the chip. This process is accompanied by miniaturizing biochemical workflows and micromechanical elements, which is often a major challenge for both engineering fields. In this work, we show that it is possible to substantially increase the runtime performance of a microfluidic affinity assay for protein interactions by simultaneously engineering fluid logics and assay chemistry. For this, synergistic effects between the micro- and chemical architecture of the chip are exploited. The presented strategy of reducing the runtime rather than size and volume of the mechanical elements and biological reagent compartments will, in general, be of importance for future analytical test systems on microfluidic chips to overcome performance barriers.

MarkerGenie: an NLP-enabled text-mining system for biomedical entity relation extraction.

Motivation: Natural language processing (NLP) tasks aim to convert unstructured text data (e.g. articles or dialogues) to structured information. In recent years, we have witnessed fundamental advances of NLP technique, which has been widely used in many applications such as financial text mining, news recommendation and machine translation. However, its application in the biomedical space remains challenging due to a lack of labeled data, ambiguities and inconsistencies of biological terminology. In biomedical marker discovery studies, tools that rely on NLP models to automatically and accurately extract relations of biomedical entities are valuable as they can provide a more thorough survey of all available literature, hence providing a less biased result compared to manual curation. In addition, the fast speed of machine reader helps quickly orient research and development. Results: To address the aforementioned needs, we developed automatic training data labeling, rule-based biological terminology cleaning and a more accurate NLP model for binary associative and multi-relation prediction into the MarkerGenie program. We demonstrated the effectiveness of the proposed methods in identifying relations between biomedical entities on various benchmark datasets and case studies. Availability and implementation: MarkerGenie is available at https://www.genegeniedx.com/markergenie/. Data for model training and evaluation, term lists of biomedical entities, details of the case studies and all trained models are provided at https://drive.google.com/drive/folders/14RypiIfIr3W_K-mNIAx9BNtObHSZoAyn?usp=sharing. Supplementary information: Supplementary data are available at Bioinformatics Advances online.

Elevated air quality index and fine particulate matter levels contribute to the poor prognosis and progression of nonsmall-cell lung cancer: A cohort study combined with external validation.

BACKGROUND: Poor air quality can result in a variety of respiratory disorders. However, the air quality index (AQI) and the level of fine particulate matter (PM2.5) on the progression and prognosis of nonsmall-cell lung cancer (NSCLC) are unclear. METHODS: We launched a cohort study focused on the relationship between air quality and overall survival as well as progression, incorporating data from 590 patients with NSCLC in our medical center between November 1, 2013 and March 1, 2016. Forty-nine patients from Sichuan Cancer Hospital were used for validation. RESULTS: Cases with poorer AQI 6 months before NSCLC diagnosis were more likely to progress to stage III to IV NSCLC than controls (OR = 2.61, 95% CI 1.35-5.24, p = 0.005). Similarly, if exposed to high levels of PM2.5 during these 6 months, overall survival was poor (HR [95% CI] = 1.53 [1.13, 2.07], p = 0.006). According to multivariate analysis, age, gender, KPS, PM2.5, hyperlipemia, and NSCLC stage were independent risk factors of overall survival. A predictive model developed by these factors above yielded a favorable agreement (C-index = 0.758) on the calibration curve. External validation was conducted by 46 patients from Sichuan Cancer Hospital displaying an AUC of 0.724 (0.684-0.763). CONCLUSIONS: PM2.5 and AQI levels affect disease progression and long-term survival in NSCLC patients. An overall survival prediction model based on the PM2.5 level can help clinicians predict the risk of death in NSCLC.

Increased diastolic blood pressure and apnea time contribute to the poor apnea and hypopnea index and life quality of primary snoring: a cohort study combined with external validation.

Obstructive sleep apnea hypoventilation syndrome (OSAHS) is a common sleep breathing disorder closely associated with cardiovascular disease. However, the respiratory sleep and related cardiovascular parameters on the apnea and hypopnea index (AHI) and life quality of primary snoring are unclear. We launched a cohort study focused on the association between respiratory sleep and cardiovascular-related parameters and apnea and hypopnea index, incorporating data from 218 patients with primary snoring in our medical center between Jun 1, 2015, and Apr 1, 2016. Thirty patients from Sichuan Cancer Hospital were used for validation. Patients with longer apnea time were more likely to progress to higher AHI (> 30) than controls (OR = 5.66, 95% CI = [2.79, 11.97], p < 0.001). Similarly, if patients have a higher value of diastolic blood pressure, they will also have a higher AHI (> 30) (HR [95% CI] = 3.42 [1.14, 13.65], p = 0.043). According to multivariate analysis, longest apnea time, the mean percentage of SaO2, and neckline length were independent risk factors of overall survival. A predictive model developed based on these factors above yielded a favorable agreement (C-index = 0.872) on the calibration curve. Thirty patients conducted external validation from Sichuan Cancer Hospital, displaying an AUC of 0.833 (0.782-0.884). Increased diastolic blood pressure and apnea time affect AHI level. An AHI prediction model based on these factors above can help clinicians predict the risk of high AHI events.

WWOX promotes apoptosis and inhibits autophagy in paclitaxel­treated ovarian carcinoma cells.

Although paclitaxel (PTX) is a first­line chemotherapeutic agent for the treatment of epithelial ovarian cancer (EOC), its clinical use is restricted by chemoresistance. Autophagy is believed to promote drug resistance, and WW domain­containing oxidoreductase (WWOX) has been predicted to serve an essential role in apoptosis induction and to suppress autophagy in various tumor cell types. In the present study, the role of WWOX was demonstrated using PTX­treated EOC cells. Cell viability and apoptosis were detected using Cell Counting Kit­8, morphological and flow cytometric analyses. WWOX and phosphorylated (p)­WWOX were highly expressed in PTX­treated sensitive EOC cells (A2780), which was accompanied by activation of the apoptosis­related proteins caspase­3 and poly (ADP­ribose) polymerase (PARP). Conversely, PTX­resistant EOC cells (A2780/T) were characterized by reduced WWOX expression and constant phosphorylation levels, as well as undetectable levels of activated caspase­3 and PARP when cells were treated with PTX. The altered expression of WWOX between the two cell types was further validated by reverse transcription­-quantitative PCR. The apoptosis­inducing role of WWOX was also confirmed by flow cytometry after WWOX overexpression was induced in PTX­treated A2780 cells. These findings indicated that WWOX activation may inhibit chemoresistance and trigger cancer cell death. The upregulated expression levels of the autophagy­related protein 12­5 complex, Beclin­1 and LC3, as well as the downregulation of P62, were also detected following PTX treatment, suggesting that PTX induced autophagic flux in both types of EOC cells. This conclusion was further supported by visualizing the accumulation of autophagosome and autolysosome vesicles, using confocal microscopy and transmission electron microscopy. PTX was also shown to inhibit mTOR signaling, indicated by a decreased level of p­mTOR and increased expression of eukaryotic translation initiation factor 4E­binding protein 1. Finally, the interaction between WWOX, mTOR and autophagy was investigated via WWOX transfection experimentation, and indicated that WWOX activated mTOR whilst inhibiting autophagy. These data indicated that WWOX may serve a critical role in PTX­induced apoptosis and could suppress autophagy by downregulating essential autophagic effectors in EOC cells via mTOR signaling.

A comprehensive characterization of the cell-free transcriptome reveals tissue- and subtype-specific biomarkers for cancer detection.

Cell-free RNA (cfRNA) is a promising analyte for cancer detection. However, a comprehensive assessment of cfRNA in individuals with and without cancer has not been conducted. We perform the first transcriptome-wide characterization of cfRNA in cancer (stage III breast [n = 46], lung [n = 30]) and non-cancer (n = 89) participants from the Circulating Cell-free Genome Atlas (NCT02889978). Of 57,820 annotated genes, 39,564 (68%) are not detected in cfRNA from non-cancer individuals. Within these low-noise regions, we identify tissue- and cancer-specific genes, defined as "dark channel biomarker" (DCB) genes, that are recurrently detected in individuals with cancer. DCB levels in plasma correlate with tumor shedding rate and RNA expression in matched tissue, suggesting that DCBs with high expression in tumor tissue could enhance cancer detection in patients with low levels of circulating tumor DNA. Overall, cfRNA provides a unique opportunity to detect cancer, predict the tumor tissue of origin, and determine the cancer subtype.

Microfluidic Western blot.

We develop a novel method for Western blot based on microfluidics, incorporating the internal molecular weight marker, loading control, and antibody titration in the same protocol. Compared with the conventional method which could detect only one protein, the microfluidic Western blot could analyze at least 10 proteins simultaneously from a single sample, and it requires only about 1% of the amount of antibody used in conventional Western blot.

Diagnostic value of miR-186-5p for carotid artery stenosis and its predictive significance for future cerebral ischemic event.

BACKGROUND: MicroRNAs (miRNAs) serve as novel promising biomarkers for the diagnosis and prognosis of many human diseases. This study investigated the diagnostic value of miR-186-5p for asymptomatic carotid artery stenosis (CAS), and its predictive value for future cerebral ischemic events (CIEs). METHODS: Sixty-seven cases with asymptomatic CAS and 60 healthy individuals were recruited. Serum levels of miR-186-5p were tested by using qRT-PCR. Receiving-operator characteristic (ROC) curve was drawn based on sensitivity and specificity analyses. All asymptomatic CAS cases were followed up for 5 years. Kaplan-Meier method was applied for the evaluation of the predictive value of miR-186-5p for the occurrence of CIE. RESULTS: The serum level of miR-186-5p was increased significantly in asymptomatic CAS patients. MiR-186-5p was the most significant factor associated with the high degree of carotid stenosis in asymptomatic CAS patients. In the ROC curve analysis, the AUC was 0.919, with the sensitivity of 89.6% and specificity of 81.7% at the cutoff value of 1.221. Kaplan-Meier method results revealed that high miR-186-5p level was associated with the occurrence of CIEs. High miR-186-5p level and high degree of carotid stenosis were independent factors for the occurrence of CIEs. CONCLUSION: MiR-186-5p serves as a potential diagnostic biomarker for patients with asymptomatic CAS, and predicts the occurrence of future CIEs.

Single-cell transcriptomic atlas of the human endometrium during the menstrual cycle.

In a human menstrual cycle the endometrium undergoes remodeling, shedding and regeneration, all of which are driven by substantial gene expression changes in the underlying cellular hierarchy. Despite its importance in human fertility and regenerative biology, our understanding of this unique type of tissue homeostasis remains rudimentary. We characterized the transcriptomic transformation of human endometrium at single-cell resolution across the menstrual cycle, resolving cellular heterogeneity in multiple dimensions. We profiled the behavior of seven endometrial cell types, including a previously uncharacterized ciliated cell type, during four major phases of endometrial transformation, and found characteristic signatures for each cell type and phase. We discovered that the human window of implantation opens with an abrupt and discontinuous transcriptomic activation in the epithelia, accompanied with a widespread decidualization feature in the stromal fibroblasts. Our study provides a high-resolution molecular and cellular characterization of human endometrial transformation across the menstrual cycle, providing insights into this essential physiological process.

Recurrently Mutated Genes Differ between Leptomeningeal and Solid Lung Cancer Brain Metastases.

When compared with solid brain metastases from NSCLC, leptomeningeal disease (LMD) has unique growth patterns and is rapidly fatal. Patients with LMD do not undergo surgical resection, limiting the tissue available for scientific research. In this study we performed whole exome sequencing on eight samples of LMD to identify somatic mutations and compared the results with those for 26 solid brain metastases. We found that taste 2 receptor member 31 gene (TAS2R31) and phosphodiesterase 4D interacting protein gene (PDE4DIP) were recurrently mutated among LMD samples, suggesting involvement in LMD progression. Together with a retrospective review of the charts of an additional 44 patients with NSCLC LMD, we discovered a surprisingly low number of KRAS mutations (n = 4 [7.7%]) but a high number of EGFR mutations (n = 33 [63.5%]). The median interval for development of LMD from NSCLC was shorter in patients with mutant EGFR (16.3 months) than in patients with wild-type EGFR (23.9 months) (p = 0.017). Targeted analysis of recurrent mutations thus presents a useful complement to the existing diagnostic tool kit, and correlations of EGFR in LMD and KRAS in solid metastases suggest that molecular distinctions or systemic treatment pressure underpin the differences in growth patterns within the brain.