Estimating the total treatment effect in randomized experiments with unknown network structure.

Randomized experiments are widely used to estimate the causal effects of a proposed treatment in many areas of science, from medicine and healthcare to the physical and biological sciences, from the social sciences to engineering, and from public policy to the technology industry. Here we consider situations where classical methods for estimating the total treatment effect on a target population are considerably biased due to confounding network effects, i.e., the fact that the treatment of an individual may impact its neighbors' outcomes, an issue referred to as network interference or as nonindividualized treatment response. A key challenge in these situations is that the network is often unknown and difficult or costly to measure. We assume a potential outcomes model with heterogeneous additive network effects, encompassing a broad class of network interference sources, including spillover, peer effects, and contagion. First, we characterize the limitations in estimating the total treatment effect without knowledge of the network that drives interference. By contrast, we subsequently develop a simple estimator and efficient randomized design that outputs an unbiased estimate with low variance in situations where one is given access to average historical baseline measurements prior to the experiment. Our solution does not require knowledge of the underlying network structure, and it comes with statistical guarantees for a broad class of models. Due to their ease of interpretation and implementation, and their theoretical guarantees, we believe our results will have significant impact on the design of randomized experiments.

Subsequent Memory Effects in Cortical Pattern Similarity Differ by Semantic Class.

Although living and nonliving stimuli are known to rely on distinct brain regions during perception, it is largely unknown if their episodic memory encoding mechanisms differ as well. To investigate this issue, we asked participants to encode object pictures (e.g., a picture of a tiger) and to retrieve them later in response to their names (e.g., word "tiger"). For each of four semantic classes (living-animate, living-inanimate, nonliving-large, and nonliving-small), we examined differences in the similarity in activation patterns (neural pattern similarity [NPS]) for subsequently remembered versus forgotten items. Higher NPS for remembered items suggests an advantage of within-class item similarity, whereas lower NPS for remembered items indicates an advantage for item distinctiveness. We expect NPS within class-specific regions to be higher for remembered than for forgotten items. For example, the parahippocampal cortex has a well-known role in scene processing [Aminoff, E. M., Kveraga, K., & Bar, M. The role of the parahippocampal cortex in cognition. Trends in Cognitive Sciences, 17, 379-390, 2013], and the anterior temporal and inferior frontal gyrus have well-known roles in object processing [Clarke, A., & Tyler, L. K. Object-specific semantic coding in human perirhinal cortex. Journal of Neuroscience, 34, 4766-4775, 2014]. As such, we expect to see higher NPS for remembered items in these regions pertaining to scenes and objects, respectively. Consistent with this hypothesis, in fusiform, parahippocampal, and retrosplenial regions, higher NPS predicted memory for subclasses of nonliving objects, whereas in the left inferior frontal and left retrosplenial regions, lower NPS predicted memory for subclasses of living objects. Taken together, the results support the idea that subsequent memory depends on a balance of similarity and distinctiveness and demonstrate that the neural mechanisms of episodic encoding differ across semantic categories.

TPQCI: A topology potential-based method to quantify functional influence of copy number variations.

Copy number variation (CNV) is a major type of chromosomal structural variation that play important roles in many diseases including cancers. Due to genome instability, a large number of CNV events can be detected in diseases such as cancer. Therefore, it is important to identify the functionally important CNVs in diseases, which currently still poses a challenge in genomics. One of the critical steps to solve the problem is to define the influence of CNV. In this paper, we provide a topology potential based method, TPQCI, to quantify this kind of influence by integrating statistics, gene regulatory associations, and biological function information. We used this metric to detect functionally enriched genes on genomic segments with CNV in breast cancer and multiple myeloma and discovered biological functions influenced by CNV. Our results demonstrate that, by using our proposed TPQCI metric, we can detect disease-specific genes that are influenced by CNVs. Source codes of TPQCI are provided in Github (https://github.com/usos/TPQCI).

Systemic Dietary Hesperidin Modulation of Osteoclastogenesis, Bone Homeostasis and Periodontal Disease in Mice.

This study aimed to evaluate the effects of hesperidin (HE) on in vitro osteoclastogenesis and dietary supplementation on mouse periodontal disease and femoral bone phenotype. RAW 264.7 cells were stimulated with RANKL in the presence or absence of HE (1, 100 or 500 µM) for 5 days, and evaluated by TRAP, TUNEL and Western Blot (WB) analyses. In vivo, C57BL/6 mice were given HE via oral gavage (125, 250 and 500 mg/kg) for 4 weeks. A sterile silk ligature was placed between the first and second right maxillary molars for 10 days and microcomputed tomography (µCT), histopathological and immunohistochemical evaluation were performed. Femoral bones subjected or not to dietary HE (500 mg/kg) for 6 and 12 weeks were evaluated using µCT. In vitro, HE 500 µM reduced formation of RANKL-stimulated TRAP-positive(+) multinucleated cells (500 µM) as well as c-Fos and NFATc1 protein expression (p < 0.05), markers of osteoclasts. In vivo, dietary HE 500 mg/kg increased the alveolar bone resorption in ligated teeth (p < 0.05) and resulted in a significant increase in TRAP+ cells (p < 0.05). Gingival inflammatory infiltrate was greater in the HE 500 mg/kg group even in the absence of ligature. In femurs, HE 500 mg/kg protected trabecular and cortical bone mass at 6 weeks of treatment. In conclusion, HE impaired in vitro osteoclastogenesis, but on the contrary, oral administration of a high concentration of dietary HE increased osteoclast numbers and promoted inflammation-induced alveolar bone loss. However, HE at 500 mg/kg can promote a bone-sparing effect on skeletal bone under physiological conditions.

TPSC: a module detection method based on topology potential and spectral clustering in weighted networks and its application in gene co-expression module discovery.

BACKGROUND: Gene co-expression networks are widely studied in the biomedical field, with algorithms such as WGCNA and lmQCM having been developed to detect co-expressed modules. However, these algorithms have limitations such as insufficient granularity and unbalanced module size, which prevent full acquisition of knowledge from data mining. In addition, it is difficult to incorporate prior knowledge in current co-expression module detection algorithms. RESULTS: In this paper, we propose a novel module detection algorithm based on topology potential and spectral clustering algorithm to detect co-expressed modules in gene co-expression networks. By testing on TCGA data, our novel method can provide more complete coverage of genes, more balanced module size and finer granularity than current methods in detecting modules with significant overall survival difference. In addition, the proposed algorithm can identify modules by incorporating prior knowledge. CONCLUSION: In summary, we developed a method to obtain as much as possible information from networks with increased input coverage and the ability to detect more size-balanced and granular modules. In addition, our method can integrate data from different sources. Our proposed method performs better than current methods with complete coverage of input genes and finer granularity. Moreover, this method is designed not only for gene co-expression networks but can also be applied to any general fully connected weighted network.

Euglycaemic ketoacidosis in pregnant women with COVID-19: two case reports.

BACKGROUND: Euglycaemic ketoacidosis (EKA) is an infrequent but serious condition which usually follows a period of starvation, severe vomiting or illness in individuals with or without diabetes. Ketoacidosis is associated with materno-fetal morbidity and mortality necessitating prompt diagnosis and management. Physiological increases in insulin resistance render pregnancy a diabetogenic state with increased susceptibility to ketosis. COVID-19 is associated with worse clinical outcomes in patients with diabetes and is an independent risk factor for ketoacidosis in normoglycaemic individuals. CASE PRESENTATIONS: We describe two cases of SARS-CoV-2 positive pregnant women presenting with normoglycaemic metabolic ketoacidosis. Both cases were associated with maternal and fetal compromise, requiring aggressive fluid and insulin resuscitation and early delivery. CONCLUSION: We discuss possible physiology and propose a management strategy for euglycaemic ketoacidosis in pregnancy.

LAmbDA: label ambiguous domain adaptation dataset integration reduces batch effects and improves subtype detection.

MOTIVATION: Rapid advances in single cell RNA sequencing (scRNA-seq) have produced higher-resolution cellular subtypes in multiple tissues and species. Methods are increasingly needed across datasets and species to (i) remove systematic biases, (ii) model multiple datasets with ambiguous labels and (iii) classify cells and map cell type labels. However, most methods only address one of these problems on broad cell types or simulated data using a single model type. It is also important to address higher-resolution cellular subtypes, subtype labels from multiple datasets, models trained on multiple datasets simultaneously and generalizability beyond a single model type. RESULTS: We developed a species- and dataset-independent transfer learning framework (LAmbDA) to train models on multiple datasets (even from different species) and applied our framework on simulated, pancreas and brain scRNA-seq experiments. These models mapped corresponding cell types between datasets with inconsistent cell subtype labels while simultaneously reducing batch effects. We achieved high accuracy in labeling cellular subtypes (weighted accuracy simulated 1 datasets: 90%; simulated 2 datasets: 94%; pancreas datasets: 88% and brain datasets: 66%) using LAmbDA Feedforward 1 Layer Neural Network with bagging. This method achieved higher weighted accuracy in labeling cellular subtypes than two other state-of-the-art methods, scmap and CaSTLe in brain (66% versus 60% and 32%). Furthermore, it achieved better performance in correctly predicting ambiguous cellular subtype labels across datasets in 88% of test cases compared with CaSTLe (63%), scmap (50%) and MetaNeighbor (50%). LAmbDA is model- and dataset-independent and generalizable to diverse data types representing an advance in biocomputing. AVAILABILITY AND IMPLEMENTATION: github.com/tsteelejohnson91/LAmbDA. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Providing visual directives via a smart watch to a student with Autism Spectrum Disorder: an intervention note.

Smart watches are discreet and wearable tools that may be repurposed to improve directive-following for individuals with autism spectrum disorder (ASD). Specifically, a mentor can transmit just-in-time (JIT) visual supports (e.g., video clips, photographs, text) that depict an upcoming directive to a learner's smart watch to prompt the learner as needed from a distance. Using a single-case multiple probe design across settings, this investigation evaluated the effectiveness of providing text-based prompts on an Apple Watch 1 to a child with ASD within a school setting. A mentor transmitted 2-step written directives via text message to the participant's Apple Watch. The participant was instructed to attend to, read, and follow directives received on the watch. Results demonstrated that the intervention improved directive-following as well as increased the instructor's distance from the learner. It is proposed that JIT supports sent to a learner's smart watch may reduce the obtrusiveness of traditional prompting while also maintaining the naturalness of ongoing social or academic interactions. Clinical limitations and implications are discussed.

In Operando Visualization of Elementary Turnovers in Photocatalytic Organic Synthesis.

We report the in operando visualization of the photocatalytic turnovers on single eosin Y (EY) through a redox-induced photoblinking phenomenon. The photocatalytic cyclization of thiobenzamide (TB) catalyzed by EY was investigated. The analysis of the intensity-versus-time trajectories of single EYs revealed the kinetics and dynamics of the elementary photocatalytic turnovers and the heterogeneity of the activity of individual EYs. The quenching turnover time showed a fast population and a slow population, which could be attributed to the singlet and triplet states of photoexcited EY. The slow quenching turnovers were more dominant at higher TB concentrations. The activity heterogeneity of EYs was studied over a series of reactant concentrations. Excess quenching reagent was found to decrease the percentage of active EYs. The method can be broadly applied to studying the elementary processes of photocatalytic organic reactions in operando.

Mitigation of BMP-induced inflammation in craniofacial bone regeneration and improvement of bone parameters by dietary hesperidin.

Based on anti-inflammatory and osteogenic properties of hesperidin (HE), we hypothesized its systemic administration could be a cost-effective method of improving BMP-induced bone regeneration. Sprague-Dawley rats were allocated into 4 groups (n = 10/group): a 5-mm critical-sized mandible defect + collagen scaffold or, scaffold + 1 µg of BMP2 with and without dietary HE at 100 mg/kg. HE was administered by oral gavage 4 weeks prior to surgeries until euthanasia at day 7 or 14 post-surgery. The healing tissue within the defect collected at day 7 was subjected to gene expression analysis. Mandibles harvested at day 14 were subjected to microcomputed tomography and histology. HE + BMP2-treated rats had a statistically significant decrease in expression of inflammatory genes compared to BMP2 alone. The high-dose BMP2 alone caused cystic-like regeneration with incomplete defect closure. HE + BMP2 showed virtually complete bone fusion. Collagen fibril birefringence pattern (red color) under polarized light indicated high organization in BMP2-induced newly formed bone (NFB) in HE-supplemented group (p < 0.05). Clear changes in osteocyte lacunae as well as a statistically significant increase in osteoclasts were found around NFB in HE-treated rats. A significant increase in trabecular volume and thickness, and trabecular and cortical density was found in femurs of HE-supplemented rats (p < 0.05). Our findings show, for the first time, that dietary HE has a remarkable modulatory role in the function of locally delivered high-dose BMP2 in bone regeneration possibly via control of inflammation, osteogenesis, changes in osteocyte and osteoclast function and collagen maturation in regenerated and native bone. In conclusion, HE had a significant skeletal bone sparing effect and the ability to provide a more effective BMP-induced craniofacial regeneration.

Improving accuracy of GPT-3/4 results on biomedical data using a retrieval-augmented language model.

Large language models (LLMs) have made a significant impact on the fields of general artificial intelligence. General purpose LLMs exhibit strong logic and reasoning skills and general world knowledge but can sometimes generate misleading results when prompted on specific subject areas. LLMs trained with domain-specific knowledge can reduce the generation of misleading information (i.e. hallucinations) and enhance the precision of LLMs in specialized contexts. Training new LLMs on specific corpora however can be resource intensive. Here we explored the use of a retrieval-augmented generation (RAG) model which we tested on literature specific to a biomedical research area. OpenAI's GPT-3.5, GPT-4, Microsoft's Prometheus, and a custom RAG model were used to answer 19 questions pertaining to diffuse large B-cell lymphoma (DLBCL) disease biology and treatment. Eight independent reviewers assessed LLM responses based on accuracy, relevance, and readability, rating responses on a 3-point scale for each category. These scores were then used to compare LLM performance. The performance of the LLMs varied across scoring categories. On accuracy and relevance, the RAG model outperformed other models with higher scores on average and the most top scores across questions. GPT-4 was more comparable to the RAG model on relevance versus accuracy. By the same measures, GPT-4 and GPT-3.5 had the highest scores for readability of answers when compared to the other LLMs. GPT-4 and 3.5 also had more answers with hallucinations than the other LLMs, due to non-existent references and inaccurate responses to clinical questions. Our findings suggest that an oncology research-focused RAG model may outperform general-purpose LLMs in accuracy and relevance when answering subject-related questions. This framework can be tailored to Q&A in other subject areas. Further research will help understand the impact of LLM architectures, RAG methodologies, and prompting techniques in answering questions across different subject areas.

Sex-based trajectories of health system use in lonely and not lonely older people: A population-based cohort study.

BACKGROUND: There is growing interest in understanding the care needs of lonely people but studies are limited and examine healthcare settings separately. We estimated and compared healthcare trajectories in lonely and not lonely older female and male respondents to a national health survey. METHODS: We conducted a retrospective cohort study of community-dwelling, Ontario respondents (65+ years) to the 2008/2009 Canadian Community Health Survey-Healthy Aging. Respondents were classified at baseline as not lonely, moderately lonely, or severely lonely using the Three-Item Loneliness Scale and then linked with health administrative data to assess healthcare transitions over a 12 -year observation period. Annual risks of moving from the community to inpatient, long-stay home care, long-term care settings-and death-were estimated across loneliness levels using sex-stratified multistate models. RESULTS: Of 2684 respondents (58.8% female sex; mean age 77 years [standard deviation: 8]), 635 (23.7%) experienced moderate loneliness and 420 (15.6%) severe loneliness. Fewer lonely respondents remained in the community with no transitions (not lonely, 20.3%; moderately lonely, 17.5%; and severely lonely, 12.6%). Annual transition risks from the community to home care and long-term care were higher in female respondents and increased with loneliness severity for both sexes (e.g., 2-year home care risk: 6.1% [95% CI 5.5-6.6], 8.4% [95% CI 7.4-9.5] and 9.4% [95% CI 8.2-10.9] in female respondents, and 3.5% [95% CI 3.1-3.9], 5.0% [95% CI 4.0-6.0], and 5.4% [95% CI 4.0-6.8] in male respondents; 5-year long-term care risk: 9.2% [95% CI 8.0-10.8], 11.1% [95% CI 9.3-13.6] and 12.2% [95% CI 9.9-15.3] [female], and 5.3% [95% CI 4.2-6.7], 9.1% [95% CI 6.8-12.5], and 10.9% [95% CI 7.9-16.3] [male]). CONCLUSIONS: Lonely older female and male respondents were more likely to need home care and long-term care, with severely lonely female respondents having the highest probability of moving to these settings.

Increased glycemic variability in pregnant women with Roux-en-Y gastric bypass compared with sleeve gastrectomy.

INTRODUCTION: Bariatric surgery is associated with adverse pregnancy outcomes such as reduced birth weight and premature birth. One possible mechanism for this is increased glycemic variability (GV) which occurs after bariatric surgery. The objective of this study was to compare the effect of Roux-en-Y gastric bypass (RYGB) versus vertical sleeve gastrectomy (SG) on GV during pregnancy and to investigate the relationships of GV, type of bariatric surgery and maternal and neonatal outcomes. RESEARCH DESIGN AND METHODS: Fourteen pregnant women after RYGB and 14 after SG were investigated with continuous glucose monitoring in their second or third trimester in this observational study carried out as part of routine clinical care. RESULTS: Pregnant women with RYGB had similar mean interstitial glucose values but significantly increased indices of GV and a lower %time in range 3.9-7.8 mmol/L (70-140 mg/dL), compared with SG. CONCLUSIONS: Pregnant women who have undergone RYGB have greater GV during pregnancy compared with those who have undergone SG. Further research is needed to establish the relationship between GV and pregnancy outcomes to determine the preferred bariatric operation in women of reproductive age, and whether interventions to reduce GV might improve outcomes.

Mechanism-centric regulatory network identifies NME2 and MYC programs as markers of Enzalutamide resistance in CRPC.

Heterogeneous response to Enzalutamide, a second-generation androgen receptor signaling inhibitor, is a central problem in castration-resistant prostate cancer (CRPC) management. Genome-wide systems investigation of mechanisms that govern Enzalutamide resistance promise to elucidate markers of heterogeneous treatment response and salvage therapies for CRPC patients. Focusing on the de novo role of MYC as a marker of Enzalutamide resistance, here we reconstruct a CRPC-specific mechanism-centric regulatory network, connecting molecular pathways with their upstream transcriptional regulatory programs. Mining this network with signatures of Enzalutamide response identifies NME2 as an upstream regulatory partner of MYC in CRPC and demonstrates that NME2-MYC increased activities can predict patients at risk of resistance to Enzalutamide, independent of co-variates. Furthermore, our experimental investigations demonstrate that targeting MYC and its partner NME2 is beneficial in Enzalutamide-resistant conditions and could provide an effective strategy for patients at risk of Enzalutamide resistance and/or for patients who failed Enzalutamide treatment.

Second-trimester uterine artery Doppler in the prediction of stillbirths.

OBJECTIVE: To examine the role of second-trimester uterine artery Doppler in the prediction of stillbirths. METHODS: Uterine artery pulsatility index (PI) was measured at 20-24 weeks' gestation in 65,819 singleton pregnancies. The PI was converted to multiples of median (MoM) and compared in live births and stillbirths. Regression analysis was used to determine the significance of association between log(10) uterine artery PI MoM and gestational age (GA) at delivery in cases of stillbirths. RESULTS: There were 306 (0.46%) stillbirths and in 159 (52.0%) of these there was pre-eclampsia (PE), placental abruption and/or birthweight below the 10th percentile (small for gestational age, SGA). In the stillbirths, the uterine artery PI MoM was significantly higher than in live births and was inversely associated with GA at delivery. The uterine artery PI MoM was above the 90th percentile in 80.6% of stillbirths with PE, abruption and/or SGA delivering at <32 weeks' gestation, in 41.9% at 33-36 weeks and in 34.3% at ≥37 weeks, and the respective percentages for stillbirths without PE, abruption or SGA were 15.8, 25.0 and 12.4%. CONCLUSION: Second-trimester uterine artery PI is effective in identifying early stillbirths in association with PE, abruption or SGA but not late deaths in the absence of PE, abruption or SGA.

High intratumoral plasma cells content in primary prostate cancer defines a subset of tumors with potential susceptibility to immune-based treatments.

BACKGROUND: Data on advanced prostate cancer (PCa) suggest more prior systemic therapies might reduce tumor immune responsiveness. In treatment-naïve primary PCa, recent work correlated intratumoral plasma cell content with enhanced tumor immune-responsiveness. We sought to identify features of localized PCa at a high risk of recurrence following local treatment with high plasma cell content to help focus future immune-based neoadjuvant trials. METHODS: We performed retrospective analyses of molecular profiles from three independent cohorts of over 1300 prostate tumors. We used Wilcoxon Rank Sum to compare molecular pathways between tumors with high and low intratumoral plasma cell content and multivariable Cox proportional hazards regression analyses to assess metastasis-free survival. RESULTS: We validated an expression-based signature for intratumoral plasma cell content in 113 primary prostate tumors with both RNA-expression data and digital image quantification of CD138+ cells (plasma cell marker) based on immunohistochemisty. The signature showed castration-resistant tumors (n = 101) with more prior systemic therapies contained lower plasma cell content. In high-grade primary PCa, tumors with high plasma cell content were associated with increased predicted response to immunotherapy and decreased response to androgen-deprivation therapy. Master regulator analyses identified upregulated transcription factors implicated in immune (e.g. SKAP1, IL-16, and HCLS1), and B-cell activity (e.g. VAV1, SP140, and FLI-1) in plasma cell-high tumors. Master regulators overactivated in tumors with low plasma cell content were associated with shorter metastasis-free survival following radical prostatectomy. CONCLUSIONS: Markers of plasma cell activity might be leveraged to augment clinical trial targeting and selection and better understand the potential for immune-based treatments in patients with PCa at a high risk of recurrence following local treatment.

Mitigation of BMP-induced Inflammation in Craniofacial Bone Regeneration and Improvement of Bone Parameters by Dietary Hesperidin.

Based on anti-inflammatory and osteogenic properties of hesperidin (HE), we hypothesized its systemic administration could be a cost-effective method of improving BMP-induced bone regeneration. Sprague-Dawley rats were allocated into 4 groups (n=10/group): a 5-mm critical-sized mandible defect + collagen scaffold or, scaffold + 1 µg of BMP2 with and without dietary HE at 100 mg/kg. HE was administered by oral gavage 4 weeks prior to surgeries until euthanasia at day 7 or 14. The healing tissue within the defect collected at day 7 was subjected to gene expression analysis. Mandibles harvested at day 14 were subjected to microcomputed tomography and histology. HE+BMP2-treated rats had a statistically significant decrease in expression of inflammatory genes compared to BMP2 alone. The high-dose BMP2 caused cystic-like regeneration with incomplete defect closure. HE+BMP2 showed virtually complete bone fusion. Red collagen fibrils were significantly higher in BMP2-induced newly formed bone (NFB) in HE-supplemented group (p<0.05) indicating high organization. Clear changes in osteocyte lacunae as well as a statistically significant increase in osteoclasts were found around NFB in HE rats. A significant increase in trabecular volume and thickness, and trabecular and cortical density was found in femurs of HE-supplemented rats (p<0.05). Our findings show, for the first time, that dietary HE has a remarkable modulatory role in locally delivered high-dose BMP2-induced bone possibly via control of inflammation, osteogenesis, changes in osteocyte and osteoclast function and collagen maturation in regenerated and native bone. In conclusion, HE has a significant skeletal bone sparing effect and the ability to provide a more effective BMP-induced craniofacial regeneration.

The "not so good" thyroid cancer: a scoping review on risk factors associated with anxiety, depression and quality of life.

The incidence of thyroid cancer has increased in recent years, leading to a growing number of survivors facing lifelong consequences. This scoping review investigated anxiety, depression, and quality of life (QoL) in thyroid cancer survivors compared to the general population, those with benign pathology, and survivors of other types of cancers. Moreover, we aimed to identify the risk factors associated with anxiety, depression, and QoL in thyroid cancer patients. A total of 727 articles were identified through PubMed, ProQuest, Cochrane, and Google Scholar databases, and 68 articles that met the criteria were selected for data extraction. Thyroid cancer survivors have a poorer QoL compared to the general population, population with benign pathology, and survivors of other types of cancer associated with worse clinical outcomes. The main risk factors are grouped into socioeconomic factors, disease-specific factors, management factors, comorbidities, and patient perceptions. Effective communication between the patient and the medical team and behavioral interventions may reduce these risks. Despite the common perception of thyroid cancer as a "good cancer," the findings of this review demonstrate the need to address the risk factors associated with increased anxiety, depression, and lower QoL in survivors.

A target discovery pipeline identified ILT3 as a target for immunotherapy of multiple myeloma.

Multiple myeloma (MM) is an incurable malignancy of plasma cells. To identify targets for MM immunotherapy, we develop an integrated pipeline based on mass spectrometry analysis of seven MM cell lines and RNA sequencing (RNA-seq) from 900+ patients. Starting from 4,000+ candidates, we identify the most highly expressed cell surface proteins. We annotate candidate protein expression in many healthy tissues and validate the expression of promising targets in 30+ patient samples with relapsed/refractory MM, as well as in primary healthy hematopoietic stem cells and T cells by flow cytometry. Six candidates (ILT3, SEMA4A, CCR1, LRRC8D, FCRL3, IL12RB1) and B cell maturation antigen (BCMA) present the most favorable profile in malignant and healthy cells. We develop a bispecific T cell engager targeting ILT3 that shows potent killing effects in vitro and decreased tumor burden and prolonged mice survival in vivo, suggesting therapeutic relevance. Our study uncovers MM-associated antigens that hold great promise for immune-based therapies of MM.