Abstract 6689: Whole genome cell-free tumor DNA mutational signatures from blood for early detection of recurrence of low stage lung adenocarcinoma

Introduction: Lung cancer remains the leading cause of cancer-related deaths. Surgery is the best option for early lung cancer, and the role of adjuvant therapy remains controversial. Liquid biopsy offers a noninvasive approach to monitor cancer burden. Targeted sequencing of circulating cell free tumor DNA (ctDNA) in blood has shown success for diagnosis; however, low tumor burden and dynamic evolution of low stage disease is challenging for targeted panels. Thus, we hypothesized that a whole genome sequencing (WGS)-derived patient specific mutational signature from a matched tumor-normal WGS can provide sensitive and specific approach to detect mutations and copy numbers in ctDNA for monitoring of lung adenocarcinoma patients.

Methods: We successfully profiled 50 Stage 1 or 2 lung adenocarcinomas. ctDNA was extracted from 1-2 mL of plasma, tumor DNA was extracted from pathology tissue and normal germline DNA from the white blood cells. WGS using was performed on matched tumor and normal DNA, and ctDNA extracted from plasma. WGS coverage was 40x for matched tumor-normal and 20x for ctDNA. We derived a personalized mutational pattern for each tumor and used an AI-based error suppression model for quantification and ultra-sensitive detection of ctDNA in plasma samples. A patient-specific personalized genome-wide compendium of somatic mutations and copy numbers was established and ctDNA tested at 3 to 18 available time points during the therapy or follow up. A personalized mutational signature for detection ctDNA from WGS was quantified and the ctDNA Tumor Fraction (TF) was compared to the clinical status and time to recurrence.

Results: Tumor specific signatures were derived from matched tumor-normal samples with >5% tumor purity and <30% duplications rate. Out of all patients, 33 patients showed no recurrence and 12 recurred. Tumor-specific signatures detected the presence of the tumor signature in plasma with TF as low as 10−5. Based on positive minimal residual disease in plasma, the recurrence prediction sensitivity was 0.75 and specificity 0.82, with positive predictive value of 0.6 and negative predictive value 0.9. WGS ctDNA predicted recurrence with a median lead time of 508 days before clinical/imaging recurrence. In one case we were able to identify the second primary by deconvoluting known and novel ctDNA mutations. ctDNA mutational profiles enabled identification of smoking mutational signature matching clinical history, and APOBEC and ageing signatures as well as tumor mutational burden.

Conclusions: Patient-specific WGS tumor signature from plasma derived ctDNA enables specific and ultrasensitive tracking of minimal residual disease in low stage lung adenocarcinoma patients. Molecularly positive status can be used to predict recurrence and identify patients with clinical low stage disease that may benefit from adjuvant therapy.

Citation Format: Ivy Tran, Alejandro Vargas, Reid Wilkins, Isabella Pizzillo, Kenneth Tokoro, Danielle Afterman, Tomer Lauterman, Maja Kuzman, Santiago Gonzalez, Dunja Glavas, James Smadbeck, Dillon Maloney, Jurica Levatic, Samuel Phillips, Sunil Deochand, Michael Yahalom, Ryan Ptashkin, Iman Tavassoly, Zohar Donenhirsh, Eric White, Ravi Kandasamy, Ury Alon, Paz Polak, Boris Oklander, Asaf Zviran, Matija Snuderl, Harvey I. Pass. Whole genome cell-free tumor DNA mutational signatures from blood for early detection of recurrence of low stage lung adenocarcinoma. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6689.

Decoding clinical biomarker space of COVID-19: Exploring matrix factorization-based feature selection methods

One of the most critical challenges in managing complex diseases like COVID-19 is to establish an intelligent triage system that can optimize the clinical decision-making at the time of a global pandemic. The clinical presentation and patients' characteristics are usually utilized to identify those patients who need more critical care. However, the clinical evidence shows an unmet need to determine more accurate and optimal clinical biomarkers to triage patients under a condition like the COVID-19 crisis. Here we have presented a machine learning approach to find a group of clinical indicators from the blood tests of a set of COVID-19 patients that are predictive of poor prognosis and morbidity. Our approach consists of two interconnected schemes: Feature Selection and Prognosis Classification. The former is based on different Matrix Factorization (MF)-based methods, and the latter is performed using Random Forest algorithm. Our model reveals that Arterial Blood Gas (ABG) O₂ Saturation and C-Reactive Protein (CRP) are the most important clinical biomarkers determining the poor prognosis in these patients. Our approach paves the path of building quantitative and optimized clinical management systems for COVID-19 and similar diseases.

Abstract 3401: Whole genome cell-free tumor DNA mutational signatures for noninvasive monitoring of pediatric brain cancers

Introduction: Liquid biopsy offers a noninvasive approach to monitor cancer burden during therapy and surveillance period. However, in pediatric brain cancers, liquid biopsy methods from the blood have been unsuccessful due to a low tumor burden and low number of mutations in coding regions. We hypothesized that a whole genome sequencing (WGS)-derived patient specific mutational signature from a matched tumor-normal WGS can provide a sensitive and specific approach to detect mutations in circulating cell free tumor DNA (ctDNA) and provide blood-based monitoring in pediatric patients with brain tumor.

Methods: All tumors were analyzed and molecularly subclassified using whole genome DNA methylation profiling and machine learning classifier. Tumor DNA was extracted from pathology tissue and normal germline DNA from the white blood cells, while ctDNA was extracted from 1-2 mL of post-surgery or follow-up plasma samples, WGS was applied to sequence DNA from matched tumor-normal and plasma samples. WGS coverage was 40x for matched tumor-normal DNA and 20x for ctDNA. Using the C2i assay, we derived a personalized mutational pattern for each tumor and used an AI-based error suppression model for quantification and ultra-sensitive detection of ctDNA in plasma samples. A patient-specific personalized genome-wide compendium of somatic mutations was established and ctDNA tested at 1 to 3 available time points during the therapy or surveillance period. An AI-based error suppression model was implemented to filter out the noise in the cell free DNA (cfDNA) while the personalized mutational signature was used to detect the ctDNA in the cfDNA and to amplify the somatic signal contained in it. The ctDNA Tumor Fraction (TF) was compared to the clinical status and MR-based imaging.

Results: We profiled 7 pediatric brain tumors, including 2 medulloblastomas (one Group 3, one Group 4), 3 pediatric glioblastomas IDH wild-type, 1 ependymoma PFA subtype and one low grade ganglioglioma. Tumor specific signatures were identified and detected in the plasma of 5 patients with clinical disease with a TF range 0.02-0.0005 but not in 2 patients with no tumor at the time of blood collection. In two children with a medulloblastoma and glioblastoma, the decrease of tumor fraction in ctDNA over 2 (TF: 0.002 to 0.0009) and 3 time points (TF: 0.0005 to undetectable), respectively, correlated with response to therapy based on imaging.

Conclusions: Patient-specific WGS tumor signature in ctDNA from blood can be used for sensitive monitoring of children with brain tumors.

Citation Format: Ivy Tran, Kristyn Galbraith, Guisheng Zhao, Robyn Borsuk, Joyce Varkey, Sharon Gardner, Jeffrey Allen, David Harter, Jeffrey Wisoff, Eveline T. Hidalgo, Sunil Deochand, Dillon Maloney, Danielle Afterman, Tomer Lauterman, Noah Friedman, Imane Bourzgui, Nidhi Ramaraj, Zohar Donenhirsh, Ronel Veksler, Jonathan Rosenfeld, Ravi Kandasamy, Iman Tavassoly, Boris Oklander, G. Praveen Raju, Theodore Nicolaides, Asaf Zviran, Matija Snuderl. Whole genome cell-free tumor DNA mutational signatures for noninvasive monitoring of pediatric brain cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3401.

Abstract 1959: Sensitive detection of circulating tumor DNA by whole genome sequencing: Validation of MRDetect using serial blood samples from stage III colorectal cancer patients

Background: While detection of circulating tumor DNA (ctDNA) is associated with poor cancer prognosis, the clinical utility for guiding treatment decisions is unresolved. Patients with minimal residual disease (MRD) often have less than one genome equivalent of ctDNA per 10 mL blood. Consequently, it is stochastic whether a 10 mL sample contains ctDNA from a particular genomic locus. Consequently, the sensitivity of ctDNA detection methods targeting a limited number of tumor loci is heavily affected by sampling bias. To overcome this challenge, we developed MRDetect; a whole genome sequencing (WGS) approach, which detects ctDNA using the patient-specific cumulative signal from tens of thousands of mutations throughout the genome. Recently, we showed how MRDetect found ctDNA fractions down to 10-4. Here, we performed a validation study to confirm the prognostic impact of MRDetect.

Aim: Validation of MRDetect for sensitive ctDNA detection to monitor residual disease in stage III colorectal cancer (CRC) patients treated with curative intent.

Methods: From a large, uniform cohort of stage III CRC patients n = 146), we had plasma samples collected every third month (n = 938, median = 9 per patient) and a median follow-up of 34 months. For each patient, a genome-wide mutational signature was established by WGS of tumor and matched normal DNA. Enhanced by an AI-based error suppression model, this signature was used to detect ctDNA in 1-2 mL plasma samples using WGS (20x coverage). We used de-novo point mutation and copy number variation analysis to investigate cancer evolution after treatment. To evaluate the reproducibility of MRDetect, aliquot samples (n = 2x190 samples) from 5 recurrence and 10 non-recurrence patients were processed and sequenced at two independent laboratories. Outcome measures: ctDNA status, tumor fraction, false positive rate, Time To ctDNA Recurrence (TTcR), and Time To radiological Recurrence (TTrR).

Results: Analysis of paired samples showed great reproducibility with high agreement between both ctDNA status calls (Cohens Kappa = 0.81) and the estimated tumor fractions (r2 = 0.99). MRDetect revealed post-operative ctDNA in all recurrence patients (5/5) with detected tumor fractions down to 2 x 10-4. Median TTcR was 0.9 month (range 0.5 - 7.3 months) while median TTrR was 12.8 months (range 11.3 - 31.1 months). The false positive rate was 1% (1/100), assessed in longitudinal samples from the 10 non-relapsing patients. Tumor evolution dynamics in plasma samples revealed novel amplification and deletions, which were absent in the primary tissue but confirmed in metachronous metastases. We will present results from the full cohort at AACR 2022.

Conclusion: MRDetect detects ctDNA with high sensitivity and specificity and enables effective postoperative assessment of MRD, cancer evolution dynamics and early relapse detection.

Citation Format: Amanda Frydendahl, Thomas Reinert, Jesper Nors, Sunil Deochand, Dillon Maloney, Noah Friedman, Tomer Lauterman, Danielle Afterman, Imane Bourzgui, Nidhi Ramaraj, Zohar Donenhirsh, Ronel Veksler, Ravi Kandasamy, Iman Tavassoly, Jonathan Rosenfeld, Anders Husted Andersen, Uffe S. Løve, Per V. Andersen, Ole Thorlacius-Ussing, Lene Hjerrild Iversen, Kåre Andersson Gotschalck, Boris Oklander, Asaf Zviran, Claus Lindbjerg Andersen. Sensitive detection of circulating tumor DNA by whole genome sequencing: Validation of MRDetect using serial blood samples from stage III colorectal cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1959.

Abstract 540: Genome-wide circulating tumor DNA for monitoring treatment response and metastatic relapse in bladder cancer

Background: Neoadjuvant chemotherapy (NAC) followed by radical cystectomy (CX), is gold standard treatment in localized muscle-invasive bladder cancer (MIBC). About 45% of patients with MIBC develop metastatic relapse within 2 years after CX. The response rate to chemotherapy and immune checkpoint inhibitors (ICI) is relatively low, and biomarker tests for monitoring response are needed. Furthermore, biomarkers for early detection of minimal residual disease (MRD) after CX is needed to enable earlier treatment initiation. Tumor-informed detection of mutations in cell-free DNA (cfDNA) from peripheral blood has shown promising results in its ability to monitor MRD. However, the low tumor fraction after surgery and limited input material obtained from a typical plasma sample limits the probability of detecting low metastatic burden scenarios. Here we implemented and applied locally a whole-genome sequencing (WGS) approach to circulating tumor DNA (ctDNA) monitoring for improving ctDNA detection.

Methods: A total of 140 MIBC patients undergoing NAC and CX were enrolled, including a test cohort (n=19) and a validation cohort (n=120). cfDNA was extracted from ~1mL plasma (n=1100) and procured from longitudinal plasma sampling during NAC (response measure), pre-cystectomy (response measure), post-surgery (relapse monitoring) and during immunotherapy (ICI treatment). WGS was applied to tumor/germline pairs (coverage >30x/20x) and plasma cfDNA (>20x) facilitating detection of genome wide genomic alterations and quantification of ctDNA using the MRDetect method.

Results: We developed a personalized tumor-informed WGS model by integrating genome-wide mutation and copy number variation data coupled with advanced signal processing and AI-based error suppression. Patient-specific somatic variant patterns were then used for detecting and measuring the ctDNA levels in low-input blood samples by WGS. The assay sensitivity allowed for detection of tumor fractions down to 8*10-5. Furthermore, in our test cohort of 19 patients, we detected ctDNA after CX in 7 of 8 patients with clinical relapse (88% sensitivity) and detected no ctDNA in 11 of 11 patients with no clinical relapse (100% specificity). We observed a positive lead-time for MRD-based recurrence detection compared to CT-based reccurence detection (9 months on average). The full dataset is currently being processed and will be presented at the AACR 2022 meeting.

Conclusions: For precision oncology, we need to develop quantitative and non-invasive methodologies to help tailor the treatments to individual patients and monitor them for further clinical decision-making. The results indicate the clinical potential of personalized genome-wide mutation integration as an ultra-sensitive, non-invasive method for MRD detection and treatment response monitoring which could aid in clinical management of patients with bladder cancer.

Citation Format: Iver Nordentoft, Karin Birkenkamp-Demtröder, Emil Christensen, Sunil Deochand, Dillon Maloney, Danielle Afterman, Tomer Lauterman, Noah Friedman, Imane Bourzgui, Nidhi Ramaraj, Zohar Donenhirsh, Ronel Veksler, Sia Viborg, Mads Agerbæk, Jørgen Bjerggaard Jensen, Jonathan Rosenfeld, Ravi Kandasamy, Iman Tavassoly, Boris Oklander, Asaf Zviran, Lars Dyrskjøt. Genome-wide circulating tumor DNA for monitoring treatment response and metastatic relapse in bladder cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 540.

Abstract 5114: Ultra-sensitive detection of minimal residual disease (MRD) through whole genome sequencing (WGS) using an AI-based error suppression model in resected early-stage non-small cell lung cancer (NSCLC)

Background: Early detection of recurrence and monitoring of MRD post-surgery is critical for clinical decision-making to tailor adjuvant therapy. In early-stage NSCLC, circulating tumor DNA (ctDNA) detection is especially challenging, requiring highly sensitive and specific assays. Therefore, we used a WGS approach (MRDetect) for ultra-sensitive ctDNA detection in NSCLC patients (pts) undergoing curative surgery.

Methods: We conducted a pilot study to evaluate the MRDetect approach in serial plasma samples (including pre-surgery, post-surgery and follow-up [f/u] timepoints) from resected stage IB-IIIA NSCLC pts. Pts underwent routine surveillance by computed tomography scans. ctDNA was extracted from ~1mL plasma. MRDetect uses WGS by a tumor-informed approach (sequencing coverage 40x for tumor, 20x for plasma DNA) combined with AI-based error suppression models (trained and calibrated with a non-cancer cohort, n=17) to increase the signal to noise ratio for precise ctDNA detection, and improve the accuracy of readouts especially for low tumor burden scenarios. The assay reports the detection and quantification of ctDNA burden in blood with a prognostic value for risk of recurrence. The ability of the assay to predict recurrence from a single sample, taken at the clinical landmark point (median 1.6 mths post-surgery, range 0.1-6.5) was evaluated.

Results: Overall, 52 NSCLC pts were enrolled (n=88 plasma samples) with median clinical f/u of 32.6 mths (range 3.1-98.6). There were 43 pts with post-surgery landmark samples, with median age 62 years, 70% were male, 79% were adenocarcinoma and 49% were EGFR mutated. 26% were stage IB and 37% each were stage II and III. There were 15/18 (sensitivity 83%) pts with confirmed radiological recurrence in which MRDetect was positive, including 6/7 (86%) EGFR mutated pts. The median RFS in MRDetect positive pts was 15.2 mths (range 3.7-33.4). Among 25 pts with no recurrence (median f/u 25.6 mths), MRDetect reported 4 pts to be MRD positive (specificity 84%). These results were consistent between EGFR mutated (sensitivity 86%, specificity 86%) and wildtype pts (sensitivity 82%, specificity 82%). For longitudinal samples (n=17 pts), negative ctDNA was associated with absence of recurrence in 14/15 pts (specificity 93%). At the AACR meeting, results from a planned larger validation study will be presented.

Conclusion: Using a robust WGS implemented AI-based computational platform (MRDetect), we demonstrate high sensitivity and specificity detection of MRD in both EGFR mutated and wildtype NSCLC. With an increasing number of therapeutic options in the adjuvant setting for NSCLC, an ultra-sensitive MRD assay has the potential to facilitate personalized clinical decision-making for tailoring both the need and choice of adjuvant therapies.

Citation Format: Aaron C. Tan, Stephanie P. Saw, Gillianne G. Lai, Kevin L. Chua, Angela Takano, Boon-Hean Ong, Tina P. Koh, Amit Jain, Wan Ling Tan, Quan Sing Ng, Ravindran Kanesvaran, Tanujaa Rajasekaran, Sunil Deochand, Dillon Maloney, Danielle Afterman, Tomer Lauterman, Noah Friedman, Imane Bourzgui, Nidhi Ramaraj, Zohar Donenhirsh, Ronel Veksler, Jonathan Rosenfeld, Ravi Kandasamy, Iman Tavassoly, Boris Oklander, Asaf Zviran, Wan-Teck Lim, Eng-Huat Tan, Anders J. Skanderup, Mei-Kim Ang, Daniel S. Tan. Ultra-sensitive detection of minimal residual disease (MRD) through whole genome sequencing (WGS) using an AI-based error suppression model in resected early-stage non-small cell lung cancer (NSCLC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5114.

Lab validation of an ultrasensitive ctDNA pan-cancer MRD assay using whole-genome sequencing

e13582

Background: Minimal residual disease (MRD) monitoring using liquid biopsy for solid tumors requires a highly sensitive and specific assay that can overcome the limitation of low abundance cfDNA in a standard blood draw. We developed a whole-genome sequencing (WGS)-based assay to detect the presence of circulating tumor DNA (ctDNA) in plasma. The C2i assay is a tumor-informed assay that uses personalized tumor signature, advanced noise models, and artificial intelligence (AI) modalities to interrogate plasma for the presence of ctDNA longitudinally. Methods: The C2i test was developed in accordance with CAP/CLIA and New York state validation principles. We used contrived samples to establish analytical validation of the assay performance, which was then validated with a large clinical cohort of early-stage patients across various cancer types. Briefly, aggregated tumor signatures derived from cancer cell lines were fragmented and spiked into a contrived healthy plasma pool; the mixed samples were used to assess the presence of tumor DNA signature down to tumor fractions of 10e-4. Positive samples are identified by tumor-derived variants detected above the noise levels. Noise modeling was established using a panel of normal (PON) approach. We assessed the analytical sensitivity, specificity, and accuracy using 348 contrived samples derived from five different cancer cell lines. Reproducibility and precision were assessed with multiple replicates, and statistical concordance was reported. This validation was complemented by a cohort of 200 patients and ̃1000 plasma samples across a variety of cancer types including, NSCLC, MIBC, CRC, GBM, Breast Cancer, and a mixture of other cancer types. Results: Cancer cell lines, representing the five most prevalent disease indications, used for determining analytical sensitivity are as follows: CRC HT-29, Breast SK-BR3, Bladder HT-1376, Lung HCI-H526, and Prostate LNCaP. The cell line DNA was enzymatically fragmented and size-selected to mimic ctDNA. This ctDNA was spiked into cfDNA extracted from healthy volunteers at various dilution levels, varying from 10e0 to 10e-4. The 95% probability of detecting ctDNA was established at 10e-4. The reproducibility of tumor signature between replicates was assessed to be greater than 90%. The assay was performed using both normal and maximum input amounts. These performance estimates were then validated on a cohort of plasma collected from early-stage (stage I-III) patients across various cancer types. Conclusions: C2i MRD test is an ultrasensitive pan-cancer MRD monitoring assay used in several clinical trials across the world. We present an extensive analytical and clinical validation of the assay supporting its high performance.

Functional Effects of Cardiomyocyte Injury in COVID-19

COVID-19 affects multiple organs. Clinical data from the Mount Sinai Health System show that substantial numbers of COVID-19 patients without prior heart disease develop cardiac dysfunction. How COVID-19 patients develop cardiac disease is not known. We integrated cell biological and physiological analyses of human cardiomyocytes differentiated from human induced pluripotent stem cells (hiPSCs) infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in the presence of interleukins (ILs) with clinical findings related to laboratory values in COVID-19 patients to identify plausible mechanisms of cardiac disease in COVID-19 patients. We infected hiPSC-derived cardiomyocytes from healthy human subjects with SARS-CoV-2 in the absence and presence of IL-6 and IL-1β. Infection resulted in increased numbers of multinucleated cells. Interleukin treatment and infection resulted in disorganization of myofibrils, extracellular release of troponin I, and reduced and erratic beating. Infection resulted in decreased expression of mRNA encoding key proteins of the cardiomyocyte contractile apparatus. Although interleukins did not increase the extent of infection, they increased the contractile dysfunction associated with viral infection of cardiomyocytes, resulting in cessation of beating. Clinical data from hospitalized patients from the Mount Sinai Health System show that a significant portion of COVID-19 patients without history of heart disease have elevated troponin and interleukin levels. A substantial subset of these patients showed reduced left ventricular function by echocardiography. Our laboratory observations, combined with the clinical data, indicate that direct effects on cardiomyocytes by interleukins and SARS-CoV-2 infection might underlie heart disease in COVID-19 patients. IMPORTANCE SARS-CoV-2 infects multiple organs, including the heart. Analyses of hospitalized patients show that a substantial number without prior indication of heart disease or comorbidities show significant injury to heart tissue, assessed by increased levels of troponin in blood. We studied the cell biological and physiological effects of virus infection of healthy human iPSC-derived cardiomyocytes in culture. Virus infection with interleukins disorganizes myofibrils, increases cell size and the numbers of multinucleated cells, and suppresses the expression of proteins of the contractile apparatus. Viral infection of cardiomyocytes in culture triggers release of troponin similar to elevation in levels of COVID-19 patients with heart disease. Viral infection in the presence of interleukins slows down and desynchronizes the beating of cardiomyocytes in culture. The cell-level physiological changes are similar to decreases in left ventricular ejection seen in imaging of patients' hearts. These observations suggest that direct injury to heart tissue by virus can be one underlying cause of heart disease in COVID-19.

High dimensionality reduction by matrix factorization for systems pharmacology

The extraction of predictive features from the complex high-dimensional multi-omic data is necessary for decoding and overcoming the therapeutic responses in systems pharmacology. Developing computational methods to reduce high-dimensional space of features in in vitro, in vivo and clinical data is essential to discover the evolution and mechanisms of the drug responses and drug resistance. In this paper, we have utilized the matrix factorization (MF) as a modality for high dimensionality reduction in systems pharmacology. In this respect, we have proposed three novel feature selection methods using the mathematical conception of a basis for features. We have applied these techniques as well as three other MF methods to analyze eight different gene expression datasets to investigate and compare their performance for feature selection. Our results show that these methods are capable of reducing the feature spaces and find predictive features in terms of phenotype determination. The three proposed techniques outperform the other methods used and can extract a 2-gene signature predictive of a tyrosine kinase inhibitor treatment response in the Cancer Cell Line Encyclopedia.

BIOM-23. A PILOT STUDY OF WHOLE GENOME SEQUENCING (WGS) OF PLASMA CELL-FREE DNA (cfDNA) FOR ULTRASENSITIVE DETECTION OF TUMOR DNA IN PATIENTS WITH GLIOBLASTOMA (GBM)

BACKGROUND

Plasma circulating tumor DNA (ctDNA) is rarely detectable by traditional methods in patients with GBM. As a result, unlike in lung and other cancers, serial next generation sequencing of ctDNA for monitoring GBM tumor burden has been challenging. In light of the low tumor fraction (TF) of DNA fragments in GBM patient plasma and the urgent need to improve upon MRI for tracking GBM tumor burden, we conducted a pilot study in patients with newly diagnosed GBM using the C2 intelligence platform (C2i Genomics), which leverages genome-wide mutational integration for highly sensitive ctDNA detection.

METHODS

Plasma was collected pre- and post-operatively in patients with newly diagnosed GBM undergoing surgical resection/biopsy. cfDNA was extracted, quantified, and analyzed for fragment size. Genomic DNA (gDNA) was extracted from matched tumor tissue. Whole genome sequencing (WGS) was performed on both gDNA and cfDNA. A specific copy number alteration (CNA) compendium was created for each patient to generate a readout of TF (Zviran, Nat Medicine 2020). We assessed the association between TF at post-operative day 1 (a surrogate for residual disease) and OS, adjusting for other prognostic factors using Cox regression.

RESULTS

37 patients were enrolled. For samples with high tumor fraction (n=5), a statistically significant (p< 1e-4) correlation between CNA profiles of tumor tissue and plasma samples was observed. Post-operative TF above the median value was associated with inferior OS (median 7.7 vs. 19.3 months, p=0.019). This association persisted after adjusting for age, O6-methylguanine-DNA methyltransferase methylation status, extent of resection, and performance status (adjusted HR 2.5, 95% CI 1.1-5.6, p=0.03).

CONCLUSION

Genome-wide mutational integration enables ultra-sensitive detection of ctDNA in GBM patient plasma. Post-operative TF measured by the C2i test is independently associated with OS in newly diagnosed GBM, providing the foundation to evaluate this technology for personalized prognostication and disease monitoring.

Pharmacological Functionalization of Protein-Based Nanorobots as a Novel Tool for Drug Delivery in Cancer

The delivery of hydrophobic therapeutic agents to tumors is a challenge in the treatment of cancers. Here, we review recent advances in coiled-coil protein origami and discuss a proposed programmable protein origami structure, switchable by a protein kinase A/phosphatase switch, as an example of functionalization for designing future protein nanorobots.

Decoding Clinical Biomarker Space of COVID-19: Exploring Matrix Factorization-based Feature Selection Methods

One of the most critical challenges in managing complex diseases like COVID-19 is to establish an intelligent triage system that can optimize the clinical decision-making at the time of a global pandemic. The clinical presentation and patients’ characteristics are usually utilized to identify those patients who need more critical care. However, the clinical evidence shows an unmet need to determine more accurate and optimal clinical biomarkers to triage patients under a condition like the COVID-19 crisis. Here we have presented a machine learning approach to find a group of clinical indicators from the blood tests of a set of COVID-19 patients that are predictive of poor prognosis and morbidity. Our approach consists of two interconnected schemes: Feature Selection and Prognosis Classification. The former is based on different Matrix Factorization (MF)-based methods, and the latter is performed using Random Forest algorithm. Our model reveals that Arterial Blood Gas (ABG) O 2 Saturation and C-Reactive Protein (CRP) are the most important clinical biomarkers determining the poor prognosis in these patients. Our approach paves the path of building quantitative and optimized clinical management systems for COVID-19 and similar diseases.

EGFR Aggregation in the Brain

Recent findings showed that preformed fibrils (PFFs) of misfolded proteins, including α-synuclein (α-syn) and amyloid-β (Aβ), activate EGFR in cell cultures and animal models of amyloid propagation. Comparing these supramolecular structures to normal EGFR ligands such as EGF and HB-EGF suggests that these PFFs might trigger the formation of high order clustering of EGFR that stimulates the aggregation of EGFR tyrosine kinase domain (EGFR-TKD) which is known to form fibrils. Subsequently, self-assembled fibril of EGFR-TKDs itself can serve as a seed to induce aggregation of monomeric forms of misfolded proteins in cytoplasm or endosomes. In this model, EGFR serves as an amyloidogenic receptor to facilitate (1) cellular uptake of exogenous PFFs and (2) seeding of endogenous misfolded proteins.

Genome-wide circulating tumor DNA monitoring for bladder cancer treatment management and organ preservation

e16527

Background: Bladder cancer (BC) is the 9th most commonly diagnosed cancer worldwide and each year responsible for 165,000 deaths. Neoadjuvant combination chemotherapy, followed by radical cystectomy, is used for the management of localized muscle-invasive bladder cancer. One of the critical challenges in this therapeutic regimen is monitoring the tumor load to assess therapeutic efficacy – this is typically performed by assessing pathological downstaging in the cystectomy specimen. A high frequency of patients presents with T0N0 at cystectomy (no indication of residual disease), and consequently, it is vital to investigate organ preservation approaches to identify those patients who may qualify for bladder preservation. For precision oncology, we need to develop quantitative and non-invasive diagnostic methodologies to help the oncologist tailor the treatments to individual patients and monitor them for further clinical decision-making. Methods: Cell-free DNA (cfDNA) mutation detection has shown significant promise in its ability to monitor minimal residual disease and disease relapse by detection of cancer mutations in the peripheral blood. However, the combination of low tumor fraction and limited input material obtained from a typical plasma sample restricts the probability of detecting low metastatic burden in cfDNA through current deep targeted sequencing methods. Results: Here we present results from applying whole-genome sequencing (WGS) of cfDNA. We integrate a genome-wide mutation and copy number monitoring approach coupled with advanced signal processing and Artificial Intelligence (AI) for measuring the tumor load from low-input blood samples (̃1mL of plasma) with ultra-sensitive detection. The increased sensitivity allowed clinical detection of tumor fraction down to 8*10-5 and recurrence detection sensitivity achieving > 65% at the first two months post-surgery. The WGS cfDNA approach is being evaluated on a patient cohort of more than 50 bladder cancer patients with longitudinal plasma sampling during neoadjuvant chemotherapy (response measure), pre-cystectomy (complete response measure), and post-surgery (relapse monitoring). Conclusions: The results indicate the clinical potential of genome-wide mutation integration as an ultra-sensitive, non-invasive diagnostic method for bladder cancer clinical management and bladder organ preservation.

Seeding Brain Protein Aggregation by SARS-CoV-2 as a Possible Long-Term Complication of COVID-19 Infection

Postinfection complications of coronavirus disease 2019 (COVID-19) are still unknown, and one of the long-term concerns in infected people are brain pathologies. The question is that severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection may be an environmental factor in accelerating the sporadic neurodegeneration in the infected population. In this regard, induction of protein aggregation in the brain by SARS-CoV-2 intact structure or a peptide derived from spike protein subunits needs to be considered in futures studies. In this paper, we discuss these possibilities using pieces of evidence from other viruses.

Physiology of cardiomyocyte injury in COVID-19

COVID-19 affects multiple organs. Clinical data from the Mount Sinai Health System shows that substantial numbers of COVID-19 patients without prior heart disease develop cardiac dysfunction. How COVID-19 patients develop cardiac disease is not known. We integrate cell biological and physiological analyses of human cardiomyocytes differentiated from human induced pluripotent stem cells (hiPSCs) infected with SARS-CoV-2 in the presence of interleukins, with clinical findings, to investigate plausible mechanisms of cardiac disease in COVID-19 patients. We infected hiPSC-derived cardiomyocytes, from healthy human subjects, with SARS-CoV-2 in the absence and presence of interleukins. We find that interleukin treatment and infection results in disorganization of myofibrils, extracellular release of troponin-I, and reduced and erratic beating. Although interleukins do not increase the extent, they increase the severity of viral infection of cardiomyocytes resulting in cessation of beating. Clinical data from hospitalized patients from the Mount Sinai Health system show that a significant portion of COVID-19 patients without prior history of heart disease, have elevated troponin and interleukin levels. A substantial subset of these patients showed reduced left ventricular function by echocardiography. Our laboratory observations, combined with the clinical data, indicate that direct effects on cardiomyocytes by interleukins and SARS-CoV-2 infection can underlie the heart disease in COVID-19 patients.

Heparin-binding Peptides as Novel Therapies to Stop SARS-CoV-2 Cellular Entry and Infection

Heparan sulfate proteoglycans (HSPGs) are cell surface receptors that are involved in the cellular uptake of pathologic amyloid proteins and viruses, including the novel coronavirus; severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Heparin and heparan sulfate antagonize the binding of these pathogens to HSPGs and stop their cellular internalization, but the anticoagulant effect of these agents has been limiting their use in the treatment of viral infections. Heparin-binding peptides (HBPs) are suitable nonanticoagulant agents that are capable of antagonizing binding of heparin-binding pathogens to HSPGs. Here, we review and discuss the use of HBPs as viral uptake inhibitors and will address their benefits and limitations to treat viral infections. Furthermore, we will discuss a variant of these peptides that is in the clinic and can be considered as a novel therapy in coronavirus disease 2019 (COVID-19) infection. SIGNIFICANCE STATEMENT: The need to discover treatment modalities for COVID-19 is a necessity, and therapeutic interventions such as heparin-binding peptides (HBPs), which are used for other cases, can be beneficial based on their mechanisms of actions. In this paper, we have discussed the application of HBPs as viral uptake inhibitors in COVID-19 and explained possible mechanisms of actions and the therapeutic effects.

Diagnosis of Pancreatic Cystic Lesions by Virtual Slicing: Comparison of Diagnostic Potential of Needle-Based Confocal Laser Endomicroscopy versus Endoscopic Ultrasound-Guided Fine-Needle Aspiration

Background:

Pancreatic cystic lesions are often challenging entities for diagnosis and management. EUS-FNA diagnostic accuracy is limited by paucicellularity of cytology specimens and sampling errors. Needle-based confocal laser endomicroscopy (nCLE) provides real-time imaging of the microscopic structure of the cystic lesion and could result in a more accurate diagnosis.

Aims and Objectives:

To determine the diagnostic utility of in vivo nCLE and EUS-FNA in the diagnosis and histologic characterization of pancreatic cystic lesions (PCL).

Materials and Methods:

All patients diagnosed with PCL who had undergone nCLE and FNA over a 10-year period within a major urban teaching hospital were included in this study. All gastroenterology reports of the nCLE images and corresponding pathologist findings from the EUS-FNA were collected and compared with, a final diagnosis prospectively collected from clinicopathological and imaging data.

Results:

A total of n=32 patients were included in this study, which consisted of n=13 serous cystadenoma (SCA), n=7 intraductal papillary mucinous neoplasms (IPMN), n=2 mucinous cystic neoplasms (MCN), n=3 well-differentiated neuroendocrine tumors, n=2 cysts, n=2 benign pancreatic lesions, n=1 adenocarcinoma, n=1 gastrointestinal stromal tumor (GIST) and n=1 lymphangioma. The overall diagnostic rate was higher in nCLE (87.5%) vs. EUS-FNA (71.9%) While the diagnostic accuracy of nCLE and EUS-FNA were comparable in characterization of benign vs. malignant lesions, the nCLE diagnosis demonstrated higher accuracy rate in identifying mucinous cystic neoplasms compared to EUS-FNA.

Conclusion:

nCLE is a useful companion diagnostic tool for pancreatic cystic lesions and could assist the cytopathologist to better triage the sample for required ancillary testing and treatment planning. The combination of nCLE and EUS-FNA may be especially helpful in reducing the proportion of cases categorized as non-diagnostic.

A Systems Biology Roadmap to Decode mTOR Control System in Cancer

Mechanistic target of rapamycin (mTOR) is a critical protein in the regulation of cell fate decision making, especially in cancer cells. mTOR acts as a signal integrator and is one of the main elements of interactions among the pivotal cellular processes such as cell death, autophagy, metabolic reprogramming, cell growth, and cell cycle. The temporal control of these processes is essential for the cellular homeostasis and dysregulation of mTOR signaling pathway results in different phenotypes, including aging, oncogenesis, cell survival, cell growth, senescence, quiescence, and cell death. In this paper, we have proposed a systems biology roadmap to study mTOR control system, which introduces the theoretical and experimental modalities to decode temporal and dynamical characteristics of mTOR signaling in cancer.

Dynamic modeling of signal transduction by mTOR complexes in cancer

Signal integration has a crucial role in the cell fate decision and dysregulation of the cellular signaling pathways is a primary characteristic of cancer. As a signal integrator, mTOR shows a complex dynamical behavior which determines the cell fate at different cellular processes levels, including cell cycle progression, cell survival, cell death, metabolic reprogramming, and aging. The dynamics of the complex responses to rapamycin in cancer cells have been attributed to its differential time-dependent inhibitory effects on mTORC1 and mTORC2, the two main complexes of mTOR. Two explanations were previously provided for this phenomenon: 1-Rapamycin does not inhibit mTORC2 directly, whereas it prevents mTORC2 formation by sequestering free mTOR protein (Le Chatelier's principle). 2-Components like Phosphatidic Acid (PA) further stabilize mTORC2 compared with mTORC1. To understand the mechanism by which rapamycin differentially inhibits the mTOR complexes in the cancer cells, we present a mathematical model of rapamycin mode of action based on the first explanation, i.e., Le Chatelier's principle. Translating the interactions among components of mTORC1 and mTORC2 into a mathematical model revealed the dynamics of rapamycin action in different doses and time-intervals of rapamycin treatment. This model shows that rapamycin has stronger effects on mTORC1 compared with mTORC2, simply due to its direct interaction with free mTOR and mTORC1, but not mTORC2, without the need to consider other components that might further stabilize mTORC2. Based on our results, even when mTORC2 is less stable compared with mTORC1, it can be less inhibited by rapamycin.

Genomic signatures defining responsiveness to allopurinol and combination therapy for lung cancer identified by systems therapeutics analyses

The ability to predict responsiveness to drugs in individual patients is limited. We hypothesized that integrating molecular information from databases would yield predictions that could be experimentally tested to develop transcriptomic signatures for specific drugs. We analyzed lung adenocarcinoma patient data from The Cancer Genome Atlas and identified a subset of patients in which xanthine dehydrogenase (XDH) expression correlated with decreased survival. We tested allopurinol, an FDA‐approved drug that inhibits XDH, on human non‐small‐cell lung cancer (NSCLC) cell lines obtained from the Broad Institute Cancer Cell Line Encyclopedia and identified sensitive and resistant cell lines. We utilized the transcriptomic profiles of these cell lines to identify six‐gene signatures for allopurinol‐sensitive and allopurinol‐resistant cell lines. Transcriptomic networks identified JAK2 as an additional target in allopurinol‐resistant lines. Treatment of resistant cell lines with allopurinol and CEP‐33779 (a JAK2 inhibitor) resulted in cell death. The effectiveness of allopurinol alone or allopurinol and CEP‐33779 was verified in vivo using tumor formation in NCR‐nude mice. We utilized the six‐gene signatures to predict five additional allopurinol‐sensitive NSCLC cell lines and four allopurinol‐resistant cell lines susceptible to combination therapy. We searched the transcriptomic data from a library of patient‐derived NSCLC tumors from the Jackson Laboratory to identify tumors that would be predicted to be sensitive to allopurinol or allopurinol + CEP‐33779 treatment. Patient‐derived tumors showed the predicted drug sensitivity in vivo. These data indicate that we can use integrated molecular information from cancer databases to predict drug responsiveness in individual patients and thus enable precision medicine.

Abstract 669: Dynamic modeling of responses to PDL-1 inhibitors in non-small cell lung cancer: implications for precision combination therapy

Immunotherapy by immune checkpoint blockade has had a promising impact on the treatment of Non-Small Cell Lung Cancer (NSCLC). Although single and combination therapy with checkpoint inhibitors such as PDL-1 inhibitors have been proved to be effective in making the survival of patients with NSCLC longer, but there is always a population of patients which shows resistance to these therapeutic modalities. The source of resistance emerges from signaling pathways and regulatory systems in tumors and the immune system, but the dynamics among tumor components, T cells, and tumor microenvironment contribute in this process as well. To understand the complex system of these interactions, a mathematical model of these pathways and interactions are presented using Ordinary Differential Equations (ODEs). The specific characteristics of NSCLC tumors including metabolic reprogramming have been considered in this model. The model is capable of decoding patient-to-patient variabilities regarding responses to PDL-1 inhibitors, and parameter-sensitivity analysis shows the vulnerable biomarkers to be targeted by combination therapy to give better results in killing the tumor cells. The model can be integrated with PK/PD models to provide an in-silico platform for investigating the therapeutic responses in NSCLC for precision oncology.

Note: This abstract was not presented at the meeting.

Citation Format: Iman Tavassoly. Dynamic modeling of responses to PDL-1 inhibitors in non-small cell lung cancer: implications for precision combination therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 669.

Abstract 4276: Analysis of sensitivity of genomic signatures of therapeutic responses of non-small cell lung cancer in patient-derived xenograft models

Genomic alterations are important characteristics of therapeutic responses and therapeutic-resistance phenotype in cancer. These genomic alterations are complex and are determinants of aberrations in signaling pathways and their dynamics. A genomic signature includes a set of genomic alterations like gene expressions, mutations, copy number variations, etc. These genomic signatures are used as biomarkers for assigning treatment to individuals benefiting from a specific treatment plan and are validated using preclinical models such as patient-derived xenograft models (PDX models). We have proposed a methodology to optimize these genomic signatures in silico and evaluate their accuracy in vivo using PDX models of non-small cell lung cancer (NSCLC). Our integrative analysis has shown that different sizes and complexities of gene sets provide different layers of information applicable in designing therapeutic regimens. To eliminate sloppy signatures, it is possible to find optimized and sensitive genomic signatures for precision oncology.

Citation Format: Iman Tavassoly, Ravi Iyengar. Analysis of sensitivity of genomic signatures of therapeutic responses of non-small cell lung cancer in patient-derived xenograft models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4276.

Abstract 5554: Metabolic reprogramming in non-small cell lung cancer: a precision oncology approach

Lung cancers are among the most common invasive cancers worldwide and annually lead to high mortality and morbidity. Genomic alterations have been known to control the evolution of hallmarks of cancer in a dynamic way. These molecular alterations combined with epigenomic and post-genomic modifications contribute to formation of these neoplasms. Multiplicity of these changes has made development of personalized therapeutic regimens for these cancers a complex problem. Metabolic reprogramming is one of the main mechanisms in progression of cancers. There have been efforts to model the metabolic reprogramming in cancer using metabolic networks of cancer cells, but there has been no computational framework to model these metabolic transitions in cancer for precision and personalized medicine. We have combined computational, mathematical and experimental methodologies to develop a platform for precision oncology in non-small cell lung cancer (NSCLC) by in silico models of metabolic switches. Our integrative analysis of genomic data from NSCLC has led to discovery of genomic signatures controlling metabolic reprogramming in NSCLC with KRAS mutations. This discovery was proved in vivo and in vitro using drugs blocking different metabolic pathways. We have shown that NSCLC cells and tumors which carry KRAS mutations and have these genomic signatures are addicted to the pentose phosphate pathway (PPP). We have verified and proved the predictive value of these genomic signatures using Patient Derived Xenograft (PDX) tumor models of NSCLC. We are developing a mathematical and computational framework to model these metabolic switches. Our platform is capable of using genomic data from a cell line or tumor to determine the metabolic dependency of them quantitatively and predict the optimized personalized treatments for modulating metabolic pathways aiming to control cancer progression.

Citation Format: Iman Tavassoly, Ravi Iyengar. Metabolic reprogramming in non-small cell lung cancer: a precision oncology approach [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 5554. doi:10.1158/1538-7445.AM2017-5554

Dynamic Modeling of the Interaction Between Autophagy and Apoptosis in Mammalian Cells

Autophagy is a conserved biological stress response in mammalian cells that is responsible for clearing damaged proteins and organelles from the cytoplasm and recycling their contents via the lysosomal pathway. In cases of mild stress, autophagy acts as a survival mechanism, while in cases of severe stress cells may switch to programmed cell death. Understanding the decision process that moves a cell from autophagy to apoptosis is important since abnormal regulation of autophagy occurs in many diseases, including cancer. To integrate existing knowledge about this decision process into a rigorous, analytical framework, we built a mathematical model of cell fate decisions mediated by autophagy. Our dynamical model is consistent with existing quantitative measurements of autophagy and apoptosis in rat kidney proximal tubular cells responding to cisplatin-induced stress.

Modelling the effect of GRP78 on anti-oestrogen sensitivity and resistance in breast cancer

Understanding the origins of resistance to anti-oestrogen drugs is of critical importance to many breast cancer patients. Recent experiments show that knockdown of GRP78, a key gene in the unfolded protein response (UPR), can re-sensitize resistant cells to anti-oestrogens, and overexpression of GRP78 in sensitive cells can cause them to become resistant. These results appear to arise from the operation and interaction of three cellular systems: the UPR, autophagy and apoptosis. To determine whether our current mechanistic understanding of these systems is sufficient to explain the experimental results, we built a mathematical model of the three systems and their interactions. We show that the model is capable of reproducing previously published experimental results and some new data gathered specifically for this paper. The model provides us with a tool to better understand the interactions that bring about anti-oestrogen resistance and the effects of GRP78 on both sensitive and resistant breast cancer cells.

Endoplasmic reticulum stress, the unfolded protein response, autophagy, and the integrated regulation of breast cancer cell fate

How breast cancer cells respond to the stress of endocrine therapies determines whether they will acquire a resistant phenotype or execute a cell-death pathway. After a survival signal is successfully executed, a cell must decide whether it should replicate. How these cell-fate decisions are regulated is unclear, but evidence suggests that the signals that determine these outcomes are highly integrated. Central to the final cell-fate decision is signaling from the unfolded protein response, which can be activated following the sensing of stress within the endoplasmic reticulum. The duration of the response to stress is partly mediated by the duration of inositol-requiring enzyme-1 activation following its release from heat shock protein A5. The resulting signals appear to use several B-cell lymphoma-2 family members to both suppress apoptosis and activate autophagy. Changes in metabolism induced by cellular stress are key components of this regulatory system, and further adaptation of the metabolome is affected in response to stress. Here we describe the unfolded protein response, autophagy, and apoptosis, and how the regulation of these processes is integrated. Central topologic features of the signaling network that integrate cell-fate regulation and decision execution are discussed.

Dynamic modelling of oestrogen signalling and cell fate in breast cancer cells

Cancers of the breast and other tissues arise from aberrant decision-making by cells regarding their survival or death, proliferation or quiescence, damage repair or bypass. These decisions are made by molecular signalling networks that process information from outside and from within the breast cancer cell and initiate responses that determine the cell's survival and reproduction. Because the molecular logic of these circuits is difficult to comprehend by intuitive reasoning alone, we present some preliminary mathematical models of the basic decision circuits in breast cancer cells that may aid our understanding of their susceptibility or resistance to endocrine therapy.

Endoplasmic reticulum stress, the unfolded protein response, and gene network modeling in antiestrogen resistant breast cancer

Lack of understanding of endocrine resistance remains one of the major challenges for breast cancer researchers, clinicians, and patients. Current reductionist approaches to understanding the molecular signaling driving resistance have offered mostly incremental progress over the past 10 years. As the field of systems biology has begun to mature, the approaches and network modeling tools being developed and applied therein offer a different way to think about how molecular signaling and the regulation of critical cellular functions are integrated. To gain novel insights, we first describe some of the key challenges facing network modeling of endocrine resistance, many of which arise from the properties of the data spaces being studied. We then use activation of the unfolded protein response (UPR) following induction of endoplasmic reticulum stress in breast cancer cells by antiestrogens, to illustrate our approaches to computational modeling. Activation of UPR is a key determinant of cell fate decision making and regulation of autophagy and apoptosis. These initial studies provide insight into a small subnetwork topology obtained using differential dependency network analysis and focused on the UPR gene XBP1. The XBP1 subnetwork topology incorporates BCAR3, BCL2, BIK, NFκB, and other genes as nodes; the connecting edges represent the dependency structures amongst these nodes. As data from ongoing cellular and molecular studies become available, we will build detailed mathematical models of this XBP1-UPR network.