Bio


Helio Costa, PhD, is a medical geneticist specializing in oncology, genomics, computational biology, data science, artificial intelligence, and product development. His work focuses on developing new cancer therapies and medical software to improve cancer patient care. He currently serves as Senior Medical Director and Head of Molecular Therapeutics & Drug Development at Natera and is an Adjunct Clinical Assistant Professor in the Department of Pathology at Stanford Medical School.

Dr. Costa’s research spans the entire drug development pipeline, from discovery to FDA-approved therapies. He leads efforts in developing innovative cancer treatments by integrating advanced genomics, computational biology, and AI-driven methodologies. His team focuses on the creation of targeted molecular therapeutics, using large-scale genetic data to identify novel treatment pathways and predict patient responses. In addition to drug discovery, Dr. Costa oversees the clinical trials necessary for validating these therapies, ensuring they meet rigorous standards for safety and efficacy. Furthermore, he manages the product development and commercialization process, guiding new treatments from initial research through to market-ready, FDA-approved cancer therapies.

In addition to his therapeutic work, Dr. Costa has led the development and implementation of genetic diagnostic tests and clinical algorithms used to support therapeutic decisions at Stanford Health Care. His contributions include the creation of DNA and RNA cancer diagnostic tests, as well as algorithms that analyze large-scale laboratory datasets and electronic medical records to predict patient outcomes. As a co-investigator with the NIH Clinical Genome Resource (ClinGen) Consortium, Dr. Costa oversaw developing FDA-recognized medical software used by healthcare providers, researchers, and biotech companies to assess the clinical relevance of genes and mutations.

He is the founding director of the Stanford Clinical Data Science Fellowship, where post-doctoral medical fellows engage in interdisciplinary clinical research and implement real-world health data solutions within Stanford Health Care. Dr. Costa is also an Attending Medical Geneticist for the Molecular Genetic Pathology Laboratory at Stanford Health Care, where he previously served as Assistant Lab Director.

Dr. Costa holds a BS in Genetics from the University of California, Davis, a PhD in Genetics from Stanford University School of Medicine, and completed his ABMGG Clinical Molecular Genetics and Genomics fellowship training at Stanford University School of Medicine.

Clinical Focus


  • Molecular Oncology
  • Molecular Pathology
  • Medical Genetics
  • Clinical Pathology

Administrative Appointments


  • Attending Medical Geneticist, Molecular Genetic Pathology Laboratory (2017 - Present)
  • Founding Director, Stanford Clinical Data Science Fellowship (2018 - 2021)
  • Assistant Lab Director, Molecular Genetic Pathology Laboratory (2017 - 2021)

Professional Education


  • Board Certification: American Society for Clinical Pathology, Technologist in Molecular Biology (2018)
  • Fellowship, Stanford University School of Medicine, ABMGG Clinical Molecular Genetics and Genomics (2017)
  • Doctor of Philosophy, Stanford University School of Medicine, Genetics (2015)
  • Bachelor of Science, University of California, Davis, Genetics (2010)

Graduate and Fellowship Programs


  • Biomedical Data Science (Phd Program)
  • Biomedical Data Science (Masters Program)

All Publications


  • Increasing Clinical Trial Accrual via Automated Matching of Biomarker Criteria. Pacific Symposium on Biocomputing. Pacific Symposium on Biocomputing Chen, J. W., Kunder, C. A., Bui, N. n., Zehnder, J. L., Costa, H. A., Stehr, H. n. 2020; 25: 31–42

    Abstract

    Successful implementation of precision oncology requires both the deployment of nucleic acid sequencing panels to identify clinically actionable biomarkers, and the efficient screening of patient biomarker eligibility to on-going clinical trials and therapies. This process is typically performed manually by biocurators, geneticists, pathologists, and oncologists; however, this is a time-intensive, and inconsistent process amongst healthcare providers. We present the development of a feature matching algorithmic pipeline that identifies patients who meet eligibility criteria of precision medicine clinical trials via genetic biomarkers and apply it to patients undergoing treatment at the Stanford Cancer Center. This study demonstrates, through our patient eligibility screening algorithm that leverages clinical sequencing derived biomarkers with precision medicine clinical trials, the successful use of an automated algorithmic pipeline as a feasible, accurate and effective alternative to the traditional manual clinical trial curation.

    View details for PubMedID 31797584

  • LitGen: Genetic Literature Recommendation Guided by Human Explanations. Pacific Symposium on Biocomputing. Pacific Symposium on Biocomputing Nie, A. n., Pineda, A. L., Wright, M. W., Wand, H. n., Wulf, B. n., Costa, H. A., Patel, R. Y., Bustamante, C. D., Zou, J. n. 2020; 25: 67–78

    Abstract

    As genetic sequencing costs decrease, the lack of clinical interpretation of variants has become the bottleneck in using genetics data. A major rate limiting step in clinical interpretation is the manual curation of evidence in the genetic literature by highly trained biocurators. What makes curation particularly time-consuming is that the curator needs to identify papers that study variant pathogenicity using different types of approaches and evidences-e.g. biochemical assays or case control analysis. In collaboration with the Clinical Genomic Resource (ClinGen)-the flagship NIH program for clinical curation-we propose the first machine learning system, LitGen, that can retrieve papers for a particular variant and filter them by specific evidence types used by curators to assess for pathogenicity. LitGen uses semi-supervised deep learning to predict the type of evi+dence provided by each paper. It is trained on papers annotated by ClinGen curators and systematically evaluated on new test data collected by ClinGen. LitGen further leverages rich human explanations and unlabeled data to gain 7.9%-12.6% relative performance improvement over models learned only on the annotated papers. It is a useful framework to improve clinical variant curation.

    View details for PubMedID 31797587

  • Genomic Evidence for Local Adaptation of Hunter-Gatherers to the African Rainforest. Current biology : CB Lopez, M., Choin, J., Sikora, M., Siddle, K., Harmant, C., Costa, H. A., Silvert, M., Mouguiama-Daouda, P., Hombert, J., Froment, A., Le Bomin, S., Perry, G. H., Barreiro, L. B., Bustamante, C. D., Verdu, P., Patin, E., Quintana-Murci, L. 2019

    Abstract

    African rainforests support exceptionally high biodiversity and host the world's largest number of active hunter-gatherers [1-3]. The genetic history of African rainforest hunter-gatherers and neighboring farmers is characterized by an ancient divergence more than 100,000 years ago, together with recent population collapses and expansions, respectively [4-12]. While the demographic past of rainforest hunter-gatherers has been deeply characterized, important aspects of their history of genetic adaptation remain unclear. Here, we investigated how these groups have adapted-through classic selective sweeps, polygenic adaptation, and selection since admixture-to the challenging rainforest environments. To do so, we analyzed a combined dataset of 566 high-coverage exomes, including 266 newly generated exomes, from 14 populations of rainforest hunter-gatherers and farmers, together with 40 newly generated, low-coverage genomes. We find evidence for a strong, shared selective sweep among all hunter-gatherer groups in the regulatory region of TRPS1-primarily involved in morphological traits. We detect strong signals of polygenic adaptation for height and life history traits such as reproductive age; however, the latter appear to result from pervasive pleiotropy of height-associated genes. Furthermore, polygenic adaptation signals for functions related to responses of mast cells to allergens andmicrobes, the IL-2 signaling pathway, and hostinteractions with viruses support a history of pathogen-driven selection in the rainforest. Finally, we find that genes involved in heart and bone development and immune responses are enriched in both selection signals and local hunter-gatherer ancestry in admixed populations, suggesting that selection has maintained adaptive variation in the face of recent gene flow from farmers.

    View details for DOI 10.1016/j.cub.2019.07.013

    View details for PubMedID 31402299

  • Structural Variation Detection by Proximity Ligation from Formalin-Fixed, Paraffin-Embedded Tumor Tissue. The Journal of molecular diagnostics : JMD Troll, C. J., Putnam, N. H., Hartley, P. D., Rice, B., Blanchette, M., Siddiqui, S., Ganbat, J., Powers, M. P., Ramakrishnan, R., Kunder, C. A., Bustamante, C. D., Zehnder, J. L., Green, R. E., Costa, H. A. 2018

    Abstract

    The clinical management and therapy of many solid tumor malignancies is dependent on detection of medically actionable or diagnostically relevant genetic variation. However, a principal challenge for genetic assays from tumors is the fragmented and chemically damaged state of DNA in formalin-fixed, paraffin-embedded (FFPE) samples. From highly fragmented DNA and RNA there is no current technology for generating long-range DNA sequence data as is required to detect genomic structural variation or long-range genotype phasing. We have developed a high-throughput chromosome conformation capture approach for FFPE samples that we call "Fix-C", which is similar in concept to Hi-C. Fix-C enables structural variation detection from archival FFPE samples. This method was applied to 15 clinical adenocarcinoma and sarcoma positive control specimens spanning a broad range of tumor purities. In this panel, Fix-C analysis achieves a 90% concordance rate with FISH assays - the current clinical gold standard. Additionally, novel structural variation undetected by other methods could be identified and long-range chromatin configuration information recovered from these FFPE samples harboring highly degraded DNA. This powerful approach will enable detailed resolution of global genome rearrangement events during cancer progression from FFPE material and inform the development of targeted molecular diagnostic assays for patient care.

    View details for PubMedID 30605765

  • Gene-specific criteria for PTEN variant curation: Recommendations from the ClinGen PTEN Expert Panel. Human mutation Mester, J. L., Ghosh, R., Pesaran, T., Huether, R., Karam, R., Hruska, K. S., Costa, H. A., Lachlan, K., Ngeow, J., Barnholtz-Sloan, J., Sesock, K., Hernandez, F., Zhang, L., Milko, L., Plon, S. E., Hegde, M., Eng, C. 2018; 39 (11): 1581–92

    Abstract

    The ClinGen PTEN Expert Panel was organized by the ClinGen Hereditary Cancer Clinical Domain Working Group to assemble clinicians, researchers, and molecular diagnosticians with PTEN expertise to develop specifications to the 2015 ACMG/AMP Sequence Variant Interpretation Guidelines for PTEN variant interpretation. We describe finalized PTEN-specific variant classification criteria and outcomes from pilot testing of 42 variants with benign/likely benign (BEN/LBEN), pathogenic/likely pathogenic (PATH/LPATH), uncertain significance (VUS), and conflicting (CONF) ClinVar assertions. Utilizing these rules, classifications concordant with ClinVar assertions were achieved for 14/15 (93.3%) BEN/LBEN and 16/16 (100%) PATH/LPATH ClinVar consensus variants for an overall concordance of 96.8% (30/31). The variant where agreement was not reached was a synonymous variant near a splice donor with noncanonical sequence for which in silico models cannot predict the native site. Applying these rules to six VUS and five CONF variants, adding shared internal laboratory data enabled one VUS to be classified as LBEN and two CONF variants to be as classified as PATH and LPATH. This study highlights the benefit of gene-specific criteria and the value of sharing internal laboratory data for variant interpretation. Our PTEN-specific criteria and expertly reviewed assertions should prove helpful for laboratories and others curating PTEN variants.

    View details for PubMedID 30311380

  • Tumor Molecular Profiling Aids in Determining Tissue of Origin and Therapy for Metastatic Adenocarcinoma in a Patient With Multiple Primary Malignancies JCO PRECISION ONCOLOGY Costa, H. A., Reyes, R., Mills, M., Zehnder, J. L., Sledge, G., Curtis, C., Ford, J. M., Suarez, C. J. 2018; 2
  • Detection and surveillance of bladder cancer using urine tumor DNA. Cancer discovery Dudley, J. C., Schroers-Martin, J. n., Lazzareschi, D. V., Shi, W. Y., Chen, S. B., Esfahani, M. S., Trivedi, D. n., Chabon, J. J., Chaudhuri, A. A., Stehr, H. n., Liu, C. L., Lim, H. n., Costa, H. A., Nabet, B. Y., Sin, M. L., Liao, J. C., Alizadeh, A. A., Diehn, M. n. 2018

    Abstract

    Current regimens for the detection and surveillance of bladder cancer (BLCA) are invasive and have suboptimal sensitivity. Here, we present a novel high-throughput sequencing (HTS) method for detection of urine tumor DNA (utDNA) called utDNA CAPP-Seq (uCAPP-Seq) and apply it to 67 healthy adults and 118 patients with early-stage BLCA who either had urine collected prior to treatment or during surveillance. Using this targeted sequencing approach, we detected a median of 6 mutations per BLCA patient and observed surprisingly frequent mutations of the PLEKHS1 promoter (46%), suggesting these mutations represent a useful biomarker for detection of BLCA. We detected utDNA pre-treatment in 93% of cases using a tumor mutation-informed approach and in 84% when blinded to tumor mutation status, with 96-100% specificity. In the surveillance setting, we detected utDNA in 91% of patients who ultimately recurred, with utDNA detection preceding clinical progression in 92% of cases. uCAPP-Seq outperformed a commonly used ancillary test (UroVysion, p=0.02) and cytology and cystoscopy combined (p is less than or equal to 0.006), detecting 100% of BLCA cases detected by cytology and 82% that cytology missed. Our results indicate that uCAPP-Seq is a promising approach for early detection and surveillance of BLCA.

    View details for PubMedID 30578357

  • Promoting appropriate genetic testing: the impact of a combined test review and consultative service. Genetics in medicine Suarez, C. J., Yu, L., Downs, N., Costa, H. A., Stevenson, D. A. 2017

    Abstract

    Genetic test misorders can adversely affect patient care. However, little is known about the types of misorders and the overall impact of a utilization management (UM) program on curbing misorders. This study aimed to identify different types of misorders and analyze the impact of a combined test review and consultative service on reducing misorders over time.Selected genetic tests were systematically reviewed between January and December 2015 at Stanford Health Care. Misorders were categorized into five types: clerical errors, redundant testing, better alternatives, controversial, and uncategorized. Moreover, consultations were offered to help clinicians with test selection.Of the 629 molecular test orders reviewed, 13% were classified as misorders, and 7% were modified or canceled. Controversial misorders constitute the most common type (42%); however, unlike the other misorder types, they were negligibly affected by test review. Simultaneously, 71 consults were received. With the introduction of the UM program, genetic test misorders went from 22% at baseline to 3% at the end of the year.Our results show that the combined approach of test review and consultative service effectively reduced misorders over time and suggest that a UM program focused on eliminating misorders can positively influence health-care providers' behaviors.Genet Med advance online publication 26 January 2017Genetics in Medicine (2017); doi:10.1038/gim.2016.219.

    View details for DOI 10.1038/gim.2016.219

    View details for PubMedID 28125079

  • Identification of a Novel Somatic Mutation Leading to Allele Dropout for EGFR L858R Genotyping in Non-Small Cell Lung Cancer Molecular Diagnosis & Therapy Costa, H. A., Neal, J. W., Bustamante, C. D., Zehnder, J. L. 2017
  • Discovery and functional characterization of a neomorphic PTEN mutation PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA Costa, H. A., Leitner, M. G., Sos, M. L., Mavrantoni, A., Rychkova, A., Johnson, J. R., Newton, B. W., Yee, M., De La Vega, F. M., Ford, J. M., Krogan, N. J., Shokat, K. M., Oliver, D., Halaszovich, C. R., Bustamante, C. D. 2015; 112 (45): 13976-13981

    Abstract

    Although a variety of genetic alterations have been found across cancer types, the identification and functional characterization of candidate driver genetic lesions in an individual patient and their translation into clinically actionable strategies remain major hurdles. Here, we use whole genome sequencing of a prostate cancer tumor, computational analyses, and experimental validation to identify and predict novel oncogenic activity arising from a point mutation in the phosphatase and tensin homolog (PTEN) tumor suppressor protein. We demonstrate that this mutation (p.A126G) produces an enzymatic gain-of-function in PTEN, shifting its function from a phosphoinositide (PI) 3-phosphatase to a phosphoinositide (PI) 5-phosphatase. Using cellular assays, we demonstrate that this gain-of-function activity shifts cellular phosphoinositide levels, hyperactivates the PI3K/Akt cell proliferation pathway, and exhibits increased cell migration beyond canonical PTEN loss-of-function mutants. These findings suggest that mutationally modified PTEN can actively contribute to well-defined hallmarks of cancer. Lastly, we demonstrate that these effects can be substantially mitigated through chemical PI3K inhibitors. These results demonstrate a new dysfunction paradigm for PTEN cancer biology and suggest a potential framework for the translation of genomic data into actionable clinical strategies for targeted patient therapy.

    View details for DOI 10.1073/pnas.1422504112

    View details for PubMedID 26504226

  • Transcriptome sequencing from diverse human populations reveals differentiated regulatory architecture. PLoS genetics Martin, A. R., Costa, H. A., Lappalainen, T., Henn, B. M., Kidd, J. M., Yee, M., Grubert, F., Cann, H. M., Snyder, M., Montgomery, S. B., Bustamante, C. D. 2014; 10 (8)

    Abstract

    Large-scale sequencing efforts have documented extensive genetic variation within the human genome. However, our understanding of the origins, global distribution, and functional consequences of this variation is far from complete. While regulatory variation influencing gene expression has been studied within a handful of populations, the breadth of transcriptome differences across diverse human populations has not been systematically analyzed. To better understand the spectrum of gene expression variation, alternative splicing, and the population genetics of regulatory variation in humans, we have sequenced the genomes, exomes, and transcriptomes of EBV transformed lymphoblastoid cell lines derived from 45 individuals in the Human Genome Diversity Panel (HGDP). The populations sampled span the geographic breadth of human migration history and include Namibian San, Mbuti Pygmies of the Democratic Republic of Congo, Algerian Mozabites, Pathan of Pakistan, Cambodians of East Asia, Yakut of Siberia, and Mayans of Mexico. We discover that approximately 25.0% of the variation in gene expression found amongst individuals can be attributed to population differences. However, we find few genes that are systematically differentially expressed among populations. Of this population-specific variation, 75.5% is due to expression rather than splicing variability, and we find few genes with strong evidence for differential splicing across populations. Allelic expression analyses indicate that previously mapped common regulatory variants identified in eight populations from the International Haplotype Map Phase 3 project have similar effects in our seven sampled HGDP populations, suggesting that the cellular effects of common variants are shared across diverse populations. Together, these results provide a resource for studies analyzing functional differences across populations by estimating the degree of shared gene expression, alternative splicing, and regulatory genetics across populations from the broadest points of human migration history yet sampled.

    View details for DOI 10.1371/journal.pgen.1004549

    View details for PubMedID 25121757

  • High-throughput Sequencing of Subcutaneous Panniculitis-like T-Cell Lymphoma Reveals Candidate Pathogenic Mutations. Applied immunohistochemistry & molecular morphology : AIMM Fernandez-Pol, S. n., Costa, H. A., Steiner, D. F., Ma, L. n., Merker, J. D., Kim, Y. H., Arber, D. A., Kim, J. n. ; 27 (10): 740–48

    Abstract

    Subcutaneous panniculitis-like T-cell lymphoma (SPTCL) is a malignant primary cutaneous T-cell lymphoma that is challenging to distinguish from other neoplastic and reactive panniculitides. In an attempt to identify somatic variants in SPTCL that may be diagnostically or therapeutically relevant, we performed both exome sequencing on paired tumor-normal samples and targeted sequencing of hematolymphoid-malignancy-associated genes on tumor biopsies. Exome sequencing was performed on skin biopsies from 4 cases of skin-limited SPTCL, 1 case of peripheral T-cell lymphoma, not otherwise specified with secondary involvement of the panniculus, and 2 cases of lupus panniculitis. This approach detected between 1 and 13 high-confidence somatic variants that were predicted to result in a protein alteration per case. Variants of interest identified include 1 missense mutation in ARID1B in 1 case of SPTCL. To detect variants that were present at a lower level, we used a more sensitive targeted panel to sequence 41 hematolymphoid-malignancy-associated genes. The targeted panel was applied to 2 of the biopsies that were evaluated by whole exome sequencing as well as 5 additional biopsies. Potentially pathogenic variants were identified in KMT2D and PLCG1 among others, but no gene was altered in >2 of the 7 cases sequenced. One variant that was notably absent from the cases sequences is RHOA G17V. Further work will be required to further elucidate the genetic abnormalities that lead to this rare lymphoma.

    View details for DOI 10.1097/PAI.0000000000000683

    View details for PubMedID 31702703