Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Correspondence
  • Published:

Data mining differential clinical outcomes associated with drug regimens using adverse event reporting data

Nature Biotechnology volume 34, pages 697–700 (2016)Cite this article

Subjects

To the Editor:

The US Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) contains millions of clinical records that detail drug exposure(s), disease indications and clinical outcomes. However, a variety of challenges limit the ability of these data to be exploited for the extraction of clinically meaningful information beyond single-agent safety signal associations. Here, we report an open access web data mining tool, AERSMine, that enables data exploration and hypothesis generation among alternatively medicated cohorts to discover patterns of differential outcomes. Cohort groups can be formed as a function of demographics, underlying disorders, drug classes and other user-driven constraints across the entire range of human diseases and approved drugs.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Figure 1: Schematic overview of AERSMine, a multi-cohort data mining platform to analyze millions of reports from the FAERS and identify subgroup-specific differential responses to therapeutics.
Figure 2: Illustration of AERSMine in hypothesis generation.

References

  1. Qureshi, Z.P., Seoane-Vazquez, E., Rodriguez-Monguio, R., Stevenson, K.B. & Szeinbach, S.L. Pharmacoepidemiol. Drug Saf. 20, 772–777 (2011).

    Article  Google Scholar 

  2. Sakaeda, T., Tamon, A., Kadoyama, K. & Okuno, Y. Int. J. Med. Sci. 10, 796–803 (2013).

    Article  Google Scholar 

  3. Sakaeda, T., Kadoyama, K., Minami, K. & Okuno, Y. Int. J. Med. Sci. 11, 461–465 (2014).

    Article  CAS  Google Scholar 

  4. Tatonetti, N.P., Ye, P.P., Daneshjou, R. & Altman, R.B. Sci. Transl. Med. 4, 125ra31 (2012).

    Article  Google Scholar 

  5. Szarfman, A., Machado, S.G. & O'Neill, R.T. Drug Saf. 25, 381–392 (2002).

    Article  CAS  Google Scholar 

  6. Bennett, C.L. et al. J. Gen. Intern. Med. 27, 1697–1703 (2012).

    Article  Google Scholar 

  7. Norén, G.N., Sundberg, R., Bate, A. & Edwards, I.R. Stat. Med. 27, 3057–3070 (2008).

    Article  Google Scholar 

  8. Norén, G.N., Hopstadius, J. & Bate, A. Stat. Methods Med. Res. 10.1177/0962280211403604 (2011).

  9. Bate, A. et al. Eur. J. Clin. Pharmacol. 54, 315–321 (1998).

    Article  CAS  Google Scholar 

  10. Böhm, R., Höcker, J., Cascorbi, I. & Herdegen, T. Nat. Biotechnol. 30, 137–138 (2012).

    Article  Google Scholar 

  11. Grigoriev, I., zu Castell, W., Tsvetkov, P. & Antonov, A.V. Pharmacoepidemiol. Drug Saf. 23, 795–801 (2014).

    PubMed  Google Scholar 

  12. Ducat, L., Philipson, L.H. & Anderson, B.J. J. Am. Med. Assoc. 312, 691–692 (2014).

    Article  CAS  Google Scholar 

  13. Butwicka, A., Frisén, L., Almqvist, C., Zethelius, B. & Lichtenstein, P. Diabetes Care 38, 453–459 (2015).

    Article  Google Scholar 

  14. Seaquist, E.R. Diabetes 59, 4–5 (2010).

    Article  CAS  Google Scholar 

  15. Sarangdhar, M., Kushwaha, A., Dahlquist, J., Jegga, A. & Aronow, B. AMIA Summits Transl. Sci. Proc. 2013, 151–155 (2013).

    PubMed  Google Scholar 

  16. Birbara, C. A. Internet J. Rheumatol. 5 http://www.ispub.com/journal/the-internet-journal-of-rheumatology/volume-5-number-2/management-of-inadequate-response-to-tnf-antagonist-therapy-in-rheumatoid-arthritis-what-are-the-options.html (2008).

  17. Moodley, I. Cardiovasc. J. Afr. 19, 102–107 (2008).

    CAS  PubMed  PubMed Central  Google Scholar 

  18. Schwaiblmair, M. et al. Open Respir. Med. J. 6, 63–74 (2012).

    Article  Google Scholar 

  19. O'Connor, N., Dargan, P.I. & Jones, A.L. QJM 96, 787–791 (2003).

    Article  CAS  Google Scholar 

  20. Zhao, S. et al. Sci. Transl. Med. 5, 206ra140 (2013).

    Article  Google Scholar 

  21. Napoli, A.A. et al. J. Int. AIDS Soc. 17, 19214 (2014).

    Article  Google Scholar 

  22. Hauben, M. & Hung, E.Y. Int. J. Med. Sci. 10, 965–973 (2013).

    Article  Google Scholar 

  23. Harpaz, R. et al. Clin. Pharmacol. Ther. 93, 539–546 (2013).

    Article  CAS  Google Scholar 

  24. Low, Y.S., Gallego, B. & Shah, N.H. J. Comp. Eff. Res. 10.2217/cer.15.53 (2015).

  25. Hoffman, K.B., Demakas, A.R., Dimbil, M., Tatonetti, N.P. & Erdman, C.B. Drug Saf. 37, 971–980 (2014).

    Article  Google Scholar 

  26. Hoffman, K.B., Dimbil, M., Erdman, C.B., Tatonetti, N.P. & Overstreet, B.M. Drug Saf. 37, 283–294 (2014).

    Article  CAS  Google Scholar 

  27. Tamura, T., Sakaeda, T., Kadoyama, K. & Okuno, Y. Int. J. Med. Sci. 9, 441–446 (2012).

    Article  CAS  Google Scholar 

  28. Sarangdhar, M., Yacyshyn, M.B., Aronow, B.J. & Yacyshyn, B. Gastroenterology 146, S–795 (2014).

    Article  Google Scholar 

  29. Knox, C. et al. Nucleic Acids Res. 39, D1035–D1041 (2011).

    Article  CAS  Google Scholar 

  30. McDonagh, E.M., Whirl-Carrillo, M., Garten, Y., Altman, R.B. & Klein, T.E. Biomarkers Med. 5, 795–806 (2011).

    Article  CAS  Google Scholar 

  31. Cai, M.-C. et al. Nucleic Acids Res. 43, D907–D913 (2015).

    Article  CAS  Google Scholar 

Download references

Acknowledgements

The authors thank G. Beekhuis, M. Kienholz and J. Williams for editing the manuscript. B.A. dedicates this to his father, Lewis Aronow, coauthor of an important milestone in molecular pharmacology, Principles of Drug Action. Supported in part by NIH NCATS CTSA grant 1UL1TR001425-01.

Author information

Authors and Affiliations

  1. Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, USA

    Mayur Sarangdhar, Scott Tabar, Charles Schmidt, Akash Kushwaha, Krish Shah, Jeanine E Dahlquist, Anil G Jegga & Bruce J Aronow

  2. Clinical and Translational Sciences Program, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, USA

    Bruce J Aronow

Authors
  1. Mayur Sarangdhar
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Scott Tabar
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Charles Schmidt
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Akash Kushwaha
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Krish Shah
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Jeanine E Dahlquist
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Anil G Jegga
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Bruce J Aronow
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

M.S. and B.J.A. planned and designed AERSMine. M.S. wrote the manuscript. M.S. and S.T. were responsible for the development and testing of all aspects of AERSMine. M.S. validated the data, designed the statistical metrics, planned and implemented the normalizations and ontological aggregations; S.T. was chiefly responsible for the software development; S.T. and C.S. integrated canvasXpress within the AERSMine framework. M.S. and K.S. integrated the drug labels. B.J.A., J.E.D. and A.G.J. provided conceptual frameworks, clinical scenarios and suggested improvements to the statistical metrics. A.K. contributed to the architectural design. B.J.A. oversaw the project. All authors edited the article. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Bruce J Aronow.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Additional information

Editor's note: This article has been peer reviewed.

Integrated supplementary information

Supplementary Figure 1 Normalization and ontology-based hierarchical grouping of FAERS drugs, indications and AEs.

A semi-supervised multi-step iterative approach is used to normalize and unify drug entries to known generic names and further categorize into therapeutic classes based on the ATC nomenclature. Drug normalization ROC curve shows that our default mapping has a 93.5% TP rate matching them with generic labels (Supplementary Table 2).

Supplementary Figure 2 Case counts of top level ontological drug classes by ATC drug categories.

Ontological aggregation of normalized drug entries across top level ATC categories shows an overview of FAERS patient drug exposures.

Supplementary Figure 3 Aggregating FAERS clinical indications and adverse events to MedDRA Ontology version 16.1.

Indications and adverse events from FAERS reports are concept-aggregated using the MedDRA ontology (version 16.1) to facilitate classification across Preferred Terms (PTs), in addition to higher-level groupings including Higher Level Term (HLT), Higher Level Global Term (HLGT) and System Organ Class (SOC).

Supplementary Figure 4 Schematic representation of the AERSMine database layout.

A schematic representation of the database structure showing the FDA-provided files and their normalizations using the ATC/MedDRA ontologies as referenced by AERSMine to facilitate various levels of aggregations across drugs, indications and adverse events.

Supplementary Figure 5 AERSMine-facilitated concept grouping.

Both set-based (non-ontological) and ontology-based aggregation of drugs, indications and/or adverse events facilitate study cohort generation via systematic partitioning of case reports across various levels within the ATC and MedDRA hierarchies.

Supplementary Figure 6 AERSMine workflow to identify potential drug candidates that minimize lithium toxicities.

Example workflow showing AERSMine-facilitated approach to identify drug candidates based on differential AEs that could be significant in managing highly reported lithium toxicities, and indicates key under-represented groups (aggregated safety signals IC < 0 and ranked by high incidence/low risk) for cardiovascular system drugs.

Supplementary Figure 7 Clustering differential toxicities as a function of mutually exclusive drug exposures showing reduced risks in combinatorial therapy.

Therapeutic class-based comparative analysis showing reduced risk of lithium-related AEs in patients on cardiovascular medications.

Supplementary Figure 8 Clustering adverse events as a function of drug exposures to suggest candidate protective agents of lithium toxicity.

Comparative analysis of differential AE rates across patients on lithium, arbs, and lithium + arbs shows that the observed reporting rate of neurological AEs is lower than the expected rate in patients on a combination of lithium and arbs (Ω < 0, FDR Benjamini and Hochberg, p < 0.001).

Supplementary Figure 9 AERSMine workflow to identify differential response to therapeutics in TNF-elevated patients.

Example workflow of a multidimensional cohort-based study enabled by AERSMine to identify candidate factors with potential to increase/decrease relative risks for anti-TNF -associated AEs in TNF-elevated disorders (by reporting frequency).

Supplementary Figure 10 Differential AE risks among population subgroups as a function of indication, demographic, and the use of NSAIDs or coxibs.

AERSMine-facilitated clinical indications based construction and grouping of patient subsets showing differential AE risks in management of pain, arthritis and hypertension. Detection of differential risk patterns across similar clinical indication demographic subgroups allows for improved hypothesis generation of alternative therapeutic regimens.

Supplementary information

Supplementary Figures and Texts

Supplementary Figures 1–10 and Supplementary Notes 1–4 (PDF 3907 kb)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Sarangdhar, M., Tabar, S., Schmidt, C. et al. Data mining differential clinical outcomes associated with drug regimens using adverse event reporting data. Nat Biotechnol 34, 697–700 (2016). https://doi.org/10.1038/nbt.3623

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nbt.3623

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing