The Lupus Foundation of America estimates that 1.5 million Americans, and at least 5 million people worldwide, have a form of Lupus.[1] Lupus is a chronic inflammatory disease that can affect any organ but varying estimates suggest 20 to 60% of patients may develop renal lupus or lupus nephritis during the course of their lupus disease.[2-5] Lupus nephritis is a major risk factor for morbidity and mortality in lupus, and 10% of patients with lupus nephritis will develop end stage renal disease.[5] Black and Hispanic/LatinX populations typically develop lupus nephritis earlier in their lupus disease course and have worse outcomes than white patients with lupus, including liver failure and death.

‍

Lupus nephritis research and real-world data‍

‍

Research using real-world datasets can provide insight into treatable patient populations but requires robust identification of patients. Two algorithms, based largely on ICD-coding and administrative claims data, have been validated on medicaid and veterans administration populations and combine data features including visits with nephrologists and treatments with relevant medications with ICD-coding for diagnosis of lupus, lupus nephritis, and common nephritis clinical features.[4,6] Even though adoption of the ICD-10 lupus nephritis-specific code (M32.14) is increasing, it is important to continue to employ a multi-pronged approach to patient identification when working with retrospective datasets.

‍

With that in mind, we’re looking forward to three new studies that will be featured at the ACR Convergence 2021 virtual conference. These studies explore how unstructured data can be used to identify lupus nephritis patients and determine risk for developing lupus nephritis using real-world data.

‍

Tierney and Rowe, scientists from PicnicHealth, examine how best to identify patients with lupus nephritis by comparing ICD-based classification abstracted from structured data to ICD-based classification combined with narrative text extraction of lupus nephritis-related terms found within clinical notes in a real-world data cohort constructed from electronic health records (EHRs).[7]

Deng, Y. et al. examine how lupus nephritis diagnoses can be gleaned from EHRs using natural language processing algorithms applied to structured and unstructured data, testing the use of curated concept unique identifiers (CUIs), the number of mentions, regular expression concepts as model features.[8]

Teixeira, B. et al. use a machine learning-based approach to classify SLE patients at high risk of developing lupus nephritis. This study uses data obtained using a point-in-time survey of rheumatologists and nephrologists and identifies positive and negative contributors to disease progression.[9]

‍

Why is it important to study patient identification?‍

‍

The treatment options for both lupus and lupus nephritis remained largely unchanged for many years, but recently the lupus nephritis community has gained two new approvals including belimumab and voclosporin, a biologic and oral treatment, respectively.[10] Additionally, there are several ongoing clinical studies in phase 2 and 3 registered on ClinicalTrials.gov.

‍

But even with new approved therapies and a healthy pipeline of therapies in development, a study by Dyball et. al. using real-world data determined that nearly half of the patients with active lupus nephritis would not be eligible for key clinical trials, making it difficult to understand the generalizability of newly approved and pipeline medications.[11] The most common exclusions in this study cohort were advanced renal disease, high background steroid use, history of malignancy, and prior cyclophosphamide use. 

‍

Accurately estimating the treatable population is important for treating physicians to properly guide care plans, for payers to plan for and negotiate coverage for their members, and for drug manufacturers to optimize their launch and access activities. It is clear that large amounts of data are available to researchers and healthcare decision makers. As real-world data is used more, it is critical to leverage complete data including, diagnoses, signs and symptoms recorded by providers, and laboratory and procedure results wherever they may be found within electronic health records.

‍

Download below Meghan Tierney and Chris Rowe’s ACR Convergence poster following the virtual conference, November 5 - 9th. 

‍

To learn more about how PicnicHealth’s scientists are exploring ways to improve the accuracy of data abstraction from narrative text at scale contact our team today.

References:

1. https://www.lupus.org/resources/lupus-facts-and-statistics accessed 10/29/2021

2. Saxena, R., Mahajan, T. & Mohan, C. Lupus nephritis: current update. Arthritis Res Ther 13, 240 (2011). https://doi.org/10.1186/ar3378.

3. Feldman, C.H., Hiraki, L.T., Liu, J., et al. Epidemiology and sociodemographics of systemic lupus erythematosus and lupus nephritis among US adults with Medicaid coverage, 2000–2004. Arthritis & Rheumatism, 65: 753-763 (2013). https://doi.org/10.1002/art.37795.

4. Li, T., Lee, I., Jayakumar, D., et al. Development and validation of lupus nephritis case definitions using United States veterans affairs electronic health records. Lupus 30(3) 518-526 (2021). https://doi.org/10.1177/0961203320973267.

5. Almaani, S., Meara, A., and Rovin, B.H. Updated on Lupus Nephritis. CJASN May 2017, 12 (5) 825-835; DOI: 10.2215/CJN.05780616.

6. Chibnik LB, Massarotti EM, Costenbader KH. Identification and validation of lupus nephritis cases using administrative data. Lupus. 2010 May;19(6):741-3. doi: 10.1177/0961203309356289.

7. Tierney, M and Rowe, C. Addition of Narrative Text Abstraction to ICD-Based Abstraction Significantly Improves Identification of Lupus Nephritis in Real-World Data [abstract]. Arthritis Rheumatol. 2021; 73 (suppl 10). Accessed October 29, 2021.

8. Deng, Y., Pacheco, J., Chung, A. et al. Natural Language Processing to Identify Lupus Nephritis Phenotype in Electronic Health Records [abstract]. Arthritis Rheumatol. 2021; 73 (suppl 10). Accessed October 29, 2021.

9. Teixeira, B., Bell, D., Holbrook, T., et al. Machine Learning: Identifying Lupus Nephritis Within Systemic Lupus Erythematosus in the Real World [abstract]. Arthritis Rheumatol. 2021; 73 (suppl 10). Access October 29, 2021.

10. https://www.the-rheumatologist.org/article/fda-approves-belimumab-voclosporin-for-lupus-nephritis/ accessed 10/29/2021

11. Dyball, S., Collinson, S., Sutton, E., et al. Lupus clinical trial eligibility in a real-world setting: results from the British Isles Lupus Assessment Group-Biologics Register (BILAG-BR) Lupus Science & Medicine 2021;8:e000513. doi: 10.1136/lupus-2021-000513.

‍

{{ln-rwd}}