Predicting response and non-response to therapy in SLE

Marta E. Alarcón-Riquelme, M.D., Ph.D., Fundación Pública Andaluza Progreso y Salud 

Standard treatments for systemic lupus erythematosus (SLE), like antimalarial and immunosuppressive drugs, do not show much success in reducing disease severity, and glucocorticoids often cause harmful side effects. Biological treatments like belimumab (brand name Benlysta) have shown promise, but in a clinical trial, only 60 percent of people with SLE responded to belimumab treatment. This may be due to the heterogeneity of lupus—how it varies from person to person. Dr. Alarcón-Riquelme previously grouped people with SLE into one of four clusters based on their molecular features. She will now test whether these previously identified molecular groups can predict how a person with SLE responds to biological therapy. 

Dr. Alarcón-Riquelme previously showed that grouping people with SLE into one of four clusters based on their cellular and molecular features allowed the team to predict a person’s response to biological therapy. She will study this model on a larger scale by analyzing molecular features in 330 people with SLE for up to one year after starting a new treatment. She will ask which cells are different when a person responds or does not respond to therapy and what features of cells make them not respond to treatment. Dr. Alarcón-Riquelme will then use this data to group each person into one of the four clusters. Lastly, the team will analyze the molecular features of cells from blood and kidney or skin tissue from people with SLE to find patterns that define responders (those who achieved improvement) versus non-responders (those who did not). 

What this study means for people with lupus 

Finding molecular and cellular features that can be used to predict a person’s response to therapy means that people with SLE could be treated earlier, in a personalized way, avoiding rounds of unsuccessful therapies that can worsen the disease. Dr. Alarcón-Riquelme’s findings could transform SLE management and improve treatment for people with lupus. 

Systemic Lupus Erythematosus (SLE) is a highly heterogeneous disease with high morbidity and mortality. Antimalarial and immunosuppressant conventional treatment, such as azathioprine, methotrexate or cyclosporine, do not show much success in ameliorating disease activity, while glucocorticoid long-standing treatment is associated with undesirable side-effects. The last years, biological treatment has become the great promise towards personalized therapy for SLE. Still, despite great success against placebo, the frequency of Belimumab responders for example, in a phase III clinical trial was 60%. The proportion of unresponsiveness could be in part, due, to the molecular heterogeneity of lupus. Our group has been using various methods to stratify autoimmune disease and lupus patients. We recently described four molecular clusters: inflammatory, lymphoid and interferon as well as an undefined cluster representing patients under remission. Our results suggest that we can use the molecular stratification to predict whether a lupus patient may respond to a biological therapy. The overall objective of this proposal is to obtain a proof-of-concept study of our in silico data and show that therapeutic responsiveness to Belimumab and other biological drugs used for the treatment of lupus, can be predicted using our novel molecular stratification. The study design will allow us to study in-depth the molecular mechanisms of response for several drugs and the potential alterations that the therapies might have on the immune system, which cells may be the major responders or non-responders, their gene features and identify markers. Lastly, on the basis of the molecular classification and the mechanisms of response, a predictive molecular model will be developed to be used in future clinical trials or even in clinical practice. For this, we present the following specific aims:

Specific Aim 1: Attain proof of concept of the in silico prediction of molecular stratification on therapy responsiveness in a real set of cases in a prospective observational study with standard-of-care therapies.

Specific Aim 2: Define single cell transcriptome patterns of response/non-response to selected therapies in paired samples of blood cells and kidney or skin tissue.

Understanding SLE heterogeneity will help define much better therapeutic responses, but also, characterize patients’ cells in a homogeneous context.

Our proposal uses the latest machine learning approaches and scRNASeq methodology, which will provide important conclusions and knowledge to the study of lupus. For us, machine-learning is not only about analysis of data, but is to put the data at the service of clinical applicability.