Detection of cytokine release syndrome using wearables and cytokine profiling following CAR-T therapy for myeloma
Sridevi Rajeeve, Matt Wilkes, Nicole Zahradka, Lewis Tomalin, Mujahid Quidwai, Darren Pan, Nicholas J. Calafat, Martin Cusack, Adolfo Aleman, Kseniya Serebryakova, Katerina Kappes, Hayley Jackson, Sarita Agte, Santiago Thibaud, Larysa Sanchez, Shambavi Richard, Joshua Richter, Cesar Rodriguez, Hearn Jay Cho, Ajai Chari, Sundar Jagannath, Alessandro Laganà, Adriana C. Rossi, Samir Parekh
Sridevi Rajeeve, Matt Wilkes, Nicole Zahradka, Lewis Tomalin, Mujahid Quidwai, Darren Pan, Nicholas J. Calafat, Martin Cusack, Adolfo Aleman, Kseniya Serebryakova, Katerina Kappes, Hayley Jackson, Sarita Agte, Santiago Thibaud, Larysa Sanchez, Shambavi Richard, Joshua Richter, Cesar Rodriguez, Hearn Jay Cho, Ajai Chari, Sundar Jagannath, Alessandro Laganà, Adriana C. Rossi, Samir Parekh
View: Text | PDF
Clinical Research and Public Health Hematology Immunology Oncology

Detection of cytokine release syndrome using wearables and cytokine profiling following CAR-T therapy for myeloma

Abstract

BACKGROUND Chimeric antigen receptor T-cell (CAR-T) therapies have revolutionized treatment for relapsed/refractory multiple myeloma (RRMM). However, cytokine release syndrome (CRS), a common and potentially severe complication, requires inpatient monitoring, limiting access and increasing costs. Wearable devices could support outpatient CAR-T delivery, but feasibility for CRS detection versus standard care remains unproven.METHODS We conducted a prospective, single-center observational pilot study to assess the feasibility of using wearable devices for monitoring vital signs and detecting CRS. Thirty patients receiving idecabtagene vicleucel (ide-cel) or ciltacabtagene autoleucel (cilta-cel) were enrolled; 25 with sufficient monitoring data were evaluable. Sensors collected skin and axillary temperature, oxygen saturation, respiratory and heart rate, and motion. Peripheral blood cytokines were analyzed pre- and postinfusion using a multiplex proteomic platform. The primary outcome was feasibility, assessed by CRS detection sensitivity and specificity; secondary outcomes included adherence, lead time, and performance of models integrating wearable and cytokine data.RESULTS CRS occurred in 20 of 25 patients. The best-performing wearable model detected 18 or 20 CRS episodes with a sensitivity of 0.72 (mean 0.75; 95% CI 0.60–0.91) and a specificity of 0.80 (mean 0.76; 95% CI 0.68–0.84), and a median lead time of 7:00 hours before nursing recognition. Median adherence during high-risk periods was 71%. Cytokine changes paralleled temperature elevations, and IFN-γ emerged as a consistent biomarker.CONCLUSION Wearable devices are feasible for early CRS detection and may support outpatient CAR-T care. Larger outpatient studies are warranted.TRIAL REGISTRATION This study did not meet the criteria for ClinicalTrials.gov registration.

Authors

Sridevi Rajeeve, Matt Wilkes, Nicole Zahradka, Lewis Tomalin, Mujahid Quidwai, Darren Pan, Nicholas J. Calafat, Martin Cusack, Adolfo Aleman, Kseniya Serebryakova, Katerina Kappes, Hayley Jackson, Sarita Agte, Santiago Thibaud, Larysa Sanchez, Shambavi Richard, Joshua Richter, Cesar Rodriguez, Hearn Jay Cho, Ajai Chari, Sundar Jagannath, Alessandro Laganà, Adriana C. Rossi, Samir Parekh

×

Figure 2

Olink analysis reveals distinct timing of peak inflammation following ide-cel and cilta-cel.

Options: View larger image (or click on image) Download as PowerPoint
Olink analysis reveals distinct timing of peak inflammation following id...
(A and B) The heatmaps show scaled model-estimated mean expression of inflammatory biomarkers following treatment with cilta-cel or ide-cel. Biomarkers showing significant changes in both treatment groups are shown in A, whereas biomarkers showing treatment-specific changes are shown in B. Longitudinal expression of each inflammatory marker was analyzed using a linear mixed-effects regression model fit with the DREAM framework, with time as a fixed effect, age as a covariate, and random intercepts for patient ID and plate ID. Comparisons of interest were performed using contrasts, and P values were adjusted for multiple testing using the Benjamini-Hochberg method. Asterisks indicate significant change from baseline at the indicated time point (fold-change > 1.3). (C) Model-estimated fold-change from baseline for IFN-γ following ide-cel or cilta-cel treatment. Fold-changes and significance were derived from the same linear mixed-effects regression/DREAM analysis described above, with Benjamini-Hochberg correction for multiple testing. Asterisks indicate significant change from baseline at each time point (* FDR < 0.05, ** FDR < 0.01, *** FDR < 0.001).