Pretreatment Prediction of Individual Rheumatoid Arthritis Patients' Response to Anti-Cytokine Therapy Using Serum Cytokine/Chemokine/Soluble Receptor Biomarkers

Towards Personalized Medicine in Rheumatoid Arthritis

Rheumatoid arthritis (RA) is a chronic, incurable, multisystem, inflammatory disease characterized by synovitis and extra-articular features. Although several advanced therapies targeting inflammatory mechanisms underlying the disease are available, no advanced therapy is universally effective. Therefore, a ceiling of treatment response is currently accepted where no advanced therapy is superior to another. The current challenge for medical research is the discovery and integration of predictive markers of drug response that can be used to personalize medicine so that the patient is started on "the right drug at the right time". This review article summarizes our current understanding of predicting response to anti-rheumatic drugs in RA, obstacles impeding the development of personalized medicine approaches and future research priorities to overcome these barriers.

Efficacy and safety of canakinumab as a second line biologic after tocilizumab treatment failure in children with systemic juvenile idiopathic arthritis: A single-centre cohort study using routinely collected health data

Background

A significant number of systemic juvenile idiopathic arthritis (sJIA) patients discontinue biologic disease-modifying antirheumatic drugs (bDMARDs) due to lack of efficacy or safety concerns. Studies of biologic therapy switch regimens in sJIA are required.

Methods

Patients with sJIA who switched from tocilizumab (due to lack of efficacy or safety) to canakinumab (4 mg/kg every 4 weeks) and were hospitalized at the rheumatology department from August 2012 to July 2020 were included. Primary efficacy outcomes were 30% or greater improvement based on the paediatric criteria of the American College of Rheumatology (ACR30), achievement of inactive disease (JADAS-71 = 0) and clinical remission (ACR sJIA clinical inactive disease criteria). Follow-up from time first canakinumab dose administered was 12 months or the closest time point (not less than 6 and not more than 18 months). Data were extracted from electronic outpatient medical records.

Results

During the study period, 46 patients with sJIA switched from tocilizumab to canakinumab. Median age at baseline was 8.2 [interquartile range (IQR) 4.0-12.9] years, with the median sJIA duration being 1.8 (IQR 0.8-5.8) years; 37 (80%) patients received at least one conventional DMARD (cDMARD; oral corticosteroids, methotrexate and/or cyclosporine A). Study outcomes were followed up in 45 patients (one patient did not attend the follow-up for an unknown reason); median follow-up was 359 (IQR 282-404) days. During the follow-up, 1 patient discontinued canakinumab due to tuberculosis detection and the dose was reduced or the injection interval increased in 4 (9%) patients. In total, 27 (60%) patients continued to receive at least one cDMARD. Improvement according to the ACR30 criteria was achieved in 43 patients [96%; 95% confidence interval (CI) 85-99], inactive disease in 42 (93%; 95%CI 82-98), and remission in 37 (82%; 95% CI 69-91); after adjustment for actual time-at-risk, the rates were 83, 85 and 73 events per 100 person-years, respectively. During follow-up, 23 AEs (most frequently infections) were reported in 19/45 (42%) patients; 5/45 (11%) patients developed macrophage activation syndrome, with a favorable outcome in all cases.

Conclusions

One-year canakinumab therapy was found to be potentially effective as second-line biologic therapy after discontinuation of tocilizumab in patients with sJIA.

Mechanisms underlying DMARD inefficacy in difficult-to-treat rheumatoid arthritis: a narrative review with systematic literature search

Management of RA patients has significantly improved over the past decades. However, a substantial proportion of patients is difficult-to-treat (D2T), remaining symptomatic after failing biological and/or targeted synthetic DMARDs. Multiple factors can contribute to D2T RA, including treatment non-adherence, comorbidities and co-existing mimicking diseases (e.g. fibromyalgia). Additionally, currently available biological and/or targeted synthetic DMARDs may be truly ineffective (‘true’ refractory RA) and/or lead to unacceptable side effects. In this narrative review based on a systematic literature search, an overview of underlying (immune) mechanisms is presented. Potential scenarios are discussed including the influence of different levels of gene expression and clinical characteristics. Although the exact underlying mechanisms remain largely unknown, the heterogeneity between individual patients supports the assumption that D2T RA is a syndrome involving different pathogenic mechanisms.

Toward Overcoming Treatment Failure in Rheumatoid Arthritis

Rheumatoid arthritis (RA) is an autoimmune disorder characterized by inflammation and bone erosion. The exact mechanism of RA is still unknown, but various immune cytokines, signaling pathways and effector cells are involved. Disease-modifying antirheumatic drugs (DMARDs) are commonly used in RA treatment and classified into different categories. Nevertheless, RA treatment is based on a "trial-and-error" approach, and a substantial proportion of patients show failed therapy for each DMARD. Over the past decades, great efforts have been made to overcome treatment failure, including identification of biomarkers, exploration of the reasons for loss of efficacy, development of sequential or combinational DMARDs strategies and approval of new DMARDs. Here, we summarize these efforts, which would provide valuable insights for accurate RA clinical medication. While gratifying, researchers realize that these efforts are still far from enough to recommend specific DMARDs for individual patients. Precision medicine is an emerging medical model that proposes a highly individualized and tailored approach for disease management. In this review, we also discuss the potential of precision medicine for overcoming RA treatment failure, with the introduction of various cutting-edge technologies and big data.

Anti-inflammatory effects of β-1,3-1,6-glucan derived from black yeast Aureobasidium pullulans in RAW264.7 cells

β-glucan derived from the black yeast Aureobasidium pullulans (A. pullulans) is one of the natural products attracting attention due to its high potential for application as a functional food and pharmaceutical. Our team of researchers obtained a highly soluble, low-molecular-weight β-glucan from the fermentation culture medium of A. pullulans via mechanochemical ball milling method, that is, the low-molecular-weight A. pullulans-fermented β-D-glucan (LMW-AP-FBG). We investigated the anti-inflammatory effect of LMW-AP-FBG using lipopolysaccharide (LPS)-stimulated murine macrophages (RAW264.7 cells) in the current study. LMW-AP-FBG altered LPS-stimulated inflammatory responses by reducing the release of inflammatory mediators such as nitric oxide (NO), interleukin (IL)-1β, IL-6 and tumor necrosis factor-α. As well, the mitogen-activated protein kinases (MAPKs) and nuclear factor-κB (NF-κB) signaling pathways were downregulated by LMW-AP-FBG. Furthermore, LMW-AP-FBG markedly reduced LPS-induced expression of cell surface molecules, CD14, CD86, and MHC class II, which mediate macrophage activation. These findings suggest that LMW-AP-FBG can be used as an effective immune modulator to attenuate the progression of inflammatory disease.

Potential clinical biomarkers in rheumatoid arthritis with an omic approach

Objective

To aid in the selection of the most suitable therapeutic option in patients with diagnosis of rheumatoid arthritis according to the phase of disease, through the review of articles that identify omics biological markers.

Methods

A systematic review in PubMed/Medline databases was performed. We searched articles from August 2014 to September 2019, in English and Spanish, filtered by title and full text; and using the terms "Biomarkers" AND “Rheumatoid arthritis".

Results

This article supplies an exhaustive review from research of objective measurement, omics biomarkers and how disease activity appraise decrease unpredictability in treatment determinations, and finally, economic, and clinical outcomes of treatment options by biomarkers’ potential influence. A total of 122 articles were included. Only 92 met the established criteria for review purposes and 17 relevant references about the topic were included as well. Therefore, it was possible to identify 196 potential clinical biomarkers: 22 non-omics, 20 epigenomics, 33 genomics, 21 transcriptomics, 78 proteomics, 4 glycomics, 1 lipidomics and 17 metabolomics.

Conclusion

A biomarker is a measurable indicator of some, biochemical, physiological, or morphological condition; evaluable at a molecular, biochemical, or cellular level. Biomarkers work as indicators of physiological or pathological processes, or as a result of a therapeutic management. In the last five years, new biomarkers have been identified, especially the omics, which are those that proceed from the investigation of genes (genomics), metabolites (metabolomics), and proteins (proteomics). These biomarkers contribute to the physician choosing the best therapeutic option in patients with rheumatoid arthritis.

Objective and noninvasive biochemical markers in rheumatoid arthritis: where are we and where are we going?

INTRODUCTION

Rheumatoid arthritis (RA) is a chronic autoimmune disease that affects approximately 1% of the adult population. RA is multi-factorial, and as such our understanding of the molecular pathways involved in the disease is currently limited. An increasing number of studies have suggested that several molecular phenotypes (i.e. endotypes) of RA exist, and that different endotypes respond differently to various treatments. Biochemical markers may be an attractive means for achieving precision medicine, as they are objective and easily obtainable.

AREAS COVERED

We searched recent publications on biochemical markers in RA as either diagnostic or prognostic markers, or as markers of disease activity. Here, we provide a narrative overview of different classes of markers, such as autoantibodies, citrulline products, markers of tissue turnover and cytokines, that have been tested in clinical cohorts or trials including RA patients.

EXPERT OPINION

Although many biochemical markers have been identified and tested, few are currently being used in clinical practice. As more treatment options are becoming available, the need for precision medicine tools that can aid physicians and patients in choosing the right treatment is growing.

Comparison of the drug retention and reasons for discontinuation of tumor necrosis factor inhibitors and interleukin-6 inhibitors in Japanese patients with elderly-onset rheumatoid arthritis-the ANSWER cohort study

BACKGROUND

This multi-center, retrospective study aimed to clarify retention rates and reasons for discontinuation of either tumor necrosis factor inhibitors (TNFi) or interleukin-6 inhibitors (IL-6i) in patients with elderly-onset rheumatoid arthritis (EORA).

METHODS

Patients with rheumatoid arthritis (RA) enrolled in a Japanese multicenter observational registry between 2011 and 2020 were included. EORA was defined as RA with onset at 60 or over. To adjust confounding by indication for treatment with TNFi or IL-6i, a propensity score based on multiple baseline characteristics variables was used to compare the drug retention and causes for discontinuation between TNFi and IL-6i. Adjusted cumulative incidence of drug discontinuation for each reason was compared between the two groups using the Fine-Gray model.

RESULTS

Among a total of 9,550 patients in the registry, 674 TNFi and 297 IL-6i initiators with EORA were identified. Age, the proportion of females, disease duration, and baseline disease activity at the time of TNFi or IL-6i initiation were similar between the two groups. After adjusting for differences in baseline characteristics between the two groups, overall drug discontinuation was significantly lower in the IL-6i as compared to the TNFi (HR = 0.71, 95%CI = 0.59-0.86, p < 0.001). The adjusted cumulative incidence of discontinuation due to lack of effectiveness was lower with the IL-6i (HR = 0.46, 95%CI = 0.33-0.63, p < 0.001) while those due to adverse events (HR = 0.82, 95%CI = 0.56-1.18, p = 0.28) or achievement of clinical remission (HR = 1.09, 95%CI = 0.62-1.91, p = 0.76) were similar between the two groups.

CONCLUSIONS

In EORA patients initiating a TNFi or IL-6i, significantly higher drug retention was observed with IL-6i. Discontinuation due to lack of effectiveness was significantly less frequent in IL-6i while discontinuations due to adverse event or achievement of clinical remission were similar between the two groups.

Predicting treatment response to IL6R blockers in rheumatoid arthritis

Patients with severe, active RA who have not responded to conventional therapy may receive biological disease modifying anti-rheumatic drugs (bDMARDs). However, 40% of cases do not achieve complete disease control, resulting in a negative impact on patient quality of life and representing a waste of healthcare resources. Ongoing research seeks to establish biomarkers, which can be used to predict treatment response to biologics in RA to enable more targeted approaches to treatment. However, much of the work has focused on one class of biologic drug, the TNF inhibitors (TNFi). Here, we will review the current state of research to identify biomarkers predictive of response to the class of bDMARDs targeting the IL6R. While success has been limited thus far, serum drug and low ICAM1 levels have shown promise, with associations reported in independent studies. The challenges faced by researchers and lessons learned from studies of TNFi will be discussed.

Association of High Serum Interleukin-6 Levels With Severe Progression of Rheumatoid Arthritis and Increased Treatment Response Differentiating Sarilumab From Adalimumab or Methotrexate in a Post Hoc Analysis

Objective

The development of biomarkers to guide treatment decisions is a major research focus in rheumatoid arthritis (RA). Patients with RA have elevated interleukin‐6 (IL‐6) levels; however, the utility of IL‐6 as a predictor of treatment response is unclear. This study was undertaken to investigate, by post hoc analysis, whether baseline IL‐6 levels are predictive of sarilumab treatment responses in 2 phase III studies.

Methods

Serum IL‐6 concentrations were measured in patients with RA prior to receiving sarilumab 200 mg (n = 148) or adalimumab 40 mg (n = 152) every 2 weeks (in the MONARCH trial; ClinicalTrials.gov identifier: NCT02332590) or sarilumab 150 mg, sarilumab 200 mg, or placebo every 2 weeks plus methotrexate (MTX) (n = 401, n = 396, and n = 397, respectively) (in the MOBILITY trial; ClinicalTrials.gov identifier: NCT01061736). Efficacy and patient‐reported outcomes were compared between and within groups according to IL‐6 tertile using linear and logistic regression.

Results

In MONARCH, patients with high baseline IL‐6 levels (all ≥3 times the upper limit of normal; n = 100) had higher disease activity at baseline than those with low IL‐6 levels (n = 100). The magnitude of clinical improvement over 24 weeks with sarilumab versus adalimumab was greater in patients with high compared to those with low baseline IL‐6 levels. In MOBILITY, compared to patients with low IL‐6 levels (n = 397), patients with high IL‐6 levels (n = 398) had higher disease activity and joint damage at baseline, were more likely to have joint progression, and had less clinical improvement over 52 weeks’ treatment with placebo plus MTX compared to sarilumab 150 mg or 200 mg plus MTX. Baseline IL‐6 and C‐reactive protein levels were both predictive of outcomes. Safety profiles were similar between defined IL‐6 tertiles.

Conclusion

IL‐6 may be a prognostic marker of disease progression and severity, and patients with high IL‐6 levels may be likely to benefit from sarilumab compared to adalimumab or MTX. Prospective validation is warranted to confirm the results of these post hoc analyses.

Multiple biomarker approach for the diagnosis and therapy of rheumatoid arthritis

The lack of specific clinical symptoms for patients in the early stage of rheumatoid arthritis (RA) has created strong interest in the laboratory diagnosis of RA. The main laboratory markers of RA, rheumatoid factor (RF) and anti-citrullinated protein antibodies (ACPAs), can be found in patients with other pathologies and in healthy donors. Even today, there is no single laboratory test that can diagnosis RA with high sensitivity and specificity. To improve the diagnosis and treatment of RA, alternative biomarkers, including 14-3-3η protein, connective tissue growth factor (CTGF), antibodies against PAD4, antibodies against BRAF, and anti-acetylated and anti-carbamylated protein antibodies have been studied extensively. The use of a multiple biomarker approach, the simultaneous measurement of a set of biomarkers, is an alternative strategy for the diagnosis of RA and for predicting the therapeutic effect of biological disease-modifying antirheumatic drugs (DMARDs). However, despite the large number of studies, only a few biomarker combinations have been validated and can be applied in clinical practice. In this article, results of studies focused on the multiple biomarker approach (both multiplex and combined single-analyte assays) to diagnose RA and to predict response to biological drug therapy are reviewed. Additionally, general factors limiting the use of multiplex analysis in RA diagnostics and therapy are discussed.

Serum level of soluble interleukin 6 receptor is a useful biomarker for identification of treatment-resistant major depressive disorder

AIM

A substantial proportion of major depressive disorder patients are treatment-resistant to antidepressant therapy, who require augmentation drugs, or other treatments including electroconvulsive therapy or transcranial magnetic stimulation. Identifying treatment-resistant major depressive disorder patients before the actual administration of antidepressant is, however, often difficult. Accordingly, the serum biomarker to identify treatment-resistant patients will be helpful in clinical settings. This study aims to clarify the appropriate biomarkers for identification of treatment-resistant major depressive disorder.

METHOD

Given that immune-inflammatory processes are involved in the pathogenesis of major depressive disorder, it is possible that certain cytokine-related molecules could serve as clinically useful biomarkers of treatment-resistant major depressive disorder patients. In this study, we measured serum levels of tumor necrosis factor-α, interleukin 6, and soluble interleukin 6 receptor after major depressive disorder patients underwent antidepressant therapy.

RESULTS

The serum level of soluble interleukin 6 receptor, but not interleukin 6 or tumor necrosis factor-α, was significantly higher in treatment-resistant major depressive disorder patients than in remitted patients, suggesting that serum soluble interleukin 6 receptor could be a good biomarker of treatment-resistant major depressive disorder. Receiver operating characteristic analysis confirmed that serum soluble interleukin-6 receptor level measurement was useful for identification of treatment-resistant major depressive disorder patients. Multiple regression analysis using the serum levels of the aforementioned cytokines as explanatory variables and the Quick Inventory of Depressive Symptomatology-Self Report score (QIDS-SR16 ) as a target variable showed that only serum soluble interleukin-6 receptor level could explain the severity of major depressive disorder.

CONCLUSION

Based on these results, we recommend measurement of serum soluble interleukin-6 receptor level to discriminate treatment-resistant major depressive disorder patients. High serum soluble interleukin-6 receptor level is associated with the pathogenesis of treatment-resistant major depressive disorder, suggesting the involvement of the interleukin 6 trans-signaling system in onset of treatment-resistant major depressive disorder.

Acupuncture Alleviates Rheumatoid Arthritis by Immune-Network Modulation

Rheumatoid arthritis (RA) is a chronic, systemic autoimmune inflammatory and debilitating disease that involves the systemic imbalance of the immune network. Previous studies have shown that acupuncture can help treat RA. However, its specific mechanisms are not fully understood. Thus, the present study was designed to clarify the mechanisms of acupuncture acted on RA via immune network modulation using complete Freund’s adjuvant (CFA)-induced arthritic rats. Results revealed that manual acupuncture (MA) could alleviate the inflammation and pain of infected joints. Moreover, MA could effectively stimulate the innate immune cytokines (IL-1[Formula: see text], IL-1[Formula: see text], IL-6, IL-7, IL-18, TNF-[Formula: see text]) and adaptive immunity cytokines (IL-2, IL-12, IFN-[Formula: see text], IL-4, IL-5, IL-10, IL-13, IL-17) as the main part of the immune response and repaired damage of RA. These complex immunomodulatory processes were analyzed quantitatively by cell–cell communication (CCC) networks. The CCC networks demonstrated that the immune networks were enhanced with the development of RA, while MA enhanced the immune networks in the early stage to act on RA and promoted the immune-network to a normal level at the late stage. Moreover, we found that monocyte/macrophage and endothelial cells were the key cells of innate immunity and body cells; TH1, TH2 and B cells were the key cells of adaptive immunity, which were also the main target cells for MA regulation.

Autoantibodies, C-reactive protein, erythrocyte sedimentation rate and serum cytokine profiling in monitoring of early treatment

Introduction

Currently used clinical scale and laboratory markers to monitor patients with early rheumatoid arthritis (RA) seem to be not sufficient. It has been demonstrated that disease- related cytokines may be elevated very early in RA development and cytokines are considered as the biomarkers potentially useful for RA monitoring.

Material and methods

The group of patients with undifferentiated arthritis (UA) developing RA (UA→RA) was identified from a total of 121 people with arthralgia. UA→RA (n = 16) and healthy control (n = 16) subjects underwent clinical and laboratory evaluation, including acute phase reactants (APRs) and autoantibodies. Cytokines IFN-γ, IL-10, TNF, IL-17A, IL-6, IL-1b, IL-2 in sera were assayed using flow cytometric bead array test.

Results

34.5% of patients with UA developed RA. DAS28 reduced as early as 3 months after initiation of treatment. No DAS28 difference between groups of autoantibody (RF, anti-CCP, ANA-HEp-2) -positive and -negative patients was observed, however, comparing groups of anti-CCP and RF-double negative and -double positive patients, the trend of sooner clinical improvement was visible in the second abovementioned group. After the treatment introduction, the ESR level reduced significantly, while CRP level reduction was not significant. Serum cytokine levels of IL-10, IL-6 and IL-17A reduced after 6 months since introduction of treatment. The positive correlations between ESR, CRP and specific cytokine levels were observed.

Conclusions

The autoantibody and APR profile is poorly connected with the RA course. The serum cytokine profile change in the course of RA and may be potentially used for optimization of RA monitoring.

Right drug, right patient, right time: aspiration or future promise for biologics in rheumatoid arthritis?

Individualising biologic disease-modifying anti-rheumatic drugs (bDMARDs) to maximise outcomes and deliver safe and cost-effective care is a key goal in the management of rheumatoid arthritis (RA). Investigation to identify predictive tools of bDMARD response is a highly active and prolific area of research. In addition to clinical phenotyping, cellular and molecular characterisation of synovial tissue and blood in patients with RA, using different technologies, can facilitate predictive testing. This narrative review will summarise the literature for the available bDMARD classes and focus on where progress has been made. We will also look ahead and consider the increasing use of 'omics' technologies, the potential they hold as well as the challenges, and what is needed in the future to fully realise our ambition of personalised bDMARD treatment.

Proteomics to predict the response to tumour necrosis factor-α inhibitors in rheumatoid arthritis using a supervised cluster-analysis based protein score

OBJECTIVE

In rheumatoid arthritis (RA), it is of major importance to identify non-responders to tumour necrosis factor-α inhibitors (TNFi) before starting treatment, to prevent a delay in effective treatment. We developed a protein score for the response to TNFi treatment in RA and investigated its predictive value.

METHOD

In RA patients eligible for biological treatment included in the BiOCURA registry, 53 inflammatory proteins were measured using xMAP® technology. A supervised cluster analysis method, partial least squares (PLS), was used to select the best combination of proteins. Using logistic regression, a predictive model containing readily available clinical parameters was developed and the potential of this model with and without the protein score to predict European League Against Rheumatism (EULAR) response was assessed using the area under the receiving operating characteristics curve (AUC-ROC) and the net reclassification index (NRI).

RESULTS

For the development step (n = 65 patient), PLS revealed 12 important proteins: CCL3 (macrophage inflammatory protein, MIP1a), CCL17 (thymus and activation-regulated chemokine), CCL19 (MIP3b), CCL22 (macrophage-derived chemokine), interleukin-4 (IL-4), IL-6, IL-7, IL-15, soluble cluster of differentiation 14 (sCD14), sCD74 (macrophage migration inhibitory factor), soluble IL-1 receptor I, and soluble tumour necrosis factor receptor II. The protein score scarcely improved the AUC-ROC (0.72 to 0.77) and the ability to improve classification and reclassification (NRI = 0.05). In validation (n = 185), the model including protein score did not improve the AUC-ROC (0.71 to 0.67) or the reclassification (NRI = -0.11).

CONCLUSION

No proteomic predictors were identified that were more suitable than clinical parameters in distinguishing TNFi non-responders from responders before the start of treatment. As the results of previous studies and this study are disparate, we currently have no proteomic predictors for the response to TNFi.

Prediction of G Protein-Coupled Receptors with SVM-Prot Features and Random Forest

G protein-coupled receptors (GPCRs) are the largest receptor superfamily. In this paper, we try to employ physical-chemical properties, which come from SVM-Prot, to represent GPCR. Random Forest was utilized as classifier for distinguishing them from other protein sequences. MEME suite was used to detect the most significant 10 conserved motifs of human GPCRs. In the testing datasets, the average accuracy was 91.61%, and the average AUC was 0.9282. MEME discovery analysis showed that many motifs aggregated in the seven hydrophobic helices transmembrane regions adapt to the characteristic of GPCRs. All of the above indicate that our machine-learning method can successfully distinguish GPCRs from non-GPCRs.

Identification of Predictive Biomarkers for Cytokine Release Syndrome after Chimeric Antigen Receptor T-cell Therapy for Acute Lymphoblastic Leukemia

Chimeric antigen receptor (CAR)-modified T cells with anti-CD19 specificity are a highly effective novel immune therapy for relapsed/refractory acute lymphoblastic leukemia. Cytokine release syndrome (CRS) is the most significant and life-threatening toxicity. To improve understanding of CRS, we measured cytokines and clinical biomarkers in 51 CTL019-treated patients. Peak levels of 24 cytokines, including IFNγ, IL6, sgp130, and sIL6R, in the first month after infusion were highly associated with severe CRS. Using regression modeling, we could accurately predict which patients would develop severe CRS with a signature composed of three cytokines. Results were validated in an independent cohort. Changes in serum biochemical markers, including C-reactive protein and ferritin, were associated with CRS but failed to predict development of severe CRS. These comprehensive profiling data provide novel insights into CRS biology and, importantly, represent the first data that can accurately predict which patients have a high probability of becoming critically ill.

SIGNIFICANCE

CRS is the most common severe toxicity seen after CAR T-cell treatment. We developed models that can accurately predict which patients are likely to develop severe CRS before they become critically ill, which improves understanding of CRS biology and may guide future cytokine-directed therapy. Cancer Discov; 6(6); 664-79. ©2016 AACR.See related commentary by Rouce and Heslop, p. 579This article is highlighted in the In This Issue feature, p. 561.