IL-10 plus the EASIX score predict bleeding events after anti-CD19 CAR T-cell therapy

Chimeric antigen receptor (CAR) T-cell-associated coagulopathy can cause bleeding events. To explore risk factors for hemorrhage after CAR T-cell therapy, we retrospectively analyzed routine indicators in 56 patients with non-Hodgkin lymphoma and B-cell acute lymphoblastic leukemia who received anti-CD19 CAR T-cell therapy. Disturbance of coagulation occurred mainly within one month post infusion, especially on day 7 and 14. The cumulative incidence of bleeding events within one month was 32.8%, with the median onset of 7 (range, 0–28) days. All bleeding events were grade 1–3. Patients who experienced bleeding events within one month had longer prothrombin time, higher IL-6, higher IL-10, and lower platelets before lymphodepletion. There were also correlations among coagulation-, inflammatory-, and tumor burden-related markers. Multi-variate analysis showed IL-10 (> 7.98 pg/mL; adjusted odds ratio [OR], 13.84; 95% confidence interval [CI], 2.03–94.36; P = 0.007) and the endothelial activation and stress index (EASIX, defined as dehydrogenase [U/L] × creatinine [mg/dL] / platelets [×109 cells/L]; >7.65; adjusted OR, 7.06; 95% CI, 1.03–48.23; P = 0.046) were significant risk factors for bleeding events. IL-10 plus the EASIX defined three risk groups for bleeding events with cumulative incidence of 100% (hazard ratio [HR], 14.47; 95% CI, 2.78–75.29; P < 0.0001), 38.5% (HR, 3.68; 95% CI, 0.82–16.67; P = 0.089), and 11.8% (reference), respectively. Future studies are needed to verify the risk assessment models for bleeding events after CAR T-cell treatment in larger cohorts. Supplementary Information The online version contains supplementary material available at 10.1007/s00277-023-05477-y.

CAR T-cell treatment also causes toxicities, most typically cytokine release syndrome (CRS) and immuneeffector cell-associated neurotoxicity syndrome (ICANS).However, treatment-related coagulopathy is also a major and frequent adverse event.CAR T-cell-associated coagulopathy (CARAC) is a clinical syndrome characterized by bleeding and/or thrombosis accompanied by decreased platelets (PLTs) and coagulopathy, which occurs mainly within 28 days after CAR-T cell infusion [4].During CAR T-cell treatment, more than 50% of patients will experience XW and CL contributed equally as co-first authors.
HM and YH contributed equally as co-senior authors.
The endothelium is important for maintaining homeostasis in the coagulation system.Its integrity contributes to maintain anticoagulant properties and control of platelet adhesion and activation [20].In addition, endothelial cells partially regulate blood vessels in response to bleeding events [20].High lactic dehydrogenase (LDH), high creatinine, and low platelets are important indicators of thrombotic microangiopathy after allogeneic transplantation [21].The endothelial activation and stress index (EASIX) score, defined as LDH (U/L) × creatinine (mg/dL) / PLTs (×10 9 /L), is used to predict endothelial complications and mortality in allogeneic transplantation and acute graft-versus-host disease [22].There are also modified EASIX variants to predict CRS and ICANS during CAR T-cell treatment [23][24][25].Because endothelial activation triggers fluctuations in coagulation, we speculate that the EASIX score could also be used to predict and guide management of hemorrhage caused by coagulopathy during CAR T-cell therapy.
In this retrospective study, we analyzed fluctuations in coagulation, recorded bleeding episodes, as well as screened coagulation-, inflammation-, and tumor burden-related indicators for risk stratification of bleeding events during CAR T-cell treatment in a single-center cohort of 56 patients.

Patient and clinical data collection
We conducted a retrospective analysis of patients with hematological malignances including non-Hodgkin lymphoma (NHL) and B-cell acute lymphoblastic leukemia (B-ALL) treated with anti-CD19 CAR T-cells from a clinical trial (NCT04008251).The production and treatment strategies of CAR T-cells in the clinical trial are consistent with previous studies [26].The second-generation CAR structure used in the study contains a humanized CD19 single chain variable fragment, CD137 costimulatory domain, and CD3ζ signal sequence [27].Patients in the study received lymphodepleting chemotherapy (fludarabine 30mg/m 2 and cyclophosphamide 250mg/m 2 ) on day (D) -5--3, followed by infusion of CAR T-cells at 1×10 6 cells/kg on D0 and D1, respectively.Fifty-six patients successfully received infusion and followup visits.Patients in the clinical trial signed informed consent consistent with the Declaration of Helsinki.

Grading of the coagulation parameters
The grading of thrombocytopenia, hypofibrinogenemia, increased INR, and prolonged APTT was according to Common Terminology Criteria for Adverse Events, version 5.0.The rating assessment of D-dimer, TT, PT, ATIII, and FDP refers to the classification as previously used (supplemental Table 1) [13].

Bleeding events and toxicity assessment
The grading of bleeding events was according to the World Health Organization (WHO) bleeding criteria [28].The individual grades were defined in order: grade 0, no bleeding; grade 1, petechiae; grade 2, mild blood loss; grade 3, gross blood loss; grade 4, debilitating blood loss.The evaluation of CRS and ICANS was according to the American Society for Transplantation and Cellular Therapy guidelines [29].Management of adverse events was based on guidelines and institutional experience [4,30,31].

Statistical analysis
For uni-variate analysis, χ 2 test or Fisher's exact test was used for categorical data, and the Mann-Whitney U test was used for continuous variants.The before-and-after analysis of the same variable in the same patient was performed using a paired Wilcoxon test.Spearman correlation coefficients were used for correlations between continuous variables.The EASIX score was calculated as previously described [22].The significance of cumulative event rates for ranked data was comparatively analyzed using the log rank test.Uni-and multi-variate analysis of categorical variables employed logistic regression.The cutoff values of risk factors were obtained by receiver operating characteristic (ROC) curves.P < 0.05 was considered statistically significant.Statistical analysis and figure drawing were generated by IBM SPSS statistics 26, and GraphPad Prism 8.

Patient characteristics
A total of 56 patients were included in the analysis, with a median age of 44 (range, 13-74) years and 55.36% of male.Nineteen patients (33.93%) were diagnosed with B-ALL and 37 (66.07%) with NHL, including diffuse large B cell lymphoma, follicular lymphoma, mantle cell lymphoma, and B-cell lymphoblastic lymphoma.All patients received humanized anti-CD19 CAR T-cells.The median lines of previous treatment were 3 (range, 1-9), and 6 patients had received transplantation.The incidences of CRS and ICANS were 51.78% (n=29) and 7.14% (n=4), respectively.The incidences of severe (grade 3-4) CRS and ICANS were 10.91% (n=6) and 1.82% (n=1), respectively.All patient characteristics between bleeding and non-bleeding group were comparable (Table 1).
Moreover, we analyzed the association between coagulation dysfunction and the development of CRS.In our study, patients with severe CRS had more dramatic changes in coagulation than those without or mild CRS (grade 0-2), especially on D14 (supplemental Figure 3 and 4).

Cumulative incidence of bleeding events
The cumulative incidence of bleeding events within one month after infusion was 32.8%, with a median onset of 7 (range, 0-28) days (Fig. 2A).Based on the WHO bleeding Grade of FDP elevation on post-infusion day D A criteria, all bleeding events were grade 1-3 (Table 2).Among them, the incidence of grade 2-3 bleeding events was 12.5%.Based on the dynamic changes of coagulation indicators after CAR T-cell infusion, we found that the correlation between the degree of abnormal PLTs on D7 or D-dimer on D14 and the cumulative incidence of bleeding events was statistically significant (Fig. 2B-C).However, we did not find a significant effect of CRS grades and bone-marrow tumor burden levels on the occurrence of bleeding events (supplemental Figure 5).7 listed pre-LD, peak, and/or minimum values of monocyte, coagulation-, inflammation-, and tumor burden-related markers for patients with or without bleeding events.Patients with bleeding events had higher PT, IL-6, IL-10, and lower PLTs before lymphodepletion than those without bleeding events.Patients with bleeding events also had a higher peak of D-dimer, IL-2, IL-10, and lower monocyte nadir within one month.

Discussion
CARAC is a frequent adverse event during CAR T-cell treatment.Its occurrence is associated with myelosuppression, thrombocytopenia, and inflammation/tumor-mediated endothelial stress or injury [32].Severe coagulopathy can lead to bleeding or thrombotic events, which can be fatal [11][12][13].In the study, we analyzed dynamic changes in coagulation indicators, the cumulative incidence of bleeding events, correlations among coagulation, inflammation, and tumor burden in a small cohort of 56 patients.Finally, we screened two statistically significant variates, IL-10 and the EASIX score, and constructed a risk-stratification model for predicting bleeding events after CAR T-cell therapy.
We comprehensively analyzed the dynamic changes of routine coagulation indicators from pre-LD to M9 after CAR T-cell treatment.Long-term follow-up results showed that abnormalities in coagulation function mainly occurred within one month and were alleviated about 3 months after treatment.Thrombocytopenia occurred frequently and severely on D14 and gradually recovered thereafter.The abnormal proportions of APTT, FDP, ATIII, TT, and INR were relatively small, but there were obvious abnormal peaks on D7 and D14, which was consistent with previous studies [8,10].In addition, the severity of PLTs and D-dimer fluctuation on D7 and D14 is correlated to the occurrence of bleeding events.Many previous studies had also used D-dimer as a marker for predicting bleeding events [18,33].These findings prompted us to closely monitor the coagulation function of patients within one month after infusion, especially within the first two weeks.We analyzed pre-LD, peak, and/or minimum clinical indicators within one month post infusion between bleeding and non-bleeding groups.Patients in bleeding group had lower pre-LD PLTs than those in non-bleeding group (P = 0.022), consistent with the previous study [13].Bonemarrow infiltration or impaired hematopoietic function due to multi-lines of previous therapies might be responsible for pre-LD thrombocytopenia, as well as the abnormalities of PLT function [34,35].PLTs played an important role in vascular endothelial generation and endothelial integrity maintenance [36].Thrombocytopenia would hinder the differentiation of endothelial progenitor cells to mature endothelial cells and the migration of mature endothelial cells [37,38].Decreased number and impaired function of PLTs led to the reduced secretion of proangiogenic cytokines and growth factors, further resulted in molecular degradation between adjacent endothelial cells and disruption of endotheliocyte homeostasis and integrity [39].Petechiae in some patients formed when red blood cells leaked from the destroyed endothelial gap to the interstitial space.Moreover, endothelial homeostasis imbalances lead to disturbances of the endogenous and exogenous coagulation systems.Furthermore, von Willebrand factor (vWF), tissue factor, and e-selectin are markers of endothelial activation, which can be included in subsequent studies [40,41].We also observed lower monocyte nadir in patients with bleeding, as mentioned by Johnsrud et al [13].Strati et al. [42] illustrated that patient with grade 3-4 ICANS were characterized by lower monocytes perhaps due to monocytes migration from the blood to central nervous tissue and mediating inflammatory effects.Similarly, monocytes might exude from the vascular endothelial space along with red blood cells when bleeding events occurred.Studies had demonstrated that clotting, inflammation, and tumors interact with each other [13,[17][18][19].The conclusions of our study also supported the above point of view.Endothelial cell activation mediated by tumor cells could induce the transformation of cell surfaces into adherent and pro-coagulable surfaces, which in turn affected the homeostasis of the coagulation system [43].Some inflammatory molecules could activate the endothelium and promote coagulation, and some factors in the coagulation system were also effective inflammatory mediators.Moreover, CRS and ICANS during CAR T-cell treatment were systemic inflammatory responses, and their occurrence had been shown to affect coagulation function [10].However, CRS and ICANS were not inevitable factors for bleeding.We did not find differences in the incidence of CRS and ICANS between bleeders and non-bleeders.The study by Johnsrud et al. found that the incidence of ICANS was significantly higher in the bleeding group than in the non-bleeding group in uni-variate analysis, while CRS was not different between the two groups.However, there was no difference in ICANS between the two groups after multi-variate analysis, indicating that CRS and ICANS were not independent risk factors for hemorrhage [13].Therefore, it is necessary to pay attention to the inflammation, tumor burden, and coagulation function, thus to adopt treatment strategies in managing coagulation disorders.Several variants of EASIX had been created to predict CRS and ICANS in CAR T-cell therapy, including simplified EASIX (s-EASIX, LDH/ PLT), modified EASIX (m-EASIX, LDH × CRP /PLT), or combined with ferritin and/or CRS for risk stratification [23][24][25].Further studies needed to be explored to create more preferable EASIX score for the prediction of coagulation and hemorrhage.When analyzing high-risk factors for bleeding events, the uni-variate analysis included major coagulation-, inflammatory-, and tumor burden-related parameters.However, only the EASIX score and IL-10 were significant in the multi-variate analysis model.The EASIX, defined as LDH × creatinine / PLT, indicated the endothelial activation and stress [22].A study had shown that hypercoagulation status characterized by significantly increased factor VIII, factor XI, vWF antigen, vWF ristocetin, ATIII, and FIB was associated with endothelial cell activation, irrespective of disease state [44].In addition, disruption of endothelial integrity and loss of vasoconstriction can lead to hemorrhage.Patients with EASIX scores higher than 7.65 and IL-10 greater than 7.98 pg/ml before lymphodepletion were at higher risk of hemorrhage, and their coagulation function during treatment should be closely monitored and preventive measures should be taken in a timely manner to prevent serious bleeding events.IL-10 was an anti-inflammatory cytokine that inhibits inflammation and promotes repair.Elevated IL-10 might represent imbalances of the immune system and inflammatory system, acting as an "alarm signal."Enhanced IL-10 promoted apoptosis by inhibiting intravenous endothelial proliferation via the STAT3 pathway [45].However, IL-10 had also been reported to promote endothelial progenitor cell infiltration and wound healing via the STAT3 pathway [46].Perhaps the effect of IL-10 is affected by concentration.In patients with severe COVID-19, IL-10 was negatively correlated with vWF, which indicated endothelium dysfunction [41].The current studies have mostly focused on IL-6 [47], and the role of IL-10 in endothelial homeostasis and coagulation disorders needed further investigation.
Compared with other predictive models, the inclusion of the EASIX score and IL-10, which are routine clinical monitoring indicators, has improved clinical practicality.Moreover, we included the EASIX for the first time in the analysis of hemorrhage after CAR T-cell therapy.However, this study was based on a single-center small cohort and needed to be validated in a larger population.

Conclusion
In summary, we described the changes in coagulation function within one month after CAR T-cell infusion.By multivariate logistic regression, we finally included the EASIX score and IL-10 as statistically significant predictors to set a risk-stratification model for hemorrhage after CAR T-cell therapy.Risk factors for bleeding events should be further prospectively explored and strategies for their prevention and management are needed to be developed.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material.If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Fig. 2
Fig. 2 Cumulative incidence of bleeding event after CAR T-cell therapy.A The cumulative incidence of bleeding events within one month after CAR T-cell infusion was 32.8%.One patient had three bleeding episodes from different sites over 30 days.B-C Cumulative incidence of bleeding events by grade of thrombocytopenia on day 7 (B) and D-dimer elevation on day 14 (C).*P < 0.05.***P < 0.001.ns no significance, CAR chimeric antigen receptor, BE bleeding events 0 1 0 2 0 3 0 0

Fig. 3
Fig.3 Correlations among coagulation-, inflammation-, and tumor burden-related markers.Spearman correlation coefficients were shown.P values for each correlation coefficient could be found in supplemental Table8

Table 1
Patient characteristics ALL acute lymphocytic leukemia, NHL non-Hodgkin lymphoma, CRS cytokine release syndrome, ICANS immune-effector cell-associated neurotoxicity syndrome

Table 2
Bleeding events within one month

Table 3
Univariate and multi-variate logistic regression analysis of bleeding events Bold indicated statistically significant results.P values in bold were less than 0.050 OR odds ratio, LD lymphodepletion, CRP C-reactive protein, IL interleukin, EASIX endothelial activation and stress index, PLT platelet, TNFα tumor necrosis factor α, CRS cytokine release syndrome Marker Subgroup No. case No. non-case Unadjusted OR (95% CI) P value Adjusted OR (95% CI) P value