Discussion

In this large, nationally representative cohort, we observed DIC among 2.1% of patients undergoing CAR T-cell therapy. Previously reported rates of 7% to 28% were observed in smaller single-center studies [10]. Importantly, we identified significant demographic disparities, with patients in the lowest income quartile and those with self-pay status disproportionately represented in the DIC cohort compared with the non-DIC cohort. These findings highlight the need for equity-focused strategies to identify at-risk populations and implement targeted monitoring.

DIC was associated with markedly worse inpatient outcomes. Patients with DIC had significantly higher in-hospital mortality (22.7% vs 2.4%), and DIC was strongly associated with increased odds of mortality (aOR, 10.96). This is consistent with prior adverse event registry reports, which documented fatality rates as high as 59.6% for DIC following CAR T-cell therapy, with higher mortality observed after axicabtagene ciloleucel than after tisagenlecleucel (66.7% vs 57.3%) [9]. These findings collectively emphasize DIC as a severe and life-threatening complication of CAR T-cell therapy, reinforcing the critical need for early recognition and aggressive intervention.

Organ dysfunction and increased need for intensive care were also disproportionately observed among CAR T-cell therapy patients who developed DIC. In this group, the odds of AKI, RF, and shock were approximately 6-fold, 10-fold, and 19-fold higher, respectively. Although causality cannot be established, these findings demonstrate a strong association between DIC and organ dysfunction in this population. A study by Helms et al. [11] in patients with septic shock found that DIC was independently associated with more severe forms of AKI, even after adjusting for illness severity and other clinical factors. This included stage 3 AKI, as defined by the Kidney Disease: Improving Global Outcomes classification, as well as cases requiring RRT. Consistent with these results, our study identified ∼10-fold higher odds of RRT use among patients with DIC. They also reported that higher International Society on Thrombosis and Hemostasis (ISTH) DIC scores correlated with a greater proportion of patients with stage 3 AKI. Further supporting this link, a study by Xu et al. [12] in patients with septic shock from intra-abdominal infection found that coagulation biomarkers such as activated partial thromboplastin time, prothrombin time, and D-dimer levels measured at intensive care unit (ICU) admission were independently associated with AKI. Although, to our knowledge, no prior studies have directly linked DIC with AKI in patients receiving CAR T-cell therapy, these findings, along with those of the present study, support a potential connection. The pathophysiological features of DIC, including microvascular thrombosis and resulting organ ischemia, may help explain this association and merit further investigation.

RF is a well-recognized complication of CAR T-cell therapy [13,14]. In our study, patients who developed DIC not only faced a higher risk of RF but also demonstrated a significantly increased need for MV, with a 23-fold rise compared with those without DIC. While a definitive causal relationship cannot be established, the data reveal a strong association between DIC and RF in this patient population. Prior work in acute respiratory distress syndrome populations has shown that DIC is closely linked to multiorgan dysfunction syndrome (MODS). In this multicenter cohort, DIC independently predicted the development of MODS, with higher mortality observed among those with persistent DIC [15]. In a multicenter ICU cohort of obstetrical patients, DIC was strongly associated with organ dysfunction, with DIC patients experiencing significantly higher rates of MODS/death and increased organ failure, including respiratory, renal, and neurologic compromise, compared with those without DIC [16]. These findings underscore DIC as a critical driver of multiorgan failure across diverse clinical settings and support our observation that CAR T-cell therapy recipients who develop DIC experience disproportionately higher rates of RF and AKI, consistent with the broader pathophysiological relationship between coagulopathy and multiorgan dysfunction.

Our findings highlight that DIC was associated with significantly higher rates of thrombotic and hemorrhagic complications, as well as increased blood product utilization. This is consistent with the pathophysiology of DIC, characterized by widespread coagulation activation, consumption of clotting factors, and secondary fibrinolysis. Although VTE and ACS were more frequent in patients with DIC, these associations were not statistically significant after adjustment, echoing prior observations that clinical thrombosis does not always mirror laboratory evidence of procoagulant activity in malignancy-associated DIC [17,18]. Johnsrud et al. [8] similarly attributed post–CAR T-cell thrombotic events more to underlying malignancy or hospitalization than to DIC itself.

Hemorrhagic complications, however, showed stronger associations. GIH had a significant aOR of 12.41 in the DIC group, aligning with reports of mucosal bleeding in consumptive coagulopathy [5]. While ICH occurred more often in the DIC group, statistical significance was not achieved, likely due to limited power, paralleling findings in the (a phase 1, first-in-human study of LCAR-B38M CAR-T cells in patients with relapsed/refractory multiple myeloma) LEGEND-2 trial [6]. Additional studies have reported increased bleeding risk with high-grade CRS and hemophagocytic syndromes [1,9].

Transfusion needs were also markedly elevated in patients with DIC, with significant associations for red blood cell, platelet, and plasma transfusions even after adjustment. This mirrors findings from the study by Mei et al. [7], who reported high transfusion demand in CAR T-cell–treated patients with coagulopathy, and from that of Xia et al. [2], who identified transfusion burden as a major resource utilization driver [2]. Variable transfusion thresholds across institutions may influence both outcomes and costs. A Cochrane review by Radford et al. [19] found that restrictive transfusion strategies in hematologic malignancies reduced product use without compromising survival, although CAR T–specific data remain limited. The DESCAR-T registry showed that over half of patients with large B-cell lymphoma required transfusions postinfusion, with early transfusion associated with poorer survival and reflective of higher toxicity burden [20].

The pathogenesis of DIC in this context likely involves cytokine-driven inflammation, endothelial injury, and tissue factor expression. Song et al. [9] identified DIC as a fatal complication linked to severe CRS and macrophage activation syndrome, both associated with endothelial disruption. Dholaria et al. [21] outlined a model where cytokine-induced tissue factor expression and platelet activation promote consumptive coagulopathy. Translational studies have reinforced the role of endothelial dysfunction in these processes. Elevated angiopoietin-2 and von Willebrand factor levels correlate with CRS severity, DIC risk, and adverse outcomes. Galli et al. [22] prospectively linked angiopoietin-2 to both CRS and coagulation dysfunction, while Gavriilaki et al. [23] emphasized immune-mediated endothelial injury as a central mechanism.

In this nationally representative cohort, DIC was independently associated with increased LOS and TOTCHG. These findings are consistent with prior studies linking DIC to adverse outcomes in hematologic malignancies. Franchini et al. [24] described DIC as both a marker and mediator of complications, including hemorrhage, organ dysfunction, and delayed recovery. Similar associations have been reported in stem cell transplantation and intensive chemotherapy settings [25,26]. In the context of CAR T-cell therapy, DIC may reflect downstream effects of severe immune-mediated toxicities. Neelapu et al. [27] and Strati et al. [28] identified CRS, immune effector cell–associated neurotoxicity syndrome, and secondary hemophagocytic lymphohistiocytosis/macrophage activation syndrome as major complications associated with prolonged hospitalization [27,28]. DIC may co-occur with or follow these syndromes, driven by cytokine excess and endothelial injury. Consistent with this, Song et al. [9] demonstrated that hemophagocytic lymphohistiocytosis and DIC following CAR T-cell therapy were associated with increased ICU utilization, mortality, and LOS [9], while Johnsrud et al. [8] linked bleeding and thrombotic complications to ICU transfer and extended monitoring. The observed increase in hospital charges in patients with DIC aligns with prior studies identifying ICU-level toxicities and care intensity as major drivers of cost [29,30]. Institutional analyses further support that complications such as DIC substantially increase health care utilization [25,26], underscoring the importance of toxicity-adapted care pathways to mitigate complications and optimize resource use [31].

Study limitations

This study used data from the NIS, a nationally representative database that captures diagnoses and procedures using ICD-10 coding. While the NIS enables large-scale epidemiologic analyses, reliance on administrative codes introduces the potential for misclassification bias, as coded diagnoses may not fully or accurately reflect underlying clinical conditions, in general, resulting in possible overrepresentation or underrepresentation of certain diagnoses. Although prior validation studies suggest that several of our ICD-10–based outcome definitions demonstrate moderate to high specificity and PPV, sensitivity is variable, and true events may be missed. For instance, our identification of DIC was based on a strategy that does not fully capture the clinical or laboratory parameters required for definitive case ascertainment. This absence of laboratory values limits the ability to apply established DIC scoring systems. Validation studies have demonstrated that ICD-based identification of DIC yields only moderate PPV, despite maintaining high specificity, indicating that coded cases are generally accurate but that a substantial proportion of true DIC events remain uncaptured [32]. Therefore, the incidence observed in this analysis likely underestimates the actual burden of DIC among CAR T-cell recipients.

Another important limitation is the inability to distinguish individual CAR T-cell products or antigen targets using administrative data. CAR T-cell exposure was identified through ICD-10 procedure coding system procedure codes, which prior to October 2021 did not reliably differentiate among Food and Drug Administration–approved CAR T-cell therapies. For example, procedure codes such as XW033C3 and XW043C3 were used for multiple products, including tisagenlecleucel (Kymriah), axicabtagene ciloleucel (Yescarta), and idecabtagene vicleucel (Abecma). Consequently, stratified analyses by CAR T-cell target (eg, CD19 vs B-cell maturation antigen) or by individual product were not feasible. This distinction is clinically relevant because CAR T-cell constructs differ in biologic targets, underlying disease populations, treatment indications, and toxicity profiles, factors that may influence the risk and outcomes of DIC. Accordingly, these findings should be interpreted within the context of this coding limitation.

Moreover, the NIS is an administrative, discharge-level database that does not capture the timing or sequence of diagnoses and procedures within a hospitalization. ICD-10-clinical modification and ICD-10 procedure coding system codes lack intrahospitalization time stamps, preventing determination of the onset of DIC relative to CAR T-cell infusion, its temporal relationship with CRS, or its duration. Therefore, our findings reflect hospitalization-level associations rather than detailed clinical trajectories.

Additionally, although the multivariable models were adjusted for known confounders, the presence of residual or unmeasured confounding cannot be entirely excluded. The retrospective nature of the study further limits causal inference. Nevertheless, the inclusion of 6 years of data from the largest all-payer hospitalization database in the United States allowed for the aggregation of a substantial cohort of patients receiving CAR T-cell therapy, facilitating the investigation of DIC in this population, a task that would be challenging to achieve through prospective data collection due to the relative rarity of this complication.

Our analysis was restricted to complete cases, with exclusion of patients with missing data in key variables, including age, sex, race/ethnicity, CCI, median household income quartile based on patient ZIP code, primary payer/insurance type, and hospital teaching status/location. Although complete case analysis is widely used, it may introduce selection bias if data are not missing completely at random, thereby potentially limiting the representativeness of the study cohort. Additionally, several outcomes involved sparse event counts, which may have contributed to inflated effect estimates and wide CIs. Finally, the charge analysis did not account for expenditures related to postdischarge care, including short-term and long-term rehabilitation services, potentially leading to an underestimation of the overall health care charges associated with DIC in the context of CAR T-cell therapy.

Clinical implications and future direction

Our findings underscore the importance of early recognition and proactive monitoring of coagulation parameters in patients undergoing CAR T-cell therapy, particularly among high-risk socioeconomic groups. Routine surveillance strategies may enable timely diagnosis and intervention for DIC, potentially improving outcomes. Future research should prioritize prospective studies to elucidate the pathophysiology of CAR T-associated DIC, identify predictive biomarkers, and guide the development of risk-stratified monitoring protocols. Evaluating the safety and efficacy of thromboprophylaxis and transfusion strategies tailored to this unique population also represents a critical area for future investigation.