This study was a retrospective analysis that consecutively enrolled 197 R/R MM patients treated at the Affiliated Hospital of Xuzhou Medical University between June 2019 and February 2024. All patients received BCMA-targeted CAR-T cell therapy, specifically involving either (1): single-target BCMA CAR-T cells, or (2) sequential infusion of BCMA and CD19 CAR-T cells (BCMA+CD19). All CAR−T products were autologous and manufactured in−house. Briefly, peripheral blood mononuclear cells were obtained from subjects by leukapheresis, and CD3⁺ T lymphocytes were isolated using anti−CD3 immunomagnetic beads. A lentiviral vector encoding the CAR construct (anti−BCMA single−chain variable fragment [scFv]/4−1BB/CD3ζ; humanized anti−CD19 scFv/4−1BB/CD3ζ; or a bispecific anti−BCMA/anti−CD19 scFv/4−1BB/CD3ζ) was transduced into CD3⁺ T cells and expanded in vitro. CAR−T cells were harvested after 10-14 days and cryopreserved until infusion. The study was approved by the hospital’s Ethics Committee and conducted in accordance with the principles of the Declaration of Helsinki. All participants provided written informed consent. The trials were registered at the Chinese Clinical Trial Registry (ChiCTR.org.cn) with registration numbers ChiCTR2000033567, ChiCTR-OIC-17011272, and ChiCTR1900026219. The detailed inclusion and exclusion criteria were consistent with previously published literature (22, 23). All patients received lymphodepleting conditioning with fludarabine (30 mg/m²/day, days -5 to -3) and cyclophosphamide (750 mg/m²/day, day -5) prior to CAR-T infusion.

Patient clinical characteristics at enrollment were collected, including age, sex, MM type, prior treatment history, and cytogenetic abnormalities. Laboratory data were obtained from the electronic medical record system. Baseline complete blood count, lactate dehydrogenase (LDH), and β2-microglobulin (β2-MG) were defined as the most recent test results within 15 days before lymphodepletion. Peak levels of ferritin, C-reactive protein (CRP), interleukin-6 (IL-6) and CAR transgene were recorded from tests within the first month after CAR-T cell infusion.

The neutrophil-to-lymphocyte ratio (NLR), monocyte-to-lymphocyte ratio (MLR), and platelet-to-lymphocyte ratio (PLR) were calculated based on complete blood count results. The optimal cut-off values for these indices were determined using receiver operating characteristic (ROC) curve analysis, maximizing the Youden index. Patients were stratified into high and low groups based on these cut-offs.

The severity of CRS and ICANS was retrospectively graded according to the American Society for Transplantation and Cellular Therapy (ASTCT) consensus criteria (24). Response to CAR-T cell therapy in R/R MM patients was assessed based on the International Myeloma Working Group (IMWG) uniform response criteria updated in 2014 (25). PFS was defined as the time from CAR-T cell infusion to disease progression or death from any cause. The objective response rate (ORR) was defined as the proportion of patients achieving a partial response (PR) or better. To address potential confounding effects on baseline inflammatory ratios, we performed a sensitivity analysis incorporating two clinically relevant variables: previous infections, defined as a documented bacterial or viral infection within 30 days prior to lymphodepletion, and previous corticosteroid use, defined as administration of systemic corticosteroids within 7 days before the baseline complete blood count. All patients were followed until June 30, 2024, with follow-up data sourced from inpatient/outpatient medical records and telephone follow-ups. The median follow−up for the entire cohort was 26.2 months (IQR: 16.5-38.4).

Statistical analyses were performed using GraphPad Prism v8.0 software. Continuous variables are presented as median (range), and comparisons between groups were made using the Mann-Whitney U test or t-test. Categorical variables are presented as frequency (percentage), and comparisons between groups were made using the χ² test or Fisher’s exact test. ROC curves were used to determine the optimal cut-off values for NLR, MLR, and PLR in predicting PFS, along with the area under the curve (AUC). Survival analysis was performed using the Kaplan-Meier method, and group comparisons were made using the log-rank test. Univariate and multivariate Cox proportional hazards regression models were used to analyze independent prognostic factors for PFS. All statistical tests were two-sided, with a p-value < 0.05 considered statistically significant.

This study included 197 R/R MM patients who received CAR-T cell therapy, with a median age of 58 years (range: 34-78), including 109 males (55.3%). Baseline clinical characteristics, including R-ISS stage, prior treatment history, and disease burden, are detailed in Table 1. ROC analysis determined the optimal cut-off values for NLR, MLR, and PLR, to be 2.55, 0.35,and 145, respectively. The AUC for NLR was 0.801 (95% CI: 0.732-0.870, p <0.001), with a sensitivity of 82.1% and specificity of 74.5%. The AUC for MLR was 0.763 (95% CI: 0.688-0.838, p < 0.001), with a sensitivity of 78.3% and specificity of 70.2%. The AUC for PLR was 0.735 (95% CI: 0.674-0.836, p < 0.001), with a sensitivity of 76.5% and specificity of 68.1%. These results suggest that all three ratios possess good discriminatory ability (Supplementary Figure S1).

Patients were stratified into high and low groups based on these cut-offs for baseline comparison. Results showed that patients in the high NLR and high MLR groups more frequently had high tumor burden (NLR: 34.1% vs. 16.9%, p = 0.007; MLR: 33.8% vs. 18.3%, p = 0.017) than that of patients with low NLR and MLR groups, while the incidence of extramedullary disease was significantly higher in the high PLR group compared to the low PLR group (49.4% vs. 21.9%, p < 0.001). Additionally, LDH levels were significantly higher in the high NLR group compared with the low NLR group (225 U/L vs. 198 U/L, p = 0.015) (Table 1). These results indicate that higher peripheral blood inflammatory ratios may be closely associated with aggressive disease features.

During dynamic monitoring post-T cell infusion, significant differences in the peak levels of CAR transgene were observed between different ratio groups. The peak levels of CAR transgene were significantly higher in the low NLR group compared to the high NLR group (p < 0.001). Similarly, the low MLR group exhibited a higher peak level of CAR transgene (p = 0.012), while no significant difference was observed between PLR groups (p = 0.109) (Figures 1A–C). This finding suggests that low peripheral blood inflammatory ratios may facilitate in vivo expansion of CAR-T cells.

The overall ORR for CAR-T therapy was 89.8%. Analysis by NLR groups indicated that in the low NLR group (≤ 2.55), 105 patients (93.8%) achieved a response, comprising 43 (38.4%) with stringent complete response (sCR), 21 (18.7%) with complete response (CR), 30 (26.8%) with very good partial response (VGPR), and 11 (9.8%) with partial response (PR). In contrast, the ORR in the high NLR group (> 2.55) was 81.2%, comprising 24 (27.8%) sCR, 14 (16.6%) CR, 19 (22.0%) VGPR, and 12 (14.1%) PR. The difference in ORR between these two groups was statistically significant (p = 0.037) (Figure 2A).

In the MLR groups, the ORR for the low MLR group (≤ 0.35) was 92.5%, comprising 38 (31.7%) sCR, 21 (17.5%) CR, 35 (29.2%) VGPR, and 17 (14.1%) PR. The ORR for the high MLR group (> 0.35) was 87.0%, comprising 22 (28.6%) sCR, 12 (15.6%) CR, 20 (25.9%) VGPR, and 13 (16.9%) PR. The difference in ORR between MLR groups had no statistical significance (p = 0.155) (Figure 2B).

In the PLR groups, the ORR for the low PLR group (≤ 145) was 91.2%, comprising 35 (30.7%) sCR, 20 (17.5%) CR, 33 (28.9%) VGPR, and 16 (14.1%) PR. The ORR for the high PLR group (> 145) was 85.5%, comprising 22 (26.5%) sCR, 16 (19.3%) CR, 20 (24.1%) VGPR, and 13 (15.7%) PR. The difference in ORR between PLR groups had no statistical significance (p = 0.206) (Figure 2C).

Overall, patients in the low NLR group not only had a higher ORR but also a significantly higher rate of deep responses (sCR + CR) compared with the high NLR group (57.1% vs. 44.4%, p = 0.019), suggesting that low peripheral blood inflammatory ratios may be associated with a strong CAR-T cell response and thorough tumor clearance.

CRS and ICANS are unique inflammation-related complications of CAR-T cell therapy. In this study, the incidence of severe CRS (≥ grade 3) was significantly higher in the high NLR group compared with the low NLR group (20.0% vs. 7.1%, p = 0.009). No statistically significant differences were observed between MLR groups (p = 0.396) or PLR groups (p = 0.081) (Table 2). Analysis of inflammatory cytokines revealed that the peak levels of IL-6 and ferritin were significantly elevated in the high NLR, MLR, and PLR groups (all p < 0.05) (Supplementary Table S2). Furthermore, NLR displayed a significant positive correlation with both the peak levels of ferritin and CRP (r = 0.295, p = 0.015) (Supplementary Table S3). Although the peak levels of CRP tended to be higher in the high inflammatory ratio groups than that of the low inflammatory ratio groups, the inter-group difference was not statistically significant (p > 0.05). These results suggest that the high peripheral blood inflammatory ratios are closely relationship with the severity of CRS and the level of systemic inflammatory response.

A total of 8 patients (4.1%) experienced ICANS in the entire cohort, including 3 cases (1.5%) of ≥ grade 3 ICANS. No significant difference in ICANS incidence was observed across NLR (p = 0.539), MLR (p > 0.999), or PLR (p = 0.536) groups (Supplementary Table S1).

Kaplan-Meier survival analysis showed that the PFS was significantly longer in the low NLR group compared to the high NLR group (median PFS: 18.6 months vs. 10.9 months, HR = 2.412, p = 0.0012) (Figure 3A). Similarly, the low MLR group exhibited longer PFS (median PFS: 16.3 months vs. 11.2 months, HR = 2.135, p = 0.0086) than that of the high MLR group (Figure 3B). Furthermore, PFS was also better in the low PLR group compared to the high PLR group (median PFS: 16.5 months vs. 13.4 months, HR = 1.825, p = 0.025) (Figure 3C).

Based on the hazard ratios (HR) for NLR, MLR, and PLR groups, a Cellular Inflammatory Prognostic Index (CIPI) was constructed: both the high NLR and MLR were assigned 1 point, but the high PLR was assigned 0.5 point. Patients were categorized into three risk groups based on CIPI score: low-risk (0-0.5 points, 98 patients, 49.7%), intermediate-risk (1-1.5 points, 67 patients, 34.0%), and high-risk (2-2.5 points, 32 patients, 16.2%). PFS differed significantly among these three groups: median PFS was 18.9 months (95% CI: 13.8-25.1 months) for the low-risk group, 13.8 months (95% CI: 10.2-17.4 months) for the intermediate-risk group, and 5.1 months (95% CI: 2.7-7.5 months) for the high-risk group (p < 0.0001) (Figure 4).

To further assess whether CIPI is an independent prognostic factor for PFS, variables including age, sex, extramedullary disease, number of prior lines of therapy, MM subtype, high-risk cytogenetic status, R-ISS stage, tumor burden, previous infections, previous corticosteroid use, and CIPI score were incorporated into a Cox proportional hazards model. Univariate analysis identified R-ISS stage (p = 0.036), high tumor burden (p = 0.015), extramedullary lesions (p = 0.028), LDH level (p = 0.043), previous infections (p = 0.024), and CIPI score (p < 0.001) as the significant factors affecting PFS. These significant variables were subsequently included in the multivariate analysis, which revealed that the high tumor burden (p = 0.041) and CIPI score (p = 0.013) were independent risk factors for PFS in R/R MM patients received CAR-T cell therapy (Table 3).