As of December 15, 2023, 189 RRMM patients underwent leukapheresis with intent to manufacture commercial CAR T. A total of 152 patients proceeded with infusion, 14 patients did not receive CAR T because of disease progression/death, and 23 were pending infusion at data cut-off. Of the 152 patients who received CAR T and were included in this analysis, 108 (71%) received standard of care (SOC) ide-cel, and 44 (29%) received SOC cilta-cel. Forty-seven (31%) patients had active EMD before the CAR T-cell infusion. Of the remaining 105 (69%) patients (non-EMD group), 26 (17%) had PMD-alone, and 79 (52%) had neither EMD nor PMD (bone marrow-contained group or BM-only group). Patients had a median of 6 (range 4–15) prior lines of therapy, and 85% had triple-class refractory disease.

Baseline patient characteristics are presented in Table 1, stratified by the groups who had active EMD (n = 47) (EMD group) prior to infusion compared to those who did not (n = 105) (non-EMD group). The median age of the EMD group was 60 (range 43–78) years, 19% had an Eastern Cooperative Oncology Group (ECOG) performance status (PS) ≥ 2, 43% had R-ISS stage III disease prior to infusion, and 35% had high-risk cytogenetics. Thirty-six (77%) patients received ide-cel and 11 (23%) patients received cilta-cel in the EMD group. The median age for patients in the EMD group was 60 years, as compared to 65 years for the non-EMD group (p = 0.0125). Otherwise groups were well matched (Table 1).

Regarding the specific location of EMD before CAR T infusion, 25 (53%) out of 47 EMD patients had visceral EMD. The most commonly involved organs are described in Supplementary Table 1. Twenty-one (45%) of EMD patients had cutaneous involvement, and seven (15%) had lymph node involvement. Nine (19%) patients had only one EMD lesion, nine (19%) patients had two EMD lesions, and the remaining 29 (62%) patients had ≥3 EMD lesions. Notably, six (17%) patients had at least one lesion measuring greater than 5 cm in longest perpendicular diameter, and 17 (36%) patients had received radiation therapy for EMD prior to CAR T.

Adverse events for both groups are summarized in Table 2. The median duration of hospitalization for the EMD group was 12 days (range 5–50), and 9 patients (19%) required intensive care unit stay during the hospitalization. The incidence of all grade CRS for the EMD group was 81%, with a median time to maximum grade CRS of four days. The incidence of all grade ICANS was 36% in the EMD group, with 21% having grade 1, 9% having grade 2, and 6% having grade 4 events. While five (11%) and three (3%) patients in the EMD and non-EMD groups experienced grade 3-4 CRS events, respectively, rates of tocilizumab use did not differ between the two groups. However, the use of steroids (40% vs 24%, p = 0.03) and anakinra (21% vs 3%, p = 0.0002) was more frequent in the EMD group. When focusing on the EMD group only, the rate of any grade (grade ≥ 3) CRS was 78% (3%) vs 91% (36%) in patients who received ide-cel versus cilta-cel, respectively. Likewise, the rate of any grade (grade ≥ 3) ICANS was 36% (0%) vs 36% (27%) in patients who received ide-cel versus cilta-cel, respectively.

At day +30 post CAR T infusion, grade ≥3 neutropenia, anemia, and thrombocytopenia were noted in 48%, 21%, and 62% of the patients in the EMD group, respectively. Compared to the non-EMD group, there was a higher incidence of any grade (95% vs 74%, p = 0.0006) and grade ≥3 thrombocytopenia (62% vs 38%, p = 0.0087) in the EMD group which translated to a higher rate of thrombopoietin agonist use (38% vs 15%; p = 0.02) for the EMD group. The incidence of documented infections (all grades) was similar between the two groups, 43% in the EMD group vs 30% in the non-EMD group.

The hematologic, radiographic, and combined hematologic and radiographic response rates for patients in the EMD group are shown in Fig. 1. Five out of the 47 patients were non-evaluable for combined overall response assessment since radiographic imaging was not obtained post infusion; these were considered as non-responders in the combined overall response analysis. In addition, patients who died within the first 30 days post infusion prior to response assessment because of toxicity were included in the response analysis and were considered as non-responders as well. Further details regarding hematologic and radiographic responses at days +30, +90, and +180 post CAR T for the EMD group are shown in Fig. 1A, B. For patients with EMD, the best combined ORR for those who received ide-cel (n = 36) and cilta-cel (n = 11) were 61% and 46%, respectively. Notably, four patients with EMD who received cilta-cel died from CAR T-related toxicities < 30 days post infusion, whereas none of the EMD patients treated with ide-cel died from toxicity < 30 days post infusion. Overall, patients with EMD had a lower combined ORR (58% vs 96%, p < 0.00001) and response of ≥CR (28% vs 56%, p = 0.002) compared to the non-EMD group (Fig. 1C).

A Hematologic response rate for the EMD group. B Radiographic response rate for the EMD group. C Best overall tumor responses for the EMD (combined hematologic and radiographic response) and non-EMD groups. ORR overall response rate, CR complete response, VGPR very good partial response, PR partial response, heme hematologic, rads radiographic. * For response assessment at any time point, patients with missing data for rads or heme response, as well as deaths from toxicities/unrelated causes were considered as non-responders. Patients who died within the first 30 days post infusion prior to response assessment because of toxicities were also considered as non-responders.

When focusing on the EMD group, we explored EMD specific factors associated with radiographic response. On univariate analysis, we did not find any significant association between radiographic response rate and EMD characteristics, including number of lesions, size of lesions, presence of visceral EMD, and radiation prior to infusion.

The median duration of follow-up was 12.5 (IQR: 9.2, 25) months for the EMD group and 12.6 (IQR: 7.6, 23.1) months for the non-EMD group. Patients in the EMD group had an inferior median PFS compared to the non-EMD group (5.1 months vs 12.4 months; p < 0.0001; Fig. 2A). Similarly, patients in the EMD group had an inferior median OS compared to patients in the non-EMD group (12.2 vs 27.5 months; p = 0.00058; Fig. 2B). To better understand the impact of true EMD on efficacy post CAR T-cell therapy, we examined PFS and OS between the EMD, PMD-only, and BM-only groups. Patients with PMD-only disease had similar median PFS (11.2 vs 13.6 months, p = 0.3798; Fig. 3A) and OS (not reached [NR] vs 27.5 months, p = 0.6446; Fig. 3B) compared to patients with BM-only disease. However, patients in the EMD group exhibited significantly inferior median PFS (5.1 vs 13.6 months, p < 0.0001; Fig. 3A) and OS (12.2 vs 27.5, p = 0.0008; Fig. 3B) compared to patients in the BM-only group.

A Progression-free survival. B Overall survival.

A Progression-free survival. B Overall survival.

Multivariable analysis of the entire patient cohort identified active EMD before CAR T as an independent predictor for both inferior PFS (HR, 2.12; 95% CI, 1.26–3.6; p = 0.005) and OS (HR, 2.09; 95% CI 1.06–4.12; p = 0.033) (Table 3). In addition to active EMD, high-risk cytogenetics, prior anti-BCMA therapy, receipt of ide-cel product and high baseline ferritin were independently associated with worse PFS (Table 3); while active EMD and ECOG PS ≥ 2 were associated with worse OS. In the EMD cohort only, multivariate analysis identified ECOG PS ≥ 2, four prior lines of therapy, prior anti-BCMA therapy exposure, absence of radiation therapy prior to infusion, and lesion size of >5 cm to be associated with worse PFS (Table 4).

Of the 47 patients in the EMD group, 32 (68%) experienced disease relapse, including 12 (25.5%) who relapsed with EMD, 8 (17%) who relapsed biochemically, and 12 (25.5%) who relapsed with both EMD and biochemically. EMD at relapse was present in the same site as prior to CAR T, in a new site, and both in the same plus new sites in seven (15%), nine (19%), and eight (17%) patients, respectively. At data cutoff, 24 (51%) out of the 47 patients in the EMD group had died; 14 (30%) from progressive myeloma; four (8.5%) from CRS/hemophagocytic lymphohistiocytosis, two (4%) from infection/sepsis and four (8.5%) from unrelated causes. Among the 105 patients in the non-EMD group, 28 (27%) patients had died at data cut-off; 24 (23%) from myeloma progression, two (2%) from infection/sepsis, one (1%) from delayed neurotoxicity, and one (1%) from unrelated causes.