Features and Factors Associated With Myeloid Neoplasms After Chimeric Antigen Receptor T-Cell Therapy

Abstract

This case-control study examines the incidence and risks of myeloid neoplasms in adults treated for B-cell lymphoproliferative disorders or multiple myeloma.

Impressive clinical trial outcomes led to US Food and Drug Administration approval of chimeric antigen receptor T-cell therapy (CART) targeting CD19 for B-cell lymphoproliferative disorders (LPDs) and B-cell maturation antigen for multiple myeloma (MM). The most common CART toxic effect, immune effector cell–associated hematotoxicity (ICAHT), generally resolves by month 3. Myeloid neoplasms occurring after CART (post-CART MN) are also recognized outcomes but with short latency. To inform counseling, risk stratification, and potential surveillance strategies, we report updated incidence and factors associated with post-CART MN events.

Methods

We identified adults who received CART between June 1, 2016, and December 31, 2021, for LPD or MM and later developed MN at Mayo Clinic in Minnesota, Florida, and Arizona. Myeloid neoplasm classification was based on the World Health Organization fifth-edition classification. To identify clinical factors associated with post-CART MN, we defined a 6:1 primary diagnosis (MM or LPD)–matched control cohort who received CART without developing subsequent MN. CAR-HEMATOTOX Score (CHS) was calculated and treated as an independent variable in multivariate models (eMethods in Supplement 1). The Mayo Clinic Institutional Review Board approved this case-control study and waived informed consent because the research involved minimal risk. We followed the STROBE reporting guideline.

Two-sided P < .05 indicated statistical significance. Data analysis was performed using BlueSky Statistics 7.3 and GraphPad Prism 9.0.

Results

Twenty patients (median [IQR] age, 67 [52-76] years; 8 males [40%], 12 females [60%]) developed post-CART MN at a median (IQR) 10 (5-18) months after CART infusion (median [IQR] follow-up, 14 [6-23] months). Estimated cumulative incidence of post-CART MN at 1, 2, and 3 years was 4%, 6%, and 9%, respectively, and was comparable between LPD and MM (Figure, A). Next-generation-sequencing analysis detected clonal hematopoiesis (variant allele frequency [VAF] ≥2%) in 7 of 11 cases (64%) with baseline samples, predominantly involving TP53 and PPM1D. Clonal link could be established for 7 of 10 evaluable cases (70%) with paired samples, although in 3 cases the baseline clone was not clinically reportable (VAF <2%).

Figure. Cumulative Incidence and Prediction Models for Post–Chimeric Antigen Receptor T-Cell Therapy (CART) Myeloid Neoplasm (MN).

CHS indicates CAR-HEMATOTOX score; LPD, lymphoproliferative disorder; MM, multiple myeloma.

Compared with cases, control patients (n = 120) were younger (median [IQR] age, 61 [21-81] years; 72 males [60%], 48 females [40%]). Univariate cause-specific hazard regression found older age, lower hemoglobin level, and lower platelet count were associated with post-CART MN (Table). Multivariate cause-specific hazard regression compared pairs of variables, with model B (age ≥65 years, platelet count ≤140 000/μL) favored by concordance index (0.776). Applied to patients receiving commercial CART, model B had post-CART MN rates of 4 per 100 person-years (PYs) or 27 per 100 PYs in the presence of 1 factor (intermediate risk; 48% of cohort) or 2 factors (high risk; 16% of cohort), respectively (P < .001) (Figure, B). No post-CART MN events occurred in the low-risk category (43% of cohort), with over 80 PYs at follow-up. Although CHS as a categorical variable (Figure, C) was associated with post-CART MN, this association was enhanced by incorporating age (Figure, D).

Table. Univariate and Multivariate Hazard Regression Analysis of Factors Associated With Myeloid Neoplasm After Chimeric Antigen Receptor T-Cell Therapy .

Variables	HR (95% CI)	P value	Concordance index	Optimal threshold
Univariate analysis: continuous variables
Age	1.06 (1.01-1.12)	.02a	0.688	≥64.0
Hemoglobin level	0.74 (0.56-0.98)	.04a	0.630	≤9.3
Platelet count	0.99 (0.99-1.00)	.04a	0.661	≤141.0
ANC	0.85 (0.63-1.15)	.29	0.544	NA
CRP	0.99 (0.98-1.01)	.48	0.525	NA
MCV	1.03 (0.97-1.09)	.33	0.537	NA
RDW	1.11 (0.91-1.35)	.31	0.564	NA
Ferritin	1.00 (0.99-1.00)	.15	0.610	NA
Prior lines of therapy	1.01 (0.84-1.23)	.09	0.554	NA
Univariate analysis: categorical and transformed variables
ASCT: yes vs no	1.22 (0.48-3.06)	.67	0.551	NA
Marrow involvement: MM or LPD	1.17 (0.48-2.83)	.72	0.506	NA
Age ≥65y vs <65 y	4.99 (1.81-13.76)	.002a	0.714	NA
Hemoglobin level ≤9.0 vs >9 g/dL	2.96 (1.21-7.25)	.02a	0.628	NA
Platelet count ≤140 000 vs >140 000 per μL	3.69 (1.34-10.18)	.01a	0.651	NA
CHS
Continuous	1.38 (1.05-1.80)	.02a	0.660	2.0
≥2	3.04 (1.24-7.44)	.02a	0.614	NA
Pairwise multivariate analysis: individual variables
Model A
Age ≥65 y	4.58 (1.66-12.65)	.003a	0.776	NA
Hemoglobin level ≤9.0 g/dL	2.60 (1.05-6.40)	.04a		NA
Model B
Age ≥65 y	4.26 (1.53-11.81)	.005a	0.776	NA
Platelet count ≤140 000 per μL	2.98 (1.07-8.30)	.04a		NA
Model C
Hemoglobin level ≤9.0 g/dL	2.28 (0.91-5.73)	.08	0.695	NA
Platelet count ≤140 000 per μL	3.02 (1.07-8.54)	.04a		NA
Model D
CHS ≥2	2.94 (1.20-7.21)	.02a	0.772	NA
Age ≥65 y	4.86 (1.77-13.43)	.002a		NA

Abbreviations: ANC, absolute neutrophil count; ASCT, autologous stem cell transplant; CHS, CAR-HEMATOTOX score; CRP, C-reactive protein; HR, hazard ratio; LPD, lymphoproliferative disorder; MCV, mean corpuscular volume; MM, multiple myeloma; NA, not applicable; RDW, red cell distribution width.

^a Indicates statistical significance.

Discussion

Intuitive baseline factors of age and thrombocytopenia can stratify MN risks occurring in patients after CART, regardless of clonal hematopoiesis, supporting counseling and guiding surveillance strategies. Although higher than posttransplant rates for MM and LPD, post-CART MN events could not be directly attributed to CART exposure because all patients received prior cytotoxic therapies. Next-generation-sequencing screening before CART remains investigational; half of cases with detected clonal link had baseline clones below reporting limits. Instead, Model B is a pragmatic surrogate of increased risk. Post-CART MN is a consequential differential diagnosis of ICAHT, and an association with CHS was enhanced by a model incorporating age.

Study limitations include lack of independent cohort validation. Additionally, the number of post-CART MN events limited multivariate models to pairs of factors, and incomplete data for clonal hematopoiesis prohibited inclusion in models. Furthermore, hazard ratios derived from controls reflect an enriched sample and are not generalizable. Instead, incidence rates calculated from patients receiving commercial CART are informative. These observations require confirmation in larger multicenter or registry-based studies.

Supplement 1.

eMethods. Cohort Definitions and Statistical Analysis

Supplement 2.

Data Sharing Statement