We reviewed published data available through PubMed search from pivotal trials and real-world experiences with anti-CD19 CART (CART19) for adults with lymphoma. We focused our review on CART19 for lymphoma, as these are the longest available products, since 2017, and have the longest follow-up and a lower rate of relapse and/or post-CART as allogenic transplant consolidations compared to other diseases and/or products.

We defined late events as occurring or persisting beyond 1 month after CART infusion. We focused our literature review on the following late-event outcomes post-CART19: cytopenia, immune reconstitution, infections, and subsequent malignancies.

The terminology for cytopenia after CAR T-cell therapy varies among studies. Usually, it is called early cytopenia for those occurring up to 30 days, “short-term” for those occurring between 30 and 90 days, and persistent or late for those occurring after 90 days (17–19). The term prolonged cytopenia was used for cytopenia beyond 21, 28, 30, and 90 days. Delayed cytopenia is used for those occurring after an initial recovery (19). Overall, grade 3–4 cytopenia beyond 30 days occurs in 30%–40% of patients and beyond 90 days in 3%–22% of patients, with neutropenia and thrombocytopenia more common than anemia (18).

In the pivotal clinical trials of third-line (3L) CART19 for relapsed/refractory (R/R) large B-cell lymphoma (LBCL), the prevalence of grade 3/4 (G3/4) cytopenia beyond day 90 (D90) ranged from 17% to 38% (2–4, 20–22). When looking at the real-world evidence (RWE) data of 3L CART19 for R/R LBCL through the Center for International Blood and Marrow Transplant Research (CIBMTR) registry studies, the prevalence of grade 4 neutropenia and/or thrombocytopenia at day 30 (D30) ranged from 10% to 22%. Among the second-line (2L) CART19 trials for R/R LBCL, the prevalence of G3/4 cytopenia at D35 was 43% with liso-cel and 29% at D30 with axi-cel, but at 6 months, the prevalence was only 5% with axi-cel and not available for liso-cel ( Table 1 ) (5, 6, 26).

The prevalence of G3/4 cytopenia at D90 was 26% in the pivotal trial of brexucabtagene autoleucel (brexu-cel) for R/R mantle cell lymphoma (MCL), and in the RWE, the prevalence of G3/4 neutropenia and thrombocytopenia at D90 was 18% and 11%, respectively. In pivotal trials of CART19 for R/R follicular lymphoma (FL), the prevalence of D90 G3/4 cytopenia was 6% with axi-cel, and the probability of G3/4 cytopenia at 6 months was 18% with tisa-cel ( Table 1 ) (1, 7, 11, 23–25, 30).

Because of the significant variability of how post-CART cytopenia is reported within these different studies, it is not possible to draw a conclusion if cytopenia is more prevalent in the real-world vs. pivotal trials, 2L vs. 3L products, or between different products. However, some studies show a trend for a higher incidence of late cytopenia with axi-cel compared to tisa-cel. In the French registry DESCAR-T, grade 3–4 cytopenia beyond 90 days was seen in 11.5% and 3.8% of patients treated with axi-cel versus tisa-cel, respectively, with neutropenia 8.6 vs. 2.9% and thrombocytopenia 8.6% vs. 1.9% (31). Of note, lymphodepletion intensity was higher for axi-cel compared to tisa-cel ( Table 1 ) (3, 4).

Reported risk factors for prolonged cytopenia are baseline cytopenia, prior treatments including bone marrow transplant, and factors related to inflammation such as C-reactive protein (CRP), ferritin, and other inflammatory markers (19, 32). Although not specific for cytopenia lasting more than 90 days, the CAR-HEMATOTOX score is interesting in predicting the risk of longer duration of neutropenia and higher incidence of severe thrombocytopenia and anemia, using baseline parameters of platelet count, hemoglobin, absolute neutrophil count (ANC), CRP, and ferritin. It was based on a large multicenter study of 235 patients treated with axi-cel and tisa-cel in a real-world setting that showed that the risk factors for late cytopenia were hematopoietic reserve and baseline inflammation (19, 32).

The impact of prolonged cytopenia is great because of infection risk (33, 34), transfusion support (33), healthcare resources for laboratory monitoring through hospitalization (18), and therapeutic limitations to treat disease relapse, especially in the clinical trial setting (35). Also, some studies tried to correlate late cytopenia with outcomes such as response, progression-free survival (PFS), and overall survival (OS), but this is not well-established (36, 37).

The etiology of cytopenia occurring beyond 30 and 90 days after CART is not completely understood but is possibly related to immune or inflammatory-mediated alterations in the bone marrow microenvironment and suppression of hematopoietic stem cell (HSC) (18). CRS plays a role that is more pronounced for early and short-term cytopenia (38, 39). When obtained, the bone marrow is usually hypocellular, and Strati et al. identified clonally expanded CX3CR1^hi cytotoxic T cells expressing IFN-gamma in the bone marrow of patients with prolonged cytopenia beyond 30 days, suggesting a possible IFN-γ-driven hematopoiesis impairment (40).

Based on the role of IFN-γ-driven hematopoiesis impairment post-CART, the pre-clinical effect of thrombopoietin receptor agonist (TPO-RA) preserving hematopoiesis in IFN-γ-mediated inflammatory conditions (41), and their effectiveness in aplastic anemia, TPO-RA have been used for the treatment of post-CART cytopenia (42–44). However, it is challenging to measure TPO-RA effectiveness in post-CART cytopenia without a randomized clinical trial, as in most cases, cytopenia will eventually improve without specific interventions.

Lastly, there are reports of the use of autologous and allogeneic stem cell boost (SCB) for patients with lymphoma with very prolonged and profound cytopenia after CART (45, 46). Rejeski et al. reported 12 cases where the median time of SCB infusion was 69 days after CART19, and the median time to neutrophil and platelet engraftment was 15 and 21 days, respectively (45). Gagelmann et al. reported SCB at a median 43 days after CART19 with neutrophil recovery in 84% and a median time to recovery of 9 days from SCB (46). Allogeneic hematopoietic CD34+ selected stem cell boost was also reported in seven patients with B-cell acute lymphoblastic leukemia (B-ALL) (47), and allogeneic transplant was reported for a patient with lymphoma, prolonged cytopenia, and ASXL1 mutation (48).

In addition to prolonged cytopenia, patients may have prolonged lymphopenia and hypogammaglobulinemia after CART19. B-cell aplasia is an expected on-target off-tumor effect and correlates with CART persistence (20, 49). Sustained polyclonal B-cell recovery was reported after tisa-cel in patients with long-duration complete responses (20) and was reported in 50% of patients with complete responses at a median time of 6.7 months after tisa-cel infusion (20). After treatment with axi-cel, approximately 30%, 40%, 50%, and 75% of patients had detectable B cells after 6 months, 12 months, 18 months, and 24 months of treatment, respectively (3, 50). Baird et al. suggested that the duration of B-cell aplasia correlates with response and that loss of B-cell aplasia is concurrent with or precedes clinical or radiological relapse after axi-cel (50).

Hypogammaglobulinemia is an expected effect after the depletion of B cells. Although 20%–30% of patients had low IgG levels even before CART (17, 51), IgG levels are expected to decline by 17% mean decrement in the first 100 days (49) and reach a nadir approximately 6 months after CART (17, 34, 49, 51). Up to 60% of patients had low IgG levels at 3 months after infusion, and up to 50% still had IgG < 400 mg/dL at 1 year without intravenous immunoglobulin (IVIG) replacement (51). The use of IVIG in clinical practice is highly variable among different institutional guidelines, so the percentage of patients requiring IVIG replacement varies from 27% to 38% ( Table 2 ). Of note, among the pivotal CART19 trials in non-Hodgkin lymphoma (NHL), the cumulative rate of use of IVIG seems lower when CART19 is used as 2L therapy, 13%–17%, vs. 3L therapy, 30%–38%, probably due to cumulative effect from prior therapies ( Table 2 ) (1, 3–7, 23).

Beyond B-cell aplasia, CD4+ T-cell count normalization can take more than 1 year, with a median CD4 count of 150–250 cells/mm³ at 1 year post-infusion (51, 52) but can remain as low as 150 cells/mm³ at 18 months. Risk factors for delayed CD4 reconstitution are older age, which may be related to thymic function and recovery in the CD19 compartment due to the participation of B cells in T CD4+ expansion (50). The practical impact of following CD4 counts is the possible increased risk of infections and reduced immune response to vaccines. It is important to note that CD4 is <200 cells/mm³ in up to 40% of patients before lymphodepletion (52). In contrast, CD56+ NK cells and CD8+ T lymphocytes recover faster and are normalized in most patients at 1 year (33, 50, 51).

Late infections are defined as infections beyond day 30 after CART. Initial long-term follow-up data of 54 patients regarding infection beyond 90 days after CART19 reported an infection density of 0.55 infection/100 days at risk (2.08 per patient-year) (17). The most common infections were upper respiratory tract infections (48%), followed by lower respiratory infections (23%), 80% of patients were treated on an outpatient basis, 20% were inpatients, and 5% were in the intensive care unit (ICU) (17). Wudhikarn et al. showed that in 60 NHL patients treated with tisa-cel or axi-cel, most infections occurred within the first 2 months after CART (52). Decreasing incidence and the grade of infection over time are consistently reported ( Table 3 ) (34, 50, 51).

We recommend prophylaxis for HSV/VZV and PJP for at least 12 and 6 months from CART19, respectively, and to check CD4 counts prior to discontinuation. We recommend neutropenic prophylaxis with levofloxacin and fluconazole while neutrophils are less than 0.5 × 10⁹/L. If prolonged severe neutropenia is more than 3 weeks, a mold-active azole should be considered. In patients with prior or chronic hepatitis B virus, entecavir should be considered for at least 6 months post-CART19 along with viral load monitoring (54).

Given the genetic editing nature of CART19, since its development, there has been a theoretical concern about the potential risk of T-cell and other malignancies due to insertional mutagenesis CAR transduction. Because of this, the FDA required that all patients need to be followed up for 15 years after treatment to assess the long-term safety and the risk of secondary malignancies. Recently, the FDA reported the occurrence of T-cell malignancy after different CART products for different diseases, warranting further investigation despite that the overall benefits of CART continue to outweigh its potential risk (77–79).

To date, there have been few reports of T-cell lymphoma (TCL) after treatment with CAR T cells. There were two cases of CART-derived T-cell malignancy among 10 patients treated with an investigational CART product manufactured by piggyback, a non-viral technology (80). There is also a case report of documented B-cell maturation antigen (BCMA) CAR+ TCL following treatment with ciltacabtagene autoleucel (cilta-cel) where the CAR insertion in the 3′ untranslated region of PBX2 along with other likely pre-existing genetic mutations (81). In another report from the University of Pennsylvania group where 449 patients received CART, only one case of TCL was reported after axi-cel, but quantitative PCR ruled out association with CAR transgene, and the T-cell clone preceded axi-cel ( Table 4 ) (82).

Moreover, much more often seen than TCL, there are many reports of myeloid neoplasms following CART (17, 27, 28, 33, 83, 84), including a recent boxed warning from the FDA of cilta-cel because 10% of the patients from their pivotal trial CARTITUDE-1 developed myeloid neoplasms (13). It is unclear if the development of subsequent myeloid malignancies is related to receipt of CART itself or to the fact that this is a heavily pretreated population already at risk of therapy-related myeloid neoplasm and perhaps now is experiencing longer survival from their underlying disease due to CART (77–79). The Mayo Clinic group reported a cumulative incidence of myeloid neoplasm of 9% at 3 years post-CART, and 70% of the patients with baseline samples had clonal hematopoiesis prior to CART (84) and median time to diagnosis of 9 months from CART (83). Similarly, the Fred Hutchinson group had also reported that among four (5%) cases of myeloid neoplasm post-CART, two of them had cytogenetic abnormality preceding CART19 ( Table 4 ).

Lastly, there are also several reports of solid tumors post-CART (17, 27, 28, 82, 85), with cumulative incidence at 5 years of 15% (82), but nearly half are non-melanoma skin cancer ( Table 4 ).

There is a report of delayed neurotoxicity 6 months after CART, with confusion, disorientation, and expressive aphasia, in which extensive workup for cerebral ischemia and infection was normal and unresponsive to dexamethasone but with response to cyclophosphamide (86). There is one case report of ascending paresthesia due to transverse myelitis at day 27, with a relapsing/remitting course over a 5-month period, treated with steroid pulse, IVIG, and plasma exchange (87). There are case reports of other neuropsychiatric events such as cerebrovascular events, Alzheimer’s dementia, peripheral neuropathy (17), and memory impairment (ZUMA-1) probably unrelated to CART.

Neurocognitive function was impaired at baseline in 33% of patients (88). After CART, 47.5% of patients reported at least one cognitive difficulty and/or clinically meaningful depression and/or anxiety, with depression before CART associated with a higher likelihood of self-reported post-CART cognitive difficulties, and younger age associated with worse long-term global mental health, anxiety, and depression (89). Neurocognitive assessments after CART show a possible small transient decline at day 90 (90, 91) with similar neurocognitive function at mid/long-term (6–24 months) compared to baseline (92, 93).