Since the first report of the efficacy of second-generation CAR T cells against CLL in 2011 [8], results have been published or reported for the injection of CAR T cells into 134 CLL patients [8–22]. The clinical status of these patients is reported in Table 1, together with the CAR T constructs and lymphodepletion schemes used.Tab1

Clinical situations and the characteristics of the CAR T cells used to treat the 134 CLL patients reported to date

The first observation to emerge from these results is that the population of treated patients had a particularly poor prognosis. The median age of the patients treated was 61 years (range: 40 to 77 years), and most were in relapse after a large number of lines of treatment. Overall, 68 patients had already received ibrutinib [14, 15, 18, 20–22], 25 had already received venetoclax [18, 20, 22], nine were in post-allograft relapse [12, 16], and 12 were treated in the context of transformation into refractory high-grade lymphoma (Richter’s syndrome) [12, 13, 18, 20]. In addition, 74 of the 108 patients evaluated (68.5%) had p53 alterations, and 41 of the 70 patients evaluated (58.6%) had a complex karyotype (see Table 1).

It is not straightforward to integrate these data, but a second observation that emerges is that efficacy is lower for CLL than for B-ALL and DLBCL: a complete response (CR), according to the IWCLL criteria, was obtained in only a minority (20–30%) of patients [14, 18], and progression-free survival (PFS) estimated at 25% at 18 months [14, 23]. Responses appear to be weaker in the lymph nodes than in the bone marrow and blood. Furthermore, these results should be considered in light of the frequency of complete bone marrow responses with undetectable minimal residual disease (MRD) reported in some series [18, 20–22], which has been correlated with PFS and OS close to 100%, with a median follow-up of 6.6 months [18]. It is difficult to determine the precise response to CAR T cells specifically in patients with Richter’s syndrome from published data, but this response is objective, with a possible decrease in lymph-node tumor syndrome. However, it appears to be partial and transient, and insufficient for the moment to improve the very poor prognosis of these patients [12, 13, 18, 20].

Promising data have also been obtained for the use of allogeneic CAR T cells derived from lymphocytes from hematopoietic stem cell donors in the context of post-allograft relapse [12, 16]. Response rates remain low in these patients with a poor prognosis, but there are signs of efficacy, and the absence of graft-versus-host disease (GVHd) is highly reassuring.

As for other CAR T cell indications, there have been many improvements in lymphodepletion schemes and the construction of chimeric receptors.

Lymphodepletion was initially achieved with cyclophosphamide treatment alone, but today, it is almost always achieved with a combination of cyclophosphamide and fludarabine (see Table 1). This lymphodepletion procedure makes it possible, in particular, to improve the expansion and persistence of CAR T cells via hypothetical mechanisms such as the decrease in residual tumor mass, the induction of inflammation, the release of tumor antigens, and the decrease in the number of regulatory cells. The immunodepression induced by such lymphodepletion may also decrease the risk of immunization against the extracytoplasmic immunoglobulin variable fragment of CAR T cells, which is mostly of murine origin.

Alternatives to the currently preferred antigenic target in B lymphoid hemopathies, CD19, exist and may prove to be more effective or safe. For example, the use of clonal anti-light chain (kappa or lambda) CAR T cells would, theoretically, spare half the B-cell compartment and limit agammaglobulinemia [17]. CD23, the receptor of the invariant fragment of IgM (FcγR), or ROR1 (tyrosine kinase-like orphan receptor 1) are also potentially interesting targets, as they are relatively specific to the B-cell tumoral compartment of CLL [24–26].

The use of CAR T cells combining the variable fragment and the CD3ζ chain with a co-stimulatory molecule of CD137 (or 4-1BB), rather than CD28, which was used in the first trials [8, 10, 14, 21], or in association with CD28 [18, 20, 22] made it possible to optimize the anti-leukemic effect of CAR T cells and to improve their long-term expansion and persistence via mechanisms that are still only partially understood [27].

The use of a variable fragment of humanized immunoglobulin in the construction of CAR T cells [21] should make it possible to limit the risk of immunization against the variable fragment, as most of the fragments used originate from mice, thereby improving the long-term maintenance of the CAR T population. Control over the CD4/CD8 ratio of the injected CAR T cells [22] could also improve the management of the CAR T cell expansion and long-term maintenance phases.

Finally, it is clear that disease persistence at time on injection has an impact on the expansion and maintenance of CAR T cells, and the composition of the expanding population: indeed, the CAR T cells of CLL patients displaying a CR at the time of injection expand more effectively and have a cytokine profile favoring their cytotoxic function and better long-term maintenance [23, 28, 29]. In addition, toxicity is lower when the residual tumor mass is limited at the time of CAR T-cell injection. These findings argue for administration earlier in the course of the disease, to ensure that the best possible response is obtained.

The lower efficacy of CAR T cells in CLL may be partly due to the intrinsic characteristics of the immune system in CLL, which is exhausted by diverse immunosubversion mechanisms, decreasing CAR T-cell activation after transduction.

Indeed, the CD4+ T cells of CLL patients have an exhausted phenotype (strong expression of PD-1, CD160, and CD244) and their CD8+ T cells have low proliferative and cytotoxic capacities [30]. These intrinsic characteristics of CLL immune cells are present at the time of diagnosis, but are also favored by previous lines of treatment (with fludarabine, in particular).

The ex vivo expansion and transduction capacities of T cells from CLL patients are clearly different from those of T cells from healthy subjects. In particular, T cells from CLL patients display less expansion of so-called “naïve” CD4+ T cells, an essential criterion for the long-term activity of CAR T cells. Moreover, the naïve CD4+ T cells that manage to expand from the autologous samples of CLL patients express more exhaustion markers [28].

These data support a rationale of developing allogeneic CAR T cells from a healthy donor, in whom the capacity of T cells to expand and their cytotoxicity are not modified by the tumor clone.

Ibrutinib has already revolutionized the routine management of CLL, but it may also improve outcomes in CLL patients receiving CAR T cells.

Indeed, particularly promising rates of response to CAR T therapy were reported in three studies. In 2016, Fraietta et al. reported their experience with this treatment, which was limited to three patients who stopped taking ibrutinib just before the leukapheresis preceding CAR T therapy. A response was observed in all three patients, including complete remission in one case, despite the absence of lymphodepletion [15]. At the last American Society of Hematology conference, two groups reported results for two series of 19 patients receiving injections of structurally different CAR T cells, in combination with ibrutinib. The overall response rate was above 80% and the frequency of complete bone-marrow response with undetectable MRD exceeded 90% [20, 21].

Many hypotheses have been put forward to explain this effect of ibrutinib, mostly based on our knowledge of the impact of ibrutinib on the immune system in CLL, which is probably still very patchy. In addition to Bruton’s tyrosine kinase, ibrutinib is known to target the IL2-inducible T-cell kinase (ITK), which orients T cells towards a Th1 cytokine secretion profile [31]. Ibrutinib may therefore be involved in redirecting the immune response of autologous T cells (before and after transduction) from a Th2 profile to a Th1 profile, more favorable for the long-term expansion and maintenance of chimeric receptor-expressing T-cell populations. Indeed, the ability of ibrutinib to promote the expansion, maintenance, and cytotoxicity of CAR T cells and to promote cellular immune responses (with, in particular, a decrease in exhaustion markers, the modification of cytokine secretion profiles, and an increase in the diversity of the T repertoire, etc.) has been demonstrated in vitro [15, 32, 33].

Cytokine release syndrome (CRS) and neurological toxicity (CRES, for CAR T cell-related encephalopathy syndrome) are, as in other indications for CAR T therapy, the most frequent complications in CLL, and their management is not different in this context [7, 34]. The incidence of these complications is variable in the small series available and it is probably still difficult to compare them: CRS occurs in 50 to 100% of patients (Grade ≥ 3 in 25 to 60% of cases), whereas neurological toxicity is less common (0 to 35% of cases) and mostly of moderate intensity. Death attributable to the CAR T cell procedure was reported for three of the 129 patients for whom clinical outcome data are available (2.3%).

CAR T cells do not appear to behave differently in CLL and in other hematological diseases in terms of the time lag to the onset of complications or the response to tocilizumab or corticosteroids, and there are, therefore, currently no specific instructions for CLL.

The use of ibrutinib before leukapheresis has been linked to a higher incidence and greater severity of CRS in first series [18], but the concomitant administration of ibrutinib and CAR T cells appears to be associated with a lower incidence of ≥ grade 3 CRS [20, 21] and lower levels of pro-inflammatory cytokines (including IL-6, IL2Rα, and MCP-1, in particular) [20].

Finally, in one case, a patient with CLL treated with CAR T cells was reported to display the proliferation of an identified population of clonal CD8+ CAR T cells carrying 1) a TET2 gene interrupted by the chimeric antigen receptor transgene and 2) a preexisting TET2 mutation in the second allele [35]. This resulted, in this particular case, in the persistence of the mutated TET2 CD8+ CAR T-cell population and complete remission of CLL more than five years after injection. This example, presented as an opportunity by the authors, should make us think carefully about the relatively moderate control we have over such genetic manipulations, particularly in elderly patients who have received a number of different treatments, in whom residual hematopoiesis is fragile and oligoclonal, and about the need to follow these patients closely in the long term.

Therapeutic strategies for CLL will be dominated, in the near future, by the use of BCRi, which will be the first-line treatment in most patients, relegating immunochemotherapy to an uncertain secondary role. BCL2i is currently indicated for patients with relapses and patients intolerant to BCRi. This new strategy certainly has major advantages in terms of response and survival, but several obstacles to its use have emerged: 1) the use of these new drugs, often continuously until relapse is associated with adverse effects, including cardiovascular effects (for BCRi) and with very high direct costs. CAR T cells could be used early in the treatment of CLL, as an alternative. 2) The treatment of patients with relapses or refractory disease after treatment with BCRi and BCL2i and the treatment of patients with Richter’s syndrome remain challenging. In these high-risk patients, CAR T cells are currently used 1) in place of HSC allografts for patients ineligible for HSC transplantation and 2) instead of HSC allografts for some patients eligible for transplantation. However, CAR T cells could ultimately be used as a complementary treatment, in addition to HSC allografts.