“Lymphorepletion” by IL-15

Optimal cytokine augmentation of chimeric antigen receptor T cells (CAR-T) could enhance their cure rates but is hampered by the potential for additive immune-mediated toxicities. In this issue of Blood, Srinagesh et al lead the field by demonstrating that interleukin-15 (IL-15) may be safely given with CAR-T.¹ 

Inflammatory cytokines like IL-2 are effective cancer immunotherapies whose use predates the clinical use of CAR-T. IL-2 contributes to the elimination of tumor cells in vivo by activating and expanding endogenous cancer-specific T cells. Indeed, high doses of IL-2 alone or combined with adoptively transferred tumor-infiltrating lymphocytes have induced unprecedented remissions in some patients with cancer. However, IL-2 administration induces serious toxicities such as capillary leak syndrome. Furthermore, prolonged IL-2 use can lead to T-cell exhaustion and the induction of regulatory T cells (Tregs). By contrast, IL-15, which shares sequence similarity with IL-2, has been shown to be well tolerated in human trials.² Not only did IL-15 induce “lymphorepletion” by activation of cancer-specific CD8⁺ T-cell and natural killer (NK)-cell subsets in these trials, but it also promoted a T-cell memory phenotype without expanding Tregs. IL15 has therefore replaced IL-2 in cultures during CAR-T manufacture and emerged as a potentially safer and superior alternative to IL-2 in the clinic.

A major mechanism of B-acute lymphoblastic leukemia (B-ALL) relapse after CD19-redirected CAR-T has been their limited persistence or their dysfunction due to exhaustion.³ Thus, the current state of the science centers on combining stimulatory cytokines with CAR-Ts to extend their survival or restore functionality. However, much like with the use of IL-2, CAR-Ts themselves are associated with well-known toxicities from supraphysiologic stimulation of immune cells such as cytokine release syndrome (CRS) and immune effector-associated neurotoxicities (ICANS). Therefore, phase 1 and 2 studies establishing the safety of the combination are paramount and are only just emerging.

Srinagesh et al administer a bispecific CD19- and CD22-targeted CAR-T product along with a polymer-conjugated recombinant IL-15 agonist known as NKTR-255. They hoped their dual innovations would counteract the problem of relapses from CD19 antigen loss or from dysfunctional CAR-Ts. NKTR255 has a considerably longer half-life compared with recombinant IL-15 alone (30 vs 2.9 hours), thus overcoming the need for a continuous infusion to achieve a biological effect.⁴ Administration of NKTR-255 on day 14 after CAR-T was well tolerated. Although fevers were noted for 1 to 2 days after NKTR-255, these were easily managed with acetaminophen. Apart from fevers, transient self-limited myelotoxicity and lymphopenia were observed post-NKTR-255. Importantly, however, there was no exacerbation of or induction of grade ≥2 CRS, ICANS, or macrophage activation syndromes. By comparing these results with those of a noncontemporaneous cohort of CD19-22 CAR-T recipients with B-ALL who did not receive NKTR-255, the authors show that the rates of severe pancytopenia (grade ≥3) were only slightly higher in the combination cohort (13% vs 22% in the NKTR-255 cohort).⁵ Moreover, 5 out of 9 patients treated with the combination went on to receive at least 1 additional infusion of NKTR-255. The additional infusions were well tolerated and did not induce prolonged cytopenias.

As expected, IL-15 levels were higher post-NKTR-255, as were levels of CXCL9 and CXCL10, which may explain the transient lymphopenia observed after day +15, as CXCL9, CXCL10, and IL-15 have all been shown to promote lymphocyte emigration into tissues.⁶ Indeed, 2 patients with prior central nervous system involvement were shown to have significantly higher levels of CAR-T within the cerebrospinal fluid after NKTR-255 administration. Further, NKTR-255 induced a rebound lymphocytosis by day +28, which is consistent with preclinical observations, where NKTR-255 injection in nonhuman primates led to expansion of CD8⁺ T and NK cells.⁴ However, in this trial, additional CAR-T expansion post-NKTR-255 was not seen. Indeed, compared with the same non-NKTR-255 cohort who received an identical CD19-22 CAR-T, there was no difference observed in the expansion of infused CAR-Ts. This might be because the comparisons were not randomized or powered to examine the effect on expansion of CAR-Ts. More likely, though, is that the inability to demonstrate expansion was because CAR-Ts may already be maximally expanding or the dosage or timing of NKTR-255 infusion was not optimized. Specifically, with regards to timing, in a mouse model of ALL, combining NKTR-255 with CD19 CAR-Ts led to a greater expansion of CAR-T and superior tumor control in mice who received NKTR-255 on day +7 but not after.⁴ Whereas the authors rightly took the approach of safety first in this trial by delaying NKTR-255 until day +14 to avoid exacerbating CRS driven by CAR-T, they may have missed the optimal timing to promote CAR-T expansion in vivo.

Recipients of NKTR-255 and CAR-T had a relapse rate of 33% (3 of 9) compared with 63% (5 of 8) in those that did not receive NKTR-255, indicating a potential therapeutic advantage of the combination. Although, as acknowledged extensively by the authors, the sample sizes were insufficient for efficacy comparisons, the analysis of both cohorts was useful to hint at safety and biological signals of the combination and perhaps to justify its further clinical development. Emerging gaps after this report include the identification of the optimal dosing of NKTR-255 (dose level 3 did not fully accrue), the mechanisms of relapse in those who failed the combination, the phenotypic changes in CAR-Ts, and conclusive proof of emigration to tumor sites post-NKTR-255.

Cytokine augmentation of CAR-T is a rational next step for improving outcomes. Ongoing CAR-T trials are thus studying exogenously administered cytokines or endowing CAR-Ts themselves with cytokine receptors or the ability to secrete cytokines.^7,8 Whether optimal cytokine combinations can selectively “replete” CAR-T function in vivo to the point that it enhances cure rates remains to be determined. To that end, the work of Srinagesh et al is of high significance, because they present one of the first of likely many forthcoming clinical experiences of combining cytokines with CAR-T.

Conflict-of-interest disclosure: P.L. is on the advisory board for Janssen Therapeutics and received research funding from Marker Therapeutics, Bristol Myers Squibb, Fate Therapeutics. J.L. declares no competing financial interests.