`DOI 10.1007/s00262-012-1254-0
`
`REVIE W
`
`Targeted immunotherapy of cancer with CAR T cells:
`achievements and challenges
`
`Grazyna Lipowska-Bhalla · David E. Gilham ·
`Robert E. Hawkins · Dominic G. Rothwell
`
`Received: 26 January 2012 / Accepted: 25 March 2012 / Published online: 22 April 2012
` Springer-Verlag 2012
`
`Abstract The adoptive transfer of chimeric antigen receptor
`(CAR)-expressing T cells is a relatively new but promising
`approach in the Weld of cancer immunotherapy. This therapeu-
`tic strategy is based on the genetic reprogramming of T cells
`with an artiWcial immune receptor that redirects them against
`targets on malignant cells and enables their destruction by
`exerting T cell eVector functions. There has been an explosion
`of interest in the use of CAR T cells as an immunotherapy for
`cancer. In the pre-clinical setting, there has been a consider-
`able focus upon optimizing the structural and signaling
`potency of the CAR while advances in bio-processing tech-
`nology now mean that the clinical testing of these gene-modi-
`Wed T cells has become a reality. This review will summarize
`the concept of CAR-based immunotherapy and recent clinical
`trial activity and will further discuss some of the likely future
`challenges facing CAR-modiWed T cell therapies.
`
`Keywords T cell · Gene modiWcation · Chimeric antigen
`receptor · Cancer · Immunotherapy
`
`G. Lipowska-Bhalla · D. E. Gilham · R. E. Hawkins
`Clinical and Experimental Immunotherapy Group,
`School of Cancer and Enabling Sciences,
`Manchester Academic Health Science Centre,
`University of Manchester, Manchester, UK
`
`G. Lipowska-Bhalla · D. G. Rothwell
`Clinical and Molecular Monitoring Laboratory,
`Clinical and Experimental Pharmacology Group,
`Manchester Academic Health Science Centre,
`School of Cancer and Enabling Sciences,
`University of Manchester, Manchester, UK
`
`D. E. Gilham (&)
`Clinical and Experimental Immunotherapy Group,
`Paterson Institute for Cancer Research, Wilmslow Road,
`Withington, Manchester M20 4BX, UK
`e-mail: dgilham@picr.man.ac.uk
`
`Introduction
`
`The concept of the chimeric antigen receptor (CAR; also
`known as T-bodies or chimeric immune receptors) was
`originally described over 15 years ago by Zelig Eshhar and
`colleagues working at the Weissman Institute in Israel [1].
`The approach was based upon the idea of expressing novel
`receptors on the T cell surface that would enable the T cell
`to identify intact protein antigens present on the surface of a
`target cell. T cells generally recognize peptide antigens that
`are presented in association with major histocompatibility
`complex (MHC) proteins by the target cell. However, one
`well-documented tumor escape mechanism is the modula-
`tion or down-regulation of MHC on the surface of the
`tumor cell [2] which thereby eVectively renders the tumor
`“invisible” to T cells, since binding of the T cell receptor to
`peptide-MHC is a pre-requisite for T cell eVector function.
`However, the direct recognition of protein antigens through
`a CAR would then make the tumor cell “visible” to T cell
`immune surveillance once more. Moreover, the use of a tar-
`geting system that functions independently of MHC means
`that the CAR can be used in a generic manner rather than
`having the restrictions that are imposed by the use of T cell
`receptor (TCR) approaches where the speciWc receptor is
`only suitable for patients expressing a speciWc MHC. In
`addition, CARs can substantially broadened the range of
`antigens recognizable by T cells to include carbohydrate [3]
`and glycolipid tumor antigens that are not within the scope
`of TCR-based recognition [4, 5]. Consequently, these fac-
`tors make the use of CARs highly attractive for adoptive
`gene-modiWed T cell therapy. The reader is also directed to
`other highly relevant and recent reviews of CAR T cell
`biology [6–8] and clinical application and focused reviews
`detailing novel directions of CAR T cell
`therapy
`approaches [9].
`
`123
`
`UPenn Ex. 2035
`Miltenyi v. UPenn
`IPR2022-00855
`
`©
`
`
`954
`
`Cancer Immunol Immunother (2012) 61:953–962
`
`The basic CAR structure: extracellular
`and transmembrane protein domains
`
`The basic CAR consists of an extracellular antigen-recogni-
`tion domain attached to an extracellular spacer domain, a
`trans-membrane region that anchors the receptor to the cell
`surface and a signaling endodomain. Single-chain antibody
`fragments (scFv) consisting of the variable heavy (VH) and
`a variable light (VL) chain isolated from an antibody linked
`by a Xexible linker have been used extensively as the anti-
`gen-recognition domain in many CARs due to their small
`size, which facilitates both the genetic manipulation and
`expression of the CAR (Fig. 1). The scFv determines the
`CAR antigen speciWcity and binds the target protein in a
`MHC-independent, non-restricted manner, most commonly
`with the speciWcity and aYnity similar to that of the anti-
`body from which it was derived [10]. More recently, non-
`scFv-based ligand-binding domains have also been suc-
`cessfully utilized in a CAR format including endothelial
`growth factor polypeptide, an integrin binding peptide,
`heregulin, IL-13 mutein, and NK cell receptor NKG2D
`[11–14].
`In most CARs, the antigen-recognition domain is con-
`nected to the transmembrane region by means of an extra-
`cellular spacer domain. The rationale for this is to distance
`the recognition domain from the membrane and to poten-
`tially make it more “accessible” to bind target. Spacer
`
`regions used tend to be comprised of common Ig-like
`domains due to the stability of the protein domain and have
`included immunoglobulin Fc or an extracellular fragment
`derived from CD28, TCR♢ chain, CD8♡, or NKG2D [13–
`19]. However, the absolute requirement for an extracellular
`spacer domain most likely depends upon the position of the
`target epitope with respect to the target cell surface. CARs
`targeting epitopes toward the proximal end of protein anti-
`gens tend to function well without spacer domains while
`CARs targeting epitopes buried closer to the target cell
`membrane appear to demonstrate improved function when
`a Xexible spacer region is included [20, 21]. These observa-
`tions suggest that there is still a requirement for the empiri-
`cal testing of CARs with varying extracellular domains in
`order to clearly identify the optimal receptor for each target
`antigen.
`In a similar manner, various transmembrane regions have
`also been employed in CARs including those derived from
`CD28, CD3♩, CD8, CD4, or Fc♦RI♤ [15, 17, 19, 22, 23].
`While CARs bearing any of these protein domains have
`been shown to function in terms of down-stream signaling
`after antigen ligation, the structural and biochemical impact
`of the transmembrane domain upon the CAR remains
`largely unknown. Recent studies by Bridgeman et al. [24]
`investigating CARs employing the CD3♩ transmembrane
`domain indicate that the biochemical interactions that occur
`between the wild-type CD3♩ transmembrane domain and
`other components of the endogenous TCR/CD3 complex are
`important for the optimal activity of the CD3♩ CAR. These
`studies suggest that the over-expression of the CD3♩ con-
`taining CAR permits an increased level of endogenous TCR
`expression on the transduced T cell since the availability of
`CD3♩ protein represents the rate-limiting factor controlling
`cell surface TCR/CD3 complex expression in T cells; this
`translates to an increased responsiveness to mitogenic anti-
`CD3 antibody stimulation. Whether this confers an advanta-
`geous or deleterious eVect upon T cells engrafted with CD3♩
`transmembrane domain containing CARs in vivo remains
`unknown. However, there is a general lack of understanding
`of the speciWc structural and biochemical nature of the
`majority of CARs and, in particular, the speciWc eVects that
`the range of extracellular and transmembrane domains
`impart upon the endogenously expressed TCR in the CAR-
`expressing T cell.
`
`Basic CAR structure: signaling domains
`
`Fig. 1 The structure of a CAR. CARs consist of a scFv derived from
`an antibody variable heavy (VH) and variable light (VL) fragments, a
`hinge region linking the scFv with a transmembrane (TM) domain and
`signaling domain, most often CD3♩ or Fc♦RI♤. CL—constant region of
`light immunoglobulin chain, CH1–CH3—constant region of heavy
`immunoglobulin chain, dotted line corresponds to disulWde bound
`
`In keeping with the plethora of options available for the
`extracellular domains of the CAR, there has been an ever-
`increasing range of intracellular signaling domains shown
`to function in the context of a CAR. Initial studies
`employed a single signaling domain, most commonly
`
`123
`
`UPenn Ex. 2035
`Miltenyi v. UPenn
`IPR2022-00855
`
`
`
`Cancer Immunol Immunother (2012) 61:953–962
`
`955
`
`derived from the CD3♩ chain or the ♤ chain of the high-
`aYnity IgE Fc receptor (Fc♦RI), referred to as FcR♤
`(Fig. 1). CARs consisting of either signaling domain
`proved capable of activating T cell eVector function includ-
`ing target cell cytotoxicity and cytokine release in response
`to target antigen binding [25]. However, head to head in
`vivo studies suggested that mouse T cells engrafted with
`CD3♩ CARs demonstrated improved anti-tumor activity as
`compared to those bearing the FcR♤ signaling domain [26].
`CARs bearing signaling domains from a single receptor
`have been subsequently termed “Wrst-generation” receptors.
`The modular nature of the CAR lends itself to further engi-
`neering
`involving combining
`intracellular
`signaling
`domains that can potentially increase the potency of the
`CAR. The full activation of T cells requires multiple sig-
`nals, and it is clear that signaling from these Wrst-generation
`CARs only supplied the so-called “signal one” that could
`drive T cell eVector functions, but in the absence of further
`signals (signal two or co-stimulation), the T cells were
`unable to fully engage its eVector machinery and therefore
`undergoes apoptosis.
`The most studied co-stimulatory pathway has been liga-
`tion of T cell CD28 receptors with the B7 family of ligands.
`Ligation of CD28 on CAR T cells through the expression of
`B7 co-stimulatory ligands on target cells [15, 27, 28] or co-
`expression of the CD28 molecule together with the scFv
`and CD3♩ domains of the CAR [29] was shown to lead to
`proliferation of CAR-modiWed T cells and enhanced anti-
`tumor activity. Exploiting the modular nature of the CAR
`enabled the engineering of “second-generation” CARs
`
`where the signaling domains of two receptors were
`expressed within one CAR (Fig. 2). The fusion of CD28
`and CD3♩ into a single CAR has proven to be eVective in
`permitting the repeated antigen stimulation and prolifera-
`tion of CAR-expressing T cells in the absence of exogenous
`co-stimulatory ligands, maintaining the ability to demon-
`strate both antigen-speciWc cytolysis and to secrete IL-2 in
`vitro [16, 28, 30, 31]. Recent Wndings have also shown that
`CARs containing the CD28 domain could also rescue acti-
`vated T cells from antigen-induced cell death (AICD) [32]
`and enhance the resistance of CAR-modiWed T cells to reg-
`ulatory T cells (Treg) activity [33].
`Experiments in animal models conWrmed these in vitro
`observations and demonstrated that T cells engrafted with
`CARs containing the CD28 signaling domain had pro-
`longed survival, produced higher levels of cytokines, pro-
`liferated vigorously, and mediated enhanced anti-tumor
`eVect compared to T cells expressing the conventional
`CARs [34–36]. These reprogrammed T cells have also been
`observed to proliferate robustly, even without administra-
`tion of exogenous IL-2 [37].
`Several other co-stimulatory receptors have been studied
`including OX40 (CD134) [38], 4-1BB (CD137) [39],
`DAP10, and ICOS [31, 40]. Although all these constructs
`showed antigen-dependent cytotolysis in vitro, only 4-1BB
`CARs showed enhanced persistence and anti-tumor activity
`in vivo [39, 41]. All other second-generation CARs failed
`to produce suYcient amounts of IL-2 to promote T cell pro-
`liferation in the absence of exogenous co-stimulation. How-
`ever, when the exogenous B7 co-stimulation was delivered,
`
`Fig. 2 Clinical protocol of
`adoptive transfer of CAR-
`expressing T lymphocytes. T
`lymphocytes are collected from
`a cancer patient and retrovirally
`transduced with CAR genes,
`then expanded ex vivo to large
`numbers and infused back into
`the patient. To facilitate engraft-
`ment of CAR-modiWed T lym-
`phocytes in vivo, the patient is
`given a lymphodepletive chemo-
`therapy regimen prior to infu-
`sion and cytokine support post-
`infusion
`
`DNA
`encoding CIR
`
`Viral vector
`
`Gene
`modification
`
`T cell
`expansion
`
`Isolation of T cells
`
`Infusion of modified
`T cell
`
`Conditional
`chemotherapy
`
`IL-2 supportive
`therapy
`
`123
`
`UPenn Ex. 2035
`Miltenyi v. UPenn
`IPR2022-00855
`
`
`
`956
`
`Cancer Immunol Immunother (2012) 61:953–962
`
`transgenic T cell expressing these CARs produced IL-2 and
`showed improved proliferation as compared to T cells
`engrafted with CD3♩ alone CARs under the same condi-
`tions [31].
`This Wnding paved the way for the engineering of “third-
`generation” CARs that contain CD28 and one other co-
`stimulatory domain, most often OX40 or 4-1BB, fused with
`the activation domain (Fig. 2). These CARs were found to
`drive equivalent levels of cytolysis as observed in second-
`generation CARs [38, 39, 42] but drove higher and more
`prolonged levels of cytokine production and cell prolifera-
`tion [38].
`The hierarchy of CAR T cell co-stimulation, reXected in
`the sequence of CAR co-stimulatory and activation
`domains, appears to directly translate into in vivo anti-
`tumor eYcacy with second-generation CARs being supe-
`rior to Wrst-generation and third-generation outperforming
`the second-generation. However, the latest Wnding by
`Cheadle and colleagues suggests that functional activity of
`CAR-expressing T cells may be dependent not only upon
`the optimal combination of CAR signaling moieties but
`also upon the endogenous physiological receptor interac-
`tions provided by target cells [43]. Murine T cells
`engrafted with a Wrst-generation CD19-speciWc CAR were
`found to produce IL-2 following co-culture with CD19+
`B-cell lymphoma cells independently of CD28 receptor
`ligation, with the production of IL-2 driven by endogenous
`CD2 receptor activity. This Wnding suggests that the opti-
`mization of CAR signaling domains according to target
`cells could possibly be the way forward in achieving the
`best anti-tumor eVect rather than use of a universal CAR
`against all target cells.
`As previously mentioned, the majority of engineered
`CARs contain the CD3♩ signaling moiety that exploits TCR
`proximal kinases signaling pathway. CARs based on alter-
`native signaling domains utilizing TCR distal signaling
`events have also been engineered. These CARs include sig-
`naling domains derived mainly from the protein tyrosine
`kinase (PTK) Syk family and are capable of triggering anti-
`gen-dependent T cell activation with IL-2 production and
`target cell lysis [44]. Although the eYciency of Syk-based
`CARs is not superior to CD3♩-based CARs, their main
`potential advantage is the ability to bypass TCR proximal
`signaling events, which are often defective in cancer
`patients, and directly trigger downstream signal transduc-
`tion machinery.
`
`Initial clinical trials with CAR-modiWed T cells
`
`Clinical protocols for CAR T cell therapy usually involve
`the isolation of autologous T lymphocytes from cancer
`patients, their ex vivo modiWcation with the CAR genes
`
`123
`
`followed by large-scale expansion. These genetically modi-
`Wed T lymphocytes are then infused back into the patient,
`usually with administration of IL-2 to support their viability
`and function. To facilitate the engraftment and persistence
`of CAR-modiWed T cells, cancer patients are also pre-con-
`ditioned prior to the cell transfer (Fig. 2).
`A number of Wrst-generation CARs showed some suc-
`cess in pre-clinical models and subsequently entered phase
`I clinical trials. The tumors targeted included ovarian
`cancer [45], renal cell carcinoma [46, 47], neuroblastoma
`[48, 49], B-cell non-Hodgkin lymphoma (NHL), and man-
`tle cell lymphoma (MCL) [50]. Unfortunately, despite
`promising preclinical results, the majority of these initial
`CAR T cell trials showed little evidence of anti-tumor
`activity with limited activation, persistence, and homing to
`tumor sites being the main barriers. However, responses
`were seen in a small number of trials with some anti-tumor
`responses being reported in B-cell lymphoma patients
`treated with ♡CD20-CD3♩ T cells, in which two patients
`were reported to maintain a previous complete response,
`one patient achieved a partial response and four patients
`achieved stable disease [50]. Further anti-tumor responses
`were also seen in patients with neuroblastoma treated either
`with CAR T cells targeting L1-cell adhesion molecule
`(L1CAM) [48] or diasialo-ganglioside (GD2) antigen [49].
`To date, though, the GD2 targeting trail is the one in which
`Wrst-generation CAR T cells have mediated a durable, com-
`plete response in cancer patients [51].
`Although clinical studies of Wrst-generation CARs
`produced rather modest clinical outcomes, they established
`the feasibility and safety of CAR-adoptive transfer therapy.
`This in turn paved the way for further improvements of
`CAR signaling capacity and resulted in the engineering of
`second- and third-generation CARs. These improved CARs
`are capable of delivering superior strength and quality acti-
`vation signal, resulting in increased proliferation, cytokine
`release, and eVector functions of CAR-modiWed T cells in
`vitro and in pre-clinical models. Both the second- and the
`third-generation CARs are now entering the clinical arena,
`and their therapeutic potential is under intense investigation
`(Table 1). Encouragingly, the Wrst clinical reports from
`these studies have been published and have shown some
`promising results [52–56].
`In a pilot clinical study with ♡CD19.4-1BB.CD3♩ T cells
`in three patients with advanced chronic lymphocytic leuke-
`mia (CLL), two complete remissions and one partial
`response that has been ongoing 10 months after the treat-
`ment has been achieved [52, 54]. The engineered T cells
`were found to expand over 3-logs in these patients, inWl-
`trated and lysed tumor cells, and persisted at high levels for
`over 6 months. Interestingly, a fraction of these cells dis-
`played a memory T cell phenotype, suggesting the potential
`for preventing tumor relapses.
`
`UPenn Ex. 2035
`Miltenyi v. UPenn
`IPR2022-00855
`
`
`
`Cancer Immunol Immunother (2012) 61:953–962
`
`957
`
`+Trial suspended temporarily due to re-location of cell processing facility
`*Trials due to open
`Williams Medical Center, Providence, Rhode Island, USA, UCL University College London, UK, UP University of Pennsylvania, Philadelphia, USA, CHMAN Christie Hospital, Manchester, UK
`MDACC M.D. Anderson Cancer Center, Houston, Texas, USA, MSKCC Memorial Sloan-Kettering Cancer Center, New York, USA, NCI National Cancer Institute, Bethesda, USA, RWMC Roger
`eBCM Baylor College of Medicine, Houston, Texas, USA, COH City of Hope Medical Center, Duarte, California, USA, FHCRC Fred Hutchinson Cancer Research Center, Seattle, Washington, USA,
`dLV lentiviral, RV ♤-retroviral, SB sleeping beauty transposon/transposase DNA plasmid system
`cEBV Epstein-barr virus-speciWc CAR-modiWed T cells
`cord blood transplantation
`bB-ALL B-lineage acute lymphoblastic leukemia, B-NHL B-lineage non-Hodgkin’s lymphoma, CLL chronic lymphocytic leukemia, HSCT hematopoietic stem cell transplantation, UCBT umbilical
`aCEA carcinoembryonic antigen, GD2 diasialoganglioside, Her2 human epidermal growth factor receptor, PSMA prostate-speciWc membrane antigen
`
`NCT00881920
`BCM
`NCT01109095
`BCM
`NCT00902044
`BCM
`NCT00889954
`BCM
`NCT00673322
`RWMC
`NCT00673829
`RWMC
`NCT01140373
`MSKCC
`NCT00664196
`RWMC
`NCT00085930
`BCM
`NCT00621452
`FHCRC
`NCT01318317*
`COH
`CHMAN
`NCT01493453+
`NCT01362452*MDACC
`MDACC
`NCT00968760
`NCI
`NCT01087294
`UP
`NCT00891215
`BCM
`NCT00840853
`NCI
`NCT00924326
`UCL
`NCT01195480*
`BCM
`NCT00709033
`BCM
`NCT00608270
`BCM
`NCT00586391
`MSKCC
`NCT01416974
`MSKCC
`NCT01044069
`MSKCC
`NCT00466531
`
`I
`
`I/II
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I/II
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I/II
`I/II
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`RV
`RV
`RV
`RV
`RV
`RV
`RV
`RV
`RV
`Electroporation
`LV
`RV
`Electroporation/SB plasmid
`Electroporation/SB plasmid
`RV
`LV
`RV
`RV
`RV
`RV
`RV
`RV
`RV
`RV
`RV
`
`CD3♩/CD28 versus CD3♩
`CD3♩/CD28 (EBV)
`CD3♩/CD28
`CD3♩/CD28
`CD3♩/CD28
`CD3♩/CD28
`CD3♩/CD28
`CD3♩
`CD3♩ (EBV)
`CD3♩/CD137/CD28
`CD3♩
`CD3♩
`CD3♩/CD28
`CD3♩/CD28
`CD3♩/CD28
`CD3♩/4-1BB versus CD3♩
`CD3♩/CD28
`CD3♩/CD28
`CD3♩ EBV
`CD3♩/CD28 versus CD3♩ (EBV)
`CD3♩/CD28 versus CD3♩
`CD3♩/CD28 versus CD3♩
`CD3♩/CD28
`CD3♩/CD28
`CD3♩/CD28
`
`First and second
`Second
`Second
`Second
`Second
`Second
`Second
`First
`First
`Third
`First
`First
`Second
`Second
`Second
`First and second
`Second
`Second
`First
`First and second
`First and second
`First and second
`Second
`Second
`Second
`
`B-NHL and B-CLL
`Glioblastoma
`Advanced osteosarcoma
`Lung malignancy
`Colorectal carcinoma
`Breast cancer
`Prostate cancer
`Prostate cancer
`Neuroblastoma
`Mantle cell lymphoma/indolent B-NHL
`B-NHL
`B-NHL
`B-lineage lymphoid malignancies post-UCBT
`B-lymphoid malignancies post-HSCT
`B-cell leukemia/CLL/B-NHL
`Lymphoma/leukemia
`ALL post-HSCT
`Lymphoma/leukemia
`ALL
`Advanced B-NHL/CLL
`B-NHL/CLL
`Lymphoma/leukemia (B-NHL)/CLL
`Leukemia
`B-ALL
`CLL
`
`Kappa light chain
`Her2/neu
`Her2/neu
`Her2/neu
`CEA
`CEA
`PSMA
`PSMA
`GD2
`CD20
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`CD19
`
`Centere
`
`gov identiWer
`Clinical trial.
`
`Phase
`
`CAR gene transferd
`
`CAR endodomainc
`
`CAR generation
`
`Diseaseb
`
`targeteda
`Antigen
`
`Table1Representative list of CAR T cell clinical trials (generated from a review of the clinical trials.gov database)
`
`123
`
`UPenn Ex. 2035
`Miltenyi v. UPenn
`IPR2022-00855
`
`
`
`958
`
`Cancer Immunol Immunother (2012) 61:953–962
`
`Although ♡CD19.4-1BB.CD3♩ T cells produced
`signiWcant therapeutic eVect, their potent cytotoxic
`activity led to the delayed development of potentially
`life-threatening tumor lysis syndrome in one of the
`patients [58]. The patient developed the syndrome
`22 days following the Wrst infusion of ♡CD19 T cells
`showing elevated levels of uric acid, lactate dehydroge-
`nase, and acute kidney injury. In two patients, elevated
`levels of certain serum cytokines were also detected;
`however, no classical symptoms of cytokine storm were
`observed [52, 54].
`Also, T cells modiWed with second-generation CD19-
`speciWc CARs containing a CD28 costimulatory endodo-
`main demonstrated enhanced expansion and persistence in
`lymphoma patients, compared to CAR T cells lacking this
`endodomain [57]. Infusions of ♡CD19.CD28.CD3♩ T cells
`were well tolerated, and no acute toxicity was seen.
`However, recently two serious adverse events were
`reported which highlight the need for caution while using
`activation potent second- and third-generation CARs. One
`occurred following the adoptive transfer of T cells modiWed
`with a CD19-speciWc second-generation CAR (♡-CD19-
`CD28-CD3♩) into a patient with advanced CLL [58], and
`the second in a patient with metastatic colon cancer treated
`with a third-generation CAR targeting HER-2/neu (♡-HER2/
`neu.CD28.4-1BB.CD3♩) [59].
`The treatment-related death in the anti-CD19 T cell
`recipient occurred shortly after cyclophosphamide-based
`chemotherapy and infusion of the anti-CD19 T cells. The
`patient developed acute sepsis, persistent fever, hypoten-
`sion, acute renal failure and expired 44 h following infusion
`of the T cells [58]. Although the cause of the patient’s death
`remains uncertain, the subsequent investigation has impli-
`cated an underlying infection which was exacerbated due to
`the cyclophosphamide-based pre-conditioning regime. Fol-
`lowing extensive re-appraisal, the trial has now been
`reopened and two subsequent patients have been treated
`with only minor modiWcations to the protocol [55]. These
`include the single infusion of CD19-reactive T cells follow-
`ing cyclophosphamide chemotherapy being split into two,
`with a 24-h delay between infusions. Both patients treated
`tolerated the treatment well, and no notable toxicity was
`seen.
`In contrast, the second fatality associated with a
`HER-2/neu-speciWc CAR seems to be a direct eVect of
`administration of large numbers of HER-2/neu-reactive
`T cells which
`targeted
`low
`levels of HER-2/neu
`expressed on pulmonary endothelium [59]. Elevated lev-
`els of IFN-♤, GM-CSF, TNF-♡, IL-6, and IL-10 observed
`in the patient serum following therapy were consistent
`with a “cytokine storm” resulting from the massive “oV-
`target” activation of infused T cells inWltrating pulmo-
`nary endothelium. Within minutes following infusion,
`
`123
`
`the patient developed acute pulmonary toxicity followed
`by cardiac arrest and died 4 days later.
`
`Challenges of CAR-modiWed T cell therapy
`
`Recent clinical trials have revealed several potential pitfalls
`to CAR-modiWed T cells that potentially still bar the way to
`an eVective CAR therapy. A major concern is the risk of
`“oV-target” toxicity, resulting in autoimmune reaction
`against self-tissues [46, 59]. This is because tumor-
`restricted antigens are very rare, and the majority of tumor-
`associated antigens that are targets for immunotherapy are
`shared with normal tissues. Although physiologically these
`antigens are usually expressed at low levels in normal tis-
`sue and are over-expressed in malignant cells, the CAR
`response is highly speciWc and can target both “on-target”
`and “oV-target” antigens. This is especially the case with
`second- and third-generation CARs containing potent com-
`bination of signaling and costimulatory molecules that have
`the potential to respond to low levels of targets physiologi-
`cally expressed, resulting in the generation of powerful
`activation signal, leading to a lethal “cytokine storm” [59].
`To adopt the lesson learned from donor lymphocyte
`infusion following hematopoietic stem cell transplantation,
`which showed dose-dependent toxicity [60], the autoim-
`mune reaction could be avoided or at least controlled by
`using split infusions of lower doses of therapeutic T cells.
`This has been already shown to abrogate the acute toxicity
`seen in the anti-CD19.CD28.CD3♩ CAR study in which,
`following a fatality, the large dose of 1010 CAR-modiWed T
`cells was split into two infusions and did not produce any
`notable toxicity in two subsequent patients [55].
`Interestingly, an autoimmune reaction was observed in
`clinical studies using a CAIX-speciWc CAR [46] and a
`HER-2/neu-speciWc CAR [59], but was not seen for a CAR
`targeting L1CAM on neuroblastoma cells [48], which is
`known to be also expressed on adrenal medulla and sympa-
`thetic ganglia, suggests that the aYnity with which the scFv
`binds target antigen could be critical in inducing an autoim-
`mune response. High-aYnity CARs are possibly more sen-
`sitive to low levels of antigen where binding generates a
`more potent activation signal compared to low aYnity
`CARs, which may require higher antigen levels to respond.
`Therefore, using CARs displaying a lower aYnity to their
`target antigen may be advantageous in preventing autoim-
`mune responses.
`Another way to obviate a potential autoimmune reaction
`is to include a suicide gene in CAR-modiWed T cells that
`could be activated in the event of acute toxicity leading to
`elimination of
`the CAR-modiWed T cells. Such an
`approach, involving the herpes simplex virus thymidine
`kinase (HV-TK) suicide gene, has already been proven to
`
`UPenn Ex. 2035
`Miltenyi v. UPenn
`IPR2022-00855
`
`
`
`Cancer Immunol Immunother (2012) 61:953–962
`
`959
`
`work in donor lymphocyte infused in patients after hemato-
`poietic stem cell transplantation to prevent graft-versus-
`host disease (GvHD) [61]. HV-TK transgenic cells are sen-
`sitive to the antiviral drugs ganciclovir and acyclovir,
`which upon phosphorylation by HV-TK interfere with
`DNA polymerase leading to cell death. This approach has
`now been utilized in CAR-modiWed T cells. Park and
`colleagues engineered CAR T cells targeting L-1CAM con-
`taining the HV-TK suicide gene and reported no HV-TK-
`related toxicity in patients with metastatic neuroblastoma
`[48]. However, the main shortcoming of the application of
`this HV-TK suicide system in humans is its potential
`immunogenicity due to the viral origin [61, 62]. This may
`result in a quick clearance of infused therapeutic T cells,
`limiting their therapeutic eYcacy. Recently, an anti-CD19
`CAR containing an alternative suicide gene, an inducible
`caspase-9 (icasp9), has also been generated [63]. The
`icasp9 is induced by the administration of a small-molecule
`dimerizer that merged two non-functional icasp9 molecules
`in the active enzyme. Activation of icasp9 was seen to rap-
`idly induce apoptosis of CAR-expressing T cells in vitro
`and in vivo. The main advantage of icasp9-based suicide
`system is the fact that icasp9 is of human origin, and its
`expression in CAR T cells is unlikely to induce unwanted
`immune response.
`Another controversial issue is related to the clinical
`safety of retroviral vectors encoding CARs that are used to
`transfect human T cells [64]. These vectors stably integrate
`into the host genome creating the risk of insertional muta-
`genesis that may lead to malignant transformation [65].
`Although no such event has been observed in patients
`treated with retroviral gene-modiWed T cells during the past
`20 years [66, 67], all possible precautions should be taken
`to minimize the risk. Thus, safer systems for the transfer of
`CAR genes are being developed including the use of lentiv-
`iral vectors [52, 54, 68, 69] and electroporation of RNA
`[70–73].
`The potential immunogenicity of the CAR molecules
`themselves is another potential limitation threatening CAR
`therapy. In cases where this occurs, most often the immune
`response is directed against the extracellular antigen-
`recognition domain that is usually derived from mouse
`monoclonal antibodies [45, 74]. Although, putative immune
`response against
`retroviral vector-encoded epitopes
`expressed by CAR-modiWed T cells has also been observed
`[74]. The immune rejection of CAR-modiWed T cells leads
`to their destruction, thereby reducing survival and anti-
`tumor eVect. This limitation, however, can be avoided by
`using humanized scFv’s or scFv’s derived from human
`monoclonal antibodies [75, 76]. Also, alternatives to viral
`transfection gene transfer techniques and modiWcation of
`clinical protocols to include immunosuppressive condition-
`
`ing regimens may be beneWcial to fully exploit the therapeu-
`tic potential of CAR-modiWed T cells.
`Finally, the use of autologous T cells requires CAR-
`modiWed T cells to be generated for each individual
`patient that is a logistically and economically demanding
`process. Therefore, much eVort is currently being directed
`to create universal eVector T cells that could serve as “oV-
`the-shelf” immunotherapeutic to be administered to
`patients irrespective of their MHC type. To this end, allo-
`geneic T cells have been the obvious candidate but their
`use is threatened by host-versus-graft (HvG) rejection and
`graft-versus-host (GvH) response [77]. However, the
`recent in vivo research by Marcus and colleagues showed
`that the GvH activity of allogeneic CAR T cells can be
`harnessed by some T cell traYcking modulators [78],
`potentially allowing exploitation of the full therapeutic
`eVect of allogeneic CAR T cells.
`
`Concluding remarks
`
`The adoptive transfer of CAR-engineered T cells represents
`a valuable and attractive therapeutic strategy that in the
`near future has the potential to give new prospects to cancer
`immunotherapy. While clinical trials with the Wrst-genera-
`tion CARs showed rather modest anti-tumor eVect, these
`trials established the feasibility and safety for the therapy
`and more importantly paved the way for further improve-
`ments of CAR function. This resulted in the engineering of
`second- and third-generation CARs capable of delivering
`greater strength and quality activation signals directly
`translating into superior anti-tumor properties of CAR-
`engineered T cells. Whether or not these second- and third-
`generations CARs will fulWll their clinical expectations is
`too early to speculate. These CARs are just entering the
`clinical arena, and results of their intense clinical assess-
`ment are much awaited in the Weld. Importantly, the further
`understanding of T cell subsets with particular respect to
`optimal T cell phenotypes for engraftment will impact upon
`future CAR T cell approaches [79–82]. So far, these Wrst
`clinical reports give some promise for tumor treatment but
`also emphasize challenges to be undertaken to establish
`CAR T cell approach as an eVective and safe cancer
`therapy.
`
`Acknowledgments DEG, REH, and DGR were funded by Cancer
`Research UK. This work was also supported by the EU FP6 pro-
`gramme ATTACK, FP7 training Network ATTRACT and the Kay
`Kendall