CAR-Ts and GBM: A Long Way Yet to Go
Are you ready for a N-valent CAR-T (N>4) with concomitant checkpoint inhibition…and the (S)AEs that come with it…to be available in perhaps 10 years?
Executive Summary
Welcome to “CAR-Ts for GBM Week” in the journals. A series of papers and short reports were just published describing results from several clinical studies using various CAR-Ts directed against GBM. We will review the studies in more detail below, but the key conclusions are:
As currently constructed CAR-Ts are minimally effective against GBM. Surprised, we are not. ORRs are minimal; transient Stable Disease is the most likely outcome; and most patients experience rGBM shortly after CAR-T therapy. While CAR-Ts have demonstrated some efficacy in treating hematological cancers, their utility with solid tumors remains elusive, even for tumors that do not sit behind the BBB.
Response to CAR-T therapies in GBM appears to be correlated to a “Hot” TME. Again, surprised we are not. CAR-Ts are genetically engineered T-cells, and just like naturally occurring T-cells, they are susceptible to exhaustion and immunosuppression in a “Cold” TME.
Univalent CAR-Ts are not a winning strategy for GBM…or for most other cancers. Particularly true if the CAR-T is directed against EGFR, and even in the presence of an anti-PD1 checkpoint inhibitor such as pembrolizumab. We will say (YELL!) it again: EGFR predominantly resides on ecDNA in HGGs. Under therapeutic pressure, the EGFR target rapidly disappears via ecDNA fungibility, rendering the therapeutic strategy useless. Until and unless an effective, brain-penetrant ecDNA pathway inhibitor is developed, we would STRONGLY advise against pursuing EGFR-targeted, monovalent approaches regardless of therapeutic agent modality.
Anti-GBM CAR-Ts delivered directly into the brain may be safe and tolerable as a monotherapy, but safety results are variable and depend on individual CAR-T constructs. Moreover, the movement towards combining CAR-Ts with checkpoint inhibitors will increase (S)AEs.
Summary of the Recent anti-GBM CAR-T Published Reports
Below we will break down the key finding from each published report, and offer some perspective.
Locoregional delivery of IL-13Rα2-targeting CAR-T cells in recurrent high-grade glioma: a phase 1 trial [Brown, CE, et al; Nature Medicine (2024); Published: 7-Mar-2024]. This paper describes in detail the results of a well-designed, multi-arm Phase 1 study involving 65 treated patients. The study and corresponding analysis is comprehensive, rigorous and rich with important information. We applaud the City of Hope group that designed and executed the trial. We wish that every clinical study in GBM were this well conceived and supported by sound scientific rationale.
The work was focused on delivering IL-13Rα2-targeting CAR-T cells to rGBM patients. Among the study’s five arms were those that examined methods of manufacturing of the CAR-T cells as well as several routes of intra-tumoral (ICT), intra-ventricular (ICV) and dual ICT+ICV administration of the CAR-T cells to the brain. The study did allow patients who had previously received bevacizumab, which is uncommon for clinical trials. While the percentage of patients for which CAR-T cell production was successful was high (97%), 24 patients from the enrolled cohort (26%) did not receive their manufactured CAR-Ts due to rapid disease progression during the multi-week manufacturing period.
The investigators offered the following key conclusions:
The therapeutic approach was generally safe and tolerable with very few high grade AEs and no ICANS reported.
The best efficacy signals were observed for the dual delivery approach using ICT + ICV administration (Arm 5)
Therapeutic efficacy was strongly correlated with the inflammatory status of the TME and the presence of T-cells therein.
As the authors stated, “Taken together, our findings suggest that a ‘hot’ TME plays a critical role in response to CAR-T therapy”.
The best mOS was documented in the ICT + ICV delivery arm (Arm 5) at 10.2 months. For comparison, the literature supports mOS figures of 5.5 to 12.6 months at first recurrence for existing rGBM therapies, and about 3.4 months for post-bevacizumab treated rGBM patients.
MissionGBM Commentary. Great study. However, one has to ask whether a therapeutic protocol that (i) is neurosurgically-intensive (multiple administrations via ICT + ICV); (ii) requires complex, lengthy and expensive individual patient CAR-T manufacturing; (iii) appears to be highly correlated to the inflammatory status of the TME (which is not influenced by the CAR-T construct); and (iv) results in an mOS extension of perhaps 4-5 months over SoC is compelling? We would expect that the investigators will refine and advance the work going forward (for example, their next trial combines checkpoint inhibitors with the CAR-T cells in an attempt to prevent T-cell exhaustion in the TME NCT04003649), but at the likely expense of significantly elevated (S)AEs.
Intrathecal bivalent CAR T cells targeting EGFR and IL13Rα2 in recurrent glioblastoma: Phase 1 trial interim results [Bagley, SJ, et al; Nature Medicine (2024); Published: 13-Mar-2024]. This paper reports data from the first six patients treated in a Phase 1 trial (NCT05168423) at the University of Pennsylvania using a bivalent CAR-T construct (EGFR + IL-13Rα2). The publication reports significant rates of early-onset neurotoxicity consistent with ICANS, but only modest reductions in tumor size and enhancement with no patient achieving an OR. One must note that this is an interim report with very few subjects at present, so it is best to continue to follow the trial to see how the data matures.
MissionGBM Commentary. We know the investigators for this trial. They are careful and rigorous scientists and clinicians, and we have confidence in their acumen. We actually considered enrolling Julie in the study, but decided to pass after a detailed discussion with the Principal Investigator because (i) the protocol has not yet yielded durable ORs; and (ii) ICANS-like SAEs requiring 1-2 weeks hospitalization (including ICU) were virtually guaranteed.
Intraventricular CARv3-TEAM-E T Cells in Recurrent Glioblastoma [Choi, BD, et al; New Engl. J. Medicine (2024); Published: 13-Mar-2024]. This Brief Report encompasses only three patients with uncompelling data. Honestly, we do not even know why the NEJM published it (no one likes to be scooped by the Nature journals?). The CAR-T construct is EGFRvIII-targeted and made via transduction of autologous T-cells with the CARv3-TEAM-E lentiviral vector containing an anti-EGFRvIII single-chain variable fragment and an intracellular 4-1BB costimulatory domain linked to CD3 zeta. While all three patients showed “dramatic radiological responses” within days of a single ICV infusion of the engineered T-cells, the response in two of the three patients was transient with rapid recurrence of the GBM characterized by EGFRvIII antigen depletion (Same As It Ever Was). One patient showed a more durable response, but the very short follow up period at the time of publication compromises the interpretability of the data. There were irAEs in all patients.
MissionGBM Commentary. This brief report was rushed into press with only three individual patient case studies and a limited post-treatment observation period. Not much else to say.
Can we please stop repeating the same monovalent EGFR-targeted experiment and expecting different outcomes. Every such trial shows transient response, and rapid recurrence with cancer cells displaying EGFR antigen depletion regardless of therapeutic modality.
CAR-T cells deal a blow to deadly brain cancer — but for how long is unclear [Ledford, H. Nature (2024); Published: 13-Mar-2024]. As it often does, Nature elected to publish a companion commentary piece alongside the research reports. The brief article summarizes the above papers, and offers some insightful quotes from Stephen J. Bagley, MD, a lead author on one of the CAR-T trials and respected neuro-oncologist at the University of Pennsylvania (Full Disclosure: We are big fans of Steve Bagley and his clinical and scientific work. He is one of the best.):
The results are promising, but the goal is to generate longer-lasting responses, says Bagley. It was exciting, he says, to watch tumours shrink in the first day after CAR-T therapy. “We hadn’t seen that before,” he says. “We were thrilled.”
But the excitement faded as participants relapsed after treatment: “It’s very humbling to go on that roller coaster ride,” he says. “One week you feel like you’ve made a real difference in their lives, and the next week the tumour is back again.”
Bottom Line
As we and others have written, GBM/HGG tumors are heterogeneous, aggressive and invasive. The tumor heterogeneity makes it imperative that combination, multi-point attacks that engage the patient’s immune system are deployed in therapeutic regimens designed to yield durable ORs (see here for our take on Therapeutic Strategies). As these recent CAR-T reports make clear, monovalent or bivalent CAR-Ts have yet to yield durable ORRs, particularly when the TME is “Cold”. If one hopes to make real progress in treating GBM/HGGs with CAR-T, the logical path forward is to engineer multi-valent CAR-T constructs, and administer them directly into the brain TME (or ventricles) with concomitant checkpoint inhibition agents to prevent the CAR-T cells from becoming exhausted in the “Hot” TME. Several research groups are heading in this direction, and we await their progress, which will likely emerge over the next decade.
Nonetheless, we have to wonder if the CAR-T view is going to be worth the climb?
Successful treatment protocols are likely to:
Require complex, lengthy and expensive manufacturing of multi-valent CAR-T cells on a patient-by-patient basis;
Be neurosurgically-intensive (multiple administrations via ICT + ICV routes);
Require independent flipping of the inflammatory status of the TME to “Hot” (which is not influenced by the CAR-T construct); and
Be co-administered with immune checkpoint inhibitors to prevent the CAR-T cells from becoming exhausted in the TME, which will significantly elevate the (S)AE profile.