Developing A Precision Approach to CAR T-Cell Toxicity Management

CancerNetwork^® spoke with Tiba Al Sagheer, PharmD, BCOP, BCACP, pharmacy quality improvement coordinator for transplant and cellular therapy at Miami Cancer Institute of Baptist Health South Florida, about her presentation titled “Thresholds and Therapeutics: A Precision Approach to CAR T Toxicity Management,” at the 2026 National ICE-T Conference in Orlando, held in July 2026.¹ She discussed where current toxicity grading criteria fall short and how biomarkers and CAR T-cell expansion inform escalation decisions.

Although new guidelines from organizations like the American Society for Transplantation and Cellular Therapy (ASTCT) may help elucidate ways to monitor and mitigate emergent toxicities associated with cellular therapy, Al Sagheer said that a “true threshold” for initiating prophylaxis still does not exist yet. She also noted that differentiating between cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) vs immune effector cell (IEC)–associated hemophagocytic lymphohistiocytosis (HLH)–like syndrome (IEC-HS) and other delayed neurotoxicities represents another ongoing challenge in the field.

CancerNetwork: What was the background or inspiration for your presentation focusing on thresholds and precision approaches to toxicity management related to CAR T-cell therapy?

Al Sagheer: The inspiration for that talk was the data that are emerging on what biomarkers to look for and how that determines what toxicities [will] emerge, and how we pull the trigger. There is a 2026 ASTCT consensus guidance document that goes over what those biomarkers are and what we should be looking at; it stratifies them into something that we must have or should have, can have, and is nice to have. It is nice to have that consensus guidance to guide us in terms of what we should really be doing as routine testing and what should be readily available vs things that should be tested just to keep for research purposes.

Where do you see current CRS or ICANS grading criteria falling short, and what would a more precision-based threshold system look like?

The ASTCT criteria are, in many ways, very helpful right now and solidified. Many institutions are using the same language to grade patients. What we are falling short on is the other toxicities that are not CRS and ICANS: the delayed or non-ICANS neurotoxicities, for instance, or IEC-HS, the enterocolitis-like syndrome that comes with some of those BCMA CAR T-cell toxicities. There are a lot of gaps that we still do not know how to address, and that was something I wanted to investigate with this presentation.

Anakinra (Kineret) appears to be increasingly used for refractory CRS and ICANS. What is driving the shift toward IL-1 blockade, and how do you decide when to escalate from tocilizumab (Actemra) or steroids to anakinra?

The nice thing with anakinra is that we can use it for many of those toxicities, so it plays nicely in the sandbox. Tocilizumab is primarily used for CRS. We try not to go over 3 doses, per se, for patients who develop CRS; in our practice, as soon as a patient receives 2 or more doses, we are usually escalating to other agents because we are not seeing a good response with tocilizumab alone.

In terms of anakinra, we are using it for CRS, and later on for ICANS as well as IEC-HS.

Speaking as a pharmacist, how do you balance early aggressive intervention against the risk of blunting CAR T-cell efficacy by dampening the immune response too soon?

When we say CAR T, there are many products and many targets. We have the CD19 products and the anti-BCMA CAR products. One of the topics I went over is that with the anti-BCMA CAR products—ciltacabtagene autoleucel [cilta-cel; Carvykti] and idecabtagene vicleucel [ide-cel; Abecma]—and CAR expansion, meaning how quickly or robustly the CAR expands in the body, there was no correlation between the efficacy of the CAR product and its expansion for cilta-cel, but there was a correlation between expansion and toxicity. Bigger expansion, more toxicity. But that is not the story with ide-cel; [with] larger expansions, we saw better clinical outcomes.

That is why it is important for us to know whether we are going to restrict the expansion of both products by looking at markers like the absolute lymphocyte count [ALC], which can tell us a patient is going to have greater toxicity. It is not just 1 thing we want to look at; it is multifactorial. There is a model called the NEMO model developed by a group in Seattle that [accounts for] ALC in addition to CRP, platelets, and ferritin at that time point, and stratifies patients to know if they will develop higher-grade toxicities and need treatment.²

What is the biggest gap between what toxicity management guidelines say on paper and what is actually feasible to do or execute at the bedside, especially outside major academic cellular therapy centers?

Not all centers have access to readily available biomarkers. When we are talking about interleukin tests that you can do as a send-out, or even flow cytometry of the CARs themselves, those are nice to have and be able to know, but they are not readily available. You may not necessarily have those results in time to guide treatment. That is the biggest gap that we have, which is why the consensus criteria published this year tell you that you can probably use things that are readily available and that you test every day on your patients, like the CBC, the CRP, and the ferritin; you can get those the same day and be able to trend them. That is the difference that makes it challenging for non-academic centers.

What do you hope those in attendance took away from your presentation?

We still do not have the answer. We do not know a true threshold, whether it is just the ALC or different markers, to tell us when you should pull the trigger for prophylaxis based on the different agents available for preventing delayed neurotoxicity because we do not have good treatments for those delayed neurotoxicities. Another thing is the differentiation between CRS and IEC-HS, which comes down to a ferritin cutoff: slightly above 7000—7420 if I recall correctly—is the cutoff where you want to switch and start treatment for IEC-HS and escalate therapy.

Another thing I did not mention is the difference between CRS and ICANS vs those delayed neurotoxicities, or non-conventional toxicities. With CRS and ICANS, we are seeing more inflammatory markers and cytokines, but with those delayed neurotoxicities, they are more CD4-positive, CAR-driven cells that are infiltrating the tissue; it is a different mechanism or pathophysiology. When biopsies were taken from patients who had those toxicities, those CARs were present there. On that note, pre-apheresis CD4-to-CD8 ratio is also important to obtain, because that can tell you that your collection is going to be skewed more toward CD4, and that patient may be more likely to have toxicities.

References

1. Sagheer TA. Thresholds and therapeutics: a precision approach to CAR T toxicity management. Presented at the 2026 National ICE-T Conference; July 18, 2026; Orlando, FL.
2. Jeon Y, Wu X, Khouderchah C, et al. NEMO: a novel prognostic model of non-ICANS neurotoxicities associated with ciltacabtagene autoleucel. Transplantation and Cellular Therapy. 2026;32(2):S12-S13. doi:10.1016/j.jtct.2025.12.028