CAR-T in prostate cancer: promise and proof

A 68-year-old retired man has metastatic castration-resistant prostate cancer. He has already used hormonal blockade, enzalutamide, docetaxel and cabazitaxel. The prostate-specific antigen has risen again, and the bone scan shows new bone lesions. His son read an article about CAR-T cells against prostate cancer and arrives at the consultation with the question ready: why can't his father have this treatment, if it cures leukemia? The resident accompanying the consultation also wants to know. The scene is fictional and was written for teaching.

This text uses the case as a clinical reasoning scenario for students and health professionals. The topic is CAR-T cell therapy in prostate cancer, based on the first phase 1 studies with targets such as PSCA and PSMA, and the question Dr. Alexei proposed for the series: what separates a promising therapy from an intervention ready for routine use? The warning that accompanies the question is direct: the therapy remains experimental and should not be presented as a proven change in practice. The text does not replace your country's guideline or your service's protocol. If you play the case of the day on NEXO, you will recognize the challenge: separating what is a signal from what is hope.

Key points

• CAR-T cells are the patient's own lymphocytes modified in the laboratory to recognize a tumor protein.
• In prostate cancer, the published studies are phase 1, with few patients, and measured mainly safety.
• In the study with CAR-T directed at PSCA, 4 of 14 patients had a drop in prostate-specific antigen above 30 out of every 100, almost always temporary.
• In the study with CAR-T directed at PSMA, one patient with a deep response died after grade 4 cytokine release syndrome with sepsis.
• Solid tumors impose barriers that leukemias do not: heterogeneous targets, an immunosuppressive environment and toxicity outside the tumor.
• Today, the path for the interested patient is a clinical trial, not supportive use.

Why this is a reasoning scenario

The son's question is reasonable. CAR-T cells have changed the treatment of some blood cancers, and the series has already covered this in the post on CAR-T in myeloma. It is natural to assume that the same technology works in other tumors. Clinical reasoning begins when you ask why this has not happened yet, and what would need to happen for it to happen.

The topic also trains critical reading. Phase 1 studies bring impressive figures: a graph with the drop in prostate-specific antigen, a CT scan with lesions that shrank. These images circulate, and the patient arrives with them. The professional needs to know what a phase 1 study can show and what it was not designed to show. The post on calibration of confidence deals with this adjustment between what is known and how much one believes.

Enthusiasm and proof

There is a known bias in reading innovations: good individual responses draw attention, and absent responses disappear from the account. One patient with a 98 out of every 100 drop in prostate-specific antigen is a strong story. The other twelve in the same study, with little or no response, are a table. The post on heuristics in clinical reasoning calls this availability: the striking case weighs more than the real frequency.

How a CAR-T cell works

The process begins with the collection of lymphocytes from the patient's blood, by apheresis. In the laboratory, these cells receive a gene that encodes a chimeric antigen receptor, an artificial protein that joins the recognition part of an antibody with the signaling parts of a T lymphocyte. The modified cells are multiplied and returned to the patient, usually after mild chemotherapy called lymphodepletion, which makes room for them to expand.

Once in the body, the CAR-T cell recognizes the target protein on the tumor surface without depending on antigen presentation by the immune system, and destroys the cell that carries it. The review by Perera and colleagues summarizes the generations of constructs and the adjustments studied to improve expansion, persistence and safety in prostate cancer (Perera and colleagues, 2022).

Choosing the target

Everything depends on a protein present in the tumor and absent, or nearly absent, in the tissues that need to be preserved. In B-cell leukemia, the CD19 target appears on malignant cells and on normal B cells, and the loss of normal B cells is tolerable. In the prostate, the candidates studied include PSMA, PSCA, STEAP1, STEAP2 and others. Each has its own distribution in normal tissues, and that distribution defines part of the toxicity.

Why the prostate is harder than leukemia

The review by Zarrabi and colleagues notes that checkpoint inhibitors had modest activity in prostate cancer, in a small fraction of patients, which already pointed to a tumor that is barely visible to the immune system (Zarrabi and colleagues, 2023). The same review describes the strategy of arming CAR-T cells against TGF beta, a molecule abundant in the prostate tumor environment that suppresses T lymphocytes (Zarrabi and colleagues, 2023).

The main barriers

• Heterogeneous expression of the target, which varies between lesions and between cells of the same lesion.
• Immunosuppressive tumor environment, with molecules that switch off lymphocytes.
• Bone metastases, where cell access and response assessment are difficult.
• Presence of the target in normal tissues, with risk of damage outside the tumor.
• Slow manufacturing, in a patient with advanced disease and little time.

Target heterogeneity appears clearly in autopsy tissues. In a study with 114 samples of metastases from castration-resistant prostate cancer, about 88 out of 100 had STEAP1 expression, against about 61 out of 100 with PSMA expression, and 32 samples had STEAP1 without PSMA (Bhatia and colleagues, 2023). A target missing in part of the lesions leaves room for the tumor to escape.

The environment that switches off the cells

Even when the CAR-T cell reaches the tumor, it finds inhibitory signals. In the study with PSMA-directed CAR-T, the researchers observed that, in patients whose cells failed, several inhibitory molecules increased in the tumor environment after infusion (Narayan and colleagues, 2022). Preclinical studies try to get around this with low-dose carboplatin to modify the tumor environment (Porter and colleagues, 2023) or with gamma delta T cells combined with zoledronate in bone metastasis models (Frieling and colleagues, 2023). These are laboratory and animal data.

The study with PSCA-directed CAR-T

The study by Dorff and colleagues was the first in humans with CAR-T cells directed at prostate stem cell antigen, PSCA. Fifty-eight men with metastatic castration-resistant prostate cancer were screened for PSCA expression in tissue, 22 underwent apheresis and manufacturing, and 14 received the infusion between 2019 and 2022 (Dorff and colleagues, 2024). The difference between 58 screened and 14 treated already says a lot about the feasibility of the process.

The starting dose was 100 million cells without lymphodepletion. At the second level, with lymphodepletion, two cases of non-infectious cystitis appeared as dose-limiting toxicity, and the protocol began using reduced lymphodepletion (Dorff and colleagues, 2024). Cytokine release syndrome grades 1 or 2 occurred in 5 of 14 patients, and none was grade 3, and there was no severe neurotoxicity attributed to the cells.

What was seen in terms of activity

Four of the 14 patients had a drop in prostate-specific antigen of more than 30 out of every 100 by day 28, and only one kept that drop after that point (Dorff et al., 2024). The median time between apheresis and infusion was 73 days, ranging from 34 to 182 days, and six patients needed bridging therapy. The authors conclude that the treatment, with reduced lymphodepletion, is safe and shows preliminary activity. That is the right language for a phase 1 study.

Toxicity beyond cytokine syndrome

The safety numbers in the study show an effect that does not always appear in the abstracts. Cytokine release syndrome began at a median of day four, and tocilizumab was used in three patients to control fever (Dorff et al., 2024). Grade 3 or higher neutropenia appeared in all 6 patients at the level with full lymphodepletion and in 3 of the 5 with reduced lymphodepletion, versus none of the 3 without lymphodepletion (Dorff et al., 2024). The chemotherapy that prepares the body for the cells is part of the risk, not a technical detail.

Cystitis deserves special attention, because it illustrates a general principle. The authors report that the appearance on cystoscopy was not characteristic, that treatment inside the bladder did not work well and that systemic corticosteroids were needed (Dorff et al., 2024). An unexpected toxicity in an organ that is not the target is exactly the kind of information that only appears when the therapy leaves the laboratory and enters people, and that is why phase 1 exists.

From 58 screened to 14 treated

The path between screening and infusion is also a result. Patients needed a biopsy to confirm expression of the target, apheresis and weeks of waiting for manufacturing, and some depended on bridging therapy with cabazitaxel, cabazitaxel with carboplatin or enzalutamide to reach infusion (Dorff et al., 2024). The delays came from mandatory pauses to assess toxicity, from protocol amendments and from waits for confirmation of the target in a new biopsy. For a man with disease resistant to several lines, every month counts.

This funnel explains why the question about routine use does not depend only on efficacy. Even if a product worked very well, it would need to be manufactured on time, at scale, and given by teams able to manage serious complications. In leukemia and myeloma, this path was built over years, with accredited centers and management protocols. In prostate cancer, it is still at the beginning.

Study	Target	Patients treated	Activity	Notable toxicity
Dorff, 2024	PSCA	14	Antigen drop above 30 out of every 100 in 4, sustained in 1	Dose-limiting cystitis; mild cytokine syndrome
Narayan, 2022	PSMA, with TGF beta resistance	13	Drop above 30 out of every 100 in 4	One death after grade 4 cytokine syndrome and sepsis
Stein, 2024	PSCA, with rimiducid activation	9 with prostate cancer	Two partial responses, one unconfirmed	Two treatment-related deaths in the highest-dose cohort

The study with PSMA-directed CAR-T

The study by Narayan et al. tested PSMA-directed CAR-T cells armed with a dominant-negative TGF beta receptor, to resist suppression by the tumor environment. Eighteen patients were enrolled and 13 received the cells, across four dose levels (Narayan et al., 2022). Five had grade 2 or higher cytokine release syndrome.

The case that draws the most attention sums up the dilemma. One patient had intense clonal expansion of the cells, a drop of more than 98 out of every 100 in prostate-specific antigen and, afterward, grade 4 cytokine release syndrome with respiratory failure and shock. He died 30 days after the infusion, with immunosuppression, enterococcal sepsis and multiple organ failure (Narayan et al., 2022). The deepest response in the study came together with the most severe toxicity.

The third study and the early closure

Another phase 1 study tested cells directed at PSCA with a switch activated by the drug rimiducide, in pancreatic and prostate cancer. Among 9 patients with prostate cancer, there were two partial responses, one of them unconfirmed, but two dose-limiting toxicities and two treatment-related deaths occurred in the highest dose cohort, and the study was closed without defining the dose for phase 2 (Stein et al., 2024). It is a reminder that activity and safety must be looked at together.

Other targets and other strategies

Research continues on several fronts. STEAP1 appeared in more metastases than PSMA in the autopsy samples, and CAR-T cells directed at it, combined with interleukin 12 localized in the tumor, had activity in animal models (Bhatia et al., 2023). STEAP2 is another target with high expression restricted to prostate tumor, and a product armed against TGF beta showed preclinical activity (Zanvit et al., 2023). The same autopsy studies show that a single target is unlikely to cover all lesions.

Bispecific antibodies

An alternative that is already further along is bispecific antibodies, which bind the patient's T lymphocyte to the tumor cell without genetic modification and can be given on an outpatient basis. In a phase 1 study with pasritamig, directed at kallikrein 2, 174 participants were treated, and grade 3 or higher treatment-related events occurred in about 10 out of every 100 (Stein et al., 2025). It is also phase 1, but with a scale and logistics very different from those of cell therapy.

What is already available

For the patient in the scene, what exists outside research are treatments with phase III evidence. One of them is lutetium PSMA, discussed in the post about lutetium PSMA and the VISION study. The conversation about CAR-T should not delay access to these options.

From promising to ready for use

Dr. Alexei's question calls for a structured answer. A promising therapy has a biological signal and some activity in a few patients. An intervention ready for routine use has gone through steps that show benefit in outcomes that matter, with known and manageable toxicity, in a process that can be reproduced outside the center that created it.

The missing steps

• A defined dose and schedule, with acceptable safety in more patients.
• Durable responses, and not just transient drops in the marker.
• Comparison with the available treatment in a randomized study.
• Benefit in survival or quality of life, not only in a marker.
• Reliable manufacturing in a timeframe compatible with the disease.
• Teams trained to recognize and treat toxicity at multiple centers.

Current studies are at the first step. The data themselves explain why: activity in a minority, short responses, treatment-related deaths in two of the three studies and a manufacturing time that reached six months. This is not a failure. It is the honest picture of a phase 1, and it is exactly what Dr. Alexei's warning asks not to hide.

Intermediate markers

The drop in prostate-specific antigen is useful to signal activity, but it does not equal living longer. The post about prognostic reasoning discusses how intermediate outcomes can mislead, and the post about pretest probability helps interpret results when the initial chance is low.

How to talk with the patient

The family in the scene deserves an honest and respectful answer. Curiosity is legitimate, and dismissing it pushes people away. The path is to explain what the technology does, what the studies have shown so far and where it is available, which is within clinical trials with specific criteria. The post about the SPIKES protocol offers a structure for difficult conversations.

What needs to be said

• CAR-T therapy in prostate cancer is experimental and is only offered in studies.
• In the published studies, most patients did not have a durable response.
• There were treatment-related deaths in some studies.
• There are approved treatments that should be considered first or in parallel.
• If there is interest, the team can look for open studies and assess inclusion criteria.

Common reasoning errors

• Transfer the leukemia or myeloma result to prostate cancer.
• Present the marker drop in one patient as proof of efficacy.
• Ignore treatment-related deaths when summarizing the studies.
• Delay approved therapies while waiting for a spot in a trial.
• Call something still in phase 1 a practice change.

How NEXO helps train this reasoning

In NEXO, the case of the day is fictional, with a real clinical presentation and up to six progressive clues, and you propose the diagnosis by looking for the ICD-10 code. Assessing a new therapy is similar: the clues are the data from each study, and the temptation is to close the conclusion with the most striking clue. The post on clinical decision-making helps separate what is available from what is proven, and the post on diagnostic uncertainty helps discuss uncertainty without losing the patient's trust.

Three exercises with cases

Change the source. Say how you would read a report that cites only the patient with a drop of 98 out of 100 on the marker.

Change the phase. Say what would need to appear in a phase 2 study for you to change the conversation with this family.

Change the target. Say why a biopsy of a single lesion may not represent all the metastases.

The limits on using the game are described on the NEXO economy page. If you research the teaching of clinical reasoning, the page for researchers explains how to get in touch. Professionals who write or review cases find information under curation and under creators, and institutions can learn about the proposal on the partnerships page.

Frequently asked questions

Is CAR-T therapy approved for prostate cancer?

No. The published studies are phase 1, with few patients, and the therapy is offered only within clinical trials (Dorff et al., 2024).

How many patients responded in the studies?

In the study targeting PSCA, 4 of 14 had a prostate-specific antigen drop above 30 out of 100, and only one maintained that drop (Dorff et al., 2024). In the study targeting PSMA, it was 4 of 13 (Narayan et al., 2022).

What are the risks?

Cytokine release syndrome, blood cell drop after lymphodepletion, infections and damage to normal tissues that carry the target, such as the cystitis seen with the PSCA target. There were treatment-related deaths (Stein et al., 2024).

Why does it work in leukemia and not in the prostate?

The prostate tumor has heterogeneous targets, an environment that switches off lymphocytes and bone metastases that are hard to reach (Zarrabi et al., 2023).

How to find a study?

The team following the patient can look for open studies in public registries and talk to research centers about inclusion criteria, such as target expression in the tumor.

Take-home messages

• I explained that CAR-T therapy in prostate cancer is experimental.
• I distinguished phase 1 from evidence for routine use.
• I cited activity and toxicity together, including the reported deaths.
• I reviewed the approved options the patient can still receive.
• I offered help to look for clinical trials, if there is interest.
• I considered the manufacturing time against the pace of the disease.
• I recorded the conversation and the patient's preferences.

Want to practice reasoning with progressive clues? Play the case of the day.

References

1. Bhatia V, Kamat NV, Pariva TE, et al. Targeting advanced prostate cancer with STEAP1 chimeric antigen receptor T cell and tumor-localized IL-12 immunotherapy. Nat Commun. 2023. PMID 37041154.
2. Dorff TB, Blanchard MS, Adkins LN, et al. PSCA-CAR T cell therapy in metastatic castration-resistant prostate cancer: a phase 1 trial. Nat Med. 2024. PMID 38867077.
3. Frieling JS, Tordesillas L, Bustos XE, et al. γδ-Enriched CAR-T cell therapy for bone metastatic castrate-resistant prostate cancer. Sci Adv. 2023. PMID 37134157.
4. Narayan V, Barber-Rotenberg JS, Jung IY, et al. PSMA-targeting TGFβ-insensitive armored CAR T cells in metastatic castration-resistant prostate cancer: a phase 1 trial. Nat Med. 2022. PMID 35314843.
5. Perera MPJ, Thomas PB, Risbridger GP, et al. Chimeric Antigen Receptor T-Cell Therapy in Metastatic Castrate-Resistant Prostate Cancer. Cancers (Basel). 2022. PMID 35158771.
6. Porter LH, Zhu JJ, Lister NL, et al. Low-dose carboplatin modifies the tumor microenvironment to augment CAR T cell efficacy in human prostate cancer models. Nat Commun. 2023. PMID 37660083.
7. Stein MN, Dumbrava EE, Teply BA, et al. PSCA-targeted BPX-601 CAR T cells with pharmacological activation by rimiducid in metastatic pancreatic and prostate cancer: a phase 1 dose escalation trial. Nat Commun. 2024. PMID 39737899.
8. Stein MN, Vinceneux A, Robbrecht D, et al. Pasritamig, a First-in-Class, Bispecific T-Cell Engager Targeting Human Kallikrein 2, in Metastatic Castration-Resistant Prostate Cancer: A Phase I Study. J Clin Oncol. 2025. PMID 40450573.
9. Zanvit P, van Dyk D, Fazenbaker C, et al. Antitumor activity of AZD0754, a dnTGFβRII-armored, STEAP2-targeted CAR-T cell therapy, in prostate cancer. J Clin Invest. 2023. PMID 37966111.
10. Zarrabi KK, Narayan V, Mille PJ, et al. Bispecific PSMA antibodies and CAR-T in metastatic castration-resistant prostate cancer. Ther Adv Urol. 2023. PMID 37359737.