PSCA-CAR T cell therapy in metastatic castration-resistant prostate cancer: a phase 1 trial

Nature Medicine

nature medicine

https://doi.org/10.1038/s41591-024-02979-8

Artice

PSCA-CAR T cell therapy in metastatic

castration-resistant prostate cancer:

a phase 1 trial

Tanya B. Dorff

, M. Suzette Blanchard

, Lauren N. Adkins

, Laura Luebbert

Neena Leggett

, Stephanie N. Shishido

, Alan Macias

, Marissa M. Del Real

Gaurav Dhapola

, Colt Egelston

, John P. Murad

, Reginaldo Rosa

Jinny Paul

, Ammar Chaudhry

, Hripsime Martirosyan

, Ethan Gerdts

Jamie R. Wagner

, Tracey Stiller

, Dileshni Tilakawardane

, Sumanta Pal

Catalina Martinez

, Robert E. Reiter

, Lihua E. Budde

, Massimo D’Apuzzo

Peter Kuhn

, Lior Pachter

, Stephen J. Forman

3,6

& Saul J. Priceman

3,6

Despite recent therapeutic advances, metastatic castration-resistant

prostate cancer (mCRPC) remains lethal. Chimeric antigen receptor

(CAR) T cell therapies have demonstrated durable remissions in hematological

malignancies. We report results from a phase 1, first-in-human study of prostate

stem cell antigen (PSCA)-directed CAR T cells in men with mCRPC. The starting

dose level (DL) was 100 million (M) CAR T cells without lymphodepletion

(LD), followed by incorporation of LD. The primary end points were safety

and dose-limiting toxicities (DLTs). No DLTs were observed at DL1, with a DLT

of grade 3 cystitis encountered at DL2, resulting in addition of a new cohort

using a reduced LD regimen + 100 M CAR T cells (DL3). No DLTs were observed

in DL3. Cytokine release syndrome of grade 1 or 2 occurred in 5 of 14 treated

patients. Prostate-specific antigen declines (>30%) occurred in 4 of 14 patients,

as well as radiographic improvements. Dynamic changes indicating activation

of peripheral blood endogenous and CAR T cell subsets, TCR repertoire

diversity and changes in the tumor i

une m

nment were observed

in a subset of patients. Limited persistence of CAR T cells was observed beyond

28 days post-infusion. These results support future clinical studies to optimize

dosing and combination strategies to improve durable therapeutic outcomes.

ClinicalTrials.gov identifier

NCT03873805

Metastatic castration-resistant prostate cancer (mCRPC) is a lethal

disease, causing more than 30,000 deaths in American men each year

Immunotherapy has largely been unsuccessful; both vaccine-based

strategies such as GVAX and Prost-VAC

and immune-checkpoint

inhibition with CTLA-4 and PD-1 inhibitors

have shown limited activ

ity. The only immunotherapy proven to prolong survival in mCRPC is

sipuleucel-T, which is an autologous cellular immunotherapy with ex

vivo incubation of dendritic cells leading to activation against prostate

acid phosphatase

; however, significant improvements are needed for

immunotherapies to effectively target mCRPC.

Reasons for the lack of immunotherapy response in prostate

cancer are multi-fold, including strong immunosuppression in

advanced prostate cancer

that limits both trafficking and effector

T cell function in the local tumor microenvironment. Despite this,

there are unique tumor-associated antigens in mCRPC that are com

monly and robustly expressed, including prostate stem cell antigen

Received: 2 October 2023

Accepted: 5 April 2024

Published online: xx xx xxxx

Check for updates

A full list of affiliations appears at the end of the paper.

e-mail:

tdorff@coh.org

;

spriceman@coh.org

Nature Medicine

Artice

https://doi.org/10.1038/s41591-024-02979-8

Results

Clinical trial design and patient characteristics

City of Hope conducted a single-center, first-in-human, phase 1 clinical

trial to evaluate safety and bioactivity of PSCA-directed CAR T cells in

patients with mCRPC (

NCT03873805

). The primary end points were

safety and dose-limiting toxicities (DLTs). The secondary end points

were expansion and persistence of CAR T cells to 28 days post-infusion

(defined as CAR T cell percentage of total CD3

cells in peripheral blood

by flow cytometry, or at least 7.5 copies per ug of DNA in peripheral

blood by qPCR), disease response (prostate-specific antigen (PSA)

decline and Response Evaluation Criteria in Solid Tumors (RECIST))

and survival described as percent of participants alive at 6 months.

Exploratory end points were phenotypes and frequencies of immune

cell subsets in the peripheral blood pre- and post-therapy, serum

(PSCA) and prostate-specific membrane antigen (PSMA), which

could be leveraged as targets for powerful cellular immunotherapy

modalities. The dramatic successes of chimeric antigen receptor

(CAR) T cell therapies in hematological malignancies have inspired

the clinical development of CAR T cell therapies for the treatment

of mCRPC.

PSCA is highly expressed in prostate cancer and increases with

advanced disease states, particularly in the setting of bone metasta-

ses

. Using xenograft and syngeneic tumor models, we demonstrated

safety and efficacy of second-generation PSCA-CAR T cells with 4-1BB

co-stimulation in eradicating bone metastatic prostate cancer

Here, we report results of our first-in-human phase 1 clinical trial to

evaluate the safety and bioactivity of PSCA-CAR T cells in patients

with mCRPC.

Participants consented and screened

for PSCA expression by IHC

(

n

= 58)

Participants enrolled and leukapheresis

(

n

= 22)

Participants received CAR T cell infusion

(

n

= 14)

DL1: 100 M

PSCA-CAR T cells

(

n

= 3)

DL2: LD + 100 M

PSCA-CAR T cells

(

n

= 6)

DL3: reduced LD + 100 M

PSCA-CAR T cells

(

n

= 5)

Cohorts

Withdrew consent after amendment

reducing LD (

n

= 1)

Patient deterioration in status, did not

meet eligibility (

n

= 7)

Low PSCA expression (

n

= 4)

Patient decision to pursue other

treatment or deterioration in status

(

n

= 32)

Leukapheresis

CAR T cell

infusion

LTFU

at 1 year

Monthly

–4

–3

Participant

consent

Screening

for PSCA

Imaging

Disease staging and correlatives

T cell manufacturing

12–17 days

Fig. 1 | Clinical trial design and CONSORT diagram.

, Illustration of clinical

trial design, including participant screening, leukapheresis, PSCA-CAR T

cell manufacturing, pre-infusion biopsy (BX), peripheral blood (PB) sample

collection before LD, bone scan and CT imaging, Flu/Cy LD, PSCA-CAR T cell

infusion, serial PB sample collection time points from day 0 to day 28, post-

infusion bone scan and CT imaging, post-infusion BX and long-term follow up

(LTFU).

, CONSORT diagram detailing participants consented and screened for

PSCA expression by immunohistochemistry (IHC) (

= 58), participants enrolled

and leukapheresis (

= 22), participants received CAR T cell infusion (

= 14). DL

cohorts, including DL1 (100 M PSCA-CAR T cells,

= 3), DL2 (Flu/Cy LD + 100 M

PSCA-CAR T cells,

= 6) and DL3 (reduced Flu/Cy LD + 100 M PSCA-CAR T cells,

= 5). CT, computed tomography.

Nature Medicine

Artice

https://doi.org/10.1038/s41591-024-02979-8

cytokine profile before and after CAR T infusion to assess potential

cytokine release syndrome (CRS) toxicity and CAR T cell effector func

tion, phenotype of tumor-infiltrating lymphocytes, gene expression

(by RNA sequencing) of circulating tumor cells (CTCs), cell-free DNA

(cfDNA) in peripheral blood by whole-exome sequencing and CAR

immunogenicity (anti-PSCA-CAR antibodies). The gene expression

in CTCs, cfDNA in peripheral blood and CAR immunogenicity are not

presented in this report.

The clinical trial design is summarized in Fig.

. Fifty-eight par

ticipants were screened for PSCA expression by immunohistochem

istry (Extended Data Table 1 details the PSCA scores), 22 participants

underwent leukapheresis and CAR T cell manufacturing and 14 par

ticipants were treated from August 2019 to July 2022 (Fig.

). The

first participant was pre-screened (tissue testing) on 23 May 2019,

the first participant was enrolled (leukapheresis) on 30 July 2019 and

the last participant was treated (CAR T cell infusion) on 25 July 2022;

the trial was then closed. Clinical and disease characteristics of the 14

participants are listed in Table

. The median age of participants in the

study was 62 years for dose level (DL)1, 70 years for DL2 and 69 years

for DL3. All participants received previous androgen receptor signal-

ing inhibitors, either enzalutamide (71%), abiraterone (79%) or both

(64%), and a majority of patients received cabazitaxel (57%), docetaxel

(86%) or both (57%) before CAR T cell infusion. Baseline PSA (median)

ranged from 16.5 to 235.3.

CAR T cell product manufacturing and characterization

The PSCA-CAR construct comprised the anti-PSCA humanized scFv

(A11 clone), ΔCH2 extracellular spacer, CD4 transmembrane domain,

4-1BB intracellular co-stimulatory domain and CD3γ cytolytic domain

as previously published (Supplementary Fig. 1)

. In brief, CAR T cell

manufacturing included depletion of CD14

and CD25

cells, CD3/28

bead stimulation, transduction with lentivirus at multiplicity of infec

tion of 0.3, removal of beads at day 7–9, followed by expansion for a

total of 12–17 days in interleukin (IL)-2 and IL-15 cytokines. There were

no manufacturing failures, with a median CAR percentage of 86.8% in

the final released product. Thawed products were characterized by

flow cytometry for expression of CD4/CD8, CD19 (for CD19t transduc

tion marker) expression and Fc (PSCA-CAR) expression (Extended

Data Fig. 1a), as well as T cell subsets demonstrating a dominant T

phenotype (Extended Data Fig. 1b). Two products fell outside the

prespecified woodchuck post-transcriptional regulatory element

(WPRE) copy number (<5) and US Food and Drug Administration (FDA)

approval was granted to proceed with the infusion. The median time

from leukapheresis to infusion of the product was 73 days (range

34–182); delays were primarily due to protocol-mandated holds

on accrual during toxicity assessments and protocol amendments,

waiting for confirmatory PSCA staining from on-study biopsies, as

well as seeking regulatory approval for the use of the product out

of parameters (as specified above). Six patients received bridging

therapy: cabazitaxel (

= 4), cabazitaxel + carboplatin (

= 1) and

enzalutamide (

= 1).

Table 1 | Patient characteristics

DL1: 100 M

CAR T cells,

= 3

DL2: LD + 100 M

CAR T cells,

= 6

DL3: reduced

LD + 100 M CAR

T cells,

= 5

Age (years) median

(range)

62 (59–69)

70 (42–73)

69 (62–72)

Race:

White

3 (100)

6 (100)

3 (60)

Black

2 (40)

Asian

Other/unknown

Previous treatment:

Enzalutamide

3 (100)

5 (83)

2 (40)

Abiraterone

3 (100)

6 (100)

2 (40)

Both

3 (100)

5 (83)

1 (20)

Previous treatment:

Docetaxel

2 (67)

5 (83)

5 (100)

Cabazitaxel

2 (67)

3 (50)

3 (60)

Both

2 (67)

3 (50)

3 (60)

Baseline PSA median

(range)

16.5

(10.7–20.4)

88.0

(11.7–590.2)

235.3

(1.79–3,260)

Lymph node only

2 (67)

1 (17)

Bone ± lymph node

4 (67)

4 (80)

Visceral

1 (33)

1 (17)

1 (20)

DL1

DL2

DL3

Months

DL1

DL2

DL3

143.95

–94.58

UPN 394 (DL3)

Before infusion

1 month after infusion

–100

–50

100

150

27.91

15.75

6.65

5.12

2.16

1.72

–5.00

–14.48

–17.86

–42.67

–43.39

–52.73

Change from baseline (%)

Death

Lost to follow-up

PI decision

Required disallowed therapy

Survival after progression

Stable disease

CAR T cell infusion to eval

Fig. 2 | Treatment response following PSCA-CAR T cell infusion.

, PSA

waterfall plot showing best PSA response in the 28 days following CAR T cell

infusion at each DL.

, Swimmer’s plot depicting response to treatment and

follow up for each participant on study. PI, principal investigator.

, CT scan of

a patient (UPN394) in DL3 showing liver metastases before infusion and disease

response 1 month after infusion of PSCA-CAR T cells.

Nature Medicine

Artice

https://doi.org/10.1038/s41591-024-02979-8

Treatment response

Declines in PSA levels from before treatment to day 28 after CAR T cell

infusion were seen in 1 of 3 participants in DL1, 3 of 6 participants in

DL2 and 3 of 5 participants in DL3 (Extended Data Table 2). A waterfall

plot of the maximum PSA change from before CAR T cell infusion to

day 28 shows 4 of 14 participants with PSA declines >30% (Fig.

). Of

these, only one individual maintained a PSA decline >30% beyond 28

days. In DL1, one of three participants treated experienced a transient

PSA response; notably this individual had evidence of early neuroen-

docrine (NE) expression in the on-study biopsy but still retained strong

PSCA expression and the RECIST response was progressive disease.

Post-treatment biopsy revealed further NE transformation (data not

shown). The first participant treated in DL2 (with lymphodepletion)

achieved a >90% PSA decline in the first 28 days post-CAR T cell infusion.

The response in this individual is characterized in greater detail below.

RECIST best objective responses are displayed in Extended Data

Table 2; rates of stable disease by RECIST were 0% (DL1), 67% (DL2)

and 60% (DL3). Swimmer plots for treated participants are shown

in Fig.

, with a 33%, 67% and 40% 6-month survival rate in DL1, DL2

and DL3, respectively (Extended Data Table 2). The first participant

treated in DL3 achieved radiographic improvement in liver metastatic

burden, but did not achieve a PSA response (Fig.

). One individual

with bone only disease who exhibited stable disease in DL3 requested

and received a second infusion of 100 million (M) CAR T cells about 6

months following initial infusion. He experienced transient relief of

cancer-related pain after the second infusion.

We also evaluated treatment response by CTC quantification in the

peripheral blood of treated participants on study using high-definition

single-cell analysis (HDSCA)

–

. Cytokeratin (CK)-positive cells were

detected in the peripheral blood of 100% of treated individuals. The

median and mean changes in CTCs are depicted in Supplementary

Table 1. Overall, there were marked declines in mean CK

cells from

baseline to 28 days after CAR T cell infusion in both lymphodepletion

(LD) cohorts (DL2 and DL3), but not in DL1.

Somatic DNA sequencing results were available for eight indi

viduals and two individuals had germline testing results (no overlap

between somatic and germline-tested patients). Of the eight with

somatic testing, the highest tumor mutational burden was 10.5, all oth

ers were deemed low. PTEN loss was noted in three of these individuals,

one of whom experienced the greatest PSA decline on study (Fig.

G-CSF

IFNy

IL-2R

IL-4

IL-6

IL-8

IP-10

100

Relative cytokine levels

(

old change from

bas

eline to

peak)

PSA

respons

DL1

DL2

DL3

0.1

100

(days)

CAR T cells (%

f CD3

in PB)

–2

–1

DL2 versus DL3

DL1 versus DL3

DL1 versus DL2

Mean dierence log

maximum

copy number per g DNA

G-CSF

IFNy

IL-2R

IL-4

IL-6

IL-8

IP-10

100

Relative cytokine levels

(

old

change from

bas

eline to

peak)

1/2

CRS grade

= 0.013

= 0.026

= 0.007

= 0.009

Fig. 3 | CAR T cell kinetics and serum cytokine analysis.

, Flow cytometric

analysis of PSCA-CAR T cells (%) within CD3

T cells in the PB for each individual.

, PSCA-CAR T cells in PB detected by quantitative PCR of WPRE copies per

μg DNA. Data shown are 95% confidence limits for the mean differences of the

three DLs in log

maximum copy number per μg of DNA.

, Box plots of relative

cytokine levels for each participant grouped by CRS grade 0 or 1/2.

, Box plots of relative cytokine levels for each individual grouped by PSA

response <0 or >0. Data are from

= 14 and presented as mean values ± s.e.m.

values indicate differences between groups using a two-tailed Student’s

-test.

NS, not significant.

Nature Medicine

Artice

https://doi.org/10.1038/s41591-024-02979-8

In DL3, the individual with radiographic improvement in liver metas-

tases had a genomic alteration in CDK12 and he had progressed on

previous immune-checkpoint inhibitor therapy.

CAR T cell persistence and bioactivity

CAR T cell persistence/expansion kinetics in peripheral blood was

assessed by flow cytometry of CD3

T cells coexpressing the CD19t

lentiviral transduction marker. CAR T cell expansion was enhanced by

fludarabine/cyclophosphamide (Flu/Cy) LD (DL2 and DL3 versus DL1)

and not significantly impaired with the reduced LD regimen in DL3

compared to DL2 (Fig.

and Supplementary Fig. 2a) acknowledging

that the sample size is too limited to establish equivalence of DL2 and

DL3. Increased expression of the exhaustion/activation marker PD-1,

activation marker CD25 and memory antigen CD95 among CAR T cells

was observed in peripheral blood in DL2 and DL3, which was greater

than in DL1 (Supplementary Fig. 2b–d). We confirmed CAR T cell kinet

ics in peripheral blood by quantitative PCR for the presence of the

WPRE within the lentiviral cassette, showing increased copies per μg

DNA in individuals in the Flu/Cy LD cohorts (DL2 and DL3) compared

to those without LD (DL1), which were not different from DL2 and DL3

(Fig.

). Despite the minimal observed differences between DL2 and

DL3 in terms of expansion of CAR T cells in peripheral blood (Fig.

reduced LD was associated with reduced CRS-related and off-tumor

toxicities (detailed below).

Toxicity assessment

Rates of DLT and CRS events are detailed in Supplementary Table 2, with

two DLTs in DL2, and a maximum grade 2 CRS among the 14 participants

on study. There were no DLTs in the first three participants treated at

DL1 (without LD). In DL2, there was one DLT (non-infectious cystitis),

resulting in enrollment of three additional participants at this DL.

A second DLT cystitis event was observed in the next three participants.

The cystoscopic appearance of cystitis was not distinctive. Intravesical

therapy was not particularly effective and systemic steroid therapy

was required in both DLT cases, after which the protocol management

team met and amended the protocol to attempt to reduce the toxicity

risk. Changes at DL3 included reducing the dose of LD chemotherapy

cyclophosphamide from 500 to 300 mg m

−2

, with mandatory mesna,

more frequent monitoring of urinalysis and enhanced guidance on

management of cystitis. DL3 received the same dose of 100 M CAR

T cells with these modifications and no DLT was seen in five participants;

the maximum grade of cystitis was grade 2.

CRS occurred in 5 of 14 participants (36%), 1 in DL1 (33%), 2 in DL2

(33%) and 2 in DL3 (40%) (Supplementary Table 2). The median onset

of CRS was 4 days (range 3–8 days). Tocilizumab was administered in

three individuals to mitigate symptomatic fever, although none of the

CRS events reached grade 3 severity. No high-grade CAR T cell related

neurologic toxicities or macrophage activation syndrome events were

noted. Grade ≥3 neutropenia was noted in 0 of 3 (0%) at DL1, 6 of 6

(100%) individuals at DL2 and 3 of 5 (60%) at DL3. Attributed adverse

events by category, MedDRA code and DL are summarized in Table

Induction of serum cytokines was significantly associated with clinical

CRS grade (Fig.

and Extended Data Fig. 2). Serum cytokine induction

was not significantly associated with PSA response (Fig.

Patient with biochemical and radiographic response

One participant in DL2 experienced a >90% PSA decline following

CAR T cell infusion, from 64.2 ng ml

−1

before LD and CAR T cells to

3.5 ng ml

−1

at day 28 after CAR T cell infusion (Fig.

). Radiographic

improvement was seen in this individual’s soft tissue metastasis (Fig.

)

although the RECIST assessment was ‘stable disease’ due to the pres

ence of bone metastases. Changes in serum cytokines in this individual

demonstrate pronounced but transient induction of inflammatory

factors, including interferon (IFN)γ, IL-6, granulocyte–macrophage

colony-stimulating factor (GM-CSF), IP-10 and MIG (Extended Data

Fig. 3a). Serum chemistry showed mild increases in C-reactive protein

(max 81 mg l

−1

), ferritin (max 555 ng ml

−1

), alanine transaminase (ALT)/

aspartate transaminase (AST) (<1.5 × upper limit of normal (ULN)),

lactate dehydrogenase (max 365 U l

−1

) and alkaline phosphatase (max

192) (Extended Data Fig. 3b–g) following CAR T cell infusion. This cor

responded to grade 2 CRS with a

max

of 39.1 on day 4, 38.8 on day 5

and tocilizumab was administered on day 6 due to persistent rigors

without fever; all aforementioned labs subsequently trended down by

day 21. CTC assessment with CK positivity was significantly reduced,

both in bone marrow (Fig.

) and in peripheral blood (Supplemen

tary Fig. 3) samples, from baseline to 28 days after CAR T infusion.

The post-CAR T cell infusion bone metastasis biopsy showed reduc

tions in PSCA

disease (Extended Data Fig. 4a) and Ki-67

expression

(Extended Data Fig. 4b), along with greater infiltration of CD3

and

cytotoxic CD8

T cells by immunofluorescence staining (Fig.

). Few

residual tumor cells in the post-treatment biopsy were observed and

were associated with increased granzyme B

and PD-L1

areas, sug

gestive of an active antitumor immune response. Quantification of

immunofluorescence staining showed increased CD8

and PD-L1

areas

in this individual, with variable results from other individuals analyzed

(Extended Data Fig. 5a,b). Notably, UPN388 also had a biopsy-proven

prostate cancer metastasis in the pancreas that necessitated stent

placement before study entry; this completely resolved after CAR T cell

infusion (Fig.

Immune landscape changes in patient with biochemical

response

Endogenous and CAR T cell populations in peripheral blood were fur-

ther characterized by flow cytometry, as well as by single-cell RNA

sequencing (scRNA-seq) and TCR repertoire analysis in this patient.

Initial assessment of T cell subsets in peripheral blood pre- and post-LD

Table 2 | Adverse events

Arm