Phase I Study of ROR1-Specific CAR-T Cells in Advanced Hematopoietic and Epithelial Malignancies

The genetic modification of T cells to express a tumor-specific T-cell receptor or a chimeric antigen receptor (CAR) is a novel therapeutic approach for cancer (1, 2). CAR-T cells that target CD19 on B-cell malignancies, or B-cell maturation antigen on multiple myeloma, can induce rapid and deep remissions in relapsed or refractory disease and are now approved for these indications (3–7). In contrast, CAR-T cells specific for antigens on solid tumors have shown limited antitumor efficacy (8–10). Several mechanisms may contribute to the poor antitumor activity of CAR-T–cell therapy in solid tumors, including heterogeneity in antigen expression, an immunosuppressive tumor microenvironment, and the propensity of solid tumors to induce T-cell dysfunction (8, 11, 12). Additionally, unlike targeting CD19 and B-cell maturation antigen, in which elimination of normal B cells and plasma cells is tolerated, antigens on solid tumors are often expressed on normal epithelial cells; hence, there is a significant risk that severe toxicities may occur (13–19).

The receptor tyrosine kinase–like orphan receptor 1 (ROR1) is highly expressed by several epithelial cancers, including triple-negative breast cancer (TNBC) and non–small cell lung cancer (NSCLC; refs. 20–22), and in some B-cell malignancies, including chronic lymphocytic leukemia (CLL), mantle cell lymphoma, and acute lymphoblastic leukemia (23–26). Normal tissue expression of ROR1 in adults is restricted primarily to the parathyroid, a subset of B-cell progenitors and alveolar type 1 cells (20, 27). Adoptively transferred ROR1 CAR-T cells mediate antitumor activity in immunodeficient mice engrafted with ROR1⁺ human tumors, and the administration of autologous ROR1 CAR-T cells did not cause toxicity in nonhuman primates, in which the expression of ROR1 on normal tissues is similar to that in humans (20, 28–30). Based on these preclinical data, we performed a phase I study to evaluate the safety of ROR1 CAR-T cells in patients with CLL, TNBC, and NSCLC. This trial allowed analysis of the in vivo behavior and antitumor activity of CAR-T cells that target the same antigen expressed on a hematologic malignancy known to be responsive to T-cell therapy and in two solid tumors that are largely resistant to T-cell therapies.

A phase I open-label study (NCT02706392) was performed at the Fred Hutchinson Cancer Center after approval by the Institutional Review Board. The primary objective was to assess the safety of ROR1 CAR-T cells administered to patients with B-cell malignancies (cohort A) or with TNBC or NSCLC (cohort B). Inclusion criteria for patients with CLL were ROR1 expression on tumor cells measured by IHC or flow cytometry and disease progression after chemoimmunotherapy and/or kinase inhibitors. Patients with TNBC and NSCLC were eligible if they had stage IV disease measurable by CT or PET, prior treatment with at least one line of therapy, and ROR1 expression by IHC on greater than 20% of cells in the primary tumor or metastasis. Complete eligibility criteria are listed in the trial protocol. This report incorporates data from all patients who received ROR1 CAR-T cells from March 2016 through September 2021. The study was conducted with approval of the Fred Hutchinson Cancer Center Institutional Review Board and in accordance with the Declaration of Helsinki. Written informed consent was obtained from each subject.

Tissues from tumor biopsies were formalin fixed and paraffin embedded per clinically validated standard operating procedures (SOPs), and 4-μm-thick unstained tissue sections were used for ROR1 IHC. Tissue slides underwent antigen retrieval and staining with primary anti-ROR1 recombinant mouse mAb, clone 6D4 (Fred Hutchinson Cancer Center). Reactivity staining was enhanced with either polymer treatment for 30 minutes or secondary goat anti-mouse UltraPolymer HRP antibody (polyclonal, Leica Biosystems, Cat. # 2MH-050, RRID: AB_3371707). Slides were counterstained by hematoxylin. ROR1 expression was interpreted by board-certified pathologists and graded on % positive tumor cells, staining characteristics (membrane, nuclear, cytoplasmic, and nonspecific), and intensity (dim, moderate, or strong). During the clinical study, uncertain cases were reviewed by a second pathologist or at consensus conference prior to sign out. The results are re-reviewed and confirmed prior to this publication.

CD8⁺ and CD4⁺ T cells were enriched from leukapheresis products obtained from each patient using the CliniMACS Plus System (Miltenyi) and subsequently activated for 5 days using Human T-Activator CD3/CD28 Dynabeads (BEAD, Thermo Fisher Scientific, Cat. # 11131D, RRID: AB_2916088) at a 3:1 (BEAD:T cell) ratio in complete T lymphocyte (CTL) media comprised of RPMI 1640, 10% human serum, 1 mmol/L L-glutamine, and 50 μmol/L β-mercaptoethanol supplemented with 50 U/mL human IL-2 (Proleukin, Prometheus). Lentiviral transduction of T cells with a CAR comprising the ROR1-specific rabbit single-chain variable fragment (scFv; R12) linked to 4-1BB and CD3ζ signaling domains and a truncated EGFR (EGFRt) transduction marker was performed 24 hours after T-cell activation (31, 32). CAR-transduced T cells were expanded for 14 days, formulated to a product containing a (1:1) ratio of (CD8:CD4) EGFRt⁺ CAR-T cells, and adoptively transferred by intravenous infusion to patients in escalating doses.

All patients in cohort A and 11 of 18 (61.1%) in cohort B received lymphodepleting chemotherapy with 300 mg/m² cyclophosphamide (Cy) and 30 mg/m² fludarabine (Flu) administered on day 4, day 3, and day 2 prior to CAR‐T–cell infusion on day 0. A subset of seven (38.9%) cohort B patients received 100 mg/m² oxaliplatin (Ox) administered on day 4 and 500 mg/m² Cy administered on day 4, day 3, and day 2. Ox/Cy was evaluated in a subset of patients in Cohort B based on data obtained in an immunocompetent murine lung cancer model, which showed that Ox/Cy activated tumor-infiltrating macrophages to produce T-cell–recruiting chemokines that promoted ROR1 CAR‐T–cell infiltration and improved antitumor activity (33). Four dose levels (DL) of EGFRt⁺ CAR-T cells were studied: 3.3 × 10⁵ (DL1), 1 × 10⁶ (DL2), 3.3 × 10⁶ (DL3), and 1 × 10⁷ (DL4) cells/kg body weight. A continuous reassessment method in which patients were treated in groups of two with a maximum increase of one dose level between groups was used to estimate the MTD, defined as a true dose-limiting toxicity (DLT) of 25%. One patient in cohort A and three patients in cohort B received a second CAR-T–cell dose either without (n = 1) or with (n = 3) Cy/Flu lymphodepletion.

Adverse events were captured using Common Terminology Criteria for Adverse Events, version 4.0 through 28 days after CAR‐T–cell infusion. DLT was defined as grade ≥3 nonhematologic toxicity occurring within 30 days of CAR‐T–cell infusion in any major organ system that is attributed to T-cell infusion and unresponsive to dexamethasone or tocilizumab. Cytokine release syndrome (CRS) and immune effector cell–associated neurotoxicity syndrome were assigned using the Lee and American Society of Transplantation and Cellular Therapy consensus grading (34, 35). The disease response in cohort A was measured by iwCLL criteria, and that in cohort B was assessed by RECIST 1.1 (36, 37). Measurement of cytokine levels and T-cell immune responses to the ROR1 CAR was performed as described.

Blood samples were obtained from patients before and at intervals after CAR‐T–cell infusion, and flow cytometry was performed to identify CD4⁺ and CD8⁺ CAR-T cells as viable CD45⁺CD3⁺CD4⁺CD8⁻EGFRt⁺ or CD45⁺CD3⁺CD4⁻CD8⁺EGFRt⁺ events, respectively. Flap-EF1 copy number per microgram DNA of total peripheral blood mononuclear cells (PBMC) from indicated time points prior to and after infusion was assessed using qPCR. WPRE copy number per milligram of DNA from eluates of tumor biopsy tissue sections was quantified using qPCR (38).

Cells were stained using Live/Dead Fixable Dye in PBS for 20 minutes at 4°C and subsequently stained in Flow Buffer (PBS, 0.5% FBS, and 3.5 μmol/L EDTA) with antibodies to cell-surface markers for 30 minutes at 4°C. For intracellular staining, cells were permeabilized for 30 minutes at room temperature in freshly prepared Perm Buffer (Foxp3 Transcription Factor Fixation kit, eBioscience, # 00-5523-00) and incubated for up to 2 hours at 4°C with antibodies specific for cytokines or transcription factors. Human reactive antibodies were purchased from BioLegend unless otherwise specified: Erbitux-PE (R&D Systems; Cat. # FAB9577P, RRID: AB_2942015, 1:100), αCD3 PE-Cy7 (Cat. # 300419, RRID: AB_2621827, 1:400), αCD4 PerCP-Cy5.5 (BD Biosciences; Cat. # 560650, RRID: AB_1727476, 1:66), αCD8 APC-Cy7 (Cat. # 344714, RRID: AB_2044006, 1:100), αCD45 BV510 (Cat. # 368526, RRID: AB_2687377, 1:25), αCD8 BUV496 (BD Biosciences; Cat. # 612943, RRID: AB_2916884, clone RPA-T8; 1:1,600), αCD4 BUV805 (BD Biosciences; Cat. # 612888, RRID: AB_2870177, clone SK3; 1:80), αEGFR BV421 (Cat. # 352991, RRID: AB_2562213, 1:100), αCCR7 BV750 (Cat. # 353254, RRID: AB_2800945, 1:40), αCD127 APC (Cat. # 351316, RRID: AB_10900804, 1:25), αCD45RO BUV395 (BD Biosciences; Cat. # 564291, RRID: AB_2744410, 1:50), αCD45RA AF700 (BD Biosciences; Cat. # 560673, RRID: AB_1727496, 1:200), αCD27 BV786 (BD Biosciences; Cat. # 740890, RRID: AB_2740539, 1:600), αCD95 BV605 (Cat. # 305628, RRID: AB_2563825, 1:100), αCD70 PerCP-Cy5.5 (Cat. # 355108, RRID: AB_2562479, 1:100), αCD25 BUV563 (BD Biosciences; Cat. # 612919, RRID: AB_2870204, 1:800), αCD69 BUV615 (BD Biosciences; Cat.# 751501, RRID: AB_2875497, 1:400), αCD122 PE-Cy7 (Cat. # 339014, RRID: AB_2562597, 1:50), αCD57 BB515 (BD Biosciences; Cat. # 565285, RRID: AB_2739155, 1:800), αCD39 BV711 (Cat. # 328228, RRID: AB_2632894, 1:200), αHLA-DR BUV661 (BD Biosciences; Cat. # 612981, RRID: AB_2916889, 1:200), αPD-1 APC/Fire810 (Cat. # 621620, RRID: AB_2910488, 1:20), αTIM3 BUV737 (BD Biosciences; Cat. # 748820, RRID: AB_2873223, 1:600), αTIGIT APC/Fire750 (Cat. # 372708, RRID: AB_2632755, 1:25), αLAG3 BV650 (Cat. # 369316, RRID: AB_2632951, 1:20), αTCF1 PE (BD Biosciences; Cat. # 564217, RRID: AB_2687845, 1:100), and αGZMB PE-CF594 (BD Biosciences; Cat. # 562462, RRID: AB_2737618, 1:800). Sample acquisition was performed on a BD Biosciences FACSymphony A5 instrument using BD FACSDiva software (v9.1; BD Biosciences, RRID: SCR_001456). Data were analyzed using FlowJo software (v10.8.1; BD Biosciences, RRID: SCR_008520).

Alternatively, apheresis product (PRE), infusion product (IP), and peak expansion after infusion (PEAK) samples were acquired on the Sony ID7000 Spectral Cell Analyzer using ID7000 Spectral Cell analyzer software (v2.0; Sony). A repeated control sample was included for each of the five sample acquisition batches. CD8⁺ T cells were gated from PRE samples, whereas CD8⁺EGFRt⁺ CAR-T cells were gated from IP and PEAK samples using FlowJo software before clustering. We adjusted for moderate batch effects by (i) aligning, for each separate marker, the mean arcsinh (mean fluorescence intensity/1500) values of the repeated controls from each batch and then (ii) applying the respective marker-specific adjustments per batch to the samples of each batch (39). Phenograph with the Leiden algorithm (RRID: SCR_016919) was then used to cluster 10,000 cells per sample (no repeated controls; K = 40, Leiden resolution parameter = 1.5; refs. 40, 41). Dimensionality reduction for visualization of single-cell flow data was performed using Uniform Manifold Approximation and Projection for Dimension Reduction (42). R version 4.3.1 was used for this analysis, and R markdown scripts are provided.

Peripheral blood concentrations of lactate dehydrogenase, calcium, and glucose were evaluated by standard laboratory tests. Serum cytokine concentrations were evaluated by Luminex assay (Luminex Corporation) according to the manufacturer’s instructions.

Responder T cells in pre- (day 0) and postinfusion (day 27 and later) PBMCs were cultured with 3,500 rad–irradiated, EGFRt-enriched R12 CAR stimulators, which have been generated using the rapid expansion method as described (Patent CA2198633C). Cells were co-cultured at a (responder: stimulator) ratio of (1:2) in RPMI supplemented with 2-[4-(2-hydroxymethyl)piperazin-1-ethanesulfonic acid] (HEPES) buffer, penicillin/streptomycin, L-glutamine, β-mercaptoethanol, and human serum (10%). On day 3 of co-culture, cells were fed with 20 U/mL human IL-2. On day 7 of co-culture, responder T cells were washed, counted, and prepared for a second stimulation cycle cultured with 3,500 rad–irradiated autologous PBMCs and 3,500 rad–irradiated EGFRt-enriched R12 CAR stimulators at a ratio of (1:2:0.5). Responder T cells were fed on days 9 and 11 with 20 U/mL human IL-2. On day 13, responder T cells were used as effectors in ⁵¹Cr-release killing assay, as previously described (43), in which unmodified and CAR-modified autologous T cells were used as target cells.

Descriptive statistics include median and range (minimum, maximum) for continuous variables and counts and/or percentages for categorical variables. Box plots show differences between groups, with P values calculated using the nonparametric tests (Wilcoxon for two-group comparison and Kruskal–Wallis for multigroup comparisons) and where applicable, adjusted for multiple comparisons. Spaghetti plots display longitudinal data over time, with separate (locally estimated scatterplot smoothing (LOESS) curves for low and high CRS grades (0–1 vs. 2–3). Reported q-values in cluster-specific box plots are adjusted for multiple comparisons, after Kruskal–Wallis tests on each cluster. A partially overlapping t test was used on incompletely paired sample means for selected markers, with violin plots showing single-cell level marker distributions. Data were analyzed using R Studio (R version 3.6.2).

The Supplementary Data S1 contains the study protocol. Data were available in the main text or supplementary figures, and additional data can be obtained from the corresponding authors upon formal request. The original code for flow cytometry analysis is accessible on the GitHub repository under (https://github.com/hughmacmillan/P9330_flow_analysis) or Code Ocean repository under capsule (https://codeocean.com/capsule/6380046/). All illustrations are original and created with BioRender.

Tumors from nine patients with CLL (n = 3), mantle cell lymphoma (n = 4), or acute lymphoblastic leukemia (n = 2) and 95 patients with TNBC (n = 60) or NSCLC (n = 35) were screened for ROR1 expression (Supplementary Fig. S1). Homogenous ROR1 expression was observed in all three CLL samples, and these patients met protocol eligibility criteria (Fig. 1A). ROR1 staining was predominantly membranous, both membranous and cytoplasmic, or cytoplasmic only in TNBC and NSCLC (Fig. 1B). Patients with only cytoplasmic ROR1 staining or with ROR1 expression on less than 20% of tumor cells were excluded. Eighteen patients in cohort B (TNBC, n = 10; NSCLC, n = 8) met protocol eligibility criteria and were enrolled (Supplementary Data S1). Patient characteristics and the number of prior therapies for metastatic disease are shown (Supplementary Tables S1 and S2; Table 1). CAR-T cells were administered at the DL and lymphodepleting chemotherapy regimen indicated (Fig. 1C; Table 1).

ROR1 CAR-T–cell manufacturing was successful for all patients, and cell dose escalation proceeded per protocol guidelines. CRS occurred in 16 patients, and immune effector cell–associated neurotoxicity syndrome occurred in two patients (Table 2). Grade 3 and higher adverse events attributed to CAR-T cells, lymphodepleting chemotherapy, underlying disease, or CRS are shown in Supplementary Table S3. Although ROR1 is expressed in parathyroid and pancreatic islets, sustained changes in serum calcium or glucose were not observed (Supplementary Figs. S2A and S2B). Of interest, 14 patients had reversible grade 3 (n = 12) or grade 4 (n = 2) pulmonary adverse events, including atelectasis (n = 1), dyspnea (n = 2), hypoxia (n = 7), pulmonary edema (n = 4), and pleural effusion (n = 1; Supplementary Table S3). Pulmonary adverse events were not associated with CAR‐T–cell expansion, disease type, or prior chest irradiation. One patient with NSCLC with a large right perihilar mass and parenchymal reticular opacification consistent with lymphangitic carcinomatosis on the pretreatment lung CT scan developed a DLT after receiving DL4 of ROR1 CAR-T cells (Fig. 2A). EGFRt⁺ CAR-T cells rapidly expanded in the blood of this patient, and progressive dyspnea developed over the first 6 days following the CAR‐T–cell infusion (Supplementary Fig. S2C and S2D). Bronchoalveolar lavage (BAL) obtained 8 days after CAR‐T–cell infusion identified EGFRt⁺ CAR-T and EGFRt⁻ bystander T cells that expressed PD1, TIM3, LAG3, TIGIT, and CD39 (Fig. 2B). A CT scan 6 days after CAR‐T–cell infusion showed a decrease in the perihilar mass but progressive lung inflammation or edema (Fig. 2A), and the patient died of respiratory failure 17 days after CAR‐T–cell infusion.

EGFRt⁺ CAR-T cells expanded in blood of all three patients with CLL, reaching 10% to 95% of total CD8⁺ and 30% to 70% of total CD4⁺ T cells 8 to 14 days after infusion (Fig. 2C). CAR‐T–cell expansion was highest in the patient with the highest frequency (60%) of CLL cells in the pretreatment bone marrow (BM). CAR-T cells remained detectable beyond 100 days after infusion in two of the three patients with CLL (Fig. 2D). In cohort B, CAR-T cells peaked in the blood at >50% of total CD8⁺ and 15% to 25% of total CD4⁺ T cells in four patients (high-expanders); 3% to 50% of total CD8⁺ and 1% to 25% of total CD4⁺ T cells in nine patients (medium-expanders); and <3% of total CD8⁺ and <1% of total CD4⁺ T cells in four patients (low-expanders; Fig. 2E). Vector copy number remained above the detection threshold at day 100 after infusion in one of two (50%) high-expanders, zero of five (0%) medium-expanders, and one of five (20%) low-expanders (Fig. 2F). IL-15 levels at day 0 were higher in patients with greater ROR1 CAR-T–cell expansion (Supplementary Fig. S3A). Comparison of lymphocyte counts in cohort B patients showed greater lymphodepletion in patients given Cy/Flu rather than Ox/Cy lymphodepletion (Supplementary Fig. S3B and S3C). Consistent with this observation, the frequency of CAR-T cells in the blood at peak expansion was significantly reduced in patients who received Ox/Cy compared with those who received Cy/Flu (Supplementary Fig. S3D–S3E).

Response assessment with BM biopsy and/or radiographic imaging was completed between days 28 and 35 after CAR-T–cell infusion. Two of three patients (66%) with CLL achieved a partial response after CAR‐T–cell infusion (Fig. 3A). Patient X645 had rapid CAR‐T–cell expansion and complete clearance of circulating tumor cells from blood, coinciding with increased serum lactate dehydrogenase and blood urea nitrogen consistent with tumor lysis (Fig. 3B and C), and elevated ferritin, C-reactive protein, and IL-6 (Fig. 3D). The response was transient because of the emergence of ROR1⁻ tumor cells. Patient X714 had a robust expansion of CD8⁺ CAR-T cells in blood and BM and a marked reduction in CLL cells in the BM from 9% prior to treatment to 0.02% at day 28 (Fig. 3E and F). The patient received a second CAR‐T–cell infusion without lymphodepletion, after which CD8⁺ CAR-T cells expanded in blood to 55% of total CD8⁺ T cells and were detected in the BM (Fig. 3G). CLL cells became undetectable in the BM 28 days after the second infusion (Fig. 3H), and the patient was in complete remission 7 months after the second infusion. CD45RO⁺CD39^HiTIM3^HiPD1^HiEGFRt⁺ CAR-T cells remained detectable in the BM 226 days after the second infusion (Fig. 3I and J). The patient relapsed with ROR1⁺ CLL 11 months after the second infusion and died 989 days after the second infusion. The third patient with CLL (X870) also had early-stage NSCLC treated by surgical resection 12 and 1 months prior to enrollment. Retrospective evaluation revealed that the lung tumor was ROR1 positive (Fig. 3K). Despite treatment at DL4, CAR‐T–cell expansion was moderate, and the patient had progression of CLL (Fig. 3A and L).

In cohort B, 16 of 18 patients (88%) had stable disease at day 28, and one patient died too early for disease response assessment (Fig. 4A). Patient X566 with TNBC had a partial response by RECIST after a second CAR‐T–cell infusion but progressed after 6 months (Fig. 4B). All remaining patients with stable disease at day 28, including one patient with initial regression of nodal and subcutaneous metastasis, progressed before 6 months (Fig. 4A and C). Unlike in patients with CLL in whom CAR-T cells were detected at tumor sites, a low level of WPRE copies was detected in biopsies from only two of the seven cohort B patients who had a posttreatment biopsy, and these two patients had high CAR‐T–cell expansion in blood (Fig. 4D). IHC of the tumor from one of these patients showed that tumor-infiltrating CD8⁺ T cells co-expressed TOX1 and PD1, markers associated with T-cell exhaustion (Fig. 4E).

Limited expansion or persistence of CAR-T cells could be due to an immune response to the ROR1 scFv. Assays of T cells from postinfusion blood obtained between days 28 and 50 after CAR‐T–cell infusion from six patients who received Ox/Cy and with low-to-medium early CAR‐T–cell expansion showed lytic activity against autologous CAR-transduced T cells but not nontransduced T cells, consistent with the development of T-cell responses to novel epitopes encoded in the CAR vector (Fig. 4F). Only one patient administered Cy/Flu developed a measurable immune response to CAR-T cells, supporting the use of fludarabine to prevent immune responses to foreign CAR sequences. CAR-T cells became undetectable by PCR in all seven patients who developed an anti-CAR immune response.

We examined the phenotype of CD8⁺ T cells in the PRE (n = 15/21) and EGFRt⁺ CD8⁺ CAR-T cells in the IP (n = 13/21) and at PEAK (n = 17/21) by flow cytometry and clustered cells by marker expression after adjustment for moderate batch effects. PRE CD8⁺ T cells were mainly in clusters 1 to 6, IP CAR-T cells were mainly in clusters 7 to 15, and PEAK CAR-T cells were mainly in clusters 16 and 17 (Supplementary Fig. S4A). All clusters in the IP had increased CD95 expression, clusters 7 and 8 of the IP were less-differentiated CD45RO^intGZMB⁻HLADR⁻ cells, and clusters 9 to 15 expressed a CD45RO^hiGZMB^hiHLADR^hi phenotype with various levels of CD69, CD25, CD39, and LAG3 activation markers. EGFRt⁺ CD8⁺ CAR-T cells in clusters 16 and 17 were discriminated mainly by CD57 expression and expressed higher levels of PD1 and LAG3 compared with IP clusters (Supplementary Fig. S4B). CD8⁺ T cells in the PRE samples of cohort B had an increased abundance of cells in cluster 1 that expressed a naïve TCF1^hiCD27^hiCCR7^hiCD45RA^hiCD45RO⁻CD25⁻ phenotype compared with those of cohort A (Supplementary Fig. S4C). Surprisingly, the higher abundance of more effector differentiated CD8⁺ T cells in the PRE samples of cohort A correlated with superior mean fold expansion during manufacturing (12.6-fold) compared with those of cohort B (5.1-fold; Supplementary Fig. S4D). The IP of cohort B showed increased abundance of less-differentiated CD8⁺ T cells (clusters 7 and 8) compared with the IP of cohort A (Supplementary Fig. S4E). However, no significant differences were detected within the IP of high-/medium-/low-expanders of cohort B, suggesting that factors other than the phenotype of the IP were responsible for driving expansion in vivo. No significant differences in cluster distribution in PEAK samples were observed in high-/medium-/low-expanders (Supplementary Fig. S4F).

This phase I study evaluated the safety and antitumor activity of CAR-T cells targeting ROR1 in patients with treatment-refractory CLL, TNBC, and NSCLC. This is the first study to examine CAR‐T–cell therapy targeting the same antigen expressed by a hematologic malignancy and by solid tumors, providing a unique opportunity to evaluate outcomes in these two settings.

The primary objective of our study was to determine the safety of ROR1 CAR-T cells, particularly because toxicity has occurred in mice treated with intensive lymphodepletion and CAR-T cells targeting murine ROR1 (44, 45). In normal human adult tissues, ROR1 is most highly expressed on parathyroid, and serial monitoring of serum calcium did not identify sustained changes after CAR‐T–cell infusion, despite significant in vivo expansion of CAR-T cells in 16 of the 21 patients. Reversible grade 3 (n = 12 patients) and grade 4 (n = 3 patients) pulmonary adverse events and a DLT in one patient with extensive pulmonary tumor involvement were observed. Neither prior thoracic irradiation nor the magnitude of CAR‐T–cell expansion correlated with the occurrence of lung adverse events. The patient with the DLT had a high tumor burden in the lung, and we detected high levels of CAR-T cells that expressed activation/inhibitory markers in a BAL obtained early after infusion, consistent with local antigen recognition. Reduction in a large lung tumor mass was observed in this patient, but the possibility that normal lung cells were also recognized by infiltrating CAR-T cells cannot be excluded. CAR affinity and structure, cell dose, and lymphodepletion intensity have been shown to alter the therapeutic window for other antigens (46); thus, the relative lack of normal tissue toxicity in this study does not ensure safety of ROR1 CAR-T cells designed with a different scFv, signaling domain(s), or administered with different lymphodepletion.

The CAR used in this trial was constructed from a rabbit scFv (32) and was more effective in preclinical models than a CAR constructed with a murine scFv (31). Rabbit antibodies are considered more similar to human antibodies than those from mice, but like any CAR derived from nonhuman sources, there is the potential to elicit immune responses. We cannot exclude the possibility that this scFv would be more immunogenic than others.

CAR-T cells expanded markedly in 16 of the 21 patients, including five patients in whom CD8⁺ CAR-T cells peaked at greater than 80% of total CD8⁺ T cells. Administration of a (1:1) formulated EGFRt⁺ CAR-T–cell product resulted in a greater expansion of CD8⁺ CAR-T cells over CD4⁺ CAR-T cells in most patients, similar to what was observed with CD19 CAR-T cells (47, 48). Five patients, all in the solid tumor cohort, had poor in vivo expansion with a peak CAR-T–cell frequency in blood of less than 3% of total lymphocytes and less than 10,000 vector copies per mg/DNA. Peak CAR‐T–cell expansion correlated with higher IL-15 levels and with the depth of lymphodepletion, as observed in studies with CD19 CAR-T cells (49–51). Ox/Cy improved CAR-T–cell infiltration and antitumor efficacy in a murine NSCLC model (33); however, CAR‐T–cell expansion was markedly lower in patients with solid tumor treated with Ox/Cy compared with those treated with Cy/Flu, suggesting that alternative dosing of Ox/Cy or inclusion of Flu may be necessary. Although the number of patients with CLL was small, CAR‐T–cell expansion was highest in the patient with the greatest tumor burden in the BM. All patients with solid tumor had extensive metastases, but the limitations of IHC on archival tissue for quantifying ROR1 expression made it difficult to ascertain whether differences in tumor antigen levels contributed to differences in CAR‐T–cell expansion in this cohort. The eligibility criteria only required the expression of ROR1 on 20% of tumor cells, and a potential strategy to increase the therapy response and ROR1 CAR‐T–cell expansion in the solid tumor cohort would be to increase the threshold of ROR1 tissue expression required for study enrollment.

Trafficking and/or T-cell exhaustion may pose greater barriers for CAR-T–cell therapy of solid tumors compared with hematologic malignancies. Seven patients with solid tumor had posttreatment tumor biopsies, and CAR-T cells were only detectable by qPCR in the two patients with the highest level of CAR-T cells in blood. The patient with extensive lung involvement and lymphangitic spread had high numbers of CAR-T cells in BAL and blood, suggesting that disseminated tumor cells in lymphatics may drive local and systemic CAR‐T–cell expansion. In CLL, CAR-T cells were present in BM in a high frequency, consistent with migration and local proliferation at a major tumor site. Consistent with CAR-T–cell access to tumor, efficacy was observed in two of the three patients with CLL but only 1 of 18 patients with TNBC or NSCLC. We previously showed that ROR1 CAR-T cells in patients with solid tumor upregulated the expression of inhibitory receptors (PD1, TIM3, LAG3, and TIGIT) and lost the ability to produce cytokines, consistent with an exhausted phenotype (33). Similar functional evaluation of ROR1 CAR-T cells after infusion was not performed in the patients with CLL, but the rapid tumor responses and elevated cytokine levels in two of the three patients indicate that CAR-T cells retained function, despite upregulating the expression of co-inhibitory receptors. Engineering T cells to resist exhaustion with novel genetic manipulations might improve the antitumor activity of ROR1 CAR-T cells in patients with a solid tumor (52–57). The transient therapeutic response in these two patients could be a result of ROR1 antigen loss or poor CAR‐T–cell persistence, as shown in the context of CD19 targeting.

In summary, ROR1 CAR-T cells were safely administered to patients with advanced CLL, TNBC, and NSCLC. Given the response rate in the small CLL cohort, ROR1 CAR-T cells may be considered for the treatment of relapsing CD19⁻ CLL or in the context of dual targeting. The treatment of TNBC and NSCLC using ROR1 CAR-T cells remains challenging, and novel strategies to improve infiltration and sustain T-cell function are required.

J.M. Specht reports grants from Juno Therapeutics during the conduct of the study as well as grants and personal fees from A2 Biotherapeutics; personal fees from Scripps Research Institution, GE Healthcare, Boehringer Ingelheim, Sensei Biotherapeutics, and Volastra, and grants from Lyell Immunopharma, Carisma Therapeutics, Celcuity, Genentech, Xencor, Seagen, Pfizer, and Merck outside the submitted work. In addition, J.M. Specht reports a patent for ROR1 CAR issued, licensed, and with royalties paid from Lyell Immunopharma. C.C.S. Yeung reports personal fees from TwinStrand outside the submitted work. S.M. Lee reports other support from Iovance, Lyell, Seagen, and Juno Therapeutics outside the submitted work. E.W. Newell reports personal fees and other support from ImmunoScape, Neogene, and Trojan Bio and personal fees from InduPro outside the submitted work. D.G. Maloney reports grants and personal fees from Bristol Myers Squibb/Juno Therapeutics during the conduct of the study as well as personal fees and other support from A2 Biotherapeutics and personal fees from Bristol Myers Squibb, Juno Therapeutics, and Lyell outside the submitted work. In addition, D.G. Maloney reports a patent for Fred Hutchinson Cancer Center pending and with royalties paid from Bristol Myers Squibb. S.R. Riddell reports grants from Juno Therapeutics, a Bristol Myers Squibb company, during the conduct of the study as well as grants from Bristol Myers Squibb; grants, personal fees, and other support from Lyell Immunopharma; other support from Ozette Technologies; personal fees and other support from Outpace Bio; and personal fees from Adaptive Biotechnologies and Nohla Therapeutics outside the submitted work. In addition, S.R. Riddell reports a patent for PCT/US2018/049812 pending to Lyell Immunopharma. No disclosures were reported by the other authors.

C.A. Jaeger-Ruckstuhl: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. J.M. Specht: Conceptualization, data curation, validation, methodology, writing–review and editing. J.M. Voutsinas: Resources, data curation. H.R. MacMillan: Resources, data curation, software, validation, visualization. Q. Wu: Resources, data curation. V. Muhunthan: Investigation. C. Berger: Investigation. S. Pullarkat: Resources, data curation. J.H. Wright: Resources, data curation. C.C.S. Yeung: Resources, validation. T.S. Hyun: Resources. B. Seaton: Resources, data curation. L.D. Aicher: Resources, data curation. X. Song: Resources. R.H. Pierce: Resources. Y. Lo: Resources, data curation. G.O. Cole: Resources, data curation. S.M. Lee: Resources. E.W. Newell: Resources, data curation, software, supervision, validation, visualization. D.G. Maloney: Conceptualization, supervision, funding acquisition, validation, writing–review and editing. S.R. Riddell: Conceptualization, supervision, funding acquisition, methodology, writing–original draft, writing–review and editing.

The study was supported by Juno Therapeutics, a Bristol Myers Squibb company; NIH R01 CA114536 (S.R. Riddell); NIH P50 CA138293; and the American Cancer Society (S.R. Riddell).