Study design and execution, patient selection and treatment
The study design comprises a dose escalation part and a dose expansion part with indication-specific cohorts including GCT, EOC and other indications to start when the recommended phase 2 dose (RP2D) obtained with the dedicated phase 2 manufacturing process is defined.
The primary endpoints for the dose escalation part are to characterize the safety and tolerability of CLDN6 CAR-T cellsCARVac and to identify the maximum tolerated dose (MTD)/RP2D. Evaluation of the antitumor activity of CLDN6 CAR-T cellsCARVac per Response Evaluation Criteria In Solid Tumors (RECIST) 1.1 and characterization of soluble immune factors (cytokines) induced by treatment are the secondary endpoints. An MTD/RP2D was not established, as dose escalation is being repeated with CAR-T cells manufactured with an automated manufacturing process.
The dose expansion part of the study enrolled adults diagnosed with advanced metastatic solid tumors of any type lacking further systemic treatment options, Eastern Cooperative Oncology Group (ECOG) 01 and measurable disease per RECIST 1.1. Patients with new or growing brain or spinal metastases during screening were excluded. Patients for the study were recruited from a CLDN6 prescreening. Additional eligibility criteria are noted in the Methods.
Of 180 patients prescreened for CLDN6 expression with a semiquantitative immunohistochemistry assay between September 2020 and November 2021, 54 (30%) met the inclusion criteria for tumor CLDN6 positivity defined as 50% of tumor cells displaying an intermediate (2+) or strong (3+) membrane staining intensity (Fig. 1a,b and Supplementary Fig. 1). The fraction of patients complying with the CLDN6-positivity threshold was highest for GCT (90%) and EOC (29%) (Fig. 1c). No significant differences in staining intensity were observed according to whether the specimens were obtained recently or were older paraffin-embedded tissue specimens; this was consistent for both primary and metastatic lesions (Supplementary Fig. 2).
a, Phase 1 bifurcated 3+3 dose escalation design. Screened patients were enrolled into either CLDN6 CAR-T monotherapy or a combination with CARVac, receiving a single dose of either 1107 (DL1) or 1108 (DL2) CLDN6 CAR-T cells. CARVac was administered with a starting dose at 25g, followed by 50g doses if tolerated. b, As of 5 November 2021 (enrollment of last patient dosed), 180 patients were prescreened and 54 classified as CLDN6-positive (50% of tumor cells 2+ CLDN6 membrane staining). Eighteen patients dropped out before the screening visit due to death (n=9), worsening of condition (n=7), loss to follow-up (n=1) or refusal of participation (n=1). Twenty-nine patients underwent full screening for trial eligibility, of which four patients did not meet eligibility criteria. Seven patients remained listed to be screened for trial eligibility, but did not consent for full trial eligibility screening, for example, because of decision to undergo other treatment approaches. CAR-T cell products were manufactured for all 25 enrolled patients. Of those, 22 were treated and included in the safety set, while one patient, who received CAR-T cells at Following successful prescreening for fulfillment of CLDN6 expression, twenty-nine patients entered screening for the other enrollment criteria of this trial. We manufactured autologous CLDN6 CAR-T cells for 25 patients who met all eligibility criteria, of which 22 were patients treated with CLDN6 CAR-T cells manufactured from autologous leukapheresis material collected between 15 December 2020 and 14 February 2022, with the day of leukapheresis defined as the day of enrollment to this trial (Fig. 1b). The drug products contained both CD4+ and CD8+ T cells, with the CD4+ subset dominating (Supplementary Fig. 3a). Overall, the proportion of nave (TN) and effector memory phenotype (TEM) was similar within the CD4+ T cell population. (Supplementary Fig. 3b). CD8+ T cells had a predominantly TN-like phenotype, followed by an effector memory re-expressing CD45RA (TEMRA) phenotype (Supplementary Fig. 3c). Dose escalation followed a 3+3 approach with patients receiving CLDN6 CAR-T cells at DL1 (1107) and DL2 (1108)CARVac. Of the 22 treated patients, 13 had GCTs (all testicular cancer with non-seminoma or mixed-type histology), four had EOCs (all serous carcinoma) and one patient each had endometrial carcinoma, serous carcinoma of the fallopian tube, desmoplastic small round cell tumor (DSRCT), gastric adenocarcinoma and cancer of unknown primary (CUP) (Table 1). GCT and EOC patients had the strongest CLDN6 expression, with an average of >80% tumor cells with 2+/3+ (intermediate/strong) staining intensity (Fig. 1d and Supplementary Fig. 1). All patients (median age of 46 years) were r/r after standard of care treatment and were heavily pretreated with a median of four previous lines of treatment, predominantly platin-based chemotherapy and three had received checkpoint inhibitor therapy. All GCT patients had received high-dose chemotherapy (HDCT) plus autologous stem cell support. The four EOC patients were platinum-refractory. All patients had measurable disease, and eight patients required bridging chemotherapy between leukapheresis and CLDN6 CAR-T cell transfer. Almost half of the patients had lung metastases (including seven of the 13 GCT patients) and about 30% had either liver involvement or peritoneal carcinosis (including three of the four EOC patients). Seven of the 12 patients with two pretreatment computed tomography (CT) scans showed rapidly progressing disease from baseline visit until infusion of the CAR-T cells (5.9 weeks on average; Table 1). Those who did not progress had all been treated with bridging chemotherapy, further underlying the advanced disease status of patients recruited to this trial. Of the 22 patients (Fig. 1b), 20 received LD (500mgm2 cyclophosphamide plus 30mgm2 fludarabine for three days) before infusion of CLDN6 CAR-T cells at DL1 (n=7, including one patient with non-conformant product with lower yield) or DL2 (n=13, including one patient with 50% reduced LD). Four patients at DL1 and seven patients at DL2 were treated in combination with CARVac. In two patients who were at risk of prolonged cytopenia, we explored omitting LD (Extended Data Tables 1 and 2). Patients treated at DL1 or with dose-reduced LD received a median of 3.5 doses of CARVac (range 16) starting at day 4 post-adoptive cell transfer (ACT). For patients treated at DL2 with full dose LD, CARVac administration was delayed to reduce the risk of accelerated and high-grade cytokine release syndrome (CRS), with CARVac starting 23 days or more post-ACT with a median of five (range 29) doses. The two patients at DL2 without prior LD were treated with two and four doses of CARVac, respectively. Two patients at DL2 crossed over and received CARVac (3 doses) starting 65 and 79 days post-ACT, respectively. Five patients were redosed at their original DL 190288 days after their first ACT in combination with CARVac starting on day 4 (DL1) or day 1516 (DL2) post redosing, respectively. One patient at DL1 was redosed without prior LD (Extended Data Table 1). Overall, assessment of CLDN6-positivity followed by manufacturing and administration of CLDN6 CAR-T cellsCARVac was feasible in this population of heavily pretreated patients including individuals with rapidly progressing disease. All 22 patients treated within the study were included in the safety analysis. The data cutoff date was 6 October 2022, with a median follow-up of five months (range: 29416 days, Extended Data Table 1). Nineteen of 22 patients had treatment-emergent adverse events (TEAEs) greater than or equal to grade (G) 3 (Extended Data Table 2). Besides pyrexia, the most frequent TEAEs observed in >20% of treated patients were hematologic toxicities related to LD or transaminase/lipase elevations without clinical correlates (Table 2), observed more often in patients experiencing CRS. TEAEs G3 attributed to the CAR-T cell treatment that occurred in >10% patients were neutropenia (23% G4) and leukopenia (5% G3, 14% G4) (Extended Data Table 3). CRS was seen in ten patients (46%), eight of whom were treated at DL2 (Extended Data Table 2). CRS typically occurred within 410 days post-ACT, was correlated with high IL-6 levels and peak CAR-T cell concentration post-ACT, was predominantly of G12, and was manageable by supportive care with antipyretics and the IL-6 receptor blocking antibody tocilizumab (Extended Data Table 2 and Extended Data Fig. 1). As CRS events were more frequently observed in the DL2 monotherapy cohort (Extended Data Table 2), the safety committee decided to delay CARVac dosing to day 24 in the DL2 combination cohort to avoid the risk of CAR-T cell-amplification related high-grade CRS. One patient treated at DL1 plus CARVac had symptoms (headache and decreased level of alertness) classified as G1 immune effector cell-associated neurotoxicity syndrome (ICANS) that occurred in conjunction with G2 CRS and resolved spontaneously within 24h (Extended Data Table 2). Dose-limiting toxicities (DLTs) emerged in two patients treated at DL2, one with monotherapy and one in combination with CARVac (Extended Data Table 2). One event was G4 hemophagocytic lymphohistiocytosis (HLH) observed in an EOC patient, who had the highest CAR-T cell peak expansion (5.8109) observed in this trial. On day 5 post-ACT, the patient developed G2 CRS (treated with tocilizumab) and on day 23 experienced HLH defined by highly elevated ferritin levels and an aplastic bone marrow, which was successfully treated with high-dose steroids. CARVac treatment was postponed to day 51 until all CRS- and HLH-related AEs had resolved without sequelae. The other DLT was prolonged G4 pancytopenia observed in a heavily pretreated GCT patient reported 21 days post-ACT. The patient was nonresponsive to granulocyte colony-stimulating factor, received an autologous peripheral blood stem cell support on day 32, upon which the patients bone marrow function recovered within 14 days. The patient was redosed with CAR-T cells 41 weeks after the first CAR-T cell administration. This DLT experience led to the decision to consider GCT patients who had recently undergone HDCT as at risk for treatment-related persistent cytopenia. Consequently, we introduced protocol amendments to make the availability of autologous peripheral blood stem cells a prerequisite for treatment of patients with HDCT within the last 12 months or with impaired bone marrow function with the full dose of LD. Alternatively, we allowed dose-reduced LD. No further DLTs were observed after enrollment of three additional patients into each of these two DL2 cohorts. The Safety Review Committee (SRC) determined that the MTD (primary endpoint) was not reached. The recommended phase 2 dose of CLDN6 CAR-T cells was not further pursued as enrollment to the planned DL3 cohorts was canceled per the study protocol amendment to facilitate a repeat of the CAR-T dose escalation component with CAR-T cells manufactured with an automated process, including potential testing of higher dose levels. All eight deaths in the study were classified as disease progression and not attributed to CAR-T cell toxicity. The safety profile of CARVac was in line with previous reports related to our RNA-LPX cancer vaccine platform16,17 and no unexpected toxicity was seen in combination with CAR-T cells (Extended Data Table 3). CARVac-related TEAEs were primarily of G12 flu-like symptoms occurring around 4h after administration. TEAEs of G3 that occurred with a frequency of 10% and were attributed to CARVac treatment were pyrexia (18% G3) and neutropenia (12% G4), which was also documented as related to LD. While dose reduction was allowed per protocol in the case of CARVac-related AEs, CARVac was generally well tolerated and no dose reductions occurred. After initial hospitalization, patients received CARVac in an outpatient setting and toxicities were managed with paracetamol. Transient release of IFN and IFN inducible protein 10, peaking 36h post intravenous CARVac administration, was observed in serum cytokine measurements (Extended Data Fig. 2), as previously reported in patients treated with RNA-LPX mRNA cancer vaccines15. In summary, the safety profile of both dose levels of CLDN6 CAR-T cells as monotherapy or in combination with CARVac was manageable (Table 2 and Extended Data Table 3) and largely in line with previous reports on approved CAR-T cell products18. Twenty-one of the 22 patients were treated per protocol and qualified for efficacy analysis for the secondary endpoints ORR, disease control rate (DCR) and duration of response (DOR). Partial responses (PRs) in six patients and one complete response (CR) resulted in an unconfirmed ORR of 33% (Fig. 2a). Four patients with a best overall response (BOR) of PR remained in PR in subsequent scan(s), while two patients experienced progressive disease (PD) at the next assessment. We observed deepening of PRs over time, with further reductions in the sum of target lesions observed over repeat assessments. The median DOR for all seven responders was 2.8months (range 1.1 10.5months) (Fig. 2b), with the CR ongoing at data cutoff for 10.5months (Extended Data Fig. 3). Seven patients had stable disease (SD) as the best response, five with quantifiable target lesion shrinkage, resulting in a DCR of 67%. One SD was ongoing at data cutoff 8.7months post-ACT. Five of the seven patients with PD had either received CLDN6 CAR-T cells at DL1 or had not received LD before DL2. The other two patients with PD were treated at DL2 with prior LD; however, they had rapid disease progression at accrual (61% and 86% increase in target sum between screening and ACT; Table 1). Five patients (three PRs, one SD, one PD as BOR to the first dosing) were redosed with CAR-T cells due to PD (Fig. 2b) resulting in one additional PR (Extended Data Fig. 4). a, Waterfall plot showing best percent change from baseline (screening, ~6 weeks before ACT) in sum of target lesion diameters. Efficacy evaluable population (n=21) contains one patient that did not reach the first CT scan and was classified as PD (confirmed by X-ray). The table below indicates applied treatment, outcome and disease status at ACT (available for 12 patients). Redosing (n=5) occurred after assessment of best response in all patients. b, Swimmer plots for all efficacy evaluable patients (n=21). The patient with dose-reduced LD is marked with an asterisk. Diamonds represent outcome at tumor assessment (CR, PR, SD). Triangles indicate redosing with CAR-T cells at the same DL (orange) or crossover from monotherapy to combination therapy (yellow). Redosing was always performed as a combination treatment. All responses were assessed according to RECIST 1.1. The patient with CR had a residual radiographic abnormality interpreted as scar tissue, as there was no abnormal radiotracer uptake according to Positron Emission Tomography/Computed Tomography (PET-CT) 12 weeks post-ACT (Extended Data Fig. 5). Cancers of other origin other than GCT and EOC were each a single case of DSRCT, GC, serous carcinoma of the fallopian tube, endometrial carcinoma and CUP. NE, non-evaluable; NR, not reached; PFS, progression-free survival. All objective responses occurred in patients with either EOC (two of four patients with PR) or GCT (four PRs plus one CR in 13 patients; Fig. 3a,c). The five patients with other tumor entities all had SD as BOR (including the patient treated a, Spider plots indicating response durability in all patients (left, n=21), GCT patients (middle, n=13) and non-GCT patients (right, n=8). Non-GCT patients include four EOC patients and one each of DSRCT, GC, serous carcinoma of the fallopian tube and endometrial carcinoma. Pretreatment CT scans are not included, as scans performed at screening (~6 weeks before ACT) served as baseline. b, Details on ORR and DCR of all patients as well as subgroups of GCT and non-GCT patients according to treatment (dose and LD). c, PFS analysis for n=7 GCT patients treated at DL2 after LD with 95% confidence interval based on a loglog transformation of the survival function (dotted lines). ORR is unconfirmed. The ORR in the subgroup of GCT patients was 38% and depended on administered CLDN6 CAR-T DL and LD (Fig. 3a,b): 25% at DL1 after LD (one PR in four patients), 57% at DL2 after LD (four PRs in seven patients) and 0% at DL2 without LD (two patients). GCT patients were those with the longest DORs, leading to a PFS (exploratory endpoint) of 42% at six months for those treated with CLDN6 CAR-T cells at DL2 after LD (Fig. 3c). As the DL2 cohort had not reached median overall survival at the time of the data cutoff, and given the heterogeneity of the patients, indications and treatment schedules, we do not report overall survival (a further exploratory endpoint). Of the eight non-responding GCT patients, five were treated at DL2. Two achieved SD as BOR, two were treated without prior LD and experienced PD, as did the fifth patient, who entered the study with a rapidly progressing tumor (86% increase in target sum from screening to ACT) having only 50% 2+/3+ CLDN6-positive tumor cells. Notably, of the two non-responding EOC patients, one was treated at DL2 and showed rapidly progressing disease (61% increase in target sum from screening to ACT) stabilizing after infusion (19% reduction from ACT to first assessment). Previous lines of treatment for patients are included in Supplementary Table 1. Tumor responses were primarily observed at DL2. However, the late timing of CARVac dosing at this DL, combined with the diverse, small cohorts, prevented analysis of how CARVac influences the antitumor activity of CLDN6 CAR-T cells. We observed encouraging signs of efficacy and disease control in a heterogenous cohort of hard-to-treat solid tumor patients, indicating GCT and EOC patients in particular as future target populations. In parallel, we developed an automated manufacturing process to scale up CAR-T production and increase the robustness of the manufacturing process. Recruitment of patients to a planned DL3 with CAR-T cells manufactured with the manual manufacturing process was therefore canceled by protocol amendment, and we have initiated a repeat of dose escalation with CAR-T cells produced with the automated manufacturing process, introducing further dose levels for both CAR-T cells and CARVac. Identification of predictive biomarkers was an exploratory endpoint. All responding patients had tumors with >80% of tumor cells expressing 2+/3+ CLDN6 at prescreening (Fig. 4a), suggesting that CLDN6 expression level may be predictive of outcome. Four patients had PD despite high CLND6 expression, three of which entered the trial with progressing disease (61%, 99% and 105% increase in target sum from start of screening to ACT). The fourth patient did not receive LD and experienced poor CAR-T cell engraftment. A positive correlation between CLDN6 CAR-T expansion and clinical response was detected for both peak expansion (CAR-T cell Cmax) and area under curve (AUC) from ACT to first staging with CT (Fig. 4b). a, Correlation of clinical outcome and CLDN6 expression of the corresponding tumor according to indication and treatment (dose and LD). b, Correlation analysis of CAR-T cell peak concentration (Cmax) (left) and AUC up to day 42 post-ACT (first tumor assessment) (right) with outcome. Cancers of other origin other than GCT and EOC were single cases of DSRCT, GC, serous carcinoma of the fallopian tube and endometrial carcinoma. Box plots show median and upper and lower quartiles, with whiskers indicating 1.5 the interquartile range. Individual data points are overlaid. Characterization of the pharmacokinetics of CLDN6 CAR-T cells was an exploratory endpoint. For lymphodepleted patients at DL1 and DL2, CAR-T cell peak expansion (Cmax) detected in peripheral blood was reached on average within 18 days (range 1724) and within 15.6 days (range 824), respectively, with higher peak engraftment seen at DL2 (Extended Data Fig. 5). The two patients treated at DL2 with CARVac but without LD displayed poor engraftment, and this cohort was therefore closed (Extended Data Fig. 6). Two patients treated at DL1 (29%) and six patients treated at DL2 (46%) showed CAR-T cell persistence for 100days post-ACT, with CAR-T cells in one patient treated once at DL2 detectable for >1year (Extended Data Fig. 3). Four patients (two at DL1, two at DL2) with decrease of CAR-T cells over time were preconditioned again and redosed with CAR-T cells in combination with CARVac 190288 days after their initial ACT (exemplified by a patient at DL1+CARVac in Extended Data Fig. 6). Robust engraftment with almost phenocopied kinetics of their first ACT was achieved in three patients; all three had detectable CAR-T cells at their last follow-up. One patient preconditioned with LD before their first ACT was redosed at DL1 without LD and reached a lower Cmax compared to their first ACT. CARVac was administered to 11 patients who had received CLDN6 CAR-T cells after LD and two patients who received CAR-T cells without prior LD (Extended Data Fig. 6). Within this group, four patients were given CARVac following CAR-T cells at DL1. A trend for greater expansion of CAR-T cells was observed at DL1 in patients receiving CARVac compared with patients receiving CLDN6 CAR-T cells alone (Extended Data Fig. 5). Notably, the patient who received CAR-T cells at Given the limited sample size, the heterogeneity of patients and unequal initial engraftment within the subgroups at DL2, the impact of CARVac on persistence of CAR-T cells cannot be statistically evaluated with the data available at this initial cutoff. However, we noted a transient increase in CAR-T cells immediately following CARVac administration in some patients (Extended Data Fig. 4c). In two patients treated with CAR-T cells at DL2 who commenced CARVac at a later stage, CARVac appeared to halt an ongoing decline in CAR-T cell frequency (Extended Data Fig. 6, bottom left). A further four patients in monotherapy cohorts crossed over to combination therapy, two after redosing with CAR-T cells at DL1 and two (indicated with a circle in Extended Data Fig. 6) at DL2 that experienced PD before crossover. No conclusions can be reached from this heterologous group that had differing CARVac vaccination schemes. Further analysis with a larger patient cohort treated with CARVac at an earlier time point is planned to provide a more conclusive understanding of the impact of CARVac on CAR-T cell dynamics. The rest is here:
