Autogene cevumeran with or without atezolizumab in advanced solid tumors: a phase 1 trial

Abstract

Effective targeting of somatic cancer mutations to enhance the efficacy of cancer immunotherapy requires an individualized approach. Autogene cevumeran is a uridine messenger RNA lipoplex-based individualized neoantigen-specific immunotherapy designed from tumor-specific somatic mutation data obtained from tumor tissue of each individual patient to stimulate T cell responses against up to 20 neoantigens. This ongoing phase 1 study evaluated autogene cevumeran as monotherapy (n = 30) and in combination with atezolizumab (n = 183) in pretreated patients with advanced solid tumors. The primary objective was safety and tolerability; exploratory objectives included evaluation of pharmacokinetics, pharmacodynamics, preliminary antitumor activity and immunogenicity. Non-prespecified interim analysis showed that autogene cevumeran was well tolerated and elicited poly-epitopic neoantigen-specific responses, encompassing CD4⁺ and/or CD8⁺ T cells, in 71% of patients, most of them undetectable at baseline. Responses were detectable up to 23 months after treatment initiation. CD8⁺ T cells specific for several neoantigens constituted a median of 7.3% of circulating CD8⁺ T cells, reaching up to 23% in some patients. Autogene cevumeran-induced T cells were found within tumor lesions constituting up to 7.2% of tumor-infiltrating T cells. Clinical activity was observed, including one objective response in monotherapy dose escalation and in two patients with disease characteristics unfavorable for response to immunotherapy treated in combination with atezolizumab. These findings support the continued development of autogene cevumeran in earlier treatment lines. ClinicalTrials.gov registration: NCT03289962 .

Fig. 1. Phase 1 study design and patient disposition.

a, Patients with locally advanced, recurrent or metastatic incurable tumors with at least five neoantigens and with limited treatment options were eligible. Autogene cevumeran: up to 20 tumor neoantigens identified by sequencing blood and tumor tissue were cloned into two RNA molecules (up to 10 neoantigens each) with backbones comprising a 5′-cap analog, 5′- and 3′-untranslated region (UTR) and poly(A) tail optimized for stability and translational efficiency. Neoantigens are flanked by an N-terminal SEC and an HLA class I trafficking domain (MITD, transmembrane and cytoplasmic domain of HLA class I) to ensure optimal antigen presentation and immunogenicity. For phase 1a, patients were treated with eight doses of autogene cevumeran administered intravenously over the first 9 weeks (induction), followed by periodic booster doses (maintenance) until disease progression. In phase 1b, autogene cevumeran was combined with 1,200 mg atezolizumab administered intravenously on day 1 of the combination regimen and given every 3 weeks until disease progression. b, Patients in phase 1a received autogene cevumeran in dose escalation cohorts ranging from 25 μg to 100 μg. In phase 1b, autogene cevumeran (25 μg to 50 μg as indicated) was combined with 1,200 mg atezolizumab in a dose escalation part, indication-specific expansion cohorts including a serial biopsy and a biomarker cohort. In the biomarker cohort, 16 of 35 patients received autogene cevumeran as monotherapy for the induction course; atezolizumab was added for the maintenance course. Patients in both cohorts underwent biopsy, leukapheresis and blood draws before and during treatment for immune monitoring. c, Participant flow diagram. The number of participants screened, enrolled, treated, discontinued and analyzed is shown for phase 1a and 1b. C1, C7, C13 and C21, cycle 1, 7, 13 and 21; D1, D8 and D15, day 1, 8 and 15; q1w, q2w, q3w, q10w and q24w, once every 1, 2, 3, 10 and 24 week(s).

Fig. 2. Safety and tolerability data of patients with advanced disease who received autogene cevumeran with or without atezolizumab.

AEs occurring in ≥5% of patients are presented as all AEs and AEs related to treatment by grade. AEs related to treatment AEs were graded in line with the NCI CTCAE, v.5.0. Alk-phos, alkaline phosphatase; AST, aspartate aminotransferase; ALT, alanine aminotransferase.

Fig. 3. Autogene cevumeran induced de novo CD4⁺ and CD8⁺ T cell responses against multiple neoantigens in the blood and tumor of most patients.

a,b, Ex vivo IFNγ ELISpot data of 90 patients and 1,404 neoantigens. a, Proportion of patients with autogene cevumeran-induced T cell responses. b, Magnitude of autogene cevumeran-induced neoantigen-specific T cells in patients with available bulk PBMC ELISpot data on single neoantigens (left) and example patient (right). Numbers above the bars: numbers of neoantigens inducing a T cell response among all encoded neoantigens. Negative control: PBMCs with medium; positive control: anti-CD3 antibody. c, Autogene cevumeran-induced T cell responses per indication and CPI status. ‘Other’ includes indications with at most five patients. The numbers on the right show the immune response rate per indication. d,e, IFNγ ELISpot analysis after IVS of enriched CD4⁺ and CD8⁺ T cells with autogene cevumeran neoantigens in 17 patients and 234 evaluable neoantigens. d, Autogene cevumeran-induced neoantigen-specific T cell responses detected after IVS and ex vivo IFNγ ELISpot (n = 15). The numbers above the bars are the total numbers of immunogenic neoantigens detected by either assay. e, Phenotype of T cell responses at the patient level (bar chart) and the population level (pie chart). f, Kinetics of neoantigen-specific CD8⁺ T cells shown using pHLA multimer staining of PBMCs from 22 patients. The dotted lines indicate autogene cevumeran administration. g, Left, de novo induced CD8⁺ T cell response of patient 11 against an autogene cevumeran-encoded INDEL, followed for 8 months and, right, memory phenotyping with PD-1 expression of the pHLA multimer-stained T cells. h, Frequency of neoantigen-specific TCRs in circulation. i, Cumulative frequency of neoantigen-specific TCRs in CD8⁺ cells in post-induction blood and on-treatment tumor tissues collected between 35 and 56 days after treatment start (147 days for patient 38; n = 10 biomarker cohort patients). The CDR3 beta sequences were used for tracking in bulk profiling data. The colors represent distinct neoantigens. Numbers above the bars: total numbers of neoantigen-specific TCRs; numbers below the bars: numbers of corresponding neoantigens. D, de novo; n.d., not done; P, preexisting; TRB, T cell receptor β; FITC, fluorescein isothiocyanate.

Fig. 4. Antitumor activity of autogene cevumeran as monotherapy or in combination with atezolizumab in patients with advanced cancers in dose escalation cohorts.

a,b, Clinical activity in patients in dose escalation cohorts with evaluable baseline and postbaseline measurements, assessed as the effect of autogene cevumeran as monotherapy (a) and in combination with atezolizumab (b) on target lesions per RECIST v.1.1 (n = 27 for phase 1a and n = 29 for phase 1b). ELISpot ‘+’ indicates that the patient had a neoantigen that indicated a positive response. HNC, head and neck cancer; EGJ, esophagogastric junction; MCC, Merkel cell carcinoma; N, no; N/A, not available; N/D, not done; SD, stable disease; STS, soft tissue sarcoma; Y, yes.

Fig. 5. Patients with objective responses showed poly-epitopic autogene cevumeran-induced T cell responses.

a, Neoantigen-specific TCRs isolated from longitudinal blood samples in patient 18. TRB CDR3 sequences of neoantigen-specific TCRs were tracked in bulk CD8⁺ (top) and CD4⁺ (bottom) TRB profiling data. The bar color represents different specificities. The vertical dashed lines represent vaccination time points. Cumulative frequency and number (in brackets) of neoantigen-specific TCRs are shown for each time point above the respective bar, above which colored dots represent evaluation of the clinical response at that time point. b–d, Lesions, indicated by red arrows, are shown before and after study treatment. Lung CT scans of patient 21 with CRC (top), with the magnitude of T cell responses to individual neoantigens detected by ex vivo IFNγ ELISpot (bottom left) and time-course analysis of a CD8⁺ response targeted against C10orf54 (T37M) using pHLA multimer staining assays (bottom right) (b). Lung CT scans from patient 13 with TNBC (top row) and longitudinal CD8⁺ response kinetics showing CD8⁺ T cells specific against GALNT6 (E579K) detected via pHLA multimer assay (middle row). Time-course analysis of neoantigen-specific CD8⁺ TCRs isolated from blood samples following the induction phase (as described in a, bottom row) (c). Lung CT scans from patient 38 with RCC (d). TCRseq, T cell receptor sequencing.

Extended Data Fig. 1. Systemic reactions during the induction course of vaccination in phase 1a (autogene cevumeran monotherapy) by dose.

Red text represents concomitant medications given before or after autogene cevumeran administration. Purple represents concomitant medications given for the treatment of adverse events. Black text represents concomitant medications given on the same day as autogene cevumeran administration but not indicated as pre or post dose administration or for treatment of adverse events. IRR and CRS events were classed according to the Common Terminology Criteria for Adverse Events, version 5.0. C, corticosteroids; CRS, cytokine release syndrome; IRR, infusion-related reaction; M, meperidine; down arrow, dose reduction.

Extended Data Fig. 2. Changes in immune-related cytokine levels prior to and during treatment.

a, Cytokine profiles of individual patients during the induction phase. b, Peak plasma cytokine levels (4–6 h after autogene cevumeran injection) for patients treated with autogene cevumeran monotherapy in phase 1a or in combination with atezolizumab in phase 1b. Boxes show the 25th to 75th quantiles with lines representing medians from timepoints 4–6 h after infusion from multiple vaccinations; whiskers show minimum to maximum values; each dot represents one patient value per dose level. Sample sizes for each cytokine analysis were: IFNα n = 28 (Ph1a), n = 137 (Ph1b); IFNγ n = 28 (Ph1a), n = 147 (Ph1b); IL-6 n = 28 (Ph1a), n = 144 (Ph1b); TNFα n = 29 (Ph1a), n = 144 (Ph1b); IL-1β n = 28 (Ph1a), n = 141 (Ph1b); IL-12p70 n = 23 (Ph1a), n = 63 (Ph1b).

Extended Data Fig. 3. Autogene cevumeran-induced T-cell responses in individual patients.

a, Concomitant autogene cevumeran-amplified CD4⁺ and de novo CD8⁺ T-cell response against LIPC (M103K) and autogene cevumeran-induced CD4⁺ or CD8⁺ responses against two other neoantigens determined by post-IVS IFNγ-ELISpot in a combination-treated patient with SCC. b, Comparison of IFNγ ELISpot detected T-cell responses against 263 autogene cevumeran-encoded neoantigens administered across 15 patients for whom both ex vivo and post-IVS ELISpot was performed. 205 neoantigens were evaluable for post-IVS and 229 for ex vivo ELISpot c, Kinetics, memory phenotype and PD-1 expression of selected autogene cevumeran-induced neoantigen-specific CD8⁺ T-cell responses from three patients. PBMCs stained with neoantigen-specific pHLA multimers (first and fourth rows). pHLA multimer+ cells stained for CD45RO and CCR7 (second row) or CD279 (PD-1, third row). d, Neoantigen-specific CD8⁺ TCRs identified from blood post-induction for patient 34. Specificity of selected TCRs was tested in vitro against patient-specific neoantigen/HLA combinations. Freq e, Neoantigen-specific TCRs identified in multiple on-treatment tumor biopsies. CDR3 beta sequences were used as molecular identifiers to track the neoantigen-specific TCRs in bulk TCRβ (TRB) profiling data from multiple on-treatment tumor biopsies of five patients from the biomarker cohort. The cumulative frequency of neoantigen-specific TCRs in all biopsies (‘merged data’) and for each biopsy is shown for each patient. Each color represents a different neoantigen target. Numbers above the bar charts show the number of neoantigen-specific TCRs tracked in each sample for each patient. D, de novo; IVS, in vitro stimulation; P, pre-existing; PBMC, peripheral blood mononuclear cells; TCR, T-cell receptor.

Extended Data Fig. 4. Deep profiling of neoantigen-specific CD8 T cells from longitudinal time points using mass cytometry (CyTOF).

a, Number of unique neoepitope specificities detected in peripheral blood of 10 patients tested in mass cytometry analysis. 3 patients were from Ph1A (autogene cevumeran only) and 7 patients were from Ph1B (autogene cevumeran and atezo combination). b, Principal Component Analysis (PCA) of neoantigen-specific CD8 T cell populations identified across all 10 patients. PCA is based on phenotypic profiling (percent of neoantigen-specific CD8 T cells positive for all phenotypic markers assessed). Data points are colored based on time of collection (left plot) or segregated into early (red) or late (blue) time points (right plot). c, Frequency of expression for each individual markers on the neoantigen specific CD8 T cells compared between all early (red) pooled or late (blue) pooled time points (as described in b) (n = 27 for induction and n = 38 for maintenance data points). Data are presented as box whisker plots depicting the median, 25% (Q1) and 75% (Q3) quantiles, with the mean added in diamond shape. Minimum and maximum points are defined as Q1-1.5*IQR and Q3 + 1.5*IQR, respectively, where IQR is the inter-quantile range between Q1 and Q3. Any data point falling above or below the maximum or minimum points are marked in red circles. P values were calculated using two-sided Wilcoxon signed rank-sum test and are marked as: * P < 0.05, **P < 0.01. (Individual p values are CD27: 0.009; CD28: 0.009; CD38:0.009; CD39:0.004; CD45RO: 0.04; CD57:0.009; EOMES:0.0007; HLA-DR:0.004; ICOS:0.016; Ki67:0.01; TIGIT:0.03). P values were adjusted for multiple testing using the Benjamini-Hochberg method to control the false discovery rate.

Extended Data Fig. 5. Anti-tumor activity of autogene cevumeran in combination with atezolizumab in patients with advanced cancers in expansion cohorts.

Clinical activity observed in the expansion cohorts, assessed as the effect of autogene cevumeran and atezolizumab combination therapy on target lesions per RECIST 1.1. Numbers in red indicate the number of responses observed over the number of patients in each cohort. SLD, sum of longest diameters (of baseline tumors); CPI, checkpoint inhibitor. UC, Urothelial cancer; RCC, renal cell cancer; NSCLC, non–small cell lung cancer; RECIST v1.1, Response Evaluation Criteria in Solid Tumours version 1.1; TNBC, triple-negative breast cancer.

Extended Data Fig. 6. Impact of selected baseline factors on progression-free survival (PFS).

Impact of selected baseline factors on PFS. Univariate Cox proportional hazard (CpxPH) analysis performed across patients with available PFS data. The forest plot shows the hazard ratios (HRs), that is, the exponentiated coefficients of the CoxPH model, with 95% confidence intervals as error bars, according to each single factor. Statistical significance was assessed by the two-sided likelihood-ratio test. P-values are based on the univariate likelihood-ratio test performed for a subset of selected baseline factors, thus p-values adjusted for multiple testing are not reported. BMI, body mass index; CPI, checkpoint inhibitor; CRP, C-reactive protein; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; IHC, immunohistochemistry; LDH, lactate dehydrogenase; SNV, single-nucleotide variant.

Extended Data Fig. 7. Correlation of clinical response and autogene cevumeran-induced T-cell response.

Correlation of clinical best overall response (BOR) with de novo and amplified neoantigen-specific T-cell responses. Either the cumulative IFNγ spot count (fitness-normalized, per 1×106 cells) of all targets with autogene cevumeran response per type of cells assayed or the number of neoantigen-specific T-cell responses detected, both measured by ex vivo ELISpot assay post-vaccination, were correlated with clinical response. T-cell responses for monotherapy were measured after ≥7 vaccinations in patients treated with autogene cevumeran monotherapy in phase 1a or participants of the biomarker substudy not yet treated with atezolizumab. Participants treated with autogene cevumeran and atezolizumab are from phase 1b. Progression-free survival (PFS) censoring events were labelled with a 1.

Extended Data Fig. 8. Phenotypic analysis of neoantigen-specific CD8 T cells from longitudinal time points using CYTOF in peripheral blood of patient 13.

a, Frequencies of GALNT6 (E579K) -specific CD8 + T cells (VPKDKEWEL) in patient 13 at day 22, 106 and 232. b, UMAP (uniform manifold approximation and projection) of CD8 + T cells from patient 13 at day 22, 106 and 232. GALNT6 (E579K) -specific cells are highlighted. UMAP plots on the right display relative expression intensities of all phenotypic markers assessed. c, Histograms showing relative expression of individual markers on GALNT6 (E579K) -specific (red), Naïve (gray) or Effector memory (black) CD8 T cells from patient 13 at different times.