一周资讯NO.103 | 聚焦医药动态0304~0310









国家药监局药审中心关于发布《化学合成多肽药物药学研究技术指导原则(试行)》的通告(2023年第12号) (







BioMarin Pharmaceutical近日宣布,FDA接受其Voxzogo(vosoritide)注射液的补充新药申请(sNDA),用以治疗5岁以下软骨发育不全(achondroplasia)儿童。FDA预计2023年10月21日前完成审查。今年1月,欧洲药品管理局(EMA)接受Voxzogo扩增适应症的申请,用以治疗2岁以下软骨发育不全孩童






近日,lonis Pharmaceuticals公司宣布,美国FDA已经接受为反义寡核苷酸疗法(ASO)eplontersen递交的新药申请(NDA),用于治疗遗传性转甲状腺素蛋白(TTR)介导的淀粉样变性多发性神经病(ATTRv-PN)患者。FDA预计在今年12月22日之前作出审评决定。






近日,强生旗下杨森(Janssen)公司宣布向FDA递交了新药申请(NDA),将PARP抑制剂niraparib和醋酸阿比特龙(abiraterone acetate)以双效片剂(DAT)的形式,联合强的松(prednisone)共同用于治疗BRCA阳性的转移性去势抵抗性前列腺癌(mCRPC)患者。新闻稿指出,如果获批,这将是美国首个可用于治疗携带BRCA突变的mCRPC患者的DAT制剂













近日,BridgeBio Pharma公司宣布,在研疗法infigratinib在治疗软骨发育不全(achondrtoplasia)儿童患者的2期临床实验中表现出潜在“best-in-class”的疗效和安全性。接受最高剂量治疗的患儿在6个月时,年平均身高生长速度(AHV)与基线相比提高3.03厘米/年。Infigratinib是一款FGFR3抑制剂,基于这一结果,该公司将于今年启动关键性3期临床试验。



近日,韩国庆熙大学药学院和Prazer Therapeutics公司的科学家合作,开发了特异性降解磷酸化p38蛋白的靶向蛋白降解剂。在阿尔兹海默病的小鼠模型中,它能够改善认知并且降低淀粉样蛋白斑块的积累。



ScienceDirect | 用于胃肠道肿瘤无创成像的靶向CLDN18.2免疫PET探针研究结果


近日,创胜集团(06628.HK),一家具备生物药品发现、研发、工艺开发和生产全流程整合能力的临床阶段的生物制药公司,近日宣布其用于胃肠道肿瘤无创成像的靶向CLDN18.2免疫PET探针【89Zr】Zr-DFO-TST001的研究结果在期刊《Journal of Pharmaceutical Analysis》上发表。



  • Claudin18.2 (CLDN18.2) is a tight junction protein that is overexpressed in a variety of solid tumors such as gastrointestinal cancer and oesophageal cancer. It has been identified as a promising target and a potential biomarker to diagnose tumor, evaluate efficacy and determine patient prognosis. TST001 is a recombinant humanized CLDN18.2 antibody that selectively binds to the extracellular loop of human Claudin18.2. In this study, we constructed a solid target radionuclide zirconium-89 (Zr) labled-TST001 to detect the expression of in the human stomach cancer BGC823 cell lines. The [Zr]Zr-DFO-TST001 showed high radiochemical purity (RCP, >99%) and specific activity (24.15 ± 1.34 GBq/μmol), and was stable in 5% human serum albumin (HSA), and phosphate buffer saline (PBS) (>85% RCP at 96 h). The concentration of 50% maximal effect (EC) values of TST001 and DFO-TST001 were as high as 0.413 ± 0.055 nM and 0.361 ± 0.058 nM (P > 0.05), respectively. The radiotracer had a significantly higher uptake in CLDN18.2-positive tumors than in CLDN18.2-negative tumors (1.11 ± 0.02 vs. 0.49 ± 0.03, P = 0.0016) 2 days post injection (p.i.). BGC823 mice models showed high T/M values 96 h p.i. with [Zr]Zr-DFO-TST001 was much higher than those of the other imaging groups. Immunohistochemistry (IHC) results showed that BGC823 tumors were highly positive (+++) for CLDN18.2, while those in the BGC823 group did not express CLDN18.2 (-). The results of ex vivo biodistribution studies showed that there was a higher distribution in the BGC823 tumor bearing mice (2.05 ± 0.16 %ID/g) than BGC823 mice (0.69 ± 0.02 %ID/g) and blocking group (0.72 ± 0.02 %ID/g). A dosimetry estimation study showed that the effective dose of [Zr]Zr-DFO-TST001 was 0.0705 mSv/MBq, which is within the range of acceptable doses for nuclear medicine research. Taken together, these results suggest that good manufacturing practices (GMPs) produced by this immuno-positron emission tomography (immuno-PET) probe can detect CLDN18.2-overexpressing tumors.89CLDN18.28950CLDN18.289CLDN18.2CLDN18.289

  • Graphical abstract

  • Keywords

  • 1. Introduction

    According to the cancer epidemiology report released in 2022, lung cancer is the primary cause of cancer death, followed by digestive tract tumors (such as stomach cancer, colorectal cancer, liver cancer, oesophageal cancer, etc.). In China, gastrointestinal cancers account for 45% of cancer-related deaths, likely because gastrointestinal cancers are mostly diagnosed in the advanced stage and patients often have a poor prognosis[[1][2][3]]. Gastrointestinal cancers have become the primary medical and economic burden for people in China. In addition to traditional chemotherapy, and immunotherapy, little progress has been made with novel chemotherapies and targeted therapies for gastrointestinal tumors[[4][5][6][7]]. Among the 70 novel first-line agents approved for cancer treatment, only 5 drugs have been approved for advanced gastrointestinal cancer and the survival rates are still low based on data from the last five years[8]. Therefore, strategies to improve the survival of patients with advanced gastrointestinal cancer remain an unmet medical necessity.

    CLDN18.2 is a tight junction protein belonging to the CLDN protein family (CLDNs) that is involved in the formation of intercellular adhesion structures, and controls cell polarity and the exchange of substances between cells[[9][10][11]]. Its expression is strictly limited to normal gastric mucosal cells, but is overexpressed in the process of proliferation, division and metastasis of tumor cells, making it an emerging therapeutic target for digestive tract tumor therapy[12,13]. Zolbetuximab (IMAB362) is the first targeted CLDN18.2 antibody that kills tumor cells through antibody-dependent cytotoxicity(ADCC) and complement-dependent cytotoxicity(CDC), and in combination with first-line epirubicin, oxaliplatin and capecitabine (EOX) to provide longer progression-free and overall survival[14]. TST001 is an anti-CLDN18.2 monoclonal antibody developed worldwide after IMAB362. Compared to IMAB362, TST001 has a higher affinity, higher FcR binding activity due to lower fucose content and stronger NK cell-mediated ADCC tumor killing activity. In a phase I clinical study of TST001 (NCT04396821) in combination with capecitabine and oxaliplatin (CAPOX) as a first-line agent for advanced gastric/gastroesophageal junction adenocarcinoma, 73.3% achieved partial response, and 26.7% achieved stable disease[15]. A phase I study (NCT03874897) of CLDN18.2 CAR-T therapy conducted by Shen et al. [16] showed that after receiving CLDN18.2 CAR-T infusion, the overall response rate (ORR) and disease control rate (DCR) reached 48.6% and 73.0%, respectively. Interestingly, both clinical studies indicate that the CLDN18.2 expression level was correlated with drug efficacy, showing more clinical benefit in patients with high CLDN18.2 expression in tumors. Therefore, patient selection based on CLDN18.2 expression level becomes critical for CLDN18.2-targeted therapy. At present, the major detection method of CLDN18.2 protein is immunohistochemistry (IHC), and other methods include molecular beacons and reverse transcription-polymerase chain reaction (RT‒PCR)[17]. IHC is invasive, and requires endoscopic biopsy, and the sampling site and number are limited. Due to the heterogeneous nature of tumor, the CLDN18.2 distribution and dynamic changes in expression levels in patients cannot be fully reflected in real-time. Molecular imaging can be used as a noninvasive diagnostic tool to detect the expression and distribution of CLDN18.2 in the lesion using the radioactive signal emitted by the radiotracer, thereby helping to clinically screen patients with potential benefit, evaluate the efficacy of CLDN18.2 targeted therapy, and guide the accurate diagnosis and treatment of tumors. A recent study showed that 18F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) parameters including maximum standard uptake value (SUVmax), metabolic tumor volume (MTV) and total lesion glycolysis (TLG) did not predict CLDN 18.2 expression status in diffuse-type gastric cancer[18]. Hu et al. [19] developed three antibodies (anti-CLDN18.2 VHH, anti-CLDN18.2 VHH-ABD and anti-CLDN18.2 VHH-Fc) of different molecular weight sizes for PET/CT imaging, and identified [89Zr]-anti-CLDN18.2 VHH-ABD as the most appropriate imaging agent (high tumor uptake and low uptake in the liver) in preclinical studies. However, in a subsequent clinical study, [89Zr]-VHH-Fc was found to be more specific and persistent than [89Zr]-anti-CLDN18.2 VHH-ABD, and was also considered to be a molecular imaging tracer with potential value for cancer diagnosis, as it contains CLDN18.2[20]. More recently, we explored a CLDN18.2-specific murine mAb 5C9 by DNA immunization, and modified 5C9 with 124I, Cy5.5 and FD1080. The results of these studies support the targeted therapy of CLDN18.2-positive tumors by using immuno-PET imaging and near-infrared fluorescent II imaging to localize tumors and guide surgery for orthotopic CLDN18.2-positive tumors[21].

    Due to the superior targeting specificity and high sensitivity of molecular imaging technology, we used the TST001 antibody produced under GMP conditions to construct the immuno-PET molecular probe [89Zr]Zr-DFO-TST001. The goal of this study was to assess the ability of [89Zr]Zr-DFO-TST001 to characterize CLDN18.2 expression.

  • 2. Material and methods

    2.1. Materials

    All reagents were obtained from Sigma‒Aldrich (Shanghai, China). P-isothiocyanatobenzyl-desferrioxamine B (p-NCS-Bz-DFO) was purchased from Macrocyclics (Plano, TX, USA). The GMP grade CLDN18.2 antibody TST001 was kindly provided by Suzhou Transcenta Therapeutics Co., Ltd. (Suzhou, China). Radionuclide 89Zr was produced and purified by the Cyclotron team of the Nuclear Medicine Department of Peking University Cancer Hospital (Beijing, China). The medium, fetal bovine serum (FBS), trypsin ethylene diamine tetraacetic acid (EDTA) and pen-strep solution were purchased from Biological Industries (Beijing, China). Radioimmunoprecipitation assay (RIPA) lysis buffer was obtained from Themo Fisher Scientific (Waltham, MA, USA). Diaminobenzidine (DAB) was provided by Jinqiao Biological Company (Beijing, China). PD-10 column was purchased from GE Healthcare (Buckinghamshire, England).

    2.2. Radiolabeling of TST001 with 89Zr

    For 89Zr labeling, 89Zr-oxalic acid was neutralized to pH 7.0 using 0.25 M 2-[4-(2-hydroxyethyl)-1-piperazinyl]ethanesulfonic acid (HEPES) and 1 M Na2CO3 buffer, and then mixed with previously described DFO-TST001 for 60 min at 37 °C. The reaction mixture was purified by a PD-10 column with 0.01 M phosphate buffer saline (PBS, 2.5 mL, pH 7.4).

    2.3. Small-animal PET imaging of [89Zr]Zr-DFO-TST001

    Normal KM mice and BGC823CLDN18.2/BGC823 model nude mice were injected with 7.4 MBq of [89Zr]Zr-DFO-TST001 via the tail vein (n = 3). Then 10 min static PET scans were acquired at each time point (2, 24, 48, and 72 h p.i.). As a non-specific control group, BGC823CLDN18.2 mice (n = 3) were fasted 6 h in advance, then injected with 7.4 MBq of 18F- FDG via the tail vein. The mice were anesthetized with 2% isoflurane before and during the 18F-FDG PET imaging. With a small-animal PET/CT scanner (Super Nova PET/CT, Pingseng Healthcare, Shanghai, China), the PET images were reconstructed by Avatar 3 (Pingseng Healthcare), and the regions of interest (ROIs)-derived SUV was calculated by drawing ROIs over these organs.

    2.4. Ex vivo biodistribution

    The KM mice were intravenously injected with 0.74 MBq of [89Zr]Zr-DFO-TST001 via the tail vein and were then sacrificed at 2, 24, 48, 72 and 144 h p.i. (n = 4). The tissues including the blood, heart, liver, spleen, lung, kidneys, stomach, intestines, muscle, bone and brain were dissected. The radioactivity of the tissues was measured using a γ-counter (PerkinElmer, Waltham, MA, USA). The radioactivity of each organ was calculated as % injected dose per gram (%ID/g). For the tumor model’s ex vivo biodistribution, female nude mice bearing BGC823CLDN18.2 and BGC823 tumor xenografts were injected by tail vein with 0.74 MBq of [89Zr]Zr-DFO-TST001 to evaluate the distribution of [89Zr]Zr-DFO-TST001 in major organs and tumors (n = 4 per group). The mice were sacrificed and dissected at 48 h p.i. (n = 4), and the tumor, kidney, blood, and other major organs were collected and weighed. The blocking study was also performed in BGC823CLDN18.2 mice by a co-injection of 0.74 MBq of [89Zr]Zr-DFO-TST001 with an excess dose of cold TST001 (1 mg). At 48 h p.i., the blocked mice were sacrificed and dissected. Then, the organ biodistribution of [89Zr]Zr-DFO-TST001 was determined.

    2.5. Dosimetry estimation

    For human radiation dosimetry, animal biodistribution data were obtained by the standard

    method of organ dissection. The human organ radiation dosimetry data were extrapolated from the biodistribution data of [89Zr]Zr-DFO-TST001 in KM mice by OLINDA/EXM 2.0 software (Vanderbilt University, Nashville, TN, USA).

    2.6. Statistical analysis

    Quantitative data are expressed as the mean ± standard deviation (SD), with all error bars denoting the SD. The means were compared using Student’s t test, and P values of less than 0.05 were considered to indicate statistical significance.

  • 3. Results and discussion

    3.1. Molecular characteristic of conjugation

    The molecular weight of the CLDN18.2 antibody, TST001, was approximately 148 kDa, which was further determined to be exactly 148,723 Da (Fig. 1A). DFO-TST001 was chelated with an approximately double-DFO chelator with a molecular weight of 150320 Da (Fig. 1B). Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS‒PAGE) showed that both TST001 and DFO-TST001 had bands at approximately 150 kDa with no other bands (Fig. 1C), which indicated that the conjugation was of excellent quality as no antibody aggregates or antibody fragments were detected. The enzyme-linked immunosorbent assays (ELISA) results showed that the EC50 value of DFO-TST001 binding to CLDN18.2 was not significantly different from that of TST001 (0.413 nM ± 0.055 nM vs. 0.361 ± 0.058 nM, P > 0.05, Fig. 1D). The binding assay demonstrated both TST001 and DFO-TST001 can form a strong bond with CLDN18.2, and the conjugation of the chelator DFO had no impact on the affinity of TST001 to CLDN18.2.

    Fig. 1. Molecular characterization of TST001 and desferrioxamine-TST001 (DFO-TST001). (A) Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) of TST001. (B) MALDI-TOF-MS of DFO-TST001. (C) Nonreducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) characterization. (D) Binding of TST001 and DFO-TST001 to human CLDN18.2 protein was evaluated by enzyme-linked immunosorbent assays (ELISA).

    3.2. Radiosynthesis, quality control, and in vitro stability

    The synthesis process of [89Zr]Zr-DFO-TST001 is shown in Fig. 2A. [89Zr]Zr-DFO-TST001 was manually prepared with a radiolabeling yield of 74.64% ± 4.41% (n = 3, nondecay corrected). The RCP of [89Zr]Zr-DFO-TST001 was more than 99% in 0.01 M PBS (pH 7.4) (Fig. 2B). The in vitro stability of [89Zr]Zr-DFO-TST001 in 0.01 M PBS or 5% human serum albumin (HSA) was demonstrated by an RCP of more than 85% after 96 h at room temperature (RT). (Fig. 2C). The excellent in vitro stability also showed that the TST001 structural modification and labeling method was feasible. Quality control results are shown in Table 1.

    Fig. 2. Synthesis, quality control and vitro stability of [89Zr]Zr-DFO-TST001. (A) Synthesis and radiolabelling of [89Zr]Zr-DFO-TST001. (B) Radio-thin-layer chromatography scanner (Radio-TLC) results of [89Zr]Zr-DFO-TST001 before and after purification. (C) In vitro stability of [89Zr]Zr-DFO-TST001

    Table 1. Quality control of [89Zr]Zr-DFO-TST001.

    ParameterQC specificationQC result
    AppearanceClear, colorlessPass
    Volume1-2 mL1 mL
    Radiochemical purity>95%>99%
    Endotoxins<15 EU/mLPass
    Specific activity18.5-296 GBq/μmol24.15 ± 1.34 GBq/μmol

    3.3. In vitro CLDN18.2 expression of cell lines

    Western blotting results confirmed that the expression of CLDN18.2 in BGC823CLDN18.2 cells was significantly different from that in BGC823 cells (Fig. 3A). The relative expression of CLDN18.2 in the BGC823CLDN18.2 and BGC823 cell lines was 1.37 ± 0.24 and 0.23 ± 0.01, respectively (P = 0.0013, Fig. 3B). Flow cytometry experiments revealed that 86.2% of cells were positively stained with anti-CLDN18.2 antibody (1D5) in the BGC823CLDN18.2 group (Fig. 3C). The differences in CLDN18.2 expression measured by western blotting and flow cytometry were then validated between the human gastric cancer cell lines BGC823 and BGC823CLDN18.2. The result of the cellular uptake experiment showed that the uptake of [89Zr]Zr-DFO-TST001 in BGC823CLDN18.2 cells increased in a time-dependent manner (7.33% ± 0.84% at 10 min, 7.97% ± 0.56% at 30 min, 11.47% ± 0.32% at 60 min and 13.37% ± 2.04% at 120 min), while no significant changes were observed in the BGC823 group (4.21% ± 0.21% at 10 min, 3.77% ± 0.53% at 30 min, 4.57% ± 0.36% at 60 min and 5.54% ± 0.21% at 120 min). The uptake by BGC823CLDN18.2 cells (CLDN18.2 positive) was significantly higher than that by BGC823 cells (CLDN18.2 negative) at each selected time point (P < 0.0004). Meanwhile, an excess of unlabeled TST001 significantly blocked the uptake of [89Zr]Zr-DFO-TST001 (11.47% ± 0.32% vs. 3.24% ± 0.36% at 60 min, 13.37% ± 2.04% vs. 5.64% ± 0.21% at 120 min) (Fig. 3D). In the cellular uptake experiment, the uptake of [89Zr]Zr-DFO-TST001 by BGC823CLDN18.2 cells at 60 min was 2.51-fold higher than that of BGC823 cells and 3.54-fold higher than that of the blocking group. The specificity of [89Zr]Zr-DFO-TST001 for CLDN18.2 was thus demonstrated at the cellular level.

    Fig. 3. CLDN18.2 expression in two cell lines, and cellular uptake of [89Zr]Zr-DFO-TST001. (A) Western blotting results of CLDN18.2 expression in the BGC823CLDN18.2 and BGC823 cell lines. (B) Relative expression of CLDN18.2 in BGC823CLDN18.2 and BGC823 cells (results are shown as the mean ± SD, n = 3). (C) Flow cytometry histogram of BGC823CLDN18.2 and BGC823 cells. (D) Cellular uptake of [89Zr]Zr-DFO-TST001 in BGC823CLDN18.2 and BGC823 cells. (**, P< 0.05, ***, P< 0.001, ****, P< 0.0001).

    3.4. Dosimetry estimation

    The biodistribution study of [89Zr]Zr-DFO-TST001 demonstrated favorable pharmacokinetics with a relatively long half-life in vivo (Fig. S1A). Human organ radiation dosimetry is shown in Table 2. The liver received the highest dose (0.360 mSv/MBq), followed by the gallbladder wall (0.155 mSv/MBq). The effective dose was 0.0705 mSv/MBq. When a patient was injected with 74 MBq of [89Zr]Zr-DFO-TST001 for imaging, its effective radiation dose was less than 5.217 mSv, which is acceptable in routine nuclear medicine research. The estimated human radiation burden due to a single i.v. [89Zr]Zr-DFO-TST001 injection is comparable to that of other 89Zr-labelled monoclonal antibodies [[22][23][24]], and is suitable for clinical research.

    Table 2. Estimates of the mean absorbed radiation dose.

    OrganmSv/MBq (10-2)
    Gallbladder Wall15.50
    Left colon3.65
    Small Intestine5.86
    Stomach Wall6.15
    Right colon4.82
    Heart Wall9.19
    Salivary Glands1.41
    Red Marrow4.61
    Osteogenic Cells10.90
    Urinary Bladder Wall0.81
    Total Body2.72
    Effective Dose7.05

    3.5. Small-animal PET/CT imaging and IHC analysis

    Small-animal PET/CT imaging at different time points (2, 24, 48, 72 and 120 h) after injection of [89Zr]Zr-DFO-TST001 into KM mice, showed high uptake in the heart, liver and spleen (Supporting Information Fig. S1B). The standard uptake value average (SUVmean) of some organs measured by ROIs is shown in Supporting Information Fig. S1C. After 2 h, the SUVmean was 2.57 ± 0.02 in the heart, 2.27 ± 0.01 in the liver and 1.86 ± 0.01 in the spleen, respectively. The ratio of heart to muscle (H/M) was 20.30 ± 0.91. After 120 h, the SUVmean in the heart, liver and spleen were 0.49 ± 0.01, 1.36 ± 0.02 and 1.21 ± 0.01, respectively, and almost no special intake was observed in the stomach. The images are consistent with the biodistribution results.

    The in vivo distribution and metabolic characteristics of [89Zr]Zr-DFO-TST001 were evaluated in real time and noninvasively via small-animal PET/CT imaging at 2, 24, 48, 72 h and 96 h p.i. of the radiotracer. Meanwhile, we set up the following three control groups, which were blocked by excess TST001, negative CLDN18.2 expression in BGC823 cells and nonspecific targeting of [89Zr]Zr-DFO-IgG (7.4 MBq), respectively. SUVmean data were collected for organs of BGC823CLDN18.2 or BGC823 mice by outlining the ROI from the immune-PET images (Fig. 4). The tumor sites in the [89Zr]Zr-DFO-TST001 group still had obvious uptake at 96 h p.i. In the BGC823CLDN18.2 model with [89Zr]Zr-DFO-TST001, the SUVmean continued to increase within 48 h p.i. and reached a maximum uptake value of 1.09 ± 0.03 at 48 h. In addition, until 96 h p.i., the SUVmean of the BGC823CLDN18.2 model was significantly different from that of the BGC823 model and blocking group (1.03 ± 0.03, 0.41 ± 0.05, 0.51 ± 0.07, respectively, P < 0.0002). Using [89Zr]Zr-DFO-IgG as a negative control probe, the results showed that in the BGC823CLDN18.2 model mice except for the tumor uptake value slightly higher than [89Zr]Zr-DFO-TST001 at 2 h after injection (0.51 ± 0.01 vs. 0.37 ± 0.02), the [89Zr]Zr-DFO-IgG tumor uptake value at all other time points (24 h, 48 h, 72 h and 96 h) was significantly lower than that of [89Zr]Zr-DFO-TST001 (0.55 ± 0.04 vs. 0.96 ± 0.12, 0.53 ± 0.02 vs. 1.10 ± 0.12, 0.54 ± 0.04 vs. 1.06 ± 0.06 and 0.47 ± 0.01 vs. 1.03 ± 0.01) (Fig. S2). Over time, compared with other imaging groups, the uptake of [89Zr]Zr-DFO-TST001 was mostly concentrated in the tumor in the BGC823CLDN18.2 model, and the uptake values of the heart, liver, and other organs were greatly reduced.

    Fig. 4. Small-animal positron emission tomography (PET) images of BGC823CLDN18.2 or BGC823 tumor mice injected with [89Zr]Zr-DFO-TST001 or [89Zr]Zr-DFO-IgG. (A) Small-animal PET images of four different groups at 2, 24, 48, 72 and 96 h. (B) Standard uptake value average (SUVmean) of [89Zr]Zr-DFO-TST001 in the organs of BGC823CLDN18.2 mice (C) SUVmean of [89Zr]Zr-DFO-TST001 in organs of BGC823CLDN18.2 mice with unlabelled TST001 blockade. (D) SUVmean [89Zr]Zr-DFO-TST001 in the organs of BGC823 mice. (E) SUVmean of [89Zr]Zr-DFO-IgG in organs of BGC823CLDN18.2 mice.

    For comparison with the gold-standard probe 18F- FDG, BGC823CLDN18.2 tumor-bearing mice were given 18F-FDG and images were collected 1 h p.i. (Fig. 5A). The results showed that the uptake of 18F-FDG in CLDN18.2-positive mice was similar to the background uptake. The tumor accumulation of [89Zr]Zr-DFO-TST001 in BGC823CLDN18.2 mice 48 h p.i. was approximately 4.15-fold that of the blocking group, 2.27-fold that of the BGC823 group, and 2.05-fold that of the [89Zr]Zr-DFO-IgG group (SUVmean values were 1.11 ± 0.02, 0.27 ± 0.01, 0.49 ± 0.03, 0.54 ± 0.06, respectively) (Fig. 5B). The tumor/heart (T/H) ratios and tumor/muscle (T/M) ratios at each time point after injection of [89Zr]Zr-DFO-TST001 were significantly higher than those of the other control groups (Figs. 5C and D), and at 96 h p.i., the T/H and T/M ratios reached their maximum of 2.37 ± 0.04, 14.95 ± 1.63, respectively.

    Fig. 5. Analysis of small-animal PET imaging. (A) Section images of tumor uptake 48 h p.i. were compared to section images of 18F-fluorodeoxyglucose (18F-FDG) in BGC823CLDN18.2 mice 1 h p.i. (B) SUVmean in the organs of different experimental group mice in organs at 48h. (C) Tumor/Heart at each point p.i. (D) Tumor/Muscle at each point p.i. (E) Immunohistochemistry (IHC) analysis of CLDN18.2 expression in BGC823CLDN18.2 (++) (e1) and BGC823 (-) (e2) tumors. (***, P< 0.001).

    The T/NT value of [89Zr]Zr-DFO-TST001 was significantly different 48 h p.i. when comparing the BGC823CLDN18.2 model to other groups. Compared with our previous research, TST001 is a humanized antibody with better immune responsiveness to the CLDN18.2 receptor. Second, the patient needs to receive iodine to block the thyroid gland before and during 124I imaging, which greatly reduces patient compliance[21]. Labelling with 89Zr would appear to be more robust and better available. Nevertheless, a remarkably high background in the liver and spleen was also noted with [89Zr]Zr-DFO-TST001, which might be a result of nonspecific binding and hepatobiliary clearance. This is very similar to previous studies on the 89Zr-labelled antibody[25,26]. From an imaging perspective, this not only results in problems for tumor localization in the liver and spleen region, but it also might lead to false-positive results when “tumor CLDN18.2 expression” and further cause erroneous selection of candidate patients for this therapy. Although the interactions between FcγR expressed on immune effector cells and the Fc region of antibodies can trigger antibody-mediated therapeutic responses, they may not be favorable in the context of molecular imaging. According to our research, there are three initial resolutions to reduce nonspecific uptake by the liver and spleen[27,28]. Firstly, the preparation of probes using antibody fragments such as Fab, F(ab)2 to replace intact antibodies not only avoids the interaction of the Fc region with the immune system, but also allows the probes to have a faster pharmacokinetic profile. Secondly, another strategy is predicated on genetically engineering the Fc region of an IgG to abrogate its binding with FcγRs on immune cells while maintaining its ability to bind FcRn. Thirdly, a more facile and modular approach may lie in manipulating the glycans of the Fc region. In addition, from the nature of the nuclide, 89Zr is a radioactive metal ion that first ligates the antibody by a suitable chelating agent (typically using a lysine group) and then indirectly labels the antibody by non-covalently chelating the radioactive metal ion. Once antibodies have been internalized into the tumor cells, they are subject to catabolism through lysosomal degradation. The catabolites of radiometal ion chelates remain trapped (residualized) inside the cells, leading to an accumulation of radiometal (and PET signal) in the target tumor t issue and metabolic organ over time. However, iodine is usually labeled directly onto antibodies through a simple and widely used procedure, and most iodine-containing catabolites are nonpolar molecules that are rapidly lost from the liver and spleen[29]. Based on this property of radionuclide iodine, we are also conducting a study related to 124I labeled TST001, which may be more suitable for clinical translation in the future.

    We also performed 18F-FDG PET/CT imaging as a reference. The tumor SUVmean of [89Zr]Zr-DFO-TST001 was higher than that of 18F-FDG (1.10 ± 0.12 vs. 0.40 ± 0.02) at the tumor sites in the BGC823CLDN18.2 model, and the T/M value of [89Zr]Zr-DFO-TST001 was also much higher than that of 18F-FDG (10.23 ± 1.30 vs. 1.80 ± 0.22).

    The results of IHC revealed high and homogenous CLDN18.2 expression in BGC823CLDN18.2 tumors, and the BGC823 xenograft tumors were negative for CLDN18.2 (Fig. 5E). The stomachs of BGC823CLDN18.2 and BGC823 tumor-bearing mice showed substantially positive expression of CLDN18.2. Neither the liver nor spleen tissue of the two types of tumor-bearing mice expressed CLDN18.2. The IHC results showed that the BGC823CLDN18.2 tumors were strongly positive for CLDN18.2 (+++), while the BGC823 tumors were negative (-), which was consistent with the imaging and western blotting results. These results prove that the [89Zr]Zr-DFO-TST001 probe we constructed has the ability to specifically target CLDN18.2. In addition, a strong positive expression of CLDN18.2 (+++) was also observed in the gastric mucosa of all mice, but neither PET/CT imaging nor biodistribution showed any obvious uptake and retention of the probe in the stomach, likely because the expression of CLDN18.2 in vivo was limited to the gastric mucosa, and monoclonal antibodies had difficulty accessing the hidden CLDN18.2 binding epitope in the gastric mucosa[30] (Fig. S3).

    3.6. Ex vivo biodistribution

    The biodistribution of [89Zr]Zr-DFO-TST001 in BGC823CLDN18.2 and BGC823 tumor–bearing mice is presented in Fig. 6. At 48 h p.i., the livers in all three groups showed relatively high uptake (8.39 ± 0.59 %ID/g in BGC823CLDN18.2 group, 9.28 ± 0.19 %ID/g in BGC823 group and 20.96 ± 0.88 %ID/g in blocking the group, respectively). The uptake value of the spleen was second to that of the liver (3.54 ± 0.26 %ID/g in BGC823CLDN18.2 group, 2.08 ± 0.29 %ID/g in BGC823 group and 1.93 ± 0.24 %ID/g in the blocking group, respectively). Tumor uptake in BGC823CLDN18.2 tumor bearing mice was higher (2.05 ± 0.16 %ID/g) than that in the BGC823 mice (0.69 ± 0.02 %ID/g) and blocking group (0.72 ± 0.02 %ID/g). (Fig. 6A). The tumor/liver (T/L) and tumor/brain (T/B) ratios of BGC823CLDN18.2 tumors were significantly higher than those of the other two control groups. (T/L: 0.075 ± 0.001 in the BGC823 group vs. 0.25 ± 0.003 in the BGC823CLDN18.2 group vs. 0.035 ± 0.002 in the blocking group, T/B: 16.03 ± 1.66 in the BGC823 group vs. 40.35 ± 3.68 in the BGC823CLDN18.2 group vs. 3.01 ± 0.53 in the blocking group, Figs. 6B and D). The tumor/stomach (T/S) ratios were not significantly different among the three groups (2.00 ± 0.13 in BGC823 vs. 2.04 ± 0.43 in BGC823CLDN18.2 vs. 1.47 ± 0.50 in blocking group, Fig. 6C). Consistent with the PET/CT results, in vitro biodistribution data at 48 h p.i. showed that [89Zr]Zr-DFO-TST001 aggregated in the liver and spleen, and the liver uptake in the blocking group was significantly higher than that in the other two groups, possibly because tumor uptake was blocked, resulting in the probes entering the liver directly through the bloodstream for metabolism. The difference in tumor uptake values in the three groups also reflects the excellent specificity of [89Zr]Zr-DFO-TST001 for CLDN18.2-positive tumors.

    Fig. 6. Biodistribution in the three different tumor models 48 h p.i. (A) Biodistribution of three different tumor models p.i. 48 h. (B) Tumor/Liver p.i. 48 h. (C) Tumor/Stomach 48 h p.i. (D) Tumor/Brain 48 h p.i. (***, P< 0.001; ****, P<0.0001; ns, no significant difference in statistics).

    • Download : Download high-res image (411KB)

    • Download : Download full-size image

    • Download : Download high-res image (1MB)

    • Download : Download full-size image

    • Download : Download high-res image (1MB)

    • Download : Download full-size image

    • Download : Download high-res image (293KB)

    • Download : Download full-size image

    • Download : Download high-res image (355KB)

    • Download : Download full-size image

    • Download : Download high-res image (340KB)

    • Download : Download full-size image

  • 4. Conclusion

    We successfully prepared 89Zr labelling of a GMP grade anti-CLDN18.2 recombinant humanized antibody TST001. [89Zr]Zr-DFO-TST001 exhibited good specificity at the cellular level and rapid tumor accumulation which remained positive from 24 to 96 h. It provides a promising molecular probe for detecting the treatment effects of therapeutic antibodies in humans in real time. It also provides a possibility for the screening and efficacy evaluation of patients targeted for CLDN18.2 therapy in the future.

  • CRediT author statement

    Yan Chen: Investigation, Methodology, Software, Formal analysis, Data curation, Writing - Original draft preparation, Reviewing and Editing, Visualization; Xingguo Hou: Investigation, Methodology, Software, Formal analysis, Data curation, Writing - Original draft preparation; Dapeng Li: Investigation, Methodology, Software, Writing - Original draft preparation; Jin Ding: Conceptualization, Investigation, Resources, Validation; Jiayue Liu: Methodology, Software, Formal analysis; Zilei Wang: Methodology, Software, Formal analysis; Fei Teng: Resources, Validation, Supervision; Hongjun Li: Resources, Validation, Supervision; Fan Zhang: Resources, Validation, Supervision; Yi Gu: Resources, Validation, Supervision; Steven Yu: Resources, Validation, Supervision; Xueming Qian: Investigation, Resources, Validation, Supervision; Zhi Yang: Conceptualization, Methodology, Investigation, Resources, Validation, Supervision; Hua Zhu: Conceptualization, Methodology, Investigation, Resources, Validation, Writing - Reviewing and Editing, Supervision.

  • Declaration of competing interest

    Intellectual properties protection have been filed by Suzhou Transcenta Therapeutics co., LTD, inventor of Xueming Qian; Fei Teng; Hongjun Li; Yi Gu, and Beijing Cancer Hospital, inventor of Hua Zhu; Yang Zhi; Jin Ding; Feng Wang. All authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  • Acknowledgments

    The research was funded by National Natural Science Foundation of China (Grant Nos.: 82171973, 82171980, and 82102092), the Capital’s Funds for Health Improvement and Research (Grant No.: 2022-1G-1021), and Beijing Millions of Talent Projects A level funding, grant number (Grant No.: 2019A38). The study was also supported by Beijing Hospitals Authority Dengfeng Project (Grant No.: DFL20191102), The Pilot Project (4th Round) to Reform Public Development of Beijing Municipal Medical Research Institute (2021−1), and the third foster plan in 2019 “Molecular Imaging Probe Preparation and Characterization of Key Technologies and Equipment” for the development of key technologies and equipment in major science and technology infrastructure in Shenzhen.

  • Appendix A. Supplementary data

    The following is the Supplementary data to this article.

    Download : Download Word document (2MB)

  • References

    • [1]

    • H. Sung, J. Ferlay, R.L. Siegel, et al.Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countriesCA Cancer J. Clin., 71 (2021), pp. 209-249

       View PDF 

      This article is free to access.

      CrossRefGoogle Scholar

    • [2]

    • W. Cao, H.-D. Chen, Y.-W. Yu, et al.Changing profiles of cancer burden worldwide and in China: a secondary analysis of the global cancer statistics 2020Chin. Med. J (Engl)., 134 (2021), pp. 783-791

      View article 

      CrossRefView in ScopusGoogle Scholar

    • [3]

    • H. Qiu, S. Cao, R. XuCancer incidence, mortality, and burden in China: a time-trend analysis and comparison with the United States and United Kingdom based on the global epidemiological data released in 2020Cancer Commun (Lond), 41 (2021), pp. 1037-1048

       View PDF 

      This article is free to access.

      CrossRefGoogle Scholar

    • [4]

    • Y.Y. Janjigian, K. Shitara, M. Moehler, et al.First-line nivolumab plus chemotherapy versus chemotherapy alone for advanced gastric, gastro-oesophageal junction, and oesophageal adenocarcinoma (CheckMate 649): a randomised, open-label, phase 3 trialLancet, 398 (2021), pp. 27-40View PDFView articleView in ScopusGoogle Scholar

    • [5]

    • E. Van Cutsem, V.M. Moiseyenko, S. Tjulandin, et al.Phase III study of docetaxel and cisplatin plus fluorouracil compared with cisplatin and fluorouracil as first-line therapy for advanced gastric cancer: a report of the V325 Study GroupJ. Clin. Oncol., 24 (2006), pp. 4991-4997

      View in ScopusGoogle Scholar

    • [6]

    • K. ShitaraChemotherapy for advanced gastric cancer: future perspective in JapanGastric Cancer, 20 (2017), pp. 102-110

       View PDF 

      This article is free to access.

      CrossRefView in ScopusGoogle Scholar

    • [7]

    • M. Pavel, K. Öberg, M. Falconi, et al.Gastroenteropancreatic neuroendocrine neoplasms: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-upAnn. Oncol., 31 (2020), pp. 844-860View PDFView articleView in ScopusGoogle Scholar

    • [8]

    • L. ShenAnticancer drug R&D of gastrointestinal cancer in China: Current landscape and challengesInnovation (Camb), 3 (2022), Article 100249View PDFView articleView in ScopusGoogle Scholar

    • [9]

    • S. Tsukita, M. Furuse, M. ItohMultifunctional strands in tight junctionsNat. Rev. Mol. Cell Biol., 2 (2001), pp. 285-293

      View in ScopusGoogle Scholar

    • [10]

    • D. Günzel, A.S.L. YuClaudins and the modulation of tight junction permeabilityPhysiol. Rev., 93 (2013), pp. 525-569

       View PDF 

      This article is free to access.

      CrossRefView in ScopusGoogle Scholar

    • [11]

    • T. Otani, M. FuruseTight Junction Structure and Function RevisitedTrends Cell Biol, 30 (2020), pp. 805-817View PDFView articleView in ScopusGoogle Scholar

    • [12]

    • S. Tabariès, P.M. SiegelThe role of claudins in cancer metastasisOncogene, 36 (2017), pp. 1176-1190

      View article 

      CrossRefView in ScopusGoogle Scholar

    • [13]

    • I. Hashimoto, T. OshimaClaudins and gastric cancer: An overviewCancers (Basel), 14 (2022), p. 290

       View PDF 

      CrossRefView in ScopusGoogle Scholar

    • [14]

    • U. Sahin, Ö. Türeci, G. Manikhas, et al.FAST: a randomised phase II study of zolbetuximab (IMAB362) plus EOX versus EOX alone for first-line treatment of advanced CLDN18.2-positive gastric and gastro-oesophageal adenocarcinomaAnn. Oncol., 32 (2021), pp. 609-619View PDFView articleView in ScopusGoogle Scholar

    • [15]

    • J. Gong, N. Li, W. Guo, et al., American Society of Clinical Oncology Annual Meeting, June 01-02, 2022, Chicago IL, USA, pp. 4062-4062Google Scholar

    • [16]

    • C. Qi, J. Gong, J. Li, et al.Claudin18.2-specific CAR T cells in gastrointestinal cancers: phase 1 trial interim resultsNat. Med., 28 (2022), pp. 1189-1198

       View PDF 

      This article is free to access.

      CrossRefView in ScopusGoogle Scholar

    • [17]

    • L. Fan, X. Chong, M. Zhao, et al.Ultrasensitive gastric cancer circulating tumor cellular CLDN18.2 RNA detection based on a molecular beaconAnal. Chem., 93 (2021), pp. 665-670

       View PDF 

      This article is free to access.

      CrossRefView in ScopusGoogle Scholar

    • [18]

    • T. Gu, H. ShiRelationship of 18F-FDG PET/CT parameters and CLDN 18.2 expression status in gastric cancerJ. Nucl. Med., 63 (2022)3028–3028Google Scholar

    • [19]

    • G. Hu, W. Zhu, Y. Liu, et al.Development and comparison of three 89Zr-labeled anti-CLDN18.2 antibodies to noninvasively evaluate CLDN18.2 expression in gastric cancer: a preclinical studyEur. J. Nucl. Med. Mol. Imaging., 49 (2022), pp. 2634-2644

      View article 

      CrossRefView in ScopusGoogle Scholar

    • [20]

    • G. Hu, W. Zhu, Y. Liu, et al.Study of 89Zr-labeled recombinant antibody VHH-Fc for noninvasive evaluation of CLDN18.2 expression in gastric cancerJ. Nucl. Med., 63 (2022)2525–2525Google Scholar

    • [21]

    • C. Zhao, Z. Rong, J. Ding, et al.Targeting Claudin 18.2 using a highly specific antibody enables cancer diagnosis and guided surgeryMol. Pharm., 19 (2022), pp. 3530-3541

      View article 

      CrossRefView in ScopusGoogle Scholar

    • [22]

    • P.K.E. Börjesson, Y.W.S. Jauw, R. de Bree, et al.Radiation dosimetry of 89Zr-labeled chimeric monoclonal antibody U36 as used for immuno-PET in head and neck cancer patientsJ. Nucl. Med., 50 (2009), pp. 1828-1836

       View PDF 

      CrossRefView in ScopusGoogle Scholar

    • [23]

    • R. Laforest, S.E. Lapi, R. Oyama, et al.[89Zr]Trastuzumab: Evaluation of radiation dosimetry, safety, and optimal imaging parameters in women with HER2-Positive breast cancerMol. Imaging Biol., 18 (2016), pp. 952-959

      View article 

      CrossRefView in ScopusGoogle Scholar

    • [24]

    • J.A. O’Donoghue, J.S. Lewis, N. Pandit-Taskar, et al.Pharmacokinetics, biodistribution, and radiation dosimetry for 89Zr-Trastuzumab in patients with esophagogastric cancerJ. Nucl. Med., 59 (2018), pp. 161-166

       View PDF 

      CrossRefView in ScopusGoogle Scholar

    • [25]

    • C.G. England, E.B. Ehlerding, R. Hernandez, et al.Preclinical pharmacokinetics and biodistribution studies of 89Zr-Labeled pembrolizumabJ. Nucl. Med., 58 (2017), pp. 162-168

       View PDF 

      CrossRefView in ScopusGoogle Scholar

    • [26]

    • N.B. Sobol, J.A. Korsen, A. Younes, K.J. Edwards, et al.Immuno-PET imaging of pancreatic tumors with 89Zr-Labeled gold nanoparticle-antibody conjugatesMol. Imaging. Biol., 23 (2021), pp. 84-94

      View article 

      CrossRefView in ScopusGoogle Scholar

    • [27]

    • D. Vivier, S.K. Sharma, P. Adumeau, et al.The impact of FcγRI binding on immuno-PETJ. Nucl. Med., 60 (2019), pp. 1174-1182

       View PDF 

      CrossRefView in ScopusGoogle Scholar

    • [28]

    • P. Adumeau, R. Raavé, M. Boswinkel, et al.Site-specific, platform-based conjugation strategy for the synthesis of dual-labeled immunoconjugates for bimodal PET/NIRF imaging of HER2-positive tumorsBioconjug. Chem., 33 (2022), pp. 530-540View articleCrossRefView in ScopusGoogle Scholar

    • [29]

    • L.E. Lamberts, S.P. Williams, A.G.T. Terwisscha van Scheltinga, et al.Antibody positron emission tomography imaging in anticancer drug developmentJ. Clin. Oncol., 33 (2015), pp. 1491-1504

      View in ScopusGoogle Scholar

    • [30]

    • J. Zhang, R. Dong, L. ShenEvaluation and reflection on Claudin 18.2 targeting therapy in advanced gastric cancerChin. J. Cancer Res., 32 (2020), pp. 263-270

      View article 

      CrossRefGoogle Scholar


Nature Medicine | 缓解率提高3倍!创新联合疗法策略对多种癌症治疗有启示


《Nature Medicine》近期发表一项由哈佛医学院麻省总医院(MGH)癌症中心和Broad研究所(The Broad Institute)团队合作的临床研究成果,研究表明,对于BRAFV600E突变型结直肠癌患者BRAF/MAPK靶向治疗联合免疫治疗可以为患者带来持久缓解,缓解率是仅靶向治疗的3倍,中位无进展生存期(PFS)也优于当前治疗选择。



While BRAF inhibitor combinations with EGFR and/or MEK inhibitors have improved clinical efficacy in BRAFV600E colorectal cancer (CRC), response rates remain low and lack durability. Preclinical data suggest that BRAF/MAPK pathway inhibition may augment the tumor immune response. We performed a proof-of-concept single-arm phase 2 clinical trial of combined PD-1, BRAF and MEK inhibition with sparatlizumab (PDR001), dabrafenib and trametinib in 37 patients with BRAFV600E CRC. The primary end point was overall response rate, and the secondary end points were progression-free survival, disease control rate, duration of response and overall survival. The study met its primary end point with a confirmed response rate (24.3% in all patients; 25% in microsatellite stable patients) and durability that were favorable relative to historical controls of BRAF-targeted combinations alone. Single-cell RNA sequencing of 23 paired pretreatment and day 15 on-treatment tumor biopsies revealed greater induction of tumor cell-intrinsic immune programs and more complete MAPK inhibition in patients with better clinical outcome. Immune program induction in matched patient-derived organoids correlated with the degree of MAPK inhibition. These data suggest a potential tumor cell-intrinsic mechanism of cooperativity between MAPK inhibition and immune response, warranting further clinical evaluation of optimized targeted and immune combinations in CRC. registration: NCT03668431.


BRAFV600E mutations occur in ~10% of colorectal cancer (CRC), driving constitutive activation of MAPK signaling. Patients with BRAFV600E CRC have unfavorable prognosis and respond poorly to standard therapies, with a median overall survival (OS) half that of BRAF wild-type CRC1,2. While BRAF inhibitors (BRAFis), including vemurafenib and dabrafenib, are highly effective in BRAFV600E melanoma (~60–80% response rate)3,4, the response rate of BRAFi monotherapy in BRAFV600E CRC is only 0–5%5,6. Previous work by our group and others identified robust adaptive feedback networks in CRC leading to rapid reactivation of MAPK signaling following BRAF inhibition, including EGFR, which acts as the dominant driver in many cases7,8,9. These data led to clinical trials of BRAFi-based therapeutic combinations designed to mitigate MAPK feedback reactivation and produce sustained MAPK suppression, yielding increased response rates in BRAFV600E CRC. Specifically, combinations of BRAFi and EGFR inhibitor (EGFRi), BRAFi and MEK inhibitor (MEKi), and BRAFi, EGFRi and MEKi have been explored clinically10,11,12. Recently, the Food and Drug Administration approved the combination of the BRAFi encorafenib plus the anti-EGFR antibody cetuximab in BRAFV600E CRC10,12. However, confirmed objective response rates (cORRs) to this regimen are only 20% and clinical benefit is not durable, with a median progression-free survival (PFS) of only 4.3 months. Thus, new effective therapies for this disease are critically needed.

Immune checkpoint blockade (ICB), particularly agents that block the programmed death (PD)-1 pathway, has revolutionized the treatment of many cancers with the potential for long-term durable responses13,14. CRC has generally responded poorly to ICB, with the exception of the ~4% of metastatic CRC with microsatellite instability (MSI)/mismatch repair deficiency, in which response rates are ~40%, likely due to increased neoantigen load15. Conversely, response rates in metastatic microsatellite stable (MSS) CRC are near 0%16. Thus, approaches to increase the immune responsiveness of MSS CRC represent a key unmet clinical need.

Interestingly, ~15–20% of BRAFV600E metastatic CRC harbors MSI17, and these patients also showed better and more durable responses to BRAF/MAPK pathway inhibition than patients who are MSS in prior clinical trials of BRAF/MEK/EGFR inhibition, despite receiving targeted BRAF-directed therapy only. Indeed, approximately one third of patients with BRAFV600E MSI exhibited durable response and/or disease control lasting >1 year in our previous clinical trial12. Conversely, no patients who are MSS remained on study >1 year. Moreover, the lone patient achieving complete response (CR) in our initial study evaluating combined BRAF/MEK inhibition with dabrafenib and trametinib (DT)—which produced a cORR of 7% (12% unconfirmed ORR)—also had MSI and remained in an ongoing durable CR for >5 years11. Given that durable responses were restricted to patients with MSI, these data suggest the possibility that BRAF pathway inhibition may enhance immune response in BRAFV600E CRC.

Preclinical models suggest that combining MAPK inhibition and ICB could enhance antitumor efficacy in BRAF and KRAS mutant cancers18,19,20. Several potential mechanisms of cooperativity have been proposed, including possible immune priming of the tumor microenvironment by a direct effect of BRAFi and/or MEKi on nontumor cells, such as antigen-specific and activated CD8 T cells and expanded memory T cells and T cell clonotypes. The potential for tumor-intrinsic effects by MAPK inhibition contributing to the immune response has also been proposed+19. However, a definitive mechanism for this potential cooperativity has not been established. Moreover, clinical trials in BRAFV600E melanoma demonstrated promising efficacy and long-lasting antitumor responses with the combined inhibition of BRAF/MEK and PD-1/PD-L1 pathways21,22,23,24.

Based on these observations, we investigated the potential cooperativity of BRAF/MAPK pathway inhibition and ICB in BRAFV600E CRC. We studied paired biopsies from previous clinical trials of patients with BRAFV600E CRC treated with BRAF-targeted combinations and preclinical immune-competent mouse models of BRAFV600E CRC. We also performed the first clinical trial, to our knowledge, of BRAF-targeted therapy combined with ICB specifically in patients with BRAFV600E CRC, evaluating the efficacy of combined BRAF, MEKi and PD-1 inhibition. All patients underwent paired pretreatment and day 15 on-treatment tumor biopsies, which were evaluated by single-cell RNA sequencing (scRNAseq) to elucidate potential mechanisms of cooperativity.


MAPK inhibition enhances immune response in BRAFV600E CRC

Our previous clinical trials with BRAF-targeted therapy combinations in BRAFV600E CRC suggested potential links between BRAF pathway inhibition and the immune response12, including durable benefit lasting >1 year in roughly one third of patients with BRAFV600E CRC also harboring MSI. To investigate this potential cooperativity, we analyzed bulk RNAseq data from 71 patients from our earlier clinical study of combined BRAF/EGFRi ± MEKi in patients with BRAFV600E CRC, including 45 paired patient biopsies (pretreatment and day 15 on treatment) and 26 separate biopsies from baseline25. Notably, RNAseq of baseline tumor biopsies revealed significantly higher T cell signatures (indicative of increased T cell levels) in responders versus nonresponders (Fig. 1a). Baseline levels of T cell and cytotoxic T cell signatures also correlated with the best percentage change in target lesion size from baseline (Extended Data Fig. 1a). Furthermore, across all patients, increases in T cell, cytotoxic T cell and other immune signatures were noted after 15 days of treatment relative to the paired baseline biopsy, suggesting increased T cell and immune infiltration in tumors following BRAF pathway inhibition (Fig. 1b). These data support a potential interaction between BRAF/MAPK inhibition and the immune response in BRAFV600E CRC.

Fig. 1: MAPK pathway inhibition enhances immune response in BRAFV600E CRC.

a, Tumor baseline T cell signature expression levels in confirmed responders (R) (n = 8) and nonresponders (NR) (n = 39) from a clinical trial of dabrafenib/trametinib/panitumumab in patients with BRAFV600E CRC (DTP treatment arm only, two-tailed Wilcoxon rank sum tests). b, Levels of immune signatures (T cell, cytotoxic T cell and phagocytic) in 45 paired day 1 and day 15 biopsies from a clinical trial of dabrafenib/trametinib/panitumumab (all DTP, DP and TP treatment arms, two-tailed Wilcoxon rank sum tests). a,b, The box plots show the median, first and third quartiles (Q1 and Q3) (25th and 75th percentiles) of the data. The upper and lower whiskers extend to the minimum and maximum values no further than 1.5× the interquartile range, respectively; outliers are plotted individually. c, Tumor volume of C57BL/6 mice bearing ABPS tumors treated with vehicle/immunoglobulin G control (n = 11), DT (n = 12), PD-1 (n = 11) and DTP (n = 12; two-tailed Wilcoxon rank sum test, data are presented as mean values ± s.e.m.). d, Representative images of CD3CD8 T cells in ABPS tumors. e, Percentage of CD3CD8 T cells in ABPS tumors across the groups control (n = 10), DT (n = 10), PD-1 (n = 11) and DTP (n = 9; two-tailed Wilcoxon rank sum test, error bars represent s.e.m.). Ctrl, control; DTP, dabrafenib/trametinib/panitumumab; DP, dabrafenib/panitumumab; TP, trametinib/panitumumab.++++

Source data

Full size image

To model this potential cooperativity, we generated a syngeneic BRAFV600E CRC mouse tumor model from C57BL/6 colon organoids with knockout of Adenomatous polyposis coli (APC), TP53 and SMAD4 and expressing BRAFV600E (APC, BRAFV600E, TP53, SMAD4 (ABPS) cells). Inhibition of MAPK signaling and sensitivity to combined BRAF/MEK inhibition with DT in ABPS cells were confirmed in vitro (Extended Data Fig. 1b,c). To assess immune effects of BRAF/MEK inhibition and the potential cooperativity with ICB, we implanted ABPS tumor cells subcutaneously into immune-competent C57BL/6 mice and treated them with vehicle, DT, anti-PD-1 antibody, or the combination. Interestingly, while BRAF/MEKi or anti-PD-1 antibody alone had minimal effect on tumor volume compared with vehicle, combined BRAF/MEK/PD-1 inhibition produced a more substantial and sustained reduction in tumor growth (Fig. 1c and Extended Data Fig. 1d). Moreover, in tumors harvested after 10 days of treatment, we detected a significant increase in the percentage of CD3CD8 T cells in tumors treated with BRAF/MEKi compared with vehicle control (Fig. 1d,e), suggesting that BRAF pathway inhibition alone can lead to increased T cell infiltration, similar to our observations in patient biopsies (Fig. 1b). Notably, treatment with PD-1 antibody alone did not lead to a significant increase in CD3CD8 T cells, although a significant increase was observed with combined BRAF/MEK/PD-1 inhibition (Fig. 1d,e). Collectively, these preclinical and translational data suggest that BRAF pathway inhibition may augment the immune response in BRAF++++V600E CRC.

Clinical efficacy

Based on these data, we initiated the first clinical trial, to our knowledge, combining targeted BRAF pathway inhibition with ICB in BRAFV600E CRC. Patients with BRAFV600E CRC were treated with the BRAFi dabrafenib, the MEKi trametinib, and the anti-PD-1 antibody spartalizumab (PDR001). While DT is not the optimal BRAF-targeted strategy for BRAFV600E CRC—producing a 7% cORR and a median PFS of 3.5 months in a previous clinical trial11—established dosing and safety data for this triple regimen from patients with melanoma allowed for more rapid initiation of this proof-of-concept clinical trial, and we reasoned that evidence of clinical cooperativity observed would provide rationale for future evaluation of ICB combinations with more effective BRAFi combinations, including anti-EGFR antibody combinations.

As of the data cutoff, 37 of a planned 40 patients with BRAFV600E CRC have been enrolled. All 32 slots for patients who are MSS accrued as well as 5 of 8 slots reserved for patients with MSI. Five patients had previous therapy with either BRAFis and/or immune checkpoint inhibitors (Extended Data Fig. 2). Median age was 63 (range 35–87), 20 (54.1%) were women (Extended Data Table 1), and median follow-up was 995 days (range 245–1,324). Overall, the regimen was well tolerated, with rash, fever and diarrhea as the most common adverse events (AEs) (Extended Data Table 2). The primary end point was ORR, and secondary end points were PFS, disease control rate (DCR), duration of response and OS. Among all 37 patients, 9 achieved a confirmed response (cORR of 24.3%; 95% confidence interval (CI) 11.9–41.2%), with 1 additional patient achieving an unconfirmed response, and the DCR was 70.3% (95% CI 53–84.1%) (Fig. 2a and Extended Data Fig. 3a), which compares favorably with the historical 7% cORR (95% CI 1.5–19.1%) of dabrafenib plus trametinib alone in BRAFV600E CRC. Two patients achieved a CR. Median PFS was 4.3 months (95% CI 3.7–7.3 months) (Extended Data Fig. 3b). Median OS was 13.6 months (95% CI 8.2–16.5 months) (Extended Data Fig. 3c). Median duration of time on treatment was 7.4 months (95% CI 4.2–7.9 months) (Fig. 2b and Extended Data Fig. 3d). Among the 32 patients without previous BRAF-directed therapy or ICB, cORR was 28.1% and DCR was 71.9% (Fig. 2c). ORR, PFS and OS of patients with previous BRAF-directed therapy or ICB are shown in Extended Data Fig. 3g.

Fig. 2: Clinical efficacy of dabrafenib, trametinib and PDR001 in patients with BRAFV600E CRC.

a, Best percentage change in the sum of the longest tumor diameters from baseline according to RECIST v.1.1 for patients in the total intention-to-treat cohort. ‘IO’, ‘MSI-H’, ‘SD’, and ‘PD’ denotes immunotherapy, microsatellite instability-high, stable disease, and progressive disease, respectively. b, Swimmer plot presenting the duration of treatment exposure and efficacy assessments in all patients. c, Best percentage change in the sum of the longest tumor diameters from baseline according to RECIST v.1.1 among patients without prior receipt of a BRAFi and/or immunotherapy in the intention-to-treat cohort. d, Best percentage change in the sum of the longest tumor diameters from baseline according to RECIST v.1.1 among patients without prior receipt of a BRAFi and/or immunotherapy and with microsatellite stability in the intention-to-treat cohort.

Full size image

Since patients with MSI CRC may respond to ICB alone, the critical focus of this study was the 28 patients without prior BRAF-directed therapy or ICB who were also MSS and would, therefore, be predicted to have negligible response rates to ICB alone. In these patients with MSS BRAFV600E CRC, cORR was 25% (95% CI 10.7–44.9%) and DCR was 75% (95% CI 55.1–89.3%) (Fig. 2d), again comparing favorably with historical controls. Median PFS was 5 months (95% CI 3.7–7.4 months), with five patients (18%) remaining on therapy for over a year (Extended Data Fig. 3e). This is in stark contrast to the earlier study of DT alone, which demonstrated a median PFS of only 3.5 months with no patients who were MSS staying on therapy greater than 1 year. One patient achieved a partial response (−100% by Response Evaluation Criteria in Solid Tumors (RECISTS) v.1.1) persisting for 2.5 years.

Analysis of baseline biopsies revealed that tumor mutational burden, BM1/BM2 transcriptional subtypes and consensus molecular subtypes (CMSs)—which were previously reported to affect prognosis and response to therapy25,26,27—did not correlate with clinical outcome (Extended Data Fig. 4 and Supplementary Table 1). Furthermore, no differences in efficacy were observed based on left versus right sidedness of the primary tumor (Extended Data Fig. 3f). Overall, these data suggest promising clinical efficacy and evidence of cooperativity between BRAF/MAPK pathway inhibition and ICB.

Tumor-intrinsic immune response and MAPK inhibition

To understand the potential interaction of BRAF/MAPK pathway inhibition and the tumor immune response, all patients underwent paired pretreatment and day 15 on-treatment biopsies of the same tumor lesion. Fresh tumor biopsies were analyzed by scRNAseq with evaluable data obtained from both paired biopsies in 23 patients. A total of 419,551 single cells passed quality control (QC) across all specimens and stromal, immune and tumor epithelial cell populations (Fig. 3a). Comparing changes in the abundance of individual cell populations in pre- versus on-treatment biopsies, we observed a significant decrease of tumor epithelial cells as well as an increase of CD45 immune cells (defined by cell-type clustering), T cells and CD8 T cells after treatment in patients with PFS > 6 months (n = 11) but not in patients with PFS < 6 months (n = 12) (Fig. 3b).++

Fig. 3: Greater tumor cell-intrinsic immune program induction and MAPK pathway inhibition in patients with better outcome.

at-distributed stochastic neighbor embedding plot of 419,551 cells color coded for the indicated cell type. ILC, innate lymphoid cell; NK, natural killer cell; Tgd, gamma-delta T cell; Tprolif, proliferating T cell. b, Percentage of indicated cell types (on- versus pretreatment biopsies) in patients with PFS > 6 months (n = 11) and patients with PFS < 6 months (n = 12; two-tailed Wilcoxon rank sum tests). c, Volcano plots showing upregulated and downregulated DEGs (on treatment versus pretreatment) in tumor epithelial cells of patients with PFS > 6 months and PFS < 6 months. Black dots on the volcano plots indicate adjusted P < 0.05 (two-tailed Wilcoxon rank sum test) and log2FC ≥ 1. Significant DEGs involved in antigen processing and presentation (gold), the IFN pathway (red) and chemokine activity (blue) are labeled. d, log2FC (on treatment versus pretreatment) of expression of ISGs involved in indicated immune pathways in tumor epithelial cells of patients with PFS > 6 months and PFS < 6 months. e, Enriched immune-related Gene Ontology terms in tumor epithelial cells of patients with PFS > 6 months and PFS < 6 months. Dotted line indicates false discovery rate (FDR) = 0.05. ‘Ag’ denotes antigen. fh, Changes of epithelial ISG score (on treatment versus pretreatment) (f), pEpiTd19 ISG score (on treatment versus pretreatment) (g) and MAPK score (on treatment versus pretreatment) (h) in tumor epithelial cells of patients with PFS > 6 months (n = 9) and PFS < 6 months (n = 10; Wilcoxon signed rank test).

Full size image

To understand the effects of BRAF/MAPK inhibition specifically in tumor cells, we characterized gene expression changes after treatment in the tumor epithelial compartment to identify potential tumor cell-intrinsic mechanisms underlying patient response. Evaluating differentially expressed genes (DEGs) in tumor cells at day 15 versus pretreatment revealed striking increases in the expression of immune-related genes in patients with PFS > 6 months that were not observed in patients with PFS < 6 months, including genes involved in interferon (IFN) response (for example, STAT1IRF1IFITM1IFITM3BST2 and TRIM31), antigen processing and presentation (for example, B2MCD74PSMB10TAP1HLA-AHLA-CHLA-FHLA-DRB1 and HLA-DPA1) and chemokine activity (for example, CXCL9CXCL10 and CXCL11) (Fig. 3c,d; all DEGs are listed in Supplementary Table 2), suggesting global upregulation of IFN-stimulated transcriptional programs and antigen processing and presentation pathways, which was also observed by gene set enrichment analyses (Fig. 3e and Supplementary Table 3). In contrast, fewer immune-related gene sets were enriched in tumor epithelial cells in patients with PFS < 6 months (Fig. 3e). DEGs in patients with PFS > 6 months that mapped to antigen processing and presentation, response to IFN-γ, response to type I IFN, and chemokine activity programs were used to create a score of epithelial interferon-stimulated genes (ISGs) (Fig. 3d and Supplementary Table 4). This score was significantly elevated on treatment in patients with PFS > 6 months but not PFS < 6 months in a patient-level analysis (in addition to the gene-level analysis shown before) (Fig. 3f). We also assessed a score for a human CRC malignant epithelial-specific ISG program we recently derived in an independent scRNAseq effort and that was associated with activated and chronically stimulated T cells28. Again, patients with PFS > 6 months showed significantly increased scores (pEpiTd19 ISG) at day 15, whereas patients with PFS < 6 months did not (Fig. 3g and Extended Data Fig. 5).

We next examined the degree of MAPK pathway inhibition after treatment in tumor cells using an MAPK gene expression signature score based on changes in the MAPK-regulated transcripts (DUSP6ETV4ETV5 and SPRY4) in tumor epithelial cells. Interestingly, the MAPK score was significantly reduced after treatment in tumor epithelial cells of patients with PFS > 6 months but not in patients with PFS < 6 months (Fig. 3h). Thus, scRNAseq analysis of tumor cell-intrinsic gene expression changes following treatment revealed that patients with longer PFS showed greater induction of tumor-intrinsic immune programs as well as greater MAPK pathway inhibition.

Enhanced immune response driven by optimized MAPK inhibition

We therefore hypothesized that the degree of MAPK inhibition achieved in tumor cells might be directly related to the degree of tumor-intrinsic induction of immune gene expression. Accordingly, we evaluated the effects of BRAF/MAPK pathway inhibition alone in patients with BRAFV600E CRC treated on previous studies with BRAF targeted therapy only. We analyzed RNAseq data from 45 paired pretreatment and on-treatment (day 15) biopsies from patients with BRAFV600E CRC from a previous clinical trial of BRAF/MEK/EGFR inhibition12,25. Notably, the induction of several immune signatures at day 15 correlated with the degree of MAPK pathway inhibition (Extended Data Fig. 6a). These included T cell and cytotoxic signatures, immune checkpoint signaling and phagocytic signatures (Extended Data Fig. 6b). Furthermore, we also found correlation between the degree of MAPK inhibition and immune signature induction in on-treatment versus pretreatment biopsies, including T cell, immune checkpoint and innate immune response signatures (Extended Data Fig. 6c). The fact that a greater increase in immune signatures was observed in patients with BRAFV600E CRC with a greater degree of MAPK pathway inhibition supports the hypothesis that the degree of MAPK inhibition in tumor cells may drive immune gene induction and aspects of the tumor immune response.

To evaluate the potential relationship between MAPK inhibition and immune program induction specifically within tumor cells, we utilized patient-derived organoid models successfully generated from baseline tumor biopsies from 10 patients, including 5 with PFS > 6 months and 5 with PFS < 6 months. Organoids were treated with DT and gene expression was measured by quantitative polymerase chain reaction (qPCR). Organoids derived from patients with PFS > 6 months showed greater increases in the expression of genes involved in IFN response (IFIT1IFIT2IFIT3 and IRF1) and chemokine activity (CXCL9CXCL10 and CXCL11) compared with organoids derived from patients with PFS < 6 months (Fig. 4a, left). Importantly, organoids from patients with PFS < 6 months also showed a lesser degree of MAPK pathway inhibition (calculated by the average DUSP6ETV4ETV5 and SPRY4 log2FC) after treatment with DT (Fig. 4a, left). The degree of ISG induction was significantly correlated with the degree of MAPK inhibition (Extended Data Fig. 7a), suggesting that insufficient MAPK pathway inhibition could potentially explain the differences in ISG upregulation. Notably, the varying degree of ISG induction and MAPK inhibition by DT in both groups of organoids (PFS > 6 months or PFS < 6 months) mirrored the differences observed from scRNAseq analysis of the tumor epithelial compartments in patients from the same groups. Similarly, global transcriptomic profiling of organoids treated with DT by RNAseq showed significant induction of more immune gene sets in organoids derived from patients with PFS > 6 months compared with organoids from patients with PFS < 6 months (Fig. 4b). Likewise, treatment of organoid models with dabrafenib and the anti-EGFR antibody panitumumab also led to a similar induction of ISGs (Extended Data Fig. 7b). Importantly, these data confirm that MAPK pathway inhibition can drive induction of immune gene expression in a tumor cell-intrinsic manner that does not depend on stimuli from the tumor immune microenvironment since organoid cultures contain tumor cells only and do not contain immune or stromal cells from the tumor microenvironment. These data also provide further correlative evidence that the degree of MAPK suppression may be directly related to the degree of immune gene induction.

Fig. 4: Enhanced tumor cell-intrinsic immune program induction is driven by optimized MAPK inhibition.

a, The log2FC of gene expression (measured by qPCR) of indicated ISGs and the average of log2FC of MAPK-regulated transcripts in patient-derived organoids treated with DT or DE for 72 h. Organoids are arranged based on the PFS data of patients (from left to right: longest to shortest PFS). b, GO term enrichment analysis of upregulated DEGs (log2FC ≥ 1, P < 0.05, Fisher exact test) in organoids treated with 72-h DT or DE versus control. Gene expression was measured by bulk RNAseq. Organoids are arranged based on patient PFS data. c, Delta of ISG scores (left) and MAPK scores (right; 72-h treatment versus control) in DT-treated (n = 10) and DE-treated (n = 10) organoids (two-tailed Wilcoxon rank sum test). d, Delta of ISG scores (72-h treatment versus control) in organoids derived from patients with PFS > 6 months (n = 5) and PFS < 6 months (n = 5; two-tailed Wilcoxon rank sum test). In c and d, the box plots show the median, Q1 and Q3 (25th and 75th percentiles) of the data; the upper and lower whiskers extend to the minimum and maximum values. e, Pearson correlation (two sided) of ISG score delta and MAPK score delta in all DT- and DE-treated organoids.

Full size image

To test whether more optimal and complete inhibition of the BRAF/MAPK pathway might enhance the degree of immune gene induction across all tumor models, including those from patients with PFS < 6 months, we utilized a combination of BRAFi and ERKi. Our earlier work showed that ERK inhibitors are able to achieve more complete MAPK pathway inhibition in combination with BRAFis compared with MEK inhibitors29. Thus, we treated the same collection of organoids with dabrafenib and the ERK inhibitor ERAS007 (DE). The concentrations of DT and DE chosen were based on inhibitory concentration (IC)90 values demonstrating equivalent efficacy to ensure an appropriate comparison (Extended Data Fig. 7c,d). Interestingly, when treated with DE, all organoids showed robust MAPK inhibition, regardless of the clinical outcome of the patient from whom the organoid was derived (Fig. 4a, right). BRAF/ERK inhibition also strikingly induced the expression of immune response genes to a comparable extent in all organoids (Fig. 4a, right), suggesting that higher ISG induction could be achieved by more complete MAPK inhibition, especially in organoids with lesser response to BRAF/MEKi. Notably, even with a higher concentration of DT, we still did not observe stronger induction of immune programs in organoids, including in representative models from PFS < 6 months (Extended Data Fig. 7e). Additionally, BRAF/ERK inhibitor (ERKi) treatment universally induced strong enrichment of immune-related Gene Onotology (GO) terms in organoids derived from patients with PFS > 6 months and PFS < 6 months (Fig. 4b), in contrast to observations with BRAF/MEKi.

Overall, BRAF/ERKi induced a significantly greater degree of ISGs upregulation and MAPK inhibition than BRAF/MEKi in all organoids, as reflected by a greater ISG score and MAPK score change with treatment versus control (Fig. 4c). Furthermore, whereas BRAF/MEKi led to a significantly greater ISG score increase in organoids derived from patients with PFS > 6 months versus patients with PFS < 6 months, BRAF/ERKi led to a comparable degree of ISG score induction in both groups of organoids that was not significantly different. Notably, the levels of ISG score induction in both groups following BRAF/ERKi were equal to or greater than the degree of induction seen with BRAF/MEKi in patients with PFS > 6 months (Fig. 4d). ISG score delta was strongly and significantly anticorrelated with MAPK score delta (Fig. 4e). Together, these data demonstrate that (1) effective MAPK inhibition alone can induce tumor cell-intrinsic immune gene expression and that (2) improving the degree of MAPK inhibition achieved can increase the degree of tumor-intrinsic immune gene expression in all tumor models to the same levels observed in patients with PFS > 6 months with BRAF/MEKi.

Overall, these data provide evidence for a tumor cell-intrinsic role of BRAF/MAPK pathway inhibition in promoting tumor immune response. Given the relationship between the degree of MAPK inhibition and induction of immune program gene expression in tumor cells, these findings also raise the possibility that improved MAPK inhibition through a more optimal BRAF-targeted combination could drive greater immune cooperativity and greater clinical efficacy in combination with ICB.


We present preclinical, clinical and translational evidence of cooperativity between BRAF/MAPK pathway inhibition and the immune response in BRAFV600E CRC. Importantly, even in patients with MSS BRAFV600E CRC, in whom a negligible response rate to ICB is expected, the combination of PD-1, BRAF and MEK inhibition yielded more than a threefold increase in cORR (25%, 95% CI 10.7–44.9%) relative to historical controls of combined BRAF/MEK alone (7%, 95% CI 1.5–19.1%) in patients without prior BRAFi. This also compares favorably with the 20% cORR of encorafenib plus cetuximab, the current Food and Drug Administration-approved standard for BRAFV600E CRC. We also observed evidence of increased durability with a median PFS of 5 months (versus 3.5 months with BRAF/MEK alone), with 57% of patients remaining on treatment for >6 months and 18% for >1 year.

The primary focus of the study was efficacy in the MSS BRAFV600E CRC population since these patients have a near-zero response to ICB and thus, the greatest clinical need. However, eight slots were reserved for patients with MSI BRAFV600E in an effort to assess whether combined BRAF pathway targeting might enhance the basal responsiveness of MSI tumors to ICB. While durable responses were observed in two of five patients with MSI enrolled, this 40% ORR does not suggest clear evidence of enhanced clinical benefit, and enrollment to remaining MSI slots is ongoing.

In what we believe is one of the first clinical trials to incorporate systematic scRNAseq analysis of paired pre- and on-treatment tumor biopsies from all patients, we identified a potential mechanism underlying the cooperativity observed between BRAF/MAPK inhibition and immune response. We find evidence of tumor cell-intrinsic induction of key immune programs (types I and II IFN response, antigen presenting genes and T cell recruiting chemokines) triggered by MAPK pathway inhibition. These gene expression changes are similar to a gene expression program recently discovered in malignant cells from immunologically active MSI CRC and associated with activated and chronically stimulated T cells28. Interestingly, induction of T cell recruiting chemokines, such as CXCL9/10/11 and certain ISGs, has also been observed in preclinical CRC mouse models following KRASG12C inhibition30, suggesting that this tumor cell-intrinsic transcriptional response may be a key mechanism more broadly linked to inhibition of the RAS–MAPK pathway. Our scRNAseq data revealed that upregulation of immune programs in tumor cells was significantly greater in patients with prolonged PFS, suggesting that this effect may be a critical mechanism mediating therapeutic benefit.

By performing parallel studies with matched patient-derived organoids generated from the same patients, we confirmed that induction of immune genes by BRAF/MAPK inhibition is tumor cell intrinsic and does not depend on cells in the tumor microenvironment. Indeed, induction of these immune programs was observed even in organoid cultures, which consist of pure tumor cell populations. These data provide important mechanistic insight, as the potential cooperativity of MAPK pathway inhibition and the immune response has previously been hypothesized to involve a direct effect of MAPK inhibition on cells in the tumor microenvironment18. One potential limitation of the current study is that it focused on the tumor cell compartment. It is still possible that MAPK inhibition within the tumor microenvironment may affect the tumor immune response, and a detailed assessment of changes in the abundance and gene expression profiles of other cell types will be undertaken through analysis of our scRNAseq data, coupled with parallel methods, in a future study. However, our data provide evidence for a robust tumor cell-intrinsic mechanism and support that BRAF/MAPK pathway activation promotes immune suppressive signals within tumor cells that can be reversed with targeted pathway inhibition. More detailed mechanistic experiments in future studies may further delineate the importance and relative contribution of this tumor cell-intrinsic mechanism to the tumor immune response.

Notably, organoids derived from patients with PFS > 6 months also retained the pattern of increased induction of immune programs upon BRAF/MEK treatment relative to organoids from patients with PFS < 6 months, as observed in our scRNAseq dataset. This observation allowed us to use these matched organoids as representative models to probe the mechanisms underlying tumor cell-intrinsic immune program induction. Interestingly, our scRNAseq data suggested that patients with PFS < 6 months failed to achieve robust MAPK inhibition in tumor cells with DT compared with patients with PFS > 6 months. This same pattern was observed in matched patient-derived organoid models, allowing us to test the hypothesis that failure to induce immune programs was a direct consequence of inadequate BRAF/MAPK inhibition and that more optimal MAPK inhibition may lead to improved induction of key immune genes, even in tumor cells from patients who did not benefit from therapy. As noted above, a limitation of the study was the use of a BRAF/MEK inhibitor combination for this proof-of-concept study, which was based on the availability of established dosing and safety data for this regimen from patients with melanoma at the time that this study was initiated; this preceded the results of the BEACON study that demonstrated the efficacy of encorafenib and cetuximab in this population. Notably, the BEACON study also showed that while the addition of an MEK inhibitor increased the overall response rate relative to the BRAF/EGFR doublet (27% versus 20%), it did not improve survival. Thus, BRAF/MEK is likely not the optimal BRAF-targeting core in this population, and these data suggest that more effective BRAF-targeting combinations could further augment the cooperativity with ICB. Remarkably, using a combination of BRAF and ERK inhibitors that led to more robust BRAF/MAPK inhibition across all organoids, we found that immune program induction could be achieved in all models, regardless of PFS. This finding has important and immediate clinical implications, suggesting that combining ICB with a more effective BRAF/MAPK-targeting core, such as BRAF/EGFR or BRAF/ERK, may further enhance the immune cooperativity and clinical benefit observed and that such strategies warrant further exploration in future clinical trials. This potential for enhanced clinical activity from targeting the BRAF pathway in combination with ICB is further supported by early data from an ongoing clinical trial adding the anti-PD-1 antibody nivolumab to the current standard of care of encorafenib plus cetuximab, yielding a 50% ORR in the first 21 patients compared with an expected 20% ORR for encorafenib plus cetuximab alone31.

Finally, it will be important to determine whether this potential mechanism of immune cooperativity is limited to BRAFV600E CRC or whether it would apply to other tumor types and to other agents targeting the MAPK pathway. For example, combinations of ICB (PD-1 or programmed death-ligand (PD-L)1) with BRAFi/MEKi in BRAFV600 melanoma have also suggested improved benefit, although each component of this combination exhibits greater activity alone in melanoma than in CRC. Notably, a large phase III study of the MEK inhibitor cobimetinib and the anti-PD-L1 antibody atezolizumab in CRC failed to achieve its primary end point, which has cast doubt on whether MAPK pathway inhibition in general has the potential to enhance the immune response. However, this trial was performed in all CRC genotypes (including CRC without MAPK-activating mutations), and data suggest that MEK inhibitors alone fail to maintain MAPK inhibition in CRC due to adaptive feedback32. Preclinical data support that induction of similar immune gene programs as observed in our study can be found in mouse models of KRASG12C CRC with KRASG12C inhibitor treatment30,33. Therefore, it is possible that the mechanism of immune cooperativity we propose in our study may be more broadly applicable to other effective RAS/BRAF/MAPK pathway inhibitors, and further studies to evaluate this possibility will be important.


Study design

This research study is a phase II clinical trial to test the safety and effectiveness of DT in combination with the anti-PD-1 antibody PDR001 in patients with metastatic CRC characterized by the BRAF V600E mutation (NCT03668431). Patients at the Massachusetts General Hospital Cancer Center and the Dana-Farber Cancer Institute were treated with spartalizumab (PDR001) 400 mg intravenous q28d (every 28 days), dabrafenib 150 mg oral administration twice a day for 28 consecutive days, and trametinib 2 mg oral administration daily for 28 consecutive days (dose safety was established in patients with melanoma). The study was conducted in accordance with the Guidelines for Good Clinical Practice and the ethical principles described in the Declaration of Helsinki, and it was approved by the local institutional review board.


Eligible patients must have histologically or cytologically confirmed metastatic CRC, have a documented BRAF V600E mutation by a CLIA-certified laboratory test and be wild type for KRAS and NRAS. Patients were required to be aged ≥18 years, have measurable disease according to RECIST v.1.1, have an Eastern Cooperative Oncology Group performance status of less than or equal to two and have adequate baseline organ function (as determined by laboratory parameters). The first patient was enrolled on 15 October 2018. The trial was amended after the first nine patients to exclude patients with prior BRAF or MEK inhibitor or immunotherapy, and this amendment was Institutional Review Board approved. Enrollment of the remaining slots reserved for patients with MSI on the trial is still ongoing at this time. Key exclusion criteria included chemotherapy or radiotherapy within 4 weeks prior to entering the study and any serious or unstable preexisting medical condition. All patients provided written informed consent before the study.

Efficacy assessments

Patients received study therapy until disease progression, unacceptable toxicity, death or discontinuation for any other reason. Safety was monitored throughout the study for all patients across cohorts via physical examinations, laboratory evaluations, vital sign and weight measurements, performance status evaluations, ocular and dermatologic examinations, concomitant medication monitoring, electrocardiograms, echocardiograms and AE monitoring (characterized and graded per Common Terminology Criteria for Adverse Events v.4.0). AEs were recorded using the standard Medical Dictionary for Regulatory Activities coding. Dose interruptions, reductions and discontinuations for all of the study drugs were monitored.

The primary end point was ORR; secondary end points were PFS, disease control rate, duration of response and OS, and the exploratory end point was scRNAseq analysis. Antitumor efficacy was assessed by CT or MRI at baseline and then, every 8 weeks until progression or death. Response determination was based on RECIST v.1.1 by the Dana-Farber/Harvard Cancer Center Tumor Metrics Core. For the subset of patients who showed a confirmed CR or PR, duration of response was defined as the time in weeks from the first documented evidence of CR or PR (the first response prior to confirmation) until the time of documented disease progression or death due to any cause, whichever was first. PFS was defined as the time in weeks between the first dose and the date of disease progression or death due to any cause. Finally, OS was defined as the time in weeks from the first dose of study drug until death due to any cause. PFS and time on treatment were summarized with Kaplan–Meier methodology using medians and 95% CIs (estimated using the Brookmeyer–Crowley method). Time on treatment was defined as the time until final treatment discontinuation. Fresh tumor biopsies were collected before dose (day 1) for scRNAseq analysis and patient-derived organoids generation as well as after dose (day 15) for scRNAseq analysis. The same tumor lesion was biopsied at baseline and at day 15. Formalin-fixed paraffin-embedded (FFPE) and flash-frozen tumor samples were collected at day 1 for genomic and molecular analyses.

Whole-exome sequencing and TMB analysis

For each biopsy, we called somatic point mutations against the patient’s matched blood normal control sample using MuTect134 for single base substitutions and Strelka235 for indels. Mutations were filtered for sequencing artifacts using the Getz Lab whole-exome analysis pipeline. We then calculated tumor mutation burden (TMB) in terms of mutation density by dividing the total number of mutations called for each biopsy by the total captured exonic territory from the TWIST Biosciences bait set.

BM1 and BM2 signatures and CMS analysis

We aligned raw RNAseq reads using STAR36 two-pass transcriptome/genome alignment and then quantified per-transcript counts using RSEM37. We then performed Bayesian nonnegative matrix factorization (BNMF)38 on the matrix of transcript counts for each sample. This expresses the vector of each sample’s transcript counts as a linear combination of vectors corresponding to transcriptional signatures common across samples, with the overall number of signatures automatically determined by the Bayesian model hierarchy. We identified one BNMF signature whose genes closely matched the gene set defining the BM1 signature26 and another BNMF signature whose genes closely matched the gene set defining the BM2 signature. We found that the sample loadings for these two signatures were nearly mutually exclusive among samples (that is, a sample with a high BM1 loading almost always had a negligible BM2 loading and vice versa), allowing us to classify the majority of samples by their BM1/BM2 status. The log2 TPM counts from RSEM were passed directly to the CMS classifier using a centroid-based predictor method, which classifies samples based on their similarity to expression clusters derived from the ground truth dataset.

Generation of ABPS and APSe cell line

Organoids were established from colon tissue of C57BL/6 mice harboring a conditional floxed Trp53 allele, infected with Cre-expressing adenovirus for Trp53 deletion, subjected to CRISPR–Cas9 knockout of Apc and Smad4 and dissociated to generate the APC, TP53, SMAD4 (APS) cell line. APS cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM)/F12 media supplemented with 10% fetal bovine serum (FBS) and 2 mM GlutaMAX (Thermo Fisher Scientific). The BRAF V600E sequence was cloned in the pMXs-Puro Retroviral Expression Vector (Cell Biolabs). Retrovirus containing the BRAF V600E sequence or empty pMXs-Puro vector was produced in HEK293 cells with packaging vector pCL-10A1 and concentrated with Retro-Concentin Retro Concentration Reagent (System Biosciences). APS cells were cultured to 50% confluence and then infected with retrovirus using 5 μg ml−1 polybrene. After 48 h of infection, 2 μg ml−1 puromycin was added to the media to select stable ABPS (contains BRAF V600E sequence) and APC, TP53, SMAD4, empty vector (APSe) cells (used as control).

Animal studies

ABPS cells were resuspended in PBS and Corning Matrigel in a 1:1 ratio and then injected (2.5 × 105 cells per injection) into the flanks of 12-week-old male C57BL/6 mice (Charles River Laboratories). When the tumor size reached 150–200 mm2, mice were randomized into four groups and treated with (1) vehicle control (0.5% hydroxypropyl-methylcellulose + 0.2% Tween 80, oral gavage) and immunoglobulin G isotype control (BioXcell, intraperitoneal injection); (2) dabrafenib (30 mg kg−1 daily, oral gavage) and trametinib (1 mg kg−1 daily, oral gavage); (3) anti-PD-1 (10 mg kg−1 twice per week, intraperitoneal injection); or (4) dabrafenib, trametinib and anti-PD-1. The mice were treated for up to 60 days, and tumor volumes were assessed twice per week and determined according to the formula length (L) × width (W)2 × π/6. Animal studies and procedures were performed in accordance with the institutional guidelines of Massachusetts General Hospital, and all experiments were conducted according to institutional animal care and use committee-approved protocols. The housing conditions for mice are 20 °C to 26 °C, a 12-h light/12-h dark cycle, and 40–60% humidity.

Immunofluorescence staining

Mouse tumors were collected and fixed in 10% formalin, embedded and sectioned (5 μm). Tumor tissue FFPE slides were then deparaffinized and rehydrated. For antigen retrieval, slides were maintained in antigen unmasking solution (Vector Laboratories; H-3300) at a subboiling temperature for 20 min using a microwave. Slides were washed three times for 5 min each in PBS supplemented with 0.1% Tween 20 and then were blocked with 5% BSA (Thermo Fisher Scientific) and 5% goat serum (Sigma-Aldrich) for 1 h at room temperature. The slides were incubated with anti-CD3 (Abcam; ab11089; 1:800 dilution) and anti-CD8 (Cell Signaling; 98941 S; 1:400 dilution) primary antibodies overnight at 4 °C. The next day, slides were washed as above and incubated with conjugated secondary antibodies (Thermo Fisher Scientific; A-11036 and A-11006) for 1 h at room temperature. The slides were then washed, stained for DAPI (Invitrogen) and mounted with SlowFade Diamond Antifade Mountant (Invitrogen). Once finished, the slides were scanned using a ZEISS Axio Scan slide scanner and analyzed using Halo software (Indica Labs). Image annotations were performed in a blinded manner. Cells stained with an intensity exceeding the settings threshold were counted as positive. The settings were set to include the full range of staining intensity (weak to strong). Halo counted the CD3CD8 cells in stromal and tumor, and data were collected as the number of positive cells divided by the total DAPI + cells in the tumor area.++

Organoid generation, culture and treatment

Patient-derived organoid generation was attempted from baseline biopsies of all patients. A total of 10 colorectal tumor organoid lines were successfully generated from tumor baseline biopsies of 10 patients enrolled in the BRAF/MEK/PD-1 inhibition trial (patients 1, 2, 4, 10, 11, 14, 16, 18, 21 and 24). Tumor biopsies were transported in ice-cold RPMI with 10% human serum and transferred into a petri dish on ice before processing. Tumor biopsies were minced and subjected to enzymatic dissociation in 4.75 ml minimum essential medium for suspension cultures (Gibco) supplemented with 250 µl Liberase for 45 min at 37 °C using a heater-shaker. Following the dissociation, tumor biopsies were centrifuged at 300g for 5 min, seeded into Matrigel in a prewarmed 24-well plate and cultured with 500 µl of basal growth media. For passaging, organoids were mechanically pipetted out of Matrigel using Corning Cell Recovery Solution (Corning), followed by a 1-h incubation at 4 °C. Organoid fragments were then centrifuged and subjected to enzymatic dissociation in Tryple E (Gibco) for 5 min at 37 °C. A 20-gauge needle was used to further disrupt the organoids mechanically. DMEM/F12 media supplemented with 10% FBS was added to the conical tube to stop the enzymatic reaction. Dissociated organoids were collected by centrifugation and seeded in Matrigel as above; 10 µM Rko-Kinase inhibitor was added to the basal growth media for organoid passaging. For drug treatment, dissociated organoids were resuspended in basal growth media supplemented with 2% Matrigel and plated in a 24-well plate coated with 250 µl Matrigel. The next day, organoids were treated with dabrafenib (100 nM) + trametinib (10 nM), dabrafenib (100 nM) + ERAS007 (100 nM), or dabrafenib (100 nM) + panitumumab (3 µg ml−1; McKesson) for 72 h. The drugs were refreshed every 24 h. After the treatment, organoids were collected and subjected to RNA extraction. For cell viability experiments, dissociated organoids supplemented with 2% Matrigel were plated in a 96-well plate coated with 70 µl Matrigel; 48 h later, organoids were treated with various doses of dabrafenib + trametinib or dabrafenib + ERAS007 for 72 h and subjected to cell viability measurement using CellTiter-Glo 3D (Promega).

Organoid basal growth media

Organoid basal growth media consist of 30% DMEM/F12 media supplemented with 20% FBS, 50% WNT3A conditioned media, 20% R-spondin conditioned media, 1× B27 (Life Technologies; 17504-044), 1; N2 (Life Technologies; 17502-048), 10 mM nicotinamide (Sigma; N0636), 1.25 mM N-acetyl-L-cysteine (Sigma; A9165), 100 µg ml−1 Primocin (InvivoGen; ant-pm,2), 0.5 µM A83-01 (Tocris; 2939), 10 nM Gastrin (Sigma; G9145), 4 nM R-spondin (R&D Systems; 4645-RS-100), 4 nM Noggin (R&D Systems; 6057-NG-100), 5 nM fibroblast growth factor (R&D Systems; 345-FG-250), 5 ng ml−1 epidermal growth factor (R&D Systems; 236-EG-200), 3 µM p38i SB202190 (Sigma; S7067) and 10 µM Rho-kinase inhibitor Y-27632 (Sigma; Y0503).

Bulk RNA sequencing and analysis in organoids

Organoids were treated with or without dabrafenib (100 nM) + trametinib (10 nM) or dabrafenib (100 nM) + ERAS007 (100 nM) for 72 h and subjected to RNA extraction using the RNeasy Kit (Qiagen). A strand-specific transcriptome library was constructed and sequenced at BGI Genomics using the DNBSEQ platform. Paired-end 100-base pair RNAseq, 20 million reads per sample, were mapped to the reference genome (GCF_000001405.38_GRCh38.p12) using HISAT. Bowtie2 was used to align the clean reads to the reference genes. GO enrichment analysis was performed using the GO enrichment analysis web-based platform (,40. Genes with significant differential expression (log2(fold change) (FC) > 1; P < 0.05) were used for the enrichment test. The gene signature score was calculated as the mean of the log2 normalized expression of all genes in each gene signature. The same genes in the epithelial ISG program and the MAPK signature from scRNAseq analysis were used to calculate the ISG score and MAPK score here. Gene signature score delta was calculated as the score in treated samples subtracted by the score in untreated samples.

Quantitative PCR

RNA extraction was performed using the RNeasy Kit (Qiagen) per the manufacturer’s protocol. Reverse transcription was performed using qScript cDNA SuperMIx (Quantabio). qPCR analysis was performed using TaqMan Gene Expression Master Mix (Thermo Fisher Scientific) on the Roche Light Cycler 480. TaqMan Gene Expression Assays of IFIT1 (Hs03027069_s1), IFIT2 (Hs01922738_s1), IFIT3 (Hs01922752_s1), IRF1 (Hs00971965_m1), CXCL9 (Hs00970538_m1), CXCL10 (Hs00171042_m1), CXCL11 (Hs00171138_m1), DUSP6 (Hs04329643_s1), ETV4 (Hs00383361_g1), ETV5 (Hs00927578_g1) and SPRY4 (Hs01935412_s1) were purchased from Thermo Fisher Scientific. B-actin (Thermo Fisher Scientific; 4326315E) was used as endogenous control.

Western blot

ABPS cells were treated with dabrafenib (100 nM) and trametinib (10 nM) for 4, 24, 48 and 72 h (drug was refreshed every 24 h) and subjected to western blotting as previously described9 using antibodies to phospho-RSK1 (abcam; ab32413; 1:1,000 dilution) and GAPDH (Millipore Sigma; MAB374; 1:1,000 dilution).

Cell viability assay

APSe and ABPS cells were seeded at concentrations of 5 × 103 and 2 × 103 cells per well, respectively, in a 96-well plate. Cells were incubated for 24 h and treated with dabrafenib (100 nM), trametinib (10 nM) or the combination of DT for 72 h. Cell viability was measured using CellTiter-Glo (Promega) according to the manufacturer’s protocol.

RNAseq analysis from the BRAF/MEK/EGFRi combination trial

Bulk RNA sequencing data in 71 patients (including 45 paired day 1 and day 15 biopsy samples and 26 separate biopsy samples from baseline) enrolled in the previous BRAF/EGFRi ± MEKi trial with dabrafenib, panitumumab and trametinib in patients with BRAFV600E CRC were obtained from Novartis25. RNA sequencing data were trimmed mean of M values normalized41. Normalized expression data were then corrected for varying levels of liver gene expression using the expression of a 22-gene score (Supplementary Table 5) in a linear model to reduce the impact of biopsy location on the expression data. All expression values are log2 of liver-corrected counts per million. Gene signature expression levels are the mean log2 of corrected counts for all genes in the signature. Genes used for gene signature score calculation are listed in Supplementary Table 5.

Tissue processing and scRNAseq

Core needle biopsies were obtained from interventional radiology at Massachusetts General Hospital and Brigham and Women’s Hospital and transported in ice-cold hypothermosol before processing. Per patient and time point, the first two to three cores from the operative procedure were allocated for scRNAseq and cut into small pieces with scissors in a 1.5-ml Eppendorf tube containing 1 ml of enzymatic digestion mix (Miltenyi; Human Tumor Dissociation kit). The Eppendorf tubes were then transferred to a rotation shaker set to 37 °C and 550 r.p.m. and shaken for 15 min. The digestion mix was subsequently filtered through a 50-μm Celltrics strainer sitting on a 15-ml falcon tube on ice and mechanically dissociated once more with the plunger of a 1-ml syringe against the screen. The filter and enzymatic mixture were washed with RPMI containing 0.5% bovine serum. The cell suspension was spun at 1,500 r.p.m. (524g) for 4 min at 4 °C in a precooled centrifuge to pellet the cells. The pellet was lysed in 300 μl ammonium–chloride–potassium buffer for 2 min on ice, transferred into an 1.5-ml Eppendorf tube and then stopped with 1.1 ml RPMI containing 0.5% bovine serum. The cell suspension was then centrifuged at 1,500 r.p.m. (524g) for 4 min at 4 °C. The resulting cell pellet was resuspended in RPMI containing 0.5% bovine serum, filtered again through a 50-μm Celltrics strainer into a new 1.5-ml Eppendorf tube, spun at 1,500 r.p.m. (524g) for 4 min at 4 °C and then resuspended in 20 μl RPMI + 0.5% bovine serum. Cells were counted and loaded as 8,000 cells per channel using the 10× Genomics Single Cell 5’ Reagent Kit v.2. If cell counts permitted, up to three channels were loaded per patient and time point; 10× libraries were constructed according to the manufacturer’s instructions and sequenced at the Broad Institute Genomics Platform.

scRNAseq preprocessing, QC filtering and clustering

CellRanger v.6.0.2 was used to align reads to the GRCh38 human genome reference and aggregate all samples into a single feature-barcode matrix. Depth normalization was turned off during the aggregation. Starting with the filtered feature-barcode matrix, cells with gene counts of <200, mitochondrial gene levels of >50% or scrublet-based doublet scores of >0.3 were filtered out, keeping ~90% of cells. Leiden clustering was performed in Scanpy and clusters were manually annotated for major cell types using canonical markers such as epithelial cell adhesion molecule for epithelial cells.

Epithelial cell-specific QC filtering

Epithelial cells were analyzed separately from the immune and stromal cells. A cluster of epithelial cells in patient 26 was identified as small intestinal epithelial cells from the adjacent nonmalignant tissue and excluded from further analysis. A minimum threshold of 1,742 genes per cell was set based on the local minimum in the observed bimodal distribution of genes per cell to exclude cells with low gene counts that likely represent dead cells.

Identification of DEGs

Differential gene expression analysis was performed using the FindMarkers function in the Seurat v.4.1 R package separately in responders (>6 months survival) and nonresponders (<6 months survival) for pre- versus on-treatment specimens. Since epithelial cell numbers per specimen were very variable, we randomly downsampled the cells in specimens with large cell numbers to the median cell count of 557 per specimen. The volcano plots call out all genes with Bonferroni corrected P values < 0.05 and |log2FC| ≥ 1.

Calculating gene signature scores

We used the AddModuleScore function of the Seurat v.4.1 R package42,43. For each cell, this calculates the average expression of genes in the module subtracted by the average expression of a randomly selected set of control genes with similar expression across the cells. As input to the function, we used normalized expression and the default setting of 100 random control genes. Genes used to calculate signature scores were listed in Supplementary Table 4.

Gene set enrichment analysis

We used g:Profiler ( for the gene set enrichment analysis. We performed an ordered query of significant upregulated genes (log2FC > 1, Bonferroni-corrected P values < 0.05) in patients with PFS > 6 months and PFS < 6 months.

Statistical analyses

Experimental data were expressed as the mean ± s.e.m. of three or more individual experiments. The two-tailed Wilcoxon rank sum test was used to evaluate differences between unpaired data; the Wilcoxon signed rank test was used to compare paired data.

As detailed in the clinical protocol, descriptive statistics were used to summarize efficacy, including response rate (with 95% CI). PFS and median overall survival were calculated via Kaplan–Meier. Descriptive statistics were also used to report rates of AEs. General power calculations for the clinical trial cohort were based on a sample size of at least n = 25 providing 80% power to detect a difference in response rate of 22% compared with historical controls using a one-sided binomial test with alpha of 0.10.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Complete deidentified patient data (including study protocol) will be available indefinitely within 2 years after the last patient’s last survival follow-up visit and will be uploaded to Sequencing data of deidentified human subject specimens are deposited at dbGaP: phs003178. Any additional information required to reanalyze the data reported in this paper is available from the corresponding author upon request from the publication of the paper. Single-cell sequencing data is available here: Requests for data sharing will be responded to within 2–3 weeks. Source data are provided with this paper.


  1. Davies, H. et al. Mutations of the BRAF gene in human cancer. Nature 417, 949–954 (2002).

    Article CAS PubMed Google Scholar 

  2. Richman, S. D. et al. KRAS and BRAF mutations in advanced colorectal cancer are associated with poor prognosis but do not preclude benefit from oxaliplatin or irinotecan: results from the MRC FOCUS trial. J. Clin. Oncol. 27, 5931–5937 (2009).

    Article CAS PubMed Google Scholar 

  3. Flaherty, K. T. et al. Inhibition of mutated, activated BRAF in metastatic melanoma. N. Engl. J. Med. 363, 809–819 (2010).

    Article CAS PubMed PubMed Central Google Scholar 

  4. Falchook, G. S. et al. Dabrafenib in patients with melanoma, untreated brain metastases, and other solid tumours: a phase 1 dose-escalation trial. Lancet 379, 1893–1901 (2012).

    Article CAS PubMed PubMed Central Google Scholar 

  5. Kopetz, S. et al. Phase II pilot study of vemurafenib in patients with metastatic BRAF-mutated colorectal cancer. J. Clin. Oncol. 33, 4032–4038 (2015).

    Article CAS PubMed PubMed Central Google Scholar 

  6. Hyman, D. M. et al. Vemurafenib in multiple nonmelanoma cancers with BRAF V600 mutations. N. Engl. J. Med. 373, 726–736 (2015).

    Article CAS PubMed PubMed Central Google Scholar 

  7. Corcoran, R. B. et al. EGFR-mediated re-activation of MAPK signaling contributes to insensitivity of BRAF mutant colorectal cancers to RAF inhibition with vemurafenib. Cancer Discov. 2, 227–235 (2012).

    Article CAS PubMed PubMed Central Google Scholar 

  8. Corcoran, R. B. et al. BRAF gene amplification can promote acquired resistance to MEK inhibitors in cancer cells harboring the BRAF V600E mutation. Sci. Signal. 3, ra84 (2010).

    Article CAS PubMed PubMed Central Google Scholar 

  9. Ahronian, L. G. et al. Clinical acquired resistance to RAF inhibitor combinations in BRAF-mutant colorectal cancer through MAPK pathway alterations. Cancer Discov. 5, 358–367 (2015).

    Article CAS PubMed PubMed Central Google Scholar 

  10. Kopetz, S. et al. Encorafenib, binimetinib, and cetuximab in BRAF V600E-mutated colorectal cancer. N. Engl. J. Med. 381, 1632–1643 (2019).

    Article CAS PubMed Google Scholar 

  11. Corcoran, R. B. et al. Combined BRAF and MEK inhibition with dabrafenib and trametinib in BRAF V600-mutant colorectal cancer. J. Clin. Oncol. 33, 4023–4031 (2015).

    Article CAS PubMed PubMed Central Google Scholar 

  12. Corcoran, R. B. et al. Combined BRAF, EGFR, and MEK inhibition in patients with BRAF(V600E)-mutant colorectal cancer. Cancer Discov. 8, 428–443 (2018).

    Article CAS PubMed PubMed Central Google Scholar 

  13. Ribas, A. et al. Association of pembrolizumab with tumor response and survival among patients with advanced melanoma. JAMA 315, 1600–1609 (2016).

    Article CAS PubMed Google Scholar 

  14. Motzer, R. J. et al. Nivolumab versus everolimus in advanced renal-cell carcinoma. N. Engl. J. Med. 373, 1803–1813 (2015).

    Article CAS PubMed PubMed Central Google Scholar 

  15. Le, D. T. et al. Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade. Science 357, 409–413 (2017).

    Article CAS PubMed PubMed Central Google Scholar 

  16. Asaoka, Y., Ijichi, H. & Koike, K. PD-1 blockade in tumors with mismatch-repair deficiency. N. Engl. J. Med. 373, 1979 (2015).

    Article PubMed Google Scholar 

  17. Cancer Genome Atlas Network Comprehensive molecular characterization of human colon and rectal cancer. Nature 487, 330–337 (2012).

    Article Google Scholar 

  18. Ebert, P. J. R. et al. MAP kinase inhibition promotes T cell and anti-tumor activity in combination with PD-L1 checkpoint blockade. Immunity 44, 609–621 (2016).

    Article CAS PubMed Google Scholar 

  19. Liu, L. et al. The BRAF and MEK inhibitors dabrafenib and trametinib: effects on immune function and in combination with immunomodulatory antibodies targeting PD-1, PD-L1, and CTLA-4. Clin. Cancer Res. 21, 1639–1651 (2015).

    Article CAS PubMed Google Scholar 

  20. Hong, A. et al. Durable suppression of acquired MEK inhibitor resistance in cancer by sequestering MEK from ERK and promoting antitumor T-cell immunity. Cancer Discov. 11, 714–735 (2021).

    Article CAS PubMed Google Scholar 

  21. Ascierto, P. A. et al. Dabrafenib, trametinib and pembrolizumab or placebo in BRAF-mutant melanoma. Nat. Med. 25, 941–946 (2019).

    Article CAS PubMed Google Scholar 

  22. Ribas, A. et al. Combined BRAF and MEK inhibition with PD-1 blockade immunotherapy in BRAF-mutant melanoma. Nat. Med. 25, 936–940 (2019).

    Article CAS PubMed PubMed Central Google Scholar 

  23. Sullivan, R. J. et al. Atezolizumab plus cobimetinib and vemurafenib in BRAF-mutated melanoma patients. Nat. Med. 25, 929–935 (2019).

    Article CAS PubMed Google Scholar 

  24. Dummer, R. et al. Combined PD-1, BRAF and MEK inhibition in advanced BRAF-mutant melanoma: safety run-in and biomarker cohorts of COMBI-i. Nat. Med. 26, 1557–1563 (2020).

    Article CAS PubMed Google Scholar 

  25. Middleton, G. et al. BRAF-mutant transcriptional subtypes predict outcome of combined BRAF, MEK, and EGFR blockade with dabrafenib, trametinib, and panitumumab in patients with colorectal cancer. Clin. Cancer Res. 26, 2466–2476 (2020).

    Article CAS PubMed PubMed Central Google Scholar 

  26. Barras, D. et al. BRAF V600E mutant colorectal cancer subtypes based on gene expression. Clin. Cancer Res. 23, 104–115 (2017).

    Article CAS PubMed Google Scholar 

  27. Guinney, J. et al. The consensus molecular subtypes of colorectal cancer. Nat. Med. 21, 1350–1356 (2015).

    Article CAS PubMed PubMed Central Google Scholar 

  28. Pelka, K. et al. Spatially organized multicellular immune hubs in human colorectal cancer. Cell 184, 4734–4752 (2021).

    Article CAS PubMed PubMed Central Google Scholar 

  29. Hazar-Rethinam, M. et al. Convergent therapeutic strategies to overcome the heterogeneity of acquired resistance in BRAF(V600E) colorectal cancer. Cancer Discov. 8, 417–427 (2018).

    Article CAS PubMed PubMed Central Google Scholar 

  30. Canon, J. et al. The clinical KRAS(G12C) inhibitor AMG 510 drives anti-tumour immunity. Nature 575, 217–223 (2019).

    Article CAS PubMed Google Scholar 

  31. Morris, V. K. et al. Phase I/II trial of encorafenib, cetuximab, and nivolumab in patients with microsatellite stable, BRAFV600E metastatic colorectal cancer. J. Clin. Oncol. 40, 12-12 (2022).

    Article Google Scholar 

  32. Eng, C. et al. Atezolizumab with or without cobimetinib versus regorafenib in previously treated metastatic colorectal cancer (IMblaze370): a multicentre, open-label, phase 3, randomised, controlled trial. Lancet Oncol. 20, 849–861 (2019).

    Article CAS PubMed Google Scholar 

  33. Fedele, C. et al. SHP2 inhibition diminishes KRASG12C cycling and promotes tumor microenvironment remodeling. J. Exp. Med. 218, e20201414 (2021).

    Article CAS PubMed Google Scholar 

  34. Cibulskis, K. et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat. Biotechnol. 31, 213–219 (2013).

    Article CAS PubMed PubMed Central Google Scholar 

  35. Kim, S. et al. Strelka2: fast and accurate calling of germline and somatic variants. Nat. Methods 15, 591–594 (2018).

    Article CAS PubMed Google Scholar 

  36. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).

    Article CAS PubMed Google Scholar 

  37. Li, B. & Dewey, C. N. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics 12, 323 (2011).

    Article CAS PubMed PubMed Central Google Scholar 

  38. Tan, V. Y. & Fevotte, C. Automatic relevance determination in nonnegative matrix factorization with the beta-divergence. IEEE Trans. Pattern Anal. Mach. Intell. 35, 1592–1605 (2013).

    Article PubMed Google Scholar 

  39. Ashburner, M. et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet. 25, 25–29 (2000).

    Article CAS PubMed PubMed Central Google Scholar 

  40. Gene Ontology Consortium The Gene Ontology resource: enriching a GOld mine. Nucleic Acids Res. 49, D325–D334 (2021).

    Article Google Scholar 

  41. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010).

    Article CAS PubMed Google Scholar 

  42. Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).

    Article CAS PubMed PubMed Central Google Scholar 

  43. Stuart, T. et al. Comprehensive integration of single-cell data. Cell 177, 1888–1902 (2019).

    Article CAS PubMed PubMed Central Google Scholar 

  44. Raudvere, U. et al. g:Profiler: a web server for functional enrichment analysis and conversions of gene lists (2019 update). Nucleic Acids Res. 47, W191–W198 (2019).

    Article CAS PubMed PubMed Central Google Scholar 

Download references


This work was supported by the NIH/NCI Gastrointestinal Cancer SPORE (P50 CA127003 to A.S. and R.B.C.), the NIH/NCI Moonshot DRSN (U54CA224068), NIH/NCI R01 CA208756 (to N.H.), Stand Up to Cancer (SU2C) Colorectal Dream Team Translational Research (SU2C-AACR-DT22-17), and the Arthur, Sandra, and Sarah Irving Fund for Gastrointestinal Immuno-Oncology. SU2C is a division of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the Scientific Partner of SU2C. Partial clinical trial funding was provided by Novartis. We are also thankful for research fellowships: NIH/NCI T32CA207021, MGH Fund for Medical Discovery, and SITC Forward Fund (to J.H.C.), and DFG, SU2C Peggy Prescott Early Career Scientist Award PA-6146, SU2C Phillip A. Sharp Award SU2C-AACR-PS-32, BroadIgnite, and NIH/NCI K99CA259511 (to K.P.).

Author information

Author notes
  1. These authors contributed equally: Jun Tian, Jonathan H. Chen, Sherry X. Chao, Karin Pelka.

Authors and Affiliations

  1. Massachusetts General Hospital Cancer Center and Harvard Medical School, Boston, MA, USA

    Jun Tian, Jonathan H. Chen, Vjola Jorgji, Princy Sindurakar, Arnav Mehta, Tomonori Oka, Mei Huang, Maxwell Spurrell, Jill N. Allen, Jeffrey W. Clark, Samuel J. Klempner, David P. Ryan, Katie Kanter, Emily E. Van Seventer, Islam Baiev, Gary Chi, Joy Jarnagin, William B. Bradford, Edmond Wong, Alexa G. Michel, Isobel J. Fetter, Giulia Siravegna, Angelo J. Gemma, Shadmehr Demehri, Aparna R. Parikh, Nir Hacohen & Ryan B. Corcoran

  2. The Broad Institute of Massachusetts Institute of Technology and Harvard, Cambridge, MA, USA

    Jonathan H. Chen, Sherry X. Chao, Karin Pelka, Julian Hess, Arnav Mehta, David Lieb, Gad A. Getz & Nir Hacohen

  3. Gladstone-UCSF Institute of Genomic Immunology, Gladstone Institutes Department of Microbiology and Immunology, UCSF, San Francisco, CA, USA

    Karin Pelka

  4. Dana Farber Cancer Institute and Harvard Medical School, Boston, MA, USA

    Marios Giannakis, Kelly Burke, Thomas A. Abrams, Andrea C. Enzinger, Peter C. Enzinger, Nadine J. McCleary, Jeffrey A. Meyerhardt & Matthew B. Yurgelun

  5. The Koch Institute, Massachusetts Institute of Technology, Cambridge, MA, USA

    Jonathan Braverman & Omer Yilmaz

  6. Department of Immunology, Blavatnik Institute, Harvard Medical School, Boston, MA, USA

    Arlene Sharpe

  7. Novartis Institute for Biomedical Research, Cambridge, MA, USA

    Rebecca Leary & Catarina D. Campbell


J.T., J.H.C., S.X.C., K.P., M.G., J.H., K.B., J.B., T.O., M.H., G.S., A.J.G., A.S., S.D., R.L., C.D.C., O.Y., G.A.G., A.R.P., N.H., and R.B.C. participated in the design, execution and/or interpretation of the reported experiments or results. J.T., J.H.C., S.X.C., K.P., V.J., P.S., A.M., D.L., M.S., J.N.A., T.A.A., J.W.C., A.C.E., P.C.E., S.J.K., N.J.M., J.A.M., D.P.R., M.B.Y., K.K., E.E.V.S., I.B., G.C., J.J., W.B.B., E.W., A.G.M. and I.J.F. participated in the sample acquisition or data analysis. J.T. and R.B.C. wrote the paper, with all authors contributing to writing and providing feedback. N.H. and R.B.C. supervised all aspects of the research.

Corresponding authors

Correspondence to Nir Hacohen or Ryan B. Corcoran.

Ethics declarations

Competing interests

S.X.C. is an employee of Google Ventures. M.G. receives research funding from Servier and Janssen. A.M. has served a consultant/advisory role for Third Rock Ventures, Asher Biotherapeutics, Abata Therapeutics, Flare Therapeutics, venBio Partners, BioNTech, Rheos Medicines and Checkmate Pharmaceuticals; is an equity holder in Asher Biotherapeutics and Abata ThPerapeutics; and has a sponsored research agreement with Bristol Myers Squibb and Olink Proteomics. P.C.E. is/has been a consultant and has received honoraria from ALX Oncology, Arcus Bioscience, Astellas, AstraZeneca, Blueprint Medicines, Chimeric Therapeutics, Celgene, Coherus, Daiichi-Sankyo, Five Prime, Ideaya, Istari, Legend, Lilly, Loxo, Merck, Novartis, Ono, Servier, Taiho, Takeda, Turning Point Therapeutics, Xencor and Zymeworks. S.J.K. has served a consultant/advisory role for Astellas, Merck, Bristol Myers Squibb, Daiichi-Sankyo, Pieris, AstraZeneca, Natera, Eli Lilly, Mersana and Sanofi-Aventis. S.J.K. owns stock in Turning Point Therapeutics. J.A.M. has served as an advisor/consultant to Merck Pharmaceutical and COTA Healthcare. M.B.Y. receives research funding from Janssen Pharmaceuticals. C.D.C. is an employee and shareholder of Novartis. N.H. receives research funding from Bristol Myers Squibb, has equity in BioNTech and advises and has equity in Related Sciences/Danger Bio. R.B.C. has received consulting or speaking fees from Abbvie, Amgen, Array Biopharma/Pfizer, Asana Biosciences, Astex Pharmaceuticals, AstraZeneca, Avidity Biosciences, BMS, C4 Therapeutics, Chugai, Cogent Biosciences, Elicio, Erasca, Fog Pharma, Genentech, Guardant Health, Ipsen, Kinnate Biopharma, LOXO, Merrimack, Mirati Therapeutics, Natera, Navire, Nested Therapeutics, N-of-one/Qiagen, Novartis, nRichDx, Remix Therapeutics, Revolution Medicines, Roche, Roivant, Shionogi, Shire, Spectrum Pharmaceuticals, Symphogen, Syndax, Tango Therapeutics, Taiho, Theonys, Warp Drive Bio and Zikani Therapeutics; holds equity in Avidity Biosciences, C4 Therapeutics, Cogent Biosciences, Erasca, Kinnate Biopharma, Interline Therapeutics, Nested Therapeutics, nRichDx, Remix Therapeutics, Revolution Medicines and Theonys; is a cofounder, equity holder and board member of Alterome Therapeutics; and has received research funding from Asana, AstraZeneca, Lilly, Novartis and Pfizer. The remaining authors declare no competing interests.


Nature | 免疫疗法对有些人没用的原因终于找到了!


免疫疗法的开发为癌症治疗带来革命性的影响,其中像是抗PD-1、PD-L1抗体等免疫检查点抑制剂已成为许多种类癌症的一线疗法,然而至今仍有许多病患对此类疗法并不产生应答。近日,加州大学洛杉矶分校(UCLA)的科学家与PACT Pharma公司合作,在《自然》期刊发表一项突破性的研究。根据新闻稿,这是首次科学家能够识别与分析免疫细胞如何“看见”癌细胞并产生相关反应,这将有助于解释免疫检查点抑制剂背后的耐药机制,并为这些不应答患者开发相对应的个体化免疫癌症疗法。





为了了解癌症患者对免疫检查点抑制剂不产生应答的原因,研究人员使用了一种之前发表在《自然》期刊并后来通过PACT公司改进、开发的专有ImPACT Isolation Technology平台技术。这种技术能够自血液与肿瘤中识别并分离对癌细胞突变有反应的T细胞(即那些能够辨认neoE的细胞)并对其所辨认的突变蛋白序列进行分析,因此科学家能够获得患者体内癌细胞突变种类、产生抗肿瘤反应的T细胞种类、数量等信息。
通过这项科技,我们能够确切知道某位癌症患者体内的免疫系统是如何区分体内的癌细胞与正常细胞,”论文作者之一,也是UCLA教授的Antoni Ribas博士说道。


“这项研究显示,对疗法不产生应答的患者体内,其抗肿瘤T细胞反应仍受到激发,”论文的第一作者,也是UCLA的助理教授Cristina Puig-Saus博士说道,“也许我们能够通过分离这些T细胞,并对其识别突变的T细胞受体(TCR)进行分析,根据所获得的信息对大量T细胞进行基因修饰,使得这些细胞能够专一识别患者肿瘤。最终通过体外扩增这些细胞,并输回患者体内来治疗患者。

根据这项假说,研究人员使用PACT司的PACT^NV™ technology平台以及CRISPR 基因置换技术,生产了带有能够专一识别病患肿瘤突变TCR的工程化T细胞,并将这些细胞扩增后输入回原本患者体内。研究发现,这些工程化的T细胞能够在病患体内发动免疫攻击肿瘤,无论这些病患之前是否对PD-1疗法产生应答。