我们在欧洲的三个中心（英国RMH（UK），贝尔法斯特市医院（英国）诺斯州诺斯特郡（Switzerland）（Switzerland）（Switzerland）（Switzerland）（Switzerland）（Switzerland）（Switzerland）（Switzerland（Switzerland），我们在欧洲三个中心（UK（RMH））进行了国际，单臂，开放标签试验：NCT03177187，Eudract：2016-003141-28）。符合条件的患者是同意18岁的患者，他们患有组织学确认的前列腺腺癌组织学的MCRPC，在试验时通过实体瘤的反应评估标准（V.1.1）和/或前列腺癌工作组2标准和PSA的癌症记录了癌症的进展。患者需要在接受雄激素剥夺治疗（Orchioto术和/或持续的黄体释放激素激素（LHRH）激动剂治疗）的过程中患有疾病进展，并确认至少在每种治疗12周的enzalutamide，Darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，darolutamide，至少接受了12周的治疗。在用其他前列腺癌治疗治疗之前：患者需要接受血清睾丸激素的雄激素剥夺治疗 <50 ng dl−1 (<2.0 nM); patients needed to be Eastern Cooperative oncology Group performance status of 0 or 1, and have adequate haematologic, renal, liver, and coagulation function; patients also needed to be willing to undergo pre- and on-treatment mCRPC biopsies, when safe and feasible.
Patients were excluded if their prostate cancer was predominantly small cell or neuroendocrine differentiated. Patients were excluded if they had any of the following: surgery, chemotherapy or other anticancer therapy (with the exception of an ARSI and gonadotropin hormone-releasing hormone analogue therapy) within 4 weeks before trial entry; limited field radiotherapy within 2 weeks or wide-field radiotherapy within 4 weeks of trial entry; hypoaldosteronism or hypopituitarism; history of seizures or predisposing factors to seizures; known central nervous system metastasis; malabsorption syndrome; prolonged QT interval > 470 milliseconds; clinically important rhythm, conduction, or ECG abnormality; predisposing factor to QT prolongation; coronary intervention, myocardial infarction, angina, or congestive cardiac failure (New York Heart Association ≥grade 2) in the past 6 months; uncontrolled hypotension or hypertension; clinically important history of liver disease (for example, Child–Pugh B or C, viral or other hepatitis, current alcohol abuse, or cirrhosis); malignancy other than prostate cancer within the past 5 years; immunocompromising disorder; thromboembolic event within the past 12 months; active or uncontrolled autoimmune disorder requiring steroids. Full eligibility criteria are described in the study protocol (Supplementary Information).
The study was conducted in accordance with the provisions of the Declaration of Helsinki and Good Clinical Practice guidelines. Regulatory approvals were obtained from the Medicines Healthcare products Regulatory Agency, Swiss Swissmedic and the institutional research ethics committee (REC; the London-Surrey Borders REC (UK sites) and Comitato Etico Cantonale Bellinzona (Switzerland)). Written informed consent was obtained from all participants. No participant compensation was provided. A safety review committee evaluated the safety and tolerability at regular intervals after recruitment of three patients to a schedule. All protocol amendments were approved by the trial sponsor, Medicines Healthcare products Regulatory Agency, Swissmedic and local UK and Swiss RECs. The study was sponsored and monitored by The Institute of Cancer Research (ICR), UK. The study was registered on ClinicalTrials.gov before commencement.
In this investigator-initiated, international, open-label, phase 1 study, we evaluated five escalating doses of orally administered AZD5069 (40 mg BD, 80 mg BD, 120 mg BD, 160 mg BD and 320 mg BD) in combination with standard, fixed-dose, orally administered enzalutamide (160 mg OD), over 28-day cycles, until disease progression, intolerance or withdrawal of consent. During the first cycle (42 days), AZD5069 was commenced 2 weeks before enzalutamide in the first four cohorts, primarily to evaluate any pharmacokinetic interactions between the two drugs. The starting dose of AZD5069 was determined on the basis of preclinical pharmacokinetics results as well as pharmacodynamic, pharmacokinetic and safety results from previous studies in humans in which the main side effect observed was dose-dependent, reversible neutropaenia in blood, without significant increase in infection rate in patients with airway disease, or healthy individuals.
The dose-escalation phase used a rule-based 3 + 3 design, with an initial three patients enrolled per dose level. If none of the first three patients experienced a DLT, dose escalation proceeded to the next dose level. If one instance of DLT was observed in three patients, up to six patients were treated at that dose level. If fewer than two of six patients at any dose level experienced a DLT, dose escalation continued to the next level. If at least two out of up to six patients experienced a DLT, dose escalation stopped and this dose level was defined as the maximum administered dose. once this maximum administered dose was defined, the maximum tolerated dose (MTD) was confirmed at the dose level below the maximum administered dose. At least six evaluable patients were required to establish the MTD at a specific dose level. only doses at which no more than one of six patients had a DLT could be defined as the MTD. Four potential dose-escalation cohorts with increasing AZD5069 doses (40 mg BD, 80 mg BD, 120 mg BD and 160 mg BD) were planned initially. The study protocol was amended on 16 December 2020 to explore a fifth dose level of AZD5069, 320 mg BD, with the option to de-escalate to 240 mg BD (dose level 4B) if dose level five was intolerable, and the study drugs were administered concurrently. This amendment occurred after previous dose levels were deemed safe and because a decrease in AZD5069 exposure was observed after adding enzalutamide. Intra-patient dose escalation was not permitted. Start of dosing between the first and second patient enrolled to each dose level was staggered by 1 week. once the MTD was determined in the phase 1 study, the recommended phase 2 dose was determined on the basis of available data, including but not limited to safety and response.
Adverse events were monitored at least weekly during cycles 1 and 2, and then once per cycle from cycle 3 onwards, and graded using the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) v.4.0, until 28 days after the discontinuation of study treatment or until the resolution of a persistent drug-related adverse event. DLTs were defined as described in the study protocol. Notably, febrile neutropaenia (neutrophil count <0.5 × 109 l−1 and fever >38.3 °C or fever ≥38 °C for >1 H），感染4级中性粒细胞膜和4级中性粒细胞7天或更长时间为DLT。不允许具有生长因子支持或抗染料药的预防。研究人员确定了不良事件是否与研究药物有关。如果患者经历了临床意义和/或不可接受的毒性，包括DLT（不归因于疾病或疾病相关的过程），剂量被中断或剂量降低并根据需要进行支持治疗。如果毒性在发病后的14天内解决或恢复为≤CTCAEV.4.0 1级，则可以在与赞助商一致后重新启动enzalutamide和AZD5069的组合处理。
在研究人员认为无法通过使用适当的医疗干预措施来改善的与恩扎拉胺相关的3级或更高毒性的患者中，enzalutamide剂量被中断，直到毒性提高到1级或较低的毒性，并允许减少剂量为1级或较低的剂量为120 mg的乙酰胺剂量。不允许进一步降低恩扎拉塔胺的剂量。在由于与AZD5069相关的不良事件引起的剂量中断期间，如果毒性在不到八天的时间内解决或恢复为CTCAE v.4.0级≤1；但是，如果毒性花费了8到14天才能解决或恢复到≤1级，则根据协议中的规格，可以将AZD5069以较低剂量重新启动（比最后剂量低的剂量水平）。恩扎拉胺剂量保持不变。对于所有其他事件，如果毒性在超过14天后无法解决CTCAE v.4.0≤1级，则停止治疗并观察到患者直到毒性解决。如果必须永久停止任何一种研究药物，则将患者取消研究。
放射学肿瘤反应是通过胸腔，腹部和骨盆，骨扫描的计算机断层扫描来测量的，并且在指示的情况下，在基线时，每3个循环一次，然后在治疗结束时，如果自上次扫描以上是8周以上，则测量全身磁共振成像。在基线，然后在每个周期的第1天以及治疗结束时测量PSA和CTC计数。使用先前描述的方法38分析了CTC计数。
收集了AZD5069和enzalutamide的药代动力学分析的血液样本，并在AZD5069（周期-14周期-14周期），第1天（周期1天）（AZD5069单疗疗法2周后）和第2天后（AZD5069单疗法）和AZD5069的所有患者（周期1天1天）和AZD5069和ENZD5069和ENED（ENZD5069循环后2周后），在所有患者的前四个剂量水平中进行分析。研究方案中列出了特定的药代动力学收集时间点。使用非室内分析（Phoenix V.8.1，Certara）计算药代动力学参数。来自匹配疾病部位（淋巴结，骨骼和软组织）的新鲜肿瘤针核活检在基线（在治疗开始后的1周内），在开始AZD5069后约2周，当时这是安全可行的，这是可行的。在经验丰富的介入放射科医生（N.T.）的计算机断层扫描或超声指导下，从18例患者获得肿瘤活检。三名患者没有进行活检，因为这不是安全的，或者患者下降了。一名患者只有基线肿瘤活检。4周后，两名患者进行了治疗活检。活检后，将肿瘤样品浸入10％中性缓冲福尔马林24小时中。通过石蜡处理样品进行组织学检查。用甲氨酸和曙红和曙红染色了福尔马林固定的石蜡（FFPE）肿瘤活检的三个3μM切片，以证实病理学家（B.G.）存在肿瘤。分析中排除了肿瘤含量不足或明显碎屑的肿瘤样品（B.G.）。通过IHC分析了基线FFPE样品，并通过使用HTG EDGESEQ对IF和靶向RNA分析分析了治疗前和治疗后的FFPE活检。所有初始分析失败的样本至少均至少休息一次。补充表4详细介绍了肿瘤样品及其分析方法的列表。
主要终点是识别DLT，估算MTD并确定与enzalutamide在160 mg OD中结合使用的建议的AZD5069剂量2剂量的AZD5069。次要终点为：（i）客观响应的速率，其定义为确认的软组织客观响应，通过验证评估标准在固体疾病的响应评估标准v.1.1中，患有可测量的疾病的人和/或PSA下降了4周或4周确认的≥50％，并且/或/或CTC计数转换为基础上7.5 ml≥5ml≥5ml≥5ml。 <5 per 7.5 ml of blood at nadir; (ii) pharmacokinetic parameters, including maximum concentration, area under the concentration–time curve; and (iii) pharmacodynamic changes including identifying patients whose blood NLR, neutrophil, and intratumour myeloid cell density decrease. Exploratory endpoints included blood cytokine levels and evaluation of tumour molecular profile on response.
The association between myeloid cell densities and NLR was evaluated in two cohorts (Supplementary Table 1). Cohort 1 consisted of 48 mCRPC biopsies from patients treated at ICR/RMH, oncology Institute of Southern Switzerland and Belfast City Hospital between 2012 and 2021. All patients provided informed consent, and enrolled onto institutional protocols approved by the local RECs (REC reference: 04/Q0801/60, 11/LO/2019). The validation cohort consisted of 57 mCRPC biopsies from patients treated at ICR/RMH between 2012 and 2016 under institutional protocols approved by the local REC (REC references: 04/Q0801/60, 2017-01002 CE TI 3237). Full blood counts were carried out using routine automated haematology analysers. NLR was defined as the quotient of the absolute peripheral blood neutrophil count divided by the absolute blood lymphocyte count. For comparisons of peripheral blood NLR with intratumour myeloid cell density, blood counts collected on the day of the biopsy, or when this was not available, within 7 days preceding the biopsy were used. Human biological samples were sourced ethically and their research use was in accordance with the terms of the informed consent provided. Studies of CXCR2 expression on immune cells and tumour cells consisted of 14 patients treated at the RMH who underwent mCRPC biopsies under a research protocol approved by The RMH REC (REC reference: 04/Q0801/60) providing consent for these analyses.
Antibodies against CXCR2, FOXP3, MUM1, CD163, CD68, HLA-DR, CD4, CD38, CD206, CD8 and GzB were validated by western blot and/or IHC comparing detection of protein expression in cells treated with either non-targeting control siRNA or ON-TARGETplus pooled siRNA against the target gene (Dharmacon) or using positive and negative control cell lines. Cells were authenticated by STR profiling and tested for mycoplasma (Venor GeM Mycoplasma Detection Kit, Minerva Biolabs). Markers were validated for appropriate tissue localization on immunohistochemical staining of relevant positive and negative tissue controls and reviewed by a certified pathologist (B.G.). Validation for PTEN, CD4, CD8, FOXP3, CD11b, CD15, CD14, CD138, CD20, Syn, CgA and AR-V7 was also previously described13,17,50,51. IHC was carried out on FFPE tissue sections using an automated staining platform (Bond RX, Leica Biosystems). Bone biopsies were decalcified using pH 7 EDTA for 48 h at 37 °C. once validated for target sensitivity and specificity, the antibodies were further optimized for IHC, multiplex IF and hyperplex IF using methods described below. The full list of antibodies, working dilutions and incubation times is in Supplementary Tables 7 and 8.
FFPE CRPC biopsies were stained using a hyperplex IF assay. For paired samples, the pre- and on-treatment biopsies from each patient, along with the positive and negative controls (tonsil, ovarian cancer, appendix, HeLa and LNCaP cell line pellet), were placed on the same slide to control for any technical variability in staining intensity and allow for comparison of the pre- and on-treatment biopsies. Samples from the tissue microarrays had been stained previously using orthogonal methods (IF and/or IHC) for /confirm/iation. Standard operating procedures were implemented to control for known factors that can impact IF staining intensity, including the use of antibodies with the same lot number, minimization of freeze–thaw of antibodies, and controlling for the temperature of the experiment. Automated hyperplex IF staining and imaging was carried out on the COMET platform (Lunaphore Technologies). Slides underwent iterative staining and imaging, followed by elution of the primary and secondary antibodies52.
FFPE tissue sections of 3 μm in thickness were baked in an oven for 60 min at 60 °C, followed by deparaffinization in xylene and rehydration in a series of ethanol solutions of decreasing concentrations. Next, tissue sections were fixed in 10% neutral-buffered formalin solution (No. BAF-0010-05A, CellPath) for 20 min at room temperature. Antigen retrieval was achieved by heating the slides in heat-induced epitope retrieval buffer H pH 9 (No. TA-999-DHBH, Epredia, Shandon Diagnostics) in the PT Module (No. A80400011, Thermo Fisher Scientific) for 60 min at 102 °C. Subsequently, slides were rinsed and stored in Multistaining Buffer (BU06, Lunaphore Technologies) until use.
The hyperplex IF protocol template was generated using the COMET Control Software (v.0.70.0.1, Lunaphore Technologies), and reagents were loaded onto the device to carry out the sequential IF (seqIF) protocol52. Secondary antibodies were used as a mix of two species’ complementary antibodies plus DAPI, Alexa Fluor Plus 647 goat anti-rabbit (No. A32733, 1:400 dilution, Thermo Scientific) and Alexa Fluor Plus 555 goat anti-mouse (No. A32727, 1:200 dilution, Thermo Scientific) diluted in Intercept T20 (TBS) antibody diluent (No. 927-65001, LI-COR Biosciences). Nuclear signal was detected with DAPI (No. 62248, dilution 1:1,000, Thermo Fisher Scientific) by dynamic incubation of 2 min. Primary antibodies were diluted in multistaining buffer (BU06, Lunaphore Technologies). For each cycle, the following exposure times were used: DAPI 80 ms, TRITC 400 ms, Cy5 200 ms. The elution step lasted 2 min for each cycle and was carried out with elution buffer (BU07-L, Lunaphore Technologies) at 37 °C. The quenching step lasted for 30 seconds and was carried out with quenching buffer (BU08-L, Lunaphore Technologies). The imaging step was carried out with imaging buffer (BU09, Lunaphore Technologies). The seqIF protocol in COMET resulted in a multi-stack ome.tiff file in which the imaging outputs from each cycle are stitched and aligned. COMET ome.tiff contains a DAPI image, intrinsic tissue autofluorescence in TRITC and Cy5 channels, and a single fluorescent layer per marker.
Elution efficiency and epitope stability of each biomarker were assessed separately through several rounds of staining, elution and imaging on positive control tissue. Antibody titration was carried out to identify the best antibody dilution and incubation time. The staining sequence was optimized through an iterative process using several positive and negative FFPE controls (appendix, tonsil, ovarian cancer and prostate cancer), cell lines (PC3, LNCaP and HeLa) and a patient-derived xenograft with a neuroendocrine phenotype (CP142)17. Images were reviewed by a pathologist (B.G.) and used to determine the final marker permutation (Supplementary Table 7 and Supplementary Figs. 1 and 2).
Six-colour OPAL-based sequential IF staining was carried out on the Bond RX automated staining platform (Leica Biosystems). FFPE tissue sections of 3 µm underwent heat-induced epitope retrieval with epitope retrieval solution 2 (pH 9.0; No. AR9640, ER2, Leica Biosystems) followed by endogenous peroxidase blocking (Novocastra Peroxidase Block, No. RE7157, Leica Biosystems) for 10 min. Nonspecific antibody binding was blocked using OPAL antibody diluent/block (ARD1001EA, Akoya Biosciences) for 10 min. Primary antibodies against CXCR2, CD15, CD11b, CD14 and HLA-DR; Supplementary Table 8) were sequentially incubated for 30 min followed by detection with the Novolink Max Polymer Detection System (RE7280-K, Leica Biosystems). IF signals for CXCR2, CD15, CD11b, CD14 and HLA-DR were visualized using TSA coumarin (NEL703001KT, Akoya Biosciences), OPAL 520 (NEL820001KT, Akoya Biosciences), OPAL 570 (NEL820001KT, Akoya Biosciences), OPAL 650 (FP1496001KT, Akoya Biosciences) and OPAL 780 (FP1501001KT, Akoya Biosciences), respectively, and counterstained with spectral DAPI. Slides were scanned using the VS200 Research Slide Scanner (Olympus).
The hyperplex and six-colour IF assay images were reviewed by a certified pathologist (B.G.) and histopathologists (M.C., A.F., I.F.). Images were analysed using Halo software (Indica Labs). Tissue segmentation was carried out using a supervised machine learning algorithm to recognize prostate cancer foci and surrounding stroma. Cell segmentation was achieved with nuclear DAPI counterstain and tumour-infiltrating immune cells were phenotypically characterized by cell surface marker. We identified CD11b+HLA-DRloCD15+CD14− and CD11b+HLA-DRloCD15−CD14+ cells using a supervised machine learning algorithm trained by a pathologist (B.G.) as previously described13.
For the hyperplex IF panel, a threshold for positivity for each marker used for cell phenotyping was set by the pathologist by referencing positive and negative control tissue or cell line pellets stained on the same slide. The same thresholds were applied to the entire slide. Manual curation and comparison with controls was essential because differences in tissue type and quality can impact the intensity of different antibodies differently, although all phenotypic markers showed excellent signal-to-noise ratio (>15）。使用这些阈值，使用Halo软件（Indica Labs）来分析标记的每个单元，以提供感兴趣的表型标记的单细胞级二进制读数。建立了一种布尔门控策略，以基于标记的强度和特异性来识别细胞的兴趣类型（补充图2）。
使用113条基因面板进行靶向NG，对16个可用的预处理肿瘤活检和3个预处理细胞的DNA样品，这些样品从在Streck管中收集的20 mL血浆中提取。使用先前描述的方法38,53进行了NGS。使用定制的Generead DNASEQ MIX-N-MAINCT v.2面板（Qiagen）从40 ng的无细胞DNA构建库，并在Miseq Sequencer（Illumina）上进行测序。在综合基因组观众（V.2.16.1，Broad Institute）中手动检查了躯体变体调用。通过对AZD5069 320 320 mg BD剂量水平进行其他三个样品进行此测定，评估了NGS的敏感性，从而评估了该测定法，该样品还进行了肿瘤活检的NGS，我们确认在肿瘤活检中也发现了所有病原性变化。
HTG Edgeseq（HTG分子诊断）使用HTG人类转录组进行了19,616个核酸酶保护探针（NPP），包括19,398基因靶标的对照50-Nuceletes，使用HTG人体转录组，使用HTG EDGESEQ（HTG分子诊断）进行了HTG EDGESEQ（HTG分子诊断），其中包括19,398基因对照50-NUCELETES，使用HTG人物转录组，使用HTG EDGESEQ（HTG分子诊断）进行了HTG EDGESEQ（HTG分子诊断），由HTG Edgeseq（HTG分子诊断）进行了，包括19,398基因对照50-核透析剂，使用HTG，对开始AZD5069之前和之后的RNA分析进行了。外部RNA控制联盟，22个测量GDNA和4个阳性对照探针的探针。该测定最少为11 mm2的FFPE肿瘤微截止切片。根据制造商的说明将部分裂解，并添加到HTG Edgeseq处理器（HTG分子诊断）上的96孔板中，并在其上进行了定量的核酸酶保护测定法。添加DNA核酸酶保护探针（NPP）是自动化的，并允许它们与目标mRNA杂交16小时。通过S1消化去除过量的非杂交DNA探针和非杂交mRNA，仅将NPPS与mRNA杂交。这产生了样品中存在的DNA检测探针与mRNA靶标的1：1之比。通过通过聚合酶链反应（PCR）将测序指数和分子条形码添加到NPP来构建文库。使用KIT KAPA库定量套件Illumina平台清理和定量PCR进行定量PCR后，将库合并并在NextSeq 500上使用高吞吐量75-Cycle v.2.5 kit（Illumina）进行测序。使用BCL2FATSQ V.2.0生成FASTQ文件，并使用HTG Edgeseq Parser软件（V.5.3，HTG Molecular Diagnostics）生成原始计数数据。使用HTG EDGESEQ揭示分析软件分析数据。进行了几个质量控制指标：QC0（样品数量不足或样品质量不足），阳性对照探针> 4％读数被标记为失败；QC1（读取深度不足），总对齐读数 <7 million per sample was marked as a failure; QC2 (high background signal) with median log10 negative control probes >2被标记为失败；QC3（DNase对GDNA的不完全消化）中位数log10 gDNA对照探针> 1被标记为故障。从随后的分析中删除了失败的任何质量控制的样本。使用HTG EDGESEQ揭示了DESEQ2分析管道和R软件（v.4.2.3），进行了预处理样品和治疗样品之间的差异基因表达。
Serum samples were collected at baseline, on day 1 of every cycle, and on cycle 1 day 15. CXCL2, CXCL5, CXCL6, CXCL7 and CXCL8 were measured in patient serum (diluted 1:2 except for CXCL7, which was diluted 1:200), then analysed using R&D Systems Luminex discovery assays using the Luminex 200 and interpolated using a five-parameter logistic curve fit.使用Perkin Elmer Envision 2103多标记板读取器使用R＆D系统在整洁的患者血清中测量CXCL1，并使用线性回归进行了插值。Luminex和ELISA分析通过ICR验证了良好的临床实践依从性，并包括血清的质量控制样品，未在每次分析运行中健康志愿者血清中使用重组蛋白标准的无重组或峰值。补充表9中提供了ELISA试剂的清单。
SU2C -PCF前列腺癌Dream 34产生的总共159个MCRPC转录组已下载并重新分配。仅使用了使用Polya+ RNA隔离的样品制备的样品（即通过捕获方法进行的图书馆制备的样品被排除在外）。共有141个MCRPC转录组具有可用于生存分析的生存数据。与在RMH/ICR治疗的患者中有一个单独的95 MCRPC转录组为27；94个MCRPC转录组用于生存分析，因为1名患者无法获得生存数据。使用TopHat2（v.2.0.7），将SU2C -PCF转录组与人参考基因组（GRCH37/HG19）排列。使用袖扣（v.2.2.1）计算基因表达为每千倍酶每千倍酶的片段（fpkm）。对RMH/ICR Bulk RNA-Seq数据集中存在的泛免疫基因26的无偏见进行了与NLR相关的质疑。MDSC签名改编自先前发布的标志10,33。使用双面Spearman的等级相关测试对关联进行分析。
对于MCRPC肿瘤活检RNA分析，使用EDGESEQ处理器处理HTG EDGESEQ数据，并包含了具有默认设置的多个步骤（解析，质量控制和归一化）。归一化计数转换为log2 [百万计数]，用于下游分析。基因集变异分析（GSVA，R封装GSVA V.1.4）用于分子签名分析。
Single-cell transcriptomic data from 15 mCRPC samples from 14 patients (https://www.nature.com/articles/s41591-021-01244-6)35 and single-cell transcriptomic data from 11 patients with localized prostate cancer (https://www.nature.com/articles/s41467-021-27322-4)36 were下载。将数据加载到R软件（v.4.1.3）中。将来自局部前列腺癌数据的原始计数归一化。两个数据集均使用Seurat（v.4.3.0）处理，并使用CellDex（v.1.4.0）库中的蓝图参考数据集对使用Singler（v.1.8.1）进行了缩放，聚类，尺寸减少和单元格分配。
务实选择样本量。根据指导剂量升级决策的基于规则的3+3设计，同类群体的大小为3例，不允许跳过剂量水平。在DLT期间完成DLT期或经历了DLT的患者被认为是可评估人群的一部分。接受至少一剂研究药物的患者被认为是安全人群的一部分。为了评估应有的反应，患者必须符合资格标准，至少接受三个试验药物的周期，并且对疾病进行了基线评估。使用两侧Mann-Whitney U检验进行了分类为响应者的患者与分类为非反应者的患者之间的基线特征的比较。使用双面配对的Wilcoxon签名级测试对配对的药代动力学和药效学参数进行了比较。安全变量以及药代动力学和药效动力学终点是描述性汇总的。
免疫细胞密度和连续的基因表达数据被描述为个体值，以及中位和四分位间范围的小提琴图或小提琴图。分析了整个幻灯片上的所有可分析区域。双面Spearman的等级相关测试用于估计连续变量之间的关联，两边的Mann-Whitney U检验用于测试未配对组之间的差异，并且使用了两侧配对的Wilcoxon签名范围测试来测试配对样品之间的差异。Kruskal – Wallis检验用于比较多种疾病部位的髓样细胞密度。进行了多变量线性回归，以确定NLR或中性粒细胞计数和髓样细胞密度之间的关联是否受活检部位影响。MaxStat方法54,55确定了用于生存分析的基因表达截止。使用Kaplan-Meier方法估算了总体存活和无进展生存。使用对数秩检验进行了生存曲线的组间比较。使用COX回归计算具有95％置信区间的危险比。所有P值≤0.05都被认为是显着的。在多种假设测试的髓样基因特征测试的背景下，将Bonferroni校正用于调整多重性，而不是NLR与免疫基因之间的关联，分析的目的是识别与NLR相关的最高排名最高的免疫基因。循环细胞因子水平被描述地提出。使用R软件（V.4.2.2）进行统计分析，并根据试验相关分析的统计分析计划。
有关研究设计的更多信息可在与本文有关的自然投资组合报告摘要中获得。