新药研发流程 | 糖尿病知识中心

一、药物发现1. Drug Discovery Discovery · 2–5 yr

靶点识别与验证Target identification & validation

找到致病的分子（受体、酶、信号通路），并证明"调节它能治病"。验证依靠人类遗传学、动物模型与人体生物学证据。糖尿病例：GLP-1 受体 —— 激动它可促进葡萄糖依赖性胰岛素分泌、抑制食欲。我们数据集中的"胰高血糖素受体拮抗 / FGF21 轴"目前正停留在这一步（临床前靶点验证）。

Identify a disease-driving molecule (receptor, enzyme, signaling pathway) and prove that modulating it treats the disease. Validation relies on human genetics, animal models, and human biology. Diabetes example: the GLP-1 receptor — agonizing it promotes glucose-dependent insulin secretion and suppresses appetite. The "glucagon-receptor antagonism / FGF21 axis" in our dataset is currently at this stage (preclinical target validation).

先导化合物：从 Hit 到 LeadFrom hit to lead compound

通过高通量筛选或结构设计找到能调节靶点的分子（hit），再优化为先导物（lead），兼顾活性、选择性与成药性。例：orforglipron 是口服小分子 GLP-1RA —— 难点正是把通常需注射的肽类靶点做成可口服的小分子。

Through high-throughput screening or structure-based design, find a molecule that modulates the target (a hit), then optimize it into a lead with balanced potency, selectivity, and druggability. Example: orforglipron is an oral small-molecule GLP-1RA — the challenge was turning a typically injectable peptide target into an orally available small molecule.

二、临床前研究2. Preclinical Research Preclinical · 1–2 yr

体外 + 体内药效：细胞实验与动物模型验证作用。
ADME / PK：吸收、分布、代谢、排泄 —— 药物在体内如何运行。
毒理学：通常两种动物，急性/慢性毒性、致癌致畸评估。
CMC（化学、生产与控制）：能否稳定、可规模化生产 —— 贯穿全程，常被低估。
产出 → 递交 IND（新药临床试验申请）：FDA 在 30 天内不反对即可进入人体试验。

In vitro + in vivo efficacy: cell assays and animal models confirm the effect.
ADME / PK: absorption, distribution, metabolism, excretion — how the drug moves through the body.
Toxicology: typically two species; acute/chronic toxicity, carcinogenicity and teratogenicity.
CMC (Chemistry, Manufacturing & Controls): can it be made stably and at scale — runs through the whole program and is often underestimated.
Output → file an IND: if the FDA does not object within 30 days, human trials can begin.

三、临床试验3. Clinical Trials Clinical · 6–7 yr

最耗时、最烧钱的主战场。分三期推进：

The most time- and cost-intensive battleground, advancing through three phases:

期别Phase	对象Subjects	规模Size	回答什么问题Question answered	糖尿病例Diabetes example
I	健康志愿者（或患者）Healthy volunteers (or patients)	20–100	安全吗？药代动力学如何？最大耐受剂量？Is it safe? What is the PK? Maximum tolerated dose?	—
II	目标患者Target patients	100–300	有效吗？最佳剂量是多少？Does it work? What is the optimal dose?	retatrutide II 期减重信号retatrutide Phase II weight-loss signal
III	大样本患者Large patient population	1,000–5,000+	对照标准疗法，确证疗效与安全性Confirm efficacy and safety vs standard of care	retatrutide TRANSCEND-T2D-1、orforglipron ATTAIN-1 / ACHIEVE-3、zimislecel FORWARDretatrutide TRANSCEND-T2D-1, orforglipron ATTAIN-1 / ACHIEVE-3, zimislecel FORWARD

关键：III 期与终点设计Key: Phase III and endpoint design

III 期通常是随机双盲对照试验（RCT），设定主终点决定成败。糖尿病常用终点：HbA1c 降幅、减重百分比、心肾结局。生物标志物也很关键：1 型糖尿病用 C-肽（β 细胞功能）与 TIR（血糖在目标范围时间）。例：zimislecel 主打"12 例中 10 例不再需要外源性胰岛素"——这是硬终点。

Phase III is usually a randomized, double-blind, controlled trial (RCT) whose primary endpoint decides success or failure. Common diabetes endpoints: HbA1c reduction, percent weight loss, cardio-renal outcomes. Biomarkers matter too: type 1 diabetes uses C-peptide (β-cell function) and TIR (time in range). Example: zimislecel's headline is "10 of 12 patients no longer need exogenous insulin" — a hard endpoint.

四、监管审批与上市后4. Regulatory Approval & Post-Marketing Approval · 1–2 yr +

递交 NDA / BLA（新药 / 生物制品申请）给 FDA、EMA 或中国 NMPA。
审评（标准约 10 个月；优先审评更快）。例：orforglipron 于 2026 年 4 月获 FDA 批准。
IV 期 / 上市后监测：更大真实世界人群，监测罕见或长期副作用。
数据集中的 Filed（CagriSema 已递交 NDA）、Approved（orforglipron、teplizumab）即处于这一段。

File an NDA / BLA (New Drug / Biologics License Application) with the FDA, EMA, or China's NMPA.
Review (standard ~10 months; priority review is faster). Example: orforglipron was FDA-approved in April 2026.
Phase IV / post-marketing surveillance: a larger real-world population to monitor rare or long-term side effects.
The Filed (CagriSema NDA submitted) and Approved (orforglipron, teplizumab) entries in our dataset sit at this stage.

加速通道Accelerated Pathways

满足"重大未满足医学需求"的药物可走加速通道压缩时间线：突破性疗法（Breakthrough）、加速批准（Accelerated Approval）、孤儿药（Orphan Drug）、优先审评（Priority Review）。例：teplizumab 延缓 1 型糖尿病发病即借助了创新监管路径。

Drugs addressing a serious unmet medical need can use accelerated pathways to compress the timeline: Breakthrough Therapy, Accelerated Approval, Orphan Drug, Priority Review. Example: teplizumab, which delays type 1 diabetes onset, benefited from innovative regulatory pathways.

为什么 90% 会失败Why ~90% Fail

疗效不足：III 期未达主终点，或不优于现有标准疗法（最常见）。
安全性 / 毒性：出现不可接受的副作用或安全信号。
药代 / 成药性：口服生物利用度、半衰期或制造问题。
商业 / 战略：市场、报销或竞争格局变化导致项目下马。

Insufficient efficacy: failing the Phase III primary endpoint, or not beating standard of care (most common).
Safety / toxicity: unacceptable side effects or safety signals.
PK / druggability: oral bioavailability, half-life, or manufacturing problems.
Commercial / strategic: shifts in market, reimbursement, or competition shelve the program.

延伸阅读Further reading

结合本站 药物研发参考（靶点与管线）与 数据集（各资产的阶段与来源）一起看。Read together with this site's Drug Development reference (targets and pipeline) and the Dataset (each asset's stage and sources).