主题
FinnGen R13 · 糖尿病并发症 endpoint 清单(中英对照)
FinnGen R13 — Diabetic Complication Outcome Endpoints (bilingual)
Exposure 暴露: UKB-PPP cis-pQTL (N=34,557 discovery, European; 面板 2,923 蛋白,1,659 有 cis 工具) · Method 方法: two-sample MR, single-SNP Wald ratio Case/control counts from the official FinnGen R13 manifest; long names are the official FinnGen endpoint definitions. 病例/对照数取自官方 FinnGen R13 manifest;长名为 FinnGen 官方 endpoint 定义。
1. Selection principle 选择原则
EN — R13 was run as a larger, updated replication of R9 (case counts are ~30–50% higher). Importantly, R13 restructured the diabetes-complication endpoint library: many R9 diabetes-specific complications no longer exist as DM_* endpoints in R13 (see §3). We therefore included only the diabetes-specific DM_* endpoints that R13 still defines, and did not substitute the missing ones with non-diabetes generic endpoints (to keep case definitions clean for publication). We additionally added R13-specific refinements: DM_RETINOPATHY_STRICT (stricter DR) and the AMD subtypes WET_AMD / DRY_AMD. Base diseases use external GWAS only (GCST90824163 for T1D, Mahajan2018-noUKBB for T2D); the FinnGen T1D_WIDE / T2D endpoints were retired on 2026-07-27 (see §2).
中文 — R13 作为 R9 的更大、更新的重复验证(病例数普遍高 30–50%)。关键:R13 重构了糖尿病并发症 endpoint 库——R9 里许多糖尿病专属并发症在 R13 中已不再作为 DM_* 终点存在(见 §3)。因此我们只纳入 R13 仍定义的糖尿病专属 DM_* 终点,未用非糖尿病的通用终点替代缺失项(以保证病例定义干净、便于发表)。另新增 R13 特有细化:DM_RETINOPATHY_STRICT(更严格的 DR)与 AMD 亚型 WET_AMD / DRY_AMD。基础病只用外部 GWAS(T1D 用 GCST90824163、T2D 用 Mahajan2018-noUKBB);FinnGen 的 T1D_WIDE / T2D 已于 2026-07-27 退役(见 §2)。
2. Endpoint inventory 结局清单
| Code (phenocode) | English name (official) | 中文名 | Category 类别 | Cases 病例 | Controls 对照 | Sig. 显著 |
|---|---|---|---|---|---|---|
| DM_RETINOPATHY_STRICT | Diabetic retinopathy, strict definition | 糖尿病视网膜病变(严格定义) | eye 眼 | 6,970 | 88,076 | 10 |
| DM_MACULOPATHY | Diabetic maculopathy | 糖尿病黄斑病变 | eye 眼 | 4,766 | 88,091 | 9 |
| DM_NEOVASCULAR_GLAUCOMA | Neovascular glaucoma | 新生血管性青光眼 | eye 眼 | 1,576 | 483,921 | 0 |
| H7_AMD | Age-related macular degeneration (dry or wet) | 老年黄斑变性 | eye 眼 | 13,947 | 458,803 | 33 |
| WET_AMD | Wet age-related macular degeneration | 湿性老年黄斑变性 | eye 眼 | 7,529 | 335,608 | 26 |
| DRY_AMD | Dry AMD (includes geographic atrophy) | 干性老年黄斑变性(含地图状萎缩) | eye 眼 | 9,533 | 333,701 | 25 |
| DM_NEUROPATHY | Diabetic neuropathy | 糖尿病神经病变 | neuro 神经 | 4,145 | 99,346 | 4 |
| DM_NEPHROPATHY_EXMORE | Diabetic nephropathy (more control exclusions) | 糖尿病肾病 | renal 肾 | 6,043 | 62,519 | 6 |
| GCST90824163 | Type 1 diabetes (external GWAS) | 1型糖尿病(外部GWAS) | base 基础病 | 20,355 | 797,363 | 57 |
| Mahajan2018 noUKBB | Type 2 diabetes (DIAMANTE European, excl. UKBB) | 2型糖尿病(外部GWAS) | base 基础病 | 55,005 | 400,308 | 28 |
Sig. = number of proteins with FDR < 0.05 in the single-SNP Wald screen(2026-07-27 全量重跑后的实测值)。 显著 = 单SNP Wald 筛查中 FDR < 0.05 的蛋白数(数值取自
results/screen_R13_*/,2026-07-27 重跑)。
不在主筛之列的 endpoint(备用 / 已退役)
| Code | 状态 | 原因 |
|---|---|---|
DM_RETINOPATHY_EXMORE | 备用(R13_reserve) | DR 主结局用更干净的 DM_RETINOPATHY_STRICT;EXMORE 保留作敏感性分析,不进主 FDR 分组 |
T1D_WIDE / T2D(FinnGen) | 已退役(R13_retired) | 基础病一律用外部 GWAS(见下),FinnGen 版定义宽、且与并发症结局同源人群会引入样本重叠 |
GCST90475661 | 已退役 | 经标志位点核验实为 T2D 主导(TCF7L2 p=9.5×10⁻⁶⁶、HLA 仅 1.28 倍),非纯 T1D,详见外部 T1D 数据源问题 |
IAMDGC_lateAMD | 外部复制专用 | 不进 FinnGen 主筛,由 17_replicate_external.R 单独跑 |
原始数据一律保留,退役项只是移出主筛清单(release 列改为 R13_retired / R13_reserve)。
对照口径(Risteys 官方定义,2026-07-27 逐条核对)
发表时这三条必须写进方法学,因为它们决定了效应量的解释:
| Endpoint | 对照是谁 | 关键含义 |
|---|---|---|
DM_RETINOPATHY_STRICT | 糖尿病人群内(对照须有 DIABETES_FG),排除 DM_MACULOPATHY / DM_RETINOPATHY | 估计的是「糖尿病人里谁会得 DR」,已扣掉糖尿病本身的效应 |
DM_NEPHROPATHY_EXMORE | 糖尿病人群内(对照须有 DIABETES_FG),排除 DM_COMPLICATIONS | 同上;与 STRICT 的差别只在排除范围,不在对照总体 |
DM_NEOVASCULAR_GLAUCOMA | 一般人群(不要求 DIABETES_FG,仅排除 DM_RETINOPATHY) | ⚠ 与前两个不是同一口径,对照数 483,921 即由此而来;该终点本课题 0 个显著蛋白,我们保留并披露,不因结果为阴性而事后删除 |
3. Endpoints present in R9 but NOT available in R13 · R9 有而 R13 无的并发症
EN — R13 removed the following diabetes-specific complication endpoints that existed in R9. R13 only offers non-diabetes-specific generic equivalents (e.g. H7_RETIVASCOCCLUSION, H7_RETINAHAEMORR), which we did not use because they change the phenotype from "diabetic complication" to "all-cause".
中文 — R13 删除了以下 R9 中存在的糖尿病专属并发症终点。R13 只提供非糖尿病特异的通用等价物(如 H7_RETIVASCOCCLUSION、H7_RETINAHAEMORR),我们未采用,因为它们会把表型从"糖尿病并发症"变成"全因"。
| R9 endpoint | 中文 | R13 status |
|---|---|---|
| DM_RETINA_PROLIF | 增殖性糖尿病视网膜病变 | removed 已删(仅有 H7_PROLIFRETINOPATHOTH 全因) |
| DM_RETINA_NOS | 未特指糖尿病视网膜病变 | removed 已删 |
| DM_RET_VEIN_OCCLU | 视网膜静脉阻塞 | removed 已删(仅 H7_RETIVASCOCCLUSION 全因) |
| DM_RET_ART_OCCLU_MAIN / _BRANCH | 视网膜动脉阻塞 | removed 已删 |
| DM_VITREOUS_BLEEDING | 玻璃体积血 | removed 已删(仅 H7_RETINAHAEMORR 全因) |
| DM_AUTONOMIC / DM_MONONEURO / DM_POLYNEURO | 神经病变亚型 | removed 已删 |
| DM_PERIPH_ANGIOPATHY | 外周血管病变 | removed 已删 |
| DM_KETOACIDOSIS | 酮症酸中毒 | removed 已删 |
| DM_HYPOGLYC | 低血糖 | removed 已删 |
| DM_SEVERAL_COMPLICATIONS | 多并发症合并 | removed 已删 |
⇒ The full complication panel exists only in R9; R13 serves as a higher-powered replication of the endpoints it retains, plus AMD subtypes. ⇒ 完整并发症谱只在 R9;R13 作为其保留终点的更高功效重复验证,另加 AMD 亚型。
4. Definition notes 定义说明
_EXMORE= "more control exclusions":controls exclude related diseases (cleaner contrast). 对照排除相关疾病。_STRICT(R13 new 新增):stricter DR case definition (fewer, better-defined cases). 更严格的 DR 病例定义(病例更少但更明确)。T1D_WIDE/T2D"combined/wide":broad register-based definitions → comparatively heterogeneous; external GCST/Mahajan remain the primary T1D/T2D reference. 宽/合并的登记定义 → 相对异质;外部 GCST/Mahajan 仍为 T1D/T2D 主参考。WET_AMD/DRY_AMD:AMD subtypes; DRY includes geographic atrophy. AMD 亚型;干性含地图状萎缩。
5. Cross-version note on T1D/T2D 关于 T1D/T2D 的跨版本提示
At near-identical case counts, FinnGen T2D yields far more significant proteins than the external DIAMANTE (R13: T2D 103 vs Mahajan 28; T1D_WIDE 43 vs 外部 T1D 21(注:2026-07-27 外部 T1D 已换源为 GCST90824163,显著数变为 57,见「外部T1D数据源问题」页)), consistent with FinnGen's broad register-based definitions inflating associations. Use Mahajan/GCST as the primary T1D/T2D result; treat FinnGen T1D/T2D as sensitivity only.
在病例数接近时,FinnGen T2D 的显著蛋白远多于外部 DIAMANTE(R13:T2D 103 vs Mahajan 28;T1D_WIDE 43 vs 外部 T1D 21(注:2026-07-27 外部 T1D 已换源为 GCST90824163,显著数变为 57,见「外部T1D数据源问题」页)),与 FinnGen 宽登记定义导致关联"虚高"一致。主结果用 Mahajan/GCST,FinnGen T1D/T2D 仅作敏感性。
6. Source & reproducibility 来源与可复现
- Endpoint definitions / full ICD codes 完整 ICD 定义: FinnGen Risteys R13 —
https://r13.risteys.finngen.fi/endpoints/<CODE> - Summary statistics 汇总统计:
https://storage.googleapis.com/finngen-public-data-r13/summary_stats/finngen_R13_<CODE>.gz - Case/control counts 病例数来源: FinnGen R13 official manifest (
finngen_R13_manifest.tsv) - Config 配置:
outcome_manifest.csv(release=R13) · Pipeline:analysis/01_screen.R R13