Data and code for: Toxicity-related treatment discontinuation and treatment-related mortality with intensive versus gemcitabine-based first-line chemotherapy in advanced/metastatic PDAC — a systematic review and meta-analysis

Data and code for: Toxicity-related treatment discontinuation and treatment-related mortality with intensive versus gemcitabine-based first-line chemotherapy in advanced and metastatic pancreatic ductal adenocarcinoma — a systematic review and meta-analysis of randomized controlled trials This repository contains the full extraction dataset, analysis code, and risk-of-bias assessments underlying the meta-analysis. It is provided to support reproducibility and in fulfilment of the journal Data Availability requirement. PROSPERO registration: CRD420261408001 Search date: inception to 25 February 2026 Included studies: 36 randomized controlled trials (14,839 randomized patients) License: CC BY 4.0 Contents File Description PDAC_MA_dataset_full.xlsx Master extraction workbook (all sheets): trial characteristics, per-arm safety extraction, analysis-ready 2×2 tables, risk-of-bias, R code, verbatim source quotes, and version/verification logs. This is the primary, authoritative data file. PDAC_MA_primary_outcomes_2x2.csv Analysis-ready 2×2 data for the co-primary outcomes: AE-related treatment discontinuation and treatment-related mortality (long format for metafor/meta). PDAC_MA_grade34_AEs_2x2.csv Analysis-ready 2×2 data for grade 3–4 adverse events and dose modifications. PDAC_trial_characteristics.csv One row per trial: design, phase, regimen class, N randomized, % metastatic, sponsorship, blinding. PDAC_RoB2_assessments.csv Cochrane RoB 2.0 judgments per trial × domain (D1–D5 + overall). PDAC_verbatim_extraction_quotes.csv Verbatim source-text quotes supporting each extracted discontinuation and mortality value. PDAC_MA_analysis.R R script reproducing all pooled estimates, subgroup analyses, sensitivity analyses, funnel plots, and forest plots. How to reproduce the analysis Install R (≥ 4.0) and the packages: meta, metafor, readxl, dplyr, ggplot2. Place PDAC_MA_dataset_full.xlsx in your working directory. Run PDAC_MA_analysis.R. Methods summary Pooled risk ratios were estimated with inverse-variance random-effects meta-analysis (DerSimonian–Laird τ² with the Knapp–Hartung–Sidik–Jonkman adjustment), a 0.5 continuity correction for single-zero studies, and exclusion of double-zero studies. Multi-arm trials sharing one gemcitabine control arm (GEST, GAMMA, ECOG E6201, GENERATE) had the shared control divided across comparisons. AViTA is excluded from the AE-related discontinuation pool because the source reports events, not patients, which is invalid for a patient-level risk ratio. Key results (for orientation) AE-related discontinuation: 24 comparisons, RR 1.34 (95% CI 1.16–1.55); Egger t=2.19, P=0.039. Treatment-related mortality: 21 comparisons, RR 1.16 (95% CI 0.95–1.41); Egger t=2.97, P=0.008. Citation If you use these data or code, please cite the associated article and this dataset. Authors: Ramy Yassa; Steven Danial Azmy Habib (contributed equally).本配套数据集与代码的研究主题为：晚期及转移性胰腺导管腺癌（pancreatic ductal adenocarcinoma, PDAC）中，强化化疗对比基于吉西他滨的一线化疗的毒性相关治疗中断及治疗相关死亡——一项随机对照试验（randomized controlled trial）的系统评价与荟萃分析（meta-analysis）。 本仓库包含本荟萃分析所依托的全部提取数据集、分析代码及偏倚风险（risk-of-bias）评估文件，旨在支持研究可重复性，并满足期刊对数据可用性的要求。 PROSPERO注册编号：CRD420261408001 检索日期：从建库至2026年2月25日 纳入研究：36项随机对照试验（共14839名随机入组患者） 许可协议：CC BY 4.0（知识共享署名4.0协议） ## 内容 | 文件 | 描述 | |------|------| | PDAC_MA_dataset_full.xlsx | 主提取工作簿（包含所有工作表）：试验特征、单臂安全性提取、可直接用于分析的2×2列联表、偏倚风险评估、R代码、原始文献逐字引用记录及版本/校验日志，为本研究的核心权威数据文件。 | | PDAC_MA_primary_outcomes_2x2.csv | 针对共同主要结局的可直接分析的2×2数据：不良反应相关治疗中断及治疗相关死亡（采用适配metafor/meta包的长格式数据）。 | | PDAC_MA_grade34_AEs_2x2.csv | 针对3~4级不良反应及剂量调整的可直接分析的2×2数据。 | | PDAC_trial_characteristics.csv | 单条记录对应一项试验，包含试验设计、分期、方案分类、随机入组样本量、转移性患者占比、资助方及设盲情况。 | | PDAC_RoB2_assessments.csv | 基于Cochrane RoB 2.0的各试验×领域（D1~D5 + 整体）的偏倚风险判断结果。 | | PDAC_verbatim_extraction_quotes.csv | 支撑每项提取的治疗中断与死亡数据的原始文献逐字引用记录。 | | PDAC_MA_analysis.R | 复现所有合并效应量、亚组分析、敏感性分析、漏斗图及森林图的R脚本。 | ## 分析复现步骤 1. 安装R（版本≥4.0）及以下依赖包：meta、metafor、readxl、dplyr、ggplot2。 2. 将PDAC_MA_dataset_full.xlsx置于您的工作目录中。 3. 运行PDAC_MA_analysis.R。 ## 方法概要 本研究采用逆方差随机效应荟萃分析（DerSimonian-Laird τ²结合Knapp-Hartung-Sidik-Jonkman校正）估算合并风险比，对含单零事件的研究采用0.5的连续性校正，并排除双零事件研究。对于共享一个吉西他滨对照组的多臂试验（GEST、GAMMA、ECOG E6201、GENERATE），将共享对照组拆分至各比较组中。由于AViTA研究仅报告事件数而非患者数，无法用于患者水平风险比的计算，因此将其排除在不良反应相关治疗中断的合并分析之外。 ## 关键结果（供参考） 不良反应相关治疗中断：共24组比较，风险比（RR）=1.34（95%置信区间[CI]：1.16~1.55）；Egger检验t=2.19，P=0.039。 治疗相关死亡：共21组比较，RR=1.16（95%CI：0.95~1.41）；Egger检验t=2.97，P=0.008。 ## 引用说明 若您使用本数据集或代码，请引用相关发表文章及本数据集。作者：Ramy Yassa；Steven Danial Azmy Habib（贡献均等）。