几项对理赔数据库的回顾性分析发现，胰岛素使用与癌症发生（CA）有关，二甲双胍可减少癌症发生。Louise Bordeleau博士在美国糖尿病学会2013年会上介绍了ORIGIN的研究结果，发现二甲双胍和甘精胰岛素对受试者的CA风险没有影响。
ORIGIN研究是一项国际性双盲随机2*2析因试验，对比在心血管（CV）风险较高的糖尿病前期或糖尿病（DM）患者中，甘精胰岛素（来得时）或标准治疗与omega-3脂肪酸或安慰剂的疗效。CA亚组研究的主要指标为随机分配前的CA发生率和死亡率。研究人员还探索了血糖控制、降血糖疗法和CA结局的联系。研究预先收集了受试者的CA事件数据，所有CA结局均经裁决。研究采用分层时序检验法（stratified log-rank test）比较各组的事件发生时间（Time-to-event）曲线，并利用Cox回归模型计算风险比，因子分配、基线糖尿病状态和随机化前CV病史作为协变量。
中位随访6.2年后，招募的12,537名受试者（平均年龄63.5，SD 7.8；4,388名女性）中有953名（7.6%）发生CA事件（甘精胰岛素组和标准治疗组中均为1.32/100人年），这部分人群年龄偏大，且吸烟、酗酒、CV病史、新发DM和他汀或阿司匹林使用频率较高。标准治疗组和甘精胰岛素组的未调整CA死亡率为0.54/100人年和0.51/100人年。两组间CA特异性结局（肺癌、乳腺癌、前列腺癌、结肠癌和黑色素瘤）没有显著差异。关键临床亚组的调整后模型仍为中性（交互作用P值均> 0.17）。分配各组的新发和复发CA风险相近。糖化血红蛋白水平、包括研究中二甲双胍治疗（持续时间和剂量）在内的降血糖疗法和身体质量指数对CA事件风险没有影响。
本项研究中，甘精胰岛素对总体和CA特异性结局（包括CA特异性死亡率）的效果为中性。研究过程中的二甲双胍暴露和糖化血红蛋白水平不会改变CA风险。
原文：
Some retrospective analyses of claim databaseshave linked insulin use to incident cancers (CA) and metformin to reduced CA.
The ORIGIN trial was an international double-blind randomized 2-by-2 factorial trial of insulin glargine (Lantus) or standard care and omega-3 fatty acids or placebo in people with either pre-diabetes or diabetes (DM) at high CV risk. The primary outcomes of the CA sub-study were CA incidence and CA mortality by randomization allocation. We also studied the associations between glycemic control glucose lowering therapies and CA outcomes. CA events were prospectively collected and all CA outcomes were adjudicated. Time-to-event curves were compared using stratified log-rank tests; and hazard ratios were calculated with Cox regression models with factorial allocation baseline diabetes status and history of CV event before randomization as covariates.
Among the 12 537 participants (mean age 63.5 yrs SD 7.8; 4388=female) 953 (7.6%) developed a CA event during the median follow-up of 6.2 years (1.32/100 person-years both in the glargine and standard groups). Participants with a CA event were older, had a higher frequency of smoking, alcohol intake, previous CV event, new diagnosis of DM and statin or aspirin use. The unadjusted CA death rates per 100 person-years were 0.54 (standard) and 0.51 (glargine). CA specific outcomes (lung, breast, prostate, colon CAs and melanoma) were not different between groups. Adjusted models for key clinical subgroups remained neutral (interaction P values all > 0.17). The risk of new and recurrent CA was similar between allocation groups. HbA1c level glucose lowering therapies including treatment with metformin (duration and dose) during study and body-mass index did not modulate risk of CA event. In this trial insulin glargine had a neutral effect on overall and CA specific outcomes including CA-specific mortality. Exposure to metformin and HbA1c levels during study did not alter CA risk.
