Sepsis results from the dysregulated host response to bacterial, viral, or fungal infections. Recent data suggest that the 28-day outcome of sepsis patients depends on the features of the immune reaction of the host. Four to 10% of patients with septic shock present an entirely pro-inflammatory course with a high risk for early death and features mimicking the macrophage activation syndrome (MALS)—that is, liver dysfunction, disseminated intravascular coagulation, and increases of blood ferritin and triglycerides. Almost 30% of patients with sepsis present with immunoparalysis, that is, failure of circulating mononuclear cells for production of cytokines upon ex vivo stimulation and defective antigen presentation, often leading to T cell exhaustion. However, the immune function of the majority lies at an intermediate state between MALS and immunoparalysis [1]. It may well be the case that this immune classification guides long-term outcomes of sepsis—a condition often called chronic critical illness (CCI) or persistent inflammation, immunosuppression, and catabolism syndrome (PICS). There are several host factors that impact the progression into CCI or PICS. The most important are age, the presence of cachexia, and the site of infection of the primary sepsis episode. These patients have profound disabilities associated with lymphopenia and increased expression of PD-1, expansion of myeloid-derived stem cells and T regulatory cells and dysfunctional erythropoiesis [2]. It is suggested that the 1-year mortality of sepsis survivors approaches 40%, and this may be associated with persisting immune dysregulations.

The article by Mageau et al. in this issue of the Journal of Internal Medicine breaks the boundaries and goes well beyond the traditional approach of long-term mortality and disabilities of sepsis survivors. Taking into consideration the complex immune dysregulation of sepsis, the authors ask themselves whether the original immune phenomena may lead to a predisposition for immune-mediated inflammatory diseases (IMIDs) among sepsis survivors. For their analysis, they used the French PMSI database, collecting information between January and November 2020 for 460,708 index cases of sepsis. They study as comparators 62,258 survivors from acute myocardial infarction (AMI). To avoid confounding bias, 62,257 sepsis survivors were selected to match with 62,257 survivors from AMI using the matching criteria of age, sex, active cancer, active malignant hematologic condition, HIV infection, and history of organ transplantation. Survivors of sepsis and AMI were followed up for 9 months, and all new diagnoses of IMIDs were recorded. Results were striking, as the incidence of IMIDs was 2.80 times higher in sepsis survivors than in AMI survivors. The time incidence for IMIDs started to increase after 16 days from the original sepsis episode. The hazard ratio was greater for immune thrombocytopenia, followed by Sjögren's syndrome, autoimmune hemolytic anemia, systemic lupus erythematosus, and ANCA-associated vasculitis. In a series of sensitivity analyses, the authors proved that the incidence of IMIDs among sepsis survivors remained high irrespective of whether the sepsis etiology was bacterial or viral and irrespective of the patient's comorbidities [3].

There are several recent publications on the long-term sequelae of sepsis survivors. The majority focus on cardiovascular events—mainly a composite of AMI, stroke, and cardiovascular death—and they report the 5-year follow-up period from the index sepsis case. In these publications, patients with preexisting cardiovascular diseases (CVDs) are excluded from the analysis. One analysis from Ontario between April 2007 and January 2017 followed-up 254,251 hospitalized patients for sepsis and matched them with a similar number of patients hospitalized for other reasons. Sepsis hospitalization was associated with a 1.30 hazard ratio for a greater incidence of CVDs. Interestingly, this was increased to 1.66 among patients aged 40 years or less [4]. This study confirms the results of one meta-analysis of 27 studies published between 2005 and 2017, of which 7 publications were population-based and the rest of the publications were based on hospitalized patients. In a retrospective analysis of almost 2.25 million adult hospital survivors from the OptumLabs Data Warehouse, the hazard ratio of new hospitalization for coronary heart disease, heart failure, stroke, and arrhythmia among sepsis survivors was 1.26, 1.51, 1.35, and 1.45, respectively [5].

At first glance, it is somehow difficult to find an association between the observed increase in the incidence of IMIDs and CVDs among sepsis survivors. As noted above, one of the recently described sepsis immunotypes is MALS. MALS is triggered by the excess production of interleukin (IL)-1β by the tissue macrophages [1]. Priming for the production of IL-1β makes patients prone to CVDs. This is supported by the results of the CANTOS trial and the results of the LoDoCo2 investigators. In the CANTOS trial, survivors from a first episode of AMI and blood high-sensitivity C-reactive protein above 2 mg/L were randomized to 5-year treatment with placebo or canakimumab once every 3 months. Canakinumab is one monoclonal antibody which targets IL-1β. Results showed a 15% decrease in the relative risk for major CVD events, including AMI, stroke, and cardiovascular death [6]. The LoDoCo2 investigator consortium randomized 5478 patients with evidence of coronary heart disease into 5-year treatment with placebo or 500 mg colchicine once daily. The primary endpoint was the incidence of an acute CVD event defined as a composite of cardiovascular death, AMI, ischemic stroke, or ischemia-driven coronary revascularization. The incidence of this endpoint was significantly decreased by 31% with colchicine treatment [7]. Both studies prove the importance of IL-1β to drive CVD complications because canakinumab directly blocks the activity of IL-1β and colchicine inhibits the formation of the NLRP3 inflammasome, which splits the proactive intracellular pro-IL-1β into the active IL-1β.

In a prospective study of our group, we monitored the over-time course of high-mobility group 1 (HMGB1) in patients with sepsis. We split patients into those with an early peak of HMGB1—notably before the first 7 days—and those with a late peak of HMGB1. HMGB1 is a danger-associated molecular pattern: It is released from dying cells, and it stimulates the production of IL-1β after binding to Toll-like receptor (TLR)-4. Indeed, patients with a late peak of HMGB1 had greater circulating levels of ferritin and of interferon-gamma, which is an indirect signature of TLR-4 activation. In patients with the late peak of HMGB1, the hazard for 28-day mortality was increased only if they had a medical history of cardiovascular comorbidities such as type 2 diabetes mellitus, congestive heart failure, or chronic renal disease [8]. The results suggest synergy between IL-1β primed by sepsis and cardiovascular comorbidities for unfavorable outcome. This synergy may also explain why the incident events of CVDs and of IMIDs arrive the first year after sepsis, when the inflammatory signature of sepsis on the organism of the host most likely persists.

Rehospitalization for infections is another major sequel of sepsis. An analysis of 159,864 sepsis patients from Germany revealed that 45.0% of survivors were rehospitalized for infection, whereas 56.6% received antibiotics as outpatients. The risk of infections post-sepsis was increased by 9.6% compared to the pre-sepsis period [9]. This predisposition to infections may well be explained by the immunoparalysis immunotype of the original sepsis episode, which often persists during the first year after sepsis [2].

The significance of the type of immune dysregulation for the future progression into CCI comes from the analysis of the SAVE-MORE randomized controlled trial. Patients with severe COVID-19 at risk for progression into severe respiratory failure due to early activation of the IL-1 cascade indicated by increased blood levels of the biomarker suPAR (soluble urokinase plasminogen activator receptor) were randomized to treatment with standard-of-care and placebo or anakinra. Anakinra is the recombinant antagonist of the IL-1 receptor and blocks the activity of both IL-1β and IL-1α. Early anakinra treatment prevented the progression into post-acute COVID syndrome [10].

The study by Mageau et al. [3] sheds light on another aspect of long-term sepsis consequences, namely, IMIDs. It is likely that IMIDs and CVDs have similar immune backgrounds greatly influenced by the type of immune dysregulation of the original sepsis episode. With this point of view, precision immunotherapy becomes a necessity because it may influence not only the 28-day outcome but also the incidence of long-term sepsis consequences. The study by Mageau et al. [3] is the first to elaborate on this necessity, and the authors are to be congratulated for this ground-breaking contribution.

败血症是由宿主对细菌、病毒或真菌感染的反应失调引起的。最近的数据表明，脓毒症患者的 28 天结果取决于宿主的免疫反应特征。4% 至 10% 的感染性休克患者表现出完全促炎症的病程，早期死亡的风险很高，其特征类似于巨噬细胞活化综合征 (MALS)，即肝功能障碍、弥散性血管内凝血以及血铁蛋白和血中铁蛋白升高。甘油三酯。近 30% 的脓毒症患者出现免疫麻痹，即循环单核细胞在离体刺激和抗原呈递缺陷时无法产生细胞因子，通常导致 T 细胞耗竭。然而，大多数人的免疫功能处于MALS和免疫麻痹之间的中间状态[ 1 ]。这种免疫分类很可能指导脓毒症的长期结果——这种疾病通常被称为慢性危重病（CCI）或持续性炎症、免疫抑制和分解代谢综合征（PICS）。有几个宿主因素会影响 CCI 或 PICS 的进展。最重要的是年龄、恶病质的存在以及原发性败血症发作的感染部位。这些患者患有与淋巴细胞减少、PD-1 表达增加、骨髓源性干细胞和 T 调节细胞扩增以及红细胞生成功能障碍相关的严重残疾 [ 2 ]。研究表明，脓毒症幸存者的 1 年死亡率接近 40%，这可能与持续的免疫失调有关。

Mageau 等人的文章。本期《内科杂志》打破了界限，远远超出了脓毒症幸存者长期死亡率和残疾的传统方法。考虑到脓毒症复杂的免疫失调，作者自问，原始的免疫现象是否可能导致脓毒症幸存者易患免疫介导的炎症性疾病（IMID）。为了进行分析，他们使用了法国 PMSI 数据库，收集了 2020 年 1 月至 11 月期间 460,708 例败血症指标病例的信息。他们对 62,258 名急性心肌梗死 (AMI) 幸存者进行了研究。为了避免混淆偏倚，使用年龄、性别、活动性癌症、活动性恶性血液病、HIV 感染和器官移植史的匹配标准，选择了 62,257 名脓毒症幸存者与 62,257 名 AMI 幸存者进行匹配。对败血症和 AMI 的幸存者进行了 9 个月的随访，并记录了所有新诊断的 IMID。结果令人震惊，败血症幸存者的 IMID 发生率比 AMI 幸存者高 2.80 倍。IMID 的发病时间在最初脓毒症发作 16 天后开始增加。免疫性血小板减少症的风险比更大，其次是干燥综合征、自身免疫性溶血性贫血、系统性红斑狼疮和 ANCA 相关血管炎。在一系列敏感性分析中，作者证明，无论脓毒症病因是细菌还是病毒，也无论患者是否有合并症，脓毒症幸存者中 IMID 的发病率仍然很高[ 3 ]。

最近有几篇关于脓毒症幸存者长期后遗症的出版物。大多数关注心血管事件——主要是 AMI、中风和心血管死亡的复合事件——并且他们报告了脓毒症病例的 5 年随访期。在这些出版物中，先前患有心血管疾病 (CVD) 的患者被排除在分析之外。安大略省 2007 年 4 月至 2017 年 1 月期间进行的一项分析对 254,251 名因脓毒症住院的患者进行了随访，并将他们与因其他原因住院的类似数量的患者进行了匹配。脓毒症住院与 CVD 发病率较高的风险比为 1.30 相关。有趣的是，在 40 岁或以下的患者中，这一比例增加至 1.66 [ 4 ]。这项研究证实了对 2005 年至 2017 年间发表的 27 项研究进行荟萃分析的结果，其中 7 项出版物以人群为基础，其余出版物以住院患者为基础。OptumLabs 数据仓库对近 225 万成年医院幸存者进行的回顾性分析显示，脓毒症幸存者因冠心病、心力衰竭、中风和心律失常新住院的风险比分别为 1.26、1.51、1.35 和 1.45。5 ]。

乍一看，在脓毒症幸存者中观察到的 IMID 和 CVD 发病率增加之间很难找到关联。如上所述，最近描述的败血症免疫型之一是 MALS。MALS 是由组织巨噬细胞过量产生白细胞介素 (IL)-1β 引发的[ 1 ]。IL-1β 的产生使患者容易患 CVD。CANTOS 试验的结果和 LoDoCo2 研究人员的结果支持了这一点。在 CANTOS 试验中，首次 AMI 发作且血液高敏 C 反应蛋白高于 2 mg/L 的幸存者被随机接受为期 5 年的安慰剂或卡纳基单抗治疗，每 3 个月一次。Canakinumab 是一种靶向 IL-1β 的单克隆抗体。结果显示，主要 CVD 事件的相对风险降低了 15%，包括 AMI、中风和心血管死亡 [ 6 ]。LoDoCo2 研究人员联盟将 5478 名有冠心病证据的患者随机分组，接受安慰剂或每天一次 500 毫克秋水仙碱治疗，为期 5 年。主要终点是急性 CVD 事件的发生率，定义为心血管死亡、AMI、缺血性中风或缺血驱动的冠状动脉血运重建的复合事件。秋水仙碱治疗后该终点的发生率显着降低了 31% [ 7 ]。这两项研究都证明了 IL-1β 在驱动 CVD 并发症中的重要性，因为卡那奴单抗直接阻断 IL-1β 的活性，而秋水仙碱抑制 NLRP3 炎症小体的形成，该炎症小体将主动细胞内前 IL-1β 分裂为活性 IL-1β。

在我们小组的一项前瞻性研究中，我们监测了脓毒症患者高迁移率组 1 (HMGB1) 的随时间推移的病程。我们将患者分为 HMGB1 早期峰值（尤其是在前 7 天之前）和 HMGB1 晚期峰值的患者。HMGB1 是一种与危险相关的分子模式：它从垂死的细胞中释放出来，与 Toll 样受体 (TLR)-4 结合后刺激 IL-1β 的产生。事实上，HMGB1 峰值较晚的患者铁蛋白和干扰素-γ 的循环水平较高，这是 TLR-4 激活的间接特征。在 HMGB1 晚峰患者中，只有当他们有心血管合并症（如 2 型糖尿病、充血性心力衰竭或慢性肾病）病史时，28 天死亡率的风险才会增加 [8 ]。结果表明，脓毒症引发的 IL-1β 与心血管合并症之间存在协同作用，导致不良结果。这种协同作用也可以解释为什么 CVD 和 IMID 的事件发生在败血症后的第一年，此时宿主机体上败血症的炎症特征很可能持续存在。

感染再住院是败血症的另一个主要后遗症。对德国 159,864 名脓毒症患者的分析显示，45.0% 的幸存者因感染而再次住院，而 56.6% 的幸存者在门诊接受了抗生素治疗。与脓毒症前期相比，脓毒症后感染的风险增加了9.6%[ 9 ]。这种感染倾向很可能是由最初脓毒症发作的免疫麻痹免疫型来解释的，这种免疫麻痹免疫型通常在脓毒症后的第一年持续存在[ 2 ]。

免疫失调类型对于未来发展为 CCI 的重要性来自 SAVE-MORE 随机对照试验的分析。患有严重 COVID-19 的患者因生物标志物 suPAR（可溶性尿激酶纤溶酶原激活剂受体）血液水平升高而导致 IL-1 级联的早期激活而面临进展为严重呼吸衰竭的风险，这些患者被随机接受标准护理治疗和安慰剂或阿那白滞素。Anakinra 是 IL-1 受体的重组拮抗剂，可阻断 IL-1β 和 IL-1α 的活性。早期阿那白滞素治疗可防止进展为急性新冠肺炎后综合征[ 10 ]。

Mageau 等人的研究。[ 3 ]揭示了脓毒症长期后果的另一个方面，即IMID。IMID 和 CVD 很可能具有相似的免疫背景，很大程度上受到原始脓毒症发作的免疫失调类型的影响。从这个角度来看，精准免疫治疗变得必要，因为它不仅可能影响 28 天的结果，还可能影响长期脓毒症后果的发生率。Mageau 等人的研究。[ 3 ] 是第一个详细阐述这种必要性的人，值得祝贺作者的这一突破性贡献。