Therapeutic potential of IL6R blockade for the treatment of sepsis and sepsis-related deathA Mendelian randomisation study

Sepsis kills roughly eleven million people a year. That's more than most cancers and more than heart attacks in most age groups. For decades, every attempt to treat it by targeting the immune system has failed — trial after trial, drug after drug. Now, a team of British geneticists believes they finally know why those trials kept missing the mark. They have identified a drug that's already available on pharmacy shelves that might work. So, what is sepsis? It's not just a severe infection. It's what happens when the body's response to infection spirals out of control. The immune system floods the organs with inflammatory signals until those organs start to fail. At the center of this story is interleukin-six, or IL-6. This cytokine, a chemical messenger, acts like a fire alarm that won't turn off during severe infection. The louder the alarm gets, the worse the outcome. The clue that IL-6 might be the right target didn't come from a sepsis ward; it came from a pandemic. When SARS-CoV-2 emerged, the sickest COVID-19 patients presented like fast-moving sepsis. They exhibited a hyperinflammatory state where IL-6 was central. Researchers observed that individuals who carried certain natural genetic variants in the IL6R gene, which encodes the IL-6 receptor, had lower odds of becoming critically ill with COVID-19. These variants dampen IL-6 signaling, essentially mimicking what the drug tocilizumab does. Tocilizumab is a monoclonal antibody that blocks the IL-6 receptor. Genetic analysis predicted this benefit, and randomized trials later confirmed it. Tocilizumab and a related drug, sarilumab, reduced mortality in critically ill COVID-19 patients and are now considered standard treatment for severe disease. This process—finding a genetic signal first and then confirming it with trials—is what Hamilton and colleagues had in mind when they applied the same method to sepsis. Their reasoning was straightforward: severe COVID-19 and bacterial sepsis share the same immunological signature. If blocking IL-6 helped in one case, it should help in the other. To test this before committing to an expensive trial, they employed a method called Mendelian randomization. The logic behind Mendelian randomization is elegant. At conception, people are randomly assigned genetic variants. Some individuals carry variants near the IL6R gene that reduce IL-6 signaling throughout their lives. Essentially, these individuals are running a lifelong low-dose version of the drug. By comparing their outcomes against those without those variants, researchers can estimate what IL-6 receptor blockade does without the confounders that make observational studies unreliable. Hamilton and colleagues built their instrument from twenty-six independent genetic variants located within 300 kilobases of IL6R, weighting each by its effect on high-sensitivity C-reactive protein, or hsCRP—a downstream marker of IL-6 activity. The minimum F-statistic among the included variants was thirty-one point one, indicating that the instrument was strong. Their primary dataset was the UK Biobank, which includes nearly four hundred eighty-six thousand participants, among which eleven thousand six hundred forty-three had sepsis. The headline result shows that genetically proxied IL-6 receptor blockade was associated with a twenty percent lower odds of sepsis overall, resulting in an odds ratio of zero point eighty with a ninety-five percent confidence interval of zero point sixty-six to zero point ninety-six. This alone is a meaningful signal. However, the real story lies in what happens as disease severity increases. For sepsis that requires admission to critical care, the odds ratio dropped to zero point forty-eight—about half the risk. For death within twenty-eight days of critical care admission with sepsis, the odds ratio was zero point thirty-seven. This indicates a sixty-three percent reduction in the odds of dying in the most severe scenario. While the confidence intervals are wider at these extremes due to smaller patient numbers in the critical care analyses, the pattern remains consistent and striking. The sicker the patient, the larger the protective signal. That type of dose-response gradient across severity is exactly what you would want to see if a biological mechanism is genuine. This pattern extends beyond sepsis as a category. For pneumonia requiring critical care, the odds ratio was zero point sixty-nine, showing that the severity gradient appears there as well. Replication in FinnGen, the Finnish biobank, yielded mixed results. The combined FinnGen sepsis estimate was close to the null at zero point ninety-eight. However, more specific phenotypes—streptococcal sepsis and sepsis due to pneumonia—showed point estimates similar to UK Biobank, albeit with wide confidence intervals. A meta-analysis across both datasets brought the summary sepsis estimate to zero point eighty-six with a confidence interval of zero point seventy-four to zero point ninety-nine. When focusing on definitions of respiratory or streptococcal sepsis, the estimates ranged roughly from zero point seventy-nine to zero point eighty. While the replication isn't perfect, it aligns with the main finding's direction. Hamilton and colleagues also conducted a significant comparison using a C-reactive protein instrument—four genetic variants near the CRP gene itself—as a parallel probe. This analysis showed an odds ratio for sepsis of zero point ninety-one and for sepsis-related death of zero point seventy-two, which is consistent with the main IL6R results. A third instrument, based on genetic variation in gp130, a co-receptor in the IL-6 signaling pathway, showed no clear evidence of protection, resulting in an odds ratio of zero point ninety-seven in the UK Biobank. This contrast is important and suggests that the protective signal is specific to the IL-6 receptor axis, rather than a general dampening of inflammation. Multiple sensitivity analyses—including MR-Egger, weighted median, leave-one-out, and outlier exclusion—did not materially change the interpretation of the results. Now, for the catch. The same genetic variants that reduced sepsis severity also increased the odds of contracting certain infections in the first place. These include upper respiratory tract infections with an odds ratio of one point sixty-seven, urinary tract infections at one point thirty-two, cellulitis at one point twenty, and cholecystitis at one point fifty-three. When IL-6 signaling is dampened, the immune system becomes less aggressive. This is protective when the immune system is overreacting to an established infection. However, it may leave the body less equipped to combat an initial pathogen effectively. The paper describes this profile as "double-edged" — providing protection from severe inflammatory disease while increasing susceptibility to some infections. This insight about drug use is critical: tocilizumab may be reasonable as a rescue treatment in established severe sepsis, but not as prophylaxis. Importantly, nosocomial, or hospital-acquired, infection rates in the COVID-19 IL-6 receptor antagonist trials were below one percent in both the REMAP-CAP and RECOVERY trials. This is reassuring yet still emphasizes the need for careful patient selection and trial design. The comparison with COVID-19 runs throughout this paper, and for valid reasons. The genetically supported protective effect in severe COVID-19 yielded an odds ratio of zero point sixty-nine. The effect on severe sepsis, particularly in critical care admissions, falls within the same range. Hamilton and colleagues emphasize that Mendelian randomization predicted a benefit from IL-6 inhibition in COVID-19, and subsequent randomized trials matched those estimates. If this sequence repeats — a genetic signal followed by trial confirmation — there is potential for a proven treatment for one of medicine's most lethal conditions. However, some uncertainties remain. Mendelian randomization has its limits. Possible pleiotropy, where genetic variants affect outcomes through pathways other than IL-6, cannot be entirely excluded. The mortality estimates carry wide confidence intervals, and translating a lifelong genetic effect into an acute drug intervention is not straightforward. Significant questions about trial design remain unresolved: which patients, at what point in their illness, at what dose, and how the drug interacts with corticosteroids— which were co-administered in the COVID-19 trials and showed some evidence of interaction with IL-6 receptor antagonists. A planned pediatric sepsis trial was halted due to a lack of funding, highlighting how much depends on the research community's ability to secure resources to answer these critical questions. The conclusion reached by Hamilton and colleagues is clear. The genetic evidence supports conducting randomized controlled trials of IL-6 receptor antagonists in sepsis. The COVID-19 story serves as proof of concept: the genetic method was effective, the biology was correct, and the drug performed well. The pressing question is whether sepsis can follow the same path—and whether that path will receive funding before another eleven million people die waiting. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.