Statisticians Weigh In: Is the LI60 Clotting Test Really a Predictor of Bleeding Stroke?

When a simple blood test promises to reveal which patients with a brain hemorrhage are about to get worse, clinicians pay attention. Spontaneous intracerebral hemorrhage, a stroke caused by a ruptured blood vessel bleeding directly into brain tissue, affects millions of people worldwide each year, and one of its most feared complications is hematoma expansion: the original bleed continues to grow in the first hours after the patient arrives at the hospital. That growth is strongly linked to worse outcomes and higher mortality, so researchers have spent decades hunting for reliable predictors. Among the candidates are point-of-care viscoelastic tests such as rotational thromboelastometry, which measures the clotting behavior of whole blood in real time rather than testing isolated clotting factors in plasma.

A recent study by Forlivesi and colleagues, published in Neurocritical Care, examined whether rotational thromboelastometry parameters recorded early after spontaneous intracerebral hemorrhage are associated with subsequent hematoma expansion. The work drew interest because it connected a rapidly available bedside coagulation measurement to a clinically decisive event. If a parameter could reliably flag patients at high risk of expansion, it might help clinicians decide who should receive aggressive hemostatic therapy, closer monitoring, or rapid transport to specialized neurointensive care, potentially reshaping the early management of a disease where minutes matter.

Now a team of Japanese anesthesiologists and intensivists has raised a pointed question about that finding. In a letter to the editor published in the same journal, Shingo Kawashima, Atsushi Kobayashi, Wakana Kawashima, Masahiko Ohashi, and Hiroyuki Kinoshita, affiliated with Hamamatsu University School of Medicine, Seirei Mikatahara General Hospital, Hamamatsu Medical Center, and Tokushima University, ask whether the reported association between the parameter known as LI60 and hematoma expansion reflects genuine biology or is instead a statistical artifact. The title of their letter, LI60 and Hematoma Expansion in Intracerebral Hemorrhage: Statistical Artefact or Novel Biology?, frames the debate precisely. The authors of the original study were offered an opportunity to reply but did not respond, leaving the question open in the published record.

To understand the concern, it helps to know what LI60 actually measures. Rotational thromboelastometry assesses how a blood sample clots under controlled shear conditions, tracing the firmness of the clot over time. Parameters derived from this tracing capture different stages of hemostasis: the speed of clot initiation, the kinetics of clot strengthening, and the stability or lysis of the clot at later time points. LI60 belongs to the last category, reflecting the integrity of the clot sixty minutes after clotting has started. A low LI60 value suggests that the clot is breaking down, a pattern consistent with enhanced fibrinolysis, the process by which the body dissolves clots. In the setting of intracerebral hemorrhage, enhanced fibrinolysis at the bleeding site is a plausible biological mechanism for ongoing bleeding, which is precisely why the parameter attracted attention as a potential predictor of expansion.

The Japanese letter writers bring substantial experience with this technology to the discussion. Members of the group previously developed a four-group classification based on fibrinogen levels and fibrin polymerization that was associated with postoperative bleeding in cardiac surgery, and they have also studied how fibrin-based rotational thromboelastometry assessments influence blood transfusion and clinical outcomes in cardiovascular surgery. Their critique therefore comes not from statisticians skeptical of viscoelastic testing in general, but from researchers who use these assays routinely and who have a practical stake in how they are interpreted in acutely bleeding patients.

At the heart of their concern is a classic problem in clinical prediction research: the ratio of events to variables in logistic regression analysis. When investigators build a multivariable model, every predictor they include consumes statistical degrees of freedom. If the number of outcome events is small relative to the number of predictors, the model becomes unstable, and the estimated effect of any single variable can be distorted by chance. A widely cited simulation study by Peduzzi and colleagues, published in the Journal of Clinical Epidemiology in 1996, examined exactly this issue and found that models generally need a sufficient number of events per variable to produce reliable coefficients. In studies of hematoma expansion, where the outcome may occur in only a minority of enrolled patients, the number of available events can be limited, and models that include several covariates alongside the parameter of interest can easily strain against this constraint. The letter suggests that the LI60 association may be vulnerable to this kind of instability, meaning that the observed relationship could reflect overfitting of the model rather than a true biological signal.

The stakes of this methodological debate extend far beyond a single journal correspondence. Large randomized trials have tested whether antifibrinolytic drugs can improve outcomes in hyperacute intracerebral hemorrhage, on the theory that stabilizing the clot would halt expansion. The international phase 3 TICH-2 trial of tranexamic acid, published in The Lancet in 2018, did not demonstrate a clear benefit, and the more recent phase 2 STOP-MSU trial, published in The Lancet Neurology in 2024, tested tranexamic acid versus placebo in patients treated within two hours of symptom onset, the earliest window yet examined, with results that continue to fuel discussion about patient selection. A reliable coagulation biomarker could potentially identify the subgroup of patients whose bleeding is driven by hyperfibrinolysis and who would benefit most from antifibrinolytic treatment. That vision underlies much of the enthusiasm for viscoelastic parameters in this disease, and it is also why a false lead would matter: treatment decisions, trial designs, and even screening criteria could be built on an association that dissolves under scrutiny.

The letter’s title also gestures toward the alternative possibility, that LI60 might represent genuinely novel biology. If the association survives methodological challenges, in larger cohorts with adequate events per variable and in external validation, it would support a mechanistic picture in which premature clot lysis contributes directly to ongoing hemorrhage. This would align with the broader scientific framework laid out in a 2023 Lancet Neurology review by Morotti and colleagues on intracerebral hemorrhage expansion, which catalogued definitions, predictors, and prevention strategies and emphasized the heterogeneity of mechanisms driving growth, from vascular causes to coagulopathy. Distinguishing among these mechanisms is one of the central tasks of current neurocritical care research, and viscoelastic testing is one of the few tools that can characterize clot stability in near real time at the bedside.

Resolving the question will require the kinds of steps the letter implicitly demands: larger datasets, prespecified models with disciplined variable selection, and independent replication before LI60 enters clinical decision-making. For now, the correspondence stands as a reminder that in fast-moving fields where observational associations are quickly translated into therapeutic hypotheses, methodological rigor is not an academic nicety but a safeguard for patients. Whether LI60 proves to be a statistical mirage or a genuine window into the coagulopathy of brain hemorrhage, the exchange illustrates how science self-corrects in public, one carefully argued letter at a time, in one of medicine’s most demanding clinical arenas. The original research article and the letter are both available in Neurocritical Care, the journal of the Neurocritical Care Society, published by Springer Nature, and the letter’s authors declared no competing interests and no use of artificial intelligence tools in preparing their manuscript.

Subject of Research: The debated association between the LI60 rotational thromboelastometry parameter and hematoma expansion after spontaneous intracerebral hemorrhage

Article Title: LI60 and Hematoma Expansion in Intracerebral Hemorrhage: Statistical Artefact or Novel Biology?

Article References: LI60 and Hematoma Expansion in Intracerebral Hemorrhage: Statistical Artefact or Novel Biology?. (n.d.). https://doi.org/10.1007/s12028-026-02681-2

Image Credits: AI Generated

DOI: 10.1007/s12028-026-02681-2

Keywords: intracerebral hemorrhage, hematoma expansion, LI60, rotational thromboelastometry, fibrinolysis, logistic regression, tranexamic acid, TICH-2, STOP-MSU, neurocritical care, coagulation, statistical artifact