Sepsis remains one of the leading causes of death in children worldwide, arising when an infection triggers a dysregulated immune response that damages the body’s own organs. In adults, clinicians have long recognized that the liver is both a victim and a signal in this process. Hepatic dysfunction in adult sepsis is considered an independent predictor of mortality, meaning that even after accounting for other factors, a failing liver marks a patient at elevated risk. Whether the same holds true for children, however, had never been systematically examined. Individual pediatric studies had reported varying rates of liver injury and conflicting associations with death, but no one had pooled the data. The Indonesian team registered their review prospectively with PROSPERO under the identifier CRD420261387360 and searched five major databases, including PubMed, Scopus, MEDLINE via EBSCO, the Cochrane Library, and ClinicalTrials.gov, from their inception through 21 May 2026, without language restrictions. Notably, Embase was not searched, a limitation the authors acknowledge.

The eligibility criteria were strict. Studies had to enroll children aged one month to 18 years, define sepsis according to consensus criteria, and report outcomes stratified by the presence or absence of liver injury. Two reviewers independently screened records and extracted data, a standard safeguard against individual error. Risk of bias was evaluated using two prespecified instruments: the Newcastle-Ottawa Scale for the observational studies included in the mortality analysis and the Quality in Prognosis Studies tool, known as QUIPS, which is specifically designed for prognostic research. Statistical pooling employed a Mantel-Haenszel random effects model for mortality odds ratios, while the proportion of children developing liver injury was estimated with a logistic-normal generalized linear mixed model, an approach better suited to proportions than naive averaging. Certainty of the evidence was graded using GRADE, the framework that rates confidence in findings from high to very low.

After screening, ten studies made the final cut. The headline result came from five of these studies, encompassing 1,717 children, that reported mortality according to liver injury status. Pooling their unadjusted data, the researchers found that children with SALI had nearly four times the odds of dying compared with those without it, with an odds ratio of 3.88 and a 95 percent confidence interval of 2.21 to 6.80. Three of the studies measured death during hospital admission, one measured death in the pediatric intensive care unit, and one did not specify the timing. On the surface, this is a striking signal: a liver under assault during sepsis appears to mark a child in grave danger.

But the statistical fine print tells a more cautious story. The prediction interval, which estimates the range of true effects one might expect in future studies, stretched from 0.63 to 23.80, crossing the value of 1.0 that would indicate no association at all. In other words, some future pediatric population might show no link between liver injury and death, while another might show an enormous one. The Hartung-Knapp adjusted confidence interval, a more conservative calculation that accounts for uncertainty in the between-study variance, widened further to 1.65 to 9.12. Combined with a GRADE rating of very low certainty, these numbers mean clinicians should treat the fourfold figure as a hypothesis-generating signal rather than a settled fact. The heterogeneity across studies, likely driven by different definitions of liver injury, different patient populations, and different measurement schedules, is substantial enough to temper any confident clinical claim.

The deeper problem is confounding by severity. Children whose livers fail during sepsis are, almost by definition, sicker than children whose livers hold out. Two of the included studies attempted to address this statistically. Saini and colleagues reported that SALI remained associated with mortality after adjustment for selected clinical covariates, but their mortality model did not include pSOFA, the pediatric Sequential Organ Failure Assessment score, or any other validated global severity measure. Without such a score, it is impossible to know whether the liver adds information beyond what general illness severity already captures. Only Koca and colleagues adjusted hepatic markers for validated pediatric severity scores, namely PRISM III, the Pediatric Risk of Mortality Score, and PELOD-2, the Pediatric Logistic Organ Dysfunction Score. In that analysis, neither alanine aminotransferase, the enzyme ALT commonly used to detect liver cell damage, nor total bilirubin, the pigment that accumulates when the liver fails to clear it, remained independently associated with death. Once a child’s overall severity was accounted for, the liver’s apparent prognostic power evaporated.

Beyond mortality, the review also estimated how often liver injury occurs in pediatric sepsis at all. Across seven studies covering 1,876 children, the pooled proportion was 29 percent, with a wide 95 percent confidence interval of 17 to 46 percent. That range is itself informative: it reveals how inconsistently SALI is defined and detected across the pediatric literature. Some studies may rely on a single elevated transaminase value, others on bilirubin thresholds, and still others on composite criteria, and the timing of measurement during the septic episode varies just as widely. The authors argue that this heterogeneity makes a strong case for standardized pediatric definitions of sepsis-associated liver injury and for agreed ascertainment schedules, so that future studies measure the same thing at the same moments. The Pediatric Organ Dysfunction Information Update Mandate, or PODIUM, represents one ongoing effort to bring consistency to organ dysfunction definitions in pediatric critical care.

The review also examined candidate biomarkers that might identify liver injury or predict outcomes earlier than conventional tests. The most promising was APRI, the AST-to-platelet ratio index, a simple calculation combining aspartate aminotransferase levels with platelet counts. In two cohorts, APRI showed areas under the receiver operating characteristic curve, or AUROCs, of 0.83 to 0.89, values generally considered good discriminative ability. Yet the authors urge caution on two fronts. First, the diagnostic thresholds differed substantially between the two cohorts, meaning a child classified as having liver injury by one study’s cutoff might be classified as normal by another’s. Second, the APRI is built on AST, while the outcome definitions in these studies were based on ALT, creating an overlap problem that limits how the biomarker’s performance can be interpreted. Other markers examined in the underlying studies, including lactate dehydrogenase, apolipoprotein A-V, and the albumin-corrected anion gap, remain at even earlier stages of evaluation. The authors conclude that no candidate biomarker is ready for clinical use without prospective multicenter validation.

What does this mean for the pediatric intensivist at the bedside? The practical takeaway is nuanced. An elevated ALT or bilirubin in a septic child should certainly prompt attention, and the unadjusted association suggests these children warrant heightened vigilance. But the evidence does not yet support treating liver injury as an independent prognostic marker that adds value beyond established severity scores such as PRISM III, PELOD-2, or pSOFA. The honest answer to whether hepatic dysfunction carries prognostic information of its own is that nobody currently knows, and the review makes that uncertainty explicit rather than papering over it. This kind of transparency is increasingly valued in evidence-based medicine, where overconfident meta-analytic estimates can mislead practice as surely as absent evidence can.

The study, published open access on 7 October 2026 and available under a Creative Commons license, received no external funding and its authors declare no competing interests. Its limitations are clearly stated: only ten studies were available, Embase was not searched, adjusted analyses were rare, and the underlying literature is dominated by observational designs vulnerable to confounding. The research team, which also included Ade Gautama and Khanyfah Syagita, notes with gratitude that Dr. Abdulkerim Elmas provided unpublished subgroup data through personal correspondence, strengthening the primary mortality analysis. For now, the review stands as both a warning and a roadmap: liver injury during pediatric sepsis is common, appears dangerous, and demands rigorous, standardized study before it can be woven into the scoring systems that guide care for the world’s sickest children.

Subject of Research: Sepsis-associated liver injury and mortality outcomes in children

Article Title: Sepsis-associated liver injury and clinical outcomes in children: a systematic review and meta-analysis

Article References: Wiriadinata, N. E. R., Gautama, A., Syagita, K., & Oswari, H. (2026). Sepsis-associated liver injury and clinical outcomes in children: a systematic review and meta-analysis. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07775-6

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07775-6

Keywords: sepsis, liver injury, pediatrics, meta-analysis, mortality, PICU, biomarkers, APRI, PRISM III, PELOD-2, pSOFA, systematic review

The study, published in CVIR Oncology, analyzed data from the CIRSE Registry for SIR-Spheres Therapy, known as CIRT, a Europe-wide observational study conducted in eight countries across 27 centers. Of the 1,036 patients enrolled with primary or secondary liver malignancies, the researchers focused on 169 adults whose liver metastases originated from neuroendocrine tumors, breast cancer, pancreatic cancer, or melanoma. All were treated with yttrium-90 resin microspheres, tiny radioactive spheres delivered through the hepatic artery that lodge in the small vessels feeding tumors and deliver high-dose internal radiation while sparing most healthy tissue.

The technical logic behind TARE is elegant. Unlike conventional external beam radiotherapy, which must pass through healthy tissue to reach its target, radioembolization exploits the fact that liver tumors draw their blood supply predominantly from the hepatic artery, while healthy liver tissue depends on the portal vein. Catheters threaded from a puncture in the groin or wrist guide the microspheres into the arteries supplying the tumors, where the beta radiation from yttrium-90, which travels only a few millimeters in tissue, kills cancer cells from within. Before treatment, teams perform careful mapping angiography and dosimetry, often using the body surface area method or a partition model, to calculate the activity to be injected and to avoid reflux of spheres into the stomach or bowel.

The survival results varied dramatically by tumor type, underscoring how different these diseases are. Patients with neuroendocrine tumors fared best, with a median overall survival of 33.3 months after treatment. Those with breast cancer liver metastases had a median survival of 10.7 months, while patients with melanoma reached 14.7 months. Pancreatic cancer, notoriously aggressive even when it has spread only to the liver, showed the poorest outcome at 5.6 months. Median follow-up ranged from 5.6 months in the pancreatic cohort to 19.3 months in the neuroendocrine group, and patients were enrolled between January 2015 and December 2017 with follow-up through the end of 2019.

Safety was the study’s most consistent message. Severe adverse events of grade three or higher, defined according to the standard CTCAE criteria, occurred in just 10 percent of all 169 patients. Procedure-related adverse events were predominantly mild to moderate, affecting roughly one in five neuroendocrine and breast cancer patients and only about six percent of those with pancreatic cancer or melanoma. Predefined serious events tracked in the registry included abdominal pain, fatigue, fever, nausea, vomiting, gastrointestinal ulceration, gastritis, and radiation cholecystitis, but these remained uncommon. For a therapy that delivers ionizing radiation directly into a vital organ, this safety profile across four very different tumor types is a notable result.

Perhaps the most scientifically intriguing findings came from the multivariable analysis, in which the researchers used Cox proportional hazards models to identify independent predictors of survival. Two laboratory-based markers emerged repeatedly. The Albumin-Bilirubin grade, or ALBI, which combines liver synthetic function with bile clearance, predicted worse overall survival in both the neuroendocrine and breast cancer cohorts, with hazard ratios above four. The Aspartate transaminase to Platelet Ratio Index, or APRI, a marker usually associated with liver fibrosis, predicted poorer survival in the neuroendocrine and melanoma cohorts. In melanoma patients the effect was striking: an APRI value above 0.40 carried a hazard ratio of 37.63 for death, an extraordinarily strong association. In breast cancer patients, an international normalized ratio above one and an elevated neutrophil-to-leukocyte ratio also predicted worse progression-free survival.

These biomarker findings carry practical weight. They suggest that the health of the underlying liver, and possibly the systemic inflammatory state reflected in blood counts, shapes how much patients benefit from internal radiation, and that the relevant markers differ by tumor type. The authors argue this supports individualized patient assessment that incorporates both the histology of the metastases and routine laboratory evaluation. It also hints that simple blood tests available before treatment could help clinicians select which patients are most likely to benefit, an increasingly important question as liver-directed therapies proliferate.

Context matters for interpreting the survival numbers. For neuroendocrine liver metastases, which develop in up to 80 percent of patients with these tumors, the 33.3 month median survival aligns closely with a systematic review of 809 patients that reported a pooled overall survival of 33 months, and with the American RESiN registry, which reported a median of 33 months. For breast cancer, the 10.7 month result is comparable to a meta-analysis of 452 patients that found a median of 11.3 months. In melanoma, where nearly 60 percent of the study’s patients received TARE as first-line liver-directed therapy, the 14.7 month median sits comfortably within the 9 to 24 month range reported across systematic reviews of radioembolization for uveal melanoma. The pancreatic cancer result, while bleak, reflects a disease in which even surgery and modern chemotherapy yield median survivals under a year in the metastatic liver setting.

The study’s authors are careful about its limits. As an observational registry without a control group, it cannot prove that TARE is superior to other options, and selection bias and residual confounding cannot be excluded. The lack of a standardized treatment protocol across 27 centers introduced heterogeneity in baseline characteristics, treatment timing, and surrounding systemic therapies. Missing data may have affected some analyses, and no multivariable model was possible for the pancreatic cohort because too few variables reached the significance threshold. What the study does provide is something rarer: prospectively collected, multicenter, real-world evidence about how TARE is actually used and how patients actually fare when multidisciplinary teams choose it for tumors that fall outside existing guidelines.

That guideline gap is precisely why the results matter. TARE is a standard of care for intermediate to advanced hepatocellular carcinoma and for metastatic colorectal cancer, but for non-colorectal liver metastases, formal recommendations have been largely absent; the European Neuroendocrine Tumour Society recently noted that no data existed on liver embolization for neuroendocrine carcinoma. By documenting safety and encouraging effectiveness in 169 patients across four tumor types, the CIRT analysis gives clinicians a firmer evidence footing for a treatment many were already offering, and it sets a clear agenda: prospective comparative studies are now needed to identify which patient groups benefit most, and simple markers like ALBI and APRI may help decide who should be offered the radioactive beads and who should be spared the procedure.

Subject of Research: Yttrium-90 transarterial radioembolization for non-colorectal liver metastases

Article Title: Transarterial radioembolization for liver metastases of non-colorectal origin: safety and effectiveness results from the prospective multicenter CIRSE registry for SIR-spheres therapy (CIRT)

Article References: Arnold, D., Chlorogiannis, D. D., Sangro, B., Kolligs, F., Maleux, G., Bilbao, J. I., Pech, M., Pfammatter, T., de Jong, N., Geyer, M., Urdániz, M., & Helmberger, T. (2026). Transarterial radioembolization for liver metastases of non-colorectal origin: safety and effectiveness results from the prospective multicenter CIRSE registry for SIR-spheres therapy (CIRT). CVIR Oncology, 2(1), Article 6. https://doi.org/10.1007/s44343-026-00039-1

Image Credits: AI Generated

DOI: 10.1007/s44343-026-00039-1

Keywords: transarterial radioembolization, yttrium-90, liver metastases, neuroendocrine tumors, breast cancer, pancreatic cancer, melanoma, SIR-Spheres, CIRT registry, ALBI grade, APRI, interventional radiology