Sirolimus, also known by its original brand name rapamycin, works by inhibiting mammalian target of rapamycin, or mTOR, a central cellular signaling hub that governs growth, proliferation and the response of lymphatic endothelial cells to injury. In vascular anomalies, particularly lymphatic and veno-lymphatic malformations, aberrant signaling through the PI3K-AKT-mTOR pathway drives the abnormal proliferation of lymphatic vessels, producing masses that can swell, leak, deform tissue and compress airways. By damping this pathway, sirolimus can shrink lesions and dramatically improve symptoms, which is why it has become a mainstay of medical therapy for complex vascular anomalies worldwide. But mTOR signaling is also indispensable to the immune system, where it coordinates T cell activation, lymphocyte proliferation and the coordination of innate and adaptive defenses. Immunosuppression is the pharmacological flip side of the drug’s therapeutic action, and it is the reason sirolimus has long been used to prevent organ transplant rejection.

The research team, led by Fadiah Alkhattabi and colleagues, assembled a cohort of 95 pediatric patients who received sirolimus between January 2019 and June 2025. The median age of the children was 12 years. Lymphatic malformations accounted for 46.3 percent of cases and veno-lymphatic malformations for 34.7 percent, reflecting the fact that these lymphatic-predominant lesions respond most reliably to mTOR inhibition. The investigators systematically collected every infection occurring within twelve months of therapy initiation, capturing outpatient episodes, emergency department encounters, hospital admissions and intensive care stays. This twelve-month window was chosen deliberately, since it spans the period of dose titration and steady-state immunosuppression during which clinicians most worry about opportunistic and community-acquired pathogens alike.

The headline number is striking: 63.2 percent of the children developed at least one infection during the observation period, and among those affected, the median count was three infections per patient. That is not a background rate of ordinary childhood illness; it is a sustained pattern of recurrent infection concentrated in a defined therapeutic window. The microbiological geography of these infections was also telling. Respiratory infections dominated, accounting for 80 percent of all infectious episodes, followed by skin and soft tissue infections at 11.7 percent and bloodstream infections at 5 percent. The respiratory predominance makes biological sense. The lung mucosa is a vast immunological interface, constantly patrolled by resident lymphocytes whose function depends on intact mTOR signaling, and lymphatic malformations themselves can compromise airway architecture and mucociliary clearance, compounding the vulnerability.

The clinical consequences were far from trivial. Infections drove emergency department visits in 70 percent of affected episodes and required hospitalization in 23.3 percent. Among children who were hospitalized, 28.6 percent deteriorated enough to require intensive care admission. One infection-related death occurred in the cohort, a sobering reminder that in children whose immune systems are pharmacologically dampened, common pathogens can occasionally follow malignant trajectories. Beyond the acute morbidity, the study documented a second, subtler cost: disruption of the treatment itself. Sirolimus doses had to be adjusted in 58.3 percent of infection episodes, and the drug was temporarily discontinued in 40 percent of cases. For a therapy whose benefit depends on continuous suppression of aberrant lymphatic growth, these interruptions risk lesion regrowth and symptom recurrence, creating a clinical dilemma in which the treatment of the infection and the treatment of the malformation pull in opposite directions.

One of the study’s most provocative findings concerns dosing. Children who developed infections had received a lower initial sirolimus dose than those who did not, with a median of 1.20 milligrams per square meter of body surface area compared with 1.60 in the infection-free group, a difference that reached statistical significance. At first glance this seems paradoxical: more drug, fewer infections. The authors’ data do not establish causation, and the retrospective design precludes firm mechanistic conclusions, but several explanations are plausible. Children prescribed higher initial doses may have been systematically different in ways the analysis could not fully capture, including the extent of their disease, their nutritional status or the intensity of monitoring they received. Alternatively, the association may reflect confounding by indication in reverse, with clinicians reserving more aggressive dosing for children they judged to be more robust. Whatever the mechanism, the finding underscores how little is known about the dose-response relationship between sirolimus exposure and infectious risk in this population, and it cautions against the assumption that lower doses are automatically safer.

Equally consequential is what the study found about antibiotic prophylaxis. Roughly a quarter to a third of children in both groups received prophylactic antibiotics, 28.3 percent of those who developed infections and 25.7 percent of those who did not, and the difference in infection rates between the two groups was not statistically significant. In other words, prophylactic antibiotics did not appear to protect the children who received them. This null result matters because prophylaxis is not a benign intervention. Chronic antibiotic exposure selects for resistant organisms, disrupts the microbiome and carries its own burden of side effects, all without demonstrated benefit in this cohort. The findings suggest that routine prophylaxis for all children on sirolimus cannot be justified by the current evidence, and that a more nuanced, risk-adapted approach is warranted, reserving preventive strategies for children with identifiable additional vulnerabilities such as profound lymphatic involvement of the airway, a history of recurrent pneumonia or coexisting immunodeficiency.

The study’s design imposes important limits on interpretation. As a single-center retrospective analysis, it depends on the completeness and consistency of clinical documentation over six years, and infection ascertainment may vary with how diligently clinicians recorded outpatient episodes. Without a control group of children with vascular anomalies who were not treated with sirolimus, the study cannot isolate the drug’s contribution to infection risk from the baseline susceptibility conferred by the malformations themselves. Children with lymphatic malformations are known to experience recurrent infections even without immunosuppressive therapy, because malformed lymphatic tissue impairs local immune surveillance and fluid drainage. The true attributable risk of sirolimus, therefore, likely lies somewhere below the raw 63.2 percent figure, though the recurrent and severe nature of the observed infections suggests the drug is a meaningful contributor.

Nevertheless, the practical implications for pediatric practice are immediate. Clinicians managing children with vascular anomalies on sirolimus should maintain a low threshold for evaluating respiratory symptoms, counsel families about early signs of infection, and coordinate care between dermatology, hematology-oncology, pulmonology and infectious disease specialists. Therapeutic drug monitoring, already standard practice for sirolimus, may need to be paired with a more explicit accounting of infection history when titrating doses, and treatment interruptions during acute illness should be weighed carefully against the risk of lesion progression. The authors argue that close monitoring and risk-adapted infection prevention strategies may help optimize sirolimus therapy, a formulation that acknowledges both the drug’s undeniable value and the infectious toll documented in their data.

As sirolimus and the newer generation of PI3K and mTOR pathway inhibitors expand into ever broader pediatric populations, studies of this kind become essential scaffolding for safe practice. The Riyadh cohort demonstrates that infections during sirolimus therapy are common, clinically significant and frequently disruptive to treatment continuity, while simultaneously showing that the reflexive answer of antibiotic prophylaxis does not work. What is needed next is prospective, multicenter surveillance with standardized infection definitions, pharmacokinetic correlates and, ideally, biomarkers of immune function that can identify which children truly need intensified protection. Until such data exist, the study stands as a clear signal that the benefits of mTOR inhibition in vascular anomalies must be actively balanced against a substantial and measurable infectious burden, one that deserves the same clinical attention devoted to the malformations themselves.

Subject of Research: Infection risk in pediatric patients with vascular anomalies treated with sirolimus

Article Title: The proportion of infections in patients with vascular anomalies treated with sirolimus: a retrospective study

Article References: Alkhattabi, F., Alsaleem, A., Aledaili, S., Alhuthil, R., & Alobieda, S. (2026). The proportion of infections in patients with vascular anomalies treated with sirolimus: a retrospective study. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07707-4

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07707-4

Keywords: sirolimus, vascular anomalies, lymphatic malformations, pediatrics, mTOR inhibitor, infections, antibiotic prophylaxis, immunosuppression, respiratory infections, drug safety, retrospective study, BMC Pediatrics

The study enrolled 41 children and adolescents—20 males and 21 females—each of whom presented with a vascular malformation, clinically appreciable overgrowth of a body region beyond the vascular lesion itself, and a confirmed pathogenic or likely pathogenic variant in genes of the PI3K/AKT/mTOR pathway or related vascular signaling pathways, with PIK3CA deliberately excluded. The mean age at first clinical assessment was 9.0 years, while the mean age at molecular diagnosis was 12.1 years, a gap that underscores how long these conditions can evade a definitive label. Participants were recruited from the Regina Margherita Children’s Hospital in Turin, the Fondazione Policlinico Universitario Agostino Gemelli and the Bambino Gesù Children’s Hospital in Rome, the Santa Maria della Misericordia Hospital in Udine, and the San Bortolo Hospital in Vicenza, with genetic analysis performed in Bari.

The molecular results reveal a landscape far more diverse than the PIK3CA-centric literature might suggest. Variants in TEK, a gene encoding the TIE2 receptor tyrosine kinase that governs venous endothelial cell survival, were the most common, found in 24.4 percent of patients. GNAQ and GNA11, two closely related G-protein genes recurrently implicated in capillary malformations and Sturge-Weber syndrome, each accounted for 21.9 percent. PTEN variants, which define the PTEN hamartoma tumor syndrome spectrum, were present in 14.6 percent, RASA1 in 12.2 percent, and single patients carried variants in AKT3 and EPHB4. Critically, 65.9 percent of the identified variants were somatic—acquired after fertilization and present only in a mosaic fraction of cells—while 34.1 percent were germline, inherited or present in every cell of the body.

Detecting these variants demanded sequencing strategies far beyond conventional diagnostic testing. The team used next-generation sequencing with a targeted 17-gene panel at a read depth exceeding 2,000 reads per base, a level of coverage that allows detection of mosaic variants present at allele fractions as low as roughly 5 percent or below. Samples were drawn from peripheral blood in 12 patients, buccal swabs in 2, and—most importantly—biopsies of affected tissue in 27. This tissue-first strategy proved decisive: somatic variants identified in lesions were absent from blood DNA in every case, confirming that blood-based testing alone would have missed nearly two-thirds of the molecular diagnoses. Conversely, in several patients whose segmental, apparently sporadic presentations strongly suggested mosaicism, germline variants were ultimately identified, a discordance the authors emphasize as a warning against inferring mutational origin from clinical pattern alone.

The phenotypic data are equally revealing. Soft tissue or muscular overgrowth was documented in 60.9 percent of patients, skeletal anomalies in 46.3 percent, and macrocrania in 19.5 percent, the latter concentrated among children with PTEN variants. Adipose overgrowth appeared in a smaller subset, and internal organ involvement in just three individuals, including one child with a germline TEK variant whose colon harbored extensive vascular malformations and two patients with GNA11 or AKT3 variants who exhibited hemimegalencephaly, a dysplastic overgrowth of one cerebral hemisphere. Perhaps most strikingly, a minority of patients displayed segmental undergrowth rather than overgrowth: in one TEK patient, serpiginous vascular overgrowth coexisted with limb hypoplasia and muscle atrophy, and a GNA11 patient with cutis marmorata telangiectatica congenita showed mild hypoplasia of the affected limb. The authors interpret these cases as evidence that disturbed vascular hemodynamics—venous hypertension, microvascular shunting, chronic hypoperfusion—can impair normal tissue development, meaning the same underlying biology can push growth in opposite directions depending on timing, location, and severity.

Genotype-phenotype correlations, however, proved frustratingly loose. A syndromic diagnosis could be assigned on clinical grounds alone in 63.4 percent of patients, but the same clinical entities mapped onto multiple genes and vice versa. GNAQ and GNA11 variants surfaced in patients labeled with Klippel-Trenaunay syndrome, diffuse capillary malformation with overgrowth, Sturge-Weber syndrome, phakomatosis pigmentokeratotica, and in several cases that fit no defined syndrome at all. TEK variants spanned Klippel-Trenaunay syndrome, blue rubber bleb nevus syndrome, and unclassifiable presentations, while RASA1 variants appeared in Parkes-Weber syndrome and beyond. Even within a single gene, severity varied enormously. The authors argue that this overlap renders phenotype-based classification unreliable and advocate a shift toward molecularly driven frameworks, in which the gene and pathway—rather than the eponym—anchor the diagnosis.

The study also probes what overgrowth actually means. Regional enlargement in these patients may reflect true mesenchymal proliferation, but it can equally represent apparent enlargement driven by venous ectasia, increased regional perfusion, or secondary deformation of neighboring structures. Macrocrania in PTEN-related disease, for instance, likely reflects hamartomatous dysplasia rather than vascular-mediated expansion, and hemimegalencephaly in GNA11 and AKT3 patients represents cortical dysplasia rather than classical somatic overgrowth. These distinctions matter clinically, because they influence how imaging findings are interpreted, how cohorts are compared across studies, and ultimately how therapeutic decisions are made. The authors acknowledge that their inclusive definition—accepting both true and apparent overgrowth—broadens the cohort but also blurs a boundary that current imaging techniques cannot reliably draw.

The clinical burden documented in the cohort was substantial and, notably, not strictly proportional to lesion size. Nearly 44 percent of patients reported functional limitations ranging from fatigue and impaired high-demand physical activity to gait abnormalities, pain episodes, and visual deficits. Four patients experienced significant hemostatic or thrombotic complications, including one GNA11 patient who suffered cerebral ischemia with subsequent epilepsy and hemiparesis, and several TEK patients with recurrent intralesional thrombosis and phlebolithiasis. Psychosocial impact was formally recorded in only one patient, but the authors suspect this dramatically underestimates the true toll of disfiguring lesions, which families often underreport to clinicians. The findings argue for multidisciplinary assessment that encompasses functional and psychological dimensions alongside anatomical characterization, and for structured transition pathways as patients move into adult care.

Treatment data carry perhaps the most consequential message. Surgery was frequently non-curative, with relapse of overgrowth in nearly every treated region. Sirolimus, an mTOR inhibitor often used off-label in vascular anomalies, showed limited or absent benefit and was discontinued in all four patients who received it, in one case because of severe recurrent infections. Yet alpelisib, a selective PI3K inhibitor developed for PIK3CA-driven disease, produced clinical and volumetric improvement in a subset of TEK-variant patients despite the absence of any PIK3CA mutation. This unexpected efficacy supports the concept of convergent downstream pathway activation: distinct upstream genetic lesions—whether in TEK, GNAQ, or PIK3CA—may funnel into shared PI3K/AKT/mTOR signaling, creating a biological rationale for pathway-targeted therapies that extends beyond the classical PROS population. One TEK patient also responded favorably to propranolol, though the authors caution that this single observation remains anecdotal, since propranolol’s established use is confined to infantile hemangiomas.

The study’s limitations are acknowledged candidly: its retrospective design, the selection bias inherent in requiring an identifiable pathogenic variant, and the absence of standardized quality-of-life measures that likely understated disease burden. Even so, the implications are clear. For children with vascular malformations and overgrowth who test negative for PIK3CA, the diagnostic journey should not end; broad gene panels, deep sequencing of affected tissue whenever feasible, and systematic parental testing offer the best route to a molecular diagnosis. And as targeted therapies mature, the gene a child carries—not the name of a syndrome—may determine whether an effective drug exists. This cohort of 41 patients makes a compelling case that the era of molecularly driven classification for vascular overgrowth disorders has arrived, and that the patients it will benefit most are precisely those the old system left unclassified.

Subject of Research: Genetic and clinical characterization of non-PIK3CA vascular malformations with overgrowth in children

Article Title: Clinical and molecular characterization of non-PIK3CA-related vascular malformations with overgrowth in a pediatric cohort

Article References: Reynolds, G., Luca, M., Coppo, P., La Selva, R., Leoni, C., Bellani, I., Ragusa, S., Massuras, S., Perrelli, S., Martino, S., Buonuomo, P. S., Rondot, F., Spinelli, A. M., Morando, C., Piglionica, M., Resta, N., & Mussa, A. (2026). Clinical and molecular characterization of non-PIK3CA-related vascular malformations with overgrowth in a pediatric cohort. World Journal of Pediatrics. https://doi.org/10.1007/s12519-026-01098-3

Image Credits: AI Generated

DOI: 10.1007/s12519-026-01098-3

Keywords: vascular malformations, overgrowth, PIK3CA, TEK, GNAQ, GNA11, PTEN, RASA1, somatic mosaicism, next-generation sequencing, alpelisib, sirolimus

The research, led by a team at Beijing Children’s Hospital of Capital Medical University, analyzed data from 257 pediatric patients treated with sirolimus, drawing on a remarkable 766 to 892 blood samples collected during routine care. Rather than treating adverse reactions as isolated events, the investigators systematically quantified the relationship between steady-state trough concentrations of sirolimus, known as Cmin, and three of the drug’s most clinically significant toxicities: myelosuppression, abnormal liver function, and dyslipidemia. Trough concentrations, measured immediately before the next dose, are the standard metric for therapeutic drug monitoring of sirolimus, because the drug’s effects and toxicities track closely with whole-blood levels.

What makes the study methodologically distinctive is its pairing of two computational approaches that rarely meet in pharmacology. The first is Bayesian kernel machine regression, or BKMR, a flexible statistical framework originally developed to assess the health effects of complex environmental exposures such as air pollutant mixtures. BKMR excels at modeling nonlinear relationships and interactions among multiple simultaneous exposures without imposing rigid functional forms, which makes it well suited to the messy reality of pediatric drug monitoring, where age, weight, disease type, concomitant medications, and treatment duration all shift the toxicity landscape. The second component is a population pharmacokinetic model, or PopPK, which describes how sirolimus is absorbed, distributed, and cleared across a heterogeneous population of children and predicts individual exposure from dosing history and patient characteristics.

By integrating the two into a unified BKMR-PopPK framework, the researchers created something more powerful than either method alone. The PopPK model translates a proposed dose into a predicted trough concentration for a specific child, while the BKMR layer converts that predicted exposure into a quantified risk of myelosuppression, liver injury, or dyslipidemia. The result is a tool for model-informed precision dosing, an approach that replaces one-size-fits-all weight-based dosing with individualized regimens designed to keep each patient inside a safety window that is narrow enough to prevent toxicity but wide enough to control the underlying vascular lesion.

The analysis identified disease complexity and duration of therapy as key modifiers of risk. Children with complex or mixed vascular malformations faced different exposure-response relationships than those with simpler, more localized lesions, and the risks of certain toxicities grew as treatment extended over months. Based on these quantitative relationships, the team recommends routine trough concentration monitoring within two distinct safety windows: 5.0 to 7.5 nanograms per milliliter for children with simple lesions or milder disease, and 8.0 to 10.2 nanograms per milliliter for those with complex or mixed vascular malformations, who may require higher exposure to achieve disease control.

The temporal dimension of the findings carries particular weight for families and clinicians committed to long-term therapy. Sirolimus treatment for vascular anomalies often continues for a year or more, and the study’s results indicate that vigilance must intensify rather than relax as months accumulate. Beyond six months of treatment, the authors call for heightened surveillance for myelosuppression and dyslipidemia through regular screening of blood counts and lipid panels. This is a meaningful departure from practice patterns in which monitoring may become less rigorous once a patient has tolerated the drug through the early treatment period.

Each of the three adverse reactions examined carries its own clinical stakes. Myelosuppression, the suppression of bone marrow activity, can lower blood cell counts and increase vulnerability to infection, anemia, and bleeding. Abnormal liver function signals hepatotoxicity that, if unchecked, can progress to more serious injury. Dyslipidemia, the elevation of blood lipids, is subtler but consequential in children, because lipid abnormalities established early in life can seed long-term cardiovascular risk. By quantifying how each toxicity relates to trough concentration, the framework allows clinicians to weigh these distinct risks against the therapeutic benefit of suppressing the malformation itself.

The broader significance of the work extends beyond vascular anomalies. Rare pediatric diseases pose a persistent methodological dilemma: patient populations are small, clinical trials are difficult to mount, and safety data are often fragmentary or anecdotal. The authors argue that their integrated framework offers a template for extracting rigorous, quantitative safety insights from the routine clinical data that even small cohorts generate. In doing so, the study addresses a genuine knowledge gap. Prior research had suggested a strong link between sirolimus exposure and adverse reactions, but the quantitative relationship, the actual probabilities of toxicity at given concentrations, had remained undefined, leaving clinicians to rely on empirical dosing inherited from transplant medicine.

Therapeutic drug monitoring itself is not new to sirolimus; the drug’s narrow therapeutic index has long demanded blood level measurement. What is new is the precision with which those measurements can now be interpreted. Instead of a single broad target range applied to every child, the study supports disease-specific and time-dependent target ranges, embedded within a predictive model that can forecast an individual patient’s exposure and toxicity risk before a dose is even administered. The retrospective design, approved by the Institutional Ethics Committee of Beijing Children’s Hospital, means the findings will need prospective validation, but the sample size and the density of pharmacokinetic sampling lend the conclusions substantial credibility.

For the children who depend on sirolimus to shrink lesions that compress airways, deform faces, or bleed unpredictably, the study promises something deceptively simple: a better chance of staying on the drug long enough for it to work. Treatment discontinuation driven by adverse reactions is one of the most common reasons vascular anomaly therapy fails, and every avoidable discontinuation represents a child whose disease resumes its advance. By converting pharmacological data into individualized risk estimates and dosing guidance, the BKMR-PopPK framework moves pediatric vascular anomaly care a decisive step closer to the era of precision medicine, where the question is no longer simply how much drug to give, but how much drug this particular child can safely carry.

Subject of Research: Quantifying sirolimus exposure-response safety relationships in pediatric vascular anomalies using an integrated BKMR-PopPK framework

Article Title: Safety of sirolimus in pediatric vascular anomalies: a BKMR-PopPK exposure–response framework for individualized risk and dosing

Article References: Liu, B., Xu, X.-L., Wang, J.-L., Li, J., Wu, Y.-X., Zhang, X.-X., Liu, Q.-Y., Zhao, Y.-M., Guo, P., Zhang, R.-Q., Zhou, H., Mao, X.-T., Jia, Y.-M., Cheng, X.-L., Wang, S.-C., & Wang, X.-L. (2026). Safety of sirolimus in pediatric vascular anomalies: a BKMR-PopPK exposure–response framework for individualized risk and dosing. World Journal of Pediatrics. https://doi.org/10.1007/s12519-026-01076-9

Image Credits: AI Generated

DOI: 10.1007/s12519-026-01076-9

Keywords: sirolimus, vascular anomalies, pediatrics, adverse drug reactions, therapeutic drug monitoring, population pharmacokinetics, Bayesian kernel machine regression, precision dosing, myelosuppression, dyslipidemia, liver function, rare diseases