Lipid dysregulation after hematopoietic stem cell transplant

Abstract

Transplant-associated thrombotic microangiopathy (TA-TMA) is a serious complication of allogeneic hematopoietic stem cell transplant (allo-HSCT), primarily driven by endothelial injury and complement activation. Statins, combined with other drugs, are commonly used as prophylaxis against endothelial injury in some parts of the world but their mechanism of action has not been clearly defined. We hypothesized that dysregulation of lipids, or their precursors, ceramides, might be an important mechanism of endothelial injury, and that statins might ameliorate that dysfunction. We measured plasma ceramide species at baseline and day 14 in pediatric and young adult allo-HSCT recipients. Ceramide species in general were increased in those who later developed endothelial injury, manifest as TA-TMA. These findings highlighted ceramides as markers of endothelial stress, prompting us to explore whether statin prophylaxis could favorably modulate lipid and ceramide pathways. A single-arm phase I trial of pravastatin prophylaxis was also performed in patients at elevated risk of endothelial injury due to high body mass index to assess lipid and ceramide modulation over time. Multiple ceramide species were elevated in patients who developed TA-TMA and showed strong correlations with ST2 but not with sC5b-9. While ceramides were associated with TA-TMA in univariate models, only ST2 remained significant in multivariable analysis. Addition of ceramide levels to ST2 only modestly improved prediction of later TA-TMA in Receiver Operating Characteristic (ROC) analysis. Pravastatin prophylaxis was associated with distinct shifts in lipoprotein and ceramide profiles, potentially reflecting modulation of endothelial function. Pravastatin may alter ceramide and lipoprotein pathways in a clinically meaningful way, contributing to their role in endothelial protection.

Introduction

Hematopoietic stem cell transplant (HSCT) is the sole curative modality for specific pediatric malignant and benign hematologic disorders. Despite advances in supportive care and transplant conditioning, life-threatening complications from endothelial damage secondary to HSCT procedures can lead to substantial morbidity and mortality. Endothelial damage plays a pivotal role in the induction of injury syndromes specific to HSCT including transplant-associated thrombotic microangiopathy (TA-TMA), veno-occlusive disease, and graft-versus-host disease (GvHD). These complications cumulatively affect up to 70% of patients receiving allogeneic-HSCT (allo-HSCT).^1-3 Inherent non-modifiable risk factors (e.g., race) also contribute to the development and progression of TA-TMA.^4,5 We have previously reported an increased risk of endothelial injury in children with high body mass index (BMI) who have received HSCT.⁵ This observation, together with a proteomics study using an obese mouse model to detect an upregulation of ceramide pathways after transplant (data not shown), drew our attention to a possible role for lipid metabolism, and ceramides in particular, in adverse outcomes after HSCT. Ceramides are central to endothelial biology and inflammatory signaling, and their modulation by statins has generated interest in their potential role as both therapeutic targets and predictive biomarkers of TA-TMA. Ceramides are a heterogeneous class of bioactive sphingolipids comprising a sphingoid base and a fatty acyl chain.^6,7 Ceramides play essential roles in maintaining endothelial barrier homeostasis and integrity.⁸ Elevated ceramides increase oxidative stress, reduce nitric oxide (NO) production, induce mitochondrial dysfunction, modify endothelial activation, and increase endothelial inflammation.⁹ The different isoforms of ceramides have varying tissue distribution and function depending on their fatty acid chain lengths.¹⁰ Ceramides are connected to numerous inflammatory diseases, including systemic lupus erythematosus (SLE), diabetes mellitus, cardiovascular disease, and are even connected to GvHD.^11-14 Interestingly, sphingosine-1-phosphate receptor (S1PR) modulators (e.g., Mocravimod) have been shown to diminish signaling that is needed for T-cell migration to lymphoid organs and has been shown to reduce GvHD in murine models and in human studies.^15,16 This work highlights ceramides as a potential therapeutic target in the allo-HSCT setting.^13,16

Alongside ceramides, other lipid mediators may influence transplant outcomes. Despite increased recognition of the role of lipids in vascular inflammation, very few studies have examined lipidomic changes in HSCT recipients (particularly in the early post-transplant period when initial endothelial injury occurs) and associations with clinical outcomes. Limited prior investigations have focused on late dyslipidemia management and long-term cardiovascular complications after transplant.^17,18 High-density lipoprotein (HDL), for example, has been widely studied in the context of atherosclerosis and is typically associated with anti-inflammatory and vasculoprotective effects.¹⁹ However, under certain pathological conditions, HDL-C may adopt proinflammatory characteristics.^20,21 To date, no studies have explored the functional role of HDL-C in HSCT or its relationship with transplant-related endothelial complications. Prior reports have demonstrated that statins can reduce the incidence and severity of TA-TMA, likely through their protective effects on the endothelium.²² Originally developed as lipid-lowering agents,²³ statins are now recognized for their pleiotropic properties, including modulation of ceramide and metabolite profiles, improvement in endothelial function by increasing production of NO and reduction in inflammatory signaling.^24-35 These effects are particularly relevant in the setting of HSCT, where endothelial injury is a central driver of TA-TMA pathogenesis. We have evaluated ceramides in the blood of children early after HSCT to determine whether similar alterations in ceramide pathways occur in human transplant recipients. We also investigated whether early post-transplant ceramide and lipid profiles differ between HSCT recipients who developed TA-TMA and those who did not, and the feasibility and impact of statin prophylaxis in children.

Methods

Patient cohort

We studied ceramide species in a consecutive cohort of 135 pediatric and young adult patients undergoing first allo-HSCT at Cincinnati Children’s Hospital Medical Center between January 2013 and December 2021. Clinical data were obtained from an Institutional Review Board (IRB)-approved database, and blood samples were collected under informed consent and stored in an IRB-approved repository. Samples were prospectively collected at baseline (pre-conditioning) and on day +14 post-HSCT.

Transplant-associated thrombotic microangiopathy diagnosis and risk stratification

Recipients were prospectively monitored for TA-TMA using weekly clinical assessments and laboratory screening (lactate dehydrogenase [LDH], schistocytes, urine protein/ creatinine ratio) from pre-conditioning through day +100. TA-TMA was diagnosed according to laboratory and clinical criteria published by Jodele et al.³⁶

Ceramide analysis using ultra-high performance liquid chromatography-mass spectrometry / mass spectrometry analysis

Plasma sphingolipids (20 μL) were extracted using a modified in-house protocol.³⁷ Quantification was performed by ultra-high performance liquid chromatography (UHPLC) coupled to triple-quadruple mass spectrometry (Waters, Milford, MA, USA) in Multiple Reaction Monitoring mode, using an Acquity UHPLC CSH C18 column (2.1x100 mm, 1.7 μm). Calibration curves were generated in charcoal-stripped human serum and processed identically to patient samples. Chromatography used a binary solvent system with gradient elution (acetonitrile / water and isopropanol / acetonitrile, each with 10 mM ammonium formate and 0.1% formic acid) at 0.4 mL/min, column temperature 55°C, total run time 20 minutes. Data were processed in MassLynx 4.1.

Safety and feasibility phase I study of pravastatin prophylaxis during hematopoietic stem cell transplant

A prospective phase I trial (2023-2025; clinicaltrials.gov NCT05524246, IND #162341) evaluated pravastatin prophylaxis during HSCT. Pravastatin (0.2-0.4 mg/kg/day; dose adjusted for calcineurin inhibitor use) was initiated at inpatient admission before conditioning and continued until day +35. Outpatients completed remaining doses at home, with compliance documented in study diaries. Eligible patients were 2-25 years old, overweight / obese according to Centers for Disease Control and Prevention (CDC) BMI criteria,³⁸ scheduled for allo-HSCT, and able to take enteral medications. Exclusion criteria included pravastatin anaphylaxis, AST/ALT >3 times ULN, GFR <50 mL/min/1.73 m², neuromuscular/metabolic disorders predisposing to rhabdomyolysis, or concurrent OATP1B1/1B3 substrate use. A consecutive cohort of HSCT patients not treated with pravastatin and matched as closely as possible by BMI were used as historical controls. Standard transplant supportive care procedures, including administration of total parenteral nutrition with intravenous lipid emulsions when clinically indicated (e.g., mucositis impairing oral intake), were followed in all patients regardless of study participation.

Primary endpoints were safety (National Cancer Institute Common Terminology Criteria for Adverse Events v5.0) and feasibility (<70% adherence or withdrawal). Compliance was assessed via inpatient records and diaries. Secondary endpoints were day +100 TA-TMA incidence, acute GvHD, and overall survival. Exploratory endpoints included lipid / lipoprotein changes, with comparisons to 20 historical elevated-BMI controls. Because the majority of patients receive overnight total parenteral nutrition with intralipid, samples were often obtained shortly after lipid infusion. To minimize confounding, analyses emphasized within-person changes, with each patient serving as their own control.

Statistical analysis

All available quantitative data were included in statistical analyses. For descriptive purposes, continuous variables are reported as median (Interquartile Range [IQR]) given the presence of skewed distributions. Continuous variables were summarized as median (IQR) and categorical variables as frequency (%). Fisher exact and Wilcoxon tests compared categorical and continuous variables, respectively; Mann-Whitney U test was used to test unpaired groups. For ceramide analyses, False Discovery Rate was controlled by the Benjamini-Hochberg method, with adjusted P<0.01 considered significant; for other analyses, P<0.05 was considered statistically significant. In the pravastatin pilot trial, the prespecified secondary endpoint was incidence of TA-TMA by day +100. Separately, in the correlative cohort not treated with pravastatin, we calculated cumulative incidence curves for TA-TMA through one year to assess longer-term outcomes. Gray’s competing risk method estimated 1-year TA-TMA cumulative incidence. Analyses were conducted in R v3.5.1.

Results

Patient and transplant characteristics

Demographic- and transplant-related characteristics in the initial HSCT cohort in which ceramides were measured, including 135 consecutive allo-HSCT recipients, are shown in Table 1. A majority of study subjects were white, with a higher proportion of males (54.1%). Most transplants were performed for non-malignant indications (71.9%). A majority of patients received an unrelated donor graft (69.6%), and bone marrow was the most common stem cell source (51.9%).

Plasma ceramides after hematopoietic stem cell transplant

Ceramide levels baseline (pre-conditioning) and at day 14 post HSCT are described in Figure 1. Ceramide species are categorized by the length of the fatty acyl chain (e.g., C16Cer, C18Cer). Baseline concentrations of C16Cer (median: 122.4 nmol/L, range: 10-562.2 nmol/L) significantly increased by day 14 (median: 326.1 nmol/L, range: 142.5-928.9 nmol/L) (P<0.0001) (Figure 1A). Significant elevation in concentrations of C18Cer, C20Cer, C24:1Cer and C26:1Cer at day 14 were also observed (Figure 1B-E).

Plasma ceramide changes in transplant-associated thrombotic microangiopathy

Given the established role of ceramides in promoting endothelial dysfunction,³⁹ we next examined whether specific ceramide alterations were associated with the development of TA-TMA, a complication driven by endothelial injury. Figure 2 shows concentrations of ceramide species at baseline and day 14 from HSCT in patients later diagnosed with TATMA compared to patients without TA-TMA. We anticipated that day 14 ceramide levels would be elevated in those with TA-TMA but were surprised to find that baseline levels of ceramide species were also elevated in patients later diagnosed with TA-TMA compared to patients who never developed TA-TMA. While day 14 ceramide levels universally increased in all patients, patients with TA-TMA had more elevated concentrations of ceramides at day 14 compared to patients who never developed TA-TMA (Figure 2A-E). Online Supplementary Tables S1 and S2 list baseline and day 14 concentrations of all measured ceramide species in patients with and without TA-TMA. Because of collinearity among ceramide species and multiple comparisons being made, we set our level of statistical significance to P<0.01, and caution should be used in considering whether particular ceramide species contribute more or less to our findings. Ceramide species that were significantly elevated at baseline in patients with later TA-TMA included C16Cer, C18Cer, C20Cer, C22Cer, and C24:1 Cer. At day 14, patients who developed TA-TMA had significantly higher levels of C16Cer, C16-OHCer, C18Cer, C20Cer, and C24:1Cer compared to patients without TA-TMA (Online Supplementary Table S2).

Table 1.Patient demographics and transplant characteristics.

Univariate and multivariate analysis of association of ceramides and transplant-associated thrombotic microangiopathy

Univariate analyses and multivariate analyses were performed to evaluate the association between individual and multiple ceramide species and the development of TA-TMA (Table 2). Univariate analysis showed C16Cer (Odds Ratio [OR]: 3.44, 95%CI: 1.62-8.34; P=0.003), C18Cer (OR: 2.44, 95%CI: 1.33-5.07; P=0.009), C20Cer (OR: 2.41, 95%CI: 1.30-5.13; P=0.012), C22Cer (OR: 2.11, 95%CI: 1.20-4.17; P=0.17), C24:1Cer (OR: 2.61, 95%CI: 1.38-5.81; P=0.009), and C26:1Cer (OR: 2.2, 95%CI: 1.28-4.10; P=0.007) were associated with increased risk of TA-TMA.

Figure 1.Representative ceramide changes in all patients at baseline and day 14 after hematopoietic stem cell transplant. Concentrations of C16Cer (A), C18Cer (B), C20Cer (C), C24:1Cer (D) and C26:1Cer (E) in plasma from all patients at baseline (purple) and day 14 (blue). Data are shown as median with interquartile range.

Multivariable logistic regression models were constructed to evaluate the association between ceramide species and TA-TMA. An initial multivariate model included C26:1Cer, C16Cer, and baseline ST2, a biomarker previously reported to be associated with TA-TMA40 (Table 2). Multivariate analysis demonstrated that only baseline ST2 was associated with increased risk for developing TA-TMA (OR: 2.21, 95%CI: 1.07-5.25; P=0.047). Due to limited sample size and potential multicollinearity between ceramide species, additional models were constructed including ST2 and one ceramide species at a time to improve model interpretability (Table 2). This multivariate analysis demonstrated that C16Cer (OR: 2.77, 95%CI: 1.16-7.37; P=0.029) and C26:1Cer (OR: 2.03, 95%CI: 1.06-4.16; P=0.04) were associated with increased risk for developing TA-TMA. Overall, none of the ceramide species were better than ST2 at predicting future risk of TA-TMA, in agreement with Receiver Operating Characteristic (ROC) analyses (Online Supplementary Figure S1). We performed a similar univariate and multivariate analysis evaluating the association between ceramide levels and GvHD but did not observe any significant associations.

Pilot study of pravastatin prophylaxis during hematopoietic stem cell transplant

We wanted to identify a prophylactic strategy that could be used routinely in children to mitigate endothelial injury mediated at least in part by elevation of ceramides. Ceramides are found in LDL particles alongside cholesterol, and lowering of cholesterol through statin treatment has been shown to decrease circulating ceramide levels in humans.⁴¹ Pravastatin was chosen as previous success has been reported in reducing TA-TMA in adult HSCT recipients using pravastatin in combination with ursodiol.²²

Figure 2.Representative ceramide changes in patients with transplant-associated thrombotic microangiopathy. Concentrations of C16Cer (A), C18Cer (B), C20Cer (C), C24:1Cer (D) and C26:1Cer (E) in plasma at baseline and day 14 from hematopoietic stem cell transplant in patients with no transplant-associated thrombotic microangiopathy (TA-TMA) (blue) and TA-TMA (red). Data are shown as median with interquartile range.

Twenty-five patients with elevated BMI (median BMI above the 85th percentile for age) were enrolled in the study during HSCT, and 20 were evaluable for study endpoints. Patient disposition is shown in Online Supplementary Figure S2. Demographics of the pravastatin cohort and historical controls matched by BMI are described in Table 3. The pravastatin cohort was balanced by sex (45% female, 55% male) and was predominantly White (85%). Most patients in the pravastatin cohort underwent transplant for malignancy (50%) or marrow failure (25%), with a median age at HSCT of 15.5 years. Most patients in the pravastatin group received fully matched grafts (65%) from unrelated donors (60%) and bone marrow or peripheral blood were used equally as stem cell sources. Conditioning regimens were split evenly between myeloablative and reduced intensity approaches. GvHD prophylaxis was primarily calcineurin inhibitor (CNI)-based (75%). Intralipid was used for parenteral nutrition in most (85%) of patients. TA-TMA occurred in 55% of the pravastatin group. Acute GvHD by day 100 occurred in 30% of the pravastatin cohort with most patients developing only grade I acute GvHD (83.3%).

A median of 37 doses of pravastatin per patient were administered (IQR: 37-44). Mean medication adherence was 83.4% (IQR: 80-100%) for the entire cohort. Five patients stopped treatment early due to low medication adherence (<70%) and they were replaced in the study.

Adverse effects

Pravastatin was found to be safe and well tolerated. No attributable severe adverse events (SAE) occurred. All SAE possibly, probably, or definitely related to pravastatin treatment are described in Online Supplementary Table S3. Only one episode of grade 4 ALT elevation could possibly be attributed to pravastatin. This patient was on other additional hepatotoxic medications including acetaminophen and posaconazole. Once these agents were discontinued, the patient had complete resolution of liver enzyme elevation making involvement of pravastatin less likely. One patient had possibly related grade 3 CPK elevation but this also resolved without any issues. No other attributable toxicities occurred. We observed grade 3 acute kidney injury (N=4), grade 3 AST/ALT elevation (N=18), and grade 4 AST/ ALT elevation (N=2), all determined to be not attributed to pravastatin.

Ceramide changes in the pravastatin pilot study

We wanted to determine the effects of pravastatin on ceramide levels in this pilot study. Targeted mass spectrometry of ceramides was performed only in pravastatin-treated patients to assess longitudinal changes; control samples were not analyzed with this platform. We measured levels of specific ceramide species that were significantly altered in patients who developed TA-TMA in our preliminary analysis (Figure 3). Day 14 ceramide levels increased from baseline in patients treated with pravastatin, similar to our original observations in the cohort of 135 consecutive HSCT patients. Day 14 concentrations of C16Cer, C18Cer, C22Cer, C24:1Cer, and C26:0Cer were markedly increased from baseline in the pravastatin-treated patients, indicating that in this limited pilot study pravastatin did not reduce ceramide elevation. There was no difference in C24Cer levels at day 14 compared to baseline.

Table 2.Univariate analysis of ceramides and transplant-associated thrombotic microangiopathy.

Lipid profile changes in patients receiving pravastatin prophylaxis during hematopoietic stem cell transplant

Ceramides are closely linked to lipoprotein metabolism and transport, and statins are well known to modify plasma lipoprotein levels, so we assessed changes in lipid profiles during pravastatin prophylaxis. First, we measured HDL levels in 20 HSCT recipients who received pravastatin and in 20 HSCT recipient controls who did not receive pravastatin. Levels were measured at baseline and days 7, 14, and 35 from HSCT (Figure 4). There was no difference in baseline concentrations of HDL in pravastatin recipients (median: 38 mg/dL, range: 19-87 mg/dL) compared with controls (median: 34 mg/dL, range: 15-62 mg/dL) (P=0.44). HDL levels declined at day 7 with a nadir at day 14 after HSCT in all pravastatin recipients. The overall trend in HDL levels in controls was similar, although there was an increase in HDL levels at day 14 compared with day 7 in 5 controls. However, there was no statistically significant difference in HDL levels between the pravastatin (median: 16.5 mg/dL, range: 5-32 mg/dL) and control (median: 16 mg/dL, range: 5-65 mg/dL) (P=0.97) groups. By day 35, HDL levels were significantly higher in pravastatin-treated patients (median: 30.5, range: 6-71 mg/dL) compared to controls (median: 23 mg/dL, range: 8-53 mg/dL) (P=0.05).

To ensure that observed associations were not confounded by lipid pathway modulation from pravastatin prophylaxis, we evaluated the relationship between lipid levels and TATMA risk in patients not exposed to pravastatin (N=86). The 1-year cumulative incidence (CI) of TA-TMA was higher in patients with baseline HDL-C below the median compared to those above the median (50% vs. 29%; P=0.20) (Figure 5D), though this difference was not statistically significant. At day 14, the CI of TA-TMA was lower in patients with HDL-C, below the median compared to those above the median (29% vs. 50%; P=0.57) (Figure 5E). Among all lipid parameters, baseline LDL-C levels below the median were significantly associated with increased TA-TMA risk (P=0.01) (Figure 5A). No significant differences in TA-TMA incidence were observed based on triglyceride levels at either time point (Figure G, H).

Additional lipoprotein and triglyceride levels are presented in Online Supplementary Figure S3. Median baseline cholesterol levels were lower in pravastatin-treated patients (115.5 mg/dL, range: 39-355 mg/dL) compared with controls (144.5 mg/dL, range: 95-193 mg/dL), although this was not statistically significant (P=0.07). Day 7 cholesterol levels increased significantly from baseline in pravastatin-treated patients (median: 154.5 mg/dL, range: 103-209 mg/dL; P=0.02), while levels in controls were similar to baseline (152 mg/dL, range: 109-273 mg/dL; P=0.19). Patients treated with pravastatin had significantly elevated total cholesterol levels at day 14 (216 mg/dL, range: 77-405 mg/dL) compared to controls (150 mg/dL, range: 71-227 mg/dL) (P=0.03). There was no difference in cholesterol levels at day 35 between cases (200 mg/dL, range 200-460 mg/dL) and controls (162.5 mg/dL, range: 104-238 mg/dL) (P=0.21). Baseline triglycerides were significantly decreased in pravastatin patients (median: 78 mg/dL, range: 39-122 mg/dL) compared to controls (median: 126.5 mg/dL, range: 55-564 mg/dL) (P=0.006). Day 7 triglyceride concentrations significantly increased overall in the pravastatin group compared to baseline (median: 167 mg/dL, range 65-479 mg/dL) (P<0.0001) and increased in controls on days 14 (median 175.5 mg/dL, range: 87-979 mg/dL) and 35 (median: 280 mg/dL, range: 74-831 mg/dL). Triglyceride levels in pravastatin-treated patients decreased at day 14 compared with day 7 (median: 66 mg/dL, range: 20-134 mg/dL) and remained lower at day 35 (median: 81 mg/dL, range: 17-171 mg/dL). A large majority of children in this study received parenteral nutrition including an infusion of lipids known to alter triglyceride levels, so caution must be used in considering these data, as we cannot distinguish the effects of this infusion from changes due to the HSCT process or the use of statins.

Table 3.Demographics and transplant characteristics of pravastatin-treated patients and controls.

Baseline low density lipoprotein (LDL) levels were higher in pravastatin treated patients (median: 144 mg/dL, range: 70-228 mg/dL) compared to controls (median 77 mg/dL, range 51-118 mg/dL) (P<0.0001), although the reasons for this are unclear. Day 7 LDL levels decreased significantly in the pravastatin group (median: 93 mg/dL, range: 29-151 mg/dL; P<0.0001) from baseline but remained similar in controls (median: 83 mg/dL, range: 44-196 mg/dL; P=0.31). LDL levels at day 14 (median: 82 mg/dL, range: 22-176 mg/ dL) and day 35 from HSCT in controls remained consistently lower than pravastatin-treated patients at the same timepoints. On day 14, pravastatin-treated patients had significantly elevated LDL level (median: 129.5 mg/dL, range: 129.5-222 mg/dL) compared to controls (median: 82 mg/ dL, range: 22-176 mg/dL) (P=0.002). Similarly, on day 35, pravastatin-treated patients had significantly increased LDL levels (median: 166.5 mg/dL, range: 84-242 mg/dL) compared to controls (median: 84.5 mg/dL, range: 42-156 mg/dL) (P<0.0001).

Figure 3.Ceramide concentration changes in pilot pravastatin study. (A-F) Ceramide species quantified by mass spectrometry in pravastatin-treated patients at baseline and day 14. Concentrations of C16Cer, C18Cer, C20Cer, C22Cer, C24Cer, C24:1Cer, and C26Cer in plasma at baseline (orange) and day 14 (black) from HSCT in all pravastatin-treated patients are shown. Data are shown as median with interquartile range. No control ceramide data were available. (E and F) Notably, long-chain ceramides C24 and C26 decreased under pravastatin while other species increased.

Discussion

In this current study, we investigated ceramides as potential mediators of TA-TMA. Our study demonstrated that specific ceramide species were markedly increased at baseline and at day 14 from HSCT in patients who developed TA-TMA compared to patients without TA-TMA. Long chain ceramides (C16Cer and C18Cer) and very long chain ceramides (C20-C24Cer) are the most altered in patients with TA-TMA, in agreement with other studies that have shown ceramides to be elevated in other disease states involving immune dysregulation and inflammation.^10,11,42,43 Ceramide accumulation is believed to contribute to endothelial dysfunction and inflammation in numerous disease states but has not previously been studied in HSCT.^12,44 Increased C16Cer and C18Cer have been associated with increased cardiovascular risk by inducing endothelial dysfunction through oxidative stress, and hindering nitric oxide production leading to vascular inflammation and endothelial cell activation.^45,46 Accumulation of very long chain ceramides (C22Cer and C24:1Cer) has been associated with endothelial dysfunction through the disruption of vascular homeostasis.⁴³ These reports support our finding of increased ceramides on day 14 in patients with TA-TMA. Intriguingly, we found that ceramides were elevated prior to HSCT in patients who were later diagnosed with TA-TMA, indicating ongoing endothelial abnormality before the start of transplant. Correlative analyses revealed robust positive associations between these ceramides and ST2, a known marker of endothelial activation, but not with sC5b-9, implicating ceramides in endothelial rather than complement-mediated injury.³⁹ We considered whether measurement of ceramide level prior to transplant might be a useful biomarker of risk of TA-TMA, but univariate, multivariate and ROC analyses did not indicate increased benefit of ceramide levels over the more easily measured, existing TA-TMA risk biomarker, ST2. Statin-based endothelial prophylaxis is widely used in Europe; we considered using statins as a prophylactic strategy that would modify ceramides.²² In our pilot study, pravastatin was well-tolerated with no SAE attributable to treatment.

Figure 4.High-density lipoprotein cholesterol dynamics in pravastatin treated patients and controls. High-density lipoprotein-cholesterol (HDL-C) concentration changes are shown for cases (purple) and controls (green) at baseline (pre-conditioning), days 7, 14, and 35 after hematopoietic stem cell transplant. Each colored line represents an individual patient. The black-dashed line represents the median HDL-C concentration at each timepoint. HDL-C changes over time are represented in (A) cases and (B) controls. Each colored line represents a single patient and their respective changes in HDL-C over time. (C) Comparisons of the concentrations of HDL-C at baseline, days 7, 14, and 35 are shown.

Medication adherence was as expected in a pediatric trial of an oral therapy, with 5 patients discontinuing pravastatin due to low adherence. We saw an increase in ceramides in pravastatin recipients from baseline to day 14, similar to those seen in patients not receiving statin, suggesting that ceramide modification may not be an important part of statin-based endothelial prophylaxis.

In parallel, pravastatin treatment was associated with fluctuations in HDL-C, LDL-C, total cholesterol, and triglycerides. Interestingly, we found that lower baseline LDL levels were significantly associated with a higher cumulative incidence of TA-TMA, a finding that runs counter to conventional understanding but that is supported by emerging literature.⁴⁷, ^{4 8} Oxidized LDL (oxLDL) is known to directly mediate endothelial activation, promote leukocyte adhesion, and impair nitric oxide signaling through eNOS uncoupling.⁴⁹ However, native LDL also plays critical physiological roles, including the delivery of antioxidant-rich cargo and membrane components that may act as a buffer against oxidative and inflammatory insults. Thus, insufficient baseline LDL may reflect a state of endothelial vulnerability that predisposes to TA-TMA following transplant-related stress. This observation underscores the need for further investigation into whether extremely low LDL levels, especially in the context of immune activation and endothelial stress, may be detrimental.

Figure 5.Lower baseline low-density lipoprotein cholesterol is associated with increased 1-year cumulative incidence of transplant-associated thrombotic microangiopathy after hematopoietic stem cell transplant. Cumulative incidence of transplant-associated thrombotic microangiopathy (TA-TMA) within one year following hematopoietic stem cell transplant (HSCT) is shown stratified by lipid levels. (A, D, G) Curves represent patients with baseline levels of high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglycerides either above the median (dashed line) or below the median (solid line). (B, E, H) Curves show TA-TMA incidence based on day 14 post-HSCT levels of the same lipid parameters. A significant association was observed between low baseline LDL-C and increased TA-TMA incidence (in panel A: P=0.01), while other lipid markers did not show statistically significant associations. Differences in cumulative incidence were evaluated using Gray’s test. (C, F, I) Overall survival curves are shown.

Our study has both strengths and limitations. In our correlative biology studies, this is the first to report a link between ceramides with endothelial injury after HSCT. In parallel, the phase I pilot trial also demonstrated that pravastatin prophylaxis is feasible and safe in a pediatric population, and that it produces measurable shifts in ceramides and lipid profiles. Although using baseline values as internal controls mitigates some confounding from TPN, residual effects of intravenous lipids on circulating lipid species cannot be fully excluded. Further limitations include the fact that the correlative study was observational, and the phase I pilot study was small and non-randomized. This underscores the need for larger, randomized phase II studies incorporating longitudinal lipoprotein and ceramide profiling to validate these findings. Such trials will be essential to establish efficacy and further dissect the mechanisms by which lipid remodeling and endothelial integrity intersect in TA-TMA pathogenesis.

Footnotes

• Received August 12, 2025
• Accepted December 17, 2025

Correspondence

J. Koo jane.koo@cchmc.org

Disclosures

No conflicts of interest to disclose.

Contributions

JK and SM were responsible for study concept, design and methodology; KRS, XZ, LL, NL and EW performed the investigations; SMD and JK were responsible for visualization; SMD, SJ, KM and JK supervised the study; AL performed the statistical analysis; JK, SMD, SJ, AL, AS, KCM, DR, KRS, XZ and EW wrote the original draft, reviewed and edited the final version of the manuscript for publication.

Acknowledgments

This research would not have been possible without the valuable contributions of the patients and families who participated in these studies.