Introduction
Despite advances in surgical techniques and perioperative management, infants undergoing cardiac surgery with cardiopulmonary bypass remain at increased risk of morbidity and mortality compared with older children. Reference Brancaccio, Villa and Girolami1–Reference Kubicki, Grohmann and Siepe4 A feared postoperative complication of cardiopulmonary bypass in neonates and infants is severe systemic inflammatory response syndrome, leading to capillary leak syndrome and organ dysfunction, resulting in delayed postoperative recovery and increased morbidity and mortality. Reference Brancaccio, Villa and Girolami1,Reference Seghaye, Grabitz and Duchateau3,Reference Seghaye, Duchateau and Grabitz6–Reference Anderson, Bond and Joffe8 Defined by the pathological migration of fluid and proteins from the vascular compartment into the interstitial space, capillary leak syndrome manifests as generalised oedema and ascites, leading to a cascade of morbidity that can delay recovery and increase mortality. Reference Kubicki, Grohmann and Siepe4 The syndrome is notoriously difficult to manage, often failing to respond to conventional diuretic therapies and complicating weaning from mechanical ventilation. Reference Stiller, Sonntag and Dähnert5,Reference Hassinger, Wald and Goodman9–Reference Saravi, Goebel and Hassenzahl11
While previous research has focused predominantly on inflammatory mediators and complement system abnormalities, investigating the mechanical factors associated with open chest surgery could provide new insights into the development of capillary leak syndrome and lead to better preventive strategies. It is noteworthy that capillary leak syndrome is usually diagnosed clinically after open-heart surgery with cardiopulmonary bypass, without direct confirmation through protein leakage from the circulatory system. Reference Tassani, Schad and Winkler12 The demanding nature of invasive cardiac surgery involving sternotomy, cardiopulmonary bypass, and delayed sternal closure places considerable stress on patients. This stress is compounded by anaesthetics, intravenous medications, vascular injury, and, notably, surgical stress from open-chest mechanical factors, all of which are thought to modulate the inflammatory response and consequently the manifestation of capillary leak syndrome—mechanisms that remain to be fully elucidated. Reference Brancaccio, Villa and Girolami1,Reference Kubicki, Grohmann and Siepe4,Reference Tárnok and Schneider13,Reference Moat, Rebuck, Shore, Evans and Finn14
The central role of cardiopulmonary bypass and open chest management in cardiac surgery for CHD, along with the scarcity of research on the mechanical factors associated with these procedures, necessitates an investigation into how such factors influence the occurrence of postoperative capillary leak syndrome. In this study, we aimed to investigate the relationship between mechanical factors specific to delayed sternal closure and the subsequent development of capillary leak syndrome. More specifically, we evaluated whether the width of the sternal opening during open chest management was associated with the occurrence of clinically significant capillary leak syndrome and consequently an increase in mortality.
Material and methods
Study population
We performed a retrospective review of clinical data of paediatric patients (aged less than 12 months) who underwent open-heart surgery with cardiopulmonary bypass and delayed sternal closure at our institution between January 2016 and December 2021. At our institution, delayed sternal closure is considered a routine management strategy in cases of postoperative haemodynamic instability. The opening of the sternum after stenting with a modified plastic tube is shown in Figure 1A. Stent length was determined at the discretion of the individual surgeon, with care taken to select a length adequate to relieve myocardial and pulmonary oedema while maintaining sufficient chest stability. Patients were divided into two groups according to the postoperative diagnosis of capillary leak syndrome. Capillary leak syndrome was defined by a paediatric ICU physician in patients with delayed sternal closure of more than three days, adhering to general and local criteria (significant subcutaneous or generalised oedema, negative fluid balance despite adequate fluid resuscitation, and recurring clinical presentation of hypotension, haemoconcentration, and hypoalbuminemia). Patients whose physical signs, symptoms, and laboratory results could be explained by an alternative diagnosis were excluded. No patients requiring mechanical circulatory support postoperatively were included, as their inclusion would likely have introduced a strong confounding effect. Postoperative anteroposterior chest radiographs were used to assess the presence of subcutaneous oedema or other signs suggestive of capillary leak syndrome. Data were obtained from the electronic health records as well as surgical and anaesthesia reports. The complexity of the CHD was categorised according to Bethesda classification. Reference Warnes, Liberthson and Danielson15 Surgical procedures were classified according to the procedure-adjusted Society of Thoracic Surgeons-European Association for Cardio-Thoracic Surgery (STAT) score Reference O’Brien, Clarke and Jacobs16 . Mortality was divided into two categories: early mortality, defined as any mortality that occurred within the first 30 days after surgery, including those after discharge from the hospital, and late mortality, defined as any mortality that occurred after the 30-day postoperative period, regardless of whether it occurred in-hospital or post-discharge.
Figure 1. Sternal opening after stenting with a modified plastic tube ( a ) and postoperative anteroposterior chest radiograph from the same patient ( b ) (stent is marked with an arrow).
Perfusion strategy and cardiopulmonary bypass management
All cardiopulmonary bypass procedures were conducted using a Dideco perfusion tubing system (LivaNova, Milan, Italy), composed of polyvinyl chloride and silicone components. To minimise haemodilution, circuits were miniaturised, with a target haematocrit (Hct) of ≥ 30% during bypass. Hollow fibre oxygenators with integrated arterial line filters were used, coated with either Xcoating™ (Terumo Europe, Leuven, Belgium) or Softline™ (Maquet Cardiopulmonary AG, Hechingen, Germany). A Stöckert S5 heart-lung machine (LivaNova, Munich, Germany) with roller pumps and a hard-shell venous reservoir was employed throughout.
In patients weighing less than 12 kg, the circuit was primed with leukocyte-reduced red blood cells (collected within 5 days), heparin, sodium chloride, mannitol, 20% albumin, and Addex-Tham (3.3 mmol/mL) as a buffering agent. For neonates under 10 days of age, plasma (10 mL/kg) replaced albumin as the colloid component, and calcium chloride was added to the prime. Prime volumes were individualised based on the patient’s preoperative haemoglobin, with the aim of maintaining a Hct ≥ 30% during cardiopulmonary bypass. Continuous ultrafiltration was performed throughout bypass, targeting a final Hct of 32–35% at separation from cardiopulmonary bypass.
All patients under 10 kg received a preoperative dose of dexamethasone (10 mg) as standard. Additional corticosteroids were administered selectively in cases of known immunodeficiency (e.g., 22q11 deletion) or during complex procedures such as total cavopulmonary connection. As part of the anti-inflammatory strategy, tranexamic acid was administered at anaesthesia induction (50 mg/kg), followed by a continuous infusion (5 mg/kg/h) until one hour postoperatively. An additional bolus dose of tranexamic acid (50 mg/kg) was added to the cardiopulmonary bypass circuit at initiation of bypass.
Quantitative measurements
Postoperative anteroposterior chest radiographs were analysed using Sectra IDS7 22.2 software (Sectra AB, Linköping, Sweden). Measurements included the transverse thoracic diameter, which was determined by measuring the greatest distance between the medial aspects of the ribs (i.e. pleura to pleura). The stent length to thoracic ratio (SL/T) was derived for each patient, normalising stent length against transverse thoracic diameter to achieve standardisation across different patient sizes. Similarly, stent length per kilogram (SL/kg) was calculated for a body weight-adjusted assessment. In patients clinically diagnosed with postoperative capillary leak syndrome, additional measurements included left and right subcutaneous oedema at the fifth rib level, which contributed to the mean subcutaneous oedema and subcutaneous to thoracic ratio (S/T). The latter ratio was determined by subtracting the proportional sum of bilateral subcutaneous oedema from the transverse thoracic diameter, according to the methodology described by Sonntag et al. Reference Sonntag, Grunert, Stöver and Obladen17 The SL/T and S/T ratios were chosen to ensure standardisation of the measurements of stent length and subcutaneous oedema, respectively. If no capillary leak syndrome was present, radiographs taken between 18 and 48 hours after surgery were selected for the measurements. In cases where capillary leak syndrome was present, radiographs taken at the time of diagnosis were used. The radiologist’s assessment of the presence of subcutaneous oedema or other signs of capillary leak syndrome was included as a variable when available, but its absence was not an exclusion criterion.
Statistical analysis
Continuous variables were reported as medians and interquartile ranges as appropriate to the data distribution. Nominal data were expressed as frequencies and percentages. Statistical analyses were performed using GraphPad Prism for Mac, version 9.5 (GraphPad Software Inc., San Diego, CA, USA). Group comparisons were performed using the Wilcoxon rank-sum test or unpaired t-test to determine statistically significant differences between patients with and without postoperative capillary leak syndrome. Statistical significance was set at p < 0.05. All reported p-values are two-sided. Logistic regression analyses for risk factors were performed using mortality as the dependent variable. Univariable Kaplan–Meier analyses were generated to show event-free rates. Receiver-operating characteristic curve analysis was performed to assess the sensitivity and specificity of SL/kg in predicting mortality and the development of capillary leak syndrome. The optimal cutoff point was determined using the Liu method, which identifies the value that maximises the product of sensitivity and specificity.
Results
Patients and group comparison
We identified 164 patients who underwent 172 surgical procedures. Comprehensive descriptive statistics are presented in Table 1. A total of 12 patients (7.3%) met the eligibility criteria mentioned in 2.1 and are listed in Supplementary Table S1 . Ten cases with suspected capillary leak syndrome were excluded due to early sternal closure. A total of 54.9% of patients were categorised as Bethesda great CHD complexity, and 64.0% underwent highly complex procedures (STAT category 4-5). Preterm birth occurred in 23 patients (14.0%).
Table 1. Clinical characteristics of patients undergoing open chest cardiac surgery with delayed sternal closure
CLS = capillary leak syndrome; n = number of patients; PICU = paediatric ICU; Data presented as number (n) (percentage) or median (inter-quartile range); * statistically significant (p < 0.05).
A comparison between capillary leak syndrome and non-capillary leak syndrome patients revealed significant differences. Capillary leak syndrome patients had a median weight at surgery of 3.2 kg (IQR 2.8–3.6 kg), lower than the 3.4 kg (IQR 3.0–3.7 kg) of non-capillary leak syndrome patients (p = 0.27). The median age at surgery was 6.0 days (IQR 4.3–7.0 days) for capillary leak syndrome patients and 7.0 days (IQR 5.0–14.0 days) for non-capillary leak syndrome patients (p = 0.25). Cardiopulmonary bypass time was longer in capillary leak syndrome patients with a median of 212 minutes (IQR 202–265 minutes) compared to 188 minutes (IQR 161–221 minutes) in non-capillary leak syndrome patients (p = 0.01). In addition, capillary leak syndrome patients had a longer length of stay with an open sternum, with a median of 5.8 days (IQR 4.6–8.4 days) compared to 2.0 days (IQR 2.0–3.0 days) in non-capillary leak syndrome patients (p < 0.001). Capillary leak syndrome patients also had a significantly longer median duration of inotrope use, chest drainage and invasive ventilation compared to non-capillary leak syndrome patients. These results are detailed in Table 1.
Radiographic analysis and group comparison
Radiographic analysis showed that capillary leak syndrome patients had higher median SL/kg (14.6 mm, IQR 11.5–16.9) compared to non-capillary leak syndrome patients (12.2 mm, IQR 10.5–14.7) (p = 0.02). An SL/kg ≥ 15.0 mm had moderate sensitivity (50%) and high specificity (78%) for identifying patients who developed capillary leak syndrome postoperatively. There was no significant difference between the two groups for median stent length (43.5 mm, IQR 39.7–49.4 vs. 41.8 mm, IQR 36.0–46.8, p = 0.29), median transverse thoracic diameter (111.7 mm, IQR 105.0–116.21 vs. 111.5 mm, IQR 103.3–117.5, p = 0.50), and median SL/T (0.42, IQR 0.35–0.44 vs. 0.37, IQR 0.32–0.42, p = 0.61). In the capillary leak syndrome group, the median S/T was 0.83 (IQR 0.77–0.85).
Logistic regression analysis for risk factors
Nine cases of postoperative mortality were observed, representing 5.5% of the cohort. Mortality was 25.0% in the CLS group vs. 3.9% in the non-CLS group (p = 0.02). The overall survival rates at 30 days and one year were 98.2% and 95.0%, respectively (Figure 2A).
Figure 2. Kaplan–Meier curves for overall postoperative survival and group postoperative survival ( a ) with corresponding number of patients at risk ( b ).
Among postoperative mortalities, higher median SL/kg was observed, with a mean difference of 2.0 mm, which was statistically significant (p = 0.04). Univariate logistic regression analysis identified increased SL/kg as a significant predictor of postoperative mortality, indicated by an OR of 1.17 (95% CI: 1.00–1.38). An SL/kg ≥ 13.8 mm demonstrated high sensitivity (89%) and moderate specificity (61%) for identifying patients who succumbed postoperatively. In addition, multivariate analysis revealed that greater SL/kg (OR: 1.28, 95% CI: 1.05–1.58) and longer postoperative paediatric ICU length of stay (OR: 1.11, 95% CI: 1.03–1.21) were associated with a higher likelihood of postoperative mortality. The data are presented in Table 2.
Table 2. Risk factors for mortality: results of univariate and multivariate logistic regression analyses
SL/kg = stent length per kilogram; PICU = paediatric ICU; * statistically significant (p < 0.05).
Discussion
The prevalence of capillary leak syndrome after cardiac surgery with cardiopulmonary bypass in paediatric patients varies across studies. Seghaye et al. Reference Seghaye, Grabitz and Duchateau3 , Stiller et al. Reference Stiller, Sonntag and Dähnert5 , and Zhang et al. Reference Zhang, Wang and Li18 reported a capillary leak syndrome prevalence of 54.2% (13 of 24 patients), 5.2% (3 of 58 patients), and 37.0% (10 of 27 patients), respectively. However, it is important to note that these studies were conducted on relatively small cohorts. In our cohort of 164 neonates and infants, 7.3% (12 patients) met the capillary leak syndrome criteria. This comparatively lower prevalence could be due to advances in surgical techniques, perioperative care, and specific criteria for the diagnosis of capillary leak syndrome. The large proportion of patients in our cohort with high STAT scores and in higher Bethesda risk categories suggests that the lower capillary leak syndrome prevalence does not necessarily reflect fewer complex cases.
It has been concluded from previous studies that severe protein extravasation is likely to occur at first within hours postoperatively, coinciding with the peak of the inflammatory response. Reference Seghaye, Grabitz and Duchateau3,Reference Levy and Kelly19 In addition to cardiopulmonary bypass, anaesthetics, reperfusion injury to the myocardium and lungs, and surgical trauma have been recognised as significant triggers of inflammation. Reference Suleiman, Zacharowski and Angelini20–Reference Prondzinsky, Knüpfer and Loppnow23 We propose that mechanical factors, including SL/kg, may also play a role in exacerbating postoperative capillary leak syndrome by affecting systemic venous return. This, in turn, could contribute to increased egress of interstitial fluid and influence the patient’s postoperative course. This hypothesis draws upon the insights of Corno et al. Reference Corno24 who, while not directly investigating the effect of stent length during open chest management on systemic venous return, emphasised the importance of surgical techniques and decisions and their mechanical effects on the venous circulation. In addition, a study by Haddad et al. Reference Haddad, Lange and Raisky25 investigated the impact of delayed sternal closure on postoperative outcomes and associated mortality and found it to be a significant risk factor for mortality after postoperative cardiac catheterisation in children. In our study, a significant difference was found, suggesting a higher SL/kg in patients who developed capillary leak syndrome. Notably, there was also a significant difference in SL/kg between patients who succumbed postoperatively and those who survived, with the former group having a higher SL/kg. Surgical stress, which has been identified as a risk factor for the development of an inflammatory response and postoperative complications, Reference Wan, Arifi and Wan21–Reference Prondzinsky, Knüpfer and Loppnow23 is thought to play a role in this context.
While we initially hypothesised that longer stents might impair venous return, this relationship appears complex and multifactorial. Theoretically, increased sternal opening could reduce intrathoracic pressure and thereby facilitate venous return. However, our data suggest that the adverse effects associated with higher SL/kg outweigh these potential benefits. One possible explanation is that longer stents may induce anatomical distortion and increase tension on the great veins and right atrium, leading to elevated central venous pressure and impaired preload augmentation. Moreover, prolonged open chest management with excessive sternal separation may further compromise cardiopulmonary interactions, exacerbating right ventricular dysfunction, venous congestion, and interstitial fluid accumulation.
It is also possible that longer stents are not causative per se, but rather reflect the underlying severity of the patient’s condition. Patients requiring wider sternal separation often present with greater haemodynamic instability and require more extensive postoperative support and prolonged intensive care. Thus, SL/kg may serve as an accompanying factor or a surrogate marker of overall illness severity, rather than an independent risk factor for capillary leak syndrome or mortality. Future studies incorporating detailed haemodynamic assessments could help clarify the interplay between stent length, venous return, and postoperative outcomes.
Importantly, SL/kg emerges as a predictive parameter for postoperative mortality in neonates and infants undergoing open-heart surgery with cardiopulmonary bypass and delayed sternal closure. The length of the stent is determined by the surgeon based on intraoperative conditions, including thoracic compliance, haemodynamic stability, and overall patient status. Therefore, the association between higher SL/kg and worse outcomes may, at least in part, reflect the underlying severity of illness rather than a modifiable surgical factor. Previous studies have reported a higher incidence of capillary leak syndrome and multiorgan failure in neonates and infants than in older children, Reference Seghaye, Grabitz and Duchateau3,Reference Stiller, Sonntag and Dähnert5,Reference Seghaye, Duchateau and Grabitz6 suggesting that immaturity of the membrane response and development of the serological system contribute to this trend. Reference Stiller, Sonntag and Dähnert5 Younger age has also been recognised as a risk factor for increased total body water retention after cardiac procedures. Reference Maehara, Novak, Wyse and Elliot26,Reference Elliott27
Extended cardiopulmonary bypass times have been associated with greater endothelial injury and postoperative capillary leak syndrome, likely due to more pronounced host dysregulation. Reference Kubicki, Grohmann and Siepe4,Reference Allan, Newburger and McGrath28–Reference Wollborn, Zhang and Gaa30 Both Stiller et al. Reference Stiller, Sonntag and Dähnert5 and Meri et al. Reference Meri, Aronen and Leijala29 have found correlations between cardiopulmonary bypass time and capillary leak syndrome, with the latter noting a linear increase in complement activation that correlated with perfusion time and postoperative complications. Given the relatively low operative mortality rate observed in our study, we did not find a significant association between cardiopulmonary bypass time and all-cause mortality in the multivariate analysis.
Our findings suggest several hypotheses for the association between higher SL/kg and capillary leak syndrome or mortality. One potential mechanism is that longer stent lengths could exacerbate capillary leak syndrome by further impeding systemic venous return and increasing interstitial fluid leakage in this group of critically ill patients. Alternatively, the inflammatory response triggered by surgical stress and prolonged cardiopulmonary bypass time could be more pronounced in patients requiring longer stents.
In addition, we retrieved data on left and right ventricular function in patients with capillary leak syndrome and found that only 2 out of 12 cases had more than mild reductions in left ventricular and/or right ventricular systolic function at the time of surgery. This finding suggests that the association between stent length and outcomes such as capillary leak syndrome or mortality is unlikely to be attributable to significant pre-existing systolic dysfunction in most patients. However, it is important to acknowledge that impaired diastolic function, rather than systolic dysfunction, may be a key contributor to capillary leak syndrome, as it can lead to impaired venous filling. Unfortunately, diastolic function is challenging to assess in this setting. Thus, while our data suggest that systolic dysfunction is unlikely to be a major factor in the observed associations, the potential role of diastolic dysfunction remains an important, yet difficult-to-measure, consideration in this patient population.
Limitations
Limitations of the study include its single-centre nature, the relatively small to moderate sample size, and the retrospective design, which limits control for exposure factors, potential confounders, and standardisation of treatment. The diagnosis of capillary leak syndrome relied on clinical documentation from electronic health records, which may lead to potential variability in patient inclusion, as the presence of capillary leak syndrome may not have been consistently documented across all cases. In addition, setting the threshold for delayed sternal closure at more than three days may have introduced selection bias by excluding cases with shorter durations of delayed sternal closure that may nevertheless have met the clinical criteria for capillary leak syndrome. The rationale for setting this threshold was to focus on patients with significant postoperative instability while minimising heterogeneity in the cohort.
A further limitation is the absence of a comparative analysis with patients undergoing primary sternal closure. While such a comparison could yield additional insights, our study was specifically designed to investigate risk factors within the delayed sternal closure cohort. This was a deliberate methodological choice, allowing us to focus on this high-risk group. Future research may benefit from a broader analysis that includes multiple patient cohorts.
It would have been ideal to control for common risk factors associated with oedema formation after cardiac surgery, such as impaired left ventricular and/or right ventricular systolic function, diastolic dysfunction, central venous pressure, and pulmonary compliance. Unfortunately, these data were not consistently available for all patients in this retrospective study. However, we were able to retrieve data on left ventricular and right ventricular systolic function at the time of surgery for the capillary leak syndrome cases and found that only 2 of 12 patients had more than mildly reduced left ventricular and/or right ventricular systolic function. Future studies could benefit from extending the study period or including multiple centres to increase the sample size and statistical power, particularly in relation to the study of early mortality. An independent assessment of capillary leak syndrome cases by an additional experienced reviewer was chosen to minimise detection and selection bias.
Conclusions
In conclusion, capillary leak syndrome after open-heart surgery in neonates and infants with delayed sternal closure is associated with postoperative morbidity. Increased SL/kg and prolonged paediatric ICU length of stay are associated with mortality. It is also possible that longer stents are not causative per se, but rather reflect the underlying severity of the patient’s condition. Patients requiring wider sternal separation often present with greater haemodynamic instability and require more extensive postoperative support and prolonged intensive care. Thus, SL/kg may function as an accompanying factor or a surrogate marker of overall illness severity, rather than as an independent risk factor for capillary leak syndrome or mortality.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1047951125101455
Acknowledgements
None.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
None.
Ethical standards
Data collection was approved under a permit for the purpose of retrospective electronic health record studies at the Department of Clinical Sciences, Paediatric Cardiology, Lund University, by the Regional Ethical Committee for Human Research, Lund, Sweden (DNR 2015/559). All patient data was securely encrypted and accessed under strict confidentiality protocols.
Open access
