Association of infections and venous thromboembolism in hospitalized children with nephrotic syndrome

Abstract

Background

Nephrotic syndrome (NS) results in hypercoagulability and increased risk of infection. Furthermore, infection increases the risk of venous thromboembolism (VTE). Our objective was to determine the prevalence of infection, VTE, and the associated outcomes among a cohort of hospitalized children with NS.

Methods

All children with NS admitted to 17 pediatric hospitals across North America from 2010 to 2012 were included. Prevalence of infection and VTE was determined. Wilcoxon rank-sum and logistic regression were performed.

Results

Seven-hundred thirty hospitalizations occurred among 370 children with NS. One-hundred forty-eight children (40%) had ≥ 1 infection (211 episodes) and 11 (3%) had VTE. Those with VTE had infection more frequently (p = 0.046) and were younger at NS diagnosis (3.0 vs. 4.0 years; p = 0.008). The most common infectious pathogen identified was Streptococcus pneumoniae. The median hospital length of stay for those with infection [10 vs 5 days (p < 0.0001)] or VTE [22 vs 6 days (p < 0.0001)] was longer than those without either complication. Of those with infection, 13% had an intensive care unit (ICU) stay compared with 3.3% of those without infection. Median ICU stay was 4 days in those with VTE compared to 0 days in those without (p < 0.001). By logistic regression, only the number of ICU days was associated with VTE (OR 1.074, 95% CI 1.013–1.138).

Conclusions

Hospitalized children with NS have high rates of infection. Presence of VTE was associated with infection. Both were associated with longer hospitalizations and ICU stays.

Introduction

Nephrotic syndrome (NS) is the most common glomerular disorder of childhood and is defined by severe proteinuria leading to hypoalbuminemia, edema, and hyperlipidemia. Although it is known that children with NS are at greater risk for complications including infection, thrombosis, and acute kidney injury, the predisposing risk factors for these complications are poorly defined in the modern era [1–3]. In particular, NS has long been recognized as a hypercoagulable state partially due to the loss of hemostatic proteins, both procoagulant and anticoagulant [4, 5]. Compensatory protein synthesis in this disorder also results in an increase of factors V and VIII, fibrinogen, and von Willebrand factor, which promote coagulation. Thrombocytosis and platelet hyperreactivity have also been demonstrated in approximately 70% of patients and may be due to NS-induced pituitary adenylate cyclase-activating polypeptide (PACAP) deficiency [6–8]. Other reported risk factors for venous thromboembolism (VTE) in this population include the common use of central venous catheters (CVCs), diuretics, and intravascular volume depletion [9, 10]. The severity of proteinuria and age ≥ 12 years have also been shown to be associated with VTE development [9].

Patients with NS also have an elevated risk for infectious complications due to loss of components of the alternative complement pathway including factors B and D into the urine [11–13]. This loss of opsonizing factors increases susceptibility to encapsulated organisms such as Streptococcus pneumoniae, Haemophilus influenzae, and Escherichia coli, which have traditionally been the most common infectious pathogens implicated [14]. Infection is present in 17% of hospitalizations for NS, most frequently pneumonia, peritonitis, cellulitis, urinary tract infection, and bacteremia [1]. The majority of studies evaluating the microbiological spectrum of infections in NS are from before 2000. Little is known about the epidemiology for infection in children with NS after the widespread introduction of pneumococcal vaccination in the USA in 2000 [15]. Furthermore, while the risk of VTE in those with infection has been described previously, this has not been described in children with NS [16, 17].

The current study was undertaken utilizing the Midwest Pediatric Nephrology Consortium to develop a multicenter cohort of hospitalized children with NS in order to better understand the rates of complications. The first study from this cohort showed that acute kidney injury occurs commonly and is associated with adverse outcomes [18]. The current study seeks to begin to understand the rates of infection and VTE in a cohort of hospitalized children with NS. The aims of this study are to (1) determine the type and prevalence of infection, (2) describe the epidemiology of VTE, and (3) evaluate the association of VTE with infection and with hospital outcomes.

Methods

The study design was previously reported by Rheault et al. [18]. Records of all hospitalized children ≤ 18 years with NS admitted to any of 17 participating pediatric hospitals from the Midwest Pediatric Nephrology Consortium across North America from January 1, 2010 to December 31, 2012 were included. Exclusion criteria were planned admissions for renal replacement therapy, kidney biopsy, or infusion therapy; known secondary nephrotic syndrome at time of admission (IgA nephropathy, lupus, Henoch Schönlein Purpura, malignancy, etc); hospital length of stay ≤ 1 day; or remission of nephrotic syndrome on admission (urine dipstick negative or trace or protein-to-creatinine ratio < 0.2 mg/mg). The Institutional Review Board at each institution approved the study.

Data obtained

Data including demographics, clinical pattern of NS, renal biopsy results, number of hospitalizations, nephrotoxic medication exposure, and presence of infection and VTE history were recorded. Infection and VTE were defined as positive if documented by a nephrologist in the medical record per local standards. “Infections” included peritonitis, pneumonia, bacteremia, and ear, nose, and throat (ENT) infections, gastroenteritis, cellulitis/skin infection, or urinary tract infections.

Statistical analysis

Descriptive statistics were used to determine prevalence of infection and VTE. Medians, interquartile ranges (IQR), and proportions were used to summarize the data including reporting the prevalence of infection and VTE. Chi-square, Fisher’s exact, and Wilcoxon rank-sum tests were used to perform comparisons between groups. Generalized estimating equation (GEE) models were used to account for within-patient correlations when looking at hospitalization level data. Logistic regression analysis was utilized to determine variables associated with having at least one infection and also developing at least one VTE. SAS version 9.4 was used for all analyses (SAS Institute Inc., Cary, NC, USA).

Results

Seven-hundred thirty hospitalizations occurred among 370 children. Demographic and NS characteristics are included in Table 1. The median age of the cohort at diagnosis was 3 years (IQR 2, 6). The median age at first hospitalization was 6.2 year (IQR 3.3, 10.7). A total of 40.5% of children did not have a biopsy diagnosis. Otherwise, minimal change disease was the most common biopsy proven diagnosis in 30% of patients and 28% of patients were steroid resistant.

Table 1.

Demographic and clinical characteristics

	Participants N (%)
Sex (N = 370)
Male	227 (61.4%)
Race/ethnicity (N =311)
White	145 (46.6%)
Hispanic	62 (19.9%)
Black	70 (22.5%)
Other	34 (10.9%)
Clinical pattern of NS (N = 354)
Steroid sensitive-infrequently relapsing	104 (29.4%)
Steroid sensitive-frequently relapsing or	146 (41.2%)
steroid dependent	99 (28.0%)
Steroid-resistant	5 (1.4%)
Congenital or infantile
Biopsy diagnosis (N = 370)
Minimal change disease (± mild mesangial	111 (30.0%)
hypercellularity)	85 (23.0%)
FSGS or FGGS	24 (6.5%)
Membranous, MPGN, others	150 (40.5%)
No biopsy

FGGS focal global glomerulosclerosis, FSGS focal segmental glomerulosclerosis, MPGN membranoproliferative glomerulonephritis, NS nephrotic syndrome

Infection

One-hundred forty-eight children (40%) had at least 1 infection with a total of 231 infections reported in 211 hospitalizations (28.9%). Eighty-three percent of infections were present at the time of admission. The most common infection was peritonitis (8.2% of hospitalizations) with bacteremia (5.2%) also frequently observed. None of the patients with peritonitis were receiving peritoneal dialysis. When an organism was identified, S. pneumoniae was by far the most common (Table 2). Pneumonia complicated 6.8% of hospitalizations, with viral pathogens including respiratory syncytial virus, influenza, and parainfluenza frequently identified. Urinary tract infection (1.6%), cellulitis or skin infection (2.9%), gastroenteritis (3.4%), and ear/nose/throat infections (3.8%) were less commonly observed. Seven children had Clostridium difficile colitis which has not been previously reported in children with nephrotic syndrome.

Table 2.

Organisms responsible for infection in children hospitalized with nephrotic syndrome

Infection type	N (% of total hospitalizations)
Peritonitis	60 (8.2%)
S. pneumoniae (N =11)
Othera(N =4)
No organism identified (N =45)
Pneumonia	50 (6.8%)
Respiratory syncytial virus (N = 5)
Influenza (N = 4)
Human metapneumovirus (N = 3)
Parainfluenza (N = 3)
Mycoplasma pneumoniae (N = 2)
No organism identified (N =33)
Bacteremia/bloodstream	38 (5.2%)
S. pneumoniae (N = 22)
Coagulase negative Staph (N = 3)
E. coli (N = 3)
S. aureus (N = 2)
Klebsiella pneumonia (N = 2)
Otherb (N = 6)
Ear/nose/throat (sinusitis, tonsillitis, etc)	28 (3.8%)
Group A Strep (N = 2)
Otherc (N =2)
No organism identified (N = 24)
GI/gastroenteritis	25 (3.4%)
C. difficile (N = 7)
Otherd (N =2)
No organism identified (N =16)
Cellulitis/skin infection	21 (2.9%)
Varicella (N =2)
Othere (N =4)
No organism identified (N =15)
Urinary tract infection	12 (1.6%)
E. coli (N = 6)
Otherf (N =5)
No organism identified (N =1 )

^aOne each Streptococcus parasanguinis, Staphylococcus epidermidis, Citrobacter freundii, and K. pneumoniae

^bOne each S. parasanguinis, C. freundii, Enterococcus sp., Salmonella type D, Candida parapsilosis, and cytomegalovirus

^cOne each Herpes simplex virus, Candida albicans

^dOne each Campylobacter sp., C. albicans

^eOne each S. aureus, K. pneumoniae, Herpes simplex virus, Acinetobacter sp.

^fOne eachK.pneumoniae, Enterococcus sp., Enterobacter sp., coagulase negative Staph, Chlamydia trachomatis

Those with infection were younger atNS diagnosis (3.0 vs. 4.0 years;p = 0.008), and steroid-resistant NS was more highly associated with infection than all other clinical phenotypes (steroid-sensitive NS, steroid-dependent NS, other) (p = 0.003) (Table 3). There were no differences between those with and without infection regarding sex or ethnicity.

Table 3.

Patients with ≥ 1 infection vs. no infection

	Infection (N = 148)	No infection (N = 222)	p value
Age (years) at first hospitalization (median, IQR)	5.9 years (3.1, 10.5)	6.6 years (3.5, 11.3)	0.008
Sex (n, %)			0.486
Male	94 (63.5)	133 (59.9)
Race/ethnicity (n, %)			0.185
White	59 (47.2)	86 (46.2)
Hispanic	29 (23.2)	33 (17.7)
Black	21 (16.8)	49 (26.3)
Other	16(12.8)	18(9.7)
Clinical pattern of NS (n, %)			0.002
Steroid sensitive-infrequently relapsing	29 (19.6)	75 (33.8)
Steroid sensitive-frequently relapsing	26 (17.6)	32 (14.4)
Steroid sensitive-steroid dependent	31 (21.0)	57 (25.7)
Steroid-resistant	53 (35.8)	46 (20.7)
Congenital or infantile	4 (2.7)	1 (0.5)
Biopsy diagnosis (n, %)			0.441
Minimal change disease (± mild mesangial hypercellularity)	48 (32.4)	63 (28.4)
FSGS or FGGS	38 (25.7)	47 (21.2)
Membranous, MPGN, others	9 (6.1)	15 (6.8)
No biopsy	53 (35.8)	97 (43.7)
Baseline eGFR ml/min/1.73m2 (median, IQR)	129.5 (99.0, 173.1)	128.9 (95.0, 166.4)	0.633
Number of hospitalizations (median, IQR)	2 (1,3.5)	1 (1,2)	<0.001
Overall hospital days (median, IQR)	10(6,22)	5 (3,8)	<0.001
ICU days (median, IQR)	0 (0, 0)	0 (0, 0)	<0.001

NS nephrotic syndrome, eGFR estimated glomerular filtration rate, ICU Intensive Care Unit, FGGS focal global glomerulosclerosis, FSGS focal segmental glomerulosclerosis, MPGN membranoproliferative glomerulonephritis

Venous thromboembolism

Eleven (3%) children had VTE including 3 with central venous sinus thrombosis, 1 with pulmonary embolism, and 1 with bilateral renal vein thrombosis. Four of the VTEs were diagnosed on the day of admission, with the remainder developing during the hospital stay (median hospital day 4, range 1–11). Four of the 11 episodes of VTE were associated with indwelling catheters. Those with VTE had more infections (p = 0.046) (Table 4). Infections associated with VTE were C. difficile (1 subject), methicillin-sensitive Staphylococcus aureus (2), S. pneumoniae (1), and unidentified pathogens (3). For six patients with both infection and thrombosis identified during the hospital stay who had dates of diagnosis available, 3 developed infection prior to thrombosis (range 1 −14 days) and 1 patient was simultaneously diagnosed.

Table 4.

Patients with VTE vs. no VTE

	VTE (N =11)	No VTE (N = 359)	p value
Age in years at first hospitalization (median, IQR)	6.8 (2.6, 10.7)	6.2 (3.4, 10.7)	0.869
Sex (n, %)			0.348
Male	5 (45.5)	222 (61.8)
Race/ethnicity (n, %)			0.369
White	6(75)	139 (45.9)
Hispanic	0 (0)	62 (20.5)
Black	1 (12.5)	69 (22.8)
Other	1 (12.5)	33 (10.9)
Clinical pattern ofNS (n, %)			0.842
Steroid sensitive-infrequently relapsing	3 (27.3)	101 (28.1)
Steroid sensitive-frequently relapsing	2 (18.2)	56(15.6)
Steroid sensitive-steroid dependent	2 (18.2)	86 (24.0)
Steroid-resistant	3 (27.3)	96 (26.7)
Congenital or infantile	0(0)	5 (1.4)
Biopsy diagnosis (n, %)			0.438
Minimal change disease (± mild mesangial hypercellularity)	6 (54.6)	105 (29.3)
FSGS or FGGS	2 (18.2)	83 (23.1)
Membranous, MPGN, others	0 (0)	24 (6.7)
No biopsy	3 (27.3)	137 (41.0)
Baseline eGFR ml/min/1.73m2 (median, IQR)	119.2(77.4, 131.5)	129.1 (97.1, 171.9)	0.086
Number of hospitalizations	2 (1,3)	1 (1,2)	0.412
(median, IQR)
Overall hospital days	22 (16, 26)	6 (3, 11)	<0.001
ICU days	4 (0, 15)	0 (0, 0)	<0.001
Number of infections	1 (0, 2)	0 (0, 1)	0.046

VTE venous thromboembolism, NS nephrotic syndrome, eGFR estimated glomerular filtration rate, ICU intensive care unit, FGGS focal global glomerulosclerosis, FSGS focal segmental glomerulosclerosis, MPGN membranoproliferative glomerulonephritis

Outcomes

Children with VTE, infection, or both also experienced significantly longer hospital stays. The median hospital stay for those without infection was 5 days vs. 10 in those with infection (p <0.001) (Table 3). Similarly, median hospital stay for those without VTE was 6 days as compared to 22 in those with VTE (p <0.001). Of those with infection, 13% had an ICU stay compared with 3.3% of those without. Those with VTE also had a median of 4 days in the intensive care unit (ICU) as compared to 0 days in those without VTE (p <0.001) (Table 4). There was no difference between those with VTE or infection and those without in admission serum albumin, admission protein to creatinine ratio, or percent usage of calcineurin inhibitor at admission (Supplemental Tables 1 and 2).

In a logistic regression model looking at hospitalizations, hospital days, and ICU days, only the number of ICU days was associated with the presence of VTE (OR 1.07, 95% CI 1.01– 1.14). When looking at the number of hospitalizations, the number of hospital days and the number of ICU days in a logistic regression model, hospitalizations (OR 1.61,95% CI 1.21,2.14), and hospital days (OR 1.04,95% CI 1.01,1.08) were found to be significantly related to having at least one infection.

Discussion

This study provides a comprehensive description of infectious and thrombotic complications in children hospitalized with NS in a contemporary US cohort. Although previous studies have defined rates of infections and VTE among this population, little is known about the relationship between infection and VTE development. Our study confirms that hospitalized children with NS have high rates of infection (40%) and thrombosis (3%) and these complications are associated with longer length of stay and ICU care. We make the novel observation that there is an association between infection and thrombosis in children hospitalized with NS. The most common infectious pathogen identified was S. pneumoniae, which is unchanged from historical studies in pediatric nephrology.

Peritonitis, pneumonia, and sepsis are common infections encountered in hospitalized children with NS [2]. Despite the widespread introduction of the pneumococcal vaccine in 2000, the most common pathogen identified in this cohort remained S. pneumoniae. From our data, we were unable to determine vaccination status. However, PCV7 (before 2010) or PCV13 (after 2010) were recommended as part of the routine vaccine series in children starting at 2 months of age, and the polyvalent PPSV23 vaccine is recommended by the CDC for all children with NS over the age of 2 years [19]. In children with NS with a high suspicion for infection, empiric antibiotic coverage should continue to include activity against S. pneumoniae. Children with steroid-resistant NS were more likely to have infection than other types of NS, possibly due to severity of proteinuria or more aggressive immunosuppressive therapy [20]. The role of immunosuppression in increasing infection risk versus the infectious risk of nephrotic syndrome itself is unclear and deserves further study. Unfortunately, however, immunosuppression history was not recorded in this database and thus could not be directly assessed for any correlation with infection risk.

This study confirms the correlation between infection and VTE in children with NS. Recently, our group identified infection as a statistically significant modifier for likelihood of thrombosis in children with chronic kidney disease (including NS) [21]. There is a complex interplay between the immune system, infectious agents, and the coagulation system which can lead to thrombosis. Exposure of tissue factor due to endothelial damage by infectious agents, microparticle formation, and the development of neutrophil extracellular traps (NETs) have all been linked to VTE [22–24]. Additionally, particular infectious agents have been shown to lead to an increased risk of thrombosis. In particular, staphylococcal infections are common among children and have a known association with thrombotic disease [16, 17]. S. aureus may carry virulence factors such as (Panton-Valentine leukocidin [pvT]) that are capable of activating the coagulation system and, thus, enhance its propensity to stimulate thromboembolic phenomena [25]. Furthermore, methicillin-resistant S. aureus strains are more likely to be pvl positive. Other S. aureus virulence factors including surface-associated adhesion (sdrD) and fibronectin-binding protein B (fnbB) may also be important in pediatric infection-associated thrombosis [26]. Of note, though S. pneumoniae was identified more commonly in the entire cohort with infection, S. aureus accounted for 2 of the 3 cases of VTE associated with infection where the pathogen was known, and for 2 of the 11 total patients with VTE. Thus, children with NS and infection, particularly S. aureus infection, are at high risk of thrombosis and a high index of suspicion for thrombosis should be maintained. Targeted VTE prophylaxis may be indicated in this subset of patients; however, further prospective studies are required given risks of anticoagulation in children.

Previous studies identified age > 12 years and severity of proteinuria as risk factors for VTE in children with nephrotic syndrome [9]. In the present study, there was no association between age, serum albumin, or urine protein-to-creatinine ratio at the time of admission and VTE (Supplemental Table 2). However, 43% of children did not have urine protein-to-creatinine ratio obtained on admission and this may have influenced the results. Additionally, prior studies included both inpatient and outpatient VTE events and were limited to a few major academic institutions and may not be comparable to the population included in the present manuscript.

Though the overall rate of VTE in this population was low (3%), this is still greater than the rate of patients without NS, and patients with VTE had significantly longer hospitalizations and were more likely to have been admitted to the ICU. We were not able to ascertain from our data whether VTE was the cause of the subjects’ ICU admissions. Previous studies of hospital-acquired VTE have identified ICU admission as an independent risk factor for this complication; therefore, we are unable to determine whether the VTE occurred as a consequence of infection associated with NS or due to interventions specific to ICU care [27]. Patients admitted to ICU often have central lines placed and have higher rates of impaired mobility compared to other hospitalized patients, both of which have been shown to be risk factors for hospital-acquired VTE [9]. Central lines should be avoided in children with nephrotic syndrome if possible.

Limitations of our study include its retrospective nature as well as the low number of VTE events recorded. Additionally, we are unable to determine from our data all of the possible risk factors for VTE or infection in this population. Identification of VTE relied on the practitioners’ concern for the diagnosis. Prior publications have shown that 35% of otherwise asymptomatic adults and children with nephrotic syndrome may have computed tomography (CT) evidence of pulmonary embolism or renal vein thrombosis [28]. Therefore, some VTE may have been missed in this cohort and the rate thus underestimated. Rates of inherited thrombophilia also were not available for this population. Immunosuppressive intensity and antimicrobial therapy data were not collected, limiting our understanding of how immune status and local antibiograms may alter infection risk. In addition, comprehensive data on the past and current CVC use were not available, limiting our ability to assess the impact of CVC use on VTE risk.

Conclusions

Children with NS who are hospitalized have high rates of infection. In this cohort, the presence of VTE was associated with infection. Both infection and VTE were associated with longer hospitalizations and intensive care unit stays. S. pneumoniae remains the most commonly identified bacterial pathogen in children with NS despite contemporary immunization strategies. However, methicillin-sensitive S. aureus was identified in 2 of 11 patients with VTE. Research in pediatric nephrotic syndrome should focus not only on immunosuppressive agents to prevent or treat relapses, but also on treatment and prevention of these common complications. Further studies are needed to identify potentially modifiable risk factors that could minimize these complications in this already high-risk population, and to identify which patients may benefit most from prophylactic anticoagulation.