Objective. Severe pneumococcal infections have been associated with hemolytic uremic syndrome (HUS), usually with a poor clinical outcome when compared with Escherichia coli O157 gastroenteritis-associated (D+) HUS. We examined our experience with 12 cases of Streptococcus pneumoniae-associated HUS (SP-HUS) and compare it with a cohort of diarrhea-associated HUS (D+ HUS).
Methods. A retrospective case survey compared 2 unrelated groups of HUS patients. Demographic factors, clinical indices of disease severity, and outcome were used to compare the 2 groups of HUS patients.
Results. Twelve children with SP-HUS were studied. Pneumococcal pneumonia with empyema was the most common precipitating illness (67%), pneumococcal meningitis was present in 17% of children, pneumonia with bacteremia in 8%, and both pneumonia and meningitis in 8%. SP-HUS patients were younger than D+ HUS patients (22.1 vs 49 months) and had more severe renal and hematologic disease than D+ HUS patients. Compared with D+ HUS patients, SP-HUS patients were more likely to require dialysis (75% vs 59%) and had a longer duration of hospitalization (33.2 vs 16.1 days) and duration of thrombocytopenia (11.6 vs 6.8 days). SP-HUS patients were also more likely to require platelet transfusions (83% vs 47%) and needed more platelet (4.7 vs 0.5) and packed red blood cell transfusions (7.8 vs 2.0). The 2 groups did not differ significantly in the incidence of extrarenal HUS complications. There were no deaths in either group. Seven patients have been seen for long-term follow-up; 2 developed end-stage renal disease, and 5 have normal renal function.
Conclusions. HUS is a rare but severe complication of invasive pneumococcal infection. Although disseminated intravascular coagulation can also occur in these children, the treatment and follow-up may be different in the 2 conditions. Children with pneumococcal disease and severe hematologic or renal abnormalities should be investigated for evidence of HUS.
- hemolytic uremic syndrome
- Streptococcus pneu-moniae
- acute renal failure
Hemolytic uremic syndrome (HUS) is one of the most common causes of acute renal failure in children.1 The vast majority of HUS (90%) occurs after acute gastroenteritis with enterotoxigenic Escherichia coli and is termed typical, or diarrhea-associated, HUS (D+ HUS).2 Outcomes from D+ HUS generally are good. Survival is >95%, and long-term morbidity is seen in <30% of patients.2 A variety of more rare causes of HUS exist. These more rare causes are collectively termed “atypical HUS” and as a group have a much higher mortality (approximately 25%) and long-term morbidity (approximately 50%).3 An increasingly recognized cause of atypical HUS is invasive Streptococcus pneumoniae infection. S pneumoniae may also lead to classic disseminated intravascular coagulation (DIC). HUS and DIC may present with similar findings of thrombocytopenia, microangiopathic anemia, and renal insufficiency. HUS is distinguished from DIC by a normal or elevated fibrinogen level and by normal or only slightly elevated prothrombin time and partial thromboplastin time. The renal failure of DIC is attributable to acute tubular necrosis, whereas HUS causes, predominantly, a thrombotic microangiopathy of renal vasculature, with or without tubular necrosis. S pneumoniae-associated HUS (SP-HUS) has historically been characterized by a high morbidity and mortality rate.4,5 We report 12 patients with SP-HUS who have survived, describe their course and outcome, and compare them with a group of D+ HUS patients.
Patients were children who were hospitalized with a history of HUS associated with a culture-proven S pneumoniae infection at the Children’s Hospital of New Mexico (Albuquerque, NM), Children’s Hospital and Medical Center (Seattle, WA), Children’s Mercy Hospital (Kansas City, MO), and Texas Children’s Hospital (Houston, TX) between 1990 and 1999. SP-HUS patients were compared with a cohort of 17 consecutive, nonepidemic, E coli O157-associated D+ HUS patients who were hospitalized at Children’s Hospital and Medical Center in between January 1 and December 31, 1996.
Patient records were reviewed to determine age, gender, ethnicity, site of pneumococcal infection, days hospitalized, duration of oliguria, type and duration of dialysis, serum creatinine, direct Coombs, Thomsen-Freidenreich antigen (T-antigen) testing, transfusion requirements, use of plasmapheresis or plasma infusions, antibiotic use, white blood cell (leukocyte) counts, and extrarenal complications. All S pneumoniae infections were confirmed by culture report of blood, cerebrospinal fluid, or pleural fluid. HUS was defined using the Centers for Disease Control and Prevention definition: thrombocytopenia (platelet count <150 000); acute onset of anemia with microangiopathic changes on blood smear; and renal injury evidenced by hematuria, proteinuria, or elevated creatinine level (above 1 mg/dL in a child younger than 14 years or a 50% increase over baseline).6 Coagulation studies were examined at the time of HUS diagnosis to rule out DIC. The presence of normal fibrinogen levels in the presence of diagnostic criteria for HUS was used to rule out DIC.
The following definitions were used:
Glucose intolerance: hyperglycemia requiring insulin therapy
Pericardial effusion: echocardiographic diagnosis
Cardiac dysfunction: need for inotropic support or cardiac arrest
Hypertension: systolic and/or diastolic pressure >95% for age and gender9
Elevated white blood cell count: >15 × 109/L
Anuria: urine production of <50 mL/d
Oliguria: <0.5 mL/kg/h during a 24-hour period
Empyema: culture-positive pleural fluid
Hepatitis: elevated transaminase levels
Pneumonia: radiographic and clinical diagnosis
Meningitis: culture-positive cerebrospinal fluid
Stroke: radiographic diagnosis
D+ HUS: HUS in the presence of culture-proven E coli O157:H7 gastrointestinal infection or bloody diarrhea.
Selected demographic and clinical information was summarized for SP-HUS and D+ HUS groups. Significance between groups was tested using the Mann-Whitney rank sum test for continuous variables and the Fisher exact test for categorical variables. The statistical program STATA 6.0 (STATA Corp, College Station, TX) was used for all analyses.
Twelve patients with documented pneumococcal infection and HUS were identified at 4 regional pediatric centers (Table 1). The mean patient age was 22 months (range: 4–62 months). Six were boys and 6 were girls. Eleven were white and 1 was black. Nine patients required dialysis (mean: 10.5 ± 17.2 days) and had a mean of 13.2 days (±16.3) of oliguria and/or anuria. Four children received hemodialysis, 2 received peritoneal dialysis, 2 received continuous venovenous hemofiltration, and 1 received both peritoneal dialysis and continuous venovenous hemofiltration. D+ HUS patients had a higher mean age (49 ± 28 months; P = .005) and a slight preponderance of boys (59%). No patient had a predisposing risk factor for pneumococcal infections, such as sickle cell disease, splenectomy, immune disorders, or other chronic disease.
Disease severity was greater in SP-HUS compared with D+ HUS patients (Table 2). Dialysis was required in 75% of SP-HUS patients but in only 59% of D+ HUS patients (P = NS). Oligoanuria duration and hospital stays were twice as long with SP-HUS. SP-HUS with patients required more platelet and red cell transfusions and had a much longer duration of thrombocytopenia. All SP-HUS children received packed red blood cell transfusions compared with only 76% of D+ HUS patients. SP-HUS patients were more likely to require 1 or more platelet transfusions than D+ HUS patients (83% vs 47%; P = .05). SP-HUS patients required a greater number (per patient) of platelet transfusions (3.6 vs 0.5; P = .005) and packed red blood cell transfusions (8.4 vs 2.0; P < .001) compared with children with D+ HUS. All patients received washed red blood cells, although 2 patients received unwashed cells as well (patients 8 and 10). Thrombocytopenia persisted for longer in SP-HUS (mean: 11.6 days) than in D+ HUS patients (6.8 days; P = .004). SP-HUS patients had a slightly lower leukocyte count at admission than D+ HUS patients (12.9 vs 17.0 × 109/L; P = NS) but a higher maximum leukocyte count (34.4 vs 20.4 × 109/L; P = .002). The mean hospital stay was longer in SP-HUS than in D+ HUS patients (33.2 vs 16.1 day; P = .004).
The primary site of S pneumoniae infection was pneumonia in 9 patients, meningitis in 2 patients, and both in 1 patient. Empyema was present in 8 of 10 patients with pneumonia. One child developed a bronchopulmonary fistula requiring a pulmonary lobectomy. Hypertension was present in 6 patients, and a pericardial effusion was detected in 3 patients (purulent in 1). Gastrointestinal complications were seen in 7 patients: 5 with pancreatitis, 1 with glucose intolerance, 1 with cholestasis, and 1 with hepatitis (also with pancreatitis). Two patients had seizures and strokes, only 1 of whom had meningitis. Cardiac dysfunction was seen in 3 patients. Mechanical ventilation was required in 4 children with pneumonia. The incidence of hypertension and gastrointestinal complications was similar. Seizure or stroke was seen in 16% of SP-HUS but in none of the D+ HUS group (Fig 1). Pericardial effusion was seen only in SP-HUS, and pleural fluid collection was seen predominantly in the SP-HUS pneumonia group. There was a marked seasonal variation between the 2 groups: 75% of SP-HUS cases occurred between November and March, whereas 82% of D+ HUS was seen between May and October10 (Fig 2).
Antibiotic use before HUS diagnosis occurred in 7 SP-HUS patients; 6 received a third-generation cephalosporin, and 1 received azithromycin. There was no discernible effect of previous antibiotic use on disease severity or duration. Patient 7 had S pneumoniae that was penicillin resistant and patient 11 had intermediate sensitivity to penicillin, but both were sensitive to third-generation cephalosporins.
All patients had coagulation studies consistent with HUS but not DIC. Fibrinogen and D-dimer (or fibrin split products) studies were normal in all patients at diagnosis. Three patients had mildly elevated prothrombin and partial thromboplastin times at diagnosis. Another 2 patients had mild increases in partial thromboplastin times alone. Six patients received fresh-frozen plasma (FFP) during their hospital stay. One patient received 2 plasmapheresis treatments that were halted after a bloody pleuracentesis. There was no discernible effect on platelet counts after plasma infusions, even when platelet counts were looked at daily for 4 days after plasma infusion. Three patients had weakly positive direct Coombs test. One of these had a positive test for the presence of T-antigen. T-antigen testing was not performed in other patients.
At hospital discharge, 9 (75%) patients had normal serum creatinine levels. There were no deaths. Seven SP-HUS patients have returned for follow-up; 5 have a normal glomerular filtration rate (by plasma iothalamate clearance in 4 and urine creatinine clearance in 1), blood pressure, and urinalysis. Quantitative urine protein testing was not done. Two (17%) patients went on to develop end-stage renal disease (ESRD), both of whom have subsequently received renal transplants 1.8 and 3 years after HUS, respectively. Neither transplant patient has had recurrent HUS, and both had functioning transplants at last follow-up. The third patient with an elevated creatinine at discharge has been lost to follow-up. Two patients experienced strokes, 1 of which had persistent seizure disorders and residual neurologic damage at last follow-up. The second child with a stroke was lost to follow-up.
HUS complicating invasive pneumococcal disease is an uncommon but severe disorder. We found only 46 previous pediatric cases reported in the literature (Table 3). The current study is the largest series of SP-HUS reported to date and highlights the severity of possible sequelae seen with this common organism. The pathophysiology of SP-HUS remains poorly understood, but the production of neuraminidase by S pneumoniae is thought to play a pivotal role. Neuraminidase cleaves n-acetyl neuraminic acid from the surface of red blood cells and endothelial cells, uncovering the T-antigen.11 Exposure of the T-antigen is postulated to result in endothelial damage leading to thrombotic microangiopathy, the pathologic hallmark of HUS. The majority of individuals possess naturally occurring immunoglobulin M (IgM) antibodies to the T-antigen, and it has been suggested that anti-T IgM may mediate hemolysis or endothelial cell damage leading to HUS. T-activation of human red blood cells has been documented to occur with a wide variety of organisms,12 and T-antigen has been demonstrated in glomerular capillaries and tubular epithelial cells in SP-HUS patients.13 Although it remains unproved whether binding of anti-T IgM mediates hemolysis or endothelial cell damage in SP-HUS,12 we recommend that all red blood cells given to these patients be washed to remove potential anti-T IgM antibodies. We did not observe any change in platelet count after FFP, but because plasma may contain IgM antibodies, FFP should also be avoided, when possible.
Pneumococcal disease remains an important public health problem in children. In a population-based study in the United States, Zangwill et al14 found that invasive pneumococcal infections occur with an annual incidence of 72/100 000 persons/y in children younger than 5 years (145/100 000/y for children younger than 2 years). Children younger than 2 years had a 4.7-fold increased risk of pneumonia compared with nonelderly adults and a 13.1-fold increased risk of meningitis compared with adults older than 65 years. In a large single-center report, HUS was associated with 0.6% of all reported invasive pneumococcal infections in the metropolitan Atlanta area between 1994 and 1996.15 Despite the relative ubiquity of invasive pneumococcal disease, HUS is a rare complication; <50 cases having been reported in the literature previously. In an early report of 3 children with SP-HUS, Novak noted free neuraminidase in their blood culture filtrate but could not find neuraminidase in isolates of the same strain of non-HUS patients.29 This suggests a unique ability to produce or release neuraminidase in the serotypes that cause HUS or perhaps a unique patient-pathogen interaction in SP-HUS patients. However, another cohort study of SP-HUS patients failed to find a relationship between neuraminidase levels and HUS.16
In the Atlanta report of 7 patients, 60% of patients with pneumonia-associated HUS had an empyema. We also found a high incidence of empyema in our patients (67%). The incidence of empyema complicating pneumococcal pneumonia in children younger than 18 years has been estimated at 3.3 cases/100 000 children/y (0.003%).17 Our review of the literature found that 51% of SP-HUS cases associated with pneumonia had concurrent empyemas present (Table 3). The high incidence of empyema in this and the Atlanta study suggests that a heavy bacterial load and suppurating disease may increase the risk of HUS, at least in the setting of a pulmonary infection.
HUS is less common after pneumococcal meningitis than pulmonary disease; of the 46 previously reported SP-HUS cases, 65% were associated with pneumonia (5% also had meningitis), 28% were associated with meningitis alone, and 7% were associated with pneumococcal bacteremia alone (Table 3). However, in a recent report from France, 8 of 11 patients with SP-HUS presented with meningitis and 88% of patients with meningitis died (4) or developed end-stage renal failure (3), whereas all of those with pneumonia alone survived.18 Neuraminidase has been shown to pass the blood-brain barrier in pneumococcal meningitis-associated HUS, suggesting a mechanism for isolated nervous system infection to lead to systemic vasculopathy.19 The lower incidence of meningitis compared with pneumonia in our patients is presumably a reflection of the lower incidence of meningitis in children compared with pulmonary disease.14 Why France sees a higher incidence of meningitis-related SP-HUS and a poorer outcome is unclear. In contrast, pneumonia or meningitis is almost never seen in D+ HUS.20
Most of the patients in this report presented to their respective institutions in discrete time periods. In Washington state, 4 of 5 children presented in a 23-month period. In Texas, 3 of 4 presented in a 17-month period, and in New Mexico, 2 presented in an 18-month period. Because our study design is in no way population based or even institution-based and because we do not have data on bacterial serotypes, we could not examine whether disease was serotype specific; however, these clusters do raise the question of whether serotype influences HUS development. Likewise, other factors that may have an impact on the severity of pneumococcal disease, such as ethnic or racial status, variability in clinical diagnostic and therapeutic decisions, or geographic variations, could not be assessed adequately in a retrospective study. A seasonal influence was seen: SP-HUS was more common in the winter months, and D+ HUS was more common in summer.
An elevated white blood cell count at presentation has been shown to be a significant risk factor for developing HUS in E coli O157 infections.21 In our SP-HUS patients, the leukocyte count was elevated in only 4 children at admission, although all developed a high leukocyte count during their disease process. The maximum leukocyte count was higher in SP-HUS than in D+ HUS patients, most likely representing the severity of the pneumococcal infection. However, we did not find a correlation between admission or maximum leukocyte count and disease severity in SP-HUS such as has been seen for D+ HUS.
Many patients received antibiotic therapy before the diagnosis of HUS, but there was no obvious difference in the clinical course or outcome of patients who received antibiotics before HUS compared with those who did not. There is evidence that antibiotic use in children with E coli O157 gastroenteritis may increase the risk of developing HUS.22 Because our report is a case study, we could not assess the risk of HUS with antibiotic use. Unlike E coli O157, which causes self-limited and localized gastrointestinal infection, invasive pneumococcal disease has grave risks. Hence, antibiotics should never be withheld when clinically indicated because of any theoretical risk of developing or worsening HUS.
Historically, morbidity and mortality in SP-HUS is high (Table 3). We had no deaths in our patients. With the exception of 2 deaths each the reports of Cabrera et al15 and Nathanson et al18, most cases of death occurred in the 1970 to 1980s. As with D+ HUS, mortality is probably lower in recent years owing largely to advances in the care of critically ill children rather than any real change in disease virulence. We did find that children with SP-HUS had more severe renal and hematologic disease compared with typical HUS patients (Table 2). Two (17%) SP-HUS patients developed ESRD, whereas the reported incidence of ESRD in D+ HUS is only approximately 5%2 and we saw none in our D+ HUS group. We also saw a higher risk of neurologic sequelae in our SP-HUS patients (16%) compared with the D+ HUS comparison group (Fig 1). However, only 1 SP-HUS patient had neurologic disease in the absence of meningeal infection, and that patient had a cardiac arrest that might explain the neurologic sequelae. Historically, D+ HUS has an incidence of neurologic sequelae of 20%,20 and our comparison group may have consisted of an unusually mild cohort of D+ HUS patients. Hence, it is not clear that the SP-HUS itself results in any higher incidence of neurologic sequelae, independent of the effects of the pneumococcal disease.
We compared our patients with D+ HUS patients who appear endemically in Washington state. D+ HUS is relatively uncommon in the 3 other regional centers in which our patients presented, making center-specific comparisons difficult.23 Historically, patients seen during large epidemics of E coli O157 infections or seen before the development of sophisticated pediatric dialysis and intensive care expertise have had higher incidences of morbidity and mortality. In D+ HUS, 70% of patients develop oliguric or anuric renal failure, 50% require dialysis, 75% require blood transfusions, and 50% require platelet transfusions.24,25 Twenty percent of D+ HUS patients develop pancreatitis, and 40% develop hepatomegaly or elevated transaminases. Although pleural effusions are seen in 20% of D+ HUS patients, empyema is very rare. Central nervous system complications occur in 20% of D+ HUS patients.20 If we had compared our patients with historical HUS reports, there would be greater risk of bias relating to patient selection and therapy. Our control group did seem to have milder disease than seen in many published series of D+ HUS. Although this is a potential source of bias, it also suggests that SP-HUS may not always be more severe than D+ HUS. A recent report comparing atypical and D+ HUS cases in the western United States found no difference in clinical severity or outcome.26 Although none of the atypical HUS patients in that series had SP-HUS, the report also questions whether atypical HUS is uniformly more severe than D+ HUS. Atypical HUS is composed of a variety of distinct conditions that share a common pathologic outcome, thrombotic microangiopathy, but that probably also exhibit a wide spectrum of disease severity and prognostic outcomes.
HUS remains a rare but severe complication of invasive pneumococcal disease. The presence of parapneumonic empyema seems to be a particular risk factor for HUS. Children with pneumococcal disease and severe hematologic or renal abnormalities should be investigated for evidence of HUS.
- Received June 4, 2001.
- Accepted February 19, 2002.
- Reprint requests to (J.B.) Children’s Hospital of New Mexico, University of New Mexico School of Medicine, Department of Pediatrics, ACC-3, 2211 Lomas Ave, NE, Albuquerque, NM 87131-5311. E-mail:
- ↵Centers for Disease Control and Prevention. Case definitions for infectious conditions under national public health surveillance. MMWR Morb Mortal Wkly Rep.1997;46(RR-10) :17
- ↵Vaziri ND, Chang D, Malekpour A, Radaht S. Pancreatic enzymes in patients with end-stage renal disease maintained on hemodialysis. Am J Gastroenterol.1988;84 :410– 412
- ↵Report of the Second Task Force on Blood Pressure Control in Children–1987. Task Force on Blood Pressure Control in Children. National Heart, Lung, and Blood Institute, Bethesda, Maryland. Pediatrics.1987;79 :1– 25
- ↵Tarr PI, Hickman RO. Hemolytic uremic syndrome epidemiology: a population-based study in King County, Washington, 1971 to 1980. Pediatrics.1987;80 :41– 45
- ↵Zangwill KM, Vadheim CM, Vannier AM, Hemenway LS, Greenberg DP, Ward JI. Epidemiology of invasive pneumococcal disease in southern California: implications for the design and conduct of a pneumococcal conjugate vaccine efficacy trial. J Infect Dis.1996;174 :752– 759
- ↵Cabrera GR, Fortenberry JD, Warshaw BL, Chambliss CR, Butler JC, Cooperstone BG. Hemolytic uremic syndrome associated with invasive Streptococcus pneumoniaeinfection. Pediatrics.1998;101(4 Pt 1) :699– 703
- ↵Butler JC, Cabrera GR, Fortenberry JD, Paton JC, Elliot JA, Facklam RR. Hemolytic uremic syndrome in children with pneumococcal bacteria. In: First International Symposium on Pneumococcal Disease; Helsingor, Denmark; June 13–17, 1998
- ↵Bell BP, Griffin PM, Lozano P, Christie DL, Kobayashi JM, Tarr PI. Predictors of hemolytic uremic syndrome in children during a large outbreak of Escherichia coli O157:H7 infections. Pediatrics.1997;100(1) . Available at:http://www.pediatrics.org/cgi/content/full/100/1/e12
- ↵Siegler RL, Pavia AT, Christofferson RD, Milligan MK. A 20-year population-based study of postdiarrheal hemolytic uremic syndrome in Utah. Pediatrics.1994;94 :35– 40
- Khodasevich LS, Val’kov AI. [Pathomorphology of pneumococcal infection in children]. Arkh Patol.1998;60 :37– 40
- Copyright © 2002 by the American Academy of Pediatrics