Effect of an Oral Shiga Toxin–Binding Agent on Diarrhea-Associated Hemolytic Uremic Syndrome in Children: Title and subTitle BreakA Randomized Controlled Trial FREE

Diarrhea-associated hemolytic uremic syndrome (HUS) is the most common cause of acute renal failure in previously healthy children in the United States.¹ Most cases are caused by an antecedent enteral infection with Shiga toxin–producing strains of Escherichia coli (STEC).^{1 - 2} Nearly 40% of patients require temporary dialysis, up to 20% develop serious extrarenal events, and the mortality rate is 3% to 5%.^{2 - 3}

Many forms of therapy have been tried in children with diarrhea-associated HUS, including plasma infusion, plasmapheresis, intravenous IgG, fibrinolytic agents, antiplatelet drugs, corticosteroids, and antioxidants. Most of them have been shown to be ineffective in controlled clinical trials. Therefore, current management relies on intensive supportive care.^{1 - 2 ,4}

Shiga toxins 1 (Stx1) and 2 (Stx2) elaborated by STEC are absorbed from the gastrointestinal tract and bind to a glycolipid receptor, globotriaosylceramide, on the endothelial cell membrane, leading to diffuse vascular injury and organ failure.⁵ Prevention of Shiga toxins binding to globotriaosylceramide is a potential means of diminishing Shiga toxin–mediated injury. SYNSORB Pk (SYNSORB Biotech Inc, Calgary, Alberta) is a novel agent composed of silicon dioxide particles covalently linked to the trisaccharide moiety of the globotriaosylceramide molecule that mediates Shiga toxin binding.⁶ Thus, this agent should compete with endothelial and epithelial receptor sites for binding of Shiga toxin. It was safely tolerated by healthy adult volunteers in a phase 1 study without any evidence of toxicity.⁷

Clinical observations suggest that there is a correlation between the severity of the antecedent enteritis and that of diarrhea-associated HUS.^{8 - 9} Moreover, many of the complications occur late in the disease.^{1 - 3} These findings raise the possibility that diarrhea-associated HUS is not a one-hit phenomenon and that interference with absorption of Shiga toxins during diarrhea-associated HUS could ameliorate the disease. This multicenter, randomized, double-blind, placebo-controlled clinical trial was conducted to determine whether oral administration of a Shiga toxin–binding agent diminishes the severity and improves the clinical course of diarrhea-associated HUS in pediatric patients.

The study was approved by the institutional review boards at the 31 participating centers in the trial, with patients enrolling at 26 sites in the United States and Canada. Informed consent for the trial and the use of stored samples for future studies was obtained from the parent or guardian before enrollment. Assent was also obtained from the patient whenever possible (ie, if the child was ≥9 years).

Children between the ages of 6 months and 18 years with diarrhea-associated HUS were eligible for inclusion. The diagnosis of diarrhea-associated HUS was made if the patient satisfied all of the following 4 criteria: (1) platelet count of less than 140 × 10³/µL; (2) fragmentation of erythrocytes on a peripheral smear (>5 per high-powered field); (3) renal injury as indicated by the presence of hematuria (>1+ by dipstick analysis) and/or proteinuria (>1+ by dipstick analysis) and/or azotemia (serum creatinine concentration, >95th percentile for age and sex); and (4) a diarrhea illness within 7 days before the identification of HUS. Evidence of microangiopathy and hemolysis rather than a specific hematocrit cutoff as an index for anemia was chosen as a diagnostic criterion to avoid the confounding effects of dehydration and parenteral fluid therapy administered before enrollment. Microbiological confirmation of STEC infection was not required before entry to expedite initiation of the study medication. Information about antibiotic use during the prodromal illness was not routinely available.

Exclusion criteria included an atypical or nondiarrhea prodrome; family history of hereditary HUS; HUS associated with bone marrow transplantation, pneumococcal infection, or human immunodeficiency virus infection; preexisting renal disease; and preexisting structural or motility disorder of the gastrointestinal tract.

Patients with diarrhea-associated HUS were randomly assigned to receive either the binding agent or cornmeal placebo (Figure 1). Although cornmeal had a slightly different texture than the binding agent, it was chosen to be the placebo rather than the Chromosorb platform molecule because it has no binding affinity for Shiga toxin. Both agents were provided by SYNSORB Biotech Inc. The study medications were packaged and labeled with a patient identification number by the manufacturer, according to the randomization schedules that were prepared by the data coordinating center for each site. Individual doses of medication were prepared on a daily basis by the pharmacist by mixing the agent with 1 to 2 oz of applesauce or banana-flavored baby food. The identity of the specific medication was blinded to the patient and physician. The study medication (500 mg/kg per day) was administered orally or via a nasogastric tube every 8 hours for 7 days. If children were discharged before 7 days, the study medication was provided to the parents with instructions to dispense it at home for the complete 7-day course.

The decision to start dialysis was made based on the occurrence of 72 consecutive hours of oligoanuria, defined as a urine flow of less than 0.5 mL/kg per hour, after hospitalization at the participating center. This is a better index of the severity of renal injury than the degree of azotemia.¹⁰ Preference was given to peritoneal dialysis, because this therapeutic modality may enhance the clearance of plasminogen activator inhibitor 1 and facilitate recovery of renal function.¹¹ However, hemodialysis and hemofiltration were acceptable alternatives if there were clinical contraindications to the use of peritoneal dialysis. If there were acute indications for dialysis, such as life-threatening hyperkalemia, this procedure could be implemented earlier than mandated by the protocol after consultation with the principal investigator. All other aspects of medical care were left to the discretion of the attending pediatric nephrologists.

Investigators were asked to record daily the presence of serious extrarenal events. These included seizure, coma, thrombotic stroke, hemorrhagic stroke, cortical blindness, bowel gangrene, bowel perforation, intussusception, pancreatitis, diabetes mellitus, gallbladder complications, liver failure, arrhythmias, congestive heart failure, and respiratory failure.³

Clinical and laboratory data were collected daily during the period of hospitalization. In addition, patients were evaluated at 7, 14, 28, and 60 days after discharge. Plasma samples were collected on days 1, 4, 7, and 10 of hospitalization and day 28 after discharge. Urine and stool samples were collected daily during the first 7 days of hospitalization, on day 10 of hospitalization, and on days 7, 14, 28, and 60 after discharge. The initial plasma sample was obtained before administration of study medication in all cases. However, this applied to the first urine and stool specimens only if they were available before initiation of therapy. The presence of proteinuria or hypertension and the glomerular filtration rate (GFR), estimated by the creatinine clearance after water loading in children older than 3 years, was assessed at the completion of the 60-day follow-up period.¹²

Each stool sample was tested for the presence of free Stx1 or Stx2, STEC, and STX1 and STX2 genes. Free Shiga toxin and the presence of STEC were determined using Premier EHEC (Meridian Biosciences, Cincinnati, Ohio). This involved direct tests on stool samples for Shiga toxin and overnight broth cultures for the detection of STEC. If broth cultures tested positive for STEC, individual STEC isolates were obtained by using a colony immunoblot as described previously.¹³ The presence of STX1 or STX2 genes in stool specimens was determined by polymerase chain reaction (PCR) on 1.5% agarose gels. This required a series of 4 10-fold dilutions of the stool sample in water and then testing 40 µL of each dilution by PCR by using the following primers: STX1 F-ATA AAT CGC CAT TCG TTG ACT AC; STX1 R-AGA ACG CCC ACT GAG ATC ATC; STX2 F-GGC ACT GTC TGA AAC TGC TCC; STX2 R-TCG CCA GTT ATC TGA CAT TCT G. Because of the importance of defining the extent of antecedent STEC infection in interpreting the efficacy of oral binding agent therapy in children with diarrhea-associated HUS, the results of the stool cultures for STEC and the assays for free Shiga toxin are provided.

Randomization was performed in a ratio of 2 oral binding agents and 1 placebo in blocks of 3 patients at each center for the purpose of maximizing safety information on the agent while maintaining a blinded randomized design. The primary end points were to determine whether treatment with the oral binding agent reduces the combined frequency of death and serious extrarenal events and lowers the need for dialysis. Based on a literature review, the event rate for the first primary outcome variable was predicted to be 20% and 50% for the second outcome variable in the placebo group.^{3 ,9} For the experimental therapy to be considered clinically beneficial, it was judged necessary to lower the event rate to 5% for the first primary end point and 25% for the second primary end point. A sample size calculation indicated that 174 patients (58 placebo and 116 binding agent) were needed to demonstrate a difference in these outcome variables with a power of 80% and an overall α of .05 for each of the 2 outcomes. Included in these calculations were 2 formal interim analyses, the first after one third and the second after two thirds of the outcome data were available. The calculations used O'Brien-Fleming stopping rules for rejecting either the null hypothesis or the alternative hypothesis and Lan and DeMets use function at the actual time of analysis. The interim analyses and stopping rules were designed using EaSt software (Cytel Software Inc, Cambridge, Mass).

Secondary outcome variables, selected on the basis of clinical relevance, included differences between the groups in transfusions of red blood cells or platelets, time to normalization of platelet count, and the number of patients with hypertension, proteinuria, or a GFR of less than 90 mL/min per 1.73 m² at the 60-day follow-up visit.

The study was monitored by the data and safety monitoring board, which convened to review the interim analyses and to determine whether any departures from protocol or unanticipated adverse events were occurring.

Comparison of the occurrence of each of the 2 primary outcomes between the treatment groups was performed by using a 2-tailed Fisher exact test. The Fisher exact test was also used to compare the sex distribution, concomitant medications by class, secondary adverse events (eg, nausea and vomiting), and the presence of hypertension and a GFR of less than 90 mL/min per 1.73 m² at the end of the follow-up period. An exact χ² test was used to compare racial distribution and route of entry into the study (emergency department, direct admission, or transfer from another hospital), as well as to explore the relationship between site and sex for the larger sites (>5 patients enrolled). A t test was used to compare the ages, weights, vital signs, and various hematology and biochemistry markers (some after transformation to achieve normality). A 1-way analysis of variance was used to compare age, platelet counts, white blood cell count, and serum creatinine concentration (after transformation) among the larger sites (>5 patients enrolled). Kaplan-Meier method curves and the log-rank test were used to compare duration of hospital stay and duration of dialysis. For dialysis, duration was set to zero for those patients who did not receive dialysis, and if dialysis was started and stopped more than once, the total duration from initiation to final stop was used. Select subgroup analyses were performed for the primary outcome variables; these results are considered exploratory. Statistical analyses were performed by using SAS version 8.0 (SAS Institute, Cary, NC).

A total of 150 patients with diarrhea-associated HUS were enrolled in the study between July 27, 1997, and April 14, 2001. The trial was stopped after 45 months by the data and safety monitoring board following the second interim analysis because of lack of efficacy on both primary outcome variables. During this period, 44 eligible patients were not enrolled (37 children declined and 7 were not offered participation) (Figure 1). Five children who were randomized never received the study medication or received only a single dose before withdrawing from the study. Therefore, this report is based on 145 patients in a modified intention-to-treat analysis. None of these patients withdrew or were lost to follow-up. The number of patients enrolled varied between 1 and 13 per site; 13 centers (50%) enrolled no more than 5 patients each. Demographic variables (age and sex), as well as platelet and white blood cell counts at entry, were comparable at all sites enrolling more than 5 patients. The mean serum creatinine concentration was lower at 1 site (n = 10) than at the other large enrolling sites (P = .046). However, if correction is made for multiple outcomes, this finding is not significant. A total of 119 (82%) of 145 patients were randomized within 24 hours of admission to the hospital.

Of the children with diarrhea-associated HUS, 57% were girls and 85% were white (Table 1). The median age of the patients was 4.2 years. Most patients (59%) were transferred from other hospitals to the participating center. Escherichia coli O157:H7 was detected in the stool samples of only 28 children (19%) at the time of study entry based on specimens obtained before enrollment or on the first day of the protocol. There were no differences between the treatment groups in any of these variables.

The clinical severity of diarrhea-associated HUS in the treatment groups was comparable at study entry based on similar values for height, weight, blood pressure, and hematological and biochemical tests (Table 2). Serum uric acid, amylase, and lactate dehydrogenase levels were also comparable among the 38, 44, and 57 children, respectively, in whom these measurements were obtained. The time course of changes in platelet count, white blood cell count, and serum creatinine concentration during hospitalization was similar in binding agent– and placebo-treated patients (data available on request).

During the 7-day treatment period, nearly 75% of all doses were successfully administered orally or via a nasogastric tube and were tolerated, defined as the absence of vomiting for 30 minutes after dispensation of the study medication. The success rate was comparable for the binding agent and placebo. The percentage of tolerated doses was slightly higher during the first 3 days of treatment than during the last 4 days. Of the nearly 25% of the per-protocol doses that were administered unsuccessfully, administration of study medication was contraindicated in 47% because of concurrent medical problems, such as an ileus or seizures. Overall, 75% of the doses were ingested, 12% were contraindicated, and only 13% were considered failures. Furthermore, the rate of successful administration of the oral binding agent or placebo was comparable in children who did and did not experience a serious extrarenal event. The most common adverse effects were nausea and vomiting, which necessitated discontinuation of drug use in approximately 5% of patients. There were no serious adverse events that were considered definitely related to the administration of the binding agent. The use of concomitant medications was similar in the 2 groups.

The prevalence of death or serious extrarenal events was similar in the 2 groups: 18% and 20% in binding agent and placebo-treated groups, respectively (P = .82) (Table 3). There were 3 deaths in the binding agent–treated group and 1 in the placebo-treated group. Similarly, the need for dialysis was 42% in the binding agent group and 39% in the placebo group (P = .86). The mean duration of dialysis was 5.2 vs 3.6 days in the binding agent and placebo-treated groups, respectively (P = .59). Exclusion of patients who experienced a serious extrarenal event on the first day of treatment and who were unlikely to have benefited from treatment did not change the findings (Table 4). Analysis of patients who were treated per protocol (n = 87), defined as receipt of at least 7 doses of study medication, yielded similar results. The results were similar in both sexes (Table 4). They were also comparable in patients who were admitted directly or via the emergency department compared with those who were transferred from another hospital, in whom the study treatment may have been initiated later in the disease. Moreover, clinical outcomes in response to study treatment were comparable in patients who had a positive stool culture for STEC at admission or demonstrable free Shiga toxin in their gastrointestinal tract vs children in whom STEC or Shiga toxin could not be detected in stool specimens collected after enrollment in the trial. Although the disease tended to be more severe during the winter (November-April) than summer (May-October) months, as indicated by the frequency of extrarenal events of 31% vs 15%, respectively (P = .07), the efficacy of the binding agent was unaffected by the time of the year when the patient was enrolled (Table 4). The overall rate of serious extrarenal events tended to be somewhat lower in centers with more than 5 patients enrolled in the study compared with centers with 5 or fewer patients (P = .10). There was also a numerical difference in the rate of serious extrarenal events between the binding agent–treated (13%) and placebo-treated groups (23%, P = .18) at sites that enrolled more than 5 children in the study. However, a follow-up power calculation indicated that the sample size needed to demonstrate this center effect, if true, would have been 548 children (ie, 365 randomized to receive the binding agent and 183 assigned to placebo). Finally, the study medication was similarly ineffective in preventing the more serious extrarenal events, defined as central nervous system or cardiovascular complications, compared with all the other extrarenal complications (P = .49).

None of the secondary outcome variables differed between the 2 groups. In particular, an equal proportion of children in the treatment groups, 76 (79%) of 96 in the binding agent–treated group vs 39 (80%) of 49 in the placebo-treated group, required transfusions of packed red blood cells or platelets. The length of time to normalization of platelet count (Figure 2) and the median duration of hospital stay (8 days) were similar in the 2 groups.

At completion of the 60-day follow-up period, most patients had recovered completely. Among the 88 patients in whom the information was available, 28 (32%) had hypertension. Among the 121 patients who had a urine test at their last follow-up visit, 10 (8%) had proteinuria. Kidney function was measured in only 46 children at the 60-day follow-up visit, and 7 (15%) had a GFR of less than 90 mL/min per 1.73 m². Presence of hypertension, proteinuria, or a reduced GFR did not differ between treatment groups (Table 5) and was not related to the occurrence of a serious extrarenal event or the need for dialysis during the acute episode.

Free Shiga toxin or STEC was demonstrated in the stool samples of 52 (35.9%) of 145 children, based on stool cultures obtained before or after enrollment, or free Shiga toxin in stool specimens obtained after enrollment. Samples containing only STX genes by PCR analysis were not considered positive. This frequency was similar in the 2 treatment groups. A more complete description of the microbiology findings, including serotyping of STEC, is under way.

The picture of diarrhea-associated HUS in the United States that emerges from this study is an accurate and current reflection of the disease, because this multicenter trial was large in scope, involving centers along the East Coast, centers in the Midwest, and a single Canadian site. This contrasts with most previous studies that describe results from a single center or region. Thus, the outcome of the trial is probably generalizable to all children with diarrhea-associated HUS.

Our case definition of diarrhea-associated HUS required evidence of acute renal injury, thrombocytopenia, and erythrocyte fragmentation. Cases of partial diarrhea-associated HUS, in which the patient satisfied 2 or fewer diagnostic criteria, were not included. Several cases had only minimal laboratory abnormalities at the time of enrollment, which reflected the intense effort to enter patients as rapidly as possible after the diagnosis. However, this does not imply that all of these cases were mild, because a shorter prodrome and less marked biochemical and hematological disturbances at the time of diagnosis may be indicative of a higher risk of developing severe diarrhea-associated HUS.¹⁴

The cases in this study are considered to have diarrhea-associated HUS, and no claim is made about the STEC origin of disease. Microbiological confirmation of STEC infection is not a routine criterion for inclusion of children in clinical trials of diarrhea-associated HUS. Because stool cultures are often negative after the diagnosis of diarrhea-associated HUS,¹⁵ children without documented STEC infection have not been excluded retrospectively from an analysis of efficacy. In this trial, 36% of the children had evidence of STEC infection based on isolation of the bacteria from stool samples before or after enrollment or detection of free Shiga toxins in stool specimens obtained after entry into the study. This rate is comparable with the results of the US National HUS Study.¹⁶ Addition of patients with positive PCR results for STX genes in stool specimens and incorporation of serological testing for antibodies to specific E coli serotypes are likely to have increased the rate of proven STEC infection.^{16 - 17} Nonetheless, the strict confirmation of a diarrheal prodrome in all cases makes STEC infection highly probable as the cause of diarrhea-associated HUS in this clinical context and justifies inclusion of all cases in this analysis.

The results of this randomized clinical trial, the largest controlled study of a potential treatment of diarrhea-associated HUS to our knowledge, demonstrate that oral therapy with a Shiga toxin–binding agent is ineffective at reducing the severity of diarrhea-associated HUS in pediatric patients. The dose of the binding agent that was administered was based on in vitro studies and in vivo evaluation of the amount of drug needed to completely neutralize Shiga toxin within the gastrointestinal lumen in experimental animals.^{6 - 7 ,18} Consideration of this fact along with the high degree of compliance with the clinical protocol, as evidenced by the percentage of successfully administered doses of both the study medication and placebo, suggest that insufficient treatment is unlikely to account for the failure of the binding agent to diminish the severity of diarrhea-associated HUS.

There are several potential explanations for the lack of efficacy of the binding agent in this trial. Therapy may have been started too late in the disease process, as suggested by the fact that nearly 60% of the patients were transferred from other hospitals to the trial sites and the finding that only 23% of patients had viable STEC or free Shiga toxin in their stool samples at the time of initiation of treatment with the binding agent. This finding is consistent with a recent study about the low yield of free Shiga toxins in stool samples at the time of diagnosis of diarrhea-associated HUS in children.¹⁹ Impaired gastrointestinal motility in patients with diarrhea-associated HUS may have limited delivery of the drug to the distal intestine, where the bulk of Shiga toxin is likely to be found, despite the remarkable tolerance of the medication in the study population. Finally, interaction between STEC and the gastrointestinal epithelium may hinder the capacity of the drug to bind Shiga toxin in vivo. However, in view of the abrupt onset of diarrhea-associated HUS, its sporadic and rare occurrence throughout the United States, and the need to transfer affected patients to tertiary centers that have the clinical expertise to manage potential complications, it is likely that our experience is typical and reflects the obstacles that confront any putative new therapy for diarrhea-associated HUS.

There are several important limitations in this study, including inability to confirm prior infection with STEC in nearly two thirds of the patients, lack of information about prior use of antibiotics or antimotility agents, lack of data about the incidence of diarrhea-associated HUS or referral patterns at each participating site, and the small number of cases with a GFR measurement at the 60-day follow-up. However, the excellent rate of enrollment of eligible patients and the degree of compliance with the protocol indicate that clinical trials can be conducted in children with life-threatening orphan diseases such as diarrhea-associated HUS.

Other biological products with enhanced avidity for Shiga toxin and increased solubility, including custom-designed modified forms of the binding agent and genetically modified bacteria, are under development.^{20 - 22} The outcome of this trial casts doubt on the likelihood of success of strategies to prevent continued gastrointestinal absorption of Shiga toxin. Recent data indicate that Shiga toxin is present in the circulation bound to polymorphonuclear leukocytes for up to 5 days after the diagnosis of diarrhea-associated HUS.²³ It may, therefore, be advisable to evaluate novel therapeutic agents, such as monoclonal antibodies, that inhibit the action of circulating Shiga toxins.^{24 - 26} Until an effective therapy is discovered for newly diagnosed diarrhea-associated HUS, efforts should be directed at disease prevention, including public health maneuvers (eg, improved food testing for STEC contamination, replacement of grain feed with hay feed before slaughter, food irradiation) and heightened consumer awareness (eg, food preparation techniques, beef cooking temperature).^{27 - 28}