IVMP+IVIG raises platelet counts faster than IVIG alone: results of a randomized, blinded trial in childhood ITP

Childhood immune thrombocytopenia (ITP) is a common and usually mild, transient disorder caused by autoantibodies against platelet membrane antigens, resulting in enhanced Fc-mediated destruction of platelets by macrophages in the reticuloendothelial system (RES) and in some cases impaired production of platelets by megakaryocytes in the bone marrow.^1,2  Most children with ITP present with low platelet counts (PCs) but with minimal bleeding.³ 

The incidence of severe life-threatening bleeds such as intracranial hemorrhage (ICH) in childhood ITP is fortunately very low. Retrospective reviews/registries have estimated the incidence of ICH in childhood ITP to be ∼0.4% to 0.6%.^4-9  Although rare, the occurrence of ICH in childhood ITP is devastating, with significant mortality (studies suggest 25% to 55% overall mortality) and morbidity.^4,10 

The only consistent risk factors for ICH include a PC of <10 × 10⁹/L and head trauma. One study also showed that gross hematuria and non-skin bleeding symptoms are much more common in patients with ITP who experience ICH vs those who do not.⁵  Rare risk factors for ICH include concomitant coagulopathies and varicella-associated ITP.

Given the rarity of ICH, controversy exists as to whether children with ITP without concomitant bleeding need any treatment other than observation. However in the context of certain situations (eg, ICH, other life-threatening bleeds, or need for urgent surgery) it is imperative that the PC be increased as quickly as possible. IV immune globulin (IVIG), corticosteroids, and IV anti-D have all been shown to increase a patient’s PC in ITP.

The main effect of corticosteroids in increasing PCs in ITP is thought to be due to suppression of platelet phagocytosis by the RES.¹¹  Corticosteroids may also improve endothelial integrity, further reducing the risk of bleeding.¹²  Several studies in adults and in children have shown that the higher the corticosteroid dose (oral/parenteral), the faster the PC increment.^13-17 

In 1981, Imbach and colleagues noted that IVIG infusions increase PCs in children with ITP.¹⁸  This was confirmed by several randomized and nonrandomized clinical trials.^12,14,15  The major proposed mechanism of action of IVIG is the inhibition of Fc-mediated phagocytosis of antibody-coated platelets by the RES.^19-21 

Both corticosteroids and IVIG have numerous adverse effects. In the case of corticosteroids, these include increased appetite, personality changes, Cushing syndrome, weight gain, hyperglycemia, and hypertension. These side effects are related to the duration of corticosteroid exposure; thus, a very short course of high-dose oral/IV corticosteroids results in fewer side effects than several weeks of low-dose oral corticosteroids.^16,17 

The main concern with IVIG is that it is a blood product and consequently carries a theoretical risk of transmitting viral diseases. Additionally, it is expensive and needs to be infused IV over 3 to 6 hours, often resulting in the need for hospitalization. Furthermore its use is associated with a considerable incidence of early (fever and chills) and/or delayed (1-3 days) complications (aseptic meningitis, headache, nausea, and vomiting).^22,23  These complications, which are more common with larger/repeated doses, can result in additional morbidity/health care costs due to prolonged hospitalizations, hospital readmissions, emergency room visits, and neuroimaging costs.^22,24 

Many investigators have reported that in ITP, IVIG has a quicker onset of action than corticosteroids, leading to a faster increment in PC.^15,25  As the mechanisms of action are different between corticosteroids and IVIG, it is theorized (but without controlled evidence) that these therapies, if used in combination, might have synergistic effects, resulting in a faster increment in PC.²⁶  There is also some evidence to suggest the concomitant administration of corticosteroids with IVIG may decrease the incidence and severity of IVIG-related side effects.^22,27 

Until now, no randomized study has been conducted to compare the effects of concomitant administration of corticosteroids with IVIG vs IVIG alone in quickly increasing PCs and reducing IVIG-related side effects. We undertook this study to answer these questions.

This was a randomized, double-blinded, placebo-controlled, multicenter, prospective study performed in 6 Canadian pediatric centers (see “Acknowledgments”). The trial was registered at www.clinicaltrials.gov (NCT00376077) and underwent research ethics board approval in all 6 participating centers. The study commenced in 2005, and patient enrollment continued until 2016. The long study duration was due to a shift in practice toward observation and less treatment of children with ITP.

The study included children (<18 years) with ITP and PCs <20 × 10⁹/L, without life-threatening bleeding, in whom a clinical decision had been made (between the patient/family and their hematologist) to treat with IVIG.

Children with acute or chronic ITP (defined using the pre-2009 definitions [<6 months and ≥6 months since diagnosis, respectively]) were eligible for participation. Children were not enrolled within the first 24 hours of initial diagnosis of ITP, as it was deemed inappropriate to burden them with a discussion of this study at that time. Previous treatment with IVIG and/or corticosteroids was not an exclusion criterion, but children who were known to not respond to IVIG/corticosteroids were excluded. Children were only studied once. For full eligibility criteria, see Table 1.

Following consent, children were randomized to receive either regimen A (saline placebo prepared to look the same as the IV methylprednisolone [IVMP] given over 1 hour) or regimen B (IVMP [Solu-Medrol, UpJohn] 30 mg/kg [maximum 1 g] over 1 hour) both immediately followed by IVIG 1 g/kg (Gamunex 10%, Bayer/Talecris/Grifols) given over 2 to 3 hours (Figure 1). To ensure that the 2 groups had an equal distribution of patients with acute and chronic ITP, block randomization was used. Randomization was done by contacting the study pharmacist at the coordinating center, who kept the randomization codes and a table with the treatments that children had received. Children/families and evaluating physicians (performing bleeding scores and clinical examinations and reviewing adverse events [AEs]) were, at all times, blinded to the patient’s management regimen. The randomization code was only broken after all enrolled children completed the study.

At baseline, the following were performed: clinical history, physical examination, determination of the Buchanan and Adix bleeding score²⁸  (supplemental Appendix 1), and Kids ITP Tool (KIT) quality of life (QoL) evaluations (child self-report for children aged 7-17 years, parent proxy report for children aged 2-17 years, and a parent impact report for all children).²⁹  The bleeding score gives an overall categorical score from 0 (no bleeding) to 5 (life-threatening or fatal bleeding). The KIT is a continuous score from 0 (worst QoL) to 100 (best QoL). KIT and bleeding scores were only used to compare the 2 groups according to baseline disease severity and were not repeated following therapy.

The following laboratory tests were performed: complete blood count, urea, creatinine, glucose, conjugated and unconjugated bilirubin, Coombs test, and urinalysis. A screening pregnancy test (urine β-human chorionic gonadotropin) was performed on all postpubertal females.

Adverse effects of treatment were recorded using a specific AE questionnaire (supplemental Appendix 2) and monitored by the study’s Data Monitoring Safety Board.

Following the intervention, children had follow-up studies performed as per supplemental Appendix 3 (schedule of observations post treatment). To minimize inconvenience for children/families certain follow-up studies (complete blood count and urinalysis) were permitted to be performed in local laboratories close to children’s homes. The last blood sample was obtained at day 21. Follow-up AE questionnaires were completed by the parents/children in clinic or by telephone by local study coordinators.

Sample size was estimated based on several assumptions: (1) mean PC in children at baseline would be 10 × 10⁹/L; (2) children receiving IVIG alone would attain a mean PC of 30 × 10⁹/L 24 hours after initiation of therapy, with substantial interpatient variation (estimated standard deviation of 20 × 10⁹/L); and (3) children receiving the combined regimen (IVMP+IVIG) would attain a PC of 50 × 10⁹/L at the 24-hour time point with similarly large interpatient variability. For 80% power to show a difference in PC of 20 × 10⁹/L (at 24 hours) with an α of 0.025 (2 sided) and a β of 0.2, it was estimated that 32 evaluable children would be required.

Two-sample Student t tests were used to compare mean PCs, mean hemoglobin (Hb) levels and KIT scores at different time points between the 2 groups. χ² testing was done to compare the proportion of children attaining PCs of 20 × 10⁹/L and 50 × 10⁹/L at the different time points and the proportions of children showing AEs at different time points. Statistical significance was set at P < .05. Statistical analyses were done using SAS 9.4.

Thirty-three children were consented into the trial, but 1 immediately dropped out of study after randomization, leaving 32 participating children (ages: mean, 9.4 years; range, 1.2-17.5 years; 22 males and 10 females; 16 acute and 16 chronic; 11 blood type O and 21 non-O blood type) (Table 2). Mean age of children with acute ITP was 8.8 years vs 10.0 years for children with chronic ITP. Most children had previously been treated for ITP; 23 had received IVIG, while 26 had received corticosteroids. Four patients had never received any treatment.

Eighteen children were randomized to regimen A (placebo+IVIG) and 14 to regimen B (IVMP+IVIG). Randomization was not equal on account of a patient who after randomization to regimen B declined to participate, and a second patient was incorrectly assigned to regimen A. There was a slight (nonstatistically significant) difference in patient ages between the 2 groups; mean/median age was 9.0/6.5 years for regimen A vs 9.9/10.5 years for regimen B (P = .65).

All children started therapy with a PC of < 20 × 10⁹/L; mean baseline PC was 9.2 × 10⁹/L, with no intergroup difference (P = .69). Mean overall baseline Buchanan and Adix bleeding score was 1.32 (corresponding to a rating of minor bleeding); 22 children had scores of 0 or 1 (no/minor bleeding), while 5 had a score of 2 (mild bleeding) and 5 had a score of 3 (moderate bleeding). There was no statistical difference in bleeding scores between the 2 groups (P = .57). Bleeding scores did not correlate with whether children had acute or chronic ITP (P = .95) or with blood group (P = .98). Older subjects and males had slightly higher bleeding scores, but these failed to reach statistical significance (P = .33 and P = .20, respectively).

KIT child self-report scores were obtained from 17 children (age 7-17.5 years), while 21 parents answered the parent proxy for children (age 2-17 years) and 26 parents answered the KIT parent impact questionnaire. Baseline KIT scores are shown in Table 2. There were no statistical differences for all 3 KIT questionnaire scores between the 2 groups. No correlation was seen between KIT scores and baseline age, gender, or type of ITP (acute/chronic).

There was a trend for lower baseline PCs being associated with higher baseline bleeding scores (Pearson correlation P = .08). There was no significant association between baseline bleeding scores and baseline KIT scores (P = .36; analysis of variance).

Five (3 regimen A; 2 regimen B) children were Coombs positive at baseline. Two continued to show a positive Coombs test result at the 24-hour time point. No additional child developed a positive Coombs test result at 24 hours. Baseline creatinine, urea, and serum glucose were normal in all children. Two children showed slight abnormalities in baseline unconjugated bilirubin (21 μmol/L in both cases [normal, <17 μmol/L]).

Baseline urine dipstick results were positive for blood in 5 children (3 “trace” and 2 “moderate/large”), and in all cases, red blood cells were seen on urinalysis. No child showed hematuria with subsequent testing. No child had glucosuria at baseline or with subsequent testing.

All children received the initial placebo/IVMP over 1 hour followed by IVIG over a mean of 2 hours 32 minutes (range, 1 hour 47 minutes to 3 hours 22 minutes). Consequently, the total mean time to administer the full treatment (placebo+IVIG or IVMP+IVIG) was 3 hours 32 minutes (range, 2 hours 47 minutes to 4 hours 22 minutes).

There was a trend (P = .12) for younger children to show a more rapid increment in PC than older children (Figure 2).

By the end of the placebo+IVIG or IVMP+IVIG therapy, mean PCs for both groups had risen to a mean of 14.6 × 10⁹/L (range, 1-31 × 10⁹/L), and 32% of all children exhibited PCs ≥20 × 10⁹/L (no intergroup difference; P = .28) (Table 3; Figures 3 and 4). By 8 hours, mean PCs for the 2 groups were 23 × 10⁹/L and 22.7 × 10⁹/L (P = .94), respectively, and 55% (17/31) of all children had a PC ≥20 × 10⁹/L, with no intergroup difference (P = .99).

The difference in mean PC approached statistical significance (P = .06) at the 24-hour time point (55 × 10⁹/L for regimen A vs 76.9 × 10⁹/L for regimen B). Given the slight difference in age between the 2 groups (regimen B children being slightly older), and given that age was a confounder in PC increment, we post-hoc adjusted the analysis for age. In doing so, the difference in mean PC at 24 hours between the 2 groups became statistically significant (P = .035), favoring the combined therapy. At 24 hours, the proportion of children having achieved a PC of ≥20 × 10⁹/L was 89% (regimen A) vs 100% (regimen B) (P = .49), while the proportion of children having achieved a PC of ≥50 × 10⁹/L was 50% (regimen A) vs 77% (regimen B), respectively (P = .15).

Mean (range) PCs for all 32 children were 143 × 10⁹/L (27-303 × 10⁹/L) at 72 hours, 161 × 10⁹/L (11-487 × 10⁹/L) at 1 week, and 40 × 10⁹/L (1-162 × 10⁹/L) at day 21 after therapy. At all of these time points, there was no statistical difference in PCs between the groups. At 72 hours posttherapy all children showed PCs >20 × 10⁹/L, and 87% of all children showed PCs ≥50 × 10⁹/L, while by 1 week posttherapy, 4 out of 30 children (2 children did not have testing at 1 week) showed PCs <20 × 10⁹/L. By day 21, 4 children (3 randomized to regimen A and 1 to regimen B) had received additional platelet-enhancing therapies as a result of a PC of <10 to 20 × 10⁹/L and/or bleeding.

There was no statistical difference in response rates between males and females (P = .43); between children with acute vs chronic ITP (P = .64), between children who previously had received IVIG (n = 23) and those who had not (n = 9) (P = .55), or between those who had previously received steroids (n = 26) and those who had not (n = 4). all 5 patients with grade 3 bleeding scores at baseline (this included a patient with large hematuria at baseline) achieved a PC of >20 × 10⁹/L at 24 hours, as did the 2 patients with moderate-to-large hematuria. The 6 patients with either grade 3 bleeding at baseline or moderate-to-large hematuria showed a mean PC of 50.2 at 24 hours and 134 at 72 hours, which was not statistically different from the overall group of patients.

Following treatment, 30 children reported AEs. The 2 children who did not report AEs were both treated with regimen B. No patient experienced a severe bleed/unexpected severe AE during the study. The most common AE was headache, followed by nausea, vomiting, dizziness, unusual taste in the mouth, abdominal pain, mood changes, and increased appetite. The most common time point to report AEs was during the 24- to 72-hour time period.

Headache was reported during the 24- to 72-hour time period by 14 out of 16 children (87.5%) who received regimen A (excluding 2 children who had acute allergic reactions; see below) vs 7 out of 14 children (50%) who received regimen B (P = .046; Fisher’s exact test). At all other time points, the incidence of headache was not’different between the 2 groups. Nausea (only reported in combination with headache) was reported by 6 children on regimen A and 2 on regimen B. There was no relationship between the occurrence of headache or nausea and patient age/gender. Two children (5-year-old male and 15-year-old female; both randomized to regimen A) developed allergic reactions upon receiving IVIG. In both cases, their reactions resolved with medical therapy; the first patient received IV diphenhydramine and acetaminophen, while the second patient received IV hydrocortisone and IV diphenhydramine.

At baseline, 31 out of 32 children had a normal Hb level. A 1.2-year-old male had mild iron-deficiency anemia (Hb 101 g/L). Overall, the mean Hb level among all children showed a statistically significant (P = .0001; paired Student t test) drop (mean drop, 8.8%) from a baseline mean of 133.3 g/L (range, 101-166 g/L) to a mean of 121.6 g/L (range, 89-149 g/L) at time 8 hours; the mean Hb drop was 7.5% in the 11 O blood type individuals vs 9.6% in the 21 non-O blood type individuals (P = .07; paired Student t test). The 5 children with a positive baseline Coombs test result showed no overall difference in Hb drop compared with the other children. By 72 hours posttreatment, the mean Hb level among all children had risen to 128 g/L, while at 1 week, it was 131.7 g/L (both not statistically different from baseline).

This study evaluated 3 main questions: (1) How fast does the PC increase with IVIG alone? (2) Does it increase faster with the combination of IVMP (1 dose) with IVIG? (3) Does the combination therapy result in fewer/more side effects than IVIG alone?

Regarding the first question, our study showed that PCs increased more rapidly than predicted following IVIG administration alone or with IVMP. By 8 hours after the initiation of therapy, 55% of all children (both regimens) had achieved a PC ≥20 × 10⁹/L, and by 24 hours after the initiation of therapy, 89% of children receiving only IVIG had achieved a PC ≥20 × 10⁹/L (vs 100% of those receiving the combination therapy). Previous reports of rapidity of PC increment with IVIG have shown slower increments. Erduran and colleagues reported in 2003 that 86% of 42 children with ITP receiving IVIG 1 g/kg per day × 2 days had achieved a PC of >20 × 10⁹/L by 48 hours.³⁰  We showed a similar percentage (89%) of children receiving only IVIG (regimen A) achieving this, but doing so by 24 hours. The more rapid increase in PCs in our study we speculate could be a function of the rapid (over 2-3 hours) IVIG infusion rate used compared with the traditional ∼6-hour infusion rate used before the mid-2000s. We must however acknowledge that by excluding previous nonresponders to IVIG, we may have selected for patients who would show a better response to IVIG. Of course, a definite benefit of a faster IVIG infusion is that it results in less time that patients/families need to spend in a clinic/hospital, thus reducing the burden to patients/families and potentially reducing health care costs.

Regarding the second question, our study showed that IVMP+IVIG led to a faster rise in PC than IVIG alone. At 24 hours after the initiation of therapy (∼20.5 hours after completion of IVIG), children receiving the combination therapy showed a PC that was on average 21.9 × 10⁹/L higher than children receiving only IVIG (76.9 × 10⁹/L vs 55 × 10⁹/L; P = .035; adjusted for age). Furthermore, a higher proportion of children treated with combination therapy showed a PC of >50 × 10⁹/L at the 24-hour time point (77% vs 50%).

Two studies, both nonrandomized and retrospective, also evaluated the rapidity of the PC increment in children with ITP when given IVMP+IVIG.^31,32  There were notable differences between those studies and ours. The first study, by Barrios and colleagues, was a noncomparative study involving 11 children who started with a higher mean PC (19 × 10⁹/L) than our children, received a lower dose of IVMP (20 mg/kg) than we used (30 mg/kg), and received IVIG over 5 hours (vs 2-3 hours in our study).³¹  They reported that by 24 hours, the mean PC in their 11 children was 111 × 10⁹/L (5.8 × baseline), while we reported a mean PC at 24 hours in the children receiving the combination therapy (regimen B) of 76.9 × 10⁹/L (8 × baseline). The second was a 4-year comparative study of 148 children with acute ITP who had been treated with IVMP (30 mg/kg per day × 3 days), IVIG (1 g/kg per day × 2 days), or the combination of both.³²  The authors reported that PCs rose fastest in patients receiving combination therapy; by 24 hours after the initiation of therapy, the mean PC was 50 × 10⁹/L in children receiving combination therapy.

Although we found a significant intergroup difference in PCs at the 24-hour time point, we did not find a difference at either earlier (end of therapy or 8 hours) or later time points (72 hours; 1 week or 3 weeks). This suggests that the effect of IVMP in raising PCs only becomes noticeable sometime after 8 hours. For this study, it was not felt that continuation of IV/oral steroids, which would have entailed prolonged hospitalization or daily clinic visits, increased venipunctures, and/or increased treatment side effects, was ethically justifiable, as subjects were not in a life-threatening situation. In the setting of an actual life-threatening hemorrhage, IV corticosteroids and IVIG would likely be continued and could potentially boost further PCs at times 24, 48, and 72 hours after the initiation of combined therapy.

Age at initial diagnosis has been shown to be a predictor of ITP resolution, with younger children (≤10 years) being more likely to resolve their ITP then older children (>10 years).^33,34  However, the effect of age on the rapidity of PC increment following therapy has not been well studied. A recent Japanese study showed that children <23 months of age (n = 24) were more likely to achieve a better PC response at the 2-week mark following therapy than older children (≥23 months; n = 29).³⁵  However, in that study, the initial rise in PC in the first 24 hours was not described. We found a trend for older children to show a less brisk increment in PC.

Regarding our third question, we found that the coadministration of 1 dose of IVMP pre-IVIG reduced AEs from IVIG. A high incidence of AEs when IVIG is given alone has previously been reported.²⁵  Our study is in keeping with several nonrandomized and nonblinded reports showing that administration of IV/oral corticosteroids reduces the incidence of IVIG-related side effects.^22,27 

Additional findings in our study include a statistically significant drop (8.8%) in mean Hb from baseline to the 8-hour mark among all children. This drop, we believe, is due to 2 processes: (1) a dilutional effect from the administration of IVIG, as 1 g Gamunex represents a volume of 10 mL, and all children received 1 g (10 mL)/kg; and (2) hemolysis resulting from anti-A or anti-B alloantibodies present in IVIG.^36,37  The former applies to all children, while the latter only to non-O blood type individuals. We found that the hemolysis arising from 1 dose of IVIG (1 g/kg) in non-O blood type individuals was relatively minor and not clinically significant.

KIT scores at study enrollment were comparable to what has been reported for children with acute ITP in the past, which tend to be in the range of 65 to 75.^38,39  We, like others, found (1) much lower parent impact scores, reflecting a considerable parental burden from childhood ITP; and (2) a lack of statistically significant correlation among PCs, bleeding scores, and KIT scores.^38,40  As we did not repeat KIT scores posttherapy, we cannot comment on the impact of treatment on follow-up KIT scores.

Our study did have limitations. It involved a relatively small number of children and unequal randomization (18 vs 14). Both of these things reduced the statistical power of our study and contributed to our initial univariate results not achieving statistical significance when comparing platelet increment with placebo+IVIG vs IVMP+IVIG. It was only after post-hoc adjustment for differences in patient age that we saw a statistical significance. Also, we excluded newly diagnosed children within 24 hours of presentation and children with life-threatening bleeding. Potentially, children who initially present with life-threatening bleeds may represent a group that might respond differently to treatment.

In our study, most of the children had previously been treated with IVIG and/or corticosteroids and consequently to be eligible for our study must have responded to these therapies, as we excluded those who previously demonstrated lack of response to IVIG/corticosteroids. Hence, our study findings, if extrapolated to all children (not just those who have demonstrated response to IVIG/corticosteroids), might overestimate the response to both study regimens.

In summary, our randomized, double-blinded study shows a more rapid response to IVIG (1 g/kg) when given as an infusion over 2 to 3 hours than previously recognized and that the combination of IVMP and IVIG leads to a more rapid increment in PC than IVIG alone. Consequently, our study is supportive of the use of combination therapy in situations of life-threatening bleeds in children with primary ITP, where it is crucial to increase the PC as quickly as possible. In such situations, we would still advocate strongly for administering a large platelet transfusion and other adjunctive hemostatic therapies (eg, recombinant factor VIIa and IV tranexamic acid⁴¹ ). Our findings also show a beneficial effect of reducing IVIG-related AEs with the use of 1 dose of IVMP. Given its demonstrable benefits in reducing side effects from IVIG, clinicians might consider adding 1 dose of IVMP (pre-IVIG) if they decide to use IVIG to treat a child with primary ITP, even in a non-life-threatening bleeding state. Our study does not, however, imply that all/most children when presenting with a low PC without significant bleeding should necessarily be treated, as many studies have shown a very low rate of bleeding in such children.

The authors thank the patients/families, nurses, and clinical research associates in our centers who participated in the study (Hospital for Sick Children, Toronto, ON, Canada; Children’s Hospital of Eastern Ottawa, Ottawa, ON, Canada; Kingston General Hospital, Kingston, ON, Canada; Ste-Justine Hôpital, Montreal, QC, Canada; IWK Health Centre, Halifax, NS, Canada; and Alberta Children’s Hospital, Calgary, AB, Canada). The authors also thank the members of the Data Monitoring Safety Board (Kaiser Ali, Sara Israels, and Shinya Ito).

Funding for this study was provided through an investigator-initiated peer-reviewed grant submitted to the Bayer-Talecris-Canadian Blood Services Partnership fund.

Contribution: M.C. designed the study, coordinated and supervised data collection, carried out analyses, interpreted data, drafted the initial manuscript, and approved the final manuscript; M. Silva, M.D., R.J.K., M. Steele, and V.P. supervised data collection, revised the manuscript, and approved the final manuscript; C.W. and L.K. coordinated data collection, revised the manuscript, and approved the final manuscript; D.S. was the statistician for the study and carried out analyses, interpreted data, revised the manuscript, and approved the final manuscript; and V.S.B. designed the study, interpreted data, revised the manuscript, and approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Manuel Carcao, Division of Haematology/Oncology, Hospital for Sick Children, Paediatrics, 555 University Ave, Toronto, ON M5G 1X8, Canada; e-mail: manuel.carcao@sickkids.ca.