Supplemental Material for Safety and Efficacy of Sodium Nitroprusside During Prolonged Infusion in Pediatric Patients

Supplemental Material for Safety and Efficacy of Sodium Nitroprusside during Prolonged Infusion in Pediatric Patients.

Protocol Inclusion/Exclusion Criteria Subjects must have met all of the following criteria: 1. Subject was less than 17 years of age. 2. An in-dwelling arterial line was clinically indicated. 3. Subject’s parent or legal guardian was willing and able to give informed parental permission signing and dating an IRB-approved informed parental permission containing all of the elements of informed consent, and subject provided assent, signing an IRB-approved and required informed assent, if applicable. 4. Subject was anticipated to require a minimum of 20 mm Hg (15 mm Hg for subjects < 2 years old) reduction in MAP for at least 12 hours using SNP [i.e., MAPB1 - MAPB2 ≥ 20 mm Hg (15 mm Hg for subjects < 2 years old)]

Subjects were excluded from the study if any of the following criteria existed: 1. Subject weighed < 3.0 kg. 2. Subject had a known allergy to SNP. 3. Subject had a known mitochondrial cytopathy with a disorder of oxidative phosphorylation or of respiratory chain enzymes. 4. Subject had a contraindication to vasodilator therapy for control of blood pressure during surgery or in the intensive care unit. 5. Subject had raised intracranial pressure. 6. Subject was anticipated to need anti-hypertensive drugs other than Sodium Nitroprusside either IV (e.g., dexmedetomidine, esmolol, etc.) or epidural (e.g., local anesthetics, clonidine, etc.) within three terminal half-lives (3X T½ receiving stable doses of an anti-hypertensive drug(s) prior to the initiation of study drug were eligible to be enrolled. 7. Subject had any serious medical condition which, in the opinion of the investigator, was likely to interfere with study procedures. 8. Subject was moribund (death likely to occur within 48 hours). 9. Subject had a positive result for the urine or serum HCG test administered at screening. 10. Subject had participated in other clinical trials for investigational drugs within 30 days prior to enrollment 11. Subject received sodium thiosulfate within 6 hours prior to the start of the open-label period. 12. Subject was either on, or anticipated to be on, ECMO.

Sample Size Determination A sample size of 30 patients in each of the two randomized groups was determined be needed to achieve 80% power to test the null hypothesis of equality in success rate between the groups using the chi square test and several plausible assumptions of the true rates in both groups (for example, 0.53 and 0.16 in the SNP and control groups, respectively). Although 63 total patients were enrolled, the sample size for the primary efficacy analysis was lower than expected due to high drop-out rates and patients who entered the blinded phase but did not have data that could be used for the primary analysis.

Time-to-failure Analysis As part of an exploratory analysis to describe differences in how quickly patients lost MAP control during the blinded phase, Kaplan-Meier estimators of the failure time distribution were calculated by treatment group. The log-rank p-value was used to compare failure time between treatment groups. In addition to participants excluded from the primary analysis, this time-to-failure analysis also excludes a patient in the placebo group who was classified as a failure before the end of the 30-minute blinded phase in the primary analysis but did not have MAP data recorded during the first 15 minutes of the blinded phase. The Kaplan-Meier curves in Supplemental Figure 3 show that failure rates were similar for the first 5 minutes of the blinded phase. In the next 5 minutes, the failure rate increases at a higher rate in the placebo group than in the SNP group before remaining parallel for the rest of the blinded phase. Across the blinded phase, the failure were significantly different overall (p-value = 0.028).

Efficacy Analysis Adjusted by Concomitant Medication Use Medication start and end dates were captured in the data system, but times were not captured. Thus, it is not known how soon before blinded phase concomitant medications were administered, although sites were recommended to not administer any vasoactive drug within 3 half-lives of the drug. In an exploratory analysis of the possible effect of antihypertensive medication use on maintaining MAP control during the blinded phase, medications were classified as having been administered near the time of the blinded phase if the medication was administered during open-label or blinded treatment and the end date was on or after the date of the blinded phase. The number of patients on cardiac medications of interest (vasodilators and other potentially antihypertensive drugs; ATC codes C01, C02, and C04) during this time frame was calculated. The percentage of patients in each randomized group on eligible cardiac medications of interest near the blinded phase was similar between treatment groups: 10/24 (42%) in the SNP group and 8/21 (38%) in the placebo group. Cardiac drugs taken by more than two blinded participants were milrinone, sildenafil, and papaverine. Failure rates in the placebo group for patients administered cardiac drug of interest near the time of the blinded phase were slightly lower (5/8; 63%) than in patients not administered these drugs (10/11; 91%). This had no effect on the failure rate in patients in the SNP group (5/11, 45%, without drugs and 4/9, 44%, with drugs). After adjustment for whether the patient received cardiac treatment, the odds of treatment failure was not substantially changed: 4.5 [(0.9, 24.4). The odds ratio for treatment failure adjusted for cardiac medication use was estimated using the Mantel-Haenszel estimate with exact 95% confidence limits. The odds ratio was similar to the unadjusted ratio: 4.5 [(0.92, 24.4)]. The number of patients on sedatives and other psycholeptic medications (ATC code N05) was also calculated. The percentage of patients in each randomized group on eligible psycholeptic concomitant medications near the blinded phase was also similar between groups: 13/24 (54%) in the SNP group and 11/21 (52%) in the placebo group. The percentage of patients in the blinded SNP group who were treatment failures were higher among patients who received psycholeptic medication; 2/9 (22%) SNP group patients not on psycholeptic medication failed compared to 7/11 (64%) SNP group patients on medication. In the placebo group, slightly lower rates of failure were found in patients on medication: 8/9 (89%) not on medication compared to 7/10 (70%) on medication. The adjusted odds ratio was again similar to the unadjusted ratio: 4 [(0.9, 25.1)].

Cyanide Dose-Response Modeling Methodology and Results Due to the limited variability in cyanide response at low infusion rates, a two-stage mixture model was fit to nonlinearly model the relationship between average infusion rate and cyanide levels at 8 hours. The response at 8 hours was modeled since that was the only cyanide collection scheduled to take place after the same infusion duration for all patients regardless of when they entered the blinded phase. For the first stage, the probability p that the cyanide level was below the LLQ was estimated using logistic regression with a logit link function with intercept b0p and infusion rate slope b1p. For the second stage, a gamma model was used to model the expected cyanide level conditioned on the cyanide value exceeding the LLQ. The mean response for these positive cyanide value was estimated through µ=exp(b0g+b1g D), where D is the dose and exp() is the exponential function. The predicted response as a function of dose is therefore f(D)=(1-p)µ. The model was fit in PROC NLMIXED in SAS version 9.3. This analysis was repeated to model the relationship between average study (open-label and blinded phase) SNP infusion rate and maximum study cyanide levels. Figure 3 plots the relationship between the average infusion rate through the first 8 hours of open-label infusion and the scheduled 8 hour cyanide values. Figure 4 plots maximum study cyanide levels against average study SNP infusion rate. At higher average infusion rates, most cyanide levels are above the LLQ. At the highest rates, variability in the observed concentrations and the low sample size lead to high variability in the model predictions. Except for higher maximum study expected levels, the shape of the relationship is the same. Alternate models with duration added and with total dose instead of infusion rate were fit, but infusion rate alone was selected as the better model based on model fit and a lower AIC value. The infusion rate at which the cyanide level is expected to be 0.5 mcg/mL was calculated using the Newton-Raphson method for both analyses. As can be seen in Figure 3, at 8 hours, this infusion rate was calculated to be 3.8 mcg/kg/min. Using the average SNP infusion rate through both the open-label and blinded phase, this infusion rate was calculated to be 3.1 mcg/kg/min. Parameter estimates for the two models are listed below.

8-Hour Overall Std Std Parameter Estimate Error Estimate Error

b0p 8.7 2.8 5.4 1.5

b1p -2.2 0.7 -1.5 0.4

b0g -1.3 1.3 -0.7 0.9

b1g 0.39 0.28 0.39 0.21

Relationship between Cyanide Levels and Creatinine Supplement Figure 7 plots the relationship between cyanide levels and change in creatinine from the baseline to study discontinuation collection. The cyanide level at the study discontinuation time point was used in the analysis. When the cyanide value at this time point was not collected or unevaluable, this value was imputed using maximum post-dose cyanide value. Multiple observations with the same cyanide and creatinine values are represented by darker circles in the plot. There is no consistent relationship between creatinine change and cyanide levels. The largest increases in creatinine of 0.3 mg/dL were experienced by two patients with baseline levels of 0.3 and 0.7 mg/dL. Two patients experienced creatinine decreases >0.3 mg/dL: from 3

0.6 to 0.2 and 1.1 to 0.4 mg/dL. These changes were found in patients with no cyanide assessments above the LLQ.

Supplemental Figures Supplemental Figure 1. Box plots of open-label infusion rates by age group (diamonds=group means).

Supplemental Figure 2. Kaplan-Meier plot of treatment failure time in each blinded phase treatment group.

Supplemental Figure 3. Scatter plot of maximum study whole blood cyanide levels and average open-label and blinded SNP infusion rate with model predictions and asymptotic 95% confidence interval.

Supplemental Figure 4. Scatter plot of 8 hour scheduled whole blood cyanide levels and average infusion rate through 8 hours with model predictions (solid line) and asymptotic 95% confidence interval.

Supplemental Figure 5. Scatter plot of creatinine change scores and study discontinuation cyanide levels with loess curve.

Supplemental Tables Supplemental Table 1. Thresholds used to classify the severity of laboratory and vital sign findings of special interest.

Finding Mild Moderate Severe Rebound Hypertension MAP rises >10% above MAP rises >20% above MAP rises >30% above baseline (Blinded Phase) baseline (MAPB1) baseline (MAPB1) (MAPB1) Hypotension MAP falls >20% below MAP falls >25% below MAP falls >30% below target target target; IV therapy required; IV therapy required; Repeated or Pharmacological therapy continuous pharmacological required therapy required Lactic Acidosis 5 -7.5 mmol/L 7.6-10 mmol/L >10 mmol/L Blood Cyanide 0.5 – 1.0 mcg/mL >1.0-1.5 mcg/mL >1.5 mcg/mL

Supplemental Table 2. Summary of the 8 hours post-dose, study drug discontinuation, and greatest study cyanide level (mean±SD or elevated ≥0.5 mcg/mL) by age group (<6 and ≥6 years) <6 Years ≥6 Years Total N Summary Elevated N Summary Elevated N Summary Elevated 8 Hours 27 0.5±0.9 7 (26%) 17 0.4±0.8 3 (18%) 44 0.4±0.9 10 (23%) Discontinuation 24 1.2±1.7 9 (38%) 18 0.7±1.5 3 (17%) 42 1±1.6 12 (29%) Greatest Study 29 1.3±2.1 11 (38%) 21 0.6±1.5 3 (14%) 50 1±1.9 14 (28%)

Supplemental Table 3. Post-treatment laboratory values and correlation with average SNP infusion rate and total dose during the open-label and blinded phases in patients who completed the open-label phase Age Group (Years) Correlation Total Total Average <6 >=6 (N=51) N Dose Rate Blood Thiocyanate (mcg/mL) N 30 21 51 45 0.6 0.62 Mean (SD) 6.8 (4.4) 7.2 (5.1) 7.0 (4.6) Median (Min, Max) 6.4 (0.0 - 17.0) 6.1 (0.0 - 18.0) 6.3 (0.0 - 18.0) Urine Thiocyanate (mcg/mL) N 30 21 51 45 0.439 0.381 Mean (SD) 4.5 (3.2) 7.3 (9.8) 5.6 (6.8) Median (Min, Max) 3.8 (0.0 - 11.0) 5.9 (0.0 - 44.0) 3.9 (0.0 - 44.0) Lactate (mmol/L) N 29 20 49 44 0.21 0.172 Mean (SD) 1.7 (1.2) 2.2 (1.2) 1.9 (1.2) Median (Min, Max) 1.3 (0.5 - 4.8) 2.0 (0.8 - 5.5) 1.6 (0.5 - 5.5) HCO3 (mEq/L) N 29 21 50 45 0.152 0.201 Mean (SD) 24.5 (2.6) 25.6 (3.7) 24.9 (3.2) Median (Min, Max) 25.0 (17.0 - 30.0) 25.0 (21.0 - 35.0) 25.0 (17.0 - 35.0) pH N 29 21 50 45 0.076 0.054 Mean (SD) 7.38 (0.04) 7.39 (0.05) 7.39 (0.04) Median (Min, Max) 7.39 (7.30 - 7.49) 7.40 (7.31 - 7.49) 7.39 (7.30 - 7.49) Methemoglobin (%) N 29 21 50 45 0.063 0.018 Mean (SD) 1.1 (0.6) 1.4 (2.3) 1.2 (1.6) Median (Min, Max) 1.2 (0.0 - 2.3) 1.0 (0.0 - 11.3) 1.1 (0.0 - 11.3) ALT (U/L) N 29 20 49 45 -0.16 -0.16 Mean (SD) 41.9 (72.4) 24.3 (14.9) 34.7 (56.8) Median (Min, Max) 18.0 (9.0 - 301.0) 19.5 (10.0 - 74.0) 19.0 (9.0 - 301.0) AST (U/L) N 29 19 48 44 -0.07 -0.11 Mean (SD) 93.9 (182.6) 41.6 (22.4) 73.2 (143.9) Median (Min, Max) 49.0 (18.0 - 37.0 (21.0 - 44.0 (18.0 - 1023.0) 112.0) 1023.0) Creatinine (mg/dL) N 27 21 48 43 -0.08 -0.1 Mean (SD) 0.4 (0.2) 0.6 (0.2) 0.5 (0.2) Median (Min, Max) 0.3 (0.2 - 0.8) 0.5 (0.2 - 1.0) 0.4 (0.2 - 1.0)