Optimal dose of adrenaline auto‐injector for children and young people at risk of anaphylaxis: A phase IV randomized controlled crossover study

Guidelines recommend intramuscular injection of 500 μg adrenaline (epinephrine) for anaphylaxis in teenagers and adults; however, most autoinjectors deliver a maximum 300 μg dose. We evaluated plasma adrenaline levels and cardiovascular parameters (including cardiac output) following self‐injection with 300 μg or 500 μg adrenaline in teenagers at risk of anaphylaxis.


| INTRODUC TI ON
Anaphylaxis is a serious systemic hypersensitivity reaction that is usually rapid in onset and may cause death. 1 Up to 5% of the population are estimated to be at risk of food-induced anaphylaxis. 1 Hospitalizations due to anaphylaxis have increased threefold in the UK since 1998, 2 a trend that is reproduced globally. 3 The firstline treatment for anaphylaxis in international guidelines is prompt administration of intramuscular (IM) adrenaline (epinephrine): Adrenaline has beneficial effects on cardiac and lung function, and reduces degranulation of mast cells and other effector cells involved in the allergic reaction. 1,4,5 Individuals at risk of anaphylaxis to food and/or venom are usually prescribed injectable adrenaline in the form of an adrenaline autoinjector (AAI) for emergency use. In the UK, AAI prescriptions increased by 336% from 1998 to 2018, representing a year-on-year increase of 11%. 2 National and international guidelines recommend 500 μg adrenaline to treat anaphylaxis in older children and adults in healthcare settings. 1,4,5 In an update of their 2014 guideline, the European Academy of Allergy and Clinical Immunology (EAACI) Anaphylaxis Task Force suggested that a dose of at least 300 μg adrenaline (as an autoinjector) should be used for adolescents and adults, although whether this applies to healthcare settings is unclear. 6 However, the majority of AAI currently available deliver a maximum of 300 μg adrenaline; as a consequence, adolescents/adults weighing >30 kg are effectively underdosed in the community setting, which might increase the risk of severe outcomes. A single dose of adrenaline is ineffective in around 10% of individuals during anaphylaxis, 7 The limited available data relating to the pharmacokinetics of injected adrenaline suggest that IM (in contrast to subcutaneous) injection of adrenaline into the thigh results in a higher and faster peak plasma adrenaline level in children 12 ; however, the subcutaneous injections were administered into the deltoid region rather than the thigh, 13 and a lower adrenaline dose was used for some of the subcutaneous injections. 12 Nonetheless, it is on this basis that international guidelines for anaphylaxis management recommend the IM K E Y W O R D S adrenaline, anaphylaxis, auto-injector, epinephrine, stroke volume

G R A P H I C A L A B S T R A C T
This crossover study assesses the pharmacokinetics/pharmacodynamics of self-injection with 300 μg or 500 μg adrenaline in food-allergic teenagers at risk of anaphylaxis. Our findings challenge the assumption that optimal cardiovascular effects are associated with the adrenaline autoinjector causing the fastest and highest peak in plasma adrenaline. These data justify the global availability of adrenaline autoinjectors which deliver 500 μg adrenaline for use in adolescents and adults over 40 kg. Abbreviations: IM, intramuscular; min, minutes route for adrenaline administration. A further concern raised by both the MHRA and CHMP related to needle length. 10,11 Recent reports have highlighted that a 16 mm needle length (such as that used in Epipen®) may be inadequate to deliver an IM injection in up to 20% of adults (typically females), 14,15 and up to one third of children. 16 To address the paucity of data relating to these issues, we undertook a crossover study to assess the pharmacokinetics/pharmacodynamics (PK/PD) of self-injection with 300 μg or 500 μg adrenaline in foodallergic teenagers at risk of anaphylaxis.

| Study design
We undertook a single-center, randomized crossover Phase IV study comparing two doses of IM adrenaline (300 and 500 μg) delivered via the same AAI device (Emerade®) in food-allergic teenagers at risk of anaphylaxis. We further explored the differences in PK/PD between 300 μg when delivered using 2 different devices with 16 and 23 mm needle length (Epipen® and Emerade®, respectively), as a secondary outcome. Written informed consent/assent was obtained in all participants, in addition to parental consent in participants under 16 years. The study sponsor was Imperial College London.

| Participants
Participants were recruited from our clinical allergy service. Eligible participants were 13-18 years with a history of immunoglobulin Emediated food allergy and weighing ≥40 kg. Patients with known cardiac, endocrine, or renal comorbidities, or poor asthma control requiring daily bronchodilator were excluded. Participants were in good health, with no recent use of short-acting beta-agonists/antagonists, and avoided all food from 2 h prior to injection. Caffeine or xanthine-containing foods were avoided in the preceding 24 h, since these can affect plasma catecholamine levels.

| Randomization and masking
Participants were randomized to 1 of 4 possible administration sequences using an online randomization tool (www.rando mizat ion.com) following a balanced, randomized-block crossover design ( Figure 1 and Appendix p 2). This was designed to reduce the influence of circadian rhythm on endogenous catecholamine levels, with respect to the primary outcome. To reduce possible bias, participants were blinded to the dose delivered by Emerade®. A separate randomization list was used to determine whether to use the left or right anterolateral thigh for the site of injection with the Emerade device (both doses were given in the same leg, on separate occasions).

| Procedures
Each participant received a total of 3 adrenaline injections over 2 separate visits, at least 28 days apart: 300 μg delivered by Emerade® (23 mm needle), 500 μg delivered by Emerade® (23 mm needle), and 300 μg delivered by Epipen® (16 mm needle), as summarized in Table S1. On the basis of previous data, a minimum washout period of 4 h was used in between the two 300 μg injections. 10,11 The injection site was marked at the midpoint of the anterolateral thigh on each side, in indelible ink. The skin to muscle depth (STMD) and skin to bone (STBD) depth were measured at this point (with and without compression) by ultrasound (Vscan, GE Healthcare, Chicago, Illinois) (see Figure S1). An intravenous cannula was sited (following application of local anesthetic cream) and venous blood samples collected F I G U R E 1 CONSORT diagram and participant allocation. V1/2: visit 1/2. am, morning, pm, afternoon. every 30 min until self-injection to assess the stability of baseline measurements.
Participants underwent intensive training and then selfadministered adrenaline under supervision, at least 60 min after cannulation, using the allocated AAI device according to the approved summary of product characteristics. IM injection was confirmed by ultrasound. Serial blood samples were collected at the following timepoints after injection: 5, 10, 15, 20, 30, 45, 60, 80, 100, 120, and 180 min. Participants were required to remain semi-recumbent from 1 h prior to injection, until the final blood sample.
Continuous cardiovascular monitoring was undertaken using a validated, FDA-approved non-invasive method to measure heart rate (HR) and stroke volume (SV) through bioreactance and peripheral systolic and diastolic blood pressure (BP) by oscillometry

| Catecholamine assay
Blood was collected into lithium heparin tubes and immediately cen-

| Outcomes
The primary outcome was the plasma catecholamine profile following injection in each patient, assessed as the maximum plasma concentration (C max ), time to C max (T max ) for each peak in plasma adrenaline identified, and area-under-curve (AUC), following 300 and 500 μg adrenaline delivered with Emerade®. Secondary outcomes were the plasma catecholamine profile following 300 μg adrenaline delivered using 2 devices with different needle lengths (i.e., Epipen and Emerade); change in cardiovascular parameters (HR, BP, SV CO) following injection; and adverse events.

| Statistical analyses
Analyses were planned prospectively as detailed in a statistical analysis plan. Time-concentration profiles were produced for each individual administration, and the following parameters calculated: C max , T max , and area under the plasma adrenaline versus time curve (AUC) from 0 to 30, 60 and 180 min. Geometric means were used for analyses, and all comparisons for AUC were baseline-corrected. C max was analyzed using both uncorrected and baseline-corrected values.
For assessment of bioequivalence, geometric means with 90% CI were used, as is established practice. 17

| Patient and public involvement
We consulted with food-allergic teenagers and their parents in setting the research question, study design and in compiling the study information leaflets. In particular, we received feedback that participants would be less likely to attend for 3 separate visits, but would consider self-injection with 2 doses on a single day. Since the PK/PD comparison of two 300 μg injections was not the primary outcome, we included in the protocol a visit with 2 injections utilizing a suitable wash-out period (rather than on 2 separate days).

| Role of the funding source
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.  Table 1. All participants completed the study protocol ( Figure 1).

Intramuscular delivery of adrenaline was confirmed in all cases
by ultrasound and resulted in a biphasic absorption profile, irrespective of dose ( Figure 2). 500 μg adrenaline caused a greater and more sustained peak plasma adrenaline level (C max ) compared to 300 μg (Table 2 and Figure 2, p = 0.03). Compared to Emerade (23G needle, length 23 mm), injection of 300 μg adrenaline by Epipen (21G needle, length 16 mm) resulted in a faster time to peak level (p = 0.006) but otherwise no significant difference in other pharmacokinetic parameters ( Table 2 and Figure 2). This was also reflected in assessing for bioequivalence ( Table 3 and Table S2), although our analysis was underpowered for this evaluation. We did not identify any order effect. No differences were seen for plasma noradrenaline levels (see Figure S2; p = 0.90). We did not identify any significant variability in baseline plasma adrenaline levels (intra-subject coefficient of variation (CV) 17%, inter-subject CV 3.4%-8.4%), although there was significant inter-individual variation with CVs ranging from 26% to 81%. There was no significant correlation between peak adrenaline levels and STMD.
Injection with all three devices resulted in an increase in HR and systolic BP within 5 min ( Figure 3A-C). 500 μg injection caused a greater and more sustained increase in HR, SV, and CO compared to 300 μg (Tables S3-S6). Adrenaline delivered by Emerade resulted in an increase in both HR and SV which was sustained for at least 60 min. In contrast, Epipen resulted in a transient increase in HR but a small decrease in in stroke volume (p = 0.036, Figure 3D) and cardiac output (p = 0.044, Figure 3E). For all three devices, the change in HR and SV mirrored the pharmacokinetic profile for adrenaline for each device (Figure 4. For individual participant plots of the change in HR and SV, see Figures S3 and S4).
Adrenaline injection was well tolerated, irrespective of dose or device used (Table S7). Finally, using ultrasound imaging, we were TA B L E 1 Baseline characteristics of the intention-to-treat population.

| DISCUSS ION
Intramuscular injection with 500 μg adrenaline resulted in both a higher and more sustained plasma adrenaline level compared to TA B L E 2 Pharmacokinetic data, intention-to-treat population. Epipen 300 μg resulted in a faster and slightly higher initial C max , we observed a negative inotropic effect (reduction in stroke volume) despite a similar plasma adrenaline profile. These PK data are consistent with those previously reported for adrenaline injection by needle/syringe and AAI device in commercially-funded studies. [18][19][20] This is the first study to evaluate changes in both PK parameters and stroke volume/cardiac output following AAI administration. The European Medicines Agency (EMA) has required manufacturers of AAI to provide further PK/PD data "in order to understand the influence of different factors on distribution, exposure and activity of adrenaline when administered via an adrenaline autoinjector device". 11 However, most studies undertaken in response to the EMA report have focused on plasma adrenaline levels following injection, rather than cardiovascular effects. [18][19][20][21] There are no data as to the optimum level of plasma adrenaline to treat anaphylaxis, 22 a knowledge gap that we sought to address in this study.
During allergic reactions, there is a fall in stroke volume: Ruiz-Garcia et al observed a mean reduction in stroke volume of ~10%, even during relatively mild systemic allergic reactions to peanut in human volunteers. 23 Therefore, stroke volume is arguably the most relevant outcome measure in PK/PD evaluations of AAI. Our observation that IM injection of 300 μg using different devices had very different effects on stroke volume reinforces this. These data directly challenge the assumption that the optimal cardiovascular effects are associated with an injection which causes the highest peak plasma adrenaline levels.
The negative inotropic effect seen following injection with Epipen is concerning, given the limited impact of a single dose of IM adrenaline (by needle/syringe) on stroke volume during peanut-induced anaphylaxis. 24 One possible explanation is the known negative inotropic effect of catecholamines when present at high concentrations. 25,26 At lower doses, adrenaline is a positive inotrope, acting via β1-and β2adrenoreceptors (AR). However, the β2-AR can also couple to an inhibitory G protein, which inhibits cAMP formation (essential for cardiac muscle contraction). At higher doses, adrenaline acts through this inhibitory pathway, becoming a negative inotrope. This may explain why adrenaline given by Epipen (and theoretically, other AAI devices) can induce Takotsubo syndrome, 27 a sudden and acute form of heart failure.
F I G U R E 3 Change in cardiovascular outcomes from baseline following adrenaline injection with 3 different devices (n = 12). Data are arithmetic mean ± SEM.
There are at least 10 reports of Takotsubo syndrome requiring hospitalization after Epipen use, which prompted the FDA to include a warning on the product information sheet for Epipen since April 2017. 28 In both this and other studies, Epipen seems to cause a faster and higher rise in plasma adrenaline. We hypothesize that this rapid rise might result in inhibitory β2-AR signaling and negative inotropy. Likewise, the slower time to peak levels with Emerade (which could be due to device mechanism, the smaller needle caliber and/or a longer needle) prevents this.
Clearly, Epipen is usually effective in treating anaphylaxis: Epipen is the most common AAI prescribed in most regions, and 90% of food-induced anaphylaxis reactions respond to a single dose of IM adrenaline (irrespective of AAI device). 7 However, most food-induced anaphylaxis involve respiratory rather than significant cardiovascular compromise. Our concern relates to reactions with cardiovascular compromise, where the risk of negative inotropyparticularly after multiple Epipen doses-could be significant. While it may be helpful to use an AAI in the first instance to treat anaphylaxis in healthcare settings for speed, convenience, and education purposes, we recommend that if more than one dose of adrenaline is needed, that these are given by ampoule/needle/syringe to mitigate against the risk of negative inotropy-a recommendation consistent with guidelines from the Resuscitation Council UK. 29 We were unable to confirm bioequivalence for any of the AAI devices; however, our study was not powered for this comparison.
Furthermore, given the inter-individual variability in PK profiles, it is likely large sample sizes would be required. However, consistent with the above, there was no evidence of bioequivalence between Emerade 500 and Emerade 300 in terms of PK profile. The act of injection (and anxiety over doing so) might have impacted plasma adrenaline levels. However, by including an analysis of plasma noradrenaline (which represents endogenous catecholamine release), we found a possible effect only during the first 10-15 min after injection, and we did not observe any order effect (which might be expected if participants were more anxious over the first rather than last AAI administration).
There are some important limitations to note. The analysis of PK and particularly PD data was hampered by significant intraindividual variation (particularly for blood pressure)-something also noted in all other PK/PD studies of AAI. [18][19][20]

AUTH O R CO NTR I B UTI O N S
PJT, NP, and EI conceived the design and wrote the protocol. PJT, NP, EI, and BD recruited patients and performed the clinical interventions. NN and JD undertook the laboratory analyses. PJT, NP, and EI were involved in the interpretation of the data, and PJT, NP, and SF did the statistical analysis. PJT wrote the first draft of the manuscript and contributed to the writing of the report. All authors had unrestricted access to all the data in the study, reviewed the manuscript drafts, and approved the final version. All authors take final responsibility for the decision to submit for publication.

ACK N OWLED G M ENTS
We thank our study participants and their families.