Pharmacokinetics of adrenaline autoinjectors

Anaphylaxis is a medical emergency with adrenaline acknowledged as the first‐line therapy. It is therefore important that patients have access to self‐injectable adrenaline in the community. Manufacturers have been requested by European Medicine Regulators to generate pharmacokinetic data for these autoinjector devices. For the first time, these data provide an insight into how individual devices work in different populations, and how they compare. We undertook a thorough literature search and also accessed grey literature, using searches of medicine regulators’ websites and freedom of information requests. The data demonstrate that it takes at least 5–10 min to achieve early peak plasma concentration for most devices. The specific autoinjector device seems to be the most important determinant of pharmacokinetics, with different devices giving rise to different plasma adrenaline profiles. Needle length does not seem to be the most important factor; rather, the force and speed of injection (which varies from one device to another) is likely to be of greater importance. In general, peak plasma adrenaline concentration is lower and time‐to‐peak concentration longer with increased skin‐to‐muscle depth. However, it is difficult to draw conclusions with the current available data, due to a lack of head‐to‐head comparisons, small numbers of study participants and the failure to acknowledge the biphasic nature of intramuscular adrenaline absorption for analysis purposes.


| INTRODUC TI ON
Anaphylaxis is an important medical emergency, with an estimated prevalence worldwide of 1-761 per 100 000 person-years for all causes. 1 Hospital admissions due to anaphylaxis are increasing globally; the most common triggers are foods such as peanut tree nuts and milk, wasp and bee stings and medications. 2,3 Presentations usually involve respiratory distress and/or cardiovascular collapse, but rarely result in fatal outcomes. [3][4][5][6] The mainstay of longer-term management is avoidance of the trigger. 7,8 This can be challenging, particularly for food allergy, with the issues around allergen labelling 9 which impact adversely on quality of life. 10 The evidence base for the acute management of anaphylaxis is weak, but there is a global consensus that intramuscular (IM) adrenaline is the treatment of choice. 7,8,11,12 In community settings, adrenaline can be provided for emergency use as an adrenaline autoinjector (AAI) device, 7,8 although these are not available in many countries. 13 Carrying an AAI enables IM adrenaline to be rapidly administered by the patient or a lay person. There have, however, been concerns that with some AAIs having shorter needle lengths, this could result in a subcutaneous rather than IM dose in many individuals. 14 In 2015, the Committee for Medicinal Products for Human Use (part of the European Medicines Agency) undertook a review in this area, 15 noting that a number of different factors could influence the delivery of adrenaline via an AAI: 'needle length, the thickness of fat under the skin, the way the autoinjector works (e.g., if it is spring loaded or not), the angle at which the device is placed on the skin and the force used to activate the device as well as how well the user follows the instructions for injection'.
The European Medicines Agency asked manufacturers to generate data to allow a better understanding of the pharmacokinetics (PK) of adrenaline delivery by autoinjectors (Box 1). These data were expected to substantially add to the previously published data.
In this review, we present a summary of the PK data now available for AAIs. We searched both published literature and grey literature to collate the evidence (Box 2) and reviewed, discussed and synthesized the available data to inform our clinical approach to managing patients at risk of anaphylaxis.

| PHARMACOK INE TI C (PK ) S TUD IE S OF ADRENALINE AUTOINJEC TOR S
A total of nine studies were identified in the literature (Table 1): six were published in peer-reviewed journals, [17][18][19][20][21][22] one is currently published as an abstract 23 while the last two were available from national regulatory bodies. [24][25][26][27][28][29][30] The four older studies were from the same group, the first 3 focusing on first-generation Epipen (future references to Epipen are second generation device unless otherwise stated), [17][18][19] and the fourth comparing second generation Epipen to Auvi-Q. 20 Included participants were either children at risk of anaphylaxis 17,19 or healthy adults. 18,20 Amongst more recent studies, all bar one 23 were undertaken by manufacturers in response to the 2015 EMA request. 21,22,[27][28][29][30] Only one study included a comparison of devices produced by different manufacturers, 27 while one compared different doses with the same device 23 ; otherwise, the comparison was to adrenaline given by needle and syringe.
Notably, a consistent feature across all studies is the considerable 'noise' in the PK parameters as evidenced by the coefficients of variation in the reported data. In addition, the three oldest studies (all by the same group) report 10-fold greater plasma adrenaline concentrations [17][18][19] than more recent studies, including a study by the same group in 2013 20 ; the reasons for this are not clear. Of note, there is a high level of consistency in peak plasma adrenaline concentrations across more recent studies, which implies a possible difference in the assay used in the earlier studies.

| IMPAC T OF S KIN -TO -MUSCLE DEP TH
Four studies-all commercially funded-attempted to evaluate the impact of body mass index, using the parameter of skin-to-muscle depth (STMD), assessed by ultrasound (Table 2). [21][22][23]27 These studies categorized participants (healthy adults) into subgroups with low (<15 mm), moderate (15-20 mm) and high (>20 mm) STMD. With the exception of the study commissioned by the manufacturer of Anapen, 21 all had a crossover design. Unfortunately, the numbers in each subgroup were too few to be powered to detect any small differences between groups.
Two studies evaluated Epipen 22,27 : one (manufacturer-funded) 22 demonstrated no discernible impact of STMD on PK parameters (Cmax, Tmax) ( Figure 1) although there was a trend towards increasing Tmax (i.e., time-to-peak adrenaline) with increasing STMD (Table 2). 22 A second study by the manufacturer of a different device found similar findings (Figure 2, Tables 2 and 3). 27

Key Messages
• The early peak plasma concentration occurs at 5-10 min for most adrenaline autoinjector devices.
• Injection force and speed may be more important than needle length for determining adrenaline pharmacokinetics.
• Peak plasma adrenaline concentrations are generally lower and time-to-peak concentration longer with increased skin-to-muscle depth.   Note: Data are mean ( ± standard error) unless otherwise stated. Units unified where possible. Many reports do not specify whether means are geometric or arithmetic, which may explain apparent discrepancies between graphs and data reported in tables. ♦Median. § Tmax reported is for first 20 mins. *Authors state that Auvi-Q and EpiPen were bioequivalent; see Table 3 for bioequivalence data for other devices.

BOX 1 Glossary box
Abbreviations: A, arm; IM, intramuscular; LT, lower thigh; MT, mid-thigh; SC, subcutaneous; SmPC, summary of product characteristics; STMD, skinto-muscle depth.  (Table 2). 27 The PK profile in the low STMD cohort was consistent with that seen in an independently funded study using Emerade in teenagers (Table 1). 23 The manufacturer's study assessing Anapen compared PK profile in 18 normal weight men with 12 overweight women (all with STMD >15 mm), in a non-crossover design ( Table 2). 21 A clear difference was seen between the 2 groups, with a faster and greater increase in plasma adrenaline in the normal weight men.

Investigated
With respect to Jext, limited information from the manufacturer's study was obtained from the UK Regulator through a freedom of information request ( Table 2). 30 There was significant evidence of a delay in adrenaline absorption with increasing STMD, prompting a rewording of the manufacturer's summary of product characteristics (SmPC) to state: 'adrenaline absorption in patients with a thick subcutaneous fat layer (i.e., STMD, skin-to-muscle depth >20 mm) is slower than in subjects with a thinner subcutaneous fat layer'. 28,29 Furthermore, this was also observed for when comparing PK parameters for Jext and IM injection using a needle/syringe. While data were comparable for the first 16 min across all cohorts, when evaluated for PK profile up to 30 min after injection, plasma adrenaline was significantly lower for Jext compared with manual IM injection in the STMD >20 mm cohort (Tables 2 and 3). This observation is also reflected in changes to the wording in the Jext SmPC and is consistent with data comparing Emerade, Epipen and Jext (in the study funded by the manufacturer of Emerade). 27 Interestingly, AUC 0-last for plasma adrenaline (which reflects overall absorption/elimination for at least 3 h after injection) is generally greater with increasing STMD: this was seen for Emerade 500 mcg (but not 300 mcg), 27 Table 2). Participants with a higher STMD seem to have a larger, delayed second peak in Cmax, around 1 h after injection. One possible explanation is that adrenaline may induce transient local vasoconstriction at the site of injection, which causes a 'modified release' phenomenon (in much the same way that co-injection with adrenaline is often used to prolong the effect of local anaesthesia). 31 This may be amplified in individuals with higher STMD, as the growth factors associated with increased adiposity also result in vascularization. 32 Skin-to-muscle depth varies with sex, with higher STMD reported in females compared with males of equivalent body mass index. 33  In summary, the available data suggest that absorption of adrenaline following percutaneous injection into the mid-thigh in patients with higher STMD is often delayed, with a lower initial peak concentration and a longer time-to-maximum concentration. This potentially means that individuals with higher STMD have a significantly lower plasma adrenaline in the first 20-30 min after injection with an AAI. Unfortunately, analysis of PK data in each report is hampered by the lack of recognition of a biphasic profile to the absorption of adrenaline given by IM injection, something wellestablished in the literature. The data are also confounded by significant intra-and inter-individual variability in PK outcomes, relatively small cohort sizes and a consequential absence of formal statistical testing.

| NEEDLE LENG TH AND INTR AMUSCUL AR VER S E S SUBCUTANEOUS INJEC TION
The impact of increasing STMD on the delivery of adrenaline may This manufacturer study also compared Epipen and Jext (which have 15mm/16mm needles, respectively) to Emerade (23 mm needle); a reasonable assumption would be that the former would not deliver an IM injection in subjects where the STMD exceeds the needle length. 27 However, Emerade did not result in a more favourable plasma adrenaline profile. In fact, absorption was significantly better and faster (i.e., not bioequivalent) with Epipen, and paradoxically, this was most evident in patients with higher STMD (Table 3). 27 Curiously, this was not seen for Jext, despite the fact that Epipen and Jext are often considered to be similar devices: the PK profile for Jext was statistically bioequivalent to Emerade ( needle (25 mm) (Figure 3). 21 Thus, needle length alone does not appear to impact on higher plasma adrenaline concentrations achieved after administration.

| IMPAC T OF DE VI CE
One possible explanation for these data is the possibility that the device mechanism is a factor in terms of PK profile. Some AAI have a springbased system (which may be under high or low tension) while others are cartridge-based (which in turn may be spring-based or gas-powered).

The manufacturers of Emerade undertook a comparison of Emerade,
Epipen and Jext (all at 0.3 mg). 27 Emerade is a (high tension) syringebased device. Epipen, a cartridge-base device, resulted in a significantly higher Cmax and quicker time-to-peak concentration compared to Emerade. 27  Jext, and about 60 min for Emerade. The difference was most apparent in those with higher STMD (Table 3). While there has not been a direct comparison with Anapen, Duvauchelle et al. 21

demonstrated that
Anapen (also a syringe-based device) results in a higher Cmax than manual injection of an equivalent dose by needle/syringe ( Figure 3).
Somewhat curiously, while the PK profile for Epipen was very consistent in both Worm et al. 22  Tmax for adrenaline given (first-generation) Epipen compared with needle/syringe. 18 In the study by Worm et al. 22 participants in the low STMD cohort had IM injection using a narrower-bore needle (which might impact on PK profile); however, the needles used in the other participants (22G or 23G) were similar to that in the Epipen (22G); therefore, this does not explain the delay in peak adrenaline seen with needle/syringe. One potential explanation is that the analysis by Worm et al failed to consider the biphasic profile of absorption usually seen with adrenaline in these types of studies (i.e., reflected in the high coefficient of variation for Tmax reported for Epipen of 102%).
In summary, differences in PK profile are apparent between different AAI, which are more obvious in individuals with increasing STMD. Adrenaline absorption seems to be most delayed by an increasing STMD for Emerade compared to Epipen, with Jext being intermediate. Unfortunately, the current data do not allow an assessment of the relative roles of needle length versus device mechanism, in understanding why Anapen and Epipen both seem to cause a rapid peak in adrenaline absorption while that seen for Emerade is more delayed.

| LO C ATI ON OF INJEC TI ON
Simons et al. 18 compared the PK profiles achieved with firstgeneration Epipen and adrenaline via needle and syringe at different locations. Surprisingly, IM injection into the deltoid muscle in the arm was not only lower than into the mid-thigh, it was apparently almost absent, with minimal absorption which seems counter-intuitive given the use of IM injection into the deltoid as Cohorts STMD <15 mm STMD ≥15 mm-≤20mm STMD >20 mm the preferred injection site for many medications (Table 1). 18 The authors do not give any explanation for this observation, and the data are challenging to interpret.
A number of studies have compared injection at the mid-versus distal anterolateral thigh. For Epipen, administration at the distal thigh gave a slightly lower peak plasma concentration than at the mid-thigh, although the difference did not imply lack of bioequivalence (0.41 vs. 0.52 ng/ml; geometric mean ratio, 0.77; 90% CI 0.63%-0.94%). 22 The time-to-peak concentration was also slightly longer (25 vs. 20 min). 22 Duvauchelle et al. 21 reported no significant difference in PK profile between mid-thigh versus distal injection with Anapen (Table 1, Figure 3). Thus, no clear conclusion can be drawn as to the most optimal site of injection in the thigh, although distal injection in individuals with a lower STMD may risk an intraosseous injection.

| IMPAC T OF DOS E
There are data comparing different dosages for the first-generation Epipen and the Emerade. In a very small study of 12 children randomized to a 0.15 mg or 0.3 mg administered with the first-generation Epipen, there was no difference in plasma adrenaline concentration between the two groups (Table 1), although more adverse events were reported with the higher dose. 19 Two studies compared Emerade 0.3 mg versus 0.5 mg in a crossover study design, one in teenagers 23 and the other in adults. 27 In both studies, the 0.5 mg dose achieved a far higher Cmax and AUC compared with 0.3 mg (Table 1) with time-to-peak Cmax around 5 min for both. However, while a higher adrenaline dose results in higher Cmax, we do not know the optimal adrenaline concentration required to treat anaphylaxis.

| Summary
We now have access to pharmacokinetic data for all the currently available AAI. These data demonstrate that it takes at least 5-10 min to achieve early peak plasma concentrations for most devices. The AAI device seems to be the most important factors in terms of pharmacokinetics, with different adrenaline plasma concentrations seen with different devices. The degree to which this is dependent on needle length is challenging to unpick. Another potentially important factor is the force and speed at which adrenaline is injected. This varies by device, with forces likely to be higher for cartridge-based (Epipen, Jext) compared to Anapen, a syringe-based system which uses a lower-tension spring. 34 It is unclear how the force of injection for Emerade (a syringe-based device with a high-tension spring) compares to cartridge-based devices. The available PK data suggest that absorption of adrenaline following percutaneous injection into the mid-thigh in patients with higher STMD is delayed with some autoinjectors, with a lower initial peak concentration and a longer time-to-maximum concentration. This does not appear to be related to the depth of injection.

| Limitations of current pharmacokinetic data
Although PK data are available, the most helpful study comparing three different devices is unpublished. 27 There is large intra-and intersubject variability in PK outcomes, which reinforce the importance of using a crossover design to minimize this issue. The published studies are also relatively small and so have limited statistical power for comparisons between subgroups; many do not include appropriate statistical assessments for bioequivalence. Many studies evaluated the PK profile on the basis of uniphasic rather than biphasic absorption kinetics for IM adrenaline: this further confounds analysis and can be potentially misleading (e.g., the assertion by Worm et al that a manual IM injection of adrenaline takes almost 60 min to reach peak absorption 22 ). Finally, most have enrolled healthy adults who are not at increased risk of anaphylaxis. This means that we have to assume that absorption of adrenaline is similar in individuals who are at risk of anaphylaxis (which is likely) or during an acute anaphylaxis episodethis may not be a valid assumption, since blood supply to muscle and subcutaneous tissues may be different during acute reactions.

| Implications for managing anaphylaxis and conclusions
We recommend the IM route as there is an absence of data demonstrating that subcutaneous injection is at least bioequivalent to IM injection. Given that the first peak in adrenaline absorption occurs at least 5-10 min after AAI administration and potentially much longer in individuals with high STMDs, it does not make sense to repeat a dose until a reassessment of the patient after 5-10 min. This, however, assumes that an adequate dose was given in the first place: according to all international guidelines, the recommended dose of IM adrenaline in older children and adults is 0.5 mg; thus, a dose of 0.3 mg given by AAI may be inadequate and should be repeated after 5-10 min in the absence of resolution of symptoms.
Arguably, the biggest limitation is that plasma adrenaline is used as a surrogate for treatment response: we do not know the optimal plasma adrenaline range we should be targeting to treat an episode of anaphylaxis. To address this, data are needed from pharmacodynamic studies undertaken on patients during anaphylaxis, to assess how treatment response relates to plasma adrenaline. 35 It may be that the optimal dose range varies according to the clinical status of the patient, with higher concentrations needed in severe anaphylaxis. 4 Further work is needed to better understand the relative importance of needle length versus device mechanism/force of injection on PK outcomes, and the mechanism by which patients with a higher STMD appear to have a lower and more delayed peak in plasma adrenaline. This is an important knowledge gap, which may be helpful in improving the design of the next generation of adrenaline autoinjectors.