Break loose force and glide force

Break-loose and glide forces were first analyzed to get an estimation for the necessary forces. Different setups were tested in smaller sample sizes to see which parameters change values on large scale. The small datasets only allow limited comparability. In this sense a t-test reveals a significant difference in break-loose force between the 3 mL (experiment #1) and the 1.5 mL (experiment #4) cartridges (p = 3.05*10− 5). This is not uninteresting as both cartridges are from the same company (Company B) and the two plungers are basically the same plungers just for different cartridge sizes.

The two large datasets of Experiments #5 and #6 would be interesting to compare as their BLF are very similar while the GF differ by almost 25%. Unfortunately, a Shapiro-Wilk test was significant for Experiment #5 such that a t-test cannot be conducted reliably. The BLF force of experiments #5 and #6 does not seem to be much different compared to the experiments #1 - #4. In literature Yoshino et al. [27] documented a change of BLF over storage time for systems containing liquid silicone oil. It seems that the baked-on silicone oil is more resistant to this change. Funke et al. [28] documented an increase for baked-on silicone oil for BLF but clearly showed that there is a difference regarding layer thickness. In the existing experiments a difference in BLF due to storage under pressure cannot be seen, but also was only tested for 2 weeks.

A further comparison between experiments reveals a significant difference (p = 4.96*10− 6) between glide forces of 3 mL and 1.5 mL cartridges. This is conclusive with a significant difference in the BLF between the two datasets, yet it was not tested consecutively. Similarly, there is a significant difference (p = 0.006) between using a test speed of 50 mm/min or 160 mm/min for measurement. Despite the significance in the two cases, the differences are so small that there is no influence for the purpose of this study. However, it should be kept in mind at other places in device development where it might make a difference.

Comparing experiments #7 and #15 we can have a look into a change due to storage under pressure. Indeed, glide forces of both experiments are significantly different (p = 0.012). To further specify the difference, experiments #7 and #13 were tested for a significant difference in a consecutive t-test (with alpha-error correction) to see whether the change is due to pressure. Unfortunately, the test was not significant such that further conclusion cannot be drawn here.

Significant differences between some of the datasets give an interesting insight into factors that contribute to changes. For the purposes of this project the maximum tolerable BLF was set to 15 N which is almost three-fold higher as the highest mean values measured. In case of GF the maximum tolerable GF was set to 10 N being 2-fold to 5-fold higher than mean GF values measured. We therefore conclude that impacts of different primary packaging on BLF and GF are too small to make an impact on the purpose of the project.

Septum pierce force under pressure

The main concerns for the pressurized septum during piercing was that either the septum blasts due to a weakness in the material introduced by the needle cutting into it, or the drug product leaking from the cartridge outside the fluid pathway. In none of the 444 experiments that were conducted either effect was observed. Therefore, we consider the issue as unproblematic. A main issue here may be the non-seat engaging surface which was lowered to an area of 3.14 mm2 and the pressure on the cartridge which held it in its mount.

Table 4 shows an excerpt of the conducted experiments with datasets that are comparable with each other. Experiments #1 and #4 are both under pressure during piercing, they differ only by the 42 days of storage under pressure. Standard deviation in the pierce force is the same; however, the average is 2 N higher. Likewise maximum and minimum values are higher as well. A t-test revealed that there is a significant difference between both datasets with p = 0.041, so pressure clearly makes a difference regarding the pierce force for this type of septum.

When comparing Experiments #2 and #5 a real difference cannot be seen, even though mean and max values are higher in the 42 days experiment, the min value is higher in the 0 days experiment. The t-test we conducted here did not conclusively proof a difference between the two samples.

Also, an interesting point is to compare experiments #4 and #5. The data is very different for both sample sets. A t-test supports the impression, indeed both samples are significantly different with p = 2.667*10− 10. So septa from different companies make a very clear difference. When looking at experiments #3 and #4, which are two different septa from the same company after 42 days of pressure, the values also appear to be different, but a t-test did not deliver a conclusive result on a difference between the two sample sets.

Septum stability

The septum bulge growth curves of all but one experiment are very similar to each other. The exception, experiment #2 (shown in Fig. 3), has a slight decrease of septum growth at day 1 (data point 3). Considering the change of pressure in graphs c), d), f), g) and h) of Figs. 3 and 4 there is always a drop in force on day 1. Interestingly for all the high-pressure experiments (1370–1420 kPa) this drop is larger than 1 N, for the low-pressure experiment #2 it was only recorded to be 0.1 N. The reason why the septum bulge reduced may be due to the low overall pressure in the experiment compared to the other experiments. However, the change in septum bulge on day 1 for #2 was on average − 0.024 mm while the standard deviation was 0.055 mm. The reason for this behavior remains unclear, but it is likely that the decrease is an error due to the optical measurement method.

The general shape of bulge growth has a large change in the beginning and is going into a steady state after about 14 days. If we look at the data points of 32 and 42 days we can see that the difference between those two in all experiments on average is ± 0.01 mm. (see Table 5) This clearly shows that the septa go into a stable state. The extent of the septum bulge is dependent on pressure that is used which can be seen when comparing experiment #2 and #7 (average septum bulge increase after 42 days of 0.29 mm vs. 0.73 mm). If we consider all experiments with pressures between 1370 and 1420 kPa to be comparable we can say that septa of company E on average bulge more (0.68–0.81 mm) than septa of company D (0.52–0.60 mm). When taking septum pierce force into account where company E septa had on average lower pierce forces than company D septa it could be assumed that company E generally uses a softer material.

The force and thus the pressure in the cartridge fluctuate in the first days after the pressurization by less than 2 N on average for all cases. After this initial fluctuation the force goes to a continuous state with ± 0.5 N difference in the first 42 days. The deviation here is likely due to the experimental setup and may originate from friction and slight changes of the angle of the cartridge in the CH. The large standard deviation may be due to the spring itself which comes with tolerance of ± 10 N (manufacturer’s information) from production. In reality, however, the measured spring force deviations are around ± 2 N.

Similar to the septa the plunger position stabilizes within the first 14 days, which can be seen in graph c) and d) of Fig. 3 and graph f), g) and h) of Fig. 4. The plunger movement of roughly 1 mm on the first day can probably be attributed to compression and dissolution of a small gas bubble which resulted from filling. Other factors may be the compression of the rubber parts and the displacement of the septum. Fluid compression (2 mL H2O, bulk modulus of 2.08 GPa, 1370 kPa pressure) may result in only 0.018 mm plunger movement and is thus not relevant.

All septa used in the presented experiments were monolayer septa. Bilayer septa were not used as in very early experiments it was shown that those septa cannot withstand the pressure. They break within a day of pressurization depending on the pressure applied. We presume that two thinner layers of rubber, as in case of the bilayer septa, is much less stable than one thicker layer of rubber, as in case of the monolayer septa. With respect to the purpose of the study this is not an issue for the device that is being developed. Monolayer septa are for single piercing while bilayer septa are made for multiple piercing into the primary container. The device that is developed is a single-use and single-dose device. Furthermore, reducing the non-seating surface of the cartridge septum to 3.14 mm² helped to further stabilize monolayer septa.

Long-term septum stability

The long-term septum stability experiment shows that it is in principle feasible to store the cartridges under 1420 kPa pressure for a longer time. A change in the septum bulge of the last two data points is within their standard deviations, while the curve follows a logarithmic pattern and still grows slightly over time we consider it to be stable for our purpose. The experiment will run until 2 years have passed which is the storage life for common medicines.

Regarding the accelerated experiment a comparison of the graphs i), k) and m) shows that there are no differences between the storage conditions regarding the shape of the curves. The graph for the 1370 kPa always has the highest septum bulge and the graph for 960 kPa is between the two other graphs. The only clearly visible difference is the extent to which the septa bulge. When comparing the slopes displayed in Table 6 it becomes visible that for the samples without pressurization the storage condition seems to have an influence on the septum bulge as well. This, however, is an artifact due to the measurement precision of the used instrument. For the two pressurized setups a steady increase of septum bulge over time is visible, while the amount by which it increases is enhanced by both factors, temperature/humidity and pressure as can be seen in Table 6. The unpressurized samples were conducted as controls. The highest measured septum bulge increase between months 1 and 12 was for a sample of the 1370 kPa under accelerated room temperature conditions. However, we do not expect it to be a problem as the bulge increase is much lower compared to other experiments which were unproblematic.

According to the ICH Q1A (R2) guideline a 6-month experiment under accelerated conditions is the minimum necessary. Yet the experiment was conducted for the double time until the last value was measured. Besides that, the bulged septum did not show any signs of damage. ICH guideline Q1E determines the extrapolation. It is stated that the change pattern in the existing data shall be approximated with a regression function with good fit. Curves of the septum bulge clearly have logarithmic shape in most experiments we conducted, especially due to the behavior in the first 14 days. But the pattern of months 1 to 12 often looks more linear then logarithmic, because they start after the initial phase of septum bulging. A logarithmic fit may thus underestimate the septum bulge in predictions. We therefore decided to fit the curves with linear regression as linear functions always have a stronger growth than logarithmic functions in the predicting area of the curve and when calculated from the same parameters. As explanation: when fitting a logarithmic curve with a linear function the fit function always intersects twice with the logarithmic function. Since the logarithmic function is rightbound and positive, the slope of a linear function must be higher than the logarithmic function and an extrapolation with the linear fit always yields a higher value compared to the logarithmic function for an x-axis value larger than the second intersection of the two curves. In order to rather overestimate than underestimate septum bulge and the fact that most curves have rather linear shape, we decided for this ‘worse case’ approximation. With this estimation, the largest increase of an average septum bulge will be the 1370 kPa sample for room temperature with ~ 0.27 mm (see Y24 value of ACC FRG in Table 6). As we have seen septum bulge changes of double amount in other experiments (e.g., the 500 days experiment) which did not turn out to be a problem, we do not expect a problem here either. We therefore conclude from our experimental data and its analysis that a shelf-life time of 2 years in room temperature or refrigerator will be feasible regarding the septum of the cartridge in the autoinjector device we proposed.

Secondary pressure failure mode

The most important finding of the secondary pressure failure mode experiments is that the lowest pressure at which a 3 mL cartridge broke is 4922 kPa which is more than 3.5 times higher than the highest pressure (1370 kPa) used for 3 mL cartridges in the autoinjector. Regarding the 1.5 mL cartridges the difference was even higher with 9226 kPa being almost 6.5 times higher than the maximum used pressure of 1420 kPa.

Regarding the differences in the cartridges, CB1 and CA1 have very different average breaking points (11,139 ± 3896 kPa vs. 7899 ± 830 kPa) even when having the same plunger (PE3). Unfortunately, the sample set of cartridge B1 with Plunger E3 came out with a significant Shapiro-Wilk test and therefore is not considered to be normally distributed. If we would conduct the t-test nonetheless, cartridge A1 and plunger E3 would have yielded a significant difference which would support the impression the values give. To prove this assumption additional experiments would have to be made. The experiments with CB1 and PD3 have a mean value of 11,375 ± 2743 kPa, which is quite similar to the CB1 + PE3 and different to CA1 + PE3. Especially when looking at the standard deviations of the different sample sets. Standard deviation of Company A cartridges is about 10% of the mean value (CA1 + PE3: ~10.5%, CA2*+PD4*: ~9.7%), standard deviation of company B cartridges is around 30% of the mean value (CB1 + PD3: ~24.1%, CB1 + PE3: ~34.9%). In this sense it seems as if company B cartridges are more resistant to pressure than company A cartridges but also with larger variation. This is especially interesting as pharma grade type I glass cartridges from different manufacturers are all formed from glass pipes they obtain from Schott AG (Mainz, Germany) according to the cartridge’s respective specifications. Therefore, the process of forming cartridges must have a significant influence on the properties of the different cartridges.

Drug product stability

The size exclusion chromatography of Adalimumab drug product determined a monomer concentration of more than 98% in all samples thus satisfying the requirement. Low amounts of high molecular weight (HMW) residues indicate that there are no agglomerates in the drug products. The requirement of less than 2% HMW is satisfied for all samples. Also, low molecular weight residues which could indicate debris can only be found in small amounts. The combined amount of mAb light chains and heavy chains is in all cases above 99% complying with the requirement of at least 95%. The amount of non-glycosylated heavy chains (NGHC) is low. Deglycosylation of the heavy chains is associated with significantly reduced bioactivity [29] and must therefore be avoided. The non-reducing electrophoresis shows IgG to be the main component of the drug product. Subvisible particles was analyzed according to USP 787 and all drug products, independent of pressure duration, comply with the regulation.

Regarding other biologic drugs we assume that other mAbs will exhibit a similar stability under pressure, as they all have a very similar structure. For our pressure experiments we choose explicitly a mAb because this class of biologics are highly sensitive due to their complex compositions. Therefore, we further assume that antibody fragments and other biologic drugs with similar or less complex structures will behave similarly under the pressure conditions we applied in our experiments. The literature on cold pressure denaturation we cited in the introduction shows that mesophilic organisms are viable up to 40 MPa [20] (the max. pressure used in the autoinjector concept is 1.42 MPa), which also gives a clear hint about the potential stability of proteins in the device. Regarding the catalytic activity of enzymes, it may depend on the enzyme. Decaneto et al. [30] studied an enzyme under pressure and found that generally enzymatic activity can be influenced by pressure for different reasons as small changes in the protein structure, hydration, influence on the rate limiting step and also changes of conformation and hydration of the ligand occur; however, we do not believe this to be an issue in our case as hydration and small conformational changes are subject to a change in environment energy levels and are likely reversible when the pressure changes back to normal, and the drug product will be used under normal pressure levels in the body. Regarding RNA and DNA single strands are covalently bonded and should not be affected by pressure. Double stranded DNA molecules are likely unaffected at used pressures [31].

Scalability and manufacturing of a new autoinjector concept

In order to make a new autoinjector concept successful on the market, we consider three main prerequisites. A new autoinjector device must outperform the state of the art with respect to human factors compliance (1) and technical versatility regarding the formulation to be delivered (2), all while its cost must be in scope of devices on the market (3).

Autoinjectors are commonly manufactured in mass production; therefore, the use of standardized primary packaging materials suitable for standard filling lines is of particular importance and was the leitmotif for the choice of components in the tests of this article and the development of the device. Also, the assembly of the autoinjector device from its subassemblies and merging the filled primary container for the final combination product is subject to this necessity, which has been successfully assessed, but exceeds the scope of this article.