Allegro™ Connect bulk fill system: Accurate aliquoting of low to high-viscosity bulk drug substance

Materials

The Cytiva materials used for this study are detailed in Table 1.

Table 1. Equipment used for the filling accuracy study

Methods

This test incorporated all four hardware system iterations, namely, the all-in-one biocontainer bag, all-in-one bottle, tower biocontainer bag, and tower bottle systems, see Figures 2 and 3. There is an option to combine the all-in-one biocontainer bag and bottle system into a ‘hybrid’ version with both sets of distribution valves present. The hybrid system was used for this study, as well as the ‘no filtration’ configuration, that is, no filters were placed in-line. Additionally, a LevMixer™ gen IV system with a 200 L tank was used to hold the viscous drug substance simulants (see Table 2) and a 200 L Allegro™ biocontainer bag was used to hold reverse osmosis water for the 1 cP test.

Fig 2. The Allegro™ Connect bulk fill system hardware elements for each system configuration.

Fig 3. The Allegro™ Connect bulk fill system configurations used during application testing.

Table 2. Product substance mimics used during this study

We installed the bulk fill system hardware and single-use system and operated as per the operating instructions. One new flow kit set was installed for each viscous feed filling assessment to ensure that no dilution of the drug substance occurred between the fill sets. For biocontainer bag fills, the Allegro™ biocontainer bags were placed in RoSS shells (Single Use Support GmbH) with the required additional foam inserts (i.e., for a 60% target fill volume, the foam insert would be sized to 40% of the biocontainer bag volume).

The filling of containers is controlled by the bulk fill system either by a target filling weight, target filling volume (with density input for weight control) or by equal distribution of the feed into a set number of containers. For this study, the filling was controlled by a target weight set point, which for biocontainer bags and bottles was set at weight of water equivalent to 60% and 80% of the nominal container volume, respectively.

The filling speed was controlled in a four-step manner, meaning the speed would reduce in four stages at set container weights. This four-step filling controller varied depending on container type and size. The maximum and minimum speeds (speeds 1 and 4) for each container type and size are listed in Table 3. An additional step that affects the filling process is the recovery of drug substance within the distribution flow kit at the end of each sub-batch (i.e., after all containers connected to the current distribution flow kit are filled). The drug substance remaining in the distribution flow kit is recovered into the final container (lowest point) filled in each sub-batch. This container is named ‘container 1’ in the recipe parameters. Additionally, the hold-up contained in the distribution branches is successively drained into each connected container and subsequently weighed to determine the final filled weight. Therefore, the filling of each container is purposely stopped short of the target, such that after product recovery, the target weight set point is achieved. This was optimized through ‘draining’ offsets entered into the recipe, see Table 3. For instructions and advice on recipe building, please see the operating instructions for the product.

We split filling batches up based on the hardware configuration (e.g., all-in-one biocontainer bag or bottle) and the viscosity of the drug substance. Firstly, the all-in-one biocontainer bag system was used with ~ 1 cP fluid for all five container sizes (1, 2, 5, 10, and 20 L) over five sub-batches. We set up the single-use flow kits by first connecting the secondary multiplier to the primary multiplier port 1. Then we connected a new distribution flow kit to the secondary multiplier ports 1 to 5 after each sub-batch, see Figure 4 for multiplier and distribution flow kit port mapping. The system was then reconfigured for all-in-one bottle filling by changing the weighing platform position, loading the bottle trolley, and removing the used secondary multiplier to connect a new multiplier at the primary multiplier port 2. A new batch was then initiated to fill all five bottle sizes over five further sub-batches, as described above. This was repeated until all container types, sizes and viscosities were tested.

To allow the determination of an offline drug substance filling weight, we established every container’s ‘empty weight’ before the first batch:

• For bottles, we numbered every bottle used—in accordance with the associated branch/valve connection—and weighed.
• We chose five new biocontainer bags of each size (1, 2, 5, 10, and 20 L) at random and the filling line Kleenpak™ sterile disconnectors were disconnected to mimic the biocontainer bag filled state and weighed. Each set of five biocontainer bag weights was then averaged and used as the ‘empty weight’.
• After each sub-batch was complete, we disconnected the containers—we disconnected bottles from their caps and biocontainer bags were disconnected using the Kleenpak™ sterile disconnectors—and all were weighed on an offline balance. By deducting the containers’ empty weights, we obtained an offline measurement of the drug substance weight in each container.

Fig 4. Multiplier and distribution port numbers on the Allegro™ Connect bulk fill system (all-in-one system shown here).

Table 3. Filling parameters used for this study

*Maximum filling speed for ≤ 1.8 m/s linear velocity

Container filling with the bulk fill system is divided into ‘sub-batches’, which represent the number of biocontainer bags or bottles to be filled by one distribution flow kit. For standard distribution flow kits, this is up to six containers for the all-in-one systems and up to 12 or 16 containers for the tower systems. Once a sub-batch is finished the containers are disconnected before the distribution flow kit is changed over and new containers are connected.

The consistency and accuracy of each fill is essential for bulk filling processes; therefore, every fill in each sub-batch must be within the process-specified tolerance. Figures 5 and 6 show the successive filling weights of six 1 L containers connected to the all-in-one system and sixteen 1 L containers connected to the tower system, each filled in one sub-batch. For all four system configurations, every container remained within the acceptance criteria of ± 2% of the weight set point, demonstrating a high degree of accuracy and consistency with the selected sterile filling strategy.

The overall range for biocontainer bags and bottles weight deviation from the setpoint was 22 and 23 g, respectively, which was within the specified tolerance for this study. However, if tighter tolerances were required, the filling controller can be further optimized for a specific drug substance and container size.

All containers were accurately filled within ± 2% of the target weight, irrespective of hardware configuration, therefore minimizing product loss, which could have resulted from under or overfilling.

Fig 5. Biocontainer bags (1 L) filled to 0.6 kg with water (~ 1 cP). We collected the data presented during one sub-batch either with the all-in-one system or the tower system. The filling set point and minimum/maximum fill weights are represented by the green and blue dashed lines, respectively.

Fig 6. Bottles (1 L) filled to 0.8 kg with water (~ 1 cP). Data presented was collected during one sub-batch with either the all-in-one system or the tower system. The filling set point and minimum/maximum filling tolerances are represented by the green and blue dashed lines, respectively.

The number of sub-batches completed in one full batch are determined based on the recipe set-up, by both the single-use flow kits used, and the specified container number. The primary multiplier (four connection ports) is a mandatory flow kit for the Allegro™ Connect bulk fill system. With the primary multiplier alone, up to four distribution flow kits can be connected, therefore, four sub-batches can be completed as shown in Figure 7. However, an optional secondary multiplier (five connection ports) can be connected to each of the four primary multiplier ports to increase the capacity up to 20 sub-batches, if required, see Figure 4 for port mapping. The filling accuracy must be not only consistent within one sub-batch but also between all sub-batches.

Figure 7 shows the filling accuracy of four sub-batches completed for bottles filled with water. The filling accuracy is determined here by the percentage deviation from the target weight set point. For standard batches, the filling set point is unlikely to change. However, the bulk fill system recipe programming allows end users to preset adjustments to the target filling weight throughout the batch. Therefore, for this batch to imitate this scenario, the target weight was changed after each sub-batch, namely from 1.6 kg (2 L bottle) to 4 kg (5 L bottle) to 8 kg (10 L bottle), and finally to 16 kg (20 L bottle).

Figure 7 shows a high degree of filling accuracy over the full batch with average set point deviations of less than 0.38% compared to the target tolerance of 2% over the four sub-batches. When comparing the set point deviation over the four sub-batches, the accuracy is shown to be comparable with ranges observed from sub-batch 1 to sub-batch 4 as follows: 6 g, 17 g, 12 g, and 7 g. Therefore, any perceived higher percentage-based accuracy from sub-batch 1 to sub-batch 4 is due to the proportional increase in filling weight set point.

All four sub-batches showed accurate container filling within ± 0.38% of the target weight confirming the consistency of filling with this automated system over a full batch.

Fig 7. Bottle filling to 80% nominal volume with water (~ 1 cP) over a full batch using the all-in-one system. Sub-batch 1 = 2 L bottle filling, sub-batch 2 = 5 L bottle filling, sub-batch 3 = 10 L bottle filling, and sub-batch 4 = 20 L bottle filling. The ⊕ symbol indicates the average set point deviation and n represents the sample number.

The bulk fill system can be used to fill both biocontainer bags and bottles with nominal volumes ranging from 1 L to 20 L. Figure 8 shows the collated data from both container types at the specified nominal volumes. All data points remained well within the specified tolerance of ± 2% set point deviation with a maximum and minimal deviation of 1.9% (1 L) and 0.15% (20 L). As specified in Section 3.1, this shows comparable results with a standardized filling controller. However this controller can be further optimized for specific processes to achieve even greater accuracy.

The process time needed to accurately fill each container is of high importance when designing a bulk filling process, both for product stability and operator shift scheduling. Table 6 shows the approximate time taken to fill each container type and size. It is worth noting that container filling time is limited by the maximum filling speed, which in this study was limited by a maximum fluid linear velocity of 1.8 m/s. If the drug substance linear velocity is not an end-user concern or larger tubing sizes can be selected through customization, filling speeds could be further increased to reduce filling time.

The bulk fill system accurately fills containers, irrespective of type or size, therefore confirming the flexibility of the system for different bulk filling processes.

Fig 8. Biocontainer bag and bottle filling to 60% and 80% nominal volume, respectively with water (~ 1 cP). Data shown was collected using 1, 2, 5, 10, and 20 L containers with the all-in-one bottle and biocontainer bag systems. The ⊕ symbol indicates the average set point deviation and n represents the sample number.

Table 4. Approximate filling time per container, excluding drain down; biocontainer bags and bottles were filled to 60% and 80% nominal volume, respectively

Bulk drug substance viscosities can range greatly from 1 cP for buffers and media to over 20 cP for high-concentration monoclonal antibody products. The bulk fill system has been verified for use with a large range of drug substance viscosities from 1 to 35 cP. Figure 9 shows the full data set of all fills conducted using the all-in-one biocontainer bag and bottle filling configurations. All data points for both container types remained within the target set point ± 2% for all drug substance viscosities, with an average set point deviation of 0.03% to 0.19% over the 60 fills for each viscous feed studied.

This confirms that the bulk fill system accurately fills containers, irrespective of drug substance viscosity, further confirming the flexibility of the system for a wide range of drug substance filling applications.

Fig 9. Biocontainer bag and bottle filling to 60% and 80% nominal volume, respectively, with water (~ 1 cP) and three mAb simulant feeds at a viscosity of 8, 20, and 35 cP. Data shown was collected using 1, 2, 5, 10, and 20 L containers with the all-in-one bottle and biocontainer bag systems. The ⊕ symbol indicates the average set point deviation, and n represents the sample number.

Offline weight measurements were taken to confirm the bulk fill system measurements were reliable and accurate. This was conducted for every container filled after each sub-batch was completed.

Table 7 details the average deviation from the filling set point, comparing weight measurements from the bulk fill system against measurements from the offline balance. This table shows all filling accuracy data for the minimum and maximum fluid viscosities studied to capture the worst-case scenario for low and high-concentration drug substance. Every fill completed with all four feeds was within the target weight ± 2% (see Fig 9 for 8 cP and 20 cP data) when measured both by the bulk fill system and the offline balance. Additionally, measurements taken from the bulk fill system are deemed comparable to those recorded from the offline balance.

Table 5. Average filling weight measurements from the Allegro™ Connect bulk fill system compared to offline measurements

* n refers to the sample number.
^†1 cP and 35 cP data are shown to cover the worst-case scenario in terms of low-concentration and high-concentration drug substance.