Removing the Saponin and increasing the production capacity by 40%.

This work was presented as the most challenging work performed so far and the most fantastic opportunity. We removed the saponin and increased the production for this client by more than 40% – a genuine crisis that became an opportunity.

During a recent client engagement, the client and I were taken aback by the new legislation passed by the country, which mandates the complete removal of saponin from vaccine products. The official reasoning behind this decision is that saponin interferes with ELISA testing, rendering effective vaccines ineffective. However, this position overlooks the technical nuances of the saponin mechanism of action and immunology. It is important to note that saponin alone cannot produce robust and efficacious immune reactions. A good antigen is required to trigger the immune response. Regrettably, such episodes are not uncommon in countries with immature regulatory status. Nonetheless, they offer the opportunity to adjust products to meet new regulatory demands and maybe increase the manufacture capacities.

Just to put in context the manufacture has a trivalent vaccine with 3 inactivated virus (O1 Campos, A24 Cruzeiro and C3 Indaial) and the vaccine is an oil in water emulsion with a 5 ml dose for cattle and buffaloes, and it’s potency measure via ELISA (polyclonal). 

Considering compliance concerns, the client promptly discontinued the use of saponin. However, given the nine-month lead time between production and field deployment, we were apprehensive about potential issues in the field and the likelihood of vaccine rejections. To determine the impact of saponin removal and devise a course of action, we conducted a rigorous evaluation. We employed a series of statistical analyses, utilizing the Miitab 15 software and way ANOVA to compare batches with and without saponin ( the raw data will not be insert here).

We also correlated the vaccine potency with he antigenic mass content of each vaccine batch. Just a remark for this particular product the best way to evaluate the antigen content is using the quantitative sucrose density gradient centrifugation (SDG) technique developed by Barteling and Meloen (1975); this technique consist of a sucrose density gradient ultracentrifugation is carried out to virus concentration liquid to be detected, and then OD259nm value of all fractions is continuously detected by an ultraviolet light spectrophotometer; next, peak area of OD value of a sample to be detected is calculated, and the 146S content in the virus liquid is calculated out according to the formula based on their sedimentation in sucrose density gradients. (Svedberg – Wikipedia).

Below we correlated the antigenic mass and potency with the following results:

Correlations: µg/dose O1; EPP O1

Pearson correlation of µg/dose O1 and EPP O1 = 0,125

P-Value = 0,276

Correlations: µg/dose A24; EPP A24

Pearson correlation of µg/dose A24 and EPP A24 = 0,239

P-Value = 0,035

Correlations: µg/dose C3; EPP C3

Pearson correlation of µg/dose C3 and EPP C3 = 0,156

P-Value = 0,173

After the statistical studies we had the following conclusions:

• The use of saponin in the FMD vaccine has significant effect enhancing the potency for at least two strains (O1 and C3).
• The A24 strain seems to be less affected by the saponin and for rason it is the strain with the highest opportunity for payload adjustment.
• When we tried to study the correlation between the antigenic mass and the potency only the A24 strain demonstrated to have some correlation. Nevertheless, the correlation value was small (0,239). This fact added to the issue of not find correlation for O1 and C3 may indicate that the saponin effect seems to be independent to the payload, in other words it cannot be replaced just by the addition of more antigen payload neither means that a payload reduction (at least between certain boundaries) will have a direct impact lowering the potency.
• Saponin, a natural glycoside found in plants, has demonstrated its efficacy as a stabilizing agent in vaccine production. Its use in vaccine formulation has been found to produce more reliable and predictable potency results. The addition of saponin to vaccines results in a more constant and consistent immune response, thereby reducing immune variability. By adding a consistent amount of saponin in each vaccine dose, the immune system appears to react in a uniform and predictable way. This leads to a consistent immune response and is confirmed by testing the vaccine with an ELISA assay using monoclonal antibodies. The use of saponin in vaccine formulation is therefore considered to be an effective approach for producing vaccines with more reliable and predictable outcomes.

Despite conducting a thorough analysis, we were unable to arrive at a definitive conclusion regarding the removal of saponin and its possible effect in the commercial vaccines. The stabilizing property of saponin is too crucial to ignore. Additionally, considering the cost-benefit of the adjuvant and the potency, it seems prudent to retain it in the formula.

We have uncovered some fascinating information that could help us make a critical decision. Our team dug deeper and unearthed some significant variations that have been negatively impacting vaccine production. Our client has been using a pre-formulated culture medium bought from two suppliers, X and Z. One of the key components of the medium, the Triptose Phosphate Broth (TPB), accounts for 40% of the media composition for BHK cells.
But hold on to your seats because this is where it gets thrilling! Earlier this year, supplier X switched the brand of TPB from A to B, without notifying the client. The company policy was to work with two suppliers alternatively, so the client did not notice any changes in the process. However, later in the year, supplier Z did the same thing, changing the TPB brand from A to B, and without prior notice! The reason? To save costs, of course! This change resulted in a significant drop of 20-30% in yields for our client. This is a massive problem because not only do we have to remove the saponin, but we also have a lower virus payload in the vaccine. The combination of these factors could have resulted in a “Katastrophe” and we need to investigate this immediately to prevent any further losses.
To gain a deeper understanding of the issue, we decided to visit the manufacturing operators and discuss the current state of the cells. Upon inquiry, we were informed that there seemed to be both “good” and “bad” batches of cell growth, with an increasing amount of variability. We conducted a correlation analysis between the cell count and internal controls to investigate the matter further. Subsequently, we identified some medium batches categorized as “good” or “bad.” These selected batches were then sent for amino acid analyses to establish a relationship between the amino acid pattern and the promotion of cell growth. Our ultimate goal is to find a solution to the problem that would enable the better development of BHK cells.

Method for Media Analysis 

At QCP (quality control pharmaceuticals) a method was available for semi-quantitative analysis of amino acids by HPLC. All available in-house standards were included in the analysis. Each individual amino concentration can be directly related to the concentration in the medium. However, between amino acids the concentration can differ because of sensitivity differences.

The “Unscrambler” (the Unscrambler) was used to perform multivariate analysis and look at the influence of amino acids on cell growth promotion of the different batches. All amino acids were given equal importance by weighing the data (1/sdev). The calculation method used was Principal Component analysis (PCA). This method looks at differences between samples (=clustering). Results are presented in plots as “scores” and “loadings”. In the score plot samples that are similar cluster together. By comparing scores and loadings we can couple the amino acids to the different clusters.

Results

The chromatogram resulted in 22 peaks of known standards, with retention times varying between 2.0 and 40.9 minutes.  The peaks of L-Alanin, L-Threonin, L-Glycin and trans-hydroxy-Prolin could not be used because of an interfering (disturbing) peak.

The data of the HPLC analysis are presented in Table 1. Values are expressed as peak area. Samples were run in single on HPLC, all samples during the same run.

Figure 1 shows that the PCA analysis resulted in 2 different clusters that could be related to the sample groups (bad, good). Two principal components (= PC, a PC is a combination of important x-values (i.e. amino acids) that are related to variation in the data) were used to explain the data. The score plot is a graph of PC1 (x-axis) versus PC2 (y-axis). Samples that are almost equal cluster together in a score plot.

It is clear that the “bad” media are in the right area of the plot and the “good” media on the left. I.e. the cell growth performance of the media is mainly explained by PC1. It is also clear that there is difference between the medium A and B.

Looking at PC1 in Figure 1 (spreading along x-axis), the bad cell promotion properties of the medium are mainly caused by excess of amino acids to the right (L-Phenylalanin, L-Methionin, L-Cystin, L-Lysin, L-Tyrosin and L-Tryptophan), and lack of amino acids to the left (L-Valin, L-Serin, L-Prolin, L-Asparagin, L-Aspartic).

Recommendation

The PCA is a powerful tool to look at the amino acid data from HPLC analyses. During HPLC analyses and PCA analyses the performance of the media is NOT KNOWN and therefore, this information does not influence the analyses.

After cluster analyses, samples are grouped by color by cell growth promotion as indicated by the manufactured. It then appeared that the clusters based on the amino acid’s differences were strongly related to cell growth performance. 

Questions can be asked to the supplier if they can trace the root cause of this difference in amino acid pattern, hence a larger dataset of “good” and “bad” media could be useful to screen future batches of MEM. Thus, we might be able to predict the cell growth performance on beforehand.

The impact of that change was a decrease in antigen production yields by approximately 23% for the three strains. Using analysis of variance (ANOVA) and control charts we evaluate the antigenic mass yield of the three strains (O1, A24 and C3) for the batches produced during the years (all the data was analyzed using Minitab 15 software), with the highlighted conclusions:

Conclusions:

• The use of different brands of TPB has significant effect for the three strains (O1, A24 and C3).
• The TPB Media B demonstrated the lowest yield in antigenic mass for three strains (O1, A24 and C3), around 23% less than the yields obtained before.
• The TPB A has the best performance increasing the antigenic mass yields 44,7% for O1; 55,3% for A24 and 67,7% for C3.

Finally we correlated the Media composition, Saponin and Antigenic mass effect all combined. 

Background information:

The client has not antigenic mass method used for antigen quantification validated which means that the results obtained presented non-controlled variations, however in order to avoid failures in potency test a conservative and empiric formula based on volume was established (8mL of O1 Campos, 11mL of A24 Cruzeiro and 4mL of C3 Indaial). The formula was defined considering the volume of inactivated antigen prior to the concentration process.

Strain	Antigenic Mass per dose
O1 Campos	8,80 µg/dose
A24 cruzeiro	6,27 µg/dose
C3 indaial	3,04 µg/dose

Later the year the antigenic mass method was validated and more reliable results were obtained. Several experiments were conducted to establish an optimum payload.  

Results of 19 ml vaccines with Saponin (PPE%)
Virus	Vaccine antigenic Content	Vaccine number	308	312	318	310	314	316
Virus O	6,6	99,54	99,60	99,56	99,42	98,68	99,42	98,68
Virus A	9,1	99,79	99,71	99,95	98,54	99,42	98,54	99,42
Virus C	3,3	99,05	99,17	98,64	97,01	96,62	97,01	96,62

Vaccine Number	ml/dose	311	315	309	313
Virus O	5,2	99,51	98,74	99,60	98,39
Virus A	7,2	99,98	98,38	99,80	99,50
Virus C	2,6	99,10	97,53	97,21	99,20

As per the regulatory guidelines, the use of Saponin in the vaccine has been prohibited. Therefore, new studies must be conducted to ensure the vaccine payload is adequately compensated. However, despite the challenges of poor media quality, removal of Saponin, and reduction of payload, we were able to observe the vaccine’s effectiveness. This has provided us with valuable data to make informed decisions, such as reducing the payload and removing Saponin without any negative impact on the vaccine’s efficacy. To gather more information, we analyzed 21 vaccine batches produced with “bad TPB” and low antigenic mass. The data gathered from this analysis will aid us in making further improvements to the vaccine.

To make sure we had the complete security in the process we conducted a test to challenge the amount of payload considering the worst conditions (without saponin and with bad TPB) we conducted 3 experimental vaccines in those conditions and 3 more with a reduced payload passed with flying colors, as you can see below:

Results in (PPE%)
Vaccine Number	Vaccine 1	Vaccine 2	Vaccine 3
Strain O1 (3mL/dose)	92,76	96,91	96,91
Strain A24 (2mL/dose)	90,58	91,96	91,96
Strain C3 (2mL/dose)	90,18	94,03	94,03

		009		0010		011		012
Vaccine Number	ml/dose	PPE%	ug/dose	PPE%	ug/dose	PPE%	ug/dose	PPE%	ug/dose
Strain O1	7	97,56	11,37	99,48	10,26	98,11	12,93	98,22	11,62
Strain A24	7	97,33	9,48	99,57	10,02	99,74	9,69	99,69	10,53
Strain C3	4	94,59	7,50	94,48	4,56	95,81	9,74	91,90	6,44

Conclusion: The four batches have passed the potency test with more than 90% (cut-off 80%).

We conclude that in spite of working without saponin and without the best TPB quality, the FMD vaccine has enough performance to pass the potency test (PPE% > 80%) even with a 25% and 50% of the industrial vaccine payload (6mL and 18 ml).

Our team conducted a comprehensive set of experiments and data analysis to assess the safety of removing saponin from our vaccine. The results of these experiments gave us the confidence to make the change. However, we also recognized the need to strengthen the use of a correct TPB for the media production process. The vaccine manufacturing industry is undoubtedly challenging and complex. Nonetheless, through our unwavering dedication and meticulous attention to detail, we are committed to producing a vaccine that is safe and highly effective.

After an extensive investigation, we determined that the client had overloaded the vaccine with too much antigen. Despite poor quality TPB and saponin removal, we have discovered that the vaccine can still be potent and efficient. Therefore, we have started brainstorming ways to reduce the payload and increase the site’s production capacity.

Unfortunately, the site was already working at maximum capacity, and increasing production would require a significant investment. This would typically take a lot of work to approve. However, we have found a solution that will allow us to reduce the initial vaccine payload from 23 ml to 18 ml, which is equivalent to almost 40% more vaccine doses!

We have discussed this with the commercial team, and they have agreed to support the production increase and adjust the forecast for the following year. As a result, preparations have begun, and our team is proud to report that we have increased the site capacity by 40% with minimal investment and low risk. 

We are confident that this improvement will lead to a more efficient and cost-effective vaccine production process, benefiting both the client and the end-users, not to mention our cost savings by increasing the number of doses they became cheaper.