Reunión del Equipo GPR

The majority of non-viral pDNA formulations encounter problems due to their aggregative nature and as yet it is not known to what extent the processing and formulation procedure affects the severity of this aggregation. The previous factorial study outlined in section 6.1 identified the factors having a statistically significant affect on the production and storage of relatively stable pDNA/PLL complexes. A second statistical study detailed in this section was designed to investigate the effect of the same processing and formulation factors on the behaviour of an aggregative formulation with the aim of further facilitating the development o f an optimised process for the production of stable non-viral gene delivery complexes. The six factors previously identified in OFaT experiments were studied along with one extra factor, NaCl concentration. It is well documented the ionic strength of the suspending solution, dictated by the NaCl concentration, has a detrimental impact on the aggregative characteristics of aqueous suspensions of plasmid complexes. The addition o f NaCl as a variable therefore served two purposes, firstly to generate more information about the influence of this factor, and any possible interactions, on the physical stability of pDNA complexes. Secondly, to induce aggregation in order to gain a greater insight into the affect of the other six factors on a naturally aggregating formulation. The seven factors comprise of formulation factors; plasmid DNA concentration and suspending solution pH and NaCl concentration, as well as process factors; tube internal diameter, mixing tube length, mixing flow rate and tube material. The previous statistical study identified three of these factors, more specifically pDNA concentration, pH and mixing flow rate, as having a statistically significant effect on the physical stability of pDNA/PLL complexes. However, at this stage the extent to which each of the seven factors, individually or in combination, would have a significant affect on the physical stability of an aggregative system was unknown. Factorial design was therefore employed to answer the question and to move towards an optimised, controllable and generic process for the production o f non-viral gene delivery complexes.

The non-viral gene delivery systems, prepared under the various treatment conditions designated by the two statistical designs used in this study, were of the same

Chapter 6.____________________________Statistical Study o f Process Variables: Results and Discussion

description as those used in the initial OFaT experiments and the previous statistical study. The complexes formed consisted of an aqueous solution o f plasmid DNA MB 113 condensed with poly-L-lysine 29300 (pDNA/PLL) at a charge ratio of +2.0. The characteristic of interest was the physical stability of the plasmid DNA complexes and to provide the relevant information three responses were measured, namely the particle size and zeta potential immediately after complex formation at time zero and in addition particle size was re-measured after the complexes had undergone a period of thirty minutes storage at 4°C. The two size measurements and the measurement of zeta potential provide a deeper insight into the short-term effect of the independent factors on the physical stability of the plasmid DNA complexes.

6.2.1 Screening Experiment: Seven Factors

The design produced at this stage of investigation was created using a quarter-factorial design and represented a screening experiment. A full factorial design of 128 runs would have created unmanageable and unnecessary complexity. A less complex design was therefore created using a quarter-factorial design and adding twelve centre points in order to improve the accuracy and to create a robust design, the details of which can be seen in table A 10.3 of appendix 10. The final design consisted of 44 runs, which were separated, into two blocks of 22 runs, and the run order randomised. The 44 runs could not be performed on the same day therefore blocking was employed to reduce the possible error that could be caused by the potential heterogeneity o f the conditions on different days. Randomising the order in which the runs were performed provided additional statistical robustness to the experiment by removing any possible bias generated from uncontrollable noise factors.

Once the design was complete the experiment was performed in run order and the relevant results collated for statistical analysis. The statistical analyses and the half normal plot of the effects of the seven factors, generated through the use of Design- Expert 5® (DX5®), examples of which are displayed in table 6.1 and figure 6.1 respectively, are then used to facilitate the identification of the factors which had the most significant effect on the responses. The procedure is explained in more detail in appendix 9 and was performed for each measured response.

Chapter 6. Statistical Study o f Process Variables: Results and Discussion

Effect Graph

C aution: Factor In volved in interaction

1 6 4 8 . 3 - 1 3 8 3 . 6 2 - 1 1 8 . 9 3 - 8 5 4 . 2 5 - o 3 5 8 9 . 5 6 7 - 3 2 4 . 8 8 3 - 6 0 . 2 G G + [N aC l] (m M )

Figure 6.25. Ejfects graph fo r the effect o f fa cto r G, NaCl concentration, on the particle size o f pDNA/FLL complexes immediately after preparation at time zero (t=0).

F o llo w in g an aly sis o f each o f the three re sp o n se s the re su lts o f the screen in g e x p e rim e n t can be d isp lay ed as effects o r in tera ctio n plots. F ig u re 6.25 show s the e ffec ts p lo t fo r fa cto r G , that is N aC l c o n c en tratio n , an d d isp lay s the in flu en c e o f that fa c to r on the p article size o f the D N A c o m p le x e s im m e d ia tely afte r m ix in g at tim e zero (t=0). F a c to r G w as fo u n d to have the h ig h est level o f in flu en c e, o f all the factors and in tera ctio n s, on the p article size o f th e p D N A c o m p le x e s at this stage. T he co m p lex e s fo rm e d at the low level o f fa cto r G are sig n ific a n tly sm a lle r than those cre a te d at the high level o f fa c to r G. T he d iffe ren c e in p a rtic le size b etw een the tw o levels is a p p ro x im a te ly 900nm and rep resen ts a statistica lly sig n ifican t d iffe ren c e.

Chapter 6. Statistical Study o f Process Variables: Results and Discussion Interaction Graph 1 6 4 8 . 3 - 1 3 6 9 . : G + ? 1 0 9 1 . 3 3 - 8 1 2 . 8 4 2 - G+_ 3 5 3 4 . 3 5 6 - 2 5 5 . 8 6 9 - - 2 2 . 6 1 6 7 - A - A + Interaction o f A :[p D N A ] (u g/m l) and G :[N aC l] (m M )

Figure 6.26. Interaction graph showing the effect o f an interaction between factors A, pD NA concentration and G, NaCl concentration on the particle size o f pDNA/FLL complexes immediately after preparation at time zero (t=0).

A t tim e zero the interactio n betw een the facto rs A , p D N A co n c en tratio n , an d G , N aCl co n c e n tra tio n , w as fo u n d to be the th ird m o st influ en tial fa c to r afte r the effect o f G and A alone. T h e in teractio n clearly affects the initial size o f the c o m p le x e s particu larly w hen fa cto r G is at its h ig h est level (F ig u re 6.26). T h e p a rtic le size o f co m p lex es fo rm e d at the high level o f fa cto r G are sm allest w hen fa c to r A is at its low level and b ec o m e larg e r as the level o f fa cto r A is in cre ased to its h ig h est level. T h e differen ce b etw een the p artic le size o f co m p lex e s fo rm e d at the low and high level o f fa c to r A w hen at the high level o f fa cto r G w as a p p ro x im a te ly TOOnm.

Chapter 6. Statistical Study o f Process Variables: Results and Discussion

3 1 0 6 . 4 -

E ffe c t G ra p h

Caution: Factor in v o lv e d in interaction

2 5 9 7 . 7 2 2 0 8 9 . 0 3 . 5 1 5 8 0 . 3 5 r 1 0 7 1 . 6 7 5 6 2 . 9 8 3 A - A + [p D N A ] (ug/m l)

Figure 6.27. Effect graph showing the influence o f fa cto r A, pDNA concentration, on the particle size o f pDNA/FLL complexes thirty minutes after preparation (t=30mins) with storage at 4 °C.

A lth o u g h facto rs A and G and the interactio n o f A and G h ad the m o st significant im p act on the p article size o f the co m p lex e s tw o o th er facto rs, B and E an d tw o o th er tw o -fa c to r in tera ctio n s betw een facto rs A and B and B and G w ere also id en tified as h av in g a statistica lly sig n ifican t im p act on the p article size at tim e zero.

A fte r a p e rio d o f 3 0 m in tu e s (t= 3 0 m in s), w ith storage at 4 °C , tw o m ain facto rs and one tw o -fa c to r in tera ctio n w ere id en tified as h av in g the m o st sig n ifican t im p act on the p artic le size o f the p D N A /P L L co m p lex es. T h e p D N A co n c e n tra tio n , th at is facto r A, w as, o n ce again , fo u n d to in flu en ce the p artic le size o f the co m p lex e s.

Chapter 6. Statistical Study o f Process Variables: Results and Discussion

Effect Graph

Caution: Factor in v o lv e d in interaction

3 1 0 6 . 4 - 2 5 9 7 . 7 2 - E ^ 2 0 8 9 . 0 3 - .Ü . 5 1 5 8 0 . 3 5 - o cn 1 0 7 1 . 6 7 - cd Ü 3 < _{5 6 2 . 9 8 3 -} 5 4 . 3 G G + [N aC l] (m M )

Figure 6.28. Ejfect graph showing the influence o f fa cto r G, NaCl concentration, on the particle size o f pDNA/PLL complexes thirty minutes after preparation (t=30mins) with storage at 4 °C.

F ig u re 6.27 show s that co m p lex e s fo rm ed at the low level o f p D N A w ere significan tly sm a lle r than th o se fo rm e d at the high level. T h is effect w as id en tical to th at seen w hen the re su lts o f the initial size m easu rem en ts w ere an a ly se d w ith the ex c ep tio n o f the specific size range. A t tim e zero the size ra n g ed from ap p ro x im a te ly 3 5 0 n m at the low level o f fa c to r A to 60 0 n m at the high level o f fa cto r A. A fte r th irty m in u te s o f storage the size ra n g ed from ap p ro x im a te ly 800nm at the low level o f fa c to r A to llO O nm at the high level o f fa cto r A. T h is in d icates that the overall p artic le size o f the com p lex es has in c re a se d a fte r the thirty m in u tes storage p erio d but the d iffe ren c e b etw een the particle size o f co m p lex e s fo rm ed at the low and high levels o f fa c to r A has rem ain ed u n ch a n g ed at ap p ro x im a te ly 300nm .

T he o th e r m ain fa c to r fo u n d to have a sig n ifican t in flu en ce on the p artic le size o f the co m p lex e s afte r th irty m in u tes o f storage w as fa c to r G , th at is N aC l co n c en tratio n . In ag reem en t w ith the initial m easu rem en t at tim e zero (t= 0) the m ain affect o f N aC l co n c en tratio n w as fo u n d to have the m ost sig n ifican t affect on the p a rtic le size o f the

Chapter 6.____________________________Statistical Study o f Process Variables: Results and Discussion

pDNA/PLL complexes after thirty minutes of storage. Figure 6.28 shows that at the low level of NaCl concentration the size of the complexes are significantly smaller than those prepared at the high level of NaCl concentration, the difference between the two being approximately 1 TOOnm. A comparison of figure 6.25 and figure 6.28 shows that although the size of complexes prepared at the low level of NaCl concentration remains relatively unchanged after thirty minutes the size of complexes prepared at the high level of NaCl concentration has increased significantly. The difference between the particle sizes of complexes prepared at the low and high level of NaCl concentration approximately double from an 800nm difference at time zero to a difference of ITOOnm after thirty minutes of storage at 4°C. The interaction between factor A and G also had a highly significant affect on the particle size of the pDNA/PLL complexes after thirty minutes of storage. The affect was virtually identical to that seen at time zero, shown in figure 6.26, the only difference being the increase in the overall particle size of complexes after thirty minutes storage, particularly at the high level of factor G, NaCl concentration.

The other response measured in this study was the zeta potential and analysis of the results identified two main factors and one two-factor interaction as having a statistically significant affect. The two main factors, factor A, pDNA concentration and factor G, NaCl concentration, had a similar affect on the zeta potential. The complexes formed at the low level of the factors had lower zeta potentials than those formed at the high level of the factors. O f the two, factor A had the most significant affect on the zeta potential. The two-factor interaction between factor A and factor G was also found to have a significant affect on the zeta potential o f the complexes. Analysis of figure 6.29 shows that moving from the low to the high level of factor A results in an increase in the zeta potential and this trend is evident at both high and low levels of factor G. The influence of factor A is, however, more significant at the low level of factor G as the increase in zeta potential moving from the low to high level of factor A is much bigger, an increase of approximately lOmV, than when factor G is at the high level, an increase of approximately 3mV.

Chapter 6. Statistical Study o f Process Variables: Results and Discussion Interaction Graph 3 7 . 7 3 4 . 7 - 3 1 . 7 - G + 2 8 . 7 - 2 5 . 7 - G + 2 2 . 7 - 1 9 . 7 - A- A + Interaction o f A :[p D N A ] (u g /m l) and G :[N aC l] (m M )

Figure 6.29. Interaction graph showing the ejfect o f an interaction between factors A, pDNA concentration and G, NaCl concentration on the zeta potential o f pDNA/PLL complexes.

W ith in the ran g e o f ex p erim en tal co n d itio n s an a ly sis o f v aria n ce (A N O V A ) (d a ta not sho w n ) o f the data from the seven facto r screen in g e x p e rim e n t id e n tifie d m ix in g tube length, p lasm id D N A co n c en tratio n , solution p H and io n ic strength as statistica lly the m ost sig n ifican t factors. T his in fo rm atio n , in c o m b in a tio n w ith p ro c ess kn o w led g e, w as used to g en e rate a re sp o n se surface desig n to stu d y th e im p act o f th ese fa cto rs on the three re sp o n se p aram eters as d etailed in the su b se q u e n t section.

Chapter 6.____________________________Statistical Study o f Process Variables: Results and Discussion

6.2.2 Response Surface Design

The initial screening experiment, based on a fractional factorial design, in combination with process knowledge reduced the number of factors from seven to four. In the subsequent response surface design presented here the values o f the three less influential factors were fixed. The response surface design was generated using the D- optimal design option in Design-Expert 5®. The aim of this design was to gain a deeper insight into the affect of four chosen factors on the physical stability of intrinsically aggregating pDNA/PLL complexes. The four factors identified as warranting further investigation were pDNA concentration, NaCl concentration, pH