An introduction to Quality by Design — Transcript

Full transcript

1. 0:00hi i'm malus peters and in this short
2. 0:03video i'm going to give you an
3. 0:04introduction to what quality by design
4. 0:06means and how this is linked to process
5. 0:08analytical technologies so if you're
6. 0:10looking for a career in the
7. 0:12pharmaceutical industry or if you're for
8. 0:14instance studying chemical engineering
9. 0:15these are really important topics so
10. 0:18i'll go over some of the terms that you
11. 0:19might come across
12. 0:21examples will follow in later videos
13. 0:24so let's compare what is the normal
14. 0:26process and how is quality by design
15. 0:29different
16. 0:31so when quality by design means that you
17. 0:32have an inbuilt quality control
18. 0:35if you just look at your normal process
19. 0:37you will see that it works that you have
20. 0:39a quality control step at the end you
21. 0:41produce your product you test whether
22. 0:43the quality is okay and then you release
23. 0:45it
24. 0:46so that means that you have relatively
25. 0:48little understanding of the system
26. 0:50because you don't know so much about the
27. 0:51variables that actually affect the
28. 0:53quality of your product and what
29. 0:55processed power majors are important
30. 0:59now for a normal
31. 1:01production of a product this is probably
32. 1:03okay
33. 1:04because it doesn't matter so much that
34. 1:06there's like a short period where you
35. 1:09have to wait until you release it
36. 1:11and also you can perhaps afford to have
37. 1:14a little bit more variation in your
38. 1:15product
39. 1:16now if you think of it in a
40. 1:18pharmaceutical industry this is actually
41. 1:20quite different
42. 1:22if you would have variations in your
43. 1:23product your finger variations in drugs
44. 1:26in vaccines this would make a very very
45. 1:29big difference so that's why quality by
46. 1:31design gives you that extra additional
47. 1:34insight into your process
48. 1:36so what you do instead you have a
49. 1:37flexible process that can be adapted
50. 1:40rather than something which is rigid and
51. 1:42fixed so this is where i meant that you
52. 1:44have your quality control inbuilt
53. 1:47so what you actually have to do you have
54. 1:49to determine parameters that are
55. 1:50important for your process and you
56. 1:52control these quite tightly within a
57. 1:54design space
58. 1:56so if we look at the terminology and
59. 1:59what qbd means you will see that this is
60. 2:01a systemic approach so you look at the
61. 2:03whole process so it's not just the
62. 2:05process how you kind of produce a
63. 2:07pharmaceutical vineyard reactor it goes
64. 2:09across the whole chain so purification
65. 2:12separation but also packaging and when
66. 2:14you look at tablets from the excipients
67. 2:17that you mix in are equally as important
68. 2:19so the whole distribution chain itself
69. 2:22it is a risk management approach
70. 2:24because what you will do you see here
71. 2:26the difference between the fixed and the
72. 2:27responsive process
73. 2:29um you will have less likely that
74. 2:32there's more variation in your product
75. 2:33so particularly for super vibrant suits
76. 2:35because you can imagine why that's
77. 2:36important and it also gives you a better
78. 2:39and more sound understanding of how the
79. 2:41process actually works
80. 2:43and therefore there are two terms that
81. 2:45you really need to know so you have
82. 2:47critical quality attributes the pca so
83. 2:50what are important attributes that
84. 2:52contribute towards equality and the
85. 2:54critical process parameters
86. 2:56so what parameters of income for
87. 2:58instance the ph the temperature
88. 3:02you can also look at certain ways of how
89. 3:04you separate things what parameters in
90. 3:07your process define the final quality of
91. 3:10the product
92. 3:11so first you will have to figure out how
93. 3:13many of these important parameters you
94. 3:15have so if you go back to i did a video
95. 3:18on modeling as well where we look at
96. 3:20principal component analysis and pls you
97. 3:22will see this is tightly linked to
98. 3:24multivariate analysis because obviously
99. 3:26you can control a certain number of
100. 3:28parameters but you don't ideally you
101. 3:30don't want to control 20. so what you
102. 3:32try and do is really reduce it down to
103. 3:34what are only the essential critical
104. 3:36process parameters
105. 3:38once you know what they are you can
106. 3:40define a certain design space so that
107. 3:43means you set up the specs into what
108. 3:45range you can tolerate it before it
109. 3:47starts to influence the quality
110. 3:49so for certain process parameters that
111. 3:52might be quite a wide range but think of
112. 3:54for instance ph where microorganisms are
113. 3:57extremely sensitive to ph this might
114. 3:59actually be quite narrow so it depends
115. 4:01on the parameter that you're dealing
116. 4:02with
117. 4:03so within the pharmaceutical industry
118. 4:05quality by design is relatively new it
119. 4:08wasn't until 2004 where the fda
120. 4:11recommended the use of process
121. 4:13analytical technologies or pat which
122. 4:16we'll come back to later and how that
123. 4:17feeds into the use of sensors
124. 4:19interquality by design
125. 4:22however the terms are not so new itself
126. 4:24if you look at design space which i said
127. 4:27was like the range that you can have or
128. 4:29the variation you can have in your
129. 4:30process parameters or within your
130. 4:32critical quality attributes this was
131. 4:34already documented in the 1950s
132. 4:37now the person who coined the term and
133. 4:39who published some very influential
134. 4:41papers in the 1990s with joseph duran
135. 4:43and i kind of started roughly around the
136. 4:4670s so you can see there was quite a bit
137. 4:48of delay in actually kind of getting
138. 4:50this to the fda
139. 4:52and but now this is common practice
140. 4:54within the pharmaceutical industry even
141. 4:56though for other industries this
142. 4:58implementation because you can imagine
143. 5:00it takes time
144. 5:02there's also a lot of resources that go
145. 5:03into understanding your process
146. 5:06so they're not implemented as widely as
147. 5:08just within the pharmaceutical industry
148. 5:10now let's say you wanted to implement
149. 5:13qbd principles there's a very specific
150. 5:16workflow that you will need to follow
151. 5:18so the first step that you start off
152. 5:20with is your target product profile
153. 5:22so think of the profile or the
154. 5:24characteristics of your product that is
155. 5:26aimed at particular disease needs to
156. 5:28have
157. 5:29once you have that you can start to have
158. 5:31a think about what
159. 5:33potentially might be critical process
160. 5:35parameters or cqas or other things that
161. 5:38you need to have a look at
162. 5:40but in the other hand you will need to
163. 5:42have done some
164. 5:44preliminary experiments and you will
165. 5:46have used some sensors to actually see
166. 5:49whether you might think that this is a
167. 5:51critical process parameter but it might
168. 5:53not be as important as you think so
169. 5:56first you will need to optimize your
170. 5:58experiment or do some small experiment
171. 6:00and figure out which ones and this is
172. 6:02usually done in smaller scale to define
173. 6:05these critical process parameters
174. 6:09now first of all you need to think about
175. 6:11what variation and product you're able
176. 6:13to tolerate
177. 6:14particularly when you look at
178. 6:16pharmaceutical products i think you can
179. 6:17imagine that this is quite a narrow
180. 6:19range when does it start to become
181. 6:20dangerous
182. 6:22there are also certain diagrams that you
183. 6:24can use you might notice more under a
184. 6:26fishbone diagram an ishikaya diagram
185. 6:29where you start to look at what are the
186. 6:32risks of certain parameters so you can
187. 6:34think of for instance and this kind of
188. 6:36goes wider than beyond your traditional
189. 6:38processes so this is also often used in
190. 6:40the business industry where you also
191. 6:42need to consider the environment that
192. 6:44you're working on not just the process
193. 6:46itself
194. 6:47when you think of your design space you
195. 6:50will based also on your risk assessments
196. 6:53you will think about the variation that
197. 6:54you're allowed to tolerate before it
198. 6:56starts to compromise product quality and
199. 6:58what the potential impact could be for
200. 7:01patients
201. 7:03now here again some of them might be
202. 7:05have like a relatively wide range some
203. 7:07of them are more narrow but you will
204. 7:09need to implement certain sensors in
205. 7:11order to make sure that you stay within
206. 7:13that specific range
207. 7:15now because this process quality by
208. 7:17design it's supposed to be in real time
209. 7:19so you're supposed to get very fast
210. 7:21feedback
211. 7:22so what you want to do is either measure
212. 7:24in c2 which is really complicated for
213. 7:26lots of bioreactor processes
214. 7:28or at least you want to look at in line
215. 7:31monitoring where you get a very quick
216. 7:32result
217. 7:34so there's only a certain number of
218. 7:35sensors that can actually fulfill these
219. 7:37requirements and don't really need
220. 7:38sophisticated lab analysis and we'll
221. 7:41come back to that when we discuss
222. 7:42process analytical technologies
223. 7:46once you know the design space we also
224. 7:48need to think about the control strategy
225. 7:50so how do you make sure that you stay
226. 7:52within that design space
227. 7:54there's a number of very common feedback
228. 7:56loops you probably are most familiar
229. 7:58with pid feedback loops where for
230. 8:00instance if you have a temperature you
231. 8:02set it at a certain range and you stay
232. 8:03within that range but there's also
233. 8:05controls where you have on off
234. 8:07strategies and there's lots of other
235. 8:09strategies that you can implement
236. 8:12finally you have your control strategy
237. 8:14you have a good understanding of your
238. 8:16process which probably would have
239. 8:18happened by doing some preliminary
240. 8:19experiments but after that the key key
241. 8:22thing that comes by quality by design is
242. 8:24that you continuously supposed to
243. 8:26improve it
244. 8:27so as you start to produce more of your
245. 8:29product you will actually gather more
246. 8:30data so you will get more insight into
247. 8:32the process itself
248. 8:34and this is fundamental within qbd that
249. 8:36you start to learn and you train people
250. 8:39and how you can improve your product
251. 8:41cycle
252. 8:43now i work in census and you will see
253. 8:45that this is a very critical aspect of
254. 8:47quality by design and the way this works
255. 8:49is by using process analytical
256. 8:52technologies and this was defined in
257. 8:53that fda document which was released in
258. 8:552004.
259. 8:57now actually what it does you control
260. 8:59your manufacturing process by measuring
261. 9:02these critical processing parameters so
262. 9:04these are the ones that can actually
263. 9:06affect your critical quality attributes
264. 9:08at the end
265. 9:09and remember this was a fast process so
266. 9:11typical sensors that you will have come
267. 9:13across of infrared spectroscopy
268. 9:16near-infrared spectroscopy is very
269. 9:18commonly used for tablets for instance
270. 9:20you can look at rama you can look at
271. 9:22certain type of biosensors
272. 9:25but you are a little bit limited because
273. 9:28it needs to be like a fast sensor and
274. 9:30ideally you want to measure it on site
275. 9:32as well
276. 9:33as you do this and because it is so fast
277. 9:35you will generate big data you will
278. 9:37generate a lot of data so you need to
279. 9:39have operators that are trained in how
280. 9:41you understand that process
281. 9:44so in order to to do that you can either
282. 9:46use principal component analysis which
283. 9:49is a method of condensing your data
284. 9:52or you can use other methods for
285. 9:54instance such as pls in order to make
286. 9:56sense of what's happening and this is
287. 9:58combined with multivariate analysis
288. 10:00techniques you will be able to look at
289. 10:03your data
290. 10:04see what is happening
291. 10:06and then that will feed into your
292. 10:08control strategy so once you control it
293. 10:11you look at your measurement data and in
294. 10:14real time you can make adaptations to
295. 10:15your process such as for instance
296. 10:17changing the ph changing the
297. 10:18temperatures in order to make sure that
298. 10:20your product quality isn't affected
299. 10:24so here i've given a very brief summary
300. 10:26of a quality by designers and how this
301. 10:28is really an inbuilt quality control
302. 10:30which is recommended by the fda for use
303. 10:33in the pharmaceutical industry i've gone
304. 10:35through some very key concepts i think
305. 10:37of for instance critical process
306. 10:39parameters critical quality attributes
307. 10:42target product profile and process
308. 10:44analytical technologies which was
309. 10:46related to sensors and then finally also
310. 10:48your design space which is the space
311. 10:50into which you can operate
312. 10:53so you will see that sensors play a very
313. 10:55fundamental part within quality by
314. 10:57design and they're really necessary to
315. 10:59fully understand your system and by
316. 11:01understanding your system you can start
317. 11:03to continuously improve your production
318. 11:05process
319. 11:06so if you want to know more about this
320. 11:08i'm going to give some case studies
321. 11:10later where you can see how this is
322. 11:12really important to improve the efficacy
323. 11:14of your process but also to guarantee
324. 11:16safety of your products