Inbuilt quality
18 Aug 2005
Driven by an initiative from the US FDA, the pharmaceutical industry is beginning to adopt process analytical technologies. Stuart Nathan looks at how sensors and spectroscopy are being used to scale up reactions from laboratory to batch scale.
Process Analytical Technologies (PAT) bring with them the promise of making quality testing of pharmaceuticals unneccesary. A process using PAT should always produce a product which meets the quality criteria. This is because all the sensors and devices used to ensure quality are incorporated into the process, and the data they produce are used to control it.
The US Food and Drug Administration (FDA) instituted its PAT Initiative in 2002, aiming to speed up drug development, improve quality assurance and, by allowing researchers to monitor and understand the physical and chemical changes within their processes more clearly, boost innovation in both research and process development.
For the manufacturers, the benefits could include reduced operating costs; reduction of defective batches and the time taken for testing them; and reduced inventories of batches waiting to be tested.
Online, in-line or at-line analysis techniques are at the heart of PAT. On-line monitoring is, of course, not new. A routine part of the petrochemical and refining industries, it is even common in household appliances to monitor programmes. Dishwashers and washing machines use online sensors to change the water temperature or rinse cycles to conserve resources, for example.
‘If the dishwasher can do it, why can’t our pharmaceutical reactor?’ asked
There are several answers to this, he said: implementing PAT requires a change in thinking about how to monitor, control, and understand pharmaceutical processes, and the new sensors and the technology needed to run them and handle the data they produce can be expensive. Scale-up is an issue — how do you know whether data gathered at small scale will translate to the large batches produced in pharmaceutical manufacture? And over and above all these, Lane said, is the question of regulation compliance. ‘Can we implement it? And do we trust it?’
There are a variety of techniques available for telling operators what is going on inside their plants, with the choice of equipment largely depending on what it is being studied. Inside the reactor, the focus may be on which compounds are being formed; in this case, spectroscopy is likely to be important, with near-infrared (NIR) giving valuable data about the chemistry of the reaction mixture. Elsewhere in the process, equipment that can monitor how, where and how fast crystals are forming will also become vital for companies bringing in PAT.
Spectroscopy’s role
The array of technologies that can be brought to bear is large. IR, NIR, Raman, fluorescence, mass and UV spectroscopy (using a device called an attenuated reflectance measurement probe which measures UV spectra in situ) can all play a part, as can high-performance liquid chromatography and gas chromatography.
For crystallisation, laser diffraction, the related focused-beam reflectance measurement (FBRM), and the microscopy-based particle vision and measurement (PVM) are increasingly being used to monitor the growth and morphology of crystals.
Lane explained that Bristol-Myers Squibb is now using PAT throughout its R&D process, with laboratory scale reactors (100ml to 20litre) up to kilolab and pilot-plant all of which can be equipped with different types of sensor. ‘Experimental time is the most precious in pharmaceutical R&D. Scientific ideas are turned into experiments within hours,’ Lane said.
In the lab, PAT-equipped reactors allow researchers to develop a rough idea of how a process might work on scale, and maximises the amount of data that can be gathered while still allowing the scientists to focus on the science.
For example, the lab has recently used Fourier transform (FT) IR to follow the progress of a Grignard reaction — where an organo-magnesium species is generated and reacted to form carbon-carbon bonds — as part of a scale-up from bench scale to multi-kilogramme. The reaction is exothermic, can generate hydrogen when quenched, and uses highly flammable materials, so PAT is a useful way of minimising risk, as it requires no sampling.
A matter of scale
The FTIR allows researchers to detect the initiation of the Grignard reaction, where the organic aryl halide reagent reacts with the magnesium to form the highly-reactive organometallic species, and the C-C bond forming coupling reaction.
On a 1kg scale, the results helped characterise the exothermic nature of the reaction, showed how it could be controlled by an initial small addition of aryl halide to initiate the reaction and confirmed that a recirculation loop could be used to monitor the reaction.
When the reaction was scaled up to 80kg, the sensor helped confirm that initiation of the Grignard reagent had occured, and that halide levels were below the point where a runaway reaction could occur.
The Lasentec FBRM, used by AstraZeneca’s researchers, counts and measures crystals by detecting their interaction with a rotating laser.
Simon Black of AstraZeneca explained how FBRM can help answer the crucial questions about crystallisation — when it starts, when it ends, and what processes take place as the crystals form. FBRM gives researchers a fingerprint of particle size distribution by detecting how crystals interact with a circle of laser light, with each ‘count’ showing how many crystals are in the path of the laser, and how much of each crystal intersects with the beam.
This helps them decide which is the best method of crystallising a product: by slow cooling, fast cooling, seeding the saturated solution with small crystals, or by ‘oiling out’, where crystals are formed at the interface of two immiscible liquids.
In the long term, PAT is likely to change the way the pharma industry operates; and moreover, it will be a boon to manufacturers of the sensors needed for the techniques.
According to market research from the ARC Advisory Group, online process IR systems (PIRS) are likely to receive the biggest boost, growing at 4.1% per year for the next five years, with a value exceeding $300million by 2009. However, good design of the systems will be crucial, says field systems analyst Paula Hollywood.
‘PIRS must be user-friendly in terms of installation, operation, validation and maintenance.’
