I am a LC practitioner of many years and I occasionally use GC when needed. There has been a lot of interest and activity in the area of LC instrument validation. The accuracy and reproducibility of many instrument functions such flowrate, temperature, wavelength, mobile phase mixing etc have been studied. However, there is almost nothing done with GC.
I would like to hear some opinion on this matter. Why is it that almost nothing has been done to validate GCs?. Is it because of the applications?, Is it because the technology is far superior and advanced?. Please let me know what you think.
By Anonymous on Wednesday, March 20, 2002 - 11:53 am:
We have an Agilent GC performance verified each year by Agilent service engineers because it's used for a pharmaceutical active. Same with 3 HPLC units.
By Benjamin on Wednesday, March 20, 2002 - 01:09 pm:
My question above goes more along the lines that there are no commonly accepted protocols for GC validation. I looked in the literature and there is almost nothing on this subject. There is a good deal published on LC validation.
I am sure Agilent and others can test the performance of their units, but do they test reproducibility of injection?, or accuracy of temperatures?, how about accuracy and precision of pressures?.
Somehow nobody seems to have paid attention to this point.
By Jim Gorum on Wednesday, March 20, 2002 - 08:09 pm:
Besides validating the equipment on installation, validating chromatography equipment is usually wasted effort, because method should incorporate checks to verify the correct performance of equipment.
Regulatory and Quality departments usually impose chromatography equipment validation and try to continue with periodic calibration. (Don't get too excited, I am an RA person.) Usually the departments do not understand why a chromatography method can not work like weighing or measuring length. After all you can weigh a sample at 10 labs and get the same value, can't you? When developers write methods requiring resolution, precision, and accuracy for the analyte rather than retention time, flow, column, etc. the nature of the discussion changes.
Your question relates to these thoughts. Ask the question, 'How much does the validation add to the accuracy and precision of the assay?' If you can't answer a define meaningful value, other discussions mean little.
By Benjamin on Thursday, March 21, 2002 - 06:56 am:
I mostly agree with you. The methods should include some test that indicate acceptable performance. However, in my experience most of the methods I receive from outside sources have very little or nothing like that, or worse, the performance limits included do not have real influence on the quality of results, and sometimes make almost imposible to implement the method.
In my opinion the value of instrument validation comes in two areas, first in the detection of problems prior to critical studies, and second, in the transferability of the methods.
I started this discussion to get opinions on why GC validation is almost unheard of, and in contrast a lot of attention has been paid to LC validation. Thus far I have not heard much on these points.
By Anonymous on Thursday, March 21, 2002 - 08:32 am:
Agilent brings in calibrated test equipment to check GC temperatures, flow rates, etc. They also have brought in a calibrated integrator too. I believe they also bring in a standard column and check repeatability of a specified number of standard injections. If necessary, I could try to sift through our documentation report, which is a large binder for each instrument per year. The cost for this is about $3K per instrument, for preventative maintenance and then performance verification. The engineer comes in first day, sets up, and bakes out overnight, finishes up the next. One year they replaced an EPC module (which had been giving us a little trouble, but we hadn't ourselves identified as bad).
By Anonymous on Thursday, March 21, 2002 - 09:14 am:
I believe the reason that GC validation has been largely ignored is that with the exception of OVI and alcohol testing, GC simply does not get much use in pharmaceutical QC labs (please note that I said "does not GET much use", not "does not "HAVE much use"). Most of the work is on LC, hence the concentrated efforts therein.
By Jim Gorum on Thursday, March 21, 2002 - 08:51 pm:
You're right about FDA regulations affecting the validation of instuments and that is why most pharm applications get extra attention. My point is the attention is misdirected. It does not help the quality of the assays and does not meet the spirit of the regulations. With thousands of assay methods, the FDA can not be on top of each one and individual inspectors might be causious of stopping work they thought useless as that is not their job.
The cure to the problem is not performing validation. Because users can not transfer methods into their lab unless the method is foolproof, does not relieve the problem of bad assays causing their employers money and their customers grief. The same problems of transferring a method occur with each breakdown of the equipment. Starting a new (to the lab or R&D) assay goes much easier with equipment known to be working- the purpose of a validation- but does not alleviate the damage caused by technicians and chemists too unfamiliar with their equipment and chromatography to recognise problems when they happen. Also they need to recognize changes that indicate a problem about to occur, but not affecting the current assay. For example, you would not like to repeat a days work because an aging column with easily sufficient resolution and peak shape had 5% increased or decreased retention time.
I have worked with chemists in some industries who slavish follow methods that know to be defective because they or their management do not believe in methods validation, but believe that any industry standard method must be followed by law. Chemists may not like this state, however, it makes more jobs for the unthinking, lower paid workers, other than that the practice benefits few others.
By Anonymous on Monday, March 25, 2002 - 11:22 am:
Validation of an instrument is usually a function of the industry it is used in. If you look at an EPA GC method there is no validation of the instrument as such, but there are performance tests that must be performed before the instrument may be used for the analysis and continuing QC checks to verify the instrument is performing properly. In my opinion, the validation as it is usually performed (measuring temperatures, pressures, etc.) does not address issues relating to GC performance. For example, you can measure the temperature of a GC oven and determine if it varies from the setpoint, but a non-reproducable temperature ramp will have more effect on quality of analysis than a 1C temperature difference between setpoint and actual measurement.
In my opinion, in most cases a performance based methodology will tell more about the suitability of an instrument for a particular analysis than testing all components to verify they meet a specification. I consider this to be as true for LC as it is for GC.
By mscdak on Monday, April 15, 2002 - 02:00 pm:
What industry do you work for? I hope your not the one making decisions for your company in terms of regulatory compliance. Try repeating clinical trials because you failed to validate/qualify the instrumentation that you are using to ensure quality. Good luck on your audits and your responces for your 483's.
By Anonymous on Tuesday, April 16, 2002 - 06:28 am:
Actually, I happen to be one of the people who has written methods for national organizations, and have run up against the problem of what is meaningful in terms of defining performance, and what is traditional but no longer has much meaning in due to advances in equipment and technology. For example, in the days of packed column GCs it was relatively easy to define an acceptable retention time window for a target analyte as the column id, mesh size of the packing, phase loading, and other column variables could be tightly controlled. In addition, it was simple to measure the flow rate, and since flow rates were high, a small variation in flow, say 0.5 mL variation in a 30 mL/min total flow would have almost no effect. Many analyses were also run isothermally. Under these conditions if temperature accuracy and proper operation of the flow controller were verified the method could easily be transferred from instrument to instrument, and if instruments were tested to meet temperature and flow specifications the methods were easily transferred from one to another.
Now consider a modern GC with electronic flow control using capillary columns running a temperature programmed analysis. You can verify all the temperatures and pressures, install a column, set the flow to 1 mL/min, then measure the flow. What you find in the majority of cases is that the measured flow is different than the set flow even though you have verified proper operation of all components. The reason is that the flow is calculated using the nominal column dimensions entered, and these usually differ from actual column dimensions. Most column manufacturers only give the nominal dimensions, so the normal variations give inaccurate results in a "validated" instrument. This also means that you can take two columns with the same nominal dimensions, install them in a properly functioning GC, and get significantly different retention times. The important parameter is retention time stability, not hitting a specific retention time.
Consider also temperature programming of a modern capillary column GC. Almost all manufacturers allow you to enter ramp rates that a GC cannot maintain at a high temperature. You may be able to set a ramp of 40C/min from 60 to 400C, and the GC will maintain that rate up to a certain temperature, say 300C, but from 300 to 400C the oven does not have enough power to maintain that rate. The retention times will be stable if the ramp rate is reproducible even though the rate achieved is not the rate set.
I stand by my statement that performance based testing is more realistic than component testing. I used to work for a small company that made specialized instruments, then tried to test quality into the finished instruments. The company no longer exists. Draw your own conclusions.
Posting is currently disabled in this topic. Contact your discussion moderator for more information.