> -----Original Message-----
> From: Ritter,Nadine [SMTP:Nadine.Ritter@add.ssw.abbott.com]
> Sent: Tuesday, August 25, 1998 5:13 PM
> To: Recipients of ABRF List
> Subject: RE: AAA-GLP
>
> Waters still sells the PicoTag workstation (as of 1997 catalog), which
> is
> what most people that I know use for hydrolysis. I have also seen a
> hydrolysis unit made by Savant that includes a sealed centrifuge rotor
> for
> spinning under high vacuum, but the lab that has it didn't like it
> very much
> and just bought a PicoTag workstation to replace it.
>
> The biggest problem we have ever had with the PicoTag unit is the
> Mininert
> valve caps used to seal the hydrolysis tubes. They have a tendency to
> crack
> and leak, with replacement caps at over $100 EACH to replace. Even if
> they
> test ok under vacuum at the beginning of the hydrolysis, they can leak
> slightly during heating. Waters has been giving us credit and
> replacing
> broken caps if they fail within the first month or so of use, but it
> is still
> frustrating to lose a set of samples due to leaking caps. One
> characteristic
> sign of leaked caps is when you do not hear the "psst" of pressure
> change as
> you open the cap's valve after the tube has cooled. We have not
> investigated
> microwave hydrolysis, which is a method that is an option.
>
> In 1994, Augustine Smith gave a workshop presentation on Amino Acid
> Validation at the ABRF meeting. One of our newsletters from that time
> should
> have a summary in it. Here are some of the issues she covered:
>
> For validation of amino acid analysis, you can run trials of the
> sample at
> various hydrolysis times (like you would for a time-zero study). You
> could
> do different hydrolysis temps, too, if you felt that temperature would
> be a
> critical variable, ie hard to control. For microwave, other variables
> might
> be critical. A plot of the individual AA pmol recoveries vs time (or
> other
> variable) would show you where the stable range is for each AA.
> Comparing
> the plots of all AAs would allow you to choose one timepoint that is
> most
> favorable to the majority of them. You will of course have lower
> recoveries
> of some of them (Ser, Thr, Met) but you should expect that.
>
> You might also want to validate the appropriate sample load by doing a
> linearity study. Test 5-7 levels of the sample, hydrolyzing at the
> optimal
> condition you determined above. Plot the recoveries of each AA and do
> linear regression on each to obtain the coefficient of determination.
> The
> range where r squared is greater than 0.98 or so is the linear range
> for each
> AA. Again, determine the "consensus range" for the majority of AAs
> to yield
> the working linear range for that sample in that buffer.
>
> If you have a troublesome buffer (ie one that contains detergents or
> other
> agents), the effect of the buffer components should also be
> investigated.
> You can do this by spiking in different concentrations of these
> elements then
> comparing the effects on AA recoveries and HPLC chromatographic
> performance. You can then determine how sensitive the analysis will
> be to
> variations in the buffer constituents. Ideally, of course, it should
> be
> pretty rugged to minor fluctuations in composition.
>
> Once you have established the appropriate hydrolysis conditions,
> sample load,
> and buffer effects, you should have the finalized method. You can then
> do
> experiments to determine precision and accuracy of the method. From
> this
> information, you will get data to support the expected performance of
> the
> method for the specific sample you are validating. You should then
> write
> this method as a protocol, then test the final protocol by running
> several
> trials (preferably with separate lots or batches) of the sample, and
> compare
> the results obtained to those you expected. If they meet your
> expectations,
> the method is validated. If they do not, you need to investigate why
> they
> are different. You may need to do some re-development in order to get
> the
> method where you need it. You should not consider the method
> validated until
> and unless it can perform as expected when following the final written
> protocol. Once it is validated, you should not make any further
> changes to
> the protocol unless you evaluate the effects the change(s) might have
> on the
> method's performance.
>
> There are many excellent references on method validation, most
> significantly
> the International Conference on Harmonisation Guidelines for Method
> Validation. There are two parts, both available on the Internet. One
> way to
> find them is to go to www.fda.gov and search the Federal Register for
> "ICH
> Guidelines". There are several guidelines on various topics, but two
> on
> method validation. These are very general, but are the benchmark
> against
> which regulatory agencies are evaluating method validations.
>
> A very easy way to dive into the validation literature is to use an
> Internet
> search engine and query "analytical + test + method + validation". I
> have
> gotten direct links to ICH and FDA documents, as well as websites with
> validation info (like Waters) and original publications.
>
> Also, there have been some terrific articles published on process and
> test
> method issues in the journal BioPharm, a free publication from
> Advanstar
> Press. Call 888-527-7088 for info on back-issues and subscriptions.
>
> I hope this helps. Please feel free to contact me with any other
> questions.
>
> Nadine Ritter
> Abbott Diagnostics Division
> Nadine.Ritter@add.ssw.abbott.com