LC Method Development Time Limit

[Vlad Orlovsky]

problem with DOTA is not polarity but rather multiple interactions which needs to be addressed. I did not give up on anything. I don't have Bi, Gd or DOTA-Gd complex and I don't have intention to buy them. I requested sample from one of our customer, but don't know when and if they are going to send us a sample. Looking at the literature I don't think that it is as complex as I suspected. Bunch of methods using RP and Silica columns and FL-detectors and I don't see why you can't use one of them.

Regarding the comparison chromatograms, I did not want to show 0% and 5% ACN because some of the chromatograms on RP column look horrible as phase collapse occurred. I can easily present even worse comparison, but I want to be fair to other columns. Our columns are compatible with 100% water and 100% ACN and they allow big mismatch between diluent and mobile phase. The main difference for mixed-mode and RP is presence of additional controllable interaction and ability to retain polar ionizable compounds much much longer than on RP column. It also allows to increase loadability of these ionizable compounds 10-50 folds.

[lmh]

Can I just add a comment? There's a big divide here between those who analyse products that contain a limited number of peaks (and are prepared to develop a method until all peaks are separated), and those who look at biological samples. The latter realistically cannot separate everything in their extract, and will settle for a situation where the peak(s) they are most interested in appear(s) to be separated from "everything else", "to an adequate extent".

For many of us who work with biological extracts, the idea of just 15 peaks is almost laughably simple, and we get a somewhat sour taste from the suggestion that changing to a magic column will suddenly resolve everything in our sample (been there, believed the rep, bought the column, spent the cash... several times...). The problem is that most of the peaks are unknowns, with unknown properties, so changing column is a bit like pressing the "hyperspace" button in that old "Asteroids" game, or ringing your bicycle bell ferociously as you approach a group of pedestrians: everything suddenly moves around, but we have absolutely no idea where anything will end up, and no guarantee that the result will be any clearer than what we had before!

The bottom line is that if you have 15 compounds whose chromatographic properties are unknown, and you run them on a column/method with sufficient resolution to allow separation of 300 peaks, assuming the compounds elute at random, unrelated times, there is still a 50% chance that two will coelute. That's just statistics.

Given that the compounds, although unknown, will often be interrelated or share some properties, the situation is usually much, much worse.

I almost don't believe that hplc is safe except in tandem with high-resolution detectors, or in simple situations where the sample is known to contain only a limited number of things, and the matrix is known never to contain anything that can disturb. And to get to the point: checking that this is the case takes time. It's much easier to "develop a method" in the sense of finding a column and conditions that separate 5 pure compounds than it is to "develop a method" in the sense of checking that in all relevant matrices there will never be anything that disturbs, and that the 5 peaks will always be measured correctly over a useful range of concentrations in all possible matrices.

[Vlad Orlovsky]

Changing to a "magic" column is not going to resolve 300 peaks, but it will help you separating more peaks. In a lot of cases you need to run few experiments in DOE even for unknown compounds. Based on this experiments you can conclude the nature of your compounds. In case of LC/MS-biological samples you don't need base-line resolution for all of the compounds. You can do 2D or 3D chromatography on any mixture ("3D" being another column). I am only saying that you have far better chances to separate compounds on mixed-mode than on RP simply because you have TWO mechanisms one of which is RP. It's up to the end user to decide what column to use, my job just to explain what can be dome. In 15 compounds application you have compounds totally different in nature, it is an example what can be achieved.

Happy Friday for everybody...even for the users of RP columns

P.S. Selling another one or two mixed-mode columns is not going to change anything in my life (I don't need boats, planes or huge houses). We already reached a steady stream of orders and bunch of customers. But I am 100% sure it will help people to save time on method development, time is one of the asset which does not come back. * years ago on this board people stated several time that nobody needs mixed-mode columns...we are still here may be even annoying some readers

[carls]

Regarding the comparison chromatograms, I did not want to show 0% and 5% ACN because some of the chromatograms on RP column look horrible as phase collapse occurred.

In the vast majority of cases phase collapse of an RPLC column (one NOT designed for use with 100% aqueous) does not occur unless you stop the flow/allow the column to depressurize. The reason for poor peak shape or performance of these "polar" compounds with RPLC is likely the injection/sample solvent was too strong and/or the injection volume was too large.

[Vlad Orlovsky]

All samples were dissoled in water and only 5 ul was injected.

[carls]

Please post the chromatograms with the gradient starting at 100% aqueous.

[unmgvar]

Vlad, DOTA is polar
it comes of column at the void and i do not want to try ion pairing
and also it was a good way to really give a chance to mixed mode column and not simply state that it is the best around or the worst.

i am still not convinced about the phase collapse, all of the columns in the example will work in this case,
I am sure that Carl from phenomenex is agreeing with me here.

i think that the best way to try (and it is really to try) to shorten the time used for method development is by using automated methodology during column screening with software. and this works well for the world of small molecules if you first also get to know what you are up against and start the screening with columns that have properties that look like they fit the task, like phase type, pore size, surface area, carbon load, end capping type, pH range.
in the "normal easy" cases it comes down to a single poor pair resolution, the nighmare is to play around with more than that,
and in the crazy world of the pharma
then have that work on different batches
then have the columns go for more than 500 injections
try to get the same with another column from another vendor- that is hard spot in many cases, especially for mixed modes, since especially Sielc has made so many different columns.
and i leave out the biotech out.

[Vlad Orlovsky]

unmgvar,

The issue with DOTA is not polarity. There is no problem to retain DOTA on mixed-mode. It should behave similar to EDTA (see links above), where retention is about 10 minutes on 150 mm column. The main issue in my opinion is analysys of Gd-DOTA compelx. If you use mixed-mode complex is not going to survive and you will end up with only DOTA analysis. Decomposition of the complex might occur on the column which will result in poor peak shape. Another issue is is that sometimes it is hard to elute ions with multiple charges (3+). I am willing to look at this in the lab but I don't have any samples. Ideally I need three things: pure DOTA, sample, Gd salt which is soluble in water and complex Gd-DOTA. Here are ways to approach this:
1. Reversed-phase/anion-exchange method for DOTA plus a separate method for Gd ions by RP/cation-exchange (only ion-exchange mechanismn present). Choice of columns Primesep D and Primesep C
2. One method for Gd and DOTA on RP/anion-exchange column - Gd will be analyzed in cation-exclusion mode in this case and DOTA in RP/AE. Choice of columns Primesep D
3. Trimodal column with a single method on RP/ion-exchnage or HILIC/ion-exchange. Choice of columns Obelisc R (RP/CE/AE) or Pbelisc N (HILIC/CE/AE).
4. Creative approach I have in mind, but I need to test it before sharing

[Vlad Orlovsky]

Some off-topic info:
I have 10 year experience at US big pharma in process development and another SIELC co-founder has 10 years in analytical method development. All our stationary phases were developed based on "problems" in pharma analysis, which we experienced. I don't think that too many people in column manufacturing companies have the same background. I disagree that using prediction software is the best way to do method development, but lets agree to disagree. We developed few phases for particular applications in bioanalysis and then some of this phases found their way to more applications. This happened not once.
For small molecules it is much easier task, because you are not dealing with 100s interactions. For small molecules you usually have 1 or 2 interaction on small molecules and they are usually defined. When you are analyzing unknown metabolite you can usually predict some of the metabolites structures based on your knowledge of chemistry, so not all the time you are doing "blind" analysis. You cannot replace human brain with piece of software. Of course you can buy 5-10 different columns and run screening overnight or weekend and then spend hours to analyze. Our approach is different, we "predict" column and MP based on the knowledge.

[carls]

Some off-topic info:
I have 10 year experience at US big pharma in process development and another SIELC co-founder has 10 years in analytical method development. All our stationary phases were developed based on "problems" in pharma analysis, which we experienced. I don't think that too many people in column manufacturing companies have the same background.

You know what they say about assumptions?

The board users can assess the level of others expertise from the quality of information provided.

[Vlad Orlovsky]

the emphasis was on "too many people" . If you look at % in each column manufacturing company.

[Vlad Orlovsky]

unmgvar,

Do you want to send me samples of DOTA and Gd-DOTA for free method development screening? I checked Aldrich and Alfa Aesar and it will cost us 300$ to buy them, I am willing to look at these if you supply samples.

[unmgvar]

hi vlad
softwares like chromswords dry lab and ACD labs are not intended to replace you. have no fear.
but the automation capabilities and the experimental advantages are huge.
we can look very fast at the chromatographic possibilities not only of the solvents, but at the pH, at the temperature, at the 3 solvents mixture. we can look at RP, NP, HILIC, Chiral, Mixed modes
with the automation we can look overnight smartly at the advantages of using a given buffer over another, at using either MeOH or ACN.
these software are also a great learning tool for new analysts i think.
based on the results it is easy for us to teach them methodology, and how to move on research or troubleshoot, because the software way of doing thinks is simply based upon a logic and step wise procedures.
see more clearly the capabilities of a column, and an experiment. you know faster if there is a need to run that column or set of solvents or to go for another solution and which solution.
once a column is calibrated into the system, we can start predicting behavior before we run it on the HPLC.

it is by no means a magical tool, but you are increasing success ration and more importantly time to success rates.and time is most expensive and non-refundable asset, always.

i cannot give you samples. i have none yet myself. we are only in the first steps of the project.
the problem of the compound is the probability to react with silica, so a very well end capped column might be interesting as well, but then again, all the compounds are polars, why make it simple. and we think it needs to be very pure silica as well, since other trace metals will be bad.
and we are still looking into the need of a derivatization step to stabilize the complex and the Gd itself

we are still waiting for the chromatograms of the first application, since you say it was not a sales trick to show at 5% organic

and when you say "% ratio" of people, you mean simply that you are a small company, so the 2 of you make a lot in the ratio in comparison to phenomenex, agilent or waters?

still, having a very high and loving attitude towards yourself is important never loose that, but humility is yet another key feature.
in the end you have created more then a dozen phases, because life is complex, you have never been able to predict a specific column but rather always at least 2 or more, because like many you are doing what we call, an intelligent guess work based on your own experience, you can give a certain column only because the application exists already out there and you recall it.
like any other column company.

and still i want to remind you that we do use mixed modes columns, we are not against them. there are cases where they work well, but like anything else they are not the only solution, far from it.

[Alexandre]

To Carlos from Brazil,

You probably already have a feeling of complexity of the problem by tsunami of discussion in Forum. Most of it is related just to separation.

There are other aspects of analytical method development. If you are operating in regulated environment e.g. GLP, FDA or ISO 17025, there are requirements to method performance that are imposed by these regulators, e.g. recovery should be no less that 70%. Uncertainty is another big area of development, do you need +/- 100% or +/-1% uncertainty. How to achieve it?

There are many guidelines how to develop and validate analytical methods. It is also necessary to test robustness. It is traditionally done at the end of the project by so called Plackett-Burman experiment.
The latest trend in method development is that one has to do estimation of robustness at the beginning, basically at the stage of setting separation and extraction.

There is also a requirement that 3 analysts have to evaluate the method.

Hope you do not need to do this, but for reference methods the inter-laboratory comparison has to be organized. You may have very good recovery and RSD of duplicates, but 100% wrong answer due to bias.

So even if you have only 1 analyte to make method you need to go through all these exercises, unfortunately (or fortunately as clients pay well for quality). From my experience method development of relatively easy setup may take 6-12 months. Development of chromatography de-novo may need 2-3 months.

To see nice chromatographic peaks is not all.

[carlos.teixeira]

Colleagues,

I am glad with your gentle comments, and would say my intention was ONLY comparing my times with yours. I know about regulatory requirements for analytical works in highly regulated companies.

I hope all the best for you.

Warm hugs from Brazil,

Carlos de Souza Teixeira
http://br.linkedin.com/pub/carlos-de-so ... 15/7b0/125
teixeiracs@yahoo.com