Ask the experts
For reprint orders, please contact [email protected]
Chromatographic baselines Bioanalysis invited a selection of leading researchers to express their views on chromatographic baseline assignment in the bioanalytical laboratory. The topics discussed include the challenges presented with ensuring automated baseline assignment is correct, when reintegration is necessary, regulation and consistency in terminology. Their enlightening responses provide a valuable insight into developing an industry consensus towards reintegration. An accompanying commentary article in this issue, authored by Howard Hill and colleagues (Huntingdon Life Sciences), provides background to this much debated topic.
Graeme Smith, Huntingdon Life Sciences, [email protected]
QQ With the emerging consensus that
all baseline assignment should be automated, what criteria do you use to determine that the first baseline assignment is correct?
The chromatography is visually checked to ensure that each peak is sufficiently well defined (i.e., adequate height or area) and that there is consistency of peak shape, resolution and baseline assignment within the batch. Example chromatograms from the method development/validation report should be available for comparative purposes.
QQ Should doubtful baselines be
repeated? If so, what are the most important criteria for applying new baselines for such repeats?
Yes, if the integration does not look correct, then it should be repeated. The objective
10.4155/BIO.14.67 © 2014 Future Science Ltd
is to try and have consistent baselines throughout the batch. The initial approach should be to adjust the automatic integration parameters and apply these to the whole batch. If this approach does not solve the problem, then the automatic integration parameters should be adjusted and applied to the problematic chromatograms. The integration should not be repeated if the difference between the initial integration and the modified integration is less than 5%.
QQ When choosing between peak area versus height, what factors do you consider? Detectors can be classified as mass flow detectors or as flow sensitive detectors and this should determine whether peak area or height is used. Flow sensitive detectors such as UV detectors do not degrade the analyte and consequently peak height should be used to quantify the analyte as it will be little changed by flow rate variability or interferences to the leading and or trailing edges. Peak area is commonly used for mass flow detectors (especially so for LC–MS).
QQ How can we get the balance between perfect peak shape and separation and optimal integration? Peak shape and separation are affected by many factors that need to be optimized during method development. Collecting the optimum number of data points to properly
Bioanalysis (2014) 6(9), 1167–1170
part of
ISSN 1757-6180
1167
Ask the experts Ask the experts define the peak is also important, as is the use of selective detectors in order to minimize interference with the analyte.
QQ What are the advantages and disadvantages
of developing all criteria a priori?
Criteria set a priori tend to be objective and this important when dealing with peak integration. The disadvantage is that flexibility can be lost, but this is not necessarily a problem when fitting baselines as it is important that operating procedures are applied as objectively as possible.
QQ When would you believe ‘manual’ integration assignment is justified? The important question is not whether manual integration is justified, but how do you define manual integration? Manual integration can be defined in several ways, for example, as a straight point and click to define the start and end of integration or by resetting the automatic integration parameters in the method.
Christopher A James, Amgen Inc., [email protected]
Until there is an industry consensus on definitions, this is a difficult question to answer.
QQ Regulators do not stipulate ‘automated’ baselines – how do you believe integration parameters can be objectively established? By separating integration from regression analysis.
QQ Should this process be defined by the regulators, industry best practice or by individual laboratories? It should be defined by industry best practice with the blessing of the regulators.
QQ Relating to terminology, how consistent are the terms between analysts and manufacturers? The current terminology is very inconsistent and can cause misunderstandings, for example, ‘manual integration’, ‘reintegration’ and ‘integration modification’ can all be interpreted differently. Harmonization of terminology is essential.
the process is rarely used. Samples where poor chromatography prevents appropriate integration may need to be repeated.
QQ When choosing between peak area versus height, what factors do you consider? Peak area is theoretically a more correct reflection of the analyte concentration and is used as a default in our laboratory.
QQ How can we get the balance between perfect peak shape and separation and optimal integration? QQ With the emerging consensus that all baseline
assignment should be automated, what criteria do you use to determine that the first baseline assignment is correct?
Visual inspection of automatically drawn baselines.
QQ Should doubtful baselines be repeated? If so, what are the most important criteria for applying new baselines for such repeats? Action should be taken if the baseline could significantly bias the result. The most important criteria is to apply the same automated approach across all study samples, quality control and calibration samples within an individual run. Our standard operating procedures allow for manual integration of an individual peak, but
1168
Bioanalysis (2014) 6(9)
All analytical methods involve compromises between various method conditions, which need to be addressed during method development. However, with continuing improvements in instrument sensitivity, good S/N can now be achieved for most methods even at the LLOQ, and this greatly facilitates proper integration.
QQ What are the advantages and disadvantages of developing all criteria a priori? A priori criteria for integration help to avoid bias and differences in approach between analysts. The disadvantage is that bioanalysis involves application of complex methodologies to measure very low analyte concentrations in complex matrices; it is therefore difficult to have a priori criteria for every possible circumstance that may be encountered with study sample analysis.
future science group
QQ When would you believe ‘manual’ integration assignment is justified? Perhaps for a method with especially challenging chromatography where automated integration procedures could not be established during method development. There would need to be a justification that this was the only feasible approach, and a priori procedures established to avoid the risk of bias.
QQ Regulators do not stipulate ‘automated’ baselines – how do you believe integration parameters can be objectively established?
Ask the experts
QQ Should this process be defined by the regulators, industry best practice or by individual laboratories? Regulators already pay special attention to integration because of examples of data manipulation they occasionally find. I think that industry best practice is the best way to define such processes, but remembering that individual laboratories are ultimately responsible for the methods they use and the data they generate.
QQ Relating to terminology, how consistent are the terms between analysts and manufacturers?
Integration parameters need to be established as part of method development, but we also have to recognize that instrumental performance (e.g., noise) will vary over time, and some adjustments to these parameters from batch to batch may be required. Review of all integrations by a second person is also important to ensure quality.
I think the terms used by different analysts can sometimes be confusing, for example, ‘manual integration’ is often used to describe a forced baseline fit in a data system although the peak is still integrated automatically.
Rebecca Scott, GlaxoSmithKline, [email protected]
the peak then it should be repeated, (i.e., if the baseline is too high, or too low or if extra peaks have been integrated).
Financial & competing interests disclosure CA James is a full time employee of Amgen, Inc.
QQ When choosing between peak area versus height, what factors do you consider? When deciding if an LLOQ is acceptable on S/N I would use peak height. I would look at the background peak in a blank sample and compare the S/N that I have on the LLOQ to the blank based on height. Peak area is normally similar anyway but you can get situations where your area is acceptable but the height S/N is not.
QQ With the emerging consensus that all baseline assignment should be automated, what criteria do you use to determine that the first baseline assignment is correct? This assignment is carried out based on the analyst experience and ensuring that the initial chromatogram that you use is representative of your assay. Ideally with a good S/N and clean upwards and downwards slope. Not too high a concentration but also not too low.
QQ Should doubtful baselines be repeated? If so, what are the most important criteria for applying new baselines for such repeats? If the automatic integrations has incorrectly integrated
future science group
QQ When would you believe ‘manual’ integration assignment is justified? When the automated software cannot pick up a peak regardless of what parameters you change then I think a manual integration is acceptable as long as the reasons why are recorded and clear.
QQ Should this process be defined by the regulators, industry best practice or by individual laboratories? Yes by best practice, within our laboratory we have guidance in our standard operating procedure as to what can be acceptable within the integrations parameters (e.g., number of smooths).
www.future-science.com
1169
Ask the experts Ask the experts
Eric Woolf, Merck, [email protected]
QQ How can we get the balance between perfect peak shape and separation and optimal integration? One needs to ensure that an adequate number of points are collected over the peak. Inadequate sampling will lead to variable peak share and variable integration.
QQ What are the advantages and disadvantages of developing all criteria a priori? Acceptance criteria need to be developed a priori to eliminate potential bias. However, it is generally not possible to specify integration parameters a priori, as they may change with column/instrument performance. For this reason, it is critical that integration adjustments be made prior to any quantitation.
QQ With the emerging consensus that all baseline assignment should be automated, what criteria do you use to determine that the first baseline assignment is correct? We review all chromatograms in a run to ensure that the baselines are drawn consistently. When deviations are observed, we first attempt to change integration parameters and apply them globally (i.e., reintegrate the entire run). Only when global parameters do not yield consistent baselines do we consider reintegration of individual chromatograms. Another key aspect of our assessment is that baselines are reviewed and accepted prior to any quantitation calculations.
QQ Should doubtful baselines be repeated? If so, what are the most important criteria for applying new baselines for such repeats?
QQ When would you believe ‘manual’ integration assignment is justified? When global integration parameters do not result in consistent baselines, manual integration is justified, provided it is performed prior to any quantitation calculations.
QQ Regulators do not stipulate ‘automated’ baselines – how do you believe integration parameters can be objectively established? Through experience, analysts learn what constitutes proper and improper baselines.
QQ Should this process be defined by the regulators, industry best practice or by individual laboratories? Each laboratory can define the exact process they use. Multiple approaches are suitable, assuming that each does not introduce bias into the assessment.
Poor chromatography (i.e., split or misshapen peaks) in a given sample is an acceptable reason for repeating a sample. If peak shape is OK, the sample should not be repeated; rather baseline should be adjusted appropriately.
QQ Relating to terminology, how consistent are the terms between analysts and manufacturers?
QQ When choosing between peak area versus
Integration parameters tend chromatographic data systems.
to
vary
between
height, what factors do you consider?
We typically always use peak area, as it tends to be less variable than peak height.
Financial & competing interests disclosure
Disclaimer
affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. No writing assistance was utilized in the production of this manuscript.
The opinions expressed in this article are those of the interviewees and do not necessarily reflect the views of Future Science Ltd.
Financial & competing interests disclosure Other than the financial & competing interests disclsosures for CA James & E Woolf, the interviewees have no other relevant
1170
Bioanalysis (2014) 6(9)
E Woolf is a full time employee of Merck.
future science group
For reprint orders, please contact [email protected]
Chromatographic baselines Bioanalysis invited a selection of leading researchers to express their views on chromatographic baseline assignment in the bioanalytical laboratory. The topics discussed include the challenges presented with ensuring automated baseline assignment is correct, when reintegration is necessary, regulation and consistency in terminology. Their enlightening responses provide a valuable insight into developing an industry consensus towards reintegration. An accompanying commentary article in this issue, authored by Howard Hill and colleagues (Huntingdon Life Sciences), provides background to this much debated topic.
Graeme Smith, Huntingdon Life Sciences, [email protected]
QQ With the emerging consensus that
all baseline assignment should be automated, what criteria do you use to determine that the first baseline assignment is correct?
The chromatography is visually checked to ensure that each peak is sufficiently well defined (i.e., adequate height or area) and that there is consistency of peak shape, resolution and baseline assignment within the batch. Example chromatograms from the method development/validation report should be available for comparative purposes.
QQ Should doubtful baselines be
repeated? If so, what are the most important criteria for applying new baselines for such repeats?
Yes, if the integration does not look correct, then it should be repeated. The objective
10.4155/BIO.14.67 © 2014 Future Science Ltd
is to try and have consistent baselines throughout the batch. The initial approach should be to adjust the automatic integration parameters and apply these to the whole batch. If this approach does not solve the problem, then the automatic integration parameters should be adjusted and applied to the problematic chromatograms. The integration should not be repeated if the difference between the initial integration and the modified integration is less than 5%.
QQ When choosing between peak area versus height, what factors do you consider? Detectors can be classified as mass flow detectors or as flow sensitive detectors and this should determine whether peak area or height is used. Flow sensitive detectors such as UV detectors do not degrade the analyte and consequently peak height should be used to quantify the analyte as it will be little changed by flow rate variability or interferences to the leading and or trailing edges. Peak area is commonly used for mass flow detectors (especially so for LC–MS).
QQ How can we get the balance between perfect peak shape and separation and optimal integration? Peak shape and separation are affected by many factors that need to be optimized during method development. Collecting the optimum number of data points to properly
Bioanalysis (2014) 6(9), 1167–1170
part of
ISSN 1757-6180
1167
Ask the experts Ask the experts define the peak is also important, as is the use of selective detectors in order to minimize interference with the analyte.
QQ What are the advantages and disadvantages
of developing all criteria a priori?
Criteria set a priori tend to be objective and this important when dealing with peak integration. The disadvantage is that flexibility can be lost, but this is not necessarily a problem when fitting baselines as it is important that operating procedures are applied as objectively as possible.
QQ When would you believe ‘manual’ integration assignment is justified? The important question is not whether manual integration is justified, but how do you define manual integration? Manual integration can be defined in several ways, for example, as a straight point and click to define the start and end of integration or by resetting the automatic integration parameters in the method.
Christopher A James, Amgen Inc., [email protected]
Until there is an industry consensus on definitions, this is a difficult question to answer.
QQ Regulators do not stipulate ‘automated’ baselines – how do you believe integration parameters can be objectively established? By separating integration from regression analysis.
QQ Should this process be defined by the regulators, industry best practice or by individual laboratories? It should be defined by industry best practice with the blessing of the regulators.
QQ Relating to terminology, how consistent are the terms between analysts and manufacturers? The current terminology is very inconsistent and can cause misunderstandings, for example, ‘manual integration’, ‘reintegration’ and ‘integration modification’ can all be interpreted differently. Harmonization of terminology is essential.
the process is rarely used. Samples where poor chromatography prevents appropriate integration may need to be repeated.
QQ When choosing between peak area versus height, what factors do you consider? Peak area is theoretically a more correct reflection of the analyte concentration and is used as a default in our laboratory.
QQ How can we get the balance between perfect peak shape and separation and optimal integration? QQ With the emerging consensus that all baseline
assignment should be automated, what criteria do you use to determine that the first baseline assignment is correct?
Visual inspection of automatically drawn baselines.
QQ Should doubtful baselines be repeated? If so, what are the most important criteria for applying new baselines for such repeats? Action should be taken if the baseline could significantly bias the result. The most important criteria is to apply the same automated approach across all study samples, quality control and calibration samples within an individual run. Our standard operating procedures allow for manual integration of an individual peak, but
1168
Bioanalysis (2014) 6(9)
All analytical methods involve compromises between various method conditions, which need to be addressed during method development. However, with continuing improvements in instrument sensitivity, good S/N can now be achieved for most methods even at the LLOQ, and this greatly facilitates proper integration.
QQ What are the advantages and disadvantages of developing all criteria a priori? A priori criteria for integration help to avoid bias and differences in approach between analysts. The disadvantage is that bioanalysis involves application of complex methodologies to measure very low analyte concentrations in complex matrices; it is therefore difficult to have a priori criteria for every possible circumstance that may be encountered with study sample analysis.
future science group
QQ When would you believe ‘manual’ integration assignment is justified? Perhaps for a method with especially challenging chromatography where automated integration procedures could not be established during method development. There would need to be a justification that this was the only feasible approach, and a priori procedures established to avoid the risk of bias.
QQ Regulators do not stipulate ‘automated’ baselines – how do you believe integration parameters can be objectively established?
Ask the experts
QQ Should this process be defined by the regulators, industry best practice or by individual laboratories? Regulators already pay special attention to integration because of examples of data manipulation they occasionally find. I think that industry best practice is the best way to define such processes, but remembering that individual laboratories are ultimately responsible for the methods they use and the data they generate.
QQ Relating to terminology, how consistent are the terms between analysts and manufacturers?
Integration parameters need to be established as part of method development, but we also have to recognize that instrumental performance (e.g., noise) will vary over time, and some adjustments to these parameters from batch to batch may be required. Review of all integrations by a second person is also important to ensure quality.
I think the terms used by different analysts can sometimes be confusing, for example, ‘manual integration’ is often used to describe a forced baseline fit in a data system although the peak is still integrated automatically.
Rebecca Scott, GlaxoSmithKline, [email protected]
the peak then it should be repeated, (i.e., if the baseline is too high, or too low or if extra peaks have been integrated).
Financial & competing interests disclosure CA James is a full time employee of Amgen, Inc.
QQ When choosing between peak area versus height, what factors do you consider? When deciding if an LLOQ is acceptable on S/N I would use peak height. I would look at the background peak in a blank sample and compare the S/N that I have on the LLOQ to the blank based on height. Peak area is normally similar anyway but you can get situations where your area is acceptable but the height S/N is not.
QQ With the emerging consensus that all baseline assignment should be automated, what criteria do you use to determine that the first baseline assignment is correct? This assignment is carried out based on the analyst experience and ensuring that the initial chromatogram that you use is representative of your assay. Ideally with a good S/N and clean upwards and downwards slope. Not too high a concentration but also not too low.
QQ Should doubtful baselines be repeated? If so, what are the most important criteria for applying new baselines for such repeats? If the automatic integrations has incorrectly integrated
future science group
QQ When would you believe ‘manual’ integration assignment is justified? When the automated software cannot pick up a peak regardless of what parameters you change then I think a manual integration is acceptable as long as the reasons why are recorded and clear.
QQ Should this process be defined by the regulators, industry best practice or by individual laboratories? Yes by best practice, within our laboratory we have guidance in our standard operating procedure as to what can be acceptable within the integrations parameters (e.g., number of smooths).
www.future-science.com
1169
Ask the experts Ask the experts
Eric Woolf, Merck, [email protected]
QQ How can we get the balance between perfect peak shape and separation and optimal integration? One needs to ensure that an adequate number of points are collected over the peak. Inadequate sampling will lead to variable peak share and variable integration.
QQ What are the advantages and disadvantages of developing all criteria a priori? Acceptance criteria need to be developed a priori to eliminate potential bias. However, it is generally not possible to specify integration parameters a priori, as they may change with column/instrument performance. For this reason, it is critical that integration adjustments be made prior to any quantitation.
QQ With the emerging consensus that all baseline assignment should be automated, what criteria do you use to determine that the first baseline assignment is correct? We review all chromatograms in a run to ensure that the baselines are drawn consistently. When deviations are observed, we first attempt to change integration parameters and apply them globally (i.e., reintegrate the entire run). Only when global parameters do not yield consistent baselines do we consider reintegration of individual chromatograms. Another key aspect of our assessment is that baselines are reviewed and accepted prior to any quantitation calculations.
QQ Should doubtful baselines be repeated? If so, what are the most important criteria for applying new baselines for such repeats?
QQ When would you believe ‘manual’ integration assignment is justified? When global integration parameters do not result in consistent baselines, manual integration is justified, provided it is performed prior to any quantitation calculations.
QQ Regulators do not stipulate ‘automated’ baselines – how do you believe integration parameters can be objectively established? Through experience, analysts learn what constitutes proper and improper baselines.
QQ Should this process be defined by the regulators, industry best practice or by individual laboratories? Each laboratory can define the exact process they use. Multiple approaches are suitable, assuming that each does not introduce bias into the assessment.
Poor chromatography (i.e., split or misshapen peaks) in a given sample is an acceptable reason for repeating a sample. If peak shape is OK, the sample should not be repeated; rather baseline should be adjusted appropriately.
QQ Relating to terminology, how consistent are the terms between analysts and manufacturers?
QQ When choosing between peak area versus
Integration parameters tend chromatographic data systems.
to
vary
between
height, what factors do you consider?
We typically always use peak area, as it tends to be less variable than peak height.
Financial & competing interests disclosure
Disclaimer
affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. No writing assistance was utilized in the production of this manuscript.
The opinions expressed in this article are those of the interviewees and do not necessarily reflect the views of Future Science Ltd.
Financial & competing interests disclosure Other than the financial & competing interests disclsosures for CA James & E Woolf, the interviewees have no other relevant
1170
Bioanalysis (2014) 6(9)
E Woolf is a full time employee of Merck.
future science group
