Peak Deconvolution With Photodiode Array

1) Peak deconvolution of photodiode array (PDA) detector data can separate peaks that are not resolved on-column, providing better detection results and reducing method development and analysis time. 2) PDA data provides spectrum information across a range of wavelengths for each moment in time during the experiment, whereas 2D UV-vis data only shows absorbance at a single wavelength. 3) Peak deconvolution uses an algorithm called multivariate curve resolution – alternating least squares (MCR-ALS) to fit peaks and spectra, unlocking the full potential of PDA data to detect coeluting impurities or reaction products.

Uploaded by

SORIN AVRAMESCU

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

64 views

Peak Deconvolution With Photodiode Array

Uploaded by

SORIN AVRAMESCU

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 2

DEDICATED DIALOGUE SPONSORED CONTENT

Unlocking the Power of 3D

Absorbance Data
Peak deconvolution with photodiode array (PDA)
detectors using a unique software function can separate
peaks that are not resolved on-column, yield better
detection results, and minimize method development
Brian Domanski
and analysis time. Strategic Collaborations
Scientist
Shimadzu Scientific
LCGC: What’s the difference between 2D UV-vis data and 3D Instruments
PDA data?
DOMANSKI: Two-dimensional UV-vis data describes the total
absorbance of a sample versus time for a single chosen wavelength
producing a chromatogram. This type of data only shows that a
Sponsored by
compound has some absorbance at the wavelength of observation
but can’t provide much more information. There is no way to tell
if we are observing the compound at a wavelength of high or low
absorbance or to predict what the compound might be. When we
collect 3D PDA data, for each moment in time during the experiment,
we can observe and record intensity across a range of wavelengths.
This provides useful spectrum information that can determine
maxima that yield the best limits of detection for the compounds of
an assay or be used for traditional PDA data analysis such as cosine-
vector peak purity.

LCGC: Beyond these uses, how does peak deconvolution unlock the
full potential of PDA data?
DOMANSKI: While traditional PDA analysis tools may use a portion
of the spectrum information available, peak deconvolution uses all
the data and applies accepted mathematical techniques to separate
peaks that are not resolved on-column. This can allow for detection
of coeluting impurities in a potency assay, indicate an unexpected
coeluting reaction product, or allow for characterization of a hard-to-
separate degradant.

LCGC: What is the key to peak deconvolution, and how did this
function get developed?
DOMANSKI: At the heart of our PDA peak deconvolution function
is a sophisticated peak-fitting and spectrum-fitting algorithm that
uses the multivariate curve resolution – alternating least squares
(MCR-ALS) approach to determining solution sets, which is often used
in benchtop spectroscopy applications such as chemometrics and
observation of reaction kinetics. Because the technique can work with
complex multi-component systems, it is ideal for LC-PDA data, where
a single experiment may contain many peaks, each with associated
SPONSORED CONTENT DEDICATED DIALOGUE

spectral data. Making the leap from accepted similarity was 0.9998 or greater, indicating a near
technique to implementation in software was done perfect match to empirical data.
in conjunction with Eisai Co. Ltd., a leading global
pharmaceutical manufacturer. Their goal was to LCGC: With these capabilities, where do you see
accelerate therapeutic development by investing in PDA peak deconvolution fitting into pharma and
technologies that would reduce analysis time and biopharma workflows?
increase awareness of coelution-related issues at DOMANSKI: With its ease of implementation,
every stage of the process. this technique can potentially replace complex
multi-dimensional separation or time-consuming
LCGC: How does MCR-ALS deconvolution compare low-flow, high-capacity columns that cannot
to other forms of peak purity assessment? provide rapid results in a user-friendly way. As
DOMANSKI: In contrast to traditional cosine- International Council for Harmonization of Technical
vector peak purity or purity-angle assessment, Requirements for Pharmaceuticals for Human
MCR-ALS deconvolution uses the entire data set Use (ICH) and other regulatory bodies continue to
from the defined time and wavelength domain to tighten requirements for impurities detection, PDA
calculate peak shape and spectrum information peak deconvolution is an additional arrow in the
for each identified chromatographic feature. If quiver of analysis with its ability to detect low-level
users have multiple impurities, up to five peaks coeluting impurities. By identifying these species
can be identified within a single deconvolution at the beginning of drug development, time can
segment, and up to 12 segments can be set in be saved throughout the downstream process. For
each chromatogram. So not only can coelution early-stage high-throughput screening, PDA peak
be spotted, but the potential impurities can be deconvolution could allow for faster injection cycles
characterized with the same tools as any other peak, while maintaining the quality of data.
including the reconstructed spectrum.
In a medicinal chemistry setting, deconvolution
LCGC: Can the deconvoluted peaks be integrated, could be used in UHPLC reaction monitoring to look
and are reconstructed spectra accurate? for hidden peaks or coeluting intermediates or be
DOMANSKI: The short answers are yes, and applied to re-injections of purification fractions as a
yes. With regards to integration, we have secondary check of purity. As the drug-development
studied a three-component system of common cycle continues, any in-silico testing that uses
pharmaceutical drugs under ultra-high- absorbance can be augmented by using PDA peak
performance liquid chromatography (UHPLC) deconvolution to provide information on coeluting
conditions that result in partial separation and peaks without the need to spend additional time on
elution in less than 20 seconds. The resulting LC method development. And for groups that may
chromatogram produces peaks and valleys, but have experience with PDA peak deconvolution, such
none of the peaks are baseline resolved. We then as analytical method research and development,
compared the results of PDA peak deconvolution the ability to use this powerful technique integrated
with the traditional integration strategy of splitting directly into the instrument software provides
the peaks with a vertical drop from the bottom of unprecedented access and ease.
the valley to baseline and with separate injections of
each standard. LCGC: Is Shimadzu planning future development
of this technique?
Traditional valley-to-valley definition of the DOMANSKI: We are developing a new generation
unresolved peaks led to as much as 13.1% deviation of the PDA peak deconvolution algorithm that will
from the single standard injection, whereas using improve performance in terms of sensitivity and
the deconvolution result, peak area was much reliability for some of the most difficult use cases.
more accurately assigned to the peaks, with the Our international team is working in cooperation
greatest error versus single standard injection with pharma users to optimize the feature for lower
being 5.6%. For the same three-component system, detection limits, even when there is minimal on-
the reconstructed spectra were compared with column separation.
the observed spectra from the single standard
injections. For all the reconstructed spectra, the For more information, visit SolutionsForPharma.com.