Antioxidant activity of green tea extracts estimated from chromatograms using partial least squares

1
Antioxidant activity of green tea extracts
estimated from chromatograms using partial least
squares regression and a colorimetric reference
method A.M. van Nederkassel, (avnederk_at_vub.ac.be),
M. Daszykowski, D.L. Massart and Y. Vander
Heyden Vrije Universiteit Brussel, Pharmaceutical
and Biomedical Analysis, Pharmaceutical
Institute, Laarbeeklaan 103, B-1090 Brussels,
Belgium
AIM OF OUR STUDY The aim is to predict the total
antioxidant capacity of green tea extracts from
their chromatograms by use of a calibration model
built between the chromatographic method and the
reference method used to measure the antioxidant
capacity. METHODOLOGY Step 1 The antioxidant
capacity of 55 green tea extracts is measured
with a colorimetric method called the TEAC assay
(Trolox Equivalent Antioxidant Capacity assay).
The antioxidant capacity is expressed as a TEAC
value (mM). Step 2 Simultaneously, the 55
green tea extracts are chromatographed on a
monolithic silica column. Step 3 The
chromatograms obtained in step 2 are aligned with
a method called Correlation Optimized Warping
(COW). Step 4 Robust PCA is used to detect
leverage objects among the chromatograms (X) and
histograms are used to detect outliers among the
TEAC values (y) The remaining objects are divided
into a calibration and test set. Step 5 Partial
Least Squares regression (PLS) and Partial Least
Squares regression with un-informative variable
elimination (UVE-PLS) are used to build a
regression model between the chromatograms (X)
and TEAC values (y). This model is then used to
predict the TEAC value (ynew) from the
chromatogram (Xnew) of a new green tea extract.

• STEP 2 Fast chromatographic analysis of the 55
  green tea extracts
• HPLC conditions (the 55 tea extracts are
  chromatographed in duplicate)
• stationary phase Chromolith SpeedROD RP-18e
  (50 x 4.6 mm)
• Chromolith Performance RP-18e (100 x 4.6 mm)
• mobile phase gradient elution (ACN/H2O both
  0.05 Trifluoroacetic acid)
• from 2 tot 26 ACN in 10 min. and from 10 to
  11 min. 26 ACN
• UV detection 280 nm
• flow rate 2 ml/min
• column temperature 30C
• STEP 3 Aligning of green tea extract
  chromatograms Correlation Optimized Warping

Chemometric treatment of the chromatograms
requires that all signals are adjusted to the
same length and that corresponding variables
(such as peak apexes) are placed in proper
columns of the data matrix. COW aligns
chromatograms by piecewise linear stretching and
compression of the time axis.
STEP 4 Elimination of bad leverage objects in X
and outliers in y subset selection
The score diagnostic plot of the green tea
chromatograms (X). The robust distance is
plotted versus the orthogonal distance. The
cut-off values are determined in the space of 5
rPCs. ? chromatograms 61 62 are leaverages ?
the corresponding tea sample is removed from the
data set
Histogram of the 55 TEAC values (y) ? 2 tea
extracts have outlying TEAC values ? both are
removed from the data set
Then, the data set is divided into a calibration
set (40 tea samples), to build the model, and a
test set (12 tea samples) to validate the model.
The calibration set was selected by uniform
sampling of sorted TEAC values - from low to
high. STEP 5 Partial least squares regression
(PLS) and UVE-PLS Regression models are built
with Partial least squares regression (PLS) and
un-informative variable elimination Partial least
squares regression (UVE-PLS). The model
complexity (Fn) is determined by leave-one-out
cross validation and the model complexity
resulting in the lowest Root Mean Squared Error
of Cross-validation (RMSECV) is chosen. The PLS
model is validated with an independent test set
of 12 objects wherefore the Root Mean Squared
Error of prediction (RMSEP) is computed. By use
of UVE-PLS, the model complexity could be reduced
to 7 and only 142 of the 3100 chromatographic
variables are retained to build a model. The
predictability of the obtained models is found to
be acceptable compared to the precision of the
TEAC assay.

• STEP 1 TEAC assay
• The TEAC assay measures the capacity of a
  compound to scavenge the blue-green ABTS cationic
  radical (ABTS? 2,2-Azino-bis-(3-ethylbenzothia
  zoline-6-sulfonic acid)
• The decolorization of the blue-green ABTS?
  solution after addition of the green tea extract
  (or Trolox solution) is measured at 729 nm
  resulting in a colorless product.
• The antioxidant capacity of the green tea
  extracts is expressed as the equivalent
  concentration (in mM) of a water-soluble vitamin
  E analogue (Trolox).
• Trolox 6-hydroxy-2,5,7,8,-tetramethylchroman-2
  -carboxylic acid
• TEAC values of the 1 (g/v) green tea extracts
  are between 2800 and 4400 mM.
• The precision of the TEAC assay is determined for
  3 teas with low (2999), intermediate (3779) and
  high (4058) TEAC value and found to be 93 mM
  (5.62), 104 mM (6.29) and 204 mM (12.33)
  respectively.

CONCLUSION A stable and reliable model is built,
able to predict the antioxidant capacity of green
tea extracts (expressed as the TEAC value) from
chromatograms with analysis times of 11 minutes
obtained on monolithic silica columns.