Database of Potential Extractables - PQRI

DATABASE OF POTENTIAL EXTRACTABLES
Status Report
Abstract
Results
Observations
“Theory guides, experiments decides” – I.M. Kolthoff
A list of potential extractables from different polymeric
materials have been compiled and analyzed to develop
predictive models that could aid in risk assessment
regarding the toxicology, detection and migration of
these compounds into Parenteral and Ophthalmic Drug
Products (PODP).
Structural Activity Relationship (SAR) and Statistical
Analysis computer programs are being applied to
develop these models. The results from the controlled
extraction experiments that are being conducted by the
PQRI-PODP working group are being used to help
“decide” the merits of the predictive models.
Introduction
A list of potential extractable compounds has been compiled
based on:
•Manufacturing sources: plasticizing and processing
agents, anti-oxidants, lubricants, etc.
• Published and non-published literature: reported
extractable compounds and related substances.
Database contains:
• CAS numbers and chemical structures
• Literature reported and SAR estimated physical
properties
• Formula weight
• Boiling point
• Partitioning value (Log P and Log D)
.
Process
Development of predictive models:
• Chemical structures – Derek and Cramer programs applied in
the evaluation of potential toxic materials.
• Physio/Chemical Properties – mathematical models applied
in the evaluation of analytical range for chromatographic
methods and the potential for the compounds to migrate from
bulk material.
Status of Investigations:
• Toxicology :
• Total of 612 compounds with structures have been
compiled
• Derek and Cramer analyses of listed compounds have
been completed
•Physio/Chemical Properties:
FW, Log P, Log D and Boiling Point data for 125
compounds have been complied
• Conducted statistical analyses (e.g., Histograms and
Quantile determinations) to assess the analytical range
of the chromatographic methods as related to analyte
partitioning.
• Applied results from the statistical model to assess the
profile of the HPLC detected peaks in the five extraction
studies to determine if improvements in the
chromatographic methods can be made.
Table 1: Example of SAR Toxicology Database
Structure CAS #
Cl
Br
Cl
H 2
ON
H
Cl
Br
Cl
O
O
OH
OH
CH 3
NH 2
OH
Cramer
Class
Cramer comment
79-34-5 3 Q3/4: Non-standard element - Chlorine (alkyl)
3018-20-0 3
88-99-3 1 Q18: No alerting features
Q33/29: Polyaromatic compound without sufficient
sulphate groups
624-20-4 3 Q3/4: Non-standard element - Bromine (alkyl)
108-45-2 3 Q33/32: Insufficient sulphate groups
1081-75-0 3
110-63-4 1 Q18: No alerting features
Q33/29: Polyaromatic compound without sufficient
sulphate groups
Table 2: Example of Physio/Chemical Property Database
CAS # FW Log P Log D pH 4 Log D pH 7 Log D pH 10 BP [°C]
100-21-0 166.13 2.00 1.44 -2.14 -2.15 392
100-41-4 106.17 3.23 3.23 3.23 3.23 136
100-51-6 108.13 1.06 1.06 1.06 1.06 205
100-52-7 106.12 1.45 1.45 1.45 1.45 180
101-77-9 198.21 1.63 -0.08 1.63 1.64 398
102-06-7 211.26 2.36 0.36 0.39 2.00 321
103-23-1 370.57 8.10 8.10 8.10 8.10 374
• Log P is a measure of the hydrophobicity (+ values) or hydrophilicity (- values) of a
neutral or non-ionizable compound.
•Log D is a measure of the hydrophobicity (+ values) or hydrophilicity (- values) of
a substance at a specified pH. If the compound is not ionizable in the aqueous than
Log D = Log P
Figure 1: 3D Contour Plot of BP, Log P and FW
Data for Listed Compounds in Database
Figure 2: Histogram Profiles of Formula Weight
[Daltons], Boiling Point [°C] and Log P Data
Distribution
FW [Daltons]
Formula Wgt.
Maximum (100%) 648.9
Quartile (75%) 281.5
Median (50%) 202.2
Quartile (25%) 126.1
Minimum (0 %) 46.0
Distribution
Log P
Log P
Maximum (100%) 17.6
Quartile (75%) 6.39
Median (50%) 3.18
Quartile (25%) 1.37
Minimum (0 %) -0.63
Distribution
BP [C°]
Boiling Point
Maximum (100%) 806
Quartile (75%) 436
Median (50%) 332
Quartile (25%) 225
Minimum (0 %) 68
Derek/Cramer identified as potentially toxic compound
Figure 3: 3D Contour Plot of Log D Data at pH 4, 7
and 10 for Listed Compounds in Database
Application: HPLC Analytical
Range
• Retention time (RT) data for a set of 8 reference compounds
was compared to reported partitioning values.
• Partitioning values included Log P * and Log D pH7
*
• Values ranged from hydrophilic (Log D pH 7 = -1) to
hydrophobic (Log P =18).
• Plotted data fitted to an exponential regression equation with
a correlation coefficient > 0.9.
• Regression equation was used to calculate partitioning values
for the reported extractable peaks in the five studies.
• Calculated partitioning values for the extraction peaks (see
Fig. 5) were compared to the range of Log P values in the
database (see Fig. 2).
* See Reference section
Figure 4: Correlation Between Partitioning Values
and Retention Times for Reference Compounds
Partition = 1.7 + 0.0111(Ret Time) 2 + 0.344x10 -3 ( Ret Time) 4
R 2 = 0.921
Figure 5: Histogram Profiles for the Calculated
Partitioning Values for HPLC Extractable Peaks
Calculated log P for HPLC peaks
Distribution
Partitioning
Maximum (100%) 21.40
Quartile (75%) 9.19
Median (50%) 7.30
Quartile (25%) 5.12
Minimum (0 %) 1.75
Profile of Database Physio/Chemical Properties
•Formula Weight
Symmetrical distribution over the range of 50 to 400 Daltons with
tailing at FW greater than 600 Daltons
Majority are between 100 to 300 Daltons
• Boiling Point
Asymmetric distribution with skewing starting at BP greater than
600°C.
Majority between 100°C to 500°C
• Log P (Partitioning)
Asymmetric distribution with skewing in the more hydrophobic
region and tailing starting at Log P = 7 extending out to Log P = 18.
Majority are between -1 to 6 which is the typical separation range
for many Reverse Phase LC methods.
• Log D (Distribution – effect of pH)
Comparison of Log D data over the pH range 4 to 10 indicate that
at values ≥8 is where Log D = Log P for these compounds as
demonstrated by the loss of spread in data for the 3D Contour Plot
(see Fig.3) .
Chromatographic separation for potential extractable with
partitioning values ≥8 are better suited for hydrophobic modifiers of
mobile phase, while for compounds with values < 8 the use of pH
modifiers maybe helpful in improving separation.
• Derek/Cramer Identified Compounds
Identified compounds evenly distributed with respect to the
properties of FW, BP and Log P. These properties do not provide an
easy approach to distinguish potential toxic compounds in the
database.
Profile of Extractable HPLC Peaks
• There is a predominance of hydrophobic compounds detected.
Median Log P value in database 3.18 vs. 7.30 in study
Maximum Log P value in database 17.6 vs. 21.4 in study
Minimum Log P value in database -0.63 vs.1.75 in study.
• Majority of chromatographic peaks have partitioning value of ≥ 7
Extraction of the detected peaks appear to be driven by Log P
compared to the Log D factors of ionic or pH effects (see Fig.3) .
Path Forward
• Continue expanding and refining the Database for Physio/Chemical
properties
•Evaluate applications of other predictive models:
.
GC method development related to BP data distribution.
Migration of ingredients and related substances from the respective
polymeric matrices.
References
Software products used in these investigations:
• JMP version 9 statistical analysis (SAS product)
• ACD version 11 Phys/Chem Database and SAR programs (used for Log D and Log P)
• ChemBioDraw Ultra version 12 with chemical properties application (used for Log P and BP)
• SciFinder web application of ACS (used for Log D, Log P and BP)
• Derek for Windows version 12 for toxicity prediction (http://www.lhasalimited.org)
• Toxtree version 1.5 for Cramer classification (http://toxtree.sourceforge.net)
Light Blue contour plot covers 97% sample population density
Red contour plot covers 50% sample population density
Light Blue contour plot covers 97% sample population density
Red contour plot covers 50% sample population density