163 / 620
Page
11
of
14
Table 2-4
Method comparison
New method
997.08
1
999.03
2
Reference 3
Number of sample preparations
and analyses needed to complete
testing
1
3
3
1
Instrument Run time (min)
45
83x3=249
NAP
65
Can test samples with
sucrose:fructan ratio >3:1
Yes
No
Yes
Yes
Requires post-column reagent
addition
No
No
NAP
Yes
Applicable to FOS,
oligofructose, and inulin
Yes
Yes*
Yes*
Yes
Requires extraction step
No
Yes
Yes
Yes
Requires SPE
No
No
No
Yes
Requires dry-down
No
No
No
Yes
Requires knowledge of
commodity type for accurate
results**
Yes
Yes
Yes
Yes
Method bias for any
commodities
No
No
Yes
No
Uses borate trap
Yes
No
No
No
LOQ (% on RTF basis)
0.03%
(90.8-115%
recovery)
0.5%
0.5%
0.014%
(83-101%
recovery)
%RSD (Intermediate
Precision?)
2.04-7.12%
5.79%
1.25%
4.22-8.37%
1
Stated method performance taken from Covance labs website effective 10/15/2010.
2
Stated method performance taken from Covance labs website effective 01/21/2005.
*Method accuracy compromised by usage of set DP factors of 4 and 10 (actual range is from 3.7-30).
**Limits utility at a regulatory level where the type of material may be unknown or testing for off-label
adulteration may be desired. The new method circumvents this via the qualitative ID methodology in the
Appendix. Reference #3 is pursuing an alternate method to circumvent this problem.
Conceptually this methodology is the similar to that of Cuany, et. al. (3) However several time savings are gained from the elimination of
extraction, SPE, and sample dry-down. An additional time savings is generated in a shorter instrument cycle time (45 minutes versus 65
minutes). Some of this is due to the incorporation of a borate trap to the analysis in place of a standard PA1 guard. Borate negatively
impacts HPAEC-PAD chromatography, creating peak distortion and other issues previously observed (3, 5). Furthermore, the addition of
the internal standard (as suggested in 997.08) allows for a non-volumetric dilution and a subsequent improvement in LOQ.
1.
Linearity –
For the 43 standard curves examined no calibration errors greater than + 5% were observed (see figure 2-3). As
expected the largest errors were observed in the lowest level standards. However the pattern of errors does not indicate any
systematic trends.
FOS-04
FOR ERP USE ONLY
DO NOT DISTRIBUTE