Gregory K.
Webster
*a,
Peter D.
Redfern
a and
Sean J.
Orlowicz
b
aGlobal Analytical R&D, Abbott Laboratories, Abbott Park, IL 60046, USA. E-mail: gregory.webster@abbott.com; Tel: +1 847 935-1295
b411 Madrid Avenue, Torrance, CA 90501, USA. E-mail: SeanO@phenomenex.com; Tel: +1 310 212-0555
First published on 16th December 2009
A simple method for the routine assessment of system performance for liquid chromatographs is presented. The method is a novel use of a commercially available test mix and simple gradient profile to challenge system precision and linearity in order to assess the chromatographic suitability of the system for GxP applications.
For chromatographic instrumentation, the IQ/OQ and PQ are initially established at the time of installation. Depending on the usage, the PQ (or equivalent) is re-performed on a scheduled basis. In the interim, pharmaceutical laboratories use method or system suitability criteria to establish the validity of analytical runs.3 Typically, this includes (1) a series of standard injections to establish suitable precision, (2) resolution, tailing and/or plate number determinations to establish suitable chromatographic performance and (3) injection of a practical detection limit or quantification limit standard to establish appropriate sensitivity.
In our laboratory, we have found using the suitability of the various methods we have running at any one time is not a good enough diagnostic tool for LC performance. Several older instruments were running with periodic failures in meeting suitability criteria. This is a significant issue for it entails not only a regulatory issue in quality audits, but incurs significant costs in terms of repeating assays, performing investigations and completing paperwork.
To address this issue, a quick instrumentation-independent suitability method was developed for use as a routine diagnostic tool to investigate instrument performance. The criteria for such a method was (1) be “simple” so that analysts would actually use it and (2) be efficient so minimal cost in analyst and instrument time would be incurred. The final criterion was that the method must be able to detect performance issues so these issues could be addressed prior to the instruments being released for routine GxP testing. An added bonus is that the sequence only takes a few hours or can run overnight. The method presented here meets proposed criteria for UV and fluorescence detection using a commercially available test mix.
2. Acetonitrile, LC Grade or equivalent.
3. Test Mix, Reversed Phase #2 Test Mix, Phenomenex (Part # ALO-3045). This is a standard test mixture containing Uracil (void volume marker), Acetophenone, Benzene, Toluene and Naphthalene in Acetonitrile.
2. A data acquisition system.
3. Phenomenex Gemini C18 HPLC column, 100 mm × 4.6 mm, 3 μm (Phenomenex, Part # 00D-4439-E0). (See Tables 1 and 2)
1 | UV Wavelength | 254 nm | |
F Wavelength | Ex: 284 nm | ||
Em: 324 nm | |||
2 | Flow Rate: | 1 mL min−1 | |
3 | Injection Volume | 2–25 μL | |
4 | Column Temperature: | 30 °C | |
5 | Mobile Phase | A: | Water |
B: | Acetonitrile |
Time (min.) | %A | % B |
0.00 | 80 | 20 |
2.00 | 80 | 20 |
20.00 | 20 | 80 |
22.00 | 20 | 80 |
22.10 | 80 | 20 |
Re-equilibrate 10 min. | ||
Run Time: 22 min |
Phase #2 Test Mixture using Acetonitrile.
2. Reprime all lines used.
NOTE: Priming of each line should require a minimum of about 20 mL of mobile phase to be drawn to waste as most degassers retain a volume of roughly 15 mL. Adjust this volume as necessary based on the specification of the instrument in question.
3. Allow the system to pump at the initial gradient conditions for at least 30 min, prior to initiating any test runs, to allow the column to equilibrate.
2. Make a second injection of Acetonitrile (Blank) and confirm that there are no peaks, other than void volume disturbances, visible in the resulting chromatogram.
3. Make six 10 μL injections of the Test Solution, calculate the %RSD of the peak areas for the six injections. Typical Retention Times are shown in Table 3 and a typical Chromatogram can be seen in Fig. 1.
Analyte | RT/min | RRT |
---|---|---|
Uracil | 1.1 (unretained) | 0.12 |
Acetophenone | 8.9 | 1.00 |
Benzene | 12.3 | 1.38 |
Toluene | 14.7 | 1.65 |
Naphthalene | 16.4 | 1.84 |
![]() | ||
Fig. 1 Representative Chromatograms of the Test Solution. (A) Uracil, (B) Acetophenone, (C) Benzene, (D) Toluene, (D) Naphthalene. Retention time. |
Requirement: The %RSD should be NMT 1.0.
4. Inject Acetonitrile (blank).
Requirement: There should be no significant carryover of any of the peaks present in the Test Solution.
5. Make a 2, 5 and 8 μL injection of the Test Solution. These are the first 3 injections used to determine injector linearity.
6. Make three 10 μL injections of the Test Solution, for use as a bracketing standard.
7. Make a 12, 15, 20 and 25 μL injection of the Test Solution. These are the final 4 injections used to determine injector linearity.
8. Make three 10 μL injections of the Test Solution, for use as a bracketing standard.
9. Perform a linearity calculation using the data from steps 5 and 7. Plot injection volume vs. peak area and calculate the correlation coefficient.
Requirement: The correlation coefficient should be NLT 0.9950.
After injection of a standard, it is important that the autosampler does not exhibit carryover greater than the practical qualification limit for the LC. This value is typically NMT 0.05% of the nominal method standard. This is where many suitability checks stop. However, systems in use need to be challenged with more than simple precision and carryover.
The procedure moves to challenge not only the linearity of the detector, but the proportionality of the injection system. While it is not critical that an injection value be exactly 10 μL, it is important that the 8 μL and 12 μL injections are 80% and 120% of the nominal 10 uL value respectively. Current LCs on the market today easily post a correlation coefficient of NLT 0.999 from 2–25 uL. For single standard methods, injector precision is more critical than injector accuracy. However, knowing the range of proportionality for each LC injector is an important diagnostic tool as well as documentation of the qualified range in which the instrument can be used in regulated methods. It works as a diagnostic for the detector and autosampler.
The recovery of each standard from the bracketed standard is a better diagnostic of the acceptable injector range than linearity. Each injected volume should be accurate to 98–102% recovery of the theoretical recovery.
In Table 4, the values for a LC system suspected of under performing are presented. Note the high RSD for the precision injections and the poor accuracy of the injections in the linearity run. It should also be noted that correlation is a poor diagnostic tool. Many labs would consider R = 0.99 good enough to illustrate an injector is performing well. Here, the injector is not suitable for GxP chromatography.
Uracil | Acetophenone | Benzene | Toluene | Naphthalene | |
---|---|---|---|---|---|
Rt | 1.166 | 8.249 | 11.527 | 14.035 | 15.706 |
Rt RSD | 0.303 | 0.411 | 0.141 | 0.126 | 0.189 |
Std | 5.881 | 253.955 | 111.047 | 140.118 | 178.658 |
Area RSD | 2.864 | 6.706 | 6.475 | 6.592 | 6.593 |
Uracil | Acetophenone | Benzene | Toluene | Naphthalene | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Area | Area | Recovery (%) | Area | Recovery (%) | Area | Recovery (%) | Area | Recovery (%) | ||
uL | 2 | 37.915 | 74.6 | 16.461 | 74.1 | 20.839 | 74.4 | 26.584 | 74.4 | |
5 | 118.217 | 93.1 | 51.639 | 93.0 | 65.173 | 93.0 | 82.936 | 92.8 | ||
8 | 124.067 | 61.1 | 54.274 | 61.1 | 68.433 | 61.0 | 87.191 | 61.0 | ||
10 | 241.912 | 95.3 | 105.963 | 95.4 | 133.587 | 95.3 | 170.329 | 95.3 | ||
12 | 299.343 | 98.2 | 130.712 | 98.1 | 164.610 | 97.9 | 209.991 | 97.9 | ||
15 | 379.953 | 99.7 | 166.005 | 99.7 | 209.651 | 99.8 | 266.993 | 99.6 | ||
20 | 512.010 | 100.8 | 222.902 | 100.4 | 281.141 | 100.3 | 358.673 | 100.4 | ||
25 | 619.105 | 97.5 | 269.649 | 97.1 | 340.619 | 97.2 | 433.644 | 97.1 | ||
k′ | 0.2 | 7.3 | 10.6 | 13 | 14.7 | |||||
T | 1.2 | 1.1 | 1 | 1 | 1 | |||||
N | 2932 | 36439 | 54630 | 88684 | 128520 | |||||
r | 0.99224 | 0.99223 | 0.99226 | 0.99220 |
The LC instrument from Table 4 was subsequently moved offline for maintenance by the vendor. The vendor determined the autosampler needed to be replaced. After the vendor was finished, the method was again run on this system. The results are listed in Table 5. A significant improvement in precision, accuracy of injection and linearity of response is found. This is the performance expected of GxP instrumentation and with our suitability method, we have documentation that our system is performing at this level.
Uracil | Acetophenone | Benzene | Toluene | Naphthalene | |
---|---|---|---|---|---|
Rt | 1.153 | 8.307 | 11.599 | 14.076 | 15.76 |
Rt RSD | 0 | 0.34 | 0.195 | 0.226 | 0.206 |
Std | 7.840 | 351.144 | 163.360 | 199.767 | 251.101 |
Area RSD | 0.384 | 0.151 | 0.077 | 0.001 | 0.002 |
Uracil | Acetophenone | Benzene | Toluene | Naphthalene | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Area | Area | Recovery (%) | Area | Recovery (%) | Area | Recovery (%) | Area | Recovery (%) | ||
uL | 2 | 71.382 | 101.6 | 32.590 | 99.7 | 40.445 | 101.2 | 51.124 | 101.8 | |
5 | 178.301 | 101.6 | 82.815 | 101.4 | 101.447 | 101.6 | 127.555 | 101.6 | ||
8 | 275.660 | 98.1 | 128.230 | 98.1 | 156.978 | 98.2 | 197.619 | 98.4 | ||
10 | 351.519 | 100.1 | 163.448 | 100.1 | 199.765 | 100.0 | 251.104 | 100.0 | ||
12 | 413.062 | 98.0 | 192.760 | 98.3 | 235.777 | 98.4 | 295.765 | 98.2 | ||
15 | 520.729 | 98.9 | 242.474 | 99.0 | 296.888 | 99.1 | 373.074 | 99.1 | ||
20 | 689.016 | 98.1 | 321.501 | 98.4 | 393.271 | 98.4 | 493.770 | 98.3 | ||
25 | 855.746 | 97.5 | 399.462 | 97.8 | 489.939 | 98.1 | 614.859 | 97.9 | ||
k′ | 0.2 | 7.3 | 10.6 | 13 | 14.7 | |||||
T | 1.2 | 1.1 | 1 | 1 | 1 | |||||
N | 2932 | 36439 | 54630 | 88684 | 128520 | |||||
r | 0.99993 | 0.99994 | 0.99996 | 0.99995 |
Another novel aspect of this method is its ability to challenge the performance of the pumps and gradient proportioning ability by monitoring the resolution of the analyte peaks. Under current evaluation is whether we can use resolution to monitor the pumps and gradient proportioning value.
Naphthalene (UV) | Naphthalene (F) | |
---|---|---|
Rt | 16.182 | 16.247 |
Rt RSD | 0.073 | 0.077 |
Std | 238.847 | 23.396 |
Area RSD | 0.562 | 0.716 |
Naphthalene (UV) | Naphthalene | ||||
---|---|---|---|---|---|
Area | Recovery (%) | Area | Recovery (%) | ||
uL | 2 | 48.240 | 101.0 | 4.742 | 101.3 |
5 | 118.802 | 99.5 | 11.661 | 99.7 | |
8 | 191.061 | 100.0 | 18.711 | 100.0 | |
12 | 287.179 | 100.2 | 28.034 | 99.0 | |
15 | 360.211 | 100.5 | 34.991 | 99.7 | |
20 | 477.044 | 99.9 | 46.236 | 98.8 | |
25 | 595.051 | 99.7 | 57.697 | 98.6 | |
r | 0.99998 | 0.99998 |
This journal is © The Royal Society of Chemistry 2010 |