Within-run precision and outlier detection for the Abbott ARCHITECT cardiac troponin I assay

Sawyer et al. ¹ assessed the rate of cardiac troponin I (cTnI) outliers on the Abbott ARCHITECT analyzers. Importantly, they focused on the critical region where the difference between the results of duplicate measurements exceeded the allowable difference, with one of the two measurements above the 99th percentile cutoff (≥0.04 µg/L) and termed this a ‘critical outlier’. ¹ Their findings are noteworthy in that the critical outliers appear to be false positives with the root cause unknown but unlikely to be related to an analyzer malfunction. Our laboratory, since November 2012, has used the Abbott ARCHITECT cTnI assay for routine clinical care. In addition to regular commercial quality control (QC) material, we monitor performance using a low cTnI concentration patient-pool QC material (approximately 0.03 µg/L).^2,3 Recently, our laboratory switched to the 500-test reagent pack from the 100-test reagent pack and through monitoring with the low-pool QC material identified possible positive cTnI outliers at this critical region. After assurances from the manufacturer that the analyzer was functioning appropriately, we set up a series of experiments utilizing a pool of patient samples with ‘normal’ cTnI concentrations to explore the extent of positive outliers and to assess if it was related to a specific analyzer or cTnI reagent lot.

Briefly, for experiment 1 we obtained ethylenediaminetetraacetic acid plasma from patient samples that had a reported concentration of 0.02 µg/L, pooled the material, and aliquoted in 10 individual cups to be analysed on an 8200 (Site 1) and 16200 (Site 2) analyzer after 60 min of instrument inactivity with the same 500-test reagent lot (Lot#17117UN13). The imprecision was doubled on the 8200 as compared with the 16200 analyzer (CV = 26.6% vs. 13.0%; Table 1, experiment 1), with the contributing factor being the first elevated result. Of note, assessing higher concentration patient pools at Site 1 (mid-pool mean (n/CV) = 0.521 µg/L (n = 9/2.2%), high-pool mean (n/CV) = 5.542 µg/L (n = 10/2.6%)) using the same 500-test reagent lot, there was no evidence of an elevated first result effect (i.e. highest concentration obtained after first result). To assess if this first result outlier at the low/normal concentration was due to the specific 500-test reagent lot stored at Site 1, experiment 2 was performed using a 100-test reagent lot (Lot#95509UN13) and a 500-test reagent (Lot#17117UN13) obtained from Site 2. Three rounds of testing were performed with both lots of cTnI reagent with at least 20 min of inactivity before each round of testing (Table 1, experiment 2). These data demonstrate that the first elevated result was apparent only with the 500- and not the 100-test reagent lot at Site 1. Experiment 3 was performed on two additional 500-test reagent lots (Lot#89576UN13 and 23070UN13) versus the 100-test reagent lot, and the first elevated result effect was also evident with the different 500-test reagent lots.

Table 1.

Within-run precision using ‘normal’ cTnI patient-pools to identify a specific lot and instrument outlier effect.

Experiment 1 at Sites 1 and 2
500-test pack Lot#17117UN13
Low patient pool (µg/L)		Site 1 8200 500-test pack		Site 2 16200 500-test pack
Cup1		0.021		0.016
Cup2		0.011		0.018
Cup3		0.010		0.016
Cup4		0.011		0.018
Cup5		0.012		0.017
Cup6		0.012		0.016
Cup7		0.011		0.013
Cup8		0.010		0.015
Cup9		0.012		0.014
Cup10		0.011		0.012
Mean		0.012		0.016
CV		26.6%		13.0%
Experiment 2 at Site 1 only
Low patient pool (µg/L)	100-test pack Lot#95509UN13			500-test pack Lot#17117UN13
	100-test pack Loaded at 12:10	100-test pack Loaded at 12:36	100-test pack Loaded at 13:03	500-test pack from Site 2 Loaded at 14:16	500-test pack from Site 2 Loaded at 14:44	500-test pack from Site 2 Loaded at 15:20
Cup1	0.024	0.024	0.028	0.030	0.029	0.024
Cup2	0.024	0.025	0.030	0.013	0.015	0.014
Cup3	0.022	0.026	0.026	0.015	0.013	0.013
Cup4	0.026	0.026	0.023	0.017	0.015	0.013
Cup5	0.021	0.024	0.025	0.015	0.019	0.018
Cup6	0.024	0.024	0.024	0.016	0.013	0.018
Cup7	0.026	0.023	0.025	0.015	0.012	0.012
Cup8	0.026	0.024	0.028	0.016	0.013	0.016
Cup9	0.023	0.023	0.028	0.014	0.014	0.015
Cup10	0.026	0.031	0.023	0.015	0.014	0.014
Mean	0.024	0.025	0.026	0.017	0.016	0.016
CV	7.5%	9.4%	9.2%	29.1%	32.2%	22.7%
Experiment 3 at Site 1 only
Pack size/Lot#	500-test/89576UN13	100-test/95553UN13	500-test/23070UN13	100-test/95553UN13
Low patient pool (µg/L)
Cup1	0.039	0.028	0.023	0.022
Cup2	0.028	0.026	0.018	0.023
Cup3	0.024	0.023	0.015	0.025
Cup4	0.023	0.040	0.015	0.018
Cup5	0.023	0.027	0.018	0.021
Mean	0.027	0.029	0.018	0.022
CV	24.8%	22.7%	18.4%	11.9%

Shaded concentration is the highest concentration observed during run.

Sawyer et al. did not indicate whether a 500- or 100-test reagent lot was used during their study, nor were they able to identify if the critical outlier was with the first result after a period of inactivity. Our experiments suggest that contributing factors to variability include a period of inactivity on the analyzer (i.e. in running mode with no samples being processed) and the reagent pack size. During our investigation, at no time did we observe differences ≥0.03 µg/L with the normal pool that may have been misinterpreted clinically as a significant change. ⁴ We were able to uncover this positive first result outlier only with a normal cTnI patient-pool, thus reaffirming the importance of monitoring cTnI assays with appropriate QC material below the 99th percentile, regardless of using a high-sensitivity or sensitive assay.^3,5 Finally, our findings indicate a renewed importance of performing within-run precision testing when evaluating cTnI reagent lots.