Impact of the COBAS AmpliPrep/COBAS AMPLICOR HIV-1 MONITOR Test, Version 1.5, on Clinical Laboratory Operations

Abstract

The COBAS AmpliPrep/COBAS AMPLICOR HIV-1 MONITOR Test, version 1.5 (CAP/CA), and the COBAS AMPLICOR HIV-1 MONITOR Test, version 1.5, were compared. CAP/CA reduced and consolidated labor while modestly increasing assay throughput without increased failure rates or direct costs, regardless of batch size and assay format.

Human immunodeficiency virus type 1 (HIV-1) RNA level in plasma is an important predictor of HIV-1 disease progression (3, 6, 7), and measurement of HIV-1 viral load has become an important tool for the management of HIV-1-infected patients (4). The semiautomated COBAS AMPLICOR HIV-1 MONITOR Test, version 1.5 (MCA; Roche Molecular Systems, Inc., Branchburg, NJ), designed specifically for the quantification of HIV-1 RNA in human plasma, utilizes manual nucleic acid extraction and purification techniques in conjunction with automated reverse transcription (RT)-PCR amplification and detection performed by the COBAS AMPLICOR instrument (CA; Roche Molecular Systems, Inc.). The recently introduced, FDA-approved COBAS AmpliPrep/COBAS AMPLICOR HIV-1 MONITOR Test, version 1.5 (CAP/CA; Roche Molecular Systems, Inc.) now combines automated nucleic acid extraction and purification performed by the COBAS AmpliPrep instrument (CAP; Roche Molecular Systems, Inc.) with RT-PCR amplification and detection performed by CA.

Use of CAP/CA eliminates the need for the 1-h ultracentrifugation step currently required when performing MCA with the ultrasensitive assay format and increases the upper quantification limit of the assay from 7.5 × 10⁵ copies/ml to 1.0 × 10⁶ copies/ml with the standard format (9, 10). CAP/CA also has the potential to improve laboratory workflow as well as reduce the hands-on time requirements and run-to-run variability associated with manual sample processing methods (1, 2, 5, 11). Previous studies evaluating the performance of CAP/CA and MCA have demonstrated good overall agreement of test results between the two assays performed with the ultrasensitive format, albeit with slightly lower HIV-1 RNA titers obtained by CAP/CA (1, 2). The current study compares workflow, assay performance characteristics, and direct costs associated with CAP/CA and MCA (standard and ultrasensitive formats) to determine the overall impact of increased use of automation on the clinical laboratory.

CAP/CA and MCA testing was performed with both the standard and the ultrasensitive assay formats and with batch sizes of 12 and 24 samples, following the manufacturer's instructions for use. Hands-on, hands-off, and total time requirements were estimated for each assay, format, and batch size, based on the average times obtained from a minimum of six runs for each combination. These data were used along with several assumptions to construct assay timelines allowing the direct comparison of assay run and specimen throughput under optimal conditions at 8-, 16-, and 24-h durations. The assumptions used in these comparisons were as follows: use of one complete CAP/CA system (a single CAP linked to a maximum of three CA instruments) for all CAP/CA testing, a maximum of three CA instruments dedicated to the performance of MCA testing, unlimited specimen availability, and the continuous effort of multiple laboratory technologists equivalent to one full-time employee per 8-h work shift to perform each assay.

Assay control and clinical specimen data obtained from six runs of 12 samples each (i.e., 6 A-rings) and six runs of 24 samples each (i.e., 12 A-rings) were combined for each assay (CAP/CA and MCA) and format (standard and ultrasensitive) to determine the overall assay control failure rate and invalid specimen result rate for each of the four assay/format combinations. All individual A-rings processed for this portion of the evaluation contained nine specimens and three assay controls. These clinical specimens, which were also used to compare the correlation between CAP/CA and original MCA results, included 162 specimens with titers ranging from <400 copies/ml to >750,000 copies/ml and 162 specimens with titers ranging from <50 copies/ml to >100,000 copies/ml as determined by MCA using the standard and ultrasensitive formats, respectively.

Cost estimates were determined on a per-specimen basis for each assay, format, and batch size, using the manufacturer's current list prices for reagent kits and consumables, while excluding labor costs. Cost comparisons were based on batch sizes of 12 (9 specimens and three assay controls) and 24 (21 specimens and three assay controls) according to the manufacturer's instructions for use of these assays.

In contrast to what was found for MCA, there was no difference between the CAP/CA standard and ultrasensitive assay formats in terms of hands-on or total time requirements (Fig. 1). CAP/CA yielded reductions in hands-on time requirements ranging from 32% (standard format with a batch size of 24) to 40% (ultrasensitive format with a batch size of 12) compared to MCA. In addition to reductions in labor, CAP/CA yielded a consolidation of hands-on time by eliminating multiple periods of nonproductive hands-off time (e.g., periods of ≤15 min each) associated with MCA with either assay format. Importantly, this feature resulted in larger blocks of hands-off time that allowed laboratory technologists to prepare additional assay runs or to perform other tasks within the laboratory. Our findings on the consolidation of hands-on time are consistent with those previously reported by Berger et al. (1) using CAP/CA with the ultrasensitive format.

FIG. 1.

Hands-on and hands-off time requirements and workflow for CAP/CA and MCA with standard (S) and ultrasensitive (U) assay formats and batch sizes of 12 and 24. For a batch size of 12, CAP sample processing required 90 min for the initial assay run (as shown) and 30 min for each of the next two consecutive runs before defaulting back to 90 min for the fourth run due to delays in amplification/detection. For a batch size of 24, CAP sample processing required 120 min for the initial assay run (as shown) and 70 min for each of the next two consecutive runs.

CAP sample processing with a batch size of 12 required 90 min for the initial assay run and 30 min for each of two subsequent runs performed without delay, while a batch size of 24 required 120 min for the initial run and 70 min for each of two subsequent runs. However, since CA instruments must be used in the “basic mode” (rather than the “parallel mode”) for MCA and CAP/CA (8, 10) and there is a limit of three linked CA instruments per CAP, delays associated with CA amplification and detection occurred by the fourth consecutive CAP/CA run. In turn, these delays resulted in a return to the default CAP processing time (i.e., initial run duration) for every fourth assay run and substantially increased CAP processing times (by 50 to 60 min), ultimately limiting the throughput of CAP/CA.

Projected run and specimen throughput were comparable for CAP/CA and MCA using the standard format with batch sizes of either 12 or 24 samples at 8-, 16-, and 24-h durations (Table 1). However, compared to that for MCA with the ultrasensitive format and batch sizes of 12 and 24 samples, the projected 16-h specimen throughputs for CAP/CA (ultrasensitive format) increased by 29% and 25%, respectively, while the 24-h specimen throughputs increased by 17% and 13%, respectively.

TABLE 1.

Projected run and specimen throughput values for CAP/CA and MCA

Duration (h)	Batch sizea	Cumulative throughput for indicated assay, assay format, and workflow parameter
		No. of runs that could be started				No. of runs that could be completed				No. of specimens processed
		Standard		Ultrasensitive		Standard		Ultrasensitive		Standard		Ultrasensitive
		CAP/CA	MCA	CAP/CA	MCA	CAP/CA	MCA	CAP/CA	MCA	CAP/CA	MCA	CAP/CA	MCA
8	12	6	6	6	6	3	2	3	1	27	18	27	9
	24	4	5	4	5	0	1	0	0	0	21	0	0
16	12	12	12	12	11	9	8	9	7	81	72	81	63
	24	9	9	9	9	5	5	5	4	105	105	105	84
24	12	18	18	18	16	14	14	14	12	126	126	126	108
	24	12	12	12	12	9	9	9	8	189	189	189	168

^aResults are based on batch sizes of 12 (9 specimens and three assay controls) and 24 (21 specimens and three assay controls).

CAP/CA yielded assay control failure rates similar to those for MCA with both the standard (5.6% versus 5.6%) and the ultrasensitive (11.1% versus 5.6%) formats (Table 2). It should also be noted that all three CAP/CA control failures were due to either the high positive control or the low positive control producing a result above the respective acceptance range and appeared to be unrelated to improper CAP setup or CAP processing failures. While there was no known cause for the differences in invalid specimen result rates observed between the standard and ultrasensitive formats for both assays, these rates were comparable for CAP/CA and MCA using either assay format. The majority of invalid specimen result flags (five of six) obtained with CAP/CA were due to the presence of clots resulting in specimen transfer failures (i.e., “S_ClotDetected”) during CAP processing, while the majority of the invalid result flags (four of five) obtained with MCA involved poor recovery or amplification of the internal quantitation standard (i.e., “Q_QS_Invalid”).

TABLE 2.

Assay control and invalid specimen result rates for CAP/CA and MCA

Assay format	Assay	No. of A-rings testeda	No. (%) of run failuresb	No. of specimens tested	No. (%) of invalid specimen results
Standard	CAP/CA	18	1 (5.6)	162	5 (3.1)
	MCA	18	1 (5.6)	162	4 (2.5)
Ultrasensitive	CAP/CA	18	2 (11.1)	162	1 (0.6)
	MCA	18	1 (5.6)	162	1 (0.6)

^aIndividual A rings contained nine specimens and three assay controls each.

^bDefined as one or more invalid assay control results per A ring.

CAP specimen transfer failures due to clots resulted in invalid results for 2.5% (4 of 162) and 0.01% (1 of 162) of the clinical specimens tested by CAP/CA using the standard and ultrasensitive formats, respectively. In contrast to a previous CAP/CA evaluation limited to the use of fresh plasma specimens due to concerns about the ability of CAP to process previously frozen plasma specimens (2), our study was conducted exclusively with previously frozen clinical plasma specimens. Our findings suggest that previously frozen specimens do not present a major problem for CAP/CA when care is taken to avoid the transfer of particulates and/or fibrin clots from the original specimen container to the CAP input tube.

Statistical analyses of the log₁₀-transformed, valid clinical specimen results (within the quantifiable range of both CAP/CA and MCA) were performed using SAS version 9.0 (SAS Institute, Cary, NC). Deming regression analyses demonstrated the same good correlation (r = 0.95) and linear association (r² = 0.90) for CAP/CA and MCA results using both standard (90 specimens) and ultrasensitive (74 specimens) formats, with linear equations of y = 1.06x − 0.02 and y = 1.02x + 0.10, respectively. Intraclass correlation coefficients of 0.96 and 0.97 were obtained for CAP/CA and MCA with the standard and ultrasensitive formats, respectively. Bland-Altman plotting further indicated that the mean differences between CAP/CA and MCA for the standard (0.21 log₁₀ copies/ml) and ultrasensitive (0.14 log₁₀ copies/ml) formats were independent of HIV-1 RNA titer (data not shown).

Despite the increased use of automation associated with CAP/CA, the direct costs per specimen, excluding labor, increased by <10% with CAP/CA compared to those for MCA, regardless of the batch size or assay format used. The total reagent and consumable cost estimates ranged from $127 (batch size of 24) to $148 (batch size of 12) for MCA and from $139 (batch size of 24) to $163 (batch size of 12) for CAP/CA using the standard assay format. The cost estimates for MCA using the ultrasensitive format were only slightly higher, ranging from $128 (batch size of 24) to $149 (batch size of 12), while the costs for CAP/CA remained unchanged.

In summary, CAP/CA performed as reliably as MCA with similar assay performance characteristics in this limited evaluation, while reducing hands-on time and consolidating labor through the automation of all major steps involved in the quantification of HIV-1 RNA by RT-PCR. The increased costs associated with CAP/CA may also be offset by reductions in labor-related costs. While assay run and specimen throughput were modestly increased with CAP/CA, workflow was interrupted by CA and CAP processing delays after just three consecutive assay runs, thereby limiting the throughput potential of the CAP/CA platform. In addition, transfer of working master mix and sample extract to individual reaction tubes (i.e., A-rings) must still be performed manually with CAP/CA. It should also be emphasized that CAP/CA remains an endpoint PCR-based assay with a limited dynamic range requiring both standard and ultrasensitive formats to accurately quantify HIV-1 RNA over the range of levels found in clinical specimens. Despite these shortcomings, CAP/CA is a reliable, cost-effective alternative to MCA for those clinical laboratories seeking a labor-saving, quantitative HIV-1 RNA assay capable of modest increases in specimen throughput.