- Email: [email protected]
The significance of matrix effects on the measurement of lactate dehydrogenase (LD) activity using kodak dry slide technology in the ontario laboratory proficiency testing program
Clin 8iochem, Vol. 23, pp. 179-181, 1990 Printed in Canada. All rights reserved.
0009-9120/90 $3.00 + .00 ~0pyright © 1990 The Canadian Society of Clinical Chemists.
The Significance of Matrix Effects on the Measurement of Lactate Dehydrogenase (LD) Activity Using Kodak Dry Slide Technology in the Ontario Laboratory Proficiency Testing Program S. A. HILL, 1'3 J. C. HEATHCOTE, 1'3 and M. J. McQUEEN 2'3 1McMaster University Medical Centre, Chedoke-McMaster Hospitals,2Hamilton General Division, Hamilton Civic Hospitals, and 3Department of Pathology, McMaster University, Hamilton, Ontario, Canada A recent lactate dehydrogenase (LD) survey of the Laboratory Proficiency Testing Program (LPTP) of Ontario showed interlaboratory coefficients of variation ranging from 6.5% to 40% for five lyophilized vials on the 12 Kodak analyzers. All the LPTP survey samples had similar protein and LD isoenzyme electrophoretic patterns which remained unchanged after reconstitution and storage for 5 days at 4 °C, although the total LD activities fell. Four Ektachem 700 analyzers were subsequently tested using LPTP material and no difference in LD activity between instruments or between two LD slide lot numbers was shown. Generation 9 slides gave higher LD activities than generation 10 on all the reconstituted lyophilized proficiency testing samples. There was no significant difference between slide generations when 19 liquid human sera were analyzed, indicating that the variability on LPTP samples was due to a matrix effect. Definition of the matrix effect of lyophUized proficiency testing material is essential before any proficiency testing program can use such material to reflect analytical performance on patient specimens.
increased variability with Kodak instruments could have been due to instrument-to-instrument variation, variations due to slide lot number or generation number, day-to-day variation resulting from change of activity of reconstituted samples during storage or variation due to interaction of the lyophilized sample with the other components of the dry slide system (the "matrix" effect). This study evaluates contributions from these sources of variability in the measurement of LD in reconstituted lyophilized material by dry slide technology and compares lyophilized material with fresh human serum.
KEY WORDS: lactate dehydrogenase; matrix effects; proficiency testing.
Four Kodak Ektachem 700's, two located at the Hamilton General Hospital and two at McMaster University Medical Centre, were used in this study. Each instrument was calibrated according to the manufacturer's instructions one day prior to the beginning of the study. Lyophilized quality assurance samples were provided by the Laboratory Proficiency Testing Program of Ontario. Fresh human sera were residues of routine samples. LD reagent slides of lot numbers 6310-0026-0474, 6310-00255200 (both generation 10), and 6309-022-5151 (gen-
Introduction
nalysis of the results of the July 1987 survey of A the Laboratory Proficiency Testing Program (LPTP) of Ontario revealed large variability in Lactate Dehydrogenase (LD) activity between those laboratories using Kodak dry slide technology. Table 1 presents a summary of these results. Large coefficients of variation (CV) are apparent in three of the five vials tested. LD results by other methods had much smaller CV's. For example, Technicon RA-1000 technology had CV's of 4.7%, 3.0%, 2.9%, 3.5% (one outlier omitted), and 3.6% for the five vials tested, and the Hitachi 704, 705, and 735's had CV's of 5.9%, 5.2%, 4.8%, 4.8% and 6.1%. The
Correspondence: S. A. Hill, Postgraduate Medical EducationOffice,McMasterUniversityMedicalCentre,Hamilton, Ontario L8N 3Z5, Canada. Manuscript received April 13, 1989; revised July 27, 1989; accepted August 6, 1989. CLINICAL BIOCHEMISTRY, VOLUME 23, APRIL 1990
Materials a n d m e t h o d s
TABLE 1
Summary of LD Result for 12 Kodak Analyzers, July 1987 LPTP Survey Vial
1
2
3
4
5
Range 248-521 468--657 528-779 852-1072 981-1233 U/L Mean
380
543
623
928
1079
SD CV%
95 40
63 12
98 12
61 6
73 7 179
HILL, HEATHCOTE, AND McQUEEN
eration 9) were provided by Eastman Kodak, Rochester, NY, USA. To assess variability between analyzers, between lot numbers, and day-to-day variability, the following protocol was used. Two Ektachem 700's were calibrated with one lot number (6310-0026-0474) and the remaining two were calibrated with the other lot number of the same generation (63100025-5200). LPTP samples were reconstituted according to the instructions provided with them and the samples were analyzed in triplicate on each instrument for each of the next 5 days. Samples were stored at 4 °C until measured. To assess variability between slide generations, one of the two Ektachem 700's at the Hamilton General Hospital was calibrated with slides of generation 9 (lot no. 6309-0022-5151), and the other with slides of generation 10 (lot no. 6310-00260474). Eleven LPTP samples and 19 h u m a n samples of various LD activities and LD isoenzyme compositions were each analyzed six times on the same day on each of the two instruments. To examine the protein and LD isoenzyme composition of the LPTP material, protein electrophoresis and LD isoenzyme electrophoresis were performed on freshly reconstituted samples. LD isoenzyme electrophoresis was repeated after 5 days of storage at 4 °C to detect any changes in the LD isoenzyme composition due to specimen aging. Comparisons were by paired t-test.
Results All LPTP samples showed a similar protein electrophoretic pattern. Albumin accounted for about 50%, alpha 1 & 2 about 10%, and beta and g a m m a fractions accounted for about 40% of the protein present. All LPTP samples showed a similar LD isoenzyme pattern (LD1 > LD2 > LD3). LD4 and LD5 were not present in the samples in detectable amounts. This pattern remained unchanged after storage of the samples for 5 days at 4 °C. All reconstituted samples displayed a significant but comparable loss of activity ranging from 8 to 12% over 5 days. We assessed the variability due to instruments by comparing results obtained on different Ektachem 700's using slides with the same lot number. There was no significant difference between results from the two instruments using slide lot number 63100025-5200 (p = 0.15) or between the two instruments using lot number 6310-0026-0474 (p = 0.46). Variability due to different lot numbers was assessed by comparing the results obtained using lot numbers 6310-0025-5200 to those using 6310-00260474 using a paired t-test. There was no significant difference between the two lot numbers (p = 0.15). Variability due to different generations of slides was evaluated on two Ektachem 700's, one calibrated with slides of generation 9 (lot number 63090022-5151) and one with generation 10 (lot number 6310-0026-0474). A significant difference was noted 180
/
1400
1200
Z
0
1000
J"/'/
o D
I
800
uJ Z uJ
600
o 400 /
200
/
/
// // /
"
I
200
I
400
I
600
800
GENERATION
I
1000
I
1200
I
1400
10
Figure 1--LD activity using reagent slides of different generations. LD activity (U/L) was determined for lyophilized quality assurance material (O--C)), and for liquid human sera (Q---O) using LD reagent slides of generations 9 and 10. Each point represents the mean of six determinations. The least squares line of best fit is shown for both types of sample.
between the results on the 11 LPTP samples when analyzed using different slide generations (p < 0.0001). Generation 9 slides gave results that were approximately 250 U/L higher in each case than those of generation 10 (Figure 1). Regression analyses of these results using the method of least squares gave the equation Gen 9 = 242 + 0.903 (Gen 10), (r = 0.985). When fresh h u m a n sera of various LD activities and isoenzyme compositions were assayed using generations 9 and 10 slides, there was no significant difference in enzyme activity (p = 0.11).
Discussion In this series of experiments we have defined some of the sources of variability in LD assays using Kodak dry-slide chemistry and reconstituted lyophilized sera. Previous studies have yielded conflicting results concerning the effect of storage on LD activity. Lawson and Haven (1) found that reconstituted lyophilized serum stored at 4 °C for 5 days showed a loss of activity of 4-15% while Wilson et al. (2) found that LD in h u m a n nonlyophilized serum is stable for up to 22 days at 4 °C. Our results indicated a significant loss of activity in the reconstituted lyophilized material stored at 4 °C over the course of the experiment. Loss of activity with storage could contribute to variability of results if individual CLINICAL BIOCHEMISTRY, VOLUME 23, APRIL 1990
MATRIX EFFECTS IN THE ASSAY OF LDH ON THE KODAK EKTACHEM
laboratories analyzed LD activity at different times after reconstitution of the lyophilized vials. The loss of activity with aging of the reconstituted specimens could not have accounted for more than one quarter of the variability seen in the initial LPTP survey. The observed variability could not be explained by variations between individual instruments or between different slide lot numbers of the same generation. A comparison of generation 9 and generation 10 slides, however, showed a significant difference with reconstituted lyophilized material but not liquid sera. There was an absolute difference of about 250 U/L of activity between the same lyophilized samples when tested on different generations of slides. This absolute difference remained relatively constant for all samples, even those with the highest LD activity, which explains why samples in the original LPTP survey with the lowest activity had the largest coefficients of variation. Different lot numbers refer to different production runs using the same formulation, while generation number changes indicate a change in reagent slide formulation. In the LPTP survey at least three different slide generations (7, 9, and 10) were in use among those labs reporting results from Kodak instruments. The lyophilized quality control material provided by Kodak (3) showed generation 9 results to be 3-12 U/L lower than generation 7, and generation 10 results 3-12 U/L higher than generation 7. Our data, derived using lyophilized LPTP control material, showed differences between these two generations that were at least 10-fold greater than with the Kodak control material. This presumably reflects differences in the source, composition or preparation of the lyophilized quality control material provided by Kodak and LPTP. Fresh patient samples values did not show the same variability as lyophilized samples. The matrix effect affects patient results only in the indirect sense that lyophilized quality control material may
CLINICAL BIOCHEMISTRY, VOLUME 23, APRIL 1990
not be usable to validate patient results. Subdividing the results by generation number would impair the statistical impact of the results in all b u t the largest proficiency testing programs. Using fresh as opposed to lyophilized samples for quality control presents problems of stability and distribution. It may be possible to identify the component causing the variability in the matrix of the sera and remove it prior to lyophilization or to alter the lyophilization procedure if this were the source of variability. Unfortunately, there would be no guarantee that removing the cause of variability between two generation numbers would assure that all future slides of new generations would yield consistent quality control results. Awareness of this source of variability, which is one among many material-method interactions manifest in proficiency testing programs, will prevent inappropriate interpretation of proficiency testing data, which increasingly are having a legislated impact on the provision of laboratory data for patient care.
Acknowledgements We are grateful to Dr. D. Wood, Director, Ontario Laboratory Proficiency Testing Program, the Enzyme Committee of that program, and Eastman Kodak for materials and advice.
References 1. Lawson NS, Haven GT. Analyte stability in control and reference material. In: Werner M, ed. CRC handbook of clinical chemistry. Pp. 371-89. Boca Ratan: CRC Press Inc, 1982. 2. Wilson SS, Guillan RA, Hocher EV. Studies on the stability of 18 chemical constituents of human serum. Clin Chem 1972; 18: 1498-503. 3. Kodak Technical Bulletin. Rochester NY, USA: Eastman Kodak Inc., April 1987.
181
0009-9120/90 $3.00 + .00 ~0pyright © 1990 The Canadian Society of Clinical Chemists.
The Significance of Matrix Effects on the Measurement of Lactate Dehydrogenase (LD) Activity Using Kodak Dry Slide Technology in the Ontario Laboratory Proficiency Testing Program S. A. HILL, 1'3 J. C. HEATHCOTE, 1'3 and M. J. McQUEEN 2'3 1McMaster University Medical Centre, Chedoke-McMaster Hospitals,2Hamilton General Division, Hamilton Civic Hospitals, and 3Department of Pathology, McMaster University, Hamilton, Ontario, Canada A recent lactate dehydrogenase (LD) survey of the Laboratory Proficiency Testing Program (LPTP) of Ontario showed interlaboratory coefficients of variation ranging from 6.5% to 40% for five lyophilized vials on the 12 Kodak analyzers. All the LPTP survey samples had similar protein and LD isoenzyme electrophoretic patterns which remained unchanged after reconstitution and storage for 5 days at 4 °C, although the total LD activities fell. Four Ektachem 700 analyzers were subsequently tested using LPTP material and no difference in LD activity between instruments or between two LD slide lot numbers was shown. Generation 9 slides gave higher LD activities than generation 10 on all the reconstituted lyophilized proficiency testing samples. There was no significant difference between slide generations when 19 liquid human sera were analyzed, indicating that the variability on LPTP samples was due to a matrix effect. Definition of the matrix effect of lyophUized proficiency testing material is essential before any proficiency testing program can use such material to reflect analytical performance on patient specimens.
increased variability with Kodak instruments could have been due to instrument-to-instrument variation, variations due to slide lot number or generation number, day-to-day variation resulting from change of activity of reconstituted samples during storage or variation due to interaction of the lyophilized sample with the other components of the dry slide system (the "matrix" effect). This study evaluates contributions from these sources of variability in the measurement of LD in reconstituted lyophilized material by dry slide technology and compares lyophilized material with fresh human serum.
KEY WORDS: lactate dehydrogenase; matrix effects; proficiency testing.
Four Kodak Ektachem 700's, two located at the Hamilton General Hospital and two at McMaster University Medical Centre, were used in this study. Each instrument was calibrated according to the manufacturer's instructions one day prior to the beginning of the study. Lyophilized quality assurance samples were provided by the Laboratory Proficiency Testing Program of Ontario. Fresh human sera were residues of routine samples. LD reagent slides of lot numbers 6310-0026-0474, 6310-00255200 (both generation 10), and 6309-022-5151 (gen-
Introduction
nalysis of the results of the July 1987 survey of A the Laboratory Proficiency Testing Program (LPTP) of Ontario revealed large variability in Lactate Dehydrogenase (LD) activity between those laboratories using Kodak dry slide technology. Table 1 presents a summary of these results. Large coefficients of variation (CV) are apparent in three of the five vials tested. LD results by other methods had much smaller CV's. For example, Technicon RA-1000 technology had CV's of 4.7%, 3.0%, 2.9%, 3.5% (one outlier omitted), and 3.6% for the five vials tested, and the Hitachi 704, 705, and 735's had CV's of 5.9%, 5.2%, 4.8%, 4.8% and 6.1%. The
Correspondence: S. A. Hill, Postgraduate Medical EducationOffice,McMasterUniversityMedicalCentre,Hamilton, Ontario L8N 3Z5, Canada. Manuscript received April 13, 1989; revised July 27, 1989; accepted August 6, 1989. CLINICAL BIOCHEMISTRY, VOLUME 23, APRIL 1990
Materials a n d m e t h o d s
TABLE 1
Summary of LD Result for 12 Kodak Analyzers, July 1987 LPTP Survey Vial
1
2
3
4
5
Range 248-521 468--657 528-779 852-1072 981-1233 U/L Mean
380
543
623
928
1079
SD CV%
95 40
63 12
98 12
61 6
73 7 179
HILL, HEATHCOTE, AND McQUEEN
eration 9) were provided by Eastman Kodak, Rochester, NY, USA. To assess variability between analyzers, between lot numbers, and day-to-day variability, the following protocol was used. Two Ektachem 700's were calibrated with one lot number (6310-0026-0474) and the remaining two were calibrated with the other lot number of the same generation (63100025-5200). LPTP samples were reconstituted according to the instructions provided with them and the samples were analyzed in triplicate on each instrument for each of the next 5 days. Samples were stored at 4 °C until measured. To assess variability between slide generations, one of the two Ektachem 700's at the Hamilton General Hospital was calibrated with slides of generation 9 (lot no. 6309-0022-5151), and the other with slides of generation 10 (lot no. 6310-00260474). Eleven LPTP samples and 19 h u m a n samples of various LD activities and LD isoenzyme compositions were each analyzed six times on the same day on each of the two instruments. To examine the protein and LD isoenzyme composition of the LPTP material, protein electrophoresis and LD isoenzyme electrophoresis were performed on freshly reconstituted samples. LD isoenzyme electrophoresis was repeated after 5 days of storage at 4 °C to detect any changes in the LD isoenzyme composition due to specimen aging. Comparisons were by paired t-test.
Results All LPTP samples showed a similar protein electrophoretic pattern. Albumin accounted for about 50%, alpha 1 & 2 about 10%, and beta and g a m m a fractions accounted for about 40% of the protein present. All LPTP samples showed a similar LD isoenzyme pattern (LD1 > LD2 > LD3). LD4 and LD5 were not present in the samples in detectable amounts. This pattern remained unchanged after storage of the samples for 5 days at 4 °C. All reconstituted samples displayed a significant but comparable loss of activity ranging from 8 to 12% over 5 days. We assessed the variability due to instruments by comparing results obtained on different Ektachem 700's using slides with the same lot number. There was no significant difference between results from the two instruments using slide lot number 63100025-5200 (p = 0.15) or between the two instruments using lot number 6310-0026-0474 (p = 0.46). Variability due to different lot numbers was assessed by comparing the results obtained using lot numbers 6310-0025-5200 to those using 6310-00260474 using a paired t-test. There was no significant difference between the two lot numbers (p = 0.15). Variability due to different generations of slides was evaluated on two Ektachem 700's, one calibrated with slides of generation 9 (lot number 63090022-5151) and one with generation 10 (lot number 6310-0026-0474). A significant difference was noted 180
/
1400
1200
Z
0
1000
J"/'/
o D
I
800
uJ Z uJ
600
o 400 /
200
/
/
// // /
"
I
200
I
400
I
600
800
GENERATION
I
1000
I
1200
I
1400
10
Figure 1--LD activity using reagent slides of different generations. LD activity (U/L) was determined for lyophilized quality assurance material (O--C)), and for liquid human sera (Q---O) using LD reagent slides of generations 9 and 10. Each point represents the mean of six determinations. The least squares line of best fit is shown for both types of sample.
between the results on the 11 LPTP samples when analyzed using different slide generations (p < 0.0001). Generation 9 slides gave results that were approximately 250 U/L higher in each case than those of generation 10 (Figure 1). Regression analyses of these results using the method of least squares gave the equation Gen 9 = 242 + 0.903 (Gen 10), (r = 0.985). When fresh h u m a n sera of various LD activities and isoenzyme compositions were assayed using generations 9 and 10 slides, there was no significant difference in enzyme activity (p = 0.11).
Discussion In this series of experiments we have defined some of the sources of variability in LD assays using Kodak dry-slide chemistry and reconstituted lyophilized sera. Previous studies have yielded conflicting results concerning the effect of storage on LD activity. Lawson and Haven (1) found that reconstituted lyophilized serum stored at 4 °C for 5 days showed a loss of activity of 4-15% while Wilson et al. (2) found that LD in h u m a n nonlyophilized serum is stable for up to 22 days at 4 °C. Our results indicated a significant loss of activity in the reconstituted lyophilized material stored at 4 °C over the course of the experiment. Loss of activity with storage could contribute to variability of results if individual CLINICAL BIOCHEMISTRY, VOLUME 23, APRIL 1990
MATRIX EFFECTS IN THE ASSAY OF LDH ON THE KODAK EKTACHEM
laboratories analyzed LD activity at different times after reconstitution of the lyophilized vials. The loss of activity with aging of the reconstituted specimens could not have accounted for more than one quarter of the variability seen in the initial LPTP survey. The observed variability could not be explained by variations between individual instruments or between different slide lot numbers of the same generation. A comparison of generation 9 and generation 10 slides, however, showed a significant difference with reconstituted lyophilized material but not liquid sera. There was an absolute difference of about 250 U/L of activity between the same lyophilized samples when tested on different generations of slides. This absolute difference remained relatively constant for all samples, even those with the highest LD activity, which explains why samples in the original LPTP survey with the lowest activity had the largest coefficients of variation. Different lot numbers refer to different production runs using the same formulation, while generation number changes indicate a change in reagent slide formulation. In the LPTP survey at least three different slide generations (7, 9, and 10) were in use among those labs reporting results from Kodak instruments. The lyophilized quality control material provided by Kodak (3) showed generation 9 results to be 3-12 U/L lower than generation 7, and generation 10 results 3-12 U/L higher than generation 7. Our data, derived using lyophilized LPTP control material, showed differences between these two generations that were at least 10-fold greater than with the Kodak control material. This presumably reflects differences in the source, composition or preparation of the lyophilized quality control material provided by Kodak and LPTP. Fresh patient samples values did not show the same variability as lyophilized samples. The matrix effect affects patient results only in the indirect sense that lyophilized quality control material may
CLINICAL BIOCHEMISTRY, VOLUME 23, APRIL 1990
not be usable to validate patient results. Subdividing the results by generation number would impair the statistical impact of the results in all b u t the largest proficiency testing programs. Using fresh as opposed to lyophilized samples for quality control presents problems of stability and distribution. It may be possible to identify the component causing the variability in the matrix of the sera and remove it prior to lyophilization or to alter the lyophilization procedure if this were the source of variability. Unfortunately, there would be no guarantee that removing the cause of variability between two generation numbers would assure that all future slides of new generations would yield consistent quality control results. Awareness of this source of variability, which is one among many material-method interactions manifest in proficiency testing programs, will prevent inappropriate interpretation of proficiency testing data, which increasingly are having a legislated impact on the provision of laboratory data for patient care.
Acknowledgements We are grateful to Dr. D. Wood, Director, Ontario Laboratory Proficiency Testing Program, the Enzyme Committee of that program, and Eastman Kodak for materials and advice.
References 1. Lawson NS, Haven GT. Analyte stability in control and reference material. In: Werner M, ed. CRC handbook of clinical chemistry. Pp. 371-89. Boca Ratan: CRC Press Inc, 1982. 2. Wilson SS, Guillan RA, Hocher EV. Studies on the stability of 18 chemical constituents of human serum. Clin Chem 1972; 18: 1498-503. 3. Kodak Technical Bulletin. Rochester NY, USA: Eastman Kodak Inc., April 1987.
181