Clinical Chemistry 44, No. 10, 1998
References 1. Graham IM, Daly LE, Refsum HM, Robinson K, Brattstrom LE, Ueland PM, et al. Plasma homocysteine as a risk factor for vascular disease. JAMA 1997;277:1775– 81. 2. Ueland PM, Refsum H, Stabler SP, Manilow MR, Andersson A, Allen RH. Total homocysteine in plasma: methods and clinical applications. Clin Chem 1993;1764 –79. 3. Fermo I, de Vecchi E, Arcelloni C, D’Angelo A, Patroni R. Methodological aspects of total plasma homocysteine measurement. Haematologica 1997; 82:246 –50. 4. Frantzen F, Faaren AL, Alfheim I, Nordhei AK. Enzyme conversion immunoassay for determining total homocysteine in plasma or serum. Clin Chem 1998;44:311– 6. 5. Fortin L-J, Genest JJ. Measurement of homocyst(e)ine in the prediction of atherosclerosis. Clin Biochem 1995;28:155– 62. 6. Araki A, Sako Y. Determination of free and total homocysteine by high performance liquid chromatography with fluorescence detection. J Chromatogr 1987;422:43–52. 7. Stabler SP, Marcell PD, Podwell ER, Allen RH. Quantitation of total homocysteine, total cysteine, and methionine in normal serum and urine using capillary gas-chromatography-mass spectrometry. Anal Biochem 1987;162: 185–96. 8. Miner SES, Evrovski J, Cole DE. Clinical chemistry and molecular biology of homocysteine metabolism: an update. Clin Biochem 1997;30:189 –201. 9. Kuo K, Still R, Cale S, McDowell I. Standardization (external, internal) of HPLC assay for plasma homocysteine. Clin Chem 1997;43:1653–5. 10. Dudman NP, Guo XW, Crooks R, Xie L, Silerberg JS. Assay of plasma homocysteine: light sensitivity of the fluorescent 7-benzo-2-oxa-1, 3-diazole4-sulfonic acid derivative, and use of appropriate calibrators. Clin Chem 1996;42:2028 –32. 11. Daskalakis I, Lucock MD, Anderson A, Wild J, Schorah CJ, Levene MI. Determination of plasma total homocysteine and cysteine using HPLC with fluorescence detection and an ammonium derivatization protocol optimized for antioxidant concentration, derivatization reagent concentration, temperature and matrix pH. Biomed Chromatogr 1996;10:202–12. 12. Gilfix BM, Blank DW, Rosenblatt DS. Novel reductant for determination of total plasma homocysteine. Clin Chem 1997;43:687– 8. 13. Reddy MN, Behnke C. A rapid and simple assay to determine total homocysteine and other thiols in pediatric samples by high pressure liquid chromatography and fluorescence detection. J Liq Chromatogr Relat Technol 1997;20:1391– 408.
Use of Real-Time Quantitative PCR to Compare DNA Isolation Methods, Jacques B. de Kok,1* Jan C.M. Hendriks,2 Wouter W. van Solinge,1,4 Hans L. Willems,1 Ewald J. Mensink,3 and Dorine W. Swinkels1 (Departments of 1 Clinical Chemistry, 2 Medical Statistics, and 3 Hematology, University Hospital Nijmegen St Radboud, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands; 4 Department of Clinical Chemistry, Eemland Hospital, P.O. Box 4150, 3800 ED, Amersfoort, The Netherlands; *author for correspondence: fax 31-243541743, e-mail
[email protected]) PCR-based diagnostics of genomic DNA and DNA from viruses, microorganisms, or tumor cells are rapidly becoming routine practice in the clinical laboratory. DNA isolation is usually the first step toward detection. Because of expanding numbers of DNA isolation methods, the choice between methods is becoming increasingly difficult. Especially when a low amount of target DNA is present, when only a limited amount of clinical sample is available, or when PCR inhibitors are expected to be present, isolation efficiency, repeatability, and removal of PCR inhibitors by the isolation method become critical factors, seriously affecting the final outcome of the experiment. Recently, a real-time quantitative PCR system [ABI PrismTM 7700 Sequence Detection System, Perkin-Elmer
2201
Applied Biosystems (PE)] was developed. Quantification is automatically performed in the exponential phase of the PCR reaction, where there is a linear relationship between the log of input target DNA quantity and the number of PCR cycles to reach an arbitrary fluorescence threshold (1, 2). The objective of this study was to evaluate this PCR system for standardization of DNA isolation methods. Therefore, we added a known amount of marker DNA copies, close to the detection limit of the PCR, to plasma, which we chose because of recent developments in diagnostic molecular oncology (3, 4). After isolating the DNA with four different methods, we used the PCR system to determine isolation efficiency and repeatability of each method. Blood from five healthy donors was collected in 10-mL EDTA tubes. Plasma was pooled and stored at 280 °C. A 12.8-kb plasmid containing a cDNA sequence construct of the bcr-abl gene translocation was used as marker DNA. The plasmid was linearized with a restriction enzyme, cleaned using a QIAquick column (QIAGEN), and quantified using GeneQuant II RNA/DNA calculator (Pharmacia). A 10-fold serial dilution of the marker DNA was prepared in 10 mmol/L Tris-HCl, pH 8.5, containing 10–105 copies of marker per microliter. The marker was amplified with primers A24 (59-AAAGGTTGGGGTCATTTTCAC-39) and BA5 [59-CGGGAGCAGCAGAAGAAGTGT-39 (316-bp fragment)]. We used bcr-abl [59-(TET)TCAGCGGCCAGTAGCATCTGACTT(TAMRA)p-39, where p indicates phosphorylation] as probe. Each 50-mL amplification reaction contained 5 mL of sample, 5 mL of 103 TaqMan buffer A (TaqMan PCR Reagent Kit, PE), 250 mmol/L each dNTP, 1.25 U of AmpliTaq Gold DNA polymerase (PE), 6 mmol/L MgCl2, 15 pmol of each primer, and 12.5 pmol of probe. MicroAmp optical tubes and caps (PE) were used to prevent light scattering or reflection. PCR cycling conditions were as follows: 10 min at 95 °C, followed by 50 two-step cycles of 95 °C for 30 s, and 60 °C for 90 s. Emission spectra of all tubes were collected every cycle in the last 60 s of the primer annealing/elongation step. The system was directly linked to a Power Macintosh 7200/ 120 (Apple Computer) containing software to program cycling conditions and analyze data. The number of PCR cycles to reach the fluorescence (DRn) threshold value is the cycle threshold (Ct) and is presented by the computer. The Ct value for each sample is proportional to the log of the initial amount of target DNA copies (5). Because of the limited capacity of the thermal cycler, isolation experiments were performed on 2 consecutive days. To eliminate between-day variations in master mixture preparation, one master mixture was prepared for both days, which was possible because AmpliTaq Gold DNA polymerase is inactive without activation for 10 min at 95 °C. A calibration curve was constructed on both days by amplification of a duplicate 10-fold dilution series of marker DNA (50 –500 000 copies). Within-run and between-day differences of the PCR system were measured by adding 1250 copies of marker DNA to sufficient PCR master mixture for 25 reactions each day
2202
Technical Briefs
(50 copies per PCR tube; Fig. 1, system). In two pilot studies of the characteristics of four DNA isolation methods, we observed that after applying the QIAamp blood kit (QIAGEN) no marker copies could be detected. This method was therefore omitted from the experimental design. For the other three DNA isolation methods, 25 000 copies of marker DNA were added to 25 mL of freshly thawed pooled plasma each day. The plasma was then divided into 50 reaction tubes (500 mL each, containing 500 marker copies), which were used in 25 replicate DNA isolations by each method. Five microliters of each sample (10%, representing a maximum of 50 copies) was used for PCR (Fig. 1, method). Control samples were used for every DNA isolation method to determine the individual contribution of loss of marker and presence of PCR inhibitors/enhancers in the sample after isolation (Fig. 1, control). Four DNA isolation methods were used initially. Isolated DNA was always dissolved in or eluted with 50 mL of 10 mmol/L Tris-HCl (pH 8.5). In the PureGene DNA isolation kit (Gentra Systems) method, DNA was isolated as described previously (6); 60 mg of glycogen was added after the protein digestion to facilitate DNA precipitation instead of poly[A] DNA. In the PureGene DNA isolation kit 1 QIAquick column method, after PureGene DNA isolation, samples were further purified according to the “PCR purification protocol” provided with the QIAquick columns. In the phenol– chloroform DNA extraction method, after overnight incubation at 55 °C with 100 mL of protein digestion buffer (600 mmol/L NaCl, 60 mmol/L Tris-HCl, pH 8.0, 120 mmol/L EDTA, 30 g/L sodium dodecyl sulfate, 200 mg of proteinase K), samples were extracted twice with 600 mL of phenol– chloroform–isoamyl alcohol (25:24:1, by volume), and DNA was precipitated with 1/10 volume of 3 mol/L sodium acetate and 1 volume of isopropanol. The DNA pellet was washed with 700 mL/L ethanol and dried. In the QIAamp blood kit method, 500 mL of AL buffer (provided in kit), 15 mL of proteinase K (20 g/L), and 3 mL of poly[A] DNA was added to the plasma samples. After overnight incubation at 55 °C, purification was performed
Fig. 1. Experimental design.
according to the “blood and body fluid protocol” provided with the kit. For all statistical analyses, the mean Ct values of each method were used to homogenize the variances. Because Ct values are proportionally related to the log of the copy number of marker, transformations were performed using a calibration curve. The calibration curves constructed on both days did not differ significantly (P ,0.05) from each other, neither in slope nor in intercept (data not shown). The calibration curve of day 1 was therefore used to obtain point estimates for the median copy number of marker from corresponding mean Ct values (Ct 5 23.444 3 log[marker copies] 1 38.510). In addition, mean Ct values of the system with corresponding standard deviations (presented in Table 1 as median copy number of marker with mean relative variance) did not differ significantly between days (Table 1). The median copy number of marker for the system is not exactly 50 copies because of the inaccuracy of the calibration curve. Because of the high repeatability of the system, we were able to compare the characteristics of isolation methods performed on 2 days. Power analyses showed (using data of pilot studies) that at least 20 measurements were needed to obtain a power of 90% for detecting statistically significant differences when the ratio of the repeatability standard deviations equals two (a 5 0.05). Differences in the mean Ct values between two methods were tested for statistical significance using the t-test in case of equal variances and the Welch test in case of unequal variances and are presented in Table 1 as median marker copies after back transformation. Differences between standard deviations of the Ct measurements of two methods were tested for statistical significance using the Bartlett test (7) and are presented as mean relative variance values. Differences in Ct values between method and control and between system and control, representing loss and inhibition, respectively (Fig. 1), were tested using the Welch test and are presented as relative loss and relative inhibition values of the median copy number of marker (Table 1). Significant differences in the performance (isolation efficiency and repeatability) were found between methods. Moreover, information was obtained about the contribution of copy number loss and PCR inhibitors or enhancers to this performance (Table 1). When the median copy number of marker of the system (day 2) is used as denominator, isolation efficiencies were 70% [95% confidence interval (CI), 60 – 83%], 34% (95% CI, 32–36%) and 51% (95% CI, 45– 60%) for the PureGene, PureGene 1 QIAquick, and phenol– chloroform methods, respectively; all methods were significantly different from each other as well as significantly different from the system. Precision (mean relative variance) of the PureGene 1 QIAquick method (0.274) was similar to that of the system and significantly lower than that of the PureGene and phenol– chloroform methods. Relative losses of marker DNA copies were 35%, 38%, and 66% for the PureGene, PureGene 1 QIAquick, and phenol– chloroform methods, respectively. For the PureGene and especially the phenol–
2203
Clinical Chemistry 44, No. 10, 1998
chloroform methods, this loss was partially compensated by the presence of PCR enhancers (210% and 252% inhibition, respectively). Although the PureGene method has the highest isolation efficiency, it has a relatively poor precision. Use of a QIAquick column after PureGene isolation markedly decreased the isolation efficiency (from 70% to 34%); however, it simultaneously increased the precision to a value equal to that of the system. Because there was not much additional relative loss when the QIAquick column was used (35% vs 38%), this suggests that the column removes variable amounts of PCR inhibitors/enhancers still present in the sample after PureGene DNA isolation, although the number of control samples should be enlarged to confirm this statistically. At the same time, a constant amount of PCR inhibitors is introduced, probably by one of the buffers in the QIAquick kit. An extra column washing step or the use of another column washing buffer may eliminate these PCR inhibitors and could make this method very robust. The additional use of a column can be especially advantageous when using plasma samples of patients, which will probably contain more diverse amounts of PCR inhibitors or enhancers than the pooled plasma of healthy donors used in this study. The QIAamp method had very poor isolation efficiency in our pilot studies when only few copies of marker DNA were present, consistent with other studies (8, 9). Al-
though our results for plasma appeared to conflict with a study performed by Dixon et al. (10), where the QIAamp method was their first choice of 13 tested DNA isolation methods from patient sera (considering DNA isolation from serum equal to plasma), both studies cannot be compared because genomic DNA copy numbers in sera from cancer patients are expected to be much higher than the copy number of marker used in our study (11, 12). Moreover, in their study ;50% of the sera that were “PCR negative” after the QIAamp method could be PCR amplified after application of another DNA isolation method. This indicates that efforts to improve isolation efficiency (loss, removal of inhibitors) using other DNA isolation methods may still be successful, especially when dealing with low copy numbers of target DNA. With this in-depth analysis of several DNA isolation methods, we try to emphasize that a well-considered choice is very important for standardization of PCR-based diagnostics when only a limited amount of target DNA may be present in a clinical sample. In conclusion, because of its high repeatability, its closed tube system, which reduces PCR product contamination and assay time, and the ability to gain information about the performance of DNA isolation methods, real-time quantitative PCR performed by the ABI Prism 7700 Sequence Detection System will provide a major step toward standardization in PCR-based diagnostics.
Table 1. Median marker DNA copies (95% CI) for the ABI Prism 7700 Sequence Detection System and different methods for the isolation of marker DNA from plasma. Methods, day 1 Systema (ABI 7700)
PureGene
Methods, day 2 PureGene 1 QIAquick
Systema (ABI 7700)
Phenol–chloroform
a
Method Median 46b 33c 16d 47b 24e (95% CI) (43, 49) (28, 39) (15, 17) (43, 51) (21, 28) n 25 25 25 25 25 Mean relative variance 0.228f 0.639g 0.274f 0.305f 0.549g Controla Median NAh 51 26 NA 70 (95% CI) (37, 68) (16, 44) (42, 118) n NA 5 4 NA 5 ..................................................................................................... Relative lossi Median NA 35%j 38% NA 66%j (95% CI) (11%, 52%) (23%, 63%) (43%, 79%) n 5 4 5 Relative inhibitionk Median NA 210% 43%j NA 252% (95% CI) (247%, 19%) (6%, 66%) (2153%, 8%) n 5 4 5 a
See Fig. 1. Same letters indicate no statistically significant difference in median copy numbers between methods (Welch test/t-test). f,g Same letters indicate no statistically significant difference in mean relative variance of copy numbers between methods (Bartlett test). h NA, not applicable. i Relative change of marker copies lost during isolation. j Significantly different from 0% (Welch test). k Relative change of marker copies caused by the presence of PCR inhibitors. Note that negative values indicate the presence of PCR enhancers in the sample. b– e
2204
Technical Briefs
We thank G. Grosveld (St. Jude’s children’s Research Hospital, Memphis, Tennessee) for providing the marker plasmid and L. van de Locht (Department of Hematology, University Hospital Nijmegen) for assistance and constructive suggestions using the ABI Prism. References 1. Gibson UE, Heid CA, Williams PM. A novel method for real time quantitative RT-PCR. Genome Res 1996;6:995–1001. 2. Heid CA, Stevens J, Livak KJ, Williams PM. Real time quantitative PCR. Genome Res 1996;6:986 –94. 3. Chen XQ, Stroun M, Magnenat JL, Nicod LP, Kurt A, Lyautey J, et al. Microsatellite alterations in plasma DNA of small cell lung cancer patients. Nat Med 1996;2:1033–5. 4. Anker P, Lefort F, Vasioukhin V, Lyautey J, Lederrey C, Chen XQ, et al. K-ras mutations are found in DNA extracted from the plasma of patients with colorectal cancer. Gastroenterology 1997;112:1114 –20. 5. Higuchi R, Fockler C, Dollinger G, Watson R. Kinetic PCR analysis: real-time monitoring of DNA amplification reactions. Biotechnology 1993;11:1026 –30. 6. de Kok JB, van Solinge WW, Ruers TJM, Roelofs RW, van Muijen GN, Willems JL, Swinkels DW. Detection of tumour DNA in serum of colorectal cancer patients. Scand J Clin Lab Investig 1997;57:601– 4. 7. Snedecor GW, Cochran WG, eds. Statistical methods, 6th ed. Ames, IA: Iowa State University Press, 1967:296 – 8. 8. Krajden M, Shankaran P, Bourke C, Lau W. Detection of cytomegalovirus in blood donors by PCR using the Digene SHARP signal system assay: effects of sample preparation and detection methodology. J Clin Microbiol 1996; 34:29 –33. 9. Kramvis A, Bukofzer S, Kew MC. Comparison of hepatitis B virus DNA extractions from serum by the QIAamp blood kit, GeneReleaser, and the phenol– chloroform method. J Clin Microbiol 1996;34:2731–3. 10. Dixon SC, Horti J, Guo Y, Reed E, Figg D. Methods for extracting and amplifying genomic DNA isolated from frozen serum. Nat Biotechnol 1998; 16:91– 4. 11. Leon SA, Shapiro B, Sklaroff DM, Yaros MJ. Free DNA in the serum of cancer patients and the effect of therapy. Cancer Res 1977;37:646 –50. 12. Shapiro B, Chakrabarty M, Cohn EM, Leon SA. Determination of circulating DNA levels in patients with benign or malignant gastrointestinal disease. Cancer 1983;51:2116 –20.
Addition of Sodium Fluoride to Whole Blood Does Not Stabilize Plasma Homocysteine But Produces Dilution Effects on Plasma Constituents and Hematocrit, Martin Patrick Hughes,1* Timothy H. Carlson,1† M. Kathleen McLaughlin,1† and Daniel D. Bankson1,2 (1 Department of Laboratory Medicine, University of Washington, Seattle, WA 98195 and 2 Veterans Affairs Puget Sound Health Care System, Seattle, WA 98108; * author for correspondence: e-mail
[email protected]; † present address: Pacific Biometrics, Inc., Seattle, WA 98119) Interest in total plasma homocysteine (Hcy) measurements has increased with the availability of evidence that even mild hyperhomocysteinemia is an independent risk factor for cardiovascular disease (1). Consequently, the variability of current Hcy results has been the subject of considerable research (2– 4). Whole blood stored at room temperature after phlebotomy shows an increase in Hcy concentrations of ;10% per hour, apparently because of Hcy synthesis and release from erythrocytes (5, 6). Previous studies have shown that the best method for avoiding falsely high Hcy results is prompt postphlebotomy centrifugation and separation of the plasma from the red blood cells (6) or storage of whole blood samples at 0 °C if prompt centrifugation is not feasible (3). The use of
anticoagulants as preservatives in whole blood has also been advocated (2– 4); among these, sodium fluoride (NaF) in particular has been described as useful when added to whole blood at suggested concentrations of up to 4 g/L (95 mmol/L) (4, 7). Many laboratories, therefore, now use NaF as a preservative in the blood collection tubes used for Hcy assays. We measured plasma Hcy, using a modified version of the HPLC method with fluorescence detection introduced by Refsum and co-workers (8), after the addition of pure NaF (2.5 g/L, 60 mmol/L) to heparinized whole blood. The studies were approved by the University of Washington Institutional Review Board and involved the donation of blood from men and women after informed consent. We found minimal inhibition of the time-dependent increase in Hcy concentrations. We also investigated the preservative abilities of Na2EDTA (1.5 g/L, 4.0 mmol/L), EGTA (1.5 g/L, 3.9 mmol/L), and thymol (1.0 g/L, 6.6 mmol/L), again added in pure form to heparinized whole blood. All chemicals were obtained from the Aldrich Chemical Co. and used without further purification. None provided satisfactory results. For example, during 8 h at 25 °C, plasma Hcy increased as follows: control (containing no preservative), 8.9% per hour; thymol, 9.8% per hour; EGTA, 8.0% per hour; Na2EDTA, 6.5% per hour; and NaF, 6.0% per hour. During the studies of these additives, we were intrigued to observe a decrease in the Hcy concentrations of the NaF-containing samples at all time points as compared with the blank control and compared with the samples that contained the other potential preservatives. During further investigations, we discovered that addition of NaF to heparinized whole blood caused an immediate NaF concentration-dependent decrease in spun hematocrit (Table 1). This led us to propose that on addition to the whole blood sample, the NaF produced hypertonic plasma with subsequent desiccation of the red blood cells. The resulting osmosis effectively diluted the plasma to such an extent that it lowered the apparent Hcy concentration by ;10% at 100 mmol/L NaF. Plasma albumin (measured by bromcresol green on the Paramax RX analyzer) also decreased linearly with increasing concentrations of NaF, confirming that the change in hematocrit was not artifactual (Table 1). Because lithium is not excluded from the erythrocyte and so does not cause the same concentration-dependent plasma dilution, we investigated the preservative properties of lithium fluoride (LiF, Fisher Scientific). LiF decreased hematocrit, albumin, and Hcy ,3% (Table 1). Next, the time-dependent effect of LiF and NaF on plasma Hcy was investigated. Whole blood with K3EDTA anticoagulant (1.8 g/L) was transferred into sample tubes containing dry LiF, dry NaF, or no preservative. These salts (50 mmol/L blood) both inhibited the rise of Hcy concentrations at 25 °C only minimally, compared with a control with no additive (Fig. 1). In fact, storage of the blood samples on ice provided the only effective inhibition of plasma Hcy increase over time. It is interesting to note that the plasma Hcy concentration of the NaF-
