| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Cancer Centre Karolinska, Radiumhemmet, Karolinska Institute & Hospital, SE-171 76 Stockholm, Sweden.
2
Department of Oncology, Uppsala University Hospital,
SE-751 85 Uppsala, Sweden.
aAuthor for correspondence. Fax 46-8-339031; e-mail torbjorn.norberg{at}cck.ki.se.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: The EMD technology uses the T4 endonuclease VII, which cleaves double-stranded DNA at sites where a DNA mismatch is present because of mispairing or an insertion/deletion of nucleotides. The EMD analyses were carried out by dividing the p53 gene into two overlapping fragments that were analyzed separately. After PCR amplification, the fragments were hybridized with wild-type p53 and subsequently exposed to the EMD enzyme. Cleavage products were analyzed and scored using an ALFTM automated DNA sequencer and ALFwin Fragment Analyzer software (Ver. 1.02).
Results: The EMD technique had sensitivities of 45% and 64% and specificities of 83% and 84% for the two fragments, respectively. Patients with EMD-positive, wild-type p53 tumors had a survival similar to that of patients with EMD-negative, wild-type p53 tumors. Node-positive patients with p53 mutated tumors according to sequencing had a statistically significantly worse overall survival than those with p53 wild-type tumors (P = 0.016), whereas this difference in survival was not detected when p53 status was determined with EMD (P = 0.47).
Conclusions: EMD had insufficient sensitivity for consideration in screening for the p53 gene in this archival material. Sequencing must still be considered as the standard procedure.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The use of sequence analysis for mutation detection is considered an accurate but high-cost and labor-intensive technique. Protein-based methods, immunohistochemistry, and the luminometric immunoassay have lower sensitivity and specificity and provide less optimal prognostic information (8)(10)(11)(12). Accordingly, an accurate mutation-screening method that minimizes sequencing is desired.
EMD was developed for the rapid scanning of DNA to find polymorphisms and mutations. The technology uses an enzyme, the T4 endonuclease VII, that is involved in the recognition of DNA mismatches. This enzyme usually is responsible for recognition and cleavage of branched DNA intermediates formed during DNA replication and packaging. It cleaves double-stranded DNA at sites where a DNA "bubble" is formed because of mispairing or an insertion/deletion of nucleotides.
The EMD technology is claimed not only to detect the presence of a mutation in a sample, but also the mutation site. EMD has been accurate and effective in studies based on genomic DNA of several different genes (13)(14)(15)(16) and has been shown to detect all types of mismatches. Proof-of-principle studies have been performed on the tumor suppressor gene p53 (13)(17)(18)(19).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. Two cell lines with defined p53 mutations, one in
each fragment (see below), were used as positive controls.
|
sequence analysis
The sequencing reactions performed on this breast cancer material
have been described previously (10). In short, using four
overlapping gene-specific primer-pairs, covering all 11 p53
exons, we amplified the p53 gene by PCR. One primer of each
pair was modified by a biotin molecule, generating biotinylated PCR
fragments. An aliquot of each reaction was used to run an agarose gel
for quality and quantity control. The remainder of each reaction was
captured on streptavidin-coated plastic combs and subjected to
solid-phase dideoxy sequencing using the Amersham Pharmacia Biotech
AutoLoad reagent set. The reactions were directly loaded onto an
ALFTM automated DNA sequencer and analyzed by
electrophoresis. Each deviation in the sequence causing an amino acid
exchange was scored as a mutation. Each mutation was verified by
reamplification and sequencing of the fragment using the cDNA
preparation as the starting material.
emd primer design and pcr amplification
The strategy chosen for the EMD analysis was to PCR amplify the
entire coding region of p53 using a nested approach with a
cDNA as the template. The first round of PCR was carried out using
primers PN-8 and PN-9, which amplify a 1379-bp fragment containing all
coding exons of the p53 gene. The second round of PCR was
carried out in two separate reactions: the 5' half of p53
(fragment 1-2) was amplified with primers GA-1 and GA-2 (759 bp) and
the 3' half (fragment 3-4) was amplified with primers P34-1 and P34-2
(703 bp). The two fragments together covered the complete coding
sequence of p53 with a 173-bp overlap.
PCR primers used.
In the first round of PCR (PCR1), the
following primers were used for both fragments: PN-8
(5'-GTGCTTTCCACGACGGTGAC-3') and PN-9
(5'-CAAATGGAAGTCCTGGGTGCTT-3'). In the second round of PCR (PCR2),
the primers used for fragment 1-2 were GA-1
(5'-CCAGACTGCCTTCCGGGTCA-3') and GA-2 (5'-CCGCCCATGCAGGAACTGTTAC-3'),
and the primers used for fragment 3-4 were P34-1
(5'-CTGGCCCCTCCTCAGCATCTTAT-3') and P34-2
(5'-TCAAAGACCCAAAACCCAAAATG-3').
Amplification conditions and cycling protocols.
The
amplification conditions for PCR1 were 1x Buffer II (Perkin-Elmer),
1.5 mM MgCl2, 0.2 mM dNTP, 4 pmol of each primer,
0.5 µL of cDNA, and 1 U of AmpliTaq Gold (Perkin-Elmer) in a total
volume of 20 µL. For PCR2 (the same for both fragments), the
conditions were 1x Buffer II (Perkin-Elmer), 1.5 mM
MgCl2, 0.2 mM dNTP, 4 pmol of each primer, 1 µL
of PCR1 product, and 0.75 U of AmpliTaq Gold in a total volume of 20
µL.
The cycling protocols for PCR1 were as follows: 1 cycle at 95 °C for 9 min; 10 cycles at 95 °C for 30 s, 72 °C for 30 s, and 72 °C for 45 s; 10 cycles at 95 °C for 30 s, 72 °C ramped to 67 °C at 0.5 °C/cycle for 30 s, and 72 °C for 45 s; 20 cycles at 95 °C for 30 s, 66 °C for 30 s, and 72 °C for 45 s; 1 cycle at 72 °C for 5 min; and then hold at 22 °C.
The corresponding protocols for PCR2 were 1 cycle at 95 °C for 9 min; 40 cycles at 95 °C for 15 s, 66 °C for 30 s, and 72 °C for 45 s; 1 cycle at 72 °C for 5 min; and then hold at 22 °C. Both reactions were performed on a Perkin-Elmer 9700 thermocycler.
After amplification, all PCR fragments were checked for quality and quantity by electrophoresis on a 1% agarose gel.
emd reaction and analysis
The human small-cell lung cancer cell line U-1285 (20)
and human breast cancer cell line MDA-MB-231 (21) were used
as positive controls for fragments 1-2 and 3-4, respectively.
MDA-MB-231 (MDA) has a point mutation in codon 280 (G/A), and U-1285
has a 21-bp deletion starting in codon 126 (T. Norberg, unpublished
data).
By amplifying fragment 1-2 from MDA and fragment 3-4 from U-1285, two probes were generated, both of which were confirmed as wild type. The probes were purified according to the manufacturer’s recommendations, and the concentration of each was determined using a spectrophotometer. Amersham Pharmacia Biotech AB (Uppsala, Sweden) supplied all protocols and EMD reaction consumables.
A negative control without the EMD enzyme was included for each sample during the EMD reaction to enable the identification of possible sample contaminants. Running two identical reactions per sample, omitting only the EMD enzyme in the control, generated the EMD-negative controls. We then set two criteria for scoring a sample as positive: (a) the generated peak should deviate from the other samples; and (b) the peak should not be present in the EMD-negative control for the tested sample.
The EMD reactions were performed according to the manufacturer’s recommendations (reagent set supplied) with a few modifications by us that provided better resolution of our samples. All samples, including the positive controls, were treated identically throughout the process. In short, the probe and sample (or positive control) were mixed and hybridized by heating to 95 °C for 5 min followed by 5 min of incubation at room temperature. The reaction was split into two equal parts (7.5 µL + 7.5 µL); one received 2.5 µL of diluted EMD enzyme (250 U per reaction), and the other received only enzyme dilution buffer. After incubation at 37 °C for 30 min, one-half of each sample mixture was transferred to a fresh test tube containing an equal volume of Stop solution (formamide and blue dextran). The samples were then denatured at 90 °C for 2 min before 8 µL of each reaction was loaded onto an ALF automated sequence analyzer equipped with short plates.
The gels were evaluated with the Amersham Pharmacia Fragment Analyzer software (Ver. 1.02). All samples with a unique peak detected were scored as positive. We confirmed each sample that scored positive with EMD by running a completely new analysis starting from cDNA. Only samples that scored positive in both reactions were considered positive and included in the final analysis (see Results).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). For 179 samples, both
the fragments could be amplified, and the resulting p53 data
could consequently be used in survival calculations. Two different
operators (T. N. and L. L.) evaluated the EMD results. All of the
samples were coded so that the operators had no prior knowledge of the
mutational status of the samples. In the first EMD round, deviating
peak patterns were found in 68 of 197 fragment 1-2 samples, whereas no
deviations were found in 129 samples. The positive samples were
reanalyzed, starting from cDNA. We could confirm the peak pattern
deviations in 40 of the 68 samples. These samples were considered as
EMD positives. For fragment 3-4, we could detect 61 positive and 121
negative samples in the first round. After the confirmation round, 40
of the 61 initial samples remained positive and were scored as EMD
positives.
After all samples were analyzed and computerized, the study code was
broken. We then compared the EMD results with the corresponding cDNA
sequencing results. If we consider sequencing as the "gold
standard" for p53 determinations, the EMD method led to 31
false-positive results and 11 false-negative samples for fragment 1-2
from a total of 40 positive and 157 negative samples. As summarized in
Table 1
, this gave a sensitivity [true positives/(true positives +
false negatives)] of 45% [95% confidence interval (CI), 23–68%]
and a specificity [true negatives/(true negatives + false-positives)]
of 82% (95% CI, 76–88%). In total, the corresponding values for
fragment 3-4 were 26 false positives among 40 positive samples and 8
false negatives among 140 negative samples. This gave a sensitivity of
64% (95% CI, 41–83%) and a specificity of 84% (95% CI, 77–89%).
|
By analyzing the sizes of the peaks in the positive samples, we found a
reasonably good correlation (approximately ± 25 bases) with the
site of mutation as determined by sequencing. This was especially true
for fragment 3-4 (Table 2
). For many of the cases, however, there were additional peaks,
and only one of the expected cleavage products was identified for some.
|
unmodified method
To investigate the possibility that our modification of the enzyme
to improve the resolution of the EMD data from our samples may somehow
have interfered with the overall results of the EMD technique, we ran
106 samples with fragment 1-2 using the recommended enzyme amounts.
This gave a sensitivity of 33% (95% CI, 8–70%) and a specificity of
87% (95% CI, 78–93%), whereas our modified procedure had a
sensitivity and a specificity of 45% (95% CI, 23–68%) and 82%
(95% CI, 76–88%), respectively.
assurance of sample quality
The sample quality in terms of PCR quantity and quality varied in
the analyzed cDNAs. We included samples for EMD analysis for which a
band the correct size could be visualized on an agarose gel. Some
samples, however, had a weak intensity or additional bands. The latter
could be compensated for by running the EMD negative controls as
described above. In addition, by calculating sensitivity and
specificity values for a subset of high-quality PCR products (Fig. 2
), we could challenge the argument that the EMD reactions were
conducted on samples of insufficient quality. By selecting the highest
quality PCR products for fragment 3-4, we identified 82 samples of the
total of 182. The sensitivity and specificity values for these
high-quality samples were 72% (95% CI, 39–94%) and 84% (95% CI,
74–92%), respectively (Table 3
), compared with the sensitivity of 64% (95% CI, 41–83%) and
specificity of 84% (95% CI, 77–89%) obtained for all fragment 3-4
samples, as stated above.
|
|
To further verify the integrity of our material, we resequenced a subset of the EMD false-negative samples, using the original method (10), and could confirm 9 of 10 mutations previously scored with sequencing. We failed to confirm one mutation (a 2-bp deletion in codon 214).
correlation with clinical outcome
For a clinical application of the EMD technique, a correlation
between the EMD-generated data and patient outcome would be the primary
objective. We therefore compared the survival curves, according to
Kaplan-Meier, obtained by p53 sequencing (wild-type and
mutated) and EMD analysis (positive and negative), respectively. We
first analyzed the overall survival for all 179 evaluable patients,
divided into four subgroups: (a) wild-type and EMD negative,
(b) wild-type and EMD positive, (c) mutated and
EMD negative, and (d) wild-type and EMD positive. As shown
in Fig. 3A
, the sequencing-generated p53 data separated, to
some degree, the four survival curves into two main groups with better
or poorer outcomes, although the differences between all four subgroups
were not statistically significant (P = 0.19). When we
analyzed the subgroup of patients that were diagnosed with metastases
in the axillary lymph nodes, we detected p53 mutations by
cDNA sequencing, which clearly conferred statistically significant
shorter survival (P = 0.016) compared with wild-type
p53 (Fig. 3B
). Such a distinct and statistically significant
difference was not observed when the p53 mutations were
detected by EMD (P = 0.47), as shown in Fig. 3C
.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The primary aim of this investigation was to examine whether EMD could function as a reliable screening tool for detection of p53 mutations because protein-based screening methods failed to fulfill our requirements.
In this study of archival cDNA samples, EMD did not show the impressive sensitivity reported previously for work with genomic DNA (13)(16) or cDNA (18) as the template. One may argue that (a) the cDNA starting materials were of lower quality because they were collected from 1987 to 1989, and (b) a more recent preparation of cDNA would possibly increase the performance of the EMD technique. However, we deliberately chose these materials because cDNA preparations cannot be generated from all materials, and we believe the quality of a majority of tumor materials used currently may be comparable to ours. A similar study (18)(19) of p53 mutations in cDNA reported a sensitivity of 91% and a specificity of 80% for EMD based on 114 samples with the ALFexpress system, which we used; however, the authors had to exclude 64 samples (36% of the 178 samples available) because they did not fulfill the PCR criteria to qualify for EMD analysis. In addition, the authors conducted their study on fragment 3-4 only, which attained the highest values in our study with respect to sensitivity and specificity. Our materials were, however, of sufficient quality to verify 9 of 10 known mutations using solid-phase sequencing. We wanted to conduct the EMD study on the same cDNA samples that we had used previously for solid-phase sequencing (8), thereby minimizing the risk of getting inconsistent results because of sample or sampling inhomogeneity. In addition, we previously have demonstrated on 100 tumor samples from these same materials that 22 of 23 p53 exon mutations were detected by cDNA sequencing compared with sequence analysis of genomic DNA derived from microdissected material (22). This demonstrates a high consistency between two markedly different sequencing approaches.
One may argue that we used the EMD technology under suboptimal conditions because we reduced the enzyme amounts in the reactions. However, after analyzing 106 samples with fragment 1-2 using the recommended enzyme amounts, we achieved a sensitivity and a specificity of 33% and 87%, respectively. Our experimental procedure had a sensitivity of 45% and a specificity of 82%. Furthermore, in all experiments, we included positive and negative controls, and the results were always as predicted. Thus, we conclude that we have analyzed the present material under conditions similar to or better than the recommended protocol.
The EMD technique appears more dependent on high-quality PCR than
sequencing methods. When we included only PCR products of the highest
quality and quantity, the sensitivity for fragment 3-4 was improved
from 64% to 72% (Table 3
). To achieve this, however, we had to
exclude 60% of the available samples compared with 10% in our initial
strategy. With an exclusion rate of that magnitude, no method can be
considered satisfactory for mutation screening.
A theoretical possibility, however, could be that the EMD technique
might have a higher sensitivity in the detection of p53
abnormalities than cDNA sequencing. The ultimate test would then be to
examine whether the prognosis is worse for patients with positive EMD
results without a verified p53 mutation compared with
patients with wild-type p53 and a negative EMD reaction.
However, that was not the case. Survival analyses showed that the
EMD-generated data did not correlate with the clinical outcome in any
significant way (Fig. 3, A and C
), unlike the sequencing-generated
p53 data, which were statistically significant in
correlation with survival in the whole patient material (data not
shown) and in the lymph node-positive group (Fig. 3B
). The clinical
value of EMD-generated p53 data thus appears limited.
Another conclusion with respect to the EMD technique is that the success rate will vary with the gene fragments analyzed. In this study, we experienced better results with fragment 3-4 than with fragment 1-2, in spite of a higher PCR success rate with fragment 1-2. This indicates that the performance of the EMD reactions will vary depending on the sequence in the fragment to be analyzed.
In conclusion, the requirements for a mutation screening technique should be high specificity, high sensitivity, high throughput, and cost-effectiveness. Our data strongly indicate that EMD does not meet these requirements when used on archival cDNA.
|
Acknowledgments |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  B. Lumbreras-Lacarra, J. M. Ramos-Rincon, and I. Hernandez-Aguado Methodology in Diagnostic Laboratory Test Research in Clinical Chemistry and Clinical Chemistry and Laboratory Medicine Clin. Chem., March 1, 2004; 50(3): 530 - 536. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
2.