Campus de Botucatu
June 23, 2008
To
Robin Cassady-Cain
Assistant Editor
The BioMed Central Editorial Team – BMC Urology
Dear Editor,
We would like to thank the editors and the reviewers for their constructive
criticisms regarding our manuscript, entitled “DNA methylation patterns in bladder
cancer and washing cell sediments: a perspective for tumor recurrence detection”, by
Priscilla Davidson Negraes et al.
According your suggestions, we alter the manuscript and included additional
information. We append each point the reviewers made and our responses. We have
indicated the page and line number in which changes or additional information has
been included in the modified text of the revised manuscript. All authors have read
and approved the revised version of this manuscript.
We look forward to hearing from you concerning the suitability of the revised
manuscript for publication.
Sincerely,
Cláudia Aparecida Rainho, PhD
Department of Genetics, Biosciences Institute
Sao Paulo State University – UNESP
Botucatu – São Paulo – Brasil
rainho@ibb.unesp.br
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
POINT-BY-POINT RESPONSES
REVIEWER 1: Jörg Ellinger
Major Compulsory Revisions
The authors used a conventional MSP (RASSF1A, RAR-beta and SFN) and a nested PCR (CDH1)
approach to detected hypermethylation at the sites of interest. The analysis were performed as
reported by others before, but I’m missing internal controls for the PCRs. Especially in case of SFN
and CDH1, which were methylated in most cancerous samples but also most normal tissue samples,
the use of an internal control is obligatory.
The authors should perform additional experiments to prove that their assay is specific (i.e. PCR
analysis of universal methylated DNA [either commercially available or by treatment of WBC DNA with
SssI methylase] and universal unmethylated DNA (either WBC DNA, or much better: unmethylated
DNA created by whole genome amplification as previously published by Umetani et al. 2005).
In our laboratory, we use Sss-I treated genomic DNA, frozen in aliquots, as a
constant reference sample for methylated DNA. Details of our protocol were published
previously (Caldeira et al., BMC Cancer, 2006, Mar 2;6:48). This information was included in
Methods section, page 9, lines 4-9, in the revised version of our manuscript.
We agree that the inclusion of the universal unmethylated DNA control in our MSP
assays is very important, specially because CDH1 and SFN genes are already methylated in
normal blood cells (Lombaerts et al., Infiltrating leukocytes confound the detection of Ecadherin promoter methylation in tumors. Biochem Biophys Res Commun 2004, 319:697704 and Bhatia et al. The tumor suppressor gene 14-3-3 sigma – SFN - is commonly
methylated in normal and malignant lymphoid cells. Cancer Epidemiol Biomarkers Prev.
2003; 12:165-9). As suggested, we have performed preliminary tests with whole amplificated
DNA by GenomiPhi DNA Amplification Kit (GE Healthcare) with the same primer set used in
CDH1 and SFN MSP analysis. After whole amplification of as little as 100pg of peripheral
blood DNA from volunteers, followed by bisulfite treatment of 2µg of amplified DNA, we have
detected just the unmethylated amplicon after MSP analysis. However, whole genomic
amplification of 10ng of starting peripheral blood DNA was enough to the co-amplification of
methylated alleles with specific primers for methylated template. Based on these results, we
believe that the MSP protocols used are very specific and sensitive.
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
The Results section is somewhat confusing. A revision is strongly required to make the results
become more understandable. I.e.:
- Subheadings to emphasize analysis of tissue and exfoliated cells.
This suggestion was accepted and included along of results section.
- I expect that a figure listing the methylation pattern of the samples in detail would be enormously
helpful. This figure could also display the methylation pattern during follow-up visits. And tissue and
exfoliated cell analysis could be parallelized.
This suggestion was accepted and we included a box in figure 2D showing the MSP
analysis of RARB and RASSF1A genes in matched TCC tissue and exfoliated cells as well
as the results of cytological evaluation of these exfoliated cells.
- Table 3: what about conventional urine cytology findings or lymph node involvement; is there any
correlation?
Table 3 details the MSP analysis of RASSF1A and RARB genes in a group of 49
fresh bladder tumors. Unfortunately, lymph node involvement was not evaluated at the
present study because this information was just available for a small number of cases. This
group of samples was collected prospectively and the post-surgical monitoring by cytology
analysis was not available.
- Table 4: I’m not sure whether the authors compared cancer tissue biopsies with exfoliated cell
analysis of healthy controls?!
Yes, cancer tissue biopsies were compared with exfoliated cells from the control
group (washouts from patients without bladder tumor). We changed the title of Table 4 to
make this information more clear.
Minor Essential Revisions
- Statistics: specify the software packages
We included this information on page 10, lanes 6-8.
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
- Recently, a multigene methylation analysis in urine sediments of patients with published (Yu et al.
2007); this study reports specificity of 100% and a sensitivity of 82% (but in a different set of analysed
genes), and should be discussed in the context of the present manuscript.
The results of Yu et al. (2007) were included in discussion section, page 17, lines
18-20. Interestingly, the cluster of reported methylation markers used in the U.S. bladder
cancers is distinctly different from that identified in this study, suggesting a possible
epigenetic disparity between the American and Chinese cases. These findings could not be
directly compared to our data, at a first moment, just because of the fact that we analyzed a
very smaller set of genes (4), restricting our possibilities of analysis. Just RASSF1A gene
was in common on both studies, and its hypermethylation was observed either for the control
group of Yu et al. evaluation.
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
REVIEWER 2: Carmen Marsit
Major Compulsory Revisions
The tables are supplied as supplementary materials, but are critical to the paper, and should be part
of the main document.
Tables are part of the manuscript, they are at the end of the document just because
of the structure asked for the magazine submission.
Why was a nested approach used for CDH1, but not for the other genes? Might this explain the
finding that all non-malignant samples were positive, as the sensitivity of this nested approach may be
far too high?
To achieve a much more sensitive MSP approach, in this study we used a nested-PCR
assay for the detection of CDH1 gene hypermethylation, similar to the conditions proposed
by Corn et al. (Clin Cancer Res 2001; 7:2765-9). This approach could explain, at least in
part, the higher frequencies of DNA methylation detected in the present samples. However, it
can’t be used as a single explanation for the finding that all non-malignant samples were
positive because a similar result was observed for SFN gene, where a nested PCR was not
used. Our results were based on a qualitative analysis (presence or absence of PCR
product) detected after electrophoresis of amplified products on polyacrylamide gels and
silver staining. These detection conditions are more sensitive than the process of
electrophoresis on agarose gels and ethidium bromide staining commonly used by others
authors. Furthermore, differences in the bisulfite modification protocols, especially the use of
urea which improves the efficiency of cytosine conversion (Paulin et al., Nucleic Acids Res
1998; 26:5009-10) could contribute to the high frequencies detected. In addition to these
experimental considerations, DNA methylation alterations may occur early in tumor formation
or in subpopulations of cells (normal or tumoral) in a given sample. Thus, genomic DNA from
tumoral tissues usually consists of a pool of molecules that may also display methylation
heterogeneity.
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
What is the overall sensitivity of the assays themselves? For example, what percent of methylated
substrate in a background of unmethylated substrate is detectable using the primers and conditions
described? If the assays are not being performed quantitatively, it is important to in some way justify
the assay sensitivity.
Some authors have described that conventional MSP analysis has the
sensitivity to detect methylated DNA molecules when they comprise as little as 5% of
the total complex DNA sample or one methylated allele in the presence of 1000-2000
unmethylated alleles.
Figure 1 is really not a very useful figure. First, it does not demonstrate any controls, and in fact, there
is no description of the use of positive or negative controls in the methods section. Was a fully
methylated control not used as a positive? How about for a negative, beyond a no template control?
The lack of these controls can also explain the findings, and must be done.
This issue has been discussed above, as questioned by the first reviewer.
Table 3 is very difficult to interpret. First, using U and M is confusing, as it can mean either methylation
negative vs. positive, respectively, or unmethylated and methylated bands (related to the PCR
primers). If it is the latter, some samples, I would bet, had both U and M bands (in fact, probably most,
if not all methylated samples showed both bands) so the interpretation is really difficult.
We agree that the use of U and M is confusing. Table 3 was completely reviewed and
we made some modifications in it. Just to elucidate, U and M were used to show methylation
negative vs. methylation positive, respectively.
I appreciate the calculation of an OR, but it would be more helpful is this was used as a multivariate
analysis (i.e. which of all of the demographics or clinical factors are truly associated with methylation
controlled for confounders). Maybe that is what these ORs represent, but that is unclear.
In Table 3 we had done a multiple analysis. Considering the parameters evaluated,
those that could contribute with a confounder effect (age and sex) did not exhibited it, since its
OR value were not significant. Hence, the factors investigated were truly associated with
methylation as presented bellow:
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
Analysis of confounders’ effect (age and sex) on parameters that truly are associated with
methylation.
Age
Variable
< 60 years
 60 years
Non-papillary
4
11
Papillary
5
28
Low
3
18
High
6
21
Noninvasive
5
25
Invasive
14
4
Absence
6
19
Presence
3
20
Sex
p value(1)
F
M
3
13
6
27
3
18
6
22
5
25
4
15
5
21
4
19
p value(1)
Growth pattern
0.432
1.000
Differentiation grade
0.712
0.714
Muscle invasion
0.711
0.720
Recorrência
0.466
1.000
(1) Fisher Exact test (α = 0.05).
Their presentation in the table is also unclear: what is the referent group?
The referent group was given in accordance with each parameter as follow:
-
age: < 60 years
-
sex: female
-
growth pattern: non-papillary
-
differentiation grade: low
-
muscle invasion: noninvasive
-
recurrence after the surgery: presence
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
Also, how was grade and stage dichotomized, as these are not clinically dichotomous measures?
Grade and stage were dichotomized in low and high, noninvasive and invasive,
respectively, because it is according to what was proposed by Epstein et al. (1998), who
dichotomized these measures in an effort to develop a universally acceptable classification
system for bladder neoplasia that could be used effectively by pathologists, urologists, and
oncologists. This classification is resulted of a consensus statement (Epstein JI, Amin MB,
Reuter VR, Mostofi FK: The World Health Organization/International Society of Urological
Pathology Consensus Classification of Urothelial (Transitional Cell) Neoplasms of the
Urinary Bladder. Bladder Consensus Conference Committee. Am J Surg Path 1998,
22:1435-1448.), as described on page 7, line 19.
In table 4, what do the ORs represent? It is unclear how those are calculated. Does the OR of 0.29 for
RASSF1A mean that having RASSF1A methylation is protective for bladder cancer?
The odds ratio (OR) is a way of comparing whether the odds of a certain event is the
same for two groups. An odds ratio of 1 implies that the event is equally likely in both groups.
An odds ratio greater than one implies that the event is more likely in the first (referent)
group. An odds ratio less than one implies that the event is less likely in the first (referent)
group.
Answering the question: an OR of 0.29 for RASSF1A implies that hypermethylation of
RASSF1A is less likely to occur in bladder cancer. However, it’s important to say that
hypermethylation of RARB is much more related to the development of tumor rather than it’s
presence for RASSF1A.
Table 5 is also problematic. In looking at the tumors, are the authors trying to say that methylation of
RASSF1A in a tumor will only predict a tumor with 83% sensitivity and 50% specificity? That is not an
appropriate comparison, as it is known that not all tumors will be methylated for each of these genes.
Maybe if a panel was used, that would be more appropriate.
According to Table 5, in our study, methylation of RARB in a tumor sample is able to
detect (not predict) the tumor presence with 83% sensitivity and 50% specificity. We agree
that the use of a panel of genes is more appropriate and we will abide this suggestion in next
studies.
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
In the discussion, this data does not prove anything towards the “seeds of methylation” hypothesis, as
the authors in no way have examined the density of methylation in the normal tissue compared to the
tumor counterpart. If anything, they have demonstrated that there is a field of epigenetic alteration in
histologically normal bladder.
We agree with this suggestion and we changed the manuscript (Discussion section,
page 14, lines 23-25.
Also, the discussion on aging is inappropriate, as they have in no way examined this issue in their
study.
We mentioned about aging because the mean age of the patients in our study was
67.85 years, which could be corroborative with what was proposed by Bornman et al. (2001)
for the hypermethylation observed in normal tissues of elderly man (CDH1 gene).
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br
Campus de Botucatu
REVIEWER 3: Sandra Mazzoli
Reviewer's report:
The Authors discuss widely and in complete way the DNA methilation patterns in bladder cancer and
washing cell sediments, with the aim to find possible new recurrence markers. Molecular biology
techniques applied to detection of DNA of differentially methylated genes should be included in follow
up of bladder cancer; this paper stress the importance to include new recurrence markers in patients
management. The discussion can be shortened, but substantially the paper is complete and
innovative. Discretionary revision.
Instituto de Biociências – Departamento de Genética
Distrito de Rubião Júnior s/n CEP 18618-000 Botucatu SP Brasil
Tel 14 3811 6229/6016 fax 14 3815 3131 genetica@ibb.unesp.br