Author's response to reviews

Transcription

1 Author's response to reviews Title: Hypermethylated sigma and ESR1 gene promoters in serum as candidate biomarkers for the diagnosis and treatment efficacy of breast cancer metastasis Authors: Mercedes Zurita Pedro C Lara (plara@dcc.ulpgc.es) Rosario del Moral (rdmoral@arrakis.es) Jose Luis Linares-Fernández (joseluislinares@ugr.es) Sandra Rios-Arrabal (sandrarioarrabal@hotmail.com) Joaquina Martínez-Galán (jmgalan22@hotmail.com) Blanca Torres (btortorres@hotmail.com) Francisco Javier Oliver (joliver@ipb.csic.es) José Mariano Ruiz de Almodóvar (jmrdar@ugr.es) Version: 5 Date: 5 April 2010 Author's response to reviews: see over

2 Dear Diana Marshall, We are pleased to attach our manuscript now entitled Hypermethylated sigma and ESR1 gene promoters in serum as candidate biomarkers for the diagnosis and treatment efficacy of breast cancer metastasis (your ref: ), which has been thoroughly revised to meet the concerns of your reviewers. We give our pointby point-responses to their comments below. We are grateful for their very helpful contributions. RESPONSE TO EDITOR 1) Abstract - please include context information in the Background section of the Abstract, in addition to the aims of the study. This has been corrected 2) We recommend that you copyedit the paper to improve the style of written English. If this is not possible, you may need to use a professional copyediting service. Examples are those provided by the Manuscript Presentation Service ( International Science Editing ( and English Manager Science Editing ( BioMed Central has no first-hand experience of these companies and can take no responsibility for the quality of their service. The paper has been copy-edited by a professional specialized (native English) translator 3) Manuscript sections - Please ensure that the manuscript sections are all in the correct order. Manuscript sections should include (in the following order): Abstract; Background; Methods; Results; Discussion; Conclusions; Abbreviations (if any); Competing interests; Authors' contributions; Acknowledgements; References; Figure legends (if any); Tables (if any); Description of Additional files (if any). The manuscript sections are in the correct order 4) Please also highlight (with 'tracked changes'/coloured/underlines/highlighted text) all changes made when revising the manuscript to make it easier for the Editors to give you a prompt decision on your manuscript. We have highlighted in red the changes introduced into the manuscript, which are also detailed in our responses to the reviewers 5) Please also ensure that your revised manuscript conforms to the journal style ( ). It is important that your files are correctly formatted. We have ensured that the manuscript conforms to the journal style RESPONSE TO REVIEWER 1 This is an interesting manuscript that could lead to a potential diagnostic test involving a novel biomarker for metastatic breast cancer. I have a number of 'minor essential revisions' that need to be addressed by the authors.

3 6) Results: last line page 7; Figure 3 shows 'all' MBC patients... and Figure 4 'all' patients with both... Please change these statements to 'Figure 3 shows 'those' MBC patients... and Figure 4 'those' patients with both...this has been corrected 7) page 8 in the same paragraph.. please clarify what you mean in the last line that the risefall pattern was associated with a 'positive predictive value of only 12%'. We have changed this paragraph, which now reads: Analysis with the Pearson chi-square test gave a value of (P = 0.017), indicating that the time course of the biomarker was determined by the clinical response to the treatment. In summary, the continuous-decline pattern of serum σ levels was associated with a positive predictive value of 65% (95% CI 38-86%), implying a favorable prognosis in two out of three patients, whereas the rise-and-fall pattern was associated with a negative predictive value of 88%, implying a poor prognosis for most of the patients with this pattern. 8) Discussion: There needs to be a better discussion regarding biological mechanism involving sigma. Specifically what is a role for stratifin in metastasis progression? Can you speculate (or is there literature evidence) concerning how long DNA from dead cells can circulate in the blood? We have addressed this point by adding the following paragraph, which contains new references: proteins are crucial in a wide variety of cell responses, including DNA damage checkpoints and apoptosis [27]. Disruption of the G2 M checkpoint also appears to contribute to the change in the sensitivity of cells to chemo- and radiotherapy [28, 29] sigma sequesters the cdc2-cyclin B1 complex in the cytoplasm, resulting in G2 arrest. Among the genes involved in the G2 M checkpoint, σ, a transcriptional target of p53, is frequently silenced by DNA methylation of the σ gene promoter or by induction of estrogen-responsive ubiquitin ligase that specifically targets σ for proteosomal degradation [24, 30]. The inactivation and reduced expression of σ have been reported in various cancers, including breast cancer [17, 27, 31]. To date, the sigma isoform of proteins has been the isoform most directly implicated in carcinogenesis and is recognized as a tumor-suppressor gene[32].although the molecular basis for the tumor-suppressor function of σ is unknown [27], it has been suggested that σ is a critical regulator of G2 M [33]. It has also been demonstrated that endogenous σ preferentially forms homodimers in cells [34]. Knocking out σ in cancer cells leads to mitotic catastrophe and cell death from DNA damage due to the absence of G2 M arrest [1]. Moreover, the highly conserved human gene family encodes proteins with either tumor-promoting or tumor-suppressing activities, suggesting that the cellular balance among different isoforms is crucial for the proper functioning of cells [33] proteins have been found in primary breast cancer, enhancing its biological activity [35]. The structure of the p53 C-terminus bound to the adaptor protein has been recently described, providing a rationale for the observed stabilizing effect of binding [36]. Consistent with these data, the G2 M checkpoint is impaired in cancer cell lines that show methylation of σ, while restoration of the expression of these genes using 5-aza-dC restores G2 M arrest induced by DNA damage [37]. This molecule also

4 contributes to mitotic catastrophe in carcinoma cells treated with chemotherapy agents [38]. Results of a recent study [39] showed that and HSP70 expression may be useful as biomarkers and targets for the diagnosis and treatment of human triplenegative breast cancer as metastasis-related proteins. Breast cancer metastasis is the main cause of treatment failure, and the goal of adjuvant therapy is to eliminate disseminated tumor cells after complete removal of the tumor. However no tool is available to monitor its efficacy [12]. Response to adjuvant treatment is usually evaluated retrospectively based on recurrence and survival rates. Therefore, the identification of metastasis biomarkers at an early stage may contribute to the early diagnosis and treatment of breast cancer patients. There is increasing recognition of the importance of epigenetic changes in the metastatic process. Cells may acquire an epigenotype that allows them to disseminate from the primary tumor mass or survive and proliferate at a secondary tissue site [40]. Overall, these results offer evidence of a difference in protein profile between metastatic and primary breast cancer. 9) Top of page 9... you state that 'It is possible that this epigenetic alteration changes in the specific environment and nutritional setting of breast cancer metastases.' I don't understand what you mean here. The sentence has been rewritten and now reads as follows: The expression profile of the metastatic tumor is known to differ between primary tumor and heterogeneous metastasis [39]. The present findings show that breast cancer methylation profiling might yield biomarkers for the diagnosis and treatment efficacy of breast cancer metastasis. Thus, we found that ROC analysis of serum levels of σ methylated gene-promoter discriminated between healthy individuals and metastatic breast cancer patients with a sensitivity of 81% (95% CI: ) and a specificity of 96.2% (95% CI: ), making this biomarker a candidate for use in metastasis screening in the follow-up of treated breast cancer patients. 10) last sentence of Discussion... this sentence is strange in that you seem to be giving evidence that your own work is invalid. Please clarify what your point is as a summation of your work. The sentence has been rewritten as follows to clarify that we are referring to studies by other authors. The above findings raise some questions about the utilization of σ gene promoter in DNA extracted from serum for metastasis screening in the follow-up of treated breast cancer patients. Further research is warranted to elucidate this issue and to establish the impact of σ on tumor progression and its potential to predict the response to treatment of metastatic breast cancer. 11) Figures: what are 'relative units'? Relative to what? We have addressed this point by adding the following paragraphs:

5 A sample of bisulfite-modified universally methylated DNA genome (CpGenomeTM Universal Methylated DNA, Intergen, New York, USA), treated in the same way as patient samples and adjusted after modification to 2 µg/ml (quantified spectrophotometrically), served as internal standard to prepare serial dilutions (from 1 to 1/128) with MiliQ water to construct a Standard Curve for Real-Time QMS-PCR. Each plate contained patient samples, serial dilutions of completely methylated DNA for constructing calibration curves, positive controls, and two wells with water blanks used as negative controls.. The fluorescence value after QMS-PCR in each sample was converted into units of universally methylated DNA (µg/ml), which we designated relative units, using the corresponding PCR Standard Curve obtained from the icycler iq software. Results obtained in a previous study [12] indicated that the method was valid for this investigation. RESPONSE TO REVIEWER 2 Zurita M et al. analyzed CpG island promoter methylation status at ESR1 and ? loci in the serum of 77 patients treated for breast cancer without evidence of disease and 34 patients with metastases by using a quantitative methylation specific PCR approach. Differences in promoter methylation were detected for ??in controls, disease free patients and between disease free and metastatic cancer group. Major Compulsory Revisions 12) The major point is the method used for determining methylation status in serum. The authors used an?absolute? quantification method with a standard curve construed with serial dilutions of a commercially available fully methylated DNA. This approach, however, does not take into consideration that the efficiency of bisulphate treatment in both control DNA and serum DNA samples cannot be 100%. The analysis of DNA methylation status by real time PCR requires a?relative? quantification approach. This method uses a reference gene sequence not containing CpGs (e.g ACTB promoter region) to normalize the amount of bisulphite modified DNA present in the samples (Jeronimo et al, Clin cancer Res 2003, Harden et al J Urol 2003, Hoque et al J Natl Canc Inst 2006, Hoque et al Canc Epidem Biom Prev, 2009, Widschwendter A Clin cancer Res 2004). In all figures authors refer to?relative units? of methylation, but they do not explain in the paper either in Materials and Methods nor in Figures legends how this measure is determined. We have addressed this point by adding the following paragraphs: Efficiency of DNA recovery after bisulfite modification was around 55% (data not shown). One microliter of the recovered bisulfite-treated DNA was used in each well for SYBR green reaction. Modified DNA of standards and samples are stable for at least 2 months at -80 C. A sample of bisulfite-modified universally methylated DNA genome (CpGenomeTM Universal Methylated DNA, Intergen, New York, USA), treated in the same way as patient samples

6 and adjusted after modification to 2 µg/ml (quantified spectrophotometrically), served as internal standard to prepare serial dilutions (from 1 to 1/128) with MiliQ water to construct a Standard Curve for Real-Time QMS-PCR. Each plate contained patient samples, serial dilutions of completely methylated DNA for constructing calibration curves, positive controls, and two wells with water blanks used as negative controls. In all cases, correlation coefficients for the calibration curves were higher than 0.98, slopes ranged from 3.2 to 3.4, and PCR efficiencies were around 100%. The reaction mixture contained 1 µl of modified serum DNA of each standard or unknown sample as template for real-time QMS-PCR, 0.5 µm of each oligonucleotide primer, 12.5 µl of 2x SYBR Green Supermix (Bio-rad), and sterile water. All PCR experiments were performed in a volume of 25 µl with 96-well plates. Primer sequences were obtained from previously published data for stratifin ( σ) [21] and estrogen receptor-α (ESR1) [22]. The fluorescence signal of the quantitative methylation-specific PCR was generated by SYBR Green Super Mix (BioRad, Hercules, CA). PCR amplification was done by using a previously reported procedure [12]. The fluorescence value after QMS-PCR in each sample was converted into units of universally methylated DNA (µg/ml), which we designated relative units, using the corresponding PCR Standard Curve obtained from the icycler iq software. Results obtained in a previous study [12] indicated that the method was valid for this investigation. 13) In the results sections authors refer to?mean serum levels?, however they used for statistical analyses non-parametric tests assuming thus samples should be described by medians and InterQuartile Ranges (IQR). It would also be better to show sample distribution in Figures using box-plot instead than scatter plots. It is clarified in Results that we refer to median values. The Figures have been corrected, and a box-plot is now used. 14) It is not clear why to measure the differences between DFBC and MBC group was used the Mann Whitney test instead than Dunn Test as it was done for others multiple comparisons. If authors use the Mann Whitney they need to make the correction for multiple comparisons whereas the Dunn test already takes into considerations multiple comparisons. We accept this comment and have used the proposed analyses. The paragraph has been rewritten accordingly: The Kruskal-Wallis test was used to study differences among groups (DFBC, MBC, and HC), and the Dunn s multiple comparison test was used for paired comparisons when results were significant 15) In the Materials and methods statistical analysis is not describe as the Z and IC value for ROC curve analysis was calculated. We have included a new paragraph to clarify this question: ROC curves: The distributions of the σ ratio values corresponding to different patient populations are, at least in part, overlapped. As a consequence,

7 the number of correct predictions of the measurable response (true positive (TP) cases identified by means of the test depends on the threshold level selected. Using different thresholds, it is possible to obtain successive pairs of false positive (FP) and TP values that can be plotted as a curve. A useful numerical parameter arising from this graph is the proportion of ROC space that lies below the ROC curve (Z). 16) Table 2 is not clearly described. The pattern of is defined empirically? Moreover the Mantel-Henzel would be more appropriate than?2 to analyze the data presented because they represent a scale (from good to worst behaviour). We considered using the Mantel-Henzel test but were advised that the chi-square test was the most appropriate approach. We have included the following new paragraph to address the other points raised. However, although initial levels appear to fall in most of the MBC patients, other patients show no change or an increase in levels. Figure 3 shows those MBC patients with a continuous decline in serum σ, and Figure 4 those patients with both rises and falls. This biomarker-based categorization has been empirically defined. Table 2 shows the contingency table that was constructed by combining this biomarker s response-pattern with the chemotherapy response scores.