Author's response to reviews Title: Estimation of groin recurrence risk in patients with squamous cell vulvar carcinoma by the assessment of marker gene expression in the lymph nodes. Authors: Magdalena Kowalewska (magdak@coi.waw.pl) Jakub Radziszewski (radziszewski@coi.waw.pl) Krzysztof Goryca (kgoryca@gazeta.pl) Mateusz Bujko (mbujko@coi.waw.pl) Malgorzata Oczko-Wojciechowska (gosiaoczko@io.gliwice.pl) Michal Jarzab (mjarzab@io.gliwice.pl) Janusz A Siedlecki (jas@coi.waw.pl) Mariusz Bidzinski (mbidzinski@coi.waw.pl) Version: 5 Date: 23 March 2012 Author's response to reviews: see over
Magdalena Kowalewska Dept. of Molecular Biology Cancer Centre and Institute of Oncology Roentgena 5, 02-781 Warsaw, Poland e-mail: magdak@coi.waw.pl Warsaw, March 23 rd, 2012 Journal Editorial Office BMC Cancer Dr. Danrolf de Jesus Dear Dr. de Jesus, M. Kowalewska, J. Radziszewski, K. Goryca, M. Bujko, M. Oczko-Wojciechowska, M. Jarzab, J. A. Siedlecki and M. Bidzinski. Estimation of groin recurrence risk in patients with squamous cell vulvar carcinoma by the assessment of marker gene expression in the lymph nodes. We thank you very much for your letter and for the reviews of the above manuscript. Please find enclosed a revised version of the manuscript. The answers and comments to the specific reviewers questions are as follows: Answers to REVIEWER 2 (Dr. Matthias Choschzick): a) In the discussion section the authors added the sentences Importantly, our microarray data were obtained for HPV-negative patients in order to exclude the interference of the infection with the expression results and to focus on metastasis-associated genes selection. However, patients with HPV infection should also be enrolled in future studies to assess the influence of HPV on the markers performance. Indeed, HPV-association of vulvar carcinomas surely influences the gene expression profile. For instance, HPV positivity of vulvar cancer is negative connected to the p53 mutation status and the occurrence of EGFR amplification. The authors should at least state throughout the manuscript including the title, the abstract and the conclusions that there results are only reliable for HPV negative patients. Furthermore, in 4 cases the HPV status was not determined and one HPV16 positive tumor was included in the study. These cases should be eliminated from the study if the authors think that the HPV status is important for the gene expression profile. The sentence in the 10 th paragraph of the Discussion section says that our microarray data were obtained for HPV-negative patients in order to exclude the interference of the infection 1
with the expression results and to focus on metastasis-associated genes selection. Indeed, we think that the HPV status is important for the gene expression profile. Accordingly, at first we searched for metastasis-specific genes purposely in the lymph nodes of HPV-negative patients in order not to clear out the possible distortion of our results by this infectious agent. Next, after these putative metastasis-specific markers were identified, some of them were chosen to be validated in subsequent qpcr analysis. As both HPV-negative and positive vulvar tumours metastase, this qpcr analysis was performed for all of the patients enrolled in our study. In our opinion, the last sentence in the 10 th paragraph of the Discussion section which reads However, patients with HPV infection should also be enrolled in future studies to assess the influence of HPV on the markers performance already shows the doubts raised by the Reviewer 2. As the reported HPV-positivity in vulvar carcinoma ranges widely (approximately from 5 to 70% 1 ) and its prognostic significance remains uncertain (see the point below), we decided not to introduce any more additional changes into our manuscript. The authors state in their answer: The results only mirror the known adverse prognostic effect of lymph node status in vulvar cancer: This is most probably true. In my view the same prognostic results with can be achieved with other tumor-associated genes like keratins or p53 without the laborious gene selection process in the presented study. To our knowledge, it is widely accepted that TP53 protein over-expression is caused by TP53 mutations and is strongly associated with the absence of HPV. Thus, TP53 over-expression evidences the HPV-independent pathway of vulvar carcinogenesis which observed in older patients with vulvar carcinoma. There are conflicting opinions on the relationship between HPV status and prognosis in vulvar carcinoma. Some studies report the same prognosis for vulvar cancers containing HPV DNA as those without 2, whereas other authors showed favorable prognosis for patients with HPV-harboring carcinomas. 3 The generally less favorable prognosis for older cancer patients with frequent comorbidities may contribute to 1 IARC 2005:90; Hampl et al. Obstet Gynecol. 2006;108:1361-1368.; Sutton et al. Mod Pathol. 2008;21:345-354. 2 Alonso et al. Gynecol Oncol. 2011;122:509-14.; Pinto et al. Gynecol Oncol. 2004;92:856-865.; Hording et al. Int J Cancer. 1993;55:394-396; Andersen et al. Am J Obstet Gynecol. 1991;165:329-335.; Bloss et al. Hum Pathol. 1991;22:711-718. 3 Sutton et al. Mod Pathol. 2008;21:345-354.; Ansink et al. Gynecol Oncol. 1994;52:180-184.; Monk et al. Obstet Gynecol. 1995;85:709-715.; Knopp et al. Am J Clin Pathol. 2006;126:266-276. 2
the latter findings. Surprisingly (in light of the above mentioned opinion of the Reviewer 2), in their recent work Choschzick et al. 4 find no relationships between TP53 mutation status and tumor stage, grading, nodal status, depth of invasion, or patient prognosis. Keratins, on the other hand, when examined immunohistochemically have diagnostic application, not prognostic to our knowledge. According to our knowledge, again, no pathological markers of prognostic significance exist. This is a reason why they have not yet earned a place in standard clinical diagnostics or treatment, as Knopp et al. 5 stated not so long ago. Moreover, none of the pathological markers excludes the necessity to omit the laborious procedure of lymph node multisectioning which improves micrometastases detection. When searching for novel markers, the point of view is quite different to this of the Reviewer 2. Even if the currently available markers would be good, it would not mean that we should not search for the better ones. Although new and laborious marker selection methods do not guarantee identifying a better marker/s, they offer some hope for achieving their goals. Therefore, it is still worthwhile to continue searching. Likewise, an example of the TP53 and keratin findings showed the laborious research that preceded these findings was worth the effort. Beside this, the number of examined cases is probably (surely) too low for prognostic statements. The results of the estimation of the effect of gene expression levels on the time to groin recurrence using Cox proportional-hazard model with Bonferroni correction for multiple hypotheses testing as well as two sided log-rank analysis used to compare Kaplan-Meier curves testify against the above mentioned opinion of the Reveiewer 2. The studied group is small indeed, but the results are statistically significant. As the threshold of statistical significance has been reached in our analysis, the conclusions regarding predictive value of our findings for the selected genes are eligible. Therefore, although the studied group is small, it is large enough to observe significant differences in the time to groin recurrences between patients with high vs low gene expression levels. As a prerequisite to the application of our knowledge on the confirmed prognostic value of the selected genes expression levels in the clinical practice, further analysis - including larger patient numbers - is required. 4 Int J Gynecol Pathol. 2011;30:497-504. 5 J Clin Pathol. 2009;62:212-8. 3
b) In their conclusions the authors write may provide a promising tool for intra-operative sentinel LN evaluation in VC patients. All these statements about the value of the applied methods for detection of lymph node metastases in comparison to histopathological examination should be eliminated from the study because the authors provide no data that their method is more reliable than pathology. Otherwise they have to provide a detailed analysis of sensitivity and specificity of their analytical method in comparison to pathological examination of lymph nodes especially the ultra-staging of groin lymph nodes. The proposal (see conclusions) to apply these methods intra-operative is also not justified by the presented results. We agree with the Reviewer 2 that we do not provide evidence that our method is more reliable than pathology. In the Discussion section, we just list some of the advantages of qrt- PCR-based tests over the histopahological means of lymph node assessment. These advantages are the reason for the use of such tests in general, as well as for seeking appropriate qrt-pcr bimomarkers, which was an attempt that we made in our work. We have not performed controlled, double-blinded, multi-centre study that would constitute a proper basis for a proposal to supplement/alter clinical practice. Therefore, in the second sentence of the two sentences written in the Conclsions section, we state that before their [markers] incorporation into clinical setting further studies are necessary to confirm their prognostic value in the qrt-pcr assays. It must be stressed that the lack of a gold standard for routine diagnostic purposes when examining lymph nodes of vulvar carcinoma patients and limitations of the microscopic evaluation of H&E-stained lymph node sections (briefly listed in our manuscript) impede a sound comparison of our method with histopathological examination. A comparison of any novel methodology with any doubtful/insufficient routine diagnostic methodology is unlikely to provide informative results. The disadvantages of our work were listed in 10 th paragraph of the Discussion secion; these include lack of ultra-staging and too limited sample size to enable the ROC curve analysis. To sum up, we believe our work advocates future investigation of qrt-pcr capabilities in vulvar carcinoma diagnostics. 4
c) I agree with the comment of the first reviewer (Omer Devaja) My oppinion is that title is a bit to far fetch. I am not covinced that expression of described marker genes can relaiable predict lymph nodal recurrence. The manuscript title has been changed according to the previous requirement of the Reviewer 1. Nevertheless, the statistical significance of the impact of gene expression levels in lymph node samples on time to recurrence (Cox's proportional hazard analysis, the log-rank test) remains valid. Please, refer to Point 8 of our answers to the Reviewer 1 in the previous covering letter. Answers to REVIEWER 3 (Dr. Marta Ewa Polak): Major compulsory revisions Although the autors explain the reasons for choosing to do in-depth analysis of one set of microarray (i.e. the pair of lymph nodes from the patient with rapidly progressing VC) it is possible, that concentrating on verification of the candidate genes chosen from that sample only they miss an opportunity to identify other valid and useful marker genes. In particular, it is possible that rapidly progressing cancer utilises different metabolic and signalling pathways than less aggressive tumours, therefore I feel it is essential to analyse the available data sets in full and discuss the results. The PUMA analysis of microarray data was performed for the four pairs of lymph nodes obtained from the four vulvar carcinoma patients. The genes that were up-regulated - according to the results of PUMA analysis - in the four metastatic lymph nodes of these four patients were subjected to subsequent qpcr analysis. Therefore, we cannot state that we missed identifying any up-regulated genes by using this methodology. In fact, patient No. 15 progressed rapidly and eventually died of cancer. However, it must noted that the four vulvar cancers from which the lymph node samples were obtained for a search of metastasis-specific genes in the initial microarray experiment were quite different in terms of the rate of disease progression. During the follow-up period, both patient No. 21 and patient 61 experienced groin recurrences, while patient No. 46 did not (please refer to Table 1 of our manuscript). Patients No. 15 and 21 experienced local recurrences, while patients 46 and 61 did not. The three patients No. 21, 46 and 61 were alive at the end of the follow-up period. Thus, we cannot say that we could have missed metastasis-associated genes specific for slower disease progression, as the lymph node samples from such patients 5
were taken into account in microarray analysis. We do hope that this information will dispel doubts raised by the Reviewer No. 3. We must regretfully admit that there were more genes indentified by the PUMA analysis jointly in metastatic lymph nodes of these four patients than just these seven genes which was further verified by qpcr. However, a list of these differentially expressed genes is provided in the Supplementary Table 3 and may be a useful source of information for future studies on vulvar carcinogenesis. Contrary to the suggestion of the Reviewer 3, the available data sets were analysed in full. Moreover, our first microarray data analysis comparing the overall gene signature of the metastatic lymph nodes with overall expression profile of the metastasis-free lymph nodes did not reveal any gene whose expression would distinguish between metastatic and histologically-negative nodes (data not shown). Obviously, it would be much better to analyse more samples obtained from patients with vulvar carcinoma in various disease stages. However, we are satisfied with our results achieved at relatively low cost that were not only concordant with the available databases but also showed significant differences in time to recurrence between patients with low and high gene expression levels for the four out of the seven genes examined. I am looking forward to hearing from you soon, Faithfully yours, Magdalena Kowalewska 6