- Email: [email protected]
A Specific miRNA Signature Correlates With Complete Pathological Response to Neoadjuvant Chemoradiotherapy in Locally Advanced Rectal Cancer
International Journal of
Radiation Oncology biology
physics
www.redjournal.org
Clinical Investigation: Gastrointestinal Cancer
A Specific miRNA Signature Correlates With Complete Pathological Response to Neoadjuvant Chemoradiotherapy in Locally Advanced Rectal Cancer Giuseppina Della Vittoria Scarpati, M.D., Ph.D.,*,1 Francesca Falcetta, Ph.D.,y,1 Chiara Carlomagno, M.D., Ph.D.,* Paolo Ubezio, Ph.D.,y Sergio Marchini, Ph.D.,y Alfonso De Stefano, M.D.,* Vijay Kumar Singh, Ph.D.,z Maurizio D’Incalci, M.D.,y Sabino De Placido, M.D.,* and Stefano Pepe, M.D., Ph.D.x From the *Department of Molecular and Clinical Endocrinology and Oncology, University of Naples Federico II, Naples, Italy; yLaboratory of Cancer Pharmacology, Department of Oncology, “Mario Negri” Institute for Pharmacological Research, Milan, Italy; zCancer Genomics Laboratory, Fondazione “Edo ed Elvo Tempia Valenta”, Biella, Italy; and x Division of Oncology, University of Salerno, Italy Received Jul 7, 2011, and in revised form Sep 8, 2011. Accepted for publication Sep 9, 2011
Summary The study evaluated the miRNA expression in tumor biopsies of patients with rectal cancer, in order to identify a specific “signature” correlating with pathological complete response (pCR) after neoadjuvant chemo-radiotherapy. A set of 13 miRNAs were found which bore this strong association and which thus have considerable clinical potential to predict response in rectal cancer patients.
Purpose: MicroRNAs (miRNAs) are small, noncoding RNA molecules that can be down- or upregulated in colorectal cancer and have been associated to prognosis and response to treatment. We studied miRNA expression in tumor biopsies of patients with rectal cancer to identify a specific “signature” correlating with pathological complete response (pCR) after neoadjuvant chemoradiotherapy. Methods and Materials: A total of 38 T3e4/Nþ rectal cancer patients received capecitabine-oxaliplatin and radiotherapy followed by surgery. Pathologic response was scored according to the Mandard TRG scale. MiRNA expression was analyzed by microarray and confirmed by real-time Reverse Transcription Polymerase Chain Reaction (qRTPCR) on frozen biopsies obtained before treatment. The correlation between miRNA expression and TRG, coded as TRG1 (pCR) vs. TRG >1 (no pCR), was assessed by methods specifically designed for this study. Results: Microarray analysis selected 14 miRNAs as being differentially expressed in TRG1 patients, and 13 were confirmed by qRT-PCR: 11 miRNAs (miR-1183, miR-483-5p, miR-622, miR-125a-3p, miR-1224-5p, miR-188-5p, miR-1471, miR-671-5p, miR-1909*, miR-630, miR-765) were significantly upregulated in TRG1 patients, 2 (miR-1274b, miR-720) were downexpressed. MiR-622 and miR-630 had a 100% sensitivity and specificity in selecting TRG1 cases.
Reprint requests to: Chiara Carlomagno, M.D., Ph.D., Dipartimento di Endocrinologia ed Oncologia Molecolare e Clinica, Universita` di Napoli Federico II, Via Sergio Pansini 5; 80131 Naples, Italy. Tel: þ390817464271; Fax: þ390812203147; E-mail: [email protected] Conflict of interest: none. Int J Radiation Oncol Biol Phys, Vol. 83, No. 4, pp. 1113e1119, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2011.09.030
Supplementary material for this article can be found at www.redjournal.org. 1 Contributed equally to this study. AcknowledgmentsdWe thank Jean Ann Gilder (Scientific Communication srl) for language editing.
International Journal of Radiation Oncology Biology Physics
1114 Della Vittoria Scarpati et al.
Conclusions: A set of 13 miRNAs is strongly associated with pCR and may represent a specific predictor of response to chemoradiotherapy in rectal cancer patients. Ó 2012 Elsevier Inc. Keywords: miRNA, Predictive factors, Preoperative chemoradiotherapy, Rectal cancer
Introduction MicroRNAs (miRNAs) are small, noncoding sequences, approximately 18e25 nucleotides long, that are post-transcriptional regulators of gene expression. They bind to complementary sequences on messenger RNA transcripts (mRNAs), usually resulting in translational repression and gene silencing. Aberrant expression of miRNAs is implicated in numerous diseases and miRNA misregulation is also linked to many types of cancer (1, 2). Depending on the genes they regulate, miRNAs can function as either oncogenes or tumor suppressors (3). Several studies showed that over- or underexpression of specific miRNAs is associated with colorectal cancer (CRC) development and progression being differently expressed in tumor cells and in normal mucosa (4e6) and in early- versus late-stage neoplasms (5, 6). Moreover, specific miRNAs aree significantly correlated with disease-free and overall survival in patients with early colon cancer (7e9) and with progression-free and overall survival in patients with metastatic CRC (9, 10). An important challenge in medical oncology is to identify the patient- or tumor-related characteristics responsible for the response to conventional or targeted drugs. In vitro experiments suggest that several miRNAs can affect sensitivity of different human tumor cell lines to chemotherapeutic agents (11). CRC is the third most frequent neoplasia in men and the second in women (12). About one-third of large bowel tumors are rectal cancers. Neoadjuvant chemoradiation followed by surgery is now the standard treatment for TNM Stage IIeIII rectal cancer. Several retrospective analyses suggest that the pathological stage of disease after neoadjuvant treatment has a significant prognostic impact on disease-free and overall survival. In particular, the subgroup of patients who achieve a complete pathological response (pCR) has a very low risk of local or distant recurrence (13, 14). This observation prompts the evaluation of chemoradiotherapy regimens that induce high response rates and the profiling of tumors that will or will not respond to specific treatments. To address this topic, we studied the expression profile of miRNAs in biopsies of patients with locally advanced rectal cancer treated with neoadjuvant chemoradiation in the attempt to identify a specific signature associated with a pCR to preoperative treatment.
Methods and Materials
patient refused surgery; however, the disease was not detectable by rectal endoscopy and multiple biopsies in the zone of the previous mass resulted negative. Surgical specimens were evaluated according to the College of American Pathologists protocol for invasive carcinomas (TNM, 6th ed.). Pathologic response was scored according to tumor regression grade (TRG) as described by Mandard (15). Details of treatment and clinical results are reported elsewhere (16).
Tissue processing Fresh biopsies of rectal tumors were collected between January 2005 and December 2007 during staging colonoscopy, frozen within 5 min in liquid nitrogen, and stored at 80 C. The collection and the use of tumor samples for research purposes was approved by the local ethical committee and a written consent was obtained from the patients. Investigators performing microRNA analysis were blind to the clinical outcome of patients.
miRNAs isolation and microarray analysis Starting in January 2010, frozen tissues (100 mg) were homogenized at 8000 rpm in a Ultraturrax T-10 at 4 C with 1 mL of Trizol until a totally liquid phase. Description of total RNA extraction and microarray assay has been performed as described elsewhere (17). Median background and fluorescence values of each spot were processed and normalized as described in the supplementary data.
Microarray data analysis Patients were divided in two groups: group A (pCR [TRG 1] including the nonoperated patient without detectable disease) and group B (any pathologic response other than complete). Because of the relatively low number of cases, we strengthen the reliability of the difference in miRNA expression between the two groups by applying six different methods of analysis. miRNAs found differentially expressed by most methods were selected for subsequent qRT-PCR (real-time Reverse Transcription Polymerase Chain Reaction) validation.
Patients and treatment
Methods I and II
Thirty-eight patients with a histologic diagnosis of rectal adenocarcinoma invading through the intestinal wall and/or with lymph node involvement as evaluated by endorectal ultrasonography (uT3-T4 and/or uNþ) were treated with neoadjuvant chemoradiotherapy (capecitabine þ oxaliplatin in combination with 45 Gy of pelvic conformal radiotherapy). Six to 8 weeks after radiotherapy completion, 37 patients underwent surgery. One
Raw data were preprocessed with (Method I) or without (Method II) background subtraction and normalized by the invariant method, implemented in R programming language (version 2.11) and the Bioconductor software suite (version 2.6). The statistical Welch t test was applied and the p values were adjusted for multiple testing by the Benjamini and Hochberg False Discovery Rate method (18); a significance level of 0.1 was used for statistical testing.
Volume 83 Number 4 2012
microRNA profiling and response to chemoradiotherapy 1115
Methods III and IV Raw data were preprocessed and normalized as described in the section “Microarray data preprocessing” (supplementary data). The statistical Welch t test was applied and the p values were calculated for each probe. Method III relies on an “miRNA adjusted p value,” which was calculated taking the median the p values of its probes and adjusting with the false discovery rate (FDR) method. A significance level of 0.05 was used for statistical testing. Considering that each probe represents a different test for a single miRNA, which has distinct binding affinities, the probability that a difference of expression of a miRNA between the two groups was due to chance is much reduced if the p values of all its probes were low. Therefore, in method IV, we calculated a score (scmiRNA) associated to an miRNA combining the p values of its probes with the formula: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p sc ZNpi p .p miRNAi
probe1i
probe2i
probeNpi
where Npi is the number of probes (1e4) of miRNA “i” and pprobe1i, pprobe2i . are the calculated p values of its probes. According to method IV, a miRNA was considered differentially expressed if scmiRNA 0.01.
from 1 mg of mirVana purified total RNA (Ambion-ABI, Milan Italy), using the miScriptÒ Reverse Transcription Kit, following the manufacturer’s instruction (Qiagen). miRNA expression was quantified by Syber Green chemistry using the commercial miScriptÒ Universal Primer, together with the miScriptÒ Primer Assay (Qiagen). Experiments were run in triplicate for each case, using 384-well reaction plates in an automatic liquid handling station (epMotion 5075LH; Eppendorf, Milan, Italy). qRT-PCR was done on an Applied Biosystems 7900HT apparatus (Ambion-ABI). Raw data were generated with SDS Relative Quantification software (version 2.3; Ambion-ABI). Fluorescence intensities for each miRNA were normalized against the normalization index calculated on RNU-6b used as housekeeping genes. To calculate the relative expression for each miRNA, we used the 2-DCT method to average the threshold cycle values of the three replicates for a single gene. For miRNA expression experiments, median values were compared using the nonparametric MannWhitney t test. Differences were considered statistically significant with a two-sided p < 0.05. Data are reported as medians with interquartile range (range, 25e75% percentile). All tests were done using GraphPad Prism Version 5.01 (GraphPad Software, La Jolla, CA).
Methods V and VI Raw data were preprocessed and normalized as described in the section “Microarray data preprocessing” (supplementary data). The normalized database was then split in two subsets (DBS1 and DBS2). While dividing the dataset in two subsets, the proportion of group A to group B patients in each subsets was kept approximately same. DBS1 included 4 group A and 14 group B patients, whereas DBs2 included 5 group A and 15 group B patients. The statistical Welch t test was applied and the p values were calculated for each probe separately within DBS1 and DBS2. According to method V, a miRNA was deemed differentially expressed if at least one of its probes was over- or underexpressed in both subsets with a p value <0.05. miRNA p values were then calculated separately in DBS1 and DBS2 taking the median of the p values of the probes of each miRNA. A miRNA was deemed differentially expressed when its p value was <0.1 in both subsets (method VI), meaning that the joint probability that a cross-validated difference was due to chance was 0.1 0.1 Z 0.01. Methods IIIeVI were fully implemented, including preprocessing of data, in Microsoft Excel.
Results The characteristics of the 38 patients are listed in Table 1. Thirty-seven patients underwent surgery; a pCR was achieved in nine cases. As of Dec. 31, 2010, 11 patients recurred, and 7 had died.
miRNA microarray analysis
The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE29298 (http://www.ncbi.nlm. nih.gov/geo/query/acc.cgi?accZGSE29298). Cluster analysis was done on the 53 miRNAs identified by the first selection of the differential expression analysis. Hierarchical dendrograms and matrix correlation were calculated with Pearson’s correlation as clustering metric. The calculation was done with the TIGR Multi-experiment Viewer feature of TM4 software suite and Microsoft Excel.
Of the 373 miRNAs with at least one probe of one array significantly higher than nonspecific fluorescence, 53 were differentially expressed in group A vs. B by at least two criteria (Table S1). This indicates that 17.4% of the miRNAs present in these tumors were differently expressed in good responders (group A) with respect to all the other patients. The trends of expression of most of these 53 miRNAs were well/highly correlated, which suggests the existence of a large cluster of interrelated miRNAs that are expressed at lower levels in less responding patients (Fig. 1). Only four miRNAs were more highly expressed in group B vs. group A. We used more stringent criteria to select a subset of these 53 miRNAs for qRT-PCR validation. For this purpose, we first excluded miRNAs with a low average expression (i.e., log fluorescence intensity <5 in both groups) and kept those differentially expressed by at least four methods. Then, we selected for qRT-PCR validation the 12 overexpressed miRNAs in group A with the highest fold change, and the only two miRNA with increasing expression in group B. Fold change and p values of these 14 miRNAs are reported in Table 2. Three other miRNAs, not differentially expressed by array analysis (negative controls), were measured by qRT-PCR.
Validation of the microarray results by qRT-PCR
miRNA signature validation by qRT-PCR
Mature miRNA expression levels were examined by RT-PCR using dedicated sets of commercial primers (Qiagen, Milan, Italy). Briefly, specific cDNA was generated in a single step reaction
Table 2 shows the median expression level of the qRT-PCReselected miRNA and the 25e75% range in group A compared withgroup B. Thirteen of the 14 selected miRNAs were
miRNA correlation and cluster analysis
International Journal of Radiation Oncology Biology Physics
1116 Della Vittoria Scarpati et al. Table 1
Patient and tumor characteristics n
Gender Males 25 Females 13 Age Median (range) 64.4 (42.7e79.3) Surgery Yes 37 No 1 Tumor regression grade (TRG; n Z 37) TRG1 8* TRG2 16 TRG3 10 TRG4 3 First site of recurrence Any 11 Local 1 Distant 10 Local þ distant 0
% 66.8 34.2
97.4 2.6 21.6 43.2 27.0 8.1 28.9 2.6 26.3 e
* 9 pathologically complete responses, including the patient not operated.
significantly differentially expressed in the two groups, whereas the viral miRNA (hsv1-miR-H1) was not found by qRT-PCR. The three negative controls were not differentially expressed, which is
coherent with the array analysis. Eleven miRNAs (miR-1183, miR483e5p, miR-622, miR-125a-3p, miR-1224e5p, miR-188e5p, miR-1471, miR-671e5p, miR-1909), miR-630, and miR-765) were upregulated, whereas two miRNAs (miR-1274b and miR720) were downregulated in group A vs. group B (Table 2). Considering the average of the medians in groups A and B as a cutoff, we built the heat map shown in Fig. 2. Each of the 13 differently expressed miRNAs individually had a high specificity and sensitivity, which reached 100% in the case of miR-622 and miR630. Considering the 13 miRNAs collectively, all group A patients had a concordance of at least 10/13 with the signature of group A, whereas no group B patient reached a score higher than 5/13.
Discussion Surgery is the most effective treatment for rectal cancer. However, in locally advanced stages, radiotherapy and chemotherapy reduce the risk of local and distant relapse. Consequently, a multidisciplinary approach is currently the standard. Complete pathologic disappearance of tumor cells after neoadjuvant chemoradiotherapy occurs in about 20% of patients and is correlated with a very good long-term prognosis. In the era of customized treatment, based on the assessment of predictive factors, we should exploit new techniques that can identify predictive factors for conventional treatments, such as chemo- or radiation therapy. The ultimate goal is to select for each patient the most effective and less toxic treatment.
Fig. 1. Correlation matrix and cluster analysis. Each row and column represents the 53 micro (mi)RNAs found to be differentially expressed by at least two criteria. The colors indicate the Pearson’s correlation between miRNAs; the value of the correlation index corresponding to each colour is shown. The two main clusters are constituted by miRNAs that are expressed at a higher level in group A (A > B) or in group B (B > A).
Volume 83 Number 4 2012 Table 2
microRNA profiling and response to chemoradiotherapy 1117
Expression of the 14 selected miRNA by microarray and qRT-PCR Microarray
miRNA
Ra
p
hsa-miR-1183 hsa-miR-1274b hsa-miR-483e5p hsa-miR-622 hsa-miR-125a-3p hsa-miR-1224e5p hsa-miR-188e5p hsa-miR-1471 hsa-miR-671e5p hsa-miR-1909* hsa-miR-630 hsa-miR-720 hsa-miR-765 hsv1-miR-H1
3.11 0.60 2.41 3.10 2.20 2.31 2.31 3.24 2.31 2.88 2.43 0.57 3.02 3.61
0.005 0.054 0.006 0.008 0.006 0.007 0.005 0.019 0.013 0.042 0.017 0.032 0.008 0.005
Not differentially expressed hsa-miR-30a* 1.46 0.628 hsa-miR-92a 0.80 0.705 hsa-miR-34a 1.13 0.753
qRT-PCR Group A* 4295 217500 7061 4275 323 3275 1753 4981 2757 4278 88.7 120900 5979 395.4
(3162e8158) (169000e315900) (5291e9534) (3363e5504) (281e538.5) (2101e5093) (1432e2593) (3539e8410) (2334e5880) (3189e8054) (71.35e118.6) (84440e203600) (3459e6972) (114.3e1465)
48.7 (22.5e330) 13150 (9295e58870) 4180 (1277e18130)
Group B* 823 550000 656 617 88.86 704 378 1023 1170 1125 12.66 580000 1276 158
(517e1730) (296600e751500) (305e1700) (250e935.5) (53.23e140.9) (348e977) (223e553.5) (757e2668) (582.5e1571) (707e2207) (8.861e22.3) (300700e1256000) (949e2307) (105.5e322.5)
32.41 (16.77e74.85) 11400 (4763e15550) 1910 (886e4271)
R
p
5.22 0.40 10.76 6.93 3.63 4.65 4.64 4.87 2.36 3.80 7.01 0.21 4.69 2.50
<0.0001 0.018 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.0006 <0.0001 0.0002 <0.0001 <0.0001 0.0004 0.131
1.50 1.15 2.19
0.216 0.114 0.099
Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed confirmed Confirmed Confirmed Not confirmed Confirmed Confirmed Confirmed
Abbreviations: p Z Mann-Whitney test; qRT-PCR Z real-time Reverse Transcription Polymerase Chain Reaction; R and Ra Z ratio of the median values (group A/group B) found by qRT-PCR and array methods respectively. * Median values of qRT-PCR normalized signal (interquartile ranges [25e75% percentiles] in brackets).
To our knowledge, the present report is the first to identify a cluster of miRNAs expressed in primary tumor that are significantly correlated with a pCR after neoadjuvant treatment with capecitabine, oxaliplatin, and radiation in rectal cancer patients. We found a large set of miRNAs that were highly expressed in patients who had a very good response, whereas very few had the opposite behavior. qRT-PCR measurement of the 14 miRNAs whose differential expression was more pronounced confirmed the association for 13, with even higher significance levels than in the array measure. This enabled us to make further data analyses using straightforward thresholds for the expression of each miRNA to identify very good responder patients. We found that two miRNAs (miR-630 and miR-622) were upregulated in all patients of group A and downregulated in all patients of group B (sensitivity and specificity: 100%). Another novelty of our report is the procedure we used to select significant miRNAs: we did not choose a priori specific miRNAs based on their known function, but we isolated all miRNAs in primary tumor tissues and selected those significantly differentially expressed in patients with or without a pCR, using specific and innovative approaches. Our working hypothesis was that miRNAs that regulate genes involved in the survival of tumor cells after damages caused by chemotherapy or radiation affect the response to neoadjuvant treatment. In a similar context, Svoboda et al. (19) studied the expression of nine miRNAs in the biopsies of 35 patients with rectal cancer before and after neoadjuvant treatment with capecitabine and radiation therapy. They found an enormous interpatient variability thus leading to not conclusive results; only miR125b and miR137 were significantly upregulated after treatment, only in patients with a weak response and a higher tumor stage. Nakajima et al. (20) evaluated the correlation between miRNA expression and response to S-1 in 46 patients with recurrent or residual colon
cancer. The expression of five miRNAs related to the p53 pathway was tested in paraffin-embedded primary tumor samples: hsa-let7g and hsa-miR-181b levels were significantly lower in patients who responded to S-1, suggesting that the let-7 family of miRNAs can affect fluoropyrimidine activity by modulating RAS, cyclines, c-myc, or E2F transcription. The potential role of miRNA in antitumor activity of platinum compounds has been studied mainly in ovarian cancer tissue. Eitan et al. (21) found that seven miRNAs were significantly differentially expressed in tumors from 38 Stage III platinumsensitive and platinum-resistant ovarian cancer patients treated with cisplatin-based chemotherapy. Moreover, low levels of hsa-miR-27a and hsa-miR-23a were significantly associated to a better progression-free and overall survival. Jang et al. (22) investigated the miRNA signature in 72 Stage IIIeIV ovarian cancer patients. Nine miRNAs were significantly differentially expressed in patients who responded or not to chemotherapy with cisplatin. In particular, low levels of let-7i significantly correlated with poor response and short progression-free survival. The most widely used method of bioinformatic prediction of miRNA targets (www.targetscan.com) shows that many of the 13 miRNAs that we found to be correlated with pCR interact with mRNA encoding: a) proteins downstream to the EGFR pathways (miR-630, miR-1274b, miR 125a-3p, miR-671-5p, miR-188-5p, miR-1183); b) proteins involved in IGFR pathway (mir-765, miR-125a-3p, miR-630); c) proteins involved in DNA repair, such as XRCC2-5-6 (miR-622, miR-630, miR188e5p, miR-1183, miR 1224-5p), PARP3 (miR-630), and ERCC3 (miR-188-5p); and c) SOD (miR-630, miR-1183, miR25a-3p). Interestingly, miR-630 and/or miR-622 can target several proteins involved in the mechanisms of DNA repair, which suggests they are involved in sensitivity to radiation and platinum treatment.
1118 Della Vittoria Scarpati et al.
International Journal of Radiation Oncology Biology Physics
Fig. 2. Heat map. Each row represents a single micro (mi)RNA that was validated by qRT-PCR chain reaction (aRT-PCR), and the columns show the patients. Black indicates miRNA expression above the cutoff, and gray below. The specificity and sensitivity of each miRNA are reported in the last two columns. “Concordance” indicates the number of miRNA concordant with the signature of group A (i.e., the first 11 miRNAs above the threshold and the last two below, shown in the first column). Recently, Galluzzi et al. (23) reported that miR-630 upregulation is strongly related to cisplatin response in nonesmall-cell lung cancer cell lines, suggesting that this miRNA blocks the upstream signaling pathways that are ignited by DNA damage, and converge to p53. Therefore, miR-630 upregulation could affect platinum-induced cancer cell death by making it difficult for tumor cells to repair efficiently DNA damage induced by the drug. In this light, in our patients, the strong correlation between miR630 upregulation and pCR might be related to the scarce ability of tumors to repair DNA damage induced by radiotherapy and oxaliplatin. In conclusion, evaluation of the 13 validated miRNAs (or a subset of them) on fresh biopsies collected at staging colonoscopy might be performed to predict complete response to neoadjuvant chemoradiotherapy. After the procedure of measurement is standardized (e.g., by adopting absolute units for the expression of each miRNA and selecting the most informative subset of miRNA), we could calculate for each new patient a score of concordance with the signature of a typical responding tumor. Our results should be validated in larger series of patients. However, we are confident that the methodologies used for the analysis are sufficiently robust to be applicable. Indeed, the wide gap between the scores of groups A and B and the limited chance of falsepositive or false-negative results suggest that the score is a valid method with which to predict complete response to neoadjuvant treatment in rectal cancer patients.
References 1. Aslam MI, Taylor K, Pringle JH, et al. MicroRNAs are novel biomarkers of colorectal cancer. Br J Surg 2009;96:702e710. 2. Faber C, Kirchner T, Hlubek F. The impact of microRNAs on colorectal cancer. Virchows Arch 2009;454:359e367. 3. Shenouda SK, Alahari SK. MicroRNA function in cancer: Oncogene or a tumor suppressor? Cancer Metastasis Rev 2009;28:369e378. 4. Michael MZ, van Holst SMOC, Pellekaan NG, et al. Reduced accumulation of specific microRNAs in colorectal neoplasia. Mol Cancer Res 2003;1:882e891.
5. Bandres E, Cubedo E, Agirre X, et al. Identification by real-time PCR of 13 mature microRNAs differentially expressed in colorectal cancer and non-tumoral tissues. Mol Cancer 2006;5:29. 6. Slaby O, Svoboda M, Michalek J, et al. MicroRNAs in colorectal cancer: Translation of molecular biology into clinical application. Mol Cancer 2009;8:102. 7. Schetter AJ, Leung SY, Sohn JJ, et al. MicroRNA expression profiles associated with prognosis and therapeutic outcome in colon adenocarcinoma. JAMA 2008;299:425e436. 8. Schepeler T, Reinert JT, Ostenfeld MS, et al. Diagnostic and prognostic microRNAs in stage II colon cancer. Cancer Res 2008;68: 6416e6424. 9. Xi Y, Formentini A, Chien M, et al. Prognostic values of microRNAs in colorectal cancer. Biomark Insights 2006;2:113e121. 10. Graziano F, Canestrari E, Loupakis F, et al. Genetic modulation of the Let-7 microRNA binding to KRAS 30 -untranslated region and survival of metastatic colorectal cancer patients treated with salvage cetuximab-irinotecan. Pharmacogenomics J 2010;10:458e464. 11. Hummel R, Hussej DJ, Haier J. MicroRNAs: Predictors and modifiers of chemo- and radiotherapy in different tumor types. Eur J Cancer 2010;46:298e311. 12. Jemal A, Bray F, Center MM, et al. Global cancer statistics. CA Cancer J Clin 2011;61:69e90. 13. Suarez J, Vera R, Bale´n E, et al. Pathologic response assessed by Mandard grade is a better prognostic factor than down staging for disease-free survival after preoperative radiochemotherapy for advanced rectal cancer. Colorectal Dis 2007;10:563e568. 14. Carlomagno C, Pepe S, D’Armiento FP, et al. Predictive factors of complete response to neoadjuvant chemoradiotherapy in patients with rectal cancer. Oncology 2010;78:369e375. 15. Mandard AM, Dalibard F, Mandard JC, et al. Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma. Cancer 1994;73:2680e2686. 16. Carlomagno C, Farella A, Bucci L, et al. Neoadjuvant treatment of rectal cancer with capecitabine and oxaliplatin in combination with radiotherapy: A phase II study. Ann Oncol 2009;20:906e912. 17. Marchini S, Cavalieri D, Fruscio R, et al. Association between miR200c and the survival of patients with stage I epithelial ovarian cancer: A retrospective study of two independent tumour tissue collections. Lancet Oncol 2011;12:273e285. 18. Benjamini Y, Hochberg Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. J Roy Stat Soc B 1995;57:289e300.
Volume 83 Number 4 2012 19. Svoboda M, Izakovicova HL, Sefr R, et al. Micro-RNAs miR125b and miR137 are frequently upregulated in response to capecitabine chemoradiotherapy of rectal cancer. Int J Oncol 2008;33:541e547. 20. Nakajima G, Hayashi K, Xi Y, et al. Non-coding MicroRNAs hsalet-7g and hsa-miR-181b are associated with chemoresponse to S-1 in colon cancer. Cancer Genomics Proteomics 2006;3:317e324. 21. Eitan R, Kushnir M, Lithwick-Yanai G, et al. Tumor microRNA expression patterns associated with resistance to platinum based
microRNA profiling and response to chemoradiotherapy 1119 chemotherapy and survival in ovarian cancer patients. Gynecol Oncol 2009;114:253e259. 22. Jang N, Kaur S, Volinia S, et al. MiscoRNA microarray indentifies Let-7i as a novel biomarker and therapeutic target in human epithelial ovarian cancer. Cancer Res 2008;68:10307e10314. 23. Galluzzi L, Morselli E, Vitale I, et al. miR-181a and miR-630 regulate cisplatin-induced cancer cell death. Cancer Res 2010;70: 1793e1803.
Radiation Oncology biology
physics
www.redjournal.org
Clinical Investigation: Gastrointestinal Cancer
A Specific miRNA Signature Correlates With Complete Pathological Response to Neoadjuvant Chemoradiotherapy in Locally Advanced Rectal Cancer Giuseppina Della Vittoria Scarpati, M.D., Ph.D.,*,1 Francesca Falcetta, Ph.D.,y,1 Chiara Carlomagno, M.D., Ph.D.,* Paolo Ubezio, Ph.D.,y Sergio Marchini, Ph.D.,y Alfonso De Stefano, M.D.,* Vijay Kumar Singh, Ph.D.,z Maurizio D’Incalci, M.D.,y Sabino De Placido, M.D.,* and Stefano Pepe, M.D., Ph.D.x From the *Department of Molecular and Clinical Endocrinology and Oncology, University of Naples Federico II, Naples, Italy; yLaboratory of Cancer Pharmacology, Department of Oncology, “Mario Negri” Institute for Pharmacological Research, Milan, Italy; zCancer Genomics Laboratory, Fondazione “Edo ed Elvo Tempia Valenta”, Biella, Italy; and x Division of Oncology, University of Salerno, Italy Received Jul 7, 2011, and in revised form Sep 8, 2011. Accepted for publication Sep 9, 2011
Summary The study evaluated the miRNA expression in tumor biopsies of patients with rectal cancer, in order to identify a specific “signature” correlating with pathological complete response (pCR) after neoadjuvant chemo-radiotherapy. A set of 13 miRNAs were found which bore this strong association and which thus have considerable clinical potential to predict response in rectal cancer patients.
Purpose: MicroRNAs (miRNAs) are small, noncoding RNA molecules that can be down- or upregulated in colorectal cancer and have been associated to prognosis and response to treatment. We studied miRNA expression in tumor biopsies of patients with rectal cancer to identify a specific “signature” correlating with pathological complete response (pCR) after neoadjuvant chemoradiotherapy. Methods and Materials: A total of 38 T3e4/Nþ rectal cancer patients received capecitabine-oxaliplatin and radiotherapy followed by surgery. Pathologic response was scored according to the Mandard TRG scale. MiRNA expression was analyzed by microarray and confirmed by real-time Reverse Transcription Polymerase Chain Reaction (qRTPCR) on frozen biopsies obtained before treatment. The correlation between miRNA expression and TRG, coded as TRG1 (pCR) vs. TRG >1 (no pCR), was assessed by methods specifically designed for this study. Results: Microarray analysis selected 14 miRNAs as being differentially expressed in TRG1 patients, and 13 were confirmed by qRT-PCR: 11 miRNAs (miR-1183, miR-483-5p, miR-622, miR-125a-3p, miR-1224-5p, miR-188-5p, miR-1471, miR-671-5p, miR-1909*, miR-630, miR-765) were significantly upregulated in TRG1 patients, 2 (miR-1274b, miR-720) were downexpressed. MiR-622 and miR-630 had a 100% sensitivity and specificity in selecting TRG1 cases.
Reprint requests to: Chiara Carlomagno, M.D., Ph.D., Dipartimento di Endocrinologia ed Oncologia Molecolare e Clinica, Universita` di Napoli Federico II, Via Sergio Pansini 5; 80131 Naples, Italy. Tel: þ390817464271; Fax: þ390812203147; E-mail: [email protected] Conflict of interest: none. Int J Radiation Oncol Biol Phys, Vol. 83, No. 4, pp. 1113e1119, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2011.09.030
Supplementary material for this article can be found at www.redjournal.org. 1 Contributed equally to this study. AcknowledgmentsdWe thank Jean Ann Gilder (Scientific Communication srl) for language editing.
International Journal of Radiation Oncology Biology Physics
1114 Della Vittoria Scarpati et al.
Conclusions: A set of 13 miRNAs is strongly associated with pCR and may represent a specific predictor of response to chemoradiotherapy in rectal cancer patients. Ó 2012 Elsevier Inc. Keywords: miRNA, Predictive factors, Preoperative chemoradiotherapy, Rectal cancer
Introduction MicroRNAs (miRNAs) are small, noncoding sequences, approximately 18e25 nucleotides long, that are post-transcriptional regulators of gene expression. They bind to complementary sequences on messenger RNA transcripts (mRNAs), usually resulting in translational repression and gene silencing. Aberrant expression of miRNAs is implicated in numerous diseases and miRNA misregulation is also linked to many types of cancer (1, 2). Depending on the genes they regulate, miRNAs can function as either oncogenes or tumor suppressors (3). Several studies showed that over- or underexpression of specific miRNAs is associated with colorectal cancer (CRC) development and progression being differently expressed in tumor cells and in normal mucosa (4e6) and in early- versus late-stage neoplasms (5, 6). Moreover, specific miRNAs aree significantly correlated with disease-free and overall survival in patients with early colon cancer (7e9) and with progression-free and overall survival in patients with metastatic CRC (9, 10). An important challenge in medical oncology is to identify the patient- or tumor-related characteristics responsible for the response to conventional or targeted drugs. In vitro experiments suggest that several miRNAs can affect sensitivity of different human tumor cell lines to chemotherapeutic agents (11). CRC is the third most frequent neoplasia in men and the second in women (12). About one-third of large bowel tumors are rectal cancers. Neoadjuvant chemoradiation followed by surgery is now the standard treatment for TNM Stage IIeIII rectal cancer. Several retrospective analyses suggest that the pathological stage of disease after neoadjuvant treatment has a significant prognostic impact on disease-free and overall survival. In particular, the subgroup of patients who achieve a complete pathological response (pCR) has a very low risk of local or distant recurrence (13, 14). This observation prompts the evaluation of chemoradiotherapy regimens that induce high response rates and the profiling of tumors that will or will not respond to specific treatments. To address this topic, we studied the expression profile of miRNAs in biopsies of patients with locally advanced rectal cancer treated with neoadjuvant chemoradiation in the attempt to identify a specific signature associated with a pCR to preoperative treatment.
Methods and Materials
patient refused surgery; however, the disease was not detectable by rectal endoscopy and multiple biopsies in the zone of the previous mass resulted negative. Surgical specimens were evaluated according to the College of American Pathologists protocol for invasive carcinomas (TNM, 6th ed.). Pathologic response was scored according to tumor regression grade (TRG) as described by Mandard (15). Details of treatment and clinical results are reported elsewhere (16).
Tissue processing Fresh biopsies of rectal tumors were collected between January 2005 and December 2007 during staging colonoscopy, frozen within 5 min in liquid nitrogen, and stored at 80 C. The collection and the use of tumor samples for research purposes was approved by the local ethical committee and a written consent was obtained from the patients. Investigators performing microRNA analysis were blind to the clinical outcome of patients.
miRNAs isolation and microarray analysis Starting in January 2010, frozen tissues (100 mg) were homogenized at 8000 rpm in a Ultraturrax T-10 at 4 C with 1 mL of Trizol until a totally liquid phase. Description of total RNA extraction and microarray assay has been performed as described elsewhere (17). Median background and fluorescence values of each spot were processed and normalized as described in the supplementary data.
Microarray data analysis Patients were divided in two groups: group A (pCR [TRG 1] including the nonoperated patient without detectable disease) and group B (any pathologic response other than complete). Because of the relatively low number of cases, we strengthen the reliability of the difference in miRNA expression between the two groups by applying six different methods of analysis. miRNAs found differentially expressed by most methods were selected for subsequent qRT-PCR (real-time Reverse Transcription Polymerase Chain Reaction) validation.
Patients and treatment
Methods I and II
Thirty-eight patients with a histologic diagnosis of rectal adenocarcinoma invading through the intestinal wall and/or with lymph node involvement as evaluated by endorectal ultrasonography (uT3-T4 and/or uNþ) were treated with neoadjuvant chemoradiotherapy (capecitabine þ oxaliplatin in combination with 45 Gy of pelvic conformal radiotherapy). Six to 8 weeks after radiotherapy completion, 37 patients underwent surgery. One
Raw data were preprocessed with (Method I) or without (Method II) background subtraction and normalized by the invariant method, implemented in R programming language (version 2.11) and the Bioconductor software suite (version 2.6). The statistical Welch t test was applied and the p values were adjusted for multiple testing by the Benjamini and Hochberg False Discovery Rate method (18); a significance level of 0.1 was used for statistical testing.
Volume 83 Number 4 2012
microRNA profiling and response to chemoradiotherapy 1115
Methods III and IV Raw data were preprocessed and normalized as described in the section “Microarray data preprocessing” (supplementary data). The statistical Welch t test was applied and the p values were calculated for each probe. Method III relies on an “miRNA adjusted p value,” which was calculated taking the median the p values of its probes and adjusting with the false discovery rate (FDR) method. A significance level of 0.05 was used for statistical testing. Considering that each probe represents a different test for a single miRNA, which has distinct binding affinities, the probability that a difference of expression of a miRNA between the two groups was due to chance is much reduced if the p values of all its probes were low. Therefore, in method IV, we calculated a score (scmiRNA) associated to an miRNA combining the p values of its probes with the formula: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p sc ZNpi p .p miRNAi
probe1i
probe2i
probeNpi
where Npi is the number of probes (1e4) of miRNA “i” and pprobe1i, pprobe2i . are the calculated p values of its probes. According to method IV, a miRNA was considered differentially expressed if scmiRNA 0.01.
from 1 mg of mirVana purified total RNA (Ambion-ABI, Milan Italy), using the miScriptÒ Reverse Transcription Kit, following the manufacturer’s instruction (Qiagen). miRNA expression was quantified by Syber Green chemistry using the commercial miScriptÒ Universal Primer, together with the miScriptÒ Primer Assay (Qiagen). Experiments were run in triplicate for each case, using 384-well reaction plates in an automatic liquid handling station (epMotion 5075LH; Eppendorf, Milan, Italy). qRT-PCR was done on an Applied Biosystems 7900HT apparatus (Ambion-ABI). Raw data were generated with SDS Relative Quantification software (version 2.3; Ambion-ABI). Fluorescence intensities for each miRNA were normalized against the normalization index calculated on RNU-6b used as housekeeping genes. To calculate the relative expression for each miRNA, we used the 2-DCT method to average the threshold cycle values of the three replicates for a single gene. For miRNA expression experiments, median values were compared using the nonparametric MannWhitney t test. Differences were considered statistically significant with a two-sided p < 0.05. Data are reported as medians with interquartile range (range, 25e75% percentile). All tests were done using GraphPad Prism Version 5.01 (GraphPad Software, La Jolla, CA).
Methods V and VI Raw data were preprocessed and normalized as described in the section “Microarray data preprocessing” (supplementary data). The normalized database was then split in two subsets (DBS1 and DBS2). While dividing the dataset in two subsets, the proportion of group A to group B patients in each subsets was kept approximately same. DBS1 included 4 group A and 14 group B patients, whereas DBs2 included 5 group A and 15 group B patients. The statistical Welch t test was applied and the p values were calculated for each probe separately within DBS1 and DBS2. According to method V, a miRNA was deemed differentially expressed if at least one of its probes was over- or underexpressed in both subsets with a p value <0.05. miRNA p values were then calculated separately in DBS1 and DBS2 taking the median of the p values of the probes of each miRNA. A miRNA was deemed differentially expressed when its p value was <0.1 in both subsets (method VI), meaning that the joint probability that a cross-validated difference was due to chance was 0.1 0.1 Z 0.01. Methods IIIeVI were fully implemented, including preprocessing of data, in Microsoft Excel.
Results The characteristics of the 38 patients are listed in Table 1. Thirty-seven patients underwent surgery; a pCR was achieved in nine cases. As of Dec. 31, 2010, 11 patients recurred, and 7 had died.
miRNA microarray analysis
The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE29298 (http://www.ncbi.nlm. nih.gov/geo/query/acc.cgi?accZGSE29298). Cluster analysis was done on the 53 miRNAs identified by the first selection of the differential expression analysis. Hierarchical dendrograms and matrix correlation were calculated with Pearson’s correlation as clustering metric. The calculation was done with the TIGR Multi-experiment Viewer feature of TM4 software suite and Microsoft Excel.
Of the 373 miRNAs with at least one probe of one array significantly higher than nonspecific fluorescence, 53 were differentially expressed in group A vs. B by at least two criteria (Table S1). This indicates that 17.4% of the miRNAs present in these tumors were differently expressed in good responders (group A) with respect to all the other patients. The trends of expression of most of these 53 miRNAs were well/highly correlated, which suggests the existence of a large cluster of interrelated miRNAs that are expressed at lower levels in less responding patients (Fig. 1). Only four miRNAs were more highly expressed in group B vs. group A. We used more stringent criteria to select a subset of these 53 miRNAs for qRT-PCR validation. For this purpose, we first excluded miRNAs with a low average expression (i.e., log fluorescence intensity <5 in both groups) and kept those differentially expressed by at least four methods. Then, we selected for qRT-PCR validation the 12 overexpressed miRNAs in group A with the highest fold change, and the only two miRNA with increasing expression in group B. Fold change and p values of these 14 miRNAs are reported in Table 2. Three other miRNAs, not differentially expressed by array analysis (negative controls), were measured by qRT-PCR.
Validation of the microarray results by qRT-PCR
miRNA signature validation by qRT-PCR
Mature miRNA expression levels were examined by RT-PCR using dedicated sets of commercial primers (Qiagen, Milan, Italy). Briefly, specific cDNA was generated in a single step reaction
Table 2 shows the median expression level of the qRT-PCReselected miRNA and the 25e75% range in group A compared withgroup B. Thirteen of the 14 selected miRNAs were
miRNA correlation and cluster analysis
International Journal of Radiation Oncology Biology Physics
1116 Della Vittoria Scarpati et al. Table 1
Patient and tumor characteristics n
Gender Males 25 Females 13 Age Median (range) 64.4 (42.7e79.3) Surgery Yes 37 No 1 Tumor regression grade (TRG; n Z 37) TRG1 8* TRG2 16 TRG3 10 TRG4 3 First site of recurrence Any 11 Local 1 Distant 10 Local þ distant 0
% 66.8 34.2
97.4 2.6 21.6 43.2 27.0 8.1 28.9 2.6 26.3 e
* 9 pathologically complete responses, including the patient not operated.
significantly differentially expressed in the two groups, whereas the viral miRNA (hsv1-miR-H1) was not found by qRT-PCR. The three negative controls were not differentially expressed, which is
coherent with the array analysis. Eleven miRNAs (miR-1183, miR483e5p, miR-622, miR-125a-3p, miR-1224e5p, miR-188e5p, miR-1471, miR-671e5p, miR-1909), miR-630, and miR-765) were upregulated, whereas two miRNAs (miR-1274b and miR720) were downregulated in group A vs. group B (Table 2). Considering the average of the medians in groups A and B as a cutoff, we built the heat map shown in Fig. 2. Each of the 13 differently expressed miRNAs individually had a high specificity and sensitivity, which reached 100% in the case of miR-622 and miR630. Considering the 13 miRNAs collectively, all group A patients had a concordance of at least 10/13 with the signature of group A, whereas no group B patient reached a score higher than 5/13.
Discussion Surgery is the most effective treatment for rectal cancer. However, in locally advanced stages, radiotherapy and chemotherapy reduce the risk of local and distant relapse. Consequently, a multidisciplinary approach is currently the standard. Complete pathologic disappearance of tumor cells after neoadjuvant chemoradiotherapy occurs in about 20% of patients and is correlated with a very good long-term prognosis. In the era of customized treatment, based on the assessment of predictive factors, we should exploit new techniques that can identify predictive factors for conventional treatments, such as chemo- or radiation therapy. The ultimate goal is to select for each patient the most effective and less toxic treatment.
Fig. 1. Correlation matrix and cluster analysis. Each row and column represents the 53 micro (mi)RNAs found to be differentially expressed by at least two criteria. The colors indicate the Pearson’s correlation between miRNAs; the value of the correlation index corresponding to each colour is shown. The two main clusters are constituted by miRNAs that are expressed at a higher level in group A (A > B) or in group B (B > A).
Volume 83 Number 4 2012 Table 2
microRNA profiling and response to chemoradiotherapy 1117
Expression of the 14 selected miRNA by microarray and qRT-PCR Microarray
miRNA
Ra
p
hsa-miR-1183 hsa-miR-1274b hsa-miR-483e5p hsa-miR-622 hsa-miR-125a-3p hsa-miR-1224e5p hsa-miR-188e5p hsa-miR-1471 hsa-miR-671e5p hsa-miR-1909* hsa-miR-630 hsa-miR-720 hsa-miR-765 hsv1-miR-H1
3.11 0.60 2.41 3.10 2.20 2.31 2.31 3.24 2.31 2.88 2.43 0.57 3.02 3.61
0.005 0.054 0.006 0.008 0.006 0.007 0.005 0.019 0.013 0.042 0.017 0.032 0.008 0.005
Not differentially expressed hsa-miR-30a* 1.46 0.628 hsa-miR-92a 0.80 0.705 hsa-miR-34a 1.13 0.753
qRT-PCR Group A* 4295 217500 7061 4275 323 3275 1753 4981 2757 4278 88.7 120900 5979 395.4
(3162e8158) (169000e315900) (5291e9534) (3363e5504) (281e538.5) (2101e5093) (1432e2593) (3539e8410) (2334e5880) (3189e8054) (71.35e118.6) (84440e203600) (3459e6972) (114.3e1465)
48.7 (22.5e330) 13150 (9295e58870) 4180 (1277e18130)
Group B* 823 550000 656 617 88.86 704 378 1023 1170 1125 12.66 580000 1276 158
(517e1730) (296600e751500) (305e1700) (250e935.5) (53.23e140.9) (348e977) (223e553.5) (757e2668) (582.5e1571) (707e2207) (8.861e22.3) (300700e1256000) (949e2307) (105.5e322.5)
32.41 (16.77e74.85) 11400 (4763e15550) 1910 (886e4271)
R
p
5.22 0.40 10.76 6.93 3.63 4.65 4.64 4.87 2.36 3.80 7.01 0.21 4.69 2.50
<0.0001 0.018 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.0006 <0.0001 0.0002 <0.0001 <0.0001 0.0004 0.131
1.50 1.15 2.19
0.216 0.114 0.099
Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed Confirmed confirmed Confirmed Confirmed Not confirmed Confirmed Confirmed Confirmed
Abbreviations: p Z Mann-Whitney test; qRT-PCR Z real-time Reverse Transcription Polymerase Chain Reaction; R and Ra Z ratio of the median values (group A/group B) found by qRT-PCR and array methods respectively. * Median values of qRT-PCR normalized signal (interquartile ranges [25e75% percentiles] in brackets).
To our knowledge, the present report is the first to identify a cluster of miRNAs expressed in primary tumor that are significantly correlated with a pCR after neoadjuvant treatment with capecitabine, oxaliplatin, and radiation in rectal cancer patients. We found a large set of miRNAs that were highly expressed in patients who had a very good response, whereas very few had the opposite behavior. qRT-PCR measurement of the 14 miRNAs whose differential expression was more pronounced confirmed the association for 13, with even higher significance levels than in the array measure. This enabled us to make further data analyses using straightforward thresholds for the expression of each miRNA to identify very good responder patients. We found that two miRNAs (miR-630 and miR-622) were upregulated in all patients of group A and downregulated in all patients of group B (sensitivity and specificity: 100%). Another novelty of our report is the procedure we used to select significant miRNAs: we did not choose a priori specific miRNAs based on their known function, but we isolated all miRNAs in primary tumor tissues and selected those significantly differentially expressed in patients with or without a pCR, using specific and innovative approaches. Our working hypothesis was that miRNAs that regulate genes involved in the survival of tumor cells after damages caused by chemotherapy or radiation affect the response to neoadjuvant treatment. In a similar context, Svoboda et al. (19) studied the expression of nine miRNAs in the biopsies of 35 patients with rectal cancer before and after neoadjuvant treatment with capecitabine and radiation therapy. They found an enormous interpatient variability thus leading to not conclusive results; only miR125b and miR137 were significantly upregulated after treatment, only in patients with a weak response and a higher tumor stage. Nakajima et al. (20) evaluated the correlation between miRNA expression and response to S-1 in 46 patients with recurrent or residual colon
cancer. The expression of five miRNAs related to the p53 pathway was tested in paraffin-embedded primary tumor samples: hsa-let7g and hsa-miR-181b levels were significantly lower in patients who responded to S-1, suggesting that the let-7 family of miRNAs can affect fluoropyrimidine activity by modulating RAS, cyclines, c-myc, or E2F transcription. The potential role of miRNA in antitumor activity of platinum compounds has been studied mainly in ovarian cancer tissue. Eitan et al. (21) found that seven miRNAs were significantly differentially expressed in tumors from 38 Stage III platinumsensitive and platinum-resistant ovarian cancer patients treated with cisplatin-based chemotherapy. Moreover, low levels of hsa-miR-27a and hsa-miR-23a were significantly associated to a better progression-free and overall survival. Jang et al. (22) investigated the miRNA signature in 72 Stage IIIeIV ovarian cancer patients. Nine miRNAs were significantly differentially expressed in patients who responded or not to chemotherapy with cisplatin. In particular, low levels of let-7i significantly correlated with poor response and short progression-free survival. The most widely used method of bioinformatic prediction of miRNA targets (www.targetscan.com) shows that many of the 13 miRNAs that we found to be correlated with pCR interact with mRNA encoding: a) proteins downstream to the EGFR pathways (miR-630, miR-1274b, miR 125a-3p, miR-671-5p, miR-188-5p, miR-1183); b) proteins involved in IGFR pathway (mir-765, miR-125a-3p, miR-630); c) proteins involved in DNA repair, such as XRCC2-5-6 (miR-622, miR-630, miR188e5p, miR-1183, miR 1224-5p), PARP3 (miR-630), and ERCC3 (miR-188-5p); and c) SOD (miR-630, miR-1183, miR25a-3p). Interestingly, miR-630 and/or miR-622 can target several proteins involved in the mechanisms of DNA repair, which suggests they are involved in sensitivity to radiation and platinum treatment.
1118 Della Vittoria Scarpati et al.
International Journal of Radiation Oncology Biology Physics
Fig. 2. Heat map. Each row represents a single micro (mi)RNA that was validated by qRT-PCR chain reaction (aRT-PCR), and the columns show the patients. Black indicates miRNA expression above the cutoff, and gray below. The specificity and sensitivity of each miRNA are reported in the last two columns. “Concordance” indicates the number of miRNA concordant with the signature of group A (i.e., the first 11 miRNAs above the threshold and the last two below, shown in the first column). Recently, Galluzzi et al. (23) reported that miR-630 upregulation is strongly related to cisplatin response in nonesmall-cell lung cancer cell lines, suggesting that this miRNA blocks the upstream signaling pathways that are ignited by DNA damage, and converge to p53. Therefore, miR-630 upregulation could affect platinum-induced cancer cell death by making it difficult for tumor cells to repair efficiently DNA damage induced by the drug. In this light, in our patients, the strong correlation between miR630 upregulation and pCR might be related to the scarce ability of tumors to repair DNA damage induced by radiotherapy and oxaliplatin. In conclusion, evaluation of the 13 validated miRNAs (or a subset of them) on fresh biopsies collected at staging colonoscopy might be performed to predict complete response to neoadjuvant chemoradiotherapy. After the procedure of measurement is standardized (e.g., by adopting absolute units for the expression of each miRNA and selecting the most informative subset of miRNA), we could calculate for each new patient a score of concordance with the signature of a typical responding tumor. Our results should be validated in larger series of patients. However, we are confident that the methodologies used for the analysis are sufficiently robust to be applicable. Indeed, the wide gap between the scores of groups A and B and the limited chance of falsepositive or false-negative results suggest that the score is a valid method with which to predict complete response to neoadjuvant treatment in rectal cancer patients.
References 1. Aslam MI, Taylor K, Pringle JH, et al. MicroRNAs are novel biomarkers of colorectal cancer. Br J Surg 2009;96:702e710. 2. Faber C, Kirchner T, Hlubek F. The impact of microRNAs on colorectal cancer. Virchows Arch 2009;454:359e367. 3. Shenouda SK, Alahari SK. MicroRNA function in cancer: Oncogene or a tumor suppressor? Cancer Metastasis Rev 2009;28:369e378. 4. Michael MZ, van Holst SMOC, Pellekaan NG, et al. Reduced accumulation of specific microRNAs in colorectal neoplasia. Mol Cancer Res 2003;1:882e891.
5. Bandres E, Cubedo E, Agirre X, et al. Identification by real-time PCR of 13 mature microRNAs differentially expressed in colorectal cancer and non-tumoral tissues. Mol Cancer 2006;5:29. 6. Slaby O, Svoboda M, Michalek J, et al. MicroRNAs in colorectal cancer: Translation of molecular biology into clinical application. Mol Cancer 2009;8:102. 7. Schetter AJ, Leung SY, Sohn JJ, et al. MicroRNA expression profiles associated with prognosis and therapeutic outcome in colon adenocarcinoma. JAMA 2008;299:425e436. 8. Schepeler T, Reinert JT, Ostenfeld MS, et al. Diagnostic and prognostic microRNAs in stage II colon cancer. Cancer Res 2008;68: 6416e6424. 9. Xi Y, Formentini A, Chien M, et al. Prognostic values of microRNAs in colorectal cancer. Biomark Insights 2006;2:113e121. 10. Graziano F, Canestrari E, Loupakis F, et al. Genetic modulation of the Let-7 microRNA binding to KRAS 30 -untranslated region and survival of metastatic colorectal cancer patients treated with salvage cetuximab-irinotecan. Pharmacogenomics J 2010;10:458e464. 11. Hummel R, Hussej DJ, Haier J. MicroRNAs: Predictors and modifiers of chemo- and radiotherapy in different tumor types. Eur J Cancer 2010;46:298e311. 12. Jemal A, Bray F, Center MM, et al. Global cancer statistics. CA Cancer J Clin 2011;61:69e90. 13. Suarez J, Vera R, Bale´n E, et al. Pathologic response assessed by Mandard grade is a better prognostic factor than down staging for disease-free survival after preoperative radiochemotherapy for advanced rectal cancer. Colorectal Dis 2007;10:563e568. 14. Carlomagno C, Pepe S, D’Armiento FP, et al. Predictive factors of complete response to neoadjuvant chemoradiotherapy in patients with rectal cancer. Oncology 2010;78:369e375. 15. Mandard AM, Dalibard F, Mandard JC, et al. Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma. Cancer 1994;73:2680e2686. 16. Carlomagno C, Farella A, Bucci L, et al. Neoadjuvant treatment of rectal cancer with capecitabine and oxaliplatin in combination with radiotherapy: A phase II study. Ann Oncol 2009;20:906e912. 17. Marchini S, Cavalieri D, Fruscio R, et al. Association between miR200c and the survival of patients with stage I epithelial ovarian cancer: A retrospective study of two independent tumour tissue collections. Lancet Oncol 2011;12:273e285. 18. Benjamini Y, Hochberg Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. J Roy Stat Soc B 1995;57:289e300.
Volume 83 Number 4 2012 19. Svoboda M, Izakovicova HL, Sefr R, et al. Micro-RNAs miR125b and miR137 are frequently upregulated in response to capecitabine chemoradiotherapy of rectal cancer. Int J Oncol 2008;33:541e547. 20. Nakajima G, Hayashi K, Xi Y, et al. Non-coding MicroRNAs hsalet-7g and hsa-miR-181b are associated with chemoresponse to S-1 in colon cancer. Cancer Genomics Proteomics 2006;3:317e324. 21. Eitan R, Kushnir M, Lithwick-Yanai G, et al. Tumor microRNA expression patterns associated with resistance to platinum based
microRNA profiling and response to chemoradiotherapy 1119 chemotherapy and survival in ovarian cancer patients. Gynecol Oncol 2009;114:253e259. 22. Jang N, Kaur S, Volinia S, et al. MiscoRNA microarray indentifies Let-7i as a novel biomarker and therapeutic target in human epithelial ovarian cancer. Cancer Res 2008;68:10307e10314. 23. Galluzzi L, Morselli E, Vitale I, et al. miR-181a and miR-630 regulate cisplatin-induced cancer cell death. Cancer Res 2010;70: 1793e1803.