Predicting progression of bladder urothelial carcinoma using microRNA expression

Authors


Correspondence: Nativ Ofer, Department of Urology, Bnai Zion Medical Center, Golomb 47 Haifa 31048, Israel.

e-mail: ofer.nativ@b-zion.org.il

Abstract

What's known on the subject? and What does the study add?

  • Recurrence and progression prediction in urothelial cancer is currently based on clinical and pathological factors: tumour grade, tumour stage, number of lesions, tumour size, previous recurrence rate, and presence of concomitant carcinoma in situ. These factors are not specific enough to predict progression and ∼50% of patients diagnosed as high risk in fact do not progress within 3 years. Patient follow-up is both expensive and unpleasant (frequent invasive cystoscopies). Molecular biomarkers, including microRNAs have been studied to provide additional prognostic information for these patients, but to date no molecular biomarker has become the ‘gold standard’ for patient diagnosis and follow-up.
  • We used Rosetta Genomics’ highly specific microRNA expression profiling platforms to study the prognostic role of microRNAs in bladder cancer. Using microdissection we chose specific tumour microRNAs to study in order to avoid background contamination. Tumour progression was associated with altered levels of microRNAs. In particular, high expression levels of miR-29c* were associated with a good prognosis. The study found that the use of microRNAs for determining progression and invasiveness for patients with urothelial cancer could potentially have a substantial impact on the treatment and follow-up individual patients.

Objective

  • To identify microRNAs that could be useful as prognostic markers for non-muscle-invasive (NMI) bladder carcinoma.

Patients and Methods

  • Formalin-fixed, paraffin-embedded samples of 108 NMI bladder carcinomas, and 29 carcinomas invading bladder muscle were collected, and microRNA expression levels were measured using microarrays.
  • For 19 samples, microdissection was performed to compare microRNA expression between the tumour and surrounding tissue.
  • MicroRNAs that were found to be unrelated to the tumour itself were excluded as potential prognostic markers.

Results

  • Expression profiles identified microRNAs that were differentially expressed in NMI tumours from patients who later progressed to carcinoma invading bladder muscle compared with NMI tumours from patients that did not progress.
  • The microRNA profile of tumours invading the bladder muscle was more similar to that of NMI tumours from patients who later progressed, than to that of the same-stage NMI tumours from patients who did not later progress.
  • The expression level of one microRNA, miR-29c*, was significantly under-expressed in tumours that progressed and could be used to stratify patients with T1 disease into risk groups.

Conclusions

  • MicroRNAs can be useful biomarkers for prognosis in patients with urothelial carcinoma.
  • In our study, expression levels of several microRNAs, including miR-29c* identified high- and low-risk groups. These biomarkers show promise for the stratification of patients with bladder cancer.
Abbreviations
NMI

non-muscle-invasive

UC

urothelial cancer

CIS

carcinoma in situ

TURBT

transurethral resection of bladder tumour

FDR

false discovery rate

ROC

receiver-operating characteristic

AUC

area under the ROC curve

TTP

time-to-progression

NP

no progression

IP

invasive progression

Introduction

Urothelial cancer (UC) of the bladder is the fourth most common cancer in the western world according to the American Cancer Society: Cancer Facts and Figures 2011 (http://www.cancer.org/). At diagnosis, 70–75% of bladder tumours are non-muscle-invasive (NMI), i.e. Ta, T1 and Tis [1, 2]. Approximately 70% of these tumours are confined to the urothelium (Ta), 20% invade the lamina propria (T1), and 10% are carcinoma in situ (CIS) [3]. Ta and T1, with their various grades, compose a heterogeneous group of tumours with respect to prognosis. Low grade Ta lesions recur at a rate of 50–70%, and progress to carcinoma invading bladder muscle within 3 years in ∼5% of cases. High grade T1 tumours, by contrast, recur in more than 80% of cases and in 50% of patients, progress within 3 years.

Recurrence and progression prediction is currently based upon clinical and pathological factors: tumour grade, tumour stage, number of lesions, tumour size, previous recurrence rate, and presence of concomitant CIS [4, 5]. Tumour progression is affected mainly by the histological grade, and also by the T category and the presence of concomitant CIS, which are important risk factors [4-6]. Low and high grade tumours are vastly different in biological behaviour and clinical outcome and are therefore often viewed as two different diseases [7]. Despite the use of clinical and pathological factors, the ability to assess patient prognosis is not satisfactory, partially because of the subjectivity of grading and staging that causes relatively high inter-observer variability [8]. For example, in the current stratification system, ∼50% of patients diagnosed as high risk (high grade T1), in fact do not progress within 3 years. Since the follow-up and treatment regimes depend on prognosis, there is a need for more accurate stratification to increase the predictive values of risk groups. With a reliable diagnostic test for progression, suitable treatments could be tailored to every specific patient. This has led to an effort to find reliable biomarkers to predict progression in patients with UC. These potential markers include genetic alterations, methylation patterns, cell adhesion molecules, proteases, growth factors and other molecular markers. To date, the markers that have been suggested lack sufficient predictive power, especially in the clinical evaluation of T1 UC [9].

MicroRNAs are a recently discovered class of small non-coding RNAs that regulate gene expression by post-transcriptional regulation of mRNA levels and translation [10]. Numerous studies link microRNAs to the pathogenesis of human cancers, through their genomic locations in fragile sites [11], altered expression profiles in cancers [12], and functional role in differentiation and metastasis [13, 14]. Studies have pointed to the potential of microRNAs as prognostic biomarkers in various cancer types [15-20], including bladder cancer. Catto et al. [21], showed that high grade UC is characterized by up-regulation of microRNA-21 that suppresses p53 function, and low grade UC is characterized by loss of microRNAs-99a/100. miR-129, miR-133b and miR-518c* were found to have prognostic potential for predicting disease progression [22], and miR-452 and miR-452* were shown to be markers for death by disease [23]. Others propose miR-200 and miR-205 silencing and DNA hypermethylation as possible prognostic markers in bladder cancer [24]. Potential tumour microRNA markers have also been detected in the urine of patients with UC [25, 26].

In the present study, we investigated microRNA expression profiles in bladder tumours from a cohort of patients with UC and for whom extensive follow-up data were available. We studied how microRNA expression is related to tumour invasiveness and invasive potential and the association of microRNA expression with NMI bladder cancer progression.

Material and Methods

Patients and Samples

Formalin-fixed paraffin-embedded specimens of tumours removed by transurethral resection of bladder tumour (TURBT) procedures during the period 1995–2005 were obtained from three medical centres (Soroka University Medical Center, Rabin Medical Center and Bnai Zion Medical Center), with the approval of the respective internal review boards. Specimens were collected for patients who were treated and followed in these institutes and had no bladder cancer history before the TURBT specimen was obtained, and no history of NMI bladder cancer. Additional independent pathological review, performed by a pathologist in Rabin Medical Center for all study cases, confirmed the stage and the grade of all studied specimens. Whenever a tumour invading the bladder muscle was detected in a re-TURBT procedure the case was excluded from the study. Clinical follow-up data were collected retrospectively for all the cases.

In addition, pathology specimens were collected by Soroka Medical Center from 29 patients who underwent radical cystectomy as a result of carcinoma invading bladder muscle between 1999 and 2006. Additional pathological review confirmed muscle invasiveness in the specimens.

RNA Extraction

Total RNA was isolated from seven to 10 10-μm-thick tissue sections per case using the extraction protocol developed at Rosetta Genomics, as previously described [27]. Briefly, the sample was incubated in xylene at 57 °C to remove excess paraffin, and was then washed several times with ethanol. Proteins were degraded by incubating the sample in a proteinase K solution at 45 °C for a few hours. The RNA was extracted using acid phenol/chloroform and then precipitated using ethanol; DNases were introduced to digest DNA.

Microarray

Custom microRNA microarrays were prepared as described previously [27]. Briefly, ∼900 DNA oligonucleotide probes representing microRNAs were spotted in triplicate on coated microarray slides (Nexterion® Slide E, Schott, Mainz, Germany). Then 3.5 μg of total RNA from each sample was labelled by ligation of an RNA-linker, p-rCrU-Cy/dye (Dharmacon, Lafayette, CO, USA; Cy3 or Cy5) to the 3′ end. Slides were incubated with the labelled RNA for 12–16 h at 42 °C and then washed twice. Arrays were scanned at a resolution of 10 μm, and images were analysed using SpotReader software (Niles Scientific, Portola Valley, CA, USA). Microarray spots were combined and signals normalized as described previously [27]

Microdissection

For microdissection, 10-μm sections (10 sections per sample) were mounted onto glass slides. Sections were deparaffinized in xylene and rehydrated through descending ethanols. Sections were stained in 0.01% methylene blue, washed in double distilled water and dipped into 10% glycerol. Areas of interest (tumour, muscle and connective tissue/lamina propria) were scraped off the slide with a sterile scalpel blade under the upright microscope by A.F., who is a histogist with extensive previous experience of TCC. Dissected tissue fragments were collected into Eppendorf tubes with proteinase K buffer. RNA was then extracted.

Agilent Platform

For the microdissection experiment custom-designed arrays from Agilent Technologies (Santa Clara, CA, USA) were used. This involved the labelling of 0.38–1 μg of total-RNA by ligation of an RNA-linker, p-rCrU-Cy/dye (BioSpring GmbH, Frankfurt, Germany; Cy3 or Cy5) to the 3’ end. Synthetic small RNA controls were spiked before labelling. Slides were incubated with the labelled RNA for 12–16 h at 55 °C and washed according to the manufacturer's protocol. Arrays were scanned using the Agilent DNA Microarray Scanner Bundle at a resolution of 5 μm, dual pass at 100% and 10% laser power. Array images were analysed using Agilent Feature Extraction software version 10.7.1.1. Triplicate spots were combined to produce one signal by taking the logarithmic mean of reliable spots. Analysis was performed in log2-space. Normalization was performed for each sample with respect to a reference vector (R), calculated by taking the median expression level over the training set. For each sample data vector S, a 2nd degree polynomial F was found so as to provide the best fit between S and R, such that R≈F(S). This was performed on a set of invariant microRNAs; remote data points (‘outliers’) were not used for fitting the polynomial. For each probe in the sample (element Si in the vector S), the normalized value (in log2-space) Mi was calculated from the initial value Si by transforming it with the polynomial function F, so that Mi = F(Si).

Data Analysis

For any group comparison only microRNAs that we considered reliable and that were expressed above background level in one of the groups were compared. P values were calculated using a two-sided unpaired t-test on the log-transformed normalized signal, and the significance level was adjusted using the false discovery rate (FDR) [28]. Fold-change for each microRNA was calculated by the change in the median values of the normalized fluorescence signal between the groups. For each microRNA, we characterized the ability to separate the two groups by the receiver-operating characteristic (ROC) curve and calculated the area under the ROC curve (AUC). Kaplan–Meier survival analysis was used to evaluate the prognostic value of the microRNAs, with the log-rank test used for significance (for tertiles, log rank between the first and third tertile).

Results

A total of 137 bladder tumours were included in the study (Table 1). Of these, 29 samples were classified as carcinoma invading the bladder muscle and 108 were classified as NMI. Out of the 108 patients with NMI bladder cancer, 79 did not progress to T2 or higher stages during the median (range) follow-up period of 53 (28–156) months, and 29 had a muscle-invasive progression within 1–117 months (median time to progression [TTP] 14 months). We termed the first group ‘no progression’ (NP) and the second group ‘invasive progression’ (IP).

Table 1. Clinical characteristics of the study cases.
 NMI tumoursTumours invading bladder muscle
NP groupIP group
  1. aData on previous tumours was not fully updated for some of the patients.
No. of patients792929
Median (range) age, years65 (38–94)73 (48–89)66 (48–82)
Gender: males, n (%)67 (85)25 (85)25 (85)
History of tumours, na   
Yes6310
No13810
Grade, n (%)   
Low27 (34)6 (21)NA
High52 (66)23 (79)NA
Stage, n (%)   
Ta45 (57)5 (17)0
T134 (43)24 (83)0
≥T20029 (100)
Median (range) follow-up, months53 (28–156)14 (1–117)NA
Number of patients recurred3514 (before progression)NA

To identify microRNAs that were predominantly expressed in non-tumour tissue, we performed a preliminary experiment (19 blocks) in which the tumour was micro dissected, and the microRNA expression was compared between the tumour-only and the full sample (Fig. 1A). In five samples, we also microdissected muscle and lamina propria and compared all the sample components. Any microRNA which was significantly overexpressed in the full sample compared with the tumour with a fold change >1.5, passing a FDR [28] of 0.1, was designated as predominantly expressed in non-tumour tissue. Our analysis identified 26 microRNAs which were predominantly expressed in non-tumour tissue (Fig. 1A). Figure 1B and C show data for two of the 26 microRNA: mir-145, which was indeed known to be expressed in smooth muscle [29] and miR-199a-3p, both highlighted in panel A. As can be seen, the expression of these microRNAs in non-tumour components causes their expression in the non-dissected sample to differ from their expression in the tumour itself, and hence these microRNAs were excluded from the prognostic marker list. By contrast, miR-200c was predominantly expressed in the tumour tissue, and could potentially be used as a prognostic marker (Fig. 1A,D). miR-200c and its family are highly enriched in epithelial tissues and are involved in cancer metastasis [30, 31]. We excluded the 26 ‘non-tumour’ microRNAs from further analysis aimed at identifying prognostic markers.

Figure 1.

Microdissection results. A, Dot-plot showing the median microRNA expression levels (normalized fluorescence signals by microarray, shown in log-scale) in 19 tumours that were microdissected (x-axis) vs samples from the same block without microdissection (y-axis). Grey crosses show control probes and microRNAs that are not used in the comparisons or whose median expression level was <100 in both groups. The three microRNAs shown in subplots B–D and miR-29c* are labelled, as well as the 26 microRNAs identified as ‘non-tumour’. B, Expression of miR-145, C, expression of miR-199a-3p, and D, expression of miR-200c in the components of five samples (sa-1 to sa-5), that were microdissected into tumour and tissue components. The expression in the full sample is shown, without microdissection (red star), the tumour-only component (yellow diamond), the muscle (blue square), and the lamina propria (green circle).

Next, we compared the IP tumours with the NP tumours. Significant differences were found in the microRNA expression levels of the two groups (Fig. 2). In all, 17 microRNAs (Table 2) had a fold change of median expression >1.2 and passed FDR of 0.1 (P < 0.029). Four of these microRNAs had a differential expression with a fold change >2 (Fig. 2). The most significant microRNA was mir-29c* with a higher expression in NP tumours (fold change of 2.1; P = 3.0E-06). Furthermore, although histologically, both IP and NP are considered to be NMI tumours, on the molecular level those that will progress are already more similar to tumours invading the bladder muscle than those tumours that will not progress (23 vs 53 microRNAs which were differentially expressed, respectively; Fig. S1). This trend can also be seen in Fig. 3.

Figure 2.

Differences in expression levels of microRNAs between NMI tumours in the IP and NP groups: dot-plot showing the median microRNA expression levels (normalized fluorescence signals by microarray, shown in log-scale) in NMI tumours in the IP group (n = 29) and those in the NP group (n = 79). Grey crosses show control probes and microRNAs that are not used in the comparisons or whose expression level was at background levels (median signal < 600) in both groups. All other microRNAs were tested for statistical differences by two-sided unpaired t-test, with significance corrected by FDR of 0.1. In all, 17 microRNAs had a P value lower than the FDR threshold of 0.029 and a fold-change of medians > 1.2, and are marked by red circles. MicroRNAs with fold change > 2 in either direction are highlighted in yellow and labelled. B, Box-plots showing the median (horizontal line), 25 to 75 percentile (box), and extent of data (‘whiskers’) of expression levels of the four microRNAs, with fold change > 2 (labelled in sub-plot A), in the two NMI bladder tumour groups as well as in the invasive group. Units show log2 of the normalized fluorescence signal. P values and fold changes are between NP and IP (Table 2).

Figure 3.

Differences in expression levels of microRNAs among NMI tumours in the IP and NP groups and invasive tumours: Box-plots showing the median (horizontal line), 25 to 75 percentile (box), and extent of data (‘whiskers’) of expression levels of the four microRNAs, with fold change > 2 (labelled in Fig. 2), in the two NMI bladder tumour groups as well as in the invasive group. Units show log2 of the normalised fluorescence signal. P values and fold changes are between NP and IP (Table 2).

Table 2. Differentially expressed microRNAs between NMI cases in the IP and those in the NP group.
 PFold changeAUC
  1. P values (two-sided unpaired t-test), fold changes (of median normalized fluorescence), and AUC are listed for comparisons between the groups. The table lists microRNAs that passed FDR of 0.1 and had changes > 1.2 fold changes in median expression levels (Fig. 2A). In fold changes, ‘+’ marks higher expression in the first group (IP) and ‘–’ marks lower expression in the first group. MicroRNAs are ordered by P value.
miR-29c*3.0E-062.1 (−)0.77
miR-331-3p1.7E-041.3 (−)0.68
miR-29c4.8E-041.2 (−)0.66
miR-1416.9E-041.2 (−)0.66
miR-2057.4E-041.6 (−)0.73
miR-19b1.1E-031.3 (+)0.67
miR-361-5p1.6E-031.3 (−)0.68
miR-312.7E-032.7 (−)0.69
miR-173.4E-031.5 (+)0.67
miR-106a3.9E-031.5 (+)0.64
miR-254.2E-031.5 (+)0.68
miR-1825.1E-032.3 (−)0.66
miR-130a5.6E-032.2 (+)0.68
miR-148a8.5E-031.4 (−)0.62
let-7b1.1E-021.3 (−)0.66
miR-20a1.9E-021.4 (+)0.65
miR-26b2.7E-021.5 (−)0.62

For assessment of the predictive value of this method we focused on miR-29c* which was the most significantly differentially expressed microRNA between the IP and NP groups (Table 2, Fig. 2). We divided the entire set of 108 NMI bladder cancers into high, intermediate or low expression of miR-29c* and we plotted Kaplan–Meier curves with TTP as an endpoint for the three groups. An increased level of miR-29c* was associated with a lower risk of progression (Fig. 4A). In the 36 cases with high expression (≥9.71, high tertile), only two cases progressed (5.6%) and only one of them progressed within 5 years (2.6%). By contrast, the group with low expression (≤8.71, low tertile), had a median progression-free survival of only 35 months, with a 50% progression rate. Thus, the expression of miR-29c* holds significant value in risk stratifying the patients with NMI bladder cancer (P < 0.001).

Figure 4.

Survival analysis using the expression of miR-29c*. A, Kaplan–Meier plot showing the progression-free survival based on expression of miR-29c*. Data are shown for the 108 NMI cases. The 108 cases are divided into tertiles according to the expression of miR-29c*. The difference in progression-free survival was highly significant (P value 5E-6 by log-rank test). B, Kaplan–Meier plot showing the progression-free survival based on expression of miR-29c* for T1 cases (P = 0.043). Mir-29c* expression threshold set by top tertile as defined in A. C, Same graph as B for high grade cases (P < 0.001): 27 are Ta and 48 are T1. D, Same graph as B for 48 T1 high grade cases (P = 0.024). ND, not determined.

A separate analysis was carried out for patients with T1 tumours (n = 58), high grade tumours (n = 75) and T1 high grade tumours (n = 48), looking at TTP as a function of mir-29c* expression. When using the threshold of 9.71 defined above, mir-29c* maintained prognostic power within each of the stage and grade categories (Fig. 4B,C,D). For example, none of the high grade cases with miR-29c* expression >9.71 progressed, and the TTP between the two resulting groups was highly significant (P < 0.001). Thus we concluded that miR-29c* has prognostic power for identifying progression risk group beyond the prognostic power available in the stage and grade information of the tumour.

Discussion

The purpose of the present study was to identify potential microRNA markers for the prediction of progression of NMI bladder carcinoma. We showed that future progression to carcinoma invading bladder muscle could be predicted using the expression levels of microRNAs. Initial analysis demonstrated that the expression of some microRNAs is highly correlated to surrounding tumour and, specifically, to the percentage of muscle in the sample. In TURBT, the depth of resection is controlled by the surgeon, and hence muscle content in the sample may in reality be affected by clinical factors such as the tumour gross appearance (solid vs papillary) during cytoscopy or the expected aggressiveness of disease as judged by the surgeon based on his previous experience. MicroRNAs which are specifically expressed in muscle (or similarly in the lamina propria), may show prognostic power which is in fact only related to the composition of the tumour and surrounding tissue in the TURBT samples. Thus, we conducted a separate experiment in which we microdissected the tumour and its surrounding tissue components to identify the microRNAs whose expression derives from the tumour, and excluded as potential markers any microRNA derived from the surrounding tissues.

Analysing tumour-related microRNAs only, we found four microRNAs that showed significant and more than twofold differences between NP and IP (miR-29c*, miR-182, miR-130a and miR-31). Significantly higher expression of miR-29c* was detected in NMI bladder tumours that did not progress than in lesions that later progressed. The lower expression of miR-29c* in patients who later progressed was similar to the expression levels seen in patients who already have a cancer invading the bladder muscle. Interestingly miR-29c, which is derived from the same hairpin, was already shown to have lower expression in bladder tumours compared with healthy tissue from thesame patient [32], suggesting that the loss of regulation by this hairpin may be critical in the development of UC. miR-182 and miR-130a also have significant differences in expression between patients that later progress and patients that do not, with miR-182, as mir-29c*, having a good prognostic effect, while the expression of miR-130a is in the opposite direction and its up-regulation is correlated with a more aggressive disease. miR-31 has substantial differences, but its expression is spread widely between patients in all groups. It was previously reported that chromosomal region 9p21 including miR-31 has frequent deletions in UC [23] and we postulate that this is why the expression of this microRNA showed high variability in the present cohort, not necessarily related to a patient's progression.

Several studies have pointed to the potential prognostic value of microRNAs in UC. Wiklund et. al [24]. proposed that the miR-200/141 family and miR-205 silencing are possibly poor prognosis markers. The present data support this suggestion for mir-205 and miR-141 (Table 2) as well as for miR-200a and miR-200c, although the fold changes were <1.2 (data not shown). Catto et al. [21], showed that high grade UC is characterized by up-regulation of miR-21, and low grade UC is characterized by loss of miR-99a/100. We found that miR-21 was indeed significantly higher in IP than in NP, but with a fold change of 1.18 (data not shown). Interestingly, miR-99a/100 were found to be part of the 26 microRNAs predominantly expressed in non-tumour tissue, and hence were removed from further analysis. miR-129, miR-133b, miR-518c* miR-452 and miR-452* which were also suggested as prognostic markers [22, 23] were not found to be expressed in our cohort in either progression groups. This could be attributable either to biological differences between the cohorts used in the different studies, or to differences in the sensitivity and/or specificity of the different platforms used for measuring microRNA expression.

A Kaplan–Meier analysis shows that a simple threshold on the expression level of miR-29c* could identify a group of NMI tumours with a very low risk for progression: only one of 36 cases in this group later progressed. NMI bladder tumours with low expression of miR-29c* had a median progression-free survival of 35 months. Moreover, we saw that the expression level of miR-29c* could provide additional information also in subsets of the cohort which belong to higher risk groups (T1 tumours, high grade tumours and T1 high grade tumours). In some of these subgroups, however, the sample size was limited and a larger validation set is required for higher statistical power. Similarly, a previous study found miR-29c* to be a prognostic marker in mesothelioma [16]. Overexpression of this microRNA in mesothelioma cell lines resulted in significantly decreased proliferation, migration, invasion and colony formation [16]. Pass et. al [16]. showed that mir-29c* acts through downregulation of DNA methyltransferases as well as upregulation of demethylating genes. In several independent studies, aberrant methylation of multiple genes was found to be abundant in bladder cancer tumours and in the urine DNA of patients with bladder cancer [33, 34]. Moreover the methylation profile was found by Brait et al. [33] to be correlated with clinicopathological features of poor prognosis. One may speculate that changes in the expression of mir-29c* may have a role in controlling these changes, acting to keep the normal methylation pattern, and its downregulation is a marker of poor prognosis in UC, and perhaps more generally in cancer.

In summary, the present study identified a pattern of microRNA deregulation in the comparison of latent and aggressive NMI urothelial carcinomas. This pattern was further enhanced in tumours invading the bladder muscle. Tumour progression was associated with altered levels of microRNAs. In particular, high expression levels of miR-29c* were associated with a good prognosis. It would be interesting to study the role of these microRNAs in cancer progression and invasiveness. The usefulness of microRNAs for determining prognosis for patients with UC should be assessed in larger prospective studies, as they could have a substantial impact on the choice of treatment and follow-up of the individual patient.

Acknowledgments

We thank Dr. Tamara Drozd from Rabin Medical Center for her help in reviewing the haematoxylin and eosin slides of all study cases.

Conflict of Interest

Source of funding: Rosette Genomics Ltd. Yael Spector, Eti Meiri, and Yaron Goren are employees of Rosetta Genomics. Yael Spector and Eti Meiri are options holders in Rosetta genomics.

Ancillary