Endobronchial miRNAs as biomarkers in lung cancer chemoprevention.
Journal: 2013/July - Cancer Prevention Research
ISSN: 1940-6215
Abstract:
Lung cancers express lower levels of prostacyclin than normal lung tissues. Prostacyclin prevents lung cancer in a variety of mouse models. A randomized phase II trial comparing oral iloprost (a prostacyclin analog) with placebo in high-risk subjects showed improvement in bronchial histology in former, but not current, smokers. This placebo-controlled study offered the opportunity for investigation of other potential intermediate endpoint and predictive biomarkers to incorporate into chemoprevention trials. Matched bronchial biopsies were obtained at baseline and at 6-month follow-up from 125 high-risk individuals who completed the trial: 31/29 and 37/28 current/former smokers in the iloprost and placebo arm, respectively. We analyzed the expression of 14 selected miRNAs by Real Time PCR in 496 biopsies. The expression of seven miRNAs was significantly correlated with histology at baseline. The expression of miR-34c was inversely correlated with histology at baseline (P < 0.0001) and with change in histology at follow-up (P = 0.0003), independent of treatment or smoking status. Several miRNAs were also found to be differentially expressed in current smokers as compared with former smokers. In current smokers, miR-375 was upregulated at baseline (P < 0.0001) and downregulated after treatment with iloprost (P = 0.0023). No miRNA at baseline reliably predicted a response to iloprost. No biomarker predictive of response to iloprost was found. MiR-34c was inversely correlated with baseline histology and with histology changes. Mir-34c changes at follow-up could be used as a quantitative biomarker that parallels histologic response in formalin-fixed bronchial biopsies in future lung cancer chemoprevention studies.
Relations:
Content
Citations
(5)
References
(34)
Grants
(488)
Diseases
(1)
Drugs
(1)
Chemicals
(4)
Organisms
(1)
Anatomy
(1)
Affiliates
(1)
Similar articles
Articles by the same authors
Discussion board
Cancer Prev Res (Phila) 6(2): 100-108

Endobronchial miRNAs as biomarkers in lung cancer chemoprevention

+6 authors

INTRODUCTION

Smoking exposes the respiratory mucosa to carcinogens in a “field cancerization” process (1). Smokers develop bronchial lesions in a multistep set of pre-invasive stages preceding the development of invasive lung cancer. The lesions are multiple and occur throughout the bronchial airways of smokers, putting the entire bronchial epithelium at risk. The administration of chemopreventive agents may thus be more effective than local treatment in reducing the risk of developing lung cancer. Smoking cessation is the only intervention proven to reduce the risk of lung cancer; however, the risk for former smokers remains elevated for many years compared to never smokers. Therefore, chemoprevention is particularly attractive for former smokers.

Epidemiologic studies have shown decreased rates of certain cancers with the chronic administration of anti-inflammatory medication such as COX-2 inhibitors that alter prostaglandin production (2, 3). Downstream of the COX-1 and COX-2 enzymes, levels of prostacyclins are determined by the expression of prostacyclin synthetase and prostaglandin E2 synthetase. In normal lung, prostacyclin synthetase is active and thus prostacyclin (PGI2) levels are high. In the majority of lung cancers, prostacyclin levels are low while prostaglandin E2 synthetase and prostaglandin E2 (PGE2) levels are high. High PGE2 levels and low PGI2 levels are associated with worse prognosis in lung cancer (4, 5). Selective pulmonary overexpression of prostacyclin synthase and oral iloprost, a stable prostacyclin analog, supplementation prevents the development of lung cancer in a variety of murine models, including cigarette smoke exposure (68). Based on this promising preclinical data, we developed a Phase II chemoprevention trial comparing the oral prostacyclin analog, iloprost, to placebo. This double-blind phase II multicenter chemoprevention trial met the pre-determined primary endpoint of improvement in endobronchial histology in former, but not current, smokers (9). Although there are no established intermediate biomarkers, histology is a logical intermediate endpoint for early chemoprevention trials (10). We have exploited the unique opportunity offered by the iloprost chemoprevention trial to explore additional potential intermediate and/or predictive biomarkers for chemoprevention trials.

MicroRNAs (miRNA) are highly stable and exhibit tissue specificity, making them attractive biomarkers (11). Moreover, we have shown that miRNAs are correlated with the earliest steps in lung squamous cell carcinogenesis (12). The latter study analyzed the expression of 374 miRNAs in biopsies representing various stages of endobronchial premalignancy and showed that 69 miRNAs were differentially expressed in the course of squamous cell lung carcinogenesis (12). This prior work yielded a number of candidate biomarkers that discriminate between squamous cell lung cancer and varying degrees of premalignancy. Thus, as they are modified at the earliest steps of lung carcinogenesis, we hypothesized that miRNAs could be both potential intermediate endpoints and/or predictive biomarkers for chemoprevention trials. We tested this hypothesis in biospecimens from the phase II iloprost lung cancer chemoprevention trial (9). In addition, we investigated associations between miRNA expression and baseline characteristics including histology and smoking status.

MATERIAL AND METHODS

1. Population

The iloprost lung cancer chemoprevention study was a multicenter, phase II, randomized, double-blind, placebo-controlled trial of iloprost in subjects at increased risk for lung cancer (defined as current or former smokers with 20 pack years or more smoking history, at least mild sputum cytologic atypia, or a history of biopsy-proven endobronchial dysplasia). The characteristics of the population and the results have been reported in detail previously (9). Autofluorescence and/or white light bronchoscopy was carried out before randomization and after 6 months of treatment, with 6 standard endobronchial sites biopsied (bronchial carinae, RUL, RML, RB6, LUL, LUDB, and LB6), along with all other visually suspicious areas. The trial sample size was 152 subjects. After obtaining written informed consent, participants were randomized based on smoking status (current versus former) and study center. Following randomization, subjects were treated with iloprost or placebo at an initial dose of 1 tablet BID (50 µg iloprost clathrate per tablet) that was escalated to 3 tablets (150 µg) BID as tolerated. Following 6 months of treatment, a second bronchoscopy was carried out with repeat biopsies at all of the baseline (BL) sites, as well as any new suspicious areas.

Pre-treatment and post-treatment bronchial biopsies were available for all 125 subjects completing the trial. All biopsies were graded according to the WHO classification for bronchial epithelium (13) and scored according to the following system: 1; normal histology; 2, reserve cell hyperplasia; 3, squamous metaplasia; 4, mild dysplasia; 5, moderate dysplasia; 6, severe dysplasia; 7, carcinoma in situ; and 8, invasive carcinoma. All the histopathologic classifications were performed by one pathologist (WAF) and a subset repeated by WAF and by a second (DTM) to determine inter- and intra-observer reproducibility. Figure 1 illustrates normal tissue and dysplasia. Histologies from the same sites were compared before and after treatment. Six to twelve biopsies, with an average of ten, were obtained for each subject.

An external file that holds a picture, illustration, etc.
Object name is nihms431175f1.jpg
Examples of biopsies illustrating histologies

Two biopsies from same patient.

Left: Normal mucosa. Cilia, intact basal cell layer, finely dispersed nuclear chromatin and nuclear membranes rounded and regular.

Right: Moderately dysplastic cells. Cilia absent, basal cell layer no long distinct, nuclear chromatin coarse and irregular an nuclear membranes convoluted.

2. Sample size calculation and sample selection for the present biomarker study

Power was greater than 80% to detect a two-fold or greater change in miRNA expression between two groups with a False Discovery Rate (FDR) of 0.10 across miRNAs. The biopsy sites corresponding to the best (lowest histology score) and worst (highest histology score) at BL were sampled. The corresponding biopsy sites at follow-up (FU) were also selected for study inclusion. We stratified all analyses by smoking status. Using data from our preliminary studies, we found that the 90 percentile of the standard deviations over miRNA and histology groups was approximately 0.71 on the log 2 scale. Assuming that a log 2 difference in miRNA expression between groups (a fold change of 2) is the smallest biological difference of interest and that there are two truly differentially expressed miRNAs in the set, we estimated that a minimum of 13 subjects per group would be needed, whereas for a smaller difference of 0.7 on the log scale a minimum of 24 subjects per group would be needed. For analyses where subjects are the unit of analysis, 125 subjects – 29 former and 31 current smokers in the iloprost arm, and 28 former and 37 current smokers in the placebo arm, were available. We also correlated changes in histology with changes in miRNA expression at the biopsy level. To identify a predictive signature of response, we compared the association between expression and response in the two treatment arms by smoking status. Finally, we correlated changes in histology at the biopsy level with changes in expression of miRNAs.

For each biopsy, a slide with eight sections from a formalin-fixed paraffin embedded (FFPE) block was used for the extraction of the miRNAs. The slide adjacent to the pathological diagnostic slide was used and histology was confirmed by the study pathologist (W.A.F.).

3. Selection of the miRNA analyzed

Fourteen miRNAs were selected from the list of 69 miRNA identified as being differentially expressed during lung squamous carcinogenesis (12) using the following criteria: 1) the miRNA were statistically differentially expressed in at least one comparison involving high grade lesions after correction for multiple testing (CMT); or 2) the miRNA were statistically differentially expressed in at least one comparison involving high-grade lesions (but not necessarily after CMT) and reported as modified in response to lung inflammation (14) and thus biologically relevant for iloprost. In total, 14 miRNA met these criteria and were analyzed: miR-9, miR-34c, miR-101, miR-135a, miR-139, miR-142-3p, miR-196a, miR-196b, miR-199a, miR-214, miR-224, miR-375, miR-452 and miR-487b.

4. Nucleic acid extraction and miRNA analyses

Total RNA was extracted from a slide stored at room temperature and containing eight 4 µm sections. The eight FFPE sections were pulled off the slides and dropped into a 1.5 ml microfuge tube. Total RNA was extracted using the MasterPure RNA Purification kit (Epicentre kit, Biozym) and the manufacturer’s protocol. RNA was stored at −80°C.

The maximum available yield of total RNA (estimated using a standard curve as ranging between 0.11 ng and 135.19 ng, with a median at 9.65 ng) was used for every sample. miRNA were retro-transcribed and amplified (PCR) using the TaqMan MicroRNA Reverse transcription kit (Applied Biosystem). MiRNA expression was analyzed by TaqMan MicroRNA assays (Applied Biosystem) using the following assays: hsa-miR-9 (Assay ID 00583), hsa-miR-34c (000428), hsa-miR-101 (002253), hsa-miR-135a (000460), hsa-miR-139-5p (002289), hsa-miR-142-3p (000464), hsa-miR-196a (241070-mat), hsa-miR-196b (002215), hsa-miR-199a (000498), hsa-miR-214 (002306), hsa-miR-224 (002099), hsa-miR-375 (000564), hsa-miR-452 (002329) and hsa-miR-487b (001285). The small nucleolar house-keeping RNA, RNU48 (SNORD48) (001006), served as the control and was run in triplicate for the large majority of the samples (451=91%) or in duplicate for 45 samples (9%). In our previous study, RNU48 proved to be very stable in bronchial biopsies (12). The amount of RNA from each sample was calibrated to the average value of the Cycle Threshold (CT) values for RNU48 in the same sample. This normalized value gives a delta CT (ΔCT) value for each miRNA relative to the RNU48 (miRNA CT value – RNU48 average CT value). The average ΔCT was then calculated for each group of samples and the delta ΔCT (ΔΔCT), which corresponds to the differences between two groups, was obtained by subtracting the average ΔCT of the second group from that of the first group. Fold differences for up-regulated miRNAs were calculated using 2, as a decrease in 1 CT value is equivalent to a two-fold increase in the amount of starting cDNA. For the down-regulated miRNAs, fold differences were calculated using 1/2. By definition, a higher ΔCT value represents a lower expression and vice versa.

In a sample exhibiting a CT value either above 37 or undetermined for RNU48 would be considered as non-evaluable. All samples were deemed evaluable. Undetermined ΔCT values for the miRNA assays were discarded from further analyses as recommended (15).

5. Statistical analyses

Welch's (i.e. unequal variance) t-test p-values were used to assess the statistical significance of the difference in miRNA expression between specific groups, i.e. between iloprost and placebo treated subjects, before and after treatment and between responders and non-responders treated with iloprost (16). We stratified all analyses by smoking status (current versus former). Logistic regression models were used to analyze the relationship between baseline or change in miRNA expression and response status. All logistic regression models were adjusted for age, sex and pack years. Models for former smokers were also adjusted for number of years since smoking cessation. Assessment of the relationships between BL histology and ΔCT and change in histology and ΔΔCT was assessed using Spearman correlation coefficients.

The False Discovery Rate (FDR) was set to 0.1 for all analyses. FDR adjustments to p-values were applied across the 14 miRNAs for each set of comparisons (e.g. responder vs. non-responder).

Histological response was evaluated using two different criteria. The first criterion termed the “worst histology score criterion” used one definition from the clinical trial: a decrease of 1 or more in worst histology score from the same patient after treatment as compared to before (9). The second is the “paired sample based criterion”, and response is defined by a decrease of one histological grade within matched paired biopsies taken at the same site before and after treatment.

The analyses for this study were generated by two biostatisticians (W.J.F and A.B.) using SAS software, Version 9.2 of the SAS System for Windows Copyright © 2011 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA.

1. Population

The iloprost lung cancer chemoprevention study was a multicenter, phase II, randomized, double-blind, placebo-controlled trial of iloprost in subjects at increased risk for lung cancer (defined as current or former smokers with 20 pack years or more smoking history, at least mild sputum cytologic atypia, or a history of biopsy-proven endobronchial dysplasia). The characteristics of the population and the results have been reported in detail previously (9). Autofluorescence and/or white light bronchoscopy was carried out before randomization and after 6 months of treatment, with 6 standard endobronchial sites biopsied (bronchial carinae, RUL, RML, RB6, LUL, LUDB, and LB6), along with all other visually suspicious areas. The trial sample size was 152 subjects. After obtaining written informed consent, participants were randomized based on smoking status (current versus former) and study center. Following randomization, subjects were treated with iloprost or placebo at an initial dose of 1 tablet BID (50 µg iloprost clathrate per tablet) that was escalated to 3 tablets (150 µg) BID as tolerated. Following 6 months of treatment, a second bronchoscopy was carried out with repeat biopsies at all of the baseline (BL) sites, as well as any new suspicious areas.

Pre-treatment and post-treatment bronchial biopsies were available for all 125 subjects completing the trial. All biopsies were graded according to the WHO classification for bronchial epithelium (13) and scored according to the following system: 1; normal histology; 2, reserve cell hyperplasia; 3, squamous metaplasia; 4, mild dysplasia; 5, moderate dysplasia; 6, severe dysplasia; 7, carcinoma in situ; and 8, invasive carcinoma. All the histopathologic classifications were performed by one pathologist (WAF) and a subset repeated by WAF and by a second (DTM) to determine inter- and intra-observer reproducibility. Figure 1 illustrates normal tissue and dysplasia. Histologies from the same sites were compared before and after treatment. Six to twelve biopsies, with an average of ten, were obtained for each subject.

An external file that holds a picture, illustration, etc.
Object name is nihms431175f1.jpg
Examples of biopsies illustrating histologies

Two biopsies from same patient.

Left: Normal mucosa. Cilia, intact basal cell layer, finely dispersed nuclear chromatin and nuclear membranes rounded and regular.

Right: Moderately dysplastic cells. Cilia absent, basal cell layer no long distinct, nuclear chromatin coarse and irregular an nuclear membranes convoluted.

2. Sample size calculation and sample selection for the present biomarker study

Power was greater than 80% to detect a two-fold or greater change in miRNA expression between two groups with a False Discovery Rate (FDR) of 0.10 across miRNAs. The biopsy sites corresponding to the best (lowest histology score) and worst (highest histology score) at BL were sampled. The corresponding biopsy sites at follow-up (FU) were also selected for study inclusion. We stratified all analyses by smoking status. Using data from our preliminary studies, we found that the 90 percentile of the standard deviations over miRNA and histology groups was approximately 0.71 on the log 2 scale. Assuming that a log 2 difference in miRNA expression between groups (a fold change of 2) is the smallest biological difference of interest and that there are two truly differentially expressed miRNAs in the set, we estimated that a minimum of 13 subjects per group would be needed, whereas for a smaller difference of 0.7 on the log scale a minimum of 24 subjects per group would be needed. For analyses where subjects are the unit of analysis, 125 subjects – 29 former and 31 current smokers in the iloprost arm, and 28 former and 37 current smokers in the placebo arm, were available. We also correlated changes in histology with changes in miRNA expression at the biopsy level. To identify a predictive signature of response, we compared the association between expression and response in the two treatment arms by smoking status. Finally, we correlated changes in histology at the biopsy level with changes in expression of miRNAs.

For each biopsy, a slide with eight sections from a formalin-fixed paraffin embedded (FFPE) block was used for the extraction of the miRNAs. The slide adjacent to the pathological diagnostic slide was used and histology was confirmed by the study pathologist (W.A.F.).

3. Selection of the miRNA analyzed

Fourteen miRNAs were selected from the list of 69 miRNA identified as being differentially expressed during lung squamous carcinogenesis (12) using the following criteria: 1) the miRNA were statistically differentially expressed in at least one comparison involving high grade lesions after correction for multiple testing (CMT); or 2) the miRNA were statistically differentially expressed in at least one comparison involving high-grade lesions (but not necessarily after CMT) and reported as modified in response to lung inflammation (14) and thus biologically relevant for iloprost. In total, 14 miRNA met these criteria and were analyzed: miR-9, miR-34c, miR-101, miR-135a, miR-139, miR-142-3p, miR-196a, miR-196b, miR-199a, miR-214, miR-224, miR-375, miR-452 and miR-487b.

4. Nucleic acid extraction and miRNA analyses

Total RNA was extracted from a slide stored at room temperature and containing eight 4 µm sections. The eight FFPE sections were pulled off the slides and dropped into a 1.5 ml microfuge tube. Total RNA was extracted using the MasterPure RNA Purification kit (Epicentre kit, Biozym) and the manufacturer’s protocol. RNA was stored at −80°C.

The maximum available yield of total RNA (estimated using a standard curve as ranging between 0.11 ng and 135.19 ng, with a median at 9.65 ng) was used for every sample. miRNA were retro-transcribed and amplified (PCR) using the TaqMan MicroRNA Reverse transcription kit (Applied Biosystem). MiRNA expression was analyzed by TaqMan MicroRNA assays (Applied Biosystem) using the following assays: hsa-miR-9 (Assay ID 00583), hsa-miR-34c (000428), hsa-miR-101 (002253), hsa-miR-135a (000460), hsa-miR-139-5p (002289), hsa-miR-142-3p (000464), hsa-miR-196a (241070-mat), hsa-miR-196b (002215), hsa-miR-199a (000498), hsa-miR-214 (002306), hsa-miR-224 (002099), hsa-miR-375 (000564), hsa-miR-452 (002329) and hsa-miR-487b (001285). The small nucleolar house-keeping RNA, RNU48 (SNORD48) (001006), served as the control and was run in triplicate for the large majority of the samples (451=91%) or in duplicate for 45 samples (9%). In our previous study, RNU48 proved to be very stable in bronchial biopsies (12). The amount of RNA from each sample was calibrated to the average value of the Cycle Threshold (CT) values for RNU48 in the same sample. This normalized value gives a delta CT (ΔCT) value for each miRNA relative to the RNU48 (miRNA CT value – RNU48 average CT value). The average ΔCT was then calculated for each group of samples and the delta ΔCT (ΔΔCT), which corresponds to the differences between two groups, was obtained by subtracting the average ΔCT of the second group from that of the first group. Fold differences for up-regulated miRNAs were calculated using 2, as a decrease in 1 CT value is equivalent to a two-fold increase in the amount of starting cDNA. For the down-regulated miRNAs, fold differences were calculated using 1/2. By definition, a higher ΔCT value represents a lower expression and vice versa.

In a sample exhibiting a CT value either above 37 or undetermined for RNU48 would be considered as non-evaluable. All samples were deemed evaluable. Undetermined ΔCT values for the miRNA assays were discarded from further analyses as recommended (15).

5. Statistical analyses

Welch's (i.e. unequal variance) t-test p-values were used to assess the statistical significance of the difference in miRNA expression between specific groups, i.e. between iloprost and placebo treated subjects, before and after treatment and between responders and non-responders treated with iloprost (16). We stratified all analyses by smoking status (current versus former). Logistic regression models were used to analyze the relationship between baseline or change in miRNA expression and response status. All logistic regression models were adjusted for age, sex and pack years. Models for former smokers were also adjusted for number of years since smoking cessation. Assessment of the relationships between BL histology and ΔCT and change in histology and ΔΔCT was assessed using Spearman correlation coefficients.

The False Discovery Rate (FDR) was set to 0.1 for all analyses. FDR adjustments to p-values were applied across the 14 miRNAs for each set of comparisons (e.g. responder vs. non-responder).

Histological response was evaluated using two different criteria. The first criterion termed the “worst histology score criterion” used one definition from the clinical trial: a decrease of 1 or more in worst histology score from the same patient after treatment as compared to before (9). The second is the “paired sample based criterion”, and response is defined by a decrease of one histological grade within matched paired biopsies taken at the same site before and after treatment.

The analyses for this study were generated by two biostatisticians (W.J.F and A.B.) using SAS software, Version 9.2 of the SAS System for Windows Copyright © 2011 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA.

RESULTS

1. Characteristics of the study population

A total of 496 out of the 500 bronchial biopsies requested from the Colorado SPORE Tissue Bank had sufficient tissue for this study: 248 samples at BL and 248 samples at FU. The distribution of bronchial histologies at BL and FU based on treatment arms and stratified by smoking status is shown in online supplemental table 1. In the whole cohort, the distribution of the histology was: 37% normal mucosa (39% and 36% of BL and FU samples, respectively), 23% hyperplasia (19%/26%), 3% metaplasia (1%/5%), 14% mild dysplasia (13%/14%), 21% moderate dysplasia (22%/19%) and 3% severe dysplasia (6%/1%). Moderate and severe dysplasias were more frequent in the current smokers (80/270 - 30%) compared to former smokers (38/226 - 17%) (p=0.001) and in BL biopsies (69/248 - 28%) compared to FU biopsies (49/248 - 20%) (p=0.0448). The distribution of histologies within the pairs of samples (best (lowest score) or worst (highest score) histology at BL) is shown in online supplemental table 2. The very large majority of biopsies (93%) with the best histology at BL have either a normal histology (78/125 - 63%) or a reserve cell hyperplasia (38/125 - 30%).

The histological response rates per arm, stratified by smoking status, are reported in online supplemental table 3. Based on the “worst histology score criterion” for response, there were more responders in former than in current smokers (48 vs. 33%) in the iloprost arm and inversely (15 vs. 24%) in the placebo arm. The response rate was lower in the placebo arm than in the iloprost arm as presented previously (9). Based on the “paired sample based criterion”, the response rate was much higher in the pairs of biopsies with the worst histology at BL than in those with best histology at BL, both in former smokers (response rate 55% vs. 7% for pairs of biopsies with the worse or the best histology at BL) and in current smokers (67% vs. 6%) in the iloprost arm. We observed the same in the placebo arm with response rate of 37% vs. 7% in former smokers and 49% vs. 11% in current smokers. This is at least partially explained by the large number of samples with normal histology in the pairs with best histology at BL that makes response impossible, as the histological score cannot improve. As shown above, the response rates are slightly better and the differences are slightly larger in the iloprost than in the placebo arm.

2. MiRNA expression associations with baseline characteristics

a. Association with baseline histology

At BL, seven miRNA were found significantly correlated to histology stages after FDR adjustment: miR-452, miR-224, miR-101, miR-139-5p, miR-199a, miR-214 and miR-34c. The expression of two miRNA was directly correlated with the stage of histology from lower to higher (CT values inversely correlated thus expression positively correlated): miR-452 (r= −0.43, p<0.001) and miR-224 (r= −0.54, p<0.001). The expression of the five other miRNAs was inversely correlated (CT values directly correlated thus expression inversely correlated) with the stage of histology: miR-101 (r=0.20, p=0.0013), miR-139-5p (r=0.18, p=0.0042), miR-199a (r=0.20, p=0.0024), miR-214 (r=0.18, p=0.0060), miR-34c (r=0.36, p<0.0001).

b. Association with current versus former smoker status

Several miRNAs showed a significantly different expression at BL in current smokers compared with former smokers. Three miRNAs had a significantly higher expression in current than in former smokers after FDR adjustment: miR-224 (p<0.0001), miR-375 (p<0.0001) and miR-452 (p<0.0001). Two miRNAs had a lower expression in current compared to former smokers after FDR adjustment: miR-142-3p (p=0.0069) and miR-34c (p=0.0550). We further restricted comparison of miRNA expression between current and former smokers to biopsies with a normal histology at BL, in order to control for the more frequent dysplasia in current smokers. In biopsies with normal histology, the up-regulation of miR-224 (p=0.0003), miR-375 (p<0.0001) and miR-452 (p<0.0013) was confirmed and miR-9 (p=0.0278) and miR-487b (p=0.0290) were also found significantly up-regulated in current smokers vs former smokers, all after FDR adjustment. The down-regulation of miR-142-3p and miR-34c in current smokers was not found to be significant in normal biopsies.

c. Association with treatment arm

There was no significant difference in miRNA expression at BL between the samples of the iloprost arm and those of the placebo arm for any of the miRNA tested.

3. Expression changes between baseline and follow-up biopsies

a. Changes specific to iloprost administration

The expression of miR-375 was not correlated with histology at BL or histology changes at FU but was significantly up-regulated in current compared to former smokers. In current smokers, the expression of miR-375 was significantly down-regulated after treatment by iloprost (p=0.0023 after FDR adjustment) (Table 1). Furthermore, the down-regulation of miR-375 in current smokers after treatment by iloprost was significantly different (p=0.0011 after FDR adjustment) from its change in current smokers who received the placebo, where a slight but not significant up-regulation was observed. Thus, the down-regulation of miR-375 in current smokers appears to be specific to iloprost treatment.

Table 1

Significant changes in miRNA expression in FU samples as compared with BL samples

miRNASignificant changes in miRNA expression in FU versus BL samples (Welch’s t-test)
AllBoth smoker groupsCurrent smokersFormer smokers
IloprostPlaceboIloprostPlaceboIloprostPlacebo
miR-9Down
P<0.0001
FDR
Down
P=0.0007
FDR
Down
P=0.001
FDR
Down
P=0.0022
FDR
Down
P=0.0476
Down
Not
significant
Down
P=0.0071
FDR
miR-375Not
significant
Not
significant
Not
significant
Down
P=0.0023
FDR
Not
significant
Not
significant
Down
P=0.0364

This table reports the significant modification of miRNA expression in follow-up (FU) samples as compared with baseline (BL) samples matched from the same site. Only miR-9 and miR-375 showed significant changes in expression level at FU. The table shows up- or down-regulation at FU compared to BL. The significant p-values using Welch’s t-test are reported. “FDR” indicates that the p value is significant after adjustment for false discovery rate (FDR).

b. Association with histologic changes

We assessed the correlation between changes in miRNA expressions and changes in histology stages between BL and FU biopsies. In the whole cohort, the changes in expression of two miRNAs, miR-34c and miR-224, were significantly correlated after FDR adjustment with histology changes between BL and FU. Including all patients in the analyses, the expression of miR-224 was up-regulated in higher histologic grades (r= −0.22, p=0.0006) (Figure 2). This was also shown in one subgroup, the former smokers in the iloprost arm (r= −0.37, p=0.0044) (Figure 2). The changes in expression of miR-34c were inversely correlated with histology changes, as miR-34c expression was down-regulated in higher histologic grades. This correlation between miR-34c expression and histology changes is shown when including all samples (r=0.23 p=0.0003). A significant correlation (but not after FDR adjustment) was also consistently seen in most subgroups (iloprost arm, current r=0.26, p=0.041 and former smokers r=0.24, p=0.0666; placebo arm, current smokers r=0.23, p=0.0466). These results are illustrated in table 2.

An external file that holds a picture, illustration, etc.
Object name is nihms431175f2.jpg
Distribution of changes in miRNA expression and changes in histology stage

This figure illustrates the distribution of the changes in miRNA expression (represented by the change in delta CT) and the change in histology stage. Each panel illustrates a different miRNA. On the X axis, “0” represents no change in histology, negative and positive numbers represent improvement or worsening in histology, respectively, with the number showing the number of stages that changed. Example: −2 = down-grading of 2 histology units from severe to mild dysplasia or from moderate dysplasia to squamous metaplasia. On the Y axis, the change in delta CT is the difference between the delta CT at follow-up and the delta CT at baseline. By definition, a higher change in delta CT value represents a lower expression and vice versa. The circles represent the different samples. The line shows the linear regression of miRNA expression on histology stage. Only two miRNA had expression changes significantly correlated with histology changes: miR-224 (r= −0.22, p=0.0006) and miR-34c (r=0.23 p=0.0003), up- and down-regulated, respectively, as histology worsens.

Table 2

Correlation between miRNA expression changes and histology changes at FU

miRNABLsamplesPaired samples: FU expression compared to baseline
Spearman
correlation
between
miRNA
expression and
histology grade
(low to high)
Spearman correlation between miRNA expression changes and histology change (up-grading) at FU
compared to baseline
AllCurrent smokersFormer smokers
iloprostPlaceboIloprostPlacebo
miR-34cDown
r=0.36
p<0.0001, FDR
Down
r=0.23
p=0.0003, FDR
Down
r=0.26
p=0.041
Down
r=0.23
p=0.0466
Down
r=0.24
Not significant
Down
r=0.15
Not significant
miR-224Up
r=−0.54
p<0.001, FDR
Up
r=−0.22
p=0.0006, FDR
Up
r=−0.15
Not significant
Up
r=0.11
Not significant
Up
r=−0.37
p=0.0044, FDR
Up
r=−0.25
Not significant

This table reports the correlations between miRNA expression and histology at BL and between miRNA expression changes and histology changes at FU for miR-34c and miR-224 as these two miRNA are the ones whose expression changes are correlated with the histology changes. The table shows up- or down-regulation with regard to histology up-grading at FU compared to baseline. The Spearman correlation coefficient (R) is negative when the CT values are inversely correlated with histology up-grading and thus reflect a positively correlated expression. The Spearman correlation coefficient is positive when the CT values are positively correlated with histology up-grading and thus reflect an inversely correlated expression. Significant p-values are reported. “FDR” indicates that the p-value is significant after adjustment for False Discovery Rate (FDR).

c. Down-regulation of miR-9 expression after bronchial biopsy

MiR-9 expression was not correlated with histology at BL (r=0.01, p=0.8600) or with histology changes between BL and FU (r=0.04, p=0.5074). However, miR-9 was found to be significantly down-regulated in FU as compared with BL biopsies. The down-regulation of miR-9 expression in FU versus BL samples was significant after FDR adjustment when analyzing all samples (Welch t test, p<0.0001) and in most of subgroups (table 1): in the iloprost arm, all patients (p=0.0007) and current smokers (p=0.0022) and in the placebo arm, all patients (p=0.001), and, former smokers (p=0.0071). In current smokers in the placebo arm (p=0.0476) and in the former smokers in the iloprost arm (p=0.0868), the same trend was observed, but did not achieve statistical significance after FDR adjustment. Thus, miR-9 is consistently down-regulated in FU biopsies 6 months later at the same site independently of histology, treatment or smoking status.

4. Comparison of miRNA expression between histological responders and non-responders

Significant differences at BL between responders and non-responders in both the iloprost and placebo arms were shown for the expression of two miRNAs that are correlated with histology. A lower expression of miR-34c at BL was detected in responders versus non-responders both in the iloprost and placebo arms and was very consistent with OR ranging between 1.38 and 1.49. A significantly higher expression of miR-224 at BL was also shown in responders compared to non-responders in several subgroups. Detailed results and OR in different subgroups are illustrated in online supplemental table 4.

In current smokers in the placebo arm, a lower expression of miR-375 at BL predicted response using the “worst histology score criterion” in all samples (p=0.0012 after FDR adjustment with OR=1.93 (95% CI: 1.10–3.39, p=0.0219) as well as in the “paired sample based criterion” (p=0.0027 after FDR adjustment and OR=3.74 (95% CI: 1.27–10.98, p=0.0165).

We tested the predictive value of miRNA expressions using interaction terms between baseline expression and treatment arm in the logistic regression models of response. No individual miRNAs were found to be predictive of response to iloprost. We examined the group of miRNAs to determine if they together or as subsets were associated with response overall and then, if they were differentially associated with response in the Iloprost vs. placebo group. We did this in the former smokers only since that was the group in which Iloprost showed a beneficial effect on histology. We evaluated three models: one in which the definition of response for each biopsy was at the participant level, one in which the definition of response was at the biopsy level, and one in which the response was defined at the biopsy level and excluded the biopsies that were normal at baseline. In the first model one miRNA was associated with response after adjusting for all of the design variables (treatment, pack years of smoking, years since quitting, age and sex) but it was not predictive (p = 0.98 for the 1 df test of the interaction between miRNA and treatment group). In the second model, there were 4 miRNA that were together independently associated with response after adjustment but were not predictive (p = 0.4 for the 4 df test of the interaction between miRNA and treatment group), and in the third model, no miRNA was associated with response after adjustment. Thus, we conclude that the miRNAs applied are not predictive of response to Iloprost, neither as single markers or in combinations.

1. Characteristics of the study population

A total of 496 out of the 500 bronchial biopsies requested from the Colorado SPORE Tissue Bank had sufficient tissue for this study: 248 samples at BL and 248 samples at FU. The distribution of bronchial histologies at BL and FU based on treatment arms and stratified by smoking status is shown in online supplemental table 1. In the whole cohort, the distribution of the histology was: 37% normal mucosa (39% and 36% of BL and FU samples, respectively), 23% hyperplasia (19%/26%), 3% metaplasia (1%/5%), 14% mild dysplasia (13%/14%), 21% moderate dysplasia (22%/19%) and 3% severe dysplasia (6%/1%). Moderate and severe dysplasias were more frequent in the current smokers (80/270 - 30%) compared to former smokers (38/226 - 17%) (p=0.001) and in BL biopsies (69/248 - 28%) compared to FU biopsies (49/248 - 20%) (p=0.0448). The distribution of histologies within the pairs of samples (best (lowest score) or worst (highest score) histology at BL) is shown in online supplemental table 2. The very large majority of biopsies (93%) with the best histology at BL have either a normal histology (78/125 - 63%) or a reserve cell hyperplasia (38/125 - 30%).

The histological response rates per arm, stratified by smoking status, are reported in online supplemental table 3. Based on the “worst histology score criterion” for response, there were more responders in former than in current smokers (48 vs. 33%) in the iloprost arm and inversely (15 vs. 24%) in the placebo arm. The response rate was lower in the placebo arm than in the iloprost arm as presented previously (9). Based on the “paired sample based criterion”, the response rate was much higher in the pairs of biopsies with the worst histology at BL than in those with best histology at BL, both in former smokers (response rate 55% vs. 7% for pairs of biopsies with the worse or the best histology at BL) and in current smokers (67% vs. 6%) in the iloprost arm. We observed the same in the placebo arm with response rate of 37% vs. 7% in former smokers and 49% vs. 11% in current smokers. This is at least partially explained by the large number of samples with normal histology in the pairs with best histology at BL that makes response impossible, as the histological score cannot improve. As shown above, the response rates are slightly better and the differences are slightly larger in the iloprost than in the placebo arm.

2. MiRNA expression associations with baseline characteristics

a. Association with baseline histology

At BL, seven miRNA were found significantly correlated to histology stages after FDR adjustment: miR-452, miR-224, miR-101, miR-139-5p, miR-199a, miR-214 and miR-34c. The expression of two miRNA was directly correlated with the stage of histology from lower to higher (CT values inversely correlated thus expression positively correlated): miR-452 (r= −0.43, p<0.001) and miR-224 (r= −0.54, p<0.001). The expression of the five other miRNAs was inversely correlated (CT values directly correlated thus expression inversely correlated) with the stage of histology: miR-101 (r=0.20, p=0.0013), miR-139-5p (r=0.18, p=0.0042), miR-199a (r=0.20, p=0.0024), miR-214 (r=0.18, p=0.0060), miR-34c (r=0.36, p<0.0001).

b. Association with current versus former smoker status

Several miRNAs showed a significantly different expression at BL in current smokers compared with former smokers. Three miRNAs had a significantly higher expression in current than in former smokers after FDR adjustment: miR-224 (p<0.0001), miR-375 (p<0.0001) and miR-452 (p<0.0001). Two miRNAs had a lower expression in current compared to former smokers after FDR adjustment: miR-142-3p (p=0.0069) and miR-34c (p=0.0550). We further restricted comparison of miRNA expression between current and former smokers to biopsies with a normal histology at BL, in order to control for the more frequent dysplasia in current smokers. In biopsies with normal histology, the up-regulation of miR-224 (p=0.0003), miR-375 (p<0.0001) and miR-452 (p<0.0013) was confirmed and miR-9 (p=0.0278) and miR-487b (p=0.0290) were also found significantly up-regulated in current smokers vs former smokers, all after FDR adjustment. The down-regulation of miR-142-3p and miR-34c in current smokers was not found to be significant in normal biopsies.

c. Association with treatment arm

There was no significant difference in miRNA expression at BL between the samples of the iloprost arm and those of the placebo arm for any of the miRNA tested.

a. Association with baseline histology

At BL, seven miRNA were found significantly correlated to histology stages after FDR adjustment: miR-452, miR-224, miR-101, miR-139-5p, miR-199a, miR-214 and miR-34c. The expression of two miRNA was directly correlated with the stage of histology from lower to higher (CT values inversely correlated thus expression positively correlated): miR-452 (r= −0.43, p<0.001) and miR-224 (r= −0.54, p<0.001). The expression of the five other miRNAs was inversely correlated (CT values directly correlated thus expression inversely correlated) with the stage of histology: miR-101 (r=0.20, p=0.0013), miR-139-5p (r=0.18, p=0.0042), miR-199a (r=0.20, p=0.0024), miR-214 (r=0.18, p=0.0060), miR-34c (r=0.36, p<0.0001).

b. Association with current versus former smoker status

Several miRNAs showed a significantly different expression at BL in current smokers compared with former smokers. Three miRNAs had a significantly higher expression in current than in former smokers after FDR adjustment: miR-224 (p<0.0001), miR-375 (p<0.0001) and miR-452 (p<0.0001). Two miRNAs had a lower expression in current compared to former smokers after FDR adjustment: miR-142-3p (p=0.0069) and miR-34c (p=0.0550). We further restricted comparison of miRNA expression between current and former smokers to biopsies with a normal histology at BL, in order to control for the more frequent dysplasia in current smokers. In biopsies with normal histology, the up-regulation of miR-224 (p=0.0003), miR-375 (p<0.0001) and miR-452 (p<0.0013) was confirmed and miR-9 (p=0.0278) and miR-487b (p=0.0290) were also found significantly up-regulated in current smokers vs former smokers, all after FDR adjustment. The down-regulation of miR-142-3p and miR-34c in current smokers was not found to be significant in normal biopsies.

c. Association with treatment arm

There was no significant difference in miRNA expression at BL between the samples of the iloprost arm and those of the placebo arm for any of the miRNA tested.

3. Expression changes between baseline and follow-up biopsies

a. Changes specific to iloprost administration

The expression of miR-375 was not correlated with histology at BL or histology changes at FU but was significantly up-regulated in current compared to former smokers. In current smokers, the expression of miR-375 was significantly down-regulated after treatment by iloprost (p=0.0023 after FDR adjustment) (Table 1). Furthermore, the down-regulation of miR-375 in current smokers after treatment by iloprost was significantly different (p=0.0011 after FDR adjustment) from its change in current smokers who received the placebo, where a slight but not significant up-regulation was observed. Thus, the down-regulation of miR-375 in current smokers appears to be specific to iloprost treatment.

Table 1

Significant changes in miRNA expression in FU samples as compared with BL samples

miRNASignificant changes in miRNA expression in FU versus BL samples (Welch’s t-test)
AllBoth smoker groupsCurrent smokersFormer smokers
IloprostPlaceboIloprostPlaceboIloprostPlacebo
miR-9Down
P<0.0001
FDR
Down
P=0.0007
FDR
Down
P=0.001
FDR
Down
P=0.0022
FDR
Down
P=0.0476
Down
Not
significant
Down
P=0.0071
FDR
miR-375Not
significant
Not
significant
Not
significant
Down
P=0.0023
FDR
Not
significant
Not
significant
Down
P=0.0364

This table reports the significant modification of miRNA expression in follow-up (FU) samples as compared with baseline (BL) samples matched from the same site. Only miR-9 and miR-375 showed significant changes in expression level at FU. The table shows up- or down-regulation at FU compared to BL. The significant p-values using Welch’s t-test are reported. “FDR” indicates that the p value is significant after adjustment for false discovery rate (FDR).

b. Association with histologic changes

We assessed the correlation between changes in miRNA expressions and changes in histology stages between BL and FU biopsies. In the whole cohort, the changes in expression of two miRNAs, miR-34c and miR-224, were significantly correlated after FDR adjustment with histology changes between BL and FU. Including all patients in the analyses, the expression of miR-224 was up-regulated in higher histologic grades (r= −0.22, p=0.0006) (Figure 2). This was also shown in one subgroup, the former smokers in the iloprost arm (r= −0.37, p=0.0044) (Figure 2). The changes in expression of miR-34c were inversely correlated with histology changes, as miR-34c expression was down-regulated in higher histologic grades. This correlation between miR-34c expression and histology changes is shown when including all samples (r=0.23 p=0.0003). A significant correlation (but not after FDR adjustment) was also consistently seen in most subgroups (iloprost arm, current r=0.26, p=0.041 and former smokers r=0.24, p=0.0666; placebo arm, current smokers r=0.23, p=0.0466). These results are illustrated in table 2.

An external file that holds a picture, illustration, etc.
Object name is nihms431175f2.jpg
Distribution of changes in miRNA expression and changes in histology stage

This figure illustrates the distribution of the changes in miRNA expression (represented by the change in delta CT) and the change in histology stage. Each panel illustrates a different miRNA. On the X axis, “0” represents no change in histology, negative and positive numbers represent improvement or worsening in histology, respectively, with the number showing the number of stages that changed. Example: −2 = down-grading of 2 histology units from severe to mild dysplasia or from moderate dysplasia to squamous metaplasia. On the Y axis, the change in delta CT is the difference between the delta CT at follow-up and the delta CT at baseline. By definition, a higher change in delta CT value represents a lower expression and vice versa. The circles represent the different samples. The line shows the linear regression of miRNA expression on histology stage. Only two miRNA had expression changes significantly correlated with histology changes: miR-224 (r= −0.22, p=0.0006) and miR-34c (r=0.23 p=0.0003), up- and down-regulated, respectively, as histology worsens.

Table 2

Correlation between miRNA expression changes and histology changes at FU

miRNABLsamplesPaired samples: FU expression compared to baseline
Spearman
correlation
between
miRNA
expression and
histology grade
(low to high)
Spearman correlation between miRNA expression changes and histology change (up-grading) at FU
compared to baseline
AllCurrent smokersFormer smokers
iloprostPlaceboIloprostPlacebo
miR-34cDown
r=0.36
p<0.0001, FDR
Down
r=0.23
p=0.0003, FDR
Down
r=0.26
p=0.041
Down
r=0.23
p=0.0466
Down
r=0.24
Not significant
Down
r=0.15
Not significant
miR-224Up
r=−0.54
p<0.001, FDR
Up
r=−0.22
p=0.0006, FDR
Up
r=−0.15
Not significant
Up
r=0.11
Not significant
Up
r=−0.37
p=0.0044, FDR
Up
r=−0.25
Not significant

This table reports the correlations between miRNA expression and histology at BL and between miRNA expression changes and histology changes at FU for miR-34c and miR-224 as these two miRNA are the ones whose expression changes are correlated with the histology changes. The table shows up- or down-regulation with regard to histology up-grading at FU compared to baseline. The Spearman correlation coefficient (R) is negative when the CT values are inversely correlated with histology up-grading and thus reflect a positively correlated expression. The Spearman correlation coefficient is positive when the CT values are positively correlated with histology up-grading and thus reflect an inversely correlated expression. Significant p-values are reported. “FDR” indicates that the p-value is significant after adjustment for False Discovery Rate (FDR).

c. Down-regulation of miR-9 expression after bronchial biopsy

MiR-9 expression was not correlated with histology at BL (r=0.01, p=0.8600) or with histology changes between BL and FU (r=0.04, p=0.5074). However, miR-9 was found to be significantly down-regulated in FU as compared with BL biopsies. The down-regulation of miR-9 expression in FU versus BL samples was significant after FDR adjustment when analyzing all samples (Welch t test, p<0.0001) and in most of subgroups (table 1): in the iloprost arm, all patients (p=0.0007) and current smokers (p=0.0022) and in the placebo arm, all patients (p=0.001), and, former smokers (p=0.0071). In current smokers in the placebo arm (p=0.0476) and in the former smokers in the iloprost arm (p=0.0868), the same trend was observed, but did not achieve statistical significance after FDR adjustment. Thus, miR-9 is consistently down-regulated in FU biopsies 6 months later at the same site independently of histology, treatment or smoking status.

a. Changes specific to iloprost administration

The expression of miR-375 was not correlated with histology at BL or histology changes at FU but was significantly up-regulated in current compared to former smokers. In current smokers, the expression of miR-375 was significantly down-regulated after treatment by iloprost (p=0.0023 after FDR adjustment) (Table 1). Furthermore, the down-regulation of miR-375 in current smokers after treatment by iloprost was significantly different (p=0.0011 after FDR adjustment) from its change in current smokers who received the placebo, where a slight but not significant up-regulation was observed. Thus, the down-regulation of miR-375 in current smokers appears to be specific to iloprost treatment.

Table 1

Significant changes in miRNA expression in FU samples as compared with BL samples

miRNASignificant changes in miRNA expression in FU versus BL samples (Welch’s t-test)
AllBoth smoker groupsCurrent smokersFormer smokers
IloprostPlaceboIloprostPlaceboIloprostPlacebo
miR-9Down
P<0.0001
FDR
Down
P=0.0007
FDR
Down
P=0.001
FDR
Down
P=0.0022
FDR
Down
P=0.0476
Down
Not
significant
Down
P=0.0071
FDR
miR-375Not
significant
Not
significant
Not
significant
Down
P=0.0023
FDR
Not
significant
Not
significant
Down
P=0.0364

This table reports the significant modification of miRNA expression in follow-up (FU) samples as compared with baseline (BL) samples matched from the same site. Only miR-9 and miR-375 showed significant changes in expression level at FU. The table shows up- or down-regulation at FU compared to BL. The significant p-values using Welch’s t-test are reported. “FDR” indicates that the p value is significant after adjustment for false discovery rate (FDR).

b. Association with histologic changes

We assessed the correlation between changes in miRNA expressions and changes in histology stages between BL and FU biopsies. In the whole cohort, the changes in expression of two miRNAs, miR-34c and miR-224, were significantly correlated after FDR adjustment with histology changes between BL and FU. Including all patients in the analyses, the expression of miR-224 was up-regulated in higher histologic grades (r= −0.22, p=0.0006) (Figure 2). This was also shown in one subgroup, the former smokers in the iloprost arm (r= −0.37, p=0.0044) (Figure 2). The changes in expression of miR-34c were inversely correlated with histology changes, as miR-34c expression was down-regulated in higher histologic grades. This correlation between miR-34c expression and histology changes is shown when including all samples (r=0.23 p=0.0003). A significant correlation (but not after FDR adjustment) was also consistently seen in most subgroups (iloprost arm, current r=0.26, p=0.041 and former smokers r=0.24, p=0.0666; placebo arm, current smokers r=0.23, p=0.0466). These results are illustrated in table 2.

An external file that holds a picture, illustration, etc.
Object name is nihms431175f2.jpg
Distribution of changes in miRNA expression and changes in histology stage

This figure illustrates the distribution of the changes in miRNA expression (represented by the change in delta CT) and the change in histology stage. Each panel illustrates a different miRNA. On the X axis, “0” represents no change in histology, negative and positive numbers represent improvement or worsening in histology, respectively, with the number showing the number of stages that changed. Example: −2 = down-grading of 2 histology units from severe to mild dysplasia or from moderate dysplasia to squamous metaplasia. On the Y axis, the change in delta CT is the difference between the delta CT at follow-up and the delta CT at baseline. By definition, a higher change in delta CT value represents a lower expression and vice versa. The circles represent the different samples. The line shows the linear regression of miRNA expression on histology stage. Only two miRNA had expression changes significantly correlated with histology changes: miR-224 (r= −0.22, p=0.0006) and miR-34c (r=0.23 p=0.0003), up- and down-regulated, respectively, as histology worsens.

Table 2

Correlation between miRNA expression changes and histology changes at FU

miRNABLsamplesPaired samples: FU expression compared to baseline
Spearman
correlation
between
miRNA
expression and
histology grade
(low to high)
Spearman correlation between miRNA expression changes and histology change (up-grading) at FU
compared to baseline
AllCurrent smokersFormer smokers
iloprostPlaceboIloprostPlacebo
miR-34cDown
r=0.36
p<0.0001, FDR
Down
r=0.23
p=0.0003, FDR
Down
r=0.26
p=0.041
Down
r=0.23
p=0.0466
Down
r=0.24
Not significant
Down
r=0.15
Not significant
miR-224Up
r=−0.54
p<0.001, FDR
Up
r=−0.22
p=0.0006, FDR
Up
r=−0.15
Not significant
Up
r=0.11
Not significant
Up
r=−0.37
p=0.0044, FDR
Up
r=−0.25
Not significant

This table reports the correlations between miRNA expression and histology at BL and between miRNA expression changes and histology changes at FU for miR-34c and miR-224 as these two miRNA are the ones whose expression changes are correlated with the histology changes. The table shows up- or down-regulation with regard to histology up-grading at FU compared to baseline. The Spearman correlation coefficient (R) is negative when the CT values are inversely correlated with histology up-grading and thus reflect a positively correlated expression. The Spearman correlation coefficient is positive when the CT values are positively correlated with histology up-grading and thus reflect an inversely correlated expression. Significant p-values are reported. “FDR” indicates that the p-value is significant after adjustment for False Discovery Rate (FDR).

c. Down-regulation of miR-9 expression after bronchial biopsy

MiR-9 expression was not correlated with histology at BL (r=0.01, p=0.8600) or with histology changes between BL and FU (r=0.04, p=0.5074). However, miR-9 was found to be significantly down-regulated in FU as compared with BL biopsies. The down-regulation of miR-9 expression in FU versus BL samples was significant after FDR adjustment when analyzing all samples (Welch t test, p<0.0001) and in most of subgroups (table 1): in the iloprost arm, all patients (p=0.0007) and current smokers (p=0.0022) and in the placebo arm, all patients (p=0.001), and, former smokers (p=0.0071). In current smokers in the placebo arm (p=0.0476) and in the former smokers in the iloprost arm (p=0.0868), the same trend was observed, but did not achieve statistical significance after FDR adjustment. Thus, miR-9 is consistently down-regulated in FU biopsies 6 months later at the same site independently of histology, treatment or smoking status.

4. Comparison of miRNA expression between histological responders and non-responders

Significant differences at BL between responders and non-responders in both the iloprost and placebo arms were shown for the expression of two miRNAs that are correlated with histology. A lower expression of miR-34c at BL was detected in responders versus non-responders both in the iloprost and placebo arms and was very consistent with OR ranging between 1.38 and 1.49. A significantly higher expression of miR-224 at BL was also shown in responders compared to non-responders in several subgroups. Detailed results and OR in different subgroups are illustrated in online supplemental table 4.

In current smokers in the placebo arm, a lower expression of miR-375 at BL predicted response using the “worst histology score criterion” in all samples (p=0.0012 after FDR adjustment with OR=1.93 (95% CI: 1.10–3.39, p=0.0219) as well as in the “paired sample based criterion” (p=0.0027 after FDR adjustment and OR=3.74 (95% CI: 1.27–10.98, p=0.0165).

We tested the predictive value of miRNA expressions using interaction terms between baseline expression and treatment arm in the logistic regression models of response. No individual miRNAs were found to be predictive of response to iloprost. We examined the group of miRNAs to determine if they together or as subsets were associated with response overall and then, if they were differentially associated with response in the Iloprost vs. placebo group. We did this in the former smokers only since that was the group in which Iloprost showed a beneficial effect on histology. We evaluated three models: one in which the definition of response for each biopsy was at the participant level, one in which the definition of response was at the biopsy level, and one in which the response was defined at the biopsy level and excluded the biopsies that were normal at baseline. In the first model one miRNA was associated with response after adjusting for all of the design variables (treatment, pack years of smoking, years since quitting, age and sex) but it was not predictive (p = 0.98 for the 1 df test of the interaction between miRNA and treatment group). In the second model, there were 4 miRNA that were together independently associated with response after adjustment but were not predictive (p = 0.4 for the 4 df test of the interaction between miRNA and treatment group), and in the third model, no miRNA was associated with response after adjustment. Thus, we conclude that the miRNAs applied are not predictive of response to Iloprost, neither as single markers or in combinations.

DISCUSSION

In the current study, we hypothesized that miRNA expression could be used as an intermediate endpoint or as a predictive biomarker in a recent lung cancer chemoprevention trial. We tested this hypothesis in the phase II iloprost lung cancer chemoprevention trial by analyzing the expression of 14 selected miRNA in matched bronchial biopsies before and after treatment with iloprost or placebo. MiRNA analyses were successfully performed from small FFPE endobronchial biopsies, providing an extremely low amount of starting RNA. Despite this technical challenge, the reproducibility of the data was very high with a low variability in RNU48 CT value triplicates.

Seven of the miRNAs analyzed in this study were correlated with histology at BL. The direct or inverse correlation to histology stage of expression of 6/7 of these miRNAs is concordant with our previously reported data (12). This study did not aim nor was it designed to validate the data of the previous study. However, the concordant results support the validity of the present study’s results.

In the current study, biopsies were collected at multiple sites before and after treatment, providing unique paired data. We analyzed the change in miRNA expression in biopsies at the same site before and after treatment. The expression of miR-34c and miR-244 correlated (inversely or directly, respectively) with histology at BL and histology changes. However, the association with histology and histology changes was stronger and more consistently seen for miR-34c. In mice, miR-34a is ubiquitously expressed, while miR-34b/c is mainly expressed in the lung. The members of the miR-34 family are direct targets of p53 and thus are key players in apoptosis and cell-cycle arrest (17). The promoter of miR-34c has been shown to be methylated and miR-34c is thus down-regulated in several cancers including NSCLC (1822). DNA hypermethylation of miR-34b/c also has a prognostic value for stage I NSCLC (21). Down-regulation was shown in rats in after exposure to cigarette smoke (23) or the tobacco associated carcinogen NNK (24). We previously showed that miR-34c is down-regulated in normal human bronchial biopsies from smokers compared to never smokers and is further down-regulated during the successive steps of lung carcinogenesis (12). In the current study, we demonstrated that changes in miR-34c expression between BL and FU correlate with histology changes in both the iloprost and placebo arms. As miR-34c expression is a reproducible and quantitative lab assay reflecting histological changes and is independent of treatment, miR-34c could be a biomarker for the quantitative measure of histological response and a potential intermediate endpoint in lung cancer chemoprevention trials. However the present study does not demonstrate that miR-34c is an intermediate endpoint, or a biomarker that could replace histology or add to histology for evaluation of response. Prospective studies designed to investigate these questions are warranted.

In addition, the results showed that a lower expression of miR-34c and a higher expression of miR-224 were detected in responders in both treatment and placebo arm. As miR-34c and miR-224 expressions are correlated, inversely and directly, respectively, with the stages of histology at baseline, lower miR-34c and higher miR-224 expressions at baseline in responders are likely to be related to higher histology at baseline, which allows response (histology downgrading) while normal histology does not. Therefore these miRNAs cannot be considered as a predictor of response.

MiR-9 was reproducibly down-regulated in FU independently of histology, smoking status and treatment. Thus the down-regulation of miR-9 expression at FU at the same site might thus be related to the biopsy and subsequent repair. Mir-9 is highly evolutionarily conserved (25) and regulates the expression of genes involved in cell motility and invasion. It down-regulates SNAIL-1 and NF-κB and up-regulates E-cadherin (26). MiR-9 also plays a role in cytoskeletal organization (26) and its down-regulation increases cell motility (26). This could perhaps explain its potential role in tissue repair, but this hypothesis requires further study.

To our knowledge, differences in miRNA expression between current and former smokers have not been previously examined. There are previous data in animal models (23, 24, 27) and in human airways analyzing miRNA profiles between never smokers and smokers (12, 28). We found that several miRNAs were up- or down-regulated in current smokers as compared with former smokers. The miRNAs that were up-regulated in current smokers in this study have not been previously described as related to smoking. The two miRNAs, miR-142-3p and miR-34c, were down-regulated in current smokers as compared with former smokers were described earlier. Mir-142-3p and miR-34c were observed to be down-regulated in the normal bronchial mucosa of smokers as compared with never smokers in our previous study (12). In addition, miR-34c has been shown to be down-regulated in rats exposed to cigarette smoke (23).

MiR-375 is up-regulated in airways by current smoking and down-regulated after iloprost treatment in current smokers. The expression of miR-375 has not been shown to be related to smoking in animal studies or in studies using human tissue and comparing never smokers to smokers. In normal lung, miR-375 has been shown to inhibit surfactant secretion by altering cytoskeletal reorganization (29). We showed in our previous study that miR-375 was down-regulated in invasive squamous carcinoma of the lung compared with non-invasive bronchial lesions (12). MiR-375 has been shown to be down-regulated in other squamous tumors (30, 31) as well as other cancers (3236). Its up-regulation in active smokers as compared with former smokers and its down-regulation in current smokers by iloprost treatment are of interest and cannot be explained by any previous functional data on miR-375.

This study failed to identify predictive markers of response to iloprost. The phase II Iloprost Trial was not designed for finding predictive biomarkers to histological response. In addition, other miRNAs, which we did not studied could potentially be predictive of response to iloprost. We could not study a larger panel of miRNAs because the amount of available RNA was limited. Therefore we selected miRNAs based on our previous findings in similar bronchial biopsies. This work must be considered as exploratory, and further investigation is required in subsequent chemoprevention trials. As no predictive biomarkers were found, it will be necessary to carry out additional similar studies with a larger panel of miRNAs.

In conclusion, no biomarkers predictive of response to iloprost were discovered. However, some interesting results were obtained. MiR-375 was down regulated in current smokers receiving iloprost. MiR-9 was down regulated in FU biopsies compared to BL biopsies and may represent a tissue repair response. Several novel miRNAs with expression modulated by current vs. former smoking status were identified. The expression of miR-34c was inversely correlated with histology at BL and with changes in histology between BL and FU, independent of the treatment arm or smoking status. Thus, the change in miR-34c expression between BL and FU biopsies may be a quantitative biomarker of histological response. Additional lung chemoprevention trials are in progress and we plan to validate the role of miR-34c and others in these trials.

Supplementary Material

1

2

3

4

1

Click here to view.(74K, docx)

2

Click here to view.(64K, docx)

3

Click here to view.(96K, docx)

4

Click here to view.(86K, docx)

Acknowledgments

Financial support: This project was supported by a Specialized Program of Research Excellence (SPORE) Career Development Award (CM) and by Lung Cancer Biomarkers and Chemoprevention Consortium (LCBCC) funds (CM and FRH). CM was also supported by an International Association for the Study of Lung Cancer (IASLC) fellowship, a grant from the Louisiana Chapter of the National Lung Cancer Partnership (NLCP) and a grant from the Gary L. and Thelissa Zollinger Early Detection of Lung Cancer Endowment Fund. Additional support was from the Colorado SPORE in Lung Cancer NCI P50 CA58187, NCI RO1 {"type":"entrez-nucleotide","attrs":{"text":"CA164780","term_id":"35082537","term_text":"CA164780"}}CA164780, Texas SPORE in Lung Cancer (P50CA070907), Department of Veterans Affairs Merit Review Program (RLK).

Department of Medicine, Division of Medical Oncology, Colorado School of Public Health and University of Colorado School of Medicine, Aurora, CO, 80045, USA
Department of Biostatistics and Informatics, Colorado School of Public Health and University of Colorado School of Medicine, Aurora, CO, 80045, USA
Department of Pathology, Colorado School of Public Health and University of Colorado School of Medicine, Aurora, CO, 80045, USA
Division of Pulmonary Sciences and Critical Care Medicine, Colorado School of Public Health and University of Colorado School of Medicine, Aurora, CO, 80045, USA
Genome and Sciences Resource, Vanderbilt University Medical Center, Nashville, TN, 37232, USA
Denver Veterans Affairs Medical Center, Denver, CO
Hamon Center of Therapeutic Oncology Research, Departments of Internal Medicine and Pharmacology, University of Texas Southwestern Medical Center, Texas 75390, USA
Corresponding author: Celine Mascaux, MD, PhD, University of Colorado Denver; Division of Medical Oncology, 12801 E. 17th Ave; Mail Stop 8177, Aurora, Colorado USA, 80045. Tel: 303-724-0252; Fax: 303-724-0714, moc.liamg@xuacsammjenilec
co-last authors

Abstract

Lung cancers express lower levels of prostacyclin than normal lung tissues. Prostacyclin prevents lung cancer in a variety of mouse models. A randomized phase II trial comparing oral iloprost (a prostacyclin analogue) to placebo in high-risk subjects demonstrated improvement in bronchial histology in former, but not current, smokers. This placebo-controlled study offered the opportunity for investigation of other potential intermediate endpoint and predictive biomarkers to incorporate into chemoprevention trials.

Matched bronchial biopsies were obtained at baseline (BL) and at 6 months follow-up (FU) from 125 high-risk individuals who completed the trial: 31/29 and 37/28 current/former smokers in the iloprost and placebo arm, respectively. We analyzed the expression of 14 selected miRNAs by qRT-PCR in 496 biopsies.

The expression of seven miRNAs was significantly correlated with histology at BL. The expression of miR-34c was inversely correlated with histology at BL (p<0.0001) and with change in histology at FU (p=0.0003), independent of treatment or smoking status. Several miRNAs were also found to be differentially expressed in current smokers as compared with former smokers. In current smokers, miR-375 was up-regulated at BL (p<0.0001) and down-regulated after treatment with iloprost (p=0.0023). No miRNA at baseline reliably predicted a response to iloprost.

No biomarker predictive of response to iloprost was found. MiR-34c was inversely correlated with BL histology and with histology changes. Mir-34c changes at FU could be used as a quantitative biomarker which parallels histologic response in formalin-fixed bronchial biopsies in future lung cancer chemoprevention studies.

Keywords: miRNAs, chemoprevention, lung cancer, iloprost
Abstract

Footnotes

Conflict of interest: PAB is consultant for Bayer. YEM and RLK have applied for a patent regarding prostacyclin for the chemoprevention of cancer.

Footnotes

References

  • 1. Slaughter DP, Southwick HW, Smejkal WField cancerization in oral stratified squamous epithelium; clinical implications of multicentric origin. Cancer. 1953;6:963–968.[PubMed][Google Scholar]
  • 2. Arber N, Eagle CJ, Spicak J, Racz I, Dite P, Hajer J, et al Celecoxib for the prevention of colorectal adenomatous polyps. N Engl J Med. 2006;355:885–895.[PubMed][Google Scholar]
  • 3. Dube C, Rostom A, Lewin G, Tsertsvadze A, Barrowman N, Code C, et al. The use of aspirin for primary prevention of colorectal cancer: a systematic review prepared for the U.S. Preventive Services Task Force. Ann Intern Med. 2007;146:365–375.[PubMed]
  • 4. Heasley LE, Thaler S, Nicks M, Price B, Skorecki K, Nemenoff RAInduction of cytosolic phospholipase A2 by oncogenic Ras in human non-small cell lung cancer. J Biol Chem. 1997;272:14501–14504.[PubMed][Google Scholar]
  • 5. Stearman RS, Dwyer-Nield L, Zerbe L, Blaine SA, Chan Z, Bunn PA, Jr, et al Analysis of orthologous gene expression between human pulmonary adenocarcinoma and a carcinogen-induced murine model. Am J Pathol. 2005;167:1763–1775.[Google Scholar]
  • 6. Nemenoff R, Meyer AM, Hudish TM, Mozer AB, Snee A, Narumiya S, et al Prostacyclin prevents murine lung cancer independent of the membrane receptor by activation of peroxisomal proliferator--activated receptor gamma. Cancer Prev Res (Phila Pa) 2008;1:349–356.[Google Scholar]
  • 7. Keith RL, Miller YE, Hoshikawa Y, Moore MD, Gesell TL, Gao B, et al Manipulation of pulmonary prostacyclin synthase expression prevents murine lung cancer. Cancer Res. 2002;62:734–740.[PubMed][Google Scholar]
  • 8. Keith RL, Miller YE, Hudish TM, Girod CE, Sotto-Santiago S, Franklin WA, et al Pulmonary prostacyclin synthase overexpression chemoprevents tobacco smoke lung carcinogenesis in mice. Cancer Res. 2004;64:5897–5904.[PubMed][Google Scholar]
  • 9. Keith RL, Blatchford PJ, Kittelson J, Minna JD, Kelly K, Massion PP, et al Oral iloprost improves endobronchial dysplasia in former smokers. Cancer Prev Res (Phila) 2011;4:793–802.[Google Scholar]
  • 10. Szabo EAltered histology provides a positive clinical signal in the bronchial epithelium. Cancer Prev Res (Phila) 2011;4:775–778.[Google Scholar]
  • 11. Mitchell PS, Parkin RK, Kroh EM, Fritz BR, Wyman SK, Pogosova-Agadjanyan EL, et al Circulating microRNAs as stable blood-based markers for cancer detection. Proc Natl Acad Sci U S A. 2008;105:10513–10518.[Google Scholar]
  • 12. Mascaux C, Laes JF, Anthoine G, Haller A, Ninane V, Burny A, et al Evolution of microRNA expression during human bronchial squamous carcinogenesis. Eur Respir J. 2009;33:352–359.[PubMed][Google Scholar]
  • 13. Travis WD, Brambilla E, Muller-Hermerlink HK, Harris CCPathology and Genetics. Tumours of the Lung, Pleura, Thymus and Heart. 2004[PubMed][Google Scholar]
  • 14. Moschos SA, Williams AE, Perry MM, Birrell MA, Belvisi MG, Lindsay MAExpression profiling in vivo demonstrates rapid changes in lung microRNA levels following lipopolysaccharide-induced inflammation but not in the anti-inflammatory action of glucocorticoids. BMC Genomics. 2007;8:240.[Google Scholar]
  • 15. Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, Kubista M, et al The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin Chem. 2009;55:611–622.[PubMed][Google Scholar]
  • 16. Moser BK, Stevens GRHomogeneity of Variance in the Two-Sample Means Test. The American Statisticians. 1992;46:19–21.[PubMed][Google Scholar]
  • 17. Hermeking HThe miR-34 family in cancer and apoptosis. Cell Death Differ. 2010;17:193–199.[PubMed][Google Scholar]
  • 18. Kozaki K, Imoto I, Mogi S, Omura K, Inazawa JExploration of tumor-suppressive microRNAs silenced by DNA hypermethylation in oral cancer. Cancer Res. 2008;68:2094–2105.[PubMed][Google Scholar]
  • 19. Lujambio A, Calin GA, Villanueva A, Ropero S, Sanchez-Cespedes M, Blanco D, et al A microRNA DNA methylation signature for human cancer metastasis. Proc Natl Acad Sci U S A. 2008;105:13556–13561.[Google Scholar]
  • 20. Toyota M, Suzuki H, Sasaki Y, Maruyama R, Imai K, Shinomura Y, et al Epigenetic silencing of microRNA-34b/c and B-cell translocation gene 4 is associated with CpG island methylation in colorectal cancer. Cancer Res. 2008;68:4123–4132.[PubMed][Google Scholar]
  • 21. Wang Z, Chen Z, Gao Y, Li N, Li B, Tan F, et al DNA hypermethylation of microRNA-34b/c has prognostic value for stage non-small cell lung cancer. Cancer Biol Ther. 2011;11:490–496.[PubMed][Google Scholar]
  • 22. Tanaka N, Toyooka S, Soh J, Kubo T, Yamamoto H, Maki Y, et al Frequent methylation and oncogenic role of microRNA-34b/c in small-cell lung cancer. Lung Cancer. 2011[PubMed][Google Scholar]
  • 23. Izzotti A, Calin GA, Arrigo P, Steele VE, Croce CM, De Flora SDownregulation of microRNA expression in the lungs of rats exposed to cigarette smoke. FASEB J. 2009;23:806–812.[Google Scholar]
  • 24. Kalscheuer S, Zhang X, Zeng Y, Upadhyaya PDifferential expression of microRNAs in early-stage neoplastic transformation in the lungs of F344 rats chronically treated with the tobacco carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone. Carcinogenesis. 2008;29:2394–2399.[Google Scholar]
  • 25. Yuva-Aydemir Y, Simkin A, Gascon E, Gao FBMicroRNA-9: functional evolution of a conserved small regulatory RNA. RNA Biol. 2011;8:557–564.[Google Scholar]
  • 26. Liu S, Kumar SM, Lu H, Liu A, Yang R, Pushparajan A, et al MicroRNA-9 up-regulates E-cadherin through inhibition of NF-kappaB1-Snail1 pathway in melanoma. J Pathol. 2012;226:61–72.[Google Scholar]
  • 27. Izzotti A, Calin GA, Steele VE, Croce CM, De Flora SRelationships of microRNA expression in mouse lung with age and exposure to cigarette smoke and light. FASEB J. 2009;23:3243–3250.[Google Scholar]
  • 28. Schembri F, Sridhar S, Perdomo C, Gustafson AM, Zhang X, Ergun A, et al MicroRNAs as modulators of smoking-induced gene expression changes in human airway epithelium. Proc Natl Acad Sci U S A. 2009;106:2319–2324.[Google Scholar]
  • 29. Zhang H, Mishra A, Chintagari NR, Gou D, Liu LMicro-RNA-375 inhibits lung surfactant secretion by altering cytoskeleton reorganization. IUBMB Life. 2010;62:78–83.[Google Scholar]
  • 30. Hui AB, Bruce JP, Alajez NM, Shi W, Yue S, Perez-Ordonez B, et al Significance of dysregulated metadherin and microRNA-375 in head and neck cancer. Clin Cancer Res. 2011;17:7539–7550.[PubMed][Google Scholar]
  • 31. Wiklund ED, Gao S, Hulf T, Sibbritt T, Nair S, Costea DE, et al MicroRNA alterations and associated aberrant DNA methylation patterns across multiple sample types in oral squamous cell carcinoma. PLoS ONE. 2011;6:e27840.[Google Scholar]
  • 32. Mazar J, DeBlasio D, Govindarajan SS, Zhang S, Perera RJEpigenetic regulation of microRNA-375 and its role in melanoma development in humans. FEBS Lett. 2011;585:2467–2476.[PubMed][Google Scholar]
  • 33. Bhatti I, Lee A, James V, Hall RI, Lund JN, Tufarelli C, et al Knockdown of microRNA-21 inhibits proliferation and increases cell death by targeting programmed cell death 4 (PDCD4) in pancreatic ductal adenocarcinoma. J Gastrointest Surg. 2011;15:199–208.[PubMed][Google Scholar]
  • 34. Ding L, Xu Y, Zhang W, Deng Y, Si M, Du Y, et al MiR-375 frequently downregulated in gastric cancer inhibits cell proliferation by targeting JAK2. Cell Res. 2010;20:784–793.[PubMed][Google Scholar]
  • 35. Mathe EA, Nguyen GH, Bowman ED, Zhao Y, Budhu A, Schetter AJ, et al MicroRNA expression in squamous cell carcinoma and adenocarcinoma of the esophagus: associations with survival. Clin Cancer Res. 2009;15:6192–6200.[Google Scholar]
  • 36. Kong KL, Kwong DL, Chan TH, Law SY, Chen L, Li Y, et al MicroRNA-375 inhibits tumour growth and metastasis in oesophageal squamous cell carcinoma through repressing insulin-like growth factor 1 receptor. Gut. 2012;61:33–42.[PubMed][Google Scholar]
Collaboration tool especially designed for Life Science professionals.Drag-and-drop any entity to your messages.