A Novel Saliva and Serum miRNA Panel as a Potential Useful Index for Oral Cancer and the Association of miR-21 with Smoking History: a Pilot Study

Head and neck cancer not related to human papillomavirus (HPV) is an aggressive malignancy and is more successfully treated if diagnosed early. Early detection of head and neck squamous cell carcinoma (HNSCC) would be improved with the identification of specific biomarkers that have been validated through large-scale studies (1). Most known biomarkers exhibit limitations in the early diagnosis and prognosis, and tissue specificity, limiting their role in screening strategies (2). Small noncoding miRNA molecules, which are often altered during head and neck carcinogenesis (3), can be defined as tissue-specific biomarkers and can have diagnostic and predictive value (4–9). Notably, recent preclinical findings from our team have documented deregulations of specific miRNAs, such as miR-21, associated with early oncogenic events in head and neck carcinogenesis induced by chronic exposure to known risk factors, such as tobacco smoke components (6, 10–14).

Although miRNAs can be detected in tissue biopsy, which is an invasive method, miRNAs can also be detected in body fluids, such as serum, plasma, or saliva, using noninvasive collection methods (15–18). In addition, extracellular miRNAs have been proven to exist in a stable cell-free form (18). Therefore, saliva or circulating (serum) miRNAs may be valuable biomarkers in cancer. Although previous studies have presented several miRNA markers detected in saliva or serum with a detection value in HNSCC, the data remain unclear due to different collection methods or miRNA analysis focus, as well as diverse types of head and neck cancer malignancies included in the studies (19, 20). To the best of our knowledge, so far, only a few studies have investigated miRNA markers in paired saliva and serum samples from the same patients with oral cancer of various risk factors, such as tobacco smoking, alcohol, and GERD, compared with healthy controls (21, 22). Such noninvasive and rapid approaches could provide promising diagnostic and prognostic biomarkers with the ability to screen and monitor high-risk patients (23).

We hypothesized that there is a specific panel of miRNAs that are altered in saliva and serum of a common HPV-negative HNSCCs, such as oral squamous cell carcinoma (OSCC), compared with nonmalignant healthy controls. To investigate our hypothesis, we performed a pilot case–control study examining the expression and the association of 20 head and neck cancer–related extracellular miRNAs, as suggested by our previous studies and proposed by others (15, 16, 19–22, 24, 25). We collected preoperative paired saliva and serum samples from patients with oral cancer and healthy individuals and performed a comparative study to assess the potential association of specific miRNAs with OSCC. Our pilot study demonstrates a novel saliva and serum set of OSCC-related miRNAs that could be included in a future validation study and suggests specific noninvasive miRNAs for OSCC risk assessment.

We analyzed data from 44 pretreatment subjects consisting of 23 patients with oral OSCC (study group) and 21 healthy volunteers without malignancy (Table 1; Supplementary Tables S1 and S2). All participants provided oral and written consent for participation in this study and donation of their biomedical samples to the Yale Head and Neck Biorepository according to the Yale IRB approved protocol (HIC#1206010419). The study was conducted in accordance to recognized ethical guidelines such as the Belmont Report, U.S. Common Rule, etc. All participants agreed to the publication of the information presented in the current study and all identifying information was removed.

Patients’ characteristics and risk factors in OSCC and healthy control groups are shown in Table 1, as well as in Supplementary Tables S1 and S2. As shown in Table 1, our OSCC study and healthy control groups included almost equal numbers of male and female subjects, as well as of smokers (current or former) and never smokers. The majority of smokers with OSCC were male (75%; 9/12), while less than half of nontobacco users with OSCC were male (36%; 4/11). These differences were not found to be statistically significant (P = 0.06; by χ² test). The nonsmoker healthy control group included a 1:1 male-to-female ratio (P = 0.52; by χ² test).

There were 52% (12/23) of patients with OSCC and 38% (8/21) of healthy control individuals who consumed alcohol, and the majority of them were also smokers. Finally, 30% (7/23) of patients with OSCC and 38% (8/21) of healthy control had a history of gastroesophageal disease (GERD). There were 28% (2/7) of patients with OSCC having GERD who consumed alcohol or had a smoking history, while 50% (5/8) of healthy control individuals with a history of GERD were smokers (Supplementary Tables S1 and S2).

Histologic characteristics of HPV-negative OSCC tumors and controls are also shown in Table 1, as well as in Supplementary Tables S1 and S2. Almost 40% (9/23) of OSCC tumors were at an early stage without regional metastases, including an almost equal number of smokers and nonsmokers.

Paired saliva and serum samples from the same patient were collected preoperatively and analyzed by our team, as follows:

We followed a previously described collection protocol by Agrawal and colleagues (24). Specifically, participants fasted for at least 1 hour and did not use mouthwash solution or drink coffee before collection. Participants swished 15–20 mL of 0.9% saline solution in their mouth for 15–30 seconds before expectorating into a collection container. The specimen container was placed immediately on ice and transferred from the preoperative area to the laboratory. Five to 10 mL of saliva solution was centrifuged in a refrigerated centrifuge at 3,000 × g for 15 minutes at 4°C. The pellet was kept at −80°C until RNA isolation.

We followed a previously established serum collection method (17). We collected 2 mL of whole blood from each participant. Blood was drawn into blood collection tubes containing no anticoagulant or preservative and kept upright at room temperature for 45–60 minutes. Clotted blood was centrifuged at 2,000 × g at room temperature for 10 minutes. Aliquots of 200–300 μL of serum were transferred in sterile tubes and kept at −80°C until RNA isolation.

We isolated miRNAs from saliva and serum using miRNeasy Kits and miRNAeasy Serum/Plasma Advanced Kits, respectively, following the manufacturer's instructions (Qiagen Inc.). Frozen saliva (200–300 μL) and serum (200 μL) samples were thawed on ice, mixed well with 700 μL and 1 mL Qiazol solution, respectively, and incubated at room temperature for 5 minutes. Then 140 μL and 200 μL of chloroform were added, respectively, to the saliva and serum mixture. The mixture was incubated for 3 minutes at room temperature and then centrifuged for 15 minutes at high speed (>8,000 × g) at 4°C. The supernatant from saliva and serum samples was transferred to a clean tube, mixed with 1.5 volumes of 100% ice-cold EtOH, and transferred to RNAeasy mini columns and RNeasy UCP MinElute columns, respectively. After centrifugation and washing with buffers, RNA from saliva and serum was eluted from the column, using respectively 30 μL and 20 μL of RNase-free water. The quality and quantity of RNA quality were determined by absorption ratios at 260/280 nm (≥2.0) and concentration ratios by absorption at 260 nm, using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific). Saliva and serum RNA were kept at −80°C until miRNA analysis.

We quantified saliva and serum miRNAs, using miRCURY LNA RT Kit, miRCURY LNA miRNA Custom Probe PCR Assays (locked nucleic acid, LNA, dual-labeled miRNA-specific probes; Supplementary Table S3), and miRCURY LNA Probe PCR Kit, according to manufacturer's instructions (Qiagen Inc.). We performed reverse transcription using equal quantities (30 ng) of RNA for each OSCC or healthy control saliva and serum sample. We analyzed a panel of 20 HNSCC-related miRNAs (2, 3, 5–8, 12–16). We used RNU6 as a reference control. Relative expression ratios for each specific miRNA (target miRNA/RNU6) were estimated by qPCR analysis (Bio-Rad Real-Time Thermal Cycler CFX96; Bio-Rad) software for all samples (6, 8, 9, 11–14) to determine miRNA expression ranges in OSCC and healthy control groups. We also determined the miRNA-relative expression ratios (OSCC/healthy control), in saliva, serum, and their combination (saliva/serum).

Using t test (Welch t test; P < 0.05) or one-way ANOVA [GraphPad Prism 7 software; (RRID:SCR_002798)] we compared miRNAs (miRNA/RNU6 levels) in saliva, serum, or their combined (saliva/serum) profiles between OSCC and healthy control and examine their differential expression between smokers and nonsmokers, as well as between early- and late-stage tumors.

The data that support the findings of this study are available in the Supplementary Data and upon request to the corresponding author.

Eighteen of 20 (90%) analyzed miRNAs were detected in the saliva and/or serum of OSCC and healthy control subjects (Fig. 1A). More than 50% of these miRNAs were found to be deregulated in saliva (66.6%; 12/18), serum (78%; 14/18), or saliva–serum profile ratios [saliva/serum ratios (89%; 16/18) of patients with OSCC compared with healthy controls).

As shown in Fig. 1B, a panel of 10 miRNAs demonstrated quantitative alterations in saliva or serum, while, only 4 of them, miR-136, miR-29B, miR-3928, and miR-21, showed statistically significant differential expression in the saliva of patients with OSCC versus healthy controls (P < 0.05, by t test; Supplementary Tables S4 and S5).

Also, as shown in Fig. 1C, “oncomir” miR-21 was detected at significantly higher levels in saliva of patients with OSCC having smoking history compared to patients with OSCC who never smoked or compared with healthy control smokers (P < 0.05; by t test). Serum miR-21 was also detected at significantly higher levels in OSCC smokers compared with nonsmokers (P < 0.05, by t test; Fig. 1C; Supplementary Tables S4A and S5B).

No association was found between smoking pack-year history [mean (±SD), 25.8 (±14.7); range, 0.5–45] and saliva miR-21 levels of patients with OSCC (r = 0.02; P = 0.49, by Pearson). However, a significant linear correlation was identified between smoking pack-year [mean (±SD), 8.9 (±5.3); range, 2.5–17.5] and saliva miR-21 levels of healthy controls (r = 0.6; P = 0.039, by Pearson; Fig. 1D; Supplementary Table S5).

Higher miR-21 levels were found in saliva of current versus former healthy control smokers [3.2 fold-change; mean (±SD): 0.046 (±0.036) vs. 0.015 (±0.023)]. However, former smokers with OSCC showed significantly higher saliva miR-21 levels than current smokers with OSCC [10-fold-change; mean (±SD): 0.22 (±0.2) vs. 0.02 (±0.048); P = 0.0139], or versus healthy control smokers (15-fold-change; P = 0.0075, by t test).

Pearson analysis revealed an inverse correlation between saliva and serum miRNAs, with a statistical significance for miR-29B (r = −0.825; P = 0.043).

To explore the potential association of miR-136, miR-3928, miR-29B, and miR-21 in saliva and/or serum of early-stage OSCC [ESC; early tumor stage (T1 or T2), negative lymph nodes (N0) and no regional metastases (M0); Table 1; Supplementary Table S1] with healthy individuals, we performed a comparative analysis, using t test or one-way ANOVA.

As shown in Table 2A, three miRNAs, miR-136, miR-3928, and miR-21, were significantly altered in saliva of 9 patients with ESC compared with controls (P < 0.05; by t test; Supplementary Table S6A).

Similar to saliva, serum miR-136 and miR-21 showed statistically significant higher levels in patients with ESC versus healthy controls (P < 0.05; Table 2A; Supplementary Table S6B).

“Oncomir” miR-21 was detected at significantly higher levels in the saliva or serum of patients with ESC and smoking history compared with never smokers or compared with healthy control with a smoking history (P < 0.05; by t test; Table 2A).

Regionally metastatic OSCC also showed significantly higher salivary miR-21, miR-3928, and miR-136 levels versus healthy controls (P < 0.05). However, ESC showed significantly higher serum miR-21 levels relative to late-stage OSCC (P = 0.017), particularly in smokers (P = 0.0027; Supplementary Table S6).

Table 2B summarizes our findings, illustrating the potential use of a set of 4 saliva and serum miRNAs associated with OSCC and smoking history for a validation study.

Using a noninvasive approach, this pilot study suggests the potential association of a novel panel of 4 miRNAs in saliva, miR-21, miR-136 and miR-3928, and miR-29B, with oral cancer, proposing cut-off levels for further analysis in a validation study. Our novel data also demonstrate the potential association of miR-21 in paired saliva and serum for oral cancer detection in patients with a smoking history, even at an early stage. Notably, our preliminary findings demonstrated a linear correlation between pack-year of tobacco smoking and salivary miR-21 levels in healthy individuals, supporting the effect of cumulative smoking history on miR-21 upregulation. On the basis of our data, more than 15 pack-year smoking history in healthy individuals reached the cut-off of miR-21 levels in saliva (miR-21/RNU6 >0.09) as found in patients with diagnosed OSCC. Although further investigation is required to reveal an association of noninvasive miRNAs with smoking status or other risk factors such as a history of alcohol consumption, miR-21 in saliva of current or former smokers is suggested as potentially associated with oral cancer.

Data from our study suggest that miRNA molecules, such as miR-3928, miR-29B, miR-136, and miR-21, detected in saliva and/or serum of patients with OSCC are selectively secreted compared with healthy individuals. Although a small cohort of nine early-stage OSCCs was included in this pilot study, based on our data, the identification of high serum miR-136 and miR-21 levels could indicate the development of cancer even at an early stage. Our pilot study also presents novel data of the combined saliva and serum miR-21 and suggests its potential application for oral cancer, especially in patients with a smoking history. Specifically, the combination of elevated miR-21/RNU6 ratios in saliva (>0.09) and serum (>0.7;Table 2B) is suggested to be further validated as a potential index of cancer detection in individuals with oral premalignant or early malignant lesions, follow-up of individuals at risk, such as individuals with a smoking history. On the basis of our findings with a small cohort, demographic diversity including biological sex, race, and ethnicity does not affect the expression levels/range of the analyzed noninvasive miRNAs.

We suggest further validation of this set of miR-3928, miR-29B, miR-136, and miR-21 in an independent group of saliva and serum samples from patients with OSCC, including more early-stage OSCC, alongside oral premalignant lesions and controls. To increase the statistical power of the recommended screening method, we also suggest further investigation and enrollment of a large group of subjects, including current or former smokers alongside nonsmokers. Understanding the tremendous clinical significance of screening miRNA biomarkers, it remains our goal to perform a broader analysis of small non-coding RNA sequencing that may reveal novel noninvasive molecular biomarkers of oral cancer.

Our current pilot study presents data supporting the consideration of salivary miRNAs, miR-136, miR-3928, miR-21, and miR-29B, as well as serum miR-21, as a candidate set of small RNAs associated with oral cancer, which could be for further validated in premalignant and early malignant lesions. Our pilot study also supports examining miR-21 in saliva or serum in a larger group of individuals with a smoking history, as a potential biomarker for risk stratification.

No disclosures were reported.

D.P. Vageli: Conceptualization, resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. P.G. Doukas: Conceptualization, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. R. Shah: Formal analysis, visualization, writing–review and editing. T. Boyi: Formal analysis, visualization, writing–review and editing. C. Liu: Formal analysis, visualization, writing–review and editing. B.L. Judson: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, project administration, writing–review and editing.

This study was partially supported by the Virginia Alden Wright Fund (to B.L. Judson).