Serum PIWI-Interacting Rnas Pir-020619 and Pir-020450 Are Promising Novel Biomarkers for Early Detection of Colorectal Cancer

Author Manuscript Published OnlineFirst on February 17, 2020; DOI: 10.1158/1055-9965.EPI-19-1148 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Serum PIWI-interacting RNAs piR-020619 and piR-020450 are promising novel biomarkers for early detection of colorectal cancer

Zhenfei Wang1#, Hao Yang2#, Daguang Ma3, Yongping Mu1, Xiaohui Tan4, Qin Hao5, Li Feng6, Junqing Liang1, Wen Xin7, Yongxia Chen1, Yingcai Wu1, Yongfeng Jia1，8, and Haiping Zhao6

1The Laboratory for Tumor Molecular Diagnosis, Affiliated People’s Hospital of Inner Mongolia Medical University, Huhhot 010020, China 2Department of Radiotherapy, Affiliated People’s Hospital of Inner Mongolia Medical University, Huhhot 010020, China 3Department of Thoracic Surgery, Affiliated People’s Hospital of Inner Mongolia Medical University, Huhhot 010020, China 4College of Traditional Chinese Medicine, Inner Mongolia Medical University, Huhhot 010010, China 5Department of Gastrointestinal Surgery, Affiliated Hospital of Inner Mongolia Medical University, Huhhot 010050, China 6Department of Abdominal Tumor Surgery, Affiliated People’s Hospital of Inner Mongolia Medical University, Huhhot 010020, China 7 TransGen Biotech Co.Ltd., Beijing 010010, China

8 Basic Medicine College, Inner Mongolia Medical University, Huhhot 010010, China # These authors equally contributed Running title: CRC-specific non-invasive early detection biomarkers. Keywords: Colorectal cancer; colorectal adenoma; early detection; biomarker; PIWI-interacting RNA, Abbreviations list: Colorectal cancer (CRC) PIWI-interacting RNAs (piRNAs) Colorectal adenoma (CRA) Carcinoembryonic antigen (CEA) Carbohydrate antigen 19-9 (CA19-9) 1

Downloaded from cebp.aacrjournals.org on September 25, 2021. © 2020 American Association for Cancer Research. Author Manuscript Published OnlineFirst on February 17, 2020; DOI: 10.1158/1055-9965.EPI-19-1148 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Long non-coding RNAs (lncRNAs) Reverse transcription quantitative PCR (RT-qPCR) Receiver operating characteristic (ROC) Area under the ROC curve (AUC) Correspondence to: Prof. Yongfeng Jia and Haiping Zhao, Affiliated People’s Hospital of Inner Mongolia Medical University, 42 Zhaowuda Road, Huhhot 010021, China. E-mail address: [email protected] (Y.F. Jia), [email protected] (H.P. Zhao). Disclosure of Potential Conflicts of Interest The authors declare no potential conflicts of interest.

Abstract Background: Early diagnosis can significantly reduce colorectal cancer (CRC) deaths. We sought to identify serum PIWI-interacting RNAs (piRNAs) that could serve as sensitive and specific non-invasive biomarkers for early CRC detection. Methods: We screened the piRNA expression profile in sera from seven CRC patients and seven normal controls using small RNA sequencing. Differentially expressed piRNAs were measured in a training cohort of 140 CRC patients and 140 normal controls using reverse transcription quantitative PCR. The identified piRNAs were evaluated in two independent validation cohorts of 180 CRC patients and 180 normal controls. Finally, the diagnostic value of the identified piRNAs for colorectal adenoma (CRA) was assessed, and their expression was measured in 50 lung cancer, 50 breast cancer and 50 gastric cancer patients. Results: The piRNAs piR-020619 and piR-020450 were consistently elevated in sera of CRC patients as compared with controls. A predicative panel based on the two piRNAs was established that displayed high diagnostic accuracy for CRC detection. The two-piRNA panel could detect small-size and early-stage CRC with an area under the receiver operating characteristic curve of 0.863 and 0.839, respectively. Combined use of the two piRNAs could effectively distinguish CRA from controls. Aberrant elevation of the two piRNAs was not observed in sera of lung, breast and gastric cancer patients. Conclusions: Serum piR-020619 and piR-020450 show a strong potential as CRC-specific early detection biomarkers. Impact: The field of circulating piRNAs could allow for novel tumor biomarker development.

Introduction Colorectal cancer (CRC) is currently the third most common malignancy worldwide with over 1.4 million new cases and over 680,000 deaths annually (1). The prognosis of CRC largely depends on the TNM stage at the time of diagnosis, with the 5-year survival rate decreasing from about 90% in early-stage disease (stages 0–II) to 12–13% in metastatic disease (stage IV) (2). Unfortunately, CRC is usually asymptomatic at the early stage. Consequently, most patients are diagnosed at the advanced stage, when tumor invasion and dissemination have already occurred and operative curative resection cannot be performed. Therefore, early detection is crucial for improving the outcome of CRC patients. Colonoscopy is the gold standard for diagnosing CRC, but its invasiveness, high cost and complicated bowel preparation makes it inappropriate for large-scale screening (3). Fecal occult blood tests and stool DNA tests are both non-invasive assays, but the former has low sensitivity and specificity and the latter is expensive (3). Blood tumor marker detection has been advocated for cancer screening since it is simple, economical and non-invasive. However, commonly used CRC biomarkers, such as carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9), are inadequate for early diagnosis because of their poor sensitivities (3). Recent studies have shown abnormal non-coding RNA profiles in the serum/plasma of CRC patients compared with healthy people (4, 5). Some researchers have attempted to develop early diagnostic biomarkers for CRC from serum/plasma long non-coding RNAs (lncRNAs) and microRNAs. However, serum/plasma lncRNAs are prone to degradation (6), substantially influencing measurement accuracy. Serum/plasma microRNAs are more stable than serum/plasma lncRNAs, but the use of microRNAs identified as tumor biomarkers has some limitations. Firstly, their performance for early detection of CRC is still unsatisfactory (7). Secondly, often times 4–5 serum/plasma microRNAs are used in combination to perform a diagnosis (8, 9), which can be financially burdensome to patients and requires extensive work from clinical laboratories. Lastly, most of the CRC serum/plasma microRNA biomarkers are discovered based on a small study population from a single medical center, and their clinically diagnostic value requires further 4

verification (9). Thus, it is urgently necessary to develop stable, sensitive, specific and cost-effective circulating biomarker assays for early detection and screening of CRC in asymptomatic populations. In addition to lncRNAs and microRNAs, PIWI-interacting RNAs (piRNAs), with a length of 26-30 nucleotides, are an important class of non-coding RNAs with increasingly demonstrated involvement in human malignancies (10). More than 30,000 piRNAs have been identified in the human genome, a number far exceeding that of microRNAs (~2,000) (10, 11). Dysregulated piRNA expression has been detected in multiple tumor tissue types, contributing to the proliferation, migration, invasion, stemness-maintenance and drug-resistance of tumor cells (12-14). Tumor cells can release piRNAs into the extracellular space in the form of free, protein-binding or exosome-enclosed molecules, which can travel to serum, plasma, saliva and other body fluids (11, 15, 16). The piRNAs in body fluids can be more resistant to oxidation and degradation than microRNAs because of the distinctive 2’-O-methyl modification at their 3’-ends (16, 17). It has been demonstrated that piR-57125 remains extremely stable in serum and plasma samples regardless of repeated freeze-thaw cycles or long-term incubation at room temperature (18). In light of these characteristics, circulating piRNAs should be considered as a novel potential source of non-invasive biomarkers for early cancer detection. In this study, for the first time, we performed a large scale, bi-center investigation to identify serum piRNA markers that can effectively distinguish CRC patients, especially early-stage (TNM stages I-II) CRC patients, from normal controls. Moreover, the expression of piRNA markers was measured in serum samples of patients with colorectal adenoma (CRA), and lung, breast and gastric cancers to further evaluate their diagnostic potential.

Material and Methods Study design The study consisted of four phases. The first phase was a discovery phase in which 14 serum samples from 7 CRC 5

patients with TNM I-II stages and 7 normal controls were collected and subjected to small RNA sequencing to explore differentially expressed piRNAs. The piRNAs with log (fold change) > 3 or < -3, p value < 0.01 and Q value < 0.001 were identified as differentially expressed piRNAs between patients and controls. The second phase was a training phase in which the differentially expressed piRNAs identified in phase 1 was further tested in an independent cohort of 140 CRC patients and 140 normal controls using reverse transcription quantitative PCR (RT-qPCR). The piRNAs with cycle threshold (Ct) values ≤ 30 in over 95% samples, fold change > 2 and p value < 0.01 were selected as potential markers. Then, a predictive piRNA panel was constructed from the potential markers based on the logistic regression model for differentiation of patients and controls. The third phase was a validation phase in which the diagnostic performance of the piRNA panel was evaluated in two independent cohorts containing 180 CRC patients and 180 normal controls from two different medical centers. In this phase, the expression of the piRNAs in the panel was also measured in 12 paired serum samples from CRC patients before and one month after surgery. The fourth phase was an external validation phase in which the expression of piRNA markers in the serum samples of 40 CRA patients was examined, and their performance to distinguish CRA from control individuals was assessed. Additionally, the expression of piRNA markers in the serum samples of 50 lung cancer, 50 breast cancer and 50 gastric cancer patients was assayed. Study population The CRC patients and control individuals in the first and second phases were from the Affiliated Hospital of Inner Mongolia Medical University. In the third phase, the 100 CRC patients and 100 control individuals in validation cohort 1 were from the Affiliated Hospital of Inner Mongolia Medical University, and the 80 CRC patients and 80 control individuals in validation cohort 2 were from the Affiliated People’s Hospital of Inner Mongolia Medical University. The 12 CRC patients donating paired serum samples before and after surgery were from the Affiliated People’s Hospital of Inner Mongolia Medical University. 6

In the fourth phase, the CRA, lung cancer and breast cancer patients were from the Affiliated People’s Hospital of Inner Mongolia Medical University, and the gastric cancer patients were from the Affiliated Hospital of Inner Mongolia Medical University. All the lesions were diagnosed by histopathology analysis and the cancers were staged according to the TNM system of the International Union against Cancer. The involved patients had not received any anticancer treatments such as surgery, chemotherapy, radiotherapy or cytotherapy before serum collection. Serum CEA and CA19-9 levels were measured in the Medical Laboratories of the two hospitals, with the cut-off values being 5 ng/mL for CEA and 27 U/mL for CA19-9. All the control individuals were recruited from the physical examination departments of the two hospitals, and were confirmed to have no indication for cancer in the physical examinations, no history of any malignancy or no diagnosis of a benign neoplasm in the colorectum. Individuals with pulmonary and gastrointestinal chronic inflammation, as well as abnormal heart rate, blood pressure and body temperature were excluded from the control group. The distribution of age and gender in the control group was similar to that in CRC group. All the patients and controls were of the same ethnicity. Demographic and clinical features of the patients and controls have been described in

Supplementary Table S1 and Supplementary Table S2. Written informed consent was obtained from all participants. This study was conducted in accordance with the Declaration of Helsinki and approved by the Medical Ethics Committee of Inner Mongolia Medical University. Serum preparation and RNA extraction Five mL fasting venous blood from each individual was drawn before breakfast and sampled in a coagulating tube. Within 1 hour, the tube was centrifuged at 4,000 rpm for 10 min at 4°C to obtain serum. Then, 1 mL aliquots of the serum were transferred to 1.5 mL RNase-free tubes and centrifuged at 12,000 rpm for 10 min at 4°C to remove cell debris. The serum samples were then transferred to RNase-free tubes and stored at −80°C until use. Total RNA was extracted from serum samples using the miRNeasy Serum/Plasma Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. For small 7

RNA sequencing, the quantity and integrity of RNA were assessed with the Qubit 2.0 (Life Technologies, Waltham, MA, USA) and Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). For RT-qPCR, the quantity and purity of RNA was assessed with the SpectraMax QuickDrop spectrophotometer (Molecular Devices, San Jose, CA, USA). Small RNA sequencing Fourteen libraries were prepared from the serum RNA of seven patients and seven controls according to the procedure described previously (19), and sequencing was conducted on the BGISEQ-500 platform (BGI-Shenzhen, Guangdong, China). The raw sequences were filtered to remove the tags with low quality, with 5’ contamination, with poly A, without 3' primer or shorter than 18 nucleotides. The obtained clean reads were mapped against piRNABank (http://pirnabank.ibab.ac.in/request.html) with Bowtie 2 allowing up to one mismatch per read. Differentially expressed piRNAs were screened using the DEGseq R package. RT-qPCR Complementary DNA was reverse-transcribed from 100 ng total RNA with the miRcute Plus miRNA First-Strand cDNA Kit (Tiangen, Beijing, China). Real-time PCR was performed on an ABI PRISM 7500 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) using the miRcute Plus miRNA qPCR Kit (Tiangen, Beijing, China). The PCR reactions were incubated at 95°C for 15 min, followed by 40 cycles of 94°C for 20 s and 60°C for 34 s, and then ramped from 60°C to 95°C to obtain melting curves. After the cycling, melting curve analyses and electrophoreses on agarose gels were conducted to evaluate the specificity of the PCR products. The Ct values of three candidate internal controls were analyzed with NormFinder to identify an internal control. Relative expression levels of piRNAs were calculated by the 2−ΔCt method, normalizing to the internal control. The piRNA-specific primers were designed and synthesized by Sangon Biotech Shanghai Co., Ltd. (Shanghai, China). Statistical analysis

Expression levels of the analyzed serum piRNAs in patients and normal controls were compared using the non-parametric Mann-Whitney test. The diagnostic 8

performance of the piRNAs for discriminating patients from controls was evaluated by receiver operating characteristic (ROC) curve analysis. The difference of the area under the ROC curve (AUC) was assessed with Medcalc 18.2.1. The comparison of gender distribution, sensitivity and specificity was performed using Pearson’s chi-squared test. The difference in age between CRC patients and controls was analyzed by unpaired unequal variance t-test. Paired samples before and after the surgery were analyzed using two-tailed Wilcoxon test. Logistic regression analysis was performed with Statistical Package for the Social Sciences version 18.0. All tests of significance were two-tailed, with a P value lower than 0.05 indicating statistical significance.

Results Discovery of differentially expressed piRNAs During the discovery phase, 14 small RNA libraries established from the serum samples of 7 CRC patients and 7 controls were subjected to small RNA sequencing. The sequenced samples contained on average 29,480,448 ± 2,283,578 raw reads and 24,208,284±1,517,389 clean reads. More than 98.9% of the clean reads had a Q-value ≥ 20. In the clean reads of each sample, 563,551,725±68,838,132 bases passed the filter, of which over 98.9% had a Q-value ≥ 20, indicating a high quality of the obtained data. Mapping the data to the piRNABank, 179,454±58,935 reads were annotated, being a proportion of 0.74±0.27% of the clean reads (Supplementary Table S3). Sequence analysis of the annotated piRNAs revealed a strong propensity for a 5’ uridine nucleoside, which is consistent with prior reports of a 5’ uridine bias in piRNAs. According to the criteria described in the study design, nine piRNAs were found aberrantly up-regulated and six piRNAs down-regulated in the sera of CRC patients compared with normal controls (Supplementary Table S4). Since we aimed to develop potential biomarkers which were practical and convenient to identify CRC in clinical application, we focused on the nine up-regulated piRNAs and further tested their expression in the training phase. Identification of an internal control The results of the small RNA sequencing analysis revealed that piR-001918, 9

piR-017716 and piR-018883 were expressed in all serum samples, and their fold change between CRC patients and normal controls was small (1.000, 1.087 and 1.094, respectively). These three piRNAs were considered as candidate internal controls for quantification of serum piRNAs, and their expression was measured in randomly selected 40 serum samples of controls and 40 serum samples of CRC patients using RT-qPCR. The stability value calculated from the qPCR data by NormFinder ranked piR-001918 as the most suitable internal control. Evaluation of the up-regulated piRNAs during the training phase The nine up-regulated piRNAs discovered in phase one was further tested on an independent training cohort of 140 CRC patients and 140 normal controls. Among the nine piRNAs, piR-000330, piR-022421, piR-017458 and piR-001169 had Ct values greater than 30 in over 5% of samples. Both piR-019544 and piR-017723 exhibited similar expression levels between CRC patients and normal controls. PiR-020619, piR-020450 and piR-020814 were confirmed to be up-regulated in CRC patients compared with normal controls (Supplementary Table S5, Fig. 1A). The diagnostic performance of piR-020619, piR-020450 and piR-020814 was evaluated by a ROC curve analysis, and the AUCs were 0.871 (sensitivity = 84.29%, specificity = 76.43%), 0.841 (sensitivity = 81.43%, specificity = 75.00%) and 0.680 (sensitivity = 58.57%, specificity = 72.86%), respectively (Supplementary Table S6, Fig. 1B). Establishment of a predictive CRC piRNA panel A stepwise logistic regression model to estimate the risk of being diagnosed with CRC that included piR-020619, piR-020450 and piR-020814 was applied to the training cohort. PiR-020619 and piR-020450 turned out to be significant predictors of CRC risk. The predicted probability of being diagnosed with CRC from the logit model based on the two-piRNA panel was calculated using the following equation: logit(p) = -3.793 + 0.592 × piR-020619 + 0.498 × piR-020450. The calculated probability was used to construct a ROC curve to evaluate the diagnostic performance of the piRNA panel. The panel yielded an AUC of 0.879, which was larger than those for CEA (0.657, P < 0.0001) and CA19-9 (0.557, P < 0.0001). The sensitivity for the panel was 86.43%, which was higher than those for CEA (37.86%, P < 0.0001) and CA19-9 (18.57%, P < 10

0.0001). The specificity for the panel was 89.29%, which was similar to those for CEA (93.57%, P = 0.0769) and CA19-9 (92.86%, P = 0.1821) (Supplementary Table S6, Fig. 1C). Validation of the piRNA panel During the validation phase, CRC patients also displayed higher levels of serum piR-020619 and piR-020450 than did normal controls (Fig. 2). The parameters estimated from the training phase were used to predict the probability of being diagnosed with CRC in the two independent validation cohorts. In validation cohort 1, the piRNA panel demonstrated an AUC of 0.883 (larger than those for CEA [0.658, P < 0.0001] and CA19-9 [0.596, P < 0.0001]), a sensitivity of 86.00% (higher than those for CEA [40.00%, P < 0.0001] and CA19-9 [27.00%, P < 0.0001]), and a specificity of 92.00% (similar to those for CEA [93.00%, P = 1.0000] and CA19-9 [95.00%, P = 0.4503]) (Fig. 2A, Supplementary Table S6). In validation cohort 2, the piRNA panel demonstrated an AUC of 0.913 (larger than those for CEA [0.681, P < 0.0001] and CA19-9 [0.625, P < 0.0001]), a sensitivity of 88.75% (higher than those for CEA [42.50%, P < 0.0001] and CA19-9 [28.75%, P < 0.0001]), and a specificity of 93.75% (similar to those for CEA [93.75%, P = 0.4795] and CA19-9 [96.25%, P = 0.4795]) (Fig. 2B, Supplementary Table S6). When the data from the two cohorts were processed in a mixed-up manner, similar results were obtained (Fig. 2C, Supplementary Table S6). Additionally, the expression of piR-020619 and piR-020450 was detected in 12 paired serum samples from CRC patients before and one month after surgery. Both the piRNAs showed decreased expression levels in post-operative samples (Fig. 3). Logistic Regression Analysis of the piRNA Panel Next, we performed a stepwise binary logistic regression analysis to evaluate whether the diagnostic usefulness of the piRNA panel was affected by clinical features of CRC cases, including age, gender, tumor location, TNM stage, lymph node metastasis and distant metastasis. The results showed that the piRNA panel remained a strong predictor of CRC regardless of these subgroupings of the patients in the training and validation cohorts (Supplementary Table S7). Diagnostic performance of the piRNA panel for small colorectal tumors 11

There were 71 CRC patients with tumors ≤ 3cm in the training and validation cohorts of this study. We assessed the performance of the piRNA panel in differentiating the small-tumor patients from normal controls. The piRNA panel presented an AUC of 0.863, which was larger than those for CEA (0.587, P < 0.0001) and CA19-9 (0.548, P < 0.0001). The sensitivity for the panel was 81.69%, higher than those for CEA (23.94%, P < 0.0001) and CA19-9 (15.49%, P < 0.0001). The specificity for the panel was 90.94%, similar to those for CEA (93.44%, P = 0.0991) and CA19-9 (94.06%, P = 0.0051) (Fig. 4A, Supplementary Table S8). Among the 71 patients with small tumor lesions, 54 patients were CEA negative and 60 patients were CA19-9 negative. The piRNA panel was able to distinguish the CEA-negative and CA19-9-negative patients from normal controls, as evidenced by an AUC of 0.844 (sensitivity = 77.78%, specificity = 90.94%) and 0.855 (sensitivity = 80.00 %, specificity = 90.94), respectively (Fig. 4B and 4C, Supplementary Table S8). Diagnostic performance of the piRNA panel for early-stage CRC There were 173 early-stage (TNM stages I-II) CRC patients in the training and validation cohorts of this study. We further assessed the performance of the piRNA panel in differentiating the early-stage CRC patients from normal controls. The piRNA panel exhibited an AUC of 0.839, which was larger than those for CEA (0.595, P < 0.0001) and CA19-9 (0.536, P < 0.0001). The piRNA panel had a sensitivity of 76.79%, which was higher than those for CEA (25.60%, P < 0.0001) and CA19-9 (13.10%, P < 0.0001), and had a specificity of 90.94%, which was similar to those for CEA (93.44%, P = 0.0991) and CA19-9 (94.06%, P = 0.0551) (Fig. 4D, Supplementary Table S8). Among the 173 early-stage patients, 125 were CEA negative and 146 were CA19-9 negative. The piRNA panel was also able to distinguish the CEA-negative and CA19-9-negative patients from normal controls, with an AUC of 0.839 (sensitivity = 76.80%, specificity = 90.94%) and 0.828 (sensitivity = 74.66 %, specificity = 90.94%), respectively (Fig. 4E and 4F, Supplementary Table S8). Diagnostic performance of piR-020619 and piR-020450 for CRA To evaluate the diagnostic performance of piR-020619 and piR-020450 for pre-cancerous CRC lesions, we measured the expression of the two piRNAs in the sera 12

of 40 CRA patients, and observed an elevated expression compared with control individuals (Fig. 5). ROC curve analyses revealed that piR-020619 and piR-020450 could discriminate CRA patients from normal controls, with an AUC of 0.701 (sensitivity = 82.50%, specificity = 55.90%) and 0.689 (sensitivity = 67.50%, specificity = 72.50%), respectively. Combining the two piRNAs resulted in an increased AUC of 0.779 with 72.50% sensitivity and 76.60% specificity (Fig. 6). Serum piR-020619 and piR-020450 levels in patients with lung, breast and gastric cancers The expression levels of piR-020619 and piR-020450 in the sera of lung, breast and gastric cancer patients were similar to those of normal controls (Fig. 5), suggesting that the two piRNAs could serve as CRC-specific early diagnostic biomarkers. Discussion PIWI-interacting RNAs were first discovered in D.melanogaster testis tissue in 2001, and were initially thought to be expressed solely in gonadal tissues, where they function to safeguard the germline genome against transposon-induced mutations. Recent studies demonstrate extensive expression of piRNAs in various human somatic tissues, where they play an important role in maintaining cellular homeostasis by silencing transposable elements, keeping normal DNA methylation and histone modification patterns, controlling mRNA stability, modulating transcription factor activity, as well as regulating protein phosphorylation and protein-protein interactions (20-25). Dysregulated piRNA expression leads to amplification and mobilization of transposable elements, hypermethylation or hypomethylation of genomic DNA, and disruption of normal histone modifications (26-30). These abnormalities further cause a variety of genomic instability events such as insertion mutations, gene amplifications and chromosomal rearrangements (31). Since genomic instability is a major driving force of carcinogenesis and cancer evolution (32), some piRNAs may be dysregulated at very early phases of cancer development. In this study, for the first time, we discovered that expression of serum piR-020619 and piR-020450 was significantly elevated in CRA and early-stage CRC patients compared with normal controls, suggesting that the 13

aberrant expression of the two piRNAs occurs early in the precancerous stages of CRC. Activation of the two piRNAs may contribute to the pathological process of CRC initiation, and their abnormal up-regulation in sera can serve as a sensitive indicator for early CRC detection. Most early-stage CRC patients do not present any phenotypic symptoms, and are often missed on routine physical examination. The conventional serum tumor markers CEA and CA19-9 are useful in therapeutic evaluation and recurrence surveillance, but are still inadequate for CRC screening mainly because of low sensitivity in early-stage patients (3). In our study cohorts, CEA and CA19-9 indeed demonstrated poor performance for early diagnosis of CRC, as their sensitivities were only 25.60% and 13.10%, respectively. In comparison, the two-piRNA panel established in our study generated a greatly improved sensitivity of 76.79% in detecting early-stage CRC. In recent years, the abnormally methylated SEPT9 gene has been tested for utility in CRC screening. However, the reported sensitivities of plasma SEPT9 methylation in diagnosing early-stage patients often vary between over 30% to less than 70% (33-35), which is lower than that for our piRNA panel. Thus, the two-piRNA panel may be a useful non-invasive biomarker for early-stage CRC screening in asymptomatic populations. Up to 85% of CRCs originate from adenomatous polyps. Timely recognition and removal of CRA can prevent the benign precancerous lesions from progressing to malignancy. Thus far, few effective blood biomarkers for CRA have been developed. In this study, we found that the expression levels of serum piR-020619 and piR-020450 were higher in CRA patients than in normal controls, and that combined use of the two piRNAs had significant diagnostic value for CRA with a sensitivity of 72.50%. Therefore, the two piRNAs may be suitable for the early detection of precancerous lesions of CRC and help reduce CRC incidence and mortality. Circulating tumor biomarkers can be organ-specific or non-specific. Organ-specific biomarkers provide the advantage of directly identifying tumor tissue sites. The serum levels of the tumor markers CEA and CA19-9 can be elevated in multiple malignant conditions, making it difficult to specify the precise lesion location. In this study, the 14

expression of serum piR-020619 and piR-020450 was increased in CRC patients but not in patients with lung, breast and gastric cancers, suggesting a potential of the two piRNAs to serve as CRC-specific circulating tumor biomarkers. One published study focusing on several types of human normal and tumor tissues discovered that the dysregulation of piRNA expression in tumor tissues was organ-specific (36). This specificity in both sera and tissues of cancer patients suggests significant value of piRNAs as tumor biomarkers. The expression levels of serum piR-020619 and piR-020450 significantly decreased after the colorectal tumors were surgically removed, suggesting that the excessive circulating piR-020619 and piR-020450 in the patients might originate from CRC tissues. They could be over-expressed in CRC cells and directly promote the cellular malignant behaviors such as proliferation, migration, invasion, stemness-maintenance or drug-resistance by disrupting the epigenetic regulation of the transposable elements, protein-coding genes or other non-coding RNAs vital for cancer risk. We will investigate the roles of the two piRNAs in CRC initiation and development in future work. In conclusion, serum piR-020619 and piR-020450 are promisingly novel CRC-specific biomarkers that can effectively detect small CRC lesions, early-stage CRC and CRA. This study not only provides potential tools for CRC screening in asymptomatic populations, but also implies that circulating piRNAs are a valuable source for developing new non-invasive tumor biomarkers.

Disclosure of Potential Conflicts of Interest The authors declare no potential conflicts of interest.

Acknowledgments This work was supported by the National Natural Science Foundation of China 81760552 (Z.F. Wang) and 81860534 (H. Yang), by the Program of the Inner Mongolia Natural Science Foundation 2016MS0824 (Z.F. Wang), by the Inner Mongolia Science & Technology Plan Project 201802152 (Z.F. Wang), by the Supporting Program for 15

Outstanding Youth in Science and Technology of Inner Mongolia Autonomous Region NJYT-17-B30 (Z.F. Wang), by the Tumor Biotherapy Collaborative Innovation Project (Y.F. Jia), by the “Keji Baiwan Gongcheng” project of Inner Mongolia medical university YKD2015KJBW008 (Z.F. Wang), and by the TransGen dream fund for life science 2018-04 (Z.F. Wang).

References 1. Douaiher J, Ravipati A, Grams B, Chowdhury S, Alatise O, Are C. Colorectal cancer—global burden, trends, and geographical variations. J Surg Oncol 2017;115:619-30. 2. Hofsli E, Sjursen W, Prestvik W, Johansen J, Rye M, Tranø G, et al. Identification of serum microRNA profiles in colon cancer. Br J Cancer 2013;108:1712-9. 3. Vatandoost N, Ghanbari J, Mojaver M, Avan A, Ghayour-Mobarhan M, Nedaeinia R, et al. Early detection of colorectal cancer: from conventional methods to novel biomarkers. J Cancer Res Clin Oncol 2016;142:341-51. 4. Chandra Gupta S, Nandan Tripathi Y. Potential of long non-coding RNAs in cancer patients: from biomarkers to therapeutic targets. Int J Cancer 2017;140:1955-67. 5. Peng H, Wang J, Li J, Zhao M, Huang S-k, Gu Y-y, et al. A circulating non-coding RNA panel as an early detection predictor of non-small cell lung cancer. Life Sci 2016;151:235-42. 6. Umu SU, Langseth H, Bucher-Johannessen C, Fromm B, Keller A, Meese E, et al. A comprehensive profile of circulating RNAs in human serum. RNA Bio 2018;15:242-50. 7. Wikberg ML, Myte R, Palmqvist R, van Guelpen B, Ljuslinder I. Plasma mi RNA can detect colorectal cancer, but how early? Cancer Med 2018;7:1697-705. 8. Guo S, Zhang J, Wang B, Zhang B, Wang X, Huang L, et al. A 5-serum miRNA panel for the early detection of colorectal cancer. OncoTargets Ther 2018;11:2603-14. 9. Hollis M, Nair K, Vyas A, Chaturvedi LS, Gambhir S, Vyas D. MicroRNAs potential utility in colon cancer: Early detection, prognosis, and chemosensitivity. World J Gastroenterol 2015;21:8284-92. 16

10. Weng W, Li H, Goel A. Piwi-interacting RNAs (piRNAs) and cancer: emerging biological concepts and potential clinical implications. BBA-Rev Cancer 2019;1871:160-9. 11. Chalbatani GM, Dana H, Memari F, Gharagozlou E, Ashjaei S, Kheirandish P, et al. Biological function and molecular mechanism of piRNA in cancer. Pract Lab Med 2018:e00113. 12. Weng W, Liu N, Toiyama Y, Kusunoki M, Nagasaka T, Fujiwara T, et al. Novel evidence for a PIWI-interacting RNA (piRNA) as an oncogenic mediator of disease progression, and a potential prognostic biomarker in colorectal cancer. Mol Cancer 2018;17:16. 13. Ai L, Mu S, Sun C, Fan F, Yan H, Qin Y, et al. Myeloid-derived suppressor cells endow stem-like qualities to multiple myeloma cells by inducing piRNA-823 expression and DNMT3B activation. Mol Cancer 2019;18:88. 14. Tan L, Mai D, Zhang B, Jiang X, Zhang J, Bai R, et al. PIWI-interacting RNA-36712 restrains breast cancer progression and chemoresistance by interaction with SEPW1 pseudogene SEPW1P RNA. Mol Cancer 2019;18:9. 15. Sadik N, Cruz L, Gurtner A, Rodosthenous RS, Dusoswa SA, Ziegler O, et al.: Extracellular RNAs: a new awareness of old perspectives. Methods Mol Biol 2018; 1740:1-15. 16. Hong Y, Wang C, Fu Z, Liang H, Zhang S, Lu M, et al. Systematic characterization of seminal plasma piRNAs as molecular biomarkers for male infertility. Sci Rep 2016;6:24229. 17. Kurth HM, Mochizuki K. 2′-O-methylation stabilizes Piwi-associated small RNAs and ensures DNA elimination in Tetrahymena. Rna 2009;15:675-85. 18. Yang X, Cheng Y, Lu Q, Wei J, Yang H, Gu M. Detection of stably expressed piRNAs in human blood. Int J Clin Exp Med 2015;8:13353-8. 19. Fehlmann T, Reinheimer S, Geng C, Su X, Drmanac S, Alexeev A, et al. cPAS-based sequencing on the BGISEQ-500 to explore small non-coding RNAs. Clin Epigenetics 2016;8:123-32. 20. Czech B, Munafò M, Ciabrelli F, Eastwood EL, Fabry MH, Kneuss E, et al. 17

piRNA-guided genome defense: From biogenesis to silencing. Annu Rev Genet 2018;52:131-57. 21. Fu A, Jacobs DI, Hoffman AE, Zheng T, Zhu Y. PIWI-interacting RNA 021285 is involved in breast tumorigenesis possibly by remodeling the cancer epigenome. Carcinogenesis 2015;36:1094-102. 22. Huang XA, Yin H, Sweeney S, Raha D, Snyder M, Lin H. A major epigenetic programming mechanism guided by piRNAs. Dev Cell 2013;24:502-16. 23. Peng L, Song L, Liu C, Lv X, Li X, Jie J, et al. piR-55490 inhibits the growth of lung carcinoma by suppressing mTOR signaling. Tumor Biol 2016;37:2749-56. 24. Yin J, Jiang XY, Qi W, Ji CG, Xie XL, Zhang DX, et al. piR‐823 contributes to colorectal tumorigenesis by enhancing the transcriptional activity of HSF 1. Cancer Sci 2017;108:1746-56. 25. Mai D, Ding P, Tan L, Zhang J, Pan Z, Bai R, et al. PIWI-interacting RNA-54265 is oncogenic and a potential therapeutic target in colorectal adenocarcinoma. Theranostics 2018;8:5213-30. 26. Yu Y, Xiao J, Hann SS. The emerging roles of PIWI-interacting RNA in human cancers. Cancer Manage Res 2019;11:5895-909. 27. Zhang H, Ren Y, Xu H, Pang D, Duan C, Liu C. The expression of stem cell protein Piwil2 and piR-932 in breast cancer. Surg Oncol 2013;22:217-23. 28. Yan H, Wu Q, Sun C, Ai L, Deng J, Zhang L, et al. piRNA-823 contributes to tumorigenesis by regulating de novo DNA methylation and angiogenesis in multiple myeloma. Leukemia 2015;29:196-206. 29. Wu D, Fu H, Zhou H, Su J, Zhang F, Shen J. Effects of Novel ncRNA Molecules, p15‐piRNAs, on the Methylation of DNA and Histone H3 of the CDKN2B Promoter Region in U937 Cells. J Cell Biochem 2015;116:2744-54. 30. Liu J, Zhang S, Cheng B. Epigenetic roles of PIWI‑interacting RNAs (piRNAs) in cancer metastasis. Oncol Rep 2018;40:2423-34. 31. Moyano M, Stefani G. piRNA involvement in genome stability and human cancer. J Hematol Oncol 2015;8:38. 32. Negrini S, Gorgoulis VG, Halazonetis TD. Genomic instability—an evolving 18

hallmark of cancer. Nat Rev Mol Cell Biol 2010;11:220-8. 33. Church TR, Wandell M, Lofton-Day C, Mongin SJ, Burger M, Payne SR, Castaños-Vélez E, et al. Prospective evaluation of methylated SEPT9 in plasma for detection of asymptomatic colorectal cancer. Gut 2014;63:317-25. 34. Lee HS, Hwang SM, Kim TS, Kim D-W, Park DJ, Kang S-B, et al. Circulating methylated septin 9 nucleic acid in the plasma of patients with gastrointestinal cancer in the stomach and colon. Transl Oncol 2013;6:290-6. 35. Wang Y, Chen P-M, Liu R-B. Advance in plasma SEPT9 gene methylation assay for colorectal cancer early detection. World J Gastrointest Oncol 2018;10:15-22. 36. Martinez VD, Vucic EA, Thu KL, Hubaux R, Enfield KS, Pikor LA, et al. Unique somatic and malignant expression patterns implicate PIWI-interacting RNAs in cancer-type specific biology. Sci Rep 2015;5:10423.

Figure legends： Figure 1. Diagnostic performance of piR-020619 and piR-020450 for CRC in the training phase. A) Expression levels of serum piR-020619 and piR-020450 were significantly higher in CRC patients than in normal controls (P < 0.0001). B) ROC curves of piR-020619 (AUC = 0.871) and piR-020450 (AUC = 0.841) for CRC detection. C) ROC curves of the two-piRNA panel, CEA and CA19-9 for CRC detection. The two-piRNA panel displayed significantly higher performance than CEA and CA19-9 for CRC detection (P < 0.0001).

Figure 2. Diagnostic performance of piR-020619 and piR-020450 for CRC in the validation phase. A) In validation cohort 1, levels of serum piR-020619 and piR-020450 were significantly elevated in CRC patients compared with normal controls (P < 0.0001); the two-piRNA panel displayed significantly higher performance than CEA and CA19-9 for CRC detection (P < 0.0001). B) In validation cohort 2, levels of serum piR-020619 and piR-020450 were also significantly elevated in CRC patients compared with normal controls (P < 0.0001); the two-piRNA panel also displayed significantly higher performance than CEA and CA19-9 for CRC detection (P < 0.0001). C) When the data from the two cohorts were mixed together, the analyses gave similar results.

Figure 3. Changes of serum piR-020619 and piR-020450 levels in 12 CRC patients before and one month after surgical removal of the tumors. The levels of serum piR-020619 and piR-020450 significantly decreased after surgical removal of the tumors (P = 0.01 and 0.008, respectively).

Figure 4. Diagnostic performance of the two-piRNA panel for small-size and early-stage CRC in the combined training and validation cohorts. A) The two-piRNA panel displayed significantly higher performance than CEA and CA19-9 for detection of small-size CRC (P < 0.0001). B) The ROC curve of the two-piRNA panel for detection of CEA-negative small CRC (AUC = 0.844). C) The ROC curve of the two-piRNA

panel for detection of CA19-9-negative small CRC (AUC = 0.855). D) The two-piRNA panel displayed significantly higher performance than CEA and CA19-9 for detection of early-stage CRC (P < 0.0001). E) The ROC curve of the two-piRNA panel for detection of CEA-negative early-stage CRC (AUC = 0.839). F) The ROC curve of the two-piRNA panel for detection of CA19-9-negative early-stage CRC (AUC = 0.828).

Figure 5. Expression of piR-020619 and piR-020450 in sera of normal controls, and patients with CRA, stage I-II CRC, stage III–IV CRC, lung cancer, breast cancer and gastric cancer. The levels of serum piR-020619 and piR-020450 in normal controls were significantly lower than those in patients with CRA, stage I-II CRC and stage III– IV CRC (P < 0.01), while were similar to those in patients with lung, breast and gastric cancers (P > 0.05).

Figure 6. ROC curves of piR-020619, piR-020450 and the combination of piR-020619 and piR-020450 for discriminating CRA from normal controls (AUC = 0.701, 0.689 and 0.771, respectively).

Serum PIWI-interacting RNAs piR-020619 and piR-020450 are promising novel biomarkers for early detection of colorectal cancer

Zhenfei Wang, Hao Yang, Daguang Ma, et al.

Cancer Epidemiol Biomarkers Prev Published OnlineFirst February 17, 2020.

Updated version Access the most recent version of this article at: doi:10.1158/1055-9965.EPI-19-1148

Supplementary Access the most recent supplemental material at: Material http://cebp.aacrjournals.org/content/suppl/2020/02/15/1055-9965.EPI-19-1148.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet Manuscript been edited.

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cebp.aacrjournals.org/content/early/2020/02/15/1055-9965.EPI-19-1148. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.