A Physically Active Status Affects the Circulating Profile of Cancer-Associated miRNAs

Circulating miRNAs are ideal diagnostics and prognostics biomarkers in cancer since altered levels of specific miRNAs have been associated to development/progression of several cancers. Physical activity is a recognized preventive strategy against several cancers, but it may also modify the baseline levels of cancer-associated miRNAs and, hence, may act as a confounding pre-analytical variable. This study aimed at understanding whether physical activity-dependent changes in cancer-associated circulating miRNAs profile could act as a confounding variable. A panel comprising 179 miRNAs was assayed in plasma from 20 highly trained and 10 sedentary men. RT-qPCR data were analyzed with the 2−2ΔΔCT methods and normalized on hsa-miR-320d, as determined by bioinformatics analysis. miRNAs associated with the diagnosis of the most prevalent cancers were considered. Only those miRNAs, relevantly associated with cancers, found ≥2-fold up- or downregulated in highly trained subjects compared to sedentary were disclosed. The results reveal that chronic physical activity determined modifications altering the baseline level of several cancer-associated miRNAs and, hence, their diagnostic and prognostic potential. In conclusion, based on our results, a physically active status emerges as an important pre-analytical variable able to alter the basal level of circulating miRNAs, and these alterations might be considered as potentially misleading the analytical output.


Introduction
MicroRNAs (miRNAs) are small non-coding RNA molecules (18)(19)(20)(21)(22) nucleotides in length), actively regulating gene expression [1,2]. Several miRNAs display a certain degree of cell-or tissue-specificity, while others are more broadly expressed [3]. Since released in biological fluids [4], they can be detected and quantified in blood [5]. Although circulating miRNAs levels are very low and, however, much lower than in the tissue/cells source [6], they give particular advantage in the case of tissues hardly subjectable to bioptic procedures (e.g., bone and cartilage) [7,8] and/or to serial monitoring the progress of diseases, such as osteoporosis, fracture risk [9] and tumors [10], as well as the response to physical activity [11,12].
During neoplastic transformation, miRNAs may be aberrantly expressed and miRNA signatures can be associated with specific tumors and, possibly, with a given tumor stage [13]. Altered circulating miRNA profiles have been associated with transformation and tumor growth, progression, metastasis and development of drug resistance.
Consequently, circulating miRNAs profiling is of great interest in cancer diagnosis and prognosis [14]. Further, circulating miRNAs do not only passively mark a pathological condition, by mirroring the tumor cell content, but they can actively act as paracrine/endocrine mediators that target other tissues to determine, for example, the appropriate microenvironment to host a metastasis [15].
Physical inactivity (PI) is the fourth leading cause of death (5.5% of deaths globally) after hypertension, cigarette smoking and hyperglycemia and just before obesity and overweight [16]. PI independently associates with an increased incidence of cancers in general and specifically for certain tumors (e.g., colorectal, lung, breast, prostate, ovarian, endometrial and esophageal) [17]. The net energy surplus consequent to the sedentary behavior is the underlying cause of the onset of these diseases [18]. The preventive role of exercise training, and particularly of endurance training, is explicated towards a plethora of not completely understood molecular mechanisms [19]. Among these, acute and chronic physical activity (PA) profoundly affects the circulating miRNA profile [20]. The anti-cancer effect of PA may thus be mediated, at least in part, by the alteration of miRNA expression profile: PA may determine a miRNA-based regulatory network that make neoplastic transformation, tumor progression and metastasis less probable [21]. From a diagnostic point of view, however, the physiological differences between sedentary and trained subjects must be considered in a clinical setting since they represent an important source of variability. As such, PA is a pre-analytical variable to be accounted in routine analysis especially when highly sensitive analytes are assayed, such as miRNAs [12,22], but thus far no published papers have investigated how PA affects the circulating miRNA-based diagnosis/prognosis on cancer [21]. Based on this background, this study aimed at determining to what extent a physically active status may act as a pre-analytical variable potentially modifying the diagnostic value of miRNAs in cancer, by comparing the circulating miRNA profile of highly trained subjects and sedentary individuals.

Study Population
We recruited 24 volunteer young adult males, preselected based on their physical activity profile. Eligibility criteria were: age range 25-40 years, healthy status, normal body mass index (BMI, 20 ≥ BMI ≥ 25 kg/m 2 ), non-smokers, no recent (in the preceding month) or chronic diseases and/or medications. According to the guidelines from the American College of Sports Medicine (ACSM) [23], participants were grouped as either sedentary or endurance trained. Sedentary individuals (n = 10) perform less than 30 min, 5 days/week (or 150 min/week) of moderate intensity PA. In our setting, endurance trained individuals (n = 14) were non-professional mountain ultra-trail athletes accustomed to vigorous-tohigh intensity, high volume, long-lasting endurance PA [23] that were recruited at the end of the 9th week of preparation (tapering week, during which the training volumes are halved [24]) for an ultra-trail mountain marathon. All individuals gave their written consent after being informed about study aim, protocol, procedures, associated benefits and eventual risks, in accordance with the Helsinki Declaration (SportMarker protocol approved by the Comitato Etico Ospedale San Raffaele, Milano, Italy and registered on ClinicalTrials.gov NCT03386981).

Blood Sampling
All participants were asked to abstain from alcohol, keep a regular diet and avoid any pro-inflammatory activity (shaving, waxing, etc.) in the week before the sampling. Sedentary participants were asked to avoid any PA in the 48 h preceding blood sampling. Trained subjects, instead, were sampled the day before an ultra-trail mountain marathon. According to the most up-to-date pre-analytical warnings, venous blood was collected in K2EDTA tubes (BD Vacutainer ® , Becton Dickinson, Franklin Lakes, NJ, USA), and plasma, obtained following centrifugation at standard conditions of 1300× g for 10 min at room temperature (22 • C), was immediately frozen at −80 • C until assayed.

miRNA Profiling
miRNA profiling analysis was performed accordingly [20]. Briefly, miRNA-enriched total RNA fraction was isolated from equal aliquot of plasma pooled from each subject using miRCURY™ RNA Isolation Kit (Exiqon A/S, Vedbaek, Denmark); isolation efficiency was checked with synthetic oligonucleotides added at known concentrations (spike-in: UniSp2, UniSp4, UniSp5). Reverse transcription reaction was conducted with miRCURY LNA™ Universal RT miRNA PCR. UniSp6 and cel-39 were used as reaction controls. qPCR was performed on a StepOne Plus (Applied Biosystem, Foster City, CA, USA) using serum/plasma miRCURY LNA™ miRNA focus panel and ExiLENT SYBR Green Master Mix (Exiqon). The qPCR reaction was conducted as follows: polymerase activation for 10 min at 95 • C, amplification (40 cycles: 10 s at 95 • C and 60 s at 60 • C) and melting curves. The analytical output, expressed as quantification cycle (Cq), was analyzed with GenEx software v6 (Exiqon). Three inter-plate (IPC) calibrators, with specific primer pairs and DNA template, were used to adjust Cq values from different plate runs of different experiments. Only miRNAs with a Cq <37 were further considered in the analysis. The 2 −∆∆CT method was applied to calculate the relative expression of each miRNA, using hsa-miR-320d as normalizers. This miRNA was identified as the most appropriate based on a study of miRNA stability, as previously reported [20]. Hemolysis was checked by hsa-miR-23-a-to-hsa-miR-451 ∆Cq difference (positive if >7). Analyses were conducted in triplicate.

Statistical Analysis
Given their distribution, data from the two cohorts were compared throughout a parametric unpaired t-test with Welch's correction. Differences were considered significant when p < 0.05 (Prism ® v6.01, GraphPad Software, La Jolla, CA, USA). Only miRNAs with a significant (p < 0.05) fold change ≥ ±2 were considered.

Selection of Cancer-Associated miRNAs
The selection strategy of cancer-associated miRNAs and signatures is summarized in Figure 1.

Discussion
Several circulating miRNAs, both singularly and grouped in signatures, have been associated with cancer development risk, detection and progression [13]. In parallel, many studies have demonstrated that regular PA reduces the global risk of developing a cancer as well as improves prognosis and reduces the risk of metastasis and side effects of the therapy in subjects treated for primitive tumor [17,94,95]. However, the impact of PA on circulating miRNAs associated with cancer is mostly neglected, as recently reviewed in [21]. Therefore, this study provides first evidence about the alteration in the circulating level of cancer associated miRNAs in response to chronic PA.
The main question this study aimed to address is whether physical active status may represent a modifier of the baseline circulating miRNA profile and, hence, may act as a pre-analytical variable that affect the diagnostic potential of miRNAs associated with cancer. In our study, a panel of 179 circulating miRNAs was analyzed. First, for each miRNA, the association with the five most frequent cancers in males (LC, PC, CC, HC and GEC) was investigated; only those miRNAs with an established diagnostic potential, defined as the ability to discriminate between cancer patients and healthy subjects and associated to a significant AUC ≥ 0.7, in ROC analysis, was considered. Based on this assumption, we identified 133 miRNAs. Second, the expression profile of the 179-miRNA panel was determined in plasma samples of highly trained subjects and their sedentary counterparts. Only miRNA significantly ≥2 fold up-or downregulated in trained subjects, compared to sedentary subjects, were further investigated. As expected, our two cohorts are featured by a different miRNA profile: the expression level of 36 miRNAs (25 upregulated and 11 downregulated) was significantly (p < 0.05) and relevantly (≥2 fold) altered in physically active subjects (Table 1). Among these, 26 miRNAs (15 upregulated and 11 downregulated) have been previously associated with at least one of the most frequent cancers in males (see the tables in the Supplementary Materials). Of the 15 upregulated miRNAs, eight have been associated with LC, two with PC and HC and five with CC and GEC. Of the 11 downregulated miRNAs, five have been associated with LC, one with PC, seven with CC, four with HC and six with GEC. These results demonstrate that the adaptation to regular exercise determines a baseline modification of several circulating miRNAs and this fact, if not considered, may limit their clinical applicability, thus confirming previous hypothesis [12]. Similarly, this consideration can be extended to miRNA signatures since 19 miRNAs that resulted significantly modulated by PA are included in diagnostic miRNA signatures from the literature. miRNA signature means a series of miRNAs that have an overall diagnostic potential. A diagnostic miRNA signature does not necessarily include all single miRNA found with diagnostic potential in a specific study; a miRNA signature can also include miRNAs that, singularly, do not have a diagnostic potential. Moreover, a signature can include both miRNAs up-and -downregulated in a specific experimental setting. In this study, signatures from the literature that include at least one of the modulated miRNAs in our trained cohort were considered. Based on our literature search, 63 signatures were identified. Of the signatures that include at least one modulated miRNA, in our experimental setting, 24 are associated with LC diagnosis, 0 with PC, 7 with CC, 4 with HC and 5 with GEC. Thereby, the presence of miRNAs affected by physical active status within a diagnostic signature may reduce its clinical reliability. Consequently, the greater is the number of miRNAs modulated by PA within a signature, the lower is the clinical reliability of the signature itself.
From these results, the training status-dependent alterations of the circulating levels of cancer-associated miRNAs in healthy individuals are remarkable. Importantly, 107 miRNAs (of the 133 circulating miRNAs associated with the five cancers) and 23 signatures resulted not affected by regular training, and, given their robustness, the analysis of these miRNAs or signatures might be privileged. Alterations of hematological and biochemical parameters in response to chronic PA are well documented [22], as well as, although to a lesser extent, for miRNAs [12]. Chronic PA alters the level of miRNAs involved in skeletal muscle and bone metabolisms [20], inflammation, immunity and angiogenesis [96], as well as those associated with metabolic syndrome and diabetes [97], cardiovascular disease [98], skeletal muscle and bone diseases [9,99]. However, the measurement of these indices is challenging. Sample-associated and subject-associated pre-analytical variables must be considered. The former includes patient preparation (timing and posture), sample collection (sample identification and labeling, type of disposables, order of draw, phlebotomy procedure and contamination), transportation (length and environmental conditions), preparation (centrifugation length, speed and temperature and aliquot preparation) and storage (length, temperature and freeze-thaw cycles) [100]. Subject-associated variables, besides the uncontrollable but considerable biological ones (e.g., age, gender, ethnicity, diseases and circannual/seasonal rhythm), include medications, physical activity status (acute and chronic) and menstrual cycle phase, based on which the sampling can be opportunely temporized [101]. As we recently demonstrated for the same set of miRNAs, many of these variables can affect their determination [102,103], and, according to the present study, the physical activity status must also be considered. Therefore, a strict definition of the pre-analytical variables affecting circulating miRNAs measurement, including PA, is needed to allow a correct interpretation of the final analytical output.
Although based on a solid experimental approach relying on a meticulous control of the pre-analytical phase, and a robust post-analytical data analysis based on a well-defined normalization strategy, this study suffers some limitations that should be taken into consideration. First, the identification of miRNAs and miRNA signatures associated with cancer diagnosis is based on a literature search that, although extensive and systematically conducted, may be not fully exhaustive. In this regard, since the rather not exhaustive information available, it was not possible to select the research articles taking into consideration important variables such as gender, age, medications and comorbidities. Second, different studies may have differently scored a miRNA, and, hence, although the criterion of AUC > 0.7 has been established, the actual diagnostic relevance of certain miRNAs may have been under-or overestimated. Third, due to the discrepancies among the studies the limit of having considered healthy subjects, it was not possible to determine whether PA may play a protective role against cancers, other than its only pre-analytical effect. Finally, the sample size could be considered rather small for a case-control study. However, this limitation would be, at least partially, overcome thanks to the very high homogeneity level of the subjects within each of the two cohorts in terms of subject-specific pre-analytical variables (e.g., age range and avoidance from physical activity for the sedentary cohort and physical activity level and training status for trained subjects).

Conclusions
Due to the PA contribution to alteration of circulating level of miRNAs, many of which are associated with cancer diagnostic, this study emphasized PA as a pre-analytical variable. The consistent variability of circulating miRNA induced by PA is of great relevance in diagnostics since miRNA variability may reflect the adaptive response of the whole organism to the PA status, rather than the presence of a pathological condition, as cancer. Therefore, further studies are needed to better standardize the PA effects in order to improve the clinical implementation of circulating miRNAs.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/diagnostics11050820/s1. Table S1: Fold-change of the 179 miRNAs analyzed in trained subjects compare with sedentary subjects. In bold are reported miRNAs significantly ≥2-fold upor downregulated in trained subjects compared to sedentary subject (p-value > 0.05). Table S2: List of the miRNAs and miRNA signature associated with diagnosis in lung cancer from the literature. Only miRNAs with significant clinically relevance (AUC ≥ 0.7) and signatures with at least one modulated miRNA from our experimental setting were considered. Table S3: List of the miRNAs and miRNA signature associated with diagnosis in prostate cancer from the literature. Only miRNAs with significant clinically relevance (AUC ≥ 0.7) and signatures with at least one modulated miRNA from our experimental setting were considered. Table S4: List of the miRNAs and miRNA signatures associated with diagnosis in colon cancer from the literature. Only miRNAs with significant clinically relevance (AUC ≥ 0.7) and signature with at least one modulated miRNA from our experimental setting were considered. Table S5: List of the miRNAs and miRNA signatures associated with diagnosis in liver cancer from the literature. Only miRNAs with significant clinically relevance (AUC ≥ 0.7) and signature with at least one modulated miRNA from our experimental setting were considered. Table S6: List of the miRNAs and miRNA signature associated with diagnosis in gastro-esophageal cancer from the literature. Only miRNAs with significant clinically relevance (AUC ≥ 0.7) and signatures with at least one modulated miRNA from our experimental setting were considered. Data Availability Statement: Raw data can be found at https://zenodo.org/record/4058363#.YC4 d5I9KiM8, accessed on 30 November 2020.