Whole blood miRNA expression analysis reveals miR-3613-3p as a potential biomarker for dedifferentiated liposarcoma

Abstract

BACKGROUND:

Liposarcoma constitute about 13% of all soft tissue sarcoma and are associated with a high risk of metastases. As the preoperative differentiation between benign and malign lipomatous tumors is restricted to magnetic resonance imaging, computed tomography and biopsy, we performed a miRNA array to distinguish dedifferentiated liposarcoma patients from healthy controls and lipoma patients.

METHODS:

Blood samples of patients with dedifferentiated liposarcoma, healthy controls and lipoma patients were collected. Whole blood RNA was extracted and samples of patients with dedifferentiated liposarcoma ($n=$ 6) and of healthy donors ($n=$ 4) were analyzed using an Affymetrix GeneChip miRNA Array v. 4.0. qRT-PCR was carried out to confirm the most differentially expressed miRNA; being further analyzed in an independent cohort of healthy controls as well as in lipoma patients.

RESULTS:

As shown by the microarray, two miRNAs (miR-3613-3p, miR-4668-5p) were shown to be significantly upregulated (fold change: $>$ 2.5; 0.05) in patients with dedifferentiated liposarcoma ($n=$ 6) as compared to healthy controls ($n=$ 4). miR-3613-3p was further validated by qRT-PCR to be significantly upregulated in dedifferentiated liposarcoma patients compared to an independent cohort of healthy controls ($n=$ 3) and lipoma patients ($n=$ 5).

CONCLUSION:

We identified a specific whole blood miRNA (miR-3613-3p) that may serve to distinguish between dedifferentiated liposarcoma patients and healthy controls, thus potentially serving as a specific biomarker for dedifferentiated liposarcoma.

1. Background

Liposarcoma are a heterogeneous group of adipocy-tic tumors which can be classified into four main subtypes: atypic lipomatous tumors, dedifferentiated liposarcoma, myxoid liposarcoma and pleomorphic liposarcoma [1]. Since these subtypes differ significantly in terms of clinical behaviour, metastasis rate and sensitivity to radio- and chemotherapy [2, 3, 4], the correct diagnosis is of utmost importance.

MicroRNAs (miRNAs) are small, non-coding mole-cules which regulate the expression of target genes through mRNA degradation and translation inhibition [5] and have been described as potential biomarkers in various malignancies [6, 7, 8]. Recently, it has been shown that several miRNAs are deregulated in liposarcoma tumor tissue [9, 10] and even in the plasma of patients with dedifferentiated liposarcoma [11]. However, we felt that it was crucial to establish a miRNA biomarker for liposarcoma which was easier to transfer into clinical practice: Thus, in this study, blood sample collection was carried out by means of PAXgene Blood RNA Tubes, which immediately stabilize intracellular RNA, even if stored at room temperature for up to 72 hours.

Since the availability of a non-invasive “liquid biopsy” for dedifferentiated liposarcoma might possibly improve survival rates due to earlier detection of sarcoma recurrence as well as allow more accurate differential diagnosis between the different subtypes, we performed a microarray-based miRNA screen of whole blood RNA of dedifferentiated liposarcoma patients compared to healthy controls, which was further validated by qRT-PCR.

2. Methods

2.1. Study population

The liposarcoma patients included in the study were patients receiving treatment from specialists in the interdisciplinary tumor board of the Comprehensive Cancer Center Freiburg (CCCF). Patients with a history of cancer other than liposarcoma, any type of systemic inflammatory disease or autoimmune disorder were excluded from the study. The control groups included healthy adults as well as lipoma patients matched to the liposarcoma groups in terms of body mass index (BMI).

2.2. Ethics, consent and permissions

Signed informed consent was obtained from all participants, allowing analysis of blood samples and all clinical data. The Ethics Committee of the Albert-Ludwigs University of Freiburg, Germany, approved the study. The design and performance of the study is in accordance with the Declaration of Helsinki.

2.3. Blood sampling

The blood samples were collected by puncture of the antecubital vein without tourniquet through a 20-gauge needle; each 2.5 ml of whole blood were collected and stabilized in PAXgene Blood RNA Tubes (PreAnalytiX, Hombrechtikon, Switzerland). The first 3 ml of blood were discarded. The RNA Tubes were left at least 2 hours at room temperature after blood collection; then being stored at $-$20${}^{\circ }$C. Before RNA extraction, the RNA tubes were equilibrated to room temperature. Total RNA $>$ 18 nucleotides (including miRNA) was purified manually according to the manufacturer’s protocol of the PAXgene Blood miRNA Kit (PreAnalytiX).

2.4. miRNA array

RNA quality and quantity was evaluated by capillary electrophoresis by the Fragment Analyzer and Standard sensitivity RNA Analysis kits (Advanced Analytical Technologies, Ames, IA, U.S.). Total whole blood RNA of patients with liposarcoma ($n=$ 6) and healthy controls ($n=$ 4) was then analyzed using an Affymetrix GeneChip miRNA Array v. 4.0 (Affymetrix, Santa Clara, CA, U.S.). 275 ng of total RNA was labeled with Biotin using a 3DNA Array Detection FlashTag™ Biotin HSR kit (Genisphere, Hatfield, PA, U.S.) following the manufacturer’s protocol, then being hybridized for 12 hours. GeneChip${}^{\mathrm{\circledR }}$  miRNA 4.0 arrays were washed and stained by the use of Affymetrix GeneChip Hybridization Wash and Stain Kit; then being scanned with the Affymetrix GeneChip Scanner 3000 7G (Affymetrix, Santa Clara, CA, U.S.).

2.5. Preparation of cDNA

For detection of miRNA expression levels, cDNA synthesis of whole blood RNA (275 ng) was carried out using the miScript II RT Kit (Qiagen) according to the manufacturer’s protocol. The reverse transcription reaction was incubated for 60 min at 37${}^{\circ }$C, followed by 5 min at 95${}^{\circ }$C to inactivate the miScript Reverse Transcriptase Mix.

2.6. qRT-PCR

qRT-PCR was carried out using the miScript SYBR Green PCR Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Cycling conditions consisted of an initial activation step of the HotStarTaq DNA Polymerase for 15 min at 95${}^{\circ }$C, followed by 45 cycles of denaturation for 15 sec at 94${}^{\circ }$C, annealing for 30 sec at 55${}^{\circ }$C and extension for 30 sec at 70${}^{\circ }$C with fluorescence data collection being performed during the extension step. MiScript Primer Assays were purchased from Qiagen. Ct-values were normalized to RNU6-2B, a frequently used reference gene [12] and $2^{-\mathrm{\Delta }\mathrm{\Delta }Ct}$ values [13] were analyzed using Student’s $t$-test for independent samples.

2.7. Statistics

CEL-files of raw data were produced with Affym-etrix GeneChip Command Console Software Version 4.0. (Affymetrix, Santa Clara, CA, U.S.), then further being analyzed by Partek Genomics Suite software (Version 6.14.0923; Partek, Inc., St. Louis, MO, USA). CEL-files were imported including control and interrogating probes. Arrays were normalized using quantile normalization. Probeset summarization was performed using Median Polish. Probe values were log2 transformed. In order to detect differential miRNA expression between the 3 groups, 1-way ANOVA was performed [14] and Fisher’s Least Significant Difference (LSD) was used as contrast method.

$2^{-\mathrm{\Delta }\mathrm{\Delta }Ct}$ and patients’ demographic data and blood counts were compared using Student’s $t$-test for independent samples. p-values were rounded to 4 significant digits; p-values below 0.05 were considered statistically significant.

3. Results

Dedifferentiated G2 and G3 liposarcoma patients ($n=$ 6) as analyzed in the microarray did not differ significantly from the healthy controls ($n=$ 4) concerning age, BMI, hemoglobin (Hb) level, platelet count and leukocyte count (Table 1). Information regarding disease and therapy status of the liposarcoma patients is depicted in Table 2. Whole blood RNA from patients with dedifferentiated liposarcoma and healthy controls was analyzed by Affymetrix miRNA 4.0 Arrays. Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under the accession number E-MTAB-5126 (http://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-5126).

Table 1.

Demographic patient data (age, BMI) and blood count of patients with G2 and G3 dedifferentiated liposarcoma and healthy donors

Patients	Age	BMI	Hb	Platelets	Leukocytes
	(years)	(kg/m${}^2$)	(g/dl)	(x10${}^8$/l)	(x10${}^8$/l)
G2 liposarcoma patients ($n=$ 3)	64.33 $\pm$ 4.41	24.91 $\pm$ 3.08	11.10 $\pm$ 1.75	272.00 $\pm$ 31.19	5.52 $\pm$ 0.97
G3 liposarcoma patients ($n=$ 3)	80.67 $\pm$ 4.67	28.05 $\pm$ 2.10	13.73 $\pm$ 1.94	244.00 $\pm$ 45.36	8.75 $\pm$ 3.26
G2 and G3 liposarcoma patients ($n=$ 6)	72.50 $\pm$ 4.64	26.48 $\pm$ 1.81	12.42 $\pm$ 1.31	258.00 $\pm$ 25.40	7.13 $\pm$ 1.68
Healthy controls ($n=$ 4)	67.50 $\pm$ 5.95	22.58 $\pm$ 2.51	14.33 $\pm$ 0.51	224.00 $\pm$ 42.93	5.98 $\pm$ 0.29
p-value of G2 liposarcoma patients vs. healthy controls	0.7068	0.5796	0.0971	0.4390	0.6241
p-value of G3 liposarcoma patients vs. healthy controls	0.1674	0.1746	0.7466	0.7652	0.3610

Data are presented as mean value $\pm$ standard error of mean (SEM). Age, BMI, Hb-, leukocyte- and platelet-values were rounded to 2 decimal digits; while p-values were rounded to 4 decimal digits. p-values were determined using a Student’s t-test for independent samples. Hb $=$ Hemoglobin. BMI $=$ Body Mass Index.

Table 2.

Demographic data (Age and Sex), grading and therapy status of dedifferentiated liposarcoma patients

Patient	Dedifferentiated	Age	Sex (F/M)	M0	M1	Current	Current	Localization
	liposarcoma grading	(years)				chemo-therapy	radio-therapy
1	G2	56	M		X			Retroperitoneal
2	G2	71	F	X				Shoulder
3	G2	66	F		X (pulm, oss)			Abdominal wall
4	G3	82	F		X (pulm)	X		Retroperitoneal
5	G3	87	F		X (nuchal)			Thigh
6	G3	72	M	X			X	Spermatic cord

M0 $=$ localized disease. M1 $=$ metastatic disease. Current chemotherapy/radiotherapy involves treatment within the last 6 weeks.

Hierarchical clustering of all covered human mature miRNAs and human pre-miRNAs separated dedifferentiated liposarcoma samples from healthy control samples (Fig. 1). Following hierarchical clustering, the array results were narrowed down to deregulated miRNAs with a fold change > $|$2.0$|$ and a p-value 0.05 of both G2 and G3 dedifferentiated liposarcoma compared to healthy controls (Table 3). Of the two miRNAs which met these criteria (miR-3613-3p and miR-4668-5p), only miR-3613-3p was validated to exhibit a significantly increased expression in whole blood of G2 dedifferentiated liposarcoma patients ($n=$ 3) compared to healthy controls ($n=$ 4; $p=$ 0.0172) and G3 dedifferentiated liposarcoma patients ($n=$ 3) compared to healthy controls ($p=$ 0.0059); (Fig. 2). Of note, G2 and G3 liposarcoma patients did not differ significantly in miR-3613-3p expression ($p=$ 0.6159; Fig. 2). An increased expression of miR-4668-5p in whole blood of dedifferentiated liposarcoma patients could not be validated by qRT-PCR.

Figure 1.

Hierarchical clustering of all covered human mature miRNAs and human pre-miRNAs separated dedifferentiated liposarcoma samples from control samples and G2 liposarcoma samples from G3 liposarcoma samples.

Table 3.

MiRNAs with a fold change of $>$ 2.0 and a p-value 0.05 of G2 and G3 dedifferentiated liposarcoma patients compared to healthy donors

Transcript ID	Fold change	Fold change	Fold change
	G2 liposarcoma patients ($n=$ 3) vs	G3 liposarcoma patients ($n=$ 3) vs	G2 and G3 liposarcoma patients ($n=$ 6)
	healthy controls ($n=$ 4)	healthy controls ($n=$ 4)	vs healthy controls ($n=$ 4)
hsa-miR-3613-3p	2.5641	3.0037	2.7752
hsa-miR-4668-5p	2.9258	3.0309	2.9779

Figure 2.

MiR-3613-3p-expression in whole blood of patients with G2 dedifferentiated liposarcoma and G3 dedifferentiated liposarcoma compared to healthy donors.

For further validation of the array, qRT-PCR was performed to confirm an upregulation of miR-3613-3p in the dedifferentiated liposarcoma samples ($n=$ 6) compared to an independent cohort of healthy controls ($n=$ 3) as well as compared to lipoma patients ($n=$ 5; Fig. 3). Interestingly, higher miR-3613-3p-values were recorded in patients with localized disease compared to patients with metastatic liposarcoma (Supplement Figs 1 and 2; expression of miR-3613-3p in localized liposarcoma patients marked in blue; expression of miR-3613-3p in metastatic liposarcoma patients marked in red).

Figure 3.

MiR-3613-3p-expression in whole blood of patients with G2 dedifferentiated liposarcoma and G3 dedifferentiated liposarcoma compared to an independent cohort of healthy donors and lipoma patients.

4. Discussion

Biomarkers for liposarcoma are urgently needed to distinguish benign tumors such as lipoma (Fig. 4A and B) from liposarcoma (Fig. 4C and D), as well as identify patients with local recurrence and metastatic disease. Although several radiological predictors differentiating malignant liposarcoma from benign lipoma have been established [15], imaging studies are often inconclusive, since both lipoma and liposarcoma can present as well-circumscribed tumors, demonstrating large lobulated components of fat-signal (Fig. 4). In this context, it has been shown that lipoma cannot always be accurately differentiated from liposarcoma using imaging studies [16, 17]. Coran et al. examined MRI findings of 54 patients with lipomatous tumors, judging accuracy in distinguishing between lipoma and malignant lipomatous tumors according to size, localization, septa, nodules and signal homogeneity. Hereby, the authors found that although there are quite a few generally known signs that can indicate malignancy of the lesion such as tumor size $>$ 5 cm, deep-sited lesions, inhomogeneous signal (especially in T2-weighted sequences), septa $>$ 2 mm, contrast-enhancement, nodules and relationship with the surrounding structures, only the presence of septa $>$ 2 mm and inhomogeneity were significantly correlated with malignancy [18]. Thornhill et al. examined the use of computer-assisted analysis (CAD) in differentiating between lipoma and liposarcoma by magnetic resonance imaging and concluded that while even experienced radiologists only manage to accurately differentiate between lipoma and liposarcoma in 77–80% of cases, CAD shows an accuracy of 91% [17]. This underlines the fact that although there are quite a few radiological criteria that differentiate lipoma from liposarcoma, imaging studies might not always result in an accurate diagnosis. Furthermore, the authors state that when imaging is inconclusive and image-guided core needle biopsy is performed, the biopsy generally targets the most nonadipose component of the lesion; thus, in soft tissue tumors, biopsy can be prone to sampling errors [17, 19]. Moreover, Robertson and Baxter argue that tumor biopsies can potentially lead to local tumor spread [20]. Therefore, “liquid biopsy” markers such as miR-3613-3p are urgently needed in order to improve early diagnosis of liposarcoma and prevent unnecessary biopsies.

Figure 4.

Magnetic resonance imaging of a 62-year male old patient with a lipoma of the right medial compartment of the thigh (A, coronal; B, axial) and an 87-year old female patient with a dedifferentiated liposarcoma (G3) of the ventromedial thigh (C, coronal; D, axial; T1-weighted images). Both tumors are well-circumscribed, demonstrating large lobulated components of fat-signal.

The aim of our study was to establish a miRNA biomarker for dedifferentiated liposarcoma which could easily be transferred into clinical practice: Using PAXgene Blood RNA Tubes for blood sample collection, samples can be stored at room temperature for up to 72 hours before being further processed, which might facilitate multi-centre studies. On the contrary, for serum or plasma miRNA profiling, immediate processing of the samples is necessary, which might lead to potential changes in miRNA levels by different processing conditions [21, 22].

Our results showed that the detection of miR-3613-3p in whole blood of dedifferentiated liposarcoma patients may serve as a ,,liquid biopsy“, a non-invasive way to differentiate between lipoma and dedifferentiated liposarcoma as well as detect distant metastasis or local recurrence of dedifferentiated liposarcoma. Interestingly, miR-3613-3p-values were higher in patients with localized disease compared to patients with metastatic liposarcoma (Supplement Figs 1 and 2). Nevertheless, the fact that localized liposarcomas yield higher miR-3613-3p-values than metastatic disease warrants further studies.

MiR-3613-3p has been described to play an important role in the histological classification and TNM staging of lung adenocarcinoma [23] and to be potentially involved in several important biological processes and functional pathways such as Wnt and Notch signaling [24]. Furthermore, bioinformatic analysis using the miRDB, Target Scan and TargetMiner databases found DFFB, APAF1, DICER, RORA, KIF3A, MAP3K1, NF1 and VHL to be possible mRNA targets binding miR-3613-3p prevalently in their 3’UTRs; however, performing qRT-PCR and western blot analyses, only target genes of the apoptotic machinery (APAF1, DFFB) were proven to be direct targets of miR-3613-3p [25].

Interestingly, the recently discovered MCPIP1 (monocyte chemoattractant protein-induced protein 1), a multidomain protein encoded by the MCPIP1 (ZC3H12A) gene which has been described to be downregulated in primary neuroblastoma, has been postulated to repress the expression of miR-3613-3p by cleavage of its precursor form [25]. Furthermore, MCPIP1 has been shown to play an important role in the immune system, being induced by inflammation-related factors, such as MCP-1, TNF$\alpha$, LPS and IL-1ß [26, 27] and acting as a negative regulator for the NF-kB signalling pathway, thus negatively regulating the cell’s response to the inflammatory state [28]. Moreover, MCPIP1 was shown to be a key regulator of adipogenesis, preventing the formation of new adipocytes when upregulated in preadipocytes [29], as well as of apoptosis, activating caspase-3 and inducing apoptotic gene families before causing cell death [26]. Thus, by implication, an upregulation of miR-3613-3p in the peripheral blood of liposarcoma patients, which has been found to inversely correlate with the expression of MCPIP1 [25], might reflect an amplification of the cell’s response to the inflammatory state as well as an impairment of apoptosis in tumor cells. Furthermore, through the interaction with MCPIP1, miR-3613-3p might be involved in the regulation of adipogenesis and thus possibly in the tumorigenesis of liposarcoma; however, the exact mechanisms remain to be elucidated.

MiR-4668-5p was shown to be upregulated in dedifferentiated liposarcoma samples as compared to heal-thy donors throughout the array; however, this miRNA was not validated to be upregulated by qRT-PCR. Nevertheless, miR-4668-5p was only shown to be of importance in IgA nephropathy and tuberculosis [30, 31], not being linked to tumorigenesis, apoptosis or tumor-immune system interactions.

Contradictory findings of miRNA profiles in the peripheral blood and tumor tissue have been shown throughout previous studies [32, 33, 34]. In this context, Fehlmann et al., who identified several miRNAs which are rather specific for certain tissues or body fluids, also argued that miRNAs which are highly expressed in solid tissues are not necessarily found in the peripheral blood [35]. Interestingly, Mookherjee and El-Gabalawy found that whole blood miRNA profiles reflect miRNA expression patterns of peripheral blood mononuclear cells (PBMC) [36]. As previously hypothesized by the authors [21], these findings point out that the miRNA signature detected in whole blood RNA constitutes a response of the innate immune system to the tumor rather than the miRNA expression profile of the tumor itself. Therefore, it is not surprising that when analyzing whole blood RNA, we did not find an upregulation of miRNAs which have been previously been described to be overexpressed in liposarcoma tissue [9, 37, 38]. Correspondingly, Keller et al. found that only a few of the miRNAs deregulated in whole blood RNA were also previously described to be deregulated in solid tissues derived from patients with the same diseases [21, 39].

However, whole blood miRNA expression might be influenced by differing quantities of circulating blood cells. Analysis of age, BMI, Hb level, platelet count and leukocyte count was not statistically different when comparing dedifferentiated liposarcoma patients to the healthy donors evaluated in the microarray (Table 1), but G2 and G3 liposarcoma patients were significantly older than the independent cohort of healthy controls ($p=$ 0.0181 and $p=$ 0.0021, respectively) and G3 liposarcoma patients were significantly older than lipoma patients ($p=$ 0.0220; Table 4). Furthermore, Hb levels of G2 liposarcoma patients were significantly lower than Hb levels of lipoma patients ($p=$ 0.0323; Table 4), which represents a limitation of our study.

Table 4.

Demographic patient data (age, BMI) and blood count of patients with dedifferentiated liposarcoma, healthy donors (Individual Cohort (IC)) and lipoma patients

Patients	Age (years)	BMI	Hb	Platelets	Leukocytes
		(kg/m${}^2$)	(g/dl)	($\times$ 10${}^8$/l)	($\times$ 10${}^8$/l)
G2 liposarcoma patients ($n=$ 3)	64.33 $\pm$ 4.41	24.91 $\pm$ 3.08	11.10 $\pm$ 1.75	272.00 $\pm$ 31.19	5.52 $\pm$ 0.97
G3 liposarcoma patients ($n=$ 3)	80.33 $\pm$ 4.41	28.05 $\pm$ 2.10	13.73 $\pm$ 1.94	244.00 $\pm$ 45.36	8.75 $\pm$ 3.26
G2 and G3 liposarcoma patients ($n=$ 6)	72.50 $\pm$ 4.65	26.48 $\pm$ 1.81	12.42 $\pm$ 1.31	258.00 $\pm$ 25.40	7.13 $\pm$ 1.68
Healthy controls (IC) ($n=$ 3)	45.33 $\pm$ 2.19	29.82 $\pm$ 3.11	14.53 $\pm$ 0.38	272.70 $\pm$ 39.46	6.06 $\pm$ 0.76
Lipoma patients ($n=$ 5)	54.80 $\pm$ 5.76	25.17 $\pm$ 2.15	14.72 $\pm$ 0.21	225.60 $\pm$ 19.98	5.87 $\pm$ 0.39
p-value of G2 liposarcoma patients vs.	0.0181	0.3249	0.1274	0.9901	0.6853
healthy Controls (IC)
p-value of G3 liposarcoma patients vs.	0.0021	0.6620	0.7066	0.6584	0.4673
healthy controls (IC)
p-value of G2 liposarcoma patients vs.	0.2957	0.9465	0.0323	0.2338	0.7039
lipoma patients
p-value of G3 liposarcoma patients vs.	0.0220	0.4100	0.5199	0.6802	0.2834
lipoma patients

Data are presented as mean value $\pm$ standard error of mean (SEM). Age, BMI, Hb-, leukocyte- and platelet-values were rounded to 2 decimal digits; while p-values were rounded to 4 decimal digits. p-values were determined using a Student’s t-test for independent samples. Hb $=$ Hemoglobin. BMI $=$ Body Mass Index. IC $=$ Independent Cohort.

In summary, although the biological significance of miR-3613-3p in the liposarcoma-immune system interaction remains to be further clarified, the detected upregulation of miR-3613-3p might constitute a sensitive biomarker to distinguish dedifferentiated liposarcoma patients from lipoma patients.

Nevertheless, a further limitation of this study constitutes its low number of patients. However, one has to consider that liposarcoma is a rare disease; furthermore, dedifferentiated liposarcoma constitutes one of the less frequent subtypes of liposarcoma, therefore making the collection of a large number of samples challenging. Thus, further studies are warranted to validate the detected miRNA deregulation in a larger cohort of independent dedifferentiated liposarcoma patients.

5. Conclusion

MiRNA-3613-3p may serve as an independent bio-marker for the diagnosis of dedifferentiated liposarcoma, distinguishing dedifferentiated liposarcoma from healthy controls as well as from lipoma patients.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-170496.

Conflict of interest

The authors declare that they have no competing interests.

List of abbreviations

BMI	Body Mass Index
CT	Computed Tomography
Hb	Hemoglobin
MiRNA	MicroRNA
MRI	Magnetic Resonance Imaging