Background
Liposarcoma is a heterogeneous group of malignant mesenchymal neoplasms with varying degrees of atypia. High-grade malignant liposarcomas, like pleomorphic liposarcoma (PLPS), DDLPS and high grade myxoid liposarcoma, have a high rate of recurrence and metastasis [
1]. Although myxoid liposarcoma respond well to radiotherapy and chemotherapy, the benefit from systemic therapy in PLPS and especially DDLPS is rather limited [
2,
3]. The well-/dedifferentiated subtype of liposarcoma (WD/DDLPS) makes up 50% of all liposarcomas and occurs mostly in two anatomical sites: in the retroperitoneum or in the extremities. In the extremities WDLPS is referred to as atypical lipomatous tumor (ALT), because the prognosis is good. A retroperitoneal/intra-abdominal location is associated with significantly worse outcome, independent of tumor size [
4]. The main treatment consists of surgical resection with negative margins for resectable disease, which is however difficult to obtain in the retroperitoneum.
Progression towards the DD form occurs in 17% of patients when WDLPS is located in the retroperitoneum and in 4% of cases when WDLPS is located in the extremities [
5]. DDLPS is reported to metastasize at a rate between 13 and 47%, and metastases are fatal, therefore DDLPS gives a sixfold higher risk of death compared to WDLPS [
6,
7].
Tools for differential diagnosis of WD versus DDLPS are radiologic examination and macroscopic and histological evaluation by a pathologist. Although the amplification of the 12q13–15 region, carrying the
MDM2 and
CDK4 genes, is used to distinguish WD/DDLPS and ALT from benign lipomas and from other types of liposarcomas, specific aberrations that can be used to distinguish between WDLPS and DDLPS subtypes have not been identified so far. Computed tomography (CT) can identify non-lipomatous area in a lipomatous tumor but does not have the resolution to reveal ongoing dedifferentiation processes within adipose-like tissue or to distinguish dedifferentiated parts of the tumor from stroma components, while PET/CT has no routine role for diagnosis [
8].
The main treatment for WD/DDLPS is surgery. The outcome depends on complete surgical resection as well as tumor location and histological subtype. Surgical outcomes are poor for patients with rapidly growing or incompletely (R1) resected tumors, in particular in the retroperitoneum. Wide surgical margins are recommended, but complete (R0) resection is particularly important for DDLPS tumors whenever possible, even at the cost of contiguous organ resection. However, because of the highly invasive nature of this surgical procedure, the post-operatory morbidity and mortality can be an issue and there is no consensus among surgeons on the best surgical strategy for WDLPS [
9‐
12]. Thus, molecular biomarkers able to accurately predict the presence of DDLPS as early as possible, would be of great value to guide the aggressiveness of the surgery.
Here we tested a metabolic gene signature as a biomarker for the differential diagnosis of ALT/WD- and DDLPS, as well as for its ability to predict malignant evolution towards the DD form. We found that this signature allowed the accurate identification of DDLPS among the analyzed samples, even in those derived from the WD part of DD tumor.
Methods
Patient material
Tumor material was collected after surgery from patients that entered the clinic between 2014 and 2017. Patients were diagnosed with ALT, WDLPS or DDLPS according to the current World Health Organization classification. The definition of an ALT or WDLPS was findings of a mature lipomatous tumor with some atypical lipoblasts with nuclear atypia and cytoplasmic multi-vacuolization and/or fibrous areas with atypical spindle cells. For DDLPS it was a biphasic appearance, where one component is WD and another is non lipomatous area. In selected cases, to differentiate from lipomas and other sarcomas, analyses of MDM2 amplification were performed. The grading is based on the French system evaluating the mitotic index, the differentiation of the cells and the amount of necrosis. The cases were reviewed for diagnosis, grade, size, location and MDM2 status if analyzed. The project (S-06133) was approved by the Regional Ethical Committee for Southern Norway, and patient participation was confirmed by written informed consent. The 6 DDLPS specimens received from the Leiden University Medical Center (LUMC) were anonymized and handled according to the ethical guidelines described in the Code for Proper Secondary Use of Human Tissue in the Netherlands of the Dutch Federation of Medical Scientific Societies, as well as our Norwegian approval. The control group included an anonymous pool of human adipose tissue (pooled total RNA from 18 individuals of different ages and genders; Clontech, cat N. 636558), healthy adipose tissue (chest) sample from one anonymous random non-sarcoma patient, and 8 lipoma (benign adipose tumor) patient samples. All tissues were fresh frozen at − 80 °C following surgery. Histological classifications were confirmed by a sarcoma reference pathologist.
RNA isolation and quantitative reverse transcription polymerase chain reaction (RT-qPCR)
Total RNA was isolated using Allprep DNA/RNA/miRNA Kit (Qiagen, cat No. 80224). In four of the 20 DDLPS samples, total RNA was extracted from the WD component. The cDNA synthesis and the RT-qPCR were performed using TaqMan gene expression Assay (Applied Biosystems). For the normalization, two house-keeping genes were initially tested, B2M and TUBA1A. Both genes gave similar relative expression level of target genes. Subsequently, for the space saving, internal control genes were reduced to only B2M, which had lower Ct values. All primers were purchased from Applied Biosystems.
Statistical analyses
Receiver operator characteristics (ROC) analyses were employed to evaluate the ability of univariate and multivariate models to correctly classify samples in the discovery cohort as either WD- or DDLPS, as defined by the pathologist. For each gene, a cut-off value was set as the mean of the two least different ΔCt values between the WD- and DDLPS groups in the discovery cohort.
Multivariate models were created using binary logistic regression with differential diagnosis as binary outcome variable (WDLPS vs. DDLPS), and the generated predictive probabilities were evaluated by ROC analysis in R (v3.4.3). Hierarchical clustering heatmaps were generated in R, using the “heatmap.2” package with Euclidean measure for distance matrix and complete agglomeration method for clustering. Gene ΔCt values were scale normalized (mean = 0 and standard deviation = 1) yielding Z-scores in R, and hierarchical clustering was applied patient-wise. Wilcoxon rank sum tests were applied to evaluate the statistical significance comparing ΔCt values from samples with differential histology. χ2 and Student’s t-tests were applied evaluate differences in distribution of clinicopathological data. A two-sided P value < 0.05 was considered statistically significant.
Next generation sequencing data analysis
High quality DNA was isolated using the Promega Wizard Genomic DNA Purification Kit (Promega, Wisconsin, United States) and the QIAamp DNA FFPE Tissue kit (Qiagen, Venlo, Netherlands) as previously described. One microgram of genomic DNA was used to produce exome-captured sequencing libraries using the Agilent SureSelect Human All Exon v5 kit (Agilent Technologies, California, United States). Paired-end 100-bp sequencing of each exome capture library was done using an Illumina HiSeq 2500 instrument and Illumina’s TruSeq SBS v3 chemistry (Illumina, California, United States).
Reads from tumor and matched normal blood sample were aligned separately to the human NCBI Build GRCh37 reference genome using Novoalign (Novocraft Technologies, Selangor, Malaysia) with default parameters. PCR duplicates, improper pairs and ambiguously mapped reads were removed using in-house scripts. SNVs were called using MuTect [
13,
14]. Variants annotation was done using Oncotator.
In silico publicly available datasets
Scale-normalized PNPLA2 expression levels, copy number variation (CNV) data and clinical parameters from DDLPS cases within the Adult Soft Tissues Sarcoma were downloaded from cBioPortal (refs 23550210 and DOI:
https://doi.org/10.1158/2159-8290.cd-12-0095) and visualized using GraphPad Prism 5. Statistical analysis (student’s t-test) on medians and variances was performed in GraphPad Prism 5 [
15].
Discussion
Here we report that the combined expression level of three adipose tissue-specific metabolic genes, namely PNPLA2, LIPE and PLIN1 (the 3M signature), could accurately distinguish WDLPS/ALT from DDLPS, and we verified this finding in an independent validation cohort.
Although it may seem obvious that the loss or down-regulation of the expression of tissue-specific genes merely reflects the dedifferentiation process, a growing body of evidence strongly suggests that this is not the case. Recently, Wu and collaborators demonstrated that double knockouts of
PNPLA2 and
LIPE in mice, but not their single knockouts, down-regulated metabolic genes, including those involved in fatty acid and lipid metabolism, and the mice spontaneously developed liposarcoma in brown adipose tissue [
19]. Interestingly, the molecular markers of WD/DDLPS,
MDM2 and
CDK4, were highly expressed in the adipose tissue of double knockout mice.
Another indication of the direct involvement of
PNPLA2 in liposarcoma pathogenesis comes from a recent study where global genome characterization of soft tissue sarcomas was performed [
15]. In this study, the 11p15.5 region, carrying
PNPLA2 gene, was recurrently deleted in DDLPS samples. The closer look at the TCGA DDLPS datasets showed that a “shallow deletion” of
PNPLA2 corresponded to the significant downregulation of its transcript level, compared to the specimens with normal copy number. The recurrent deletion of the 11p15.5 region in DDLPS was confirmed by a very recent study by Beird et al. [
20], comparing CNV profiles of DDLPS specimens with the matched WDLPS specimens. Strikingly, the deletion of the 11p15.5 region was detected only in DDLPS specimens, but not in the WDLPS specimens. Interestingly, when we tested
PNPLA2 expression in WD and DD components of the same tumor, the expression was lost in both. It seems likely that
PNPLA2 loss precedes the loss of the WD phenotype, and that the deletion of
PNPLA2 may be an important event in the transition of WDLPS towards DDLPS. Beird et al. also found a low fraction of mutations shared between the paired WD and DDLPS subclones, indicative of the development of the propensity to dedifferentiate as an early process [
20].
However, although the deletion of PNPLA2 gene or surrounding regions may be a mechanism of gene expression loss in some samples, the absence of such a deletion in other samples suggests the existence of alternative mechanisms, like hypermethylation or activation of transcriptional repressors.
In another study by Lyu et al., knock-out of
PLIN1 in mice caused down-regulation of adipogenic pathways despite the near normal level of PPARγ [
21]. Strikingly, Horvai and collaborators reported that the PPARγ protein can be detected in the vast majority of dedifferentiated liposarcomas, with specific nuclear staining in 93% of DDLPS tested [
22]. Altogether, these data demonstrate that the loss of these adipose tissue-specific metabolic genes triggers the downregulation of other specific metabolic genes independently of the master regulator of adipose differentiation PPARγ. In this respect, the loss of
PNPLA2 expression may be a primary event in the DDLPS pathogenesis.
The histological detection of the transformation to DDLPS is dependent on the visual observation of perhaps a minor focal DD component which could easily be missed, especially since these tumors are usually very large. Importantly, the 3M signature appears to be able to provide correct diagnosis of tumors with progression to DDLPS even when samples from the phenotypically well-differentiated parts were investigated. At the same time, in those samples, the higher expression of the lipid droplet coating protein
PLIN1 was consistent with the presence of lipid droplets, showing that the decrease and loss of PNPLA2 and LIPE expression was malignancy-specific and not connected to the simple loss of fat part. This makes the 3M signature useful for diagnostic biopsies, where small samples are randomly collected, although it remains to be seen how homogeneously the early loss of this signature is distributed in dedifferentiating tumors. Moreover, it has been shown that the response of DDLPS to chemotherapy is underestimated by current analytical tools [
23]. This implies that the earliest detection possible of the DD component would be important for pharmacological patient management.
Lipid storage and release from adipose tissue is highly coordinated and dependent on several metabolic enzymes, characteristic of mature adipocytes [
24]. PNPLA2 hydrolyzes triacylglycerols, while LIPE hydrolyzes diacylglycerols, both acting coordinately within the lipolytic cascade. When these two lipases were inactivated, lipolysis is almost completely suppressed. Because ALT, WD- and DDLPS all have marker chromosomes with multiple amplified segments and share the same amplification of
MDM2 in 12q13–15, their etiology appears to be closely related, pointing to a common origin. However, the mechanism underlying the transition from WD- to DDLPS is unknown, although Beird et al. identify the frequent loss of the let-7-binding part of amplified HMGA2 transcripts in DDLPS to be a possible candidate [
20,
25]. The new finding reported by Wu et al. and our data together indicate that the pathogenic mechanism may involve PNPLA2 and LIPE lipases. Importantly, LIPE is a major retinyl ester hydrolase (REH) in white adipose tissue (WAT). REH activity of LIPE-null mice was abolished and accompanied by increased levels of retinyl esters and decreased levels of retinol, retinaldehyde and all-trans RA [
26]. Also, the differentiation of WAT in LIPE-null mice was suppressed [
26]. This is a very relevant aspect for the possible implication of LIPE in the pathogenic mechanism, as retinoic acid (RA) pathway has well-established tumor suppressor function and its loss contributed to the loss of differentiation in WAT. Similarly, a recent review considers non-energetic tumor-suppressive functions of PNPLA2 in cancer [
27].
In addition to the 3M signature, another metabolic gene, SCD1, was strongly-down-regulated in DDLPS. However, SCD1 displayed more variation across lipoma and WDLPS samples, and for this reason it was not included in the final diagnostic panel. Yet, in a validation phase on a bigger cohort of patients, if necessary, SCD1 may be used together with the 3M panel to reinforce the differential diagnosis. Although still not validated in a larger cohort, the 3M signature would be expected to be valuable in the diagnostic work-up of WD/DDLPS tumors, together with the routinely used histological analysis and detection of amplified MDM2 by FISH, PCR or immunohistochemistry.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.