ELSEVIER
Surgery
journal homepage: www.elsevier.com/locate/surg
SURGERY
MWEMBER 2018
=====
$
A
=
== 2
Identification of novel lipid metabolic biomarkers associated with poor adrenocortical carcinoma prognosis using integrated bioinformatics
Chitra Subramanian, PhD, MBAª, Mark S. Cohen, MD, FACS, FSSOa,b,*
a Department of Surgery, Michigan Medicine, Ann Arbor, MI
b Departments of Pharmacology and Biomedical Engineering, University of Michigan, Ann Arbor, MI
Check for updates
ARTICLE INFO
Article history: Accepted 9 April 2021 Available online 2 August 2021
ABSTRACT
Background: Adrenocortical carcinoma while rare, often presents with advanced metastatic disease carrying a 5-year survival of <15%. Despite adrenocortical carcinoma tumors having high avidity for cholesterol, the role of lipids in adrenocortical carcinoma has not been well described. Therefore, we performed an integrated bioinformatic analysis to identify novel lipid biomarkers correlating with poor survival that may help identify adrenocortical carcinoma tumor progression or therapy resistance. Methods: A meta-analysis of collated adrenocortical carcinoma studies from the correlation engine identified lipid metabolism genes differentially expressed between adrenocortical carcinoma and the normal adrenal, which were then selected for enrichment analysis by the Database for Annotation, Visualization and Integrated Discovery database. A protein-protein interaction network of genes was constructed using Search Tool for the Retrieval of Interacting Genes/Proteins and Cytoscape. Top hub genes identified were validated using the Xena database. Survival analysis of hub genes was performed in the R2 genomic analysis platform using The Cancer Genome Atlas program data set.
Results: Examination of pathways by correlation engine identified a unique subset of lipid metabolism- related genes that are differentially regulated in adrenocortical carcinoma tumors versus normal tissues (P < . 01). Enrichment pathway analysis in Database for Annotation, Visualization and Integrated Dis- covery indicated that genes involved in sphingolipid, steroid, and peroxisome proliferator-activated re- ceptor-o. metabolism is upregulated in adrenocortical carcinoma, whereas glycerol phospholipid, fatty acid, and phosphatidylinositol metabolism are downregulated. Survival analysis of differentially regu- lated genes indicated that upregulation of SGPL1, FDFT1, SQLE and downregulation of PIK3C2B, PIK3CD, SYNJ2, DGAT1, PLA2G16, PLD1, GPD1 are all significantly associated with poor overall survival (P < . 05) in adrenocortical carcinoma patients.
Conclusion: Upregulation of sphingolipid and steroid synthesis genes and downregulation of phospha- tidylinositol and glycerol phospholipid metabolism are associated with worse survival in patients with adrenocortical carcinoma.
@ 2021 Elsevier Inc. All rights reserved.
Introduction
Adrenocortical carcinoma (ACC) is a rare endocrine neoplasm of the adrenal cortex with a poor long-term prognosis and limited
therapeutic options.1 This aggressive cancer has a global annual incidence of 0.5 to 2 cases per million.2 The clinical presentation of ACC typically involves 3 different forms. Approximately 60% of patients with ACC present with symptoms relating to hormone excess. Another 20% present symptoms of abdominal pain or fullness owing to tumor growth, and the remaining 20% of ACC patients are identified by abdominal imaging owing to unrelated medical indications.3-5 As the clinical manifestations of ACC are often difficult to determine, most patients are diagnosed at an advanced metastatic stage of the disease. Although ACC patients with the locoregional disease are treated with surgery, approxi- mately 75% of patients develop distant metastasis and recurrence
Accepted for Presentation at the 41st Annual Meeting of the American Asso- ciation of Endocrine Surgeons April 25-27, 2021.
* Reprint requests: Mark S. Cohen, MD, FACS, FSSO, Department of Surgery, University of Michigan Hospital and Health Systems, 2920K Taubman Center, SPC 5331, 1500 East Medical Center Dr, Ann Arbor, Michigan 48109-5331.
E-mail address: cohenmar@med.umich.edu (M.S. Cohen);
Twitter: @MarkCohenFACS
after treatment.6 Currently, the first line of therapy for patients with advanced disease involves systemic therapy with mitotane (M), either alone or in combination with multidrug chemotherapy (Etoposide, Doxorubicin, and cisPlatin [EDP]) known as the Italian protocol. The second line of treatment after progression following the first line of therapy with EDP regimen plus mitotane involves streptozotocin ± mitotane or gemcitabine + 5-fluorouracil or capecitabine.7 Despite these treatments being effective in pro- longing the progression-free survival, disease progression and drug-related toxicities are common. For advanced, inoperable metastatic disease, median overall survival is poor at 12 to 15 months with a 5-year overall survival of <15%.3,8,9 Given these poor prognoses for most patients, there is a great need for the identification of novel biomarkers in ACC that carry prognostic and diagnostic value as well as future opportunities for the development of novel therapeutics for this disease.
Lipid metabolic reprogramming of cancer cells is emerging as one of the hallmarks of cancer.10 Lipids not only function as structural components of cellular membranes, but are also involved in energy storage (through oxidation of B-fatty acids), control of redox homeostasis, and as signaling molecules for modulating several cancer processes.11 Altered lipid metabolism affects a myriad of cellular processes in cancer including proliferation, angiogenesis, migration, invasion, transformation, tumor micro- environment reshaping, and inflammation.11,12 Additionally, cancer cells have an altered lipid profile and lipid membrane composition compared with normal cells.1º Furthermore, enzymes involved in lipid metabolism are expressed differentially depending on the tumor type, stage, and grade. As such, recent studies have identified key lipid metabolism-related genes that are diagnostic and prog- nostic biomarkers of tumor progression and therapy resistance for several cancers.1º Evaluation of lipid metabolic dysregulation in pan-cancer using omics data revealed that overexpression of genes involved in fatty acid metabolism (FADS1, FADS2, FASN, and ACOT7), arachidoic acid metabolism (GPX8, PTGES3, and PTGIS), cholesterol mechanism (SQLE, VDAC1, CD36, LDLR, LRP1, and VAPA), peroxisome proliferator-activated receptor (PPAR) signaling (MMP1, OLR1) as well as downregulation of genes, such as CYP4A11, PLA2G4A, PLA2G3, LTC4S, CYP27A1, HMGCS2, and PDPK1 are associated with poor overall survival of patients in many can- cers.13 Therefore, drugs targeting some of these lipid markers including FASN, PPAR, CPTI, PTGS2, DGAT, ACC, and ACLY are already in clinical trials.11 Although many cancer cells including ACC cells are known to have high avidity for cholesterol and lipid drop- lets14,15 in them, the role of these lipids and lipid metabolic bio- markers in ACC tumorigenesis and prognosis have not been well characterized to date.
High throughput gene expression profiling of microarray and RNA sequencing data from patient with ACC samples has enabled access for the first time to a large data set from different cohorts of patients with ACC. Although these studies have identified several key protein-signaling pathways in ACCs, the role lipids or lipid biomarkers play in ACC pathogenesis and therapy response is not yet defined. Therefore, in the present study, we performed an integrated bioinformatic analysis of 3 ACC Gene Expression Omnibus data sets downloaded from the BaseSpace Correlation Engine. Enrichment analysis, protein-protein network, validation of hub genes, and the survival analysis were carried out using Database for Annotation, Visualization and Integrated Discovery (DAVID), STRING, Xena, and R2 data bases. Our hypothesis is that there are several novel lipid biomarkers in ACC that significantly correlate with disease prognosis that can be used in the future to identify more aggressive tumor variants or tumor progression risk.
Methods
Data acquisition
The BaseSpace Correlation engine is an online searchable data- base having raw experimental data from high throughput gene expression analysis deposited in global repositories such as Gene Expression Omnibus and Array Express. Studies related to ACC were selected using the keyword “cancer of the adrenal cortex” for the search engine and “Homo sapiens” for the organism. To define the data sets for the current study, inclusion criteria of primary ACC tumors versus normal adrenal; and exclusion criteria like child- hood ACC, adenoma, and drug treatment, were applied to the 108 studies generated by the search engine. Defined filtration criteria resulted in the identification of data sets such as GSE19750 (44 ACC and 4 normal tissue); GSE10927 and GSE33371 (33 ACC, and 10 normal adrenal); and GSE12368 (12 ACC and 6 normal adrenal) for the analysis of lipid metabolism-related genes in ACC.
Identification of differentially expressed genes
A meta-analysis of the collection of individual bio sets in cor- relation engine and canonical pathway filter identified the highly correlated lipid metabolic genes in each of the 3 bio sets. The RNAseq datasets were subjected to statistical threshold of a mini- mum of 1.2-fold change and a maximum adjusted P value of .05 in correlation engine. A proprietary (Illumina) specialized gene set enrichment algorithm (running on Fisher) then converts the experimental data sets to statistically significant differentially expressed genes between ACC and normal adrenal tissue with fold change values.
Enrichment Analysis of lipid pathways
Differentially expressed genes were subjected to functional annotation and gene enrichment analysis to extract the major metabolic and signaling pathways in DAVID functional bio- informatic tool (https://david.ncifcrf.gov/summary.jsp). The DAVID functional clustering annotation tool enables enrichment analysis of genes in the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases to measure and rank the most relevant biological function and pathways that are associated with a given set of genes. Based on the overall Expression Analysis Systematic Explorer scores of all enriched annotations, the enrichment score (ES) ranks the biological significance of the given data set of genes. It first runs a user’s gene list against the DAVID functional annotation chart to calculate the P value for each enriched annotations. Next, the geometric mean of Expression Analysis Systematic Explorer score (modified Fisher exact test) for the annotations involved in the given data set of genes was calcu- lated. ES score represents the relative importance of a given set of genes in the enriched biological pathways. The higher the ES score, the more enriched are the biological pathways in the given set of genes.
Protein-protein network and the identification and validation of hub genes
STRING database uses the collection of publicly available infor- mation and a computer algorithm to calculate protein-protein in- teractions. Therefore, we used the STRING database (https://string- db.org) to construct a protein-protein interaction network of differentially regulated genes to identify the connection between proteins involved in the enriched biological function. The
B
PLA2G1B-
FDFT1-
SLC44A5-
SQLE
NCOA3-
ACADL-
PRKAB2-
GSE12368
GSE10927
NCOA2-
SRD5A1-
0
PRKACA-
NR1D1-
SGPLI-
MED13L-
CRLS1-
ETNK2-
14
26
27
CDK19-
GM2A-
ARSK-
COL4A3BP-
140
CDK8-
-10
CYP11B1-
HSD3B2-
PLA2G4A-
25
53
CYP11B2-
APOE-
CYP17A1-
HSD11B1-
CTGF-
PLTP-
93
ALASI-
ABCG1-
ABCC3-
-20
PLIN2-
PLPP3-
TM7SF2-
GSE19750
SREBF1-
CERK-
ACOX2-
PLPP1-
CPT1A-
ACADVL
PLA2G16-
DGAT1-
-30
PLD1-
GSE19750
GSE12368
GSE10927
A
B
Upregulated Lipid Genes
Downregulated Lipid Genes
Fatty acid beta-oxidation
Phosphatidic acid biosynthetic process
Phosphatidylinositol biosynthetic process
Sphingolipid biosynthetic process
Glucocorticoid biosynthetic process
Phosphatidylinositol biosynthetic process
Cholesterol metabolic process
Phosphatidylglycerol acyl-chain remodeling
Phosphatidic acid biosynthetic process
Phospholipid metabolic process
Intracellular receptor signaling pathway
Phosphatidylcholine biosynthetic process
Phospholipid biosynthetic process
Triglyceride biosynthetic process
CDP-diacylglycerol biosynthetic process
Phosphatidylinositol phosphorylation
Phosphatidylethanolamine biosynthetic process
Carnitine shuttle
Phosphatidylinositol acyl-chain remodeling
C21-steroid hormone biosynthetic ..
Ceramide metabolic process
Positive regulation of cholesterol efflux
Phosphatidylserine acyl-chain remodeling
Phosphatidylethanolamine acyl-chain ..
Leukotriene biosynthetic process
Phosphatidylcholine acyl-chain ..
Mediator complex
Triglyceride metabolic process
Endoplasmic reticulum membrane
Glycerol-3-phosphate catabolic process
Nucleoplasm
Endoplasmic reticulum membrane
Organelle membrane
Peroxisome
Nucleotide-activated protein kinase complex
Mitochondrion
Transcription coactivator activity
Phosphatidylinositol 3-kinase complex
RNA polymerase II transcription cofactor ..
Cytosol
Nuclear receptor transcription coactivator ..
Intracellular membrane-bounded ..
1-acylglycerol-3-phosphate O-acyltransferase ..
Fatty-acyl-CoA binding
Core promoter sequence-specific DNA binding
Acyl-CoA dehydrogenase activity
Histone acetyltransferase activity
BP
Phospholipase A2 activity
Steroid 11-beta-monooxygenase activity
BP
Phosphotransferase activity
CC
CC
Calcium-dependent phospholipase A2 activity
MF
Glycerol-3-phosphate dehydrogenase ..
Sphingosine-1-phosphate phosphatase ..
MF
Phosphatidylinositol-3,5-bisphosphate 3 -..
Phosphatidylinositol phosphate kinase ..
Phosphatidylinositol-3-phosphatase activity
Phosphatidate phosphatase activity
0
2
4
6
8
10
12
0
2
4
6
8
-Log (P)
-Log (P)
confidence score cutoff of 0.4 was used for constructing the network.16 Cytoscape data integration, analysis and visualization network (https://cytoscape.org) plugin Molecular Detection (MCODE) was used to identify significant modules as well as hub genes.
Survival analysis of hub genes and their validation
Differential expression levels of hub genes in ACC compared with the normal adrenal in The Cancer Genome Atlas (TCGA) pro- gram data set was validated using University of California Santa
| Term | Count | Rich factor (%) | P value | Genes |
|---|---|---|---|---|
| hsa00564: glycerophospholipid metabolism | 26 | 18.4397163 | 3.13E-23 | PLA2G1B, PLA2G5, PLD1, SELENOI, PLD3, AGPAT4, PTDSS2, GPAT4, GPAT3, CEPT1, PLA2G12A, PCYT1A, CHKB, PLA2G4A, GPCPD1, CRLS1, PLA2G16, GPAM, ETNK2, GPD1, PEMT, CHPT1, GPD1L, PLPP3, PLPP1, CDS2 |
| hsa00565: ether lipid metabolism | 13 | 9.21985816 | 1.08E-11 | PLA2G12A, PLA2G1B, PLA2G4A, PLA2G5, PLD1, SELENOI, PLD3, PLA2G16, AGPS, CHPT1, CEPT1, PLPP3, PLPP1 |
| hsa01212: fatty acid metabolism | 12 | 8.5106383 | 4.92E-10 | HADHB, HADHA, ACADVL, CPT1A, ACADL, ELOVL5, ACOX1, TECR, ACADM, HSD17B12, ACAA1, ACAT1 |
| hsa04919: thyroid hormone signaling pathway | 14 | 9.92907801 | 1.06E-07 | NCOA1, NCOA2, CREBBP, NCOA3, PIK3CD, MED16, PIK3CG, PIK3R5, MED13L, MED14, NCOR1, RXRA, SIN3A, PRKACA |
| hsa05231: choline metabolism in cancer | 13 | 9.21985816 | 1.93E-07 | SLC44A5, CHKB, PCYT1A, SLC44A1, PLA2G4A, PIK3CD, GPCPD1, PLD1, PIK3CG, PIK3R5, CHPT1, PLPP3, PLPP1 |
| hsa00562: inositol phosphate metabolism | 11 | 7.80141844 | 4.25E-07 | MTMR1, MTMR3, PTEN, PIP4K2A, PIK3CD, PIP4K2B, SYNJ2, MTMR4, MTMR6, PIK3CG, PIK3C2B |
| hsa00600: sphingolipid metabolism | 9 | 6.38297872 | 1.46E-06 | GALC, ARSA, SGPL1, SPTLC1, CERK, NEU1, PLPP3, SGPP1, PLPP1 |
| hsa00592: alpha-linolenic acid metabolism | 7 | 4.96453901 | 3.88E-06 | PLA2G16, PLA2G12A, ACOX1, PLA2G1B, PLA2G4A, PLA2G5, ACAA1 |
| hsa00140: steroid hormone biosynthesis | 9 | 6.38297872 | 7.54E-06 | HSD11B1, STS, HSD3B2, CYP11B2, SRD5A1, CYP11B1, HSD17B12, HSD17B7, CYP17A1 |
| hsa03320: PPAR signaling pathway | 9 | 6.38297872 | 2.24E-05 | CPT1A, RXRA, ACADL, ACOX2, ACOX1, ACADM, PPARA, ACAA1, PLTP |
| hsa00561: glycerolipid metabolism | 8 | 5.67375887 | 6.95E-05 | DGAT1, GPAM, GPAT4, AKR1B1, GPAT3, PLPP3, PLPP1, AGPAT4 |
| hsa04146: peroxisome | 9 | 6.38297872 | 1.07E-04 | HMGCL, MVK, ACOX2, ACOX1, IDH1, AGPS, PMVK, CROT, ACAA1 |
| hsa04931: insulin resistance | 10 | 7.09219858 | 1.23E-04 | SREBF1, PRKAB2, CPT1A, PTEN, PIK3CD, TRIB3, PPARA, ACACB, PIK3CG, PIK3R5 |
| hsa04071: sphingolipid signaling pathway | 8 | 5.67375887 | 0.00552586 | SGPL1, SPTLC1, PTEN, PIK3CD, PLD1, SGPP1, PIK3CG, PIK3R5 |
Cruz Xena (https://xena.ucsc.edu). Genomic analysis and visuali- zation platform R2 (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi) was used for the evaluation of the potential function of differen- tially expressed lipid metabolic genes in modulating clinical pro- gression and prognosis of ACC. Expression of hub genes that are significantly up- and downregulated in ACC compared with normal adrenal cortex were validated in R2 from the TCGA data set. The overall Kaplan-Meier survival of the key differentially regulated genes, presence of necrosis, and combination of genes with sig- nificant P values (P <. 05) was identified from the R2 platform using the TCGA data set of 79 ACC patient samples. Cox hazard model analysis of TCGA genes was performed in RStudio.
Real-time polymerase chain reaction analysis for the validation of hub genes
RNA was prepared from validated adrenocortical cell line NCI- H295R and normal fibroblast cell line grown in 2-dimensional culture in appropriate growth medium using Qiagen RNA isola- tion kit (Qiagen Sciences, Hilden, Germany) using the manufac- turer’s protocol. The RNA was reverse transcribed using the Superscript RT kit from Thermo Fisher Scientific (Waltham, MA) and then reverse transcriptase polymerase chain reaction (RT-PCR) analysis of hub genes was performed using gene specific primer sets as published previously in QunatStudio Real-time PCR system (Thermo Fisher Scientific, Waltham, MA). Delta-delta threshold cycle method was used to calculate the relative gene expression levels after normalization with internal glyceraldehyde 3- phosphate dehydrogenase controls. Given this is a retrospective review of de-identified genomic data from ethically approved da- tabases, this research was considered exempt status.
Results
Identification of the metabolism of lipid and lipoprotein genes in ACC by meta-analysis
Meta-analysis in correlation engine using a canonical pathway filter identified a total of 472 genes in ACCs associated with the
metabolism of lipid and lipoproteins. The bio sets GSE19750, GSE12368, GSE10927 had a total of 311 (P = 3.0e-13), 205 (P = 3.7e-9), and 246 (P = 1.0e-12) common lipid related genes respectively between ACCs and normal adrenal gland tissues. Analysis of the intersection of differentially regulated lipid meta- bolism genes among GSE19750, GSE12368, and GSE10927 by Venn diagram (using http://bioinformatics.psb.ugent.be/cgi-bin/liste/ Venn/calculate_venn.htpl) is shown in Fig 1, A. Intersectional analysis demonstrates that 140 lipid metabolism genes are differ- entially expressed in ACC versus normal adrenal in all the 3 data sets. Of these 64 genes are differentially upregulated in ACC while 76 genes are differentially downregulated. The heat map of these differentially regulated lipid metabolic genes in ACC are shown in Fig 1, B.
DAVID functional annotation of differentially regulated lipid genes in KEGG and GO enrichment pathway
To understand the biological pathways and function of differ- entially regulated lipid metabolism genes, we used GO and KEGG enrichment analysis in DAVID. Within the GO enrichment analysis, the top 10 significantly upregulated lipid-mediated biological pathways in ACC with a P < . 05 versus normal adrenal tissues include biosynthetic processes of phosphatidic acid, sphingolipid, phosphatidylinositol, phospholipid, CDP-diacylglycerol, phospha- tidylethanolamine, leukotriene, phosphatidylglycerol, and serine acyl-chain remodeling as well as ceramide metabolic processes (Fig 2, A). Conversely, the top lipid metabolic biological pathways that are uniquely enriched in the downregulated genes include: fatty acid beta-oxidation, cholesterol, sterol, phospholipid metabolic processes, triglyceride, glucocorticoid, and phosphatidylcholine biosynthetic processes, phosphatidylethanolamine acyl-chain remodeling, and lipid homeostasis. Additionally, phosphatidic acid and phosphatidylinositol biosynthetic pathways are also enriched in downregulated lipid genes (Fig 2, B). GO cellular component analysis revealed that both the endoplasmic reticulum and mitochondrion are enriched with these up- and down- regulated differentially expressed genes in ACCs (Fig 2, A and B). Differentially regulated genes enriched in GO Molecular function
MED26
MEDG
MED23
CKS
MED14
MEDIG
CDK 19
RP4KCZA
MTMRA
NCOAS
NED23
MEDSI
PRALKAIZ
CYP17A1
PIPAK 20
RKICD
TRES
NCOAZ
MDIZ
PK BOG
CY PLIAZ
CREMEP
CYFL 184
SYNG
PRK ACA
SINIA
PIKICAS
NCOAG
PIERS
KOR
1601181
TELIXR1
PTEN
CTCF
NEYA
AZM
1011
CTWICZ
TEA04
FDFTI
PW
MK
THISF2
TEAD2
PLTP
NOI
9.C2 TANO
AICAL
ACO2
RORA
CYP32AI
SQLE
DI
ACATI
RXRA
MED19
NOTE
POCA
HACHE
APDE
HADHA
NGOẠI
HSD1787
ACOKI
NOORI
COLAASEP
OPTIA
SHEFFI
ACLY
KADI
MOC 1
ALAGI
HID1 7012
SAIDS AL
AKRIRI
ACARI
ACACA
ACADVL
PPARA
CROT
AG’S
PLINZ
ARNTL
GLOVES
MCCL
CPAM
OGAT 1
AGPATG
MOCS
ELOVLI
SPILCI
ACADL
CIDIL
PPAPZA
SCPPI
CMZA
TECH
CO2
PEMT
PPAPOS
CPOL
PTESS2
CALC
CHILS1
SCPLI
RLAIC 12A
R.DI
NEUE
VAPE
EPTI
PLAZCI G
0KR
GIFTI
PCYT LA
PLAZCIA
PLDS
CEPTI
STS
POOL
ACPAT4
ETNA2
SLC4-44.1
ML SK
CON
ACPAT 9
PLAZCS
PLAZGIN
CHPT1
PLA2G12A
CDK19
PLA2G16
PLA2G4A
MED26
PEMT
MED13L
PLD1
MED16
PTDSS2
MED23
DGAT1
MED31
CDK8
AGPAT6
NCOA3
GPAM
ELOVLS
MED29
MED14
SREBF1
NCOA2
PLIN2
ACACB
ACOX1
CREBBP
ACADVL
ACAA1
MED6
HADHB
ACAT1
SLC25A20
HADHA
ACOX2
involve transcriptional coactivator, oxidoreductase, phospholipase A2, phosphatase and kinase activities (Fig 2, A and B). KEGG pathway enrichment analysis of differentially regulated genes revealed that phopholipid, phosphatidylinositol, PPAR signaling, fatty acid, ether lipid, and steroid metabolism are the top enriched pathways (Table I). Although ACC tumors are known to have high cholesterol, our enrichment analysis revealed downregulation of cholesterol and upregulation of genes involved in steroid meta- bolism. This may be owing to active transport of cholesterol by steroidogenic acute regulatory protein in ACC.
Protein-protein network and hub gene identification
The function of protein-protein interactions in up- and downregulated genes was next analyzed using the STRING
network (Fig 3). Using the MCODE plugin using Cytoscape soft- ware, significant modules were MCODE identified using the se- lection criteria of degree cut-off = 2, node score cut-off = 0.2, K- score = 2, maximum depth = 100, and MCODE scores >5. This generated 5 downregulated gene clusters and 3 up-regulated lipid gene clusters. The highest node of the downregulated gene cluster with 22 nodes and 84 edges was associated with glycerol phospholipid and fatty acid metabolism whereas the highest node of the upregulated gene cluster with 13 nodes and 78 edges was associated with PPAR-gamma metabolism (Fig 4). From this cluster analysis, we identified a novel set of 15 lipid metabolism-related hub genes significantly modulated in ACCs. These include: DGAT1, PLA2G16, PLD1, PEMT, FDFT1, SQLE, COL4BP, SGPL1, SPTLC1, PIP4K2A, SYNJ2, PIK3CD, PIK3C2B, GPDL1, and GPD1.
15
Primary Tumor
Normal Tissue
Log2(norm_count+1)
10
5
0
PIK3C2B
PIK3CD
SYNJ2
DGAT1
PLA2G16
PLD1
GPD1
SGPL1
FDFT1
SQLE
Hub Genes
Validation and correlation of ACC-modulated hub lipid metabolism genes with Kaplan-Meier survival and prognostic analysis
The expression levels of the hub genes identified were further validated in the TCGA data set using the University of California Santa Cruz Xena functional genome explorer. The violin plot of expression of the top 10 hub genes significantly and differentially expressed in adrenal cancer compared with normal adrenal tis- sue (P < . 05) is shown in Fig 5. To further evaluate the role of differentially regulated lipid metabolism genes involved in the overall survival of patients with ACC , an R2 database was used. On correlative analysis with Kaplan-Meier survival data in ACC patients we identified 10 novel lipid metabolism genes that significantly (P < . 05) correlate with poor overall survival in ACC. These include (Fig 6, A) the following differentially upregulated lipid metabolism genes (genes involved in sphingolipid meta- bolism [Sphingosine-1-phosphate Lyase 1; SGPL1] and steroid biosynthetic pathway [FDFT1, SQLE]) as well as the following differentially down-regulated lipid metabolism genes (genes in phosphatidylinositol [PIK3C2B, PIK3CD, SYNJ2] and glycerol phospholipid metabolism [DGAT1, PLA2G16, PLD1, GPD1] path- ways). Clinicopathological parameter necrosis is frequently observed in ACC tumors. Therefore, we next evaluated how presence of necrosis influence the overall survival of patients with ACC owing to the differential regulation of hub lipid meta- bolic genes. In addition to evaluating the overall survival of pa- tients with ACC owing to differential expression of individual genes, we also examined the overall survival of combination of genes. The results from the analysis of TCGA data sets revealed that combination of genes involved in upregulation of sphingo- lipid metabolism (bonf. P = 1.8e-03) as well as downregulation phosphatidylinositol (bonferroni P = . 02) and glycer- ophospholipid metabolism (bonferroni P = . 01) significantly decreased the overall survival of ACC patients (Fig 6, B). Survival analysis of the differentially regulated genes indicated that higher expression levels of SGPL1, and FDFT1 as well as lower expression levels of PLD1, GPD1, DGAT1, PLA2G16, CHPT1, and PIK3C2B significantly reduced the overall survival of ACC patients
(Fig 6, C). Also expression of these lipid genes significantly cor- relates with Ki-67 immunostaining. Although up regulation of SQLE; and downregulation of SYNJ2 and PIK3CD decreased the overall survival (Fig 6, C) of patients with ACC, it was not sig- nificant in the presence of necrosis. To further evaluate the cor- relation between the expression of all the differentially regulated hub genes and the survival of ACC patients, we performed Cox Hazard model in R Studio. The results indicated that upregulation of SGPL1, and FDFT1 as well as lower expression levels of PLD1are significant, whereas the rest of the downregulated genes even though not significant had higher hazard ratio (Fig 6, D and Table II). Finally, we evaluated the expression of the hub genes in ACC cell line NCI-H295R by RT-PCR. The results shown in Fig 7 indi- cated that compared with normal fibroblast cell lines, ACC cell line had upregulation of Sphingolipid metabolism gene (SGPL1) and downregulation of phosphotidyl inositol metabolism gene (SYNJ2) as well as glycerophospholipid metabolism gene (PLA2G16).
Discussion
Recent advances in transcriptomics and proteomics have identified several gene and protein biomarkers for development of targeted therapies for ACC.17 Early DNA microarray studies identified overexpression of insulin-like growth factor (IGF2) in >90% of ACC tumors.18 The role of IGF2/insulin growth factor receptor (IGF1R) cascade in cellular growth pathways like phosphatidylinositol 3-kinase/Akt/mTOR, and mitogen-activated protein kinase fueled preclinical and clinical studies targeting the IGF2/IGF1R pathway. Several phase 1 to 3 clinical trials with IGF2/IGF1R pathway inhibitors like cixutumumab, figitumumab, or linsitinib in patients with advanced ACC revealed only 3% to 5% of patients responded with long-term regression.19-21 Inte- grated comprehensive multiplatform “omic” studies by Assie et al22 and ACC-TCGA data23 confirmed the overexpression of IGF2 and further identified several pathway drivers of ACC pathogenesis. These include recurrent somatic mutations involved in the cell cycle (RB1, TP53, CCNE1, CDK4, CDK2A),
A
SGPL1
SQLE
FDFT1
PIK3C2B
PIK3CD
1.UU
1.UU
1.UU
1 183.
1.LU
0.90.
high (n=23) Icw (n=56)
0.90.
high (n=17)
high (n=12)
high (n=56)
high (n=38) low (n=41)
low (n=62)
0.90.
low (n=67)
0.90
low (n=23)
I
0.90
overall survival probability
0.80.
O RO
overall survival probability
overall survival probability
o no
0.00
0.70-
0.70
0.70.
overall survival probability
0.70
overall survival probabilry
0.00-
0 70-
0 80-
O RO
O RO
0.00-
0.60-
0.50-
0.50.
0.50.
0.50-
0.50
0.40-
0.40
0.40-
0.40
0.40.
0.30-
0.30
0.30
0.30
0.30-
0.20-
0.20
0.20-
0 20-
0.20-
0.10-
Tp25-7
0.10
0.10-
0.10-
0 10
0.00-
bunt pu.01.
maw y 8.30-03
0.00
hont = 0.533
saw p 2.5e-or
0.00-
sont a 1.dle-05
FEW p2.7e-05 tout p 1.70-03
0
2
5
₹
8
5
₣
1
24
=
₹
&
0
£
2
=
2
=
0.00-
0.00
៛
茶
3
2º
கீ
3
&
6
2
=
A
€
Follow up in months
5
Follow up in months
Follow up in months
8
Follow up in months
$
Follow up in months
PLA2G16
DGAT1
PLD1
GPD1
CHPT1
high (n=28) low (n=51)
1.00
1
1.UU
H
1.00
1.00
1.VV
0.90
high (1-66)
0.90.
high (n=39)
0.90-
high (n=71)
0.90
high (n=45) low (n=34)
0.90
om.
low (n=13)
o no.
low (n=40)
low (n=8)
0.80
overall survival probability
0.70-
overall survival probability
0.70
overall survival probability
O BO-
0.80-
0.70-
overall survival probability
0.70
overall survival probability
0.70-
o An-
0 50-
0 50.
O FO.
0.60.
0.50
0.50.
0.50-
0.50
0.50
0.40
0.40-
0.40
0.40
0.40
0.30
0.30
0.30.
0.30
0.30
0.20
0.20
0.20
0.20
0.20
0.10
saw p 5.50-04
0.10
120 p 2.80-03
0.10
bont : 0.035
tontp 0.178
Raw p.o.011
0.10
0.00
0.00
0.00
Bonfp 0.05G
1>>
0.10-
bond p 1.20g
o no.
Ka
2
=
S
3
8
£
0
A
=
S
5
®
玉
2
=
3
5
0.00
8
E
3
2
=
FS
$
§
£
S
วิ
5
S
8
8
Follow up in months
Follow up in months
Follow up in months
Follow up in months
Follow up in months
B Sphingolipid metabolism genes (SGPL1+FDFT1+SQLE)
Phosphatidylinositol metabolism genes (PIK3C2B+PIK3CD+SYNJ2)
Glycerophospholipid metabolism genes (PLA2G16+GPD1+PLD1+DGAT1)
1.00
77
I
1.00
1.00
0.90
0.90
0.90
0.80.
0.80
0.80
0.70-
high (n=26)
0.70.
high (n=51)
survival probability
low (n=53)
survival probability
low (=28)
0.70-
high (n=58)
0.60-
LI
0.60
survival probability
low (r=21)
0.60
0.50.
0.50
0.50
0.40.
0.40.
0.40
0.30-
0.30
0.30
0.20.
0.20
0.20-
0.10-
D 1.30-03
0.10.
row p 3.4c-04
0.10
0.00
0.00
bonf > 0.022
*#|17H €4
Rentp.D.011
6
-
3
00
A
0
0
4
NO
00
1
Un
0.00
5
1
1
1
-
Follow up in months
Follow up in months
Follow up in months
mv p 0.051
hent p 1 and
£
mw p 7/19-04
bonif p 0.047
초
wnt/B-catenin signaling pathway (CTNNB1, ZNRF3, APC), Chro- matin remodeling (MEN1, DAXX), protein kinase A signaling (PRKAR1A), and other pathways. Integrated analysis of the ACC- TCGA data set additionally demonstrated the presence of 3 distinct molecular cluster of clusters, which vary in prognosis, CpG island methylator phenotype, and expression of steroid machinery. Although the development of biomarkers from comprehensive “omic” studies and immunotherapy paved the way for some novel therapeutic options for patients with ACC, their overall survival and prognosis remains poor, warranting the search for other novel pathways and biomarkers. To address this need, and the lack of information around lipid metabolism gene changes in these tumors as potential biomarkers, we performed an integrated bioinformatic analyses to evaluate the role of lipid metabolic genes in ACC pathogenesis and identified a novel set of 10 lipid metabolic genes/pathways that are associated with poor survival in ACC. These novel ACC- modulated lipid genes may serve as useful biomarkers both
for prognostic evaluation of tumors and their response to treatment, and as new targets to evaluate for future therapeutic development.
Lipids, the structural component of the membrane, play an active role in cancer development.11 Although dysregulated lipid metabolism contributes to several cellular processes in cancer tissues, the role of lipids in cancer development, pro- gression, and metastasis in cancer, including, in ACC is not well understood.24 Hence, we examined lipid and lipid protein metabolism genes by meta-analysis in correlation engine and identified differentially regulated lipid metabolism genes in ACC compared with normal tissue. Overall, 140 differentially regulated genes were significantly modulated in ACC by inte- grated analysis of the bio sets. Enriching this analysis with DAVID, STRING protein-protein network, and Cytoscape network identified differentially expressed genes belonging to phospholipid, phosphatidyl inositol, peroxisome, sphingolipid, and steroid metabolic pathways as potential biomarkers for
C
SGPL1
FDFT1
PLD1
GPD1
1.00
1.00
1.00
:
1.00
0.90
0.90
0.90
0.90
0.80-
high (n=29)
0.80-
0.80
0.80
overall survival probability
0.70-
low (n=28)
overall survival probability
0.70-
overall survival probability
0.70.
overall survival probability
0.70-
high (n=29)
0.60
0.60-
high (n=10)
0.60
high (n=37)
0.60
low (n=28)
0.50-
0.50
low (n=47)
0.50
low (n=20)
0.50
0.40-
0.40-
0.40
0.40
0.30-
0.30-
0.30
0.30
0.20
0.20-
0.20
0.20
0.10-
₱ 0.022
0.10-
raw p 1.00-03
0.10
0.10
bonf p 0.042
raw p 4.90-03
0.00-
0.00-
0.00
bonf p 0.205
₱ 0.019
8
2
=
N
8
120
E
0
Nº
8
2
&
8
套
6
2
=
2
&
0.00
8
£
8
ส
க்
2
&
8
Follow up in months
Follow up in months
茶
Follow up in months
Follow up in months
PLA2G16
DGAT1
CHPT1
PLA2C3B
1.00
1.00-
1.00
1.00
0.90
0.90
0.90
0.90
0.80
0.80
overall survival probability
0.80
0.80
0.70.
overall survival probability
0.70-
overall survival probability
high (n=29)
0.70
overall survival probability
0.70
0.60.
high (n=49)
0.60-
low (n=28)
0.60
high (n=40)
high (n=40)
0.60
low (n=17)
0.50
low (n=8)
0.50
0.50
low (n=17)
0.50
0.40
0.40-
0.40-
0.40
0.30
0.30
0.30
0.30
0.20
0.20
0.20
0.20
0.10
0.10-
raw 2.40-06
₱0.013
0.10-
bont p 1.00-04
raw p 9.80-03
0.10
0.00.
0
2
%
2
8
0.00
0
2
=
₹
&
0.00
bont p 0.411
**= 9 3.40-05
bondp 1.50-03
2
茶
5
₣
0
2
க்
₪
g
Follow up in months
£
0.00
6
=
=
A
Follow up in months
8
Follow up in months
8
五
Follow up in months
ACC. Upregulation of lipid genes that regulated sphingolipid (SGPL1) and steroid metabolism (FDFT1and SQLE) in ACC tu- mors was associated with poor overall survival. Additional analysis of pathologic features such as hormone excess, pres- ence of necrosis and mitotic rate more than 5 to 50 HPF, and upregulation of SGPL1 corroborates our finding that SGPL1 might play an essential role in ACC pathogenesis. SGPL1 binds to sphingosine-1 phosphate (SIP) and catalyzes the final step of the sphingolipid metabolic pathway.25 As the majority of ACCs are known to produce excessive steroid hormones,26 the identification of upregulated steroidogenesis genes (FDFT1and SQLE) as well as SGPL1 (that modulates steroidogenesis and cellular functions that are hallmarks of cancer like growth, migration, survival, and drug resistance), suggests that target- ing sphingolipid metabolism may represent a novel therapeutic strategy for ACC. Another finding from our study is the asso- ciation of poor survival in ACC with downregulation of several genes involved in phosphatidylinositol (PIK3C2B, PIK3CD, SYNJ2) and glycerol phospholipid metabolism (DGAT1, PLA2G16, PLD1, GPD1) pathways. Despite preclinical studies showing the efficacy of phosphatidylinositol 3-kinase/mTOR inhibitors, the phase I clinical trial with temsirolimus in com- bination with an IGF-1R inhibitor in advanced ACC resulted in tumor inhibition only in 4 of the 10 patients.27 Therefore, a better understanding of the role of phospholipid metabolism
(phosphatidylinositol and glycerol phospholipid) in ACC may provide new insights to aid in the development of novel therapeutic strategies for ACC.
Although our integrated bioinformatic analysis had shed light on novel lipid metabolic pathways and identified 10 candidate genes involved in the pathogenesis and progression of ACC, this analysis has several limitations. First, we have considered all pri- mary ACC tumors irrespective of stage or grade or patient charac- teristics to identify the role of global lipidomic profiling. Second, as ACC is a rare and heterogeneous tumor, our sample size of <100 has statistical power limitations. Furthermore, owing to small sample size as well as lack of functional, non-functional and oncolytic characterization of the tumor limits subgroup evaluation. However, validation of identified lipid metabolic hub genes in the TCGA co- horts, the largest dataset available for ACC, corroborates our find- ings as still significant even in a relatively small cohort of patients. Although the association of hub metabolic genes with poor overall survival of ACC patients in our integrated bioinformatic approach is encouraging, it will require further evaluation in preclinical models of ACC for target validation and efficacy evaluation of novel ther- apeutics for future clinical translation.
In conclusion, integrated gene expression microarray analysis identified several lipid metabolism genes and pathways that might play an essential role in the pathogenesis of ACC. Additional pre- clinical validation of these findings are needed before clinical
| Genes | coef | exp(coef) | exp(-coef) | se(coef) | z | P value |
|---|---|---|---|---|---|---|
| SGPL1Low | -0.9208 | 0.3982 | 2.5113 | 0.4461 | -2.064 | .03903 |
| SQLELow | -0.2084 | 0.8119 | 1.2317 | 0.6449 | -0.323 | .74661 |
| FDFT1Low | -1.4743 | 0.2289 | 4.368 | 0.6266 | -2.353 | .01864 |
| PIK3C2BLow | 0.2393 | 1.2704 | 0.7872 | 0.5442 | 0.44 | .66009 |
| PIK3CDLow | -0.8002 | 0.4493 | 2.2259 | 0.5596 | -1.43 | .15276 |
| SYNJ2Low | -0.0669 | 0.9353 | 1.0692 | 0.5044 | -0.133 | .89447 |
| DGAT1Low | 0.637 | 1.8909 | 0.5289 | 0.4821 | 1.321 | .18639 |
| PLD1Low | 1.9035 | 6.7093 | 0.149 | 0.6106 | 3.117 | .00183 |
| PLA2G16Low | 1.0422 | 2.8355 | 0.3527 | 0.5621 | 1.854 | .0637 |
| GPD1Low | -0.5815 | 0.5591 | 1.7887 | 0.6318 | -0.92 | .35736 |
D
Hazard ratio
SGPL1
High (N=29)
reference
Low (N=49)
0.40 (0.166 - 0.95)
0.039 *
SQLE
High (N=26)
reference
Low (N=52)
0.81 (0.229 - 2.87)
A
0.747
FDFT1
High (N=27)
reference
Low (N=51)
0.23 (0.067 - 0.78)
I
0.019 *
PIK3C2B
High (N=33)
reference
Low (N=45)
1.27
(0.437 - 3.69)
I
0.66
PIK3CD
High (N=24)
reference
Low (N=54)
0.45 (0.150 - 1.35)
I
0.153
SYNJ2
High (N=28)
reference
Low (N=50)
0.94 (0.348 - 2.51)
0.894
DGAT1
High (N=28)
reference
Low (N=50)
1.89 (0.735 - 4.86)
0.186
PLD1
High (N=32)
reference
Low (N=46)
6.71
(2.027 - 22.20)
0.002 **
PLA2G16
High (N=26)
reference
Low (N=52)
2.84
(0.942 - 8.53)
0.064
GPD1
High (N=59)
reference
Low (N=19)
0.56
(0.162 - 1.93)
1
0.357
# Events: 26; Global p-value (Log-Rank): 0.00056963
AIC: 187.56; Concordance Index: 0.82
0.05
0.1
0.2
0.5
1
2
5
10
20
2.5
Normal
ACC
2
Fold Change vs Normal
1.5
1
0.5
0
SGPL1
PLA2G16
SYNJ2
translation. From 140 differentially expressed lipid metabolic genes common to all 3 datasets, we identified 10 hub genes that signifi- cantly correlated with poor survival in ACC. Upregulation of genes involved in sphingolipid metabolism and steroidogenesis, as well as, downregulation of genes in phosphatidylinositol and glycerol phospholipid metabolism showed a significant correlation with poor survival in patients with ACC. These results suggest that un- derstanding the mechanism through which sphingolipid meta- bolism modulates steroidogenesis either alone or in combination with current therapies could lead to novel therapeutic strategies for treating ACC patients. As therapeutics targeting sphingolipid metabolism including fingolimod inhibiting SGPL1 are already in clinical trials,28,29 preclinical evaluation of these drugs alone or in combination with current therapeutics would provide both a rationale and validation for future clinical translation for patients with advanced ACC.
Funding/Support
This research was funded in part by the National Institutes of Health (R01 CA173292 and R01 CA216919 [M.S.C. and B.S.J.B.], 3U01 CA120458 [M.S.C. and B.S.J.B.]), the University of Michigan Comprehensive Cancer Center Support Grant P30-CA-046592, and the University of Michigan Department of Surgery.
Conflict of interest/Disclosure
None of the authors have any conflicts of interest (financial or otherwise) related to the material presented in this manuscript.
References
1. Kebebew E, Reiff E, Duh QY, Clark OH, McMillan A. Extent of disease at pre- sentation and outcome for adrenocortical carcinoma: have we made progress? World J Surg. 2006;30:872-878.
2. Golden SH, Robinson KA, Saldanha I, Anton B, Ladenson PW. Clinical review: prevalence and incidence of endocrine and metabolic disorders in the United States: a comprehensive review. J Clin Endocrinol Metab. 2009;94:1853-1878.
3. Else T, Kim AC, Sabolch A, Raymond VM, et al. Adrenocortical carcinoma. Endocr Rev. 2014;35:282-326.
4. Puglisi S, Perotti P, Pia A, Reimondo G, Terzolo M. Adrenocortical carcinoma with hypercortisolism. Endocrinol Metab Clin North Am. 2018;47:395-407.
5. Fassnacht M, Allolio B. Clinical management of adrenocortical carcinoma. Best Pract Res Clin Endocrinol Metab. 2009;23:273-289.
6. Glenn JA, Else T, Hughes DT, et al. Longitudinal patterns of recurrence in pa- tients with adrenocortical carcinoma. Surgery. 2019;165:186-195.
7. Ardolino L, Hansen A, Ackland S, Joshua A. Advanced adrenocortical carcinoma (ACC): a review with focus on second-line therapies. Horm Cancer. 2020;11: 155-169.
8. Fassnacht M, Kroiss M, Allolio B. Update in adrenocortical carcinoma. J Clin Endocrinol Metab. 2013;98:4551-4564.
9. Baur J, Buntemeyer TO, Megerle F, Deutschbein T, et al. Outcome after resection of adrenocortical carcinoma liver metastases: a retrospective study. BMC Cancer. 2017;17:522.
10. Butler LM, Perone Y, Dehairs J, et al. Lipids and cancer: emerging roles in pathogenesis, diagnosis and therapeutic intervention. Adv Drug Deliv Rev. 2020;159:245-293.
11. Fernandez LP, Gomez de Cedron M, Ramirez de Molina A. Alterations of lipid metabolism in cancer: implications in prognosis and treatment. Front Oncol. 2020; 10:577420.
12. Orita H, Coulter J, Lemmon C, et al. Selective inhibition of fatty acid synthase for lung cancer treatment. Clin Cancer Res. 2007;13:7139-7145.
13. Hao Y, Li D, Xu Y, et al. Investigation of lipid metabolism dysregulation and the effects on immune microenvironments in pan-cancer using multiple omics data. BMC Bioinformatics. 2019;20(Suppl 7):195.
14. Beloribi-Djefaflia S, Vasseur S, Guillaumond F. Lipid metabolic reprogramming in cancer cells. Oncogenesis. 2016;5:e189.
15. Gazdar AF, Oie HK, Shackleton CH, et al. Establishment and characterization of a human adrenocortical carcinoma cell line that expresses multiple pathways of steroid biosynthesis. Cancer Res. 1990;50:5488-5496.
16. Szklarczyk D, Morris JH, Cook H, et al. The STRING database in 2017: quality- controlled protein-protein association networks, made broadly accessible. Nucleic Acids Res. 2017;45(D1): D362-D38.
17. Mohan DR, Lerario AM, Hammer GD. Therapeutic targets for adrenocortical carcinoma in the genomics era. J Endocr Soc. 2018;2:1259-1274.
18. Barlaskar FM, Spalding AC, Heaton JH, et al. Preclinical targeting of the type I insulin-like growth factor receptor in adrenocortical carcinoma. J Clin Endo- crinol Metab. 2009;94:204-212.
19. Haluska P, Worden F, Olmos D, et al. Safety, tolerability, and pharmacokinetics of the anti-IGF-1R monoclonal antibody figitumumab in patients with re- fractory adrenocortical carcinoma. Cancer Chemother Pharmacol. 2010;65: 765-773.
20. Lerario AM, Worden FP, Ramm CA, et al. The combination of insulin-like growth factor receptor 1 (IGF1R) antibody cixutumumab and mitotane as a first-line therapy for patients with recurrent/metastatic adrenocortical carci- noma: a multi-institutional NCI-sponsored trial. Horm Cancer. 2014;5: 232-239.
21. Fassnacht M, Berruti A, Baudin E, et al. Linsitinib (OSI-906) versus placebo for patients with locally advanced or metastatic adrenocortical carcinoma: a double-blind, randomised, phase 3 study. Lancet Oncol. 2015;16:426-435.
22. Assie G, Letouze E, Fassnacht M, et al. Integrated genomic characterization of adrenocortical carcinoma. Nat Genet. 2014;46:607-612.
23. Zheng S, Cherniack AD, Dewal N, et al. Comprehensive pan-genomic charac- terization of adrenocortical carcinoma. Cancer Cell. 2016;30:363.
24. Baenke F, Peck B, Miess H, Schulze A. Hooked on fat: the role of lipid synthesis in cancer metabolism and tumour development. Dis Model Mech. 2013;6: 1353-1363.
25. Maharaj A, Williams J, Bradshaw T, et al. Sphingosine-1-phosphate lyase (SGPL1) deficiency is associated with mitochondrial dysfunction. J Steroid Biochem Mol Biol. 2020;202:105730.
26. Ponnusamy S, Meyers-Needham M, Senkal CE, et al. Sphingolipids and cancer: ceramide and sphingosine-1-phosphate in the regulation of cell death and drug resistance. Future Oncol. 2010;6:1603-1624.
27. Tacon LJ, Prichard RS, Soon PS, Robinson BG, Clifton-Bligh RJ, Sidhu SB. Current and emerging therapies for advanced adrenocortical carcinoma. Oncologist. 2011;16:36-48.
28. Shaw J, Costa-Pinheiro P, Patterson L, Drews K, Spiegel S, Kester M. Novel sphingolipid-based cancer therapeutics in the personalized medicine era. Adv Cancer Res. 2018; 140:327-366.
29. Rupp T, Pelouin O, Genest L, Legrand C, Froget G, Castagne V. Therapeutic potential of Fingolimod in triple negative breast cancer preclinical models. Transl Oncol. 2021;14:100926.
Discussion
Dr. James Howe (Iowa City): How did you arrive at the cutoff for high and low expression? Did you try to fit that best to survival?
Second, what would you propose to do with these three genes that were found to be significant by multivariate analysis? Do you think they could be used to look at tissue sections after resection to decide who gets adjuvant therapy?
Dr. Chitra Subramanian: Thank you so much for the very interesting questions. First, the analysis itself has up- and down- regulated genes. Since we had only a total of 140 genes that were significantly differentially regulated, we used all the genes in the analysis. To your second point, we used a standard cutoff of at least 5X change in expression and that was related to survival analysis. We then looked at the genes that are specifically associated with a poorer overall survival of ACC.
We plan to use all these three genes from the multivariate analysis to screen in future, especially for sphingolipid metabolic genes. It would be interesting to look at whether they are useful or not, because there already is a drug targeting the sphingolipid metabolism, Fingolimod, that is used for MS in clinical trials. If we could figure this out correctly, it could more easily enter clinical trials in ACC because the drug is already in use in other clinical trials.
Dr. Martha Zeiger (Bethesda): Are there any known potential targets within these pathways?
Check for updates
Dr. Chitra Subramanian: Yes, there are other genes that are involved in phospholipid metabolism down regulation. We can see whether they are down regulated or not. FDFT-1 also is one of the steroidogenesis pathway regulatory genes, but there are several other genes that can be used to identify which pathways are clin- ically associated with lipid metabolism.
Dr. Juan Pablo Pantoja (Ciudad de Mexico): Do you think you can actually validate this in tissue samples by going back and figuring it out in an independent cohort or with RT-PCR?
Dr. Chitra Subramanian: I think we can very easily do this. I have started looking at some of these genes by RT-PCR in cell lines and already noted SGPL1 up-regulation even in cell lines compared to normal cells.
Dr. Naira Baregamian (Nashville): How do tumor lipid me- tabolites correlate with these gene expression profiles?
Dr. Chitra Subramanian: We have not looked specifically at that yet and it is an interesting question. We can, however, comment about the PI3K kinase pathway, because previously in clinical trials where PI3K inhibitors were used, only a partial response was observed. What we found in vitro is that the PI3K pathway is down- regulated based on the lipid metabolism, so maybe this could explain some of the partial response observed clinically. Further correlation studies are needed in the future though to look at this question in more detail.
Surgery is abstracted and/or indexed in Index Medicus, Science Citation Index, Current Contents/ Clinical Medicine, Current Contents/Life Sciences, and MEDLINE.
This Journal has been registered with Copyright Clearance Center, Inc, 222 Rosewood Dr, Danvers, MA 01923. Consent is given for the copying of articles for personal or internal use of specific clients. This consent is given on the condition that the copier pay directly to the Center the per-copy fee stated on the first page of each article for copying beyond that permitted by US Copyright Law. This consent does not extend to other kinds of copying, such as for general distribution, resale, advertising and promotional purposes, or for creating new collective works. All inquiries regarding copyrighted material from this publication other than those that can be handled through Copyright Clearance Center should be directed to Journals Permission Department, Elsevier Inc, 3521 Riverport Lane, Maryland Heights, MO 63043; (314) 447-8871