Targeted Molecular Analysis in Adrenocortical Carcinomas: A Strategy Toward Improved Personalized Prognostication
Juliane Lippert,1 Silke Appenzeller,2 Raimunde Liang,3 Silviu Sbiera,3 Stefan Kircher,4 Barbara Altieri,3,5 Indrajit Nanda,1 Isabel Weigand,3 Andrea Gehrig,1 Sonja Steinhauer,3 Renzo J. M. Riemens,1,6 Andreas Rosenwald,4, Clemens R. Müller,1 Matthias Kroiss,3,7 Simone Rost,1 Martin Fassnacht,3,7,8 and Cristina L. Ronchi3,9,10
1Institute of Human Genetics, University of Würzburg, 97074 Würzburg, Germany; 2Core Unit Bioinformatics, Comprehensive Cancer Center Mainfranken, University Hospital of Würzburg, 97080 Würzburg, Germany; 3Department of Medicine I, Division of Endocrinology and Diabetes, University Hospital, University of Würzburg, 97080 Würzburg, Germany; 4Institute for Pathology, University of Würzburg, 97080 Würzburg, Germany; 5Division of Endocrinology and Metabolic Diseases, Catholic University of the Sacred Heart, 00168 Rome, Italy; 6Department of Psychiatry and Neuropsychology, School for Mental Health and Neuroscience, Maastricht University, 6211 LK Maastricht, Netherlands; 7Comprehensive Cancer Center Mainfranken, University of Würzburg, 97080 Würzburg, Germany;
8Central Labor, University Hospital of Würzburg, 97080 Würzburg, Germany; 9Institute of Metabolism and System Research, University of Birmingham, B15 2TT Birmingham, England; and 1ºCentre for Endocrinology, Diabetes and Metabolism, Birmingham Health Partners, B15 2TT Birmingham, England
Context: Adrenocortical carcinoma (ACC) has a heterogeneous prognosis, and current medical therapies have limited efficacy in its advanced stages. Genome-wide multiomics studies identified molecular patterns associated with clinical outcome.
Objective: Here, we aimed at identifying a molecular signature useful for both personalized prognostic stratification and druggable targets, using methods applicable in clinical routine.
Design: In total, 117 tumor samples from 107 patients with ACC were analyzed. Targeted next- generation sequencing of 160 genes and pyrosequencing of 4 genes were applied to formalin-fixed, paraffin-embedded (FFPE) specimens to detect point mutations, copy number alterations, and promoter region methylation. Molecular results were combined with clinical/histopathological parameters (tumor stage, age, symptoms, resection status, and Ki-67) to predict progression-free survival (PFS).
Results: In addition to known driver mutations, we detected recurrent alterations in genes not previously associated with ACC (e.g., NOTCH1, CIC, KDM6A, BRCA1, BRCA2). Best prediction of PFS was obtained integrating molecular results (more than one somatic mutation, alterations in Wnt/ B-catenin and p53 pathways, high methylation pattern) and clinical/histopathological parameters into a combined score (P < 0.0001, x2 = 68.6). Accuracy of prediction for early disease progress was 83.3% (area under the receiver operating characteristic curve: 0.872, 95% confidence interval 0.80 to 0.94). Furthermore, 17 potentially targetable alterations were found in 64 patients (e.g., in CDK4, NOTCH1, NF1, MDM2, and EGFR and in DNA repair system).
ISSN Print 0021-972X ISSN Online 1945-7197 Printed in USA
First Published Online 2 August 2018
Abbreviations: ACC, adrenocortical carcinoma; CN, copy number; CNA, copy number alteration; CNV, copy number variation; COMBI, combined; DFS, disease-free survival; ENSAT, European Network for the Study of Adrenocortical Tumors; FC, fold change; FFPE, formalin fixed, paraffin embedded; FISH, fluorescence in situ hybridization; GRAS, tumor grade, R status, age, symptoms; HR, hazard ratio; Mb, megabase; mGRAS, modified tumor grade, R status, age, symptoms; MMR, mismatch repair; NGS, next-generation sequencing; PFS, progression-free survival; R, resection; TMB, tumor mutational burden.
Conclusions: This study demonstrates that molecular profiling of FFPE tumor samples improves prognostication of ACC beyond clinical/histopathological parameters and identifies new potential drug targets. These findings pave the way to precision medicine in this rare disease. (J Clin Endocrinol Metab 103: 4511-4523, 2018)
A drenocortical carcinoma (ACC) is a rare tumor with a generally poor but heterogeneous prognosis [5-year survival rate ranging from 13% to 80% (1, 2)]. Tumor stage according to the European Network for the Study of Adrenocortical Tumors (ENSAT) classification, which has now been used also by the American Joint Committee on Cancer TNM classification system (3), is one of the most relevant prognostic factors (1). However, ~10% of patients with metastatic disease at diagnosis are still alive after 10 years, and >20% of patients with tumor stages I to III die within the first 3 years (1). Resection (R) status of the primary tumor (4, 5) and Ki- 67 index (6) represent additional prognostic factors. A recent study also proposed a combination of clinical/ histopathological parameters (i.e., tumor grade, R status, age, symptoms = GRAS score) to improve prognostica- tion in patients with advanced ACC (7).
Genome-wide studies have identified molecular pat- terns associated with clinical outcome (8, 9). Among these, a specific gene expression pattern (i.e., high BUB1B-PINK1 levels) (8, 10), specific copy number (CN) alteration (CNA) (9), and CpG island methylation patterns (8, 11) have been associated with a poor prognosis. However, these studies have been performed on fresh-frozen tumor samples that are difficult to collect in routine clinical settings. Moreover, cost-intensive genome-wide technologies and complex bioinformatics workup were required, which precludes the adoption of the proposed prognostic biomarkers in clinical practice.
At present, few effective pharmacological therapies are available for ACC (12). Mitotane (Lysodren; Bristol- Myers Squibb, Princeton, NJ) is the only approved drug, but an objective response is observed in only ~20% of cases, and treatment is limited by severe adverse reactions (13, 14). Even combined therapies of mitotane and cyto- toxic chemotherapies, like etoposide-doxorubicin-cisplatin (15), streptozotocin (15), and gemcitabine plus capecitabin (16, 17), exhibit response rates <25%. Although some studies provided some promising insights into potential pharmacological targets (18-20), effective targeted thera- pies have not been identified yet (2, 21).
The main aim of the current study was to identify a molecular tumor signature for a prognostic classification of patients with ACC that may be easily transferred into clinical practice. To this end, we used 117 standard formalin-fixed, paraffin-embedded (FFPE) tumor tissue specimens to investigate the prognostic power of both
previously proposed or new molecular markers and potential drug targets, which we evaluated by targeted next-generation sequencing (NGS).
Patients and Methods
Study protocol
This is a single-center retrospective study designed and conducted in accordance with the Declaration of Helsinki. We followed the recommendations for tumor prognostic markers studies reported in the Reporting Recommendations for Tu- mour Marker Prognostic Studies (REMARK) (22). Moreover, we searched for known drug targets using the OncoKB website (23). The study protocol was approved by local ethics com- mittee (#88/11), and written informed consent was obtained from all patients prior to study enrollment.
Patient cohort and clinical data
In total, 107 patients were selected for the study. Inclusion criteria were histologically confirmed diagnosis of ACC and availability of FFPE tumor specimens collected between 2002 and 2016 and corresponding blood samples. Initial clinical/ histopathological parameters and follow-up data were collected through the ENSAT registry (Table 1).
A modified version of the GRAS classification (7) (mGRAS score) was used to merge prognostically relevant clinical/ histopathological data: tumor stage (ENSAT 1 to 2 = 0 points, 3 = 1 point, 4 = 2 points), grading (Ki-67 proliferation index 0 to 9 = 0 points, 10 to 19 = 1 point, ≥20 = 2 points), R status (RO = 0 points, RX = 1 point, R1 = 2 points, R2 = 3 points), age (<50 years = 0 points, ≥50 years = 1 point), and symptoms due to steroid autonomous secretion or tumor mass (no = 0 points, yes = 1 point).
Progression-free survival (PFS) was the major outcome being defined as the time from tumor resection (primary surgery) to first radiological evidence of disease relapse or disease- related death.
Material collection and DNA isolation
The final series included 117 FFPE samples (89 primary tumors, 10 local recurrences, and 18 distant metastases). In 10 cases, tumor tissues were available from consecutive surgeries of the same patients (seven with primary + metastasis, one with primary + local recurrence, one with local recurrence + me- tastasis, and one with two metastases). For survival analyses, only the chronologically first sample of a patient was used (either primary tumor or local recurrence/metastasis). The tu- mor cell content in each FFPE slide was assessed by hematoxylin and eosin staining and reached a high fraction (median, 90%; range, 60% to 95%). DNA was isolated from tumors with the GeneRead DNA FFPE Kit (Qiagen, Hilden, Germany) and from peripheral blood with the NucleoSpin Blood L Kit (Macherey- Nagel, Bethlehem, PA) according to the manufacturer’s
| Parameter | Value (N = 107) |
|---|---|
| Sex, male/female, no. | 46/61 |
| Baseline (at time of diagnosis) | |
| Age, median (range), y | 49 (18-87) |
| <50 y | 57 (53.3) |
| ≥50 y | 50 (46.7) |
| Clinical presentation (available data) | 107 |
| Incidentally | 31 (29.0) |
| Tumor related | 45 (42.0) |
| Hormone related | 31 (29.0) |
| Hormone secretion (available data) | 81 |
| Glucocorticoids alone | 24 (29.6) |
| Mixed secretion | 25 (30.8) |
| Endocrine inactive | 25 (30.8) |
| Others (androgens, mineralocorticoids, | 7 (8.6) |
| estrogens) | |
| Unknown, no. | 26 |
| Initial ENSAT tumor stage (available data) | 107 |
| 1-2 | 56 (52.3) |
| 3 | 28 (26.2) |
| 4 (metastatic) | 23 (21.5) |
| Resection status (available data) | 104 |
| R0 | 74 (71.1) |
| RX | 16 (15.4) |
| R1 | 5 (4.8) |
| R2 | 9 (8.6) |
| Unknown, no. | 3 |
| Ki-67 proliferation index, median (range) | 12 (1-90) |
| 0-9 | 31 (29.0) |
| 10-19 | 33 (30.8) |
| ≥20 | 43 (40.2) |
| Type of tumor | |
| Primary | 89 (83.2) |
| Local recurrence | 9 (8.4) |
| Metastasis | 9 (8.4) |
| During follow-up | |
| Duration of follow-up, median (range), mo | 31 (3-274) |
| Deaths | 54 (50.5) |
| Local therapeutic approaches | |
| Additional surgeries | 26 (24.3) |
| Radiotherapy (tumor bed or metastasis) | 19 (17.8) |
| Iodometomidate | 7 (6.5) |
| Mitotane | 73 (68.2) |
| Adjuvant setting | 39 (36.4) |
| Palliative setting | 34 (31.8) |
| Cytotoxic chemotherapies | |
| None | 45 (42.1) |
| Platinum compounds | 52 (48.6) |
| Streptozotocin | 44 (41.1) |
| Gemcitabine plus capecitabin | 36 (33.6) |
Values are presented as number (%) unless otherwise indicated.
Abbreviations: R0, complete resection; R1, microscopic incomplete re- section; R2, macroscopic incomplete resection; RX, uncertain resection.
instructions. Qualitative and quantitative evaluation of DNA fragmentation was assessed by the GeneRead DNA Quanti- MIZE Assay Kit (384) (Qiagen). Quantitative PCRs were performed with a SYBRGreen mix according to the manu- facturer’s protocol and measured with a ViiA7 RT-PCR System (Thermo Fisher Scientific, Manassas, VA). Data were analyzed
with QuantStudio RT-PCR Software (Applied Biosystems, Foster City, CA). The quality of all DNA samples was calcu- lated with GeneRead DNA QuantiMIZE_384_DataAnalysis (Qiagen) in comparison with a control DNA included in the kit. Only DNA with a quality control (QC) score (indicator of sample damage/fragmentation) ≤0.04 was sequenced.
Targeted DNA sequencing
Tumor and leukocyte DNAs were enriched with the Gen- eRead DNAseq Human Comprehensive Cancer Panel V2 and GeneRead DNAseq Panel PCR Kit V2 (both Qiagen), according to the manufacturer’s protocol. This panel includes coding regions of 160 genes (7951 amplicons and 744,835 bases of target regions), many of them known or suspected to be in- volved in adrenocortical tumorigenesis or known drug targets. NGS was performed on a NextSeq500 with NextSeq Mid Output Reagent Kit V2 and 150-bp paired end reads (Illumina, San Diego, CA). Raw data were aligned and analyzed with GensearchNGS (Phenosystems S.A., Wallonia, Belgium). For detection of somatic point mutations and small insertions and deletions (small indels) in tumor samples, the called variants were filtered as followed: coverage >100, exon distance <21, frequency of appearance >0.1, minor allele frequency <0.02, and variant balance >0. Variants found also in the matched blood samples were excluded. Intronic and synonymous vari- ants have been considered for calculating tumor mutational burden (TMB) but not for further analysis, except of those with a predicted influence on splice sites.
Impact on splicing for intronic and synonymous variants was analyzed with Alamut software (Interactive Biosoft- ware, Rouen, France) using five prediction algorithms: SpliceSiteFinder-like, MaxEntScan (24), NNSPLICE (25), GeneSplicer (26), and Human Splicing Finder (27). All other variants were evaluated for predicted pathogenicity by the Polymorphism Phenotyping v2 algorithm tool (PolyPhen-2) (28), Sorting Tolerant From Intolerant algorithm, and Mutation Taster (29). The Catalogue of Somatic Mutations in Cancer was used as a reference of cancer-related somatic variants. Inter Var was used as an additional tool for the interpretation of variants (30).
ZNRF3, which was previously reported to be involved in the pathogenesis of ACC (8, 9, 31), was evaluated separately by direct Sanger sequencing. PCR primers for the coding region of ZNRF3, except exon 1, were designed with Primer3 (version 4.0.0) software (3, 32). Sequencing data were gen- erated with an ABI 3730 or an ABI 3130xl capillary sequencer under standard conditions and analyzed with Gensearch (Phenosystems S.A.).
TMB was calculated by summing all detected somatic variants and dividing this number by the size of the target region. Values are specified in variations per megabase (Mb).
To analyze CNAs, we used a combination of two different approaches: the CNV analysis tool from GensearchNGS and an in-house pipeline. Single CN gains or losses have been identified by comparing the CN of matched tumor and blood samples. With GensearchNGS, a fold change (FC) of 1 was considered “normal.” Hence, genes with all amplicons having an FC be- tween 1.25 and 1.75 were considered as “heterozygous du- plicated” and as “homozygous duplicated” with an FC >1.75. Genes were considered deleted when all amplicons showed an FC <0.75. For the second approach, an initial quality as- sessment was performed using FastQC, v0.11.3 (Illumina).
Adapters and low-quality reads were trimmed using TrimGalore, v0.4.0 (Babraham Bioinformatics, Cambridge, United Kingdom) powered by Cutadapt, v1.8 (Department of Computer Science, TU Dortmund, Dortmund, Germany). The reads were aligned against the UCSC hg19 human reference genome with BWA mem, v0.7.12 (Wellcome Trust Sanger Institute, Cambridge, United Kingdom) using default parameters (33). Sorted BAM files were created using Picard v1.125 (Broad Institute, Cam- bridge, MA) and indexed using SAMtools v1.3 (Wellcome Trust Sanger Institute, Cambridge, United Kingdom; Broad Institute of MIT; and Harvard, Cambridge, MA) (34). Local realignment around indels was executed with GATK, v3.5 (Broad Institute) (35). For CNV, calling the number of reads of each amplicon was determined using the multiBamCov-Tool in the BEDTools suite, v2.26.0 (Eccles Institute for Human Genetics, University of Utah, Salt Lake City, UT) (36). Only markers covered with an average of at least 200 reads in control samples in the re- spective panel were considered. For normalization, the reads for each amplicon were divided by the total number of reads for each sample. Log2 FC was calculated for each amplicon passing QC using the corresponding amplicon in the matched control. A gene was considered amplified or deleted if at least 80% of all markers in a tumor covering the gene were amplified or deleted at least 1.5-fold. In both approaches, CNAs were only investigated for genes covered by at least six probes. Only CN alterations detected with both approaches were considered.
Targeted DNA methylation analysis
Bisulfite pyrosequencing was used for quantitative methyl- ation analysis of four tumor suppressor genes, PAX5, PAX6, PYCARD, and GSTP1, that have been demonstrated to play a substantial prognostic role in ACC (11). A total of 500 ng DNA from tumor and matched blood samples was used to perform bisulfite conversion and cleanup of converted DNA with the EpiTect Fast 96 DNA Bisulfite Kit (Qiagen) according to the manufacturer’s protocol. Target regions of the assays were selected to include the regions accessible with the MLPA ME002 tumor suppressor 2 probe mix (MRC-Holland, Amsterdam, Netherlands) (3). PCR and sequencing primers were designed with PyroMark Assay Design 2.0 software (Qiagen) (3). Bisulfite-treated DNA was amplified in 25-uL reactions containing 2.5 µL 10× PCR buffer with 20 mM MgCl2, 0.5 µL 10 mM dNTP mix, 1.0 µL (10 pmol) of each forward and reverse primer, 0.2 uL FastStart Taq DNA Poly- merase (5 U/pL), 18.8 L PCR-grade H2O, and 1 µL bisulfite- converted DNA. PCR was carried out with an initial denaturation step at 95℃ for 5 minutes, followed by 45 cycles at 95℃ for 30 seconds, primer-specific annealing temperature (58℃ for PAX5 and PYCARD, 59℃ for PAX6, and 60℃ for GSTP1) for 30 seconds, and elongation at 72℃ for 30 seconds and a final extension step at 72℃ for 7 minutes. Bisulfite pyrosequencing was performed on a PyroMark Q96 MD Pyrosequencing System with the PyroMark Gold Q96 CDT Reagents Kit (Qiagen). Pyro Q-CpG software (Biotage, Uppsala, Sweden) was used for data analysis.
Fluorescence in situ hybridization analysis
To validate CDK4 CN gains, we investigated six repre- sentative 2-um-thick FFPE slides by Fluorescence in situ hybridization (FISH) analysis (two samples with “homozy- gous” CDK4 amplification, two with “heterozygous” CDK4
amplification, and two with normal CDK4 allele status at NGS). CDK4 gene amplification was visualized through hybridization of a Zytolight SPEC CDK4/CEN12 Dual Color Probe (ZytoVision GmbH, Germany) (D12Z3) according to the manufacturer’s recommendation. At least 200 nonoverlapping nuclei per sample were evaluated by fluorescence microscopy (Zeiss Axioskop, Jena, Germany) using the appropriate filter sets. Only nuclei with a distinct nuclear border showing clear hy- bridization signals were evaluated. CDK4 gene was considered heterozygous amplified when the FISH signal ratio of CDK4/ CEN12 was between 1.0 and 2.0 or homozygous amplified when the ratio was ≥2.0. Ratios may nevertheless differ when gains affect whole chromosome 12.
Targeted gene expression analysis
The mRNA expression of BUB1B and PINK1 was evaluated by quantitative real-time RT-PCR only in samples with high- quality RNA and cDNA (n = 38). All baseline clinical/ histopathological characteristics as well as follow-up data of this subgroup of patients did not differ from those of the entire series. RNA was isolated from tumors by miRNeasy FFPE (Qiagen). High RNA quality was tested using an Agilent 2100 Bioanalyzer (RNA integrity number >7.5). RNA was reverse transcribed by the Quantitec Reverse Transcription Kit (Qiagen). A quantitative RT-PCR for ß-actin and GAPDH was performed, and only samples with a cycle threshold of <35 were included for further analysis (n = 38). The expression of BUB1B and PINK1 was evaluated by quantitative RT-PCR using Taqman BUB1B (Hs01084828_m1) and PINK1 (Hs00260868_m1) probes with expressed B-actin (Hs9999903_m1) as reference (Applied Bio- systems, Darmstadt, Germany). Each PCR reaction was done with 40 ng cDNA, and each analysis was performed in duplicate. Transcript levels were determined using the TaqMan Gene Ex- pression Master Mix (Applied Biosystems), the CFX96 real-time thermocycler (Bio-Rad, Hercules, CA), and the Bio-Rad CFX Manager 2.0 software. Cycling conditions were 95℃ for 3 minutes, followed by 40 cycles of 95℃ for 30 seconds, 60℃ for 30 seconds, and 72℃ for 30 seconds. The ACT method was applied for normalization of gene expression levels to those of ß-actin. The ACT(BUB1B)-ACT(PINK1) expression was then calculated (10). Statistical analysis with different cutoff values was performed. The best cutoff value for a high BUB1B-PINK1 differential expression was 6.3, which was already previously suggested by De Reynies et al. (10).
Statistical analysis
A Fisher exact or X test was used to investigate dichot- omic variables, whereas a two-sided t test (or Mann-Whitney nonparametric test) was used to compare two groups of continuous variables as appropriate. A nonparametric Kruskal-Wallis test, followed by Bonferroni post hoc test, was used for comparison among several groups for non- normal distributed variables. Correlations and 95% CIs between different parameters were evaluated by linear regression analysis. Survival curves were obtained by Kaplan-Meier esti- mates, and the differences between two or more curves were assessed by the log-rank (Mantel-Cox) test. Multivariate re- gression analysis was performed by the Cox proportional hazard regression model to identify those factors that might in- dependently influence survival.
To assess and compare the prognostic accurateness and performance of different markers or scores, we used two
approaches: (i) We used the x2 (log-rank) values (deviance x2 test) to determine the goodness-of-fit statistic of the regression model, representing a surrogate of a likelihood ratio test; (ii) We calculated the sensitivity, specificity, and accuracy of different models categorizing patients with or without disease recurrence/progress within 24 months from primary surgery as affected/nonaffected. Finally, we considered the area under the receiving operating characteristic curve and 95% CI for pre- dicted probability of disease progression within 24 months from primary surgery.
Statistical analyses were made using GraphPad Prism (ver- sion 6.0; GraphPad Software, La Jolla, CA) and SPSS software (version 23; SPSS, Inc., Chicago, IL). P values <0.05 were considered statistically significant.
Results
Targeted molecular analysis of ACC: overview
The clinical and histopathological characteristics of the 107 patients selected for the study are shown in the Table 1 (see also “Patients and Methods”).
By performing targeted NGS in 117 ACC samples, we found a median TMB of 1.3/Mb (range: 0 to 22.8/Mb). Altogether, we found 237 somatic genetic variants (single nucleotide variants and small indels). The complete list of alterations and their characteristics is shown in an online repository (3). Considering the 10 cases with FFPE samples from consecutive surgical interventions, most variants in driver genes were conserved in samples obtained from the same patient (3). Thus, we considered only the first available sample from each of the 107 patients with ACC. Among them, 30 presented no mutations, 25 had one mutation, and 52 had at least two mutations (median per sample: 1; range: 0 to 14; ≥5 mutations in 13 cases). Overall, 215 protein-altering somatic variations were found, affecting 69 of 161 evaluated genes. Among the affected genes, 40 were mutated in at least two samples and 17 in at least three samples (frequency ≥2.8%) (3). The frequency of recurrent mutations previously de- scribed in ACC (8, 9, 31) and in our series is shown in Fig. 1A. The most frequently mutated genes were TP53 (22%), CTNNB1 (17%), NF1 (11%), APC (8.4%), ZNRF3 (8.4%), MEN1 (7.4%), GNAS (6.5%), and ATRX (6.5%). We also discovered novel recurrent mutations not clearly associated with ACC yet, such as in NOTCH1, CIC, KDM6A, BRCA1, and BRCA2 (all ≥2.8%) [Fig. 1A and (3)].
We then evaluated somatic CNVs in the same series. Most frequent CN gains were observed in CDK4 (43%), NOTCH1 (19%), TERT (12%), FGFR3 (12%), and MDM2 genes (7.4%) and CN losses at RB1 (5.6%), as expected (Fig. 1B). The presence of amplifications at the CDK4 locus was confirmed by FISH analysis (see Fig. 2). We also found CN alterations that were not previously reported in ACC, such as gains in STK11 (31%) and GNA11 (17%)
and losses in TNFRSF14 (30%), SMARCB1 (22%), FLCN (20%), and CHEK2 (13%) (Fig. 1B).
Using our targeted sequencing approach, we iden- tified three different CN patterns, consistent with a previous report (9). Accordingly, we defined them as “chromosomal” when at least three large chromosomal regions were affected by amplifications or deletions, “quiet” when fewer than three regions were altered, and “noisy” when several small regions were affected [modified from Zheng et al. (9)]. An example of each CN pattern is reported in an online repository (3). Thirty-eight ACC samples were recognized to present a “chromosomal” pattern, 44 a “quiet” pattern, and 25 a “noisy” pattern.
The two most frequently affected pathways were p53/ Rb signaling (59.8%; including alterations in CDKN2A, CDK4, MDM2, RB1, and TP53) and Wnt/B-catenin pathway (33.6%, including alterations in APC, CTNNB1, MED12, MEN1, and ZNRF3). In 22 of 107 samples (20.6%), both pathways were involved. Three of these patients (2.8%) had variations in CTNNB1 and TP53. Another frequently altered pathway was the chromatin remodeling pathway (29.9%) (Fig. 1C). In a lower per- centage of cases, genetic variations in genes of the DNA repair (7.4%) or the mismatch repair (MMR) systems (4.5%) were observed (Fig. 1C).
The methylation pattern of promoter regions of four preselected genes was also evaluated. The median percentage of methylated promoter regions in the tumor material was 11% at PAX5 (range, 1% to 98%), 22% at PAX6 (range, 2% to 97%), 17% at PYCARD (range, 1% to 94%), and 3% at GSTP1 (range, 1% to 74%). Considering all genes, the median value of mean methylation was 21% (range, 2% to 77%). Thirty-three tumors presented a promoter meth- ylation status of “high” (31% of cases).
A high BUB1B-PINK1 differential expression is a known negative prognostic marker in ACC (10). Thus, we evaluated BUB1B and PINK1 mRNA expression levels in a subgroup of 38 FFPE tumor specimens with good RNA quality (32.5%). The analysis of this series revealed a high BUB1B-PINK1 differential expression in 16 cases [42%, (3)].
Prognostic stratification
To evaluate the benefit of applying a molecular clas- sification to prognosticate clinical outcome, we first in- vestigated the prognostic effectiveness of ENSAT tumor staging classification in our series. As expected, the median PFS was shorter for patients with metastatic disease (ENSAT 4, n = 23) than for those with intermediate (ENSAT 3, n = 28) or early tumor stages (ENSAT 1 to 2, n = 58) (P < 0.0001, x2= 35.6; Fig. 3A). However, using the mGRAS score (see Methods), we obtained an
A)
C)
25.0
22.5
Somatic mutations (current series, n=107)
& Somatic mutations (literature, n=182)
p53/Rb1 signaling 59.8% of cases altered
DNA mismatch repair 4.5% of cases altered
Chromatin remodelling 29.9% of cases altered
20.0
Frequency somatic mutations (%)
17.5
CDKN2A
MLH1
ATRX
DAXX
0%
1.8%
6.5%
4.7%
15.0
Deleted NA, mutated 0 %
Mutated
Mutated
Mutated
12.5
CDK4
MDM2
MSH2
KDM6A
SETD2
10.0
44.8%
7.6%
1.8%
2.8%
1.8%
7.5
Amplified 43 %, mutated 1,8 %
Amplified
Mutated
Mutated
Mutated
5.0
RB1
TP53
MSH6
MEN1
DNMT3A
9.9%
22.4%
0.9%
7.4%
1.8%
2.5
P
Deleted 5.2 %, mutated 4.7 %
Mutated
Mutated
Mutated
Mutated
0.0
MA
MSH
TP53
DNA damage response/ genomic stability
TERT 12.4%
FB
Cell cycle progression
KMT2D
1.8%
B)
Amplified
Mutated
50
Wnt/beta catenin 33.6% of cases altered
NOTCH-Signalling 27.1% of cases altered
45
CN loss heterozygosis %
ZNRF3
APC
FBXW7
BRCA-DNA repair 7.4% of cases altered
DCN gain heterozygosis %
8.4%
8.4%
1.9%
40
CN gain homozygosis %
Deleted NA, mutated 8,4 %
Mutated
Mutated
ATM
Frequency copy number alterations (%)
35
1.8%
CTNNB1
NOTCH1
Mutated
30
16.8%
25.2%
Mutated
Amplified 20.5%, mutated 4.7 %
BRCA1
25
2.8%
Cell cycle progression
Cell-cell communication
Mutated
20
15
Percentage of Cases (%)
BRCA2
50
0
50
2.8%
10
Inactivated
Activated
Inhibition Activation
Mutated
5
☐ Variable
DNA damage response/ genomic stability
0
ABL1
CDC73
CSF1R
ECT2L
EPCAM
EZH2
MAP3K1
MLH1
NPM1
PHF6
SPOP
FGFR2
MSH2
SLC7A8
VHL
GNAQ
GNAS
PIK3R1
SDHB1
BRIP1
CIC
IL7R
RB1
SMAD4
MDM2
MAP2K4
FGFR3
TERT
CHEK2
GNA11
FLCN
NOTCH1
SMARCB1
TNFRSF14
STK11
CDK4
improved prognostic stratification by recognizing four subgroups with a different clinical outcome, from fa- vorable prognosis (median PFS = 54 months) to poor prognosis (median PFS = 3 months) (P < 0.0001, x2= 49.0, Fig. 3B).
Considering the results of the targeted molecular analysis, five events predicted a shorter PFS in univariate analysis: (i) presence of more than one mutation [P = 0.0015, hazard ratio (HR) = 2.12, 95% CI = 1.3 to 3.4], (ii) noisy CNA pattern (P = 0.0038, HR = 2.46, 95% CI = 1.3 to 4.5), (iii) presence of alterations in Wnt/B-catenin sig- naling alone or together with p53/Rb (P < 0.0001), (iv) “high” promoter methylation status (P = 0.0002, HR = 2.9, 95% CI = 1.7 to 5.0), and (v) high BUB1B-PINK1 differential expression (n = 38, P = 0.0037, HR = 2.56, 95% CI = 1.16 to 5.67). To investigate the applicability of a molecular prognostic classification in a clinical setting, we developed a simplified score excluding parameters that cannot be reliably and easily analyzed by targeted analysis in FFPE samples (i.e., CNA pattern and mRNA ex- pression). At multivariate analysis including clinical/ histopathological parameters, presence of alterations at
Wnt/B-catenin alone or with p53/Rb signaling and “high” promoter methylation status remained significant (P = 0.026, HR = 1.39, 95% CI = 1.04 to 1.87 and P = 0.003, HR = 2.03, 95% CI = 1.27 to 3.25, respectively). We then combined genetic items in a molecular score as follows: number of somatic mutations (0 to 1 = 0 points, >1 = 1 point), alterations in the Wnt/B-catenin and p53/Rb pathways (none = 0 points, only Wnt/B-catenin = 1 point, Wnt/B-catenin + p53/Rb = 2 points), and promoter region methylation pattern (≤25% = 0 points, >25% = 1 point) (overall points 0 to 4). This allowed us to separate four groups with PFS as end point: score 0 (n = 35, median PFS = 36 months), score 1 (n = 30, median PFS = 9 months), score 2 (n = 22, median PFS = 6 months), and score 3 to 4 (n =20, median PFS = 4 months) (P < 0.0001, x2 = 34.4; for definition, see Fig. 3C).
By merging mGRAS and molecular score into a combined (COMBI) score, we obtained a further im- provement in the progression risk stratification. In par- ticular, we better distinguished a group of patients with a really favorable prognosis (median PFS = 54 months) and also three groups with good (median PFS = 13 months),
A)
B)
C)
GYWU1267
GYWU1267
GYWU1267
chr1
chr3
chr5
chr7
chr9
chr11
chr13
chr15
chr17
chr21
chrX
4
chr2
chr19
chr4
chrã
chr8
chr10
chr12
chr14
chris
chr18
chr2
4
.
chr22
:
2
:
:
2
Log2(FC)
.
.
Log2(FC)
0
O
-2
-2
$
-4
-4
68.140
8.14
58.142
41 58
5 58.143
8.144 5
44 58.145 58.
58.146 58.14
Pos [10^6]
D)
E)
F)
GYWU699
GYWU699
GYWU699
4
chr1
chr3
chr5
chr7
chỉ9
chr11
cher13
chr15
chr21
chr2
chrX
chr4
chru
chr8
.
chr10
chr12
chr17
chr19
chr14
chri6
chr18
cbr20
4
.
chr22
:
2
·
2
Log2(FC)
Log2(FC)
0
0
-2
-2
-4
-4
58.140
58.141
142 58.143 58.
58.142
58.144
58.145
45
58.146 58.147
Pos [10^6]
G)
H)
1)
GYWU949
GYWU949
GYWU949
chr1
chra
chro
chry
chry
chr11
chr13
chr15
chr17. .
ch/19
chr21 chrX
4
:
chr2
:
chr4
.
chr6
chrB
chr10
chr12
chr14.
chr18
chr18
chr
B
.
chr22
4
·
2
2
Log2(FC)
Log2(FC)
0
-2
-2
i
-4
:
-4
58.140 58.141 58.142 58.143 58.144 58.145 58.148 58.147
Pos [10^6]
intermediate (median PFS = 6 months), and poor (median PFS = 3 months) prognosis (P < 0.0001, x2 = 68.6; for definition, see Fig. 3D). When we tested the superiority of COMBI with respect to mGRAS score by discriminating patients with the best clinical out- come (at least 24 months free of disease progression), COMBI score showed a better prognostic perfor- mance, proven by superior specificity (58.6% vs 31.0%) and accuracy (83.3% vs 74.5%). Moreover, the area under the receiving operating characteristic curve was higher for COMBI than for mGRAS score (0.872, 95% CI = 0.800 to 0.943 vs 0.780, 95% CI = 0.689 to 0.871) (3).
A heatmap sorted for prognosis including mGRAS score, molecular parameters, and COMBI score is shown in Fig. 4.
We then decided to compare the prognostic power of mGRAS and COMBI score evaluating the disease-free survival (DFS) in those 74 patients with ACC who were successfully operated (R0). In this subgroup, only COMBI score was able to identify a category of patients with an extremely longer DFS: median DFS for COMBI 0 to 2 (n = 23) was 243 months, COMBI 3 to 4 (n = 30)
was 13 months, COMBI 5 to 7 (n = 18) was 5.5 months, and COMBI 8 to 13 (n = 3) was 3 months [P < 0.0001, x2 = 50.98; see (3)].
Prediction of response to therapy
In patients treated with adjuvant mitotane (n = 39), a low COMBI score (0 to 2) was slightly more powerful to predict a longer DFS than a low mGRAS (0 to 1) (P = 0.0001, x2 = 21.5 vs P = 0.0058, x2 = 12.5). However, similar results were obtained considering patients with superimposable disease stages who did not receive ad- juvant mitotane (n = 49) (COMBI score: P = 0.0001, x2 = 27.5; mGRAS: P = 0.0008, x2 = 16.8), thus suggesting no specific relationship between molecular alterations and response to mitotane.
In patients with advanced ACC, none of the single molecular events showed a significant predictive role for response to mitotane monotherapy (n = 34), etoposide- doxorubicin-cisplatin (n = 52), gemcitabine plus capecitabine (n = 36), and/or streptozotocin (n = 44). These analyses were performed by considering both objective response to the investigated drugs and time to progression during treatment.
A)
B)
Percent progression-free survival
Percent progression free survival
P<0.0001
100
Chi-square=35.6
100
P<0.0001
Chi-square=49.0
75
75
0 -…
0- ---- ,
50
50
25
25
0
0
0
12
24
36
48
60
72
0
12
24
36
48
60
72
Time (months)
Time (months)
ENSAT tumor stage 1 (n=6)
ENSAT tumor stage 3 (n=28)
m GRAS score 0-1 (n=16)
m GRAS score 4-5 (n=32)
ENSAT tumor stage 2 (n=50)
ENSAT tumor stage 4 (n=23)
m GRAS score 2-3 (n=44)
m GRAS score 6-9 (n=15)
C)
D)
Percent progression -free survival
Percent progression free survival
P<0.0001
P<0.0001
100
Chi-square=34.4
100
Chi-square=68.6
75
75
50
50
0 0- - - - -
25
25
0
0
0
12
24
36
48
60
72
0
12
24
36
48
60
72
Time (months)
Time (months)
molecular score 0 (n=35)
molecular score 2 (n=22)
COMBI score 0-2 (n=24)
COMBI score 5-7 (n=22)
molecular score 1 (n=30)
molecular score 3-4 (n=20)
COMBI score 3-4 (n=32)
COMBI score 8-13 (n=19)
Actionable molecular alterations
Having chosen an NGS panel that includes several known pharmacologically targetable genetic alterations allowed us to directly look for their presence in ACC. According to the level of evidence (OncoKB website), we found at least one alteration in a drug-targetable gene in 64 of our 107 patients. The list and specifics of 17 ac- tionable genetic alterations are reported in Table 2. The most interesting ones are CN gains at gene CDK4 (43% of cases) that are accessible by different CDK4/6 in- hibitors already approved for other types of solid tumors. Moreover, recurrent alterations at NOTCH1, targeted by y secretase inhibitors; NF1, targeted by MEK in- hibitors; or MDM2, targeted by MDM2 inhibitors, were recognized. Mutations in other known druggable genes, such as those coding for receptor tyrosine kinases (EGFR, KIT, and RET), members of the DNA repair system (ATM, BRCA1, and BRCA2), PTCH1, and TSC1/TSC2, were detected in a small percentage of
samples (<3%). In two ACC samples, we identified the well-known Val600Gly activating mutation in the gene BRAF, which is found in ~50% of papillary thyroid carcinomas and is directly actioned by BRAF and/or MEK inhibitors. Finally, mutations and/or CN losses were also observed in MMR genes MLH1, MSH2, and MSH6, which are associated with response to immune checkpoint inhibitors such as PD1/PDL1 inhibitors.
Discussion
The current study represents the largest study combining targeted NGS and methylation analysis on ACC samples (n = 117) using FFPE tissue specimens that are easily obtainable during routine histopathological workup. Our results clearly demonstrate that these analyses are feasi- ble on FFPE material. Furthermore, we propose, to our knowledge, a new combined histological, clinical, and molecular score that improves the prognostic stratification
| Prognosis | |
| mGRAS Score | |
| COMBI Score | |
| Number of Mutations | |
| Wnt/p53 Pathway | |
| Methylation | |
| CNA Pattern | |
| BUB1B-PINK1 |
Legend:
| Prognosis: | bad | intermediate | good | best | not applicable |
|---|---|---|---|---|---|
| mGRAS Score: | 0-1 | 2-3 | 4-5 | 6-8 | |
| COMBI Score: | 0-2 | 3-4 | 5-7 | >7 | |
| Number of Mutations: | 0-1 | 2-4 | >4 | ||
| Wnt/p53 Pathway: | none | Wnt only | Wnt and p53 | ||
| Methylation: | <= 25 % | > 25 % | |||
| CNA Pattern: | noisy | chromosomal/quiet | |||
| BUB1B-PINK1: | <= 6.3 | >6.3 | not available | ||
in this rare disease (COMBI score). Finally, we identify actionable molecular events in >50% of patients.
Interestingly, we could evaluate the genetic profile of consecutive tumors from 10 patients. In these cases, we found a good concordance between primary and re- current tumors in terms of both TMB and mutated genes, similar to what is described for other cancer types (37). Thus, we considered only the first available tumor sample for each single patient (n = 107). Overall, we confirmed the presence of frequent ACC-associated alterations (Fig. 1A). Notably, we also detected in a smaller per- centage of cases alterations previously not clearly asso- ciated with ACC (>2.5%; i.e., mutations at NOTCH1, CIC, and BRCA1/2; amplifications in STK11 and GNA11; and deletions in TNFRSF14 and SMARCB1). In terms of signaling pathways, the most frequently in- volved were p53/Rb and Wnt/B-catenin, as expected. In 22 samples (20.6%), we observed alterations in both signaling pathways, representing an important negative prognostic marker-a rate that was already reported in literature (8, 9). In the group with the worst prognosis, three patients (2.8%) with alterations in CTNNB1 and TP53 were observed. Although Ragazzon et al. found
alterations in CTNNB1 and TP53 mutually exclusive (38), a small number of patients in the cohort of Assié et al. (8) and Zheng et al. (9) had variants in both genes, thus also supporting our data. We also found alterations in genes involved in chromatin remodeling, as expected (8, 9, 31, 39). More surprisingly, we also observed recurrent genetic alterations affecting members of the MMR (i.e., MLH1, MSH2, MSH6) or homologous recombination DNA re- pair system (i.e., ATM, BRCA1, BRCA2).
Concerning the prognostic role of molecular markers, we could confirm in our FFPE series the impact of mo- lecular markers already proposed in studies on fresh-frozen material (8, 9, 31). However, the investigation of the CN pattern was not easily achievable starting from targeted analysis in FFPE material. Similarly, the isolation of high- quality RNA from FFPE tissue was successful in only 32.5% of samples, allowing investigation of mRNA ex- pression in only a subset of patients. Therefore, we ex- cluded these markers from further analysis. A simplified molecular prognostic score was then devised that includes mutational load, alterations in the p53/Rb and Wnt/ ß-catenin pathway, and “high” promoter methylation status. However, importantly, only by merging molecular
| Gene Symbol | Description | Type of Observed Alteration | % Samples | Potential Targeted Therapy | Level of Evidenceª |
|---|---|---|---|---|---|
| DNA level | |||||
| CDK4 | Cyclin-dependent | CN gains | 43 | CDK4/6 inhibitor (palbociclib/ | 2A (liposarcoma) |
| kinase | Missense mutation | 1.8 | abemaciclib/ribociclib) | ||
| NOTCH1 | NOTCH signaling | CN gains | 20.5 | y secretase inhibitor | 4 (all tumors) |
| Missense mutation | 4.7 | (PF-03084014) | |||
| NF1 | RAS/MAPK regulation | Del/Dup or missense mutation | 11.2 | MEK inhibitor (trametinib/ cobimetinib) | 4 (glioblastoma / melanoma)b |
| MDM2 | P53 pathway | CN gain | 7.0 | MDM2 inhibitors (DS-3032b, RG7112) | 3A (liposarcoma) |
| EGFR | Receptor tyrosine kinase | Missense mutation | 2.8 | TKI (afatinib/erlotinib/gefitinib) | 1 (NSCLC)6 |
| BRCA1 | DNA repair system | Del or missense mutation | 2.8 | PARP inhibitor (rucaparib/ olaparib/nirapanib) | 1-2A (ovary cancer)b |
| BRCA2 | Missense mutation | 2.8 | (synthetic lethality) | ||
| ATM | DNA repair system | Missense mutation or delins | 1.8 | PARP inhibitor (olaparib) (synthetic lethality) | 4 (prostate cancer)b |
| BRAF | Ser/thr kinase | Missense mutation | 1.8 | BRAF inhibitor (vemurafenib/ dabrafenib) | 1 (cutaneous melanoma) |
| MEK inhibitor (trametinib/ cobimetinib) | 2A (NCSLC) | ||||
| 4 (thyroid cancer) | |||||
| PTCH1 | Sonic hedgehog receptor | Missense mutation | 1.8 | Hedgehog inhibitor (sonidegib) | 3A (skin cancer)b |
| TSC1 | mTOR pathway | Del (frameshift) | 1.8 | mTOR inhibitor (everolimus) | 2A (CNS + renal cancer)b |
| TSC2 | Missense mutation Missense mutation | 0.9 | |||
| KIT | Receptor tyrosine kinase | Missense mutation | 0.9 | TKI (imatinib/sunitinib) | 1 (GIST)b 2A (melanoma) |
| RET | Receptor tyrosine kinase | Missense mutation | 0.9 | TKI (cabozantinib) | 2A (NSCLC)b |
| ESR1 | Estrogen receptor | Missense mutation | 0.9 | AZD9496 (fulvestrant) | 3A (breast cancer)b |
| EZH2 | Histone N-methyl- transferase | Nonsense mutation | 0.9 | GSK126 (tazemetostat) | 4 (B-cell lymphoma)b |
Abbreviations: CNS, central nervous system; GIST, gastrointestinal stromal tumor; NCSLC, non-small cell lung cancer; PARP, poly (ADP-ribose) poly- merase; TKI, tyrosine kinase inhibitor.
ªEvidence by OncoKB website (21): level 1 = US Food and Drug Administration-approved biomarker; level 2A, standard care biomarker in this indication; level 2B, standard care biomarker in another indication; level 3A, predictive biomarker according to clinical evidence in this indication; level 3B, predictive biomarker according to clinical evidence in another indication; level 4, predictive biomarker according to biological evidence. bNot the same molecular alteration.
alterations with clinical/histopathological parameters in- cluded in mGRAS into a COMBI score did we obtain the best discrimination among patients with ACC with dif- ferent prognostication. The COMBI score was particularly supportive to identify patients with an extremely favorable clinical outcome, showing the best predictive accuracy for discriminating patients without disease recurrence/progress within the first 24 months after primary surgery compared with the mGRAS score. The superiority of the COMBI score was even more evident when considering the capa- bility to predict DFS in patients successfully operated. These findings might play a key role in clinical practice, helping to better select patients who do not need aggressive treatment, thus sparing unnecessary side effects to patients and costs for the community.
A targeted approach to molecular analysis has been recently proposed by Garinet et al. (40), who validated
targeted NGS for simultaneously calling mutations, chro- mosome alterations, and DNA methylation status. Such analysis might have clinical benefits but still needs to be validated in FFPE material. Considering other cancer types, genetic analysis by targeted NGS and methylation analysis by pyrosequencing have been performed in FFPE tumor specimens, obtaining good results (41-43). Nevertheless, this kind of approach had not been tested in ACC samples until now. In general, it is now the task to prove that proposed molecular-driven scores are clinically helpful to guide clinicians in patient care. To this end, only a mul- ticenter, prospective, and randomized trial will provide reliable answers, but the international ACC community seems to be well connected to perform such effort.
Furthermore, we intended to investigate the potential predictive role of molecular alterations for response to systemic chemotherapies. However, none of the
evaluated alterations were associated with the response to any standard pharmacological therapy in ACC. This might have different explanations, including the hetero- geneity of treatments usually used in this kind of patients and the complexity of the molecular background of ACC.
Finally, we intended to identify potentially druggable molecular events. A similar approach has been used in a few previous studies in a small series of patients (up to 40) demonstrating the presence of potentially actionable ge- nomic alterations in a subset of ACC (19, 20). In our study, we concentrated on molecular events targeted by drugs already available for solid tumors (OncoKB). Based on our analysis, the most promising candidate is the gene CDK4. Specifically, CN gains at the CDK4 locus already have been reported in the literature on ACC (8, 19, 20), but we observed them in an even higher percentage of cases (>40%). These alterations were confirmed with FISH analysis. Our findings may be clinically relevant because selective CDK4/6 inhibitors palbociclib and ribociclib have been approved by US Food and Drug Administration for the treatment of breast cancer (44). Phase I to III studies are now ongoing with other CDK4/6 inhibitors in solid tumors (44). Moreover, although they have not yet been tested in patients with ACC, CDK4/6 inhibitors have been shown to reduce cell viability in ACC cell lines (46, 47).
Another promising drug target is the NOTCH1 gene, which was gained in >20% of cases in the present series and in >40% in a previous study (47). The Notch pathway might represent an interesting target as it was reported to be activated in ACC (49) and can be actioned by different y secretase inhibitors or monoclonal antibodies (50). For instance, the y secretase inhibitor PF-03084014 has already been tested in a phase I study in patients with advanced solid tumors (51). The presence of CN gains at MDM2 (7% of cases) might also be considered encouraging targets as MDM2 inhibitors such as DS-3032b or RG7112 have been reported to reduce cell proliferation in MDM-amplified liposarcoma (52). An interesting therapeutic option is also represented by targeting the BRCA-related DNA repair system (altered in >7% of cases) by PARP [poly (ADP-ribose) polymerase] inhibitors (i.e., olaparib, nirapanib, and rucaparib) (53) that are approved for treatment of BRCA-mutant ovarian cancer. Moreover, mutations in targetable genes coding for receptor tyrosine kinases (EGFR, KIT, RET), members of the mTOR path- way (TSC1/2), and BRAF were detected in rare cases. Fi- nally, in 4.5% of cases, we observed mutations or CN losses in members of the MMR system (MSH2, MSH6, MLH1), which have been reported as predictive biomarkers for an- titumor effects of checkpoint PD1/PDL1 inhibitors [i.e., pembrolizumab or novolumab (54, 55)]. Our findings on actionable targets open up a new therapeutic avenue for subsets of patients with ACC.
In conclusion, our study demonstrates that molecular classification based on targeted genetic analysis is able to improve the prognostication of patients with ACC when combined with clinical/histopathological parameters. This approach paves the way to a personalized management of ACC, allowing better decisions about the need for ad- juvant therapies and/or frequency of periodical post- operative monitoring. In addition, our targeted panel can at the same time identi druggable targets. In some cases, these results may be used to select patients for clinical trials or off-label use of specific anticancer drugs. The fact that all this is possible in readily available FFPE material is a major step toward precision medicine in this rare disease.
Acknowledgments
The authors thank Ms. Martina Zink for excellent technical support and Ms. Michaela Haaf for coordinating the ENSAT Registry. This work has been carried out with the help of the Interdisciplinary Bank of Biomaterials and Data of the University Hospital of Würzburg and the Julius Maximilian University of Würzburg (IBDW). The implementation of the IBDW has been supported by the Federal Ministry for Education and Research (grant FKZ: 01EY1102).
Financial Support: This work has been supported by the Deutsche Forschungsgemeinschaft (DFG) within the CRC/ Transregio 205/1 (to M.F. and M.K.) and the Comprehensive Cancer Center Mainfranken, University Hospital of Würzburg.
Correspondence and Reprint Requests: Cristina L. Ronchi, MD, PHD, Division of Endocrinology and Diabetes, Univer- sity Hospital of Würzburg, Oberduerrbacher-Str 6, 97080 Würzburg, Germany. E-mail: Ronchi_C@ukw.de.
Disclosure Summary: The authors have nothing to disclose.
References
1. Fassnacht M, Johanssen S, Quinkler M, Bucsky P, Willenberg HS, Beuschlein F, Terzolo M, Mueller HH, Hahner S, Allolio B; German Adrenocortical Carcinoma Registry Group; European Network for the Study of Adrenal Tumors. Limited prognostic value of the 2004 International Union Against Cancer staging classification for adrenocortical carcinoma: proposal for a revised TNM classification. Cancer. 2008;115(2):243-250.
2. Else T, Kim AC, Sabolch A, Raymond VM, Kandathil A, Caoili EM, Jolly S, Miller BS, Giordano TJ, Hammer GD. Adrenocortical carcinoma. Endocr Rev. 2014;35(2):282-326.
3. Lippert J, Appenzeller S, Liang R, Sbiera S, Kircher S, Altieri B, Nanda I, Weigand I, Gehrig A, Steinhauer S, Riemens RJM, Rosenwald A, Müller CR, Kroiss M, Rost S, Fassnacht M, Ronchi CL. Targeted molecular analysis in adrenocortical carcinomas: a strategy towards improved personalized prognostication. figshare. Deposited 19 June 2018. 10.1530/endoabs.56.OC11.1. www. figshare.com/s/82bc6478458020a9c4db.
4. Erdogan I, Deutschbein T, Jurowich C, Kroiss M, Ronchi C, Quinkler M, Waldmann J, Willenberg HS, Beuschlein F, Fottner C, Klose S, Heidemeier A, Brix D, Fenske W, Hahner S, Reibetanz J, Allolio B, Fassnacht M; German Adrenocortical Carcinoma Study Group. The role of surgery in the management of recurrent ad- renocortical carcinoma. J Clin Endocrinol Metab. 2013;98(1): 181-191.
5. Margonis GA, Kim Y, Prescott JD, Tran TB, Postlewait LM, Maithel SK, Wang TS, Evans DB, Hatzaras I, Shenoy R, Phay JE, Keplinger K, Fields RC, Jin LX, Weber SM, Salem A, Sicklick JK, Gad S, Yopp AC, Mansour JC, Duh QY, Seiser N, Solorzano CC, Kiernan CM, Votanopoulos KI, Levine EA, Poultsides GA, Pawlik TM. Adrenocortical carcinoma: impact of surgical margin status on long-term outcomes. Ann Surg Oncol. 2015;23(1):134-141.
6. Beuschlein F, Weigel J, Saeger W, Kroiss M, Wild V, Daffara F, Libé R, Ardito A, Al Ghuzlan A, Quinkler M, Oßwald A, Ronchi CL, de Krijger R, Feelders RA, Waldmann J, Willenberg HS, Deutschbein T, Stell A, Reincke M, Papotti M, Baudin E, Tissier F, Haak HR, Loli P, Terzolo M, Allolio B, Müller HH, Fassnacht M. Major prognostic role of Ki67 in localized adrenocortical carci- noma after complete resection. J Clin Endocrinol Metab. 2015; 100(3):841-849.
7. Libé R, Borget I, Ronchi CL, Zaggia B, Kroiss M, Kerkhofs T, Bertherat J, Volante M, Quinkler M, Chabre O, Bala M, Tabarin A, Beuschlein F, Vezzosi D, Deutschbein T, Borson-Chazot F, Hermsen I, Stell A, Fottner C, Leboulleux S, Hahner S, Mannelli M, Berruti A, Haak H, Terzolo M, Fassnacht M, Baudin E; ENSAT Network. Prognostic factors in stage III-IV adrenocortical carci- nomas (ACC): an European Network for the Study of Adrenal Tumor (ENSAT) study. Ann Oncol. 2015;26(10):2119-2125.
8. Assié G, Letouzé E, Fassnacht M, Jouinot A, Luscap W, Barreau O, Omeiri H, Rodriguez S, Perlemoine K, René-Corail F, Elarouci N, Sbiera S, Kroiss M, Allolio B, Waldmann J, Quinkler M, Mannelli M, Mantero F, Papathomas T, De Krijger R, Tabarin A, Kerlan V, Baudin E, Tissier F, Dousset B, Groussin L, Amar L, Clauser E, Bertagna X, Ragazzon B, Beuschlein F, Libé R, de Reyniès A, Bertherat J. Integrated genomic characterization of adrenocortical carcinoma. Nat Genet. 2014;46(6):607-612.
9. Zheng S, Cherniack AD, Dewal N, Moffitt RA, Danilova L, Murray BA, Lerario AM, Else T, Knijnenburg TA, Ciriello G, Kim S, Assie G, Morozova O, Akbani R, Shih J, Hoadley KA, Choueiri TK, Waldmann J, Mete O, Robertson AG, Wu HT, Raphael BJ, Shao L, Meyerson M, Demeure MJ, Beuschlein F, Gill AJ, Sidhu SB, Almeida MQ, Fragoso MCBV, Cope LM, Kebebew E, Habra MA, Whitsett TG, Bussey KJ, Rainey WE, Asa SL, Bertherat J, Fassnacht M, Wheeler DA, Hammer GD, Giordano TJ, Verhaak RGW; Cancer Genome Atlas Research Network. Comprehensive pan- genomic characterization of adrenocortical carcinoma. Cancer Cell. 2016;29(5):723-736.
10. de Reyniès A, Assié G, Rickman DS, Tissier F, Groussin L, René- Corail F, Dousset B, Bertagna X, Clauser E, Bertherat J. Gene expression profiling reveals a new classification of adrenocortical tumors and identifies molecular predictors of malignancy and survival. J Clin Oncol. 2009;27(7):1108-1115.
11. Jouinot A, Assie G, Libe R, Fassnacht M, Papathomas T, Barreau O, de la Villeon B, Faillot S, Hamzaoui N, Neou M, Perlemoine K, Rene-Corail F, Rodriguez S, Sibony M, Tissier F, Dousset B, Sbiera S, Ronchi C, Kroiss M, Korpershoek E, de Krijger R, Waldmann J, K D, Bartsch, Quinkler M, Haissaguerre M, Tabarin A, Chabre O, Sturm N, Luconi M, Mantero F, Mannelli M, Cohen R, Kerlan V, Touraine P, Barrande G, Groussin L, Bertagna X, Baudin E, Amar L, Beuschlein F, Clauser E, Coste J, Bertherat J. DNA methylation is an independent prognostic marker of survival in adrenocortical cancer. J Clin Endocrinol Metab. 2017;102(3):923-932.
12. Fassnacht M, Dekkers O, Else T, Baudin E, Berruti A, de Krijger RR, Haak HR, Mihai R, Assie G, Terzolo M. European Society of Endocrinology Clinical Practice Guidelines on the Management of Adrenocortical Carcinoma in Adults, in collaboration with the European Network for the Study of Adrenal Tumors [published online ahead of print July 24, 2018]. Eur J Endocrinol.
13. Reidy-Lagunes DL, Lung B, Untch BR, Raj N, Hrabovsky A, Kelly C, Gerst S, Katz S, Kampel L, Chou J, Gopalan A, Saltz LB. Complete responses to mitotane in metastatic adrenocortical carcinoma-a new look at an old drug. Oncologist. 2017;22(9): 1102-1106.
14. Megerle F, Herrmann W, Schloetelburg W, Ronchi CL, Pulzer A, Quinkler M, Beuschlein F, Hahner S, Kroiss M, Fassnacht M; German ACC Study Group. Mitotane monotherapy in patients with advanced adrenocortical carcinoma. J Clin Endocrinol Metab. 2018;103(4):1686-1695.
15. Fassnacht M, Terzolo M, Allolio B, Baudin E, Haak H, Berruti A, Welin S, Schade-Brittinger C, Lacroix A, Jarzab B, Sorbye H, Torpy DJ, Stepan V, Schteingart DE, Arlt W, Kroiss M, Leboulleux S, Sperone P, Sundin A, Hermsen I, Hahner S, Willenberg HS, Tabarin A, Quinkler M, de la Fouchardière C, Schlumberger M, Mantero F, Weismann D, Beuschlein F, Gelderblom H, Wilmink H, Sender M, Edgerly M, Kenn W, Fojo T, Müller HH, Skogseid B; FIRM-ACT Study Group. Combination chemotherapy in advanced adrenocortical carcinoma. N Engl J Med. 2012;366(23):2189-2197.
16. Sperone P, Ferrero A, Daffara F, Priola A, Zaggia B, Volante M, Santini D, Vincenzi B, Badalamenti G, Intrivici C, Del Buono S, De Francia S, Kalomirakis E, Ratti R, Angeli A, Dogliotti L, Papotti M, Terzolo M, Berruti A. Gemcitabine plus metronomic 5-fluorouracil or capecitabine as a second-/third-line chemotherapy in advanced adrenocortical carcinoma: a multicenter phase II study. Endocr Relat Cancer. 2010;17(2):445-453.
17. Henning JEK, Deutschbein T, Altieri B, Steinhauer S, Kircher S, Sbiera S, Wild V, Schlötelburg W, Kroiss M, Perotti P, Rosenwald A, Berruti A, Fassnacht M, Ronchi CL. Gemcitabine-based che- motherapy in adrenocortical carcinoma: a multicenter study of efficacy and predictive factors. J Clin Endocrinol Metab. 2017; 102(11):4323-4332.
18. Costa R, Carneiro BA, Tavora F, Pai SG, Kaplan JB, Chae YK, Chandra S, Kopp PA, Giles FJ. The challenge of developmental therapeutics for adrenocortical carcinoma. Oncotarget. 2016; 7(29):46734-46749.
19. De Martino MC, Al Ghuzlan A, Aubert S, Assié G, Scoazec JY, Leboulleux S, Do Cao C, Libè R, Nozières C, Lombès M, Pattou F, Borson-Chazot F, Hescot S, Mazoyer C, Young J, Borget I, Colao A, Pivonello R, Soria JC, Bertherat J, Schlumberger M, Lacroix L, Baudin E. Molecular screening for a personalized treatment ap- proach in advanced adrenocortical cancer. J Clin Endocrinol Metab. 2013;98(10):4080-4088.
20. Ross JS, Wang K, Rand JV, Gay L, Presta MJ, Sheehan CE, Ali SM, Elvin JA, Labrecque E, Hiemstra C, Buell J, Otto GA, Yelensky R, Lipson D, Morosini D, Chmielecki J, Miller VA, Stephens PJ. Next- generation sequencing of adrenocortical carcinoma reveals new routes to targeted therapies. J Clin Pathol. 2014;67(11):968-973.
21. Fassnacht M, Berruti A, Baudin E, Demeure MJ, Gilbert J, Haak H, Kroiss M, Quinn DI, Hesseltine E, Ronchi CL, Terzolo M, Choueiri TK, Poondru S, Fleege T, Rorig R, Chen J, Stephens AW, Worden F, Hammer GD. 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(4):426-435.
22. McShane LM, Altman DG, Sauerbrei W, Taube SE, Gion M, Clark GM; Statistics Subcommittee of the NCI-EORTC Working Group on Cancer Diagnostics. REporting recommendations for tumor MARKer prognostic studies (REMARK). Nat Clin Pract Urol. 2005;2(8):416-422.
23. Chakravarty D, Gao J, Phillips SM, Kundra R, Zhang H, Wang J, Rudolph JE, Yaeger R, Soumerai T, Nissan MH, Chang MT, Chandarlapaty S, Traina TA, Paik PK, Ho AL, Hantash FM, Grupe A, Baxi SS, Callahan MK, Snyder A, Chi P, Danila D, Gounder M, Harding JJ, Hellmann MD, Iyer G, Janjigian Y, Kaley T, Levine DA, Lowery M, Omuro A, Postow MA, Rathkopf D, Shoushtari AN, Shukla N, Voss M, Paraiso E, Zehir A, Berger MF, Taylor BS, Saltz LB, Riely GJ, Ladanyi M, Hyman DM, Baselga J, Sabbatini P, Solit DB, Schultz N. OncoKB: a precision oncology knowledge base [published online ahead of print May 16, 2017]. JCO Precis Oncol.
24. Yeo G, Burge CB. Maximum entropy modeling of short sequence motifs with applications to RNA splicing signals. J Comput Biol. 2004;11(2-3):377-394.
25. Reese MG, Eeckman FH, Kulp D, Haussler D. Improved splice site detection in Genie. J Comput Biol. 1997;4(3):311-323.
26. Pertea M, Lin X, Salzberg SL. GeneSplicer: a new computational method for splice site prediction. Nucleic Acids Res. 2001;29(5): 1185-1190.
27. Desmet FO, Hamroun D, Lalande M, Collod-Béroud G, Claustres M, Béroud C. Human Splicing Finder: an online bioinformatics tool to predict splicing signals. Nucleic Acids Res. 2009;37(9):e67.
28. Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork P, Kondrashov AS, Sunyaev SR. A method and server for predicting damaging missense mutations. Nat Methods. 2010;7(4): 248-249.
29. Schwarz JM, Cooper DN, Schuelke M, Seelow D. MutationTaster2: mutation prediction for the deep-sequencing age. Nat Methods. 2014;11(4):361-362.
30. Li Q, Wang K. InterVar: clinical interpretation of genetic variants by the 2015 ACMG-AMP guidelines. Am J Hum Genet. 2017; 100(2):267-280.
31. Juhlin CC, Goh G, Healy JM, Fonseca AL, Scholl UI, Stenman A, Kunstman JW, Brown TC, Overton JD, Mane SM, Nelson-Williams C, Bäckdahl M, Suttorp AC, Haase M, Choi M, Schlessinger J, Rimm DL, Höög A, Prasad ML, Korah R, Larsson C, Lifton RP, Carling T. Whole-exome sequencing characterizes the landscape of somatic mutations and copy number alterations in adrenocortical carcinoma. J Clin Endocrinol Metab. 2015;100(3):E493-E502.
32. Rozen S, Skaletsky H. Primer3 on the WWW for general users and for biologist programmers. Methods Mol Biol. 2000;132:365-386.
33. Fujita PA, Rhead B, Zweig AS, Hinrichs AS, Karolchik D, Cline MS, Goldman M, Barber GP, Clawson H, Coelho A, Diekhans M, Dreszer TR, Giardine BM, Harte RA, Hillman-Jackson J, Hsu F, Kirkup V, Kuhn RM, Learned K, Li CH, Meyer LR, Pohl A, Raney BJ, Rosenbloom KR, Smith KE, Haussler D, Kent WJ. The UCSC Genome Browser database: update 2011. Nucleic Acids Res. 2010; 39(Database):D876-D882.
34. Li H, Durbin R. Fast and accurate short read alignment with Burrows- Wheeler transform. Bioinformatics. 2009;25(14):1754-1760.
35. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, Garimella K, Altshuler D, Gabriel S, Daly M, DePristo MA. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010;20(9):1297-1303.
36. Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 2010;26(6):841-842.
37. Meric-Bernstam F, Frampton GM, Ferrer-Lozano J, Yelensky R, Pérez-Fidalgo JA, Wang Y, Palmer GA, Ross JS, Miller VA, Su X, Eroles P, Barrera JA, Burgues O, Lluch AM, Zheng X, Sahin A, Stephens PJ, Mills GB, Cronin MT, Gonzalez-Angulo AM. Con- cordance of genomic alterations between primary and recurrent breast cancer. Mol Cancer Ther. 2014;13(5):1382-1389.
38. Ragazzon B, Libe R, Gaujoux S, Assie G, Fratticci A, Launay P, Clauser E, Bertagna X, Tissier F, De Reynies A, Bertherat J. Transcriptome analysis reveals that p53 and ß-catenin alterations occur in a group of aggressive adrenocortical cancers. Cancer Res. 2010;70:8276-8281.
39. Assié G, Jouinot A, Bertherat J. The ‘omics’ of adrenocortical tumours for personalized medicine. Nat Rev Endocrinol. 2014; 10(4):215-228.
40. Garinet S, Néou M, de La Villéon B, Faillot S, Sakat J, Da Fonseca JP, Jouinot A, Le Tourneau C, Kamal M, Luscap-Rondof W, Boeva V, Gaujoux S, Vidaud M, Pasmant E, Letourneur F, Bertherat J, Assié G. Calling chromosome alterations, DNA methylation sta- tuses, and mutations in tumors by simple targeted next-generation sequencing: a solution for transferring integrated pangenomic studies into routine practice? J Mol Diagn. 2017;19(5):776-787.
41. Tiedje V, Ting S, Herold T, Synoracki S, Latteyer S, Moeller LC, Zwanziger D, Stuschke M, Fuehrer D, Schmid KW. NGS based identification of mutational hotspots for targeted therapy in ana- plastic thyroid carcinoma. Oncotarget. 2017;8(26):42613-42620.
42. Walter RFH, Rozynek P, Casjens S, Werner R, Mairinger FD, Speel EJM, Zur Hausen A, Meier S, Wohlschlaeger J, Theegarten D, Behrens T, Schmid KW, Brüning T, Johnen G. Methylation of L1RE1, RARB, and RASSF1 function as possible biomarkers for the differential di- agnosis of lung cancer. PLoS One. 2018;13(5):e0195716.
43. Einaga N, Yoshida A, Noda H, Suemitsu M, Nakayama Y, Sakurada A, Kawaji Y, Yamaguchi H, Sasaki Y, Tokino T, Esumi M. Assessment of the quality of DNA from various formalin-fixed paraffin-embedded (FFPE) tissues and the use of this DNA for next- generation sequencing (NGS) with no artifactual mutation. PLoS One. 2017;12(5):e0176280.
44. Ramos-Esquivel A, Hernández-Steller H, Savard MF, Landaverde DU. Cyclin-dependent kinase 4/6 inhibitors as first-line treatment for post-menopausal metastatic hormone receptor-positive breast cancer patients: a systematic review and meta-analysis of phase III randomized clinical trials. Breast Cancer. 2018;25(4):479-488.
45. Deng Y, Ma G, Li W, Wang T, Zhao Y, Wu Q. CDK4/6 inhibitors in combination with hormone therapy for HR+/HER2- advanced breast cancer: a systematic review and meta-analysis of randomized controlled trials [published online ahead of print May 4, 2018]. Clin Breast Cancer.
46. Hadjadj D, Kim SJ, Denecker T, Ben Driss L, Cadoret JC, Maric C, Baldacci G, Fauchereau F. A hypothesis-driven approach identifies CDK4 and CDK6 inhibitors as candidate drugs for treatments of ad- renocortical carcinomas. Aging (Albany NY). 2017;9(12):2695-2716.
47. Fiorentini C, Fragni M, Tiberio GAM, Galli D, Roca E, Salvi V, Bosisio D, Missale C, Terzolo M, Memo M, Berruti A, Sigala S. Palbociclib inhibits proliferation of human adrenocortical tumor cells. Endocrine. 2017;59(1):213-217.
48. Ronchi CL, Sbiera S, Leich E, Henzel K, Rosenwald A, Allolio B, Fassnacht M. Single nucleotide polymorphism array profiling of adrenocortical tumors-evidence for an adenoma carcinoma se- quence? PLoS One. 2013;8(9):e73959.
49. Ronchi CL, Sbiera S, Altieri B, Steinhauer S, Wild V, Bekteshi M, Kroiss M, Fassnacht M, Allolio B. Notch1 pathway in adreno- cortical carcinomas: correlations with clinical outcome. Endocr Relat Cancer. 2015;22(4):531-543.
50. Lamy M, Ferreira A, Dias JS, Braga S, Silva G, Barbas A. Notch-out for breast cancer therapies. N Biotechnol. 2017;39(Pt B):215-221.
51. Messersmith WA, Shapiro GI, Cleary JM, Jimeno A, Dasari A, Huang B, Shaik MN, Cesari R, Zheng X, Reynolds JM, English PA, McLachlan KR, Kern KA, LoRusso PM. A phase I, dose-finding study in patients with advanced solid malignancies of the oral y-secretase inhibitor PF-03084014. Clin Cancer Res. 2014;21(1): 60-67.
52. Ray-Coquard I, Blay JY, Italiano A, Le Cesne A, Penel N, Zhi J, Heil F, Rueger R, Graves B, Ding M, Geho D, Middleton SA, Vassilev LT, Nichols GL, Bui BN. Effect of the MDM2 antagonist RG7112 on the P53 pathway in patients with MDM2-amplified, well-differentiated or dedifferentiated liposarcoma: an exploratory proof-of-mechanism study. Lancet Oncol. 2012;13(11):1133-1140.
53. Dedes KJ, Wilkerson PM, Wetterskog D, Weigelt B, Ashworth A, Reis-Filho JS. Synthetic lethality of PARP inhibition in cancers lacking BRCA1 and BRCA2 mutations. Cell Cycle. 2014;10(8): 1192-1199.
54. Le DT, Uram JN, Wang H, Bartlett BR, Kemberling H, Eyring AD, Skora AD, Luber BS, Azad NS, Laheru D, Biedrzycki B, Donehower RC, Zaheer A, Fisher GA, Crocenzi TS, Lee JJ, Duffy SM, Goldberg RM, de la Chapelle A, Koshiji M, Bhaijee F, Huebner T, Hruban RH, Wood LD, Cuka N, Pardoll DM, Papadopoulos N, Kinzler KW, Zhou S, Cornish TC, Taube JM, Anders RA, Eshleman JR, Vogelstein B, Diaz LA Jr. PD-1 blockade in tumors with mismatch-repair deficiency. N Engl J Med. 2015; 372(26):2509-2520.
55. Feng YC, Ji WL, Yue N, Huang YC, Ma XM. The relationship between the PD-1/PD-L1 pathway and DNA mismatch repair in cervical cancer and its clinical significance. Cancer Manag Res. 2018;10:105.