Chromosome 19 amplification correlates with advanced disease in adrenocortical carcinoma
Jill C. Rubinstein, MD, PhD,ª Taylor C. Brown, MD,a Gerald Goh, PhD,b,+ C. Christofer Juhlin, MD, PhD,“ Adam Stenman, MD,“ Reju Korah, PhD,ª and Tobias Carling, MD, PhD,a New Haven, CT, and Stockholm, Sweden
Background. Familial syndromes with specific genetic drivers account for a subset of adrenocortical carcinomas (ACCs), but the genomic underpinnings of sporadic cases remain poorly understood. Recent advances in copy number variation (CNV) prediction from exome sequencing are facilitating exploration of genomic rearrangements common to these carcinomas.
Methods. ACC and matched, nontumor samples underwent exome sequencing. CNVs were predicted using coverage-depth comparison. Clinicopathologic characteristics of amplification- and deletion- dominant samples were compared and pathway enrichment analysis performed for regions with significant variation.
Results. CNVs are distributed broadly across the ACC genome. Individual signatures demonstrate amplification or deletion dominance. Areas of recurrent amplification include chromosomes 5, 12, 19, and 20, whereas chromosomes 1, 10, 18, and 22 are deletion prone. Large-scale amplification of chromosome 19 occurred in 12 of 19 cases (63%), including 6 of 8 amplification-dominant samples (75% ) and was associated with stage III/IV disease (P = . 002). Genes within this amplified region are overrepresented among the adrenal hyperplasia and steroid biosynthesis pathways (P = 4.2’ and 2.5 ”, respectively). Conclusion. CNV detection via exome sequencing allows high-resolution cataloging of structural variations in ACC. Large-scale, recurrent amplifications encompassing known adrenal-specific gene pathways correlate with tumor stage. Further functional analysis of individual genes within these regions could provide mechanistic insight into specific drivers underlying pathogenesis and progression of ACC. (Surgery 2015 ;:. )
From the Department of Surgery & Yale Endocrine Neoplasia Laboratory” and the Department of Genetics,b Yale University School of Medicine, New Haven, CT; and the Department of Oncology-Pathology, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden
ADRENOCORTICAL CARCINOMA (ACC) is a rare and aggressive malignancy of the adrenal cortex that portends a poor prognosis. The incidence is be- tween 0.5 and 2 cases per million per year and
The authors have no disclosures.
T.C. is a Damon Runyon Cancer Research Foundation clinical investigator and is supported by the Damon Runyon Cancer Research Foundation. The study was also supported by the Ohse Research Foundation.
Presented at the Annual Meeting of the American Association of Endocrine Surgeons on May 17-19, 2015, Nashville, TN.
+Current Address: Bill Lyons Informatics Centre, University Col- lege London, London, UK.
Accepted for publication September 2, 2015.
Reprint requests: Tobias Carling, MD, PHD, Yale School of Med- icine, 333 Cedar Street, FMB130A, P.O. Box 208062, New Ha-
ven, CT 06520. E-mail: tobias.carling@yale.edu.
0039-6060/$ - see front matter
@ 2015 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.surg.2015.09.001
the 5-year survival rate is 16-38%.1,2 Syndromes with specific known genetic drivers of ACC tumor- igenesis include the Li-Fraumeni (TP53) and Beckwith-Wiedemann syndromes.3 The majority of cases, however, are sporadic, and the genomic underpinnings of these tumors remain poorly un- derstood. Reported recurrent somatic events include loss of heterozygosity and occasional muta- tion of TP53.4 There is also increasing evidence to support the role of mutations in the activating Wnt/B-catenin pathway.5,6
In addition to these somatic mutations, genomic instability in the form of DNA copy number variation (CNV) is found to be a common characteristic of the ACC genome, but a rare occurrence in adrenocortical adenomas. 4,7-10 Comparative genomic hybridization studies report frequent large-scale CNVs, but with a degree of variability in reported results that might partly be explained by methodologic limits. Recent
A
CNV Frequency
B
CNV Type by Sample
Deletion-Dominant
Amplification-Dominant
60
100
Frequency (% of Samples)
Amplifications
40
20
75
CNV (%)
· Amplifications
· Heterozygous Deletions
0
50
Homozygous Deletions
20
Deletions
40
25
60
0
.
2
0
A
&
7
&
9
10
11
12
13
14
15
16
17
18
19
ลอส
20
G545
G516
G540
G546
Chromosome
G551
G550
G506
G539
G543
G523
G509
G544
G542
G547
G548
G538
G549
G522
G541
advances in bioinformatics methods for identifying CNVs from high-throughput sequencing data are enabling a more unbiased approach, facilitating comprehensive exploration of informative genomic rearrangements in ACC. Juhlin et al6 recently performed exome sequencing on a cohort of ACCs, identifying recurrent TERT amplifica- tions and ZNRF3 deletions. The focus of the cur- rent study is the association of individual CNV profiles with the clinical characteristics of the ma- lignant neoplasms.
METHODS
After approval by the Yale University Institu- tional Review Board and after obtaining written informed consent from all patients, samples were selected from formalin-fixed, paraffin-embedded and fresh frozen tissues at Yale New Haven and Karolinska University Hospitals. None of the pa- tients had undergone chemotherapy or radiation therapy at the time of surgical excision. Exome capture and sequencing were performed on 41 matched ACC and adjacent, nontumor tissues from patients with sporadic ACC as described previously.6 Tumor stage was determined via the staging system of the European Network for the Study of Adrenal Tumors (ENSAT).11
Using depth of read coverage comparison be- tween matched tumor and nontumor tissues with 500-kb windows, CNVs were discernible from 19 sample pairs. GISTIC 2.0 was used to assess statis- tical significance via comparison with 100 random permutations of CNV distributions and to define peak boundaries through an arbitrated peel-off algorithm.12 CNVs were reported on the gene level and filtered to include only the top quartile of pre- dicted CNVs on a per sample basis.
Clinical characteristics including age, sex, tu- mor diameter, stage, the profile of hormone
secretion, and outcome were compared between subgroups of ACC using the Welch, 2 sample t test and analysis of variance from the R Project for Sta- tistical Computing. Pathway enrichment analysis was performed for regions with significant varia- tion using ConsensusPathDB.13,14
RESULTS
The mean read coverage was 243- and 124-fold for tumor and normal samples, respectively. There were a mean of 3,053 amplified (median, 4,469; range, 0-6,022), 3,655 heterozygously deleted (me- dian, 4,358; range, 0-6,728), and 7.7 homozygous- ly deleted (median, 4; range, 0-34) genes per sample. There was no correlation between the total number of genes with CNVs and patient age, sex, tumor stage, tumor diameter, or hormonal profile.
The genome-wide frequency of each CNV as a percent of the cohort demonstrated a broad distribution of low-frequency events by location, punctuated by specific chromosomes with more frequent alterations (Fig 1, A). Specifically, large- scale, recurrent amplifications were observed on chromosomes 5, 12, 19, and 20, and deletions on chromosomes 1, 10, 18, and 22. To assess the rela- tive frequency of each CNV type in each neoplasm, the percent of CNVs composed of amplifications, heterozygous deletions, and homozygous deletions were plotted on a per sample basis (Fig 1, B). There is a clear division into distinct patterns, whereby 9 of 19 tumors (47%) were dominated by deletions, and the remainder were amplifica- tion dominant. Homozygous deletions were rare events.
Having established a pattern wherein certain chromosomes and certain neoplasms were each prone to specific CNV types, Fig 2 combines these observations to demonstrate that recurrent, large- scale CNVs on a few specific chromosomes are
CNV Frequency by Sample
Amplifications
Deletions
G522
G541
G538
G549
G548
G542
G547
G544
G509
G523
G543
G546
G550
G545
G516
G551
G506
G540
G539
-
2
3
*
5
CƠ
A
00
₼
0
=
2
$ =
1
6
=
8
:
2
5 2
Chromosome
| All samples (n = 19), n (%) | Deletion-dominant samples (n = 9), n | Amplification-dominant samples (n = 10), n (%) | P value | |
|---|---|---|---|---|
| Age, mean (y) | 55 | 60 | 51 | .119 |
| Sex | 1.00 | |||
| Female | 12 (63) | 6 | 6 (60) | |
| Male | 7 (37) | 3 | 4 (40) | |
| Tumor diameter (cm), mean ± SD | 12.6 ± 4.6 | 12.9 ± 4.2 | 12.3 ± 5.2 | .808 |
| ENSAT stage | .087 | |||
| II | 10 (53) | 7 | 3 (30) | |
| III | 5 (26) | 2 | 3 (30) | |
| IV | 4 (21) | 0 | 4 (40) | |
| Hormonal profile | 1.00 | |||
| Nonhyperfunctional | 6 (32) | 3 | 3 (30) | |
| Cortisol | 5 (26) | 2 | 3 (30) | |
| Cortisol/aldosterone | 3 (16) | 1 | 2 (20) | |
| Aldosterone | 1 | 1 | 0 (0) | |
| Androgen | 4 (21) | 2 | 2 (20) | |
| Outcome | 1.00 | |||
| Alive, no recurrence | 6 (32) | 3 | 3 (30) | |
| Alive, recurrence | 3 (16) | 1 | 2 (20) | |
| Dead, unknown/other | 2 (11) | 1 | 1 (10) | |
| Dead, recurrence | 8 (42) | 4 | 4 (40) | |
| CNV count, mean | ||||
| Total | 6,716 | 6,154 | 7,221 | .060 |
| Amplifications | 3,053 | 263 | 5,563 | <. 0001 |
| Deletions | 3,655 | 5,881 | 1,651 | <. 0001 |
CNV, Copy number variation; ENSAT, European Network for the Study of Adrenal Tumors; SD, standard deviation.
responsible largely for amplification or deletion dominance of each neoplasm. For example, large-scale amplification of chromosome 19 was a
frequent event, occurring in 12 of 19 cases (63%), including 6 of 8 amplification-dominant samples (75%).
A
CNV Count per Chromosome: Stage II vs. III/IV (t-test)
2.5
p=0.002
Total SCNV Count
Amplifications
2.0
Deletions
-log10(p-value)
1.5
p = 0.05
1.0
0.5
0.0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Chromosome
B
Chromosome 19 CNVs
Amplifications
Deletions
G522
G542
G548
Stage IV
G547
G509
G523
G538
Stage III
G506
G516
G544
G543
G539
G540
G541
Stage II
G551
G549
G550
G546
G545
Enriched Pathway Genes: Steroid Biosynthesis Adrenal Hyperplasia
Dividing the cohort by dominant CNV type, clinical characteristics were compared but demon- strated no difference in sex, tumor diameter, hormonal profile, or outcome (Table I). There were more patients of older age in the deletion- dominant samples (mean 60 vs 52 years; P = . 119) and a correlation between amplification dominance and advanced ENSAT stage (P= . 087). The cohort was divided by ENSAT stage (II vs III/IV) and CNV burden compared between groups on a per chromosome basis. The per chro- mosome -log10(P value) for total CNV count,
amplifications, and heterozygous deletions illus- trated a correlation between chromosome 19 amplification and advanced stage disease (Fig 3, A; P = . 002). Examining chromosome 19 in closer detail revealed that 4 of 4 stage IV and 3 of 5 stage III tumors (100% and 60%, respectively) carried large-scale amplifications compared to 2 of 10 stage II neoplasms (20%; Fig 3, B).
To investigate potential functional conse- quences of chromosome 19 amplification, pathway enrichment analysis was performed on all genes on the chromosome. These genes, which are
| Gene Symbol | Samples amplified (n = 19), n (%) | Samples deleted (n = 19), n (%) | Function |
|---|---|---|---|
| COPE | 8 (42) | 3 (16) | Essential for retrograde Golgi-to-endoplasmic reticulum transport of dilysine-tagged proteins |
| CYP2A13 | 8 (42) | 3 (16) | CYP family genes: monooxygenases which catalyze many reactions involved in drug metabolism and synthesis of cholesterol, steroids and other lipids |
| CYP2A6 | 8 (42) | 3 (16) | |
| CYP2A7 | 8 (42) | 3 (16) | |
| CYP2B6 | 8 (42) | 3 (16) | |
| CYP2F1 | 8 (42) | 3 (16) | |
| CYP2S1 | 8 (42) | 3 (16) | |
| CYP4F11 | 7 (37) | 4 (21) | |
| CYP4F12 | 7 (37) | 4 (21) | |
| CYP4F2 | 7 (37) | 4 (21) | |
| CYP4F22 | 7 (37) | 4 (21) | |
| CYP4F3 | 7 (37) | 4 (21) | |
| CYP4F8 | 7 (37) | 4 (21) | |
| FDX1L | 7 (37) | 4 (21) | Essential for heme A and Fe/S protein biosynthesis. Mutation in FDX1L gene is associated with mitochondrial muscle myopathy |
| RDH8 | 7 (37) | 4 (21) | Catalyzes the reduction of all-trans-retinal to all-trans-retinol, the first reaction step of the rhodopsin regeneration pathway |
differentially altered in early versus advanced stage disease, were shown to be significantly overrepre- sented among adrenal hyperplasia and steroid biosynthesis pathways (P= 4.2-5 and 2.5-5, respec- tively). The location of the implicated genes is indicated in Fig 3, B, and individual genes are listed in Table II. The list is dominated by various members of the CYP gene family that are impor- tant in the metabolism of foreign substrates. These enzymes have been identified as potentially impor- tant to the cancer genome owing to their ability to activate carcinogens from dietary and environ- mental sources and also for their role in activating or inactivating chemotherapeutic agents.15
DISCUSSION
Previous studies employing array-based tech- niques have identified myriad CNV events in ACC, with chromosome 19 amplifications included among lists of recurrent changes.4,6,8-10 A relationship exists between total burden of CNVs and both tumor size and malignancy.7 Further association has been demonstrated between increased CNV burden and decreased survival.1º To our knowledge, however, no clinical correlations have been drawn with recur- rent, large-scale chromosome 19 alteration.
The current study used whole-exome sequencing data to characterize CNVs in ACC, identifying amplification- and deletion-prone
chromosomes as well as amplification- and deletion-prone tumors. Using these 2 observations to derive a chromosome-based comparison of ACCs by their clinical characteristics uncovered a strong correlation between chromosome 19 amplification and ENSAT stage. The observed large-scale, recurrent amplification events on chromosome 19 encompassed known adrenal- specific gene pathways, providing a short list of specific candidates for further targeted investiga- tion. Future studies will focus on assessment of individual CYP gene candidates in an expanded ACC cohort to determine the global nature of these findings. Targeted validation assays to confirm individual CNVs in this expanded cohort and real-time quantitative polymerase chain reac- tion analysis correlating CNVs with gene expres- sion levels could provide mechanistic insight into the specific drivers underlying progression of ACCs. Additionally, an expanded cohort will allow validation of alternative CNVs that seem to corre- late with disease stage (specifically chromosomes 9 and 10) and allow investigation into the potential utility of combining panels of CNVs to improve tumor classification. Ultimately, the goal is to identify biomarkers, be they individual genes or complex amalgams of variants that serve to stratify risk and assist in tailoring treatment the specific genomic profile of the individual ACC.
REFERENCES
1. Javadpour N, Woltering EA, Brennan MF. Adrenal neo- plasms. Curr Probl Surg 1980;17:1-52.
2. Allolio B, Fassnacht M. Clinical review: adrenocortical carci- noma: clinical update. J Clin Endocrinol Metab 2006;91: 2027-37.
3. Soon PS, McDonald KL, Robinson BG, et al. Molecular markers and the pathogenesis of adrenocortical cancer. Oncologist 2008;13:548-61.
4. Assié G, Letouzé E, Fassnacht M, et al. Integrated genomic characterization of adrenocortical carcinoma. Nat Genet 2014;46:607-12.
5. Gaujoux S, Grabar S, Fassnacht M, et al. 6-Catenin activa- tion is associated with specific clinical and pathologic characteristics and a poor outcome in adrenocortical carci- noma. Clin Cancer Res 2011;17:328-36.
6. Juhlin CC, Goh G, Healy JM, et al. Whole-exome sequencing characterizes the landscape of somatic mutations and copy number alterations in adrenocor- tical carcinoma. J Clin Endocrinol Metab 2015;100: E493-502.
7. Kjellman M, Kallioniemi OP, Karhu R, et al. Genetic aberra- tions in adrenocortical tumors detected using comparative genomic hybridization correlate with tumor size and malig- nancy. Cancer Res 1996;56:4219-23.
8. Dohna M, Reincke M, Mincheva A, et al. Adrenocortical car- cinoma is characterized by a high frequency of
chromosomal gains and high-level amplifications. Genes Chromosomes Cancer 2000;28:145-52.
9. Sidhu S, Marsh DJ, Theodosopoulos G, et al. Comparative genomic hybridization analysis of adrenocortical tumors. J Clin Endocrinol Metab 2002;87:3467-74.
10. Stephan EA, Chung TH, Grant CS, et al. Adrenocortical car- cinoma survival rates correlated to genomic copy number variants. Mol Cancer Ther 2008;7:425-31.
11. Fassnacht M, Johanssen S, Quinkler M, et al. Limited prog- nostic value of the 2004 International Union Against Can- cer staging classification for adrenocortical carcinoma: proposal for a revised TNM classification. Cancer 2009; 115:243-50.
12. Mermel CH, Schumacher SE, Hill B, et al. GISTIC2.0 facili- tates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome Biol 2011;12:R41.
13. Kamburov A, Wierling C, Lehrach H, et al. Consensus- PathDB-a database for integrating human functional inter- action networks. Nucleic Acids Res 2009;37:D623-8.
14. Kamburov A, Pentchev K, Galicka H, et al. Consensus- PathDB: toward a more complete picture of cell biology. Nucleic Acids Res 2011;39:D712-7.
15. Rodriguez-Antona C, Gomez A, Karlgren M, et al. Molecular genetics and epigenetics of the cytochrome P450 gene fam- ily and its relevance for cancer risk and treatment. Hum Genet 2010;127:1-17.