EJE

Targeted Next Generation Sequencing molecular profiling and its clinical application in adrenocortical cancer

Francesca Cioppi, 1,2,*+ Giulia Cantini,2,3,4,+ Tonino Ercolino,5 Massimiliano Chetta,6 Lorenzo Zanatta,2,3,5 Gabriella Nesi,7 Massimo Mannelli,2,3,4 Mario Maggi,2,3,4,5 Letizia Canu,2,3,4,5[D and Michaela Luconi2,3,4,*[D

1Department of Experimental and Clinical Medicine, University of Florence, 50139 Florence, Italy

2European Network for the Study of Adrenal Tumours (ENSAT) Centre of Excellence, University of Florence, 50139 Florence, Italy

3Department of Experimental and Clinical Biomedical Sciences, Endocrinology Section, University of Florence, 50139 Florence, Italy 4Centro di Ricerca e Innovazione sulle Patologie Surrenaliche, AOU Careggi, 50139 Florence, Italy

5Azienda Ospedaliero-Universitaria Careggi, (AOUC), 50139 Florence, Italy

6Medical Genetics, Azienda Ospedaliera di Rilievo Nazionale (A.O.R.N.) Cardarelli, Padiglione, 80131 Naples, Italy

7Department of Health Sciences, University of Florence, 50139 Florence, Italy

*Corresponding author: Department of Experimental and Clinical Medicine, University of Florence, viale Pieraccini 6, Florence 50139, Italy. Email: francesca. cioppi@unifi.it; Department of Experimental and Clinical Biomedical Sciences, University of Florence, viale Pieraccini 6, Florence 50139, Italy. Email: michaela. luconi@unifi.it

Abstract

Objective: Adrenal cortical carcinoma (ACC) is a rare malignancy with a generally poor but heterogeneous prognosis, especially depending on the tumour stage at diagnosis. Identification of somatic gene alterations combined with clinical/histopathological evaluation of the tumour can help improve prognostication. We applied a simplified targeted-Next-Generation Sequencing (NGS) panel to characterise the mutational profiles of ACCs, providing potentially relevant information for better patient management.

Design and methods: Thirty frozen tumour specimens from a local ACC series were retrospectively analysed by a custom-NGS panel (CDKN2A, CTNNB1, DAXX, MED12, NF1, PRKAR1A, RB1, TERT, TP53, ZNRF3) to detect somatic prioritised single-nucleotide variants. This cohort was integrated with 86 patients from the ACC-TCGA series bearing point-mutations in the same genes and their combinations identified by our panel. Primary endpoints of the analysis on the total cohort (113 patients) were overall survival (OS) and progression-free survival (PFS), and hazard ratio (HR) for the different alterations grouped by the signalling pathways/combinations affected.

Results: Different PFS, OS, and HR were associated to the different pathways/combinations, being NF1 + TP53 and Wnt/B-catenin + Rb/p53 combined mutations the most deleterious, with a statistical significance for progression HR which is retained only in low-(I/II) stages-NF1 + TP53 combination: HR =2.96[1.01-8.69] and HR= 13.23[3.15-55.61], all and low stages, respectively; Wnt/B-catenin + Rb/p53 combined pathways: HR = 6.47[2.54-16.49] and HR = 16.24[3.87-68.00], all and low-stages, respectively.

Conclusions: A simplified targeted-NGS approach seems the best routinely applicable first step towards somatic genetic characterisation of ACC for prognostic assessment. This approach proved to be particularly promising in low-stage cases, suggesting the need for more stringent surveillance and personalised treatment.

Keywords: ACC, somatic mutational profile, routine analysis, progression-free survival, genetics, targeted-NGS, molecular oncology, ACC-TCGA

Significance

Several genes have been identified as recurrently mutated in ACC tumours, offering potential DNA-based biomarkers for clin- ical outcomes. However, tumour molecular profiling has not yet been introduced in routine management of ACC. Starting from a small monocentric local cohort then integrated with data from ACC-TCGA patients, we demonstrated that a light custom targeted-NGS panel, including only those genes clinically relevant in ACC provided prognostication ability, especially in patients diagnosed with low-risk disease on the basis of the tumour stage. These findings underline how molecularly guided patient stratification using NGS analysis in everyday clinical practice may improve individualised treatment strategies.

Introduction

Adrenal cortical carcinoma (ACC) is a rare endocrine malig- nancy originating from the adrenal cortex, with an annual worldwide incidence of .5-2:1 million1,2 and a bimodal age

distribution.3,4 Approximately 50%-60% of cases display clin- ical evidence of hormonal excess, mainly hypercortisolism.5,6 This tumour behaves aggressively and survival rate is poor, par- ticularly when it is diagnosed at advanced stages.5,7 Prognosis is heterogeneous and pathological stage proves to be the main prognostic factor7 along with the proliferation index Ki67,8,9

+ F.C. and G.C. contributed equally.

possibly implemented by the S-GRAS score, which in addition to Ki67 and stage, includes age, tumour resection status, and symptoms as independent prognostic factors.1º Cortisol excess11 and Weiss score parameter prioritisation12 can be con- sidered additional predictive factors. Surgery is the only cura- tive treatment, although local recurrence and metastases are common, even after resection with negative margins.5 In the ad- juvant setting, mitotane monotherapy in high-risk patients (stage III and Ki67> 10%),13 or with etoposide-doxorubicin- cisplatin (EDP) in unresectable cases is recommended.14,15

ACC molecular classification on the basis of somatic altera- tions in the cancer genome may integrate clinico-pathological prognosticators, and help predict individual outcomes while supporting the development of more effective and personalised ther- apies. In recent years, understanding of ACC genetics has improved, leading to the identification of driver genes involved in the pathogen- esis of this malignancy. Two large pan-genomic studies have shown several recurrently altered genes in the tumour of ACC patients, in- cluding CDKN2A, CTNNB1, DAXX, MED12, MEN1, NF1, PRKAR1A, RB1, TERT, TP53, and ZNRF316,17 validated as the major ACC driver genes in omics-based independent studies, inte- grating expression, epigenetic, miRNA, chromosomal, copy- number variation (CNV) profiles.18,19 Since genome-wide analysis is time-consuming, expensive, and generates a sizable quantity of complex data, a targeted-Next-Generation Sequencing (NGS) ap- proach is advisable. Targeted-sequencing is now an important rou- tine technique in both clinical and research settings, offering advantages (acceptable turnaround times, low costs, fewer compu- tational burden, high confidence and accuracy20).

Here, we assessed the ability of a custom targeted-NGS pan- el, consisting of 10 of the most frequently mutated genes in ACC and applied to a small local cohort of ACC patients, to identify the mutation profiles. Integration of the local cohort with data from ACC-TCGA allowed the evaluation of the identified point mutations in a prognostic factor analysis, in- creasing the prognostication power, in particular in low-stage patients, generally considered at low risk of survival.

Materials and methods

Local cohort patients

This study was designed and conducted in accordance with the Declaration of Helsinki. We retrospectively analysed a local series of 30 conventional primary ACCs operated at Careggi University Hospital (European Network for the Study of Adrenal Tumours-ENSAT Centre of Excellence), between 2000 and 2021. A formal sample size calculation was not performed due to the rarity of the tumour. All suspected lesions underwent laparo- scopic surgery according to ENSAT ACC Guidelines.5 The study was approved by the local Ethical Committee (Prot. 2011/ 0020149), and recruited patients gave their written informed con- sent. At surgery, resected specimens were formalin-fixed and paraffin-embedded (FFPE) or snap frozen. Blood samples were drawn before surgery for Sanger sequencing of variants on germline DNA. The characteristics of the local series are indicated in Table 1.

Pathological analysis of ACC samples from the local cohort

Histological diagnosis was carried out by two independent ref- erence pathologists on the tumour tissue removed at surgery. Tumour specimens were evaluated according to the Weiss scoring system.21 The Ki67 labelling-index (LI) was estimated

using the anti-human Ki67 antibody (1:40 dilution, MIB-1, Dako, CA, USA).22 Ki67-positive nuclei were counted in 1000 tumour cells, and Ki67-LI expressed as labelled-cell per- centage. Tumour stage was assessed according to ENSAT classification.23 Tumour margin status is expressed as no re- sidual tumour (R0), microscopic residual tumour (R1), macro- scopic residual tumour (R2), uncertain resection status (Rx).

DNA extraction from ACC specimens and blood samples

Patient genomic DNA was extracted from the local cohort of 30 frozen tumour specimens and blood samples using DNeasy Blood & Tissue Kit (Qiagen, Germany), according to the man- ufacturer’s instructions. DNA quality and quantity were measured by the Qubit ds assay (Thermo Fisher Scientific, MA, USA).24

Gene panel design and sequencing

A novel custom targeted-panel of 10 genes (Table 2), previous- ly identified as driver genes for ACC,15,16 was created with SureDesign software (Agilent Technologies, UK), encompass- ing a targeted region of 115.17 kpb utilising 891 amplicons with mean sequence coverage = 98.53 at 20x priori coverage of 99.41%, on the basis of our expertise in designing targeted-NGS panels.27,28 Library amplification and variant filtering are detailed in Data. The SNVs obtained were manu- ally filtered as in Figure 1A.29

Protein folding in silico analysis

Modifications in protein tertiary structure were analysed using Phyre2 software (Protein Homology/analogY Recognition Engine V 2.0 Server; www.sbg.bio.ic.ac.uk/phyre2/) to assess the potential functional effect of VUS observed in TERT, ZNRF3, PRKAR1A and RB1 genes. The 3D-structures of wild type and mutant proteins generated by Phyre2 were com- pared and displayed with Chem3D 20.1.1.125 (Cambridge Software, PerkinElmer, Inc., Massachusetts, USA).

ACC-TCGA cohort

From the ACC-TCGA cohort,17 we selected 86 conventional ACCs patients (https://www.cbioportal.org,30) including only those patients with or without point mutations in those genes found altered in more than one patient in the local cohort (CTNNB1, ZNRF3, NF1, TP53, RB1 genes). We limited our query to CTNNB1, ZNRF3, NF1, TP53, RB1 genes; TERT was not considered since the alterations reported in the ACC-TCGA were promoter amplification instead of point mu- tations in the encoding sequence. We also excluded PRKAR1A as low represented and never alone. All these exclusion criteria have been adopted in order to limit any variability between the two cohorts in terms of candidate point mutations and their combinations. Clinical and pathological characteristics ex- tracted from https://www.cbioportal.org and chemotherapy information derived from https://portal.gdc.cancer.gov/ exploration are listed in Table 3.

Statistical analysis

Statistical analysis was performed with SPSS28.0 for Windows (Chicago, IL, USA). Continuous variables with normal distri- bution were presented as mean ± standard deviation (SD) and nonparametric variables as median [interquartile range

Table 1. Clinical characteristics of the local monocentric cohort of 30 ACC patients.
ACC Local CohortN=30WTN=13Patients with Mutation/VariantN=17
Age (years)50± 1130/3050±1013/1351 ± 1217/17
Male sex (%)11 (37)30/307 (54)13/134 (24)17/17
Secretion (%)28/3013/1315/17
Cortisol12 (40)7 (54)5 (29)
Androgens6 (20)2 (15)4 (24)
Aldosterone1 (3)0 (0)1 (6)
NS9 (30)4 (31)5 (29)
n.a.2 (7)0 (0)2 (12)
ENSAT Stage (%)29/3013/1316/17
I3 (10)2 (15)1 (6)
II12 (40)3 (23)9 (53)
III11 (37)6 (46)5 (29)
IV3 (10)2 (15)1 (6)
n.a.1 (3)0 (0)1 (6)
Tumour size-diameter (cm)9.7±4.929/3010.2±4.913/139.3±5.016/17
Total Weiss score6±230/306±213/136±117/17
Ki67 LI18±1830/3016±1113/1320±2217/17
Resection status (%)30/3013/1317/17
R018 (60)3 (23)15 (88)
R16 (20)5 (38)1 (6)
R21 (3)1 (8)0 (0)
Rx5 (17)4 (31)1 (6)
MTT28 (30)13 (13)15/17
Yes27 (90)13 (100)14 (82)
No1 (3)0 (0)1 (6)
n.a2 (7)0 (0)2 (12)
Chemotherapy (%)26/3011/1315/17
No17 (57)7 (54)10 (59)
EDP7 (23)4 (31)3 (18)
Other2 (7)0 (0)2 (12)
n.a4 (13)2 (15)2 (12)
Progression (%)29/3013/1316/17
Yes11 (37)5 (39)6 (35)
No18 (60)8 (61)10 (59)
n.a.1 (3)0 (0)1 (6)
Death (%)29/3013/1316/17
Yes6 (20)4 (31)2 (12)
No23 (77)9 (69)14 (82)
n.a.1 (3)0 (0)1 (6)
Progression Free Survival Time (months)38.0[20.5-82.0]29/3034[10-83.5]13/1338[21.5-88.2]16/17
Overall Survival Time (months)53.0[32.5-84.5]29/3062[21.5-83.5]13/1353[34.2-89.8]16/17

Data are expressed as mean + SD or median[IQR] according to their continuous normal or nonparametric distribution; categorical variables are expressed as number or percentage (%). NS: non-secreting; n.a .: not available; Ki67 LI: Ki67 Labelling Index; Tumor margin status is expressed as no residual tumor (RO), microscopic residual tumor (R1), macroscopic residual tumor (R2), uncertain resection status (Rx); MTT: Mitotane; EDP: etoposide, doxorubicin and cisplatin.

(IQR)]. Categorical variables were expressed as counts and percentages. Comparison between two groups of data was per- formed using Student’s t-test for parametrically distributed variables and Mann-Whitney U’s test for nonparametric vari- ables, while multiple comparisons were accomplished by one- way ANOVA test followed by post hoc Kruskal-Wallis’ test for nonparametric variables. For the statistical analyses, the refer- ence group is considered wild type (wt), consisting of those pa- tients in the total cohort with ACC without any of the point mutations identified by the custom targeted-NGS panel. Overall survival (OS) has been defined as the probability that a patient di- agnosed with the disease is still alive. Progression-free survival (PFS) has been defined as the probability of absence of any recur- rence or metastasis or positive lymph nodes in stages < IV, or of any recurrence or increase in the number or diameter of distant le- sions (including lymph nodes) in stage IV, according to the RECIST criteria31 and ENSAT guidelines3 by imaging (perform- ing thorax plus complete abdomen Computed Tomography, every 3 months for the first 2 years and every 6 months till 5th

year, and then annually till 10th year). Survival analysis was ac- complished using the Kaplan-Meier method, and statistically sig- nificant differences between curves were estimated by the log-rank test. A P value <. 05 was considered statistically significant and was also applied to the small samples deriving from subgrouping analysis. Hazard ratio of death from the tumour and disease pro- gression was calculated with multivariable Cox regression, includ- ing point alterations in genes and their involved pathways, as well as age as independent discrete variables, considering wt or age <50 years to have risk = 1.

Results

Local cohort

The local cohort consisted of 30 conventional ACCs, with a gender ratio of 11 males to 19 females and an average age of 50 ± 11 years at diagnosis. The distribution of ENSAT stages was similar, with 50% classified as low-stage and 47% as high-stage. Over a follow-up period of 53.0 months (IQR:

Table 2. List of the driver genes identified in the literature as associated to ACC.
GeneCytogenetic locationRefseq transcriptPathway affectedAlteration frequency in ACCReferences
CDKN2A9p21.3NM_000077p53 apoptosis/Rb1 cell cycle11%-15%16-18
CTNNB13p22.1NM_001098210Wnt/ß-catenin signalling15%-16%16,25
DAXX6p21.32NM_001141969Chromatin remodelling/maintenance6%-7%16,26
MED12Xq13.1NM_005120Chromatin remodelling/maintenance5%16
NF117q11.2NM_000267Ras-cAMP/ERK MAP kinase cascade (Ras-ERK)3%-5%17,25
PRKAR1A17q24.2NM_001276289cAMP/PKA signalling8%17
RB113q14.2NM_000321p53 apoptosis/Rb1 cell cycle7%16
TERT5p15.33NM_001193376Chromatin remodelling/maintenance6%-14%16,17
TP5317p13.1NM_000546p53 apoptosis/Rb1 cell cycle16%-21%16,17,25
ZNRF322q12.1NM_001206998Wnt/ß-catenin signalling1.9%-21%16-18,25
Figure 1. NGS genetic analysis of the local ACC cohort. Panel A: Inverted pyramid representation of bioinformatics filtering of the variants detected by targeted-NGS panel. MAF: Minor Allele Frequency in Non-Finnish Europeans based on GnomAD database; VAF: Variant Allele Frequency; Tier I: variants with strong clinical significance; Tier II: variants with potential clinical significance; Tier III: variants with unknown clinical significance. Panel B: Distribution of gene variants identified in the local ACC samples. Panel C: Mutation frequency of driver genes in the local ACC cohort of 30 patients (black columns) compared with the frequencies we calculated from the training cohort of 107 ACC patients (grey columns) in Lippert et al.

A

B

Total number of variants from 30 tumor samples (n=181)

A total of 1 variant classified as III class

PRKAR1A

≥250x read depth and VAF≥5% (n=116)

A total of 2 variants classified as III class

ZNRF3

4%

MAF≤0.01 or not available in GnomAD_NFE (n=68)

A total of 2 variants, one of which classified as III class and the other one as Il class

9%

RB1

9%

TP53

36%

A total of 8 variants classified as Il class

Exonic (non-synonymous) and splice-region variants (n=36)

A total of 3 variants classified as III class

14%

TERT

14%

14%

Tier I. II. III (n=22)

A total of 3 variants classified as Il class

NF1

CTNNB1

A total of 3 variants classified as I class

C

25,0

Somatic Mutation Frequency

20,0

15,0

10,0

5,0

0,0

TP53

CTNNB1

NF1

TERT

RB1

ZNRF3

PRKARIA

Present paper

@Lippert et al. 2023

32.5-84.5), 11 patients experienced recurrence or tumour pro- gression, and 6 died of the disease. Local series’ characteristics are detailed in Table 1.

NGS analysis

Applying the custom NGS panel to the local cohort, for 9 out of the 10 ACC driver genes investigated we identified 181 var- iants that were filtered to a total of 116 on the basis of read depth and variant allele frequency (VAF). For clinical setting

purposes, we started performing NGS on frozen tumour speci- mens to avoid any potential DNA degradation and fragmenta- tion due to fixation. However, finding the same mutations in both frozen and FFPE tumour material in one patient, who carried mutations in 3 different candidate genes, confirmed that NGS can be applied to FFPE without losing any variants (not shown).32 The overall variant filtering is illustrated in Figure 1A. Twenty-two rare variants were selected within cod- ing regions and exon-intron boundaries including splicing, missense and nonsense variants, and classified of strong/

Table 3. Clinical characteristics of the 86 patients extracted from the ACC-TCGA cohort.
ACC TCGA CohortN= 86WTN=54Patients with Mutation/ VariantN=32
Age (years)47±1686/8646 ±1754/5449 ± 1632/32
Male sex (%)30 (35)86/8620 (37)54/5410 (31)32/32
Secretion (%)6/862/544/32
Cortisol3 (4)0 (0)3 (9)
Androgens2 (2)1 (2)1 (3)
Aldosterone1 (1)1 (2)0 (0)
NS0 (0)0 (0)0 (0)
n.a.80 (93)52 (96)28 (88)
Stageª (%)84/8653/5431/32
I9 (11)6 (11)3 (9)
II44 (51)33 (61)11 (34)
III11 (13)7 (13)4 (13)
IV20 (23)7 (13)13 (41)
n.a.2 (2)1 (2)1 (3)
Tumor size diameter (cm)10.8 ±3.983/8610.5±4.053/5411.1 ±3.730/32
Total Weiss score6±247/865±229/546±218/32
KI67 LI20±1827/8613±814/5427± 2313/32
Resection Status (%)18/8611/547/32
R015 (17)11 (20)4 (13)
R13 (4)0 (0)3 (9)
R20 (0)0 (0)0 (0)
Rx0 (0)0 (0)0 (0)
n.a.68 (79)43 (80)25 (78)
Chemotherapy80/8651/5429/32
Yes58 (67)33 (61)25 (78)
No22 (26)18 (33)4 (13)
n.a.6 (7)3 (6)3 (9)
Progression (%)86/8654/5432/32
Yes48 (56)24 (44)24 (75)
No38 (44)30 (56)8 (25)
Death (%)86/8654/5432/32
Yes33 (38)14 (26)19 (59)
No53 (62)40 (74)13 (41)
Progression Free Survival Time (months)22.5[7.4-44.5]85/8631.2[14.1-61.1]53/5413.7[4.9-19.5]32/32
Overall Survival Time (months)38.5[19.6-66.4]85/8644.5[28.5-74.5]53/5422.0[15.5-45.5]32/32

Data are expressed as mean ± SD or median[IQR] according to their continuous normal or nonparametric distribution; categorical variables are expressed as number or percentage (%). Stage is reported according to the American Joint Committee on Cancer Tumour Stage Codeª. NS: non-secreting n.a .: not available; Ki67 LI: Ki67 Labelling Index; Tumour margin status is expressed as no residual tumour (RO), microscopic residual tumour (R1), macroscopic residual tumour (R2), uncertain resection status (Rx).

potential/unknown clinical significance (files) according to the literature. 33,34 To confirm the “somatic” nature of the tu- mour alterations (occurring during the life time in the tumour tissue) rather than the “germline” presentation (present in normal tissue), each variant was excluded by Sanger sequen- cing from the blood DNA extracted from leukocytes (not shown).

The majority of variants were missense (n= 16), n = 5 non- sense substitutions and 1 splicing variant (Table 2); 15/22 (68%) were classified as damaging and 7/22 (32%) of unknown significance (VUS, 31). All filtered variants in CTNNB1, NF1 and TP53 genes showed strong clinical significance (pathogenic), while those in PRKAR1A, TERT and ZNRF3 genes (n=7) were VUS. A CTNNB1 variant (c.133T > C; p.Ser45Pro) was de- tected in more than one patient, confirming its role as a mutation- al hotspot. The final variants are all reported in the Catalogue of Somatic Mutations in Cancer (COSMIC), with the exception of the NF1 splicing variant and the three TERT variants. TP53 ac- counted for 36% (n = 8/22) of all alterations, encompassing 6 missense and 2 nonsense private variants, each classified as Tier II-with potential clinical significance (Figure 1B). CTNNB1, NF1, and TERT variants totalled 14% (n = 3/22).

Gene mutation frequency was 23% for TP53, followed by CTNNB1 (17%), NF1 (10%) and TERT (10%) (Figure 1C). RB1 and ZNRF3 inactivating variants were detected in 7% of the cases, while the PRKAR1A gene was mutated in only 1 (3%) patient (Figure 1C). No interesting point mutations were identified in CDKN2A and MED12 genes. Thus, 17 (57%) tumours harboured at least one variant affecting ACC-related pathways. The mutation frequency of the 7 genes in our cohort is consistent with the data reported by Lippert and colleagues,29 with the exception of TERT (Figure 1C). Interestingly, mutational signature of the local cohort was char- acterised by a predominance of C> T, followed by T > C and C > A transitions (Table 4, Table S2).

In silico prediction of VUS

In silico analysis of the identified VUS (Table 4) revealed no im- pact of the three TERT mutations on the RNA-dependent DNA polymerase domains and on the location of the RNA-binding domain of the protein, with no alteration in the three-dimensional structure (Figure 2A). The p.Arg245Ter vari- ant contributed to the deletion of the zinc-finger domains

Table 4. List of the selected gene variants found in the cohort (n= 22).
GeneTranscriptcDNAproteinVariant typeTierACMG classificationLegacy ID (COSMIC)Number of carriers
CTNNB1NM_001098209c.95A > Gp. Asp32GlymissenseIPathogenicCOSM294171
CTNNB1NM_001098209c.133T > Cp.Ser45PromissenseIPathogenicCOSM56633
CTNNB1NM_001098209c.133T > Gp.Ser45AlamissenseIPathogenicCOSM56851
NF1NM_000267c. 1885G > Ap.Gly629ArgmissenseIIPathogenicCOSM2200891
NF1NM_000267c.4270-1G> CsplicingIIPathogenicn.a1
NF1NM_000267c.574C> Tp.Arg192TernonsenseIIPathogenicCOSM427941
PRKAR1ANM_001276289c.797C>Tp.Thr266MetmissenseIIIVUSCOSM56372271
RB1NM_000321c.2042G > Ap.Trp681TernonsenseIIPathogenicCOSM69085921
RB1NM_000321c.2650G > Ap.Glu884LysmissenseIIIVUSCOSM57868721
TERTNM_001193376c.430G > Tp. Val144LeumissenseIIIVUSn.a1
TERTNM_001193376c.237G > Tp. Glu79AspmissenseIIIVUSn.a1
TERTNM_001193376c.26C>Tp.Ala9ValmissenseIIIVUSn.a1
TP53NM_000546c.584T> Cp. Ile195ThrmissenseIIPathogenicCOSM3297431
TP53NM_000546c.824G > Tp. Cys275PhemissenseIIPathogenicCOSM60229061
TP53NM_000546c.1024C>Tp.Arg342TernonsenseIIPathogenicCOSM2200891
TP53NM_000546c.799C> Tp.Arg267TrpmissenseIIPathogenicCOSM11695381
TP53NM_000546c.376T > Gp.Tyr126AspmissenseIIPathogenicCOSM60246091
TP53NM_000546c.541C> Tp.Arg181CysmissenseIIPathogenicCOSM110901
TP53NM_000546c.1031T > Cp.Leu344PromissenseIIPathogenicCOSM440701
TP53NM_000546c.916C>Tp.Arg306TernonsenseIIPathogenicCOSM106631
ZNRF3NM_001206998c.992C>Tp.Prp331LeumissenseIIIVUSCOSM97268441
ZNRF3NM_001206998c.733C> Tp.Arg245TernonsenseIIIVUSCOSM10330891

For each variant, transcript, changes on cDNA and protein, variant type, Tier classification, ACMG classification, COSMIC Identifier (ID) and number of carriers are reported. n.a .: not available; VUS: Variant of Unknown Significance.

necessary for the negative regulation of the Wnt/ß-catenin signal- ling pathway. The p.Pro331Leu ZNRF3 change modified the lo- cation of the zinc-finger domains in relation to the cytoplasmic cadherin domain, leading to a significant protein compression from 126.69 to 119.23Å (Figure 2B). The p.Thr266Met substitu- tion in the PRKAR1A protein resulted in cyclic nucleotide- binding domain misfolding, changing the spatial distribution (Figure 2C). The p.Glu884Lys RB change altered the spatial dis- tribution of the pocket domains causing protein stirring from 71.04 to 79.30 Å, preventing Rb1 transcriptional repressor activ- ity on cell-cycle genes (Figure 2D). Therefore, the variants identi- fied in ZNRF3, PRKAR1A, RB1 genes were now be considered as likely pathogenic, while those in TERT likely benign.

Survival analysis

Somatic alterations and clinical characteristics for each patient of the local cohort are reported in Table 5 and Figure 3A. The mutated genes were clustered into main signalling pathways, the most frequently altered being Rb/p53 pathway (47%: TP53, RB1), followed by the Wnt/B-catenin pathway (41%: CTNNB1, ZNRF3), then NF1 (always present with TP53 mu- tation, 18%) and TERT (always alone, 18%) genes. The only PRKAR1A variant was associated with CTNNB1.

The preliminary survival evaluation performed by Kaplan-Meier analysis of the local ACC cohort stratified in 4 classes according to the mutated signalling pathways or presentation in combined gene mutations, displayed a statistically significant difference in the vel- ocity of tumour progression (pooled Log rank =. 012). Patients with combined NF1 + TP53 alterations showed the most rapid progression (PFS =20[3-20] months, n =3), followed by patients with mutations in the Wnt/B-catenin (PFS = 38.0[24.5-47.0] months, n=5) and Rb/p53 (PFS=51.0[22.5-101.5] months, n = 5) pathways, while patients with TERT variants exhibited the slowest tumour progression (PFS = 99[68-99] months, n = 3),

being these differences between the median PFS statistically signifi- cant (P =. 038, Kruskal-Wallis’ test). The statistical significance was retained even when Kaplan-Meier analysis was limited to the low- stage ACCs (I-II) with a log rank = . 014. Differences in OS did not reach any statistical significance (pooled Log rank = . 147), though mortality in patients with altered Wnt/B-catenin pathway was 100% by month 74.

To increase the power of the survival analysis, we extended the local cohort including data from patients of the ACC-TCGA cohort17 with conventional tumours bearing point mutations in the same genes we identified with the targeted-NGS panel. The new cohort consisted of 113 patients, with n = 46 (41%) bearing at least one pathogenic/likely pathogenic point mutation in the selected genes and their combinations, and n = 67 without any of the above point mutations (wild type: wt). ACC-TCGA char- acteristics are reported in Table 3 and Figure 3B. An additional mutation class was present only in the ACC-TCGA, consisting of combined single mutations in Rb/p53 + Wnt/B-catenin pathways.

Kaplan-Meier analysis of the total cohort confirmed the pre- liminary results of the local cohort highlighting differences in the behaviour of the survival curves (Figure 4A-C for PFS and Figure 4G-I for OS). For PFS, statistically significant differences in tumour progression were present vs wt for NF1 + TP53 com- bination (pairwise log rank =. 039) and for Rb/p53 + Wnt/ B-catenin pathway (pairwise log rank <. 001), Figure 4A, which were associated to a significant reduction in PFS vs wt (P =. 009 and P = . 005, respectively), Figure 4D. A similar survival behav- iour was appreciated when patients were grouped in low-(I/II) vs high-(III-IV) stages (Figure 4B vs 4C); the statistical significance in PFS time reduction was maintained only in the low (Figure 4E) but not in the high stages (Figure 4F), except for Wnt/ß-catenin pathway. Any mutation in Wnt/ß-catenin + Rb/p53 pathways conferred the worst progression behaviour (Figure 4A, B, D, E), also compared with mutations in the

Figure 2. In silico prediction of the tertiary structures of the protein encoded by mutated genes. Phyre2 software (Protein Homologyfold Recognition Server; www.sbg.bio.ic.ac.uk/phyre2/) was used to assess potential functional effects of the variants seen in TERT (panel A), ZNRF3 (panel B), PRKAR1A (panel C) and RB1 (panel D) genes. The amino acid variants associated with each mutation are indicated, and predicted 3D structures of mutated proteins are compared with the predicted conformation of each wt encoding gene. Specific functional domains are indicated with different colours.

TERT

A

WT

Val9

Asp79

Leu144

RNA binding domain

RVT1 domain

RVT1 domain

ZNRF3

B

WT

Leu331

Arg245ter

ZNRF3 Ecto domain

Cadherin domain

Zinc finger

126,69 Å

89,23 Å

PRKARIA

C

WT

Met266

cNMP binding

PKA R-subunit

cNMP binding

RB1

D

WT

Lys884

Pocket domain A

71,04 Å

79,30 Å

Pocket domain B

CycB domain

CycA domain

single pathway (P = . 029 vs Rb/p53 Figure 4D and P = . 027 vs Wnt/B-catenin pathways, Figure 4E). For OS curves, when considering all stages, Rb/p53 + Wnt/ß-catenin and Wnt/ B-catenin mutated pathways had the worst behaviour (Figure 4G), which was however statistically significant for OS median time (Figure 4J) only for the two combinations NF1 + TP53 (P =. 025) and Rb/p53 +Wnt/ß-catenin path- ways (P =. 024), Figure 4J. This statistical significance was also maintained for the two combinations in low-(Figure 4K)

but not in high-(Figure 4L) stages. Interestingly, while the com- bined mutation in Rb/p53 + Wnt/ß-catenin pathways had the worst prognosis for OS and PFS and in both low and high-stage patients, PFS seemed to be more affected by the combination NF1 + TP53 in low-stage patients (Figure 4B), while the Wnt/ ß-catenin pathway was crucial in high stage patients (Figure 4C).

A multivariable survival-risk analysis associated with point mutations in the different signalling pathways compared with

Table 5. Genotype and clinical features of each ACC patient in the cohort.
CaseSexAge (ys)SecretionENSAT stagesize (cm)Ki67 (%)WeissGenotypeVAF (%)AMPACMG
3M64CII7,5307CTNNB1 (NM_001098209):c.95A>G; p.Asp32Gly32,8IPathogenic
5F62NSII2,5106TERT (NM_001193376):c.430G> T; p.Val144Leu27,65IIIVUS
6M26C + ANDROII14,5157Non relevant variants
7F59NSII7,5308Non relevant variants
8M47CIII18,0158Non relevant variants
9F51ALDOIII3,0205TP53 (NM_000546):c.584T> C; p.Ile195Thr22,62IIPathogenic
10F58ANDROIII9,5307TERT (NM_001193376):c.237G> T; p.Glu79Asp19,72IIIVUS
11F58NSIII13,0908TP53 (NM_000546):c.824G> T; p.Cys275Phe17,18IIPathogenic
13F58CIII7,0403Non relevant variants
16F45ANDROII7,556Non relevant variants
17F37n.a.IIn.a.56TERT (NM_001193376):c.26C> T; p.Ala9Val13,12IIIVUS
21M61CI3,053CTNNB1 (NM_001098209):c.133T>C; p.Ser45Pro26,9IPathogenic
22F71NSIV9,0107CTNNB1 (NM_001098209):c.133T> G; p.Ser45Ala68,21IPathogenic
23F46Cn.a.7,514CTNNB1 (NM_001098209):c.133T> C; p.Ser45Pro34,85IPathogenic
24F36CIII6,7155Non relevant variants
25M54NSI4,5105Non relevant variants
26F27n.a.II9,556ZNRF3 (NM_001206998):c.992C> T; p.Pro331Leu67,93III/IVVUS
27M56CIV15,0308Non relevant variants
28M51NSIV17,0158Non relevant variants
29F62ANDROIII4,8258NF1 (NM_000267):c.1885G>A; p.Gly629Arg90,5IIPathogenic
TP53 (NM_000546):c.916C> T; p.Arg306Ter85,25IIPathogenic
ZNRF3 (NM_001206998):c.733C> T; p.Arg245Ter81,17III/IVVUS
30F54CIII13,5508RB1 (NM_000321):c.2042G> A; p.Asp681Ter50IIPathogenic
TP53 (NM_000546):c.799C> T; p.Arg267Trp9,96IIPathogenic
TP53 (NM_000546):c.1024C> T; p.Arg342Ter44,61IIPathogenic
31M46CIII15,0156Non relevant variants
32F56NSIII9,7106Non relevant variants
33M52ANDROI4,624Non relevant variants
34M38CII13,0208CTNNB1 (NM_001098209):c.133T>C; p.Ser45Pro39,38IPathogenic
PRKAR1A (NM_001276289):c.797C>T;18,14IIIVUS
p.Thr266Met
35F41ANDROII10,065RB1 (NM_000321):c.2650G> A; p.Glu884Lys5,1IIIVUS
36F48ANDROII18,056NF1 (NM_000267):c.574C> T; p.Arg192Ter40,89IIPathogenic
TP53 (NM_000546):c.541C> T; p.Arg181Cys13,3IIPathogenic
37F40NSII6,5106NF1 (NM_000267):c.4270-1G>C70,85IIPathogenic
TP53 (NM_000546):c.376T> G; p.Tyr126Asp72,35IIPathogenic
38F66CIII5,5106Non relevant variants
39M37n.a.II6,510n.a.Non relevant variants
40F52NSII19205TP53 (NM_000546):c.1031T> C; p.Leu344Pro80,45IIPathogenic

C: cortisol; ALDO: aldosterone; ANDRO: androgens; NS: non secreting; n.a .: not available; the genes interested by the variant are indicated in bold.

the wt condition and adjusted for age as a dummy variable (<50 or ≥50 years) was run, Table 6. Among the different mu- tations, NF1 + TP53 combination and Rb/p53 + Wnt/ B-catenin pathways displayed a statistically significant in- creased risk of progression compared with wt, which further increased when considering the low-stages. An even higher risk of death was associated to these two mutational groups in low stage patients. Conversely, in the high-stage, the prog- nostication relevance of these mutations was lost. Age effect has a different trend in low- vs high-stages.

Finally, the relevance of the number of point mutations for the survival outcomes (PFS and OS) was assessed with Kaplan-Meier (Figure S1) and multivariable Cox regression (Table S1). The presence of more than one pathogenic alter- ation in any genes was associated to significantly reduced PFS and OS time, even compared with having a single mutation (Figure S1) and increased HR (Table S1), but only when considering the low stages. Young age was con- firmed also in this multivariable analysis to behave as a pro- tective or detrimental factor in low or high-stages, respectively (Figure S1).

Discussion

Targeted-NGS has revolutionised medical genetic research by cutting sequencing costs while increasing the throughput, allow- ing simultaneous analysis of several genes with decreased allele dropout. Thus, multi-gene panels are becoming the standard ap- proach for the molecular analysis of solid tumours. 29,35 Recently, a large multicentre study on 194 ACC samples showed that DNA-based biomarkers, evaluated by targeted se- quencing of 160 cancer-specific genes for the training cohort as well as two smaller panels including 100 and 33 genes for the validation cohort, can improve prognostication beyond routine- ly available clinical and histopathological parameters.29 29

Here, starting from a local series of 30 ACCs, we demon- strated the clinical utility of tumour mutational analysis using a light custom targeted-NGS panel. Moreover, by incorporating data from the ACC-TCGA series, we compared the prognostic power of the different signalling pathways/combinations involv- ing those genes harbouring any point mutations. In the local ser- ies, 116 variants were identified, filtered according to gene location, and classified as somatic variants of strong/potential/

Figure 3. NGS point mutation signature and clinical profiles of mutated ACC from the local and ACC-TCGA cohorts. Heatmap describing distribution of the cancer-specific somatic point mutations found in each gene clustered according to the specific signalling pathways or combinations (upper panels) and the main clinical-pathological characteristics (lower panels) in the local cohort of n= 17 mutated patients (panel A) and in the n= 32 mutated patients from ACC-TCGA series (panel B). The colour legend is given below the heatmap. Cases are ordered by increasing tumour stages.

A

B

ACC40

ACC21

ACC03

ACC23

ACC34

ACC26

ACC35

ACC36

ACC37

ACC17

ACC05

ACC09

ACC10

ACC11

ACC29

ACC30

ACC22

TCGA-PK-A5HB

TCGA-OR-A5JE

TCGA-OR-A5JL

TCGA-OR-A5L3

TCGA-OR-A5JS

TCGA-OR-A5LE

TCGA-OR-A5LJ

TCGA-OR-A5JP

TCGA-OR-A5KZ

TCGA-OR-A5KB

TCGA-OR-A5K6

TCGA-OR-A5K9

TCGA-OR-A5KU

TCGA-OR-A5L1

TCGA-OR-A5L8

TCGA-OR-A5J2

TCGA-OR-A5LF

TCGA-OR-A5J8

TCGA-OR-A5JJ

TCGA-OR-A5KY

TCGA-OR-A5JB

TCGA-OR-A5JA

TCGA-OR-A5JG

TCGA-OR-A5K4

TCGA-OR-A5KO

TCGA-PK-A5HC

TCGA-OR-A5K5

TCGA-OR-A5J5

TCGA-OR-A5JM

TCGA-OR-A5JY

TCGA-OR-A5LB

TCGA-P6-A5OF

Point mutations

CTNNB1

ZNRF3

PRKAR1A

TP53

RB1

NF1

TERT

PATHWAYS/COMBINATIONS

ßCAT Pathway

TP53 Pathway

NF1+TP53

TP53+B CAT Pathways

Patients

characteristics

SEX

AGE

STAGE

I

I

II

II

II

II

II

II

II

II

Il

III

III

III

III

III

IV

I

I

1

II

II

II

11

II

II

11

II

=

II

III

III

III

IV

2

2

4

2

Z

2

2

Z

2

2

DEATH

PROGRESSION

Tumor

characteristics

SIZE ≥ 4

WEISS ≥ 6

☒ ☒ ☒ ☒

Ki67 ≥ 10

☒ ☒

☒ ☒

☒ ☒

☒ ☒

☒ ☒

☒ ☒

☒ ☒ ☒ ☒

☒ ☒ ☒

☒ ☒ ☒ ☒ ☒ ☒

SECRETION

☒ ☒ ☒ ☒ ☒ ☒ ☒

☒ ☒ ☒ ☒ ☒

☒ ☒ ☒ ☒ ☒

☒ ☒

Pathogenic

F

Cortisol

VUS

☒ M

☒ Androgens

Likely Benign

Aldo

< 50 ys

BCAT Pathway

≥ 50 ys

No event

TP53 Pathway

☒ PROGRESSION/DEATH

NF1+TP53

SIZE <4 cm or WEISS<6 or Ki67<10%

TP53+BCAT Pathways

SIZE ≥ 4 cm or WEISS≥ 6 or Ki67≥ 10%

☒ not available

unknown clinical significance.35 This stringent filtering limited the otherwise very high mutational burden previously found with a tar- geted approach,36 and 22 variants in 7 genes (CTNNB1, NF1, PRKAR1A, RB1, TERT, TP53 and ZNRF3) were obtained. Notably, 10% of our cases showed an allele frequency ≥70%. Given that germline conditions were excluded by blood analysis, the far higher VAFs could implicate loss-of-heterozygosity mecha- nisms in the cancer development. As previously reported, no point mutation was detected in the CDKN2A gene.16,17 In line with the very low mutation frequency (1%) described by Lippert and col- leagues, no point mutation for the MED12 gene was found in our cohort.29 The DAXX gene was removed from our analysis for insufficient sequence coverage (<50x). TP53 was the most fre- quently mutated gene (23%), with a frequency higher than in other but similar to the 20% we calculated from Lippert’s data.29 Eight pathogenic TP53 variants were identified, 6 missense and 2 nonsense, the majority affecting the DNA-binding domain,

in agreement with the literature.16,37 CTNNB1 was the second most frequently mutated gene (16%).16,17 All mutations were mis- sense and confined to exon-3, which encodes the regulatory N-terminal amino acids.38 Four of the five CTNNB1-mutated tu- mours harboured a codon 45 substitution, and the p.Ser45Pro hot- spot was the most common CTNNB1 somatic alteration, seen in three cases. NF1 and TERT alterations were observed in 10% of the cases, in accordance with Ross and colleagues.26 Mutations in the protein coding sequence of TERT gene have been assessed only recently, occurring in approximately 3% of cases.29 We iden- tified 3 missense VUS not affecting TERT protein function accord- ing to the in silico modelling performed. RB1 and ZNRF3 point alteration frequency was 6%, in line with the literature.18,29 We found only one missense PRKAR1A variant with unknown signifi- cance (3%) as recently described.29 In silico modelling of the novel VUS found, enabled us to better classify them as likely pathogenic and likely benign (Figure 3).

Figure 4. Kaplan-Meier of survival analysis in the total cohort of ACCpatients. Patients were stratified into five groups according to the pathways involving the mutated genes or the combination of the mutated genes or pathways. WT: nonmutated; BCAT pathway: Wnt/B-catenin pathway; TP53 pathway: Rb/ p53 pathway; TP53+BCAT: combination of Rb/p53 + Wnt/B-catenin pathways; NF1 + TP53: combination of mutations in NF1 and TP53 genes. Panels A-C and G-I: Kaplan-Meier survival curves for the Progression Free Survival (PFS) and Overall Survival (OS) were calculated for all tumor stages (panels A, G: pooled Log Rank P <. 001 for PFS & pooled Log Rank P< . 001 for OS, respectively), low-stage (panels B, H: pooled Log Rank P < . 001 for PFS & pooled Log Rank P <. 001 for OS, respectively) and high-stage (panels C, I: pooled Log Rank P= . 008 for PFS & pooled Log Rank P= . 018 for OS, respectively) patients. The number of patients in each group is indicated in brackets, along with the statistically significant pairwise Log Rank values for each curve vs the WT curve. Panels D-F & J-L: Box charts indicate the median value of PFS (panels D-F) and OS (panels J-L) time in each group. Statistical significance was evaluated among groups by One-way ANOVA followed by post-hoc Kruskal-Wallis' test towards WT or as indicated, for all stages (panels D, J), low-stage (panels E, K) and high-stage (panels F, L) patients; P < . 05 is considered statistically significant and indicated in bold Italics for comparison vs WT and underlined for comparison between the mutated genes and/or pathways.

A

All stages

B

Low-stages

C

High-stages

1,0

1,0

1,0

TP53 pathway (14)

NF1+TP53 (2)

0,8

0,8

0,8

PFS probability

PFS probability

BCAT pathway (10)

PFS probability

TP53 pathway (9)

0,6

WT (66)

0,6

0,6

pathway

(5)

WT (44)

0,4

0,4

0,4

BCAT pathway (17), Log Rank p=0.049

WT

(21)

0,2

NF1+TP53 (7), Log Rank p=0.039

0,2

TP53+BCAT pathways (3), Log Rank p< 0.001

0,2

BCAT pathway (7), Log-Rank p=0.008

0,0

TP53+BCAT pathways (6), Log Rank p< 0.001

NF1+TP53 (4), Log Rank p< 0.001

0,0

0,0

TP53+BCAT pathways (3)

0

25

50

75

100

125

150

175

200

0

25

50

75

100

125

150

175

200

0

20

40

60

80

100

120

D

All stages

E

Low-stages

F

High-stages

200

200

120

175

175

PFS Time (months)

150

100

0.029

PFS Time (months)

150

0.027

PFS Time (months)

125

125

80

100

100

60

75

75

0.006

0.009

50

0.009

40

0.005

50

25

25

20

0

0

WT

ßCAT pathway

TP53 pathway

NF1+ TP53

0

BCAT+TP53 pathways

WT

ßCAT pathway

TP53 pathway

NF1+ TP53

BCAT+TP53 pathways

WT

ßCAT pathway

TP53 pathway

NF1+ TP53

BCAT+TP53 pathways

G

All stages

H

Low-stages

High-stages

1,0

1,0

1,0

NF1+TP53 (2)

0,8

Survival probability

NF1+TP53 (7)

WT (44)

Survival probability

0,8

0,8

WT (66)

0,6

NF1+TP53 (4), Log Rank p=0.009

Survival probability

0,6

TP53 pathway (14)

BCAT pathway (10)

0,6

TP53 pathway (9)

0,4

0,4

0,4

BCAT pathway (17), Log Rank p=0.012

TP53 pathway (5)

WT (21)

0,2

TP53+BCAT pathways (6) Log Rank p< 0.001

0,2

0,2

TP53+BCAT pathways (3)

TP53+BCAT pathways (3), Log Rank p< 0.001

0,0

BCAT pathway (7),

0,0

0,0

Log Rank p=0.011

0

25

50

75

100

125

150

175

200

0

25

50

75

100

125

150

175

200

0

20

40

60

80

100

120

J

All stages

K

Low-stages

L

High-stages

200

200

120

175

175

OS Time (months)

150

100

OS Time (months)

150

OS Time (months)

125

125

80

100

100

60

75

0.025

0.024

75

0.034

0.031

40

50

50

25

25

20

0

WT

BCAT pathway

TP53 pathway

NF1+ TP53

0

0

BCAT+TP53 pathways

WT

ßCAT pathway

TP53 pathway

NF1+ TP53

BCAT+TP53 pathways

WT

ßCAT pathway

TP53 pathway

NF1+ TP53

₿CAT+TP53 pathways

Table 6. Multivariable survival analysis including the mutated gene pathways/combinations and age grouping in the total cohort of ACC patients.
PROGRESSIONDEATH
Total cohort HR [95%CI]Low-stages HR [95%CI]High-stages HR [95%CI]Total cohort HR [95%CI]Low-stages HR [95%CI]High-stages HR [95%CI]
Wnt/ß cateninHR = 1.97 [.98-3.98]HR = 1.16 [.38-3.52]HR = 6.24 [1.98-19.67]HR =2.413 [1.06-5.48]HR = 1.23 [.22-6.80] P =. 810HR = 3.24 [1.12-9.41]
P =. 058P =. 788P =. 002P =. 035P =. 031
Rb/p53 NF1+ TP53HR = 1.38 [.62-3.06]HR = 2.25 [.49-10.30]HR = 1.24 [.42-3.65]HR = 1.10 [.36-3.30]HR = 8.00 [.74-86.39] P =. 087HR = . 49 [.13-1.84] P= . 288
P =. 429P =. 296P =. 694P =. 870
HR = 2.96 [1.01-8.69]HR = 13.23 [3.15-55.61]HR = 2.37 [.53-10.54]HR = 54.81 [3.33-902.61]HR = 1.38 [.16-11.89]
P =. 048P <. 001P =. 259P =. 005P= . 770
Rb/p53+ Wnt/ß cateninHR = 6.47 [2.54-16.49] P <. 001HR = 16.24 [3.87-68.00] P <. 001HR =2.73 [.72-10.36] P =. 141HR = 7.44 [2.64-21.01] P <. 001HR = 39.11 [4.10-373.21] P =. 001HR = 2.84 [.76-10.67] P =. 121
AGEHR = 1.19 [.67-2.03] P =. 521HR = 2.42 [1.03-5.64] P= . 042HR =. 41 [.16-1.01] P =. 053HR = 1.94 [.98-3.81] P =. 055HR = 5.97 [1.26-28.36] P =. 025HR = 1.28 [.51-3.21] P =. 598

The hazard ratio (HR) of tumour progression and death was calculated for each mutated signalling pathways/combinations through a multivariable Cox regression compared with those patients bearing no point mutations in the analysed genes (wt). Age was considered as a discrete covariate (senior vs young age, with a cut off = 50 years). Risk analysis was considered for patients of all stages, as well as stratified in low- and high-stages. HR: hazard ratio; CI: confidence interval; (-) indicates the incapability of performing statistical analysis for the subgroup. Statistical significance corresponding to P <. 05 is indicated in bold Italics.

Among the signalling pathways affected by point mutations, the most frequently altered in the local cohort was the Rb/p53 cell cycle (TP53, RB1, 47%), followed by Wnt/ß-catenin (CTNNB1, ZNRF3, 41%) and Ras/MAPK (NF1, 18%) signal- ling. We showed here that CTNNB1 and ZNRF3 mutations were mutually exclusive and TP53 can associate with RB1 or NF1 variants. Interestingly, by including the ACC-TCGA, an associated mutation in the Rb/p53 + Wnt/B-catenin pathways emerged, as previously described.17,29,39,40 Finally, tumours

with NF1 impairment always harboured a TP53 mutation in both ACC series, implying an association between these two genes mapping on ch17. The high frequency of somatic NF1 mutations in sporadic tumours indicates that neurofibromin may play a role in cancer far beyond the predisposition evident in NF1 tumour syndrome.41 NF1 + TP53 associated mutation deserves further investigations. Notably, in the local cohort, we observed a mutational signature of the retained somatic var- iants, characterised by a predominance of C> T, followed by T>Cand C> A transitions. Suggestively, the most represented single-base substitution mutational signatures (SBS) concerning C> T transitions was SBS1 (C> T at NCG nucleotides) demon- strated as associated with age and found in all types of ancer. 42,43 SBS4, characterised by C > A transition, directly as- sociated to tobacco smoke induced DNA mutagenesis, has pre- viously been described in ACC.17,44

In the total cohort of 113 ACC patients, we were able to per- form a solid survival analysis. To the best of our knowledge, this is the first time that a direct comparison of the prognostic ability of point mutations in different signalling pathways has been undertaken. In the ACC-TCGA, 3 clusters of genes derived from integration of a multi-genomic analysis were identified dis- playing different prognostication power and associated with dif- ferent mutational pathways.17,45 However, that analysis was performed using an untargeted multi-genomic complex approach, whereas, here, we propose a light custom targeted-NGS panel eas- ily transferrable to routine analysis.

By Kaplan-Meier univariate survival analysis, we showed a different relative velocity of tumour progression and death, ac- cording to the point mutational pathways identified: the com- bined alterations in TP53 + NF1 genes, and even to a higher extent, in the Wnt/B-catenin + Rb/p53 pathways, displayed a significantly reduced time to progression and OS time, likely cooperating in exacerbating the malignant trait of the single mutations, as previously demonstrated for the latter combined pathways.16,17,29,40 This is coherent with the findings observed in ACC-genetic mouse models where the concurrent Wnt/ß-catenin mutational activation cooperates with the loss of p53 to promote murine ACC tumourigenesis and progression.46,47 The importance of identifying point muta- tional pathways in ACC patients, independently from the tu- mour stage, was further confirmed by the observed increase in HR for OS and PFS associated to all the pathways bearing pathogenic gene alterations compared with wt. HRs associated to combined mutations in the TP53 + NF1 genes and Wnt/ B-catenin + Rb/p53 pathways were statistically significant and even higher when calculated in the low-stages (I, II), while no significance was evident in the high-stage groups in a multi- variable Cox regression analysis adjusted for age. Of note, the Wnt/ß-catenin pathways appeared to have a significantly in- creased HR for OS outcome in advanced stages (III-IV). These findings suggest that in low-stages, point mutations in these genes play a relevant role for progression and OS, while in advanced stages, probably other risk factors are more

important, except for mutations in the Wnt/ß-catenin path- ways which, even when alone, is associated to 100% of death.

Alterations in the Wnt/B-catenin pathway were associated with death in all the mutated cases, while there were no deaths in those harbouring mutations in other pathways. Consequently, patients with Wnt/B-catenin mutations should receive a stringent follow-up and aggressive treatment, independently of the tu- mour stage. Of note, the multi-genomic CoC analysis identified alterations in Wnt/B-catenin and cell cycle pathways, as mainly associated with the most aggressive COC3 (Cluster-of-Cluster 3), characterised by the worst prognosis.17,45

Similar to previous studies,29,40 we also confirmed the rele- vance of the number of point mutations, associated to a stat- istically significant increase in HR for both death and progression, in our analysis, when limited to the low-stages.

Interestingly, in all multivariable Cox regression analyses, aging appeared detrimental or protective in low- or in high- stages, respectively.

Our study recognizes some limitations: (1) the panel de- signed lacks some driver genes and should be improved; (2) NGS analysis was performed on DNA extracted from frozen samples and not from FFPE; (3) the synonymous variants are excluded from our analysis, although some may have func- tional effects48,49; (4) our genetic analysis targeted point muta- tions without encompassing CNV and DNA methylation; (5) we cannot exclude that the combination of the two datasets (local and TCGA cohort) have introduced potential biases; (6) due to ACC rarity, the number of patients in the monocen- tric local cohort screened by the custom targeted-NGS panel is low and could affect the results. However, the inclusion of data from the ACC-TCGA allowed a solid survival analysis, though some information (Ki67-LI, R status) were not extract- able for the majority of those patients; (7) the small sample size introduced by sub-grouping could result in a bias in the statis- tical analysis; (8) the study is retrospective and even benefiting from the TCGA, the sample size can still be increased, there- fore validation in a larger prospective study and on routine FFPE tumour material is mandatory to confirm our prelimin- ary findings.

Conclusions

Based on our findings, parallel sequencing of multiple targeted driver genes appears to be the first step towards routine genetic characterisation of surgically treated ACC for prognostic pur- poses. Targeted-NGS analysis may improve the clinical man- agement of low-risk patients by identifying the need for more stringent surveillance and personalised treatment.

Acknowledgments

We thank the COST Action CA20122 Harmonisation for sup- portive networking.

This work is generated within the European Network for rare Endocrine Conditions (Endo-ERN) and ERN- EURACAN.

Supplementary material

Supplementary material is available at European Journal of Endocrinology online.

Funding

This work has been supported by Ministero dell’Universita’ e della Ricerca (PRIN 20222KAYY5, PNRR M4.C2.1.1, f -

Next Generation EU funded by the European Community to M.L.) and by Associazione Italiana per la Ricerca sul Cancro AIRC (grant IG2015-17691 to M.L.)

Conflict of interest: The authors declare no conflict of interest.

Authors’ contributions

Francesca Cioppi (Conceptualisation [equal], Investigation [equal], Methodology [equal], Writing-original draft [equal], Writing- review & editing [equal]), Giulia Cantini (Conceptualisation [equal], Data curation [equal], Investigation [equal], Methodology [equal], Supervision [equal], Writing-original draft [equal], Writing-review & editing [equal]), Tonino Ercolino (Formal ana- lysis [equal], Investigation [equal], Methodology [equal], Writing- review & editing [equal]), Massimiliano Chetta (Investigation [equal], Methodology [equal], Software [equal], Writing-review & editing [equal]), Lorenzo Zanatta (Formal analysis [equal], Writing-review & editing [equal]), Gabriella Nesi (Investigation [equal], Writing-review & editing [equal]), Massimo Mannelli (Conceptualisation [equal], Writing-review & editing [equal]), Mario Maggi (Writing-review & editing [equal]), Letizia Canu (Conceptualisation [equal], Data curation [equal], Writing-review & editing [equal]), and Michaela Luconi (Conceptualisation [equal], Funding acquisition [equal], Supervision [equal], Writing-original draft [equal], Writing-review & editing [equal])

References

1. Bilimoria KY, Shen WT, Elaraj D, et al. Adrenocortical carcinoma in the United States: treatment utilization and prognostic factors. Cancer. 2008;113(11):3130-3136. https://doi.org/10.1002/cncr.23886

2. Kerkhofs TMA, Verhoeven RHA, Van der Zwan JM, et al. Adrenocortical carcinoma: a population-based study on incidence and survival in The Netherlands since 1993. Eur J Cancer. 2013;49(11):2579-2586. https://doi.org/10.1016/j.ejca.2013.02.034

3. Ng L, Libertino JM. Adrenocortical carcinoma: diagnosis, evalu- ation and treatment. J Urol. 2003;169(1):5-11. https://doi.org/10. 1016/S0022-5347(05)64023-2

4. Sharma E, Dahal S, Sharma P, et al. The characteristics and trends in adre- nocortical carcinoma: a United States population based study. J Clin Med Res. 2018;10(8):636-640. https://doi.org/10.14740/jocmr3503w

5. Fassnacht M, Dekkers OM, Else T, et al. European society of endo- crinology clinical practice guidelines on the management of adreno- cortical carcinoma in adults, in collaboration with the European network for the study of adrenal tumors. Eur J Endocrinol. 2018;179(4):G1-G46. https://doi.org/10.1530/EJE-18-0608

6. Stigliano A, Chiodini I, Giordano R, et al. Management of adreno- cortical carcinoma: a consensus statement of the Italian society of endocrinology (SIE). J Endocrinol Invest. 2016;39(1):103-121. https://doi.org/10.1007/s40618-015-0349-9

7. Jouinot A, Bertherat J. MANAGEMENT OF ENDOCRINE DISEASE: adrenocortical carcinoma: differentiating the good from the poor prognosis tumors. Eur J Endocrinol. 2018;178(5): R215-R230. https://doi.org/10.1530/EJE-18-0027

8. Libé R, Borget I, Ronchi CL, et al. Prognostic factors in stage III-IV adrenocortical carcinomas (ACC): an European network for the study of adrenal tumor (ENSAT) study. Ann Oncol. 2015;26(10): 2119-2125. https://doi.org/10.1093/annonc/mdv329

9. Beuschlein F, Weigel J, Saeger W, et al. Major prognostic role of Ki67 in localized adrenocortical carcinoma after complete resec- tion. J Clin Endocrinol Metab. 2015;100(3):841-849. https://doi. org/10.1210/jc.2014-3182

10. Elhassan YS, Altieri B, Berhane S, et al. S-GRAS score for prognostic classification of adrenocortical carcinoma: an international, multi- center ENSAT study. Eur J Endocrinol. 2021;186(1):25-36. https://doi.org/10.1530/EJE-21-0510

11. Berruti A, Fassnacht M, Haak H, et al. Prognostic role of overt hy- percortisolism in completely operated patients with adrenocortical cancer. Eur Urol. 2014;65(4):832-838. https://doi.org/10.1016/j. eururo.2013.11.006

12. Luconi M, Cantini G, van Leeuwaarde RS, et al. Prognostic value of microscopic tumor necrosis in adrenal cortical carcinoma. Endocr Pathol. 2023;34(2):224-233. https://doi.org/10.1007/s12022-023- 09760-6

13. Terzolo M, Fassnacht M, Perotti P, et al. Adjuvant mitotane versus sur- veillance in low-grade, localised adrenocortical carcinoma (ADIUVO): an international, multicentre, open-label, randomised, phase 3 trial and observational study. Lancet Diabetes Endocrinol. 2023;11(10): 720-730. https://doi.org/10.1016/S2213-8587(23)00193-6

14. Fassnacht M, Terzolo M, Allolio B, et al. Combination chemotherapy in advanced adrenocortical carcinoma. N Engl J Med. 2012;366(23): 2189-2197. https://doi.org/10.1056/NEJMoa1200966

15. Terzolo M, Angeli A, Fassnacht M, et al. Adjuvant mitotane treat- ment for adrenocortical carcinoma. N Engl J Med. 2007;356(23): 2372-2380. https://doi.org/10.1056/NEJMoa063360

16. Assié G, Letouzé E, Fassnacht M, et al. Integrated genomic charac- terization of adrenocortical carcinoma. Nat Genet. 2014;46(6): 607-612. https://doi.org/10.1038/ng.2953

17. Zheng S, Cherniack AD, Dewal N, et al. Comprehensive pan- genomic characterization of adrenocortical carcinoma. Cancer Cell. 2016;29(5):723-736. https://doi.org/10.1016/j.ccell.2016.04.002

18. Maharjan R, Backman S, Åkerström T, Hellman P, Björklund P. Comprehensive analysis of CTNNB1 in adrenocortical carcinomas: identification of novel mutations and correlation to survival. Sci Rep. 2018;8(1):8610. https://doi.org/10.1038/s41598-018-26799-2

19. Lavoie JM, Csizmok V, Williamson LM, et al. Whole-genome and transcriptome analysis of advanced adrenocortical cancer high- lights multiple alterations affecting epigenome and DNA repair pathways. Cold Spring Harb Mol Case Stud. 2022;8(3):a006148. https://doi.org/10.1101/mcs.a006148

20. Pei XM, Yeung MHY, Wong ANN, et al. Targeted sequencing ap- proach and its clinical applications for the molecular diagnosis of human diseases. Cells. 2023;12(3):493. https://doi.org/10.3390/ cells12030493

21. Lau SK, Weiss LM. The Weiss system for evaluating adrenocortical neoplasms: 25 years later. Hum Pathol. 2009;40(6):757-768. https://doi.org/10.1016/j.humpath.2009.03.010

22. Poli G, Ruggiero C, Cantini G, et al. Fascin-1 is a novel prognostic biomarker associated with tumor invasiveness in adrenocortical carcinoma. J Clin Endocrinol Metab. 2019;104(5):1712-1724. https://doi.org/10.1210/jc.2018-01717

23. Fassnacht M, Johanssen S, Quinkler M, et al. Limited prognostic value of the 2004 International Union Against Cancer staging clas- sification for adrenocortical carcinoma: proposal for a revised TNM classification. Cancer. 2009;115(2):243-250. https://doi. org/10.1002/cncr.24030

24. Cantini G, Niccolai E, Canu L, et al. Intratumour microbiota mod- ulates adrenocortical cancer responsiveness to mitotane. Endocr Relat Cancer. 2023;30(10):e230094. https://doi.org/10.1530/ ERC-23-0094

25. Crona J, Beuschlein F. Adrenocortical carcinoma-towards genom- ics guided clinical care. Nat Rev Endocrinol. 2019;15(9):548-560. https://doi.org/10.1038/s41574-019-0221-7

26. Ross JS, Wang K, Rand JV, et al. Next-generation sequencing of adrenocortical carcinoma reveals new routes to targeted therapies. J Clin Pathol. 2014;67(11):968-973. https://doi.org/10.1136/ jclinpath-2014-202514

27. Provenzano A, Chetta M, De Filpo G, et al. Novel germline PHD2 variant in a metastatic pheochromocytoma and chronic myeloid leukemia, but in the absence of polycythemia. Medicina (Kaunas). 2022;58(8):1113. https://doi.org/10.3390/medicina58081113

28. Toledo RA, Burnichon N, Cascon A, et al. Consensus statement on next-generation-sequencing-based diagnostic testing of hereditary phaeochromocytomas and paragangliomas. Nat Rev Endocrinol. 2017;13(4):233-247. https://doi.org/10.1038/nrendo.2016.185

29. Lippert J, Dischinger U, Appenzeller S, et al. Performance of DNA-based biomarkers for classification of adrenocortical carcin- oma: a prognostic study. Eur J Endocrinol. 2023;189(2):262-270. https://doi.org/10.1093/ejendo/lvad112

30. Cerami E, Gao J, Dogrusoz U, et al. The cBio cancer genomics por- tal: an open platform for exploring multidimensional cancer gen- omics data. Cancer Discov. 2012;2(5):401-404. https://doi.org/10. 1158/2159-8290.CD-12-0095

31. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evalu- ation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer. 2009;45(2):228-247. https://doi.org/10.1016/j. ejca.2008.10.026

32. Jennings LJ, Arcila ME, Corless C, et al. Guidelines for validation of next-generation sequencing-based oncology panels: a joint consensus recommendation of the association for molecular pathology and col- lege of American pathologists. J Mol Diagn. 2017;19(3):341-365. https://doi.org/10.1016/j.jmoldx.2017.01.011

33. Sukhai MA, Misyura M, Thomas M, et al. Somatic tumor variant fil- tration strategies to optimize tumor-only molecular profiling using targeted next-generation sequencing panels. J Mol Diagn. 2019;21 (2):261-273. https://doi.org/10.1016/j.jmoldx.2018.09.008

34. Li MM, Datto M, Duncavage EJ, et al. Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the association for molecular pathology, American society of clinical oncology, and college of American pathologists. J Mol Diagn. 2017;19(1):4-23. https://doi. org/10.1016/j.jmoldx.2016.10.002

35. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the in- terpretation of sequence variants: a joint consensus recommendation of the American college of medical genetics and genomics and the as- sociation for molecular pathology. Genet Med. 2015;17(5):405-424. https:/doi.org/10.1038/gim.2015.30

36. Vatrano S, Volante M, Duregon E, et al. Detailed genomic character- ization identifies high heterogeneity and histotype-specific genomic profiles in adrenocortical carcinomas. Mod Pathol. 2018;31(8): 1257-1269. https://doi.org/10.1038/s41379-018-0042-6

37. Harms KL, Chen X. Histone deacetylase 2 modulates p53 transcrip- tional activities through regulation of p53-DNA binding activity. Cancer Res. 2007;67(7):3145-3152. https://doi.org/10.1158/0008- 5472.CAN-06-4397

38. Kim S, Jeong S. Mutation hotspots in the ß-catenin gene: lessons from the human cancer genome databases. Mol Cells. 2019;42(1): 8-16. https://doi.org/10.14348/molcells.2018.0436

39. Assié G, Jouinot A, Fassnacht M, et al. Value of molecular classifica- tion for prognostic assessment of adrenocortical carcinoma. JAMA Oncol. 2019;5(10):1440-1447. https://doi.org/10.1001/jamaoncol. 2019.1558

40. Lippert J, Appenzeller S, Liang R, et al. Targeted molecular analysis in adrenocortical carcinomas: a strategy toward improved personal- ized prognostication. J Clin Endocrinol Metab. 2018;103(12): 4511-4523. https://doi.org/10.1210/jc.2018-01348

41. Philpott C, Tovell H, Frayling IM, Cooper DN, Upadhyaya M. The NF1 somatic mutational landscape in sporadic human cancers. Hum Genomics. 2017;11(1):13. https://doi.org/10.1186/s40246- 017-0109-3

42. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Signatures of mu- tational processes in human cancer. Nature. 2013;500(7463): 415-421. https://doi.org/10.1038/nature12477

43. Alexandrov LB, Kim J, Haradhvala NJ, et al. The repertoire of mu- tational signatures in human cancer. Nature. 2020;578(7793): 94-101. https://doi.org/10.1038/s41586-020-1943-3

44. Fojo T, Huff L, Litman T, et al. Metastatic and recurrent adreno- cortical cancer is not defined by its genomic landscape. BMC Med Genomics. 2020;13(1):165. https://doi.org/10.1186/s12920-020- 00809-7

45. Lerario AM, Mohan DR, Hammer GD. Update on biology and gen- omics of adrenocortical carcinomas: rationale for emerging therapies. Endocr Rev. 2022;43(6):1051-1073. https://doi.org/10.1210/endrev/ bnac012)

Downloaded from https://academic.oup.com/ejendo/article/191/1/17/7698912 by National Library of Medicine user on 03 April 2026

46. Batisse-Lignier M, Sahut-Barnola I, Tissier F, et al. P53/Rb inhib- ition induces metastatic adrenocortical carcinomas in a preclinical transgenic model. Oncogene. 2017;36(31):4445-4456. https://doi. org/10.1038/onc.2017.54

47. Borges KS, Pignatti E, Leng S, et al. Wnt/B-catenin activation coop- erates with loss of p53 to cause adrenocortical carcinoma in mice. Oncogene. 2020;39(30):5282-5291. https://doi.org/10.1038/s41388- 020-1358-5

48. Supek F, Miñana B, Valcárcel J, Gabaldón T, Lehner B. Synonymous mutations frequently act as driver mutations in human cancers. Cell. 2014;156(6):1324-1335. https://doi.org/10.1016/j.cell.2014.01.051

49. Zeng Z, Bromberg Y. Inferring potential cancer driving synonym- ous variants. Genes (Basel). 2022;13(5):778. https://doi.org/10. 3390/genes13050778

50. Buzdin A, Skvortsova II, Li X, Wang Y. Editorial: next generation sequencing based diagnostic approaches in clinical oncology. Front Oncol. 2021;10:635555. https://doi.org/10.3389/fonc.2020. 635555

51. de Biase D, Acquaviva G, Visani M, et al. Molecular diagnostic of solid tumor using a next generation sequencing custom-designed multi-gene panel. Diagnostics (Basel). 2020;10(4):250. https://doi. org/10.3390/diagnostics10040250

Downloaded from https://academic.oup.com/ejendo/article/191/1/17/7698912 by National Library of Medicine user on 03 April 2026