Low Protein Expression of both ATRX and ZNRF3 as Novel Negative Prognostic Markers of Adult Adrenocortical Carcinoma

Adrenocortical carcinoma (ACC) is a rare malignancy that is associated with a dismal prognosis. Pan-genomic studies have demonstrated the involvement of ATRX and ZNRF3 genes in adrenocortical tumorigenesis. Our aims were to evaluate the protein expression of ATRX and ZNRF3 in a cohort of 82 adults with ACC and to establish their prognostic value. Two pathologists analyzed immuno-stained slides of a tissue microarray. The low protein expression of ATRX and ZNRF3 was associated with a decrease in overall survival (OS) (p = 0.045, p = 0.012, respectively). The Cox regression for ATRX protein expression of >1.5 showed a hazard ratio (HR) for OS of 0.521 (95% CI 0.273–0.997; p = 0.049) when compared with ≤1.5; for ZNRF3 expression >2, the HR for OS was 0.441 (95% CI, 0.229–0.852; p = 0.015) when compared with ≤2. High ATRX and ZNRF3 protein expressions were associated with optimistic recurrence-free survival (RFS) (p = 0.027 and p = 0.005, respectively). The Cox regression of RFS showed an HR of 0.332 (95%CI, 0.111–0.932) for ATRX expression >2.7 (p = 0.037), and an HR of 0.333 (95%CI, 0.140–0.790) for ZNRF3 expression >2 (p = 0.013). In conclusion, low protein expression of ATRX and ZNRF3 are negative prognostic markers of ACC; however, different cohorts should be evaluated to validate these findings.


Introduction
Adrenocortical carcinoma (ACC) is a rare and aggressive cancer deriving from the adrenal cortex gland with an incidence of 0.7-2 cases/million/year [1][2][3]. Despite a dismal prognosis, the five-year overall survival (OS) rate varies, ranging from 60 to 80% when the tumor is restricted to the adrenal gland and from 0 to 28% in the advanced stages of the disease [3][4][5][6][7][8]. These data suggest that ACC is a heterogeneous adrenal disorder, presenting different behaviors, probably due to its different biology and genetic signatures [2,8,9].
In previous years, considerable advances in the genetic field of ACC tumorigenesis investigation have occurred, allowing patients to be stratified according to the combination of clinical-hormonal-pathological and genetic alterations.
Prognostic factors have an essential role in determining the aggressiveness of ACC. We can classify tumors according to (A) clinical data (age, hormone-related symptoms, and tumor stage); (B) pathological findings, genetic/protein background, and category of surgical procedure (Weiss system, modified Weiss system, Helsinki score, mitotic count, Ki-67 proliferation marker, SF-1, P53, β-catenin, and resection status); (C) clinical and pathological parameter associations (staging system of European Network for the Study of Adrenal Tumors modified by grade, resection status, age and symptoms-mENSAT plus GRAS); and (D) molecular classifications (chromosomal aberrations, methylome, gene expression alterations, microRNAs dysregulation, and gene mutations) [2][3][4]6,.
Pan genomic studies have identified new genes related to adrenocortical carcinogenesis, providing a refinement of ACC taxonomy, and improving the development of new biomarkers for early diagnosis and staging, conducive to the development of molecularly targeted therapies.
ATRX is a chromatin structure regulator and telomere structure maintainer. Somatic pathogenic variants and structural variations, which cause the deletion of multiple exons of ATRX were associated with germline TP53 pathogenic variants in a pediatric cohort [36]. Pinto et al. reported alterations in TP53 and ATRX in 32% of their cohort, and these alterations were associated with advanced disease and poor event-free survival in pediatric patients with adrenocortical tumor [36]. Assie et al. identified somatic ATRX alterations in only 4% of their adult ACC cohort. In contrast, ZNRF3 was the most frequently altered gene, corresponding to 21% of cases [27]. This gene encodes a cell surface transmembrane E3 ubiquitin ligase that acts as a negative feedback regulator of the canonical Wnt signaling (Wnt/β-catenin signaling). Zheng et al. studied an adult ACC cohort and reported a frequency of 19.3% for ACCs with deletions or non-silent somatic pathogenic variants in the ZNRF3 gene. In comparison, ATRX alterations were present in only four cases (4.3%) [31].
In this study, we aimed to evaluate the protein expression of the ATRX and ZNRF3 genes in a cohort of 82 adult ACC patients from a unique Complex tertiary center to investigate their potential role as prognostic markers.

ATRX Protein
Patients with ACC showing high ATRX expression, defined as tumor scored 2+ or higher, represented 47.5% of cohort ( Figure 1). ATRX high expression group had a median age at diagnosis 52.5 years. In contrast, patients with ACC with low ATRX expression had a median age of 34.7 years (Z = −3.349; p = 0.001; Mann-Whitney U test). ATRX was highly expressed in 90% of cases without hypercortisolism at initial presentation, while it was expressed at a low level in 59.2% of cases with hypercortisolism (X 2 (2) = 8.759; p = 0.013; Chi-squared test). The group with hypercortisolism presented a median ATRX expression score of 0.5 (ranging from 0 to 4), while the non-hypercortisolism group presented a median score of 3.08 (ranging from 0 to 4) (X 2 (2) = 6.728; p = 0.035; Kruskal-Wallis test). In this study, we aimed to evaluate the protein expression of the ATRX and ZNRF3 genes in a cohort of 82 adult ACC patients from a unique Complex tertiary center to investigate their potential role as prognostic markers.

ATRX Protein
Patients with ACC showing high ATRX expression, defined as tumor scored 2+ or higher, represented 47.5% of cohort ( Figure 1). ATRX high expression group had a median age at diagnosis 52.5 years. In contrast, patients with ACC with low ATRX expression had a median age of 34.7 years (Z = −3.349; p = 0.001; Mann-Whitney U test). ATRX was highly expressed in 90% of cases without hypercortisolism at initial presentation, while it was expressed at a low level in 59.2% of cases with hypercortisolism (X 2 (2) = 8.759; p = 0.013; Chi-squared test). The group with hypercortisolism presented a median ATRX expression score of 0.5 (ranging from 0 to 4), while the non-hypercortisolism group presented a median score of 3.08 (ranging from 0 to 4) (X 2 (2) = 6.728; p = 0.035; Kruskal-Wallis test).  The quantitative analysis of ATRX expression showed a positive correlation with age at diagnosis (coefficient 0.391; p < 0.001; Spearman's rank correlation coefficient) and negative correlations with tumor size (coefficient -0.263; p = 0.02; Spearman's rank correlation coefficient) and Weiss score (coefficient -0.245; p = 0.033; Spearman's rank correlation coefficient).
Thirty-nine patients from the cohort died due to ACC or its clinical complications during the study period. The ATRX expression cut-off point was statistically determined (a) ACC case 13 presenting high ATRX nuclear expression (score 4). Female patient, aged 28.7 y with virilizing syndrome, a Weiss score 3, and a European Network for the Study of Adrenal Tumors (ENSAT) system score of 2. (b) ACC case 32 with low ATRX nuclear expression (score 0). Female patient, aged 26.5 y with mixed syndrome (hypercortisolism and androgen excess), a Weiss score of 7, and an ENSAT system score of 4. (c) ACC case 77 with high ZNRF3 cytoplasmic expression (score 5; extent = 4, intensity = 1). Male patient, aged 65.6 y with mixed syndrome (hypercortisolism and androgen excess), a Weiss score of 6, and an ENSAT system score of 4. (d) ACC case 5 with low ZNRF3 cytoplasmic expression (score 0). Female patient, aged 85.4 y with hypercortisolism syndrome, a Weiss score of 7, and an ENSAT system score of 3. Scale bar = 50 µm.
Thirty-nine patients from the cohort died due to ACC or its clinical complications during the study period. The ATRX expression cut-off point was statistically determined and used for the OS analysis. The cut-off value ≤1.5 for ATRX expression had an impact on the Kaplan-Meier survival curve (X 2 (1) = 4.021; p = 0.045; Kaplan-Meier method and log rank test) ( Figure 2). In addition, the Cox regression of OS showed a hazard ratio (HR) of 0.521 (95%CI, 0.273-0.997) for ATRX expression of >1.5 when compared with expression of ≤1.5 (Wald = 3.881; df1; p = 0.049; Cox's proportional hazards model). and used for the OS analysis. The cut-off value ≦1.5 for ATRX expression had an impact on the Kaplan-Meier survival curve (X 2 (1) = 4.021; p = 0.045; Kaplan-Meier method and log rank test) ( Figure 2). In addition, the Cox regression of OS showed a hazard ratio (HR) of 0.521 (95%CI, 0.273-0.997) for ATRX expression of >1.5 when compared with expression of ≦1.5 (Wald = 3.881; df1; p = 0.049; Cox's proportional hazards model).

ZNRF3 Protein
The high expression of ZNRF3 protein was defined as the sum of extent and intensity parameters equaling ≥3+ (Figure 1). High ZNRF3 expression was present in 46.25% of the cohort and was associated with a small tumor size (average size of 9.7 cm), while low expression was associated with a large tumor size (average size of 13.3 cm) at diagnosis (Z = −2.665; p = 0.008; Mann-Whitney U test). Tumors with high ZNRF3 expression presented with lower Weiss score (average 4.7) than the group with low ZNRF3 expression (average 6.1) (Z = −2.942; p = 0.003; Mann-Whitney U test). ZNRF3 expression was related to the ENSAT staging system. High ZNRF3 expression was present in almost 89 % of patients with an ENSAT staging system score of 1 at diagnosis, 42.9% with an ENSAT score of 2, 29.4% with an ENSAT score of 3, and 44.4% with an ENSAT score of 4 (p = 0.033, Fisher's exact test). ZNRF3 protein expression had no relationship with hypercortisolism at clinical presentation (X 2 (2) = 2.035; p > 0.05, Chi-squared test). Regarding the outcome analysis, 64.1% of patients with ACC who progressed to death presented with a low level of ZNRF3 protein expression and had a median survival of 19 months (X 2 (1) = 4.716; p = 0.030; Chi-squared test).

ZNRF3 Protein
The high expression of ZNRF3 protein was defined as the sum of extent and intensity parameters equaling ≥3+ (Figure 1). High ZNRF3 expression was present in 46.25% of the cohort and was associated with a small tumor size (average size of 9.7 cm), while low expression was associated with a large tumor size (average size of 13.3 cm) at diagnosis (Z = −2.665; p = 0.008; Mann-Whitney U test). Tumors with high ZNRF3 expression presented with lower Weiss score (average 4.7) than the group with low ZNRF3 expression (average 6.1) (Z = −2.942; p = 0.003; Mann-Whitney U test). ZNRF3 expression was related to the ENSAT staging system. High ZNRF3 expression was present in almost 89 % of patients with an ENSAT staging system score of 1 at diagnosis, 42.9% with an ENSAT score of 2, 29.4% with an ENSAT score of 3, and 44.4% with an ENSAT score of 4 (p = 0.033, Fisher's exact test). ZNRF3 protein expression had no relationship with hypercortisolism at clinical presentation (X 2 (2) = 2.035; p > 0.05, Chi-squared test). Regarding the outcome analysis, 64.1% of patients with ACC who progressed to death presented with a low level of ZNRF3 protein expression and had a median survival of 19 months (X 2 (1) = 4.716; p = 0.030; Chi-squared test).
The quantitative expression of ZNRF3 showed negative correlations with tumor weight (coefficient −0.282; p = 0.026, Spearman's rank correlation coefficient), tumor size (coefficient −0.313; p = 0.005; Spearman's rank correlation coefficient), and the Weiss score (coefficient −0.345; p = 0.002; Spearman's rank correlation coefficient). The ZNRF3 expression cut-off of ≤2 had an impact on the OS analysis ((X 2 (1) = 6.277; p = 0.012; Kaplan-Meier method and log rank test) ( Figure 3). The Cox regression of OS showed an HR of 0.441 (95% CI, 0.229-0.852) for ZNRF3 expression >2 compared with the value for an expression level of ≤2 (Wald = 5.942; df1; p = 0.015; Cox's proportional hazards model). ZNRF3 is considered a suppressor of the Wnt/β-catenin pathway. Due to this known function, we analyzed the status of β-catenin in the tumoral cells. Active Wnt/β-catenin signaling was characterized by a cytoplasmic/nuclear staining for β-catenin [37]. The analysis demonstrated a positive correlation between ZNRF3 expression and β-catenin expression in the cell membrane (coefficient 0.202; p = 0.004; Spearman's rank correlation coefficient). These data support the function of ZNRF3 as a tumor suppressor, keeping βcatenin attached in the cell membrane ( Figure 4). ZNRF3 is considered a suppressor of the Wnt/β-catenin pathway. Due to this known function, we analyzed the status of β-catenin in the tumoral cells. Active Wnt/β-catenin signaling was characterized by a cytoplasmic/nuclear staining for β-catenin [37]. The analysis demonstrated a positive correlation between ZNRF3 expression and β-catenin expression in the cell membrane (coefficient 0.202; p = 0.004; Spearman's rank correlation coefficient). These data support the function of ZNRF3 as a tumor suppressor, keeping β-catenin attached in the cell membrane ( Figure 4). ZNRF3 is considered a suppressor of the Wnt/β-catenin pathway. Due to this known function, we analyzed the status of β-catenin in the tumoral cells. Active Wnt/β-catenin signaling was characterized by a cytoplasmic/nuclear staining for β-catenin [37]. The analysis demonstrated a positive correlation between ZNRF3 expression and β-catenin expression in the cell membrane (coefficient 0.202; p = 0.004; Spearman's rank correlation coefficient). These data support the function of ZNRF3 as a tumor suppressor, keeping βcatenin attached in the cell membrane ( Figure 4).

Recurrence-Free Survival
The recurrence-free survival (RFS) was defined as the period between the date of complete tumor resection and the date of the first radiological evidence of local or distant recurrence [38]. ATRX protein expression was associated with a better RFS when a cut-off >2.7 was statistically established (X 2 (1) = 4.920; p = 0.027; Kaplan-Meier method and log rank test) (Figure 7). The Cox regression of RFS showed an HR of 0.332 (95% CI, 0.111-0.932) for ATRX expression >2.7 compared with the value for an expression level of ≦2.7 (Wald = 4.365; df1; p = 0.037; Cox's proportional hazards model). The presence of high ZNRF3 protein expression (score > 2) was positively associated with RFS (X 2 (1) = 6.963; p = 0.008; Kaplan-Meier method and log rank test) (Figure 8), which means that the cases with high protein expression presented a greater recurrence-free survival than the group with low protein expression. This corroborates to ZNRF3 function of tumor suppressor. The Cox regression of RFS showed an HR of 0.333 (95% CI, 0.140-0.790) for ZNRF3 expression >2 compared with the value for an expression level of ≦2 (Wald = 6.222; df1; p = 0.013; Cox's proportional hazards model). Categorization of the Ki-67 proliferation marker into groups (<10%, ≧10% and <20%, ≧20%) impacted the RFS curve (X 2 (2) = 16.357; p < 0.001; Kaplan-Meier method and log rank test), as seen in Figure 9.

Recurrence-Free Survival
The recurrence-free survival (RFS) was defined as the period between the date of complete tumor resection and the date of the first radiological evidence of local or distant recurrence [38]. ATRX protein expression was associated with a better RFS when a cut-off >2.7 was statistically established (X 2 (1) = 4.920; p = 0.027; Kaplan-Meier method and log rank test) (Figure 7). The Cox regression of RFS showed an HR of 0.332 (95% CI, 0.111-0.932) for ATRX expression >2.7 compared with the value for an expression level of ≤2.7 (Wald = 4.365; df1; p = 0.037; Cox's proportional hazards model). The presence of high ZNRF3 protein expression (score > 2) was positively associated with RFS (X 2 (1) = 6.963; p = 0.008; Kaplan-Meier method and log rank test) (Figure 8), which means that the cases with high protein expression presented a greater recurrence-free survival than the group with low protein expression. This corroborates to ZNRF3 function of tumor suppressor. The Cox regression of RFS showed an HR of 0.333 (95% CI, 0.140-0.790) for ZNRF3 expression >2 compared with the value for an expression level of ≤2 (Wald = 6.222; df1; p = 0.013; Cox's proportional hazards model). Categorization of the Ki-67 proliferation marker into groups (<10%, ≥10% and <20%, ≥20%) impacted the RFS curve (X 2 (2) = 16.357; p < 0.001; Kaplan-Meier method and log rank test), as seen in Figure 9.
The Cox regression analysis revealed the following independent variables for death due to ACC: tumor weight (HR 39 (Table S1).

Discussion
Malignancies develop protector mechanisms for self-survival. Generally, defects are harbored in evolutionarily conserved pathways, such as the Wnt/β-catenin pathway [39]. However, nine acquired biological capabilities, known as cancer hallmarks, are involved in a multistep tumorigenesis process [40].
Telomeres are responsible for protecting genetic material and avoiding end-to-end fusion due to the repair process [41]. Telomeric shortening results from ineffective replication due to a problem in the DNA replication machinery [41][42][43]. When telomeres reach

Discussion
Malignancies develop protector mechanisms for self-survival. Generally, defects are harbored in evolutionarily conserved pathways, such as the Wnt/β-catenin pathway [39]. However, nine acquired biological capabilities, known as cancer hallmarks, are involved in a multistep tumorigenesis process [40].
Telomeres are responsible for protecting genetic material and avoiding end-to-end fusion due to the repair process [41]. Telomeric shortening results from ineffective replication due to a problem in the DNA replication machinery [41][42][43]. When telomeres reach a critical length, the cell enters a state of senescence, which is considered to be a natural tumor suppressor mechanism in humans [44,45]. Some malignancies present an alternative lengthening of telomeres (ALT) pathway to avoid senescence and apoptosis [41,46]. The ATRX gene encodes a chromatin remodeler (ATRX protein), which functions in nucleosome stability, DNA replication, transcription, maintenance of telomere and heterochromatin structure and stability [46][47][48][49][50]. Lovejoy et al. suggested that the inactivation of ATRX is a significant step in the ALT pathway [51]. Bower et al. showed that the loss of wild-type ATRX expression in somatic cell hybrids results in the development of the ALT pathway [52]. Barthel et al. analyzed 9127 patients and 31 cancer types and showed that inactivation of ATRX is associated with telomere length elongation, reinforcing the idea of an association between ATRX and ALT [53]. ATRX is suggested to be a suppressor of the ALT pathway [47]. Cohorts of different malignancies have shown that ATRX alterations have prognostic value [54][55][56][57][58][59][60][61][62][63][64][65].
The Wnt/β-catenin pathway is one of the central molecular mechanisms of embryonic development [66]. Although it is essential for the embryonic phase, the Wnt/β-catenin has a role in various malignancies. ZNRF3 is a transmembrane E3 ubiquitin ligase molecular target of R-spondin that is localized to the plasma membrane [67]. Hao et al. suggested that ZNRF3 inhibits Wnt/β-catenin signaling by promoting decreases in the membrane levels of frizzled and LRP6 [67]. Basham et al. reported ZNRF3 homeostatic regulation of the adrenal cortex, showing that its loss leads to severe adrenocortical hyperplasia [68]. In the same way as the ATRX gene, multiple cohorts have reported the role of ZNRF3 in tumorigenesis and its prognostic value in different malignancies [69][70][71][72][73][74][75][76][77].
We presented a younger cohort with median age 38.17 years (range 15.38-85.46 years), with more female patients (63,76.8%), and more deaths (39, 55%), compared to TCGA-ACC and ENSAT and Assie et al. cohorts [27,37]. However, the frequency of ENSAT system in the cohort was very similar with those of both cohorts. Unfortunately, we did not have access to genetic samples to enable determination of the frequency of alterations in these genes or to calculate correlations between molecular alterations and protein expression levels. Nevertheless, we were able to show the protein expression profiles of ATRX and ZNRF3 in our cohort.
Our objective was to analyze the ATRX and ZNRF3 protein expression in an ACC cohort; we chose the immunohistochemistry technique because of its availability and ease of use, aiming to bring pan-genomic studies closer to the clinical practice. The ATRX analysis considered the extent of cells. Unlike Mete et al. [32] who registered only the global loss of ATRX expression, we adapted the ATRX analysis from other neoplasia studies, intending to be more descriptive [78,79]. The ZNRF3 evaluation was based on studies from Qiu et al. and Yu et al. [71,73], based their characterization of the extent and intensity of ZNRF3 cytoplasmic staining.
ATRX protein expression is seen in the nuclear compartment of ACC cells (Figure 1). Mete et al. studied ATRX protein expression in 43 ACCs in a tissue microarray (TMA) and showed a global loss of nuclear ATRX expression in 48% of the cohort [32]. We registered the global loss of ATRX expression (score 0) in 47.5% (n = 38) of our cohort. The loss of ATRX protein expression and its association with disease-free survival or with the presence of adverse outcomes was not proven in the study by Mete et al. [32]. We demonstrated that low expression of ATRX, present in 52.5% of our cohort, impacted negatively on OS and RFS. Moreover, we proved the positive effect of ATRX expression (score >1.5) on prognosis with an HR of 0.521 (95% CI, 0.273-0.997) (p = 0.049).
Hypercortisolism has a negative impact on the OS and RFS of ACC, acting as a predictor of recurrence and death [26,[80][81][82]. Although cortisol-secreting ACCs have been associated with a worse OS [83], this was not shown in our cohort. ATRX was highly expressed (median score of 3.08) in the non-hypercortisolism group, and it showed a low expression (median score of 0.5) in the hypercortisolism group. The role of ATRX in cortisol secretion is unknown.
The high expression of ZNRF3 was associated with smaller tumors in our cohort. ZNRF3 expression was negatively correlated with tumor weight (p = 0.026) and tumor size (p = 0.005). Qin et al. assessed ZNRF3 expression in a cohort of gastric cancer patients and showed an association of protein expression with smaller tumor size [70]. They also revealed that ZNRF3 overexpression causes significantly more apoptosis and lowered proliferation of cancer cells by reducing the level of Lgr5, a component of Wnt/β-catenin signaling, while also reducing Gli1, a component of Hedgehog signaling (SHH) [70]. Gomes et al. showed that the SHH pathway is upregulated in adult ACC and downregulated in pediatric adrenocortical tumors [90]. If ZNRF3 has a role in the adrenal SHH pathway, it remains unclear.
Qiu et al. demonstrated the downregulation of ZNRF3 protein expression in human papillary thyroid carcinoma cell lines [73]. They showed that ZNRF3 overexpression strongly inhibits the migratory and invasive capacities of the cancer cells [73]. Yu et al. revealed that strong expression of ZNRF3 had a significant association with disease-free survival and OS in a colorectal carcinoma cohort [71]. We showed that low ZNRF3 expression impacted negatively on OS analysis in this ACC cohort. The Cox regression showed an HR of 0.441 (95% CI, 0.229-0.852) for tumors with high ZNRF3 expression; furthermore, this factor was positively associated with a better RFS than the loss of its expression.
Ki-67 proliferation marker is considered a prognostic factor in localized ACC and is a powerful tool for predicting recurrence after complete surgical resection [28]. Some studies have documented the importance of Ki-67 as a marker of cellular proliferation, providing a value in the ACC scenario [13,20,91,92]. We categorized the Ki-67 proliferation marker into groups (<10%, ≥10% and <20%, ≥20%), as suggested by Beuschlein et al. [28], and we confirmed the association of Ki-67 with OS and RFS. However, the expressions levels of ATRX and ZNRF3 were not shown to be associated with the Ki-67 proliferation marker.
Since ATRX and ZNRF3 are present in independent pathways, they were considered prognostic markers in this cohort, so we analyzed the association of low expression levels of both. When low expression levels were combined, a stronger negative prognostic value was found for this cohort (HR 0.314; 95% CI, 0.136-0.725; p = 0.007).
Tumors with active Wnt/β-catenin signaling due to CTNNB1, ZNRF3, or APC mutations present a dismal prognosis, according to TCGA Research Network. TCGA reported a high expression of ZNRF3 transcript by RNA sequencing (RNA-seq) levels is associated with a low survival rate in the ACC cohort. On the contrary, we showed that a low ZNRF3 protein expression is linked to poor survival.
The protein and transcript levels are in agreement in most cases, and some hypotheses may explain the difference in the ZNRF3 and ACC context. ZNRF3 is a target gene of Wnt/β-catenin signaling, and the levels of ZNRF3 transcript can reflect the level of Wnt/βcatenin signaling activity. However, ZNRF3 deletions may influence its transcript levels, which could be undetectable due to biallelic deletion; simultaneously, the Wnt/β-catenin signaling could be highly active due to the biallelic loss of its ZNRF3 suppressor. The TCGA analyzed the ZNRF3 transcript levels without excluding the biallelic deletions cases, and this could be a reason for the difference between transcript and protein expression. Another hypothesis is based on the function of ZNRF3 negative regulator by R-spondin proteins in the Wnt/β-catenin signaling. ZNRF3 protein expression may be more dependent on R-spondin proteins activity, through LGR-dependent and LGR-independent mechanisms, reflecting their activity, than the absolute quantity of its transcript.
TCGA also reported ATRX transcript expression by RNA-seq levels in the ACC cohort; however, the ATRX expression did not impact the survival rate.
Patients classified as having a low or moderate risk of recurrence (ENSAT 1 or 2, presenting microscopically complete resection and Ki-67 ≤10%), according to Fassnacht et al. [8], may benefit from the use of this novel prognostic tool. The presence of low expression levels of ATRX and ZNRF3, which presented a negative prognostic value, could guide the medical approach used for these particular patients to early adjuvant treatment, since these prognostic markers showed associations with OS and RFS.
Regarding germline TP53 pathogenic variants, two patients presented TP53 p.R337H variant (cases 51 and 58), and one patient presented TP53 p.R273H (case 68) (Table S2). Although all of them presented with hypercortisolism at initial presentation and died due to ACC (OS 7, 36, and 12 months, respectively), they presented with variable expression of ATRX (score 2, 0, and 0, respectively) and variable expression of ZNRF3 (score 3, 2, and 5, respectively). It remains unclear as to whether an association between the loss of ATRX expression and germline TP53 pathogenic variants exists and could impact OS in the adult ACC population. However, the association between genes has already been published in the pediatric adrenocortical tumor cohort and in other malignancy cohorts [36,57,[93][94][95].
Tumor biology knowledge is an important step in the delivery of personalized medicine, allowing the design of better clinical approaches and targeted therapies. This study brought pan-genomic studies closer to clinical practice with an easily accessible tool to refine prognostication in adult patients with ACC. In conclusion, we demonstrated that low levels of protein expression of both ATRX and ZNRF3 are negative prognostic markers of ACC; however, different cohorts should be evaluated in future studies to validate these findings.

Materials and Methods
The Ethics Committees of the Hospital das Clínicas-University of São Paulo approved this study (n.2.394.934/2017). Informed written consent was obtained from all patients.
All patients underwent an adrenalectomy due to ACC, and they were followed in a unique tertiary center by a team specialized in adrenal disorders. Detailed clinical data, including clinical and hormonal presentations, treatments, follow-up and survival data, imaging, and histopathological results, were collected from medical records. Laboratory results were adjusted according to the delta of reference value, aimed at carrying out a comparison since the laboratory tests were not performed in the same laboratory. ACCs were diagnosed according to Weiss Score (Weiss ≥ 3) [10]. The tumor stage was classified according to the ENSAT and Helsinki systems [5]. Enrolled patients were followed up in the Adrenal Ambulatory-HC/FMUSP and in the Instituto do Cancer do Estado de Sao Paulo-ICESP for an average time of 85.35 months (1.4-406.83 months) ( Table 1). A total of 82 adrenocortical carcinomas were evaluated in terms of the protein expression of ATRX, ZNRF3, and β-catenin. Detailed information about the cohort is presented in the Supplemental Material (Table S2).

Tissue Microarrays (TMA) and Immunohistochemistry
The tumor series included a total of 82 formalin-fixed paraffin-embedded surgical cases of ACC. Representative areas of the ACCs (viable tumor tissue without necrosis) were identified on hematoxylin-and eosin-stained slides and marked on paraffin donor blocks by two experienced pathologists (I.C.S., M.C.N.Z.). The spotted areas of the donor blocks were punched (1.0 mm punch) and mounted into recipient paraffin blocks using a precision microarray instrument (Beecher Instruments, Sun Prairie, WI, USA), with a total of 3 cores per ACC case. The organization of cores was guided by a cartesian map constituting the three TMAs. Control samples (kidney and liver) were included for TMA orientation. One set of three slides was selected (one slide from each of the three TMA paraffin blocks of the triplicate) for staining with anti-ATRX rabbit polyclonal antibody (titer 1:1000; code HPA001906; lot BH115094; 100 µL (0.2 mg/mL); Sigma-Aldrich, Darmstadt, Germany, anti-ZNRF3 rabbit polyclonal antibody (titer 1:100; code HPA036703; lot A86511; 100 µL (0.1 mg/mL); Sigma-Aldrich, Germany), and anti-β-catenin mouse monoclonal antibody (titer 1:200; clone 14/β-catenin; code 610154; lot 4351813, 150 µg (0.6 mL, 250 µg/mL) BD Transduction Laboratories). The positive control for immunostaining with ATRX and ZNRF3 antibodies was an ACC sample with preserved ATRX immunoexpression and with a known wild type ATRX status, and a colon carcinoma, respectively. Tonsil tissue was used as a positive control for b-catenin and Ki-67 antibodies. A modified immunoperoxidase immunohistochemical method with humid heat antigen retrieval was used as previously described and validated [38,100]. The immunostaining evaluations were performed by two independent observers (I.C.S. and M.C.N.Z.), who were unaware of clinical data, who independently evaluated ATRX, ZNRF3, and β-catenin staining. The average value of the two evaluations was taken for statistical analysis [22,97]. TMA samples were included in the analysis only if two or more evaluable cores were available after the staining procedure. Two ACC cases were excluded from analysis due to the loss of tissue sample (1.62%). The interobserver agreement coefficient (kappa) for the staining evaluation was 0.854 (T = 11.76; p < 0.0001; Cohen's kappa coefficient). A kappa coefficient >0.6 is considered to represent substantial agreement [101].
Nuclear staining for Ki-67 was also evaluated, and another set of three slides were stained with mouse monoclonal anti-human Ki-67 antigen (titer 1:40; clone MIB-1; code M7240; lot 95324; 0.2 mL (46 mg/L); Dako, Glostrup, Denmark). The stained slides were scanned by the Scanner of histological slides through Panoramic Viewer 1:15 software (3DHISTECH, Budapest, Hungary). Image Pro Plus 4.5 software (MediaCybernetics, Rockville, MD, USA) was used to process the images, and the positive and total nuclei were automatically counted.

Statistical Analysis
The baseline patient characteristics were expressed as absolute and relative frequencies for qualitative variables and as the average, standard deviation, median, minimum, and maximum for quantitative variables. The association among qualitative variables were evaluated by the chi-squared test or Fisher's exact test. The Spearman's rank correlation coefficient was calculated for the quantitative variables. The Kaplan-Meier estimator of the survival function was applied for survival analysis, and the log-rank test was used to compare the survival function between groups. Regarding the expression of ZNRF3 and ATRX, the determination of two groups of observations for a simple cut-off point was estimated using the maximum of the standardized log-rank statistic proposed by Lausen and Schumacher in 1992 [105]. The Cox semiparametric proportional hazards model was fitted to describe the relationship between OS or RFS and the covariates [106]. OS was defined as the time from the date of ACC diagnosis to the date of disease-related death or the last follow-up visit. RFS was defined as the time from the date of complete tumor resection to the date of the first radiological evidence of local and/or distant recurrence [38].
The assumption of proportional hazards was assessed based on the so-called Schoenfeld residuals [107,108]. There was evidence that covariates had a constant effect over time in all cases. The significance level was fixed at 5% for all tests. Statistical analyses were performed using IBM SPSS Statistics version 23.0 (IBM Corp., Armonk, NY, USA) and R software version 3.6 (R Foundation for Statistical Computing, Vienna, Austria).  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available in Supplementary Material (Tables S1 and S2).

Conflicts of Interest:
The authors declare no conflict of interest.